In silico design and cell-based evaluation of two dual anti breast cancer compounds targeting Bcl-2 and GPER

According to WHO statistics, breast cancer (BC) disease represents about 2.3 million diagnosed and 685,000 deaths globally. Regarding histological classification of BC, the Estrogen (ER) and Progesterone (PR) receptors negative-expression cancer, named Triple-Negative BC (TNBC), represents the most aggressive type of this disease, making it a challenge for drug discovery. In this context, our research group, applying a well-established Virtual Screening (VS) protocol, in addition to docking and molecular dynamics simulations studies, yielded two ligands identified as 6 and 37 which were chemically synthesized and evaluated on MCF-7 and MDA-MB-231 cancer cell lines. Strikingly, 37 assayed on MDA-MB-231 (a TNBC cell model) depicted an outstanding value of 18.66 μM much lower than 65.67 μM yielded by Gossypol Bcl-2 inhibitor whose main disadvantage is to produce multiple toxic effects. Highlighted above, enforce the premise of the computational tools to find new therapeutic options against the most aggressive forms of breast cancer, as the results herein showed.

Deregulated apoptosis is a feature now considered as a hallmark of cancer cells.It results from an adaptation to high levels of oncogenic signaling by two main processes: disabling their senescence and de-regulating apoptosisinducing circuitry 1,2 .Rewired apoptosis machinery considers modification of the Bcl-2 regulatory proteins.Bcl-2 protein (archetype of the family) and other members as Bcl-x L , Bcl-w, and Mcl-1 A1, inhibit apoptosis by binding (in part) to two proapoptotic triggering proteins (Bax and Bak) 3,4 .Certainly, death or cell survival depends largely on the ratio of Bcl-2 family anti-and pro-apoptotic proteins.As occurs in most tumor and cancer cell lines, there is an overexpression of the anti-apoptotic Bcl-2 proteins, for example, it is well-known that prostate, colorectal and breast cancers overexpressed Bcl-2 proteins [5][6][7] .Besides, although there are potent inhibitors of the Bcl-2 family proteins as ABT-737, its molecular structure contributes to its poor drug-profile related to stability, affinity, and metabolism, which led to its failure in preclinical studies 8,9 .In the same way, Gossypol, a polyphenol derived from the cottonseed plant, has demonstrated behavior as pan-Bcl-2 family inhibitor, which, in combination with ABT-737 has overcoming resistance assayed on a clinically relevant leukemia model evaluating a Bcl-2 inhibitor alone 10 .Recently, a new estrogen membrane-receptor named GPER has been described in breast cancer cells, which promotes both cancer cell viability and Tamoxifen-resistance 11,12 .Several studies have strengthened the premise of the pro-oncogenic role of GPER.Firstly, it is required the GPER-activation for survival of xenograft-derived cancer stem cells and progression to metastatic disease, secondly, GPER leads

Rational design of new dual GPER/Bcl-2 ligands
On the chemical basis of a GPER pharmacophore composed of a tetrahydroquinoline scaffold, we previously developed a GPER ligand named G1-PABA [26][27][28] which contains in its structure a carboxylic acid group capable of being modified by an amide bond formation.Taking advantage of that, we proposed functionalizing it by forming an amide bond with nucleophilic amines, which contain chemical moieties in its structure, capable of reaching the Bcl-2 binding site 29,30 and, consequently, blocks both GPER and Bcl-2 aberrant activity.

