VIROLOGY

185,689-697

Processing

(1991)

of Nonstructural

Proteins

NS4A and NS4B of Dengue 2 Virus

FRANK PREUGSCHAT Division

of Biology,

156-29, Received

California July

AND

Institute

in Vitro

and

in Viva

JAMES H. STRAUSS’ of Technology,

16, 199 1; accepted

August

Pasadena,

California

9 1125

2 1. 199 1

The production, from polyprotein precursors, of two hydrophobic nonstructural proteins of dengue 2 (DEN2) virus, NS4A and NS4B. was analyzed both in cell-free systems and in infected cells. In DENa-infected cells, NS4B is first produced as a peptide of apparent size 30 kDa; NS4B is then post-translationally modified, in an unknown way, to produce a polypeptide of apparent size 28 kDa. The rate and extent of NS4B modification was found to be cell-dependent; in BHK cells the half-time for the conversion of the 30-kDa form to the 28-kDa form was 90 min. N-terminal sequence analysis of NS4B suggests that the N-terminus is produced by an enzyme with a specificity similar to that of signalase. Low levels of a putative polyprotein, NS4AB, were also found in mammalian cells, but not mosquito cells, infected with DENP, suggesting that a small proportion of DEN2 4A/4B cleavage can occur post-translationally or that some nonstructural polyproteins escape normal processing. Cleavage of the 4AI4B bond in infected cells required expression of DEN2 sequences in addition to those in NS4A and NS4B, as NS4AB produced in cells by a vaccinia expression system was not cleaved. NS4AB produced in cells by a vaccinia expression system was modified posttranslationally, presumably in the same way as NS4B. We show that upon translation of DEN2 polyproteins in a cell-free system, the N-terminus of NS4A is generated by cleavage by the viral nonstructural proteinase NS3 and that processing of DEN2 polyproteins occurs with a preferred, but nonobligatory order. 0 1991 Academic Press, Inc.

structural proteins. Specific biochemical roles for NS4A and NS4B have not yet been demonstrated, but they could be involved in the establishment of membrane-bound replication complexes (Rice et a/., 1986), as flavivirus RNA replication occurs in close association with host cell membranes (Stohlman et a/., 1975). The flavivirus genomic RNA is the only messenger in infected cells and is translated into a large polyprotein in the order C-prM(M)-E-NSl -NS2A-NS2B-NS3-NS4ANS4B-NS5. This polyprotein is both co- and post-translationally processed by the viral NS3 proteinase and host cell proteinases into at least the 10 different polypeptides described above (reviewed in Chambers et a/., 199Oc). Host cell signal peptidase (signalase), which usually cleaves cotranslationally and functions in the lumen of the endoplasmic reticulum, has been implicated in the processing of the structural proteins, including the cleavage to produce the N-terminus of NSl , both from cell-free translation studies and from considerations of the nature of the cleavage site. The viral NS3 proteinase, which presumably functions in the cytosol, has been shown to generate the N-termini of the nonstructural proteins NS2B and NS3, and may cleave the polyprotein to produce the N-termini of NS4A and NS5 as well (Preugschat et al., 1990; Chambers et a/., 1990b; Rice and Strauss, 1990). The proteinase that separates NSl from NS2A has been more difficult to identify. Although this cleavage site has characteristics of a signalase-like site, there is no upstream hydrophobic amino acid sequence and, fur-

INTRODUCTION The family Flaviviridae contains about 70 positivestranded RNA viruses, many of which are important human or veterinary pathogens (Monath, 1986; Strauss et a/., 1991). Flavivirus virions are enveloped and contain three protein species and a single RNA molecule. The major structural protein of the virion is a membrane-bound envelope (E) protein that is capable of eliciting neutralizing antibodies and is responsible for hemagglutination and cell fusion activities. The membrane (M) protein is a small, hydrophobic, integral-membrane protein whose role in virion structure is not well understood. A small, basic capsid (C) protein complexes with the - 11 -kb genomic RNA to form a nucleocapsid (reviewed in Rice et al., 1986). At least seven nonstructural proteins are produced in infected cells. NS3 is a proteinase (Preugschat eta/., 1990; Chambers et al., 1990b) and a putative RNA-dependent RNA helicase (Gorbalenya ef a/., 1989) and NS5 contains motifs characteristic of RNA polymerases (Rice et al., 1986). NSl is a glycoprotein of unknown function but may be required for virus assembly. NS2A, NS2B, NS4A, and NS4B are hydrophobic peptides that may be membrane-associated. NS2A (Falgout et a/., 1989) and NS2B (Preugschat et al., 1990; Falgout et a/., 1991) have been proposed to play a role in modulating proteolytic processing of the non’ To whom

reprint

requests

should

be addressed. 689

0042-6822191 Copyright All rights

$3.00

0 1991 by Academic Press. Inc. of reproduction in any form resewed.

PREUGSCHAT

690

thermore, the cleavage appears to occur POSt-tradationally in cells infected with yellow fever virus (YF) (Chambers et a/., 1990a). Sequences within NS2A have been found to be required for proper cleavage of the l/ZA junction, and it was proposed that cleavage may occur autocatalytically involving residues near the cleavage site and located at least partially in NS2A (Falgout et al., 1989). The cleavage site between NS4A and NS4B has all of the known properties of a signalase site and it is believed that signalase does in fact make this cleavage. However, a polyprotein believed to be NS4AB has been observed in YF-infected cells, suggesting that the cleavage that separates YF NS4A from NS4B can occur post-translationally, which would be unusual for a eukaryotic signalase cleavage, or that a proportion of NS4AB cannot be processed (Chambers eta/., 1990a). It remains formally possible that the 4A/4B cleavage is mediated by a viral or cell-encoded proteinase of unique specificity. In this paper we report studies on the processing of the 3/4A and 4A/4B sites of the dengue 2 (DEN2) virus polyprotein, both in cell-free systems and in infected cells. We show that the N-terminus of NS4A is generated by the viral NS3 proteinase. We also found that NS4B is post-translationally modified in infected cells. MATERIALS

