Vaccinia Virus A21 Virion Membrane Protein Is Required for Cell Entry and Fusion

We provide the initial characterization of the product of thevaccinia virus A21L (VACWR140) gene and demonstrate that itis required for cell entry and low pH-triggered membrane fusion.The A21L open reading frame, which is conserved in all sequencedmembers of the poxvirus family, encodes a protein of 117 aminoacids with an N-terminal hydrophobic domain and four invariantcysteines. Expression of the A21 protein occurred at late timesof infection and was dependent on viral DNA replication. TheA21 protein contained two intramolecular disulfide bonds, theformation of which required the vaccinia virus-encoded cytoplasmicredox pathway, and was localized on the surface of the lipoproteinmembrane of intracellular mature virions. A conditional lethalmutant, in which A21L gene expression was regulated by isopropyl-ß-D-thiogalactopyranoside,was constructed. In the absence of inducer, cell-to-cell spreadof virus did not occur, despite the formation of morphologicallynormal intracellular virions and extracellular virions withactin tails. Purified virions lacking A21 were able to bindto cells, but cores did not penetrate into the cytoplasm andsynthesize viral RNA. In addition, virions lacking A21 wereunable to mediate low pH-triggered cell-cell fusion. The A21protein, like the A28 and H2 proteins, is an essential componentof the poxvirus entry/fusion apparatus for both intracellularand extracellular virus particles.

INTRODUCTION

Top

Introduction

The mechanism used by poxviruses to enter cells is poorly understood,partly because of the extreme complexity of the members of thisfamily. Vaccinia virus (VACV), the prototypic member of thePoxviridae, contains a large double-stranded DNA genome of approximately195 kbp that encodes nearly 200 proteins with functions relatingto viral ingress, transcription, genome replication, morphogenesis,egress, and host interactions (23). Two major types of infectiousVACV particles have been characterized. The most abundant form,called the intracellular mature virion (IMV), can be isolatedfollowing cell lysis and consists of a nucleoprotein core surroundedby a lipid envelope that contains more than a dozen viral proteins(19). IMVs are very stable and may mediate spread from hostto host. A second membrane, derived from trans-Golgi or endosomalcisternae, with at least five unique viral proteins, surroundsthe IMV in the extracellular cell-associated enveloped virions(CEVs) and in the dissociated extracellular enveloped virions(EEVs) (43). CEVs and EEVs may mediate spread to adjacent andmore distal cells, respectively (6, 7, 26). There is evidencethat IMVs penetrate cells by fusing with the plasma membraneat neutral pH (8, 11, 18), although other mechanisms have beensuggested (22). In contrast, there is no evidence that the EEVmembrane fuses with the cell. Two studies suggest that EEVsenter cells by endocytosis, presumably followed by acid-mediateddisruption of the EEV membrane within intracellular vesicles(17, 46).

Many of the IMV membrane proteins have roles in virus assembly,while others have been implicated in VACV attachment and entry.The products of the A27L, D8L, and H3L genes (note that originalnames for VACV genes was based on location within a HindIIIfragment followed by L or R indicating left or right directionof transcription; the latter can be omitted when referring tothe protein product of the gene) enhance virus attachment bybinding to cell surface glycosaminoglycans (9, 15, 16, 21).An additional role for the A27 protein in fusion was proposed(47), although recent studies with a deletion mutant demonstratethat this protein is not essential for neutral pH virus entryor low pH-triggered cell-cell fusion (37, 49). In contrast,two other IMV membrane proteins, namely A28 and H2, are eachrequired for virus entry and cell-cell fusion (37, 39). BecauseA28 and H2 are necessary for infection mediated by IMVs andcell-to-cell spread mediated by CEVs and EEVs, there is likelyto be a single fusion mechanism involving only the IMV membrane(37). Furthermore, since A28 and H2 are conserved, all poxvirusesare likely to penetrate cells in a similar way. Thus far, however,no cellular receptor proteins necessary for fusion have beenidentified.

We have continued our analysis of poxvirus gene products inorder to identify and characterize additional conserved IMVtransmembrane proteins, which may have essential roles in virusassembly or entry. Here we provide the initial characterizationof the product of the A21L gene (VACWR140; accession numberAY243312) of the Western Reserve (WR) strain of VACV. First,we showed that the protein was expressed late in infection andincorporated into the membrane of IMV. Next, we constructeda recombinant VACV with an inducible A21L gene and demonstratedthat the virus has a conditional lethal phenotype. Further analysisindicated that the A21 protein was not required for virus assemblybut was essential for virus entry into cells and cell-cell fusion.Therefore, A21, like A28 and H2 (37, 39), forms part of thepoxvirus fusion apparatus required for cell entry of intracellularand extracellular virions.

MATERIALS AND METHODS

Top

Materials and Methods

Propagation and purification of VACV. All studies were carried out with the WR strain of VACV (ATCCVR-1354) or recombinant viruses derived from this strain. Virusamplification and titration were performed as previously described(12) with the addition of 50 µM isopropyl-ß-D-thiogalactopyranoside(IPTG) to the medium for vA21Li unless otherwise indicated.IMVs were purified from HeLa S3 cells by sedimentation twicethrough a 36% sucrose cushion and banded once or twice on a25 to 40% sucrose gradient essentially as described previously(12).

