ProteoMutaMetrics: machine learning approaches for solute carrier family 6 mutation pathogenicity prediction.

The solute carrier transporter family 6 (SLC6) is of key interest for their critical role in the transport of small amino acids or amino acid-like molecules. Their dysfunction is strongly associated with human diseases such as including schizophrenia, depression, and Parkinson's disease. Linking single point mutations to disease may support insights into the structure-function relationship of these transporters. This work aimed to develop a computational model for predicting the potential pathogenic effect of single point mutations in the SLC6 family. Missense mutation data was retrieved from UniProt, LitVar, and ClinVar, covering multiple protein-coding transcripts. As encoding approach, amino acid descriptors were used to calculate the average sequence properties for both original and mutated sequences. In addition to the full-sequence calculation, the sequences were cut into twelve domains. The domains are defined according to the transmembrane domains of the SLC6 transporters to analyse the regions' contributions to the pathogenicity prediction. Subsequently, several classification models, namely Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XGBoost) with the hyperparameters optimized through grid search were built. For estimation of model performance, repeated stratified k-fold cross-validation was used. The accuracy values of the generated models are in the range of 0.72 to 0.80. Analysis of feature importance indicates that mutations in distinct regions of SLC6 transporters are associated with an increased risk for pathogenicity. When applying the model on an independent validation set, the performance in accuracy dropped to averagely 0.6 with high precision but low sensitivity scores.

The macrocycle inhibitor landscape of SLC-transporter.

In the past years the interest in Solute Carrier Transporters (SLC) has increased due to their potential as drug targets. At the same time, macrocycles demonstrated promising activities as therapeutic agents. However, the overall macrocycle/SLC-transporter interaction landscape has not been fully revealed yet. In this study, we present a statistical analysis of macrocycles with measured activity against SLC-transporter. Using a data mining pipeline based on KNIME retrieved in total 825 bioactivity data points of macrocycles interacting with SLC-transporter. For further analysis of the SLC inhibitor profiles we developed an interactive KNIME workflow as well as an interactive map of the chemical space coverage utilizing parametric t-SNE models. The parametric t-SNE models provide a good discrimination ability among several corresponding SLC subfamilies' targets. The KNIME workflow, the dataset, and the visualization tool are freely available to the community.

Experimental and Computational Analysis of Newly Identified Pathogenic Mutations in the Creatine Transporter SLC6A8.

Creatine is an essential metabolite for the storage and rapid supply of energy in muscle and nerve cells. In humans, impaired metabolism, transport, and distribution of creatine throughout tissues can cause varying forms of mental disability, also known as creatine deficiency syndrome (CDS). So far, 80 mutations in the creatine transporter (SLC6A8) have been associated to CDS. To better understand the effect of human genetic variants on the physiology of SLC6A8 and their possible impact on CDS, we studied 30 missense variants including 15 variants of unknown significance, two of which are reported here for the first time. We expressed these variants in HEK293 cells and explored their subcellular localization and transport activity. We also applied computational methods to predict variant effect and estimate site-specific changes in thermodynamic stability. To explore variants that might have a differential effect on the transporter's conformers along the transport cycle, we constructed homology models of the inward facing, and outward facing conformations. In addition, we used mass-spectrometry to study proteins that interact with wild type SLC6A8 and five selected variants in HEK293 cells. In silico models of the protein complexes revealed how two variants impact the interaction interface of SLC6A8 with other proteins and how pathogenic variants lead to an enrichment of ER protein partners. Overall, our integrated analysis disambiguates the pathogenicity of 15 variants of unknown significance revealing diverse mechanisms of pathogenicity, including two previously unreported variants obtained from patients suffering from the creatine deficiency syndrome.

Privacy-preserving techniques for decentralized and secure machine learning in drug discovery.

Data availability, data security, and privacy concerns often hamper optimal performance efficiency of machine learning (ML) techniques. Therefore, novel techniques for the utilization of private/sensitive data in the field of drug discovery have been proposed for ML model-building tasks. Some examples of the different techniques are secure multiparty computation, distributed deep learning, homomorphic encryption, blockchain-based peer-to-peer networking, differential privacy, and federated learning, as well as combinations of such techniques. In this paper, we present an overview of these techniques for decentralized ML to illustrate its benefits and drawbacks in the field of drug discovery.

Structural Insights into Neonicotinoids and N-Unsubstituted Metabolites on Human nAChRs by Molecular Docking, Dynamics Simulations, and Calcium Imaging.