AND

METHODS

Cells and virus stocks

Stocks of DEN2 PR159 Sl (Bancroft et a/., 1981; Hahn et al., 1988; Preugschat et al., 1990) were prepared on Aedes albopictus C6/36 cells. Stocks of YF 17D were prepared on SW13 cells. C6/36, BHK-21 clone 15 and BSC-40 cells were propagated as described (Hahn et al., 1990; Preugschat et a/., 1990). The recombinant vaccinia virus vTF7-3 (obtained from B. Moss) (Fuerst et al., 1986) was twice plaque purified on BSC-40 cells prior to generating a high-titer stock on mouse L-cells. Cell lysates were monitored for T7 RNA polymerase activity as described (Elroy-Stein and Moss, 1990). Expression

constructs

cDNA clones of the DEN2 PR159 strain were used for all plasmid constructions (Hahn et al., 1988) using standard recombinant DNA techniques (Maniatis eta/., 1982). Recombinant clones were routinely screened by analysis of protein expression patterns, restriction endonuclease digestion, and DNA sequencing of miniprep DNAs. To create a gene fusion between trpE and DEN2 NS4B, vector pATH2 (Spindler et al., 1984; Hardy and

AND

STRAUSS

Strauss, 1988) was digested with BarnHI and ligated to a Bg/ll fragment containing nucleotides 6916 to 7284 of the DEN2 genome, to generate the expression plasmid pTNS4. Upon induction, bacteria containing pTNS4 produce a NS4B-trpE fusion protein containing amino acids 2273 to 2399 of the DEN2 polyprotein. In order to analyze processing at the 3/4A and 4A/4B cleavage sites, we constructed plasmids that expressed polyproteins which extended through NS4A or NS4B. Briefly vector pT10 (Preugschat et al., 1990) was digested with EcoRI, the 5’ overhanging ends were made blunt using Klenow fragment, and the product was digested with Sail. cDNA inserts containing NS4A or NS4B sequences were prepared by PCR. Clone pDN5 (Hahn et al., 1988) was used as the PCR template with a plus-strand primer complementary to nucleotides 4915-4934 of the DEN2 genome and a minus-strand primer complementary to nucleotides 6805-6825 or 7550-7569, respectively. The minusstrand primers introduced stop codons at the predicted C-termini of NS4A or NS4B. The amplified cDNA fragments were digested with SalI and cloned into pTl0 prepared as above to generate plasmids pT8 (extends through NS4A) and pT9 (extends through NS4B). Templates and RNAs for in vitro translations were prepared as described (Preugschat et al., 1991). Samples were boiled in SDS loading buffer (Laemmli, 1970) and were immediately subjected to polyacrylamide gel electrophoresis (PAGE). To construct a plasmid for the expression of polyprotein NS4AB in v&o, a plus-strand primer which introduced an initiating methionine at the N-terminus of NS4A was used in conjunction with the NS4B minusstrand primer described above to amplify cDNA containing NS4AB sequences. The plus-strand primer introduced a BarnHI restriction site at the 5’ end of the cassette to facilitate cloning into vector p5’3’PL (see Results). PCR products were gel purified, digested with BarnHI, and ligated to Bglll and Smal-digested p5’3’PL to generate the expression plasmid p5’3’4AB. Characterization

of viral proteins

The NS4B-trpE fusion protein was purified and injected into New Zealand white rabbits, and antiserum was prepared as described (Preugschat et a/., 1990). Western blot analyses were conducted using serum dilutions of 1: 150, as described (Johnson et a/., 1984). [35S]Methionineor [3H]leucine-labeled lysates were prepared by solubilizing infected cell monolayers in denaturing or nondenaturing lysis buffer (Hardy and Strauss, 1988; Preugschat et a/., 1990). Radiolabeled NS4B was purified from denatured BHK cell lysates by preparative immunoprecipitation and was sequenced

FLAVIVIRUS

NONSTRUCTURAL

NSl-

prM-

.NS2B

FIG. 1. Characterization of (uNS~B antiserum. (A) Immunoprecipitates of [%]methionine-labeled DEN2-infected BHK cells analyzed by SDS-PAGE. DEN2-infected cells (1 X 1 05) were labeled for 12 hr beginning at 30 hr postinfection. Lysates were immunoprecipitated using a polyvalent mouse hyperimmune ascitic fluid under denaturing (s-HIS) or nondenaturing conditions (t-HIS) or using monospecific (xNS2B or cuNS4B immune (I) or preimmune (P) sera under denaturing conditions, Locations of vrral proteins NS5, NS3, E, NSl, prM, NS2B. and NS4B are indicated. (B) Western blot analysis of mock- or DEN2-infected BHK cells. Cells (1 x 1 04) were solubilized in SDSPAGE loading buffer, and proteins were separated by SDS-PAGE and electroblotted onto nitrocellulose membranes. aNS26 or (YNS~B immune sera were used as a primary antibody and ‘z51-labeled Protein A was used for secondary detection of immune complexes.

on an Applied Biosystems as described (Preugschat Transfections

477A gas phase sequenator et al., 1990).

