The CD63-Syntenin-1 Complex Controls Post-Endocytic Trafficking of Oncogenic Human Papillomaviruses

Human papillomaviruses enter host cells via a clathrin-independent endocytic pathway involving tetraspanin proteins. However, post-endocytic trafficking required for virus capsid disassembly remains unclear. Here we demonstrate that the early trafficking pathway of internalised HPV particles involves tetraspanin CD63, syntenin-1 and ESCRT-associated adaptor protein ALIX. Following internalisation, viral particles are found in CD63-positive endosomes recruiting syntenin-1, a CD63-interacting adaptor protein. Electron microscopy and immunofluorescence experiments indicate that the CD63-syntenin-1 complex controls delivery of internalised viral particles to multivesicular endosomes. Accordingly, infectivity of high-risk HPV types 16, 18 and 31 as well as disassembly and post-uncoating processing of viral particles was markedly suppressed in CD63 or syntenin-1 depleted cells. Our analyses also present the syntenin-1 interacting protein ALIX as critical for HPV infection and CD63-syntenin-1-ALIX complex formation as a prerequisite for intracellular transport enabling viral capsid disassembly. Thus, our results identify the CD63-syntenin-1-ALIX complex as a key regulatory component in post-endocytic HPV trafficking.


Results
Tetraspanin CD63 controls infectivity of HPV. We have previously reported that HPV16 pseudoviruses (PsV) colocalise with tetraspanin CD63 on the plasma membrane and in endosomes of infected HeLa cells 22 . Similarly, HPV16 L1 colocalises with CD63 in primary keratinocytes (NHEK) (Fig. 1a). To investigate whether CD63 plays a role in HPV infection, we performed a series of infection assays using various keratinocyte cell models depleted of CD63. These experiments demonstrated that CD63 knockdown with different siRNAs decreased HPV16 infection of HeLa, HaCaT and primary keratinocytes by > 50% (Fig. 1b). Control experiments showed that CD63 depletion does not affect luciferase gene expression ( Supplementary Fig. S1). Importantly, reexpression of CD63 in siRNA-depleted HeLa cells restored viral infectivity, excluding possible off-target effects from siRNA-based knockdowns (Fig. 1c). To extend these observations we analysed the role of CD63 in infection of two other oncogenic HPV types: HPV18 and HPV31. Immunofluorescence staining demonstrated that HPV18 and HPV31 L1 proteins also strongly colocalise with CD63 suggesting that these viruses were internalised into CD63-positive endosomes (Fig. 1d). Detailed quantification revealed that ~60% of HPV 16, 18, and 31 L1-positive endosomes were labelled with anti-CD63 mAb (Fig. 1e). Moreover, depletion of CD63 inhibited HPV18 and HPV31 infection in all cell lines tested (Fig. 1f). These data demonstrate that CD63 plays a general role in infection of epithelial cells with various human papillomaviruses.

CD63 regulates post-endocytic steps in HPV infection. To investigate the involvement of CD63 in
HPV infection in more detail we first examined the interaction of HPV16 PsV with CD63. L1-specific coimmunoprecipitation assay confirmed that HPV16 is physically linked to the CD63-containing protein complex (Fig. 2a). Next, we analysed the functional requirement of CD63 for virus cell binding, endocytosis, and post-entry processing of HPV16 PsV in CD63-depleted cells. As shown by Western blotting (Fig. 2b,c) and flow cytometry (Figs 1h and 2d), binding of HPV16 PsV to the surface of HeLa/CD63(-) cells was similar to that of control cells. Furthermore, flow cytometry experiments demonstrated that depletion of CD63 had no apparent effect on internalisation of HPV16: staining intensity of polyclonal anti-L1 antibody K75, which detects surface bound PsV, was comparable in control and CD63-depleted cells 24 hours after infection (Fig. 2d). To investigate the effect of CD63 depletion on virus post-uncoating processing, we examined the amount of disassembled capsids in endosomes using the disassembly-specific L1 antibody 33L1-7 (short L1-7) 22,29,45 which detects L1 capsomeres, unfolded L1 46 , and also post-disassembly degraded L1 14 . This antibody binds a linear epitope (residues 303-313) that is inaccessible in intact viral particles 46,47 and recognizes L1 protein after capsid disassembly 22 . Notably, the number of endosomes positive for the L1-7 epitope was dramatically decreased in CD63-depleted cells (Fig. 2e,f), while staining with the polyclonal L1-antibody K75 was unaffected (Fig. 2d,e). These findings indicate that CD63 is not involved in HPV cell binding or endocytosis. Instead, CD63 appears to control a step in HPV infection, which follows endocytosis but precedes virus uncoating and post-uncoating processes.
