Porcine Epidemic Diarrhea Virus 3C-Like Protease-Mediated Nucleocapsid Processing: Possible Link to Viral Cell Culture Adaptability

ABSTRACT

Porcine epidemic diarrhea virus (PEDV) causes severe diarrhea and high mortality rates in newborn piglets, leading to massive losses to the swine industry worldwide during recent epidemics. Intense research efforts are now focusing on defining viral characteristics that confer a growth advantage, pathogenicity, or cell adaptability in order to better understand the PEDV life cycle and identify suitable targets for antiviral or vaccine development. Here, we report a unique phenomenon of PEDV nucleocapsid (N) cleavage by the PEDV-encoded 3C-like protease (3Cpro) during infection. The identification of the 3Cpro cleavage site at the C terminus of N supported previous observations that PEDV 3Cpro showed a substrate requirement slightly different from that of severe acute respiratory syndrome coronavirus (SARS-CoV) 3Cpro and revealed a greater flexibility in its substrate recognition site. This cleavage motif is present in the majority of cell culture-adapted PEDV strains but is missing in emerging field isolates. Remarkably, reverse-genetics-derived cell culture-adapted PEDV_AVCT12 harboring uncleavable N displayed growth retardation in Vero E6-APN cells compared to the wild-type virus. These observations altogether shed new light on the investigation and characterization of the PEDV nucleocapsid protein and its possible link to cell culture adaptation.

IMPORTANCE Recurrent PEDV outbreaks have resulted in enormous economic losses to swine industries worldwide. To gain the upper hand in combating this disease, it is necessary to understand how this virus replicates and evades host immunity. Characterization of viral proteins provides important clues to mechanisms by which viruses survive and spread. Here, we characterized an intriguing phenomenon in which the nucleocapsids of some PEDV strains are proteolytically processed by the virally encoded main protease. Growth retardation in recombinant PEDV carrying uncleavable N suggests a replication advantage provided by the cleavage event, at least in the cell culture system. These findings may direct us to a more complete understanding of PEDV replication and pathogenicity.

INTRODUCTION

Porcine epidemic diarrhea virus (PEDV) infection results in up to 100% mortality in neonatal piglets and has caused substantial economic damage to the swine industry worldwide. Since the end of 2010, outbreaks in many countries in Asia have caused severe economic losses (1). In 2013 to 2014, a particularly large epidemic swept through all of North America, killing more than 7 million pigs (2, 3). Presently, PEDV has evolved into two distinct clades. Classical PEDV strains are CV777-like strains, while emerging PEDV strains are those appearing after 2010, including highly virulent field strains circulating in Asia (4). Antigenic variations between these two groups were thought to contribute to the severity of recent outbreaks (5, 6). This disease is characterized by severe watery diarrhea, dehydration, and anorexia. Deceased piglets presented with thin and almost transparent small intestines containing undigested milk curdles (7–9).

PEDV is an alphacoronavirus in the Coronaviridae family, and like other coronaviruses (CoVs), it possesses a large positive-sense RNA genome of >28 kb (10). The PEDV genome is composed of two overlapping open reading frames (ORFs) encoding two polyproteins, pp1a and pp1ab, and five other ORFs encoding the following five proteins: spike (S), ORF3, envelope (E), membrane (M), and nucleocapsid (N) (11). The polyproteins are then processed into individual nonstructural proteins by the following virally encoded proteases: papain-like proteases (PLPs) (PLP1/PLP2; nsp3) and 3-chymotrypsin-like protease (3Cpro) (nsp5) (12, 13).

As one of the most abundant and multifunctional PEDV proteins, the PEDV N protein plays a key role in organizing the viral genome through viral RNA (vRNA) binding and self-multimerization (14, 15). Although PEDV replicates exclusively in the cytoplasm, PEDV N has been shown to localize in the nucleolus of infected cells and possesses both nuclear localization and export signals for its nucleocytoplasmic shuttling (16). Besides genome organization, the N protein has been shown to be involved in PEDV pathogenesis and host cell manipulation. For example, stable expression of N in porcine intestinal epithelial cells (IECs) could induce endoplasmic reticulum stress, prolong the S phase of the cell cycle, and upregulate interleukin-8 expression via modulation of NF-κB activity (17). PEDV N was shown to activate NF-κB via Toll-like receptor signaling pathways in IECs (18). In contrast, transient PEDV N expression was also found to inhibit Sendai virus-induced NF-κB activation in HEK 293T cells (19). Moreover, PEDV N has been shown to inhibit interferon beta (IFN-β) production and interferon-stimulating gene expression (19). These data suggest layers of complexity and multiple roles played by PEDV N during the course of PEDV infection.

