(C) PLOS One
This story was originally published by PLOS One and is unaltered.
. . . . . . . . . .



Ebola virus VP35 interacts non-covalently with ubiquitin chains to promote viral replication [1]

['Carlos A. Rodríguez-Salazar', 'Department Of Microbiology', 'Immunology', 'University Of Texas Medical Branch', 'Galveston', 'Texas', 'United States Of America', 'Molecular Biology', 'Virology Laboratory', 'Faculty Of Medicine']

Date: 2024-03

To further confirm direct, non-covalent, binding between Ub and VP35 and to identify the type of polyUb chains involved in these interactions, a cell-free in vitro binding assay with purified Flag-VP35 and recombinant purified unanchored K48- or K63-Ub chains (a mix of 2 to 7 Ub chains) was conducted. We found that VP35 strongly interacts with unanchored K63- but not K48-Ub chains ( Fig 1B ). Furthermore, unanchored K63-polyUb chains interacted mostly with the C-terminal IID of VP35 ( Fig 1C ). Finally, since we previously showed that VP35 K309R or K309G mutants lose covalent ubiquitination on the K309 residue [ 6 ], we asked whether these mutants can still bind unanchored Ub. We found that these mutants interacted with unanchored Ub at similar levels compared to WT VP35 ( Fig 1D ), suggesting that non-covalent interactions with Ub do not require covalent ubiquitination on the K309 residue.

(A) WCEs from HEK293T cells transfected with Flag-VP35 (VP35) and HA-Ub WT or HA-Ub ΔGG (cannot conjugate proteins) were used for HA IP under non-denaturing conditions (RIPA washes), followed by IB. (B) Purified recombinant K48 or K63 polyUb chains (mix of 2–7 Ub chains) were mixed in vitro with Flag-VP35, followed by Flag IP. Interacting proteins were eluted with Flag peptide. (C, D) Experiments performed as in (B) but using the C-terminal IID domain of VP35 (C), or the VP35 K309R or K309G mutants, which are not covalently ubiquitinated (D). * No ubiquitinated VP35, possibly phosphorylation. The data underlying the graphs shown in the figure can be found in S1 Data . EBOV, Ebolavirus; IB, immunoblot; IID, IFN-inhibitory domain; IP, immunoprecipitation; WCE, whole-cell extract; WT, wild type.

We previously reported that VP35 is covalently ubiquitinated on K309 using co-immunoprecipitation (co-IP) assays. Intriguingly, these experiments also consistently showed a non-modified fraction of VP35 that co-immunoprecipitated with Ub, suggesting a non-covalent interaction between Ub and VP35 [ 4 ]. In this new work, we postulate that VP35 interacts with unanchored or free polyUb chains and that the interaction between these Ub chains and VP35 is functionally relevant. Here, to first confirm binding between VP35 and Ub, we performed a co-IP assay in which we pulled down ectopically expressed wild type (WT) Ub or an Ub mutant lacking the C-terminal di-glycine residues (HA-Ub-ΔGG), which renders Ub unable to form covalent linkages. The terminal -GG on Ub is required for the formation of covalent conjugates of Ub with other proteins [ 37 , 38 ]. This approach allows testing non-covalent interactions between monomeric, non-conjugated Ub, and VP35. Consistent with our previous observation, in the presence of WT Ub, multiple migrating forms corresponding to the molecular weight of covalently ubiquitinated VP35, as well as monomeric VP35 (non-covalent interaction with Ub), were detected by immunoblot (IB). In contrast, monomeric, non-conjugated VP35, co-immunoprecipitated with HA-Ub-ΔGG ( Fig 1A and IP). As expected, the HA-Ub-ΔGG runs at the predicted molecular weight of monomeric Ub (approximately 8.5 kDa) and is unable to form the characteristic smear corresponding to cellular ubiquitinated proteins ( Fig 1A , whole-cell extract [WCE]). This indicates that VP35 associates non-covalently with Ub, either directly or indirectly.

(A) WCE from HEK293T cells transfected with His-IsoT WT, His-IsoT C335A, VP35 WT, and HA-Ub were used for IP with anti-HA beads. (B) Polymerase minigenome assay. HEK293T cells transfected with a monocistronic firefly luciferase-expressing minigenome, including VP30, L, and REN-Luc/pRL-TK, in the presence or absence of IsoT-WT or C335A mutant. Data are expressed as mean + SEM of 3 independent assays in triplicate. Tukey’s multiple comparisons tests. ** p < 0.001. The percent of activity from the luciferase and renilla (Luc/ren) ratio was calculated. The data underlying the graphs shown in the figure can be found in S1 Data . IP, immunoprecipitation; WCE, whole-cell extract; WT, wild type.

