Abstract
Flavivirus is a genus of emerging and re-emerging arboviruses which include many significant human pathogens. Non-structural protein 3 (NS3), a multifunctional protein with N-terminal protease and C-terminal helicase, is essential in viral replication. The NS3 protease together with NS2B cofactor is an attractive antiviral target. A construct with an artificial glycine linker connecting the NS2B cofactor and NS3 protease has been used for structural, biochemical and drug-screening studies. The effect of this linker on dynamics and enzymatic activity of the protease was studied by several biochemical and NMR methods but the findings remained inconclusive. Here, we designed constructs of NS2B cofactor joined to full length DENV4 NS3 in three different manners, namely bNS2B47NS3 (bivalent), eNS2B47NS3(enzymatically cleavable) and gNS2B47NS3 (glycine-rich G4SG4 linker). We report the first crystal structures of linked and unlinked full-length NS2B-NS3 enzyme in its free state and also in complex with Bovine Pancreatic Trypsin Inhibitor (BPTI). These structures demonstrate that the NS2B-NS3 protease mainly adopts a closed conformation. BPTI binding is not essential to but favors the closed conformation by interacting with both NS2B and NS3. The artificial linker between NS2B and NS3 tends to induce the open conformation and interfere with the protease activity. This negative impact on the enzyme structure and function is restricted to the protease domain as the ATPase activities of these constructs are not affected.
Introduction
Flaviviruses include many significant human pathogens such as dengue virus (DENV), West Nile virus (WNV) and recently re-emerging Zika virus (ZIKV). Recent outbreak of ZIKV infections in America has caused global health concern since the infections were linked to neuropathic Guillain-Barré syndrome in adults and microcephaly in infants [1–3]. DENV, has been emerging in the past decade and is a global healthcare burden. The emergence of pandemic DENV and epidemic ZIKV infections in the past years due to globalisation and urbanisation call for countermeasures such as the development of potent antivirals against these infections.
Flaviviruses are enveloped viruses which contain a single-stranded positive-sense RNA genome of about 11 kb, with 3’ and 5’ untranslated regions (UTR) [4]. The genome encodes a poly-protein precursor which is cleaved into three structural proteins and seven non-structural proteins by host and viral proteases [4, 5]. Non-structural protein 3 (NS3) plays essential roles in viral replication and polyprotein processing and is an attractive anti-viral target [6]. The N-terminal domain of NS3 (residues 1-168) is a serine protease responsible for cleavage of polyprotein precursors into mature functional proteins [7–12]. The C-terminal domain of NS3 is an NTPase/Helicase involved in viral replication and virion assembly [13–15]. Recently, several drugs targeting the Hepatitis C virus (HCV) NS3 protease have been approved by the U.S. Food & Drug Administration (FDA) [16]. However, no NS3 inhibitor for DENV has advanced to clinical trials [7, 17, 18].
The N-terminal protease contains a catalytic triad formed by residues Ser-135, His-51, and Asp-75 [19] and requires NS2B protein as cofactor for endoplasmic reticulum (ER) membrane anchorage, proper folding, and protease activity [9, 10, 19]. Soluble and catalytically active recombinant NS2B47-G4SG4-NS3 protease (heraftercalled gNS2B47NS3 Pro) was designed by tethering central NS2B cofactor to NS3 protease by a flexible artificial glycine linker [20]. Structural studies of the flavivirus NS3 protease have been done utilizing this construct design except for the recent ZIKV protease studies [12, 21–24]. These studies using conventional glycine-linked constructs demonstrated that the NS2B N-terminus contributes to the folding of protease by inserting a β-strand to the N-terminal β-barrel of protease [11, 12, 22, 25–27]. The C-terminus of NS2B is flexible and is only observed in crystal structures where the protease is bound to an inhibitor or a substrate, suggesting that it is acting as a flap closing upon substrate binding [11, 12, 22, 26]. The free protease structures with flexible NS2B C-terminus are said to adopt an “open” conformation, while the protease-inhibitor structures with NS2B contributing to substrate binding site show a “closed” conformation. Although all crystal structures of free gNS2B47NS3 protease are reported to adopt the open conformation, NMR studies have shown that in solution, gNS2B47NS3 protease interonverts between the open and closed conformations even in the absence of an inhibitor [28–32]. These studies also showed that when NS2B and NS3 are separate polypeptides, the NS2B-NS3 protease complex is mainly in the closed conformation without the substrate [29, 32].
