FtsZ is an essential bacterial protein abundantly studied as a novel and promising target for antimicrobials. FtsZ is highly conserved among bacteria and mycobacteria, and it is crucial for the correct outcome of the cell division process, as it is responsible for the division of the parent bacterial cell into two daughter cells. In recent years, the benzodioxane–benzamide class has emerged as very promising and capable of targeting both Gram-positive and Gram-negative FtsZs. In this study, we explored the effect of including a substituent on the ethylenic linker between the two main moieties on the antimicrobial activity and pharmacokinetic properties. This substitution, in turn, led to the generation of a second stereogenic center, with both erythro and threo isomers isolated, characterized, and evaluated. With this work, we discovered how the hydroxy group slightly affects the antimicrobial activity, while being an important anchor for the exploitation and development of prodrugs, probes, and further derivatives.

The long, complex, and costly process of drug development makes early, reliable toxicity assessment a central challenge in pharmaceutical research. Accurately predicting drug‐induced toxicity and mutagenicity is crucial for guiding decision‐making and reducing late‐stage attrition. Deep learning approaches can predict the results of the Ames mutagenicity test, a standard screening method; however, most existing models rely on molecular descriptors that limit structural representation, lack interpretability, and do not account for Ames results of individual bacterial strains. Graph neural networks (GNNs) are chemically intuitive and overcome these limitations. This work presents a deep learning, multitask GNN model for predicting Ames mutagenicity. The proposed approach outperforms several existing models, including those from the Ames/QSAR International Challenge, on the dataset presented in this work. Furthermore, the model is highly interpretable. Explainability analysis of atom‐level contributions reveal that the model is capable of learning known structural alerts associated with mutagenicity, along with additional molecular patterns and 3D structural properties. Importantly, these insights are captured in a strain‐specific manner, highlighting the value of a multitask approach. Finally, the model is able to identify previously unreported molecular substructures potentially associated with mutagenicity, supporting its utility in drug design and development.

The long, complex, and costly process of drug development makes early, reliable toxicity assessment a central challenge in pharmaceutical research. Accurately predicting drug-induced toxicity and mutagenicity is crucial for guiding decision-making and reducing late-stage attrition. Deep learning approaches can predict the results of the Ames mutagenicity test, a standard screening method; however, most existing models rely on molecular descriptors that limit structural representation, lack interpretability, and do not account for Ames results of individual bacterial strains. Graph neural networks (GNNs) are chemically intuitive and overcome these limitations. This work presents a deep learning, multitask GNN model for predicting Ames mutagenicity. The proposed approach outperforms existing models, including those from the Ames/QSAR International Challenge. Furthermore, the model is highly interpretable. Explainability analysis of atom-level contributions reveal that the model is capable of learning known structural alerts associated with mutagenicity, along with additional molecular patterns and three-dimensional structural properties. Importantly, these insights are captured in a strain-specific manner, highlighting the value of a multitask approach. Finally, the model is able to identify previously unreported molecular substructures potentially associated with mutagenicity, supporting its utility in drug design and development.

(DYRK1A) plays a key role in various diseases, including DYRK1A syndrome, cancer, diabetes, and neurodegenerative disorders such as Alzheimer’s disease (AD), making it a compelling therapeutic target. This work integrates multiple Artificial Intelligence (AI) methods, including predictive models and generative algorithms, to design non-toxic DYRK1A inhibitors. A dual-target drug discovery framework is constructed by combining AI-driven approaches with classical techniques. An ensemble Quantitative Structure-Activity Relationship (QSAR) model predicts compound affinities, while Directed Message Passing Neural Networks (DMPNN) assess toxicity. In the generative phase, a Hierarchical Graph Generation (HGG) model facilitates the design of potential inhibitors, which are further refined through docking studies, synthesized, and experimentally validated. This approach led to the identification of the pyrazolyl-1H-pyrrolo[2,3-b]pyridine 1 as a potent DYRK1A inhibitor, prompting the synthesis of a novel derivative series. Enzymatic assays confirmed nanomolar-level inhibitory activity, while ORAC assays and LPS-induced pro-inflammatory response evaluations in BV2 microglial cells demonstrated antioxidant and anti-inflammatory properties. Overall, these compounds exhibit strong DYRK1A inhibition alongside promising pharmacological effects.

Drug repurposing is an efficient strategy to accelerate the identification of thera- peutic compounds by finding new uses for existing drugs. Here, we leveraged artifi- cial intelligence (AI) to optimize this process and discover protein-protein interaction (PPI) modulators targeting the Neuronal Calcium Sensor 1. AI models were trained on large datasets of known PPIs, protein structures, and small molecular graphs to predict binding score. Using virtual screening of an FDA-approved drug library, the models generated a prioritized list of candidates. The most promising compounds were selected for experimental validation. This integrated approach efficiently reduced ex- perimental workload, time, and cost, leading to the identification of dipyridamole as a selective modulator of the interaction between NCS-1 and the dopamine D2 receptor. The crystal structure of NCS-1 in complex with dipyridamole elucidates its mechanism of modulation. Our findings highlight the potential of AI to streamline and accelerate the discovery of novel therapeutic PPI modulators

Extracellular protein disulfide isomerase (PDI) is a critical mediator of thrombus formation and a promising target for novel antithrombotic therapies aiming to overcome the bleeding risks associated with current antiplatelet agents. This study aimed to design and evaluate novel, targeted peptide-based PDI inhibitors with a safe physiological profile. We utilized a bioinformatics-driven rational design approach, based on the evolutionary conservation of the PDI active site, to generate a library of 34 targeted peptides. The inhibitory mechanism and biological efficacy of the lead candidate, peptide p23, were evaluated using 500 ns molecular dynamics (MD) simulations, flow cytometry, and in vitro microfluidic thrombosis models under arterial shear stress. Based on bioinformatic analyses, p23 is predicted to interact with the PDI catalytic cleft and potentially form a highly stable inhibitory complex. Functionally, p23 maintained consistent antiplatelet efficacy in the complex matrix of human platelet-rich plasma. Flow cytometry demonstrated that p23 selectively suppresses integrin GPIIb/IIIa conformational activation without impairing P-selectin and CD63 surface expression. Crucially, p23 significantly reduced collagen-dependent three-dimensional thrombus formation in whole blood under biomimetic arterial flow, while still allowing for necessary residual platelet adhesion. Unlike direct integrin antagonists, p23 preserved physiological clot retraction and did not prolong plasma coagulation parameters. These findings validate the rational design of covalent PDI inhibitors and highlight peptide p23 as a promising candidate for preclinical in vivo evaluation, offering a favorable preliminary hemostatic profile and the potential for a reduced impact on hemostasis.