Combination toxicity of etoposide (VP-16) and photosensitisation with a water-soluble aluminium phthalocyanine in K562 human leukaemic cells.

Etoposide (VP-16) is an anti-cancer drug commonly used against several types of tumours and leukaemia, either alone or in combination chemotherapy. Photodynamic therapy (PDT) is another, relatively new modality for treatment of various malignancies. The interactions between VP-16 and PDT, using aluminium tetrasulphophthalocyanine as photosensitiser, in K562 human leukaemic cells were investigated. Cell responses to individual and combined drug treatment under different experimental conditions revealed synergistic drug toxicity. The latter was evident from various events of cell response, including supra-additive accumulation of cells in G2/M cell cycle phase and endonucleolytic DNA fragmentation (apoptosis). The involvement of the cellular antioxidant system in the synergistic interactions of photosensitisation and VP-16 is proposed.

agent used against several tumour types, either alone (Issel et al., 1984), or in combination therapy (Aisner and Lee, 1991). Etoposide is also commonly used for the treatment of acute myelogenous leukaemia (Champlin and Gale, 1987). The cytotoxicity of VP-1 6 is generally believed to be based on introduction of DNA damage by drug interference with breakagereunion reaction of DNAtopoisomerase II (formation of DNA -protein cross-links) (Glisson and Ross, 1987;Liu, 1989) and/or induction of direct DNA strand breaks and adducts (van Maanen et al., 1988a, b;Mans et al., 1991). In contrast to the topoisomerase II (topo II) poisoning by VP-16 itself, the direct inactivation of DNA was suggested to be conjugated with oxidation-reduction activation of VP-16 in the cellular environment (Mans et al., 1990(Mans et al., , 1992. In particular, cytochrome P450-dependent mono-oxygenases, peroxidases, prostaglandin synthetase, tyrosinase, etc. may be involved in VP-16 metabolic transformation (van Maanen et al., 1987;Haim et al., 1991;Gantchev et al., 1994a). Evidence has been found that peroxidative metabolic products of VP-16, e.g. the ortho-quinone derivative of etoposide, as well as the short-lived intermediates (phenoxyl and semi-quinone free radicals) are involved in various oxidative reactions and DNA damage (Mans et al., 1990(Mans et al., , 1991Sinha et al., 1990). Recent studies suggest that the interactions of VP-16 free radicals with intracellular reductants (thiols, ascorbic acid, etc.) might play an essential role in the cytotoxic activity of the drug as well (Mans et al., 1992;Kagan et al., 1994;Yokomizo et al., 1995).
Light activation of photosensitisers that have been accumulated in tumours is the basis of the photodynamic therapy (PDT) of cancer. Cytotoxic action of photosensitisers may involve oxidative damage to different cell constituents, including depletion of the pool of cell antioxidants (free and protein-bound thiols, ascorbic acid, a-tocopherol, etc.) (Buettner, 1984, Shopova and Gantchev, 1990, Gantchev and van Lier, 1995. Metallo-phthalocyanines (MePc) constitute a class of dyes proposed as second-generation photodynamic agents to supplant Photofrin, a mixture of porphyrin derivatives, currently used in the clinical treatment of various malignancies. Highly water-soluble phthalocya-Correspondence: TG Gantchev Received 28 March 1996; revised 4 June 1996; accepted 12 June 1996 nines, such as di-through tetrasulphonated derivatives (MePcS2 4) partially localise in cytoplasm and photosensitise intracellular generation of hydroxyl and organic free radicals (Gantchev et al., 1994b). Among different effects of photosensitised cell damage, tetrasulphonated Al-and ZnPcS4 have been shown to weaken cell viability by inactivation of catalase (Gantchev and van Lier, 1995) and inflicting damage to DNA (Hunting et al., 1987;Gantchev et al., 1994c). We have also shown that these phthalocyanines can initiate oxidative transformations of VP-16 in solution via photosensitised generation of VP-16 phenoxyl radical (Gantchev et al., 1994a).
