Annexin A2 and S100A10 Regulate Human Papillomavirus Type 16 Entry and Intracellular Trafficking in Human Keratinocytes

Abstract

Human papillomaviruses (HPVs) cause benign and malignant tumors of the mucosal and cutaneous epithelium. The initial events regulating HPV infection impact the establishment of viral persistence, which is requisite for malignant progression of HPV-infected lesions. However, the precise mechanisms involved in HPV entry into host cells, including the cellular factors regulating virus uptake, are not clearly defined. We show that HPV16 exposure to human keratinocytes initiates epidermal growth factor receptor (EGFR)-dependent Src protein kinase activation that results in phosphorylation and extracellular translocation of annexin A2 (AnxA2). HPV16 particles interact with AnxA2 in association with S100A10 as a heterotetramer at the cell surface in a Ca²⁺-dependent manner, and the interaction appears to involve heparan-sulfonated proteoglycans. We show multiple lines of evidence that this interaction promotes virus uptake into host cells. An antibody to AnxA2 prevents HPV16 internalization, whereas an antibody to S100A10 blocks infection at a late endosomal/lysosomal site. These results suggest that AnxA2 and S100A10 have separate roles during HPV16 binding, entry, and trafficking. Our data additionally imply that AnxA2 and S100A10 may be involved in regulating the intracellular trafficking of virus particles prior to nuclear delivery of the viral genome.

INTRODUCTION

Human papillomaviruses (HPVs) cause a number of benign and malignant tumors, including cancers of the anogenital and oropharyngeal mucosa, and some skin malignancies (reviewed in references 1 and 2). HPVs are some of the many pathogens known to nonspecifically bind to glycoproteins, specifically syndecan-1 heparan-sulfonated proteoglycans (HSPGs), as a means of initial cellular interaction (3, 4). Although epithelial wounding is known to potentiate infection in vivo (5–7), the precise role that skin abrasion plays is ill defined. A prevailing model of early HPV infection events proposes that HSPG interactions result in conformation changes in the L1 major capsid protein that lead to exposure and proteolytic processing of the L2 minor capsid protein. These changes are thought to permit capsid dissociation from HSPGs and transfer of the virions to an unidentified cellular uptake receptor (8, 9). A variety of proteins have been identified as HPV-binding partners in the context of human keratinocytes (HKs), including syndecan-1 (10), alpha-6-integrin (11), tetraspanin CD151 (12), and laminin-332 (laminin-5) (13). However, since all of these cellular factors contain or associate with HSPGs at the cell surface and on the extracellular matrix (ECM) (10, 14–16), it is possible that their involvement in HPV infection may reflect HPV particle associations with HSPGs.

Our laboratory recently showed evidence for an alternate model to explain the movement of HPV capsids from HSPGs to signaling receptors important for infection. We found that HPV16 and HPV31 could be liberated from HSPG-containing syndecan-1 core proteins via the action of cellular proteinases. The released high-molecular-weight virion complexes are not dissociated from the HSPGs, but rather are decorated with HSPGs, syndecan-1 ectodomains, and epidermal growth factor receptor (EGFR) ligands like EGF, amphiregulin, and heparin-binding EGF. The growth factor (GF)-HSPG-virus complexes activate signaling through cognate GFRs, including EGFR (17). The proteinase-mediated cleavage of HSPG-GF complexes to activate cognate GFRs is a normal cell process, but complete physical release of these complexes from the cell is not requisite; the HSPG-GF complexes more simply can be transferred to nearby GFRs on the same or adjacent cells upon ectodomain cleavage (18). Interestingly, a number of intracellular pathogens, including Chlamydia spp., human immunodeficiency viruses, enterovirus 71, hepatitis C virus, and others, employ complex host cell interactions, many using soluble cell factors to bridge to cell surface receptors and, in several cases, engaging multiple interacting proteins to promote uptake (reviewed in reference 19). Thus, our model for HPV infection proposes that the association of the virus with different cell factors permits HPVs to utilize more than a single receptor and entry route.

The two described models for HPV infection need not be mutually exclusive. Yet the mechanism(s) of entry and the specific receptors directly involved in the internalization of oncogenic HPVs remain mysterious. Clathrin- and caveola-mediated endocytosis, the two main pathways used by nonenveloped viruses for cell invasion (20), have been shown to function in HPV entry. A majority of HPV types studied so far, including HPV16, are reported to enter the cell via clathrin-dependent endocytosis; however, the details are rather difficult to reconcile due to the use of varied methods as well as both keratinocyte and nonhost cell lines (21–24). In contrast, a newer report shows that HPV16 entry can occur in HKs via a macropinocytosis-related endocytic pathway that is dependent on actin dynamics and tyrosine kinase signaling but independent of clathrin and caveolin (25). Many entry routes have been discounted because specific inhibitors fail to give preponderant effects on infection (25). However, an underlying assumption of most studies is that a single or predominant route is used and/or that HPV virions will be equally susceptible to the inhibitors shown to be effective for other viruses; such assumptions may be unsound. Prominent difficulties that impede a straightforward determination of HPV internalization routes include the variety of identified viral interacting proteins and the unusually long time the virus particles spend attached to the ECM and/or plasma membrane prior to entry. The process of entry has been much studied, but never with conclusive evidence of any robust primary route. Again, the use of multiple receptors with different entry routes could help explain these difficulties.

Annexin A2 (AnxA2) is a 36-kDa protein whose gene belongs to a multigene family with over 160 members. These Ca²⁺- and phospholipid-binding proteins play multiple roles in membrane trafficking (26). Annexins are characterized by type II Ca²⁺ binding motifs within the annexin conserved core domain that enable Ca²⁺-dependent interaction with phospholipid-containing membranes. The core domain is made up of four annexin repeats and forms a disc-shaped protein with a convex surface as a membrane-binding site. The AnxA2 core domain interacts with F-actin and modulates actin polymerization and dynamics (27). The N terminus of AnxA2 harbors phosphorylation sites at Tyr23 and Ser25, which are substrates for pp60-src (28) and protein kinase C (29), respectively. AnxA2 Tyr23 phosphorylation is a well-studied modification that regulates cellular distribution, endosome recruitment, and association with a variety of ligands (30–34). AnxA2 exists in cells in two physical states, as a monomer or in a stable heterotetramer with S100A10 (p11). This complex, called A2t, forms when two molecules of AnxA2 bind to the S100A10 homodimer (reviewed in reference 26). The AnxA2 N terminus also contains an atypical membrane binding motif (amino acids 15 to 24) that regulates Ca²⁺-independent association of AnxA2 with endosomes (35).

AnxA2 functions at the early stages of the endocytic pathway, including both clathrin (36)- and caveola-mediated internalization (37), macropinocytic rocketing (38), multivesicular endosome biogenesis, and subsequent fusion with lysosomes (39). Association with lipid raft microdomains is mediated via a direct interaction of AnxA2 with phosphatidylinositol (4,5)-bisphosphate [PtdIns(4,5)P2] (40–43). AnxA2 interacts functionally with proteins of the endocytic machinery (e.g., AP1/AP2) (44) and trafficking molecules such as EGFR (45). This suggests that AnxA2 may directly initiate coated-pit formation and facilitate endosome engulfment. It has been proposed that AnxA2 mediates a function in endocytosis through association with the actin cytoskeleton (40, 43). However, interaction with p11/S100A10 is needed neither for AnxA2's binding to endosomal membranes nor for its actions in endosomal trafficking (34, 46).

In contrast to the role of AnxA2 in endocytosis, intracellular S100A10 participates in the trafficking of several proteins, including AnxA2, to the plasma membrane (reviewed in reference 47). The A2t at the cell surface is a prominent plasminogen receptor and activator; other known actions of S100A10 that may or may not be separate from A2t interactions involve immune cells and brain functions (47). S100A10 appears to be required for AnxA2 activities at the cell surface in cultured cells, and S100A10 is undetectable unless AnxA2 is also expressed (48). Thus, actions specific to either AnxA2 or S100A10 are generally difficult to separate and often are not obvious in the literature. AnxA2 and/or A2t has been implicated in the process of entry and replication of multiple viruses, including rabbit vesivirus (49), cytomegalovirus (50, 51), enterovirus 71 (52), and human immunodeficiency virus type 1 (53–55).

AnxA2 and S100A10 are expressed in many tissues, including human epithelia (56, 57). AnxA2 is predominantly expressed in the endocervix and basal layers of epithelia and the ectocervix (56, 58), whereas S100A10 is expressed in both basal and suprabasal cells (57). Upon HK wounding in culture, AnxA2 expression increases at the cell surface (58), and knockdown of AnxA2 alters wound-induced HK migration (59). In wounded human skin, S100A10 expression remains high throughout all epidermal layers, whereas AnxA2 expression is concentrated in basal layers (57).

Given the expression patterns in the epithelium and the roles of AnxA2 in GF shedding, endocytosis, GFR trafficking, and entry of other viruses, we hypothesized that AnxA2 may regulate the entry of GF-associated HPV16 particles into host keratinocytes. Herein we show the mechanisms for AnxA2 activation and mobilization for HPV16 binding at the plasma membrane. We demonstrate that recruitment of AnxA2 to the external leaflet of the plasma membrane is EGFR and Src kinase dependent and is regulated by HPV16 exposure. Our data indicate that A2t influences not only HPV16 binding and entry but also the intracellular trafficking of the particles. The findings are in partial agreement with a recent study by Woodham et al. describing the interaction of S100A10 with HPV16 particles in an L2-dependent manner as a requirement for HPV16 pseudovirus infection (60). However, in contrast to the conclusions made by Woodham and coworkers, our data indicate that HPV16 associates with A2t in a Ca²⁺-dependent and likely an HSPG-dependent manner. Whereas an interaction of S100A10 with HPV16 L2 is not incompatible with our findings, our additional data suggest that AnxA2 and S100A10 have separable roles in HPV16 binding and entry.

