Feature engineered embeddings for classification of molecular data.

The classification of molecules is of particular importance to the drug discovery process and several other use cases. Data in this domain can be partitioned into structural and sequence/text data. Several techniques such as deep learning are able to classify molecules and predict their functions using both types of data. Molecular structure and encoded chemical information are sufficient to classify a characteristic of a molecule. However, the use of a molecule's structural information typically requires large amounts of computational power with deep learning models that take a long time to train. In this study, we present an alternative approach to molecule classification that addresses the limitations of other techniques. This approach uses natural language processing techniques in the form of count vectorisation, term frequency-inverse document frequency, word2vec and Latent Dirichlet Allocation to feature engineer molecular text data. Through this approach, we aim to make a robust and easily reproducible embedding that is fast to implement and solely dependent on chemical (text) data such as the sequence of a protein. Further, we investigate the usefulness of these embeddings for machine learning models. We apply the techniques to two different types of molecular text data: FASTA sequence data and Simplified Molecular Input Line Entry Specification data. We show that these embeddings provide excellent performance for classification.

Non-active site mutations in the HIV protease: Diminished drug binding affinity is achieved through modulating the hydrophobic sliding mechanism.

The global HIV/AIDS epidemic still currently affects approximately 38 million individuals globally. The protease enzyme of the human immunodeficiency virus is a major drug target in antiviral therapy, however, under the influence of reverse transcriptase and in the context of drug pressure, the rapid PR mutation rate contributes significantly to clinical failure. The set of cooperative non-active site mutations, I13V/I62V/V77I, have been associated with reduced inhibitor susceptibility and are the focus of the current study. When compared to the wild-type protease the mutant protease exhibited decreased binding affinities towards ATV and DRV by 64- and 12-fold, respectively, and decreased the overall favourable Gibbs free energy for ATV, DRV, RTV and SQV. Moreover, these mutations decreased the thermal stability of the protease when in complex with ATV and DRV by approximately 6.4 and 4.2 °C, respectively. The crystal structure of the mutant protease revealed that the location of these mutations and their effect on the hydrophobic sliding mechanism may be crucial in their role in resistance.

Elasticity-Associated Functionality and Inhibition of the HIV Protease.

HIV protease plays a critical role in the life cycle of the virus through the generation of mature and infectious virions. Detailed knowledge of the structure of the enzyme and its substrate has led to the development of protease inhibitors. However, the development of resistance to all currently available protease inhibitors has contributed greatly to the decreased success of antiretroviral therapy. When therapy failure occurs, multiple mutations are found within the protease sequence starting with primary mutations, which directly impact inhibitor binding, which can also negatively impact viral fitness and replicative capacity by decreasing the binding affinity of the natural substrates to the protease. As such, secondary mutations which are located outside of the active site region accumulate to compensate for the recurrently deleterious effects of primary mutations. However, the resistance mechanism of these secondary mutations is not well understood, but what is known is that these secondary mutations contribute to resistance in one of two ways, either through increasing the energetic penalty associated with bringing the protease into the closed conformation, or, through decreasing the stability of the protein/drug complex in a manner that increases the dissociation rate of the drug, leading to diminished inhibition. As a result, the elasticity of the enzyme-substrate complex has been implicated in the successful recognition and catalysis of the substrates which may be inferred to suggest that the elasticity of the enzyme/drug complex plays a role in resistance. A realistic representation of the dynamic nature of the protease may provide a more powerful tool in structure-based drug design algorithms.

Cantilever-centric mechanism of cooperative non-active site mutations in HIV protease: Implications for flap dynamics.

The HIV-1 protease is an important drug target in antiretroviral therapy due to the crucial role it plays in viral maturation. A greater understanding of the dynamics of the protease as a result of drug-induced mutations has been successfully elucidated using computational models in the past. We performed induced-fit docking studies and molecular dynamics simulations on the wild-type South African HIV-1 subtype C protease and two non-active site mutation-containing protease variants; HP3 PR and HP4 PR. The HP3 PR contained the I13V, I62V, and V77I mutations while HP4 PR contained the same mutations with the addition of the L33F mutation. The simulations were initiated in a cubic cell universe containing explicit solvent, with the protease variants beginning in the fully closed conformation. The trajectory for each simulation totalled 50 ns. The results indicate that the mutations increase the dynamics of the flap, hinge, fulcrum and cantilever regions when compared to the wild-type protease while in complex with protease inhibitors. Specifically, these mutations result in the protease favouring the semi-open conformation when in complex with inhibitors. Moreover, the HP4 PR adopted curled flap tip conformers which coordinated several water molecules into the active site in a manner that may reduce inhibitor binding affinity. The mutations affected the thermodynamic landscape of inhibitor binding as there were fewer observable chemical contacts between the mutated variants and saquinavir, atazanavir and darunavir. These data help to elucidate the biophysical basis for the selection of cooperative non-active site mutations by the HI virus.

