Role of Proteases in the Release of Porcine Epidemic Diarrhea Virus from Infected Cells

Abstract

Porcine epidemic diarrhea virus (PEDV), a causative agent of pig diarrhea, requires a protease(s) for multicycle replication in cultured cells. However, the potential role of proteases in the infection process remains unclear. In order to explore this, we used two different approaches: we infected either Vero cells in the presence of trypsin or Vero cells that constitutively express the membrane-associated protease TMPRSS2 (Vero/TMPRSS2 cells). We found that PEDV infection was enhanced, and viruses were efficiently released into the culture fluid, from Vero cells infected in the presence of protease, while in cells without protease, the virus grew, but its release into the culture fluid was strongly hampered. Cell-to-cell fusion of PEDV-infected cells and cleavage of the spike (S) protein were observed in cells with protease. When infected Vero cells were cultured for 3 days in the absence of trypsin but were then treated transiently with trypsin, infectious viruses were immediately released from infected cells. In addition, treatment of infected Vero/TMPRSS2 cells with the protease inhibitor leupeptin strongly blocked the release of virus into the culture fluid. Under electron microscopy, PEDV-infected Vero cells, as well as PEDV-infected Vero/TMPRSS2 cells treated with leupeptin, retained huge clusters of virions on their surfaces, while such clusters were rarely seen in the presence of trypsin and the absence of leupeptin in Vero and Vero/TMPRSS2 cells, respectively. Thus, the present study indicates that proteases play an important role in the release of PEDV virions clustered on cells after replication occurs. This unique observation in coronavirus infection suggests that the actions of proteases are reminiscent of that of the influenza virus neuraminidase protein.

INTRODUCTION

Porcine epidemic diarrhea virus (PEDV), a member of the group I coronaviruses (CoVs), is an enveloped virus with a single-stranded, positive-sense RNA genome of about 30 kb (21). PEDV, a causative agent of pig diarrhea, induces loss of appetite and weight in adult pigs and is lethal in piglets (35).

The spike (S) protein of the CoVs, a large glycoprotein of ca. 180 to 200 kDa, is a class I fusion protein (5). Three molecules of the S protein (homotrimer) constitute a spike on the virion. The CoV S proteins are cleaved by host-derived proteases into two subunits: the N-terminal S1 subunit and the C-terminal membrane-anchored S2 subunit. S1 binds to its receptor, while S2 is responsible for fusion activity (43, 45). This cleavage is thought to be essential for the induction of cell-to-cell fusion and virus entry into cells (5, 18, 26, 41, 43). Several different proteases are known to be utilized for cleavage of the S protein of each CoV. For example, the S protein of the murine CoV mouse hepatitis virus (MHV), with a basic amino acid cluster in the middle of its molecule, is cleaved by a protease, furin, during its biogenesis, and the cleaved S protein is retained on the virion and infected-cell surfaces, inducing cell-to-cell fusion (14, 43, 45). In contrast to the MHV S protein, S proteins of some other CoVs, such as those of severe acute respiratory syndrome CoV (SARS-CoV), nonfusogenic MHV-2, and human CoV 229E (HCoV-229E), have no furin recognition site, and accordingly, their S proteins are not cleaved during their biogenesis (18, 27, 36, 41, 48, 49).

However, the S protein of CoVs without a furin recognition site is thought to be cleaved by proteases in the endosome, cathepsins, and other proteases active only in a low-pH environment (4, 15, 27, 40). These CoVs, once bound to the receptor, are transported to the endosome, where the S protein is cleaved and activated for fusion, which, in turn, results in the release of the virus genome into the cytoplasm from the endosome (41). Thus, these CoVs fail to induce syncytia in infected cells, and the S protein on the virion is not in a cleaved form (4, 27, 41). However, those S proteins are cleaved by exogenous proteases, such as trypsin, into two subunits similar to MHV S1 and S2 (19, 45, 48), and this cleavage event leads to the activities of cell-to-cell and cell-to-virus fusion (18, 36, 41, 48, 49), although the efficiency of infection of those CoVs is not highly influenced by exogenous proteases (18, 26, 27). PEDV has uncleaved S protein (10), and PEDV-infected cells produce syncytia only after treatment with an exogenous protease, features similar to those of the CoVs described above. However, without the exogenous protease trypsin, PEDV cannot grow efficiently in cultured cells (20, 34). This finding differs from the results of studies of other CoVs with uncleaved S protein and strongly suggests to us that exogenous proteases play an important role in PEDV infection, but one entirely different from that found in other CoVs.

