NS2 Proteins of GB Virus B and Hepatitis C Virus Share CommonProtease Activities and Membrane Topologies

Célia Boukadida,a,b,c Caroline Marnata,a,b,c Roland Montserret,d,e,f Lisette Cohen,a,b,c Brigitte Blumen,a,b,c Jérôme Gouttenoire,g

Darius Moradpour,g François Penin,d,e,f Annette Martina,b,c

Institut Pasteur, Unit of Molecular Genetics of RNA Viruses, Paris, Francea; CNRS UMR3569, Paris, Franceb; Université Paris Diderot-Sorbonne Paris Cité, Paris, Francec;Institut de Biologie et Chimie des Protéines, Bases Moléculaires et Structurales des Systèmes Infectieux, Lyon, Franced; CNRS UMR 5086, Lyon, Francee; Labex Ecofect,University of Lyon, Lyon, Francef; Division of Gastroenterology and Hepatology, Centre Hospitalier Universitaire Vaudois, University of Lausanne, Lausanne, Switzerlandg

ABSTRACT

GB virus B (GBV-B), which is hepatotropic in experimentally infected small New World primates, is a member of the Hepacivi-rus genus but phylogenetically relatively distant from hepatitis C virus (HCV). To gain insights into the role and specificity ofhepaciviral nonstructural protein 2 (NS2), which is required for HCV polyprotein processing and particle morphogenesis, weinvestigated whether NS2 structural and functional features are conserved between HCV and GBV-B. We found that GBV-BNS2, like HCV NS2, has cysteine protease activity responsible for cleavage at the NS2/NS3 junction, and we experimentally con-firmed the location of this junction within the viral polyprotein. A model for GBV-B NS2 membrane topology was experimen-tally established by determining the membrane association properties of NS2 segments fused to green fluorescent protein (GFP)and their nuclear magnetic resonance structures using synthetic peptides as well as by applying an N-glycosylation scanning ap-proach. Similar glycosylation studies confirmed the HCV NS2 organization. Together, our data show that despite limited aminoacid sequence similarity, GBV-B and HCV NS2 proteins share a membrane topology with 3 N-terminal transmembrane seg-ments, which is also predicted to apply to other recently discovered hepaciviruses. Based on these data and using trans-comple-mentation systems, we found that intragenotypic hybrid NS2 proteins with heterologous N-terminal membrane segments wereable to efficiently trans-complement an assembly-deficient HCV mutant with a point mutation in the NS2 C-terminal domain,while GBV-B/HCV or intergenotypic NS2 chimeras were not. These studies indicate that virus- and genotype-specific intramo-lecular interactions between N- and C-terminal domains of NS2 are critically involved in HCV morphogenesis.

IMPORTANCE

Nonstructural protein 2 (NS2) of hepatitis C virus (HCV) is a multifunctional protein critically involved in polyprotein process-ing and virion morphogenesis. To gain insights into NS2 mechanisms of action, we investigated whether NS2 structural andfunctional features are conserved between HCV and GB virus B (GBV-B), a phylogenetically relatively distant primate hepacivi-rus. We showed that GBV-B NS2, like HCV NS2, carries cysteine protease activity. We experimentally established a model forGBV-B NS2 membrane topology and demonstrated that despite limited sequence similarity, GBV-B and HCV NS2 share an or-ganization with three N-terminal transmembrane segments. We found that the role of HCV NS2 in particle assembly is genotypespecific and relies on critical interactions between its N- and C-terminal domains. This first comparative analysis of NS2 proteinsfrom two hepaciviruses and our structural predictions of NS2 from other newly identified mammal hepaciviruses highlight con-served key features of the hepaciviral life cycle.

The Hepacivirus genus of the Flaviviridae family was originallycreated to uniquely classify hepatitis C virus (HCV), a virus

that chronically infects an estimated 130 million to 170 millionindividuals worldwide. Until recently, only one other virus wastentatively classified along with HCV within the Hepacivirus ge-nus, GB virus B (GBV-B), a virus of unknown origin identified in1995 in experimentally infected small New World primates (1).GBV-B infects tamarins (Saguinus species) and marmosets (Cal-lithrix species) and generally causes acute self-resolving hepatitis(2–4) but in some cases also causes protracted or chronic infec-tions (5–7), making GBV-B an attractive surrogate animal modelfor HCV infection. In the past 3 years, several hepaciviral homo-logues have been identified in various mammalian species, includ-ing horses (8–10), rodents (11, 12), bats (13), and Old Worldmonkeys (14). Based on phylogenetic relationships, these recentlyidentified viruses were tentatively classified within the Hepacivirusgenus. Although none of these new viruses has been conclusivelyidentified as being hepatotropic, they represent valuable models to

decipher conserved salient features of the hepacivirus life cycleand hepacivirus-host interactions and may guide future efforts inestablishing useful surrogate rodent models.

The HCV genome encodes a polyprotein precursor that iscleaved co- and posttranslationally by cellular and viral proteasesto yield the capsid protein (C) and two envelope glycoproteins (E1and E2) that form the viral particle as well as seven nonstructuralproteins (15). GBV-B shares with HCV a common genomic orga-nization, including a 5= nontranslated region (5=NTR) containing

Received 4 March 2014 Accepted 12 April 2014

Published ahead of print 16 April 2014

Editor: M. S. Diamond

Address correspondence to Annette Martin, annette.martin@pasteur.fr.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JVI.00656-14

7426 jvi.asm.org Journal of Virology p. 7426 –7444 July 2014 Volume 88 Number 13

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

an internal ribosome entry site (16–19) and conserved key enzy-matic functions such as NS3-4A serine protease and helicase ac-tivities as well as NS5B RNA-dependent RNA polymerase activity(20–22). Mechanisms of capsid protein maturation and lipiddroplet targeting involved in particle morphogenesis (23–25) aswell as the role of NS3-4A protease in disrupting host innate im-mune responses through the cleavage of the adaptor proteinMAVS (26, 27) are also shared by GBV-B and HCV. Althoughexperimental studies with the newly identified hepaciviruses arenot as extensive, it has been reported that equine and some rodenthepacivirus genomes bear 5=-NTR organizations (8, 11) and theability to cleave MAVS (28) similarly to HCV and GBV-B. Amongnonstructural proteins, NS3 to NS5B were shown to be necessaryand sufficient to ensure genome replication by establishing auton-omously replicating subgenomic HCV and GBV-B RNAs (26, 29,30), and the small ion channel protein p7 as well as NS2 of HCVwere found to be essential for particle morphogenesis (31–33).HCV NS2 is a 217-amino-acid (aa) transmembrane (TM) proteinthat associates with the endoplasmic reticulum (ER) membrane ofinfected hepatoma cells (34, 35). The p7/NS2 junction is cleavedby host signal peptidases, indicating that the NS2 N terminus re-sides in the ER lumen. A recently proposed topological model ofHCV NS2 described three putative TM segments located within itsN-terminal region (34) and a C-terminal globular cytosolic do-main. The HCV NS2 C-terminal subdomain carries cysteine pro-tease activity responsible for cleavage at the NS2/NS3 junction(36, 37), and its crystal structure indicates that the NS2 protease isactive as a dimer with composite active sites (38). Moreover, effi-cient processing at the NS2/NS3 junction requires the NS3 N-ter-minal protease domain that acts as a cofactor (39). In addition toits proteolytic activity, NS2 is involved in virion assembly andegress, although its precise mechanisms of action in these pro-cesses remain elusive. Several studies suggested that particle as-sembly could take place in specialized microdomains of the ERmembrane, close to lipid droplets and replication sites (40, 41),and requires the concerted action of structural but also nonstruc-tural proteins that may be coordinated by NS2 (34, 35, 42–45).

In order to gain further insight into the mechanisms of actionof NS2 during the HCV life cycle, we investigated whether NS2structural and functional features are conserved among distantlyrelated hepaciviruses such as GBV-B and HCV. In this study, anal-ysis of the proteolytic processing of GBV-B NS2-NS3 precursorsin transiently transfected cells revealed that GBV-B NS2 is also acysteine protease responsible for the cleavage of the polyprotein atthe NS2/NS3 junction, which we experimentally mapped betweenThr 940 and Ala 941 of the viral polyprotein. GBV-B and HCVNS2 membrane topologies were comparatively studied throughanalyses of the membrane association properties of GBV-B NS2segments and comparison with the proposed HCV NS2 topolog-ical model (31, 34), determinations of GBV-B and HCV NS2segment orientations with respect to the ER membrane using anN-glycosylation acceptor site tagging approach, as well as deter-minations of nuclear magnetic resonance (NMR) structures ofselected GBV-B NS2 membrane binding segments. Overall, ourdata demonstrated that GBV-B and HCV NS2 proteins share sim-ilar topological organizations in spite of limited amino acid con-servation. Based on these data, we investigated whether the mech-anisms of action of NS2 during morphogenesis were conservedbetween GBV-B and HCV and across HCV genotypes by using atrans-complementation approach.

MATERIALS AND METHODSPlasmids. GBV-B NS2 sequences were amplified from pGBV-B/2, whichcontains a genome-length GBV-B cDNA (6). HCV genotype 1b strainCon1 (HCV 1b-Con1) and 1b-N NS2 sequences were amplified fromplasmids pCMVNS2-GFP (34) (see below) and pNNeo/C-5B (46), whichwas kindly provided by S. M. Lemon (University of North Carolina, Cha-pel Hill, NC), respectively. HCV 2a-JFH1 and 2a-J6 NS2 sequences wereamplified from plasmids pJFH1 (47), kindly provided by T. Wakita (Na-tional Institute of Infectious Diseases, Tokyo, Japan), and FL-J6/JFH-5=C19Rluc2AUbi (48), kindly provided by C. Rice (The Rockefeller Uni-versity, New York, NY), respectively.

Plasmid pCMV/GFP, which allows the expression of enhanced greenfluorescent protein (GFP), was generated by inserting between the EcoRIand BamHI restriction sites of pCMV-KEB-GFP (49) the hybridizationproduct of two complementary, 5=-phosphorylated oligonucleotides de-signed to create an EcoRI restriction site, followed by ACC-ATG nucleo-tides that include a start codon and a BamHI restriction site. PlasmidpCMV/sp-GFP was derived from pCMV-KEB-GFP by inserting betweenits EcoRI and BamHI restriction sites a PCR-amplified fragment contain-ing ACC-ATG nucleotides followed by the signal peptide (sp) sequencederived from the CD5 cellular gene. The pCMV/SX-GFP expression plas-mids, where SX represents the various GBV-B NS2 segments depicted inFig. 3A, were derived from pCMV-KEB-GFP by inserting between thesame sites a PCR fragment containing ACC-ATG nucleotides followed byGBV-B NS2 segments and framed by EcoRI and BamHI restriction sites.These plasmids allow the expression of GBV-B NS2 segments C-termi-nally fused to a Gly-Ser linker and GFP. Plasmid pCMV/sp-S1-GFP con-taining the GBV-B NS2 segment 1 (S1) sequence (Fig. 1) downstream ofthe CD5 sp sequence was generated by overlapping PCR. PlasmidspCMVNS21-27-GFP, pCMVNS227-59-GFP, and pCMVNS2-GFP con-taining HCV Con1 NS2 sequences were described previously (34).

Plasmids pCMV/GB-NS2(1-141)-gt(N), -gt(31), -gt(67), -gt(96), and-gt(141) were generated by an overlapping-PCR strategy designed to in-troduce an N-glycosylation acceptor coding sequence (AAC-AGT-ACC,coding for NST) flanked by 36 nucleotides encoding flexible linkers (GGSGGSGGSGSG and SQGSAPVAGSQG, respectively) (50) downstream ofeither the N-terminal residue (N) or the amino acid position indicated inparentheses within the GBV-B NS2 aa 1 to 141 sequence. The resultingPCR fragments contained EcoRI and BamHI restriction sites at their 5=and 3= extremities, respectively, which were used for cloning into pCMV-KEB-GFP at the corresponding sites. This same strategy was applied togenerate pCMV/Con1-NS2(1-145)-gt(N), -gt(27), -gt(70), -gt(99), and-gt(145), which contained the glycosylation tag coding sequence down-stream of codon 1, 27, 70, 99, or 145, respectively, within the HCV Con1NS2 aa 1 to 145 sequence. Primer-based mutagenesis and overlappingPCR were used to introduce a nucleotide substitution encoding the N(AAC)-to-Q (CAG) change within the glycosylation tag in pCMV/GB-NS2(1-141)-gt(N), pCMV/GB-NS2(1-141)-gt(67), and pCMV/Con1-NS2(1-145)-gt(N), yielding pCMV/GB-NS2(1-141)-gt(Nmut), pCMV/GB-NS2(1-141)-gt(67mut), and pCMV/HCV-NS2(1-145)-gt(Nmut),respectively.

The expression plasmids pCMV/GB-NS2-3pro-ST and pCMV/JFH1-NS2-3pro-ST were generated by amplifying overlapping PCR fragmentsthat allowed the fusion of GBV-B and HCV JFH1 sequences encodingfull-length NS2 and 190 or 213 N-terminal amino acids of NS3, respec-tively, to the CD5 sp sequence at their 5= ends and a twin Strep-tag (ST)sequence (WSHPQFEK-[GGGS]3-WSHPQFEK) at their 3= ends. Primersused for these reactions also contained a 5= extension comprising anEcoRI restriction site followed by ACC-ATG nucleotides that include astart codon and a 3= extension containing a TAA stop codon followed bya BamHI restriction site. Additionally, Ser139 codons (TCA and TCG) ofGBV-B and HCV JFH1 NS3 protease sequences, respectively (numberingrefers to amino acid positions within NS3), were mutated into GCT trip-lets by primer-based mutagenesis to replace Ser residues of the NS3 pro-tease catalytic sites with Ala residues. Derivatives of plasmids pCMV/

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7427

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

GB-NS2-3pro-ST and pCMV/JFH1-NS2-3pro-ST containing point sub-stitutions of Cys177 (TGT) and Cys184 (TGT) codons into Ala (GCC)codons within GBV-B and HCV JFH1 NS2 protease catalytic sites, respec-tively (numbering refers to amino acid positions within NS2), were alsogenerated by primer-based mutagenesis, yielding plasmids pCMV/GB-NS2(C177A)-3pro-ST and pCMV/JFH1-NS2(C184A)-3pro-ST.

