Porcine deltacoronavirus enters cells via two pathways: A protease-mediated one at the cell surface and another facilitated by cathepsins in the endosome

Porcine deltacoronavirus (PDCoV) is a pathogen belonging to the genus Deltacoronavirus that in 2014 caused outbreaks of piglet diarrhea in the United States. To identify suitable therapeutic targets, a more comprehensive understanding of the viral entry pathway is required, particularly of the role of proteases. Here, we identified the proteases that activate the viral spike (S) glycoprotein to initiate cell entry and also pinpointed the host-cellular pathways that PDCoV uses for entry. Our results revealed that cathepsin L (CTSL) and cathepsin B (CTSB) in lysosomes and extracellular trypsin in cell cultures independently activate the S protein for membrane fusion. Pretreating the cells with the lysosomal acidification inhibitor bafilomycin-A1 (Baf-A1) completely inhibited PDCoV entry, and siRNA-mediated ablation of CTSL or CTSB expression significantly reduced viral infection, indicating that PDCoV uses an endosomal pathway for entry. Of note, trypsin treatment of cell cultures also activated PDCoV entry, even when the endosomal pathway was inhibited. This observation indicated that trypsin-induced S protein cleavage and activation in cell cultures enables viral entry directly from the cell surface. Our results provide critical insights into the PDCoV infection mechanism, uncovering two distinct viral entry pathways: one through cathepsin L and cathepsin B in the endosome and another via a protease at the cell surface. Because PDCoV infection sites represent a proteases-rich environment, these findings suggest that endosome inhibitor treatment alone is insufficient to block PDCoV entry into intestinal epithelial cells in vivo. Therefore, approaches that inhibit viral entry from the cell membrane should also be considered.

The causative agent of porcine diarrhea that first caused outbreaks in 2014 in the United States is a newly identified porcine deltacoronavirus (PDCoV), 3 which causes severe diarrhea and vomiting followed by high morbidity and mortality in piglets (1)(2)(3)(4)(5). The virus belongs to the genus Deltacoronavirus, and family Coronaviridae (6). Like porcine epidemic diarrhea virus (PEDV) and transmissible gastroenteritis virus (TGEV) in the genus Alphacoronavirus, the jejunum and ileum are the primary sites of PDCoV replication (3). PDCoV was first reported in Hong Kong in 2012 (6) and has been subsequently reported in the United States (4), Canada (7), South Korea (8), China (9), Thailand (10), and Vietnam (11).
PDCoV is an enveloped virus that expresses spike (S) proteins with a high molecular weight. The coronavirus S protein has been recognized as a type I transmembrane glycoprotein with heavy glycosylation (12,13). The cryo-EM structure of the PDCoV S protein has been determined, and indicates that the structures of the S1 domain are more similar to those of alphacoronaviruses (e.g. PEDV and TGEV) than those of betacoronaviruses (e.g. severe acute respiratory syndrome coronavirus (SARS-CoV), mouse hepatitis virus, and Middle East respiratory syndrome coronavirus (MERS-CoV)) (14). In addition, the S protein of mouse hepatitis virus or MERS-CoV is often posttranslationally cleaved into S1 and S2 by endogenous cellular proteases (e.g. furin) (15,16). A number of features of the PEDV or TGEV S proteins in alphacoronaviruses and PDCoV in deltacoronavirus are conserved; however, they have a low amino acid identity, similar to the S protein between SARS-CoV and those of the other coronaviruses (17). The protein covers the virion surface in a trimeric, integral, and uncleaved format (13). The S trimer is generated in a locked conformation to prevent proteolytic activation triggering membrane fusion (18,19), which is similar to studies of other coronaviruses (e.g. SARS-CoV (20,21), PEDV (18), TGEV (22), and human coronavirus 229E (23,24)). The S1 subunit contains receptor-binding sites, which are responsible for the recognition and binding of its cellular receptor (14,(25)(26)(27). After binding to the receptor, conformational changes occur between S1 and S2, which expose the cleavage site to proteases (28). The spike protein is separated into a surface unit, S1, and a transmembrane unit, S2 after cleavage by protease. The cleavage of S protein is the key step for the membrane fusion. The cleaved S2 subunit contains an N-terminal fusion peptide, which can be inserted into the cell membrane and induce virus-cell membrane fusion, leading to viral entry (29,30).
Host proteases play a crucial role in virus infection and the different proteases used by viruses determine the virus entry pathway to some extent. Four proteases participate in the process of viral infection: 1) membrane-binding proteases, like transmembrane serine protease, which appear to mediate viral entry following virus attachment to cell receptors (31,32); 2) lysosomal proteases, cathepsin L or cathepsin B activated virus entry after virus endocytosis in virus-targeted cells; 3) extracellular proteases (e.g. intestinal proteases), which are essential for PEDV entry (33); and 4) proprotein convertases (e.g. furin). The S protein is cleaved by furin after production in virus-infected cells.
Although the mechanism of PDCoV entry remains unclear, the functional virus receptor (porcine aminopeptidase N) has been identified as an important factor critical for PDCoV entry into cells (34,35). However, another crucial factor required for viral entry, in which proteases function as activators of the viral S glycoprotein to activate cell entry, has not been determined. Moreover, which pathway was used by PDCoV for its entry remains unknown. In the present study, we examined the pathways used by PDCoV for cell entry, and the data suggest that cathepsins (i.e. CTSL or CTSB) activate the S protein for fusion activity. The results indicate that PDCoV used an endosomal pathway for its entry and cell infection. Moreover, the function of trypsin in virus infection was also evaluated. We found that trypsin was not necessary for the continuous passage of PDCoV in ST cells, which differed from PEDV. The propagation of PEDV in cell culture requires exogenous trypsin (18,33). Therefore, whether trypsin plays a role in the entry of PDCoV remains unknown. Our results suggest that treatment with trypsin allowed PDCoV to bypass the endosomal pathway and directly enter the cells from the cell surface. Together, these results suggest that PDCoV used two pathways for its entry. Furthermore, cathepsins (e.g. CTSL and CTSB) and trypsin activated the S protein for viral entry in the endosomal and cell surface pathways, respectively.

