Programmed ribosomal frameshifting is a key event during translation of the SARS-CoV-2 RNA genome allowing synthesis of the viral RNA-dependent RNA polymerase and downstream proteins. Here we present the cryo-electron microscopy structure of a translating mammalian ribosome primed for frameshifting on the viral RNA. The viral RNA adopts a pseudoknot structure that lodges at the entry to the ribosomal mRNA channel to generate tension in the mRNA and promote frameshifting, whereas the nascent viral polyprotein forms distinct interactions with the ribosomal tunnel. Biochemical experiments validate the structural observations and reveal mechanistic and regulatory features that influence frameshifting efficiency. Finally, we compare compounds previously shown to reduce frameshifting with respect to their ability to inhibit SARS-CoV-2 replication, establishing coronavirus frameshifting as a target for antiviral intervention.
Ribosomal frameshifting, a process during which the reading frame of translation is changed at the junction between open reading frames 1a and 1b, is one of the key events during translation of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) positive sense single-stranded RNA genome. This programmed -1 translational frameshifting is conserved in all coronaviruses and is necessary for synthesis of viral RNA-dependent RNA polymerase (RdRp or Nsp12) and downstream viral non-structural proteins encoding core enzymatic functions involved in capping of viral RNA, RNA modification and processing, and RNA proof-reading (1). Although the translational machinery typically prevents frameshifting as a potential source of one of the most disruptive errors in translation (2, 3), many viruses rely on programmed ribosomal frameshifting to expand and fine-tune the repertoire and stoichiometry of expressed proteins (4).
Programmed -1 frameshifting in SARS-related coronaviruses occurs at the slippery sequence U_UUA_AAC in the context of a 3′ stimulatory RNA sequence that was predicted to form a 3-stemmed pseudoknot structure (5), and in parallel was independently tested by our lab and others (6–8). The frameshifting occurs with high efficiency (25-75%) depending on the system used (6, 7, 9–11) and changes the reading frame to UUU_AAA_C (12) (Fig. 1A). Consequently, two viral polyproteins are synthesized, one encoded by the ORF1a when frameshifting does not take place, whereas ORF1ab is expressed as a result of frameshifting. Translation of ORF1a produces polyprotein 1a ending with Nsp10 followed by the short Nsp11. On the other hand, when the frameshift occurs, the polyprotein 1ab is generated, which contains almost 2700 additional amino acids and in which the viral RdRp, Nsp12, is produced after Nsp10 as a consequence of translation in the -1 frame. A putative secondary structure element in the viral RNA that forms a loop upstream of the shift site has been proposed to play an attenuating role in frameshifting and is referred to as the 5′ attenuator loop (8). Maintaining the precise level of coronavirus frameshifting efficiency is crucial for viral infectivity, evidenced by the remarkable fact that mutation of a single nucleotide in the frameshifting region of the SARS-CoV-1 RNA results in a concomitant abrogation of viral replication (13). Therefore, the importance of 3-stemmed pseudoknot-dependent -1 ribosomal frameshifting for the propagation of SARS-related coronaviruses, a process that has not been seen to occur on any endogenous human transcript in human cells, presents itself as an opportune drug-target with minimal tolerance for drug-resistant mutations.
(A) Schematic of the SARS-CoV-2 main ORF. In the close up view of the frameshift event, codons and corresponding amino acids are shown. During -1 frameshifting, the ‘slippery site’ codons UUA (Leu) and AAC (Asn) are the last codons decoded in the 0 frame. Upon -1 frameshifting of the AAC codon to AAA, translation resumes at the CGG (Arg) triplet, where elongation proceeds uninterrupted to produce full-length Nsp12. (B) In vitro translation reaction depicting pausing at the frameshift site. Efficient frameshifting is observed for the WT template, consistent with our dual luciferase assays (see methods). Samples for cryo-EM originally intended to be trapped by dominant negative eRF1 (AAQ) show a tRNA-bound pause in proximity of the frameshift site. The tRNA-associated band is lost upon RNase treatment. Reactions without added eRF1 (AAQ) produce a similarly paused product. (C) Overview of the density low pass filtered to 6Å with the pseudoknot found close to the entry of the mRNA channel on the small subunit (SSU). The SSU proteins are colored in yellow, the large subunit (LSU) proteins in blue and the rRNA in grey. The pseudoknot is colored according to its secondary structure as in (F), and the P-site tRNA is colored in green. (D) Close-up view of the pseudoknot from the solvent-exposed side of the SSU. Helix h16 of the 18S rRNA interacts with the base of Stem 1. Unpaired loop-forming nucleotides are colored in cyan. (E) P-site codon-anticodon interactions reveal a Phe (UUU) codon interacting with tRNA(Phe). (F) Schematic of the revised secondary structure elements in the pseudoknot necessary for -1 PRF with different functional regions labeled and colored accordingly.
Due to its importance in the life cycle of many important viruses and coronaviruses in particular, programmed frameshifting has been extensively studied using a range of structural and functional approaches (4). The structure of a 3′ stimulatory pseudoknot in isolation or in context of the viral genome has been proposed recently by various groups using techniques that include molecular dynamics, nuclease mapping, in vivo selective 2′-hydroxyl acylation analyzed by primer extension (SHAPE), nuclear magnetic resonance (NMR) and cryo-electron microscopy (cryo-EM) (7, 14–17). Furthermore, a ribosomal complex with a frameshift stimulatory pseudoknot from the avian infectious bronchitis virus was reported at low resolution (18). Here, to provide a structural and mechanistic description of the events during ribosomal frameshifting, we investigated mammalian ribosomes captured in distinct functional states during translation of a region of SARS-CoV-2 genomic RNA where -1 programmed frameshifting occurs.