ACS Central Science最新文献

Generative machine learning models for exploring chemical space have shown immense promise, but many molecules that they generate are too difficult to synthesize, making them impractical for further investigation or development. In this work, we present a novel approach by fine-tuning Meta’s Llama3 Large Language Models (LLMs) to create SynLlama, which generates full synthetic pathways made of commonly accessible building blocks and robust organic reaction templates. SynLlama explores a large synthesizable space using significantly less data and offers strong performance in both forward and bottom-up synthesis planning compared to other state-of-the-art methods. We find that SynLlama, even without training on external building blocks, can effectively generalize to unseen yet purchasable building blocks, meaning that its reconstruction capabilities extend to a broader synthesizable chemical space than those of the training data. We also demonstrate the use of SynLlama in a pharmaceutical context for synthesis planning of analog molecules and hit expansion leads for proposed inhibitors of target proteins, offering medicinal chemists a valuable tool for discovery.

Fine-tuning on synthetic reactions from commercial building blocks and high-fidelity reactions creates a versatile LLM, SynLlama, for key drug discovery tasks.

Tumor resistance to immune checkpoint blockade (ICB) therapy is frequently driven by adaptive metabolic reprogramming in the hypoxic tumor microenvironment (TME). The key N4-acetylcytidine (ac4C) RNA modification mediator N-acetyltransferase 10 (NAT10) emerges as a promising therapeutic target, despite the lack of potent targeting agents. Here, we engineered NP1192, a PROTAC degrader targeting NAT10. NP1192 achieved nearly 70% NAT10 degradation and a 26.8% lower IC₅₀ than canonical NAT10 inhibitor Remodelin in cervical cancer cells, outperforming Remodelin in antitumor effect in vivo, in vitro, and across three tumor organoids. It abrogated ac4C modifications on HIF1A mRNA and translation, reducing hypoxic lactate production and depleted ATP, and suppressed HIF-1α-mediated PD-L1 upregulation. In xenograft models, NP1192 combined with anti-PD-L1 inhibited subcutaneous xenograft growth and reduced tumor-core lactate gradients by > 80%. Furthermore, scRNA-seq and in vitro coculture experiments identified expansion of IFN-γ⁺ effector CD8⁺ T cells (T_eff) and decline in exhausted CD8⁺ T cells (T_ex). NP1192 in combination with anti-PD-L1 enhanced proliferation and effector function of CD8⁺ T_eff cells, thereby reversing resistance to anti-PD-L1 blockade therapy and synergizing with immunotherapy. These findings establish PROTAC-mediated NAT10 degradation as a dual metabolic-immune strategy to enhance checkpoint blockade efficacy.

PROTAC NP1192 degrades NAT10 inhibiting HIF-1α via ac4C modification. It reduces hypoxic glycolysis, enhances CD8⁺ T_eff function, and synergizes with anti-PD-L1 to reverse immunotherapy resistance.

The efficacy of cancer immunotherapy is often limited by the immunosuppressive tumor microenvironment (TME) and insufficient immune activation in tumor-draining lymph nodes (TDLN). Since the TME and TDLN form a dynamic axis crucial for tumor metastasis and resistance to immune checkpoint blockade, strategies that effectively modulate both sites are critical. Here, we present a dissolving microneedle (MN) system that generates nanomicelles (NMCs) for localized delivery of a newly identified dual-functional macrocyclic trichothecene, Roridin E (R.E). R.E induces cancer cell-autonomous secretion of IFN-β and immunogenic cancer cell death (ICD). Direct delivery of R.E to the TDLN via the MN platform reshapes the local immune landscape to suppress cancer while minimizing off-target toxicity. In a B16F10 melanoma model, MN-guided R.E. delivery significantly improved tumor control, reduced lung metastases, and extended overall survival. This approach provides a minimally invasive and effective strategy for integrating natural-product-based therapies with advanced drug delivery systems to target the TME–TDLN axis, thereby improving outcomes in metastatic cancer.

Microneedle-directed delivery of a dual-functional immune modulator reprograms tumor−lymph node immunity to enhance cancer immunotherapy.

Machine learning (ML) is rapidly turning into a key technology for biocatalysis. By learning patterns in amino acid sequences, protein structures, and functional data, ML models can help navigate complex fitness landscapes, uncover new enzymes in databases, and even design biocatalysts de novo. Along with advances in DNA synthesis and sequencing, laboratory automation, and high-throughput screening, ML is increasing the speed and efficiency of enzyme development. In this Outlook, we highlight recent applications of ML in the fields of enzyme discovery, design, and engineering, with a focus on current challenges and emerging solutions. Furthermore, we discuss barriers that impede a broader and faster adoption of ML-based workflows in the biocatalysis community. We conclude by suggesting best practices for fostering effective collaborations in this interdisciplinary field.

We highlight how machine learning accelerates enzyme discovery, design, and engineering, outlining recent advances, key challenges, and emerging opportunities in biocatalysis.

