Steroidal glycoalkaloids from Solanum nigrum target cytoskeletal proteins: an in silico analysis

Background Solanum nigrum (black nightshade; S. nigrum), a member of family Solanaceae, has been endowed with a heterogeneous array of secondary metabolites of which the steroidal glycoalkaloids (SGAs) and steroidal saponins (SS) have vast potential to serve as anticancer agents. Since there has been much controversy regarding safety of use of glycoalkaloids as anticancer agents, this area has remained more or less unexplored. Cytoskeletal proteins like actin play an important role in maintaining cell shape, synchronizing cell division, cell motility, etc. and along with their accessory proteins may also serve as important therapeutic targets for potential anticancer candidates. In the present study, glycoalkaloids and saponins from S. nigrum were screened for their interaction and binding affinity to cytoskeletal proteins, using molecular docking. Methods Bioactivity score and Prediction of Activity Spectra for Substances (PASS) analysis were performed using softwares Molinspiration and Osiris Data Explorer respectively, to assess the feasibility of selected phytoconstituents as potential drug candidates. The results were compared with two standard reference drugs doxorubicin hydrochloride (anticancer) and tetracycline (antibiotic). Multivariate data obtained were analyzed using principal component analysis (PCA). Results Docking analysis revealed that the binding affinities of the phytoconstituents towards the target cytoskeletal proteins decreased in the order coronin>villin>ezrin>vimentin>gelsolin>thymosin>cofilin. Glycoalkaloid solasonine displayed the greatest binding affinity towards the target proteins followed by alpha-solanine whereas amongst the saponins, nigrumnin-I showed maximum binding affinity. PASS Analysis of the selected phytoconstituents revealed 1 to 3 violations of Lipinski’s parameters indicating the need for modification of their structure-activity relationship (SAR) for improvement of their bioactivity and bioavailability. Glycoalkaloids and saponins all had bioactivity scores between −5.0 and 0.0 with respect to various receptor proteins and target enzymes. Solanidine, solasodine and solamargine had positive values of druglikeness which indicated that these compounds have the potential for development into future anticancer drugs. Toxicity potential evaluation revealed that glycoalkaloids and saponins had no toxicity, tumorigenicity or irritant effect(s). SAR analysis revealed that the number, type and location of sugar or the substitution of hydroxyl group on alkaloid backbone had an effect on the activity and that the presence of α-L-rhamnopyranose sugar at C-2 was critical for a compound to exhibit anticancer activity. Conclusion The present study revealed some cytoskeletal target(s) for S. nigrum phytoconstituents by docking analysis that have not been previously reported and thus warrant further investigations both in vitro and in vivo.

264 iGEMDOCK version 2.1, is a graphical, automated software developed by the Institute of 265 Bioinformatics, National Chiao Tung University, Taiwan for integrated docking, screening and post-266 analysis (Yang and Chen, 2004). For docking, preparation of protein and ligand files was done. With the 267 help of the software, binding sites for the ligand were defined and prepared. GEMDOCK uses a generic 268 evolutionary method (GA) to compute a ligand conformation and orientation relative to the binding site 269 of protein target. Therefore, docking performance is directly related to the selected GA parameters. The

326
The overall MolProbity score of homology models COR1A and TYB4

327
MolProbity score provides a single number that represents the overall protein quality statistics.
328 MolProbity score combines the clashscore, rotamer, and Ramachandran evaluations into a single score, 329 normalized to be on the same scale as X-ray resolution. Currently, the percentile scores are given for 330 clashscore and for MolProbity score relative to the cohort of PDB structures within 0.25A° of the file's 331 resolution. Figure 1(A) and (B) depict the modeled 3D structures of human coronin 1-A (COR1A) and 332 thymosin beta-4 (TYB4) respectively. In the present study, the molprobity scores of COR1A and TYB4 9 333 ( Figure 1) were determined to be 0.91 and 0.86 respectively, which were well within the allowed range 334 (Suppl. Tables 1 and 2). Ramachandran plots of the modeled COR1A revealed that 95% residues lay 335 inside the 'favored 98% contour' (Suppl. Figure 1) and there were 2 out of 393 Ramachandran outliers Manuscript to be reviewed 336 (Gly 129 and Val 229). In case of TYB4, 100% residues lay within the 'favored 98% contour' (Suppl. Figure   337 2) and there were 0 out of 34 Ramachandran outliers. Ramachandran plots for Gly, Pro and pre-Pro 338 residues have also been shown for both COR1A and TYB4. In case of COR1A, the C β deviation kinemage 339 showed each residue's C β position relative to an ideal C β and its three bond vectors (gray lines, Suppl. 340 Figure 3A). Circles mark the deviation distances, with the yellow circle at the 0.25 A ° cutoff for outliers.
341 Most of the distribution was found to be good, but Gly 129 and Val 229 formed part of an outlier cluster 342 and probably reflect distortions caused by a local fitting problem. In case of TYB4, there were no 343 outliers, and all residues lay within the yellow circle at the 0.25 A ° cutoff for outliers (Suppl. Figure 3B).  Table 2  387 of a majority of phytoconstituents from S. nigrum were found to be lower than those for ATP with 388 respect to the studied target protein(s). Suppl. Figure

555
In the present study, additional prospective targets of solasonine and solamargine have been 556 elucidated using molecular docking. It was found that both the spirosolane SGAs showed potent binding 557 to the newly proposed molecular targets ( Figure 2, Table 1). The mechanism of action may be related to 558 structural alteration/modification of cytoskeletal network, actin stabilization/destabilization, 559 intermediate filament assembly/disassembly, EMT inhibition along with induction of apoptosis.

561
This study is the first of its kind using in silico docking to explore the specific molecular targets to 562 which these compounds may bind prospectively in vivo and thereby exert their anticancer or 563 antibacterial activity. The study revealed some new targets for S. nigrum phytoconstituents, other than 564 those already known. Solanidine, solasodine, alpha-solanine, solasonine and uttroside B showed potent 565 binding to human coronin-1A, whereas degalactotigonin and nigrumnin-I showed strongest binding to 566 human thymosin beta-4. These phytoconsituents also showed strong bindings to other selected target 567 proteins as well, albeit to a lower extent as compared to these two proteins. However, there is a need   Best Docking Pose(s) of major components in S. nigrum with target proteins 1