Recombinant virus construction. vA21i was constructed from vT7LacOI, a recombinant VACV thatpossesses an IPTG-inducible T7 RNA polymerase gene and an Escherichiacoli lac repressor gene situated within the dispensable thymidinekinase locus (2), essentially as described for vE10i (40). Thefinal PCR product used for transfection to generate vA21i consistedof the entire A21L open reading frame (ORF) (nucleotides 127905to 128258) placed under the control of a lac operator-regulatedT7 promoter containing a consensus sequence for initiation oftranslation (CGAAATTAATACGACTCACTATAGGGAATTGTGAGCGCTCACAATTCCCGCCGCCACCATG),adjacent to a copy of the enhanced green fluorescent proteingene (EGFP) (accession number AAG27429) placed under the controlof a VACV synthetic early-late promoter (AAAAATTGAAATTTTATTTTTTTTTTTTGGAATATAAATG).Initiation codons are underlined. In addition, flanking sequencesof approximately 650 bp were incorporated into the recombinantPCR product to permit homologous recombination. The final productwas assembled by recombinant PCR using Accuprime Pfx (Invitrogen)as previously described (2). BS-C-1 cells in 24-well plateswere infected with 1 PFU per cell of vT7LacOI in Optimem medium(Invitrogen) and 2 h later were transfected with 0.3 µgof purified PCR product and Lipofectamine 2000 (Invitrogen).Infected cells were harvested 24 h after infection and subjectedto three freeze-thaw cycles, and dilutions were plated on BS-C-1monolayers. Virus exhibiting green fluorescence was purifiedby five rounds of plaque isolation in the presence of 50 µMIPTG.

vA21-V5 was constructed using essentially the same protocolas described for the generation of vA21i. The recombinant PCRproduct comprised a copy of the A21L gene encoding a C-terminalV5 epitope tag (GKPIPNPLLGLDST) under the control of the nativeA21L promoter as well as a copy of the EGFP gene under the controlof a VACV late promoter and VACV flanking sequences to allowhomologous recombination. Following transfection of VACV-infectedcells, recombinant virus was identified by green fluorescenceand clonally purified during three rounds of plaque isolation.All gene insertions and modifications were confirmed by PCRand sequencing of viral DNA.

One-step virus growth curve. BS-C-1 monolayers in 24-well plates were incubated with 5 PFUper cell of vA21i or vT7LacOI for 1 h at 4°C. Followingadsorption, the cells were washed three times before incubationat 37°C in medium with or without 50 µM IPTG. Cellswere harvested and subjected to three freeze-thaw cycles andsonicated three times for 30 sec each. Titers were determinedby plaque assay as previously described (12).

Antibodies. The following mouse monoclonal antibodies (MAbs) were used:7D11 to the L1 protein (53), C3 to the A27 protein (31), anti-V5epitope (Invitrogen), and anti-protein disulfide isomerase (AffinityBioReagents, Golden, CO). Rabbit polyclonal antibodies to thefollowing VACV proteins were used: L1 (28), A4 (10), A10 (R.Doms and B. Moss, unpublished), A3 (3), H4 (1), A28 (G. Nelsonand B. Moss, unpublished). For detection of CEV, rat anti-B5RMAb (19C2) was used (35). Cy5-conjugated donkey anti-rat antibody,fluorescein isothiocyanate-conjugated goat anti-mouse antibody,and rhodamine red-X-conjugated goat anti-rabbit antibody werepurchased from Jackson Immunoresearch and used according tothe manufacturer's recommendations.

Western blot analysis. Virus- or mock-infected BS-C-1 cells were harvested, washedin phosphate-buffered saline (PBS), incubated in cell lysisbuffer containing 0.5% NP-40, 10 mM Tris-HCl, pH 7.4, 10 mMCaCl₂, 10 mM NaCl, 8 µg/ml micrococcal nuclease, and proteaseinhibitor cocktail (complete mini EDTA-free; Roche) with orwithout 20 mM N-ethylmaleimide (NEM; Sigma) on ice for 30 min,and then solubilized in lithium dodecyl sulfate sample buffer(Invitrogen) with or without NuPage sample reducing agent (Invitrogen).For some experiments, cell pellets were washed with PBS andtreated with 20 mM NEM or 20 mM 4-acetamido-4'-maleimidylstilbene-2,2'-disulfonicacid (AMS; Invitrogen) prior to cell lysis. In other cases,cell lysates were treated with the reducing agent Tris-(2-carboxyethyl)phosphine (TCEP; Sigma) prior to alkylation with NEM or AMS.All lysates were incubated at 70°C for 10 min and resolvedby sodium dodecyl sulfate-polyacrylamide gel electrophoresis(SDS-PAGE) in 10%, 12%, or 4 to 12% NuPage Bis-Tris gels (Invitrogen).For Western blot analysis of purified virions, indicated amountswere solubilized in lithium dodecyl sulfate sample buffer withor without the NuPage sample reducing agent. Lysates were transferredto nitrocellulose membranes and blocked with Tris-buffered salinesupplemented with 5% nonfat dried milk and 0.05% Tween-20 for1 h at room temperature. Subsequently, membranes were incubatedwith the appropriate primary antibody, washed, incubated withhorseradish peroxidase-conjugated secondary antibodies (Pierce),and analyzed with chemiluminescence reagents (Pierce).

Confocal microscopy. Infected HeLa cells were fixed with 3% paraformaldehyde in PBSfor 20 min at room temperature, washed three times with PBS,and quenched with 2% glycine. Subsequently, cells were permeabilizedwith 0.1% Triton X-100 for 10 min at room temperature and washedthree times with PBS. Cells were incubated with the appropriateprimary antibody or Alexa Fluor 568 phalloidin (Invitrogen)diluted in 10% complement-inactivated fetal bovine serum inPBS for 1 h at 37°C. Cells were washed three times withPBS and incubated with the appropriate fluorophore-conjugatedsecondary antibody for 1 h at 37°C. DNA was visualized withdiamidino-2-phenylindole dihydrochloride (DAPI; Invitrogen)according to the manufacturer's directions. Images were obtainedusing a Leica TCS-NT/SP2 inverted confocal microscope with attachedargon laser (Coherent Inc.).

