Cytotoxic properties of a 4-nitroimidazole (NSC 38087): a radiosensitizer of hypoxic cells in vitro.

5-Phenoxysulphonyl-1-methyl-4-nitroimidazole (NSC 38087) can act as a sensitizer of hypoxic mammalian cells to radiation in vitro. The degree of sensitization achieved is greater than would be predicted from the drug's electron affinity. Cytotoxicity studies have shown that 5 microM NSC 38087 can reduce the surviving fraction of log-phase V79 cells in air at 37 degrees C to 10(-2) after 2 h exposure. This toxicity is considerably increased by small rises in temperature. In contrast to other nitro-heterocyclic radiosensitizers, NSC 38087 and related 5-substituted 4-nitroimidazoles show greater toxicity towards aerobic than to hypoxic cells. Log-phase cells show the highest sensitivity to the toxic action of NSC 38087, with plateau-phase cells, cells with a history of chronic hypoxia, and thermotolerant cells showing greater resistance. These toxicity data are compared to those for the 2-nitroimidazole hypoxic-cell sensitizer misonidazole.

Summary.-5-Phenoxysulphonyl-l-methyl-4-nitroimidazole (NSC 38087) can act as a sensitizer of hypoxic mammalian cells to radiation in vitro. The degree of sensitization achieved is greater than would be predicted from the drug's electron affinity. Cytotoxicity studies have shown that 5pM NSC 38087 can reduce the surviving fraction of log-phase V79 cells in air at 37°C to 10-2 after 2 h exposure. This toxicity is considerably increased by small rises in temperature. In contrast to other nitroheterocyclic radiosensitizers, NSC 38087 and related 5-substituted 4-nitroimidazoles show greater toxicity towards aerobic than to hypoxic cells. Log-phase cells show the highest sensitivity to the toxic action of NSC 38087, with plateau-phase cells, cells with a history of chronic hypoxia, and thermotolerant cells showing greater resistance. These toxicity data are compared to those for the 2-nitroimidazole hypoxic-cell sensitizer misonidazole. THE NITROIMIDAZOLES misonidazole (MISO) and metronidazole have been used in clinical trials as potential sensitizers of the radio-resistant hypoxic cells in tumours. In instances where clinical data have been evaluable some improvement in radiation response has been observed with these compounds (Urtason et al., 1976;Thomlinson et al., 1976;Dawes et al., 1978;Ash et al., 1979). However, these drugs show some neurotoxic effects which prevent their use at doses sufficient to give optimal sensitization (Dische et al., 1977;Urtason et al., 1977). This has led to a search for sensitizers more effective than MISO or metronidazole. Interest has centred on modifying the nitroimidazole compounds to produce sensitizers more efficient than MISO but with similar toxicity (Adams et al., 1979a, b;Adams et al., 1980a) or alternatively to produce a drug as good a sensitizer as MISO but with lower toxicity (Brown & Workman, 1980;Brown & Lee, 1980). Using mammalian cells cultured in vitro, it has been shown that electron affinity is the predominant factor which determines both the cytotoxicity and sensitizing effectiveness of hypoxiamediated drugs. There are, however, a number of exceptions to these general relationships which provide grounds for optimism that other parameters may be identified that can affect sensitization by and toxicity of nitroimidazoles. The compound 5-chloro-1 -methyl-4-nitroimidazole has previously been shown to sensitize at concentrations about 100-fold lower than would have been predicted on the basis of electron affinity (Watts & Jacobs, 1978). In a subsequent study this anomalously high sensitizing efficiency was demonstrated for a range of 4-nitroimidazoles substituted in the 5-position with a sulphonate or sulphonamide group (Adams et al., 1980a. Among these compounds, NSC 38087, with the structure shown in Fig. 1, which has an electron affinity only slightly greater than that of MISO, shows a sensitizing efficiency similar to that of 02. In this paper we examine  . All these cell populations had plating efficiencies 9000 before cytotoxic treatment. Cytotoxicity experimen.ts.-250ml spinner flasks were fitted with a gas inlet/outlet system and a sidearm through which samples could be withdrawn. Cells at a concentration of 2 x 105 cells/ml were suspended in MEM+ 7.50o FCS and held in a water bath at 37°C. The compounds were dissolved in MEM + 7-50o FCS and added to the suspension, which was buffered with bicarbonate to pH 7-4. When appropriate, the spinner containing cells was de-aerated by flowing N2 plus 50o CO2 (< 10 pt/106 02; BOC Ltd.) at 500 ml per min over the surface of the stirred suspension throughout the experiment. Samples of cells were withdrawn at appropriate times, centrifuged, resuspended, counted and diluted. For each test condition various numbers of cells were plated, in triplicate, in MEM+ 150% FCS and incubated for 7-10 days at 37°C before scoring for colony formation. Further details of the technique are described elsewhere (Stratford & Adams, 1977).
