Lei Yang

Ultralarge virtual screenings (ULVSs) evaluate billions of molecules for drug discovery but face cost, flexibility and scalability limits. We introduce AdaptiveFlow, an open-source platform that makes ULVSs more accessible, scalable and efficient and supports artificial intelligence (AI) and machine learning (ML) method development. AdaptiveFlow provides a screening-ready version of the Enamine REAL Space, to our knowledge the largest library of ready-to-dock, drug-like molecules, comprising 69 billion compounds, also available in SELFIES format. An 18-dimensional grid of molecular properties prioritizes promising chemical subspaces, with optional active learning, reducing computational costs by orders of magnitude. AdaptiveFlow integrates >1,500 docking protocols, including GPU-accelerated and ML-based methods, and achieves near-linear scaling on up to 5.6 million CPUs in the Amazon Web Services cloud. We identified nanomolar inhibitors of two disease-relevant targets, ferroptosis suppressor protein 1 (FSP1) and poly(ADP-ribose) polymerase 1. Co-crystal structures provided mechanistic insights into FSP1 inhibition. AdaptiveFlow enables drug discovery at unprecedented scale and supports the development of AI-driven methods.

Benzo[a]pyrene (BaP), a representative polycyclic aromatic hydrocarbon (PAH), is a widespread environmental toxicant and potent ligand of the aryl hydrocarbon receptor (AHR). Yet, how early developmental exposure to BaP influences human neurodevelopment remains poorly understood. We first examined AHR expression dynamics during human embryonic stem cell (ESC)-derived cerebral organoid development and found that AHR expression was highest at the ESC stage and declined during subsequent differentiation, suggesting a potential window of heightened susceptibility to AHR-mediated environmental perturbations. Based on this observation, ESCs were exposed to BaP (0.1, 1 μM) for 7 days prior to organoid generation. BaP exposure did not alter pluripotency, proliferation, cell death, or global transcription of ESCs but increased expression of a subset of AHR target genes. Remarkably, however, organoids derived from BaP-exposed ESCs exhibited profound morphological defects resulting from premature neurogenesis, characterized by disrupted neural rosette organization, reduced EOMES⁺ intermediate progenitors, and increased BCL11B⁺ neurons. Pharmacological inhibition of AHR with CH-223191 attenuated AHR activation and rescued the progenitor-neuron imbalance. These findings identify AHR signaling as a critical upstream mediator of BaP-induced developmental neurotoxicity and highlight the vulnerability of early pluripotent stages to environmental insults.

Introduction: The stability of cobimetinib prepared in a simple aqueous suspension was evaluated. This evaluation aimed to determine how well the compound maintains its integrity over time in a non-complex vehicle for pediatric applications. By monitoring concentration and purity over a defined period, the study provides insight into optimal storage conditions and potential limitations of the preparation. Methods: Cobimetinib tablets were dispersed in purified water to prepare oral suspensions with final concentrations of 1 mg/mL and 12 mg/mL. Each suspension was equally aliquoted and stored under two conditions: room temperature (20 ± 2 °C) and refrigerated conditions (4 °C) (n = 3 per condition). Samples were prepared and analyzed at predefined time points (days 0, 1, 2, 3, 7, 9, and 14) to evaluate stability over time. Physical stability was assessed by pH measurement, as well as visual inspection. Cobimetinib concentrations were quantified using ultra-performance liquid chromatography coupled with ultraviolet detection and mass spectrometry (UPLC-UV-MS). Results: No changes in pH were observed throughout the study. Visual inspections revealed no color change. The cobimetinib oral suspensions at concentrations of 1 mg/mL and 12 mg/mL retained 90 to 110% of their initial concentrations over 14 days when stored at both room temperature and 4 °C. Conclusion: The extemporaneously compounded cobimetinib oral suspensions, at concentrations of 1 mg/mL and 12 mg/mL, demonstrated acceptable stability for up to 14 days under both room temperature and 4 °C. These findings support its potential as a viable alternative dosage form for short-term use in clinical settings.

