Influenza A Virus Protein NS1 Exhibits Strain-Independent Conformational Plasticity

IAV is responsible for several pandemics over the last century and continues to infect millions annually. The frequent rise in drug-resistant strains necessitates exploring novel targets for developing antiviral drugs that can reduce the global burden of influenza infection. Because of its critical role in the replication and pathogenesis of IAV, nonstructural protein 1 (NS1) is a potential target for developing antivirals. Previous studies suggested that NS1 adopts strain-dependent “open,” “semiopen,” and “closed” conformations. Here we show, based on three crystal structures, that NS1 irrespective of strain differences can adopt an open conformation. We further show that NS1 from different strains primarily exists in an open conformation in solution and binds to cellular proteins with a similar affinity. Together, our findings suggest that conformational polymorphism facilitated by a flexible linker is intrinsic to NS1, and this may be the underlying factor allowing NS1 to bind several cellular factors during IAV replication.

ABSTRACT

Influenza A virus (IAV) nonstructural protein 1 (NS1), a potent antagonist of the host immune response, is capable of interacting with RNA and a wide range of cellular proteins. NS1 consists of an RNA-binding domain (RBD) and an effector domain (ED) separated by a flexible linker region (LR). H5N1-NS1 has a characteristic 5-residue deletion in the LR, with either G (minor group) or E (major group) at the 71st position, and non-H5N1-NS1 contains E71 with an intact linker. Based on the orientation of the ED with respect to the RBD, previous crystallographic studies have shown that minor group H5N1-NS1(G71), a non-H5N1-NS1 [H6N6-NS1(E71)], and the LR deletion mutant H6N6-NS1(Δ80-84/E71) mimicking the major group H5N1-NS1 exhibit “open,” “semiopen,” and “closed” conformations, respectively, suggesting that NS1 exhibits a strain-dependent conformational preference. Here we report the first crystal structure of a naturally occurring H5N1-NS1(E71) and show that it adopts an open conformation similar to that of the minor group of H5N1-NS1 [H5N1-NS1(G71)]. We also show that H6N6-NS1(Δ80-84/E71) under a different crystallization condition and H6N6-NS1(Δ80-84/G71) also exhibit open conformations, suggesting that NS1 can adopt an open conformation irrespective of E or G at the 71st position. Our single-molecule fluorescence resonance energy transfer (FRET) analysis to investigate the conformational preference of NS1 in solution showed that all NS1 constructs predominantly exist in an open conformation. Further, our coimmunoprecipitation and binding studies showed that they all bind to cellular factors with similar affinities. Taken together, our studies suggest that NS1 exhibits strain-independent structural plasticity that allows it to interact with a wide variety of cellular ligands during viral infection.

IMPORTANCE IAV is responsible for several pandemics over the last century and continues to infect millions annually. The frequent rise in drug-resistant strains necessitates exploring novel targets for developing antiviral drugs that can reduce the global burden of influenza infection. Because of its critical role in the replication and pathogenesis of IAV, nonstructural protein 1 (NS1) is a potential target for developing antivirals. Previous studies suggested that NS1 adopts strain-dependent “open,” “semiopen,” and “closed” conformations. Here we show, based on three crystal structures, that NS1 irrespective of strain differences can adopt an open conformation. We further show that NS1 from different strains primarily exists in an open conformation in solution and binds to cellular proteins with a similar affinity. Together, our findings suggest that conformational polymorphism facilitated by a flexible linker is intrinsic to NS1, and this may be the underlying factor allowing NS1 to bind several cellular factors during IAV replication.

INTRODUCTION

Influenza A viruses (IAVs) belong to the Orthomyxoviridae family, are responsible for acute, highly contagious respiratory disease, and have been linked to over 80,000 deaths in the 2017-2018 season in the United States (Centers for Disease Control and Prevention). IAVs encode nonstructural protein 1 (NS1), a multifunctional protein capable of interacting with various host cellular ligands, which is essential for the virus’s replication, spread, and pathogenesis (1, 2). NS1 blocks the host immune response through several mechanisms, including inactivation of the 2′-5′-oligoadenylate synthetase (OAS)/RNase L pathway by binding to double-stranded RNA (dsRNA) (3), blocking apoptosis by interacting with the p85β regulatory subunit of phosphoinositide 3-kinase (PI3K) (4, 5), preventing protein kinase R (PKR) activity (6), and inhibiting antiviral mRNA (mRNA) maturation by inhibiting the functioning of the cleavage and polyadenylation specificity factor 30 (CPSF30) (7).

Full-length (FL) structures of NS1 have been determined from H5N1 (A/Vietnam/1203/2004) and H6N6 (A/blue-winged teal/MN/993/1980) strains (8, 9). NS1 contains two well-defined domains: the N-terminal RNA-binding domain (RBD; residues 1 to 73) and the C-terminal effector domain (ED; residues 84 to 220) connected through a flexible linker region (LR). Although the overall polypeptide fold of each domain is conserved, the relative orientations of the ED with respect to the RBD in these two structures are altered due to changes in the linker region (see below). In addition to the FL-NS1 structures, crystal structures of individual RBD (10–12) and ED (13–17) from various IAV strains, and also that of RBD with RNA oligomer (10), and of ED in complex with the F2F3 domain of CPSF30 have been determined (18). These studies show that both RBD and ED individually form dimers, as also observed in the FL-NS1 structures, and that the polypeptide fold of these two domains remains the same in all these structures. While the structures of RBD show a conserved dimeric interface (10–12), crystal structures of ED show variations in the homodimeric interfaces involving either strand-strand or helix-helix interactions (13–17). The structure of ED in complex with F2F3 shows an interesting variation in which ED dimer formation involves a head-head interaction (18). These findings appear to suggest that ED itself can have transient homodimeric interfaces to allow interaction with several host antiviral factors (15).

