Interferon-Induced Transmembrane Protein 3 Is a Virus-Associated Protein Which Suppresses Porcine Reproductive and Respiratory Syndrome Virus Replication by Blocking Viral Membrane Fusion

Porcine reproductive and respiratory syndrome (PRRS), which is caused by PRRS virus (PRRSV), is of great economic significance to the swine industry. Due to the complicated immune escape mechanisms of PRRSV, there are no effective vaccines or therapeutic drugs currently available against PRRS. Identification of cellular factors and underlying mechanisms that establish an effective antiviral state against PRRSV can provide unique strategies for developing antiviral vaccines or drugs. As an interferon (IFN)-stimulated gene, the role of IFN-induced transmembrane 3 (IFITM3) in PRRSV infection has not been reported as of yet. In the present study, it was shown that IFITM3 can exert a potent anti-PRRSV effect, and PRRS virions are trafficked to IFITM3-containing cell vesicles, where viral membrane fusion is impaired by cholesterol accumulation that is induced by IFITM3. Additionally, both endogenous and exogenous IFITM3 are incorporated into newly assembled progeny virions, and this decreased their intrinsic infectivity.

endogenous and exogenous IFITM3 are incorporated into newly assembled progeny virions, and this decreased their intrinsic infectivity. KEYWORDS IFITM3, PRRSV, cholesterol, cellular vesicles, membrane fusion I nnate antiviral immunity is orchestrated by the interferon (IFN) system, which plays a pivotal role in early perception and confrontation of invading viruses by the mammalian host, and the interaction between the virus and the host IFN system largely determines the outcome of most viral diseases. IFN, whether endogenous or exogenous, induces the expression of a large number of IFN-stimulated genes (ISGs), the primary host effectors mediating the establishment of an antiviral status against viral infection. Hundreds of ISGs have been identified thus far, only a few of which have had their function characterized (1,2). Understanding how individual ISGs restrict viral infection is of broad interest for designing antiviral therapeutics that can be targeted to specific or more general classes of pathogens.
The IFN-induced transmembrane proteins (IFITMs) are a family of ubiquitously expressed transmembrane proteins that respond differentially to IFN induction and viral infections. Humans have five IFITM proteins, IFITM1, 2, 3, 5, and 10 (3). IFITM1, 2, and 3 are expressed in a wide range of tissues, whereas IFITM5 expression appears to be limited to skeletal tissue, and the function of IFITM10 remains ambiguous (4). Thus, most studies have focused on IFITM1, 2, and 3. IFITM1, 2, and 3 have emerged as broad-acting restriction factors capable of disturbing the replication of multiple RNA viruses that enter the host cell through endocytosis, including influenza A virus (IAV), Dengue virus (DENV), West Nile virus, SARS coronavirus (SARS-CoV), and hepatitis C virus (5,6). Although all three IFITM proteins are able to restrict viral replication to various degrees, IFITM3 alone mediates a large extent of the antiviral effects of IFN in vitro (7). IFITM3 is the most potent IFITM family member in restricting IAV replication in vitro (8,9). Notably, IFITM3 Ϫ/Ϫ mice are more susceptible to IAV infection (10,11). IFITM1, 2, and 3 have been shown to restrict Mycobacterium tuberculosis replication in human monocytes, while IFITM3 seems to play a central role in this process (12). For certain livestock viruses, IFITMs display antiviral effects as well. For example, IFITMs are reported to suppress replication of African swine fever virus, classical swine fever virus, and avian tembusu virus in vitro (13)(14)(15). However, some viruses are resistant to IFITM-mediated restriction. For example, IFITMs do not restrict infection of mouse leukemia virus, Machupo virus, Lassa virus, or lymphocytic choriomeningitis virus (16), highlighting the potential dual roles of IFITM3 in inhibiting viral replication.
The underlying mechanism by which IFITMs inhibit infection of a range of viruses, however, is still largely unknown. Accumulating evidence suggests that IFITMs may interfere with virus-endosome fusion to block enveloped virus entry (17)(18)(19)(20). In mammalian cells, IFITM1 is primarily located in the plasma membrane, while IFITM2 and IFITM3 are predominantly localized to early and late endosomes and lysosomes, as shown by immunofluorescence and live-cell imaging studies (21). The cellular localization of the IFITMs may be a crucial determinant of their specificities toward viruses, as IFITM1 is more potent against viruses that enter through the plasma membrane or early endosomes, whereas IFITM2 and IFITM3 are more potent against viruses that enter cells through late endosomal compartments. Effective restriction of viruses that enter from the late endosome, such as IAV, Ebola virus, and SARS-CoV seems consistent with the cellular localization of IFITM2 and IFITM3 proteins (16). Indeed, mutation of IFITM3 that redistributes the late endosome/lysosome-resident protein to the cytoplasmic membrane abolishes its antiviral activity against IAV (22). However, IFITMs also restrict vesicular stomatitis virus, which appear to fuse with early endosomes (9). As for the mechanism associated with the inhibition of membrane fusion, IFITMs have been reported to curtail viral infection, in part by resulting in the accumulation of cholesterol in late endosomes as a result of IFITM-mediated disruption of the interaction between the vesicle membrane-protein-associated protein A (VAPA) and oxysterol binding protein (OSBP) (23). A recent study provided evidence of the antiviral effect of choles-terol accumulation in late endosomes/lysosomes and confirmed accumulation of cholesterol in these membrane-associated compartments upon IFITM3 expression (13,23). Due to the important role of lipids in membrane fusion, these findings offer an attractive paradigm for a broad antiviral defense mechanism that involves altering the lipid composition of cellular membranes. Yount et al. recently showed that human IFITM3 undergoes both ubiquitination and S-palmitoylation modification in cells and that these posttranslational modifications strikingly regulate IFITM3 cellular localization and its anti-influenza activities (24). In addition to the aforementioned antiviral mechanisms, certain studies have shown that IFITMs abolished viral replication through merging with the viral envelope protein during virion assembly and are eventually incorporated into newly produced particles that display decreased infectivity compared with their wild-type (WT) counterparts (25,26).
Porcine reproductive and respiratory syndrome (PRRS), caused by PRRS virus (PRRSV), is characterized by reproductive failure in sows and respiratory diseases in pigs of all ages and results in huge economic losses to the swine industry worldwide (27). PRRSV is an enveloped, single-stranded RNA virus which belongs to the genus Porartevirus, family Arteriviridae, and order Nidovirales (28,29). PRRSV has highly restricted cell tropism whereby it infects the monocyte-macrophage lineage cells, including porcine alveolar macrophages (PAMs), the primary target of PRRSV infection in vivo (30,31). The African green monkey kidney cell line MA-104 and its subclone Marc-145 are also susceptible to PRRSV infection and have been frequently used in PRRSV studies in vitro (32,33). Although the mechanism of PRRSV particle entry into host cells remains to be fully understood, it has been shown that the envelope glycoproteins (GP3, GP4, GP5, and M) on the surface of the viral envelop regulate cell entry via binding to cell surface receptors, such as scavenger receptor (CD163), sialoadhesin (Sn), and heparan sulfate (HS) among others (34)(35)(36). After receptor clathrin-dependent endocytosis, the endocytosed virions further undergo low-pH-dependent viral membrane fusion, which leads to the nucleocapsid disintegration and the release of viral genomes into the cytoplasm, initiating its replication process. Previous studies concluded that after entering the host cells, PRRSV particles are transported to early endosomes without being further transported to acidified late endosomes and lysosomes (37), exhibiting an intracellular transport pathway that appears to be less consistent with other enveloped RNA viruses. Thus far, the effect of IFITM3 on PRRSV replication and the relevant mechanism is unknown, to the best of our knowledge. Since the antiviral activity of IFITM3 is likely mediated by preventing endosome fusion and subsequent viral entry into the cytosol (19,38), or by leading to the production of virions with reduced infectivity, we hypothesize that swine IFITM3 is able to abrogate PRRSV replication by preventing virions escaping from the endocytic pathway during an early stage of the life cycle and by incorporation into the envelope of newly produced PRRS virions that lead to decreased infectivity.
