Does Risk of Hyperhomocysteinemia Depend on Thiol-Disulfide Exchange Reactions of Albumin and Homocysteine?

Significance: Increased plasma concentrations of total homocysteine (tHcy; mild-moderate hyperhomocysteinemia: 15-50 µM tHcy) are considered an independent risk factor for the onset/progression of various diseases, but it is not known about how the increase in tHcy causes pathological conditions. Recent Advances: Reduced homocysteine (HSH ∼1% of tHcy) is presumed to be toxic, unlike homocystine (∼9%) and mixed disulfide between homocysteine and albumin (HSS-ALB; homocysteine [Hcy]-albumin mixed disulfide, ∼90%). This and other notions make it difficult to explain the pathogenicity of Hcy because: (i) lowering tHcy does not improve pathological outcomes; (ii) damage due to HSH usually emerges at supraphysiological doses; and (iii) it is not known why tiny increments in plasma concentrations of HSH can be pathological. Critical Issues: Albumin may have a role in Hcy toxicity, because HSS-ALB could release toxic HSH via thiol-disulfide (SH/SS) exchange reactions in cells. Similarly, thiol-disulfide exchange processes of reduced albumin (albumin with free SH group of Cys34 [HS-ALB]) or N-homocysteinylated albumin are plausible alternatives for initiating Hcy pathological events. Adverse effects of albumin and other data reviewed here suggest the hypothesis of a role of albumin in Hcy toxicity. Future Directions: HSS-ALB might be involved in disruption of the antioxidant/oxidant balance in critical tissues (brain, liver, kidney). Since homocysteine-albumin mixed disulfide is a possible intermediate of thiol-disulfide exchange reactions, we suggest that homocysteinylated albumin could be a new pathological factor, and that studies on the redox role of albumin and mixed disulfide production via thiol-disulfide exchange reactions could offer new therapeutic insights for reducing Hcy toxicity.

Homocysteinemia control by cysteine in cerebral vascular patients after methionine loading test: evidences in physiological and pathological conditions in cerebro-vascular and multiple sclerosis patients.

The toxicity risk of hyperhomocysteinemia is prevented through thiol drug administration which reduces plasma total homocysteine (tHcy) concentrations by activating thiol exchange reactions. Assuming that cysteine (Cys) is a homocysteinemia regulator, the hypothesis was verified in healthy and pathological individuals after the methionine loading test (MLT). The plasma variations of redox species of Cys, Hcy, cysteinylglycine, glutathione and albumin (reduced, HS-ALB, and at mixed disulfide, XSS-ALB) were compared in patients with cerebral small vessels disease (CSVD) (n = 11), multiple sclerosis (MS) (n = 12) and healthy controls (n = 11) at 2-4-6 h after MLT. In MLT-treated subjects, the activation of thiol exchange reactions provoked significant changes over time in redox species concentrations of Cys, Hcy, and albumin. Significant differences between controls and pathological groups were also observed. In non-methionine-treated subjects, total Cys concentrations, tHcy and thiol-protein mixed disulfides (CSS-ALB, HSS-ALB) of CSVD patients were higher than controls. After MLT, all groups displayed significant cystine (CSSC) increases and CSS-ALB decreases, that in pathological groups were significantly higher than controls. These data would confirm the Cys regulatory role on the homocysteinemia; they also explain that the Cys-Hcy mixed disulfide excretion is an important point of hyperhomocysteinemia control. Moreover, in all groups after MLT, significant increases in albumin concentrations, named total albumin (tALB) and measured as sum of HS-ALB (spectrophometric), and XSS-ALB (assayed at HPLC) were observed. tALB increases, more pronounced in healthy than in the pathological subjects, could indicate alterations of albumin equilibria between plasma and other extracellular spaces, whose toxicological consequences deserve further studies.

Platelet function tests: a comparative review.