PDCoV uses two pathways for its entry
A previous study confirmed that MERS-CoV and SARS-CoV infected cells through two different routes: 1) the endosomal pathway relying on a low pH environment and lysosomal cysteine proteases or 2) directly from the cell surface via fusion of the virus envelope and the cell membrane when the S protein binds to receptors on the cell surface (36,37). To determine the role of lysosomal cysteine proteases in PDCoV entry, various inhibitors, including bafilomycin A1 (Baf-A1, lysosomal acidification inhibitor), E64d (lysosomal cysteine protease inhibitor), CA-074 (cathepsin B-specific inhibitor), and Z-FY-CHO (CTSLI, cathepsin L-specific inhibitor) were used. The optimal inhibitor concentrations were determined in ST and IPI-2I cells using a CCK-8 -based cell viability assay (Fig. 1, A and B, respectively) and virus titer assay by measuring the TCID 50 (Fig. 1, C and D, respectively). The results indicate that the inhibitors exhibited cellular toxicity following an increase in the drug concentration. The optimal functional concentrations were 50 M for E64d, CA-074, and CTSLI and 50 nM for Baf-A1. Furthermore, when ST and IPI-2I cells were treated with inhibitors, PDCoV infection was dramatically inhibited. The results indicate that PDCoV cell entry was activated by lysosomal proteases.
To determine whether PDCoV directly enters from the cell surface, IPI-2I cells were treated with inhibitors (50 M for E64d, CA-074, CTSLI and 50 nM for Baf-A1) for 1 h at 37°C to inhibit the endosomal pathway and infected with PDCoV at an m.o.i. of 1 for 30 min at 4°C (absorption to the cell surface did not allow viral entry into the cells). The cells were subsequently treated with various concentrations of trypsin at room temperature for 10 min and cultured with inhibitors for 6 h. The entry of PDCoV was detected with RT-qPCR for sgNS7a. As shown in Fig. 1E, viral entry was dramatically inhibited following treatment with inhibitors compared with the DMSO-treated group, whereas treatment with 100 g/ml trypsin extensively facilitated viral entry, showing no differences compared with the mock-treated group. Virus entry in the presence of cathepsin L or cathepsin B inhibitors was also detected using IFA (Fig. 1F) and Western blotting (Fig. 1G). The results suggest that treatment with trypsin could reverse the inhibitory effect of the inhibitors and allow PDCoV entry to bypass the endosomal pathway.
A previous study confirmed that the peptides derived from the heptad repeat (HRP) region of the SARS-CoV S protein can strongly inhibit virus entry from the cell surface induced by proteases but has little effect on the endosomal pathway (38,39), which has also been found in PEDV as previously described (40). Therefore, to confirm that PDCoV can directly enter cells from the cell surface, the HR2 peptides (GIYNNTILNLT-VEINDLQERSKNLSQIADRLQNYIDNLNNTLVDLEWL) of the PDCoV S protein were synthesized. IPI-2I cells were treated with 50 nM Baf-A1 for 1 h at 37°C, infected with PDCoV for 30 min at 4°C, and treated with 100 g/ml trypsin in DMEM containing various concentrations of HR2 peptides. Viral entry was detected using RT-qPCR for sgNS7a at 6 h post-infection. As shown in Fig. 1H, 50 nM Baf-A1 dramatically inhibited virus entry via the endosomal pathway, whereas trypsin treatment reversed this effect, as in the above-mentioned results. However, when the endosomal pathway was inhibited, 10 M HR2 peptide completely inhibited viral entry, even if trypsin was added. These results indicate that PDCoV bypassed the endosomal pathway and entered the cells directly from the cell surface following trypsin treatment.

