Some reactions and properties of nitro radical-anions important in biology and medicine.

Nitroaromatic compounds, ArNO2 have widespread actual or potential use in medicine and cancer therapy. There is direct proof that free-radical metabolites are involved in many applications, and an appreciation of the conceptual basis for their therapeutic differential; however, an understanding of the detailed mechanisms involved is lacking. Redox properties control most biological responses of nitro compounds, and the characteristics of the one-electron couple: ArNO2/ArNO2- are detailed. The "futile metabolism" of nitroaryl compounds characteristic of most aerobic nitroreductase systems reflects competition between natural radical-decay pathways and a one-electron transfer reaction to yield superoxide ion, O2-. Prototropic properties control the rate of radical decay, and redox properties control the rate of electron transfer to O2 or other acceptors. There are clear parallels in the chemistry of ArNO2- and O2-. While nitro radicals have frequently been invoked as damaging species, they are very unreactive (except as simple reductants). It seems likely that reductive metabolism of nitroaryl compounds, although generally involving nitro radical-anions as obligate intermediates (and this is required for therapeutic selectivity towards anaerobes), results in biological damage via reductive metabolites of higher reduction order than the one-electron product.


Introduction
Although the other papers in this issue attest to the widespread interest in free-radical internediates in the action of several classes of medically important compounds, it is arguable that nitrocompounds are the one class of drug in which direct proof of radical production in intact target organisms has been demonstrated (1)(2)(3) and in which the free-radical reaction almost certainly responsible for the therapeutic selectivity has been observed directly (4). The combination ofidentification and measurement of steady-state concentrations of radicals by electron spin resonance (ESR) and spectrophotometric monitoring of reaction kinetics following radical generation by pulse radiolysis (5) has provided considerable insight into the most important reactions and properties of nitro radical-anions important in biology and medicine.
Although there may be little or no net nitroreduction in aerobic systems because of reaction (2), there may be a stimulation of respiration, a feature which Biaglow, Durand, Sutherland, et al. (25)(26)(27)(28)(29) recognized may be important in the potentially widespread use of nitroaryl compounds in cancer therapy (31)(32)(33)(34). The oxygen tension in tumor cells may control the radiotherapeutic response, and nitro compounds (and other oxidants) are able to sensitize hypoxic cells to radiation, with a negligible effect on the radiosensitivity of well-oxygenated tissues. In addition to this application, there is considerable interest in combination chemotherapy involving nitro compounds, or "chemosensitization" (33,34).
There is currently much interest (35) in the enzyme, superoxide dismutase and the possible toxic properties of 02, or, more plausibly, subsequent products obtained via metal-catalyzed Fenton-type chemistry (36) and in the invoking of "redox cycling" (37) [reactions (1) and (2)] in the analogous chemistry with anthracyclines and other quinones. Since the flux of O2is stimulated by most nitro compounds via reaction (2), one might have expected the biological properties of nitro compounds to reflect possible "superoxide" toxicity. However, most nitro compounds are less toxic towards mammalian cells in air than in hypoxia (38)(39)(40), which must reflect the much more damaging, competing reactions such as (3) and subsequent reductive pathways. The details of the mechanism oftoxicity ofantiparasitic nitro compounds is the subject of an article by Moreno and Docampo in this issue (41), and the present paper is restricted to the more basic chemical properties of nitro radical-anions which control the rate of reactions (1)-(3) and hence their biological properties.
Nitroimidazoles are the class ofnitro compounds most widely used in medicine, and the review ofJosephy and Mason (42) covers the literature on the reduction products of nitroimidazoles to 1982. More recent work has provided further information, e.g., on the instability of reduction intermediates (43,44) and the reaction of reductive fragments with glutathione (45), nucleosides (46), and protein (47,48). Although there is much emphasis on higher reduction, the nitro radical-anion is thought to be the obligate intermediate involved in nitroreduction by most (but not all) organisms, and its radical chemistry is therefore central to the use of nitro compounds in medicine.