Structural studies on ZIKV NS2B47NS3 protease shed new light on this unsolved issue. Zhang et al have reported a crystal structure of unlinked ZIKV protease (bZiPro) in closed conformation without an inhibitor [33]. The biochemical studies of ZIKV NS2B-NS3 protease constructs with glycine linker (gZiPro), NS2B-NS3 enzymatic cleavage site linker (eZiPro) and bivalent unlinked NS2B NS3 protease (bZiPro) have revealed that the flexible glycine linker interferes with the protease activities resulting in lower kcat [24]. Shannon et al posited that reduced product release could be the possible mechanism behind the lower activity of glycine-linked constructs based on studies carried with the DENV2 bivalent co- expressed NS2B NS3 protease (bNS2B47NS3 Pro) and glycine-linked NS2B NS3 protease (gNS2B47NS3 Pro) [34]. Optimising construct designs to obtain biologically relevant crystal structures is an important factor for structure-based drug discovery. The crystal structures of separate domains of DENV NS3 have been reported, as well as full-length NS3 together with an 18-residues cofactor region of NS2B (NS2B18NS3)[12, 25, 35–37]. Full-length gNS2B47NS3 from Murine valley encephalitis virus (MVEV) has also been reported to adopt an open conformation in the absence of inhibitor [38]. Although the protease and NTPase/helicase domains of full length DENV4 NS2B18NS3 showed similar folds to those in MVEV NS2B47-NS3, domain arrangements between helicase and protease were found to be different, consistent with the flexibility of the linker region between the two functional domains. Here we designed bivalent, enzymatic cleavage site linked and conventional flexible glycine linked NS2B cofactor with full length DENV4 NS3 constructs namely bNS2B47NS3, eNS2B47NS3, and gNS2B47NS3 similar to those in our previous studies on ZIKV protease [24]. We report three crystal structures of full length DENV4 NS2B47NS3 constructs, eNS2B47NS3 and gNS2B47NS3 in free form and two in complex with Bovine Pancreatic Trypsin Inhibitor (BPTI). The structural analysis suggests that the NS2B-NS3 protease has a preformed active site with NS2B cofactor wrapped around NS3 participating in substrate binding. The biochemical studies of the ATPase activities of full length NS3 demonstrate uncoupled enzymatic activities for the full length NS3 protein.
Results
Design and preparation of unlinked and linked full length NS2B47NS3 proteins
To overcome the problem of poor expression of wild type gNS2B47NS3 proteins, we mutated the protease at either (1) S135 to alanine (S135A) or (2) hydrophobic residues on the surface, L30 and F31 to serine (L30S-F31S) [35]. The protease activity of gNS2B47NS3 Pro and of gNS2B47NS3 Pro (L30S-F31S) are comparable indicating that L30S-F31S mutation does not interfere with the proteolytic activity of NS3 unlike the S135A mutation which completely abolished protease activity (S1 Fig). We replaced the glycine linker of gNS2B47NS3 with the NS2B C-terminal penta-peptide (VKTQR) resulting in endogenous enzyme cleavable NS2B-NS3 constructs - eNS2B47NS3 (S135A) and eNS2B47NS3 (L30S-F31S) (Fig 1A). We name this eNS2B47NS3 L30S-F31S construct as unlinked eNS2B47NS3 since the NS2B/NS3 cleavage site was fully cleaved by the protease resulting in heterodimers of NS2B cofactor peptide-NS3. The bivalent construct bNS2B47NS3 was designed by co-expressing NS2B cofactor and NS3 sequences which fold as a heterodimer, similar to bZiPro [24, 33].SDS- PAGE analysis of proteins showed that eNS2B47NS3 L30S-F31S undergoes complete proteolysis resulting in unlinked full length NS3 similar to ZIKV eZiPro [24] (Fig 1B). The constructs and mutations are listed in Figure 1C. All full length NS2B47NS3 proteins were soluble and monomeric in solution as shown by the size exclusion chromatography profiles (S2 Fig A). Internal proteolysis at NS3 was observed during and after purification for all constructs with active protease, and gNS2B47NS3 degraded slightly more slowly than the bNS2B47NS3 and eNS2B47NS3 (S2 Fig B).
