Chemistry and Materials Science

Abstract: Gonadotropin Releasing Hormone receptors (GnRHRs) are overexpressed in several hormone-dependent malignancies, making them attractive molecular targets for selective anticancer drug delivery. Peptide drug conjugates (PDCs) are a promising therapy for cancer and autoimmune diseases with high specificity and reduced toxicity. In this study, the three-dimensional structures of two previously synthesized mitoxantrone-GnRH conjugates, con3 and con7, were elucidated using high-resolution NMR spectroscopy in combination with molecular dynamics (MD) simulations. Complete ¹H and ¹³C resonance assignments were achieved in DMSO-d₆ through two-dimensional NMR experiments. NOESY-derived distance restraints were subsequently used to refine the conformational ensembles obtained from MD simulations performed in water and DMSO. Both conjugates exhibited compact bent conformations with a U-shaped peptide backbone. The mitoxantrone moiety is positioned close to the peptide backbone in water simulations and NMR-refined structures, while it is positioned farther away in DMSO, without affecting the orientation of key residues involved in GnRH receptor binding. Importantly, His², Trp³, and Arg⁸ remain solvent-exposed, whereas the disulfide bond is easily accessible to the solvent, consistent with the proposed drug release mechanism by the thioredoxin system. NMR-restrained molecular modeling confirmed the dominant conformational features predicted by the unconstrained theoretical simulations. Overall, these findings provide better structural understanding of the molecular organization of mitoxantrone–GnRH conjugates, highlighting key receptor-recognition residues and supporting both the proposed thioredoxin-mediated drug release mechanism and their previously reported biological properties. These insights may facilitate the rational design and optimization of improved GnRH peptide–drug conjugates for targeted therapy.

Abstract:

Nerium oleander L is a medicinal plant of significant ethnopharmacological importance, yet its well-documented toxicity necessitates rigorous phytochemical characterization and standardization. This study comprehensively evaluated the influence of eight extraction solvents of varying polarity (water, methanol, ethanol, acetone, dichloromethane, chloroform, ethyl acetate, and hexane) and three analytical methods (foam-forming, optical activity, and spectrophotometric vanillin-acetic acid) on saponin quantification in N. oleander leaf extracts. Plant material was collected, dried, pulverized, and extracted via maceration, with saponin content determined using the three analytical approaches. Data were analyzed using two-way ANOVA with replication and Tukey’s HSD post-hoc testing. Results demonstrated statistically significant variation in saponin yield attributable to both extraction solvent (F = 143.44, p < 0.001) and analytical method (F = 90.53, p < 0.001), with a significant interaction effect (F = 6.35, p < 0.001). Polar solvents, particularly methanol (18.22 ± 0.57%) and ethanol (16.32 ± 3.94%), exhibited superior extraction efficiency, consistent with the glycosidic nature of saponins. Hexane yielded anomalously elevated content (18.00 ± 2.57%), potentially due to unique hydrogen-bonding interactions. The spectrophotometric method consistently produced the highest saponin values due to enhanced sensitivity, while the foam-forming method proved suitable only for preliminary screening, and optical activity lacked specificity due to interference from other chiral constituents. This study conclusively demonstrates that methodological selection critically influences saponin quantification, with significant implications for ethnopharmacological standardization and quality control. Based on these findings, methanol extraction followed by spectrophotometric vanillin-acetic acid analysis is recommended as the optimal protocol for routine quantitative analysis, providing the highest sensitivity, reproducibility, and practical feasibility. These findings provide an evidence-based framework for method selection and underscore the necessity of harmonized analytical protocols to ensure accuracy, reproducibility, and comparability in natural product research, thereby supporting the safe and effective development of N. oleander-based therapeutics and cosmetic formulations.