Since we previously found that ubiquitination on the VP35 IID can regulate polymerase activity, we asked whether unanchored Ub would also affect this function. To test the function of unanchored Ub in relation to VP35 polymerase cofactor activity, we used the unanchored Ub-specific protease Isopeptidase T (IsoT, also called USP5), which can cleave unanchored Ub by interacting with the free di-glycine residue of Ub chains [ 39 ]. Ectopic expression of IsoT-WT cleaved polyUb chains, as observed in WCE, and correlated with reduced association of Ub with VP35 ( Fig 2A ). In contrast, a catalytically inactive mutant (C335A), which does not cleave unanchored Ub [ 39 ], did not affect the association between VP35 and Ub chains ( Fig 2A ). These results support that VP35 interacts with unanchored Ub chains. Furthermore, the effects of IsoT correlated with decreased EBOV polymerase activity evaluated in a minigenome assay ( Fig 2B ), suggesting that unanchored Ub may promote virus replication. However, although increasing concentration of IsoT further reduced polymerase activity, the highest concentrations do not completely abolish minigenome activity ( S1 Fig ), suggesting that either IsoT is unable to fully remove unanchored Ub bound to VP35, or unanchored Ub plays a partial role in promoting viral polymerase activity. The effect on the minigenome can be partially explained by enhanced interactions between VP35 and NP in the presence of unanchored K63-linked polyUb chains ( S2A Fig ).

The predicted structure of the VP35-Ub complex was used as a template to superpose the structure of VP35 bound to RNA (PDB ID 3KS8). PolyUb was modeled using as a template the structure of K63 Di-Ubiquitin (PDB ID 2JF5). The residues K63, K48, and G76 of the central Ub bound to VP35 are labeled in magenta and contribute favorably to RNA binding in this model. (C) In vitro competition binding assay. Increasing amounts of purified recombinant Ub K63 [ 2 – 12 ] were incubated with VP35 and Biotin-polyI:C, followed by IP with anti-flag beads. (D) The mixes from (C) containing VP35-Ub-PolyI:C were treated with or without Rnase III followed by IP. The data underlying the graphs shown in the figure can be found in S1 Data . IP, immunoprecipitation; PDB, Protein Data Bank.

Since our model indicates that Ub interacts with the basic patch of VP35 that modulates polymerase activity, it would not be likely that Ub affect dsRNA binding to VP35, based on previous reports [ 5 , 11 ]. To test this possibility, we used the core model in Fig 3A as a template to create a model of the ternary complex of VP35, dsRNA, and a tri-Ub chain of K63-linked polyubiquitin ( Fig 5 ). Interestingly, in this model K48 PolyUb would clash with the dsRNA-binding site while the position of K63 points away from the dsRNA-binding site ( Fig 5A ), in agreement with the experimental data (shown in Fig 1 ). The model complex suggests that the central Ub bound to VP35 makes contact with RNA. Not only is K48 occluded by the RNA in the model but it in fact makes favorable interactions with the RNA ( Fig 5B ). Using Surfaces to identify per-residue contributions to this extended interface with RNA, we identified that K48, R54, Y59, and A46 among others favorably contribute to binding RNA. These contribute to strengthen the overall estimated binding free energy by 25% relative to that of the interface between VP35 residues and RNA but lacking the interaction with Ub. In support of this model, increasing concentrations of purified unanchored K63-linked polyUb chains did not compete with the dsRNA mimic poly(I:C) for interaction with VP35. Instead, the presence of Ub chains enhanced co-immunoprecipitation of poly(I:C) with VP35 ( Fig 5C ). Furthermore, treatment with RNase III, which specifically degrades dsRNA, reduced but not eliminated the interactions between VP35 and Ub ( Fig 5D ), further supporting a complex between Ub and VP35 that also favors interactions with dsRNA. Taken together, the interaction between Ub and RNA is likely to be functionally important and suggests that VP35-Ub interaction does not block the ability of VP35 to bind dsRNA and therefore should not affect VP35 IFN-I antagonist function.

