Changes in the chikungunya virus E1 glycoprotein domain II and hinge influence E2 conformation, infectivity, and virus-receptor interactions

ABSTRACT

In a previous study to understand how the chikungunya virus (CHIKV) E1 glycoprotein β-strand c functions, we identified several attenuating variants at E1 residue V80 and the emergence of second-site mutations in the fusion loop (E1-M88L) and hinge region (E1-N20Y) with the V80 variants in vivo. The emergence of these mutations led us to question how changes in E1 may contribute to CHIKV infection at the molecular level. Here, we use molecular dynamics to understand how changes in the E1 glycoprotein may influence the CHIKV glycoprotein E1-E2 complex. We found that E1 domain II variants lead to E2 conformational changes, allowing us to hypothesize that emerging variants E1-M88L and E1-N20Y could also change E2 conformation and function. We characterized CHIKV E1-M88L and E1-N20Y in vitro and in vivo to understand how these regions of the E1 glycoprotein contribute to host-specific infection. We found that CHIKV E1-N20Y enhanced infectivity in mosquito cells, while the CHIKV E1-M88L variant enhanced infectivity in both BHK-21 and C6/36 cells and led to changes in viral cholesterol-dependence. Moreover, we found that E1-M88L and E1-N20Y changed E2 conformation, heparin binding, and interactions with the receptor Mxra8. Interestingly, the CHIKV E1-M88L variant increased replication in Mxra8-deficient mice compared to WT CHIKV, yet was attenuated in mouse fibroblasts, suggesting that residue E1-M88 may function in a cell-type-dependent entry. Taken together, these studies show that key residues in the CHIKV E1 domain II and hinge region function through changes in E1-E2 dynamics to facilitate cell- and host-dependent entry.

IMPORTANCE

Arboviruses are significant global public health threats, and their continued emergence around the world highlights the need to understand how these viruses replicate at the molecular level. The alphavirus glycoproteins are critical for virus entry in mosquitoes and mammals, yet how these proteins function is not completely understood. Therefore, it is critical to dissect how distinct glycoprotein domains function in vitro and in vivo to address these gaps in our knowledge. Here, we show that changes in the CHIKV E1 domain II and hinge alter E2 conformations leading to changes in virus-receptor and -glycosaminoglycan interactions and cell-specific infection. These results highlight that adaptive changes in E1 can have a major effect on virus attachment and entry, furthering our knowledge of how alphaviruses infect mammals and insects.

INTRODUCTION

Alphaviruses are single-stranded positive-sense RNA viruses responsible for causing arthritic and encephalitic diseases worldwide (1–3). This genera includes arthropod-borne viruses (arboviruses) such as eastern equine encephalitis virus, chikungunya virus (CHIKV), and Mayaro virus. As there are no effective antiviral therapeutics available, understanding the mechanisms of how alphaviruses replicate and enter cells is crucial for preventing and controlling future epidemics.

CHIKV is a re-emerging alphavirus spread by Aedes (Ae.) mosquitoes. CHIKV causes disease consisting of rash and arthralgia and can cause neurological and cardiac manifestations in severe cases (4–8). Importantly, in 2023, there has been a significant increase in CHIKV cases in South America, emphasizing the importance of further studying the mechanisms, transmission, and spread of this virus (9, 10). The alphavirus genome consists of two open reading frames, the first encoding four nonstructural proteins (nsP1—nsP4) and the latter encoding the structural proteins, including capsid, E3, 6K, E2, TF, and the class II fusion glycoprotein E1 (11). Embedded in the virion membrane, the CHIKV E1 and E2 glycoproteins are essential in the early stages of receptor-mediated endocytosis of alphaviruses. The E1-E2 heterodimer is disrupted as the endosomal acidic pH sparks a conformational change where E1 trimerizes and further facilitates the fusion of viral and endosomal membranes (12–15). The alphavirus E1 glycoprotein is comprised of three domains (E1-I, E1-II, and E1-III), where domain II contains the fusion loop, ij loop, and bc loop that have been previously studied for their importance in attachment, fusion, and infectivity (16–18). Notably, during a significant outbreak in 2005 on La Réunion Island, an E1-A226V mutation in the domain II ij loop was identified and caused an increase in fitness and change in host vector to Ae. albopictus mosquitoes, thus leading to an increased widespread distribution of the virus (10, 19).

In previous studies to understand CHIKV E1 function, we generated a panel of amino acid substitutions in the β-strand c residue V80 and showed that the CHIKV E1 glycoprotein β-strand c plays an important role in the entry, infectivity, fusion, and dissemination of CHIKV both in vitro and in vivo (20). However, we do not understand exactly how these changes in the E1 glycoprotein are driving these phenotypes. Moreover, in these studies, we identified two novel second-site E1 mutations that were found in Ae. aegypti mosquitoes infected with the CHIKV E1 variant V80Q. These mutations (E1-M88L and E1-N20Y) were initially identified by an increased plaque size and were then hypothesized to play important roles in the stability of the E1-V80Q variant as well as potentially for the evolution and transmission of CHIKV.

In this study, we used molecular dynamic simulations of the E1-E2 molecules to begin understanding how changes in the E1 fusion loop, β-strand c, and ij loop can impact the glycoproteins. We found that these changes in E1 can alter the conformations of both E1 and E2, indicating changes in E1 can have significant impacts on the entire glycoprotein spike. To test this hypothesis, we generated the CHIKV E1-M88L and E1-N20Y variants and used both in vitro and in vivo approaches to characterize how these residues contribute to CHIKV entry and pathogenesis. In line with our in silico results, we found that CHIKV E1-M88L and E1-N20Y significantly changed the host- and cell-type infection of CHIKV in vitro, as well as interactions with heparin and Mxra8, functions of the E2 protein. When evaluating the usage of the CHIKV receptor Mxra8 in vivo, the E1-M88L variant showed increased viral titers in Mxra8-deficient mice, suggesting it may be using alternative routes of infection compared to wild-type CHIKV. Taken together, these studies highlight how changes in the E1 domain II and hinge can impact virus binding and infection, and how changes in the E1 protein can contribute to multiple steps in the entry process beyond fusion.

RESULTS

Molecular computational simulations of CHIKV E1 β-strand c and fusion loop variants reveal changes in E1 and E2 dynamics

In a previous study to understand how the CHIKV E1 glycoprotein β-strand c contributes to infection, we mutated the CHIKV E1 β-strand c residue V80 to every possible amino acid and identified a variant of V80 (E1-V80L) that is significantly attenuated in vitro and in vivo (20). To gain mechanistic insight on how the CHIKV E1-V80L variant may be contributing to viral entry, we took a molecular dynamics approach to understand how changes in E1 may be influencing the dynamics of the CHIKV E1-E2 heterodimer in silico. By performing molecular dynamic simulations of the pre-fusion conformation of the E1-E2 heterodimer and doing a principal component analysis (PCA) of the concatenated trajectory (Fig. 1A), we observed that the first principal component (PC) was associated with the bending of the envelope glycoproteins of CHIKV and the second PC was associated with the opening of E2 domain B, which is present in a stable conformation between open and closed states (Fig. 1B). When we introduced the attenuated E1-V80L variant into the simulation, we observed a shift in both the second PC, with E2 now in more open conformation (Fig. 1C—light green), suggesting that the E1-V80L variant can influence E2. Importantly, when we introduced E1-V80L with the E1-V226A variant, which we found to experimentally rescue the attenuation of E1-V80L (20), we found that E2 was restored to the wild-type conformation (Fig. 1C—dark green). As the major difference was observed in PC2, which involves mainly residues in E2-B, we performed a PCA of E2 alone. Consistently in the PC1, we observed the E2 wild-type protein to be in two distinct states consisting of one major closed state and an open sub-state, with a shift to the open conformation in the E1-V80L mutant which was restored to the WT phenotype by the E1-V226A second-site mutation (Fig. 1C). To confirm these findings, we ran simulations on several other E1-V80 variants we characterized in previous studies to increase virulence and transmission (E1-V80I) or genetically unstable (E1-V80K, V80F, and V80E) (Fig. 1D). Although each of these variants significantly changed the dynamics of E2 in different ways, there seems to be an association of a shift to the open state of E2 domain B with a reduced phenotype (E1-V80L, V80K, V80F, and V80E). These proof-of-principle simulations show that changes in E1 β-strand c and the ij loop can impact E2 conformations and highlights that we can understand glycoprotein function through molecular simulations coupled with experimentation.

