Host Cell Entry of Respiratory Syncytial Virus Involves Macropinocytosis Followed by Proteolytic Activation of the F Protein

Respiratory Syncytial Virus (RSV) is a highly pathogenic member of the Paramyxoviridae that causes severe respiratory tract infections. Reports in the literature have indicated that to infect cells the incoming viruses either fuse their envelope directly with the plasma membrane or exploit clathrin-mediated endocytosis. To study the entry process in human tissue culture cells (HeLa, A549), we used fluorescence microscopy and developed quantitative, FACS-based assays to follow virus binding to cells, endocytosis, intracellular trafficking, membrane fusion, and infection. A variety of perturbants were employed to characterize the cellular processes involved. We found that immediately after binding to cells RSV activated a signaling cascade involving the EGF receptor, Cdc42, PAK1, and downstream effectors. This led to a series of dramatic actin rearrangements; the cells rounded up, plasma membrane blebs were formed, and there was a significant increase in fluid uptake. If these effects were inhibited using compounds targeting Na+/H+ exchangers, myosin II, PAK1, and other factors, no infection was observed. The RSV was rapidly and efficiently internalized by an actin-dependent process that had all hallmarks of macropinocytosis. Rather than fusing with the plasma membrane, the viruses thus entered Rab5-positive, fluid-filled macropinosomes, and fused with the membranes of these on the average 50 min after internalization. Rab5 was required for infection. To find an explanation for the endocytosis requirement, which is unusual among paramyxoviruses, we analyzed the fusion protein, F, and could show that, although already cleaved by a furin family protease once, it underwent a second, critical proteolytic cleavage after internalization. This cleavage by a furin-like protease removed a small peptide from the F1 subunits, and made the virus infectious.


Introduction
Human respiratory syncytial virus (RSV) belongs to the Paramyxoviridae, a family of enveloped viruses with a negativestranded RNA genome. It is a ubiquitous human pathogen that causes severe respiratory tract infections affecting mainly children and the elderly worldwide. Despite ongoing efforts, there are no available vaccines or treatments except passive immunoprophylaxis [1]. A better understanding of virus/host cell interactions is critical for the development of new therapeutic strategies.
RSV particles produced in tissue culture are heterogeneous in size and shape. Some are rounded with a diameter of 100-300 nm, others filamentous with a length up to 10 mm [2]. The nucleocapsid is helical and contains in addition to the RNA the nucleoprotein N, the viral polymerase L, its cofactor-phosphoprotein P, and the transcription processivity factor M2-1. The matrix protein M is believed to form a layer on the inside of the viral envelope [3]. The lipid envelope is derived from the plasma membrane (PM) of the infected host cell, and contains three viral glycoproteins; the major attachment protein G, the fusion protein F, and a small hydrophobic protein SH.
Cell attachment of RSV is mediated by G and F, which bind to cellular glycosaminoglycans [4]. That G and SH are not essential for replication in cell culture [5], indicates that the F protein can support both attachment and fusion. In vivo, RSV targets airway epithelial cells, and in the human mucociliary epithelium it infects ciliated cells from the apical surface [6].
Previous studies on RSV entry employing a lipid-dequenching assay suggested that RSV, as most other paramyxoviruses, fuses its membrane directly with the PM of target cells [7]. That RSV entry is pH-independent is consistent with this view [8]. On the other hand, Kolokoltsov and coworkers concluded, that RSV uses clathrin-mediated endocytosis (CME) to infect HeLa cells because a targeted siRNA screen revealed clathrin light chain, Eps-15, and AP-2 as important cellular factors in RSV infection [9]. In a recent publication, San-Juan-Vergara et al. [10] argued that in primary NHEB cells RSV entry is a two-step process; RSV docks to cholesterol-rich PM domains facilitating hemifusion between the viral envelope and the PM followed by endocytosis and complete fusion in endosomes.
To determine the pathway of RSV entry into HeLa and A549 cells, we developed quantitative fluorescence-activated cell sorting (FACS) assays and complemented them with confocal microscopy to monitor cell binding of RSV, endocytosis, fusion, and infection. We tested the effects of inhibitors and other perturbants. Our results indicated that RSV infected the cells by an endocytosismediated mechanism that fulfilled the criteria of macropinocytosis. After uptake into macropinosomes, a second proteolytic cleavage in F served as a 'cue' for penetration by membrane fusion.

Purified RSV is efficient in cell binding and infection
In our studies, we used a recombinant RSV strain called rgRSV that expresses GFP [11] enabling us to quantify infection by FACS. The virus was grown in HEp-2 cells, and to minimize exposure to broken cells, harvested from the cell supernatant before cytopathic effects were observed. The quality of virus purified by gradient centrifugation was confirmed by SDS-PAGE (Fig. 1A).
When the purified particles were examined by indirect immunofluorescence (IIF) using antibodies to the F and the N proteins, we found three different particle populations. Half of the particles represented intact virions because in addition to F (green) they contained N (red) (Fig. 1B). Of these, 30% also stained with phalloidin (blue, pseudocolored white) indicating the presence of actin filaments as previously reported (Fig. 1B arrowheads) [12]. The remaining particles constituted capsid-free envelopes (VLPs). They stained for F but not for N. Since we did not detect free capsids that would stain only for N or P (data not shown), we used the presence of the capsid antigens to distinguish between intact RSVs and VLPs.
When purified virus preparations were incubated with HeLa cells at 4uC, immunoblotting after SDS-PAGE showed that more than half of the input N and P associated with the cells indicating that RSV binding in the cold was efficient (Fig. 1C).
To measure infection, RSV was added to HeLa cells for 1 h and infection was continued for additional 3-8 h before measuring GFP expression by FACS (Fig. 1D). The fraction of cells

