Regulation of coronavirus nsp15 cleavage specificity by RNA structure

Abstract

SARS-CoV-2, the etiologic agent of the COVID-19 pandemic, has had an enduring impact on global public health. However, SARS-CoV-2 is only one of multiple pathogenic human coronaviruses (CoVs) to have emerged since the turn of the century. CoVs encode for several nonstructural proteins (nsps) that are essential for viral replication and pathogenesis. Among them is nsp15, a uridine-specific viral endonuclease that is important in evading the host immune response and promoting viral replication. Despite the established endonuclease function of nsp15, little is known about other determinants of its cleavage specificity. In this study we investigate the role of RNA secondary structure in SARS-CoV-2 nsp15 endonuclease activity. Using a series of in vitro endonuclease assays, we observed that thermodynamically stable RNA structures were protected from nsp15 cleavage relative to RNAs lacking stable structure. We leveraged the s2m RNA from the SARS-CoV-1 3'UTR as a model for our structural studies as it adopts a well-defined structure with several uridines, two of which are unpaired and thus highly probable targets for nsp15 cleavage. We found that SARS-CoV-2 nsp15 specifically cleaves s2m at the unpaired uridine within the GNRNA pentaloop of the RNA. Further investigation revealed that the position of uridine within the pentaloop also impacted nsp15 cleavage efficiency suggesting that positioning within the pentaloop is necessary for optimal presentation of the scissile uridine and alignment within the nsp15 catalytic pocket. Our findings indicate that RNA secondary structure is an important determinant of nsp15 cleavage and provides insight into the molecular mechanisms of RNA recognition by nsp15.

Fig 1. Secondary structure protects RNA from cleavage by SARS-CoV-2 nsp15.

A. Predicted minimum free energy structures (RNAfold) of 100 nt RNAs from the SARS-CoV-2 genome of varying thermodynamic stability (high, medium, and low). Unpaired uridines are highlighted in blue, and paired uridines are outlined in blue. All RNA structures were predicted in RNAfold and designed in RNA2Drawer [34, 36]. B. The locations within the SARS-CoV-2 genome are indicated for each structure, and thermodynamic stability is indicated as -ΔG. Frequency of MFE and ensemble diversity is listed as predicted by RNAfold. C. Endonuclease assays of RNAs from (A) at a nsp15 hexamer:RNA ratio of 1:600. Full-length (uncut) RNAs are indicated by the red arrows. Diminishing cleavage products (cleavage intermediates) are indicated by green arrows. Accumulating cleavage products (cleavage end products) indicated by blue arrows. RNAs were run on a 15% TBE-Urea Gel. Representative images from one of three total experiments are shown. D. Quantitation of data from (C). The percentage of full-length RNA remaining was measured by densitometry. Most thermodynamically stable RNA (RNA 1) was cleaved least rapidly relative to other RNAs. Percentage of uncut RNA was calculated by normalizing to a denatured nsp15 control (indicated by (-) in the far-right lane). Area under the curve (AUC) was calculated for each RNA, and one-way ANOVA with multiple comparisons was performed on the AUC. Data represents three independent experiments.

Fig 2. SARS-CoV-2 nsp15 cleaves structured RNAs at specific sites.

A. Endonuclease assays of full-length s2m mutant RNAs at a nsp15 hexamer:RNA ratio of 1:30. RNAs were run on a 15% TBE-Urea Gel. Full-length bands indicated at the top of the gel with cleavage products below. Representative images from one experiment. B. Percent of full-length RNA remaining as measured by densitometry. Percentage of uncut RNA was calculated by normalizing to a denatured nsp15 control (-). RNAs with uridines present in the pentaloop (s2m and s2m_mut2) cleaved more efficiently than s2m_mut1 and s2m_mut3. AUC was calculated for each RNA, and one-way ANOVA with multiple comparisons was performed on the AUC. AUC of s2m mutant RNAs are compared to wt s2m. Data represents three independent experiments. C. Thermodynamic stability, frequency of MFE, and ensemble diversity of s2m and mutant RNAs indicated as predicted by RNAfold. D. Endonuclease assays of modified s2m RNAs at a nsp15 hexamer:RNA ratio of 1:6. RNAs were run on a 22.5% TBE-Urea Gel. Full-length bands indicated at the top of the gel with cleavage products below. Representative images from one experiment. E. Percent of full-length RNA remaining as measured by densitometry. Percentage of uncut RNA was calculated by normalizing to a denatured nsp15 control (-). RNAs with uridines present in the pentaloop (∆s2m and ∆s2m_mut2) were cleaved more efficiently than ∆s2m_mut1 and ∆s2m_mut3. AUC was calculated for each RNA, and one-way ANOVA with multiple comparisons was performed on the AUC. AUC of ∆s2m mutant RNAs are compared to wt ∆s2m. Data represents three independent experiments. F. Nsp15 binding curves of Δs2m RNAs as compared to wt s2m. ∆s2m RNA binding curves overlap with each other while wt s2m curve is shifted left indicating improved binding with nsp15. Dissociation constant (K_D) was calculated for each RNA, and one-way ANOVA with multiple comparisons was performed on the K_D. AUC of wt full-length s2m and ∆s2m mutant RNAs are compared to wt ∆s2m. Data represents three independent experiments.

Fig 3. Uracil position in loop structures impacts SARS-CoV-2 nsp15 cleavage efficiency.

A. Endonuclease assays of s2m pentaloop mutant RNAs. Reactions were run at a nsp15 hexamer:RNA ratio of 1:6 run on a 22.5% TBE-Urea Gel. Full-length bands indicated at the top of the gel with cleavage products below. Representative images from one experiment. Cleavage products that accumulate over time are indicated by arrows. B. Thermodynamic stability, frequency of MFE, and ensemble diversity of ∆s2m mutant RNAs indicated as predicted by RNAfold. C. Percent of full-length RNA remaining as measured by densitometry. Percentage of uncut RNA was calculated by normalizing to a denatured nsp15 control (indicated by (-) in the far-right lane). RNAs with uridines positioned closer to the apex of the pentaloop cleaved more rapidly than ∆s2m_mut5 in which the uridine is at the end of the pentaloop. AUC was calculated for each RNA, and one-way ANOVA with multiple comparisons was performed on the AUC. Data represents three independent experiments.