Introduction

Eastern equine encephalitis virus (EEEV) is a member of the Alphavirus genus, which is comprised mostly of mosquito-borne positive-sense single-stranded RNA viruses in the family Togaviridae1. EEEV is endemic in North America. EEEV disease in humans is uniquely neurovirulent among alphaviruses, marked by limited febrile prodrome and rapid onset of encephalitis with a case fatality rate of 30–70% and a high rate of long-term neurological sequelae1. Currently, there are no licensed vaccines or therapeutics to prevent or treat human EEEV infections. Similar to humans, mice infected with EEEV show limited signs of a prodrome, often with encephalitis being the first signs of disease before fatalities2,3,4. Consistent with the disease signs, EEEV infection exhibits limited replication in lymphoid tissues, does not induce a robust systemic innate immune response, and replication is poorly controlled in central nervous system (CNS) tissues3,4. These phenotypes have been linked, in part, to the ability of wild-type (WT) EEEV to efficiently bind heparan sulfate (HS) proteoglycan attachment receptors on cells3,4,5,6.

To date, EEEV is the only arbovirus for which amino acid residues in attachment proteins of naturally circulating, unpassaged, viruses have been shown to promote efficient binding to HS3,7. Previous infection-based studies identified positively charged lysine (K) residues (E2 K71, K74, and K77) in a cluster in domain A of the EEEV E2 glycoprotein as critical for HS-dependent infection3,4. Subsequently, cryo-electron microscopy (cryo-EM) identified two sites involving lysine (K) and arginine (R) residues in the EEEV E2 protein that bind a low-molecular-weight HS analog, 6 kilodalton (kDa) heparin, as putative HS-binding sites (E2 R84 and R119 coordinating with HS in one site and K156, R157 in another)8. However, it remains unclear whether the heparin-binding sites identified by cryo-EM contribute to HS-dependent infection or replication of EEEV in vertebrates or mosquito vectors.

More recently, several low-density lipoprotein (LDL) binding proteins, including the very-low-density lipoprotein receptor (VLDLR), LDLR, and apolipoprotein E receptor 2 (ApoER2), were identified as EEEV attachment and entry receptors9,10. Unexpectedly, the positively charged residues in the K156 and R157 HS-binding site were among the VLDLR-E2 contact residues mapped by structural analysis11,12,13. It has been hypothesized that enhanced infectivity promoted by HS proteoglycans results from concentration of virus particles on cell surfaces such that interaction with bona fide “entry receptors” is non-specifically enhanced by HS7,14,15,16. However, the relationship of HS proteoglycan attachment receptors to protein attachment and entry receptors is unknown for RNA viruses.

Using charged-to-alanine (A) mutagenesis, we show that each of the three HS-binding sites on E2 contributes to EEEV-HS interactions at the cell surface, although to different degrees; that each site may interact with HS in a functionally unique manner; and that mutation at each of the three HS-binding sites decreases neurovirulence in mice and mosquito infectivity. Unexpectedly, mutagenesis of each HS-binding site either diminishes or abrogates interactions with VLDLR, ApoER2, and LDLR receptors as measured by infectivity assay and binding to receptor-overexpressing cells, or by competition with a VLDLR-derived infection inhibitor both in vitro and in vivo. Thus, the sites present in naturally circulating EEEV glycoproteins are polyfunctional, coordinating interactions with both HS and protein receptors. Notably, abrogation of known protein receptor interactions with a 156–157 (E2 K156A and R157A) mutant eliminated oral mosquito infection, but did not diminish mosquito cell infectivity in vitro and had only a modest effect on virulence in mice, suggesting the presence of additional entry mediating factors in both viral hosts. Finally, mutations that enhanced both protein receptor- and HS-dependent infectivity could be selected through in vitro passage of attenuated chimeric mutant EEEV viruses. Together, these data suggest structural and/or functional convergence between EEEV engagement with HS and protein receptors and provide a potential mechanism for historical observations of in vitro selection of single-site substitution mutations that confer HS-binding17,18.

