NMR Structure of the N-terminal Coiled Coil Domain of the Andes Hantavirus Nucleocapsid Protein*

The hantaviruses are emerging infectious viruses that in humans can cause a cardiopulmonary syndrome or a hemorrhagic fever with renal syndrome. The nucleocapsid (N) is the most abundant viral protein, and during viral assembly, the N protein forms trimers and packages the viral RNA genome. Here, we report the NMR structure of the N-terminal domain (residues 1–74, called N1–74) of the Andes hantavirus N protein. N1–74 forms two long helices (α1 and α2) that intertwine into a coiled coil domain. The conserved hydrophobic residues at the helix α1-α2 interface stabilize the coiled coil; however, there are many conserved surface residues whose function is not known. Site-directed mutagenesis, CD spectroscopy, and immunocytochemistry reveal that a point mutation in the conserved basic surface formed by Arg22 or Lys26 lead to antibody recognition based on the subcellular localization of the N protein. Thus, Arg22 and Lys26 are likely involved in a conformational change or molecular recognition when the N protein is trafficked from the cytoplasm to the Golgi, the site of viral assembly and maturation.

Hantaviruses can cause two emerging infectious diseases known as the hantavirus cardiopulmonary syndrome (HCPS) 3 and the hantavirus hemorrhagic fever with renal syndrome (1). Annually, there are over 150,000 cases of hantaviral infections reported world wide (2). Rodents are the primary reservoir of hantaviruses, and humans are normally infected by inhalation of aerosol contaminated with the excreta of infected rodents. The first reported cases of HCPS in North America (3) was caused by a novel hantaviral species (4,5), the Sin Nombre virus, and had an initial mortality rate of 78%. HCPS has since been reported throughout the United States with a current mortality rate of 35% when correctly diagnosed (6). The major cause of HCPS in South America is the Andes virus, and person-to-person transmission of the Andes virus was reported in Argentina and Chile (7). Hantaviruses are known to invade and replicate primarily in endothelial cells, including the endothelium of vascular tissues lining the heart (8 -10).
The genome of hantaviruses consists of three negativestranded RNAs, which encode the nucleocapsid (N) protein, two integral membrane glycoproteins (G1 and G2), and an RNAdependent RNA polymerase (L protein). The N protein is highly immunogenic (11,12) and elicits a strong immune response, which confers protection in mice (13)(14)(15). It is highly conserved and is the most abundant viral protein, and it plays important roles in viral encapsidation, RNA packaging, and host-pathogen interaction (16). The N protein binds to viral proteins (16), host proteins (17)(18)(19)(20)(21)(22)(23), and viral RNA (24 -28). The self-association of the N protein into trimers was shown by gradient fractionation and chemical cross-linking (29). Deletion mapping identified that regions at the N and C termini are important in N-N interaction (29 -31), and a model of trimerization was proposed based on the head-to-head and tail-to-tail association of the N-terminal and C-terminal domains, respectively (30,32).
The N-terminal region in the Sin Nombre virus N protein (residues 3-73) (33) and the Tula virus (residues 1-77) (34) were predicted to form coiled coil domains. Recently, the structure of the N-terminal coiled coil domain (residues 1-75 and 1-93) of the Sin Nombre virus was determined by crystallography (35). The highly conserved hydrophobic residues stabilize the structure of the coiled coil; however, there are highly conserved polar residues that appear to have no function in stabilizing the coiled coil domain. Here, we report the solution structure of the N-terminal 1-74 residues of the Andes virus N protein, which also forms a coiled coil domain. Further, we identified that the coiled coil contains distinct regions of positively and negatively charged surfaces involving conserved polar residues. We hypothesize that these regions are also important in N protein function. We used site-directed mutagenesis to alter the surface of the N protein and assayed for the subcellular localization of the N protein by immunocytochemistry. We used CD spectroscopy to confirm that mutations did not alter the coiled coil structure of the N 1-74 (residues 1-74 of the N protein) domain. However, immunocytochemistry showed that despite the N protein being present throughout the cytoplasm, a monoclonal antibody only recognized the Arg 22 and Lys 26 mutants when nucleocapsids are associated with the Golgi, the site of viral assembly and maturation. We propose that the conserved surface residues Arg 22 and Lys 26 are important in the proper conformation or molecular recognition of the N protein.