Redox Properties of Nitroaromatic Compounds
Most biological properties of nitroaryl compounds reflect the ease of nitroreduction in a remarkably similar way (28,49,50), although the apparent simplicity of most redox dependences is unfortunately not extended to the organisms of most interest in medicine (51)(52)(53). Hence the thermodynamic parameter characterizing the relative ease of reduction is of major importance in defining the likely biological properties of different nitro compounds; in fact, the parameter also controls the rate of reaction (2), and other reactions with electron acceptors, as well as (1). Reaction (1) involves the one-electron couple: ArNO2/ArNO2L, and it is the reduction potential E of this couple in water at physiological pH which is the most appropriate index of the redox properties of nitroaromatic compounds in the present context.
Since ArNO2is unstable in aqueous solution at pH -7 (see below), conventional electrochemical measurements such as polarographic half-wave potentials, E1/2 cannot be equated with potentials for the one-electron couple, E(ArNO2/ArNO21). However, electrochemical measurements of E½ using, e.g., cyclic voltametry in aprotic solvents (54) or polarography in water (55) generally parallel the thermodynamically reversible oneelectron potentials in water at pH 7 such that the values are numerically similar when E is expressed on the hydrogen scale (NHE) and E½ on the calomel reference (SCE). Thus, generally, E (E½ -0.24 V) to a fair approximation. The higher the value (more positive), the more electron-affinic (more powerful oxidant) the nitro compound.
The most powerful and reliable method to determine E(ArNO2/ArNO2-) is from pulse radiolysis measurements of the equilibrium constant for one-electron transfer equilibria involving ArNO2 and a reference compound of known reduction potential such as a quinone or bipyridinium compound (56,57): (7) E(ArNO2/MANO2 ) -E(Q/Q ) -0.059 log K5 (8) if E is in volts. The yields and reactions of the species produced upon radiolysis of aqueous solutions are so well established that the design of such experiments is a matter of routine (5). The radicals: ArNO21, Q are usually generated within a microsecond or so of the end of a radiolysis pulse of equally short duration, and the equilibrium (5) attained and the equilibrium constant K5 measured spectrophotometrically within (typically) 10-200 ,usec, i.e., before the unstable radicals can decay via routes such as (3) (5,56,57).
A scale of reduction potential spanning the range appropriate for virtually all nitroaryl compounds of medical or biological interest is shown in Figure 1. The potentials of three common nitroheterocycic pharmaceuticals are seen to be significantly lower than that of oxygen. [Note: E(02/02-) is correctly given as -0.33 V vs. NHE at pH 7, since the thermodynamic standard state for oxygen is unit fugacity, i.e., 1 atmosphere pressure. Use ofthe Nernst relationship and converting to a nonstandard state of 1 mole/dn3 O2/O2-(the same standard state as ArNO2/ArNO2t) gives an effective E(02/02-) of ca. -0.15 V, a value more appropriate for direct comparison with E(ArNO2/ArNO2 ] (58).
Also shown in Figure 1 are the potentials of compounds which illustrate the effects of additional, electron-withdrawing substituents in (in this case) the 2nitroimidazole ring system. Such effects can be readily  predicted once the potentials of two or three compounds in a given series have been measured, using predictive relationships based upon Hammett substituent constants (59). Thus for 5(R)-1-methyl-2-nitroimidazoles: ElV -0.406 + 0.146 (R) and for 4(R) -nitrobenzenes: ElV -0.484 + 0.168 (R) (9) (10) in water at pH 7, -298°K. Figure 2 illustrates the variation of estimated values of E(ArNO2/ArNO2-) (under physiological conditions) for some of the more common nitroaryl ring systems, and Table 1 lists values of E for the compounds of most interest in biology, medicine, and cancer therapy. [The author is preparing a more complete compilation of reduction potentials of couples involving free radicals in aqueous solutions, to be published in the U.S. National Standard Reference Data Series, which will give full details and a bibliography; most of the values shown have been published in a variety of references (4,20,30,49,53,56,57).] The position of the important one-electron transfer equilibrium (2): can be simply calculated from these values of E and the relationship: . Estimated values of reduction potential, E(ArNO2/ ArNO2) vs. NHE in water at pH 7 of some typical nitroaryl systems. The examples are based upon measurements of the compounds or simple derivatives where R = alkyl or hydroxyalkyl, etc., and the values may be significantly different when additional ring substituents are present (see Fig. 1).