MATERIALS AND METHODS

Plasmids and antibodies.

The plasmid for capsid production, HPV16-L1/L2 (pXULL), was obtained from J. Schiller's laboratory (NIH) (61), and the pGL3-control plasmid expressing luciferase was purchased from Promega. Plasmids for AnxA2 expression (pcDNA3-AnxAwt), the AnxA2-EGFP fusion plasmid (pEGFP-C1-wild type annexin II), and the control green fluorescent protein (GFP) expression plasmid (pEGFP-C1) were kind gifts from J. K. Vishwanatha (University of North Texas) and were described previously (62). The p11 expression plasmid (S100A10 cDNA) was purchased from OriGene (Rockville, MD). Rabbit polyclonal anti-AnxA2 antibody (Ab) (rabbit annexin II [H-5] polyclonal Ab) and mouse monoclonal anti-AnxA2 Ab (mouse annexin II [H-5] monoclonal Ab [MAb]), both directed against amino acids 1 to 50 of the AnxA2 N terminus, mouse monoclonal anti-phosphorylated tyrosine 23 of AnxA2 Ab (mouse p-annexin II [85.Tyr24] monoclonal Ab), mouse monoclonal anti-p11 Ab (S-100A10 [4E7E10]), and mouse anti-PCNA Ab were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). A rabbit polyclonal anti-mouse secondary Ab (control Ab, C1) was obtained from eBiosciences. Mouse monoclonal anti-HPV16 L1 Ab (CamVir) was purchased from Abcam. Rabbit polyclonal anti-HPV16 virus-like particle (VLP) Ab was generated by our laboratory and affinity purified via protein A column chromatography. Rabbit polyclonal anti-EGFR Ab [EGF receptor (D38B1) XP(R) rabbit MAb] was purchased from Cell Signaling Technology. Mouse monoclonal anti-heparin/heparan sulfate Ab (Ms X heparin/heparan sulfate) was obtained from Millipore. Alexa Fluor 488-labeled epidermal growth factor (Alexa Fluor 488-EGF) was obtained from Invitrogen. Mouse anti-LAMP1 antibody (H4A3), rabbit anti-Src antibody, and mouse anti-EEA1 Ab were obtained from Abcam. Mouse antiphosphotyrosine (anti-pTyr) Ab (4G10 Platinum) was purchased from Millipore. Mouse anti-α-tubulin Ab was purchased from Zymed. Mouse anti-Na/K ATPase α 1 was obtained from Abcam. DyLight 594-conjugated affinity pure donkey anti-rabbit IgG and DyLight 488-conjugated affinity pure donkey anti-mouse IgG were obtained from Jackson Immunochemicals.

Cells, PsV production, and infections.

HaCaT cells (a gift from N. Fusenig, Deutsche Krebsforschungszentrum, Heidelberg, Germany) and HEK-293TT cells (a gift from C. Buck, NIH, USA) were maintained as previously described (63, 64). C4-2 prostate adenocarcinoma cells are derivatives from the LNCaP epithelial cell line, which lacks AnxA2 expression due to the hypermethylation of the AnxA2 promoter (65, 66). They were gifts from G. N. Thalmann (University of Bern) and were maintained as reported previously (67). HPV16 pseudovirions (PsVs) encapsidating a luciferase expression plasmid were generated and purified as described previously (63, 68, 69). PsVs were quantified for pseudoviral genome equivalents (vge) by quantitative PCR (qPCR), and L1/L2 capsids were subjected to SDS-PAGE and Coomassie brilliant blue staining to assess the quantity of capsids and purity of virus stocks (69).

HaCaT cells were routinely seeded at 2 × 10⁵ cells/35-mm dish for a next-day confluence of ≈60%. Luciferase reporter genomes containing HPV16 PsVs were sonicated for 60 s, added to culture media, and incubated on cells at 4°C for 1 h to allow attachment. Unbound viruses were removed, cells were washed three times with ice-cold phosphate-buffered saline (PBS), fresh media were added, and cells were shifted to 37°C to promote virus entry. Cells were analyzed for luciferase expression as a measure of infection 24 h after the temperature shift. Cells were washed three times with PBS, lysed (Promega luciferase lysis buffer) for 10 min at room temperature (RT), and clarified by centrifugation. Luciferase expression was analyzed using the luciferase assay system (Promega) by following the manufacturer's protocol. Raw data (relative light units) were standardized against total protein concentration, determined by Bradford assay (Bio-Rad). Data were graphed as the percentage of infection relative to control infections, which were set as 100%. For phosphorylation studies, coimmunoprecipitation (co-IP), and confocal analysis, infections were terminated at earlier time points after the temperature shift, as specified for each experiment. In selected experiments cells were pretreated for 30 to 60 min and infection continued in the presence of rabbit anti-AnxA2 Ab (5 μg/ml, 10 μg/ml, and 15 μg/ml), mouse anti-p11 Ab (5 μg/ml and 10 μg/ml), rabbit anti-mouse secondary Ab (15 μg/ml), mouse anti-PCNA Ab (15 μg/ml), PP2 inhibitor (Tocris), PD168393 (Calbiochem), 0.01% NaN₃ (Sigma), or EGTA (Sigma-Aldrich). In one experiment HPV16, prior to pseudovirus infection, was treated with 1 mM EGTA for 1 h at 37°C, followed by buffer exchange to remove free EGTA using Amicon columns (100 kDa).

Small interfering RNA (siRNA) transfection was performed using Lipofectamine 2000 reagent (Invitrogen), with AnxA2 siRNA (100 nM; Santa Cruz Biotechnology) according to the manufacturer's recommendations. A nonspecific siRNA (siGENOME nontargeting siRNA pool 1; Thermo Scientific) was used as a negative control. Transfection was monitored by fluorescence-activated cell sorter (FACS) using fluorescein-conjugated siRNA (Cell Signaling).

Src protein kinase and AnxA2 tyrosine 23 phosphorylation.

HaCaT cells starved for 24 h in serum-free medium (SFM) were either mock exposed or exposed to HPV16 for 0, 5, 15, 30, and 60 min in the absence or presence of inhibitors: PP2 (1 μM), PD168393 (100 nM), and rabbit anti-AnxA2 Ab (10 μg/ml) or EGF (5 ng/ml). Cells were lysed with radioimmunoprecipitation assay (RIPA) buffer containing 1× Halt protease and phosphatase inhibitor cocktail and 1 mM sodium orthovanadate and subjected to immunoprecipitation (IP). AnxA2 was pulled down with rabbit anti-AnxA2 Ab (1 μg/100 μg cell lysate), followed by immunoblotting with mouse anti-pTyr23-AnxA2 Ab (1:500) and mouse anti-AnxA2 Ab (1:500) to verify equal loading of samples and to normalize data. Src protein kinase was IP with rabbit anti-Src Ab (1 μg/100 μg cell lysate) followed by immunoblotting with mouse antiphosphotyrosine Ab and anti-rabbit Ab to detect heavy and light chains as a loading control. Data were analyzed with ImageJ (http://rsbweb.nih.gov/ij/) and expressed as the percentage of normalized area, with error bars representing standard deviations (SD).

Cell fractionation.

HaCaT cell monolayers in 75-cm² flasks, either mock exposed or HPV16 exposed for 30 min, were harvested and homogenized with a Dounce homogenizer in hypotonic buffer (25 mM Tris-HCl, pH 7.5, 1× Halt protease and phosphatase inhibitor cocktail) with the tight (B) pestle (Kontes Glass Co., Vineland, NJ). Cell breakage was judged by phase-contrast microscopy. Intact cells and nuclei were removed by centrifugation at 4°C for 5 min at 1,000 × g. The membranes were isolated from precleared supernatants by centrifugation for 60 min at 100,000 × g. The membrane-containing pellet was separated from the cytosolic fraction, washed with hypotonic buffer, and dissolved in 5 mM Ca²⁺-containing RIPA buffer. Protein concentrations were measured by Bradford assay. Samples were stored at −20°C until use. Immunoblotting of anti-α-tubulin was used as a control for cytosolic fraction loading, and Na/K ATPase α 1 immunoblotting was used as a loading control for membrane fractions.

Analysis of HPV16 binding to the cell surface of HaCaT cells in the presence of EGTA.

HaCaT cells were either left untreated or preincubated with various concentrations of EGTA (0.1 mM, 0.5 mM, 1 mM) for 1 h at 37°C. Cells were next either mock or HPV16 exposed (100 vge/cell) for 1 h at 4°C. After binding, medium was collected and subjected to 25- to 20-fold concentration with Amicon columns (100 kDa). Total protein concentration was determined using the Bradford assay, and samples were subjected to SDS-PAGE analysis. Unbound HPV16 present in cell medium was detected using anti-HPV16 L1 Ab. In addition AnxA2 release to the media after 1 h of treatment with increasing concentration of EGTA at 37°C was determined using anti-AnxA2 Ab.