The results of the present study show that both endogenous and exogenous IFITM3 serves an important role in prohibiting PRRSV replication, and S-palmitoylation and ubiquitination modifications benefit both anti-PRRSV activities of IFITM3. Furthermore, IFITM3 is essential in mediating antiviral activity of IFN-␣. Further study suggests that PRRSV particles are first transported into early endosomes and then into late endosomes and lysosomes, which is inconsistent with previous reports that the virus entered only the early endosomes but not late endosomes or lysosomes in the early stage of infection. By using single virus particle fluorescent labeling, we demonstrate that IFITM3 restricts PRRSV membrane fusion by disturbing intracellular cholesterol homeostasis. Both exogenous and endogenous IFITM3 are incorporated into virions and reduce viral infectivity, and intercellular transmission of PRRSV is abolished by IFITM3 as well. In conclusion, the results of the present study are the first to demonstrate that IFITM3 restricts replication of one member of the arteritis virus family PRRSV that utilizes endosome-dependent cell entry mechanisms. Additionally, IFITM3 interferes with PRRSV replication by resulting in the production of virions with reduced infectivity.
The dual inhibitory effect also emphasizes the importance of IFITM3 in the host anti-PRRSV response.

IFITM3 inhibits PRRSV replication in vitro.
To examine the expression of IFITM3 following PRRSV infection, PAMs were infected or mock infected with 0.1 multiplicity of infection (MOI) PRRSV, and IFITM3 was detected using reverse transcriptase quantitative PCR (RT-qPCR). The results showed that PRRSV infection induced IFITM3 expression in PAMs (Fig. 1A). To understand the role of IFITM3 on PRRSV replication, recombinant cell lines stably expressing swine IFITM3 were first established. As shown in Fig. 1B, expression of IFITM3 in Marc-145-IFITM3-flag cells (lower panel) was notably overex- pressed compared with the Marc-145-Vector cell line (upper panel). Western blotting using an anti-flag antibody further confirmed the overexpression of IFITM3 protein (Fig.  1C), suggesting that the recombinant cell line was successfully established.
Subsequently, the effect of IFITM3 on PRRSV infection was assessed. According to the Western blotting and immunofluorescence assay (IFA) results, expression of PRRSV N protein was notably decreased compared with the control group after 24, 36, and 48 hours postinfection (hpi) (Fig. 1D and E). Overexpression of IFITM3 markedly decreased supernatant virus titers as well (Fig. 1F). As the natural host cell of PRRSV, the effect of IFITM3 on PRRSV replication in PAMs was also determined. Consistently, overexpression of IFITM3 reduced N protein expression in PAMs compared with the control group ( Fig.  1G and H) and simultaneously decreased supernatant virus titers (Fig. 1I). In addition, Western blotting results suggested that IFITM3 suppressed GD-HD-, JXA1-, and VR2332-PRRSV N protein expression, suggesting that IFITM3 suppression of PRRSV infection was not strain dependent (Fig. 1J). Collectively, these results suggest that overexpression of IFITM3 can inhibit PRRSV replication in vitro.
PRRSV particles traffic along the endosome-lysosome pathway. The intracellular transportation of PRRSV particles was first investigated to determine on which stage of the PRRSV replication cycle IFITM3 exerted its antiviral effects. Colocalization was defined as any observed merge of the red fluorescence of viral protein with the green fluorescence of cell organelles. As shown in Fig. 2A, colocalization was observed between PRRSV particles and early/late endosomes or lysosomes in Marc-145 cells. To further confirm that PRRSV required early/late endosomes and lysosomes for transportation, subcellular localization of PRRS virions was analyzed in PAMs as well as the virus' in vivo target cells. Confocal results showed that PRRS virions occurred in early/late endosomes and lysosomes similar to the Marc-145 cells (Fig. 2C), and colocalization analysis results further confirmed the above results ( Fig. 2B and D). The red spots at the periphery of the cells represent viruses that were sticking to the cell but were not yet internalized ( Fig. 2A and C). Previous reports showed that PRRSV particles were trafficked only to early endosomes but not to late endosomes and lysosomes for productive infection (37). The above results suggest that PRRS virions are trafficked along the endosome-lysosome pathway, similar to other enveloped viruses. As PRRSV particles undergo a similar transport process within both Marc-145 cells and PAMs, Marc-145 cells were selected as the research model for the subsequent experiments to examine the underlying molecular mechanism of anti-PRRSV activity exerted by IFITM3, unless otherwise indicated.
IFITM3 does not block PRRSV attachment, entry, or access to cell vesicles. To determine the stage of the PRRSV life cycle that was restricted by IFITM3, we first characterized the kinetics of the viral N protein and NSP2 gene expression during the course of PRRSV infection. As shown in Fig. 3A, IFA results indicated that viral N protein was initially detected at ϳ8 hpi. RT-qPCR results showed that NSP2 gene levels began to increase from ϳ8 hpi (Fig. 3B), consistent with the IFA results, suggesting that at this time point, viral transcription and replication were already under way, and the replication cycle of PRRSV is ϳ8 h. Compared with the viral replication kinetics in Marc-145-Vector cells, both N protein and NSP2 gene levels began to lower in Marc-145-IFITM3-flag cells compared with control cells from 8 h onward, and the gap gradually expanded, as illustrated by the IFA and RT-qPCR results ( Fig. 3A and B). Taken together, these results suggest that IFITM3 exerted its antiviral effects during a single viral replication cycle.
To further investigate at which stage of the single cycle of PRRSV replication IFITM3 exerted its antiviral effects, the effect of IFITM3 on PRRSV attachment and internalization was assessed. RT-qPCR results of attachment and entry assay both showed that there was no significant difference in the amount of viral RNA associated with control or IFITM3-overexpressing cells ( Fig. 3C and D), indicating that IFITM3 does not modulate viral attachment and internalization into cells. Subsequently, whether IFITM3 prevented PRRSV particles from translocating to endosomes and lysosomes was investigated using confocal analysis. A total of 10 different fields of view were selected for each sample, and colocalization analysis was performed using ImageJ; representative images are presented in Fig. 3E. Confocal analysis results indicated that overexpression of IFITM3 did not block PRRS virion access to acidic endosomes or lysosomes (Fig. 3E). Taken together, these results indicate that IFITM3 does not block PRRSV attachment, entry, or access to endosomes or lysosomes during the early stages of replication.