Fig 1.

CHIKV E1 variants influence E2 molecular dynamics in silico. (A) Free energy landscape along the first two principal components (PC1 and PC2) obtained from a PCA of the Cα atoms of the MD trajectory of the WT E1-E2 heterodimer. A dark dot represents the minimum energy conformation. The explained variance of each PC is shown in the X and Y labels as a percentage of the total variance. (B) Collective motion of the Cα atoms represented by each PC. The extreme conformations are colored in dark or light cyan for E2 and orange for E1 and intermediate conformations are depicted as transparent tubes. (C) Free energy landscape along the first two PCs obtained from the MD trajectory for the WT, E1-V80L, or E1-V80L/V226A variants. The top row is a representation of the projection along the first two PCs for the E1-E2 heterodimer and the bottom row is the projection only for E2 protein. (D) Free energy landscape along the first two PCs obtained for E2 protein with WT (gray), E1-V80I (blue), E1-V80K (red), E1-V80F (brown), and E1-V80E (yellow) variants.

In addition to the variants above, we also identified a highly attenuated variant E1-V80Q that presented with a small plaque phenotype. When we infected Ae. aegypti mosquitoes with CHIKV E1-V80Q, we observed two emerging second-site E1 glycoprotein variants associated with increased plaque size. One variant, E1-M88L, is located in the conserved E1 fusion loop of domain II, and the other, E1-N20Y, is located in the E1 hinge region of domain I. Given, our results with E1-V80L:E1-V226, we asked whether the E1-V80Q:E1-M88L double variant changed the simulation of E1-V80Q. When modeled alone, we found that the E1-V80Q variant shifted E2 more to the closed conformation and E1-M88L further shifted E2 to this closed state yet did not restore the wild-type phenotype (Fig. 2A). Unfortunately, when we attempted to generate the E1-V80Q single and double variants to address virus biology in the lab, we found that the E1-V80Q single variant was genetically unstable, reverting to the wild-type valine or obtaining the E1-M88L mutation for stability. Therefore, without a proper control, we could not explore interactions between E1-V80Q and M88L or N20Y. Nonetheless, when we introduced the E1-M88L variant into the molecular dynamics simulation, we observed that the E2 open state was largely absent compared to wild-type E2, suggesting that changes in the E1 fusion loop can place the E2 protein into a more closed conformation (Fig. 2B). Taken together, these molecular simulations highlight dramatic changes in E2 dynamics imparted by E1 variants and provide molecular insight into how E1 can potentially drive cell binding and infectivity in multiple ways.

Fig 2.

Molecular dynamic simulations of CHIKV E1-V80Q and E1-M88L variants. (A) Free energy landscape along the first two PCs obtained from a PCA of the Cα of the MD trajectory for E1-E2 heterodimer (top) or E2 protein (bottom) for CHIKV WT (gray), V80Q (beige), and V80Q-M88L (purple). (B) Free energy landscape along the first PC obtained for E2 protein of WT and E1-M88L variant.

CHIKV E1-M88L and E1-N20Y enhance infection in a host-specific manner in vitro

Given the emergence of E1-M88L and E1-N20Y, the location of these residues in the E1 glycoprotein (Fig. 3A) and our molecular simulation data showing changes in E2, we hypothesized that E1-M88 and E1-N20 are important residues for virus entry. To characterize how these residues contribute to CHIKV infection, we first aligned the fusion loop and domain I hinge region across multiple alphaviruses. We found that E1-M88 is conserved in 75% of the alphaviruses of our alignment (Fig. 3A). Interestingly, the leucine variant is found in Aura virus (AURV), Middelburg virus (MIDV), and Una virus (UNAV), suggesting this residue may provide some advantage in other alphaviruses. On the contrary, E1-N20 is highly variable among alphaviruses (Fig. 3A). To begin to characterize these variants in vitro, we introduced the E1-M88L and E1-N20Y variants into the wild-type (WT) CHIKV infectious clone, rescued virus, and looked at the growth kinetics of each mutant in both mammalian BHK-21 (Fig. 3B) and insect C6/36 Aedes (Ae.) albopictus (Fig. 3C) cells, yet found no growth difference between the viruses over 24 h. However, when we quantified the amount of viral RNA in the supernatant at 24 h post-infection and calculated the specific infectivity of each virus (Fig. 3D and E), we observed that while the specific infectivity was similar in BHK-21, the viruses in C6/36 cells had increases in specific infectivity, suggesting an increase in infectious virus. Finally, we addressed protein accumulation by western blotting. We electroporated BHK-21 cells with each viral in vitro transcribed RNA and lysed the cells after 48 h. We observed that while there were no major differences in nsP1, capsid, or E2 protein accumulation between wild-type (WT) CHIKV and either E1 variant, there was a consistent reduction in E1 accumulation in the E1-N20Y variant (Fig. 3F).

Fig 3.

CHIKV E1-M88L and E1-N20Y variant conservation and growth kinetics. (A) PyMol structure (PDB: 3N42) depicting the CHIKV E1 and E2 glycoprotein. Domains are color-coded as follows: E1-I (red), E1-II (orange), Fusion Loop (pink), E1-III (yellow), E2-A (light blue), E2-B (dark blue), E2-C (purple), E2-β ribbon (green). The E1-M88 and E1-N20 positions are in black and labeled. Part of the fusion loop and part of the E1-I domain alignments across 12 alphaviruses are shown as a composite logo, with a yellow star indicating the CHIKV E1-M88 and E1-N20 position. BHK-21 cells (B) and C6/36 cells (C) were infected at an MOI of 0.1 for 1 h, washed with PBS, and supernatants were collected at the indicated timepoints. Viral titers were quantified via plaque assay. The dotted line and gray-shaded area represent the limit of detection (LOD). Data represent three independent trials, and no statistical significance was found via two-way ANOVA. Viral supernatants from BHK-21 (D) or C6/36 cells (E) that were collected from cells infected at an MOI of 0.1 were used to determine infectious titers via plaque assay and RNA genomes via qPCR. Specific infectivity was then calculated by dividing the infectious titers (PFU/mL) by the viral genomes (RNA molecules/mL). Data represent two independent trials in duplicate, and statistical significance was found via Kruskal-Wallis multiple comparisons test. (F) BHK-21 cells were electroporated with in vitro transcribed RNA of each virus or mock transfected. Cells were lysed 48 h post-electroporation and CHIKV E1, E2, Capsid, nsP1, and Actin accumulation was visualized by western blotting. The bar graph represents the adjusted volume of protein bands that were quantified via Image Lab 6.0.1 software from all three independent trials. The average and standard error of the mean (SEM) are shown for all data.

