Journal of Inorganic Biochemistry 139 (2014) 38–48

Contents lists available at ScienceDirect

Journal of Inorganic Biochemistry journal homepage: www.elsevier.com/locate/jinorgbio

Curcumin derivatives as metal-chelating agents with potential multifunctional activity for pharmaceutical applications Erika Ferrari a,⁎, Rois Benassi a, Stefania Sacchi a, Francesca Pignedoli a, Mattia Asti b, Monica Saladini a a b

Department of Chemical and Geological Sciences, University of Modena and Reggio Emilia, via Campi 183, 41125 Modena, Italy Nuclear Medicine Unit, Advanced Technology Department, Santa Maria Nuova Hospital-IRCCS, via Risorgimento 80, 42100 Reggio Emilia, Italy

a r t i c l e

i n f o

Article history: Received 23 January 2014 Received in revised form 29 May 2014 Accepted 2 June 2014 Available online 11 June 2014 Keywords: Curcumin derivatives Cu2 + complexes Ga3 + complexes DFT calculation NMR spectroscopy Alzheimer's disease

a b s t r a c t Curcuminoids represent new perspectives for the development of novel therapeutics for Alzheimer's disease (AD), one probable mechanism of action is related to their metal complexing ability. In this work we examined the metal complexing ability of substituted curcuminoids to propose new chelating molecules with biological properties comparable with curcumin but with improved stability as new potential AD therapeutic agents. The K2T derivatives originate from the insertion of a -CH2COOC(CH3)3 group on the central atom of the diketonic moiety of curcumin. They retain the diketo-ketoenol tautomerism which is solvent dependent. In aqueous solution the prevalent form is the diketo one but the addition of metal ion (Ga3+, Cu2+) causes the dissociation of the enolic proton creating chelate complexes and shifting the tautomeric equilibrium towards the keto–enol form. The formation of metal complexes is followed by both NMR and UV–vis spectroscopy. The density functional theory (DFT) calculations on K2T21 complexes with Ga3+ and Cu2+ are performed and compared with those on curcumin complexes. [Ga(K2T21)2(H2O)2]+ was found more stable than curcumin one. Good agreement is detected between calculated and experimental 1H and 13C NMR data. The calculated O\H bond dissociation energy (BDE) and the O\H proton dissociation enthalpy (PDE), allowed to predict the radical scavenging ability of the metal ion complexed with K2T21, while the calculated electronic affinity (EA) and ionization potential (IP) represent yardsticks of antioxidant properties. Eventually theoretical calculations suggest that the protontransfer-associated superoxide-scavenging activity is enhanced after binding metal ions, and that Ga3+ complexes display possible superoxide dismutase (SOD)-like activity. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Curcumin is a homodimer of feruloyl-methane containing a 3methoxy and 4-hydroxy phenyl group, a heptadiene with two Michael acceptors, and an α,β-diketone (Scheme 1). The importance given to this phytocompound, extracted from Curcuma longa L. (turmeric), to related curcuminoids and their conjugates, is attested by the everincreasing in vitro and in vivo tests reported in literature. In the past few decades, numerous studies on curcumin have demonstrated its multiple pharmacological activities and promising application as a novel drug in various diseases [1]. Curcumin is well studied due to its putative cancer prevention and anti-cancer activities which are mediated influencing multiple signaling pathways [2,3]. Recently, curcumin has been considered as a key molecule for the development of novel therapeutics for Alzheimer's disease (AD) [4]. In fact several studies revealed that curcumin is able to destabilize the preformation of βamyloid (Aβ) fibrils in the central nervous system. Based on the assay

⁎ Corresponding author. Tel.: +39 059 2055026; fax: +39 059 373543. E-mail address: [email protected] (E. Ferrari).

http://dx.doi.org/10.1016/j.jinorgbio.2014.06.002 0162-0134/© 2014 Elsevier Inc. All rights reserved.

results of mitochondrial metabolic markers, curcumin was found to protect human neuroblastoma SH-SY5Y cells against Aβ-induced damage of mitochondrial energy metabolism [5]. The utility of curcumin is however limited by its poor watersolubility, fast degradation and relatively low in vivo bioavailability. The recent quest for a ‘supercurcumin’ has led to different proposed formulations with various oils and with inhibitors of metabolism (e.g., piperine), liposomal and polymeric nanoparticles encapsulations [6] and conjugation of curcumin prodrugs [7]. PEGylation has been an instrumental avenue to increase water-solubility, stability and bioactivity of curcumin [8,9]. An alternative way is to functionalize curcumin in order to improve its bioavailability without compromising its beneficial properties. In this way we have recently synthesized some new curcumin derivatives with greater chemical stability; among these the tert-butyl ester derivatives (Scheme 2) have shown cytotoxicity against different tumorogenic cell lines, with IC50 values similar or in many cases lower than curcumin itself [10]. The presence, in these compounds, of a substituent on the central C atom of the β-diketo moiety of curcumin does not significantly alter the ability of the molecules to undergo a keto–enol tautomerism which is a fundamental feature of curcumin.

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

39

2. Experimental section 2.1. General procedure and chemicals All chemicals were reagent grade and used without further purification unless otherwise specified. They were purchased from SigmaAldrich. Compounds K2T21, K2T23, K2T24, and K2T33 were synthesized and characterized as reported in ref. [10]. K2T31 [(3Z-5E)-tert-butyl-4-hydroxy-6-(4-hydroxyphenyl)-acrolyl) hexa3,5-dienoate] was here synthesized according to ref. [10]. Orange– red powder, 40% yield; KE 60%: 1H NMR (MeOD-d4) δ 3.60 (s, 2H; H-2), 7.11 (d, 2H; H-5, J = 15.2 Hz), 7.68 (d, 2H; H-6, J = 15.2 Hz), 6.85 (d, 4H; H-8, J = 8.1 Hz), 7.54 (d, 4H; H-9, J = 8.1 Hz), 1.45 (s, 9H; –COO(CH3)3); 13C NMR (MeOD-d4) δ 171.0 (C-1), 32.7 (C-2),104.8 (C-3), 183.9 (C-4), 118.5 (C-5), 142.9 (C-6), 126.8 (C-7), 116.7 (C-8), 131.2 (C-9), 160.2 (C-10), 81.0 (–COOC(CH3)3), 26.8 (–C(CH3)3). DK 40%: 1H NMR (MeOD-d4) δ 2.88 (d, 2H; H-2), 4.85 (t, 1H; H-3), 6.88 (d, 2H; H-5, J = 15.9 Hz), 7.70 (d, 2H; H-6, J = 15.9 Hz), 6.85 (d, 4H; H-8, J = 8.1 Hz), 7.54 (d, 4H; H-9, J = 8.1 Hz), 1.45 (s, 9H; –COO(CH3)3); 13C NMR (MeOD-d4) δ 171.0 (C-1), 33.4 (C-2), 59.0 (C-3), 194.2 (C-4), 122.6 (C-5), 146.1 (C-6), 129.8 (C-7),116.7 (C-8), 131.2 (C-9), 160.2 (C-10), 81.0 (–COOC(CH3)3), 26.8 (–C(CH3)3).

