|Home | About | Journals | Submit | Contact Us | Français|
Poxviruses have an elaborate system for infecting cells comprising several proteins for attachment and a larger number dedicated to membrane fusion and entry. Thus far, 11 proteins have been identified as components of the vaccinia virus (VACV) entry-fusion complex (EFC), and 10 of these proteins have been shown to be required for entry. J5, the remaining functionally uncharacterized component of the complex, is conserved in all poxviruses, has a predicted C-terminal transmembrane domain, and is an N-terminally truncated paralog of two other EFC proteins. To determine the role of J5, we constructed a mutant that inducibly regulates J5 transcription. Although the virus yield was reduced only about 80% without inducer, the inability to isolate a J5 deletion mutant suggested an essential function. To enhance stringency, we employed RNA silencing alone and together with transcriptional repression of the inducible mutant. The yield of infectious virus was reduced 4- to 5-fold by repression, 2-fold by silencing, and 60-fold by the combination of the two. Virus particles made under the latter conditions appeared to contain a full complement of proteins excluding J5 but had very low infectivity. Further studies indicated that after binding to cells, J5-deficient virions had a defect in core entry and an inability to induce syncytium formation. In addition, we confirmed that J5 is associated with the EFC by affinity purification. These data indicate that J5 is a functional component of the EFC and highlights the advantage of combining transcriptional repression and RNA silencing for stringent reduction of gene expression.
The members of the virus family Poxviridae are linear double-stranded DNA viruses that replicate exclusively in the cytoplasm of the host cell (23). Vaccinia virus (VACV), the prototype poxvirus, has a 190-kbp genome predicted to encode nearly 200 proteins of which approximately 80 have been identified as components of the mature virion (MV) (8, 29, 48). MVs enter cells by neutral- and low-pH fusion with plasma and endosomal membranes, respectively (3, 38) utilizing macropinocytosis or fluid phase uptake (16, 22). Four proteins have been reported to participate in attachment by binding to glycosaminoglycans and laminin (6, 7, 15, 20). Analysis of conditional lethal mutants established roles in entry and membrane fusion for 11 viral proteins with homologs in all poxviruses: A16 (28), A21 (37), A28 (34), F9 (5), G3 (17), G9 (27), H2 (32), I2 (26), L1 (4), L5 (36), and O3 (30). Each of these proteins, except I2, has been shown to be associated in a stable complex that can be extracted from membranes and purified. However, the complex could not be isolated if either A16, A21, A28, G3, G9, H2, L5, or O3 was repressed even though the remaining entry proteins were still incorporated into virions, suggesting that the complex is held together by multiple protein interactions. L1 and F9 are not required for isolation of the complex, which suggests that they are peripherally associated with the EFC. On this basis, A16, A21, A28, G3, G9, H2, L5, and O3 have been considered core components of the entry fusion complex (EFC) and L1 and F9 have been considered EFC-associated proteins although there is no difference in the phenotypes of mutants.
An additional protein, J5, is associated with the EFC as determined by mass spectroscopy (33). J5 is expressed postreplicatively (47, 49) and incorporated into MVs (29). Interestingly, the C-terminal cysteine-rich domains of J5, G9, and A16 are homologous, although there is extensive sequence divergence. Nevertheless, a distinct role for J5 in entry has not been demonstrated. Here we show that J5 participates in entry and membrane fusion.
