Title: Functional Characterization of cis and trans Activity of the Flavivirus NS2B-NS3 Protease
Abstract: Flaviviruses are serious human pathogens for which treatments are generally lacking. The proteolytic maturation of the 375-kDa viral polyprotein is one target for antiviral development. The flavivirus serine protease consists of the N-terminal domain of the multifunctional nonstructural protein 3 (NS3) and an essential 40-residue cofactor (NS2B40) within viral protein NS2B. The NS2B-NS3 protease is responsible for all cytoplasmic cleavage events in viral polyprotein maturation. This study describes the first biochemical characterization of flavivirus protease activity using full-length NS3. Recombinant proteases were created by fusion of West Nile virus (WNV) NS2B40 to full-length WNV NS3. The protease catalyzed two autolytic cleavages. The NS2B/NS3 junction was cleaved before protein purification. A second site at Arg459↓Gly460 within the C-terminal helicase region of NS3 was cleaved more slowly. Autolytic cleavage reactions also occurred in NS2B-NS3 recombinant proteins from yellow fever virus, dengue virus types 2 and 4, and Japanese encephalitis virus. Cis and trans cleavages were distinguished using a noncleavable WNV protease variant and two types of substrates as follows: an inactive variant of recombinant WNV NS2B-NS3, and cyan and yellow fluorescent proteins fused by a dodecamer peptide encompassing a natural cleavage site. With these materials, the autolytic cleavages were found to be intramolecular only. Autolytic cleavage of the helicase site was insensitive to protein dilution, confirming that autolysis is intramolecular. Formation of an active protease was found to require neither cleavage of NS2B from NS3 nor a free NS3 N terminus. Evidence was also obtained for product inhibition of the protease by the cleaved C terminus of NS2B. Flaviviruses are serious human pathogens for which treatments are generally lacking. The proteolytic maturation of the 375-kDa viral polyprotein is one target for antiviral development. The flavivirus serine protease consists of the N-terminal domain of the multifunctional nonstructural protein 3 (NS3) and an essential 40-residue cofactor (NS2B40) within viral protein NS2B. The NS2B-NS3 protease is responsible for all cytoplasmic cleavage events in viral polyprotein maturation. This study describes the first biochemical characterization of flavivirus protease activity using full-length NS3. Recombinant proteases were created by fusion of West Nile virus (WNV) NS2B40 to full-length WNV NS3. The protease catalyzed two autolytic cleavages. The NS2B/NS3 junction was cleaved before protein purification. A second site at Arg459↓Gly460 within the C-terminal helicase region of NS3 was cleaved more slowly. Autolytic cleavage reactions also occurred in NS2B-NS3 recombinant proteins from yellow fever virus, dengue virus types 2 and 4, and Japanese encephalitis virus. Cis and trans cleavages were distinguished using a noncleavable WNV protease variant and two types of substrates as follows: an inactive variant of recombinant WNV NS2B-NS3, and cyan and yellow fluorescent proteins fused by a dodecamer peptide encompassing a natural cleavage site. With these materials, the autolytic cleavages were found to be intramolecular only. Autolytic cleavage of the helicase site was insensitive to protein dilution, confirming that autolysis is intramolecular. Formation of an active protease was found to require neither cleavage of NS2B from NS3 nor a free NS3 N terminus. Evidence was also obtained for product inhibition of the protease by the cleaved C terminus of NS2B. Most flaviviruses, including West Nile virus (WNV), 2The abbreviations used are: WNV, West Nile virus; YFV, yellow fever virus; JEV, Japanese encephalitis virus; ER, endoplasmic reticulum; YFP, yellow fluorescent protein; CFP, cyan fluorescent protein; DV, dengue virus. yellow fever virus (YFV), dengue viruses, and Japanese encephalitis virus (JEV), cause severe human diseases. The plus-sense RNA genome of flaviviruses is a single open reading frame encoding a polyprotein precursor of ∼3400 amino acids, consisting of three structural proteins (C, prM, and E) and seven nonstructural replication proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5) (Fig. 1A) (1Brinton M.A. Annu. Rev. Microbiol. 2002; 56: 371-402Crossref PubMed Scopus (305) Google Scholar, 2Kuhn R.J. Strauss J.H. Adv. Protein Chem. 2003; 64: 363-377Crossref PubMed Scopus (12) Google Scholar, 3Mukhopadhyay S. Kuhn R.J. Rossmann M.G. Nat. Rev. Microbiol. 2005; 3: 13-22Crossref PubMed Scopus (899) Google Scholar). Signal sequences direct the polyprotein into the host endoplasmic reticulum (ER) membrane so that NS1 and the exogenous domains of prM and E are in the lumen; C protein, NS3 and NS5 are cytoplasmic; and proteins NS2A, NS2B, NS4A, and NS4B are predominantly trans-membrane. Post-translational processing of the polyprotein, which is required for virus replication, is performed by a viral NS3 protease (4Chambers T.J. Weir R.C. Grakoui A. McCourt D.W. Bazan J.F. Fletterick R.J. Rice C.M. Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 8898-8902Crossref PubMed Scopus (287) Google Scholar, 5Preugschat F. Yao C.W. Strauss J.H. J. Virol. 1990; 64: 4364-4374Crossref PubMed Google Scholar) in the cytoplasm and by host proteases in the ER lumen. NS3 protease activity is dependent upon association with an NS2B cofactor (NS2B40), a central 40-amino acid hydrophilic domain within the largely hydrophobic NS2B protein (6Chambers T.J. Grakoui A. Rice C.M. J. Virol. 1991; 65: 6042-6050Crossref PubMed Google Scholar, 7Falgout B. Pethel M. Zhang Y.M. Lai C.J. J. Virol. 1991; 65: 2467-2475Crossref PubMed Google Scholar, 8Falgout B. Miller R.H. Lai C.J. J. Virol. 1993; 67: 2034-2042Crossref PubMed Google Scholar). The viral NS2B-NS3 protease cleaves the viral polyprotein precursor at the NS2A/NS2B, NS2B/NS3, NS3/NS4A, and NS4B/NS5 junctions (Fig. 1A), as well as at internal sites within C, NS2A, NS3, and NS4A (9Lin C. Amberg S.M. Chambers T.J. Rice C.M. J. Virol. 1993; 67: 2327-2335Crossref PubMed Google Scholar, 10Arias C.F. Preugschat F. Strauss J.H. Virology. 1993; 193: 888-899Crossref PubMed Scopus (155) Google Scholar, 11Nestorowicz A. Chambers T.J. Rice C.M. Virology. 1994; 199: 114-123Crossref PubMed Scopus (70) Google Scholar, 12Teo K.F. Wright P.J. J. Gen. Virol. 1997; 78: 337-341Crossref PubMed Scopus (63) Google Scholar). In general, the viral protease has specificity for two basic residues (Lys-Arg, Arg-Arg, Arg-Lys or occasionally Gln-Arg) at the canonical P2 and P1 positions immediately preceding the cleavage site, followed by a small amino acid (Gly, Ser, or Ala) at the P1′ position (Fig. 1B). The activity of NS2B-NS3 protease has been studied in vitro using purified, recombinant protease domains of dengue virus type 2 (DV2) and WNV NS3. These studies employed fusions of the NS2B cofactor peptide to truncated forms of NS3 (NS3pro) that included only the N-terminal protease domain, one-third of the full-length NS3, and excluded the C-terminal helicase domain (13Yusof R. Clum S. Wetzel M. Murthy H.M. Padmanabhan R. J. Biol. Chem. 2000; 275: 9963-9969Abstract Full Text Full Text PDF PubMed Scopus (256) Google Scholar, 14Khumthong R. Angsuthanasombat C. Panyim S. Katzenmeier G. J. Biochem. Mol. Biol. 2002; 35: 206-212PubMed Google Scholar, 15Ganesh V.K. Muller N. Judge K. Luan C.H. Padmanabhan