Abstract: Background: Colorectal cancer (CRC) remains a major cause of cancer-related mortality worldwide and ranks third in Colombia. Despite therapeutic advances, limitations such as reduced efficacy, adverse effects, and drug resistance persist. In the search for novel ther-apeutic strategies, molecular hybridization of melatonin and furanochalcone—compounds with antioxidant and antitumor properties—led to the synthesis of a novel hybrid mole-cule Mel-Fur (6f), a promising candidate for CRC treatment. Objectives: The aim of this study was to develop and validate an HPLC-DAD analytical method for quantification of Mel-Fur in serum and murine organ matrices. Methods: Chromatographic analysis used an Agilent Series 1200 system with a diode array detector and a C30 column under opti-mized conditions: acetonitrile: water (85:15, v/v) as mobile phase, flow rate 0.8 mL/min, detection at 342 nm, retention time 4.3 min. Results: The method fulfilled ICH Q2 (R1) and FDA validation guidelines, showing high selectivity, excellent linearity (R² > 0.999), sensitivity, precision, accuracy (RE% and CV% < 15%), recovery above 93%, and analyte stability for up to 8 days. The validated method was applied in a pilot in vivo biodistribu-tion study. Following oral administration of a single dose of Mel-Fur (1000 mg/kg) in BALB/c mice, rapid absorption and elimination were observed, with measurable systemic exposure and effective distribution into peripheral tissues. Pharmacokinetic analysis re-vealed preferential accumulation in lungs and liver, with sustained presence in colon. Conclusions: These findings provide a robust analytical tool and preliminary pharmaco-kinetic insights supporting further preclinical development of Mel-Fur as a potential therapeutic candidate for CRC.

Abstract: As machine learning (ML) becomes increasingly integrated into drug discovery, reliance on legacy datasets and superficial performance metrics threatens to stall genuine progress. This perspective examines common pitfalls in solubility modeling, specifically overreliance on flawed public datasets and insufficient similarity analysis between training and test sets. By comparing performance on "real-world" datasets with consistent experimental conditions, specifically the Biogen and ASAP Discovery sets, we demonstrate that inflated correlations can mask poor generalizability. We propose new guidelines for authors, reviewers, and journals to elevate the standard of ML validation.

Abstract: Background/Objectives: Cancer resistance, pathway redundancy, and compensatory signaling challenge traditional therapies. Dual-target strategies address this by engaging two disease-relevant proteins within a single molecule. This review compares classical dual inhibitors with dual proteolysis-targeting chimeras (dual PROTACs) to evaluate the therapeutic advantages of degradation over occupancy. Methods: We examine oncology target pairs featuring documented examples of both dual inhibitors and dual PROTACs. The biological rationale for co-targeting is analyzed alongside a comparative assessment of their chemical frameworks, focusing heavily on the synthetic strategies, length, and structure of linkers required for dual-PROTAC ternary complex formation. Results: While dual inhibitors rely on active-site occupancy, dual PROTACs leverage the ubiquitin–proteasome system for catalytic target elimination. Transitioning from dual inhibition to dual degradation frequently enhances antitumor efficacy, extends duration of action, and overcomes resistance mutations. Optimizing linker design remains the critical factor in balancing the simultaneous degradation kinetics of two distinct proteins. Conclusions: Dual PROTACs provide distinct advantages over traditional inhibitors by completely destroying target proteins rather than merely blocking them. This comparison offers a practical entry point and actionable synthetic strategies for medicinal chemists designing multi-target protein degraders.

Abstract: Colloidal instability remains a dominant cause of product failure, manufacturing attrition, and safety risk across biopharmaceutical modalities including monoclonal antibodies (mAbs), bispecific antibodies, antibody-drug conjugates (ADCs), and mRNA-lipid nanoparticle systems. Protein aggregation is a critical quality attribute (CQA) under ICH Q6B, linked to immunogenicity, reduced potency, and adverse patient outcomes. Despite the transformative impact of machine learning (ML) on protein structure prediction and molecular design, its application to formulation-dependent colloidal stability prediction remains fragmented, poorly benchmarked, and largely disconnected from regulatory frameworks. This review systematically examines ML approaches for predicting aggregation propensity, viscosity, solubility, liquid-liquid phase separation, and shelf-life across biopharmaceutical modalities. We critically assess experimental data sources, feature engineering strategies, and ML architectures spanning classical models, deep learning, graph neural networks, and protein language models, alongside the emerging role of explainable AI (XAI). No standardised, cross-modality ML benchmarking framework for colloidal stability currently exists -- a gap that constrains generalisation, reproducibility, and regulatory acceptance. Principal unresolved challenges include dataset scarcity, label noise, external validation deficits, and proprietary data silos. A decade roadmap for integrating physics-informed ML, autonomous formulation laboratories, and foundation models into next-generation biologics development is proposed.