Neonicotinoid pesticides were initially designed in order to achieve species selectivity on insect nicotinic acetylcholine receptors (nAChRs). However, concerns arose when agonistic effects were also detected in human cells expressing nAChRs. In the context of next-generation risk assessments (NGRAs), new approach methods (NAMs) should replace animal testing where appropriate. Herein, we present a combination of in silico and in vitro methodologies that are used to investigate the potentially toxic effects of neonicotinoids and nicotinoid metabolites on human neurons. First, an ensemble docking study was conducted on the nAChR isoforms α7 and α3ß4 to assess potential crucial molecular initiating event (MIE) interactions. Representative docking poses were further refined using molecular dynamics (MD) simulations and binding energy calculations using implicit solvent models. Finally, calcium imaging on LUHMES neurons confirmed a key event (KE) downstream of the MIE. This method was also used to confirm the predicted agonistic effect of the metabolite descyano-thiacloprid (DCNT).

Paralog-dependent isogenic cell assay cascade generates highly selective SLC16A3 inhibitors.

Despite being considered druggable and attractive therapeutic targets, most of the solute carrier (SLC) membrane transporters remain pharmacologically underexploited. One of the reasons for this is a lack of reliable chemical screening assays, made difficult by functional redundancies among SLCs. In this study we leveraged synthetic lethality between the lactate transporters SLC16A1 and SLC16A3 in a screening strategy that we call paralog-dependent isogenic cell assay (PARADISO). The system involves five isogenic cell lines, each dependent on various paralog genes for survival/fitness, arranged in a screening cascade tuned for the identification of SLC16A3 inhibitors. We screened a diversity-oriented library of ∼90,000 compounds and further developed our hits into slCeMM1, a paralog-selective and potent SLC16A3 inhibitor. By implementing chemoproteomics, we showed that slCeMM1 is selective also at the proteome-wide level, thus fulfilling an important criterion for chemical probes. This study represents a framework for the development of specific cell-based drug discovery assays.

Identifying Differences in the Performance of Machine Learning Models for Off-Targets Trained on Publicly Available and Proprietary Data Sets.

Each year, publicly available databases are updated with new compounds from different research institutions. Positive experimental outcomes are more likely to be reported; therefore, they account for a considerable fraction of these entries. Established publicly available databases such as ChEMBL allow researchers to use information without constrictions and create predictive tools for a broad spectrum of applications in the field of toxicology. Therefore, we investigated the distribution of positive and nonpositive entries within ChEMBL for a set of off-targets and its impact on the performance of classification models when applied to pharmaceutical industry data sets. Results indicate that models trained on publicly available data tend to overpredict positives, and models based on industry data sets predict negatives more often than those built using publicly available data sets. This is strengthened even further by the visualization of the prediction space for a set of 10,000 compounds, which makes it possible to identify regions in the chemical space where predictions converge. Finally, we highlight the utilization of these models for consensus modeling for potential adverse events prediction.

Ligand- and Structure-based Approaches for Transmembrane Transporter Modeling.

The study of transporter proteins is key to understanding the mechanism behind multi-drug resistance and drug-drug interactions causing severe side effects. While ATP-binding transporters are well-studied, solute carriers illustrate an understudied family with a high number of orphan proteins. To study these transporters, in silico methods can be used to shed light on the basic molecular machinery by studying protein-ligand interactions. Nowadays, computational methods are an integral part of the drug discovery and development process. In this short review, computational approaches, such as machine learning, are discussed, which try to tackle interactions between transport proteins and certain compounds to locate target proteins. Furthermore, a few cases of selected members of the ATP binding transporter and solute carrier family are covered, which are of high interest in clinical drug interaction studies, especially for regulatory agencies. The strengths and limitations of ligand-based and structure-based methods are discussed to highlight their applicability for different studies. Furthermore, the combination of multiple approaches can improve the information obtained to find crucial amino acids that explain important interactions of protein-ligand complexes in more detail. This allows the design of drug candidates with increased activity towards a target protein, which further helps to support future synthetic efforts.

Systematic screening identifies ABCG2 as critical factor underlying synergy of kinase inhibitors with transcriptional CDK inhibitors.