Scheme 1. Curcumin.

Another interesting domain of investigation is curcumin metalchelation capacity, bearing probable correlation with its cytoprotective potency. Iron complexes of curcumin seem to have high potential in the treatment of cancer [11], while gallium complexes have remarkable antiviral effects on HSV-1 in cell culture [12]. 1 H NMR data state that the dissociated keto–enol moiety of the ligand is involved in metal chelation and the formed complexes have high stability at physiological pH [13,14]. Various computational and analytical reports were published on curcumin interaction with Cu2+ ion, too [15–17]. The interaction of curcumin with Cu+ and Cu2+ ions may be involved in the observed anticancer properties [18,19] while the interactions of Cu2+ with Aβ are generally accepted to be a crucial process in the development of neurotoxicity in AD [20,21]. In fact drugs for the treatment of this disease seem to be correlated with metal complexation, acting as superoxide scavengers. Thus the skill of curcumin to chelate metal ions such as iron and copper could be a useful feature in developing new treatment for AD. In addition Zn-curcumin was found to increase superoxide dismutase (SOD) activity in chronic gastric ulcers in rats [22]. In summary, the curcumin derivatives here investigated are designed as potential neuroprotective agents, acting as metal-chelators with improved stability at physiological pH and increased cytotoxic activity with respect to curcumin. The aim of this work is to investigate by means of experimental and theoretical approach, the ability of these compounds (Scheme 2) to interact with Ga3+ and Cu2+ ions, to compare the results with the lead compound curcumin, and to evaluate their potential application as therapeutics.

2.2. UV–vis spectroscopy UV–vis spectrophotometric measurements were performed using Jasco V-570 spectrophotometer at 25.0 ± 0.1 °C in the 200–600 nm spectral range employing 1 cm quartz cells. Owing to the poor water solubility of the compounds, a methanol mother solution (5 × 10−3 M) was diluted in water in order to obtain 5 × 10−5 M solutions used for pHmetric titrations. The pH value was varied by adding small amounts of concentrated NaOH or HCl to obtain at least 25 different values in the pH range 4–9. The metal–ligand solutions were obtained by adding appropriate quantities of GaCl3 or CuCl2 water solutions as to obtain 1:1, 1:2, 1:3 and 1:4 M:L molar ratios ([L] = 5 × 10−5 M). A constant ionic strength of 0.1 M (NaNO3) was maintained in all experiments. Each titration was performed at least three times. The overall protonation constants (logβLH) and the overall stability constants of metal complexes

O

O 3

O

R1

1

R

O

4 2

6 5

7

12

11 10

8

R

R1

9

OH O

R1

O

O R

O

R

R1

R1

OH

O R

O

R

R= R1= H K2T33

R = OCH3, R1 = OCOCH3 K2T24

R= OCH3, R1 = H K2T23

R = H, R1 = OH K2T31

R1

R = OCH3, R1 = OH K2T21 Scheme 2. General scheme of tautomeric equilibrium between keto–enol (KE) and di-keto (DK) forms of curcuminoids.

40

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

(logβMLH) were evaluated from spectrophotometric data using the software HypSpec [23]. 2.3. Nuclear magnetic resonance NMR spectra were recorded on a Bruker Avance AMX-400 spectrometer with a Broad Band 5 mm probe in inverse detection. Nominal frequencies were 100.13 MHz for 13C and 400.13 MHz for 1H. For each sample 0.5 mL of mM solution was prepared in MeOD-d4. The typical acquisition parameters for 1 H were used. 2D homonuclear shift correlation (H,H-COSY) spectra were acquired using the Bruker pulse sequence “cosygpqf” implying gradient pulses for selection with second flip angle being 90°. The phase sensitive 2D 1 H, 13 C HSQC (hetero-nuclear single quantum coherence) was performed via double INEPT transfer using Echo/Antiecho-TPPI gradient selection; decoupling during acquisition was performed using TRIM pulses in INEPT transfer with multiplicity editing during the selection step [24]. For 2D H,X-Hetero Correlated Spectroscopy HMBC (heteronuclear multiple bond coherence) proper parameters were used as suggested by literature [25]. The investigations of Ga3 +:ligand systems were performed in MeOD-d4, by adding to mM ligand solution (0.5 mL) small additions of Ga(NO3)3 solution (10 μL each) until reaching the equimolar ratio. Spectra were registered after few minutes from each addition. 2.4. Superoxide dismutase (SOD) assay Superoxide was generated by enzymatic methods employing a xanthine/xanthine oxidase assay [26]. A mixture of 50 μM xanthine, 300 μM NBT2+ (nitro blue tetrazolium), 0.04 unit/ml xanthine oxidase and 0–120 μM M(K2T21)2 complex (M = Ga3+, Cu2+) in 20 mM phosphate buffer (pH 7.4) was incubated at 25 °C for 10 min and the NBT+ formed was measured spectrophotometrically at 560 nm. The percentage of inhibition (%I) of NBT+ formation was calculated from %I = (A0 − As)100 / A0 where A0 and As are the maximum absorbance values due to NBT+ at 560 nm in the absence and in presence of the complex. The IC50 value was also calculated. 2.5. Computational details All calculations were performed with the Gaussian 03 package of programs [27] and GaussView 03 [28] was used as the plotting tool for data visualization. The computations were performed by density functional theory (DFT) approach, the structures were fully optimized using hybrid-functional B3LYP with 6-311G** basis set (B3LYP/6311G**). The selection of DFT basis set was made considering the dimension of molecules and taking into account our previous experience on these systems [29]. The magnetic isotropic shielding tensors (σ) for 1H and 13C NMR were calculated using the standard GIAO (Gauge-Independent Atomic Orbital) [30] at B3LYP/6-311G** approach.