Estimates of K2 are included in Table 1.
The positions of other electron-transfer equilibria of interest may be calculated similarly, replacing Q/Qby the appropriate couple in eq. (5) and using the relationship (8). Thus ascorbate, AHhas frequently been of interest as a potential electron donor to nitroaryl compounds, e.g., with 4-nitroquinoline-N-oxide (27,(60)(61)(62)(63): ArNO2 + AH-± ArNO-+ AH (A + H+) (12) The reduction potential of the ascorbyl radical at pH 7, E(AH/AH-) can be estimated reliably at 0.30 V vs. NHE from either measurements of one-electron transfer equilibria at pH 13.5 (64) or calculations (65) based upon the semiquinone formation constant from ESR measurements (66). Hence: (13) and K12 is estimated to be 9 x 10-9 for 4-nitroquinoline-N-oxide, 3 x 10-10 for typical 5-nitrofurans, 2 x 10-12 for simple 2-nitroimidazoles and 5 x 10-14 for metronidazole and analogs. These equilibria are, of course, pHdependent since E(AH/AH -) is decreased, e.g., to 0.015 V at pH 13.5 (64), at which pH K12 is increased to ca. 1 X 10-7 for simple 2-nitroimidazoles. In spite of equilibrium (12) being thermodynamically so unfavorable (K12 < 10-8 even for the most electronaffinic nitro compound), abscorbate can still be a potential one-electron reductant because the products of reaction (12) are unstable and are being continuously removed by, e.g., disproportionation or reaction (3). A further reaction may also be considered. The one-electron reduction potentials for addition to a second electron to nitroaryl compounds [the reduction potential of the nitro radical-anion, E(ArNO2"/ArNO22-) or E(ArNO2"/ArNO)] are unknown for aqueous solutions at pH 7, although there is some evidence (see below) that ArNO21 is a poorer oxidant than O2at physiological pH; apparently ArNO21 is unable to oxidize the Cu(I) form of Cu-Zn superoxide dismutase, whereas O2is able to do so (67). However, it is not inconceivable that oxidation of AHby ArNO2' may play a significant role; the reaction would be expected to be at least as complex as the analogous oxidation by O2- (68). Indirect evidence for nitro radical formation from a nitrofuran with ascorbate (pH 7) as electron donor has been obtained (63), making use of a diagnostic cis-trans chain isomerization reaction (69,70) which yields up to -200 molecules isomerized per radical-anion produced (71) (see below): cis -AF -2 + "nitroreductases" --trans -AF -2 (14) Similar considerations apply to the thermodynamics of formation of nitro radical-anions from other reductants, e.g. reduced flavin, FMNH2: ArNO2 + FMNH2 a± ArNO; + FMNH (+H+).
These calculations are valid only for pH 7, since the potentials of both couples may vary with pH. In the case of the couple, ArNO2 , the potential Ei at any pH = i may be calculated from: (17) provided there are no additional substituents with prototropic properties. Since pK4 < 7 for all known, simple nitroaryl compounds (see below), this pH dependence of E(ArNO2/ArNO2-) is frequently unimportant in respect of physiological conditions. However, acidic or basic substituents complicate the issue considerably (24,56,57,59).