FL-NS1 structures have been broadly categorized into two classes, H5N1 and non-H5N1 types (9). The H5N1 strain has a deletion of amino acids 80 to 84 in the LR, which has been linked to increased viral replication and pathogenicity (19, 20). The H5N1 strains can be further classified into a major group and a minor group. The major group of H5N1-NS1 contains glutamate (E) at the 71st residue, while the minor group of H5N1-NS1 contains a glycine (G) residue (9) (Fig. 1A). Non-H5N1-NS1 has no deletion in the LR and contains a glutamate (E) residue at the 71st position. The crystal structure of a minor group of H5N1-NS1 [A/Vietnam/1203/2004; referred to as H5N1-NS1(G71)] showed that full-length NS1 exists as a dimer via the RBD, whereas the two EDs flank the side of the RBD dimer, adopting an “open” conformation (8) (Fig. 1B). Interestingly, the crystal structure of H6N6-NS1 [A/blue-winged teal/MN/993/1980; referred to as H6N6-NS1(E71)], representing a non-H5N1-NS1, also exhibits a dimer via the RBD similar to H5N1-NS1(G71), but its EDs are in closer proximity, referred to as a “semiopen” conformation (9). To understand the effect of the LR deletion, a construct with a five-amino-acid deletion (residues 80 to 84) in the LR, referred to as H6N6-NS1(Δ80-84/E71), was made, which arguably is a mimic of the major group of H5N1-NS1. Interestingly, in contrast to the expected open conformation, the crystallographic structure of H6N6-NS1(Δ80-84/E71) showed a “closed” conformation (9), where two EDs in the dimer move closer to each other. These observations appear to suggest that NS1 exhibits a strain-dependent, i.e., open, semiopen, or closed, conformation based on the orientation of the ED relative to the RBD. Since H6N6-NS1 with a deletion of amino acids 80 to 84 and with E at the 71st position represents the major group of H5N1-NS1, Carrillo et al. (9) suggested that the major group of H5N1-NS1 should adopt a closed conformation. However, the structure for an original major group of H5N1-NS1 has not yet been reported.

FIG 1.

(A) Multiple sequence alignment of H5N1-NS1 and H6N6-NS1 showing overall conservation of amino acid residues. RBD, LR, and ED are highlighted in blue, red, and green. The linker region (LR) is further zoomed in, with the residue at the 71st position (highlighted in red) showing its conservation among H5N1-NS1 and non-H5N1-NS1 strains. A minority of H5N1-NS1 has a characteristic 5-amino-acid deletion in the linker region and a G at the 71st position. A majority of H5N1-NS1 contains a 5-amino-acid deletion but an E at the 71st position. Non-H5N1-NS1 has no deletion in the linker region and contains E71. (B) Structural plasticity of NS1: open conformations of H5N1-NS1(G71) dimer (green; PDB ID 3F5T); semiopen conformation of H6N6-NS1(E71) dimer (salmon; PDB ID 4OPH); and closed conformation of H6N6-NS1(Δ80-84/E71) dimer (cyan; PDB ID 4OPA).

To investigate the conformational state of a naturally occurring major group of H5N1-NS1, we determined the structure of the first full-length major group of H5N1-NS1 [H5N1-NS1(E71)]. Interestingly, the crystal structure of the major group revealed an open conformation more similar to that of the minor group of H5N1-NS1. This indicates that NS1 can adopt an open conformation irrespective of the residue (E or G) present at the 71st position. To further examine the role of the 71st residue (E or G) in the H6N6 background, we mutated E to G at the 71st position and determined the crystallographic structure of H6N6-NS1(Δ80-84/G71). As expected, this structure also adopted an open conformation. To further understand the conformational dynamics of NS1 constructs in solution, single-molecule Förster/fluorescence resonance energy transfer (smFRET) studies were performed with H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71), and they revealed that NS1 exists in an equilibrium state primarily in an open conformation in solution. We also validated our findings by coimmunoprecipitation (co-IP) and isothermal titration calorimetry (ITC). Our study revealed that all NS1 constructs with various crystallographically observed conformations bind to cellular ligands with similar binding properties. These findings strongly suggest that the NS1 of any particular strain has structural plasticity, by virtue of the inherent flexibility in the LR, to adopt context-dependent conformations that enable it to interact with a wide range of host cellular proteins.

RESULTS

H6N6-NS1 Δ80-84 exists in an open conformation with G at the 71st position.

H6N6-NS1(E71) with a characteristic longer LR adopts a semiopen conformation (9). With the expectation that a shorter LR, as in H5N1-NS1, is the key to an open conformation, Carrillo et al. (9) deleted amino acids 80 to 84 in the LR of H6N6-NS1. However, the structure of H6N6-NS1(Δ80-84/E71) showed a closed conformation, in contrast to the minor group of H5N1-NS1 (shorter LR and G71) that adopts an open conformation. Based on these observations, we hypothesized that the nature of the residue at the 71st position (G or E) is critical for determining the conformation of NS1 and that the E-to-G mutation of H6N6-NS1 Δ80-84 would favor an open conformation.

To test the above-mentioned hypothesis, we mutated E to G in H6N6-NS1 Δ80-84 and determined the crystal structure at a resolution of ∼3.2 Å (Table 1). This E71G construct crystallized in the P6₅22 space group with one NS1 molecule in the crystallographic asymmetric unit (AU) coincidently in the same space group as that observed in the crystals of minor group H5N1-NS1 (Fig. 2A). The structural features of the RBD and ED in H6N6-NS1(Δ80-84/G71) are well conserved, as indicated by superposing it with H6N6-NS1(Δ80-84/E71), showing root-mean-square deviations (RMSD) of 0.5964 Å and 0.7021 Å for RBD and ED, respectively. As in the structure of H6N6-NS1(Δ80-84/E71), the structure of H6N6-NS1(Δ80-84/G71) exists as a dimer stabilized by similar RBD-RBD dimeric interactions (Fig. 2B), although in the H6N6-NS1(Δ80-84/E71) structure these interactions are between the two molecules in the same crystallographic asymmetric unit, and in the H6N6-NS1(Δ80-84/G71) structure they are between the molecules related by the crystallographic 2-fold axis. However, in the context of the full-length NS1, structural alignment of H6N6-NS1(Δ80-84/G71) and H6N6-NS1(Δ80-84/E71) show significant differences in the relative orientations between the RBD and ED, clearly indicating that NS1(Δ80-84/G71) adopts an open conformation similar to that observed in the structure of the minor group H5N1-NS1 (with G in position 71) (Fig. 2C and D).

