Adenosine deaminases acting on RNA (ADAR) are RNA-editing enzymes that may restrict viral infection. We have utilized deep sequencing to determine adenosine to guanine (A→G) mutations, signifying ADAR activity, in clinical samples retrieved from 93 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)–infected patients in the early phase of the COVID-19 pandemic. A→G mutations were detected in 0.035% (median) of RNA residues and were predominantly nonsynonymous. These mutations were rarely detected in the major viral population but were abundant in minor viral populations in which A→G was more prevalent than any other mutation (P < 0.001). The A→G substitutions accumulated in the spike protein gene at positions corresponding to amino acids 505 to 510 in the receptor binding motif and at amino acids 650 to 655. The frequency of A→G mutations in minor viral populations was significantly associated with low viral load (P < 0.001). We additionally analyzed A→G mutations in 288,247 SARS-CoV-2 major (consensus) sequences representing the dominant viral population. The A→G mutations observed in minor viral populations in the initial patient cohort were increasingly detected in European consensus sequences between March and June 2020 (P < 0.001) followed by a decline of these mutations in autumn and early winter (P < 0.001). We propose that ADAR-induced deamination of RNA is a significant source of mutated SARS-CoV-2 and hypothesize that the degree of RNA deamination may determine or reflect viral fitness and infectivity. Significance Viral RNA may be edited by enzymes of the ADAR family that deaminate adenosine residues with ensuing A→G mutations. We found multiple A→G mutations in minor viral populations of the SARS-CoV-2 genome. A→G mutations accumulated in the receptor binding domain of the spike gene, which may cause structural changes by altering binding to the ACE2 receptor. Presence of A→G mutations in minor viral populations was associated with reduced viral load, implying that ADAR may limit viral replication. Analyses of >250,000 European samples from 2020 revealed that A→G mutations in SARS-CoV-2 RNA were inversely correlated with mortality as a reflection of incidence. ADAR may thus be important in providing new variants of SARS-CoV-2 with altered infectivity and transmissibility.

Hepatocytes infected by hepatitis B virus (HBV) produce different HBV RNA species, including pregenomic RNA (pgRNA), which is reverse transcribed during replication. Particles containing HBV RNA are present in serum of infected individuals, and quantification of this HBV RNA could be clinically useful. In a retrospective study of 95 patients with chronic HBV infection, we characterised HBV RNA in serum in terms of concentration, particle association and sequence. HBV RNA was detected by real-time PCR at levels almost as high as HBV DNA. The HBV RNA was protected from RNase and it was found in particles of similar density as particles containing HBV DNA after fractionation on a Nycodenz gradient. Sequencing the epsilon region of the RNA did not reveal mutations that would preclude its binding to the viral polymerase before encapsidation. Specific quantification of precore RNA and pgRNA by digital PCR showed almost seven times lower ratio of precore RNA/pgRNA in serum than in liver tissue, which corresponds to poorer encapsidation of this RNA as compared with pgRNA. The serum ratio between HBV DNA and HBV RNA was higher in genotype D as compared with other genotypes. The results suggest that HBV RNA in serum is present in viral particles with failing reverse transcription activity, which are produced at almost as high rates as viral particles containing DNA. The results encourage further studies of the mechanisms by which these particles are produced, the impact of genotype, and the potential clinical utility of quantifying HBV RNA in serum.