Cotranslational Membrane Insertion of the Serine Proteinase Precursor NS2B-NS3(Pro) of Dengue Virus Type 2 Is Required for Efficient in Vitro Processing and Is Mediated through the Hydrophobic Regions of NS2B*

Polyprotein processing of dengue virus type 2, a positive strand RNA virus, is carried out by the host signal peptidase and a novel two-component viral proteinase of the serine proteinase family, NS2B/NS3(Pro), in the endoplasmic reticulum. Using an in vitro processing system, we examined the cis andtrans cleavages of the 2B/3 and 4B/5 sites by NS2B/NS3(Pro), respectively. Lysates of BHK-21 cells coexpressing NS2B and NS3(Pro) mediated trans cleavage of the 4B/5 sitein vitro, and the protease activity was associated with the membrane fraction. To study the role of membranes in the protease activity of NS2B/NS3(Pro), labeled precursors, NS2B-NS3(Pro), and the mutant ndNS2B-NS3(Pro) in which the functional hydrophilic domain of NS2B was deleted, were analyzed using a coupled in vitrotranscription/translation system (TnT). The results showed that cotranslational addition of microsomal membranes to the TnT reaction markedly enhanced the cis cleavage of the 2B/3 site in a dose-dependent manner. NS2B synthesized in the presence of membranes also facilitated trans cleavage of the 2B/3 site in the mutant precursor. The cleavage products, NS2B and NS3(Pro), were membrane-associated. Furthermore, this membrane requirement was dictated by the hydrophobic regions of NS2B. Deletion of hydrophobic regions of NS2B, leaving only the conserved hydrophilic domain of 40 amino acids, resulted in highly efficient processing of the 2B-3 sitein vitro in the absence of microsomal membranes.

genomic RNA contains 10,723 nucleotides that contains a single open reading frame encoding a polyprotein precursor of 3391 amino acid residues with a gene order of 5Ј-C-prM-E-NS1-NS2A-NS2B-NS3-NS4A-NS4B-NS5-3Ј (in New Guinea Cstrain (1)). The 5Ј-end of the genomic RNA has a type I cap, and the 3Ј-end is devoid of a poly(A) tail (for review, see Ref. 2). The polyprotein precursor is processed into three structural proteins that are assembled into the virion (C, prM, and E) and at least seven nonstructural proteins, NS1 to NS5, which are expressed in infected cells (2).
The processing of the N-terminal region of the polyprotein precursor that encodes the structural proteins (C, prM, and E) is carried out by the host signal peptidase associated with the endoplasmic reticulum (3)(4)(5). The role of NS3 as the putative viral protease was established subsequent to the identification of a serine proteinase domain within the N-terminal 180 amino acid residues (6,7). Further studies showed that the protease activity of NS3 was dependent on the presence of NS2B. Moreover, NS2B and NS3 form a complex in virus-infected cells (8 -10). This two-component proteinase is required for the rapid cis cleavage of the 2A/2B and 2B/3 sites as well as for trans cleavage of the 3/4A and 4B/5 sites (11)(12)(13)(14)(15)(16)(17). In addition, the viral encoded protease mediates cleavages within the C (18 -20), NS2A (21), NS3 (8,14,22), and NS4A proteins (23). These cleavage sites have the consensus sequence of two amino acids (KR, RR, RK, and occasionally QR) at the Ϫ2 and Ϫ1 positions, followed by Gly, Ala, or Ser at the ϩ1 position (2). The cleavage that converts prM to M protein occurs at a late step during viral morphogenesis and is mediated by a cellular protease located in a post-Golgi acidic compartment of the cell (24).
The goal of the current study was to develop an in vitro processing system for characterization of the role of NS2B in the activation of NS3 protease. When NS2B and NS3 were coexpressed in mammalian cells using a recombinant vaccinia virus expression system, the protease activity that cleaved the NS4B-NS5(93 aa) substrate in vitro partitioned with the membrane fraction. Using the coupled in vitro transcription/translation system, we demonstrated that cotranslational insertion of the NS2B-NS3(Pro) precursor into the endoplasmic reticulum membranes markedly enhanced the NS3-dependent cis and trans cleavages. However, this membrane-dependent enhancement was nullified by deletion of the hydrophobic regions of NS2B. Under these conditions, the cleavage of the 2B/3 site in the precursor occurred with increased efficiency in the absence of membranes.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes and DNA modifying enzymes were purchased from New England Biolabs (Beverly, MA). Rabbit reticulocyte coupled transcription/translation system and the dog pancreatic * This research was supported in part by Grant AI 32078 from the National Institutes of Health and by funds from the Johnson & Johnson Foundation (to R. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Supported in part by a predoctoral fellowship from Kansas Health Foundation.
