Analysis of Foot-and-Mouth Disease Virus Internalization Events in Cultured Cells

It has been demonstrated that foot-and-mouth disease virus (FMDV)can utilize at least four members of the ${alpha}$ _V subgroup of the integrinfamily of receptors in vitro. The virus interacts with thesereceptors via a highly conserved arginine-glycine-aspartic acidamino acid sequence motif located within the ßG-ßHloop of VP1. While there have been extensive studies of virus-receptorinteractions at the cell surface, our understanding of the eventsduring viral entry into the infected cell is still not clear.We have utilized confocal microscopy to analyze the entry oftwo FMDV serotypes (types A and O) after interaction with integrinreceptors at the cell surface. In cell cultures expressing boththe ${alpha}$ _Vß₃ and ${alpha}$ _Vß₆ integrins, virus adsorbedto the cells at 4°C appears to colocalize almost exclusivelywith the ${alpha}$ _Vß₆ integrin. Upon shifting the infectedcells to 37°C, FMDV capsid proteins were detected within15 min after the temperature shift, in association with theintegrin in vesicular structures that were positive for a markerof clathrin-mediated endocytosis. In contrast, virus did notcolocalize with a marker for caveola-mediated endocytosis. Virusremained associated with the integrin until about 1 h afterthe temperature shift, when viral proteins appeared around theperinuclear region of the cell. By 15 min after the temperatureshift, viral proteins were seen colocalizing with a marker forearly endosomes, while no colocalization with late endosomalmarkers was observed. In the presence of monensin, which raisesthe pH of endocytic vesicles and has been shown to inhibit FMDVreplication, viral proteins were not released from the recyclingendosome structures. Viral proteins were not observed associatedwith the endoplasmic reticulum or the Golgi. These data indicatethat FMDV utilizes the clathrin-mediated endocytosis pathwayto infect the cells and that viral replication begins due toacidification of endocytic vesicles, causing the breakdown ofthe viral capsid structure and release of the genome by an as-yet-unidentifiedmechanism.

INTRODUCTION

Top

Introduction

Foot-and-mouth disease is a highly contagious viral diseaseof cloven-hoofed livestock caused by Foot-and-mouth diseasevirus (FMDV), the type species of the genus Aphthovirus of thefamily Picornaviridae. The virus is a positive-strand RNA viruscontaining a genome of approximately 8,400 nucleotides (reference33 and references therein).

FMDV initiates infection in cultured cells by binding to anyof four members of the ${alpha}$ _V subgroup of the integrin family ofcellular receptors ( ${alpha}$ _Vß₁, ${alpha}$ _Vß₃, ${alpha}$ _Vß₆,and ${alpha}$ _Vß₈) (16, 27, 36, 39, 40, 60) via a highly conservedarginine-glycine-aspartic acid amino acid sequence motif locatedwithin the ßG-ßH loop of VP1 (6, 30, 46,52). We have recently shown that the ${alpha}$ _Vß₆ integrinacts as a high-affinity receptor for the virus while ${alpha}$ _Vß₃interacts with virus with a much lower affinity (28). In addition,viruses of the type A serotype utilize both the ${alpha}$ _Vß₃and ${alpha}$ _Vß₆ integrins as receptors in cultured cells,while serotype O viruses have an affinity for the ${alpha}$ _Vß₆integrin (27).

While the initial events of FMDV-receptor interactions havebeen studied in detail, the subsequent events of virion entryand release of the viral genome are still not well defined.Interaction of enteroviruses with their receptors causes a conformationalrearrangement of the virion, resulting in the release of VP4and externalization of the N-terminal extension of VP1. Thisparticle has been called an altered, or A, particle (31, 32,42, 90). The A particle appears to degrade further to an 80Sparticle by interaction with membranes and the release of theRNA genome (13). In contrast, FMDV interactions with receptorsdo not result in structural changes to the virion (5, 28). Rather,this event occurs during viral internalization and results inthe breakdown of the virus to 12S pentameric subunits and therelease of the RNA (3-5, 21). Compounds which raise intracellularpH inhibit the breakdown, indicating that it occurs within acidicendocytic vesicles (3, 19, 20, 57). Interestingly, the breakdownof the virion within the vesicle, while necessary for productiveinfection, is not sufficient, suggesting that additional stepsare required (45). Direct analysis of FMDV entry and eventsoccurring subsequent to the virion entering the cell, however,has not been reported. Most nonenveloped viruses enter cellsvia endocytic mechanisms that are either clathrin, caveola,or lipid raft mediated (63, 78). Analysis of the internalizationof human parechovirus 1 (HPEV-1) and human adenoviruses, whichutilize ${alpha}$ _V integrins as internalization receptors, have shownthat both viruses use the clathrin-mediated endocytosis pathwayfor entry into the cell (41, 61, 95), while the enterovirusechovirus 1, which utilizes the integrin ${alpha}$ ₂ß₁ as areceptor, enters via a caveola-mediated, lipid raft-dependentmechanism (49, 64).

In this study, we utilized confocal microscopy to follow viralentry after adsorption and to determine the cell structureswhich play roles in these events.

MATERIALS AND METHODS

Top

Materials and Methods

Cell lines, viruses, and plasmids. Human mammary gland epithelial cells (MCF-10A) were obtainedfrom the American Type Culture Collection (catalogue no. CRL-10317)and were maintained in a mixture of Dulbecco's minimum essentialmedium and F12-Ham's media (1:1; Life Technologies) containing5% heat inactivated fetal bovine serum (FBS), 20 ng/ml epidermalgrowth factor, 100 ng/ml cholera toxin, 10 µg/ml insulin,and 500 ng/ml hydrocortisone. COS-1 cells were maintained inDulbecco's minimum essential medium containing 10% FBS, an additional2 mM L-glutamine, and 1 mM sodium pyruvate.