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification of N 1-74 -The N 1-74 domain of the Andes virus (strain 23) nucleocapsid protein was subcloned into pET151 (Invitrogen), which appends a 33-residue His 6 tag and a TEV (tobacco etch virus) protease cleavage site at the N terminus. Isotopically ( 15 N, 13 C) labeled protein was overexpressed in Escherichia coli BL21(DE3) (DNAY) grown in 1 liter of M9 minimal medium with [ 15 N]ammonium chloride and [ 13 C]glucose. The cells were grown at 37°C to A 600 0.8, induced with 1 mM isopropyl-␤-D-thiogalactopyranoside, and incubated overnight (16 h) in a 15°C shaker. The cells were harvested by centrifugation, resuspended in 30 ml of binding buffer (20 mM Tris-HCl, pH 8.0, 500 mM NaCl, 5 mM imidazole), and lysed by sonication. The cells were centrifuged at 22,500 ϫ g for 15 min, and the supernatant was loaded on a Ni 2ϩ affinity column (Sigma), washed with 35 ml of binding buffer, and eluted with elution buffer (500 mM NaCl, 20 mM Tris-HCl, pH 8.0, 250 mM imidazole). The purified His-tagged N 1-74 was dialyzed into buffer (10 mM sodium phosphate, pH 6.9, 10 mM NaCl) and used for NMR structure determination. Typical NMR samples contained 1-1.4 mM N 1-74 . For CD spectroscopy, the His tag was cleaved by adding 0.08 mM TEV protease into purified His-tagged N 1-74 and dialyzing the mixture in an 8000 molecular mass cut-off dialysis tubing in buffer (50 mM Tris-HCl, pH 8.0, 0.5 mM EDTA, 1 mM dithiothreitol) for 16 h at room temperature.
Mutagenesis of N 1-74 -Site-specific mutations in the N 1-74 domain were introduced by PCR using the Stratagene QuikChange kit in two plasmids: (i) pET151-N 1-74 , used to overexpress recombinant His-tagged N 1-74 in E. coli, and (ii) pcDNA3.1-AND-N, used to express full-length Andes virus N protein in a mammalian cell line for immunocytochemistry (see below). The mutations were confirmed by DNA sequencing.
Peak volumes were analyzed to identify residues with slower hydrogen-deuterium exchange rates.
Structure Calculation-NOE distance restraints were classified into upper bounds of 2.7, 3.5, 4.5, and 5.5 Å and lower bound of 1.8 Å based on peak volumes. Backbone dihedral angles in the ␣-helical regions were restrained to (Ϫ60 Ϯ 20°) and (Ϫ40 Ϯ 20°). Hydrogen bonding distance restraints were used for ␣-helical residues that showed slow hydrogen-deuterium exchange rates. Initial structures were generated by torsion angle dynamics in CYANA (47), followed by molecular dynamics and simulated annealing in AMBER7 (48), first in vacuo and then with the generalized Born potential to account for the effect of solvent during structure calculation. CYANA and AMBER structure calculation protocols have been described elsewhere (49). Iterative cycles of AMBER calculations followed by refinement of NMR-derived restraints were performed until the structures converged with low restraint violations and good statistics in the Ramachandran plot. A family of 20 lowest energy structures was analyzed using PRO-CHECK (50), and graphics were generated using Pymol. The surface electrostatic potentials were calculated using APBS (51) and visualized in Pymol.
CD Spectroscopy-N 1-74 samples for CD spectroscopy contained 5-10 M protein in buffer (25 M Tris-HCl, pH 8, 3 M EDTA, and 5 M dithiothreitol). CD spectra were collected on a Jasco J-815 spectropolarimeter in triplicate. Wavelength scans were collected at 25°C at a scanning rate of 50 nm/min. Thermal denaturation scans at 222 nm were acquired with a temperature ramp rate of 1°C/min to a final temperature of 80°C, followed by cooling at 1°C/min to 25°C. The melting temperature (T m ) was determined from calculating the first derivative of thermal denaturation plots using the Jasco CD software.

RESULTS
NMR Structure Determination of N 1-74 -The His-tagged N 1-74 expressed well in soluble form in E. coli and yielded an excellent two-dimensional 1 H-15 N HSQC spectrum that showed distinct and well dispersed peaks (Fig. 1A). Nearly complete backbone assignments were obtained from three-dimensional HNCA, CBCA(CO)NH, HNCACB, and 15 N-edited NOESY-HSQC. The histidine residues of the His tag were overlapped and could not be assigned unambiguously. The C ␣ , H ␣ , C ␤ , and CЈ secondary chemical shifts (supplemental Fig. S2) showed that the first 33 residues, which were part of the His tag, were in random coil orientation, and the native N 1-74 sequence contained two ␣-helices (42). Side chain assignments were completed using two-dimensional 1 H-13 C HMQC, three-dimensional HBHA(CO)NH, and three-dimensional 13 C-edited HMQC-NOESY. Manual analysis of three-dimensional 15 Nand 13 C-edited NOESY spectra identified 1432 unambiguous interproton distance restraints. The NOE restraints together with 73 and 62 dihedral angle restraints and 38 hydrogen bond restraints (supplemental Table S1) were used in structure calculation and refinement in CYANA and AMBER. The 20 low energy NMR structures of N 1-74 converged into a family of structures ( Fig. 1B) with low restraint violations and good Ramachandran plot statistics (supplemental Table S1).
