INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The genome of influenza A virus is composed of eight strands of negative-sense RNA, which in virions and infected cells occurs in the form of ribonucleoprotein (RNP) complexes with the virus-encoded PB1, PB2, and PA proteins and nucleoprotein (NP). The RNP complexes represent the functional templates for influenza virus RNA synthesis, and during infection they contain both negative-sense (vRNA) and positive-sense (cRNA) replicative intermediate RNA but not virus mRNA (reviewed in reference 31). The four RNP-associated proteins are the only viral polypeptides necessary for transcription and replication of a genome segment (24).

PB1, PB2, and PA associate to form a highly regulated RNA-dependent RNA polymerase capable of switching between two modes of transcription to produce either capped and polyadenylated mRNA or replicative copies of the genome segments (31). In contrast, the major function associated with NP is a nonenzymatic single-strand-RNA-binding activity, which has no apparent sequence specificity (3, 27, 43, 49). This RNA-binding activity is reflected in the stoichiometric quantity of the protein in RNPs, from which it has been estimated that one polypeptide interacts with around 20 nucleotides of RNA (11). Although NP encapsidates the RNA, it does not protect it from digestion with RNase (3, 14, 39), suggesting that the RNA is bound on the outside of the structure. NP is apparently responsible for the higher-order structure of RNPs, which occur in the form of helices twisted back into a double-helical hairpin (25, 39), since the structures can persist following removal of the RNA (39, 41). This implies protein-protein contacts between NP monomers (41), which is consistent with the observed cooperative nature of the NP-RNA interaction (49).

Other functions have been postulated for NP; although the polymerase complex can efficiently transcribe short templates, NP is required for synthesis of longer RNAs, leading to the suggestion that it may act as a processivity factor (23). NP may also play a role in the regulation of polymerase activity, since cRNA and vRNA (but not mRNA) synthesis depends on a supply of soluble (i.e., not bound to RNA) NP (4, 45). Intriguingly, NP has recently been shown to make direct protein-protein contacts with the PB1 and PB2 subunits of the polymerase (5, 35), but whether these contacts play a role in regulating polymerase activity is not yet known. NP also interacts with at least two cellular proteins. As expected for a protein that contains nuclear localization signals, it interacts with polypeptides of the importin  family (36, 47), and this proximal interaction is capable of directing the nuclear import of NP-RNA complexes (37). NP also binds filamentous actin, an interaction that influences the cellular localization of NP (13).

A molecular characterization of this multifunctional protein is therefore of some interest, with an analysis of how the polypeptide interacts with RNA being a logical starting point. Two studies have identified an amino-terminal region of NP that is capable of binding RNA (1, 29), but this fragment binds RNA with much lower affinity than the intact polypeptide does, suggesting that other NP sequences are important for high-affinity binding (1). Also, little information is available at the amino acid level on how NP interacts with RNA, and the protein does not contain sequences with homology to previously characterized RNA-binding motifs (1, 9). However, the protein as a whole is very basic (48), and this has led to a widely accepted hypothesis that the positive charges of lysine and arginine residues are involved in contacting the negatively charged phosphate backbone. Since our long-term aim is to understand the mechanisms by which NP exerts its regulatory effects on polymerase transcription, we decided to further investigate how the protein interacts with RNA. Our strategy was to first identify which regions of the full-length polypeptide make direct contact with RNA, to then test the importance of various classes of amino acid for binding, and to finally employ site-directed mutagenesis to identify the important residues. We show here that direct protein-RNA contacts occur throughout the length of the polypeptide and not just at the N terminus. Intrinsic fluorescence spectroscopy and chemical modification experiments indicate the involvement of tryptophan and arginine residues in binding RNA, and consistent with this, mutational analysis identified two arginine and six aromatic residues essential for RNA binding. These mutationally sensitive residues were not localized to any particular region of the polypeptide but were found throughout the protein. Thus, high-affinity RNA binding by NP requires contributions from multiple regions of the polypeptide.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmids, RNA, antisera, and viruses. Plasmids capable of directing the expression of authentic NP in eukaryotic cells (pKT5) or fused to maltose-binding protein (MBP) (pMAL-NP) in Escherichia coli have been described previously (13). Certain NP point mutants have been described previously (13), while others were constructed by oligonucleotide-directed mutagenesis of plasmids pKT5 or pMAL-NP by standard procedures (30). For in vitro RNA-binding assays, a radiolabelled 178-nucleotide synthetic RNA target corresponding to influenza virus A/PR8/34 segment 8, but with nucleotides 84 to 795 (inclusive) deleted, and a C-to-A transversion of the penultimate base was transcribed from plasmid pKT8-3'5' as previously described (13). Rabbit antisera raised against amino acids 340 to 498 of A/PR8/34 NP (13) have been described previously. Recombinant vaccinia viruses expressing the three influenza virus A/PR8/34 P proteins (46) or bacteriophage T7 RNA polymerase (vTF-7) (18) have been described previously.