Benzo[a]pyrene (BaP), a representative polycyclic aromatic hydrocarbon (PAH), is a widespread environmental toxicant and potent ligand of the aryl hydrocarbon receptor (AHR). Yet, how early developmental exposure to BaP influences human neurodevelopment remains poorly understood. We first examined AHR expression dynamics during human embryonic stem cell (ESC)-derived cerebral organoid development and found that AHR expression was highest at the ESC stage and declined during subsequent differentiation, suggesting a potential window of heightened susceptibility to AHR-mediated environmental perturbations. Based on this observation, ESCs were exposed to BaP (0.1, 1 μM) for 7 days prior to organoid generation. BaP exposure did not alter proliferation, cell death, or global transcription of ESCs but increased expression of a subset of AHR target genes. Remarkably, however, organoids derived from BaP-exposed ESCs exhibited profound morphological defects resulting from premature neurogenesis, characterized by disrupted neural rosette organization, reduced EOMES⁺ intermediate progenitors, and increased BCL11B⁺ neurons. Pharmacological inhibition of AHR with CH-223191 attenuated AHR activation and rescued the progenitor-neuron imbalance. These findings identify AHR signaling as a critical upstream mediator of BaP-induced developmental neurotoxicity and highlight the vulnerability of early pluripotent stages to environmental insults.

Label-free high-throughput screening (HTS) technologies have greatly accelerated early drug discovery by allowing direct measurement of biochemical activity without the need for chemical labels. Techniques like MALDI-TOF, RapidFire, desorption electrospray ionization, and acoustic-mist ionization mass spectrometry have shown strong analytical performance. More recently, acoustic droplet ejection mass spectrometry (AEMS) has emerged as a next-generation platform capable of analyzing over a thousand data points per hour. In this study, we thoroughly evaluated an AEMS workflow using cytochrome P450 3A4 (CYP3A4) as a model enzyme. Screening a library of 9702 compounds, this work represents the first successful label-free HTS campaign for CYP3A4 inhibitors in a 384-well format using AEMS. The result using nifedipine as a substrate was comparable to the widely used P450-Glo luminescence assay but with fewer inhibitor hits identifications. To further compare and parse differences between the methods and the substrates, 819 hits were retested using AEMS with luciferin isopropyl acetal, the same substrate used in the luminescence assay. The findings confirmed that both the assay method and the substrate influence which inhibitors were detected. Overall, this study established AEMS as a robust and reliable tool for CYP3A4 inhibitor screening. It offers high analytical precision, fewer false positives, compatibility with multiple substrates, and direct detection of metabolites, providing a practical way to validate hits and improve confidence in early drug discovery efforts.

Effective therapies for high-risk medulloblastoma (MB), particularly MYC-driven Group 3 (G3) MB, remain elusive due to limited druggable mutations, poor blood–brain barrier (BBB) penetration, and rapid resistance. We developed SINBA (Synergy Inference by Data-driven Network-Based Bayesian Analysis), a systems biology framework that predicts synergistic, BBB-permeable drug combinations by identifying hidden drivers that sustain oncogenic programs without detectable genetic or transcriptional alterations. Integrating MB-specific gene networks, transcriptomic data, and drug–gene interactions, SINBA prioritized 32 candidate combinations, of which 19 were experimentally validated as synergistic. The MEK inhibitor mirdametinib and p38 inhibitor regorafenib emerged as the top brain-penetrant pair, suppressing G3 MB progression and extending survival in human xenograft and immunocompetent mouse models, with efficacy enhanced by low-dose radiation. Single-cell analysis revealed selective targeting of the developmental origins of G3 MB, accompanied by immune microenvironment reprogramming. These findings establish hidden-driver inference as a generalizable framework for rational drug combination discovery.

Molecular glues powerfully control protein proximity but have largely eluded direct screening. A promising avenue for addressing this challenge lies within pinpointing the fundamental features for function-first identification of molecular gluing events. In the widely accepted mechanism, a molecular glue stabilizes two proteins within a ternary complex─here, we show how differences in affinity for ternary and binary complexes directly categorize glues from nonglues. We leverage these differences together with high-throughput chemical synthesis and affinity-selection mass-spectrometry to discover a molecular glue from a suite of over 20,000 crude chemical reaction mixtures. Orthogonal assays robustly support the identification of molecular glues via ternary complex stability. Our findings suggest that a roadmap for de novo molecular glue discovery lies within kinetic profiling of unpurified mixtures of small organic molecules against protein pairs.

Supplementary Table S4 from Combination of Ribociclib with BET-Bromodomain and PI3K/mTOR Inhibitors for Medulloblastoma Treatment In Vitro and In Vivo