§ To whom correspondence should be addressed: Dept. of Biochemistry and Molecular Biology, University of Kansas Medical Center, 3901 Rainbow Blvd., Kansas City, KS 66160-7421. Tel.: 913-588-7018; Fax: 913-588-7440; E-mail: rpadmana@kumc.edu. 1 The abbreviations used are: DEN-2, dengue virus type 2; NS, nonstructural protein denoting the virus-specific proteins expressed only in the infected cells as opposed to being a component of the virion; PBS, phosphate-buffered saline; vv, vaccinia virus; PAGE, polyacrylamide gel electrophoresis; PFH, a 27-amino acid region containing a protein kinase A, a synthetic Flag epitope, and a histidine tag fused to the C terminus; Pro, protease domain of NS3; PCR, polymerase chain reaction; aa, amino acid(s); TnT, transcription/translation system; nt, nucleotide(s); YF, yellow fever. microsomal membranes were purchased from Promega (Madison, WI). trans-[ 35 S]Methionine label (1000 Ci/mmol) was purchased from ICN (Costa Mesa, CA). The pET-PFH-nef vector was a gift from L. J. Zhao (see Ref. 25). The monoclonal antibody against the FLAG epitope was from IBI-Kodak (Kodak Scientific Imaging Systems, New Haven, CT). The pTM1 vector was a gift from B. Moss (26). The pJK3 vector was derived from the pTM1 to include the C-terminal 27 amino acid residues encoding the protein kinase A site, FLAG epitope, and hexahistidine tag (PFH), cloned into the polylinker region as described (27).
In Vitro Transcription/Translation of Expression Constructs-The rabbit reticulocyte coupled transcription/translation system (TnT) was used for expression of proteins in vitro. Plasmid DNAs purified by CsCl-ethidium bromide ultracentrifugation (1 g/reaction) were used for the TnT reactions carried out at 30°C for 90 min in a total volume of 25 l. T7 RNA polymerase, trans-[ 35 S]methionine label (20 Ci) and canine pancreatic microsomal membranes (3 l unless otherwise indicated) were added to the TnT reactions. When microsomal membranes were isolated, TnT reactions were centrifuged at 15,000 ϫ g for 15 min, and the isolated membrane pellet was subsequently washed with phosphate-buffered saline (PBS). The TnT reactions were terminated by addition of cycloheximide (0.6 mg/ml), cold methionine (1 mM), and RNase (1 g) and incubation at 30°C for 20 min. To determine the effect of post-translational addition of membranes, the TnT reactions after termination were incubated for 3 h at 30°C.
Transient Expression of Proteins in Vivo-Subconfluent monolayers of HeLa cells (3 ϫ 10 6 in a 25-cm 2 tissue culture flask) were infected with 10 plaque forming units/cell of vTF7-3 and transfected with 5 g of the indicated expression constructs using liposome-mediated DNA uptake. The cells were incubated at 37°C for 4 h after which time fresh media containing fetal bovine serum (10%) was added. Incubation was continued for an additional 15 h at 37°C. Cells were subsequently harvested and washed with PBS, pH 7.4, and lysed with 0.1 ml of sodium dodecyl sulfate (SDS) lysis buffer (65 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 5% 2-mercaptoethanol, and 0.001% bromphenol blue). Proteins were resolved on 10% SDS-polyacrylamide gels, transferred to nitrocellulose membranes, and subjected to immunoblot analysis using the commercially available FLAG monoclonal antibody.
Cell Extracts for in Vitro Processing of the NS4B-NS5 Precursor-BHK-21 monolayers were infected with vTF7-3 and the specified re- combinant vaccinia viruses at 10 plaque-forming units/cell. Approximately 16 h postinfection, the cells were harvested by scraping and washed with PBS. The cell pellet was then resuspended in 50 l of lysis buffer (50 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM EDTA, 0.25 mg/ml cycloheximide, and 0.5% Triton X-100) and incubated on ice for 1 h. The lysate was then centrifuged at 15,000 ϫ g for 20 min, and the resulting pellet was resuspended in 50 l of the above buffer.
For trans cleavage assays, 25 l of the soluble or the resuspended pellet fraction obtained by centrifugation was incubated with the NS4B-NS5(93 aa) precursor. The NS4B-NS5(93 aa) precursor was synthesized in the presence and absence of microsomal membranes as indicated. Reactions were incubated at 30°C for 3 h and subsequently terminated by addition of the SDS loading buffer.
Extraction of Proteins from Membranes by Treatment with pH 11.5 Buffer-Microsomal membranes were isolated by centrifugation at 15,000 ϫ g for 15 min following TnT and were washed once with PBS, pH 7.2. The washed membrane pellets were then resuspended in either 80 l of PBS, pH 7.2, or 80 l of 100 mM sodium carbonate, pH 11.5, and incubated on ice for 30 min. Membranes were then pelleted by centrifugation at 200,000 ϫ g for 1 h at 4°C. Proteins associated with the membrane pellet and supernatant fractions were analyzed by SDS-PAGE followed by autoradiography of the gel (Fujiki et al. (31)).
Analysis of membrane-associated NS3(Pro)-PFH and NS2B-PFH expressed from recombinant vaccinia viruses in BHK-21 cells was carried out as follows. Cell pellets (from 6-well plates) were resuspended in 80 l of H buffer (20 mM HEPES, pH 7.5, 5 mM KCl, 0.5 mM MgCl 2 , 0.5 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride) and lysed by passage through a 27.5-gauge needle. The lysate was subjected to centrifugation at 200,000 ϫ g for 1 h at 4°C. The soluble fraction was saved. The pellet was resuspended in 80 l of 100 mM sodium carbonate, pH 11.5, and sonicated (4 ϫ 30-s bursts in ice) using a cup-horn sonicator (Heat Systems Ultrasonics, model 185 F) at power setting of 7. The lysate was centrifuged as before to obtain the soluble and pellet fractions. Proteins from each fraction were analyzed by SDS-PAGE followed by Western blot analysis using the FLAG monoclonal antibody.