Author Summary
Respiratory Syncytial Virus (RSV) is a highly pathogenic paramyxovirus. We developed assays for RSV endocytosis, intracellular trafficking, membrane fusion, and infection. The results showed that RSV was rapidly and efficiently internalized, and that acid-independent membrane fusion occurred intracellularly after endocytosis. Cell biological studies demonstrated that endocytosis was macropinocytic, and that it was required for infection. The process involved activation of the EGF receptor and its downstream effectors including Cdc42, Pak1, and myosin II. RSV induced transient actin rearrangements accompanied by plasma membrane blebbing, elevated fluid uptake, and internalization of intact RSV particles into large macropinosomes. Expression of a dominant negative Rab5 mutant but not Rab7 decreased infection indicating that RSV penetration is intracellular, and takes place in Rab5 positive macropinosomes before fusion with endolysosomal compartments. The reason why RSV, unlike most paramyxoviruses, depended on endocytic entry was found to be the need for activation of the F protein by a second proteolytic cleavage. It occurred after endocytosis, and involved most likely a furin-like, vacuolar enzyme. . Gradient purifier RSV (,10 5 particles/50 ml) was resolved on the SDS-PAGE, followed by the blue silver gel staining. (B). After binding to a glass slide, purified RSV was stained with anti-F-AF488 (green), anti-N-AF594 (red), and phalloidin-AF647 (pseudocolored white). The particles (n = 457) were imaged with a confocal microscope and analyzed for colocalization by Imaris. Arrowheads show particles with all three stains. (C). Equal volumes of the virus input (moi 10), the cell bound virus lysates, and the unbound virus (sup) were resolved by a SDS-PAGE. Western blots were developed with anti-P or anti-N RSV specific antibody. (left) Representative western blots. (right) Densitometry quantification of the P-and N-protein bands intensities for the virus input and cell bound virus samples. (D). HeLa cells were infected with RSV moi (3-10) for 1 h at 37uC. Virus inoculum was replaced with medium and the infection was carried for indicated times. The percentage of infected cells expressing GFP was measured by FACS. doi:10.1371/journal.ppat.1003309.g001 expressing GFP increased with time and with increasing multiplicity of infection (moi). In cells infected at moi of 10, GFP expression was detected as early as 3 h post-infection (hpi) (28% GFP positive cells) and after 5 hpi 80% of the cells were infected. At a moi of 3, GFP expression was delayed by about 3 h.

RSV is endocytosed
To follow the fate of the cell-bound particles in the cold after warming to 37uC, IIF with anti-F and anti-N antibodies was used. Actin filaments were labeled with phalloidin to visualize cell boundaries. Confocal Z-stack image series in the orthogonal view revealed that after 30 min virus particles containing N and F were present not only on the cell surface but also deep inside the cytoplasm ( Fig. 2A). This indicated that viral particles and VLPs were endocytosed.
After binding in the cold, cell-associated viruses and VLPs can be removed from the cell surface by brief incubation with trypsin in the cold that does not affect cell attachment (Fig. 2B, 0 min) [10]. We found that when cells after virus binding in the cold were incubated at 37uC, an increasing fraction of the cell-associated particles became trypsin resistant (Fig. 2B, 15-60 min). Quantitation using spot detection software Imaris showed that after 60 min, 77% VLP and 70% RSV-containing spots were, in fact, resistant to trypsin (Fig. 2C). That the total number of RSV-and VLP-containing spots decreased over time was probably caused by the accumulation of multiple particles in common endocytic vacuoles that represented single spots. Of the anti-F stained spots, 47% stained for N indicating that they were intact viruses.

RSV endocytosis is followed by delayed intracellular fusion
To confirm that RSV was endocytosed in an intact form, it was important to determine whether the endocytosed particles also contained the viral lipid envelope. Purified RSV was therefore labeled with a lipophilic fluorescent dye, DiOC, which partitions into the viral membrane. It is fixable with formaldehyde, and can be quenched by the membrane-impermeable dye, trypan blue (TB) [13]. After labeling, 80% of re-purified particles contained detectable DiOC (data not shown).
When added to cells and incubated at 37uC, the RSV-DiOC particles were visible as discrete fluorescent spots, and of these some were quenched when TB was added (Fig. 3A). FACS analysis showed that, 50% of the fluorescence was resistant to TB after 30 min, and full resistance was reached in about 180 min ( Fig. 3B top), indicating that internalization of RSV and VLPs was rapid and complete. Importantly, when the intracellular accumulation of F and N proteins was measured in parallel ( Fig. 3B bottom), internalization of both antigens and DiOC (Fig. 3B top) followed similar kinetics.
To monitor fusion of RSV with cellular membranes, we used a method developed by Sakai and coworkers [14]. In this case, RSV was labeled with two fluorescent lipids, R18 (red) and DiOC (green). Concentrations were used at which the R18 quenches the fluorescence signal emitted by the DiOC. Therefore, when allowed to bind to cells and viewed live by confocal microscopy, the labeled viruses were initially all red (Fig. 3C, 0 min). However, after about 60 min at 37uC, yellow and green intracellular spots became apparent increasing in numbers over time, because after fusion, the two lipids were diluted out and the green fluorescence of DiOC was no longer quenched by R18 (Fig. 3C). Some of the spots showed a ring-like fluorescence indicating that the DiOC was localized in the limiting membrane of intracellular vacuoles.
Quantitative FACS analysis showed that the dequenching of DiOC became detectable already after 30 min at 37uC (Fig. 3D). It reached a half maximal level at 90 min, and plateaued after 240 min. Treatment of cells with TB during FACS analysis revealed that more than 90% of the fluorescent DiOC failed to be quenched by this membrane impermeable agent confirming that the DiOC was localized in intracellular organelles. From the time course, it was apparent that the fusion events occurred on the average 50 min after endocytosis.
Our interpretation of these results was that the virus particles and VLPs that bound to the cell surface were endocytosed. Endocytosis was rapid and efficient, and the internalized viruses accumulated in endocytic vacuoles. After a lag period, the viral envelopes underwent fusion with vacuolar membranes.
To bring infection into the picture, cells with virus bound in cold were transferred to 37uC. At indicated times, they were placed on ice, incubated with trypsin to strip away surfaceattached RSV, re-plated and incubated for 10 additional hours to allow infection to proceed and GFP to be expressed. FACS analysis demonstrated that in cultures that had been incubated at 37uC before trypsinization, the fraction of GFP-expressing cells increased with time. Maximum infection was reached within 180 min, and the half time was around 30 min (Fig. 3E). That the time course coincided perfectly with the time course of virus endocytosis (Fig. 3B) implied that productive infection depended on endocytosis.

Endocytosis and infection are clathrin-, dynamin-and pHindependent
If RSV entry and infection depended on endocytosis as indicated by our experiments, we expected perturbants that inhibit endocytosis to block internalization and infection. In the experiments that followed, endocytosis of RSV was quantified by measuring the amount of the incoming N protein that was trypsin resistant using FACS analysis 1 h after warming. Infection was scored as a percentage of cells expressing GFP 6 hpi. It is important to mention that the dose-dependence (Fig. S1) and toxicity (data not shown) of each inhibitor was carefully determined. For clarity for most of the inhibitors we will present data at a single concentration where we found a strong effect but low cytotoxicity.
RSV infection has been reported to be insensitive to an increase in endosomal pH [9]. This was confirmed by the lack of influence of bafilomycin A, ammonium chloride, and monensin on RSV infection (Fig. 4C). As expected all three agents blocked infection by SFV, which needs low endosomal pH to trigger fusion [19]. The small reduction in RSV infection observed for ammonium chloride and monensin may reflect the importance of a balanced vacuolar environment for productive RSV infection.
Taken together, the results indicated that RSV endocytosis and infection did not depend on CME nor did it require acidification.