Vaccinia virus stocks were digested for 30 min with 0.25% w/v trypsin in Eagle’s medium for 30 min prior to infection. BSC-40 cells were grown to near confluency in 35-mm plates in Eagle’s medium containing 10% fetal calf serum. The cells were infected with vTF7-3 at a multiplicity of 10 for 30 min at 37” and then transfected with 5 pg of p5’3’4AB and 30 pg Lipofectin (GIBCO-BRL). RESULTS Characterization of arNS4B antiserum and identification of NS4B To study the metabolism of NS4B, we generated a rabbit polyclonal antiserum monospecific for NS4B by using as antigen a bacterial fusion protein containing amino acids 31 to 157 of NS4B. When the aNS4B antiserum was used to immunoprecipitate SDS-denatured proteins from DEN2-infected BHK cells labeled for 12 hr with [35S]methionine, a protein of approximately 28kDa was specifically precipitated (Fig. 1A). A small

PROTEIN

PROCESSING

691

amount of NS5 also precipitated in this experiment; the amount of NS5 which coprecipitated varied from experiment to experiment and the reason for its precipitation is not clear (but see also Chambers et a/., 1990a). Preimmune serum did not detectably react with viral proteins. By comparison, DEN2 hyperimmune ascitic fluid did not precipitate denatured NS4B, but did react weakly with NS4B under nondenaturing conditions (Fig. 1A). The titer of the aNS4B antiserum rose steadily with each injection of antigen, suggesting that NS4B is highly immunogenic in rabbits under the conditions used. lmmunoprecipitation with aNS2B antiserum (Preugschat et a/., 1990) is shown for comparison. Western blot analysis of a DEN2-infected BHK cell extract identified a 28-kDa protein that reacted with the immune serum but that was not present in mock-infected BHK cells (Fig. lB), confirming the results obtained by immunoprecipitation. A Western blot with (YNS~B antiserum is shown for comparison. NS4B is produced as a 30-kDa precursor mammalian and mosquito cells

in

The monospecific aNS4B antiserum was used to analyze the appearance of viral proteins during infection of BHK cells. NS4B was first detected at 6 to 9 hr postinfection, just before the release of mature virions from infected cells (data not shown). Other monospecific antisera (Preugschat eta/., 1990) produced similar results for other viral proteins. Changes in the pattern of DEN2 proteins and precursors present were not observed during the infection cycle (data not shown). During short-term labeling experiments, a novel 30kDa form of NS4B was observed in addition to the 28kDa form found upon long-term labeling. In order to investigate the relationship between the 30- and the 28-kDa forms of NS4B, a pulse-chase experiment was performed in both mammalian and mosquito cells (Fig. 2). The 30-kDa protein is clearly a precursor to the 28kDa form of NS4B. After 30 min of labeling, the 30-kDa form is the predominant form of NS4B present (Fig. 2, lanes “0.5”). In mammalian cells the 30-kDa form is then chased into the 28-kDa protein with an apparent half-time of approximately 90 min in BHK cells (Fig. 2A) or 60 min in LLC-MK, cells (data not shown), and the 28-kDa form is stable. Equivalent results were obtained with both BHK and LLC-MK, cells using pulse times of 5 or 15 min (not shown). In mosquito cells, in contrast, the 30-kDa protein was processed more slowly and both the 30-and the 28-kDa forms of NS4B appeared to be unstable, as the amounts of immunoprecipitable material decreased with time (Fig. 2B). In addition to the 30- and 28-kDa forms of NS4B, a 39-kDa polypeptide was immunoprecipitated by the

PREUGSCHAT

692

B 0.5

1

2

3

5

8

M IO.6

PO.51

2

3

5 8 H

AND

STRAUSS

that the post-translational modification diated by a cellular enzyme. N-terminal

NS4B

f

FIG. 2. NS4B metabolism in mammalian and mosquito cell lines. (A) Pulse-chase analysis of DEN2-infected BHK cells. DEN2infected (0.5, 1, 2, 3. 5, 8) or mock-infected (M) BHK cells (1 X IO? were labeled for 30 min beginning at 30 hr postinfection with [35S]methionine and then chased for various lengths of time with excess unlabeled methionine. Cells were lysed in denaturing lysis buffer and immunoprecipitated wrth aNS4B immune serum. The trmes indicated are in hours and represent time after addition of label, i.e., 1 hr is 30 min of labeling with a 30-min chase. A 30.min pulse-labeling of DENZ-infected LLC-MK, cells (lane 0.5, right edge of A) is shown for comparison of NS4AB and NS4B levels in another mammalian cell line. Two forms of NS4B and the putative polyprotein NS4AB are indicated. (6) Pulse-chase analysis of DEN2-infected mosquito cells. [35S]Methionine-labeled C6/36 cell lysates were analyzed by immunoprecipitation as described in A. Lane P is a DENZinfected C6/36 cell lysate immunoprecipitated using (YNS~B preimmune serum, and lane H is a DEN2-infected BHK cell lysate immunoprecipitated using a hyperimmune ascitic fluid serum. The molecular weights of 14C-labeled protein standards (Amersham) are indicated.

aNS4B antiserum from infected BHK and LLC-MK, cells (Fig. 2A), as well as from BSC-40 cells (see below), but not from DEN2-infected mosquito cells (Fig. 2B). From its molecular weight and immunoreactivity, we believe this to be polyprotein NS4AB, containing both NS4A and NS4B. This polyprotein appears during the pulse and disappears during the chase. Chambers et al. (1990a) also found NS4AB in YF-infected cells and cite evidence that it is a precursor to NS4B. In our hands, the amount of DEN2 NS4AB relative to that of the 30-kDa NS4B did not vary with length of pulse, suggesting that NS4AB may not be a precursor to NS4B in this system but might represent a small amount of polyprotein precursor that escapes normal processing. The kinetics and extent of modification of NS4B, and its stability, were cell-dependent. In addition we found that NS4B appeared to be modified in the absence of other viral proteins (see below). These results suggest