Plasmid pJad was derived from pJFH1 to include cDNA substitutionsresulting in three amino acid changes in the HCV polyprotein(Val2153Ala, Val2440Leu, and Val2941Met) that have been shown to in-crease viral production in cell culture (51). Mutated sequences were de-rived from an NsiI-EcoRV restriction fragment of pJFH1-2EI3-adapt,kindly provided by R. Bartenschlager (University of Heidelberg, Heidel-berg, Germany). Primer-based mutagenesis and overlapping PCR wereused to replace the NotI-AvrII restriction fragment of pJad by a corre-sponding fragment containing an engineered mutation of the Ser168(AGT) codon within NS2 into a Glu codon (GAA), resulting in pJad/S168E.

Plasmids pHelper/NS2JFH1, -/NS2J6, -/NS2N, and -/NS2GB were gen-erated by replacing the BglII-SwaI restriction fragment of plasmidpFK_i341helperSpNS2(Jc1) (31), kindly provided by R. Bartenschlager,by nucleotide sequences encoding the C-terminal residues of p7/p13 act-ing as signal peptides and full-length NS2 amplified from HCV 2a-JFH1,HCV 2a-J6, HCV 1b-N, and GBV-B viral cDNAs and framed by BglII andSwaI engineered restriction sites. Plasmid pHelper/E2p7NS2JFH1 was sim-ilarly constructed by amplifying 2a-JFH1 nucleotide sequences encodingthe C-terminal residues of E1 acting as a signal peptide and E2–p7-NS2.Plasmids encoding chimeric NS2, pHelper/E2p7NS2-TM(123)-MAJ6,-TM(123)-MAN, -TM(123)J6, -TM(123)N, -TM(1)J6, -TM(1)N, and-TM(1)GB, were generated by using overlapping PCR and the same strat-egy for insertion between the BglII and SwaI restriction sites. These plas-mids contain substitutions of JFH1 NS2 codons 1 to 140 [TM(123)-MA],1 to 97 [TM(123)], or 1 to 28 [TM(1)] by the equivalent NS2 codons fromthe indicated viruses (HCV 2a-J6, HCV 1b-N, or GBV-B).

The sequences of all primers used and further details of these cloningprocedures will be provided upon request. PCR-amplified DNA frag-ments from selected plasmid clones were analyzed by automated nucleo-tide sequencing using capillary electrophoresis (Applied Biosystems).

Sequence analyses and structure predictions. Sequence analyseswere performed by using the Network Protein Sequence Analysis (NPSA)website (http://npsa-pbil.ibcp.fr/) (52) and the European HCV Database(53). Multiple-sequence alignments and calculations of amino acid iden-tities and similarities were performed with Clustal W (54). Protein sec-ondary structures were deduced from a large set of prediction methodsavailable at the NPSA website, including DSC, HNNC, MLRC, PHD,Predator, and SOPM. Various methods were combined for the predictionof TM segments, including DAS, HMMTOP, SOSUI, TMHMM, Tmpred,TOPPRED, and PHD-TM (available at the Expasy website [http://www.expasy.org/proteomics]). Interfacial hydrophobicity plots were gener-ated with the MPEx program, which uses whole-residue hydrophobicityscales for hydropathy analysis of membrane proteins (http://blanco.biomol.uci.edu/mpex/). These hydrophobicity scales were determinedexperimentally by measuring the partitioning of host-guest peptides intobilayer interfaces and n-octanol (55). The MPEx algorithm was demon-strated to accurately predict the TM segments of membrane proteins ofknown structure (56).

Peptide synthesis and purification. GBV-B NS2(2–32), NS2(32–57),and NS2(113–137) peptides, containing aa 2 to 32, 32 to 57, and 113 to 137of GBV-B NS2, respectively, were synthesized on a Milligen 9050 appara-tus, employing Fmoc (N-[9-fluorenyl]methoxycarbonyl) chemistry. Thepeptides were highly purified by reversed-phase high-performance liquidchromatography on a Nucleosil C18 column (120 Å; 5 �m) using a water-acetonitrile gradient containing 0.1% trifluoroacetic acid. The peaks wereidentified by mass spectroscopy as the peptides with the respective ex-pected molecular masses.

Circular dichroism. Far-UV circular dichroism (CD) spectra wererecorded on a Chirascan spectrometer (Applied Photophysics) calibratedwith 1S-(�)-10-camphorsulfonic acid. Measurements were carried out at25°C in a 0.1-cm-path-length quartz cuvette. Spectra were measured at awavelength range of 180 to 260 nm with increments of 0.2 nm, a band passof 0.5 nm, and an integration time of 1 s. Spectra were processed, baselinecorrected, smoothed, and converted with Chirascan software. Spectralunits were expressed as the mean molar ellipticity per residue by using thepeptide concentrations determined by measuring the UV light absor-bance of tyrosine and tryptophan at 280 nm. Estimation of the secondary-structure content was carried out with the DICHROWEB server (http://dichroweb.cryst.bbk.ac.uk/) (57).

Nuclear magnetic resonance spectroscopy. The purified GBV-B pep-tides were dissolved in a mixture of 50% (vol/vol) 2,2,2-trifluoroethanol-d2 (TFE-d2) (�99%) in H2O, and 2,2-dimethyl-2-silapentane-5-sulfo-nate was added to the NMR samples as an internal 1H chemical shiftreference. Multidimensional experiments were performed at 25°C on aBruker Avance 500-MHz spectrometer using standard homonuclearpulse sequences, including nuclear Overhauser enhancement (NOE)spectroscopy (NOESY) (mixing times of between 100 and 250 ms) andclean total correlation spectroscopy (TOCSY) (isotropic mixing time of80 ms), as detailed previously (58, 59). Water suppression was achieved bypresaturation. Bruker Topspin software was used to process all data, andSparky was used for spectral analysis (http://www.cgl.ucsf.edu/home/sparky/). Intraresidue backbone resonances and aliphatic side chainswere identified from homonuclear 1H TOCSY experiments and con-firmed with 1H-13C heteronuclear single-quantum correlation (HSQC)experiments in 13C natural abundance. Sequential assignments were de-termined by correlating intraresidue assignments with interresidue crosspeaks observed by bidimensional 1H NOESY. NMR-derived 1H� and 13C�

chemical shifts are reported relative to the random-coil chemical shifts inTFE (60).

NMR-derived constraints and structure calculation. NOE intensitiesused as the input for structure calculations were obtained from theNOESY spectrum recorded with a 150-ms mixing time and checked forspin diffusion on spectra recorded at lower mixing times (100 ms). NOEswere partitioned into three categories of intensities that were convertedinto distances ranging from a common lower limit of 1.8 Å to upper limitsof 2.8 Å, 3.9 Å, and 5.0 Å, respectively. Protons without stereospecificassignments were treated as pseudoatoms, and the correction factors wereadded to the upper distance constraints (61). No hydrogen bond or dihe-dral angle constraints were introduced. Three-dimensional structureswere generated from NOE distances with the standard torsion angle mo-lecular dynamics protocol in the X-plor-NIH 2.30 program (62), usingstandard force fields and default parameter sets. A set of 50 structures wasinitially calculated to widely sample the conformational space, and thestructures of low energy with no distance restraint violations were re-tained. The selected structures were compared by pairwise root meansquare deviation (RMSD) over the backbone atom coordinates (N, C�,and C=). Statistical analyses, superimposition of structures, and structuralanalyses were performed with MOLMOL (63), and the quality of theselected structures was checked with the Protein Data Bank (PDB) vali-dation server.

Cell lines. Cos-1 cells were cultured in Dulbecco’s modified Eagle’smedium (Invitrogen) supplemented with 10% fetal calf serum, 100 U/mlpenicillin, and 100 �g/ml streptomycin (DMEM-10%) at 37°C in a 5%CO2 environment. Huh-7.5 human hepatocellular carcinoma cells (64),kindly provided by C. Rice, were cultured in DMEM-10% supplementedwith nonessential amino acids and 1 mM sodium pyruvate (completeDMEM).

Antibodies. Monoclonal antibodies specific for HCV core protein(Hep C HBcAg 1851) and HCV NS3 (clone 2E3) were purchased fromSanta Cruz Biotechnology and BioFront Technologies, respectively. NS2-1519 rabbit polyclonal antibodies raised against HCV JFH1 NS2 syntheticpeptides (31) were kindly provided by R. Bartenschlager. Monoclonal

Boukadida et al.

7428 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

antibodies specific for GFP (JL-8), the V5 epitope, the Strep-tag epitope,p63 (G1/296), and GM130 (clone 35) were purchased from Clontech,Invitrogen, Qiagen, Enzo Life Sciences, and BD Biosciences, respectively.

DNA transfection. Cos-1 cells were seeded at a concentration of 2 �104 cells per well into 24-well plates and transfected 24 h later with 0.4 �gof plasmid DNA by using FuGENE 6 transfection reagent (Promega), asrecommended by the manufacturer.

In vitro transcription. Genome-length JFH1-based plasmids werelinearized with XbaI and treated with mung bean nuclease (New EnglandBioLabs), while plasmids containing helper subgenomic cDNAs were lin-earized with MluI. Linearized DNAs served as the templates for in vitrotranscription using a T7 RiboMAX Express Large Scale RNA productionsystem (Promega) according to the manufacturer’s instructions. Syn-thetic RNAs were then purified by phenol-chloroform extractions andprecipitated with isopropanol. RNA quality and quantity were monitoredfollowing 1% agarose gel electrophoresis and measurement of the absor-bance at 260 nm.

RNA transfection. Huh-7.5 cells (2 � 106 cells in 0.4 ml Opti-MEM[Life Technologies]) were transfected by electroporation with 5 �g of invitro-transcribed genome-length RNA or cotransfected with 5 �g of ge-nome-length RNA and 1.6 to 1.8 �g of helper subgenomic RNA. Cellswere electroporated in 4-mm-gap-width cuvettes by applying one pulse at240 V at 900 �F (EasyjecT Plus; Equibio) and then immediately resus-pended in complete DMEM and seeded at 6 � 105 cells per well in 6-wellplates.

HCV TCID50 titration. Huh-7.5 cells were seeded at a concentrationof 4 � 103 cells per well in 96-well plates and infected 24 h later with serialdilutions of transfected-cell supernatants harvested at 48 h posttransfec-tion. After a 3- to 4-day incubation at 37°C, cells were fixed with methanoland incubated with phosphate-buffered saline (PBS) containing 0.3%(vol/vol) H2O2 for 5 min at room temperature. Cells were then blockedfor 30 min in 2.5% horse serum, and HCV antigen-expressing cells weredetected following 1 h of incubation in PBS containing 0.1% Tween(PBS-T) and anti-NS3 monoclonal antibody at a final concentration of0.3 �g/ml. Bound anti-NS3 antibody was detected by a 30-min incubationwith peroxidase-conjugated antibodies specific for murine IgG(ImmPRESS; Vector Laboratories) diluted 1:2 in PBS-T. Peroxidase ac-tivity was detected by using the liquid DAB� substrate chromogen system(Dako). Viral titers (50% tissue culture infectious doses [TCID50]/ml)were calculated according to the method of Reed and Muench (65).

Immunoblot analysis. RNA- and DNA-transfected cells were lysed at48 h and 32 to 40 h posttransfection, respectively, in NuPAGE LDS samplebuffer (Life Technologies) containing 0.71 M 2-mercaptoethanol andheated for 10 min at 95°C. Proteins were loaded onto NuPAGE 10% or12% Bis-Tris gels (Life Technologies), separated by sodium dodecyl sul-fate-polyacrylamide gel electrophoresis (SDS-PAGE) in morpholinepro-panesulfonic acid (MOPS) SDS running buffer (Life Technologies), andtransferred onto polyvinylidene difluoride (PVDF) membranes (GEHealthcare). Membranes were saturated in PBS-T containing 5% dryskimmed milk at room temperature for 1 h, prior to incubation for 1 h atroom temperature or overnight at 4°C with one of the following mono-clonal antibodies or antisera diluted in PBS-T containing 1% dryskimmed milk: anti-C (0.1 �g/ml), anti-NS2JFH1 (1:1,000), anti-V5 (1:4,000), anti-Strep-tag (0.05 �g/ml), anti-GFP (0.125 �g/ml), or anti-p63(0.5 �g/ml). Following incubation for 1 h at room temperature with per-oxidase-conjugated anti-mouse or anti-rabbit antibodies, proteins werevisualized by using ECL Prime Western blotting detection reagent fol-lowed by exposure to Hyperfilm MP (all reagents were from GE Health-care).

Immunofluorescence. Cos-1 cells grown on glass coverslips andtransfected with pCMV-KEB-GFP-derived plasmid DNAs were fixed at40 h posttransfection with 4% paraformaldehyde. Cells were permeabil-ized with 0.2% Triton X-100 and incubated for 15 min with 5% donkeyand fetal calf sera and then for 1 h at room temperature with antibodiesspecific for p63 (1 �g/ml) or GM130 (1.25 �g/ml) diluted in PBS contain-

ing 1% donkey and fetal calf sera. Bound primary antibodies were revealedwith Alexa Fluor 555 donkey anti-murine Ig antibody (Life Technolo-gies). Coverslips were mounted in ProLong Gold Antifade reagent with4=,6-diamidino-2-phenylindole (DAPI) (Life Technologies) and exam-ined by using a fluorescence microscope with a 40� oil immersion objec-tive and an ApoTome optical sectioning device (Zeiss).

Membrane flotation assays. At 40 h posttransfection, DNA-trans-fected Cos-1 cells were subjected to Dounce homogenization in hypotoniclysis buffer (20 mM HEPES [pH 7], 100 mM NaCl) containing Completeprotease inhibitor cocktail (Roche), followed by centrifugation at 1,000 �g for 5 min to pellet nuclei, unlysed cells, and large debris. In controlexperiments (membrane fragmentation), Triton X-100 was added topostnuclear supernatants to a final concentration of 1%. Sucrose wasadded to postnuclear supernatants to a final concentration of 54% (wt/wt)in lysis buffer. The sucrose-containing lysate (0.5 ml) was overlaid with 2ml of lysis buffer containing 45% sucrose (wt/wt) and 1.2 ml of lysis bufferin an open-top thick-wall polycarbonate tube (Beckman Coulter). Equi-librium centrifugation was carried out at 285,000 � g for 100 min at 4°C inan SW55 Ti rotor (Beckman Coulter). Subsequently, 1 fraction of 1 ml, 1fraction of 0.5 ml, and 5 fractions of 0.44 ml were collected from the top ofthe gradient, and equal volumes of each fraction were subjected to immu-noblot analysis.

Glycosylation assays. At 32 h posttransfection, Cos-1 cells were lysedin glycoprotein denaturing buffer (0.5% SDS, 40 mM dithiothreitol[DTT]), heated for 10 min at 100°C, and incubated for 1 h at 37°C in theabsence or in the presence of peptide-N-glycosidase F (PNGase F) (NewEngland BioLabs) according to the manufacturer’s instructions. Reactionmixtures were then supplemented with NuPAGE LDS sample buffer con-taining 2-mercaptoethanol to a final concentration of 0.71 M, heated for 3min at 100°C, and sonicated for 10 s, and proteins were analyzed by im-munoblotting.