Results

Mutations to each HS interaction site of E2 independently alter EEEV-HS interactions

Previously, residues of EEEV E2 (K71, K74, and K77) were shown to be critical for EEEV-HS interactions, and a charge-to-alanine mutant, referred to as EEEV 71–77, ablated EEEV-HS interactions3,4. We, with collaborators, recently demonstrated that residues in two other sites of E2 (R84 and R119 or K156 and R157) interact with 6 kDa heparin8. To determine the extent that residues in these newly identified E2 sites contribute to EEEV-HS interactions and how they compare to the 71–77 mutant, we tested the infection characteristics of charged-to-alanine mutants at each HS/heparin-binding site with a series of HS infectivity dependence assays.

Using chimeric viruses derived from a cDNA clone encoding Sindbis virus (SINV) nonstructural proteins and RNA replication control structures19,20, and expressing EEEV structural proteins (SINV/EEEV), we compared the infectivity of WT and each mutant on WT CHO-K1 cells versus glycosaminoglycan (GAG)-deficient CHO-pgsA-745 and HS-deficient CHO-pgsD-677 cells21. Consistent with previous results3,4, WT SINV/EEEV infectivity was significantly reduced (p < 0.001) in the absence of all GAGs or HS, whereas the 71–77 mutant was unaffected. The 84–119 (E2 R84A, R119A) mutant exhibited reduced infectivity like the WT, whereas the 156–157 mutant was similar to 71–77, showing no significant reduction in infectivity for the mutant cells (Fig. 1A and Supplementary Fig. 1A). GAG-dependent infectivity was recapitulated in cellular binding assays, with WT and the 84–119 mutant showing significantly decreased binding to CHO-psgA-745 compared to CHO-K1 cells. The 71–77 and 156–157 mutants exhibited decreased binding to CHO-K1 cells compared to the WT virus and similar binding to both cell lines (Fig. 1B). We then compared the infection rates of each mutant when incubated in media with increased NaCl concentration, which disrupts ionic interactions such as protein-HS binding22,23,24,25. Consistent with previous data4, the WT showed a significant decrease in infectivity starting at the lowest concentration of supplemented NaCl, whereas the 71–77 mutant was not susceptible to ionic disruption at the highest concentration tested (Supplementary Fig. 1B). The 84–119 and the 156–157 mutants exhibited intermediate phenotypes, with significantly reduced infectivity compared to the medium control but higher infectivity compared to WT at the higher NaCl concentrations (Supplementary Fig. 1B).

Fig. 1: E2 mutations independently and uniquely alter EEEV-HS interactions in vitro.
figure 1

A Relative infectivity of WT or mutant SINV/EEEV for CHO-K1, GAG-deficient CHO-pgsA-745, and HS-deficient CHO-pgsD-677 cells (n = 9 from 3 independent experiments). B Binding of chimeric WT and mutant viruses to CHO-K1 and CHO-pgsA-745 cells (n = 6 from 2 independent experiments, except for SINV/EEEV WT where n = 9 from 3 independent experiments). Viruses were allowed to incubate with cells on ice for 75 min, then cells were washed, and the bound virus was quantified by ddPCR as a ratio of genomes bound to cells versus the host Mmadhc gene. C Relative infectivity of WT or mutant SINV/EEEV on BHK-21 cells treated with heparinases (n = 6 from 2 independent experiments). D Relative indirect binding to heparin-agarose beads by WT and mutant SINV/EEEV. Viruses were incubated with collagen-agarose or heparin-agarose beads; unbound viruses were titered on BHK-21 cells (n = 6 from 2 independent experiments). E Apparent binding affinity showing the nonlinear fit of WT or mutant SINV/EEEV for heparin, with KD apparent and Hill slope (h) as a measure of cooperativity, where h > 1 shows increased cooperative binding and h < 1 shows decreased cooperative binding (n = 6 for mutants and n = 5 for WT from 2 independent experiments). F Genome-to-BHK PFU ratios for EEEV WT and mutant viruses (n = 3 independent experiments). G BHK-21 cells were infected with equal genomes of EEEV WT and mutant EEEVs, corresponding to a multiplicity of infection of 1 for WT (n = 6 from 2 independent experiments). Limit of detection indicated with dashed line. Means ± SD are shown. Significance determined by (A–C) two-way ANOVA with Tukey’s post-hoc tests, D Brown-Forsythe and Welch ANOVA with Dunnett’s post-hoc tests, F one-way ANOVA with Tukey’s post-hoc tests on log-transformed data, or G two-way repeated measures ANOVA with Tukey’s post-hoc tests. Exact p-values are indicated. Source data provided as a Source Data file.

