Antioxidant potential of curcumin-related compounds studied by chemiluminescence kinetics, chain-breaking efficiencies, scavenging activity (ORAC) and DFT calculations

Summary This study compares the ability to scavenge different peroxyl radicals and to act as chain-breaking antioxidants of monomers related to curcumin (1): dehydrozingerone (2), zingerone (3), (2Z,5E)-ethyl 2-hydroxy-6-(4-hydroxy-3-methoxyphenyl)-4-oxohexa-2,5-dienoate (4), ferulic acid (5) and their corresponding C 2-symmetric dimers 6–9. Four models were applied: model 1 – chemiluminescence (CL) of a hydrocarbon substrate used for determination of the rate constants (k A) of the reactions of the antioxidants with peroxyl radicals; model 2 – lipid autoxidation (lipidAO) used for assessing the chain-breaking antioxidant efficiency and reactivity; model 3 – oxygen radical absorbance capacity (ORAC), which yields the activity against peroxyl radicals generated by an azoinitiator; model 4 – density functional theory (DFT) calculations at UB3LYP/6-31+G(d,p) level, applied to explain the structure–activity relationship. Dimers showed 2–2.5-fold higher values of k A than their monomers. Model 2 gives information about the effects of the side chains and revealed much higher antioxidant activity for monomers and dimers with α,β-unsaturated side chains. Curcumin and 6 in fact are dimers of the same monomer 2. We conclude that the type of linkage between the two “halves” by which the molecule is made up does not exert influence on the antioxidant efficiency and reactivity of these two dimers. The dimers and the monomers demonstrated higher activity than Trolox (10) in aqueous medium (model 3). A comparison of the studied compounds with DL-α-tocopherol (11), Trolox and curcumin is made. All dimers are characterized through lower bond dissociation enthalpies (BDEs) than their monomers (model 4), which qualitatively supports the experimental results.

Starting materials (vanilin (12), zingerone (3), ferulic acid (5)), reagents and solvents were obtained from commercial suppliers (Sigma-Aldrich Chemie GmbH and E. Merck, Germany) and were used without further purification. Acetone was freshly distilled from CaCl 2, tetrahydrofuran (THF) from sodium wire. Melting points were determined on a Büchi 530 apparatus (Assago, Italy) and are uncorrected. All 1 H and 13 C NMR spectra were recorded in CDCl 3 and DMSO-d 6 solutions with a Varian Mercury Plus (Palo Alto, CA) spectrometer at 400 MHz and 100.57 MHz, respectively. Chemical shifts () are given in ppm; multiplicates are indicated by s (singlet), d (doublet) or dd (double of doublets). Elemental analyses were performed using an elemental analyzer Perkin-Elmer model 240 C (Waltham, Massachusetts). Flash chromatography was carried out with silica gel 60, 230-400 mesh (Aldrich, Milano, Italy) eluting with appropriate solution in the stated v:v proportions. Analytical TLC was performed with either 0.25 mm thick silica gel plates (Polygram SilG/UV 254 , Macherey-Nagel, VWR International, Italy) or 0.2 mm thick silica gel plates (60 5254 Merck, Italy). The purity of all new compounds was judged to be >98% by 1 H NMR and 13 C NMR spectral determination. Monomers 2 and 3 and dimers 6 and 7 were previously prepared by us following straightforward methods as depicted in Scheme S1 [s1]. Dimer 9 was prepared according to the known procedure [s2].
1.1.Synthesis of C 2 -symmetric hydroxylated biphenyls 6, 7, 9 and of monomer 2 Biphenyl 7 was prepared in 65% yield by C-C coupling of zingerone (3) in the presence of methyl tributyl ammonium permanganate (MTBAP) in dichloromethane at room temperature for 1 h (Scheme S1). Unfortunately, the procedure turned out to be unsuccessful in preparation of dimer 6 that was obtained in 83% yield by Claisen-Schmidt condensation of vannilin dimer 14 in the presence of large excess of LiOH in acetone at room temperature. With a slight modification of the above synthetic procedure, dehydrozingerone (2) was achieved starting from vanillin (12) in acetone using NaOH as base. Dimer 9 was prepared by Perkin reaction of diacetate dimer of vanillin (15) in the presence of malonic acid and bases and further hydrolysis of the acetate groups. All compounds prepared were solid, air-stable and they were fully characterized. In the synthesis of unsaturated compounds 2, 6, and 9 trans-configuration was exclusively obtained at the olefinic double bond and detected by NMR spectroscopy.

