Biochimica et Biophysica Acta 1780 (2008) 989–994

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r.com/ locate /bbagen

A comparative biochemical analysis of the NS2B(H)–NS3pro protease complexfrom four dengue virus serotypes

Tawin Iempridee, Ratchanu Thongphung, Chanan Angsuthanasombat, Gerd Katzenmeier ⁎Laboratory of Molecular Virology, Institute of Molecular Biology and Genetics, Mahidol University, Salaya Campus, Phutthamonthon 4 Road, Nakornpathom 73170, Thailand

Abbreviations: acmc, 7-amino-3-carbamoylmethyl-zoyl; BPTI, bovine pancreatic trypsin inhibitor; DENhaemorrhagic fever; DSS, dengue shock syndrome; Gmethylcoumaryl-7-amide; n, norleucyl; NS, non-strustructural viral proteins 2B and 3, respectively; NS3prNS3 protein; NS2B(H), the central hydrophilic activation3-14, N-tetradecyl-N,N-dimethyl-3-ammonio-1-propanby-overlap extension polymerase chain reaction⁎ Corresponding author. Tel.: +662 800 3624 1211; fa

E-mail address: frkgz@mahidol.ac.th (G. Katzenmeie

0304-4165/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.bbagen.2008.03.018

A B S T R A C T

A R T I C L E I N F O

Article history:

The two-component protea Received 25 January 2008Received in revised form 17 March 2008Accepted 28 March 2008Available online 7 April 2008

Keywords:Dengue virusNS3 serine proteaseSerotypeRefoldingAssayActive-site titration

se NS2B–NS3 of dengue virus mediates proteolytic processing of the polyproteinprecursor and therefore represents a target for the development of antiviral drugs. The amino acid sequencesof the NS3 serine protease and the NS2B cofactor exhibit relatively low degrees of conservation among the 4serotypes thus suggesting that differences in enzyme activity exist which could modulate their susceptibilityto future protease inhibitors. In this study we have addressed the question of functional similarity among theNS2B(H)–NS3pro proteases from 4 dengue virus serotypes by employing a uniform approach to clone, purifyand assay proteolytic activity of these enzymes. Significant differences were observed for patterns of proteinformation and expression levels in the E. coli host. Renaturation of the NS2B(H)–NS3pro precursors fromdengue virus serotypes 2, 3 and 4 mediated by artificial chaperone-assisted refolding yielded enzymaticallyactive proteases, whereas the enzyme from serotype 1 was obtained as soluble protein. Kinetic experimentsusing the GRR-amc substrate revealed comparable Km values while kcat values as obtained by active-sitetitration experiments displayed minor variations. Denaturation experiments demonstrated significantdifferences in half-life of the NS3 proteases from serotypes 2, 3 and 4 at 50 °C, whereas pH optima for all 4enzymes were comparable.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

The genus Flavivirus in the Flaviviridae family represents a largenumber of viral pathogens causing severe morbidity and mortality inhumans and animals [1]. In endemic areas, dengue virus istransmitted by the main mosquito vector, Stegomyia aegypti (formerlyAedes), and circulates as a complex of 4 serotypes, DEN1 to 41, whichall are causative agents of dengue fever, dengue haemorrhagic fever(DHF) and dengue shock syndrome (DSS). The high burden on publichealth care and the limited therapy options for dengue diseases havenow stimulated efforts to characterize viral drug targets such as theNS3 serine proteases and to design and evaluate antiviral inhibitorswhich are equally effective against the 4 dengue virus serotypes andalso related members of the Flaviviridae (for review see [2,3] andreferences herein).

4-methyl coumarin; Bz, ben-, dengue virus; DHF, dengueRR-amc, tBoc-Gly-Arg-Arg-4-ctural; NS2B and NS3, non-o, the protease domain of thedomain of the NS2B protein; SBesulfonate; SOE-PCR, splicing-

x: +662 441 9906.r).

l rights reserved.

The dengue virus genome encodes uninterrupted open readingframes of 10188, 10173, 10170 and 10158 nucleotides in DEN1, 2, 3 and4, respectively, which are translated to polyprotein precursors of 3396,3391, 3390 and 3386 amino acid residues with individual proteinsarranged in the order C-prM-E-NS1-NS2-NS2A-NS2B-NS3-NS4A-NS4B-NS5. The polyprotein is co- and posttranslationally cleaved byhost cell proteases and the viral proteaseNS3 into 3 structural proteins,capsid, premembrane and envelope protein, which are components ofthe virion and at least seven non-structural proteins, which areinvolved in replication and maturation of the virus [4]. The virus-encoded protease complexNS2B–NS3 is responsible for cleavage at theNS2A/NS2B, NS2B/NS3, NS3/NS4A and NS4B/NS5 junctions and in thecapsid protein.

The N-terminal 180 amino acid residues of the 69 kDa NS3 proteincontain a serine protease with a classic catalytic triad comprised ofresidues H51, D75 and S135, whereas the C-terminal two-thirds of NS3are associated with the enzymatic activities of a nucleoside triphos-phatase, a 5′-RNA triphosphatase (RTPase) and a NTP-stimulatedhelicase, which are presumably required for capping of viral geno-mic RNA and unwinding of double-stranded replicative intermediates[5–7]. Autoproteolytic cleavage of the NS2B(H)-NS3pro molecule atthe NS2B/NS3 site results in the formation of an enzymatically activenon-covalent complex between NS2B(H) and NS3pro [5].