BACKGROUND: Triple-negative breast cancer (TNBC) is a subtype of breast cancer with limited treatment options and poor clinical prognosis. Inhibitors of transcriptional CDKs are currently under thorough investigation for application in the treatment of multiple cancer types, including breast cancer. These studies have raised interest in combining these inhibitors, including CDK12/13 inhibitor THZ531, with a variety of other anti-cancer agents. However, the full scope of these potential synergistic interactions of transcriptional CDK inhibitors with kinase inhibitors has not been systematically investigated. Moreover, the mechanisms behind these previously described synergistic interactions remain largely elusive. METHODS: Kinase inhibitor combination screenings were performed to identify kinase inhibitors that synergize with CDK7 inhibitor THZ1 and CDK12/13 inhibitor THZ531 in TNBC cell lines. CRISPR-Cas9 knockout screening and transcriptomic evaluation of resistant versus sensitive cell lines were performed to identify genes critical for THZ531 resistance. RNA sequencing analysis after treatment with individual and combined synergistic treatments was performed to gain further insights into the mechanism of this synergy. Kinase inhibitor screening in combination with visualization of ABCG2-substrate pheophorbide A was used to identify kinase inhibitors that inhibit ABCG2. Multiple transcriptional CDK inhibitors were evaluated to extend the significance of the found mechanism to other transcriptional CDK inhibitors. RESULTS: We show that a very high number of tyrosine kinase inhibitors synergize with the CDK12/13 inhibitor THZ531. Yet, we identified the multidrug transporter ABCG2 as key determinant of THZ531 resistance in TNBC cells. Mechanistically, we demonstrate that most synergistic kinase inhibitors block ABCG2 function, thereby sensitizing cells to transcriptional CDK inhibitors, including THZ531. Accordingly, these kinase inhibitors potentiate the effects of THZ531, disrupting gene expression and increasing intronic polyadenylation. CONCLUSION: Overall, this study demonstrates the critical role of ABCG2 in limiting the efficacy of transcriptional CDK inhibitors and identifies multiple kinase inhibitors that disrupt ABCG2 transporter function and thereby synergize with these CDK inhibitors. These findings therefore further facilitate the development of new (combination) therapies targeting transcriptional CDKs and highlight the importance of evaluating the role of ABC transporters in synergistic drug-drug interactions in general.

Toxicity prediction using target, interactome, and pathway profiles as descriptors.

In silico methods are essential to the safety evaluation of chemicals. Computational risk assessment offers several approaches, with data science and knowledge-based methods becoming an increasingly important sub-group. One of the substantial attributes of data science is that it allows using existing data to find correlations, build strong hypotheses, and create new, valuable knowledge that may help to reduce the number of resource intensive experiments. In choosing a suitable method for toxicity prediction, the available data and desired toxicity endpoint are two essential factors to consider. The complexity of the endpoint can impact the success rate of the in silico models. For highly complex endpoints such as hepatotoxicity, it can be beneficial to decipher the toxic event from a more systemic point of view. We propose a data science-based modelling pipeline that uses compounds` connections to tissue-specific biological targets, interactome, and biological pathways as descriptors of compounds. Models trained on different combinations of the collected, compound-target, compound-interactor, and compound-pathway profiles, were used to predict the hepatotoxicity of drug-like compounds. Several tree-based models were trained, utilizing separate and combined target, interactome and pathway level variables. The model using combined descriptors of all levels and the random forest algorithm was further optimized. Descriptor importance for model performance was addressed and examined for a biological explanation to define which targets or pathways can have a crucial role in toxicity. Descriptors connected to cytochromes P450 enzymes, heme degradation and biological oxidation received high weights. Furthermore, the involvement of other, less discussed processes in connection with toxicity, such as the involvement of RHO GTPase effectors in hepatotoxicity, were marked as fundamental. The optimized combined model using only the selected descriptors yielded the best performance with an accuracy of 0.766. The same dataset using classical Morgan fingerprints for compound representation yielded models with similar performance measures, as well as the combination of systems biology-based descriptors and Morgan fingerprints. Consequently, adding the structural information of compounds did not enhance the predictive value of the models. The developed systems biology-based pipeline comprises a valuable tool in predicting toxicity, while providing novel insights about the possible mechanisms of the unwanted events.

A Structure-Based View on ABC-Transporter Linked to Multidrug Resistance.

The discovery of the first ATP-binding cassette (ABC) transporter, whose overexpression in cancer cells is responsible for exporting anticancer drugs out of tumor cells, initiated enormous efforts to overcome tumor cell multidrug resistance (MDR) by inhibition of ABC-transporter. Because of its many physiological functions, diverse studies have been conducted on the mechanism, function and regulation of this important group of transmembrane transport proteins. In this review, we will focus on the structural aspects of this transporter superfamily. Since the resolution revolution of electron microscope, experimentally solved structures increased rapidly. A summary of the structures available and an overview of recent structure-based studies are provided. More specifically, the artificial intelligence (AI)-based predictions from AlphaFold-2 will be discussed.

POLθ processes ssDNA gaps and promotes replication fork progression in BRCA1-deficient cells.

