Natural metalloenzymes harness the sophisticated synergy between canonical amino acids and metal cofactors to catalyze life-sustaining reactions. However, the limited chemical diversity of the twenty standard amino acids often restricts non-natural transformations. To overcome these constraints, artificial metalloenzymes (ArMs) integrate synthetic cofactors into protein scaffolds to enable unnatural catalytic functions. This review highlights how noncanonical amino acids (ncAAs) bridge the gap between natural enzyme architecture and synthetic chemistry. We examine genetic code expansion and codon optimization technologies as robust strategies for the site-specific incorporation of ncAAs. By introducing unique chemical moieties—such as chelating groups, photo-responsive elements, and bio-orthogonal handles—ncAAs enable precise mechanistic probing of active sites and the rational design of biomimetic centers. Furthermore, we discuss the strategic use of ncAAs in anchoring metal ions and large-sized synthetic cofactors via bio-orthogonal reactions, allowing for the creation of hybrid catalysts that maintain structural integrity while exhibiting novel reactivity. Ultimately, the combinatorial design of ncAAs and synthetic cofactors provides a transformative toolkit for protein engineering. Integrating these structural elements transcends natural limitations, laying a foundation for ArMs with enhanced catalytic efficiency, expanded substrate scope and programmable activity for industrial and therapeutic applications.
A key challenge in drug development is identification of druggable targets, the modulation of which attenuates disease progression, while avoiding inhibition of proteins that lead to dose-limiting toxicities. Here, we investigate a drug target, casein kinase 2 (CK2) - a serine/threonine kinase implicated in multiple cancers - for which existing inhibitors have so far failed in clinical trials. Using molecular and pharmacoepidemiology approaches, we show that small molecules targeting cyclin-dependent kinase (CDK) family members CDK1/2/7/9, including the existing CK2 inhibitors, have a higher risk to induce adverse effects or failing in clinical trials. We then establish a machine learning (ML)-assisted discovery pipeline to redesign more specific and allosteric lead compounds against CK2, with more selective on-target binding and favourable off-target profile. Importantly, we show that such selective design is possible when standard molecular docking, CDK-liability filtering and ML algorithms are combined with an error prediction model. Taken together, we report a simple yet efficient ML-powered drug discovery pipeline and novel submicromolar inhibitors targeting the clinically relevant CK2 kinase with no clinically approved antagonists currently available. Our prediction pipeline was able to achieve a 90% hit-rate, significantly reducing the need for subsequent wet-lab validation.
Recent deep learning-based methods have advanced chemical retrosynthesis. However, they neglect side reactions, strong competitors to desired synthesis reactions that reduce the yield of anticipated product molecules. To identify side-reaction sites, we introduce an atom representation learning framework, SideReact, which explores synthesis reactions from a top-down chemical perspective (i.e., from reaction to motif, to property, and to quantum w.r.t. atoms). Its multi-tier atom encoder (MtAE) extracts top-down observations as atom representations, including reaction paradigms, motif reactivities and co-reactivities, atomic identities and bonding circumstances, and atoms' electron-attracting abilities. Moreover, its hierarchical cross-attention module (HCA) captures the chemical hierarchy among encoding tiers in a bottom-up way. Especially, HCA enables an interpretable path through the hierarchy to reveal how to find reaction sites among atoms in reactants. The comparison with ten state-of-the-art methods demonstrates the superiority of SideReact in both side reactions and desired reactions. Furthermore, a real case study, including ten synthetic examples in textbooks, a complex synthetic reaction of Atorvastatin with side-reaction sites, and a synthetic reaction for a drug-like molecule, substantiates the effectiveness of SideReact. More importantly, analyses demonstrate how its MtAE works and how its HCA unveils the interpretable inference of side-reaction sites, highlighting its potential in retrosynthesis planning.
The complex formed between the HIV-1 protein Rev and the Rev Response Element (RRE) in the genomic RNA of the virus represents an attractive target for antiretroviral therapy. To date, however, there are no RRE-Rev inhibitors in clinical use. In this study we have applied a fluorescence resonance energy transfer (FRET) assay based on placing Cy3- and Cy5-labeled oligomeric probes at distal domain locations within the RRE to identify small-molecule compounds capable of perturbing the spatial organization of RRE domains and disrupting the association of Rev monomers. After screening a library of bioactive compounds and approved drugs, we identified two molecules, myricetin and raloxifene, that recognize the RRE and inhibit the RRE-Rev interaction in vitro and in cellulo. Both compounds exhibited antiretroviral activity with favorable selectivity indexes, and further analyses involving RRE mutants suggested that myricetin modulated the conformation of the RRE through binding to a four-helix intersection. Collectively, these findings provide proof of concept for targeting the RRE-Rev complex and identify myricetin and raloxifene as promising starting points for the development of novel anti-HIV agents.
Laboratory automation has benefited substantially from well-defined protocols and machine-readable representations of chemical procedures. Yet a long tail of context-dependent physical tasks, including setup, adjustment, and cleanup, continues to require manual intervention or bespoke robotic implementations tailored to individual workflows and platforms. This study examines whether this apparent diversity can be organized into a structured and reusable action space. TARMAC (TAxonomy for Robot MAnipulation in Chemistry), an empirical taxonomy of laboratory actions, is derived through a bottom-up analysis of instructional chemistry practice. The taxonomy comprises 18 physically meaningful primitives organized by three dimensions: wrench dependence, actuation directness, and motion periodicity. Evaluation through cross-corpus protocol decomposition and robotic instantiation addresses expressive coverage and proof-of-concept executability. Across 23 procedures, 904 of 917 manipulative steps map to the primitive set, while robotic tests instantiate 8 primitives in five tasks across four robotic configurations. Together, these results indicate that laboratory manipulation can be organized as a structured action space across the examined contexts. TARMAC therefore provides a descriptive basis for organizing and comparing manipulation requirements across experimental workflows and robotic embodiments. Laboratory automation has made well-defined experimental protocols increasingly executable by machines, yet many physical manipulations remain difficult to generalize, including both those surrounding automated protocols and those not readily accommodated by current automation. Here, the authors propose TARMAC, a taxonomy that organizes these actions into reusable primitives, demonstrating its applicability across various platforms and workflows.
