Proteolytic processing of Middle East respiratory syndrome coronavirus spikes expands virus tropism

Significance

Middle East respiratory syndrome coronavirus (MERS-CoV) can cause lethal pneumonia by infecting lung epithelial cells. Infection requires viral “spike” proteins, which catalyze virus–cell membrane fusion during cell entry. Fusion catalysis requires that spikes be first cleaved by cellular proteases. MERS-CoV spikes are cleaved in virus-producing cells before subsequent virus-cell entry, whereas other human coronaviruses have spikes that are cleaved in virus-infecting cells during virus-cell entry. Here, we found that the early MERS-CoV cleavages are required for the subsequent infection of human lung-like cells, but are dispensable for infection of several other cell types. Our findings demonstrate that the spike cleavage status of MERS-CoVs dictates cell tropism, and points to spike proteolytic processing as a correlate of MERS-CoV virulence.

Abstract

Middle East respiratory syndrome coronavirus (MERS-CoV) infects humans from zoonotic sources and causes severe pulmonary disease. Virions require spike (S) glycoproteins for binding to cell receptors and for catalyzing virus–cell membrane fusion. Fusion occurs only after S proteins are cleaved sequentially, first during their secretion through the exocytic organelles of virus-producing cells, and second after virus binding to target-cell receptors. To more precisely determine how sequential proteolysis contributes to CoV infection, we introduced S mutations obstructing the first cleavages. These mutations severely compromised MERS-CoV infection into human lung-derived cells, but had little effect on infection into several other cell types. These cell type-specific requirements for proteolysis correlated with S conformations during cell entry. Without the first cleavages, S proteins resisted cell receptor-induced conformational changes, which restricted the second, fusion-activating cleavages. Consistent with these findings, precleaved MERS viruses used receptor-proximal, cell-surface proteases to effect the second fusion-activating cleavages during cell entry, whereas the more rigid uncleaved MERS viruses trafficked past these cell-surface proteases and into endosomes. Uncleaved viruses were less infectious to human airway epithelial and Calu3 cell cultures because they lacked sufficient endosomal fusion-activating proteases. Thus, by sensitizing viruses to receptor-induced conformational changes, the first S cleavages expand virus tropism to cell types that are relevant to lung infection, and therefore may be significant determinants of MERS-CoV virulence.

Enveloped viruses deposit their genomes into host cells by coalescing their membranes with the cell. These functions are executed by virion envelope-anchored glycoprotein trimers termed “membrane-fusion proteins.” In virus-infected cells, these proteins are synthesized as inactive forms, structured such that they can maintain their membrane-fusion potential throughout their residence on extracellular virus particles. The proteins then transit into fusion-competent forms during virus-cell entry. Various environmental stimuli control these cell entry-related structural transitions. Proteolysis is central, as fusion proteins cleaved by host proteases are frequently liberated to undergo transitions into fusion-competent forms (1–3). Knowledge of the proteolytic cleavages and host proteases regulating virus infections can be used to predict viral tropism and pathogenesis (4, 5), and can also reveal antiviral strategies (6).

Coronaviruses (CoVs) are enveloped, positive-stranded RNA viruses in the order Nidovirales. These viruses infect mammals and birds, and are mainly associated with respiratory and enteric tract disorders (7). Of six known human CoVs, the severe acute respiratory syndrome (SARS)-CoV and Middle East respiratory syndrome (MERS)-CoV are the most recent to have emerged from zoonotic reservoirs, which include bats (8–10), Himalayan palm civets and other animals found in wet markets in China (11, 12), and Dromedary camels (13). Infection can cause acute respiratory distress with high mortality, particularly in elderly individuals and those with underlying pulmonary dysfunctions (14, 15). It is unknown whether the spread of these highly pathogenic CoVs between animals and humans, as well as within the infected respiratory system, relates to their propensity to use particular lung proteases as entry factors.

CoV fusion proteins, called spike (S) proteins, are integral-membrane ∼500-kDa trimers that project ∼20 nm from viral envelopes (Fig. 1A). Atomic resolution structures of two CoV S protein ectodomains have recently been obtained (16, 17). These structures can be broadly divided into receptor-binding domains (RBDs in Fig. 1A), and fusion-catalyzing domains (FDs in Fig. 1A). For MERS-CoV, the S RBDs bind to dipeptidyl peptidase 4 (DPP4/CD26) receptors (18). Following receptor binding, the FDs are revealed through unfolding transitions, such that viral fusion peptides (FP in Fig. 1B) intercalate into host cell membranes. Subsequent refolding transitions bring viral and cellular membranes so close that they coalesce.

Fig. 1.

S1/S2 cleavage is required for MERS-CoV infection of Calu3 cells. (A) CoV particles have ∼100 S trimers. A single CoV S monomer is highlighted. A globular RBD (S1) and a stem-like FD (S2) is separated by domains that include the S1/S2 cleavage site. (B) A linear depiction of the MERS-CoV S protein includes a RBD, FP, two heptad repeats (HR1 and HR2), and a transmembrane domain (TMD). Two known proteolytic cleavage sites (S1/S2 and S2′) are depicted. Additional putative furin/proprotein convertase (black arrow heads) and cathepsin (white arrow heads) cleavage sites were predicted by ProP 1.0 Server (www.cbs.dtu.dk/services/ProP/) and SitePrediction (www.dmbr.ugent.be/prx/bioit2-public/SitePrediction/index.php), respectively. (C) Vero81, Huh7, and Calu3 cells were infected with WT and S1/S2 mutant (YSAS and SSVR) MERS-CoVs. After 5 h, subgenomic N mRNAs were determined by real-time PCR, and plotted relative to N mRNA levels in WT MERS-CoV–infected Vero81 cells. (D) Vero81, Huh7, and Calu3 cells were infected with WT and S1/S2 mutant MERS-CoVs. Progeny were collected at 20-h postinfection and titrated by plaque assay in Vero81 cells. (E) Six HAE cultures were infected with WT and S1/S2 mutant MERS-CoVs. Progeny were collected at 20-h postinfection and titrated by plaque assay in Vero81 cells. Error bars present SD from the mean (n = 3 for cultured cells, n = 6 for HAE cultures). Statistical significance was assessed by Student's t test. ^†P < 0.01; ^‡P < 0.001; ns, not significant. (F) Vero81, Huh7, and Calu3 cells were infected with WT and S1/S2 mutant MERS-CoVs and S proteins within cells were analyzed by Western blot. Uncleaved (S_unc) and cleaved (S1 and S2) positions are indicated. The numbers at the left indicate molecular mass in kilodaltons.