Recently, a trypsin-like serine protease, transmembrane type II serine protease 2 (TMPRSS2), was reported to cleave the hemagglutinin (HA) protein of influenza virus and to allow multicycle replication of influenza virus in the absence of trypsin (6). Shirogane et al. also reported, in studies using cells constitutively expressing TMPRSS2, that TMPRSS2 cleaved the fusion protein of human metapneumovirus and enhanced virus replication (38). More recently, it was shown that TMPRSS2 induced not only cell fusion but also infection of cells with SARS-CoV (12, 25, 39). These findings may indicate that TMRSS2 could be associated with the biological functions of several viruses. However, the effects of TMPRSS2 on infections caused by other viruses that require protease have not been studied so far.

To address the role played by proteases in PEDV infection, we used two different proteases, trypsin and membrane-bound TMPRSS2, and clarified that the proteases play an important role in the release of viruses from infected cells. We determined that in PEDV-infected cells without proteases, virions formed clusters or aggregates on the infected-cell surface, which were released by subsequent protease treatment. The phenomenon is unique and indicates that proteases are crucial for the efficient spread of PEDV to neighboring cells. This protease action is reminiscent of the role played by neuraminidase (NA) in influenza virus infection, in that NA facilitates the release of influenza virus virions attached to their receptor on the cell surface (1, 23, 31, 32).

MATERIALS AND METHODS

Cells, virus, and serum.

Vero cells were obtained from the American Type Culture Collection (ATCC; Manassas, VA). COS7 cells were obtained from the Health Science Research Resources Bank (HSRRB; Osaka, Japan). Vero cells stably expressing TMPRSS2 (Vero/TMPRSS2 cells) were used in this study (38). Those cells were maintained with Dulbecco's modified Eagle's medium (DMEM; Sigma, St. Louis, MO) containing 5% fetal calf serum (FCS). PEDV strain MK was kindly provided by Tetsuo Nunoya (The Nippon Institute for Biological Science, Tokyo, Japan) and was propagated by using Vero cells. Briefly, Vero cells were inoculated with PEDV and were incubated at 37°C for 1 h of virus adsorption. Cells were then washed with phosphate-buffered saline, pH 7.4 (PBS), and DMEM containing 10% tryptose phosphate broth (TPB), and 2.5 μg/ml of trypsin (Sigma) was added to the cells. After 24 h of incubation, the cells and supernatant were collected together and were stored at −80°C until use.

Virus titration.

Virus infectivity was determined by a plaque assay. Vero cell monolayers grown in 24-well plates coated with type I collagen (AGC Techno Glass Co. Ltd., Chiba, Japan) were inoculated with 125 μl of serially diluted virus samples. After 1 h of virus adsorption, the inoculants were aspirated, and the cells were first washed with PBS and then cultured in 0.5 ml of DMEM containing 10% TPB and 1.25 μg/ml of trypsin. After 15 h of incubation at 37°C, cells were fixed with 20% formalin and were stained with PBS containing 0.1% crystal violet. Syncytia were counted under the microscope as plaques. The virus titer was expressed as PFU.

Real-time PCR.

Virus amounts were also determined by the real-time PCR assay as follows. Viral RNA was extracted from samples with TRIzol LS reagent (Invitrogen, Carlsbad, CA) by following the manufacturer's protocol. First-strand cDNA was synthesized by using Moloney murine leukemia virus (M-MLV) reverse transcriptase (Takara Bio Inc., Shiga, Japan) and oligo(dT)₁₆ (Applied Biosystems, Foster City, CA). To determine the copy number of viral RNA, real-time PCR was performed in duplicate with synthesized cDNA by using the LightCycler 480 system and Probe Master reagents (Roche, Basel, Switzerland) with a specific primer and probe set (sense, 5′-ACGGCGACTACTCAGC-3′; probe, 6-carboxyfluorescein [FAM]-5′-CCGCAAACGGGTGC-3′-minor groove binder [MGB]; antisense, GGGCATAAAGGGATAAT) constructed to amplify the nucleocapsid gene of PEDV. Real-time PCR was performed under the following conditions: 5 min at 95°C, followed by 45 cycles of 10 s at 95°C, 10 s at 40°C, and 20 s at 72°C. The copy numbers of viral RNA were calculated with reference to the standard curve obtained from a serially diluted plasmid that contained a target region sequence.

Effects of proteases and a protease inhibitor on virus replication.

