Combined approaches of EPR and NMR illustrate only one transmembrane helix in the human IFITM3

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

Interferon-inducible transmembrane protein IFITM3 was known to restrict the entry of a wide spectrum of viruses to the cytosol of the host. The mechanism used by the protein to restrict viral entry is unclear given the unavailability of the membrane topology and structures of the IFITM family proteins. Systematic site-directed spin labeling (SDSL) and electron paramagnetic resonance (EPR) studies of IFITM3 in detergent micelles identified a single, long transmembrane helix in the C-terminus and an intramembrane segment in the N-terminal hydrophobic region. Solution NMR studies of the same sample verified the secondary structure distribution and demonstrated two rigid regions interacting with the micellar surface. The resulting membrane topology of IFITM3 supports the mechanism of an enhanced restricted membrane hemi-fusion.

Related collections

Most cited references 25

Record: found
Abstract: found
Article: found

Is Open Access

IFITM3 Inhibits Influenza A Virus Infection by Preventing Cytosolic Entry

Eric Feeley, Jennifer S. Sims, Sinu P. John … (2011)

Introduction The 2009 H1N1 pandemic provided a strong reminder of the threat that influenza A virus poses to world health (http://www.cdc.gov/h1n1flu/cdcresponse.htm). The most effective means of protection against influenza is the seasonal vaccine. However, if the vaccine does not match the viral strains, its effectiveness can be reduced to 50% or less [1], [2]. Among small molecules, only two approved influenza drugs remain effective, zanamivir (Relenza) and oseltamivir (Tamiflu). Although resistance to zanamivir is rare, there has been an increase in oseltamivir-resistant flu strains [3]. Of concern, both drugs target viral neuraminidase (NA), precluding combinatorial therapy to minimize resistance [4], [5]. Thus, research to identify new anti-influenza strategies would be useful. The influenza A virus is 50–100 nm in size, encodes for up to 11 proteins, and contains eight segments of negative single-stranded genomic RNA (3). Influenza A virus infection initiates with the cleavage and activation of the viral hemaglutinnin (HA) envelope receptor by host proteases [6], [7], [8], [9]. HA then binds to sialylated proteins on the cell surface, eliciting endocytosis of the viral particle. Endocytosed viruses are transported through the early and late endosomes, with late endosomal acidification triggering a conformational change in HA which results in viral-host membrane fusion [6], [10]. Fusion transitions from a hemifusion intermediate into a fusion pore through which the virus' eight viral ribonucleoproteins (vRNPs) enter the cytosol. The vRNPs are subsequently guided by the host cell's karyopherins into the nucleus [11], [12], [13], wherein the viral RNA-dependent RNA polymerase synthesizes viral genomes (vRNA) and mRNAs, both of which are exported to the cytosol, culminating in the production of viral progeny. Genetic screens have identified multiple host factors and pathways which modulate influenza A virus infection in vitro [14], [15], [16], [17]. Using such a genetic screen, we identified the IFITM protein family members IFITM1, 2 and 3 as antiviral factors capable of blocking influenza A viruses [14]. We further tested the antiviral activity of IFITM3 protein using the seasonal influenza A strains, A/Uruguay/716/07 (H3N2) and A/Brisbane/59/07 (H1N1), and found similar levels of IFITM3-mediated viral inhibition [14]. IFITM3 accounts for a significant portion (50–80%) of IFN's (type I or II) ability to decrease influenza A virus infection in vitro, and IFITM3 resides in vesicular compartments that are IFN-inducible [14]. In addition, the IFITM family inhibits infection by the flaviviruses, dengue virus and West Nile virus [14], [18], as well as the filoviruses, Ebola and Marburg, and the SARS coronavirus [19]. The IFITM proteins also block vesicular stomatitis virus-G protein (VSV-G)-mediated entry, but do not substantially alter the replication of Moloney leukemia virus (MLV), several arena viruses, or hepatitis C virus (HCV, [14], [20]). The human IFITM proteins were identified 26 years ago based on their expression after IFN stimulation [21], [22], [23]. The IFITM1, 2, 3 and 5 genes are clustered on chromosome 11, and all encode for proteins containing two transmembrane domains (TM1 and 2), separated by a conserved intracellular loop (CIL, [22]), with both termini extra-cellular or intra-vesicular [24], [25]. TM1 and the CIL are well conserved between the IFITM proteins and a large group of proteins representing the CD225 protein family. CD225 family members exist from bacteria (125 members) to man (13 members, with 156 members in chordata), with no in depth functional data available for any member other than the IFITM proteins. IFITM1, 2 and 3 are present across a wide range of species including amphibians, fish, fowl and mammals. The IFITM proteins have been described to have roles in immune cell signaling and adhesion, cancer, germ cell physiology, and bone mineralization [25], [26], [27], [28], [29], [30]. IFITM3 expression can inhibit the growth of some IFN-responsive cancer cells [31]. Genetic evidence also points to IFITM5/Bril being required for early bone mineralization [30], [32]. IfitmDel mice, which are null for all five of the murine Ifitm genes, display a 30% perinatal mortality among null pups, but thereafter grow and develop normally in a controlled setting [26]. However, cells derived from these IfitmDel mice are more susceptible to influenza A virus infection in vitro [14]. IFITM3 inhibited infection by all influenza A virus strains tested including a 1968 pandemic isolate and two contemporary seasonal vaccine viruses [14]. We have found IFITM3 to be the most potent of the IFITM protein family members in decreasing influenza A virus replication [14]. Viral pseudoparticles are differentially inhibited by the IFITM proteins based on the specific viral receptors expressed on their surfaces [14], [19]. Therefore, we have hypothesized that IFITM proteins inhibit susceptible virus families (Orthomyxoviridae, Flaviviridae, Rhabdoviridae, Filoviridae, and Coronaviridae) during the envelope-dependent early phase of the infection cycle, which extends from viral binding to cell surface receptors through the creation of the fusion pore between viral and host membranes [14], [19], [20]. In support of this notion, recent work demonstrated that IFITM protein overexpression did not prevent influenza A virions from accessing acidified compartments [19]. Consistent with its acting on endocytosed viruses, a portion of IFITM3 resides in structures that contain host cell endosomal and lysosomal proteins [19]. Furthermore, inhibition of influenza A virus infection depends on the palmitoylation of IFITM3, a post-translational modification that targets proteins to membranous compartments [33]. Here we directly test the idea that IFITM3 restricts influenza A viral infection during the envelope-dependent early phase of the viral lifecycle. Consistent with previous studies, we find that IFITM3 inhibits influenza A viral infection after viral-host binding and endocytosis, but prior to primary viral transcription [19], [20]. Moreover, using a combination of assays, we find that either IFN or high levels of IFITM3 impede influenza A viruses from transferring their contents into the host cell cytosol, and that IFITM3 is necessary for this IFN-mediated action. Therefore, we conclude that IFN is acting predominantly through IFITM3 to block viral fusion. We also find that IFN expands the late endosomal and lysosomal compartments, and that IFITM3 overexpression is sufficient for this phenotype. This study also presents data showing that IFITM3 overexpression leads to the expansion of enlarged acidified compartments consisting of lysosomes and autolysosomes. Interestingly, we observe that viruses trapped in the endocytic pathway of IFITM3-overexpressing cells are trafficked to these expanded acidified compartments. Based on these results and those of others [19], [20], we present a model whereby IFN acts via IFITM3 to prevent viral fusion, thereby directing endocytosed viruses to lysosomes and autolysosomes, for subsequent destruction. Collectively this study expands our understanding of how IFITM3 restricts a growing number of viruses by exploiting a shared viral vulnerability arising from their use of the host's endocytic pathway. Results IFITM3 inhibits influenza A viral infection after viral-host binding but prior to viral transcription The inhibition of HA-expressing pseudoparticles by the IFITM proteins pointed towards restriction occurring during the envelope-dependent phase of the viral lifecycle [14]. Therefore we tested IFITM3's impact on the most proximal phase of infection, viral binding, by incubating influenza A virus A/WSN/33 H1N1 (WSN/33, multiplicity of infection (moi) 50) with A549 lung carcinoma cells either stably overexpressing IFITM3 (A549-IFITM3) or an empty vector control cell line (A549-Vector, Fig. 1A). Samples were incubated on ice to permit viral binding but prevent endocytosis. After incubation, cells were washed with cold media, fixed and stained for HA. When analyzed by flow cytometry, we observed no appreciable difference in surface bound HA between the vector and IFITM3 cells. There was also no difference in surface-bound virus over a series of ten-fold dilutions of viral supernatant (data not shown). We also determined that the stable expression of IFITM3 did not alter the surface levels of (α2, 3) or (α2,6) sialylated cell-surface proteins (Fig. S1). 10.1371/journal.ppat.1002337.g001 Figure 1 IFITM3 inhibits infection after viral binding but before viral transcription. A) A549 cell lines were incubated on ice with H1N1 WSN/33 to permit viral-host binding. Cells were washed, fixed and immunostained for surface-bound HA protein, then analyzed by flow cytometry. Values given are percentage of cells staining for surface HA. Values are representative of three independent experiments. B) MDCK cells transduced with the empty vector control (Vector) or IFITM3 were incubated with A/Puerto Rico/8/34 H1N1 (PR8) on ice. Warm media was added at time zero. Cells were then fixed at the indicated time points and hybridized with RNA probes against the viral NP mRNA (red) and stained for DNA (blue), then imaged by confocal microscopy. Images are representative of three independent experiments. (Scale bar: 20 µm). To investigate IFITM3's impact on initial viral mRNA production, we infected canine kidney cells, either expressing IFITM3 (MDCK-IFITM3) or the empty vector (MDCK-Vector), with influenza A virus (A/Puerto Rico/8/34 H1N1 (PR8), moi 500). We used PR8 because of the purified high titer stocks available. Next, the viral supernatant was removed and warm media was added (0 min). At the indicated times, cells were processed and stained for the positive stranded NP mRNA of PR8 using a specific RNA probe set (red, Fig. 1B), then imaged on a confocal microscope. Based on NP mRNA staining, primary viral transcription begins by 60 min. p.i. in the vector control, with the NP mRNA signal increasing through to 180 min., when the export of viral mRNAs to the cytosol can be observed. A decrease in primary viral transcription can be seen when comparing the IFITM3 cells to the vector control line. Therefore, IFITM3 inhibits influenza A viral infection after viral-host binding but before primary viral mRNA transcription. IFN interferes with vRNP nuclear entry and IFITM3 is necessary and sufficient for this antiviral defense We next used confocal imaging to track the nuclear translocation of vRNPs (Fig. 2 [34], [35]). At the start of infection, the NP within infected cells is complexed with viral genomic RNA forming vRNPs. Therefore, immunostaining for NP permitted us to follow vRNP distribution intracellularly [16], [34], [36]. Normal diploid human lung fibroblasts (WI-38 cells) were stably transduced with empty vector (Vector), IFITM3 cDNA (IFITM3), or short hairpin RNAs (shRNA) either against IFITM3 (shIFITM3) or a scrambled non-targeting control (shScramble, Fig. 2, S2). WI-38s were chosen because of their normal karyotype and relatively larger and flatter morphology. Cells were first incubated on ice with PR8 (moi 500). Next, the viral supernatant was removed and warm media was added (0 min). At the indicated times after warming, cells were fixed, permeabilized, stained for NP and DNA, and imaged on a confocal microscope. Image analysis software was used to create an outline of each cell's periphery (white lines) and nucleus (blue lines). Based on NP staining, vRNPs arrive in the nuclei by 90 min in the vector control, shIFITM3, and in the shScramble cells, with the NP signal increasing through to 240 min (Fig. 2A, S2A–D). 10.1371/journal.ppat.1002337.g002 Figure 2 IFN prevents vRNP nuclear entry, and IFITM3 is necessary and sufficient for this action. A) Normal diploid human fibroblasts (WI-38 cells) were stably transduced with retroviruses containing either IFITM3 (IFITM3), a shRNA against IFITM3 (shIFITM3), an empty viral vector alone (Vector), or a non-targeting control shRNA (shScramble, Fig. S2). Cells were incubated with PR8 on ice, and then warm media was added at time zero. Cells were fixed at the indicated times p.i. and stained for NP (green) and DNA and analyzed by confocal microscopy. Image analysis software was used to define each cell's cytosolic (white lines) and nuclear peripheries (blue lines, based on DIC images and DNA staining, respectively). Red arrows: cytosolic compartments containing NP. Images are representative of four independent experiments. (Scale bar: 15 µm). B) As in (A) except that cells were treated with IFN-α prior to infection. In contrast, we observed decreased nuclear and increased cytosolic NP staining in the IFITM3 cells (Fig. 2, S2C). Moreover, in the IFITM3 cells greater than 60% of the cytosolic NP colocalized with Lysotracker Red (LTRed), a dye which marks acidic cellular compartments (late endosomes, lysosomes, pH≤5.5), and which was added to the warm media at time zero (Fig. S2A, D). The increased NP in the cytosol of the IFITM3 cells likely arises in part from an increase in the local concentration of viruses because α-NP Western blots (after trypsinizing the cells to remove adherent NP) did not show substantial differences in internalized NP levels between cell lines for up to 90 min post infection (p.i., data not shown). Because IFITM3 is required for the anti-viral actions of IFN in vitro [14], we performed a companion experiment with the WI-38 cells treated with IFN-α (Fig. 2B). IFN-α treatment also decreased NP nuclear staining in the WI-38-Vector cells, however this block was not as complete nor was it associated with similar levels of cytosolic NP staining as those seen with high levels of IFITM3. Consistent with the gain-of-function data, the depletion of IFITM3 decreased IFN's ability to block vRNP trafficking to the nucleus (Fig. 2A and B, compare top and bottom rows). Similar results were obtained either using A549 cells (Fig. S3) or using MDCK cells, with the latter experiments employing additional influenza A viral strains (X:31, A/Aichi/68 (Aichi H3N2), Fig. S4A–C, WSN/33 and A/Victoria/3/75 H3N2, data not shown). It is important to note that the levels of IFITM3 protein in the A549-IFITM3 cells are higher than those seen after treatment with IFN-α or -γ (Fig. S3C). However, we have not observed that other overexpressed proteins have either protected against viral infection or expanded the lysosome/autolysosome compartment (data not shown), arguing that this is a specific effect. To better assess the expanded LTRed compartments observed with IFITM3 overexpression, we created MDCK cells stably expressing the lysosomal protein, LAMP1, fused to a red fluorescence protein (LAMP1-RFP) and IFITM3. As compared to control cells, the IFITM3 cells demonstrated extensive colocalization (>60%) between the NP and LAMP1-RFP signals, revealing that the entering viruses are trafficked to lysosomal compartments (Fig. S5). We extended this analysis by directly tracking the location of the vRNA contained in the incoming vRNPs. MDCK cells stably expressing an empty vector or IFITM3, were used in time-course experiments as above (Fig. 3A–D). At the indicated times, cells were processed and stained for the negative stranded NP vRNA of PR8 using a specific RNA probe set (green). As seen with the WI-38 cells, we observed the nuclear translocation of vRNA by 80 min p.i. in the MDCK-vector cells (Fig. 3A). The nuclear vRNA signal was strongly decreased with IFITM3 overexpression based on the average number of vRNA particles present per nucleus (Fig. 3C). Consistent with the WI-38 results, the vRNAs accumulated in the cytosol of the IFITM3 cells, with >50% co-localizing with LTRed-staining acidic structures (Fig. 3D). Similar levels of retained cytosolic vRNPs were observed in experiments without LTRed (data not shown). Interestingly, we observed the loss of the vRNA signal in the acidic inclusions of the MDCK-IFITM3 cells between 80 and 240 min. p.i. (Fig. 3B). By comparison, the vRNAs in the control cells increased in number in both the nucleus and cytosol, as would be expected with the nuclear export of newly synthesized viral genomes [36]. 