Identification of TRAPPC8 as a Host Factor Required for Human Papillomavirus Cell Entry

Human papillomavirus (HPV) is a non-enveloped virus composed of a circular DNA genome and two capsid proteins, L1 and L2. Multiple interactions between its capsid proteins and host cellular proteins are required for infectious HPV entry, including cell attachment and internalization, intracellular trafficking and viral genome transfer into the nucleus. Using two variants of HPV type 51, the Ma and Nu strains, we have previously reported that MaL2 is required for efficient pseudovirus (PsV) transduction. However, the cellular factors that confer this L2 dependency have not yet been identified. Here we report that the transport protein particle complex subunit 8 (TRAPPC8) specifically interacts with MaL2. TRAPPC8 knockdown in HeLa cells yielded reduced levels of reporter gene expression when inoculated with HPV51Ma, HPV16, and HPV31 PsVs. TRAPPC8 knockdown in HaCaT cells also showed reduced susceptibility to infection with authentic HPV31 virions, indicating that TRAPPC8 plays a crucial role in native HPV infection. Immunofluorescence microscopy revealed that the central region of TRAPPC8 was exposed on the cell surface and colocalized with inoculated PsVs. The entry of Ma, Nu, and L2-lacking PsVs into cells was equally impaired in TRAPPC8 knockdown HeLa cells, suggesting that TRAPPC8-dependent endocytosis plays an important role in HPV entry that is independent of L2 interaction. Finally, expression of GFP-fused L2 that can also interact with TRAPPC8 induced dispersal of the Golgi stack structure in HeLa cells, a phenotype also observed by TRAPPC8 knockdown. These results suggest that during viral intracellular trafficking, binding of L2 to TRAPPC8 inhibits its function resulting in Golgi destabilization, a process that may assist HPV genome escape from the trans-Golgi network.


Introduction
Human papillomavirus (HPV) is a non-enveloped virus composed of a circular double-stranded DNA genome of approximately 8000 base pairs (bp) encapsulated by a capsid composed of two structural proteins: the L1 major capsid protein and the L2 minor capsid protein. The capsid is formed from 360 molecules of L1 organized into 72 pentamers, and 12-72 molecules of L2 localized in the central internal cavity of the L1 pentamer [1]. To date, 170 HPV genotypes have been identified in proliferative lesions of the skin or mucosa and classified based on L1 gene sequence homology [2]. HPVs that infect the genital mucosal epithelia are grouped into two types: the low-risk types (i.e., types 6 and 11) found primarily in condyloma, and the highrisk types (i.e., types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, 68, and 73) found in cervical cancer, the second most common type of cancer in women worldwide [3].
The endocytotic pathways utilized by HPV for cell entry are dependent on actin dynamics, the Na + /H + exchanger and signaling through a variety of kinase pathways, including RTKs, MAP kinase, phosphatidylinositol-3 kinase, protein kinase C, and p21-activated kinases, but are independent of clathrin, caveolin, lipid rafts, or dynamin-2 [19,31]. After internalization of the virion, the capsid undergoes conformational rearrangements and uncoating in endosomal compartments by acidification [32], and the uncoated L2-DNA complex is delivered to the trans-Golgi network (TGN) by the retromer complex [33,34]. The L2-DNA complex subsequently travels through an unknown microtubuledependent route to the nucleus, where replication of the viral genome occurs [11,17]. L2 is important for the escape of the L2-DNA complex from the endosomal and lysosomal compartments, which is mediated by L2 interaction with sorting nexin 17 [35][36][37][38][39][40][41]. Although L2 is essential for infectious entry of HPV, precise functional interactions between L2 and host factors are not well defined.
Successful identification of the cellular proteins necessary for a specific phase of a viral life cycle is greatly facilitated by the use of viral variant proteins that are deficient in such processes. L2dependent infection by HPV type 51 (HPV51) can be investigated using a unique pair of functionally distinct variants: the Ma strain isolated by Matsukura and Sugase, and the Nu strain isolated by Nuovo et al. [42]. We have previously reported that pseudovirus (PsV) comprised of the HPV51 MaL1, MaL2 and a reporter plasmid induce efficient reporter expression in inoculated HEK293FT cells, whereas that containing NuL2, instead of MaL2, does not [42], thus suggesting an essential role for MaL2 in HPV infection. Since the Nu strain was originally cloned from a cervical condyloma biopsy [46], it may represent a noninfectious viral genome that was integrated into the cellular genome or maintained only as an episome. In this study, using the two forms of HPV51 L2 as bait, we have performed a proteomic search for cellular proteins responsible for L2-dependent infection and report that the transport protein particle (TRAPP) complex subunit 8, TRAPPC8, coprecipitates with 51MaL2, but not with 51NuL2.
The transport protein particle (TRAPP) complex, also known as a trafficking protein particle complex, is a highly conserved multimeric guanine nucleotide-exchange factor (GEF) that regulates multiple membrane trafficking pathways [43]. Yeast TRAPP complexes exist as three forms: TRAPPI, TRAPPII, and TRAPPIII. TRAPPI and TRAPPII tether coated vesicles during traffic from the endoplasmic reticulum (ER) to the Golgi and during intra-Golgi traffic, respectively. TRAPPIII is required for membrane expansion events during autophagy [44]. TRAPPI is thought to be a core component of the TRAPP complex because both TRAPPII and TRAPPIII contain TRAPPI. In contrast, the mammalian TRAPP complex (mTRAPP) is poorly characterized, and its subunit compositions and functions remain unclear. Among the TRAPP subunits, TRAPPC8 (previously called KIAA1012) is ubiquitously expressed in various human tissues ><(), and its N-terminal sequence (amino acids [aa] 1-603) is homologous to yeast Trs85, a subunit of the yeast TRAPPIII complex. Several recent studies have revealed that TRAPPC8 is required for autophagy, Golgi stack formation, and ricin susceptibility [45][46][47].
