3  Conclusion
We synthesized 16 adenine dinucleosides which were designed as bisubstrate SAM analogs to mimic the transition state of 2′-O-methylation of the cap RNA. Both adenosines were connected by various NH or N-substituted linkers between the 2′-O-position of the adenosine representing the 5′-end nucleoside of mRNA and the 5′-position of the adenosine mimicking the SAM cofactor in the methylation reaction. None of these bisubstrate SAM analogs were found to inhibit 2′-O MTases of several flaviviruses, coronavirus or pox-virus at 200 μM. Conversely, six of them inhibited SARS-CoV nsp14 N7-MTase in micromolar range concentration and one in submicromolar range. Additionally, we also observed that these compounds barely inhibit the human RNA N7 MTase, an important feature given that the lack of antiviral specificity represents a common issue in coronavirus antiviral discovery. Indeed, the structural homology between viral and cellular MTases often impairs the discovery of specific inhibitors for CoV N7-MTase. Nevertheless, our work did identify one dinucleoside (13) bearing a 4-chloro-3-nitrobenzenesulfonamide moiety in the N-linker between both adenosines, that blocks the activity of SARS-CoV nsp14 at the submicromolar concentration in the same range than sinefungin but with a significant specificity. Thermal shift assays and molecular modeling indicate that the inhibitory activity is likely due to the binding of 13 to both SAM and mRNA binding pockets of nsp14. It is quite interesting to note that all residues of SARS-CoV nsp14 involved in the binding of 13 are fully conserved in the SARS-CoV-2 nsp14 protein (Fig. S1, Supporting Information). Indeed, the genome sequence of SARS-CoV-2 nsp14 exhibits about 95% sequence similarity with SARS-CoV nsp14 [38]. The lead compound 13 and the most potent derivatives will serve as starting building blocks for the development of SARS-CoV-2 nsp14 inhibitors.