Structural and Antioxidative Features of Chlorogenic Acid

This work contributes to the structure clarification of chlorogenic acid (5-O-caffeoylquinic acid, 5CQA) by comparing the experimental and simulated IR, Raman, 1H-NMR, 13C-NMR, and UV-vis spectra. The lowest-energy conformers in the gas-phase and solution were used for all calculations. Very good agreement between all experimental and simulated spectra indicates correct arrangement of the atoms in the 5CQA molecule. In addition, the bond dissociation enthalpies, proton affinities, electron transfer enthalpies, ionization potentials, and proton dissociation enthalpies for 5CQA were used for thermodynamic consideration of the major antioxidative mechanisms: HAT (Hydrogen Atom Transfer), SPLET (Sequential Proton-Loss Electron-Transfer), and SET-PT (Single Electron Transfer – Proton Transfer). It was found that HAT may be the predominant mechanism in nonpolar solvents, while HAT and SPLET are competitive pathways in polar media. All quantum-chemical calculations were carried out by means of the MN12-SX method. Its performance is similar to those of the B3LYPD2, B3LYP-D3, and M06-2X functionals.


INTRODUCTION
HLOROGENIC acid (5-O-caffeoylquinic acid, 5CQA), an ester formed between caffeic and quinic acids (Figure 1), is a natural polyphenol that can be isolated from various fruits and vegetables.This compound is an important secondary metabolite with various roles in plants.It has been demonstrated that enhanced levels of 5CQA in transgenic tomato plants improve protection from the UV radiation, and increase microbial resistance. [1,2]In addition, it has been shown that 5CQA acts as pest resistance factor in ornamental plants. [3][13][14] However, HAT (eq.1), SPLET (eqs 2 and 3), and SET-PT (eqs 4 and 5) are considered as major antioxidative mechanisms, as they are most often obeyed by various polyphenols.
ArOH ArO e In the reactions (1)-( 5) ArO  , ArO  , and ArOH + represent the free radical, anion, and radical cation of the phenolic compound ArOH.The corresponding reaction enthalpies are called: bond dissociation enthalpy (BDE), proton affinity (PA), electron transfer enthalpy (ETE), ionization potential (IP), and proton dissociation enthalpy (PDE), respectively.5CQA is a white powder for which the X-ray powder diffraction has not been carried out.So, its structural features have not been completely elucidated.[17][18] A preferential conformation of 5CQA in methanol solution has recently been determined on the basis of the 1 H-NMR, 13 C-NMR, DQF-COSY, z-filtered TOCSY, ROESY, phase-sensitive HMBC, and HSQC experiments. [17]V-vis absorption spectroscopy has been used to investigate complex formation of some metal ions with 5CQA. [19,20]In our recent work, we used different spectroscopic methods (IR, Raman, UV-vis, 1 H-and 13 C-NMR) in order to contribute to clarification of the 5CQA structure. [21]In addition, we performed a comprehensive study of the antioxidative activity of 5CQA by focusing on the thermodynamics of the three major antioxidative mechanisms: HAT, SPLET, and SET-PT.It was shown that HAT may be the predominant mechanism in nonpolar solvents, while HAT and SPLET are competitive pathways in polar media. [22]ll these tasks were realized by performing necessary quantum chemical calculations at the B3LYP-D2, B3LYP-D3 and M06-2X levels of theory.It is well-known that numerous density functionals are nowadays available.In this work we focus our attention towards the MN12-SX method.This method is a screened exchange hybrid meta-NGA (nonseparable gradient approximation) functional that uses a finite amount of the Hartree-Fock (HF) exchange at short-range, and none in the large-range limit. [23]The aim of this approximation is to reduce high computational costs resulting from the HF exchange.Our aim is to examine its performance in the investigations devoted to the spectral and antioxidative properties of 5CQA.Our additional goal is to carry out a comparison of the results obtained with different density functionals.