The unfolding of the FDs is thought to require S protein cleavage by host proteases (19). Although the MERS-CoV S proteins have several proteolytic cleavage sites between the RBDs and FDs, a scission at the S1/S2 site (Fig. 1B) takes place in virus-producing cells, by furin or related proprotein convertases (20). Subsequent proteolyses at more C-terminal sites, notably the S2′ site (Fig. 1B), takes place later in the virus transmission process, after virions have been released from producing cells and bound to target-cell receptors (1, 21, 22). These proteases include members of type II transmembrane serine proteases (TTSPs) (23, 24), furin/proprotein convertases (21, 25), and cathepsins (23, 26).

Proteolysis at S2′ is thought to trigger S-mediated membrane fusion. Therefore, the abundance and subcellular distribution of the TTSPs, furin/proprotein convertases, and cathepsins quite likely determines host cell susceptibility to CoV infection, as well as the tempo and site of CoV fusion into cells (19). In contrast, the relevance of the S1/S2 proteolysis to CoV infection is more ambiguous. For example, several infectious CoVs lack S1/S2 proteolytic processing motifs and therefore secrete from producer cells as uncleaved forms (27, 28). For those CoVs with S1/S2 processing sites, mutation of the sites to block the cleavages often does not abrogate viral infectivity (20, 29). Pharmacologic inhibition of proprotein convertases diminished preliminary MERS-CoV S1/S2 cleavages, but did not reduce viral infectivity into several cell types (30).

We hypothesized that S1/S2 proteolysis in virus-producing cells facilitates subsequent S2′ cleavage by fusion-activating proteases in virus target cells. An extension of this hypothesis was that the S1/S2 proteolysis might determine which target-cell proteases (TTSPs, furins, cathepsins) are required for fusion activation. Given that these proteases are uniquely present in particular cell types, the S1/S2 cleavages might influence MERS-CoV cell tropism.

Results

S1/S2 Cleavage Is Necessary for MERS-CoV to Infect Calu3 Cells.

To evaluate the consequences of S proteolytic cleavage on virus infection, we engineered site-specific S1/S2 mutations (RSVR₇₅₁/S in Fig. 1B). The WT S1/S2-encoding region was replaced with the codons (YSAS₇₅₁/S) from a MERS-CoV–related bat CoV-HKU4 (Hong Kong University 4). Of note, HKU4 S proteins are secreted from cells as uncleaved forms (31). Additionally, we mutated the R₇₄₈ codon to serine (S₇₄₈SVR₇₅₁/S), expecting that these S proteins would secrete as uncleaved forms but remain susceptible to extracellular serine protease cleavage at S1/S2 sites.

These mutations were incorporated into the MERS-CoV genome and recombinant viruses were generated (32). WT and mutant MERS-CoVs, produced in Vero81 cells, were infected into several different target cells, including Vero81, Huh7, and Calu3. These target cells have differing amounts and types of cellular proteases (23, 25, 30, 33), and therefore each cell type might uniquely process MERS-CoV S proteins, generating differential infection. At 5-h postinfection, total RNAs were extracted from the infected cells and the relative amounts of subgenomic nucleocapsid (N) mRNA were quantified as a measure of infection. In Vero81 and Huh7 cells, infections by WT and mutant MERS-CoVs were comparable, with only a modestly reduced infection of the SSVR mutant into Huh7 cells (Fig. 1C). However, in Calu3 cells, the mutants produced ∼4 log₁₀ less viral mRNA than WT (Fig. 1C). Consistently, WT and mutant MERS-CoVs produced comparable amounts of progeny viruses in Vero81 and Huh7 cells, whereas in Calu3 cells, mutant virus titers were nearly 3 log₁₀ less virus than those of the WT (Fig. 1D). Calu3 are human lung adenocarcinoma cells (34). To assess susceptibility of normal human respiratory cells, viruses were infected into primary cultures of well-differentiated human airway epithelia (HAE). Average progeny virus yields from HAE were ∼10^4.4 for WT and mutant SSVR viruses, but were significantly lower at ∼10^3.6 for mutant YSAS viruses (Fig. 1E). These findings demonstrated that the engineered mutations specifically disabled infection into lung-derived Calu3 cells, with replacement of all basic S1/S2 residues also compromising infection into HAE cultures.

Infected cells were also evaluated for cleaved S proteins by Western blotting (35). Partial WT S1/S2 cleavage was evident in all cell types, but the mutant S proteins were either uncleaved or absent (Fig. 1F). Collectively, these findings suggested that viruses must undergo a minimal extent of S1/S2 cleavage in producer Vero81 cells to be infectious to target Calu3 cells.

S1/S2 Cleavage Increases MERS S-Mediated Virus-Cell Entry.

To discern the mechanisms by which preliminary S1/S2 cleavage increases MERS-CoV infection, we produced HIV- and vesicular stomatitis virus (VSV)-based MERS pseudoparticles (pps) in 293T cells. Upon S-mediated transduction of target cells, these pps express firefly luciferase (Fluc), with Fluc levels providing quantitative measurements of virus-cell entry.

First, the pps were pelleted, and particle-associated S proteins were evaluated by detecting C-terminal epitopes on Western blots. WT MERS-CoV S proteins were mostly cleaved, whereas both mutant S proteins were uncleaved (Fig. 2A). When inoculated at equivalent input multiplicities, the uncleaved mutants were specifically compromised for transduction into lung-derived Calu3 cells (Fig. 2A) and HAE cells (Fig. 2B). These results were comparable to those obtained with infectious viruses, demonstrating that the MERS pseudoparticle (pp) transductions reflect authentic virus entry, and indicating that the S1/S2 cleavages affect infection at the level of virus-cell entry.

Fig. 2.