The CHIKV E1-V80 residue contributes to CHIKV infectivity and E1-M88L is found in the fusion loop, allowing us to hypothesize that these variants may contribute to virus infectivity. To test this hypothesis, we performed an infectivity assay, where we adsorbed virus to BHK-21 and C6/36 cells for 1 h at 4°C and then incubated the cells at 37°C for mammalian cells and 28°C for insect cells to induce entry. At different timepoints, entry was stopped by the addition of 20 mM ammonium chloride to neutralize the endosomal and lysosomal pH and block further viral infection and spread (Fig. 4A and B). We found that E1-M88L led to faster infection and enhanced CHIKV infectivity in BHK-21 cells while E1-N20Y behaved like WT CHIKV (Fig. 4A). When we completed the same experiment with insect cells, we found that there was minimal infection in C6/36 cells until the 60 min post-infection timepoint, at which we see an increased infectivity for both E1-M88L and E1-N20Y (Fig. 4B). These results suggest that the E1-M88L variant increases infection in both hosts, while E1-N20Y is insect cell specific. To investigate if this increased infectivity was due to cell binding, we incubated both mammalian and insect cells with each virus for 30 min at 4°C in the presence of 20 mM ammonium chloride and quantified the number of membrane-bound virus via qPCR (Fig. 4C). We found no significant difference in cell-bound RNA but did find that the E1-M88L variant led to 2.5-fold (BHK-21 cells) and 1.8-fold (C6/36 cells) increases in RNA compared to WT. Lastly, as CHIKV entry is cholesterol-dependent (20, 21), we sought to determine if there were differences in cholesterol-dependent entry in our E1 variants. We depleted BHK-21 cells of cholesterol using varying concentrations of methyl-beta-cyclodextrin (MβCD), infected cells with each virus, and then added 20 mM ammonium chloride to block spread and, therefore, determine the entry efficacy of each virus. We found that E1-M88L showed a decrease in cholesterol dependency, as the virus was still able to infect cells at higher concentrations of MβCD compared to E1-N20Y and WT (Fig. 4D). These results show that while E1-N20Y had an increased infectivity in C6/36 cells, E1-M88L was able to enter both mammalian and insect cells more efficiently compared to WT.

Fig 4.

CHIKV E1-M88L and E1-N20Y influence infectivity in a host-dependent manner. (A) BHK-21 cells were infected with WT CHIKV or each E1 variant expressing ZsGreen at an MOI of 1 and treated with 20 mM ammonium chloride at the indicated timepoints post-infection. Cells were fixed and stained with DAPI 24 h post-infection. Infected cells were quantified using a CX7 high-content microscope, and representative images were shown for the 60 min post-infection timepoint. Data represent three independent trials with internal duplicates. Statistical significance was found by two-way ANOVA and indicated with P-values shown. (B) C6/36 cells were infected with WT CHIKV or each E1 variant expressing ZsGreen at an MOI of 0.1 for 1 h and then treated with 20 mM ammonium chloride. Cells were fixed and stained at 24 h post-infection, and infected cells were quantified as above with representative images. Data represent four independent trials with internal duplicates. Statistical significance was found by Kruskal-Wallis one-way ANOVA test and indicated with P-values shown. (C) Binding assays were performed on BHK-21 and C6/36 cells. Cells were incubated on ice with each virus at an MOI of 100 (based on viral genomes) for 1 h. After a cold PBS wash, virus-bound cells were collected and RNA genomes were quantified using qPCR. A fold change increase is labeled for E1-M88L RNA genomes bound percent when normalized to WT. Data represent three independent experiments, with internal duplicates. A Mann-Whitney test was used to determine statistical significance and indicated P-values are shown. (D) BHK-21 cells were pre-treated with methyl-beta-cyclodextrin (MßCD) for 1 h and washed once with PBS, and cholesterol-depleted cells were infected with each CHIKV virus at an MOI of 1 for 1 h before adding 20 mM ammonium chloride. Cells were fixed and stained 24 h post-infection and quantified as above. Data represent three independent trials with internal duplicates. Statistical significance was found by two-way ANOVA and indicated with P-values shown. The average and SEM are shown for all data.

One explanation for the enhanced infectivity of the CHIKV E1-M88L variant may be that while the CHIKV E1-M88L particle is more infectious, there is a defect in another step in the life cycle, such as RNA replication. To test this hypothesis, we introduced the E1-M88L variant into a virus expressing a Firefly luciferase reporter that is expressed under active replication through a subgenomic promoter (Fig. 5A). We transfected BHK-21 cells with each CHIKV in vitro transcribed (IVT) RNA and measured intracellular luciferase activity at 4, 6, 8, and 24 h post-transfection (Fig. 5A and B). Prior to the luciferase readout, we transferred the culture supernatant to naïve cells and measured the intracellular luciferase activity after 24 h to evaluate infectious particle production (Fig. 5A and C). While there were no differences in luciferase activity between WT CHIKV and E1-M88L (Fig. 5B), we did observe an increase in luciferase activity from the supernatant at 8 h post-transfection compared to WT CHIKV (Fig. 5C). These results suggest that while there are no differences in subgenomic replication, CHIKV E1-M88L may be producing more particles faster than WT or more infectious particles, which may lead to enhanced infection.

Fig 5.

WT CHIKV and E1-M88L subgenomic replication and particle production. (A) A schematic of the CHIKV genome expressing the firefly luciferase reporter (top) and schematic of the experiment (bottom). (B) BHK-21 cells were transfected with WT CHIKV, nsp4-GNN, or E1-M88L firefly IVT RNAs using Lipofectamine 2000. At each indicated timepoint, the supernatant was transferred to naïve BHK-21 cells, and the cells were lysed to quantify luciferase activity. (C) Twenty-four hours post-infection of the naïve BHK-21 cells, the cells were lysed and luciferase activity was quantified. Data represent four independent trials. No statistical significance was found via two-way ANOVA. The average and SEM are shown for all data.

The CHIKV E1-M88L variant shows no major advantage in mosquitoes or WT mice

As CHIKV E1-M88L was found as a second-site mutation from a CHIKV E1-V80Q infection in Ae. aegypti mosquitoes and E1-M88L enhanced infectivity and binding in C6/36 cells, we wanted to see if there were differences in infection and dissemination of CHIKV E1-M88L in a mosquito model. We infected Ae. aegypti mosquitoes with 10⁶ PFU of WT or CHIKV E1-M88L, collected the mosquito bodies and legs and wings after 7 days post-infection, and assessed infectious virus by plaque assay. We found no differences in infection or dissemination to the legs and wings between WT CHIKV and E1-M88L (Fig. 6A and B). Since CHIKV E1-M88L enhanced infectivity and binding in BHK-21 cells, we hypothesized this residue may play a role in infection in WT mice. We infected C57BL/6J mice with 1000 PFU of each virus via the footpad, and harvested footpad, ipsilateral calf and quadricep muscle, and serum at 2, 3, and 5 days post-infection to quantify viral particles by plaque assay (Fig. 6C through E). We found that CHIKV E1-M88L had no significant advantage in WT mice at most timepoints with the exception of a slight attenuation at 3 days post-infection (dpi) in the footpad and serum. These results suggest that the function of the CHIKV E1-M88L variant plays cell-specific roles that could be masked in whole organism models.

Fig 6.

WT CHIKV and E1-M88L replication in Aedes aegypti mosquitoes and C57BL/6 J mice. Aedes aegypti mosquitoes were fed a viral blood meal containing 10⁶ PFU/mL of WT CHIKV or E1-M88L virus. Mosquitoes were maintained for 7 days post-infection, and viral titers were quantified via plaque assay in bodies (A) or legs and wings (B). Data represent two independent infectious with at least n = 43 mosquitoes. Male and female 5- to 7-week-old C57BL/6J mice were infected with 1000 PFU of WT CHIKV or E1-M88L virus via footpad injection. Mice were euthanized at the corresponding timepoint and the footpad (C), the ipsilateral calf and quadricep muscle (D) and serum (E) harvested. Infectious titers were determined via plaque assay. Data represent at least two independent infections with at least n = 9 mice. The dotted line and gray-shaded area represent the LOD. No statistical significance was found via Mann-Whitney test. The average and SEM are shown for all data.