BS-C-1 cells were grown in minimum essential medium with Earl's balanced salts (EMEM) (Quality Biological, Gaithersburg, MD) supplemented with 2 mM l-Gln and 10% fetal bovine serum (FBS) and hereafter is referred to as S-EMEM (10% FBS). The Western Reserve (WR) strain of VACV (ATCC VR-1354; GenBank accession number AY243312) and recombinant viruses were propagated, the titers of the viruses were determined, and the viruses were purified as described previously (11–13).
vJ5i was constructed by replacing the J5 promoter in vT7LacOI (1) with the phage T7 promoter and Escherichia coli lac operator and appending DNA encoding the octopeptide flag tag immediately after the start codon of the J5 open reading frame (ORF) essentially as described for other recombinant viruses (37). vV5L5iStrepJ5 was constructed by appending the sequence coding for the strepIII tag (31) to the 5′ end of the J5 ORF in vV5L5i (36) and inserting DNA encoding Discosoma sp. DsRed fluorescent protein under the control of the VACV p11 late promoter into the intergenic region between J4R and J5L with DsRed transcription leftward. The recombinant virus vA16iflagJ5 was derived from vA16i (28) by replacement of the original J5R gene by enhanced green fluorescent protein (GFP) regulated by the I1L intermediate promoter followed by a new J5R gene containing three tandem copies of the sequence encoding the flag epitope immediately after the N-terminal methionine as described for vA28iG93XFlag (42). Recombinant viruses were clonally purified by several rounds of fluorescent-plaque isolation.
BS-C-1 cells in a 6-well plate were infected with 5 PFU of unpurified virus per cell. After 1-h adsorption at 37°C, the cells were washed with EMEM and suspended in S-EMEM (2.5% FBS) with or without 100 μM isopropyl-β-d-thiogalactopyranoside (IPTG). The cells were harvested 22 h later and suspended in 1 ml S-EMEM (2.5% FBS). Following freeze-thawing (3 times) and 1 min of sonication (3 times), virus infectivity was determined by plaque assay. Plaque assays were carried out by infecting BS-C-1 cells in a 6-well plate with 10-fold serial dilutions of virus. Following incubation at 37°C for 1 h, nonadsorbed virus was removed by washing the cells, and 0.5% methylcellulose in S-EMEM (2.5% FBS) with 100 μM IPTG was added. After 48 h, infected cells were stained with crystal violet, and the plaques were counted.
For determination of specific infectivity, the number of sucrose gradient-purified virus particles was estimated by light scattering at A260 (1 unit = 1.2 × 1010 particles), and infectivity was determined by plaque assay on BS-C-1 cells (13).
To suppress J5 expression, four small interfering RNAs (siRNAs) (Thermo Fisher Scientific, Dharmacon RNAi Technologies, Lafayette, CO) were designed to specifically cleave J5 mRNA. These double-stranded RNAs were 21 nucleotides (nt) long and targeted the following regions of the 402-nucleotide J5 transcript: nt 32 to 50, nt 81 to 99, nt 133 to 151, and nt 357 to 375. The control siRNA was On-TARGETplus Non-Targeting pool siRNA (Thermo Fisher Scientific).
For analytical experiments, 0, 13.6, 27.2, or 54.4 pmol of siRNA in 0.25 ml of Opti-mem I (Invitrogen) was added to 0.25 ml of Opti-mem I containing 5 μl of RNAiMAX Lipofectamine (Invitrogen). After 10- to 20-min incubation at room temperature, the mixtures were added to wells containing 2 ml of S-EMEM (10% FBS) to give siRNA concentrations of 0, 5.4, 10.9, and 21.8 nM. Cells were infected with virus 18 to 20 h later.
For large-scale silencing to make J5-deficient virions, 3,360 pmol of a pool of four J5 siRNAs, siRNA-1 to siRNA-4 (siRNA1-4) (840 pmol of each J5 siRNA) in 15.5 ml Opti-mem I was combined with 0.4 ml of RNAiMAX Lipofectamine in 15 ml Opti-mem I. After 10 to 20 min, 6 ml was added to each T150 flask containing 107 BS-C-1 cells at 50% confluence in 30 ml of growth medium for a final siRNA concentration of 18.7 nM. After 18 h at 37°C, 5 PFU per cell of vJ5i was added.