Expression and purification of NP. NP fused to MBP-NP was purified from extracts of E. coli TG1 containing plasmid pMAL-NP by affinity chromatography on amylose resin (New England Biolabs) as previously described (13). To remove the MBP moiety, the fusion protein was digested with 0.5% (wt/wt) factor Xa protease (New England Biolabs). After an 18-h incubation at 14°C, the protein was loaded onto a MonoQ ion-exchange column (Pharmacia) equilibrated with 50 mM Tris-Cl (pH 7.6), and the column was eluted with a linear gradient of 0 to 2 M NaCl in 50 mM Tris-Cl (pH 7.6). MBP eluted mainly in the flowthrough fractions, while NP eluted as a broad peak at around 800 mM NaCl.

RNA filter-binding and UV cross-linking assays. Filter-binding assays were performed by incubating protein samples with 20 fmol of RNA (around 5,000 cpm) in 25 mM Tris-Cl (pH 7.6)-50 mM NaCl-5 mM MgCl₂-0.5 mM dithiothreitol DTT-5% glycerol at room temperature (unless otherwise specified) for 20 min. The reaction mixtures were passed through nitrocellulose filters equilibrated in 20 mM Tris-Cl (pH 7.6)-50 mM NaCl and washed three times with 200 µl of the same buffer. Bound radioactivity was quantified by liquid scintillation counting. For UV cross-linking experiments, reaction mixtures (generally containing 125 nM MBP-NP) were incubated for 20 min as above and then irradiated at 254 nm for 5 min at an intensity of 4 mW/cm² in a Spectronics XL-1500 UV cross-linker. Free RNA was removed by digestion with 2 µg of RNase A per ml for 30 min at room temperature before being subjected to analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and autoradiography.

Chemical modification of NP. All modification procedures included control protein samples that were treated in parallel as described below, except that the reactive chemical was omitted. Spectrophotometric measurements were taken with reactions carried out in the absence of protein as a baseline.

1

1

1

1

Fluorescence spectroscopy. Intrinsic fluorescence measurements were recorded at 25°C in a Perkin-Elmer LS 50 B luminescence spectrometer with a thermostatically controlled cuvette holder. Wavelength scans were taken from protein samples diluted to 1 µg/ml in 50 mM NaCl-20 mM Tris-Cl (pH 7.6) with an excitation wavelength of 280 nm (2.5-nm slit width) and an emission slit width of 5 nm. Scans were averaged from at least five repetitions. Quenching experiments were performed with 5 µg of protein per ml, allowing a 5-min equilibration time after the addition of RNA aliquots. Fluorescence output was then integrated over 1 min, before the addition of further RNA.

Polymerase-binding assays. Xenopus laevis oocytes were maintained and microinjected with synthetic mRNAs encoding the influenza virus P proteins (transcribed from plasmids pKT1 to pKT3) as previously described (6, 12). For precipitation reactions, 2.5 µl of oocyte lysate (corresponding to one-quarter of an oocyte) was incubated with 0.5 µg of fusion protein in 100 µl of IP buffer (100 mM KCl, 50 mM Tris-Cl [pH 7.6], 5 mM MgCl₂, 1 mM dithiothreitol, 0.1% Nonidet P-40) for 1 h at room temperature. A 50-µl volume of a 50% (vol/vol) slurry of amylose resin (New England Biolabs) in phosphate-buffered saline was added, and incubation was continued for a further 30 min with gentle mixing. The solid phase was collected by centrifugation and washed three times with 750 µl of IP buffer. Bound material was eluted by boiling in 40 µl of SDS-PAGE sample buffer and analyzed by SDS-PAGE and autoradiography.