H2O

Fig. 1. UV–vis spectra of K2T21 in different solvents at 298 K ([K2T21] = 5 × 10−5 M).

effects on the value of the molar extinction coefficients (ε) [33] we can calculate fDK from the following equation: fDK × εDK / εKE = Abs(λDK) / [Abs(λDK) + Abs(λKE)] where fDK represents the fraction of molecules in the diketo form; εDK and εKE are the molar extinction coefficients of the diketo conformer at λDK and of the enol conformer at λKE respectively. The calculated f DK values for K2T21 are in the order H 2 O (0.63) N DMSO (0.61) N ACN (0.57) N MeOH (0.54); the other K2T compounds display the same trend. Polar solvents were found to stabilize the DK form of substituted β-diketones [34] and an increase of solvent dielectric constant value (εMeOH = 32.7, εACN = 37.5, εDMSO = 46.7, εΗ2Ο = 81.1,) fits with the increment observed in fDK. From the comparison of fDK in aqueous medium we may derive a general prevalence of DK for all compounds (fDK[K2T24] = 0.53; fDK [K2T33] = 0.57; fDK[K2T21] = 0.58; fDK[K2T23] = 0.63; fDK[K2T31] = 0.82]. In order to assess the acidity of the compounds and predict the most abundant species in physiological pH-range, pH-metric titrations were performed by means of UV–vis spectroscopy rather than by potentiometry, due to the low water solubility of the ligands (~10−4 M). As pH is increased from 4 to 9, simultaneous equilibria, involving keto-enolic tautomerism and acid dissociations, take place; the plotting of absorbance at λmax vs. pH fits with a titration trend characterized by an equivalent point, corresponding to keto–enol deprotonation. For K2T21 and K2T31 the dissociation of phenolic hydrogens also takes place. Overall protonation constants were calculated from spectrophotometric data; logβ and pKa values are summarized in Table 1. Previous study on curcumin demonstrated by NMR data the prevalence of the enolic tautomer in methanolic solution and attributed the first deprotonation step to enolic dissociation [13]. As reported in the present study, tautomeric equilibrium is strongly affected by the

3. Results and discussion 3.1. Spectroscopic characterization of the ligands As it was previously observed, K2T curcumin derivatives undergo keto–enol tautomerism [10] and the two absorption maxima at λ around 350 nm and 420 nm are attributed to the diketo (DK) tautomer, whose λ ranges from 270 nm to 350 nm, and to the keto–enol (KE) tautomer with λ in the range 420–450 nm, depending on the substituent on the aromatic ring [31,32]. Fig. 1 reports the UV–vis spectra of K2T21 in different solvents. A comparison of the absorbance values at λDK and λKE yields an estimation of the relative abundance of DK versus KE conformer. In the reasonable hypothesis that tautomerism has minor

Table 1 Logarithms of protonation constants (logβLH) and pKa values calculated with HypSpec [23] from spectrophotometric titrations performed at 25 °C, I = 0.1 M (NaNO3).

Logβ11 Logβ12 Logβ13 pKa1 pKa2 pKa3

K2T33

K2T23

K2T21

K2T24

K2T31

10.49(2) / / 10.49(2)

10.38(3) / / 10.38(4)

11.72(2) 20.67(2) 28.81(2) 8.14(3) 8.95(2) 11.72(2)

10.03(1)

10.06(1) 19.82(2) 27.63(2) 7.81(2) 8.76(2) 10.06(1)

/ 10.03(1)

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

3.2. Metal complexation

solvent, and the behavior in aqueous solution should be different from that observed for curcumin in MeOD-d4. Nevertheless, low water solubility of curcuminoids prevents their investigation in this medium by NMR. In addition electronic effects need to be considered; on one side the diketo group influences the dissociation of phenolic hydrogen; on the other substituents on aromatic rings exert an electronic effect which is experienced through the aliphatic backbone, as showed by the chemical shifts of H-5 and H-6 (Table S1 — Supplementary Information). For these reasons UV–vis data don't allow to assign unambiguously the pKa values. Nonetheless the pKa1 of K2T21 and K2T31 may be probably attributed to phenolic dissociation, strongly diminished with respect to unsubstituted phenol by the presence of the diketo moiety. The lower acidity of K2T21 with respect to K2T31 is ascribed to the formation of an intra-molecular hydrogen bond involving OH and OCH3 substituent on the aromatic ring as it was found in curcumin [35]. This experimental finding is consistent with data previously found for p-cumaric and ferulic acids (pKa of p-phenolic group 8.98 and 9.39 respectively; pKa of carboxylic group 4.36 and 4.52 respectively) [36].

H-6 *

41

3.2.1. NMR data 1 H NMR spectrum of K2T21 in MeOD-d4 shows the presence of two spectral patterns corresponding to the DK and KE tautomers (Fig. 2). By adding Ga3 + to the ligand solution at acidic pH (~ 5) (Fig. 3, panel A, bottom), a new set of down field shifted signals in slow chemical exchange in NMR time scale originate. They are attributed to the M:L 1:2 complexed KE tautomer, in which the dissociated keto–enol moiety is involved in metal coordination. The DK tautomer is not yet affected by metal addition. By adding Ga3+ till the 1:1 M:L molar ratio the formation of a complex species is indicated by the appearance of new downfield shifted signals, which are due to the 1:1 coordinated KE species; simultaneously the disappearance of the 1:2 M:L KE complex is observed, together with the shift of tautomeric equilibrium in favor of KE form, until the complete disappearance of the DK signals within few hours. Complete 1H and 13C NMR characterization of GaL+ 2 species is reported in Table 2. Fig. 3 — panel B shows the 1H 13C HMBC spectrum for Ga3+:K2T21 at the 1:1 molar ratio after the complete conversion of the DK into the KE complexed form, as stressed by the presence of a

H-2 *

H-5 * H-2 *

*

ppm 20 40

O

OH 6 3 4

HO

ROOC 1 OCH 3

60

2

O

OH OCH 3

O 6 3

80

5

HO OCH 3

ROOC 1

4

5

2

OH OCH 3

100 120 140 160 180

C-4 (KE) C-4 (DK)

200 8

7

6

5

4

3

2

ppm

Fig. 2. 1H–13C HMBC spectrum of K2T21 in MeOD-d4, typical DK (■) and KE (*) signals are showed, aromatic signals of the two tautomeric species are almost superimposed, and are not explicitly reported for clarity.