Cathepsin L and cathepsin B activate PDCoV entry from the endosome pathway
To determine whether cathepsin L or cathepsin B activated the entry of PDCoV by cleaving the S protein, we expressed the S protein in 293T cells. An enzyme cleavage assay was performed with recombinant cathepsin L and cathepsin B in Dul-

Cell entry of porcine deltacoronavirus
becco's phosphate-buffered saline (pH 5.6), respectively. The cleavage of the S protein by these two proteases was detected by Western blotting. As shown in Fig. 2, A and B, the extracellular domain of the S protein was ϳ180 kDa, as revealed by an anti-FLAG mAb reacting with the C-terminal fraction of the S protein. Cathepsin L cleaved the S protein into two S2 bands of ϳ50 kDa, whereas only one S2 band (ϳ45 kDa) was detected following cathepsin B cleavage.
To confirm that the cathepsins (CTSL and CTSB) activated PDCoV entry, specific siRNAs were designed and transfected into IPI-2I cells to knock down endogenous CTSL or CTSB expression. Western blotting results imply that endogenous CTSL or CTSB expression was successfully knocked down ( Fig.  2, C and D). As shown in Fig. 2E, viral entry was inhibited by ϳ10-fold lower compared with that of the mock-treated group. The virus yield was also dramatically decreased, as shown by Cell entry of porcine deltacoronavirus measuring the TCID 50 (Fig. 2F). These results also suggest that CTSL or CTSB activated the S protein of PDCoV, which facilitated viral infection from the endosomal pathway.

PDCoV infection up-regulates the expression of CTSL and CTSB
To further understand the relationship between cathepsins (CTSL and CTSB) and PDCoV infection, we examined the changes in protein expression and enzyme activity following viral infection. As determined by RT-qPCR, the levels of CTSL or CTSB mRNA were significantly up-regulated in PDCoVinfected IPI-2I cells at 15 h post-infection (Fig. 3A). Moreover, with regard to the protein levels, virus infection increased the expression of CTSL and CTSB (Fig. 3B). We also examined the changes in cathepsin enzyme activity following PDCoV infec-

Cell entry of porcine deltacoronavirus
tion, as described in Fig. 3, C and D. Although viral infection altered the enzyme activity of CTSB, it had no effect on CTSL enzyme activity (data not shown). To examine the changes of CTSL and CTSB in pigs following PDCoV infection in vivo, intestinal tissues from five specific pathogen-free pigs (three infected pigs and two control pigs) used to study the pathogenicity of PDCoV were prepared and subjected to Western blot analyses. As shown in Fig. 3E, PDCoV infection promoted CTSL and CTSB expression in the pig intestinal tissues. The intensity band ratio of CTSL/GAPDH and CTSB/GAPDH is shown in Fig. 3, F and G. To investigate whether these changes contributed to PDCoV infection, we overexpressed CTSL and CTSB in IPI-2I cells with the pLVX-IRES-EGFP plasmid, a lentiviral vector expressing EGFP proteins. As observed in Fig. 3H, more than 90% of the cells displayed green fluorescence, suggesting that the CTSL or CTSB genes had been successfully transduced into the ST cells with high efficiency. Western blot analyses also indicated that CTSL and CTSB, respectively, were overexpressed in IPI-2I cells (Fig. 3I). The relative efficiency of PDCoV entry into CTSL-or CTSB-overexpressing IPI-2I cells was ϳ2-fold higher than that of the vector-only transduced cells (Fig. 3J). We also examined the viral yield in the overexpressed cells. The results indicate that the viral yield was higher in CTSL (6.92 Ϯ 0.38 log 10 TCID 50 /ml) or CTSB (6.93 Ϯ 0.28 log 10 TCID 50 /ml)-overexpressing IPI-2I cells than in the mock cells (5.77 Ϯ 0.5 log 10 TCID 50 /ml) (Fig. 3K). These results indicate that PDCoV infection could up-regulate the expression of CTSL and CTSB both in vitro and in vivo, and these changes promoted viral infection.