S1/S2 cleavage is required for MERS pp entry into Calu3 cells. (A) Vero81, Huh7, and Calu3 cells were inoculated with WT, S1/S2 mutant (YSAS and SSVR) MERS pps, or with pp-lacking S proteins (Bald). (B) Three HAE cultures were inoculated with VSV-based WT and S1/S2 mutant MERS pseuodoparticles. (C) MERS pps were pretreated with trypsin and then used to inoculate Calu3 cells. (D) MERS pps were produced in the presence of PCI, cleared free of residual PCI, and used to inoculate Calu3 cells. In all experiments, virus entry was quantified by measuring luciferase levels at 18 h (B) or at 48 h (A, C, D) posttransduction. Lower panels depict S proteins on MERS pps after Western blotting. Uncleaved (S_unc) and cleaved (S2) positions are indicated. The numbers at the left indicate molecular mass in kilodaltons. Error bars present SD from the mean (n = 3). Statistical significance was assessed by Student's t test. *P < 0.05; ^†P < 0.01; ^‡P < 0.001; ns, not significant.

Second, the MERS pps were prepared in serum-free 293T media (SFM) and incubated with trypsin before transduction. Trypsin processed the uncleaved fraction of WT S proteins, as well as the uncleaved SSVR mutant S proteins, into fragments at ∼95 kDa (Fig. 2C), consistent with scission at the R₇₅₁/S cleavage site. In contrast, YSAS S proteins were cleaved at an alternative more N-terminal cleavage site, most likely RSTR₆₉₄/S (30), as inferred from the ∼100-kDa C-terminal fragments (Fig. 2C). On transduction into Calu3 cells, this trypsin processing was associated with a >100-fold increase in SSVR pp transduction (Fig. 2C). In contrast, the WT and YSAS pp transductions were not affected by trypsin (Fig. 2C). The fact that the YSAS pps remained incompetent for transduction even after trypsin proteolysis to an ∼100 kDa form suggested that cleavage at a precisely defined S1/S2 site is a prerequisite for Calu3 infection.

Third, MERS pps were produced in the presence of a proprotein convertase inhibitor (PCI, also known as a dec-RVKR-cmk) (36). These pps were concentrated, and pp-associated S proteins were evaluated by Western blotting. Compared with control MERS pps, those secreted from PCI-treated cells contained primarily uncleaved S proteins and were compromised for Calu3 transduction (Fig. 2D). These results confirmed that MERS-CoV depends on S1/S2 cleavage by proprotein convertases within 293T producer cells to efficiently infect lung-derived Calu3 target cells.

Preliminary Cleavage Increases SARS S-Mediated Virus-Cell Entry.

In contrast to MERS-CoV, which is naturally secreted with cleaved S proteins, the other five human CoV S proteins mediating human lung infections are produced as uncleaved forms (28). Thus, we asked whether preliminary S cleavage facilitates entry by these naturally uncleaved CoVs. HIV-based pps bearing SARS-CoV or human CoV strain 229E S proteins were produced in SFM and treated with trypsin before virus entry. Trypsin processed the SARS S, but not the 229E S proteins, into ∼100-kDa fragments, consistent with SARS S cleavage into S1 and S2 (Fig. 3A). Without trypsin, the SARS and 229E pps did not transduce Calu3 cells; trypsin exposure before virus inoculation significantly and specifically increased only the SARS pp transduction (Fig. 3B). The correlation between S1/S2 proteolysis and Calu3 transduction thus applied to both MERS and SARS pps, suggesting that these two pathogenic human CoVs expand their infection through preliminary S cleavage.

Fig. 3.

SARS-CoV S cleavage by trypsin facilitates entry into Calu3 cells. (A) SARS and HCoV-229E pps were treated with trypsin. S proteins on pps were analyzed by Western blot. Uncleaved (S_unc) and cleaved (S2) positions are indicated. The numbers at the left indicate molecular mass in kilodaltons. (B) Calu3 cells were incubated with trypsin pretreated pps. Virus entry was quantified by measuring luciferase levels at 48-h posttransduction. Error bars present SD from the mean (n = 4). Statistical significance was assessed by Student's t test. ^‡P < 0.001.

Preliminary Cleavage Facilitates Subsequent S Proteolysis After Receptor Binding.

For SARS-CoV and some murine CoVs, the target-cell proteases triggering membrane fusion operate only after the viruses have bound to receptors (2), suggesting that receptor binding induces conformational changes that display S cleavage sites to proteases. We hypothesized that the MERS-CoV S proteins also respond to receptor binding by changing conformation, in ways that are facilitated by prior S1/S2 cleavages. To address this hypothesis, we used an in vitro assay that imitates the structural changes occurring during virus-cell entry. In the assays, MERS pps were incubated on ice with fivefold molar excess of membrane-anchored DPP4 (DPP4 pps), allowing S proteins to bind their cognate receptors. The mixed pps were then digested with trypsin, a surrogate for MERS-CoV activating proteases (33), and S fragmentation patterns were analyzed by Western blotting. In the absence of DPP4, trypsin cleaved the WT S proteins at the S1/S2 cleavage site, and at higher doses, cleaved further to form an ∼40-kDa fragment (Fig. 4A). However, in the presence of DPP4, trypsin uniquely generated a stable ∼72-kDa fragment (Fig. 4A), consistent with cleavage at the fusion-activating S2′ site (25). These findings suggested that specific biologically meaningful cleavage sites are exposed in response to receptor binding.

Fig. 4.

S1/S2 cleavage facilitates the receptor-induced exposure of a second S2′ cleavage site. (A) MERS (WT) pps were mixed with SFM or with DPP4 pps. The mixed pps were treated with increasing concentrations of trypsin. S and DPP4 proteins on pps were analyzed by Western blotting. Uncleaved (S_unc) and cleaved (S2, S2′, and 40K) positions are indicated. The numbers at the left indicate molecular mass in kilodaltons. (B) Similar experiments to A were carried out in parallel with YSAS MERS pps.

We carried out similar experiments with YSAS MERS pps, expecting that these uncleaved pps would be less susceptible to the receptor- and trypsin-driven S2′ cleavage. Indeed, we observed only barely detectable amounts of YSAS S2′ fragments, even at the high trypsin doses that cleave the majority of receptor-bound WT S proteins into S2′ fragments (Fig. 4B). These data indicate that the receptor-promoted exposure of the S2′ cleavage site depends on a prior S1/S2 cleavage.

S1/S2 Cleavage Facilitates Early Virus Entry.

