Screening and Identification of Salicin Compound from Desmodium gangeticum and its in vivo Anticancer Activity and Docking Studies with Cyclooxygenase (COX) Proteins from Mus musculus

Plants have been utilized as medicines for thousands of years [1], initially as crude drugs like tinctures, teas, poultices, powders and other herbal formulations [1,2]. The identity medicinal plants and the methods of their use were passed down through oral history, but eventually this information was recorded in herbals, and subsequently active compounds were isolated beginning with morphine from opium in the early 19th century [1,3], followed by cocaine, codeine, digitoxin and quinine [1,4,5]. Isolation and characterization of pharmacologically active compounds from medicinal plants continue today, and drug discovery techniques are now being used to standardize herbal medicines and to elucidate analytical marker compounds. Methods used to acquire compounds for drug discovery include isolation from plants and other natural sources, chemical synthesis, combinatorial chemistry and molecular modeling [6-8]. Drug discovery from medicinal plants has contributed to cancer treatment, and most new clinical applications during the last half century relate to cancer [4,5,9]. By 2020, approximately 15 million new cancer cases will be diagnosed, and 12 million these patients will die [10]. Cancer is caused by both internal factors such as inherited mutations, hormones, and immune conditions, and environmental/acquired factors like tobacco, diet, radiation, and infectious organisms [11]. The attractiveness of natural compounds as drugs partly stems from their potential ability to influence multiple components of the carcinogenesis pathway.


Introduction
Plants have been utilized as medicines for thousands of years [1], initially as crude drugs like tinctures, teas, poultices, powders and other herbal formulations [1,2]. The identity medicinal plants and the methods of their use were passed down through oral history, but eventually this information was recorded in herbals, and subsequently active compounds were isolated beginning with morphine from opium in the early 19 th century [1,3], followed by cocaine, codeine, digitoxin and quinine [1,4,5]. Isolation and characterization of pharmacologically active compounds from medicinal plants continue today, and drug discovery techniques are now being used to standardize herbal medicines and to elucidate analytical marker compounds. Methods used to acquire compounds for drug discovery include isolation from plants and other natural sources, chemical synthesis, combinatorial chemistry and molecular modeling [6][7][8]. Drug discovery from medicinal plants has contributed to cancer treatment, and most new clinical applications during the last half century relate to cancer [4,5,9]. By 2020, approximately 15 million new cancer cases will be diagnosed, and 12 million these patients will die [10]. Cancer is caused by both internal factors such as inherited mutations, hormones, and immune conditions, and environmental/acquired factors like tobacco, diet, radiation, and infectious organisms [11]. The attractiveness of natural compounds as drugs partly stems from their potential ability to influence multiple components of the carcinogenesis pathway.
Natural products are typically isolated in quantities insufficient for lead optimization, lead development, and clinical trials. Therefore, possibilities of their synthesis or semi-synthesis need to be explored [8,12]. In addition, libraries of natural products and natural-productlike compounds including their features important for combinatorial chemistry may be created [13][14][15].
There are two well established isoforms of the cyclooxigenase (COX) enzyme that differ in their distribution in the body and in physiological function. COX-1 is constitutively expressed in normal tissues and it is involved in maintaining mucosal integrity, platelet aggregation and gastric cytoprotection [16]. In contrast, COX-2 is not expressed in normal mucosa, but is expressed very early in response to neoplastic and inflammatory stimuli, and is extensively overexpressed in different neoplasms, making it an attractive therapeutic target. Besides the role of COX-2 in the production of inflammatory prostaglandins, its momentous participation in the initiation/propagation of cancer [17][18][19][20][21] and in the development of multidrug resistance is well explored [22,23]. Over-expression of COX-2 probably occurs from the first genetically altered cell, through hyperplasia, dysplasia, carcinoma, and even metastasis of colorectal cancer [24][25][26]. Number of nonsteroidal anti-inflammatory drugs (NSAIDs) and selective COX-2 inhibitors have been investigated for anticancer activities [27][28][29][30][31].
White Willow Bark and Meadowsweet are sources of salicin, which has analgesic as well as anti-inflammatory properties. Salicylic acid, released from Salicin in the body, provides anti-inflammatory and painrelieving actions [46], and the same COX-2 inhibition properties as aspirin, but unlike aspirin it does not function as an anticoagulant [47]. Salicin from white willow bark extract showed modest effectiveness in treating pain associated with knee and/or hip osteoarthritis [48] and back pain [49], when administered over a period of weeks in dose of up to 240 mg/day. Our main objective is to optimize Salicin as a specific inhibitor of COX-1 and COX-2 enzymes in the hope that this molecule may be further explored as novel anticancer, especially anti-colorectal cancer, lead-candidates. As in our other report under in vivo condition salicin is proved a potent anticancer drug. Our strategy is intended to obtain selective inhibition of COX-2 using traditional medicinal chemistry techniques motivated by the comparative modeling of a COX-1 and COX-2 complexed with Salicin together with the available pharmacophore. The modern modeling and docking programs/ software packages e.g. Discovery Studio module DS Modeler, Docking Server and Q-site Finder have been used to determine the active sites of COX-1 and COX-2 proteins. The structure of ligand against this active binding site can be found by Q-site Finder. The exact conformation and configuration of the ligand can be calculated to find the best molecule with minimum binding energy and it can be used to develop potential drug molecules against the disease. This knowledge may be important for the development of novel therapies for the treatment of infectious and other viral diseases in the future.