CHIKV E1-M88L and N20Y alter interactions with E2-specific monoclonal antibodies and heparin

The molecular simulations and increased infectivity of E1-M88L and E1-N20Y in vitro allowed us to hypothesize that the E1-M88L and E1-N20Y variants may be changing E2 conformation leading to these phenotypes. E2 has been shown to be important for glycosaminoglycan (GAG) and receptor interactions (22–24). Therefore, we hypothesized that if E1-M88L and E1-N20Y were changing E2 conformation, we may be able to observe changes in structure and/or GAG interactions. To test this hypothesis, we first used CHIKV monoclonal antibodies that target residues in E2 domain B and addressed the ability of these antibodies to neutralize CHIKV (Fig. 7A—orange and green). We incubated ZsGreen reporter WT CHIKV and each E1 variants with increasing concentrations of bovine serum albumin (BSA) as a control (Fig. 7B) or the mouse monoclonal antibodies CHK263 (Fig. 7C) and RRV-12 (Fig. 7D). We observed that the CHIKV E1-N20Y variant was significantly more sensitive to the CHK263 antibody compared to WT virus and E1-M88L, yet we found no difference in neutralization with the RRV-12 antibody. Together, these results show that the E1-N20Y variant alters E2 conformation, leading to increased neutralization. Finally, to test if the E1 variants cause changes in CHIKV-GAG interactions, we looked at the direct binding of purified CHIKV particles with heparin-agarose beads (Fig. 7E). We incubated each virus with either heparin-agarose beads or protein A/G-agarose beads as a control for 1 h. We isolated and washed the beads and addressed bound proteins by western blotting. Interestingly, we observed that both E1-M88L and E1-N20Y increased heparin binding compared to WT CHIKV (Fig. 7E). Taken together, these results provide support that both E1-M88L and E1 N20Y can lead to changes in E2 function.

Fig 7.

CHIKV E1-M88L and E1-N20Y change antibody and heparin binding. (A) CHIKV E1 and E2 crystal structure (PDB ID: 6NK7), where E1 is yellow and E2 is light blue. Monoclonal antibodies CHK263 (green) and RRV-12 (orange) binding sites are shown. E1-M88 and E1-N20 are in red. E2-R82 is in magenta. Each ZsGreen reporter virus was incubated with the indicated concentration of BSA (B), CHK263 (C), or RRV-12 (D) for 1 h at 37°C and then used to infect BHK-21 cells for 18 h. Plates were then fixed and stained with DAPI, and the number of ZsGreen positive cells was quantified. Data represent the average and SEM. Each experiment was completed in triplicate with internal technical duplicates. A two-way ANOVA was used to determine statistical significance, with *P < 0.05 and **P < 0.001. (E) Each sucrose gradient-purified virus was incubated with heparin-agarose or protein A/G-agarose beads for 1 h at room temperature. Beads were isolated, washed twice, and resuspended in Laemmli buffer. Proteins were separated by SDS-PAGE and analyzed by western blotting. Protein bands were quantified by densitometry using Image Lab. Images represent one of two independent experiments.

CHIKV E1-M88L and E1-N20Y alter virus-Mxra8 interactions

The Matrix Remodeling Associated 8 (Mxra8) protein is an important receptor for CHIKV and other arthritogenic alphaviruses (2, 22, 25). A major function of CHIKV E2 is to interact with Mxra8 for entry (2, 3, 22). CHIKV residue E1-M88 is in proximity to the Mxra8 receptor binding domain (Fig. 8A), and molecular dynamic simulations of the Mxra8 binding sites for the E1-M88L and E1-N20Y variants show the binding site deviating from the WT structure (Fig. 8B and C), suggesting the E1 variants could influence CHIKV-Mxra8 interactions. To test direct binding of mouse and human Mxra8 with WT CHIKV and each variant, we performed an ELISA assay by incubating each virus with mouse or human Mxra8-Fc, a positive control RRV-12 antibody, or negative control West Nile virus (WNV) antibody (Fig. 8D). We found no difference in binding between viruses. While the ELISA showed no differences in Mxra8-virus binding, it does not provide a functional readout of virus-Mxra8 interactions. To provide a functional readout, we looked at the ability of each virus to be neutralized by purified mouse or human Mxra8 (25). We incubated each virus with increasing concentrations of mouse or human Mxra8 and addressed virus infection on BHK-21 cells. We found that although Mxra8-virus binding was equal, the E1-M88L and E1-N20Y variants were more resistant to Mxra8 neutralization (Fig. 8E and F), suggesting that the interaction between each virus and Mxra8 was modified.

Fig 8.

CHIKV E1 variants alter virus-Mxra8 interactions. (A) Crystal structure of CHIKV E1-E2 with Mxra8 (PDB ID: 6NK7), with labeled E1 (yellow), E2 (purple), and Mxra8 (magenta), and E1-M88 (red). (B and C) Molecular dynamics simulation of the E1-E2 Mxra8 binding site of WT CHIKV (n = 3), E1-M88L (n = 3), and E1-N20Y (n = 2). (D) Sucrose gradient-purified viruses were bound to ELISA plates with either mouse or human CHIKV antibodies and incubated with mouse Mxra8-Fc, human Mxra8-Fc, positive control antibody (human RRV-12), or negative control antibody (human WNV). Absorbance was read at OD 450 nm. Data represent the average and SEM of two independent experiments with internal triplicates. Each ZsGreen reporter virus was incubated with the indicated concentrations of mouse Mxra8-Fc (E) or human Mxra8-Fc (F) for 1 h at 37°C and added to BHK-21 cells for 18 h. Plates were fixed and stained with DAPI, and ZsGreen positive cells were quantified. Data represent the average and SEM of three independent experiments with internal duplicates. A two-way ANOVA was performed, with *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

CHIKV E1-M88L replication at the site of infection is Mxra8-independent in vivo, but not in mouse fibroblasts in vitro

Given that the E1-M88L variant is the most resistant to Mxra8 neutralization, we hypothesized that E1-M88L entry may be Mxra8-independent. To test this hypothesis, we infected control NIH-3T3 mouse fibroblast cells, Mxra8-deficient 3T3 cells, or Mxra8-deficient cells expressing Mxra8 in trans with each virus and harvested virus containing supernatants at 24 and 48 h (Fig. 9A and B). We found that in control 3T3 (Fig. 9A—red lines), the E1-M88L variant was attenuated in growth over 2 days. In addition, the E1-M88L variant replicated slightly worse than WT CHIKV in Mxra8-deficient cells, suggesting that the CHIKV E1-M88L variant did not have an advantage in mouse 3T3 cells. When Mxra8 was introduced in trans, we found that WT CHIKV and E1-M88L replication was enhanced, yet the E1-M88L variant was still attenuated (Fig. 9B—green lines). These results were confirmed by fixing and staining the cells at the 48 h timepoint, where we found that E1-M88L had significantly less infected cells in our control and overexpressed Mxra8 cells (Fig. 9C and D). Finally, we wanted to test whether there was a difference in infectivity of the viruses in 3T3 cells, similar to what we saw in BHK-21 cells (Fig. 4A), We infected each 3T3 cell line with the viruses for 1 h and then replenished the cells with 20 mM ammonium chloride in complete media for 24 h. After fixing, staining, and quantifying the number of infected cells, we found that again the CHIKV E1-M88L virus was attenuated in both control cells and overexpressed Mxra8 cells (Fig. 9E).

Fig 9.

CHIKV E1-M88L replicates in Mxra8-deficient mice but is attenuated in mouse fibroblasts. (A and B) Virus growth in wild-type NIH 3T3 mouse embryonic fibroblasts (MEF), Mxra8-deficient MEFs, or MEFs expressing Mxra8 in trans. NIH 3T3 MEF cells were infected with each virus expressing ZsGreen with an MOI of 5 for 1 h, washed with PBS, and complete media added. Supernatants were collected at the indicated timepoints and infectious titers were quantified via plaque assay. At 48 h post-infection, cells were fixed and stained with DAPI. The number of infected cells was imaged (C) and quantified (D) using a CX7 high-content microscope. (E) Each cell line was incubated with each virus at an MOI of 5 for 1 h followed by the addition of 20 mM ammonium chloride. At 24 hpi, cells were fixed and stained for DAPI, and infected cells were quantified as above. Data represent at least two independent trials in triplicate. Multiple Mann-Whitney tests were performed with P-values representing **P < 0.01 and ****P < 0.0001. Male and female 5- to 7-week-old WT or Mxra8^Δ8/Δ8 mice were infected with 1,000 PFU of WT CHIKV or E1-M88L virus via footpad injection. Mice were euthanized at 3 days post-infection. The footpad (F) and ipsilateral calf and quadricep muscle (G) were harvested to quantify infectious titers via plaque assay. Data represent at least three independent infections with n > 13. The dotted line and gray-shaded area represent the LOD. The average and SEM are shown for all data.