Influenza virus gene expression assay. For in vivo RNA transcription assays, a synthetic influenza virus genome segment (flu-CAT) containing an antisense chloramphenicol acetyltransferase (CAT) gene was produced by in vitro transcription of plasmid pPB2CAT9 (a generous gift of Mark Krystal). BHK cells (in 35-mm-diameter dishes) were infected for 2 h at 37°C with recombinant vaccinia viruses expressing the three subunits of the influenza virus polymerase (PB1-VAC, PB2-VAC, and PA-VAC) (46) and the bacteriophage T7 RNA polymerase (VTF-7) (18) at a multiplicity of infection of 5 of each virus per cell. The cells were washed three times with serum-free medium before being subjected to transfection with up to 1 µg of plasmid pKT5 encoding NP (or mutant derivatives), 0.5 µg of pPB2CAT9 in vitro-transcribed RNA, and 10 µg of a cationic liposome mixture (Escort; Sigma-Aldrich) as specified by the manufacturer. The cells were incubated at 37°C for 20 h, washed three times with ice-cold phosphate-buffered saline, and solubilized in 1 ml of CAT enzyme-linked immunosorbent assay lysis buffer (Boehringer Mannheim). The lysate was clarified by centrifugation at 14,000 × g for 10 min at 4°C, and CAT expression was quantified relative to known standards by a commercial enzyme-linked immunosorbent assay (Boehringer Mannheim).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of NP sequences in direct contact with RNA. Previously, we have shown that while MBP alone does not detectably bind RNA, a fusion of E. coli MBP and influenza virus NP (MBP-NP) binds RNA with similar affinity to authentic NP purified from influenza virus virions (K_d  20 nM) (3, 13). To further examine the interaction of NP with RNA, we digested the fusion protein with factor Xa protease to separate the MBP and NP domains and purified the NP to near homogeneity by ion-exchange chromatography (Fig. 1b, lane 1). Purified NP and MBP-NP were then tested for their RNA-binding activity by using a UV cross-linking assay. Protein samples were incubated with a radiolabelled RNA and subjected to UV irradiation to form cross-links between molecules in direct contact. Following removal of free RNA by digestion with RNase, polypeptides were separated by SDS-PAGE, visualized by staining with Coomassie blue dye, and examined for bound RNA by autoradiography (Fig. 1a). Both MBP-NP and NP became radiolabelled when subjected to this procedure (lanes 8 and 10, respectively) but did not acquire radiolabel when incubated with RNA in the absence of UV irradiation (lanes 6 and 7). In addition, binding could be competed with unlabelled RNA from viral and cellular sources (data not shown), indicating that the reaction was specific and that the recombinant NP showed similar sequence-independent binding properties to those of authentic NP (3, 19, 49). Moreover, when MBP-NP was irradiated in the presence of RNA and subsequently incubated with factor Xa protease to separate the MBP and NP moieties, only NP was radiolabelled (Fig. 1a, lane 9). Therefore, the RNA-binding activity of MBP-NP depends solely on the NP portion of the fusion protein, and UV cross-linking provides a convenient and specific means of examining this activity.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   RNA-binding activity of recombinant NP. (a) UV cross-linking analysis of RNA binding by MBP-NP and NP. Protein samples were incubated with radiolabelled RNA and subjected to UV irradiation. Excess RNA was removed by RNase digestion, and samples were separated by SDS-PAGE. Radiolabelled proteins were detected by autoradiography (32P; lanes 6 to 10), and total protein was detected by staining with Coomassie brilliant blue (Coomassie; lanes 1 to 5). Lanes: 1 and 6, NP; 2 and 7, MBP-NP (samples in lanes 1 and 2 were not irradiated); 3 and 8, MBP-NP; 4 and 9, MBP-NP incubated with factor Xa protease after cross-linking; 5 and 10, NP. Open arrowheads mark the indicated polypeptides. Solid arrowheads mark a contaminant protein in the RNase preparation (top) or factor Xa (bottom). Also shown are the molecular masses (in kilodaltons) of marker proteins. (b) Chemical fragmentation of NP. NP was cross-linked to a radiolabelled RNA and analyzed by SDS-PAGE before () or after (+) incubation in 70% formic acid. Total and radiolabelled proteins were detected as in panel a or by Western transfer to a polyvinylidene difluoride membrane and immunoblotting with an antiserum directed against the C terminus of NP (-C) or by amino acid sequencing. Major NP fragments are labelled i to v, and the N-terminal residue (NH2 residue) is given. (c) Formic acid cleavage map of NP. Numbered arrows indicate the positions of aspartyl-prolyl dipeptides in NP, and the bracket gives the sequences against which anti-NP sera were raised. Thin lines represent the identities of the numbered NP fragments.

