A Novel Cellular Protein, VPEF, Facilitates Vaccinia Virus Penetration into HeLa Cells through Fluid Phase Endocytosis

HeLa cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum. The wild-type WR strain of vaccinia virus was used. IMV virions were purified on a 36% sucrose cushion, followed by 25 to 40% sucrose gradient centrifugation as described previously (32), and stored at −70°C. Purified rabbit anti-VPEF immunoglobulin G (IgG), purchased from QCB (Hopkinton, MA), was raised against a peptide sequence of VPEF (KSTGVFQDEELLFSHKLQKDNDPD) which is conserved in human, mouse, and CHO cells. Rabbit anti-A4L antiserum was raised against recombinant A4L protein purified from bacteria. Rabbit anti-vaccinia virus antiserum was raised against IMVs. The mouse monoclonal antibody (MAb) clone 2D5, recognizing the L1R protein, was obtained from Y. Ichihashi (30). A mouse anti-human transferrin receptor (TfR) MAb was purchased from Zymed Laboratories, Inc. Rabbit anti-TfR antibodies (Abs) were purchased from Chemicon and Millipore. Mouse anti-CD107a (Lamp1) Abs and rabbit anticaveolin Abs were purchased from BD Biosciences Inc. Mouse anticaveolin MAb was purchased from BD Transduction Laboratories Inc. Tetramethylrhodamine-conjugated goat anti-rabbit IgG Abs were purchased from Invitrogen. Cy5-conjugated goat anti-mouse IgG Abs were purchased from Jackson ImmunoResearch Laboratories, Inc. The cholera toxin B subunit (CTB) conjugated to Alexa Fluor 594, Texas Red- or fluorescein isothiocyanate (FITC)-conjugated transferrin, and Texas Red- or FITC-conjugated dextran (molecular weight, 10,000) were purchased from Molecular Probes and Invitrogen. The pEGFP-C1 vector was purchased from Clontech Inc. The Rab11 plasmid was purchased from Missouri S&T cDNA Resource Center (Rolla, MO). Eps15D95/295 was provided by Alice Dautry-Varsat (L'Institut Pasteur, Paris, France). Plasmids encoding wild-type dynamin 1 (WT-Dyn1), dynamin 1 with a K44A mutation (DN-dyn1 [Dyn1K44A]), WT-Dyn2, and DN-dyn2 (Dyn2K44A) were from Sandra Schmid (Scripps Research Institute, CA). pEGFP-C3-Rab5 was provided by Marino Zerial (Max Planck Institute, Germany). The mCherry plasmid was obtained from Roger Y. Tsien (Howard Hughes Medical Institute, University of California at San Diego). Dynasore was a gift from Tomas Kirchhausen (Harvard Medical School and the CBR Institute for Biomedical Research, Inc., MA).

Fusion of mCherry to the N terminus of A4L was achieved by PCR splicing by overlap extension (25) using vaccinia virus strain WR genomic DNA as the template. Oligonucleotides 1, 5′-CTCCGTTGAATTCGATGACTATAGGACAAGAACCCTCCTC-3′, and 2, 5′-ATCCTCCTCGCCCTTGCTCACCATTTAAGGCTTTAAAATTGAATTGCG-3′, were used to generate a 463-bp PCR product located upstream of the A4L gene open reading frame (ORF). Oligonucleotides 3, 5′-GGCATGGACGAGCTGTACAAGGACTTCTTTAACAAGTTCTCACA GGGG-3′, and 4, 5′-CGTACTCCAAGCTTGTGTAGATGCTACTTCGTCGATGG-3′, were used to generate a 1,215-bp PCR fragment containing the A4L gene ORF and the 346 bp downstream of the A4L gene ORF. The mCherry gene ORF was amplified using pmCherry (51) as the template and oligonucleotides 5, 5′-ATGGTGAGCAAGGGCGAGGAGGAT-3′, and 6, 5′-CTTGTACAGCTCGTCCATGCC-3′, generating a 711-bp fragment. Oligonucleotides 1 and 4 introduced EcoRI and HindIII restriction sites, respectively (underlined), while oligonucleotides 2 and 3 contained mCherry gene sequences enabling the individual fragments to be assembled into a single 2,389-bp gene. This PCR product was cloned into TOPO pCDNA3.1 to form pA4L-mCherry-N.