On the basis of the in vitro findings in Fig. 4, we hypothesized that S1/S2-cleaved MERS-CoVs would be further processed by target cell proteases shortly after receptor binding, whereas uncleaved MERS-CoVs would traffic onward to endosomes without being sufficiently processed, and would thus be dependent on late endosomal proteases. The proteases cleaving S proteins early after virus binding to target cells include serine-class transmembrane proteins (TMPRs) and furin/proprotein convertases (24, 37, 38), whereas those cleaving S proteins later, after virus endocytosis, include cysteine-class cathepsins (26). Therefore, early and late CoV-cell entry can be distinguished by their sensitivities to serine- and cysteine-specific protease inhibitors, respectively.

We evaluated MERS pp transductions in the presence of camostat, a serine-class TMPR protease inhibitor (23), PCI, an inhibitor of kexin-type proprotein convertases including furin (36), and E64d, an inhibitor of cysteine-class proteases including endosomal cathepsins (39). These inhibitors had no effect on VSV glycoprotein (G)-mediated transductions, but they did reduce MERS S transductions in cell-type–specific fashion (Fig. 5A and Fig. S1). In Calu3 cells, camostat, PCI, and E64d blocked transductions ∼100-, ∼10-, and ∼2-fold, respectively, indicating utilization of the early-acting proteases (Fig. 5A). Similar patterns were observed during transduction of HAE cultures, with camostat and PCI, but not E64d blocking virus entry (Fig. 5B). Conversely, in Huh7 cells camostat was inert, and PCI and E64d effected ∼fivefold reductions, a reverse pattern relative to Calu3 cells that suggested utilization of the later-acting proteases (Fig. 5A). MERS pp transductions into Caco2 and Vero81 cells were also sensitive to inhibitors of later-acting proteases (Fig. S1).

Fig. 5.

S1/S2 cleavage facilitates early virus entry. (A) Calu3 and Huh7 cells were incubated with protease inhibitors [100 µM camostat (camo), 50 µM PCI, 10 µM E64d], or with DMSO, and then transduced with MERS or VSV pps. (B) Three HAE cultures were incubated with indicated protease inhibitors, and then transduced with WT and S1/S2 mutant MERS pps. (C) Huh7 and Huh7/TMPRSS2 cells were incubated with protease inhibitors, and then transduced with WT or S1/S2 mutant (YSAS or SSVR) MERS pps. Virus entry was quantified by measuring luciferase levels in infected cells at 48-h (A, C) or 18-h (B) posttransduction. Error bars present SD from the mean (n = 3). Statistical significance was assessed by Student's t test. *P < 0.05; ^†P < 0.01; ^‡P < 0.001; ns, not significant.

Fig. S1.

Effect of protease inhibitors on MERS pp entry into Caco2 and Vero81 cells. Caco2 and Vero81 cells were incubated with protease inhibitors [100 µM camostat (camo), 50 µM PCI, 10 µM E64d], or with DMSO. Cells then were transduced with MERS or VSV pps. Virus entry was quantified by measuring luciferase levels at 48 h posttransduction and plotted relative to DMSO control values. Error bars present SD from the mean (n = 3). Statistical significance was assessed by Student’s t test. *P < 0.05; ^‡P < 0.001.

In determining whether uncleaved MERS pps were differentially sensitive to these protease inhibitors, we chose to use Huh7 cells, as they supported WT and uncleaved mutant MERS-CoVs nearly equally (Figs. 1 and 2). In these cells, the WT and uncleaved mutant MERS-CoVs were similar in their resistance to camostat, and in their modest sensitivity to PCI, but were set apart by the mutants’ hypersensitivity to E64d (Fig. 5C). This pattern, in which E64d potently blocked the mutant pp transductions, was also observed in Caco2 cells (Fig. S2). These findings indicated that the uncleaved viruses required a late endosomal cell entry pathway. We considered whether the uncleaved mutant pps might use early-acting proteases if they were provided in abundance. To this end, we overexpressed the TTSP family TMPR serine 2 (TMPRSS2) in Huh7 cells, and then evaluated MERS pp transductions. TMPRSS2 overexpression increased Fluc accumulation (about eightfold) and made all MERS pp transductions slightly sensitive to camostat, and also rendered WT MERS pp transduction completely resistant to E64d (Fig. 5C), indicating ample supplies of the TMPRSS2 proteases. However, the overexpressed TMPRSS2 did not render the uncleaved YSAS and SSVR MERS pp transductions resistant to E64d (Fig. 5C), indicating that the uncleaved MERS-CoVs are dependent on late-acting cathepsin proteases even when early-acting proteases are abundantly available.

Fig. S2.

Effect of protease inhibitors on WT and mutant MERS pp entry into Caco2 cells. (A) Caco2 cells were incubated with WT, S1/S2 mutant (YSAS and SSVR) MERS pps or with pp-lacking S proteins (Bald). (B) Caco2 cells were incubated with protease inhibitors (100 µM camostat, 50 µM PCI, 10 µM E64d), or with DMSO. Cells were then transduced with WT or S1/S2 mutant (YSAS or SSVR) MERS pps. Virus entry was quantified by measuring luciferase levels at 48-h posttransduction and plotted relative to DMSO control values. Error bars present SD from the mean (n = 3). Statistical significance was assessed by Student's t test. ^†P < 0.01; ^‡P < 0.001.

Cathepsin L Sensitizes Calu3 Cells to Uncleaved MERS Viruses.

Because uncleaved MERS viruses require late proteases and do not infect Calu3 cells, we inferred that Calu3 cells are depleted in endosomal cathepsins, at least relative to infectable Huh7 cells. This inference was validated by measuring protease-encoding transcripts by quantitative RT-PCR. Huh7 cells contained DPP4, furin, cathepsin L, and cathepsin B, but very few TMPRSS2 transcripts (Fig. 6A). Relative to the Huh7 cells, the lung and airway-derived Calu3 and HAE cultures had far more TMPRSS2, but significantly fewer furin and cathepsin L transcripts (Fig. 6B). To determine whether the relatively low levels of late-acting proteases in Calu3 cells accounted for virus resistance, we transduced the cells with human cathepsin L genes and then evaluated MERS pp entry. Transduced cells were highly sensitized to uncleaved MERS pp entry, and were made resistant by E64d (Fig. 6C). Similar augmentation of uncleaved MERS transduction was achieved by exposing cell-bound MERS pps to purified human cathepsin L (Fig. 6D). These results indicate that uncleaved viruses can only infect cells containing sufficient late-acting endosomal proteases, and they make it clear that CoV–cell tropism is related to the abundance and distribution of proteases in both virus-producing and virus-targeting cells.