Structures of full length NS2B47NS3
The NS2B47NS3 proteins were crystallised using very similar conditions (S1 Table). The crystalswere assigned into three groups that are related to the protein conformation: (1) Open conformation in which the C terminal region of the cofactor NS2B was disordered (Fig 2A); (2) -Closed conformation in which the C-terminus of NS2B loosely forms a beta hairpin (Fig 2B, 2D);(3) BPTI-bound closed conformation with similar but less dynamic NS2B C- terminus beta hairpin (Fig 2C, 2E). The correlation between the overall conformations and the unit cell dimensions is apparent from Table S1. The open conformation was only captured in gNS2B47NS3 (Fig 2A), which closely resembles the structure of DENV4 gNS2B18NS3 conformation I (PDB id: 2VBC) [35], while the remaining free enzyme structures are in closed conformation (Fig 2B, 2D). The gNS2B47NS3-BPTI structure and eNS2B47NS3-BPTI structures adopt the same conformation (Fig 1E, G). On the other hand, the bNS2B47NS3 protein crystals diffracted poorly to about 4 Å and as a result, we failed to find a convincing structure solution by molecular replacement. The unit cell dimensions of bNS2B47NS3 crystals were similar to the closed conformations of gNS2B47NS3 and of unlinked eNS2B47NS3, suggesting that the bNS2B47NS3 was also in a closed conformation (S1 Table).The data collection and refinement statistics are summarized in Table 1.
Overall, the NS2B47NS3 structures adopt an extended shape where the N terminal protease and the C terminal helicase domains are loosely connected through a flexible interdomain linker similar to DENV4 gNS2B18NS3 structures (PDB id: 2VBC, 2WHX, 2WZQ) and the MVEV gNS2B47NS3 (PDB id: 2WV9) (Fig 2, S3 Fig) [35, 37, 38]. In all NS2B47NS3 crystals, major crystal contacts are formed between the neighbouring helicase domains, which allow the protease domain to adopt multiple conformations (S4 Fig). From the open to the closed conformation, the protease domain is translated by about 0.7 Å. From the closed free enzyme conformation to the closed protease-BPTI conformation, the protease is rotated by an angle of 52.9°. Although of low resolution, the solvent shells around the protease domain are discernible confirming that the structure solution is correct (S5 Fig). Compared to MVEV NS2B47NS3 structure, when their helicase domains are superimposed, the protease domain are rotated by an angle of 177.1° and a translation of 17 Å (S3 Fig) with respect to the superimposed helicase domain.
We captured three free enzyme full length NS2B47NS3 structures. For the gNS2B47NS3 construct, the NS2B cofactor is captured in both open and closed conformations (Fig 3A, Fig 3B) while for unlinked eNS2B47NS3 constructs, the NS2B co-factor is captured only in closed conformation (Fig 3D, Fig 3E). This indicates that the presence of a flexible glycine linker increases the population adopting an open conformation. The RMSDs of protease domain between gNS2B47NS3, linked eNS2B47NS3, unlinked eNS2B47NS3 are less than 0.7 Å for 160 Cα atoms. In the free enzyme structures with closed NS2B conformation, the last 10-12 amino acids from the cofactor NS2B, the linker, and the first 20 amino acids of NS3 are flexible. In eNS2B47NS3 structures, the NS2B N terminal 8 residues and 3 residues are flexible for linked and unlinked structures respectively. The electron density of NS2B C- terminus for free enzyme closed NS2B conformation structures are relatively weak indicating that the C-terminus of NS2B is dynamic when the active site is not occupied (Fig. 3).
Interestingly in the unlinked eNS2B47NS3 structure, the NS2B/NS3 cleavage peptide (VKTQR) is not occupying the substrate binding site, unlike in the similar ZIKV protease structure, (eZiPro) (PDB accession code 5GJ4) [24]. We also report the crystal structures of gNS2B47NS3 and of unlinked eNS2B47NS3 in complex with BPTI. The RMSD between the two NS3-BPTI structures are 0.44 Å for 618 Cα atoms indicating that mode of binding of BPTI is conserved in both gNS2B47NS3 and eNS2B47NS3. The protease domain rotates by 52.9° in the NS3-BPTI full length structure to accommodate the BPTI in the crystal. The detailed interactions between the BPTI and NS2B cofactor and NS3 protease in both structures are conserved (S6 Fig). The three NS2B47NS3 free enzyme structures reveal a more dynamic NS2B cofactor and NS3 protease compared to the two NS2B47NS3-BPTI structures indicating that substrate binding stabilises the protease (Fig 3, S7 Fig). The ATPase/helicase domains of both gNS2B47NS3 and eNS2B47NS3 are identical with RMSD of less than 0.5 Å. The overall conformation of helicase is similar to the helicase structures with no NTP or RNA bound except for the residues 461-471 as mentioned before by [38] and residues 243-253. This surface loop is in close proximity to NS2B β hairpin and to NS3 residues 66PSWAD71, and changes conformation when the BPTI binds to the protease domain. In eNS2B47NS3-BPTI structure, movement of the protease domain results in the P- loop moving away from these residues.
