Structural elucidation and molecular docking of ferulic acid from Parthenium hysterophorus possessing COX-2 inhibition activity

In this study, isolation of ferulic acid from Parthenium hysterophorus L. followed by the structural characterization using elemental analysis, FT-IR, NMR, ESI–MS and XRD has been carried out. The molecular geometry, harmonic vibrational frequencies and structural parameters were computed by density functional theory with the 6-311G** basis set. Comparisons between experimental and simulated data of spectroscopic analysis and geometrical parameters was accomplished for their statistical validation and the values of correlation coefficient for 1H and 13C-NMR chemical shifts, bond lengths and bond angles were found to be 0.934, 0.951, 0.943 and 0.961, respectively. The HOMO and LUMO analyses were used to find out the charge transfer within the ferulic acid. Thermal studies were carried out by thermogravimetry (TG), differential thermogravimetric analysis (DTA), and derivative thermogravimetry (DTG) to confirm the effect of temperature upon the ferulic acid. Furthermore, the binding study of the optimized geometry of ferulic acid has been envisioned with cyclooxygenase-2 for its activity inhibition by molecular docking. Electronic supplementary material The online version of this article (doi:10.1007/s13205-014-0253-6) contains supplementary material, which is available to authorized users.


Introduction
Parthenium hysterophorus L. is an annual herbaceous, exotic weed which belongs to the asteraceae family. It grows abundantly in agricultural lands, orchards, forest lands, overgrazed pastures, flood plains, wasteland, around residential colonies, railway tracks and along roadsides (Patel 2011). This noxious weed was introduced in India with the contaminated PL-480 wheat, imported from USA in 1950s. Exposure to P. hysterophorus L. in human manifests the symptoms of hay fever, eczema, skin inflammation, allergic rhinitis, black spots, burning, asthma, blisters around the eye, diarrhea, breathlessness, severe papular erythematous eruptions and choking Maishi et al. (1998), systemic toxicity in livestock by irreparable damage to liver and kidney, and inhibits the liver dehydrogenases in buffalo and sheep (Gunaseelan 1987;Rajkumar et al. 1988). However, P. hysterophorus L. possesses useful chlorogenic acids such as caffeic, p-coumaric, ferulic, vanillic and p-anisic and many more which has been exploited for their biomedical and industrial purposes (Ferguson et al. 2005;Kikugawa et al. 1983;Kumar and Pruthi 2014;Meng et al. 2013). Among them, ferulic acid (FA) is a ubiquitous phenolic cinnamic acid derivative, commonly found in the cell wall of commelinid plants, grasses, grains, beans, seeds of coffee, artichoke, and nuts (Rosazza et al. 1995). FA exhibits wide variety of biomedical activities viz. antioxidant, anti-inflammatory, antimicrobial, antiallergic, hepatoprotective, anticarcinogenic, antithrombotic, increase sperm viability, antiviral, increases binding of IgE to peanut allergens, improve the stability of cytochrome c and inhibit the apoptosis induced by cytochrome c (Chung and Champagne 2011; Mori et al. 1999;Kumar and Pruthi 2014;Middleton et al. 2000;Ou and Kwok 2004;Ou et al. 1999;Toshihiro et al. 2000;Yang et al. 2007). Recent reports proved that FA also acts as a b-secretase modulator with therapeutic potential against Alzheimer's disease and also improves the structure and function of heart, blood vessels, liver, and kidneys in hypertensive rats (Alam et al. 2013;Mori et al. 2013). Due to the wide variety of functional aspects of FA, there is an urgent need to study its geometry and other structural properties. Further, to study the role of FA in anti-inflammatory reactions, molecular docking of FA and cyclooxygenase-2 (COX-2) has been carried out. The COX-2 is also known as prostaglandin-endoperoxide synthase (PTGS), an enzyme (EC No.: 1.14.99.1) which is responsible for the formation of key biological mediators such as prostanoids (prostaglandins, prostacyclin and thromboxane). The action of most of the anti-inflammatory drugs is due to their binding ability within the active sites of COX-2, preventing the catalysis of arachidonic acid to prostaglandins (Shorrock and Rees 1988). Pharmacological inhibition of COX-2 can provide relief from the symptoms of inflammation and pain. To the best of our knowledge, this is the first report of its kind of statistical analysis of experimental and simulated data for structural parameters and inhibition of COX-2 by FA.