In physiological hemostasis a prompt recruitment of platelets on the vessel damage prevents the bleeding by the rapid formation of a platelet plug. Qualitative and/or quantitative platelet defects promote bleeding, whereas the high residual reactivity of platelets in patients on antiplatelet therapies moves forward thromboembolic complications. The biochemical mechanisms of the different phases of platelet activation - adhesion, shape change, release reaction, and aggregation - have been well delineated, whereas their complete translation into laboratory assays has not been so fulfilled. Laboratory tests of platelet function, such as bleeding time, light transmission platelet aggregation, lumiaggregometry, impedance aggregometry on whole blood, and platelet activation investigated by flow cytometry, are traditionally utilized for diagnosing hemostatic disorders and managing patients with platelet and hemostatic defects, but their use is still limited to specialized laboratories. To date, a point-of-care testing (POCT) dedicated to platelet function, using pertinent devices much simpler to use, has now become available (ie, PFA-100, VerifyNow System, Multiplate Electrode Aggregometry [MEA]). POCT includes new methodologies which may be used in critical clinical settings and also in general laboratories because they are rapid and easy to use, employing whole blood without the necessity of sample processing. Actually, these different platelet methodologies for the evaluation of inherited and acquired bleeding disorders and/or for monitoring antiplatelet therapies are spreading and the study of platelet function is strengthening. In this review, well-tried and innovative platelet function tests and their methodological features and clinical applications are considered.

Effects of cryoablation and radiofrequency ablation on endothelial and blood clotting activation.

Cryoablation (CA) emerged as an alternative procedure to radiofrequency (RF). The aim of this study was to compare haemostatic system alterations in patients undergoing RF or CA for atrioventricular nodal reentrant tachycardia ablation. von Willebrand factor (vWF), spontaneous whole blood platelet aggregation, prothrombin fragment F1 + 2 (F1 + 2), thrombin-antithrombin complex (TAT), plasminogen activator inhibitor type-1 (PAI-1), and clot lysis time (CLT) were determined in 48 patients (27 CA; 21 RF; 19M/29F, mean age 49.6 ± 17.6 years). Blood samples were obtained before the procedure (T0), immediately after (T1), and 24 h later (T2). At T1 both procedures were associated with a significant increase in levels of the endothelial activation marker vWF. At T2 vWF levels were lower in CA than in RF group. No changes in whole blood platelet aggregation before and after ablation procedures were observed. At T1 both groups determined an increase in blood clotting activation markers, F1 + 2, TAT, and DD. At T2 F1 + 2, TAT and DD levels were similar to baseline values. The comparison between RF and CA showed no significant differences in F1 + 2 and TAT levels, whereas at T1 DD levels were higher in CA group than in RF group. Both procedures induced a significant decrease in CLT, whereas no changes in PAI-1 levels were found. There were no significant differences in CLT and PAI-1 levels. The fibrinolytic efficiency analysis showed that at T1 DD/TAT and DD/F1 + 2 ratios were lower in RF group and remained lower in RF than in CA group at T2. CA procedure may be associated with a lower degree of endothelial damage and with a higher fibrinolytic capacity respect to RF.

The control of hyperhomocysteinemia through thiol exchange mechanisms by mesna.

In hyperhomocysteinemic patients, after reaction with homocysteine-albumin mixed disulfides (HSS-ALB), mesna (MSH) forms the mixed disulfide with Hcy (HSSM) which can be removed by renal clearance, thus reducing the plasma concentration of total homocysteine (tHcy). In order to assess the HSS-ALB dethiolation via thiol exchange reactions, the distribution of redox species of cysteine, cysteinylglycine, homocysteine and glutathione was investigated in the plasma of healthy subjects: (i) in vitro, after addition of 35 µM reduced homocysteine (HSH) to plasma for 72 h, followed by MSH addition (at the concentration range 10-600 µM) for 25 min; (ii) in vivo, after oral treatment with methionine (methionine, 200 mg/kg body weight, observation time 2-6 h). In both experiments the distribution of redox species, but not the total amount of each thiol, was modified by thiol exchange reactions of albumin and cystine, with changes thermodynamically related to the pKa values of thiols in the corresponding mixed disulfides. MSH provoked a dose-response reversal of the redox state of aged plasma, and the thiol action was confirmed by in vivo experiments. Since it was observed that the dimesna production could be detrimental for the in vivo optimization of HSSM formation, we assume that the best plasma tHcy lowering can be obtained at MSH doses producing the minimum dimesna concentration in each individual.

In vitro inhibition of human and rat platelets by NO donors, nitrosoglutathione, sodium nitroprusside and SIN-1, through activation of cGMP-independent pathways.

