Explore the vast range of titles available.

A variety of machine learning methods such as naive Bayesian, support vector machines and more recently deep neural networks are demonstrating their utility for drug discovery and development. These leverage the generally bigger datasets created from high-throughput screening data and allow prediction of bioactivities for targets and molecular properties with increased levels of accuracy. We have only just begun to exploit the potential of these techniques but they may already be fundamentally changing the research process for identifying new molecules and/or repurposing old drugs. The integrated application of such machine learning models for end-to-end (E2E) application is broadly relevant and has considerable implications for developing future therapies and their targeting.This Perspective describes the application of machine learning models in the design, synthesis and characterisation of molecules at different stages in the drug discovery and development process.

We currently have an incomplete understanding of why only a fraction of human antibodies that bind to flaviviruses block infection of cells. Here we define the footprint of a strongly neutralizing human monoclonal antibody (mAb G9E) with Zika virus (ZIKV) by both X-ray crystallography and cryo-electron microscopy. Flavivirus envelope (E) glycoproteins are present as homodimers on the virion surface, and G9E bound to a quaternary structure epitope spanning both E protomers forming a homodimer. As G9E mainly neutralized ZIKV by blocking a step after viral attachment to cells, we tested if the neutralization mechanism of G9E was dependent on the mAb cross-linking E molecules and blocking low-pH triggered conformational changes required for viral membrane fusion. We introduced targeted mutations to the G9E paratope to create recombinant antibodies that bound to the ZIKV envelope without cross-linking E protomers. The G9E paratope mutants that bound to a restricted epitope on one protomer poorly neutralized ZIKV compared to the wild-type mAb, demonstrating that the neutralization mechanism depended on the ability of G9E to cross-link E proteins. In cell-free low pH triggered viral fusion assay, both wild-type G9E, and epitope restricted paratope mutant G9E bound to ZIKV but only the wild-type G9E blocked fusion. We propose that, beyond antibody binding strength, the ability of human antibodies to cross-link E-proteins is a critical determinant of flavivirus neutralization potency.

Adenosine (ADO) is an extracellular signaling molecule generated locally under conditions that produce ischemia, hypoxia, or inflammation. It is involved in modulating a range of physiological functions throughout the brain and periphery through the membrane-bound G protein-coupled receptors, called adenosine receptors (ARs) A 1 AR, A 2A AR, A 2B AR, and A 3 AR. These are therefore important targets for neurological, cardiovascular, inflammatory, and autoimmune diseases and are the subject of drug development directed toward the cyclic adenosine monophosphate and other signaling pathways. Initially using public data for A 1 AR agonists we generated and validated a Bayesian machine learning model (Receiver Operator Characteristic of 0.87) that we used to identify molecules for testing. Three selected molecules, crisaborole, febuxostat and paroxetine, showed initial activity in vitro using the HEK293 A 1 AR Nomad cell line. However, radioligand binding, β-arrestin assay and calcium influx assay did not confirm this A 1 AR activity. Nevertheless, several other AR activities were identified. Febuxostat and paroxetine both inhibited orthosteric radioligand binding in the µM range for A 2A AR and A 3 AR. In HEK293 cells expressing the human A 2A AR, stimulation of cAMP was observed for crisaborole (EC 50 2.8 µM) and paroxetine (EC 50 14 µM), but not for febuxostat. Crisaborole also increased cAMP accumulation in A 2B AR-expressing HEK293 cells, but it was weaker than at the A 2A AR. At the human A 3 AR, paroxetine did not show any agonist activity at 100 µM, although it displayed binding with a K i value of 14.5 µM, suggesting antagonist activity. We have now identified novel modulators of A 2A AR, A 2B AR and A 3 AR subtypes that are clinically used for other therapeutic indications, and which are structurally distinct from previously reported tool compounds or drugs.

