INTRODUCTION
Influenza A virus, the causative agent of human pandemics and annual epidemics, is a negative-sense, single-stranded RNA virus in the
Orthomyxoviridae family. Within the pleomorphic lipid envelope of the virion are eight segments of virion RNA (vRNA). Each vRNA segment interacts with multiple nucleoproteins (NPs) and a heterotrimeric polymerase complex (3P, comprised of PA, PB1, and PB2) to form a viral ribonucleoprotein (vRNP) that functions in transcription, replication, and packaging of the viral genome (
1). Approximately 24 nucleotides (nt) of vRNA associate with each NP molecule (
2,
3). Electron microscopy showed that influenza virus RNPs are not linear but rather are hairpin-like double-helical structures (
2,
4–6). The viral polymerase, which binds to the common 3′- and 5′-terminal sequences of the vRNA segments, is located at the hairpin termini (
7). The location of the viral polymerase at the juxtaposed 3′ and 5′ termini was also observed in a reconstituted mini-RNP (
8). In contrast to influenza virus RNPs, the RNPs of nonsegmented, negative-strand RNA viruses (e.g., rhabdoviruses) are linear and form single coils (
9). The molecular basis for the double-helical structure of influenza virus RNPs has not been determined.
The atomic structures of the NPs of two influenza A virus strains, H1N1 (influenza A/WSN/33 [WSN]) and H5N1 (influenza A/Hong Kong/483/97 [HK]), both in the form of a trimer, have been determined to 3.2- and 3.5-Å resolutions, respectively (
10,
11). These two structures show that NP-NP interaction is mediated largely by a tail loop consisting of amino acid residues from positions 402 to 429. A highly positively charged groove is found at the exterior of the NP trimer, indicating that RNA is bound at the outer periphery of the viral RNPs (
4,
12). This mode of RNA binding is consistent with previous findings that the viral RNAs in influenza virus RNPs are readily digested by RNase and are exposed to the solvent (
12) and that polyvinylsulfate (PVS), a negatively charged polymer, is able to completely displace RNA from influenza virus RNPs (
13). The putative RNA-binding groove of the influenza NP is lined with a large number of basic residues scattered along the NP primary sequence (
10). Mutagenesis analysis identified two polypeptide regions in the groove that are essential for RNA binding, one containing residues R74 and R75 and the other containing residues R174, R175, and R221 (
11,
14,
15).
NP that is not associated with RNA is required for viral RNA replication, which occurs in two steps (
16). First, a full-length copy of vRNA, termed complementary RNA or cRNA, is made and is then copied to produce vRNA. Whereas viral mRNA synthesis is initiated with capped RNA primers derived from cellular pre-mRNAs, the synthesis of cRNA and vRNA is initiated without a primer. Three amino acids (R204, W207, R208) in the loop at the top of the head domain of NP are required for both its binding to the viral polymerase and its ability to support viral RNA synthesis catalyzed by the viral polymerase (
17). Several roles for the NP-polymerase interaction have been proposed, but no role has been definitively established (
18–22). NP is deposited along the newly synthesized cRNA and vRNA chains during viral RNA synthesis (
23,
24), so that cRNA and vRNA synthesis and RNP assembly are coupled. This coupling leads to the selective binding of NP to cRNA and vRNA but not to viral mRNA or cellular RNAs in infected cells. By itself, NP binds RNA nonselectively
in vitro (
12).
It has not been determined how the transition of NP from monomer to oligomer is regulated in infected cells. Several observations suggested that the oligomeric state of NP might be the predominant form of NP. For example, oligomeric ring structures of NP were predominantly present in purified, recombinant NP that was overexpressed in both
Escherichia coli and insect cells (
10,
11). In addition, when vRNA was removed from vRNPs, a large proportion of the NP molecules still remained associated with each other in an RNP-like structure (
25,
26). Because of these findings, Ruigrok and Baudin (25) raised an intriguing issue about RNP assembly: “it will be interesting to know whether NP in the infected cells is monomeric and, if so, whether other proteins are involved in preventing NP polymerization.” In comparison, members of the
Mononegavirales often encode a phosphoprotein, P, that keeps their N proteins in a soluble form prior to RNP encapsidation (
27), but a comparable viral protein is lacking in orthomyxoviruses.
To resolve this issue and to provide a coherent model for the assembly of the double-helical influenza virus RNP structure, we performed structural and biochemical characterization of NP in three different molecular forms, namely, oligomer, dimer, and monomer. In particular, we solved the crystal structure of an NP dimer to 2.8-Å resolution, which revealed a previously unknown NP-NP interface that is likely responsible for organizing the double-helical viral RNP structure. Mutational disruption of the NP dimer interface resulted in NP molecules that were not able to support viral RNA synthesis, as determined in minigenome assays, indicating the biological significance of the NP dimer interaction that organizes influenza virus RNPs into their double-helical structure.