In this paper we demonstrate that TRAPPC8 plays several important roles during HPV infection and that it is an absolute requirement for successful HPV cell entry. We discuss the significance of the interaction between TRAPPC8 and L2 during endocytosis and propose a novel mechanism for its mode of action.

Immunoprecipitation
HEK293FT cells, which had been seeded in a 10-cm culture dish 16 h before transfection, were transfected with 20-mg p3xFLAG14-51MaL2, p3xFLAG14-51NuL2, p3xFLAG14-51Ch4L2, p3xFLAG14-51Ch5L2 p3xFLAG14-16L2, or p3xFLAG14-31L2 using 80-ml Fugene HD (Roche, Diagnostics GmbH, Mannheim, Germany). Two days later, the cells were washed with PBS and harvested using a cell scraper. The cells were then suspended in 2-ml IP buffer (20 mM Tris-HCl pH 8.0, 10% glycerol, 5 mM MgCl 2 , 0.1% Tween 20, 0.1 M KCl, 0.5 mM DTT and protease inhibitor cocktail; Complete Mini, Roche Diagnostics GmbH), and briefly sonicated with a UP 50H ultrasonic processor (Hielscher Ultrasonics GmbH, Teltow, Germany). The homogenized cells were centrifuged at 50006g for 10 min at 4uC. The supernatant was mixed with 100-ml anti-FLAG M2 agarose beads (A2220, Sigma-Aldrich) and rotated for 1 h at room temperature. Proteins bound to the beads were eluted with 100-ml IP buffer containing 250 mg/ml 3xFLAG peptides (Sigma-Aldrich). Identification of proteins precipitated with 51MaL2, but not 51NuL2 The proteins released from the anti-FLAG M2 agarose beads were boiled in 66 SDS sample buffer for 10 min and fractionated using 5-20% gradient SDS-polyacrylamide gel electrophoresis (PAGE) (ATTO, Tokyo, Japan), followed by SYPRO Ruby (Life Technologies) staining. The 150-kDa protein band that coprecipitated with 51MaL2 was excised from the gel and subjected to in-gel trypsin digestion. The resultant peptide mixtures were analyzed by MALDI-QIT-TOF MS (AXIMA-QIT, Shimazu Biotech, Japan). Mascot software (Matrix Science) was used for protein identification.

Production of HPV31b virions and infection assay
HPV31b virions were prepared from CIN612 9E raft tissues [50]. The raft tissues grown for 14 days at the air-liquid interface were suspended in suspension buffer (D-PBS containing 0.85 M NaCl), and stored at 280uC. Frozen tissues were homogenized with a Mixer Mill MM300 (Qiagen, Hilden, Germany) at 25.0/s frequency twice for 5 min each. Homogenates were centrifuged at 8,0006g for 10 min at 4uC, and the supernatants were collected. Pellets were resuspended in suspension buffer and homogenized as above. Homogenates were centrifuged and the supernatants were collected. These steps were repeated three times. The pooled supernatant was placed on an Optiprep (Axis-Shield PoC AS, Oslo, Norway) gradient (from top to bottom, 27%, 33%, and 39% in PBS containing 1 mM CaCl 2 , 10 mM MgCl 2 , and 0.85 M NaCl) and centrifuged at 237,0006g for 3.5 h at 16uC in an SW55Ti rotor (Beckman Coulter, Fullerton, CA, USA). Fractions (400 ml each) were obtained by puncturing the bottom of the tube. Aliquots (5 ml per fraction) were analyzed by Western blot using mouse anti-HPV16L1 antibody (554171, BD Pharmingen, San Diego, CA, USA), which can also recognizes HPV31 L1. The fraction in which L1 was most abundant was used as a stock of HPV31b virions.
HaCaT cells (2610 5 /well) in a 24-well plate were inoculated with HPV31b virions diluted with growth medium to 1:100. The medium was changed daily. After incubation for 3 days, total RNA was extracted from the cells using an RNeasy mini prep kit (Qiagen). Reverse transcription (RT) and quantitative PCR (qPCR) were performed using a TaqMan Gene-Expression Cells-to-Ct TM kit (Life Technologies) with HPV31E1E4 primers (E7.4A and E4.3B) [51] (final concentration 200 nM) and HPV31 E1E4 probe (6FAM-CAG TGA CGA AAT ATC CTT TGC TGG GAT TGT T-TAMRA) [52] (final concentration 100 nM) on a 7900HT fast real-time PCR system (Life Technologies). The qPCR cycling profile was as follows: 50uC for 2 min, 95uC for 10 min, 45 cycles at 95uC for 15 s, and 60uC for 1 min. The standard curve was constructed using serial 10-fold dilutions of an E1E4 cDNA template from a cloned sequence. Relative HPV31E1E4 expression was normalized to beta-actin mRNA using TaqMan beta-actin control reagents (401846, Life Technologies).

Flow cytometric analysis
HeLa cells post-transfected with siRNAs in a 24-well plate were detached with PBS containing 2.5 mM EDTA. The cells (2610 5 ) were suspended in 100-ml PBS containing 10% FBS and 5-ml anti-P880/894, anti-P1250/1270, or anti-N1/603 antibodies. Cells were then incubated at 4uC for 1 h with tapping every 10 min. After washing with PBS containing 10% FBS, the cells were suspended in 100-ml PBS containing 10% FBS and 0.1-ml antirabbit IgG conjugated to Alexa Fluor 488 (Life Technologies). The cells were then incubated at 4uC for 1 h, with tapping every 10 min. After washing with PBS containing 10% FBS, the fluorescence was measured with a flow cytometer (PERFLOW; Furukawa Electric Co. Ltd., Tokyo, Japan).