In view of the above properties of phthalocyanines and the interrelation between the rate of VP-16 oxidative transformation, its cytotoxicity, and the activity of intracellular antioxidant systems, we hypothesised that PDT, in conjunction with etoposide, could result in enhanced cytotoxicity. Analysis of different effects of combination therapy with VP-16 and AlPcS4 photosensitisation against K562 chronic myelogenous leukaemia cells is the subject of the present work.
Materials and methods Chemicals Etoposide (VP-16) was purchased from Sigma (St Louis, MO, USA). The compound was dissolved in dimethyl sulphoxide (DMSO) at 5 mM, aliquoted and stored at -20°C. Further dilutions were made in RPMI medium immediately before use. Although the final DMSO concentrations of 1-2% in cell cultures were not toxic, all control groups received an equivalent amount of DMSO. Aluminium tetrasulphonated phthalocyanine (AlPcS4) was synthesised in our laboratory, purified by high-performance liquid chromatography (HPLC) and dialysis to homogeneity and dissolved in phosphatebuffered saline (PBS) to yield a stock solution of 1.5-2.0 mm, as measured spectrophotometrically (8695 nm = 2.5 x 104 M-' cm-' in dimethylformamide). All other chemicals used were of the highest available purity.
Cell culture and drug treatments Human chronic myelogenous leukaemia cells, K562 (ATCC CCL 243), were grown in RPMI-1640 medium, supplemented with glutamine, 10% fetal bovine serum (FBS), 50 ,ug ml-' gentamycin and 10 mM Hepes. All experiments were performed with exponentially growing (asynchronous) cells. Usually 12-14 h before the experiment, the cells were seeded in a complete RPMI medium (3:1 v/v parts fresh to conditioned medium) at cell density of approximately 0.4 x 106 cells ml-'. On the next day the cell density was adjusted by a small volume of fresh medium to give 0.5 x 106 cells ml-' and cells were exposed to drugs (VP-16 and AlPcS4 alone, or mixed). All experiments were performed in triplicate flasks. In a typical protocol, after incubation with drugs, cells were first washed once with PBS, and a second time with RPMI (without FBS). After resuspending in a complete RPMI medium (conditioned to fresh medium = 1: 3 v/v), cells were exposed to broad spectrum red light (A> i580 nm; I = 10 mW cm -2). Cell suspensions (12 -15 ml) were exposed to red light in ventilated (green-cap Falcon, 75 cm-2) incubation flasks and were gently shaken during irradiation. Cells were incubated at 5 x 105 cells ml-1, or diluted to 5 x 104 cells ml-' to obtain a longer exponential phase. Control experiments performed on cells that were first photosensitised in PBS and then resuspended in culture medium did not show any significant difference in cell survival. Dark incubation with AlPcS4 was not cytotoxic. Exposure to red light of cells incubated with VP-16, in the absence of AlPcS4, did not result in altered VP-16 toxicity.
Cytotoxicity assays and drug interaction analysis Cell growth inhibition was monitored by means of dye exclusion (staining with 0.4% Trypan blue) and/or by Coulter counter measurements. Under our experimental conditions, the doubling time of control cells was 22-24 h. The growth suppression activity of drugs was compared by estimation of the exponential rate constants using initial portions of growth curves (e.g. 48-60 h after treatment). Clonogenic assay was performed in a standard fashion by plating cells in soft (0.33%) agar in 35 mm Petri dishes (usually 300, 600 and 1200 cells per dish in triplicate).
