Characterization of the Hantaan Nucleocapsid Protein-Ribonucleic Acid Interaction*

The nucleocapsid (N) protein functions in hantavirus replication through its interactions with the viral genomic and antigenomic RNAs. To address the biological functions of the N protein, it was critical to first define this binding interaction. The dissociation constant, K d , for the interaction of the Hantaan virus (HTNV) N protein and its genomic S segment (vRNA) was measured under several solution conditions. Overall, increasing the NaCl and Mg2+ in these binding reactions had little impact on the K d . However, the HTNV N protein showed an enhanced specificity for HTNV vRNA as compared with the S segment open reading frame RNA or a nonviral RNA with increasing ionic strength and the presence of Mg2+. In contrast, the assembly of Sin Nombre virus N protein-HTNV vRNA complexes was inhibited by the presence of Mg2+ or an increase in the ionic strength. TheKd values for HTNV and Sin Nombre virus N proteins were nearly identical for the S segment open reading frame RNA, showing weak affinity over several binding reaction conditions. Our data suggest a model in which specific recognition of the HTNV vRNA by the HTNV N protein resides in the noncoding regions of the HTNV vRNA.

Hantaviruses are tripartite negative-sense RNA viruses in the Bunyaviridae family (1). The viral RNA segments, L, M, and S, encode an RNA dependent RNA polymerase, two envelope glycoproteins, and nucleocapsid (N) 1 protein, respectively. In mature virions, multiple copies of the N protein coat each of the genomic segments (vRNAs) to form three distinct ribonucleoprotein complexes, which may also include the viral polymerase. The N protein has been suggested to play a functional role in replication and/or transcription, perhaps by providing a mechanism for controlling the amount or rate of macromolecular synthesis. Precedent exists for such activities with analogous proteins of other negative-strand RNA viruses such as those in the Rhabdoviridae, Paramyxoviridae, and Orthomyxoviridae families. The N protein of vesicular stomatitis virus (VSV) modulates the transition from transcription to replica-tion (2), while the influenza nucleoprotein is required for RNA synthesis (3)(4)(5), and antitermination during viral mRNA synthesis (6). In contrast to influenza nucleoprotein (7,8), selective encapsidation of viral genomic or leader RNAs has been reported for the N proteins of VSV (9,10) and rabies virus (11). For hantaviruses, however, little is known regarding the mechanism for encapsidating viral RNA or the role that N protein plays in regulating viral transcription and replication.
The proposed activities of the hantavirus N protein in the virus life cycle rely on differential interactions with the three types of viral RNAs, the vRNA, the virus complementary (cRNA) and the messenger RNA (mRNA). The mechanism for selective encapsidation of the vRNA and cRNA by the N protein remains a central question in the hantavirus replication cycle as well as those of other members of the family Bunyaviridae. Cell culture-based experiments with Bunyamwera virus suggested that encapsidation requires full-length vRNA or cRNA because mRNA was not found in nucleocapsids (12). The mRNA is quite distinct in structure and sequence from the vRNA; with truncated 3Ј termini and extended 5Ј termini observed on all members of the Bunyaviridae. The 3Ј truncation prevents base pairing of the complementary sequences at the 3Ј and 5Ј termini of the vRNA and cRNA and results in the inability of the mRNA to form panhandle structures. Thus, while vRNA and cRNA can form noncovalently, closed circular structures, mRNA cannot (13). This observation led to the suggestion that this gene region may play a role in assembly of the nucleocapsids, however, no sequences or secondary structures in any of the viral RNAs have been defined (14). Further, no readily identifiable RNA binding motifs have been located in N proteins, although deletion mapping of Hantaan (HTNV) and Puumala virus (PUUV) N proteins identified a nonspecific RNA binding domain in the carboxyl-terminal 93 amino acids (15). This region showed no apparent specificity for viral RNA.
The N protein clearly functions in the viral replication cycle through its interactions with the viral nucleic acids and with other structural proteins. The biochemical and molecular determinants that promote interactions between the N protein and the viral RNAs have not been defined. To begin to address the biochemical basis of N protein and viral RNA interactions, we investigated the first event in the assembly process, the binding step. We report procedures for obtaining highly purified hantavirus N protein and for measuring its interactions with RNA. Using these methods, we defined the dissociation constant for HTNV N protein-RNA interactions under several solution conditions. In addition, we explored the ability of a heterologous N protein from Sin Nombre virus (SNV) to interact with HTNV-derived RNAs and a nonviral RNA.