A note of caution is also appropriate concerning the rates of reactions which may have readily calculable thernodynamic parameters but which are catalyzed either by enzymes in vitro or even by simple, free metal ions. The reduction of ArNO2 by flavins or thiols are catalyzed by trace quantities of Fe(II) (63,(73)(74)(75)(76). Oneelectron reduction by free thiols is thermodynamically much less favorable than by ascorbate, since E(RS/ RS-) must be much more positive than E(A/AH-) at pH-7, indeed higher than E(PZ+)/PZ) 0.8-0.9 V where PZ = common phenothiazines (77).

Prototropic Properties of Nitro Radical-Anions
In addition to being ofpotential importance in defining the pH-dependence of redox properties, prototropic equilibria control the natural lifetimes of nitroaryl radical-anions in aqueous solution. Equilibrium (4) is written as a dissociation of an oxygen acid, but protonation at sites other than oxygen may be important, so that Eq. (18) may be considered a general representation, of which Eq. (4) is a specific example.
Thus even simple nitroimidazole radicals have two sites for protonation in the readily-accessible range of pH: N02 oxygen and ring nitrogen. When ring substituents carry groups with prototropic properties, e.g., nitrogenous bases or carboxylic acids, the pKa for dissociation of these additional proton sites may be significantly different in the ground state and radical even when, e.g., the side-chain nitrogen is "insulated" from the nitroaryl ring by a 2-or 3-carbon aliphatic chain. A typical example in the important 2-nitro-1-imidazolylalkylamine series has been discussed (59). More dramatic shifts in pKa between ground state and one-electron adduct are found with the protonation of the unsubstituted (N-3) imidazolyl nitrogen in some simple N-1 alkyl/alkanol substituted 2-, 4-and 5-nitroimidazoles, where the increase in pKa upon electron addition is around 6.2, 5.5, and 3.6 units, respectively (24,81). ESR studies have characterized the dissociation of acidic ring N-H protons in other nitroaryl radical-anions lacking carbon substitution at nitrogen (80,82).

Natural Lifetimes of Nitro-Radical Anions
The normal decay pathway of most nitro radical-anions at pH -7 in water is that of second-order disproportionation: Figure 3 illustrates the typical, pH-dependent second-order rate constant for the decay of metronidazole radicals measured by pulse radiolysis. More extensive studies (21) using pH 7.4, isotonic ionic strength, -2980K, have proven the radicals decay by accurate second-order kinetics with 2kobs = 4.2 x 104 dm3/mole-sec under these conditions. The values for radicals from other 5-nitroimidazoles such as ornidazole and nimorazole were within a factor oftwo ofthat for metronidazole (21), and independent ESR observations of the disproportionate rate of 4-nitrobenzoate radicals at RH 7.4 gave a value of 2kob8 [eq. (25)] = 8.5 x 103 dmi/molesec (9,84)]. More extensive, pulse radiolysis measurements ofsubstituted nitrobenzene radicals (20) provided estimates of 2kobh at pH 7 in the range 7 x 104 to 3.3 x 108 dm3/mole-sec, although most were < 107 dm3/ mole-sec (values at pH 7.4 would be expected to be 2.5 times lower than those at pH 7). The decay kinetics of some other nitroimidazole radicals (e.f., 5-chloro-1methyl-4-nitroimidazole, 2kob8 = 1 x 10 dm3/mole-sec at pH 7) (81), but not 2-nitroimidazole radicals (21)   other observations (see below). The dependence of the steady-state concentration of radicals on the square root of the (reductase) protein concentration (3,9,10,13,16) indicates the general second-order decay pathway, Eq. (20), is common to nitrobenzenes, -furans, and -imidazoles (except 2-nitroimidazoles). The pH-dependent lifetimes shown in Figure 3 raise an interesting point. That the transition in the value for 2kob8 for 4-nitroacetophenone radicals occurs at a pH significantly higher than pK4 (= 2.7 (77)) reflects the relative values: k?2 > k2l, a similar situation to that seen with HOW/O2 (83).