Abstract: Molecular hydrogen (H₂) has emerged as a potential redox-active molecule with distinctive physicochemical and biological properties. Due to its small molecular size and rapid diffusion, H₂ readily penetrates biological membranes and selectively interacts with highly reactive oxygen and nitrogen species particularly hydroxyl radicals (•OH) and peroxynitrite (ONOO⁻) while largely preserving physiological redox signaling. Experimental and clinical studies further suggest that H₂ may influence intracellular signaling pathways associated with oxidative stress and inflammation, including activation of the nuclear factor erythroid 2–related factor 2 (Nrf2) pathway and suppression of nuclear factor κB (NF-κB) signaling. Human studies employing hydrogen-rich water (HRW), inhaled hydrogen gas, or hydrogen-enriched dialysate have reported improvements in biomarkers related to oxidative stress, inflammation, cardiometabolic risk, and immune function, although effect sizes and reproducibility vary across studies. To date, however, no peer-reviewed investigations have evaluated sublingual delivery of molecular hydrogen. This review synthesizes current mechanistic and clinical evidence supporting the biological activity of H₂ and examines the physicochemical rationale for exploring sublingual administration as a potential alternative delivery route. Considerations related to dissolved hydrogen concentration, oxidation–reduction potential, stability, and safety are discussed, alongside key translational gaps that must be addressed. Rigorous pharmacokinetic studies and randomized controlled trials will be essential to determine the feasibility, bioavailability, and clinical relevance of sublingual hydrogen delivery.

Abstract:

This review consolidates current knowledge on the phytochemical composition, traditional uses, pharmacological properties, and industrial application of Ricinus communis L. This plant belongs to the Euphorbiaceae family and is a globally distributed plant of considerable medicinal and industrial importance. It is rich in bioactive compounds, notably ricinoleic acid as the dominant fatty acid in seed oil, as well as ricin, ricinine, phenolic acids and flavonoids distributed across different plant parts. Variations in phytochemical profiles among cultivars and tissues are influenced by genetic and environmental influences. Traditional medicinal uses of the leaves, roots, seeds, and oil particularly for inflammatory conditions, pain, infections, wound healing, and gastrointestinal disorders are critically examined in relation to experimental pharmacological evidence. Castor oil extracted from the R. communis plant remains central to the plant’s industrial value, serving as a renewable feedstock for pharmaceuticals, cosmetics, polymers, lubricants, and biofuels due to the unique hydroxyl functionality of ricinoleic acid. However, the presence of the highly toxic protein ricin in unprocessed seeds necessitates strict processing and safety controls. Overall, R. communis emerges as a chemically versatile species with significant therapeutic and industrial potential, warranting further research into cultivar-specific chemistry, standardisation of extraction and testing methods, and safe value-adding applications.

Abstract: The continuous emergence of SARS-CoV-2 variants necessitates the identification of effective multi-target antiviral agents with enhanced stability and binding efficiency. This study employed an integrated computational approach, including molecular docking, molecular dynamics (MD) simulations, free energy landscape (FEL) analysis, and Molecular Mechanics Poisson-Boltzmann Surface Area (MM-PBSA) calculations, to evaluate the inhibitory potential of twenty natural compounds against SARS-CoV-2 proteins 7XJW, 8DRW, and 9PFH. Molecular docking identified Amentoflavone as the most promising candidate, exhibiting strong binding affinities toward all targets through favorable hydrogen-bond and hydrophobic interactions within the active sites. Interaction analysis revealed that its biflavonoid scaffold promoted extensive ligand–protein complementarity through hydroxyl and aromatic functional groups. MD simulations demonstrated stable protein-ligand complexes, characterized by low fluctuations in RMSD, RMSF, SASA, and radius of gyration values throughout the 100 ns trajectories. Persistent hydrogen-bond interactions further supported complex stability. FEL analysis revealed compact low-energy conformational basins, indicating thermodynamically favorable binding states. MM-PBSA calculations confirmed favorable binding free energies primarily driven by van der Waals and electrostatic contributions, with the 7XJW-Amentoflavone complex exhibiting the most favorable energetic profile. Overall, these findings highlight Amentoflavone as a promising multi-target inhibitor and potential lead compound for future antiviral drug development and experimental validation against SARS-CoV-2.