sequence

of NS4B is me-

of NS4B

The CXNS~B antiserum was used to purify NS4B from DEN2-infected BHKcelis byimmunoaffinitychromatography. Edman degradation of [35S]methionineor [3H]leucine-labeled NS4B released peaks of radioactivity in cycles 3 or 6, respectively (Fig. 3A), confirming that Asn-2243 of the DEN2 polyprotein is the N-terminal residue of NS4B (Fig. 3B). The predicted or identified N-termini of NS4B from several flaviviruses are aligned in Fig. 38. Asn has been identified as the N-terminal residue for the three NS4B species sequenced to date and by homology is presumed to be the N-terminal residue in all flavivirus NS4Bs. The 4A/4B cleavage site is preceded by a stretch of hydrophobic amino acids, which suggests that this cleavage is mediated by a signalase-like enzyme (von Heijne, 1986; Speight et al., 1988). If the conserved N-terminus of NS4B is generated in the lumen of the endoplasmic reticulum (ER), then both NS4A and NS4B are likely to be integral membrane proteins.

The N-terminus proteinase

of NS4A is generated

by the NS3

It has been proposed that the N-terminus of NS4A is generated by the nonstructural proteinase NS3 (Rice et al., 1986) whereas the 4A/4B cleavage is mediated by an enzyme with a specificity resembling that of signalase (Speight et a/., 1988). In order to investigate the processing of NS4A and NS4B in a cell-free system, we designed two expression plasmids, pT8 and pT9. Plasmid pT8 contains a T7 RNA polymerase promoter followed by a cDNA copy of the 5’ nontranslated region (NTR) of DEN2 RNA and the coding region for the first 37 amino acids of the capsid protein fused in-frame to the sequence beginning with amino acid 101 of NS2A and extending through NS4A, with a stop codon introduced at the end of NS4A. A second plasmid pT9 extended the DEN2 sequence through NS4B, with the stop codon placed at the end of NS4B. Transcription of the plasmids with T7 RNA polymerase and translation of the resulting RNAs in reticulocyte lysates supplemented with microsomal membranes gave the patterns shown in Fig. 4A. Only low levels of the full-length polyproteins, P2AB34A or P2AB34AB, are seen, suggesting that in this experiment the polyproteins were efficiently cleaved at the 2A/2B junction. This suggests that the first cleavage in a polyprotein containing the 2A/2B, 2BI3, and 3l4A cleavage sites occurs between

FLAWVIRUS

NONSTRUCTURAL

A DPM

5

lb

=

%SMet

-

sHI&u

PROTEIN

PROCESSING

693

The 4A/4B cleavage site was not processed in these experiments (pT9 translations, Fig. 4A), which precluded-the use of an in vitro approach to study processing at the 4A/4B site. However, the extra sequence contributed by NS4B in the various polyproteins allowed us to unambiguously identify processing intermediates on the basis of their gel mobilities. Polyproteins containing NS4B sequences were also identified by immunoreactivity with aNS4B immune serum (data not shown). An 85kDa doublet consisting of the processing intermediates NS2B3 and NS34A is produced in pT8-programmed translations (Fig. 4A). A single 85kDa protein (NS2B3) is observed in pT9-programmed

15

CYCLE

B

NS2B24AB NS34AB

NS4B Ns2B34A Ns34Ai

KuN* YF* JE TBE

NS2BS

NS2BS

DEN2* DENS DEN4

NSS-

..L.M.....E..F . .Y.M.. . . .A. .K . . . . . . . . . . A. .S

FIG. 3. N-terminal sequence of NS4B. (A) Data are plotted as radioactivity per Edman degradation cycle versus cycle number. The burst of 35S observed in the first sequencing cycle of immunoaffinitypurified NS4B is presumably contributed by contaminants which coprecipitate with NS4B. (B) The identified (*) or predicted N-termini of NS4B species from several flaviviruses are aligned relative to the DEN2 sequence. The methionine and leucine residues identified by Edman degradation of radiolabeled NS4B are highlighted by shaded boxes. A putative hydrophobic, membrane-spanning segment at the C-terminus of NS4A is enclosed by an open box, and the N-terminus of NS4B is indicated by a rightward facing arrowhead. The single letter amino acid code is used. Abbreviations and sequences used for flaviviral NS4B alignments are as follows: DEN2, dengue type 2 PR159 Sl strain (Hahn et a/., 1988); DENB, dengue type 3 (Csatomi and Sumiyoshi, 1990); DEN4, dengue type 4 (Mackow et al., 1987); MVE, Murray Valley encephalitis (Lee eta/., 1990); KUN, Kunjin (Coia et al., 1988); YF, yellow fever 17D strain (Rice et al., 1985); JE, Japanese encephalitis Beijing 1 strain (Hashimoto et al., 1988); TBE, tickborne encephalitis virus western subtype (Mandl et al., 1989).

NS2A and NS2B. In previous studies in which DEN2 polyproteins containing only the 2fV2B and 2B/3 cleavage sites were translated in reticulocyte lysates, we also found that the NS3 proteinase first cleaved the 2AI2B site to produce a processing intermediate, NS2B3, which was then processed to yield NS2B and NS3 (Preugschat et a/., 1990, 1991).