Fluorescence spectroscopy of NP. Next, we wished to examine the involvement of specific classes of amino acids in NP for RNA binding. There is much evidence from experiments with other nucleic acid-binding proteins that aromatic amino acids can interact with single-stranded nucleic acids either by polar interactions or by planar stacking with exposed bases (2, 7, 8, 20, 26, 38, 40, 44). To test whether this was the case for NP, we used fluorescence spectroscopy. Tryptophan and to a lesser extent tyrosine residues absorb light at wavelengths around 280 to 300 nm and subsequently emit some of the absorbed energy at an intensity and wavelength dependent on the environment of the fluorescing side chain (42).

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Spectroscopic analysis of the interaction of NP with RNA. (a) Fluorescence spectra of native and denatured NP. Spectra of recombinant (thick line) and virion (thin line) NP were recorded in the absence and presence (recombinant NP only [dashed line]) of 8 M guanidinium chloride. All spectra have been scaled to the same maximum intensity. (b) Effect of RNA on the intrinsic tryptophan fluorescence of NP. Solutions of NP in the presence (thin line) and absence (thick line) of saturating amounts of RNA were excited at 296 nm, and fluorescent emission was measured at the plotted wavelengths. (c) Titration of fluorescence quenching by RNA. Aliquots of RNA (plotted in terms of nucleotides of RNA added per protein molecule) were added to a solution of MBP-NP, and fluorescence emission was measured at 335 nm after excitation at 280 nm. (d). Salt-dependent reversal of fluorescence quenching. Aliquots of NaCl were added to a sample of NP containing saturating amounts of RNA, and fluorescence emission was measured as in panel c.

Chemical modification of NP. NP as a whole is extremely basic (containing 21 lysine and 49 arginine residues [48]), consistent with the hypothesis that positive charges on the protein interact with the negatively charged phosphate backbone of RNA (19). To test this, we chemically modified the protein with reagents specific for arginine or lysine residues. We chose treatments that could be used under relatively mild buffer conditions, since this approach would be more likely to preserve the polypeptide structure (avoiding nonspecific loss of RNA-binding activity) and therefore would preferentially modify surface-exposed residues. Also, to be able to monitor the degree of protein modification, we chose reagents that allowed spectrophotometric quantification of the reaction. However, since these procedures required relatively large quantities of protein for accurate determinations, we used MBP-NP as a substrate since it was easier to prepare the necessary quantities of the fusion protein and since the MBP moiety had no apparent effect on the affinity of the polypeptide for RNA (13).

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Chemical modification of basic residues in MBP-NP. (a and b) Modification of arginine side-chains with p-HPG. (a) Samples treated (T) or mock treated (M) with p-HPG were analyzed by SDS-PAGE and staining with Coomassie blue. The arrow indicates MBP-NP. Also shown are values measured for the amount of derivatization of the polypeptide (as a percentage and as residues modified per protein). Sizes of molecular mass markers (in kilodaltons) are also shown. (b) The ability of the polypeptides to bind to a radiolabelled RNA was measured in a nitrocellulose filter-binding assay. (c and d) Modification of lysine side chains with TNBS. (c) Samples were treated (or mock treated) with TNBS at the indicated pH values and analyzed as in panel a. (d) The RNA-binding activity of TNBS-treated samples was measured as in panel b. (e and f) Modification of lysine side chains with acetic anhydride. (e) Samples were treated (or mock treated) with 60- and 250-fold molar excesses of acetic anhydride and analyzed as in panel a. (f) The RNA-binding activity of acetylated samples was measured as in panel b.