Virtual screening protocol
Following a well-established protocol developed by our work group, we based our virtual screening (VS) procedure on the analysis of several features ranging from protein structural parameters as binding modes and energetics yielded by molecular docking, and physicochemical, toxic-biological, and metabolism-prediction by combining various online servers, to select the most promising ligands for their subsequent obtaining by chemical synthesis.In detail, the most relevant values for the VS protocol were the yielded by docking simulations; in this sense, we set for that the binding energy value (in absolute value = |BE| ) and percentage of similarity in the bind- ing mode value ( %similarityBM ), which is the result of the comparison of the tested ligands versus compounds with a demonstrable affinity towards our protein targets, for such control compounds were G1-PABA for GPER and Gossypol for Bcl-2, such analysis considered the results yielded from the following equation: Results of both parameters are defined as |BE| and %similarityBM were summative to the final score for the selection of the best ranked ligands.Besides, following the VS protocol, we considered the physicochemical properties analyzed with Molinspiration server https:// www.molin spira tion.com/, considering for that, the parameters encompassed into Lipinsky rules.Accordingly, for the final score we considered the values yielded by the tested ligands according to: molecular weight 200-500 g/mol, ligands with values out this range would add 0 points, meanwhile, ligands with values into this range would add 10 points; Log P 2-5, ligands with values out this range would add 0 points, meanwhile, ligands with values into this range would add 10 points; H bond acceptors 5-10, ligands with values out this range would add 0 points, meanwhile, ligands with values into this range would add 10 points; H bond donors 2-5, ligands with values out this range would add 0 points, meanwhile, ligands with values into this range would add 10 points.Regarding to toxic-biological properties, these were examined with Osiris Property explorer https:// www.organ ic-chemi stry.org/ prog/ peo/ and considered for the VS protocol setting for all the tested parameters (mutagenicity, teratogenicity, tumorigenicity and irritability) the follow scoring: ligands with high risk would add 0 points, intermedium risk would add 10 points and low risk would add 20 points.Ultimately, all the concerning to predictive properties of the first step metabolism (mediated by the super metabolizing family enzymes CYP450) was conducted with Metaprint2D included in Bioeclipse % similarityBM = interacting residues of control compounds − 100 interacting residues of tested ligands − X

Ligand construction and molecular docking simulations
Ligand drawing was carried out employing ACD/ChemSketch (Freeware) 2020.1.2 31generating in this software a previous 3D pre-optimization to each ligand .molfile, conversion of the .molfile to a Z-matrix was conducted with GaussView 5.0 32 setting an AM1 semiempirical method to develop a complete energy and geometrical minimization, and a correct proton assignation to each atom that requires it, employing the Gaussian 09 33 quantic-chemical package, the resulting output files were employed as .pdbinput files to molecular docking simulations.Molecular docking simulations were conducted with AutoDock 4.2.6 34 software employing GPER and Bcl-2 3D models previous generated and validated by our research group and published elsewhere [26][27][28][29][30] , setting the follow parameters prior to docking simulations: polar hydrogens were adequately placed over the atoms capable of establishing polar contacts in the ligand/receptor complexes, and Kollman charges were assigned to the whole proteins.For the ligands, Gasteiger charges were assigned, and the affinity maps were generated using a grid-based procedure delimiting a grid box covering most of the protein with dimensions of 126 Å 3 in each XYZ coordinates (blind docking procedure) with a grid spacing of 0.375 Å 3 .As scoring sampling was selected the Lamarckian genetic algorithm including a randomized initial population of 100 individuals delimiting a unique as top individual that survived, and a maximum number of energy evaluations of 1 × 10 7 cycles, all above as search parameters.The chemical interactions formed in the ligand/receptor complex, were analyzed with Pymol v0.99 35 molecular viewer and Discovery Studio 2020 36 .

Physicochemical and ADME-Tox properties analysis and metabolism prediction
Lipinski's rule of five is a statement of thumb which encompasses several parameters aimed to evaluate, for compounds designed de novo, drug-likeness, or the possibility to be orally administered.Thereby, physicochemical properties of the designed ligands were assessment employing Molinspiration web server (https:// www.molin spira tion.com/), emphasizing in key attributes of their chemical structures which could affect its water-lipid solubilization and biological activity such as: molecular weight, Log P, h-bond donors and acceptors and total polar surface area (TPSA).Aside, toxic-biological risks that influence on the safety of the designed ligands, were analyzed using Osiris Property Explorer (https:// www.organ ic-chemi stry.org/ prog/ peo/), pointing in such features as mutagenic, tumorigenic, and irritant and reproductive effects.Results yielded of these analyses (physicochemical and ADME-Tox) were included as summation scores for the virtual screening and are included as part of Table S1 on Supplementary Information.Furthermore, metabolism mediated by the CYP450 metabolic enzyme superfamily, was conducted with MetaPrint2D included in Bioeclipse 2.6.2 (https:// www.biocl ipse.net/ metap rint2d), for that, were only included as possible metabolism sites, those contained in the moiety added by the amide bond formation (nucleophilic amines or as appropriate, with possibility to reach the Bcl-2 binding sites).Results yielded of this analysis (metabolism prediction) were included as subtraction scores for the virtual screening and are included as part of Table S1 on Supplementary Information.

