|Home | About | Journals | Submit | Contact Us | Français|
Ebola viruses (EBOV) are causative agents of lethal hemorrhagic fever in humans and non-human primates. Among the filoviruses characterized thus far, Reston Ebola virus (REBOV) is the only Ebola species that is non-pathogenic in humans despite the fact that REBOV can cause lethal disease in non-human primates. Previous studies also suggest that Reston EBOV is less effective at inhibiting host innate immune responses, compared with Zaire EBOV or Marburg virus. Virally encoded VP35 protein is critical for immune suppression, but an understanding of the relative contributions of VP35 proteins from REBOV and other filoviruses is currently lacking. In order to address this question, we characterized REBOV VP35 IFN inhibitory domain (IID) using structural, biochemical, and virological studies. These studies reveal differences in dsRNA binding and IFN inhibition between the two species. These observed differences are likely due to increased stability and loss of flexibility in REBOV VP35 IID as demonstrated by thermal-shift stability assays. Consistent with this finding, our 1.71 Å crystal structure of the REBOV VP35 IID reveal that the structure is highly similar to ZEBOV VP35 IID with an overall backbone r.m.s.d. of 0.64 Å, but contains an additional helical element at the linker between the two subdomains of VP35 IID. Mutations near the linker, including swapping sequences between REBOV and ZEBOV, reveal that the linker sequence has limited tolerance for variability. Together with the previously solved ligand free and dsRNA bound forms of ZEBOV VP35 IID structures, our current studies of REBOV VP35 IID reinforce the importance of VP35 in immune suppression. Functional differences observed between REBOV and ZEBOV VP35 proteins may contribute to observed differences in pathogenicity, but these are unlikely be the major determinant. However, the high similarity in structure and the low tolerance of sequence variability, coupled with the multiple critical roles played by EBOV VP35 proteins, highlight the viability of VP35 as a potential target for therapeutic development.
Filoviruses, including Ebola virus (EBOV) and Marburg virus (MARV), are enveloped negative-sense RNA viruses associated with zoonotic infections in humans1; 2; 3; 4. Only a single species of MARV (Lake Victoria) has been identified thus far, but currently five known species of EBOV have been reported, including Zaire, Sudan, Ivory Coast, Bundibugyo and Reston1; 2; 3; 4. Reston Ebola virus (REBOV) is the only known non-pathogenic filovirus to humans. The ability to cause lethal disease in humans and non-human primates has resulted in outbreaks primarily due to Zaire EBOV (ZEBOV) with human fatality rates near 90 percent4. In contrast, only a few documented cases of human infection with REBOV have been documented. Moreover, these infections were not associated with illness or death, suggesting that REBOV may be attenuated and likely avirulent in humans. The recent discovery of the presence of REBOV in samples collected from domestic swine in the Philippines5; 6; 7 suggests that REBOV has a greater potential to enter the human food chain than previously recognized. Comparison of global cellular transcriptional responses to ZEBOV, REBOV, and MARV suggest that MARV is less efficient than other EBOVs in inhibition of host cell IFN responses8, although interpretation of these cellular responses is complicated by the fact that REBOV replicates more slowly in cell culture than ZEBOV9. However, the molecular basis for species variation among EBOVs is currently lacking. Therefore, studies directed at examining structural and functional comparisons may shed light on this highly lethal family of native-stranded RNA viruses and suggest novel antiviral strategies against EBOVs.