RESULTS
In Vitro Processing of the NS4B-NS5 Precursor-The goal of this study was to develop an in vitro processing system for biochemical characterization of the viral encoded NS2B/NS3 protease. The initial strategy for developing this in vitro processing system was to express NS2B and the N-terminal protease domain of NS3 (NS3(Pro)) using the recombinant vaccinia virus expression system. To facilitate purification of NS2B and NS3(Pro), the coding sequences of both proteins were modified by a C-terminal fusion of 27 amino acid residues (PFH) consisting of a protein kinase A phosphorylation site, a synthetic FLAG epitope recognized by a monoclonal antibody, and a hexahistidine tag (25). First, we verified that this C-terminal modification of NS2B and NS3 did not affect their biological activity in the processing of NS3-NS4A-NS4B-NS5 polyprotein substrate in vivo. When the protease components and the substrate were coexpressed in vivo, the efficiency of processing was similar to that of unmodified NS2B and NS3(Pro) reported earlier (16) (data not shown).
To establish an in vitro processing system that can carry out trans cleavages, BHK-21 cells were infected with recombinant vaccinia viruses encoding the T7 RNA polymerase (vTF7-3), NS2B-PFH, and NS3(Pro)-PFH. NS2B and NS3 coding sequences in these recombinant viruses are under the control of T7 promoter/encephalomyocarditis virus 5Ј-untranslated region (30) which requires T7 RNA polymerase (from vTF7-3 infection) for expression. Infected cells were then treated with Triton X-100 lysis buffer, and the lysates were fractionated into cytoplasmic and membrane fractions by centrifugation (15,000 ϫ g). NS2B-PFH and NS3(Pro)-PFH fractionated into the pellet fraction ( Fig. 2A, lanes 1-3). The pellet fraction was resuspended in a buffer containing Triton X-100 and assayed for protease activity using the substrate NS4B-NS5(93 aa)-PFH labeled with [ 35 S]methionine in the TnT reaction. Processing of the NS4B-NS5(93 aa)-PFH substrate was observed only when the substrate was incubated with the membraneassociated pellet fraction but not with the supernatant fraction (Fig. 2B, compare lanes 3 and 4).
In this experiment the detection of NS5(93 aa) was obscured by the endogenously produced globin. Since NS4B is a membrane-associated protein, the NS4B-NS5(93 aa)-PFH precursor was synthesized in vitro in the presence of microsomal membranes. In this manner, it was possible to make the NS4B-NS5(93 aa)-PFH substrate associated with the membranes and FIG. 2. In vitro processing of the NS4B-5 substrate. A, BHK-21 cells were infected with the recombinant vaccinia viruses, vTF7-3 (encoding T7 RNA polymerase) alone or a mixture of vTF7-3, vvNS2B-PFH, and vvNS3(Pro)-PFH, and extracts were prepared using a Triton X-100 lysis buffer. The lysates were fractionated into pellet (P) and soluble (S) fractions by centrifugation as described under "Experimental Procedures." Each fraction was analyzed by SDS-PAGE followed by Western blot analysis using the FLAG monoclonal antibody. Lane 3, pellet fraction from vTF7-3-infected cells. Lanes 1 and 2, soluble and pellet fractions from cells infected with a mixture of vTF7-3, NS2B-PFH, and NS3(Pro)-PFH recombinant vaccinia viruses. B, 1 ϫ 10 6 cells were infected with vTF7-3 alone or a mixture of vTF7-3, NS2B-PFH, and NS3(Pro)-PFH recombinant vaccinia viruses, and the soluble and membrane pellet fractions were prepared and resuspended in 50 l. The NS4B-5(93 aa)-PFH substrate was expressed in the TnT system in the absence of microsomal membranes. For in vitro processing reactions, 25 l of the soluble lysate or 25 l of resuspended pellet fraction was incubated at 30°C for 3 h with 1.5 l of the labeled NS4B-5(93)-PFH substrate. The products were analyzed by SDS-PAGE and autoradiography. Lanes 1 and 2, soluble and pellet fractions from vTF7-3-infected cell lysates, respectively. Lanes 3 and 4, soluble and pellet fractions of lysates, respectively, prepared from cells infected with a mixture of vTF7-3 ϩ vvNS2B-PFH ϩ vvNS3(Pro)-PFH. C, the NS4B-NS5(93 aa)-PFH substrate was expressed in the TnT system supplemented with microsomal membranes, and the membrane pellet fraction was isolated as described under "Experimental Procedures." It was incubated with the pellet fraction from vTF7-3-or vvNS2B ϩ vvNS3(Pro)coinfected cell lysates as described in B. The products of processing were analyzed by SDS-PAGE and autoradiography. Lanes 1 and 2, membrane pellet fractions from BHK-21 cells infected with vTF7-3 alone or a mixture of vTF7-3, vvNS2B-PFH, and vvNS3(Pro)-PFH. remove the endogenous globin by centrifugation (15,000 ϫ g). This substrate was incubated with the microsomal pellet fraction isolated from cells expressing the NS2B and NS3(Pro) or from the vTF7-3-infected cells as control. As shown in Fig. 2C, processing of the NS4B-NS5(93 aa)-PFH substrate was observed by the membrane-associated NS2B-PFH and NS3(Pro)-PFH. These results taken together indicated that NS2B-PFH and NS3(Pro)-PFH are associated with the membrane fraction of the infected cell lysates and are active in trans cleavage of the 4B/5 site.