Similar to other trypsin-like serine proteases, the 3-dimensionalstructure of dengue virus NS3pro encompassing the N-terminal 185amino acid residues comprises 2β-barrels, each formedby6β-strands,

990 T. Iempridee et al. / Biochimica et Biophysica Acta 1780 (2008) 989–994

in perpendicular arrangement with the catalytic apparatus positionedat the two domain interface [8]. The formation of a heterodimericcomplex between the NS3 protease and the 14 kDa NS2B cofactor is aprerequisite for optimal catalytic activity of the protease and site-directed mutagenesis experiments have recently identified individualresidues within the 40-residues NS2B activation domain presumablyinvolved in the association of the NS2B–NS3 complex and thestructural activation of the protease [9–11].

Crystal structures of the NS2B–NS3pro co-complex fromWest Nilevirus and dengue virus have been resolved recently and have providedinsight into a novel mechanism of cofactor-dependent proteaseactivation, which involves a rearrangement of the NS3pro globalfold conducive to a catalytically productive conformation of the activesite [11,12]. Based on structural similarities between the substrate-freeconformations of the NS2B cofactors in dengue virus and West Nilevirus, it was proposed that substrate binding leads to an induced fitmechanism of catalysis and it was speculated that the flaviviral pro-teases possess an alternative conformational state in vivo which isproteolytically inactive and terminates polyprotein processing aftergeneration of sufficient amounts of non-structural proteins [11].

The substrate specificity of the dengue virus serotype 2 NS3 pro-tease has been investigated by functional profiling of randomizedpeptide libraries andbyusing internally quenchedfluorescent peptides[13,14]. These studies have established preferences of the NS3 proteasefor dibasic residues (RR, KR and RK) at the P1 and P2 positions followedby a small side chain amino acid, with serine being preferred overglycine and alanine, at the P1′ position of the cleavage site sequence. Ina recent study an extended substrate binding site was suggested withinteractions that extend from the P6 to the P6′ positions [15].

Previously published studies on the dengue virus NS3 protease hadpredominantly focused on the enzyme from serotype 2 as a modelsystem for mechanistic and structural analysis [5,8,11,13]. It is con-ceivable, however, that – in addition to host-derived factors – diseaseseverity and haemorrhagic symptoms during an outbreak of denguefever are modulated by serotype or strain-specific factors. Notwith-standing the observation that multiple serotypes circulate in endemicareas (hyperendemicity), seroepidemiological data from various stu-dies have confirmed the association of one predominant serotype ofdengue virus with disease outbreaks [16].

In this report we have addressed the question of biochemicalsimilarities and divergences between the NS2B(H)–NS3pro serineproteases from4 dengue virus serotypes.We demonstrate that despitepossessing obvious similarities such as pH optima and overall tertiarystructure, the proteases from the 4 serotypes appear to have somedisparate properties such as catalytic efficiency, substrate discrimina-tion, thermostability, expression yields and protein stability in E. coli.

2. Materials and methods

2.1. Construction of pTrcHisA/NS2B(H)–NS3pro of DEN1, DEN3 and DEN4 by SOE-PCR

The DNA encoding the NS2B(H)–NS3pro precursor containing NS2B residues 48–95followed by residues 121–130 and the N-terminal 180 amino acids of the NS3 proteinfrom dengue virus serotypes 1, 3 and 4 was generated in analogy to proceduresdescribed previously for serotype 2 [10]. The constructs were cloned into the pTrcHisAexpression vector (Invitrogen) to yield the polyhistidine-tagged fusion proteins. Thesequence of all constructs was confirmed by automated DNA sequence analysis using aBigDye Terminator Cycle Sequencing kit (Perkin-Elmer, Norwalk, USA).

2.2. Construction of catalytically inactive mutants of DEN1, DEN3 and DEN4

Alanine substitutions were introduced in the NS3 protease domain at position S135to generate the catalytically inactive protease using the QuickChange site-directedmutagenesis kit (Stratagene) as described earlier [10].

2.3. Expression and purification of DEN1–4 NS2B(H)-NS3pro protease complexes

Protease complexes were purified by 2-step procedure using a Hitrap chelatingcolumn (Pharmacia) and a Superdex 75 HR 10/300 gel filtration column (Pharmacia) asdescribed earlier [10]. Protein samples were analyzed by Western blotting using anti-

Xpress antibodies (Invitrogen) directed against the N-terminal polyhistidine purifica-tion tag.

2.4. Renaturation of DEN2–4 NS2B(H)–NS3pro recombinant proteins

Renaturation of purified enzymes from inclusion bodies was initiated by artificialchaperone-assisted refolding with cycloamylose (Ezaki Glico Co., Japan; degree ofpolymerization from 22 to 50). Two non-ionic detergents, Tween-40 and Tween-60,and two ionic detergents, CTAB and SB 3-14, were explored for refolding. Purifiedproteins (3–10 mg ml−1) were mixed with 1.4 ml of 0.1 M Tris–HCl, pH 9.0, containing0.05% (v/v) detergent and incubated at room temperature for 1 h. Cycloamylose wasadded to the protein–detergent complex, giving a 0.6% (w/v) final concentration and themixtures were incubated at room temperature ranging from 1.5–48 h or for 24 h at 4 °C.Denatured proteins were diluted with 0.1 M Tris–HCl, pH 9.0 containing 0.05% Tween-40 for DEN2 and 3, and 0.05% Tween-60 for DEN4. Preparations of refolded enzymeswere centrifuged (10,000 ×g, 30 min, 4 °C) and the resultant supernatant wasconcentrated by Centricon YM-3 devices (3 kDamolecular weight cutoff, Millipore). Therefolded enzymewas stored at −20 °C in 0.1M Tris–HCl, pH 9.0, 50% (v/v) glycerol for upto 1 week.