Accession numbers. The atomic coordinates for the NMR structureof GBV-B NS2(2–32), NS2(32–57), and NS2(113–137) and the corre-sponding NMR restraints in 50% TFE are available in the RCSB PDBunder accession numbers 2LZP (RCSB identification code 103026), 2LZQ(RCSB identification code 103027), and 2MKB (RCSB identification code103716), respectively. The chemical shifts of all peptide residues have beendeposited in BioMagResBank (BMRB) under accession numbers 18768,18769, and 19765, respectively. Viral strains considered in this study andthe respective GenBank accession numbers (in parentheses) are as fol-lows: bat hepacivirus (BHV) PDB-112 (KC796077), BHV PDB-452(KC796090), BHV PDB-829 (KC796074), GBV-B (AY243572), Guerezahepacivirus 1 (GHV-1) BWC08 (KC551800), GHV-2 BWC04(KC551802), HCV 1a-H77 (NC_004102), HCV 1b-Con1 (AJ238799),HCV 1b-N (AF139594), HCV 2a-JFH1 (AB047639), HCV 2a-J6(D00944), HCV 2b-MD2b-1 (AF238486), HCV 3a-CB (AF046866), HCV3b-TrKj (D49374), HCV 4a-ED43 (NC_009825), HCV 5a-SA13(AF064490), HCV 6a-EUHK (Y12083), HCV 6b-Th580 (NC_009827),HCV 7a-QC69 (EF108306), nonprimate/equine hepacivirus (NPHV)B10-022 (JQ434004), NPHV EF369/11 (JX948116), NPHV G1-073(JQ434002), NPHV NZP-1 (JQ434001), rodent hepacivirus (RHV) RHV-339 (KC815310), RHV NLR07-oct70 (KC411784), RHV RMU10-3382(KC411777), and RHV SAR-46 (KC411807).

RESULTSPredictions of GBV-B NS2 secondary structure. Based on se-quence alignments between GBV-B and HCV polyproteins (17),GBV-B NS2 is predicted to comprise 208 aa residues, while HCVNS2 comprises 217 aa. The GBV-B NS2 amino acid sequenceshares limited similarity with its HCV counterpart, e.g., approxi-mately 21% amino acid identity (50 and 54% amino acid similar-ities) with HCV Con1 (genotype 1b) and JFH1 (genotype 2a) NS2proteins, respectively (Fig. 1). A greater degree of amino acid iden-tity (29%) is found in the C-terminal one-third of NS2 from thesetwo hepaciviruses, including conservation of His, Glu, and Cys

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7429

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

residues that compose the catalytic triad of the HCV NS2 cysteineprotease (at positions 143, 163, and 184, respectively) (Fig. 1). Inaddition, a Pro residue with a cis conformation at position 164 ofHCV NS2, which is conserved in all HCV sequences and suspectedto have a critical role in establishing the correct geometry of theGlu 163 side chain for catalysis (38), is also conserved in GBV-BNS2 (Fig. 1). By using various secondary-structure and TM seg-ment prediction algorithms as well the MPEx algorithm, whichcalculates the likelihood of an amino acid insertion within an apo-lar environment (see Materials and Methods), five segments ofGBV-B NS2 were predicted to exhibit propensity to partition intothe membrane bilayer. These include three segments with pre-dicted TM helical passages (aa 4 to 27, 71 to 93, and 116 to 135)and two segments with potential peripheral membrane associa-tion properties (aa 33 to 47 and 99 to 109). Only two predicted TMsegments would thus be located in the GBV-B N-terminal do-main, while three TM segments have been identified in the N-ter-minal domain of HCV Con1 NS2 by subcellular localization ofGFP fusion proteins and NMR analysis of corresponding peptidestructures (31, 34). These bear roughly equivalent positions to theTM1 and TM3 segments in the HCV NS2 topological model,

whereas no TM segment equivalent to HCV NS2 TM2 was pre-dicted in GBV-B NS2 by most algorithms except partially byMPEx (Fig. 1). Surprisingly, the third TM segment in GBV-B NS2was predicted to lie in the C-terminal half of the protein (Fig. 1), ina region homologous to the cytosolic protease domain of HCVNS2, whose structure was solved by X-ray crystallography (38).Several complementary experimental approaches were under-taken in this study to assess putative membrane association andmembrane-spanning properties of the five GBV-B NS2 segmentspredicted to be membranotropic by these algorithms.

GBV-B NS2 is an autoprotease. The GBV-B NS2 N-terminalboundary was previously experimentally determined by proteinsequencing as the Phe 733 residue of the viral polyprotein (16),whereas its C terminus was only putatively predicted based onsequence alignments between GBV-B and HCV polyproteins. Theexperimental determination of the GBV-B NS2 C-terminalboundary and the identification of the protease responsible forNS2/NS3 cleavage were therefore important prerequisites for thestudy of GBV-B NS2 topology. In order to investigate whetherGBV-B NS2 also exhibits proteolytic activity based on residueshomologous to the HCV NS2 catalytic triad (Fig. 1), we designed

FIG 1 Comparison of the predicted GBV-B NS2 secondary structure with the experimentally defined HCV NS2 secondary structure. Sequences of NS2 proteinsfrom GBV-B (top half of each block) and from the Con1 and JFH1 strains of HCV subtypes 1b and 2a, respectively (bottom half of each block) were aligned withClustal W. The alignment is shown in two sequential blocks, and amino acids are numbered with respect to their positions within GBV-B and HCV NS2 proteins.Identical, highly similar, and similar residues at each position of GBV-B and HCV NS2 proteins are indicated with asterisks, colons, and dots, respectively(similarity), according to Clustal W conventions. Gaps between GBV-B and HCV NS2 sequences are indicated by hyphens. The predicted secondary structure(Pred. Struct.) is indicated in italics above the GBV-B amino acid sequence (c, coiled; e, extended; h, �-helix; t, turn; ?, undefined). Residues exhibiting apropensity to partition into the lipid bilayer according to the interfacial hydrophobicity plot calculated with MPex are underlined. Predicted TM segments(consensus TM) are indicated as a succession of T’s above the secondary-structure predictions. See Materials and Methods for details concerning the algorithmsused for these predictions. The HCV NS2 secondary structure (Sec. Struct.) is depicted below the HCV JFH1 sequence. It was established based on the NMRstructure of selected synthetic peptides representing TM segments of HCV 1b-Con1 NS2 (31, 34) as well as the crystal structure of the HCV 1a-H77 NS2C-terminal subdomain (38) and membrane association properties of selected segments of this subdomain (67). Solid and dotted boxes delineate hypothetical(GBV-B) or experimentally confirmed (HCV) TM and membrane-associated segments, respectively. Residues comprising the HCV NS2 protease catalytic triadare boxed in gray, and these conserved residues are also highlighted in the GBV-B NS2 sequence. Selected segments of GBV-B NS2 (S1 to S7) that were fused toGFP in this study are schematically represented by shaded boxes above the GBV-B sequence, with the positions of amino acid boundaries indicated at bothextremities.

Boukadida et al.

7430 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

plasmids that allowed the transient expression of NS2-NS3 pre-cursors from a cytomegalovirus (CMV) promoter. Plasmid DNAsensured the expression of wild-type (wt) GBV-B or HCV NS2-NS3 precursors (NS2-3pro-ST) spanning the 208 or 217 aa of NS2fused to the N-terminal 180 or 213 aa of NS3 that contain theNS3-4A serine protease domains of GBV-B and HCV JFH1, re-spectively, in which the catalytic Ser 139 residues were mutatedinto Ala residues. These polypeptide precursors were expresseddownstream of the CD5 signal peptide to mimic ER translocationof the NS2 N terminus and were fused at their C termini to a twinStrep-tag (ST) for identification purposes. Proteins extractedfrom transfected Cos-1 cells were separated by SDS-PAGE, andST-reactive products were detected by immunoblotting. As ex-pected, the HCV JFH1 wt NS2-3pro-ST precursor was cleaved tocompletion at the NS2/NS3 junction to generate a 25-kDa productcorresponding to NS3pro-ST (Fig. 2, right). The mutation of theCys residue of the HCV NS2 catalytic triad into an Ala residue(C184A) resulted in the detection of a 49-kDa product corre-sponding to the uncleaved precursor (Fig. 2, right). These datashowed that this approach was adapted to conclusively demon-strate NS2 proteolytic activity at the NS2/NS3 junction, as previ-ously reported for HCV H77 NS2 in in vitro translation systems(36, 37) or in transient-expression systems in U-2OS or BHK21cells (38, 66). A product that migrated slightly faster than thefull-length precursor was also detected in the case of the mutatedHCV catalytic triad (Fig. 2, right). It is compatible with posttrans-lational cleavage by an unidentified protease acting at a cryptic sitelocated within the NS2 N-terminal TM segment, as previouslyreported (31).

Similarly, upon expression of the GBV-B wt NS2-3pro-ST pre-cursor, a 23-kDa product was detected, which likely correspondedto NS3pro-ST (Fig. 2, left). This strongly suggested that cleavagealso occurred in the GBV-B NS2-3pro-ST precursor at the NS2/NS3 junction. The contribution of GBV-B NS3 protease to thiscleavage event could be ruled out thanks to the mutation of itscatalytic Ser residue. In addition, this cleavage was abrogated

when the Cys residue of the putative catalytic site of GBV-B NS2was mutated into an Ala residue (C177A), leading to the detectionof a product with a molecular mass compatible with that of thefull-length NS2-3pro-ST precursor (Fig. 2). This demonstratedthat GBV-B NS2, like HCV NS2, contains cysteine protease activ-ity with C177 as a catalytic residue.

To experimentally confirm the predicted GBV-B NS2/NS3junction, a pCMV-derived plasmid DNA containing GBV-B NS2-3pro-ST codon-optimized sequences was transfected into HEK293T cells. This permitted the expression of high levels of theGBV-B NS2-3pro-ST polypeptide, which was subsequentlycleaved by NS2 protease to give rise to NS3pro-ST, as observed forCos-1 cells (data not shown). Proteins extracted from transfectedcells were affinity purified by using Strep-Tactin beads and sepa-rated by SDS-PAGE, and the cleaved NS3pro-ST product was gelextracted and submitted for analysis on a matrix-assisted laserdesorption ionization–time of flight (MALDI-TOF) mass spec-trometer following trypsin digestion. The analysis of trypsin-cleaved peptides confirmed the location of the NS2/NS3 junctionbetween the Thr and Ala amino acids at positions 940 and 941 ofthe polyprotein (data not shown), hence providing an experimen-tal demonstration that GBV-B NS2 comprises 208 aa correspond-ing to residues 733 to 940 of the viral polyprotein.

Analysis of membrane association properties of GBV-B NS2segments. According to computer-aided secondary-structurepredictions (Fig. 1), we defined a first set of six 32- to 96-aa-longsequences that spanned one or several of the five segments withputative membrane association properties within the N-terminaltwo-thirds of GBV-B NS2 (Fig. 3A, top, black segments). To de-termine the capacity of these segments as well as of full-length NS2to associate with membranes, we generated pCMV-derived plas-mids allowing the expression of GBV-B NS2 segments fused togreen fluorescent protein (GFP). Following transient expressionin Cos-1 cells, the subcellular localization of these fusion proteinswas determined by fluorescence microscopy. While native GFPwas diffusely distributed within the cell cytosol and nucleus (Fig.3B), the fusion of GBV-B NS2 segments or of full-length NS2 toGFP led to the relocalization of the GFP fluorescence to intracel-lular membranous compartments (data not shown). These resultsconfirmed the membrane association of the GBV-B NS2 N-termi-nal segment spanning aa 1 to 141 and revealed the presencetherein of at least two membranotropic sequences comprised of aa32 to 67 and 110 to 141, respectively.

Based on the boundaries of this first set of segments, we nextdesigned a second set of 6 contiguous, 17- to 33-aa-long segments,designated S1 to S6 (Fig. 1 and 3A, bottom, blue segments), thatspanned GBV-B NS2 aa 1 to 141 and contained either a computer-predicted TM segment (S1, S4, and S6), a segment with putativeperipheral membrane association (MA) properties (S2 and S5), orpresumably no membrane-associated segment (S3). A seventhsegment (S7) spanned the C-terminal subdomain of GBV-B NS2that harbors the proteolytic activity (aa 132 to 208). Plasmids al-lowing the expression of recombinant proteins comprised of eachGBV-B NS2 segment fused to GFP were generated, and the integ-rity of all fusion proteins expressed in Cos-1 cells was verified byimmunodetection with anti-GFP antibodies (Fig. 4). The subcel-lular localization of the fusion proteins was determined by bothdirect fluorescence microscopy as well as colocalization with ER-or Golgi apparatus-resident proteins (p63 and GM130, respec-tively). A known membrane-spanning segment comprising NS2

FIG 2 GBV-B NS2 autoprotease activity. Cells were transiently transfectedwith pCMV/GB-NS2–3pro-ST, pCMV/JFH1-NS2–3pro-ST, pCMV/GB-NS2(C177A)-3pro-ST, or pCMV/JFH1-NS2(C184A)-3pro-ST DNA, respec-tively, driving the expression of native (wt) polypeptide precursors(NS2-3pro) spanning NS2 and the protease domain of NS3 from GBV-B orHCV-JFH1 fused at their C termini to a twin Strep-tag (ST) or precursorsbearing an Ala substitution of the Cys residue comprising the putative(GBV-B) (C177A) or established (HCV) (C184A) NS2 catalytic triad, as indi-cated above the gels. Uncleaved precursors and cleaved products were immu-nodetected with anti-ST antibodies following SDS-PAGE separation of trans-fected-cell extracts prepared at 40 h posttransfection and are indicated byclosed and open arrowheads, respectively. Molecular mass markers are indi-cated at the left.