Among the seven structural proteins encoded by the negative single stranded RNA genome of Ebola virus, VP24 and VP35 display immune antagonism. In particular, VP24 functions through inhibition of the JAK/STAT pathway, while VP35 functions to inhibit IFN-α/β signaling10; 11; 12. Additionally, VP35 is critical for RNA-dependent protein kinase (PKR) inhibition13; 14, viral replication15; 16, and RNA silencing suppression17. As shown in Fig. 1a, VP35 contains an N-terminal coiled-coil domain that is important for oligomerization18; 19 and a C-terminal interferon inhibitory domain (IID) that binds dsRNA15; 20; 21; 22. The coiled-coil domain is required for several VP35-mediated functions, including viral replication and nucleocapsid formation18; 19; 23; 24; 25; 26 as well as immune suppression, as the EBOV VP35 IID is unable to suppress IFN induction to the same level as the full length protein19; 20. However, we and others have shown that VP35 binds dsRNA through the C-terminal IID, which alone is sufficient for interferon (IFN) inhibition15; 20; 21; 22. Our recent crystal structure of the ZEBOV VP35 IID revealed a cluster of conserved basic residues that are important for dsRNA binding27. From this first structure of a VP35 IID and associated solution NMR experiments, we demonstrated that VP35 IID consists of two subdomains that form a single independently folded unit. The α-helical subdomain forms a four helix bundle whereas the β-sheet subdomain forms a 4 stranded β-sheet with a short helix. Examination of the structure identified two distinguishable charged patches, termed first and central basic patch. Interestingly, this region was previously identified as a critical element for IFN inhibition and many residues within these charged patches display high sequence similarity among members of the filoviral family (Fig. 1a) 21. Mutation of R312 in ZEBOV VP35 IID (corresponding to R301 in REBOV VP35 IID), which is located at the center of the central basic patch, prevents dsRNA binding and display attenuated IFN inhibition. Several studies have extensively characterized mutations from the ZEBOV central basic patch and demonstrated the functional importance of residues within this patch16; 21; 15; 22. Consistent with these reports, recent mouse-28 and guinea pig-adapted29; 30 ZEBOV viruses show that growth rates of R312A and K319A/R322A mutation were greatly attenuated compared to viruses with wildtype VP3531. Moreover, K319A/R322A mutant virus was avirulent in a lethal guinea pig model30. High sequence similarity, including the presence of identical residues at key sites, within the central basic patch suggests that VP35 proteins in all EBOVs may function is a similar manner9; 20.
Previous studies have shown that nearly all components from the REBOV and ZEBOV nucleocapsid (viral polymerase L, NP, VP30, and VP35) are interchangeable in a minigenome assay that mimics viral replication9. Only the combination of ZEBOV VP35 and REBOV viral polymerase L displayed diminished replication function9. Therefore, despite significant differences in their virulence in humans, limited information about functional comparisons between REBOV and ZEBOV gene products is currently available32. Our studies described below address these questions in the context of critical VP35 proteins through combined biochemical, structural, and virological studies by comparing the VP35 IID proteins from REBOV and ZEBOV. Biochemical and virological studies show that VP35 IID proteins from REBOV and ZEBOV function similarly, but display different structural stabilities. Consistent with these findings, our crystal structure of REBOV VP35 IID, solved to 1.71 Å resolution, reveal similar folds, with a single difference at the interface of the α-helical and β-sheet subdomains. While this difference is likely responsible for the higher structural stability observed for REBOV VP35 IID, structural and functional studies show that only sequence swapping between REBOV and ZEBOV near the linker are tolerated as additional mutations result in loss of structure and function. While the observed differences in structure, dsRNA binding, and IFN inhibition likely result from loss of conformational plasticity, our results suggest that VP35 proteins from REBOV and ZEBOV probably function in a similar manner. Therefore, while VP35 may contribute to observed pathogenic differences, it is unlikely to be the main determinant. Given the structural and functional similarities, coupled with a low tolerance for mutations at the subdomain interface, our results suggest that VP35 IID is viable candidate for antiviral development and the availability of a high resolution crystal structure from our study should facilitate these efforts.