--3

Tertiary Cll?iWage

f?l

A

FIG. 4. The N-terminus

of NS4A is generated by NS3. (A) SDSPAGE analysis of pT8- and pT9-programmed in vitro translations. Aliquots of pT8- or pT9-programmed translations were removed at the time points indicated (in minutes) and were resolved by SDSPAGE (10% acrylamide, acrylamide:bis ratio of 6O:l). The gel was dried and directly autoradiographed. The positions of processing intermediates (NS2B34AB, NS2B34A, NS34AB. NS34A, and NS2B3) and products (NS3) are indicated. (B) Processing pathway of the pT8 polyprotein. The full-length translate P2AB34A contains sequences from capsid (checkered box), NS2A (open box), NS2B (shaded box), NS3 (closed box), and NS4A (diagonally shaded box). The relative positions of the four homology boxes proposed by Bazan and Fletterick (1989) are designated by four small closed boxes at the N-terminus of NS3. Primary processing at the 2AJ2B cleavage site is designated by a large downward facing arrowhead. Secondary processing at either the 2B/3 or the 3/4A cleavage sites is designated by small downward facing arrowheads. Tertiary processing generates mature NS2B, NS3, and presumably NS4A.

PREUGSCHAT

694

translations, as NS34A was not generated, and a higher molecular weight band of NS34AB is seen. The presence of both NS2B3 and NS34A, as well as NS3, in the pT8 translation, and of NS2B3 and NS34AB as well as NS3 in the pT9 translation, clearly shows that cleavage at both the 2B/3 and the 3/4A sites occurred in these experiments and that the order of these two cleavages was not fixed. Figure 4B presents a schematic of the pT8 polyprotein processing pathway. Primary cleavage of the fulllength polyprotein P2AB34A occurs at the 2AI2B junction to generate NS2B34A. Secondary cleavage occurs at either the 2B/3 or the 3/4A cleavage sites to generate two different NS3-containing processing intermediates, NS2B3 or NS34A. Tertiary cleavage then occurs to generate the mature nonstructural proteins NS2B, NS3, and presumably NS4A. It is not clear in these experiments whether both NS2B3 and NS34A can be further processed to produce NS3 or if only NS2B3 is actively processed (see Falgout et al., 1991; Preugschat et a/., 1990). The efficient primary processing of the 2A/2B cleavage site and the generation of both NS2B3 and NS34A in pT8-programmed translations suggest that the DEN2 polyprotein is processed with a preferred but nonobligatory order (see also Preugschat et a/., 199 1). Inactivation of the NS3 proteinase by a change of Asn-152 + Lys abolished all processing of the pT9 polyprotein (data not shown). Cleavage of the 3/4A site in both pT8- and pT9-programmed translations (Fig. 4A) and its abolition in the N 152K mutant provided evidence that this site is cleaved by the NS3 proteinase. In viva modification

of NS4AB

The 4A/4B cleavage site was not processed in pT9-programmed translations. The addition of nonionic detergents to release signalase from the lumen of the ER did not result in processing of the 4A/4B junction (data not shown). Previous experiments have shown that some internal signal sequences that are not processed in vitro can be processed in viva (Rottier et al., 1987). Therefore in an attempt to study the processing of the 4A/4B junction, the T7 expression system of Fuerst et al. (1986) was used to produce NS4AB in mammalian cells. Our initial attempts using the T7 expression system failed to detect expression of DEN2 proteins in transfected cells. vTF7-3-infected cell extracts contained active T7 RNA polymerase and vTF7-3 infection did not affect DEN2 protein stability in coinfection experiments (data not shown), suggesting that transcribed RNAs were either inefficiently translated or were unstable in viva. As capped or uncapped mRNAs which contain

AND

STRAUSS

NSS-

-93 kDa

NSS.

$9 kDa

E-

NSl-

-46 kDa xi ,.

NS4Bf

L

-31 kDa

_

^

FIG. 5. In viva modification of NS4AB. Mock (M)- or DEN2 (D)-infected BSC-40 cells were labeled with [35S]methionine for 2 hr, beginning at 22 hr postinfection. SDS-denatured cell lysates were immunoprecipitated using DEN2 hyperimmune ascitic fluid (HIS) or olNS4B immune serum. Mock (T7)- p5’3’4AB-transfected (T7T) vTF7-3-infected BSC-40 cells were labeled with [35S]methionine for 2 hr, at 22 hr postinfection. SDS-denatured cell lysates were immunoprecipitated using aNS4B immune or preimmune (P) serum. Locations of bands containing the viral proteins NS5, NS3, E, NSl , NS4B, and NS4AB are indicated. The molecular weights of “C-labeled protein standards (Amersham) are indicated.

the flavivirus 5’ NTR are efficiently translated in reticulocyte lysates (Preugschat et a/., 1990; Ruiz-Linares et a/., 1989), we assumed that the mRNA stability was limiting expression in v&o. An expression vector (p5’3’PL) was constructed that contained both the DEN2 5’ NTR and the DEN2 3’ NTR modified to terminate with a 21-nt poly(A) tract. mRNAs containing the DEN2 5’ and 3’ NTRs were efficiently translated in vitro (data not shown) and were sufficiently stable in vivo to allow detection of protein expression (see below). Cells were infected with vaccinia vTF7-3 and transfected with p5’3’4AB, and denatured cell lysates were immunoprecipitated with (rNS4B antiserum to assay for the production of NS4B (Fig. 5). Mature NS4B was not detected in p5’3’4AB-transfected cells, but a polyprotein consistent in molecular weight and immunoreactivity with NS4AB was specifically immunoprecipitated. This putative NS4AB was present as a doublet, similar to the results with NS4B, suggesting that the NS4B sequence within NS4AB had undergone the same post-translational modification that occurred in NS4B. The electrophoretic mobility of NS4AB immuno-