To examine CD63-dependent trafficking of HPV16 through the endosomal pathway, we also analysed L1 cleavage products. It was shown that most of L1 cleavages occur in acidified endosomes generating products of degradation 14 . In these experiments we found that CD63 depletion reduced the generation of lower molecular weight L1 cleavage products (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) (Supplementary Fig. S1). In further control experiments, we analysed whether depletion of CD63 would affect trafficking of transferrin or epidermal growth factor (EGF). Transferrin is a bona fide cargo for the clathrin-dependent endocytic pathway 48 and EGF is internalized by endocytosis followed by trafficking through the various endocytic compartments including multivesicular endosomes (MVEs) for degradation 49 . CD63 did not colocalise with transferrin under steady state conditions and its depletion did not affect the amount or intracellular distribution of endocytosed transferrin ( Supplementary Fig. S2). Similarly, CD63 knockdown had no effect on EGF cell entry or trafficking of the growth factor to MVEs (marked with anti-LBPA mAb, Supplementary Fig. S3). These results emphasise that CD63 specifically acts on virus-containing endosomes.

CD63 regulates HPV infection via syntenin-1.
We have previously reported that CD63 directly interacts with the PDZ domain-containing adaptor protein syntenin-1, which is known to control various aspects of endocytic trafficking including trafficking of CD63 50 . Therefore, we investigated whether syntenin-1 is involved in CD63-dependent steps of HPV infection. Western blot analysis showed that syntenin-1 together with endosomal marker Rab5 and the HPV L1 was detectable in specific sucrose density fractions of endosomal preparations of infected HeLa cells (Fig. 3a). In addition, Western blot (Fig. 3a) and quantitative mass spectrometry (Fig. 3b) of endosomal preparations revealed that syntenin-1 levels steadily increased on HPV-containing endosomes when assessed at four and seven hours post-infection ( Fig. 3b and Supplementary Fig. S4). By contrast, endosomal distribution of CD63 was not affected (data not shown), suggesting the interaction of virus with CD63 induces recruitment of syntenin-1 to endosomal membranes. Supporting these findings, immunofluorescence experiments demonstrated colocalisation of CD63, syntenin-1 and HPV16 L1 protein in HPV16-infected cells (Fig. 3c). Moreover, colocalisation of syntenin-1 and CD63 was significantly increased after 7 hours HPV infection when compared to uninfected controls (Fig. 3d,e) and, conversely, was reduced in CD63 depleted cells compared to control siRNA treated cells (Fig. 3f,g) further emphasizing that CD63 and syntenin-1 interact at HPV-containing Figure 2. CD63 associates with L1 and is involved in HPV16 post-uncoating processing but not in cell surface binding or virus endocytosis. (a) CD63 is coprecipitated by L1 pAb K75 immunoprecipitation (IP) from lysates of HPV16 PsV infected HeLa cells. Expressed (left) and precipitated proteins (right) were detected by Western blotting using L1 mAb 321-F (upper panels) and mAb CD63 (lower panels). (b-d) CD63 knockdown does not affect HPV16 cell binding. (b) HeLa cells were pretreated with CD63 or control siRNA and incubated with HPV16 PsV. Cell-bound PsV were detected in cell lysates by Western blotting (WB) using L1 mAb 312F. Polyethylenimine (PEI) treated cells were used as control for binding deficiency 75 . (c) Quantification of WB bands shows HPV16 cell surface binding independent of CD63. (d) PsV binding