All chemical reagents used in this study were of analytical grade and purchased from Sigma and Merck Co. The drugs were dissolved in 10 mM Tris-HCl-KCl (pH 7.4).

The blood samples were obtained from normal healthy donors and do not use NSAID and young people beween 20 - 30 years old. The blood (10 ml) was drawn from an antecubital vein into EDTA anticoagulant solution (20%). The samples were then centrifuged to separate plasma from the blood cells. Following, the erythrocytes were washed twice with 10 mM Tris-HC1, 150 mM KCl (pH 7.4) and packed erythrocytes were prepared in 3000 RPM at 25˚C. All procedures were approved by Human Research Ethics Committee/UFPA (065/06 CEP-CCS/UFPA).

The method of Gutteridge et al. [16] was used, after slight modification, to investigate the effect of the paracetamol and salicylic acid on membrane lipid peroxidation induced by H₂O₂. A volume of 0.2 mL of the packed erythrocytes were added to 0.8 mL of Tris-HCl azide (10.0 mM, pH 7.4) in 150.0 mM KCl (10mM sodium azide) and incubated with 1.0 mL of the appropriate amount of test drug in a shaking waterbath at 37.0˚C for 30 minutes to allow the catalase to be inactivated by azide within the cells. After adding 0.5 mL of 30.0 mM H₂O₂ (in 150.0 mM NaCl), the cells were incubated for 1 (one) hour at the same conditions. The reaction was terminated by adding 0.5 mL of 28.0% TCA (w/v). All tubes were centrifuged at 4000 x g for 10 minutes and 2.0 mL of the supernatant was collected. A volume of 1.0 mL of TBA 1.0% (w/v; in 0.05 M NaOH) was added to the portion of supernatant and the mixture was heated at 100˚C for 15 minutes. After cooling, the mixture absorbance measured was of 532 nm.

The data obtained, presented as mean values ± the standard error of the mean (SEM), were analyzed by one-way of variance (ANOVA) and Student-Newman-Keuls post hoc tests for the significant interrelation between the various groups using the SigmaPlot software. The value of p ≤ 0.05 was considered to be significantly different from the control.

Prior to any Density-Functional Theory (DFT) [17] calculation, both the paracetamol and the salicylic acid structures were submitted to a semi-empirical PM3 [18] geometry conformational search. Among the physical and the electronic descriptors (variables) with potential relevance for the hypothetical mechanism examined for both compounds, only the ionization potential (IP) of their neutral and anionic forms were selected. After the initial optimization with PM3, the geometry of the paracetamol and of the salicylic acid were re-optimized with DFT using B3LYP (three parameter hybrid exchange functional of Becke with the Lee-Yang-Parr correlation functional) [19] [20] and 6-31G(d) basis sets [21] . Only the most stable conformation of each compound was then used in our study.

The IP was calculated as the energy difference between a neutral molecule and the respective cation free- radical as showed in Equation (1).

The bond dissociation energies of the phenol group (BDE_OH) or semiquinone formation was calculated as the energy difference between a neutral molecule and the respective semiquinone plus the hydrogen radical as defined in Equation (2).

Radical stability is calculated by electron transfer related to trolox. In this case, the radical stability by electron transfer is determined by stabilization energies calculation (ΔEiso), as showed in Equation (3), where paracetamol and salicylic acid derivatives is ArOH and tolox is PhOH.

Figure 2 shows the antioxidant effect of NSAID phenol derivatives paracetamol (PAR) and salicylic acid (SAC) on H₂O₂ (30.0 mM)-induced lipid peroxidation in human erythrocytes compared to trolox (TLX). The TBARS concentrations for the TLX are significantly smaller than that for PAR or SAC. On 0.25 mM (M1) the TBARS concentration was increased by the no significantly amount of 8% for the PAR and of 43% for the SAC (p ≤ 0.01). However, in 0.5 mM (M2), the TBARS concentrations were reduced by 25% for PAR (p ≤ 0.01) and it was increased to no significantly amount of 6% for SAC. The same tendency is observed for 1.0 mM (M3) of concentration, for which the TBARS level was reduced by 43% for PAR (p ≤ 0.01) and increased by 8% for the SAC (p ≤ 0.05). The TBARS concentrations for the TLX were significantly reduced in all concentration, i.e. 42% (M1), 66% (M2), and 80% (M3) (p ≤ 0.001).

The antioxidant capacity of these phenol derivatives can be considered using theoretical methods. Hence, a structure and activity relationship (SAR) study was performed here in order to correlate biological values and theoretical property, such as electronic parameters by electron or hydrogen transfers. These parameters can be observed in Table 1.