The Role of Microsomal Membranes in the Protease Activity of the NS2B/NS3(Pro) Components-The association of the protease activity of NS2B-PFH and NS3(Pro)-PFH with the membrane pellet fraction of recombinant vaccinia virus-infected BHK-21 cell lysates could be explained by two possible scenarios. First, overexpression of proteins in the recombinant vaccinia virus expression system could result in aggregation and partitioning of the protein into the membrane pellet fraction. Second, NS2B-PFH and NS3(Pro)-PFH have intrinsic properties for association with the membranes, and this membrane association is required for optimal protease activity. To investigate these possibilities, the cis cleavage of the 2B/3 site was examined using an expression plasmid encoding the NS2B-NS3(Pro)-PFH precursor.
The NS2B-NS3(Pro)-PFH precursor was expressed in HeLa cells by plasmid transfection followed by infection with vTF7-3 to provide T7 RNA polymerase (30). The processing was monitored by Western blot analysis using the FLAG monoclonal antibody. As shown in Fig. 3A (lane 2), the processing of NS2B-NS3(Pro)-PFH precursor in vivo was very efficient as little unprocessed precursor was detected.
Expression of the NS2B-NS3(Pro)-PFH precursor in vitro using the TnT reaction yielded markedly different results. In contrast to the efficient processing of the NS2B-NS3(Pro)-PFH precursor observed in vivo, only a trace amount of precursor underwent processing in vitro in the absence of microsomal membranes. Therefore, the TnT reactions were conducted in the presence or absence of canine pancreatic microsomal membranes to determine whether membranes influenced the cleavage efficiency. As shown in Fig. 3B, only a faint band was observed in the 36-kDa size range which is the expected size of NS3(Pro)-PFH liberated by cis cleavage of the 2B/3 junction (Fig. 3B, lane 2). However, in vitro processing assays conducted in the presence of microsomal membranes significantly enhanced the efficiency of cleavage of the 2B/3 site (Fig. 3B, lane  3). The possibility that a contaminating protease in the microsomal membrane preparation was cleaving the NS2B-NS3(Pro)-PFH precursor was considered. To eliminate this possibility the precursor NS2B-NS3*(Pro) containing a His 51 3 Ala mutation within the catalytic triad of NS3 was translated in the presence of microsomal membranes. Expression of the NS2B-NS3*(Pro) precursor in the presence of microsomal membranes did not result in any detectable cleavage of the precursor (Fig. 3B, lane 1). These results indicated that processing of the NS2B-NS3(Pro)-PFH precursor is markedly stimulated on microsomal membranes and the wild type protease domain of NS3.
When total lysates were resolved by SDS-PAGE, NS2B was obscured by co-migration of endogenously produced globin. To definitively identify both NS2B and NS3(Pro)-PFH, the lysates were subjected to immunoprecipitation using a mixture of anti-NS2B and anti-NS3 antibodies (Fig. 3C, lanes 1-3) or anti-NS3 alone (Fig. 3C, lanes 4 -6). Analysis of the immunoprecipitates by SDS-PAGE revealed the identity of both proteins. Immunoprecipitates from translation reactions in which the NS2B-NS3(Pro)-PFH precursor was expressed in the presence of membranes resulted in increased amounts of immunoprecipitated NS2B and NS3(Pro)-PFH. Furthermore, NS2B and NS3(Pro) were not detected in the immunoprecipitates of TnT reactions in which the NS2B-NS3*(Pro) precursor containing FIG. 3. In vivo and in vitro processing of the NS2B-3(Pro)-PFH precursor. A, HeLa cells were infected with recombinant vaccinia virus, vTF7-3 (to provide T7 RNA polymerase) and transfected with pTM1-NS2B-NS3(Pro)-PFH plasmid as described under "Experimental Procedures." The total cell extracts were analyzed by SDS-PAGE followed by Western blot analysis using the FLAG monoclonal antibody. Lanes 1 and 2, extracts were prepared from cells infected with vTF7-3 alone (control) or subjected to "infection and transfection" with vTF7-3 and the plasmid. B, NS2B-NS3(Pro)-PFH or NS2B-NS3*(Pro) (which contains the His 3 Ala mutation in the catalytic site) was expressed in vitro using the TnT reaction. Incubations with or without canine pancreatic microsomal membranes at 30°C for 90 min are as indicated. The products of processing were analyzed by SDS-PAGE and autoradiography. C, in vitro processing of the precursors were carried out as described in B. Reaction mixtures were then immunoprecipitated with either a mixture of anti-NS2B and anti-NS3 antibodies (lanes 1-3) or anti-NS3 antibody (lanes 4 -6). Lane 1, NS2B-NS3*(Pro) (ϩ) microsomal membranes. Lane 2, NS2B-3(Pro)-PFH (Ϫ) microsomal membranes. Lane 3, NS2B-3(Pro)-PFH (ϩ) microsomal membranes. Lanes 4 -6, same order as lanes 1-3. the His 51 3 Ala mutation was expressed (Fig. 3C, lanes 1 and 4).