10.1371/journal.ppat.1002337.g003 Figure 3 IFITM3 overexpression leads to both a retention of viral genomes in the cytosol, and a decrease in viral genomes entering the nucleus. MDCK cells transduced with the empty vector control (A) or IFITM3 (B) were incubated with PR8 on ice. Warm media containing lysotracker red dye (LTRed, red) was added at time zero. Cells were then fixed at the indicated time points and hybridized with RNA probes against the viral NP genome (NP vRNA, green) and stained for DNA, then imaged by confocal microscopy. Image analysis software was used to define the nuclear boundaries (blue lines) based on DNA staining. Images are representative of four independent experiments. (Scale bar: 20 µM). C) Quantitation of nuclear vRNA particles. The number of viral RNA particles per nucleus of the MDCK-Vector and IFITM3 cells at the indicated time points are shown. Values represent the mean +/− the SD of three independent experiments. D) Percent colocalization of vRNA and LTRed-containing compartments for MDCK-Vector and IFITM3 cells lines treated as in A and B, at the indicated time points. We next evaluated vRNP translocation in murine embryonic fibroblasts (MEFs) derived from animals that have had all five Ifitm genes deleted (IfitmDel−/−, [14], [26]). Compared to wild-type (WT) matched litter mate controls, the IfitmDel−/− MEFs displayed 5–10 fold more nuclear NP staining, with or without IFN-γ treatment (Fig. 4, S6C). IFN-mediated viral restriction was restored when we transduced the null MEFs with a retrovirus expressing Ifitm3 (IfitmDel−/− Ifitm3, Fig. S6). Similar to what was observed with the IFITM3 overexpressing cell lines, the majority of the vRNP signal in the IFN-γ-treated WT and Ifitm3-rescued cells localized to acidic compartments (red, Fig. S6B). An increase in acidic compartments occurred after IFN-γ treatment with either the WT or the IfitmDel−/−Ifitm3 MEFs, but not in the IfitmDel−/− cells, suggesting that Ifitm3 is required for this event (Fig. 4, S6). Similar results were obtained with IFN-α (data not shown). We conclude from these experiments using orthologous reagents (cell lines and influenza A viruses) and methods, that IFN impedes vRNP nuclear entry, and IFITM3 is necessary and sufficient for this activity. 10.1371/journal.ppat.1002337.g004 Figure 4 Ifitm knockout cells are more vulnerable to vRNP nuclear entry and are rescued by the restoration of Ifitm3 expression. MEFs, either A) wild type (WT) or B) IfitmDel−/−, which are missing all five of the mouse Ifitm proteins, were either left untreated (left panels, Buffer), or treated (right panels) with IFN-γ. The following day cells were incubated with PR8 on ice. Cells were next incubated in warm media containing LTRed. Cells were then fixed at the indicated times and immunostained with anti-NP antibodies (green), stained for DNA (blue), and imaged by confocal microscopy. Image analysis software was used to define the nuclear boundaries (blue lines). Images are representative of three independent experiments. (Scale bar: 12 µm). Viral pseudoparticle fusion mediated by either HA or VSV-G envelope proteins is decreased by IFN, and IFITM3 is necessary and sufficient for this activity To further characterize the mechanism of IFITM3-mediated restriction, we used an established viral fusion assay [37], [38]. Lentiviral pseudoparticles containing the β-lactamase protein fused to the HIV-1 accessory protein Vpr (BLAM-Vpr) and expressing either HA and NA (H1N1, WSN/33), or VSV-G envelope proteins, were incubated for 2 h with cells, which were then loaded with the β-lactamase flourogenic substrate, CCF2. Upon viral pseudoparticle fusion, BLAM-Vpr enters the cytosol and cleaves CCF2, producing a wave length shift in emitted light (from green to blue) when analyzed by flow cytometry (Fig. 5A, [37]). In MDCK-IFITM3 cells we observed a decrease in both HA- and VSV-G-directed fusion, which was comparable to the block produced by poisoning of the host vacuolar ATPase (vATPases) with a low dose of bafilomycin A1 (Baf, Fig. 5B). The inhibition of vATPases prevents the low-pH activation required by these two viral envelope proteins to produce membrane fusion. A block to fusion of pseudoparticles expressing H1 (PR8), H3 (A/Udorn/72), H5 (A/Thai/74) or H7 (A/FPV/Rostock/34) subtypes of HA was also detected with MDCK cells or with chicken embryonic fibroblasts (ChEFs), in which IFITM3 strongly inhibited viral replication (Fig. S7A, B, C). In the case of the MDCK cells, the block to fusion closely paralleled the level of inhibition seen when the pseudoparticles were tested for productive infection using HIV-1 p24 expression as a readout (Fig. S7E). Consistent with earlier findings, pseudoparticles expressing an amphotropic MLV envelope protein were insensitive to IFITM3, showing the specificity of these results (Fig. S7D). Similarly to its effect on H5-expressing pseudoparticles, IFITM3 inhibited replication of infectious avian H5N1 influenza A virus, A/Vietnam/1203/04 (VN/04), isolated from a fatal human infection (Fig. S7F–H). 10.1371/journal.ppat.1002337.g005 Figure 5 HA or VSV-G-mediated fusion is inhibited by IFN or IFITM3. A) Schematic model of the established viral fusion assay [37], [38] comprised of lentiviral pseudoparticles (pps) containing the β-lactamase protein fused to the HIV-1 accessory protein Vpr (BLAM-Vpr, shown in orange/red) and expressing HA and NA (WSN/33) on their surfaces. The H1N1pps were incubated for 2 h with cells, which were subsequently loaded with the β-lactamase flourogenic substrate, CCF2. Upon viral fusion, BLAM-Vpr enters the cytosol and can cleave CCF2, producing a wavelength shift from green to blue in emitted light when analyzed by flow cytometry ([37]). B) MDCK cells stably overexpressing IFITM3 (MDCK-IFITM3) or empty vector control cells (MDCK-Vector) were exposed for 2 h to viral pseudoparticles containing a BLAM-Vpr and expressing either the HA and NA envelope proteins of Influenza A virus (WSN/33, H1N1pp) or the VSV-G envelope protein (VSV-Gpp), then loaded with CCF2. After incubation with the indicated pseudoparticles, the cells were fixed and assayed for cleavage of CCF2 by determining the conversion of the fluorescence emission from 520 nm (uncleaved CCF2) to 447 nm (cleaved CCF2) using flow cytometry. Fusion of the pseudoparticles was inhibited by bafilomycin A1 (Baf). These results are representative of six independent experiments. C) IFITM3 inhibits fusion of H1N1pps in normal diploid fibroblasts. WI-38 fibroblasts stably transduced with IFITM3 (WI-38 M3) or the empty vector (WI-38 V) were exposed for 2 h to serial dilutions of H1N1pps containing BLAM-Vpr, with or without Baf. These results are representative of four independent experiments. D) Fusion of H1N1pps increases after IFITM3 knockdown. WI-38 fibroblasts stably transduced with a shRNA against IFITM3 (WI-38 shM3), a shRNA control with a scrambled sequence (WI-38 shScr), or the IFITM3 cDNA (WI-38 M3) were exposed to either no virus, H1N1pps or VSV-Gpps containing BLAM-Vpr. These results are representative of two independent experiments. E) Fusion of H1N1pps is inhibited by IFN-γ. WI-38 fibroblasts were treated with IFN-γ for 24 h or buffer alone prior to incubation with H1N1pps containing BLAM-Vpr. These results are representative of three independent experiments. To enhance our analysis, we tested two additional cell lines, WI-38 and HeLa cells. A strong block to fusion in WI-38-IFITM3 cells, similar to that of the Baf and uninfected control samples, was seen at a range of serial dilutions of pseudoparticles, as well as an increase in fusion with IFITM3 depletion (shIFITM3, Fig. 5C, D). IFN treatment inhibited fusion of the H1N1 pseudoparticles, albeit to a lesser extent than IFITM3 overexpression (Fig. 5E), and this effect was largely absent when IFITM3 was stably depleted in HeLa cells (Fig. S8). Similar results were obtained with IFN-α (data not shown). Based on these experiments using multiple cell lines and HA, VSV-G, and MLV envelope-expressing pseudoparticles, we conclude that IFITM3 is required and sufficient for an IFN-mediated block of viral pseudoparticle fusion. Importantly, the increase in pseudoparticle fusion seen when endogenous IFITM3 was depleted in either the HeLa or WI-38 shIFITM3 cell lines argues that fusion inhibition underlies the first line defense provided by endogenous, as well as overexpressed, IFITM3. MxA is an IFN-inducible large GTPase which interferes with secondary transcription during influenza A viral replication [39]. A549 cells express MxA and have been used extensively in influenza A viral replication studies [40]. Therefore to clarify the antiviral roles of IFITM3 and MxA, we tested the levels of viral replication in A549 cells stably expressing one of three shRNAs targeting IFITM3 (shIFITM3-1, -2, or -3). All three shIFITM3 cell lines showed increased infection (WSN/33 strain) and strong IFITM3 knockdown, when compared to the negative control cell line expressing a shRNA against firefly luciferase (shLuc), with or without IFN treatment (Fig. S9A, B). The majority of the protective effect of either IFN-α or γ was lost in the shIFITM3 cell lines. We next confirmed both the baseline levels, as well as the IFN-inducibility of MxA in the A549 cells (Fig. S9C). We also determined that MxA was both present and IFN-inducible in WI-38 normal fibroblasts, another cell line used in loss-of-function experiments in this work (Fig. S9D). Furthermore, IF studies of WI-38 cells showed that MxA is expressed in an IFN-inducible vesicular pattern and that these structures did not appreciably co-localize with vesicles containing IFITM3 (Fig. S9E, [39]). We conclude that MxA is expressed in the A549 and WI-38 cell lines, but cannot fully compensate for loss of the antiviral actions of IFITM3. IFITM3 is present in endosomes and lysosomes and these compartments are expanded with IFITM3 overexpression or IFN treatment Our data demonstrate that IFN or IFITM3 inhibit viral fusion. Influenza A virus fuses with the host membrane in late endosomes when the pH decreases to 5 [6], [7], [41]. Rab7 is a late endosomal/lysosomal small GTPase that is required for the fusion of many pH-dependent viruses, including influenza A virus [6], [41]. Previous reports have shown that IFITM3 colocalizes with LAMP1 and CD63, components of lysosomes and multivesicular bodies, respectively [19]. However, the relationship of IFITM3 and Rab7 within the host cell infrastructure remains unknown. Therefore we investigated the location of IFITM3, by undertaking immunoflourescence (IF) studies using antibodies that recognize IFITM3, Rab7, or LAMP1 [42]. Although the baseline level of IFITM3 in the A549-Vector cells was low, there was partial colocalization observed with either Rab7 or LAMP1 (Fig. 6A–D, 7A,). IFITM3 also partially colocalized with LAMP1 and LTRed-containing structures seen with IFITM3 overexpression (Fig. 6A, B, 7A). Interestingly, either IFITM3 overexpression or IFN increased the staining intensity of Rab7 and LAMP1 (Fig. 7A, B, S10A). Partial colocalization of IFITM3 was also seen with either endogenous LAMP1, or an exogenously expressed Rab7-yellow fluorescence fusion protein (Rab7-YFP) in MDCK cells (Fig. 6E–I). However, in all cases, co-localization was not complete because cells contained areas that uniquely labeled for each of the proteins. Western blots indicated that IFITM3 over-expression led to modest increases in both LAMP1 and Rab7 proteins in the A549-IFITM3 cells (Fig. 7C). However, these blots also showed that while IFN treatment of the A549-Vector cells increased IFITM3 protein levels as expected, the amount of Rab7 and LAMP1 remained unchanged. We conclude that IFITM3 partially resides in the late endosomal and lysosomal compartments along with Rab7 and LAMP1, and that IFITM3 overexpression or IFN treatment expands these compartments through a mechanism that cannot be fully explained by increased protein expression alone. 10.1371/journal.ppat.1002337.g006 Figure 6 IFITM3 partially colocalizes with Rab7 and LAMP1, and compartments containing these proteins are amplified with IFITM3 overexpression. A) A549 cells stably transduced with either IFITM3 or with the empty vector alone, were incubated with LTRed (red) at 37°C, then fixed and immunostained for confocal imaging of IFITM3 (endogenous and overexpressed, gold), and LAMP1 (endogenous, green). DNA = blue. (Scale bars: 20 µM throughout). B) Percent colocalization of IFITM3, LTRed and LAMP in A549-Vector (blue) or IFITM3 (red) cells in (A). C) A549 cells stably transduced with either IFITM3 or with the empty vector alone were immunostained for confocal visualization of IFITM3 (endogenous and overexpressed, red) and Rab7 (endogenous, green). DNA = blue. D) Percent colocalization of IFITM3 and Rab7 in either the A549-Vector or A549-IFITM3 cells in (C). E) MDCK-Vector or MDCK-IFITM3 cells stained for exogenous IFITM3 (overexpressed, red) and LAMP1 (endogenous, green). DNA = blue. F) Percent colocalization of IFITM3 and LAMP1 in MDCK-Vector or MDCK-IFITM3 cells in (E). G) MDCK cells stably overexpressing Rab7-YFP and either IFITM3 (MDCK-IFITM3) or the empty vector control (MDCK-Vector) were immunostained and confocally imaged for IFITM3 (overexpressed, red) and Rab7-YFP (fluorescent signal from exogenous protein, green). Nuclear peripheries are represented by blue lines. H) Percent colocalization of Rab7-YFP and IFITM3 in either the MDCK-Vector or IFITM3 cells in (G). I) Enlarged view of images outlined by white boxes shown in (G), with MDCK-IFITM3 cells stably overexpressing both IFITM3 (red) and Rab7-YFP (green). 10.1371/journal.ppat.1002337.g007 Figure 7 IFN treatment or IFITM3 overexpression expands late endosomes and lysosomes. A549 cells stably expressing IFITM3 (IFITM3) or empty vector (Vector) were (A) left untreated (Buffer) or (B) treated with IFN-α, then fixed, permeabilized and immunostained for IFITM3 (endogenous and overexpressed, red), Rab7 (endogenous, gold), LAMP1 (endogenous, green), and for DNA (blue, merged image). Images were obtained using a confocal microscope. Similar results were observed with IFN-γ (data not shown). (Scale bars: 20 µm throughout). C) Whole-cell lysates from A549-IFITM3 or A549-Vector cells in (A) and (B) treated or untreated with IFN-α or γ were subjected to immunoblotting against the proteins indicated. GAPDH levels are provided to demonstrate comparable protein loading. Molecular weights in kDa are provided to the left. These images are representative of three independent experiments. D) A549 cells stably expressing Rab7-YFP (fluorescent signal from exogenous protein, green) were left untreated (Buffer) or treated with IFN-α, then fixed, permeabilized and immunostained for IFITM3 (endogenous, red) and imaged confocally. DNA = Blue. Similar results were obtained for IFN-γ (data not shown). E) Percent colocalization of IFITM3 and Rab7-YFP in the A549 cells in (D), with or without IFN-α treatment. IFITM3 overexpression leads to the expansion of the host cell's acidified compartments Our assays showed that incoming influenza A viruses were retained in the expanded acidic compartments of both the IFITM3 overexpressing cell lines as well as the IFN-γ-treated MEFs, and that IFITM3 partially localized to these structures (Fig. 2– 4, S2–4, S6). Therefore, we extended our investigation of these compartments. An increase in acidic structures was seen in MDCK and A549 cells overexpressing IFITM3 as compared to control cell lines, using either the vital acidophilic stain, acridine orange (AO), LTRed, or a cathepsin-L substrate that fluoresces only after it is proteolyzed, when compared to the corresponding vector control cells (Fig. 8A, B, a, b). Cathepsins are a family of lysosomal zymogens active in acidic environments (pH≤5.5) which are required for both the degradation of endocytic substrates and for the entry of several IFITM3-susceptible viruses [19]. Flow cytometry revealed an increase in the total LTRed fluorescent signal in both the MDCK and A549 IFITM3 cell lines when compared to controls (Fig. 8C). This expanded compartment represents a heterogeneous population of lysosomes and autolysosomes, based on confocal imaging showing the colocalization of the autophagosome marker, microtubule-associated protein 1 light chain 3 (LC3), with either LTRed or with CD63, with the latter being a resident of multivesicular bodies, amphisomes and autolysosomes (Fig. 8D, E). Furthermore, MDCK-IFITM3 cells stably transduced with an LC3 protein fused to both a red fluorescent protein (mCherry) and an enhanced green fluorescence protein (EGFP) showed a predominantly red signal, which occurs when the mCherry-EGFP-LC3 protein resides inside the acidified interior of an autolysosome (Fig. 8F, [43]). In keeping with previous reports that IFN-γ induces autophagy [44], [45], we detected enhanced LTRed staining in either IFN-γ treated MEFs or A549 cells (Fig. 4A, S10A). We conclude that increases in IFITM3 levels expand the lysosomal/autolysosomal compartment. 