To investigate the replication kinetics, Vero and Vero/TMPRSS2 cells were seeded in 24-well, type I collagen-coated plates, and the cells were infected with PEDV at a multiplicity of infection (MOI) of 0.01. After virus adsorption, cells were washed with PBS twice and were then cultured with DMEM containing 10% TPB. Trypsin (1.25 μg/ml) was added to the medium at different times after infection, depending on the experiment. At the indicated hours after infection at 37°C, supernatants and cells were collected separately and were stored at −80°C until use. To observe syncytium (cell-to-cell fusion) formation, Vero and Vero/TMPRSS2 cells were infected with PEDV at an MOI of 0.1 and were incubated in the presence or absence of trypsin. After the indicated hours of incubation, the culture medium was replaced with PBS, and cell images were captured with a DS-Fi1 camera (Nikon, Tokyo, Japan). For the detection of virus released from cells after transient trypsin treatment, Vero cells were infected with PEDV at an MOI of 0.1, incubated for 1 h at 37°C for virus adsorption, and cultured with DMEM containing 5% FCS. After 3 days of incubation at 37°C without trypsin, cells were washed with PBS twice, and 300 μl of DMEM containing either no trypsin or the indicated concentration of trypsin (200 or 20 μg/ml) was added to the cells. After 5 min of incubation at room temperature, supernatants were collected. Cells were also collected with 300 μl fresh medium and were ultrasonicated. For the detection of virus after protease inhibitor treatment, Vero/TMPRSS2 cells were infected with PEDV at an MOI of 0.1. After virus adsorption, cells were washed with PBS twice, and 300 μl of DMEM containing 10% TPB was added. After another 4 h of incubation at 37°C, DMEM containing 10% TPB and either no leupeptin or the indicated concentration of leupeptin (500, 100, or 20 μM; Roche) were added. Twenty hours later, the supernatant and cells were collected separately as described above.

Western blot analysis.

To determine the cleavability of the S protein by trypsin, Vero cells were infected with PEDV and were incubated without trypsin. After 3 days of incubation, cells were collected; then they either were treated with the indicated concentration of trypsin at room temperature for 5 min or were first washed twice with PBS and then treated with trypsin at room temperature for 5 min. After trypsin treatment, samples were mixed with sodium dodecyl sulfate (SDS) sample buffer containing the protease inhibitor Complete Mini (Roche). To examine the cleavability of S protein by TMPRSS2, PEDV strain MK S protein was cloned into plasmid pTargeT (pT-MK-S; Promega, Fitchburg, WI). COS7 cells were transfected with the indicated amount (in micrograms) of pT-MK-S or empty pTargeT and 1 μg of a TMPRSS2-expressing plasmid (pcDNA/TMPRSS2) (38) by using the DMRIE-C reagent (Invitrogen) according to the manufacturer's protocol. After 3 days of incubation, cells were collected with SDS sample buffer. Plasmid pCAG/HA, expressing the influenza virus HA protein (subtype H3), was used as a positive control. All samples collected were separated by SDS-polyacrylamide gel electrophoresis (PAGE), and proteins were transferred to a nitrocellulose membrane. PEDV S protein was detected with rabbit antisera, which recognized the cytoplasmic tail region sequence of PEDV S protein (residues 1366 to 1380; CRGPRLQPYEAFEKV), and anti-rabbit IgG (H+L) guinea pig serum (Rockland Immunochemicals, Inc., Gilbertsville, PA). HA protein was detected by using an anti-HA rabbit serum specific to influenza virus strain Udorn and an anti-rabbit IgG (H+L) guinea pig serum.

Electron microscopy.

For electron microscopic examination, PEDV- and mock-infected Vero cells were collected after 3 days of incubation at 37°C. PEDV-infected Vero/TMPRSS2 cells cultured with 500 μM leupeptin were also collected. PEDV-infected Vero cells cultured with trypsin were collected after 15 h of incubation. Cells were detached with Cell Dissociation Solution Non-enzymatic (Sigma) and were washed with PBS twice. For ultrastructural studies, cells were centrifuged at 15,000 rpm for 1 min, fixed with 2.5% glutaraldehyde with 2% paraformaldehyde at 4°C, postfixed in 1% osmium tetroxide, dehydrated, and embedded in epoxy resin. Ultrathin sections were stained with uranyl acetate and lead citrate and were examined with a JEM-1400 transmission electron microscope (JEOL Ltd., Tokyo, Japan).

Immunofluorescence assay.