FMDV type A₁₂ strain 119ab (A₁₂) was derived from the infectiouscDNA clone pRMC₃₅ (71), and type O₁ strain Campos (O₁C) wasderived from the vesicular fluid of an infected steer. Viralstocks were grown in BHK-21 cells and purified as previouslydescribed (51). Titers were determined by plaque assay on BHK-21cell monolayers using standard techniques (71).

cDNA plasmids encoding the full-length bovine ${alpha}$ _V, ß₃,and ß₆ subunits have been described (27, 59). Transientexpression of bovine integrin subunits in COS-1 cells was performedas described previously (59). Briefly, cells were transfectedwith 2 µg each of cDNA plasmids encoding the bovine ${alpha}$ _Vsubunit and the appropriate ß subunit, using the transfectionreagent FuGene6 (Roche Molecular Biochemicals). After overnightincubation, the transfected cultures were infected with FMDVas described below.

Viral growth curve. Monolayers of MCF-10A or BHK-21 cells were infected with typeA₁₂ or O₁C virus at a multiplicity of infection (MOI) of 10PFU/cell for 1 h at 37°C. At the end of the adsorption period,the inoculum was removed and the cells were rinsed with ice-coldMES (2-morpholinoethanesulfonic acid)-buffered saline (25 mMMES, pH 5.5, 145 mM NaCl) to remove residual virus particles.The monolayers were then rinsed with minimum essential mediumcontaining 1% FBS and 25 mM HEPES, pH 7.4, and incubated at37°C. At appropriate times postinfection, the cells werefrozen at –70°C, and the thawed lysates were usedto determine titers by plaque assay on BHK-21 cell monolayers.

Antibodies and reagents. Monoclonal antibodies (MAbs) 6HC4, directed against FMDV typeA₁₂, and 12FB, directed against type O₁, have been previouslydescribed (8, 80). A rabbit polyclonal antiserum to the C-terminalregion of the FMDV 3A protein has been previously described(62). MAb 2D11, directed against the FMDV nonstructural protein3D^pol, was obtained from E. Brocchi, Instituto ZooprofilatticoSperimentale della Lombardia e dell Emilia-Romagna, Brescia,Italy. MAbs LM609 (catalogue no. MAB1976), which recognizesthe ${alpha}$ _Vß₃ heterodimer (23), and CSß6 (catalogueno. MAB2076), which recognizes the ß₆ integrin subunit(97) and the ${alpha}$ _Vß₆ heterodimer (28), were purchasedfrom Chemicon. To study virus entry, antibodies against thefollowing human cellular structures were used: a rabbit polyclonalantibody against early endosomal antigen 1 (EEA-1) (AffinityBioreagents) was used to identify early endosomes, a rabbitpolyclonal antibody against the cation-independent mannose-6-phosphatereceptor (CI-MPR) (Affinity Bioreagents) was used to identifylate endosomes, and a rabbit polyclonal antibody against caveolin-1and a MAb against clathrin (clone X22) were used as markersfor caveolae and clathrin endocytosis pathways, respectively(Affinity Bioreagents). In addition, MAbs directed against theendoplasmic reticulum (ER) marker, protein disulfide isomerase(PDI; clone RL77; Affinity Bioreagents), Golgi zone area (clone371-4; Sigma), ß-COP (clone maD; Sigma), and transferrinreceptor (TfnR; clone RVS-10; Chemicon) were used.

The following chemicals and inhibitors were used: monensin ionophore(Sigma), a lysosomotropic agent which raises the pH in endocyticvesicles (3), was prepared as a 10 mM stock solution in 95%ethanol and used at a 50 µM concentration diluted in culturemedium; nystatin (Gibco), which sequesters cholesterol and disruptslipid rafts (1, 77, 99), was used at a 25 µM concentrationdiluted in culture medium; chlorpromazine (Sigma), which causesthe loss of coated pits from the surface of the cell and theappearance of clathrin coats composed of the same subunits onendosomal membranes (41, 96), was used at a 12.5 µM concentrationdiluted in culture medium.

Infection of cells for confocal microscopy. Subconfluent monolayers of MCF-10A cells or transfected COS-1cells expressing either bovine ${alpha}$ _Vß₃ or ${alpha}$ _Vß₆,grown on 12-mm glass coverslips in 24-well tissue culture dishes,were infected with FMDV (MOI, 100 PFU/cell) for 1 h at 4°Cin minimum essential medium containing 0.5% FBS and 25 mM HEPES,pH 7.4. After the adsorption period, the inoculum was removed,the monolayers were washed with medium, fresh medium was added,and the cells were incubated at 37°C. At the appropriatetimes after the temperature shift, cells were fixed with 4%paraformaldehyde and processed for immunofluorescence and confocalmicroscopy as described below. In all of the figures, the timesare listed as either minutes postadsorption (p.a.) or hoursp.a. These times refer to the amount of time elapsed after thetemperature was shifted from 4° to 37°C.

To investigate the effects of the inhibitory compounds (monensin,nystatin, and chlorpromazine), cells were incubated with thecompounds for 30 min at 37°C prior to infection, and thecompounds were present during the entire experimental period.Virus infection of the cells was monitored by counting the numberof immunofluorescence (IF)-positive cells stained with MAbsreactive with viral structural proteins.