To determine if mutations at each site impact qualitative aspects of the EEEV interaction with cell surface HS, we examined the infectivity of chimeric WT and each mutant for BHK-21 cells following treatment with different microbial heparinases, which target different HS modifications based on sulfonation patterns26. Infectivity of the WT SINV/EEEV and the 84–119 mutant was significantly reduced following each heparinase treatment, but WT was more susceptible to heparinase II or III treatments whereas the 84–119 mutant was more susceptible to heparinase I or II treatments (Fig. 1C). In contrast, the infectivity of the 71–77 mutant was not decreased following treatment with any heparinase. Although the 156–157 mutant was not significantly decreased in infectivity or binding for GAG-deficient CHO cells (Fig. 1A, B), it was susceptible to heparinase II and III treatment like the WT (Fig. 1C). Together, CHO cell infectivity, salt disruption, and heparinase digestion data suggest that each of the three mutations alter EEEV-HS interactions in a unique manner.

To determine whether the mutations impacted the ability of virions to bind to HS/heparin in the absence of other cellular receptors, we performed indirect binding assays using heparin-agarose beads (a highly sulfonated analog of HS27) with collagen-agarose beads serving as a control. Decreased infectivity for viruses incubated with heparin versus collagen beads demonstrates increased heparin binding. The WT bound most efficiently of the four viruses to the heparin-agarose beads, with the 84–119 mutant binding slightly, but not significantly, less. The 71–77 and 156–157 mutants exhibited significantly reduced binding to heparin beads compared to WT SINV/EEEV (Fig. 1D). When direct binding to heparin was assessed via ELISA, the WT and 84–119 mutant showed similar apparent KD and cooperative binding, whereas both 71–77 and 156–157 mutants exhibited significantly reduced strength (p < 0.001 for 71–77 and p < 0.01 for 156–157 by Kruskal–Wallis test) and cooperativity of binding (p < 0.001 for 71–77 and 156–157 by Kruskal–Wallis test) compared to the WT (Fig. 1E).

Increased HS-binding of cell-adapted viral mutants has been associated with increased infectivity for cells and decreased genome/particle-to-plaque forming unit (PFU) ratios6,25,28. We determined the genome-to-PFU ratios of each virus on BHK-21 cells. The 84–119 mutant was similar to the WT, whereas the ratios for the 71–77 and 156–157 mutants were approximately 100-fold higher than the WT (Fig. 1F), suggesting reduced infectivity per particle. To ensure that all mutants could replicate effectively, single-step replication curves of EEEV viruses were performed using BHK-21 cells (Fig. 1G). Overall, the 84–119 mutant replicated similarly to the WT EEEV; however, there was a significant decrease at 6 h post-infection (hpi) (Fig. 1G). The 71–77 and 156–157 mutants clustered together, replicating more slowly and to a significantly lower titer than the WT throughout the replication curve (Fig. 1G). However, due to differences in genome-to-PFU ratios and reduction in HS-dependent infectivity of the 71–77 and 156–157 mutants for BHK-21 cells (Fig. 1F), plaque titers exaggerate replication differences and differences in HS interactions most likely explain the clustering of the viruses into two groups.

Mutations that disrupt E2 HS binding sites also interfere with EEEV-protein receptor interactions