The artificial glycine linker interferes with the protease activity of NS3
Latest studies using biochemical and NMR of flaviviral protease have shown that the flexible glycine linker affects the enzymatic and binding activities of the protease [23, 24]. To determine the effect of artificial glycine linker on the enzymatic activities of full length DENV NS3, we measured the protease activity of eNS2B47NS3, gNS2B47NS3, and bNS2B47NS3 using Benzoyl-Nle-Lys-Arg-Arg-Aminomethylcoumarin (Bz-NKRR-AMC) fluorescent substrate [39]. Our enzymatic assays showed that while the glycine linker does not affect the substrate apparent affinity (Km), its presence slows down the rate of catalysis (kcat) (Fig 4A). It is possible that the glycine linker introduces steric hindrance on the NS2B- NS3 conformational transitions compared to unlinked construct. Although eNS2B47NS3 has a slightly higher Km and lower kcat compared to bNS2B47NS3, presence of NS2B/3 cleavage site does not have the similar inhibitory effect on the protease enzymatic activity as reported for eZiPro [24]. This could be due to the sub-optimal cleavage site found at NS2B/NS3 in all DENV serotypes where the P2 residue is glutamine instead of a strongly basic lysine or arginine found in other flaviviruses. The inhibition activity assays with BPTI and with small peptidic inhibitor, Benzoyl-Lys-Arg-Arg-H, shows that the half maximal inhibitory value, IC50, was lowest for bivalent bNS2B47NS3 (Fig 4B,C) and highest for gNS2B47NS3 indicating a slightly tighter association with the former. In addition, the thermal shift assay of these constructs shows Tm of gNS2B47NS3 is 2°C lower than that of bNS2B47NS3 and of eNS2B47NS3, further suggesting that the presence of artificial flexible linker between NS2B cofactor and NS3 may interfere with the protein stability (S2 Fig).
The kinetics of ATP hydrolysis by full length NS3 constructs are similar
Next, to determine the effect of linkers on the NTPase activities of NS3, we carried out the NADH coupled ATPase assay for g-, e-, bNS2B47NS3 full length constructs. These constructs show similar ATPase activity demonstrating that the different linkers between NS2B and NS3 protease do not interfere with NTPase activity of the helicase. The helicase activity of NS3 requires the energy provided by ATP hydrolysis. The NTP binding site of helicase is situated right on top of the protease domain while the RNA binding groove of the helicase domain is spatially separated from the protease domain. Therefore, these different linker constructs are unlikely to have an effect on the helicase activity if the ATPase activity is unaffected by the presence of different linkers between NS2B and NS3. To test if the binding of substrate to the protease domain affect the NTPase activity of helicase domain, we measured the ATPase activity of bNS2B47NS3 in the presence and absence of BPTI. The rate of ATP hydrolysis remains unchanged when BPTI is bound to the protease domain, demonstrating that the substrate binding on protease domain does not affect the ATPase activity of helicase domain (Fig 5B). The ATPase activity of DENV4 NS3 helicase was measured in the presence and absence of BPTI as the control (Fig 5B). Both kcat and Km of ATP hydrolysis of bNS2B47NS3 is slightly slower compared to those of helicase alone, and hence the catalytic efficiencies of both enzymes are similar (Fig 5B).