42

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

ppm

B

50

100

150

200 8

7

6

5

4

3

ppm

2

A Ga3+ : K2T21 1:2 (after 8h)

Ga3+ : K2T21 1:1

Ga3+ : K2T21 1:2

K2T21 8.0

7.5

7.0

6.5

6.0

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

ppm

Fig. 3. Panel A — 1H NMR titration of K2T21 in MeOD-d4 solution (bottom spectrum) with increasing amount of Ga3+, until reaching 1:1 M:L molar ratio (top spectrum). Vertical stacked lines evidence characteristic protons of the DK form (H-6, H-5 and H-2), which disappear as gallium ion is added. Panel B — 1H 13C HMBC spectrum of Ga3+:K2T21 at the 1:1 molar ratio in the same solvent system; ellipsis highlights the protons' correlations with C-4 of the enolic tautomer, while square box underlines the correlation 2JCH between H-2 and C-3.

cross peak at 107 ppm due to the scalar coupling between H-2 and C-3, and by the appearance of cross peaks between H-5 and H-6 with the keto–enol carbon at 185 ppm (C-4). There is no evidence of cross peaks corresponding to a carbonyl carbon around 190–200 ppm typical of a di-keto species. These data confirm the coordination of Ga3+ ion by

the keto-enolic moiety of K2T21, the other investigated ligands behave in the same way. Furthermore, no significant shifts were observed for aromatic protons of K2T21 and K2T31, these outcomes suggest the incapability of phenolic groups to bind gallium ion, thus excluding the formation of polymeric species.

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

43

Table 2 1 H and 13C chemical shifts (δ) of GaL+ 2 in MeOD-d4 at 298 K.

Ga(K2T21)+ 2 Ga(K2T23)+ 2 Ga(K2T24)+ 2 Ga(K2T31)+ 2 Ga(K2T33)+ 2

δ (1H) δ (13C) δ (1H) δ (13C) δ (1H) δ (13C) δ (1H) δ (13C) δ (1H) δ (13C)

1

2

3

4

5

6

7

8

9

10

11

12

/ 172.7 / 172.9 / 172.2 / 172.7 / 172.7

3.73 33.4 3.2 34.8 3.79 34.7 3.67 35.4 3.61 34.9

/ 107.5 / 107.6 / 108.1 / 106.6 / 107.0

/ 184.4 / 184.5 / 184.8 / 184.2 / 184.8

7.25 119.3 7.33 122.1 7.43 121.9 7.21 119.7 7.37 121.6

7.88 145.4 7.96 143.0 7.91 144.1 7.85 143.5 7.98 143.1

/ 127.2 / 136.9 / 134.2 / 126.8 / 135.1

7.30 110.8 7.16 112.7 7.38 111.5 6.85 115.6 7.64 127.3

/ 148.5 / 160.0 / 152.0 7.54 130.2 7.64 127.4

/ 150.3 6.98 115.5 / 142.3 / 160.1 7.44 130.1

6.88 115.3 7.36 129.5 7.11 111.8 7.54 130.2 7.64 127.4

7.21 123.9 7.16 120.6 7.29 121.7 6.85 115.6 7.64 127.3

3.2.2. UV–vis data The interaction of ligand molecules with metal ions was studied also by means of UV–vis spectroscopy in aqueous solution. By adding (Ga3+, Cu2+) solution to the free ligand under acidic conditions (pH ~5), the decrease in absorbance at λKE (420 nm) is due to metal complexation, and by plotting the absorbance vs. metal to ligand molar ratio, the formation of M:L 1:2 and 1:1 complex species is suggested. On increasing pH in a 1:2 M:L molar ratio system for all metal ions, a further decrease in absorbance at λDK is observed (Fig. 4) and by plotting the absorbance vs. pH a titration trend is found. The pKa calculated from the curves is in the range 6–8.0 for all the systems, with a lowering with respect to the free ligand (Fig. 4, inset). This confirms that the metal has great affinity for the ligand being able to anticipate the deprotonation of the keto-enolic proton. Copper containing systems behave in the same manner. The stability constants of the complexes were calculated taking into account the possible complexes in M:L 1:1, 1:2 for M = Cu2+ and also 1:3 molar ratio for M = Ga3+ but the refinement converged excluding the 1:3 M:L complex probably due to the great steric hindrance of three coordinated ligand molecules. Previous study on Ga3+ complexes involving β-diketo compounds with structure like “half curcumin” [14] has revealed the existence of ML3 complexes but in that case the steric hindrance of the ligands around the metal ion was significantly reduced. Also the mixed hydroxo-complexes were explored but without success at least in the investigated pH range (4–9),

nevertheless we do not exclude that at higher pH the dissociation of coordinated H2O molecules may take place. Polinuclear species were excluded according to NMR data. Table 3 reports the stability constants of the complex species and Fig. 5 shows the species distribution curves of M:L systems (M = Ga3+, Cu2+). The stability constants of Ga3+ complexes are of the same magnitude of those found in K2A compounds [29] and the greater stability of gallium complexes with respect to copper ones is also retained. At physiological pH, one of our major interest, the metal ions are totally complexed and in Ga3+ containing system the GaL2 species reaches 100% of formation for all curcuminoids, while in Cu2+ containing systems the CuL2 species prevails only in K2T21 and K2T31 which are the most acidic ligands. 3.3. Theoretical DFT calculations The theoretical calculation confirms that in vacuum the keto–enol tautomer is the most stable one as previously observed for curcumin [37] and other substituted curcuminoids [34]. Fig. 6 shows the optimized structures for different conformers of keto–enol tautomer of K2T21. Table 4 reports the total energy values (E) and relative energy values (ΔE) of the different conformers for both K2T21 and its monoanion originated from the dissociation of the enolic proton. The most stable conformer is c, for both the neutral and monoanionic forms, hence it was selected as a representative to construct metal

pH

Fig. 4. pH-metric spectrophotometric titration of Ga3+:K2T21 system in 1:2 molar ratio, inset shows the plot of absorbance vs. pH at λ 350 nm; [Ga3+] 2.5 × 10−5 M.