10.1371/journal.ppat.1002337.g008 Figure 8 IFITM3 overexpression results in the expansion of acidified organelles. A) MDCK or (B) A549 cell lines, stably overexpressing IFITM3 or the empty vector alone, were incubated with either the acidophilic dye acridine orange (AO), LTRed, or a flourogenic cathepsin-L substrate (Cath-L). All cells were also stained for DNA (blue). After incubation cells were imaged on a confocal microscope. Middle panels show enlarged images of the IFITM3 cells. (Scale bars: 20 µm throughout). C) Vector (blue) or IFITM3 (red) transduced cell lines, either MDCK (left) or A549 (right), were incubated with LTRed then analyzed by flow cytometry. D) A549 cells stably transduced with IFITM3 or with the vector alone were incubated with LTRed (red), then immunostained for confocal imaging of LC3 (endogenous, green). DNA = blue. E) MDCK cells stably transduced with IFITM3 or with the vector alone were immunostained for confocal imaging of LC3 (endogenous, red) and CD63 (endogenous, green). DNA = blue. F) Confocal images of MDCK cells overexpressing IFITM3 or the empty vector alone showing the distribution and fluorescence intensities of a stably expressed mCherry-EGFP-LC3B fusion protein using fluorescence channels that detect light emitted from the mCherry protein, EGFP or both (merge). DNA = blue. G) Model of IFITM3-mediated restriction of virus replication. Endocytosed viruses enter late endosomes where IFITM3 is present. IFITM3 prevents viral fusion within the endosomes and likely lysosomes via an unknown mechanism, perhaps by altering pH, membrane characteristics, lipid composition, transport speed or destination. Trapped viruses are trafficked to lysosomes and/or autolysosomes where they undergo degradation. Discussion Here we report several novel findings regarding the antiviral actions of IFN and the transmembrane IEG, IFITM3. First, this study demonstrates that IFN inhibits the nuclear translocation of vRNPs, and that IFITM3 is required for this IFN-mediated block, with both endogenous and overexpressed IFITM3 inhibiting vRNP nuclear entry. Second, either endogenous or overexpressed IFITM3, as well as IFN treatment, block the fusion of viral pseudoparticles expressing various influenza A virus envelope proteins (H1, H3, H5 and H7 subtypes of HA), or the VSV-G envelope protein; this block is specific because the fusion of pseudoparticles expressing MLV envelope is not inhibited by IFITM3. Third, our work reveals that IFITM3 partially resides with Rab7 in late endosomes, thus placing it in position to block influenza A virus' cytosolic access. Fourth, IFITM3 overexpression or IFN induce the expansion of late endosomal and lysosomal compartments containing Rab7 and LAMP1. Fifth, we show that similar to IFN-γ treatment, IFITM3 overexpression expands the number and size of autolysosomes, and it is into these compartments that trapped viruses are trafficked and subsequently degraded. Consistent with previous reports, our data show that high levels of IFITM3 do not prevent viral access to acidified compartments and that IFITM3 colocalizes with CD63 and LAMP1 [19]. This is in contrast to a report noting the exclusion of overexpressed IFITM3 from LAMP1-containing structures [33]. Therefore, this work adds substantially to our interpretation of previous reports by demonstrating that key downstream events in the viral lifecycle, fusion and vRNP nuclear translocation, are prevented by either IFN or IFITM3. IFITM3 thus represents a previously unappreciated class of anti-viral effector that permits viral entry into the endosomal compartment, but prevents egress into the cytosol. These studies also raise new questions including i) how do IFN and IFITM3 prevent viral fusion? ii) how do IFN and IFITM3 alter the endosomal and autolysosomal compartments? and iii) is the latter action required for viral restriction, or alternatively does it arise as an outcome of IFITM3's potential cellular role? Based on the substantial loss in IFN's potency observed when IFITM3 is depleted (50–80% loss of viral inhibition, Fig. S9A, B, [14]) we conclude that inhibition of viral emergence from the endosomal pathway is a prominent component of IFN's antagonism of influenza A virus replication in vitro. Our data also show that MxA cannot fully compensate for the loss of IFITM3 in IFN-treated cells challenged with influenza A virus. Recent work by Dittmann et al. [46] and Zimmermann et al. [47] reveal that human influenza A viral strains have evolved a means to evade MxA, suggesting a possible explanation for the cellular reliance on IFITM3 for protection in vitro. Similarly the IEG, IFIT1, prevents viral replication by targeting viral 5′ triphosphate-RNAs (PPP-RNA) for destruction [48], [49]. Given that IFITM3 is necessary for the majority of IFN-mediated restriction of influenza A virus in vitro, it may be that the virus has also evolved a means to at least partially nullify IFIT1, perhaps via the massive production of short “decoy” PPP-RNAs, as previously postulated [49], [50]. IFITM3 primarily resides in the endosomal compartment and partly colocalizes with Rab7 and LAMP1. IFITM3 overexpression or IFN stimulation caused the endocytosed viruses to accumulate in acidic compartments that contained both IFITM3 and LAMP1. Together with the BLAM-Vpr fusion assay data, these results reveal that IFITM3 prevents viral-host membrane fusion within late endosomes, and likely within lysosomes as well, in light of studies showing IFITM-mediated restriction of filoviruses and coronaviruses, which depend on cathepsin-mediated activation prior to fusion [19]. In doing so, IFITM3 traps the virus on a path which terminates in a degradative environment [51]. In support of this, our experiments show the eventual loss of a detectable vRNA signal in the LTRed-positive compartments of the IFITM3-transduced cells, thus revealing the fate of viral fitness under those conditions. These studies also reveal that elevated levels of IFITM3 correlate with the expansion of host cell structures containing Rab7 and LAMP1, and that IFITM3 was also present in these structures. In the MEF and A549 experiments, IFN produced increased Rab7 and LAMP1 immunostaining, in addition to an increase in acidic structures. At present, we cannot explain the increased Rab7 and LAMP1 signals seen after IFN stimulation or IFITM3 overexpression solely on the slight elevations in the abundance of these proteins detected by immunoblotting. Two possible explanations for the increased immunostaining observed, are that IFN stimulation induced these proteins to cluster together or alternatively unmasked sequestered epitopes; we find the latter possibility less likely since LAMP1 and Rab7 flourescent fusion proteins also showed larger and more intense signals under similar conditions. We envision that IFITM3-mediated clustering of organelles and their protein cargoes might contribute to the host cell's antiviral state. Earlier work reported no correlation between the size of the IFITM3-induced acidified compartments and the level of viral restriction [19], however, we observe that increasing levels of IFITM3 result in both an expansion of lysosomes/autolysosomes and increased viral inhibition. These observations might be explained by a common mechanism underlying the increase in these structures and viral inhibition, in addition to raising the possibility that they play a role in IFITM-mediated viral restriction. Is there a common characteristic shared by IFITM3-susceptible viruses? The late endosomal- and lysosomal-associated small GTPase, Rab7, is required for influenza A virus infection [7], [41]. The IFITM3-resistant viruses previously tested (MLV, the arena viruses and the hepacivirus, HCV) are all Rab7-independent, while the entry of the IFITM3-susceptible viruses (influenza A, dengue, Ebola, Marburg, and SARS) relies on Rab7 [14], [19], [41], [52], [53], [54]. Standing against this hypothesis, is the lack of effect on VSV-G-mediated entry with expression of a dominant negative Rab7 [41], [55], [56]). However, additional studies have shown that VSV-G-directed entry is dependent on transport to the late endosomes [57], [58]; these latter results, together with those of Huang et al. and Weidner et al. [19], [20], are consistent with the prediction that viruses that fuse in late endosomes or lysosomes are vulnerable to IFITM3's actions, while viruses whose genomes enter at the cell surface or in the early endosomes may avoid IFITM3's full effect. Of note, we have been unable to demonstrate that IFITM3 blocks HIV-1 replication using TZM-bl HeLa cells and are working to address these differences with a published study ([59], data not shown). This study, together with previous work, demonstrates that IFITM3 permits endocytosis of viruses, but prevents viral fusion and the subsequent entry of viral contents into the cytosol [19], [20]. While the BLAM-Vpr fusion assay demonstrates inhibition of fusion by IFN or by IFITM3, we note that this assay uses an indirect readout to assess entry of viral contents. Therefore several possibilities could explain the containment and neutralization of viruses within the endosomal pathway, including alterations in endosomal trafficking, acidification, or the host membrane's fusion characteristics (bending modulus, elasticity). While additional work is required to further define the mechanism, the lack of toxicity seen with cells stably overexpressing high levels of IFITM3 suggests that gross alterations in endogenous trafficking or pH control are unlikely (data not shown). Therefore overexpressing or activating IFITM3 to produce an enhanced antiviral state may be an effective prevention strategy during high risk periods in vulnerable populations. We propose that IFN causes the degradation of endocytosed viruses by preventing their contents from entering the host cytosol, and that IFITM3 is necessary and sufficient for this defense (Figure 8G). IFITM3's mode of defense could be envisioned as an effective means to neutralize pathogens during an organism-wide threat. Such actions might confer an advantage to the host because if IFITM3 simply decreased viral attachment and/or entry, the repulsed viruses would be free to attack neighboring cells. Of course while there are considerable differences between this simple scenario and the directed phagocytosis of pathogens by specialized immune cells, i.e. macrophages, the similarities none-the-less suggest an early prototype for a more evolved defense mechanism. Materials and Methods Cell lines and culture conditions U2OS, A549, MDCK, HeLa cells (all from ATCC), and chicken embryonic fibroblasts (ChEFs, from Charles River Labs) were grown in complete media (DMEM, Invitrogen Cat#11965) with 10% FBS (Invitrogen). WI-38 cells (ATCC) were cultured in DMEM (Invitrogen Cat#10569), containing non-essential amino acids (Invitrogen Cat#11140) and 15% FBS. Wild type and matched IfitmDel−/− MEFs were from adult IfitmDel+/− mice [26] that were intercrossed and MEFs derived from embryos at day 13.5 of gestation, as described previously [14]. The MEFs were genotyped by PCR and Western blot, and the generation of the IfitmDel−/− Ifitm3 cells have been previously described [14]. Plasmids The IFITM3 retroviral vector, pQCXIP-IFITM3 and empty vector control (Clontech) have been previously described [14]. The shRNA lentiviral vectors, pLK0.1-Scramble and pLK0.1-shIFITM3-3 (clone ID HsSH00196729) are available from the Dana Farber DNA core, Harvard Medical School, Boston, MA. pCAGGS-HA WSN/33 and pCAGGS-NA WSN/33 were kind gifts of Dr. Donna M. Tscherne and Dr. Adolpho Garcia-Sastre, Microbiology Dept., Mt. Sinai School of Medicine, NY, NY [38]. pBABE-mCherry-EGFP-LC3B was from Addgene (Plasmid #22418) and was kindly deposited by Jayanta Debnath. pLZS-Rab7-YFP and pLVX-RFP-LAMP1 were generously provided by Walther Mothes, Section of Microbial Pathogenesis, Yale University School of Medicine. The following shRNA sequences (sense strand sequence provided) were cloned into the pAPM shRNA-expression lentiviral vector [60], to create the viruses used to generate the A549 IFITM3 knockdown cell lines in Fig. S9: IFITM3-1: 5′-TCCTCATGACCATTCTGCTCAT-3′ IFITM3-2: 5′-CCCACGTACTCCAACTTCCATT-3′ IFITM3-3: 5′-TTTCTACAATGGCATTCAATAA-3′ Viral propagation and titration Influenza A virus A/Puerto Rico/8/1934 (H1N1) (PR8, Charles River Labs) and A/WSN/33 (H1N1) (kind gift of Dr. Peter Palese, Microbiology Dept., Mt. Sinai School of Medicine, NY, NY) were propagated and assessed for viral infectivity as previously described [14]. Influenza A virus A/Vietnam/1203/2004 (H5N1) was propagated and characterized as previously described [61]. Cytokines Human interferon (IFN)-γ (Invitrogen) was used at 100–300 ng/ml, human IFN-αA2 (PBL Interferon Source) was used at 500–2500 U/ml. Cells were incubated with cytokines for 16–24 h prior to IF or viral infection experiments unless otherwise noted. Murine IFN-γ (PBL Interferon Source) was used at 100–300 ng/ml. Western analysis Whole-cell extracts were prepared by cell lysis, equivalent protein content boiled in SDS sample buffer, resolved by SDS/PAGE, transferred to Immobilon–P membrane (Millipore), and probed with the indicated antibodies. Time course infection experiments and confocal microscopy Cells were seeded on glass coverslips for Influenza A virus infection experiments. Cells were incubated on ice with PR8 for 40 min. At time zero, the viral supernatant was removed and 37°C media was added with or without Lysotracker Red DND-99 (Invitrogen). At the indicated time points post-warming, cells were washed twice with D-PBS (Sigma) and incubated for 30 seconds with room temperature 0.25% trypsin (Invitrogen). The cells were then washed with complete media twice and fixed with 4% formalin (PFA, Sigma) in D-PBS. Image analysis for quantitation of vRNP nuclear translocation was done using Imaris 7.1 (bitplane scientific software). We generated a mask of the nucleus and applied this mask to the channel containing the viral signal (puncta) to determine vRNA puncta contained in each nucleus. Live cell imaging experiments Cells were incubated at 37°C and 5% CO2 for 60 min. with either Lysotracker Red DND-99 or acridine orange (ImmunoChemistry Technologies). Hoechst 33342 (DNA stain, Invitrogen) was incubated (1∶10,000) with the cells for the final 15 min. The Cathepsin L flourogenic substrate assay was performed as per the manufacturer's instructions (Cathepsin L -Magic Red, ImmunoChemistry Technologies). Cells were visualized live by confocal microscopy. Immunoflourescence protein Cells were fixed in 4% PFA in D-PBS, and then incubated sequentially in 0.25% Tween 20 (Sigma), then 1% BSA with 0.3 M glycine (Sigma), both in D-PBS. Primary and secondary antibodies are listed below. Slides were mounted in Vectashield with DAPI counterstain (Vector Labs). Slides were imaged using a Zeiss LSM 510, laser scanning inverted confocal microscope equipped with the following objectives: 40× Zeiss C-APOCHROMAT UV-Vis-IR water, 1.2NA, 63× Zeiss Plan-APOCHROMAT DIC oil, 1.4NA, and 100× Zeiss Plan-APOCHROMAT DIC oil, 1.46NA. Image analysis was performed using ZEN software (Zeiss). Laser intensity and detector sensitivity settings remained constant for all image acquisitions within a respective experiment. Nuclear outlines were generated using Metamorph software suite (Molecular Devices) using the Kirsch/Prewitt filter to define boundaries and then subtracting out the original binary images. Antibodies The following antibodies were used in this study for either Western blotting (WB) or immunoflourescence (IF), or both as indicated, along with their respective source and catalogue number: Primary antibodies: Actin (Sigma A5316, WB), CD63 (Developmental Studies Hybridoma Bank (DSHB) clone H5C6, IF), Fragilis (mouse Ifitm3) (Abcam ab15592, WB, IF), GAPDH (BD Biosciences 610340, WB), HA (Wistar collection, Coriell Institute, clone H18-S210, WC00029, IF), IFITM3 (Abgent AP1153a, WB, IF), IFITM3 (Abgent AP1153c, IF), LAMP1 ((DSHB) clone H4A3, WB, IF), LC3 (Nanotools Mab LC3-5F10, WB, IF), MX1 (Proteintech 13750-1-AP, WB, IF), NP (Millipore clone H16-L10-4R5 MAB8800, IF), RAB7 (Abcam 50533, WB, IF). Secondary antibodies for IF (all from Invitrogen): Alexa Fluor 488 and 647 (goat anti-rabbit and goat anti-mouse). The LAMP1 [H4A3] and CD63 [H5C6] antibodies were developed by J.T. August and J.E.K. Hildreth and were obtained from the DSHB and developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA. Immunoflourescence RNA These experiments employ the QuantiGene ViewRNA slide-based assay kit from Affymetrix (Cat #QV0096) with all components from that source unless noted. RNA was visualized following a modified manufacturer protocol; changes made include the omission of the ethanol dehydration step, and use of Vectashield mounting media. Post-fixation with 4% PFA, cells adherent on coverslips were incubated with 1× detergent solution or incubated in 0.25% PBS-Tween20. Cells were then incubated with Proteinase K. Next cells were incubated at 40°C in hybridization solution A containing a viewRNA probe set designed against either the negative stranded RNA NP genome (vRNA) of PR8 (Affymetrix VX1-99999-01 QG ViewRNA TYPE 1 Probe Set against NP Influenza A virus (A/PuertoRico/8/34(H1N1)) at 1∶100) or a probe set against the positive stranded NP mRNA. Cells were then incubated in hybridization preamplifiers (1∶100 in hybridization buffer B) at 40°C. Finally cells were incubated with labeled probes (1∶100 in hybridization buffer C), washed and imaged as above. All steps were followed by two D-PBS washes. BLAM-Vpr pseudoparticle fusion assays Pseudotyped lentiviral particles expressing the HA envelope were produced by plasmid transfection of HEK 293T cells with an HIV-1 genome plasmid derived from pBR43IeG-nef+ (NIH AIDS Research and Reference Reagent Program (Division of AIDS, NIAID, NIH, Cat#11349, from Dr. Frank Kirchhoff) modified with a deletion which abolishes expression of Env without disrupting the Rev-responsive element, pCAGGS-HA WSN/33, pCAGGS-NA WSN/33 and pMM310, which encodes a hybrid protein consisting of β-lactamase fused to the HIV accessory protein, Vpr (NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH (Cat#11444) from Dr. Michael Miller). pCG-VSV-G together with pBR43IeG-nef+ and pMM310 were transfected to produce VSV-G pseudotyped lentiviral particles. For the H5N1, H3N1, and H7N1 pseudoparticles, pCAGGS-HA5 (A/Thailand2(SP-33)/2004) pCAGGS-HA3 (A/Udorn/72), and pCAGGS-HA7 (A/FPV/Rostock/34) expression plasmids were co-transfected with the pCAGGS-NA WSN/33, pMM310, and the pBR43IeG-nef+ lentiviral backbone. Cultures for pseudoparticle fusion assays, including stably transduced MDCK cells and WI-38 fibroblasts, were plated in 24-well dishes with 90,000 cells per well at the beginning of each assay. At the time of assay, 0.5 mL of virus stock was added to cells and incubated for 2–3 h (depending on cell type) at 37°C. In experiments using bafilomycin A1 (Sigma), the inhibitor was added at 0.1 nM final concentration (low dose) at 37°C for 1 h prior to incubation with virus. After infection, viral media was then aspirated and replaced with complete DMEM containing CCF2-AM (Invitrogen) along with 1.7 µg/mL probenecid (Sigma). Cells were incubated in the dark for 1 h, followed by dissociation from the dish using Enzyme Free PBS-based Dissociation Buffer, and fixation in 2% PFA. Flow cytometry was conducted on a Becton Dickinson LSRII using 405 nm excitation from the violet laser, and measuring 450 nm emission in the Pacific Blue channel and 520 nm emission in the Pacific Orange channel. Data was analyzed using FACSDiva and FlowJo8.8.7. Sialic acid linkage expression studies A549 cells stably transduced to overexpress IFITM3 or with empty expression vector (pQCXIP, Clontech) were grown to ∼50% confluency, dissociated with trypsin-free EDTA-based dissociation buffer (Invitrogen) for 10 min. at 37°C. Cells were incubated at 4°C with FITC-conjugated Sambucus nigra lectin (SNA, Vector Labs #FL-1301) to detect (α-2,6) sialic acid linkages, and biotinylated Maackia amurensis lectin II (MAL, Vector Labs #B-1265) to detect (α-2,3) sialic acid linkages, followed by streptavidin-PE-Cy7 (Invitrogen). Cells were incubated with lectins individually and in combination, and the results of staining were indistinguishable. All cells were stained with violet cell-impermeable dye (Invitrogen #L34955), and cells were included in the analysis if viable by FSC/SSC and viability dye. Binding assay A549 cells transduced with IFITM3 or the empty vector pQXCIP were detached using Enzyme Free PBS-based Dissociation Buffer, and then washed in cold PBS extensively. Cells and virus (WSN/33) were pre-chilled on ice for 30 min. and mixed at a moi of 50 and incubated at 4°C for 1 h with rotation. Cells were washed extensively with ice cold PBS and then fixed using 4% PFA. The cells were then probed with anti-HA mouse monoclonal antibody (Wistar collection, Coriell Institute, clone H18-S210, WC00029, IF) for 1 h at room temperature, followed by anti-mouse AlexaFlour-488 conjugated antibody (Invitrogen) for 1 h with PBS washes in between, then analyzed by flow cytometry. Supporting Information Figure S1 IFITM3 overexpression does not alter the surface levels of (α-2,3) or (α-2,6) sialylated proteins. A549 cells stably transduced with IFITM3 or the empty vector were incubated with biotinylated Maackia Amurensis lectin II (MAL) to detect (α-2,3) sialic acid linkages, followed by streptavidin-PE-Cy7, as well as FITC-conjugated Sambucus Nigra lectin (SNA) to detect (α-2,6) sialic acid linkages. A) The percentage of IFITM3 or vector cells staining positive for both sialic acid linkages (upper right hand quadrant), compared to unstained controls. B) IFITM3 overexpressing and vector cells are compared with regard to each sialic acid linkage in the double-stained populations. (PDF) Click here for additional data file. Figure S2 IFITM3 arrests influenza A virus in acidic cytosolic inclusions preventing vRNP nuclear translocation. A) Normal diploid human fibroblasts (WI-38 cells) were stably transduced with retroviruses containing IFITM3 (WI-38 IFITM3) or (B) a non-targeting control shRNA (WI-38 shScramble). Cells were incubated with PR8 on ice, and then warm media containing LTRed (red) was added at time zero. Cells were fixed at 150 min. p.i. and stained for NP (green) and DNA, then analyzed by confocal microscopy. Image analysis software was used to define each cell's cytosolic (white lines) and nuclear peripheries (blue lines, based on DIC images and DNA staining, respectively). Images are representative of four independent experiments. (Scale bar: 12 µM). C) Quantitation of nuclear vRNP particles. The number of vRNP particles per nucleus of the WI-38 cell lines (with or without IFN treatment) at the indicated time points are shown. Values represent the mean +/− the SD of three independent experiments. D) Percent colocalization of vRNPs and LTRed compartments in WI-38 shScramble, shIFITM3 or IFITM3 expressing cells at the indicated times p.i. Values represent the mean +/− the SD of three independent experiments. E) Western blot of lysates from WI-38 cells probed with the indicated antibodies. shIFITM3-3 is referred to as shIFITM3 in the preceding figures and was selected for use based on its superior knockdown of the target protein. (PDF) Click here for additional data file. Figure S3 A549 cells overexpressing IFITM3 inhibit vRNP nuclear entry. A549 cells overexpressing the empty vector control (A) or IFITM3 (B) were incubated with PR8 on ice (moi 500). At time zero warm media was added along with LTRed (red). At the indicated times, cells were processed and stained for NP (green) and DNA (blue lines represent the nuclear periphery based on staining), then imaged using a confocal microscope. (Scale bar: 20 µM). These images are representative of three independent experiments. C) Whole cell lysates of A549 cells used in (A) and (B) treated with either buffer, IFN-α or IFN-γ, were subjected to immunoblotting using the indicated antibodies. (PDF) Click here for additional data file. Figure S4 IFITM3 overexpression halts H3N2 influenza A virus in acidic cytosolic inclusions prior to vRNP nuclear translocation. MDCK cells stably expressing (A) the empty vector control or (B) IFITM3 were incubated with Aichi H3N2 virus on ice, and then warm media was added at time zero along with LTRed (red). Cells were then fixed at the indicated times p.i. and stained for NP (green), and DNA (blue lines denote nuclear periphery), then imaged by confocal microscopy. Images are representative of three independent experiments. (Scale bar: 20 µm). B) Quantitation of nuclear vRNP particles. The number of vRNP particles per nucleus of the MDCK cell lines at the indicated time points are shown. Values represent the mean +/− the SD of three independent experiments. C) Percent colocalization of vRNP and LTRed compartments in MDCK-Vector and MDCK-IFITM3 cell lines at the indicated times p.i. (PDF) Click here for additional data file. Figure S5 vRNPs are retained in LAMP1-containing organelles in cells overexpressing IFITM3. A) MDCK-Vector or IFITM3 cells stably expressing a LAMP1-red fluorescence protein (LAMP1-RFP) were challenged with PR8 as in Fig. S4. Cells were immunostained for NP (green), stained for DNA, and then imaged confocally along with the collection of LAMP1-RFP fluorescence (orange). Images are representative of three independent experiments. Blue lines represent the nuclear margins based on DNA staining. (Scale bar: 20 µM). B) Quantitation of nuclear vRNP particles. The number of vRNP particles present per nucleus of the MDCK cell lines at the indicated time points are shown. Values represent the mean +/− the SD of three independent experiments. C) Percent colocalization of vRNP particles and LAMP1-RFP-containing compartments in MDCK-Vector and MDCK-IFITM3 cell lines at the indicated times p.i. Values represent the mean +/− the SD of three independent experiments. (PDF) Click here for additional data file. Figure S6 Ifitm3 expression rescues IFN-γ-mediated inhibition of vRNP nuclear translocation in Ifitm Del −/− MEFs. A) IfitmDel−/− MEFs stably overexpressing Ifitm3 (IfitmDel−/−Ifitm3), were left untreated (left panels, Buffer), or treated (right panels) with IFN-γ. The following day cells were incubated on ice with PR8 (moi 500). Cells were next incubated in warm media containing LTRed (0 min.). Cells were then fixed at the indicated times p.i., immunostained with anti-NP antibodies (green) and imaged by confocal microscopy. Image analysis software was used to define the nuclear boundaries (blue lines). Images are representative of three independent experiments. (Scale bar 12 γM). B) Percent colocalization of vRNP and LTRed compartments in the indicated MEF cell lines, with or without IFN-γ treatment, are shown for the indicated times p.i. C) Quantitation of nuclear vRNP particles. The number of vRNP particles per nucleus of the MEF cell lines, with or without IFN-γ treatment, at the indicated time points are shown. Values represent the mean +/− the SD of three independent experiments. D) Western blot of whole cell lysates from the indicated MEFs probed with anti-mouse Ifitm3 and using GAPDH as a loading control. (PDF) Click here for additional data file. Figure S7 Fusion of viral pseudoparticles expressing HA envelope subtypes, but not a MLV envelope, is decreased by IFITM3. IFITM3 inhibits the replication of infectious H5N1 virus. A) MDCK cells stably transduced with IFITM3 or empty vector were incubated with pseudoparticles expressing N1 and HA subtypes (H1N1pp, H3N1pp, or H5N1pp). Cells were then fixed and assayed for cleavage of CCF2 using flow cytometry. These results are representative of three independent experiments. B) Chicken embryonic fibroblasts (ChEF) cells stably expressing the empty vector control or IFITM3 were incubated with pseudoparticles expressing N1 and either of the two avian influenza A viral HA subtypes, H5 or H7, as in (A). These data are representative of three independent experiments. C) ChEF cells stably transduced with the empty vector control or overexpressing IFITM3, were infected with WSN/33 for 12 h then stained for HA protein (red) and DNA (blue). Average percent infection is given for three independent experiments +/− SD. 4× magnification. D) MDCK-Vector or MDCK-IFITM3 cells were incubated with pseudoparticles expressing the amphotropic MLV envelope protein (MLVpp) and then assayed for cleavage of CCF2 using flow cytometry. These results are representative of two independent experiments. E) Infectivity of HA-expressing pseudoparticles is decreased by IFITM3. MDCK-Vector or MDCK-IFITM3 cell lines were infected with the indicated pseudoparticles for 48 h. Cells were then immunostained for expression of HIV-1 p24 protein expressed from the integrated lentiviral genomes. Percent infection is provided. These results are representative of three independent experiments. 4× magnification. F) A549 cells were stably transduced with retroviruses containing IFITM3 or empty viral vector alone, then infected with A/Vietnam/1203/04 (H5N1) influenza A virus (VN/04). After 12 h, the cells were fixed and stained for viral NP expression (green) and for DNA (blue). Values given are percentage infected cells and are representative of two independent experiments. 4× magnification. G) Western blot of lysates from A549-IFITM3 or A549-Vector cell lines probed with the indicated antibodies. H) A549 cell lines were infected with increasing amounts of H5N1 VN/04. Twelve hours after infection the cells were immunostained for NP expression and scored for infection status. Values are representative of two independent experiments. (PDF) Click here for additional data file. Figure S8 IFITM3 is required for IFN's inhibition of HA-mediated fusion. A) HeLa cells were stably transduced with retroviruses containing either IFITM3, a shRNA against IFITM3 (shIFITM3), or a non-targeting control shRNA (shScr). Cells were left untreated (left panels), or treated with IFN-γ (right panels), then exposed for 2 h to H1N1pps (WSN/33) containing BLAM-Vpr. After incubation with the pseudoparticles, the cells were fixed and assayed for cleavage of CCF2 by flow cytometry. These results are representative of three independent experiments. B) The indicated HeLa cell lines were treated with IFN-γ for 24 h then infected with increasing amounts of WSN/33. After 12 h of infection the cells were stained for HA expression. These results are representative of three independent experiments. C) Western blot of the indicated HeLa cell line lysates probed with the indicated antibodies. (PDF) Click here for additional data file. Figure S9 A549 cells depleted of IFITM3 show increased susceptibility to influenza A virus infection. MxA is expressed and is IFN-inducible in A549 and WI-38 cells. A) A549 cells, stably transduced with retroviruses expressing IFITM3, a negative control shRNA against firefly luciferase (shLuc), or one of three shRNAs against IFITM3 (1, 2 or 3), were treated with buffer, IFN-α or IFN-γ for 24 h, then challenged with WSN/33. After 12 h of infection, the cells were fixed and immunostained for HA and stained for DNA. IF images were captured and the percentage of infected cells determined based on HA staining. Values represent the average of three independent experiments +/−SD. B) Western lysates of A549 cells from (A) probed with the indicated antibodies. Western lysates of (C) A549 cells or (D) WI-38 cells, treated with buffer, IFN-α or -γ, then probed with the indicated antibodies. E) Confocal images of WI-38 cells treated with buffer or IFN-α, then fixed, permeabilized and immunostained for either IFITM3 (endogenous, red), or MxA (endogenous, green), and for DNA (blue, scale bar: 20 µM). (PDF) Click here for additional data file. Figure S10 IFN treatment both expands Rab7- and IFITM3-containing structures, and increases the size and number of acidified organelles. A) Confocal images of WI-38 cells treated with buffer, IFN-α, or IFN-γ, and then immunostained for either IFITM3 (endogenous, red) or Rab7 (endogenous, green), and DNA (blue). Arrows denote larger structures staining for Rab7 and IFITM3 that were seen predominantly with IFN-γ treatment. (Scale bar: 20 µM). B) A549 cells treated with either buffer or IFN-γ, then incubated with LTRed before fixation and DNA staining (blue) followed by confocal imaging. Images in this figure are representative of three independent experiments. (PDF) Click here for additional data file.

