Human embryonic kidney cells (HEK-293T), baby hamster kidney cells (BHK-21), African green monkey kidney epithelial cells (Vero), human cervical carcinoma cells (HeLa), and lung epithelial (A549) cells were grown in complete DMEM (HyClone) with 10% fetal bovine serum (Thermo) at 37°C in a 5% CO₂ incubator. A vaccinia virus Tian Tan mutant strain carrying green fluorescent protein gene (VTT-EGFP) was constructed by homologous recombination as described previously (35). VTT strain and VTT-EGFP were propagated in BHK-21 cells, and handed according to the standard biosecurity procedures in a BSL-2 (Biosafety level 2) laboratory and purified on a sucrose gradient as described previously (33). The infectious titer was determined by plaque assay on Vero cells. The concentrations of purified viruses were determined by measuring the optical density (OD) at 260 nm (1 OD = 1.2 × 10¹⁰ virions) (36).

Human IFITM3 gene was amplified from the cDNA synthesized from THP-1 cells (ATCC TIB-202). The plasmid pLV-IFITM3 containing human IFITM3 gene with a N-terminal FLAG tag was constructed based on lentivirus-based plasmids, pLV-Puro (Inovogen, Beijing, China). Recombinant Human IFN-α2b was purchased from PeproTech Asia (Rehovot, Israel). Anti-FLAG M2 antibody was purchased from Sigma-Aldrich; anti-IFITM3, anti-GAPDH (glyceraldehyde 3-phosphate dehydrogenase), and anti-β-actin antibodies from Proteintech (Chicago, IL, USA); Anti-VACV D8 (WR113) antibody was purchased from Immune Tech (New York, NY, USA). Anti-GFP and Alexa Fluor 555-conjugated secondary antibodies and 4,6-diamidino-2-phenylindole (DAPI) were purchased from Beyotime (Haimen, China). Long-chain dialkylcarbocyanines (DiD) were procured from Life Technologies (Grand Island, NY, USA).

HeLa, 293T, or A549 cells were infected with VTT at 0.1 PFU/cell, and after indicated time intervals of infection, the total cell proteins were extracted with RIPA buffer with protease inhibitor cocktail (Roche, Basel, Switzerland). The extracted proteins were electrophoresed by SDS-PAGE, and then electrophoretically transferred to Nitrocellulose blotting membrane (GE Healthcare, Germany). After incubating in blocking buffer (5% bovine serum albumin in Tris-buffered saline) at room temperature for 1 h, the samples of membrane were probed with the indicated specific primary antibodies and corresponding HRP-conjugated IgG secondary antibodies. The bands were then visualized using an enhanced chemiluminescence detection kit (Pierce Biotechnology, Rockford, IL, USA) and X-ray film and quantified using Image J software normalized to GAPDH.

HeLa cells grown to 80% confluence in 12-well culture plates were treated with human IFN-α2b (10,000 or 20,000 U/mL) for 24 h and then infected with VTT-EGFP at PFU/cell of 0.1. After 1 h, the medium was removed, and cells were washed and cultured in fresh medium at 37°C. At specified time points, EGFP-positive cells were visualized by fluorescence microscopy and quantified by flow cytometry.

The siRNA oligonucleotides targeting IFITM3 and universal negative control siRNA were designed and synthesized from Ribobio (Guangzhou, China), their sequences are: siIFITM3-1: 5′-CCCACGUACUCCAACUUCC [dT][dT]-3′, siIFITM3-2: 5′-UGUCCAAACCUUCUUCUCU [dT][dT]-3′. Cells were transfected with siRNAs (50 nM each well) using RNAiMax (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s instructions. Transfected cells were treated with IFN-α2b (10,000 U/mL) or medium alone for 24 h and then infected with 1 PFU/cell of VTT-EGFP. At 24 h postinfection, the percentage of infected cells was determined by flow cytometry. Alternatively, IFITM3 expression or silencing was confirmed by Western blot.

Cells were treated with siRNA for 48 h prior to infection with VTT at PFU/cell of 0.1. At indicated time points, cell viability was assessed by using the MTS/PMS method as described by the manufacturer (Promega, Madison, WI, USA) and determined by measuring the absorbance at 490 nm using a 96-well plate absorbance reader. A time-response curve was created by nonlinear regression analysis. Each assay was designed with six replicates.