Previously, contact residues were mapped for chikungunya virus (CHIKV) and Venezuelan equine encephalitis virus (VEEV) binding to the matrix remodeling associated 8 (MXRA8) and low-density lipoprotein receptor class A domain 3 (LDLRAD3) receptors, respectively29,30,31,32. At least one of the residues of all three EEEV HS-binding sites is either immediately adjacent to or directly overlaps contact residues for either or both receptors (Supplementary Fig. 2). Furthermore, two recent structural studies that characterized the molecular basis of EEEV-VLDLR interactions, identified the E2 K156 and R157 residues as critical for engagement with the LA1 domain of VLDLR, making these residues central to a shared receptor binding motif across multiple alphaviruses11,12,13. We hypothesized that all three HS-binding mutations, especially the 156–157 mutation, may impact the ability of EEEV to interact with one or more protein receptors. We first determined whether the E2 mutations affected SINV/EEEV infectivity for K562 cells ectopically expressing either an empty vector (EV), VLDLR or ApoER2 isoform 1 or 29 (Fig. 2A), or THP-1 cells expressing an EV or LDLR10 (Fig. 2B). Infectivity of the 156–157 mutant and, to a lesser degree, the 71–77 and 84–119 mutants was significantly decreased versus the WT for cells over-expressing VLDLR, LDLR, or ApoER2 isoform 1 and 2, with the 84–119 mutant showing least decrease in infectivity compared to the 71–77 and 156–157 mutants (Fig. 2A, B). The infectivity results (Fig. 2A) were largely recapitulated when virus binding to K562-EV and K562-VLDLR cells was assessed, with the WT SINV/EEEV showing significantly higher binding to the K562-VLDLR versus K562-EV cells (Fig. 2C). The binding of the 71–77 and 156–157 mutants was not significantly different on the two cell types and was significantly decreased with K562-VLDLR cells compared to the WT (Fig. 2C). However, the 84–119 mutant showed a reproducible but non-significant increase in binding to the K562-VLDLR cells (Fig. 2C). Given the significant difference in infectivity (Fig. 2A), this non-significant decrease for the 84–119 mutant binding to K562-VLDLR cells (Fig. 2C) may reflect limited sensitivity of the binding assay. While this manuscript was in preparation, another group reported that the K156A mutation disrupts VLDLR and ApoER2 binding12.

Fig. 2: E2 mutations that disrupt E2 HS-binding sites also interfere with EEEV-protein receptor interactions.
figure 2

A Infectivity of K562 cells expressing empty vector (EV), VLDLR, ApoER2 isoform1, or ApoER2 isoform2 (n = 9 from 3 independent experiments, except for EV where n = 12 from 4 independent experiments). B Infectivity of THP-1 cells expressing EV or LDLR (n = 9 from 3 independent experiments). C Binding of chimeric WT and mutant viruses to K562-EV or K562-VLDLR cells (n = 12 from 4 independent experiments for WT, n = 9 from 3 independent experiments for 84–119 and 156–157, and n = 6 from 2 independent experiments for 71–77). Viruses were incubated with cells on ice for 75 min, then cells were washed, and the bound virus was quantified by qPCR, then expressed as a ratio to GAPDH and normalized to EV. D Apparent binding affinity showing the nonlinear fit of WT or mutant SINV/EEEV for VLDLR LA(1-2)-Fc, with KD apparent and Hill slope (h) (n = 6 from 2 independent experiments). E Neutralization of chimeric eGFP reporter WT and mutant viruses by VLDLR LA(1-2)-Fc (100–0.1 µg/mL in tenfold dilutions and 0 µg control) in Vero cells (n = 4 from 2 independent experiments, except for controls where n = 8 for WT and n = 12 for mutants from 2 independent experiments). F Neutralization of chimeric eGFP reporter WT and mutant viruses by VLDLR LA(1-2)-Fc in CHO-K1 and CHO-pgsA-756 cells (n = 4 from 2 independent experiments, except for WT where n = 9 from 3 independent experiments). G Neutralization of chimeric eGFP reporter WT by heparin or BSA control (2000–2 µg/mL in tenfold dilutions and control) in cells overexpressing protein receptors (n = 6 from 2 independent experiments, except for N2a ΔB4galt7-LDLR cells where controls are n = 6 and samples are n = 4 from 2 independent experiments) quantified by flow cytometry. Means ± SD are shown. Significance was determined by two-way ANOVA with Tukey’s post-hoc tests. Exact p-values are indicated. Source data provided as a Source Data file.