and endocytosis is unaffected by CD63 knockdown. HeLa cells were pretreated as in (b). Cells were incubated with HPV16 PsV as indicated. The amount of surface associated PsV was assessed by flow cytometry using L1 pAb K75. The difference between the amount of surface bound PsV at 1 h and 24 h serves as a measure for endocytosis. Mean fluorescence intensity of cells incubated with PsV for 1 h was adjusted to 100%. (e,f) L1-7 reactivity is reduced after CD63 knockdown. Representative pictures of immunofluorescence with HPV16 L1 pAb K75 (blue), mAb 33L1-7 (green) and CD63 mAb (red) are shown in (e). Hela cells were treated with siRNAs as in (b) and infected for 7 hours. K75 Ab recognises bound and internalized virus particles. 33L1-7 Ab recognises HPV16 epitope only accessible after capsid disassembly. (f) Shows quantification of disassembled viral capsids performed by analysis of L1-7 positive pixels of at least 15 images using ImageJ-script. Shown are the results of three independent experiments using CD63 siRNAs normalized to control siRNA treated cells. *P < 0.05 compared to control. endosomes. These endosomes also contain the minor capsid protein L2 (Fig. 3h) suggesting syntenin-1 association takes place prior to capsid disassembly and dissociation of the capsid proteins.
While knockdown of syntenin-1 decreased infectivity of HPV16 PsV in HeLa, HaCaT and NHEK cells (Fig. 4a, control in Supplementary Fig. S1), overexpression of the protein increased viral infectivity (Fig. 4b). The inhibitory phenotype of the knockdown could be reversed when syntenin-1 was reexpressed in siRNA-depleted cells, excluding possible off-target effects and indicating that syntenin-1 plays a key role in HPV infection. Similarly, infectivity of HPV18 and HPV31 PsV was also decreased in syntenin-1-depleted cells (Fig. 4c), emphasising that the syntenin-1 dependent pathway is crucial for the infection cycle of various HPV types. As with CD63, control experiments showed that trafficking of transferrin and EGF was not affected in cells depleted of syntenin-1 ( Supplementary Figs S2 and S3).
To examine whether CD63 endosomal localisation or formation of the CD63-syntenin-1 complex is important in HPV infection, L1-7 reactivity assays were performed by immunofluorescence after knockdown of endogenous CD63 and subsequent DNA transfection with CD63 wild type or CD63-Δ C, CD63-GY→ AA and CD63-T7 mutants. We previously reported that the deletion of the last two C-terminal cytoplasmic amino acids of CD63 (CD63-Δ C mutant) abolished syntenin-1 interaction and CD63 internalization 50 . The CD63 GY→ AA mutant in which critical glycine and tyrosine residues in the Tyr-based sorting motif were substituted for alanines also predominantly localise at the plasma membrane 50 . Therefore, we examined the activity of the CD63-T7 mutant, in which the positions of the last two amino acids of CD63 (i.e. V-M was changed to M-V) were interchanged. We found that while colocalising with the endogenous CD63 (Fig. 5a), this mutant did not interact with syntenin-1 (Fig. 5b). As illustrated in Fig. 5d,e, neither CD63-Δ C nor CD63-GY→ AA nor CD63-T7 mutants could recover L1-7 reactivity in CD63-depleted HeLa cells. By contrast, staining with the polyclonal L1-antibody K75 was unaffected ( Supplementary Fig. S5). Syntenin-1 depletion led to comparable reduction of PsV infection (Fig. 4a) and L1-7 reactivity (Fig. 5e). Together, these experiments reveal that CD63 endosomal localisation and the association with syntenin-1 is critical for CD63-dependent steps needed for HPV16 infectivity.

