NMR analysis of the dynamic exchange of the NS2B cofactor between open and closed conformations of the West Nile virus NS2B-NS3 protease

Abstract

Background: The two-component NS2B-NS3 proteases of West Nile and dengue viruses are essential for viral replication and established targets for drug development. In all crystal structures of the proteases to date, the NS2B cofactor is located far from the substrate binding site (open conformation) in the absence of inhibitor and lining the substrate binding site (closed conformation) in the presence of an inhibitor.

Methods: In this work, nuclear magnetic resonance (NMR) spectroscopy of isotope and spin-labeled samples of the West Nile virus protease was used to investigate the occurrence of equilibria between open and closed conformations in solution.

Findings: In solution, the closed form of the West Nile virus protease is the predominant conformation irrespective of the presence or absence of inhibitors. Nonetheless, dissociation of the C-terminal part of the NS2B cofactor from the NS3 protease (open conformation) occurs in both the presence and the absence of inhibitors. Low-molecular-weight inhibitors can shift the conformational exchange equilibria so that over 90% of the West Nile virus protease molecules assume the closed conformation. The West Nile virus protease differs from the dengue virus protease, where the open conformation is the predominant form in the absence of inhibitors.

Conclusion: Partial dissociation of NS2B from NS3 has implications for the way in which the NS3 protease can be positioned with respect to the host cell membrane when NS2B is membrane associated via N- and C-terminal segments present in the polyprotein. In the case of the West Nile virus protease, discovery of low-molecular-weight inhibitors that act by breaking the association of the NS2B cofactor with the NS3 protease is impeded by the natural affinity of the cofactor to the NS3 protease. The same strategy can be more successful in the case of the dengue virus NS2B-NS3 protease.

Figure 1. Crystal structures of the West Nile and dengue NS2B-NS3 proteases.

NS2B is shown in magenta, with the N- and C-termini labeled. The polypeptide chain of NS3 is shown in rainbow colors ranging from blue (N-terminus) to red (C-terminus). Full-length NS2B comprises 131 residues . All constructs used in crystal structure determinations include the NS2B cofactor segment and exclude the N-terminal 48 and C-terminal 35 residues of NS2B which contain hydrophobic membrane anchors. The constructs used in the present work closely resembled the constructs used for crystallography. In denoting residues 49–60 as the N-terminal segment (NTS) of NS2B and residues 75–96 as the C-terminal segment (CTS) of NS2B, we refer to the cofactor segment of NS2B only. (A) WNV NS2B-NS3pro in the presence of the inhibitor BPTI (PDB accession code 2IJO). The substrate binding site is indicated by the BPTI segment Pro13-Arg17 shown in black. The location of the active-site histidine is labeled with a blue star. (B) WNV NS2B-NS3pro in the absence of inhibitors (PDB code 2GGV). (C) Dengue virus NS2B-NS3pro in the absence of inhibitors (PDB code 2FOM). The present text refers to the protein fold in (A) as closed conformation, while the folds in (B) and (C) are referred to as open conformations.

Figure 2. Chemical structures of inhibitors used.

Throughout the text, the inhibitors in the top and bottom panels are referred to as compounds 1 and 2, respectively.

Figure 3. Conformational exchange of NS2B highlighted by inhibitor-induced spectral changes.

(A) Superimposition of ¹⁵N-HSQC spectra of a 50 µM solution of selectively ¹⁵N-Ile labeled wt NS2B-NS3pro in the presence (blue spectrum) and absence (red spectrum) of 0.3 mM inhibitor 1. The spectra were recorded at 25°C in a buffer containing 20 mM HEPES, pH 7.0, and 2 mM DTT, using a Bruker 800 MHz NMR spectrometer. The sequence-specific resonance assignments are indicated for the spectrum recorded in the presence of 1. The star identifies a peak that could not be attributed to any of the isoleucines in the protease. Its relatively narrow line shape suggests its origin from a low-molecular weight impurity. (B) Locations of the isoleucine residues in the crystal structure of the WNV NS2B-NS3 protease (PDB code 2IJO). The nitrogen atoms are drawn as balls, using blue and red colour to identify the residues with, respectively, little and large spectral changes in the ¹⁵N-HSQC spectra caused by 1. NS2B is shown in magenta and its N- and C-termini are identified. The isoleucine residues of NS2B are labeled in italics. The arrow identifies the site of Cys89 in NS2B-NS3pro^C. (C) Same as (B), except for the open conformation (PDB code 2GGV). Only the isoleucine residues of NS2B are labeled for improved visual presentation. (D) Same as (B), except that results obtained with uniformly ¹⁵N-labeled WNV NS2B-NS3pro are displayed by highlighting selected backbone nitrogens. Red: ¹⁵N-HSQC cross-peaks were assigned in the presence of inhibitor but seem to be missing in the absence of inhibitor. Yellow: ¹⁵N-HSQC cross-peaks shifting more than 0.05 ppm in the ¹H dimension between the spectra recorded with and without inhibitor 1. Blue: ¹⁵N-HSQC cross-peaks shifting less than 0.02 ppm in the ¹H dimension between spectra with and without inhibitor.

Figure 4. Protein flexibility probed by transverse ¹⁵N relaxation rates R ₂.

The data were measured using a 0.9 mM solution of ¹⁵N/¹³C-labeled NS2B-NS3pro in the presence (black triangles) and absence (red squares) of 3 mM 2. Data points are plotted versus residue number and connected by lines for improved visual appearance. For NS3, data are shown only for residues for which R ₂ data could be measured in both states. All points are shown for NS2B in the presence of the inhibitor 2 as a guide for the mobility of the NS2B CTS. Error bars show the error reported by the fitting routine in Sparky .

Figure 5. Attenuation of cross-peak intensities by the MTSL spin-label and comparison with proximity to the amide protons in crystal structures of WNV NS2B-NS3pro.

(A) Relative peak intensities observed in ¹⁵N-HSQC spectra of WNV NS2B-NS3pro^C with MTSL versus those of WNV NS2B-NS3pro^C without spin-label. In order to adjust for uncertainties in protein concentration, the intensity ratios were normalized by setting the largest I_para/I_dia ratio to 1. The black and red data points were measured in the presence and absence of inhibitor 2, respectively. (B) Distance between the N89C^α (as a proxy for the position of the spin-label) and the amide nitrogens in the closed conformation (PDB code 2IJO). (C) Same as (B), except for the open conformation (PDB code 2GGV). (D) Plot of the data in (A), as obtained without inhibitor, on the closed conformation of the WNV NS2B-NS3 protease (PDB code 2IJO). NS2B is shown in magenta. The position of Cys89 in WNV NS2B-NS3pro^C carrying the spin-label is identified by an arrow. Amide protons are highlighted with spheres of different colour depending on the I_para/I_dia ratio in (A): 0–0.2 (red), 0.2–0.4 (pink), 0.4–0.6 (yellow), 0.6–0.8 (cyan), 0.8–1.0 (blue). (E) Same as (D), but for the data in the presence of inhibitor 2. (F) Stereoview of the closed conformation in the representation of (D) and (E). Red spheres identify amide protons of residues for which small concentration dependence in the absence of inhibitor indicates intramolecular PRE effects. Pink spheres identify the location of amide protons for which PRE were significantly concentration dependent, indicating an intermolecular PRE mechanism. NS2B residues are marked in italics.