2.1.Model 1. Estimation of rate constant of antioxidant reaction with peroxyl radicals (k A ) using kinetic chemiluminescence (CL) method
Similarly as described in [s3] the chemiluminescence measurements were performed with the Hamamatsu photosensor module H7467 supplied with the RS-232C interface. The probe chemiluminescent hydrocarbon 'cocktails' for antioxidant assay consisted of chlorobenzene solutions of ethylbenzene (RH) being oxidized by molecular oxygen in the presence of 2,2'azobisisobutyronitrile (AIBN) as free radical source used to initiate the oxidation process. All the chemicals used in this study were obtained from standard suppliers and purified by known procedures [s4].
Analogous to the description in [s3] the strategy behind the chemiluminescence methodology used in this work is based on the peroxy radical-mediated excited-state generation in a probe solution of a model hydrocarbon substrate (RH, in the present case, ethylbenzene) upon oxidation of the latter. The advantage of such a kind of hydrocarbon as chemiluminescent substrate resides in a well-understood and straightforward mechanism of its oxidation at moderate temperatures (20-80 °C). Besides, under these conditions, its oxidation rate is very easy to control, simply by choosing an appropriate concentration of the reaction initiator (Y, see Scheme S2), a peroxide or azo compound (in the present case, AIBN), which works as thermal source of initiating free radicals (Y  ). A relatively low initiation rate of the free radical oxidation process enables the consumption of both the hydrocarbon substrate and the initiator to be neglected during the course of reaction. As a matter of fact, the quasi-stationary reaction conditions allow the chemiluminescence intensity to be kept constant over the time of the experiment in the absence of antioxidative reactants.
The chemiluminescent assay for antioxidant monitoring is based on the competition between the self-reaction of peroxyl radicals (reaction step "k t " in Scheme S2), giving rise to light emission, and scavenging the peroxyl radicals by antioxidants (monomers 1 and dimers 2 ), thus inhibiting the oxidation process and thereby quenching the light emission. The kinetic analysis of the reaction sequence affords suitable expressions which relate to the time profile of the chemiluminescence signal with the pertinent characteristics of an antioxidant being studied. The strength of an antioxidant (its most important characteristics in the context of the present work) is quantified by the rate constant k A , whose value may be acquired from the slope of the chemiluminescence time profile at the inflection point according to Equation 1 [s5,s6] i rel is a dimensionless light intensity given by the ratio of the intensities in the presence (I) and in the absence (I 0 ) of an antioxidant, i rel = I/I 0 (it is noteworthy that consumption of an antioxidant in the probe chemiluminescence solution after the induction period results in recovery of the lightemission intensity, I ∞ , i.e., I ∞ = I 0 and i rel = I/I 0 = I/I ∞ [s5,s6] and R IN stays for the reaction initiation rate defined by Equation 2. In the latter expression, k dec is the rate constant of the initiator (Y) decomposition to generate initiating radicals (Y  in Scheme S2) and γ c is their probability to escape the solvent cage. The 2γ c k dec data are available for most of "standard" initiators (such as AIBN, for instance) in a number of organic solvents and in a temperature range of 20 to 80 °C. Calculation of k Am and k Ad from Equation 1 was made taking into account that the value of 2k t obtained experimentally in the same system (chemiluminescence of ethylbenzene) and temperature of 50 °C by Belyakov et al.
1 M i OH in the kinetic schemes and in the DFT calculation details. 2 D i (OH) 2 in the kinetic schemes and in the DFT calculation details.