Transmission electron microscopy.

HPV16 PsVs (200 ng) were negatively stained using 2% uranyl acetate by following a standard protocol. Particles were visualized by transmission electron microscopy (Hitachi 7500) at 80 kV following adsorption to a carbon-coated electron microscopy grid.

Immunofluorescence.

For colocalization studies, HaCaT cells (1 × 10⁵ cells/35-mm dish) grown on glass coverslips were mock or HPV16 exposed for 30 min after the temperature shift as described above. Cells were fixed with 3.7% paraformaldehyde (PFA) for 10 min. HPV16 was detected with rabbit anti-HPV16 VLP Ab (1:200) followed by DyLight 594-conjugated affinity pure donkey anti-rabbit IgG (1:200). AnxA2 was detected with mouse anti-AnxA2 Ab (1:200) in 0.2% fish gelatin blocking buffer (Sigma-Aldrich) followed by donkey anti-mouse Alexa Fluor 488-IgG secondary Ab (1:200). S100A10/p11 was detected with mouse anti-p11 Ab (1:200) followed by anti-mouse Alexa Fluor 488-IgG secondary antibody (1:200). For EGFR visualization Alexa Fluor 488-labeled EGF (5 μl/ml; Invitrogen) was added for 10 min at 4°C prior to temperature shift. Cells were mounted using Prolong Gold with DAPI (4′,6-diamidino-2-phenylindole; Invitrogen) for nuclear visualization. Images were taken using a Zeiss LSM510 Meta confocal microscope at the University of New Mexico (UNM) Cancer Center Fluorescence Microscopy Facility, supported as detailed at http://hsc.unm.edu/crtc/microscopy/. Three-dimensional (3D) images were generated with Zen 2009 software (Zeiss), after acquiring z-stacks.

To determine the effect of anti-AnxA2 Ab treatment on PsV entry, the HPV16 particle localization within the cell was studied, in the absence and presence of specific and nonspecific antibodies, at 24 h postinfection (p.i.) using confocal microscopy. Cells were either left untreated or preincubated for 1 h at 37°C with 10 μg/ml of rabbit anti-AnxA2 Ab, 10 μg/ml of anti-p11 Ab, or 10 μg/ml of control antibody (rabbit anti-mouse IgG antibody). Cells were infected next with HPV16 PsVs for a period of 24 h in the absence or presence of specific treatments. The cells were fixed, the plasma membrane was stained with wheat germ agglutinin (WGA) conjugated to Alexa Fluor 488 (Invitrogen) (5 μl/ml), and cells were permeabilized with 0.1% Triton X-100. HPV16 was detected with mouse anti-HPV16 L1 Ab. Confocal image analysis (SlideBook software 5.0.0.24) was performed on 6 cells per treatment to determine the colocalization of virus with the plasma membrane. Side views (i.e., cross sections) of the imaged cells were primarily used for colocalization analysis. Data are presented as the average percentages of colocalization between red HPV16 pixels and green plasma membrane pixels relative to the total HPV16 red pixels detected in the cell. In some experiments, anti-p11 Ab-treated cells were immunostained at 24 h p.i. with rabbit anti-HPV16 VLP Ab and mouse anti-AnxA2 Ab (1:200), mouse anti-LAMP1 Ab (1:200), or mouse anti-EEA1 Ab (1:200).

The abundance of AnxA2 on the cell surface of mock- and HPV16-infected HKs was assayed 30 min after the temperature shift. Nonpermeabilized cells were immunoblotted with mouse anti-AnxA2 Ab (1:200) and with rabbit anti-HPV16 VLP Ab (1:200). Z-stack images were collected randomly for mock- and HPV16-infected slides based on DAPI staining. SlideBook image analysis software was used to analyze z-stacks of 51 cells for each sample for mean fluorescence intensity of the fluorescein isothiocyanate (FITC) channel, representing the AnxA2 surface staining. For HPV16- infected samples after analysis, we verified that cells included in analysis were bound by virus particles. Error bars represent standard errors of the means (SEM) for the FITC channel intensity.

AnxA2 expression in C4-2 cells was detected with mouse anti-AnxA2 Ab on either Triton X-100-treated cells to verify intracellular protein expression or on nonpermeabilized cells to determine the presence of cell surface-bound AnxA2 at 24 h posttransfection. HPV16 binding and internalization were analyzed at 6 h postinfection using rabbit anti-HPV16 VLPs. In selected experiments cell were stained with WGA (green) to determine the location of HPV16 relative to the cell surface at 6 h p.i.

Immunoprecipitation and immunoblotting.

Mock- or HPV16-exposed cells were collected at 30 min and 1 h after the temperature shift and lysed using RIPA buffer (150 mM NaCl, 50 mM Tris HCl, 1% NP-40, 0.5% sodium deoxycholate, 1 mM EDTA, 1× Halt protease and phosphatase inhibitor cocktail [Thermo Scientific]) or modified RIPA buffer (150 mM NaCl, 50 mM Tris HCl, 1% NP-40, 0.5% sodium deoxycholate, 5 mM CaCl₂, 1× Halt protease and phosphatase inhibitor cocktail without EDTA). For protein cross-linking, 1% formaldehyde (FA; JD Baker; 10 min at RT) or 1 mM DTSSP (3,3′-dithiobis [sulfosuccinimidylpropionate]; Thermal Scientific; 2 h on ice) was used prior to protein extraction. Cell lysates were cleared by centrifugation (5 min, 1,000 × g), and supernatants were collected and incubated with rabbit anti-AnxA2 Ab for at least 2 h at 4°C and subsequently incubated for 1 h with 1% bovine serum albumin (BSA)-preblocked protein A-Sepharose CL-4B beads (GE Healthcare). Beads were washed three times with ice-cold lysis buffer, solubilized with SDS-sample loading buffer, and heated for 5 min at 95°C. Samples were resolved by 10% SDS-PAGE and transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blotted with mouse anti-HPV16 L1 Ab (1:10,000), followed by rabbit anti-mouse horseradish peroxidase (HRP)-conjugated secondary antibody (1:10,000). In selected experiments rabbit anti-HPV16 VLP primary Ab was used to IP virus, followed by immunoblotting with mouse anti-AnxA2 Ab (1:500). For analysis of EGFR and HSPG copurification with AnxA2, mouse or rabbit anti-AnxA2 Ab, respectively, was used. Membranes were blotted with either rabbit anti-EGFR Ab (1:2,000) or mouse anti-HSPG Ab (1:1,000). Mock-infected cells and protein A beads only without primary antibody were used as controls in experiments. To analyze the role of HSPG in AnxA2-HPV16 interaction low-molecular-weight heparin (500 μM; Sigma) was used as a specific competitor. Cells were preincubated with heparin for 30 min prior to and during virus binding or heparin was added after HPV16 binding for 30 min, followed by IP of AnxA2 and immunoblotting for HPV16 L1 protein. In one experiment cells were incubated with heparinase I (5 U/ml; Sigma) for 30 min prior to and during virus binding, followed by IP of AnxA2 and immunoblotting for HPV16 L1 protein.

For IP of AnxA2, samples were generated as stated above except 0.1% SDS was present in RIPA buffer. Rabbit anti-AnxA2 Ab was used to pull down AnxA2, and membranes were immunostained with mouse anti-pTyr23-AnxA2Ab (1:500) and mouse anti-AnxA2 Ab (1:500) for a loading control. For IP of Src kinase, rabbit anti-Src Ab was used and membranes were immunostained with mouse antiphosphotyrosine Ab (1:1,000); mouse anti-rabbit Ab (1:10,000) was used to detect heavy and light chains as a loading control.

Expression of AnxA2 in the C4-2 cell line.

C4-2 cells were transfected with AnxA2 expression plasmid, control EGFP expression plasmid, or p11 expression plasmid using Lipofectamine 2000 (Invitrogen) by following the manufacturer protocol. Briefly 4 μg of each plasmid was transfected with 8 μl of lipids for 24 h. At 24 h posttransfection the cells were refed with fresh media prior to infection with HPV16 (100 vge/cell) by following a standard protocol. Samples were analyzed 24 h later for luciferase expression. Data are expressed as percentages of infection; HPV16 infection of control vector-transfected cells is set to 100%. To verify AnxA2 expression levels, separate samples collected 24 h posttransfection were used for IP with mouse anti-AnxA2 Ab and analyzed by immunoblotting using rabbit anti-AnxA2 Ab. IgG blotting was used as a loading control.

Cell cycle analysis of HaCaT cells treated with anti-AnxA2, anti-p11 antibodies, EGTA, PP2 inhibitor, and siRNA.

HK cells (2 × 10⁵) were either left untreated or treated for 24 h with anti-AnxA2 Ab (15 μg/ml), anti-p11 Ab (15 μg/ml), control Ab (15 μg/ml), and various concentrations of EGTA and PP2 inhibitor, as indicated on the figures. For siRNA knockdown, cells were transfected for 5 days prior to cell cycle analysis. Cells were trypsinized, ethanol fixed for 1 h at 4°C, and stained with propidium iodide (PI) (30 min, RT) (70). Samples were analyzed using propidium iodide staining and FACSCalibur to determine the numbers of cells at G₁, S, and G₂/M phases (Flow Cytometry Shared Resource Center supported by the University of New Mexico Health Sciences Center and the University of New Mexico Cancer Center). The number of cells in each cell cycle phase is expressed as a percentage of the total cell count.

