L858R/L718Q and L858R/L792H Mutations of EGFR Inducing Resistance Against Osimertinib by Forming Additional Hydrogen Bonds.

Acquired resistance to first-line treatments in various cancers both promotes cancer recurrence as well as limits effective treatment. This is true for epidermal growth factor receptor (EGFR) mutations, for which secondary EGFR mutations are one of the principal mechanisms conferring resistance to the covalent inhibitor osimertinib. Thus, it is very important to develop a deeper understanding of the secondary mutational resistance mechanisms associated with EGFR mutations arising in tumors treated with osimertinib to expedite the development of innovative therapeutic drugs to overcome acquired resistance. This work uses all-atom molecular dynamics (MD) simulations to investigate the conformational variation of two reported EGFR mutants (L858R/L718Q and L858R/L792H) that resist osimertinib. The wild-type EGFR kinase domain and the L858R mutant are used as the reference. Our MD simulation results revealed that both the L718Q and L792H secondary mutations induce additional hydrogen bonds between the residues in the active pocket and the residues with the water molecules. These additional hydrogen bonds reduce the exposure area of C797, the covalent binding target of osimertinib. The additional hydrogen bonds also influence the binding affinity of the EGFR kinase domain by altering the secondary structure and flexibility of the amino acid residues in the domain. Our work highlights how the two reported mutations may alter both residue-residue and residue-solvent hydrogen bonds, affecting protein binding properties, which could be helpful for future drug discovery.

Artesunate acts through cytochrome c to inhibit growth of pediatric AML cells.

Artesunate is a derivative of artemisinin, an active compound isolated from Artemisia annua which has been used in Traditional Chinese Medicine and to treat malaria worldwide. Artemisinin derivatives have exhibited anti-cancer activity against both solid tumors and leukemia. The direct target(s) of artesunate are controversial; although, heme-bound proteins in the mitochondria have been implicated. We utilized computational modeling to calculate the predicted binding score of artesunate with heme-bound mitochondrial proteins and identified cytochrome c as potential artesunate target. UV-visible spectroscopy showed changes in the absorbance spectrum, and thus protein structure, when cytochrome c was incubated with artesunate. Artesunate induces apoptosis, disrupts mitochondrial membrane potential, and is antagonized by methazolamide in pediatric AML cells indicating a probable mechanism of action involving cytochrome c. We utilized a multi-disciplinary approach to show that artesunate can interact with and is dependent on cytochrome c release to induce cell death in pediatric AML cell lines.

Crizotinib induces Par-4 secretion from normal cells and GRP78 expression on the cancer cell surface for selective tumor growth inhibition.

Lung cancer is the leading cause of cancer-related deaths. Lung cancer cells develop resistance to apoptosis by suppressing the secretion of the tumor suppressor Par-4 protein (also known as PAWR) and/or down-modulating the Par-4 receptor GRP78 on the cell surface (csGRP78). We sought to identify FDA-approved drugs that elevate csGRP78 on the surface of lung cancer cells and induce Par-4 secretion from the cancer cells and/or normal cells in order to inhibit cancer growth in an autocrine or paracrine manner. In an unbiased screen, we identified crizotinib (CZT), an inhibitor of activated ALK/MET/ROS1 receptor tyrosine kinase, as an inducer of csGRP78 expression in ALK-negative, KRAS or EGFR mutant lung cancer cells. Elevation of csGRP78 in the lung cancer cells was dependent on activation of the non-receptor tyrosine kinase SRC by CZT. Inhibition of SRC activation in the cancer cells prevented csGRP78 translocation but promoted Par-4 secretion by CZT, implying that activated SRC prevented Par-4 secretion. In normal cells, CZT did not activate SRC and csGRP78 elevation but induced Par-4 secretion. Consequently, CZT induced Par-4 secretion from normal cells and elevated csGRP78 in the ALK-negative tumor cells to cause paracrine apoptosis in cancer cell cultures and growth inhibition of tumor xenografts in mice. Thus, CZT induces differential activation of SRC in normal and cancer cells to trigger the pro-apoptotic Par-4-GRP78 axis. As csGRP78 is a targetable receptor, CZT can be repurposed to elevate csGRP78 for inhibition of ALK-negative lung tumors.

Open Data and transparency in artificial intelligence and machine learning: A new era of research.