To further assess whether or not the E2 mutations disrupted EEEV-VLDLR interactions, we performed direct binding and neutralization assays using chimeric WT and E2 mutants and VLDLR LA(1-2)-Fc, a two-domain truncated “receptor decoy” version of the VLDLR receptor fused to the constant region of human IgG111. Despite no significant difference in binding to or infectivity for K562-VLDLR versus K562-EV cells, when binding to the VLDLR LA(1-2)-Fc decoy was assessed, the 71–77 mutant showed significantly higher apparent affinity of binding compared to the WT (p = 0.0299 by Kruskal–Wallis test) (Fig. 2D). The 84–119 mutant exhibited similar binding affinity to the WT (Fig. 2D), and the 156–157 mutant showed minimal binding to VLDLR LA(1-2)-Fc (Supplementary Fig. 1C). To measure neutralization, Vero cells were infected with the WT SINV/EEEV and mutant viruses following incubation with increasing concentrations of VLDLR LA(1-2)-Fc (Fig. 2E) or LDLRAD3 LA1-Fc19 used as a negative control (Supplementary Fig. 1D). The WT virus was significantly neutralized at the lowest dilution of VLDLR LA(1-2)-Fc tested, with increased inhibition at higher concentrations (Fig. 2E). The 84–119 mutant exhibited a neutralization phenotype like the WT but was significantly more resistant to neutralization at the 1 µg/mL concentration. Despite more avid binding for the VLDLR LA(1-2)-Fc, the 71–77 mutant was significantly less susceptible to neutralization than the WT at all concentrations and was significantly inhibited only at 100 µg/mL of VLDLR LA(1-2)-Fc (Fig. 2E). In comparison, the 156–157 mutant was completely resistant to neutralization by VLDLR LA(1-2)-Fc (Fig. 2E). Therefore, despite VLDLR LA(1-2)-Fc being able to bind to the 71–77 and 84–119 viruses, mutations at each of the three sites interferes with EEEV-VLDLR LA(1-2)-Fc neutralization, with the 156–157 mutation having the greatest impact.

Given the overlap in the E2 mutations affecting HS and protein receptor interactions, we hypothesized that inhibition of EEEV infection by VLDLR LA(1-2)-Fc could, in part, reflect interference with EEEV-HS interactions. To test this, we determined VLDLR LA(1-2)-Fc inhibition of WT SINV/EEEV and the 71–77 and 84–119 mutants on CHO-K1 and CHO-pgsA-745 cells. Since the 156–157 mutant was not neutralized by the decoy, it was omitted from these experiments. The CHO-K1 and CHO-pgsA-745 cells both expressed similar, low levels of VLDLR but did not express ApoER2 or LDLR (Supplementary Fig. 3A–D); thus, differences in neutralization on these cells would primarily be driven by cellular HS expression. While the WT was significantly neutralized by all concentrations of the VLDLR decoy on both cell types, at higher concentrations of inhibitor, neutralization was significantly reduced on CHO-pgsA-745 cells versus the CHO-K1 cells (Fig. 2F). The 84–119 mutant was less inhibited by the VLDLR decoy on CHO-pgsA-745 cells versus the CHO-K1 cells at all concentrations, only showing significant inhibition by the decoy at higher concentrations on CHO-pgsA-745 cells (Fig. 2F). As with Vero cells (Fig. 2E), the 71–77 mutant was only significantly neutralized at the highest concentration of the decoy on CHO-K1 cells and showed no difference in neutralization on CHO-K1 versus CHO-pgsA-745 cells (Fig. 2F). Together, these data suggest that neutralization by VLDLR LA(1-2)-Fc is aided by blockade of cellular HS-binding.

To determine whether, inversely, heparin diminishes EEEV-protein receptor interactions, we tested heparin neutralization of the WT SINV/EEEV virus on cells that over-expressed EEEV-protein receptors. The WT virus was significantly inhibited by 20 or 200 µg/mL heparin, with increasing neutralization at higher concentrations on K562-VLDLR, K562-ApoER2, and THP-LDLR cells (Fig. 2G). To determine whether the high degree of neutralization was due to heparin blocking of EEEV-HS interactions and reducing overall cell attachment efficiency versus directly impacting protein receptor engagement, we also tested neutralization on N2a ΔB4galt7-LDLR cells10, which are deficient in HS and chondroitin sulfate but over-express LDLR. In these experiments, heparin only inhibited infection by the WT virus at the highest (2000 µg/mL) concentration (Fig. 2G). Finally, treatment of K562-EV and K562-VLDLR cells with high concentrations of heparinase II demonstrated that EEEV infectivity for K562-VLDLR cells was not primarily dependent upon attachment to HS (Supplementary Fig. 1E). Together, these data demonstrate that heparin contributes to HS-mediated neutralization of SINV/EEEV WT on GAG+ receptor overexpressing cells, and heparin can compete with protein receptor binding, albeit only at high concentrations.

