Report

The Structure of Native Influenza Virion Ribonucleoproteins

See allHide authors and affiliations

Science  21 Dec 2012:
Vol. 338, Issue 6114, pp. 1634-1637
DOI: 10.1126/science.1228172

Abstract

The influenza viruses cause annual epidemics of respiratory disease and occasional pandemics, which constitute a major public-health issue. The segmented negative-stranded RNAs are associated with the polymerase complex and nucleoprotein (NP), forming ribonucleoproteins (RNPs), which are responsible for virus transcription and replication. We describe the structure of native RNPs derived from virions. They show a double-helical conformation in which two NP strands of opposite polarity are associated with each other along the helix. Both strands are connected by a short loop at one end of the particle and interact with the polymerase complex at the other end. This structure will be relevant for unraveling the mechanisms of nuclear import of parental virus RNPs, their transcription and replication, and the encapsidation of progeny RNPs into virions.

The influenza A viruses belong to the Orthomyxoviridae family and cause annual epidemics and occasional pandemics of respiratory disease. The latest pandemic was originated by low-pathogenic H1N1 viruses (1), but fears remain that high-pathogenic H5N1 avian viruses may cause a catastrophic pandemic. The viral genome contains eight single-stranded negative-polarity RNAs assembled into ribonucleoprotein particles (RNPs) containing the viral polymerase and multiple nucleoprotein (NP) monomers (2, 3); within each RNP, the RNA adopts a closed conformation by interaction of both ends with the polymerase. Viral RNPs act as independent molecular machines for transcription and replication in the nucleus, both processes being mediated by the heterotrimeric polymerase complex. Viral mRNAs are produced by cap-snatching, whereas RNA replication generates progeny RNPs (24), which are encapsidated in an ordered manner into virus particles (5, 6). The RNPs are flexible, helical particles (7, 8) with general features determined by the NP (9), which contain the polymerase at one end (10, 11) and a closing loop at the other. The atomic structure of isolated RNP components, including NP and several polymerase subunit domains, has been solved recently (3, 4, 12, 13), but its detailed quaternary organization to form virion RNPs remains unknown.

Previous studies using recombinant RNPs identified circular, elliptical, or helical structures depending on the RNA length (14). Circular RNPs were studied by cryogenic electron microscopy (cryo-EM), providing the first description of biologically relevant NP-NP and NP-polymerase interactions (15, 16). However, these particles lacked the helical features typical of virion RNPs that might be important for proper replication, transcription, and packaging. Unfortunately, the length heterogeneity and high flexibility of virion RNPs have hampered any structural analysis so far. To overcome these problems, we performed separate analyses of termini and central regions of RNPs isolated from purified A/WSN virions (Fig. 1A) by either negative staining or cryo-EM. Images from central regions or RNP termini were separately classified by using a pattern-free algorithm, and representative two-dimensional (2D) averages are presented in Fig. 1, B to D, and fig. S1. The class-average images from central regions were compatible with a helical structure (Fig. 1B), whereas those obtained from RNP termini suggested two possible structures: one that might contain the polymerase complex (Fig. 1C) whereas the other could correspond to the closing loop (Fig. 1D).

Fig. 1

Structure of influenza virion RNPs. (A) Gallery of images of virion RNPs obtained from negative-stained samples. The blue and red squares illustrate the areas used for processing helical and terminal sections of the RNPs, respectively. (B) Two-dimensional average obtained from images of frozen-hydrated samples of RNP central regions (contrast inverted for comparison). (C and D) Two-dimensional averages of polymerase-containing and loop-containing RNP termini, respectively. (E) Three-dimensional cryo-EM reconstruction of the central portion of a virion RNP showing the two opposite-polarity NP strands in red and blue. (F) Docking of two NP monomers showing the interstrand connection. (G) Docking of two NP monomers showing the intrastrand connection. One of the monomers is shown in dark blue to illustrate the swapping loop and the distances of the flexible connections. Scale bar represents 100 Å in (B) to (D) and 50 Å in (E) to (G).

Several classes from central sections of frozen-hydrated RNPs were composed of straight RNPs images, whereas other showed curved structures not amenable to further analysis. To perform the 3D reconstruction, we obtained an estimation of helical parameters from a symmetry-free refinement using images of straight RNPs and a featureless cylinder as starting model (17). Upon refinement using the iterative helical real space reconstruction protocol (IHRSR) (18), reconstructions of various classes converged to quasi-equivalent structures showing a dihedral helix with two opposite-polarity NP strands (Fig. 1E and fig. S2A). The double-helical structure defined a minor groove between the connected NP strands and a major groove between the NP strands not physically in contact. The helical parameters reached upon convergence of all structures (fig. S2B) were very similar: The rise step per monomer was constant at 28.4 Å, and the rotation angle between monomers ranged from 57° to –64°, forming helices with around 12 NP monomers per turn (six pairs). The basic features of the NP structure, including the head and body domains, were visible in the helical structure and could be compared to the mini-RNP structure (15) (fig. S3). The docking of the NP atomic structure (13) into the helix places the head domain of each NP facing the major groove and the N-terminal regions contacting each other in the minor groove (Fig. 1, F and G). The excess density at the NP-NP interstrand connection could derive from the first 22 NP amino acids not represented in the crystal structure (fig. S4). For intrastrand NP-NP interaction, the 402-to-428 loop would be inserted into the neighboring NP in a way equivalent to that found in NP crystallographic structures (Fig. 1G). The high flexibility in the connections of the loop to the rest of the NP (fig. S5) allows the various dispositions among NP monomers observed in the helical RNPs, the crystal structures (12, 13), and the mini-RNP (15). Although the atomic structure of NP is conserved in the helical RNP and the distance between the loop and the body, 25 and 41 Å (Fig. 1G), is compatible with extended amino acid chains, some local rearrangements of the structure could not be ruled out.