Additional information regarding the roles and characterization of PEDV N could also be deduced from studies of other related and more comprehensively studied coronaviruses such as severe acute respiratory syndrome coronavirus (SARS-CoV) or transmissible gastroenteritis virus (TGEV). As replication ensues, CoV N proteins interact with M proteins for viral assembly and localize with replicase components for viral replication (14, 20–23). CoV N proteins play an important role in viral RNA synthesis and demonstrate RNA chaperone and RNA silencing suppressor activities (24–26). Additionally, CoV N has been shown to modulate other cellular activities such as cell cycle regulation, host translational shutoff, immune system interference, and host cell signal transduction (14). Host cells also mount antagonistic responses to CoV N. During apoptosis induced by TGEV infection, TGEV N has been shown to be a viral substrate for caspase-dependent degradation (27). For SARS-CoV, N is cleaved by caspases in a cell type-specific manner, and N cleavage seems to be associated with viral titers and cytopathic effects (CPEs) (28). In fact, processing of CoV N seemed to be common in cells infected with coronaviruses, as lower-molecular-mass species of N-derived polypeptides have been observed in cells infected with murine (mouse hepatitis virus [MHV]), feline (feline infectious peritonitis virus [FIPV]), bovine (bovine coronavirus [BCV]), and avian (infectious bronchitis virus [IBV] and turkey coronavirus [TCV]) coronaviruses (15, 27). Whether PEDV N is processed in the same manner has not been reported to date. If so, the role of N processing in the PEDV life cycle still needs further investigation.

Here, we demonstrated that PEDV N was posttranslationally processed by characterizing a major cleavage band (∼43 kDa) observed in lysates prepared from cells infected with the PEDV_AVCT12 strain. Unlike other coronaviruses whose nucleocapsids are especially prone to degradation or cleavage by caspases (15, 27, 28), this PEDV N cleavage product was released by PEDV 3Cpro. The identified cleavage site is found in only some of the classical PEDV strains that have been adapted in the laboratory but not in recently circulating field isolates. Interestingly, when the uncleavable variant of N was introduced into PEDV, we observed disadvantages in PEDV propagation in Vero E6-APN cells. These data add to the molecular characterization of N and may suggest new research avenues regarding the pathogenesis and cell adaptation of PEDV isolates.

RESULTS

PEDV N is cleaved in the presence of the PEDV genome.

During characterization of our reverse-genetics-derived PEDV_AVCT12, we consistently observed multiple bands in robustly infected Vero E6-APN cell lysates when they were probed with anti-N antibodies, indicating further processing of the PEDV N protein during infection (Fig. 1A). However, only the topmost band corresponding to full-length PEDV N appeared when the supernatant containing PEDV particles was analyzed, indicating that only full-length N was packaged into the virions (Fig. 1A). These observations prompted us to further investigate the occurrence of multiple species of N during the course of infection. First, we used the reverse-genetics system for PEDV established in our laboratory to recapitulate infection in a more controlled manner. HEK 293T cells were transfected with the infectious clone encoding the PEDV genome, pSMART-PEDV_AVCT12, and tested for PEDV N cleavage. Besides the 55-kDa band corresponding to full-length N, a major band was observed between 43 and 55 kDa (Fig. 1B). Of note, the cleavage of PEDV N proteins observed during infection of Vero E6-APN cells was much more extensive than that observed during transfection of a PEDV infectious clone in HEK 293T cells (Fig. 1). We also noticed that the extent of PEDV N cleavage varied with experiments and was determined primarily by the extent of viral infection or the conditions for viral propagation (e.g., trypsin addition) in different cell types (data not shown). Interestingly, PEDV N expressed from an expression plasmid (pHW-AVCT12 N) revealed only one major band at the expected size of full-length PEDV N (Fig. 1B). Upon longer exposure, degradation products of PEDV N could be seen, consistent with the observation that coronavirus N proteins are prone to degradation during overexpression (15). Nevertheless, the prominent cleavage band just above the 43-kDa marker (referred to here as the ∼43-kDa band) appeared only in the context of infection or the expression of other viral proteins. Although multiple translation products of N could not be ruled out, we were inclined to believe that these smaller species of N proteins were proteolytic cleavage products, as previously reported for several coronaviruses (15).

FIG 1.

PEDV N is cleaved in the presence of the PEDV genome. Shown are data from Western blot analysis with anti-PEDV N antibodies of Vero E6-APN cells infected with PEDV_AVCT12 (A) or HEK 293T cells that were untransfected (−) or transfected with pSMART-PEDV_AVCT12 (PEDV) or pHW-AVCT12 N (N) (B). Asterisks indicate the major cleavage product (∼43 kDa) observed for both infection and transfection of the infectious clone. cell, infected-cell lysate; sup, supernatant. A longer exposure in another experiment shown in panel B revealed minor degradation events during the overexpression of PEDV N.

The major ∼43-kDa cleavage product is not released by caspases.