Northern blot analysis. Infected cells were washed in PBS, and total RNA was isolatedwith an RNeasy Minikit (QIAGEN). Total RNA (5 µg) wasfractionated in glyoxal-agarose gels prepared according to themanufacturer's directions (Northernmax-Gly; Ambion). RNA specieswere transferred to Hybond N+ membranes in the presence of 20x SSC (1 x SSC is 0.15 M NaCl plus 0.015 M sodium citrate).[ ${alpha}$ -³²P]dCTP-labeled double-stranded DNA probes were generatedfrom purified PCR products by using Ready-to-go DNA labelingbeads (Amersham). Hybridization was performed in Ultrahyb (Ambion)according to the manufacturer's directions.

In vitro RNA synthesis. Purified virions were incubated at 37°C for 1 h in 50 mMTris-HCl (pH 8.0), 5 mM dithiothreitol (DTT), 10 mM MgCl₂, 0.05%NP-40, 5 mM ATP, 1 mM GTP, 1 mM CTP, 0.02 mM UTP, and 1 µCiof [ ${alpha}$ -³²P]UTP (3,000 Ci/mmol). Reaction mixtures were spottedonto DE81 filter papers, air dried, and washed five times inice-cold 10% trichloroacetic acid followed by three washes in96% ethanol. The incorporation of [ ${alpha}$ -³²P]UMP into trichloroaceticacid-insoluble material was determined by scintillation counting.

Biotinylation of purified virions. Sucrose gradient-purified virions were biotinylated with 1 mgof EZ-Link sulfo-NHS-SS-biotin [sulfosuccinimidyl 2-(biotinamido)ethyl-1,3-dithiopropionate;Pierce] per ml for 30 min at 4°C. Excess labeling reagentwas quenched according to the manufacturer's directions, andvirions were pelleted by centrifugation at 20,000 x g for 30min at 4°C. Virions were solubilized with SDS-PAGE samplebuffer, and the resulting lysates were incubated with neutravidinagarose beads (Pierce) for 1 h at 4°C. The supernatant fractionwas removed, and the biotinylated proteins were recovered byincubation of the beads with SDS-PAGE sample buffer containing50 mM DTT for 1 h at 4°C. The pellet and supernatant fractionswere analyzed by reducing SDS-PAGE and Western blotting.

Transient expression of A21. A PCR product of the A21L gene fused to DNA encoding a C-terminalV5 tag under the control of its native promoter was generatedwith the following primer pair: ATGTTCCCCATTCTACCAAAGAATTATAC(VACV nucleotides 128346 to 128374) and GGGGTCACGTAGAATCGAGACCGAGGAGAGGGTTAGGGATAGGCTTACCGGTAGTAAAAAATAAGTCAGAATATGCCC(VACV nucleotides 127908 to 127936). Nucleotides noncomplementaryto VACV sequences are underlined. The PCR product was amplifiedusing Accuprime Pfx polymerase and VACV DNA as a template, purifiedwith a QIAGEN PCR purification kit, and subcloned into pCR-BluntII-TOPO (Invitrogen). Constructs were verified by PCR and sequencing.For transfections, BS-C-1 cells in 24-well dishes were infectedwith vE10i or vT7LacOI at a multiplicity of 5. After 2 h, theinfected cells were transfected with 0.25 µg of vectorDNA and Lipofectamine 2000. At 18 h after infection, cells werecollected by centrifugation, washed in PBS, and treated withNEM or AMS.

Low pH-induced cell-cell fusion. For "fusion from without," BS-C-1 monolayers were incubatedfor 1 h at room temperature with 200 PFU per cell of purifiedvirions from cells infected with vA21i in the presence of IPTGor the corresponding optical density at 260 nm (OD₂₆₀) of virionsfrom cells infected in the absence of IPTG. Following this,cell monolayers were washed extensively and incubated with eitherpH 5.3 or pH 7.4 buffer for 2 min at room temperature beforethe buffer was replaced with medium containing 300 µgof cycloheximide (Sigma) per ml for 3 h to allow cell-cell fusion.Cells were subsequently fixed and stained with Alexa Fluor 568phalloidin and DAPI for visualization by confocal microscopy.For "fusion from within," BS-C-1 monolayers were infected with5 PFU per cell of vA21i in the presence or absence of 50 µMIPTG for 18 h and treated with either pH 5.3 or pH 7.4 bufferfor 2 min at room temperature before the buffer was replacedwith complete Eagle's minimal essential medium. Cells were stainedwith DAPI and visualized by phase and fluorescence microscopy.

RESULTS

Top

Results

Expression of the A21L gene product. The A21L (VACWR140) gene was selected for study because it belongsto the subset of ORFs that are present in all sequenced poxvirusesand encode proteins with putative transmembrane domains. Inaddition, the gene has been retained in poxviruses such as modifiedvaccinia virus Ankara (4), which have undergone numerous deletionmutations while propagated in vitro. A multiple alignment ofrepresentative A21L orthologs from each sequenced genus of vertebrateand insect poxviruses is shown in Fig. 1. Overall, the N-terminalhalves of these proteins are more highly conserved than theC-terminal halves. Four cysteine residues are notable amongthe completely conserved amino acids, and there is an additionalconserved cysteine in the transmembrane domain of chordopoxvirusorthologs. The degree of amino acid identity of the predictedN-terminal transmembrane domains of the vertebrate poxvirusA21L orthologs is relatively high, suggesting some specificinteractions in addition to membrane insertion.