Both PRRS virions and IFITM3 colocalize with endosomes and lysosomes. Since our results showed that PRRSV was transported to endosomes and lysosomes at the beginning of infection (Fig. 2), the subcellular localization of IFITM3 was further determined. Marc-145-IFITM3-flag cells were transiently transfected with pcDNA3.1-Rab5-GFP, pcDNA3.1-Rab7, and pcDNA3.1-LAMP1-YFP plasmids, Rab7 was labeled with  an anti-Rab7 antibody, and IFITM3 was labeled with an anti-IFITM3 antibody. Colocalization analysis results showed that the vast majority of IFITM3 colocalized with early/late endosomes and lysosomes (Fig. 4A), and PRRSV infection did not change the subcellular distribution of IFITM3 (Fig. 4C). We next investigated whether internalized PRRSV particles were transported to IFITM3-positive endosomes or lysosomes. PRRSV was first bound and allowed to enter the Marc-145-IFITM3-flag cells by incubating at 37°C for 45 min. Cells were fixed and labeled for N and IFITM3-flag proteins and , and -LAMP1-YFP plasmids for 48 h were fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton X-100. Cells were labeled with anti-IFITM3 and -Rab7 antibodies followed by Alexa Fluor 594-conjugated goat anti-mouse IgG (H&L) antibody for IFITM3 (red) and Alexa Fluor 488-conjugated goat anti-rabbit IgG (H&L) antibody for Rab7 (green), and cell nuclei were counterstained using DAPI. Rab5 and LAMP1 (green) were observed directly using a confocal microscope. (B) PRRSV were allowed to bind to prechilled Marc-145-IFITM3-flag cells for 1 h on ice, unbound virions were washed, and cells were warmed to 37°C for 45 min. IFITM3 was stained using an anti-IFITM3 antibody and the corresponding fluorescent second antibody (blue), and PRRSV was stained using an anti-N antibody and a fluorescent second antibody (red). (C) Marc-145-IFITM3-flag cells were infected with 0.1 MOI of PRRSV for 12 h, and then cells were transfected with pcDNA3.1-Rab5-GFP, -Rab7, and -LAMP1-YFP plasmids for 36 h and fixed with 4% paraformaldehyde and permeabilized with 0.3% Triton X-100. Cells were labeled as described in visualized by confocal microscopy. Representative images of PRRSV particle colocalization with IFITM3-containing vesicles are presented in Fig. 4B. These data suggest that endocytosed PRRSV particles are delivered to IFITM3-positive endosomes and lysosomes in the initial stages of infection.

IFITM3 restricts membrane fusion between PRRSV and cell vesicles.
To further investigate the molecular mechanism of IFITM3 inhibition of PRRSV infection, a direct virus-cell fusion assay was employed to evaluate the extent of restriction. To visualize single PRRSV-endosome fusion under a fluorescence microscope in Marc-145 cells, the PRRSV lipid membrane was labeled with two lipophilic probes, DiOC18 and R18. In the labeled virus, DiOC18 probes are incorporated into viral membranes at high concentrations, and this leads to suppression of the green fluorescence to a level similar to that of the red fluorescence by both self-quenching of DiOC18 and fluorescent resonance energy transfer from DiOC18 to R18. Virus-endosome fusion decreases the density of the green probe and consequently increases the dequenching of the fluorescence of the probe (Fig. 5A). The green (488-nm) and red (594-nm) fluorescence images of cells were obtained simultaneously using two detectors on a confocal microscope. Time-lapse images showed the initial trafficking of a representative DiOC18/R18 labeled virus in Marc-145-Vector cells, until fusion occurred, as evidenced by the enhancement of green fluorescence, while the red signal remained relatively constant ( Fig. 5B and C). Importantly, PRRSV membrane fusion was also readily detected in Marc-145-IFITM3-flag cells ( Fig. 5B and D). From these traces, the time required for complete dequenching (Δt) and the extent of dequenching (ratio of the initial and final mean intensities I f /I i ) can be determined; Δt was 5 min in Marc-145-Vector cells, which was shorter than the Marc-145-IFITM3-flag cells (13 min; Fig. 5C and D). Considering that PRRS virions colocalized with IFITM3-positive endosomes and lysosomes ( Fig. 4), we hypothesized that a sufficiently high local IFITM3 concentration was required for restriction of PRRSV fusion. To test this hypothesis, the number of virus particles that underwent membrane fusion events in both Marc-145-Vector and Marc-145-IFITM3-flag cells were counted for as long as the time-lapse imaging was performed. Analysis of the 532 particles revealed that only 18 single viral fusion events (3.38%) occurred after extensive PRRS virion colocalization with IFITM3-positive endosomes or lysosomes (Fig. 5E), while 32 single viral fusion events (6.82%) were observed in a total of 469 analyzed particles in Marc-145-Vector cells. Redistribution of DiOC18 was mediated by low pH-dependent virus-endosome fusion, as evidenced by potent inhibition of membrane fusion by NH 4 Cl treatment (1.53%; Fig. 5E), while AmphoB, which destabilizes lipid membranes, prominently reversed inhibition of membrane fusion elicited by IFITM3 (7.62%; Fig. 5E). The rate at which PRRSV particles underwent membrane fusion was longer in Marc-145-IFITM3-flag cells than in Marc-145-Vector cells (Fig. 5F), with a t 1/2 of approximately 62 min for the former, and 47 min for the latter. In AmphoB-treated Marc-145-IFITM3-flag cells, the DiOC18-PRRSV dequenching rate was quicker than that of untreated cells (Fig. 5F), and the t 1/2 for fusion occurred ϳ10 min earlier, even though virus trafficking times may vary widely even in synchronized infections (Fig. 5F). Additionally, AmphoB partially reversed the inhibitory effect of IFITM3 on PRRSV replication, as shown by the Western blot and fluorescenceactivated cell sorter (FACS) results ( Fig. 5G and H). Inhibition of intracellular vesicle acidification can also inhibit PRRSV replication, since concentration-dependent NH 4 Cl suppressed PRRSV replication in Marc-145 cells (Fig. 5I). Together, these findings suggest that the mechanism of IFITM3-mediated restriction arises likely from the disruption of membrane fusion required for PRRSV vesicle transport.

IFITM3-induced endosome/lysosome cholesterol accumulation impairs PRRSV membrane fusion.