Electron-Transfer Reactions of Nitro Radical-Anions
The values of the couple: E(ArNO2/ArNO2!) may be used to assess not only the thermodynamic feasibility of one-electron reduction of ArNO2 by any potential "nitroreductase" (flavin, ascorbate, Fe/S protein, etc.) but also the likelihood of electron donation from ArNO2to potential acceptors. In the present context the most important of these is obviously oxygen [reaction, (2)]. Equation (11) defines the position of equilibrium but says nothing about the kinetics of reaction (2) in particular: ArNO`+ 02 --ArNO2 + 02 be measured by ESR and are typically of the order of 1 ,umole/dm3 in many experiments (7,9). The concentrations in intact, target organisms depend on the substrate (2) and likely in vivo values are difficult to predict. However, typical natural lifetimes [with respect to decay by reaction (20) only] of nitro radicals under physiological conditions could well be as long as 20 sec [from Eq. (26), using a steady-state concentration of 1 ,umole/ di3 and 2kob8 = 5 x 104 dm3/mole-sec].
Lifetimes of nitroimidazole radicals are many seconds at high pH (4,21,24,81), similar to the behavior of radicals from nitrobenzene and derivatives (18,19,85,86). A report (23) that the second-order rate constant for the decay of the 5-nitrofuroic acid radical anion remained essentially unchanged (2k = 2.2 x 109 dm3/ mole-sec) from pH 10.5 to 3.3 is completely inconsistent in both magnitude and pH-independence with numerous with Ea all in the range 30-39 kJ/mole; the algorithms for the temperature dependences of k2 for metronidazole and misonidazole have been reported (4,21).
There seems no question that the product of reaction (2) is indeed 02with restoration ofArNO2. Preliminary spectral evidence that O2is a product (4) is substantiated by the spectra shown in Figure 4, which are the results of repeating the earlier work (4) with further precautions taken to eliminate artefacts from scattered light (or at least, to ensure identical artefacts in the control spectrum of 021). Pulse radiolysis measurements always measure the change in absorbance introduced by the conversion of ground state to radical; that the final spectrum in Figure 4 agrees with that of 02requires restoration of the nitro radical to ground state, (2)  which absorbs strongly in this spectral region. ESR spin-trapping of O2- (17) provides further evidence that O2is a product. Since an earlier report (87) indicated a rate constant for reaction of the nifuroxime radical-anion around 4 orders ofmagnitude higher than our value (4) (and about 50-fold higher than that for any other nitro compound so far reported, let alone for a compound with high electron affinity, i.e., with relatively low "driving energy"), it seems appropriate to present some raw data to justify our claim. Figure 5 shows the absorption changes produced upon generating the nitro radical in N2,-, airor 02-saturated solutions of formate (to scavenge H and OH) (5); tert-butanol, an alternative OH scavenger was used by Greenstock and Dunlop (87). There are rapid spectral changes in the tert-butanol system, presumably resulting from reaction of tert-butanol radicals with nifuroxime, or radical-radical reactions; the natural lifetimes of simple nitro radicals are without exception much longer than implied by the data of Greenstock and Dunlop (87). Although the first-order dependence of radical decay on oxygen concentration reported by Greenstock and Dunlop is impressive, it appears that the tert-butanol system is not satisfactory for studying reaction (2) in this instance.
Since the reduction potential, E of even the most electron-affinic nitroaryl compound so far reported (4- nitroquinoline-N-oxide) is -0.18 V, it is not surprising that electron transfer from nitro radicals to more powerful oxidants such as Fe(III) or Cu(II) is at rates approaching the diffusion-controlled limit. Thus for ArNO2 = misonidazole or 5-chloro-1-methyl-4-nitroimidazole: has k29 = 2.3 x 108 and 3.6 x 108 dm3/mole-sec, respectively at zero ionic strength and room temperature (24,81) (both values would be ca. 1 x 109 dm3/mole-sec in isotonic saline). Electron transfer from ArNO21 to Fe(III)-cytochrome c has rate constants of the same order (88).