R. Murthy K.H. Bioorg. Med. Chem. 2005; 13: 257-264Crossref PubMed Scopus (86) Google Scholar, 16Niyomrattanakit P. Yahorava S. Mutule I. Mutulis F. Petrovska R. Prusis P. Katzenmeier G. Wikberg J.E. Biochem. J. 2006; 397: 203-211Crossref PubMed Scopus (49) Google Scholar). Among several fusion peptides tested, Gly4-Ser-Gly4 was found to be the optimal linker for DV2 NS2B-NS3pro (17Leung D. Schroder K. White H. Fang N.X. Stoermer M.J. Abbenante G. Martin J.L. Young P.R. Fairlie D.P. J. Biol. Chem. 2001; 276: 45762-45771Abstract Full Text Full Text PDF PubMed Scopus (278) Google Scholar, 18Nall T.A. Chappell K.J. Stoermer M.J. Fang N.X. Tyndall J.D. Young P.R. Fairlie D.P. J. Biol. Chem. 2004; 279: 48535-48542Abstract Full Text Full Text PDF PubMed Scopus (101) Google Scholar), and several groups have used this Gly4-Ser-Gly4 linker peptide for a variety of biochemical and mutagenesis studies of protease specificity (19Chappell K.J. Nall T.A. Stoermer M.J. Fang N.X. Tyndall J.D. Fairlie D.P. Young P.R. J. Biol. Chem. 2005; 280: 2896-2903Abstract Full Text Full Text PDF PubMed Scopus (54) Google Scholar, 20Li J. Lim S.P. Beer D. Patel V. Wen D. Tumanut C. Tully D.C. Williams J.A. Jiricek J. Priestle J.P. Harris J.L. Vasudevan S.G. J. Biol. Chem. 2005; 280: 28766-28774Abstract Full Text Full Text PDF PubMed Scopus (219) Google Scholar, 21Chappell K.J. Stoermer M.J. Fairlie D.P. Young P.R. J. Biol. Chem. 2006; 281: 38448-38458Abstract Full Text Full Text PDF PubMed Scopus (79) Google Scholar, 22Shiryaev S.A. Ratnikov B.I. Chekanov A.V. Sikora S. Rozanov D.V. Godzik A. Wang J. Smith J.W. Huang Z. Lindberg I. Samuel M.A. Diamond M.S. Strongin A.Y. Biochem. J. 2006; 393: 503-511Crossref PubMed Scopus (86) Google Scholar). Crystal structures of DV2 and WNV protease domains have been reported (23Murthy H.M. Clum S. Padmanabhan R. J. Biol. Chem. 1999; 274: 5573-5580Abstract Full Text Full Text PDF PubMed Scopus (114) Google Scholar, 24Erbel P. Schiering N. D'Arcy A. Renatus M. Kroemer M. Lim S.P. Yin Z. Keller T.H. Vasudevan S.G. Hommel U. Nat. Struct. Mol. Biol. 2006; 13: 372-373Crossref PubMed Scopus (432) Google Scholar). Crystal structures have also been reported for the helicase domains of YFV and DV2 (25Wu J. Bera A.K. Kuhn R.J. Smith J.L. J. Virol. 2005; 79: 10268-10277Crossref PubMed Scopus (137) Google Scholar, 26Xu T. Sampath A. Chao A. Wen D. Nanao M. Chene P. Vasudevan S.G. Lescar J. J. Virol. 2005; 79: 10278-10288Crossref PubMed Scopus (180) Google Scholar). The protease belongs to the chymotrypsin family with a classic Ser-His-Asp catalytic triad. The catalytic triad is arranged identically in structures with (24Erbel P. Schiering N. D'Arcy A. Renatus M. Kroemer M. Lim S.P. Yin Z. Keller T.H. Vasudevan S.G. Hommel U. Nat. Struct. Mol. Biol. 2006; 13: 372-373Crossref PubMed Scopus (432) Google Scholar) and without (23Murthy H.M. Clum S. Padmanabhan R. J. Biol. Chem. 1999; 274: 5573-5580Abstract Full Text Full Text PDF PubMed Scopus (114) Google Scholar) the NS2B40 cofactor. The NS2B40 cofactor contributes to the binding site for the P2 residue of the substrate, based on the structure of WNV NS2B-NS3 protease domain in complex with a substrate-based inhibitor (24Erbel P. Schiering N. D'Arcy A. Renatus M. Kroemer M. Lim S.P. Yin Z. Keller T.H. Vasudevan S.G. Hommel U. Nat. Struct. Mol. Biol. 2006; 13: 372-373Crossref PubMed Scopus (432) Google Scholar). Here we report single polypeptide variants of NS2B-NS3 protease from five flaviviruses. To our knowledge, this is the first study addressing flavivirus protease activity using the NS2B40 cofactor and full-length NS3. Detailed experiments on the WNV protease demonstrated that the NS2B/NS3 junction and the internal NS3 site are cleaved in cis only. In some variants, product inhibition by the NS2B C terminus was also observed. Plasmid Construction—Constructs encoding WNV (strain NY99) proteins were derived from pWN-CG (27Beasley D.W. Whiteman M.C. Zhang S. Huang C.Y. Schneider B.S. Smith D.R. Gromowski G.D. Higgs S. Kinney R.M. Barrett A.D. J. Virol. 