(A) HEK293T cells were transfected with minigenome plasmids and VP35 WT, VP35 R225E, or VP35 R225K, followed by Luciferase assay. (B) HEK293T cells were transfected with plasmids encoding Flag-VP35 WT, VP35 R225E, or VP35 R225K. WCE were then used to isolate Flag-tagged proteins using anti-Flag beads. After washes, the beads containing VP35 were mixed with the WCE containing HA-Ub to test binding. (C) As in B, but instead of mixing with WCE, binding was performed using purified recombinant K63-linked polyUb chains, followed by Flag elution. Quantification by densitometry of 3 independent experiments is shown. The data underlying the graphs shown in the figure can be found in S1 Data . WCE, whole-cell extract; WT, wild type.

We utilized the Surfaces software to determine the top contributing interactions between VP35 and Ubiquitin. This analysis detected the R298-E24, R225-E18, R305-S57, R305-D58, and Y229-E18 as the top contributions to the VP35-Ub interaction ( Fig 3A and S1 Table ). We utilized gRINN to validate the Surfaces result. The gRINN analysis confirmed all 5 interactions as the top contributions (see S1 Table ). Both methods suggest that R225-E18 is among the top contributors to the interaction. Analysis of all possible mutations at position R225 with Surfaces ( S2 Table ), suggested that the R225E mutation in VP35 would disrupt the interaction ( Fig 3B and 3C ). The R225E mutation was experimentally tested and has functional effects, abrogating polymerase activity in minigenome assays ( Fig 4A ), which is consistent with previous studies [ 11 ]. Importantly, the mutation led to a decrease in K63 polyUb binding in a cell-free in vitro co-IP assay ( Fig 4B ), or by mixing lysates from cells expressing VP35 and Ub ( Fig 4C ). In contrast, the mutation R225K, which maintains the positive charge on this residue and was predicted to not completely disrupt the interaction ( S1 Table ), partially rescued binding with Ub ( Fig 4C ). Therefore, the reduced binding of VP35-R225E and its reduced polymerase activity further supports a functional role for non-covalent interactions between VP35 and Ub in promoting viral polymerase activity and suggests that the modeled complex structure is correct.

(A) The complex of VP35 (PDB ID 3JKE) and Ubiquitin (PDB ID 1UBQ) modeled using a combination of protein docking and molecular dynamics simulations. Within the complex, VP35 is shown on the left and Ubiquitin on the right. The K48 and K63 Ub residues are shown in cyan on the bottom left and C-terminal on the right within Ub. (B) One of the strongest interactions contributing to the stability of the complex is ARG225-GLU18. (C) Mutation of ARG225 to GLU affects interactions. PDB, Protein Data Bank.

Identification of small-molecules that inhibit VP35–unanchored K63 Ub Interactions

To test whether the non-covalent interactions between VP35 and Ub have functional relevance, we first employed a computational approach with the objective of identifying compounds that could potentially disrupt the Ub-VP35 complex. A cavity within the putative VP35-Ub interface was used as a target to dock 36,000 small molecules with known complex structures using the small-molecule protein docking program FlexAID. Two criteria were used to detect potential binders: A combination of highly favorable docking score relative to the average of all molecules and a large level of binding-site similarities measured using the IsoMIF program between the targeted VP35 cavity and the original protein where the compound is known to bind (Fig 6A). The docking scores (CF) for the 36,000 molecules had a mean value around −100 AU. The z-score of the top 10% varied from −5.0 to −8.0. The top-scored molecules were evaluated to identify those molecules among the top 10% likely to have favorable pharmacological properties. Two molecules emerged from this analysis, pCEBS, 3-[4-(aminosulfonyl) phenyl] propanoic acid—a molecule developed to inhibit carbonic anhydrase [40], and SFC, 2,5-dimethyl-4-sulfamoyl-furan-3-carboxylic acid—a molecule developed as a Metallo-β-lactamase inhibitor [41]. The 2 candidates pCEBS and SFC, had a CF value of −321AU and −278AU, equivalent to a Z-score of −6.5 and −4.8, respectively. The binding site analysis with IsoMIF of the cavities of the crystal structures of the complexes containing SFC (PDB: 6KXO, 6KXI, and 6LBL) revealed binding site similarities of 0.25, 0.32, and 0.35 with VP35, respectively, and the cavities of the crystal structures of the complexes containing pCEBS (PDB: 2NN0 and 2NN1) showed binding site similarities of 0.24 and 0.28 to VP35, respectively. The mean binding site similarity for the top 10% of molecules in the docked dataset is 0.21. Thus, the chosen 2 molecules have a docking score considerably lower (more favorable) than the average and the binding sites known to bind these molecules are more similar to the targeted VP35 cavity than cavities of other top-scoring molecules. This suggests that important interactions responsible for binding pCEBS and SFC are also exploited in the VP35 cavity. Although independently selected, the 2 compounds share a common sulfonamide group (-S0 2 NH 2 ) linked to an aromatic ring system and a carboxyl group that interacts with the same VP35 residues and both are nearly perfectly superimposed (Fig 6B and 6C). Interestingly, at least 1 X-ray structure of VP35 (PDB: 4IBG) shows a sulfate ion from the crystallization buffer bound in very close proximity to the position where the sulfonamide group from pCEBS and SFC are predicted to interact with VP35 based on the docked ligand poses (Fig 6D). The experimental observation that a sulfate ion at that position has favorable interactions with VP35 serves as indication that the docked structures with their sulfonamide groups located at that approximate position are taking advantage of interactions that were experimentally validated.