RNA-binding activity of mutant NP polypeptides. The experiments described above implicated tryptophan and arginine side chains as being important in RNA binding. However, they did not address the question of which particular residues are involved. Since NP contains only 6 tryptophans (48) (Table 1), we mutated each codon in an attempt to define the residue(s) involved in RNA binding. A similar approach did not seem feasible for arginine residues, since the protein contains 49. However, by examining sequence alignments of NPs from influenza A, B, C, and Dhori orthomyxoviruses (data not shown), we were able to identify several conserved arginine residues at positions 8, 150, 156, 175, 199, 204, 208, 213, 267, 391, and 416 (Table 1). These and the tryptophan residues encoded by pMAL-NP were altered to alanine. In addition, to further probe the importance of aromatic amino acids in binding RNA, we mutated conserved tyrosine and phenylalanine residues at positions 148 and 412, respectively (Table 1).

View larger version (45K): [in this window] [in a new window]   FIG. 4.   RNA-binding activity of NP mutants containing single-amino-acid changes. WT and mutated MBP-NP (as labelled) were expressed and purified from E. coli. Protein samples were adjusted to contain approximately equal amounts of full-length fusion protein and assayed for their ability to bind radiolabelled RNA in a UV cross-linking assay. Polypeptides were analyzed by SDS-PAGE and autoradiography (a) and staining with Coomassie brilliant blue (b). The autoradiograms in panel a were taken from the stained gels in panel b. Also indicated by arrowheads are molecular size markers (in kilodaltons) and the position of MBP-NP.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   Effect of temperature on the NP-RNA interaction. (a and b) MBP-NP fusion proteins were prepared from E. coli cultures grown at the indicated temperatures and assayed for their ability to bind a radiolabelled RNA as in Fig. 4. (a and b) The autoradiogram (a) and the equivalent Coomassie brilliant blue-stained gel (b) are presented. (c) Thermal denaturation of RNA-binding activity. The indicated polypeptides (100 nM) were tested for their ability to bind a radiolabelled RNA (1 nM) at the indicated temperatures by filtration through a nitrocellulose filter. Protein was heated in the presence (solid symbols) or absence (open symbols) of RNA. Values are plotted as a percentage of the amount bound by each protein at 30°C.

Interaction of the NP RNA-binding mutants with the influenza virus polymerase. The mutational analysis described above identified specific residues whose substitution reduced RNA-binding activity, potentially through direct alteration of an amino acid side chain normally in contact with RNA. Alternatively, some or all of the alterations may have exerted an indirect effect on RNA binding by perturbing the polypeptide structure, a possibility which seemed especially plausible for the mutations which rendered RNA binding ts. However, defects which caused gross malfolding of the polypeptide structure would be expected to affect other functions of NP. Previously, we have shown that the W120 and R267 mutations have minimal effect on the ability of NP to bind filamentous actin (13). Similarly, the W120, W139, R267, W330, W386, and F412 mutations have little effect on the ability of the polypeptides to form NP-NP oligomers (16). Recent work has shown that NP also makes direct protein-protein contacts with the PB1 and PB2 subunits of the influenza virus polymerase (5, 35). We therefore tested the ability of the RNA-binding mutants to interact with the polymerase complex. Previously, we have shown that the three influenza virus P proteins assemble into a complex in X. laevis oocytes microinjected with the appropriate mRNAs and that this system provides a convenient source of soluble radiolabelled protein for binding studies (6, 12, 35). Accordingly, oocyte lysates containing the polymerase complex were incubated with MBP or with WT and mutant MBP-NP polypeptides (synthesized at 37°C). After the addition of amylose resin, the solid phase was collected, washed, and examined for bound polypeptides. All three P proteins were precipitated by WT MBP-NP but not by MBP alone (Fig. 6, lanes 2 and 3), indicating a specific association between NP and the polymerase complex. Furthermore, similar quantities of the polymerase complex were precipitated by all the mutant MBP-NPs defective for RNA binding (lanes 4 to 10). Therefore the point mutations which disrupt RNA-binding activity are not associated with the general loss of other NP functions.