To test Mxra8-dependence in vivo, we infected Mxra8-deficient and WT litter mate control mice with 1,000 PFU of either WT CHIKV or E1-M88L virus via footpad injection. At 3 days post-infection, we harvested the footpad and ipsilateral calf and quadricep muscle and quantified infectious virus by plaque assay. We found that while WT and E1-M88L replicated to the same levels in the footpad of wild-type mice (Fig. 6C), the E1-M88L variant led to a statistically significant increase in infectious particles in the footpad of Mxra8-deficient mice (Fig. 9F and G). However, this phenotype was not seen in the ipsilateral calf and quadricep muscle. These findings suggest that while attenuated in our in vitro model, E1-M88L is able to infect and replicate within Mxra8-deficient mice in vivo, suggesting a cell-type specific mechanism present in mice may be important for virus binding and entry.

DISCUSSION

CHIKV is an alphavirus that has caused significant outbreaks worldwide, including recent explosive outbreaks in South America (9, 10). Our previous studies use CHIKV as a model alphavirus to examine how the E1 class II fusion glycoprotein contributes to viral fusion, infectivity, and evolution (20, 26, 27). We identified two second-site E1 glycoprotein variants (E1-M88L and E1-N20Y) in the bodies of Ae. aegypti mosquitoes infected with an attenuated E1 variant, CHIKV E1-V80Q, that corresponded with an increased plaque size phenotype (20). This initial observation allowed us to hypothesize that perhaps the emergence of E1-M88L and/or E1-N20Y was able to rescue the attenuation of CHIKV E1-V80Q. Unfortunately, due to the genetic instability of E1-V80Q, we were unable to answer this specific question and instead used in vitro and in vivo approaches to understand how E1-M88 and E1-N20 contribute to the CHIKV life cycle.

When we focus on the location and conservation of E1-N20 and E1-M88 across different alphaviruses, we find that in the E1 domain I hinge, E1-N20 is the most variable of the two with the majority of alphaviruses having a charged residue at this position. The domain I hinge has been shown to be important for virus entry and domain swiveling during membrane fusion (28, 29). Given that the E1-N20Y variant increases infection in C6/36 cells and not mammalian cells, it suggests that this residue is important for insect-specificity between alphaviruses for entry. In contrast, residue E1-M88, in the fusion loop, is highly conserved among alphaviruses and is flanked by a conserved phenylalanine that has been shown to be critical for infectivity in multiple alphaviruses (30). Interestingly AURV, UNAV, and MIDV all encode a leucine at this position, suggesting there may be an evolutionary role for the residue in alphaviruses.

Using cell culture approaches, we observed minimal differences in RNA replication and growth kinetics of E1-M88L and E1-N20Y in BHK-21 and C6/36 cells compared to wild-type CHIKV. However, we did find changes in specific infectivity and E1 accumulation, suggesting a potential role for these residues in virus assembly. In addition, we observed that E1-N20Y enhanced infection specifically in C6/36 cells, while E1-M88L was able to enhance binding and infectivity in BHK-21 and C6/36 cells. These results suggest that the fusion loop and hinge region may contribute to the cell- and/or host-specific entry of CHIKV. Interestingly, previous work has shown that Semliki Forest virus E1-M88L had minimal effects on membrane fusion, surface expression, antibody binding, and glycosylation (31). These results suggest that the function of residue E1-M88 may be virus-specific or study-specific depending on lab cell lines and culture conditions. Nonetheless, the idea of cell- and/or host-specificity is supported by experiments in NIH-3T3 cells, where we found that E1-M88L was attenuated in these cells, suggesting that there are host and/or cell-dependent differences. This hypothesis may also explain why we observe no differences in growth kinetics in BHK-21 and C6/36 cells, even with enhanced infectivity of the variants. These results may be due to how the growth curves were tittered on Vero cells, and variations in Vero-specific phenotypes of these variants may contribute to differences in reported growth kinetics as compared to viral infectivity. Future work focusing on how the CHIKV E1 fusion loop and hinge impact infectivity of multiple cell types may address these questions and shed light on cell- and host-dependent entry.

A major finding of this study is that the CHIKV E1-M88L and E1-N20Y variants lead to changes in heparin binding and Mxra8 interactions, processes thought to be driven by the E2 glycoprotein. These results along with our molecular simulations suggest that changes in E1 can have effects on E2 conformation driving changes in virus-host interactions and infection. In particular, E1-M88 is located in the fusion loop which is in close proximity to the Mxra8-binding site and, therefore, may alter Mxra8 binding. In this same location is the E2-D71 residue shown to be important for Mxra8 interactions (22) as well as E2 R82 which mediates GAG binding (23, 24). Therefore, changes in E2 domain B may change these entire regions, modifying E2 function. E1-M88L altering Mxra8-CHIKV interactions may also explain why we observe a phenotype with E1-M88L in CHIKV, yet not in SFV. While SFV does use Mxra8 to an extent, it is much more dependent on the very low-density lipoprotein receptor (32, 33). Finally, E1-N20Y also leading to changes in E2 conformation and function is intriguing. These findings suggest that the hinge region may be important for proper glycoprotein complex assembly, or in critical movement of E2 during entry. Future studies looking at the role of E1 in particle assembly and how changes in E1 can drive changes in GAG and receptor usage are critical to understand how alphaviruses enter cells and how the adaptation of alphaviruses may influence receptor usage.

Given our enhanced infection of BHK-21 and C6/36 cells with E1-M88L, we also addressed whether CHIKV E1-M88L had an advantage in mosquitoes and mice. We found no major differences between E1-M88L and wild-type CHIKV in Aedes aegypti mosquitoes or C57BL/6J mice, with the exception of decreased viral titers in footpad and increased clearance from the serum at 3 days post-infection. One explanation of these findings may be that, as we saw with in vitro NIH 3T3 experiments, E1-M88 is critical for cell and/or host-dependent entry, and therefore, results on BHK-21 cells do not reflect what happens in mice. In addition, our insect cells used for in vitro experiments were C6/36 Aedes albopictus cells, and we infected Ae. aegypti mosquitoes in vivo. The infection results between in our in vitro and in vivo insect experiments with our E1 mutants suggest that there could be species differences between Aedes mosquitoes. Interestingly, CHIKV E1-M88L led to increased viral titers in the footpad of Mxra8-deficient mice, suggesting that E1-M88L can facilitate entry in the absence of this receptor. These results are in line with the increased cell binding and infectivity we observed in vitro. While E1-M88L was dependent on Mxra8 and attenuated in NIH 3T3 cells in vitro, the complex in vivo environment could suggest that E1-M88L is able to better infect other cell types or bind to other host molecules for entry. Future studies exploring the broad cell tropism of CHIKV E1-M88L in mammals and mosquitoes will be critical in mapping out how this fusion loop residue contributes to entry.

Finally, the molecular dynamic simulations have been a powerful tool to dissect how CHIKV glycoprotein changes may alter the glycoprotein complex. Importantly, we observed that modeling changes in CHIKV E1 domain II, the fusion loop and β-strand c, led to changes in not only E1, but also E2 dynamics, supporting what we see experimentally. In addition to these studies, we have observed changes in cell binding and GAG interactions with CHIKV E1 variants in the E1-E1 interface (26), supporting the idea that changes in E1 can contribute to glycoprotein structure and dynamics. It will be crucial to understand how E1 and E2 work together using complementary in silico and laboratory experiments, as this will help us better understand the mechanisms of CHIKV entry.

Alphaviruses are a diverse genus that exhibits high epidemic potential due to their prevalence and transmission globally. These viruses are a leading cause of arthritic and encephalitic disease, yet there are limited antiviral therapeutics available. Taken together, our studies highlight that understanding the molecular mechanisms of how alphaviruses infect both insects and mammals, as well as interact with host molecules and specific cell types during infection, is essential to our understanding of viral emergence.

