RLSynC: Offline-Online Reinforcement Learning for Synthon Completion.

Retrosynthesis is the process of determining the set of reactant molecules that can react to form a desired product. Semitemplate-based retrosynthesis methods, which imitate the reverse logic of synthesis reactions, first predict the reaction centers in the products and then complete the resulting synthons back into reactants. We develop a new offline-online reinforcement learning method RLSynC for synthon completion in semitemplate-based methods. RLSynC assigns one agent to each synthon, all of which complete the synthons by conducting actions step by step in a synchronized fashion. RLSynC learns the policy from both offline training episodes and online interactions, which allows RLSynC to explore new reaction spaces. RLSynC uses a standalone forward synthesis model to evaluate the likelihood of the predicted reactants in synthesizing a product and thus guides the action search. Our results demonstrate that RLSynC can outperform state-of-the-art synthon completion methods with improvements as high as 14.9%, highlighting its potential in synthesis planning.

Stereochemical Definition of the Natural Product (6R,10R,13R, 14R,16R,17R,19S,20S,21R,24S,25S,28S,30S,32R,33R,34R,36S,37S,39R)-Azaspiracid-3 by Total Synthesis and Comparative Analyses.

The previously accepted structure of the marine toxin azaspiracid-3 is revised based upon an original convergent and stereoselective total synthesis of the natural product. The development of a structural revision hypothesis, its testing, and corroboration are reported. Synthetic (6R,10R,13R,14R,16R,17R,19S,20S,21R,24S,25S,28S,30S,32R, 33R,34R,36S,37S,39R)-azaspiracid-3 chromatographically and spectroscopically matched naturally occurring azaspiracid-3, whereas the previously assigned 20R epimer did not.

Total Synthesis of (6R,10R,13R,14R,16R,17R,19S,20R,21R,24S, 25S,28S,30S,32R,33R,34R,36S,37S,39R)-Azaspiracid-3 Reveals Non-Identity with the Natural Product.

A convergent and stereoselective total synthesis of the previously assigned structure of azaspiracid-3 has been achieved by a late-stage Nozaki-Hiyama-Kishi coupling to form the C21-C22 bond with the C20 configuration unambiguously established from l-(+)-tartaric acid. Postcoupling steps involved oxidation to an ynone, modified Stryker reduction of the alkyne, global deprotection, and oxidation of the resulting C1 primary alcohol to the carboxylic acid. The synthetic product matched naturally occurring azaspiracid-3 by mass spectrometry, but differed both chromatographically and spectroscopically.

Tomatidine targets ATF4-dependent signaling and induces ferroptosis to limit pancreatic cancer progression.

Pancreatic ductal adenocarcinoma (PDAC) is an aggressive cancer with high metastasis and therapeutic resistance. Activating transcription factor 4 (ATF4), a master regulator of cellular stress, is exploited by cancer cells to survive. Prior research and data reported provide evidence that high ATF4 expression correlates with worse overall survival in PDAC. Tomatidine, a natural steroidal alkaloid, is associated with inhibition of ATF4 signaling in multiple diseases. Here, we discovered that in vitro and in vivo tomatidine treatment of PDAC cells inhibits tumor growth. Tomatidine inhibited nuclear translocation of ATF4 and reduced the transcriptional binding of ATF4 with downstream promoters. Tomatidine enhanced gemcitabine chemosensitivity in 3D ECM-hydrogels and in vivo. Tomatidine treatment was associated with induction of ferroptosis signaling validated by increased lipid peroxidation, mitochondrial biogenesis, and decreased GPX4 expression in PDAC cells. This study highlights a possible therapeutic approach utilizing a plant-derived metabolite, tomatidine, to target ATF4 activity in PDAC.

Structural and Mechanistic Bases of Viral Resistance to HIV-1 Capsid Inhibitor Lenacapavir.