EXPERIMENTAL PROCEDURES
Hantavirus N Protein Purification-The S segment open reading frame (S ORF) representing the HTNV N protein was cloned by polym-erase chain reaction amplification with 5Ј and 3Ј primers engineered with flanking NdeI and XhoI sites. The amplified HTNV S ORF was cloned into the NdeI and XhoI sites of pET23b (Novagen) to create the expression plasmid, pHTNV-N, and was sequenced with an ABI automated sequencer. Cloning into the XhoI site created a C-terminal fusion of a hexahistidine tag (predicted mass ϭ 49.06 kDa). The hexahistidine tag has been shown previously by others to not effect N protein RNA binding reactions (15). However, experiments were also performed using the HTNV N protein expressed from the pFLAG vector (Sigma) (16). The Hantaan virus S segment ORF was cloned into the EcoRI site of pFLAG1 to create an N-terminal fusion of the omp signal peptide followed by the FLAG epitope. Enterokinase can cleave the FLAG peptide from the HTNV N protein and will leave an amino-terminal Leu, (predicted mass ϭ 48.2 kDa). The plasmid expression vector, pSNV-N pET-1, (gift of Brian Hjelle, University of New Mexico) contains the entire ORF representing the SNV N protein (isolate 3H226). The ORF is cloned into the HindIII and XhoI sites of the pET23b expression vector, with an N-terminal fusion of the T7 leader and a C-terminal fusion of a hexahistidine tag; thus, the SNV fusion protein had a predicted size of 51.4 kDa.
The expression, purification and enterokinase cleavage of soluble HTNV N-FLAG were performed as described previously (16). The enterokinase-cleaved HTNV N protein fractions were stored at Ϫ80°C in 20 mM Tris, pH 8.0, 150 mM NaCl, and 0.1 M glycine. Prior to analysis in RNA filter binding reactions, the cleaved protein was concentrated, and washed and resuspended in binding buffer (40 mM HEPES, pH 7.4, 80 mM NaCl, 20 mM KCl, and 1.5 mM DTT). HTNV and SNV N proteins, expressed in BL21DE3 from the pET23b vector, were purified using methods previously described (17). BL21DE3 harboring pET23b was included as a negative control during expression and purification. Column fractions of HTNV N protein, SNV N protein and pET23b purifications were adjusted to an A 595 of 0.2 and refolded by dialysis through a stepwise dilution of urea from 3 to 0 M urea as described previously (17). Final dialysis of the protein fractions was conducted in Buffer C (200 mM NaCl, 40 mM Hepes, pH 7.4, 0.1 mM EDTA, 1 mM DTT, 0.4% Nonidet P-40, and 10% glycerol). The final buffers were rendered RNase-free by pretreatment of the components with diethylpyrocarbonate before autoclaving, or by preparing the solutions from diethylpyrocarbonate-treated water and RNase-free molecular biology grade reagents (Sigma). The refolded protein fractions were flash frozen on dry ice and stored at Ϫ80°C. Column fractions were followed by 12% SDS-PAGE and Western blot analysis. Protein concentrations were measured using the Bradford method (18) with Bio-Rad Micro-Assay reagents as recommended. The primary antibody for detecting the SNV N protein was inactivated human serum from a patient diagnosed with acute hantavirus pulmonary syndrome (Chiron). The secondary antibody for detecting the SNV N protein was affinity-purified biotinylated goat anti-human IgG (HϩL) (Vector Laboratories). The primary antibody for detecting HTNV N protein was hyperimmune mouse ascitic fluid to authentic HTNV 76 -118 (19). The secondary antibody for detecting HTNV N protein was affinity-purified biotinylated goat antimouse IgG (HϩL) (Vector Laboratories).