Incubation with the 2 highest concentrations of either of the 2 molecules leads to a decrease in Ub-VP35 interactions detected in co-IP assays (Fig 7A) and these correlated with a decrease in luciferase activity in minigenome experiments (Fig 7B). These effects from the compounds on minigenome activity can be partially explained by reduced interactions between VP35 and NP as observed in a co-IP assay (S2B Fig). The compounds did not affect interactions between unanchored K63-linked Ub and RIG-I, which is also known to interact with free Ub [35] (S3 Fig), and served as a control for specificity. Although the inhibition did not show a perfect dose-response, possibly due to other important interactions between Ub and viral polymerase proteins, these results, at high concentrations, further suggest that Ub-VP35 non-covalent interactions may contribute to efficient EBOV polymerase function. In contrast, the compounds did not affect the ability of VP35 to antagonize IFNβ in a luciferase reporter assay (S4 Fig). The compounds showed less than 5% cytotoxicity (Fig 7C) and did not cause significant cell death (apoptosis or necrosis), by flow cytometry (S5 Fig). Importantly, both molecules lead to a decrease of infectious EBOV replication in cells, as observed in plaque reduction (PR) and virus yield reduction (VYR) assays, (Figs 7D, 7E, S6A, and S6B). The effect of pCEBS and SFC are within the same range as that observed for the nucleoside analog Favipiravir (T-705), used as a positive control for its broad-spectrum reported activity against Filoviruses [42]. However, pre-treatment with the compounds did not inhibit virus replication (S6C Fig).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. pCEBS and SFC compounds inhibit interactions between VP35 and K63-linked polyubiquitin chains and correlate with reduced viral polymerase activity and virus replication. (A) Flag-VP35 bound to anti-Flag beads were incubated for 1 h at room temperature with different concentrations of pCEBS or SFC, followed by incubation with recombinant purified unanchored K63-linked polyUb chains [2–7]. VP35-Ub complexes were eluted with Flag-peptide and analyzed by Immunoblot. (B) 293T cells were transfected with minigenome components and 4 h post-transfection cells were treated with pCEBS and SFC compounds at different concentrations, and 50 h later cells were lysed for luciferase assay. (C) Cytotoxicity test (CyQUANT MTT Cell Viability Assay Thermo Fisher) using pCEBS and SFC at different dilutions (D) PR and (E) VYR assays, the cells were infected by 1 h and after 1 h the treatment was made with pCEBS, SFB compound, or DMSO: Dimethyl sulfoxide with the overlay. The number of plaques in each set of compound dilution were converted to a percentage relative to the untreated virus control. (F, G) 6-week-old BALB/c females uninfected and treated with PBS (Mock vehicle) (n = 5), uninfected treated with 100 mg/kg of SFC (n = 5) (Mock SFC), infected intraperitoneal (IP) with 100 PFU of maEBOV and treated with either vehicle (EBOV vehicle) (n = 10) or SFC (EBOV SFC) (n = 10). (F) Viral titers in serum of infected mice at days 2, 4, and 6 post-infection. No plaques were detected in the mock groups. (G) Clinical presentation of disease scored as 1: Healthy; 2: Ruffle fur and/or Lethargic; 3: scoring 2 + hunched posture; 4: Weight loss over 20% of initial weight or scoring 3 + unable to move when stimulated, unable to access food/water, or displaying a moribund appearance. The percent of activity from the ratio of luciferase and renilla (Luc/ren) was calculated. Data are depicted as mean + SEM of the 2 independent assays in triplicate. Tukey’s multiple comparisons tests. p < 0.001 **, p < 0.0001 ***, p < 0.00001 ****. The data underlying the graphs shown in the figure can be found in S1 Data. EBOV, Ebolavirus; PR, plaque reduction; VYR, virus yield reduction. https://doi.org/10.1371/journal.pbio.3002544.g007