IF and confocal microscopy. After fixation, the paraformaldehyde was removed, and the cellswere permeabilized with 0.5% Triton X-100 for 5 min at roomtemperature (RT) and incubated in blocking buffer (phosphate-bufferedsaline [PBS], 5% normal goat serum, 2% bovine serum albumin,10 mM glycine, 0.01% thimerosa) for 1 h at RT. The fixed cellswere then incubated with the primary antibodies overnight at4°C. When double labeling was performed, cells were incubatedwith both antibodies together. The dilutions of the primaryantibodies were as follows: anti- ${alpha}$ vß3 (1/100), anti-ß6(1/100), anti-FMDV (1/5), anti-caveolin-1 (1/200), anti-clathrin(1/100), anti-ß-COP (1/200), anti-TfnR (1/50), anti-EEA-1(1/200), anti-CI-MPR (1/100), anti-PDI (1/200), and anti-Golgi(1/100). After being washed three times with PBS, the cellswere incubated with the appropriate secondary antibody, goatanti-rabbit immunoglobulin G (IgG) (1/400; Alexa Fluor 594;Molecular Probes) or goat anti-mouse isotype-specific IgG (1/400;Alexa Fluor 488 or Alexa Fluor 594; Molecular Probes), for 1h at RT. Following this incubation, the coverslips were washedthree times with PBS, counterstained with the nuclear stainTOPRO-iodide 642/661 (Molecular Probes) for 5 min at RT, washedas before, mounted, and examined in a Leica scanning confocalmicroscope. Data were collected utilizing appropriate preparedcontrols lacking the primary antibodies, as well as using anti-FMDVantibodies in uninfected cells to give the negative backgroundlevels and to determine channel crossover settings. The capturedimages were adjusted for contrast and brightness using AdobePhotoshop software.

RESULTS

Top

Results

FMDV replication in MCF-10A cells. Since foot-and-mouth disease affects livestock species suchas cattle and pigs and can replicate to high titers in the hamstercell line BHK-21, it would have been advantageous to utilizeeither bovine cells, porcine cells, or BHK-21 cells for theseexperiments. However, since almost all of the available antibodiesagainst intracellular organelles react only with human cells,we decided to perform most of these studies with the human breasttissue epithelial cell line MCF-10A. Since it has been shownthat FMDV can utilize the human ${alpha}$ _Vß₃ and ${alpha}$ _Vß₆integrins as viral receptors in vitro (40, 60), we analyzedthe expression of these integrins in MCF-10A cells by IF-confocalmicroscopy. The ${alpha}$ _Vß₃ integrin was distributed on thecell surface as small discrete structures (Fig. 1a). Examinationof the distribution of ${alpha}$ _Vß₆, which was done by usinga MAb that recognizes only the ß₆ subunit, showeda cytoplasmic localization for the subunit (Fig. 1b). Both integrinswere present in only a subset of cells. The distribution ofthe integrins in MCF-10A cells was similar to the distributionobserved when primary cultures of fetal bovine kidney cellswere examined (not shown). In addition, MCF-10A cells also expressthe ß₁ integrin subunit and the ${alpha}$ _Vß₅ and ${alpha}$ ₅ß₁ integrins (not shown). The last two integrinsare not FMDV receptors (27, 40, 60). We did not analyze thesecells for the expression of the other two FMDV integrin receptors, ${alpha}$ _Vß₁ (39) and ${alpha}$ _Vß₈ (36).

View larger version (27K):
[in this window]
[in a new window]
FIG. 1. Distribution of ${alpha}$ _Vß₃ and ß₆ integrins in MCF-10A cells. Monolayers of uninfected MCF-10A cells were processed for IF staining as described in Materials and Methods. MAbs LM609, which recognizes the ${alpha}$ _Vß₃ heterodimer (a), and CSß6, which recognizes the ß₆ integrin subunit (b), were used as primary antibodies. Alexa Fluor 488-conjugated antibodies were used as secondary antibodies. The bars represent 16 µm in panel a and 20 µm in panel b.

The replication cycle of FMDV in MCF-10A cells was analyzedby growth curve and immunofluorescence. The growth kineticsof types A₁₂ (Fig. 2a) and O₁C (Fig. 2b) in MCF-10A cells weresimilar to their kinetics in BHK-21 cells; however, the humancell line produced 2 to 3 log units lower viral titers thandid BHK-21 cells (Fig. 2a and b). Analysis of the localizationof the viral capsid protein VP1 during the replication cycleshowed that the protein was distributed near the perinucleararea by 1 h p.a., and newly synthesized viral protein was distributedthroughout the cytoplasm by 4 h p.a. (Fig. 2 bottom). In addition,the nonstructural proteins 3A and 3D^pol were also detected inthe cytoplasm by 4 h p.a. (data not shown). These results indicatethat the MCF-10A cells are suitable for the study of the earlyevents in the viral replication cycle in vitro.

View larger version (34K):
[in this window]
[in a new window]
FIG. 2. FMDV replication in MCF-10A cells. Monolayers of MCF-10A and BHK-21 cells were infected with type A₁₂ (a) or O₁C (b) at an MOI of 10 PFU/cell for 1 h at 37°C. After the adsorption period, the cells were washed with MES-buffered saline (see Materials and Methods) and incubated at 37°C. At the times indicated, the plates were removed to –70°C. Samples were thawed, and titers were determined by plaque assay on BHK-21 cells. Parallel cultures of MCF-10A cells were processed for IF-confocal microscopy as described in Materials and Methods at the times indicated after the adsorption period (bottom). Viral protein synthesis was visualized with a specific MAb against type A₁₂ or type O₁C VP1. Anti-mouse isotype-specific IgG Alexa Fluor 594-labeled conjugate was used as a secondary antibody. The bars represent 8 µm.

