A Single Point Mutation Creating a Furin Cleavage Site in the Spike Protein Renders Porcine Epidemic Diarrhea Coronavirus Trypsin Independent for Cell Entry and Fusion

Entry of the enveloped coronaviruses (CoVs) is mediated by the spike (S) glycoprotein. The S protein can be functionally divided into two domains, S1 and S2, which enable the consecutive stages of receptor binding and membrane fusion, respectively. Like other class I viral fusion proteins, the CoV S proteins rely on proteolytic cleavage by host proteases for fusion priming/activation; cleavage enables the release of the fusion peptide and its insertion into the target cell membrane in a controlled manner. To this end, most CoVs exploit endogenous cellular proteases, such as furin, plasma membrane proteases, and endolysosomal proteases, that cleave the S protein during virus exit from the infected cell or during its entry at or in the target cell (1, 2). Cleavage of CoV S proteins has been shown to occur at the junction of receptor binding domain S1 and membrane fusion domain S2 and just upstream of the putative fusion peptide (S2′ site) in the S2 domain, with the latter cleavage supposed to be the most critical for fusion, as it liberates the fusion peptide (1, 2).

Porcine epidemic diarrhea virus (PEDV) is an emerging coronavirus that causes acute diarrheal disease in swine in Asia and since 2013 in the Americas. In contrast to other CoVs, PEDV requires the presence of an exogenous protease, trypsin, in the cell culture medium to activate the S protein for membrane fusion and for virus propagation in cell culture (3). This trypsin dependency in vitro is consistent with the tissue tropism of PEDV in vivo, where PEDV is confined to the trypsin-rich small intestine. Trypsin activates the PEDV S protein for membrane fusion after the binding of the virion to the host cell (4). Genetic and mutational analyses have pointed to a conserved arginine just upstream of the fusion peptide (S2′ site) in membrane fusion domain S2 as the trypsin cleavage site for fusion activation, yet direct evidence of functional cleavage at this position is lacking (4).

To assess the cleavage position required for PEDV S protein fusion activation and to test whether activation by exogenous trypsin can be bypassed by an endogenous host protease, we introduced an artificial cleavage recognition site for the furin protease into the PEDV spike protein (Fig. 1). The proprotein convertase furin is a type I membrane protein that is ubiquitously expressed in eukaryotic cells. Furin is located primarily in the trans-Golgi network but also occurs at the cell surface and as an extracellular, truncated soluble form (5). It cleaves cellular and viral proteins C terminal of a substrate recognition motif containing basic amino acids, with R-X-X-R and R-X-K/R-R (R, arginine; K, lysine; X, any amino acid) representing the minimal and highly favored motifs, respectively. We constructed a mutant PEDV S protein (strain CV777) with a valine-to-arginine single-residue substitution at amino acid position 888 (V888R), thereby creating a furin cleavage sequence site (VQKR→RQKR) N terminal of the predicted fusion peptide (6). Bioinformatic analysis predicted that this position can be cleaved by furin with a prediction score of 0.875, relative to 0.611 for the wild-type S (S^WT) protein (Fig. 1a).

First, we assessed whether the introduction of the furin cleavage site in S (S^FCS) enables trypsin-independent cell-cell fusion. Therefore, we transiently expressed the S^FCS protein and the S^WT protein, each provided with a C-terminally appended Flag tag, in Vero cells. At 6 h posttransfection, the culture medium was replaced by fresh medium supplemented either with a furin inhibitor or with a soybean trypsin inhibitor, the latter to ensure that trypsin activity was completely absent. At 47 h posttransfection, cells were treated with trypsin for 1 h or left untreated. In the presence of trypsin, both the S^WT and S^FCS spike proteins were able to efficiently form syncytia (Fig. 1b). As expected, the S^WT spike protein was unable to mediate cell-cell fusion when trypsin was omitted from the cell culture medium. However, the formation of multiple, large syncytia were seen after expression of the S^FCS spike protein in the absence of trypsin activity. Trypsin-independent syncytium formation by S^FCS may be inhibited by the inclusion of furin inhibitor in the culture medium (Fig. 1b). These data indicate that the creation, through a single amino acid substitution, of a furin cleavage site at the S2′ position renders the PEDV S protein prone to activation of its membrane fusion capacity by endogenous furin.

To assess whether the furin cleavage site at the S2′ position of the S protein enables the virus to infect cultured cells independently of trypsin, we introduced the mutation encoding the S^V888R substitution into the viral genome. We generated recombinant viruses encoding either the wild-type CV777 S protein (PEDV-S^WT) or the S^FCS mutant (PEDV-S^FCS) using our recently established PEDV reverse-genetics system (7). To facilitate the analyses, the nonessential ORF3 gene in the genomes of both recombinant viruses was replaced by the green fluorescent protein (GFP) gene. Recombinant PEDV-S^WT and PEDV-S^FCS were successfully rescued in the presence of trypsin, and the relevant S gene region of their genomes was confirmed by sequencing. Next, we inoculated Vero cells in parallel with PEDV-S^WT and PEDV-S^FCS in the absence and presence of trypsin for 2 h. The infection was subsequently continued in the absence or presence of trypsin for another 10 h, after which infected (i.e., GFP-positive) cells were visualized by fluorescence microscopy. In the absence of trypsin, hardly any infection was seen on Vero cells for PEDV-S^WT, whereas clear infection was observed for the PEDV-S^FCS mutant (Fig. 2a). Both viruses efficiently infected Vero cells in the presence of trypsin. Syncytia were abundantly observed when trypsin was absent after virus inoculation for the PEDV-S^FCS mutant but not for PEDV-S^WT (Fig. 2a and b). These results confirm our earlier observation that the furin recognition site creating mutation S^V888R confers trypsin-independent cell-cell fusion.