Preparation of RNA Substrates-HTNV vRNA was transcribed from the vector pGEM1-HTNVS (19). The plasmid contains the genomic HTNV S segment cloned into the PstI site of the pGEM1 plasmid (Promega). An S ORF RNA was generated by transcription from the HTNV S cDNA (pHTNV-N), and a nonviral 67-nucleotide control RNA was transcribed from pGEM7Zf ϩ . In preparation for the in vitro transcription reactions, pGEM1-HTNV S was digested with XbaI, pHTNV-N was digested with XhoI, and pGEM7Zfϩ was digested with SmaI. [ 32 P]UMP radiolabeled transcripts were produced from linearized plasmids using the MaxiScript SP6/T7 RNA Transcription Kit (Ambion). Transcription of the pGEM1-HTNV S template creates an additional 20 and 12 nucleotides derived from pGEM1 on the 5Ј and 3Ј ends, respectively. Purification of the RNA transcripts was performed using the RNeasy Kit (Qiagen). Purified RNA was stored at Ϫ20°C in 25-l aliquots for up to 2 weeks. Typically, RNA transcripts had specific activities of 1 ϫ 10 8 cpm/g.
UV Cross-linking Assay-The following RNase-free components were combined on ice in a final volume of 20 l: 40 mM HEPES, pH 7.4, N protein, and 1 l of [ 32 P]UTP-labeled RNA (1 ng) along with variable concentrations of MgCl 2 , NaCl, and KCl as noted in figure legends. In standard reactions, 29 ng/l SNV N protein or 175 ng/l HTNV N protein were added to reaction buffer and incubated for 10 min at 37°C. One g of heparin was then added, and reactions were incubated for 15 min at 37°C. RNA-protein complexes were covalently cross-linked by exposing binding reactions to 1.8 kJ of UV light in a UV cross-linker (UVC500, Hoeffer). Unbound RNA was digested by adding 50 units of RNase T1 (Ambion) to the SNV-RNA or 1 unit of RNase V1 (Amersham Pharmacia Biotech) to HTNV N-RNA binding reactions, and incubating for 30 min at 37°C. Reactions were stopped with 2ϫ loading buffer (0.125 M Tris-HCl, pH 6.8, 0.4% sodium dodecyl sulfate (w/v), 0.28 M ␤-mercaptoethanol, 20% glycerol (v/v), and 0.01 mg/ml bromphenol blue), heated to 95°C, and separated by 12% SDS-PAGE. Gels were dried, and the results visualized by exposing to autoradiographic film or by using a Molecular Dynamics PhosphorImager. Signals were quantitated using ImageQuaNT TM version 4.2 software (Molecular Dynamics).
Filter Binding Assay-HTNV RNAs were prepared by in vitro transcription in the presence of [␣-32 P]UTP as described above. HTNV N protein was serially diluted in binding buffer (40 mM HEPES, pH 7.4, 40 mM NaCl, 20 mM KCl, and 1.5 mM DTT) to give a final concentration range of 3.5 ϫ 10 Ϫ9 to 3.5 ϫ 10 Ϫ6 M. SNV N protein was serially diluted in binding buffer to give a final concentration range of 5.6 ϫ 10 Ϫ10 to 5.6 ϫ 10 Ϫ7 M. Additional NaCl in the protein storage buffer brought the final NaCl concentration to 80 mM. One ng of [␣-32 P]UTP-labeled RNA was added, and the reactions were incubated for 10 min at 37°C. The reactions were slot-blotted (Bio-Rad) onto nitrocellulose filters as described by the manufacturer, and each well was washed with 200 l of binding buffer. Limitations of filter binding assays include nonspecific retention of RNA on the membrane. Nonspecific retention of RNA was measured by filtering complete reaction mixtures in the absence of protein. Signals were quantitated by scintillation counting or by using a Molecular Dynamics PhosphorImager. Dissociation constants (K d ) were averaged from a minimum of four independent experiments, and were calculated by fitting a nonlinear binding curve to the empirical data using the Origin program (MicroCal). The apparent K d corresponds to the concentration of N protein required to obtain half-saturation, assuming the complex formation obeys a simple bimolecular equilibrium. Interpretation of the data as a simple binding equilibrium between RNA and protein requires that the complex does not dissociate during the filtration process (20). Another limitation of the filter binding assay is the possible dissociation of the complex during filtration. The stability of the complex was experimentally confirmed (data not shown). A second inherent assumption in the assay is that the fractional retention of the different RNA-protein complexes is not due to the incomplete retention of the protein. In our assays, the fractional retention of the complexes ranged from 38% to 94%, depending on the RNA substrate. Therefore, we assumed the plateau in the percentage of binding of the RNA represented complete binding of the RNA substrate.