Given that cleavage of other CoV N proteins, primarily by caspases during apoptosis, has also been reported (15, 27, 28), we examined N cleavage during PEDV infection in the presence of caspase inhibitors. To this end, Vero E6-APN cells were infected with PEDV_AVCT12. After virus adsorption, infection of cells was continued in infection medium containing either 100 μM Z-Val-Ala-Asp-(OMe)-fluoromethyl ketone [Z-VAD(OMe)-FMK], a pancaspase inhibitor, or dimethyl sulfoxide (DMSO), a solvent for the inhibitor. We observed slightly more robust CPEs and PEDV infection in the presence of Z-VAD(OMe)-FMK (Fig. 2A). At the indicated time points after infection, lysates were analyzed by Western blotting with anti-PEDV N antibodies, and supernatants were analyzed for PEDV RNA contents. In the presence of Z-VAD(OMe)-FMK, the PEDV N degradation pattern observed at late time points was more extensive than that in the presence of DMSO (Fig. 2B). Consistent with CPEs, slightly more robust PEDV replication occurred in the presence of Z-VAD(OMe)-FMK, as indicated by a semiquantitative reverse transcription-PCR (RT-PCR) assay of viral RNAs obtained from equal volumes of infection supernatants (Fig. 2B). Although N degradation seems to correlate with the extent of PEDV infection (Fig. 1 and 2 and data not shown), the release of the ∼43-kDa cleavage products was unperturbed by caspase inhibitors; their intensities were comparable under both conditions at every time point (Fig. 2B). Altogether, these data suggested that during infection, there are at least two mechanisms for PEDV N processing: caspase-independent degradation that is triggered by PEDV N accumulation or the stage of infection and caspase-independent specific cleavage that releases the ∼43-kDa cleavage product. In this study, we focused on the characterization of the latter mechanism.

FIG 2.

Caspases are not responsible for the major cleavage product of PEDV N. Shown are data for Vero E6-APN cells infected with PEDV_AVCT12 in the presence of DMSO or 100 μM Z-VAD(OMe)-FMK. (A) CPEs were observed and photographed at the indicated time points after infection. Bars, 100 μm. (B) Western blot analysis with anti-PEDV N antibodies of lysates prepared at different time points. Asterisks indicate the major cleavage product (∼43 kDa). PEDV vRNAs were extracted from equal volumes of supernatants collected at the indicated time points and analyzed by RT-PCR with PEDV N-specific primers. An ethidium bromide-stained DNA gel is marked as “RT-PCR.” Data are representative of results from two independent experiments.

PEDV 3Cpro cleaves PEDV N.

Since the presence of the PEDV genome was strictly required for PEDV N cleavage, we sought to determine whether PEDV N was cleaved by the PEDV proteases PLP and 3Cpro. These proteases are well conserved within the coronavirus family and have been quite well studied in other coronaviruses (29). Especially for the human pathogen SARS-CoV, 3Cpro is a target for antiviral drug development, and its substrate specificity and mechanism of action have been extensively mapped out (30–32). In addition to these biochemical characterizations, bioinformatics analysis based on known CoV genome sequences and identified 3Cpro cleavage sites led to the generation of the NetCorona 1.0 prediction program for CoV 3Cpro cleavage site prediction (33). With this program, we identified several possible cleavage sites near the C terminus of PEDV N (Fig. 3A). The most likely site, ETTLQ^{^}QHEEA, was given a substantial possibility score (0.508) and had a perfect match to the consensus substrate specificity for the P2 and P1 sites, which are the most important recognition sites for the 3Cpro enzymes (29). Cleavage at this site would release a product with the size of the observed band in our experiments. Based on these observations, we hypothesized that PEDV N was cleaved by its 3Cpro.

FIG 3.

PEDV 3Cpro cleaves PEDV N. (A) Schematic of AVCT12 N with possible 3Cpro recognition sites (black stripes) predicted by NetCorona 1.0. The carets designate the predicted cleavage sites. Numbers in parentheses indicate the scores given by the prediction program. (B and C) Western blot analysis of HEK 293T cells transfected with pCAGGS-GFP^5/6 (B) or pHW-AVCT12 N (C) in the presence or absence of pCAGGS-3Cpro-FLAG variants (wild type or C144A).

To this end, we first generated a working system to detect 3Cpro activity. A pCAGGS-based plasmid expressing PEDV 3Cpro with a C-terminal FLAG tag was constructed. pCAGGS-GFP^5/6 was used as a substrate for PEDV 3Cpro functional assays. This plasmid encoded green fluorescent protein (GFP) harboring an 8-amino-acid natural cleavage site derived from the junction between the nsp5 and nsp6 proteins in the PEDV genome (YGVNLQ^GG) (GFP^5/6). Cleavage by coexpressed PEDV 3Cpro would result in a faster-migrating band (∼22 kDa) in addition to the full-length GFP band (∼28 kDa) when the lysates were probed with anti-GFP antibodies. HEK 293T cells were cotransfected with pCAGGS-3Cpro-FLAG and pCAGGS-GFP^5/6. Western blot analysis with anti-GFP antibodies on lysates prepared from these transfected cells revealed a smaller, fast-migrating band whose size was consistent with that of the cleavage product of GFP^5/6 (Fig. 3B). As a negative control, 3Cpro-FLAG harboring the C144A protease active-site mutation did not release the GFP-based cleavage product, indicating the specificity of the assay (34, 35).

Next, we determined whether PEDV 3Cpro was responsible for processing PEDV N. HEK 293T cells were cotransfected with pHW-AVCT12 N and either the empty vector or pCAGGS-3Cpro-FLAG. Cell lysates were analyzed for cleavage events by Western blotting with anti-PEDV N antibodies. The cleavage product could be visualized only in the presence of 3Cpro-FLAG but not 3Cpro-C144A-FLAG (Fig. 3C). These data suggest that the major cleavage product of PEDV N resulted from the protease activity of PEDV 3Cpro.

PEDV 3Cpro cleaves N at position Q382.