44

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

Table 3 Logarithms of formation constants of complex species with Cu2+ and Ga3+ at 25 °C, I = 0.1 M (NaNO3). The phenolic groups of K2T21 and K2T31 are considered nondissociated in the 4– 9 pH interval. K2T33

Logβ11 Logβ12

K2T23

K2T21

K2T24

K2T31

Cu2+

Ga3+

Cu2+

Ga3+

Cu2+

Ga3+

Cu2+

Ga3+

Cu2+

Ga3+

9.72(1) 15.02(1)

11.91(2) 20.05(2)

8.63(3) 13.31(4)

11.64(2) 18.31(3)

6.81(2) 11.10(2)

8.26(4) 13.61(6)

7.79(2) 13.44(2)

9.89(2) 16.76(2)

7.19(2) 11.18(2)

8.74(2) 12.36(3)

complexes. According to previous studies on Cu2+-curcumin complex [15,38], the 1:2 M:L chelating mode with a square planar arrangement was used as starting point of our study on Cu2+ complexes, while for Ga3 + an octahedral environment was reached by adding two water molecules to the 1:2 M:L Ga3+ complex [39]. The complex M(K2T21)2 × nH2O [n = 2 for M = Ga3+ and n = 0 for M = Cu2+] originates different structural isomers named A and B which are shown in Fig. 7. Table 4 reports their total energy (E), relative energy (ΔE), binding energy (BE) and relative binding energy (ΔΒE) compared to Cu2+ and Ga3+ curcumin complexes calculated using the same basis set. By comparing BE values of the same metal complexes, we can observe that A isomers are more stable than B, with population of 99.9% for Ga3+, 47.1% and 49.4 for A1 and A2 Cu2+ complexes respectively. Addicoat et al. [38] indicated that steric hindrance has a destabilizing influence on Cu(II)(curcumin)2 complex, correlated with the dihedral angle ϕC9-C3-C3′-C9′ involving the two carbon atoms bearing the methoxy substituent (C9, C9′) of the two coordinated molecules on the same side with respect to central C atoms (C3, C3′) (Fig. 7). Our data (Table 4) confirm that steric hindrance is the driving force in destabilizing metalcomplexes. In fact in B isomer the bulky substituent on central carbon atom diminishes the binding energy of the complex even if the dihedral angle is smaller than that of A isomer. In our case the higher stability of A complex is achieved by a distortion from the planarity of the curcumin

ML

skeleton that gives also an inequality of the two dihedral angles. Gallium complexes have greater BE than copper ones, demonstrating their higher stability confirmed by the experimentally obtained logK values (Table 3). This trend was observed also in metal complexes with human serum transferrin [40]. Comparing K2T metal complexes with those of curcumin we notice that Ga 3 + complex is stabilized (ΔBE = 11.09 Kcal/mol) while Cu2 + complex is destabilized (ΔBE = − 2.03 Kcal/mol). This behavior is attributed to the presence of the ester C_O groups, each of which forms two strong hydrogen bonds with coordinated water molecules in A conformer of Ga 3 + complex (C_O⋯H distance: 1.733 Å; C\O⋯H angle 174.20°). This effect is minor in B conformer of Ga3+ complex where the same water molecule forms weaker hydrogen bonds with the two C_O groups (C_O⋯H distances: 1.795 and 1.782 Å; C\O⋯H angles: 177.29° and 177.65°). In contrast, Cu2 +-complexes display a square planar geometry which can't be stabilized by intra-molecular hydrogen bonds. The GIAO (Gauge-Independent Atomic Orbital) isotropic magnetic shielding tensors (σcalc) were calculated for Ga3+ complexes; bearing in mind the experimental molecular symmetry of K2T21, the averaged values of σcalc (σ calc ) between the A and B conformers were computed on the basis of relative species populations. 1H and 13C experimental chemical shift assignments were supported by 2D COSY, HSQC and HMBC experiments. Plotting the experimental 1H and 13C chemical

ML2

ML

ML ML2

ML ML

MOH

ML

M ML2

ML2

ML M

M

ML2

ML

ML2

ML2 ML2

MOH M M

Fig. 5. Species distribution curves for M:L systems [M]tot = 1 × 10−3; [L]tot = 4 × 10−3. Solid lines refer to Cu2+ species, while dotted lines represent Ga3+ species; complex charges are omitted for clarity.

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

a

bI

bII

c Fig. 6. The final optimized structures of K2T21 conformers at B3LYP/6-311G** level.

shifts (δexp) versus σ calc a linear relationship is obtained (Fig. 8) with a good correlation between experimental data and calculated isotropic shielding constants for both 1H and 13C. In order to predict antioxidant properties, in particular the ability to scavenge DPPH (2,2-diphenyl-1-picrylhydrazyl) radical and superoxide anion, we calculated the O\H bond dissociation energy (BDE), the O\H proton dissociation enthalpy (PDE), the electron affinity (EA) and

Table 4 B3LYP/6-311G** total energy E (a.u.) and relative energy ΔE (kcal/mol) for different conformers of neutral K2T21 and for respective monoanions. a, bI, bII and c conformers refer to Fig. 7. K2T21 neutral molecule

a bI bII c

K2T21 monoanion

E (a.u.)

ΔE (kcal/mol)

E (a.u.)