293T cells were infected with vaccinia virus WR at 5 PFU per cell and transfected using Lipofectamine (Invitrogen) with the DNA fragment produced by EcoRI and HindIII digestion of pA5L-EGFP-N. The cell lysates were collected at 24 h postinfection. Recombinant viruses expressing mCherry were identified by fluorescence and subsequently underwent three rounds of plaque purification on BSC-40 cells.

HeLa cells grown to 70% confluence on coverslips were infected at 100 PFU/cell at 37°C for 30 min with purified A4L-mCherry-expressing IMV. Where indicated (see the Fig. 1 legend), 0.5 μM latrunculin A (Invitrogen) and 2.5 μM taxol (Sigma) were added for 30 min at 37°C before and during virus infection. The cells were maintained in dishes in DMEM without phenol red medium (Gibco-Invitrogen); the dishes were placed on a microscope stage for preheating to 37°C in 5% CO₂ before virus addition. The cells were observed on a Zeiss Axiovert 200 M microscope with a ×63, 1.25-numerical-aperture Plan-NEOFluar oil lens using Zeiss AxioVision Rel. 4.6 software. Images were acquired at 30-s intervals for 60 frames at times between 0 and 30 min postinfection.

Cellular mRNA was isolated from total RNA purified from CHO cells using oligo(dT) beads as described previously (47a) and was used to generate a cDNA phagemid expression library in a lambda ZAP vector as described by the manufacturer (Stratagene cDNA synthesis kit) (52). Screening with MAb B2 (4) for a cDNA gene product was performed as described previously (52). Out of 2 million clones, identical cDNA clones were repeatedly isolated, and their sequences were determined. The cDNA was 4.2 kb and contained a full-length VPEF gene ORF that was fused in frame with the coding sequence for the upstream N-terminal portion of lacZ in the ZAP vector, confirming the validity of Ab screening. The VPEF gene ORF encoded a polypeptide of 1,318 amino acids. Its homologues were found in other species, such as humans (KIAA0592) and mice (NM_026585).

To generate a glutathione S-transferase (GST)-VPEF fusion construct, the VPEF cDNA was digested with SalI to release the insert VPEF cDNA, which was subcloned into pGEN-KG to generate the GST-VPEF plasmid (12). To generate the VPEF-green fluorescent protein (GFP) construct, the full-length cDNA was PCR amplified and cloned in frame into TOPO vector pcDNA3.1/CT-GFP (Invitrogen) so that GFP was fused at the C terminus. Sequencing was performed to ensure that no mutation was introduced during PCR. Wild-type caveolin-1 DNA was generated by reverse transcription-PCR and cloned into pcDNA3.1/CT-GFP to give the caveolin-GFP plasmid. Construct Cav1Δ1-81 DNA was generated as described previously for construct DN-Cav3 (47).

For protein expression, Escherichia coli strain BL21(DE3) was transformed with GST-VPEF or control GST plasmids and then the cultures were induced with 0.2 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 20 h at 17°C and harvested. GST and GST-VPEF fusion proteins were purified using glutathione beads as described previously (12). The purified proteins were dialyzed against phosphate-buffered saline (PBS) at 4°C and concentrated on Amicon ultracentrifugal filters (Millipore) before use.

Cell morphology during vaccinia IMV entry into HeLa cells was monitored as described previously, with some modifications (31). A suspension of purified wild type IMVs (200 PFU/cell) or G3 mutant IMVs in 0.3 ml of DMEM was applied to a HeLa monolayer in a 12-well tissue culture plate that had been equilibrated for 30 min at 4°C. The plate was placed in a (no. 5782) bucket in a GP8R centrifuge (International Equipment Company), and virus was pelleted onto the cells at 350 × g for 1 h at 4°C. Subsequently, the cells were cultured in normal medium at 37°C for 0, 5, 15, or 30 min before fixation. The fixed cells were treated with 1% OsO₄, dehydrated, and embedded as described previously (56). After embedding, the cells were stained with uranyl acetate and lead citrate and analyzed under a Zeiss 902 transmission electron microscope (46).