Entry of PDCoV from the cell surface is facilitated by trypsin
To compare the efficiency of the two pathways or cell types with regard to virus entry, the infection kinetics of PDCoV in cells pretreated with Baf-A1 and subsequently treated with Intestinal tissues from five specific pathogen-free pigs (three infected pigs and two control pigs) were prepared and subjected to Western blot analysis. F and G, the intensity band ratios of CTSL/GAPDH (F) and CTSB/GAPDH (G) were calculated and compared. The overexpression of CTSL or CTSB facilitates PDCoV entry. H, CTSL or CTSB was overexpressed in IPI-2I cells. IPI-2I cells were transfected with CTSL-or CTSB-expressing plasmids and the transfection efficiency was detected by EGFP expression. I, the protein expression was detected by Western blotting. J, overexpression of CTSL or CTSB promoted viral entry. K, the virus yield as determined by measuring the TCID 50 at 36 h post-infection. Data are expressed as the mean Ϯ S.D. for triplicate samples. Level of significance was determined by Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001.

Cell entry of porcine deltacoronavirus
100 g/ml trypsin were conducted, in which the virus only enters via the cell surface pathway. The ST and IPI-2I cells treated with trypsin (viral entry from the cell surface) or untreated (viral entry only via the endosome pathway) were used as controls. The time of viral entry from the cell surface was ϳ1 h earlier via only the endosomal pathway in ST and IPI-2I cells (Fig. 4, A and B, respectively). The results also indicated that viral entry at any given time was always about 10 -100-fold higher in the trypsintreated cells than in the untreated cells. The viral growth curve was also determined to assess the efficiency of virus spread via the two entry pathways in the presence of trypsin or Baf-A1. The virus titer could be determined in the trypsin or mocktreated cells at 12 h post-infection, whereas the viral dose in the cell cultures containing trypsin and Baf-A1 could be examined at 18 h post-infection. Moreover, the viral titer in the cultures treated with only trypsin was 10-to 100-fold higher than that in the ST cells treated with trypsin and Baf-A1 (Fig. 4C). In IPI-2I cells, the effect of trypsin on PDCoV infection was more obvious than that in the ST cells, especially 24 h post-infection, and the virus yield in trypsin-treated IPI-2I cells (6.5 log 10 TCID 50 / ml) was ϳ10,000-fold higher than that in the untreated IPI-2I cells (2.8 log 10 TCID 50 /ml) (Fig. 4D). We also assessed the function of trypsin in the presence of other inhibitors by determining the viral titer. As shown in Fig. 4, E and F, trypsin treatment increased the viral load by about 100-to 1000-fold when compared with that of the untreated cells, and could reverse the

Activation of cell fusion and S1-S2 cleavage by trypsin treatment
ST cells were infected with PDCoV at an m.o.i. of 1, and the infected cells were treated with 20 g/ml trypsin or left untreated at room temperature for 20 min. The uninfected cells were used as a control. Cell fusion was detected by IFA and the CPE was observed. The results indicated that PDCoV infection with trypsin treatment induced cell fusion, and the cells displayed several syncytia and obvious CPE. In contrast, no syncytia were observed in the control or untreated ST cells (Fig. 5A). To confirm that cell fusion was induced by the PDCoV S protein after trypsin treatment, the plasmid encoding the S gene (pAAV-PDCoV opti-S) was transfected into ST cells, which were treated with trypsin as described above. At 48 h post-transfection, large cell fusion was observed with trypsin treatment (Fig. 5A), whereas the untreated or mock cells remained largely mononucleated. However, no obvious cell fusion was detected in the PDCoV-infected or transfected IPI-2I cells with the pAAV-PDCoV opti-S plasmid, although the IPI-2I cells were treated with trypsin. The number of nuclei in the syncytia was counted to semi-quantify the extent of cell fusion (Fig. 5B). To further verify the role of trypsin in PDCoV infection, the SBTI was added to the cell culture in the presence of Baf-A1 to block trypsin activity and as shown in Fig. 5C, virus infection was significantly inhibited compared with the trypsin treatment group in the presence of Baf-A1 (Fig. 5C). Large syncytial foci were observed after trypsin treatment in the presence of Baf-A1 (Fig. 5C, white arrow), whereas no syncytium was observed after SBTI treatment, even though in the presence of trypsin. To provide direct biochemical evidence that trypsin cleaved PDCoV S, ST cells were infected with PDCoV to

