Natural compounds as angiogenic enzyme thymidine phosphorylase inhibitors: In vitro biochemical inhibition, mechanistic, and in silico modeling studies

Natural flora is the richest source of novel therapeutic agents due to their immense chemical diversity and novel biological properties. In this regard, eighteen natural products belonging to different chemical classes were evaluated for their thymidine phosphorylase (TP) inhibitory activity. TP shares identity with an angiogenic protein platelet derived endothelial cell growth factor (PD-ECGF). It assists tumor angiogenesis and is a key player in cancer progression, thus an ideal target to develop anti-angiogenic drugs. Eleven compounds 1–2, 5–10, 11, 15, and 18 showed a good to weak TP inhibitory activity (IC50 values between 44.0 to 420.3 μM), as compared to standards i.e. tipiracil (IC50 = 0.014 ± 0.002 μM) and 7-deazaxanthine (IC50 = 41.0 ± 1.63 μM). Kinetic studies were also performed on active compounds, in order to deduce the mechanism of ligand binding to enzyme. To get further insight into receptor protein (enzyme) and ligand interaction at atomic level, in- sillico studies were also performed. Active compounds were finally evaluated for cytotoxicity test against mouse fibroblast (3T3) cell line. Compound 18 (Masoprocol) showed a significant TP inhibitory activity (IC50 = 44.0 ± 0.5 μM). Kinetic studies showed that it inhibits the enzyme in a competitive manner (Ki = 25.6 ± 0.008 μM), while it adopts a binding pose different than the substrate thymidine. It is further found to be non-toxic in MTT cytotoxicity assay. This is the first report on TP inhibitory activity of several natural compounds, some of which may serve as leads for further research towards drug the development.


Introduction
Cancer is one of the primary causes of death worldwide, with approximately one in six person dies because of some malignant conditions. In the last few decades, early diagnosis and treatment have improved the patient's life quality, and survival rates. However, there are still high rates of recurrence, invasiveness, and metastases. To improve efficacy, new therapeutic interventions are required which target different and important stages of tumor progression. In this context, angiogenesis has emerged as a hallmark of tumor growth. Since the last five decades, it has been hypothesized that blocking angiogenesis could be an effective way to combat the disease's progression. Thenceforth, a large number of molecules, targeting the process of angiogenesis, have been under pre-clinical and clinical trials. Some of these molecules are already approved as drugs by the U.S. Food and Drug Administration (FDA). Thymidine phosphorylase (TP) is among the several factors that stimulate the growth of blood vessels, and it described as a validated target for the antiangiogenic compounds development [1].
Enzyme thymidine phosphorylase (TP) is involved in pyrimidine salvage pathway, regulating the nucleotide homeostasis which is required for DNA repair and replication. It catalyzes the reversible phosphorolysis of thymidine to thymine and 2-deoxyribose-1-phosphate. 2-Deoxyribose-1-phosphate then undergoes dephosphorylation to 2-deoxyribose, which actually triggers tumor angiogenesis. 2-Deoxyribose affects endothelial-cell migration by activating the downstream integrin signaling pathway. It also augment the expression and/or release of several angiogenic growth factors, such as VEGF, MMPs, cytokines, interleukins, and others, which in the tumor microenvironment enhance the angiogenesis, and cancer metastasis. TP is reported to share sequence homology with an angiogenic growth factor (Platelet derived endothelial cell growth factor; PD-ECGF), which promotes angiogenesis by facilitating endothelial cells proliferation, and migration [2,3].
Our research group has previously reported several synthetic and few natural inhibitors of TP [21][22][23][24][25][26]. In the present study, an attempt was made to screen selected natural compounds for their TP inhibitory potential. Natural compounds present a long-standing tradition as valuable starting points in drug discovery program. Structural diversity makes them privileged sources for chemical probes. Medicinal plants are rich source of producing natural products and many of these are used as active ingredient of modern medicine [27][28]. Only less than 10% of the natural flora is investigated so far, therefore, nature holds a great promise for discovery of leads against most common, and emerging diseases [29].
Based on reported medicinal importance particularly the anti-cancer activity, some natural compounds belonging to class coumarin, flavonoid, carboxylic acid, alkaloid, and lignan were selected randomly and evaluated for their TP inhibitory activity [30][31][32][33][34][35][36][37]. These natural compounds are not reported earlier for the TP inhibitory activity. Isolation and spectroscopic data for the compounds are presented in the S1 Table. The current study led to the identification of new potential TP inhibitors, as summarized in Figs 1-4.  assayed by HPLC techniques (See S1 Information and S1 Table for details about extraction and spectroscopic data).
Thymidine phosphorylase inhibition assay. In the current study, we have used commercially available recombinant E. coli TP enzyme, as human TP is not easily accessible. Substantial similarities in terms of structural and active site residues exist between E. coli and mammalian TP enzymes, therefore E. coli TP generally serves as a primary model for the identification of lead inhibitors of TP [3].
Thymidine phosphorylase inhibition assay was carried out spectrophotometrically [45]. Briefly, 0.058 U of TP enzyme (E.C. No. 2.4.2.4, E. coli) was incubated with 500 μM of test compounds at 30˚C for 10 minutes. After that 1.5 mM of substrate thymidine was added to 96-well plate, and changes in absorbance were monitored for 10 minutes at 290 nm in microtiter plate reader (Spectramax, M5, Molecular Devices, CA, USA). Enzyme and substrate solution were prepared in 50 mM phosphate buffer of pH 7.0, while test compounds were prepared in DMSO (final concentration 4.5%). Tipiracil and 7-deazaxanthine were used as standard inhibitors, and each experiment was performed in triplicate.
Mechanistic studies. In mechanistic studies, TP enzyme was incubated with different concentrations of inhibitors at 30˚C for 10 min. Reaction was then initiated by adding four different concentrations of substrate, i.e. thymidine (0.1875-1.5 mM). Enzyme activity was measured under steady state conditions by measuring changes in absorbance for another 10 min at 30˚C at 290 nm, on microtiter plate reader (Spectramax, M5, Molecular Devices, CA, USA). Every experiment is performed in triplicate.