Plant material
The plant D. gangeticum was collected in the month of Nov 2008 from Ayurvedic Garden, Institute of Medical Sciences, B.H.U, Varanasi. This plant occurs naturally on the lower hills and in the plains throughout India. The plants were taxonomically authenticated at the site and collected locally and from surrounding areas as well as from the ayurvedic garden of Banaras Hindu University, Varanasi.

Isolation of Salicin
Dry and finely powdered leaves (5 kg) of D. gangeticum were extracted with methanol (6×5 L) for 36 h using a Soxhlet apparatus at 60-70°C. The residue (536 g) obtained after in vacuo concentration was further fractionated in n-hexane (2 L×2), chloroform (1 L×1), and ethyl acetate (1 L×3) using a mechanical stirrer followed by concentration under reduced pressure to afford crude residue of 152 g, 34 g and 92 g, respectively ( Figure 1). Systematic chemical investigation of the methanolic leaf extract enabled isolation of known glycoside, 2-(hydroxymethyl)phenyl hexopyranoside (DG-1), also known as 'salicin' which is conventionally isolated from the willow bark [50]; this is the first report of isolation of salicin from leaves of D. gangeticum.

Ligand optimization
Salicin is a glycoside, which acts as a precursor for the synthesis of acetyl salicylic acid. For ligand molecule, the structure of salicin, SDF file was retrieved from PubChem site (http://www.ncbi.nlm. nih.gov/pccompound). Salicin.sdf file was converted into PDB file using Discovery Studio visualize and geometry was cleaned by 'clean geometry' menu of Discovery Studio 3.0 and saved for further computational analysis. This retrieved ligand file was imported to the Docking Server program for docking with COX-1 and COX-2 proteins of Mus musculus obtained from NCBI database sequence (http://www. ncbi.nlm.nih.gov/) for homology modeling using Discovery Studio 3.1 DS Modeler.

3D Preparation of receptor molecule
The amino acid sequence of COX-1 and COX-2 was used to search template structure using PDB Database (http://www.pdb.org/pdb/ home/home.do). After getting the templates 1CQE with 85% similarity for COX-1 and 1CVU with 97% similarity for COX-2 ( Figure 2), they were taken for homology modeling using Discovery Studio 3.1 module DS MODELER [51].