NS2B and NS3(Pro)-PFH Generated by the Cis Cleavage of the 2B/3 Site in the TnT Reaction Are Membrane-associated-
The results obtained thus far established that processing of the NS2B-NS3(Pro)-PFH precursor is markedly enhanced by microsomal membranes. Next, it was examined whether the precursor protein or the products of processing were associated with the membranes. The NS2B-NS3*(Pro) and NS2B-NS3(Pro)-PFH precursors were expressed in the TnT reaction in the presence and absence of microsomal membranes. The reactions were then subjected to centrifugation (15,000 ϫ g) to recover the microsomal membranes. As shown in Fig. 4, in the absence of microsomal membranes both the NS2B-NS3*(Pro) and NS2B-NS3(Pro)-PFH precursors were recovered predominantly in the supernatant fraction (Fig. 4, compare lanes 1 and  5, 3 and 7, respectively). Only a faint band was observed in the 36-kDa region in the NS2B-NS3(Pro)-PFH reactions indicating that very little processing of the precursor occurred in the absence of microsomal membranes confirming the results shown in Fig. 3 (Fig. 4, lanes 3 and 7).
When the NS2B-NS3*(Pro) and NS2B-NS3(Pro)-PFH precursors were translated in the presence of microsomal membranes, polypeptides corresponding to the unprocessed precursors were recovered from both the supernatant and membrane pellet fractions (Fig. 4, lanes 2 and 6, 4 and 8, respectively). However, the cleavage products, NS3(Pro)-PFH and NS2B, were found predominantly in the membrane pellet fraction (Fig. 4, compare lane 4 and 8). Furthermore, both NS3(Pro)-PFH and NS2B were clearly resolved when the membrane pellet fractions were analyzed.
Analysis of the membrane pellet from reactions expressing the NS2B-NS3*(Pro) precursor revealed only a band corresponding to the unprocessed precursor. Thus, in the absence of a functional protease domain, the precursor was still associated with the membrane, but the precursor did not undergo processing (Fig. 4, lane 6), suggesting that the increased processing efficiency of the NS2B-NS3(Pro)-PFH precursor in membranesupplemented reactions was not the result of a contaminating protease.

In Vitro Processing Efficiency of the NS2B-NS3(Pro)-PFH Precursor Increased with Increasing Amounts of Microsomal
Membranes-The previous experiments clearly showed that microsomal membranes increased the processing efficiency of the NS2B-NS3(Pro)-PFH precursor and that proteins liberated during processing were essentially membrane-associated. The presence of a large amount of unprocessed precursor suggested that microsomal membranes were limiting in the TnT reaction. To examine this possibility, the NS2B-NS3(Pro)-PFH precursor was expressed in the TnT reaction in the presence of increasing amounts of microsomal membranes, and both the total lysate and the membrane pellet fractions obtained by centrifugation were analyzed. Analysis of the total lysates revealed that processing efficiency increased with increasing amounts of microsomal membranes added to the TnT reactions as shown by the increasing amounts of NS3(Pro)-PFH produced (Fig. 5, lanes  1-4). Detection of NS2B in total lysates was obscured by comigration of endogenously produced globin (Fig. 5, lanes 1-4). Therefore, aliquots of total cell lysates were fractionated into soluble and membrane pellet fractions by centrifugation (15,000 ϫ g). Analysis of the microsomal pellet fractions showed that increasing amounts membranes added cotranslationally proportionately increased the cleavage products NS3(Pro)-PFH and NS2B (Fig. 5, lanes 5-8).
Microsomal Membranes Are Required Cotranslationally for Increased Processing Efficiency of the NS2B-NS3(Pro)-PFH Precursor-Next, we sought to determine whether the membrane-dependent enhancement of the processing of NS2B-NS3(Pro)-PFH precursor is a cotranslational or post-translational event. To address this question, the NS2B-NS3(Pro)-PFH precursor was expressed using the TnT reaction in the absence of microsomal membranes. RNase, cycloheximide, and cold methionine were added to terminate the TnT reaction. To one-half of the TnT reaction microsomal membranes were added post-translationally, and the reactions were incubated for an additional 3 h at 30°C. As a positive control, the NS2B-NS3(Pro)-PFH precursor was expressed cotranslationally with microsomal membranes, and following termination of the TnT reaction under the same conditions incubation was continued for 3 h at 30°C (Fig. 6, lane 3). The reaction mixtures were analyzed by SDS-PAGE and autoradiography. Addition of microsomal membranes post-translationally failed to enhance cleavage of the NS2B-NS3(Pro)-PFH precursor (Fig. 6, compare  lanes 1 and 2). These data suggested that microsomal membranes must be present cotranslationally for increased processing efficiency of the NS2B-NS3(Pro)-PFH precursor.
