Scavenging of Nitrogen Dioxide, Thiyl, and Sulfonyl Free Radicals by the Nutritional Antioxidant (cid:98) -Carotene*

Mechanisms of free radical scavenging by the nutri- tional antioxidant (cid:98) -carotene have been investigated by pulse radiolysis. Free radicals, which can initiate the chain of lipid peroxidation, including nitrogen dioxide (NO 2 (cid:122) ), thiyl (RS (cid:122) ), and sulfonyl (RSO 2 (cid:122) ) radicals, are rapidly scavenged by (cid:98) -carotene. Absolute rate constant k [NO 2 (cid:122) (cid:49) (cid:98) -carotene] (cid:53) (1.1 (cid:54) 0.1) (cid:51) 10 8 M (cid:50) 1 s (cid:50) 1 and for the glutathione thiyl radical k [GS (cid:122) (cid:49) (cid:98) -carotene] (cid:53) (2.2 (cid:54) 0.1) (cid:51) 10 8 M (cid:50) 1 s (cid:50) 1 have been determined. The mechanisms however are mutually exclusive, the former in- volving electron transfer to generate the radical-cation [ (cid:98) -carotene] . (cid:49) and the latter by radical-addition to generate an adduct-radical [RS (cid:122)(cid:122)(cid:122) (cid:98) -carotene] (cid:122) . Rate constants

␤-Carotene (provitamin A) is a major lipid soluble antioxidant nutrient present in a range of fresh fruit and vegetable produce (1). Epidemiological evidence that correlates ␤-carotene dietary supplementation with a reduced risk of contracting cancer and heart disease (2) has provided the impetus for ongoing human intervention trials (3)(4)(5)(6), particularly on smokers (7,8). The antioxidant properties of ␤-carotene have been implicated in the molecular basis for disease prevention (9,10), primarily because of the putative role of oxidative stress in disease initiation and progression (11)(12)(13)(14). In vitro studies on cellular membrane damage (15,16) and low density lipoprotein (LDL) 1 oxidation (17) indicate that ␤-carotene may modulate free radical processes in vivo by behaving as a chain-breaking antioxidant in lipid peroxidation (9,18) Paradoxically, other studies have shown that ␤-carotene offers little protection from metal-catalyzed LDL oxidation (19,20) and can kill tumor cells via prooxidative pathways (21). ␤-Carotene in combination with ␣-tocopherol has recently been found to reduce radiation effects in normal tissues, suggesting a potential application as a radioprotector in clinical cancer radiotherapy (22).
Although the role of ␤-carotene as a singlet oxygen quencher has been thoroughly characterized (23)(24)(25)(26)(27), surprisingly fewer mechanistic studies on radical-scavenging properties of the carotenoid have been pursued. Pulse radiolysis has shown that halogenated peroxyl radicals (e.g. CCl 3 OO ⅐ ) are rapidly scavenged by ␤-carotene (28 -30) and that the reactivity toward the superoxide radical-anion (O 2 . ) is much less significant than for the related carotenoid lycopene (31). Nitric oxide, a major component of cigarette smoke (ϳ500 -1000 ppm), may contribute to the development of oral/lung cancers and heart disease in smokers by generating more damaging nitrogen oxides, including nitrogen dioxide (NO 2 ⅐ ) radicals (32). In vitro studies have demonstrated that NO 2 ⅐ radicals can initiate the autoxidation of polyunsaturated fatty acids (33) and that cigarette smoke can induce LDL oxidation (34). Paradoxically, although ␤-carotene does not appear to prevent the oxidation of LDL by cigarette smoke (34), it can protect lymphocytes from NO 2 ⅐ radical-induced membrane damage (16). A number of thiols, including the endogenous thiol antioxidant glutathione (GSH), have been shown to enhance nitrogen dioxide-induced lipid peroxidation in model systems by generating thiyl (RS ⅐ ) and thiyl-derived radical species (35). Thiyl radicals are synonymous with the classical repair reactions of thiols in vivo (36,37) and are also generated by peroxidasecatalyzed oxidation of thiols (38,39). Although the fate of RS ⅐ radicals within cells remains a matter for debate (40), there is increasing evidence to suggest that they are capable of initiating the chain of lipid peroxidation at least in model systems (41)(42)(43). Thiyl radicals undergo conjugation with molecular oxygen to generate the thiyl peroxyl (RSOO ⅐ ) radical (44) which can rearrange to thiyl-sulfonyl (RSO 2 ⅐ ) radical (45), another potent initiator of lipid peroxidation (46). A thiol-specific antioxidant enzyme considered widely distributed in mammalian tissues which catalyzes the removal of thiyl radicals has provided evidence for thiyl radical-induced damaging reactions (47). Moreover, human (THP-1) macrophages oxidize LDL by a thiol-dependent mechanism, which is believed to involve thiyl and thiyl-derived radical species.