MATERIALS AND METHODS

Cell lines

Baby hamster kidney cells (BHK-21, ATCC CCL-10) and NIH-3T3 cells [a gift from Dr. Michael Diamond at Washington University (25)] were grown in Dulbecco’s modified Eagle medium (DMEM, Corning) with 10% fetal bovine serum (FBS; Atlanta Biologicals), 1% nonessential amino acids (NEAA, Fisher Scientific), and 10 mM HEPES (Invitrogen). Vero cells (ATCC CCL-81) were grown in DMEM with 10% newborn calf serum (NBCS, Gibco). Mammalian cells were maintained at 37°C with 5% CO₂. Aedes albopictus cells (C6/36, ATCC CRL-1660) were grown in L-15 media (Corning) supplemented with 10% FBS, 1% NEAA, and 1% tryptose phosphate broth (Invitrogen), and were maintained at 28°C with 5% CO₂. All cell lines were confirmed mycoplasma free.

Biosafety

All work with CHIKV and CHIKV E1 variants were completed under Biosafety level 3 (BSL3) conditions at the NYU Grossman School of Medicine. Mouse and mosquito work were completed in the NYU Grossman School of Medicine ABSL3 animal and insect facility. Animal work was complaint with NYU Grossman School of Medicine Institutional Animal Care and Use Committee (IACUC) protocol #IA16-01783.

Viruses

Wild-type CHIKV (strain 06-049), E1 glycoprotein variants, and viral derivatives expressing ZsGreen or Firefly luciferase reporter proteins under the subgenomic promoter were produced from infectious clones as previously described (34). E1 glycoprotein variants were introduced in the wild-type CHIKV infectious clone by site-directed mutagenesis using the following primers: E1-M88L; Forward GTCTACCCATTTTTGTGGGGCGGCG, Reverse CGCCGCCCCACAAAAATGGGTAGAC, E1-N20Y; Forward GTATAAGACTCTAGTCTATAGACCTGGCTACAG, Reverse CTGTAGCCAGGTCTATAGACTAGAGTCTTATAC. The E1 variants were introduced into an infectious clone expressing ZsGreen or Firefly luciferase by subcloning the XhoI/NotI restriction fragment from the unmarked CHIKV clone into the same restriction sites of the reporter plasmids. All plasmids were Sanger sequenced to confirm the E1 variants and with no second-site mutations.

To generate infectious virus, 10 µg of each infectious clone plasmid was linearized overnight with NotI restriction endonuclease (Invitrogen), purified by phenol:chloroform extraction and ethanol precipitation, and resuspended in nuclease-free water (20, 34). CHIKV RNA was in vitro transcribed using the SP6 mMessage mMachine kit (Invitrogen) following the manufacturer’s instructions. RNA was purified by phenol:chloroform extraction and ethanol precipitation, diluted to 1 µg/µL in nuclease-free water, aliquoted, and stored at −80°C. BHK-21 cells (10⁷ cells/mL) were electroporated with 10 µg of in vitro transcribed RNA by 1 pulse of 1,200 V, 25 Ω, and infinite resistance. Electroporated cells were added to a T25 flask in complete media (DMEM, 10% FBS, 1% NEAA, and 1% HEPES). After incubation at 37°C for 72 h, the passage 0 (p0) supernatant was centrifuged at 1,200 rpm for 5 min and was used to infect a T-175 flask of BHK-21 cells to produce a passage 1 (p1) working stock. After incubation at 37°C for 24 h, p1 virus was centrifuged at 1,200 rpm for 5 min, aliquoted, and stored at −80°C. Viral RNA was extracted, and all viruses were Sanger sequenced to address genetic stability as described below. Infectious virus titers were quantified by plaque assay for all stocks as described below. Ultracentrifugation was used to generate purified virus stocks. Briefly, viruses were pelleted over a 20% sucrose cushion by centrifugation at 25,000 rpm for 4 h. Purified virus particles were resuspended in media consisting of DMEM containing 2% FBS. Viral titers were quantified by plaque assay, and viral genomes were quantified by RT-qPCR as described below.

Plaque assay

Infectious virus was quantified via plaque assay. Supernatants containing virus were diluted 10-fold in DMEM and added to a monolayer of Vero CCL81 cells (300,000 cells/well) in a 12-well plate. After a 1 h incubation at 37°C, media comprised of DMEM, 2% FBS, and 0.8% agarose was added to each well and incubated at 37°C for 72 h. For all mosquito and mouse samples, 1% antibiotic-antimycotic (Gibco) was added to the media. Cells were fixed with 4% formalin for 1 h, agarose plugs were removed, and plaques were visualized using crystal violet. Viral titers were calculated using the lowest countable dilution.

RNA extractions and RT-qPCR

CHIKV RNA for qPCR assays was extracted and purified using TRIzol reagent (Fisher Scientific) following the manufacturer’s protocol. To quantify the number of viral RNA genomes/mL, a TaqMan RNA-to-CT one-step kit (Applied Biosystems, Fisher Scientific) was used with the following primers targeting the nsP4 region of the genome: Forward 5′ TCACTCCCTGCTGGACTTGATAGA 3′. Reverse 5′ TTGACGAACAGAGTTAGGAACATACC 3′. Probe 6919-FAM 5′ AGGTACGCGCTTCAAGTTCGGCG 3′. A standard curve was generated from 10-fold dilutions of in vitro transcribed RNA was included for all samples and all were run in technical duplicate.

Sanger sequencing

Viral RNA was extracted from stocks as described above. cDNA was generated using a Maxima H Minus Enzyme First Strand cDNA Synthesis Kit (Thermo Scientific) following the manufacturer’s instructions. The CHIKV genome was amplified in 5 PCR fragments using the primers listed in Table 1 with the Phusion High-Fidelity PCR Master Mix Kit with HF Buffer (Thermo Scientific). CHIKV fragments were confirmed by an agarose gel and sent to Sanger Sequencing using the primers in Table 1 for confirmation of mutations in all working CHIKV viral stock mutations (Genewiz).

TABLE 1.

PCR and sequencing primers used in this study

CHIKV fragment	Primer sequence (5′ to 3′)	Primer use
F1	AGCAGACGTCGCTATATAC	Phusion PCR
F1	TCTCAGTAGGGTCAACGC	Sanger Sequencing
F1	GAGGACTAGAATCAAATGG	Sanger Sequencing
F1	GAAGCAGAGGAAGAACGAG	Sanger Sequencing
F1	CACCGACGTGATGAGAC	Sanger Sequencing
F1/F2	CACTGTTCTTAAAGGACTC	Phusion PCR/Sanger Sequencing
F2	GCATACTCACCTGAAGTAGCC	Phusion PCR
F2	ACTGAATGCAGCCTTCGTAG	Sanger Sequencing
F2	GATGAGCACATCTCCATAG	Sanger Sequencing
F2	CATTACATTTGGGGACTTCAAC	Sanger Sequencing
F2	CCTCTCTTTAGTCTCTGG	Sanger Sequencing
F2	CACTACAGGAAGTACCAATGG	Sanger Sequencing
F2	CCACATAGGTATGCTGTCGCC	Sanger Sequencing
F2	AGCAGTAAGCGCAAGTGAATC	Phusion PCR
F2.5	TCATCATACCAAATTACCGACG	Phusion PCR
F2.5	TATTTAGGACCGCCGTACAAAG	Sanger Sequencing
F2.5	GGTTTCATTACTTTGTCCCCC	Phusion PCR
F3	TTAAACTGGGCAAACCGC	Phusion PCR
F3	GAGGCCGGTTCACCATCC	Sanger Sequencing
F3	GTTTCCTCCGGTTCCTTTTC	Sanger Sequencing
F3	GAAGCGACAGACGGGACG	Sanger Sequencing
F3/F4	GCCTCTTGGTATGTGGCCGC	Phusion PCR/Sanger Sequencing
F4	CAACGAGCCGTATAAGTATTGG	Phusion PCR
F4	GAACGAGCAGCAACCTTTG	Sanger Sequencing
F4	CTATGCAACTGTCTGAG	Sanger Sequencing
F4	TACGAACACGTAACAGTGATCC	Sanger Sequencing
F4	GCACCATCTGGCTTTAAGTATTG	Sanger Sequencing
F4	CTATTCAGGGGTTGCGTAG	Phusion PCR

Cell-Insight CX7 quantification of infected cells

Cells infected with CHIKV-ZsGreen virus were fixed with 4% paraformaldehyde (PFA) for 1 h at room temperature (20). Following incubation, fixed cells were stained with 4′,6-diamidino-2-phenylindole (DAPI, Thermo Scientific) diluted 1:1,000 in phosphate-buffered saline (PBS, Corning) for 1 h at room temperature. A Cell-Insight CX7 high-content microscope (Thermo-Scientific) was used to quantify the percent of infected cells in each well, by dividing the number of ZsGreen positive cells with the total number of DAPI stained cells.