The virion binding assay measuring the amount of cell surface-bound virions was performed as described previously (13, 50, 63). The virion penetration assay measuring the intracellular uncoated cores, which can be stained only with anti-A4 core Abs after virus entry, was also performed as described previously (13, 49, 62). In brief, HeLa cells were infected for 1 h at 4°C with wild-type virus at a multiplicity of infection (MOI) of 40 PFU/cell, washed three times with PBS, and either fixed immediately (for the virion binding assay) or incubated for 2 h at 37°C in the presence of cycloheximide (30 μg/ml) and then fixed (for the virus penetration assay). The cells were fixed, permeabilized in PBS-0.2% saponin, and stained with rabbit anti-A4L Abs or mouse anti-L1R MAb, followed by FITC-conjugated goat anti-rabbit IgG Abs or Cy5-conjugated goat anti-mouse IgG Abs, respectively. DNA was visualized by staining with 0.5 μg/ml of DAPI (4′,6-diamidino-2-phenylindole dihydrochloride; Molecular Probes). Cell images were collected with an LSM510 Meta confocal laser scanning microscope (Carl Zeiss, Germany) using a 63× objective lens and confocal microscopy software (release 2.8; Carl Zeiss). The fluorescent particles were counted from multiple photos, and the averaged numbers of surface-bound virions and uncoated cores per cell were determined.

Fifty, 100, or 200 μg/ml of GST or GST-VPEF was preincubated with virus (MOI = 40) at 4°C for 30 min. The mixture was then added to HeLa cells to initiate virus infection at 4°C for 60 min (for virus binding assays) or at 37°C for 2 h (for the penetration assay). These cells were washed, fixed and stained for viral L1 and core protein as described above for virion binding and penetration assays (11, 49, 61).

One, 5, 10, or 50 μg/ml of anti-M2 Abs or anti-TfR Abs was preincubated with HeLa cells at 37°C for 1 h, and then the cells were washed with PBS, and used for virus binding and penetration assays as described above for virion binding and penetration assays.

Patching of cell surface proteins with Abs was performed as described previously (55). HeLa cells were seeded on coverslips in 12-well plates and then were infected on the next day with vaccinia virus at an MOI of 50 PFU/cell in DMEM for 1 h, washed, and transferred to 12°C for patching of lipid rafts. The cells were incubated at 12°C for 1 h with a primary Ab (anti-vaccinia virus [diluted 1:500], anti-L1R MAb [diluted 1:400], anti-VPEF [3.3 μg/ml], or anti-TfR [50 μg/ml]) or Alexa Fluor 594-CTB (10 μg/ml) (Molecular Probes, Invitrogen), and then tetramethylrhodamine-conjugated goat anti-rabbit IgG (diluted 1:500) or Cy5-conjugated goat anti-mouse IgG (diluted 1:500) (1.5 mg/ml; Jackson ImmunoResearch Laboratories) was added for 30 min. The cells were then fixed for 5 min with 3.7% paraformaldehyde in PBS on ice and then methanol for 10 min at −20°C, mounted in Vectashield medium (Vector Laboratories, Burlingame, CA), and visualized by confocal laser scanning microscopy with an LSM 5 Pascal instrument (Carl Zeiss) using a 63× objective lenses and accompanying software.

HeLa cells were seeded on coverslips 1 day before the experiment to allow growth to 80% confluence. For transferrin uptake analyses (modified from that in reference 6), cells were starved in serum-free DMEM for 2 h at 37°C and then incubated in complete medium containing 25 μg/ml of Cy5- or Texas Red-conjugated transferrin (Molecular Probes, Invitrogen) at 37°C for 30 min before fixation for confocal microscopy analyses. For CTB and dextran uptake (procedure modified from those in references 5 and 53, respectively), cells were incubated at 37°C in complete medium with 8 μg/ml of Alexa Fluor 594-conjugated CTB or 10 mg/ml of FITC-conjugated dextran (molecular weight, 10,000; Molecular Probes, Invitrogen) for 40 min before fixation for confocal microscopy analyses. Alternatively, these ligand-treated cells were cooled on ice, washed twice with cold PBS, fixed in 4% paraformaldehyde, and permeabilized in PBS-0.2% saponin. Cells were stained with anti-M2 Abs (3.3 μg/ml) at room temperature for 1 h, with tetramethylrhodamine-conjugated goat anti-rabbit IgG Abs (diluted 1:1,000) at room temperature for 30 min, and with 0.5 mg/ml of DAPI.