Analysis of the Nature of Membrane Association of NS2B and NS3(Pro)-Further analysis of the cotranslational requirement of membranes for efficient processing of NS2B-NS3(Pro) precursor was carried out to establish whether NS2B or NS3(Pro)-PFH or both are integral or peripheral membrane proteins. One of the methods used to distinguish between integral versus peripheral membrane association is by treatment of membrane fractions containing the protein(s) of interest with high pH (pH Ն 11.0) (31). After the in vitro processing of the NS2B-NS3(Pro)-PFH precursor by TnT reactions was carried out in the presence of microsomal membranes, the mem-

FIG. 4. In vitro processing of the NS2B-NS3(Pro)-PFH and NS2B-NS3*(Pro) precursors: fractionation and analysis of cleavage products.
NS2B-NS3(Pro)-PFH and NS2B-NS3* (Pro) precursors were expressed in vitro using the TnT system in the presence and absence of microsomal membranes as indicated. The reaction mixtures were centrifuged at 15,000 ϫ g for 20 min, and the supernatant and membrane pellet fractions were analyzed by SDS-PAGE followed by autoradiography. brane pellet fractions containing the unprocessed and processed products were isolated. Equal amounts of pellet fractions were treated either with PBS, pH 7.4, or with sodium carbonate buffer, pH 11.5, followed by centrifugation at 200,000 ϫ g. The pellet and soluble fractions obtained after these treatments were analyzed by SDS-PAGE and autoradiography (Fig. 7A). The results shown in Fig. 7A indicate that under the conditions of extraction with PBS, pH 7.4, the unprocessed precursor and the processed products were exclusively associated with the membrane pellet fraction (lane 2) and none were solubilized (lane 1). However, the treatment of the membrane pellet fraction with a high pH buffer (pH 11.5) resulted in solubilization of a small fraction of NS2B (lane 3). The majority of the precursor and the products still remained associated with the pelleted membranes (lane 4). The association of NS2B with the membranes was more stable to high pH treatment than that of NS3, suggesting that NS2B behaves more like an integral membrane protein. The observation that both NS2B and NS3(Pro) produced in the in vitro processing reaction were associated with the membranes was independently confirmed by expressing NS2B-PFH and NS3(Pro)-PFH individually using the recombinant vaccinia virus expression system. BHK-21 cells were coinfected with vTF7-3 (to provide T7 RNA polymerase) and vvNS2B-PFH or vvNS3(Pro)-PFH. The membrane pellet fractions isolated from the infected cell lysates were treated with sodium carbonate buffer, pH 11.5. Subsequently, the portions of NS2B and NS3(Pro)-PFH that were solubilized and those remained in the pellet fractions were analyzed by SDS-PAGE and Western blot. The results shown in Fig. 7, B and C, indicate that major portions of both NS3(Pro)-PFH and NS2B-PFH still remained associated with the membrane pellet fractions after the pH 11.5 treatment (Fig.  7B, compare lanes 1 and 2 versus C, lanes 2 and 3). These results independently confirmed the results shown in Fig. 7A.
In Vitro Processing of the ndNS2B-NS3(Pro)-PFH Precursor by Trans Supply of NS2B-The results of this study showed that membranes were required for efficient cis cleavage of the precursor NS2B-NS3(Pro)-PFH and that both NS2B and NS3(Pro)-PFH were associated with the membranes during the course of the in vitro TnT. However, these results did not show how membranes were influencing the functionality of NS2B or NS3(Pro)-PFH or both proteins. To examine the contribution of membrane association of NS2B independent of NS3(Pro) to the processing efficiency, the ndNS2B-NS3(Pro)-PFH precursor was constructed. In this construct the functional domain of NS2B was deleted requiring addition of the wild type NS2B for the cleavage of the 2B/3 site. The in vitro expression of the ndNS2B-NS3(Pro)-PFH alone in the presence of microsomal membranes did not yield detectable cleavage at the 2B/3 site as expected (Fig. 8, lane 3). This result is consistent with the in vivo studies of mutational analysis of NS2B in the processing of precursor protein (17). Coexpression of the ndNS2B-NS3(Pro)-PFH precursor with NS2B in the absence of microsomal membranes (Fig. 8, lane 5) also did not yield detectable cleavage at the 2B/3 cleavage junction confirming the membrane requirement established in this study (Fig. 3B). When the NS2B was translated in the presence of membranes and the membraneassociated NS2B was incubated with the ndNS2B-NS3(Pro)-PFH substrate expressed in the absence of membranes, no FIG. 5. Dose-dependent enhancement of in vitro processing of the NS2B-3(Pro)-PFH precursor by microsomal membranes. The TnT reactions set up for expression of the NS2B-NS3(Pro)-PFH and NS2B-NS3*(Pro) precursors in vitro were supplemented with increasing amounts of microsomal membranes as indicated. Following the 90-min reaction, an aliquot of each reaction was collected, and the remaining reaction mixture was centrifuged at 15,000 ϫ g for 20 min. Total TnT lysate (aliquots saved before centrifugation; see lanes [1][2][3][4] and the membrane pellet fractions (lanes 5-8) were analyzed by SDS-PAGE followed by autoradiography. processing of the substrate was observed (data not shown). However, if NS2B and the ndNS2B-NS3(Pro)-PFH precursor were coexpressed in the TnT reaction in the presence of microsomal membranes, processing of the 2B/3 cleavage junction was restored (Fig. 8, lane 4). Additionally, processing of the ndNS2B-NS3(Pro)-PFH precursor was also observed if NS2B was first expressed alone in the presence of microsomal membranes, and the membrane fraction from this TnT reaction was used in a second TnT reaction programmed for expression of the ndNS2B-NS3(Pro)-PFH substrate without additional membranes added (Fig. 8, lane 2). The 36-kDa band as a cleavage product generated in the presence of microsomal membranes was specific as it was not observed in the control reaction where NS2B or ndNS2B-NS3(Pro)-PFH were expressed individually (Fig. 8, lanes 1 and 3, respectively). These results indicated that insertion of both NS2B and the substrate ndNS2B-NS3(Pro)-PFH into the membranes was required for processing and that the membrane-associated NS2B was able to activate the NS3(Pro) domain in the processing of the 2B/3 site.