ΔE (kcal/mol)

−1648.7628 −1648.7641 −1648.7629 −1648.7641

0.84 0.02 0.75 0

−1648.1783 −1648.1805 −1648.1790 −1648.1810

1.69 0.32 1.29 0

45

ionization potential (IP). These theoretical parameters have been recognized appropriate for measuring the H atom donating, proton donating, electron accepting and electron donating abilities of antioxidants [15]. The lower the parameters (BDE, PDE, EA and IP) are the higher the antioxidant activities. Table 5 reports the above mentioned parameters for K2T21 and for the more stable A complexes in comparison with curcumin ones. Curcumin and K2T21 show almost the same value of BDE although K2T21, in previous experimental studies, showed to display less efficacy in scavenging the DPPH radical [10]. This finding suggests that, in addition to hydrogen dissociation ability, showed by BDE, the stabilization by resonance of the formed radical plays a fundamental role in radical scavenging properties. In fact the prevalence in K2T21 of the DK form in water solution probably reduces the charge delocalization, destabilizing the formed radical thus diminishing its activity against DPPH radical. Since, an enhancement of charge delocalization could be expected after complexation, an improvement of the radical scavenging ability of K2T21 up to the curcumin metal complex level, could be expected too. Surprisingly, the BDE value was higher in metal complexes than in the corresponding free ligands (Table 6), therefore a moderately lower ability to scavenge DPPH radical by metal complexes may be derived. This effect was more evident for Cu2 + than Ga3+ complexes. On the other hand, the PDE value of Ga3+ complexes was significantly lower than that of the free ligands suggesting that the proton-transfer-associated superoxide-scavenging activity is enhanced after Ga3+ binding while, for Cu2+, appreciable effect was not showed. Since the presence of electron-donating groups in the molecule causes a decrease of BDE and an increase of PDE, while the electronwithdrawing groups have an opposite effect, we can conclude that the metal ion exhibits an electron-withdrawing effect in the present case, although Cu2+ in a minor extent. In order to predict SOD-like activity of metal complexes we calculated the electron affinity (EA): the lower the EA, the higher the SODlike activity. According to results reported in Table 5 EA values of Ga3 + complexes are lower than those of the corresponding free ligands although they are higher than those of known SOD mimics (~− 180 kcal/mol), nevertheless both Cu2 + and Ga3 + complexes have much higher binding energies than those of some previously synthesized SOD mimics (BE 460–466 kcal/mol) [41]. This implies that K2T complexes are more stable, a feature which is of great significance in pharmaceutical applications. In polar solvents the ROS-scavenging mechanism is mainly related to electron transfer process [42] and the ionization potential (IP) allows to estimate the electron donating ability of metal complexes: the lower the IP, the more active the electron donor. As shown in Table 5 only Cu2+ complexes display IP values lower than those of the ligands therefore we may suggest that copper complexes are more active than the parent compounds in ROS-scavenging mechanism. 3.4. Superoxide scavenging ability Although there are many methods, direct and indirect, to determine the SOD-like activity for small molecules, and indirect ones are not the best choice in terms of reliability of the results, we decided to use NBT+ 2 with superoxide radical generated by xanthine/xanthineoxidase system [26], as this method was used to determine SOD activity of Cu(curcumin)2 complex [43]. Fig. 9 reports the percentage of inhibition of NBT+ against metal complex concentration; the value of IC50, i.e., the concentration of the complex required to reduce the absorbance of NBT+ to half of its initial value was estimated at 22 ± 1 μM for Ga(K2T21)2 and 78 ± 2 μM for Cu(K2T21)2 complexes. The value of Ga3 + complex is not much dissimilar to that previously found for Cu(curcumin)2 complex [43] and for Cu-complexes with SOD activity [41]. The results here reported confirm the trend found in theoretical calculation: the lower is the EA value the higher is the predicted SOD activity. The experimental data support this prediction, in fact Cucomplexes with SOD-like activity showed EA ~− 180 kcal/mol and

46

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

[Ga(K2T21)2(H2O)2]+ (A)

[Ga(K2T21)2(H2O)2]+ (B)

H O C Ga; Cu

[Cu(K2T21)2] (A1)

[Cu(K2T21)2] (A2)

[Cu(K2T21)2] (B)

Fig. 7. The final optimized structures of Ga3+ and Cu2+ complexes at B3LYP/6-311G** level.

IC50 b10 μM [41] while Ga(K2T21)2 complex exhibits EA at − 79.97 kcal/mol with IC50 22 ± 1 μM and Cu(K2T21)2 complex exhibits EA at −14.85 kcal/mol with IC50 78 ± 2 μM. 4. Conclusions The introduction of the substituent [CH2COOC(CH3)3] on the central carbon atom of the β-diketo moiety of curcumin leads to the formation of K2T derivatives that in previous studies showed to have biological activity comparable to or, in some cases, better than curcumin itself [10]. In this study we have investigated the binding capacity of these molecules towards gallium(III) and copper(II), two metals which are very important from a biological point of view. Although the predominant tautomer of these ligands in aqueous solution is the DK one, complexation with the metal ion shifts the equilibrium towards the KE tautomer as demonstrated by NMR data. The metal ion is able to promote the dissociation of the enolic proton and the formed complex remains stable in a wide range of pH, confirming the ability of these ligands to act as metal chelators in physiological conditions. DFT calculations on K2T21 and its

Cu2+ and Ga3+ complexes allow to predict their structure and to confirm experimental data. A good correlation was found between the experimental 1H and 13C chemical shifts and the calculated GIAO isotropic magnetic shielding constants, in addition the predicted order of complexes stability is retrieved in solution as confirmed by logβ values calculated from UV–vis spectroscopic data. Furthermore the calculated BDE, PDE, EA and IP values for K2T21 and its complexes compared to curcumin ones shed new light on the mechanism of radical scavenging of these molecules, predicting the antioxidant properties of metal complexes. In fact the BDE values of the ligands, compared to previously reported antioxidant activity against DPPH radical, suggest that charge delocalization plays an important role in determining the scavenging ability. On the basis of BDE value we may predict the antioxidant properties of metal complexes: in particular metal complexation is expected to maintain almost unaffected radical scavenging ability of the ligands while it seems to enhance the antioxidant properties. Therefore these complexes, in particular gallium(III) ones, look very promising as superoxide-scavengers suggesting their possible use as SOD mimics. In addition the good affinity

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

47

Table 6 B3LYP/6-311G** calculated O\H (phenolic) bond dissociation energy BDE (kcal/mol), proton dissociation energy PDE (kcal/mol), electron affinity EA (kcal/mol) and ionization potential IP (kcal/mol) for metal complexes and free ligands.