Materials
Plant samples have been collected from the area nearby Indian Institute of Technology Roorkee. The Plant was approved by the Forest Research Institute (FRI), Dehradun (Uttarakhand, India) as P. hysterophorus L. Plants were thoroughly washed with boiled distilled water and their parts such as stem, leaves and flowers have been separated. The samples were kept to become dry in oven at 35°C for 24 h. Finally, the dried samples were ground to powder for experimental analysis. All the chemicals and reagents used in this study were purchased from Himedia (Mumbai, India) and of analytical grade.

Extraction and purification
The FA has been extracted by the modifications in the earlier reported method (Tilay et al. 2008). Briefly, 2.0 g of each plant sample (root, stem and leaves) was taken in a 250-ml round bottom flask charged with 60 ml of NaOH (2M). To prevent the oxidation of FA during alkali treatment, 0.001 g NaHSO 3 was added in each flask and kept within a rotary shaker at 25°C, 180 rpm for 24 h. Then, all the samples were centrifuged at 12,000 rpm for 10 min and the supernatant so obtained was acidified to a pH of B2 by HCl (2 M) solution. The acidified samples were treated three times with 60 ml ethyl acetate and concentrated for the extraction of FA. The concentrated extracts were dissolved in equal volume of acetonitrile/water for further analysis. FA bands from TLC plate were scraped and dissolved in 2.0 ml acetonitrile. The quantitative analysis of all samples was performed on a HPLC column (Merck, Darmstadt, Germany). The isocratic procedure was applied using a mobile phase of acetonitrile/ water (80:20) and 0.1 % acetic acid, 20 ll injection volume and flow rate of 1.0 ml/min and analyzed at 320 nm (Zupfer et al. 1998). The melting point of purified FA was found to be 173-175°C. The FT-IR spectrum of FA (KBr, Fig. 1a) showed the characteristic peaks of carboxylic group (-OH stretching at 3,437 cm -1 , C=O at 1,689 cm -1 , C-O at 1,278 cm -1 ), aromatic system (C-H stretching at 3,080-3,030 cm -1 , combination band at 2,000-1,650 cm -1 , C=C stretching at 1,600-1,450 cm -1 , C-H bend at 900-650 cm -1 ), alkene (1,690 cm -1 ) merged with C=O of -COOH group and methoxy group (C-O-C stretching at 1,275-1,200 cm -1 ) also merges with C-O of COOH group.  68,116.04,116.16,123.36,126.30,145.05,148.44,. The mass spectrum of isolated compound, shown in Fig. 1d  Single crystal X-ray diffraction data were collected at 100 K on a Bruker Kappa-CCD diffractometer using graphite monochromated MoKa radiation (k = 0.71070 Å ). The empirical absorption corrections were applied in the reduction of data Lorentz and polarization corrections (Sheldrick 1996). The SHELXTL program was used for the structure solution, refinement and data output (Sheldrick 1990(Sheldrick , 2000. Non-hydrogen atoms were refined anisotropically, while the hydrogen atoms were placed in geometrically calculated positions using a riding model. The images and hydrogen bonding interactions were created with diamond and mercury (Brandenburg 1999). Refinement parameters and data collection conditions for FA during crystallography experiment are given in the Table 1. Thermogravimetry (TG), differential thermal analysis (DTA) and derivative thermogravimetry (DTG) were carried out using a mass of 0.045 g at 10°C/min under the nitrogen at 200 ml/min flow rate on a thermogravimetric analyzer (PerkinElmer's, CA, USA).

Quantum chemical calculations and statistical analysis
The quantum chemical calculations for the geometry optimization of FA were performed by Gaussian 09 with a hybrid function B3LYP at DFT/6-311G** basis set (Becke 1993;Frisch et al. 2009;Peng et al. 1996;Stephens et al. 1994;Goel and Singh 2013). The structural parameters for optimized structure of the FA viz. dipole moment, total energy, bond lengths, bond angles, and HOMO-LUMO energy were computed. The 1 H and 13 C-NMR chemical shifts were also calculated by the Gauge-Independent Atomic Orbital method at corresponding basis set. Theoretically computed chemical shifts were converted to TMS scale prior their comparison with the experimental data. The electronic absorption spectra for FA with timedependent DFT (TD-DFT) in solution (ethanol and DMSO) and gas phase were also computed. Different physicochemical properties viz. electronegativity (v), chemical hardness (g), chemical softness (S), chemical potential and electrophilic index (x) for FA were calculated using Koopmans theorem for a closed shell molecule (Koopmans 1934). The experimental and simulated values for chemical shifts ( 1 H and 13 C-NMR), bond lengths and bond angles were statistically tested for their mathematical significance in the MATLAB R2010a toolbox by curve fitting analysis (Goel and Kumar 2014;Kumar and Bhalla 2011). The correlation coefficient (R) value which measures the potency and direction of a linear association between two variables was also calculated between simulated and experimental data.