View larger version (78K):
[in this window]
[in a new window]
FIG. 1. Conservation of the A21L gene in Poxviridae. Multiple alignment of A21 orthologs, including a representative from each sequenced genus of the Chordopoxvirinae and Entomopoxvirinae subfamilies. Invariant cysteines are denoted by white text upon a black background. Other identical or similar residues are indicated by shading. VAC, vaccinia virus (Orthopoxvirus); MYX, myxoma virus (Leporipoxvirus); LSD, lumpy skin disease virus (Capripoxvirus); SWP, swinepox virus (Suipoxvirus); YLD, Yaba-like disease virus (Yatapoxvirus); ORF, ORF virus (Parapoxvirus); MOC, molluscum contagiosum virus (Molluscipoxvirus); FWP, fowlpox virus (Avipoxvirus); AME, Amsacta moorei entomopoxvirus (Entomopoxvirus B); MSE, Melanoplus sanguinipes entomopoxvirus (Entomopoxvirus B).

Late promoter motifs were discerned upstream of orthologousA21L ORFs, leading to a suggestion that the proteins are expressedat late times of infection (36). To demonstrate that the predicted13.6-kDa A21 protein is expressed during VACV infection, wegenerated the recombinant vA21-V5 in which the A21L gene remainedunder the control of its putative promoter but was modifiedby addition of codons specifying a C-terminal V5 epitope tag.Selection of the recombinant VACV was facilitated by coinsertionof the EGFP ORF under the control of a late promoter adjacentto the modified A21L gene (Fig. 2A). vA21-V5 was clonally purified,and the predicted genomic modifications were verified by DNAsequencing. The epitope tag had no discernible adverse effecton virus replication as determined by plaque size or virus yield(data not shown). In lysates of BS-C-1 cells infected with vA21-V5,a protein with an apparent mass of approximately 14 kDa wasdetected by Western blotting with antibody to the V5 tag (Fig.2B). The 14-kDa protein was detected at 6 h postinfection andincreased in amount until 24 h. The protein was not detectedeven at 24 h when viral DNA replication was prevented by theaddition of AraC (Fig. 2B). Western blotting with antibody toprotein disulfide isomerase was performed as a loading control.Both the kinetics of expression and the requirement for DNAreplication correlated well with the presence of a late stagepromoter, as predicted from sequence analysis.

View larger version (22K):
[in this window]
[in a new window]
FIG. 2. Expression of A21 and formation of intramolecular disulfide bonds. (A) Diagram of a portion of the genome of recombinant vA21-V5 containing a V5 epitope encoded at the end of the A21L gene. P_A21L and P₁₁ refer to the natural promoter of the A21L gene and the promoter for the gene encoding the 11K protein, respectively. Arrows over ORFs indicate the direction of transcription. (B) Western blot showing A21 expression. BS-C-1 cells were either mock-infected (M) or infected with 5 PFU of vA21-V5 per cell in the absence or presence of AraC and disrupted in SDS-PAGE loading buffer at indicated hours postinfection. Cell lysates were analyzed by SDS-PAGE and Western blotting with an anti-V5 MAb and anti-protein disulfide isomerase (anti-PDI) as a loading control. (C) Analysis of intramolecular disulfide bonds. BS-C-1 cells were infected with 5 PFU per cell of vA21-V5 for 18 h and then disrupted in SDS-PAGE loading buffer containing NEM or AMS. Some cells were treated with the reducing agent TCEP prior to alkylation with either NEM or AMS. SDS-PAGE and Western blotting were performed using an anti-V5 MAb. (D) Dependence of disulfide bond formation on expression of E10 protein. BS-C-1 cells were infected with vE10i or vT7LacOI in the absence of IPTG. Two hours later, the cells were transfected with a plasmid encoding A21 with a C-terminal V5-epitope tag under the control of the natural A21 promoter. At 18 h postinfection, cells were disrupted in SDS-PAGE loading buffer containing NEM or AMS. SDS-PAGE and Western blotting were carried out with an anti-V5 MAb. The positions of molecular mass standards in kDa are shown on the left axis of each panel.

A21 contains intramolecular disulfide bonds. The VACV A21 protein contains 5 cysteine residues that couldpotentially form inter- and intramolecular disulfide bonds.Intermolecular disulfides seemed unlikely, as there was onlya small difference in the mobility of unreduced and TCEP-reducedA21 (Fig. 2C). The slightly more rapid mobility of the unreducedspecies, however, suggested the presence of intramolecular disulfidebonds, which can cause a more compact structure in the presenceof SDS. To investigate the presence of intramolecular disulfidebonds, cells infected with vA21-V5 were lysed in the presenceof the alkylating agents NEM or AMS. Alkylation of a singlesulfhydryl group with NEM or AMS increases the molecular massof a protein by 0.125 or 0.536 kDa, respectively. The slightretardation of mobilities observed when A21 was alkylated byNEM or AMS was most consistent with a single reactive cysteineand two intramolecular disulfides (Fig. 2C). The presence ofintramolecular disulfide bonds was supported by an additionalincrease in mobility when A21 was first reduced with TCEP andthen alkylated with NEM or AMS (Fig. 2C). Under the latter conditions,a ladder of bands extending to the maximal 2.7-kDa increasepredicted by alkylation of all five cysteines with AMS was partiallyresolved.