Artificial Intelligence (AI) and machine learning are the current forefront of computer science and technology. AI and related sub-disciplines, including machine learning, are essential technologies which have enabled the widespread use of smart technology, such as smart phones, smart home appliances and even electric toothbrushes. It is AI that allows the devices used day-to-day across people's personal lives, working lives and in industry to better anticipate and respond to our needs. However, the use of AI technology comes with a range of ethical questions - including issues around privacy, security, reliability, copyright/plagiarism and whether AI is capable of independent, conscious thought. We have seen several issues related to racial and sexual bias in AI in the recent times, putting the reliability of AI in question. Many of these issues have been brought to the forefront of cultural awareness in late 2022, early 2023, with the rise of AI art programs (and the copyright issues arising from the deep-learning methods employed to train this AI), and the popularity of ChatGPT alongside its ability to be used to mimic human output, particularly in regard to academic work. In critical areas like healthcare, the errors of AI can be fatal. With the incorporation of AI in almost every sector of our everyday life, we need to keep asking ourselves- can we trust AI, and how much? This Editorial outlines the importance of openness and transparency in the development and applications of AI to allow all users to fully understand both the benefits and risks of this ubiquitous technology, and outlines how the Artificial Intelligence and Machine Learning  Gateway on F1000Research meets these needs.

The RelB-BLNK Axis Determines Cellular Response to a Novel Redox-Active Agent Betamethasone during Radiation Therapy in Prostate Cancer.

Aberrant levels of reactive oxygen species (ROS) are potential mechanisms that contribute to both cancer therapy efficacy and the side effects of cancer treatment. Upregulation of the non-canonical redox-sensitive NF-kB family member, RelB, confers radioresistance in prostate cancer (PCa). We screened FDA-approved compounds and identified betamethasone (BET) as a drug that increases hydrogen peroxide levels in vitro and protects non-PCa tissues/cells while also enhancing radiation killing of PCa tissues/cells, both in vitro and in vivo. Significantly, BET increases ROS levels and exerts different effects on RelB expression in normal cells and PCa cells. BET induces protein expression of RelB and RelB target genes, including the primary antioxidant enzyme, manganese superoxide dismutase (MnSOD), in normal cells, while it suppresses protein expression of RelB and MnSOD in LNCaP cells and PC3 cells. RNA sequencing analysis identifies B-cell linker protein (BLNK) as a novel RelB complementary partner that BET differentially regulates in normal cells and PCa cells. RelB and BLNK are upregulated and correlate with the aggressiveness of PCa in human samples. The RelB-BLNK axis translocates to the nuclear compartment to activate MnSOD protein expression. BET promotes the RelB-BLNK axis in normal cells but suppresses the RelB-BLNK axis in PCa cells. Targeted disruptions of RelB-BLNK expressions mitigate the radioprotective effect of BET on normal cells and the radiosensitizing effect of BET on PCa cells. Our study identified a novel RelB complementary partner and reveals a complex redox-mediated mechanism showing that the RelB-BLNK axis, at least in part, triggers differential responses to the redox-active agent BET by stimulating adaptive responses in normal cells but pushing PCa cells into oxidative stress overload.

KEAP1 Is Required for Artesunate Anticancer Activity in Non-Small-Cell Lung Cancer.

Artesunate is the most common treatment for malaria throughout the world. Artesunate has anticancer activity likely through the induction of reactive oxygen species, the same mechanism of action utilized in Plasmodium falciparum infections. Components of the kelch-like ECH-associated protein 1 (KEAP1)/nuclear factor erythroid 2-related factor 2 (NRF2) pathway, which regulates cellular response to oxidative stress, are mutated in approximately 30% of non-small-cell lung cancers (NSCLC); therefore, we tested the hypothesis that KEAP1 is required for artesunate sensitivity in NSCLC. Dose response assays identified A549 cells, which have a G333C-inactivating mutation in KEAP1, as resistant to artesunate, with an IC50 of 23.6 µM, while H1299 and H1563 cells were sensitive to artesunate, with a 10-fold lower IC50. Knockdown of KEAP1 through siRNA caused increased resistance to artesunate in H1299 cells. Alternatively, the pharmacological inhibition of NRF2, which is activated downstream of KEAP1 loss, by ML385 partially restored sensitivity of A549 cells to artesunate, and the combination of artesunate and ML385 was synergistic in both A549 and H1299 cells. These findings demonstrate that KEAP1 is required for the anticancer activity of artesunate and support the further development of NRF2 inhibitors to target patients with mutations in the KEAP1/NRF2 pathway.

Machine learning and ligand binding predictions: A review of data, methods, and obstacles.