TABLE 1.

Data collection and refinement statistics^a

Parameter	Valueb for:
	H5N1-NS1(E71)	H6N6-NS1(Δ80-84/G71)	H6N6-NS1(Δ80-84/E71)
Data collection statistics
Space group	P6522	P6522	P6522
Unit cell dimensions
a, b, c (Å)	105.57, 105.57, 82.40	105.19, 105.19, 80.21	106.16, 106.16, 79.90
α, β, γ (o)	90.00, 90.00, 120.00	90.00, 90.00, 120.00	90.00, 90.00, 120.00
Resolution (Å)	30.60–3.00 (3.77∼3.00)	34.43–3.2 (3.315∼3.20)	28.61–3.89 (4.00∼3.89)
Rmerge	0.099 (1.640)	0.135 (1.401)	0.136 (0.94)
Avg I/σ	18.5 (2.1)	14.6 (2.1)	12.8 (4.0)
Completeness (%)	99.9 (100)	99.61 (100)	95.2 (99.4)
Multiplicity	19.8 (20.3)	16.2 (17.1)	10.9 (17.8)
Refinement statistics
Resolution range (Å)	30.60–3.00	34.43–3.2	28.61–3.89
No. of reflections	5,786	4,652	2,536
Rwork/Rfree	0.229/0.279	0.243/0.259	0.244/0.295
No. of atoms
Proteins	1,520	1,533	1,375
Ligand/ion	0	0	0
Water	0	0	0
B-factors (Å2)
Proteins	66.0	122.0	153.0
Ligand/ion	NA	NA	NA
Water	NA	NA	NA
RMSD
Bond lengths (Å)	0.005	0.004	0.003
Bond angles (o)	0.94	0.901	0.560

^aOne crystal was used for each structure.

^bValues in parentheses are for the highest-resolution shell. NA, not applicable.

FIG 2.

H6N6-NS1(Δ80-84/G71) crystal structure. (A) Cartoon representation of the H6N6-NS1(Δ80-84/G71) monomer structure with RNA-binding domain (RBD), linker region (LR), and effector domain (ED). (B) H6N6-NS1(Δ80-84/G71) dimer mediated by RBD-RBD interfaces forming an open conformation. (C) Superposition of H6N6-NS1(Δ80-84/E71) (closed conformation in cyan; PDB ID 4OPA) and H6N6-NS1(Δ80-84/G71) (open conformation in gold; PDB ID 6NRL) monomers demonstrates significant differences in ED orientations with respect to the RBD. (D) Superposition of H6N6-NS1(Δ80-84/E71) (cyan) and H6N6-NS1(Δ80-84/G71) (gold) dimers demonstrates the conformational shift of ED orientations between closed and open conformations.

In the closed conformation [H6N6-NS1(Δ80-84/E71)] determined by Carrillo et al. (9), residues at the 74th to 77th positions in the linker region form a type I β-turn, allowing the linker regions of the two subunits to cross over, bringing the EDs of the dimeric subunits closer to form a closed conformation. In this conformation, both the ED and the RBD participate in additional intersubunit dimeric interactions. These include stacking interactions between the Tyr84 residues of the two ED subunits and hydrogen bond interactions between Gln25 of RBD and Glu71 of LR. These stabilizing interactions result in a clearly defined density for the LR. However, in the structure of H6N6-NS1(Δ80-84/G71), which adopts an open conformation, with the absence of these interactions, the LR is more flexible, as indicated by a higher B-factor and a poorly defined density in this region.

The major group of H5N1-NS1 takes an open conformation similar to that of the minor group.

The structures of H6N6-NS1(Δ80-84/G71) and the minor group NS1 with G at the 71st residue showing an open conformation and the structure of the H6N6-NS1(Δ80-84/E71) showing a closed conformation appear to imply that the nature of the residue at the 71st position acts as a switch between open (with G71) and closed (with E71) conformations. To ascertain if this is indeed the case, we determined the crystal structure of an authentic major group of H5N1-NS1 [H5N1-NS1(E71)] to a resolution of 3 Å (Table 1).

This major group of H5N1-NS1, similar to the minor group of H5N1-NS1, also crystallized in the P6₅22 space group with one molecule in the AU (Fig. 3A). In the crystals, the RBD of the monomeric subunits related by the crystallographic 2-fold axis interacts closely to form a dimer with the EDs positioned on either side, resulting in an open conformation similar to that observed in the case of minor group NS1 (Fig. 3B). The structural comparison of major and minor group NS1 reveals that the overall secondary structure and peptide folds of the RBD and ED remain conserved, with RMSD values of 0.52 Å and 1.42 Å, respectively. However, when the full-length major (E at position 71) and minor (G at position 71) groups of H5N1-NS1 structures in their entirety were superposed, the overall RMSD value was 5.578 Å because of significant repositioning of the ED with respect to the RBD (Fig. 3C). The ED of the major group is shifted away from the ED of the minor group by ∼10 Å (Cα backbone distance) with respect to the RBD. Unlike in the closed conformation, as observed in H6N6-NS1(Δ80-84/E71), the 74th to 77th residues in the LR of H5N1-NS1(E71) do not form a β-turn and lack similar interdomain interactions despite the presence of E71. A lack of stabilizing interactions between the ED and the RBD imparts flexibility in the LR, as indicated by a high B-factor and poor density similar to that observed in the crystal structures of either H6N6-NS1(Δ80-84/G71) or the minor group NS1 with G71. Thus, these structures with E at the 71st position indicate that NS1 can take either an open or closed conformation due to the intrinsic flexibility in the LR.

FIG 3.

Major group of H5N1-NS1 crystal structures. (A) Cartoon representation of the H5N1-NS1(E71) monomer structure containing the RNA-binding domain (RBD) and the effector domain (ED) connected through the linker region (LR). (B) H5N1-NS1(E71) dimer mediated by RBD-RBD interface forming an open conformation. (C) Superposition of major (magenta; PDB ID 6O01) and minor (green; PDB ID 3F5T; disordered residues in the LR are shown as a broken green line) groups of H5N1-NS1 monomers demonstrates a shift of ED with respect to the RBD. The inset shows that Asp119 in the ED domain of H5N1-NS1(E71) has shifted by 11.5 Å from Asp119 in the ED domain of H5N1-NS1(G71). (D) The ED of the major group shifts significantly in comparison to the minor group when major (magenta) and minor (green) groups of H5N1-NS1 dimers are superposed.