Molecular docking studies
Ligand preparation. Prior to docking, the molecules (ligands) were processed via Ligprep module in Maestro Schrödinger 10.5. It involves the generation of low energy 3-D structures from 2-D structures of compounds in SD format. Possible ionization states and correct protonation were generated using Epik module which predict the tautomeric state, and generate energetic penalties for each molecule conformation it predicts [46].
Protein preparation. X-Ray crystallographic structure of E. coli TP was used for docking studies (PDB ID: 4LHM). Maestro Schrödinger software was used to prepared protein by employing the Protein Preparation Wizard 10.5 [47,48]. OPLS-2005 force field was used to add missing hydrogens, and for the assignment of partial charges. Optimization of heavy atoms and hydrogens was then carried out by subjecting the structure to restrained minimization. The co-crystallized water molecules were retained because they were present in the active site, involving the formation of general hydrogen bond network. Since the sulfate ion was replaced with phosphate, it occupied the same place in active site in crystal structure as that of phosphate ion.
Searching for allosteric binding sites and molecular docking analysis. To find out the allosteric site for non-competitive and uncompetitive inhibitors, site recognition software Site-Map 3.7 [49,50] Maestro version 10.5 from Schrödinger was run on crystal structure to identify the top 5 ranked potential ligand-binding pockets.
The grid box with dimensions of 15Ǻ x 15Ǻ x 15Ǻ was defined to confine the mass of centre of each docked ligand. Extra precision (XP) mode of Glide based on OPLS-2005 force field was run for rigid receptor docking protocol [51][52][53][54]. Molecular mechanics-generalized Born surface area (MM-GBSA) method in Prime was used for rescoring the docked pose of ligand [55]. These poses were taken as inputs for the energy minimization of the protein-ligand complexes (E complex ), the free protein (E receptor ), and the free ligands (E ligand ). The binding free energy ΔG bind was determined according to the following equation:

In vitro cytotoxicity assay
Cytotoxicity of active compounds was evaluated spectrophotometrically by standard MTT (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyl-tetrazolium bromide) assay, following the method of Demas et al. In cytotoxicity assay, reduction of MTT dye to formazan by mitochondrial enzyme (succcinate dehydogenase) is measured. The reduction of MTT can only occur in metabolically active cell, so the enzyme activity is actually a measure of cell viability [24,56]. Briefly, the mouse fibroblast cells (3T3) were cultured in DMEM media containing 5% FBS, 100 IU/ mL penicillin, and 100 μg/ mL streptomycin. Cell suspension (5×10 4 cells/ mL) was dispensed in flat-bottomed 96-well plates, and incubated at 37˚C and with 5% CO 2 . After the overnight incubation, old media was aspirated and fresh media containing different concentrations of compounds (prepared in DMSO) were added, and plate was further incubated for 48 hrs. After that, MTT dye (0.5 mg/ mL) was added into each well, and the plate was further incubated for 4 hrs. Following this, DMSO (100 μL) was added into each well, and the level of MTT reduction to formazan with in the cells was calculated by taking the absorbance at 540 nm using a micro plate reader (Spectra Max plus, Molecular Devices, CA, USA).

Analysis of the experiments
In vitro TP spectrophotometric assay. Enzyme activity was measured under steady-state conditions by observing changes in absorbance for 10 min (at 290 nm) on microtiter plate reader (Spectra Max M5, Molecular Devices, CA, USA). Changes in absorbance upon addition of test compounds were analyzed with the SoftMax Pro 4.8 software which was purchased from Molecular Devices, CA, USA. To calculate the percent inhibition, following formula was used: Inhibitory potential of test compounds is represented in Figs 1-4 as IC 50 value (Inhibitory concentration). The IC 50 represents concentration of compounds that inhibit the degradation of thymidine to thymine by 50%. It was calculated by measuring the effects of different concentrations of inhibitors on the degradation of thymidine. EZ-Fit, Enzyme kinetics software (purchased from Perrella Scientific, Inc., USA) was used to deduce the IC 50 values.

Analysis of mechanistic assay
Mechanistic studies were performed to find out the mechanism of inhibition of active compounds. Lineweaver-Burk plot was used to determine the type of inhibition, where the reciprocal of rate of the reaction were plotted against the reciprocal of substrate concentration. Dissociation constant (K i ) were determined by secondary re-plot of Lineweaver-Burk plots. Kinetic data was analysed by GraFit 7 data analysis software (purchased from Erithacus Software Limited, UK) [57].

Analysis of docking studies
To find out the allosteric site for non-competitive and uncompetitive inhibitors, site recognition software SiteMap 3.7 Maestro version 10.5 from Schrödinger was run on crystal structure to identify the top 5 ranked potential ligand-binding pockets. Molecular mechanics-generalized Born surface area (MM-GBSA) method in Prime was used for rescoring the docked pose of ligand. The binding free energy ΔG bind was determined according to the-equation given earlier.

Analysis of cytotoxicity assay
The cytotoxicity was recorded as concentration causing 50% growth inhibition (IC 50 ) for 3T3 cells. Cellular viability was measured for 10 min, at 570 nm on microtiter plate reader. Changes in absorbance upon addition of test compounds were analyzed with the SoftMax Pro 4.8 software which was purchased from Molecular Devices, CA, USA. Percent inhibition was calculated using the formula mentioned earlier.
Cytotoxicity potential of test compounds is represented in Figs 1-4 as IC 50 values. The IC 50 represents concentration of compounds that inhibit the cell growth by 50%. It was calculated by measuring the effects of different concentrations of inhibitors on the growth of cells. EZ-Fit, Enzyme kinetics software was used to deduce the IC 50 values.

Results and discussion
Eighteen secondary metabolites of medicinal plant origin were evaluated for their in vitro TP inhibitory activity via biochemical method. Eleven compounds 1, 2, 5-10, 11, 15, and 18 exhibited good to weak TP inhibitory activity with IC 50 values between 44.0 to 420.3 μM, as compared to the standard inhibitor, 7-deazaxanthine (7-DX) (IC 50 = 41.0 ± 1.63 μM). The results indicate that these compounds are effective at high micromolar concentration thus could serve as starting point for the development of new anti-TP molecules. Limited structureactivity relationship was carried out (explained below) as the current study is based on the small number of compounds.