Membrane Enhancement of NS2B-mediated Activation of the NS3(Pro) Domain Is Nullified by Deletion of Hydrophobic Regions of NS2B-Since NS2B was associated with the mem-branes cotranslationally, we sought to determine whether the hydrophobic regions of NS2B played any role in the membranedependent enhancement of the activity of NS2B. To address this question, the expression clone pTM1-NS2B(H)-NS3(Pro)-PFH was constructed (Fig. 1F) in which three hydrophobic regions flanking the hydrophilic domain of 40 amino acid residues of NS2B were deleted. To establish that processing is due to the NS3 protease domain, another clone pTM1-NS2B(H)-NS3*(Pro) was also constructed in which the NS2B(H) has the same deletion of hydrophobic sequences as in Fig. 1F, but NS3(Pro)-PFH has the His 3 Ala mutation in the catalytic site of the protease domain (Fig. 1G). The processing of this precursor polypeptide was examined using the in vitro TnT reaction in the absence of microsomal membranes. The results shown in Fig. 9 indicate that deletion of the hydrophobic domains from NS2B (lanes 2 and 5) relieves the membrane requirement for its function in the activation of NS3(Pro)-PFH activity and that the 2B/3 site was processed very efficiently leaving very little unprocessed precursor. Lane 3 shows the migration of authentic NS3(Pro)-PFH used as a control. Lanes 4 and 5 show the same lysates used in lanes 1 and 2 but after immunoprecipitation with rabbit polyclonal anti-NS3 antibod- ies. The origin of the doublet in lane 2 is unknown. The results shown in Fig. 9 taken together indicate that the hydrophilic domain of NS2B alone is sufficient for in vitro cis cleavage of the 2B/3 site in the absence of microsomal membranes.

DISCUSSION
Flavivirus NS2B/NS3 proteinases belong to the class of twocomponent proteinases that includes the adenovirus cysteine proteinase (32), hepatitis C virus serine proteinase (33), and the mitochondrial processing peptidase of Neurospora crassa (34). However, flavivirus NS2B/NS3 proteinases bear some resemblance to the ␣-lytic serine proteinase (35) and the serine proteinase subtilisin (36) as well. The propeptide of subtilisin and ␣-lytic proteinases is essential for production of active enzyme in vivo and is autoprocessed from the precursor of the mature protease by a cis cleavage. In vitro experiments demonstrated that the propeptides need not be physically linked to the proteinase domain. It was suggested that they function as an intramolecular chaperone assisting in the folding of the proteinase domain (35)(36)(37). The requirement of flavivirus NS2B for generating an active NS3 protease in vivo, the physical link of NS2B to the protease domain of NS3 (N-terminal to the NS3 protease), and the property of NS2B to activate NS3 in trans are analogous to the subtilisin and ␣-lytic proteinase systems. However, how NS2B activates the protease domain of flavivirus NS3 is currently unknown.
The hydrophobicity plot of NS2B using the Kyte-Doolittle program (38) shows that NS2B contains a central hydrophilic domain flanked by two hydrophobic domains at the N terminus (I and II) and a single hydrophobic domain (III) at the C terminus of NS2B followed by a 10-aa region upstream of the 2B/3 cleavage site (Fig. 10). The central hydrophilic region contains 40 amino acids which is conserved among flaviviruses. Deletion analysis of the NS2B region in DEN-4-encoded NS2B-NS3(Pro) precursor and transient expression using the recombinant vaccinia virus expression system in vivo revealed that this hydrophilic region is required for activation of protease activity (17). Results of those studies showed that the 40-aa conserved hydrophilic region of NS2B is the only region required for the function of NS2B, which led the authors to conclude that membrane association of DEN-4 NS2B or NS3 may not be required for protease activity (17). However, that study was focused on analysis of in vivo processing of the wild type and mutant NS2B-NS3(Pro) precursors. Therefore, the effect of membranes in the in vitro processing efficiency of the NS2B-NS3(Pro) was not addressed.