FMDV colocalizes with integrins at the cell surface during adsorption. Both types A₁₂ and O₁C were adsorbed to cells at 4°C, andlocalization of virus on the membrane in relation to the integrins ${alpha}$ _Vß₃ and ${alpha}$ _Vß₆ was determined by confocal microscopy.To locate the ${alpha}$ _Vß₆ integrin on the cell surface, weutilized a MAb (CSß6) which recognizes the ß₆subunit (97). However, we have recently shown that this antibodyalso reacts with the intact heterodimer (28). In nonpermeabilizedcells, we can detect the ß6 subunit on the cell surface(not shown), and since the ligand binding domains of activesurface integrins consist of regions of both the ${alpha}$ and ßsubunits (100), in this report we will consider that stainingof this subunit on the cell membrane will identify the intact ${alpha}$ _Vß₆ heterodimer.

Upon adsorption of either type A₁₂ or O₁C to MCF-10A cells,virus colocalized with the ${alpha}$ _Vß₆ integrin on the cellmembrane (Fig. 3b). However, neither virus appeared to interactwith the ${alpha}$ _Vß₃ integrin on these cells (Fig. 3a). Wefound a few cells that faintly reacted with anti-FMDV antibodiesand displayed positive staining for ${alpha}$ _Vß₃ at the cellsurface, but we were not able to detect any colocalization.Most of the cells that displayed strong staining for the viralcapsid protein did not express ${alpha}$ _Vß₃ at the cell surface(Fig. 3a). Since we have shown that type A viruses can utilizeboth ${alpha}$ _Vß₃ and ${alpha}$ _Vß₆ integrins as receptorswhile type O viruses have a preference for the ${alpha}$ _Vß₆integrin (27), we transiently expressed the bovine ${alpha}$ _Vß₃or ${alpha}$ _Vß₆ integrin in COS-1 cells and analyzed viruslocalization by confocal microscopy. Our results showed thattype A₁₂ colocalized with both integrins while type O₁C colocalizedonly with ${alpha}$ _Vß₆ (not shown), confirming our previouslyreported results (27). In addition, these results suggest thatwhen both integrins are expressed on cultured cells, both typesA₁₂ and O₁C appear to have a higher affinity for the ${alpha}$ _Vß₆integrin.

View larger version (25K):
[in this window]
[in a new window]
FIG. 3. Distribution of FMDV virions and integrin receptors during the viral adsorption period. Monolayers of MCF-10A cells were infected with FMDV type A₁₂ or O₁C at an MOI of 100 PFU/cell for 1 h at 4°C and processed for double IF staining as described in Materials and Methods. FMDV virions were localized with specific MAbs against the capsid protein VP1 and visualized with Alexa Fluor 594 (red). Integrins were stained with anti- ${alpha}$ _Vß₃ (a) or anti-ß₆ (b) MAbs and visualized with Alexa Fluor 488 (green). The bars represent 20 µm in panel a and 8 µm in panel b.

FMDV internalizes in association with the ${alpha}$ _Vß₆ integrin. To begin to analyze the internalization process, we followedboth the virion and the integrin receptor after binding thevirus to MCF-10A cells at 4°C and shifting the temperatureto 37°C. Figure 4 shows that viral capsid protein can beseen entering the cell cytoplasm from the membrane as earlyas 15 min p.a. By 1 h p.a., the virion had translocated to theperinuclear region of the cell (Fig. 4). These micrographs alsoreveal that ${alpha}$ _Vß₆ is also redistributed and appearsto internalize along with the virion. When uninfected cellswere examined, we did not detect integrin internalization at37°C (not shown), suggesting that virus binding to ${alpha}$ _Vß₆resulted in its internalization. This association can be observedfor as long as 1 h p.a. Interestingly, the amount of integrinon the cell surface is decreased drastically by the internalizationprocess (Fig. 4) and does not appear to return to preinfectionlevels throughout the infectious cycle (not shown). Similarresults were obtained when these experiments were performedwith either fetal bovine kidney cells or COS-1 cells expressingthe bovine ${alpha}$ _Vß₆ integrin (not shown).

View larger version (38K):
[in this window]
[in a new window]
FIG. 4. Internalization of FMDV and ${alpha}$ _Vß₆. FMDV type O₁C was adsorbed to MCF-10A monolayers at an MOI of 100 PFU/cell for 1 h at 4°C. The cells were washed with medium and incubated at 37°C. At the times indicated after the temperature shift, the cells were processed for IF-confocal microscopy as described in Materials and Methods. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red), and the ${alpha}$ _Vß₆ integrin was stained with an anti-ß₆ MAb and visualized with Alexa Fluor 488 (green). The bars represent 8 µm.

FMDV internalizes via a clathrin-dependent mechanism. In order to determine the endocytic mechanism utilized by FMDV,we followed virion entry into MCF-10A cells, along with thedistribution of either clathrin or caveolin-1. Figure 5a showscolocalization of both type A₁₂ and type O₁C with clathrin within5 min p.a., which has almost completely disappeared by 30 minp.a. In contrast, virus was not observed to colocalize withcaveolin-1, a marker for caveolae, even though the surfacesof these cells have a high concentration of the protein (Fig.5b).