RESULTS
Purification of Hantavirus N Proteins-BL21(DE3) cells harboring the expression vector, pHTNV-N or pSNV-N, were induced with isopropyl-1-thio-␤-D-galactopyranoside to express hantaviral N proteins. In addition, a mock purification with bacteria transformed with the parent vector, pET23B, was included as a negative control. Total protein from each of the induced cell cultures were similarly extracted under denaturing conditions, and solubilized protein was separated using nickel-affinity chromatography. Proteins bound to the nickelaffinity resin were eluted by a stepwise decrease in the pH of the running buffer from pH 6.3 to 5.9 and, finally, pH 4.5. High levels of SNV N protein expression were obtained and readily extracted as shown in the SDS-PAGE of the pelleted fraction recovered after detergent lysis and centrifugation (Fig. 1, lane 1) and load (Fig. 1, lane 2). SNV N was recovered from the column in the pH 6.3 elution (Fig. 1, lane 5), the first seven fractions of the pH 5.9 step gradient (Fig. 1, lanes 6 -8 and  10 -12), and the first six fractions of the pH 4.5 step gradient (Fig. 1, lanes 13-18). Similar elution profiles were observed during the purification of the HTNV N protein. In the mock purification, no protein bands were observed at the expected molecular weight for hantaviral N proteins. UV cross-linking experiments showed the pH 4.5 fractions gave the greatest level of RNA binding activity (21); thus, these were used in the various assays described herein.
Encapsidation of HTNV vRNA with HTNV and SNV N Proteins-The interaction between the purified HTNV N protein and HTNV vRNA was first investigated using an assay in which the full-length S segment vRNA was cross-linked to N protein by exposure to ultraviolet light (UV cross-linking as-say). After removing unbound RNA by RNase digestion, the reaction was examined by SDS-PAGE. The optimal conditions for forming complexes were determined earlier by extensive titrations of ionic strength, metal, and reducing agents in these UV cross-linking assays with the N protein (21). Titrations experiments revealed an increase in the concentration of ribonucleoprotein complexes with increasing amounts of HTNV N protein and a constant level of HTNV vRNA ( Fig. 2A). Complexes were not detectable at 0.73 ng/l ( Fig. 2A, lane 2); however, efficient cross-linking was observed at 7.3 ng/l ( Fig.  2A, lane 3) and 14.5 ng/l ( Fig. 2A, lane 4). At 29 ng/l, the reaction became saturated, as determined by quantitation from the PhosphorImager analysis ( Fig. 2A, lane 5). We estimate that the concentration of the N protein required to saturate the reaction is greater than a 4000-fold excess of what would be necessary to encapsidate the vRNA with 1 protein per 10 nucleotides. These results are similar to encapsidation experiments previously published for VSV nucleocapsid interactions (10).
Using the same approach and binding conditions, a titration of purified SNV N protein to a constant amount of HTNV vRNA was examined (Fig. 2B). The highest level of signal was observed when the concentration of SNV N was 58 ng/l (Fig. 2B, lane 5). In addition, efficient cross-linking was observed at the concentrations of 7.3 ng/l (Fig. 2B, lane 2), 14.5 ng/l (Fig. 2B,  lane 3), and 29 ng/l (Fig. 2B, lane 4). The protein concentrations that provided saturation levels for HTNV and SNV N proteins were used in all subsequent UV cross-linking experiments. Finally, the UV cross-linking studies revealed that the structure of the ribonucleoprotein complexes promoted by HTNV and SNV N proteins were clearly distinct on the HTNV vRNA template. HTNV complexes were sensitive to digestion with the double-stranded nuclease, RNase V1, but not with the single-stranded nuclease RNase T1. 2 In contrast, SNV complexes were only sensitive to RNase T1 digestion.
Effect of Ionic Strength on the Dissociation Constant of HTNV Ribonucleoprotein Complexes-The UV cross-linking experiments confirmed the absence of other RNA-binding proteins in N protein preparations, and defined the basic conditions suitable for further studies of N protein-RNA interactions. However, the UV cross-linking assay is limited in that the accessibility of the nucleocapsid to RNase digestion as well as the activity of the RNase may vary under different ionic strength conditions. As exemplified by the nucleocapsid proteins of Sendai virus and VSV, increases in the ionic strength of the microenvironment may increase the tightness of nucleocapsid coiling, which can increase the resistance of the nucleocapsid to digestion with RNase (22). Therefore, using the reaction conditions defined in the UV cross-linking experiments, we developed and used a filter binding assay to explore the dissociation constant (K d ) for formation of N protein-RNA complexes under several ionic strength conditions (Fig. 3A). For comparison, we examined the effect of ionic strength on nucleocapsid formation using the UV cross-linking assay (Fig. 3B).
