key: cord-0713214-8u1sk5si
authors: Neupane, Krishna; Munshi, Sneha; Zhao, Meng; Ritchie, Dustin B.; Ileperuma, Sandaru M.; Woodside, Michael T.
title: Anti-Frameshifting Ligand Active against SARS Coronavirus-2 Is Resistant to Natural Mutations of the Frameshift-Stimulatory Pseudoknot
date: 2020-09-11
journal: J Mol Biol
DOI: 10.1016/j.jmb.2020.09.006
sha: 80526752018dc79e56930c5f67e5ffb3beaaaced
doc_id: 713214
cord_uid: 8u1sk5si

SARS-CoV-2 uses − 1 programmed ribosomal frameshifting (− 1 PRF) to control expression of key viral proteins. Because modulating − 1 PRF can attenuate the virus, ligands binding to the RNA pseudoknot that stimulates − 1 PRF may have therapeutic potential. Mutations in the pseudoknot have occurred during the pandemic, but how they affect − 1 PRF efficiency and ligand activity is unknown. Studying a panel of 6 mutations in key regions of the pseudoknot, we found that most did not change − 1 PRF levels, even when base-pairing was disrupted, but one led to a striking 3-fold decrease, suggesting SARS-CoV-2 may be less sensitive to − 1 PRF modulation than expected. Examining the effects of a small-molecule − 1 PRF inhibitor active against SARS-CoV-2, it had a similar effect on all mutants tested, regardless of basal − 1 PRF efficiency, indicating that anti-frameshifting activity can be resistant to natural pseudoknot mutations. These results have important implications for therapeutic strategies targeting SARS-CoV-2 through modulation of − 1 PRF.

The severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) causing the pandemic that is currently sweeping the globe features a −1 programmed ribosomal frameshifting (−1 PRF) site (1) . −1 PRF, which involves a shift in the reading frame of the ribosome at a specific location in the RNA message to generate an alternate gene product, is stimulated by a structure in the mRNA-typically a pseudoknot-that is located 5-7 nucleotides downstream of the 'slippery' sequence where the reading-frame shift occurs (2, 3) . The −1 PRF signal is essential to SARS-CoV-2: the frameshift gene products include the RNA-dependent RNA polymerase that is required for viral replication. Previous work on SARS-CoV showed that mutations suppressing −1 PRF significantly attenuated viral propagation in cell culture, by up to several orders of magnitude (4) (5) (6) . Indeed, work on other viruses has found more generally that −1 PRF levels must typically be held within a relatively narrow range to avoid attenuation (7, 8) .

As a result, frameshift-stimulatory structures are potential targets for anti-viral drugs (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) .

The pseudoknot stimulating −1 PRF in SARS-CoV-2 has a 3-stem architecture that is characteristic of coronaviruses, in contrast to the more common 2-stem architecture of most frameshift-stimulatory pseudoknots in viruses (19) . Although the pseudoknot sequence is highly conserved among coronaviruses (20) , several mutations have been identified in the pseudoknot from viral samples isolated from COVID-19 patients during the pandemic, tracked in the To examine the functional effects of these mutations on frameshifting, we selected a panel of 6 mutations (Fig. 1, magenta) , consisting of U20C, G29U, C43U, U47C, U58C, and C62U.

These mutations were chosen for their prevalence or their potential functional and/or structural importance: C62U and C43U were the two most common mutations; C62U is also located adjacent to an adenine bulge (A63) critical to −1 PRF in SARS-CoV (4); C43U and U47C are part of a domain in loop 2 that promotes pseudoknot dimerization (22); U20C disrupts the crossjunction base-pair stacking between stems 1 and 2 that is important for −1 PRF in many pseudoknots (23) , and U20 is also protected against nuclease digestion in SARS-CoV (4); G29U destabilizes the end of stem 1 near the junction with stem 3, which may facilitate threading of the 5′ end of the RNA through the stem 1/stem 3 junction as proposed in structural models (24, 25) ; and finally U58C pairs with G31 (nuclease-protected in SARS-CoV) and stabilizes the end of stem 3 near the stem 1/stem 3 junction, again possibly affecting the structure. Note that we included in this study only single-nucleotide polymorphisms found from patient samples, rather than any of the engineered mutations found to suppress −1 PRF in SARS-CoV (4, 22) , because the former are most relevant from a therapeutic perspective. For each mutant, we measured the −1 PRF efficiency induced by the pseudoknot using cell-free translation of a dual-luciferase reporter system consisting of the Renilla luciferase gene in the 0 frame upstream of the firefly luciferase gene in the −1 frame and separated by the SARS-CoV-2 frameshift signal (1). The −1 PRF efficiency was obtained from the ratio of luminescence emitted by the two enzymes, compared to controls with 100% and 0% firefly luciferase read-through.