Cell entry of porcine deltacoronavirus
express S protein in the cells. The cells were collected and lysed by sonication and then incubated with various concentrations of trypsin followed by Western blot analysis with S polyclonal antibody. As shown in Fig. 5D, two extra bands (150 and 50 kDa) were detected after trypsin treatment. The results suggested that trypsin could activate PDCoV spike protein to induce cell to cell fusion.

Arg-672 in the S protein is critical for trypsin-induced cell fusion
The structure of the PDCoV S protein was determined by cryo-EM (PDB ID: 6B7N). The monomer structure of the S protein is shown in Fig. 6A. A loop structure was observed between the S1 and S2 domains which contains a variety of proteases cleavage sites (e.g. furin protease (16,41,42), trypsin (12), and cathepsin L (44)). Whether the loop structure of the PDCoV S protein could be cleaved by proteases remains unknown. A multiple sequence alignment of the S gene was conducted among coronaviruses, which indicated that the S2Ј (Arg-672 in PDCoV S protein) cleavage site was conserved (Fig.  6B). This amino acid site was confirmed to be cleaved by trypsin in SARS-CoV and PEDV. We also conducted a cell to cell fusion assay to analyze the effect of Arg-672 in the S-meditated cell fusion induced by trypsin. The results indicate that the extent of cell fusion was dramatically decreased with the mutated S compared with the WT S protein (Fig. 6C). The results were confirmed by S-or S RN -meditated cell fusion, as shown in Fig. 6D; a large amount of cell fusion was detected in the S-transfected ST cells when the cells were treated with trypsin, whereas little or no obvious cell fusion was detected in S RN -transfected ST cells. To determine whether this position (Arg-672) was cleaved by trypsin, we mutated residue Arg-672 to N (S RN ) and no detectable cleaved S2 bands were observed (Fig. 6F) with S polyclonal antibody. The cleavage of WT S protein was used as a control and S2 bands could be detected, as shown in Fig. 6E.

Discussion
The S protein of coronavirus is a type I membrane glycoprotein with high molecular weight which determines the viral tropism, host range, and pathogenicity by receptor specificity and activation meditated by host proteases during virus entry into cells (13,16,27). Because of the essential role of S cleavage for coronavirus infection, targeting of relevant host cell proteases may be a promising therapeutic strategy for coronavirus infection (41,45,47). However, the viral entry pathway of PDCoV, particularly of the role of proteases in viral entry has not been determined. In this study, we found that PDCoV used two pathways (CTSL and CTSB in the endosome and trypsin at the cell surface) for its entry.
The roles of CTSL and CTSB have been extensively studied in virus entry, such as SARS-CoV, MERS-CoV, and PEDV (48 -50). Here, we found that purified CTSL and CTSB can also cleave S proteins into a fusion-ready formation for membrane fusion under a low pH environment. Furthermore, PDCoV infection could increase CTSL and CTSB expression in vivo and in vitro, and the enzyme activity of CTSB was increased following PDCoV infection. These findings are consistent with those of a previous study showing that porcine reproductive and res-piratory syndrome virus (PRRSV) infection up-regulates CTSL expression, but PDCoV infection increased the proteolytic activity of intracellular CTSB instead of the CTSL following PRRSV infection (51). The mechanisms of the up-regulation of CTSL and CTSB to promote PDCoV infection should be further investigated. The enzyme activity was increased by PDCoV infection, and this may represent the best method of efficiently activating viral entry.
PDCoV is enteropathogenic and the primary sites of PDCoV replication are the jejunum and ileum (3). The jejunum and ileum belong to a portion of the small intestine that contains several digestive enzymes (53,54). Therefore, the small intestine fluid, including trypsin or other proteases, may contribute to viral infection during the viral entry step (18). In fact, the findings of a previous study suggest that the isolation or repeated passage of PEDV and PDCoV in vitro, and the addition of trypsin or small intestinal contents is necessary during cell culture (33,55). However, whether trypsin plays an important role in virus entry has not been determined. In this study, we found that trypsin activated the membrane fusion induced by the PDCoV S protein in ST cells. This indicated that treatment with trypsin meditated the cell-membrane fusion induced by the S protein in ST cells but not in IPI-2I cells. Moreover, it may meditate fusion of the viral envelope to the cell membrane, which contributes to viral infection. Therefore, we formulated the hypothesis that PDCoV could bypass the endosomal pathway and directly enter cells from the cell surface. The addition of trypsin rendered viral entry by bypassing the endosomal pathway with high efficiency. To further verify this assumption, we compounded the HR2 regions of the PDCoV S protein and found that this peptide could efficiently inhibit viral entry from the cell surface but had little effect on entry from the endosomal pathway as described in a previous study of SARS-CoV (38). Moreover, PDCoV entry was significantly inhibited by the HR2 peptide even though the cells were treated with trypsin. These results confirm the hypothesis that PDCoV could enter the cells via a trypsin-mediated cell surface pathway. We compared the entry efficiency of PDCoV from 2 to 6 h post-infection. The direct entry of PDCoV from the cell surface meditated by trypsin was considered the most effective method of viral infection, and was about 1 h in advance of that via the endosomal pathway. Abundant proteases in the environment of the small intestine provide powerful cleavage for the S proteins of PDCoV, and entry from the cell surface is the most appropriate choice for virus infection. Considering several Toll-like receptors locate in intracellular compartments such as endosomes, bypassing the endosomal pathway for cell entry may minimally induce the body's immune response, which may also be a method of viral immune escape (56, 57).
As described above, PDCoV entry from the cell surface directly meditated by proteases in small intestine is the most efficient method of viral infection. Thus, the question remains as to why the virus retains an inefficient method (endosomal pathway) of entry. Recent reports have shown that an alternative pathway for PEDV infection exists from the nasal cavity to the intestinal mucosa in swine (58). This study provides evidence of airborne transmission for a gastrointestinal coronavirus in which entry via the endosomal pathway may be an impor-