Structure-activity relationship of natural compounds
Among compounds 1-5 belonging to flavonoid class, 1, 2, and 5 showed a good to weak TP inhibiting ability with IC 50 values of 76.8 to 195.1 μM, as compared to standard 7-deazaxanthine (IC 50 = 41.0 ± 1.63 μM), while compounds 3, and 4 were found to be inactive.
Compound 1 with a basic flavonoid skeleton showed a weak TP inhibiting activity (IC 50 = 195.1 ± 1.1 μM). Compound 2 with OH groups at C-5, C-2', and C-3', and OCH 3 groups at C-7 and C-6' showed a good TP inhibition (IC 50 = 113.0 ± 0.9 μM), as compared to compound 1. However, presence of OH groups at C-5, C-7 and C-4', as in compound 3, and at C-5 and C-7, as in compound 4 were found unfavourable for TP inhibition. Presence of OCH 3 at C-6 and OH at C-3, C-5, C-7, C-4', and C-5' in compound 5 enhanced the TP inhibitory potential (IC 50 = 76.8 ± 1.1 μM). Structure-activity relationship (SAR) indicated that number and position of electron donating groups on flavonoid nucleus play role in TP inhibition. Hydroxyl and methoxy groups are expected to form hydrogen bonds and/or hydrophobic interactions with amino acid residues at active site or hydrophobic pocket of the TP enzyme.
Compounds 6-10 of coumarins class showed a weak TP inhibiting activity with IC 50 values in the range of 181.1-420.3 μM, as compared to the standard 7-deazaxanthine (IC 50 = 41.0 ± 1.63 μM) ( Table 1). Among the group, compound 6 showed the weakest TP inhibition (IC 50 = 420.3 ± 6.15 μM). Presence of furan and methoxy moieties at coumarin skeleton in compound 7 increased the inhibitory potency (IC 50 = 344.2 ± 1.2 μM), as compared to compound 6. TP inhibition increased further due to the presence of pyran and di-methyl groups at coumarin skeleton, as in compound 8 (IC 50 = 181.1 ± 0.6 μM). Compound 9 with OH and OCH 3 groups on coumarin skeleton also showed the increased TP inhibitory potential in comparison to compound 6 (IC 50 = 193.4 ± 1.0 μM). However, replacement of hydroxyl group at C-7 with apiose glucoside moiety in compound 10 decreased the TP inhibition potency (IC 50 = 295.2 ± 2.0 μM). This indicates that OH group is critical for inducing enzyme inhibition.
SAR suggested that compound 6 (coumarin) is mainly interacting with the enzyme via hydrophobic interaction with critical residues at the active site or hydrophobic pocket of enzyme TP. The presence of other electron donating groups (hydroxyl and methoxy) in compounds 7-10, found to enhance their potential of inhibiting the TP enzyme, in comparison to compound 6. Hydroxyl may likely to be involve in forming hydrogen bond with the amino acids at active site or hydrophobic pocket of the TP enzyme.
Among the alkaloids 11-14, compound 11 exhibited a weak TP inhibitory activity (IC 50 = 238.5 ± 2.1 μM). Methoxy groups of compound 11 proposed to interact with the enzyme's hydrophobic pocket via hydrogen bonding with critical amino acid residues. Compounds 12-14 were found to be inactive against the TP. Among the compounds 15-17 belonging to benzoic acid class, only compound 15 showed a weak TP inhibitory activity with IC 50 of 349.7 ± 1.4 μM. This compound is likely to be engaged in hydrogen bonding with the residues present at the active site or hydrophobic pocket of the enzyme. However, the phenyl moiety can also be involved in π-π interaction with aromatic residues. Compounds 16, and 17 were found to be inactive.
Compound 18, a lignan, exhibited a significant TP inhibitory activity (IC 50 = 44.0 ± 0.5 μM). Hydroxyl groups are apparently forming hydrogen bonds with critical amino acid residues of the active site.