Three different NO donors, S-nitrosoglutathione (GSNO), sodium nitroprusside (SNP) and 3-morpholino-sydnonimine hydrochloride (SIN-1) were used in order to investigate mechanisms of platelet inhibition through cGMP-dependent and -independent pathways both in human and rat. To this purpose, we also evaluated to what extent cGMP-independent pathways were related with the entity of NO release from each drug. SNP, GSNO and SIN-1 (100 µM) effects on platelet aggregation, in the presence or absence of a soluble guanylate cyclase inhibitor (ODQ), on fibrinogen receptor (α(IIb)ß(3)) binding to specific antibody (PAC-1), and on the entity of NO release from NO donors in human and rat platelet rich plasma (PRP) were measured. Inhibition of platelet aggregation (induced by ADP) resulted to be greater in human than in rat. GSNO was the most powerful inhibitor (IC(50) values, µM): (a) in human, GSNO=0.52±0.09, SNP=2.83 ± 0.53, SIN-1=2.98 ± 1.06; (b) in rat, GSNO = 28.4 ± 6.9, SNP = 265 ± 73, SIN-1=108 ± 85. GSNO action in both species was mediated by cGMP-independent mechanisms and characterized by the highest NO release in PRP. SIN-1 and SNP displayed mixed mechanisms of inhibition of platelet aggregation (cGMP-dependent and independent), except for SIN-1 in rat (cGMP-dependent), and respectively lower or nearly absent NO delivery. Conversely, all NO-donors prevalently inhibited PAC-1 binding to α(IIb)ß(3) through cGMP-dependent pathways. A modest relationship between NO release from NO donors and cGMP-independent responses was found. Interestingly, the species difference in NO release from GSNO and inhibition by cGMP-independent mechanism was respectively attributed to S-nitrosylation of non-essential and essential protein SH groups.

A structurally driven analysis of thiol reactivity in mammalian albumins.

Understanding the structural basis of protein redox activity is still an open question. Hence, by using a structural genomics approach, different albumins have been chosen to correlate protein structural features with the corresponding reaction rates of thiol exchange between albumin and disulfide DTNB. Predicted structures of rat, porcine, and bovine albumins have been compared with the experimentally derived human albumin. High structural similarity among these four albumins can be observed, in spite of their markedly different reactivity with DTNB. Sequence alignments offered preliminary hints on the contributions of sequence-specific local environments modulating albumin reactivity. Molecular dynamics simulations performed on experimental and predicted albumin structures reveal that thiolation rates are influenced by hydrogen bonding pattern and stability of the acceptor C34 sulphur atom with donor groups of nearby residues. Atom depth evolution of albumin C34 thiol groups has been monitored during Molecular Dynamic trajectories. The most reactive albumins appeared also the ones presenting the C34 sulphur atom on the protein surface with the highest accessibility. High C34 sulphur atom reactivity in rat and porcine albumins seems to be determined by the presence of additional positively charged amino acid residues favoring both the C34 S⁻ form and the approach of DTNB.

The role of protein sulfhydryl groups and protein disulfides of the platelet surface in aggregation processes involving thiol exchange reactions.

Blood platelets are central to haemostasis and platelet aggregation is considered to be a direct index of platelet function. Although protein disulfides (PSSP) are structural components of most proteins, current evidence suggests that PSSP work together with protein SH groups (PSH) to activate various platelet functions in dynamic processes involving thiol/disulfide exchange reactions. Based on these assumptions, we performed experiments to demonstrate how PSH and PSSP are involved in platelet aggregation and how modifications of PSH and PSSP concentrations on the platelet surface by N-ethylmaleimide (NEM) (a PSH-blocking reagent) and dithiothreitol (DTT) (a PSSP-reducing reagent), respectively, may condition platelet susceptibility in protein rich plasma and washed platelets and integrin αIIbß3 conformation. Our data strongly suggest that the PSH blockage and the PSSP reduction of the platelet surface are deeply involved in aggregation processes evoked in protein rich plasma and washed platelets by ADP and collagen; that endogenous thiols (e.g. GSH) may interfere with NEM actions; that NEM and DTT, acting on preexisting PSH and PSSP of active platelets have opposite conformational changes on integrin αIIbß3 conformation. Although the precise mechanism and the populations of specific PSH and PSSP involved remain unresolved, our data support the notion that PSH and PSSP of the platelet surface are involved in platelet activation by thiol exchange reactions. A plausible molecular mechanism of PSH and PSSP recruitment during thiol exchange reactions is here proposed.