Following a preliminary communication (48), we now report a detailed pulse radiolysis study on the oxidation of ␤-carotene by NO 2 ⅐ , RS ⅐ , and RSO 2 ⅐ radicals and characterization of the resultant ␤-carotene-derived radical species.
Methods-The pulse radiolysis technique is a useful way of generating specific radical species and studying their reactions with ␤-carotene by kinetic spectrophotometry (49). The pulse radiolysis facility at the Gray Laboratory Cancer Research Trust, including methodology, data acquisition, and analysis, have been described elsewhere (50). In this study 30-ns pulses of 3.5 MeV electrons were used to deliver doses of typically 0.3-2 Gy as determined by thiocyanate dosimetry (51). A tungsten lamp with a photodiode detector allowed the decay kinetics of ␤-carotene radical species to be observed over 0.2 s. Free radical-induced bleaching of the ␤-carotene ground-state absorption was also measured by a xenon lamp utilizing narrow (1.25-mm slits) to eliminate photobleaching.
The nitrogen dioxide radical (NO 2 ⅐ ) was generated by radiolysis of N 2 -saturated tert-butyl alcohol/water mixtures (50%, v/v) containing 0.1 mM NO 3 Ϫ in 1 mM phosphate buffer at pH 5. Under these conditions the radiolysis of tert-butyl alcohol/water mixtures generates the primary radical species (52) in Reaction 1.
Both hydroxy ( ⅐ OH) and hydrogen (H ⅐ ) radicals are scavenged by tertbutyl alcohol to generate a radical that is unreactive toward ␤-carotene and does not absorb above 300 nm.
Thiyl radicals were generated by radiolysis of N 2 O-saturated tert-butyl alcohol/water mixtures (50%, v/v) containing 10 mM thiol and 1 mM phosphate buffer at pH 5. In this case hydrated electron is converted to ⅐ OH radicals that becomes the principal product of water radiolysis in Reaction 5.
Thiyl radicals (RS ⅐ ) are then rapidly formed (Ͻ1 s) after the pulse by the repair of the carbon-centered radicals (Reaction 6) generated in Reactions 1 and 4 by the thiol (54 -56).
Prior to experimentation a N 2 -saturated stock solution of ␤-carotene in toluene was prepared and kept in the dark. Solutions were prepared immediately before experimentation using purified water from a Milli-Q system (Millipore). Solutions required for pulse radiolysis experiments contained no more than 20 mol dm Ϫ3 ␤-carotene to allow adequate light transmission in the 300 -500 nm region, where ␤-carotene exhibits a strong ground-state absorption (59). Before irradiation solutions were bubbled with nitrous oxide (N 2 O, oxygen content Ͻ 10 ppm) or with nitrogen (N 2 , oxygen content Ͻ 10 ppm) from the British Oxygen Co., United Kingdom. All experiments were performed at ambient room temperature (20 Ϯ 2°C).