Purified J5+ virions (10 PFU per cell), produced by infection with vJ5i in the presence of IPTG, and equivalent numbers of purified J5− and J5−/− virions, produced in the absence of IPTG with or without J5 siRNA, respectively, were spinoculated at 650 × g for 1 h at 4°C onto the BS-C-1 cells. The cells were washed with EMEM and then incubated for 2 h in S-EMEM (2.5% FBS) containing 300 μg/ml of cycloheximide. After washing with Dulbecco's phosphate-buffered saline with Ca2+ and Mg2+ (DPBS), the cells were fixed for 20 min with 4% paraformaldehyde in DPBS followed by washing with DPBS and permeabilized with 0.1% Triton X-100 for 15 min. The cells were blocked for 30 min with 10% heat-inactivated FBS in DPBS and incubated for 1 h with mouse monoclonal antibody (MAb) 7D11 to the VACV L1 membrane protein (45) and rabbit polyclonal antibody to the A4 core protein in blocking solution. Primary antibodies were removed by washing, and the cells were incubated for 1 h in a 1:200 dilution of fluorescent anti-rabbit Alexa Fluor 488 and anti-mouse Alexa Fluor 594 secondary antibodies (Molecular Probes, Eugene, OR) in blocking solution for 1 h. Following washing, the coverslips were inverted and mounted on a glass slide using Prolong Gold (Invitrogen).
BS-C-1 cells on coverslips (50% confluence) were incubated for 20 h with 28 nM control or J5 siRNA prior to infection with 5 PFU per cell of vJ5i and incubation in S-EMEM (2.5% FBS) with or without 50 μM IPTG for an additional 20 h. The cells were washed with fresh S-EMEM (2.5% FBS) and then incubated with prewarmed pH 5.25 buffer or DPBS for 2 min and then washed again to raise the pH. For preparation of the low-pH buffer, 20 ml of concentrated HCl was added to 180 ml of 0.3 M 2-(N-morpholino)ethanesulfonic acid (MES). MES-HCl was then added to DPBS with stirring until the pH was lowered to 5.25. The cells were incubated in S-EMEM (2.5% FBS) for 3 h at 37°C and then washed and fixed with 4% paraformaldehyde in DPBS for 20 min. The cells were blocked with 4% bovine serum albumin in DPBS for 30 min and stained for 30 min with 1:100 phalloidin conjugated to Alexa Fluor 594 in 4% bovine serum albumin in DPBS. The cells were then washed, mounted with Prolong Gold (Invitrogen) and 4′,6′-diamidino-2-phenylindole (DAPI), and examined by fluorescence microscopy.
Approximately 200 PFU per cell of purified J5+ or equivalent numbers of J5− or J5−/− virions were spinoculated onto BS-C-1 cells on coverslips at 650 × g for 1 h at 4°C. After washing with EMEM, cells were treated for 2 min with either pH 5.25 or pH 7.4 prewarmed buffer as described above, washed, and incubated in S-EMEM (2.5% FBS) containing 300 μg/ml cycloheximide for 3 h, washed, fixed with 4% paraformaldehyde, blocked with 4% bovine serum albumin, stained with phalloidin, and mounted as described above.
BS-C-1 cells were infected with vV5L5i or vV5L5istrepJ5 with or without 100 μM IPTG at 37°C for 22 h. After harvesting, cells were lysed and the proteins were bound to Strep-Tactin Superflow beads and eluted with biotin as described previously (43).
Whole-cell lysates or purified virions were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) on 4 to 12% Novex NuPAGE gels with 2-(N-morpholino)ethanesulfonic acid–SDS running buffer and transferred to nitrocellulose membranes using miniblot gel transfer stacks (Invitrogen). The membrane was blocked with 5% nonfat milk in phosphate-buffered saline containing 0.05% Tween 20 (PBS-T), incubated with primary antibody in blocking buffer overnight at 4°C, and thoroughly washed with PBS-T. For chemiluminescence detection, the blot was incubated for 1 h with the appropriate secondary antibody conjugated to horseradish peroxidase (Pierce, Rockford, IL) in blocking buffer. The blot was washed and developed using Dura or Femto chemiluminescent substrate (Pierce). Rabbit polyclonal antibodies against the following were used: A3 (R. Doms and B. Moss, unpublished data), A17 (44), A21 (37), A28 (24), H2 (25), L2 (21), and L5 (36).