Western blotting

BHK-21 cells were electroporated with 10 µg of in vitro transcribed RNA and seeded in a T-25 flask for 48 h at 37°C. After incubation, cells were washed once with PBS and scraped off the flask into cold PBS. Cells were pelleted by centrifugation at 1,200 rpm for 5 min at 4°C and resuspended in lysis buffer [1× Tris-buffered saline (TBS; 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% TX-100, and 1× Halt protease inhibitor cocktail (Thermo) for 1 h on ice]. After lysis, debris was removed by centrifugation at 10,000 × g for 5 min and mixed 1:1 with 2× Laemmli buffer containing 10% β-mercaptoethanol. Samples were boiled at 95°C for 10 min and centrifuged at 10,000 × g for 5 min. Proteins were separated by SDS-PAGE on a Mini-PROTEAN TGX Stain-Free pre-cast polyacrylamide gel (Bio-Rad), transferred to a hydrophobic polyvinylidene fluoride membrane (Immobilon), and incubated overnight at 4°C in blocking buffer comprised of 1× TBS-T (TBS + 0.1% Tween-20) and 5% dry milk. Blots were then incubated for 2 h at room temperature with primary CHIKV antibodies including anti-rabbit CHIKV-E1 (gift from Dr. Gorben Piljman), anti-mouse CHIKV-E2 (CHIK-48, BEI Resources), anti-rabbit CHIKV-Capsid (gift from Dr. Andres Merits), anti-rabbit CHIKV-nsP1 (gift from Dr. Andres Merits), and anti-mouse Actin (MA5-11869, ThermoFisher). After multiple washes with 1× TBST, blots were incubated with anti-rabbit or anti-mouse IgG HRP secondary antibodies (Invitrogen) for 1 h at room temperature. After another round of washes, blots were developed using a SuperSignal West Pico plus chemiluminescence substrate kit (Thermo) and imaged using a ChemiDoc MP imaging system (Bio-Rad). Images were analyzed and quantified using Image Lab (Bio-Rad).

Growth kinetic analysis

BHK-21 cells (50,000 cells/well) or C6/36 cells (200,000 cells/well) were seeded in a 24-well plate 24 h prior to infection. Cells were infected with each virus diluted in DMEM at an MOI of 0.1 for 1 h at 37°C (BHK-21) or 28°C (C6/36). After incubation, cells were washed twice with PBS, and complete BHK media or insect media was added to the cells. Supernatants were collected from each corresponding timepoint, and infectious virus was quantified via plaque assay as described above.

Infectivity assays

BHK-21 cells (15,000 cells/well) were seeded in a Costar 96-well plate and incubated at 37°C 24 h prior to infection. After the seeded plate was incubated at 4°C for 1 h, ZsGreen expressing viruses were added to the cells at an MOI of 1 and incubated at 4°C for another hour. A final concentration of 20 mM ammonium chloride (NH₄Cl) was diluted in complete media, then added to each well at its corresponding timepoint and incubated overnight at 37°C. The following day, infected cells were fixed with 4% PFA and quantified by the CX7 as stated above.

For C6/36 cells (100,000 cells/well), cells were seeded in a 96-well plate at 28°C for 24 h. Cells were then infected with WT CHIKV or an E1 mutant virus at an MOI of 0.1 and incubated at 28°C for 1 h. A final concentration of 20 mM ammonium chloride was diluted in insect cell media and treated on the cells for incubation at 28°C. After 24 h, cells were fixed with 4% PFA, stained with DAPI, and quantified using a CX7 high-content microscope.

Virus-binding assay

BHK-21 cells (100,000 cells/well) or C6/36 cells (350,000 cells/well) were seeded in a 12-well plate and incubated the day before infection at 37°C for mammalian cells or 28°C for insect cells. Cells were pre-treated with binding buffer media compromised of DMEM, 0.2% BSA, 10 mM HEPES, and 20 mM NH₄Cl at 4°C for 1 h. After incubation, cells were washed once with PBS and then incubated with purified virus at an MOI of 100 (based on RNA genomes) diluted in binding buffer and left for 30 min on ice. Virus media was removed, and cells were washed with cold PBS three times. Cells were then collected in Trizol and RNA was extracted to quantify viral RNA genomes via RT-qPCR as described above.

Cholesterol-depletion assay

BHK-21 cells (20,000 cells/well) were seeded in a 96-well plate and incubated for 24 h at 37°C. Cells were pre-treated with increasing concentrations of methyl-β-cyclodextrin (MβCD) for 1 h at 37°C. Following the incubation, the cells were washed once with PBS and then infected with each corresponding CHIKV-ZsGreen virus at an MOI of 1 at 37°C for 1 h. A final concentration of 20 mM ammonium chloride was added to the cells and was then incubated at 37°C for 24 h. Infected cells were fixed with 4% PFA, stained with DAPI, and quantified using a CX7 high-content microscope.

Luciferase assay

BHK-21 (20,000 cells/well) were seeded in a 96-well plate 1 day prior to transfection. Cells were transfected with 10 μg of WT CHIKV, CHIKV nsP4-GNN, or E1-M88L RNA using Lipofectamine 2000. Lipofectamine was mixed with Opti-MEM (Thermo Scientific) and incubated at room temperature for 10 min. This was then added to a mixture of RNA and Opti-MEM at a 1:2 ratio of Lipofectamine to RNA. After incubating for 5 min at room temperature, 10 µL of this mix was added to cells and was incubated at 37°C. At 4, 6, 8, and 24 hours post-transfection, supernatants were collected and cells were lysed with Steady-Glo firefly buffer reagent (Pierce) in a white 96-well plate. Collected supernatants were used to infect BHK-21 cells (20,000 cells/well) for 24 h, and luciferase activity was read as before. Luminescence was read using a SpectraMax M3 plate reader (Molecular Devices), and raw data were quantified as relative luminescence.

NIH 3T3 virus growth and infectivity assays

Control NIH-3T3 cells (Control + Vector), 3T3 ΔMxra8 receptor knockout cells (ΔMxra8 + Vector), and 3T3 ΔMxra8 cells overexpressing Mxra8 (ΔMxra8 + Mxra8) in trans were seeded at 20,000 cells/well in a 96-well plate 1 day prior to infection (25). Cells were infected with a purified WT CHIKV-ZsGreen or E1-M88L-ZsGreen at an MOI of 5 and incubated at 37°C for 1 h. Cells were then either given 20 mM ammonium chloride with complete media and fixed after 24 h post-infection (hpi) or were given complete media and fixed at 48 hpi. Supernatants were collected at both 24 and 48 hpi from both experiments and were used for a plaque assay to determine infectious particles. Once cells were fixed, they were stained with DAPI for 30 min and then imaged and quantified using a Cell-Insight CX7 high-content microscope.

Virus neutralization assays

The RRV-12 human CHIKV monoclonal antibody was a gift from Dr. James Crowe at Vanderbilt University (35), and the CHK-263 mouse monoclonal antibody was obtained through BEI Resources (NR-44003). The human Mxra8 (mouse Fc) was purchased from Native Antigen (REC31649), and the mouse Mxra8 (human Fc) was a gift from Dr. Michael Diamond (25). In brief, two-fold serial dilutions of each antibody, Mxra8 protein, or BSA as a control were made in DMEM containing 2% FBS. Antibody, Mxra8, or BSA mixes were incubated with 5,000 PFU of CHIKV-ZsGreen viruses for 1 h at 37°C. Following incubation, virus-protein mix was added to a 96-well plate containing pre-seeded 10,000 BHK-21 cells/well and incubated for 18 h at 37°C. Plates were then fixed and stained, and infected cells were quantified as described above.