Molecular dynamics simulations with MMGBSA approach.
Anchoring of the GPER-ligand into the membrane Best ligands resulting from VS protocol and subsequently obtained by chemical synthesis, were employed to carried out molecular dynamics simulations, thus, compound 6 and 37 orientation with respect to the bilayer membrane was carried out with OPM (Orientations of Proteins in Membranes) server 37 .The oriented complexes were immersed into a preequilibrated 320 POPC phospholipids bilayer using the membrane-builder tool of CHARM 38,39 with the following xyz dimensions 110.437 × 110.437 × 122.485Å.The GPER-ligand-membrane systems were solvated using the TIP3 water model and neutralized with an ionic strength of 0.15 M NaCl.

MD simulations
The GPER-ligand-membrane and Bcl2-ligand systems were submitted to 100 ns (ns) MD simulations using the Amber 16 package 40 .Ligand forcefields were constructed considering the Generalized Amber Force Field (GAFF) and the AM1-BCC method.Topology for the GPER-ligand-membrane systems were constructed with a Leap module using ff14SB 41 , Lipid14 42 and GAFF 43 .Topology for the Bcl2-ligand systems were constructed with a Leap module using ff14SB 41 , and GAFF 43 .MD simulations were performed under periodic boundary conditions using an NPT ensemble at 310 K and 1 bar pressure.Long-range electrostatic interactions were determined with the PME 44 , considering a 10 Å cutoff for van der Waals interactions.Bond lengths were restricted at the equilibrium using the SHAKE algorithm 45 .The temperature was kept using Langevin dynamics.For GPER-ligand-membrane systems the pressure was preserved using a semi-isotropic constant surface tension to maintain the area per lipid.The area per lipid, root means squared deviation (RMSD), radius of gyration (Rg) and clustering examination were carried out using the cpptraj tool in Amber16.Maps of contacts in 2D from the ligand-receptor analysis were made using Maestro Schrödinger version10.5 46.The Molecular Mechanics Generalized Born Surface Area (MMGBSA) 47 approach was considered to determine the binding free energy (ΔGbind).The last 40 ns of converged simulation were considered, gathering a total of 4000 snapshots, utilizing implicit solvation models 48 , and a 0.15 M NaCl concentration.The ΔG bind values were determined as previously described 49 .

Chemical synthesis
All the solvents and reagents were chemical grade obtained from commercial suppliers and employed without further purification.Progress of the reactions were monitored by TLC (thin-layer chromatography) on aluminum-backed sheets with silica gel 60 GF254 (HX805651) and a fluorescent indicator (visualized with UV light of 254 nm).Flash chromatography was performed using silica gel 60 (230-400 mesh) for further purification of the products obtained.Melting points (mp) were determined on an Electrothermal IA 91,000 apparatus (Electrothermal, Bibby Scientific, Staffordshire, ST15 OSA, UK), and are reported as uncorrected.IR spectroscopy was obtained on a Perkin-Elmer 16 FPC FT infrared spectrophotometer and only significant absorption bands are reported as wavenumbers (cm-1) for the representative chemical groups. 1 H and 13 C NMR spectra were obtained on a Varian Mercury 300 MHz spectrometer or a Bruker Avance III 750 MHz spectrometer using deuterated chloroform (CDCl3) as the solvent and TMS as the internal standard.Chemical shifts (δ) are reported in ppm downfield from the internal standard and coupling constants are reported in Hertz (Hz) (supplementary material).Electrospray ionization high resolution mass spectrometry positive mode (ESI-HRMS) was performed with an Agilent 6545 QTOF LC/MS instrument (Agilent Technologies, Santa Clara CA, United States).