Interestingly, the major groups of H5N1-NS1 and H6N6-NS1(Δ80-84/G71), similar to the minor group of H5N1-NS1, crystallize in the space group P6₅22. In each case, the crystal packing analysis shows that neighboring NS1 monomers interact with alternating RBD-RBD and ED-ED interactions to form a linear chain, and three such chains related by the crystallographic 3₁ screw axis form a tubular structure with a central hole having a diameter of ∼20 Å (Fig. 4). This type of tubular structure is only possible with the open conformational state. It is suggested that such a tubular structure may be important for sequestering dsRNA (8).

FIG 4.

Three chains of NS1 in an open conformation related by crystallographic 3₁-screw axis exhibit a tubular structure. (A) A tubular structure is formed by a minor group of H5N1-NS1 (PDB ID 3F5T) chains colored in red, blue, and green. (B) Three chains of a major group of H5N1-NS1 (PDB ID 6O01) colored in marine blue, sand, and magenta also form a higher-ordered tubular structure. (C) Chains of H6N6-NS1(Δ80-84/G71) (PDB ID 6NRL) colored in firebrick, bright orange, and gray also exhibited a tubular structure. A 90° rotation of the NS1 tubule exhibits the RBD and ED of each NS1 dimer related by crystallographic 2-fold axes interacting with the respective domains of neighboring NS1 molecules, resulting in long chains of NS1. Due to variations in the open conformation, W182 (yellow spheres) of H5N1-NS1(E71) and H6N6-NS1(Δ80-84/G71) dimers have moved ∼10 Å apart in comparison to the minor group of H5N1-NS1. Irrespective of slight variations in the open conformations formed by H5N1-NS1(G71), representing the minor group H5N1-NS1, H5N1-NS1(E71), representing the major group H5N1-NS1, and H6N6-NS1(Δ80-84/G71) (a mimic of major group H5N1-NS1), they all can form long filamentous oligomeric chains with a conserved tunnel diameter (20 Å) sufficient enough to accommodate dsRNA. Conserved Arg38 implicated in dsRNA binding is highlighted (cyan).

Interaction of H5N1-NS1 and H6N6-NS1 constructs with cellular ligands.

Based on three different crystallographically observed conformations (open, semiopen, and closed) of H5N1-NS1 and H6N6-NS1, it would be expected that these NS1 proteins have differential binding with different substrates (Fig. 1B). However, the crystal structures of the major groups of H5N1-NS1(E71) and H6N6-NS1(Δ80-84/G71) showed that both NS1s take up open conformations irrespective of E/G at the 71st position and further confirmed the overall conformational flexibility intrinsic to NS1. To investigate whether crystallographically observed conformations have any effect on binding to the cellular ligands, we determined the binding of various strains/mutants of NS1 constructs with known cognate cellular partners, such as PI3K (4, 5), CPSF30 (7, 18), and Crk/L (21, 22), using coimmunoprecipitation. 293T cells were transfected with FLAG-tagged FL-NS1, including H6N6-NS1(E71), H6N6-NS1(Δ80-84/E71), H6N6-NS1(Δ80-84/G71), H5N1-NS1(G71), and H5N1-NS1(E71). Immunoblotting detected the presence of all three cellular ligands (CPSF30, PI3K, and Crk/L) irrespective of crystallographically observed NS1 conformations (Fig. 5). However, we observed a differential level of CPSF30 binding to NS1 among different conformations. These findings indicate that crystallographically observed conformations of NS1 might not have a substantial impact on binding to cellular ligands. There might be two possibilities that could be inferred from these observations: (i) strain-specific conformations interact with all three cellular ligands but with different affinities, which in turn can impact the level of pathogenicity they induce, or (ii) more likely, three conformations are interconvertible due to intrinsic flexibility in the LR, as demonstrated in the structures to ensure a context-dependent interaction with cellular ligands.

FIG 5.

Coimmunoprecipitation assays of NS1 constructs with cellular ligands. 293T cells were transfected with FLAG-tagged NS1s and a negative control (empty vector). At 48 h posttransfection, cell extract was prepared and NS1 was pulled down from the extract by performing a binding assay with FLAG antibody, and cellular proteins bound to NS1 were coimmunoprecipitated. Products of the binding assay were resolved in SDS-PAGE and examined by immunoblotting. Cellular PI3K, CPSF30, and Crk/L were immunoprecipitated with NS1 and detected with corresponding antibodies in the immunoblot. β-Actin was analyzed in cell extracts to verify that equal amounts of extracts were used for the immunoprecipitation assay.

FL-NS1s with different conformations interact with F2F3 with similar binding affinities.

Although we saw a consistent open conformation in FL-NS1 irrespective of the strain, there were variations in the open conformation itself, as evident from available crystal structures. The major and minor groups of H5N1-NS1 formed an open conformation, but there was an overall ∼10-Å shift in the ED with regard to the RBD and a further shift (semiopen) in the ED with a longer LR, as in the H6N6-NS1 structure. We investigated further whether the variability in the open conformations impacts the binding affinity to the F2F3 domain of CPSF30. We should note that if H6N6-NS1(Δ80-84/E71) adopted a closed conformation in solution, also as observed crystallographically, there would be a steric hindrance if this NS1 interacted with CPSF30, as depicted in the previously reported crystal structure ED-F2F3 complex (18). To analyze whether different NS1s show variations in binding affinity for the F2F3 domain, we performed binding analysis using ITC (Fig. 6). The H6N6-NS1(E71) (semiopen conformation) showed a dissociation constant (K_d) of 30 nM, whereas, with H6N6-NS1(Δ80-84/E71) (closed conformation), the estimated K_d was 86.9 nM. For H6N6-NS1(Δ80-84/G71) and H5N1-NS1(E71) (both open conformations), the K_ds were 44 and 65.3 nM, respectively. These results indicate that irrespective of the conformations observed crystallographically, NS1 interacts with the cellular ligand F2F3 with a similar binding affinity, strengthening the notion that NS1 of the same strain can undergo conformational changes relevant for the interaction with a specific cellular protein.