RESULTS

HPV16 PsV exposure to HaCaT cells induces Src-dependent AnxA2 phosphorylation at Tyr23.

Pseudovirions that encapsidate a reporter plasmid are a safe and biologically relevant alternative to working with infectious HPV16 with oncogenic potential (63, 68). We previously showed that HPV PsVs, upon initial attachment to cell surface HSPGs, become decorated with HSPGs and GFs, such as EGF and other EGFR ligands, and can be released as soluble, high-molecular-weight complexes. These HPV-associated GFs activate signaling cascades essential for virus entry, suggesting that the cognate signaling GFRs, such as EGFR, are regulators of HPV uptake (17, 69). As Src is phosphorylated as a result of EGFR activation (71, 72), we tested the induction of Src protein kinase activation following HPV16 exposure. Src became phosphorylated within 5 min of HPV16 exposure to HaCaT cells and persisted for 30 min (Fig. 1A). PD168393, a specific inhibitor of EGFR activation and HPV infection (17), inhibited this HPV-mediated Src phosphorylation (Fig. 1B and C). PD168393 blunted total Tyr phosphorylation of Src, when normalized to total protein loaded, by nearly 3-fold compared to that for HPV16-exposed cells in the absence of treatment (Fig. 1C). We previously showed that HPV-induced signaling elicited through EGFR is slower and more prolonged than that from free EGF; the latter signaling peaks at 5 min and returns to basal levels by 30 min postexposure (17). Similarly, Src activation returns to basal levels at 30 min after exposure to EGF (Fig. 1B, lane 4), but the PsV-stimulated activation of Src is more prolonged and mirrors that seen for PsV-induced activation of EGFR (Fig. 1A) (17). We recently reported that PP2, a Src kinase-specific inhibitor, reduces HPV16 infectivity. PP2 was found to have a 50% inhibitory concentration of 1.5 × 10⁻⁶ M without affecting PsV attachment to HaCaT cells (69). Herein, we confirmed the importance of Src signaling for HPV16 infection, showing that specific inhibition by PP2 reduced HPV infection in a dose-dependent manner in serum-containing media (Fig. 1D). As HPV16 infectious entry into HKs requires cell cycle progression (73), we evaluated the effects of PP2 inhibitor treatments on the cell cycle. The data showed that the alterations in the cell cycle under these conditions could not fully account for the reduction in infection levels (Fig. 1H). For example, the 1.0 and 5.0 mM PP2 treatments caused an increase of 5% to 15% in cells in G₁ compared to mock-treated cells. However, cell infectivity was decreased by ≈50% with both treatments (Fig. 1D).

Fig 1.

HPV16 exposure induces Src kinase-dependent AnxA2 phosphorylation at Tyr23. HaCaT cells serum starved for 4 to 6 h were mock (M) or HPV16 (H16) exposed (100 vge/cell) for the indicated periods of time (A, E, and F) before harvest for immunoprecipitation (IP). (A) IP was performed with rabbit anti-Src Ab followed by immunoblotting with mouse pTyr Ab (upper blot). IgG was detected as a loading control (lower blot). (B) Cells were treated with 100 nm PD168393 prior to and during virus exposure (30 min) or with 5 ng/ml EGF (30 min postaddition) as a positive control. IP was performed with anti-Src Ab followed by immunoblotting for anti-pTyr Ab. IgG was detected as a loading control (lower blot). (C) Levels of pTyr Src normalized to IgG loading controls for various treatments based on data in panel B from three independent IP experiments; bars represent standard deviations (*, P = 0.0297 as determined by Student's t test). Ø, no treatment. (D) HaCaT cells were mock or HPV16 infected (100 vge/cell) in the presence of different concentrations of PP2. Infectivity was determined 24 h later, and data were graphed from three independent experiments; bars represent standard deviations (*, P = 0.0268 by Student's t test). (E) IP was performed with rabbit anti-AnxA2 Ab followed by immunoblotting with mouse anti-pTyr23 AnxA2 Ab (upper blot). Mouse anti-AnxA2 Ab was used to detect total protein as a sample loading control (lower blot). (F) Levels of pTyr 23 AnxA2 normalized to total AnxA2 protein for each time based on data from three independent IP experiments from panel E; bars represent standard deviations (*, P = 0.0038 by Student's t test). (G) Cells were treated with 1 μM PP2 inhibitor, 10 μg/ml N terminus anti-AnxA2 Ab, or 100 nM PD168393 (PD) prior to and during virus exposure for 30 min. IP was performed with rabbit anti-AnxA2 Ab followed by immunoblotting with mouse anti-pTyr23 AnxA2 Ab (upper blot). Mouse anti-AnxA2 Ab was used to detect total protein (lower blot). (H) The fractions of cells in G₁ (1n), S (intermediate), and G₂/M (2n) phases were expressed as percentages of the total cells counted. HaCaT cells were untreated (M) or treated with anti-AnxA2 Ab (15 μg/ml), anti-p11 Ab (15 μg/ml), control Ab (15 μg/ml), EGTA (0.1, 0.5, or 1.0 mM), or PP2 (0.1, 0.5, 1.0, or 5.0 mM) for 24 h.

As AnxA2 is a prominent substrate for Src kinase (28), we tested whether AnxA2 becomes phosphorylated upon low-dose exposure to HPV16. Serum-starved HaCaT cells were either mock or HPV16 exposed for 0, 5, 15, 30, and 60 min, which led to AnxA2 Tyr23 phosphorylation within 5 min, reaching a maximum by 30 min (Fig. 1E, F). AnxA2 Tyr23 remained phosphorylated at higher levels for 2 h after HPV16 exposure (data not shown). PP2 fully blocked AnxA2 phosphorylation at Tyr23 (Fig. 1G, lane 3), indicating that AnxA2 activation is downstream of Src kinase. Additionally, PD168393, a specific inhibitor of EGFR tyrosine kinase, also completely prevented AnxA2 phosphorylation (Fig. 1G, lane 5). These data indicate that HPV16-induced phosphorylation of AnxA2 is catalyzed by Src kinase, which is activated by EGFR signaling. EGFR and Src signaling is more prolonged in the presence of HPV and is important for HPV infection of HKs.

HaCaT cells exposed to HPV16 initiate AnxA2 translocation to the outer surface of the plasma membrane.

Consequences of AnxA2 phosphorylation at Tyr23 include increased AnxA2 tetramer (A2t) translocation to the extracellular side of the plasma membrane (30, 31) and association with assembling endosomes (34). To determine whether low-dose HPV16 exposure that leads to AnxA2 phosphorylation at Tyr23 altered the localization of AnxA2, we prepared cytoplasmic and membrane fractions from cells mock exposed and HPV16 exposed. AnxA2 was detected in both cytoplasmic and membrane fractions of HaCaT cells (Fig. 2A and B), but there was an average of ≈2.5-fold-higher association of AnxA2 with the total membrane fractions in cells exposed to virus for 30 min than in mock-exposed cells (Fig. 2A, lanes 3 and 4, and B). Confocal microscopy was used to visualize the AnxA2 present at the extracellular surface of the plasma membrane in mock- and HPV16-exposed cells. Quantitative analysis of all planes from 3D images showed that, upon HPV16 exposure, nearly 1.6-fold more AnxA2 was cell surface bound than in mock-exposed cells at 30 min p.i. (Fig. 2C and D). These data imply that HPV16-induced EGFR and Src kinase activation leads to AnxA2 phosphorylation, resulting in increased AnxA2 translocation to membranes and, in particular, to the extracellular leaflet of the HK plasma membrane.

Fig 2.

AnxA2 translocates to the cell surface in HPV16-exposed cells. (A) Immunoblot of AnxA2 levels in the cytoplasmic and membrane fractions isolated from mock- and HPV16-exposed (100 vge/cell) HaCaT cells at 30 min p.i. α-Tubulin was detected as a cytoplasmic protein marker and Na/K ATPase α 1 as a membrane protein indicator. (B) The AnxA2 bands from immunoblots in panel A were analyzed with ImageJ, and the value of the raw integrated density (Raw IntDen) was plotted from two independent experiments (*, P = 0.0107 by Student's t test). (C) Confocal microscopy images of representative mock- and HPV16-exposed (100 vge/cell) cells at 30 min p.i. Nonpermeabilized cells were immunostained for AnxA2 (green) and HPV16 (red; anti-HPV16 VLP Ab). Arrows indicate yellow colocalization of red virus particles with green AnxA2. Bars = 10 μm. (D) Acquired confocal microscopy z-stack images were analyzed for FITC channel intensity representing AnxA2 staining at the plasma membrane. Cells were randomly selected for analysis based on the DAPI staining (n = 51 cells per group quantified). Error bars represent standard errors of the means (***, P = 0.0009 by Student's t test).

HPV16 PsVs associate with A2t at the plasma membrane in a Ca²⁺-dependent manner.

In prior work we showed that HSPG/GF-decorated HPV particles associate with cognate GFRs to activate signaling cascades important for virus entry (17). To determine whether HPV16 PsVs interact with AnxA2 on the cell surface of HKs, we used confocal microscopy and coimmunoprecipitation (co-IP). Figure 3A shows evidence for colocalization of HPV16 with AnxA2 at the plasma membrane after cells were incubated with PsVs for 30 min to promote the infectious process. For comparison, a mock-exposed cell stained for AnxA2 and HPV16 is shown. Consistent with the fact that membrane-bound AnxA2 exists in a tetramer (A2t) composed of two AnxA2 proteins and two S100A10/p11 proteins, HPV16 also colocalized with S100A10/p11 on the external cell plasma membrane (Fig. 3B).