MATERIALS AND METHODS
The vibrational spectra, NMR spectra in DMSO-d6 with TMS as the internal standard, and UV-vis spectrum in methanol of 5CQA were recorded in a manner described in our related work. [21]There, a detailed conformational analysis of 5CQA has been performed using the gas-state B3LYP/6-31G(d) calculations.Twenty conformers of the lowest total energy have been reoptimized in the gas-state and solution using the B3LYP-D2, B3LYP-D3 and M06-2X functionals in combination with the 6-311+G(d,p) and 6-311++G(d,p) basis sets. [21,22]In the present work we used the MN12-SX/6-311+G(d,p) theoretical model to reoptimize the same twenty conformers of 5CQA.The calculations revealed the gasphase geometries, as well as those in the DMSO (dielectric constant ε = 46.8260)and methanol (ε = 32.6130)solutions.These solvents were selected to mimic the conditions of experimental measurements, and their influence was taken into account by applying the CPCM polarizable continuum solvation model. [24]Frequency calculations were included; there were no imaginary frequencies.A comparison of the free energy values resulting from the four methods: B3LYP-D2, B3LYP-D3, M06-2X, and MN12-SX, revealed that there are only minor differences in the orders of conformer stability in all three media.Furthermore, the most stable conformations in the gas-phase and in both solutions for all four methods are exactly the same.The optimized geometries of the most stable conformations of 5CQA obtained with the MN12-SX method in the gas-phase and DMSO solution are presented in Figure 2.
The gas-phase geometry of 5CQA was used to simulate the IR and Raman spectra, while its solvated geometry was used to predict the NMR and UV-vis spectra.The vibrational modes of 5CQA were assigned on the basis of the potential energy distribution (PED) analysis [25] using the VEDA 4 software. [26]The gauge independent atomic orbital (GIAO) method was used for prediction of the The free radicals, anions, and radical cation of 5CQA were issued from the solvated geometry.The structures of all species in the solvents of different polarity: benzene (ε = 2.2706), methanol, and water (ε = 78.3553),were obtained by full optimizations and frequency calculations.Taking into account that free radicals, including hydrogen atoms, were considered, these calculations were carried out by employing the MN12-SX/6-311++G(d,p) level of theory in combination with the CPCM solvation model.The so-obtained enthalpies were used to compute the BDE, PA, ETE, IP, and PDE values, required for a description of the HAT, SPLET, and SET-PT mechanisms of 5CQA in the three solvents.
Gaussian 09 program package was employed for all quantum chemical calculations. [27]

RESULTS AND DISCUSSION
As in the case of B3LYP-D3 and M06-2X, [21] the MN12-SX geometries of 5CQA in DMSO and methanol are mutually very similar.The solvated structure is in perfect agreement with the finding of Forino et al. based on a detailed NMR study. [17]This structure is characterized with the O4-H4•••O10' hydrogen bond.The only significant difference of the gas-phase geometry is that it is characterized with the O4-H4•••O9' hydrogen bond (Figure 2).In both structures there are very strong O8-H8•••O1 and O1-H1•••O3 hydrogen bonds that influence the orientation of all hydroxyl groups in the quinic moiety of 5CQA.

