In-Silico Drug Discovery Through Interaction-Based Perspectives for ATAD3A

In-Silico Drug Discovery Through Interaction-Based Perspectives for ATAD3A | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article In-Silico Drug Discovery Through Interaction-Based Perspectives for ATAD3A Kevin Gong, Patrick McKenna This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3280889/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The overexpression of ATAD3A, a mitochondrial membrane oncoprotein, is correlated with worsened prognosis of many prevalent cancers and has been identified as an attractive target for drug development. This work investigates drug development for ATAD3A through a site-specific, interaction-based, in-silico framework bypassing conventional drug development obstacles, notably the inability to experimentally resolve the ATAD3A structure. In our approach, we target the critical ATAD3A-S100B binding interaction, which facilitates the cytoplasmic processing of ATAD3A. Relying on the canonicality of the S100B binding mechanism of ATAD3A, and the ATAD3A C-terminal sequence being highly conserved and homologous to existing structures, we reduce the necessity for an accurate model from the whole target protein to a single, well-established domain. In our in-silico framework, a model of the ATAD3A S100B binding domain was constructed, followed by drug discovery targeting the S100B binding domain utilizing virtual screening and hit identification. QSAR modeling and physicochemical filters were then subsequently employed to assess the hit compounds. Further analyzing the specific binding contacts between the hit compounds and ATAD3A compared to ATAD3A-S100B binding contacts from protein-protein docking, we were able to determine that two hit compounds, ZINC6235062 and ZINC20728831, strongly occupy the critical residues established in previous literature as necessary for S100B binding, and thus indicate great potential in inhibiting ATAD3A oncoprotein function through disrupting the ATAD3A-S100B binding interaction in competitive inhibition. Biological sciences/Biochemistry Biological sciences/Cancer Biological sciences/Computational biology and bioinformatics Biological sciences/Drug discovery Drug Development Virtual Screening Molecular Docking Homology Modeling Quantitative Structure-Activity Relationship Modeling Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 INTRODUCTION In 2020, around 18.1 million people were diagnosed with cancer, and cancers accounted for nearly 10 million deaths, making cancer the leading cause of death for humans around the world ( 1 ). The development of treatment for all types of cancer has been a massive front in scientific research for decades, with billions of dollars and countless researchers devoted to its cause ( 1 ). In the world of modern anticancer therapeutic drug development, numerous mitochondrial enzymes have been identified as attractive targets for drug development, as they have been found to be overexpressed and upregulated in tumor cells and critical in oncogenic signaling and pathways necessary to support cancer metastasis and rapid proliferation ( 2 ). Mitochondrial membrane oncoprotein ATPase family AAA domain containing protein 3A (ATAD3A) has been found to be upregulated and implicated for cancer proliferation, oncogenic signaling, tumor growth and metastasis in certain types of cancer, consequently resulting in its upregulation being associated with worsened prognosis for cancer patients ( 2 ). Specifically, studies have found that the abnormal activation of ATAD3A in cancer cells causes mitochondrial oncogenic signaling, enhancing activities and functions that promote tumor growth ( 3 ). Among these, overexpression of ATAD3A leads to breast cancer metastasis by increasing the stability of WASF3 protein, a known tumor metastasis promoter, with the ER GRP78 protein, whose expression is reduced by ATAD3A knockdown ( 4 ). In prostate cancer, upregulation of ATAD3A correlates with serum prostate-specific antigen (PSA) levels and androgen receptor expression ( 5 ). In hepatocellular carcinoma, urothelial carcinoma, prostate cancer, glioma, breast cancer and lung adenocarcinoma, ATAD3A upregulation is correlated with reduced overall patient survivability, worsened prognosis, proliferation, and tumor growth ( 2 ). Based on the dire consequences of the oncogenic effects of ATAD3A and as previously suggested in a 2020 study by Lang & Teng, the targeting of ATAD3A for drug development could yield great potential, and two studies have already shown that knockdown of ATAD3A with shRNA has been found to significantly induce repression of tumor growth and metastasis for breast and colon cancer in mice lung models ( 2 , 6 ). Currently, however, an effective, ATAD3A specific inhibitor has yet to be developed, and most studies up to date focus on the process of developing transcriptional inhibitors as there is no existing experimentally resolved structure for ATAD3A and there is little understanding on its molecular mechanisms ( 6 ). ATAD3A is a mitochondrial transmembrane protein of the AAA+ (ATPases associated with various cellular activities) superfamily ( 2 , 7 ). As an AAA protein, ATAD3A is thus a mechanochemical enzyme, or motor protein, in its hexameric aggregate. In terms of in-depth structure, ATAD3A is localized to the mitochondrial inner membrane ( 7 ). Within its highly conserved mitochondrial matrix C terminal end from (aa 264–634), ATAD3A contains an S100B binding site and ATPase domain with the Walker A (P loop) and Walker B motifs associated with phosphate binding and ATP hydrolysis respectively ( 8 , 9 ). Beyond the inner mitochondrial membrane, which is spanned by a transmembrane domain from aa 247–264, the lack of conservation with other protein sequences and lack of an experimentally resolved structure of ATAD3A has resulted in the lack of concrete structural information of the N terminal region ( 2 , 7 , 10 , 11 ). The lack of structural information on the N terminal and absence of an experimentally resolved structure have further led ATAD3A to be deemed infeasible for activity inhibitor design ( 2 ). S100B, a zinc and calcium-binding protein with a chaperone like function in ATAD3A folding and mitochondrial targeting, binds with ATAD3A at its S100B binding site (aa 290–310) ( 9 , 12 ). This study seeks to investigate the potentiality of drug development for a conventionally undruggable target, the ATAD3A oncoprotein, through exploring a novel approach employing a site specific, interaction-based perspective. Specifically, the ATAD3A oncoprotein has been labeled undruggable due to its difficulty to purify and crystallize as a mitochondrial cristae membrane protein, resulting in its lack of experimentally resolved structure and inability to be assessed in-vitro as a whole protein. With these limitations, a novel approach for drug development is required which enables an in-silico framework and bypasses the structural information obstacle. In our approach, the well-established and conserved S100B-ATAD3A interaction region is modeled with high confidence and independent of the other regions of ATAD3A. The modeled interaction is then targeted to induce conformational change and improper topology of the ATAD3A C terminal ATPase, which theoretically results in the suppression of ATPase enzymatic and hexameric activity. This proposed mechanism is supported by literature on ATAD3A synthesis, where complex topology, aggregation prone regions of ATAD3A, notably the ATPase domain, require S100B to facilitate cytoplasmic protein folding and processing and mitochondrial targeting ( 9 ). Consequently, as suggested by previous literature, with decreased levels of ATAD3A mitochondrial targeting and conformationally altered ATPase domains, the oncogenic functions of ATAD3A can be theoretically mitigated, as studies attribute ATAD3A ATPase function with oncogenic pathways and ATAD3A known to scaffold tumor promoters ( 2 , 4 ). In this study, we discovered compounds that can act as inhibitors of ATAD3A through competitive inhibition of the ATAD3A S100B binding domain and potentially suppress the oncogenic pathways of ATAD3A. RESULTS ATAD3A Model Development for Drug Discovery With the structure of ATAD3A currently difficult to be experimentally resolved, an in-silico approach had to be utilized to develop a model for ATAD3A drug discovery. However, it is difficult to develop a sufficiently accurate model utilizing in-silico methodologies to serve as the basis of drug discovery due to the regions of ATAD3A extending from across the intermembrane space to the outer mitochondrial membrane being difficult to model accurately due to the lack of homologous sequences as well as a lack of well-defined tertiary structures, with the exception of two coiled coil domains ( 7 ). Without a sufficiently accurate structural model of the target protein, conventional drug discovery methodology becomes infeasible. To overcome this, we focus on the S100B binding site of the ATAD3A oncoprotein as the binding site for drug discovery and the ATPase domain as the region to be suppressed. The process of target validation in the drug discovery pipeline, the validation of the target’s therapeutic value, is substituted by established experimental characteristics in previous literature deducing the ATPase function to be critical for oncogenic effect as well as the necessity of S100B binding for proper ATPase folding and function ( 2 , 9 ). With these deductions, the regions of the structure of ATAD3A which necessitates accuracy becomes reduced to within the highly conserved C terminal domain containing the S100B binding domain, enabling the construction of a model to high accuracy due to the homologous structures, enabling sufficient confidence to initiate drug discovery. Indeed, when assessed with a Ramachandran plot, few Ramachandran outliers were identified for the C terminal and none identified in the S100B binding domain of the model of ATAD3A developed through homology modeling in the Modeller software,, suggesting the accuracy of the model developed in homology modeling (Fig. 1 b ) ( 13 , 14 ). This confidence in accuracy was further validated through a protein model refinement web server, Feig Lab PREFMD, where the refined model was virtually identical to the inputted model (Fig. 1 a ) ( 15 ). With an accurate model of the ATAD3A C terminal S100B binding site (Fig. 1 a), drug discovery targeting the S100B binding of ATAD3A could be initiated. Drug Development and Binding Analysis Large scale virtual screening through in-silico docking with AutoDock Vina on the Zinc Chemical Library was utilized to conduct drug discovery ( 16 , 17 ). All screened ligands in the virtual screening library were then sorted by the binding affinity of their best binding conformation, with the top 1% (ranging from − 8.2 to -10.1 kcal/mol) selected for subsequent phases and identified as compounds with high binding affinities for ATAD3A in the 20x20x20 angstrom search grid, or area on the target protein where docking is conducted. These compounds were then evaluated with physicochemical filters from RDKit (Lipinski Rule of 5, Veber, Drug-like, Ghose and REOS filters) to filter out compounds with unsatisfactory properties such as large molecular weight, LogP and number of hydrogen bond acceptors/donors ( 18 ). All compounds that passed the filters were identified as hit compounds and selected for Quantitative Structure-Activity Relationship (QSAR) model evaluation (shown in Fig. 2 ). Hit compounds possessing binding affinities stronger than − 8 kcal/mol for the S100B binding site, indicating very strong binding, can competitively inhibit S100B binding of ATAD3A by occupying the S100B binding site and preventing the ATAD3A-S100B interaction, and the mitochondrial targeting of ATAD3A as well as proper folding of ATAD3A are consequently inhibited through the loss of the calcium dependent mechanism ( 9 ). Thus, it becomes critical to determine the ability for the hit compounds to occupy the residues critical for S100B binding. To better determine the ability for the compounds to occupy the residues critical for S100B binding and thus inhibit ATAD3A-S100B binding, we utilized the HDOCK web server, a protein-protein docking utility, to analyze the S100B binding of ATAD3A ( 19 ). Analyzing the best docking mode between ATAD3A and S100B predicted by HDOCK in PyMOL, we identified the residue sites of the binding contacts between ATAD3A and S100B (Fig. 3). Binding contact length was cut off at 3.5 angstroms to show only significant hydrogen and disulfide bonds, with the exception of V341, which formed < 4.0 angstrom C-H bonds with every point on the aromatic