Reaction (32) is the basis for the facile, chain cis/trans isomerization of the (5-nitro-2-furyl)acrylamide, AF-2, known to occur via nitro radical anion intermediates (11,13,63,(69)(70)(71). The chain reaction arises because the reduction potential of the trans/trans couple is -34 mV lower than that of the cis/cis" couple; the chain is propagated via: trans + cis --trans + cis- (35) competing efficiently with: cs-transm since ks5 = 2 x 106 dm3/molesec and k36 5 -40/ sec (71). When AF-2 is used as an indicator of "nitroreductase" activity with other nitroaromatic compounds also present, competing for the (enzyme) electron donor (90), the electron exchange reaction (35) can occur in competition with Eqs. (36) and (2), and the redox-related competition is a reflection of all of these reactions and not just the relative efficiency of electron donation from the enzyme (71).  (38) rate of decay where [ArNO2'] is the value at a steady state. If this is of the order of micromolar, (see above) then Eq. (38) (39) and even submicromolar levels of 02 (continually replenished) will be sufficient to inhibit nitroreduction very efficiently if other decay pathways are not available.
Just such a simple situation was modeled by Rauth et al. (95), producing metronidazole radical-anions radiolytically at a zero-order rate of ca. 10 nmole/dm3-sec in the presence of low concentrations of 02. They found that e.g., 100 ppm 02 (gas phase; ca. 120 nmole/dm3 02 in solution) was sufficient to inhibit nitroreduction al-most completely. Using the steady-state approximation: rate of production ofArNO2-= rate ofdecay, it is easily calcuated that the steady-state concentration of nitro radical-anion is ca. 11 nmole/dm3 under these conditions. Then, from Eq. (38), a ratio: rate of restitution/rate of decay 2000 is expected.
In the same study, it was found that 02 inhibited nitroreduction of four typical 2-nitroimidazoles very much less efficiently than the effect on the reduction of metronidazole or 4-nitroacetophenone. Thus ca. 2-7 ,umole/dm3 02 was required to inhibit effectively the reduction of the 2-nitroimidazoles. The authors pointed out (95) that the differences could arise if there was a first-order pathway for "natural" decay, as indeed had been observed experimentally (21). If the competing reactions are those of Eqs. (2) and (27) with (e.g., misonidazole) k2 = 4.2 x 106 dm3/mole-sec and k23 = 10/ sec, then an 02 concentration of ca. 20 ,umole/dm would be required for the rate of Eq.
(2) to be ten times faster than the rate of Eq. (27).

Some Other Reactions of Nitro Radicals
The pH-dependent decay of ArNO2' via reactions (21) and (22)  [A2-j) < 1. Thus O2is a more powerful oxidant than 02; however, as discussed above, ArNO2seems to be a much weaker oxidant than 02since a typical example cannot reoxidize Cu(I).
The reaction of 02-with thiols, RSH has been the subject of several studies (96,97), with the more direct measurements (97) being (in the author's view) the most reliable, especially since the reaction is probably thermodynamically unfavorable (see below). Thus: L-Cysteine + 0-°2products (40) was estimated to have k4o < 15 ± 2 dm3/mole-sec at pH 10.9 (97). One would expect the reduction by thiolate anions of the weaker oxidant, ArNO2' to be much slower than this latter value. There is some evidence (50,98) that the reduction potential, E(RS/RS-) does not differ by more than 0.10-0.15 V for the common thiols, so a study with one thiol should be reasonably predictive of the behavior of another.