A recent study has shown that IFITM3 caused cholesterol accumulation in late endosomes through disrupting the interaction between VAPA and OSBP (23). Change in endosomal distribution is a characteristic phenotype of altered cholesterol endosomal efflux, which is required for correct endosomal function. Thus, we speculated that IFITM3 might cause accumulation of cholesterol in endosomes or lysosomes, which results in obstruction of PRRSV fusion. Marc-145-Vector/-IFITM3-flag cells transfected with Rab5, Rab7, or LAMP1 expression plasmids were incubated with filipin for detection of intracellular cholesterol levels. Confocal analysis results showed that cholesterol was diffusely distributed in the plasma membrane, cytoplasm, and endosomes in Marc-145-Vector cells (Fig. 6A), while intense intracellular cholesterol accumulation was observed at the perinuclear area, and cholesterol also colocalized with the IFITM3-positive endosomes and lysosome vesicles in Marc-145-IFITM3-flag cells (Fig. 6B). These findings indicate that IFITM3 overexpression results in cholesterol accumulation in endosomal and lysosomal compartments. To further determine whether cholesterol accumulation in cell compartments was correlated with PRRSV fusion, U18666A, an amphipathic steroid which is widely used to block intracellular trafficking of cholesterol (39), was utilized to evaluate its effect on viral fusion. First, the subcellular distribution of cholesterol after U18666A treatment was detected in Marc-145 cells. Confocal results showed that in the presence of U18666A, cholesterol accumulated in the perinuclear swollen early/late endosomes and lysosomes (Fig. 6C). To evaluate the effect of cholesterol on PRRSV membrane fusion, viral envelope-endosome fusion assay utilizing DiOC18 and R18 labeled PRRSV was

IFITM3 Suppresses PRRSV Replication In Vitro
Journal of Virology employed to identify potential fusion defects. U18666A treatment-induced cholesterol accumulation resulted in a pronounced decrease in DiOC18 dequenching compared with the DMSO treatment group (6.55% versus 2.93%; Fig. 6D), indicating that PRRSV membrane fusion within cholesterol-laden endosomes or lysosomes was impaired. BafA1, an inhibitor of the vacuolar ATPase required for endosomal acidification, which can block endosomal acidification, an essential prerequisite for successful PRRSV membrane fusion, prominently abolished DiOC18 dequenching (6.55% versus 1.35%; Fig.  6D). To further clarify the effect of vesicle cholesterol on viral fusion, dequenching rates of DiOC18 were analyzed. The DiOC18-PRRSV dequenching rate was delayed in U18666A-treated Marc-145 cells compared with untreated cells, with a t 1/2 of ϳ11 min longer than that of untreated cells (Fig. 6E). The effect of U18666A treatment on PRRSV infection in Marc-145 cells was subsequently determined. Western blotting results showed that U18666A suppressed PRRSV N protein expression in a concentrationdependent manner (Fig. 6F), and FACS analysis of GFP-PRRSV-positive cells also showed similar results (Fig. 6G), indicating that cholesterol accumulation-induced membrane fusion disorder effectively impeded PRRSV replication. Both S-palmitoylation and ubiquitination of the IFITM3 protein are critical for anti-PRRSV activity. To determine whether swine IFITM3 was S-palmitoylated and the role of S-palmitoylation in its anti-PRRSV activity, amino acid sequence analysis was performed. The amino acid alignment of the IFITM3 from human, pig, and monkey showed that IFITM3 exhibited conserved amino acid sequences between the three species, such as the two transmembrane domains and the palmitoylation sites at positions 71, 72, and 105 (Fig. 7A), suggesting that swine IFITM3 may have posttranslational modification characteristics similar to the human counterpart. Thus, immunoprecipitation together with an acyl-biotin exchange (ABE) assay was performed to determine the S-palmitoylation of swine IFITM3. Western blotting results indicated that swine IFITM3 underwent S-palmitoylation in the Marc-145-IFITM3-flag cell line ( Fig. 7B and C), while the palmitoylation modification was significantly weakened after cysteine was mutated to alanine and almost completely disappeared when all the three cysteines were mutated to alanine ( Fig. 7D and E). To assess whether S-palmitoylation of IFITM3 was important for anti-PRRSV activity, single (C105A), double (CC71/72AA), and triple (CCC71/72/105AAA) cysteine to alanine mutant cell lines were infected with GFP-PRRSV for further analysis. Both Western blotting and FACS results showed that the single cysteine residue palmitoylation of IFITM3 was important for anti-PRRSV activity, as an almost complete loss of anti-PRRSV activity was observed when any of the cysteine residues were mutated ( Fig. 7F and G). Then colocalization between mutant IFITM3 and endosomes or lysosomes was analyzed. Confocal results indicated that IFITM3-ΔPal (CCC71/72/105AAA) still largely colocalized with early/late endosomes and lysosomes in Marc-145-IFITM3-ΔPal-flag cells (Fig. 7M), consistent with the results of Fig. 4A. However, according to the Western blotting results of Fig. 7D and F, mutation of cysteine to alanine markedly decreased modified IFITM3 expression and abolished its antiviral effect, suggesting that S-palmitoylation of IFITM3 is critical for its anti-PRRSV activity.
Sequence alignment of IFITM3 showed that lysine residues are highly conserved between human, swine and monkey (Fig. 7A). Thus, whether swine IFITM3 underwent ubiquitin modification was determined. Compared with the Marc-145-Vector cells, IFITM3 ubiquitination was notably increased in Marc-145-IFITM3-flag cells (Fig. 7H and I). To further explore the specific sites where ubiquitination occurred, cell lines with individual lysine mutants as well as IFITM3 mutants in which three of the four lysines or all four lysines were mutated to alanine (ΔUb) were generated using a lentiviral transduction system. Western blotting after immunoprecipitation showed that all four lysine residues of swine IFITM3 were ubiquitinated (Fig. 7J), and almost complete loss of ubiquitinated bands could only be visualized when all four lysines were mutated to alanine (Fig. 7J). Subsequently, whether swine IFITM3 ubiquitination influenced its anti-PRRSV activity was investigated. Both Western blotting and FACS analysis of IFITM3 mutants deficient in ubiquitination for activity against PRRSV infection showed that Journal of Virology mutation of any lysine to alanine resulted in IFITM3 almost completely losing its antiviral properties (Fig. 7K and L). Thus, whether ubiquitinated lysine influenced IFITM3 targeting to endosomes or lysosomes was determined. Indeed, IFITM3-ΔUb showed notably decreased colocalization with Rab5 and Rab7 compared with WT IFITM3 in Fig.  4A (Fig. 7M). Nevertheless most IFITM3-ΔUb still colocalized with lysosomes (Fig. 7M), suggesting that ubiquitination of IFITM3 might affect its anti-PRRSV activity by influencing its subcellular localization. Then the subcellular localization of PRRSV particles were determined. The results showed that PRRSV colocalized with early/late endosomes or lysosomes in both Marc-145-IFITM3-ΔPal-flag and Marc-145-IFITM3-ΔUb cells (Fig. 7N). Thus, in contrast to a previously reported role of IFITM3 posttranslational modifications on IAV (24), S-palmitoylation and ubiquitination of IFITM3 exhibited synergistic effects on regulation of its anti-PRRSV activity. IFITM3 is incorporated into PRRSV particles and diminishes viral infectivity. Studies have described an additional mechanism by which IFITMs are incorporated into newly produced HIV-1 particles that display reduced infectivity compared with their WT counterparts (25,26). However, whether the incorporation of IFITM3 into virions was restricted to HIV or whether it applied to PRRSV was not established. To focus only on PRRSV intrinsic infectivity, all the subsequent assays were performed after purification and normalization of viral particles produced in IFITM3-overexpressing or control cells (Fig. 8A).
Cell lysates and viral preparations obtained from both Marc-145-IFITM3-flag and PAM-IFITM3-flag and corresponding control cells were analyzed using Western blotting. Under these conditions, the primary PRRSV envelop proteins GP3 and GP5 and the capsid protein N in all cell lysates exhibited robust expression ( Fig. 8B and C). When purified and normalized virions were analyzed, there were no significant differences between the expression of GP3, GP5, and N protein in the preparations produced in the IFITM3-overexpressing or control cells ( Fig. 8B and C). However, readily detectable quantities of IFITM3 were observed in viral fractions from both Marc-145-IFITM3-flag cells and PAM-IFITM3-flag ( Fig. 8B and C), suggesting that IFITM3 could be a virionassociated protein. To further validate the finding that IFITM3 was a bona fide virionassociated protein, viruses produced from IFITM3-overexpressing or control cells were further subjected to immunogold electron microscopy using an antibody against the flag tag. Immunogold labeling results revealed that IFITM3 distributed at or closely to the virion envelope (Fig. 8D). Then, protein G MagBeads were coated with porcine antiserum and coincubated with supernatants collected from PRRSV-infected Marc-145-Vector or Marc-145-IFITM3-flag cells, and immunoprecipitation was performed. Western blotting results revealed that GP3, GP5, and N protein were readily detectable, and IFITM3 was detected simultaneously (Fig. 8E), further confirming the results shown in Fig. 8B and C. Next, the ability of IFITM3 to be incorporated in normalized purified viral preparations of different strains (GD-HD, JXA1, and VR2332) was assessed by Western blotting. In agreement with the above results, IFITM3 was readily detected in viral preparations from the other three PRRSV strains (Fig. 8F). These results indicate that IFITM3 is packaged in substantial quantities in PRRS virions.
