We describe a fully data driven model that learns to perform a retrosynthetic reaction prediction task, which is treated as a sequence-to-sequence mapping problem. The end-to-end trained model has an encoder–decoder architecture that consists of two recurrent neural networks, which has previously shown great success in solving other sequence-to-sequence prediction tasks such as machine translation. The model is trained on 50,000 experimental reaction examples from the United States patent literature, which span 10 broad reaction types that are commonly used by medicinal chemists. We find that our model performs comparably with a rule-based expert system baseline model, and also overcomes certain limitations associated with rule-based expert systems and with any machine learning approach that contains a rule-based expert system component. Our model provides an important first step toward solving the challenging problem of computational retrosynthetic analysis.

New strategies are urgently needed to target MRSA, a major global health problem and the leading cause of mortality from antibiotic-resistant infections in many countries. Here, we report a general approach to this problem exemplified by the design and synthesis of a vancomycin–D-octaarginine conjugate (V–r8) and investigation of its efficacy in addressing antibiotic-insensitive bacterial populations. V–r8 eradicated MRSA biofilm and persister cells in vitro, outperforming vancomycin by orders of magnitude. It also eliminated 97% of biofilm-associated MRSA in a murine wound infection model and displayed no acute dermal toxicity. This new dual-function conjugate displays enhanced cellular accumulation and membrane perturbation as compared to vancomycin. Based on its rapid and potent activity against biofilm and persister cells, V–r8 is a promising agent against clinical MRSA infections.

The emergence of multidrug-resistant Gram-negative bacteria, including carbapenem-resistant Enterobacteriaceae, is a major health problem that necessitates the development of new antibiotics. Vancomycin inhibits cell-wall synthesis in Gram-positive bacteria, but is generally ineffective against Gram-negative bacteria and unable to penetrate the outer membrane barrier. In an effort to determine whether vancomycin and other antibiotics effective against Gram-positive bacteria could, through modification, be rendered effective against Gram-negative bacteria, we discovered that covalent attachment of a single arginine to vancomycin yielded conjugates with order-ofmagnitude improvements in activity against Gram-negative bacteria, including pathogenic E. coli. The vancomycin-arginine conjugate (V-R) exhibited efficacy against actively-growing bacteria, induced loss of rod cellular morphology, and resulted in intracellular accumulation of peptidoglycan precursors, all consistent with cell-wall synthesis disruption as its mechanism of action. Membrane permeabilization studies demonstrated enhanced outer membrane permeability of V-R as compared to vancomycin. The conjugate exhibited no mammalian cell toxicity or hemolytic activity in MTT and hemolysis assays. Our study introduces a new vancomycin derivative effective against Gram-negative bacteria and underscores the broader potential of generating new antibiotics through combined mode-of-action and synthesis-informed design studies.

Bryostatin 1 is an exceedingly scarce marine-derived natural product that is in clinical development directed at HIV/AIDS eradication, cancer immunotherapy, and the treatment of Alzheimer’s disease. Despite this unique portfolio of indications, its availability has been limited and variable, thus impeding research and clinical studies. Here, we report a total synthesis of bryostatin 1 that proceeds in 29 total steps (19 in the longest linear sequence, >80% average yield per step), collectively produces grams of material, and can be scaled to meet clinical needs (~20 grams per year). This practical solution to the bryostatin supply problem also opens broad, facile, and efficient access to derivatives and potentially superior analogs.

Circular RNAs (circRNAs) are stable and prevalent RNAs in eukaryotic cells that arise from back-splicing. Synthetic circRNAs and some endogenous circRNAs can encode proteins, raising the promise of circRNA as a platform for gene expression. In this study, we developed a systematic approach for rapid assembly and testing of features that affect protein production from synthetic circRNAs. To maximize circRNA translation, we optimized five elements: vector topology, 5′ and 3′ untranslated regions, internal ribosome entry sites and synthetic aptamers recruiting translation initiation machinery. Together, these design principles improve circRNA protein yields by several hundred-fold, provide increased translation over messenger RNA in vitro, provide more durable translation in vivo and are generalizable across multiple transgenes.