Fig. 6.

Uncleaved MERS entry depends on endosomal cathepsins. (A) DPP4, TMPRSS2, furin, cathepsin L (Cat. L), cathepsin B (Cat. B), and hypoxanthine phosphoribosyltransferase (HPRT) mRNA levels were measured in Huh7 cell extracts by quantitative RT-PCR, and plotted relative to HPRT. (B) The same mRNA levels were measured in Calu3 and HAE cell extracts and plotted relative to those in Huh7 cells. (C) Calu3 cells were transduced with lentiviral vectors encoding human cathepsin L or GFP. After 3 d, cells were treated with or without 10 µM E64d and then transduced with WT or S1/S2 mutant MERS pps. (D) Calu3 cells were incubated with WT or YSAS MERS pps at 4 °C, and then treated with cathepsin L. Virus entry was quantified by measuring luciferase levels at 48-h posttransduction. Error bars present SD from the mean (n = 3). Statistical significance was assessed by Student's t test. ^†P < 0.01.

Discussion

CoV–cell entry can be viewed in the context of a proteolytic cascade that includes at least two cleavages in S proteins, first at S1/S2, then at S2′. For MERS-CoV, the cascade can begin shortly after virus morphogenesis in virus-producing cells. Extensive S1/S2 cleavage at this beginning stage allows for a similarly extensive S2′ cleavage shortly after virus binding to receptors in virus-target cells, and the cascade ends when a sufficient number of adjacent S proteins are triggered for fusion activation by early-acting proteases. However, for several CoVs, and for some variant forms of MERS-CoV, the cascade begins when viruses bind target cells, and then ends much later after virus endocytosis and late-acting intraendosomal proteolysis to activate a sufficient number of adjacent S proteins into fusion competence. These variations in the beginning and ending stages of the CoV infection-related proteolytic cascade are illustrated in Fig. 7.

Fig. 7.

MERS-CoV–cell entry model. In some producer cell types, MERS-CoV S proteins are cleaved by furin/proprotein convertases in the exocytic pathway. Cleaved MERS-CoV S proteins change their conformations rapidly after receptor binding, exposing subsequent proteolytic cleavage sites, which are processed by proteases (i.e., TMPRs, found at or near cell surfaces). Early cell-surface entry is achieved when several adjacent S proteins are processed. In other producer cell types, MERS-CoV S proteins are not cleaved. Uncleaved MERS-CoV S proteins slowly change their conformations after receptor binding. MERS-CoVs having uncleaved S proteins traffic to the late endosomes/lysosomes and late endosomal entry is achieved when several adjacent S proteins are eventually processed by cathepsins.

The beginning and ending points of this proteolytic cascade correlated with cell tropism. MERS-CoVs that began proteolysis at S1/S2 in virus-producing cells could infect Calu3 cells, but mutant MERS-CoVs that remained uncleaved could not (Figs. 1 and 2). Uncleaved MERS viruses bypassed early proteases (Fig. 5C) and required late proteases (Fig. 6 C and D), making it clear that S1/S2 cleavage promotes early entry. Of note, uncleaved viruses were able to infect the more cathepsin protease-enriched Huh7 and Vero81 cells (Figs. 1 and 2). Earlier reports used similar cathepsin-enriched target cell types in their experiments, and thus came to conclusions that the preliminary S1/S2 cleavages had a limited relevance to infection (20, 29, 30). It was the paucity of virus-activating endosomal proteases in Calu3 cells that revealed a key importance for the S1/S2 cleavages in cell tropism.

The relevance of the S1/S2 cleavages extended to infection of primary HAE cultures, as evidenced by the poor growth of the YSAS uncleaved mutant MERS-CoVs (Fig. 1E) and the relatively inefficient entry of the YSAS and SSVR uncleaved pseudoviruses (Fig. 2B). The WT-level growth of the SSVR mutants might argue against the importance of S1/S2 cleavages in HAE infections; however, we note that HAE are frequently collected from patients with underlying pulmonary diseases (e.g., chronic obstructive pulmonary disease, cystic fibrosis, pulmonary fibrosis, α-1 antitrypsin deficiency), conditions associated with increased levels of mucus-associated proteases (40–42). Airway mucus-associated proteases are known to cleave SARS-CoV S proteins (43), and therefore we suggest that the SSVR mutant MERS-CoVs are similarly cleaved by extracellular proteases, making their infection resemble that of WT viruses at the point of entry into some HAE cells.

Once on Calu3 or HAE target cells, the precleaved viruses are presumably processed to fusion-ready S2′ products by early cell-surface serine proteases (Fig. 5A) (23). That uncleaved viruses are not similarly processed by these proteases is best explained by their failure to respond to receptor binding, which limits downstream S2′ processing. MERS S protein interactions with DPP4 were necessary to reveal a proteolytic cleavage site at or near the S2′ position (Fig. 4A), demonstrating that a receptor-induced allosteric transition makes the S2′ cleavage site available to a protease. This DPP4-induced structural transition was far less effective in the absence of the preliminary S1/S2 cleavage (Fig. 4B). All of these findings fit well with the model in Fig. 7, which depicts the precleaved viruses transiting rapidly from receptor binding to fusion, whereas uncleaved viruses require more protease and more time to make this transit, and therefore do so only after being trafficked into protease-enriched endosomes.

The recent 4A resolution structures of murine CoV (MHV) and human CoV (HKU-1) S proteins depict uncleaved ectodomain trimers, without ligation to receptors (16, 17). Although images of S proteins in complex with cellular receptors will add important insights, the current structures do suggest that receptor binding locks relatively dynamic S proteins into conformations that expose cleavage sites to host proteases. S1/S2 cleavage may increase the dynamic properties of S proteins, giving them more opportunities to assume locked receptor-bound conformations, making the S2′ sites more rapidly available to proteases. A theme here is that S protein dynamics, achieved through proteolysis or through destabilizing mutations (44), may generally allow for early, cell-surface cleavage at the activating S2′ sites. Of note, such destabilizing mutations are liable to be counter-selected during in vitro virus growth in many cell cultures. In vitro conditions typically select for extracellular virion stability, which maintains infectivity in culture fluids. Therefore, the adaptive mutations fixed into CoV S proteins during in vitro virus propagation may restrict dynamic structural transitions (45). These in vitro mutations are frequently attenuating in in vivo CoV infection models (46). We suggest that in vivo attenuation arises because the stabilized viruses have reduced capacities to assume the receptor-bound conformations enabling S2′ proteolysis and fusion activation. The stabilized, cell culture-adapted CoVs are therefore directed to the late endosomal entry routes, which may correlate with diminished capacities to infect many of the target cells found in in vivo environments.