Vero and Vero/TMPRSS2 cells were infected with PEDV at an MOI of 0.01 and were cultured in the absence of proteases as described above. At the indicated hours postinfection, cells were fixed with methanol-acetone for 5 min and were then washed with PBS. For the first staining, cells were treated with PEDV-infected pig serum for 60 min. After washing with PBS, cells were treated with a fluorescein isothiocyanate (FITC)-conjugated anti porcine IgG (H+L) rabbit polyclonal antibody (Invitrogen) for the second staining. Following washing, fluorescence images were captured by using a fluorescence microscope.

Statistical analysis.

Statistical significance was determined by using an unpaired t test. P values of <0.05 were considered statistically significant.

RESULTS

PEDV replication in Vero cells with or without proteases.

The effect of proteases on PEDV infection was examined by using trypsin in Vero cell culture and Vero/TMPRSS2 cells (Vero cells that constitutively express TMPRSS2) (38). Vero and Vero/TMPRSS2 cells were infected with PEDV strain MK, and Vero cells were cultured in the presence or absence of trypsin. Intracellular and extracellular virus titers were examined separately at intervals. Intracellular virus titers in Vero cells and Vero/TMPRSS2 cells were not remarkably different (Fig. 1 b); however, trypsin or TMPRSS2 enhanced replication slightly (ca. 10-fold). In contrast, the virus titers in the extracellular fractions were very distinct. PEDV-infected Vero cells cultured in the presence of trypsin and PEDV-infected Vero/TMPRSS2 cells displayed extracellular virus titers 100- to 1,000-fold higher than those of Vero cells cultured in the absence of trypsin (Fig. 1a). We also monitored the levels of viral RNA; extracellular viral RNA levels were 100- to 1,000-fold higher in cells infected with PEDV and cultured in the presence of trypsin, or in Vero/TMPRSS2 cells, than in Vero cells cultured in the absence of trypsin (Fig. 1c). However, no such notable differences were observed in the intracellular viral RNA levels (Fig. 1d). These findings indicated that proteases play an important role in the efficient release of viruses into extracellular culture fluid.

Fig. 1.

Replication kinetics of PEDV in cells in the presence or absence of protease. Vero and Vero/TMPRSS2 cells were infected with PEDV strain MK at an MOI of 0.01. One hour after virus adsorption, cells were washed with PBS and were incubated in DMEM containing 10% TPB. Infected Vero cells were then cultured in the presence or absence of trypsin (1.25 μg/ml), and culture fluid and cells were collected separately at the indicated hours postinfection. (a and b) PEDV titers in culture fluid (extracellular) (a) and cells (intracellular) (b) were determined by a plaque assay using Vero cells and are expressed as log PFU/ml for extracellular titers and log₁₀ PFU/well for intracellular titers. (c and d) Levels of PEDV-specific RNA in culture fluid (extracellular) (c) or in cells (intracellular) (d) were also determined by real-time reverse transcription-PCR. The copy numbers were calculated by referring to the standard curve made by serially diluted positive-control plasmids and are expressed as log₁₀ RNA copies/2.5 μl. Results of six experiments are shown. Symbols: ⧫, Vero cells without trypsin; □, Vero cells cultured in the presence of trypsin; ▵, Vero/TMPRSS2 cells.

Under the conditions employed in the experiments described above, we examined syncytium (cell-to-cell fusion) formation. PEDV infection induced cell fusion in Vero cells cultured with trypsin and in the Vero/TMPRSS2 cell culture, while syncytium formation was never seen in PEDV-infected Vero cells cultured in the absence of trypsin (Fig. 2 a). Cell fusion in Vero/TMPRSS2 cells expanded with time, while the entire Vero cell culture was included in the fusion at 24 h of culture in the presence of trypsin (Fig. 2a). Additionally, we examined the expansion of the PEDV infection by immunofluorescence (Fig. 2b). The spreading of virus in Vero cells was limited in the absence of trypsin, while the virus spread rapidly through cell-to-cell fusion in Vero/TMPRSS2 cells. A huge area of cell fusion was observed in PEDV-infected Vero cells in the presence of trypsin (data not shown).

Fig. 2.

Syncytium formation and antigen expansion after virus infection. (a) Syncytium formation. Vero and Vero/TMPRSS2 cells were infected with PEDV and were cultured in the presence or absence of trypsin, as described in the Fig. 1 legend. After the indicated hours of incubation, the culture medium was replaced with PBS, and images were captured. (b) Expansion of the PEDV antigen in Vero and Vero/TMPRSS2 cells. Vero and Vero/TMPRSS2 cells were infected with PEDV as described in the legend to Fig. 1. At the indicated hours postinfection, cells were fixed with methanol-acetone for 5 min and were stained with an anti-PEDV pig serum and an FITC-conjugated anti-porcine IgG (H+L) rabbit polyclonal antibody.