Abstract:

Background/Objectives: Alzheimer’s disease (AD) is a neurodegenerative disorder with a complex pathomechanism. Acetylcholinesterase (AChE) and monoamine oxidase-B (MAO-B) are key targets regulating neurotransmitter levels, and dual inhibitors (compounds 1–46) were designed as experimental candidates for AD therapy. Methods: Drug-likeness parameters were estimated using pkCSM, SwissADME web tools, and MoloVol software (v1.2.0). SwissADME predicted gastrointestinal absorption and blood–brain barrier penetration, whereas pkCSM evaluated P-glycoprotein recognition and CYP450 inhibition. Toxicological profiles of compounds (1–46) were assessed with DataWarrior software (v06.05.04), which classified them as mutagenic, carcinogenic, reproductive, or irritant. Results: Most compounds complied with Lipinski’s rule (excluding 12 and 35) indicating favorable absorption and permeability. All compounds showed TPSA < 140 Å2, indicating good intestinal absorption, while compounds 1, 3–6, 8, 11-16, 18, 19, 27, 30, 31, 34, 36-38, and 44–46 displayed TPSA < 60 Å2, suggesting blood–brain barrier penetration. The majority of compounds were predicted P-glycoprotein substrates, potentially limiting oral absorption and blood-brain barrier penetration. Metabolic profiling revealed inhibition of CYP1A2, 2C19, 2C9, 2D6, and 3A4, highlighting drug–drug interaction risks. Toxicological analysis identified mutagenicity (compounds 4, 5, 19, 20 and 27), carcinogenicity (compounds 4, 5, 8, 18 and 19), reproductive toxicity (compounds 15, 16 and 19–23), and irritant effects (compounds 7, 11, 17 and 20). Conclusions: Computational findings support further in vitro and in vivo evaluation of compounds 1, 3, 6, 13, 14, 30, 31, 34, 36–38, and 44–46 as dual AChE/MAO-B inhibitors and potentially new drugs for AD treatment.

Abstract: The regeneration of the dentin-pulp complex remains a major challenge in regenerative endodontics. While conventional therapeutic approaches are effective in eliminating infection and preserving dental structure, they fail to restore the biological functionality of the pulp tissue. In recent years, three-dimensional (3D) printing and biopolymer-based bioprinting have opened unprecedented opportunities in dental tissue engineering, enabling the fabrication of biomimetic scaffolds with precisely controlled structural and bioactive properties. This review synthesizes current advances in bioprinting technolo-gies, the diversity of biomaterials and bioinks employed, and the various stem cell sources utilized in pulp regeneration. It further examines how the three-dimensional microenvironment modulates cell viability, odontogenic differentiation, and the pro-motion of angiogenesis and neurogenesis, emphasizing the role of scaffold composition, mechanical properties, and internal architecture in influencing regenerative outcomes. Additionally, persistent challenges are discussed, including the optimization of bioink formulations, the achievement of functional vascular integration, and long-term valida-tion of regenerated tissues, underscoring the need for multidisciplinary strategies to fa-cilitate clinical translation. By integrating recent evidence, this review establishes a conceptual framework for the development of personalized and predictable approaches to dentin-pulp complex reconstruction.

Abstract: Alzheimer’s disease (AD) is a multifactorial neurodegenerative disorder driven by complex interactions between protein aggregation, oxidative stress, neuroinflammation, and cellular dysfunction. Among plant-derived compounds, curcumin has emerged as one of the most extensively studied polyphenols due to its broad spectrum of biological activities. This review provides a critical synthesis of mechanistic, preclinical, and clinical evidence on curcumin in AD. Experimental studies consistently demonstrate that curcumin modulates key pathogenic processes, including neuroinflammatory signaling, oxidative stress, and amyloid-β aggregation, with more limited evidence for effects on tau pathology. While in vitro studies offer detailed mechanistic insights, in vivo models provide more integrated evidence, including improvements in cognitive performance and reductions in pathological markers. Despite this strong preclinical foundation, clinical evidence remains limited and inconsistent. Randomized controlled trials have not demonstrated clear therapeutic efficacy, with outcomes strongly influenced by formulation, bioavailability, and study design. Poor solubility, rapid metabolism, and limited brain exposure remain key translational barriers. In response, increasing attention has been directed toward formulation strategies and structurally related compounds. Emerging curcuminoids, such as bisdemethoxycurcumin (BDMC), are discussed as potential next-generation candidates. Preliminary evidence suggests that BDMC may modulate oxidative stress, autophagy, astrocyte senescence, and amyloid-related processes, although data remain largely preclinical. Overall, curcumin represents a mechanistically rich and preclinically promising multi-target compound, but with unresolved translational limitations. Future research should prioritize pharmacokinetic optimization, formulation-dependent validation, and exploration of novel curcuminoid strategies to bridge the gap between experimental findings and clinical application in AD.