View larger version (61K): [in this window] [in a new window]   FIG. 6.   Ability of RNA-binding mutants to interact with the polymerase complex. Radiolabelled cell lysate containing the three influenza virus P proteins was analyzed by SDS-PAGE and autoradiography before (Total) or after binding to the indicated polypeptides. Open arrowheads indicate molecular mass markers (in kilodaltons); solid arrows are as labelled.

Ability of the NP RNA-binding mutants to support virus gene expression. NP is known to be essential for virus RNA synthesis (reviewed in reference 31), and an artificial system has been established where NP and the three subunits of the influenza virus RNA polymerase supplied from recombinant vaccinia viruses are sufficient to drive transcription and replication of a synthetic influenza virus genome segment in cells (24). We used an adaptation of this system where NP is supplied from transfected plasmids (13) to test the effects of point mutations disruptive of RNA binding on the ability of NP to support virus gene expression. The mutant NP genes were subcloned into plasmid pKT5 (13), which directs the expression of unfused NP under the control of the bacteriophage T7 RNA polymerase promoter. BHK cells were multiply infected with vaccinia viruses expressing the influenza virus and bacteriophage T7 RNA polymerase subunits (18, 46) and transfected with WT or mutant pKT5 plasmids and the synthetic flu-CAT RNA, and 20 h later the cells were lysed and the accumulation of CAT polypeptide was measured. Figure 7 shows the activities of the NP RNA-binding mutants relative to the WT polypeptide. All four mutants which were unable to bind RNA in vitro (Fig. 7a, R267, W330, F412, and R416) failed to support appreciable levels of virus gene expression. Of the three mutants which displayed reduced RNA-binding activity (RNA +), one also failed to produce significant levels of CAT polypeptide (Fig. 7a, W120) whereas the other two (W139 and W386) supported appreciable but diminished levels of gene expression relative to the WT protein. Western blot analysis of parallel transfections showed that the majority of the NP mutants (with the exception of W120) accumulated in similar quantities to WT NP (Fig. 7b), indicating that their diminished ability to support virus RNA synthesis was not simply the result of poor expression. Thus, the in vitro phenotypes of the mutants correlate with their in vivo function, and as expected, the RNA-binding activity of NP is essential for virus gene expression.

View larger version (16K): [in this window] [in a new window]   FIG. 7.   Ability of NP RNA-binding mutants to support virus gene expression. (a) BHK cells were infected with recombinant vaccinia viruses expressing the three subunits of the influenza virus RNA polymerase and T7 RNA polymerase and transfected with 1 µg each of a synthetic influenza virus segment containing an antisense CAT gene and plasmid encoding WT or mutant NPs, and the resulting accumulation of CAT polypeptide was quantified. Results are expressed as the percent activity ± standard error (n  3) for the mutants relative to WT NP. (b) Accumulation of WT and mutant NP in cells. Cells transfected with plasmids encoding WT or mutant NP molecules (or with empty plasmid vector, pKT0) were examined for NP accumulation by Western blot analysis with anti-NP serum.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Previously, two studies mapped the minimal sequences of NP capable of binding RNA to the N-terminal one-third of the polypeptide (Fig. 8) (1, 29). However, this truncated fragment of NP bound to RNA with lower affinity than did the WT protein, indicating the necessity for other sequences for full binding activity (1). In this study, we show by UV cross-linking and chemical fragmentation that at least two sequences within the C-terminal two-thirds of NP directly contact RNA (Fig. 1b). Furthermore, we found that single-amino-acid replacements throughout much of the length of NP were capable of disrupting the interaction of the polypeptide with RNA (summarized in Fig. 8). The widespread mutational sensitivity of RNA-binding activity is in contrast to the sequence requirements for the characterized protein-protein interactions mediated by NP, in which for any one ligand, disruptive point mutations are generally more localized in distribution (Fig. 8).

View larger version (14K): [in this window] [in a new window]   FIG. 8.   Summary of single-amino-acid replacements that perturb macromolecular interactions mediated by NP. Arrows give the locations of important residues (as labelled). (Top) Mutations affecting protein-protein interactions. The grey arrow indicates binding to cellular importin  (47); hatched arrows indicate self-association (16); solid arrows indicate actin binding (13). (Bottom) Mutations affecting RNA binding (this study and reference 35). The minimal region of NP capable of binding RNA (1, 29) is shaded.