View larger version (50K):
[in this window]
[in a new window]
FIG. 5. Analysis of the FMDV internalization pathway. FMDV types A₁₂ and O₁C were adsorbed to monolayers of MCF-10A cells (MOI, 100 PFU/cell) for 1 h at 4°C. The cells were washed, overlaid with warm medium, and transferred to 37°C. At the times indicated after the temperature shift, cells were processed for confocal microscopy as described in Materials and Methods. Virus was stained with anti-VP1 MAbs and visualized with Alexa Fluor 594 (red). Clathrin (a) was stained with an anti-clathrin MAb, and caveolin-1 (b) was stained with a rabbit polyclonal anti-caveolin-1 antibody. Both proteins were visualized with Alexa Fluor 488 (green). Only the merged photographs are shown. In panel c, cells were pretreated with chlorpromazine (Cpz; 12.5 µM) for 30 min at 37°C prior to infection with type O₁C (MOI, 100 PFU/cell). The cells were incubated at 37°C in the presence of the drug until 4 h postinfection, when they were processed for confocal microscopy. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red). The bars represent 8 µm in panels a and b and 40 µm in panel c.

To confirm these results biochemically, we infected cells inthe presence of chlorpromazine, which inhibits clathrin-mediatedendocytosis (41, 96), and observed the number of virus-positivecells at 4 h p.a. by IF-confocal microscopy. We found that chlorpromazinemarkedly reduced the number of cells expressing viral antigen(Fig. 5c). By counting infected and noninfected cells in multiplefields, we determined that there were about 50% infected cellsper field in the control cultures and only about 4% infectedcells per field in the chlorpromazine-treated cultures. In addition,in the presence of the drug, we were able to detect viral capsidprotein at the surfaces of a small number of cells for as longas 30 min p.a., but no colocalization with clathrin (not shown).By 1 h p.a., we were unable to detect virus either on the cellsurface or in the cytoplasm (not shown). We have previouslyshown that FMDV, after binding to the surface of the cell at4°C, begins to elute from the cell surface as early as 5to 10 min after the temperature is shifted to 37°C (5);thus, we did not expect to see virus on the cell surface atthese later time points. These results, therefore, suggest thatin the presence of chlorpromazine the virus can bind to thecellular receptor but is unable to internalize.

Although we had already seen that FMDV did not colocalize withthe caveolin-1 marker, we investigated if the cholesterol-sequesteringand lipid raft-disrupting compound nystatin (1) could inhibitviral infection. By using IF-confocal microscopy, we observedno decrease in the number of virus-infected cells when infectionwas done in the presence of the drug compared to the control(not shown). Overall, these data indicate that FMDV entry intoMCF-10A cells is clathrin dependent.

FMDV traffics through early endosomes after entry. To evaluate the movement of virions through early endosomes,we utilized an antibody against the early endosomal proteinEEA-1. Because this antibody did not detect the protein in MCF-10Acells, we used COS-1 cells expressing the bovine integrin ${alpha}$ _Vß₆.Examination of virus-infected cells showed colocalization ofthe viral capsid protein with the EEA-1 protein at 15 min p.a.which disappeared by 30 min p.a., indicating that the virushad already been translocated from the early endosomes (Fig.6). Interestingly, we were not able to detect virions in conjunctionwith CI-MPR, a marker of late endosomes (Fig. 7).

View larger version (42K):
[in this window]
[in a new window]
FIG. 6. Movement of virus into early endosomes. Monolayers of COS-1 cells were cotransfected with cDNA plasmids encoding the bovine ${alpha}$ _V and ß₆ subunits as described in Materials and Methods. At 24 h posttransfection, cells were infected with type O₁C (MOI, 100 PFU/cell) for 1 h at 4°C. After being washed, the cells were overlaid with warm medium and moved to 37°C. At the times indicated after the temperature shift, cells were processed for confocal microscopy. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red). Early endosomes were stained with a rabbit polyclonal anti-EEA-1 antiserum and visualized with Alexa Fluor 488 (green). The bars represent 8 µm.

View larger version (28K):
[in this window]
[in a new window]
FIG. 7. Interaction of FMDV with late endosomes. Monolayers of MCF-10A cells were infected with type A₁₂ (MOI, 100 PFU/cell) for 1 h at 4°C. After being washed, the cells were overlaid with warm medium and moved to 37°C. At the times indicated after the temperature shift, cells were processed for confocal microscopy. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red). Late endosomes were stained with a rabbit polyclonal anti-CI-MPR antiserum and visualized with Alexa Fluor 488 (green). The bars represent 8 µm.

Following internalization, some recycling receptors, such asTfnR, are known to pass through the early endosomes and thenaccumulate in recycling endosomes before they return to theplasma membrane (69, 92). To further study FMDV entry, we examinedthe localization of FMDV capsid protein with TfnR. Double labelingwith anti-FMDV capsid protein and anti-TfnR antibodies revealedcolocalization for up to 1 h p.a. (not shown), resembling thedistribution observed for FMDV with ${alpha}$ _Vß₆ (Fig. 4).