Abstract: Introduction: Tetrazole and thiazolidine-4-one derivatives are important heterocyclic scaffolds with diverse pharmacological activities, including antimicrobial and antioxidant effects. This study focuses on the design and synthesis of novel Schiff base–derived analogues using a green synthetic approach to improve biological efficacy and reduce environmental impact. Methods: Schiff bases (2a–2h) were synthesized using tetrabutylammonium iodide as a green catalyst in aqueous medium. These were further converted into tetrazole (3a–3h) and thiazolidine-4-one (4a–4h) derivatives using sodium azide and thioglycolic acid. Structures were confirmed by FTIR, ¹H NMR, and ¹³C NMR spectroscopy. Antioxidant activity was evaluated using the DPPH assay, while antimicrobial activity was assessed by the zone of inhibition method. Molecular docking was performed against Penicillin-Binding Protein 4 (3ZG8), CYP51 (5V5Z), and 1OAF. Results: Compounds 2a, 2b, 3a, and 4a showed strong antifungal activity, exceeding standard drugs. Compounds 2d, 3b, and 4b exhibited superior antibacterial activity. Several derivatives demonstrated higher antioxidant activity than ascorbic acid. Docking studies confirmed stable ligand–protein interactions, with compound 4f showing the highest binding affinity. Discussion: Substituent variation influenced biological activity. Electron-donating and withdrawing groups affected potency. Docking results supported experimental findings and confirmed target interactions. The green synthesis improved efficiency and reduced environmental risk. Conclusion: These derivatives show promising antimicrobial and antioxidant potential. Compound 4f emerged as a lead candidate for further optimization and drug development.

Abstract: The hexane extract of Psoralea drupacea Bunge fruits was initially evaluated for antivi-ral activity against SARS-CoV-2 based on GC-MS data indicating high bakuchiol con-tent (87.74%). Unexpectedly, the extract showed no antiviral effect in Vero E6 cells due to cytotoxicity (CC₅₀ = 7.5 μg/mL), while purified bakuchiol demonstrated moderate antiviral activity (IC₅₀ = 6.2 ± 0.8 μg/mL; SI = 2.9). Quantitative NMR revealed that the actual bakuchiol content in the extract was 44.3% — approximately half the GC-MS value — explaining the lack of efficacy at non-cytotoxic concentrations. Both the ex-tract and purified bakuchiol effectively blocked the RBD-ACE2 interaction in a com-petitive ELISA (71.3% inhibition at 50 μM for bakuchiol; IC₅₀ = 18.5 μM). Notably, the extract also inhibited the viral main protease 3CLpro (IC₅₀ = 32.0 ± 3.5 μg/mL), while purified bakuchiol showed no such activity. These findings reveal a dual mechanism: bakuchiol inhibits viral entry via RBD-ACE2 blockade, while other extract components (e.g., angelicin, psoralen) suppress viral replication via 3CLpro inhibition.

Abstract: Aspergillus nidulans, a model filamentous fungus endowed with well-established genetic tools and a repertoire of cryptic secondary metabolite biosynthetic gene clusters (BGCs), is extensively exploited as a microbial chassis for heterologous biosynthesis. Mining of its secondary metabolites facilitates the discovery of novel bioactive compounds and the development and application of chassis cells. In the course of heterologous expression of exogenous genes in A. nidulans, we unexpectedly observed the activation of cryptic host BGCs, which resulted in substantial alterations to its secondary metabolic profile. Four previously undescribed compounds (1–4), together with six known analogs (5–10), were isolated from three recombinant A. nidulans strains. Notably, compounds 1–3 are the first naturally occurring examples of diketopiperazine-isoindolinone hybrid alkaloids, while compound 4 is a previously unreported benzofuran carboxylic acid derivative. Their structures and absolute configurations were assigned by interpretation of a combination of spectroscopic data and electronic circular dichroism calculations. Compounds 4 and 5 exhibited potent DPPH radical scavenging activity (IC₅₀, 6.01 and 7.00 μg·mL^-1, respectively). This study uncovers a "metabolic perturbation" effect on the host metabolic network during heterologous expression and offers a new strategy for activating silent gene clusters and discovering novel natural products through genetic manipulation.

