Substrate Specificity and Kinetics of RNA Hydrolysis by SARS-CoV-2 NSP10/14 Exonuclease

Abstract

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the virus that causes COVID-19, continues to evolve resistance to vaccines and existing antiviral therapies at an alarming rate, increasing the need for new direct-acting antiviral drugs. Despite significant advances in our fundamental understanding of the kinetics and mechanism of viral RNA replication, there are still open questions regarding how the proofreading exonuclease (NSP10/NSP14 complex) contributes to replication fidelity and resistance to nucleoside analogs. Through single turnover kinetic analysis, we show that the preferred substrate for the exonuclease is double-stranded RNA without any mismatches. Double-stranded RNA containing a 3'-terminal remdesivir was hydrolyzed at a rate similar to a correctly base-paired cognate nucleotide. Surprisingly, single-stranded RNA or duplex RNA containing a 3'-terminal mismatch was hydrolyzed at rates 125- and 45-fold slower, respectively, compared to the correctly base-paired double-stranded RNA. These results define the substrate specificity and rate of removal of remdesivir for the exonuclease and outline rigorous kinetic assays that could help in finding next-generation exonuclease inhibitors or nucleoside analogs that are able to evade excision. These results also raise important questions about the role of the polymerase/exonuclease complex in proofreading during viral replication. Addressing these questions through rigorous kinetic analysis will facilitate the search for desperately needed antiviral drugs to combat COVID-19.

Figure 1

Duplex RNA without mismatches is the preferred substrate for the SARS-CoV-2 NSP10/NSP14 exonuclease complex. A solution of 1 μM NSP10/NSP14 was mixed with 100 nM FAM-ssRNA (FAM-LS2, blue), FAM-primer/template double-strand RNA (dsRNA) with an A:A mismatch (FAM-LS2/LS1.2.2, green), or FAM-primer/template dsRNA (FAM-LS2/LS1, red) to start the reaction. Data are shown on a logarithmic time scale, fit to single-exponential functions, with observed rates of 0.016, 0.044, and 2 s^–1 for ssRNA, A:A mismatch dsRNA, and primer/template dsRNA, respectively, as summarized in Table 1. Note that the template strands LS1 and LS1.2.2 differ by three nucleotides in the single-strand region of the template strand, which is unlikely to have any nearest-neighbor effects, which are due to duplex stability and structure.

Figure 2

Processive exonuclease kinetics observed with double-stranded RNA. (A) Excision upon mixing the enzyme and RNA to start the reaction. A solution of 1 μM NSP10/NSP14 complex was mixed with 100 nM FAM-LS2/LS1 RNA to start the reaction, monitored using a quench-flow instrument. Both syringes contained 5 mM Mg²⁺. Data for the loss of the 20 nt primer fit a single exponential with an observed rate of 2 ± 0.1 s^–1. Data for the formation and decay of the 19 nt product best fit a double-exponential function with both rates at approximately 2 s^–1, given in Table 1. (B) Excision upon the Mg²⁺ addition to E-RNA complex. A solution of 1 μM NSP10/NSP14, 2.5 mM EDTA, and 100 nM FAM-LS2/LS1 RNA was mixed with 7.5 mM Mg²⁺ to start the reaction. Samples were quenched with EDTA, and products were resolved by capillary electrophoresis. Fitting the fast phase of the loss of 20 nt starting material gave an observed rate of 6.5 ± 0.9 s^–1. A double-exponential fit of the data for the formation and decay of the 19 nt intermediate is summarized in Table 1. Note that on the time scale of the experiment in panel (B) a fraction of the starting material (∼10%) fails to react. We could not account for this fraction of slow-reacting RNA by any simple model based on RNA–enzyme equilibration, so we have focused on the faster reaction phase accounting for 90% of the reaction.

Figure 3

Excision of incorporated remdesivir monophosphate. (A) Scheme for the enzymatic synthesis of RMP containing substrate. SARS-CoV-2 RdRp complex and remdesivir triphosphate were added to enzymatically incorporate RMP into the primer strand (see the Materials and Methods section). The RdRp complex was heat-denatured, and then the RNA was reannealed before measuring excision by NSP10/NSP14 in the quench flow. (B) Structure of remdesivir monophosphate. Minor modifications relative to ATP in the ring and the addition of a 1′ cyano group that causes delayed chain termination by the RdRp.^, (C) Time course of remdesivir excision from the RNA. Data are shown fit to a double-exponential function with rates of 2.4 and 0.42 s^–1 for the fast and slow phases, respectively. Amplitudes for the two phases are approximately equal.