Lenacapavir (LEN) is a long-acting, highly potent HIV-1 capsid (CA) inhibitor. The evolution of viral variants under the genetic pressure of LEN identified Q67H, N74D, and Q67H/N74D CA substitutions as the main resistance associated mutations (RAMs). Here, we determined high-resolution structures of CA hexamers containing these RAMs in the absence and presence of LEN. Our findings reveal that the Q67H change induces a conformational switch, which adversely affects the inhibitor binding. In the unliganded protein, the His67 side chain adopts the closed conformation by projecting into the inhibitor binding pocket and thereby creating steric hindrance with respect to LEN. Upon the inhibitor binding, the His67 side chain repositions to the open conformation that closely resembles the Gln67 side chain in the WT protein. We propose that the switch from the closed conformation to the open conformation, which is needed to accommodate LEN, accounts for the reduced inhibitor potency with respect to the Q67H CA variant. The N74D CA change results in the loss of a direct hydrogen bond and in induced electrostatic repulsions between CA and LEN. The double Q67H/N74D substitutions exhibited cumulative effects of respective single amino acid changes. An examination of LEN binding kinetics to CA hexamers revealed that Q67H and N74D CA changes adversely influenced the inhibitor binding affinity (KD) by primarily affecting the dissociation rate constant (koff). We used these structural and mechanistic findings to rationally modify LEN. The resulting analog exhibited increased potency against the Q67H/N74D viral variant. Thus, our studies provide a means for the development of second-generation inhibitors with enhanced barriers to resistance. IMPORTANCE LEN is an investigational long-acting agent for future HIV-1 treatment regimens. While ongoing clinical trials have highlighted a largely beneficial profile of LEN for the treatment of HIV-1 infected people with limited therapy options, one notable shortcoming is a relatively low barrier of viral resistance to the inhibitor. Cell culture-based viral breakthrough assays identified N74D, Q67H, and N74D/Q67H capsid changes as the main resistance associated mutations (RAMs). N74D and Q67H capsid substitutions have also emerged in clinical trials in some patients who received subcutaneous LEN. Understanding the structural basis behind viral resistance to LEN is expected to aid in the rational development of improved inhibitors with enhanced barriers to resistance. Here, we report high resolution structures of the main drug resistant capsid variants, which provide mechanistic insight into the viral resistance to LEN. We used these findings to develop an improved inhibitor, which exhibited enhanced activity against the viral Q67H/N74D capsid phenotype compared with that of parental LEN.

Identification and Optimization of a Novel HIV-1 Integrase Inhibitor.

Human immunodeficiency virus-1 (HIV-1) is the causative agent of acquired immunodeficiency syndrome (AIDS). HIV-1, like all retroviruses, stably integrates its vDNA copy into host chromatin, a process allowing for permanent infection. This essential step for HIV-1 replication is catalyzed by viral integrase (IN) and aided by cellular protein LEDGF/p75. In addition, IN is also crucial for proper virion maturation as it interacts with the viral RNA genome to ensure encapsulation of ribonucleoprotein complexes within the protective capsid core. These key functions make IN an attractive target for the development of inhibitors with various mechanisms of action. We conducted a high-throughput screen (HTS) of ∼370,000 compounds using a homogeneous time-resolved fluorescence-based assay capable of capturing diverse inhibitors targeting multifunctional IN. Our approach revealed chemical scaffolds containing diketo acid moieties similar to IN strand transfer inhibitors (INSTIs) as well as novel compounds distinct from all current IN inhibitors including INSTIs and allosteric integrase inhibitors (ALLINIs). Specifically, our HTS resulted in the discovery of compound 12, with a novel IN inhibitor scaffold amenable for chemical modification. Its more potent derivative 14e similarly inhibited catalytic activities of WT and mutant INs containing archetypical INSTI- and ALLINI-derived resistant substitutions. Further SAR-based optimization resulted in compound 22 with an antiviral EC50 of ∼58 µM and a selectivity index of >8500. Thus, our studies identified a novel small-molecule scaffold for inhibiting HIV-1 IN, which provides a promising platform for future development of potent antiviral agents to complement current HIV-1 therapies.

Structural and mechanistic bases for a potent HIV-1 capsid inhibitor.

The potent HIV-1 capsid inhibitor GS-6207 is an investigational principal component of long-acting antiretroviral therapy. We found that GS-6207 inhibits HIV-1 by stabilizing and thereby preventing functional disassembly of the capsid shell in infected cells. X-ray crystallography, cryo-electron microscopy, and hydrogen-deuterium exchange experiments revealed that GS-6207 tightly binds two adjoining capsid subunits and promotes distal intra- and inter-hexamer interactions that stabilize the curved capsid lattice. In addition, GS-6207 interferes with capsid binding to the cellular HIV-1 cofactors Nup153 and CPSF6 that mediate viral nuclear import and direct integration into gene-rich regions of chromatin. These findings elucidate structural insights into the multimodal, potent antiviral activity of GS-6207 and provide a means for rationally developing second-generation therapies.