Filter binding experiments were performed with increasing concentrations of HTNV N protein and a constant amount of HTNV vRNA (Fig. 3A). NaCl concentrations were titrated from 0.05 to 1.0 M. From the resulting Klotz plot, a K d of 25 nM was calculated for the 50 mM NaCl binding isotherm. At 100 mM NaCl, the smallest dissociation constant, 14 nM, was observed. The K d values observed at 0.25, 0.50, and 1.0 M NaCl were 32, 42, and 47 nM, respectively (Table I). These experiments were repeated with a 3-fold higher concentration of vRNA (the upper limit allowed in our filter binding assay) and one-half the amount of N protein (Table I). The results were similar to those obtained with the standard assay conditions. In both experiments, the K d for the HTNV N protein-vRNA interactions did not vary substantially as the ionic strength was increased in the binding reaction buffer. This suggests nonelectrostatic con-tributions to formation and/or stability of the ribonucleoprotein complex.
Using conditions identical to those in the filter binding assays, NaCl concentrations were titrated and evaluated for their effect on complex formation of the HTNV N protein and HTNV vRNA by UV cross-linking. PhosphorImager analysis of the resulting N protein-RNA complex revealed the highest level of RNA binding at 100 mM NaCl (Fig. 3B, lane 3). The level of binding decreased minimally, 1.2-fold, as the ionic strength was decreased to 50 mM. As the NaCl concentration was increased from 0.25 to 0.50 M NaCl the level of N protein-RNA complex fell 1.1-fold (Fig. 3B, lane 4) and 1.5-fold (Fig. 3B, lane 5), respectively. RNase V1 displays optimal activity in reactions containing up to 0.5 M NaCl; therefore, as expected, no complex was observed at the 1.0 M NaCl concentrations (Fig.  3B, lane 6). In summary, both UV and filter binding assays showed a similar level of binding of the vRNA by the N protein over a wide range of ionic strength, 50 -500 mM. While the UV cross-linking assay was limited to these lower NaCl concentrations, the filter binding assay clearly revealed a similar stability of the N protein-RNA complex at 100 mM and 1 M NaCl concentrations. Thus, the results suggest that electrostatic interactions are not a predominant factor in complex formation. In conjunction with the differential susceptibility of the HTNV N-vRNA complex to RNase V1 and RNase T1, these results suggest the major contributions to complex stability may be through short range van der Waals or dielectric interactions between the N protein and the phosphodiester backbone or nucleobases within single-stranded regions of the vRNA.
Effect of Magnesium on the Formation of HTNV N Protein-vRNA Complexes-In addition to monovalent ions, divalent ions can strongly influence the structure and stability of RNA (23,24). Alternatively, metal ions such as Mg 2ϩ can compete with a RNA-binding protein for electrostatic-based interactions along the phosphodiester backbone of an RNA molecule. The ability of the metal ion to compete with the protein essentially reflects nonspecific protein-RNA interaction, as recently demonstrated in the interactions of the HIV-1 nucleocapsid and the tRNA 3 Lys (25). To explore the requirement for a metal ion in the in vitro binding reaction of the HTNV N-protein and vRNA, we examined the effects of MgCl 2 on the RNA binding activity of the HTNV N proteins (Fig. 4, A and B).
In experiments in which binding reactions contained a constant level of HTNV vRNA and a range of HTNV N protein, as the concentration of MgCl 2 was increased, we noted a small shift in the binding isotherms toward an increase in K d (Fig.  4A). The K d for each HTNV N protein-RNA complex was calculated from the isotherms presented in Fig. 4A, and plotted against the MgCl 2 concentration at which it was determined (Fig. 4B). The K d values determined at concentrations of MgCl 2 ranging from 1 to 8 mM were similar. The K d observed in the absence of MgCl 2 , 53 nM, was only slightly smaller. The highest K d was observed when 10 mM MgCl 2 was included in the reactions, and represented a 2.4-fold increase in the K d as compared with the K d at 8 mM MgCl 2 . Overall, these studies do FIG. 3. Ionic strength dependence of HTNV N protein-vRNA complex formation. The effect of increasing ionic strength on the vRNA binding activity of HTNV N protein was measured using filter binding (A) and UV cross-linking assays (B). In panel A, binding reactions were assembled in 5 mM MgCl 2 in addition to standard reaction components as described under "Experimental Procedures." 32 P-Labeled RNA was incubated with the indicated molar concentrations of N protein (x axis). The amount of radioactively labeled N protein retained on the filter was calculated relative to the maximum radioactivity retained in each experiment (y axis). In panel B, binding reactions were assembled in 5 mM MgCl 2 , subjected to UV cross-linking, separated by 12% SDS-PAGE, and subjected to autoradiography and PhosphorImager analysis as described under "Experimental Procedures." Lane numbers are indicated below each panel. The total salt concentrations are indicated above each lane. Lane 1 is a no protein control reaction. not support a major role for MgCl 2 in promoting the interaction of the HTNV N protein and HTNV vRNA.