One of the hallmarks of EBOV infections is the inhibition of host immune responses that impair innate and adaptive immunity. In vivo studies comparing REBOV and ZEBOV infections show differences, suggesting marked differences in the immune suppression capacities. Although REBOV and ZEBOV show very different disease outcomes in humans, examination of the corresponding VP35 IID sequences reveal that only 7 residues are different between the two proteins within the IID region (Fig. 1a). Since, VP35 is an important EBOV encoded virulence factor responsible for immune suppression and evasion, we examined the functional properties of REBOV VP35 IID. Protein samples were prepared by modifying previously published protocols for ZEBOV VP35 IID33. Given the high sequence homology, we were able to identify constructs that were well behaved based on hydrodynamic characterization, including size exclusion chromatography and dynamic light scattering (data not shown). Based on these studies, we identified a REBOV VP35 IID construct containing residues 204–329, which corresponds to ZEBOV VP35 IID residues 215–340.
We recently determined the dsRNA bound structure of ZEBOV VP35 IID, which revealed that the VP35 IID protein uses multiple strategies to engage in protein-protein and protein-RNA interactions in order to facilitate viral infection and replication22. We also demonstrated that dsRNA binding is important for VP35-mediated antagonism of host immune responses that originate from cytosolic RIG-I like receptors (RLRs). As an initial test of the REBOV VP35 characterization and comparison, we assessed the ability of REBOV VP35 IID to bind 8 base pair (bp) dsRNA by isothermal titration calorimetry (ITC) (Fig. 1b). Both VP35 IIDs bound 8 bp dsRNA with a stoichiometry of 1:4 ratio of dsRNA:VP35 IID. However, comparison of binding constants revealed that REBOV VP35 IID displayed approximately 2–3 fold lower dsRNA binding for the same dsRNA when compared to ZEBOV VP35 IID. Moreover, the relative heats released upon binding were ~25% lower for REBOV VP35 IID than ZEBOV VP35 IID titrations, indicating that dsRNA interactions with REBOV and ZEBOV VP35 IIDs may not be identical despite the highly similar primary sequence, particularly near the central basic patch.
We next tested the inhibition of Sendai virus mediated activation of the IFN-β promoter by REBOV and ZEBOV VP35 proteins. As shown in Fig. 1c, both REBOV and ZEBOV VP35 proteins can inhibit IFN-β promoter activation, but REBOV VP35 inhibits this activation to levels lower than that observed for ZEBOV VP35. Examination of corresponding western blots indicate that the expression levels of REBOV and ZEBOV VP35 proteins are similar, suggesting that the differences observed in dsRNA binding and IFN inhibition may result from the reduced ability of REBOV VP35 to sequester dsRNA leading to higher levels of IFN promoter activation.
To further examine the physical basis for the difference in REBOV and ZEBOV VP35 function, we conducted Thermofluor assays34; 35 with REBOV and ZEBOV VP35 IID proteins. Using a temperature gradient from 30 to 90 °C, we monitored the changes in fluorescence upon SYPRO Orange binding to unfolded hydrophobic regions. Analysis of the derivative of the fluorescence change, shown in Fig. 1d, revealed that the average melting temperature (Tm) for REBOV VP35 IID is 63.4 ± 0.3 °C while ZEBOV VP35 IID displayed a Tm value of 57.0 ± 0.1 °C. Given that the reported values are an average of at least 15 independent measurements, our results suggest that the REBOV VP35 IID is thermally more stable than ZEBOV VP35 IID by at least 6 °C. However, the source of the difference in thermal stability is not immediately evident as these two proteins only differ at 7 residues in the region encoded for VP35 IID (Fig. 1a).
In an effort to structurally characterize the REBOV VP35 IID protein, we determined the crystal structure of REBOV VP35 IID containing residues 204–329. Initial crystallization conditions were obtained from a commercial crystal screen (Hampton Research) and these conditions were subsequently optimized to obtain single crystals that were suitable for diffraction data collection at the synchrotron. Our data collection statistics are shown in Table 1. Data reduction and indexing revealed that the REBOV VP35 IID crystals belong to the P41 space group, with unit cell dimensions of a=b=55.55, c=50.87 and α=β=γ=90. The asymmetric unit contained one molecule of REBOV VP35 IID and the Matthews coefficient was calculated at 2.55 with a solvent content of 51.72%.