Abstract: Cosmetic preservatives should have reduced percutaneous absorption to lower the risk of systemic exposure and skin irritation. In this work, Escherichia coli β-galactosidase was used to enzymatically modify several of the commonly used cosmetic preservatives to produce their corresponding galactosylated derivatives: benzyl alcohol β-D-galactopyranoside 7, 2-phenoxyethanol β-D-galactopyranoside 8, chlorphenesin β-D-galactopyranoside 9, 1,2-hexanediol β-D-galactopyranoside 10, 1,2-octanediol β-D-galactopyranoside 11, and 2-phenylethyl β-D-galactopyranoside 12. HPLC and NMR spectroscopy were used to analyze the synthesized derivatives. The Franz diffusion cell assay was used to evaluate skin penetration. 2-phenoxyethanol (PE), chlorphenesin (CPN), and 2-phenylethanol (PhE), exhibited measurable skin penetration with flux values ranging from 3.82 to 7.34 µg·h⁻¹·cm⁻² and permeability coefficients (Kp) between 1.38 and 3.00 ×10⁻³ cm·h⁻¹. In contrast, their galactosylated derivatives showed markedly reduced permeation under the same experimental conditions. Moreover, brine shrimp lethality assays indicated that galactosylated derivatives had significantly higher LD₅₀ values (1.6–2.1 mg/mL) than their parent compounds (0.1–0.79 mg/mL), suggesting lower cytotoxicity. These findings suggest that enzymatic galactosylation can significantly decrease skin permeability and the toxicity of cosmetic preservatives, highlighting its potential as a strategy to improve the safety of cosmetic ingredients.

Abstract: Quantitative structure–activity relationship (QSAR) modeling has traditionally relied on expert-designed molecular descriptors to encode chemical structures. DeepSnap is a descriptor-free QSAR approach that converts three-dimensional molecular structures into image representations and feeds them directly into convolutional neural networks for activity prediction. The method generates a conformer for each molecule, renders it as a color-coded molecular image, and captures omnidirectional snapshots from systematically varied viewing angles. This review traces DeepSnap from its introduction in 2018 to its current state. The method has been applied to 35 nuclear receptor endpoints from the Tox21 10K library (mean AUC 0.884), 59 molecular initiating event models spanning the full Tox21 target panel, rat hepatic clearance (ensemble AUC 0.943), and blood–brain barrier penetration (ensemble AUC 0.936). An ensemble strategy combining image-based and descriptor-based predictions has consistently outperformed either approach alone. The computational pipeline has evolved from a DIGITS/Caffe/Jmol system to a TensorFlow/Keras/PyMOL framework. Limitations include endpoint-dependent parameter sensitivity, class imbalance effects, the absence of direct comparisons with graph neural networks, and an interpretability gap addressed in part by CAM-family visualization in the AI-SHIPS platform and S-COPHY. Future directions include systematic application of explainable AI methods, automated hyperparameter optimization, and integration with graph-based approaches.

Abstract: The development of new chronobiotics, substances capable of selectively modulating the parameters of circadian rhythms, is hampered by the fragmented nature and limited volume of available experimental data.In the present study, a comprehensive evaluation of the applicability of the SMILES-Transformer architecture to the classification of circadian rhythm modulators was performed using the specialised ChronobioticsDB resource, and the first systematic virtual screening of the SAVI (Synthetically Accessible Virtual Inventory) library of synthetically accessible compounds for chronobiotic activity was carried out. Rigorous protocols were applied for model training and validation: Data-Efficient Modeling (DEM) assessment with 20 repeats, repeated scaffold validation (5 × 5), and a comparative analysis of training strategies (feature-based vs. end-to-end fine-tuning). The influence of three variants of circadian-effect labelling (raw, aggregated, and expert-curated) and three loss functions (BCE, Focal Loss, and Asymmetric Loss) on the quality of multi-label classification was investigated. The results demonstrate that systematic hyperparameter optimisation in end-to-end mode provides the best predictive performance (ROC-AUC 0.666 for the effect_coarse task), whereas standard fine-tuning without optimisation leads to overfitting (ROC-AUC 0.470). Scaffold validation confirmed the ability of the model to generalise to structurally novel compounds (ROC-AUC 0.587). Expert aggregation of labels improved the recognition of rare classes (F1-macro 0.254 versus 0.148 for the raw labelling). Based on the trained models, a consensus virtual screening of the SAVI library was performed using four independent classifiers (classf, effect_coarse, target, mechanism). From more than five million compounds, 10,000 of the most promising candidates were selected, among which 34 super-candidates (consensus score > 0.9) and 435 strong candidates (> 0.8) were identified. Analysis of the predicted targets revealed dominance of the CLOCK-BMAL1 complex (60.49%), while among effects the circadian phase shift prevailed (37%). All identified candidates are synthetically accessible and are recommended for prioritised experimental verification.

