Proteolysis of the Ebola surface glycoprotein by cathepsins in the host cell endosome is a critical step for viral entry (5
). In this study, we characterized the products generated by CatL cleavage of ZEBOV-GP and examined their relevance to entry and membrane fusion. Using N-terminal sequencing and mass spectrometry analyses of CatL-cleaved ZEBOV-GP, we identified three CatL cleavage sites, confirmed the furin cleavage site, and observed three cleavage products. The first CatL cleavage occurs at Glu201 between β-strands 13 and 14, the second between Ser210 and Ala222, most likely after β-strand 14, and a third potential site may occur close to Gly647 near the C-terminal end of GP2 but is likely protected by the close proximity of the viral membrane. The known furin cleavage was also observed at Glu502. The first two cleavages result in the removal of the membrane distal “glycan cap” and the entire MUC domain. This confirms a hypothesis by Lee et al., who predicted a CatL cleavage site within the disordered β13-14 loop (residues 190 to 213) based on their crystal structure of the GP ectodomain (10
). However, in contrast to thermolysin cleavage of EBOV-ZEBOV-GP, it appears that the outer strand of the head, β-strand 14, remains as part of the trimer after CatL cleavage. Two cleavage products were observed by MS under nonreducing conditions (43 kDa and 4 kDa), and three main products were observed under reducing conditions (23.5, 19.4, and 4 kDa). These products result from disulfide-bonded GP1 and GP2 and a non-covalently associated 4-kDa fragment. Most of the extensive glycosylation of GP is removed by CatL cleavage, with MS analysis of PNGaseF-treated, CatL-cleaved GP showing that only 7.9 kDa of glycosylation per GP1:GP2 monomer remains after cleavage.
A recent study by Dube and coworkers characterized ZEBOV-GP cleavage and priming by thermolysin, a nonphysiological metalloprotease (6
). They observed a single cleavage site between residues 190 and 194, generating a 19-kDa GP1 fragment beginning at residue 33, with the GP2 fragment extending up to the transmembrane domain. However, we observe the more physiological CatL cleavage at two different locations. It is not surprising that we observed a different cleavage pattern, since CatL and thermolysin are members of completely different protease families (cysteine and metalloprotease, respectively) and function at different pHs (5.5 and 7.5, respectively). Whether further processing of the CatL-cleaved GP by CatB generates a cleavage site at residue 190 remains to be determined in future studies. Dube and coworkers also predicted a 20-kDa fragment, likely to include residues 33 to 200. The difference between the reported 20-kDa fragment and our 23.5-kDa fragment is likely due to the contribution of an N-linked glycosylation site contained within the GP1 N-terminal fragment. Nevertheless, modeling or predictions of the processed intermediate structure by us and others agree that the MUC domain, the glycan cap, and β-strand 15 of the head domain are removed upon CatL processing.
CatL cleavage of the ZEBOV-GP trimer is required for viral entry (5
). One way in which CatL likely facilitates entry is that removal of the glycan cap domain and the MUC domain and extensive glycosylation provide much greater access to the receptor-binding domain. This is evident from studies showing higher levels of cell binding of Ebola virus upon CatL treatment (7
) and is also supported by our modeling of the CatL-cleaved GP. As CatL cleavage optimally occurs in the acidic milieu of the endosome, it is also possible that a direct interaction between CatL-cleaved EBOV-GP and a cellular receptor may take place in the endosome; however, future studies are required to determine the cellular receptor(s) and site of binding. The EBOV GP1 residues critical to virus entry have been proposed by multiple groups (2
), with the recently reported ZEBOV-GP trimer structure identifying a cluster of six residues (K114, K115, K140, G143, P146, and C147) located in an approximately 20- by 15-Å footprint on the inner surface of the GP1 trimer (10
). These six residues, when mapped onto our CatL-processed GP model, become completely exposed on the surface of the trimer intermediate. Furthermore, these six residues are conserved among all EBOV species, and two of the six residues are conserved in Marburg virus strains (Fig. ). Recently, Dube and coworkers employed mutational studies to confirm the importance of residues K114, K115, and K140 and also identified K95 as another critical residue (6
). However, receptor binding was not completely obliterated by the combined mutations of these four lysine residues, indicating that other residues of GP1 may also be involved in EBOV entry (6
Our study is the first to report the immunological properties of the CatL-cleaved ZEBOV-GP intermediate. Importantly, the cleaved ZEBOV-GP elicited 3-fold more potent neutralization activity in the sera of vaccinated mice, slightly greater than that of uncleaved EBOV GP (Fig. ). CatL cleavage, through exposure of the RBD and residues critical for virus entry, likely facilitates access of neutralizing antibodies to the GP receptor-binding site. We employed uncleaved EBOV-GP pseudovirus in evaluating the neutralizing activity of mouse sera following immunization, principally to mimic the initial interaction between EBOV GP and the immune system. The neutralizing sera elicited by immunization with CatL-cleaved EBOV GP may well have been more potent had matched CatL-treated pseudovirus been used to analyze the neutralizing activity. Furthermore, the polyclonal neutralizing immune serum response was generated against the entire CatL-cleaved EBOV-GP surface; thus, a monoclonal antibody directed specifically against the conserved core residues may provide a broader neutralizing response in Ebola species than an antibody directed to regions with higher sequence variability.
Our results also indicate that binding to the neutralizing antibody KZ52 was not disrupted by CatL cleavage. This strongly argues that another cathepsin, CatB, and/or secondary factors are likely required to trigger fusion peptide exposure and GP-mediated membrane fusion (Fig. ). This proposal is supported by two differing models of GP-mediated fusion: that of Chandran et al., who propose that CatL and B processing alone is sufficient to trigger membrane fusion, and that of Schornberg et al., who propose that CatL and CatB as well as additional secondary factors are required to trigger the membrane fusion event. However, all reported models and our study agree that CatL processing alone is insufficient to trigger and initiate the required conformational change that leads to membrane fusion. A recent study examining cathepsins in the endosome suggested that α5
integrins play a role in regulating the activation of cathepsins and may thus be important for EBOV virus entry (15
). Interestingly, the regulation was shown to occur at the level of cathepsin processing, after binding and prior to membrane fusion (15
FIG. 7. Schematic model of ZEBOV-GP entry and fusion events. Attachment and binding of the virion (purple) to the host cell membrane is the first step required for EBOV-GP virus entry and replication, followed by endocytosis of the virion into vesicles. The critical (more ...)
This study provides the first reported specific CatL-cleavage sites in ZEBOV-GP and demonstrates that exposure of the conserved core of GP likely facilitates receptor binding and elicits a potent neutralizing antibody response. Our results confirm that these conserved determinants are required for viral entry and suggest that CatB and/or other secondary factors are involved in triggering the conformational changes that lead to EBOV GP-mediated membrane fusion.