Computational predictions of ligand binding is a difficult problem, with more accurate methods being extremely computationally expensive. The use of machine learning for drug binding predictions could possibly leverage the use of biomedical big data in exchange for time-intensive simulations. This paper reviews current trends in the use of machine learning for drug binding predictions, data sources to develop machine learning algorithms, and potential problems that may lead to overfitting and ungeneralizable models. A few popular datasets that can be used to develop virtual high-throughput screening models are characterized using spatial statistics to quantify potential biases. We can see from evaluating some common benchmarks that good performance correlates with models with high-predicted bias scores and models with low bias scores do not have much predictive power. A better understanding of the limits of available data sources and how to fix them will lead to more generalizable models that will lead to novel drug discovery.

Ensemble Docking in Drug Discovery: How Many Protein Configurations from Molecular Dynamics Simulations are Needed To Reproduce Known Ligand Binding?

Ensemble docking in drug discovery or chemical biology uses dynamical simulations of target proteins to generate binding site conformations for docking campaigns. We show that 600 ns molecular dynamics simulations of four G-protein-coupled receptors in their membrane environments generate ensembles of protein configurations that, collectively, are selected by 70?99% of the known ligands of these proteins. Therefore, the process of ligand recognition by conformational selection can be reproduced by combining molecular dynamics and docking calculations. Clustering of the molecular dynamics trajectories, however, does not necessarily identify the protein conformations that are most often selected by the ligands.

Characterization of Squamous Cell Lung Cancers from Appalachian Kentucky.

BACKGROUND: Lung cancer is the leading cause of cancer mortality in the United States (U.S.). Squamous cell carcinoma (SQCC) represents 22.6% of all lung cancers nationally, and 26.4% in Appalachian Kentucky (AppKY), where death from lung cancer is exceptionally high. The Cancer Genome Atlas (TCGA) characterized genetic alterations in lung SQCC, but this cohort did not focus on AppKY residents. METHODS: Whole-exome sequencing was performed on tumor and normal DNA samples from 51 lung SQCC subjects from AppKY. Somatic genomic alterations were compared between the AppKY and TCGA SQCC cohorts. RESULTS: From this AppKY cohort, we identified an average of 237 nonsilent mutations per patient and, in comparison with TCGA, we found that PCMTD1 (18%) and IDH1 (12%) were more commonly altered in AppKY versus TCGA. Using IDH1 as a starting point, we identified a mutually exclusive mutational pattern (IDH1, KDM6A, KDM4E, JMJD1C) involving functionally related genes. We also found actionable mutations (10%) and/or intermediate or high-tumor mutation burden (65%), indicating potential therapeutic targets in 65% of subjects. CONCLUSIONS: This study has identified an increased percentage of IDH1 and PCMTD1 mutations in SQCC arising in the AppKY residents versus TCGA, with population-specific implications for the personalized treatment of this disease. IMPACT: Our study is the first report to characterize genomic alterations in lung SQCC from AppKY. These findings suggest population differences in the genetics of lung SQCC between AppKY and U.S. populations, highlighting the importance of the relevant population when developing personalized treatment approaches for this disease.

Incorporating Protein Dynamics Through Ensemble Docking in Machine Learning Models to Predict Drug Binding.

Drug discovery is an expensive, lengthy, and sometimes dangerous process. The ability to make accurate computational predictions of drug binding would greatly improve the cost-effectiveness and safety of drug discovery and development. This study incorporates ensemble docking, the use of multiple protein conformations extracted from a molecular dynamics trajectory to perform docking calculations, with additional biomedical data sources and machine learning algorithms to improve the prediction of drug binding. We found that we can greatly increase the classification accuracy of an active vs a decoy compound using these methods over docking scores alone. The best results seen here come from having an individual protein conformation that produces binding features that correlate well with the active vs. decoy classification, in which case we achieve over 99% accuracy. The ability to confidently make accurate predictions on drug binding would allow for computational polypharamacological networks with insights into side-effect prediction, drug-repurposing, and drug efficacy.

Polypharmacology Within the Full Kinome: a Machine Learning Approach.

Protein kinases generate nearly a thousand different protein products and regulate the majority of cellular pathways and signal transduction. It is therefore not surprising that the deregulation of kinases has been implicated in many disease states. In fact, kinase inhibitors are the largest class of new cancer therapies. Understanding polypharmacology within the full kinome, how drugs interact with many different kinases, would allow for the development of safer and more efficacious cancer therapies. A full understanding of these interactions is not experimentally feasible making highly accurate computational predictions extremely useful and important. This work aims at making a machine learning model useful for investigating the full kinome. We evaluate many feature sets for our model and get better performance over molecular docking with all of them. We demonstrate that you can achieve a nearly 60% increase in success rate at identifying binding compounds using our model over molecular docking scores.

