key: cord-1012795-3f50ydww
authors: Aldhumani, Ali H.; Hossain, Md Ismail; Fairchild, Emily A.; Boesger, Hannah; Marino, Emily C.; Myers, Mason; Hines, Jennifer V.
title: RNA sequence and ligand binding alter conformational profile of SARS-CoV-2 stem loop II motif
date: 2021-01-14
journal: Biochem Biophys Res Commun
DOI: 10.1016/j.bbrc.2021.01.013
sha: 599ff48802e887c23c49fcbce0b4596397d5e25d
doc_id: 1012795
cord_uid: 3f50ydww

Antiviral drug discovery continues to be an essential complement to vaccine development for overcoming the global pandemic caused by SARS-CoV-2. The genomic RNA of SARS-CoV-2 contains structural elements important for viral replication and/or pathogenesis making them potential therapeutic targets. Here we report on the stem-loop II motif, a highly conserved noncoding RNA element. Based on our homology model we determined that the G to U transversion in the SARS-CoV-2 stem-loop II motif (S2MG35U) forms a C–U base-pair isosteric to the C-G base-pair in the early 2000's SARS-CoV (S2M). In addition, chemo-enzymatic probing and molecular dynamics simulations indicate the S2MG35U conformational profile is altered compared to S2M in the apical loop region. We explored S2MG35U as a potential drug target by docking a library of FDA approved drugs. Enzymatic probing of the best docking ligands (aminoglycosides and polymyxins) indicated that polymyxin binding alters the conformational profile and/or secondary structure of the RNA. The SARS-CoV-2 stem-loop II motif conformational differences due to nucleotide transversion and ligand binding are highly significant and provide insight for future drug discovery efforts since the conformation of noncoding RNA elements affects their function.

Antiviral drug discovery continues to be an essential complement to vaccine development for overcoming the global pandemic caused by SARS-CoV-2. The genomic RNA of SARS-CoV-2 contains structural elements important for viral replication and/or pathogenesis making them potential therapeutic targets. Here we report on the stem-loop II motif, a highly conserved noncoding RNA element. Based on our homology model we determined that the G to U transversion in the SARS-CoV-2 stem-loop II motif (S2M G35U ) forms a C-U base-pair isosteric to the C-G base-pair in the early 2000's SARS-CoV (S2M). In addition, chemo-enzymatic probing and molecular dynamics simulations indicate the S2M G35U conformational profile is altered compared to S2M in the apical loop region. We explored S2M G35U as a potential drug target by docking a library of FDA approved drugs. Enzymatic probing of the best docking ligands (aminoglycosides and polymyxins) indicated that polymyxin binding alters the conformational profile and/or secondary structure of the RNA. The SARS-CoV-2 stem-loop II motif conformational differences due to nucleotide transversion and ligand binding are highly significant and provide insight for future drug discovery efforts since the conformation of noncoding RNA elements affects their function.

J o u r n a l P r e -p r o o f

The COVID-19 disease and global pandemic caused by the SARS-CoV-2 coronavirus continues to threaten human health and wellbeing [1] . Considerable effort has been focused on developing vaccines and small molecule therapeutics [2] including repurposing of existing drugs [3] . The SARS-CoV-2 coronavirus has genomic RNA structural elements, many in noncoding regions, which play key roles in viral replication and/or pathogenesis [4] . These types of functionally important noncoding RNA structural elements offer an opportunity for small-molecule interactions and the development of RNA-targeted therapeutics [5] .

Relatively few small-molecule therapeutics targeting RNA are in the overall drug discovery pipeline [6] . For treating COVID-19, considerable effort is focused on identifying small molecule therapeutics to target SARS-CoV-2 viral proteins, developing vaccines and developing siRNA that target viral RNA to silence viral protein expression [2] . Less focus has been placed on targeting the viral RNA structural elements, although a few studies identified small molecules targeting the frame-shifting RNA element in SARS-CoV [7] and SARS-CoV-2 [8] .