2005; 79: 8339-8347Crossref PubMed Scopus (255) Google Scholar), a kind gift of Richard Kinney, Centers for Disease Control and Prevention. For the fusion of the C-terminal 79 residues (residues 53-131) of NS2B with full-length NS3, a 2154-nucleotide fragment was amplified from pWN-CG, using primers WNV-NS2B53-F and WNV-NS3FL619STOP-R (supplemental Table 1), and cloned between NdeI and BamHI restriction sites in pET28 (Strat-agene) to generate pNS2B53-131-NS3FL (simply pWNV-NS2B79-NS3FL). Other WNV plasmids were constructed in two steps. First, an 1857-nucleotide fragment was amplified from pWN-CG by primers WNV-NS31-F and WNV-NS3619STOP-R and cloned into pET28 between NdeI and BamHI restriction sites to generate pWNV-NS3FL encoding full-length NS3. Then fragments encoding the 41-residue protease cofactor within NS2B (residues 53-93) were amplified from pWN-CG using primers WN-NS2B53-F and either WNV-NS2B93-R or WNV-NS2B95-G4SG4-R. The fragments were digested with NdeI and cloned into NdeI-restricted pWNV-NS3FL to generate either pWNV-NS2B53-93-HM-NS3FL (simply pWNV-NS2B40-HM-NS3FL) or pWNV-NS2B53-95-G4SG4-HM-NS3FL (simply pWNV-NS2B40-G4SG4-HM-NS3FL). Plasmids encoding the active proteases for YFV (pYFV-NS2B40-G4SG4-HM-NS3FL), DV2 (pDV2-NS2B40-G4SG4-ASR-NS3FL), dengue virus type 4 (DV4, pDV4-NS2B40-G4SG4-HM-NS3FL), and JEV (pJEV-NS2B40-G4SG4-ASR-NS3FL) were constructed in two steps, as described for the WNV protease. PCR primers are listed in supplemental Table 1. For the YFV, and DV4 constructs, an NdeI site between the sequences for the NS2B cofactor and NS3 encoded the amino acids His-Met. For the DV2 and JEV constructs, an NheI site was used, corresponding to Ala-Ser and followed by a basic residue "R," corresponding to the C-terminal residue of NS2B. The plasmid for YFV, pACNR/FLYF (28Bredenbeek P.J. Kooi E.A. Lindenbach B. Huijkman N. Rice C.M. Spaan W.J. J. Gen. Virol. 2003; 84: 1261-1268Crossref PubMed Scopus (170) Google Scholar), was provided by Charles Rice, Laboratory of Virology and Infectious Disease, The Rockefeller University, and plasmids for DV2 and DV4 were from Richard Kinney, and a partial cDNA clone of JEV was from Tsutomu Takegami, Medical Research Institute of Kanazawa Medical University. Inactive proteases with alanine substitutions at Ser135 (WNV, DV2, DV4, and JEV) or Ser138 (YFV) in the catalytic triad of NS3 were made by site-directed mutagenesis. For WNV, the substitution was generated by PCR from pWNV-NS2B40-G4SG4-HM-NS3FL and primers WNV-S135A-F and WNV-S135A-R, yielding pWNV-NS2B40-G4SG4-HM-NS3FL/S135A. The PCR product was digested with DpnI before transformation. The plasmid pCYFP28 encoding cyan (CFP) and yellow fluorescent proteins (YFP) separated by multiple restriction sites was provided by Todd W. Geders (from our laboratory), in the pET28 vector with a C-terminal His tag. The annealed product of S-WNV-2B/3-1 and S-WNV-2B/3-2 (supplemental Table 1) encoding the sequence LQYTKR/GGVLWD between BamHI and HindIII restriction sites was ligated into pCYFP28 to generate pWNV-CFP-LQYTKR/GGVLWD-YFP (or simply pWNV-CFP-2B/3-YFP). Constructs encoding other substrates were made similarly using primers listed in supplemental Table 1. The composition of all constructs was verified by DNA sequencing. Gene Expression and Protein Purification—The plasmid, pWNV-NS2B40-G4SG4-HM-NS3FL, was used for high level, inducible expression