Energy minimization
The 3D modeled structures of COX-1 and COX-2 were used for energy minimization using CHARMm force field based on Conjugate Gradient (CONJ) method that exhibits better convergence than the steepest descent method.

Validation of the modeled protein structures
The validity of protein models was tested using DS Protein Health tool, which verifies a protein structure derived from modeling studies or experimental methods. Profiles-3D Verify program based on Kabsch-Sander method was used to evaluate the likelihood that an amino acid should be present within its current environment. It allows us to browse and correct a suggested list of structural violations, which are mapped and colored to the 3D structure. Then CHARMm-based structural refinement of loops and side chains was performed using DS Protein Refine tool. LOOPER algorithm was used for loop refinement that quickly generated energy optimized variants of the structure and provided a list of proposed loop conformations that have been scored using the CHARMm Energy function. The stereochemical quality of modeled protein was checked by Ramchandaran plot provided by online PDBsum analysis. Model quality assessment was done using RAMPAGE (http://mordred.bioc.cam.ac.uk/). Qmean server (http://swissmodel.expasy.org/qmean/cgi/index.cgi) and quantitative evaluation of protein structure quality were done with VADAR (Volume, Area, Dihedral Angle Reporter) server (http://vadar. wishartlab.com/).

Calculating the active site sequence
Active site analysis was done using 'detect cavity' function of Q-site Finder (http://www.modelling.leeds.ac.uk/qsitefinder/). It is an energybased method for the prediction of protein-ligand binding sites.

Docking
The binding of ligand molecule with the COX-1 and COX-2 protein molecules was performed using Docking Server (http:// www.dockingserver.com/web) and SwissDock (http://swissdock. vital-it.ch/), which integrates a number of computational chemistry software specifically aimed at correctly calculating parameters needed at different steps of the docking procedure, i.e., accurate ligand geometry optimization, energy minimization, charge calculation, docking calculation and protein-ligand complex representation, and high-quality docking based on a novel optimization technique combined with a user interface experience focusing on usability and productivity. Its advanced visualization and analysis examined ligandreceptor interactions and finely tuned the docking solutions. Docking calculation, coordination and interaction were done using Discovery Studio 3.1.

Treatment schedule
Swiss albino mice were divided into 5 groups (n=20) and were injected with EAC cells (2×10 6 cells/mouse) intraperitoneally except for the normal group. This was taken as day zero. On the first day normal saline (0.85%, w/v, NaCl) 5 ml/kg/mouse/day i.p. served as Group-I and EAC control (without any treatment) served as Group-II. salicin at 100 mg/kg body weight/day in Group-III and at 200 mg/kg/day in group IV, and the standard drug 5-Flurouracil (5-FU) at 20 mg/kg/day was injected in Group-V. After twenty-four hours from the last dose and 18 h of fasting, 10 animals of each group were sacrificed by cervical dislocation to measure antitumor and hematological parameters. The rest of the animals of each group were maintained to assess their lifespan, and they were provided food and water ad libitum. The effect of isolated and characterized salicin on tumor growth and host's survival time were examined by studying the parameters like tumor volume, tumor cell count, mean survival time, increase in lifespan of EAC bearing mice.

Antitumor and hematological parameters
At the end of the experimental period, the next day after an overnight fasting blood was collected from freely flowing tail vein and from eye and used for the estimation of hemoglobin (Hb) content, red blood cell (RBC) count and white blood cell (WBC) count and differential count of WBC. Along with ascitic fluid was collected from the peritoneal cavity for tumor volume, tumor weight, percentage increase in life span, tumor cell count, viable/nonviable tumor were measured.