Cells were transfected with pLV-IFITM3 using X-tremeGENE HP DNA transfection reagent (Roche, Basel, Switzerland). After 48 h, cells were selected in complete medium containing 4 µg/mL puromycin (Sigma, Saint Louis, MO, USA). To assess the expression of IFITM3, cell lysates were harvested stably and examined by Western blot analysis with anti-FLAG or IFITM3 antibodies.

All cells used in the study were infected with VTT or VTT-EGFP at the indicated multiplicity of infection. The inocula were removed after 1 h and the cells were washed with PBS and replaced with fresh complete culture medium. Cells were harvested at various times after inoculation, and virus infection were determined by fluorescence imaging, flow cytometry and quantitative PCR. Alternatively, the infected cell lysates were harvested and examined by Western blot with anti-VACV D8 or EGFP antibodies.

Cells were absorbed with VTT at the designated PFU/cell for 1 h at 4°C, washed, overlaid with complete DMEM medium containing 1% methylcellulose and cultured at 37°C for 2–3 days. Cells were fixed with 10% formaldehyde and plaques were visualized and their counts were estimated with crystal violet staining and subjected to statistical analysis.

Prechilled BHK-IFITM3 or control cells were incubated with 5 MOI of VTT on ice for 1 h to permit virus binding but impede their cell entry. Unbound virus was removed; cells were washed, fixed and processed for immunofluorescence analysis with an anti-D8 antibody and visualized by a confocal laser-scanning microscope (60 × objective lens/oil). Alternatively, virus-bound cell genome was extracted to determine the amount of viral DNA accumulation by quantitative PCR of EGFP and VACV E3L genes. To assess VACV entry into cells, virus inocula were removed after binding on ice for 1 h, and cells were washed and cultured with the growth medium for 30, 60, 90, and 120 min at 37°C. Total cellular RNA was extracted to measure the relative quantity of VACV early or late gene transcription (E3L or D8) by qRT-PCR in order to determine the early postentry steps of viral life cycle and infection.

Vaccinia virus particles were labeled with DiD for 20 min at room temperature in the dark, followed with removal of non-incorporated DiD as described previously (33). DiD-labeled virus particles were incubated with prechilled BHK or 293T cell lines on ice for 1 h. Cells were then washed, trypsinized and then fixed with 4% paraformaldehyde/PBS to quantify using a FACSCalibur flow cytometer.

Cells were adsorbed with VTT-EGFP at indicated MOIs at 4°C for 1 h. After adsorption, cells were washed with cold PBS to remove unbound virions. Thereafter, the cells were either mock-treated or treated with 10 or 20 mM NH₄Cl and cultured at 37°C as described previously (32). After 24 h, the infection was analyzed by measuring the EGFP-positive cells with a FACSCalibur flow cytometer.

Total cellular RNA was extracted using TRIzol reagent (Life technologies, USA), and then was reverse transcribed with M-MLV reverse transcriptase (Promega) to cDNA. Quantitative real-time PCR was performed using SYBR Green Master Mix (TOYOBO, Japan) on an ABI 7500 Real-Time PCR System (Applied Biosystems, USA). For relative quantitation analysis, samples were normalized based on the expression of the gene encoding human GAPDH as a reference. The specific primers used were as follows:

IFITM3 (forward, 5′-ATGTCGTCTGGTCCCTGTTC-3′ and reverse, 5′-GTCATGAGGATGCCCAGAAT-3′);

The membrane proteins from cells were isolated with Minute™ plasma membrane protein isolation kit following the instruction of the manufacturer (Invent Biotechnologies, Eden Prairie, USA). Cells were collected and resuspended with buffer A supplemented with protease inhibitor cocktail (Roche, Basel, Switzerland). The total membrane proteins including organelles and plasma membrane were isolated and evaluated by Western blot as described previously (37), with β-actin as cytosolic marker and CD81 as membrane protein marker.

Cells were cultured on glass coverslips, fixed with 4% paraformaldehyde in PBS for 30 min and permeabilized with 0.25% Triton X-100 according to the specific requirement. Blocking was performed for 1 h with 3% BSA in PBS containing 0.3 M glycine. Cells were next incubated with VACV D8 antibody, IFITM3 antibody or LBPA antibody at room temperature for 2 h. After washing, the cells were incubated with anti-mouse IgG conjugated to Alexa Fluor 555 (AF555), anti-rabbit IgG conjugated to FITC and anti-mouse IgG conjugated to Cy3, respectively, at room temperature for 1 h. Intracellular cholesterol was stained with Amplex^® Red Cholesterol Assay Kit (Molecular Probes, Inc., Invitrogen) based on the protocol of the manufacturer. Nuclei were stained with DAPI (Invitrogen). Images were detected with a fluorescence microscope or the Leica TCS SP5 Confocal microscope (63 × objective lens).