Subsequently, the impact of IFITM3 on the infectivity of progeny virions was determined by infecting Marc-145 cells with normalized quantities of purified virions harvested at 24, 36, or 48 hpi, followed by determination of GFP-PRRSV positive cell percentage or N protein expression, which was indicative of virion infectivity. Notably, virus recovery from supernatants of Marc-145-IFITM3-flag cells revealed reduced virion infectivity compared with the WT at all three time points (Fig. 8G). To determine whether the infectivity deficit of PRRSV particles produced in the presence of IFITM3 was target cell type specific, normalized PRRS virions were used to challenge PAMs, and the percentage of GFP-PRRSV-positive PAMs or N protein expression was determined. Even using different target cells, the infectivity of viruses produced from PAM-IFITM3- cell lysates and supernatants purified by ultracentrifugation through sucrose density gradient were collected and analyzed by Western blotting. (C) As described above, PAM lysates and purified virions were analyzed. (D) Virions produced as described above were analyzed using immunogold electron microscopy. Unfixed viral preparations produced from IFITM3-overexpressing or control cells were purified and incubated with anti-flag antibodies, followed by incubation with 6 nm gold-conjugated secondary antibody. Representative pictures are presented. Scale bar, 100 nm. (E) Viral supernatants produced from IFITM3-overexpressing or control cells were concentrated and purified, followed by immunoprecipitation using swine antiserum. Western blotting was performed using an anti-flag antibody to detect IFITM3 and using swine antiserum to detect PRRSV GP3 and GP5. (F) Virions of different PRRSV strains obtained as described above were subjected to Western blotting using anti-IFITM3 antibody, and N protein was used as the loading control. (G and H) Incorporation of IFITM3 into PRRS virions decreased viral infectivity. Newly produced virions from IFITM3-overexpressing or control cells were collected 24, 36, or 48 hpi, purified, and normalized prior to infectivity analysis. (I) IFITM3 did not affect PRRSV major envelope protein processing and incorporation into viral particles. Equal amounts of N protein were loaded into each lane, and Western blotting was performed using anti-flag, swine antiserum, and anti-N antibodies. PRRSV, porcine reproductive and respiratory syndrome virus; IFITM, interferon-induced transmembrane; MOI, multiplicity of infection; PAM, porcine alveolar macrophage; hpi, hours postinfection. flag also displayed a characteristic decrease (Fig. 8H), indicating that this phenomenon was independent of the target cell. Human IFITMs have been reported to impair HIV-1 infectivity by decreasing Env processing in the host cell and reducing virions' Env incorporation (40). To determine the impact of IFITM3 on PRRSV envelope protein incorporation, GP3 and GP5 were determined in normalized purified virions from IFITM3-overexpressing or control cells of different strains. Notably, IFITM3 did not affect the level of viral envelope glycoproteins incorporated into virions of different PRRSV strains (Fig. 8I), which appeared to be inconsistent with the reported mode of action of IFITM3 on HIV (40).

Endogenous IFITM3 is incorporated into PRRS virions, and knockdown of IFITM3 increases viral replication and infectivity.
Since overexpression of IFITM3 abrogated PRRSV replication (Fig. 1), we hypothesized that basal levels of IFITM3 possessed the potential to limit PRRSV replication as well. Marc-145 cells stably transfected with shIFITM3 or PAMs transfected with small interfering RNA (siRNA) targeting IFITM3 both significantly reduced endogenous IFITM3 expression (Fig. 9A). Western blotting results revealed that knockdown expression of endogenous IFITM3 significantly enhanced PRRSV infection in both Marc-145 cells and PAMs (Fig. 9B), and FACS analysis also showed similar results (Fig. 9C and D), further confirming that basal levels of IFITM3 were effective as an anti-PRRSV factor.
To further assess the role of endogenous IFITM3 in PRRSV infection, Marc-145-shIFITM3 cells or PAMs transfected with IFITM3 siRNA were challenged with 0.1 MOI of GFP-PRRSV, and newly produced virions were then purified and normalized, and their infectivity was assessed (according to the scheme provided in Fig. 8A). Intracellular reduction of IFITM3 led to a small concomitant reduction of virion-associated IFITM3 when sufficient N protein was loaded (Fig. 9E). When newly produced virions were normalized and used to challenge Marc-145 cells or PAMs, the infectivity of virions produced in IFITM3-knockdown cells significantly increased compared with the control cells ( Fig. 9F and G). Thus, depletion of endogenous IFITM3 resulted in an increase in the intrinsic infectivity of progeny PRRS virions, corroborating the results presented in Fig. 8G and H.
Given that IFITM3 is an ISG, the role of IFITM3 in mediating the anti-PRRSV activity of IFN-␣ was investigated. Marc-145 cells or PAMs were stimulated for 24 h with IFN-␣ prior to infection with GFP-PRRSV. Western blotting results suggested that the abundance of IFITM3 protein in both Marc-145 cells and PAMs increased markedly in an IFN-␣ dose-dependent manner (Fig. 9H and I), suggesting that expression of IFITM3 was induced by IFN-␣ in both PRRSV permissive cells. In contrast to the upward trend of IFITM3, expression of PRRSV N protein decreased gradually with the increase in IFN-␣ ( Fig. 9H and I). Additionally, following IFN-␣ stimulation, IFITM3 incorporation of newly produced virions was readily detectable (Fig. 9J). When virions produced under these conditions were purified, normalized, and used to challenge Marc-145 cells or PAMs, a significant decrease in infectivity was observed from virions produced in IFN-␣-treated cells versus viruses produced in the control cells ( Fig. 9K and L).
To verify whether IFITM3 mediated the anti-PRRSV activity of IFN-␣, Marc-145-shNC/-shIFITM3 cells treated with IFN-␣ were infected with GFP-PRRSV and then analyzed via Western blotting and FACS. Western blotting results showed that the N protein level in IFN-␣-treated shIFITM3 cells was significantly increased compared with the shNC cells treated with IFN-␣ (Fig. 9M), and FACS analysis showed similar results (Fig. 9N), suggesting that IFITM3 partially mediated the anti-PRRSV activity of IFN-␣. In the following conditions, purified and normalized virions were inoculated to Marc-145 cells to evaluate their infectivity. Virions from IFN-␣-treated Marc-145-shNC cells showed dramatically decreased infectivity compared with virions from Marc-145-shNC cells (Fig.  9O), while knockdown of IFITM3 expression partially increased the virions' infectivity compared with shNC cells treated with IFN-␣ (Fig. 9O). Together, these results suggest that endogenous IFITM3 is incorporated into PRRSV particles and positively interferes with their infectivity.