Discussion
Due to absence of NS2B/NS3 crystal structures in closed conformation without substrate or inhibitor, NS2B was proposed to convert from open and closed conformations upon substrate binding [11, 12, 25–27]. Early NMR studies of glycine linked DENV and WNV proteases showed crowded cross peaks due to conformational exchanges [28, 30]. The use of unlinked constructs in the followed-up NMR studies has improved the spectral quality and backbone assignment [29, 32]. The unlinked DENV protease constructs are obtained by 1) replacement of glycine linker with NS2B/NS3 cleavage site (EVKKQR) similar to eNS2B47NS3 and 2) by co-expression of NS2B cofactor peptide and NS3 protease similar to bNS2B47NS3 [29, 32]. The NMR studies of these unlinked DENV proteases confirmed that the NS2B cofactor is predominantly in a closed conformation. Likewise, NMR studies for similar ZIKV protease, eZiPro, bZiPro, and gZiPro, also showed that the unlinked protease is in a closed conformation [23, 24, 40]. The presence of glycine linker between NS2B and NS3 shifts the population towards open NS2B conformation, leading to crowded peaks in NMR spectra, whereas for unlinked NS2B-NS3 protease, well-resolved spectra are obtained due to the dominant closed NS2B conformation [29, 32, 40]. Here, we report a series of crystal structures of DENV4 NS2B47NS3 protease-ATPase/helicase which were designed in different formats and captured as free enzyme and inhibitor-bound complexes. These structures for the first time clearly confirm that both gNS2B47NS3 and unlinked eNS2B47NS3 could adopt the closed NS2B conformation in the absence of any substrate or inhibitor. These results therefore demonstrate that NS2B47NS3 protease has a preformed ligand binding site which becomes further stabilized upon substrate binding. For unlinked eNS2B47NS3, the NS2B/NS3 cleavage site pentapeptide (VKTQR) is not found at the active site, in contrast to the otherwise comparable protease structure from ZIKV [24]. All the structures reported here are crystallised under similar crystallization conditionsand the major crystal contacts are formed by the helicase domain (S6 Fig). This implies that these constructs could be further engineered to study the structural properties of NS2B-NS3 protease. The NS2B/3 cleavage pentapeptide of eNS2B47NS3 could be replaced by othercleavage sites present in the viral polyprotein. Determination of the crystal structures of the above mentioned constructs could be useful in understanding how different polyprotein cleavage sites bind to the NS2B-NS3 protease. Moreover, the binding loop of BPTI could be mutated as reported by Lin et al and subsequently co-crystallised with eNS2B47NS3 as a scaffold to understand the prime site interactions between the inhibitor and the protease [41].
The protease activity assays of different constructs show that bNS2B47NS3 and unlinked eNS2B47NS3 have comparable kcat while gNS2B47NS3 displays the lowest kcat (Fig 4A). The ATPase activity of these constructs are similar. This indicates that the lower kcat in the protease activity observed for gNS2B47NS3 is unlikely due to other factors, such as small differences in enzyme concentrations or in protein stability (Figure 5A). The flexible glycine linker might introduce steric hindrance on NS2B dynamics and therefore lower the kcat. Both BPTI and peptidomimetic inhibitor, Bz-KRR-H, inhibit the protease activity of all three constructs with similar range of affinity (Fig 4B,C) indicating that the flexible linker is not interfering with inhibitor or substrate binding. Hence, it is plausible that the dynamics of NS2B and NS3 are involved at the post-catalytic/product-release stage rather than simply during substrate binding. This is in accordance with the single molecule enzymatic studies performed by Shannon et al, where the kcat of the enzyme was affected rather than Km when the NS2B-NS3 interactions were disrupted [34]. In agreement with the crystal structure, the enzymatic activities of eNS2B47NS3 are similar to bNS2B47NS3 again in contrast to that of eZiPro and bZiPro [24]. These results suggest that DENV NS2B NS3 cleavage site is released from the active site upon cleavage, whereas for ZIKV, it remains bound at the active site. It is possible that different flavivirus are employing the different polyprotein cleavage site and specificity to regulate the protease activity of NS3 in vivo. From our structures, we propose that NS2B/NS3 protease mainly stays in closed conformation regardless of the presence of a substrate. During polyprotein processing, NS2B is anchored to ER membrane by N and C-terminal hydrophobic regions [9, 42]. The complete dissociation of NS2B C- terminus from NS3 protease would not be favourable spatially due to the NS2B membrane anchorage, whereas stable tight association of whole NS2B cofactor to NS3 will place the active site of NS3 close to the membrane, shielding it from substrate binding or interfering with substrate release Therefore, a rather loosely associated NS2B-cofactor appears to be the optimal conformation for NS2B-NS3 in vivo.