Results
Cancer is one of the life-threatening diseases, and identification of active drug targets against proteins involved in cancer like COX is of great interest. Medicinal compounds like Salicin have antiinflammatory activity, due to which they may serve as an active drug target against many toxic/pathogenic proteins like COX-2 in our study. Interaction of COX proteins with drugs like aspirin and ibuprofen shows pharmacological inhibition of this protein and thus relieve many symptoms of inflammation and pain. Salicin is found to be as active drug target in our study as it shows inhibition of COX-1 and particularly COX-2, indicated that it may be useful in chemoprevention of some cancers like colorectal cancer. The detailed results are discussed here. were assigned for C-1 and C-2, respectively. The signal for hydroxyl methylene resonance was observed at δ 64.5, while the signals at δ 127.8, δ 120.0, δ 129.4 and δ 116.4 were identified for tertiary carbons C-3, C-4, C-5 and C-6, respectively.

Isolation and identification of salicin
Finally, the compound DG-1 was identified as 2-(hydroxymethyl) phenyl hexopyranoside, also known as 'salicin', on the basis of comparison of physical (melting point and elemental analysis) and spectroscopic data (UV, IR, 1 H NMR, 13 C NMR and mass spectra) to literature reports [52][53][54][55]. In the absence of an authentic sample, a direct comparison was not possible. This is the first report of isolation of salicin, from leaves of D. gangeticum.

Homology modeling
The three dimensional structures for COX-1 and COX-2 proteins were constructed using PDBID 1CQE X-ray diffraction with resolution of 3.10 Å with E-Value: 0.0 and Score 2596, Identity 84%, Positivity 89% for COX-1 and PDBID 1CVU X-ray diffraction with resolution of 2.40 Å with E-Value 0.0, Score: 2882 Identity 97%, Positivity 97% for COX-2. Based Template 1CQE for COX-1 and 1CVU for COX-2 was used for model construction using homology modeling tool DS MODELER (Figure 7). The Predicted 2D and 3D structures provide valuable insight into functional regulatory region in secondary elements and also enable the identification of possible interaction site for a suitable inhibitor. Among the three conformations generated the one with the least modeller objective function value was considered to be thermodynamically stable and was chosen for further refinement and validation.
Based on the structural alignment of the amino acid sequences of the COX-1 and COX-2, a theoretical model of these proteins was obtained, corresponding with amino acid residues 34-586 for COX-1 and residues 18-569 for COX-2 of the primary structure ( Figure 8).

Evaluation and refinement
The rough models for COX-1 and COX-2 were subjected to energy minimization using conjugate gradient algorithm with maximum steps 200 and RMS gradient 0.1 to eliminate bad contacts between amino acid atoms using simulation tool of Discovery Studio 3.1. The backbone conformation of the rough model was inspected using the Phi/Psi Ramchandaran plot obtained in the PDBSum server (http://www.ebi.ac.uk/pdbsum/). The results of Ramchandaran plot indicate that the rough model generated for COX-1 had no residue in the disallowed region whereas that for COX-2 had two residues, viz., Glu384 and Ser482, in the disallowed region, occurring in the loop. Loop refinement was done using looper and CHARMm based molecular mechanics to generate multiple energy optimized variants of the selected segments of the protein structure. Side chain refinement was done using chi-rotator, a CHARMm based energy minimization, a routine tool to optimize the conformation of the selected amino acid side chain atoms ( Table 2).
The Ramchandaran plot statistics showed that 89.9% residues were in the most favored regions with Φ and Ψ angles in the core of favored regions and 10.1% of residues were in additional allowed regions for COX-1, while for COX-2 90.7% residues were in the most favored regions and 9.3% of the residues occurred in additional allowed regions ( Figure 9). This result was also verified by RAMPAGE server (http:// mordred.bioc.cam.ac.uk/~rapper/rampage.php). There are no residues either in the generously allowed region and or in disallowed region in COX-1 and COX-2 had two residues, viz., Glu384 and Ser482, in the disallowed region, after evaluation and refinement.