Cell entry of porcine deltacoronavirus
tant route for viral infection from nasal epithelial cells. A study of PDCoV pathogenicity also found that infection could occur indirectly by contact in piglets. The control gnotobiotic piglets were infected by indirect contact and displayed high levels of viral RNA similar to those of the infected piglets (59). We have sufficient reason to believe that PDCoV may use a pathway similar to PEDV for its infection, although the molecular mechanisms remain to be elucidated.
In summary, our study found that PDCoV uses two pathways for cell entry. Moreover, PDCoV entry from the cell surface was found to be more efficient than that from the endosome. Therefore, treatment with trypsin dramatically increased the virus yield, which may explain why the addition of trypsin contributes to viral isolation. These results suggest that treatment with inhibitors of the cell entry step may be insufficient for blocking only PDCoV entry into intestinal epithelial cells in vivo from endosomes, given that PDCoV can directly enter cells from the cell surface following protease treatment. Therefore, targets designed to inhibit viral entry from the cell membrane also should be considered.

Plasmids
The codon-optimized PDCoV S gene (PDCoV strain NH, GenBank TM : KU981062.1) was cloned into the pCAGGS and pAAV-IRES-hrGFP vector. The constructs had a FLAG tag at the C terminus and were designated pCAGGS-PDCoV opti-S and pAAV-PDCoV opti-S. Two pairs of specific primers were used for PCR amplification of the gene Sus scrofa cathepsin L (CTSL) (NCBI Reference Sequence: NM_213892.1) and Sus scrofa cathepsin B (CTSB) (NCBI Reference Sequence: NM_001097458.1) from the cDNA of ST cells, respectively. The PCR products were each cloned into the pLVX-IRES-Zs-Green1 vector with an HA tag at the C terminus to generate the plasmids pLVX-CTSL/pLVX-CTSB. Luciferase plasmids (pcDNA3.1-T7 and pET-32a-IRES-luc) were constructed as described previously to detect cell-fusion (60). The pCDNA3.1-T7 vector, which encodes the T7 polymerase, was PCR amplified from Escherichia coli BL21 (DE3) competent cells. The pET-32a-IRES-luc expressing luciferase was controlled by a T7 promoter, in which the internal ribosome entry site (IRES) sequence and the luciferase sequence were cloned into the BamH I/EcoR I and EcoR I/NotI sites of the pET-32a vector, respectively.