Polymerase theta (POLθ) is an error-prone DNA polymerase whose loss is synthetically lethal in cancer cells bearing breast cancer susceptibility proteins 1 and 2 (BRCA1/2) mutations. To investigate the basis of this genetic interaction, we utilized a small-molecule inhibitor targeting the POLθ polymerase domain. We found that POLθ processes single-stranded DNA (ssDNA) gaps that emerge in the absence of BRCA1, thus promoting unperturbed replication fork progression and survival of BRCA1 mutant cells. A genome-scale CRISPR-Cas9 knockout screen uncovered suppressors of the functional interaction between POLθ and BRCA1, including NBN, a component of the MRN complex, and cell-cycle regulators such as CDK6. While the MRN complex nucleolytically processes ssDNA gaps, CDK6 promotes cell-cycle progression, thereby exacerbating replication stress, a feature of BRCA1-deficient cells that lack POLθ activity. Thus, ssDNA gap formation, modulated by cell-cycle regulators and MRN complex activity, underlies the synthetic lethality between POLθ and BRCA1, an important insight for clinical trials with POLθ inhibitors.

Using Jupyter Notebooks for re-training machine learning models.

Machine learning (ML) models require an extensive, user-driven selection of molecular descriptors in order to learn from chemical structures to predict actives and inactives with a high reliability. In addition, privacy concerns often restrict the access to sufficient data, leading to models with a narrow chemical space. Therefore, we propose a framework of re-trainable models that can be transferred from one local instance to another, and further allow a less extensive descriptor selection. The models are shared via a Jupyter Notebook, allowing the evaluation and implementation of a broader chemical space by keeping most of the tunable parameters pre-defined. This enables the models to be updated in a decentralized, facile, and fast manner. Herein, the method was evaluated with six transporter datasets (BCRP, BSEP, OATP1B1, OATP1B3, MRP3, P-gp), which revealed the general applicability of this approach.

CACHE (Critical Assessment of Computational Hit-finding Experiments): A public-private partnership benchmarking initiative to enable the development of computational methods for hit-finding.

One aspirational goal of computational chemistry is to predict potent and drug-like binders for any protein, such that only those that bind are synthesized. In this Roadmap, we describe the launch of Critical Assessment of Computational Hit-finding Experiments (CACHE), a public benchmarking project to compare and improve small molecule hit-finding algorithms through cycles of prediction and experimental testing. Participants will predict small molecule binders for new and biologically relevant protein targets representing different prediction scenarios. Predicted compounds will be tested rigorously in an experimental hub, and all predicted binders as well as all experimental screening data, including the chemical structures of experimentally tested compounds, will be made publicly available, and not subject to any intellectual property restrictions. The ability of a range of computational approaches to find novel binders will be evaluated, compared, and openly published. CACHE will launch 3 new benchmarking exercises every year. The outcomes will be better prediction methods, new small molecule binders for target proteins of importance for fundamental biology or drug discovery, and a major technological step towards achieving the goal of Target 2035, a global initiative to identify pharmacological probes for all human proteins.

KNIME workflow for retrieving causal drug and protein interactions, building networks, and performing topological enrichment analysis demonstrated by a DILI case study.

As an alternative to one drug-one target approaches, systems biology methods can provide a deeper insight into the holistic effects of drugs. Network-based approaches are tools of systems biology, that can represent valuable methods for visualizing and analysing drug-protein and protein-protein interactions. In this study, a KNIME workflow is presented which connects drugs to causal target proteins and target proteins to their causal protein interactors. With the collected data, networks can be constructed for visualizing and interpreting the connections. The last part of the workflow provides a topological enrichment test for identifying relevant pathways and processes connected to the submitted data. The workflow is based on openly available databases and their web services. As a case study, compounds of DILIRank were analysed. DILIRank is the benchmark dataset for Drug-Induced Liver Injury by the FDA, where compounds are categorized by their likeliness of causing DILI. The study includes the drugs that are most likely to cause DILI ("mostDILI") and the ones that are not likely to cause DILI ("noDILI"). After selecting the compounds of interest, down- and upregulated proteins connected to the mostDILI group were identified; furthermore, a liver-specific subset of those was created. The downregulated sub-list had considerably more entries, therefore, network and causal interactome were constructed and topological pathway enrichment analysis was performed with this list. The workflow identified proteins such as Prostaglandin G7H synthase 1 and UDP-glucuronosyltransferase 1A9 as key participants in the potential toxic events disclosing the possible mode of action. The topological network analysis resulted in pathways such as recycling of bile acids and salts and glucuronidation, indicating their involvement in DILI. The KNIME pipeline was built to support target and network-based approaches to analyse any sets of drug data and identify their target proteins, mode of actions and processes they are involved in. The fragments of the pipeline can be used separately or can be combined as required.