Endocytosis is actin-dependent
When RSV was bound to HeLa cells in the cold and the cells warmed to 37uC, rapid and dramatic changes in cell shape and actin distribution were observed (Fig. 5A). The number of actin stress fibers decreased, the cells rounded up, and transient blebs filled with actin formed on the cell surface (Fig. 5A, 30 min). These changes were clearly visualized by live cell imaging (Movie S1).
The cell morphology and actin distribution returned to normal within 2 hpi. When the ratio of G (globular) and F (fibrous) actin was determined, it was found that 30 min after addition of RSV the ratio of G to F actin was 2:1 compared to 1:2 in control cells . HeLa cells without (top) or with bound RSV (moi ,0.5) at 4uC (bottom) were transferred for 30 min to 37uC before fixation. Cells were processed for confocal microscopy with anti-F-AF488 (green), anti-N-AF594 (red), and phalloidin-AF647 (blue), and Zstack image series acquired. The orthogonal views of Z-stack projections (pseudocolored white) were generated with ImageJ. (B). RSV (moi ,1) was bound to HeLa cells at 4uC, unbound virus was removed and cells transferred to 37uC. At indicated times cells were placed on ice and treated with PBS or trypsin for 3 min (PBS wash vs. trypsin treated). Samples were processed for confocal microscopy as in (A). (C). Quantification of the image series represented in (B). Analysis was performed with Imaris to detect spots with anti-F staining only (left) or anti-F and -N staining (right) at different times after warming the cells to 37uC. doi:10.1371/journal.ppat.1003309.g002 Disruption of actin filaments with cytochalasin D and latrunculin A as well as filament stabilization by jasplakinolide were found to strongly reduce RSV infection, whereas SFV infection was enhanced (Fig. 5C). RSV endocytosis was also significantly decreased (Fig. 5D). Inhibition of Cdc42 (pirl1), inhibited RSV infection effectively, while inhibitors of Rac1 (NSC23766), RhoA (CT04) and its effector ROCK (Y27632) had only a moderate effect (Fig. 5E). Both infection and endocytic uptake were reduced when some of Cdc42's downstream effectors were inhibited including PAK1 (IPA-3), N-Wasp (wiskostatin), and moderately when Arp2/3 (CK-869) was targeted (Fig. 5E, F). Nocodazole and taxol that interfere with microtubules had no effect on RSV or SFV infection (Fig. 5D).
These results demonstrated that actin and its regulators played a critical role during RSV endocytosis and infection. F-actin was transiently depolymerized, resulting in the formation of blebs. In addition, Cdc42, PAK1, and N-Wasp were required for RSV internalization and infection.

RSV induces fluid phase uptake
The formation of blebs, the involvement of actin, and the role of Cdc42 and PAK1 suggested that infectious entry of RSV occurred by macropinocytosis as recently shown for several other viruses [for reviews see [20,21]]. One of the characteristic features of macropinocytosis is an elevation in the uptake of extracellular fluid [22]. Indeed, when serum-starved HeLa cells were exposed to RSV, we observed that the internalization of 10 kDa dextran-AF488, a soluble, fluorescent tracer added to the medium, increased by 50% and 120% at low and high moi, respectively, over mock treated cells (Fig. 6A). The elevation was significantly higher than in serum-stimulated cells. IIF showed that majority of endocytosed viruses stained by anti-F and -P antibodies were localized in large, dextran-AF488 filled, intracellular vacuoles ( Fig. 6B).
Macropinosome formation requires Na + /H + exchanger (NHE) activity to modulate Rho GTPases at the PM [23]. Inhibition of NHE by EIPA (an amiloride derivative) has become one of the diagnostic criteria for macropinocytosis. EIPA inhibited RSV internalization and infection by 90% (Fig. 6C). In addition, pretreatment of cells with EIPA blocked the increase in fluid phase uptake induced by RSV (Fig. 6D). Taken together, these results demonstrated that RSV induces macropinocytosis and uses it for virus endocytosis and infection.

Cellular regulators of macropinocytosis
Macropinocytosis is usually initiated by activation of receptor tyrosine kinases (RTKs) or integrins, followed by the activation of a spectrum of cellular signaling factors [24]. Accordingly, we found that RSV infection was significantly decreased by two broad range protein kinase inhibitors; staurosporine (Ser/Thr kinases) and a multi-target protein tyrosine kinase inhibitor, genistein ( Fig. 7C, D).
To test whether RTKs were involved, we used a human phospho-RTK array comprising antibodies against 42 different phosphorylated RTKs. Lysates from cells exposed to RSV for 15 min, and lysates from mock-treated control cells were used as probes. The epidermal growth factor receptor (EGFR) was the only RTK for which activation was detected; a five-fold increase in phosphorylation compared to control (Fig. 7A).
When the EGFR was depleted using siRNA, greater than 50% reduction in infection was observed (Fig. 7B). We found, moreover, that EGFR inhibitors (CAS879127-07-8, iressa) significantly decreased RSV infection (Fig. 7C). Inhibition of PI3K (wortmannin, Ly294002, PI-103), a downstream effector of EGFR, also reduced infection (Fig. 7C). EGFR inhibitors had little effect on SFV. That the PI3K inhibitors boosted SFV infection was consistent with a distinct entry mechanism for this virus. In addition, inhibition of PKC (rottlerin, calphostin C) decreased RSV (Fig. 7D). Although the effect on SFV was smaller, it suggested a role for PKC in the entry of both viruses. Finally, since non-muscle myosin II is thought to mediate closure of macropinosomes, we tested the effects of a myosin II inhibitor (blebbistatin), and a myosin light chain kinase inhibitor, (ML-7)   . Serum starved HeLa cells were incubated with 20% FCS or purified RSV (moi ,10, 30), at 4uC for 1 h. The inoculum was replaced with medium containing 10 kDa dextran-AF488 and transferred for 15 min to 37uC before fixation. The MFI of AF-488 measured by FACS. (B). Purified RSV (moi ,10) was bound at 4uC to serum-starved HeLa cells. The input virus was replaced with medium containing 10 kDa dextran-AF488 (green) and transferred to 37uC. At indicated times cells were fixed, permeabilized and stained with anti-F-AF594 (red) and anti-P-AF647 (blue) antibody. Images represent a Z-stack projection acquired with the same confocal microscope settings. (C). HeLa cells were pretreated with solvent (MOCK) or EIPA at indicated concentration. (left) Cells were infected with RSV (moi ,3) at 37uC for 6 hours before FACS analysis of GFP expressing cells. (right) RSV (moi ,3) was bound to the cells at 4uC followed by 1 h of internalization at 37uC. Cells were trypsinized, fixed and stained with anti-N-AF488 antibody, and the MFI of AF-488 measured by FACS. (D). Serum starved HeLa cells were pretreated with EIPA at indicated concentration and incubated with purified RSV (moi , 30) or no virus control at 4uC. The inoculum was replaced with medium containing 10 kDa dextran-AF488 and EIPA and transferred for 15 min to 37uC. Cells were fixed and the MFI of AF-488 measured by FACS. doi:10.1371/journal.ppat.1003309.g006 [25]. Both reduced RSV infection with little effect on SFV (Fig. 7E).
All the inhibitors that decreased infection also reduced RSV endocytosis (Fig. 7C-E bottom). Depending on the compound, RSV internalization was reduced by 60-90%. None of the inhibitors affected RSV cell binding (Fig. S2). Thus, we concluded that infectious RSV cell entry and endocytosis were associated with activation of EGFR and its downstream signaling partners including PI3K and PKC. Combined with the requirement for myosin II, these findings were consistent with productive RSV internalization by macropinocytosis.
In addition, we performed a series of experiments in A549 cells (Fig. S3). They revealed changes in actin morphology and polymerization after addition of RSV, and a role of EGFR, NHE, Cdc42, Pak1, and other factors similar to HeLa cells in RSV infection. That internalization and infection were clearly dependent on the same cellular processes and factors in A549 cells indicated that entry by macropinocytosis was not HeLa cell specific.