The structure of REBOV VP35 IID was determined by molecular replacement using our previously determined structure of ZEBOV VP35 IID as a search model (PDB: 3FKE) 27 (Fig. 2a). REBOV VP35 IID shows remarkable similarity to ZEBOV VP35 IID (backbone r.m.s.d. value of 0.64 Å) (Fig. 2b). Overall, the REBOV VP35 IID structure retained the α-helical subdomain and the β-sheet subdomains present in the ZEBOV VP35 IID structure, which aligned with r.m.s.d. values of 0.3 Å and 0.5 Å for the α-helical and β-sheet subdomains, respectively27. These results suggest that other filoviral VP35 IID proteins are likely to contain a similar fold and, more importantly, that the VP35 IID fold is substantially different from the canonical dsRBD fold of αβββα. Comparisons using the DALI server35 reveal that the α-helical subdomain has a topology similar to many functionally unrelated proteins (DALI z-score >2). However, like the ZEBOV IID structure, we were unable to identify other known structurally or functionally related proteins, confirming the uniqueness of the REBOV VP35 IID fold.
Previous studies have shown that mutation of individual residues within the central basic patch results in loss of dsRNA binding and a concomitant loss of IFN inhibition. Initial solution NMR analysis revealed that mutant VP35 proteins retain their overall fold. In contrast, recent crystal structures of three ZEBOV VP35 IID central basic patch mutants, R312A (PDB: 3L27), K319A/R322A (PDB: 3L29), and K339A (PDB: 3L28), showed that substantial changes to the electrostatic surface are evident as a result of mutating charged residues. However, in these mutant structures, only subtle changes to the overall protein conformation are observed. Comparison of the electrostatic surface of the REBOV and ZEBOV IID structures, shown in Fig. 2c, confirm our predictions that residues previously known to be important for dsRNA binding in ZEBOV VP35 IID are forming a similar charged surface in REBOV VP35 IID.
Despite the remarkable similarities at the level of global structure, we observe one notable difference between the REBOV and ZEBOV VP35 IID structure at the subdomain interface. In the REBOV VP35 IID structure, the linker between the α-helical and β-sheet subdomains forms an additional short 5 residue alpha helix in the REBOV VP35 IID structure (α4b in Fig. 2a). This helix is absent in the ZEBOV VP35 IID structure and is replaced by a loop. We have previously demonstrated that the subdomain interface, formed by highly conserved residues, is critical for folding and stability of ZEBOV VP35 IID 27. It is likely then that the presence of the α4b helix provides additional stabilization of the REBOV VP35 IID structure near the subdomain interface and contributes to the increased thermal stability observed in the Thermfluor assays. As a consequence of this helix, the relative orientation between the α-helical and β-sheet subdomains is rotated by 1.0–4.0° when the structures of dsRNA-VP35 complex (PDB: 3L25, 3L26) as well mutant structures are compared by aligning the α-helical subdomain (or the β-sheet subdomain) (Fig. 3a–b). Moreover, we observed that the ZEBOV VP35 IID was able to bind both dsRNA backbone and the blunt dsRNA ends forming an “end cap” and this interaction also depends on the ability of the penultimate Lysine residue (Lys 328 in REBOV or Lys339 in ZEBOV) to coordinate the terminal caboxylate group. As shown in Fig. 3c, charge neutralization of the terminal carboxylate is less efficient in REBOV VP35 IID due to the relative orientation (Fig. 3c). Thus, the helical nature of the linker in REBOV VP35 IID likely limits the conformational flexibility required for multiple interactions between VP35 IID and dsRNA, such as those observed in the ZEBOV IID/dsRNA complex 22.