Cell culture
Antiproliferative activity evaluation of the synthesized compounds was assayed in two breast cancer cell lines.Thus, MCF-7 and MDA-MB-231 (American Type Tissue Culture Collection (ATTC), Rockville, MD, USA) were grown in Dulbecco′s Modified Eagle′s Medium (DMEM) media (Life Technologies, Gaithersburg, MD, USA), with 5% FBS (BioWest, Miami, FL, USA), 2 mM glutamine, 100 U/mL penicillin, and 100 mg/mL streptomycin.Cell cultures were incubated at 37 °C in a humidified atmosphere supplemented with 5% CO 2 and 95% air.Prior to the treatment of the cells, MDA-MB231 and MCF-7 were grown in phenol red-free DMEM media (Life Technologies, Gaithersburg, MD, USA) containing 5% FBS.The FBS was charcoal-stripped (to eliminate the estrogenic effects) only for the MCF-7 cells.

Results and discussion
Virtual screening converge on two potential GPER and Bcl-2 dual-inhibitors Results yielded from the series of combined protocols applied to obtain potential inhibitors of GPER and Bcl-2 showed crucial chemical features of the best ranked molecules which are found in other compounds with antiproliferative, and inhibitory growth activity demonstrated mainly in cancer cell lines.Indeed, from the analyzed results, there are a group of molecules with the premise of contain a phenyl sulfonamide moiety as pharmacophore towards GPER and Bcl-2 binding sites and additionally, contains in the aromatic ring of this moiety different o-, m-and p-substituents of electron withdrawing nature as the compounds 6, 7, 8, 9 and 37 (Table S1, column 2), structures in concordance with potent Bcl-2 inhibitors as ABT-737 and ABT-263 which contain a N-(phenylsulfonyl)benzamide as pharmacophoric structure [51][52][53] , highly similar to the structures here designed.Concerning to the mentioned group of phenyl sulfonamide ligands, is worth of mention that, most of them, meet the requirements settled as part of the VS protocol, with only violations in the evaluations encompassed in physicochemical properties, specifically with a Log P and molecular weight slightly increased than is consider optimum according to the Lipinski´s rule of five and, in regards to metabolism prediction which shown a medium probability for ligands 6, 7, 8, and 37, of being modified in specific atoms of the phenyl sulfonamide moiety, impacting in the final value yielded of the VS protocol.Lastly, as a result of the VS protocol and, according to the availability from the chemical suppliers, just ligands 6 and 37 were obtained by chemical synthesis and evaluated on cell-based assays.