Immunofluorescence microscopy
For observation of cell-surface TRAPPC8, HeLa cells (4610 4 ) post-transfected with siRNAs in an 8-well chamber glass slide were inoculated with 51PsVs (MOI of ,2000 particles/cell) in 200-ml growth medium. The cells were then incubated at 4uC for 1 h and washed with the medium. Cells were incubated in the medium with mouse anti-51L1 antiserum and rabbit anti-P880/894 antibody at 4uC for 1 h, followed by staining with Alexa Fluor 488-conjugated goat anti-mouse IgG and Alexa Fluor 546conjugated goat anti-rabbit IgG (Life Technologies) at 4uC for 1 h. The cells were washed with the medium and fixed with 4% paraformaldehyde (PFA) in PBS for 15 min at room temperature. After washing with PBS, the cells were mounted using a Prolong Gold anti-fade reagent with DAPI (Life Technologies).
For investigation of subcellular localization of L1, packaged DNA, and TGN46 in cells inoculated with 51PsVs, HEK293FT cells in an 8-well chamber glass slide were inoculated with PsVs (MOI of ,2000 particles/cell) containing 5-ettynil-29-deethynil-29-deoxyuridine (EdU)-labeled DNA in 200 ml growth medium [48]. The cells were incubated at 4uC for 1 h and washed with the medium, incubated in the medium at 37uC, fixed with 4% PFA in PBS for 15 min at room temperature and permeabilized with 0.5% Triton X-100 in PBS for 20 min at room temperature. The cells were washed with 3% BSA in PBS and incubated with a Click-it reaction cocktail containing Alexa Fluor 488 (Click-it TM EdU imaging kit; Life Technologies) as described previously [48]. The cells were incubated with mouse anti-51L1 antiserum and rabbit anti-TGN46 antibody (ab50595) (Abcam, Cambridge, UK), followed by staining with Alexa Fluor 555-conjugated goat antimouse IgG (Life Technologies) and Alexa Fluor 647-conjugated goat anti-rabbit IgG (Life Technologies), and washed and mounted as described above. To visualize the Golgi, the cells  were incubated with mouse anti-GM130 antibody (610822, BD Biosciences), followed by staining with Alexa Fluor 555-conjugated goat anti-mouse IgG (Life Technologies).
To assess the effect of L2 on Golgi stacks in cells, HeLa cells were transfected with ph51NuL2-GFP, ph51MaL2-GFP, ph51Ch4L2-GFP, ph51Ch5L2-GFP, ph16L2-GFP or ph31L2-GFP using Lipofectamine 2000 (Life Technologies) and incubated at 37uC for 24 h. These cells were fixed with 4% PFA in PBS for 15 min at room temperature and permeabilized with 0.5% Triton X-100 in PBS for 20 min at room temperature. The cells were washed with 3% BSA in PBS and incubated with mouse anti-GM130 antibody (610822, BD Biosciences), followed by staining mouse IgG and Alexa Fluor 546-conjugated anti-rabbit IgG at 4uC. Cells were fixed, permeabilized, and mounted with Prolong Gold anti-fade reagent with DAPI. Fluorescence was examined using confocal microscopy. The boxed area is enlarged in the right panel. doi:10.1371/journal.pone.0080297.g003  with Alexa Fluor 555-conjugated goat anti-mouse IgG (Life Technologies), and washed and mounted as described above.
All fluorescent images were obtained using a FluoView FV1000 confocal microscope (Olympus, Tokyo, Japan).

Internalization assay with transferrin and cholera toxin
HeLa cells at 72 h post-transfection with control or TRAPPC8 (KIAA1012-04) siRNA were incubated with 20 mg/ml transferrin (Tf) conjugated with Alexa Fluor 568 (T23365, Life Technologies) at 4uC for 1 h then followed by incubation at 37uC for 15 min, or incubated with 1 mg/ml cholera toxin subunit B (CtxB) conjugated with Alexa Fluor 555 (C34776, Life Technologies) at 4uC for 1 h then followed by incubation at 37uC for 1 h. Endocytotic uptake of the ligands was terminated by washing the cells with ice-cold PBS, and surface-bound Tf and CtxB were removed by acid treatment with DMEM containing 25 mM sodium acetate (pH 2.0) on ice for 2 min [53]. The cells were fixed, permeabilized, and incubated with mouse anti-GM130 antibody (610822, BD Biosciences) and anti-mouse IgG conjugated with Alexa Fluor 488 (Life Technologies). Fluorescence derived from incorporated Tf and CtxB was measured by confocal microscopy and quantified with FV1000 software (Olympus, Tokyo, Japan).

Identification of TRAPPC8 as an L2-interacting protein
We used HPV51 L2s derived from strains Ma and Nu as bait in the identification of the cellular proteins mediating L2-dependent HPV infection. HEK293FT cells were transfected with an expression plasmid for FLAG-tagged NuL2 (51NuL2-FLAG) or MaL2 (51MaL2-FLAG). Two days later the cells were homogenized to yield lysates. 51NuL2-FLAG or 51MaL2-FLAG were immunoprecipitated with anti-FLAG-antibody, followed by protein separation by SDS-PAGE. A comparison of the resolved proteins on a SYPRO Ruby-stained gel revealed a 150-kDa protein band present only in the 51MaL2-FLAG precipitate ( Figure 1B). Peptide mass fingerprint analysis identified this protein as TRAPPC8. Western blotting analysis confirmed this identification; an anti-TRAPPC8 antibody (anti-P880/894), which targeted a TRAPPC8 peptide corresponding to aa 880-894 (P880/894), recognized a 150-kDa protein in the 51MaL2-FLAG precipitate, but not in the 51NuL2-FLAG precipitate ( Figures 1C,  middle panel).