Model quality estimation
ProSA-web server (https://prosa.services.came.sbg.ac.at/prosa. php) shows overall model quality by comparing the potential of mean forces derived from a large set of known NMR and X-ray deciphered structures of similar sizes and group. The model quality assessment is graphically presented in form of Z score; in our study Z scores were found to be −8.86 and −9.23 for COX-1 and COX-2 respectively,     QMEAN server (http://swissmodel.expasy.org/qmean/cgi/index. cgi), model quality estimation, was used to analyze QMEAN score, residue error, energy profiles and plot and volume area dihedral angle for fractional accessible surface area, residue volume, 3D profile and stereo/packing quality index were done with VADAR (http://vadar. wishartlab.com/). QMEAN and VADAR were specially designed for quantitative and qualitative assessment of protein structures determined by X-ray crystallography, NMR spectroscopy, 3D-threading or Homology modeling. QMEAN score of the whole model reflecting the predicted model reliability range from 0 to 1. In this study, the predicted COX-1 model QMEAN score was 0.654 with global scores estimated absolute quality Z-score of -1.32; these results show that the model is reliable. Fractional accessible surface area volumes of all residues close to 1.0 ± 0.1, statistics of hydrogen bonds of predicted model show equal to expected mean H bond distance score of 2.2 sd=0.4 equal to the expected value, Mean H bond energy observed -1.7 sd=1.1 close to the expected -2.0 sd=0.8. Dihedral Angles were observed close to the expected, Total Accessible Surface Area score 23870.3 Angs**2 with expected score 20013.5 Angs**2, Total volume (packing) score observed 85910.2 Angs**3 with expected 77534.2 Angs**3, Stereo/Packing, 3D quality index results show no error residues in the predicted model. In this predicted COX-2 model QMEAN score 0.603 with global scores estimated absolute quality Z-score of -1.94 indicate that the model is reliable. Fractional accessible surface area volumes of all residues is close to 1.0±0.1, statistics of hydrogen bonds of predicted model is equal to the expected mean H bond distance score of 2.2 sd=0.4, Mean H bond energy observed was -1.7 sd=1.1 that is closes to the expected -2.0 sd=0.8. Dihedral Angles were observed to be closes to the expected, Total Accessible Surface Area score was 23652.7 Angs**2 with expected score of 19948.1 Angs**2, Total volume (packing) score was observed 83467.8 Angs**3 with expected of 77179.1 Angs**3, Stereo/Packing, 3D quality index results shows that no error residues occur in the predicted model. After complete reliability test based on quantitative and qualitative assessment, the predicted Models of COX-1 and COX-2 from M. musculus were deposited in Protein Model Database (PMDB) with PMDBID PM0077492 (http://mi.caspur.it/PMDB/user/model_ info.php?id=77492) and PM0077493 (http://mi.caspur.it/PMDB/user/ model_info.php?id=77493), respectively.
Superimposition of the 3D predicted model with template were done with Combinatorial Extension (CE) Method based server (http:// cl.sdsc.edu/). The superimposed backbone traces displayed Z-Score=8.4 and Rmsd=0.3Å with sequence identity=89.5% for COX-1, while for COX-2 Z-Score=8.4 and Rmsd=0.4 Å with sequence identity=99.5% for all atoms calculated locally within the two polypeptide chains (Figure 2). Most bond lengths, bond angles, and torsion angles were between values expected for a naturally folded protein.