Several CoVs, including SARS- and human 229E-CoVs, remain uncleaved throughout their morphogenesis and secretion from virus-producing cells (27, 28). Thus, several CoV S proteins remain in their more rigidly structured uncleaved forms until target cell entry. For these less flexible viruses to transit into protease-sensitive and fusion-competent forms, they may need the energy derived from high-affinity interactions with receptors. Indeed the affinity of SARS-CoV S with its receptor angiotensin converting enzyme 2 (ACE2) is 10- to 20-fold higher than that of MERS-CoV S with its receptor DPP4 (47, 48). Thus, one can suggest that S1/S2-cleavages may reduce the need for high-specificity receptor interactions, and in doing so, may allow CoVs to bind adaptably to receptor orthologs, fostering zoonoses. Consistent with this hypothesis, the human-circulating MERS-CoVs have not undergone significant adaptations in their RBDs, at least not toward higher-affinity binding to DPP4. In fact, one lineage of human MERS-CoVs acquired reduced affinity to DPP4 (49). This finding contrasts with the SARS-CoV epidemics of 2003–2004, where adaptive changes increased binding affinities to ACE2 (50). Receptor utilization in relation to S proteolytic processing will inform us on CoV transmissions.

Recent excellent reports have made connections between CoV S proteolytic processing and CoV virulence. For example, CoV S protein proteolytic processing, principally by TMPRSS2, was suggested to increase viral pathogenesis by generating S “decoy” fragments that bind and sterically inactivate antiviral antibodies (24). Additionally, S protein proteolysis, again by TMPRSS2, was suggested to increase pathogenesis by allowing viruses to bypass IFN-induced transmembrane protein 3, an innate antiviral effector that blocks virus-endosome membrane fusions (51, 52). Here we claim that MERS-CoVs, and other CoVs, will preferentially use TMPRSS2 if they have been precleaved; that is, preliminary S1/S2 proteolysis gives infecting viruses the facility to use TMPRSS2 and other early-acting proteases for fusion activation. Thus, the preliminary S1/S2 cleavages may be the more proximal determinants of pathogenesis. Of the six known human CoVs, only MERS-CoV secretes from virus-producing cells with cleaved S proteins, and SARS-CoV, secreted uncleaved, can be processed by extracellular proteases before encountering target cells (Fig. 3) (1). Therefore, these viruses have special facility for using early-acting entry proteases. Preliminary S cleavages may also contribute to the tissue tropism and pathogenesis of feline and murine CoVs (4). As human CoVs continue to infect humans, it will be important to consider possible adaptive changes in their S protein proteolytic processing cascades, as these may contribute to CoV disease.

Experimental Procedures

Recombinant MERS-CoV Production and Infection.

Recombinant WT and mutant MERS-CoVs were generated from bacterial artificial chromosomes and infected into Vero81, Huh7, and Calu3 cells or HAE cultures. HAE cultures were obtained from the University of Iowa Cell Culture Core Facility, which acquired tissue by informed consent under an Institutional Review Board-approved organ research donation protocol. Virus infectivities were analyzed by real-time RT-PCR and by plaque assay.

HIV and VSV Pseudoparticle Transduction.

Viral psueudoparticles were transduced into Vero81, Huh7, Calu3, or HAE cultures, with or without prior protease or protease inhibitor exposures. Fluc levels were measured posttransduction.

In Vitro S Fragmentation Assay.

MERS pps and DPP4 pps were incubated in 1:5 M ratios, digested with graded doses of trypsin, and S fragments visualized by Western blotting.

For additional information, see SI Experimental Procedures. See Tables S1–S3 for primers used for MERS-CoV, real-time PCR, and mutant MERS pps.

Table S1.

Primers for generating S1/S2 mutant authentic MERS-CoV

Primer name	Primer sequences
MERS YSAS F	5′ CTCACACCTTACAGTGCGAGCTCTGTTCCA 3′
MERS YSAS R	5′ TGGAACAGAGCTCGCACTGTAAGGTGTGAG 3′
MERS SSVR F	5′ CTCACACCTTCCAGTGTGCGCTCTGTTCCA 3′
MERS SSVR R	5′ TGGAACAGAGCGCACACTGGAAGGTGTGAG 3′
10443pBAC-F5 F	5′ CTCGGGTATGGTCAGGTAATGAACGT 3′
10443pBAC-F5 R	5′ ACGTTCATTACCTGACCATACCCGAG 3′
M13_pUC F	5′ CCTGTGTGAAATTGTTATCCGCTCAC 3′
M13_pUC R	5′ GTGAGCGGATAACAATTTCACACAGG 3′

Table S3.

Primers for generating S1/S2 mutant MERS pps

Primer name	Primer sequences
MERS YSAS F	5′ CTGACCCCTTACTCAGCGTCAAGTGTCCCC 3′
MERS YSAS R	5′ GGGGACACTTGACGCTGAGTAAGGGGTCAG 3′
MERS SSVR F	5′ CTGACCCCTAGTTCAGTGCGAAGTGTCCCC 3′
MERS SSVR R	5′ GGGGACACTTCGCACTGAACTAGGGGTCAG 3′
M13_pUC F	5′ CCTGTGTGAAATTGTTATCCGCTCAC 3′
M13_pUC R	5′ GTGAGCGGATAACAATTTCACACAGG 3′

SI Experimental Procedures

Cells.