RT-qPCR for sgNS7a and calibration line
The NS7a subgenome (sgNS7a) was amplified using RT-qPCR to quantify the efficiency of viral entry (61). This method was also used with other coronaviruses (e.g. SARS-CoV and MERS-CoV) as described previously (36,37). Detection of the level of CTSL or CTSB mRNA in IPI-2I cells or pig small intestinal tissue was performed by SYBR Green-based RT-qPCR with GAPDH as the internal control. Relative quantification was performed using the cycle threshold (⌬⌬CT) method. The One Step PrimeScript™ RT-PCR Kit (Perfect Real Time) or One Step SYBR ® PrimeScript™ RT-PCR Kit II (Perfect Real Time) was used to perform RT-qPCR analysis according to the manufacturer's instructions under the following conditions: reverse transcription (42°C for 10 min; 95°C for 10 s), one cycle; PCR (95°C for 5 s; 60°C for 20 s), and 45 cycles with a LightCycler 480 instrument (Roche Diagnostics). All of the primers and probes are listed in Table 1. To measure the level of viral entry, IPI-2I cells in 12-well culture plates were infected with 10-fold stepwise diluted PDCoV, from 10 2 to 10 6 , at a 50% tissue culture infectious dose (TCID 50 ). 6 h later, the total cellular RNA was isolated using 1 ml TRIzol ® reagent according to the established protocol and dissolved in 40 l DNase/RNase-free ddH 2 O. The amounts of sgNS7a were determined by RT-qPCR. The calibration line was established according to the cycle value and virus titer. The relationship between the viral titer (x axis) and realtime PCR cycles (y axis) was demonstrated to reach a positive correlation (data not shown).

Indirect immunofluorescence assay (IFA)
ST and IPI-2I cells in 24-well plates were infected with PDCoV at a multiplicity of infection (m.o.i.) of 0.1 in DMEM. After adsorption at 37°C for 30 min, the virus was removed, and the cells were treated with 100 g/ml trypsin in DMEM for 10 min at room temperature, or left untreated. After the trypsin had been removed, the cells were washed three times in DMEM. The cells were cultured in DMEM with or without 10 g/ml trypsin for 24 h. When the cytopathic effect (CPE) was obvious, the cells were fixed with 4% paraformaldehyde and blocked (5% nonfat dry milk in PBS) overnight. The cells were subsequently incubated for 60 min with a mouse anti-PDCoV N protein mAb, followed by Alexa Fluor ® 488-conjugated sheep anti-mouse IgG for 60 min. Nuclei were stained with 4Ј,6-diamidino-2-phenylindole (DAPI) and the results were observed using a fluorescence microscope.
a Primers used for RT-qPCR. b Primers used for the amplication of CTSL and CTSB genes. The products were cloned into EcoR I (bold) and Xho I (underlined) sites of the pAAV-IRES-hrGFP vector.

Cell entry of porcine deltacoronavirus
ST and IPI-2I cells grown to 90% confluence in 12-well plates were transfected with a pAAV-IRES-hrGFP-PDCoV NH opti-S plasmid using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's instructions. At 48 h post-transfection, the cells were treated with 100 g/ml trypsin in DMEM for 10 min at room temperature, or left untreated. After removing the trypsin, the cells were washed three times in DMEM. The cells were cultured in DMEM with or without 10 g/ml trypsin for 24 h. The cells were fixed and stained with DAPI. The number of nuclei in the syncytia was counted.
ST cells were pretreated with 50 nM Baf-A1 at 37°C for 1 h, then inoculated with PDCoV (m.o.i. ϭ 0.1) in the presence 10 g/ml trypsin or trypsin and 40 g/ml soybean trypsin inhibitor type I (SBTI, Sigma) for 24 h (18). The cells were fixed and staining with mouse anti-PDCoV N mAb, following with Alexa Fluor ® 633-conjugated goat anti-mouse IgG (HϩL) antibody to detect PDCoV-positive cells.

Cell to cell fusion assay
To quantify the fusion induced by trypsin, a cell-fusion assay based on luciferase expression was performed as described previously (52). Briefly, effector cells (293T) were co-transfected with WT (pCAGGS-opti S) and pCDNA3.1-T7 vectors. The target cells (ST) were transfected with a pET-32a-IRES-luc plasmid. At 6 h post-transfection, the ST cells were quickly trypsinized and overlaid with 293T cells; 24 h later, the cocultured cells were treated with various concentrations of trypsin in serum-free DMEM at room temperature for 30 min to induce cell fusion. The cells were incubated for another 6 h with fresh DMEM with 10% FBS and then lysed. Luciferase activity was measured using a luciferase assay kit (Promega) with an EnSpire ® multifunctional microplate reader (PerkinElmer).