The aa sequence of MaL2 (GenBank GQ487712) is strikingly different from that of NuL2 (GenBank M62877) in two regions: aa 95-99 (region I) and aa 179-187 (region II) ( Figure 1A). Using HPV51 PsVs containing a series of chimeric Ma and Nu L2s, we have previously reported that region I of MaL2, which shows a high degree of identity among mucosal HPVs, is indispensable for PsV-mediated gene transduction [42]. Western blotting analysis of proteins co-immunoprecipitated with FLAG-tagged chimeric L2s, Ch5L2 (Ch5L2-FLAG), in which region II of MaL2 was replaced by that of NuL2, and Ch4L2 (Ch4L2-FLAG), in which region I of MaL2 was replaced by that of NuL2, showed the presence of TRAPPC8 in the Ch5L2-FLAG precipitate but not in the Ch4L2-FLAG precipitate ( Figures 1C, middle panel). Flow cytometric analyses of HeLa cells inoculated with PsVs containing these chimeric L2s and a GFP-expression plasmid revealed that 80% of HeLa cells inoculated with PsV containing the Ch5L2 were GFPpositive, whereas only 4% were GFP-positive in cells inoculated with PsV containing the Ch4L2 ( Figure 1A). These results indicate that the transgene expression levels generated by PsVs containing chimeric L2s correlate with the protein level of TRAPPC8 that coprecipitated with L2, suggesting that the TRAPPC8 interaction with MaL2 through region I is essential for transgene expression.
Furthermore, TRAPPC8 co-immunoprecipitated with FLAGtagged L2s of HPV16 and HPV31 at levels similar to those observed with FLAG-tagged MaL2 ( Figures 1C, middle panel), suggesting that the interaction between L2 and TRAPPC8 is a general property of L2 in various HPV types.
Since TRAPPC8 is supposed to be a subunit of the TRAPPIII complex, we further examined coprecipitation of TRAPPC12, another subunit of TRAPPIII subunits, with L2s. TRAPPC12 coprecipitated with FLAG-tagged L2s able to bind to TRAPPC8 ( Figure 1C, right panel).

Requirement of TRAPPC8 for the early stages of HPV infection
We examined the effect of TRAPPC8 knockdown on gene transduction with HPV51 PsV containing MaL1, MaL2, and the GFP-expression plasmid (51PsVMaL2). HeLa cells were individually transfected with four siRNAs against TRAPPC8 (KIAA1012-01, -02, -03, or -04), and two days later the TRAPPC8 levels were determined by Western blotting with an anti-TRAPPC8 antibody (anti-N1/603) purified from the serum of a rabbit immunized with recombinant TRAPPC8 N-terminal protein (aa 1-603, named N1/603). Expression of endogenous TRAPPC8 in cells transfected with TRAPPC8 siRNAs decreased at 48 h post-transfection compared to cells transfected with control siRNA (Figure 2A, upper panel). These TRAPPC8 knockdown cells were further inoculated with 51PsVMaL2, and 2 days later the number of cells expressing GFP was measured by flow cytometry. As shown in Figure 2A, the proportion of GFP-positive cells was reduced in TRAPPC8 knockdown cells compared to that in cells transfected with control siRNA. TRAPPC8 knockdown HeLa cells were viable 96 h post-transfection, (Figure 2A, right panel), indicating that the reduced GFP positivity was not due to impaired viability of the knockdown cells. As shown in Figure 2B, the number of GFP-positive cells also decreased in TRAPPC8 knockdown cells inoculated with HPV16 PsV (16PsV) and HPV31 PsV (31PsV).
We then tested the effect of TRAPPC8 knockdown on infection with authentic HPV virions. HaCaT cells were transfected with TRAPPC8 siRNAs (KIAA1012-03 or -04), and 2 days later the cells were inoculated with HPV31b virions prepared from CIN612 9E raft tissues. After an additional 3 day incubation, the levels of spliced viral E1E4 transcripts were quantified by RT-qPCR. The levels of E1E4 transcripts decreased in the TRAPPC8 knockdown cells compared to cells transfected with control siRNA, and this reduction correlated with the levels of TRAPPC8 knockdown ( Figure 2C). These results suggest that TRAPPC8 is required for the early stages of native HPV infection. . EdU-labeled DNA was visualized with Alexa Fluor 488 azide (green). Fluorescent images were obtained using confocal and differential interference contrast (DIC) microscopy. The boxed area is enlarged below. Nucleus is shown as ''N''. (B) Subcellular localization of EdU-labeled DNA, 51L1 and TGN46. HEK293FT cells were incubated with 51PsVNuL2 or 51PsVMaL2 in growth medium at 37uC for 8 h. The cells were fixed and permeabilized. EdU-labeled DNA was visulalized with Alexa Fluor 488 azide (green). L1 was visualized with mouse anti-51Ll VLP antiserum and Alexa Fluor 555-conjugated anti-mouse IgG (red). TGN46 (trans-Golgi marker) was visualized with rabbit anti-TGN46 antibody (ab50595, Abcam) and Alexa We further examined the effect of TRAPPC12 knockdown on gene transduction with 51PsVMaL2. The proportion of GFPpositive cells was similarly reduced in TRAPPC12 knockdown cells ( Figure 2D), implying the importance of TRAPPIII for HPV infection.

Localization of TRAPPC8 on cell surface
To investigate potential roles for TRAPPC8 in HPV entry, we probed the plasma membrane for TRAPPC8 by using flow cytometry with three anti-TRAPPC8 antibodies: anti-N1/603, anti-P880/894, and anti-P1270/1285 ( Figure 3A and S1A), which targets a TRAPPC8 peptide (aa 1270-1285, named P1270/1285; Figure 3A). HeLa cells were transfected with control siRNA ( Figure 3A, green and blue lines) or TRAPPC8 siRNA (KIAA1012-04) ( Figure 3A, yellow and red lines). Two days later, the cells were detached with EDTA and incubated with individual anti-TRAPPC8 antibodies ( Figure 3A, blue and red lines) at 4uC for 1 h. The cells were then incubated with Alexa Fluor 488conjugated anti-rabbit IgG, and fluorescence derived from the cells was analyzed by flow cytometry. As shown in Figure 3A, overall fluorescence intensity was increased in control siRNAtransfected cells probed with anti-P880/894 and anti-P1270/ 1285, and was slightly decreased in TRAPPC8 siRNA-transfected cells probed with anti-P880/894, but not with anti-P1270/1285. These results indicated that a proportion of TRAPPC8 is localized in the plasma membrane and that the epitope region of anti-P880/ 894 (aa 880-894) is exposed on the cell surface.