Fig 3.

HPV16 associates with AnxA2 at the cellular membranes in the presence of calcium. (A, B) HPV16 colocalizes with AnxA2 and p11 at the plasma membrane of nonpermeabilized cells. HaCaT cells were either mock or HPV16 exposed (5,000 vge/cell) at 37°C for 30 min after the temperature shift. Images are two-dimensional (2D) views, with cell cross sections shown to the right of each frame. Bars = 10 μm. Yellow indicates overlap of green and red signals. (A) Mock- and HPV16-exposed cells were immunostained for AnxA2 (green) and HPV16 (red, anti-HPV16 VLP Ab). (B) HPV16-exposed cells were immunostained for S100A10/p11 (green) and HPV16 (red, anti-HPV16 VLP Ab). (C to E) HaCaT cells were mock or HPV16 exposed (100 vge/cell) at 37°C for 1 h after the temperature shift prior to cell lysis and IP. Immunoblotting was performed to detect HPV16 L1 and AnxA2. (C) Cell lysates were used for IP with anti-HPV16 VLP Ab followed by immunoblotting for AnxA2 in the presence of 5 mM CaCl₂. (D) Cells were either directly lysed in the absence or presence of calcium (−Ca2+ and +Ca2+) or cross-linked with DTSSP or formaldehyde (FA) in the absence of calcium prior to lysis. Cell lysates were used for IP with anti-AnxA2 Ab, followed by analysis of copurifying HPV16 L1. (E) Cell lysates were fractionated by differential centrifugation to separate membranes and cytosol. Each fraction was subjected to IP with anti-AnxA2 Ab. Blots were probed with anti-HPV 16 L1 Ab. Asterisks, nonspecific bands (D and E). Input lysates (prior to IP) and supernatants that remained unbound to the Ab–protein A-Sepharose beads (unb) are indicated. (F and G) HaCaT cells were pretreated with the indicated concentrations of EGTA for 1 h at 37°C, then were either mock or HPV16 exposed in the presence of EGTA for 1 h at 4°C to allow for cell surface binding. (F) Media were analyzed for unbound virus particles (top) and AnxA2 released from cell surface (bottom) upon EGTA treatment. (G) Infections were analyzed 24 h later; additionally, HPV16 PsVs were EGTA treated for 1 h at 37°C, repurified, and used to infect HaCaT cells (white bar). Data were graphed from three independent experiments. Error bars represent standard deviations (*, P = 0.0173 by Student's t test). (H) Morphology of the untreated (left) and EGTA-treated HPV16 (right) visualized by transmission electron microscopy. Bars = 100 nm.

The predominant plasma membrane-associated A2t form requires Ca²⁺ for plasma membrane binding, and AnxA2 associates with other ligands, like the glycosoaminoglycans heparin and heparan sulfate, in Ca²⁺-dependent modes (74). However, the phosphorylated form of AnxA2 is also reported to bind membranes in a Ca²⁺-independent manner (39, 75). We reasoned that, if Ca²⁺ is required for AnxA2 and HPV16 association, IP conditions in the presence of EDTA might disrupt the Ca²⁺-dependent AnxA2 structure and/or interactions of AnxA2 with potential binding partners. Consistent with this, there was no detectable interaction between AnxA2 and HPV16 when EDTA was included during the AnxA2-specific IP (Fig. 3D, lane 3). However, in the absence of EDTA and presence of 5 mM Ca²⁺, we detected an HPV16 and AnxA2 association (Fig. 3D, lane 7); also, reciprocal IP of HPV16 in the presence of Ca²⁺ copurified AnxA2 (Fig. 3C). These findings indicate that Ca²⁺ is needed for the interaction of HPV16 and AnxA2 at the HK plasma membrane.

To test if the AnxA2-HPV16 interaction takes place at the extracellular leaflet of the plasma membrane prior to virus entry, cross-linking agents were employed before cell lysis. DTSSP is membrane impermeable and cross-links proteins within a distance of ≈12 Å, making it useful for analysis of protein interactions at the cell surface. DTSSP cross-linking facilitated co-IP of AnxA2 with HPV16 (Fig. 3D, lane 9), suggesting that AnxA2 and HPV16 interact on the plasma membrane prior to virus uptake by the cell. In contrast, formaldehyde, a cell-permeable, reversible cross-linker with ∼2- to 3-Å cross-linking ability, failed to permit copurification of HPV16 with AnxA2 by co-IP (Fig. 3D, lane 13). Instead, HPV16 was found in the unbound supernatant following IP for AnxA2 (Fig. 3D, lane 12). These data confirm the results with microscopy that suggested colocalization of HPV16 and AnxA2. The larger distance allowed for protein cross-linking with DTSSP and not formaldehyde suggests that the AnxA2-HPV16 interaction is not direct and that Ca²⁺ may be a bridging molecule. However, other molecules in addition to Ca²⁺ could also be involved in the interaction between AnxA2 and HPV16.

To ascertain whether the association between AnxA2 and HPV16 takes place only at cellular membranes, cells were fractionated into membrane and cytoplasmic components in the presence of Ca²⁺. HPV16 failed to co-IP with AnxA2 from the cytosolic fraction (Fig. 3E, lane 8). However, an interaction between AnxA2 and virus was detected in the membrane fraction (Fig. 3E, lane 6), suggesting that membrane-associated AnxA2 is the main form targeted for HPV16 binding. Together, these data suggest that HPV16 and AnxA2 associate at the HK plasma membrane and Ca²⁺, if not other factors, mediate this interaction.

As an additional means of assessing the need for Ca²⁺ in HPV16 binding to the surface of the cell membrane, EGTA was used to chelate Ca²⁺ from HaCaT cells. We observed an increase of unbound virus in media collected from virus-exposed cells in the presence of increasing EGTA concentrations (Fig. 3F, lanes 2 to 5). We found a dose-dependent, albeit slight increase in AnxA2 released from the cell surface upon EGTA treatment, consistent with previous reports (76). The low level of AnxA2 release is expected since phosphorylated AnxA2 binding to membranes is independent of Ca²⁺ (39, 75). This experiment corroborates the co-IP finding that Ca²⁺ is necessary for HPV16 association with AnxA2 at the plasma membrane.

Additionally, HPV16 infection of HaCaT cells was inhibited in an EGTA dose-dependent manner, consistent with the loss of membrane-bound HPV16 from cells (Fig. 3G). As EGTA can alter cell proliferation (77), we evaluated the effects of EGTA treatments on the cell cycle. The data showed that the EGTA treatments did not cause significant cell cycle perturbations (Fig. 1H); at 1 mM EGTA, a <10% change in the cells in G₁ could not fully account for the >45% reduction in infection (Fig. 1H and 3G, respectively). We also confirmed that EGTA treatments did not affect the viability of cells using trypan blue exclusion staining (data not shown). To rule out the effects of EGTA on virion integrity, we treated HPV16 PsVs with EGTA prior to cell exposure and investigated infectivity and morphology. The results revealed that virions treated with EGTA exhibited normal infectivity (Fig. 3G) and, by transmission electron microscopy, no gross morphological alterations compared with untreated virions (Fig. 3H). Although EGTA could affect many cellular interactions important for virus binding and infection, the strong correlation between decreased HPV16 infection and inhibited HPV16 binding to the cell surface supports a role for Ca²⁺-dependent A2t-HPV16 interaction in the process of infection.

AnxA2 and S100A10 regulate HPV16 entry, proper intracellular trafficking, and efficient infection of HKs.

The importance of AnxA2 in HPV16 infection of HaCaT cells was investigated using an affinity-purified rabbit anti-AnxA2 Ab specific to N-terminal amino acids 1 to 50 in the PsV infectious-entry assay. Cell treatment with this antibody did not appreciably inhibit or compete with AnxA2 phosphorylation at Tyr23 (Fig. 1G, lane 4). HaCaT cell monolayers were left untreated or pretreated for 1 h with increasing concentrations of anti-AnxA2 Ab or with control antibody. Cells were then exposed to HPV16 PsVs for 24 h in medium containing each respective antibody. The infection level in cells treated with the control antibodies did not differ from that of the untreated cells (Fig. 4A). In contrast, HPV16 pseudovirus infection in the presence of the N-terminal anti-AnxA2 Ab was severely diminished in a dose-dependent fashion, with nearly 90% inhibition at 15 μg/ml. Woodham et al. showed only 30% block of HPV16 infection in HaCaT cells using 20 μg/ml of a core domain-specific AnxA2 antibody; with 40 μg/ml of the same antibody, infection was reduced by about 70% (60). It is unclear whether differences between the two antibodies are related to antibody-antigen affinity or the mechanism of infection inhibition. Since membrane-bound AnxA2 exists predominantly in a heterotetrameric form with S100A10/p11, we tested the effect of mouse anti-p11 Ab treatment on HPV16 infection. The p11 antibody also blocked HPV16 pseudovirus infection of HK cells (Fig. 4A). The monomeric form of AnxA2 may affect DNA synthesis and cell proliferation, and its expression is strictly regulated during the mammalian cell cycle (78). Thus, we analyzed whether the infection inhibition observed might result from cell cycle deregulation. Neither the anti-AnxA2 Ab nor the anti-p11 Ab treatment significantly affected the percentages of cells in G₁, S, and G₂/M (Fig. 1H). All antibodies used contain the stabilizing compound NaN₃, which at a concentration of ≥0.05% inhibits bovine papillomavirus (PV) infection (23). Although NaN₃ was present at a final concentration of ≤0.001% in our infectious-entry assays, we tested the susceptibility of HPV16 infection of HaCaT cells to NaN₃ suppression. HPV16 infectivity remained unchanged in the presence of up to 0.01% NaN₃ (Fig. 4A), a level 10-fold higher than that present in the AnxA2, p11, and control antibody-treated samples. The data strongly suggest that the AnxA2 tetramer is involved in HPV16 infection of HKs.