The J5L (WR097) ORF predicts a 15-kDa protein with a C-terminal transmembrane domain spanning amino acids 110 to 132. A multiple-sequence alignment of J5 orthologs from each known genus of chordopoxvirus and entomopoxvirus is shown in Fig. 1A. In all orthopoxviruses, the J5 ORF is exactly 133 amino acids with an identity of 96 to 100%. The length of the protein is more variable, and the amino acid identity ranges from 62 to 80% in other chordopoxvirus genera and is about 30% in entomopoxviruses. Most striking is the conservation of eight cysteine residues and a predicted C-terminal α-helical transmembrane domain (Fig. 1A). The previously noted (33) similarity between J5 and the C-terminal portions of A16 and G9 is shown by the alignment in Fig. 1B. Again, the outstanding features are conservation of cysteines and predicted C-terminal α-helical transmembrane domains. A16, G9, and J5 likely arose by gene duplications in a common ancestor of all poxviruses.
The conservation of J5 among poxviruses and its homology to G9 and A16 suggested that J5 plays a role in viral replication as part of the entry fusion complex. To investigate such a role, homologous recombination was used to create vJ5i, a virus in which transcription of the J5L ORF is conditionally regulated by IPTG. In this construct, the T7 RNA polymerase transcribes J5L with an appended N-terminal flag epitope tag and the E. coli lac operator regulates transcription of both the T7 RNA polymerase gene and J5L to provide stringency (Fig. 2A). To aid in isolating vJ5i, the ORF for GFP preceded by the VACV p11 late promoter was inserted in the intergenic region between J5L and J6R. The recombinant virus was clonally purified by multiple rounds of fluorescent-plaque purification in the presence of IPTG to allow J5 expression.
To determine whether expression of J5 was required for virus replication, BS-C-1 cells were infected with vJ5i with IPTG and without IPTG, and the yield of infectious virus was determined subsequently by plaque assay. The amount of infectious virus was reduced approximately 80% in the absence of IPTG (Fig. 2B). Although the reduction was significant, this result suggested either incomplete repression or that J5 was not absolutely required for replication. Western blotting failed to detect J5 in cells infected with vJ5i in the absence of IPTG (Fig. 2C), suggesting good stringency. However, a previous failed attempt to delete the J5L ORF showed that this gene is essential for virus replication (49). We therefore considered that repression was incomplete and that the small amount of J5 needed for infectivity was below the threshold detection limit for the flag antibody.
Silencing with siRNA was tried as an alternative to repression of J5 expression. Short double-stranded RNAs matching four different regions of the J5 ORF were designed for selective recognition and degradation of J5 transcripts. Following transfection of these siRNAs or nonspecific control siRNAs, the cells were infected with virus encoding flag-tagged J5 (vA16iflagJ5) to determine the efficacy of silencing by Western blotting with flag antibody. (The inducible A16L gene in the recombinant virus was irrelevant for this experiment, as the infection was done in the presence of IPTG.) Individual J5 siRNAs or a combination of four J5 siRNAs effectively knocked down J5 expression below the threshold limit for detection by Western blotting (Fig. 3A and data not shown). Nevertheless, the individual and pooled siRNAs reduced the yield of virus only by 40 to 50% compared to cells transfected with control siRNA (Fig. 3B). The discrepancy between the apparently stringent reduction of J5 expression determined by Western blotting and the incomplete reduction in virus yield was reminiscent of the effect obtained by repression with vJ5i.