Gene-based association study of genes linked to hippocampal sclerosis of aging neuropathology: GRN, TMEM106B, ABCC9, and KCNMB2.

Hippocampal sclerosis of aging (HS-Aging) is a common neurodegenerative condition associated with dementia. To learn more about genetic risk of HS-Aging pathology, we tested gene-based associations of the GRN, TMEM106B, ABCC9, and KCNMB2 genes, which were reported to be associated with HS-Aging pathology in previous studies. Genetic data were obtained from the Alzheimer's Disease Genetics Consortium, linked to autopsy-derived neuropathological outcomes from the National Alzheimer's Coordinating Center. Of the 3251 subjects included in the study, 271 (8.3%) were identified as an HS-Aging case. The significant gene-based association between the ABCC9 gene and HS-Aging appeared to be driven by a region in which a significant haplotype-based association was found. We tested this haplotype as an expression quantitative trait locus using 2 different public-access brain gene expression databases. The HS-Aging pathology protective ABCC9 haplotype was associated with decreased ABCC9 expression, indicating a possible toxic gain of function.

Genetic Comparison of Symptomatic and Asymptomatic Persons With Alzheimer Disease Neuropathology.

OBJECTIVE: The objective was to determine whether symptomatic and asymptomatic persons with Alzheimer disease (AD) neuropathology have different allele counts for single-nucleotide polymorphisms that have been associated with clinical late-onset AD. METHODS: Data came from the National Alzheimer's Coordinating Center Uniform Data Set and Neuropathology Data Set, and the Alzheimer's Disease Genetics Consortium (ADGC). Participants had low to high AD neuropathologic change. The 22 known/suspected genes associated with late-onset AD were considered. "Symptomatic" was defined as Clinical Dementia Rating global score >0. RESULTS: Sixty-eight asymptomatic and 521 symptomatic participants met inclusion criteria. Single-nucleotide polymorphisms associated with ABCA7 [odds ratio (OR)=1.66; 95% confidence interval (CI), 1.03-2.85] and MAPT (OR=2.18; CI, 1.26-3.77) were associated with symptomatic status. In stratified analyses, loci containing CD2AP (OR=0.35; 95% CI, 0.16-0.74), ZCWPW1 (OR=2.98; 95% CI, 1.34-6.86), and MAPT (OR=3.73, 95% CI, 1.30-11.76) were associated with symptomatic status in APOE e4 carriers. CONCLUSIONS: These findings potentially explain some of the variation in whether a person with AD neuropathology expresses symptoms. Understanding why some people remain cognitively normal despite having AD neuropathology could identify pathways to disease heterogeneity and guide treatment trials.

Automated quality control for genome wide association studies.

This paper provides details on the necessary steps to assess and control data in genome wide association studies (GWAS) using genotype information on a large number of genetic markers for large number of individuals. Due to varied study designs and genotyping platforms between multiple sites/projects as well as potential genotyping errors, it is important to ensure high quality data. Scripts and directions are provided to facilitate others in this process.

Ensemble-based docking: From hit discovery to metabolism and toxicity predictions.

This paper describes and illustrates the use of ensemble-based docking, i.e., using a collection of protein structures in docking calculations for hit discovery, the exploration of biochemical pathways and toxicity prediction of drug candidates. We describe the computational engineering work necessary to enable large ensemble docking campaigns on supercomputers. We show examples where ensemble-based docking has significantly increased the number and the diversity of validated drug candidates. Finally, we illustrate how ensemble-based docking can be extended beyond hit discovery and toward providing a structural basis for the prediction of metabolism and off-target binding relevant to pre-clinical and clinical trials.

Multi-conformer ensemble docking to difficult protein targets.

Large-scale ensemble docking is investigated using five proteins from the Directory of Useful Decoys (DUD, dud.docking.org ) for which docking to crystal structures has proven difficult. Molecular dynamics trajectories are produced for each protein and an ensemble of representative conformational structures extracted from the trajectories. Docking calculations are performed on these selected simulation structures and ensemble-based enrichment factors compared with those obtained using docking in crystal structures of the same protein targets or random selection of compounds. Simulation-derived snapshots are found with improved enrichment factors that increase the chemical diversity of docking hits for four of the five selected proteins. A combination of all the docking results obtained from molecular dynamics simulation followed by selection of top-ranking compounds appears to be an effective strategy for increasing the number and diversity of hits when using docking to screen large libraries of chemicals against difficult protein targets.