Recent advances in machine learning (ML) have led to newer model architectures including transformers (large language models, LLMs) showing state of the art results in text generation and image analysis as well as few-shot learning (FSLC) models which offer predictive power with extremely small datasets. These new architectures may offer promise, yet the ‘no-free lunch’ theorem suggests that no single model algorithm can outperform at all possible tasks. Here, we explore the capabilities of classical (SVR), FSLC, and transformer models (MolBART) over a range of dataset tasks and show a ‘goldilocks zone’ for each model type, in which dataset size and feature distribution (i.e. dataset “diversity”) determines the optimal algorithm strategy. When datasets are small ( < 50 molecules), FSLC tend to outperform both classical ML and transformers. When datasets are small-to-medium sized (50-240 molecules) and diverse, transformers outperform both classical models and few-shot learning. Finally, when datasets are of larger and of sufficient size, classical models then perform the best, suggesting that the optimal model to choose likely depends on the dataset available, its size and diversity. These findings may help to answer the perennial question of which ML algorithm is to be used when faced with a new dataset. Machine learning (ML) is a powerful tool in the field of drug discovery, with the continuous development of new models, however, rational selection of the most appropriate model based on the task remains challenging. Here, the authors explore the capabilities of classical ML algorithms and newer models over a range of dataset tasks and show an optimal zone for each model type, developing a predictive model to aid in the selection of a modeling method based on dataset size and diversity.

PurposeIndividuals with the rare genetic disorder Pitt Hopkins Syndrome (PTHS) do not have sufficient expression of the transcription factor 4 (TCF4) which is located on chromosome 18. TCF4 is a basic helix-loop-helix E protein that is critical for the normal development of the nervous system and the brain in humans. PTHS patients lacking sufficient TCF4 frequently display gastrointestinal issues, intellectual disability and breathing problems. PTHS patients also commonly do not speak and display distinctive facial features and seizures. Recent research has proposed that decreased TCF4 expression can lead to the increased translation of the sodium channel Nav1.8. This in turn results in increased after-hyperpolarization as well as altered firing properties. We have recently identified through a drug repurposing screen an FDA approved dihydropyridine calcium antagonist nicardipine used to treat angina, which inhibited Nav1.8.MethodsWe have now performed behavioral testing in groups of 10 male Tcf4(± ) PTHS mice dosing by oral gavage at 3 mg/kg once a day for 3 weeks using standard methods to assess sociability, nesting, fear conditioning, self-grooming, open field and test of force.ResultsNicardipine returned this spectrum of behavioral deficits in the Tcf4(± ) PTHS mouse model to WT levels and resulted in statistically significant results.ConclusionsThese in vivo results in the well characterized Tcf4(± ) PTHS mice may suggest the potential to test this already approved drug further in a clinical study with PTHS patients or suggest the potential for use off label under compassionate use with their physician.

Stimulator of interferon genes (STING) is a receptor in the endoplasmic reticulum that propagates innate immune sensing of cytosolic pathogen-derived and self DNA 1 . The development of compounds that modulate STING has recently been the focus of intense research for the treatment of cancer and infectious diseases and as vaccine adjuvants 2 . To our knowledge, current efforts are focused on the development of modified cyclic dinucleotides that mimic the endogenous STING ligand cGAMP; these have progressed into clinical trials in patients with solid accessible tumours amenable to intratumoral delivery 3 . Here we report the discovery of a small molecule STING agonist that is not a cyclic dinucleotide and is systemically efficacious for treating tumours in mice. We developed a linking strategy to synergize the effect of two symmetry-related amidobenzimidazole (ABZI)-based compounds to create linked ABZIs (diABZIs) with enhanced binding to STING and cellular function. Intravenous administration of a diABZI STING agonist to immunocompetent mice with established syngeneic colon tumours elicited strong anti-tumour activity, with complete and lasting regression of tumours. Our findings represent a milestone in the rapidly growing field of immune-modifying cancer therapies. A small-molecule agonist for the cGAS–STING pathway has systemic activity in a mouse model of colon cancer.

CLN1 disease, also called infantile neuronal ceroid lipofuscinosis (NCL) or infantile Batten disease, is a fatal neurodegenerative lysosomal storage disorder resulting from mutations in the CLN1 gene encoding the soluble lysosomal enzyme palmitoyl-protein thioesterase 1 (PPT1). Therapies for CLN1 disease have proven challenging because of the aggressive disease course and the need to treat widespread areas of the brain and spinal cord. Indeed, gene therapy has proven less effective for CLN1 disease than for other similar lysosomal enzyme deficiencies. We therefore tested the efficacy of enzyme replacement therapy (ERT) by administering monthly infusions of recombinant human PPT1 (rhPPT1) to PPT1-deficient mice (Cln1-/-) and CLN1R151X sheep to assess how to potentially scale up for translation. In Cln1-/- mice, intracerebrovascular (i.c.v.) rhPPT1 delivery was the most effective route of administration, resulting in therapeutically relevant CNS levels of PPT1 activity. rhPPT1-treated mice had improved motor function, reduced disease-associated pathology, and diminished neuronal loss. In CLN1R151X sheep, i.c.v. infusions resulted in widespread rhPPT1 distribution and positive treatment effects measured by quantitative structural MRI and neuropathology. This study demonstrates the feasibility and therapeutic efficacy of i.c.v. rhPPT1 ERT. These findings represent a key step toward clinical testing of ERT in children with CLN1 disease and highlight the importance of a cross-species approach to developing a successful treatment strategy.