Analysis of the molecular interactions established in the ligand/receptor complexes
As an essential step to reach a total understanding of protein modulation by explicitly designed ligands, the depiction of the chemical interactions established in the ligand/receptor complexes turns into a mandatory requirement for that aim.For such, the molecular docking analysis of the only two synthesized and assayed compounds yield key insights of 6 and 37 regarding reaching the binding sites described for GPER and Bcl-2.Interestingly, both ligands largely share the same binding mode on GPER binding site (Fig. 2A), reaching entirely the phenylalanine cluster formed by Phe206, Phe208, Phe223, and Phe278, described as key orchestrator of the GPER activity 28 .Additionally, the subtle differences among both binding modes lie in the presence of Gln215 www.nature.com/scientificreports/establishing a h-bond with one of the oxygen atoms from the sulfonamide moiety of 6 at a close contact of 3.11 Å (Fig. 2B).In contrast, 37 lacks this interaction but establishes contact with Asn276 (Fig. 2C); both binding modes are in an orientation slightly biased to adopt a conformation showed by antagonist compounds of GPER as G15 and G36, which is characterized by modifications on the h-bond occupancy that impacts directly on the helix 2, 3 and 7 rearrangements by establishing interactions with pivotal residues as Phe206, Cys207, Phe208, Phe278 and importantly with Asn310 54 .Furthermore, it is important to note that the phenylalanine cluster stabilizes the ligands mainly through pi-alkyl interactions, with the only exception of Phe278 that forms a dual pi-pi-T-shaped interaction with the benzene moiety of the bromobenzo dioxol carboxaldehyde from both ligands, contrary to commonly described for agonist ligands in which, is the benzene moiety of the tetrahydroquinoline scaffold of the pharmacophore structure that is establishing a pi-pi interaction with Phe208 and additionally, reaching the Cys207 to conduct the putative conformational rearrangement between helix 3 and 5 that leads to the activation process of the receptor 54 .Interestingly, 37 depicts no-conventional interactions that arise from the addition of the 4-chloro benzene sulfonamide; in detail, benzene ring of the outlined above moiety, interacts with Glu218 by a pi-anion bond and Phe223 with their sulfur atom by a pi-sulfur bond, both interactions affecting the binding mode of the ligand on the receptor promoting better binding energy as is possible to note in their ∆G value of − 10.93 kcal/mol.Remarkably, the binding mode overlapping analysis from both ligands converge on a binding site well-defined both for the tetrahydroquinoline scaffold interacting with their benzene moiety by a pi-sigma interaction with Ile279 and, additionally, with Phe206 and Phe208 by pi-alkyl interactions with the piperidine and cyclopentene ring, consecutively, and lastly with the bromobenzo dioxol carboxaldehyde throughout h-bonds with Gln53 and His282 and the corresponding oxygen atoms of the dioxolane ring and, by a pi-alkyl interactions between Ala202 and Pro303 and the carbon atom from the acetal group of the same ring.Regarding the analysis of docking simulations on Bcl-2 protein, results yield encouraging insights to place the synthesized ligands as potential inhibitors similar to BH3 mimetics.Indeed, 6 and 37 reach a binding site bounded by residues in close contact to BH1, and predominantly BH3 domains as Phe130, Ala131, Thr132, Val133, Leu137, Phe153, and Met157 (Fig. 3A), which is the expected for pure BH3 mimetic inhibitors, such as the ligands that deactivate the antiapoptotic family members by filling up their BH3-binding pockets 55,56 , and how is possible to note in the binding mode depicted by Gossypol bounded on Bcl-2 protein, characterized by forming interactions with Asp111, Phe112, Met115, Gln118, Leu119, His120, and very important with the same residues previously described for 6 and 37 as Phe130, Ala131, Thr132, Val133, Leu137, Phe153, Met157 30 .Tellingly, both of the sulfonamide moieties from 6 and 37, adopt an orientation towards a sulfonamide-acceptor subcavity not described previously and formed by Leu121, Thr125, Ala126, Gly128, Phe153, Met157, and Trp176 (Fig. 3B), this last is the residue most important when interacting both methyl from 6 and chlorine from 37 groups, stabilizing them by an alkyl-chlorine and alkyl-alkyl interactions, respectively, and, for both ligands, by a pi-pi interaction between the aromatic ring of the sulfonamide moiety and the aromatic ring from the indole group of tryptophan (Fig. 3C).It is worth mentioning that Phe130 and Arg129 are in close contact with both ligands stabilizing them by two h-bonds established between the CO and NH groups from the peptide bond, respectively, and SO and NH from the sulfonamide moiety of the ligands at a distance ranging from 1.71 and 3.02 Å, respectively, placing this residue as a part of a seemingly funnel that connects the sulfonamide moiety with S1, S2 and S4 sub-pockets previously described on Bcl-2 towards ligands that reaches additional binding   sites on the protein 51,57 .Intriguingly, Arg129 is an important residue that forms part of the hydrophobic groove present on the surface of the Bcl-2 protein, which, its analogous in Bcl-XL corresponds to a region on the protein with the ability to bind to proapoptotic members of the Bcl-2 family 58,59 , specifically, this structural feature confers to Bcl-XL a higher affinity for Bad and Bak proteins due to a more electronegative state than the Bcl-2 owing to the role of Arg129 to neutralize the electronegativity of the groove.In that sense, it is worth mentioning that 6 and 37 binding modes (Fig. 4A,B), displace the protonable moiety of Arg129, being able to impact its ability to adopt a protonated state modifying its affinity to dimerize with other Bcl-2 family members as Bad and Bak or homodimerizing 60,61 .