0 comments Cited 211 times – based on 0 reviews      Review now

Bookmark

Record: found
Abstract: found
Article: found

Is Open Access

IFITM Proteins Restrict Viral Membrane Hemifusion

Kun Li, Ruben Markosyan, Yi-Min Zheng … (2013)

Introduction The interferon (IFN) system is the first line of host defenses against pathogen invasion, including viral infections. It protects by producing hundreds of IFN-stimulated genes (ISGs) that modulate diverse biological functions. A number of ISGs (such as PKR, RNase L, ISG 15, etc.) have been characterized and shown to suppress viral replication, the mechanisms of which are still poorly defined (reviewed in reference [1]). One exciting development in the last few years has been the discovery of some novel ISGs, also known as cellular restriction factors (such as APOBEC3G, Trim5α and Tetherin, etc.), which intrinsically block different steps of retroviral replication [2], [3], [4], [5], [6]. It is notable that many viruses, including retroviruses, have evolved to acquire a variety of strategies that evade IFN-mediated restrictions [7], [8], [9]. This type of intrinsic immunity is believed to play crucial roles in virus-host co-evolution and viral pathogenesis [7], [10]. The interferon-inducible transmembrane (IFITM) protein family belongs to a group of small ISGs (∼15 kD) that has recently been shown to block early stages of viral replication [11], [12]. Originally identified through RNAi genetic screening and shown to inhibit infections by influenza A virus (IAV), West Nile virus and Dengue virus, the IFITM proteins are now known to potently restrict entry and infections by a number of highly pathogenic viruses, including HIV-1, filovirus, and SARS coronavirus [12], [13], [14], [15], [16], [17], [18]. In humans, there are at least 4 functional members of IFITM proteins; IFITM1, 2 and 3 are expressed in a variety of human tissues and cell lines, IFITM5 is limited to the bone and is involved in mineralization [11]. All of these human IFITM proteins have been shown to restrict viral entry and infection, with IFITM3 being generally thought to be the most potent [12], [13], [14], [15], [17], [18]. A recent study demonstrated that the IFITM3 protein significantly restricts the morbidity and mortality associated with influenza, further underscoring the crucial role of IFITM3 in vivo [19]. Yount and colleagues recently showed that the mouse IFITM3 protein is not only palmitoylated but also ubiquitinated, and that these posttranslational modifications distinctly regulate the cellular localization of IFITM3 and its anti-influenza activities [20], [21]. While it has been suggested that viral membrane fusion may be blocked by IFITMs [12], [13], [15], direct evidence is still lacking and exactly how IFITM proteins restrict virus entry and infection is currently not known. Membrane fusion is an essential step for enveloped viruses to enter host cells and initiate infection, a process that is mediated by the viral fusion proteins present on the surface of virions [22]. To prevent premature activation, viral fusion proteins in the mature viral particles are normally metastable and exist at a high-energy state. Once triggered by specific cellular stimuli, such as receptor binding, a low pH, or both, they undergo a series of conformational changes, resulting in the insertion of the fusion peptide of the viral fusion protein into the target cell membrane, leading to hemifusion, pore formation, expansion, and ultimately, complete fusion [23], [24], [25], [26]. While the general principle of viral membrane fusion has been extensively studied, the detailed molecular mechanisms governing this process are still poorly defined [22]. In particular, how viral membrane fusion is modulated by cellular factors other than the specific triggers (such as receptor binding, low pH, cathepsin cleavage, etc.) remains an emerging subject that needs to be explored. In this work, we sought to determine the mechanisms by which cellular IFITM proteins restrict viral membrane fusion and entry. We chose the Jaagsiekte sheep retrovirus (JSRV) envelope (Env) and IAV hemagglutinin (HA) proteins as the model system of study because of some of their advantages. JSRV is a simple retrovirus, with Env-mediated membrane fusion and entry requiring both receptor-binding and low pH; an initial receptor binding primes the subsequent low pH-dependent conformational changes required for full activation [27], [28], [29], [30]. This unconventional two-step triggering mechanism was originally discovered in the avian sarcoma leukosis virus (ASLV) [31], and has now been suggested to operate in other enveloped viruses, including HCV [32]. Compared to most pH-dependent viruses, JSRV has a relatively high pH threshold (∼pH 6.3) for fusion, the process of which likely occurs in a GPI-anchored-protein-enriched endosomal compartment (GEEC) or caveolae [27], [33], [34]. Thus, study of JSRV Env-mediated fusion should lead to new insights into the mechanism of action of the IFITM proteins. IAV is a prototype pH-dependent virus, the entry and infection of which has been shown to be significantly restricted by IFITM proteins, particularly IFITM3, both in vitro and in vivo [12], [13], [19]. In addition to JSRV Env and IAV HA, which belong to class I fusion proteins, we also explored the inhibitory effects of IFITM proteins on membrane fusion induced by the Semliki Forest virus (SFV) E1/E2 and vesicular stomatitis virus (VSV) G proteins, which represent class II and III viral fusion proteins, respectively [22]. Hence, the mechanisms uncovered from this study are likely applicable to other viral fusion proteins, and collectively provide critical new insight into our understanding the mechanism by which IFITM proteins restrict viral membrane fusion and entry. Results Overexpression of IFITM proteins differentially restricts JSRV, IAV, 10A1 MLV and VSV entry Prior studies focused on IFITM3, and have suggested that it mainly acts in late endosomes or lysosomes to restrict viruses that fuse at lower pH (∼pH 5.5) than is present in early endosomes [12], [13], [14], [18]. Here we examined if IFITM proteins also restrict entry of JSRV, whose Env-mediated membrane fusion readily occurs at pH 6.3 or even higher [27], [33]. We did so by creating several stable lines expressing human IFITM1, 2 or 3 and testing their effects on JSRV entry, along with that of several other viruses. We observed that all three IFITM proteins effectively inhibited the infections of MoMLV pseudovirions bearing either IAV HA/NA or VSV-G in HTX cells (a subclone of the HT1080 cell line), with approximately equivalent efficiency (Fig. 1A; p 0.05). shRNA did not significantly reduce IFITM2 in K562 cells, and thus we could not assess the consequences of reducing of this protein (data not shown). Overall, these results suggest that endogenous IFITM proteins intrinsically restrict JSRV, IAV and VSV entry. IFITM expression does not affect binding of JSRV Env to its Hyal2 receptor or receptor-mediated priming for fusion activation IFITM1 has been previously shown to be associated with caveolin-1, a protein that is known to play an essential role in caveolin-mediated endocytosis [37], [38]. Given that the JSRV receptor, hyaluronidase 2 (Hyal2), is a GPI-anchored protein that is localized in lipid rafts and that JSRV may use a GPI-anchored-protein-enriched endosomal compartment (GEEC) and/or a caveolar pathway for entry [33], [34], [39], [40], we considered the possibility that IFITM1 could preferentially interfere with the binding of JSRV Env to Hyal2, thereby restricting viral entry. We took advantage of a soluble form of the JSRV SU-human IgG fusion protein (JSU-hFc) we previously created, and performed an in vitro binding assay on HTX cells expressing individual IFITM proteins and functional human Hyal2 [27], [28], [29], [41]. Flow cytometry analysis revealed that the fluorescence shifts in HTX cells expressing IFITM proteins, including that of IFITM1, were similar to those of parental cells (Figs. 2A and B). This indicates that expression of IFITM proteins did not affect the binding of JSRV Env to HTX cells expressing the Hyal2 receptor. We also performed virus binding assays using Gag-YFP-expressing MoMLV (kind gifts of Walter Mothes) pseudoviral particles bearing JSRV Env [28], [42]; again, similar fluorescence intensities were observed among cells expressing IFITM proteins and parental cells (Figs. 2C and D). The expression of IFITM proteins on the surface of HTX cells was also examined by using an anti-FLAG antibody. IFITM1 had a relatively higher level of surface expression as compared to IFITM2 and 3, but overall the fluorescence signals were low and their differences were not statistically different (Figs. 2E and F; data not shown). We conclude that expression of IFITM proteins, including IFITM1, does not affect the binding of JSRV Env to its Hyal2 receptor on the cell surface. 10.1371/journal.ppat.1003124.g002 Figure 2 Expression of IFITM proteins does not affect binding of JSRV Env to cells expressing the Hyal2 receptor, nor does it perturb receptor-mediated priming for fusion activation. (A) Examination of JSRV Env binding to cells expressing IFITMs. HTX cells stably expressing indicated IFITM proteins were incubated with purified JSRV SU-human IgG Fc proteins at 4°C; following incubation with FITC conjugated anti-human Fc antibody, cells were analyzed by flow cytometry. Representative histograms from one typical experiment are shown. Arrow indicates the secondary antibody alone control. (B) Quantitative analysis of JSRV Env binding data shown in (A). The fluorescence intensities (geometric means) obtained from (A) were averaged and normalized to those of mock controls. The means ± SD of at least three independent experiments are shown. (C) Examination of the JSRV pseudovirion binding to cells expressing IFITMs. HTX cells expressing IFITM proteins were incubated with purified JSRV pseudovirions containing MLV Gag-YFP to allow virus binding. Cells were washed, fixed and analyzed by flow cytometry. “Control” indicates cells incubated with MLV Gag-YFP pseudovirions in the absence of JSRV Env. Representative flow cytometry profiles are shown. (D) Quantitative analysis of the JSRV pseudovirion binding experiments shown in (C). The fluorescence intensities of three independent experiments were averaged and plotted. (E) Expression of IFITM proteins on the surface of HTX cells. The expression was examined by an anti-FLAG antibody and analyzed by flow cytometry. (F) Quantitative analysis of IFITM expression on the cell surface shown in (E). Values are the means ± SD of at least five independent experiments. (G and H) Examination of the effect of IFITMs on JSRV SU shedding. 293T cells were co-transfected with plasmids encoding FLAG tagged-JSRV Env and FLAG-tagged IFITMs. Cells were metabolically labeled and chased in the presence of indicated amounts of sHyal2. Cell lysates and culture media were harvested and immunoprecipitated with anti-FLAG beads. Samples were resolved by SDS-PAGE and subjected to autoradiograthy. (G) Expression of JSRV Env and IFITM in transfected cells. Env: the full length of JSRV Env; SU: surface subunit; TM: transmembrane subunit. (H) Shedding of JSRV SU into culture medium. Note the increased SU shedding in cells expressing JSRV Env with increasing amounts of sHyal2; no significant differences in shedding among cells expressing IFITM and mock controls were observed. The relative intensities of signals for shed SU were calculated by setting the signals of the mock control without sHyal2 stimulation as 1.0; three independent experiments were used for the quantification. JSRV Env uses a dual triggering mechanism in which receptor binding primes the Env to undergo low pH-dependent conformational changes that lead to fusion [28], [29]. This unusual feature allowed us to examine if IFITM proteins may affect receptor-mediated priming for fusion. We performed metabolic labeling of 293T cells co-expressing IFITMs and JSRV Env, and determined shedding of JSRV SU in the presence or absence of a soluble form of Hyal2 (sHyal2). In our previous studies, we had established that shedding of JSRV SU into culture media is an important indicator of Hyal2 receptor-mediated triggering for the fusion activation of JSRV Env [28], [29], [30]. Here we observed that the levels of JSRV SU harvested from the culture media of 293T cells expressing IFITM1, 2 or 3 were comparable to those of parental cells, and that they all increased with the presence of sHyal2 in a dose-dependent manner (Fig. 2H). The total levels of JSRV Env expression in these radiolabeled cells were approximately equivalent, as evidenced by the intensities of Env precursors and processed TMs (Fig. 2G). Collectively, we conclude that overexpression of IFITM proteins, including IFITM1, does not affect the expression and trafficking of JSRV Env, nor does it impair receptor-mediated priming for fusion activation. IFITM1 profoundly inhibits syncytia formation induced by JSRV Env and IAV HA; the effect occurs over a broad range of pH Syncytia formation and cell-cell fusion assays have been instrumental in understanding membrane fusion, including viral membrane fusion [43]. We sought to obtain direct evidence that IFITM proteins may restrict viral membrane fusion mediated by JSRV Env and other viral fusion proteins. As JSRV Env requires Hyal2 overexpression for membrane fusion to be detected at low pH [27], we generated stable HTX and 293 cell lines overexpressing Hyal2 and IFITM1, 2 or 3, which served as target cells for the syncytia formation and cell-cell fusion assays described below. For parental 293 cells overexpressing Hyal2 (293/LH2SN, mock), we observed almost complete syncytia formation (∼100%) induced by JSRV Env and IAV HA within 5–10 min after a pH 5.0 pulse (Fig. 3A). In sharp contrast, very little syncytia formation was detected in 293/LH2SN cells expressing IFITM1, even after a 1 h recovery period (Fig. 3A). Syncytia formation was also substantially reduced in 293/LH2SN cells expressing IFITM2, but the reduction was much less in cells expressing IFITM3, especially in the case of JSRV Env (Fig. 3A). There is much less or no inhibitory effect of these IFITM proteins on entry of 10A1 MLV (a virus that fuses at neutral pH, Fig. 2B) [12], [13]. We therefore measured syncytia formation induced by 10A1 MLV Env (with its R peptide deleted) at neutral pH and found, as predicted, that it was not significantly affected by IFITMs (Fig. 3A). The fusion efficiency of 10A1 MLV Env in IFITM1, 2 and 3-expressing cells, as quantified using fusion index (0.51±0.06, 0.50±0.05 and 0.52±0.04, respectively), was comparable to that in parental cells (0.52±0.06). A similar order of syncytia inhibition by the IFITMs, i.e., IFITM1>IFITM2>IFITM3, on JSRV Env and IAV HA was also obtained in cells expressing WT IFITM proteins (without the N-terminal FLAG tags) (Fig. S3). The differential inhibitory effects of IFITMs on syncytia formation of JSRV Env and IAV HA in 293/LH2SN cells were unlikely due to their levels of IFITM expression, which were examined by Western blots (Fig. 3B). Flow cytometry analysis of 293T cells co-expressing JSRV Env and WT IFITM proteins showed that the levels of JSRV Env on the surface of IFITM-expressing cells were comparable to that of the mock control (Fig. 3C), indicating that the reduced syncytia formation was not due to a change in the Env surface expression. 10.1371/journal.ppat.1003124.g003 Figure 3 Expression of IFITM proteins or treatment of cells with IFN suppresses syncytia formation induced by JSRV Env and IAV HA. (A) 293/LH2SN cells stably expressing the indicated IFITM proteins were transfected with plasmid DNA encoding JSRV Env, IAV HA or 10A1 MLV Env with the R peptide deleted (10A1 Env R−); cells were treated with a pH 5.0 buffer for 1 min (for JSRV Env and IAV HA) or left untreated (for 10A1 MLV Env) and analyzed for syncytium formation using fluorescence microscopy. Note the stronger inhibition of IFITM1 relative to that of IFITM2 and 3 for JSRV Env and IAV HA; no apparent inhibition was observed for 10A1 MLV Env. (B) Expression of IFITM proteins in 293/LH2SN cells was determined by immunoblotting with an anti-FLAG antibody. β-actin was used as a loading control. (C) Expression of IFITM proteins does not downregulate the JSRV Env expression on the cell surface. 293T cells were co-transfected with plasmid DNAs encoding FLAG-tagged JSRV Env (at both the N- and C-termini) plus indicated wildtype IFITMs. Cells were incubated with an anti-FLAG antibody on ice, and the surface expression of JSRV SU was determined by flow cytometry. A second antibody alone was used control. (D) 293/LH2SN cells were transfected with plasmids encoding JSRV Env, IAV HA, or 10A1 MLV Env with the R peptide deleted; 6 h after transfection, cells were treated with indicated amounts of IFN-α2b or medium for 24 h. Cells were exposed to a pH 5.0 buffer for 1 min (for JSRV Env and IAV HA) or left untreated and examined for syncytia formation. To evaluate if the differential effects of IFITM proteins on syncytia formation induced by JSRV Env were dependent over a limited pH range, we treated the JSRV Env-expressing cells with different pH values, i.e., pH 6.2, 5.7 and 5.0, respectively. Under these pH conditions, IFITM1 consistently exhibited the strongest inhibition on syncytia formation induced by JSRV Env (Fig. S4). Further lowering the pH (pH 4.0) or incubating 293 cells with an increased concentration of sHyal2 (up to 30 µg/ml) did not overcome the IFITM1-mediated restriction on fusion (data not shown), suggesting that the block by IFITM1 does not occur at the triggering step. It has been previously established for influenza HA that progressively lowering pH causes the activation of more fusion proteins; rather than causing each individual protein to undergo increasingly extensive conformational changes [44], [45]. Based on the results of JSRV Env described here, we conclude that the mechanism of IFITM inhibition is independent of fusion protein density. We also assessed if treatment of cells with IFN could suppress viral membrane fusion. We observed that, following a 24-h incubation of 293/LH2SN cells with IFN-α2b, syncytia formation induced by JSRV Env or IAV HA was greatly reduced in a dose-dependent manner (Fig. 3D). In contrast, 10A1 MLV Env-mediated syncytia formation was not significantly affected by the IFN-α2b treatment (Fig. 3D). No cytotoxicity was observed during the 24-h IFN treatment period; nor were there any changes in the expression of JSRV Env and Hyal2, as examined by immunoblotting and flow cytometry (data not shown). Since 293 cells do not express significant amounts of IFITMs, especially IFITM1 and 3 (Fig. S2), we have been unable to unambiguously determine if depletion of individual endogenous IFITMs by shRNA in 293 cells enhances syncytia formation. Nonetheless, these experiments clearly demonstrated that IFN can block viral membrane fusion. Cell-cell fusion induced by JSRV Env is inhibited by IFITM1; inhibition is the same for expression in target as in effector cells We applied a more quantitative cell-cell fusion assay to evaluate the effects of IFITM proteins on JSRV Env-mediated fusion as well as to understand the possible mechanisms for inhibition. We expressed JSRV Env in 293T effector cells stably expressing GFP, and labeled HTX/LH2SN target cells expressing IFITM proteins with a red-fluorescent dye, CMTMR; cell-cell fusion was measured by a fluorescence microscope and flow cytometry [27]. Similar to the syncytia formation results (Fig. 3A), IFITM1 exhibited the strongest inhibition, reducing the cell-cell fusion efficiency of JSRV Env by ∼50% (Figs. 4A, B and C; p 0.05). (F) Low pH does not overcome IFITM1-mediated block of JSRV entry. HTX or HTX cells expressing IFITM1 were pretreated with 20 nM BafA1 (middle and right columns) or with 0.01% DMSO (left columns, as controls) for 2 h and then spininoculated with JSRV pseudovirions at 4°C for 1 h. Cells were washed with cold PBS to remove unbound viruses before incubation at 37°C for 1 h to allow endocytosis (see references 33 and 34). Cells were then treated with either a pH 7.5 or pH 5.0 buffer at 37°C for 5–10 min, followed by an additional incubation with 0.01% DMSO or 20 nM BafA1 for 4 h to allow infection. Three days after infection, viral titers were determined by counting AP-positive foci, and relative infectivity was calculated by normalizing all titers relative to those in parental HTX cells treated with DMSO. Values are the means ± standard deviations of three independent experiments. We explored whether IFITM proteins expressed in effector cells co-expressing the viral fusion proteins could also restrict cell-cell fusion. We expressed IFITM1 in 293T/GFP effector cells, in HTX/LH2SN target cells (the stable cell lines described above), or in both, and compared their effects on JSRV Env-induced cell-cell fusion. Expressing IFITM1 in effector cells (Fig. 4D, column 3; Fig. 4E) was as effective in reducing cell-cell fusion by JSRV Env as expressing IFITM1 in target cells (Fig. 4D, column 2; Fig. 4E), and co-expressing IFITM1 in both effector and target cells enhanced this inhibitory effect (Fig. 4D, column 4; compare columns 2 and 3 with column 4; p<0.05 in both cases; Fig. 4E). Thus, the inhibition of cell-cell fusion by IFITM1 is not specifically related to its expression in effector or target cells. This result suggests that IFITM proteins are unlikely to suppress cell-cell fusion by directly acting on specific viral fusion proteins or their corresponding receptors, but rather through a common physical mechanism(s) (see below). A critical question is whether or not the cell-cell fusion assay employed here is relevant to endosomal fusion; this is particularly important, given that IFITM proteins have been shown to predominantly restrict viruses that require low pH for membrane fusion and entry. To address this question, we took advantage of our previous finding that JSRV pseudovirions are virtually resistant to low pH inactivation and that an extracellular low pH pulse can overcome proton pump inhibitor–mediated block of JSRV entry [27], [33], [34]. We pretreated JSRV pseudovirion-bound HTX or HTX/IFITM1 cells with 20 nM bafilomycin A1 (BafA1), followed by a pH-5.0 pulse for 5–10 min; cells were then allowed for infection for 4 h in the presence of BafA1. We observed that, while low pH substantially rescued the BafA1-mediated block in both the parental HTX and HTX/IFITM1 cells, as would be expected [33], [34], the low pH treatment did not increase the JSRV titer in HTX/IFITM1 cells to a level that was similar to that of parental HTX cells (Fig. 4F). This result apparently differed from that of SARS coronavirus, whose fusion inhibition by IFITMs had been previously shown to be bypassed by trypsin [13]. Overall, our data suggest that IFITM1 expressed on the plasma membrane effectively blocks the forced entry of JSRV rendered by the low pH pulse, and this is consistent with the cell-cell fusion data. IFITM proteins suppress cell-cell fusion induced by all three classes of viral fusion proteins; the inhibition efficiency can be cell type dependent In order to assess possible broad effects of IFITMs on viral membrane fusion, we employed another cell-cell fusion assay for examination of IAV HA, SFV E1/E2, and VSV-G, which represent class I, II and III fusion proteins, respectively [22]. In these experiments, effector cells were the NIH 3T3-derived HAB2 cell line stably expressing IAV HA (kind gift of Judy White) [46] or COS7 transiently transfected with plasmids encoding SFV E1/E2 or VSV-G; we labeled these cells with calcein-AM. We loaded the target 293/LH2SN cells expressing individual IFITM proteins (the same cell lines as used for the syncytia formation assay shown in Fig. 3) with CMAC, allowing aqueous dye transfer during fusion to be monitored by a fluorescence microscope. We found that, somewhat surprisingly, both IFITM1 and 3 strongly inhibited membrane fusion induced by all three classes of viral fusion proteins, and their efficiencies were almost the same. IFITM2 also inhibited viral membrane fusion, but the efficiency was generally low (Figs. 5A, B and C), consistent with the results of syncytia formation (Fig. 3A) and the GFP-transfer cell-cell fusion assay described above (Fig. 4). Even more surprisingly, we observed that when the JSRV Env protein was expressed in COS7 cells as effector, Env-mediated fusion was also markedly suppressed by IFITM3 (Fig. 5D). This was in sharp contrast to the situation in which IFITM3 had essentially no effect on fusion when 293T cells were used as effector cells to express JSRV Env (Fig. 5E). 