To analyse the dose -effect relationships in combination treatment of cells with VP-16 and AlPcS4 photosensitisation, two algorithms were used. A simple estimate was performed using the fractional product method (Veleriote and Liu, 1975). This 'multiplicative' model predicts additivity, antagonism and/or synergism of two drugs based on the comparison between the individual drug effect on cell survival, fu (expected) = (fu)I x (fu)2 and the experimentally obtained value of cell survival after combination treatment, (f),2. The symbol (f0),2 stands for 'fraction unaffected' and is equal to 1-(f)L,2, where fa ('fraction affected') refers to the fraction of cells responding to various concentrations of the two drugs in combination. By definition, (u)1,2 lower, equal to or larger than fu (expected) determines the border lines of synergistic (supra-additive), additive and antagonistic drug interactions respectively. In a specially designed set of experiments, we also performed a detailed analysis of drug interactions using the median effect principle. Unlike other methods often used to predict drug interactions in biological systems (usually applicable only to mutually exclusive interactions), the median effect principle may be used to analyse both mutually exclusive and mutually non-exclusive interactions (Chou and Talalay, 1983). We used the linearised median effect equation in the form of log [(fu) -'-1] = m log (D) -m log (Din), where 'm' is the Hill-type coefficient determining the sigmoidality of the dose -effect curve; 'D' is drug(s) dose; and Dm = IC50 is the dose required to produce the median effect. The combination index (CI) was calculated at fa levels of 0.1 intervals, as described by Chou and Talalay (1983). According to this algorithm, CI values <1.0 indicate synergism; CI > 1.0, antagonism; and CI approximately 1.0, additivity. Flow cytometric cell cycle analysis After given time periods following drug removal, cells were washed and resuspended in 200 PBS containing 5.5 mM glucose and 1 mM EDTA. While vortexing, 1 ml of cold 70% ethanol was added and cells were incubated at 4°C for 30 min. Thereafter, the cells were transferred to PBSglucose-EDTA buffer containing 0.5% Tween-20 and incubated for 1 h at 4°C to rehydrate. Following a brief exposure to RNAase, DNA was stained with 50 pg ml-' propidium iodide. Before cytoflow analysis, cells were filtered through nylon mesh to avoid clumps. The readings were performed on a Hewlett-Packard computer-controlled Becton Dickinson FACScan instrument. DNA chromatograms were analysed by the SOBR software supplied by the manufacturer. In cases in which high levels of cell debris were present, appropriate gate cut-off settings were used.
Analysis of internucleosomal DNA fragmentation The integrity of DNA from drug-treated and control cells was assessed by agarose electrophoresis. After removal of culture medium at different post-treatment times, cells (0.5 2 x 106) were resuspended in 150 pl PBS-glucose-EDTA buffer and transferred to Eppendorf microfuge tubes. Immediately after that, 0.5 ml of lysis buffer [0.8% sodium dodecyl sulphate (SDS) and 100 mM EDTA in 50 mM Tris-HCI, pH=7.4] was added and cells were incubated at room temperature for 20 min. After addition of 125 pl of 5 M sodium chloride to each tube, samples were gently mixed and left overnight at 4°C. On the next day, samples were microcentrifuged at 12 000 x g for 30 min at 4°C and the pellets were carefully removed and discarded. Supernatants were further incubated with 0.45 mg ml-' proteinase K for 2 h at 50°C. After addition of 50 p1 of 5 M sodium chloride and 1 ml isopropanol, samples were left for 1 h at -20°C and microfuged. The pellet was washed with 70% ethanol and dried under vacuum. Resuspended in 20 pl TE buffer, pellets were incubated with 0.8 mg ml-' RNAase A at 37°C for 3 h. After mixing with 4 p1 of 6 x loading buffer (TE/ bromophenol/glycerol), samples were transferred onto 1.2% agarose gel and electrophoresed at 40 V for 4 h. Gels were stained for 40 min with 10 ng ml-' ethidium bromide, destained for 30-60 min in distilled water and photographed under UV light. Series of experiments were performed to selected post-treatment time periods and number of cells required to give early and easily detectable DNA fragmentation in combination with selected multiple drug dose equivalents to reveal drug interaction.

Results
Cell growth and loss of clonogenicity To determine the toxicity range of individual drugs and their combination effects, several parameters were examined: incubation times of cells with drugs, drug concentrations and the applied light dose for AlPcS4 activation. The cellular response to drug treatment was followed by monitoring the growth rate and cell clonogenic activity. Growth curves shown in Figure 1 exemplify the evolution of immediate drug toxicity with respect to drug incubation time, concentrations and light dose. For individual drugs and their combinations, growth curves show a proliferation lag period, followed either by predominant cell regrowth or cell death. At low levels of drug treatment (Figure la and b), and during the initial proliferation lag-times (e.g. at post-treatment times shorter than 24 h), the number of dead cells did not exceed 5-8% of the total, but cell cycle progression was largely inhibited (see below). about 20 jgM (D =9 J cm-2), and increased when cells were exposed to photosensitiser for a shorter time (e.g. 30-35 guM at 2 h). The simple 'multiplicative' model and data from cloning experiments (e.g. shown in Figure 2) largely predict supra-additive (synergistic) combined toxicity [i.e. (fu)l,2<fu Materials and methods]. A complete analysis of drug interaction, however, was performed using the median effect principle, and is described below.