An Isoquinoline Scaffold as a Novel Class of Allosteric HIV-1 Integrase Inhibitors.

Allosteric HIV-1 integrase inhibitors (ALLINIs) are a new class of potential antiretroviral therapies with a unique mechanism of action and drug resistance profile. To further extend this class of inhibitors via a scaffold hopping approach, we have synthesized a series of analogues possessing an isoquinoline ring system. Lead compound 6l binds in the v-shaped pocket at the IN dimer interface and is highly selective for promoting higher-order multimerization of inactive IN over inhibiting IN-LEDGF/p75 binding. Importantly, 6l potently inhibited HIV-1NL4-3 (A128T IN), which confers marked resistance to archetypal quinoline-based ALLINIs. Thermal degradation studies indicated that at elevated temperatures the acetic acid side chain of specific isoquinoline derivatives undergo decarboxylation reactions. This reactivity has implications for the synthesis of various ALLINI analogues.

HIV-1 integrase tetramers are the antiviral target of pyridine-based allosteric integrase inhibitors.

Allosteric HIV-1 integrase (IN) inhibitors (ALLINIs) are a promising new class of antiretroviral agents that disrupt proper viral maturation by inducing hyper-multimerization of IN. Here we show that lead pyridine-based ALLINI KF116 exhibits striking selectivity for IN tetramers versus lower order protein oligomers. IN structural features that are essential for its functional tetramerization and HIV-1 replication are also critically important for KF116 mediated higher-order IN multimerization. Live cell imaging of single viral particles revealed that KF116 treatment during virion production compromises the tight association of IN with capsid cores during subsequent infection of target cells. We have synthesized the highly active (-)-KF116 enantiomer, which displayed EC50 of ~7 nM against wild type HIV-1 and ~10 fold higher, sub-nM activity against a clinically relevant dolutegravir resistant mutant virus suggesting potential clinical benefits for complementing dolutegravir therapy with pyridine-based ALLINIs.

HIV-1 inserts its genetic code into human genomes, turning healthy cells into virus factories. To do this, the virus uses an enzyme called integrase. Front-line treatments against HIV-1 called "integrase strand-transfer inhibitors" stop this enzyme from working. These inhibitors have helped to revolutionize the treatment of HIV/AIDS by protecting the cells from new infections. But, the emergence of drug resistance remains a serious problem. As the virus evolves, it changes the shape of its integrase protein, substantially reducing the effectiveness of the current therapies. One way to overcome this problem is to develop other therapies that can kill the drug resistant viruses by targeting different parts of the integrase protein. It should be much harder for the virus to evolve the right combination of changes to escape two or more treatments at once. A promising class of new compounds are "allosteric integrase inhibitors". These chemical compounds target a part of the integrase enzyme that the other treatments do not yet reach. Rather than stopping the integrase enzyme from inserting the viral code into the human genome, the new inhibitors make integrase proteins clump together and prevent the formation of infectious viruses. At the moment, these compounds are still experimental. Before they are ready for use in people, researchers need to better understand how they work, and there are several open questions to answer. Integrase proteins work in groups of four and it is not clear how the new compounds make the integrases form large clumps, or what this does to the virus. Understanding this should allow scientists to develop improved versions of the drugs. To answer these questions, Koneru et al. first examined two of the new compounds. A combination of molecular analysis and computer modelling revealed how they work. The compounds link many separate groups of four integrases with each other to form larger and larger clumps, essentially a snowball effect. Live images of infected cells showed that the clumps of integrase get stuck outside of the virus's protective casing. This leaves them exposed, allowing the cell to destroy the integrase enzymes. Koneru et al. also made a new compound, called (-)-KF116. Not only was this compound able to tackle normal HIV-1, it could block viruses resistant to the other type of integrase treatment. In fact, in laboratory tests, it was 10 times more powerful against these resistant viruses. Together, these findings help to explain how allosteric integrase inhibitors work, taking scientists a step closer to bringing them into the clinic. In the future, new versions of the compounds, like (-)-KF116, could help to tackle drug resistance in HIV-1.

Synthesis of the C22-C40 Domain of the Azaspiracids.

An efficient synthesis of the C22-C40 domain of the azaspiracids is described. The synthetic route features a Nozaki-Hiyama-Kishi (NHK) coupling and chelation controlled Mukaiyama aldol reaction to access an acyclic intermediate and a double-intramolecular-hetero-Michael addition (DIHMA) to provide the FG-ring system bridged ketal.