2.5. Enzyme assays

Autoproteolytic cleavage at the NS2B/NS3 junction of the refolded NS2B(H)-NS3prowas analyzed by SDS-PAGE analysis. Enzymatic activity of the NS2B(H)–NS3prorecombinant proteases was assayed with the fluorogenic peptide substrate, tBoc-Gly-Arg-Arg-4-methylcoumaryl-7-amide (Peptides International) as described previously[14]. Initial velocities were corrected for inner-filter effects as described in the literature[17]. At least three independent experiments were performed for each set of datapoints. Fluorescence spectrawere obtained by emission scanning of samples containingapprox. 0.2 mg ml−1 purified NS2B(H)–NS3pro proteases in quartz cuvettes (0.5 cmpath length) using a Jasco FP-6300 (Jasco Corp., Tokyo, Japan) spectrofluorometer atexcitation wavelength λ=295 nm and emission spectra were recorded from 300 to550 nm.

2.6. Determination of active enzyme concentration

The concentration of active NS2B(H)-NS3(pro) protease was determined bytitration with aprotinin (synonymous to bovine pancreatic trypsin inhibitor, BPTI).Proteases (0.15–0.4 µM) in 50 mM Tris–HCl, pH 9.5, 20% glycerol, 1 mM CHAPS weremixed with different amounts of aprotinin (Sigma) and were pre-incubated at 37 °C for30 min. The reaction was initiated by adding 150 µM of GRR-amc. The concentration ofactive enzyme was determined by Henderson equation

I½ �= 1� vi=v0ð Þ ¼ Kappi v0=við Þ þ E½ � ð1Þ

where vi is initial velocity in the presence of inhibitor; v0 is initial velocity in theabsence of inhibitor; Ki

app is the apparent value of Ki at substrate concentration [S]; [E] isthe active enzyme concentration; and [S] is the substrate concentration. Concentrationsof active enzymes were obtained graphically from Henderson plots where [I]/(1−vi /v0)is plotted versus the reciprocal fractional velocity (v0 /vi) for a series of inhibitorconcentrations [18].

pH-dependence of enzymatic hydrolysis was determined by titration with variousbuffers in the pH range from 6.0 to 11.0 using 0.3 µM enzyme and 100 µM GRR-amcsubstrate in 50 mM buffer [MES (pH 6.0), MOPS (pH 7.0), Tris–HCl (pH 7.5–9.5) andCAPS (pH 10.0–11.0)] containing 20% glycerol and 1 mM CHAPS.

2.7. Thermal stability assays

Reaction mixtures (100 µl), containing 0.15 µM purified NS2B(H)–NS3pro proteasein 50mM Tris–HCl, pH 9.0, 20% glycerol, 1 mM CHAPS, were pre-incubated for 10min attemperatures ranging from 20 °C–60 °C. Assays were started by the addition of GRR-amc (100 µM) and incubation was continued for 20 min followed by end-pointmeasurement of released fluorescence. To determine the half-lifes of the proteases at50 °C, reactions were pre-incubated at 50 °C, samples were withdrawn at different timepoints and the remaining activities were measured at 37 °C. Half-life (t1/2) wascalculated from plots of log percentage remaining activity versus incubation time. Dataare reported as means of triplicate measurements±standard error.

3. Results and discussion

3.1. Cloning, expression and purification of the NS3pro serine proteasesfrom 4 dengue serotypes

The NS2B–NS3 two-component proteases of the Flaviviridae arecurrently being intensively investigated with the objective to discoversmall molecule inhibitors which are suitable as treatment option for anumber of severe diseases caused by these human pathogens [5,8,3,14]. In this study we have expressed and purified the NS2B(H)–

Fig.1. Schematic representation of the dengue virus polyprotein and construction of theNS2B(H)–NS3pro precursor by PCR. The figure shows the organization of the viralstructural (C, prM and E) and non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A,NS4B and NS5) in the viral polyprotein precursor. The NS2B(H)-to-NS3pro encodingsequence was obtained by RT-PCR on viral RNA samples as described in Materials andMethods. The NS2B(H)–NS3pro fusion protein was generated by SOE-PCR and wascloned into the pTrcHis vector downstream of the N-terminal (His)6-purification tag.The NS2B/NS3 cleavage site is indicated by a black triangle.