Fig 4.

AnxA2 regulates HPV16 infection. (A) HaCaT cells were pretreated with increasing concentrations of antibodies to the AnxA2 N terminus (amino acids 1 to 50), p11, or nonspecific control antibodies C_a (affinity-purified rabbit anti-mouse IgG) and C_b (mouse anti-PCNA antibody), followed by HPV16 PsV exposure (100 vge/cell). Virus was allowed to attach at 4°C in the presence of antibodies. Unbound viruses were removed, and cells were refed with fresh medium containing the corresponding antibody. Infections were continued at 37°C for 24 h. Alternatively HPV16 infection in the presence of 0.01% NaN₃ was monitored. Graphs were generated based on three independent experiments. Error bars represent standard deviations (***, P < 0.0001 as determined by Student's t test). (B) HaCaT cells were transfected with control or AnxA2-specific siRNAs for 6 days; AnxA2 protein knockdown was determined by immunoblotting, with tubulin detected as a control. The left graph represents quantified AnxA2 levels normalized to tubulin from immunoblots from three separate transfections. Error bars represent the average AnxA2 levels normalized to α-tubulin in control and AnxA2 siRNA-transfected cells from the three independent experiments (*, P = 0.029 by Student's t test). The middle graph shows data for siRNA-transfected cells infected with HPV16 PsVs (100 vge/cell) 6 days posttransfection. Error bars represent the averages of luciferase readings from the three independent infections (***, P = 0.0001 by Student's t test). The right graph represents cell cycle analyses of HaCaT cells transfected with control and AnxA2 siRNA (100 nM) for a period of 5 days. The fractions of cells in G₁ (1n), S (intermediate), and G₂/M (2n) phases were expressed as percentages of the total cells counted. (C to J) HaCaT cells were exposed to HPV16 PsVs (5,000 vge/cell) in the absence of Ab (C, I) or presence of control antibody (C_a) (D), rabbit anti-AnxA2 N terminus Ab (E), or anti-p11 Ab (F, G, H, J). All antibodies were used at a 10-μg/ml concentration. Confocal microscopy was performed 24 h p.i. by staining for HPV16 (red), plasma membrane (WGA, green), and nuclei (DAPI, blue) (C to F), for HPV16 (red) and EEA1 (green) (G), for LAMP1 (green) (H, I), or for AnxA2 (green) (J). Quantification of HPV16 and plasma membrane colocalization is depicted at each image (C to F). Numbers represent the percentages of the total HPV16 signal that colocalizes with plasma membrane signal as an average of 6 cells per group. Side views of images were used for colocalization analysis. Yellow indicates colocalization of green and red signal. Bars = 10 μm.

Inhibiting AnxA2 expression using siRNA knockdown in HaCaT cells similarly reduced HPV16 infection. The transfection efficiency of HaCaT cells was determined to be ≈70% using fluorescein-labeled control siRNA. Immunoblot analysis revealed maximal AnxA2 knockdown at 6 to 7 days after siRNA transfection, but total AnxA2 levels were lowered by ≈45% (Fig. 4B, lower left). Consistent with the decreased level of AnxA2 expression, HPV16 PsV infection levels at 6 days posttransfection were reproducibly and significantly decreased by ≈40% (Fig. 4B, lower middle). AnxA2 siRNA knockdown caused only slight (≈5%) cell cycle perturbations compared to control siRNA-transfected cells (Fig. 4B, lower right). The decreased levels of AnxA2 and reduced infection upon siRNA knockdown shown in Fig. 4B closely parallel those reported by Woodham et al. in HeLa cells stably transduced with short hairpin RNA (shRNA) against AnxA2 (60). This further substantiates a role for AnxA2 and/or A2t in HPV16 infection of normal HKs.

Confocal microscopy was used to investigate the location of the antibody-mediated block in HPV16 infection under conditions used in the infectious-entry assay (Fig. 4A). Quantitative colocalization analysis of microscopy cross sections imaging HPV16 PsVs with a plasma membrane marker revealed that HPV16 PsVs were localized predominantly at locations internal to the stained plasma membrane in untreated infected cells at 24 h p.i.; only 12% ± 3% of the red PsV signal colocalized at the plasma membrane (Fig. 4C). The control antibody that did not affect infection yielded similar findings (Fig. 4D). In contrast, when cells were pretreated with anti-AnxA2 N terminus antibody, 51% ± 5% of the bound HPV16 was retained at the plasma membrane (Fig. 4E). The anti-AnxA2 antibody did not seem to prevent HPV16 binding to the cell surface; however, it severely inhibited or slowed virus internalization and infection. These findings agree fully with results showing that genetic knockdown of AnxA2 prevents HPV16 PsV entry in HeLa cells (60). Interestingly, when HPV16 particle localization was visualized in the presence of anti-p11 Ab, no block in entry was observed. HPV16 PsVs colocalized with the plasma membrane at low levels, similar to those in untreated cells (11% ± 4%) (Fig. 4F). However, compared to the control cells at 24 h p.i. (Fig. 4C and D), the anti-p11 Ab altered the distribution of intracellular virus, causing HPV16 capsid protein to accumulate at perinuclear regions (Fig. 4F). Colocalization between HPV16 and early endosomal marker EEA1 also was observed in the presence of anti-p11 treatment (Fig. 4G). Immunostaining showed that HPV16 PsVs in the presence of anti-p11 Ab were predominantly retained in the late endosomal/lysosomal compartment, localizing with LAMP1 and, to a lesser extent, with AnxA2 (Fig. 4H and J, respectively). This differed from the distribution of PsVs at 24 h p.i. in the absence of anti-p11 Ab, where less HPV16 L1 was colocalized with LAMP1 (Fig. 4I). The pattern of intracellular AnxA2 in the presence of p11 Ab resembled that observed with a transdominant mutant chimeric AnxA2-p11 protein, which caused endosomal aggregation (79). Results suggest that anti-p11 Ab treatment may affect membrane sorting or fusions and/or inhibit capsid disassembly at the late endosomal/lysosomal compartment, thus preventing nuclear delivery of viral genomes. Together, these data suggest that the AnxA2 and S100A10/p11 proteins not only interact with HPV16 prior to cell entry but also may regulate proper HPV16 trafficking through the endosomal pathway to initiate infection.

Exogenous expression of AnxA2 increases susceptibility of the AnxA2-deficient C4-2 epithelial cell line to HPV16 infection.

C4-2 cells are derived from the prostate adenocarcinoma LNCaP cell line, which lacks AnxA2 p36 expression as the result of promoter hypermethylation (65). The cells are also resistant to HPV16 pseudovirus infection compared to HaCaT cells (Fig. 5A). To determine whether exogenous AnxA2 expression could increase the susceptibility of the C4-2 cell line to HPV16 PsV infection, cells were transfected with an AnxA2 expression plasmid encoding a full-length, tag-free protein. The transfection efficiency of C4-2 cells was estimated to be ≈7 to 8% in side-by-side transfections based on flow cytometry analysis of intracellular expression of an AnxA2-EGFP fusion protein (data not shown). At 24 h posttransfection, C4-2 cells were exposed to HPV16 PsVs and pseudovirus infection was analyzed 24 h later. Exogenous expression of AnxA2 (Fig. 5B, lane 4) resulted in a statistically significant 2- to 3-fold increase in HPV16 infectivity in C4-2 cells (Fig. 5A), which was noteworthy given that transfection efficiency was quite low. As expected, transfection of C4-2 cells with a p11 expression plasmid alone had no effect on HPV16 infection (Fig. 5A). We were unable to verify p11 expression by immunoblotting as the anti-p11 antibody does not recognize SDS-denatured forms of S100A10. However, results from immunofluorescence confocal microscopy suggested that the S100A10/p11 protein was present in C4-2 cells at very low levels and that p11 levels increased over 3-fold upon AnxA2 expression (data not shown). More importantly, AnxA2 p36 was detected at the cell surface in AnxA2 plasmid-transfected C4-2 cells (Fig. 5E). This implies that the cells contain S100A10/p11, which is stabilized by AnxA2 via a posttranslational mechanism that enables the surface localization of Tyr23-phosphorylated AnxA2 (31, 48).

Fig 5.