To analyze the interstrand interaction, we compared the replication in vivo of a helical chloramphenicol acetyltransferase (CAT)-expressing RNP or a circular mini-RNP (15), in which the NP N terminus is not involved in NP-NP interaction. Because no atomic structure is available for the NP N terminus (12, 13), we assayed mutants lacking amino acids 2 to 8 (mutant N7) or 2 to 21 (mutant N20). Mutation N7 significantly reduced the accumulation of CAT RNPs but only slightly that of mini-RNPs, whereas mutation N20 strongly reduced replication of both (fig. S6). However, no gross alteration in the helical structure of recombinant RNPs containing N7 mutant NP was observed (fig. S6E). The relevance for intrastrand interaction of the critical residues at positions 412 and 416 at the swapping loop was also verified in both helical and circular RNPs (15) (fig. S6).

To determine the structure of the RNP termini, we performed a direct 3D classification of images from negative-stained samples using the maximum-likelihood procedures of Xmipp (19) and a filtered helical segment as initial model. The classification yielded six groups, two of which were assigned to the terminal loop and merged to generate the final structure (Fig. 2A). The electron density map showed a small loop where only three NPs are needed to connect the two helix strands (Fig. 2A, yellow). These results are consistent with the NP ability to form small, closed structures (9). Two other image groups showed the presence of the polymerase and revealed the existence of two conformations of the complex. The main structural difference is illustrated in gray (Fig. 2B) and suggested a rotation and displacement of the gray and green portions of the polymerase (Fig. 2B, diagram). The proportion of both conformations is roughly 50%, and their biological relevance is not clear. When compared to the polymerase in a mini-RNP (15), all structures showed the same appearance with a large central groove, but part of the polymerase is twisted and tilted in the helical RNPs in comparison with the mini-RNP (fig. S7, green). These structural changes are probably due to the minimal length of the mini-RNP template, which does not allow the formation of a helix.

Fig. 2

Three-dimensional structure of the termini of influenza virion RNPs. (A) Structure of the terminal loop of a virion RNP juxtaposed to a portion of the helical section as a reference. The three NPs present in the loop are depicted in yellow, and the NP atomic structure is docked into one of them to serve as a size guide. (B) Structure of the polymerase-containing terminus of a virion RNP juxtaposed to a portion of the helical section as a reference. Two conformations of the polymerase are shown in top and bottom images, and the structural changes are outlined on the right.

The structure of the RNPs inside virions was determined by cryogenic electron tomography and subvolume averaging. The spikes, the envelope, and the inner M1 layer were visible as were individual RNPs in the internal density (Fig. 3A). RNP segments were selected and subjected to iterative alignment and averaging to improve resolution and to avoid the information loss resulting from the missing wedge. The helical parameters were determined in the averaged subvolume by using Xhelicals (18), and the values obtained at the minimum, a rotation angle of –64° and a rise step of 28.5 Å per monomer (fig. S8), were used to apply helical symmetry, although dihedral symmetry was not imposed (Fig. 3B). The symmetrized subvolume was similar to soluble RNPs and showed a major and a minor groove, verifying that the structure of the RNPs inside virions is the same as the isolated ones. The final volumes were then placed back into the cryo-tomograms by using the coordinates and orientation of the helical axis (Fig. 3C), verifying their close interaction in bundles (20, 21).

Fig. 3

Three-dimensional structure of influenza RNPs within intact virions. (A) Central sections of cryo–electron tomograms of frozen-hydrated A/WSN virions. RNPs are highlighted in red rectangles. (B) Averaged volume obtained after alignment and symmetrization of 288 volumes extracted from tomograms. (C) Volumetric representation of the segmented influenza virus showed in the bottom right image in (A). Averaged RNP volumes were placed back on the position and orientation computed in the alignment [densities marked with purple arrow heads in (A)]. Scale bar represents 1000 Å.