Ga(K2T21)2(A) Ga(curcumin)2 Cu(K2T21)2(A2) Cu(curcumin)2 K2T21e Curcumin

O\H BDEa

O\H PDEb

EAc

IPd

89.85 90.10 107.46 110.15 86.29 86.76 86.83

299.02 295.44 347.35 346.90 343.55 346.97 344.80

−79.97 −84.70 −16.13 −14.85 −20.03

180.95 184.32 137.75 142.67 144.87

−20.96

147.23

TER = total electronic energy of the radical derived from H-abstraction from the phenolic hydroxyl. TEH = total electronic energy of H-atom (−0.500273 Eh). TEC = total electronic energy of metal complex. TEA = total electronic energy of anion derived from proton abstraction of phenolic hydroxyl. TEH+ = total electronic energy of proton (0.0 Eh). TEAR = total electronic energy of anion radical derived from electron-accepting reaction. TECR = total electronic energy of cation radical derived from electron-donating reaction. a O\H BDE = TER + TEH − TEC. b O\H PDE = TEA + TEH+ − TEC. c EA = TEAR − TEC. d IP = TECR − TEC. e O\H BDE and O\H PDE values are calculated for the two unequivalent free radicals formed from H abstraction.

Fig. 8. Plot of 1H and 13C δexp versus σcalc for K2T21.

of K2T series for Cu2 +, which is implicated in the self-assembly of amyloid-β (Aβ) peptides [44], hints their potential development in therapeutic applications. Abbreviations

AD Aβ BDE COSY DFT

DK DPPH EA HSQC HMBC IP KE NBT2+ PDE ROS SOD GIAO

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.jinorgbio.2014.06.002.

Alzheimer's disease β-amyloid O\H bond dissociation energy correlation spectroscopy density functional theory

[GaL2]+ ( M) 100

Table 5 B3LYP/6-311G** total energy E (a.u.), relative energy ΔE (kcal/mol), binding energy BE (kcal/mol), relative binding energy ΔBE (kcal/mol) and ϕC9-C3-C3′-C9′ for metal complexes with K2T21 and curcumin. A and B conformers are referred to Fig. 8. ΔE

BEa

ΔBEb

Ga(K2T21)2 (A)

−5372.136765

0.00

1358.82

11.09

Ga(K2T21)2 (B)

−5372.130552

3.90

1354.92

7.19

Ga(curcumin)2

−4601.827558

1347.73

0.00

Cu(K2T21)2(A1)

−4936.699950

0.03

705.81

−2.06

Cu(K2T21)2(A2)

−4936.699995

0.00

705.84

−2.03

Cu(K2T21)2(B)

−4936.697503

1.56

704.28

−3.59

Cu(curcumin)2

−4166.411704

707.87

0.0

ϕ 16.12 −17.23 8.89 10.53 8.71 8.83 52.23 32.97 47.18 27.70 −0.44 −0.49 −42.56 −42.57

BE = TES − TEC where TES = sum of the total energy of the bare metal ion and the relative isomer of the complexing anion; TEC = total electronic energy of complexes. b ΔBE = BE[M(K2T21)2] − BE[M(curcumin)2].

0

5

10

15

20

25

30

80

% Inhibition

E

a

diketo tautomer di(phenyl)-(2,4,6-trinitrophenyl)iminoazanium radical calculated electronic affinity heteronuclear single quantum coherence heteronuclear multiple bond coherence ionization potential keto–enol tautomer nitro blue tetrazolium O\H proton dissociation enthalpy reactive oxygen species superoxide dismutase Gauge-Independent Atomic Orbital

60 40

[CuL2] [GaL2]+

20 0 0

20

40

60

80

100

120

140

[CuL2] ( M) Fig. 9. Percentage of inhibition (%I) of superoxide radical formed by xanthine/xanthine oxidase enzyme by different concentrations of complexes assayed by NBT+ absorption at 560 nm in presence of 1:2 Cu 2 + -K2T21 complex (■) and in presence of 1:2 Ga3+-K2T21 complex (▲).

48

E. Ferrari et al. / Journal of Inorganic Biochemistry 139 (2014) 38–48

Acknowledgments We are thankful to the “Laboratorio di Calcolo Scientifico Avanzato Interdipartimentale dell'Università degli Studi di Modena e Reggio Emilia” for computing facilities. We are grateful to the “Centro Interdipartimentale Grandi Strumenti — C.I.G.S.” of the University of Modena and Reggio Emilia and to the “Fondazione Cassa di Risparmio di Modena” for supplying NMR spectrometer. The authors are also thankful to the Arcispedale Santa Maria Nuova-IRCSS (Reggio Emilia — Italy). References [1] K. Kumar, A.K. Rai, J. Pharm. Res. 5 (2012) 254–256. [2] G. Bar-Sela, R. Epelbaum, M. Schaffer, Curr. Med. Chem. 17 (2010) 190–197. [3] P. Anand, C. Sundaram, S. Jhurani, A.B. Kunnumakkara, B.B. Aggarwal, Cancer Lett. 267 (2008) 133–164. [4] D. Yanagisawa, N. Shirai, T. Amatsubo, H. Taguchi, K. Hirao, M. Urushitani, S. Morikawa, T. Inubushi, M. Kato, F. Kato, K. Morino, H. Kimura, I. Nakano, C. Yoshida, T. Okada, M. Sano, Y. Wada, K.N. Wada, A. Yamamoto, I. Tooyama, Biomaterials 31 (2010) 4179–4185. [5] H. Han-Chang, X. Ke, J. Zhao-Feng, J. Alzheimers Dis. 32 (2012) 981–996. [6] F.-L. Yen, T.-H. Wu, C.-W. Tzeng, L.-T. Lin, C.-C. Lin, J. Agric. Food Chem. 58 (2010) 7376–7382. [7] Y. Zhang, X. Qu, Q. Chen, H. Zhou, Adv. Mater. Res. (2012) 1857–1860. [8] C.Y. Kim, N. Bordenave, M.G. Ferruzzi, A. Safavy, K.H. Kim, J. Agric, Food Chem. 59 (2011) 1012–1019. [9] V. Kumar, B. Gupta, G. Kumar, M.K. Pandey, E. Aiazian, V.S. Parmar, J. Kumar, A.C. Watterson, J. Macromol. Sci. A Pure Appl. Chem. 47 (2010) 1154–1160. [10] E. Ferrari, F. Pignedoli, C. Imbriano, G. Marverti, V. Basile, E. Venturi, M. Saladini, J. Med. Chem. 54 (2011) 8066–8077. [11] Z. Kovacevic, S.D. Kalinowski, B.D. Lovejoy, Y. Yu, S.Y. Rahmanto, C.P. Sharpe, V.P. Bernhardt, R. Richardson, Curr. Top. Med. Chem. 11 (2011) 483–499. [12] K. Zandi, E. Ramedani, K. Mohammadi, S. Tajbakhsh, I. Deilami, Z. Rastian, M. Fouladvand, F. Yousefi, F. Farshadpour, Nat. Prod. Commun. 5 (2010) 1935–1938. [13] M. Borsari, E. Ferrari, R. Grandi, M. Saladini, Inorg. Chim. Acta 328 (2002) 61–68. [14] B. Arezzini, M. Ferrali, E. Ferrari, C. Frassineti, S. Lazzari, G. Marverti, F. Spagnolo, M. Saladini, Eur. J. Med. Chem. 43 (2008) 2549–2556. [15] L. Shen, H.Y. Zhang, H.F. Ji, J. Mol. Struct. THEOCHEM 757 (2005) 199–202. [16] H.M. Mandy, D.T.P. Leung, F.L. Stephen, W.K. Tak, Phys. Chem. Chem. Phys. 14 (2012) 13580–13587. [17] J. Rajesh, M. Rajasekaran, G. Rajagopal, P. Athappan, Spectrochim. Acta A Mol. Biomol. Spectrosc. 97 (2012) 223–230. [18] M. Yoshino, M. Haneda, M. Naruse, H.H. Hatay, R. Tsubouchi, S.L. Qiao, W.H. Li, K. Murakami, T. Yokochi, Toxicol. in Vitro 18 (2004) 783–789. [19] K. Sakano, S. Kawanishi, Arch. Biochem. Biophys. 405 (2002) 223–230. [20] C.S. Atwood, R.D. Moir, X. Huang, R.C. Scarpa, N.M.E. Bacarra, D.M. Romano, M.A. Hartshorn, R.E. Tanzi, A.I. Bush, J. Biol. Chem. 273 (1998) 12817–12826.

