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.

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.

AIDS is a pandemic disease caused by HIV that affects 37 million people worldwide. Current antiretroviral therapy slows disease progression but does not eliminate latently infected cells, which resupply active virus, thus necessitating lifelong treatment with associated compliance, cost, and chemoexposure issues. Latency-reversing agents (LRAs) activate these cells, allowing for their potential clearance, thus presenting a strategy to eradicate the infection. Protein kinase C (PKC) modulators—including prostratin, ingenol esters, bryostatin, and their analogs—are potent LRAs in various stages of development for several clinical indications. While LRAs are promising, a major challenge associated with their clinical use is sustaining therapeutically meaningful levels of the active agent while minimizing side effects. Here we describe a strategy to address this problem based on LRA prodrugs, designed for controllable release of the active LRA after a single injection. As intended, these prodrugs exhibit comparable or superior in vitro activity relative to the parent compounds. Selected compounds induced higher in vivo expression of CD69, an activation biomarker, and, by releasing free agent over time, significantly improved tolerability when compared to the parent LRAs. More generally, selected prodrugs of PKC modulators avoid the bolus toxicities of the parent drug and exhibit greater efficacy and expanded tolerability, thereby addressing a longstanding objective for many clinical applications.

Bryostatin 1 is a marine natural product under investigation for HIV/AIDS eradication, the treatment of neurological disorders, and enhanced CAR T/NK cell immunotherapy. Despite its promising activity, bryostatin 1 is neither evolved nor optimized for the treatment of human disease. Here we report the design, synthesis, and biological evaluation of several close-in analogs of bryostatin 1. Using a function-oriented synthesis approach, we synthesize a series of bryostatin analogs designed to maintain affinity for bryostatin's target protein kinase C (PKC) while enabling exploration of their divergent biological functions. Our latestage diversification strategy provides efficient access to a library of bryostatin analogs, which per our design retain affinity for PKC but exhibit variable PKC translocation kinetics. We further demonstrate that select analogs potently increase cell surface expression of CD22, a promising CAR T cell target for the treatment of leukemias, highlighting the clinical potential of bryostatin analogs for enhancing targeted immunotherapies.

RNA oligonucleotides containing a phenyl selenide derivative of 5-methyluridine were chemically synthesized by solid-phase synthesis. The phenyl selenide is rapidly converted to an electrophilic, allylic phenyl seleneate under mild oxidative conditions. The phenyl seleneate yields interstrand cross-links when part of a duplex and is useful for synthesizing oligonucleotide conjugates. Formation of the latter is illustrated by reaction of an oligonucleotide containing the phenyl selenide with amino acids in the presence of mild oxidant. The products formed are analogous to those observed in tRNA that are believed to be formed posttranslationally via a biosynthetic intermediate that is chemically homologous to the phenyl seleneate.