FIG 6.

ITC analysis of F2F3 binding to different NS1 constructs. The original raw data (top) and the best-associated fit after buffer subtraction and integration (lower graph) for a one-site model show F2F3 binding to NS1. Binding affinity is reported in the inset. ITC analysis of the major group of H5N1-NS1 [H5N1-NS1(E71) with open conformation] (A), H6N6-NS1(Δ80-84/G71) (open conformation) (B), H6N6-NS1(E71) (semiopen conformation) (C), andH6N6-NS1(Δ80-84/E71) (closed conformation) interacting with F2F3 (D). All NS1 constructs bind with the cellular substrate with nanomolar affinities.

NS1 constructs exist primarily in an open conformation in solution.

smFRET spectroscopy provides a tool to directly detect the different NS1 conformations that are in a dynamic equilibrium in solution. We performed smFRET experiments on freely diffusing H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) to determine their preferred conformational state in solution. For this experiment, we singly labeled NS1 monomer constructs at C111 with either Alexa Fluor 488 or Alexa Fluor 594 (FRET donor and acceptor probes) (Fig. 7A). Because NS1 is a homodimer, only the ones labeled with both donor and acceptor dyes will register a FRET signal, whereas donor-only and acceptor-only molecules will show zero peaks or will be invisible, respectively. smFRET histograms of both H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) show single FRET populations in solution. The measured FRET efficiency (E_FRET) for H5N1-NS1(E71) is 0.65 (±0.04), which corresponds to an interdye distance of around ∼50 Å (Fig. 7B), using a Forster distance (R₀) of 55 Å and a γ of unity as previously described for the Alexa Fluor 488-Alexa Fluor 594 dye pair (23). Upon binding to the F2F3 domain (Fig. 7C), H5N1-NS1(E71) undergoes expansion, as evidenced by the decrease in E_FRET (0.55 ± 0.02; ∼53 Å) (Fig. 7D, composite figure). E_FRET of unbound H6N6-NS1(Δ80-84/E71) is measured to be 0.60 ± 0.08 (∼51 Å) (Fig. 7E). Similar to the major group of H5N1-NS1, H6N6-NS1(Δ80-84/E71) undergoes conformational expansion upon binding to F2F3 (Fig. 7F), as indicated by the decrease in E_FRET (0.39 ± 0.02; ∼59 Å) (Fig. 7G, composite figure). The measured E_FRET values for unbound H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) suggest similar conformations for the two constructs in solution. The observed E_FRET for H6N6-NS1(Δ80-84/E71) suggests that it exists in a more open conformation than what is expected on the basis of its crystal structure (Protein Data Bank [PDB] ID 4OPA). To further substantiate this, we also determined the structure of H6N6-NS1(Δ80-84/E71) under a crystallization condition (PDB ID 6OQE) different from the previously reported condition (9). Interestingly, the H6N6-NS1(Δ80-84/E71) structure also showed an open conformation (Fig. 8), in contrast to the previously reported closed conformation (9), consistent with the E_FRET signatures of H6N6-NS1(Δ80-84/E71) and H5N1-NS1(E71). The very similar E_FRET values observed for the unbound forms of both H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) suggest that the conformations adopted by the two proteins in solution are also similar. Similarly, F2F3 binding to either of the constructs induces an open conformation, although with H6N6-NS1(Δ80-84/E71), it is a more drastic shift.

FIG 7.

Binding-induced conformational changes in NS1 constructs detected by single-molecule fluorescence resonance energy transfer (FRET). FRET efficiency (E_FRET) histograms were fit as Gaussian distributions by nonlinear least-squares (NLS) analysis; cyan and red areas represent free and bound forms of the proteins, respectively. (A) FRET pairs labeled on C111 (green and red balls) on H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71). Crystal structures of H5N1-NS1(E71) (left, PDB ID 6O01) and H6N6-NS1(Δ80-84/E71) (right, PDB ID 4OPA) are shown in cartoon representations. (B and C) smFRET histograms for the free and F2F3-bound species of H5N1-NS1(E71), respectively. (D) Superposition of smFRET histograms of free and ligand-bound H5N1-NS1(E71) showing clear conformational shift upon binding evidenced by a decrease in E_FRET. (E and F) smFRET histograms for the free and F2F3-bound forms of H6N6-NS1(Δ80-84/E71), respectively. (G) Superposition of smFRET histograms of free and bound H6N6-NS1(Δ80-84/E71), similarly exhibiting an E_FRET reduction due to ligand binding.

FIG 8.

H6N6-NS1(Δ80-84/E71) crystal structure. (A) Structural superposition of the dimers of H6N6-NS1(Δ80-84/E71) crystallized under different conditions exhibiting different ED orientations (“closed” in cyan, “open” in gray) with respect to the RBD. (B) H5N1-NS1(E71), H6N6-NS1(Δ80-84/G71), and H6N6-NS1(Δ80-84/E71) in magenta, gold, and gray, respectively, show NS1 can consistently adopt an open conformation with a slight variability in ED orientations.

DISCUSSION

Three different crystallographic conformations of NS1 (open, semiopen, and closed) have been previously reported (8, 9). These structural observations raised the possibility that LR length and the nature of the residue at the 71st position (E or G) may be the determinants of the conformational states of NS1. In this study, the crystal structures of H5N1-NS1(E71), H6N6-NS1(Δ80-84/G71), and H6N6-NS1(Δ80-84/E71) clearly show that irrespective of E or G at the 71st position, NS1 with a shorter linker can adopt an open conformation (Fig. 9). From the previously reported FL-NS1 crystal structures, and the crystal structures reported in our study, we can surmise that the polypeptide fold of the individual domains and the RBD dimer remain invariant. The main variant is the degree of flexibility in the LR of these structures. In the context of the dimeric structure of the FL-NS1, it is evident that as the EDs move closer to one another, the flexibility of the LR is restricted, as observed in the semiopen H6N6-NS1 and closed H6N6-NS1(Δ80-84/E71) structures. This restricted flexibility, as evident from the well-defined density in the LR, is perhaps due to the formation of the type 1 β-turn by residues 74 to 77 in the LR. In contrast, in the structures with an open conformation, invariably the LR exhibits a higher degree of flexibility, resulting in poorer electron density in the LR and consequently higher B-factors. It is also noticeable that the degree of flexibility in the LR varies among the structures that exhibit open conformations. In the open conformation of the previously reported minor group H5N1-NS1(G71), the density in the LR overall is poor, and there is no visible density between amino acids 70 and 74 (8). In the open conformation structures of H5N1-NS1(E71) and H6N6-NS1(Δ80-84/G71) reported here, the LR density is poor but strong enough to trace Cα backbone density. Although the crystallization conditions and crystal packing may be the factors affecting this variability, these observations nonetheless reinforce the adaptability and flexibility that is inherent in the LR.