Measurement of mixed disulfides including glutathionylated proteins.

Mixed disulfides between protein cysteines and low-molecular-weight thiol cysteine or glutathione lead to the formation of cysteinylated proteins or glutathionylated proteins. These types of posttranslational modification are of great importance in the so-called redox regulation, by which changes in the redox state of the cell regulate a number of biochemical processes. We describe the methods for quantitatively measuring the various redox states of cellular thiols including protein cysteines and these mixed disulfides. These include spectrophotometric methods, which do not distinguish between protein-cysteine and protein-glutathione disulfides, and HPLC methods that make such distinction. Finally, we report a method for labeling proteins susceptible to glutathionylation with biotin, to allow their visualization by Western blot after electrophoretic separation, which is used to identify proteins undergoing this posttranslational modification.

Regulation of redox forms of plasma thiols by albumin in multiple sclerosis after fasting and methionine loading test.

Increases in plasma concentrations of total homocysteine (tHcy) have recently been reported in multiple sclerosis (MS) as the alteration of the methionine cycle for the onset of autoimmune diseases. Homocysteine (Hcy) and cysteine (Cys) are generated by the methionine cycle and transsulfuration reactions. Their plasma levels are subjected to complex redox changes by oxidation and thiol/disulfide (SH/SS) exchange reactions regulated by albumin. The methionine loading test (MLT) is a useful in vivo test to assay the functionality of the methionine cycle and transsulfuration reactions. Time courses of redox species of Cys, cysteinylglycine (CGly), Hcy, and glutathione have been investigated in plasma of MS patients versus healthy subjects after an overnight fasting, and 2, 4, and 6 h after an oral MLT (100 mg/kg body weight), to detect possible dysfunctions of the methionine cycle, transsulfuration reactions and alterations in plasma distribution of redox species. After fasting, the MS group showed a significant increase in cysteine-protein mixed disulfides (bCys) and total Cys (tCys). While plasma bCys and tCys in MS group remained elevated after methionine administration when compared to control, cystine (oxCys) increased significantly with respect to control. Although increased plasma concentrations of bCys and tCys at fasting might reflect an enhance of transsulfuration reactions in MS patients, this was not confirmed by the analysis of redox changes of thiols and total thiols after MLT. This study has also demonstrated that albumin-dependent SH/SS exchange reactions are a potent regulation system of thiol redox species in plasma.

The control of S-thiolation by cysteine via gamma-glutamyltranspeptidase and thiol exchanges in erythrocytes and plasma of diamide-treated rats.

Protein thiol modifications including cysteinylation (CSSP) and glutathionylation (GSSP) in erythrocytes of rat treated with diamide have been reported, but mechanism and origin of CSSP formation are unknown. Experiments were performed to relate CSSP formation to GSH hydrolysis via gamma-glutamyltranspeptidase (gamma-GT) and know whether cysteine may act as deglutathionylation factor. Time-dependent variations of redox forms of glutathione and cysteine were investigated in erythrocytes, plasma, liver and kidney of diamide-treated rats (0.4 mmol/kg by infusion for 45 min followed by 135 min of washout) in the presence and absence of acivicin (10 mg/kg administered twice 1 h before diamide) a known gamma-GT inhibitor. Diamide-treated rats showed decreased concentrations of erythrocyte GSH and increased levels of GSSP and CSSP. The rate of CSSP formation was slower than that of GSSP. Besides the entity of CSSP accumulation of erythrocytes was high and equivalent to approximately 3-fold of the normal plasma content of total cysteine. The result was paradoxically poorly related to gamma-GT activity because the gamma-GT inhibition only partially reduced erythrocyte CSSP. After gamma-GT inhibition, a large concentration fluctuation of glutathione (increased) and cysteine (decreased) was observed in plasma of diamide-treated rats, while little changes were seen in liver and kidney. There were indications from in vitro experiments that the CSSP accumulation in erythrocytes of diamide-treated rats derives from the coexistence of GSH hydrolysis via gamma-GT and production of reduced cysteine via plasma thiol exchanges. Moreover, reduced cysteine was found to be involved in deglutathionylation processes. Mechanisms of protein glutathionylation by diamide and deglutathionylation by cysteine were proposed.