Cell cycle arrest It has been shown previously that continuous exposure to low concentrations of VP-16 allows cells of different origin to traverse the cell cycle until a predominant number of them are blocked with the DNA content of G2 cells, but cannot progress to the mitotic stage (Krishan et al., 1975). In the present study, the cells were incubated with VP-16 for given time periods, and, thereafter, the drug was removed. Under perhaps, apoptotic cells). In contrast, when the toxic effects of individual drugs were low (growth curve shown in Figure  la), combined treatment induced a supra-additive (synergistic) accumulation of cells in G2/M-phase, as also proved by the 'multiplicative' model ( Figure 4b). Synergistic increase of the number of cells accumulated in G2/M-phase was reached under various other conditions, but especially when the toxicity of at least one of the drugs was kept low (Figure 4c).
Internucleosomal DNA degradation K562 leukaemic cells are known to be relatively resistant to VP-16-induced internucleosomal DNA fragmentation (apoptosis) (Dubrez et al., 1995). Thus, exposure of cells to VP-16 of concentrations around or higher than IC50 (5 gM and 10 gM for 1 or 2 h) did not reveal any significant DNA fragmentation during post-treatment incubation times up to 36 h, as assessed by agarose gel electrophoresis ( Figure 5, lanes 4 and 5). DNA fragmentation was induced when cells were treated with higher VP-16 concentrations (e.g. 5 x IC5o), but the pattern was obscured even when DNA was isolated from 2 x 106 or more cells (not shown). In contrast, AlPcS4 photosensitisation was a much more potent inducer of DNA fragmentation (apoptosis). Internucleosomal DNA damage was readily detected 4-6 h post treatment in DNA samples isolated from 0.5-1.0 x 106 cells treated with drug dose <IC50 ( Figure 5, lane 3). This pattern was persistent and progressively augmented during the next 48-72 h of posttreatment incubation. Figure 5 also shows the results from a typical experiment when cells were exposed to two drug concentrations individually, or to their combination. Under the specified conditions (see figure legend), no extensive DNA laddering took place after treatment with either 5 gM AlPcS4 or 5 and 10 gM VP-16 alone (lanes 2, 4 and 5).
Internucleosomal cleavage, however, was well pronounced in all samples when cells were subjected to combined treatment. It is also noteworthy that in every case when DNA internucleosomal fragments were easily seen, they were superimposed on DNA smears. This is probably caused by the occurrence of two parallel processes of DNA degradation when cells were more heavily damaged: unprogrammed (necrosis) and programmed (apoptosis) cell death. The technique of agarose electrophoresis, as applied in this study, does not permit a direct quantitative estimate of the levels of internucleosomal damage induced by single and combined treatment. From the selected drug doses and the results shown in Figure 5, however, it is evident that simultaneous employment of VP-16 with photodynamic treatment induces supra-additive increase of DNA fragmentation.