In spite of this background, and in spite of the total lack of experimental demonstration, several authors have postulated that the protective role ofthiols in, e.g., the cytotoxicity of nitroaryl compounds, arises from re-action of ArNO2with RSH/RS-, presumably the reaction: ArNO2 + RS-+ 2 H+ -. ArNO +RS7 + H20 (41) This reaction seems exceedingly unlikely thermodynamically. Since thiyl radicals, RS oxidize phenothiazines with k > 3 x 107 dm3/mole-sec (77), E(RS/RS-) is probably > 1.1 V at pH= 3 and at least 0.9 V atypH 7, (probably significantly higher). Since E(02-/02 ) = 0.865 V at pH 7 (58), reaction (40) is probably thermodynamically unfavorable. The couple ArNO21/ArNO will only be reversible at high pH, but the difficulty in oxidizing Cu(I) points to its potential being at least several tenths of a volt lower than that of RS/RS-. Reaction (41) thus seems most unfavorable, unless irreversibility facilitates the forward reaction. This analysis is entirely consistent with two independent, experimental studies. Polnaszek et al. (99) found that 0.1 mole/ di3 glutathione (GSH) had no effect on the steady-state concentration of ArNO2" produced via rat hepatic microsomal or xanthine oxidase reducing systems, using three different 5-nitrofurans. The lifetime of misonidazole or metronidazole radical-anions was not detectably changed by the presence of GSH (2 mmole/dm3) at pH 7.3 or 9.4 [the higher pH should favor reaction (41)]; thus k41 < 5 dm3/mole-sec at pH 9.4 (21) and is very probably < 0.05 dm3/mole-sec at pH 7.4. The well-characterized reaction of nitrosoaryl compounds with GSH (100) is, therefore, probably >6 orders of magnitude faster than Eq. (41) and the most likely explanation of the biological role of GSH in nitroaryl cytotoxicity, etc. However, nitroaryl compounds, when coreduced with DNA, do cause extensive damage to the macromolecule (101) in spite of a lack of effect of DNA, RNA, ribose, nucleotides or protein on the steady-state concentration of ArNO21 in reductase systems (99). We have failed to demonstrate any oxidizing properties of ArNO2with some of the most favorable possible reductants, e.g., N,N,N',N'-tetramethyl-p-phenylenediamine (TMPD) (50): ArNOr + TMPD + 2 H+ 4 ArNO + TMPD+ + H20 (42) Oxidation of e.g., guanine sites, G3 would be much less favorable than Eq. (42) since E(G+/G) >> E(TMPD+/ TMPD). We have speculated that nitroso radicals could be more powerful oxidants than ArNO21 (50), since nitrosobenzene is a more powerful oxidant than nitrobenzene (18): ArNO + ArNO -ArNO2 + ArNO, (43) and indeed than oxygen (30): 02 + ArNO -02 +ArNO (44) with kv = 4.1 x i07 dm3/mole-sec (18)  ArNO21 is, of course, protonated at pH 7 and this recalls a note of caution (50) concerming the potential reactivity of ArNO21. Since HOi is several orders of magnitude more reactive than 02 in some circumstances (102), it is possible that the protonated conjugate of ArNO2 may play a role in its biological activity.

Conclusions
Although a detailed understanding ofthe mechanisms of cytotoxicity of nitroaryl compounds in both procaryotic and eucaryotic cells still eludes us, there is no question that the use of these compounds in medicine and cancer therapy relies upon free-radical mechanisms. The redox properties ofthe one-electron couple: ArNO2/ ArNO2" define virtually all the biological properties of these compounds. Disproportionation of the radicals controls their natural lifetime in most model chemical and biological systems (though with important exceptions). In all cases, the rates of these natural radicaldecay processes is a function of pH and the prototropic properties of the radical. Most simple electron-transfer reactions can be rationalized in terms of both equilibrium and kinetics.
However, there are still many, important questions unanswered. There are some parallels in the chemistry of ArNO2" and O2-. The enormous, widespread interest in the biological role of 02is somewhat paradoxial since O2itself is really rather an unrective species. It seems that ArNO21 is even less reactive than O2towards likely biological targets (except of course readily definable electron acceptors). It is hoped that this short article will help clarify the likely role of nitro radicals in biological systems and help point experimentalists towards identifying the critical reactions, which may well involve ArNO2~as an obligate intermediate but probably not as the direct, damaging toxin. This work was supported by the Cancer Research Campaign.