Intracellular trafficking and role of Rabs
Intracellular trafficking of macropinosomes is not well characterized, but it has been shown that like endosomes, they acidify and acquire Rab5 followed by Rab7 before fusing with endolysosomes [26]. Wild type (WT) GFP-Rab5 and GFP-Rab7 as well as various constitutively active (C/A) and dominant negative (D/N) mutants of the Rabs were transiently expressed in HeLa cells. After 15 min post warming, we observed that some of the incoming RSV colocalized with GFP-Rab5 WT positive vacuoles (Fig. 8A). Colocalization was even more evident in cells expressing the C/A GFP-Rab5 mutant Q79L that exhibits enlarged Rab5-positive vacuoles that fail to undergo further maturation [27]. There was no detectable colocalization with Rab7 at this time. After 120 min, some colocalization with GFP-Rab5 WT was still observed. In cells expressing the D/N GFP-Rab5 S34N, we noted accumulation of F-and P-stained particles inside large vacuoles in the perinuclear space. Colocalization of RSV with GFP-Rab7 WT was also detected.
To determine whether Rab5 and Rab7 played a role in infection, we infected cells expressing GFP-tagged constructs of Rab5, Rab7, and their mutants with RSV-A2. To determine the fraction of infected cells among cells expressing GFP-tagged Rabs, we stained the cells with anti-N-AF647. FACS analysis revealed that the D/N -GFP Rab5 (S34N) was the only Rab construct that caused a significant decrease in RSV infection when overexpressed (Fig. 8B). We confirmed this result by imaging of rrRSV expressing red fluorescent protein (Fig. S4).
Together with our imaging data, these results indicated that RSV depends on Rab5 GTP for infection but does not require Rab7. Infectious penetration is thus likely to be determined during early stages of macropinosome maturation. It is noteworthy that expression of the C/A Rab5 (Q71L), which is known to generate enlarged Rab5-containing endosomes and prevent endosome maturation and trafficking to lysosomes [27], did not affect infection. Pretreatment of cells with PIKfyve inhibitor (CAS 371942-96-7) had no effect on the RSV infection while SFV infection was decreased by 50% (Fig. 8C). By generating PtdIns(3,5)P2, PIKfyve is involved in the maturation of endosomes and macropinosomes [28]. This suggested that full maturation of macropinosomes was not required for RSV.
These results demonstrated that after pinching off from the PM, macropinosomes containing RSV acquired Rab5 and later Rab7. Maturation of macropinosomes involving Rab5 was evidently a critical step in infection, whereas later stages in maturation coordinated by Rab7 and PIKfyve were not essential.