These observations collectively suggest that proteases such as trypsin and membrane-bound TMPRSS2 contribute to the enhancement of PEDV infection, most likely by releasing the cell-associated virions into the extracellular environment and by rapid cell-to-cell fusion.

Effects of proteases on S protein cleavage.

We then investigated whether those proteases induce the cleavage of PEDV S protein. Vero cells infected with PEDV as described above were incubated at 37°C for 3 days without trypsin. After washing with PBS, cells were treated with trypsin at room temperature for 5 min and were then collected (Fig. 3 a). To determine the cleavage of PEDV S protein by TMPRSS2, COS7 cells were transfected first with an expression vector containing the PEDV S gene and then with a vector expressing the TMPRSS2 gene. Cell lysates were prepared at 3 days after transfection (Fig. 3b). Cleavage was monitored by Western blotting using an antibody against synthetic peptides encompassing the PEDV S protein C-terminal region. As a positive control for TMPRSS2-mediated cleavage of the protein, we used the influenza virus HA protein expressed in COS7 cells in a manner similar to that in which the PEDV S protein was used (Fig. 3c). As shown in Fig. 3, the S protein of PEDV was cleaved by trypsin treatment, and ca. 140-kDa and 60-kDa proteins were observed (Fig. 3a). Since an antibody reacting with the C-terminal end of the PEDV S protein was used for Western blot analysis, the results described above imply that two sites on the S protein were cut by trypsin. The site of the cleavage producing the 140-kDa protein is far upstream of the cleavage site observed in the S protein of MHV cleaved by furin. The 60-kDa protein could have resulted from cleavage at a region far downstream of the putative cleavage site found in cleaved MHV S protein. The region contains small clusters of basic amino acids, as seen in the SARS-CoV S protein (24). Two additional bands found by treatment with high concentrations of trypsin would not be involved in fusion activity, since cells were fused without these bands after treatment with low concentrations of trypsin. The S protein was also cleaved by coexpression with TMPRSS2 (Fig. 3b), but the cleavage was very weak compared to that of the influenza virus HA protein (Fig. 3c). The fraction of ca. 160-kDa protein was detected after the cleavage of the PDEV S protein with TMPRSS2. We also investigated whether S proteins in Vero/TMPRSS2 cells could be cleaved or not, since fusion as well as efficient release of PEDV was observed in those cells. In contrast to the results with trypsin treatment, only trace amounts of a cleavage product were detected (data not shown). This observation is similar to the finding of SARS-CoV S protein cleavage by TMPRSS2 (25). These results suggest that the protease-induced cleavage of the S protein may be responsible for syncytium formation in PEDV infection. However, the cleavage site for fusion activation was not precisely restricted, since both trypsin and TMPRSS2 induced cell fusion in spite of their different cleavage patterns. This mechanism of fusion by distinct cleavage products of S protein is similar to the fusion mechanisms of SARS-CoV and other coronaviruses (3, 18, 26, 47, 48).

Fig. 3.

Cleavage of PEDV S protein by treatment with proteases. (a) Vero cells were infected with PEDV and were cultured in the absence of trypsin for 3 days. Then the cells were treated with the indicated concentrations of trypsin and were analyzed by Western blotting. (b) COS7 cells were transfected with pT-MK-S or with the empty pTargeT vector and pcDNA/TMPRSS2. After 3 days of incubation, cells were collected and lysed, and S proteins in the lysates were detected by Western blotting. (c) pCAG/HA was used as a positive control for TMPRSS2 digestion in a manner similar to that for the expressed PEDV S protein. Arrows indicate either S or HA protein cleavage products.

Virus release from PEDV-infected Vero cells by trypsin treatment.