FIG 9.

Structural plasticity of NS1. Open conformations of H5N1-NS1(G71) dimer (green; PDB ID 3F5T), H5N1-NS1(E71) dimer (magenta; PDB ID 6O01), H6N6-NS1(Δ80-84/E71) dimer (gray; PDB ID 6OQE), and H6N6-NS1(Δ80-84/G71) dimer (gold; PDB ID 6NRL); semiopen conformation of H6N6 NS1(E71) dimer (salmon; PDB ID 4OPH); and closed conformation of H6N6-NS1(Δ80-84/E71) dimer (cyan; PDB ID 4OPA) demonstrating various ED orientations with respect to the RBD. Intrinsic flexibility of the linker region allows NS1 to adopt different conformational states suitable for binding to specific cellular ligands.

A fascinating observation from our studies is that crystals of NS1, in which NS1 adopts an open conformation, although crystallized under different conditions, all showed the same P6₅22 space group and the formation of a similar tubular structure in the crystal lattice. Based on cryoelectron microscopy (cryo-EM) images of the H1N1 NS1 with dsRNA showing filamentous structures with a diameter similar to that of the crystallographically observed tubular structure, it is suggested that such tubular structures could form a basis for sequestration of dsRNA (8). In the crystallographically observed tubular structure, although the α-helix in the RBD responsible for binding to dsRNA, based on the crystal structure of the RBD in complex with dsRNA (10), is positioned facing the central hole, the location of conserved Arg38 that is critical for dsRNA binding is not optimally positioned (Fig. 4). For optimal interaction with dsRNA, the RBD has to reorient slightly to allow Arg38 to protrude toward the center of the hole, as proposed by Aramini et al. (24). It is conceivable that such reorientation can be achieved due to the flexibility of the LR. It is clear from the crystal structures that the ED-ED interactions play a critical role in the formation of the linear chains and subsequent tubular structures. Previous studies have shown that although the RBD alone can bind to dsRNA, the binding affinity is considerably lower than that of FL-NS1 (24). This suggests that although ED does not bind to RNA directly, the formation of long filamentous structures mediated by ED-ED interactions plays a crucial role in the cooperative binding of dsRNA (8, 24).

Considering that the adoption of an open conformation and the formation of a tubular structure by NS1 has been observed only in crystals of NS1 with a 5-residue deletion in the LR, including H5N1-NS1 (major and minor) and H5N1-NS1 mimics [H6N6-NS1(Δ80-84/E71) and H6N6-NS1(Δ80-84/G71)], a relevant question is whether non-H5N1-NS1, i.e., without a 5-residue deletion in LR, can form such tubular structures. The only crystal structure of a non-H5N1-NS1 reported to date is that of H6N6-NS1 (9), which forms a semiopen conformation. Although H6N6-NS1 does not form a tubular structure, as observed with H5N1-NS1, it does form higher-order chain-like structures involving alternating RBD-RBD and ED-ED interactions. In order for H6N6-NS1 to form a tubular structure, it has to switch from a semiopen to an open conformation. Similarly, any NS1 adopting a closed conformation has to switch to an open conformation to form a tubular structure. The observation from the cryo-EM images that H1N1-NS1, with an intact linker, forms filamentous structures in the presence of dsRNA (8) suggests that a non-H5N1 NS1 also can adopt an open conformation to form tubular structures. It is to be noted here that tubular structures observed in the crystal lattice involve a crystallographic 3₁ screw axis; however, in the context of dsRNA binding, the positioning of the RBD and possibly the ED are likely to be influenced by the conformation of dsRNA. The precise structural details of how FL-NS1 responds to binding dsRNA requires further structural studies. Overall, these observations are consistent with the notion that although the RBD is responsible for interaction with dsRNA, the cooperative ED-ED interactions leading to the formation of tubular structures facilitate sequestration of dsRNA intermediates formed during virus replication to prevent an antiviral immune response.

An important observation from our studies is that NS1 irrespective of the strain can adopt any of the conformations, starting from a closed, semiopen, to open conformation, because of the flexibility of the LR while maintaining the same polypeptide folds of the RBD and ED. This is clearly demonstrated in our crystallographic studies of H6N6-NS1(Δ80-84/E71). Although this construct was used as a surrogate mimicking the major group of H5N1 NS1, our crystal structural analysis shows that the same construct can adopt a closed conformation or an open conformation depending upon the crystallization condition. Although an open conformation is likely the most preferred state in solution, our in vivo studies and binding analysis suggest that NS1 can switch to a suitable conformational state that allows for optimal interactions with the ligand. This is clearly exemplified by the recently reported structure of NS1 in complex with the tripartite motif-containing protein 25 (TRIM25) coiled-coil domain (25). In this structure, NS1 from an H1N1 strain which is expected to form a semiopen conformation like H6N6-NS1, because it has a longer LR and E at the 71st position, adopts a conformation that is between the semiopen and closed conformations.

Together, our findings clearly suggest that the conformational state of NS1 is not strain dependent and that it can form different conformations to sample wide conformational space, allowing it to bind a broader range of substrates. The crystallographically observed conformational states thus represent snapshots in this conformational space from a closed to an open state. Although the closed state is observed only in H6N6-NS1(Δ80-84/E71), which is designed only to mimic the major group of H5N1-NS1, our observation that it can transit to an open conformation indicates that perhaps the closed state is also accessible to NS1 under certain conditions. Such conformational adaptability and interconvertibility mediated by the flexibility inherent in the LR perhaps explain the mechanism of how a small protein such as NS1 interacts with as many as 30 cellular proteins and provides the basis for the multifunctionality of NS1 contributing to its virulence.