S6
2RO 2 •  R=O* + ROH +O 2 (2k t ) Scheme S2: Basic kinetic scheme of initiated oxidation of ethylbenzene (RH) in CL, in absence and in presence of studied compounds.

Model 2. Estimation of the chain-breaking antioxidant activity of the studied compounds during lipid autoxidation
Lipid samplesimilarly as described in [s7] triacylglycerols of commercially available sunflower oil (TGSO) were cleaned from pro-and antioxidants by adsorption chromatography [s8] and stored under nitrogen atmosphere at −20 °C. Fatty acid composition of the lipid substrate was determined according to Christy [s9] by GC analysis of the methyl esters of the total fatty acids obtained with a GC-FID Hewlett-Packard 5890 equipment (Hewlett-Packard GmbH, Austria) and a capillary column HP INNOWAX (polyethylene glycol mobile phase, Agilent Technologies, USA) 30 m  0.25 mm  0.25 mm. The temperature gradient started from 165 °C increased to 230 °C with 4 °C/min and held at this temperature for 15 min; injection volume was 1 µL. Injector and detector temperatures were 260 and 280 °C, respectively. Nitrogen was the carrier gas at a flow rate of 0.8 mL/min. The analyses were performed in triplicate. Six different fatty acids were present in TGSO: 16:0 -6.7%; 18:0 -3.6%; 18:1 -25.1%; 18:2 -63.7%; 20:0 -0.2%; 22:0 -0.7%. Lipid samples containing various inhibitors were prepared directly before use. Aliquots of the antioxidant solutions in purified acetone were added to the lipid sample. Solvents were removed under a nitrogen flow.
Lipid autoxidationanalogous to the description in [s7] the process was carried out in a thermostatic bath at 80 °C (± 0.2 °C) by blowing air through the samples in special vessels and was monitored by withdrawing samples at measured time intervals and subjecting them to iodometric determination of the primary products (lipid hydroxyperoxides, LOOH) concentration, i.e., the peroxide value (PV) [s10,s11]. All kinetic data are expressed as the average of two independent measurements.
Triacylglycerols of sunflower oil (TGSO) are a mixture of different types of fatty acids glycerol esters, but only those fatty acids with pentadiene structures are vulnerable to oxidation with atmospheric oxygen. Phenolic antioxidants inhibit or retard lipid oxidation by interfering with either chain propagation or initiation by readily donating hydrogen atoms to lipid peroxyl radicals [s12-s14]. Recently, Kancheva et al. [s15] have published the detailed mechanism of lipid autoxidation in homogeneous solutions under sufficient oxygen pressure in the presence of monomer M i OH and dimer D i (OH) 2 (Scheme S3). In this mechanism: LH is linoleic acid with its allylic hydrogen; M i OH -the studied monomeric antioxidant; LO 2 • -lipid peroxide radicals; M i O • -phenoxyl radicals; P , P A , P 1 -P 4corresponding products formed; QP-quinoidal peroxides; R IN -the rate of initiation; k p , k t , k b , k Am, k' Am , k tMO , k pMO , k QPcorresponding rate constants of different reactions. In the presence of an antioxidant, the rate of hydroperoxides formation is related to the ratio of the lipid [LH]

Rate of inhibited oxidation (R Ad )
R Ad = k p [LH]R IN /nk Ad [D i (OH) 2 ] 0 (5) In Equation 3-5: k p /(2k t ) 0.5 is the oxidizability of the lipid substrate (linoleic acid glycerol ester); n-is the number of radicals trapped per inhibitor for molecule (the stoichiometric factor).
Determination of the main kinetic parameters of the studied compounds [s12-s15] Antioxidant efficiency means the potency of an antioxidant to increase the persistence towards oxidation of a lipid substrate by blocking the radical-chain process. It was presented with a protection factor (PF) meaning how many times the antioxidant increases the persistence against oxidation of the lipid sample, determined as a ratio between the induction periods in presence (IP A ) and in absence (IP C ) of an antioxidant, i.e., PF = IP Am /IP C for the monomers and PF = IP Ad /IP C for the dimers.
Inhibition degree (ID) is a measure of the antioxidant reactivity, which manifests how many times the antioxidant shortens the oxidation chain length, i.e. ID = R C /R Am in the case of monomers and ID = R C /R Ad in the case of dimers. For that reason, it is one of the most important kinetic parameters. Initial rates of lipid autoxidation in the absence (R C ) and in the presence of the antioxidant (R Am or R Ad ) were found from the tangent at the initial phase of the kinetic curves of hydroperoxides accumulation.