Bleaching of the ␤-carotene ground-state absorption in the 300 -500 nm has proven a useful method of probing the mechanism of free radical and enzymatic destruction of carotenoid (9,10,60). To assess the possibility of ␤-carotene-derived products interfering at these wavelengths. Steady-state radiolysis experiments were performed in a field of a 2000 Ci of 60 Co ␥-source at a dose rate of 5 Gy min Ϫ1 (total dose ϳ 25 Gy equals Ͻ 30% conversion of ␤-carotene to products) as determined by Fricke dosimetry (61). EDTA 10 mmol dm Ϫ3 was included to prevent metal-catalyzed degradation of either ␤-carotene or ␤-carotene-derived products (60), otherwise the experimental conditions were as for pulse radiolysis. Products formed on irradiation of ␤-carotene were separated by high performance liquid chromatography using a Hypersil 5ODS 125 ϫ 4.6-mm column. Samples were separated isocratically with a methanol/acetonitrile/chloroform (25:60:15%, v/v) eluent (62) and components detected spectrophotometrically at 451 and 300 nm by a Waters 486 variable wavelength detector. Although NO 2 ⅐ , RS ⅐ , RSO 2 ⅐ radicals depleted ␤-carotene in a dose-dependent manner no products of the reactions were detected above 420 nm which would interfere with the pulse radiolysis kinetic measurements.

RESULTS
Oxidation of ␤-Carotene by Nitrogen Dioxide-The absorption spectra obtained by pulse radiolysis of N 2 -saturated tertbutyl alcohol/water mixture (50%, v/v) containing 0.1 mM NO 3 Ϫ , 1 mM phosphate buffer, and 10 M ␤-carotene at pH 5 are shown in Fig. 1. These are attributed to the ␤-carotene radical species generated by the oxidation of ␤-carotene by NO 2 ⅐ radicals. Electron pulses of low dose typically 1 Gy were used to eliminate competition from the fast bimolecular decay of NO 2 ⅐ radicals via Reaction 8 (2k 8 ϭ 4.5 ϫ 10 8 M Ϫ1 s Ϫ1 ) and to ensure that all NO 2 ⅐ radicals generated by Reactions 1-3 reacted solely with ␤-carotene.  The absorption of the solution measured 10 ms after the electron pulse showed a decrease in the wavelength range 300 -600 nm (where ␤-carotene absorbs) accompanied by an increase in the range 600-1000 nm. These spectral changes are ascribed to the reaction of NO 2 ⅐ radicals with ␤-carotene causing a bleaching of the parent chromophore and the formation of a new absorbing species with max Ϸ 910 Ϯ 5 nm. The extinction coefficient of the radical species was calculable from the radical yield of G⑀ 910 ϭ 8.5 Ϯ 0.2 ϫ 10 4 M Ϫ1 cm Ϫ1 . Under the experimental conditions employed, the radiation chemical yield of nitrogen dioxide radicals was G(NO 2 ⅐ ) ϭ 0.9 giving an extinction coefficient for the ␤-carotene radical species of ⑀ 910 ϭ 9.4 Ϯ 0.2 ϫ 10 4 M Ϫ1 cm Ϫ1 . By comparison with an analogous transient generated by flash photolysis (25) and one-electron oxidation of ␤-carotene by the halogenated peroxyl radical CCl 3 OO ⅐ (28, 29), the transient species was identified as the ␤-carotene radical-cation [carotene] . ϩ generated by electron abstraction from ␤-carotene by NO 2 ⅐ radicals via Reaction 9. The extinction coefficient of the [carotene] .