Formation of intramolecular disulfide bonds in A21 is dependent on the VACV-specific redox pathway. The VACV genome encodes three oxidoreductases (E10, A2.5, andG4) that act sequentially to catalyze the formation of intramoleculardisulfide bonds within the cytoplasmic domains of certain IMVmembrane proteins (38, 41, 42, 50). Repression of any one ofthe redox proteins prevents the formation of disulfide bondsin the target protein. The following experiment was designedto establish whether formation of the intramolecular disulfidesof A21 depends on the VACV-encoded redox pathway. A plasmid,comprising the entire A21L ORF with a C-terminal V5 epitopetag under the control of its native promoter, was transfectedinto BS-C-1 cells infected with a conditional lethal VACV mutantvE10i (42) in the absence of inducer so that expression of theE10L gene product did not occur. As a control, cells infectedwith the vE10i parental virus vT7LacOI were transfected in parallel;vT7LacOI expresses the E10 protein at wild-type levels in theabsence of inducer. The alkylation procedure described in theprevious section was used to discriminate between free reactivecysteines and unreactive disulfides. Cells were disrupted inthe presence of NEM or AMS, and the resulting cell lysates wereanalyzed by Western blotting. A21 was completely reduced incells infected with vE10i in the absence of inducer as shownby the complete mobility shift caused by alkylation with AMS(Fig. 2D). In contrast, there was a mixture of oxidized andreduced A21 in cells infected with the vT7LacOI control virusas seen by the double bands after alkylation with NEM or AMS.In vT7LacOI-infected cells, the unoxidized A21 probably resultedfrom over expression, as was previously reported when othergenes were expressed by transfection (38). Thus, the VACV redoxsystem is required for the formation of two intramolecular disulfidebonds in A21.

A21 is associated with the IMV membrane of mature virions. The association of A21-V5 with sucrose gradient-purified virionswas demonstrated by Western blotting (Fig. 3A). The releaseof some A21-V5 into the supernatant upon addition of the nonionicdetergent NP-40 and the release of the majority in the presenceof NP-40 and DTT paralleled that of L1, a well-characterizedIMV membrane protein (Fig. 3A). To determine whether A21 ison the surface of IMV, purified virions were treated or mock-treatedwith the membrane-impermeant biotinylating agent sulfo-NHS-SS-biotin,which selectively reacts with exposed primary amines. Afterexcess reagent was quenched, the proteins were dissociated withSDS and incubated with neutravidin beads, which bind biotin.The bound and unbound proteins were subjected to SDS-PAGE andWestern blot analysis. Binding of proteins to neutravidin wasspecific as no proteins of the mock-treated virions bound (Fig.3B). The proportion of biotinylated A21 was similar to thatof the other IMV surface proteins A27 (48), A28 (38), and D8(24) and considerably more than the three core proteins tested(Fig. 3B). The core proteins were not intrinsically resistantto biotinylation, as this occurred after the membrane was removedby NP-40 treatment (Fig. 3B).

View larger version (24K):
[in this window]
[in a new window]
FIG. 3. A21 is a membrane protein exposed on the surface of IMV. (A) Extraction of A21 with a nonionic detergent. Purified VACV (WR) or vA21-V5 virions were suspended in Tris buffer, pH 7.4, and incubated at 37°C for 30 min with or without 1% NP-40 or 50 mM DTT. The virus suspensions were then centrifuged at 20,000 x g for 30 min at 4°C and separated into pellet (P) and supernatant (S) fractions. The fractions were analyzed by SDS-PAGE and Western blotting with anti-V5 and anti-L1 MAbs. (B) Biotinylation of IMV surface proteins. Untreated or NP-40-treated purified virions were reacted with 1 mg of sulfo-NHS-SS-biotin per ml for 30 min at 4°C (BIOTIN) or mock-labeled with PBS (MOCK). Excess reagent was quenched and virions were solubilized with SDS-PAGE sample buffer and incubated with neutravidin gel for 1 h at 4°C. The unbound (U) and the bound (B) proteins were treated with SDS-PAGE sample buffer containing 50 mM DTT and analyzed by SDS-PAGE and Western blotting with antibodies to membrane proteins A28, A27, and D8 and to core proteins A10, H4, and A3.

A21 expression is required for plaque formation and the production of infectious virus. The conservation of A21 among vertebrate and invertebrate poxvirusessuggested that A21 is essential for poxvirus replication. Totest this hypothesis, we constructed vA21i, which requires IPTGfor expression of A21. As diagrammed in Fig. 4A, vA21i containsa constitutively expressed E. coli lac repressor, bacteriophageT7 RNA polymerase regulated by an E. coli lac operator, andthe A21L gene regulated by a T7 promoter and lac operator. Consequently,in the presence of IPTG, the lac repressor is inactivated andT7 RNA polymerase transcribes A21. In the absence of IPTG, thelac repressor prevents expression of T7 RNA polymerase, whichis needed for transcription of the A21L ORF. For additionalstringency, a lac operator also regulates A21 transcription.The EGFP gene was coinserted adjacent to the A21L ORF to distinguishvA21i from parental vT7lacOI (2) and aid in picking plaqueswith a fluorescence microscope. vA21i formed large plaques atthe optimal IPTG concentration (50 µM), as shown by fluorescenceor crystal violet staining (Fig. 4B). In contrast, plaque formationwas completely abolished in the absence of IPTG. Upon closeinspection, single EGFP-positive cells were detectable by fluorescencemicroscopy in cells infected without IPTG (Fig. 4B), demonstratingthat vA21i has a defect in virus replication or spread. To determinethe nature of the defect, viral replication was assessed undersingle-step growth conditions in the absence or presence ofIPTG. As illustrated in Fig. 4C, the formation of infectiousvirus was dependent on IPTG, as no significant increase in virustiter was evident 24 h postinfection in the absence of inducer.In contrast, the titer of vA21i increased nearly 100-fold overthe same period in the presence of 50 µM IPTG.

