Abstract
The nucleocapsid (N) protein is the most expressed protein in later stages of SARS-CoV-2 infection with several important functions. It is translated from a subgenomic mRNA (sgmRNA) formed by template switching during transcription. A recently described translation initiation site (TIS) with a CTG codon in the leader sequence (TIS-L) is out of frame with most structural and accessory genes including the N gene and may act as a translation suppressor. We analyzed multiple sequenced samples infected by SARS-CoV-2 and found that any single variant of this virus produces multiple isoforms of the N sgmRNA. The main isoform starting at TIS-L is out of frame, but two secondary dominant isoforms (present in nearly all samples) were found to restore the reading frame and likely involved in the regulation of N protein production. Analysis of sequenced samples infected by other coronaviruses revealed that such isoforms are also produced in their transcriptomes. In SARS-CoV, they restore the reading frame for a putative TIS (also a CTG codon) in the same relative position as in SARS-CoV-2. Positions of junction breakpoints relative to stem loop 3 in the 5'-UTR suggest similar mechanisms in SARS-CoV, SARS-CoV-2, and OC43, but not in MERS-CoV. These observations may be pertinent for antisense-based antiviral strategies.