Median effectcombination index analysis
The results shown in previous sections indicate that drug combination treatment in most experimental conditions is supra-additive (synergistic). To evaluate modes of drug interactions, we designed experiments suitable for application of combination index analysis. This analysis is based on the median effect principle and is a statistical technique that allows formal evaluation of the nature of interaction between two cytotoxic agents. Cells were incubated with different concentrations of drugs, but at constant molar ratio (VP-16/ AlPcS4 = 1/2). Before irradiation, the cells were exposed to AlPcS4 for 1.5 h, followed by co-incubation with VP-16 for 0.5 h. Thereafter, the cells were washed and irradiated (D = 7.2 J cm-2). Drug toxicity was assessed by their effect on growth rate and clonogenicity. Figure 6 demonstrates that, under these conditions, the AlPcS4 phototoxicity was low, as estimated by both growth rate inhibition and loss of clonogenicity. However, growth rate assessment of VP-16 toxicity underestimates the reproductive toxicity of the drug ( Figure 6). The median effect plots derived from cloning experiments and linear regression data fit are shown in Figure  7. The dose-effect relationships of the individual drugs are strictly linear (r > 0.999, lines 1 and 2), while the plot obtained for the mixture of the two drugs tends to concave upward (r = 0.99) and intersects the plot of the more active drug (Figure 7, line 3). It is, therefore, apparent that the combined drug action is synergistic. This was further confirmed by calculating the combination index values (CI) for a wide range offa ('fraction affected', Figure 8). Since the plots for the individual drugs are almost parallel (lines 1 and 2 in Figure 7), and the plot for the combined treatment (line 3) intersects the plot of the more active drug, it is likely that the two drugs interact as mutually non-exclusive. The combination index was calculated assuming both possibilities, mutually exclusive and non-exclusive interactions. The plots shown in Figure 8 indicate that these two models predict synergistic toxicity (CI < 1) under conditions in which combined treatment results in fa >0.10 or fa >0. 15 respectively. For a comparison, CI values obtained from growth inhibition data were also calculated and included in Figure 8. It can be seen that synergism in the latter case is somewhat overestimated, probably due to the underestimation of VP-16 toxicity by this criterion.

Discussion
We have investigated the effects of individual toxicity of etoposide (VP-16) and photodynamic treatment with AlPcS4, as well as the efficacy of drug combination against K562 human leukaemic cells. Similarly, in our previous studies with Namalva Burkitt's lymphoma cells (Gantchev et al., 1994b, photosensitisation of K562 cells results in division block and, depending on the extent of the treatment, is followed either by cell regrowth or by predominant cell death. Unlike photosensitisation with Photofrin, which induces S-phase cell cycle arrest (Gantchev et al., 1994d), in the present study we show that AlPcS4 photosensitisation arrests cells in G2/Mphase. Under the conditions of relatively mild photodynamic treatment employed, no evidence for immediate membrane disintegration was found. In contrast, most of the toxic effects were delayed and emerged as a function of posttreatment incubation time, as seen, for example, by the time course of the accumulation of G2/M-arrested cells and internucleosomal DNA cleavage. Similar cell growth inhibition effects were observed after short ( (2) AlPcS4 and (3) their combination. Experimental data from clonogenic survival (Figure 6).  DNAprotein cross-links (Hunting et al., 1987;Gantchev et al., 1994c), and impairment of cell organelles, including peroxisomes (Peng et al., 1991). All these oxidative processes may account for AlPcS4 cytotoxicity and represent primary events in AlPcS4 -PDT-mediated internucleosomal DNA cleavage.
hle cytostatic/cytotoxic action Of VP-1 iS generally believed to involve formation of DNA-protein cross-links and breaks. However, it is likely that drug toxicity is not limited to topoisomerase II (topo II) poisoning only. Thus, a recent study argued on the lack of correlation between the level of DNA-protein cross-links and cytotoxicity in several leukaemic cell lines (Dubrez et al., 1995). Previously, other studies have also indicated that the interaction of VP-16 with topo II and cytotoxicity of the drug can be uncoupled, e.g. drug interaction with topo II might be not intrinsically cytotoxic, and therefore it has been suggested that VP-16 cytotoxic action might involve one, or more, intervening metabolic steps (Kaufmann, 1989;Walker et al., 1991). Fraction affected (fa) K562 cells to VP-16. The cytotoxicity however, evolved faster compared with VI lar, PDT-induced growth and clonogenic oped at similar rates, while VP-16 displaye growth suppression, compared with the clonogenic inhibition ( Figure 6). When ap 16 did not induce any significant internu degradation (apoptosis). Apoptotic cell dez response to AlPcS4-PDT and was augmente drug treatment ( Figure 5). Synergistic ( combined drug treatment was attained conditions, as clearly demonstrated by g rates (Figure la and d), loss of clonogenici accumulation of cells in G2/M cell cycle pha c). In particular, synergism (supra-additiv arrest took place when the fraction of G2/M single drug treatment did not exceed a v mately 60%. Quantitatively, this value co normal S-phase fraction of exponentially gr and implies that drug treatment exclusi' during the stage of DNA synthesis. Aftei treatment and during the accumulation o phase, a progressive increase of DNA fragmentation was observed. It is, there when cells are more extensively damaged b treatment, G2/M-arrested cells tend to fall cycle and die (eventually by apoptosis).