By a combination of cryo-EM and cryo–electron tomography, we show the complete structure of native influenza virus RNPs. The composite model resulting from the juxtaposition of the polymerase-containing terminus, the helical internal portion, and the terminal loop portrays the most common of the quasi-equivalent structures present in the sample (Fig. 4A). Because this RNP structure derives from purified virions, it represents the incoming RNPs in the infected cell and the RNPs to be encapsidated in progeny virus. This RNP structure has implications for various steps in the infection cycle. After uncoating, parental RNPs are transported to the nucleus in a movement governed by two nuclear localization signal sites present in NP (2224) whose locations in the RNP structure document their accessibility, in agreement with previous results (23, 25). The positions of the RNA template was inferred from the surface potential of the docked NP monomers and the positions of mutations that reduce the RNA binding capacity of NP (12). The RNA could be located along the positively charged path present in each NP strand, as shown in alternative views of three NP monomers from the same strand (Fig. 4B and movie S1). The length of the RNA included into the path was around 140 Å per NP, in agreement with the 20 to 24 nucleotides per NP value previously estimated (7, 14, 16). Also, the RNP structure described here imposes constraints and suggests mechanisms for virus transcription and replication. The disposition of the RNA along the NP strand implies that the interstrand interaction should be disrupted for template copy, but the polymerase may only require local RNP destabilization because the template RNA is accessible to the solvent (26). The NP N terminus appears important for in vivo RNP replication (fig. S6) but not for in vitro transcription efficiency (fig. S9). These results are consistent with the trans/cis model for replication/transcription of virion RNPs (3, 27), which imply the replication of the template RNP by a nonresident polymerase while the resident polymerase would transcribe in cis. Mutation of the NP N terminus could affect recognition of the helical RNP or destabilize the binding of a nonresident polymerase, without affecting template accessibility for the resident polymerase during transcription. The progeny RNPs are assembled by protection of the nascent 5′ RNA terminus by an additional polymerase and sequential incorporation of NPs (3, 27, 28), but it is unclear how they would adopt the double-helical structure. One possibility is that the association of assembling and replicating polymerases during RNA synthesis (29) facilitates the 3′- and 5′-RNA termini interaction and induces a zippering reaction along the RNP. Moreover, the structure of helical RNPs suggests a potential mechanism for the generation of influenza defective-interfering particles. Because the NP strands are in close proximity along the structure, the replicating polymerase could easily switch strands and continue reading the template at a disparate position in its primary sequence, leading to internal-deletion defective RNPs (30).

Fig. 4

Model for the structure of an influenza virus helical RNP. (A) Composite volume generated by combination of the central portion of an RNP and the RNP ends showing the polymerase (green and brown) and the terminal NP loop (yellow). This structure represents the complete model of the segment 8 RNP. (B) Model for the localization of the viral RNA (yellow thread). Four views of the same NP strand are represented as surface potential and show, highlighted in blue, the residues whose mutation reduce NP-RNA interaction (12) (C) Model for the localization of the template RNA (yellow thread) in the helical structure of an RNP, showing one of the NP strands as surface potential and the opposite strand as a ribbon representation.

Lastly, encapsidation into progeny virions is an ordered process, whereby the eight different RNPs associate to become enveloped at the plasma membrane, and relies on specific cis signals at the 5′- and 3′-proximal regions of each RNA (5, 6). According to the structure presented here, only a fraction of the RNA sequence would be exposed at the surface of the RNP, and the precise position of each cis signal would translate into a specific spatial disposition in the RNP that could determine its interaction with other(s) RNP(s). The combination of mutational data and the structure shown here will allow understanding the packaging code of influenza virus and will provide insight into how reassortment among influenza viruses of different subtypes occurs.

Supplementary Materials

www.sciencemag.org/cgi/content/full/science.1228172/DC1

Materials and Methods

Figs. S1 to S9

References (3147)

Movie S1

References and Notes

  1. Information on materials and methods is available on Science Online.
  2. Acknowledgments: The technical assistance of Y. Fernández and N. Zamarreño is gratefully acknowledged. We thank D. Luque for help with the generation of movies. This work was supported by the Spanish Ministry of Science and Innovation (Ministerio de Ciencia e Innovación) grants BFU2010-17540/BMC (J.O.) and BFU2011-25090/BMC (J.M.-B.), by the FLUPHARM strep project (FP7-259751) (J.O.), and by Fundación Marcelino Botín (J.O.). J.M.-B. and J.O. designed the experiments. R.A., J.M-B, R.C. F.J.C. and J.J.C. carried out the experiments. J.O. and J.M.-B. wrote the paper with contributions from R.A., R.C., F.J.C., and J.J.C. and discussions with J.L.C. and J.M.V. The data presented in this manuscript are tabulated in the main paper and in the supplementary materials. EM maps have been deposited in the EM Data Bank (www.ebi.ac.uk/pdbe/emdb/), accession numbers EMD-2205, 2206, 2207, and 2208. The atomic coordinates of the docking have been deposited in the Protein Data Bank (www.rcsb.org/), accession code 4bbl.
View Abstract

Navigate This Article