Fig. 2. A: Western blot analysis of protein lysates from E. coli expressing the recom-binant NS2B(H)–NS3pro constructs from 4 dengue virus serotypes. Lanes 1–4: lysatesfrom E. coli containing pTrcHisA/NS2B(H)–NS3pro from dengue virus serotypes 1, 2, 3and 4, respectively. Lane 5: protein lysate from E. coli transformed with pTrcHis/chloramphenicol acetyl transferase; lane 6: sample from E. coli containing pTrcHisA.The 29.8 kDa NS2B(H)–NS3pro precursor displays anomalous migration on SDS-PAGEgels at an apparent molecular weight of approx. 37 kDa. Molecular weights of themarker proteins are indicated on the left. B: Western blot analysis of protein lysatesfrom recombinant E. coli overexpressing the NS2B(H)–NS3pro precursor from 4 denguevirus serotypes and the corresponding inactive mutants S135A. Lane 1: lysate of E. colicontaining pTrcHisA vector; lanes 2, 4, 6, 8: lysate of E. coli expressing the wild-typeNS2B(H)–NS3 precursor of dengue virus serotype 1, 2, 3, and 4, respectively; lanes 3, 5,7, 9: lysates from E. coli expressing the NS2B(H)–NS3pro S135A inactive mutants fromdengue virus serotypes 1, 2, 3, and 4, respectively.

991T. Iempridee et al. / Biochimica et Biophysica Acta 1780 (2008) 989–994

NS3pro serine proteases from 4 dengue virus serotypes in order toanalyze possible changes in biochemical properties caused by theirgenetic divergence. The following strains of dengue virus were used:DEN1 strain 16007 [19], DEN2 strain 16681 [20], DEN3 strain 16562[20] and DEN4 strain 1036 [21].

Paired alignments of the amino acid sequences for the dengue virusNS2B(H)–NS3pro precursor indicate identities in the range of 61–71%and similarities ranging from 78–86% between the 4 serotypes.Analysis of sequence similarities in the NS2B(H) and NS3pro regionsshows a lower degree of conservation for the NS2B cofactor than theNS3 protease with about 44–61% identity, whereas the proteasesequences exhibit 64–73% identity between 2 serotypes. Phylogeneticanalysis based on amino acid sequences of NS2B and NS3pro revealsthat DEN1 and DEN3 are closely related, whereas DEN4 exhibits alarger evolutionary distance to the remaining 3 serotypes [1].

All 4 NS2B(H)–NS3pro proteases were cloned, expressed and puri-fied by a uniform methodology using SOE-PCR and ligation of theresulting products into the pTrcHisAvector downstreamof a N-terminalhexahistidine purification tag (Fig. 1). The sequence of NS2B–NS3profrom dengue virus serotype 4, strain 1036, was submitted to GenBankaccession no. AY253850.

Expression of the recombinant proteases from the 4 dengue virusserotypes was analyzed byWestern immunoblotting withmonoclonalantibodies directed against the polyhistidine tag. Contrary to ourexpectations, substantial differences in the amount of proteins formedin the E. coli host cell and the presence of immunoreactive by-productswere observed (Fig. 2). In particular, an unusual high degree ofbioproduct formation was observed with the NS2B(H)–NS3proprotease from DEN4, which was associated with the presence of anumber of immunoreactive protein signals atmolecularweights lowerthan 30 kDa. To address the question as of whether autoproteolyticcleavage of NS2B(H)–NS3pro at secondary sites produced the lowermolecular weight impurities, enzymatically inactive mutants S135Awere generated by site-directed mutagenesis. Expression and electro-phoretic analysis of samples containing the inactive proteasesdemonstrated that a number of products observed with wild-typesamples (for example 25 kDa in DEN1 and 32, 11 and 7 kDa in DEN4)originate from proteolytic autocleavage of the precursor duringexpression. We have used a homology model for the DEN1 proteasebased on the crystal structure of DEN2 NS3prowhich suggests that theputative cleavage sites are located at solvent-exposed surface loopswhich are easily accessible to proteolytic enzymes (R. Thongphung,unpublished data). A number of lowmolecularweight impurities like a35 kDa product in DEN2 and a 32 kDa product in DEN3 were observed

in samples of the inactive S135A mutants and therefore do notrepresent products of autoproteolysis. Earlier studies had suggestedthat premature termination of translation and/or proteolytic degrada-tion by host cell proteases could be the source of these heterogeneousfragments [22].

An interesting observation is that in all cases expression levels ofNS2B(H)–NS3pro appeared to be higher for the inactive S135Amutants than for the wild-type, thus suggesting cytotoxic effects ofa catalytically active dengue virus NS2B(H)–NS3pro protease ongrowth of the host cells. Expression of a number of non-structuralproteins including NS2B–NS3 of Japanese encephalitis virus resultedin altered membrane permeability and reduced cell growth rates inthe E. coli system [23].

3.2. Refolding and assay of activity

Recombinant proteases of DEN2, 3 and 4 overproduced in E. coliwere obtained predominantly as insoluble aggregates in the form ofinclusion bodies whereas the DEN1 product appeared to be containedin the soluble fraction of the cell lysate. NS2B(H)–NS3pro proteinswere purified by a 2-step process usingmetal affinity chromatographyand size-exclusion chromatography on Superdex 75 HR 10/300 gelfiltration columns (Fig. 3). Purity of proteins eluted frommetal affinitycolumns was estimated to be 70–90% based on electrophoreticanalysis on 15% SDS-PAGE gels. Further purification by gel filtrationyielded products of N95% purity as estimated from SDS-PAGE analysis

Fig. 3. Purification of the DEN2 NS2B(H)–NS3pro protein by size-exclusion chromato-graphy. A: Elution profile of samples obtained by Ni2+-affinity chromatography using aSuperdex 75 HR 10/300 column. Elution was monitored by UV detection at 215 (■) and280 (▼) nm. The column was eluted with 100 mM Tris–HCl, pH 8.0, 300 mM NaCl, 8 Murea at a flow rate of 0.3 ml min−1 and fractions of 0.2 ml were collected. The peakfraction containing NS2B(H)–NS3pro at a retention volume of 8.4 ml is indicated by anasterisk. B: SDS-PAGE analysis of fractions obtained by gel filtration. Lane 1: molecularweight marker proteins; lane I: protein sample as obtained from metal chelate affinitychromatography; lanes 1–6: elution fractions of size-exclusion chromatography.