Change history: In this Letter, author Ana Puhl was inadvertently omitted; this error has been corrected online. An amendment to this paper has been published and can be accessed via a link at the top of the paper

The portfolio of SARS-CoV-2 small molecule drugs is currently limited to a handful that are either approved (remdesivir), emergency approved (dexamethasone, baricitinib) or in advanced clinical trials. We have tested 45 FDA-approved kinase inhibitors in vitro against murine hepatitis virus (MHV) as a model of SARS-CoV-2 replication and identified 12 showing inhibition in the delayed brain tumor (DBT) cell line. Vandetanib, which targets the vascular endothelial growth factor receptor (VEGFR), the epidermal growth factor receptor (EGFR), and the RET-tyrosine kinase showed the most promising results on inhibition versus toxic effect on SARS-CoV-2-infected Caco-2 and A549-hACE2 cells (IC50 0.79 uM) while also showing a reduction of > 3 log TCID50/mL for HCoV-229E. The in vivo efficacy of vandetanib was assessed in a mouse model of SARS-CoV-2 infection and statistically significantly reduced the levels of IL-6, IL-10, TNF-a;, and mitigated inflammatory cell infiltrates in the lungs of infected animals but did not reduce viral load. Vandetanib rescued the decreased IFN-1b; caused by SARS-CoV-2 infection in mice to levels similar to that in uninfected animals. Our results indicate that the FDA-approved vandetanib is a potential therapeutic candidate for COVID-19 positioned for follow up in clinical trials either alone or in combination with other drugs to address the cytokine storm associated with this viral infection. Competing Interest Statement SE is CEO of Collaborations Pharmaceuticals, Inc. ACP is an employee at Collaborations Pharmaceuticals, Inc. All other c-authors have no conflicts of interest.

Abstract SARS-CoV-2 is a newly identified virus that has resulted in over 1.3 M deaths globally and over 59 M cases globally to date. Small molecule inhibitors that reverse disease severity have proven difficult to discover. One of the key approaches that has been widely applied in an effort to speed up the translation of drugs is drug repurposing. A few drugs have shown in vitro activity against Ebola virus and demonstrated activity against SARS-CoV-2 in vivo. Most notably the RNA polymerase targeting remdesivir demonstrated activity in vitro and efficacy in the early stage of the disease in humans. Testing other small molecule drugs that are active against Ebola virus would seem a reasonable strategy to evaluate their potential for SARS-CoV-2. We have previously repurposed pyronaridine, tilorone and quinacrine (from malaria, influenza, and antiprotozoal uses, respectively) as inhibitors of Ebola and Marburg virus in vitro in HeLa cells and of mouse adapted Ebola virus in mouse in vivo. We have now tested these three drugs in various cell lines (VeroE6, Vero76, Caco-2, Calu-3, A549-ACE2, HUH-7 and monocytes) infected with SARS-CoV-2 as well as other viruses (including MHV and HCoV 229E). The compilation of these results indicated considerable variability in antiviral activity observed across cell lines. We found that tilorone and pyronaridine inhibited the virus replication in A549-ACE2 cells with IC50 values of 180 nM and IC50 198 nM, respectively. We have also tested them in a pseudovirus assay and used microscale thermophoresis to test the binding of these molecules to the spike protein. They bind to spike RBD protein with Kd values of 339 nM and 647 nM, respectively. Human Cmax for pyronaridine and quinacrine is greater than the IC50 hence justifying in vivo evaluation. We also provide novel insights into their mechanism which is likely lysosomotropic. Competing Interest Statement SE is CEO of Collaborations Pharmaceuticals, Inc. ACP and TRL are employees at Collaborations Pharmaceuticals, Inc. Collaborations Pharmaceuticals, Inc. has obtained FDA orphan drug designations for pyronaridine, tilorone and quinacrine for use against Ebola. CPI have also filed a provisional patent for use of these molecules against Marburg and other viruses.