Spectroscopic and Structural Properties of Chlorogenic Acid
Due to the fact that 5CQA possesses six hydroxyl groups, which have a significant impact on the vibrational spectra appearance, these spectra are very complex with in total 123 spectral modes.The vibrational scaling factors for the number of different computational methods can be found in literature. [28]However, to our best knowledge, the scaling factor for the MN12-SX/6-311+G(d,p) theoretical model has not been reported.In present study, the scaling factor for this method was determined on the basis of the experimental data for the IR spectrum using the least squares method, and it amounts to 0.923.The calculated values of the wavenumbers were scaled and assigned to the corresponding experimental values.The IR and Raman spectra of 5CQA are depicted in Figures 3 and 4, whereas the most important vibrational modes of the compound are presented in Table 1.
The quality of the linear correlation between the experimental and calculated wavenumbers was evaluated by means of three descriptors: the correlation coefficient (R), average absolute error (AAE), and average relative error (ARE).The R value for the IR and Raman spectra are identical and amount to 0.9990.The AAE and ARE values for the IR spectrum are equal to 24.3 cm −1 and 1.8 %, whereas for the Raman spectrum AAE and ARE amount to 23.2 cm −1 and 1.8 %.The AAE and ARE values obtained with the MN12-SX method for both spectra are very similar to those obtained with B3LYP-D3 and M06-2X, indicating reasonable capability of the applied method to reproduce this kind of spectra.It should be pointed out that the simulated vibrational spectra refer to an isolated molecule of 5CQA, implying that intermolecular interactions, including hydrogen bonds, are neglected.Certainly, this fact is the most important source of slight disagreement between the experimental and calculated band positions.
Very massive and intense overlapping bands in the IR spectrum, which appear in the region of 4000-2000 cm −1 , were assigned to the different modes of the O-H vibrations.The high frequency region is also characterized by the C-H stretching modes of acyclic chain and quinic moiety.The bands of the medium and strong intensities at 1732 (IR) and 1690 cm −1 (IR) were assigned to the C-O stretching modes of the acyclic chain and quinic moiety.In addition, the bands of the medium and strong intensities at 1639 (IR), 1632 (R), 1604 (R), 1602 (IR) and 1443 cm −1 (IR, R) were mostly assigned to the C-O stretching modes of the acyclic chain and benzene moiety.
Due to the fact that the chemical shifts in the simulated 13 C-NMR spectrum were underestimated, and in the 1 H-NMR spectrum were overestimated, the calculated values were scaled.As in the case of the vibrational spectra, the scaling factors were determined by means of the least squares method, and they amount to 1.066 and 0.928 for 13 C-NMR and 1 H-NMR.The obtained results are presented in Table 2.A comparison between the experimental and calculated chemical shifts shows that the correlation coefficients for 13 C-NMR and 1 H-NMR are equal to 0.997 and 0.996, whereas the AAE values are equal to 2.26 and 0.17 ppm, respectively.
By comparing the results obtained with the B3LYP-D3 and M06-2X methods [21] with the results shown in Table 2, it can be concluded that the MN12-SX functional achieves the best agreement with the experimental NMR spectra of 5CQA.Yet, the deviation of the calculated chemical shifts from the experimental values is notable for some carbon atoms: C2', C6' and C8'.This occurrence is a consequence of facile rotation of the benzene moiety around the single C1'-C7' bond.Very good agreement between the experimental and calculated chemical shifts in the 1 H-NMR spectrum indicates decreased flexibility of the quinic moiety, which is a consequence of the directed hydrogen bonds.
The experimental UV-vis spectrum of 5CQA in methanol consists of four bands at 329, 298, 244, and 218 nm. [21]The simulated spectrum, obtained at the TDDFT/MN12-SX/6-311+G(d,p) level of theory, shows only one intensive absorption band at 328 nm.This band corresponds to a  * electronic transition from the HOMO to the LUMO (Figure 5).There is a large sharing region between the HOMO and LUMO of 5CQA, as both orbitals are delocalized over the caffeic moiety.These findings, as well as the small HOMO-LUMO gap of 0.13564 au, explain the pronounced intensity of this band.Interestingly, this analysis of the HOMO→LUMO transition, based on the Kohn-Sham orbitals, agrees very well with that based on the NLMO clusters. [29]Namely, the HOMO cluster of 5CQA is delocalized over the caffeic moiety, whereas the LUMO cluster is delocalized over the acyclic chain (more precisely -over the C9'-O9' carbonyl group and adjacent carbon and oxygen atom). [21]ble 1.Characteristic experimental and calculated wavenumbers () in the vibrational spectra of chlorogenic acid Mode assignment IR Raman MN12-SX Exp. [21]/ cm −1 Calcd./ cm (a) b, a, and q denote benzene moiety, acyclic chain, and quinic moiety, respectively (b) vs, s, m, and w stand for very strong, strong, medium, and weak vibrations Table 2. Experimental and calculated NMR properties of chlorogenic acid in DMSO.Chemical shifts for the 13 C-and 1 H-NMR spectra are given in ppm C Calcd.
Exp. [21] 7 Although the agreement between the calculated and experimental wavelengths of the first band is excellent, the other bands were not reproduced.While B3LYP-D3 was very successful, and M06-2X underestimated all four bands, [21] the MN12-SX functional failed to reproduce the entire UV-vis spectrum of 5CQA.