Specificity of the N Protein for Viral and Nonviral
RNAs-A central question in the hantavirus replication cycle is how the N protein selectively encapsidates vRNA and cRNA, but not mRNA. To address this question, we explored the K d for HTNV and SNV N proteins complexed with two different viral RNAs, the full-length HTNV vRNA and an RNA representing the open reading frame of the HTNV S segment (S ORF RNA). The S ORF RNA is derived from the cRNA and does not contain the terminal complementary regions, which are expected to form the panhandle structure characteristic of vRNA and cRNA. The K d for the formation of a complex between the HTNV N protein and a nonspecific 67-nucleotide RNA (control RNA) was also examined. As described above, the addition of 1-8 mM MgCl 2 in the binding reaction did not reduce the affinity of the HTNV N protein for HTNV vRNA. However, in the following, we decided to examine the binding of the three RNAs in the presence and absence of MgCl 2 to explore its effect on alternative RNA substrates.
Binding isotherms for the HTNV N protein and the three RNA substrates were defined for four different reaction conditions: 1) 40 mM NaCl (Fig. 5A), 2) 40 mM NaCl and 1 mM MgCl 2 (Fig. 5B), 3) 80 mM NaCl and 20 mM KCl (Fig. 5C), and 4) 80 mM NaCl, 20 mM KCl, and 1 mM MgCl 2 (Fig. 5D). The dissociation constants for each binding isotherm are summarized in Table  II. At low ionic strength, we observed little difference in the K d for HTNV N protein and the three RNA substrates, although the K d for the control RNA substrate was 3-fold higher (Fig. 5A, Table II). The addition of 1 mM MgCl 2 to the binding reaction with 40 mM NaCl enhanced the affinity of the HTNV N protein for the vRNA as compared with the S ORF and control RNAs (Fig. 5B, Table II). The greatest difference in binding of the templates by the HTNV N protein was observed in experiments that increased the ionic strength of the binding reaction to 100 mM (4:1 NaCl:KCl). In these assays, we noted an approximate 5-fold increase in the preference of the HTNV N protein for its vRNA as compared with the S ORF and control RNAs (Fig. 5D, Table II).
We used the same approach to examine the effects of ionic strength on the stability of the heterologous SNV N protein and HTNV RNAs. In contrast to the HTNV N protein-vRNA interactions, increasing the ionic strength of the binding buffer increased the dissociation of the SNV N protein-vRNA interaction approximately 3-fold (Table III). A comparison of the K d for binding reactions containing 80:20 mM NaCl/KCl with no MgCl 2 and those with 1 mM MgCl 2 revealed an approximate 10-fold increase in the K d in the presence of the metal. The interaction of both the HTNV and SNV N proteins with the S ORF reflected a similar dissociation constant (compare Tables  II and III).