ring of the S100B residue H85 (Fig. 3b). As demonstrated in Fig. 3 , the main interfaces of interaction between ATAD3A and S100B are situated at the site consisting of residues I339, T340, V341 & A344 and the site of residue L345 with the S100B carboxyl terminal domain residues F87 and H85 respectively, as well as the site consisting of residues I349, S352 & R353 with the S100B linker domain residues E46 and L44. To further assess the results of the protein-protein docking, we revisited the mutagenesis annotations on UniProt from established ( 9 ) Notably, of the main sites of contact identified in protein-protein docking, three binding residues matched with mutagenesis annotations on UniProt, which described the complete loss of S100B binding with the mutagenesis of both the residues V341 and L345 and decrease in S100B binding with the mutagenesis of residue I349 ( 9 ). We here utilize the reports of S100B binding loss or decrease as a result of residue mutagenesis as an indicator of the necessity of availability of the residues in ATAD3A-S100B binding. Thus, the mutagenesis annotations correlate to the interactions demonstrated in the protein-protein docking, affirming its validity and emphasize the importance of the availability of the residues V341 and L345 in the S100B binding of ATAD3A ( 9 , 20 ). Thus, the ability of inhibitor compounds to occupy and bind to these residues is critical to the inhibition of ATAD3A-S100B binding, which enables the endogenous processing mechanism of S100B ( 9 ). Of the hit compounds developed, two of them have their highest affinity binding modes occupying these exact sites. The hit compound ZINC20728831 (N-benzhydryl-4-oxoquinazoline-2-carboxamide) occupies the binding site at V341, establishing contacts with the residues T340, V341 and L342 as demonstrated in Fig. 4 a. The hit compound ZINC6235062 (3-hydroxy-4-[(2S)-3-hydroxy-4-oxo-1,2-dihydroquinazolin-2-yl]naphthalene-2-carboxylic acid), binds to the adjacent T340, a similarly significant residue in ATAD3A S100B binding, and is spatially oriented to block access to V341, as seen in Fig. 4 b. However, many of the hit compounds were also found to have high binding affinity binding poses with the ATPase region of ATAD3A adjacent to the S100B binding site, a region that also happened to be within the defined AutoDock Vina search grid for docking. Upon further inspection, this region was identified as the Walker B motif, with a similar LLFVDE sequence as the Walker B motif identified in the 586 aa isoform and coinciding with the hhhhDE of the consensus Walker B sequence. The proximity of the S100B binding site to the Walker B motif is thus of note, as it is possible that S100B binding plays a critical role in the folding and proper ATP hydrolysis function of the Walker B motif. All highest affinity binding modes of each hit molecule are demonstrated in Fig. 5 . From observations of all the binding modes calculated by AutoDock Vina, it becomes clear that the main residues responsible for binding contacts with the hit compounds within the S100B binding site are at the side chains of the T340 residue and the R338 residue, which are both characteristically associated as polar binding residues. Similarly significant in binding contacts were the V341 and L342 residues, which were also commonly identified as ligand sites in PyMOL. Though our protein-protein docking results indicate that there are indeed contacts between the R338 residue and the S100B protein, such contact is greater than 4 angstroms in length and associated with the S100B linker between the two EF hands, not the carboxyl terminal domain associated with protein-protein interactions and binding ( 9 ). Additionally, as the V341 residue is well documented as being critical to S100B binding and is shown in protein-protein docking to closely interact with the S100B carboxyl terminal domain, the occupancy of this particular residue is especially crucial, and it is found that ZINC20728831 consistently binds to this residue in its 5 highest binding affinity binding modes, rendering it the most promising compound for the inhibition of ATAD3A-S100B binding. The adjacent T340 is similarly critical for ATAD3A S100B binding with its projected hydrogen bonding with the S100B residue H85 (Fig. 3d), which ZINC20728831 and ZINC6235062 bind to consistently, further yielding promise for both compounds (Fig. 4 a, 4 b). Statistical Inference To assess the statistical significance of the binding affinities of the hit compounds developed compared the average binding affinity of all compounds, Z tests were conducted. The Z test results demonstrated that each hit compound had a Z-score of around 6, corresponding to P-values significantly less than the alpha of .01, thus demonstrating that the hit compounds were statistically significantly stronger in binding with ATAD3A in the defined search grid compared to the average compound in the virtual screening library (Table 1 ). Table 1. Z-Test Results Compound Name Z-Score P-value ZINC401340 -6.411925508 ZINC5082118 -6.241072954 ZINC6235062 -6.241072954 ZINC6235061 -6.582778063 ZINC5048139 -6.0702204 ZINC5082118 -6.0702204 ZINC20728831 -6.411925508 The Z-score and P-value of each hit compound in Z-test with the average binding affinity of all compounds in virtual screening library (800,000 randomly compounds from ZINC20 library). An significance level of α = 0.01 was utilized to determine significance. QSAR and CHARMM-GUI Results Quantitative Structure-Activity Relationship (QSAR) models trained on PubChem data were utilized to assess the activity of the hit compounds. The trained QSAR models, Multiple Linear Regression, K-Nearest Neighbor, Random Forest Regression and Associative Neural Network developed on the OCHEM web platform, were evaluated, in which the model with the highest combination of accuracy, balanced accuracy, and area under the curve (AUC) was selected to predict the activity of the hit compounds. This comparison revealed that the Associative Neural Network was the most successful model by each measure of accuracy (Table 2 ). This model was thus applied to perform activity prediction for the hit compounds. Table 2 QSAR Models Model Type Accuracy Balanced Accuracy AUC Associative Neural Network 76% 76% 81% Random Forest Regression 73% 75% 83% K-Nearest Neighbor 63% 79% 80% Multiple Linear Regression 65% 74% 76% All QSAR models trained within OCHEM, includes their respective Accuracy, Balanced Accuracy, and AUC. The best model was the one with the highest combination of all three of these criteria. The free energy of binding values calculated by CHARMM-GUI along with AutoDock Vina binding affinity values indicate strong binding between the developed hit compounds and ATAD3A within the AutoDock Vina search box. Additionally, QSAR activity prediction condition for all but compound ZINC6235061 demonstrates the high likelihood of binding activity between the active compounds and ATAD3A (Table 3 ). Table 3 QSAR and CHARMM-GUI Results Compound Name Activity Prediction Molecular Weight (kDa) Binding Affinity Free Energy of Binding (KJ/mol) ZINC401340 Active 338.318 -8.5 -10.1 ZINC5082118 Active 380.488 -8.4 -7.9 ZINC6235062 Active 350.33 -8.4 -9.1 ZINC6235061 Inactive 350.33 -8.5 -9.1 ZINC5048139 Active 394.471 -8.6 -8.3 ZINC5082118 Active 374.318 -8.3 -8.9 ZINC20728831 Active 355.397 -8.3 -9.1 QSAR activity prediction, molecular weight, binding affinity, and free energy of binding of all hit compounds is shown above. DISCUSSION Our study successfully investigated a pragmatic approach in drug development utilizing new site specific, interaction-based perspectives, employing a basis of previous literature to bypass technical barriers of conventional orthosteric drug development for the ATAD3A oncoprotein. Instead of pursuing conventional approaches of resolved structure to direct inhibition of an orthosteric site, our study utilized an allosteric approach of indirect inhibition, where critical binding interactions are suppressed to directly induce conformational change and mechanistic changes which then consequentially mitigate oncogenic pathways, as described in previous literature ( 2 , 4 , 9 ). Allowing the target site to be an established critical binding interaction site and supporting the allosteric inhibitory effect on the oncogenic domain (ATPase) with established mechanisms, we bypass the barrier of the necessity of an experimentally resolved structure of the whole target protein and specific active site architecture by initiating drug discovery for a site which we can confidently construct a model of and occupy, thus dismantling the critical binding interaction. Methodologically, we constructed a model for ATAD3A in homology modeling and utilized refinement web servers and Ramachandran plot to assess the accuracy of the developed model. With a constructed model, we employed in-silico drug discovery in virtual screening through large scale docking. Developed hit compounds were then filtered and assessed in subsequently applied physicochemical filter and QSAR modeling. Through this in-silico framework we were able to substantially substitute the conventional drug discovery pipeline with completely computational alternatives, as necessitated by the inability for in-vitro work to be conducted on membrane proteins such as ATAD3A. This investigation in computational alternatives in the drug development framework thus expands the feasibility of new perspectives such as the one outlined in this study and drug development relying heavily on computational methodologies for conventionally undruggable targets, many of which are difficult to purify for in-vitro work. The developed hit compounds ZINC6235062 (3-hydroxy-4-[(2S)-3-hydroxy-4-oxo-1,2-dihydroquinazolin-2-yl]naphthalene-2-carboxylic acid) and ZINC20728831 (N-benzhydryl-4-oxoquinazoline-2-carboxamide) possess high values of binding affinity and low free energy of binding for critical residues necessary for S100B binding at the S100B binding domain, most notably the residues T340 and V341, of which V341 has been experimentally determined to be critical for S100B binding ( 9 ). Both compounds share similar characteristics in their docking with the S100B binding domain, hydrogen bonding with the oxygen atom of the hydroxyl group on the T340 side chain through the hydrogen at nitrogen atom in the 4-quinazolinone group, which has been of interest in medicinal chemistry ( 21 ). The oxygen of carbonyl group in 4-quinazolinone of ZINC20728831 also forms hydrogen bonds with the hydrogen of amines of both V341 and L342, and the hydrogen in amide bond which links the 4-quinazolinone and naphthalene moiety also binds with the same T340 hydroxyl group oxygen. These series of hydrogen bonds in the docking of ATAD3A with the critical S100B binding domain residues reinforce the deduction of strong binding affinity between ZINC20728831 and the ATAD3A S100B binding domain. Of note, the epimers ZINC6235061 and ZINC6235062, which are sterically different in the link between their quinazolinone and 3-hydroxy-2-naphthoic acid groups, were found to dock to different sites, and the activity predicted by QSAR modeling indicates ZINC6235061 as inactive and ZINC6235062 as active. This suggests the difference stereochemistry between the two epimers being of great importance to their binding activities, especially as their 3-hydroxy-2-naphthoic acid groups adopt similar spatial orientations in the binding domain region, and the steric differences between the two compounds resulting in different orientations of the critical quinazolinone group. Taken together with the findings on ZINC20728831, it appears that the quinazolinone group is critical for the compounds ZINC20728831 and ZINC6235062 to dock to the critical S100B binding domain residues, which indicates the ability to competitively inhibit S100B binding in ATAD3A and the consequent oncogenic effects of endogenous ATAD3A through occupancy of the S100B binding domain. Coupled with the known severity of ATAD3A oncogenic functions in multiple of the deadliest cancers, the ATAD3A inhibitors developed in this study, thus, in their potential direct suppression of S100B binding and indirect inhibition of the oncogenic pathways of ATAD3A, yield potential for therapeutic development and can be further assessed and characterized for inhibitory effect in cellular in-vitro testing. Thus, as an in-silico study into the ATAD3A oncoprotein, we successfully present a new approach in drug development for ATAD3A, and demonstrate the development of inhibitors along such an approach, as well as expand on previous studies investigating drug development options for targeting ATAD3A. Broadly speaking, we believe our