Two-Step Preparation of Highly Pure, Soluble HIV Protease from Inclusion Bodies Recombinantly Expressed in Escherichia coli.

Heterologous expression of exogenous proteases in Escherichia coli often results in the formation of insoluble inclusion bodies. When sequestered into inclusion bodies, the functionality of the proteases is minimized. To be characterized structurally and functionally, however, proteases must be obtained in their native conformation. HIV protease is readily expressed as inclusion bodies, but must be recovered from the inclusion bodies. This protocol describes an efficient method for recovering HIV protease from inclusion bodies, as well as refolding and purifying the protein. HIV protease-containing inclusion bodies are treated with 8 M urea and purified via cation-exchange chromatography. Subsequent refolding by buffer exchange via dialysis and further purification by anion-exchange chromatography produces highly pure HIV protease that is functionally active. © 2020 by John Wiley & Sons, Inc. Basic Protocol: Recovery, refolding, and purification of HIV protease from inclusion bodies Support Protocol 1: Expression and extraction of inclusion bodies containing HIV protease expressed in Escherichia coli Support Protocol 2: Determination of the active site concentration of HIV protease via isothermal titration calorimetry.

Esterase-Catalyzed Production of Hyperpolarized 13C-Enriched Carbon Dioxide in Tissues for Measuring pH.

13C Magnetic resonance imaging of hyperpolarized (HP) 13C-enriched bicarbonate (H13CO3-) and carbon dioxide (13CO2) is a novel and sensitive technique for tissue pH mapping in vivo. Administration of the HP physiological buffer pair is attractive, but poor polarization and the short T1 of 13C-enriched inorganic bicarbonate salts are major drawbacks for this approach. Here, we report a new class of mixed anhydrides for esterase-catalyzed production of highly polarized 13CO2 and H13CO3- in tissue. A series of precursors with different alkoxy and acyl groups were synthesized and tested for chemical stability and T1. 13C-enriched ethyl acetyl carbonate (13C-EAC) was found to be the most suitable candidate due to the relatively long T1 and good chemical stability. Our results showed that 13C-EAC can be efficiently and rapidly polarized using BDPA. HP 13C-EAC was rapidly hydrolyzed by esterase to 13C-enriched monoacetyl carbonate (13C-MAC), which then decomposed to HP 13CO2. Equilibrium between the newly produced 13CO2 and H13CO3- was quickly established by carbonic anhydrase, producing a physiological buffer pair with 13C NMR signals that can be quantified for pH measurements. Finally, in vivo tissue pH measurements using HP 13C-EAC was successfully demonstrated in the liver of healthy rats. These results suggest that HP 13C-EAC is a novel imaging probe for in vivo pH measurements.

A general chemical shift decomposition method for hyperpolarized (13) C metabolite magnetic resonance imaging.

Metabolic imaging with hyperpolarized carbon-13 allows sequential steps of metabolism to be detected in vivo. Potential applications in cancer, brain, muscular, myocardial, and hepatic metabolism suggest that clinical applications could be readily developed. A primary concern in imaging hyperpolarized nuclei is the irreversible decay of the enhanced magnetization back to thermal equilibrium. Multiple methods for rapid imaging of hyperpolarized substrates and their products have been proposed with a multi-point Dixon method distinguishing itself as a robust protocol for imaging [1-(13) C]pyruvate. We describe here a generalized chemical shift decomposition method that incorporates a single-shot spiral imaging sequence plus a spectroscopic sequence to retain as much spin polarization as possible while allowing detection of metabolites that have a wide range of chemical shift values. The new method is demonstrated for hyperpolarized [1-(13) C]pyruvate, [1-(13) C]acetoacetate, and [2-(13) C]dihydroxyacetone. Copyright © 2016 John Wiley & Sons, Ltd.