Active site residue details
Among the ten sites obtained from Q-site finder, only 3 sites were selected, since they were conserved among different species, and the other sites are not further discussed. In both the proteins, site 2 is highly conserved in the active sites of the template and predicted COX-1 and COX-2 models (Figure 10). Results from multiple sequence alignment and Motif analysis revealed that at COX-1 site 2, Tyr150, Ala201, Phe202, Ala204, Gln205, Thr208, His209, Phe212, Lys 213, Thr214, Leu296, Tyr350, Asn384, Tyr387, His388, Trp389, His390, Leu392, Met393, Phe397, Tyr406, Phe409, Leu410 and Val446 residues are conserved, and at the COX-2 site 2, Tyr 134, Thr192, His193, Phe196, Lys197, Thr198, Asp199, His200, Lys201, Arg208, Asn217, His 218, Gly221, Glu222, Thr223, Arg226, Gly274, Gln275, Glu276, Val277, Asn368, His372 are conserved (Table 3). Sequence analysis of the proteins depicts low substitution rate and less gap penalty, which indicates that they belong to the same protein family. Thus, site 2 of both COX-1 and COX-2 proteins has been found to be the most favorable site for docking studies. Motifs obtained from MEME tool, 29 for COX-1 and 21 for COX-2 out of selected maximum number of 30 motifs with length 20 to 50 ( Figure 11). In which motif 2 and motif 5 represent the functional motif for both COX proteins containing the active site residues and catalytic residues, which are highly conserved and most representative among different species (Figure 12). The sequences which are highly representative and conserved throughout evolutionary studies are marked by red box in figure 11 (FAFFA in COX-1 and QHFTHQ and FFA in COX-2).

Docking study of COX-1 and COX-2 receptors with salicin inhibitor
Docking of COX-1 and COX-2 was performed with salicin inhibitor (2-(hydroxymethyl) phenyl hexopyranoside). The final docked conformation obtained for salicin was evaluated based on the number of hydrogen bonds formed and bond distances between atomic coordinates of the active site and inhibitor. To evaluate the structural similarity of M. musculus COX-1 and COX-2 with related COX proteins of different species, multiple sequence alignment (22 species for COX-1 and 25 for COX-2) was performed by UPGMA. The UPGMA (Unweighted Pair Group Method with Arithmetic Mean) phylogenetic profile based on reliability of an inferred tree is based on Felsenstein's [56] bootstrap test, which is evaluated using Efron's [57] bootstrap resampling technique (Figure 3). It was observed that the following stretch Ala201-Phe202-Ala204-Gln205, Thr208-His209, Phe212-Lys213-Thr214, Tyr387-His388-Trp389-His390, Leu392-Met 393 for COX-1 and Ala188-Gln189, Thr192-His193, Phe196, Asn368, Tyr371-His372, His374, Leu377 for COX-2 were conserved and present in catalytic active residues. Among these stretches the Thr208, Tyr387 and Trp389 amino acids residues may be involved in hydrogen bond interaction with salicin in the case of COX-1, while in case of COX-2 Ala 188, Gln189, Thr192, His193, Phe196, Asn368, Tyr371, His372, His374, Leu377 amino acid residues appear to be involved. Based on docking of COX proteins with salicin inhibitor COX-2 having more hydrogen bond with greater affinity interaction rather than COX-1. The hydrogen bond interactions between inhibitor and COX proteins along with the bond distances are shown in table 4 and figure 13.

Effect of salicin on tumor volume and survival time
There were no gross behavioral changes and mortality upto a dose level of 200 mg/kg body weight. The LD50 value of salicin compound was found to be > 2g/kg body weight of mice indicating that it has low toxicity to the animal. Treatment with salicin at the dose of 100 and 200 mg/kg body weight increased the lifespan (ILS) and nonviable cell count and significantly reduced the tumor volume, tumor weight and viable tumor cell count when compared to that of EAC control group ( Table 5).

The effect of salicin on hematological studies
The haemoglobin content, RBC count, lymphocyte (%) and monocyte (%) in EAC bearing mice given salicin at the dose of 100 and 200 mg/kg increased significantly compared with those in EAC control, whereas WBC count and neutrophil (%) showed significant      (Table 6). Treatment with salicin restored the hematological parameters to more or less normal values. The number of RBC count and hemoglobin content also increased, while the WBC and the differential count decreased as compared to that of EAC control. Treatment with salicin illustrated the percent increase in tumor cell volume and numbers of viable tumor cells were found to be significantly less when compared to those of the EAC control. Hence, it can be concluded that the extracts by their cytotoxic effect and arresting the tumor growth, increased the life span of EAC bearing mice. The percentage increase in life span in response to the 200 mg/kg body weight of salicin administration was indicating its potent anticancer nature (Table 6). In acute toxicity studies, the administration of salicin at the dose of 100 mg/kg and 200 mg/kg for 14 days did not exhibit any adverse effect.