of N-terminal hexahistidine-tagged recombinant proteins. Cultures of Escherichia coli strain Rosetta2 (Novagen) transformed with the expression plasmids were grown in 1 liter of LB medium containing 35 μg/ml chlor-amphenicol and 50 μg/ml kanamycin at 37 °C until the A600 = 0.5. The temperature was reduced to 18 °C, and expression of the recombinant protein was induced by addition of isopropyl β-d-thiogalactopyranose to a final concentration of 0.4 mm, cultures were incubated for an additional 12 h at 18 °C, and cells were harvested by centrifugation. Cell pellets were resuspended in 30 ml of lysis buffer (25 mm sodium phosphate, pH 6.5, 300 mm NaCl, 20 mm imidazole, 5% glycerol), lysed by three passes through a French press at a pressure of 1000 pascals, and centrifuged at 15,000 rpm for 30 min at 4 °C. The supernatant was loaded onto a 5-ml HiTrap chelating column (GE Healthcare) pre-equilibrated with lysis buffer. The column was washed with 30 ml of Buffer A (25 mm Tris-HCl, pH 8.0, 300 mm NaCl, 5% glycerol) containing 50 mm imidazole. The protein was eluted with a linear gradient of 50-300 mm imidazole in Buffer A. Fractions containing NS2B-NS3 proteins, determined by 12% SDS-PAGE, were pooled and dialyzed against Buffer A, first with and then without 2 mm EDTA, concentrated to 10 mg/ml using Centriprep-50 (Millipore), increased to 15% glycerol, and stored at -20 °C. Yields for all WNV protease variants that included 40 residues of the NS2B cofactor were reproducibly 12 mg of purified protein per liter of E. coli culture. The variant with 79 residues of NS2B was produced in lower yield. Some proteins (using plasmids pWNV-NS2B79-NS3FL, pDV2-NS2B40-G4SG4-ASR-NS3FL, and pYFV-NS2B40-G4SG4-HM-NS3FL) were further purified by anion-exchange chromatography. Proteins were dialyzed in 25 mm Tris-HCl, pH 9.0, 75 mm NaCl and 5% glycerol, loaded into a 1-ml HiTrap Q HP column (GE Healthcare), and eluted with a linear gradient of 75-300 mm NaCl. Protease Assay—For self-cleavage reactions, purified proteins were diluted to the stated concentration (0.25-5.0 mg/ml) in assay buffer (25 mm Hepes, pH 8.5, 50 mm NaCl, 35% glycerol (v/v)), incubated at 37 °C for the indicated times, and quenched by addition of an SDS-PAGE loading buffer to a final concentration of 2% SDS. For intermolecular cleavage reactions, substrate and enzyme were diluted separately to 1 mg/ml in assay buffer, mixed in a 1:4 ratio (enzyme:substrate), unless indicated otherwise, and incubated for the indicated time. Mass Spectrometry Analysis—The matrix-assisted laser desorption ionization-mass spectrometric results were obtained in the Purdue Campus-wide Mass Spectrometry Facility using an Applied Biosystems (Framingham, MA) Voyager DE PRO mass spectrometer, which uses a nitrogen laser (337 nm) for ionization with a time-of-flight mass analyzer. The positive-ion mass spectra were obtained in the linear mode with an accelerating voltage of 25 kV, a grid voltage of 85%, an extraction delay time of 98 ns, and 150 laser shots per spectrum. The matrix was 2,5-dihydroxybenzoic acid (1 mg/ml in 75% ethanol). Equal volumes of the sample and matrix were mixed and air-dried prior to analysis. The acquisition mass range was 500-10,000 daltons with an instrument error of 0.1%. Design and Production of Recombinant Flavivirus NS2B-NS3—Flavivirus protease activity is dependent on the association of NS3 with a cofactor (NS2B40), a central 40-amino acid hydrophilic domain within the largely hydrophobic NS2B protein. To produce active NS2B-NS3 proteases from five flaviviruses, we engineered proteins in which the 40-residue cofactor from NS2B (NS2B40, residues 54-93 of WNV NS2B) was fused to the N terminus of full-length NS3 (NS3FL) by a