CD63-syntenin-1-complex controls viral trafficking to multivesicular endosomes.
To examine the role of the CD63-syntenin-1 complex in intracellular trafficking in more detail, we compared distribution of internalised HPV16 PsV in control and CD63-or syntenin-1-depleted cells using electron microscopy. These experiments revealed that in control cells 7 hours post-initiation of HPV infection a significant proportion of viral particles was found in multivesicular endosomes whose diameter exceeded 400 nm (Fig. 7a). By contrast, in CD63-or syntenin-1-depleted cells the viral capsids were located in smaller endocytic vesicles (i.e. < 250 nm in diameter) in the cell periphery (Fig. 7b). Consistently, endosomes larger than 400 nm in diameter contained lower numbers of viral particles in CD63-or syntenin-1-depleted cells as compared to control cells (Fig. 7c), underscoring the involvement of the CD63-syntenin-1 complex in trafficking of internalised HPV to MVEs. The overall number and size of MVEs were not affected in CD63-, or syntenin-1-depleted cells and independent of PsV infection (Fig. 7c). In agreement with the role of CD63-syntenin-1 complex in endocytic trafficking of HPV we observed that colocalisation between L1 and LBPA, a well-established marker of MVEs 51 , was markedly decreased in both CD63-and syntenin-1-depleted cells when compared to control cells (Fig. 8a-c). Furthermore, in CD63and syntenin-1-depleted cells, PsV were distributed more towards the cell periphery and colocalisation with early endosomal marker EEA1 was increased (Fig. 8b,c), providing an additional confirmation that HPV16 trafficking to perinuclear MVEs depends on the CD63-syntenin-1-complex. Moreover, transport of viral DNA to the Golgi apparatus was significantly decreased in CD63 and synenin-1-depleted cells, indicating that CD63-syntenin-1 mediated HPV trafficking steps to MVEs precede the retrograde trafficking of viral DNA to Golgi (Fig. 8d,e).
Previous studies demonstrated that ALIX also interacts with components of the ESCRT machinery and that these interactions may be critical for the biogenesis of MVEs 44,52 . As ALIX is also a syntenin-1 interacting protein, we hypothesised that ALIX is recruited to the CD63-syntenin-1 complex during the transition of viral particles to infectivity can be restored after syntenin-1 reexpression. HeLa cells were treated with syntenin-1 siRNAs, transfected with control or syntenin-1-expression plasmid and then infected with HPV16 PsV. Infectivity of HPV16 PsV was analysed as above. Upper panels show the efficiency of syntenin-1 knockdown and reexpression. *P < 0.05 significant decrease compared to control, $ P < 0.05 significant increase compared to cells transfected with syntenin 3′ UTR siRNA and control plasmid. (c) Syntenin-1 knockdown correlates with reduced HPV18 and HPV31 infectivity. HeLa cells were treated as in (a) and infectivity of HPV18 or HPV31 PsV was analysed as above. MVEs. Indeed, when analysed by immunofluorescent stainings, ALIX showed a prominent colocalisation with syntenin-GFP and HPV16 L1 in vesicular structures (Fig. 9a). These data suggest that ALIX may function in the context of the CD63-syntenin-1 complex to ensure viral trafficking through MVEs. Accordingly, we found that (a) Representative pictures of wild-type (wt) and mutant CD63. Rat1 cells were transfected with a plasmid encoding CD63 mutant T7 (human) and stained with species-specific CD63 antibodies, sc-5275 (human) and AD1 (rat). CD63-T7 mutant (green) colocalises with endogenous rat CD63 (red). (b) CD63-T7 mutant cannot interact with syntenin-1. CHO cells were cotransfected with plasmids encoding flag-tagged syntenin-1 and human CD63 as indicated. Immunoprecipitation was carried out using mouse anti-human CD63 mAb 6H1. Precipitated proteins were detected by Western blotting using rabbit Flag (lower panels) or CD63 antibodies (upper panels). Lysates were loaded as positive controls for transfection. Asterisks mark non-specific bands. (c-e) Recovery of L1-7 reactivity using CD63 mutants. CD63 depleted HeLa cells were transfected with control or mutant CD63 plasmid and infected with HPV16. (c) CD63 mutants efficiently reexpress CD63 after respective knockdown. (d) Shows representative pictures of CD63 and L1-7 double staining in different CD63 mutants. (e) L1-7 reactivity was quantified as in Fig. 2e,f. *P < 0.05 significant decrease compared to control, $ P < 0.05 significant increase compared to cells transfected with CD63 siRNA and control plasmid.