We further confirmed the localization of TRAPPC8 on the cell surface by confocal microscopy using anti-P880/894 ( Figure 3B). HeLa cells were stained with anti-P880/894 and Alexa Fluor 546conjugated anti-rabbit IgG; several areas of anti-P880/894 bound to the membrane surface were observed [ Figure 3B, upper left panels (-)]. We then examined TRAPPC8 on the surface of cells inoculated with 51PsVMaL2. The cells were inoculated with 51PsVMaL2 at 4uC for 1 h, and stained with anti-P880/894 and anti-51L1 antiserum. A greater number of anti-P880/895 reactive locations were detected, which colocalized with anti-51L1 immunoreactivity ( Figure 3B, upper left panels (+51PsVMaL2)). In contrast, no anti-P880/894 reactive locations were observed on the surface of cells transfected with TRAPPC8 siRNA ( Figure 3B, upper right panels). Anti-N1/603 and anti-P1270/1285 immunoreactivity on the cell surface was barely detectable and did not colocalize with anti-51L1 immunoreactivity ( Figure S1B), although anti-N1/603 immunoreactivity in permeabilized cells did colocalize with anti-51L1 and anti-P880/894 immunoreactivity ( Figure  S1C). Similar increases in anti-P880/894 immunoreactive locations, colocalized with anti-51L1 immunoreactivity, were observed on the surface of cells inoculated with 51PsVNuL2 or 51PsV lacking L2 (51PsVL2-) ( Figure 3B, lower panels). These results suggest that the aa 880-894 region of TRAPPC8 is exposed on the cell surface in a manner that is L1 capsid inoculation-dependent, that it colocalizes with PsV and that this colocalization does not require interaction between L2 and TRAPPC8. Similar results were obtained with the commercially available anti-TRAPPC8 antibody, sc-8519 (Santa Cruz Biotechnology Inc.), which targets the C-terminal peptide of TRAPPC8 ( Figure S1D); anti-TRAPPC8 immunoreactivity colocalized with L1 spots of on the cell surface when inoculated with 51PsVMaL2 ( Figure S1E).

Role of TRAPPC8 in PsV entry into cells
Since our results indicated that TRAPPC8 knockdown severely impaired HPV infection and that TRAPPC8 colocalized with PsV on the cell surface following PsV inoculation, the potential role(s) of TRAPPC8 in PsV entry were investigated. Firstly, we examined the attachment of PsV in TRAPPC8 knockdown cells. HeLa cells transfected with TRAPPC8 siRNA (KIAA1012-04) were inoculated with 51PsVMaL2, 16PsV or 31PsV ( Figure S2) and incubated for 1 h at 4uC. After removing unbound PsVs, the cells were detached with EDTA (Trypsin -, 0 h). PsVs bound to the cell surface were analyzed by Western blotting with anti-L1 antibodies. The L1 level was reduced by 0%-5% in cells transfected with TRAPPC8 siRNA compared to those transfected with control siRNA (Trypsin -at 0 h, Figures 4A and S3B). Although a decrease in L1 levels of 10%-30% was also observed in TRAPPC8 knockdown cells inoculated with HPV16 or HPV31 PsV (Trypsin -at 0 h, Figures 4A and S3B), these results suggest that TRAPPC8 does not play a major role in HPV cell attachment.
Next, we examined whether TRAPPC8 has a role in internalization of PsV. HeLa cells were incubated with 51PsVMaL2, 16PsV or 31PsV, in growth medium at 37uC for 0, 1, 2, 4, or 8 h. The cells were detached with trypsin, and L1 in the cell lysates analyzed by Western blotting with anti-L1 antibodies. As shown in Figure 4A, L1 was not detected in the lysates of cells harvested at time 0, indicating that PsV bound to the cell surface had not yet entered the cell. Following incubation with medium, trypsin-resistant L1 was detected, indicating that PsV had entered the cell (Figures 4A and S3B). In contrast, the Fluor 647-conjugated anti-rabbit IgG (blue). In lower panels, instead of L1 staining, GM130 (Golgi marker) was visualized with mouse anti-GM130 antibody and Alexa Fluor 555-conhugated anti-rabbit IgG (red). doi:10.1371/journal.pone.0080297.g006 level of trypsin-resistant L1 was severely reduced in cells transfected with TRAPPC8 siRNA (KIAA1012-04). This result was reproduced in cells transfected with another TRAPPC8 siRNA (KIAA1012-03) ( Figure S3A). In addition, we performed the same experiments using 51PsVNuL2, 51PsVL2-, 16PsV lacking L2 (16PsVL2-), and 31PsV lacking L2 (31PsVL2-) ( Figure  S2). As shown in Figure 4B, all PsVs exhibited reduced levels of trypsin-resistant L1 in TRAPPC8 knockdown cells when compared with cells transfected with control siRNA. These results indicate that TRAPPC8 plays a crucial role in PsV internalization, but that the process is independent of L2.