Data were calculated as the mean ± SD. Student’s t-tests were performed for all analyses using GraphPad Prism 6 software (GraphPad Software Inc., CA, USA). Differences among groups were determined by one-way ANOVA with repeated test. Statistical significance was determined by two-tailed P values: *P < 0.05, **P < 0.01, and ***P < 0.001.

To determine the role of endogenous IFITM3 in the interaction between the host cells and vaccinia virus, we first estimated the expression of IFITM3 in HeLa, A549, and 293T cells infected by vaccinia virus, respectively. Notably, we observed that endogenous IFITM3 expression was only transiently and moderately upregulated in A549 and 293T cells but not in HeLa after 2–6 h infection, and then shut down in a timely manner following infection with VACV at 0.1 PFU per cell (Figure 1A) in HeLa, A549, and 293T cells. These findings led us to reinvestigate whether VACV infection affected the signal transduction pathway of ISGs expression. First of all, we found that the expression of IFITM1 or IFITM3 in HeLa cells was further induced following IFNα2b treatment (Figure 1B) and subsequently IFNα2b markedly inhibited VACV infection (Figure 1C). Obviously, this finding indicate that IFNα can induce the activation of signal transduction pathway of ISG expression in HeLa cells and thus exhibit a strong antiviral activity. In order to further investigate how vaccinia virus affected the pathway of type I interferon production in HeLa or 293T cells, the expression of IFNα and IFNβ was measured in mRNA and protein levels. Vaccinia virus could trigger the upregulation of IFNα and IFNβ transcription at 24 or 48 h postinfection (Figures 1D,E), but the corresponding protein expression could not be detected by Western blot. Interestingly, vaccinia virus infection stimulated IRF3 phosphorylation, but simultaneously down regulated its expression in HeLa cells (Figure 1F). However, we were unable to detect STAT1 phosphorylation in HeLa cells infected with vaccinia virus (data not shown). Therefore, we preliminarily judged that HeLa cells are unable to produce type I interferon, and down regulation of IFITM3 is independent of IRF3 phosphorylation activated by vaccinia virus infection. Thus, we speculated that poxviruses had to reprogram host cell translation to benefit the viral life cycle (38). During the viral infection and replication, the majority of host transcripts and its expression into proteins was suppressed or even shut off (39). Therefore, the regulation of expression of IFITM3 as an intrinsic host defense factor that are often induced by virus infection, might also be associated with vaccinia virus.

Earlier studies showed that the depletion of IFITM3 increased IAV (H1N1) infection in primary lung fibroblasts and HeLa cells (4), so we first assessed whether endogenous IFITM3 also interrupts poxvirus infection. In order to accomplish this, the effect of siRNA on the expression of endogenous IFITM3 in HeLa and 293T cells was evaluated, and as expected IFNα2b induced the expression of IFITM3 (Figures 2A,C). In IFNα2b-treated cells, the percentages of infection were significantly suppressed in both HeLa and 293T cells, and knockdown of IFITM3 slightly increased the VTT infection (Figures 2B,D, gray bars). Interestingly, the anti-VACV activity of IFNα2b in HeLa cells was significantly stronger compared to that in 293T cells. However, in the absence of IFNα2b, knockdown of endogenous IFITM3 had no impact on virus infection in both HeLa and 293T cells (Figures 2B,D, black bars). Consistent with other previous studies (40), perhaps VACV had resistance to the inhibition of endogenous IFITM3, which may play a major role in the antiviral effects of interferon.

Although depletion of IFITM3 had no affect on the percentage of infected cells, the effect of IFITM3 on the cytopathogenicity of the virus was uncertain. We therefore investigated whether depletion of IFITM3 had any impact on the viral cytopathogenecity. The siRNA treatments or knockdown of IFITM3 had no significant influence on the cell proliferation (Figures S1A–C in Supplementary Material). Notably, depletion of IFITM3 enhanced the cell death caused by VTT infection (Figures 2E,F), and overexpression of IFITM3 in 293T cells enhanced the cellular antiviral activity (Figures S1D,E in Supplementary Material), indicating that endogenous IFITM3 modulated viral cytopathicity.