Stability of simulated systems
The area per lipid, RMSD, and Rg analysis plots display the mobility of the GPER-ligand complexes (Fig. 5), and RMSD and Rg plots show the mobility of Bcl2-ligand systems (Fig. 6).The area per lipid values for GPERligand complexes showed convergence between 50 and 60 ns with average values of 59.49 ± 0.40 and 59.5 ± 0.53 for GPER-37 and GPER-6, respectively (Fig. 5A).The GPER-ligand complexes showed RMSD (Fig. 5B) and Rg (Fig. 5C) values that achieved convergence between 20 and 50 ns with average Rg values of 25.0 ± 0.10 and 25.9 ± 0.30 Å for GPER-37 and GPER-6, respectively (Fig. 5C).The Bcl2-ligand complexes showed RMSD (Fig. 6A) and Rg (Fig. 6B) values that equilibrated between 20 and 40 ns with average Rg values of 17.7 ± 0.16 and 18.0 ± 0.20 Å for Bcl2-37 and Bcl2-6, respectively (Fig. 6B).RMSF analysis was determined over the equilibrated simulation time for GPER-ligand complexes (Fig. 5D) showing two high fluctuation peaks corresponding to N-ter and C-ter fragments of both complexes, meanwhile, for Bcl2-ligand complexes (Fig. 6C) there are two regions with increased fluctuations corresponding to regions intrinsically disordered 29,30 .Based on these results, only the last 40 ns and 50 ns were considered for further analysis of GPER-ligand and Bcl2-ligand complexes, respectively.