Discussion
In the early 1990s, cyclooxygenase (COX) was demonstrated to exist as two distinct isoforms. COX-1 is constitutively expressed as a housekeeping enzyme in nearly all tissues, and mediates physiological response, e.g., cytoprotection of the stomach, platelet aggregation. COX-2, on the other hand, is expressed by cells that are involved in inflammation, e.g., macrophages, monocytes, synoviocytes, and has emerged as the isoform that is primarily responsible for synthesis of  Each point represents the mean (n=10 mice per groups).
*The values marked with the different letters show significant difference (Duncan's multiple range test, P<0.05). the prostanoids involved in pathological processes, such as acute and chronic inflammatory states [58]. The two known COX isoforms show a high degree of similarity in their amino-acid sequences [59][60][61][62] and structural topology [63][64][65].
Classical NSAIDs like aspirin, ibuprofen, naproxen, but not nimesulide, are non-selective inhibitors of both the COX isozymes (IC 50 for COX-1 is similar to that for COX-2) and their prolonged use can cause gastric bleeding and renal failure [66][67][68]. Hence, there have been sustained effort to identify selective COX-2 inhibitors, i.e., compounds whose IC 50 for COX-1 inhibitory activity is significantly greater than that for COX-2. Some COX-2 inhibitors have been evaluated in clinical trials, but some of them showed increased cardiovascular toxicity; celecoxib, however, seems to be relatively safe COX-2 inhibitor [58]. It has meanwhile been hypothesized that there might be other isoforms of the COX enzyme yet to be discovered [67].
Nexrutine is a herbal alternative to COX inhibitor drugs for pain and soreness, and offers a number of advantages over both broad COX-1/2 inhibitors like aspirin and selective COX-2 inhibitors like Celebrex. Nexrutine inhibits the inflammatory COX-2 connected with pain without inhibiting the protective COX-1; thereby having a lower risk of producing gastrointestinal and bleeding side effects compared aspirin and Celebrex [69]. Synthetic compounds like mono-, di-, and triaryl substituted tetrahydropyrans were also reported as COX-2 and tumor growth inhibitors. These compounds exhibit IC50 for COX-2 in the range 0.57-4.0 nM, and their selectivity for COX-2 over COX-1 is better than that of celecoxib and rofecoxib [70].
A number of docking procedures based on different search and scoring algorithms have been proposed [71], but none can treat all biological systems with the same accuracy and efficacy [72][73][74]. It is thus advisable once the biological target has been selected, to set up an adequate system strategy for the study goal. The bark from white willow and some legumes contain salicin which is a natural pain reliever, is very easy on the stomach and kidneys, while acetylsalicylic acid is known to upset the stomach and in some cases damage kidneys. Scientists believe that this is because salicin is converted to acetylsalicylic acid after the stomach has absorbed it [75]. Putting acetylsalicylic acid directly into the stomach damages its lining and bleeding ulcers can result. Thus salicin is a pro-drug that is gradually transported to the lower part of the intestine, hydrolysed to saligenin by intestinal bacteria, and converted to salicylic acid after absorption. It thus produces an antipyretic action without causing gastric injury [76].
COX-2 produces inflammation causing compounds that lead to swelling for curative and protection. Blocking the COX-2 enzyme completely is not good because COX-2 activity is required for cardiovascular health. Many pain killers in the past ended up blocking all COX-2 activity, which led to heart problems for the patients taking them. Today most pain killers block a part of COX-2 activity but they also block COX-1, which is vital for the health and structure of the stomach lining. The white willow bark derived pain killer does not block COX-1, but it does block COX-2; as a result, it has few reported side effects [77].