Scientific RepoRts | 6:32337 | DOI: 10.1038/srep32337 depletion of ALIX significantly reduced HPV infectivity rate (Fig. 9b, control in Supplementary Fig. S1). The N-terminal 3 LYPSL 7 sequence of syntenin-1 was shown to be critical for its interaction with ALIX 44 , a finding which we could reproduce using surface plasmon resonance (SPR) experiments. Here, ALIX shows strong binding to the peptide corresponding to the first fourteen amino acids of syntenin-1 including the LYPSL sequence (Fig. 9c). This binding was impeded when we used syntenin-1-derived peptides with point mutations at tyrosine 4 (Y4F) or serine 6 (S6A) of the LYPSL sequence (Fig. 9d). Accordingly, Synt-Δ N (a mutant lacking the first 100 AA), Synt-Y4F, and Synt-S6A mutants were unable to restore early steps of viral infection when reexpressed in syntenin-1 depleted cells (Fig. 9e,f, Supplementary Fig. S6 and controls in Supplementary Fig. S7). Interestingly, we found that overexpression of Synt-Δ N and Synt-S6A has a suppressive effect on L1-7 reactivity when overexpressed in control cells (Supplementary Fig. S6) thus further supporting the importance of the ALIX-binding site in HPV trafficking. These results demonstrate that the CD63-syntenin-1 complex requires ALIX to control the post-entry pathway in HPV infection.

Discussion
Previous data revealed that various oncogenic human papillomaviruses (HPV) enter cells via a novel endocytosis mechanism 22,24,26,40 . Here, the post-endocytic intracellular pathway controlling virus capsid disassembly and infection was investigated. We identified the CD63-syntenin-1-ALIX complex as a novel factor in the viral replication cycle regulating post-endocytosis trafficking of viruses to multivesicular endosomes. Having previously established the contribution of tetraspanin-enriched microdomains in HPV cell entry, our current data emphasise the critical role of tetraspanin-based protein platforms in regulation of various aspects of HPV infection.
We have recently reported that the association with other tetraspanins is essential in the CD151-dependent HPV infection pathway 23 . As CD151 controls viral uptake, other tetraspanins may either facilitate this process, or function downstream by directing trafficking of internalised viruses along the endocytic pathway. Here, we present evidence that the latter is regulated by a complex of tetraspanin CD63 and its interaction partner syntenin-1. These results provide a novel mechanistic insight into the poorly characterised process of post-endocytosis HPV trafficking. We demonstrated that HPV16 is physically linked to a CD63-containing protein complex and that the delivery of endocytosed viral particles to multivesicular endosomes was repressed in cells depleted of CD63 or syntenin-1 while MVE biogenesis was not affected. These observations are in line with an earlier report describing that lack of CD63 does not lead to any marked morphological alterations of the late endosomal/lysosomal  53 . Furthermore, depletion of these trafficking components led to decreased transport of the viral DNA to the Golgi implying that the CD63-syntenin-1-complex controls the transition of virus-containing endosomes to acidic cellular compartments, a critical step preceding retrograde trafficking of HPV to the Golgi.