The majority of the RNA-binding mutations do not affect other functions of NP, arguing against global malfolding of the polypeptide. None of the lesions described here decrease the ability of NP to interact with the polymerase complex (Fig. 6). With the exception of R416, which also decreases oligomerization of the polypeptide, none of the mutations affecting RNA binding block the ability of the polypeptides to form NP-NP contacts (16). In addition, the W120 and R267 mutations have little effect on the interactions of the proteins with filamentous actin (13). Thus, the majority of the mutations reduce RNA-binding activity specifically. However, some of them do so in a temperature-dependent fashion, affecting the function significantly at 37°C but much less so at 30°C. Thus, for these mutants at least, it seems unlikely that the sequence alterations have removed an amino acid side chain normally in direct contact with RNA. Yet, as discussed above, the mutations are clearly not disrupting the global folding of NP at the elevated temperature, since the polypeptides retain the other activities associated with NP. Furthermore, any effects of the mutations on the overall structure of NP were too subtle to be detected as changes in the resistance of the polypeptides to partial digestion with trypsin or chymotrypsin proteases (data not shown).

Overall, these data are not consistent with the hypothesis that RNA binding by NP is mediated through a discrete N-terminal region of the protein (1, 29). Instead, we propose that high-affinity binding of RNA by NP requires the concerted interaction of multiple regions of the protein with RNA. This hypothesis is consistent with the demonstration of direct protein-RNA contacts outside of the N terminus (Fig. 1b) and the identification of inhibitory mutations throughout the length of NP (Fig. 4). This model is also consistent with the proposed wrapping of RNA around NP in a nucleosome-like pattern (25, 49), as well as with the increased thermal stability of WT protein-RNA complexes and the ability of RNA to rescue the binding activity of a ts mutant at elevated temperatures (Fig. 5c). Within this framework, several possibilities exist to explain the ability of widely separated point mutations to ablate RNA binding without affecting other functions of NP. One possibility is that high-affinity binding results from the multiplicative contribution of independent binding sites, so that the loss of any one interaction causes a relatively large decrease in binding affinity resulting from what is effectively a loss of avidity. However, we have recently found evidence from circular dichroism spectroscopy that NP undergoes a conformational change on binding RNA (data not shown). Therefore, an alternative hypothesis is that multiple protein-RNA contacts are required to drive this conformational change necessary for high-affinity binding and that the loss of any one interaction interrupts the process.

Within the model of polyvalent binding, we propose that the individual protein-RNA interactions are mediated by a combination of electrostatic interactions between positively charged residues and the phosphate backbone and planar interactions between aromatic side chains and bases. This is consistent with the sensitivity of the protein to chemical modification of arginine residues (Fig. 3), the changes in the environment of tryptophan residues upon RNA binding (Fig. 2), and the requirement for specific arginine and aromatic residues (Fig. 4). In addition, this mode of single-stranded-nucleic-acid binding is well supported by mutational and structural studies of other DNA- and RNA-binding proteins (7, 10, 20, 28, 38, 40, 44). Two studies which probed influenza virus RNP complexes by measuring the protection afforded the RNA from chemical and enzymatic attack found a general protection of RNA phosphates but sensitivity of the bases (3, 28). This was interpreted as indicating that the protein binds to the phosphate backbone and not to the Watson-Crick positions of the bases. However, such findings do not rule out contacts between the protein and other positions of the bases. Indeed, NP binds preferentially to pyrimidine homopolymers, which suggests an interaction between the protein and bases (1). Also, considering that NP is not a sequence-specific RNA-binding protein, RNA modification experiments may well reflect an average state of protection for any one position, rather than a static interaction.

Recently, we have determined that the mutations defined as the ts lesions in the NPs of viruses A/WSN/33 ts56 (S314-N) (33) and A/FPV/Rostock/34 tsG81 (A332-T) (34) specifically induce ts RNA-binding by NP (35). Since these viruses are considered specifically defective for cRNA and vRNA transcription, this finding has implications for the function of NP during replicative transcription. The panel of NP mutants described here should therefore prove useful tools for improving our understanding of the functions of NP in influenza virus transcription.