The results of this study showed that microsomal membranes contributed significantly in enhancing the biological activity of DEN-2 NS2B in the activation of NS3 protease in mediating cis and trans cleavages of the 2B/3 and 4B/5 sites in vitro. This membrane association of NS2B and its activation of NS3 protease was a cotranslational event and was stable to pH 11.5 treatment. These results suggest that membrane-associated NS2B may influence the activity of the protease complex by allowing the interaction of the hydrophilic domain of NS2B with the NS3(Pro) domain through a conformational change. Analysis of fractionated lysates of TnT reactions support this view in that the products of cis cleavage of the 2B/3 site, NS2B and NS3(Pro)-PFH, were predominantly associated with the membranes.
Our results show that the mutant NS2B in which the hydrophobic regions were deleted was the most efficient in the activation of the NS3(Pro) domain and the resultant cis cleavage of the 2B/3 site. This finding implicates a role for the hydrophobic domains in stipulation of the membrane requirement for the wild type NS2B in the activation of the NS3 protease. This is further supported by the results shown in Fig. 8. Only when both NS2B and ndNS2B-NS3(Pro) precursor were cotranslationally inserted into membranes were they active in the cleavage of the 2B/3 site. This precursor could not undergo cis cleavage of the 2B/3 site because of the deletion of the hydrophilic domain which is required for NS2B activity. It should be noted that the ndNS2B-NS3(Pro)-PFH precursor contains the entire hydrophobic domain III (Fig. 10). Even in the presence of membranes the efficiency of cleavage of the 2B/3 site in the ndNS2B-NS3(Pro)-PFH precursor by trans supply of NS2B was significantly reduced when compared with the cis cleavage of the 2B/3 site in the NS2B-NS3(Pro) precursor.
It should be noted that the membrane requirement for NS2B function in the activation of NS3 protease is not conserved in different flaviviruses. In vitro processing studies of tick-borne encephalitis virus and yellow fever virus (YF) revealed that the FIG. 9. Effect of membranes in the enhancement of NS2B activity is nullified by deletion of hydrophobic regions of NS2B. The NS2B(H)-NS3(Pro)-PFH and NS2B(H)-3*(Pro)-PFH precursors were expressed using the TnT system in vitro at 30°C for 90 min in the absence of microsomal membranes. H represents the hydrophilic domain, and the 3* refer to the His 3 Ala mutation in the catalytic site. Both the total lysates (lanes 1 and 2) and immunoprecipitated lysates (lanes 4 and 5) using anti-NS3 antibody were analyzed. Lane 3 is the NS3(Pro) expressed in vitro as a control.
FIG. 10. Kyte-Doolittle hydrophobicity plot of NS2B. The amino acid sequence of NS2B is shown beneath the hydrophobicity plot derived using the Kyte-Doolittle program (38). Positive values represent hydrophobic regions, and the negative values represent hydrophilic regions. I, II, and III refer to the hydrophobic regions in NS2B, and the underlined region indicate the conserved hydrophilic domain.
NS2B-NS3 precursors undergo processing in vitro in the absence of membranes (11,39,40). However, the in vitro processing of a precursor polyprotein NS2A-NS2B-NS3 encoded by West Nile virus requires microsomal membranes (15,42). In these studies whether the membranes were targeting the NS2A or NS2B/NS3 components for efficient processing was not established. In this regard it is worth noting that NS2A itself is an integral membrane protein (43).
It is possible that the conformation of the DEN-2 NS2B-NS3(Pro) precursor might be different from that of YF precursor which does not require membranes. In support of this view, pulse-chase analysis of YF and DEN-2 precursors revealed that YF precursors were processed first at the 2B/3 site followed by the 2A/2B site, whereas in the DEN-2 precursor, cleavage of the 2A/2B site preceded the cleavage of the 2B/3 site (11,12).
The results of this study showed that even though the hydrophilic domain is the only region required for processing at 2B/3 site in vitro, the three hydrophobic regions of the wild type NS2B are likely to be involved in anchoring the NS2B into membranes in vivo. This membrane insertion could facilitate the interaction of the hydrophilic region with the NS3 protease domain. Moreover, another possible function of the membraneanchored NS2B via its hydrophobic regions is the localization of NS3 to the membranes through protein-protein interaction. Thus the membrane-localized NS2B⅐NS3 complex could additionally recruit NS5 to the membranes as part of an RNA replicase complex where RNA replication process is localized (2,44). NS3 is a multifunctional protein that contains the protease and the NTPase/RNA helicase domains. It was previously reported that NS3 and NS5 exist as a complex in extracts from DEN-2-infected cells (27). Additional support for the notion that the hydrophobic regions of NS2B may play a role in viral replication comes from the evidence that small deletions in both the N-and C-terminal hydrophobic domains of YF NS2B when introduced in an infectious clone were found to be deleterious for viral replication (9). Thus, the hydrophobic regions of NS2B, which are conserved in position but not in amino acid sequence in different flaviviruses, may play a role in viral replication.