CoV 3Cpro enzymes exhibit a strong preference for leucine at the P2 position (29, 33). To ascertain that the cleavage site computationally predicted by the NetCorona 1.0 program indeed served as a substrate for PEDV 3Cpro, we mutated the leucine at the P2 position (L395) into a glutamine and expected that the mutant N protein would be tolerant of cleavage by PEDV 3Cpro. Surprisingly, efficient cleavage was still observed with PEDV N-L395Q compared to the wild-type protein, suggesting that PEDV 3Cpro might not recognize the predicted cleavage site (Fig. 4A). We additionally corroborated this unexpected result with a GFP cleavage assay. We transplanted the identified cleavage site, RETTLQ^{^}QH, onto GFP (GFP^AVCT12-396) and tested if PEDV 3Cpro could recognize and cleave this substrate in the same manner as in the GFP^5/6 construct. Lysates prepared from HEK 293T cells cotransfected with 3Cpro-FLAG and each of the GFP constructs were analyzed with anti-GFP antibodies for cleavage events. In the presence of 3Cpro-FLAG, GFP^5/6 could be cleaved, but GFP^AVCT12-396 remained intact (Fig. 4B). These data indicate that PEDV 3Cpro did not cleave at the site predicted by NetCorona 1.0.

FIG 4.

PEDV N is not cleaved at the computationally predicted site. (A) Western blot analysis of HEK 293T cells transfected with pHW-AVCT12 N (wild type [WT] or L395Q mutant) in the presence or absence of pCAGGS-3Cpro-FLAG. (B) Western blot analysis of HEK 293T cells transfected with pCAGGS-GFP plasmids containing cleavage sites from the nsp5-nsp6 junction (5/6) or the AVCT12 N protein around residue 396 (AVCT12³⁹⁶) in the presence or absence of pCAGGS-3Cpro-FLAG.

In an effort to identify the cleavage site on PEDV N, we examined whether N proteins from other PEDV strains were also cleaved by 3Cpro. An expression plasmid containing the N gene derived from a field isolate was constructed. This PEDV strain was isolated from a swine farm in Thailand and was sequence matched to a Korean isolate, KNU-141112 (KR) (36). Lysates prepared from HEK 293T cells cotransfected with plasmids expressing 3Cpro and either AVCT12 N or KR N were probed for N cleavage with anti-PEDV N antibodies. As expected, AVCT12 N was cleaved. Interestingly, KR N did not show any cleavage pattern (Fig. 5A). Alignment of amino acid sequences from these two strains revealed differences near the C terminus that could be a potential cleavage site (Fig. 5B). Besides the LQ motif at position 396 that was already disqualified (Fig. 4 and 5B, white triangle), another LQ motif at position 382 is present in the cleavage-sensitive AVCT12 N protein but absent in the cleavage-resistant KR N protein (Fig. 5B, black triangle).

FIG 5.

PEDV 3Cpro cleaves N at residue 382. (A) Western blot analysis of HEK 293T cells transfected with pHW-AVCT12 N or pHW-KR N in the presence or absence of pCAGGS-3Cpro-FLAG. (B) Alignment of amino acid sequences at the C termini of N proteins from strains AVCT12 and KR. The white triangle marks the NetCorona 1.0-predicted cleavage site. The black triangle marks an alternative possible cleavage site. Differences in amino acids are highlighted in gray. (C) Western blot analysis of HEK 293T cells transfected with pHW-AVCT12 N (wild type or L381P mutant) or pHW-KR N (wild type or P381L mutant) in the presence or absence of pCAGGS-3Cpro-FLAG. (D) Western blot analysis of HEK 293T cells transfected with pCAGGS-GFP plasmids containing the cleavage sites from the nsp5-nsp6 junction (5/6), the AVCT12 N protein around residue 382 (AVCT12³⁸²), or the KR N protein around residue 382 (KR³⁸²) in the presence or absence of pCAGGS-3Cpro-FLAG.

To nail down the cleavage site, we tested cleavage on AVCT12 N carrying an L381P mutation and KR N carrying a corresponding mutation, P381L. HEK 293T cells cotransfected with plasmids expressing 3Cpro-FLAG and either AVCT12 N-L381P or KR N-P381L were lysed and analyzed by Western blotting for cleavage patterns. KR N-P381L could now be cleaved in the presence of 3Cpro, whereas AVCT12 N-L381P could no longer serve as a substrate for 3Cpro (Fig. 5C). Note that the faint band appearing below full-length AVCT12 N-L381P in the presence of 3Cpro might indicate an alternative cleavage site. However, this possible cleavage site is not likely, as this slightly larger cleavage product was never observed when the major cleavage site at residue 382 was present. Moreover, GFP containing the 8-amino-acid stretch around the cleavage site from AVCT12 N (GNAKLQ^{^}RK) (GFP^AVCT12-382) could be cleaved, while GFP containing the 8 amino acids derived from KR N (GNAKPQ^{^}RK) (GFP^KR-382) was resistant to cleavage by 3Cpro (Fig. 5D). These data indicate that PEDV 3Cpro processed the nucleocapsid at Q382.

N proteins from some classical PEDV strains harbor the 3Cpro cleavage site.