Radiation experiments-.Cells were harvested from exponential-phase cultures, diluted appropriately and added to glass Petri dishes containing 2-5 ml MEM, supplemented with 15% FCS. Cells were allowed to attach at 37°C for 1 -1 h before the medium was removed and replaced with fresh medium containing drug. Irradiations with 60Co y-rays were carried out in "dural" containers which can hold 4 Petri dishes (Cooke et al., 1976). These vessels could be rendered hypoxic by purging with N2 + 5% CO2 (BOC Ltd.) for 1 h, when the vessels were sealed and irradiated at room temperature at a dose rate of 4-2 Gy/min. For irradiations in air, cells were equilibrated with air+500 CO2. After irradiation, the medium was removed and replaced with fresh MEM+ 150% FCS and the cells incubated for 7-9 days before scoring for colony formation.
Compounds.-NSC 38087 was provided by the Drug Synghesis and Development Branch of NCI, U.S.A. NSC 326151 was synthesized in our laboratories. compound is considerably more efficient as a sensitizer than MISO. Not only does NSC 38087 show very high sensitizing efficiency, but it also shows considerable cytotoxicity towards V79 cells held under aerobic conditions at 37TC. Fig. 3 illustrates the toxicity of varying concentrations of NSC 38087 plotted as a function of time of exposure to drug. Data in this and other figures (unless indicated otherwise) come from individual experiments; replicate experiments were always done and provided similar results. It is apparent that the cytotoxicity of NSC 38087 is both concentration and time-dependent. However, there is an indication that, depending on the drug concentration, survival reaches a minimum level with longer exposures, producing no further cell killing (see, e.g., data for 3 5,uM NSC 38087). Further experiments showed that cells exposed to 3 5/tM NSC 38087 for up to 23 h gave a similar surviving fraction to cells exposed than hypoxic cells. This contrasts with all previous data for nitro-heterocyclic radiosensitizers, which have shown increased toxicity to hypoxic cells (Sutherland, 1974;Hall & Roizin-Towle, 1975;Mohindra & Rauth, 1976;Moore et al., 1976). Other 5-substituted 4-nitroimidazoles which have sensitizing efficiencies much greater than would be expected from the electron affinities also show increased toxicity towards aerobic cells. Typical of such data are those given in Fig. 6 for 1 -methyl-4nitro -5 -morpholinosulphonylimidazole, NSC 326151. Interestingly, the concentration of NSC 326151 used in the above experiment is 10 times greater than for NSC 38087; yet similar amounts of cell killing are seen in both air and N2. This may be due to the higher electron affinity of NSC 38087 (see Table).  26x10-3 1 9x 10-1 90x 10-4 1-9x 10(-0 35 -406 1 5xIt-2 62xx10-1 2 0xJ10-3 66x 10-1 0-13 * For comparison, values of Cc/C1.6 for AIPSO range from 0 3 (Adams et al., 1979b) to 1 0 (WNatts et al., 1980). t E71 is the one-electron redluction potential measured at pH 7 0; C1.6, C1-6 arc concentrations required to gixve a radiosensitizer enhancement ratio of 1-6; Cc and Cc are concentrationis requiredl to reduce plating efficiency by 50% on incubation of cells with drug for 7-10 days. Data for values of E71 and C1.6 are from Adams et al. (1980aAdams et al. ( , 1981. C1.6 and Cc are values of sensitizing efficiency andl chronic aerobie toxicity pre(lictedc from previous electron affinity correlations (Adams et al., 1979a, b). carried out on cells harvested from exponential-phase cultures. In order to determine the likely spectrum of effectiveness of these drugs, cytotoxicity experiments were done on V79 cells with different culture histories. Fig. 7 shows the effect of 5jM NSC 38087 on exponential, plateau and thermotolerant cells, and in addition on cells with a history of chronic hypoxia. All treatments were in air at 37TC. Clearly, the different cell populations respond to NSC 38087 in widely differing ways, with exponential cells showing the greatest, and thermotolerant cells the least sensitivity.

DISCUSSION
NSC 38087 has the typical properties of a series of 5-sulphonyl derivatives of 1-methyl-4-nitroimidazole. These compounds show much greater sensitizing efficiency (Adams et al., 1980a and greater toxicity than would be suggested from their electron affinity. Some sensitization and toxicity data for NSC 38087 and NSC 326151, together with their oneelectron reduction potentials (E17) are recorded in the Table. Previously it has been shown for uncharged 5-nitrofurans, 2-nitro-and 5-nitroimidazoles, nitrobenzenes, quinones and some unsubstituted 4-nitroimidazoles that sensitizing efficiency and aerobic toxicity correlate extremely well with the electron affinity of the compounds (Adams et al., 1976a(Adams et al., , b, 1979a. Sensitizing efficiencies were defined as the concentration of drug required to achieve a given level of enhancement, usually 1-6 (C1.6). These values for the compounds used in the present work are given in the table, together with those values (n1. 6) which would have been predicted from previous electron-affinity correlations. There are orders-of-magnitude differences between the observed and predicted sensitizing efficiency. Values for the chronic aerobic toxicity of NSC 38087 and NSC 326151 are also given. These values are defined as the concentration of drug required to reduce plating efficiency to 50% after 7 days' exposure. As for sensitization, these compounds show much greater toxicity than would be suggested from their electron affinity. Nevertheless, within the series of 5-substituted 4-nitroimidazoles, the most electron-affinic compounds still appear the most efficient sensitizers and also the most toxic.
