Davide Zeppilli

We designed a supervised machine learning framework to predict standard Gibbs free energies, ΔG°, of formal hydrogen atom transfer (f-HAT) for phenolic antioxidants across different radicals and media, enabling rapid and chemically interpretable screening. We curated a DFT dataset of 71 molecules (phenolic compounds and anthocyanidins), with 207 reaction sites, 10 radical reactive oxygen/sulfur species, and three environments (leading to a total of 6210 ΔG° values). The models amass 106 numerical RDKit descriptors, augmented with one-hot encodings of medium, site, radical, and structural class, and were evaluated through a leave-one-molecule-out protocol. Among the tested regression algorithms, the random forest regressor provides the best balance of accuracy and robustness with both R² test (≈0.94) and MAE (2.74 kcal mol⁻¹; RMSE (≈5.0 kcal mol⁻¹)), close to DFT chemical accuracy. The feature-importance analysis revealed that “electronic” and “experimental” (site/group) descriptors primarily drive predictions, with the radical’s maximum absolute partial charge being the most important descriptor in the prediction of a radical’s ΔG°. These results suggest that descriptor-driven RF (Random Forest) models can generalize across chemical space to provide interpretable ΔG° predictions, providing a path for chemists towards a scalable route to prioritize antioxidant candidates for broader molecular families.

In this work, we employ a systematic quantitative computational approach to study the scavenging activity via formal hydrogen atom transfer (f-HAT) of a series of phenolic compounds. The most common features present in natural polyphenols are included to analyze their effect in the reactivity with radicals. All molecules are classified according to the analyzed effect: aromaticity size (class A), number and position of multiple hydroxyl (OH) groups (class B), and type and position of various substituents inspired by natural systems (classes C–E). The analysis also includes the corresponding anions for systems with pK_a values that permit deprotonation under physiological conditions. Finally, all the identified criteria for selecting effective polyphenolic antioxidants are validated in real systems, such as anthocyanidins, providing insightful tools for drug design.

Recent advancements in electro-photoredox catalysis (e-PRC) and consecutive photoinduced electron transfer (conPET) have pushed the energy limits of conventional photocatalysis. Both methods produce open-shell intermediate catalysts that, upon light absorption, become highly reducing or oxidizing, enabling challenging reactions. Despite their widespread use, the mechanisms of e-PRC and conPET reactions remain debated, in part due to a lack of quantitative data in most studiesparticularly single-electron transfer rate constants (k _SET) between excited-state catalysts and substrates. We present a straightforward electrochemical method for determining k _SET using cyclic voltammetry (CV) under light irradiation, paired with electrochemical simulation. Using inexpensive LEDs and standard potentiostats, we investigated the reactivity of excited-state anions of a perylene diimide dye (PDI), the seminal catalyst of conPET reactions. CV was used to study the photochemical reactivity of both reduced species of PDI, *PDI^•– and *PDI^2–, in the reductive cleavage of carbon–halogen bonds in alkyl and aryl halides. The extreme reactivity of these excited-state anions is confirmed, with quenching rate constants of 10⁷ and 10¹⁰ M^–1 s^–1 for *PDI^•– and *PDI^2–, respectively, consistent with theoretical and experimental data. The voltammetric approach presented here provides a rapid and reliable tool for studying the excited-state reactivity of labile intermediates utilized in e-PRC and conPET systems, including both radical anions and dianions.

Phenolic compounds are important antioxidants with great ROS scavenging potential and the presence of the hydroxyl groups is fundamental for this chemical activity. Therefore, changing the chalcogen atom (oxygen) with any of its siblings of group 16 (sulfur, selenium and tellurium) may affect the reactivity of these compounds. In this work, the ROS scavenging activity and mechanism of phenyl chalcogenols was evaluated in silico, unravelling better performance with heavier chalcogens, both thermodynamically and kinetically. Furthermore, a scavenging mechanism switch is reported, moving from Concerted Proton Electron Transfer (CPET) in phenols to Hydrogen Atom Transfer (HAT) in the other phenyl chalcogenols. Both kinetic trends and mechanistic features are rationalized in the framework of Activation Strain Analysis (ASA). Lastly, the role of aromaticity is evidenced by analyzing the differences between the phenol/phenoxyl and methanol/methoxyl self-exchange reactions, as well as between the corresponding processes with the other chalcogens.

The combination of intrinsic bond orbital and activation strain analysis allows the correct identification and rationalization of the operative mechanism between HAT and CPET, while the singly occupied molecular orbital does not fit in this picture.