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7431

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

aa 27 to 59 (TM2 segment) of the Con1 strain of HCV genotype 1b(34) was used as a control to demonstrate the relocalization of thefused GFP fluorescence to the ER compartment in these cells (Fig.3C). Among GBV-B NS2 segments, three segments showed dif-fuse fluorescence (S3, S5, and S7) (Fig. 3D), like GFP (Fig. 3B);three other segments showed fluorescence confined predomi-

nantly to the Golgi apparatus (S2, S4, and S6) (Fig. 3D); and onesegment showed ER-confined fluorescence (SP-S1) (Fig. 3D). TheS1-GFP fusion protein was expressed downstream of the signalpeptide of the cellular protein CD5 to mimic the predicted trans-location of the GBV-B NS2 N terminus into the ER lumen. In thecontext of the native polyprotein, this translocation occurs via a

FIG 3 Fluorescence analysis of the membrane association properties of GBV-B NS2 segments fused to GFP. (A) Schematic representation and membraneassociation properties of GBV-B NS2 segments expressed as a fusion with GFP. The series of GBV-B NS2 segments expressed as a C-terminal fusion with a Gly-Ser(G-S) linker and GFP, as depicted at the top, are represented in the blow-up scheme by horizontal lines spanning the corresponding NS2 sequence, with aminoacid boundaries (numbering refers to the amino acid position in NS2) indicated at both extremities. S1-GFP was expressed downstream of a heterologous signalpeptide (sp) derived from the cellular protein CD5. Putative TM and membrane-associated (MA) segments are depicted as dark and light gray boxes, respectively,with amino acid boundaries indicated at the top. Membrane association (�) or a lack of membrane association (�) properties, determined primarily byimmunofluorescence patterns in transiently transfected cells (panel D and data not shown), are indicated at the right of each segment. S1 to S7 (in blue) werestudied further. (B to D) Subcellular localization of NS2 segments fused to GFP in transiently transfected cells. Cells were transiently transfected with plasmidDNAs ensuring the expression of GFP (pCMV/GFP) (B) or fusion proteins comprised of GFP C-terminally fused to aa 27 to 59 of HCV Con1 NS2 (pCMVNS227-59-GFP DNA) (C) or fused to GBV-B NS2 sp-S1 and S2 to S7, as indicated above the images (pCMV/SX-GFP) (D). Forty hours later, cells were fixed and stainedfor ER (p63) or Golgi (GM130) compartments, as indicated at the top left corner of the images, for indirect immunofluorescence analysis. Cell nuclei werecounterstained with DAPI. Cells were observed by using an Observer Z1 inverted fluorescence microscope (Zeiss) in the ApoTome mode using a 40� objective.Representative images showing autofluorescent GFP (green), ER or Golgi compartments (red), and cell nuclei (blue) are superimposed. Bar, 20 �m.

Boukadida et al.

7432 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

signal peptide located at the C terminus of p13. Of note, the ex-pression of S1-GFP in the absence of any signal peptide presenteda diffuse distribution within transfected cells (data not shown).This could be due to the fact that the N-terminal 32 residues ofGBV-B NS2 do not exhibit intrinsic properties to associate withthe lipid bilayer in the absence of addressing to the ER membranetranslocation machinery.

Next, to confirm the data obtained by fluorescence micros-copy, the membrane association properties of the seven GBV-BNS2 segments (S1 to S7) were examined on the basis of flotation ofthe corresponding GFP fusion proteins in density gradients. Forthis purpose, transfected cells were lysed by Dounce homogeniza-tion in hypotonic buffer and placed below discontinuous 54 to 0%sucrose gradients that were subsequently ultracentrifuged to equi-librium prior to the collection of the seven fractions. An ER-resi-dent TM protein, p63, was used as a control for membrane flota-tion in all gradients. As shown by immunodetection of eachfraction with anti-p63 antibodies (Fig. 4, top gels), p63 was found

in low-density gradient fractions (fractions 2 and 3) in all cases. Itshould be noted that a proportion of p63 remained in high-den-sity fractions (fractions 6 and 7), maybe as a result of partial mem-brane fragmentation upon Dounce homogenization. In contrastto p63, native GFP was detected only in high-density fractions,corroborating the absence of a membrane association for this pro-tein (Fig. 4A, bottom). In addition, we verified that the expressionof GFP downstream of the CD5 signal peptide did not alter itsdiffuse localization, regardless of whether or not the signal peptidewas cleaved from GFP (Fig. 4A, middle). In contrast, GFP fused toa known TM segment comprising aa 1 to 27 of the Con1 strain ofHCV genotype 1b was detected in light fractions (Fig. 4B), con-firming that membrane associations of ectopically expressed pro-teins are efficiently detected by this method, as previously re-ported for Nycodenz gradients (59). GFP fusion downstream ofsp-S1, S2, S4, and S6 (Fig. 4C, D, F, and H, respectively) resulted inproteins that tightly interacted with membranes and floated. Incontrast, S5-GFP and S7-GFP (Fig. 4G and I, respectively) were

FIG 4 Membrane association properties of GBV-B NS2 segments examined by flotation gradients. Discontinuous sucrose gradients were layered on top oflysates prepared at 40 h posttransfection from Cos-1 cells transfected with plasmid DNAs ensuring the expression of GFP (pCMV/GFP) or GFP downstream ofthe CD5 signal peptide (pCMV-sp-GFP) (A) or fusion proteins comprised of GFP C-terminally fused to aa 1 to 27 of HCV Con1 NS2 (pCMVNS21–27-GFPDNA) (B) or fused to the indicated segment (sp-S1 and S2 to S7) of GBV-B NS2 (pCMV/SX-GFP) (C to I). In panel D, extracts from S2-GFP-expressing cells werealso treated with Triton X-100 (� Triton X-100) (bottom gel) prior to loading below sucrose gradients. Seven fractions (fractions 1 to 7) were collected from thetop of the density gradients following equilibrium ultracentrifugation, and the triangles at the bottom of the panels indicate the percent sucrose density gradient.The ER-resident p63 and GFP fusion protein contents of each fraction were determined following SDS-PAGE separation and immunodetection with thecorresponding antibodies, as shown in the top and bottom gels, respectively. Cellular p63, GFP, GFP fused to the CD5 signal peptide (sp-GFP), GBV-B S1-GFPfused to the CD5 signal peptide (sp-S1-GFP), HCV fusion protein [(1-27)-GFP], GBV-B SX-GFP fusion proteins (S1-GFP to S7-GFP), as well as truncated (t)forms of these fusion proteins are identified at the right. Molecular weight markers (in thousands) are indicated at the left.

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7433

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

found almost exclusively in dense fractions, confirming the ab-sence of a membrane association of these two segments. The den-sity pattern observed for the S3-GFP fusion revealed its predom-inant association with dense fractions (Fig. 4E), in agreement withits diffuse distribution within the cell cytoplasm (Fig. 3D). A mi-nor fraction of S3-GFP was found in low-density fractions, whichmay be explained by the propensity of this NS2 segment to parti-tion into an apolar environment, as suggested by interfacial hy-drophobicity predictions (Fig. 1). Moreover, the alternation ofhydrophobic and negatively charged residues, together with thepresence of several Phe residues in the S3 sequence, may be indic-ative of an amphipathic helix that might insert in-plane at themembrane interface.

Comparative analysis of transmembrane properties and ori-entation of GBV-B and HCV NS2 segments. In order to deter-mine whether GBV-B NS2 S1, S2, S4, and S6, shown to interactwith intracellular membranes (Fig. 3D and 4), were able to spanthe lipid bilayer, we used an N-glycosylation scanning approach.This approach was also applied to HCV Con1 NS2 with the objec-tive to ascertain the HCV NS2 topology model initially based onthe membrane association properties of isolated NS2 segments(34). A sequence containing an Asn-Ser-Thr (NST) N-glycosyla-tion acceptor site flanked by 12-aa linkers was inserted on bothsides of each potential TM segment within GBV-B and HCV NS2.We generated a series of pCMV-derived plasmids that allowed thetransient expression downstream of the CD5 signal peptide ofNS2 residues 1 to 145 from HCV Con1 or residues 1 to 141 fromGBV-B that contained a unique NST sequence and was C-termi-nally fused to GFP. Since the addition of oligosaccharides to NSTacceptor sites occurs specifically in the ER lumen, the determina-tion of the glycosylation status of each NS2 protein containing aunique NST sequence was expected to provide information on theglobal topology of full-length NS2 with respect to the ER mem-brane.

We first confirmed by fluorescence microscopy that the inser-tion of an NST sequence in all fusion proteins did not alter theirER association in transfected Cos-1 cells (data not shown). Pro-teins extracted from transfected cells were separated by SDS-

PAGE, and GFP-reactive products were detected by immuno-blotting. Additional products harboring an �2.5-kDa shift inmolecular mass were detected in the case of glycosylation accep-tor tag insertions at the N terminus as well as downstream ofresidues 70 and 67 of HCV and GBV-B NS2 proteins, respectively(Fig. 5A and B, lanes 1 and 7, asterisks). These additional productswere abrogated by incubation of the cellular extracts with peptide-N-glycosidase F (PNGase F) prior to SDS-PAGE (Fig. 5A and B,lanes 2 and 8), showing that the increase in molecular mass re-sulted from the addition of N-linked oligosaccharides. Moreover,glycosylation was abolished when the Asn residue within the N-glycosylation acceptor site was mutated to a Gln residue (Fig. 5A,lanes 3 and 4, and B, lanes 3, 4, 9, and 10), confirming that theinserted sequence was specifically glycosylated. It should be notedthat glycosylation remained incomplete at all ER lumen-facingsites, possibly as a result of steric hindrance, only partial accessi-bility of the acceptor sequence to the glycosylation complex,and/or saturation of the glycosylation machinery. The lack of gly-cosylation of sequences inserted downstream of residues at posi-tions 27, 99, and 145 of HCV NS2 or at positions 31, 96, and 141 ofGBV-B NS2 indicates that these residues are likely located at thecytoplasmic face of the ER membrane (Fig. 5A and B, lanes 5, 6,and 11 to 14). Furthermore, GBV-B proteins bearing two N-gly-cosylation tags either at the N terminus and downstream of aa 31or at the N terminus and downstream of aa 67 were glycosylated atonly one or both inserted sites, respectively (data not shown). Thisindicated that the length of the inserted sequences did not alter theNS2 topology and confirmed data obtained for proteins bearingthe single respective insertions. Altogether, data from this analysisof the luminal versus cytosolic localization of inserted NST se-quences strongly suggest that GBV-B NS2 S1, S2, and S4 span thelipid bilayer but that S6 does not. Importantly, glycosylation dataobtained here for HCV Con1 NS2 are in full agreement with thepreviously proposed topological model (34) and support the pres-ence of three TM segments in the N-terminal region of NS2 fromboth GBV-B and HCV (see Fig. 8).

Experimental structural analyses of GBV-B NS2 S1, S2, andS6 and deduced membrane topology model for GBV-B NS2. To

FIG 5 Glycosylation pattern of NS2-GFP fusion proteins bearing an N-glycosylation acceptor sequence. Protein extracts were prepared at 40 h posttransfectionfrom cells transfected with plasmid DNAs encoding fusion proteins comprised of GFP C-terminally fused to aa 1 to 145 of HCV Con1 NS2 (A) or aa 1 to 141 ofGBV-B NS2 (B), in which an Asn-Ser-Thr consensus acceptor sequence for N-glycosylation framed by two flexible hinges was introduced downstream of theN-terminal residue (N) or the amino acid position within the respective NS2 proteins indicated above the gels. A mutated glycosylation tag (mut) harboring Glninstead of Asn within the glycosylation acceptor site served as a control. Extracted proteins, either untreated (�) or subjected to deglycosylation by treatment withpeptide-N-glycosidase F (PNGase F) (�), were separated by SDS-PAGE and identified following immunodetection with anti-GFP antibodies. Molecular weightmarkers (in thousands) are indicated at the left. Glycosylated proteins are marked with an asterisk. Dotted lines indicate where noncontiguous lanes that belongto the same immunoblot image have been brought together, and solid lines delineate different immunoblots.

Boukadida et al.

7434 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

gain further insight into the structure and membranotropic prop-erties of S1, S2, and S6, the corresponding peptides, designatedGBV-B NS2(2–32), NS2(32–57), and NS2(113–137), respectively,were chemically synthesized and purified to homogeneity, andtheir structures were analyzed by CD and NMR spectroscopy inmembrane-mimetic environments. Attempts to chemically syn-thesize a peptide corresponding to membrane-associated S4 (aa67 to 96) failed, likely due to the presence of three cysteine residuesin its sequence.

The CD spectra of the NS2(2–32), NS2(32–57), and NS2(113–137) peptides in water indicate limited folding, with the presenceof some poorly defined secondary structures (Fig. 6A to C, left).However, solubilization of the NS2(2–32) and NS2(113–137)peptides in membrane-mimetic medium including a lysophos-pholipid (�-lysophosphatidyl choline [LPC]), detergents (n-do-decyl phosphocholine [DPC], n-dodecyl--D-maltoside [DDM],and sodium dodecyl sulfate [SDS]), or a 2,2,2-trifluoroethanol(TFE)–water mixture resulted in spectra typical of �-helical fold-ing, with two minima at 208 and 222 nm and a maximum at 192nm. The NS2(32–57) peptide also exhibits a tendency to fold intoan �-helix, at least in LPC, DPC, and DDM, but the aromatic ringsof the four Trp residues in this peptide are expected to seriouslyinterfere with the far-UV CD signal and thus prevent accuratespectrum interpretations.

Peptide samples prepared in deuterated SDS or DPC micellesdisplayed broad, poorly resolved NMR spectra, likely due to themicelle sizes. Therefore, the NMR structures of these peptideswere studied in 50% TFE-d2, which yielded well-resolved homo-and heteronuclear multidimensional spectra (data not shown).The sequential attribution of all spin systems was complete, andan overview of the sequential and medium-range NOE connec-tivities is shown in Fig. 6 (right). The patterns of NOE connectivi-ties clearly demonstrate that the main body of these peptides dis-plays most characteristics of �-helices, including strong dNN(i, i� 1) and medium d�N(i, i � 1) sequential connectivities as well asweak d�N(i, i � 2), medium or strong d�N(i, i � 3) and d�(i, i� 3), and weak d�N(i, i � 4) medium-range connectivities. TheNOE-based indications of �-helical conformation were also sup-ported by the deviation of the 1H� and 13C� chemical shifts fromrandom-coil values (data not shown). Based on the NOE-derivedinterproton distances, sets of 50 structures were calculated withX-plor, and final sets of 39, 36, and 28 low-energy structures thatfully satisfied the experimental NMR data were retained forNS2(2–32), NS2(32–57), and NS2(113–137), respectively. The re-markable structural features of the three GBV-B NS2 peptides areillustrated in the representative three-dimensional models shownin Fig. 7. The relatively short helix (aa 13 to 29) of the NS2(2–32)peptide is globally hydrophobic but exhibits a more polar sideincluding small polar residues (Fig. 7A), suggesting that it is likelyinvolved in intra- and/or intermolecular interactions with otherTM segments. The presence of some large hydrophobic residuesin its polar, unstructured, N-terminal segment (aa 5 to 12) sug-gests that this region could also be involved in interactions withother membrane-associated segments in the protein context.Overall, the physicochemical and structural features of this pep-tide are reminiscent of those observed for the N-terminal TMsegment of HCV NS2 (31).