View larger version (22K):
[in this window]
[in a new window]
FIG. 4. Construction of vA21i, a conditional lethal VACV with an inducible A21L gene. (A) Diagram of a portion of the genome of vA21i. Arrows over ORFs indicate the direction of transcription. Abbreviations: lacO, E. coli lac operator; P₁₁, VACV late promoter; P_7.5, VACV early/late promoter; lacI, E. coli lac repressor gene; P_T7, bacteriophage T7 promoter. (B) Plaque formation by vA21i in the presence and absence of IPTG. BS-C-1 monolayers were infected with vA21i in the presence or absence of IPTG. At 24 h postinfection, cells were observed by fluorescence microscopy for the presence of EGFP. At 48 h postinfection, monolayers were fixed and stained with crystal violet. (C) Yields of vA21i in the presence or absence of IPTG. BS-C-1 monolayers were infected with vA21i or vT7LacOI at a multiplicity of 5 PFU per cell in the absence or presence of IPTG. The cells were harvested at indicated hours after infection, and virus titers were determined by plaque assay in the presence of 50 µM IPTG. Experiments were performed in duplicate, and data points represent the mean ± standard error.

A21 expression is not required for morphogenesis and formation of IMV and CEV. To determine if vA21i has an assembly defect under nonpermissiveconditions, BS-C-1 cells were infected with vA21i in the presenceor absence of IPTG and analyzed by transmission electron microscopy.Morphogenesis appeared entirely normal under both conditions;e.g., immature virions, IMVs, intracellular enveloped virions,and CEVs were abundant (data not shown). Confocal microscopywas used to determine whether the CEVs were associated withactin tails, which are important for cell-to-cell spread (33,34, 51, 54). HeLa cells were infected with vA21i in the presenceor absence of IPTG and stained with a MAb to the B5 protein,an outer membrane component of CEVs. The cells were then permeabilizedand stained with phalloidin to detect actin. CEVs were presentat the tips of actin-containing microvilli in the presence orabsence of IPTG (Fig. 5), indicating that the defect in spreadof vA21i could not be attributed to a failure at this step ora preceding one.

View larger version (31K):
[in this window]
[in a new window]
FIG. 5. Visualization of CEVs with actin tails. HeLa cell monolayers were infected at a multiplicity of 5 PFU per cell in the absence or presence of 50 µM IPTG for 20 h, fixed with 3% paraformaldehyde, and quenched with 2% glycine. Cell surface CEVs were immunolabeled with anti-B5R MAb and Cy5-conjugated goat anti-rat secondary antibody. Following this, cells were permeabilized by the addition of 0.1% Triton X-100 and stained with DAPI to visualize cellular and viral DNA and with Alexa Fluor 568 phalloidin to visualize filamentous actin. Arrows point to representative CEVs at the tips of actin tails.

Comparison of the structure and protein composition of A21⁺ and A21^– virions. Until now, all preparations of vA21i were made in the presenceof IPTG and then used to infect cells in the presence or absenceof inducer. Thus, the infecting virus particles always containedthe A21 protein, though no further A21 protein synthesis occurredin the absence of IPTG. In the next experiments, we purifiedIMVs from cells infected with vA21i in the absence or presenceof IPTG to produce A21^– or A21⁺ virions, respectively.The virions were indistinguishable by electron microscopy (Fig.6A) and SDS-PAGE electrophoresis (Fig. 6B), except for the absenceof the A21 protein in A21^– virions (Fig. 6C). The particle/PFUratio of A21^– virions, however, was >50-fold higherthan that of A21⁺ virions, indicating that most A21^– virionswere noninfectious. This low infectivity, which may have beendue to traces of the residual A21⁺ inoculum used to preparethe A21^– virions, was sufficient to account for the decreasein virus yield during one-step growth experiments (Fig. 4C).

View larger version (40K):
[in this window]
[in a new window]
FIG. 6. Electron microscopy and SDS-PAGE of A21^– and A21⁺ IMVs. (A) Electron micrographs of negatively stained sucrose gradient-purified A21^– and A21⁺ IMVs. Virions were deposited on grids, washed with water, and stained with 7% uranyl acetate in 50% ethanol for 30 sec. (B) SDS-PAGE of sucrose gradient-purified A21^–, A21⁺, and VACV WR IMVs. The numbers of particles were determined from the optical density and equal amounts of the three types of virions were analyzed by SDS-PAGE and silver staining. The positions and molecular masses in kDa of marker proteins are shown on the left. (C) Western blot of A21^– and A21⁺ virions. Following SDS-PAGE, Western blotting was performed with rabbit polyclonal antibody to the A21 peptide DKDRDDIPGFARSC.

A21 virions are unable to penetrate cells and mediate early gene expression. The low infectivity of A21^– virions could have many differentcauses, manifested at various stages of infection. However,a block at binding or entry seemed most likely, as A21 is anIMV membrane protein. Because poxviruses incorporate the enzymesand factors needed for transcription of early genes, viral RNAsynthesis is a good indicator of the presence of cores in thecytoplasm. Early transcription was analyzed at 2 h after infectionin the presence of AraC by Northern blotting using ³²P-labeledDNA probes complementary to the viral growth factor (C11R) orDNA polymerase (E9L) genes. Prominent bands corresponding toC11R and E9L transcripts were detected in extracts of cellsinfected with A21⁺ virus and wild-type VACV but were barelydetectable in cells infected with A21^– virus (Fig. 7A).Actin mRNA served as a loading control (Fig. 7A). We also notedthat cells infected with A21^– virions exhibited littleto no cytopathic effect in comparison to cells infected withA21⁺ virions; this observation is consistent with a reductionin early gene expression since VACV-induced cell rounding isdependent upon viral early protein synthesis (5).