mechanisms that are involved in AlP apoptosis were not addressed in this stuc suggest that they do not encompass the phos A2, signalling pathway shown to operat sensitisation with cytoplasmic membrane-l stituted) AlPc (Agarwal et al., 1993). In th primary target is the cytoplasmic membrane Maanen et at., 1987; Mans et al., 1990;Haim et al., 1991).
Furthermore, the intermediate species from VP-16 metabolism: VP-16 phenoxyl and semi-quinone radicals, and one of 0.8 1.0 the final products of drug demethylation (VP-16 orthoquinone) have been implicated as important cytotoxic agents action affected (fa) (Mans et al., 1991(Mans et al., , 1992Ritov et al., 1995). It is, however, kic toxicity of the noteworthy that the intracellular accumulation of these as calculated for products depends strongly on the presence of cellular Kclusive drugs (Fi, antioxidants (e.g. reduced ascorbate and thiols) (Gantchev et al., 1994a;Kagan et al., 1995). The combination index analysis of AlPcS4 photosensitisation and VP-16 treatment as performed in the present work largely predicts synergistic toxicity of the two drugs. The of AlPcS4-PDT, experiments throughout this study were performed so that P-16. In particu-effects, such as cell synchronisation, were avoided and cannot inhibition develaccount for the observed synergy. Moreover, the median J a lower level of effect plot of combined toxicity (Figure 7) suggests that the more effective drugs are likely to interact as mutually non-exclusive. Non-)plied alone, VP-exclusivity of drug interaction would imply that AlPcS4 and .icleosomal DNA VP-16 share a common intracellular target and/or interact ath was an early chemically. It is not known if AlPcS4 photosensitisation can d after combined directly interfere with the topo II/DNA breakage-reunion cell response to process, but it is recognised that photosensitisation (Hunting under various et al., 1987;Gantchev et al., 1994c) and VP-16 metabolites ,rowth inhibition (van Maanen et al., 1988b;Mans et al., 1990Mans et al., , 1991; Sinha et ty ( Figure 2) and al., 1990) can instantly induce damage to DNA. Therefore, se (Figure 3b and based on the assumption that metabolic oxidoreductive ity) in cell cycle transformations of VP-16 are important for its cytotoxic I cells arrested by action, we suggest that photosensitisation-induced depletion ralue of approxi-of intracellular reductants facilitates VP-16 metabolism, trresponds to the probably on the level of VP-16 phenoxyl free radical -owing K562 cells transformations. Generation of the VP-16 phenoxyl radical vely affects cells is the primary step in the enzymatic metabolism of the drug r combined drug and the interactions of the phenoxyl radical with cell tf cells in G2/Mantioxidants prevents etoposide from further transformainternucleosomal tions and results in recovery of its original form. Also, the fore, likely that VP-16 phenoxyl radical may itself initiate oxidative cell ty combined drug damage (Ritov et al., 1995). A parallel drug interaction out from the cell mechanism may involve direct AlPcS4-mediated photo- The molecular oxidation of VP-16 to yield the phenoxyl radical (Gantchev cS4-PDT-induced et al., 1994a). Complete elucidation of the role of these ly. However, we different pathways and the involvement of additional pholipases C and processes that may explain the observed synergistic drug ;e during phototoxicity will require further experimental work. It is relevant ocalising (unsub-to note that a recent study showed synergism in cells e latter case, the subjected to gamma radiation and etoposide treatment , and it has been (Haddock et al., 1995). Since the oxidative processes induced by PDT and radiotherapy are similar, it is possible that the synergistic interactions with etoposide are based on the same mechanisms. Although the presented results only involve in vitro synergy between photodynamic treatment and etoposide toxicity, our findings may provide alternative perspectives for clinical PDT and etoposide applications.