992 T. Iempridee et al. / Biochimica et Biophysica Acta 1780 (2008) 989–994

of elution fractions. Refolding of the NS2B(H)–NS3pro precursor bysuccessive dialysis has been reported in the literature but is associatedwith the risk of precipation and poor recovery [24]. In analogy topublished protocols [25], we have therefore attempted to increaserecovery of the renatured proteases by the use of an artificialchaperone, cycloamylose, and we have found that refolding of theDEN2 and DEN3 proteases was more efficient by using Tween-40,whereas refolding of DEN4 NS2B(H)–NS3pro was more optimal in thepresence of Tween-60. Refolding of the enzymes in the presence oftween and cycloamylose reached a maximum after approx. 2 h and adecrease of activity was observed upon prolonged incubation. TheDEN4 protease demonstrated only 50% of initial activity after

Fig. 4. Fluorescence emission spectra for purified and refolded NS2B(H)–NS3pro pro-teases. Intrinsic tryptophan fluorescence was recorded for DEN1, 2, 3 and 4 as shown inthe inset from 300 to 550 nm at an excitation wavelength 295 nm.

incubation for 6 h. It is conceivable that this reduction in enzymeactivity is associated with a competing nonproductive process leadingto increased protein aggregation. A similar observation for thetemperature-dependent renaturation of bovine carbonic anhydraseII was explained by the existence of slow-folding intermediates withdifferent renaturation kinetics [26]. Consistent with the aggregation-prone behaviour of the DEN4 protease, the protein exhibits a markedsensitivity to heat inactivation and the lowest half-life at 50 °C of allinvestigated proteases. Intrinsic tryptophan fluorescence emissionspectra of the refolded proteases revealed a close similarity betweenthe 4 serotypes with an emission peak at 348 nm thus indicatingconserved overall conformations (Fig. 4).

Processing of the precursor proteases was incomplete uponpurification and we expected that only a proportion of the proteasemoleculeswould be enzymatically active.Wehave thereforemeasuredkinetic parameters for the fluorogenic substrate GRR-amc by usingactive-site titration in the presence of the tight-binding inhibitor,aprotinin [13]. Dose–response plots indicated a quasi-linear decreaseof fractional velocities in the presence of increasing inhibitorconcentrations as predicted for the case where virtually all of theinhibitor molecules are bound to the protease (data not shown). Wehave not attempted to determine the inhibition constant for aprotinin,however, dose–response plots for the case of E/KiN100 suggest valuesfor Ki in the picomolar range. In an earlier literature report, highaffinity binding of aprotinin has been observedwith Ki values of 79, 25,88 and 6.4 pM for the DEN1–4 proteases, respectively [13]. Fractions ofcatalytically active enzymes obtained by refolding were estimated to11.7, 12.6 and 4% for DEN2, 3 and 4, respectively, thus indicating poorrecovery and limited efficiency of the renaturation process. However,rate constants obtained for substrate cleavage using inner-filter effectcorrection [17] were in excellent agreement with Michaelis–Mentenkinetics and steady-state kinetics for NS2B(H)–NS3pro from DEN2, 3and 4 indicated similar binding affinities (Km), whereas catalytic rates(kcat) displayed larger variations. Kinetic parameters for the 4 proteasesare summarized in Table 1. The rank order for catalytic efficiencies wehave found is DEN4NDEN2NDEN3NDEN1, however, differences arerelatively small with DEN4 exhibiting only a 1.5- and 3.0-fold greaterkcat/Km when compared to DEN2 and 3, respectively.

Recently, Bessaud et al. have reported kinetic constants for the NS3proteases from seven flaviruses including DEN1 and DEN3 containinga C-terminal polyhistidine purification tag [27]. The DEN1 proteasedisplayed a 5.5-fold lower Km value for the GRR-amc substrate whencompared to the DEN3 enzyme, whereas Km observed for the peptideGKR-amc was approx. 20-fold lower for DEN3. The reasons for thesediscrepancies are not entirely clear at the moment, however, therelatively large differences in binding affinity suggest the possibility ofsignificant conformational effects of additional sequences introducedfor cloning and purification purposes on enzymatic specificity of theNS2B–NS3 proteases.