View larger version (32K):
[in this window]
[in a new window]
FIG. 7. In vivo and in vitro transcription mediated by A21^– and A21⁺ virions. (A) Northern blot analysis. BS-C-1 cell monolayers were infected with 5 PFU per cell of wild type VACV WR, A21⁺ virions, or the corresponding OD₂₆₀ of A21^– virions. Total RNA was harvested at 3 h after infection and analyzed by gel electrophoresis and Northern blotting with [ ${alpha}$ -³²P]dCTP-labeled double-stranded DNA probes to C11R, E9L, and ß-actin transcripts. (B) In vitro transcription. Purified virions were incubated in reaction buffer containing NP-40, DTT, Mg⁺⁺, ribonucleoside triphosphates, and [ ${alpha}$ -³²P]UTP. Incorporation of [ ${alpha}$ -³²P]UMP into trichloroacetic acid-insoluble material was determined by scintillation counting and is expressed as counts per minute (cpm).

Although a primary defect in transcription seemed unlikely,it was necessary to demonstrate that A21^– virions arecapable of synthesizing RNA. We used an in vitro system in whichvirions are permeabilized with NP-40 detergent and RNA synthesisis measured by incorporation of ³²P-labeled ribonucleoside triphosphates(20). Little difference was found in the ability of permeabilizedA21^– and A21⁺ virions to incorporate radiolabeled ribonucleosidetriphosphates into RNA (Fig. 7B). In both cases, RNA synthesiswas proportionate to the amount of virions added to the reaction.

The data thus far suggested that the A21 protein might be requiredfor virus binding or entry. To investigate this, we used a procedureadapted from Vanderplasschen et al. (46), which depends on theinability of antibodies to bind core proteins even after permeabilizationunless the envelope is removed during entry. We incubated equivalentphysical amounts of purified A21^– or A21⁺ virions withHeLa cells at 4°C for 1 h to allow cell attachment but notpenetration. Unattached viruswas removed by washing the cellmonolayer, and the cells were incubated for 2 h at 37°Cin the presence of cycloheximide, an inhibitor of protein synthesis.Under the latter conditions, cores accumulate in the cytoplasmbut are not further uncoated. Cells were fixed and permeabilizedfollowing the 4°C and 37°C incubations and allowed tobind antibodies to the L1 membrane protein and A4 core protein.L1 staining indicated similar numbers of A21^– and A21⁺virions attached to cells after the 4°C adsorption period,demonstrating no obvious defect in cell attachment (Fig. 8).At this temperature, no cores were detected in the cytoplasm,consistent with the temperature dependence of this process (11,27). However, considerable cytoplasmic A4 staining was evidentin A21⁺-infected cells after incubation at 37°C, whereasonly rare staining occurred in A21^–-infected cells. Thus,A21^– virions are defective in a postadsorption step necessaryfor the entry of viral cores into the cytoplasm.

View larger version (28K):
[in this window]
[in a new window]
FIG. 8. Inability of A21^– virions to release cores into the cytosol. HeLa cell monolayers were incubated with 20 PFU per cell of A21⁺ virions or the corresponding OD₂₆₀ of A21^– virions at 4°C for 1 h. Cells were washed extensively and fixed or incubated for a further 2 h at 37°C in the presence of cycloheximide before fixation. Autofluorescence was quenched with 2% glycine, and cells were permeabilized with 0.1% Triton X-100. Cells were incubated with mouse anti-L1 MAb to detect IMVs on the cell surface and rabbit anti-L4 antibody to detect cytoplasmic cores, followed by fluorescein isothiocyanate-conjugated goat anti-mouse (green) and rhodamine Red-X-conjugated goat anti-rabbit (red) antibody, respectively. DNA was visualized by staining with DAPI. Optical sections were obtained by confocal microscopy and are displayed as maximum intensity projections.

A21 virions are unable to induce low pH-triggered cell-cell fusion. VACV is able to induce cells to undergo two types of low pH-triggeredsyncytium formation, termed fusion from within and fusion fromwithout (11, 14). The terms are somewhat misleading becausethere is abundant evidence to indicate that in each case fusionis mediated by virions on the cell surface. The difference isthat fusion from within is triggered by briefly lowering thepH at late times of infection when progeny CEVs are on the surface,whereas fusion from without is accomplished by adding largeamounts of purified IMVs to cells and then briefly loweringthe pH. To analyze fusion from without, BS-C-1 cells were incubatedwith 200 PFU per cell of A21⁺ virions or a corresponding physicalamount of A21^– virions, treated briefly with pH 5.3 buffer,and incubated at 37°C for 3 h in the presence of cycloheximideto prevent viral protein synthesis and cytopathic effects. Cellswere then fixed and stained with DAPI to visualize nuclear DNAand with Alexa Fluor 568 phalloidin to visualize filamentousactin, since syncytia formation is accompanied by extensiveactin rearrangement. BS-C-1 cell monolayers inoculated withA21⁺ virions exhibited extensive cell-cell fusion after lowpH triggering, whereas cell monolayers that had been inoculatedwith A21^– virions did not (Fig. 9A). To analyze fusionfrom within, cells were infected with 5 PFU per cell of vA21iin the presence or absence of IPTG, and a brief low pH treatmentwas administered at 20 h postinfection. Cells were returnedto neutral pH, incubated for 3 h, fixed, and viewed by phaseand fluorescence microscopy. As vA21i constitutively expressesEGFP, content mixing in syncytia was detected as uniform greenfluorescence throughout the cytoplasm of cells that were infectedin the presence of IPTG (Fig. 9B). In contrast, fusion of cellsinfected in the absence of IPTG was not observed. Thus, A21was required for entry of viruses into cells and for low pH-triggeredcell-cell fusion.