Post-endocytic cleavage of F
Since acidification of macropinosomes was not needed for infection, we speculated that RSV required some other intracellular cue to trigger fusion. The F is unique among paramyxovirus fusion proteins in having two cleavage sites for furin-like proteases generating in addition to F1 and F2 (50 and 20 kDa, respectively) a soluble 27 amino acid (aa) peptide (p27) [29,30]. The p27 peptide is located between F2 and F1 N-terminal to the fusion peptide in F1 (Fig. 9D). We hypothesized that removal of this peptide after endocytosis might be required to activate the F protein.
Experiments in which cells were pretreated with a membrane permeable furin inhibitor, dec-RVKR-CMK, prior to addition of RSV indicated that a protease was indeed involved. Dec-RVKR-CMK treatment reduced infection by about 80% (Fig. 9A). That a membrane impermeable furin inhibitor, a-PDX, had no effect on infection suggested that the activating proteolysis did not occur on the cell surface. A broad range protease inhibitor leupeptin (serine, cysteine, threonine proteases) caused only a slight increase in infection.
That the inhibition of infection by dec-RVKR-CMK involved the F protein, was confirmed using a recombinant virus strain (RSVDSHDG) that lacks the SH and G glycoproteins [5]. Infection by this mutant virus was also blocked by dec-RVKR-CMK (Fig. 9A). When dec-RVKR-CMK inhibitor was applied 1 h after the virus inoculation, there was no inhibition indicating that the critical proteolytic step coincided with entry (Fig. 9B). Fusion assays with R18/DiOC labeled RSV and RSVDSHDG revealed that dec-RVKR-CMK impaired viral fusion (Fig. 9C). Its inhibitory effect was comparable to the effect of EIPA, which blocked virus internalization (Fig. 6C).
To confirm the presence of the p27 peptide in the purified virus, we used a targeted mass spectrometry approach based on selected reaction monitoring (SRM) (Fig. 9E). As a negative control, we analyzed HEp-2 cells extracts used to produce the virus. In trypsin digested virus preparations, we detected 2 peptides corresponding aa 113-131 and 109-123 of F protein both spanning p27 peptide (Fig. 9D). Neither peptide was present in HEp-2 control samples. SRM transitions of the targeted peptides are included as supporting information in Table S1.
Finally, SDS-PAGE was used to monitor changes in the F protein itself during entry. Blotting with an anti-F1 antibody revealed a protein band in the isolated virus migrating with a molecular weight of 50 kDa, confirming that the F protein had been cleaved at least once already in the producer cells (Fig. 9F). When virus bound to cells in the cold were allowed cell enter for 1.5 h at 37uC, the mobility of the cell-associated F1 became faster (48 kDa) indicating that further processing had occurred. The reduction in size of around 2 kDa was consistent with the loss of p27 at the N terminus of F1. Importantly, in cells pretreated with dec-RVKR-CMK, the processing of the 50 kDa F1 protein did not occur. As expected, a-PDX and leupeptin did not influence the processing step. When the time course of F1 processing was followed, we observed that some processed F1 was detectable already 15 min after cell warming (Fig. 9G). It peaked at 60 min when the precursor was fully consumed. That the amount of F1 protein gradually decreased at later time points was probably due to lysosomal degradation explaining the decrease in band intensity in the endocytosis lanes in Fig. 9F.  . HeLa cells were pretreated with dec-RVKR-CMK, a -PDX or leupeptin at indicated concentration for 1 h before experiment and inhibitors were continuously present during following steps of the experiment. Cells were infected with RSV or RSVDGDSH for 6 h before FACS analysis of GFP expressing cells. (B). HeLa cells were infected with RSV or RSVDGDSH for 1 h. Virus inoculum was replaced with medium containing 100 mM dec-RVKR-CMK and incubated for 6 before FACS analysis of GFP expressing cells. (C). Versene detached HeLa cells were pretreated with solvent (MOCK), dec-RVKR-CMK or EIPA at indicated concentrations and inhibitors were continuously present during following steps of the experiment. RSV-R18/DiOC or RSVDGDSH-R18/DiOC (moi ,5) was bound to cells at 4uC. Unbound virus was removed and cells were incubated at 37uC for 2 h in the presence of inhibitor. Cells were fixed and the MFI of DiOC fluorescence measured by FACS and normalized to mock no inhibitor controls. (D). RSV F protein (574 aa) is proteolytically processed by furin like protease at the two sites (A aa-106 and B aa-136) to generate disulfide bonds linked F1+F2 and small peptide p27 (aa sequence depicted above). At the N-terminus of F1 is a FP (fusion peptide) and at the C-terminus TM (transmembrane domain), numbers indicate aa position and red underlines specify peptide sequences detected in mass spectrometry. (E). Proteomic analysis of HEp-2 cells and purified RSV particles. The N-terminal sequence of the p27 peptide (F protein) was quantified by a targeted mass spectrometry based on the selected reaction monitoring (SRM). Representative SRM peaks of peptides (left) FMNYTLNNAKKTNVTLSK 3+ and (right) ELPRFMNYTLNNAK 3+ peptides, corresponding to the aa 113-131 and 109-123 of F protein, respectively. Different SRM transitions for a peptide shown in different colors (see supporting information Table S1). The bar graphs show the results of the targeted peptide quantitation, presented as the sum of the areas of all the SRM peaks for a given peptide. Where no peptide peak was detectable, noise values were reported as a reference. RT retention time, and Cps counts per second. (F). HeLa cells were pretreated or MOCK treated with dec-RVKR-CMK, a-PDX, or leupeptin at indicated concentration. RSV (input control in the first line) was bound for 1 h in cold (B-binding) or after washaway unbound virus was internalized for 1.5 h at 37uC before processing (I-internalized). Lysed cell samples were resolved by SDS-PAGE and blots were probed with anti-F1 antibody. (G). RSV was bound for 1 h in cold to HeLa cells; unbound virus wash away and cells were placed at 37uC for indicated times before, lysis, SDS-PAGE, and processing for western blot probed with anti-F1 antibody. doi:10.1371/journal.ppat.1003309.g009 These results indicated that to become fusion competent and infectious, the F1 protein underwent a second, highly efficient cleavage by a furin-like convertase in an endocytic compartment. The time course indicated that the processing of F1 occurred soon after endocytosis preceding fusion by about 30 min.

RSV enters polarized bronchial epithelial cells by macropinocytosis
To address whether our results applied to cell types infected by RSV in vivo, we tested polarized epithelial cells 16HBE14o obtained from human bronchial biopsies [31]. After 9 days in culture, the distribution of the tight junction marker, ZO-1, showed that the cells had reached a polarized phenotype (Fig. 10A). After making certain that RSV could infect 16HBE14o cells from the apical side (Fig. 10A), we tested the effects of nine diagnostic inhibitors previously used in HeLa cell experiments (Fig. 4B, 5CE, 6C, 7C, 9A). They inhibited dynamin, macropinocytosis, and furin proteases. RSV infection was quantified by an image-based approach that detected the fraction of GFP-expressing cells.
In agreement with our findings in HeLa cells, inhibition of dynamin by dynasore had no effect on RSV infection in 16HBE14o cells; it even boosted infection. EIPA and seven other inhibitors of macropinocytosis decreased infection in a dosedependent fashion indicating that macropinocytosis was involved in entry. The need for F1 processing was confirmed by dec-RVKR-CMK, which was found to decrease infection by as much as 95%.