Comparing the results for each mutant to the −1 PRF efficiency seen for the consensus (wildtype) pseudoknot sequence, we found that 5 of the 6 mutations left the −1 PRF efficiency effectively unchanged within error ( Fig. 2A) . Mutations that disrupted base-pairing in a stem J o u r n a l P r e -p r o o f Journal Pre-proof (e.g. G29U) or the dimerization domain (C43U, U47C) as well as for those that left all basepairing intact (U58C, C62U) were both observed to have the same lack of effect, suggesting that most natural mutations do not alter the −1 PRF efficiency characteristic of the virus. Such a result is consistent with previous work suggesting that −1 PRF levels are regulated in a narrow range, outside of which the virus is significantly attenuated (5) (6) (7) (8) , and that they are not determined directly by characteristics such as stability or specific static structures but rather are most closely related to the conformational heterogeneity of the RNA (26, 27) . Remarkably, however, the U20C mutation, which disrupted the A:U base-pair at the end of stem 2 in the stem 1/stem 2 junction, caused a very significant decrease in −1 PRF efficiency, suppressing it over 3fold ( Fig. 2A, red) . This result is particularly striking because in SARS-CoV, a ~3.5-fold reduction of −1 PRF efficiency was shown to cause an over 1,000-fold attenuation of the virus (5, 6) . The fact that the U20C mutant was sampled from a COVID-19 patient implies that SARS-CoV-2 can survive at a wider range of −1 PRF levels than expected based on work on other viruses.

Finally, we explored if these natural mutations might lead to resistance to the effects of an

amino}-benzoic acid ethyl ester, denoted as MTDB (Fig. 2B, inset) . MTDB was previously found to bind to the pseudoknot from SARS-CoV and inhibit −1 PRF (9,11). MTDB was also recently shown to have a similar effect on −1 PRF in wild-type SARS-CoV-2 (1). We repeated the dual-luciferase assays of −1 PRF for all 6 mutants described above while adding 5 μM MTDB, to quantify the reduction in −1 PRF caused by the ligand. Comparing to the effect of MTDB on −1 PRF when using the wild-type pseudoknot (Fig. 2B, black) , we found that for all of the mutants, 5 μM MTDB reduced −1 PRF efficiency by roughly half. This result was J o u r n a l P r e -p r o o f Journal Pre-proof obtained even for the U20C mutant, which already stimulated −1 PRF at a substantially lower efficiency. Hence none of mutations significantly altered the inhibition of −1 PRF by MTDB, regardless of whether the mutations altered the basal −1 PRF efficiency.

Considering how the mutations affected −1 PRF levels, the surprising effect of the U20C mutation on −1 PRF may arise from a combination of key properties of U20 in the SARS-CoV-2 pseudoknot. For one thing, this mutation destabilizes an inter-stem junction; stacking of paired bases across junctions is known to play an important role in the stimulatory power of pseudoknots (23) . Structural modeling also indicates that U20 plays an important role in triplexlike interactions with loop 1 that organize the stem 1/stem 2 interface (24), and previous work showed that removing triples in pseudoknots can reduce −1 PRF efficiency (28) . The relatively sparse network of triples predicted in this pseudoknot (24) compared to others, in combination with the effects on the stem junction, could explain the sensitivity of −1 PRF to mutation of U20.

Interestingly, G29U also disrupts an inter-stem junction, yet it leaves −1 PRF unchanged, suggesting that the stem 1/stem 3 interface is less important than the stem 1/stem 2 junction, consistent with work on SARS-CoV showing that stem 3 is not essential for frameshifting (4).

The results presented here have important implications for therapeutic strategies targeting −1 PRF modulation. First, the example provided by U20C suggests that SARS-CoV-2 can tolerate a substantial reduction in −1 PRF levels. Hence any putative drug suppressing −1 PRF may need to do so at quite dramatic levels to be effective therapeutically. In fact, 5 μM MTDB induces a smaller decrease in −1 PRF than does the U20C mutation; significantly higher concentrations would be needed to abolish −1 PRF, and MTDB is thus unlikely to be effective as a drug for treating COVID-19. Just as important, however, is the evidence that the effects of an anti-J o u r n a l P r e -p r o o f Journal Pre-proof 7 frameshifting ligand can be insensitive to a wide range of natural mutations. The fact that the details of how the mutations affected the pseudoknot-whether they stabilized or destabilized the structure, disrupted base-pairing, affected one region of the structure or another, or even perturbed the pseudoknot enough to alter the basal −1 PRF rate significantly-were effectively immaterial to the activity of MTDB suggests that the inhibitory effect of the ligand is not easily evaded by simple mutations. Even though such an insensitivity to mutation was seen for only a single anti-frameshifting ligand, this behavior shows the promise of the pseudoknot as a therapeutic target, and holds out hope that a ligand with higher anti-frameshifting activity may be found in future work.

J o u r n a l P r e -p r o o f Methods Preparation of mRNA constructs: A dual luciferase reporting system was created by cloning the sequence corresponding to Renilla luciferase and the multiple cloning site (MCS) from the plasmid pMLuc-1 (Novagen) upstream of the firefly luciferase sequence in the plasmid pISO (addgene), as described previously (1) . The frameshift signal from SARS-CoV-2 was then cloned into MCS between the restriction sites PstI and SpeI. Three different types of constructs were made. First, a construct for measuring −1 PRF stimulation was made, containing the frameshift signal with consensus (wild-type) slippery sequence (U UUA AAC) and consensus or mutant pseudoknot sequence, placing the downstream firefly luciferase gene in the −1 frame so that its expression was dependent on −1 PRF. Next, two controls were derived from this construct: (1) a negative control for measuring the background firefly luciferase luminescence (0% firefly luciferase read-through), in which the slippery sequence was mutated to include a stop codon (U UGA AAC); and (2) a positive control for measuring 100% firefly luciferase readthrough, in which the slippery sequence was disrupted (U AGA AAC) and the firefly luciferase gene was shifted into the 0 frame. Sequences for all constructs are listed in Table S1 .