We further examined whether TRAPPC8 knockdown causes general defects in the endocytic uptake of non-HPV molecules like transferrin (Tf) and cholera toxin subunit B (CtxB). While Tf is internalized through dynamin-dependent, clathrin-mediated endocytosis [53], CtxB is internalized via both dynamin-dependent, clathrin-or caveolae-mediated endocytosis and dynamin-independent endocytosis [53,54]. TRAPPC8 knockdown in HeLa cells did not affect endocytic uptake of Tf ( Figures 5A and 5C), suggesting that TRAPPC8 is not involved in dynamin-dependent, clathrinmediated endocytosis. By contrast, transfection of TRAPPC8 siRNA caused partial defects in uptake of CtxB compared to control siRNA transfection ( Figures 5B and 5C), implying that TRAPPC8 has a general role in either dynamin-independent or dynamin-dependent, caveolae-mediated endocytosis.

Subcellular localization of L1 and packaged DNA in cells inoculated with HPV51 PsVs
Because TRAPPC8 knockdown inhibited the internalization of PsVs in a manner independent of L2 interaction, we then investigated which stage of HPV infection requires L2-TRAPPC8 interaction. To this end, we monitored intracellular trafficking of 51PsVNuL2 or 51PsVMaL2 in the cell. HEK293FT cells were inoculated with these PsVs packaged with 5-ethynyl-29-deoxyuridine (EdU)-labeled DNA. L1 and EdU-labeled DNA were visualized with the anti-51L1 antiserum and Click-it chemistry, respectively. In both 51PsVNuL2 and 51PsVMaL2-inoculated cells, EdU-labeled DNA and L1 colocalized (yellow dots, Figure 6A) on the cell surface at 0 h indicating cell attachment of the PsVs. After incubation for 8 h, EdU-labeled DNA (green dots, Figure 6A) was observed in the cytoplasm of 51PsVMaL2inoculated cells, suggesting that PsV had been internalized and that its packaged DNA was separated from the L1 capsid ( Figure 6A, right panel). In contrast, EdU-labeled DNA and L1 in the 51PsVNuL2-inoculated cells colocalized in the cytoplasm at 8 h. At 24 h, EdU-labeled DNA was transferred into the nucleus in 51PsVMaL2-inoculated cells, whereas EdU-labeled DNA and L1 in 51PsVMuL2-inoculated cells remained colocalized in the perinuclear region as large clusters without DNA translocation into the nucleus.
Since recent studies have revealed HPV trafficking to the trans-Golgi network (TGN) before genome translocation into the nucleus [33,34], we examined whether localization of EdU- The central region of TRAPPC8 is exposed on the cell surface upon virion attachment. (iii) The capsid is internalized into the cell by TRAPPC8-dependent endocytosis. (iv) The capsid undergoes conformational rearrangements and exposes hidden domains when exposed to low pH, leading to an exposure of the L2 N-terminal region. (v) Endosomal vesicles containing HPV51 Nu or HPV that fails to interact with TRAPPC8 are elongated or fused to each other, and these virions are eventually degraded by lysosomes and autolysosomes. (vi) HPV that interacts with TRAPPC8 through L2 inhibits TRAPPC8 functions, such as vesicle fusion and membrane expansion, both of which are necessary for the subsequent degradation process, thereby leading to the release of the viral genome from the trans-Golgi network. doi:10.1371/journal.pone.0080297.g008 labeled DNA and L1 to the TGN differ between NuL2-and MaL2-containing PsVs. HEK293FT cells were inoculated with 51PsVNuL2 or 51PsVMaL2 and, after 8-h incubation, the TGN was visualized with anti-TGN46 antibody (trans-Golgi marker) together with staining of L1 and EdU-labled DNA. As shown in Figure 6B, while L1 staining was adjacent to or colocalized with the TGN similarly in both 51PsVNuL2 and 51PsVMaL2 ( Figure 6B, upper panels, red dots), more EdU-labeled DNA signals were adjacent to the TGN in 51PsVMaL2-inoclated cells than in 51PsVNuL2-inoculated cells ( Figure 6B, upper panels, green and yellow dots). Similar results were obtained by staining the cells with the Golgi marker, anti-GM130 antibody ( Figure 6B, lower panels), suggesting that EdU-labeled DNA separated from L1 is more successfully localized in the Golgi in 51PsVMaL2inoculated cells. These results raise the possibility that L2-TRAPPC8 interaction may be required for the viral genome localization in the TGN/Golgi compartment and that this L2 function could be related to the impaired infectivity observed in NuL2.

Golgi dispersal caused by L2 expression and TRAPPC8 knockdown
To further gain insight into roles for L2-TRAPPC8 interaction in the Golgi, we investigated Golgi morphology in cells expressing L2. HeLa cells were transfected with expression plasmids for a series of L2-GFP fusion proteins, and the Golgi bodies inside the cell visualized using anti-GM130 antibody. As shown in Figure 7, Golgi stacks were dispersed in the cells expressing MaL2-GFP, but not in the cells expressing NuL2-GFP. In contrast, immunofluorescence assays performed on cells expressing MaL2-GFP using the anti-early endosome antigen 1 (early endosome marker), anti-LAMP2 (late endosome marker) and anti-protein disulfide isomerase (ER marker) antibodies, did not reveal obvious structural changes in the corresponding organelles ( Figure S4C). Golgi dispersal was also observed in cells expressing L2s that were capable of interacting with TRAPPC8, such as Ch5L2, HPV16 and HPV31 L2s (Figure 7, right panels). Intriguingly, TRAPPC8 knockdown in HeLa cells caused Golgi dispersal that was indistinguishable from that observed in MaL2-GFP expressing cells (Figure 7, left panels). Similarly, MaL2-GFP expression and TRAPPC8 knockdown did not affect early endosomes, late endosomes, or the ER ( Figure S4A). Although we were unable to detect colocalization of PsV with TRAPPC8 in the Golgi compartment because of high anti-TRAPPC8 background staining of intracellular compartments (data not shown), these results suggest that the interaction between L2 and TRAPPC8 disrupts normal Golgi structure through inhibition of TRAPPC8 function.