Our preliminary experiment showed that transient expression of IFITM3 in vitro could affect VACV infection (data not shown). In order to precisely evaluate the role of IFITM3 in VACV infection, A549, 293T, HeLa, Vero, and BHK-21 cells stably expressing Flag-N-tagged IFITM3 were generated, and the expression of IFITM3 was analyzed by Western blot (Figure S2A in Supplementary Material). Thereafter, the cells were infected with VTT/EGFP at different multiplicities of infection (PFU/cell) of 0.2, 0.1, and 0.05 for 24 h. We found that the viral cytopathic effect and percentages of infected cells were significantly suppressed by IFITM3 expression in 293T and Vero cells (Figures 3A–C). Furthermore, the infection was also inhibited by the expression of IFITM3 in BHK-21 and A549 cells (Figures S2B,C,E in Supplementary Material). Surprisingly, this restriction role of IFITM3 expression was not observed in HeLa cells (Figures S2D,E in Supplementary Material), and the antiviral activities of IFITM3 expression in 293T, BHK-21, and Vero cells were superior to that in A549 cells against VACV infection. Consistent with the previous report (11), the efficiency of IFITM3-mediated restriction in this study was associated with the cell type and differences in expression. The inhibitory effect of IFITM3 was further confirmed by expression analysis of VTT D8 (cell surface-binding protein) and EGFP reporter in Vero cells simultaneously (Figure 3D). Together, these results showed that IFITM3 expression could inhibit the initial infection and the efficiency of restriction was cell type dependent.

To further address whether IFITM3 could inhibit the VACV replication cycle in host cells, the extent of virus replication in IFITM3-overexpressing cells was analyzed by quantifying the viral gene expression at the transcriptional and translational levels (e.g., viral late gene D8 and inserted gene EGFP) at different time points postinfection with VTT-EGFP. The transcriptional level of viral D8 in IFITM3-positive cells was significantly lower than that in control cells (Figures S3A,E in Supplementary Material). Theoretically, potent inhibition of viral gene transcription should interfere with viral translation. As expected, IFITM3 delayed and blocked the expression of viral D8 protein (Figures S3C,D,F–H in Supplementary Material), suggesting that the exogenous expression of IFITM3 suppressed VACV replication.

To further confirm the inhibitory role of IFITM3 on virus cell-to-cell transmission and diffusion, plaque formation analysis was carried out by crystal violet staining in Vero cells. The expression of IFITM3 in Vero cells strongly suppressed the cytopathic effect (CPE) and plaque formation of sequential rounds of virus infection (Figure 4).

Prior studies have revealed that the IFITM3 protein restricts virus infection at the early stage of the viral replication cycle of several RNA virus, such as IAV, WNV, and DENV (4, 11, 24). We therefore speculated that the inhibition of IFITM3 might also occur at the early steps of the VACV lifecycle. To verify this speculation, we first tested the impact of IFITM3 on the initial phase of infection by incubating BHK-IFITM3 with VTT (5 PFU/cell) on ice for 1 h to permit viral binding but with preventing viral entry into the cells (33). Compared with the vector cells, we observed the loss of VTT D8 signal in IFITM3-positive cells at 30, 60, and 90 min postinfection (Figure S4 in Supplementary Material). Interestingly, the D8 signal observed in IFITM3-positive cells seemed to be lower than that in vector cells after adsorption (Figure S4 in Supplementary Material). Next, we exploited the characteristics of the poxvirus replication cycle to evaluate the inhibition of IFITM3 on virus entry. As described in Figure 5A, the VACV early gene transcription and translation occurred within the viral core, and viral DNA replication initiated within a few hours (about 2–12 h) after virus entering the cytoplasm leading to intermediate and late phases of gene expression (34, 36, 38). To verify this speculation, 293T or BHK-21 cells expressing IFITM3 were incubated with VTT (5 PFU/cell) at 4°C for 1 h to permit viral binding (Figure 5A), and the early (e.g., E3L) or late (e.g., D8) gene transcription was quantified by real-time PCR to serve as an indicator to determine the inhibition efficiency of IFITM3 to VACV entry. Interestingly, we found that expression of IFITM3 inhibited VACV binding (Figure 5B). Moreover, the transcription of the early gene E3L and late gene D8 in IFITM3-expressing cells was delayed and restricted (Figures 5C–F), suggesting that IFITM3-mediated inhibition of VACV infection and replication occurs at the virus entry stage.