The long helix (aa 34 to 54) observed for the GBV-B NS2(32–57) peptide (Fig. 7B) clearly exhibits an amphipathic character,with numerous polar and charged residues located on one helix

side, while the hydrophobic side includes bulky hydrophobic ar-omatic residues, notably an unusually large number of Trp resi-dues (four). These features suggest that this amphipathic helixmight interact with the membrane interface in an in-plane topol-ogy, at least transiently. Based on physicochemical considerations,a TM orientation of such an amphipathic �-helix could beachieved only upon an interaction with another complementaryTM segment neutralizing the polar and charged residues locatedin the hydrophobic core of the membrane. The overall structuralfeatures of this peptide are reminiscent of those observed for thesecond TM segment of HCV NS2 (34).

In the case of the GBV-B NS2(113–137) peptide, the hydro-phobic helix at residues 116 to 135 includes a short stretch of polarand negatively charged amino acids (GENG; aa 129 to 132), whichis likely flexible because of the presence of two glycine residues(Fig. 7C). The overall hydrophobic nature of this helix likely ex-plains its prediction as a TM segment (Fig. 1). However, the hy-drophobicity together with the dispersion of charged residues inthis peptide are compatible with its location in the NS2 cytosolicdomain (Fig. 5B), in interaction with the membrane interface (S6)(Fig. 3D and 4H).

The relatively high amino acid sequence similarities betweenthe C-terminal cytosolic domain of GBV-B NS2 and that of HCVNS2 allowed us to construct a three-dimensional homologymodel for this region (Fig. 8A), using the crystal structure of HCVNS2 as the template (38). Remarkably, the �-helix at residues 116to 135 revealed in the NS2(113–137) peptide by NMR (Fig. 7C)fits with that observed in this homology model (helix 2 [H2]) (Fig.8A, dark pink cylinder). This helix has a counterpart in HCV NS2(H2; aa 123 to 136) (Fig. 8B, dark blue cylinder), which has alsobeen shown to associate with membranes (67).

Altogether, three-dimensional structural findings for thethree GBV-B NS2 peptides analyzed (Fig. 7) are in agreementwith data concerning NS2 segment orientation with respect toER membranes (Fig. 5B). All experimental findings describedin this study concurred to demonstrate the existence of threeTM segments in the N-terminal regions of both HCV andGBV-B NS2, revealing that these two proteins share commontopological organizations in spite of limited sequence similar-ity, as modeled in Fig. 8.

Role of NS2 in morphogenesis is virus and genotype specific.The lack of a GBV-B reverse-genetics system has precluded directinvestigations of a potential involvement of NS2 in GBV-B parti-cle morphogenesis and/or egress, while HCV NS2 was shown toplay a crucial role during these steps of the viral cycle (reviewed inreference 42). Since GBV-B and HCV NS2 proteins share similartopological organizations and proteolytic activities, we investi-gated whether GBV-B NS2 was able to functionally substitute forHCV NS2 in viral morphogenesis. A first approach based on HCVJFH1/GBV-B chimeras encoding part or all of the GBV-B p13-NS2 sequences in place of the corresponding HCV p7-NS2 se-quences failed to give rise to cell culture-adapted viruses (data notshown). Toward using a trans-complementation approach, wegenerated a genome-length HCV cDNA (Jad/S168E) encoding anassembly-deficient mutation within NS2 and derived from a JFH1variant (Jad) of HCV genotype 2a, which contains highly cell cul-ture-adaptive mutations in NS5A and NS5B (51). The inactivatingS168E substitution in NS2 was chosen since it targets the NS2protease cytosolic domain and was previously shown to lead to asevere defect in virion morphogenesis in the Jc1 backbone without

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7435

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

altering genome replication (68). Similar levels of core and NS2were detected in Huh-7.5 cells transfected with in vitro-tran-scribed parental Jad or mutant Jad/S168E RNAs (Fig. 9A, top andbottom, lanes 2 and 3), indicating that the Jad/S168E RNA wasfully replicative and that the S168E substitution did not alter NS2

stability or cleavage efficiency at the NS2/NS3 junction. TCID50

titration of supernatants from Huh-7.5 cells transfected with theparental Jad RNA or with the mutated Jad/S168E RNA showedthat the NS2 S168E mutation completely abrogated infectious-virus production in this context (Fig. 9B).

FIG 6 Circular dichroism and NMR analyses of GBV-B NS2 synthetic peptides. (A to C) CD spectra and NMR data for GBV-B NS2(2–32), NS2(32–57), andNS2(113–137) synthetic peptides, respectively. Far-UV CD spectra (left) were recorded in water (H2O) or complemented with either 50% 2,2,2-trifluoroethanol(TFE), 1% L-�-lysophosphatidyl choline (LPC), or the following detergents: 100 mM sodium dodecyl sulfate (SDS), 100 mM N-dodecyl--D-maltoside (DDM),or 100 mM dodecyl phosphocholine (DPC). Summaries of sequential (i, i � 1) and medium-range (i, i � 2 to i, i � 4) NOEs of GBV-B NS2 peptides in 50% TFEare shown in the right panels. Sequential NOEs allowing the assignment of proline residues are indicated in red. Asterisks indicate that the presence of an NOEcross peak was not confirmed because of overlapping resonances or the lack of H assignment. Intensities of NOEs are indicated by the height of the bars.

Boukadida et al.

7436 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

To investigate the impact of NS2 genetic distance on trans-complementation of Jad/S168E, we studied whether the Jad/S168E mutant could be rescued by the expression in trans of NS2not only from GBV-B but also from HCV strains of the samegenotype (genotype 2a) or another genotype (genotype 1b). Wethus generated four helper RNAs encoding homologous NS2 fromHCV JFH1 or heterologous NS2 from either the J6 strain of HCVgenotype 2a, the N strain of HCV genotype 1b, or GBV-B down-stream of their respective signal peptides located at the p7 or p13 Cterminus within the backbone of a previously described bicis-tronic, subgenomic JFH1 replicon (31). These heterologous pro-teins share 87% (99%), 57% (85%), and 21% (54%) amino acididentities (amino acid similarities) with JFH1 NS2, respectively. Inorder to be able to compare the intracellular levels of these NS2

proteins expressed in trans, we introduced a V5 epitope down-stream of the NS2 N-terminal residue, which was previouslyshown to have no effect on particle assembly within the JFH1backbone (data not shown).

In Huh-7.5 cells cotransfected with the Jad/S168E assembly-deficient RNA and subgenomic helper RNAs, V5-NS2 proteinsexpressed from JFH1, J6, N, and GBV-B NS2 helper RNAs weredetected at similar levels by immunoblotting with anti-V5 anti-bodies (Fig. 9A, middle, lanes 4 to 7). This indicated that all helperRNAs replicated efficiently and that V5-NS2 proteins provided intrans were stable, warranting the use of these RNAs in rescue ex-periments.

Although NS2–S168E and V5-NS2 proteins were both de-tected in cotransfected cells (Fig. 9A, bottom, lanes 4 and 5), de-

FIG 7 NMR structure models of synthetic peptides representing putative membrane-associated segments of GBV-B NS2. (A to C) Ribbon-and-stick models ofrepresentative NMR structures of GBV-B NS2(2–32) (A), NS2(32–57) (B), and NS2(113–137) (C) peptides. The amino acid sequences of the correspondingpeptides are depicted at the top of each panel, and boxes indicate �-helix residues. Based on the NOE-derived interproton distances reported in Fig. 6, sets of 50structures were calculated with X-plor, and final sets of 39, 36, and 28 low-energy structures that fully satisfied the experimental NMR data were retained forNS2(2–32), NS2(32–57), and NS2(113–137) peptides, respectively. Superimpositions of each set of structures (data not shown) highlight the most well-organized helical part for each peptide, i.e., segments spanning residues 13 to 28, 37 to 51, and 118 to 132, exhibiting RMSDs of 0.97, 0.63, and 0.71 Å for thebackbone atoms (C=, C�, and N) for NS2(2–32), NS2(32–57), and NS2(113–137) peptides, respectively. Representative NMR structures of each peptide (PDBaccession numbers 2LZP [A], 2LZQ [B], and 2MKB [C]) are shown as ribbon-and-stick models, and the molecular surface of helices highlighting theirhydrophilic and hydrophobic sides is shown in the middle and left views. The corresponding helix projections are shown on the right. Residues are color codedaccording to Wimley-White hydrophobicity scales (55), as follows: hydrophobic residues are gray; Phe and Trp are black; the polar residues Gly, Ala, Ser, Thr,Asn, and Gln are yellow; positively and negatively charged groups of basic and acidic residues are blue and red, respectively; and His is cyan. Figures weregenerated from structure coordinates by using VMD (http://www.ks.uiuc.edu/Research/vmd/) and rendered with POV-Ray (http://www.povray.org/).

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7437

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

creases in core and NS2 protein levels expressed from the genomicJad/S168E RNA were observed, compared to protein levels in cellstransfected with Jad/S168E RNA alone (Fig. 9A, top and bottom,compare lanes 4 to 7 to lane 3). Furthermore, Jad infectious titerswere �0.5 logs lower in cells cotransfected with NS2 helper RNAsthan in cells transfected with Jad alone (Fig. 9B). This could be dueto the cytotoxicity of the helper RNA and/or to competition be-tween the two transfected RNAs, as previously described (31).

The trans-complementation efficiency was determined byTCID50 titration of cell supernatants collected at 48 h posttrans-fection. Expression of homologous JFH1 V5-NS2 in trans led to apartial rescue of infectious Jad/S168E particle production incotransfected cells (Fig. 9B), confirming that the function of NS2in particle assembly could be reliably complemented in trans, aspreviously reported (31, 69). V5-NS2 from the J6 isolate of HCVgenotype 2a was able to trans-complement the Jad/S168E defect tothe same degree as homologous JFH1 NS2 (Fig. 9B). In contrast,neither NS2 from the N isolate of HCV genotype 1b nor GBV-BNS2 could rescue the Jad/S168E assembly defect. These resultsindicate that the role of NS2 in particle morphogenesis is virus andgenotype specific but is not strain dependent.

Analysis of the trans-complementation activities of chimericNS2 proteins. Residues from both the membrane N-terminal andcytosolic C-terminal NS2 domains have been involved in HCVparticle assembly (34, 68). Our goal here was to investigatewhether the two domains could be functionally uncoupled in thisprocess. Based on the common TM1-2-3 segment organization ofGBV-B and HCV NS2 N-terminal domains and the shared pres-

ence of one membrane-interface-associated helix in the cytosolicNS2 domains (Fig. 8), we generated chimeric GBV-B/JFH1 NS2proteins in which part or all of the N-terminal TM anchors andcytosolic membrane-interacting helices were replaced byequivalent structural segments from GBV-B NS2. We then as-sessed their ability to trans-complement the assembly defect ofJad/S168E NS2, which mapped to the NS2 C-terminal region.Either the N-terminal TM segment only [TM(1); aa 1 to 28], allthree N-terminal TM segments [TM(123); aa 1 to 97], or TMand membrane-associated segments [TM(123)-MA; aa 1 to140] were replaced by GBV-B NS2 counterparts. Similar de-signs were applied to substitute JFH1 sequences for HCV se-quences from the J6 strain of genotype 2a or the N strain ofgenotype 1b, creating chimeric NS2 proteins that all containthe native JFH1 protease domain.

Corresponding helper RNAs were generated, in which chime-ric NS2 coding sequences were inserted downstream of JFH1 se-quences encoding the last C-terminal residues of E1 acting as asignal peptide followed by E2–p7, so as to express NS2 from E2–p7-NS2 precursors properly translocated into the ER lumen. Pre-vious studies indeed suggested that E2–p7-NS2 performed betterthan NS2 in trans-complementation studies (31). All chimericintra- and intergenotypic HCV NS2 proteins were stably ex-pressed, as shown by immunoblotting using antibodies directedagainst the JFH1 NS2 cytosolic domain (Fig. 10A). In contrast,among GBV-B/JFH1 chimeric NS2 proteins, only that bearing thesmallest substitution, TM(1)GB NS2, was stable (Fig. 10A and datanot shown). This suggested that the genetic distance between

FIG 8 Deduced model of GBV-B NS2 membrane topology. (A) In the model of the GBV-B NS2 membrane topology deduced from this study, TM segments 1and 2 are shown in a ribbon-and-stick representation, according to representative NMR structures of the corresponding peptides. Residues are color codedaccording to their physicochemical properties, as described in the legend of Fig. 7. TM3, for which no NMR structure is available, is represented by an orangecylinder. The three TM segments are tentatively positioned as separate entities in the membrane, and the boundaries of TM �-helices are indicated on each sideof the helices. Connecting sequences between TM segments are tentatively represented within membrane interfaces by dotted lines. The C-terminal cytosolicdomain of GBV-B NS2 is represented by a three-dimensional homology model of one dimer subunit constructed by using the crystal structure of the HCV NS2cytosolic domain as the template (PDB accession number 2HD0) (38) and the Swiss-Model automated protein structure homology modeling server (http://www.expasy.org/spdbv/). The resulting tentative model includes an �-helix spanning residues 116 to 135 (H2) with demonstrated membrane association properties,schematized as a dark pink cylinder within the plane of the membrane. (B) The HCV NS2 membrane topology model relies on the NMR structure of peptidesrepresenting the three N-terminal TM segments (31, 34) confirmed by the determination of the TM segment orientation with respect to the ER membrane (thisstudy), the three-dimensional crystallographic structure of one dimer subunit of the C-terminal cytosolic domain (38), and the membrane association of thisdomain via �-helices H1 and H2 (67).

Boukadida et al.

7438 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

GBV-B and HCV JFH1 precluded the folding and/or the stabilityof chimeric NS2 comprising an N-terminal GBV-B membraneanchor and a C-terminal JFH1 cytosolic domain. In cells cotrans-fected with the assembly-deficient Jad/S168E genomic RNA andeach E2–p7-NS2 helper RNA, Jad/S168E RNA replicated effi-ciently, as documented by the even production of core (Fig. 10B).TM(1)GB NS2, like TM(1)N NS2, was unable to trans-complementthe assembly defect of Jad/S168E (Fig. 10C), while TM(1)J6 effi-ciently rescued Jad/S168E virus production to a level �1.5 logslower than that of parental Jad (Fig. 10C). Likewise, only intra-genotypic chimeric TM(123)J6 and TM(123)-MAJ6 NS2 proteinswere able to trans-complement Jad/S168E assembly, like nativeJFH1 NS2 (Fig. 10C). Of note, TM(123)-MAJ6 NS2 acted with a�1-log-lower efficiency than that of TM (123)J6 NS2 (Fig. 10C),although they were expressed at similar levels (Fig. 10A), high-lighting the role of membrane-interface-associated segments

within the cytosolic domain of NS2 in JFH1 morphogenesis.In contrast, cotransfection with intergenotypic TM(123)N orTM(123)-MAN helper RNAs did not lead to virus production.These results highlight a genotype-specific, intramolecularconcerted action between TM(1) and the other N-terminal TMsegments and/or between the N-terminal TM region and the C-terminal protease domain of HCV NS2 during virion morphogen-esis.

