Virtual screening of novel compounds as potential ER-alpha inhibitors

Majority of breast cancers diagnosed today are estrogen receptor (ER)-positive, however, progesterone receptor-positive (PR-positive) is also responsible for breast cancer. Tumors that are ER/PR-positive are much more likely to respond to hormone therapy than tumors that are ER/PR-negative. Nearly 105 ERa inhibitors from literature when docked resulted in 31 compounds (pyrazolo[1,5-a]pyrimidine analogs and chromen-2-one derivatives) with better binding affinities. The maximum score obtained was -175.282 kcal/mol for compound, [2-(4- Fluoro-phenylamino)-pyridin-3-yl]-{4-[2-phenyl-7- (3, 4, 5-trimethoxy-phenyl)-pyrazolo[1,5-a]pyrimidine-5-carbonyl]-piperazin-1-yl}-methanone. The major H-bond interactions are observed with Thr347. In pursuit to identify novel ERa inhibitory ligands, virtual screening was carried out by docking pyrazole, bipyrazole, thiazole, thiadiazole etc scaffold analogs from literature.34 bipyrazoles from literature revealed Compound 2, ethyl 5-amino-1-(5-amino-3-anilino-4-ethoxycarbonyl-pyrazol-1-yl)-3-anilino-pyrazole-4-carboxylate, with -175.9 kcal/mol binding affinity with the receptor, where a favourable H-bond was formed with Thr347.On the other hand, screening 2035 FDA approved drugs from Drug Bank database resulted in 11 drugs which showed better binding affinities than ERa bound tamoxifen. Consensus scoring using 5 scoring schemes such as Mol Dock score, mcule, SwissDock, Pose&Rank and DSX respectively resulted in better rank-sumsfor Lomitapide, Itraconazole, Cobicistat, Azilsartanmedoxomil, and Zafirlukast.


Background:
Majority of breast cancers diagnosed today are estrogen receptor (ER)-positive, where, estrogen binds to estrogen receptors on the surface of the cell [1]. According to the American Cancer Society, about 2 out of every 3 cases of breast cancer is hormone receptorpositive. However, in certain cases, progesterone receptor-positive (PR-positive) is also responsible for breast cancer [2]. Tumors that are ER/PR-positive are much more likely to respond to hormone therapy than tumors that are ER/PR-negative. ERα-positive breast cancer is more resistant to chemotherapy than ERα-negative cancer [3]. Estrogen-receptor status and outcomes of modern chemotherapy for patients with node-positive breast cancer is known. ERα plays an important role in determining the sensitivity of breast cancer cells to chemotherapeutic agents in vitro [4]. Down regulation of Aurora-A overrides estrogen-mediated growth and chemo resistance in breast cancer cells. Patients with ER-α-positive tumors have a slightly better survival rate than patients with ER-αnegative. However, both the ER and PR respond to the drug tamoxifen, designed to interfere the function of ER-α [5]. Tamoxifen decreases the incidence of invasive and non-invasive breast cancer. In spite of the tamoxifen administered side effects, its use as a breast cancer preventive agent is appropriate in many women at increased risk for the disease [6]. ER-α is thought to function as a ligand-activated transcription factor. Extracellular signals can also stimulate ER-α-mediated transcription in the absence of estrogen. Stimulated ER-α can influence gene expression by associating with other transcription factors without binding directly to DNA 322 ©Biomedical Informatics (2019) Estrogen receptor alpha rapidly activates the IGF-1 receptor pathway [7][8]. Specific binding sites for estrogen at the outer surfaces of isolated endometrial cells are known. Estrogens stimulate growth of many breast cancer cells. Reducing estrogen levels or blocking often leads to a clinical response in patients with receptor-positive disease. In premenopausal women, estrogen production is high and in postmenopausal women relatively small amounts of estrogens are produced. These low levels of estrogens can be inhibited either by blocking the estrogen receptor, or by inhibiting the peripheral conversion of androgens to estrogens [9]. The most widely accepted pharmacologic endocrine therapies for breast cancer are treatment with anti estrogens [10]. Tamoxifen has been shown to be effective in both premenopausal women as well as in postmenopausal women [11]. Tamoxifen is the most widely used and extensively studied anti estrogen and its role in the management of patients with breast cancer is well established [12]. However, extensive evaluation of tamoxifen treatment revealed significant side effects such as endometrial cancer, blood clots and the development of acquired resistance. Hence, there is a pressing need for the improvement and/or development of new antiestrogens for the prevention and treatment of breast cancer.  [15]. Moreover, a co-crystallized bound ligand represents better geometric orientation within the active site space of the protein. Therefore, the 3D structure of ERα bound with an antagonist, i.e. 4hydroxytamoxifen (PDB ID: 3ERT), was selected as the preferred docking target protein.