Exogenous expression of AnxA2 increases the susceptibility of the AnxA2-deficient C4-2 cell line to HPV16 infection. (A) C4-2 cells were transfected with a p11 expression plasmid, an AnxA2 expression plasmid, or an EGFP expression plasmid as a control (C) for 24 h prior to infection. Cells were mock or HPV16 infected (100 vge/cell) for 24 h, at which point they were analyzed for pseudovirus infection. Infection of C4-2 cells is expressed as % of infection relative to HPV16 infection of untreated HaCaT cells (set as 100%). Graphs were generated based on four independent experiments. Error bars represent standard deviations (*, P = 0.0442 by Student's t test). (B) Representative immunoblot with specific rabbit anti-AnxA2 Ab of untransfected C4-2 cells, control- and p11 vector-transfected C4-2 cells, and AnxA2 plasmid-transfected C4-2 cells. (C to J) 2D confocal microscopy images and side views of C4-2 cells. C4-2 cells untransfected or transfected with AnxA2 expression plasmid for 24 h were mock or HPV16 exposed for 6 h postbinding. (C) Untransfected C4-2 cells stained for AnxA2 in permeabilized cells. (D, E) AnxA2 plasmid-transfected mock-exposed C4-2 cells analyzed for intracellular AnxA2 expression (green) and cell surface AnxA2 presence in permeabilized cells (D) and AnxA2 cell surface expression in transfected, HPV16 exposed, nonpermeabilized cells (E). (F to H) HPV16 PsV localization (red) at 6 h p.i. in AnxA2-negative C4-2 cells (F) and transfected-AnxA2-expressing C4-2 cells (green; G and H). (I, J) C4-2 cells were WGA stained (green) prior to permeabilization. HPV16 PsV localization (red) in AnxA2-negative C4-2 cells (I) and AnxA2-transfected C4-2 cells (J). Cells in panels C, D, and F to J were permeabilized with 0.1% Triton X-100. Bars = 10 μm. Yellow indicates colocalization of green and red signal. Blue arrows, cell surface-bound HPV16; white arrows, intracellularly located HPV16.

Immunofluorescence microscopy analysis confirmed the lack of AnxA2 expression in C4-2 cells (Fig. 5C) and the presence of exogenous AnxA2 in transfected cells 24 h later (Fig. 5D). As indicated above, AnxA2 translocated to the cell surface of AnxA2 plasmid-transfected C4-2 cells upon HPV16 exposure (Fig. 5E). Yet, in the absence of HPV16 exposure, exogenous expression of AnxA2 in C4-2 cells was detected at the cell surface at lower levels (data not shown). This finding is consistent with our data that HPV16-dependent stimulation of AnxA2, likely thorough Tyr23 phosphorylation (as in Fig. 1D), results in AnxA2 translocation across the plasma membrane, as observed in HaCaT cells (Fig. 2C and D). Confocal microscopy cross sections revealed that HPV16 PsVs bound to the cell surface of both AnxA2-null and AnxA2-expressing C4-2 cells (Fig. 5F and I and G, H, and J, respectively) at 6 h after virus exposure. However, HPV16 PsVs were internalized only in AnxA2-expressing cells (Fig. 5G and H). Plasma membrane staining with WGA confirmed that HPV16 PsVs bound at the cell surface of AnxA2-null cells (Fig. 5I) but were internalized in AnxA2-expressing cells (Fig. 5J). These findings strongly support the idea that AnxA2 protein, possibly with S100A10/p11 as A2t, is required for efficient HPV16 uptake.

AnxA2 associates with HPV16 and EGFR at the cell surface of HKs.

As reported previously, we showed that HPV16 particles become decorated with various cellular components such as syndecan-1, HSPGs, and GFs and associate with cognate GFRs at the plasma membrane (17). Since AnxA2 links actin filaments with membrane domains in a dynamic and regulated fashion to enable lateral movements and/or endocytosis of EGFR upon activation (45), we tested whether AnxA2 and EGFR colocalize at the surface of HaCaT cells in the presence of HPV16 PsVs. Confocal microscopy showed that AnxA2 and EGFR colocalize at the plasma membrane of HaCaT cells in the absence (data not shown) and presence (Fig. 6A) of HPV16 at 30 min p.i. Likewise, at 30 min p.i., HPV16 and EGFR colocalize substantially (Fig. 6B) compared to lower colocalization at 15 min p.i. after serum starvation, as previously reported (17). Co-IP analysis showed that AnxA2 and EGFR associate in HaCaT cells in the absence and presence of HPV16 (Fig. 6C, lanes 3 and 4, respectively). Treatment of cells with the impermeable cross-linker DTSSP prior to cell lysis confirmed that these interactions take place at the cell surface (Fig. 6C, lanes 5 and 6). As AnxA2 reportedly associates with heparin (74, 80), we tested this interaction in HaCaT cells and found that AnxA2 and HSPGs are associated in the presence and absence of HPV16 particles (Fig. 6D). To determine whether HSPGs may act as bridging molecules between AnxA2 and HPV16, we treated cells with heparinase I and achieved an ≈50% decrease in the HSPG levels at the cell surface as detected by FACS (data not shown). This correlated well with reduced HPV16 binding to AnxA2 (Fig. 6E). We cannot exclude the possibility that the reduced infection under these conditions results from decreased capsid conformational changes proposed to be needed prior to secondary receptor binding (8, 9). Additionally, we found that heparin, used as a specific competitor before and during or after HPV16 attachment to cells reduced HPV16 binding to AnxA2 (Fig. 6F, lanes 3 and 4). These data suggest that heparin not only competes away the HPV16 association with the initial HSPG attachment factors, as has been demonstrated (3, 4), but also competes with the subsequent transfer and/or association with secondary internalization receptors like AnxA2. Together, these data support a role for HSPGs in bridging AnxA2 and HPV16 and further support our model that virion decoration with HSPGs is important after initial HPV-cell interactions.

Fig 6.

AnxA2 associates with HPV16 and EGFR at the cell surface of HPV16-exposed HaCaT cells. (A, B) 2D confocal microscopy images and side views of nonpermeabilized cells. Bars = 10 μm. (A) Immunostaining of EGFR (green) and AnxA2 (red) at the cell surface 30 min after HPV16 PsV exposure. (B) Immunostaining of HPV16 (red, anti-HPV16 VLP Ab) with EGFR (green) at the cell surface 30 min after infection with HPV16 (5,000 vge/cell). (C) Co-IP of EGFR and HPV16 with AnxA2. Mouse anti-AnxA2 Ab was used to IP AnxA2 from cell lysates of mock- and HPV16-infected HaCaT cells at 30 min p.i. Immunoblots were probed with rabbit anti-AnxA2, rabbit anti-HPV16 VLP, and rabbit anti-EGFR antibody. EGFR was detected in untreated cells and cells cross-linked with DTSSP. (D) Co-IP of HSPG and AnxA2 at 30 min p.i. in mock- and HPV16-exposed HaCaT cells. (E, F) Co-IP of HPV16 with AnxA2. Rabbit anti-AnxA2 was used to IP AnxA2 from HPV16-infected cells at 30 min p.i. Immunoblots were probed with mouse anti-HPV16 L1. IgG was used as a loading control. (E) Cells were untreated or treated with 5 U/ml of heparinase I (Hep. I) 30 min prior to virus exposure. (F) Heparin (500 μM, low molecular weight) was added to culture medium 30 min prior to and during HPV16 binding (Pre) or after virus binding (Post).

DISCUSSION

The process of HPV infection in vitro is unusually lengthy and is strictly regulated by the viral capsid proteins and a number of host factors. Such host elements include proteins that directly interact with the oncoming virions at the plasma membrane, as well as components of cellular endocytic machinery that the virus usurps for entry (reviewed in reference 8). Viruses commonly manipulate and enhance signaling pathways, as they rely on cellular endocytic mechanisms for uptake and genome routing, as well as the intracellular environment for their propagation (reviewed in reference 81). Previously, our laboratory showed that HPV16 particles activate EGFR-induced mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K) pathways, both of which are essential for infection of human keratinocytes (17, 69). However, the cell machinery and the precise steps of HPV entry into host epithelial cells remain to be clearly defined. Recently, Woodham et al. reported that HPV16 PsVs interact with A2t at the cell surface. They showed that an AnxA2 antibody inhibits HPV16 pseudovirus infection and that genetic knockdown of A2t reduces infectious entry. These functions were attributed to HPV L2 protein interaction with S100A10 (60). Our work corroborates a role for A2t in infection, but some of our results and conclusions differ in key ways from the former report. Herein we demonstrate that HPV16 exposure activates signaling cascades to mobilize A2t translocation to the epithelial cell surface to promote virus internalization. Our experiments reveal three important findings with regard to the relationship between A2t and HPV16 infection that provide a common link with previously identified virus-interacting proteins, such as EGFR, EGFR ligands, and HSPGs (17, 25, 80). Furthermore, our findings place HPV-A2t interactions into a physiological context of keratinocyte wound response and provide additional insight into why the bulk of the HPV uptake process in vitro is prolonged (Fig. 7).

Fig 7.

Proposed model for AnxA2 function in HPV16 infection. (A) HSPG- and EGF (also heparin-binding EGF- or amphiregulin)-decorated HPV16 binds EGFR and activates MAPK and PI3K signaling cascades (17, 69). (B, C) Src protein kinase is subsequently activated, which in turn phosphorylates AnxA2 at Tyr23. (D) Association of AnxA2 with p11 and Tyr23 phosphorylation induces the translocation of AnxA2 to the extracellular leaflet of plasma membrane (31) (shown as a bent arrow as the translocation mechanism is not entirely clear). (E) Phosphorylated A2t adopts a conformation wherein the membrane interaction is independent of Ca²⁺. (F) HPV16 recognizes AnxA2 and binds to it in a Ca²⁺-dependent manner. This interaction may be directed, in part, through virus-associated HSPGs, as the common heparin-binding site is situated at the convex face of AnxA2's domain IV (74). AnxA2 associates with cholesterol-containing lipid rafts to regulate endocytic pit formation and initiate endocytosis of AnxA2-bound ligands in an EGFR-dependent or independent mode (94).