[21] Y. Yoshiike, K. Tanemura, O. Murayama, T. Akagi, M. Murayama, S. Sato, X. Sun, N. Tanaka, A. Takashima, J. Biol. Chem. 276 (2001) 32293–32299. [22] X.-T. Mei, D.-H. Xu, S.-K. Xu, Y.-P. Zheng, S.-B. Xu, Food Chem. Toxicol. 60 (2013) 448–454. [23] P. Gans, A. Sabatini, A. Vacca, Talanta 43 (1996) 1739–1753. [24] W. Wilker, D. Leibfritz, R. Kersebaum, W. Bermel, Magn. Reson. Chem. 31 (1993) 287–292. [25] A. Bax, M.F. Summers, J. Am. Chem. Soc. 108 (1986) 2093–2094. [26] J. Robak, R.J. Gryglewski, Biochem. Pharmacol. 37 (1988) 837–841. [27] Gaussian 03, Revision D.01, M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, J. A. Montgomery, Jr., T. Vreven, K. N. Kudin, J. C. Burant, J. M. Millam, S. S. Iyengar, J. Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G. A. Petersson, H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J. E. Knox, H. P. Hratchian, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, P. Y. Ayala, K. Morokuma, G. A. Voth, P. Salvador, J. J. Dannenberg, V. G. Zakrzewski, S. Dapprich, A. D. Daniels, M. C. Strain, O. Farkas, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. V. Ortiz, Q. Cui, A. G. Baboul, S. Clifford, J. Cioslowski, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, C. Gonzalez, and J. A. Pople, Gaussian, Inc., Wallingford CT (2004) [28] R. Dennington II, T. Keyth, J. Millam, K. Eppinnett, W.L. Hovell, R. Gilliand, Gaussview, Version 3.0, Semichem, Inc., Shawnee Mission, KS, 2003. [29] E. Ferrari, M. Asti, R. Benassi, F. Pignedoli, M. Saladini, Dalton Trans. 42 (2013) 5304–5313. [30] K. Wolinski, J.F. Hinton, P.J. Pulay, J. Am. Chem. Soc. 112 (1990) 8251–8260. [31] L. Nardo, A. Maspero, M. Selva, M. Bondani, G. Palmisano, E. Ferrari, M. Saladini, J. Phys. Chem. A 116 (2012) 9321–9330. [32] M. Caselli, E. Ferrari, C. Imbriano, F. Pignedoli, M. Saladini, G. Ponterini, J. Photochem. Photobiol. A 210 (2010) 115–124. [33] J. Toullec, in: Z. Rappoport (Ed.), The Chemistry of Enols, Wiley J. & Sons, New York, 1990, pp. 323–398. [34] E. Ferrari, M. Saladini, F. Pignedoli, F. Spagnolo, R. Benassi, New J. Chem. 35 (2011) 2840–2847. [35] R. Benassi, E. Ferrari, S. Lazzari, F. Spagnolo, M. Saladini, J. Mol. Struct. 892 (2008) 168–176. [36] F.Z. Erdemgil, S. Sanli, N. Sanli, G. Ozkan, J. Barbosa, J. Guiteras, J.L. Beltran, Talanta 72 (2007) 489–496. [37] K. Balasubramanian, J. Agric, Food Chem. 54 (2006) 3512–3520. [38] M.A. Addicoat, G.F. Metha, T.W. Kee, J. Comput. Chem. 32 (2011) 429–438. [39] M. Xu, X. Dou, Y. Bu, Y. Zhang, Chem. Phys. Lett. 537 (2012) 101–106. [40] T. Sakajiri, H. Yajiman, T. Yamamura, ISRN Biophys. (2012), http://dx.doi.org/10. 5402/2012/124803 (Article ID 124803, 5 pages). [41] H.-F. Ji, H.-Y. Zhang, Bioorg. Med. Chem. Lett. 15 (2005) 21–24. [42] H.Y. Zang, Curr. Comput. Aided Drug Des. (2005) 257–273. [43] A. Barik, B. Mishra, A. Kunwar, R.M. Kadam, L. Shen, S. Dutta, S. Padhye, A.K. Satpati, H.-Y. Zhang, K.I. Priyadarsini, Eur. J. Med. Chem. 24 (2007) 431–439. [44] P. Faller, C. Hureau, O. Berthoumieu, O. Inorg. Chem. http://dx.doi.org/10.1021/ ic4003059.