Mechanistic studies
Mechanistic studies were performed on the selected natural products. It was found that they inhibit TP enzyme in a concentration-dependent manner with dissociation constant (Ki) values between 25.6-419.0 μM. Natural products 5, 9, and 18 exhibited mixed, non-competitive, and competitive modes of inhibition, respectively, while natural compounds 1, 6, 8, 10, and 11 showed an uncompetitive type of inhibition (Table 1, Figs 5-8). Lineweaver-Burk plot was plotted to determine the inhibition type, secondary re-plot of Lineweaver-Burk plot was used to deduce (Ki) values.

Molecular docking studies
Sitemap analysis was run for non-competitive, uncompetitive, and mixed inhibitors to identify the potential allosteric binding site of the enzyme. Among the 5 top ranked sites, the druggable site with highest site score of 1.05 was chosen as allosteric site with respect to the thymidine for docking studies.
The docking analysis revealed that all of the in silico predicted lowest energy complexes were stabilized by stacking interaction, and intermolecular hydrogen bonds. Compound 1 having the primary flavonoid scaffold, form hydrogen bonds with Arg 171, while compound 5 with OH and OCH 3 substitutions form π-cation interaction with Arg 171. π-π interactions were also observed with Phe 210 (Fig 9A and 9B). Thymidine phosphorylase inhibitors of natural origin MMGBSA score provides the predictive binding affinity of the ligand binding to the protein. A more negative score is indicative of better binding affinity towards corresponding ligand molecule. Analysis of MMGBSA score revealed that compounds 1, and 5 have predictive binding energies of -40.45, and -61.99 kcal/mol, respectively, which is in agreement with their experimental Ki values i.e., 225 and 150 μM, respectively.
Molecular docking of coumarins with TP showed that compound 6 form only π-π interaction through its pyran ring with Phe210 ( Fig 10A). The dimethyl pyran ring in compound 8 made π-cation interaction with Arg171 ( Fig 10B). Presence of hydroxyl and methoxy groups in compound 9 form hydrogen bonds with Arg171, and Ser186, along with π-π interaction with Phe 210 (Fig 10C). Compound 10 showed π-cation interaction via its benzyl moiety to    Molecular docking studies also showed that compound 11 had only one hydrogen bond with Lys 190 via one of its dimethoxy benzyl moieties (Fig 11). The predictive binding energy (ΔG bind ) was found to be -41.79 kcal/mol. Compound 18 of carboxylic acid class was found to be competing with thymidine for binding (competitive inhibitor); therefore it was docked into the substrate binding site of TP. Analysis of docking result showed that compound 18 adopted a different binding pose than substrate thymidine. One of the dihydroxy benzyl moieties interacted via π-cation interaction to Tyr 168. While the attached OH groups form hydrogen bonds with Arg 171, and Ile 183. The other dihyroxy benzyl group form hydrogen bonds with phosphate ion in substrate binding site (Fig 12). The compound was rescored via MMGBSA, and turned out to give predictive binding energy of about -60 kcal/mol.

Cytotoxicity assay
Compounds 1, 6-11, 15, and 18 were finally subjected to cytotoxicity MTT assay on 3T3 cell line. Compound 1 showed a weak growth inhibition of 3T3 cells (IC 50 = 29.05 ± 0.61μM) in comparison to standard cycloheximide (IC 50 = 0.20 ± 0.10 μM), while compounds 6-11, 15, and 18 were found to be non-cytotoxic. These compounds are, therefore, promising leads to be investigated further for in vivo anti-cancer and other relevant therapeutic activities.

Conclusions
Many solid tumors have over-expression of thymidine phosphorylase (TP). TP assists cancer progression by inducing angiogenesis and preventing apoptosis. These critical functions make it an ideal target for the development of anti-angiogenic compounds. This systematic study identifies several natural products as anti-TP molecules with inhibitory activity good to weak, in comparison to the standard 7-deazaxanthine. Masoprocol (18), a dicatechol isolated from the Larrea divaricata Cav., showed the significant TP inhibitory activity, among all the natural products evaluated. It was found to interact with the active site residues of TP in kinetic studies (competitive inhibitor); however it exhibited a different binding pose than the natural