In conclusion, we provide crystallographic evidence that the NS2B cofactor loosely assumes closed conformation around NS3 protease in the full length NS3 in the absence of substrate. In contrast to the unlinked ZIKV protease, eZiPro, the substrate pocket of eNS2B47NS3 is not occupied and therefore may be useful for co-crystallisation with inhibitors for antiviral drug discovery. Due to slightly better protease activities, bNS2B47NS3 and eNS2B47NS3 appear to be better suited for more sensitive high-throughput screening of potential drugs.
Materials and methods
Plasmid preparation
The bacterial expression plasmid containing wild type NS3 linked to cofactor NS2B residues 49—95 was generated by site directed mutagenesis method by inserting NS2B 68-96 to the gNS2B18NS3 construct from Luo et al [35]. The eNS2B47NS3 construct was generated by replacing the glycine linker with residues 126-130 of NS2B C-terminus which is the enzymatic cleavage site of NS2B/NS3. The eNS2B47NS3 L30S F31S and gNS2B47NS3 L30S F31S are mutated from eNS2B47NS3 WT and gNS2B47NS3 WT by site directed mutagenesis. The bivalent full length construct (bNS2B47NS3) was synthesized by biobasic.
Expression and purification
The plasmids containing bNS2B47NS3, gNS2B47NS3, eNS2B47NS3 or mutants were transformed into Escherichia coli BL21(T1R). The transformants were grown in Luria Broth (LB) medium supplemented with suitable antibiotics (ampicillin (100 mg/L) or kanamycin (50 mg/L) and chloramphenicol (37 mg/L)), 40-50 mM Potassium Phosphate buffer pH 7.4 and 2.5% glycerol at 37°C until OD600 of 0.8 was reached. The culture was cooled to 18°C, subsequently induced with 1mM Isopropyl β-D-1-thiogalactopyranoside, and the proteins were overexpressed overnight at 18°C shaking at 200 rpm. Cells were harvested after 15 hours by centrifugation at 5000 rpm for 20 minutes at 4°C. Cells were resuspended in lysis buffer (25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.5, 500 mM Sodium chloride (NaCl), 5 mM β-Mercaptoethanol (β-ME), 5% glycerol, 10 mM imidazole). Cells were lysed by passing though NIRO SOAVI PANDA HIGH PRESSURE HOMOGENIZER at pressure 700-900 bars. The soluble fraction was separated by centrifugation of the lysate at 40000 RPM for 40 minutes. The soluble proteins were purified by metal affinity chromatography using Ni-NTA beads (Thermofisher). The N terminal Histidine-tag was cleaved by Tobacco Etch Virus (TEV) protease while the eluted fraction was dialyzed against Size exclusion chromatography (SEC) buffer (25mM HEPES, pH 7.5, 150 mM NaCl, 2 mM DTT, 5% glycerol) overnight at 4°C. The His-tag cleaved proteins were further purified by running through HiTrap Heparin HP 5 ml column (GE Healthcare) and were finally polished with size exclusion chromatography using HiLoad 16/600 Superdex 200 (GE Healthcare).
Crystallization, data collection and refinement
Crystals were grown by mixing 1 µL of proteins at a concentration of 8.5 mg/ml with 1 µL of precipitant by hanging drop vapour diffusion method (S1 Table). Cluster of thin plate crystals grew after 2 days of incubation at 20 °C. Crystals are separated into single plates, transferred to cryoprotected reservoir solution with 20% glycerol and cooled down to 100 K in liquid nitrogen before mounting.
Diffraction intensities were recorded on PILATUS 2M-F detector at PXIII beamline at the Swiss Light Source, Paul Scherrer Institut, Villigen, Switzerland and on ADSC Quantum 210r Detector at MX1 beamline at Australian Synchrotron. Diffraction intensities were integrated using iMOSFLM or XDS [43–45]. Scaling and merging of the intensities were done using POINTLESS and AIMLESS from CCP4 suite [46–49].Data collection statistics are summarized in (Table 1). For gNS2B47NS3 and eNS2B47NS3 unlinked datasets, the multiplicity was higher due to the smaller oscillation of the Φ.