MATERIALS AND METHODS

Plasmid construction.

H6N6-NS1(E71) (A/blue-winged teal/MN/993/1980), H6N6-NS1(Δ80-84/E71), and H5N1-NS1(G71) (A/Vietnam/1203/2004) were cloned into the bacterial expression vector pET-46-Ek/LIC (Novagen). An N-terminal thrombin cleavage site was present between the hexahistidine (6×His) tag and the NS1 coding region for the 6×His tag removal by thrombin. R38A and K41A mutations were introduced in both NS1s to prevent protein precipitation. The pET46-H6N6-NS1 Δ80-84 R38A/K41A construct was used to incorporate an E71G mutation by site-directed mutagenesis using the QuikChange mutagenesis kit (Agilent Biotechnologies). The primers used to generate the E71G mutation are as follows: forward primer, 5′-GGTGGTGCGGCCGCGATGGCACATCACCACCAC-3′, and reverse primer, 5′-GGTGGTGTCGACTCAAACTTCTGACTCAATTGTTCTC-3′. The QuikChange mutagenesis kit was further used to generate a G71E mutation in the pET46-VN/04 H5N1-NS1 construct. Primers used to generate this construct are as follows: forward primer, 5′-GCCTTATCAGACTCCTCCTCCAGAATCCGCT-3′, and reverse primer, 5′-AGCGGATTCTGGAGGAGGAGTCTGATAAGGC-3′. H5N1-NS1 and the mutant derivatives were also cloned into pFLAG-CMV2. The primers utilized to amplify the constructs prior to their insertion into pFLAG-CMV2 are as follows: forward primer, 5′-GGTGGTGCGGCCGCGATGGCACATCACCACCAC-3′, and reverse primer, 5′-GGTGGTGTCGACTTACCGTTTCTGATTTGGAGG-3′. To PCR amplify H6N6-NS1 and its mutant derivatives into pFLAG-CMV2, we utilized the same forward primer used in the case of H5N1-NS1 and reverse primer 5′-GGTGGTGTCGACTTACCGTTTCTGATTTGGAGG-3′. The CPSF30 construct provided by B. Carrillo was used to PCR amplify 60 to 120 residues of the F2F3 domain by using forward primer 5′-GACGACGACAAGATCATGGCACATCACCACCAC-3′ and reverse primer 5′-GAGGAGAAGCCCGGTTTAAATCTTGGACTCGGGGTC-3′. The PCR-amplified product was cloned into pET-46-Ek/LIC as well.

Protein expression and purification.

Escherichia coli BL21(DE3) (Novagen) cells containing the recombinant plasmid were grown at 37°C, and once the absorbance at 600 nm (A₆₀₀) reached 0.7, recombinant protein was expressed by inducing the cells with 500 μM IPTG (isopropyl β-d-thiogalactopyranoside) overnight at 25°C. Fifty milliliters of lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 8.0) was used to dissolve the bacterial pellet that was lysed by a microfluidizer (high-shear fluid processor microfluidizer LM20; ATS Scientific). Affinity chromatography with Ni-NTA (Ni-nitrilotriacetic acid) agarose beads (Qiagen) was used to bind the 6×His tag-NS1. The beads were washed with wash buffer (50 mM NaH₂PO₄, 300 mM NaCl, 30 mM imidazole, pH 8.0) to reduce the contaminant load and subsequently eluted with elution buffer (50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole, pH 8.0). Affinity-purified 6×His tag-NS1 was treated with thrombin while being dialyzed in lysis buffer overnight at 4°C. Thrombin-treated protein was rebound on Ni-NTA beads to remove the cleaved 6×His tag. NS1 was then passed through a Q-column (GE Healthcare) to reduce the contaminant load. Further purification of the protein was carried out by size exclusion chromatography using a Superdex 75 preparatory-grade column attached to an ÅKTA purifier (GE Biosciences) preequilibrated with gel filtration buffer (10 mM HEPES, 200 mM NaCl, pH 8.0). All of the constructs were expressed and purified by following the same purification protocol.

For F2F3 expression, the same protocol was followed; however, cells induced at 15°C with 750 μM IPTG overnight led to maximum expression. Affinity purification of the F2F3 protein to reduce the contaminant load was directly followed by size exclusion chromatography in the gel filtration buffer.

Immunoprecipitations.

293T cells were transfected with Flag-tagged plasmids and lysed 48 h posttransfection using Pierce immunoprecipitation (IP) lysis buffer. Protease inhibitor cocktail (Sigma-Aldrich) was also added to the lysis buffer. An anti-FLAG M2 affinity gel (Sigma-Aldrich) was used to perform immunoprecipitation by following the manufacturer’s protocol. An anti-FLAG M2 affinity gel was incubated with cell lysate at 4°C overnight. Beads were washed with lysis buffer following incubation. Sample loading buffer was added to the beads the following the wash and heated for 5 min at 95°C to elute the bound proteins. Samples were centrifuged and loaded into a 4-to-20% Tris gradient gel (Bio-Rad). Following the SDS-PAGE run, the gels were transferred to a nitrocellulose membrane and Western blotting was performed, followed by probing with appropriate antibodies.

ITC.

The Microcal Auto-iTC200 system (GE Life Sciences) was used to determine the binding affinities of H5N1-NS1(G71), H5N1-NS1(E71), H6N6-NS1(Δ80-84/E71), and H6N6-NS1(Δ80-84/G71) to the F2F3 domain of CPSF30 by isothermal titration calorimetry (ITC). Purified proteins were dialyzed in buffer (10 mM HEPES, 200 mM NaCl, pH 8.0) overnight before the ITC experiments were performed. Both protein and ligand were centrifuged (2 min at 2,000 × g) to remove bubbles and thus reduce large unexpected peaks in the titration. NS1 proteins (7 to 12 μM) were titrated with F2F3 (110 to 130 μM). Titration was carried out with 5-min interval injections of 1 μl for 30 injections at 25°C. A control experiment was performed for each titration by titrating F2F3 into the final buffer under the same conditions. Data from the first injection were discarded for each experiment. The data obtained were transformed into *.itc files and analyzed using Origin7 software.