To assess whether the S^V888R mutation would also enable PEDV to propagate in the absence of trypsin, the growth kinetics of PEDV-S^WT and PEDV-S^FCS on Vero cells were compared in the absence and presence of trypsin. Vero cells were inoculated in parallel with PEDV-S^WT and PEDV-S^FCS at a multiplicity of infection (MOI) of 0.01 or 0.1. Both viruses displayed similar growth kinetics and reached similar titers in the presence of trypsin (Fig. 2c). In the absence of trypsin, no infectious progeny was detected with either virus after inoculation at a low MOI (0.01 [data not shown]). At a 10-fold-higher MOI, PEDV-S^FCS—but not PEDV-S^WT—yielded low infectious virus titers starting from 12 h postinfection (p.i.). To check the cleavage status of the spike protein as it occurs on PEDV-S^FCS virions, we performed Western blot analysis. The results indicate that the introduction of the furin cleavage site at the S2′ site did not result in detectable cleavage (Fig. 2e).

Huh-7 human hepatoma cells are known to express high levels of furin protease (8). We therefore analyzed whether these cells could support infection by PEDV-S^FCS. Of note, infection of Huh-7 cells in the presence of trypsin could not be assessed since these cells are affected too strongly by trypsin at the concentrations required for propagation of PEDV in cell culture. Inoculation of Huh-7 cells with PEDV-S^FCS resulted in trypsin-independent infection and the development of syncytia (Fig. 3a and b). Virus entry as well as cell-cell fusion could be inhibited when the furin inhibitor was present during these processes. Interestingly, some PEDV-S^WT infection was also observed on Huh-7 cells, which could be inhibited by furin inhibitor, yet no syncytia were seen. Whether this infectivity correlates with the predicted suboptimal proprotease cleavage site at the S2′ position in the PEDV S^WT protein (Fig. 1) remains to be seen.

The proteolytic activation process of the CoV spike fusion protein has long been rather enigmatic. Whereas processing at the S1/S2 junction by furin was documented already in the early 1980s (9, 10), this cleavage occurs only in a subset of coronaviruses and does not liberate the putative fusion peptide at the N terminus of the membrane-anchored subunit, as it does in other class I viral fusion proteins. The second, more universal, as well as more appropriately located, S2′ cleavage site was identified only recently, and evidence for its general importance in CoV infection has since been accumulating (4, 11–13; for a recent review, see reference 1). Cleavage at the S2′ position is generally carried out by cellular proteases occurring at the plasma membrane or in the endo-/lysosomal system, depending on the particular target cell and S2′ sequence. In the case of PEDV, however, in which the S protein lacks a canonical furin cleavage site at the S1/S2 and the S2′ position, activation supposedly occurs by trypsin-like enzymes in the gut, and the virus hence requires supplementation of trypsin for propagation in vitro. Earlier, we mapped this trypsin requirement to the S2′ cleavage site in the PEDV S protein and demonstrated the critical importance of the characteristic arginine at this site for the viability of the virus and for the cell fusion capacity of the S protein (4).

In the present study, we aimed to demonstrate the requirement for cleavage at the S2′ site and to alleviate the trypsin dependence of PEDV infection in vitro. Thus, we show that introduction—through a single point mutation—of an artificial furin cleavage motif N terminal of the spike fusion peptide confers cell-cell fusion and PEDV entry in a trypsin-independent manner. Both processes were blocked by a furin-specific inhibitor, thereby confirming the functionality of furin cleavage at the S2′ position. The observations add further evidence that cleavage just upstream of the fusion peptide is a general and essential requirement for activation of CoV spike proteins for membrane fusion (11–14). Propagation of PEDV-S^FCS in the absence of trypsin was less efficient than in its presence, which might be due to trypsin also being required for virus release (15). Moreover, besides cleavage N terminal of the fusion peptide, additional cleavage(s) may be required to increase the S protein's membrane fusion efficiency. Cleavage of the S protein at the S1/S2 junction of the coronaviruses Middle East respiratory syndrome (MERS) CoV, severe acute respiratory syndrome (SARS) CoV, and infectious-bronchitis virus (IBV) has been implicated to precede and promote cleavage at the S2′ position (11, 13, 14). For PEDV, the lack of efficient cleavage of virion-incorporated S^FCS proteins indicates that the introduced furin cleavage site at the S2′ site is rather inaccessible for furin. Whether cleavage at the S1/S2 junction or binding to the receptor enhances the efficiency of cleavage at the S2′ cleavage site awaits further investigation.

While field strains of PEDV strictly require exogenous trypsin for propagation in vitro, serial passaging of the virus on cultured cells can lead to trypsin independency (3, 4). Vero cells are the most commonly used cells for PEDV studies because of their resistance to the high concentrations of trypsin required for PEDV infection. Hence, the protease requirement of PEDV, in addition to the specific virus receptor, may function as a critical tropism determinant in vitro as well as in vivo. As a consequence, the rational or evolutionary adaptation of CoVs to the use of ubiquitous, endogenous proteases like furin for the activation of their S proteins may expand their tropism in vitro and in vivo (1).

This study was supported by a grant from the Natural Science Foundation of China (31272572) provided to Qigai He.