HeLa cells were preincubated with 0.8% dimethyl sulfoxide (DMSO) or dynasore (40 μM or 80 μM) in serum-free DMEM at 37°C for 30 min and then directly cooled to 4°C and infected with vaccinia IMV (MOI of 40) at 4°C for 1 h. Cells were washed with cold PBS to remove free virions and further incubated at 37°C for 2 h before fixation for virus penetration assays as described above. The drug was present in the medium throughout the experiments. Alternatively, HeLa cells were treated with dynasore as described above and then pulsed with transferrin for a further 30 min before fixation for confocal microscopy and quantification using MetaMorph Offline, version 6.1 (Molecular Devices, CA).

The control cyclophilin B (CypB) small interfering RNA (siRNA) duplex and the two VPEF siRNA duplexes, 3-378 (AAGUAGAUCUCAUUCCUAAUU) and 1-525 (CAAAGGAUCUAUACAUUGAUU), which, respectively, target the 378 to 398 and 525 to 545 regions of human KIAA0592 RNA, were purchased from Dharmacon Inc. HeLa cells were transfected with siRNA duplexes (50 pM) using Lipofectamine 2000 (Invitrogen), the process was repeated, and then the cells were used for virus binding and penetration assays as described above.

The VPEF gene ORF was submitted to the NCBI under accession number AAK 53434.

We examined the entry pathway of vaccinia IMV into HeLa cells. Electron microscopy (EM) showed that IMVs bound to long and short filopodia (see Fig. S1A in the supplemental material), as previously reported by Locker et al. (37). To understand the relevance of these filopodia to virus entry, we turned to live-cell imaging, which allows a more dynamic observation of the interaction between virus particles and cells. Our attempt to construct a fluorescent vaccinia IMV expressing an envelope protein fused to the mCherry protein (51) was unsuccessful due to loss of infectivity. However, a fluorescent virus expressing the core protein A4L fused to mCherry, named mCherry-VV, was viable and subsequently purified (see Fig. S1B in the supplemental material). Within minutes after infection, some of the inoculated mCherry-VV bound to long filopodia of HeLa cells and started moving along the surface of the filopodia toward the cell body before disappearing at the base of filopodia or out of the focal plane (Fig. 1A; see Movies S1 and S2 in the supplemental material). Some of the viruses bound to cell edges or active ruffling areas and were transported to the cell body during ruffling retraction (see Movie S3 in the supplemental material), whereas other viruses were “grabbed” by waving actin protrusions and brought back to the cell body (see Movie S4 in the supplemental material). When HeLa cells were treated with latrunculin A to interrupt actin polymerization, as described previously (39), IMV attachment to filopodia was not affected; however, many of the bound particles became still (Fig. 1B; see Movie S5 in the supplemental material) or showed only slight local movement, demonstrating the importance of actin dynamics in the regulation of vaccinia IMV movement prior to cell entry. As expected, virus penetration into HeLa cells, as determined by measuring the amount of uncoated core A4L protein using previously described virus penetration assays (62), was significantly inhibited by latrunculin A (Fig. 1C). Taxol, previously reported not to affect vaccinia IMV entry (37), was included here to stabilize microtubules so that cells remained attached to plates, and this drug alone did not affect IMV movement (see Movie S6 in the supplemental material). Together, the results suggest that these dynamic actin-containing structures could serve as effective ways to recruit viruses for cell entry.

Due to the limitations of live-cell imaging, we were unable to follow individual viral particles after they reached the cell body. EM analyses showed that viruses which had reached the cell body were more associated with filopodium-rich surfaces than with smooth surfaces and were surrounded by filopodia (see Fig. S2A in the supplemental material). Interestingly, unlike Locker et al. (37), who reported no evidence of IMVs in endocytic vesicles, we observed single or multiple IMV particles that were enveloped by the intracellular vesicular membrane during viral infection (Fig. 1D), and the number of these vesicles increased steadily over the first 30 min of virus infection (Fig. 1E), suggesting endocytosis of IMVs by HeLa cells. Indeed, we found that viral core exposure was significantly inhibited in HeLa cells treated with bafilomycin (BFLA) (Fig. 1F), which blocks the endocytic route previously reported for virus entry into BSC-1 cells (58). We also used a mutant vaccinia virus in which expression of a component of the viral fusion complex, G3, was under IPTG regulation; mutant G3⁻ vaccinia viruses prepared from cells without IPTG induction produce IMVs that bind to cells, but the fusion step is blocked (31). In EM analyses, G3⁻ mutant virions accumulated within the cytoplasmic vesicles, suggesting that endocytic vesicles containing IMVs were already present in cells prior to the membrane fusion step (see Fig. S2B in the supplemental material). Taken together, our results show that vaccinia IMV was able to enter HeLa cells through an endocytic pathway, although we did not exclude plasma membrane fusion, as previously reported (37).