flexible 9-residue linker followed by a noncleavable dipeptide, Gly4-Ser-Gly4-His-Met (G4SG4-HM) (Fig. 2). A hexahistidine tag was engineered at the N terminus to facilitate purification of NS2B-NS3 protease. Similar recombinant forms of NS2B-NS3 protease were engineered by fusion of the respective NS2B cofactors with NS3 from YFV, DV2, DV4, and JEV. All recombinant proteins were produced in an E. coli expression system at 18 °C. The major proportion of the expressed protein was present in the soluble fractions of cell lysate, indicating that recombinant proteins were likely to be folded correctly. The proteins were purified by immobilized metal affinity chromatography via the His6 tags (in some cases followed by anion-exchange chromatography). SDS-PAGE analysis of purified proteins revealed fragments of lower molecular weight in addition to the expected products (Fig. 3, lanes 2, 5, 8, 11, and 14). These lower molecular weight products increased upon longer incubation at 37 °C (lanes 3, 6, 9, 12, and 15), indicative of proteolysis. However, none of the lower molecular weight fragments was detected in preparations of recombinant proteins in which the protease catalytic serine was substituted with alanine (Fig. 3, lanes 4, 7, 10, 13, and 16) and in which each protein was produced in identical yield to its corresponding parent. Thus, the recombinant NS2B-NS3 protease is subject to autolytic cleavage. At least two sites of autolysis are apparent in most of the NS2B-NS3 proteins, although the five proteases differ in their susceptibility to autolytic cleavage. Further analysis of cleavage was performed using the WNV NS2B40-G4SG4-HM-NS3FL protein.FIGURE 3Analysis of five purified recombinant flavivirus NS2B-NS3 proteases. SDS-PAGE of purified NS2B-NS3 from West Nile virus (lanes 2-4), yellow fever virus (lanes 5-7), dengue virus type 2 (lanes 8-10), dengue virus type 4 (lanes 11-13), and Japanese encephalitis virus (lanes 14-16). For each recombinant protein, the 1st lane is freshly purified protein, the 2nd lane is protein following a 2-h incubation at 37 °C; the 3rd lane is the purified protease-inactive mutant (S135A or S138A). The active and inactive forms of each protease were produced in equivalent yields. Equivalent amounts of total protein were loaded in all lanes of the gel. The predicted size of the intact recombinant protein is 77 kDa. Molecular mass markers are in lane 1. Autolytic cleavage of each of the proteases is apparent.View Large Image Figure ViewerDownload Hi-res image Download (PPT) Sites of Autolytic Cleavage of WNV NS2B-NS3—Fragments of 8, 18.5, 50.5, and 69 kDa were detected in the recombinant NS2B-NS3 protein from WNV, indicative of two cleavage sites in the full-length protein of 77 kDa (Figs. 3 and 4). These sites were cleaved at different rates. One site was cleaved completely prior to purification, yielding fragments of 69 and 8 kDa (Fig. 4A, lane 2). Cleavage of the 69-kDa fragment at a second site was slower, as seen at time points during incubation at 37 °C (Fig. 4A, lanes 2-6). All fragments were associated with the natively folded protein because they did not dissociate upon additional steps of nickel-affinity, anion-exchange, and size-exclusion chromatography of the monomeric protein (supplemental Fig. 1). To identify the site of rapid cleavage, NS2B-NS3 was analyzed by mass spectrometry. A fragment of 7870 ± 8Dawas identified, which corresponds to cleavage at the "noncleavable" junction following the Gly4-Ser-Gly4-His-Met linker between NS2B and NS3 (compared with 7863 Da, calculated mass; data not shown). The cleavage site was also identified by N-terminal sequencing of the 69-kDa fragment, which yielded the sequence Gly-Gly-Val-Leu-Trp-Asp, corresponding