Inhibitors
Various inhibitors were used as described in previous studies (43,46,49). To determine the optimal inhibitor concentration, ST and IPI-2I cells were pretreated with various concentrations of lysosomal acidification inhibitor (bafilomycin-A1, Abcam) and protease inhibitors (CA-074, Sigma-Aldrich; cathepsin L inhibitor Z-FY-CHO, (CTSLI), Santa Cruz Biotechnology; and E64d, Sigma-Aldrich) for 1 h at 37°C, after which fresh DMEM with 10% FBS was added for 24 h. The effect of inhibitors on the viability of ST and IPI-2I cells was determined with a CCK-8 kit (Dojindo Molecular Technologies, Gaithersburg, MD). The inhibitor-treated ST and IPI-2I cells were infected with PDCoV at an m.o.i. of 1 for 6 h because a previous study showed that one cycle of replication for most coronaviruses takes 6 h. The total cellular RNA was isolated, and RT-qPCR was performed to measure viral entry.

Western blotting
IPI-2I cells were pretreated with 50 M CA-074 or 50 M CTSLI for 1 h at 37°C, and the cells were infected with PDCoV at an m.o.i. of 1 for 30 min at 4°C, followed by treatment with 100 g/ml trypsin at room temperature for 20 min and washed three times with DMEM. At 6 h post-infection, the cells were prepared for Western blotting. The proteins were transferred to a nitrocellulose membrane and incubated with mouse antipig CTSL (Abcam) or rabbit anti-pig CTSB (Cell Signaling Technology) antibodies, a mouse anti-PDCoV N protein mAb, and mouse anti-GAPDH antibody, followed by horseradish peroxidase (HRP)-conjugated sheep anti-mouse IgG (Sigma-Aldrich) or HRP-conjugated sheep anti-rabbit IgG (Thermo Fisher Science). The expression of proteins was detected using the enhanced chemiluminescence Western blotting detection system (GE Healthcare Life Sciences).

Knockdown
The siRNAs were designed by GenePharma (Shanghai, China) ( Table 2). CTSL-or CTSB-specific siRNA pools or control siRNA were used to transfect IPI-2I cells. After 24 h, the cells were prepared for the detection of gene silencing using Western blotting. In some experiments, after 24 h of transfection, IPI-2I cells were infected with PDCoV at an m.o.i. of 1 and the cells were subjected to RT-qPCR analysis 6 h later.

Enzymatic activity assay
CTSL or CTSB enzymatic activity after PDCoV infection was measured with a Magic Red ® Cathepsin L or a Cathepsin B Detection Kit (ImmunoChemistry Technologies, Bloomington, MN) (51). Briefly, IPI-2I cells in 12-well plates were infected with PDCoV at an m.o.i. of 1. At different times (6, 12, and 24 h), the cells were treated with Magic Red ® staining solution, which could be cleaved by CTSL or CTSB. The products were observed using a fluorescence microscope equipped with an excitation filter of 550 nm and a long pass Ͼ610 nm emission filter pair. The fluorescence intensity of the red fluorescence could be detected at an optimal excitation and emission wavelength of 592 nm and 628 nm, respectively. The experiment was repeated three times.

Growth curves
IPI-2I cells were pretreated with inhibitors for 1 h at 37°C, and infected with PDCoV at an m.o.i. of 0.01 for 1 h at 37°C. The cells were treated with 100 g/ml trypsin at room temperature for 20 min and washed three times with DMEM, followed by incubation at 37°C. Supernatant samples were collected at different time points and stored at Ϫ70°C. The TCID 50 of the virus was quantified after three freeze-thaw cycles (62).

Cell entry of porcine deltacoronavirus Western blot analysis of spike cleavage by trypsin
ST cells were infected with PDCoV at an m.o.i. of 1, and 24 h later, the cells were collected and lysed by sonication as described previously with some modifications (49). Cell lysates were then incubated with various concentrations of trypsin at 37°C for 30 min and subjected to Western blot analysis. For RN mutation assay, plasmids encoding S or S RN gene were transfected to 293T cells. 48 h after transfection, the cells were collected, washed with PBS, and lysed by sonication. Trypsin cleavage was performed as described above and Western blot analysis was conducted with S polyclonal antibody.

Statistical analysis
All data are representative of at least three independent experiments. Data were analyzed using a Student's t test, and values are reported with mean Ϯ S.D. A threshold of p Ͻ 0.05 was considered significant.