In order to further assess the differences between the REBOV and ZEBOV VP35 IID linker regions and to identify structural characteristics that may be functionally relevant, we generated a series of mutations for REBOV and ZEBOV VP35 IID near the linker region. These include mutation of the linker sequence to a flexible linker (linker 1, Table 2), swapping of the linker region between ZEBOV and REBOV (linker 2, Table 2), and insertion of a flexible linker (linker 3, Table 2). As shown in Fig. 4, swapping the linker sequence between REBOV and ZEBOV IID did not affect overall thermal stability (Fig. 4a), structure (Fig. 4b, left), or function (Fig. 4c). However, substitution of the linker sequences (linker 1) or insertion (linker 3) with a linker composed of Gly and Ser residues resulted in highly destabilized VP35 IID domains. All linker 1 constructs were highly unstable when expressed in E. coli under conditions similar to wildtype protein. Although we were able to purify the linker 3 proteins, they show diminished stability (Fig. 4a, compare black (wildtype) with blue (linker 3)) and a large number of structural changes, highlighted for REBOV linker 3 (Fig. 4b, right). Consistent with these observations, expression of both linker 1 and linker 3 in the context of full length REBOV and ZEBOV proteins displayed diminished IFN inhibitory activity in the Sendai virus infection assay (Fig. 4c). In contrast, even at lower plasmid concentrations than those in Fig. 4d, linker 2 constructs show wildtype behavior. Together, the results described above highlight the importance of the linker sequence connecting the two subdomains in VP35 IID to activity and demonstrate that EBOV VP35 sequences have very low sequence tolerance at the subdomain interface.
The critical role played by VP35 in immune suppression and the importance to viral replication and pathogenesis of Ebola viruses is well established. We have recently provided additional experimental evidence that links the ability of ZEBOV VP35 to bind dsRNA and its ability to antagonize host immune responses22. Consistent with these observations, recent reports show that antiviral pathways initiated by cytosolic RIG-I-like helicases are important for EBOV infections36; 37; 38; 39; 40; 41; 42 as preactivation of RIG-I alone can reduce EBOV infections up to ~1000-fold43. Therefore, inhibition RIG-I like receptor (RLR) function, which detect viral nucleic acids such as dsRNA during infection and replication, is likely important for EBOV propagation. Comparison of EBOV VP35 IID structures with those of other viruses suggests that VP35 is significantly different from other viral and cellular proteins, but these properties are highly conserved among filoviral VP35 proteins. Therefore, the availability of multiple structures of Ebola VP35 IID proteins, including the REBOV VP35 IID structure described here, coupled with the high sequence similarity among VP35 from different filoviruses and the low tolerance for sequence variation provides unique information that will facilitate new opportunities for therapeutic and diagnostic drug design.
We thank ISU Biotechnology Facilities and Drs. J. Hoy for providing access to instrumentation and support. We also thank P. Ramanan, J. Binning, L. Tantral, D. Peterson, and C. Brown for lab assistance; and Drs. S. Ginnell, N. Duke, F. Rotella, M. Cuff, and J. Lazarz at APS Sector 19. Use of Argonne National Laboratory Structural Biology Center beamlines at the Advanced Photon Source, was supported by the U.S. D.O.E. under contract DE-AC02-06CH11357. This work is supported in part by NIH grants (1F32AI084324 to D.W.L., R01AI059536 to C.F.B., and R01AI081914 to G.K.A.); by the German Research Foundation (SFB 535 to E.M.); MRCE Developmental Grant (U54AI057160-Virgin(PI) to G.K.A.); Roy J. Carver Charitable Trust (09-3271 to G.K.A.).
Accession numbers: Coordinates and structure factors for Reston Ebola VP35 IID protein has been deposited in the Protein Data Bank under the PDB ID: 3L2A.
Note added during manuscript preparation: While this manuscript was in preparation, a structure of the Reston Ebola VP35 IID free and bound to 18 bp dsRNA has appeared online and based on the published images, the two structures appear similar48. However, we were unable to compare as the coordinates for these structures were not released at the time of initial submission.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.