Since we discovered that the N protein derived from a cell-adapted CV777-like strain was cleaved, while that from the field-isolated KR strain was not, we speculated whether this property might be associated with cell adaptability. We acquired three other field isolates from intestines of infected piglets from various farms in central Thailand. These viruses were circulating in Thailand during the years 2012 to 2015. Notably, these viruses exhibited poor growth in Vero E6-APN cells (data not shown). The N genes of these strains were cloned into an expression vector and tested for cleavage. None of the PEDV N proteins from the field strains showed processing by 3Cpro (Fig. 6A). The results correlated with the observation that their amino acid sequences have a proline at position 381 (Fig. 6B). These results suggest that cleavage of N was specific to some PEDV strains and might be able to serve as a marker of cell adaptation.

FIG 6.

Cleavage of PEDV N is strain specific. (A) Western blot analysis of HEK 293T cells transfected with plasmids expressing N proteins from PEDV_AVCT12 or field isolates from different regions in central Thailand (Nakorn Pathom, Saraburi, or Banpong) in the presence or absence of pCAGGS-3Cpro-FLAG. (B and C) Nucleocapsid sequence comparisons between PEDV_AVCT12 and field strains (B) or among classical PEDV strains (C). Black triangles mark the cleavage site. Residues at position 381 are highlighted in gray. The GenBank accession numbers for each strain are as follows: LC053455 for AVCT12, KR061458 for ITA7239, JQ023162 for attDR13, AB618622 for 83P-5, JX560761 for SD-M, KR610991 for EAS1, EF185992 for LZC, AF353511 for CV777, and GU937797 for SM98.

We expanded the search for the 3Cpro cleavage site to a wider array of PEDV strains. We sampled a total of 100 representative strains from both classical (pre-2010) and emerging (post-2010) strains, as categorized recently (4). We discovered that residue 381 is a conserved proline in all of the emerging PEDV strains in both North America and Asia. Interestingly, the classical PEDV strains were split into two groups regarding the identity of residue 381 (Fig. 6C). Historical CV777, SM98, and Chinese LZC strains, together with a CV777-like strain isolated in 2014 in Thailand (EAS1), harbored a leucine at this position, while some vaccine strains, such as attDR13 and 83P-5 (P100), and field isolates SD-M and ITA7239 possessed a proline (Fig. 6C). Some of the strains that circulated in Fujian, China, in 2012 to 2014 also carried a leucine at position 381 (Y. S. Lin, L. B. Wang, L. J. Zhou, and C. Y. Wang, unpublished data). These observations suggest that L381 alone might not serve as a cell adaptation marker, as some tissue culture-adapted PEDV strains also carried P381. However, the obvious shift to proline at this position in emerging strains might bear some physiological significance in terms of PEDV evolution during field circulation.

Cleavable PEDV N confers a slight growth advantage to PEDV in Vero E6-APN cells.

To pinpoint the biological significance of the cleavage event for PEDV replication in tissue culture, we rescued PEDV_AVCT12 harboring the cleavage site mutation in its N gene. To easily monitor viral growth, we utilized the infectious clone backbone harboring the mCherry coding sequence in place of ORF3 and efficiently expressing mCherry as a marker of viral growth (37). Reverse genetics (rg) was performed by the transfection of pSMARTBAC-mCh-PEDV_{ACVT12-N-L381P} into HEK 293T cells. Viral particles in the supernatant were further propagated and titrated in Vero E6-APN cells as previously described (37). The mutation was confirmed by sequencing of the viral RNA extracted from rgPEDV_{ACVT12-N-L381P} (passage 3).

Vero E6-APN cells were infected with mCherry-expressing wild-type or mutant PEDV viruses (mCh-PEDV; multiplicity of infection [MOI] of 0.001), and samples of the supernatant were taken at various time points postinfection for the determination of viral replication kinetics by RT-quantitative PCR (RT-qPCR) with primers specific for the PEDV M gene (37). A difference in growth kinetics was observed for the two variants (Fig. 7A). Especially at the early time point (24 h postinfection [hpi]), mCh-PEDV_{ACVT12-N-L381P} yielded substantially lower virus titers than did mCh-PEDV_ACVT12 (Fig. 7A). The difference was reduced at later time points but was still notable.

FIG 7.

PEDV_{AVCT12-N-L381P} shows a slight growth defect in Vero E6-APN cells. (A) Growth kinetics of mCh-PEDV_AVCT12 carrying wild-type or L381P mutant N as determined by RT-qPCR. D.L. denotes the detection limit of the RT-qPCR assay. GE, genome equivalents. Error bars indicate SD from duplicate infections. The bar graph is representative of results from two independent experiments. (B) Fluorescent images of Vero E6-APN cells infected with mCh-PEDV_AVCT12 carrying wild-type or L381P mutant N at 48 hpi. Bars, 100 μm. Data in the bar graph represent average diameters measured for five plaques under each condition. Error bars represent SD. (C and D) Western blot analysis of Vero E6-APN cells infected with mCh-PEDV_AVCT12 carrying wild-type or L381P mutant N at an MOI of 0.001 (C) or 0.01 (D). Blots are representative of results from two independent experiments.