Discussion
In this study we have identified TRAPPC8, a specific subunit of the mammalian TRAPPIII complex (mTRAPPIII), as an L2interacting protein, and demonstrated that the knockdown of TRAPPC8 in human epithelial cells reduces both gene transduction with PsV and infectivity of authentic HPV virions. These observations strongly suggest that TRAPPC8 plays a central role in the early stages of HPV infection. However, TRAPPC8 knockdown inhibited the internalization of PsV independently of L2, thus excluding the requirement for L2-TRAPPC8 interaction during the initial entry step. Further, we find that the L2-TRAPPC8 interaction can induce Golgi dispersal and that this Golgi phenotype strongly correlates with the gene transduction efficiency of PsV. Thus, we propose two non-overlapping, L2-independent and -dependent roles for TRAPPC8 in the HPV cell entry mechanism.
For the L2-independent role of TRAPPC8, we propose a model in which the HPV virion, trapped by cell-surface L1-binding proteins such as HSPGs, is internalized by TRAPPC8-dependent endocytosis (Figure 8). The central region of TRAPPC8 was exposed on the surface of HeLa cells and colocalized with inoculated PsV (Figure 3B). These observations suggest the possibility that TRAPPC8 is located adjacent to HSPGs on the cell surface. Syndecans (syndecan-1 to -4) are a major class of membrane-spanning HSPGs, and their transmembrane (TM) and C-terminal cytosolic domains are highly conserved. Among the syndecans, syndecan-1 acts as an initial attachment protein for HPV entry into keratinocytes [55]. Given that TRAPPC4, one of the TRAPP core subunits, binds to the conserved EFYA motif at the C-terminus of syndecan-2 [56], it is possible that mTRAPPIII containing TRAPPC4 also binds to syndecan-1 in association with virion. Furthermore, TRAPPC8-containing vesicles beneath the plasma membrane may fuse with the plasma membrane trapping HPV, increasing exposure of its central region on the cell surface ( Figure 3B). When considering the mechanism of TRAPPC8dependent endocytosis, it is worth noting that TRAPP complexes regulate the function of small-GTPases via their GEF activity and by the combination of individual TRAPP subunits [43]. Thus, mTRAPPIII may regulate the GTPase activity of Rab5 and/or other unknown small GTPases that are necessary for the endocytosis of HPV [19,32].
TRAPPC8 knockdown did not affect uptake of transferrin ( Figure 5), excluding the possibility that the effects of TRAPPC8 knockdown are non-specific and act on multiple endocytotic pathways. However, TRAPPC8 knockdown partially decreased uptake of cholera toxin, which is internalized via both dynamindependent and -independent endocytosis ( Figure 5). Since TRAPPC8 knockdown reduces the cell's susceptibility to ricin [45], which enters the cell via a dynamin-independent pathway [54], and HPV also enters the host keratinocyte via a dynaminindependent pathway [19], TRAPPC8 may have a general role in this endocytotic pathway.
TRAPP complexes are involved in both the transport of vesicles from the ER to the Golgi and the endosome to the Golgi [43], and mTRAPPIII plays important roles in Golgi stack formation [47]. Recently, Day et al. and Lipovsky et al. reported that HPV16 is delivered to the TGN or the Golgi during the entry process [33,34], highlighting the importance of the TGN/Golgi compartment as a route for infectious HPV entry. Intriguingly, the Golgi dispersal phenotype observed in cells expressing L2s able to bind TRAPPC8 was almost identical to that seen in TRAPPC8 knockdown cells (Figures 7 and S4), suggesting that the interaction of L2 with TRAPPC8 inhibits its function in the Golgi. In contrast, L2s that can interact with TRAPPC8 did not affect early endosomes, late endosomes, or the ER ( Figure S4C), which suggests a specific effect of L2-TRAPPC8 interaction on the Golgi structure. In summary, we speculate that the L2-TRAPPC8 interaction inhibits TRAPPC8 function in the Golgi and causes Golgi destabilization, thereby facilitating escape of the HPV genome from the Golgi compartment.
Since most regions in L2 molecules are buried in the capsid and the N-terminus of L2 becomes exposed on the capsid surface during the entry process [1,11], the L2 N-terminus including region I may interact with TRAPPC8 in the lumen of endocytotic vesicles. Although we were unable to detect direct interactions between recombinant TRAPPC8 and L2 (data not shown), TRAPPC12, another specific subunit contained in mTRAPPIII, showed similar L2-binding profiles to TRAPPC8 ( Figure 1C, right panel), suggesting that mTRAPPIII binds to L2 as an entire complex. Further support for this mechanism comes from the observation that the knockdown of TRAPPC12 in HeLa cells reduced susceptibility to gene transduction with 51MaL2PsV ( Figure 2D). Similar results were obtained in HeLa cells transfected with siRNA against TRAPPC11, another subunit in mTRAPPIII (data not shown). These results suggest that mTRAPPIII as a whole plays an important role in HPV infection. In contrast, HeLa cells transfected with siRNA against TRAPPC9, one of the mTRAPPII subunits, exerted little or no effect on gene transduction with PsV (data not shown), suggesting that mTRAP-PII is not involved in HPV infection. Moreover, the effects of siRNAs for each TRAPP subunit (TRAPPC8, 9, 11, and 12) on HPV infection show similar susceptibility to ricin, suggesting that HPV and ricin share the same entry pathway, a pathway that is dependent on mTRAPPIII.