ELISA assays

CHIKV ELISA assays were performed similar to those in Zhang et al. (22). For mouse Mxra8 ELISAs, Maxisorp plates (ThermoFisher) were incubated with 2 mg/mL of mouse CHK-263 and mouse CHK-48 (BEI Resources; NR-44002) in coating buffer (ThermoFisher) overnight at 4°C. For human Mxra8 ELISAs, Maxisorp plates were incubated with 2 μg/mL of human RRV-12 antibody in coating buffer (ThermoFisher) overnight at 4°C. The next day, plates were washed four times with PBS and blocked for 2 h with PBS containing 4% BSA at room temperature. Each variant was diluted to 2.0 × 10⁶ PFU/mL in PBS containing 2% BSA and incubated with each condition for 1 h at room temperature. Plates were then washed five times with PBS and incubated with 10 μg/mL of human anti-WNV monoclonal antibody (BEI Resources; NR-31042), mouse Mxra8 (human Fc), human Mxra8 (mouse Fc), or human RRV-12 for 1 h at room temperature. Plates were washed four times with PBS and incubated with goat anti-human-HRP secondary antibody (1:2,000) or goat anti-mouse-HRP secondary antibody (1:2,000) in PBS with 2% BSA for 1 h at room temperature. Plates were then washed five times with PBS, and 100 μL of TMB reagent was added for 30 min, followed by the addition of 2N H₂SO₄. Plates were then read at 450 nM using a SpectraMax M3 plate reader.

Heparin-agarose pull-down assay

Heparin-agarose (Fisher) and Protein A/G-agarose (Pierce) beads were washed twice with PBS and once with DMEM containing 2% FBS. Sucrose gradient-purified viruses were diluted to 10⁶ PFU/mL in DMEM containing 2% FBS and incubated with 100 μL of each washed bead mix for 1 h at room temperature with continuous spinning. After incubation, beads were isolated by centrifugation at 10,000 × g for 1 min and washed twice with cold PBS. Following the second wash, Laemmli buffer containing 10% beta-mercaptoethanol was added, the beads boiled for 10 min, and proteins analyzed by SDS-PAGE and western blotting as described above. Ten percent of the initial input virus was added directly to Laemmli buffer as an input control.

Mouse infections

Wild-type C57BL/6J (Jackson Laboratory) and Mxra8^Δ8/Δ8 mice [a gift from Dr. Michael Diamond (22)] were bred and maintained in the NYU Grossman School of Medicine BSL2 vivarium and moved to the ABSL3 facility prior to and during infection. Mxra8^Δ8/Δ8 mice were rederived from sperm at the NYU Grossman School of Medicine. WT and ΔMxra8 littermate controls were generated by first crossing Mxra8^Δ8/Δ8 mice to WT C57BL/6J mice to generate heterozygote Mxra8^+/Δ8 mice and subsequently crossing these heterozygotes to obtain WT and Mxra8^Δ8/Δ8 mouse lines. All mice were genotyped (Transnetyx) prior to each experiment.

For virus infections, male and female 5- to 7-week-old mice were anesthetized using isoflurane and infected with 1,000 PFU of WT CHIKV or the E1-M88L variant via footpad injection. At each corresponding timepoint, mice were euthanized via CO₂ inhalation. Serum, ipsilateral footpad, and ipsilateral muscle from each mouse were harvested and placed in separate 2 mL tubes containing 500 µL PBS and two 5 mm stainless steel balls. Samples were homogenized using a Tissue-Lyser II (Qiagen) for 2 min at 30 Hz, centrifuged at 8,000 rpm for 8 min, and then used to quantify infectious particles via plaque assay.

Mosquito infections

Aedes aegypti (Poza Rica, Mexico) mosquitoes were reared and maintained in a 28°C incubator with 70% humidity in an ABSL3 insectary (26, 36). Mosquitoes 4–7 days post-emergence were infected with virus via blood meal through a pig intestine membrane. Each virus was mixed with PBS-washed defibrinated sheep blood (Fisher Scientific) containing 5 mM ATP to a concentration of 10⁶ PFU/mL and fed to ~50–60 female mosquitoes. After 1 h, engorged mosquitoes were sorted and maintained for 7 days, after which their bodies, legs, and wings were separated into 2-mL tubes containing 200 µL PBS and one 5-mm stainless steel balls. Samples were then homogenized and ground using a Tissue-Lyser II (Qiagen) for 2 min at 30 Hz, centrifuged at 8,000 rpm for 8 min, and then used to quantify infectious particles via plaque assay.

Molecular dynamic simulations

To study the impact of E1-M88L mutation in the envelope glycoproteins function at a molecular level, we performed unbiased atomistic Molecular Dynamics simulations in explicit solvent. As an initial structure, we used the pre-fusion conformation of the E1-E2 heterodimer (PDB: 3N42). To simulate the variants, we introduced the point mutations in the WT protein using PyMol (Schrödinger L, DeLano W. PyMOL. In: http://www.pymol.org/pymol). All simulations were done using the GROMACS 2021.2 software package (37). For each variant, the protonation state at pH 7 of all protonable residues was determined using PROPKA (38). The Amber99SB*-ILDN force field (39, 40) was employed to describe the protein. A cutoff value of 10 Å was applied for short-range electrostatic and van der Waals interactions. Long-range electrostatic interactions were treated with PME (Particle Mesh Ewald). The system was solvated with a dodecahedral box of TIP3P water (41). The water box extended 12 Å from the protein surface and the system was neutralized with 0.15 M NaCl. The system was minimized using the Steepest Descent algorithm. Subsequently, it was heated to 310 K and equilibrated for 200 ps in the NVT ensemble using the V-rescale thermostat (42). A second equilibration step was performed for 1 ns in the NPT ensemble using the Berendsen barostat (43). Finally, a production run was conducted in the NPT ensemble using the V-rescale thermostat and the Parrinello-Rahman barostat (44). In both equilibration and production runs, the LINCS algorithm (45) was used and a time step of 2 fs was employed. For each variant, we performed two independent runs of 250 ns starting from the same minimized structure. All simulations were performed on the high-performance computing centers of NYU Langone and CCAD (https://ccad.unc.edu.ar/).

We analyzed the final 200 ns of each MD run. We concatenated the trajectory of both replicas of all variants and we performed a PCA over the concatenated trajectory. In order to compare the conformational space sampled by each variant, we projected the trajectory of each one on PC1 and PC2. Using the simulation done with E1-E2 heterodimer, we performed the PCA over both E1 and E2 or over E2 alone. The pre-processing of the trajectories and the PCA analysis were done using GROMACS tools. The 2D projections and graphics were done using an in-house Python script using Pandas, Scipy, Seaborn, and Matplotlib packages.

Protein structures and sequencing analysis

The CHIKV E1 and E2 glycoproteins (PDB: 3N42) were rendered in PyMol (Version 2.5.2). CHIKV nucleotide sequencing alignments were generated with SeqMan Ultra (DNA Star), and alphavirus amino acid alignments were generated with MegAlign Pro (DNA Star). The following alphavirus were used for E1 amino acid alignments: CHIKV (06-049; AM258994), Mayaro virus (NP_740694), Sindbis virus (AWT57845), Ross River virus (QTC33398), Eilat virus (YP_006732328), Semliki Forest virus (CAA52444), Middelburg virus (AAL35777), O’nyong n’yong virus (YP_010775618), Aura virus (AWQ38331), Una virus (AAL35783), western equine encephalitis virus (QEX51909), eastern equine encephalitis virus (AAU95735).

Statistics and data analysis

All data were analyzed using GraphPad Prism (Version 9). All in vitro experiments were completed with at least two biological replicates and internal technical duplicates (exact details can be found in the figure legends). All in vivo experiments were completed with at least n = 6 for mice and n = 43 for mosquito infections. A P-value < 0.05 is considered statistically significant.

DATA AVAILABILITY

No new data sets were generated in this study.