One RNA element in SARS-CoV-2 not previously investigated for small molecule targeting is the stem-loop II motif. This element is located in the 3′-UTR close to the poly-A tail [4] . The sequence of the stem-loop II motif is highly conserved across coronaviruses and astroviruses that contain it and its tertiary fold is structurally complex [9] . While the function of this RNA element is unknown [10] , it may provide an evolutionary advantage [11] and/or be important for viral pathogenesis [9] . Consistent with this possibility is a recent genomic RNA mapping J o u r n a l P r e -p r o o f study which identified secondary structural differences in the apical loop of the SARS-CoV-2 stem-loop II motif in cells compared to that found in vitro indicating the RNA element may be involved in further molecular interactions in the context of an infected cell [12] . Based on the high sequence conservation, complex tertiary structure and cell-dependent structural differences, the stem-loop II motif has the characteristics of an important potential therapeutic target.

In this paper, we investigate the SARS-CoV-2 stem-loop II motif RNA structure and its ligand binding. We developed a computational homology model and used this for both molecular dynamics simulations and ligand docking. We complemented these computational studies with chemo-enzymatic RNA probing experiments. Together, the data indicate that the stem-loop II motif in SARS-CoV-2 has a different conformational profile compared to that of SARS-CoV due to a G to U base transversion and that this conformational profile is altered upon ligand binding. Consequently, since the conformation of noncoding RNA elements often determines their function [13] , we suggest that the coronavirus stem-loop II motif may be a druggable target.

All reagents were molecular biology grade, RNase/DNase-free or highest purity grade possible.

RNases were from ThermoFisher Scientific. Ligands were from Sigma-Aldrich with purity >90%.

Custom synthesized HPLC-purified RNAs fluorescently labeled with 5′-DY547 were obtained J o u r n a l P r e -p r o o f fully deprotected from Horizon Discovery. RNAs 5′-labeled with DY547 are a substitute for 32 P 5′-labeled RNAs in chemo-enzymatic probing experiments [14] . All RNAs were dialyzed in 5 mM MOPS, pH 7.0, 0.01 mM EDTA and renatured prior to use. All computational work was done using the Owens Cluster at the Ohio Supercomputer Center [15] with the exception of the standard precision docking done on a MacPro and the sequence analysis done on a MacBook Pro.

We used the NCBI Viral Database 25, 2020 to October 28, 2020 (9,302 retrieved). These records were then searched by accession number using BLAST-n [17] for sequences similar to the stem-loop II motif core sequence [9] (nucleotides 10-40, Figure 1 ). The retrieved sequences were then processed using Jupyter Notebook accessed through Anaconda Navigator and BioPython [18] to parse them according to host and nucleotide identity at the position corresponding to nucleotide 35 in Figure 1 and complementarity to the core sequence.

The computational model RNAs were derived from PDB ID 1XJR, the 2005 crystal structure for a stem-loop II motif (S2M) [9] . In order to facilitate comparison of biochemical and computational results, only the single G35U mutation difference in the core was modeled (i.e., J o u r n a l P r e -p r o o f S2M vs. S2M G35U , Figure 1 ). Using Protein Preparation Wizard (Schrödinger), we deleted the G1 GTP and refined the structure of S2M still with the hydrated Mg 2+ . The structure for S2M G35U with hydrated Mg 2+ was prepared by deleting the G35 nucleobase, replacing it with a U followed by refinement. For the S2M G35U docking studies, the metal ions and water were deleted and the RNA refined. All of the above refinements used the Protein Preparation Wizard 

We surveyed genomic databases to investigate the stem-loop II motif RNA element that was previously shown to be rigorously conserved in SARS-related viruses and astroviruses [9, 10] .

We used the NCBI Viral database to retrieve SARS-CoV related entries (taxid:694009 ) and

BLAST-n to search these for sequences somewhat similar to the core sequence in the stem-loop genome study [4] .