investigation into a more pragmatic approach in drug development exploring new allosteric interaction-based, site specific perspectives allows for an innovative insight into more flexible thinking and strategy in drug development, which has the potential to enable the drug development for more targets previously labeled infeasible. MATERIALS AND METHODS Required Materials A Sam Houston State University Computer Science Department Intel 32 core workstation was utilized through the generosity of Computer Science Department Professor Dr. Frank Liu. Online and software resources utilized in this study include the NCBI BLAST program, Modeller 10.3 software, AlphaFold2, Autodock Vina software, AutoDock Tools software, PyMOL software, CHARMM-GUI, OCHEM Web Server, UniProt Database, Feig Lab PrefMD Web Server and Python 3.11 ( 14 – 16 , 20 , 22 – 29 ). The ZINC20 Chemical Library was accessed for the virtual screening library of compounds, and the rdkit repository on GitHub was utilized to implement physicochemical filters ( 17 , 18 ). Homology Modeling and Model Construction The structure of ATAD3A was constructed utilizing homology modeling and AlphaFold’s statistical probabilities based on known structures with sequence similarity ( 23 ). To identify analogous templates ATAD3’s amino acid sequence was identified from UniProt and uploaded into NCBI BLAST (Basic Local Alignment and Search Tool) ( 22 ). The PDB files of the sequences with the highest sequence homology (similar sequences with resolved structures) were then downloaded to be utilized as template sequences, all of which met the requirement of at least 30% sequence similarity, the standard accepted minimum percent similarity for homology modeling to be employed accurately. As it was discovered that nearly all homologous template sequences possessed sequence similarity with the ATAD3A C terminal ATPase, the other regions of the protein were isolated to find homologous sequences, particularly the N terminal domain (aa 1–70) and Intermembrane domain (aa 71–294). BLAST was then run on each of these segments separately to identify homologous regions for each region specifically, with the same requirement of 30% sequence similarity ( 22 ). Regions that continued to lack a homologous model (disordered regions), most notably the intermembrane domain, were inputted into Alphafold 2 to generate candidate structures, which were incorporated into the structure with the other PDB templates for each of the other regions ( 23 ). Each template sequence of ATAD3A was then aligned, combined, and built within the Modeller software ( 14 ). The model was then evaluated using a Ramachandran plot and outliers were noted ( 13 ). Finally, the model was refined with the Feig Lab PREFMD webserver ( 15 ). In-silico Drug Discovery In in-silico drug discovery, the S100B binding site was extracted from the ATAD3A model through PyMOL, compared with published structural characteristics of canonical binding ( 9 , 25 ). As determined by an experimentally assessed ATAD3A S100B binding site in complex with S100B, the S100B binding mechanism of ATAD3A is a canonical p53-S100B binding interaction, and thus the canonical S100B-p53 bond was utilized to assess the validity of our S100B binding site structure, with the contacts identified in the 2010 study by Gilquin et al serving as reference ( 9 ). Subsequently, 800,000 randomly sampled compounds from the Zinc20 in stock chemical library were downloaded and prepared in pdbqt file format as the ligand library for AutoDock Vina ( 16 , 17 , 24 ). The ATAD3A S100B binding site was then prepared as the receptor in PyMOL and AutoDock Tools. A python script was then written to iterate through all the ligand files and adapt Vina for virtual screening ( 29 ). The ligands were then screened by AutoDock Vina through docking with the S100B binding site of ATAD3A. Following the screen, a python script was written to sort all the output ligand files by their binding affinity and the top 1% identified. These output ligand files were then moved into a different folder and screened in a miniconda environment with the Veber, Ghose, Lipinski Rule of 5, Rule of 3, REOS and Drug Like filters imported from rdkit repository through executing a python implementation script modified to iterate through the folder and output log files for each hit compound ( 18 ). A python script was then written to identify the compounds that passed all filters as hit compounds ( 29 ). The free energy of binding was then calculated with the use of CHARMM-GUI ( 26 , 27 ). Creation of QSAR for Activity Prediction The QSAR (Quantitative Structure-Activity Relationship) was constructed with the use of a publicly available data set on PubChem, referenced as “Blockade of RAGE activation in vascular endothelium” and accession ID: 482 ( 28 ). This assay measured the level of phosphorylated ERK1/2 in endothelial cells in the presence of 25 microg/mL S100b. This assay utilized colorimetry to quantitate phosphorylated ERK1/2 in an in-situ cell-based ELISA. In this assay, an active compound was defined as Phospho-ERK1/2 activity 60%. Through this, 25 compounds were identified as “active” and 1015 were identified as “inactive”. This data set was then downloaded and uploaded into OCHEM. Compound structures underwent preprocessing with Standardization, Neutralize, Remove Salts, and Clean Structure using the ChemAxon Standardizer, and variable selection was completed on the initial data set to clean the data set, remove redundant features, and conduct feature selection. The data set was divided into two sets: “training set” (80% of data) and “test Set” (20% of data). Then based on feature selection, the descriptors selected were ALogPS, OEState, and ChemAxon descriptors, and with parameters set to remove over exhausted descriptors. They were then filtered by grouping descriptors with a Pearson correlation coefficient of over 0.95 to prevent overfitting and remove highly correlated features. Since 3D descriptors were being utilized, Corina was employed to optimize the structures of the molecules. Multiple models were then constructed consisting of Associative Neural Network (ASNN), Multiple Linear Regression (MLR), K-Nearest Neighbor (KNN), and Random Forest Regression (RFR), with 10-fold cross validation. These models were retrained 4 times and the one with the highest combination of accuracy, balanced accuracy, and area under the curve (AUC) was then selected to predict the activity of each of the developed hit compounds. Application of QSAR Model to Predict Activity of Hit Compounds The QSAR model with the highest combination of accuracy, balanced accuracy, and area under the receiver operating curve (AUC) was then selected to predict the activity of the hit compounds. The compounds were then downloaded from the ZINC chemical library and the SMILES formats were uploaded into the OCHEM database in which the model was then selected to predict the activity of each of the molecules ( 28 ). This process was repeated several times to assure that the accuracy of the QSAR prediction was maintained. Declarations ACKNOWLEDGEMENTS The computational resources which enabled the virtual screening conducted in this study were kindly provided by Dr. Frank Liu from the Computer Science Department at Sam Houston State University. We greatly appreciate his technical assistance for our study. We would also like to greatly express our appreciation for Mr. Tyler Daniel and Dr. Qiuyue Nie from the Department of Chemical and Biomolecular Engineering at Rice University for their guidance in revisions and assistance with figure creation and nomenclature. We especially thank Dr. Qiuyue Nie for her guidance in compound descriptions. We would also like to thank Mr. Caleb Chang from the Department of Biosciences at Rice University for his mentorship and guidance with regards to protein target model development and drug discovery provided throughout the study. AUTHOR CONTRIBUTIONS K.G. conducted in silico drug discovery, including virtual screening and physicochemical filtering. K.G. wrote all sections of the manuscript and prepared Figures 1-5 and Table 1. K.G. also conducted all revisions of the manuscript. P.M. conducted homology modeling and QSAR modeling and wrote the sections of the manuscript related to those aspects. P.M. also prepared Tables 2 and 3. COMPETING INTERESTS The authors declare no competing interests. DATA AVAILABILITY STATEMENT No datasets were generated or analyzed during the current study. References Ferlay J, Colombet M, Soerjomataram I, Parkin DM, Piñeros M, Znaor A, et al. Cancer statistics for the year 2020: An overview. Int J Cancer. 2021. Epub 20210405. doi: 10.1002/ijc.33588 . PubMed PMID: 33818764. Lang L, Loveless R, Teng Y. Emerging Links between Control of Mitochondrial Protein ATAD3A and Cancer. Int J Mol Sci. 2020;21(21). Epub 20201025. doi: 10.3390/ijms21217917 . PubMed PMID: 33113782; PubMed Central PMCID: PMC7663417. Teng Y, Lang L, Shay C. ATAD3A on the Path to Cancer. Adv Exp Med Biol. 2019;1134:259 – 69. doi: 10.1007/978-3-030-12668-1_14 . PubMed PMID: 30919342. Teng Y, Pi W, Wang Y, Cowell JK. WASF3 provides the conduit to facilitate invasion and metastasis in breast cancer cells through HER2/HER3 signaling. Oncogene. 2016;35(35):4633–40. Epub 20160125. doi: 10.1038/onc.2015.527 . PubMed PMID: 26804171; PubMed Central PMCID: PMC4959990. Huang KH, Chow KC, Chang HW, Lin TY, Lee MC. ATPase family AAA domain containing 3A is an anti-apoptotic factor and a secretion regulator of PSA in prostate cancer. Int J Mol Med. 2011;28(1):9–15. Epub 20110406. doi: 10.3892/ijmm.2011.670 . PubMed PMID: 21584487. Lang L, Teng Y. Using Genome-Editing Tools to Develop a Novel In Situ Coincidence Reporter Assay for Screening ATAD3A Transcriptional Inhibitors. Methods Mol Biol. 2020;2138:159–66. doi: 10.1007/978-1-0716-0471-7_8. PubMed PMID: 32219745; PubMed Central PMCID: PMC8552142. Peralta S, Goffart S, Williams SL, Diaz F, Garcia S, Nissanka N, et al. ATAD3 controls mitochondrial cristae structure in mouse muscle, influencing mtDNA replication and cholesterol levels. J Cell Sci. 2018;131(13). Epub 20180704. doi: 10.1242/jcs.217075 . PubMed PMID: 29898916; PubMed Central PMCID: PMC6051345. Baudier J. ATAD3 proteins: brokers of a mitochondria-endoplasmic reticulum connection in mammalian cells. Biol Rev Camb Philos Soc. 2018;93(2):827–44. Epub 20170920. doi: 10.1111/brv.12373 . PubMed PMID: 28941010. Gilquin B, Cannon BR, Hubstenberger A, Moulouel B, Falk E, Merle N, et al. The calcium-dependent interaction between S100B and the mitochondrial AAA ATPase ATAD3A and the role of this complex in the cytoplasmic processing of ATAD3A. Mol Cell Biol. 2010;30(11):2724–36. Epub 20100329. doi: 10.1128/mcb.01468-09 . PubMed PMID: 20351179; PubMed Central PMCID: PMC2876520. Zhao Y, Sun X, Hu D, Prosdocimo DA, Hoppel C, Jain MK, et al. ATAD3A oligomerization causes neurodegeneration by coupling mitochondrial fragmentation and bioenergetics defects. Nat Commun. 2019;10(1):1371. Epub 20190326. doi: 10.1038/s41467-019-09291-x . PubMed PMID: 30914652; PubMed Central PMCID: PMC6435701. Arguello T, Peralta S, Antonicka H, Gaidosh G, Diaz F, Tu YT, et al. ATAD3A has a scaffolding role regulating mitochondria inner membrane structure and protein assembly. Cell Rep. 2021;37(12):110139. doi: 10.1016/j.celrep.2021.110139 . PubMed PMID: 34936866; PubMed Central PMCID: PMC8785211. Li S, Clémençon B, Catty P, Brandolin G, Schlattner U, Rousseau D. Yeast-based production and purification of HIS-tagged human ATAD3A, A specific target of S100B. Protein Expr Purif. 2012;83(2):211–6. Epub 20120421. doi: 10.1016/j.pep.2012.04.005 . PubMed PMID: 22542587. Ramachandran GN, Ramakrishnan C, Sasisekharan V. Stereochemistry of polypeptide chain configurations. J Mol Biol. 1963;7:95–9. doi: 10.1016/s0022-2836(63)80023-6. PubMed PMID: 13990617. Sali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol. 1993;234(3):779–815. doi: 10.1006/jmbi.1993.1626 . PubMed PMID: 8254673. Feig M, Mirjalili V. Protein structure refinement via molecular-dynamics simulations: What works and what does not? Proteins. 2016;84 Suppl 1(Suppl 1):282 – 92. Epub 20150817. doi: 10.1002/prot.24871. PubMed PMID: 26234208; PubMed Central PMCID: PMC5493977. Trott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455–61. doi: 10.1002/jcc.21334 . PubMed PMID: 19499576; PubMed Central PMCID: PMC3041641. Irwin JJ, Tang KG, Young J, Dandarchuluun C, Wong BR, Khurelbaatar M, et al. ZINC20-A Free Ultralarge-Scale Chemical Database for Ligand Discovery. J Chem Inf Model. 