2.3.Model 3. Estimation of oxygen radical absorbance capacity (ORAC)
ORAC was measured according to the method of Ou et al. [s16] with some modifications, analogous to the description in [s17]. The method affords the antioxidant scavenging activity against peroxyl radical generated by 2,2'-azobis (2-amidinopropane) dihydrochloride (AAPH, Y A ) at 37 °C. Fluorescein 3 was used as the fluorescent probe. AAPH and fluorescein were obtained from Sigma-Aldrich Chemie (Steinheim, Germany). The experiments were performed on FLUOstar OPTIMA fluorimeter (BMG LABTECH, Offenburg, Germany). The loss of fluorescence of fluorescein was an indication of the extent of fluorescein damage in its reaction with the peroxyl radical. The protective effect of an antioxidant was measured by assessing the area under the fluorescence decay curve (AUC) as compared to that of blank in which no antioxidant is present. Solutions of AAPH, fluorescein and Trolox were prepared in a phosphate buffer (75 mM, pH 7.4). Samples were diluted in phosphate buffer as well. Reaction mixture (total volume 200 μL) contained fluorescein (170 μL, final concentration 5.36 × 10 −8 M), AAPH -(20 μL, final concentration 51.51 mM), and sample (10 μL). Fluorescein solution and sample were incubated at 37 °C for 20 min, and AAPH (dissolved in 37 °C buffer) was added. The mixture was incubated for 30 s before the initial fluorescence was measured. After that, the fluorescence readings were taken at the end of every cycle after shaking. For the blank, 10 μL of phosphate buffer was used instead of the sample. Antioxidant activity was expressed in Trolox equivalents. For defining the standard curve 10 μL of 3.13, 6.25, 12.50, 25 and 50 μM Trolox solutions (final concentrations 0.16, 0.31, 0.63, 1.25 and 2.50 μM, respectively) were used instead of the sample. One ORAC unit is assigned to the net protection area, provided by 10 µL, 1 μM Trolox solution (final concentration -0.05 μM). Results were expressed as Relative Trolox Equivalents (RTE).
According to the basic kinetic consideration of Huang et al. [s18] the new reaction sequence (Scheme 4) is proposed: Scheme S4: Basic kinetic scheme of initiated oxidation in ORAC.
The net AUC was obtained by subtracting the AUC of the blank from that of the sample. The relative ORAC value (Relative Trolox Equivalents) was calculated as [s16]: [(AUC Sample -AUC Blank )/(AUC Trolox -AUC Blank )] × (molarity of Trolox/molarity of sample) (6).

2.4.Model 4. DFT calculations
In this work, DFT calculations have been performed using GAUSSIAN 09 program package [s19]. The geometries of curcumin, hydroxylated biphenyls 6-9, corresponding monomers 2-5, fluorescein, DL-α-tocopherol, Trolox and their radicals were optimized using unrestricted open-shell approach (UB3LYP) and 6-31+G(d,p) basis sets [s20-s22] without symmetry constraints (C 1 symmetry was assumed) with the default convergence criteria. The reliability of the used combination of the above-mentioned hybrid functional and double-zeta basis set with polarization and diffusion functions in the description of the phenoxyl radicals was proved by Anouar et al. for ferulic acid [s23]. No spin contamination is found for radicals. Local minima were verified by establishing that the Hessians have zero negative eigenvalues. Unscaled thermal corrections to enthalpy were added to the total energy values. The homolytic bond dissociation enthalpy (BDE) was used like descriptor of the free-radical scavenging activity. The BDEs for the generation of the respective radicals/biradicals from the parent compounds are