Administration of minor polar compound-enriched extra virgin olive oil decreases platelet aggregation and the plasma concentration of reduced homocysteine in rats.

We investigated the effect of extra virgin olive oil (EVOO) on platelet aggregation and plasma concentrations of homocysteine (Hcy) redox forms in rats in relation to the minor polar compound (MPC) concentration of EVOO. We used 3 olive oil samples with similar fatty acid but different MPC concentrations: refined olive oil (RF) with traces of MPC (control oil), native EVOO with low MPC concentration (LC), and EVOO with high MPC concentration (HC) enriching LC with its own MPC. Oil samples were administered to rats by gavage (1.25 mL/kg body weight) using 2 experimental designs: acute (24-h food deprivation and killed 1 h after EVOO administration) and subacute (12-d treatment, a daily dose of oil for 12 d, and killed after 24 h of food deprivation). Platelet aggregation was induced by ADP (ex vivo tests) and a reduction in platelet reactivity occurred in cells from rats given LC in the subacute study and in cells from rats administered HC in both studies as indicated by an increase in the agonist half maximal effective concentration. HC inhibited platelet aggregation induced by low ADP doses (reversible aggregation) in cells of rats in both the acute and subacute studies, whereas LC had this effect only in the subacute experiment. Moreover, in rats administered HC in both experiments, the plasma concentration of free reduced Hcy (rHcy) was lower and Hcy bound to protein by disulfide bonds (bHcy) was greater than in RF-treated rats. bHcy was also greater in rats given LC than in RF-treated rats in the subacute experiment. Plasma free-oxidized Hcy was greater in rats given LC and HC than in those administered RF only in the subacute experiment. In conclusion, these results show that MPC in EVOO inhibit platelet aggregation and reduce the plasma rHcy concentration, effects that may be associated with cardiovascular protection.

Protein-thiol substitution or protein dethiolation by thiol/disulfide exchange reactions: the albumin model.

Dethiolation experiments of thiolated albumin with thionitrobenzoic acid and thiols (glutathione, cysteine, homocysteine) were carried out to understand the role of albumin in plasma distribution of thiols and disulfide species by thiol/disulfide (SH/SS) exchange reactions. During these experiments we observed that thiolated albumin underwent thiol substitution (Alb-SS-X+RSH<-->Alb-SS-R+XSH) or dethiolation (Alb-SS-X+XSH<-->Alb-SH+XSSX), depending on the different pK(a) values of thiols involved in protein-thiol mixed disulfides (Alb-SS-X). It appeared in these reactions that the compound with lower pK(a) in mixed disulfide was a good leaving group and that the pK(a) differences dictated the kind of reaction (substitution or dethiolation). Thionitrobenzoic acid, bound to albumin by mixed disulfide (Alb-TNB), underwent rapid substitution after thiol addition, forming the corresponding Alb-SS-X (peaks at 0.25-1 min). In turn, Alb-SS-X were dethiolated by the excess nonprotein SH groups because of the lower pK(a) value in mixed disulfide with respect to that of other thiols. Dethiolation of Alb-SS-X was accompanied by formation of XSSX and Alb-SH up to equilibrium levels at 35 min, which were different for each thiol. Structures by molecular simulation of thiolated albumin, carried out for understanding the role of sulfur exposure in mixed disulfides in dethiolation process, evidenced that the sulfur exposure is important for the rate but not for determining the kind of reaction (substitution or dethiolation). Our data underline the contribution of SH/SS exchanges to determine levels of various thiols as reduced and oxidized species in human plasma.

GSH depletion, protein S-glutathionylation and mitochondrial transmembrane potential hyperpolarization are early events in initiation of cell death induced by a mixture of isothiazolinones in HL60 cells.