Discussion
Paramyxoviruses are generally thought to infect cells by fusing directly with the PM [32,33]. That paramyxovirus particles can also be endocytosed is, however, also clear. This has most recently been documented for Sendai, Nipah, RSV, Newcastle Disease viruses and for a lentivirus vector pseudotyped with measles virus glycoproteins [9,34,35,36,37,38]. Which of the two pathways -fusion at PM or fusion after endocytosis -leads to infection is not clear.
In our experiments, we found that intact RSV was rapidly and efficiently endocytosed with capsid, glycoproteins, and the lipid envelope intact. The RSV and capsid-free VLPs accumulated within cytoplasmic vacuoles with a half time of about 30 min followed by fusion in the vacuoles with the half time of around 90 min. A sensitive fusion assay using R18/DiOC-labeled fluorescent viruses showed that fusion occurred intracellularly. No fusion of viruses with the PM was detected, and no formation of syncytia by fusion-from-without was observed even after exposing cells to high moi (data not shown). The significant delay between RSV internalization and fusion could at least in part be explained by the requirement for post-endocytic cleavage of F protein.
Perturbations that inhibited endocytic uptake caused a dramatic reduction in infection confirming a role for endocytosis in infection. EIPA inhibited both endocytosis and infection by 95%, and a similar level of inhibition was observed for agents that interfered with actin dynamics, a variety of kinases, and myosin II. The inhibitory effect of Rab5 D/N expression was also consistent with a role for endocytosis in infectious entry. Proteolytic activation of the F protein necessary for fusion and infection occurred in intracellular compartments. We concluded that RSV and VLPs were efficiently endocytosed, that penetration by membrane fusion occurred in endocytic vacuoles, and that at least 90% of infection was caused by endocytosed viruses.
That the endocytic mechanism responsible for the entry was macropinocytic was demonstrated by the following observations:  observations satisfied all the main criteria currently used to define macropinocytosis [20,21].
When inhibitor studies were performed using polarized physiologically relevant epithelial cells (human bronchial epithelium cells, 16HBE14o), infectious entry of RSV was found to depend on the actin cytoskeleton, on cell signaling, and on a furinlike protease activity as also observed in HeLa cells. The results indicated that infection of these polarized epithelial cells monolayers derived from human bronchial tissue involved macropinocytosis and proteolytic activation of the F protein.
Macropinocytosis is a clathrin-independent mechanism for the uptake of fluid and cell-associated particles within large, uncoated vesicles formed at the PM [24]. In most cell types, it is transiently induced by the activation of RTKs and downstream signaling factors [39]. In recent years, several viruses have been shown to use it for infectious cell entry. As recently reviewed [21], the bestdescribed examples include large viruses such as vaccinia, Ebola, adeno 35, and Kaposi sarcoma-associated viruses. Interestingly, Nipah virus, a paramyxovirus of the Henipavirus subfamily, also belongs to this group. It uses EphrinB2A as a receptor, the phosphorylation of which is required for macropinocytic internalization and infection in CHO-K1 and VeroE6 cells [38].
We found that EGFR phosphorylation was activated by RSV, and that inhibitors such as iressa targeting this receptor blocked endocytosis and infection. It is noteworthy that iressa was only inhibitory when present during the first hour of virus cell contact confirming that its effect was entry-specific (data not shown). When the EGFR was depleted using siRNA, infection decreased only by 50% suggesting that other RTKs may be able to compensate in long-term experiments.
Downstream effectors of EGFR such as PI3K and PKC were also important for RSV endocytosis and infection. Infection has previously been shown to promote cell survival mediated by PI3K/NFkB. In A549 cells, PI3K activation and phosphorylation of its effector Akt occurs within 30 min after RSV addition [40]. Interestingly, it has also been demonstrated that RSV binding to NHEB cells induces PKC-a phosphorylation and translocation to the PM, while inhibition of PKC-a, as confirmed here, blocks RSV uptake and infection [41].
Our results contradicted a previous report proposing that RSV entry in HeLa cells occurs by CME [9]. The authors based their interpretation on hits such as clathrin and associated proteins in a targeted siRNA silencing screen against factors involved in endocytosis. (The hits also included actin modulators such as PAK1 but their function in entry was not addressed). However, since the read-out was infection after 20 h, a role for CME in postendocytic steps in the RSV infectious cycle could not be excluded. In our experiments, we did not observe inhibition of RSV endocytosis or infection by five different agents that block CME: chlorpromazine, dynasore, pitstop-2, dyngo-4a, and dynol-34-2. Importantly these agents efficiently inhibited SFV, a virus that enters via CME. That dynasore fails to inhibit RSV infection was also recently reported by others [10].
While macropinosomes are still poorly characterized, there is evidence that they undergo a maturation process similar to that of endosomes involving acidification, association with Rab5 and Rab7, and fusion with late endosomes or endolysosomes [42]. We noted that some of the vacuoles containing RSV were in fact Rab5-and later Rab7-positive. Over-expression of a D/N Rab5 mutant inhibited infection suggesting that RSV penetration required passage through 'early' macropinosomes that contained Rab5. The lack of inhibition by Rab7 mutants, a PIKfyve inhibitor, and nocodazole, all known to inhibit vacuolar maturation, implied that macropinosome maturation beyond the Rab5 positive stages was not necessary.
Finally, our results provided a likely molecular explanation for the endocytosis requirement exhibited by RSV. Unlike other paramyxoviruses, the F protein in RSV has two activating cleavage sites [29]. Our mass spectroscopy analysis and western blots showed that while F in the isolated virus had been cleaved in the A-site (RARR) generating F2 and F1, it had not been cleaved at the more C-terminal B site (KKRKR). The second cleavage occurred after endocytosis. Inhibition of the second cleavage by dec-RVKR-CMK inhibited RSV fusion and infection. That dec-RVKR-CMK is a furin inhibitor suggested that the protease in question belonged to the furin family of convertases. The enzyme was evidently acid-independent, and active in early Rab5 -positive macropinosomes. Cleavage at the B site was most likely important because after removal the p27 peptide 'cap' from the N-terminus, the hydrophobic fusion peptide is rendered the most N-terminal sequence in F1. In other class I viral fusion proteins, including other paramyxovirus F proteins, the fusion sequence is invariably N-terminal [43].
In conclusion, we confirmed that RSV requires two cleavages in its F protein for infectivity and showed that the second cleavage occurs during cells entry. Infectious entry depends on endocytosis, which the virus induces by transiently activating macropinocytosis. The virus most likely meets the enzyme that generates the second cleavage in Rab5 positive macropinosomes, and fusion occurs after some delay in these vacuoles. In this respect, the virus resembles Ebola and SARS viruses, the fusion proteins of which are also activated within endocytic vacuoles by proteases [44,45]. It is interesting to note that the F of Nipah virus, which has a single monobasic cleavage site in its F, is activated after endocytic uptake by cathepsin [46,47]. Inhibitors of cathepsins block infection, and cathepsin double knock-out cells are not infected. The infectious entry of other paramyxoviruses (and other viruses that have pHindependent membrane fusion) may thus be endocytosis-dependent and the mechanisms more complex than previously assumed.
For RSV, it will now be important to analyze the molecular features of the entry process in more detail, to identify the protease(s), and to determine whether the intracellular route is relevant also in vivo. Being inducible and highly regulated, the macropinocytic process may prove more amenable to inhibition than other endocytic mechanisms, and therefore more easily targeted by therapeutics.
All RSV stocks were titered by infecting HEp-2 cells with serial dilutions of the virus in 96 well plates. Infection was allowed to proceed for 18-22 h at 37uC. Fixed cells were assessed by microscopy for GFP expression, or stained with RSV anti-N antibody (AF-488) to detect infected cells of rgRSV or RSV-A2 respectively. Protein assay (Bio-Rad) was used to measure the amount of protein.