For each variant, an identical infection (MOI = 0.001) was performed in two separate wells in a 6-well plate to monitor viral protein production and plaque formation. In wells for monitoring plaque formation, viral adsorption medium was replaced by infection medium containing trypsin and 0.9% agar. At 48 hpi, plaques were imaged under a fluorescence microscope. We noticed that the sizes of the plaques produced by wild-type and mutant viruses were different (Fig. 7B). On average, wild-type mCh-PEDV_ACVT12 produced larger and more uniformly round plaques than those produced by the mCh-PEDV_{ACVT12-N-L381P} mutant (Fig. 7B). In other wells for monitoring viral protein production, viral adsorption medium was replaced by infection medium containing trypsin. Infected cells were lysed at 48 hpi and analyzed by Western blotting for cleavage of PEDV N. As expected, the lysates prepared from cells infected with wild-type mCh-PEDV_ACVT12 showed a distinct cleavage product of N, while those prepared from cells infected with the mCh-PEDV_{ACVT12-N-L381P} mutant displayed a single band corresponding to full-length N (Fig. 7C). Interestingly, cells infected with the mCh-PEDV_{ACVT12-N-L381P} mutant showed significantly fewer cytopathic effects and produced less N protein, as detected by Western blotting (Fig. 7C). When supernatants were probed for the presence of N in viral particles, significantly less N protein was observed in the mutant virus, consistent with the reduced numbers of viral progenies observed by using the RT-qPCR method (Fig. 7C). To address the possibility that the level of accumulation of the N-L381P mutant might not be high enough to trigger cleavage, we performed infection at a higher starting MOI (0.01) to increase the extent of infection and examined N cleavage in the infected-cell lysates at different time points postinfection. While clearly visible during infection with wild-type PEDV, the ∼43-kDa cleavage product released by 3Cpro was never observed during infection with the mCh-PEDV_{ACVT12-N-L381P} mutant, even at late time points (Fig. 7D). Taken together, these results suggest that processing of PEDV N by 3Cpro may positively contribute to the replication of PEDV_ACVT12 in the tissue culture system.

DISCUSSION

With the tremendous economic consequences of global PEDV outbreaks, there has been a large collective effort in research on PEDV. To mount a proper and effective response through vaccination or antiviral therapy, we must understand the pathogenesis of PEDV through the characterization of PEDV proteins to fully appreciate the roles that they play during the course of PEDV infection. In this study, we described cleavage of PEDV N by 3Cpro and showed that cleavage occurs in a strain-specific manner and confers a replication advantage in cell culture over the uncleaved PEDV variant.

Previous studies have shown that N proteins of other coronaviruses are also similarly processed upon infection. For example, TGEV N is a substrate of caspase-6 and -7 upon apoptosis induced by infection; about 5 kDa is cleaved from the C-terminal side of the N protein (27). SARS-CoV N is also cleaved by caspase-6 and/or caspase-3 but only in a cell type-specific manner and in association with virulence. SARS-CoV infection of Vero E6 cells yielded clear caspase cleavage of N, high viral titers, and severe cytopathic effects, whereas that of Caco-2 cells produced undetectable N cleavage, low viral titers, and moderate cytopathic effects (28). Furthermore, smaller N-derived polypeptides with a molecular mass that is 2 to 5 kDa lower than the expected molecular mass have been reported during infection of murine (MHV), feline (FIPV), bovine (BCV), and avian (IBV and TCV) coronaviruses, but the proteases responsible for the processing of these N proteins have not been identified (15). Based on these observations, it has been proposed that CoV N processing by caspases might be one of the common mechanisms employed by hosts to counteract virus infection (27). Interestingly, we showed in this study that caspases were not responsible for the observed PEDV N processing, as pancaspase inhibitors had no effect on N cleavage during PEDV infection. In fact, slightly more robust N processing and PEDV infection were observed in the presence of pancaspase inhibitors. In line with our findings, a recent study showed that PEDV infection was unperturbed by pancaspase inhibitors and further demonstrated that, unlike other coronaviruses, PEDV-induced apoptosis occurred via a caspase-independent pathway and did not activate the main caspase, caspase-3, and the following cascade (38). While the mechanism mediating PEDV N degradation during late infection remains to be identified, we characterized the caspase-independent release of the ∼43-kDa major cleavage product during infection by PEDV_AVCT12. With further analyses of N sequences and biochemical studies, our results clearly indicate that this cleavage product was achieved through a distinct mechanism dependent on the function of virally encoded 3Cpro. To our knowledge, this finding is the first to illustrate the role of 3Cpro in the modification of structural proteins during coronavirus replication.