Since repression and silencing each produced a partial reduction in virus infectivity, we investigated the effects of the combination. We considered that small amounts of J5 mRNA produced by vJ5i in the absence of IPTG would be efficiently degraded by siRNA. The pooled J5 siRNAs (siRNA1-4) or control siRNAs were transfected into BS-C-1 cells, which were then infected with vJ5i in the presence or absence of IPTG. When the cells were transfected with 0 to 21.8 nM control siRNA and infected with vJ5i in the absence of IPTG, the yield of infectious virus was reduced approximately 3- to 4-fold due to repression compared to cells infected in the presence of IPTG (Fig. 4A). However, when cells were transfected with J5-targeted siRNAs and infected with vJ5i in the absence of IPTG, the yield of infectious virus was reduced more than 60-fold compared to cells transfected with control siRNA and infected with vJ5i in the presence of IPTG (Fig. 4A). A combined effect of repression and silencing was also seen by the decrease in vJ5i plaque formation in the absence of IPTG when J5-targeted siRNAs were transfected (Fig. 4B). The few plaques that formed under the latter conditions might have resulted from nonuniform transfection of the siRNA. Thus, transcriptional repression and RNA silencing had a synergistic effect, reducing the production of infectious virus by about 1.5 log units.
The formation of virions with very low infectivity is a characteristic feature of conditional lethal EFC mutants. We infected BS-C-1 cells with vJ5i in the presence or absence of IPTG with or without J5-targeted siRNAs. After 24 h, the cells were lysed and virions were purified from the postnuclear supernatant by sedimentation through a sucrose cushion followed by a sucrose density gradient. The yields of virus particles determined by optical density assuming that 1 unit equals 1.2 × 1010 particles (13) were similar in each case (Table 1). However, infectivity of the particles was reduced in the absence of IPTG and further reduced with the addition of siRNAs. The specific infectivities (number of PFU/number of virus particles) were 65, 194, and 3,526 for virions made in the presence of IPTG (vJ5+), absence of IPTG (vJ5−), and absence of IPTG plus siRNAs (vJ5−/−), respectively (Table 1). Nevertheless, the polypeptide compositions of the three MV preparations as observed by SDS-PAGE and silver staining were similar (Fig. 5A). Furthermore, immunoblot analysis of the virions (Fig. 5B) showed comparable amounts of representative membrane (A17, A21, A28, and L5) and core (A3 and I7) proteins. The cleaved forms of A17 and A3 were predominant, indicating no apparent defect in morphogenesis.
An entry block is the hallmark of EFC-deficient virions. To examine whether J5 is required for entry, we adapted the method of Vanderplasschen et al. (39) that discriminates virions on the surfaces of cells and cores in the cytoplasm by immunostaining. An important feature of the method is that following fixation, core proteins cannot be detected in whole virions by antibody even after permeabilization of the viral membrane with detergent. For this experiment, equivalent numbers of virions produced in the presence of IPTG or in the absence of IPTG with or without siRNAs were spinoculated at 4°C onto coverslips containing a monolayer of BS-C-1 cells. Following washing to remove unbound virions, the cells were incubated at 37°C for 2 h in medium containing cycloheximide, which allows penetration and accumulation of cores in the cytoplasm. The cells were then fixed, permeabilized, and stained with antibody to the L1 membrane protein and the A4 core protein. In each case, similar numbers of virions were bound to cells as evidenced by binding of L1 antibody (Fig. 6A, red dots). Numerous cytoplasmic cores (small green dots) that stained with the A4 antibody were present in cells infected with vJ5+ virions (Fig. 6A). In contrast, cytoplasmic cores were moderately reduced when cells were infected with vJ5− virions made in the absence of IPTG and greatly reduced when cells were infected with vJ5−/− virions made in the absence of IPTG and presence of siRNAs (Fig. 6A).
The reduction in core entry was also assessed by analyzing early gene expression. For this experiment, equivalent numbers of vJ5+, vJ5−, and vJ5−/− virions were spinoculated onto cells at 4°C. After removal of unbound virions, the cells were incubated at 37°C for 2 or 4 h. Synthesis of the L2 early protein was analyzed by Western blotting. Immunoblots showed a modest reduction of L2 in cells infected with vJ5− virions compared to those infected with vJ5+ virions at 2 and 4 h after infection. However, L2 was barely detected at 2 h after infection and strongly reduced at 4 h after infection in cells infected with vJ5−/− virions (Fig. 6B).