In addition to the conserved heptad repeats mentioned above, there are other highly conserved residues whose side chains are pointed toward the helix ␣ 1 -␣ 2 interface. These residues are nonpolar (Leu 7 and Leu 54 ), aromatic (His 14 ), polar (Gln 17 , Asn 40 , and Thr 43 ), or charged (Glu 6 , Glu 15 , Glu 29 , Lys 41 , Arg 47 , and Lys 57 ), and their side chains are pointed toward the helix ␣ 1 -␣ 2 interface (Fig. 2C). The polar and charged residues in this group do not participate in any salt bridge or hydrogen bonding contacts; however, their polar moieties are pointed toward the surface of the coiled coil, whereas the aliphatic portion of their side chains are involved in hydrophobic interaction that contribute to the stabilization of the hydrophobic core. The methyl groups of two invariant alanines, Ala 21 and Ala 28 , in helix ␣ 1 (Fig. 2C) are oriented toward the helix ␣ 1 -␣ 2 interface but do not contact any other residues on helix ␣ 2 , indicating that small side chains are required in those positions.

JOURNAL OF BIOLOGICAL CHEMISTRY 28299
Asp 35 , Asp 37 , and Asp 38 ) forms an acidic surface (Fig. 2D). We mutagenized many residues in this group (see below) to test the hypothesis that these residues are important in molecular recognition rather than in stabilizing the coiled coil structure.
Electrostatic Surface of N 1-74 -The N 1-74 domain is acidic (theoretical pI of 5.8), and the surface electrostatic potential map of N 1-74 shows distinct regions of negatively charged (red) and positively charged (blue) surfaces (Fig. 2D). The tip of the coiled coil, where the loop connecting the two helices are located, is negatively charged (Fig. 2D) because of clustering of conserved acidic residues (Asp 27 , Glu 29 , Glu 33 , Asp 35 , Asp 37 , and Asp 38 ) and polar residues (Asn 40 and Thr 43 ). Although the N 1-74 domain is acidic, there are conserved basic residues (Arg 22 , Lys 24 , Lys 26 and Arg 47 ) that form a positively charged surface just below the negatively charged tip (Fig. 2D). Point mutations in this positively charged surface have a dramatic effect on the antibody recognition of the N protein in vivo (see below).
In addition, there is a smaller negatively charged surface formed by Glu 9 and Glu 6 (Fig. 2D). Residue 9 could be acidic (Glu or Asp) or basic (Arg or Lys). Residue 9 is acidic among American hantaviruses (which cause the cardiopulmonary syndrome) and Old World hantaviruses that are nonpathogenic or cause a milder form of hemorrhagic fever with renal syndrome. Residue 9 is basic among Old World hantaviruses that causes the severe form of hemorrhagic fever with renal syndrome. Circular Dichroism Spectroscopy of N 1-74 -Point mutations were introduced in the basic (Arg 22 , Lys 24 , Lys 26 , and Arg 47 ) and acidic (Glu 33 and Asp 38 ) surfaces. In addition, we mutated Gln 23 , which is near the basic region, and Pro 36 , which is near the acidic region. These residues are surface-exposed (Fig. 2D) and are nearly invariant among hantaviruses (supplemental Fig.  S1). CD spectroscopy was used to assess the folding and stability of N 1-74 mutants. Wild type and point mutants showed  nearly identical CD spectra (Fig. 3A), indicating that the ␣-helical structure of N 1-74 was preserved. In addition, the ratio of ellipticity at 222 and 208 nm can be used to characterize ␣-helices. A 222 / 208 ratio of ϳ1.0 indicates ␣-helices with extensive interhelical contacts as in coiled coils and helical bundles, whereas a 222 / 208 ratio of ϳ0.8 indicates extended ␣-helices with little interhelical contacts (53)(54)(55). All N 1-74 constructs have a 222 / 208 ratio higher than 0.9 ( Table 1), suggesting that all mutants have the intact coiled coil structure. Further insight was provided by acquiring the CD melting temperatures (Fig.  3B and Table 1). Compared with wild type N 1-74 , the majority of mutants showed lower T m , with D38R having the lowest value, whereas two mutants (K24A and R47A) showed higher T m (Table 1). Nevertheless, all mutations were within Ϯ5°C of wild type T m ( Table 1), indicating that the mutations did not drastically alter the thermal stability of N 1-74 . Thus, the point mutations maintained the structural integrity of the N 1-74 coiled coil.