ϩ radical-cation compares well with that measured in aqueous Triton X-100 (⑀ max ϳ 1 ϫ By analogy to the reversible addition reactions of nitrogen dioxide with the double bonds of alkenes to form nitro alkyl radicals (64), the nitrogen dioxide ␤-carotene adduct-radical [NO 2 ⅐⅐⅐␤-carotene] ⅐ was considered a possible precursor to radical-cation formation (16). Indeed, a characteristic shoulder at 650 -800 nm in the [␤-carotene] . ϩ radical-cation absorption spectrum was observed and thought to be a separate radical species quite possibly a [NO 2 ⅐⅐⅐␤-carotene] ⅐ radical. However, the reaction kinetics between 600 and 1000 nm were identical, and it was therefore concluded that the entire absorption in this region was due to the [carotene] . ϩ radical-cation alone. Both the decline in the ␤-carotene ground-state absorption at 451 nm (left inset, Fig. 1) and the increase in the [␤-carotene] . ϩ radical-cation absorption at 910 nm (right inset, Fig. 1) were exponential and first-order in ␤-carotene concentration. The rate constant for the reaction obtained from the slope of the linear plot of the first-order rate constant of the build-up of absorption at 910 nm versus the concentration of ␤-carotene was identical to the corresponding plot for the decline in ␤-carotene absorption at 451 nm (not shown here), both giving k 9 ϭ (1.1 Ϯ 0.1) ϫ 10 8 M Ϫ1 s Ϫ1 . This provided further confirmation that no other radical species are generated under these experimental conditions and that NO 2 ⅐ radicals react with ␤-carotene exclusively by electron abstraction to form the carotenoid radical-cation. If intermediate [NO 2 ⅐⅐⅐␤-carotene] ⅐ radicals are formed they must be highly unstable, rapidly eliminating NO 2 Ϫ to generate the radical-cation in Ͻ10 s.
After 0.1 s the [␤-carotene] . ϩ radical-cation absorption has decayed almost to zero (see in Fig. 1). The absorption at 910 nm decays predominantly by second-order kinetics with a half-life which decreased with increased radiation dose (0.2-2 Gy) as shown in Fig. 2 (i.e. with increasing initial concentration of radicals). This is attributed principally to the bimolecular decay of the radical-cation according to Reaction 10.
The reciprocal of the first half-life of the [␤-carotene] .
ϩ radicalcation varied linearly with the initial radical concentration and from the slope of the fitted straight line the rate constant 2k 10 ϭ 4.1 ϫ 10 5 M Ϫ1 s Ϫ1 was determined. An intercept of 1 ⁄2 Ͻ 6 s Ϫ1 indicated that a small fraction of the [␤-carotene] .
ϩ radicalcation decays by a first-order process, quite possibly solvation or base-catalyzed formation of the carotene hydroxy adductradical, Reactions 11 and 12, respectively.

͓␤-Carotene͔
However, due to the constraints of pH on NO 2 ⅐ radical (see Reactions 2 and 3) and ␤-carotene concentration within this experimental system, this particular aspect of the mechanism could not be pursued. The decay of the radical-cation absorption after 0.1 s paralleled a partial restitution of the ␤-carotene ground-state absorption in the 300 -600 nm region. These spectral changes were consistent with the partial regeneration of ␤-carotene as a consequence of Reaction 10.
␤-Carotene Scavenging of RS ⅐ Radicals-The absorption spectra for the reaction of the ␤-mercaptoethanol thiyl radical with ␤-carotene in Fig. 3 was obtained by pulse radiolysis of an N 2 O-saturated tert-butyl alcohol/water mixture (50%, v/v) containing ␤-mercaptoethanol 10 mM, 10 M ␤-carotene, and 1 mM phosphate buffer at pH 5. In marked contrast to the absorption spectrum obtained from NO 2 ⅐ radical attack on ␤-carotene in Fig. 1, no absorption was observed above 600 nm, indicating a complete absence of the ␤-carotene radical-cation. However, a decline in the ␤-carotene ground-state absorption was observable, indicating that the RS ⅐ radicals are scavenged by ␤-carotene by an alternative mechanism to that observable with nitrogen dioxide.
The decline in absorption at 451 nm is biphasic with a fast step complete in Ϸ50 s and a much slower step complete some 80 ms later. The fast step was exponential and first-order in ␤-carotene concentration, while the slower phase was second- order, independent of ␤-carotene concentration, but dependent on the radiation dose per pulse.
The observations have been attributed to a radical-addition process in which RS ⅐ radicals are scavenged by ␤-carotene to generate a resonance stabilized adduct-radical [RS⅐⅐⅐␤-carotene] ⅐ via Reaction 13. Two isobestic points are located near 380 and 520 nm, respectively, thus indicating that the adductradical absorbs in the same region of the spectrum as that of the ␤-carotene ground-state absorption.