Table 1Numerical constants for the hydrolysis of the peptide substrate GRR-amc by the NS2B(H)-NS3pro proteases of four serotypes of dengue virus

Serotype Km (μM) kcat (s−1) kcat/Km (M−1 s−1)

DEN1⁎ 45±1.3 0.034±0.0006 759±20DEN2 72±4.3 0.21±0.01 2,905±31DEN3 79±3.1 0.12±0.003 1,490±66DEN4 64±1.9 0.3±0.01 4,554±169

Protease activity was assayed in 50 mM Tris–HCl (pH 9.0), 20% glycerol, 1 mM CHAPS,for 20 min at 37 °C with the fluorogenic peptide GRR-amc at concentrations rangingfrom 10 to 600 µM. Standard reactions contained protease at a concentration of 0.15 µM.Kinetic parameters, Km and kcat, were calculated by non-linear fitting of data to aMichaelis–Menten equation, v=Vmax[S]/[S]+Km. kcat values were calculated by usingconcentrations of enzymes as determined by active-site titration with the exception ofthe NS2B(H)–NS3pro protease from serotype 1, indicated by ⁎, which was purified fromthe soluble fraction of the E. coli lysate.

Fig. 6. A: Activity profiles of DEN2, 3 and 4 in the temperature range between 20 °Cand 60 °C. Shown is the percentage of relative velocities for NS2B(H)–NS3pro fromDEN2 (□), DEN3 (▵) and DEN4 (○). Activity at the optimum temperature was set to100%. Error bars indicate standard error of the mean (SEM) from 3 experiments.B: Shown are half-life times of the NS2B(H)–NS3pro proteases from dengue virusserotypes 2, 3, and 4 at 50 °C. Purified enzymes were pre-incubated with 100 μMsubstrate GRR-amc in 50 mM Tris–HCl, pH 9.0, 20% glycerol, 1 mM CHAPS at 50 °C fordifferent periods of time. Remaining activity was assayed at 37 °C and half-life timeswere calculated by plotting the log of remaining activity against the incubation time.Error bars indicate the standard error of the mean (SEM) from triplicate measurements.

993T. Iempridee et al. / Biochimica et Biophysica Acta 1780 (2008) 989–994

It is noteworthy that depending on the substrate analyzed, the NS3proteases from 4 dengue virus serotypes can exhibit much largerdifferences in specificity thanwe have seen for the GRR-amc substrate.Li et al. have found a kcat/Km for Bz-nKTR-acmc of 93,900 M−1 s−1 withthe DEN4 protease, which is 19.5-fold higher than that of the DEN1enzyme and for the substrate peptide Bz-nKRR-acmc, kcat/Km of theDEN4 protease is approx. 7.5-fold higher than for DEN1 [13]. Therecombinant construct used by Li et al. contains a non-cleavable linkerinserted between the NS2B(H) activation domain and the NS3proprotease domain. Therefore the possibility exists that the differencesobserved for substrate specificity between this constitutively activatedsingle chain protease and the NS2B(H)-NS3pro co-complex relate tothese structural variations.

The effect of pH on enzymatic activity for the serotypes 2, 3 and 4proteases was determined by assays using different buffers in the pHrange from 6.0–11.0 (Fig. 5). The optimal pH for cleavage of thesynthetic peptide substrate GRR-amc was found to be 9.5. It isnoteworthy that reactivity of the enzymes at physiological pH is onlyapprox. 4% of the activity observed at pH 9.5 and that for the NS3protease from dengue virus serotype 3, a biphasic titration curve wasobserved in an earlier study which displayed a second peak of activityaround pH 10.5 [27]. The structural stability of the NS2B(H)–NS3proproteases was characterized by assaying proteolytic activity atelevated temperatures. Proteolytic activity of DEN2, 3 and 4 proteasesappeared to be maintained over a range of temperatures with N70% ofactivity between 25 °C and 50 °C (Fig. 6, panel A). The DEN2 proteaseappeared to possess greater resistance to increasing temperaturesthan the DEN3 and DEN4 enzymes. The half-life times of the enzymeswere determined by incubation at 50 °C and remaining activities aftervarious periods of incubation were assayed with GRR-amc at 37 °C.These experiments revealed significant differences in half-life at 50 °Cfor the 3 enzymes with the DEN2 protease displaying a 3.5- and 12-fold greater recovery of activity when compared to the DEN3 andDEN4 proteases, respectively (Fig. 6, panel B). Proteolytic activitywithin a broad range of temperatures has recently been observed forthe NS3 proteases from Zika virus and St. Louis encephalitis virus [27].

A number of residues within the active site of the West Nile virusNS3 protease probably involved in enzyme-substrate interactionswere identified recently [28,29]. Residues V115, D129, Y150, S160 andS163 were predicted to line the S1 subsite of the protease, while N152was proposed to serve as key residue in the S2 subsite. With theexception of the residues at positions 115 and 160, all these residuesare strictly conserved among all flaviviruses. In a structure-guidedmutagenesis study, Shiryaev et al. have identified residues R76, P131and T132 as critical determinants for the cleavage specificity of theWest Nile virus NS3 protease, and mutation of these residues to R76L,P131K and T132P resulted in a proteasewith cleavage preferences akin

Fig. 5. Effect of pH on the proteolytic activity of NS2B(H)–NS3pro from dengue virusserotypes 2, 3 and 4. Shown are activities of the dengue proteases from serotypes 2 (■),3 (▼) and 4 (●) as assayed by using 100 μM substrate GRR-amc at 37 °C in the pH rangefrom 6.0–11.0 by using different buffers (50 mM MES, MOPS, Tris, CAPS). Hydrolysis ofGRR-amc in the absence of enzyme was used as a control (✖). Error bars indicate thestandard error of the mean (SEM) from three independent measurements.

to dengue virus NS3pro [30]. It is interesting to note that even thesespecificity-determining residues are not strictly conserved among the4 dengue virus serotypes. Within the NS3 protease sequence ofserotype 4, L76 is replaced by M, and with serotype 2, K131 issubstituted by S, whereas only P132 is invariantly present in all 4serotypes. Whether these variations could give rise to significantchanges in enzymatic specificity is still an unresolved question.