10.1371/journal.ppat.1003124.g005 Figure 5 IFITM proteins inhibit cell-cell fusion induced by representatives of three classes of viral fusion proteins. (A to E) Effector cells expressing indicated viral fusion proteins were loaded with calcein-AM, and were bound to target 293/LH2SN cells (Mock) or to cells expressing indicated IFITM proteins that were prelabeled by CMAC. pH was then lowered to 5.0 for JSRV Env, 4.8 for IAV HA, 5.7 for VSV G, and 5.4 for SFV E1/E2. Following reneutralization of cells to 7.2, fusion between pairs of effector and target cells was scored under fluorescence microscopy. (A) IAV HA (a class I fusion protein). (B) VSV G (class III). (C) SFV E1/E2 (class II). (D) JSRV Env (class I), with COS7 as effector cells. (E) JSRV Env, with 293T cells as effector cells. Note distinct effects of IFITM3 on JSRV fusion shown in (D) and (E). (F) Restriction of JSRV and IAV entry by IFITM proteins in COS7/LH2SN cells. COS7/LH2SN cells (Mock) or derivatives expressing indicated IFITM proteins were infected with GFP-encoding MoMLV pseudovirions bearing JSRV Env or IAV HA/NA, and viral infectivity was determined by flow cytometry as described in Fig. 1. Note that IFITM3 inhibited JSRV entry as effectively as did the IFITM1; representative flow cytometry profiles are shown in Fig. S4A. Perhaps COS7 cells express a specific cellular factor(s) that functionally promotes the inhibitory effect of human IFITM3 on viral membrane fusion. Consistent with this notion, COS7 cells expressing human IFITM3 (also engineered to co-express human Hyal2, because COS7 cells are not permissive to JSRV infection) drastically suppressed JSRV entry, with efficiency almost equivalent to that of IFITM1 (Figs. 5F and S5A). This observation was in sharp contrast to the situations in HTX and 293 cells, where IFITM3 had much less effect (Figs. 1A and B). Similarly, the syncytia forming activity of JSRV Env, as well as that of IAV HA, was almost equally inhibited by human IFITM3 and IFITM1 in COS7 cells (Fig. S5B) as compared to that in 293/LH2SN cells (Fig. 3A). This cell type-dependent effect on cell-cell fusion mirrored prior reports showing that the IFITM-mediated restriction on viral entry and infection is also cell type dependent [13], [15]. Collectively, our data demonstrated that these three human IFITM proteins potently suppress cell-cell fusion induced by all three classes of viral fusion proteins. Chlorpromazine (CPZ) does not overcome the block of cell-cell fusion induced by IFITM proteins We next examined which steps of the viral membrane fusion process are inhibited by IFITM proteins. For this purpose, we used conditions that, in the absence of IFITM proteins, allow fusion to proceed up to and through the point of hemifusion while preventing the steps that lead to pore formation. We did so by creating an intermediate of fusion, referred to as a cold arrested state (CAS). It has been previously shown that for viral fusion proteins that induce fusion at low pH, lowering pH at the low temperature of 4°C yields hemifusion; raising temperature at neutral pH then leads to pore formation and growth [47], [48], [49]. We created CAS for JSRV Env and IAV HA, and found, as expected, that membrane fusion did not occur (Figs. 6A and B, middle panels). Adding CPZ at neutral pH and low temperature to the parental cells (mock) resulted in a significant aqueous dye transfer (Figs. 6A, C, and D). Any further dye transfer upon subsequently raising of temperature to 37°C was not statistically significant (Figs. 6A, C and D). For target cells expressing IFITM proteins, aqueous dye spread upon CPZ addition at CAS was much less than for the parental cells (Figs. 6B, C and D). In contrast, raising the temperature led to a significant increase in dye spread; but the total dye spread was still less than for the parental cells (Figs. 6B, C and D). The fact that the amount of dye spread induced by CPZ was relatively small showed that little hemifusion occurred for the IFITM-expressing cells. The greater amount of dye spread upon raising temperature indicated that the IFITM proteins in the target cell membrane block the creation of hemifusion. Temperature dependence of protein conformational changes is generally greater than for those of lipids that are not near a phase transition. We therefore assume that the presence of IFITM proteins blocked the conformations changes in JSRV Env and IAV HA needed for hemifusion. But it is possible that the decrease in membrane fluidity caused by IFITM proteins (see below) inhibits lipids from rearranging into a hemifusion configurations. 10.1371/journal.ppat.1003124.g006 Figure 6 CPZ does not rescue the restriction of IFITMs on cell-cell fusion of viral fusion proteins. COS7 cells expressing JSRV Env or HAB2 cells expressing IAV HA (images not shown) were loaded with calcein-AM (green) and bound to target cells (unlabeled), either parental 293/LH2SN (Mock) or derivatives expressing IFITM1 (IFITM1). Cells were treated with a pH 5.0 buffer at 4°C for 1 min to create a cold arrested state (CAS), at which aqueous dye had not transferred. Cells were then switched to 37°C or treated with CPZ, cell-cell fusion were monitored under a fluorescence microscope. (A) In mock cells (JSRV Env-mediated fusion): Upon raising temperature from the 4°C of CAS to 37°C, the two target cells of the image became labeled by calcein-AM (arrows), illustrating that fusion was now extensive. Similarly, adding CPZ to cells at CAS also led two target cells receiving calcein-AM, illustrating that fusion was as extensive upon addition of CPZ as upon raising temperature. (B) In IFITM1-expressing cells (JSRV Env-mediated fusion): Raising temperature led to calcein transfer to only one (arrow) of the four target cells. Addition of CPZ did not lead to calcein-AM transfer to any of the three target cells. (C–D) The quantifications of these phenomena are presented in JSRV Env (C) and for IAV HA (D). Similar experimental procedures were applied to IFITM2 and 3-expressing cells, and the data were plotted as show in (C) and (D). The negative curvature-promoting lipid, oleic acid (OA), effectively rescues IFITM-mediated suppression of viral membrane fusion Hemifusion is promoted by negative spontaneous curvature of monolayers that contact each other in binding [50], [51]. Therefore, making the spontaneous curvature more negative should oppose the inhibitory actions of IFITMs and thereby promote hemifusion. We tested this by adding oleic acid (OA) to aqueous solutions for 15 min to allow them to incorporate into plasma membranes. We observed that creating CAS in the presence of OA led to more fusion both when adding CPZ (Fig. 7A, compare second columns with first columns in each cell line) and raising temperature (Fig. 7B, compare second columns with first columns in each cell line). For the parental target cells, the addition of OA resulted in fusion between almost all cell pairs (Figs. 7A and B). For target cells expressing IFITM proteins, OA-addition led to a great increase in the percentage of cells pairs that fused (Figs. 7A and B). In sharp contrast, incorporating OA into membranes subsequent to creating CAS was much less effective in promoting fusion upon either the addition of CPZ or the raising of the temperature (Figs. 7A and B, compare third columns with second columns in each cell line). The almost complete rescue of IFITM-mediated restriction on hemifusion by OA further supports the notion that the major block in fusion caused by ITIFM proteins in target membranes occurs upstream of hemifusion. The presence of IFITMs likely blocks hemifusion by making the spontaneous curvature of outer leaflets of plasma membranes more positive. 10.1371/journal.ppat.1003124.g007 Figure 7 Making spontaneous curvature more negative prior to creating CAS promotes JSRV Env-mediated hemifusion. CAS was created as described in Fig. 6. OA was added prior or subsequent to CAS (see details in Materials and Methods). The addition of OA before creating CAS (middle columns of groups of three) promoted aqueous dye transfer upon either addition of CPZ (A) or raising temperature to 37°C from CAS (B). In contrast, adding OA after creating CAS (third columns), did not affect the extents of dye transfer caused by either CPZ addition or by raising temperature as compared to control (first columns). This was the general pattern, independent of whether target cells contained indicated IFITM proteins or not (Mock). Overexpression of IFITM proteins results in increased lipid order of cell membranes In principle, IFITM proteins could alter spontaneous monolayer curvatures and thereby restrict the creation of hemifusion by influencing the membrane molecular order and membrane fluidity. To explore these possibilities, we labeled 293/LH2SN cells expressing IFITM1, 2 or 3 or mock control with Laurdan, a hydrophobic fluorescent probe that is highly sensitive to lipid phases, and measured their generalized polarization (GP) values and florescence lifetimes using 2-photon laser scanning and fluorescence-lifetime imaging microscopy (FLIM) [52], [53]. Because of its large excited state dipole moment to align surrounding water molecule in the energy dissipation process, Laurdan has been commonly used to report the extent of water penetration into the lipid bilayer, which correlates with lipid packing [52], [54]. In general, higher GP values and longer lifetimes indicate that the membranes are more molecularly ordered, while lower GP values and shorter lifetimes mark membranes as less molecularly ordered [52]. In mock control cells, the GP distribution was characterized by two peaks, one with a lower GP, associated with intracellular membranes, and another with a higher GP, identified with plasma membranes; the less ordered populations were predominant in the mock controls (Figs. 8A–D; first row). In cells treated with methyl-β-cyclodextrin (MβCD), a cholesterol-depleting reagent known to make the membrane less ordered [55], we observed dramatically reduced GP signals, resulting in an almost complete loss of the higher GP peak in the histogram (Figs. 8A–D; second row). In sharp contrast, cells expressing IFITM1, 2 or 3 all exhibited marked increases in the GP value, which was particularly evident in the higher GP peak (Fig. 8A–D; third, fourth and fifth rows), suggesting that these cell membranes are more ordered than those of mock controls. 10.1371/journal.ppat.1003124.g008 Figure 8 Expression of IFITM proteins increases the lipid order of cell membranes. 293/LH2SN cells stably expressing IFITM1, 2 or 3, or mock controls were incubated with 1.8 µM Laurdan for 40 min at 37°C, and were imaged using two-photon fluorescence microscope. Parental 293/LH2SN cells were treated with 10 mM MβCD for 1 h to serve as controls. (A) Fluorescence intensity images. The fluorescence signal was acquired from 416 nm to 474 nm for the blue channel and from 474 nm to 532 nm for the green channel. (B) Generalized polarization (GP) images. The GP image was generated according to the GP function which is a normalized ratio between the blue and the green channels (GP scale from −1 to +1). According to the calculated GP values, we restricted the GP scale from −0.3 to 0.5. (C) GP histograms. The GP histogram was fitted using two Gaussian distributions. The lower GP values are associated with internal membranes, and the higher GP values are associated with plasma membranes. (D) Pseudo-colored GP images. The lower and higher GP distributions were pseudo-colored in green and red, respectively. (E) Averaged GP values. The GP values of individual cell lines were averaged and plotted; for each cell lines, a total of 12–18 images were used for statistical analysis. Significant differences were observed between mock control and IFITM1 (p = 0.00672), IFITM3 (p = 0.00107) in the plasma membrane. See text for details. Quantitative analysis showed that the averaged GPs values of IFITM1 and 3 in the plasma membranes were significantly different from that of mock controls (p = 0.00672 and p = 0.00107, respectively) (Fig. 8E). The averaged GP values for the intracellular membranes of IFITM1 and 3-epxressing cells were also increased, albeit not statistically significant from those of mock controls (p = 0.08∼0.37) (Fig. 8E). Unfortunately, we have been unable to distinguish the lipid order of endosomal membranes from that of total intracellular membranes in this analysis. Noticeably, IFITM2 exhibited modest increases in the GP value (p = 0.1388) (Fig. 8E), which correlated its less inhibitory effect on syncytia formation and cell-cell fusion (Figs. 3, 4 and 5). The increased lipid order in cells expressing IFITM proteins was also evidenced by their longer FLIM lifetimes as compared to those of mock controls and MβCD-treated cells (Fig. S6). Collectively, these results showed that overexpression of IFITM proteins dramatically increases the lipid packing order of cell membranes and makes them less fluid and possibly less competent for membrane fusion (see below the Discussion). Discussion The IFITM protein family is the first and thus far only restriction factor known to block viral entry [12], [56]. Previous studies have suggested that these proteins, particularly IFITM3, predominantly restrict viruses that fuse in the late endosomal or lysosomal compartments at a lower pH [12], [13], [14], [17], [18]. Here we provide evidence that this family of proteins can also effectively restrict viruses that fuse with a higher pH threshold, such as JSRV (∼pH 6.3) which requires both receptor binding and low pH to co-trigger membrane fusion activation. This activation process likely occurs in the GPI-anchored-protein-enriched endosomal compartment or caveolin-associated compartments [27], [29], [33]. Interestingly, we found that, among the three human IFITM proteins examined, IFITM1 was more active than IFITM2 and 3 in restricting JSRV Entry (Figs. 1A and B). This was not because of downregulation of JSRV Env or its Hyal2 receptor, nor due to a perturbation of receptor-mediated priming for fusion activation (Fig. 2). Instead, results of syncytia formation and cell-cell fusion experiments showed that JSRV Env-mediated fusion at low pH was profoundly inhibited by IFITM1 (Figs. 3, 4 and 5). Given that an extracellular low pH pulse cannot bypass the IFITM1-mediated inhibition of endosomal entry of JSRV (Fig. 4F), and that IFITM proteins do not inhibit 10A1 MLV Env-mediated fusion at neutral pH (Fig. 3), we conclude that the syncytia formation and cell-cell fusion assays employed in this study can reflect the situation of viral membrane fusion in endosomes, which is believed to be the predominant site of IFITM-mediated inhibition of viral entry. Somewhat surprisingly, we observed that IFITM1 was also generally more effective than IFITM2 and 3 in suppressing syncytia formation induced by IAV HA (Fig. 3A) and VSV-G (not shown), even though these proteins restrict viral entry with almost comparable efficiency (Figs. 1A and B). The exact mechanism underlying these observations is currently unknown, but could be related, in part, to the relatively higher levels of IFITM1 expression on the cell surface in 293 cells as compared to that of IFITM2 and 3 (Fig. 1E and F). Quantitative cell-cell fusion analysis confirmed the syncytia formation data (Fig. 5), but also revealed that IFITM3 dramatically inhibited viral membrane fusion when COS7, rather than 293T, was used as effector cells (Figs. 4, 5 and Fig. S5). These results indicate that the effects of IFITM proteins on viral membrane fusion can be cell type dependent, which agrees with previous observations on viral entry [13], [15]. Because all three human IFITM proteins tested exhibited potent restriction of viral membrane fusion induced by all three classes of viral fusion proteins that have different structures (Fig. 5), we suggest that a common physical mechanism, rather than specific interactions with viral fusion proteins, is responsible. Results from our series of experiments support this hypothesis. Our first line of evidence is the effect of CPZ on cell-cell fusion (Fig. 6). It has been previously established that CAS is a state of hemifusion, and that the addition of CPZ to cells at the CAS intermediate leads to full fusion [47], [57]. It is also known that for cells brought to CAS by fusion proteins that are triggered by acidic conditions, raising the temperature at neutral pH leads to full fusion [47], [48], [58]. We observed that the extent of aqueous dye spread in target cells expressing IFITM proteins was much less upon CPZ addition or upon raising the temperature from CAS than was fusion induced by lowering pH at 37°C (Fig. 6). These findings provide strong evidence that IFITM proteins inhibit the creation of hemifusion. Because the effects caused by adding CPZ or raising the temperature were qualitatively similar for JSRV Env and IAV HA (Figs. 6C and D), we conclude that the mechanism of inhibition by IFITM proteins is not dependent on the precise fusion protein. Although IFITM proteins may affect pore formation and/or expansion, their primary mechanism appears to be the prevention of hemifusion. The second line of evidence for the IFITM-mediated block on hemifusion came from the OA experiments (Fig. 7). It has been repeatedly shown that hemifusion is promoted by negative spontaneous curvature and is inhibited by positive spontaneous curvature [51], [57]. Consequently, if IFITM proteins conferred positive spontaneous curvature to membranes that contain them, these proteins would naturally block hemifusion. As OA has a large negative spontaneous curvature [57], we reasoned that it should overcome the inhibitory actions of IFITM proteins if the curvature was at the core of the action of IFITMs. Experimentally, we observed that the addition of OA virtually overcame all of the block of fusion by IFITM proteins (Figs. 7A and B). That is, when hemifusion was induced by the addition of OA, pore formation readily resulted without any inhibition despite the expression of IFITM proteins in the target membrane. The fact that the addition of OA after establishing CAS had no apparent effect on IFITM-mediated inhibition further supports the conclusion that the block occurs at steps prior to the creation of hemifusion (Figs. 7A and B). How can IFITM proteins block hemifusion? Our Laurdan labeling experiments showed that IFITM-expressing cell membranes were more ordered than those of mock controls, as evidenced by their increased GP values and longer FLIM lifetimes (Figs. 5, 8, and S6). The increase in the lipid order of IFITM-expressing cells, particularly in their plasma membranes, correlates with the potency of IFITMs in suppressing viral membrane fusion (Fig. 8 and Fig. S6). Thus, IFITM proteins may block hemifusion by decreasing the fluidity of the membrane that contains them: a decreased fluidity would reduce the ability of lipids to undergo movements necessary for achieving hemifusion. It is also possible that the increased exclusion of water from the bilayer, as indicated by the higher GP values in the presence of the proteins (Fig. 8) is due to an increased average area occupied by lipid headgroups relative to the area swept out by their acyl chains. This would be equivalent to a greater positive spontaneous curvature. It remains possible that expression of IFITM proteins alters the lipid composition of cell membranes, thereby influencing their fluidities and spontaneous curvatures. We