Abstract: Oleoresins are complex natural lipophilic matrices traditionally analyzed using chromatographic techniques that require extensive sample preparation, derivatization, and authentic standards. Amazonian oleoresins from Copaifera and Eperua species (Fabaceae) represent valuable bioresources with recognized pharmacological potential, largely attributed to diterpenoids such as copalic and hardwickiic acids, as well as bioactive sesquiterpenes, including the cannabinoid b-caryophyllene. In this study, we present a proof-of-concept application of Direct Analysis in Real Time coupled with High-Resolution Mass Spectrometry (DART-HRMS) as a rapid, direct, and environmentally friendly approach for chemical fingerprinting and semi-targeted screening of the two most important amazonian oleoresins from these two genera: Eperua oleifera and Copaifera multijuga. Analyses were performed using a Q Exactive Orbitrap coupled to a DART ion source under after conditions optimization. Hardwickiic acid was used as a model compound for method optimization, with optimal performance achieved at 200 °C and 100 V, yielding stable signal intensities (CV &lt; 10%) and high mass accuracy (&lt; 1 ppm). The method enabled reproducible detection of diterpenic acids in both oleoresins, allowing differentiation of their chemical profiles and assessment of short-term stability under ambient conditions. In addition to diterpenes, free fatty acids were also detected, expanding the compositional characterization of these matrices. Compound annotation was performed based on accurate mass measurements and literature comparison, corresponding to Level 5 confidence according to established metabolomics criteria. Although the absence of chromatographic separation limits isomer discrimination and absolute quantification, DART-HRMS provides a rapid and solvent-free strategy for chemical fingerprinting and preliminary characterization of oleoresins. This approach aligns with Green Chemistry principles and shows strong potential as a screening and triage tool for quality control, chemotaxonomic studies, and sustainable valorization of Amazonian natural products.

Abstract: Oleanolic acid (OA) is a hydrophobic pentacyclic triterpene widely distributed in the plant kingdom and characterized by broad biological activity, including antioxidant, anti-inflammatory, neuroprotective, renoprotective, and anticancer effects. Increasing evidence suggests, however, that many of these actions are better explained not by single molecular targets, but by OA-dependent modulation of an integrated organelle stress network involving mitochondria, the endoplasmic reticulum (ER), autophagy, mitophagy, and apoptosis. This review critically analyzes the available evidence on the effects of OA on the mitochondria–ER–autophagy–apoptosis axis, with particular emphasis on mechanisms governing the transition between cellular adaptation and cell death. The available literature indicates that, in non-cancer models, OA most commonly lowers reactive oxygen species (ROS), stabilizes mitochondrial function, attenuates the ER stress signature, and promotes adaptive autophagy and mitophagy. In contrast, in many cancer models, OA may enhance mitochondrial dysfunction, lower the threshold for mitochondrial apoptosis, and induce autophagy that can be either protective or cytotoxic depending on the biological context. Overall, the current evidence supports a model in which OA acts as a context-dependent modulator of the organelle stress threshold rather than as a uniformly cytoprotective or uniformly proapoptotic compound. At the same time, the literature remains heterogeneous with respect to models, doses, exposure times, and markers used, while poor aqueous solubility and limited bioavailability continue to constrain translation. Future studies should therefore integrate analyses of mitochondria, ER, mitochondria–ER contact sites (MERCS), autophagy, apoptosis, pharmacokinetics, formulation, and safety in order to define the true potential of OA as a modulator of biological stress.