As the main protease for coronaviruses, 3Cpro is critical for the cleavage of the C-terminal regions of the polyproteins ORF1a and ORF1ab, giving rise to mature nonstructural proteins and viral replicase complexes (12). In addition, PEDV 3Cpro was recently demonstrated to act as an IFN antagonist by cleaving the NF-κB essential modulator (NEMO) and thereby attenuating host antiviral immunity (39). Based on other well-studied coronaviruses, the cleavage motif recognized by CoV 3Cpro has been mapped out. The general substrate specificity of 3Cpro is determined primarily by the residues in the P1, P2, and P1′ sites. The P1 position contains a conserved glutamine (Q), while the P2 position usually contains a hydrophobic residue, preferably leucine (L), and the P1′ position contains a small aliphatic residue such as serine, glycine, or alanine (12, 29, 34, 35, 40). For the PEDV N cleavage site that we identified, the P1 and P2 positions strictly conform to the CoV 3Cpro substrate specificity. However, the P1′ position is occupied by arginine and not the usual small amino acids suggested by previous reports. However, if the P1′ position is occupied by glutamine, PEDV 3Cpro does not recognize this as its cleavage site (Fig. 4). A previous report also showed that the PEDV 3Cpro P1′ subsite on NEMO contains valine instead of small aliphatic residues (39). Analysis of the structure of PEDV 3Cpro reconciles this conflict by showing that the S1′ subsite could tolerate a larger hydrophobic P1′ residue, such as valine, in NEMO (34). Our results expand the pool of tolerable P1′ residues to include a large, positively charged side chain of arginine, suggesting an even greater degree of flexibility of the S1′ subsite of PEDV 3Cpro.

Although their main function is to organize the viral genomes, CoV N proteins are involved in a myriad of other functions (14). However, despite extensive findings that CoV N proteins are proteolytically processed, no definitive functions have been ascribed to the cleavage events or products. Consistent with what we observed here, it is generally accepted that only full-length N is incorporated into the virions (15, 27). Although the N-terminal RNA binding motifs are still preserved following C-terminal cleavage, oligomerization motifs in the C terminus might be impaired for cleaved N, rendering them unable to form the helical capsid viral RNP (vRNP) complexes. Additionally, the C termini of CoV N proteins have been shown to interact with M proteins (41, 42). The removal of this segment might also affect normal functions of PEDV N during virus assembly. Possibly, cleavage may serve as a strategy to regulate the pool of full-length N for balancing or optimizing virus production in host cells. Moreover, it could be speculated that the leftover proteins (on both the N- and C-terminal sides) could interact with partner proteins and modulate N′s functions. Alternatively, lessons from many viruses have demonstrated that cleavage of numerous viral proteins by caspase could lead to many possible outcomes. For some viruses, cleavage of viral proteins caused virus attenuation; for others, cleavage triggered viral amplification (43). These outcomes might be similarly applicable to the case of cleavage of viral proteins by a virally encoded protease.

Another intriguing aspect regarding N cleavage is its strain specificity. For SARS-CoV, caspase cleavage of N was associated with cell types, cytopathic effects, and viral titers (28). Influenza viruses also showed a strain specificity of nucleoprotein cleavage by caspases: nucleoproteins of avian influenza viruses contained no cleavage motif and were not cleaved by caspases (44). Among PEDV strains with sequence information in the NCBI database, the cleavage motif is found in certain classical PEDV strains that can be propagated to high titers in tissue culture and a few PEDV strains circulating in the Fujian area in 2012 to 2014, with no information on viral adaptability in cell culture. More importantly, our reverse-genetics system allows us to modify the cleavage motif of a cell-adapted PEDV and study its effect during PEDV growth in cell culture. PEDV bearing uncleaved N exhibited a noticeable growth disadvantage compared to the wild type, suggesting that N cleavage may play a role in supporting PEDV growth in cell culture. However, whether this is an isolated event for PEDV_AVCT12 or a more general phenomenon for other PEDV strains remains to be determined.

In summary, we demonstrated, for the first time, the modification of a PEDV structural protein by a virally encoded main protease, 3Cpro, and identified the PEDV 3Cpro cleavage site on PEDV N. Remarkably, the cleavage motif is found only in PEDV strains that are highly adapted in cell culture, and an uncleavable mutant affected the growth of the virus. Emerging field strains do not possess the cleavage motif in their N genes and are difficult to propagate in laboratory settings. Although N cleavage is not the sole marker of PEDV adaptation to cell culture, it will be interesting to further investigate the ramifications of PEDV N cleavage for PEDV replication and cell adaptability.

MATERIALS AND METHODS

Biological materials.

Human embryonic kidney 293T (HEK 293T) and Vero E6-APN cells were maintained in Opti-MEM supplemented with 10% fetal bovine serum (FBS). pSMART-PEDV_AVCT12, reverse-genetics-derived PEDV_AVCT12, and mCh-PEDV_AVCT12 containing mCherry in place of ORF3 were described previously (37). pSMART-PEDV_{AVCT12-N-L381P} was assembled and mCh-PEDV_{AVCT12-N-L381P} was rescued and propagated as described previously (37). Z-VAD(OMe)-FMK (cell-permeable pancaspase inhibitor) was purchased from Abcam and dissolved in DMSO according to the manufacturer's recommendations. The following antibodies were obtained from commercial sources and used according to the manufacturers' recommendations: mouse anti-PEDV N (Medgene Labs), rabbit anti-FLAG (Cell Science Technology), rabbit anti-GFP (Santa Cruz Biotechnology), and horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG and goat anti-rabbit IgG (Santa Cruz Biotechnology).

Plasmid construction.