FLAVIVIRUS

NONSTRUCTURAL

precipitated from p5’3’4AB-transfected cell lysates differed slightly from that of NS4AB immunoprecipitated from DEN-2-infected BSC-40 cell lysates, presumably because of the addition of a methionine residue at the N-terminus of NS4A. Modification of NS4AB appeared to occur more slowly than that of NS4B, suggesting that the efficiency of NS4B modification differed when present as a NS4AB polyprotein. In order to rule out the possibility that the lack of 4A/4B processing was due to mutations introduced by PCR amplification, several individual clones were analyzed by expression and by DNA sequencing. No differences were observed in the level or pattern of expression and the DNA sequence obtained was identical to the published sequence, suggesting that no mutations had been introduced during cloning procedures. We can not rule out the possibility that mutations were introduced by transfection into vTF7-3-infected BSC40 cells, as it is known that transfected DNA is extensively replicated in vaccinia-infected cells (DeLange and McFadden, 1986; Merchlinshy and Moss, 1988). Further studies will be necessary to determine if 4AJ4B processing can be rescued by viral proteins in cis or in trans. DISCUSSION Cleavage between NS3 and NS4A is effected by the NS3 proteinase Our results indicate that the N-terminus of DEN2 NS4A is generated by the nonstructural proteinase NS3. The predicted site of cleavage from homology with the site of cleavage determined for Kunjin virus (Speight and Westaway, 1989b) is GRK 4 S, typical of sites cleaved by this enzyme which, in most cases, cleaves following two basic residues that are flanked by amino acids with short side chains. We found that NS3-mediated processing was abolished by the Asn152 --* Lys mutation. Asn-152 lies within the putative substrate binding pocket described by Bazan and Fletterick (1989) and is one of five residues predicted to be in direct contact with bound substrate. It is likely that the structure of the substrate binding pocket was perturbed by the mutation. Identification of processing intermediates provided evidence that the DEN2 polyprotein containing the 2A/ 2B, 2BI3, and 3/4A cleavage sites is cleaved with a preferred but nonobligatory order. Mechanistically these results imply that after the 2A/2B cleavage site is processed, the conformational flexibility of the DEN2 polyprotein increases in order to allow processing of either the 2B/3 or the 3/4A cleavage sites. Trans processing has not been observed in vitro at the 2A/2B, 2B/3 (Chambers eta/., 1990b; Preugschat eta/., 1990)

PROTEIN

PROCESSING

695

or 3/4A sites (data not shown), or at the 3/4A site in vivo (Falgout eta/., 1991). We conclude that in vitro the 2A/2B, 2B/3, and 3/4A cleavage sites are probably processed in cis. Experiments in which cleavage sites within the Sindbis alphavirus nonstructural polyprotein were selectively inactivated suggested that the Sindbis polyprotein is processed with an obligatory order (Shirako and Strauss, 1990). In contrast, the results presented in this paper and those obtained with chimeric proteinases, where mutagenesis of the substrate binding pocket changed both the order and the efficiency of processing (Preugschat et a/., 1991), support the hypothesis that the flavivirus polyprotein is processed with a nonobligatory order. Similarly, the HIV proteinase processes the gag polyprotein with a preferred, nonobligatory order (Tritch et a/., 1991). These data suggest that inactivation of one cleavage site, such as 2B/3 or 3/4A, should not affect the ability of the proteinase to process other wild-type cleavage sites unless the structure of the polyprotein was altered. Post-translational

modification

of NS4B

It is clear from our analysis that NS4B is post-translationally modified. We found that NS4B is produced as a 30-kDa form that is chased into a 28-kDa form. Chambers et al. (1989) also reported that YF NS4B exhibited heterogeneity during long-term labeling of mammalian cells, but this heterogeneity was not observed in YF-infected mosquito cells. Thus our results and the results of Chambers et al. (1989) suggest that the kinetics and the extent of NS4B post-translational processing vary from cell line to cell line. it is tempting to speculate that the variation in post-translational modification of NS4B may contribute to the variability in the levels of RNA replication observed among cell lines. The nature of the modification to NS4B is obscure. As NS4B is likely to be an integral membrane protein that spans the endoplasmic reticulum, modification probably occurs in close association with host cell membranes. The modification leads to an increase in the mobility of the protein, but may not involve truncation at either the N-terminus or the C-terminus. C-terminal sequence analysis of Kunjin NS4B suggested that Kunjin NS4B is not truncated post-translationally (Speight and Westaway, 1989a), and we assume the same should be true for DEN2 NS4B. Although the N-terminal sequence of the precursor 30-kDa form of NS4B has not been obtained for comparison with that of the 28-kDa product (only the 28-kDa form of NS4B was sequenced in our experiments because of the difficulty of obtaining sufficient amounts of the short-lived

696

PREUGSCHAT

30-kDa form), N-terminal modification seems unlikely because of the findings with NS4AB. The modification seen in NS4AB appears to be the same as that which occurs in NS4B, and the N-terminus of NS4B is not available for truncation in NS4AB. It is possible, therefore, that NS4B is internally modified in some way that leads to an increase in its mobility. Further experiments are needed to identify the site and type of post-translational modification in NS4B. Cleavage

of the 4A/4B

site

Chambers et a/. (1990a) reported that NS4AB was produced in YF-infected cells and suggested that this could serve as a precursor to NS4B. This would imply that some proportion of 4A/4B cleavage can occur post-translationally. We also observed small amounts of NS4AB in DEN2-infected cells but were unable to demonstrate a precursor-product relationship with NS4B, and in the case of DEN2 it may be that a small proportion of polyprotein escapes normal processing. The lack of 4A/4B cleavage in p5’3’4AB-transfected cells suggests that additional virus-encoded sequences or virus-induced cellular factors are required either for targeting NS4AB to the correct cellular compartment or for processing this cleavage site. The fact that NS4AB appears to be modified in a manner similar to that observed for NS4B suggests that the correct cellular compartment has been reached at least for this modification. Although the site of cleavage fits well all known requirements for cleavage by signalase, it is formally possible that a virus-encoded or virus-induced cellular proteinase is responsible for 4A/4B cleavage. Complementation experiments designed to rescue cleavage in transfected cells should help delineate which viral proteins are directly or indirectly involved in processing of the 4A/4B cleavage site. ACKNOWLEDGMENTS The authors thank Ellen Strauss, Richard Kuhn, and Ron Weir for stimulating discussions and editorial expertise; David Teplow and Tammy Bauer for their expertise and assistance with protein sequencing; and Joel Dalrymple for furnishing cells, virus, and hyperimmune antiserum. This work was supported by Grant V22/181/10 from the World Health Organization and by Grant Al2061 2 from the Natlonal Institutes of Health.