Reassessment of risk genotypes (GRN, TMEM106B, and ABCC9 variants) associated with hippocampal sclerosis of aging pathology.

Hippocampal sclerosis of aging (HS-Aging) is a common high-morbidity neurodegenerative condition in elderly persons. To understand the risk factors for HS-Aging, we analyzed data from the Alzheimer's Disease Genetics Consortium and correlated the data with clinical and pathologic information from the National Alzheimer's Coordinating Center database. Overall, 268 research volunteers with HS-Aging and 2,957 controls were included; detailed neuropathologic data were available for all. The study focused on single-nucleotide polymorphisms previously associated with HS-Aging risk: rs5848 (GRN), rs1990622 (TMEM106B), and rs704180 (ABCC9). Analyses of a subsample that was not previously evaluated (51 HS-Aging cases and 561 controls) replicated the associations of previously identified HS-Aging risk alleles. To test for evidence of gene-gene interactions and genotype-phenotype relationships, pooled data were analyzed. The risk for HS-Aging diagnosis associated with these genetic polymorphisms was not secondary to an association with either Alzheimer disease or dementia with Lewy body neuropathologic changes. The presence of multiple risk genotypes was associated with a trend for additive risk for HS-Aging pathology. We conclude that multiple genes play important roles in HS-Aging, which is a distinctive neurodegenerative disease of aging.

Accelerating Virtual High-Throughput Ligand Docking: current technology and case study on a petascale supercomputer.

In this paper we give the current state of high-throughput virtual screening. We describe a case study of using a task-parallel MPI (Message Passing Interface) version of Autodock4 [1], [2] to run a virtual high-throughput screen of one-million compounds on the Jaguar Cray XK6 Supercomputer at Oak Ridge National Laboratory. We include a description of scripts developed to increase the efficiency of the predocking file preparation and postdocking analysis. A detailed tutorial, scripts, and source code for this MPI version of Autodock4 are available online at http://www.bio.utk.edu/baudrylab/autodockmpi.htm.

Detection of Bacillus anthracis DNA in complex soil and air samples using next-generation sequencing.

Bacillus anthracis is the potentially lethal etiologic agent of anthrax disease, and is a significant concern in the realm of biodefense. One of the cornerstones of an effective biodefense strategy is the ability to detect infectious agents with a high degree of sensitivity and specificity in the context of a complex sample background. The nature of the B. anthracis genome, however, renders specific detection difficult, due to close homology with B. cereus and B. thuringiensis. We therefore elected to determine the efficacy of next-generation sequencing analysis and microarrays for detection of B. anthracis in an environmental background. We applied next-generation sequencing to titrated genome copy numbers of B. anthracis in the presence of background nucleic acid extracted from aerosol and soil samples. We found next-generation sequencing to be capable of detecting as few as 10 genomic equivalents of B. anthracis DNA per nanogram of background nucleic acid. Detection was accomplished by mapping reads to either a defined subset of reference genomes or to the full GenBank database. Moreover, sequence data obtained from B. anthracis could be reliably distinguished from sequence data mapping to either B. cereus or B. thuringiensis. We also demonstrated the efficacy of a microbial census microarray in detecting B. anthracis in the same samples, representing a cost-effective and high-throughput approach, complementary to next-generation sequencing. Our results, in combination with the capacity of sequencing for providing insights into the genomic characteristics of complex and novel organisms, suggest that these platforms should be considered important components of a biosurveillance strategy.

VinaMPI: facilitating multiple receptor high-throughput virtual docking on high-performance computers.

The program VinaMPI has been developed to enable massively large virtual drug screens on leadership-class computing resources, using a large number of cores to decrease the time-to-completion of the screen. VinaMPI is a massively parallel Message Passing Interface (MPI) program based on the multithreaded virtual docking program AutodockVina, and is used to distribute tasks while multithreading is used to speed-up individual docking tasks. VinaMPI uses a distribution scheme in which tasks are evenly distributed to the workers based on the complexity of each task, as defined by the number of rotatable bonds in each chemical compound investigated. VinaMPI efficiently handles multiple proteins in a ligand screen, allowing for high-throughput inverse docking that presents new opportunities for improving the efficiency of the drug discovery pipeline. VinaMPI successfully ran on 84,672 cores with a continual decrease in job completion time with increasing core count. The ratio of the number of tasks in a screening to the number of workers should be at least around 100 in order to have a good load balance and an optimal job completion time. The code is freely available and downloadable. Instructions for downloading and using the code are provided in the Supporting Information.