The rate constant, k 13 ϭ 2.5 Ϯ 0.1 ϫ 10 9 M Ϫ1 s Ϫ1 for the ␤-mercaptothiyl radical plus ␤-carotene was obtained from the slope of the linear plot of the first-order rate constant for the fast step versus ␤-carotene concentration. At such low doses per pulse (ϳ1 Gy), the slower decline in absorption at 451 nm could have been easily mistaken for first-order processes, including hydrogen abstraction from ␤-carotene (Reaction 14), radicaladdition at alternative sites along the polyconjugated backbone of ␤-carotene or as a consequence of a reverse repair reaction, e.g. Reaction 15 in which ␤-carotene could perturb a possible equilibrium reaction (Reaction 15) by irreversibly scavenging thiyl radicals via Reaction 13. In the latter case the observed biphasic bleaching of the ␤-carotene chromophore could reflect the fast formation of the [RS⅐⅐⅐␤-carotene] ⅐ radical-adduct via Reactions 13 and 15 followed by a slower formation of the [RS⅐⅐⅐␤-carotene] ⅐ radical-adduct reflecting the contribution from the reverse repair reaction (Reaction 15).
However, the radical species decayed slowly by pure secondorder kinetics with a half-life which decreased with increased radiation dose (0.2-2 Gy), suggesting the [RS⅐⅐⅐␤-carotene] ⅐ radicals undergo bimolecular decay to products via Reaction 16. In Fig. 4 the plot of the reciprocal of the first half-life of the radical species varied linearly with the initial radical concentration and from the slope of the fitted straight line the rate constant 2k 16 ϭ (2.1 Ϯ 0.1) ϫ 10 6 M Ϫ1 s Ϫ1 was determined.
2͓RS· · ·␤-carotene͔ ⅐ 3 product͑s͒ REACTION 16. No intercept is observed in Fig. 4, ruling out any contribution from the aforementioned first-order processes. The O 2 . radicalanion has been shown to undergo addition-elimination reactions with ␤-carotene (31). An analogous reversible reaction can be envisaged for the reaction of thiyl radicals with ␤-carotene, Reaction 17.
However, in this system the elimination of the thiyl radical from the [RS⅐⅐⅐␤-carotene] ⅐ radical-adduct is unable to compete with Reaction 16, which effectively pulls the equilibrium position (Reaction 17) toward [RS⅐⅐⅐␤-carotene] ⅐ radical-adduct formation.
No evidence was obtained for thiol repair of the [RS⅐⅐⅐␤carotene] ⅐ adduct-radical, since hydrogen atom transfer via Reaction 18 would have regenerated another RS ⅐ radical-capable of initiating a chain reaction and further bleaching of the ␤-carotene via Reaction 13.
Under the same experimental conditions the glutathione thiyl radical (GS ⅐ ) gave the same biphasic bleaching of the ␤-carotene ground-state absorption and exhibited similar spectral changes previously observed for the ␤-mercaptothiyl radical (see Fig. 5, left and right insets). Again no evidence was obtained for carotenoid radical-cation formation by the GS ⅐ radical, which clearly favors the radical-addition pathway. In this case, however, the rate constant for GS ⅐ radical-addition to ␤-carotene, k 13 ϭ (2.2 Ϯ 0.1) ϫ 10 8 M Ϫ1 s Ϫ1 is almost an order of magnitude slower than that previously measured for the ␤-mercaptoethanol thiyl radical. Rate constants, k 13 for a variety of thiyl radical reactions with ␤-carotene (together with other rate constants measured during this study) are displayed in Table I. Values for k 13 vary by almost 2 orders of magnitude, depending on the degree of hydrophilicity of the thiyl radical under study. The lowest rate constant, k 13 ϭ (4.2 Ϯ 0.3) ϫ 10 6 mM Ϫ1 s Ϫ1 , was measured for the thiyl radical of WR-1065, which contains a doubly protonated polyamino side chain (65). A similar trend has been observed for rates of thiyl radical abstraction of biallylic hydrogen atoms from PUFAs (42).