To further validate experimentally direct interactions between VP35 and Ub, and VP35 and the compounds, microscale thermophoresis (MST) binding assays were performed with tagged full-length purified WT VP35 and K63-linked polyUb chains as well as with pCEBS and SFC. The MST binding experiments detected binding in all 3 instances (S7 Fig). Furthermore, MST titration curve experiments determined a K d of 15 nM for Ub (S8A Fig) and estimated 375 nM for SFC (S8B Fig). It is interesting to note that the K d for Ub falls well within the broad range of K d observed for protein–protein interactions and can be considered as a strong interaction [43]. These results demonstrate that there are direct interactions between VP35 and Ub, as well as between VP35 and both pCEBS and SFC.

In order to add further evidence for the direct interaction between pCEBS and SFC, we modeled and evaluated computationally potential mutations that would disrupt their interactions without impairing the interface with ubiquitin and therefore not compromising the studied mechanism. We modeled 228 mutants, for all possible single substitutions in residues that constitute the binding cavity (Y229, G234, F235, G236, T237, H240, Q241, Q244, I303, P304, R305, A306). The interactions between VP35 mutants and Ub, pCEBS, or SFC (in their positions modeled to bind WTVP35) were evaluated with Surfaces [44]. Based on these results, the mutation F235H (S9 Fig) was selected as likely to not affect Ub binding, while decreasing the strength of interactions with pCEBS and SFC. We utilized the molecular docking program FlexAID [45,46] to dock the 2 molecules to the mutant structures for further evidence of weaking the interaction between the predicted mutated VP35 cavity and the 2 compounds. The FlexAID scoring function (CF) gave a more negative result for more favorable binding interactions. As FlexAID utilizes a probabilistic optimization method, we performed 5 simulations with each of the molecules. The F235H mutation decreases the CF values to −118.9 +- 2.4 AU and −128.9 +- 4.2 for pCEBS and SFC, respectively, relative to the WTVP35 values of −150.1 +- 5.3 AU and −161.5 +- 16.8 AU, respectively. Therefore, we obtain a positive ΔCF = CF mut -CF wt of 31.2 +- 5.82 AU and 32.6 +- 17.32 AU for pCEBS and SFC, respectively. Whereas the docking score is in arbitrary units (AUs), the negative or positive sign in the ΔCF reflects an increase or decrease in favorable interactions, respectively.

To validate the model, we then generated a vector expressing the VP35 F235H mutation and tested whether the compounds are now unable to inhibit VP35 interactions with free Ub chains by coIP. This mutation was previously shown to have reduced minigenome activity [11,20]. As predicted, while the compounds reduced binding of K63-linked polyUb chains with WT VP35, the compounds did not inhibit binding of polyUb chains with VP35-F235H (S10 Fig).

Finally, we then tested whether SFC has antiviral activity in an in vivo mouse model of EBOV infection. We chose SFC because previous studies have shown that it was nontoxic in mice and has activity against carbapenem-resistant Enterobacteriaceae by inhibiting their metallo-ß-lactamases in vitro and in vivo [41]. Intraperitoneal administration of SFC daily for 6 days (S11 Fig) significantly reduced EBOV replication in the serum of infected mice as compared to PBS treated mice (Fig 7F). This reduced viremia in SFC treated mice also correlated with significantly less symptoms of disease such as ruffle fur, lethargy, hunched posture, or moribund appearance (Fig 7G).

Taken together, these results suggest that VP35 non-covalent interaction with Ub promotes EBOV replication by enhancing the function of VP35 as a cofactor of the polymerase. Furthermore, the identification of chemical compounds that block these VP35-Ub interactions could serve as starting point for the development of novel antivirals.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002544

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.