2020;60(12):6065–73. Epub 20201029. doi: 10.1021/acs.jcim.0c00675 . PubMed PMID: 33118813; PubMed Central PMCID: PMC8284596. RDKit: Open-source cheminformatics; http://www.rdkit.org . Yan Y, Zhang D, Zhou P, Li B, Huang SY. HDOCK: a web server for protein-protein and protein-DNA/RNA docking based on a hybrid strategy. Nucleic Acids Res. 2017;45(W1):W365-w73. doi: 10.1093/nar/gkx407 . PubMed PMID: 28521030; PubMed Central PMCID: PMC5793843. UniProt: the Universal Protein Knowledgebase in 2023. Nucleic Acids Res. 2023;51(D1):D523-d31. doi: 10.1093/nar/gkac1052 . PubMed PMID: 36408920; PubMed Central PMCID: PMC9825514. Jafari E, Khajouei MR, Hassanzadeh F, Hakimelahi GH, Khodarahmi GA. Quinazolinone and quinazoline derivatives: recent structures with potent antimicrobial and cytotoxic activities. Res Pharm Sci. 2016;11(1):1–14. PubMed PMID: 27051427; PubMed Central PMCID: PMC4794932. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403–10. doi: 10.1016/s0022-2836(05)80360-2 . PubMed PMID: 2231712. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, et al. Highly accurate protein structure prediction with AlphaFold. Nature. 2021;596(7873):583–9. Epub 20210715. doi: 10.1038/s41586-021-03819-2 . PubMed PMID: 34265844; PubMed Central PMCID: PMC8371605. Eberhardt J, Santos-Martins D, Tillack AF, Forli S. AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model. 2021;61(8):3891–8. Epub 20210719. doi: 10.1021/acs.jcim.1c00203 . PubMed PMID: 34278794. The PyMOL Molecular Graphics System, Version 2.5, Schrödinger, LLC. Jo S, Kim T, Iyer VG, Im W. CHARMM-GUI: a web-based graphical user interface for CHARMM. J Comput Chem. 2008;29(11):1859–65. doi: 10.1002/jcc.20945 . PubMed PMID: 18351591. Jo S, Jiang W, Lee HS, Roux B, Im W. CHARMM-GUI Ligand Binder for absolute binding free energy calculations and its application. J Chem Inf Model. 2013;53(1):267–77. Epub 20121220. doi: 10.1021/ci300505n . PubMed PMID: 23205773; PubMed Central PMCID: PMC3557591. Sushko I, Novotarskyi S, Körner R, Pandey AK, Rupp M, Teetz W, et al. Online chemical modeling environment (OCHEM): web platform for data storage, model development and publishing of chemical information. J Comput Aided Mol Des. 2011;25(6):533–54. Epub 20110610. doi: 10.1007/s10822-011-9440-2 . PubMed PMID: 21660515; PubMed Central PMCID: PMC3131510. Van Rossum G, Drake Jr. FL. Python reference manual. Centrum voor Wiskunde en Informatica Amsterdam; 1995. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3280889","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":228044752,"identity":"8cb40a96-ca38-4ed6-81d8-a6dcb0339921","order_by":0,"name":"Kevin Gong","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA80lEQVRIiWNgGAWjYBACCSA+AMQ8YN4HIGZjJ1YLSA/jDJAWZiK0gAFICzPYLkJaJNt7DA8XMNTJ2LP3Hn5t82ubPB8zA+OHjzm4tUjznDE4PIPhMA8Pz7k069y+24ZtzAzMkjO34dYiJ5GWcJiH4QAPj0SOmXFuz21GoBY2Zl58WuSfgbTUQbRY9ty2J6hFWoL5AFALM0iL8WOGH7cTCWqR7EkGajEA+uXMGTPG3obbyW3MjM14/SJx/GDzZ56KOnv29h7jDz/+3Lad39588MNHPFogwABMskkwtoFoxgZC6uGA+QPDH6IVj4JRMApGwQgCANu9R2s0JQxzAAAAAElFTkSuQmCC","orcid":"","institution":"Rice University","correspondingAuthor":true,"submittingAuthor":false,"prefix":"","firstName":"Kevin","middleName":"","lastName":"Gong","suffix":""},{"id":228044753,"identity":"0f488bfc-fc81-4bc7-9c12-d07ed7f4a42d","order_by":1,"name":"Patrick McKenna","email":"","orcid":"","institution":"CISD Academy of Science \u0026 Technology","correspondingAuthor":false,"submittingAuthor":false,"prefix":"","firstName":"Patrick","middleName":"","lastName":"McKenna","suffix":""}],"badges":[],"createdAt":"2023-08-21 03:29:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3280889/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3280889/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":42018482,"identity":"dd42cd89-f8bc-4fa8-9f8c-d20e07a929f6","added_by":"auto","created_at":"2023-08-23 14:48:08","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":218600,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eS100B Binding Site of ATAD3\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConstructed model and Ramachandran plot of ATAD3A S100B binding domain. \u003cstrong\u003ea) \u003c/strong\u003eModel of S100B binding domain of ATAD3A developed with the Modeller Software and refined with the Feig Lab PREFMD web server. Binding domain side chains are shown as sticks. Key S100B binding residues are labeled. \u003cstrong\u003eb) \u003c/strong\u003eRamachandran plot of ATAD3A C terminal. No Ramachandran Outliers are found in the S100B binding domain (aa 338-353)\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/581e0f16a67ab0c398181a3c.png"},{"id":42019249,"identity":"774c1e49-d578-45b1-b74f-38fa6dcc901a","added_by":"auto","created_at":"2023-08-23 14:56:08","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":71356,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHit Compounds\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCompounds identified to possess high binding affinity for ATAD3A (-8.2 kcal/mol to -10.1 kcal/mol) within the AutoDock Vina search grid and to have passed physicochemical filters including the Lipinski Rule of 5, Veber Filter, Ghose Filter, REOS Filter and Drug-like Filter.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/81a6e31759d64a57c573e075.png"},{"id":42018483,"identity":"06f973ae-c8d1-43fd-8291-3d945a6b78b0","added_by":"auto","created_at":"2023-08-23 14:48:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":462942,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eATAD3A-S100B Binding Visualized in HDOCK\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eVisualization of binding between the ATAD3A S100B binding site and Dimeric S100B through protein-protein docking in HDOCK. Binding residues are shown as sticks, while other residues are shown in cartoon. The ATAD3A S100B binding site is colored in magenta, and S100B residues are colored in yellow. Length of binding contacts are labeled in Angstroms. (\u003cstrong\u003ea)\u003c/strong\u003e Overall structure of the binding site is depicted. \u003cstrong\u003e(b) \u003c/strong\u003eBinding contacts between the residue V341 of the S100B binding site and residue H85 of the carboxyl terminal domain of S100B. \u003cstrong\u003e(c) \u003c/strong\u003eBinding contacts between the residue L345 of the S100B binding site and residue F87 of the carboxyl terminal domain of S100B. \u003cstrong\u003e(d) \u003c/strong\u003eBinding contacts between the residues I339, T340 and A344 of the S100B binding site and residue H85 of the carboxyl terminal domain of S100B. \u003cstrong\u003e(e) \u003c/strong\u003eBinding contacts between the residues I349, S352 and R353 of the S100B binding site and residues L44 and H46 of the S100B linker domain.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/ddaf6795bc4ab082d78ea1cc.png"},{"id":42018484,"identity":"90cb6bb8-99ff-415e-a0cf-4482d98b6106","added_by":"auto","created_at":"2023-08-23 14:48:08","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":542188,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eZINC20728831 \u0026amp; ZINC6235062 Docked in S100B Binding Domain\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePromising hit compounds are shown docked in their best binding mode. The hit compounds are shown in green, and the binding site residues are shown as sticks. The S100B binding site residues are colored magenta, and the rest of the protein is colored blue. \u003cstrong\u003e(a) \u003c/strong\u003eHit compound ZINC20728831 (N-benzhydryl-4-oxoquinazoline-2-carboxamide) with the residues T340, V341 and L342 of the S100B binding domain. \u003cstrong\u003e(b) \u003c/strong\u003eHit compound ZINC6235062 (3-hydroxy-4-[(2S)-3-hydroxy-4-oxo-1,2-dihydroquinazolin-2-yl]naphthalene-2-carboxylic acid) docked with residue T340 within the S100B binding domain.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/415cf2886dd190e071d7a890.png"},{"id":42018487,"identity":"6fb68a58-87c5-45b8-8fca-119938a0c154","added_by":"auto","created_at":"2023-08-23 14:48:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":495499,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOher Hit Compounds Docked in ATAD3A\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe best binding mode of each hit compound is shown. Binding residues are shown as sticks, while the rest of the protein is shown as cartoon. The S100B binding site is colored magenta, and the Walker B motif is colored salmon (light pink). The remainder of the protein is colored blue. \u003cstrong\u003e(a) \u003c/strong\u003eHit compound ZINC6993962 docked with residue R452 at the Walker B motif adjacent to the S100B binding site in the Docking grid. (\u003cstrong\u003eb) \u003c/strong\u003eHit compound ZINC5048139 docked with residues R338, D420, R451 and R452 at portions of the Walker B motif and S100B binding site. \u003cstrong\u003e(c) \u003c/strong\u003eHit compound ZINC6235061 docked with residues E343, H347 and R354 of the S100B binding site. \u003cstrong\u003e(d) \u003c/strong\u003eHit compound ZINC401340 docked with ATAD3A residue T336.\u003cstrong\u003e (e) \u003c/strong\u003eHit compound ZINC5082118 docked with residues R338 and R451, contacting both the S100B binding sequence and Walker B motif of ATAD3A.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/1ec84acc5586922c43303063.png"},{"id":46840496,"identity":"8f9d7370-5670-44bf-80d3-75032fbec02d","added_by":"auto","created_at":"2023-11-21 10:29:50","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2023145,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3280889/v1/39278466-ea36-46f8-b7be-c95260ef7b36.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"In-Silico Drug Discovery Through Interaction-Based Perspectives for ATAD3A","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eIn 2020, around 18.1\u0026nbsp;million people were diagnosed with cancer, and cancers accounted for nearly 10\u0026nbsp;million deaths, making cancer the leading cause of death for humans around the world (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). The development of treatment for all types of cancer has been a massive front in scientific research for decades, with billions of dollars and countless researchers devoted to its cause (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). In the world of modern anticancer therapeutic drug development, numerous mitochondrial enzymes have been identified as attractive targets for drug development, as they have been found to be overexpressed and upregulated in tumor cells and critical in oncogenic signaling and pathways necessary to support cancer metastasis and rapid proliferation (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMitochondrial membrane oncoprotein ATPase family AAA domain containing protein 3A (ATAD3A) has been found to be upregulated and implicated for cancer proliferation, oncogenic signaling, tumor growth and metastasis in certain types of cancer, consequently resulting in its upregulation being associated with worsened prognosis for cancer patients (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Specifically, studies have found that the abnormal activation of ATAD3A in cancer cells causes mitochondrial oncogenic signaling, enhancing activities and functions that promote tumor growth (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Among these, overexpression of ATAD3A leads to breast cancer metastasis by increasing the stability of WASF3 protein, a known tumor metastasis promoter, with the ER GRP78 protein, whose expression is reduced by ATAD3A knockdown (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). In prostate cancer, upregulation of ATAD3A correlates with serum prostate-specific antigen (PSA) levels and androgen receptor expression (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). In hepatocellular carcinoma, urothelial carcinoma, prostate cancer, glioma, breast cancer and lung adenocarcinoma, ATAD3A upregulation is correlated with reduced overall patient survivability, worsened prognosis, proliferation, and tumor growth (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Based on the dire consequences of the oncogenic effects of ATAD3A and as previously suggested in a 2020 study by Lang \u0026amp; Teng, the targeting of ATAD3A for drug development could yield great potential, and two studies have already shown that knockdown of ATAD3A with shRNA has been found to significantly induce repression of tumor growth and metastasis for breast and colon cancer in mice lung models (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Currently, however, an effective, ATAD3A specific inhibitor has yet to be developed, and most studies up to date focus on the process of developing transcriptional inhibitors as there is no existing experimentally resolved structure for ATAD3A and there is little understanding on its molecular mechanisms (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eATAD3A is a mitochondrial transmembrane protein of the AAA+ (ATPases associated with various cellular activities) superfamily (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). As an AAA protein, ATAD3A is thus a mechanochemical enzyme, or motor protein, in its hexameric aggregate. In terms of in-depth structure, ATAD3A is localized to the mitochondrial inner membrane (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Within its highly conserved mitochondrial matrix C terminal end from (aa 264\u0026ndash;634), ATAD3A contains an S100B binding site and ATPase domain with the Walker A (P loop) and Walker B motifs associated with phosphate binding and ATP hydrolysis respectively (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Beyond the inner mitochondrial membrane, which is spanned by a transmembrane domain from aa 247\u0026ndash;264, the lack of conservation with other protein sequences and lack of an experimentally resolved structure of ATAD3A has resulted in the lack of concrete structural information of the N terminal region (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). The lack of structural information on the N terminal and absence of an experimentally resolved structure have further led ATAD3A to be deemed infeasible for activity inhibitor design (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). S100B, a zinc and calcium-binding protein with a chaperone like function in ATAD3A folding and mitochondrial targeting, binds with ATAD3A at its S100B binding site (aa 290\u0026ndash;310) (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis study seeks to investigate the potentiality of drug development for a conventionally undruggable target, the ATAD3A oncoprotein, through exploring a novel approach employing a site specific, interaction-based perspective. Specifically, the ATAD3A oncoprotein has been labeled undruggable due to its difficulty to purify and crystallize as a mitochondrial cristae membrane protein, resulting in its lack of experimentally resolved structure and inability to be assessed in-vitro as a whole protein. With these limitations, a novel approach for drug development is required which enables an in-silico framework and bypasses the structural information obstacle.\u003c/p\u003e \u003cp\u003eIn our approach, the well-established and conserved S100B-ATAD3A interaction region is modeled with high confidence and independent of the other regions of ATAD3A. The modeled interaction is then targeted to induce conformational change and improper topology of the ATAD3A C terminal ATPase, which theoretically results in the suppression of ATPase enzymatic and hexameric activity. This proposed mechanism is supported by literature on ATAD3A synthesis, where complex topology, aggregation prone regions of ATAD3A, notably the ATPase domain, require S100B to facilitate cytoplasmic protein folding and processing and mitochondrial targeting (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Consequently, as suggested by previous literature, with decreased levels of ATAD3A mitochondrial targeting and conformationally altered ATPase domains, the oncogenic functions of ATAD3A can be theoretically mitigated, as studies attribute ATAD3A ATPase function with oncogenic pathways and ATAD3A known to scaffold tumor promoters (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). In this study, we discovered compounds that can act as inhibitors of ATAD3A through competitive inhibition of the ATAD3A S100B binding domain and potentially suppress the oncogenic pathways of ATAD3A.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eATAD3A Model Development for Drug Discovery\u003c/h2\u003e \u003cp\u003eWith the structure of ATAD3A currently difficult to be experimentally resolved, an in-silico approach had to be utilized to develop a model for ATAD3A drug discovery. However, it is difficult to develop a sufficiently accurate model utilizing in-silico methodologies to serve as the basis of drug discovery due to the regions of ATAD3A extending from across the intermembrane space to the outer mitochondrial membrane being difficult to model accurately due to the lack of homologous sequences as well as a lack of well-defined tertiary structures, with the exception of two coiled coil domains (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Without a sufficiently accurate structural model of the target protein, conventional drug discovery methodology becomes infeasible. To overcome this, we focus on the S100B binding site of the ATAD3A oncoprotein as the binding site for drug discovery and the ATPase domain as the region to be suppressed. The process of target validation in the drug discovery pipeline, the validation of the target\u0026rsquo;s therapeutic value, is substituted by established experimental characteristics in previous literature deducing the ATPase function to be critical for oncogenic effect as well as the necessity of S100B binding for proper ATPase folding and function (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). With these deductions, the regions of the structure of ATAD3A which necessitates accuracy becomes reduced to within the highly conserved C terminal domain containing the S100B binding domain, enabling the construction of a model to high accuracy due to the homologous structures, enabling sufficient confidence to initiate drug discovery. Indeed, when assessed with a Ramachandran plot, few Ramachandran outliers were identified for the C terminal and none identified in the S100B binding domain of the model of ATAD3A developed through homology modeling in the Modeller software,, suggesting the accuracy of the model developed in homology modeling (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb\u003cb\u003e)\u003c/b\u003e (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). This confidence in accuracy was further validated through a protein model refinement web server, Feig Lab PREFMD, where the refined model was virtually identical to the inputted model (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea\u003cb\u003e)\u003c/b\u003e (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). With an accurate model of the ATAD3A C terminal S100B binding site (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea), drug discovery targeting the S100B binding of ATAD3A could be initiated.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eDrug Development and Binding Analysis\u003c/h2\u003e \u003cp\u003eLarge scale virtual screening through in-silico docking with AutoDock Vina on the Zinc Chemical Library was utilized to conduct drug discovery (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). All screened ligands in the virtual screening library were then sorted by the binding affinity of their best binding conformation, with the top 1% (ranging from \u0026minus;\u0026thinsp;8.2 to -10.1 kcal/mol) selected for subsequent phases and identified as compounds with high binding affinities for ATAD3A in the 20x20x20 angstrom search grid, or area on the target protein where docking is conducted.\u003c/p\u003e \u003cp\u003eThese compounds were then evaluated with physicochemical filters from RDKit (Lipinski Rule of 5, Veber, Drug-like, Ghose and REOS filters) to filter out compounds with unsatisfactory properties such as large molecular weight, LogP and number of hydrogen bond acceptors/donors (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). All compounds that passed the filters were identified as hit compounds and selected for Quantitative Structure-Activity Relationship (QSAR) model evaluation (shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eHit compounds possessing binding affinities stronger than \u0026minus;\u0026thinsp;8 kcal/mol for the S100B binding site, indicating very strong binding, can competitively inhibit S100B binding of ATAD3A by occupying the S100B binding site and preventing the ATAD3A-S100B interaction, and the mitochondrial targeting of ATAD3A as well as proper folding of ATAD3A are consequently inhibited through the loss of the calcium dependent mechanism (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Thus, it becomes critical to determine the ability for the hit compounds to occupy the residues critical for S100B binding.\u003c/p\u003e \u003cp\u003eTo better determine the ability for the compounds to occupy the residues critical for S100B binding and thus inhibit ATAD3A-S100B binding, we utilized the HDOCK web server, a protein-protein docking utility, to analyze the S100B binding of ATAD3A (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). Analyzing the best docking mode between ATAD3A and S100B predicted by HDOCK in PyMOL, we identified the residue sites of the binding contacts between ATAD3A and S100B (Fig.\u0026nbsp;3). Binding contact length was cut off at 3.5 angstroms to show only significant hydrogen and disulfide bonds, with the exception of V341, which formed\u0026thinsp;\u0026lt;\u0026thinsp;4.0 angstrom C-H bonds with every point on the aromatic ring of the S100B residue H85 (Fig.\u0026nbsp;3b). As demonstrated in \u003cb\u003eFig.\u0026nbsp;3\u003c/b\u003e, the main interfaces of interaction between ATAD3A and S100B are situated at the site consisting of residues I339, T340, V341 \u0026amp; A344 and the site of residue L345 with the S100B carboxyl terminal domain residues F87 and H85 respectively, as well as the site consisting of residues I349, S352 \u0026amp; R353 with the S100B linker domain residues E46 and L44. To further assess the results of the protein-protein docking, we revisited the mutagenesis annotations on UniProt from established (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e) Notably, of the main sites of contact identified in protein-protein docking, three binding residues matched with mutagenesis annotations on UniProt, which described the complete loss of S100B binding with the mutagenesis of both the residues V341 and L345 and decrease in S100B binding with the mutagenesis of residue I349 (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). We here utilize the reports of S100B binding loss or decrease as a result of residue mutagenesis as an indicator of the necessity of availability of the residues in ATAD3A-S100B binding. Thus, the mutagenesis annotations correlate to the interactions demonstrated in the protein-protein docking, affirming its validity and emphasize the importance of the availability of the residues V341 and L345 in the S100B binding of ATAD3A (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThus, the ability of inhibitor compounds to occupy and bind to these residues is critical to the inhibition of ATAD3A-S100B binding, which enables the endogenous processing mechanism of S100B (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Of the hit compounds developed, two of them have their highest affinity binding modes occupying these exact sites. The hit compound ZINC20728831 (N-benzhydryl-4-oxoquinazoline-2-carboxamide) occupies the binding site at V341, establishing contacts with the residues T340, V341 and L342 as demonstrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003ea. The hit compound ZINC6235062 (3-hydroxy-4-[(2S)-3-hydroxy-4-oxo-1,2-dihydroquinazolin-2-yl]naphthalene-2-carboxylic acid), binds to the adjacent T340, a similarly significant residue in ATAD3A S100B binding, and is spatially oriented to block access to V341, as seen in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003eb. However, many of the hit compounds were also found to have high binding affinity binding poses with the ATPase region of ATAD3A adjacent to the S100B binding site, a region that also happened to be within the defined AutoDock Vina search grid for docking. Upon further inspection, this region was identified as the Walker B motif, with a similar LLFVDE sequence as the Walker B motif identified in the 586 aa isoform and coinciding with the hhhhDE of the consensus Walker B sequence. The proximity of the S100B binding site to the Walker B motif is thus of note, as it is possible that S100B binding plays a critical role in the folding and proper ATP hydrolysis function of the Walker B motif. All highest affinity binding modes of each hit molecule are demonstrated in \u003cb\u003eFig.