Off-targetP ML: an open source machine learning framework for off-target panel safety assessment of small molecules.

Unpredicted drug safety issues constitute the majority of failures in the pharmaceutical industry according to several studies. Some of these preclinical safety issues could be attributed to the non-selective binding of compounds to targets other than their intended therapeutic target, causing undesired adverse events. Consequently, pharmaceutical companies routinely run in-vitro safety screens to detect off-target activities prior to preclinical and clinical studies. Hereby we present an open source machine learning framework aiming at the prediction of our in-house 50 off-target panel activities for ~ 4000 compounds, directly from their structure. This framework is intended to guide chemists in the drug design process prior to synthesis and to accelerate drug discovery. We also present a set of ML approaches that require minimum programming experience for deployment. The workflow incorporates different ML approaches such as deep learning and automated machine learning. It also accommodates popular issues faced in bioactivity predictions, as data imbalance, inter-target duplicated measurements and duplicated public compound identifiers. Throughout the workflow development, we explore and compare the capability of Neural Networks and AutoML in constructing prediction models for fifty off-targets of different protein classes, different dataset sizes, and high-class imbalance. Outcomes from different methods are compared in terms of efficiency and efficacy. The most important challenges and factors impacting model construction and performance in addition to suggestions on how to overcome such challenges are also discussed.

Structure-based peptide ligand design for improved epidermal growth factor receptor targeted gene delivery.

The epidermal growth factor receptor EGFR allows targeted delivery of macromolecular drugs to tumors. Its ligand, epidermal growth factor, binds EGFR with high affinity but acts mitogenic. Non-mitogenic peptides are utilized as targeting ligands, like the dodecapeptide GE11, although its low binding affinity warrants improvement. We applied a two-step computational approach with database search and molecular docking to design GE11 variants with improved binding. Synthesized peptides underwent binding studies on immobilized EGFR using surface plasmon resonance. Conjugates of peptides coupled via heterobifunctional PEG linker to linear polyethylenimine (LPEI) were used for transfection studies on EGFR-overexpressing cells using reporter gene encoding plasmid DNA. Docking studies unraveled similarities between GE11 and the EGFR dimerization arm. By skipping non-overlapping amino acids, a less hydrophobic segment (YTPQNVI) was identified to be directly involved in EGFR binding. By replacing valine by tyrosine, a full-length version with proposed enhanced binding (GE11m3) was developed. While hydrophobic or hydrophilic segments and variations thereof exhibited low binding, GE11m3 exhibited 3-fold increase in binding compared to GE11, validating in silico predictions. In transfection studies, polyplexes with GE11m3 induced a significantly higher reporter gene expression when compared to GE11 polyplexes both on murine and human cancer cells overexpressing EGFR.

Evaluation of the Success of High-Throughput Physiologically Based Pharmacokinetic (HT-PBPK) Modeling Predictions to Inform Early Drug Discovery.

Minimizing in vitro and in vivo testing in early drug discovery with the use of physiologically based pharmacokinetic (PBPK) modeling and machine learning (ML) approaches has the potential to reduce discovery cycle times and animal experimentation. However, the prediction success of such an approach has not been shown for a larger and diverse set of compounds representative of a lead optimization pipeline. In this study, the prediction success of the oral (PO) and intravenous (IV) pharmacokinetics (PK) parameters in rats was assessed using a "bottom-up" approach, combining in vitro and ML inputs with a PBPK model. More than 240 compounds for which all of the necessary inputs and PK data were available were used for this assessment. Different clearance scaling approaches were assessed, using hepatocyte intrinsic clearance and protein binding as inputs. In addition, a novel high-throughput PBPK (HT-PBPK) approach was evaluated to assess the scalability of PBPK predictions for a larger number of compounds in drug discovery. The results showed that bottom-up PBPK modeling was able to predict the rat IV and PO PK parameters for the majority of compounds within a 2- to 3-fold error range, using both direct scaling and dilution methods for clearance predictions. The use of only ML-predicted inputs from the structure did not perform well when using in vitro inputs, likely due to clearance miss predictions. The HT-PBPK approach produced comparable results to the full PBPK modeling approach but reduced the simulation time from hours to seconds. In conclusion, a bottom-up PBPK and HT-PBPK approach can successfully predict the PK parameters and guide early discovery by informing compound prioritization, provided that good in vitro assays are in place for key parameters such as clearance.