Molecular Docking Analysis:
Molecular docking is a study of non-bonded, non-covalent interactions between a receptor or active site region of a protein and a drug or chemical molecule forming an intermolecular complex [16]. Docking is carried out to dock various conformations of small molecules to a receptor followed by evaluation of the molecules with respect to the geometrical orientation and complementarity in terms of shape and properties, such as electrostatics [17]. The outcome of a docking routine includes affinity prediction (scoring) for the molecules investigated, yielding a relative rank ordering of the docked compounds with respect to affinity, reported as kcal/mol [18].

Molegro Virtual Docker:
Molegro Virtual Docker is an integrated platform for predicting protein -ligand interactions [19]. All default options including preparation of the molecules to determination of the potential binding sites of the target protein, and prediction of the binding modes of the ligands were employed.

Results and Discussion:
The crystal structure of human estrogen receptor alpha ligand binding domain in complex with 4-hydroxytamoxifen (PDB ID: 3ERT) was used for the docking. A thorough analysis of the X-ray crystal structure of estrogen receptor revealed that the active site regions has flexible amino acid side chains and hence could accommodate different chemical scaffolds. The amino acid residues lining active site are: Phe404, Glu419, Leu428, Met343, Gly420, Met421, Leu525, Gly521, Thr347, Leu387, Asp351, Ala350, Glu353, Trp383, Arg394, Leu346, respectively. The protein was prepared using Molegro software. All bond orders and hybridization were assigned, hydrogen and other missing atoms were added to the residues and charges were assigned. The co-crystallized water molecules were excluded from docking. Cavities in the protein were evaluated by Cavity detection algorithm using Expanded Van der Waals molecular surface with default parameters such as minimum and maximum cavity volume set at 10 Figure 1 and the h-bond interactions are given in Table 3.

Set-2: ERα Non-tested ligands from literature
A thorough literature search was made on structural features of ligands that would fit into the active site region of ERα, which resulted in pyrazole, bipyrazole, thiazole, thiadiazoleetc scaffold analogs. Bipyrazoles are known to possess inhibitory properties against several classes of enzymes. Moreover, preliminary docking analysis revealed better inhibition of ERα with bipyrazoles. Other classes of compounds displayed reduced inhibition. Hence, bipyrazoles are considered for further analysis.

Computational molecular docking and structural specificity of bipyrazoles as inhibitors of ERα
Docking of all 34 bipyrazoles from literature was carried out to evaluate the best conformer based on the lowest docked energy (kcal/mol) ( Table 4), in other words, it should possess highest affinity towards the binding site [42].
From the bipyrazole Vs ERα docking analysis output, it is evidenced that the bipyrazoles are able to bind and fit into the geometrical space provided by the active site region of ERα. The binding orientations of all bipyrazoles were similar to the cocrystallized ligand, tamoxifen (Figure 2). The best compound 2 Table-7 displayed a score of -175.9 kcal/mol which is much better than the ERα bound ligand (-149.8 kcal/mol). A favourable H-bond was formed with Thr347 ( Figure 3) as observed with chromene derivatives. The next best compound 29 resulted in dock score (-167.1 kcal/mol), however two favourable H-bonds were found to interact with compound 29, via Thr347 (Figure 4).

