Human SARS-CoV-2 has evolved to reduce CG dinucleotide in its open reading frames.

The outbreak of COVID-19 has brought great threat to human health. Its causative agent is a severe acute respiratory syndrome-related coronavirus which has been officially named SARS-CoV-2. Here we report the discovery of extremely low CG abundance in its open reading frames. We found that CG reduction in SARS-CoV-2 is achieved mainly through mutating C/G into A/T, and CG is the best target for mutation. Meanwhile, 5'-untranslated region of SARS-CoV-2 has high CG content and is capable of forming an internal ribosome entry site (IRES) to recruit host ribosome for translating its RNA. These features allow SARS-CoV-2 to reproduce efficiently in host cells, because less energy is consumed in disrupting the stem-loops formed by its genomic RNA. Notably, genomes of cellular organisms also have very low CG abundance, suggesting that mutating C/G into A/T occurs universally in all life forms. Moreover, CG is the dinucleotide related to CpG island, mutational hotspot and single nucleotide polymorphism in cellular organisms. The relationship between these features is worthy of further investigations.