The G35U mutation is notable because it is in a region previously determined to be base-paired and rigorously conserved across coronavirus and astrovirus with minimal structural flexibility [9] . In the stem-loop II motif of the early 2000's SARS-CoV outbreak, G35 base-pairs with C16 [9] . Of the 16,352 SARS-CoV-2 human host sequences collected December 2019 to Oct 28, 2020 fewer than 2% had an additional mutation or deletion in the core region beyond G35U. Only 6 of the sequences had a mutation at the position corresponding to C16 (a T or Y) indicating that J o u r n a l P r e -p r o o f there is no significant occurrence of a corresponding compensatory mutation. The C16-G35 base-pair of S2M of the early 2000's SARS-CoV is immediately adjacent to G34 and A33, which form key long-range hydrogen bonds that contribute to the core tertiary fold [9] . Consequently, we investigated the structural effect the G35U mutation has on this important anchoring region in S2M G35U of the current SARS-CoV-2.

We probed S2M and S2M G35U model RNAs using single-strand specific RNases T1 and A. We found that both model RNAs have comparable regions of single-stranded character consistent with the 2005 crystal structure of S2M ( Figure 1 ). The most significant differences in S2M G35U compared to S2M were focused in the apical loop region including a decrease in single-strand character at G24 and an increase in single-stranded character at C27 and G28. Notably, no significant single-strand character was observed at position 35 for either model RNA indicating that position 35 is likely base-paired.

To investigate the implications of the G35U transversion on the tertiary structure of the stemloop II motif, we prepared a homology model based on the 2005 crystal structure of an S2M (PDBID: 1XJR [9] ). The resulting core region of the S2M G35U homology model RNA (nucleotides 10-40) superimposed closely with that of the S2M crystal structure (Figure 2, 2.68 RMS) . In particular, a C16-U35 base-pair with two hydrogen bonds forms in S2M G35U that is similar to the J o u r n a l P r e -p r o o f C16-G35 base-pair in S2M (Figure 2a inset) and is comparable to an exemplar model [21] for a C-U base-pair (1.65 heavy atom RMS with C764-U900 in PDB ID 1S72 [22] , Figure 2b ).

While the in silico studies indicated that the G35U mutation could form a conformationally isosteric base-pair, the hydrogen bonding differences could potentially affect the overall structure / conformational flexibility of the core region. Consequently, we used chemical probing and molecular dynamics simulations to compare S2M and S2M G35U model RNAs.

We used magnesium facilitated in-line cleavage to investigate the conformational profile of each model RNA. This in-line cleavage method identifies nucleotides that are conformationally flexible enough to result in in-line attack of the 2′OH to cleave the phosphodiester bond [23] .

In-line probing of S2M G35U resulted in a two-fold decrease in cleavage at positions U25, U30 and A33 along with a 50% increase at positions C27 and G28 compared to S2M (Figure 3) . Notably, while a slight increase in cleavage was observed at position 35 in S2M G35U , the differences were not statistically significant. These in-line probing results indicate a likely change in the conformational profile of S2M G35U compared to S2M, particularly in the apical loop region.

We also used molecular dynamics simulations to investigate the effect of the G35U mutation.

Molecular dynamics simulations are an effective strategy for exploring RNA conformational sampling and dynamics ranging from bond rotations or sugar puckering occurring on the nanosecond timescale to base-pair opening on the microsecond timescale and more global J o u r n a l P r e -p r o o f changes occurring at longer times [19] . The extent of fluctuation during a 10 ns simulation of the hydrated magnesium model RNAs was greater for S2M than it was for S2M G35U particularly in the apical loop region (e.g., U25), consistent with the conformational profile differences observed in the in-line probing experiments (Figure 3 ). For U30, however, longer timescale global motions may be why the flexibility differences observed in the in-line probing were not reflected in the molecular dynamics simulations.