HEK293T cells were maintained in 293T media [DMEM, supplemented with 10% (vol/vol) FBS (Atlanta Biologicals), 10 mM Hepes, 100 mM sodium pyruvate, 0.1 mM nonessential amino acids, 100 U/mL penicillin G, and 100 µg/mL streptomycin]. Calu3 cells were maintained in Minimum Essential Media (MEM) supplemented with 20% FBS, 100 U/mL penicillin G, and 100 µg/mL streptomycin. Caco2 cells were obtained from Susan Uprichard of Loyola University Chicago, Chicago, IL, and maintained in Calu3 media. Huh7 cells were obtained from Susan Uprichard and maintained in DMEM supplemented with 10% FBS, 10 mM Hepes, 0.1 mM nonessential amino acids, 100 U/mL penicillin G, and 100 µg/mL streptomycin. Vero81 cells were maintained in DMEM supplemented with 10% FBS, 100 U/mL penicillin G, and 100 µg/mL streptomycin. Cell culture materials and reagents were obtained from Corning and HyClone, unless otherwise noted.

HAE cultures were dissociated into cell suspensions and seeded onto semipermeable collagen-coated membranes with a pore size of 0.4 μm, and maintained in Ultroser G medium at 37 °C under 5% CO₂. Millicell inserts were placed in 24-well plastic cell culture plates (Costar). Twenty-four hours after seeding, the mucosal medium was removed and the cells were allowed to grow at the air-liquid interface as reported previously (53).

Generation and Infection of Recombinant MERS-CoVs.

A MERS-CoV infectious cDNA clone (EMC12 strain) in pBeloBAC11 [designated as pBAC-MERS (32)] was kindly provided by Luis Enjuanes of the Centro Nacional de Biotecnología–Consejo Superior de Investigaciones Cientificas, Madrid, Spain. To introduce mutations in the S gene of pBAC-MERS, three fragments were PCR-amplified from the plasmid pUC-MERS-1 (a pUC plasmid containing the MERS-CoV nucleotides 20,898–25,836), with the mutagenic primers listed in Table S1. The fragments were assembled using Gibson Assembly (New England Biolabs). Sanger sequencing was used to confirm mutations in the assembled constructs. Mutated pUC-MERS-1 fragments were digested with SwaI and PacI and used to replace corresponding SwaI-PacI in pBAC-MERS. To recover infectious virus, Vero81 cells were grown to 95% confluence and transfected with 1 µg/10⁶ cells of the infectious cDNA clone, using Lipofectamine 2000 (Invitrogen). At 5-d posttransfection, cells and media were collected and used to inoculate Vero81 cells. Stock viruses were obtained after one cycle of amplification in Vero81 cells. Infectivities were determined by plaque titration in Vero81 cells.

Cells in 48-well plates were incubated with recombinant WT or S1/S2 mutant MERS-CoVs at multiplicity of infection (MOI) at 2 for 1 h. At 5-h postinfection, cells were rinsed, homogenized in TRIzol, and total RNA was isolated by Direct-zol RNA MiniPrep (Zymo Research). Total RNA (200 ng) was used for cDNA synthesis using High Capacity cDNA Reverse Transcription (Applied Biosystems). Real-time PCR was performed to quantify mRNA expression levels of subgenomic nucleocapsid (N) and HPRT using primers listed in in Table S2 and Power SYBR Green PCR Master Mix (Applied Biosystems). Data were analyzed by ∆∆CT method and shown as relative mRNA levels to WT MERS-CoV infected Vero81 cells. For virus titration, cultured cells and HAE cultures were incubated with WT or S1/S2 mutant MERS-CoVs at MOI = 0.1 for 1 h. At 20-h postinfection, infected cells were frozen and thawed. Infectious viruses in cell extracts or the apical washes of infected HAE cultures were titrated by plaque assay in Vero81 cells.

Table S2.

Primers for real-time PCR

Target mRNA	Primer name	Primer sequences
MERS-CoV N	MERS-Leader	5′ CTCGTTCTCTTGCAGAACTTTG 3′
	MERS-N R	5′ TGCCCAGGTGGAAAGGT 3′
DPP4	DPP4 F	5′ TACAAAAGTGACATGCCTCAGTT 3′
	DPP4 R	5′ TGTGTAGAGTATAGAGGGGCAGA 3′
TMPRSS2	TMPRSS2 F	5′ CTCTACGGACCAAACTTCATC 3′
	TMPRSS2 R	5′ CCACTATTCCTTGGCTAGAGTA 3′
Furin	Furin F	5′ CCTGGTTGCTATGGGTGGTAG 3′
	Furin R	5′ AAGTGGTAATAGTCCCCGAAGA 3′
Cathepsin L	CTSL F	5′ GTGGACATCCCTAAGCAGGA 3′
	CTSL R	5′ CACAATGGTTTCTCCGGTC 3′
Cathepsin B	CTSB F	5′ AGAGTTATGTTTACCGAGGACCT 3′
	CTSB R	5′ GATGCAGATCCGGTCAGAGA 3′
HPRT	HPRT F	5′ CTCGTTCTCTTGCAGAACTTTG 3′
	HPRT R	5′ TGCCCAGGTGGAAAGGT 3′

Plasmid Constructions.

DNA encoding codon-optimized MERS-CoV EMC/2012 S (GenBank accession no. JX869059) and containing a C-terminal C9 tag (pcDNA3.1-MERS S-C9) was purchased from GenScript. The MERS S-C9 was PCR-amplified with mutagenic primers listed in Table S3, and PCR fragments were assembled using In-Fusion HD (Clontech). All constructs were confirmed by Sanger sequencing. Plasmids encoding C-terminal flag-tagged hDPP4 (GenBank accession no. NM_001935, pCMV6-Entry-hDPP4-flag) were purchased from OriGene. Plasmids encoding SARS-CoV S and HCoV-229E S containing a C-terminal C9 tag were obtained from Michael Farzan of The Scripps Research Institute, Jupiter, FL. Plasmids encoding VSV G Indiana, pHEF-VSV G, were provided from BEI Resources. pNL4.3-Luc R- E- was obtained from the NIH AIDS Research and Reference Program, cat. # 3418. Lentiviral packaging plasmid psPAX2 and lentiviral transfer plasmid pReceiver-Lv205-human cathepsin L were purchased from Addgene and GeneCopoeia, respectively. Lentiviral transfer plasmid pLvx-GFP was obtained from Edward Campbell of Loyola University, Chicago, IL.

HIV Pseudoparticle Preparation and Transduction.