to the N terminus of NS3, indicating that cleavage occurs at the site G4SG4HM↓GGVLWD. Mutagenesis at this site, for experiments described below, provided further confirmation of the site identity. We designate this cleavage as an "N-terminal cleavage." Cleavage at this site was unexpected because the sequence at the site, HM↓GG, is unlike the natural cleavage sequences at the critical P1 residue (Met versus Arg, see Fig. 1B), and substitution of Leu for Arg in the P1 position of the YFV NS2B-NS3 protease domain led to no detectable cleavage (29Chambers T.J. Nestorowicz A. Rice C.M. J. Virol. 1995; 69: 1600-1605Crossref PubMed Google Scholar). In addition, autolytic cleavage of a similar NS2B40-fused construct of WNV NS3pro was reported at NS3 residue Lys15 (22Shiryaev S.A. Ratnikov B.I. Chekanov A.V. Sikora S. Rozanov D.V. Godzik A. Wang J. Smith J.W. Huang Z. Lindberg I. Samuel M.A. Diamond M.S. Strongin A.Y. Biochem. J. 2006; 393: 503-511Crossref PubMed Scopus (86) Google Scholar), in a sequence (KK↓G) similar to natural flavivirus protease sites. We anticipated cleavage at Lys15-Gly16 when we detected the 8-kDa fragment, and we engineered a protein in which the nonconserved Gly16 was substituted with Leu. This substitution had no impact on appearance of the 8-kDa fragment (data not shown). We conclude that the Lys15-Gly16 site is less accessible to the protease active site in full-length NS3 than it is in the isolated protease domain (NS3pro). We also attempted to isolate the full-length NS2B-NS3 protein by including protease inhibitors in the lysis buffer for cells from a fresh culture, but the protein was fully cleaved when it emerged from E. coli (supplemental Fig. 2). The second, slower autolytic cleavage was presumed to occur within the helicase domain, where NS3 cleavage has been reported in cells infected with DV2 (10Arias C.F. Preugschat F. Strauss J.H. Virology. 1993; 193: 888-899Crossref PubMed Scopus (155) Google Scholar, 12Teo K.F. Wright P.J. J. Gen. Virol. 1997; 78: 337-341Crossref PubMed Scopus (63) Google Scholar), at a site corresponding to Arg459-Gly460 in WNV NS3. Cleavage at this site was confirmed by N-terminal sequencing of the C-terminal fragment. The resulting sequence, Gly-Arg-Ile-Gly-Arg, corresponded to NS3 residues 460-464. The cleavage site was also confirmed by mutagenesis to produce the G460L substitution. The 8- and 69-kDa fragments were present in the purified proteins, but the 18.5- and 50.5-kDa fragments were not detected for NS2B40-G4SG4-HM-NS3FL/G460L (Fig. 4C, lanes 14-18), confirming that the slower cleavage occurs in the helicase region at Arg459↓Gly460. We designate this as the internal helicase site. No autolytic cleavages were observed with enzymatically inactive NS2B40-G4SG4-HM-NS3FL/S135A (Fig. 4B, lane 8-12). An Artificial Substrate to Study Cleavage Specificity—To examine the unexpected N-terminal cleavage, we created an artificial substrate containing several variants of the natural NS2B-NS3 cleavage sequence (Fig. 1B). In this construct, a dodecapeptide encompassing the natural NS2B/NS3 junction was engineered between two fluorescent proteins, CFP and YFP, to create "CFP-2B/3-YFP" (LQYTKRP1↓GP1′ GVLWD). Proteolytic activity with this substrate was assayed by SDS-PAGE analysis of the reaction mixture, in which substrate (59.3 kDa, Fig. 5, lane 3), enzyme (lane 2), and two products (29.5 and 29.8 kDa, Fig. 5, lane 4) were easily distinguished, or by cleavage-induced loss of fluorescence resonance energy transfer between CFP and YFP (data not shown). We also made several artificial substrates containing substitutions in the NS2B/NS3 junction to check cleavage specificity. The artificial substrate (CFP-LQYTKR↓GGVLWD-YFP) containing