View larger version (67K):
[in this window]
[in a new window]
FIG. 9. Inability of virions lacking A21 to mediate low pH-triggered cell-cell fusion. (A) Fusion from without. B-SC-1 monolayers were infected with either 200 PFU per cell of A21⁺ virions or the corresponding OD₂₆₀ of A21^– virions. Following this, a transient low pH treatment (pH 5.3) or control treatment (pH 7.4) was administered, and then the buffer was replaced with regular medium containing 300 µg of cycloheximide per ml. After 3 h, the cells were fixed and stained with Alexa Fluor 568 phalloidin and DAPI before visualization by confocal microscopy. (B) Fusion from within. BS-C-1 monolayers were infected with 5 PFU per cell of vA21i in the presence or absence of 50 µM IPTG for 18 h and transiently treated with either pH 5.3 or pH 7.4 buffer. The buffer was replaced with regular medium, and after 3 h the cells were stained with DAPI and visualized by phase microscopy and fluorescence microscopy for DNA and constitutively expressed EGFP.

DISCUSSION

Top

Discussion

We have taken a genetic approach to identify and characterizeproteins that are essential for poxvirus replication. Our strategyhas been to identify genes with orthologs in all sequenced poxvirusesand then to construct inducible mutants. This approach successfullyled to the identification of the A28 and H2 entry/fusion proteins(37, 38). We noted that the previously uncharacterized A21Lgene is conserved in all poxviruses and that the predicted productclosely resembles A28 and H2 in having an N-terminal hydrophobicdomain and four invariant cysteine residues. Our studies indicatedfurther similarities in that A21 was expressed late in infection,contained two intramolecular disulfide bonds formed by the VACV-encodedredox system (41), and was incorporated into the IMV membrane.

A functional relationship between A21, A28, and H2 was establishedby making an inducible A21L mutant. Replication and spread ofthe latter mutant was entirely dependent on the addition ofan inducer. Under nonpermissive conditions, many mutants withdefects in expression of IMV membrane proteins exhibit blocksin virus assembly or maturation. Examples include the A9 (55),A14 (32, 45), A17 (30, 52), and L1 (29) proteins. A28 and H2mutants, however, are notable exceptions as morphologicallynormal but noninfectious virions are formed (37, 38). Similarly,we found that the A21 mutant exhibited no obvious defect inmorphogenesis but produced noninfectious IMV, intracellularenveloped virion, and CEV progeny in the absence of inducer.Virions lacking A21 could bind to cells, but cores were notdetected in the cytosol and viral RNA was not synthesized. Furthermore,virions lacking A21 did not induce low pH-triggered cell-cellfusion. Thus, the phenotype of the A21 null mutant was indistinguishablefrom that of the previously described A28 and H2 null mutants(37, 38), suggesting that all three are part of the poxvirusentry/fusion apparatus.

Another interesting finding is that each of the entry/fusionproteins identified so far has intramolecular disulfide bondsthat are formed by the poxvirus cytoplasmic redox pathway (41).An ancient relationship between the fusion and redox proteinsis suggested by their conservation in all poxviruses. The redoxpathway was discovered during efforts to explain how the intramoleculardisulfide bonds of the L1 IMV membrane protein are formed (42).Antibodies to L1 block virus entry (25, 53), but we do not knowwhether the L1 protein is directly involved in entry/fusionbecause null mutants are impaired in virus assembly (29).

Our finding of a third poxvirus entry/fusion protein locatedin the IMV membrane reinforces our contention that only thismembrane is fusogenic (37, 38). Evidently, the outer membraneof extracellular virions is disrupted prior to fusion of IMVswith the cell membrane. Nevertheless, how A28, H2, and A21 mediatefusion is not yet understood. The physical interaction of A28and H2 was demonstrated previously (37), and preliminary dataindicate that A21 is also a part of the complex (T. G. Senkevich,unpublished). A requirement for three or more viral gene productsfor entry and fusion is unusual but not unprecedented, as thisoccurs with herpesviruses. For example, cooperative interactionbetween gB, gD, gH, and gL is required for herpes simplex virusentry though not for cell attachment (44). We presume that thepoxvirus fusion complex exists in a metastable prefusion state,as do fusion complexes of other viruses (13). There are twowell-recognized methods of activation: exposure to low pH andspecific interaction with target cell receptors at neutral pH.Low pH triggers VACV-induced syncytia formation (11, 14) andaccelerates virus entry (17; A. C. Townsley, unpublished). Nevertheless,Vanderplasschen et al. (46) presented evidence that IMV entryis not dependent on a low pH pathway, although EEV entry is.These data suggest that poxviruses may use both low pH and receptor-activatedmechanisms. However, cellular receptors for poxviruses havenot yet been identified. The result of activation is insertionof a viral fusion peptide into the cell membrane (13). Whetherthe putative poxvirus fusion peptide exists in one of the poxvirusfusion proteins identified thus far or in another protein yetto be discovered remains to be determined.

ACKNOWLEDGMENTS

We thank Norman Cooper for providing cells, Andrea Weisbergfor electron microscopy, and Owen Schwartz for help with confocalmicroscopy.

FOOTNOTES

* Corresponding author. Mailing address: Laboratory of Viral Diseases, National Institutes of Health, 4 Center Dr., MSC 0445, Bethesda, MD 20892-0445. Phone: (301) 496-9869. Fax: (301) 480-1147. E-mail: bmoss{at}nih.gov.

REFERENCES

Top