IFITM3 effectively blocks PRRSV cell-to-cell transmission. Since virions from virus-producing cells in the presence of IFITM3 exhibited decreased infectivity, we evaluated the antiviral effects of IFITM3 in different infection modes. Purified and normalized virions from IFITM3-overexpressing or control cells were inoculated to Marc-145-Vector or Marc-145-IFITM3-flag cells, and then the GFP-PRRSV positive cell percentage under different infection modes was analyzed using FACS at different time points. Virions from Marc-145-Vector cells showed stronger infective activity when inoculated with either Marc-145-Vector or Marc-145-IFITM3-flag cells (Fig. 10A). Virions in the presence of IFITM3 resulted in the infection of fewer target cells overall compared with virions in the absence of IFITM3 (Fig. 10A), and a more potent inhibition was observed when both virions and target cells contained IFITM3 (Fig. 10A). Thus, whether IFITM3 impeded newly assembled IFITM3-containing virion attachment or entry into host cells was determined. Purified and normalized virions were inoculated to Marc-145 cells, and viral attachment and an internalization assay were performed by evaluating viral genome content. However, virions from IFITM3-overexpressing or control cells did not exhibit any differences in the ability of adsorption or entry into host cells (Fig. 10B and C). A previous study showed that IFITM3 incorporated into HIV virions could effectively impair virion spread in target cells (26), and for PRRSV, intercellular transmission was an important means for establishment of sustained and effective infection (41). This led us to ask whether IFITM3 exhibited its anti-PRRSV function by impairing intercellular virion spread.
To determine whether IFITM3 could restrain PRRSV intercellular spread during infection, we performed a dynamic analysis of PRRSV spread using a coculture system. GFP-PRRSV-positive cells sorted by flow cytometry were used as the donor cells in different coculture systems, whereas cells stably expressing a red fluorescent protein When ϳ60% of the cells were GFP-PRRSV positive, the cells were digested, sorted, and counted as donor cells to coculture with target cells to produce four coculture conditions. Images were taken every 5 min for 60 hpi. The area of GFPϩRFPϩ pixels was quantified and used as a marker of target cell infection. A total of three fields of view in two independent experiments were analyzed and plotted as the mean Ϯ the standard error of the mean; one representative experiment is shown. PRRSV, porcine reproductive and respiratory syndrome virus; IFITM, interferon-induced transmembrane; MOI, multiplicity of infection; RT-qPCR, reverse transcription-quantitative PCR; hpi, hours postinfection.
(RFP) marker (tomato red) were utilized as the target cells; the various combinations are illustrated in Fig. 10D. For the different coculture systems, target cells were plated in a 35-mm glass-bottom petri dish and visualized by time-lapse fluorescence microscopy with images acquired every 5 min for up to 60 hpi. Representative videos of PRRSV intercellular transmission in different coculture modes are presented in Movies S1 and S2. Live cell time-lapse analysis results showed that in the absence of IFITM3 in donor cells, PRRS virions efficiently spread to target cells, as shown by the presence of GFPϩRFPϩ cells, and eventually formed cluster-like cell clumps (Fig. 10E, upper panel and Fig. 10F), and a slight decrease (0.157-fold) of GFPϩRFPϩ double positive cells was observed when Marc-145-IFITM3-flag cells were utilized as target cells versus Marc-145-Vector cells as the target cells 48 hpi (Fig. 10E and F). In addition, the presence of IFITM3 in the donor cells further abolished the spread of virions, as a 0.77-fold decrease in the number of GFPϩRFPϩ Marc-145-IFITM3-flag target cells was observed at 48 hpi (Fig.  10F). Under these experimental conditions, the strongest inhibitory effect on PRRSV intercellular transmission was observed in the presence of IFITM3 in both donor and target cells (Fig. 10E and F). Real-time time-lapse microscopy analysis of PRRSV spread also showed that IFITM3 could effectively hamper PRRSV intercellular transmission (Fig.  10G).

DISCUSSION
Several studies have extensively explored the transport process of viruses following entry into host cells, and the endosome-lysosome pathway, although not the exclusive entry mechanism, is considered to be a classic early intracellular transport process. Receptor-mediated PRRSV entry into cells occurs through adsorption and endocytosis; however, its intracellular transport process is still unclear. Previous studies have suggested that PRRSV particles are trafficked through early endosomes, but not late endosomes or lysosomes, for productive infection (37). In the present study, it was shown that, similar to several other enveloped viruses, PRRSV particles were transported to early/late endosomes and lysosomes and then underwent membrane fusion to release genomic RNA and initiate their replication process. Since subcellular localization is coincident for both Marc-145 cells and PAMs, our results are likely to objectively reflect the intracellular transport process of PRRSV following entry into host cells. Although no significant differences were observed in restriction of PRRSV attachment and entry after overexpression of IFITM3, taking into consideration the PRRSV replication kinetics in IFITM3-overexpressing or control cells and the localization of IFITM3 at both endosomes and lysosomes, we speculate that IFITM3 is following the established paradigm of acting at the later stages of entry of PRRSV. Colocalization of IFITM3 with PRRSV particles and endosomes or lysosomes showed that PRRSV particles are likely cotrafficked with IFITM3-containing endosomes (Fig. 4), which is consistent with previous reports showing that the IFITM3-sensitive virus IAV was cotrafficked with IFITM3-positive endosomes during intracellular transportation (42). Virus restriction by IFITM3 was mediated partially by its ability to prevent virus-endosomal membrane fusion (also known as lipid mixing) (42). Furthermore, IFITM3-induced cholesterol accumulation in cellular compartments blocks virus entry mediated by lipid mixing (23), although a previous study also documented unimpeded IAV lipid mixing activity in IFITM3-expressing cells (38). According to the results of the present study, overexpression of IFITM3 resulted in notable accumulation of cholesterol in endosomes and lysosomes, and blockage of cholesterol efflux effectively impaired PRRSV membrane fusion (Fig. 6). Analysis of lipid dye dequenching showed significant differences in the rate of lipid mixing in IFITM3-overexpressing or control cells, in agreement with reduced IAV lipid dequenching in IFITM3-expressing cells (43). Thus, we postulate that PRRS virions are trapped in endosomes or lysosomes following endocytosis due to IFITM3-induced alterations of cholesterol at endosomes and lysosomes, which prevents viral fusion and, subsequently, genomic release into the host cell's cytoplasm.
Posttranslational modifications of IFITM3 have been shown to serve a crucial role in subcellular localization and antiviral function. S-palmitoylation of Cys-72 of human IFITM3 is essential for restriction of IAV and DENV (44,45). The addition of palmitoyl groups is associated with IFITM3 stability, since mutation of any of the cysteines to alanine resulted in a notable decrease in total IFITM3 expression and almost completely abolished the anti-PRRSV activity of IFITM3, thus explaining the importance of S-palmitoylation for the antiviral activity of the IFITM3. Interestingly, no changes were observed in IFITM3 subcellular localization when all cysteine residues were mutated, suggesting that S-palmitoylation of swine IFITM3 primarily maintains protein stability without affecting its subcellular localization. A previous study showed that S-palmitoylation and ubiquitination served different roles in suppressing IAV replication (24). The present discovery of IFITM3 ubiquitination and analysis of a ubiquitinationdeficient mutant on PRRSV infection showed that ubiquitination is a necessary posttranslational modification of IFITM3 and is essential for effective inhibition of PRRSV infection. Complete mutation of lysine to alanine of IFITM3 results in reduced localization with endosomal markers but not lysosomal markers (Fig. 7M); considering the almost complete abrogation of its antiviral activity, we speculate that appropriate subcellular localization of IFITM3 is necessary for its anti-PRRSV activity. In disagreement with a previous study (24), the effects of IFITM3 posttranslational modification on IAV replication, S-palmitoylation, and ubiquitination served a synergistic role in inhibiting PRRSV replication via different molecular mechanisms in the present study.