Most docking algorithms consider that the enzyme structure is rigid, according to the high computational cost induced by the flexibility of big molecules, while the ligand is free to move. Usually, in the first step, a library of ligand conformers is generated and in the second step, these conformers are docked into the target, each conformer being treated as a new ligand. Any algorithm able to generate in a correct manner the ligand conformers could be used to generate this library. One of them is represented by genetic algorithms. Genetic algorithms attempt to use the rules of natural selection to subset computationally demanding tasks [78,79].
Qualitative and quantitative analyses of predicted 3D-structure of PMDBID PM0077492 and PM0077493 show that total Accessible Surface Area and total volume (packing) have significant reliable scores, which were compared with the experimental data available in VADAR server [80]. Secondary structure element statistics describe 44% and 46% of helices, 13% and 13% of Beta sheets, 41% and 40% of coils and 25% and 26% of turns in PM0077492 and PM0077493, respectively. After complete validation and refinement, stereochemical properties reveal no residues in the disallowed region in COX-1 and COX-2 had two residues, viz., Glu384 and Ser482, in the disallowed region. A good quality model would be expected to have over 90% residues in the most favored regions. In this study PMDBID PM0077492 and PM0077493 contain 89.9% and 90.7% residues in the most favored regions with overall average G-factor value of 0.14 and 0.17, respectively, indicating that the model quality is good.
Hillarp et al. [82] reported that in case of mutations within the COX-1 gene in aspirin non-responders with recurrence of stroke and carriers and non-carriers of one of the mutations behaved similarly when aggregation and granule content release function were studied using collagen, ADP and arachidonic acid as agonists. Thus their hypothesis do not support that common variants of the COX-1 gene results in unblocked COX-1 molecules in aspirin non-responders.
The present study illustrated the effect of isolated compound salicin from D. gangeticum leaves on EAC bearing mice, which significantly increased the life span of treated animal as compared to the EAC control. The reliable conditions for evaluating the value of any anticancer drug are prolongation of life span and decrease in the WBC count [83]. The ascitic fluid is the direct nutritional source to tumor cells and the rapid increase in ascitic fluid with tumor growth could possibly be a means to meet more nutritional requirements of tumor cells [84]. Furthermore the reduced volume of EAC and increased survival time of mice suggest the delaying impact of extracts on cell division [85]. The treatment with salicin inhibited the tumor volume, viable cell count and enhanced survival time of EAC bearing mice. These finding suggest the antitumor effects of salicin compound against EAC cell line.

Conclusion
The protein-ligand interaction plays a significant role in structural based drug designing. It has been clearly demonstrated that the approach utilized in this study is successful in finding novel anticancer inhibitor from medicinal plant D. gangeticum. The ligand salicin, in particular, from D. gangeticum showed high binding affinity against COX-2 protein (PDB ID: PM0077493) (-5 Kcal/mol) and lesser interaction with COX-1 (PDB ID: PM0077492) (-3.79 Kcal/mol) based on docking score. Therefore, this study states the importance of natural plant compound protein-ligand interaction studies, in silico. From the docking results, we conclude that salicin could be a potential COX inhibitor. Arachidonic acid (mmu00590) and VEGF signaling (mmu04370) pathway key enzymes play an important role during tumor angiogenesis and metastasis in context of COX proteins so it is possible that proteins related to these metabolisms are actively involved in interaction with salicin. However, additional biological and mutational studies would help in prediction of anti-cancerous compounds. The obtained results are useful to understand the structural features inhibitory activities to COX proteins. The extraction, isolation and characterization of bioactive compound salicin from leaves of this plant and its in vivo anticancer activity confirm salicin as potent anticancer drug.