REFERENCES BANCROFT, W. H., TOP, F. H., ECKELS, K. H., ANDERSON, J. H., MCCOWN. J. M., and RUSSELL, P. K. (1981). Dengue-2 vaccine: Virological, immunological, and clinical responses of six yellow fever-immune recipients. Infect. Immun. 31, 698-703. BAZAN. I. F., and FLETTERICK. R. 1. (1989). Detection of a trypsin-like serine protease domain in flaviviruses and pestiviruses. wfology 171, 637-639.

AND

STRAUSS CHAMBERS, T. J.. MCCOURT, D. W., and RICE, C. M. (1989). Yellow fever virus proteins NS2a, NS2b and NS4b: Identification and parteal N-terminal amino acid sequence analysis. Virology 169, 1 OO109. CHAMBERS, T. J., MCCOURT, D. W., and RICE, C. M. (1990a). Production of yellow fever virus proteins in infected cells: Identification of discrete polyprotein species and analysis of cleavage kinetics using region-specific polyclonal antisera. Virology 177, 159-l 74. CHAMBERS, T. J., WEIR, R. C., GRAKOUI, A., MCCOURT. D. W., BAZAN, J. F., FLE~ERICK. R. J., and RICE, C. M. (1990b). Evidence that the N-terminal domain of yellow fever virus NS3 is a serine protease responsible for site-specific cleavages in the viral polyprotein. Proc. Nat/. Acad. Sci. USA 87, 8898-8902. CHAMBERS, T. J., HAHN, C. S., GALLER, R., and RICE, C. M. (199Oc). Flavivirus genome organization, expression, and replication. Annu. Rev. Microbial. 44, 649-688. COIA, G., PARKER, M. D., SPEIGHT, G., BYRNE, M. E., and WESTAWAY, E. G. (1988). Nucleotide and complete amino acid sequences of Kunjin virus: Definitive gene order and characteristics of the virusspecified proteins. /. Gen. Viral. 69, l-2 1. DERANGE, A. M., and MCFADDEN, G. (1986). Sequence-nonspecific replication of transfected plasmid DNA in poxvirus-infected cells. Proc. Nat/. Acad. SC;. USA 83, 614-618. ELROY-STEIN, O., and Moss, B. (1990). Cytoplasmic expression system based on constitutlve synthesis of bacteriophage T7 RNA polymerase In mammalian cells. Proc. Nat/. Acad. Sci. USA 87, 6743-6747. FALGOUT, B., CHANOCK, R., and LAI, C. 1. (1989). Proper processing of dengue virus nonstructural glycoprotein NSl requires the N-terminal hydrophobic signal sequence and the downstream nonstructural protein NS2a. J. viral. 63, 1852-l 860. FALGOUT, B., PETHEL, M., ZHANG, Y.-M., and LAI, C.-J. (1991). Both nonstructural proteins NS2B and NS3 are required for the proteolytic processing of dengue virus nonstructural proteins. J. Viral. 65, 2467-2475. FUERST. T. R.. NILES. E. G., STUDIER, F. W., and Moss, B. (1986). Eukaryotic transient-expression system based on recombinant vaccinia virus that synthesizes bacteriophage T7 RNA polymerase. Proc. Nat/. Acad. Sci. USA 83, 8122-8126. GORBALENYA, A. E., KOONIN, E. V., DONCHENKO. A. P., and BLINOV, V. M. (1989). Two related superfamilies of putative helicases involved in replication, recombination, repalr and expression of DNA and RNA genomes. Nucleic Acids Res. 17, 47 13-4729. HAHN, Y. S., GALLER. R., HUNKAPILLER, T., DALRYMPLE, J., STRAUSS, J. H.. and STRAUSS, E. G. (1988). Nucleotide sequence of dengue 2 RNA and comparison of the encoded proteins with those of other flavlviruses. Virology 162, 167-l 80. HAHN, Y. S., LENCHES, E. M., GALLER, R., RICE, C. M., DALRYMPLE, J., and STRAUSS, J. H. (1990). Expression of the structural proteins of dengue 2 virus and yellow fever virus by recombinant vaccinia viruses. Arch. Viral. 115, 251-265. HARDY, W. R.. and STRAUSS, J. H. (1988). Processing of the nonstructural polyprotelns of Sindbis virus: Study of the kinetics in vivo using monospeciflc antibodies. /. Viral. 62, 998-l 007. HASHIMOTO, H., NOMOTO, A., WATANABE, K., MORI. T., TAKEZAWA, T.. AIZAWA, C., TAKEGAMI, T., and HIRAMATSU. K. (1988). Molecular cloning and complete nucleotlde sequence of the genome of Japanese encephalitis virus Beijing-1 strain. Virus Genes 1, 305-317. JOHNSON, D. A., GAUXCH, J. W., SPORTSMAN, J. R.. and ELDER, J. H. (1984). Improved technique utilizing nonfat dry milk for analysis of proteins and nucleic acids transferred to nitrocellulose. Gene Anal. Tech. 1, 3-8. LAEMMLI, U. K. (1970). Cleavage of structural proteins during assembly of the head of bacteriophage 14. Narure 227, 680-685.