Finally, we wanted to determine whether the histidine tag affected binding to RNA and the observed dissociation constants. The HTNV N protein was expressed in Escherichia coli, and purified from the soluble fraction as a FLAG fusion protein (16). After purification, the FLAG peptide was removed by enterokinase cleavage, and binding isotherms for the three RNA substrates were determined. Filter binding assays were performed with the reaction conditions noted as optimal for binding in Table II (80 mM NaCl, 20 mM KCl). The K d values, calculated from the resulting binding isotherms, were 32 Ϯ 14 for the vRNA, 107 Ϯ 26 for the ORF RNA, and 250 Ϯ 38 for the control RNA. These results were similar to those defined for the HTNV N protein with the hexahistidine tag (Table II). However, the K d obtained for the control RNA with the enterokinase-cleaved HTNV N protein was approximately 2-fold lower than the K d obtained for the hexahistidine tag HTNV N protein. This suggests that the hexahistidine tag may provide a slight amount of stability toward the binding of nonspecific nonviral RNA. DISCUSSION The N proteins of various RNA viruses have distinct biological roles in the viral replication cycle such as encapsidation, interactions with membrane proteins for virus assembly, as well as regulation of viral RNA synthesis. In addition to encapsidation of the viral RNAs, the hantavirus N protein may also FIG. 4. Effect of magnesium concentration on HTNV N protein-vRNA interactions. A, the binding isotherms for several Mg 2ϩ concentrations were measured for the HTNV N protein-vRNA interactions. Reactions were assembled for HTNV with 80 mM NaCl, 20 mM KCl in addition to standard reaction components as described under "Experimental Procedures." The Mg 2ϩ concentrations examined are indicated above each lane. In panel A, 32 P-labeled RNA was incubated with the indicated molar concentrations of N protein (x axis). The amount of radioactively radiolabeled N protein retained on the filter was calculated relative to the maximum radioactivity retained in each experiment (y axis). In panel B, the dissociation constants calculated in panel A were plotted against the concentration of Mg 2ϩ used in the reaction set.
play an important role in the regulation of replication and transcription (1). Little is known, however, regarding the structure of the RNA binding domain and its intrinsic biochemical properties and requirements for RNA interaction. To initiate detailed biochemical studies of this protein and its interactions, we developed methods to purify the hantavirus N protein. Our purification scheme resulted in greater than 95% homogeneity of the full-length HTNV and SNV N proteins. The protocols developed to purify the N protein are substantially different from those reported in an earlier brief report (15). A major  difference in the two protocols is the use of NaCl throughout our purification scheme. During development of the optimal strategy for purification of HTNV and SNV N proteins, we found that addition of NaCl to the extraction buffer greatly increased the solubility of the N protein (21). In addition, our protocol allows for the complete refolding of the N by eliminating urea with extensive dialysis, and the inclusion of EDTA and the reducing agent, DTT, in the refolding of the protein. Thus, our strategy should result in a protein with a more native conformation than was reported earlier. In support of this, the same strategy has been used with great success to obtain highly active retrovirus enzymes such as the HIV-1 RNase H (26), as well as the murine leukemia virus (27) and human T-cell leukemia virus (17) integrases.
No comprehensive studies have addressed N protein binding in the family Bunyaviridae. Gel electrophoretic mobility shift assays (GEMSA) were used to characterize the RNA binding activity of N proteins of the Tospovirus tomato spotted wilt virus, and the hantavirus PUUV N proteins (15,28). The RNA binding activity of the HTNV N protein was also examined following refolding of the protein on nitrocellulose filters (15,28). The effect of ionic strength, metal ion, and redox conditions on the binding of RNA has not been previously reported for any hantavirus N protein. Therefore, in the development of these assays, we were interested in obtaining quantitative tools to measure conditions affecting RNA binding and to compare the binding of various RNAs under a variety of solution conditions. Herein, we have presented two assays, UV cross-linking and filter binding, with which to study hantavirus N protein-RNA interactions. A wide range of protein concentrations were noted to be effective for RNA binding in both assay systems and were found to be similar to the concentrations used in PUUV N protein GEMSA (15). The effect of metal ions on the interaction of the HTNV N protein and its vRNA was not notable until large concentrations of Mg 2ϩ were added to the binding buffer. However, when we compared the binding affinity of the HTNV N protein with alternative substrates, a clear preference was observed for the vRNA in the presence of the metal ion. Metal ions can play a role in RNA-protein interactions by enabling specific RNA folding and stability (24,29). We suggest that addition of Mg 2ϩ may have increased secondary structures in the vRNA template and thereby reduced the nonspecific interactions based on structural affinities. Alternatively, Mg 2ϩ may have competed with the protein for nonspecific van der Waals interactions with the phosphodiester backbone (25). The metal ion may act to mimic the presence of a protein factor (e.g. hantavirus L protein), and thereby enhance the binding specificity of the N protein with vRNA and cRNA as opposed to mRNA. Precedent for this type of mechanism has been shown with the VSV and rabies virus P proteins, which increase the specificity of N protein encapsidation for their genomic or leader viral RNAs (11,30). Examination of the effect of ionic strength on the binding reactions by filter binding and UV cross-linking assays showed that HTNV ribonucleoprotein complexes were stable to a wide range of ionic strength. This suggests that close range nonelectrostatic forces such as van der Waals and not electrostatic interactions play a predominant role in N protein-vRNA interactions. The lack of salt dependence in complex formation has been reported for other RNA-binding proteins such as the R17 coat protein (31), and the S15 (32) and L11 ribosomal proteins (33). Clarification of the mechanisms underlying these finding awaits a more thorough kinetic and thermodynamic analysis of the N protein and the vRNA.