Passage of HS/protein receptor binding site mutants in cultured cells selects for mutations that impact binding of both receptor types

As HS- and protein receptor-binding residues appeared to overlap functionally in that both receptors enhanced infectivity, we determined whether selection for partial or full restoration of HS- or protein receptor-binding would affect one or both interactions. Rapid acquisition of positively charged amino acid mutations that confer increased HS-binding after arbovirus passage in cultured cells has been extensively documented (reviewed in ref. 7). We sought to mimic this process by passaging the receptor-binding domain SINV/EEEV mutants on BHK-21 cells, which abundantly express HS but have low expression of VLDLR, ApoER2, or LDLR (thus, HShi protein receptorlow), and K562-VLDLR cells (Supplementary Fig. 3A–D). Notably, despite expression, HS is not a dominant factor for productive infection of K562-VLDLR cells as they are still highly infectable following heparinase II digestion (Supplementary Fig. 1E). Therefore, the K562-VLDLR cells functionally act as complementary HSlow VLDLRhigh cells.

Sequencing of supernatants from the K562-VLDLR cells at passage 5 and 10 revealed that all three viruses retained the original mutant residues (Table 1), and by passage 5 the 71–77 and 156–157 mutants each acquired an additional charge-reversing substitution mutation (71–77-E244K and 156–157-E206K), with acquired residues physiochemically similar to documented passage-associated changes in arboviruses found to increase HS-binding7. In addition, the 206K residue was reported to allow EEEV strain PE-6 to use an additional E2 domain B shelf VLDLR-binding site11,12,13. By passage 10, both replicates for the 84–119 mutant also acquired the E206K mutation. Sequencing of BHK-21 cell supernatants at passages 5 and 10 revealed the 71–77 and 156–157 mutants had each acquired a mutation to a positively charged amino acid while, again, retaining the original mutant residues (71–77-E147K and 156–157-Q8K) (Table 1). No mutations or reversions were observed for the 84–119 mutant (Table 1). Notably, all acquired mutations except for E244K are within three amino acid residues of MXRA8-, LDLRAD3-, and/or VLDLR-binding residues (Supplementary Fig. 2)11,12,13,29,30,33.

Table 1 Mutations acquired after serial passaging chimeric HS-binding mutants

To test whether these additional mutations partially or fully restored HS-binding and/or VLDLR receptor usage, we determined the infectivity and binding of the chimeric passaging mutants on CHO-K1 versus GAG- or HS-deficient cells, binding to heparin, and the increase of infectivity and binding for cells that overexpress the protein receptors versus control cells (Fig. 3A–F). Since the 156–157 mutant has more pronounced effects on these phenotypes compared to the 84–119 mutant (Figs. 1A, B and E and 2A–C), we evaluated the impact of the addition of the E206K mutation using the 156–157/206K mutant only. Both mutants that arose from passaging on BHK-21 cells (71–77/147 K and 156–157/8 K) and K562-VLDLR cells (71–77/224 K and 156–157/206 K) exhibited significantly reduced infectivity for both CHO-pgsA-745 and CHO-pgsD-677 cells compared to CHO-K1 (Fig. 3A and Supplementary Fig. 4A). Therefore, selection on both HShi protein receptorlow and HSlow VLDLRhigh cells yielded mutations that increased HS-dependent infectivity, presuming that protein receptor-mediated infectivity is limited on CHO cells and equal between mutant and WT cells. Both 71–77/147 K and 156–157/8 K mutants showed increased binding to CHO-K1 cells compared to their parent mutation (Fig. 3B). However, the diminution of infectivity observed for the mutants on GAG/HS-deficient cells (Fig. 3A) was only recapitulated for cell binding with the 71–77/147 K mutant, which exhibited a significant decrease in binding to CHO-psgA-745 cells compared to CHO-K1 (Fig. 3B). When direct binding to heparin was assessed, the 71–77/147 K, 156–157/8 K, and 156–157/206 K mutants exhibited apparent binding affinities close to that of WT, while the 71–77/244 K mutant exhibited a minor, twofold, increase compared to the 71–77 parent (Figs. 1E and 3C). Thus, it is possible that the cell binding assay may not be sufficiently sensitive to detect subtle differences in binding that affect infectivity, or these mutations may allow for binding to a different structure on cells that does not promote productive infection.