Pharmacological inhibitors treatment
When the influence of pharmacological inhibitors was tested, cells were preincubated with a medium containing inhibitors for 30 min at 37uC before virus binding or infection (except Rho GTPase and protease inhibitors that required 1-5 h of preincubation, and pitstop-2 which was preincubated for 10 min). Inhibitory compounds at indicated concentration were continuously present during all of the steps of infection and internalization experiments.

RSV DiOC labeling and endocytosis assay
DiOC (6 mg) was added to freshly purified RSV (25 mg) in 1 ml of HBSS-Hepes buffer and incubated at room temperature while gently shaking for 1 h. To remove unincorporated dye, virus was filtered through 0.45 mm syringe filter (Millipore) and snap frozen as 100 ml aliquots, stored in 280uC, and used within 2 weeks. Labeled RSV-DiOC was titered in HEp-2 cells as described above.
For the RSV-DiOC endocytosis assay, cells were detached with versene solution (0.53 mM EDTA, pH 8.0), washed with PBS and chilled on ice. Purified RSV-DiOC, diluted in DMEM-Hepes was bound to the cells on ice for 1 h. The inoculum was washed away with PBS; the cells were resuspended in complete medium and transferred to 37uC. After desired times, the cells were fixed in 4% formaldehyde, washed, and resuspended in FACS buffer (PBS, 0.2% FCS, 5 mM EDTA). For the FACS data acquisition (BD Biosciences, Canto II) samples were divided into two. One sample was acquired directly and the other after addition of trypan blue to a final concentration of 0.01% (Invitrogen).
Alternatively, after internalization cells were treated with 0.5% trypsin for 10 min on ice. Than washed with PBS, fixed in 4% formaldehyde, permeabilized with 0.01% Triton X-100, stained with anti-F or anti-N antibody at 4uC over night followed by the AF-647 secondary antibody staining. Mean fluorescence intensity measured by flow cytometry (BD Biosciences, Canto II) was normalized to the mock control without bound virus.

RSV R18/DiOC labeling and fusion assay
Purified RSV (25 mg) was resuspended in 0.5 ml of HBSS-Hepes buffer before adding of DiOC (3.3 mM) and R18 (6.7 mM) mixture. Labeling was performed for 1 h at room temperature while gently shaking. To remove unincorporated dye, virus was filtered through 0.45 mm syringe filter (Millipore) and used freshly for the fusion assay.
For the RSV-R18/DiOC fusion assay, cells were detached with versene solution, washed with PBS and chilled on ice. Purified RSV-R18/DiOC, diluted in DMEM-Hepes was bound to the cells on ice for 1 h. The inoculum was washed away with PBS; the cells were resuspended in complete medium and transferred to 37uC. After desired times, the cells were fixed in 4% formaldehyde, washed, and resuspended in FACS buffer. For the FACS data acquisition (BD Biosciences, Canto II) samples were divided into two. One sample was analyzed directly and the other after addition of trypan blue to a final concentration of 0.01% (Invitrogen).

FACS based RSV internalization assay
RSV (moi ,3) was bound to subconfluent cells for 1 h at 4uC. Than cells were washed with PBS, supplemented with medium containing 1 mM cycloheximide and incubated for 1 h at 37uC. Cells were moved on ice and treated with ice cold 0.5% trypsin for 10-20 min, washed with PBS, fixed in 4% formaldehyde, permeabilized with 0.1% Triton X-100 and stained with anti-N antibody over night at 4uC followed by the goat anti-mouse AF staining. The MFI (mean fluorescence intensity) of AF staining determined by FACS (BD Bioscience, Canto II) and analyzed by For the Trf-AF488 (2 mg/ml) internalization was performed in parallel in identical conditions with the exception that uptake was carried for 20 min.

FACS based RSV binding assay
Adherent cells were detached with versene solution, washed with PBS, chilled on ice, and pelleted. Cell pellets were resuspended in DMEM-Hepes, containing the virus inoculum, and incubated 1 h at 4uC. When the effect of inhibitors was tested, cells were preincubated with media containing inhibitor for 30 min, and inhibitor was continuously present at each step of the experiments. Cells were washed with PBS to remove unbound virus, fixed with 4% formaldehyde, permeabilized with 0.01% Triton X-100, and stained with RSV anti-F over night at 4uC, followed by the secondary antibody labeling. Samples were reconstituted in FACS buffer and analyzed by flow cytometry (BD Biosciences, Canto II). Mean fluorescence intensity values were normalized to the mock control without bound virus.

Quantification of the RSV binding by western blot
Cells were detached with versene solution, washed with PBS, chilled on ice, and pelleted. Virus input (purified virus stock (moi ,10) diluted in DMEM-Hepes) was divided in two portions. One was mixed with the cells and the other left for further analysis as virus input. Virus was incubated with the cells for 1 h at 4uC. The cells were pelleted by centrifugation and the supernatant collected as an unbound viruses sample. The cell pellets were washed in PBS and resuspended in DMEM-Hepes of equal volume to the collected supernatant. The SDS-PAGE loading buffer was added to all samples, followed by denaturation for 10 min at 95uC. An equal volume of each sample was separated by the SDS-PAGE and subjected to western blot with anti-P or anti-N antibody. Western blots were quantified by densitometry with QuantityOne software (Bio-Rad).

Fluid phase uptake
Subconfluent cells seeded in 24 well plates were starved over night with serum free DMEM-Hepes medium. Purified RSV was bound to pre-chilled cells for 1 h at 4uC. Virus inoculum was washed and cells were pulsed with the serum free DMEM-Hepes medium containing 10 kDa dextran-AF488 (0.5 mg/ml) for 15 or 60 min at 37uC. The cells were washed with 10 mM NaOAc, 50 mM NaCl pH 5.5 followed by PBS wash. The cells were trypsinized, fixed in 4% formaldehyde and analyzed by flow cytometry (BD Biosciences, Canto II). For confocal microscopy, cells were fixed in 4% formaldehyde, permeabilized with 0.1% Triton X-100, and stained with RSV specific antibodies. Images were acquired with Zeiss LSM510 laser scanning confocal microscope.

Transfection and transient expression
For transfection, cells were trypsinized, pelleted, and electroporated with 2 mg of plasmid DNA using Amaxa (Nucleofactor solution R, program I-13). Cells were seeded on 12 mm cover glass in 24 well plates for imaging or in the 6 well plates for FACS analysis. Experiments were performed after 12 h of transient expression.