The results of the present study highlight IFITM3 as an anti-PRRSV factor that acts primarily through the sequestration of infective virions in endosomes or lysosomes. On the other hand, an additional mechanism of viral interference with which IFITMs could lead to the formation of virions with decreased infectivity has been previously reported in HIV and some other viruses (25,26,46,47). By utilizing overexpression or induction of IFITM3 in PRRSV permissive cells, we showed for the first time that IFITM3 was a PRRSV-associated host protein that can diminish intrinsic infectivity of progeny viral particles through wrapping as part of the virus envelope. Furthermore, knockdown of IFITM3 expression affected viral sensitivity to IFN-␣, suggesting the important role of IFITM3 in mediating the antiviral activity of IFN-␣. Previous reports suggested that incorporation of IFITM3 into HIV virions resulted in impaired envelope protein Env processing and incorporation into HIV-1 virions (40); however, no differential effects of swine IFITM3 on PRRSV envelop protein GP3 and GP5 processing and incorporation in PRRS virions was observed in the present study. IFITMs have been proposed to sterically rigidify membranes in which they are inserted, leading to fusion inhibition (38). The finding that IFITMs can interact with themselves strongly supports the above hypothesis, although the extent of this multimerization has not been fully examined (48). Whether these mechanisms of inhibition are at play in target cells and in virion particles remains to be determined. The present study expands our understanding of how IFITM proteins contribute to intrinsic immunity, and it could prove fruitful to study these functions in the context of other enveloped arteriviruses, which may reveal whether viruses sensitive to IFITM3 in the target cell are also susceptible to this donor cell-specific mode of antiviral function, that is, virion incorporation and decreased infectivity.
Nevertheless, it is still unclear whether the physical incorporation of IFITM3 into PRRSV particles is indispensable for effective anti-PRRSV activity or whether the abovedescribed two properties can be dissociated. However, we are more inclined to the view that IFITM3 exerts it antiviral effect through two different molecular mechanisms, since viral membrane fusion, which is essential for infection, was blocked during one viral replication cycle when WT PRRSV was utilized, and this process does not appear to be associated with viral assembly or production of new virions. Additionally, the most conspicuous anti-PRRSV effect was observed when IFITM3 was present in both target and donor cells. In this infection mode, although no marked difference was observed in absorption or entry capability of newly produced virions from IFITM3overexpressing or control cells, it is very likely that the reduced virion infectivity as well as the membrane fusion ability blocked by the presence of IFITM3 after entering the endosomes/lysosomes both served key roles. Identification of IFITM3 mutants that have lost their ability to be incorporated into viral particles may help us validate the above hypothesis. Given that the protection offered by antiviral factors is often incomplete, this dual mechanism of inhibition is likely to potentiate the leverage of IFITM3 on PRRSV replication.

MATERIALS AND METHODS
Cells, viruses, and reagents. PAMs were prepared from 4-to 6-week-old PRRSV-negative pigs as previously described (49). PAMs were maintained in RPMI 1640 medium (Gibco, Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS) (Biological Industries), 100 U/ml penicillin, and 100 g/ml streptomycin. Marc-145 and HEK293T cells were cultured in DMEM (Gibco, Thermo Fisher Scientific, Inc.) supplemented with 10% FBS. All cells were cultured at 37°C with humidity and 5% CO 2 . The animal experiments were performed in strict accordance with the guidelines of the Institutional Animal Care and Use Committee and approved by the Animal Care and Use Committee of Henan Agricultural University Zhengzhou (Henan, China).
Two highly pathogenic PRRSV strains, SD16 (GenBank accession no. JX087437.1) and JXA1 (GenBank accession no. EF112445.1), and PRRSV-2 prototype strain VR-2332 (GenBank accession no. EF536003.1) were used in the present study. Enhanced green fluorescent protein (EGFP)-PRRSV based on the genetic background of the SD16 strain that expressed EGFP as an additional open reading frame (ORF) (rHP-PRRSV/SD16/EGFP) was kindly provided by En-min Zhou at Northwest A&F University (50). Viruses were propagated and titrated in Marc-145 cells by calculating the median tissue culture infective dose as previously described (33). SD16 PRRSV was used for the majority of the experiments, and thus is indicated as PRRSV unless otherwise specified.
Overexpression and knockdown of IFITM3 in PRRSV permissive cells. Swine IFITM3 cDNA with an N-terminal flag tag was cloned into pTRIP-puro lentiviral expression vector to generate pTRIP-IFITM3flag. The recombinant plasmid was verified by sequencing. A blank pTRIP-puro vector was generally used as a control. pTRIP-IFITM3-flag expression vector (1.3 g) was cotransfected into HEK293T cells, with packaging vectors psPAX (1.8 g) and pMDG.2 (1.0 g) to produce pseudotyped lentiviral vectors using X-tremeGENE HP DNA transfection reagent (Roche Diagnostics). Supernatants containing the pseudotyped lentiviruses were collected 48 h posttransfection, followed by centrifuging at 1,000 rpm for 5 min at 4°C to remove cell debris.
To create cell lines stably expressing IFITM3, Marc-145 cells were infected with pseudovirions in the presence of 1 g/ml Polybrene (Sigma-Aldrich, Merck KGaA). A total of 48 h after infection, cells were selected in 10% FBS ϩ Dulbecco modified Eagle medium (DMEM) containing 8 g/ml puromycin (Sigma-Aldrich, Merck KGaA). For the production of red fluorescence cell lines stably expressing IFITM3, Marc-145-IFITM3-flag cells were first inoculated with pseudovirions expressing red fluorescence (packaged using pLenti-tomato-neo expression vector together with psPAX and pMDG.2) and further screened using 600 mg/ml neomycin, and the control cell lines were established as described above. Marc-145 cell lines with different mutant versions of IFITM3 (Cys¡Ala, CC71/72AA, C105A, CCC71/72/105AAA; Lys¡Ala, K24A, K83A, K88A, K104A, K(24)A, K(83)A, K(88)A, K(104)A, KKKK24/83/88/104AAAA) were constructed by cloning mutant genes into pTRIP-puro using overlap PCR with mutant primers, and all cell lines were selected using puromycin screening and identified by indirect immunofluorescence assay (IFA) and Western blotting. The sequences of the primers used in the present study are listed in Table 1.
Purification of PRRS virions. Marc-145-Vector or Marc-145-IFITM3-flag cells were infected with PRRSV at a multiplicity of infection (MOI) of 0.1, and supernatants were harvested 48 h postinfection (hpi) and purified by centrifugation (4,000 ϫ g, 4°C, 30 min). The supernatant was concentrated by ultracentrifugation (40,000 rpm, 3 h, 4°C; Beckman Coulter). The concentrated virus was then resuspended in phosphate-buffered saline (PBS; pH 7.4) and purified over a 30 to 60% sucrose gradient (40,000 rpm, 3 h, 4°C), and the banded virus was collected, diluted with PBS, pelleted (40,000 rpm, 3 h, 4°C), and resuspended in 1 ml PBS for further use. Taking into consideration that IFITM3 was reported to be associated with exosomes and given that exosomes may share numerous characteristics with virion particles (51), anti-CD63 immuno-magnetic beads (Invitrogen, Thermo Fisher Scientific, Inc.) were used to exclude free exosome contamination and further purify virions, according to the manufacturer's protocol.