emphasize that while changes in lipid order and membrane fluidity likely account for the general inhibitory effect of IFITMs on viral membrane fusion, they do not fully explain the virus-specific and somewhat cell type-dependent inhibition of IFITMs on viral entry as reported in this and previous studies (Figs. 1, 4 and 5; Fig. S5) [12], [13], [15]. Additional factors are likely to be involved, such as specific IFITM-binding partners and possibly viral elements that modulate IFITM-mediated inhibition of viral entry. In this respect, IFITM proteins may or may not influence cell-cell fusion mediated by developmental and cellular fusogens, depending on the specific cell systems that express IFITMs and cellular fusogens. The reason IFITM proteins promote greater lipid order remains unclear, but we offer a suggestion. IFITM proteins may directly change membrane curvature by adopting an unconventional membrane topology or topologies that function as a wedge to generate positive spontaneous curvature. There is an increased appreciation that lipid-binding proteins, along with lipids themselves, can influence membrane curvature, which has been shown to be crucial for vesicular trafficking and membrane fusion [59], [60]. Results of continuum membrane mechanics show that the spontaneous curvature of the monolayer of the target membrane proximal to the membrane expressing the fusion proteins (i.e., outer leaflets) affects hemifusion, but the spontaneous curvature of the distal monolayer (i.e., inner leaflets) does not [61]. Given our experimental data showing that IFITM proteins block hemifusion (Figs. 6 and 7), we suggest that IFITM proteins affect the outer monolayer, probably by spanning part or all of the outer monolayer. This suggestion is in line with the predicted membrane topology of IFITM proteins [11], in which N- and C-termini face the lumens of vesicles. It is also supported by prior and our current work showing that both the N and C-terminally tagged epitopes of IFITM3 can be, though not prominently, detected by flow cytometry and immunostaining without permeabilization (Figs. 2E and F) [12], [14]. A recent study suggested that the mouse IFITM3 protein adopt an alternate topology [20]. In this report, the authors provided evidence that the originally predicted transmembrane domains of IFITM3 fold into a hairpin loop and span only the inner leaflets, resulting in an intramembrane topology with both N- and C-termini facing the cytosol [20]. While our data does not unambiguously demonstrate the existence of this alternate topology, it is possible that IFITM proteins adopts multiple and dynamic topologies. In fact, it has been shown that some transmembrane proteins, including those of viral glycoproteins, adopt dual or dynamic topologies because of “lipid flip-flop” and/or changes in the net charge of their cytosolic sequences [62], [63], [64]. Dynamic topologies could result from the cleavages of IFITM proteins that have been observed at both the N- and C-termini in mammalian cells [11], [19], [65] (our unpublished data). It is therefore possible that IFITM proteins, including their orthologs in different species which differ significantly at the N- and C-termini [11], adopt distinct topologies in mammalian cells; these sequence and topologic differences may account for, and contribute to, their somewhat distinct phenotypes in suppressing viral membrane fusion and entry into host cells. Materials and Methods Cells 293T, 293, HTX (a subclone of HT1080), COS7, 293T/GFP (stably expressing GFP), HAB2 (expressing IAV HA, kind gift of Judy White, University of Virginia, Charlottesville, VA), 293/LH2SN (stably expressing Hyal2), HTX/LH2SN (stably expressing Hyal2), and 293/GP-LAPSN (expressing MLV Gag-Pol and alkaline phosphatase (AP)) cells have all been described previously [27], [40], [46], [66]. COS7 cells expressing human Hyal2 were generated by transduction with PT67/LH2SN retroviral vector encoding human Hyal2 [40]. 293, HTX, 293/LH2SN, COS7/LH2SN, HTX/LH2SN cells stably expressing IFITM proteins were generated by transduction with pQCXIP (Clontech, Mountain View, CA) retroviral vectors encoding IFITM1, 2 or 3 (see below). K562 cells stably expressing control shRNA or shRNA targeting IFITM1 or 3 mRNA were kind gifts of Michael Farzan and I-Chueh Huang (Harvard Medical School, Boston, MA). All mammalian cells used were grown in Dulbecco's modified Eagle's (DMEM) medium with 10% FBS (Hyclone, Logan, UT). Plasmids and reagents The human IFITM1, 2 and 3 genes, with or without an N-terminal FLAG tag, were amplified by PCR from pRetro-Tet-IFITM constructs [15]. PCR products were digested and ligated into the EcoRI/BamHI restriction sites of pQCXIP vector, resulting in pQCXIP-IFITMs. Retrovirus packaging plasmid encoding the MoMLV Gag-Pol (pCMV-gag-pol-MLV) and transfer vector encoding the GFP (pCMV-GFP-MLV) were kind gifts of Francois-Loic Cosset (INSERM U758-ENS, Lyon, France). Plasmids encoding JSRV Env with both N- and C-terminal FLAGs, the 10A1 amphotropic MLV Env, the vesicular stomatitis virus G protein (VSV-G) and SFV E1/E2 have been described previously [27], [29], [67], [68]. The 10A1 MLV Env construct with the R peptide deleted was created by removing the last 16 amino acid of the R peptide using PCR. Plasmids encoding the IAV HA and NA (Thailand KAN-1/2004 H5N1 strain) were kind gifts of Gary Nabel (NIH, Bethesda, MD). The MLV Gag-YFP construct was a kind gift of Walter Mothes (Yale University, New Heaven, CT). The anti-FLAG monoclonal antibody beads (EZview™-red), anti-FLAG antibody, anti-β-actin monoclonal antibody, anti-Tubulin, secondary anti-mouse immunoglobulin G conjugated to FITC, TRITC or HRP, chlorpromazine (CPZ), bafilomycin A1 (BafA1), and oleic acid (OA) were all purchased from Sigma (St. Louis, MO). Anti-IFITM1, anti-IFITM2 and IFITM3 were purchased from Proteintech Group (Chicago, IL). IFNα-2b, CMAC (7-Amino-4-Chloromethylcoumarin), calcein-AM, Methyl-β-cyclodextrin (MβCD), CMTMR (5-(and-6)-(((4-Chloromethyl)Benzoyl)Amino)Tetramethylrhodamine), and Lipofectamine 2000 were purchased from Invitrogen (Carlsbad, CA). The Express 35S-Met/Cys protein labeling mix was purchased from Perkin Elmer (Boston, MA). The JSRV SU-human IgG Fc fusion protein and sHyal2 were produced and purified as previously described [27], [41]. Pseudovirion production, transduction and infection The GFP- and AP-expressing MLV pseudovrions bearing JSRV Env, 10A1 MLV Env, IAV HA/NA and VSV-G were produced as previously described [33]. Target cells were infected with appropriate amounts of virus stock in the presence of 5 µg/ml Polybrene (Sigma), and assessed for GFP expression by flow cytometry 48 h after infection or for AP activity by staining of cells 72 h after infection. To test the effect of interferon (IFN) on viral entry, 293 cells were treated with 200–1000 units of IFN-α2b or medium alone for 24 h before pseudovirus infection. Typically, an MOI of 0.05 to 0.2 was used for all infections. To create cell lines stably expressing IFITM1, 2 or 3, we produced retroviral pseudotypes by transfecting 293T cells with plasmids encoding IFITMs (pQCXIP-IFITMs), MLV-Gag-Pol (pCMV-gag-pol-MLV) and VSV-G (pMD.G) using the calcium phosphate method. Supernatants were harvested 48–72 h post-transfection and centrifuged at 3,200 g to remove cell debris. Cells were infected with pseudovirions in the presence of 5 µg/ml Polybrene. Twenty-four hour after infection, cells were selected in growth medium containing 1 µg/ml puromycin (Sigma). For production of MLV Gag-YFP pseudovirions bearing JSRV Env, 293/GP-LAPSN cells were co-transfected with plasmids encoding Gag-YFP and JSRV Env by the calcium phosphate method. Virus binding assay MLV pseudovirions bearing JSRV Env and Gag-YFP were concentrated by centrifugation at 185,000 g on a 2 ml 20% sucrose cushion for 3 h, and were resuspended in phosphate-buffered saline (PBS). Cells were detached with PBS plus 5 mM EDTA, and incubated with different amounts of purified pseudovirions on ice for 3 h. The cells were washed with PBS for 5 times and fixed with 3.7% paraformaldehyde before being analyzed by using flow cytometry. Cell surface staining Cells were detached by PBS containing 5 mM EDTA and resuspended in PBS plus 2% FBS. To examine the binding of JSRV SU to cells expressing Hyal2, 5×105 HTX cells were incubated with 10 µg purified JSRV SU-human IgG Fc proteins on ice for 3 h, washed 3 times, and incubated with FITC conjugated anti-human IgG Fc antibody for another 1 h. Cells were then washed, fixed and analyzed by flow cytometry. For detection of the expression of the FLAG-tagged JSRV Env and IFITMs, similar procedures were used except that cells were incubated with an anti-FLAG antibody on ice for 1 h, followed by incubation with FITC conjugated anti-mouse IgG antibody for 1 h before cells were analyzed by flow cytometry. Metabolic labeling Metabolic labeling was performed as previously described [27]. Briefly, 293T cells were transiently transfected with plasmids encoding JSRV Env and/or IFITMs by the calcium phosphate method. Twenty-four hours post-transfection, cells were starved in DMEM without cysteine and methinonine (MP Biomedicals, Cost Mesa, CA) for 30 min and pulse-labeled with a 62.5 µCi mixture of cysteine plus methionine (Perkin Elmer, Waltham, WA) for 1 h, followed by chase-labeling in complete growth medium. To examine shedding of JSRV SU, 3 h after the chase period, indicated amounts of sHyal2 were added and cells were incubated for another 3 h. Cell lysates and culture media containing the 35S-labeled JSRV Env were harvested and immunoprecipitated with anti-FLAG beads. Samples were then resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) and applied to autoradiography. Band intensities were quantified using Quantity One (Bio-Rad, Hercules, CA). Syncytium induction assay Syncytia induction assays were performed as described previously [27]. Briefly, 293/LH2SN or COS7/LH2SN cells, either parental (mock) or derivatives stably expressing IFITMs, were seeded in 6-well plates and cotransfected with 2 µg plasmids encoding JSRV Env or 10A1 MLV Env with the R peptide deleted (10A1 Env-R−), or 0.5 µg plasmids encoding IAV HA, plus 0.5 µg peGFP-N1 (Clontech) using the calcium phosphate method. Twenty-four hours post-transfection, cells were treated with pre-warmed buffer (pH 6.2, pH 5.5, pH 5.0 etc.) for 1 min and were examined for syncytium formation under a fluorescence microscope. If applicable, cells were incubated with indicated doses of IFN-α2b for 24 h before being treated with a low pH buffer to induce syncytia. For 10A1 Env-R−, IFN-α2b was added upon transfection and was maintained throughout the entire fusion assay. The fusion index was calculated by using f = [1-(C/N)], where C is the number of cells per phase field after fusion, and N is the total number of nuclei [69]. At least five phase-contrast microscopy fields were used for the analysis, with means and standard deviations calculated. Cell-cell fusion assay Two cell-cell fusion assays were employed in this study. The first cell-cells fusion assay was used for JSRV Env as previously described [27]. In brief, 293T/GFP cells were transfected with 2 µg plasmids encoding JSRV Env alone or plus IFITM1 by Lipofectamine 2000. Twenty-four hours after transfection, cells were detached with PBS plus 5 mM EDTA and co-cultured with CMTMR prelabeled effector HTX/LH2SN cell lines, either parental (mock) or derivatives expressing IFITM proteins. After co-culture for 1 h, cells were treated with a pH 5.0 buffer for 1 min, and recovered in complete growth medium for another 1 h. Cells were analyzed for fusion under a fluorescence microscope or by flow cytometry. For a second cell-cell fusion assay, the viral fusion proteins (JSRV Env, VSV G, and SFV E1/E2) were separately expressed in COS7 cells to generate effector cells. In a few experiments, 293T cells were employed as effector cells to test whether the type of effector cell was of functional consequence. For IAV, the cell line HAB2 stably expressing influenza virus HA was used as the effector. 293/LH2SN cells stably expressing IFITM proteins, as described above, were the target cells. Effector cells were loaded with the fluorescent dye calcein-AM (Invitrogen) and targets were labeled by the dye CMAC (Invitrogen). For fusion experiments, cells were allowed to bind for 30 min at room temperature and pH was then lowered, through exchange of aqueous solutions, for 10 min, to 5.0 for JSRV Env, 4.8 for IAV HA, 5.7 for VSV G, and 5.4 for SFV E1/E2. The pH was then reneutralized to 7.2, and 30 min later fusion between pairs of effector and target cells was scored by the transfer of both aqueous dyes, as observed by fluorescence microscopy. Low pH rescue of BafA1 or IFITM1-mediated block on vial entry Experiments were performed as previously described [33], [34]. Briefly, HTX or HTX/IFITM1 cells were pretreated with 20 nM BafA1 for 2 h and spininoculated with JSRV or IAV pseudovirions at 4°C for 1 h. Following three washes with cold PBS, the virion-cell complexes were either directly exposed to a pH 5.0 solution (for IAV) 5–10 min or were preincubated at 37°C for 1 h in the presence of 20 nM BafA1 (for JSRV) and then incubated with a pH 5.0 for 5–10 min. In both cases, the total period of infection in the presence of 20 nM BafA1 was 4 h. Noninternalized virus was inactivated using citrate buffer (pH 3.0) after the infection period, and viral infectivity was determined by counting AP-positive foci 72 h after the initiation of infection. A pH 7.5-PBS buffer and 0.01% DMSO served as controls for the pH 5.0 buffer and 20 nm BafA1, respectively. Creating and using CAS After binding labeled effector and target cells on cover slips within culture dishes at room temperature for 30 min, the dishes were placed on ice, bringing the solutions bathing the cells to 4°C. The pH was lowered (∼pH 5.0) for 10 min before reneutralizing at 4°C to pH 7.2. At this point, the cells were in a “cold-arrested” stage (CAS). After 3 min, one of two operations was performed. In the first, 0.5 mM CPZ was added (through exchange of solutions at 4°C) and 1 min later the CPZ was washed out with a solution containing delipidated-BSA. Each dish was kept on ice, and each cover slip was removed to monitor aqueous dye transfer by fluorescence microscopy. In the second manipulation, cells at CAS were placed in a 37°C incubator for 30 min and dye transfer was then monitored. In order to make spontaneous monolayer curvatures more negative, 285 µM oleic acid (OA, Sigma) was incorporated into cell membranes either prior to or subsequent to creating CAS. For prior incorporations, effector and target cells were incubated together at room temperature for 20 min and OA was then added; 15 min later, the solutions were lowered to 4°C and CAS was created. OA was removed at 4°C by washing the cells with a solution containing delipidated-BSA. CPZ was then added or temperature was raised to 37°C. To incorporate OA subsequent to CAS, OA was added to cells at CAS, and CPZ was added or the temperature was raised without removing OA. Laurdan labeling The membrane probe Laurdan (6-dodecanoyl-2-dimethylamino naphthalene, Invitrogen) was dissolved in DMSO (dimethylsulfoxide) to make a stock concentration of 1.8 mM. 293/LH2SN cells expressing IFITM1, 2 or 3, or none (Mock) were incubated with 1.8 µM Laurdan for 40 min at 37°C. To deplete cholesterol, a 50 mM stock solution of methyl-β-cyclodextrin (MβCD, Sigma-Aldrich) was prepared by dissolving in nanopure water. Cells were incubated with 10 mM MβCD for 1 h at 37°C. All cells were rinsed with PBS once before being processed for imaging. Microscope handling, imaging, and data analysis involved in Laurdan labeling In order to quantitatively assess the membrane order, a ratiometric method known as generalized polarization (GP) was developed [52]. The GP function or value characterizing the spectral properties of Laurdan is calculated through the following expression: (1) where Iblue and Igreen are the respective intensities conventionally centered at 440 nm (the emission maximum for more ordered lipid bilayer) and centered at 490 nm (the emission maximum for less ordered lipids). GP and FLIM data were acquired with a Zeiss LSM710 META Laser scanning microscope, coupled to a 2-Photon Ti:Sapphire laser (Spectra-Physics Mai Tai, Newport Beach, CA) producing 80 fs pulses at a repetition of 80 MHz and a ISS A320 FastFLIMBox for the lifetime data. A 40× water immersion objective 1.2 N.A. (Zeiss, Oberkochen, Germany) was used for all experiments. The excitation wavelength was set at 780 nm. A SP 760 nm dichroic filter was used to separate the fluorescence signal from the laser light. For FLIM data, the fluorescence signal was directed through a 495 LP filter and the signal was split between two photo-multiplier detectors (H7422P-40, Hamamatsu, Japan), with the following bandwidth filters in front of each: blue channel 460/40 and green 540/25, respectively. For image acquisition, the pixel frame size was set to 256×256 and the pixel dwell time was 25.61 µs/pixel. The average laser power at the sample was maintained at the mW level. For GP data the fluorescence signal was acquired from 416 nm to 474 nm for the blue channel and from 474 nm to 532 nm for the green channel, using the spectral detector of the LSM 710 by joining 6 channels of the detector each having a bandwidth of 9.7 nm. For image acquisition, the pixel frame size was set to 256×256 and the pixel dwell time was 177.32 µs/pixel. The average laser power at the sample was maintained at the mW level. SimFCS software developed at the Laboratory for Fluorescence Dynamics (www.lfd.uci.edu) was used to acquire FLIM data and to process FLIM and GP data. Calibration of the system and phasor plot for FLIM data was performed by measuring fluorescein (pH 11), which has a known single exponential lifetime of ∼4.04 ns. Calibration for GP data was performed with Laurdan in DMSO which has a reference GP value equal to zero. GP histograms of the images were calculated using SimFCS, and the histograms fitting was performed by using a MatLab routine. GP histograms were characterized by the presence of two Gaussian distributions. After highlighting the corresponding pixels in the images, it was possible to couple the first Gaussian distribution (centered at lower GP values) with the intracellular membranes of the cell, and the second distribution (centered at higher GP values) with the plasma membranes [70]. The two histogram distributions were used to calculate the averaged GP values of intracellular and plasma membranes. Statistical analysis Paired student t test was used for statistical analysis unless otherwise noted. Typically data from three to eight independent experiments were used for the analysis. Supporting Information Figure S1 Effect of wildtype IFITMs on JSRV, 10A1 MLV, IAV and VSV entry. Experiments were performed as described in Fig. 1 except that HTX cells expressing the wildtype IFITM1, 2 or 3 were used for infection. (TIFF) Click here for additional data file. Figure S2 Examination of the expression of IFITMs in 293 cells. 293 or 293 cells stably expressing IFITM1, 2 or 3 were treated with IFN-α2b (500 units/ml) for 24 h or left untreated, and cell lysates were examined for IFITM expression by Western blot using anti-IFITM1, anti-IFITM2, anti-IFITM3, and anti-FLAG antibodies, respectively. Tubulin served as a loading control, which was determined by an anti-Tubulin antibody. Note that the levels of endogenous IFITM expression induced by IFN-α2b in 293 cells were much less than those of IFITM overexpression. (TIFF) Click here for additional data file. Figure S3 Effect of different pH on syncytia formation induced by JSRV Env. Experiments were performed similarly as described in Fig. 3, except that indicated pH buffers were applied. Experiments were performed three times, with similar results obtained. Representative images are shown. (TIFF) Click here for additional data file. Figure S4 Effect of wildtype IFITMs on syncytia formation induced by JSRV Env and IAV HA. Assays were carried out as described in Fig. 3, except 293/LH2SN cells expressing the wildtype IFITM1, 2 or 3 were used. Experiments were repeated at least 4 times, with similar results obtained. Representative images are presented. (TIFF) Click here for additional data file. Figure S5 Human IFITM3 inhibits JSRV Env-mediated entry and cell-cell fusion in COS7 cells. (A) COS7/LH2SN cells expressing human IFITM1, 2 or 3 were infected with GFP-encoding MLV pseudovirions bearing JSRV Env or IAV HA/NA; 24 h after infection, infectious titers were determined by flow cytometry. Flow cytometry profiles from one typical experiment are shown. (B) Experiments were performed as described in Fig. 3, except that COS7/LH2SN cells expressing IFITM1, 2 or 3 were transfected with plasmid encoding JSRV Env or IAV HA, and that syncytia formation was examined following a pH 5.0 treatment for 5 min. For each cell line, the representatives of both phase-contrast and GFP images are shown; arrows indicate syncytia induced by JSRV Env. (TIFF) Click here for additional data file. Figure S6 Effect of IFITM expression on the lipid order of cell membranes examined by FLIM. Cells were analyzed by fluorescence-lifetime imaging microscopy (FLIM). FLIM images were acquired by using ISS A320 FastFLIMBox. SimFCS software developed at the Laboratory for Fluorescence Dynamics (University of California, Irvine) was used to acquire FLIM data and to process FLIM and GP data. The Phasor approach was used to directly visualize the Laurdan lifetime distribution and to associate a color map to lifetime values (see reference 53). Note that green cursors are associated with shorter lifetimes or less ordered lipid membranes (e.g. MβCD-treated cells), while red cursors correspond to longer lifetimes, ordered lipid membranes (e.g., IFITM-expressing cells). (A) Fluorescence intensity image. (B) FLIM image in the green channel. (C) Phasor plot. (D) Phasor color palette distribution. (TIFF) Click here for additional data file.