Although the NS3 proteins of several flaviviruses display signifi-cant levels of sequence homology and astounding similarities in theirtertiary structures, the available data for the substrate specificity ofthe flaviviral NS3 proteases imply that subtle variations exist whichcould probably modulate proteolytic activity in vivo and accordinglykinetics of virus replication and maturation. In this report, we haveaccumulated preliminary evidence that the NS2B(H)–NS3pro pro-teases from 4 dengue virus serotypes display a number of discreteproperties such as expression yield, pattern of product formation and(auto)proteolysis in E. coli, refolding kinetics, thermostability andproteolytic activity. Although the physiological significance of thesedifferences remains to be investigated, we believe that the informa-tion presented here nevertheless provides an incentive to furtherexplore the biochemical dissimilarities between the flaviviral NS3proteases with a view to the rationale-based design of antiviralinhibitors.

Acknowledgments

We are grateful to Prof. Dr. Duncan R. Smith, Institute of MolecularBiology and Genetics, Mahidol University, Thailand, for a generous giftof dengue virus strains. We thank Drs. Santad Chanprapaph, Dept. ofPharmaceutical Chemistry, Chulalongkorn University, Thailand andSupa Hannongbua, Dept. of Chemistry, Kasetsart University, Thailand,for the helpful discussions. Anchalee Nirachanon is acknowledged for

994 T. Iempridee et al. / Biochimica et Biophysica Acta 1780 (2008) 989–994

her excellent secretarial assistance. This work was supported by grantBRG49800008 (to G.K.) from the Thailand Research Fund (TRF).

References

[1] E.G.Westaway, J. Blok, Taxonomy and evolutionary relationships of flaviviruses, in:D.J. Gubler, G. Kuno (Eds.), Dengue and Dengue Hemorrhagic Fever, CABInternational, New York, 1997, pp. 147–173.

[2] S. Melino, M. Paci, Progress for dengue virus diseases. Towards the NS2B-NS3proinhibition for a therapeutic-based approach, FEBS J. 274 (2007) 2986–3002.

[3] M.D. Ryan, S. Monaghan, M. Flint, Virus-encoded proteinases of the Flaviviridae,J. Gen. Virol. 79 (1998) 947–959.

[4] T.J. Chambers, C.S. Hahn, R. Galler, C.M. Rice, Flavivirus genome organization,expression, and replication, Annu. Rev. Microbiol. 44 (1990) 649–688.

[5] R. Yusof, S. Clum, M. Wetzel, H.M. Murthy, R. Padmanabhan, Purified NS2B/NS3serine protease of dengue virus type 2 exhibits cofactor NS2B dependence forcleavage of substrates with dibasic amino acids in vitro, J. Biol. Chem. 275 (2000)9963–9969.

[6] G. Bartelma, R. Padmanabhan, Expression, purification, and characterization of theRNA 5′-triphosphatase activity of dengue virus type 2 nonstructural protein 3,Virology 299 (2002) 122–132.

[7] D. Luo, T. Xu, C. Hunke, G. Grüber, S.G. Vasudevan, J. Lescar, Crystal structure of theNS3 protease-helicase from dengue virus, J. Virol. 82 (2008) 173–183.

[8] H.M. Murthy, S. Clum, R. Padmanabhan, Dengue virus NS3 serine protease. Crystalstructure and insights into interaction of the active site with substratesby molecular modeling and structural analysis of mutational effects, J. Biol.Chem. 274 (1999) 5573–5580.

[9] B. Falgout, R.H. Miller, C.J. Lai, Deletion analysis of dengue virus type 4 non-structural protein NS2B: identification of a domain required for NS2B-NS3protease activity, J. Virol. 67 (1993) 2034–2042.

[10] P. Niyomrattanakit, P. Winoyanuwattikun, S. Chanprapaph, C. Angsuthanasombat,S. Panyim, G. Katzenmeier, Identification of residues in the dengue virus type 2NS2B cofactor that are critical for NS3 protease activation, J. Virol. 78 (2004)13708–13716.

[11] A.E. Aleshin, S.A. Shiryaev, A.Y. Strongin, R.C. Liddington, Structural evidence forregulation and specificity of flaviviral proteases and evolution of the Flaviviridaefold, Protein Sci. 16 (2007) 795–806.

[12] P. Erbel, N. Schiering, A. D'Arcy, M. Renatus, M. Kroemer, S.P. Lim, Z. Yin, T.H. Keller,S.G. Vasudevan, U. Hommel, Structural basis for the activation of flaviviral NS3proteases from dengue and West Nile virus, Nat. Struct. Mol. Biol. 13 (2006)372–373.

[13] J. Li, S.P. Lim, D. Beer, V. Patel, D. Wen, C. Tumanut, D.C. Tully, J.A. Williams, J. Jiricek,J.P. Priestle, J.L. Harris, S.G. Vasudevan, Functional profiling of recombinant NS3proteases from all four serotypes of dengue virus using tetrapeptide andoctapeptide substrate libraries, J. Biol. Chem. 280 (2005) 28766–28774.