FLAWVIRUS NONSTRUCTURAL LEE, E., FERNON,C., SIMPSON, R., WEIR, R. C., RICE,C. M., and DALGARNO, L. (1990). Sequence of the 3’ half of the Murray Valley encephalitis virus genome and mapping of the nonstructural proteins NSl , NS3 and NS5. Virus Genes 4, 197-213. MACKOW, E., MAKINO, Y., ZHAO, B., ZHANG, Y.-M., MARKOFF, L., BUCKLER-WHITE,A., GUILER, M. CHANOCK, R. M., AND LAI, C.-J. (1987). Nucleotide sequencing of dengue type 4 virus: Analysis of genes coding for nonstructural proteins. Virology 159, 217-228. MANDL, C. W., HEINZ,F. X., STOEKL, E., and KUNZ,C. (1989). Genome sequence of tick-borne encephalitis virus (western subtype) and comparative analysis of nonstructural proteins with other flaviviruses. Virology 173, 29 l-30 1. MANIATIS, T., FRITSCH,E. F., and SAMBROOK, J. (1982). “Molecular Cloning, a Laboratory Manual.” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. MERCHLINSHY,M., and Moss, B. (1988). Sequence-independent replication and sequence-specific resolution of plasmids containing the vaccinia virus concatemer junction: Requirements for early and late trans-acting factors. Cancer Cells 6, 87-93. MONATH, T. (1986). Pathology of the flaviviruses. ln “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, Eds.), pp. 375-440. Plenum, New York. OSATOMI, K., and SUMIYOSHI,H. (1990). Complete nucleotide sequence of dengue type 3 virus genome RNA. Virology 176, 643647. PREUGSCHAT,F., YAO, C.-W., and STRAUSS,J. H. (1990). In vitro processing of dengue 2 nonstructural proteins NS2A, NS2B and NS3. J. Viral. 64, 4364-4374.

PREUGSCHAT,F., LENCHES,E. M., and STRAUSS,1. H. (1991). Flavivirus enzyme-substrate interactions studied with chimeric proteinases: Identification of an intragenic locus important for substrate recognition. J. Viral. 65, 4749-4758. RICE,C. M., LENCHES,E. M., EDDY,S. R., SHIN, S. J., SHEETS,R. L., and STRAUSS,1. H. (1985). Nucleotide sequence of yellow fever virus: Implications for flavivirus gene expression and evolution. Science 229, 726-733. RICE,C. R., STRAUSS,E. G., and STRAUSS,J. H. (1986). Structure of the flavivirus genome. In “The Togaviridae and Flaviviridae” (S. Schlesinger and M. J. Schlesinger, Eds.), pp. 279-327. Plenum, New York.

PROTEIN PROCESSING

697

RICE,C. M.. and STRAUSS,J. H. (1990). Production of flavivirus polypeptides by proteolytic processing. Sem. Viral. 1, 357-367. ROTTIER, P. J. M., FLORKIEWICZ,R. Z., SHAW, A. S., and ROSE,J. K. (1987). An internalized amino-terminal signal sequence retains full activity in vivo but not in vitro. 1. Biol. Chem. 262, 8889-8895. RUIZ-LINARES.A., BOULOY, M., GIRARD,M., and CAHOUR,A. (1989). Modulations of the in vitro translational efficiencies of yellow fever virus mRNAs: Interactions between coding and noncoding regions. Nucleic Acids Res. 17, 2463-2474. SHIRAKO,Y., and STRAUSS,J. H. (1990). Cleavage between nsP1 and nsP2 initiates the processing pathway of Sindbis virus nonstructural polyprotein P123. Virology 177, 54-64. SPEIGHT, G., COIA, G., PARKER,M. D., and WESTAWAY,E. G. (1988). Gene mapping and positive identification of the non-structural proteins NS2A, NS2B. NS3, NS4B and NS5 of the flavivirus Kunjin and their cleavage sites. J. Gen. Viral. 69, 23-34. SPEIGHT,G., and WESTAWAY,E. G. (1989a). Carboxy terminal analysis of nine proteins specified by the flavivirus Kunjin: Evidence that only the intracellular core protein is truncated. J. Gen. Viral. 70, 2209-2214. SPEIGHT,G., and WESTAWAY,E. G. (198913). Positive identification of NS4a, the last of the hypothetical nonstructural proteins of flaviviruses. Virology 170, 299-301. SPINDLER,K. R., ROSSER,D. S. E., and BERK,A. J. (1984). Analysis of adenovirus transforming proteins from early regions 1A and 1 B with antisera to inducible fusion agents produced in Escherichia coli. J. Viral. 49, 132-l 41. STOHLMAN, S. A., WISSEMAN, C. L., EYLAR.0. R., and SILVERMAN,D. J. (1975). Dengue virus induced modifications of host cell membranes. J. l&o/. 16, 1017-1026. STRAUSS,1. H., PREUGSCHAT,F., and STW\USS,E. G. (1991). Structure and function of the flavivirus and pestivirus genomes. ln “Viral Hepatitis and Liver Disease” (F. B. Hollinger, S. M. Lemon, and H. S. Margolis, Eds.), pp. 333-344. Williams &Wilkins, Baltimore, MD. TRITCH, R. J., CHENG, Y.-S. E., YIN, F. Y., and ERICKSON-VIITANEN (1991). Mutagenesis of protease cleavage sites in the human immunodeficiencyvirus type 1 gag polyprotein. 1. Viral. 65,922-930. VON HEIJNE,G. (1986). A new method for predicting signal sequence cleavage sites. Nucleic Acids Res. 14, 4683-4690.