Crystallization of H5N1-NS1(E71), H6N6-NS1(Δ80-84/G71), and H6N6-NS1(Δ80-84/E71).

H5N1-NS1(E71), H6N6-NS1(Δ80-84/G71), and H6N6-NS1(Δ80-84/E71) were crystallized by the hanging-drop vapor diffusion method using the Mosquito robot (TTP LabTech) at 20°C, and crystal plates were periodically monitored to visualize crystal formation using Rock Imager (Formulatrix). Crystals of H5N1-NS1(E71) were formed by mixing a 1:1 ratio of the protein (10 mg/ml) and the well solution containing 0.1 M sodium acetate trihydrate and 0.1 M Tris hydrochloride, pH 8.5. Crystals were grown for almost 1 month, harvested and cryoprotected in 0.1 M sodium acetate trihydrate, 0.1 M Tris hydrochloride (pH 8.5), and 22% glycerol, and flash-frozen in liquid nitrogen. H6N6-NS1(Δ80-84/G71) (12 mg/ml) was crystallized by mixing it at a 1:1 ratio with well solution containing 0.2 M sodium citrate, 0.1 M Tris hydrochloride (pH 8.5), and 30% (vol/vol) polyethylene glycol (PEG) 400. A 0.01 M concentration of TCEP [Tris(2-carboxyethyl)phosphine hydrochloride] (Hampton Additive Screen) was used as an additive. Crystals were formed at 20°C after almost 1 month. Crystals were harvested and placed in cryoprotectant solution (0.2 M sodium citrate, 0.1 M Tris hydrochloride [pH 8.5], 30% [vol/vol] PEG 400, 0.01 M TCEP, and 16% glycerol) and flash-frozen in liquid nitrogen. Crystals of H6N6-NS1(Δ80-84/E71) (9.5 mg/ml) were also produced by mixing a 1:1 ratio of protein and the well solution containing 0.2 M sodium citrate tribasic dehydrate, 0.1 M Tris hydrochloride (pH 7.0), and 15% (vol/vol) PEG 400. Crystals formed at 20°C after almost 2 months were harvested and placed in cryoprotectant solution (0.2 M sodium citrate tribasic dihydrate, 0.1 M Tris hydrochloride [pH 7.0], 15% [vol/vol] PEG 400, and 18% glycerol) and flash-frozen in liquid nitrogen.

Data collection and processing.

Data for H5N1-NS1(E71), H6N6-NS1(Δ80-84/G71), and H6N6-NS1(Δ80-84/E71) were collected at the Advanced Light Source beamline 8.2.1 using a Quantum 315R detector, at 3 Å , 3.2 Å, and 3.89 Å, respectively. Diffraction data for all crystals were processed with the iMOSFLM program (26) and SCALA from the CCP4 package (27). Space groups were identified by the QuickSym subprogram in iMOSFLM. For H5N1-NS1(E71), the molecular replacement (MR) solution was found in space group P6₅22, with one molecule in the asymmetric unit (AU). Similarly, the MR solutions for H6N6-NS1(Δ80-84/G71) and H6N6-NS1(Δ80-84/E71) were found in space group P6₅22, with one molecule in the AU. The PR8 H1N1-NS1 ED (PDB ID 2GX9) and the H5N1-NS1 RBD (PDB ID 3F5T) were used separately as search models for MR in the PHASER program, as implemented in the CCP4 6.0.2 software package (28). Iterative cycles of refinement and model building were performed using Rosetta (29, 30), PHENIX (31), and COOT (32). Statistical values for data collection and refinement are shown in Table 1. Figures for the structures were generated using PyMOL (The PyMOL Molecular Graphics System, v2.2.3).

smFRET analysis.

Unlabeled H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) proteins were expressed and purified as described above. Purified H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) were labeled using a 1:3 to 1:4 protein-to-dye (Alexa Fluor 488 and Alexa Fluor 594 maleimide; Invitrogen) molar ratio by incubating mixtures at room temperature for 30 min. Excess unconjugated dyes were removed using NAP-5 G-25 (GE Healthcare) columns. Fluorescently labeled proteins were further purified by analytical size exclusion chromatography using smFRET buffer (10 mM HEPES, 200 mM NaCl, pH 8.0). To reduce donor-only “zero peak” contributions in smFRET experiments, chromatography fractions with higher Alexa Fluor 594-to-Alexa Fluor 488 ratios were selected. Donor and acceptor fluorescence signals were recorded at ∼100 pM of Alexa Fluor 488/Alexa Fluor 594-labeled H5N1-NS1(E71) and H6N6-NS1(Δ80-84/E71) using a custom-built ISS Alba confocal laser microscopy system described previously (33). The binning time used for individual smFRET measurements was 500 μs. The leakage of donor emission into the acceptor channel (1%) and acceptor emission due to direct excitation (3%) were taken into account in the data analysis. A lower threshold of 42 counts, i.e., the sum of signals from the donor and acceptor channels, was used to separate background noise from single-molecule fluorescence signals. FRET efficiencies (E_FRET) were determined from the corrected donor (I_D) and acceptor (I_A) fluorescence intensities as

$$E_{\text{FRET}}=\frac{I_A}{I_A+I_D}$$

The value of gamma was approximated as unity based on previous experiments (23). FRET efficiency histograms were generated using VistaVision 4.2.099.0 (ISS, Inc.), and the distributions were fitted by Gaussian function using OriginPro 9.0. Interdye distances (r) were estimated from E_FRET from the following relation:

$$R=\frac{R_0}{{\left(\frac{1}{E_{\text{FRET}}}-1\right)}^6}$$

where R₀ is the Förster distance between the donor-acceptor dye pair.

Accession number(s).

The coordinates and structure factors for the three crystal structures determined in this study have been deposited in the PDB under accession codes 6O01 for H5N1-NS1(E71), 6NRL for H6N6-NS1(Δ80-84/G71), and 6OQE for H6N6-NS1(Δ80-84/E71).