Skip to main content

Drug Discovery Targeting SARS-CoV-2 RNA

      The rapid spread of corona virus has lift up global concerns, especially with no effective treatment available so far. Different therapeutic approaches have been proposed, including developing vaccines and targeting protein machinery employed by virus RNA replication. Besides those effective yet conventional approaches, directly targeting RNA with small molecules can be a potential direction to develop antivirals.  RNA targeting with small molecules is an emerging field. Some recent work from academia, the Disney group in particular, and industry demonstrated that RNAs, just like proteins, can be selectively targeted by small molecules, which offers an alternative approach of finding effective drugs against COVID-19. Recently, a preprint on BioRvix from the Das group at Stanford University utilized computational prediction of different regions within SARS-CoV-2 RNA and offered a 3D structure dataset called ‘FARFAR2-SARS-CoV-2’. This dataset is proposed in this work as a starting point of virtual screening for small molecule candidates.

      Due to the limited information about SARS-CoV-2, most of the work in this paper is based on the structure of related β-coronavirus. It is already proved that the β-coronavirus genome contains conserved regions, which are essential for the life cycle and also have stable secondary structures that can be targeted. In this work, several regions in the virus genome were modeled using Rosetta, including 5’ UTR, the frameshifting element (FSE) and 3’ UTR.


Figure 1: A) 5′ UTR secondary structure, with each modeled stem colored. In bold are positions that are completely conserved across a set of SARS-related virus sequences. B) Top 10 clusters for SL1. The top-scoring cluster member in each case is depicted with solid colors, and the top cluster members (up to 10) are depicted as transparent structures.

      For SARS-CoV-2 extended 5’ UTR, stem loops 1-7 (SL1-7) were modeled based on the previous dissection of β-coronavirus secondary structure (Chen, 2010; Yang, 2015). Some of the stem loops that have structural preferences are indicated with high potential for small molecule binding, which can possibly alter conformations and thus interrupt RNA functions. For example, mutations in SL1 can destabilize the lower part of this stem, which interrupts the long-range RNA contact, leading to a failed initiation of translation by the human genome. More than 200,000 FARFAR2 models were generated for each stem loop in 5’ UTR with excellent modelling convergence (Fig.1). Top 10 scored clusters of each stem loop were selected, indicating their higher potential of binding small molecules (examples of SL1 were shown in Fig. 1B). To test the accuracy of this model, DMS and SHAPE reactivity data of SL5 predicted from RNA structures was also provided, which recovered the same secondary structure as shown in the homology model to SL5 (Fig.2).


Figure 2. Secondary structures for the SARS- CoV-2 SL5 predicted from RNAstructure guided by DMS and SHAPE reactivity data validate the SL5 secondary structure modeled based on homology to literature models (Fig. 1A). DMS and SHAPE reactivity are colored for each position, with red indicating high reactivity, white indicating low reactivity, and grey indicating positions that cannot be modified.

      A similar process was also conducted for SARS-CoV-2 frameshifting elements and 3’ UTR region. Due to the larger size of those elements and the presence of hyper-variable region, they were not able to observe adequate convergence in those models. The results are expected, as pointed out by the authors, due to the limited computational power and lack of experimental characterizations.

      The authors have also listed the caveats using this model. Those models can be biased only using the Rosetta function. The models were also generated based on solvation effects, nucleobase stacking and other physical forces, which can vary when it comes to structure prediction. More extensive computational work is definitely needed to reach high convergence and accuracy.

      Even with those caveats, this dataset can still serve as a good starting point for more physically realistic conformations of virtual ligand screening and more sophisticated modeling. All the data and models have been uploaded to Github as resources for future directions. This dataset can provide a novel path for virtual screening with RNA computational models, which has not been used before. It is both a challenge and opportunity to develop small molecules directly targeting the RNA of SARS-CoV-2.

References
1.     Rangan. R, Watkins. A, Kladwang. W, Das. R; BioRxiv 2020.04.14.041962; doi: http://doi.org/10.1101/2020.04.14.041962
2.     https://github.com/DasLab/FARFAR2-SARS-CoV- 2

Comments

Popular posts from this blog

Lysine-Targeted Covalent Inhibitors Gaining Traction

In the past decade, targeted covalent inhibitors (TCIs) has experienced a resurgence due to its sustained target engagement and improved proteome-wide selectivity. Currently, targeting cysteine residues with acrylamides and other α, β-unsaturated carbonyl compounds is the predominant strategy in TCI development. Although lysine is more prevalent than cysteine and possesses lower mutation rates, lysine has been much less frequently considered and more challenging as a residue targeted by TCIs due to its low nucleophilicity. Recently, Jack Taunton and his colleagues reveled their two related work in JACS .   Based on the co-crystal structure of Hsp90 with a reversible inhibitor PU-H71, they designed two arylsulfonyl fluorides 1 and 2 and co-incubated them with Hsp90 protein (Figure 1a, b). Mass spectrometry detected the formation of a 1:1 covalent adduct. Mutation of Lys58 to arginine abrogated the covalent modification, indicating that the TCI selectively modified Lys58 (Fig...

Grafting Alkyne Handles on Cysteines

One of the most common ways to study biomolecules is to label specific amino acids, mostly cysteine, with chemical handles for further modification. Azide-alkyne click reaction is a very popular strategy due to its high reactivity and reaction rate. Therefore, numerous reagents have been developed to specifically label proteins with terminal alkynes, including iodoacetamide alkynes and maleimide derivatives. However, limitations like insufficient selectivity and low stability remain as long-lasting concerns. To address these problems, one common strategy is introducing a linker between the targeting residue and the alkyne moiety , which potentially brings in unwanted structural changes. A reagent that can introduce small-sized terminal alkyne with one step under mild conditions is in demand. Scheme 1 . Cysteine-labeling study in this work In this study, the Waser group presented a series of reagents that selectively ethynylate cysteine in a one-pot manner ( Scheme 1 ). The ...

First PROTAC Degradation of Mulit-Pass Transmembrane Protein

Cancer cells exploit proteins involved in energy metabolism to maintain a steady supply of building blocks necessary for increased proliferation. Many transporters play a role in the process of shuttling water, nutrients, ions, and various metabolic products across the cell membrane, including the solute carrier (SLC) group of proteins. The SLC protein group consists of more than 400 proteins organized into more than 50 families. For example, the SLC9 family is involved in intracellular pH homeostasis, with increased SL9 activity resulting in an elevated intracellular pH and cytosolic alkalinization in cancer. Currently there 12 FDA approved drugs targeting SLC proteins for conditions ranging from hypertension to depression, but recent studies suggest SLC proteins may be attractive targets for cancer drug development. With this end goal, Bensimon et al., aimed to investigate if multi-pass transmembrane proteins like SLCs can be chemically degraded. To first explore this aim, th...