Molecular docking
In the present work, molecular docking of FA and COX-2 enzyme has been carried out with the help of AutoDock 4.2.3 suite, which uses the genetic algorithm (GA) for its internal conformation search and produces an assembly of conformations by applying Lamarckian genetic algorithm to study the interactions between the ligand and receptor (Morris et al. 2009;Garg et al. 2013). The structure of COX-2 (PDB ID: 6COX) used in this study is available at protein data bank (http://www.rcsb.org). During the docking experiment, water molecules and heteroatoms have been removed, while the hydrogen atoms were added at appropriate geometry groups. The COX-2 was ionized as required at the physiological pH and its protonated form was used for the final docking experiment. The docking parameters used in the present research work are: number of GA runs: 30, population size: 150, maximum number of   The energy-based AutoDock scoring function includes terms accounting for short range Van der Waals forces, electrostatic interactions, loss of entropy upon ligand binding, hydrogen bonding, and solvation. The input files (.pdb files) of COX-2 and FA were converted into the subsequent.pdbqt format output files, which contain the necessary parameters for docking such as atom coordinates, partial charges, and salvation. To know the binding sites in COX-2, blind docking was done with the grid size of 126, 126, and 126 along with the X, Y, and Z axis and 0.697 Å of the grid spacing. The center of the grid was set to 29.5, 31.8, and 23.5 Å .

Results and discussion
Isolation and purification Isolation of FA from P. hysterophorus L. samples (viz. root, stem, leaves and whole plant) has been carried out by alkali treatment method. The qualitative and quantitative analysis was carried out by chromatographic techniques.
The TLC chromatograms of all the samples with  commercially purchased ferulic acid (Sigma-Aldrich, India) indicate the presence of FA in each sample when sprayed with 10 % ferric chloride solution (Fig. S1). The HPLC analysis of spots scraped from TLC plates confirmed the purification of FA with retention time of 29.07 min (Fig. S2). The content of FA in whole plant was found to be 123-145 mg/100 g (38-45, 53-60 and 32-40 mg in root, stem and leaves, respectively).

Crystal structure description
After spectroscopic characterization of the isolated compound as FA (4-hydroxy-3-methoxycinnamic acid), we crystallize the FA to study its structural properties such as bond lengths, bond angles, inter(intra)-molecular H-bonding and many more. FA was crystallized in the monoclinic space group P 21/n. The crystal packing of FA (Fig. S3) showed that the dominant O-HÁÁÁO hydrogen bond was formed via the interaction of OH of -COOH group of one FA molecule with the oxygen atom of the C=O group of adjacent molecule and vice versa forming a dimer which is responsible for the formation of R 2 2 (8) motif via strong [O 3 -H 3 ÁÁÁO 4 , 1.882(1) Å ] non-covalent intramolecular interactions (Table 2). On the other side, the phenolic group (-OH) of FA also exhibited a strong O-HÁÁÁO [O 1 -H 1 ÁÁÁO4, 2.148 (1) Å ] non-covalent interaction with the C=O group of adjacent acid molecule (Fig. 2a, b).

Thermal analysis
The thermal stability of FA has been demonstrated by TG-DTA-DTG study. The TGA-DTG curves for FA as shown in Fig. 3 clearly indicated that it was stable up to 100°C but at higher temperatures curves for thermal study showed irregular pattern. The decomposition of FA occurs in one step in which it leaves out with 99.1 % weight loss (DTG peak at 260°C) in between 200 and 432°C, corresponding to endotherm at 173°C in DTA thermogram.