The results described above indicated that proteases increase extracellular virus titers and also suggested to us that proteases may promote cell fusion in PEDV infection. We then focused on the observation that efficient release of the virus into culture fluid was facilitated by proteases. Vero cells infected as described above were incubated for 3 days in the absence of trypsin, and the intracellular and extracellular virus titers were examined. The findings were similar to those seen as a result of virus growth kinetics (shown in Fig. 1), in that high titers of infectious virus were found in intracellular fractions, while extracellular virus titers were near the detection limit (Fig. 4 a and b). When those cells were treated with varying concentrations of trypsin for 5 min at room temperature, we began to detect significantly large amounts of infectious viruses in culture fluid (Fig. 4b), as well as large amounts of viral RNA (detected by real-time PCR) (Fig. 4c). Extracellular virus titers increased with increasing concentrations of trypsin, apparently indicating that trypsin treatment enhanced the virus titer in the culture fluid. Such an observation suggests to us either that viruses that had been stuck to the cell surface became free and were released from infected cells by trypsin treatment or that PEDV-infected cells are prone to trypsin treatment, and viruses were released from the cytoplasm of cells by trypsin treatment.

Fig. 4.

Virus release from PEDV-infected Vero cells after transient treatment with trypsin. Vero cells were infected with PEDV at an MOI of 0.1. After virus adsorption, cells were washed with PBS twice and were incubated with DMEM containing 5% FCS without trypsin. After 3 days of incubation, cells were washed with PBS three times, and DMEM containing trypsin (200 or 20 μg/ml) was added to the cells. After 5 min of incubation, fluids and cells were collected separately. (a and b) PEDV titers in cells (intracellular) (a) and in culture fluid (extracellular) (b) were determined by plaque assays. (c) PEDV RNA copy numbers in culture fluid (extracellular) were determined by real-time PCR. Results of six experiments are shown. *, P < 0.01.

Effect of a TMPRSS2 inhibitor on virus release.

We then tried to confirm that protease facilitates the release of viruses into the culture fluid by using another method in which Vero/TMPRSS2 cells infected with the virus released high titers into the culture fluid, as shown in Fig. 1. We reasoned that if virus release could be attributed to TMPRSS2, then we should be able to suppress virus release with an inhibitor of TMPRSS2, the serine and cysteine protease inhibitor leupeptin. Therefore, Vero/TMPRSS2 or Vero cells were infected with PEDV as described above and were incubated with DMEM containing various concentrations of leupeptin. Virus titers in culture fluids and cells were monitored 24 h after infection (Fig. 5). Intracellular virus titers in Vero/TMPRSS2 cells were higher than those in Vero cells, results similar to those shown in Fig. 1, although the titers were slightly decreased by treatment with leupeptin, depending on its concentration (Fig. 5a). Infectious virus was detected in the culture fluid of virus-infected Vero/TMPRSS2 cells, but the titers were decreased in a dose-dependent manner by leupeptin treatment (Fig. 5b). When Vero/TMPRSS2 cells were cultured in the presence of 500 μM leupeptin, the infection was strongly inhibited. This phenomenon was also confirmed by quantification of viral RNA by real-time PCR assays (Fig. 5c). These findings also suggested to us that serine proteases, such as TMPRSS2, play an important role in the release of virus into the culture fluid.

Fig. 5.

Inhibition of virus release from PEDV-infected Vero/TMPRSS2 cells after treatment with a serine and cysteine protease inhibitor (leupeptin). Vero and Vero/TMPRSS2 cells were infected with PEDV at an MOI of 0.1. After virus adsorption, cells were washed with PBS twice and were incubated with 10% TPB-DMEM for 4 h. Then the medium was replaced with 10% TPB-DMEM containing the indicated concentration of leupeptin (500, 100, or 20 μM). After 24 h of incubation, culture fluid and cells were collected separately. PEDV titers and viral RNA copy numbers were determined as described in the Fig. 4 legend. (a and b) PEDV titers in cells (intracellular) (a) and in culture fluid (extracellular) (b). (c) Viral RNA copies in culture fluid (extracellular). Results of four experiments are shown. *, P < 0.01. Shaded bars, Vero/TMPRSS2 cells; open bars, Vero cells.

Electron microscopic observation of PEDV-infected cells.