In addition to defining the parameters for encapsidation of the HTNV vRNA by HTNV N protein, we were interested in determining whether the HTNV vRNA could be encapsidated by a heterologous N protein from a distant member of the Hantavirus genus, SNV. HTNV and SNV N proteins share 63% identity, and represent two distinct phylogenetic classes of hantaviruses (34). The N complexes formed by HTNV and SNV were distinct as revealed by their differential susceptibility to RNase V1 and RNase T1. The comparison of the binding isotherms for the HTNV or SNV N proteins and three RNA substrates revealed a preference for the HTNV N protein for its vRNA as compared with the other RNA substrates tested. Our results suggest that specific interactions, albeit with moderate affinity, exist between the HTNV N protein and the terminal regions of the vRNA and cRNA. These results are similar to those reported for the rabies virus N protein (11), in which a 3-fold greater binding was shown for the viral leader RNA than the N mRNA. In their analysis, Yang et al. also examined the binding of the rabies N protein to a nonviral control RNA. As observed for the hantavirus N protein, the rabies N protein had a greater affinity for a viral than a nonviral RNA. The studies reported herein are also in agreement with previous observations with La Crosse virus in which the signal for encapsidation was suggested to be in the terminal complementary nucleotides (13) and in a separate study, which showed the mRNA is not a preferred substrate for encapsidation (14). However, further analysis is required to define where within the untranslated region the specificity occurs. Previously, it was reported that the binding of the PUUV N protein to a vRNA was nonspecific as noted in a tRNA competition assay, and when compared with binding to a PUUV RNA, which contained only the PUUV ORF (15). We offer a different interpretation of the PUUV N protein GEMSA results. First, the authors observed an increase in the retardation of genomic S-RNA with the addition of increasing amounts of protein. We interpret this finding to suggest the presence of multiple binding sites and cooperativity (35). In addition, in the previous study, the authors indicated that GEMSA showed reduced binding of PUUV N protein and vRNA in the presence of tRNA. However, without quantitation, we do not find this interpretation compelling. Further, the presence of the tRNA appeared to have little impact on the formation of the higher ordered complexes as the concentration of N protein was increased. The second line of reasoning for their conclusions for an overall lack of specificity of the N protein for its vRNA template came from comparing its binding with the PUUV ORF RNA, for which they reported little change in the GEMSA pattern when compared with the vRNA or tRNA competition. If one examines the retardation of all the complexes, however, one can clearly observe a reduction in the mobility of the PUUV N protein-PUUV S ORF RNA complexes as compared with the others. Overall, we suggest that the conditions for vRNA specificity by the PUUV N protein were impossible to accurately ascertain with the GEMSA analysis performed. Finally, in the earlier study, the authors seemed to discount the importance of the GEMSA using a construct representing the flanking regions of the PUUV ORF and the PUUV N protein to show substrate preference. Interestingly, at all concentrations of PUUV N protein examined, a large aggregate is observed (well shift) in the case of the PS2-/PS2ϩ RNA, a partially double-stranded RNA. We offer that these experiments suggest a region of the vRNA with a higher affinity for the N protein. Further, we suggest that these interpretations taken in conjunction with our data support a model in which binding of the N protein to its vRNA proceeds through cooperative interactions among the N proteins that initiate in the terminal regions of the RNA, possibly at the 5Ј end.
These studies suggest the HTNV N protein relies primarily on nonelectrostatic forces to recognize structural and/or se-quence determinants in its vRNA and cRNA. Further, the reduced affinity of the SNV N protein for the HTNV vRNA suggests unique structures and/or sequences in the RNA and/or protein have evolved for these two viruses. Future work will address the site of encapsidation of the vRNA and the thermodynamics and kinetics underlying the interaction between the N protein and viral RNAs. The solution conditions we have defined herein will also be useful for mapping the RNA binding domain that renders specificity to the vRNA.