Geometry optimization
The optimized geometry of FA showed the positive harmonic vibrational frequencies, which indicate that it attained the global minimum on the potential energy surface. Singlepoint energy calculations and zero-point corrected total energies were also computed. The optimized structure of FA at DFT/6-311G** basis set as shown in Fig. 4 matches with the structure obtained by single crystal diffraction. The comparison between experimental and simulated chemical shifts for proton and carbon is given in Table 3, while the structural parameters such as bond lengths and bond angles are provided in Table 4. From the statistical analysis of data, it was found that the value of correlation coefficient (R) for four parameters, 1 H and 13 C-NMR chemical shifts, bond lengths and bond angles is 0.934, 0.951, 0.943 and 0.961, respectively. The statistical analysis was performed to validate the theoretical and experimental data. The plots for curve fitting analysis are shown in Fig. 5, which clearly indicate the significant resemblance between theoretical and experimental data obtained from NMR and X-ray crystal structure. Commenting on the statistical results, we inferred that simulated values lie statistically closed to the values obtained from experimental results and outcome of all the four cases to be worthy with respect to their statistical significance for correlation.
Frontier molecular orbital (FMOs) and ultraviolet spectra analysis Highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) are known as frontier molecular orbital (FMOs), which plays a major role in the computation of UV-Vis spectra, electronic and optical properties of a molecule. HOMO represents the ability to donate an electron, while LUMO corresponds to the ability of a molecule to obtain an electron. The energy gap between these two orbitals helps to compute the physico-chemical properties viz. chemical hardness, chemical softness, kinetic stability, optical polarizability of the molecule. To evaluate the energetic behavior of FA, the calculations were done both in solution (ethanol and DMSO) and gas phase. The UV-Vis data for FA obtained by TD-DFT simulations have been complying with previously reported experimental results and found comparable (Asiri et al. 2011;Kosar and Albayrak 2011). The calculated values for the physico-chemical properties with the energies of the highest (HOMO), second highest (HOMO-1), third highest (HUMO-2), lowest (LUMO), second lowest (LUMO ? 1) and third lowest (LUMO ? 2) for FA are summarized in Table 5. These parameters are prerequisites for QSAR calculations of a compound. The molecular orbital diagrams for FA, computed at DFT/6-311G** basis set and shown in Fig. 6, confirmed that LUMO is localized on almost whole molecule, while HOMO is mainly localized on the side chain and near carboxylic acid. The values of energies for HOMO were calculated as -5.75, -5.76 and -5.73 eV, while for LUMO these values were computed as -2.28, -2.29 and 2.09 eV in ethanol, DMSO and gas phase, respectively. The energy gap between HOMO and LUMO also indicates the chemical stability of the molecule. The values of energy gap of FMOs in FA were found to be -3.46, -3.46 and -3.64 eV in ethanol, DMSO and gas phase, respectively. Theoretical absorption wavelength (k), excitation energies (E) and oscillator strengths (f) for FA have been computed and given in Table 6. Data from theoretical calculations for physico-chemical properties of FA showed that the values are close in solution phase (ethanol and DMSO); while in gas phase, these are slightly deviated from those in solution phase.

Analysis of docking results
All the docking poses of complex (COX-2 and FA) created by the AutoDock were loaded into PyMOL. The docking poses were ranked according to their docking scores and both the rank list of docked ligands with their corresponding binding poses have been given in Table S1. The analysis of the complex with the best docking conformation (with the least binding energy) showed that FA binds to the COX-2 mainly in the hydrophobic pocket. The stabilization of the FA-COX-2 complex has been carried out through the hydrogen bond interactions of different bond lengths between FA and amino acids of COX-2 enzyme. This study shows that the FA-COX-2 complex is stabilized by two hydrogen bonds of bond length 1.945 and 1.801 Å with the residues Gln372 and Lys532 of COX-2, respectively. The binding energy of the lowest energy conformer of FA-COX-2 complex was calculated computationally and found to be -5.25 kcal/mol. These results show the high affinity of FA with COX-2 enzyme. The binding pose of FA with COX-2 is illustrated in Fig. 7.

Conclusions
This investigation started with the extraction and purification of FA from P. hysterophorus L. and its structural characterization has been carried out experimentally using different spectroscopic techniques and single crystal diffraction. FA was crystallized in the monoclinic space group (P 21/n) forming dimer via the formation of a strong hydrogen bond between -OH of -COOH group of one FA molecule with the oxygen atom of the C=O group of adjacent molecule and vice versa. A quantum chemical calculation for the optimization of FA geometry was done by Gaussian 09. The electronic and physico-chemical properties for the optimized geometry of FA have been also calculated, which showed degree of freedom, total energy and dipole moment to be 66, -688.012 Hartree and 2.50 Debye, respectively. The developed model comprises of geometry optimization at DFT/6-311G** basis set and the optimized geometry parameters and NMR chemical shifts of FA showed good correlation with the values obtained from X-ray crystal structure and experimental NMR. The  calculated value for UV spectra of FA also matches with the experimental data. The theoretical as well as experimental data depicted structural resemblance of FA in solution and gaseous phase, which was confirmed by statistical significance of the data. The thermal studies carried out using non-isothermal TG-DTA-DTG inferred that FA undergoes single-step thermal decomposition. Further, molecular docking of FA with cyclooxygenase-2 was performed by AutoDock 4.2.3 program and we demonstrated that it binds with COX-2 enzyme in the hydrophobic pocket stabilized mainly by two hydrogen bonds. This study has tremendous scope in the pharmaceutical industry and biology for the designing of novel inhibitors for COX-2 enzyme.