To define which endocytic machinery was used for vaccinia IMV entry, we obtained several dominant-negative (DN) mutant gene constructs known to specifically block different cellular endocytic pathways and tested their ability to block vaccinia IMV entry into HeLa cells. A dynamin 1 K44A DN mutant (DN-Dyn1) unable to bind GTP has been shown to inhibit the function of both endogenous dynamin 1 and dynamin 2 isoforms and to block major endocytic pathways, including clathrin- or caveola-mediated endocytosis, and even fluid phase endocytosis (7, 17, 18, 21, 40, 42, 60). In addition, a K44A mutant dynamin 2 (DN-Dyn2), the isoform that is ubiquitously expressed in nonneuronal cells, was included (15, 54). The more specific DN mutant Eps15 construct (Eps15Δ95/295 [DN-Eps15]), which blocks only clathrin-mediated endocytosis (6), and a DN caveolin construct (Cav1Δ1-81 [DN-Cav1]), which blocks caveola-mediated endocytosis (47), were also included. HeLa cells were transfected with each of these constructs and subsequently infected with vaccinia IMV as described in Materials and Methods; then the cells were fixed, permeabilized, and stained for the presence of intracellular core A4 protein after virus penetration, as described previously (61, 62). Ectopic expression of GFP-fused DN-Dyn1, but not of GFP-fused WT-Dyn1 or DN-Dyn2 or of GFP alone, blocked vaccinia IMV entry by ∼70% (Fig. 2A; data are quantified in Fig. 2G). In control experiments, both GFP-fused DN-Dyn1 and DN-Dyn2 effectively blocked uptake of transferrin into HeLa cells (Fig. 2B), showing an isoform-related variation in blockade of IMV entry but not of the coat-mediated endocytosis pathway. To directly inhibit the GTPase activity of endogenous dynamin in HeLa cells, we then treated cells with 40 and 80 μM concentrations of a cell-permeating inhibitor of dynamin, dynasore, that has been shown to specifically interfere with endocytic functions that depend on dynamin without affecting dynamin-independent functions (38). Dynasore blocked transferrin uptake as expected and, more importantly, blocked vaccinia IMV penetration (Fig. 2C). We also knocked down endogenous dynamin 2 expression in HeLa cells using a siRNA approach, and virus penetration was blocked (Fig. 2D). We thus concluded that vaccinia IMV entry into HeLa cells is dynamin dependent. Furthermore, expression of the GFP-fused Eps15Δ95/295 or Cav1Δ1-81 construct had little effect on the amount of uncoated intracellular A4 core protein (Fig. 2E; data are quantified in Fig. 2G), showing that vaccinia IMV does not utilize a clathrin- or caveolin-dependent endocytic pathway. In control experiments, DN-Eps15 and DN-Cav1 readily blocked uptake of transferrin and CTB, respectively, into HeLa cells (Fig. 2F). These results show that vaccinia IMV entry into HeLa cells does not rely on the formation of clathrin-coated pits or caveolae. Instead, IMV has evolved a different strategy for internalization into HeLa cells through an actin- and dynamin-dependent process. Although we could call it “dynamin-dependent macropinocytosis,” we feel that “fluid phase endocytosis,” as described by Cao et al. (7), fits better to our observation, so we decided to use the latter term. Although fluid phase endocytosis is consistent with active cellular protrusions recruiting viruses observed in our imaging analyses, more data are required to sustain this conclusion.