First, we showed that the EGFR signaling caused by HPV16 exposure to human keratinocytes (17) initiates Src protein kinase activation, which is important for AnxA2 phosphorylation and increased A2t translocation to the plasma membrane to promote infection (Fig. 7A to C). Src kinase catalyzes tyrosine phosphorylation of various cytoskeletal and cell adhesion proteins, including AnxA2 (45, 82). The EGFR-Src signaling cascade leads to AnxA2 Tyr23 phosphorylation and, along with S100A10/p11 association, promotes A2t translocation to the extracellular leaflet of the plasma membrane by way of a nonclassical endoplasmic reticulum-Golgi route (31, 82, 83) (Fig. 7D). Tyr23 phosphorylation not only regulates AnxA2 extracellular translocation but also influences protein association with endosomal membranes in a Ca²⁺- and phospholipid-independent but cholesterol-dependent manner (35, 84). We propose that virus-induced, Src-mediated AnxA2 Tyr23 phosphorylation may increase the number of AnxA2 tetrameric platforms for HPV16 binding and/or mobilize AnxA2 recruitment to lipid microdomains to stimulate endosome assembly for HPV uptake, analogous to that induced by vesicular stomatitis virus nucleation of clathrin (85) (Fig. 7E). In vitro, extended time periods are likely needed for a population of PsVs to become decorated with HSPGs and GFs, activate EGFR signaling to cause prolonged Src-mediated translocation of A2t platforms for HPV16 binding, promote endosomal assembly, and induce entry. The kinetics of this process will no doubt be asynchronous, giving rise to protracted entry of virions into the cell.

Second, we demonstrated that HPV16 interacts with A2t at the plasma membrane prior to cell entry. Whereas this observation is in agreement with data shown by Woodham and colleagues (60), our findings emphasize the role of AnxA2, not just S100A10, as an important interacting partner and regulator of HPV16 entry. We suggest that the AnxA2 core domain is involved in binding HPV16 at the cell surface. This is based on the finding that the virus-AnxA2 interaction requires Ca²⁺ ions, which interact with the AnxA2 core domain. Chelating Ca²⁺ in cell-based assays and in in vitro immunoprecipitation experiments disrupts HPV16 association with the cell surface, specifically with surface-resident AnxA2. This implies that Ca²⁺ is involved in bridging HPV16 and AnxA2 either directly or indirectly, through other AnxA2 ligands. Indirect AnxA2-HPV16 interactions could involve HSPGs, EGFR ligands or EGFR itself. An HSPG bridge between HPV and A2t is logical in the sense that heparin interacts in a Ca²⁺-dependent manner with the convex face of AnxA2 (74, 80), and we showed that AnxA2 interacts with HSPGs in HaCaT cells whether HPV was present or not. Additionally, HPVs are well known to associate with HSPGs on cells, and we showed that HPV virions are decorated with HSPGs and GFs when released from cells upon proteinase processing (17).

The third important finding in this study is that AnxA2 functions in the entry and trafficking of HPV16 for infection of HKs. Our data showing the interaction of virus with AnxA2 at the membrane as well as within endocytic vesicles support an active and direct role for AnxA2 in viral entry. Since the AnxA2 N-terminal antibody does not prevent Tyr23 phosphorylation but blocked HPV16 internalization, it is possible that the antibody may prevent AnxA2 association with lipid microdomains of forming endosomes by masking the N-terminally located atypical membrane binding motif (amino acids 15 to 24) (39, 82). Consequently, AnxA2 and its bound cargo, in this case PsVs, would become trapped at the plasma membrane as we observed. The antibody could hinder other AnxA2 functions at the plasma membrane as well. Perhaps the antibody prevents the function AnxA2 has in regulating the processing of EGFR ligands (59), which we showed are present in complex with HPV-HSPG and are important for infection (17). The antibody may function to block HPV entry by preventing AnxA2 from properly engaging in actin-mediated membrane remodeling in preparation for, or in the function of, macropinocytosis or endocytic vesicle rocketing (38, 82). Recent work suggests that HPV16 can enter cells via a macropinocytosis-related route (25); thus, inhibition of normal AnxA2-actin interactions could strand HPV16 on the cell surface. Grewal and Enrich make the case that active AnxA2 might connect actin filaments with membrane domains to promote lateral motion and/or endocytosis of EGFR (45). If EGFR is involved in HPV16 endocytosis, for which we recently provided evidence (17), AnxA2 inactivation could prevent this process. Certainly, understanding the mechanics of how the AnxA2 antibody prevents entry will reveal important insights into how HPV16 gains entrance into human keratinocytes.

We were surprised to find that the antibody to S100A10 inhibited infection by trapping viral particles in LAMP1-positive compartments alongside AnxA2. It therefore appears that HPV16 binds AnxA2 at the cell surface and remains associated with A2t during internalization in the presence of p11 antibody. This implies, although is certainly not proof, that S100A10 and AnxA2 can have distinct roles in HPV infection. S100A10 is essential for AnxA2 transport to the cell surface (31), yet the interaction of AnxA2 with p11/S100A10 is not required for AnxA2 to bind endosomal membranes or for endosomal trafficking (34, 46). It seems likely that the AnxA2-S100A10 tetramer is important for endosome formation and initial endocytosis since a dominant negative mutant AnxA2-S100A10 fusion protein allows endocytosis (79), whereas p11 function in endosomal trafficking with AnxA2 is less clear. If AnxA2 and p11 normally separate during endocytosis, the antibody to p11 may prevent dissociation to hold the A2t and HPV16 cargo hostage in dysfunctional trafficking that prevents proper viral uncoating and endosomal escape of the viral genome. Alternatively, the p11 antibody may simply inhibit L2-mediated virion uncoating and endosomal escape, leading to an accumulation of endosomal viral particles. It is plausible that the antibody-p11 interaction also inhibits the function of AnxA2 in membrane trafficking, leading the virus to a noninfectious end in a LAMP-1-positive vesicle. Regardless of the specific intracellular activities, both AnxA2 and S100A10 have important roles in the entry and trafficking of HPV16 in keratinocytes.

As AnxA2 regulates membrane dynamics at early stages of endocytosis and trafficking (reviewed in references 39 and 82), we propose that it is a key controller of HPV16 entry into human keratinocytes. AnxA2 not only binds actin and endosome-forming proteins but also can directly bind trafficking molecules such as EGFR (45), perhaps with its associated ligands, thereby facilitating their engulfment. We showed that HPV16 requires functional AnxA2 to enter human keratinocytes; whether the virus remains in complex with HSPGs and/or EGF/EGFR to stimulate AnxA2 association with endosome machinery to regulate internalization of AnxA2-bound ligands is not yet clear (Fig. 7F). AnxA2 is recruited to various endocytic pathways, including clathrin (36)- and caveola-mediated internalization (37), and is involved in macropinocytic rocketing (38), and this could explain why AnxA2-bound HPV16 might enter cells using multiple routes. As discussed above, the involvement of HPV16 with AnxA2 helps clarify the protracted nature of HPV16 uptake in laboratory infections. Understanding the mechanisms underlying diverse AnxA2 recruitment to distinct endocytic pathways is likely to shed light on the complexity of HPV entry and permit the reconciliation of seemingly disparate data on HPV entry.

The proposed role of A2t in HPV16 entry fits well into the context of epithelial wounding and access to the basal cells, which potentiates the efficiency of papillomavirus infection in vivo (5–7, 86). Both AnxA2 and S100A10/p11 are expressed in the stratified epithelium (56–58), but AnxA2 expression in intact and wounded skin is more concentrated to the surface of basal cells (57, 58). Currently, there is no direct evidence to suggest that a cellular receptor dictates cell- or tissue-specific tropism for papillomaviruses. However, the expression of A2t and upregulation of GFRs and HPSGs, like syndecan-1, in the basal epithelial layer are consistent with putative targeting of HPVs to the cells where the virus is most likely to find a permissive, mitogenic cell state, especially in a wounded environment. Together, these tissue-relevant observations, along with our identifying a mechanism whereby HPV recruits A2t to the extracellular membrane, as well as a role for A2t in HPV16 entry into human keratinocytes, strongly implicate AnxA2 and S100A10 as uptake receptors for HPV16.

In the larger view of intracellular pathogens, it is worth emphasizing that Chlamydia and Neisseria spp., as well as viruses from the families Retroviridae, Herpesviridae, Flaviviridae, and Picornaviridae interact with HSPGs at the cell surface (87, 88). Chlamydia trachomatis, cytomegalovirus, herpes simplex virus, hepatitis C virus, and others activate GFR pathways, in many cases bridging via soluble factors (89–93). Further, cytomegalovirus, enterovirus 71, and human immunodeficiency virus type 1 also employ annexin A2 interactions during infection (50–53, 55). These observations prompt speculation as to whether these pathogens, as well as some HPVs, might use similar, albeit complex, interactions at the host cell surface to gain entry. The data herein provide a deeper understanding of the complexity of HPV entry and reveal an additional way that pathogens usurp normal cell functions during host cell infections, as well as new targets for intervention.