F/G actin extraction
The experiment was performed according to the protocol provided by the F/G actin in vitro assay kit producer (Cytoskeleton Inc. cat # BK037). In brief, subconfluent cells in 3.5 cm dishes were inoculated with purified rgRSV (moi 30) for 30 or 120 min at 37uC. Cells were washed, lysed, and clarified by centrifugation. Supernatants were centrifuged (53000 rpm, 1 h, 37uC, TLA120.2 rotor, Beckman Optima TLX ultracentrifuge), the resulting supernatants were collected (G-actin), and the pellets (F-actin) were resuspended in equal to supernatant volume of water containing actin depolymerizing reagent provided in the kit. Equal volume of each sample was resolved by SDS-PAGE and western blots were developed with an anti-actin antibody. To measure ratio of the G and F actin western blots were analyzed by densitometry with QuantityOne software (Bio-Rad).

Image-based infection analysis
For the infection assays cells were plated in optical bottom 96 well plates. After infection with rgRSV for 18-20 hours, the cells were fixed with 4% formaldehyde and counterstained with DAPI. Nine images per well were acquired with automated MD2 microscope that autofocuses at each image acquisition (106 objective). The total cells number and the number of infected cells expressing GFP were scored by the MatLab-based infection counter software described previously [52].

Human phospho-RTK array
Human phospho-RTK array kit was obtained from R&D Systems (cat # ARY001) and experimental procedure was followed according the manufacturers guidelines. In brief, serum-starved subconfluent cells in 3.5 cm dish were inoculated with purified RSV (moi 30) or null virus prep extract for 15 min at 37uC. Then, cells were lysed and pre-clarified by centrifugation. The supernatant samples were incubated with the antibody array, and developed with provided reagents. Developed arrays were analyzed by densitometry with QuantityOne software (Bio-Rad).

Detection of the F by western blot
Subconfluent cells in 12 well plates were chilled on ice for 20 min. RSV (moi 10) in DMEM-Hepes was bound to cells for 1 h at 4uC. Post binding samples were washed and lysed in RIPA buffer (50 mM Tris, 150 mM NaCl, 2 mM EDTA, 1% NP-40, 0.1% SDS, pH 7.4). Remaining samples were washed with PBS supplemented with complete medium and transferred to 37uC for 1 h; following cells were trypsinized, washed and lysed with RIPA buffer. All samples were separated on the 10% Bis-Tris gels (Invitrogen). Western blots were probed with anti-F antibody (1:500), HRP conjugated goat ant-mouse (1:2000) and developed with super signal west pico (Pierce).
Immunofluorescence staining and analysis 0.8610 5 cells were seeded on 12 mm coverslips in the 24 well plate 24 h prior experiment. Plates were chilled on ice and then RSV was bound to cells in DMEM-Hepes for 1 h at 4uC. Viruses were washed away and cells were supplemented with complete medium and transferred to 37uC. At desired times cells were fixed in 4% formaldehyde, permeabilized with 0.1% triton X-100 and incubated with 10% goat serum for 30 min. Cells were then stained with appropriate primary and secondary antibody. Coverslips were mounted to the glass slides with prolong gold anti-fade reagent (Invitrogen).

Microscopy
Immunofluorescence images were captured with LSM Zeiss 510 microscope with the confocal laser scanning set up (objectives 60 or 1006). Per experiment, 10-15 images were captured and processed by ImageJ. For the virus particles detection Imaris software was used set up to detect particles larger than 0.5 mm and quality greater than 15.
Live cell imaging was performed with Olympus CellR, 206 objective with DIC setting, equipped with 37uC incubator.

Sample preparation for mass spectrometry
Hep2 cells or gradient purified RSV particles were lysed in denaturing buffer containing 8 M urea and 100 mM NH 4 HCO 3 . Lysates were briefly sonicated and proteins were reduced with 12 mM DTT for 30 min at 32uC and alkylated with 40 mM iodoacetamide for 45 min at 25uC. The samples were diluted 1:5 with 100 mM NH 4 HCO 3 and digested with sequencing-grade porcine trypsin (Promega) at an enzyme/substrate ratio of 1:100. The digestion was performed overnight and stopped with formic acid at final concentration of 2%. The peptide mixtures were desalted on Sep-Pak C18 cartridges (Waters), eluted with 80% acetonitrile, vacuum centrifuged until dryness and resuspended in 0.15% formic acid.
For each peptide, Q1 and Q3 masses, as well as collision energies (CE) for peptide fragmentation were calculated using Skyline (v1.3, MacCoss Lab Software). Double and triple charged product ions from the y-and b-series and transitions in a mass range of 350-1250 Da were considered (see full list of SRM transitions in the supporting information Table S1). The peptide samples were measured in SRM mode on a triple-quadrupole/ion trap mass spectrometer (5500QTrap, ABSciex) equipped with a nano-electrospray ion source. For the chromatographic separation of the peptides, the instrument was coupled with an Eksigent Nano LC system (ABSciex) connected to a 15-cm fused silica column, 75 mm inner diameter (BGB Analytik), packed in-house with Magic C18 AQ, 5 mm beads (Michrom Bioresources). The peptide mixtures were loaded from an autosampler cooled to 4uC (ABSciex) and separated with a linear gradient of acetonitrile/ water containing 0.1% formic acid from 5 to 35% acetonitrile in 30 min, with a flow rate of 350 nl/min. SRM analysis was conducted with Q1 and Q3 operated at unit resolution (0.7 m/z half maximum peak width) with up to 70 transitions per run (dwell time, 30 ms; cycle time, 2.5 s). Data were analyzed with the software Skyline. Peak area of the SRM peaks was used for quantitation, after confirming co-elution and shape similarity of the transitions monitored for each peptide. Outlier transitions (e.g., shouldered or noisy transition traces) were not considered in the calculations. Results are presented as the sum of the areas of all SRM peaks for a given peptide.

Statistical analysis
All experiments were performed at least in triplicate, and presented as normalized values with +/2 standard deviation (SD). . RSV (moi ,0.5) was bound to A549 cells at 4uC followed by 30 min at 37uC. Cells were by IIF with anti-F-AF488 (green), anti-N-AF594 (red), and phalloidin-AF647 (pseudocolored white) for confocal microscopy as, and Z-stack image series acquired. The orthogonal views of image Z-stacks (pseudo-colored white) were generated with ImageJ. (B). RSV (moi ,0.5) was bound to A549 cells at 4uC, virus inoculum was washed, cells warmed to 37uC, fixed at indicated times, and stained with phalloidin-AF488 (pseudo colored white) and anti-F-AF647 (red) antibody. Images represent Z-stack projections acquired with a confocal microscope. Movie S1 RSV induces transient blebbing of HeLa cells. HeLa cells were inoculated with a purified RSV (moi 30) and immediately imaged with Olympus CellR microscope with DIC settings with the 206 objective, 1 frame per 10 sec speed at 37C. (AVI)