The extent of charge transfer between an adsorbate and thermocatalytic surface plays a key role in determining catalytic activity, but direct and quantitative measures have remained elusive. Here, we report the method of isopotential electron titration (IET), an approach that directly measures charge transfer between adsorbates and catalytic surfaces. Charge transfer between Pt and adsorbed hydrogen adatoms was investigated using a catalytic condenser, where the Pt surface was separated from a p-type silicon layer by a hafnia dielectric film. By forcing the Pt and Si layers into isopotential conditions, charge transfer between the adsorbate and Pt surface was titrated through an external circuit. Hydrogen atoms donated electrons to Pt upon adsorption, which was quantitatively reversed upon desorption. Across a temperature range of 125–200 °C (surface hydrogen fractional coverages of 80–100%), the charge transferred to Pt by an adsorbed hydrogen atom was measured to be 0.19 ± 0.01% |e|/H. Bader charge analysis of the extent of charge transfer was in agreement with experimental measurements, with a calculated net donation of 0.4% |e|/H. The ability to experimentally quantify surface charge transfer provides an electronic-based approach to characterize catalytic surfaces, the adsorbed moieties residing on them, and the chemical reactions they accelerate.

The electrostatic generosity of hydrogen to metals: Isopotential electron titration measures the nonfaradaic charge transfer between adsorbates and catalytic surfaces

The site-selective C–H functionalization to install meta-C–S bonds on aniline derivatives is highly desirable, due to the preponderance of resulting compounds in numerous medicinally relevant compounds. However, the execution of the same is far from being trivial, due to the intrinsic electronic bias of anilines and concerns associated with the ready availability of an appropriate and odorless sulfur source. Accordingly, we demonstrate a metal- and additive-free, one-pot, multicomponent reaction between p-anisidines/anilines, carbon disulfide, and aliphatic amines to install an otherwise difficult meta-C–S bond on anilines with exclusive regioselectivity, while furnishing an array of biologically relevant anisidine-derived S-aryl dithiocarbamates. The method exhibits broad scope with appreciable functional group tolerance, as demonstrated through late-stage modification of a variety of amino acids, pharmaceuticals, and natural products. Importantly, final S-aryl dithiocarbamates are amenable to further synthetic manipulations, furnishing highly valuable and medicinally relevant sulfur-containing functional moieties, such as thiols, thioethers, and sulfones. Furthermore, in vitro evaluations demonstrate that many of the synthesized dithiocarbamates exhibit promising drug-like properties, demonstrating antiproliferative activity on a nanomolar level for breast cancer cell lines by affecting microtubule dynamics.

We report a metal-free multicomponent reaction that constructs meta-C−S bonds on anilines to afford S-aryl dithiocarbamates, featuring broad scope, facile diversification, and potent anticancer activity.

Inspired by biological rebound processes, radical ligand transfer (RLT) has emerged as a powerful and versatile strategy for the selective functionalization of alkyl radicals. RLT enables direct C–X bond formation through homolytic substitution at a metal-bound ligand (M–X) and demonstrates broad functional group tolerance and high potential for catalysis. Despite growing interest and mechanistic understanding, including recent insights into asynchronous concerted ion–electron transfer (cIET), the broader application of RLT strategies remains underdeveloped. In parallel, the closely related S_H2 (bimolecular homolytic substitution) mechanism has gained increasing utility in C–C bond formation, where low-valent metals capture transient radicals and facilitate selective coupling with persistent radical partners─a process referred to as radical sorting. Herein, we present a comprehensive perspective of the evolving landscape of RLT and S_H2 chemistry, emphasizing recent advances. We highlight key bioinspired and computationally guided approaches that have enhanced mechanistic understanding and broadened the substrate scope, including landmark contributions by Kochi, Groves, Shaik, MacMillan, and others. To complement these studies and encourage further development, we also report DFT-based thermodynamic analyses of radical ligand transfer across first-row transition metal complexes bearing porphyrin and BOX ligands. By unifying these mechanistic frameworks, this perspective aims to provide a roadmap for designing novel, selective, and sustainable radical-based transformations.

Radical ligand transfer (RLT) and bimolecular homolytic substitution (S_H2), employed in synergy with other catalytic platforms, enable selective C−X and C−C bond formation through radical strategies.

More drugs are entering aquatic habitats. Scientists are teasing apart how they influence the behavior, reproduction, and biology of organisms that live there.

Covalent ligands contain an electrophilic moiety that reacts with a nucleophilic residue on a target protein, following an initial reversible binding event. Covalent ligand development typically involves efforts to increase on-target selectivity by maximizing the ligand binding affinity and minimizing intrinsic electrophile reactivity. Problematically, this limits labeling kinetics and requires high affinity ligands. The concept of “latency” describes the potential for “turn-on” activation of electrophiles upon target engagement. Here, we investigate the potential intrinsic latency of covalent electrophiles and test the hypothesis that diverse electrophiles can be differentially activated by proximity effects. We develop a kinetic effective molarity (EM_k) approach to quantitatively characterize kinetics associated with diverse electrophilic reaction mechanisms, both with and without binding proximity effects. We observe that different electrophiles are associated with significantly different EM_k parameters, with SuFEx and acrylamide electrophiles associated with the highest intrinsic latency. Eyring transition state analysis revealed that all covalent ligands, independent of electrophile, benefit from significant transition state entropic stabilization. Strikingly, electrophiles associated with the highest latency are associated with greater relative transition state stabilization with different enthalpic and entropic contributions. These findings quantitatively describe electrophile latency and will aid the mechanism-guided development of next-generation covalent ligands associated with “turn-on” reactivity.

A kinetic effective molarity analysis reveals mechanism-dependent differences in proximity-induced reactivity of covalent electrophiles.