The molecular demonstration of virus entry pathways often requires the identification of cellular proteins that are important in virus entry. We previously generated MAb B2, which recognizes an unknown cell surface molecule(s) and can block vaccinia IMV infection (4). MAb B2 was used to screen a phage cDNA expression library, where it identified a single CHO cell cDNA clone (accession number AAK 53434) containing a full-length ORF encoding a 1,318-amino-acid protein, which we tentatively named VPEF. VPEF has homologues in other species, including humans (KIAA0592) and mice (NM_026585), but none contain any known structural domains or have a known function. Because MAb B2 is a low-affinity IgM, which gives a high background in immunofluorescence analyses, we generated a rabbit antiserum (anti-M2) using peptide KSTGVFQDEELLFSHKLQKDNDPD (M2) from VPEF. The anti-M2 Abs recognized an endogenous cellular protein with an apparent molecular mass of about 200 kDa on sodium dodecyl sulfate gels and not vaccinia IMV proteins (Fig. 3A). The 200-kDa protein size on gel was consistent with that of the in vitro-translated product of VPEF cDNA (data not shown), although larger than the predicted molecular mass of 145 kDa. The anti-M2 Abs also recognized the ectopically expressed VPEF-GFP fusion protein, but not GFP, transfected in HeLa cells (Fig. 3B). In immunofluorescence analyses, the anti-M2 Abs also reacted specifically with exogenous and endogenous VPEF protein in cytoplasm and not with vaccinia IMV particles (see Fig. S3 in the supplemental material).

Although abundant fluorescence of VPEF-GFP fusion protein appeared as intracellular speckles in HeLa cells, we could enhance the surface detection of VPEF-GFP using the anti-M2 Abs to cross-link surface VPEF-GFP as patches on the plasma membrane (Fig. 3C). Immunofluorescence microscopy showed that these VPEF-GFP-containing patches were colocalized with the plasma membrane lipid raft marker CTB, and not with TfR, suggesting that VPEF-GFP associates with lipid raft microdomains on the plasma membrane (Fig. 3C). Consistent with the data for ectopically expressed VPEF-GFP, endogenous VPEF was also detected on cell surface microdomains (Fig. 3D). It is worth noting that endogenous VPEF was present in, but not limited to, microdomains, as VPEF also exhibited intracellular-vesicle-like staining in permeabilized cells (see Fig. S3 in the supplemental material).

We had previously demonstrated vaccinia virus clustering at plasma membrane lipid rafts during virus penetration (13). Since VPEF-GFP was also present on lipid rafts, we investigated whether VPEF-GFP protein colocalized with vaccinia IMV particles during cell entry. Immunofluorescence analyses showed that VPEF-GFP colocalized mostly with IMVs that stained positive for core protein, suggesting that these IMVs were in the process of penetration (Fig. 4A). To eliminate the possibility that the observed viral core staining was an artifact of virion rupture during cell fixation, we performed kinetic analyses of vaccinia IMV infection in VPEF-GFP-transfected HeLa cells. Cells were infected with vaccinia IMV for 60 min at 4°C and, after a wash to remove unbound virus, were incubated at 37°C for a range of time periods from 0 to 2 h and then harvested for copatching analyses with anti-vaccinia virus and anticore Abs (Fig. 4B; data are quantified in Fig. 4C) as described previously (13). Cycloheximide was added to the medium to block viral early gene expression so that the virus life cycle was arrested immediately after entry (13, 61). At 0 min, as expected, >95% of bound IMV particles were intact and stained negative for core proteins. Initially, only little colocalization between IMV and VPEF was detected (Fig. 4C, left). As virus penetration progressed at 37°C from 30 min to 2 h, less IMV remained on the cell surface and more and more viral signals indicative of penetration intermediates were detected, i.e., cells stained positive with both anti-L1 and anticore Abs (Fig. 4C, middle). Interestingly, most of these penetration intermediates were also associated with VPEF. After penetration, viral cores were no longer associated with the viral envelope (Fig. 4C, right). The majority of these uncoated cores were not colocalized with VPEF, suggesting a transient association of VPEF with IMV during virus entry and a possible role of VPEF in IMV penetration.