The results show that the antioxidant activity of paracetamol in the human erythrocytes is more expressive than that of salicylic acid. However, it is known that paracetamol can cause severe hepatic necrosis at high doses both in human and animals [22] , since it can be activated by peroxidases of the cytochrome P-450 producing quinones and semiquinones forms, both active oxygen free-radical species [23] [24] . Furthermore, it has been shown that one of the main effects of the paracetamol toxicity is the depletion of cellular gluthatione (GSH) [24] , a necessary compound for the termination step of the lipid peroxidation as antioxidant defense. It has been suggested that paracetamol toxic dose treatment induces metabolic and membrane alteration making RBCs prone to hemolysis [25] . On the contrary, the salicylic acid does not present this effect at the same condition and concentration. Nevertheless, the paracetamol has showed antioxidant activity in vitro while the acetylsalicylic acid does not present it. Moreover, the salicylic acid is a compound derived from hydrolysis of acetylsalicylic acid (aspirin) [26] .

It has recently been shown that the paracetamol administration in rats induces an increasing of serotonin, norepinephrine and lipid peroxidation levels in the brain [27] . This drug alone or combined with aspirin exhibits antioxidant effects in the brain using the rat brain homogenates [28] .

The study performed in this work shows that the paracetamol and the salicylic acid have the potential to limit these undesirable effects, qualities that satisfy requirements for using them as cytoprotective agents in degenerative cells disorders, for example. We found that the potent free-radical scavenger and lipid peroxidation inhibition of paracetamol affords only a protection against H₂O₂-induced oxidative stress in human erythrocytes. The salicylic acid, which has less antioxidant ability, as shown in the present study, was capable of fully reviving the erytrhocytes damage caused by H₂O₂ only when combined with other phenol derivatives.

Our results are in agreement with previous reports where paracetamol has been shown to significantly reduce the KCN-induced superoxide anion generation as well as the KCN-induced lipid peroxidation. When paracetamol and salicylic acid are used in combination, the cytoprotective effect is increased [29] . Similarly, when administered prior to 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), the aspirin has been shown to attenuate striatal dopamine depletion in mice [30] - [32] . The salicylic acid, an aspirin metabolite, has also been shown to afford protection against MPTP- or 1-methyl-4-phenyl pyridinium (MPP+)-induced dopaminergic toxicity in mice [33] and rats [34] .

The theoretical results for the hypothetical mechanism for paracetamol (PAR) and salicylic acid (SAC) by hydrogen transfers using ΔEiso are related to trolox (TLX) [35] . Thus, paracetamol is approximately by 11.82 kcal∙mol⁻¹ and salicylic acid is 33.11 kcal∙mol⁻¹ (Table 1). Therefore, the paracetamol is more reactive than the salicylic acid. In fact, the spin density calculation reveals that the semiquinone free-radical of paracetamol has more resonance structures in the aromatic ring than that of the salicylic acid. Therefore, DFT calculations indicated that the paracetamol presents larger stability for the semiquinone free-radical formation compared to the salicylic acid [26] [36] - [40] .

Therefore, electron-donating groups as phenol and acetamide ortho or para substituted decrease the IP and BDE_OH values, while electron-withdrawing as carboxyl acid substituent increases the IP and BDE_OH values. Consequently, the salicylic acid showed the highest IP and BDE_OH values, while the paracetamol showed the lowest IP and BDE_OH values.

These results show the reason why the oxidation of the paracetamol is more favored than that of the salicylic acid when the electron or hydrogen abstractions are taken into account. Nonetheless, the DFT calculations indicate that paracetamol present the lowest energy barriers when compared to trolox. Therefore, these results can explain the possible mechanism for both salicylic acid and paracetamol on the forced oxidative stress in human erythrocytes. All theoretical parameters, such as HOMO, LUMO, ionization potentials (IP), and hydroxyl bond dissociation energies (BDE_OH) of the studied molecules showed good correspondence with the antioxidant activity evaluated by TBARS levels.

The trolox has shown the lowest values for HOMO (−5.07 eV), LUMO (0.07 eV), IP (158.49 kcal∙mol⁻¹), and BDE_OH (89.84 kcal∙mol⁻¹). Moreover, the salicylic acid has the highest values for HOMO (−6.24 eV), LUMO (−1.46 eV), IP (192.61 kcal∙mol⁻¹), and BDE_OH (111.82 kcal∙mol⁻¹). Nonetheless, for the paracetamol all values were intermediates HOMO (−5.48 eV), LUMO (−0.08 eV), IP (171.32 kcal∙mol⁻¹), and BDE_OH (93.54 kcal∙mol⁻¹) compared to trolox and salicylic acid.