Set-3: Drugs from DrugBank Database
Owing to the output from bipyrazole dataset, which showed better inhibitory than tamoxifen, the next step utilized was to search DrugBank database because it was observed that certain drugs which are specific against a particular disease were found to be effective against other disease conditions as well, for example, Pioglitazone, a drug used for type 2 diabetes, may prevent recurrent stroke and heart attacks in people with insulin resistance but without diabetes [43][44]. Several studies indicate that persons with type-2 diabetes are at higher risk of cancer of the pancreas, liver, endometrium, breast, colon, rectum and urinary bladder [45]. however, the use of metformin was associated with decreased risk of the occurrence of various types of cancers, especially of pancreas and colon and hepatocellular carcinoma [46] evidence suggested that metformin might reduce breast cancer incidence in postmenopausal women [47] In another study, by screening already approved drugs, researchers identified calcium channel blockers, which are used to treat hypertension, can efficiently stop cancer cell invasion in vitro [48]. Preliminary investigations revealed that Gleevec blocked the progression and development of rheumatoid arthritis in laboratory mice [49]. Therefore, in this context DrugBank database was accessed to select 2035 FDA approved drugs and subjected to molecular docking. Analysis resulted in 15 drugs, which showed better binding affinities than ERα bound tamoxifen, tabulated in Table 5.   Figure 5 where it is evidenced that all drugs occupied clearly within the geometric space of the protein. From the table, out of 15 drugs, only 11 are finalized to consider for further analysis. This is because the four drugs viz., Bazedoxifene, Lapatinib, Raloxifene and Dabrafenib found to be anti-cancer drugs and hence omitted from the list.

Consensus Scoring to enrich drugs active against ERα:
It has been reported recently that consensus scoring, which combines multiple scoring functions, leads to higher hit-rates in virtual library screening studies [50] and presented an idealized computer experiment to explore how consensus scoring works based on the assumption that the error of a scoring function is a random number in a normal distribution. Many studies suggested that implementing consensus-scoring approaches enhances the performance by compensating for the deficiencies of the scoring functions with each other [51] [52] [53] The possibility that several scoring methods might have their own strengths and weaknesses and combined use of more than one method might increase the overall signal-noise ratio and might perform better than the average of the individual scoring functions [54] presented computer-aided analysis where they implemented an intersection-based consensus approach to group few scoring functions. Stahl and Rarey [55] reported the performance of four scoring functions on seven target proteins.
Screening analysis of DrugBank database drugs against ERα resulted in 11 drugs and all these drugs are subjected to consensus scoring using 5 scoring schemes such as MolDock score of Molegro, mcule, SwissDock, Pose & Rank and DSX respectively. Here, we chose the "rank-by-number" strategy to pool the output of multiple scoring functions. This is because, this strategy was reported to outperform the other techniques such as "rank-by-rank" and "rankby-vote" as the rank-by-number strategy summarized most of the information [56] Each scoring function was applied to generate three classes based on the obtained dock scores followed by ranking the best conformations. Classes were generated for all scoring functions and instead of taking an average, rank-bynumber technique [57] was employed to finalize best compounds. The ranks obtained from each of the scoring functions were added to give the rank-sum. The benefit of rank-by-number technique is that the each individual score involvement for a rank can certainly be split out for illustrative purposes [58]. The rank sums obtained for 11 drugs against five scoring functions were in the range 5 to 15, with 5 being low rank and 15 being first and best rank, respectively (Table 6). Therefore, finally from 11 drugs, the top five compounds with rank-sums 15 -12 (Lomitapide, Itraconazole, Cobicistat, Azilsartanmedoxomil, and Zafirlukast) are finalized. Further work shall be carried out to study their affinity of binding and inhibitory characteristics against ERα in a breast cancer cell line MCF-7.