To test whether VPEF is important for vaccinia virus entry into cells, recombinant GST-VPEF fusion protein was expressed in E. coli and tested for its ability to block vaccinia virus entry into cells. Vaccinia IMV binding to cells was not affected by GST-VPEF or GST, whereas heparin readily blocked virus attachment as expected (Fig. 5A; see Fig. S4A in the supplemental material). In contrast, GST-VPEF, but not control GST, blocked vaccinia virus core uncoating in a concentration-dependent manner (Fig. 5A; see Fig. S4B in the supplemental material), suggesting that VPEF plays a role in the postattachment penetration step. Although the GST-VPEF blocking effect was only partial at the highest dosage used (200 μg/ml), its blocking activity was comparable to those of several other recombinant viral envelope proteins previously used in virus entry blocking assays (26, 28). The anti-M2 Abs were used in further experiments similar to those above (Fig. 5B), which showed that, while control anti-TfR Abs did not block vaccinia IMV entry, anti-M2 Abs effectively reduced IMV penetration without affecting virus binding to cells, consistent with the results of the soluble protein blocking assays.

To determine if endogenous VPEF is required for vaccinia virus infection of permissive HeLa cells, siRNAs targeting human VPEF were transfected into HeLa cells to silence endogenous VPEF expression (VPEF-KD cells). In immunoblot analyses, a pool of four siRNAs (mix) and the single siRNA 3-378 both effectively reduced VPEF levels, whereas another siRNA, 1-525, had only a slight effect (Fig. 5C). Cellular CypB was independently knocked down as a control (Cyp-KD cells). Vaccinia IMV attachment to either VPEF-KD or CypB-KD cells was not affected (Fig. 5D; see Fig. S4C in the supplemental material), whereas vaccinia virus penetration was significantly reduced in VPEF-KD 3-378-treated cells, mildly reduced in VPEF-KD 1-525-treated cells, and not affected in CypB-KD cells (Fig. 5D; see Fig. S4D in the supplemental material). Quantitatively, more-effective siRNA silencing of VPEF expression resulted in less vaccinia virus penetration.

Our above results suggested that IMV penetration into HeLa cells may operate through a fluid phase endocytic pathway and that VPEF is important for cellular penetration by IMV. Furthermore, the intracellular distribution of endogenous VPEF was reminiscent of that of intracellular vesicles, suggesting that VPEF may be part of the system for vesicle transport between the cell surface and intracellular organelles. Could it be that the vaccinia IMV somehow recruits and/or diverts a cellular vesicle protein for its own benefit? We hypothesized that the cellular function of VPEF is involved in the uptake or transport of certain physiological ligands. To test this, we incubated HeLa cells with either fluorescence-labeled transferrin or dextran. Transferrin undergoes clathrin-mediated endocytosis, whereas dextran is taken up by fluid phase endocytosis (7, 64). Each of these fluorescent ligands was added to HeLa cells, which were then fixed, permeabilized, and stained with anti-M2 Abs. Transferrin showed little colocalization with VPEF, whereas dextran exhibited clear colocalization with intracellular VPEF (Fig. 6A) 40 min after addition. Four hours later, when the dextran had been chased into late endosomes/lysosomes, no colocalization between dextran and VPEF was observed (Fig. 6B), suggesting that dextran was transported in VPEF-containing vesicles during the early phase after being taken up into HeLa cells. Most importantly, when VPEF expression was knocked down using siRNA, dextran transport in cells was affected, showing a more restricted pattern close to the perinuclear areas (Fig. 6C), suggesting again that VPEF plays a role in the regulation of the fluid phase endocytic pathway in HeLa cells.

To further characterize the membrane organization of the VPEF-containing vesicles, we tested several intracellular vesicle markers: Rab5 for early endosomes (48), caveolin-1 for caveolae/caveosomes (43), and Rab11a for recycling endosomes (59). The VPEF-containing vesicles did not colocalize with Rab5 (Fig. 7A) or caveolin-1 (Fig. 7B) but showed a partial colocalization with the recycling endosome marker Rab11a (Fig. 7C), supporting our hypothesis that VPEF-containing vesicles are distinct from early endosomes and caveolae/caveosomes and participate in vesicle recycling between the cell surface and intracellular organelles. Finally, though both cellular VPEF and dynamin are important for vaccinia virus penetration, immunofluorescence studies revealed no colocalization of VPEF protein with either WT-Dyn1 or DN-Dyn1 and DN-Dyn2 in HeLa cells (Fig. 8), implying that VPEF and dynamin mediate IMV penetration through different mechanisms.