An interesting feature of the cytotoxic properties of these compounds is their greater toxicity to aerobic than to hypoxic cells. Previously it has been shown that many other nitro-aromatic compounds are preferentiallv toxic to hypoxic cells (Sutherland, 1974;Hall & Roizin-Towle, 1975;Mohindra & Rauth, 1976;Moore et al., 1976). This has been attributed to metabolic reduction of the nitro group (Varghese et al., 1976). Formation of the nitro-radical anion (RNO-) was shown by ESR to be the first step in the biological reduction of nitro-aromatic compounds (Mason & Holtzman, 1975a). In the presence of 02, reduction of RNO2 is inhibited by electron transfer from RNO2 to 02 (Mason & Holtzman, 1975b;Wardman & Clarke, 1976). Recently it was proposed that 5-chloro-1 -methyl-4-nitroimidazole and related 5-substituted 4-nitroimidazoles, such as those reported here, would be more toxic towards hypoxic cells than would be expected on an electron-affinitv basis (Clarke & Wardman, 1980). These authors proposed that this could be due to initial formation of RNO2 followed by dissociative electron attachment with the resulting formation of a free radical or other species, which would be toxic.
This scheme would be inhibited in the presence of 02 (Clarke & Wardman, 1980). Therefore, it is now apparent that the greater toxicity shown by NSC 38087 and 326151 to aerobic cells is Inot wholly consistent with their proposal. Nevertheless, it is possible that the results reported here would occur if reductive metabolism of the nitro group proceeded via an 02insensitive mechanism, but this would be unique to this series of compounds in this cell line.
One reason for the observed difference in toxicity to hypoxic and aerobic cells could be that under hypoxic conditions metabolism of the drug is more rapid, so that the form of the drtug that is toxic is removed from the cellular environment.
The shapes of some of the hypoxic survival curves lend support to this argument. However, we have shown, using HPLC analysis, that the concentration of free NSC 326151 in the cell suspensions does not vary during the course of these experiments either in air or N2 (Gibson & Stratford, unpublished). Therefore other reason(s) may have to be sought to explain the shapes of the survival curves.
At present there is no explanation for the differential toxicity of these compounds to aerobic cells, nor for their greater toxicity (in both air and N2) than would be predicted from their electron affinities (Adams et al., 1979b(Adams et al., , 1980b. However, it is known that related compounds such as Imuran (azathioprine, 6-(1 -methyl-4-nitro-5-imidazolyl) thiopurine) are cleaved rapidly in vivo to 6mercaptopurine and 1-methyl-4-nitro-5thiolimidazole (Bresnick, 1959;Elion, 1967). This is thought to be due to reaction of free sulphydryl groups at the 5 position of the imidazole ring. We are currently carrying out studies to determine whether toxicity in vitro, in air and/or N2, is a consequence of such a reaction.
Increases in temperature enhance the toxicity of nitro-aromatic compounds to hypoxic and aerobic cells (Stratford et al., 1978). Potentiation of toxicity by 41°C hyperthermia is also seen for NSC 38087. However, any other similarity with the toxicity of previously studied nitro compounds (e.g. MISO, metronidazole, nitrofurazone) end there. The shape of the hypoxic survival curve for MISO shows a shoulder region after which survival decreases exponentially with time; this clearly differs from many of the curves shown here. The toxicity of MISO is similar in exponential phase, plateau phase and in cells which have been cultured at 41°C. Further, cells with a history of chronic hypoxia show a slightly greater sensitivity to MISO than to exponentialphase cells (Williamson & Smith, unpublished). This is not the trend in the present work with NSC 38087, where cells can be ranked in their sensitivity to this drug as exponential phase > plateau phase > chronically hypoxic > cells held previously at 41°C. The change in sensitivity is unlikely to be cell-cycle-associated, since cells held at 41°C continue to progress through the cell cycle. Interestingly, the above order of effectiveness for the toxicity of NSC 38087 is similar to that seen for heat killing . Changes in heat sensitivity may be due to membrane alteration. Therefore we could speculate that the changes in sensitivity to NSC 38087 may be membrane-associated, which could subsequently effect drug penetration and/or metabolism. This is currently being studied.
We have shown that NSC 38087 and 326151 are highly cytotoxic members of a series of 5-substituted 4-nitroimidazoles, and also show very high radiosensitizing efficiency in vitro. However, as can be seen from the table, the value C,/Cl.6 is no better than for MISO. The in vitro data alone would suggest little therapeutic advantage over MISO by using these compounds. However, the value of these in vitro studies is to give a guide to the activity of a drug as a sensitizer or cytotoxic agent. Current in vivo work is exploiting metabolic and pharmacokinetic differences within this series of compounds to take full advantage of the cytotoxic and radiosensitizing properties of these drugs.