With its physical proximity to HPV receptors within tetraspanin-enriched microdomains (e.g. EGFR, annexin A2, laminin-binding integrins 6,41,[54][55][56] ) and interaction capacity with syntenin-1 at virus containing endosomes, CD63 provides a critical physical link between internalised viral particles and intracellular endocytic machinery represented in the complex by syntenin-1 (see below). CD63 has been previously implicated in endocytic trafficking of various transmembrane proteins (e.g. CXCR4, VEGFR2, synaptotagmin VII) via mechanisms involving its tyrosine-based sorting signal (G-Y-E-V-M sequence) [57][58][59] . As expected 50 , we found that mutation of this sequence (in CD63-GY→ AA mutant) stabilised CD63 on the plasma membrane and precludes recruitment of the protein to HPV-containing endosomes. Importantly, the CD63-GY→ AA mutant was unable to support HPV infection thereby suggesting that the CD63-dependent step in virus infection relies on the endosomal CD63 protein pool.
Progressive recruitment of syntenin-1 to virus-containing endosomes and retention of internalised HPV in small peripheral endosomes in syntenin-depleted cells supports the idea that the CD63-syntenin-1 complex controls the pathway directing internalised viruses towards multivesicular endosomes. Accordingly, we observed that CD63 and syntenin-1 knockdown lead to reduced degradation of viral capsid proteins (data not shown). The role of syntenin-1 in various aspects of endocytic trafficking is well established 42,43,60,61 . Its ability to link numerous transmembrane partners to key components of the endocytic molecular machinery (i.e. phospholipids, small GTPases, ubiquitylated proteins) is likely to be central in the diverse endocytic functions of syntenin-1. Our results demonstrate that syntenin-1 partner protein ALIX is required for productive HPV infection. Interestingly, interaction between syntenin-1 and ALIX is critical for the production of exosomes, nanovesicles released by cells after MVEs fuse with the plasma membrane 44 . This, and other overlapping functions of CD63, syntenin-1 and ALIX such as viral egress from cells, ubiquitin-independent sorting of transmembrane cargos to MVEs 52,62 , suggest diverse cellular functionality of the complex. Previous studies demonstrated that ALIX interacts with TSG101 63,64 , an ESCRT-I protein binding to the HPV minor capsid component L2 32 . Therefore, spatial  We propose that syntenin-1 has a central role in regulating the functionality of the CD63-syntenin-ALIX complex in HPV infection. Indeed, recent reports demonstrated that syntenin-1 can be transiently phosphorylated at Tyr4 by the Src kinase 65 and downstream of activated FGFR2 66 , both of which were previously linked to HPV infection 18,21,26,67 . Our data indicate that this modification is likely to control the dynamics of syntenin-1-ALIX binding preventing interaction during the earlier steps of viral cell entry. Likewise, the assembly of the syntenin-1-ALIX complex may be suppressed by ULK1-induced phosphorylation of syntenin-1 68 . ULK1 is a serine-threonine kinase which was shown to play a critical role in the initiation of autophagy 69,70 . Autophagy pathway activation and subsequent dissociation of the CD63-syntenin-ALIX complex may divert trafficking of internalised viral particles from MVEs to other endolysosomal compartments formed under environmental insults. The proposed regulatory role of syntenin-1 phosphorylation is in agreement with Schelhaas and colleagues finding that broad range inhibitors of tyrosine and serine/threonine phosphatases suppressed HPV entry 26 . Finally, syntenin-1 was shown to interact with Rab5 and Rab7 71 , key regulators of endocytic trafficking, which are suggested to control endocytic routes of internalised HPV 26,33 .
In conclusion, our study presents the CD63-syntenin-ALIX complex as a principal regulator of post-endocytic trafficking of HPV to multivesicular endosomes (Fig. 10). These compartments contain factors such as the V-ATPase, required for virus disassembly 30,31 , and cyclophillins, required for separation of L1 and L2 29 , which enables interaction of L2 with cytosolic proteins to facilitate retrograde transport to the nucleus 30,[34][35][36]39,72,73 . Within the CD63-syntenin-1-ALIX-complex, syntenin-1 may represent the key factor integrating intracellular signalling pathways and ensuring their spatio-temporal coordination during early steps of HPV infection. Further analysis of the molecular network involving syntenin-1 is needed to draw a comprehensive map of trafficking routes directing endocytosed HPV to MVEs for productive infection.