Virions on the surfaces of infected cells can mediate syncytium formation in response to a brief lowering of pH (10, 14). The virions can either be progeny from the infected cell (fusion from within) or virus inoculum deposited on the cell at high multiplicity (fusion from without). To test whether J5 is required for pH-induced fusion from within, BS-C-1 cells on coverslips were transfected overnight with either control or J5-specific siRNA and then infected with vJ5i in medium with or without IPTG for 20 h. The cells were then pulsed with either pH 5.25 or pH 7.4 buffer and allowed to recover in regular medium for 3 h to allow fusion. The cells were then fixed and stained with DAPI and analyzed by fluorescence microscopy. Syncytia were observed following the low-pH pulse in cells transfected with control siRNA and infected in the presence of IPTG but not after infection with vJ5i in the absence of IPTG whether control or J5-specific siRNA had been transfected (Fig. 7). The cell clumps that formed in the absence of IPTG could be clearly distinguished from syncytia with nuclei that were massed together. Syncytia were also not observed under any of the conditions when the cells were pulsed with pH 7.4 buffer (Fig. 7).
To test whether J5 is required for fusion from without, equivalent numbers of vJ5+, vJ5−, or vJ5−/− virions were spinoculated onto BS-C-1 cells. After washing and a brief treatment with pH 5.25 or pH 7.4 buffer, the cells were incubated for 3 h at 37°C in medium containing cycloheximide to prevent viral gene expression and cytopathic effects. The preparations were analyzed by confocal microscopy after fixing and staining with phalloidin and DAPI. Syncytia were observed in cell cultures infected with vJ5+ virions and pulsed with low-pH buffer but not in cell cultures infected with vJ5− or vJ5−/− virions in which the individual cell outlines were clearly visible (Fig. 8). Pulsing with pH 7.4 buffer did not induce syncytium formation in any of the samples (data not shown).
J5 was originally described as an EFC protein because it was detected by mass spectrometry following affinity purification of A28 and associated proteins (33). We now investigated the association by affinity chromatography with tagged J5 itself. In order to accomplish this, we constructed the recombinant virus vV5L5istrepJ5 by inserting DNA encoding the strepIII tag (31) at the N terminus of J5L in the genome of vV5L5i, an L5-inducible virus (36). BS-C-1 cells were infected with the parental vV5L5i or vV5L5istrepJ5 in the presence or absence of IPTG to regulate L5 expression. The cells were then harvested, lysed with Triton X-100, clarified, and incubated with streptactin beads. The input proteins and bound proteins eluted with biotin were analyzed by Western blotting with antibodies to the representative EFC proteins A21, A28, H2, and L5. All four proteins were affinity purified with strepJ5 from cells infected with virus able to express L5 (in the presence of IPTG), confirming that J5 is associated with the EFC (Fig. 9). Under the latter conditions, we also detected A16 and G9 as well as J5 by mass spectroscopy (data not shown). Copurification of EFC proteins with J5 did not occur when L5 was repressed and the complex was destabilized (Fig. 9).
J5 is the 11th protein shown to be a functional component of the EFC. Determination of the role of J5 was more difficult than for other VACV entry proteins. Indeed, this study was stalled for more than 5 years because of an apparent contradiction: a deletion mutant could not be isolated, suggesting that J5 is essential, yet replication of a J5 mutant was incompletely inhibited in the absence of inducer even though the repression appeared complete based on Western blotting. We suspected that some J5 was made but the amount was below the detection threshold of the antibody and that the stringency of repression was insufficient for complete inhibition of virus replication. Furthermore, we considered it possible that there could be excess EFCs on the surfaces of VACV MVs and that entry into cultured cells could still occur with a greatly reduced number. Unfortunately, we cannot test this hypothesis, as methods to determine the EFC density on virions have not yet been devised. In this regard, however, it is worth noting that a single envelope trimer is sufficient for HIV entry (46).