\u0026nbsp;5\u003c/b\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFrom observations of all the binding modes calculated by AutoDock Vina, it becomes clear that the main residues responsible for binding contacts with the hit compounds within the S100B binding site are at the side chains of the T340 residue and the R338 residue, which are both characteristically associated as polar binding residues. Similarly significant in binding contacts were the V341 and L342 residues, which were also commonly identified as ligand sites in PyMOL. Though our protein-protein docking results indicate that there are indeed contacts between the R338 residue and the S100B protein, such contact is greater than 4 angstroms in length and associated with the S100B linker between the two EF hands, not the carboxyl terminal domain associated with protein-protein interactions and binding (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Additionally, as the V341 residue is well documented as being critical to S100B binding and is shown in protein-protein docking to closely interact with the S100B carboxyl terminal domain, the occupancy of this particular residue is especially crucial, and it is found that ZINC20728831 consistently binds to this residue in its 5 highest binding affinity binding modes, rendering it the most promising compound for the inhibition of ATAD3A-S100B binding. The adjacent T340 is similarly critical for ATAD3A S100B binding with its projected hydrogen bonding with the S100B residue H85 (Fig.\u0026nbsp;3d), which ZINC20728831 and ZINC6235062 bind to consistently, further yielding promise for both compounds (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e4\u003c/span\u003eb).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Inference\u003c/h2\u003e \u003cp\u003eTo assess the statistical significance of the binding affinities of the hit compounds developed compared the average binding affinity of all compounds, Z tests were conducted. The Z test results demonstrated that each hit compound had a Z-score of around 6, corresponding to P-values significantly less than the alpha of .01, thus demonstrating that the hit compounds were statistically significantly stronger in binding with ATAD3A in the defined search grid compared to the average compound in the virtual screening library (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1. Z-Test Results\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"613\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eCompound Name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eZ-Score\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC401340\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.411925508\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC5082118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.241072954\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC6235062\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.241072954\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC6235061\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.582778063\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC5048139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.0702204\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC5082118\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.0702204\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"45.02446982055465%\" valign=\"top\"\u003e\n \u003cp\u003eZINC20728831\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.427406199021206%\" valign=\"top\"\u003e\n \u003cp\u003e-6.411925508\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"28.548123980424144%\" valign=\"top\"\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eThe Z-score and P-value of each hit compound in Z-test with the average binding affinity of all compounds in virtual screening library (800,000 randomly compounds from ZINC20 library). An significance level of \u0026alpha; = 0.01 was utilized to determine significance.\u003c/p\u003e\n\u003ch3\u003eQSAR and CHARMM-GUI Results\u003c/h3\u003e\n\u003cp\u003eQuantitative Structure-Activity Relationship (QSAR) models trained on PubChem data were utilized to assess the activity of the hit compounds. The trained QSAR models, Multiple Linear Regression, K-Nearest Neighbor, Random Forest Regression and Associative Neural Network developed on the OCHEM web platform, were evaluated, in which the model with the highest combination of accuracy, balanced accuracy, and area under the curve (AUC) was selected to predict the activity of the hit compounds. This comparison revealed that the Associative Neural Network was the most successful model by each measure of accuracy (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). This model was thus applied to perform activity prediction for the hit compounds.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eQSAR Models\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel Type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBalanced Accuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAssociative Neural Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e76%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e76%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e81%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom Forest Regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e73%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e75%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e83%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eK-Nearest Neighbor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e79%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e80%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMultiple Linear Regression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e65%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e74%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e76%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eAll QSAR models trained within OCHEM, includes their respective Accuracy, Balanced Accuracy, and AUC. The best model was the one with the highest combination of all three of these criteria.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe free energy of binding values calculated by CHARMM-GUI along with AutoDock Vina binding affinity values indicate strong binding between the developed hit compounds and ATAD3A within the AutoDock Vina search box. Additionally, QSAR activity prediction condition for all but compound ZINC6235061 demonstrates the high likelihood of binding activity between the active compounds and ATAD3A (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e\u003cb\u003eQSAR and CHARMM-GUI Results\u003c/b\u003e\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCompound Name\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActivity Prediction\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMolecular Weight (kDa)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eBinding Affinity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eFree Energy of Binding (KJ/mol)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC401340\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e338.318\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-10.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC5082118\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e380.488\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-7.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC6235062\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e350.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-9.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC6235061\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInactive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e350.33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-9.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC5048139\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e394.471\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-8.3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC5082118\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e374.318\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-8.9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eZINC20728831\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eActive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e355.397\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-8.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-9.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eQSAR activity prediction, molecular weight, binding affinity, and free energy of binding of all hit compounds is shown above.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eOur study successfully investigated a pragmatic approach in drug development utilizing new site specific, interaction-based perspectives, employing a basis of previous literature to bypass technical barriers of conventional orthosteric drug development for the ATAD3A oncoprotein. Instead of pursuing conventional approaches of resolved structure to direct inhibition of an orthosteric site, our study utilized an allosteric approach of indirect inhibition, where critical binding interactions are suppressed to directly induce conformational change and mechanistic changes which then consequentially mitigate oncogenic pathways, as described in previous literature (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Allowing the target site to be an established critical binding interaction site and supporting the allosteric inhibitory effect on the oncogenic domain (ATPase) with established mechanisms, we bypass the barrier of the necessity of an experimentally resolved structure of the whole target protein and specific active site architecture by initiating drug discovery for a site which we can confidently construct a model of and occupy, thus dismantling the critical binding interaction.\u003c/p\u003e \u003cp\u003eMethodologically, we constructed a model for ATAD3A in homology modeling and utilized refinement web servers and Ramachandran plot to assess the accuracy of the developed model. With a constructed model, we employed in-silico drug discovery in virtual screening through large scale docking. Developed hit compounds were then filtered and assessed in subsequently applied physicochemical filter and QSAR modeling. Through this in-silico framework we were able to substantially substitute the conventional drug discovery pipeline with completely computational alternatives, as necessitated by the inability for in-vitro work to be conducted on membrane proteins such as ATAD3A. This investigation in computational alternatives in the drug development framework thus expands the feasibility of new perspectives such as the one outlined in this study and drug development relying heavily on computational methodologies for conventionally undruggable targets, many of which are difficult to purify for in-vitro work.\u003c/p\u003e \u003cp\u003eThe developed hit compounds ZINC6235062 (3-hydroxy-4-[(2S)-3-hydroxy-4-oxo-1,2-dihydroquinazolin-2-yl]naphthalene-2-carboxylic acid) and ZINC20728831 (N-benzhydryl-4-oxoquinazoline-2-carboxamide) possess high values of binding affinity and low free energy of binding for critical residues necessary for S100B binding at the S100B binding domain, most notably the residues T340 and V341, of which V341 has been experimentally determined to be critical for S100B binding (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Both compounds share similar characteristics in their docking with the S100B binding domain, hydrogen bonding with the oxygen atom of the hydroxyl group on the T340 side chain through the hydrogen at nitrogen atom in the 4-quinazolinone group, which has been of interest in medicinal chemistry (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e). The oxygen of carbonyl group in 4-quinazolinone of ZINC20728831 also forms hydrogen bonds with the hydrogen of amines of both V341 and L342, and the hydrogen in amide bond which links the 4-quinazolinone and naphthalene moiety also binds with the same T340 hydroxyl group oxygen. These series of hydrogen bonds in the docking of ATAD3A with the critical S100B binding domain residues reinforce the deduction of strong binding affinity between ZINC20728831 and the ATAD3A S100B binding domain. Of note, the epimers ZINC6235061 and ZINC6235062, which are sterically different in the link between their quinazolinone and 3-hydroxy-2-naphthoic acid groups, were found to dock to different sites, and the activity predicted by QSAR modeling indicates ZINC6235061 as inactive and ZINC6235062 as active. This suggests the difference stereochemistry between the two epimers being of great importance to their binding activities, especially as their 3-hydroxy-2-naphthoic acid groups adopt similar spatial orientations in the binding domain region, and the steric differences between the two compounds resulting in different orientations of the critical quinazolinone group. Taken together with the findings on ZINC20728831, it appears that the quinazolinone group is critical for the compounds ZINC20728831 and ZINC6235062 to dock to the critical S100B binding domain residues, which indicates the ability to competitively inhibit S100B binding in ATAD3A and the consequent oncogenic effects of endogenous ATAD3A through occupancy of the S100B binding domain.