It had been shown that the use of lysosomotropic agents, suchas monensin, which raise the pH of endosomal vesicles, inhibitedthe replication of FMDV (3, 19, 20, 57). To further understandthis phenomenon, we examined FMDV-infected MCF-10A cells inthe presence of monensin at 4 h p.a. Figure 8 shows that inthe presence of the drug, both type A₁₂ (Fig. 8b) and type O₁C(Fig. 8a) virions were found in small structures resemblingendosomes, quite similar to the viral staining seen at veryearly times after the temperature shift (Fig. 4 and 5). In contrast,viral proteins were distributed throughout the cytoplasm inuntreated infected cells (Fig. 8). In addition, in infectedtreated cells, the virus was still colocalizing with TfnR andthe ${alpha}$ _Vß₆ integrin as late as 4 h p.a. (Fig. 8) andwas no longer colocalizing with the clathrin marker, which wasnot affected by monensin treatment (not shown). Thus, raisingthe pH of the early endosomes probably prevents the viral genomefrom being released to the cytoplasm to begin the replicationcycle. This result confirms earlier results, which showed thatmonensin prevented the breakdown of the virus to pentamericsubunits and RNA in infected cells (3), and is in agreementwith the above-mentioned results on viral trafficking throughclathrin-coated vesicles.

View larger version (48K):
[in this window]
[in a new window]
FIG. 8. Effect of monensin on virus internalization. Monolayers of MCF-10A cells were incubated with monensin (50 µM) for 30 min at 37°C prior to infection with type O₁C (a) or type A₁₂ (b) (MOI, 100 PFU/cell) for 1 h at 4°C. After being washed, the cells were overlaid with warm medium in the presence of monensin and moved to 37°C. At 4 h p.a., cells were processed for confocal microscopy. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red). The integrin ${alpha}$ _Vß₆ was stained with an anti-ß₆ MAb (a), and the TfnR was stained with an anti-TfnR MAb (b). Both proteins were visualized with Alexa Fluor 488 (green). The bars represent 8 µm.

FMDV does not colocalize with the ER or the Golgi. We also examined the roles of both the ER and the Golgi apparatusduring virus internalization. We used anti-PDI antibody as amarker for the ER and compared the distribution with that ofthe FMDV capsid protein in MCF-10A cells at early times p.a.The infected cells showed no colocalization of the viral capsidprotein with the ER (Fig. 9). Similarly, viral proteins werenot found colocalized with the Golgi apparatus during earlytimes p.a. (Fig. 9), suggesting that virus is not being deliveredto the ER or Golgi during internalization.

View larger version (34K):
[in this window]
[in a new window]
FIG. 9. Interaction of FMDV with ER and Golgi apparatus. Monolayers of MCF-10A cells were infected with type O₁C (MOI, 100 PFU/cell) for 1 h at 4°C. After being washed, the cells were overlaid with warm medium and moved to 37°C. At the times indicated after the temperature shift, cells were processed for confocal microscopy. Virus was stained with an anti-VP1 MAb and visualized with Alexa Fluor 594 (red). The ER was stained with an anti-PDI MAb, and the Golgi was stained with an anti-Golgi zone area MAb. Both proteins were visualized with Alexa Fluor 488 (green), and only the merged photographs are shown. The bars represent 8 µm.

We have previously demonstrated that BHK-21 cells infected withFMDV or transfected with a cDNA plasmid encoding the nonstructuralprotein 3A disrupted the Golgi apparatus (62), resembling thedisruption reported in studies examining the mechanism of actionof brefeldin A (47, 48). We have shown that FMDV replicatesin the presence of brefeldin A (62), which affects the buddingof coatomer protein I (COPI)-coated membranes (44). To furtherstudy if FMDV uses the retrograde transport system, we examinedby IF-confocal microscopy whether a component of COPI, ß-COP,colocalizes with FMDV capsid proteins during the internalizationprocess. No colocalization was observed in MCF-10A cells infectedwith FMDV (not shown), supporting the data shown above.

DISCUSSION

Top

Discussion

In general, picornaviruses utilize endocytic pathways, whichare either clathrin mediated, caveola mediated, or lipid raftdependent, to enter cells after binding to their receptors (11,24, 25, 35, 49, 63, 76, 78, 81, 84). The current data suggestthat the nature of the viral receptor determines the pathwayof entry. At least four different picornaviruses utilize integrinreceptors to infect cells. Echovirus 1, which utilizes the integrin ${alpha}$ ₂ß₁ as a receptor (14, 15, 89), has been shown toutilize a caveola-mediated entry pathway (49, 64), while HPEV-1,which utilizes the ${alpha}$ _Vß₃ or ${alpha}$ _Vß₁ integrin asa receptor (67, 83, 86), uses the clathrin-mediated endocytosispathway to enter the cell (41). Coxsackievirus A9 can utilizethe integrin ${alpha}$ _Vß₃ as a receptor (74, 87, 88) and glucose-regulatedprotein 78 as a coreceptor (82). The latter molecule appearsto associate with a major histocompatibility complex class Imolecule that is responsible for virus internalization via alipid raft-dependent mechanism (85). The results presented inthis study indicate that FMDV, which utilizes the ${alpha}$ _V integrinsas receptors (references 9, 10, and 38 and references therein),enters the cell via a clathrin-mediated endocytic pathway.

Most of these studies were performed with the human breast epithelialcell line MCF-10A, since almost all the reagents available todetect cellular structures react only with human cells. StainingMCF-10A cells for the expression of the ${alpha}$ _Vß₃ and ${alpha}$ _Vß₆integrins showed a heterogeneous cell population for integrinexpression (Fig. 1). While the two integrins appear to be expressedin different subpopulations of cells, we were not able to determineif there was coexpression in individual cells. MCF-10A cellswere shown to be susceptible to infection with both serotypes,A₁₂ and O₁C, with growth kinetics similar to those of BHK-21cells (Fig. 2). In addition, IF studies showed strong stainingfor newly synthesized capsid proteins after 4 h p.a. (Fig. 2),as well as for nonstructural proteins 3A and 3D^pol (not shown),indicating that this cell line is suitable for study of theearly events during the viral replication cycle in vitro.