Immunocytochemistry of N Protein-Hantaviruses are believed to mature intracellularly; specifically, in the Golgi complex (56). During infection, the N protein was shown to localize cytoplasmically in the endoplasmic reticulum-Golgi intermediate compartment, presumably as they traffic from the endoplasmic reticulum to the Golgi (22). In addition, immunofluorescence of Cos-7 cells transfected with the N protein alone showed a granular pattern of staining in the perinuclear region (32,34), suggesting colocalization with the Golgi. To test our hypothesis that the conserved surface residues of N 1-74 are important in molecular interaction, we introduced point mutations designed to keep the N 1-74 coiled coil domain intact while altering only specific surface residues and transfected fulllength N protein in mammalian cells to observe the subcellular localization of the N protein. We used two types of anti-nucleocapsid antibodies, rabbit polyclonal and mouse monoclonal antibodies. The polyclonal antibody detected that wild type N and mutants (Arg 22 , Gln 23 , Lys 24 , and Lys 26 ) were located throughout the cytoplasm (Fig. 4A). The monoclonal antibody also detected wild type N and the Gln 23 and Lys 24 mutants throughout the cytoplasm in a similar pattern of staining as the polyclonal antibody (Fig. 4A). However, the monoclonal antibody showed a dramatic difference between the recognition of wild type N and the Arg 22 and Lys 26 mutants (Fig. 4A). Using the monoclonal antibody, Arg 22 and Lys 26 mutants were observed in a compact location lateral to the nucleus (Fig. 4A). To further define the subcellular localization of these N mutants, a Golgi-specific antibody (targeting the Golgi matrix protein GM130) was used (Fig. 4B). The Arg 22 and Lys 26 mutants were only detected by the monoclonal antibody when the N protein colocalized with the Golgi (Fig. 4B); however, these mutants were also present throughout the cytoplasm as shown by the polyclonal antibody (Fig. 4A). Thus, for the Arg 22 and Lys 26 mutants, the monoclonal antibody was able to distinguish between two populations of the N protein based on its subcellular localization in the cytoplasm or in the Golgi, the site of viral assembly and maturation (56). Other mutants (Glu 33 , Asp 35 , Pro 36 , Asp 37 , Asp 38 , and Arg 47 ) did not show this localization-dependent antibody recognition (supplemental Fig. S3).

DISCUSSION
The NMR structure of the Andes virus N 1-74 domain is similar to the recent crystal structure of the Sin Nombre virus nucleocapsid protein N-terminal coiled coil (N 1-75 ) (35). The C␣ root mean square deviation between the two structures is 1.3 Å. The crystal structure determination of the N protein addressed the issue of the trimerization of the N protein (35) because earlier models suggested the trimerization of the nucleocapsid N-terminal domain (29,30,32,33). A proposed model of N protein trimerization involves, first, the association of three N-terminal domains, followed by the association of three C-terminal domain (34). However, crystallography revealed that the Sin Nombre nucleocapsid N-terminal domain was monomeric and formed a coiled coil structure, and conserved hydrophobic residues participate in helix-helix interaction that stabilize the coiled coil (35). Our NMR structure of the Andes virus N 1-74 supports the crystallographic results; even at 1.4 mM, N 1-74 remained monomeric in solution. Our results, however, do not preclude the trimerization of full-length N protein in vivo by another mechanism.