Materials and Methods
Cell lines and Pseudovirions. Normal human epidermal keratinocytes (NHEK) were purchased from PromoCell, Heidelberg, Germany and cultivated according to manufacturer's instructions. The human cervical carcinoma cell line HeLa was purchased from the German Resource Centre for Biological Material (DSMZ), Braunschweig, Germany. HaCaT cells (human non-virally immortalised keratinocytes) were obtained from Cell Lines Services (CLS), Eppelheim, Germany. The cells were grown at 37 °C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% foetal calf serum (FCS), 1% Glutamax I (Invitrogen, Carlsbad, CA, USA), 1% modified Eagle's medium nonessential amino acids and antibiotics. CHO-K1 cells were grown in Ham's F-12 media supplemented with 10% fetal calf serum.
Coimmunoprecipitation assay. HeLa cells were seeded on 100-mm dishes in duplicate and cultured overnight. Cells of one dish were infected with HPV16 PsVs for five hours. Dishes were put on ice, cells were washed with Hepes buffer (25 mM Hepes, 150 mM NaCla, 5 mM MgCl 2 ) and lysed in 1% Chaps/Hepes buffer with protease inhibitors Aprotinin and Leupeptin (10 μ g/ml each). Lysates were incubated for 15 minutes at 4 °C on a rotating wheel, treated with an ultrasonifier three times for 20 seconds each (30% duty cycle; output control, 30%; Branson Sonifier 250, Emerson Industrial Automation, St. Louis, MO, USA) and incubated for 30 minutes at 4 °C on a rotating wheel. Dynabeads (M280 Sheep anti-Rabbit IgG, Invitrogen, Carlsbad, CA, USA) were prepared according to manufacturer's instructions and incubated with K75 rabbit IgG for 20 minutes at room temperature on a rotating wheel. Cell lysates were incubated with K75-coupled Dynabeads overnight at 4 °C on a rotating wheel. Beads were washed in lysis and Hepes buffer. Likewise, CHO cells were seeded and transfected with Flag-tagged syntenin-1 and/or human CD63 plasmids. At 48 hours post-transfection, cells were lysated by immunoprecipitation buffer (0.5% Brij 98-0.5% Triton X-100-PBS, 2 mM phenylmethylsulfonyl fluoride, 10 μ g/ml Aprotinin and Leupeptin) for 4 to 16 hours at 4 °C. Undissolved material was pelleted at 12,000 rpm for 10 min. Immune complexes were collected by incubation of cellular lysate with anti-CD63 mAb (6H1) pre-bound to protein G agarose beads for 16 hours and washed four times with immunoprecipitation buffer. Hela and CHO cell precipitates were boiled in Laemmli sample buffer and processed for Western blotting with anti-bodies as indicated in figure legends. CD63 Western blots were performed under non-reducing conditions. Cell binding assay by Western blotting. HeLa cells grown in 12-well plates were transfected with siRNAs for 48 hours. Subsequently, cells were incubated with HPV16 PsV for 1 hour at 37 °C, washed four times with PBS and then collected in SDS sample buffer for Western blotting. Cell-bound HPV16 particles were stained with anti-L1 antibody 312F and β -Actin was stained as input control.
Detection of surface bound particles by flow cytometry. HeLa cells were transfected with the indicated siRNAs and/or plasmids and infected with HPV16 PsV for 1 hour (cell binding assay). Subsequently, cells were extensively washed with PBS to remove unbound virions and analysed by flow cytometry or incubated for additional 23 hours. Cells were trypsinized with 0,25% Trypsin/2,5mM EDTA. Surface-bound particles were stained using the anti-L1 (K75) antibody and secondary anti-rabbit Alexa Fluor-488 antibody. Background staining was determined using a non-specific IgG serum. Measurements of surface bound particles at 1 hour post virus addition served as virus-cell binding analysis (see above) and comparison of bound particles at 1 h versus 24 h to show virus disappearance from the cell surface by endocytosis.