For PEDV N expression plasmids, the coding sequence for N was reverse transcribed and amplified from PEDV RNA derived from PEDV_AVCT12 (CV777-like) (37) or field isolates retrieved from affected farms in central Thailand. The resulting PCR products were assembled into the pHW2000-based vector. The plasmids were verified by sequencing. Point mutations at possible cleavage sites (position 396 or 382) were constructed by site-directed mutagenesis.

For pCAGGS-3Cpro-FLAG, the coding sequence for 3Cpro, based on that of PEDV_AVCT12, was synthesized and codon optimized for expression in human cells. The sequence was tagged with the FLAG epitope at the C terminus. The synthesized fragment was assembled into the pCAGGS vector, and the resulting plasmid was verified by sequencing. For functional substrates of PEDV 3Cpro, an 8-amino-acid stretch corresponding to the P6-P2′ (numbering around the cleavage site follows the conventions described in reference 45: P6-P5-P4-P3-P2-P1^P1′-P2′) cleavage site of PEDV 3Cpro derived from the junction of the nsp5 and nsp6 genes (YGVNLQ^GG) in the PEDV genome was implanted into the coding sequence of GFP between amino acids G190 and D191. This position is in the loop previously identified as being prone to proteolysis (46). The coding sequence of GFP with the inserted nsp5-nsp6 junction was assembled by two PCRs. The first PCR product (N-GFP) was amplified with a forward primer harboring the MluI site and a reverse primer harboring the N-terminal half of the cleavage site coding sequence (YGVNL) and the BsmBI restriction site. The second PCR product (C-GFP) was amplified with a forward primer harboring the BsmBI site complementary to that in the N-GFP fragment followed by the C-terminal half of the cleavage site coding sequence (LQGG) and the reverse primer containing the NotI restriction site. pCAGGS-GFP^5/6 was then assembled by three-way ligation with MluI/NotI-digested pCAGGS, the MluI/BsmBI-digested N-GFP PCR product, and the BsmBI/NotI-digested C-GFP PCR product. The final plasmid was verified by sequencing. Based on the backbone of pCAGGS-GFP^5/6, its derivatives containing different possible 8-amino-acid cleavage sites from PEDV N were constructed by using the same three-way ligation method. These plasmids are named pCAGGS-GFP^AVCT12-396 (RETTLQ^{^}QH), pCAGGS-GFP^AVCT12-382 (GNAKLQ^{^}RK), and pCAGGS-GFP^KR-382 (GNAKPQ^{^}RK).

Cleavage visualization.

For infection experiments, Vero E6-APN cells were plated to confluence in 6-well plates in Opti-MEM with 10% FBS and infected with variants of reverse-genetics-derived PEDV_AVCT12 at the indicated MOIs for 1 h. After washing with phosphate-buffered saline (PBS), 2 ml of FBS-free Opti-MEM supplemented with trypsin (2 μg/ml) was added. Infected cells were monitored daily for signs of syncytium formation as a CPE and harvested when a >80% CPE was observed or at the time points indicated for each experiment. For transfection experiments, HEK 293T cells were plated to confluence in 6-well plates in Opti-MEM supplemented with 10% FBS and transfected with 1 μg each of the indicated plasmids by using FuGene HD (Promega). At 48 h posttransfection (hpt), cells were lysed on ice for 10 min with 150 μl of lysis buffer (25 mM Tris-HCl [pH 7.4], 150 mM NaCl, and 1% Triton X-100 supplemented with a protease inhibitor cocktail). Clarified lysates were analyzed by Western blotting using the indicated antibodies.

Plaque size determination.

Vero E6-APN cells were plated to confluence in 6-well plates and infected with mCh-PEDV_AVCT12 or mCh-PEDV_{AVCT12-N-L381P} for 1 h at 37°C (MOI = 0.001). After PBS washes, the infected cells were covered with Opti-MEM in 0.9% agar supplemented with a 1× final concentration of TrypLE select (Thermo). Plaques were imaged at 48 hpi, and diameters were measured with CellSens software measurement tools (Olympus Life Science). Numbers shown are averages of data for five representative plaques. Error bars represent standard deviations (SD).

One-step RT-qPCR.

Vero E6-APN cells were plated to confluence in T-25 flasks and infected with either mCh-PEDV_AVCT12 or mCh-PEDV_{AVCT12-N-L381P} for 1 h at 37°C (MOI = 0.001). After rigorous washing with PBS, the infected cells were covered with infection medium containing a 1× final concentration of TrypLE Select (Thermo). At 0, 24, 48, and 72 hpi, supernatants were collected and stored at −80°C. For quantification of viral RNA, one-step RT-qPCR was performed as described previously (37). Briefly, vRNA was extracted from 200-μl supernatant samples with a viral nucleic acid extraction kit (GeneAid). In each 20-μl reaction mixture, 5 μl vRNA was mixed with primers specific for the PEDV M gene and iTaq Universal SYBR green mix (Bio-Rad). One-step RT-qPCR was performed in a CFX96 thermal cycler under the following conditions: 50°C for 30 min, 95°C for 1 min, and 40 cycles of 95°C for 15 s and 60°C for 60 s. Data were analyzed with CFX Manager software. Conversion to copy numbers (in units of log genome equivalents per milliliter of supernatant) was performed with reference to a standard curve generated from a serially diluted plasmid containing the PEDV M gene (47).