Since the stem-loop II motif is phylogenetically conserved and likely important for viral pathogenesis [9] , we investigated the ability of small molecules to target this noncoding RNA element using docking methods previously applied to ligand-RNA complexes [24] . We conducted a high-throughput virtual screening (HTVS) of a library of 2309 FDA approved drugs docking to S2M G35U . Approximately 2% of the library (49 compounds) returned no structures during ligand preparation, however dropped/incorrect structures are not uncommon when working with in silico compound libraries [25] . The resulting set of 3236 prepared structures was docked to the S2M G35U model RNA without the Mg 2+ since one key mode by which small molecules bind RNA is via displacement of magnesium ions [26] and since the authors of the original crystal structure suggested a potential site for ligand binding that is available in the RNA without Mg 2+ [9] . The resulting ligand-RNA poses obtained (2628 valid poses) were then ranked based on the calculated E model value as we have done for other ligand-RNA docking studies [24] .

Ligand poses within 10% of the lowest E model pose (polymyxin B, -492 kcal/mol) were selected for further analysis: tobramycin (-453 kcal/mol), kanamycin (-425 kcal/mol), colistin (-406 J o u r n a l P r e -p r o o f kcal/mol) and capreomycin (-488 kca/mol). In the case of capreomycin, the isomer with the lowest energy docking pose was not a natural stereoisomer due to the original structure data file for this compound not having all the stereocenters specified, consequently, this ligand was not included in further analyses. Compound library structure data files with undefined stereocenters are often encountered [25] . The remaining four compounds were redocked to S2M G35U using a standard precision mode and energy minimized. These compounds docked S2M G35U in the tunnel formed by the long-range G11-A33 base-pair of the complex tertiary fold ( Figure 4 ). This tunnel had previously been proposed as a possible binding site for small molecules [9] .

It was not surprising that tobramycin and kanamycin were amongst the best docking compounds since they are aminoglycosides, compounds known to bind RNA electrostatically, frequently in divalent metal ion binding sites [26] . In contrast, we were initially surprised that polymixin B and colistin were amongst the best binders since their antibacterial mechanism of action is primarily thought to be via cell wall disruption, however a recent study determined that it might also involve binding to and disrupting ribosomal RNA [27] . Both the aminoglycosides and polymyxins make extensive contacts with S2M G35U primarily with the nonbridging phosphate oxygens lining the tunnel formed by the G11-A33 base-pair and flanking regions. This is the same location where the two hydrated magnesium ions are bound in the crystal structure [9] and is consistent with the electrostatic RNA binding characteristics of these compounds.

To explore the effects of these compounds on the RNA structure, we looked at ligand-induced changes in the RNase A profile of S2M G35U . Tobramycin and kanamycin had no effect at lower concentrations and only showed an overall, nonspecific protection of enzymatic cleavage at 

RNA conformation and dynamics play a significant role in the functional activity of noncoding RNA elements. This is due to the fact that RNA-ligand binding events take place on a slower time-scale than protein-ligand binding, occurring via a complex interplay of tertiary structure capture and induced-fit [28] . The observed change in the conformational profile of S2M G35U from the current SARS-CoV-2 outbreak may result in a change in its functional activity compared to that of S2M from the early 2000's SARS-CoV outbreak. Previous studies indicate that the stem-loop II motif may have a host-specific target [11] , possibly recruiting the 40S ribosomal subunit to the viral genomic RNA and/or sequestering translation initiation factors [9] . Consistent with these possibilities is a recent report of structural differences concentrated in the apical loop of the stem-loop II motif when in a cellular context [12] . This is in contrast to the global change in the 3'-UTR secondary structure predicted solely based on sequence J o u r n a l P r e -p r o o f analysis [29] . Consequently, the stem-loop II motif of the current SARS-CoV-2 outbreak may favor a subset of conformations that facilitate interactions with host or viral factors related to its function and human pathogenesis.

