A Structure-Function Diversity Survey of the RNA-Dependent RNA Polymerases From the Positive-Strand RNA Viruses

Jia, Hengxia; Gong, Peng

contains several basic residues and is known to interact with the triphosphate and base moieties of the NTP substrate. Among these residues, one lysine and one arginine (corresponding to the PV RdRP residues K159 and R174) have the highest conservation level (Bruenn, 2003) . Hence, we use the absolutely conserved motif C aspartic acid (the first D in the signature sequence XGDD) as the reference point to align all RdRP sequences and use the aforementioned conserved residues to label motifs A/B/C/F (Figure 1) . In this way, the relative spacing of these key motifs can be compared in all representative sequences. Typically, the seven motifs appear in the order of G-F-A-B-C-D-E and follow the same protein folding topology. Very interestingly, the RdRPs from the Permutotetraviridae family has a different motif order of G-F-C-A-B-D-E. While its spatial organization of the motifs is consistent with that of other RdRPs, the folding topology is permutated (Gorbalenya et al., 2002; Ferrero et al., 2015) . A similar situation was found in RdRPs from the Birnaviridae family in the double-stranded (ds) RNA virus category (Gorbalenya et al., 2002; Pan et al., 2007) . These exceptions suggest that the swapping of the motifs could occur during protein evolution, while the catalytic function could remain largely unaffected. Besides the similarity in the order of RdRP catalytic motifs between the Birnaviridae and the Permutotetraviridae, the Birnaviridae viruses also use a VPg (viral protein genome linked)-mediated initiation mechanism for genome replication and a polyprotein coding strategy that are often found in the positive-strand RNA viruses (Lee et al., 1977; Pan et al., 2007) . These observations suggest that the evolutionary boundary between the positive-strand and ds RNA viruses are not definite. Next, we use two representative RdRPs, the PV 3D pol and the hepatitis C virus (HCV) NS5B to help estimate the boundaries of the catalytic module using conserved residues in motifs A/B/C/F as the reference points (Figure 1) . The first reason for choosing these two representatives is that these two proteins are known not to contain functional regions beyond the catalytic module except that the NS5B protein has a 21-residue membrane anchor at its C-terminus (Schmidt-Mende et al., 2001) . The second reason is that 3D pol and NS5B represent RdRPs that utilize primerdependent and de novo mechanisms to initiate the RNA synthesis, respectively (Wimmer and Nomoto, 1993; Zhong et al., 2000) . Both of these proteins have its N-terminus ∼150 or ∼230 residues away from the conserved motif F lysine or motif A aspartic acid (corresponding to the yellow and green bars in Figure 1) , respectively, while the residue distances between C-terminal boundary of the catalytic module and the conserved motif C aspartic acid are different (∼130 residues for 3D pol vs. ∼250 residues for NS5B). The thumb domain usually starts from 50-60 residues after the XGDD sequence and ends at the C-terminal boundary of the catalytic module. The primer-dependent 3D pol contains four helices in the thumb, while the de novo NS5B contains seven. If compared to 3D pol , NS5B has one insertion between the third and fourth helices, three extra helices after the fourth helix, and a C-terminal extension (Hansen et al., 1997; Lesburg et al., 1999) . It has been suggested that the insertion and the extension together form a priming platform, interacting with the 3 -end of the template and the initiati