Structural and energetical analysis of receptor-ligand complexes
Representative GPER-ligand (Fig. 7) and Bcl2-ligand (Fig. 8) conformations were obtained through cluster analysis, considering a cut-off of 2.5 Å. Compound 37 on GPER is coordinated by a series of residues in which the main type of interactions is hydrophobic established between residues Leu59, Val116, Leu119, Met141, Phe206, Phe208, Val219, Phe223, Trp272, Phe278, Ile279, Pro303, Leu311, Phe314, and Ala313 with the compound.(Fig. 7A).Noteworthy, Phe206, Phe208, Phe223, and Phe278 residues that form the phenylalanine cluster into GPER are interacting with the ligand mainly through pi-alkyl interactions, detailing, is Phe208 and Phe278, the residues that confer an improved stabilization to the ligand/receptor complex, owing to non-polar interactions between the aromatic ring of F208 and the chlorine atom and the pentadiene ring of 37, whilst, Phe278 interacts through a pi-pi interaction with the aromatic ring and with a pi-alkyl with the piperidine ring, both from the tetrahydroquinoline moiety of 37, respectively.Additionally, this compound formed polar contacts with His52, Gln53, Gln54, Gly58, S62, Gln138, Gln215, Glu218, Glu275, and His307.Remarkably, from the hydrogen bonds established between side chains of Asn276 and Asn310 with 37, it is interesting that sulfonamide moiety of the ligand is stabilized by two h-bonds provided by Asn276 at a distance lesser than 3 Å.In regard to compound 6, it is bound by most of the same interacting residues and through non-polar interactions than 37, as it is Leu59, Leu119, Leu137, Met141, Ala204, Phe206, Phe208, Phe223, Trp272, Phe278, Ile279, and Pro303.Strikingly, is the phenylalanine cluster that stabilizes compound 6 through a binding mode very similar to that depicted for compound 37, with the main stabilizing interactions provided by Phe208 and Phe278.Besides, the polar interactions correspond to those given by Glu51, His52, Gln53, Gln215, Glu275, Asn276, His282, and Asn310.Similarly analyzed for compound 37, Asn276 is the residue stabilizing the sulfonamide moiety of 6 by forming two hydrogen bonds (Fig. 7B).Besides, analysis of Bcl-2/ligand complexes yielded structural insights into the ability of both compounds to reach and maintain key interactions with residues closed to subpockets previously described as binding sites on Bcl-2 protein.Thus, compound 37 by 12 hydrophobic residues: Met115, Leu119, Val133, Leu121, Ala126, Phe130, Ala131, Val134, Phe153, Met157, Ile173, and Trp176.And 5 polar residues: Thr125, Gly128, Arg127, Arg129, and Thr132 (Fig. 8A), it is worth emphasizing that several residues previously listed are encompassed into BH1-BH3 domains, which are the proteinic target of pure BH3 mimetics inhibitors.Moreover, a striking find of the MD results is the maintaining of the sulfonamide-acceptor subcavity formed by Leu121, Thr125, Ala126, Gly128, Arg129, Phe153, Met157 and Trp176, with the inclusion now of Ile173 as an additional residue stabilizing along the time to compound 37 by forming an alkyl interaction with the chlorine atom from the sulfonamide moiety of this ligand, further interactions towards stabilize this moiety are established by pi-alkyl and pi-pi interactions provided by Met157, and Phe153 and Trp176, respectively, at an average distance lesser to 5 Å.A last detail regarding to polar interactions, is the pivotal role of Arg129 to forms a funnel that connects some residues from the S1, S2 and S4 subpockets on Bcl-2 with the sulfonamideacceptor subcavity owing to the formation of two h-bonds at 2.74 and 2.96 Å with the CO and SO from the sulfonamide bond of 37, respectively.Finally, compound 6 on the Bcl2-ligand complex shown a binding mode very similar to the depicted for 37, with slight differences as far as the sulfonamide-acceptor subcavity, in detail, Met157, Phe153 and Trp176 are found establishing the same interactions with the chloro-phenyl moiety than 37, with the only exception that Ile173 is lacking in the structural analysis.Additionally, total description of the interactions established in the receptor-ligand complex shown that 6 was coupled by 14 hydrophobic residues:  www.nature.com/scientificreports/Phe112, Met115, Leu119, Leu121, Ala126, Phe130, Ala131, Val133, Val134, Leu137, Ala149, Phe153, Met157, and Trp176 (Fig. 8B).Polar interactions were taken by Thr125, Arg127, Gly128, Arg129, Thr132 and Arg146.Arg129 also formed one hydrogen bond with the sulfonamide bond of 6, very similar to the described for the 37 and the apparently funnel formed by this residue.Lastly, comparative analysis revealed that 37 reached a higher number of interactions with GPER than 6.In contrast, is this last compound that showed a more optimized map of receptor-ligand interactions for Bcl-2 than 37.

Binding Free Energy decomposition of receptor-ligand complexes
The binding affinity of the ligands with GPER or Bcl-2 was determined over the equilibrate simulation time using the MMGBSA method.Table 1 shows that the two compounds exhibited favorable binding free energies on GPER or Bcl2.Comparative analysis between the different contributions to the binding free energy (ΔG bind ) points that the Van der Waals energy (ΔE vdw ) was the major force stabilizing the ligand on both receptors and, specifically for Bcl-2, the chemical nature of 6 and 37 containing in both structures pi electrons able to interact throughout pi-alkyl interactions with most of the residues from the groove contained on Bcl-2 protein, might affecting directly on the displacement that suffer Arg129 (on its protonable moiety) and in the affinity for dimerize with proapoptotic proteins or homodimerization.This analysis suggests a more favorable binding affinity of 6 to Bcl-2 than GPER.Whereas a similar affinity was observed for 37 on both receptors.