Small molecules binding to noncoding RNA elements can directly block the binding of endogenous factors and also can alter the conformational profile of the RNA, thus disrupting the RNA element's cellular function [13] . The enzymatic probing changes we observed in the presence of polymyxin B and colistin are consistent with a ligand-induced change in the conformational profile of S2M G35U . Given the possible functional importance of the stem-loop II motif conformational profile discussed above, the identification of ligands which alter the secondary structure and/or conformational profile of the stem-loop II motif in SARS-CoV-2 is significant. These ligand-induced changes indicate the druggability of the stem-loop II motif and may provide an avenue for future drug discovery efforts to treat COVID-19 and other diseases caused by viruses containing the stem-loop II motif. 

Coronavirus disease 2019 (COVID-19): current status and future perspective

Research and development on therapeutic agents and vaccines for COVID-19 and related human coronavirus diseases

To counter the pandemic, clinicians bank on repurposed drugs

RNA genome conservation and secondary structure in SARS-CoV-2 and SARS-related viruses: a first look

Principles for targeting RNA with drug-like small molecules

RNA therapeutics on the rise

Small molelcules targeting viral RNA

Targeting the SARS-CoV-2 RNA Genome with Small Molecule Binders and Ribonuclease Targeting Chimera (RIBOTAC) Degraders

The Structure of a Rigorously Conserved RNA Element within the SARS Virus Genome

Distribution and Evolutionary History of the Mobile Genetic Element s2m in Coronaviruses. Diseases, 4

A mobile genetic element with unknown function found in distantly related viruses

Comprehensive in-vivo secondary structure of the SARS-CoV-2 genome reveals novel regulatory motifs and mechanisms

Understanding the contributions of conformational changes, thermodynamics and kinetics of RNA-small molecule interactions

Association of a peptoid ligand with the apical loop of pri-miR-21 inhibits cleavage by Drosha

Ohio Supercomputer Center. Columbus OH: Ohio Supercomputer Center

NCBI BLAST: a better web interface

Biopython: freely available Python tools for computational molecular biology and bioinformatics

RNA structural dynamics as captured by molecular simulations: A comprehensive overview

Molecular Dynamics Simulations Identify Time Scale of Conformational Changes Responsible for Conformational Selection in Molecular Recognition of HIV-1

Nucleic Acid Database RNA Basepair Catalog

The Roles of Ribosomal Proteins in the Structure, Assembly and Evolution of the Large Ribosomal Subunit

Relationship between internucleotide linkage geometry and the stability of RNA

Synthesis and stereospecificity of 4,5-disubstituted oxazolidinone ligands binding to Tbox riboswitch RNA

The Significance of Chirality in Drug Design and Development

Drug-target interactions that involve the replacement or displacement of magnesium ions

Polymyxins and Analogues Bind to Ribosomal RNA and Interfere with Eukaryotic Translation in Vitro

Slow molecular recognition by RNA

Implications of SARS-CoV-2 Mutations for Genomic RNA Structure and Host microRNA Targeting

We wish to thank the National Institutes of Health (grants #1R15GM132841 and 1C06 RR14575-01), the Ohio Supercomputer Center and Ohio University for support of this work and Dr. P.Hines and Prof. S. Bergmeier for helpful discussions.J o u r n a l P r e -p r o o f

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.☐ The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Please note that all Biochemical and Biophysical Research Communications authors are required to report the following potential conflicts of interest with each submission. If applicable to your manuscript, please provide the necessary declaration in the box above.(1) All third-party financial support for the work in the submitted manuscript.(2) All financial relationships with any entities that could be viewed as relevant to the general area of the submitted manuscript.(3) All sources of revenue with relevance to the submitted work who made payments to you, or to your institution on your behalf, in the 36 months prior to submission. (4) Any other interactions with the sponsor of outside of the submitted work should also be reported. (5) Any relevant patents or copyrights (planned, pending, or issued). (6) Any other relationships or affiliations that may be perceived by readers to have influenced, or give the appearance of potentially influencing, what you wrote in the submitted work. As a general guideline, it is usually better to disclose a relationship than not.