Several recent studies indicated that siRNAs can inhibit VACV replication, although they were not used to analyze the roles of the silenced genes (9, 19, 40, 41). In a screen of siRNAs to 12 monkeypox virus ORFs, knockdown of the J5 homolog reduced the virus yield by approximately 60% (2). Although we did not expect siRNA alone to reduce the amount of J5 more efficiently than the inducible mutant, it occurred to us that the combination might work synergistically. Thus, if J5 mRNA could be greatly reduced by repressing synthesis, then the siRNA might effectively degrade the small remainder. We found that the yield of infectious virus was reduced 4- to 5-fold by repression alone, about 2-fold by siRNA alone, but by more than 60-fold together. The siRNAs were most effective when transfected prior to infection (C.L.W., unpublished data) probably because the RNA-induced silencing complex had additional time to form and was distributed throughout the cytoplasm before establishment of specialized DNA factories where VACV transcription occurs (18). We are unsure why repression more efficiently prevents infectivity of other EFC mutants than the J5 mutant. However, the effectiveness of the inducible system appears to vary with gene location and the stringency needed to produce a functional defect, so that supplementation with siRNA may be a generally useful strategy.
The phenotype resulting from stringent reduction of J5 was similar to that occurring when expression of other EFC proteins is repressed, indicating that the J5 paralogs A16 and G9 cannot substitute for J5. When J5 was reduced, a normal yield of MVs formed, indicating the absence of effects on viral gene expression or morphogenesis. The protein composition of J5-deficient virions appeared similar to that of the wild type by staining SDS-polyacrylamide gels and Western blotting of representative core and membrane proteins, including components of the EFC. Nevertheless, the purified J5-deficient virions had very low infectivity. Further studies indicated that the block was at core entry rather than MV binding and that as a consequence early gene expression was severely inhibited.
Interestingly, repression of J5 was sufficient to prevent syncytium formation from within or without; the added stringency of siRNA was not required. This result suggests that more EFCs are needed to fuse cells together than to allow entry of virus particles.
We confirmed an earlier report that J5 is physically associated with the EFC (33). The prior study used an affinity tag attached to the A28 EFC protein to capture the complex, and here we used a tag attached to the J5 protein itself to show association with other EFC proteins. An interesting feature of J5 is its homology with two other proteins of the EFC, A16 and G9. The three proteins are conserved in all poxviruses, suggesting that they arose from two gene duplications prior to the diversification of present poxvirus genera and subsequently evolved distinct functions in virion entry. The conservation of cysteine residues in the three proteins is striking. Evidence for intramolecular disulfide bonds was obtained for A16 (28), and it is likely that these bonds are also formed in the related G9 and J5 proteins. The intramolecular disulfide bonds of A16 and other EFC proteins are formed by the VACV-encoded cytoplasmic redox system, which is also conserved in all poxviruses (35), suggesting that the redox and entry systems arose together during evolution.
With 11 functional components of the EFC known, are there more to be identified? The small I2 protein is a good candidate, since the phenotype of a conditional lethal mutant is similar to that of EFC proteins with the exception that repression decreases the amounts of EFC components in MVs (26). Because of its small size and hydrophobicity, the I2 protein, like the O3 EFC protein (30), may have been missed by mass spectroscopy of the purified complex (33). The overall architecture of the EFC has not been determined, although a few protein-protein interactions that resist destabilization of the complex have been identified (25, 42, 43).
We thank Catherine Cotter for help with cell culture and P. S. Satheshkumar, Jason Laliberte, and Tatiana Senkevich for helpful discussions.
The work was supported by funds from the Division of Intramural Research, NIAID, NIH.
Published ahead of print 19 October 2011