Since thiyl radicals are also capable of abstracting a hydrogen atom from the activated C-H bonds of alcohols and ethers (42), it was necessary to rule out the possibility of hydrogen abstraction from ␤-carotene (Reaction 14). Provided the experimental conditions remained constant the rate constants for the bimolecular decay of [␤-carotene] ⅐ radicals following Reaction 14 were expected to be independent of the nature of the attacking thiyl radical. However, further evidence supporting the assignment of the observed thiyl radical-mediated biphasic bleaching of the ␤-carotene ground-state absorption to formation and decay of the [RS⅐⅐⅐␤-carotene] ⅐ adduct-radical via Reactions 13 and 16 was derived from comparing second-order rate constants, 2k 16 for thiyl radicals of ␤-mercaptoethanol and WR-1065. Addition of the WR-1065 thiyl radical to ␤-carotene effectively lowered the rate of bimolecular decay of the adductradicals due to mutual electrostatic repulsion of the polyamino side chains (65), i.e. where R ϭ ϩ H 3 N(CH 2 ) 3 ϩ H 2 N(CH 2 ) 2 -. The measured rate constant 2k 16 ϭ (5.2 Ϯ 0.1) ϫ 10 5 M Ϫ1 s Ϫ1 was slower than the glutathione thiyl radical-adduct (1.3 Ϯ 0.1) ϫ 10 6 M Ϫ1 s Ϫ1 and almost an order of magnitude slower than that of the more hydrophobic ␤-mercaptoethanol thiyl radical-adduct (2.1 Ϯ 0.1) ϫ 10 6 M Ϫ1 s Ϫ1 . Thiyl radical-addition to ␤-carotene is kinetically favored relative to hydrogen abstraction from the polyconjugated C-H bonds.
Scavenging of RSO 2 ⅐ Radicals by ␤-Carotene-The absorption spectra obtained by pulse radiolysis of an N 2 -saturated tertbutyl alcohol/water mixture (50%, v/v) containing 10 mM C 2 H 5 SO 2 Cl and 1 mM phosphate buffer at pH 5 are shown in Fig. 6. The observed spectra were attributed to the rapid scav-enging of the thiyl-sulfonyl C 2 H 5 SO 2 ⅐ radical by ␤-carotene to generate carotenoid radical species. The characteristic absorption of the [␤-carotene] .
ϩ radical-cation peaking at 910 nm is accompanied by a biphasic decline in the ␤-carotene groundstate absorption between 300 and 600 nm, characteristic of a carotenoid radical-adduct. Thiyl-sulfonyl radicals clearly react by both electron transfer and radical-addition pathways, Reactions 19 and 20.
The radiation chemical yield of the [␤-carotene] . ϩ radical-cation at 910 nm was determined as G⑀ 910 ϭ 6.1 Ϯ 0. ⅐ ) ϭ 0.9. Formation of the [C 2 H 5 SO 2 ⅐⅐⅐␤-carotene] ⅐ radical-adduct therefore accounts for the remaining 29% of the C 2 H 5 SO 2 ⅐ radical yield. The fast decline in absorption at 451 nm was exponential and first-order in ␤-carotene concentration and reflects the contributions from both Reactions 19 and 20. The overall rate constant for C 2 H 3 SO 2 ⅐ radical attack on ␤-carotene was obtained from the slope of the plot of observed first-order rate constant (k obs ) versus ␤-carotene concentration and was found to be k 19 /k 20 ϭ 5.5 Ϯ 0.1 ϫ 10 9 M Ϫ1 s Ϫ1 . From the relative yields of [␤-carotene] .