Immunogold labeling of purified virions for transmission electron microscopy. Transmission electron microscopy was performed as previously described (52) with the following modifications: purified PRRSV particles were diluted 1:20 with PBS and adsorbed onto carbon-coated 200-mesh copper palladium grids. After washing with PBS for 5 min, the samples were blocked with 3% BSA in PBS for 1 h at room temperature (RT). Mouse anti-flag tag monoclonal antibody (Beyotime Institute of Biotechnology) was diluted in 1% BSA/PBS (1:500) and adsorbed to the grid for 1 h at RT. Following 3 washes with PBS, a goat anti-mouse IgG conjugated with 6-nm gold particles (Jackson ImmunoResearch Laboratories, Inc.) was added and incubated at RT for 1 h. After labeling, the grids were washed 3 times using PBS, and then the grids were negatively stained with aqueous 3% uranyl acetate solution for 15 sec. Images of stained virions were captured on a Hitachi H-7650 120 kV transmission electron microscope.
Cell-to-cell transmission assay. A cell coculture system was utilized to assess intercellular viral spread. Marc-145-Vector or Marc-145-IFITM3-flag cells were inoculated with 0.1 MOI GFP-PRRSV for 48 h to achieve ϳ60% GFP-PRRSV-positive cells. Cells were adequately digested with trypsin to ensure that a single cell suspension was obtained, and then the GFP-PRRSV-positive cells were separated using a FACSAria cell sorter and were used as the donor cells after counting using trypan blue staining. For the virus cell-to-cell transmission assay utilizing Marc-145-Vector-RFP cells as the target cells, 5 ϫ 10 5 donor cells were mixed with 5 ϫ 10 5 Marc-145-Vector-RFP target cells (ϳ1:1 donor:target) in 6-well plates. After coculturing for 2 h at 37°C, cells were adequately digested, and ϳ4 ϫ 10 5 target cells (red) were separated from donors by flow cytometry sorting and plated in 35-mm culture dishes and continued to incubate for 12 h at 37°C. Then time-lapse video imaging was performed. Transmission and fluorescence images were taken every 5 min for up to 60 hpi at 37°C with 5% CO 2 , using a Nikon Biostation IMQ. At 48 hpi, another batch of parallel samples was subjected to flow cytometry detection, and data were analyzed using FlowJo version 7.6 (FlowJo LLC). When Marc-145-IFITM3-flag-RFP cells were used as the donor cells, the other two coculture systems were similar to those described above.
Virion double fluorescence labeling and single virion imaging and analysis. PRRSV particles' lipid envelopes were labeled using DiOC18 (Thermo Fisher Scientific, Inc.) and R18 (Thermo Fisher Scientific, Inc.) as previously described (53,54) with some modifications. Purified PRRSV particles were resuspended in 1 ml PBS, followed by addition of DiOC18 and R18 to final concentrations of 0.2 and 0.4 M, respectively. The mixture was agitated vigorously and incubated in the dark for 1 h at RT and subsequently filtered through a 0.22-m-pore size filter (EMD Millipore) to remove any large lipid and/or virus aggregates. The labeled virions were purified from excess dyes through a NAP-5 gel filtration column (GE Healthcare) that was equilibrated with 50 mM HEPES (pH 7.4) and 145 mM NaCl solution. The virions were aliquoted into tubes and stored at Ϫ80°C until use.
For single virus imaging, Marc-145 cells were seeded onto 35-mm collagen-coated glass-bottom petri dishes and were prechilled on ice for 30 min and then incubated with labeled virus on ice for 30 min to allow virus to fully adsorb to the cell surface. The cells were washed 3 times using ice-cold PBS to remove any unbound viruses. Viral entry was initiated by adding 2 ml of prewarmed (37°C) 3% FBSϩDMEM, and then the cells were imaged immediately on the stage of a confocal microscope. Two ranges of fluorescence wavelength, 510 to 525 nm and 575 to 640 nm, were detected simultaneously. Every 15 sec, three Z-stacks separated by ϳ2 m were acquired to cover the thickness of cells using a Zeiss 63x/1.4NA oil lens objective. For the viral membrane fusion experiments, a DiOC18 dequenching (fusion) curve for the individual virion was acquired by tracking particles using the R18 (red) channel and the DiOC18 (green) channel. The fusion events were manually obtained from fluorescence intensity that showed at least 2-fold increases in DiOC18 fluorescence. The dequenching time (fusion time) was calculated as the time taken to reach an intensity plateau from the time of initial rise in intensity of DiOC18. To block PRRSV fusion, experiments were performed in medium supplemented with 70 mM NH 4 Cl (pH 7.6) or containing 200 nM of the endosomal acidification inhibitor BafA1. For quantitative analysis, PRRSV particle total fluorescence intensities were analyzed using ImageJ. Colocalization analysis. Marc-145 cells or PAMs were seeded on coverslips in 24-well plates at a density of 1 ϫ 10 4 cells/well 1 day prior to prechilling on ice for 30 min, followed by infecting with 0.1 MOI of PRRSV and further incubation on ice for 30 min. At 10, 30, and 60 min postinfection, cells were fixed with 4% paraformaldehyde for 10 min at RT, permeabilized with 0.1% Triton X-100 for 30 min, and blocked with 2% BSA for 1 h at RT. Cells were stained using mouse anti-PRRSV N antibody (1:300; produced in our lab) together with rabbit anti-Rab5 (1:1,000; ProteinTech Group, Inc.), -M6PR (1:1,000; ProteinTech Group, Inc.), or -LAMP2 antibody (1:1,000; ProteinTech Group, Inc.) for 1 h at RT. Then, cells were incubated with Alexa Fluor 594-conjugated goat anti-mouse IgG (H&L) and Alexa Fluor 488conjugated goat anti-rabbit IgG (H&L) (1:500; Abcam) for 1 h at RT in the dark. Coverslips were mounted on glass slides with DAPI counterstain (Sigma-Aldrich, Merck KGaA). Coverslips were imaged using a Zeiss LSM510 laser scanning inverted confocal microscope with a Zeiss 63x/1.4NA oil lens objective.
Measurement of viral particle infectivity. Viral infectivity assays were performed using purified, normalized viral preparations. Prior to the viral infectivity assay, a plasmid containing a 372-bp fragment of the PRRSV ORF7 sequence was utilized to create a standard curve to conduct absolute reverse transcription-quantitative PCR (RT-qPCR) to ascertain supernatant virus copies for PRRS virion normalization. To approximate a single cycle of PRRSV infection and facilitate detection, Marc-145 cells were inoculated with an MOI of 1.0 of purified and normalized GFP-PRRSV from supernatants of cells of the indicated treatment, and the percentage of infected cells was monitored at one viral replication cycle (ϳ10 h) by FACS to avoid multiple cycles of infection. Viral infection activity was expressed as the normalized proportion of GFP-PRRSV positive cells.
Statistical analysis. Statistical analysis was performed using GraphPad Prism (GraphPad Software, Inc.). Data are presented as the mean Ϯ standard deviation. Differences between two groups were assessed using an unpaired Student's t test, while a one-way analysis of variance (ANOVA) was used to compare differences between three or more groups. P Ͻ 0.05 was considered to indicate a statistically significant difference. All experiments were performed in triplicate.