Detection of L1-7 epitope by immunofluorescence. HeLa cells were grown on coverslips and transfected with siRNAs. 48 hours later cells were infected with HPV16 pseudovirions (with approximately 200 particles per cell) and incubated for 7 hours at 37 °C. Subsequently, cells were fixed with methanol and processed for staining with mAb 1-7 as described previously 22 . This mAb recognizes a specific epitope located in the interior of the pseudovirion capsid and is not accessible in intact virions. The samples were analysed by fluorescence microscopy using a Zeiss Axiovert 200M inverted microscope (Carl Zeiss, Jena, Germany) and quantified by ImageJ software. For quantification, the relative amount of internalised particles was determined based on the L1-7 positive pixels relative to the cell nucleus signal (DNA/Hoechst 33342 positive pixels) out of 100 randomly selected cells (knockdown experiments) and 100 CD63 or syntenin-1 expressing cells (recovery experiments) from at least four independent experiments. A threshold value was set to exclude background.
Preparation of endosomes. 4.5 × 106 HeLa cells were either left untreated or were infected with 50 vge/cell of HPV16 PsV for 4 or 7 hours. Endosomal fractions were prepared as described 86,87 . Briefly, cells were harvested and homogenized, a post-nuclear supernatant was prepared and adjusted to 40.6% sucrose, 3 mM imidazole, pH 7.4, loaded at the bottom of an SW60 tube, and overlaid sequentially with 35% and 25% sucrose solutions in 3 mM imidazole, pH 7.4, and homogenization buffer (HB; 8.5% sucrose, 3 mM imidazole, pH 7.4). The gradient was centrifuged for 90 min at 14 000 × g. Thirteen fractions were collected. Early endosomal fractions (35%/25% interface) were identified by immunoblotting using the endosomal marker Rab5 87,88 .
Quantitative mass spectrometry. For protein digest early endosomes were pelleted by ultracentrifugation (100.000 × g, 1 h, 4 °C), proteins reduced by adding 5 mM DTT, free cysteines alkylated with iodoacetamide (Sigma-Aldrich, St. Louis, MO, USA), and proteins digested with 0.2 μ g porcine sequencing grade trypsin (Promega, Mannheim, Germany) 89 . Nanoscale liquid chromatography of tryptic peptides was performed with a Waters NanoAcquity UPLC system equipped with a 75 μ m × 150 mm BEH C18 reversed phase column and a 2.6 μl PEEKSIL-sample loop (SGE, Darmstadt, Germany). Mass spectrometry analysis of tryptic peptides was performed using a Waters Q-TOF Premier API system, operated in V-mode with typical resolving power of at least 10,000. All analyses were performed using positive mode ESI using a NanoLockSpray source. For data processing and protein identification the continuum LCMSE data were processed and searched using the IDENTITYE-Algorithm of ProteinLynx Global Server (PLGS) version 2.3. The resulting peptide and protein identifications were evaluated by the software using statistical models similar to those described by Skilling et al. 90 . Protein identifications were assigned by searching the UniProtKB/Swiss-Prot Protein Knowledgebase Release 52.3. Electron Microscopy. HeLa cells were infected with HPV16 PsV (approx. 500 p/cell) and after 7 hours fixed with 2.5% glutaraldehyde in PBS for 45 minutes at room temperature. Cells were washed with PBS, harvested and embedded in Epon 812 according to standard procedures. 70 nm ultrathin sections were cut, stained with 1% lead citrate and 2% uranyl acetate and finally analysed in a Zeiss CEM 902 electron microscope (Carl Zeiss, Jena, Germany), equipped with TRS digital camera. For quantification, 20 pictures of each sample were analysed.