FIG 9 trans-complementation of the assembly-deficient Jad/S168E mutantcontaining a point mutation in the NS2 protease domain by NS2 proteins fromHCV strains of various genotypes or from GBV-B. (A) Helper NS2 expressionlevels. Proteins were extracted at 48 h posttransfection from cells transfected inthe absence of RNA (mock), transfected with cell culture-adapted Jad genomicRNA or with Jad/S168E genomic RNA encoding the S168E substitution withinNS2, or cotransfected with Jad/S168E RNA and subgenomic V5-NS2 helperRNAs (in boldface type) expressing NS2 proteins from the indicated HCVstrains (JFH1, J6, and N) or from GBV-B (GB), which were N-terminallytagged with a V5 epitope. Proteins were separated by SDS-PAGE and analyzedwith antibodies specific for JFH1 core (top), V5 (middle), or JFH1 NS2 (bot-tom). Molecular weight markers (in thousands) are indicated at the left. Notethat NS2 expressed by Jad/S168E genomic RNA (NS2JFH1) is distinguishablefrom NS2 expressed by the subgenomic helper RNA that is fused to the V5 tag(V5-NS2JFH1/J6), therefore migrating with a higher apparent molecular mass.Also note that NS2 antibodies recognize only HCV genotype 2a (JFH1 and J6)NS2 proteins. (B) Infectivity rescue. Huh-7.5 cells were cotransfected witheither Jad (gray bars) or Jad/S168E (black bars) genomic RNA and the indi-cated V5-NS2 helper RNAs (HCV genotypes are shown in parentheses) ortransfected with genomic RNAs in the absence of helper RNA (�). Release ofinfectious viral particles was determined at 48 h posttransfection by TCID50

titration of transfected-cell supernatants in Huh-7.5 cells. Means and standarddeviations of �3 independent transfections are shown.

FIG 10 trans-complementation of the Jad/S168E assembly-deficient mutantby HCV intra- or intergenotypic chimeric NS2 or by GBV-B/HCV chimericNS2 with the native JFH1 protease domain. (A) NS2 expression from helperRNAs. Proteins were extracted at 48 h posttransfection from cells transfectedin the absence of RNA (mock) or transfected with Jad or Jad/S168E genomicRNA or with the indicated subgenomic helper replicons (in boldface type)encoding E2–p7-NS2 from JFH1 (NS2JFH1) or E2–p7-NS2 precursors in whichchimeric NS2 proteins are comprised of N-terminal TM segments 1, 2, and 3and membrane-associated segments [TM(123)-MA]; N-terminal TM seg-ments 1, 2, and 3 [TM(123)]; or N-terminal TM segment 1 [TM(1)] from theJ6 strain of HCV genotype 2a; the N strain of HCV genotype 1b, or GBV-B(GB) and the remaining NS2 sequences derived from JFH1 NS2 (see schematicrepresentations at the bottom of panel C). Proteins were separated by SDS-PAGE and analyzed with antibodies specific for JFH1 NS2. Note that NS2antibodies recognize the C-terminal protease domain of JFH1 NS2 and hencedetect all chimeric NS2 proteins. (B) Core expression from genomic RNAs.Proteins extracted from cells transfected in the absence of RNA (mock), trans-fected with Jad or Jad/S168E genomic RNAs, or cotransfected with Jad/S168ERNA and the indicated helper RNAs (in boldface type) were separated bySDS-PAGE and analyzed with antibodies specific for core. (C) Infectivity res-cue. Huh-7.5 cells were cotransfected with either parental Jad (gray bars) orassembly-deficient Jad/S168E (black bars) genomic RNAs and no helper RNA(�) or subgenomic helper replicons encoding E2–p7-NS2 precursors whichcomprise JFH1 NS2 or the indicated chimeric NS2s. These are described abovefor panel A and are schematically represented at the bottom with JFH1 andheterologous sequences depicted by gray and black boxes, respectively. Releaseof infectious viral particles was determined at 48 h posttransfection by TCID50

titration of transfected-cell supernatants in Huh-7.5 cells. Means and standarddeviations of �2 independent transfections are shown.

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7439

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

DISCUSSION

In recent years, genetic and protein-protein interaction studieshave revealed that membrane-associated NS2 is a key organizer ofHCV morphogenesis (34, 35, 43–45). However, the exact mecha-nism of action of NS2 during infectious-particle assembly andpossibly egress remains undetermined. NMR studies of short seg-ments of its hydrophobic N-terminal region previously ascer-tained the TM anchoring of NS2 from the Con1 strain of HCVgenotype 1b (31, 34). The identification of the ER lumen versusthe cytosolic localization of NS2 sequences following the insertionof N-glycosylation acceptor sites shown here (Fig. 5 and 8) fullysupports the previously proposed model for HCV NS2 membranetopology (34). Furthermore, the three approaches that were usedhere to determine the topological organization of NS2 from adistantly related hepacivirus, small New World primate GBV-B,all concurred to show that GBV-B NS2 also contains three mem-brane-spanning segments in its N-terminal region and at least onesegment peripherally associated with membranes in its C-termi-nal region. Experimental approaches included (i) the determina-tion of the membrane association properties of NS2 segments byboth fluorescence microscopy (Fig. 3) and flotation gradients (Fig.4), (ii) the identification of the topological orientation of NS2sequences with respect to the ER membrane following the inser-tion of N-glycosylation acceptor sites (Fig. 5), and (iii) resolutionof the NMR structures of selected NS2 peptides identified as beingmembrane-associated segments (Fig. 6 and 7). Therefore, theseresults revealed that, in spite of only limited amino acid sequencesimilarities with the homologous HCV protein (in the amino acididentity and similarity ranges of �20% and �50%, respectively),GBV-B NS2 remarkably shares a similar membrane topology withHCV NS2. Interestingly, the NMR structure of the TM2 segmentfrom both HCV NS2 (34) and GBV-B NS2 (Fig. 7B) revealedsimilar atypical membranous �-helices comprised of numerouspolar and charged residues located on one helix side, suggestingthat the membrane insertion of these segments is likely to be me-diated through interactions with complementary �-helices. Al-though genetic studies were not possible for GBV-B given the lackof a reverse-genetics system for this virus, in the case of HCV,direct or indirect interactions of NS2 TM2 with NS2 TM1 andwith the N-terminal TM segment of the upstream p7 protein havebeen documented by genetic analyses (34) and NMR studies (70).NMR studies of the GBV-B NS2(2–32) peptide indicated that theTM1 �-helix extends from aa 12 to 31, while its N terminus isessentially disordered (Fig. 6A and 7A). Although this �-helixdoes not fully align with HCV TM1, which extends from aa 2 to 24and includes a helix-breaking Gly residue at position 10 (31), theoverall structures of both TM1s appear similar. Together, theseresults suggest that, in spite of minor �-helix TM differences, themembrane topology summarized in Fig. 8 for GBV-B and HCV islikely to also be shared by NS2 proteins from new members of theHepacivirus genus that are phylogenetically either closer to HCV(e.g., equine hepaciviruses [8, 11]) or closer to GBV-B (e.g., ro-dent, bat, and Old World monkey hepaciviruses [11, 13, 14]) (Fig.11A). Indeed, although the amino acid sequences in the NS2membrane binding domain appear to be species specific, with theexception of some patches of well-conserved residues (aa 28 to 47and 65 to 69), the good conservation of the overall hydropathyprofiles and of secondary-structure predictions suggests the con-

servation of TM segments across all hepacivirus NS2 proteins (Fig.11B).

We also showed that NS2 from GBV-B (Fig. 2) shares withHCV NS2 (36–38) a cysteine protease activity responsible forcleavage at the NS2/NS3 junction. In recently identified hepacivi-ruses, the catalytic-dyad residues (His and Cys) as well as the thirdnegatively charged partner (Glu/Asp) and the adjacent cis-Pro res-idue are conserved (Fig. 11B), suggesting that NS2 protease func-tion is likely shared by different members of the Hepacivirus ge-nus. Subcellular localization and flotation gradient analyses ofNS2-GFP fusion proteins revealed that the C-terminal half ofGBV-B NS2 contains one segment (S6) exhibiting a propensity toassociate peripherally with membranes (Fig. 3D and 4H). Thissequence was further identified by NMR studies as an �-helixspanning aa 116 to 135, which is mainly hydrophobic but alsoincludes charged residues (Fig. 6C and 7C). The equivalent �-he-lix revealed in the HCV NS2 crystal structure (H2; aa 123 to 136)(38) (Fig. 8B) was recently shown to also interact with the mem-brane interface, most likely together with the antiparallel �-helixH1 (aa 102 to 114) (Fig. 8B), and to contain residues that arecritical for NS2 stability and efficient polyprotein processing (67).Similar helices are also predicted for NS2 proteins of recentlyidentified hepaciviruses (Fig. 11B). The conservation of such amembrane binding determinant across distantly related hepacivi-ruses is indicative of the importance of the membrane anchoringof the NS2 cytosolic domain for NS2 functions.

trans-complementation approaches revealed that an assemblydefect due to a lethal substitution in the NS2 C-terminal domain(S168E) within the backbone of a highly cell culture-adapted virusderived from the JFH1 strain of HCV genotype 2a could not becomplemented by NS2 from another hepacivirus (GBV-B) or astrain of another HCV genotype (genotype 1b) provided in transby dicistronic subgenomic HCV replicons but could be rescued byNS2 from another strain of the same HCV genotype (Fig. 9). Thesefindings extend data from a previous report showing that NS2 ofHCV genotype 1a expressed via an alphavirus replicon was notable to trans-complement an NS2 defect within the backbone ofwild-type JFH1 (69). This lack of intervirus and intergenotypecross-functionality may be explained by the disruption of NS2interactions with E2, p7, and/or NS3 that have been shown to beinvolved in HCV morphogenesis by genetic and coimmunopre-cipitation studies (34, 35, 44, 45). In addition, we found that thedefect of Jad/S168E could not be rescued by the expression in transof chimeric NS2 comprised of the JFH1 C-terminal cytosolic do-main containing the native S168 residue and various lengths of theN-terminal membrane-associated domain [TM(123)-MA, TM(123), and TM(1)] of HCV genotype 1b (Fig. 10C), although thestability of these hybrid NS2 proteins (Fig. 10A) argues against adefect in their membrane insertion or subcellular localization. Be-sides, these chimeric NS2 proteins are all expected to be able toform homodimers with the NS2–S168E mutant, given that theyshare identical protease domains, which contain residues involvedin NS2 homodimerization (38, 71). Our data thus suggest that theN-terminal membrane anchor and C-terminal cytosolic domainsof NS2 cannot be functionally uncoupled, underscoring the exis-tence of important, genotype-specific intramolecular interactionsbetween these domains during HCV morphogenesis. However,some degree of cross-genotype functionality was reported for ge-notype 1a, as shown by the rescue of JFH1–S168A mutant by chi-meric helper NS2 comprising a 1a-H77 TM(123)-MA segment

Boukadida et al.

7440 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

FIG 11 Multiple-sequence alignment of NS2 proteins from GBV-B, HCV, and new members of the Hepacivirus genus. (A) NS2 amino acid sequences of selectedmembers of the Hepacivirus genus (for details, see Materials and Methods) were aligned with Clustal W, and a phylogenetic tree was constructed by neighborjoining with the MEGA6 program. Bootstraps were performed from 2,000 replicates, and values of �70% are shown. The bar indicates the average number ofamino acid substitutions per site. The putative N- and C-terminal boundaries of NS2 proteins from BHV, GHV, NPHV, and RHV were based on predicted signalpeptidase cleavage sites by using the SignalP 4.0 server (http://www.cbs.dtu.dk/services/SignalP/) and/or sequence homology with HCV and GBV-B. Coloredcircles cluster viruses according to their experimental hosts (GBV-B) or natural hosts (other viruses), as indicated at the surface of the circle. Viral strainsconsidered for NS2 alignment (in panel B) are highlighted in the respective species-coded colors. (B) The sequences of the indicated six hepacivirus speciesrepresentatives were aligned with the T-Coffee multiple-sequence alignment tool (74), using the corresponding Web server facility (http://tcoffee.vital-it.ch/).The alignment is shown in two sequential blocks, and amino acids are numbered with respect to their positions within NS2 proteins of GBV-B (top) and HCV(bottom). Gaps between NS2 sequences are indicated by hyphens. Identical, highly similar, and similar residues at each position of the six hepacivirus NS2proteins are symbolized in the Reliability line by asterisks, colons, and dots, respectively, according to Clustal W conventions. This line also displays the localreliability of the sequence alignment according to a color code from low (blue) to high (red) reliability. Residues forming the HCV NS2 protease catalytic triad,which are identical (His and Cys) or highly conserved (Glu/Asp), are highlighted in cyan. The secondary-structure consensus (Sec.Struct.Cons) is indicated ashelical (h) (blue), extended (e) (red), or undetermined (coil [c]) (orange). In the NS2 membrane binding domain (top block), this consensus was deduced fromthe experimentally determined peptide structures for GBV-B NS2 (this study) and HCV NS2 (31, 34) and secondary-structure predictions for GHV, BHV, RHV,and NPHV. The latter predictions were made by using the Web-based algorithms HNNC, MLRC, PHD, Predator, SOPM, and SIMPA96, available at the NPSAwebsite (http://npsa-pbil.ibcp.fr). In the NS2 cytosolic domain (bottom block), the secondary-structure consensus was deduced from the crystal structure of theHCV NS2 cytosolic domain (PDB accession number 2HD0) (38) and the three-dimensional homology models for GBV-B, GHV, BHV, RHV, and NPHV NS2proteins, constructed by using the HCV NS2 crystal structure as the template and the Swiss-Model automated protein structure homology modeling server(http://www.expasy.org/spdbv/) (75). For each viral sequence, amino acids predicted to belong to hydrophobic membrane segments by dense alignment surface(DAS) analysis (76) (http://www.sbc.su.se/�miklos/DAS/) are shown in orange and red according to their DAS scores (1.7 and 2.2, respectively). Underlinedresidues correspond to predictions or experimental determinations of �-helices (blue underlining) and -sheets (red underlining). Plain and dotted boxesdelineate putative TM segments and membrane-associated �-helices (H), respectively, deduced from experimental studies of HCV NS2 (31, 34, 67) and GBV-BNS2 (this study).