the natural sequence of the NS2B/NS3 junction was cleaved efficiently (Fig. 5, lane 4). Substitution of Asp for Arg at the P1 position (Fig. 5, lane 5) or Asp for Gly at the P1′ position (lane 6) abolished the cleavage activity. Two artificial substrates containing HM↓GG, the site that was cleaved rapidly in the full-length enzyme, were also tested. Neither CFP-LQYTHMGGVLWD-YFP, containing the natural NS2B/NS3 junction with KR replaced by HM, nor CFP-GGGGHMGGVLWD-YFP, containing the junction sequence from our recombinant enzyme, was cleaved in the artificial substrate (Fig. 5, lanes 7 and 8). This suggests that either HM↓GG is not a substrate for intermolecular cleavage or that a cleavage-competent conformation of HM↓GG is not accessible in the context of the artificial substrate. The natural cleavage sites (Fig. 1B) for flavivirus NS2B-NS3 proteases generally have two basic residues (Arg-Arg, Arg-Lys or Lys-Arg) at the P2 and P1 positions and a small amino acid (Gly, Ser, or Ala) at the P1′ position. In contrast to the N-terminal HM↓GG cleavage site, the internal helicase cleavage site has a sequence very similar to the natural inter-protein sites (Fig. 1B). We also made an artificial substrate, CFP-SAAQRR↓GRIGRN-YFP, containing the Arg459↓Gly460 cleavage site observed in the recombinant protein. However, cleavage of this artificial substrate was barely detectable (Fig. 5, lane 9). In the context of the artificial substrate, efficient cleavage of the natural NS2B-NS3 site is in striking contrast to the lack of cleavage of the two sites of autolysis, suggesting that the autolytic reactions may occur in cis. Thus we designed materials to test this possibility explicitly. Intramolecular Proteolytic Cleavage of WNV NS2B-NS3—To determine whether the autolytic cleavage is intramolecular (cis) or intermolecular (trans), we designed, expressed, and purified a natural substrate for trans cleavage (Fig. 6A). The protein, NS2B40-G4SG4-HM-NS3FL/S135A, has intact N-terminal and helicase cleavage sites but lacks the protease catalytic side chain (Ser135). NS2B40-G4SG4-HM-NS3FL/S135A undergoes no autolytic cleavage (Fig. 6, lane 3) in 2 h at 37 °C, whereas autolysis by the parent enzyme (NS2B40-G4SG4-HM-NS3FL) results in complete cleavage at the N-terminal site and more than 50% cleavage at the internal helicase site under the same conditions (lane 9). We also designed, expressed, and purified an enzyme lacking both of the N-terminal and internal cleavage sites (NS2B40-G4SG4-HD-NS3FL/G460L). This protein undergoes no autolytic cleavage (Fig. 6, lanes 6 and 7), as expected, but is an efficient enzyme against the artificial substrate CFP-LQYTKR↓GGVLWD-YFP (lanes 17 and 18). To detect intermolecular hydrolysis, the substrate, NS2B40-G4SG4-HM-NS3FL/S135A, was incubated with the enzyme, NS2B40-G4SG4-HD-NS3FL/G460L. We observed no trans cleavage of substrate NS2B40-G4SG4-HM-NS3FL/S135A (Fig. 6, lanes 10 and 11) by enzyme NS2B40-G4SG4-HD-NS3FL/G460L under conditions in which this enzyme completely hydrolyzed the artificial substrate (lanes 17 and 18). These results clearly indicate intramolecular cleavage of both the N-terminal site (GGGGHM↓GGVLWD) and the internal helicase site (SAAQRR↓ GRIGRN) of recombinant NS2B-NS3. The natural cleavage sequence at the NS2B/NS3 junction was also tested for trans cleavage by construction of the substrate NS2B40-G4SG4-KR-NS3FL/S135A. A low level of N-terminal cleavage was seen using the trans substrate (Fig. 6, lanes 12 and 13), but this was far below the rate of cis cleavage of the parent construct (lanes 8 and 9). Further confirmation of cis cleavage was provided by a dilution experiment using purified NS2B40-G4SG4-HM-NS3FL. The rate of cleavage