\u003c/p\u003e \u003cp\u003eCoupled with the known severity of ATAD3A oncogenic functions in multiple of the deadliest cancers, the ATAD3A inhibitors developed in this study, thus, in their potential direct suppression of S100B binding and indirect inhibition of the oncogenic pathways of ATAD3A, yield potential for therapeutic development and can be further assessed and characterized for inhibitory effect in cellular in-vitro testing. Thus, as an in-silico study into the ATAD3A oncoprotein, we successfully present a new approach in drug development for ATAD3A, and demonstrate the development of inhibitors along such an approach, as well as expand on previous studies investigating drug development options for targeting ATAD3A. Broadly speaking, we believe our investigation into a more pragmatic approach in drug development exploring new allosteric interaction-based, site specific perspectives allows for an innovative insight into more flexible thinking and strategy in drug development, which has the potential to enable the drug development for more targets previously labeled infeasible.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eRequired Materials\u003c/h2\u003e \u003cp\u003eA Sam Houston State University Computer Science Department Intel 32 core workstation was utilized through the generosity of Computer Science Department Professor Dr. Frank Liu. Online and software resources utilized in this study include the NCBI BLAST program, Modeller 10.3 software, AlphaFold2, Autodock Vina software, AutoDock Tools software, PyMOL software, CHARMM-GUI, OCHEM Web Server, UniProt Database, Feig Lab PrefMD Web Server and Python 3.11 (\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan additionalcitationids=\"CR23 CR24 CR25 CR26 CR27 CR28\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). The ZINC20 Chemical Library was accessed for the virtual screening library of compounds, and the rdkit repository on GitHub was utilized to implement physicochemical filters (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eHomology Modeling and Model Construction\u003c/h2\u003e \u003cp\u003eThe structure of ATAD3A was constructed utilizing homology modeling and AlphaFold\u0026rsquo;s statistical probabilities based on known structures with sequence similarity (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). To identify analogous templates ATAD3\u0026rsquo;s amino acid sequence was identified from UniProt and uploaded into NCBI BLAST (Basic Local Alignment and Search Tool) (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). The PDB files of the sequences with the highest sequence homology (similar sequences with resolved structures) were then downloaded to be utilized as template sequences, all of which met the requirement of at least 30% sequence similarity, the standard accepted minimum percent similarity for homology modeling to be employed accurately. As it was discovered that nearly all homologous template sequences possessed sequence similarity with the ATAD3A C terminal ATPase, the other regions of the protein were isolated to find homologous sequences, particularly the N terminal domain (aa 1\u0026ndash;70) and Intermembrane domain (aa 71\u0026ndash;294). BLAST was then run on each of these segments separately to identify homologous regions for each region specifically, with the same requirement of 30% sequence similarity (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Regions that continued to lack a homologous model (disordered regions), most notably the intermembrane domain, were inputted into Alphafold 2 to generate candidate structures, which were incorporated into the structure with the other PDB templates for each of the other regions (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Each template sequence of ATAD3A was then aligned, combined, and built within the Modeller software (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e). The model was then evaluated using a Ramachandran plot and outliers were noted (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Finally, the model was refined with the Feig Lab PREFMD webserver (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eIn-silico Drug Discovery\u003c/h2\u003e \u003cp\u003eIn in-silico drug discovery, the S100B binding site was extracted from the ATAD3A model through PyMOL, compared with published structural characteristics of canonical binding (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e). As determined by an experimentally assessed ATAD3A S100B binding site in complex with S100B, the S100B binding mechanism of ATAD3A is a canonical p53-S100B binding interaction, and thus the canonical S100B-p53 bond was utilized to assess the validity of our S100B binding site structure, with the contacts identified in the 2010 study by Gilquin et al serving as reference (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). Subsequently, 800,000 randomly sampled compounds from the Zinc20 in stock chemical library were downloaded and prepared in pdbqt file format as the ligand library for AutoDock Vina (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). The ATAD3A S100B binding site was then prepared as the receptor in PyMOL and AutoDock Tools. A python script was then written to iterate through all the ligand files and adapt Vina for virtual screening (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). The ligands were then screened by AutoDock Vina through docking with the S100B binding site of ATAD3A. Following the screen, a python script was written to sort all the output ligand files by their binding affinity and the top 1% identified. These output ligand files were then moved into a different folder and screened in a miniconda environment with the Veber, Ghose, Lipinski Rule of 5, Rule of 3, REOS and Drug Like filters imported from rdkit repository through executing a python implementation script modified to iterate through the folder and output log files for each hit compound (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). A python script was then written to identify the compounds that passed all filters as hit compounds (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e). The free energy of binding was then calculated with the use of CHARMM-GUI (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eCreation of QSAR for Activity Prediction\u003c/h2\u003e \u003cp\u003eThe QSAR (Quantitative Structure-Activity Relationship) was constructed with the use of a publicly available data set on PubChem, referenced as \u0026ldquo;Blockade of RAGE activation in vascular endothelium\u0026rdquo; and accession ID: 482 (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). This assay measured the level of phosphorylated ERK1/2 in endothelial cells in the presence of 25 microg/mL S100b. This assay utilized colorimetry to quantitate phosphorylated ERK1/2 in an in-situ cell-based ELISA. In this assay, an active compound was defined as Phospho-ERK1/2 activity\u0026thinsp;\u0026lt;\u0026thinsp;60% of S100b stimulated level for the plate tested and inactive was \u0026gt;\u0026thinsp;60%. Through this, 25 compounds were identified as \u0026ldquo;active\u0026rdquo; and 1015 were identified as \u0026ldquo;inactive\u0026rdquo;. This data set was then downloaded and uploaded into OCHEM. Compound structures underwent preprocessing with Standardization, Neutralize, Remove Salts, and Clean Structure using the ChemAxon Standardizer, and variable selection was completed on the initial data set to clean the data set, remove redundant features, and conduct feature selection. The data set was divided into two sets: \u0026ldquo;training set\u0026rdquo; (80% of data) and \u0026ldquo;test Set\u0026rdquo; (20% of data). Then based on feature selection, the descriptors selected were ALogPS, OEState, and ChemAxon descriptors, and with parameters set to remove over exhausted descriptors. They were then filtered by grouping descriptors with a Pearson correlation coefficient of over 0.95 to prevent overfitting and remove highly correlated features. Since 3D descriptors were being utilized, Corina was employed to optimize the structures of the molecules. Multiple models were then constructed consisting of Associative Neural Network (ASNN), Multiple Linear Regression (MLR), K-Nearest Neighbor (KNN), and Random Forest Regression (RFR), with 10-fold cross validation. These models were retrained 4 times and the one with the highest combination of accuracy, balanced accuracy, and area under the curve (AUC) was then selected to predict the activity of each of the developed hit compounds.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eApplication of QSAR Model to Predict Activity of Hit Compounds\u003c/h2\u003e \u003cp\u003eThe QSAR model with the highest combination of accuracy, balanced accuracy, and area under the receiver operating curve (AUC) was then selected to predict the activity of the hit compounds. The compounds were then downloaded from the ZINC chemical library and the SMILES formats were uploaded into the OCHEM database in which the model was then selected to predict the activity of each of the molecules (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e). This process was repeated several times to assure that the accuracy of the QSAR prediction was maintained.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eACKNOWLEDGEMENTS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe computational resources which enabled the virtual screening conducted in this study were kindly provided by Dr. Frank Liu from the Computer Science Department at Sam Houston State University. We greatly appreciate his technical assistance for our study. We would also like to greatly express our appreciation for Mr. Tyler Daniel and Dr. Qiuyue Nie from the Department of Chemical and Biomolecular Engineering at Rice University for their guidance in revisions and assistance with figure creation and nomenclature. We especially thank Dr. Qiuyue Nie for her guidance in compound descriptions. We would also like to thank Mr. Caleb Chang from the Department of Biosciences at Rice University for his mentorship and guidance with regards to protein target model development and drug discovery provided throughout the study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAUTHOR CONTRIBUTIONS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eK.G. conducted in silico drug discovery, including virtual screening and physicochemical filtering. K.G. wrote all sections of the manuscript and prepared Figures 1-5 and Table 1. K.G. also conducted all revisions of the manuscript. P.M. conducted homology modeling and QSAR modeling and wrote the sections of the manuscript related to those aspects. P.M. also prepared Tables 2 and 3.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCOMPETING INTERESTS\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDATA AVAILABILITY STATEMENT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo datasets were generated or analyzed during the current study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eFerlay J, Colombet M, Soerjomataram I, Parkin DM, Pi\u0026ntilde;eros M, Znaor A, et al. Cancer statistics for the year 2020: An overview. Int J Cancer. 2021. Epub 20210405. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/ijc.33588\u003c/span\u003e\u003cspan address=\"10.1002/ijc.33588\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 33818764.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLang L, Loveless R, Teng Y. Emerging Links between Control of Mitochondrial Protein ATAD3A and Cancer. Int J Mol Sci. 2020;21(21). Epub 20201025. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ijms21217917\u003c/span\u003e\u003cspan address=\"10.3390/ijms21217917\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 33113782; PubMed Central PMCID: PMC7663417.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeng Y, Lang L, Shay C. ATAD3A on the Path to Cancer. Adv Exp Med Biol. 2019;1134:259 \u0026ndash; 69. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/978-3-030-12668-1_14\u003c/span\u003e\u003cspan address=\"10.1007/978-3-030-12668-1_14\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 30919342.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTeng Y, Pi W, Wang Y, Cowell JK. WASF3 provides the conduit to facilitate invasion and metastasis in breast cancer cells through HER2/HER3 signaling. Oncogene. 2016;35(35):4633\u0026ndash;40. Epub 20160125. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/onc.2015.527\u003c/span\u003e\u003cspan address=\"10.1038/onc.2015.527\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26804171; PubMed Central PMCID: PMC4959990.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang KH, Chow KC, Chang HW, Lin TY, Lee MC. ATPase family AAA domain containing 3A is an anti-apoptotic factor and a secretion regulator of PSA in prostate cancer. Int J Mol Med. 2011;28(1):9\u0026ndash;15. Epub 20110406. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3892/ijmm.2011.670\u003c/span\u003e\u003cspan address=\"10.3892/ijmm.2011.670\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 21584487.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLang L, Teng Y. Using Genome-Editing Tools to Develop a Novel In Situ Coincidence Reporter Assay for Screening ATAD3A Transcriptional Inhibitors. Methods Mol Biol. 2020;2138:159\u0026ndash;66. doi: 10.1007/978-1-0716-0471-7_8. PubMed PMID: 32219745; PubMed Central PMCID: PMC8552142.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeralta S, Goffart S, Williams SL, Diaz F, Garcia S, Nissanka N, et al. ATAD3 controls mitochondrial cristae structure in mouse muscle, influencing mtDNA replication and cholesterol levels. J Cell Sci. 2018;131(13). Epub 20180704. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1242/jcs.217075\u003c/span\u003e\u003cspan address=\"10.1242/jcs.217075\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 29898916; PubMed Central PMCID: PMC6051345.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaudier J. ATAD3 proteins: brokers of a mitochondria-endoplasmic reticulum connection in mammalian cells. Biol Rev Camb Philos Soc. 2018;93(2):827\u0026ndash;44. Epub 20170920. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/brv.12373\u003c/span\u003e\u003cspan address=\"10.1111/brv.12373\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 28941010.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGilquin B, Cannon BR, Hubstenberger A, Moulouel B, Falk E, Merle N, et al. The calcium-dependent interaction between S100B and the mitochondrial AAA ATPase ATAD3A and the role of this complex in the cytoplasmic processing of ATAD3A. Mol Cell Biol. 2010;30(11):2724\u0026ndash;36. Epub 20100329. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1128/mcb.01468-09\u003c/span\u003e\u003cspan address=\"10.1128/mcb.01468-09\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 20351179; PubMed Central PMCID: PMC2876520.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao Y, Sun X, Hu D, Prosdocimo DA, Hoppel C, Jain MK, et al. ATAD3A oligomerization causes neurodegeneration by coupling mitochondrial fragmentation and bioenergetics defects. Nat Commun. 2019;10(1):1371. Epub 20190326. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-019-09291-x\u003c/span\u003e\u003cspan address=\"10.1038/s41467-019-09291-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 30914652; PubMed Central PMCID: PMC6435701.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArguello T, Peralta S, Antonicka H, Gaidosh G, Diaz F, Tu YT, et al. ATAD3A has a scaffolding role regulating mitochondria inner membrane structure and protein assembly. Cell Rep. 2021;37(12):110139. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.celrep.2021.110139\u003c/span\u003e\u003cspan address=\"10.1016/j.celrep.2021.110139\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 34936866; PubMed Central PMCID: PMC8785211.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi S, Cl\u0026eacute;men\u0026ccedil;on B, Catty P, Brandolin G, Schlattner U, Rousseau D. Yeast-based production and purification of HIS-tagged human ATAD3A, A specific target of S100B. Protein Expr Purif. 2012;83(2):211\u0026ndash;6. Epub 20120421. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.pep.2012.04.005\u003c/span\u003e\u003cspan address=\"10.1016/j.pep.2012.04.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 22542587.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRamachandran GN, Ramakrishnan C, Sasisekharan V. Stereochemistry of polypeptide chain configurations. J Mol Biol. 1963;7:95\u0026ndash;9. doi: 10.1016/s0022-2836(63)80023-6. PubMed PMID: 13990617.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol. 1993;234(3):779\u0026ndash;815. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1006/jmbi.1993.1626\u003c/span\u003e\u003cspan address=\"10.1006/jmbi.1993.1626\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 8254673.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeig M, Mirjalili V. Protein structure refinement via molecular-dynamics simulations: What works and what does not? Proteins. 2016;84 Suppl 1(Suppl 1):282 \u0026ndash; 92. Epub 20150817. doi: 10.1002/prot.24871. PubMed PMID: 26234208; PubMed Central PMCID: PMC5493977.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTrott O, Olson AJ. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem. 2010;31(2):455\u0026ndash;61. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jcc.21334\u003c/span\u003e\u003cspan address=\"10.1002/jcc.21334\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 19499576; PubMed Central PMCID: PMC3041641.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIrwin JJ, Tang KG, Young J, Dandarchuluun C, Wong BR, Khurelbaatar M, et al. ZINC20-A Free Ultralarge-Scale Chemical Database for Ligand Discovery. J Chem Inf Model. 2020;60(12):6065\u0026ndash;73. Epub 20201029. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.jcim.0c00675\u003c/span\u003e\u003cspan address=\"10.1021/acs.jcim.0c00675\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 33118813; PubMed Central PMCID: PMC8284596.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRDKit: Open-source cheminformatics; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.rdkit.org\u003c/span\u003e\u003cspan address=\"http://www.rdkit.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan Y, Zhang D, Zhou P, Li B, Huang SY. HDOCK: a web server for protein-protein and protein-DNA/RNA docking based on a hybrid strategy. Nucleic Acids Res. 2017;45(W1):W365-w73. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gkx407\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkx407\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 28521030; PubMed Central PMCID: PMC5793843.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUniProt: the Universal Protein Knowledgebase in 2023. Nucleic Acids Res. 2023;51(D1):D523-d31. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nar/gkac1052\u003c/span\u003e\u003cspan address=\"10.1093/nar/gkac1052\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 36408920; PubMed Central PMCID: PMC9825514.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJafari E, Khajouei MR, Hassanzadeh F, Hakimelahi GH, Khodarahmi GA. Quinazolinone and quinazoline derivatives: recent structures with potent antimicrobial and cytotoxic activities. Res Pharm Sci. 2016;11(1):1\u0026ndash;14. PubMed PMID: 27051427; PubMed Central PMCID: PMC4794932.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAltschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403\u0026ndash;10. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s0022-2836(05)80360-2\u003c/span\u003e\u003cspan address=\"10.1016/s0022-2836(05)80360-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 2231712.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, et al. Highly accurate protein structure prediction with AlphaFold. Nature. 2021;596(7873):583\u0026ndash;9. Epub 20210715. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41586-021-03819-2\u003c/span\u003e\u003cspan address=\"10.1038/s41586-021-03819-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 34265844; PubMed Central PMCID: PMC8371605.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEberhardt J, Santos-Martins D, Tillack AF, Forli S. AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model. 2021;61(8):3891\u0026ndash;8. Epub 20210719. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/acs.jcim.1c00203\u003c/span\u003e\u003cspan address=\"10.1021/acs.jcim.1c00203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 34278794.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThe PyMOL Molecular Graphics System, Version 2.5, Schr\u0026ouml;dinger, LLC.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJo S, Kim T, Iyer VG, Im W. CHARMM-GUI: a web-based graphical user interface for CHARMM. J Comput Chem. 2008;29(11):1859\u0026ndash;65. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jcc.20945\u003c/span\u003e\u003cspan address=\"10.1002/jcc.20945\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 18351591.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJo S, Jiang W, Lee HS, Roux B, Im W. CHARMM-GUI Ligand Binder for absolute binding free energy calculations and its application. J Chem Inf Model. 2013;53(1):267\u0026ndash;77. Epub 20121220. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1021/ci300505n\u003c/span\u003e\u003cspan address=\"10.1021/ci300505n\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 23205773; PubMed Central PMCID: PMC3557591.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSushko I, Novotarskyi S, K\u0026ouml;rner R, Pandey AK, Rupp M, Teetz W, et al. Online chemical modeling environment (OCHEM): web platform for data storage, model development and publishing of chemical information. J Comput Aided Mol Des. 2011;25(6):533\u0026ndash;54. Epub 20110610. doi: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10822-011-9440-2\u003c/span\u003e\u003cspan address=\"10.1007/s10822-011-9440-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 21660515; PubMed Central PMCID: PMC3131510.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Rossum G, Drake Jr. FL. Python reference manual. Centrum voor Wiskunde en Informatica Amsterdam; 1995.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Drug Development, Virtual Screening, Molecular Docking, Homology Modeling, Quantitative Structure-Activity Relationship Modeling","lastPublishedDoi":"10.21203/rs.3.rs-3280889/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3280889/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe overexpression of ATAD3A, a mitochondrial membrane oncoprotein, is correlated with worsened prognosis of many prevalent cancers and has been identified as an attractive target for drug development. This work investigates drug development for ATAD3A through a site-specific, interaction-based, in-silico framework bypassing conventional drug development obstacles, notably the inability to experimentally resolve the ATAD3A structure. In our approach, we target the critical ATAD3A-S100B binding interaction, which facilitates the cytoplasmic processing of ATAD3A. Relying on the canonicality of the S100B binding mechanism of ATAD3A, and the ATAD3A C-terminal sequence being highly conserved and homologous to existing structures, we reduce the necessity for an accurate model from the whole target protein to a single, well-established domain. In our in-silico framework, a model of the ATAD3A S100B binding domain was constructed, followed by drug discovery targeting the S100B binding domain utilizing virtual screening and hit identification. QSAR modeling and physicochemical filters were then subsequently employed to assess the hit compounds. Further analyzing the specific binding contacts between the hit compounds and ATAD3A compared to ATAD3A-S100B binding contacts from protein-protein docking, we were able to determine that two hit compounds, ZINC6235062 and ZINC20728831, strongly occupy the critical residues established in previous literature as necessary for S100B binding, and thus indicate great potential in inhibiting ATAD3A oncoprotein function through disrupting the ATAD3A-S100B binding interaction in competitive inhibition.\u003c/p\u003e","manuscriptTitle":"In-Silico Drug Discovery Through Interaction-Based Perspectives for ATAD3A","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-08-23 14:48:03","doi":"10.21203/rs.3.rs-3280889/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"440d9c7e-66bc-427e-9271-f82630f71648","owner":[],"postedDate":"August 23rd, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":24200257,"name":"Biological sciences/Biochemistry"},{"id":24200258,"name":"Biological sciences/Cancer"},{"id":24200259,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":24200260,"name":"Biological sciences/Drug discovery"}],"tags":[],"updatedAt":"2023-11-21T10:29:30+00:00","versionOfRecord":[],"versionCreatedAt":"2023-08-23 14:48:03","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3280889","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3280889","identity":"rs-3280889","version":["v1"]},"buildId":"FbvkV6FR0MCFSLy54lSbu","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}