The solution for gNS2B47NS3 with BPTI was solved using PHASER MR (CCP4 suite) using 2VBC as search model [37]. The solutions for full length NS3 (gNS2B47NS3 and eNS2B47NS3) were solved by using PHASER MR (CCP4 suite) using gNS2B47NS3 free enzyme structure as search model. The dataset for unlinked eNS2B47NS3 has ice rings and therefore the diffractions spots at the resolution shells around 3.4 Å were removed to reduce the noise. This has resulted in lowered completeness of the overall dataset. The structure solutions were subject to rounds of refinement using Phenix.refine program and manual refinement using WinCoot[50–54]. Rotational and translational movements of domains were carried out using DynDom (CCP4 suites) and Superpose (CCP4 suites)[55, 56]. Figs were generated using Pymol and electron density maps were generated using FFT (CCP4 suites)[57, 58]
Protease activity assay
The protease activity assays were carried out using 7-amino-4-methylcoumarin (AMC) fluorophore, Benzyonyl-Nle-Lys-Arg-Arg-AMC (Peptide Institute, Japan) modified from [39]. The Bz-NKRR-AMC substrate with starting concentration of 300 µM was serially diluted in assay buffer (20 mM Tris hydrochloric acid, pH 8.5, 10% glycerol, 0.01% Triton X-100, 2 mM DTT) and added to Corning® 96 Well black plates with 3 nM protein in same buffer. Assays were carried out as duplicates or triplicates at 37°C. The rate of AMC released was monitored at Synergy™ HTX Multi-Mode Microplate Reader at excitation wavelength 380 nm and emission wavelength 460 nm over 5-10 minutes at 1 minute intervals. To determine the amount of AMC released, standard AMC curve was plotted with over different concentrations of AMC (data not shown). Initial velocities were calculated using linear regression function using GraphPad Prism version 5.0 for Windows. The relative fluorescence units (RFU) were converted to amount of AMC using the standard curve. Data were analysed and plotted using Michalis-Menten equation with GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, California, USA).
Protease inhibition assay
The protease inhibition assays were carried out using the same substrate used in enzymatic assay at 30 µM concentrations. The inhibitors of different concentrations were added to the wells with 3 nM of proteins and were incubated for 30 minutes at room temperature. The reaction was initiated by addition of 30 µM substrate and initial velocities were measured at 1 minute intervals at 37°C for 10 minutes. Data were analysed using function Log inhibitor vs normalized response function in GraphPad Prism.
ATPase assay
ATPase activity assay was carried out based on Kiianitsa et al[59]. 50 nM of enzymes were incubated in assay buffer (25 mM MOPS pH 7.4, 150 mM potassium chloride, 2 mM DTT, 0.01% Triton X-100) with 50 µM of BPTI for an hour in Corning® 96 Well clear plates. NADH mixture (NADH 1mM, Phosphoenol pyruvate 2.5mM Pyruvate Kinase 500 U/ml and lactic dehydrogenase 100 U/ml in ATPase assay buffer) was added to reaction and incubated for 30 minutes more. Reaction was started by addition of various ATP concentrations. Depletion of NADH was measured by change in absorbance at 340 nm and was plotted against time using Cytation 3 Mulitmode plate reader (BioTek). After determining the path length, molar extinction coefficient for the given path length (Kpath) was calculated. Initial velocities were calculated using linear regression function using GraphPad Software version 5.0 for Windows. Data were plotted using Michaelis-Menten equation in GraphPad Prism.
Thermal shift assays
The Thermofluor assay was carried out as described previously [60]. The samples contained 10 µM protein and 5x SYPRO Orange dye in buffer containing 20 mM HEPES pH 7.5, 150 mM NaCl, 2 mM DTT and 5% glycerol. The samples were subject to temperature increments of 1°C from 20°C to 95°C over 20 minutes using real-time PCR machine Bio-Rad CFX96. The fluorescence intensities were recorded and analysed using GraphPad Prism. The melting curves were generated using Boltzmann-sigmoidal function.
Supporting information
Acknowledgment
We thank scientists from Australian Light Source MX beam-line and Swiss Light Source PX beam-line for their help with diffraction data collection. This work was supported by (1) the start-up grant to DL lab from Lee Kong Chian School of Medicine, Nanyang Technological University, (2) Ministry of Education grant MOE2016-T2-2-097 to DL lab, (3) National Medical Research Council grant CBRG14May051 to JL, (4) National Medical Research Council grant NMRC/CBRG/0103/2016 to SV lab, (5) National Research Foundation grant NRF2016NRF-CRP001-063. Ms. Wint Wint Phoo is supported by Nanyang Research Scholarship, Nanyang Technological University.