All of the data discussed above seem to indicate that PEDV can infect and grow in Vero cells without trypsin or TMPRSS2; however, the release of viruses is blocked in the absence of such proteases. These findings point to the possibility that PEDV fails to detach or be released from cells without trypsin or TMPRSS2. To address this hypothesis, cells infected with PEDV were observed under electron microscopy (Fig. 6). In PEDV-infected Vero cells cultured in the absence of trypsin, huge clusters of virions were seen on the cell surface (Fig. 6a and b), while clusters were rarely seen on the surfaces of PEDV-infected Vero cells cultured in the presence of trypsin (Fig. 6c and d). Similarly, only small clusters of virions were seen on the surfaces of PEDV-infected Vero/TMPRSS2 cells (Fig. 6e and f), although large clusters of virions were seen widely distributed on PEDV-infected Vero/TMPRSS2 cells cultured in the presence of leupeptin (Fig. 6g and h). To quantify the virions released from clusters, we counted the virions included in the clusters. As shown in Fig. 6i, the mean numbers of virions in untreated PEDV-infected Vero cells and PEDV-infected Vero/TMPRSS2 cells treated with leupeptin were much higher than those in PEDV-infected Vero cells treated with trypsin or untreated PEDV-infected Vero/TMPRSS2 cells (102 ± 39 and 149 ± 43 versus 23 ± 21 and 43 ± 12 virions, respectively [n = 4; P < 0.01]). These observations indicate that the viruses stick to the cell surface or form aggregates associated with the cell surface, and they lead us to suggest that PEDV requires proteases for release from the cell surface, but not from the cytoplasm, as a result of cell destruction by proteases.

Fig. 6.

Electron microscopic examination of PEDV-infected cells. PEDV-infected Vero (a and b) and Vero/TMPRSS2 (e and f) cells were collected after 3 days of incubation at 37°C. PEDV-infected Vero cells cultured with 1.25 μg/ml of trypsin for 15 h (c and d) and PEDV-infected Vero/TMPRSS2 cells cultured with 500 mM leupeptin for 3 days (g and h) were also collected. Cells were detached with Cell Dissociation Solution Non-enzymatic (Sigma) and were washed twice with PBS, and ultrathin sections were prepared. High magnification, ×6,000; low magnification, ×3,000. (i) Virions in the high-magnification fields were counted. Results of four experiments are shown. *, P < 0.01.

DISCUSSION

Host cell-derived proteases play critical biological roles in a variety of virus infections. It is well known that some viral envelope glycoproteins, such as influenza virus HA, HIV gp160, paramyxovirus glycoprotein, and MHV S protein, are cleaved by host proteases, most likely by furin, and that this cleavage is required for the proteins to execute their biological roles (8, 13, 28, 37, 46). The cleaved fractions are associated with each other through covalent or noncovalent linkage and work to enable the entry of the virus into cells. Influenza virus HA, for example, is synthesized as a single molecule, HA0, and is then cleaved during biogenesis into two subunits: the N-terminal HA1 and the C-terminal HA2 subunit. HA1 is responsible for receptor binding, while HA2 is critical for envelope-cell membrane fusion, which facilitates the entry of the viral genome into cells (42, 44). In other viruses, such as Ebola virus and some coronaviruses, such host-derived proteases as cathepsins are thought to be critical for the activation of envelope glycoproteins via cleavage of the molecules expressed on the virion envelope, which occurs in an endosome at the very last stage of the infection cycle (4, 7, 15, 17, 18, 26). Thus, host cell-derived proteases are important factors for virus replication and infection in host animals, although no exogenous proteases are required for the efficient replication of these viruses.

On the other hand, PEDV has been known to require a protease, trypsin, to produce cell-to-cell fusion and efficient multicycle infection (20). However, the precise mechanism underlying these effects has remained unknown. In the present study, we showed a novel role of proteases in PEDV infection: causing the release of virions from infected cells into the culture fluid. Both infectious and noninfectious virions were rarely detected in the culture fluid of PEDV-infected cells in the absence of proteases, as revealed by a plaque assay and real-time PCR. However, after treatment of those cells with proteases, infectious free virions became detectable in culture fluids. Electron microscopic observations revealed that the clusters or aggregates of virions found on the surfaces of PEDV-infected cells were removed by those proteases. These findings strongly indicate that proteases can free the cell surface-attached virus and promote efficient virus spread toward neighboring cells.

The activity and localization of the proteases are important factors for the pathogenesis of the viruses. Some of the influenza viruses and paramyxoviruses fail to exhibit high pathogenicity because of a lack of furin-mediated cleavage. In SARS-CoV infection, the protease affects the entry pathway of the virus and the efficiency of virus infection (27), possibly exacerbating pneumonia as a result of SARS-CoV infection (2). PEDV causes an infection of the small intestine in pigs (35), which would be influenced by the presence or absence of proteases that promote the release of free viruses from the surfaces of infected cells. Trypsin exists in the intestine under physiological conditions and (along with certain other similar proteases) could contribute to the dissemination of infectious viruses in the intestine. Thus, the proteases may have an influence on the outcome of PEDV infection in animals, as is expected in the exacerbation of SARS by a protease produced in the lungs (2).