July 2014 Volume 88 Number 13 jvi.asm.org 7441

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

(69). Our data are in line with a recent report showing direct orindirect interactions between the cytosolic loop joining TM1 andTM2 segments and the protease domain of NS2 that are critical forvirus production (72). It is interesting to note, however, that in-dividual residues identified by de la Fuente and colleagues in theseinteractions (L30, R32, and Y39) (72) are conserved between HCV2a-J6, 2a-JFH1, and 1b-N, suggesting that other determinants inthis region are involved. The replacement of JFH1 NS2 TM1 oronly the �-helix of TM1 by the corresponding HCV 1b-N seg-ments (Fig. 10 and data not shown) yielded stable hybrid proteinsthat did not provide functional trans-complementation. This re-sult contrasts with the successful recovery, albeit with decreasedtiters, of a JFH1/1b-Con1 chimeric virus encoding a hybrid NS2protein with the HCV 1b-Con1 TM1 �-helix sequence (34), againindicating some degree of intergenotypic cross talk depending onthe system used. In comparison, the C-terminal membrane an-chor of NS5B proved more permissive, since it could be replacedby the equivalent GBV-B sequence within the backbone of anHCV 1b-Con1 subgenomic replicon without abolishing replica-tion activities (73). Adaptive compensatory mutations within themutant replicon RNA were, however, necessary to restore parent-like activity.

In conclusion, we showed that the NS2 membrane topology,with 3 TM segments in its N-terminal region and 1 or 2 mem-brane-associated segments in its C-terminal region, is a commonorganization of distantly related hepaciviruses, pointing to com-mon functions of NS2 in the hepacivirus infectious life cycle. Thegenotype-specific, intramolecular interactions between N- and C-terminal domains of NS2 that are critical during viral particleassembly or release, but not for NS2 proteolytic activity, under-score the complex mode of action of this at least bifunctionalprotein. Further investigations are needed to determine how thesehomotypic NS2 interactions interconnect with known critical in-teractions of NS2 with other viral proteins and possibly host fac-tors during the HCV life cycle.

ACKNOWLEDGMENTS

This work was supported in part by a grant (10081/11131) from the ANRSFrensh (France Recherche, Nord & Sud, Sida-HIV et Hépatites), an au-tonomous agency at Inserm, France, and by Swiss National Science Foun-dation grant 31003A-138484. C.B. and C.M. were the recipients of Uni-versité Paris Diderot-Paris 7 and ANRS doctoral fellowships, respectively.We gratefully acknowledge financial support from the TGE RMN THCFR3050 for conducting NMR studies.

We gratefully acknowledge Eric Diesis for peptide chemical synthesesand purifications; Lelli Moreno for NMR experiments at 1 GHz; PascalLenormand (Plateforme Protéomique, IP) for mass spectrometry analy-sis; Ralf Bartenschlager, Anastassia Komarova, and Nicolas Escriou forhelpful discussions and advice; Sylvie van der Werf for continuous sup-port; and the following colleagues for sharing valuable materials: TakajiWakita (JFH1 cDNA), Ralf Bartenschlager (JFH1-2EI3-adapt cDNA andNS2 antibodies), Stanley M. Lemon (N-NS2 coding cDNA), and CharlesM. Rice (J6-NS2 coding cDNA and Huh-7.5 cells). Peptide synthesis andCD experiments were performed with the platforms Centre Commun deMicroanalyse des Protéines and Production et Analyze de Protéines fromthe UMS 3444 BioSciences Gerland-Lyon Sud.

REFERENCES1. Simons JN, Pilot-Matias TJ, Leary TP, Dawson GJ, Desai SM, Schlauder

GG, Muerhoff AS, Erker JC, Buijk SL, Chalmers ML, Van Sant CL,Mushahwar IK. 1995. Identification of two flavivirus-like genomes in theGB hepatitis agent. Proc. Natl. Acad. Sci. U. S. A. 92:3401–3405. http://dx.doi.org/10.1073/pnas.92.8.3401.

2. Bright H, Carroll AR, Watts PA, Fenton RJ. 2004. Development of a GBvirus B marmoset model and its validation with a novel series of hepatitisC virus NS3 protease inhibitors. J. Virol. 78:2062–2071. http://dx.doi.org/10.1128/JVI.78.4.2062-2071.2004.

3. Bukh J, Apgar CL, Govindarajan S, Purcell RH. 2001. Host range studiesof GB virus-B hepatitis agent, the closest relative of hepatitis C virus, inNew World monkeys and chimpanzees. J. Med. Virol. 65:694 – 697. http://dx.doi.org/10.1002/jmv.2092.

4. Lanford RE, Chavez D, Notvall L, Brasky KM. 2003. Comparison oftamarins and marmosets as hosts for GBV-B infections and the effect ofimmunosuppression on duration of viremia. Virology 311:72– 80. http://dx.doi.org/10.1016/S0042-6822(03)00193-4.

5. Iwasaki Y, Mori K, Ishii K, Maki N, Iijima S, Yoshida T, Okabayashi S,Katakai Y, Lee YJ, Saito A, Fukai H, Kimura N, Ageyama N, YoshizakiS, Suzuki T, Yasutomi Y, Miyamura T, Kannagi M, Akari H. 2011.Long-term persistent GBV-B infection and development of a chronic andprogressive hepatitis C-like disease in marmosets. Front. Microbiol. 2:240.http://dx.doi.org/10.3389/fmicb.2011.00240.

6. Martin A, Bodola F, Sangar DV, Goettge K, Popov V, Rijnbrand R,Lanford RE, Lemon SM. 2003. Chronic hepatitis associated with GB virusB persistence in a tamarin after intrahepatic inoculation of synthetic viralRNA. Proc. Natl. Acad. Sci. U. S. A. 100:9962–9967. http://dx.doi.org/10.1073/pnas.1731505100.

7. Takikawa S, Engle RE, Faulk KN, Emerson SU, Purcell RH, Bukh J.2010. Molecular evolution of GB virus B hepatitis virus during acute re-solving and persistent infections in experimentally infected tamarins. J.Gen. Virol. 91:727–733. http://dx.doi.org/10.1099/vir.0.015750-0.

8. Burbelo PD, Dubovi EJ, Simmonds P, Medina JL, Henriquez JA,Mishra N, Wagner J, Tokarz R, Cullen JM, Iadarola MJ, Rice CM,Lipkin WI, Kapoor A. 2012. Serology-enabled discovery of geneticallydiverse hepaciviruses in a new host. J. Virol. 86:6171– 6178. http://dx.doi.org/10.1128/JVI.00250-12.

9. Kapoor A, Simmonds P, Gerold G, Qaisar N, Jain K, Henriquez JA,Firth C, Hirschberg DL, Rice CM, Shields S, Lipkin WI. 2011. Charac-terization of a canine homolog of hepatitis C virus. Proc. Natl. Acad. Sci.U. S. A. 108:11608 –11613. http://dx.doi.org/10.1073/pnas.1101794108.

10. Lyons S, Kapoor A, Sharp C, Schneider BS, Wolfe ND, Culshaw G,Corcoran B, McGorum BC, Simmonds P. 2012. Nonprimate hepacivi-ruses in domestic horses, United Kingdom. Emerg. Infect. Dis. 18:1976 –1982. http://dx.doi.org/10.3201/eid1812.120498.

11. Drexler JF, Corman VM, Muller MA, Lukashev AN, Gmyl A, CoutardB, Adam A, Ritz D, Leijten LM, van Riel D, Kallies R, Klose SM,Gloza-Rausch F, Binger T, Annan A, Adu-Sarkodie Y, Oppong S,Bourgarel M, Rupp D, Hoffmann B, Schlegel M, Kummerer BM,Kruger DH, Schmidt-Chanasit J, Setien AA, Cottontail VM, Hema-chudha T, Wacharapluesadee S, Osterrieder K, Bartenschlager R, Mat-thee S, Beer M, Kuiken T, Reusken C, Leroy EM, Ulrich RG, DrostenC. 2013. Evidence for novel hepaciviruses in rodents. PLoS Pathog.9:e1003438. http://dx.doi.org/10.1371/journal.ppat.1003438.

12. Kapoor A, Simmonds P, Scheel TK, Hjelle B, Cullen JM, Burbelo PD,Chauhan LV, Duraisamy R, Sanchez Leon M, Jain K, Vandegrift KJ,Calisher CH, Rice CM, Lipkin WI. 2013. Identification of rodent ho-mologs of hepatitis C virus and pegiviruses. mBio 4(2):e00216-13. http://dx.doi.org/10.1128/mBio.00216-13.

13. Quan PL, Firth C, Conte JM, Williams SH, Zambrana-Torrelio CM,Anthony SJ, Ellison JA, Gilbert AT, Kuzmin IV, Niezgoda M, OsinubiMO, Recuenco S, Markotter W, Breiman RF, Kalemba L, Malekani J,Lindblade KA, Rostal MK, Ojeda-Flores R, Suzan G, Davis LB, BlauDM, Ogunkoya AB, Alvarez Castillo DA, Moran D, Ngam S, Akaibe D,Agwanda B, Briese T, Epstein JH, Daszak P, Rupprecht CE, Holmes EC,Lipkin WI. 2013. Bats are a major natural reservoir for hepaciviruses andpegiviruses. Proc. Natl. Acad. Sci. U. S. A. 110:8194 – 8199. http://dx.doi.org/10.1073/pnas.1303037110.

14. Lauck M, Sibley SD, Lara J, Purdy MA, Khudyakov Y, Hyeroba D,Tumukunde A, Weny G, Switzer WM, Chapman CA, Hughes AL,Friedrich TC, O’Connor DH, Goldberg TL. 2013. A novel hepaciviruswith an unusually long and intrinsically disordered NS5A protein in a wildOld World primate. J. Virol. 87:8971– 8981. http://dx.doi.org/10.1128/JVI.00888-13.

15. Bartenschlager R, Lohmann V, Penin F. 2013. The molecular and structuralbasis of advanced antiviral therapy for hepatitis C virus infection. Nat. Rev.Microbiol. 11:482–496. http://dx.doi.org/10.1038/nrmicro3046.

16. Ghibaudo D, Cohen L, Penin F, Martin A. 2004. Characterization of GB

Boukadida et al.

7442 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

virus B polyprotein processing reveals the existence of a novel 13 kDaprotein with partial homology to hepatitis C virus p7 protein. J. Biol.Chem. 279:24965–24975. http://dx.doi.org/10.1074/jbc.M401148200.

17. Muerhoff AS, Leary TP, Simons JN, Pilot-Matias TJ, Dawson GJ, ErkerJC, Chalmers ML, Schlauder GG, Desai SM, Mushahwar IK. 1995.Genomic organization of GB viruses A and B: two new members of theFlaviviridae associated with GB agent hepatitis. J. Virol. 69:5621–5630.

18. Rijnbrand R, Yang Y, Beales L, Bodola F, Goettge K, Cohen L, LanfordRE, Lemon SM, Martin A. 2005. A chimeric GB virus B with 5= nontrans-lated RNA sequence from hepatitis C virus causes hepatitis in tamarins.Hepatology 41:986 –994. http://dx.doi.org/10.1002/hep.20656.

19. Warter L, Cohen L, Benureau Y, Chavez D, Yang Y, Bodola F, LemonSM, Traboni C, Lanford RE, Martin A. 2009. A cooperative interactionbetween nontranslated RNA sequences and NS5A protein promotes invivo fitness of a chimeric hepatitis C/GB virus B. PLoS One 4:e4419. http://dx.doi.org/10.1371/journal.pone.0004419.

20. Ranjith-Kumar CT, Santos JL, Gutshall LL, Johnston VK, Lin-Goerke J,Kim MJ, Porter DJ, Maley D, Greenwood C, Earnshaw DL, Baker A, GuB, Silverman C, Sarisky RT, Kao C. 2003. Enzymatic activities of the GBvirus-B RNA-dependent RNA polymerase. Virology 312:270 –280. http://dx.doi.org/10.1016/S0042-6822(03)00247-2.

21. Scarselli E, Urbani A, Sbardellati A, Tomei L, De Francesco R, TraboniC. 1997. GB virus B and hepatitis C virus NS3 serine proteases sharesubstrate specificity. J. Virol. 71:4985– 4989.

22. Zhong WD, Ingravallo P, Wright-Minogue J, Skelton A, Uss AS, ChaseR, Yao NH, Lau JYN, Hong Z. 1999. Nucleoside triphosphatase and RNAhelicase activities associated with GB virus B nonstructural protein 3. Vi-rology 261:216 –226. http://dx.doi.org/10.1006/viro.1999.9871.

23. Hope RG, Murphy DJ, McLauchlan J. 2002. The domains required todirect core proteins of hepatitis C virus and GB virus-B to lipid dropletsshare common features with plant oleosin proteins. J. Biol. Chem. 277:4261– 4270. http://dx.doi.org/10.1074/jbc.M108798200.

24. McLauchlan J, Lemberg MK, Hope G, Martoglio B. 2002. Intramem-brane proteolysis promotes trafficking of hepatitis C virus core protein tolipid droplets. EMBO J. 21:3980 –3988. http://dx.doi.org/10.1093/emboj/cdf414.

25. Targett-Adams P, Schaller T, Hope G, Lanford RE, Lemon SM, MartinA, McLauchlan J. 2006. Signal peptide peptidase cleavage of GB virus Bcore protein is required for productive infection in vivo. J. Biol. Chem.281:29221–29227. http://dx.doi.org/10.1074/jbc.M605373200.

26. Chen Z, Benureau Y, Rijnbrand R, Yi J, Wang T, Warter L, Lanford RE,Weinman SA, Lemon SM, Martin A, Li K. 2007. GB virus B disruptsRIG-I signaling by NS3/4A-mediated cleavage of the adaptor proteinMAVS. J. Virol. 81:964 –976. http://dx.doi.org/10.1128/JVI.02076-06.

27. Meylan E, Curran J, Hofmann K, Moradpour D, Binder M, Barten-schlager R, Tschopp J. 2005. Cardif is an adaptor protein in the RIG-Iantiviral pathway and is targeted by hepatitis C virus. Nature 437:1167–1172. http://dx.doi.org/10.1038/nature04193.

28. Parera M, Martrus G, Franco S, Clotet B, Martinez MA. 2012. Caninehepacivirus NS3 serine protease can cleave the human adaptor proteinsMAVS and TRIF. PLoS One 7:e42481. http://dx.doi.org/10.1371/journal.pone.0042481.