We have previously shown that ${alpha}$ _Vß₆ acts as a high-affinityreceptor for the virus, while ${alpha}$ _Vß₃ interacts with thevirus with a much lower affinity (28). We have also shown thattype A viruses can utilize both the ${alpha}$ _Vß₃ and the ${alpha}$ _Vß₆integrins as receptors, while the type O viruses appear to preferentiallyutilize the ${alpha}$ _Vß₆ integrin (27). The experiments withMCF-10A cells support these data, showing that in cultures ofcells which express either ${alpha}$ _Vß₃ or ${alpha}$ _Vß₆, bothtypes A₁₂ and O₁C preferentially bind to the ${alpha}$ _Vß₆ integrin(Fig. 3). However, when COS-1 cells were transfected with cDNAsencoding either the bovine ${alpha}$ _Vß₃ or ${alpha}$ _Vß₆ integrin,we were able to detect colocalization of type A₁₂ with bothintegrins, while type O₁C colocalized only with ${alpha}$ _Vß₆(not shown).

Following the binding of FMDV to the cell surface, we analyzedthe entry route of the virus into the cell. FMDV appears tobe internalized, in association with the ${alpha}$ _Vß₆ integrin,into a structure resembling an endocytic vesicle and is translocatedfrom the plasma membrane to the perinuclear region within 1h p.a. (Fig. 4). Interestingly, in contrast to the results presentedhere, the ${alpha}$ _Vß₃ integrin does not appear to internalizeor colocalize with HPEV-1 during the early phases of internalization(41). Under normal conditions, ${alpha}$ _Vß₃, under the controlof cellular growth factors, intracellular kinases, and GTPases,is internalized, traffics through endosomes, and then recyclesto the plasma membrane for reutilization (72, 73, 98). Similarstudies have not been done with the ${alpha}$ _Vß₆ integrin,but it appears that the integrin receptor does not recycle backto the surface, since we cannot detect surface integrin at 2h p.a. (Fig. 4) and even at later times in the infectious cycle(not shown). It is not clear whether the surface integrin utilizedto internalize the virus is degraded; however, since the virusinhibits cap-dependent cellular protein synthesis (26, 43),it is unlikely that any newly synthesized integrin would begenerated.

In order to determine the endocytic pathway used by FMDV toenter the cell, the distribution of virus with markers for theclathrin or caveola pathway was examined. After adsorption at4°C, virus appears to colocalize with clathrin as earlyas 5 min after the temperature is shifted to 37°C (Fig.5). By 30 min after the temperature shift, little or no colocalizationof virus and clathrin can be observed (Fig. 5), which probablyis the result of the uncoating of clathrin from the clathrin-coatedpit after separation from the plasma membrane (55). In addition,virions did not colocalize with caveolin-1, a marker for thecaveola-mediated endocytosis pathway, during the internalizationprocess. To verify these results biochemically, we utilizedchlorpromazine, a member of a class of compounds that inhibitsthe formation of clathrin-coated pits and causes pits to disappearfrom the cell surface (41, 96). In the presence of this drug,viral infection was markedly inhibited. Finally, viral infectionwas not inhibited in the presence of nystatin, a cholesterol-sequesteringand lipid-raft disrupting compound (1, 77, 99). Clathrin-mediatedendocytosis is generally not dependent on lipid rafts (63).

It has also been shown that the virus can utilize at least twosurrogate receptors to infect cells, Fc receptors (7, 51, 52)and a chimeric receptor consisting of a single-chain anti-FMDVMAb fused to ICAM-1 (70). While Fc receptors have been shownto internalize via a clathrin-mediated route (50, 56), ICAM-1,which is a receptor for the major group of human rhinoviruses(HRV) (12), does not contain endocytosis signals, and neitherthe transmembrane nor the cytoplasmic domain is essential forHRV infection (29, 79). It has been shown, however, that major-groupHRV is internalized into intracellular endosomes (76) via adynamin-dependent mechanism (25), but it is not clear whetherthe clathrin or caveola pathway is used. Cell culture-adaptedFMDV has also been shown to utilize cell surface heparan sulfate(HS) as a receptor in cultured cells (37, 60), which resultsin a loss of virulence for cattle (75). Although the pathwayof HS internalization of ligands is not clear, some recent resultssuggest that HS-mediated internalization of growth factors occursvia a caveola-dependent pathway (68). Thus, the pathway utilizedby FMDV for internalization appears to be a function of thereceptor and not the virus.

We have not examined the role of dynamins during FMDV internalization.Dynamins are a family of GTPases that facilitate the buddingof clathrin-coated pits, leading to the formation of coatedvesicles (91), and also mediate the caveola-dependent pathwayof internalization (34). Although the role of dynamin in theinternalization of HPEV-1 or CAV9, both of which utilize ${alpha}$ _V integrinsas receptors, has not been determined, the internalization ofhuman adenovirus, which is mediated by the ${alpha}$ _V integrins, is dynamindependent (53, 54, 95).