ϩ radical-cation to the [C 2 H 5 SO 2 ⅐⅐⅐␤-carotene] ⅐ radical-adduct (i.e. 71 to 29%, respectively), the first-order rate constants k 19 and k 20 were determined. The observed rate constants for electron transfer (k ET ) and radical-addition (k RA ) are given by k ET ϭ (0.71 ϫ k obs ) and k RA ϭ [(1 Ϫ 0.71) ϫ k obs ], respectively. Fig. 7 shows the linear plots of k ET and k RA versus [␤-carotene] the slopes of which correspond to the rate con- ϩ radical-cation which decays via a bimolecular charge transfer process. In marked contrast all the RS ⅐ radicals under study undergo rapid addition reactions to generate [RS⅐⅐⅐␤-carotene] ⅐ adduct-radicals. Thiyl radicals are particularly electrophilic, and it is therefore not surprising that they add to centers of relatively high electron density such as the polyconjugated -system of ␤-carotene. This electrophilicity may be enhanced by the ability of sulfur to use d orbitals to accommodate negative charge. Thiyl-sulfonyl radicals can abstract an electron to generate the polyene radical-cation (71%) or add to the polyconjugated double bonds to generate [RSO 2 ⅐⅐⅐␤-carotene] ⅐ adduct-radicals (29%).
The antioxidant properties of ␤-carotene will of course not only reflect rates of free radical scavenging, but also the reactivity of the resultant ␤-carotene-derived radicals. Carotenoid radical-cations and adduct-radicals are highly resonance stabilized and must therefore be relatively unreactive compared with the attacking free radical species (10). Within lipid-rich environments, such as biomembranes or LDL carotene, the bimolecular decay of radical-adducts and radical-cations will generate non-radical products probably incapable of exerting prooxidative effects. Carotenoid radical-adducts are also likely to contribute to the antioxidant properties of ␤-carotene by scavenging PUFA radicals, thereby terminating the chain of lipid peroxidation as exemplified by Reactions 21 and 22.
PUFAOO ⅐ ϩ ͓RS· · · ␤-carotene͔ ⅐ 3 RS· · · ␤-carotene· · ·OOPUFA REACTION 22. The antioxidant efficiency of ␤-carotene against peroxyl radical-initiated lipid peroxidation in homogeneous solutions is diminished with increasing partial pressure of oxygen (pO 2 ) (18) Corroborating observations in liposomes (69,70) and in LDL (19) suggest that carotenoid-peroxyl radicals may prove important in vivo particularly in tissues of high pO 2 such as the lung. Based on the classical mechanisms of peroxyl radical formation, the conjugation of molecular oxygen with ␤-carotene adduct-radicals to generate a carotenoid-peroxyl radical (Reaction 23) represents a most feasible decay pathway following RS ⅐ radical oxidation of ␤-carotene. Due to resonance stabilization within the carotenoid adduct-radical, the reaction with oxygen may be represented as a reversible equlibrium Reaction 23.
High pO 2 is likely to drive Reaction 23 to the right, thereby promoting the autoxidation of ␤-carotene or PUFAs in a similar manner to other carotenoid-peroxyl radicals (9,18). Theoretically, any radical species likely to generate carotenoid adductradicals when scavenged by ␤-carotene, including thiyl-sulfonyl radicals, may initiate prooxidative pathways at high pO 2 . No evidence was obtained for the reaction of carotenoid radical-  cations with molecular oxygen. In the case of nitrogen dioxide, ␤-carotene is likely to behave as a chain-breaking antioxidant, since the carotenoid radical-cations may undergo bimolecular decay by charge transfer to generate non-radical products, which are unlikely to exhibit prooxidative effects. However, GSH has been shown to enhance nitrogen dioxide-induced lipid peroxidation and DNA strand breaks in model systems (35), possibly via the formation of thiyl and thiyl-derived radical species according to .
Reaction 24 represents a stoichiometric conversion of nitrogen dioxide to the glutathione thiyl radical (53,71), which in the context of free radical scavenging by ␤-carotene may produce a corresponding shift from carotenoid radical-cation to adductradical formation. Reactions 25-27 represent the formation of glutathione thiylperoxyl (GSOO ⅐ ), sulfonyl (GSO 2 ⅐ ), sulfonylperoxyl (GSO 2 OO ⅐ ), and sulfinyl (GSO ⅐ ) radicals, moderately strong oxidants which may exert damaging effects in vivo (44 -46, 72, 73). The interplay of free radical reactions in oxidative stress has been largely overlooked in studies of the antioxidant effects of ␤-carotene (9, 10), a factor that may contribute to some of the contradictory results obtained from in vitro models.