[14] P. Niyomrattanakit, S. Yahorova, I. Mutule, F. Mutulis, R. Petrovska, P. Prusis, G.Katzenmeier, J.E. Wikberg, Probing the substrate specificity of the dengue virustype 2 NS3 serine protease by using internally quenched fluorescent peptides,Biochem. J. 397 (2006) 203–211.

[15] I.E. Gouvea, M.A. Izidoro, W.A.S. Judice, M.H.S. Cezari, G. Caliendo, V. Santagada,C.N.D. dos Santos, M.H. Queiroz, M.A. Juliano, P.R. Young, D.P. Fairlie, L. Juliano,Substrate specificity of recombinant dengue 2 virus NS2B-NS3 protease:

influence of natural and unnatural basic amino acids on hydrolysis of syntheticfluorescent substrates, Arch. Biochem. Biophys. 457 (2007) 187–196.

[16] S.B. Halstead, Epidemiology of dengue and dengue hemorrhagic fever, in: D.J.Gubler, G. Kuno (Eds.), Dengue and Dengue Hemorrhagic Fever, CAB International,New York, 1997, pp. 23–44.

[17] Y. Liu, W. Kati, C.M. Chen, R. Tripathi, A. Molla, W. Kohlbrenner, Use of afluorescence plate reader for measuring kinetic parameters with inner filter effectcorrection, Anal. Biochem. 267 (1999) 331–335.

[18] P.J.F. Henderson, A linear equation that describes the steady-state kinetics ofenzymes and subcellular particles interacting with tightly bound inhibitors,Biochem. J. 127 (1973) 321–333.

[19] C.Y. Huang, S. Butrapet, D.J. Pierro, G.J. Chang, A.R. Hunt, N. Bhamarapravati, D.J.Gubler, R.M. Kinney, Chimeric dengue type 2 (vaccine strain PDK-53)/dengue type1 virus as a potential candidate dengue type 1 virus vaccine, J. Virol. 74 (2000)3020–3028.

[20] S.B. Halstead, P. Simasthien, Observations related to the pathogenesis of denguehemorrhagic fever. II. Antigenic and biologic properties of dengue viruses andtheir association with disease response in the host, Yale J. Biol. Med. 42 (1970)276–292.

[21] D.J. Gubler, W. Suharyono, Sumarmo, H. Wulur, E. Jahja, J. Sulianti Saroso,Virological surveillance for dengue hemorrhagic fever in Indonesia usingthe mosquito inoculation technique, Bull, World Health Organ. 57 (1979)931–936.

[22] T.L. Arakaki, N.X. Fang, D.P. Fairlie, P.R. Young, J.L. Martin, Catalytically activeDengue virus NS3 protease forms aggregates that are separable by size exclusionchromatography, Protein Expr. Purif. 25 (2004) 241–247.

[23] Y.S. Chang, C.L. Liao, C.H. Tsao, M.C. Chen, C.I. Liu, L.K. Chen, Y.L. Lin, Membranepermeabilization by small hydrophobic nonstructural proteins of Japaneseencephalitis virus, J. Virol. 73 (1999) 6257–6264.

[24] D. Leung, K. Schroder, H. White, N.X. Fang, M.J. Stoermer, G. Abbenante, J.L. Martin,P.R. Young, D.P. Fairlie, Activity of recombinant dengue 2 virus NS3 protease in thepresence of a truncated NS2B co-factor, small peptide substrates, and inhibitors,J. Biol. Chem. 276 (2001) 45762–45771.

[25] D.L. Daugherty, D. Rozema, P.E. Hanson, S.H. Gellman, Artificial chaperone-assistedrefolding of citrate synthase, J. Biol. Chem. 273 (1998) 33961–33971.

[26] Y. Xie, D.B. Wetlaufer, Control of aggregation in protein refolding: thetemperature-leap tactic, Protein Sci. 5 (1996) 517–523.

[27] M. Bessaud, B.A.M. Pastorino, C.N. Peyrefitte, D. Roland, M. Grandadam, H.J. Tolou,Functional characterization of the NS2B/NS3 protease complex from seven virusesbelonging to different groups inside the genus Flavivirus, Virus Res. 120 (2006)79–90.

[28] K.J. Chappell, T.A. Nall, M.J. Stoermer, N.X. Fang, J.D. Tyndall, D.P. Fairlie, P.R.Young, Site-directed mutagenesis and kinetic studies of the West Nile Virus NS3protease identify key enzyme-substrate interactions, J. Biol. Chem. 280 (2005)2896–2903.

[29] K.J. Chappell, M.J. Stoermer, D.P. Fairlie, P.R. Young, Insights to substrate bindingand processing by West Nile Virus NS3 protease through combined model-ing, protease mutagenesis, and kinetic studies, J. Biol. Chem. 281 (2006)38448–38458.

[30] S.A. Shiryaev, B.I. Ratnikov, A.E. Aleshin, I.A. Kozlow, N.A. Nelson, M. Lebl, J.W.Smith, R.C. Liddington, A.Y. Strongin, Switching the substrate specificity of thetwo-component NS2B-NS3 flavivirus proteinase by structure-based mutagenesis,J. Virol. 81 (2007) 4501–4509.