HEK293T cells were transfected with pNL4.3-Luc R- E- in conjunction with plasmids encoding MERS-CoV S, SARS-CoV S, HCoV-229E S, VSV G, or DPP4. For transfections, plasmid DNAs were incubated with polyethylenimine (PEI; Polysciences), at 1:3 DNA:PEI ratios, in Opti-MEM (Life Technologies) for 15 min at room temperature, then added dropwise to adherent cells (2 µg DNA per 10⁶ cells). Cell-free supernatants containing the pps were collected at 48-h posttransfection, filtered through 0.45-µm syringe filters (Pall Life Sciences), and stored at −80 °C until use. Where indicated, pps were secreted into SFM, filtered, and stored as described above. For PCI treatment, transfected cells were incubated in SFM containing 50 µM PCI (EMD Millipore) from 6-h posttransfection until harvest. Cell-free supernatants were collected, and viruses were concentrated ∼100-fold by ultrafiltration, using Amicon Ultra-15 centrifugal filter units with Ultracel-30 membranes (EMD Millipore). The concentrated viruses were rediluted back to their original volumes with PCI-free SFM. The residual PCI in the virus preparations was below 0.5 μg/mL, lower than that required to inhibit virus entry.

Target cells were transduced with pps at normalized inputs, based on S protein levels, for 1 h at 37 °C, then cells were washed and further incubated for an additional 48 h. In one experiment, target cells overexpressed TMPRSS2. The Huh7-TMPRSS2 cells were generated by transduction with adenovirus serotype 5 (Ad5)-TMPRSS2 (35). At the end of MERS pp transduction periods, cells were dissolved into cell culture lysis buffer [25 mM Tris-phosphate (pH 7.8), 2 mM DTT, 2 mM 1,2-diaminocyclohexane-N,N,N′,N′-tetraacetic acid, 10% (vol/vol) glycerol, 1% Triton X-100] and luciferase levels were measured by addition of Fluc substrate [1 mM d-luciferin, 3 mM ATP, 15 mM MgSO₄·H₂O, 30 mM Hepes (pH 7.8)] using a Veritas microplate luminometer (Turner BioSystems).

For production of lentiviral vectors encoding cathepsin L, HEK293T cells were transfected with HEF-VSV G, psPAX2, and pReceiver-Lv205-human cathepsin L or pLvx-GFP at a ratio of 4:3:1. Cell-free supernatants containing pps were collected at 48-h posttransfection and filtered through 0.45-µm syringe filters. Calu3 cells were transduced with lentiviral particles and selected using 4 µg/mL puromycin (InvivoGen).

VSV Pseudoparticle Preparation and HAE Transduction.

HEK293T cells were transfected with plasmids encoding MERS-CoV S, VSV G, or empty vector. At 24-h posttransfection, cells were transduced with VSVluc-VSV∆G complemented with Junin virus glycoproteins for 1 h. Cell-free supernatants were collected for 24∼72 h posttransduction and filtered through 0.45-µm filter. The pps in the supernatants were pelleted by centrifugation at 10,000 × g for 10∼18 h, suspended in complete media, and kept at −80 °C until use. HAEs were inoculated with 150× concentrated VSV pps on apical sides. At 18-h posttransduction, cells were lysed and luciferase levels quantified as a measure of VSV pp entry.

Protease Inhibitors.

Target cells used for transduction were exposed to 100 µM camostat, 50 µM PCI, or 10 µM E64d (all from Sigma-Aldrich) from 1 h before to 1 h after pp inoculations. DMSO served as vehicle controls. At 1-h posttransduction, inhibitor-free media were applied.

Proteases.

MERS pps prepared in SFM were incubated with 3 µg/mL trypsin (EMD Millipore) for 10 min at 37 °C. SARS and 229E pps prepared in SFM were incubated with 10 µg/mL trypsin for 10 min at 37 °C. Trypsin was inactivated with 100 µg/mL soybean trypsin inhibitor (STI; Sigma-Aldrich) for 5 min at room temperature. MERS pps were incubated with cells for 1 h at 4 °C, rinsed, and treated with 10 µg/mL cathepsin L (Calbiochem) at pH 6.0 for 10 min at 37 °C.

In Vitro S Fragmentation Assay.

MERS pps and DPP4 pps prepared in SFM were incubated in 1:5 M ratios for 1 h at 4 °C. The mixtures were digested with graded doses of trypsin for 10 min at 37 °C, and digestions were halted with 100 µg/mL STI.

Western Blotting.

MERS pps were pelleted through 20% (wt/vol) sucrose in HNE (20 mM Hepes, 100 mM NaCl, 1 mM EDTA) by centrifugation at 80,000 × g for 2 h at 4 °C. MERS-CoV–infected cells were lysed in Nonidet P-40 lysis buffer [1% Nonidet P-40, 50 mM Tris⋅HCl (pH 7.4), 150 mM NaCl, 1 mM EDTA] on ice and cleared by centrifugation at 1,000 × g for 10 min at 4 °C. Pelleted pps and infected cell lysates were mixed with SDS solubilizer to final concentrations of 0.0625 M Tris⋅HCl (pH 6.8), 10% glycerol, 0.01% bromophenol blue, 2% (wt/vol) SDS, 1% 2-mercaptoethanol. Samples were heated at 95 °C for 5 min, separated in 10% (wt/vol) polyacrylamide-SDS gels, transferred to PVDF membranes, probed with polyclonal mouse anti-MERS S (35), monoclonal mouse anti-C9 (EMD Millipore), or monoclonal mouse anti-p24 antibodies (Abcam), or HRP-conjugated anti–β-actin (Sigma-Aldrich). Membranes were then probed with horseradish peroxidase-conjugated goat anti-mouse IgG (Perkin-Elmer), incubated with ECL substrate (Thermo Fisher Scientific), and signals detected with FluorChem E (ProteinSimple).

Real-Time RT-PCR.

Total cellular RNA was isolated from Huh7, Calu3, and HAE cells by using the RNeasy Mini Kit (Qiagen). Total RNA (500 ng) was used for cDNA synthesis using RevertAid RT kit reagents (Thermo Fisher Scientific). Real-time PCR was performed to quantify mRNA expression levels of DPP4, TMPRSS2, furin, cathepsin L, cathepsin B, and HPRT, using the primers described in Table S2, and RT² SYBR Green qPCR Mastermix (Qiagen). Data were analyzed by ∆∆CT method.

Statistical Analyses.

All experiments were independently repeated at least three times. Data were presented as mean ± SD. Statistical significance was calculated using the Holm–Sidak multiple Student’s t test procedure. A P value of <0.05 was considered statistically significant.