Transcription templates were amplified from these plasmids by PCR, using a forward primer that contained the T7 polymerase sequence as a 5′ extension to the primer sequence (1). The mRNA for dual-luciferase measurements was produced from the transcription templates by invitro transcription (MEGAclear). Luciferase luminescence was then measured using a microplate reader (Turner Biosystems).

First, 20 µL of the reaction mixture was mixed with 100 µL of Dual-Glo Luciferase reagent (Promega) and incubated for 10 min before reading firefly luminescence, then 100 µL of Dual-Glo Stop and Glo reagent (Promega) was added to the mixture to quench firefly luminescence, and the reaction was incubated for 10 min before reading Renilla luminescence. The −1 PRF efficiency was calculated from the ratio of firefly to Renilla luminescence (F:R), subtracting the background F:R measured from the negative control and then normalizing by F:R measured from the positive control.

To quantify the effects of MTDB on −1 PRF, 5 μM MTDB was added to the reaction volume for each construct (wild-type and mutant pseudoknots, positive and negative controls). At least five replicates were measured and the results averaged, as described previously (30) . 

Structural and functional conservation of the programmed -1 ribosomal frameshift signal of SARS coronavirus 2 (SARS-CoV-2)

Frameshifting RNA pseudoknots: structure and mechanism

Pseudoknot-dependent programmed-1 ribosomal frameshifting: Structures, mechanisms and models. In Recoding: expansion of decoding rules enriches gene expression

A three-stemmed mRNA pseudoknot in the SARS coronavirus frameshift signal

Achieving a golden mean: mechanisms by which coronaviruses ensure synthesis of the correct stoichiometric ratios of viral proteins

Ribosomal frameshifting efficiency and gag/gag-pol ratio are critical for yeast M1 double-stranded RNA virus propagation

Decreasing the frameshift efficiency translates into an equivalent reduction of the replication of the human immunodeficiency virus type 1

Identification of RNA pseudoknot-binding ligand that inhibits the− 1 ribosomal frameshifting of SARS-coronavirus by structurebased virtual screening

Interference of ribosomal frameshifting by antisense peptide nucleic acids suppresses SARS coronavirus replication

Antiframeshifting ligand reduces the conformational plasticity of the SARS virus pseudoknot

Cell cycle control (and more) by programmed− 1 ribosomal frameshifting: implications for disease and therapeutics

High-affinity recognition of HIV-1 frameshift-stimulating RNA alters frameshifting in vitro and interferes with HIV-1 infectivity

N-Methylation as a selectivity of RNA-binding peptides: application to the HIV-1 frameshift-stimulating RNA

RNA-targeted triazoles inhibit propagation of replication-competent and multi-drug-resistant HIV in human cells

Enhancing the ligand efficiency of anti-HIV compounds targeting frameshift-stimulating RNA

Targeting frameshifting in the human immunodeficiency virus

Structure of the HIV-1 frameshift site RNA bound to a small molecule inhibitor of viral replication

The role of programmed-1 ribosomal frameshifting in coronavirus propagation

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

Data, disease and diplomacy: GISAID's 13

dimerization plays a role in ribosomal frameshifting of the SARS coronavirus

The global structures of a wild-type and poorly functional plant luteoviral mRNA pseudoknot are essentially identical

Modeling the structure of the frameshift stimulatory pseudoknot in SARS-CoV-2 reveals multiple possible conformers

Cryo-electron microscopy and exploratory antisense targeting of the 28-kDa frameshift stimulation element from the SARS-CoV-2 RNA genome

Complex dynamics under tension in a high-efficiency frameshift stimulatory structure

Programmed− 1 frameshifting efficiency correlates with RNA pseudoknot conformational plasticity, not resistance to mechanical unfolding

Triplex structures in an RNA pseudoknot enhance mechanical stability and increase efficiency of-1 ribosomal frameshifting

A dual-luciferase reporter system for studying recoding signals

Systematic analysis of bicistronic reporter assay data

writing-review & editing, supervision. Sneha Munshi: resources, investigation, analysis, writing-review & editing. Meng Zhao: investigation, analysis, visualization, writing-review & editing. Dustin Ritchie: resources, investigation, analysis, writing-review & editing. Sandaru Ileperuma: investigation, writing-review & editing. Michael Woodside: conceptualization, writing-original draft

We gratefully acknowledge the originating laboratories that obtained SARS-CoV-2 specimens and submitting laboratories that generated and shared genetic sequence data via the GISAID Initiative, and all associated authors. This work was supported by Canadian

The authors declare that they have no conflicts of interest with the contents of this article.