Principles of Virology. Jane Flint

and PB2 (green) subunits (PDB file 4WSB). (B) Cartoon model of the trimeric influenza virus RdRP colored as above with PB1 (cyan), PA (magenta), and PB2 (green) subunits. (C) Model for activation of RdRP by virion RNA. The three P proteins form a multisubunit assembly that can neither bind to capped primers nor synthesize mRNAs. Addition of a sequence corresponding to the 5′-terminal 11 nucleotides of the viral RNA, which is highly conserved in all eight genome segments, activates the cap-binding activity of the P proteins. The PB1 protein binds this RNA sequence and activates the cap-binding PB2 subunit, probably by conformational change. Concomitantly with activation of cap binding, the PB1 protein acquires the ability to bind to a conserved sequence at the 3′ ends of genomic RNA segments. This second interaction activates the endonuclease that cleaves host cell RNAs 10 to 13 nucleotides from the cap, producing the primers for viral mRNA synthesis. The RNA polymerase can then carry out initiation and elongation of mRNAs. p, polymerase active site. 5′ and 3′ indicate the binding sites for the 5′ and 3′ ends, respectively, of (−) strand genomic RNA. Gray indicates an inactive site, and red indicates an active site.

      In most cases, the RdRP first binds the RNA template-primer such that the templating base is above the active site. In this state, the RdRP conformation is the same as in the unbound form. Nucleotides enter the catalytic site via a large opening on one side of the enzyme. NTP selection is via interactions between the ribose 2′ and 3′ hydroxyl groups and three conserved residues on motifs B and A. These interactions cause a subtle restructuring of the palm domain, closing the active site. Incorrect NTPs can bind, but their ribose hydroxyls will not be properly positioned to cause active-site closure, and hence they will be inefficiently incorporated. After catalysis, the active site is opened by movement of motif A, and the template moves one base to place the next base in the active site.

Closure of the active site by movements of the palm domain appears to be a feature of the RdRPs of all (+) RNA viruses but not (–) strand or double-stranded RNA viruses; their palm domains are already structured in the unbound form. This simple nucleotide selection mechanism greatly influences polymerase fidelity. In T7 RNA polymerase and Taq DNA polymerase, a pre-insertion site is utilized to first bind the incoming NTP to the templating base. Next, the template-NTP base pair undergoes a major movement that places the triphosphate into the active site. These higher-fidelity enzymes therefore select for proper template-NTP pairing at two different binding sites, in contrast to the single site used by (+) strand RNA virus RdRPs.

Functions of Additional Polymerase Domains

Additional N- and C-terminal domains that surround the RdRP cores are often encoded in the genomes of larger RNA viruses (Fig. 6.13). The flavivirus RdRP has an extra N-terminal domain that has 5′-methyltransferase activity that contributes to mRNA capping. The core RdRPs of double-stranded RNA viruses are flanked by large N- and C-terminal domains. The former surrounds the fingers and thumb subdomains, closing the enzyme in a cage-like structure. The C-terminal domains are shaped like bracelets and resemble the sliding clamps that contribute to the efficiency of DNA polymerases. In contrast to other RdRPs, these enzymes have four channels. Two are in equivalent positions to the template and nucleotide entry channels of other RdRPs, but the other two serve as RNA exit pathways. One extends through the bracelet domain and is the pathway for release of new double-stranded RNA to the particle interior. The other serves to guide newly synthesized (+) single-stranded RNAs out of the core.

      The RdRP of (–) strand RNA bunyaviruses has an N-terminal endonuclease domain that is essential for procuring capped mRNA primers. The minimal RdRP of vesicular stomatitis virus is surrounded by three globular domains with three enzymatic activities required for mRNA 5′-cap synthesis: 2′-O-methyltransferase, guanine-N7-methyltransferase, and polyribonucleotidyl transferase.

      Not all RdRPs have other functions encoded in extra N- and C-terminal domains. The influenza virus RdRP consists of three individual polypeptides, PA, PB1, and PB2, each of which has the distinct activity described above.

RNA Polymerase Oligomerization

RNA polymerases of multiple (+) and (–) strand RNA viruses have been reported to form dimers and higher-order oligomers (Fig. 6.14). There is evidence that such arrangements may increase the stability and catalytic activity of these enzymes. In many cases, deletions of amino acids that prevent oligomerization also inhibit or cause complete loss of enzyme activity.

      The first poliovirus 3D^pol structure revealed that the polymerase molecules interacted in a head-to-tail manner and formed fibers; subsequently the protein was shown to form a lattice. The head-to-tail fibers were formed by an interface comprising parts of the thumb of one polymerase and the back of the palm of another. Amino acid changes in the back of the thumb that disrupt this interface impaired replication. Repetition of this interaction in a head-to-tail fashion results in long fibers of polymerase molecules 50 Å in diameter. The presence of a second interface, formed by N-terminal polypeptide segments, may lead to a network of polymerase fibers. These interacting N-terminal polypeptide segments may originate from different polymerase molecules and are required for enzyme activity. Intermolecular cross-linking has been observed between cysteines engineered at Ala29 and Ile441 of poliovirus 3D^pol, and disruption of these interactions led to reduced infectivity. Polymerase-containing oligomeric structures resembling those seen with purified 3D^pol were observed on the surface of vesicles isolated from poliovirus-infected cells. Because picornavirus RNA synthesis occurs on membranous vesicles, the concept of a catalytic lattice is attractive mechanistically.

Figure 6.13 Functional N- and C-terminal extensions of RNA polymerases. The smallest known RdRP is encoded in picornavirus genomes and consists of a core catalytic unit made of thumb (green), fingers (red), and palm (yellow) domains. An N-terminal extension of the flavivirus dengue virus RNA polymerase (tan) has methyltransferase activity. The rhabdovirus RNA polymerase has both N- (blue) and C-terminal (light gray) extensions; the latter contain the capping and methyltransferase domains.

Figure 6.14 Oligomerization of RNA-dependent RNA polymerases. Ribbon diagrams of dimers of murine norovirus RdRP (PDB ID: 3QID), tetramers of influenza virus RdRP (PDB ID: 3J9B), and cryo-electron microscopy density data showing tubular arrangement of sheets of poliovirus 3D^pol (EMD ID: 2270).

Template Specificity

      Viral RdRPs must select viral templates from among a vast excess of cellular mRNAs and then initiate correctly to ensure accurate RNA synthesis. Different mechanisms that contribute to template specificity have been identified. Initiation specificity may be regulated by the affinity of the RdRP for the initiating nucleotide. For example, the RdRPs of bovine viral diarrhea virus and bacteriophage ϕ6 prefer 3′-terminal C. Reovirus RdRP prefers a G at the second position of the template RNA. This preference is controlled by hydrogen bonding of carbonyl and amino groups of the G with two amino acids of the enzyme. Both preferences would exclude initiation on cellular mRNAs, the great majority of which end in poly(A).

      Template specificity may also be conferred by the recognition of RNA sequences or structures at the 5′ and 3′ ends of viral RNAs by viral proteins. RNA synthesis initiates specifically within a polypyrimidine tract in the 3′ untranslated region of hepatitis C virus RNA. The 3′ noncoding region of polioviral genomic RNA contains an RNA pseudoknot structure that is conserved among picornaviruses (Fig. 6.9).