We recently described that brief exposure of HL60 cells to a mixture of 5-chloro-2-methyl-4-isothiazolin-3-one (CMI) and 2-methyl-4-isothiazolin-3-one (MI) induces apoptosis at low concentrations (0.001-0.01%) and necrosis at higher concentrations (0.05-0.1%). In this study, we show that glutathione (GSH) depletion, reactive oxygen species generation, hyperpolarization of mitochondrial transmembrane potential (DeltaPsim) and formation of protein-GSH mixed disulphides (S-glutathionylation) are early molecular events that precede the induction of cell death by CMI/MI. When the cells exhibit common signs of apoptosis, they show activation of caspase-9, reduction of DeltaPsim and, more importantly, decreased protein S-glutathionylation. In contrast, necrosis is associated with severe mitochondrial damage and maximal protein S-glutathionylation. CMI/MI-induced cytotoxicity is also accompanied by decreased activity of GSH-related enzymes. Pre-incubation with L-buthionine-(S,R)-sulfoximine (BSO) clearly switches the mode of cell death from apoptosis to necrosis at 0.01% CMI/MI. Collectively, these results demonstrate that CMI/MI alters the redox status of HL60 cells, and the extent and kinetics of GSH depletion and S-glutathionylation appear to determine whether cells undergo apoptosis or necrosis. We hypothesize that S-glutathionylation of certain thiol groups accompanied by GSH depletion plays a critical role in the molecular mechanism of CMI/MI cytotoxicity.

Biochemical and biological aspects of protein thiolation in cells and plasma.

Protein thiolation is elicited by oxidation by different mechanisms and is involved in a variety of biological processes. Thiols, protein SH (PSH) and non-protein SH groups (NPSH, namely GSH), are in competition in all biological environments in the regulation of oxidant homeostasis because oxidants thiolate proteins, whereas GSH dethiolates them (e.g., GSSG + PSH --> GSSP + GSH). Although poorly investigated, the elimination of disulfides from thiolated proteins to regenerate critical PSH is important. These aspects are poorly known in cells, where glutaredoxin and peroxiredoxin operate as enzymes or potential chaperones to accelerate dethiolation. On the contrary, studies with plasma or albumin have highlighted the importance of protein conformation in dethiolation processes and have clarified the reason why homocysteine (thiol with potential toxicity) is preferentially bound to albumin as protein-thiol mixed disulfide with respect to other NPSH. Here we provide an overview of protein thiolation/dethiolation processes, with an emphasis on recent developments and future perspectives in this field.

The effects of age and hyperhomocysteinemia on the redox forms of plasma thiols.

We assayed the redox forms of cysteine (reduced [CSH], oxidized [CSSC], and bound to protein [CS-SP]), cysteinylglycine (CGSH; cysteinylgycine disulfide [CGSSGC] and cysteinylglycine-protein mixed disulfide [CGS-SP]), glutathione (GSH; glutathione disulfide [GSSG] and glutathione-protein mixed disulfide [GS-SP]), homocysteine (Hcy; homocystine [HcyS] and homocystine-protein mixed disulfides [bHcy]), and protein sulfhydryls in the plasma of healthy subjects (divided into 8 groups ranging in age from birth to 70 years) and patients with mild hyperhomocysteinemia associated with cardiovascular disease (heart-transplant patients) or vascular atherosclerosis, with or without renal failure. In healthy individuals, levels of disulfides and protein-mixed disulfides were more abundant than those of thiols, and those of protein-thiol mixed disulfides were higher than disulfides. Concentrations of CSH, GSH, and CGSH in the various groups had profiles characterized by a maximum over time. The concentration of Hcy was unchanged up to the age of 30 years, after which it increased. CSSC concentration increased gradually with age, whereas concentrations of the other disulfides were essentially unchanged. By contrast, the concentrations of all protein-thiol mixed disulfides, especially those with CSH, increased gradually with age. Ranks of distribution of the reduced forms changed with age (at birth, CSH > CGSH > GSH > Hcy; in 1- to 2-year-olds, CSH > GSH > CGSH > Hcy; and in 51- to 70-year-olds, CSH > CGSH = GSH > Hcy), whereas those of disulfides and protein-thiol mixed disulfides were substantially unchanged (in all age groups, CSSC > CGSSGC > GSSG = HcyS and CS-SP > CGS-SP > bHcy > GS-SP). In patients with pathologic conditions, plasma levels of disulfide forms CSSC, HcyS, CS-SP, and bHcy were significantly increased, whereas other redox forms of thiols were unchanged or showed variations opposite (increasing or decreasing) to control values. Maximal increases in disulfides and protein-thiol mixed disulfides were associated with renal failure. Our data suggest that increases in plasma bHcy concentrations in subjects with pathologic conditions were more likely the result of activation of thiol-disulfide exchange reactions between free reduced Hcy and CS-SP than of a direct action of reactive oxygen species.