To further determine whether expression of IFITM3 affected VACV attachment, labeling assays with fluorescence-labeled antibody against VACV D8 protein or fluorescent probe DiD to virus particle were performed as mentioned previously in the methods (Figure 6A) (32, 41). Immunofluorescence images suggested that the presence of stably expressed IFITM3 could block viral-cell binding (Figure 6B). To better quantify these effects, purified VTT particles were labeled with DiD, a lipophilic membrane dye, and the labeling rate was determined to be approximately 99.21% (Figure 6C). DiD-labeled virus particles were incubated with IFITM3-positive BHK and vector cells on ice for 1 h, and then DiD-positive cells and DiD mean fluorescence intensities were quantified by flow cytometry. Compared to vector cells, the percentage of IFITM3-positive cells and DiD mean fluorescence (DiD fluor.) intensities (MFI) of IFITM3-positive cells bound with virus particles was significantly lower (Figures 6D,E), indicating that IFITM3 interrupted VTT binding consistent with the data in Figure 5B.

Several reports provided the evidences that IFITM3 is localized in the late endosomal and lysosomal structures (42, 43) with CD63, Rab7, LAMP1, or LAMP2, and thus inhibit the entry of viruses, such as IAV, WNv, and EBOV into these cellular organelles. It was previously reported that the Poxviruses enter the cells either through the plasma membrane at neutral pH or through a low pH-dependent endocytic route (32). In order to elucidate the mechanism of IFITM3 against VACV, we investigated its subcellular localization. IFITM3 was mainly localized in the cell membrane structures by extracting the membrane and cytosol proteins of IFITM3-positive 293T and HeLa cells (Figure 7A). Confocal microscopy revealed that IFITM3 is partially localized in LBPA-positive late endosomes and lysosomes in HeLa cells (Figure 7B). However, LBPA, a phospholipid that has an important role in cholesterol homeostasis, was absent in 293 cells (Figure 7C), whereas, high levels of cholesterol was detected in these cells (Figures 7D,E). Moreover, IFITM3-positive cellular compartments showed much higher degrees of cholesterol (Figure 7F), but the expression of IFITM3 displayed little or no effect on the accumulation of endosomal cholesterol (Figure 7E). Of note, IFITM3 was expressed at cellular membrane lipid raft, and localized at late endosomes and lysosomes, providing a significant correlation of its physical location with its antiviral mechanism.

Since IFITM3 localizes to endosomal compartments including macropinosomes required for VACV entry triggered by a low pH, we further characterized its impact on VACV entry. An endosomal acidification inhibition assay was performed in the absence or presence of ammonium chloride (NH₄Cl), which neutralizes endosomal pH and consequently prevents the low-pH-dependent activation and triggering of fusion between the viral envelope and endosomal membrane. The results showed that NH₄Cl treatment reduced the infectivity of VTT by approximately 60% in 293T or BHK cells and inhibited plaque formation of VTT in Vero cells (Figures 8A–C). Moreover, IFITM3-positive cells exhibited lower levels of infection than control cells in the presence of NH₄Cl, and no difference in the percentages of IFITM3-positive cells infected by VTT-EGFP was observed in the presence or absence of NH₄Cl (Figures 8A,B). These results suggested that IFITM3 did not affect VACV entry in a low-pH independent pathway. Therefore, we assumed that IFITM3 might exert antiviral activity during viral endosomal trafficking. However, the infection of A549 cells with VTT did not show any significant affect by NH₄Cl (Figure 8D), perhaps suggesting that VACV infects A549 cells mainly in a pH-independent fusion. To understand this observation, A549 cell lines were incubated with virions at 4°C for 1 h to permit attachment but with preventing viral entry. The cells were then exposed briefly to buffer at pH 7.4, pH 5.0, or pH 4.0 for 5 min before incubation with regular medium at neutral pH for 24 h. The results showed that the infection of VTT in A549 cells was not accelerated by low pH, and had no affect the antiviral activity of IFITM3 (Figure 8E), indicating that VTT may enter A549 cells either through the plasma membrane or a neutral pH endocytic route. However, we performed a similar experiment in 293T cell-derived lines and found an enhancement of infection in 293T cells following low-pH treatment of cell-binding virions in the presence or absence of IFITM3. Moreover, the percentage of infected 293T cells overexpressing IFITM3 was higher than that of vector control cells infected with VTT-EGFP (Figure 8F). These results indicated that a low pH could antagonize the antiviral activity of IFITM3 in 293T cells. Taken together, these findings suggested that IFITM3 could mainly block the cytosolic entry of the low pH-dependent virus.