Over the past 6 years, significant progress has been made in understanding how filoviruses gain entry into cells and a model of infection based on this progress has been proposed (10
). EBOV-May particles attach to lectins on the cell surface and are taken up by macropinocytosis into vesicles that are transported to LE/LY-containing endosomal cysteine proteases and NPC1 (5
). One function of endosomal cysteine proteases is to cleave the carboxyl-terminal “cap” region of GP1 from the mushroom-shaped GP protruding from the virus membrane and expose the stalk containing the protease-resistant N-terminal domain in GP1 that is the ligand for the NPC1 protein (4
). A second function may be to biochemically destabilize the GP, thereby sensitizing it to triggering for viral membrane fusion (3
). Further studies are needed to determine the roles of NPC1 (5
) and additional events, including further cleavage of GP1 in this process (3
In this report, we provide evidence that key aspects of this scheme are conserved among other filoviruses. The new findings show that as in EBOV-May, the carboxyl-terminal domain of SUDV GP1 is a substrate for proteolytic cleavage, that cleaved particles are infectious and NPC1 dependent, and that the protease-resistant N-terminal domain of GP1 binds to purified LE/LY membranes in an NPC1-dependent manner. These findings are consistent with the recent report showing that the domain organization of EBOV-May is conserved in SUDV (11
). Moreover, we find that the sensitivity of the C-terminal domain of GP1 to cleavage and the dependence of cleaved particles on NPC1 is conserved among other ebolaviruses. Consistent with this view, we find that isolates from each species of Filoviridae
are E-64 sensitive, including the growth of EBOV-May, SUDV, and MARV. While more work needs to be done, including studies of other proteases and MARV, the results of the experiments described in this report, coupled with alignment of primary amino acid sequences of GPs which indicate that the domain structure is likely to be conserved (data not shown), suggest that cleavage and binding are key steps in the filovirus entry pathway. In this model, endosomal cysteine proteases are required for the efficient removal of the carboxyl-terminal domain of GP1 to expose the NPC1 binding domain. Endosomal cysteine proteases may mediate the additional steps necessary for orderly deployment of the virus membrane fusion activity (21
). Indeed, two recent reports suggest that cysteine protease cleavage of the β-13-14 loop in GP1 promotes release of the GP2 fusion peptide (3
). The studies in this report provide the basis for future studies to compare the virus requirements for specific proteases to cleave GP1, to bind to NPC1, and to release the GP2 fusion peptide.
The endosomal cysteine proteases are a family of 11 acid-dependent proteases that reside in LE/LY (15
). Little is known about the roles individual members of this family play during filovirus infection. Previously, we showed that EBOV-May requires cathepsin B, but not cathepsin L, for entry (7
). We confirm this finding and show that CIEBOV has the same requirement. Additional studies are needed to determine if the virus's sensitivity to the absence of cathepsin B activity is due to a specific requirement for the double-chain isoform of cathepsin B (36
). Although cathepsin L is not essential for any of the filoviruses studied, we found that it works in concert with cathepsin B to enhance the infection of EBOV-May, CIEBOV, and RESTV. Furthermore, cathepsin L activity is required for MARV infection of MEF cells but not Vero cells, suggesting that the role of cathepsin L may be shared by other cysteine proteases with endopeptidase activity. Remarkably, RESTV infection is sensitive to E-64 but is not sensitive to the loss of cathepsins B and L. This suggests that one or more additional E-64-sensitive proteases can support RESTV infection.
The substrate specificity of endosomal cysteine proteases is governed largely by the accessibility of the polypeptide chain to the active site of the protease and not by a strong preference for specific sequence motifs (32
). Thus, one consequence of the extensive variation in the sequence of the carboxyl-terminal domain of GP1 is that it may alter the repertoire of cysteine proteases that are able to cleave GP1. Therefore, the presence of multiple proteases with overlapping substrate preferences in late endosomes and lysosomes of host cells may provide for redundancy in the conditions for cleavage of GP and might explain the virus-specific differences in dependence on cathepsins B and L observed in our studies. One advantage of this scheme may be that effective cleavage of the GP1 cap and/or fusion peptide release is maintained in the presence of selective pressure from host immune recognition for sequence diversification, analogous to the function of variable loops in HIV gp120 (2
). In this model, adaptation to loss of cathepsin B activity by a change to a single amino acid residue (i.e., D47 or I584) may provide a means for a rapid response to changes in endosomal cysteine protease expression between hosts or cell types. Sequence polymorphisms that specifically predict host factor preference have been identified in other virus envelope GPs, including SARS GP N479K interactions with human or civet ACE2 receptor, influenza virus HA1 interactions with α-2 glycan linkage in human or α-2-6 glycan linkage in avian sialic acid, and HIV gp120 binding to receptor CXCR4 and/or CCR5 (2
). Analysis of the filoviruses identified in future outbreaks will provide further tests of the utility of using the D47/I584 polymorphisms in GP in determining virus preference for host endosomal cysteine proteases.