Olanzapine is an antipsychotic drug that has been reported to suppress ferroptosis, a recently discovered form of regulated cell death. In this work, the scavenging activity of olanzapine and some of its metabolites is investigated in silico using state-of-the-art density functional theory calculations (level of theory: (SMD)-M06-2X/6-311+G(d,p)//M06-2X/6-31G(d)). Indeed, this reactivity is linked to the therapeutic activity of many antipsychotic drugs and ferroptosis inhibitors. Furthermore, the distinction between hydrogen atom transfer (HAT) and concerted proton coupled electron transfer (cPCET) is elucidated for the most reactive sites of the studied molecules. Then, a promising experimentally guided anti-ferroptotic cyclic mechanism is proposed for ferrostatin-1, a well-known ferroptosis inhibitor, involving the oxidation of Fe^II to Fe^III, the quenching of hydroperoxyl radicals, and the subsequent regeneration of the reactant (level of theory: M06/6-311+G(d,p),def2TZVP//M06/6-31G(d),LANL2DZ). An analogous cyclic process is investigated for liproxstatin-1 and olanzapine, whose activity has been reported in the literature and compared to ferrostatin-1. Finally, the effect of water solvation is evaluated unveiling that the anti-ferroptotic activity of olanzapine is likely less efficient in polar media.

The reactivity of Zn²⁺ tetrahedral complexes with H₂O₂ was investigated in silico, as a first step in their disruption process. The substrates were chosen to represent the cores of three different zinc finger protein motifs, i. e., a Zn²⁺ ion coordinated to four cysteines (CCCC), to three cysteines and one histidine (CCCH), and to two cysteines and two histidines (CCHH). The cysteine and histidine ligands were further simplified to methyl thiolate and imidazole, respectively. H₂O₂ was chosen as an oxidizing agent due to its biological role as a metabolic product and species involved in signaling processes. The mechanism of oxidation of a coordinated cysteinate to sulfenate‐κS and the trends for the different substrates were rationalized through activation strain analysis and energy decomposition analysis in the framework of scalar relativistic Density Functional Theory (DFT) calculations at ZORA‐M06/TZ2P ae // ZORA‐BLYP‐D3(BJ)/TZ2P. CCCC is oxidized most easily, an outcome explained considering both electrostatic and orbital interactions. The isomerization to sulfenate‐κO was attempted to assess whether this step may affect the ligand dissociation; however, it was found to introduce a kinetic barrier without improving the energetics of the dissociation. Lastly, ligand exchange with free thiolates and selenolates was investigated as a trigger for ligand dissociation, possibly leading to metal ejection; molecular docking simulations also support this hypothesis.

PhSeZnCl, which is also known as Santi's reagent, can catalyze the reduction of hydrogen peroxide by thiols with a GPx-like mechanism. In this work, the first step of this catalytic cycle, i.e., the reduction of H₂O₂ by PhSeZnCl, is investigated in silico using state-of-the-art density functional theory calculations. Then, the role of the metal is evaluated by replacing Zn with its group 12 siblings (Cd and Hg). The thermodynamic and kinetic factors favoring Zn are elucidated. Furthermore, the role of the halogen is considered by replacing Cl with Br in all three metal compounds, and this turns out to be negligible. Finally, the overall GPx-like mechanism of PhSeZnCl and PhSeZnBr is discussed by evaluating the energetics of the mechanistic path leading to the disulfide product.

We set up an in silico experiment and designed a chimeric compound integrating molecular features from different efficient ROS (Reactive Oxygen Species) scavengers, with the purpose of investigating potential relationships between molecular structure and antioxidant activity. Furthermore, a selenium centre was inserted due to its known capacity to reduce hydroperoxides, acting as a molecular mimic of glutathione peroxidase; finally, since this organoselenide is a precursor of a N-heterocyclic carbene ligand, its Au(I) carbene complex was designed and examined. A validated protocol based on DFT (Density Functional Theory) was employed to investigate the radical scavenging activity of available sites on the organoselenide precursor ((SMD)-M06-2X/6-311+G(d,p)//M06-2X/6-31G(d)), as well as on the organometallic complex ((SMD)-M06-2X/SDD (Au), 6-311+G(d,p)//ZORA-BLYP-D3(BJ)/TZ2P), considering HAT (Hydrogen Atom Transfer) and RAF (Radical Adduct Formation) regarding five different radicals. The results of this case study suggest that the antioxidant potential of chemical motifs should not be considered as an additive property when designing a chimeric compound, but rather that the relevance of a molecular topology is derived from a chemical motif combined with an opportune chemical space of the molecule. Thus, the direct contributions of single functional groups which are generally thought of as antioxidants per se do not guarantee the efficient radical scavenging potential of a molecular species.

Due to its endogenously high oxygen consumption, the central nervous system (CNS) is vulnerable to oxidative stress conditions. Notably, the activity of several CNS-targeting compounds, such as antidepressant and hypnotic drugs, or endogenous mediators, such as melatonin, is indeed linked to their ability of mitigating oxidative stress. In this work, we report the synthesis of two organoselenium compounds of which the structure was inspired by CNS-targeting psychotropic drugs (zolpidem and fluoxetine) and an endogenous mediator (melatonin). The molecules were designed with the aim of combining the ROS-scavenging properties, which were already assessed for the parent compounds, with a secondary antioxidant action, a glutathione peroxidase (GPx) mimic role empowered by the presence of selenium. The compounds were obtained through a facile three-step synthesis and were predicted by computational tools to passively permeate through the blood–brain barrier and to efficiently bind to the GABA A receptor, the macromolecular target of zolpidem. Of note, the designed synthetic pathway enables the production of several other derivatives through minor modifications of the scheme, paving the way for structure–activity relationship studies.