0 comments Cited 192 times – based on 0 reviews      Review now

Bookmark

Record: found
Abstract: found
Article: not found

Interferon-induced cell membrane proteins, IFITM3 and tetherin, inhibit vesicular stomatitis virus infection via distinct mechanisms.

X. Pan, Qing-Dong Jiang, Irene J. Chang … (2010)

Tetherin and IFITM3 are recently identified interferon-induced cellular proteins that restrict infections by retroviruses and filoviruses and of influenza virus and flaviviruses, respectively. In our efforts to further explore their antiviral activities against other viruses and determine their antiviral mechanisms, we found that the two antiviral proteins potently inhibit the infection of vesicular stomatitis virus (VSV), a prototype member of the Rhabdoviridae family. Taking advantage of this well-studied virus infection system, we show that although both tetherin and IFITM3 are plasma membrane proteins, tetherin inhibits virion particle release from infected cells, while IFITM3 disrupts an early event after endocytosis of virion particles but before primary transcription of incoming viral genomes. Furthermore, we demonstrate that both the N-terminal 21 amino acid residues and C-terminal transmembrane region of IFITM3 are required for its antiviral activity. Collectively, our work sheds light on the mechanisms by which tetherin and IFITM3 restrict infection with rhabdoviruses and possibly other pathogenic viruses.

0 comments Cited 155 times – based on 0 reviews      Review now

Bookmark

All references

Author and article information

Journal

Journal ID (nlm-ta): Sci Rep

Journal ID (iso-abbrev): Sci Rep

Title: Scientific Reports

Publisher: Nature Publishing Group

ISSN (Electronic): 2045-2322

Publication date (Electronic): 05 April 2016

Publication date Collection: 2016

Volume: 6

Electronic Location Identifier: 24029

Affiliations

[1 ]Hefei National Laboratory of Microscale Physical Sciences, School of Life Science, University of Science and Technology of China , Hefei, 230027, P. R. China

[2 ]High Magnetic Field Laboratory, Chinese Academy of Sciences , Hefei, 230031, P. R. China

Author notes

[a ] zlhustc@ 123456ustc.edu.cn

[b ] cltian@ 123456ustc.edu.cn

[*]

These authors contributed equally to this work.

Article

Publisher Item ID: srep24029

DOI: 10.1038/srep24029

PMC ID: 4820770

PubMed ID: 27046158

SO-VID: 748c7cf4-5da7-434c-8816-c4cded999b25

License:

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Combined approaches of EPR and NMR illustrate only one transmembrane helix in the human IFITM3

Read this article at

Abstract

Related collections

Alternative approaches to Alzheimer's Disease

Most cited references 25

IFITM3 Inhibits Influenza A Virus Infection by Preventing Cytosolic Entry

IFITM Proteins Restrict Viral Membrane Hemifusion

Interferon-induced cell membrane proteins, IFITM3 and tetherin, inhibit vesicular stomatitis virus infection via distinct mechanisms.

Author and article information

Journal

Affiliations

Author notes

Article

History

Categories

Comments

Comment on this article

Similar content 560

Cited by 32

Most referenced authors 440