Human TRAPPC8 has been shown to contribute to the autophagy system [46] and its yeast homolog, Trs85, is required for the formation of pre-autophagosomal structure and the membrane expansion of autophagosomes [43,44,57]. Thus the L2-TRAPPC8 interaction may lead to the arrests of vesicle fusion or membrane expansion at the TGN. These arrests are likely to protect HPV trapped in the lumen of the vesicle from subsequent degradation by lysosomes or autophagosomes. This model is further supported by recent findings that autophagy responses are induced in the initial stages of HPV16 infection [58], and that most internalized HPVs are degraded by autolysosomes [59,60]. We hypothesize that while most HPVs are degraded by the cell defense machinery, including the autophagy system, a proportion of internalized HPVs, i.e. those in which L2 binds to TRAPPC8, inhibit its function and thus escape from the TGN through Golgi destabilization (see Figure 8). NuL2, which lacks the capacity to bind to TRAPPC8, may not be able to avoid such degradation. Although further experiments are needed to substantiate our hypothetical models for the roles of TRAPPC8 in infectious HPV entry, our findings will contribute to a better understanding of the mechanisms by which HPV enters host cells and escapes from the TGN. After washing with medium, the cells were incubated in medium with mouse anti-51L1 VLP antiserum and rabbit anti-N1/603 (left panel) or anti-P1270/1285 (right panel), followed by staining with Alexa Fluor 488-conjugated anti-mouse IgG and Alexa Fluor 546conjugated anti-rabbit IgG. The cells were fixed and permeabilized, then mounted with Prolong Gold anti-fade reagent with DAPI. (C) HeLa cells were incubated with 51PsVMaL2 (MOI of ,2000 particles/cell) in growth medium at 4uC for 1 h. After removing unbound PsVs, the cells were incubated in medium with mouse anti-51L1 VLP antiserum and rabbit anit-P880/894, followed by staining with Alexa Fluor 488-conjugated anti-mouse IgG and Alexa Fluor 546-conjugated anti-rabbit IgG. The cells were fixed and permeabilized, then incubated with or without rabbit anti-N1/603, followed by staining with Alexa Fluor 647conjugated goat anti-rabbit IgG. Fluorescence was visualized by confocal microscopy. The boxed areas are enlarged in the right panels. (D) Western blot analysis using commercial anti-TRAPPC8 antibody, sc-85191 (Santa Cruz Biotechnology Inc.). Truncated TRAPPC8 proteins, aa 1-603 (N1/603), aa 604-1435 (C604/1434), aa 604-747 (P604/747), aa 737-886 (P737/886), aa 876-1025 (P876/1025), aa 1015-1164 (P1015/1164), aa 1154-1303 (P1154/1303), and aa 1293-1435 (P1293/1435), were expressed in E.coli Rosetta-gami B (Takara Bio Inc.) by using the pCold II vector system (Takara Bio Inc.) and purified by nickel affinity chromatography. These proteins were electrophoresed and stained with CBB (upper panel). The proteins were analyzed by Western blotting using sc-85191 (lower panel). (E) Immunofluorescence microscopy analysis for cell-surface TRAPPC8 using sc-85191. HeLa cells were incubated with 51PsVMaL2 (MOI of ,2000 particles/cell) in growth medium at 4uC for 1 h. After removing unbound PsVs, the cells were incubated in medium with mouse anti-51L1 VLP antiserum and goat anti-TRAPPC8 antibody, sc-85191, followed by staining with Alexa Fluor 488conjugated anti-mouse IgG and Alexa Fluor 546-conjugated antigoat IgG. The cells were fixed and permeabilized, then incubated with rabbit anti-N1/603, followed by staining with Alexa Fluor 647-conjugated goat anti-rabbit IgG. Fluorescence was visualized by confocal microscopy. The boxed areas are enlarged in the right panels. control or TRAPPC8 siRNAs (KIAA1012-03 or -04) were inoculated with 51PsVMaL2, 51PsVNuL2, 51PsVL2-, 16PsV, 16PsVL2-, 31PsV, or 31PsVL2-(MOI of ,2000 particles/cell) and incubated for 1 h at 4uC. After washing with PBS, the cells were incubated in medium at 37uC for additional 0, 1, 2, 4 or 8 h. The cells were detached with PBS containing EDTA (Trypsin -) or PBS containing trypsin and EDTA (Trypsin +) at the indicated time points. The detached cells were lysed and boiled. Type 51L1, 16L1, 31L1, TRAPPC8, or a-tubulin were detected by Western blotting using anti-51MaL1 VLP antiserum, anti-HPV16L1 antibody (554171; BD Biosciences), anti-TRAPPC8 (anti-N1/ 603) and anti-a-tubulin antibodies, respectively. Asterisks: unknown protein that reacted with the anti-HPV16L1 antibody. Alpha-tubulin was detected as a loading control. (TIF) Figure S4 Effects of TRAPPC8 knockdown or 51MaL2 expression on intracellular organelles. (A) Effects of TRAPPC8 knockdown on early endosomes, late endosomes, or the endoplasmic reticulum (ER). HeLa cells transfected with control or TRAPPC8 siRNA (KIAA1012-04) were incubated in medium at 37uC for 2 days. The cells were fixed, permeabilized, and incubated with anti-EEA1 (early endosome marker, 610457; BD Biosciences), anti-LAMP2 (late endosome marker, 555803; BD Biosciences) or anti-PDI (ER marker, ab2729; Abcam) antibody, followed by staining with Alexa Fluor 555-conjugated anti-mouse IgG, and mounted with Prolong Gold with DAPI. Fluorescence in the cells was examined by confocal microscopy. (B, C) Effects of expression of 51MaL2-GFP on early endosomes, late endosomes, or the ER. HeLa cells transfected with pCMV-GFP (B) or pCMV-51MaL2-GFP (C) were incubated in medium at 37uC for 24 h. The cells were fixed, permeabilized, and incubated with anti-GM130 (Golgi marker, 610822; BD Biosciences), anti-EEA1, anti-LAMP2, or anti-PDI antibody, followed by staining with Alexa Fluor 555-conjugated anti-mouse IgG, and mounted as described above. Fluorescence in the cells was examined by confocal microscopy. White arrows indicate cells expressing GFP or 51MaL2-GFP. (TIF)