A feature of the N 1-74 domain that had not been addressed in the literature is the role of many conserved polar residues whose side chains are pointed away from the coiled coil. Furthermore, the majority of these surface-exposed residues are not involved in polar interactions (Fig. 2D). Point mutations of these polar residues maintained the structural integrity and high thermal stability of the coiled coil ( Fig. 3 and Table 1). For example, Arg 22 , which forms a salt bridge with a conserved residue Glu 55 , can be mutated (R22F or R22M) without disrupting the coiled coil structure of N 1-74 (Table 1). R22F, which replaced arginine with a bulkier aromatic side chain, decreased the overall melting temperature by ϳ3°C (Table 1). This change is likely attributed to increased steric clash between phenylalanine and Glu 55 . However, the observation that R22M melts at a temperature comparable with that of wild type suggests that the salt bridge between Arg 22 and Glu 55 does not play a significant role in helix-helix interaction and that hydrophobic interaction is the major force stabilizing the coiled coil. A mutation in a nonpolar residue, Pro 36 , which is at the turn connecting the two ␣-helices of the coiled coil, had a T m approximately four degrees lower than wild type, which is consistent  (Fig. 3 and Table 1). Thus, there is no compelling structural reason for the high sequence conservation of surface residues. Furthermore, these The cell nuclei were stained blue using 4Ј,6-diamidino-2-phenylindole. Point mutations in Arg 22 and Lys 26 showed a dramatic difference in the monoclonal antibody recognition of Golgi-associated N protein, suggesting that the conformation or molecular interaction (or both) of the N protein is different when it is in the cytoplasm or when it is associated with the Golgi. WT, wild type. polar residues are clustered together on the surface of the N 1-74 domain and form distinct positively and negatively charged regions (Fig. 2D). We hypothesize that the reason for the clustering of conserved polar residues on the surface of N 1-74 is that they are sites of molecular recognition involved in the proper function of the N protein. Our mutagenesis and immunocytochemistry data suggest that point mutations in this group had a dramatic effect on the antibody recognition of the N protein with respect to its subcellular localization (Fig. 4).
During infection, nucleocapsids are trafficked to the cytoplasm (22) to assemble into mature virions (56). Mammalian cells transfected with the N protein alone show a granular pattern of immunofluorescence (32,34). This localization pattern is thought to be necessary for the nucleocapsid to perform its many functions in the establishment of an effective infection (22). We questioned whether the conserved polar surface residues in the coiled coil domain are important in the proper functioning of the N protein and reasoned that defects in the conformation or molecular recognition of the N protein will be manifested in the antibody recognition of the N protein in the context of its subcellular localization. CD spectroscopy confirmed that the mutant forms of N 1-74 maintained the structural integrity of the coiled coil structure ( Fig. 3 and Table 1); thus, the mutations altered only the surface property of the N protein.
Immunocytochemistry (Fig. 4) indicates that mutations in a conserved basic surface formed by Arg 22 and Lys 26 show monoclonal antibody recognition depending on the subcellular localization of the N protein. Polyclonal antibodies show that Arg 22 and Lys 26 mutants are present in the cytoplasm and Golgi; however, only Golgi-associated mutant nucleocapsids are detected by the monoclonal antibody (Fig. 4). Mutation of Arg 22 or Lys 26 changes the presentation of the N-terminal coiled coil to the monoclonal antibody. This change is dependent on the subcellular localization of the N protein.
There are two possible scenarios that could account for this differential monoclonal antibody recognition of the Arg 22 and Lys 26 mutants. First, there may be a difference in the conformation of the N-terminal coiled coil depending on whether the N protein is localized in the cytoplasm or in the Golgi, and this conformational change upon binding to the Golgi exposes the epitope, which is somewhere between residues 1-45 (comprising helix ␣ 1 , the interhelical loop, and part of helix ␣ 2 of the N 1-74 coiled coil; Fig. 1B), thereby allowing the monoclonal antibody to recognize the N protein associated with the Golgi. Second, the epitope may be masked differently by molecular interactions when the N protein is localized in the cytoplasm or in the Golgi. In addition to self-association, several host proteins such as SUMO-1 (17)(18)(19), Ubc9 (17,18), Daxx (20), actin (21), microtubules (22), and MxA (23) were reported to bind the N protein. Binding of the N protein with SUMO-1 and Ubc9 was required for localization of the N protein in the perinuclear region (17,19). Furthermore, because the N protein is not known to be a membrane protein, its localization in the Golgi must involve interaction with a Golgi-associated protein. Any of these molecular interactions could potentially alter the epitope presentation of the N 1-74 coiled coil and needs to be experimentally verified.
In summary, our structural results revealed that the highly conserved polar residues in the N-terminal coiled coil domain of the hantavirus nucleocapsid protein form distinct acidic and basic surfaces, and point mutations of the conserved basic surface formed by Arg 22 and Lys 26 allowed a monoclonal antibody to distinguish between two populations of the N protein based on its subcellular localization. Thus, in the Arg 22 or Lys 26 mutants, the conformation or molecular interaction of the N protein is different when it is in the cytoplasm or in the Golgi, the site of viral assembly and maturation.