We detected virus in early endosomes within 15 min after shiftingthe temperature to 37°C (Fig. 6). By 30 min, however, therewas no staining of viral proteins in early endosomes, indicatingthat the virus was already translocated from the vesicle. Wewere not able to detect virus within late endosomes at timesafter 30 min, suggesting that virus was not being translocatedto this cellular structure. To further examine this, we studiedthe distribution of FMDV with the TfnR, which passes sequentiallythrough clathrin-coated early endosomes and recycling endosomesbefore returning to the plasma membrane in a dynamin-dependentmanner (69, 92). We found that FMDV and TfnR colocalized inthe presence of monensin, which inhibits receptor recycling(2), indicating that FMDV is moving from the early endosomesto recycling endosomes rather than to the late endosomes, aswas described for HPEV-1 (41).

We were not able to detect any colocalization of FMDV capsidproteins with either ER or Golgi markers (Fig. 9). Again, theseresults are in contrast to those observed with HPEV-1 (41).However, these data confirm our previous results, demonstratingthat FMDV replicates in the presence of brefeldin A, a compoundthat disrupts the Golgi apparatus and affects the budding ofCOPI-coated membranes (62). We also found that virus did notcolocalize with the COPI complex or with tubulin, a microtubulemarker, during the internalization process (not shown). In addition,nocodazole, which disrupts microtubules and inhibits endosomaltransport from early to late endosomes, did not inhibit FMDVreplication (S. J. Berryman, S. Clark, A. Burman, P. Monaghan,and T. Jackson, Abstr. Seventh Int. Symp. Pos. Strand RNA Viruses,abstr. P1-B4, 2004), indicating that virus is not transportedto the ER and Golgi apparatus. Thus, the data presented hereindicate that neither the ER nor the Golgi is used by the virusduring the internalization event.

FMDV is structurally unaffected by interaction with its receptor(5, 28). The initial alteration of the 140S virion to 12S pentamericsubunits probably occurs within acidified endocytic vesiclesonce the virus is internalized, resulting in the release ofthe RNA genome (4, 5, 21). Further evidence of this is the findingthat agents that raise the pH of endocytic vesicles inhibitFMDV replication and prevent this initial alteration (3, 19,20, 57). In the present work, we found that in the presenceof monensin FMDV is adsorbed to the cell surface and internalizednormally but remains in structures that resemble endocytic vesiclesfor as long as 4 h after infection (Fig. 8). These data furtherconfirm previous experiments and indicate that preventing theacid-induced virion breakdown inhibits the release of the RNAgenome to the cytosol. Thus, FMDV resembles the minor-groupHRV, which are also dependent on low-pH-induced virion degradationfor release of the viral genome through virus-induced poresin the endosomal membrane (65, 66).

The role, if any, of integrin cytoplasmic domain internalizationsignals in FMDV internalization has not been determined. Integrinß subunit cytoplasmic domains contain a tetramericNPXY sequence, which has been shown to be an internalizationsignal for the human low-density lipoprotein receptor (22).It has been suggested, however, that while this sequence isnot required for integrin internalization (93), it is requiredfor integrin signaling functions and adhesion (17, 18). We havepreviously reported that deletion of almost all of the cytoplasmicdomains of the ${alpha}$ _V and ß₃ subunits does not preventthe integrin from mediating FMDV infection in cultured cells(58). In addition, we also observed that mutation of the tyrosinewithin the NPXY sequence of the ß₃ subunit, whichabolishes phosphorylation of the subunit and affects integrinavidity (17), did not affect the ability of the ${alpha}$ _Vß₃integrin to function as a receptor for type A virus (not shown).In contrast, Miller and coworkers demonstrated that removalof at least 80% of the C-terminal region of the ß₆cytoplasmic domain, or deletion of the central region containingthe NPXY motif, resulted in loss of the ability of the ${alpha}$ _Vß₆integrin to mediate viral infection subsequent to adsorption(57). In addition, exchange of the ß₆ cytoplasmicdomain with the ß₈ cytoplasmic domain abolished theability of the ${alpha}$ _Vß₆ integrin to both bind virus andmediate infection (36). Taken together, these results indicatethat the ${alpha}$ _Vß₃ integrin is active in virus receptoractivity even when its activity for cell attachment to vitronectinis abolished by either removing or mutating the cytoplasmicdomain. There is a possibility that viral entry mediated by ${alpha}$ _Vß₃ might not use the clathrin-dependent pathway.In contrast, the ${alpha}$ _Vß₆ integrin requires the cytoplasmicdomain for binding, internalization, or both. Removal of thethree C-terminal residues of the ß₅ subunit preventedthe release of the human adenovirus genome mediated by the ${alpha}$ _Vß₅integrin to the cytosol but did not prevent virus internalization(94).

The data presented here indicate that FMDV interacts at thecell surface with integrins, and a mechanism that regulatesreceptor recycling is used by the virus during the internalizationprocess. The virus enters the cell using the clathrin-mediatedendocytosis pathway, trafficking throughout the acidified endocyticvesicles, where its capsid rapidly dissociates into pentamers,resulting in the release of the RNA genome.

ACKNOWLEDGMENTS

We thank Terry Jackson for sharing his data with us prior topublication.

This work was supported by the U.S. Department of Agriculture,Agricultural Research Service, through CRIS project no. 1940-32000-035-00D.

FOOTNOTES

* Corresponding author. Mailing address: USDA, ARS, Plum Island Animal Disease Center, PO Box 848, Greenport, NY 11944-0848. Phone: (631) 323-3354. Fax: (631) 323-3006. E-mail: bbaxt{at}piadc.ars.usda.gov.

REFERENCES

Top