Cell proliferation analysis
MCF-7 cells and MDA-MB-231 cell were subjected to treatments with 6 and 37 to corroborate these compounds' ability to inhibit cell proliferation.Outlined above, MCF-7 (ER and PR ( +) and MDA-MB-231 (ER, PR, and HER2 (−) cell lines were considered as models of the most common and aggressive subtype of breast cancer according to their molecular expression pattern, in addition to a deregulated Bcl-2 activity promoting an antiapoptotic effect 62,63 .As is depicted in Fig. 9, in MCF-7 both compounds showed a better performance, measured as a decrease in the percentage of cell viability, than Gossypol, in a concentration-dependent manner without involving the cell number seeded; this finding is an outstanding insight into the probable mechanism of action   conducted by 6 and 37 which points mainly to a probable Bcl-2 inhibition, reinforced for a GPER modulation, since, from the energetical analyzes yielded by in silico simulations over the two pharmacological targets, the affinity of both compounds is slightly more representative for GPER in comparison with Bcl-2 (< onefold), however, from the individual results analysis, 37 decreased cell proliferation in a better way than 6 probably due to the slightly finer binding energy of this ligand towards Bcl-2 as is possible to contrast it both docking calculations and molecular dynamics simulations.Astoundingly, MDA-MB-231 cancer cell were more sensitive to the assayed compounds than MCF-7 cells, owing largely to a over expression of Bcl-2 protein on MDA-MB-231 cell line 64,65 , added to a slight GPER expression that in a basal state, triggers proliferation and cancer cell survival 66,67 , although, from the individual results obtained for MCF-7 cancer cells, 37 depicted an IC 50 value significantly lower than Gossypol (18.66 vs. 65.67 μM, respectively), probably due to multiple effects of this ligand at inhibit both estrogen receptors (ERs α > β and GPER) and the anti-apoptotic protein Bcl-2, triggering a synergistic effect which leads to a best antiproliferative performance than showed by Gossypol, besides, we were unable to determine the IC 50 value for G-15, due to solubilization problems starting at a 20 μM concentration (Data not showed).The highlighted above place both compounds as potential therapeutic options against most aggressive cancer type TNCB (Triple-Negative Breast Cancer) represented by the two cancer cell lines assayed.IC 50 values of 6 and 37 assayed on MDA-MB-231 and MCF-7 cancer cell lines are collected on Table 2.

Figure 1 .
Figure 1.Schematic procedure for the obtention of the tetrahydroquinoline derivatives 6 (R = CH 3 ) and 37 (R = Cl) from G1-PABA.Reactants and chemical synthesis conditions are listed up (first step)-down (second step) from the arrow.

Figure 2 .
Figure 2. (A) Binding modes overlapping of 6 and 37 on GPER binding site.(B) Close-up of the 6 and (C) 37 binding modes.Pi-pi interactions are depicted as dark orange dashes and points a right-turn of 45° of the ligand for a better understanding.

Figure 3 .
Figure 3. Binding modes overlapping of 6 and 37 on Bcl-2 binding sites.B) Isosurface representation of the residues forming the sulfonamide-acceptor subcavity.C) 2D map of contacts of the interactions between benzene sulfonamide moiety and the interacting residues.

Figure 4 .
Figure 4. Binding modes of 6 and 37 on Bcl-2.(A) Interacting residues for 6 and, (B) interacting residues for 37 on the ligand/receptor complex.H-bonds are depicted as red dashes.

Figure 5 .
Figure 5. Equilibrium properties of 6 and 37 in complex with GPER through MD simulations.(A) Area per lipid, (B) Root mean square deviation (RMSD), and (C) Radius of gyration (Rg).6 and 37 behaviors are depicted in black and red lines, respectively.

Figure 6 .
Figure 6.Equilibrium properties of 6 and 37 in complex with Bcl-2 through MD simulations.(A) Root mean square deviation (RMSD), and (B) Radius of gyration (Rg).6 and 37 behaviors are depicted in black and red lines, respectively.

Figure 9 .
Figure 9.Effect of 6 and 37 on cell viability of breast cancer cells.(A) MCF-7 and (B) MDA-MB-231.Each data point was performed in sextuplicate in three different experiments, and the results were reported as the mean absorption ± SD.

Table 1 .
Binding free energy components of GPER or Bcl-2 in complexes with 6 and 37 (kcal/mol).

Table 2 .
IC 50 values of 6 and 37 on the cancer cell lines assayed.