Antioxidative Activity of Chlorogenic Acid: Thermodynamic Approach
In order to assume which antioxidative mechanism -HAT, SPLET or SET-PT, is a predominant pathway in 5CQA, we determined the corresponding thermodynamic parameters: BDE, PA, ETE, IP, and PDE.These parameters were calculated as the enthalpies of the reactions (1)-( 5) in benzene, methanol, and water; namely: For this purpose, the enthalpies of the parent molecule, its free radicals in the positions 3' and 4', anions in the positions 3' and 4', and radical cation were calculated at the MN12-SX/6-311++G(d,p) level of theory.The enthalpy values of the solvated proton and electron in benzene (−898.7 and −14.4 kJ mol −1 ), methanol (−1064.6 and −78.3 kJ mol −1 ), and water (−1052.7 and −98.8 kJ mol −1 ) were taken from literature. [30]The calculated reaction enthalpies are collected in Table 3.By comparing these enthalpies one can suppose which antioxidative mechanism of the 5CQA pre-vails.Namely, the mechanism which is characterized with the smallest value of the reaction enthalpy is supposed to be the operative reaction pathway.
A careful inspection of Table 3 shows that 5CQA will not obey SET-PT mechanism in any solvent, due to the highly endothermic first step of this mechanism.As expected, the BDE values in all three solvents are mutually very similar, indicating that the HAT mechanism is almost independent on the solvent polarity.In benzene the PA and ETE values are larger than the BDE values, so one can assume that the HAT mechanism will be predominant in nonpolar solvents.In polar solvents the PA values are smaller than the BDEs, while the ETE values are larger, but still comparable to the BDE values.Thus, one can suppose that the HAT and SPLET mechanisms are competitive in polar solvents.All these results are in excellent agreement with our previous findings obtained by employing the B3LYP-D2/6-311++G(d,p) and M06-2X/6-311++G(d,p) theoretical models, [22] thus indicating very similar performances of the three functionals in the thermodynamic investigations of antioxidative activity.

CONCLUSIONS
The performance of the MN12-SX functional in the investigations of the structural, spectroscopic, and antioxidative properties of 5CQA was examined and compared to those of the B3LYP-D2, B3LYP-D3, and M06-2X methods.In the simulated UV-vis spectrum the band at 329 nm was excellently reproduced, but the other three bands do not appear at all in the calculated spectrum.This fact indicates that MN12-SX is inferior to B3LYP-D3.However, this method is the best performer in reproducing the NMR spectra of 5CQA.Capability of the tested functional to reproduce the vibrational spectra, and to approximate thermodynamic quantities necessary for description of the antioxidative mechanisms, is comparable to those of other three methods. [21,22]he MN12-SX calculations perfectly agree with previously revealed most stable conformations of 5CQA in the gas phase [21] and solution. [17,21,22]The quinic moiety is characterized with directed hydrogen bonds, where the carboxylic hydrogen is not oriented towards the carbonyl oxygen of the carboxylic group, but towards the oxygen of the proximate hydroxyl group.
Thermodynamic approach to the investigation of the antioxidative action of 5CQA showed that HAT may be the predominant mechanism in nonpolar solvents, while HAT and SPLET are competitive pathways in polar media.Mechanistic investigation of the HAT and SPLET reaction pathways are under intensive scrutiny.

Figure 2 .
Figure 2. Gas-phase (left) and solvated (right) structures of chlorogenic acid.The hydrogen bond lengths are given in Å.