The observation that proteases facilitate the release of PEDV from the infected-cell surface may indicate that some molecules sensitive to proteases are involved in the blockade of virus release. Recently, tetherin was reported to block the release of HIV-1 or virus-like particles of Ebola virus by trapping those virions on the cell surface (16), but they were released from the cell surface by proteases (11, 16, 29, 30). These findings lead us to suggest that a tetherin-like molecule could be responsible for trapping the PEDV virion on the cell surface. In our preliminary examination in the present study, tetherin seemed not to affect the release of PEDV, since Vero/TMPRSS2 cells were shown not to express tetherin; nevertheless, TMPRSS2 enhanced the release of virions (data not shown).

Alternatively, the electron microscopic examination hinted at the possibility that PEDV aggregates are formed on the cell surface by binding to the PEDV receptor, in a manner similar to that of the formation of virion clusters in influenza viruses deficient in NA activity. It is generally accepted that influenza virus NA promotes the release of influenza virus virions attached to the cell surface via receptors by cleaving the receptor sialic acid (1, 31, 32). Liu et al. reported that NA-deficient influenza virus showed an aggregation of progeny viruses on the cell surface (23). Similar findings have also been reported for influenza virus infection when infected cells were treated with such anti-influenza virus NA compounds as oseltamivir or zanamivir (9). This formation of aggregates on the cell surface by influenza viruses lacking NA activity resembles the production of PEDV aggregates on the cell surface in the absence of proteases, suggesting that a host protease may play a role in PEDV infection similar to that of NA in influenza virus infection, by digesting the virion receptor binding protein and thus releasing clustered virions from infected cells. Although it has been reported that aminopeptidase N (APN) might be a cellular receptor of PEDV, this notion is still controversial, since cells without APN, such as Vero cells, are susceptible to PEDV (22). Clarifying the cellular receptor is a prerequisite for testing the above possibility.

It is also possible to speculate that the cleavage of the S protein, which is responsible for cell-to-cell fusion, simultaneously promotes the release of virions from clusters located on the cell surface. The cleavage of the S protein could result in conformational changes of that protein, as recently revealed in the case of MHV-2 infection (26). Such changes may, in turn, trigger the dissociation of the virion-cell or virion-virion interaction, thereby releasing free virions from infected cells. This possibility implies that the cleavage of S protein by proteases has two important biological roles in PEDV infection: the induction of cell fusion and the release of virions from infected cells. Since the S proteins expressed on cells induced fusion when treated with trypsin (data not shown), it is clear that only S protein is involved in syncytium formation, while some other viral protein could be involved in the release of virions from the cell surface. Such a possibility should be studied in the future.

The results of this study also showed that expression of human TMPRSS2 on Vero cells enhanced the multicycle replication of PEDV. Recently, expression of TMPRSS2 was shown to facilitate the multicycle replication of influenza virus (6) and human metapneumovirus (38) without additional proteases in the cell culture medium, and their glycoproteins were revealed to be cleaved by TMPRSS2 (6, 38). TMPRSS2 is expressed in respiratory tissue, but also in the intestines (33). These findings suggest the possibility that porcine TMPRSS2-like proteases are involved in the tissue tropism and pathogenesis of PEDV, although no porcine homologue of human TMPRSS2 has been identified yet. We would like to examine whether proteases such as TMPRSS2 actually contribute to the pathogenesis of PEDV.

PEDV-infected cells exhibited syncytium formation when treated with trypsin. PEDV is similar to SARS-CoV, HCoV-229E, and MHV-2 in that there is an uncleaved S protein on the virion. Trypsin induced the cleavage of the S proteins of these three coronaviruses, as well as that of the PEDV S protein, as shown in the present study. These observations suggest to us that PEDV enters into cells in a manner similar to that of these other coronaviruses. As expected, PEDV seems to take an endosomal pathway, and the acidic conditions of the endosome are required for cell entry; however, involvement of the proteases (cathepsins) utilized for the entry of SARS-CoV, HCoV-229E, and MHV-2 has not been clearly demonstrated for the entry of PEDV (data not shown). At the moment we are studying the detailed entry mechanism of PEDV.

In summary, the results of this study suggested to us that proteases play a critical role in the release of viruses from the cell surface and the enhancement of PEDV infection. This might result in the exacerbation of PEDV-mediated diarrhea in host animals. If so, a protease inhibitor may be one of the good candidates for developing an anti-PEDV compound to combat this infectious disease.