29. Lohmann V, Korner F, Koch J, Herian U, Theilmann L, BartenschlagerR. 1999. Replication of subgenomic hepatitis C virus RNAs in a hepatomacell line. Science 285:110 –113. http://dx.doi.org/10.1126/science.285.5424.110.

30. De Tomassi A, Pizzuti M, Graziani R, Sbardellati A, Altamura S,Paonessa G, Traboni C. 2002. Cell clones selected from the Huh7 humanhepatoma cell line support efficient replication of a subgenomic GB virusB replicon. J. Virol. 76:7736 –7746. http://dx.doi.org/10.1128/JVI.76.15.7736-7746.2002.

31. Jirasko V, Montserret R, Appel N, Janvier A, Eustachi L, Brohm C,Steinmann E, Pietschmann T, Penin F, Bartenschlager R. 2008. Struc-tural and functional characterization of nonstructural protein 2 for its rolein hepatitis C virus assembly. J. Biol. Chem. 283:28546 –28562. http://dx.doi.org/10.1074/jbc.M803981200.

32. Jones CT, Murray CL, Eastman DK, Tassello J, Rice CM. 2007. HepatitisC virus p7 and NS2 proteins are essential for production of infectiousvirus. J. Virol. 81:8374 – 8383. http://dx.doi.org/10.1128/JVI.00690-07.

33. Steinmann E, Penin F, Kallis S, Patel AH, Bartenschlager R, Piet-schmann T. 2007. Hepatitis C virus p7 protein is crucial for assembly andrelease of infectious virions. PLoS Pathog. 3:e103. http://dx.doi.org/10.1371/journal.ppat.0030103.

34. Jirasko V, Montserret R, Lee JY, Gouttenoire J, Moradpour D, Penin F,Bartenschlager R. 2010. Structural and functional studies of nonstruc-tural protein 2 of the hepatitis C virus reveal its key role as organizer ofvirion assembly. PLoS Pathog. 6:e1001233. http://dx.doi.org/10.1371/journal.ppat.1001233.

35. Popescu CI, Callens N, Trinel D, Roingeard P, Moradpour D, Des-camps V, Duverlie G, Penin F, Heliot L, Rouille Y, Dubuisson J. 2011.NS2 protein of hepatitis C virus interacts with structural and non-structural proteins towards virus assembly. PLoS Pathog. 7:e1001278.http://dx.doi.org/10.1371/journal.ppat.1001278.

36. Grakoui A, McCourt DW, Wychowski C, Feinstone SM, Rice CM. 1993.A second hepatitis C virus-encoded proteinase. Proc. Natl. Acad. Sci.U. S. A. 90:10583–10587. http://dx.doi.org/10.1073/pnas.90.22.10583.

37. Hijikata M, Mizushima H, Akagi T, Mori S, Kakiuchi N, Kato N,Tanaka T, Kimura K, Shimotohno K. 1993. Two distinct proteinaseactivities required for the processing of a putative nonstructural precursorprotein of hepatitis C virus. J. Virol. 67:4665– 4675.

38. Lorenz IC, Marcotrigiano J, Dentzer TG, Rice CM. 2006. Structure ofthe catalytic domain of the hepatitis C virus NS2-3 protease. Nature 442:831– 835. http://dx.doi.org/10.1038/nature04975.

39. Schregel V, Jacobi S, Penin F, Tautz N. 2009. Hepatitis C virus NS2 is aprotease stimulated by cofactor domains in NS3. Proc. Natl. Acad. Sci.U. S. A. 106:5342–5347. http://dx.doi.org/10.1073/pnas.0810950106.

40. Miyanari Y, Atsuzawa K, Usuda N, Watashi K, Hishiki T, Zayas M,Bartenschlager R, Wakita T, Hijikata M, Shimotohno K. 2007. The lipiddroplet is an important organelle for hepatitis C virus production. Nat.Cell Biol. 9:1089 –1097. http://dx.doi.org/10.1038/ncb1631.

41. Roingeard P, Hourioux C, Blanchard E, Prensier G. 2008. Hepatitis Cvirus budding at lipid droplet-associated ER membrane visualized by 3Delectron microscopy. Histochem. Cell Biol. 130:561–566. http://dx.doi.org/10.1007/s00418-008-0447-2.

42. Bartenschlager R, Penin F, Lohmann V, Andre P. 2011. Assembly ofinfectious hepatitis C virus particles. Trends Microbiol. 19:95–103. http://dx.doi.org/10.1016/j.tim.2010.11.005.

43. Boson B, Granio O, Bartenschlager R, Cosset FL. 2011. A concertedaction of hepatitis C virus p7 and nonstructural protein 2 regulates corelocalization at the endoplasmic reticulum and virus assembly. PLoS Pat-hog. 7:e1002144. http://dx.doi.org/10.1371/journal.ppat.1002144.

44. Ma Y, Anantpadma M, Timpe JM, Shanmugam S, Singh SM, LemonSM, Yi M. 2011. Hepatitis C virus NS2 protein serves as a scaffold for virusassembly by interacting with both structural and nonstructural proteins. J.Virol. 85:86 –97. http://dx.doi.org/10.1128/JVI.01070-10.

45. Stapleford KA, Lindenbach BD. 2011. Hepatitis C virus NS2 coordinatesvirus particle assembly through physical interactions with the E1–E2 gly-coprotein and NS3-NS4A enzyme complexes. J. Virol. 85:1706 –1717.http://dx.doi.org/10.1128/JVI.02268-10.

46. Ikeda M, Yi M, Li K, Lemon SM. 2002. Selectable subgenomic andgenome-length dicistronic RNAs derived from an infectious molecularclone of the HCV-N strain of hepatitis C virus replicate efficiently in cul-tured Huh7 cells. J. Virol. 76:2997–3006. http://dx.doi.org/10.1128/JVI.76.6.2997-3006.2002.

47. Wakita T, Pietschmann T, Kato T, Date T, Miyamoto M, Zhao Z,Murthy K, Habermann A, Krausslich HG, Mizokami M, BartenschlagerR, Liang TJ. 2005. Production of infectious hepatitis C virus in tissueculture from a cloned viral genome. Nat. Med. 11:791–796. http://dx.doi.org/10.1038/nm1268.

48. Tscherne DM, Jones CT, Evans MJ, Lindenbach BD, McKeating JA,Rice CM. 2006. Time- and temperature-dependent activation of hepatitisC virus for low-pH-triggered entry. J. Virol. 80:1734 –1741. http://dx.doi.org/10.1128/JVI.80.4.1734-1741.2006.

49. Brass V, Berke JM, Montserret R, Blum HE, Penin F, Moradpour D.2008. Structural determinants for membrane association and dynamicorganization of the hepatitis C virus NS3-4A complex. Proc. Natl.Acad. Sci. U. S. A. 105:14545–14550. http://dx.doi.org/10.1073/pnas.0807298105.

50. Gouttenoire J, Castet V, Montserret R, Arora N, Raussens V, Ruyss-chaert JM, Diesis E, Blum HE, Penin F, Moradpour D. 2009. Identifi-cation of a novel determinant for membrane association in hepatitis Cvirus nonstructural protein 4B. J. Virol. 83:6257– 6268. http://dx.doi.org/10.1128/JVI.02663-08.

51. Kaul A, Woerz I, Meuleman P, Leroux-Roels G, Bartenschlager R. 2007.Cell culture adaptation of hepatitis C virus and in vivo viability of an

GBV-B NS2 Protease Activity and Membrane Topology

July 2014 Volume 88 Number 13 jvi.asm.org 7443

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from

adapted variant. J. Virol. 81:13168 –13179. http://dx.doi.org/10.1128/JVI.01362-07.

52. Combet C, Blanchet C, Geourjon C, Deleage G. 2000. NPS@: networkprotein sequence analysis. Trends Biochem. Sci. 25:147–150. http://dx.doi.org/10.1016/S0968-0004(99)01540-6.

53. Combet C, Garnier N, Charavay C, Grando D, Crisan D, Lopez J,Dehne-Garcia A, Geourjon C, Bettler E, Hulo C, Le Mercier P, Barten-schlager R, Diepolder H, Moradpour D, Pawlotsky JM, Rice CM, TrepoC, Penin F, Deleage G. 2007. euHCVdb: the European hepatitis C virusdatabase. Nucleic Acids Res. 35:D363–D366. http://dx.doi.org/10.1093/nar/gkl970.

54. Thompson JD, Higgins DG, Gibson TJ. 1994. CLUSTAL W: improvingthe sensitivity of progressive multiple sequence alignment through se-quence weighting, position-specific gap penalties and weight matrixchoice. Nucleic Acids Res. 22:4673– 4680. http://dx.doi.org/10.1093/nar/22.22.4673.

55. Wimley WC, White SH. 1996. Experimentally determined hydrophobic-ity scale for proteins at membrane interfaces. Nat. Struct. Biol. 3:842– 848.http://dx.doi.org/10.1038/nsb1096-842.

56. Snider C, Jayasinghe S, Hristova K, White SH. 2009. MPEx: a tool forexploring membrane proteins. Protein Sci. 18:2624 –2628. http://dx.doi.org/10.1002/pro.256.

57. Whitmore L, Wallace BA. 2004. DICHROWEB, an online server forprotein secondary structure analyses from circular dichroism spectro-scopic data. Nucleic Acids Res. 32:W668 –W673. http://dx.doi.org/10.1093/nar/gkh371.

58. Favier A, Brutscher B, Blackledge M, Galinier A, Deutscher J, Penin F,Marion D. 2002. Solution structure and dynamics of Crh, the Bacillussubtilis catabolite repression HPr. J. Mol. Biol. 317:131–144. http://dx.doi.org/10.1006/jmbi.2002.5397.

59. Penin F, Brass V, Appel N, Ramboarina S, Montserret R, Ficheux D,Blum HE, Bartenschlager R, Moradpour D. 2004. Structure and func-tion of the membrane anchor domain of hepatitis C virus nonstructuralprotein 5A. J. Biol. Chem. 279:40835– 40843. http://dx.doi.org/10.1074/jbc.M404761200.

60. Merutka G, Dyson HJ, Wright PE. 1995. ‘Random coil’ 1H chemicalshifts obtained as a function of temperature and trifluoroethanol concen-tration for the peptide series GGXGG. J. Biomol. NMR 5:14 –24. http://dx.doi.org/10.1007/BF00227466.

61. Wüthrich K. 1986. NMR of proteins and nucleic acids. John Wiley &Sons, Inc, New York, NY.

62. Schwieters CD, Kuszewski JJ, Tjandra N, Clore GM. 2003. The Xplor-NIH NMR molecular structure determination package. J. Magn. Reson.160:65–73. http://dx.doi.org/10.1016/S1090-7807(02)00014-9.

63. Koradi R, Billeter M, Wuthrich K. 1996. MOLMOL: a program fordisplay and analysis of macromolecular structures. J. Mol. Graph. 14:51–55, 29 –32. http://dx.doi.org/10.1016/0263-7855(96)00009-4.

64. Blight KJ, McKeating JA, Rice CM. 2002. Highly permissive cell lines forsubgenomic and genomic hepatitis C virus RNA replication. J. Virol. 76:13001–13014. http://dx.doi.org/10.1128/JVI.76.24.13001-13014.2002.

65. Reed LJ, Muench H. 1938. A simple method of estimating fifty per centendpoints. Am. J. Hyg. (Lond.) 27:493– 497.

66. Reed KE, Grakoui A, Rice CM. 1995. Hepatitis C virus-encoded NS2-3protease: cleavage-site mutagenesis and requirements for bimolecularcleavage. J. Virol. 69:4127– 4136.

67. Lange CM, Bellecave P, Dao Thi VL, Tran HTL, Penin F, MoradpourD, Gouttenoire J. 2014. Determinants for membrane association of thehepatitis C virus NS2 protease domain. J. Virol. 88:6519 – 6523. http://dx.doi.org/10.1128/JVI.00224-14.

68. Phan T, Beran RK, Peters C, Lorenz IC, Lindenbach BD. 2009.Hepatitis C virus NS2 protein contributes to virus particle assembly viaopposing epistatic interactions with the E1–E2 glycoprotein and NS3-NS4A enzyme complexes. J. Virol. 83:8379 – 8395. http://dx.doi.org/10.1128/JVI.00891-09.

69. Yi M, Ma Y, Yates J, Lemon SM. 2009. Trans-complementation of anNS2 defect in a late step in hepatitis C virus (HCV) particle assembly andmaturation. PLoS Pathog. 5:e1000403. http://dx.doi.org/10.1371/journal.ppat.1000403.

70. Cook GA, Dawson LA, Tian Y, Opella SJ. 2013. Three-dimensionalstructure and interaction studies of hepatitis C virus p7 in 1,2-dihexanoyl-sn-glycero-3-phosphocholine by solution nuclear magnetic resonance.Biochemistry 52:5295–5303. http://dx.doi.org/10.1021/bi4006623.

71. Gorzin AA, Ramsland PA, Tachedjian G, Gowans EJ. 2012. Identifica-tion of residues involved in NS2 homodimerization and elucidation oftheir impact on the HCV life cycle. J. Viral Hepat. 19:189 –198. http://dx.doi.org/10.1111/j.1365-2893.2011.01504.x.

72. de la Fuente C, Goodman Z, Rice CM. 2013. Genetic and functionalcharacterization of the N-terminal region of the hepatitis C virus NS2protein. J. Virol. 87:4130 – 4145. http://dx.doi.org/10.1128/JVI.03174-12.

73. Brass V, Gouttenoire J, Wahl A, Pal Z, Blum HE, Penin F, MoradpourD. 2010. Hepatitis C virus RNA replication requires a conserved structuralmotif within the transmembrane domain of the NS5B RNA-dependentRNA polymerase. J. Virol. 84:11580 –11584. http://dx.doi.org/10.1128/JVI.01519-10.

74. Notredame C, Higgins DG, Heringa J. 2000. T-Coffee: a novel methodfor fast and accurate multiple sequence alignment. J. Mol. Biol. 302:205–217. http://dx.doi.org/10.1006/jmbi.2000.4042.

75. Guex N, Peitsch MC. 1997. SWISS-MODEL and the Swiss-PdbViewer:an environment for comparative protein modeling. Electrophoresis 18:2714 –2723. http://dx.doi.org/10.1002/elps.1150181505.

76. Cserzo M, Wallin E, Simon I, von Heijne G, Elofsson A. 1997. Predic-tion of transmembrane alpha-helices in prokaryotic membrane proteins:the dense alignment surface method. Protein Eng. 10:673– 676. http://dx.doi.org/10.1093/protein/10.6.673.

Boukadida et al.

7444 jvi.asm.org Journal of Virology

on June 5, 2018 by guesthttp://jvi.asm

.org/D

ownloaded from