Fig. 3: Passage of HS/protein receptor binding site mutants on cultured cells selects for mutations that impact binding of both receptors.
figure 3

A Infectivity of double SINV/EEEV mutants found in passaging experiment on CHO-K1, CHO-pgsA-745, or CHO-pgsD-677 cells. Relative infectivity compared to CHO-K1 was determined for all viruses (n = 9 from 3 independent experiments). B Binding of chimeric passaging mutant viruses to CHO-K1 and CHO-pgsA-745 cells(n = 6 from 2 independent experiments). (Data for single mutant parent viruses are the same as in Fig. 1B.) C Apparent binding affinity showing nonlinear fit of chimeric passaging mutants for heparin, with KD apparent and Hill slope (h) (n = 6 from 2 independent experiments). D, E Infectivity of passaging mutants in (D) K562 cells expressing EV, VLDLR, ApoER2 isoform 1, or ApoER2 isoform 2 (n = 9 from 3 independent experiments, except for the 71–77 and 156–157 mutants on K562-EV cells where n = 12 from 4 independent experiments) or E THP-1 EV or LDLR cells (n = 9 from 3 independent experiments) quantified by flow cytometry. (Data for single mutant parent viruses are the same as in Fig. 2A, B). F Binding of chimeric WT and mutant viruses to K562-EV or K562-VLDLR cells, performed as described in Fig. 2C (n = 6 from 2 independent experiments, except for 1561-57 where n = 9 from 3 independent experiments). Binding is expressed as a ratio to GAPDH and normalized to EV. (Data for single mutant parent viruses are the same as in Fig. 2C.) G Apparent binding affinity showing nonlinear fit of chimeric passaging mutants for VLDLR LA(1-2)-Fc, with KD apparent and h (n = 6 from 2 independent experiments). H Neutralization of chimeric passaging mutant viruses by VLDLR LA(1-2)-Fc (100–0.1 µg in tenfold dilutions and 0 µg control) in CHO-K1 and CHO-pgsA-745 cells (n = 4 from 2 independent experiments). Means ± SD are shown. Significance was determined by (A) one-way ANOVA with Bonferroni’s post-hoc tests and (B, D–F, and H) two-way ANOVA with Tukey’s post-hoc tests. Exact p-values are indicated. Source data provided as a Source Data file.

When protein receptor-dependent infection was assessed, of the mutants that arose during BHK-21 (HShi protein receptorlow) cell passage, the 71–77/147 K mutant exhibited increased infectivity versus the 71–77 parent only on THP-1 LDLR cells (Fig. 3D, E). Although exhibiting a small but significant infectivity increase for THP-1 LDLR cells versus THP-1 EV cells, the 156–157/8 K mutant showed no significant difference in infectivity versus the 156–157 parent for any of the over-expression cells (Fig. 3D, E). For the mutants that arose during K562-VLDLR (HSlow VLDLRhigh) cell passage, the 156–157/206 K mutant exhibited increased infectivity for the K562-VLDLR and THP-1 LDLR cells compared to the 156–157 parent, whereas the 71–77/244 K mutant exhibited increased infectivity compared to its parent on K562-VLDLR, K562-ApoER2 isoform 1, and THP-1 LDLR cells (Fig. 3D, E). These infectivity results were recapitulated when virus binding to K562-EV and K562-VLDLR was assessed. The 71–77/147 K and 156–157/8 K mutants exhibited similar binding to their parental viruses with K562-EV and K562-VLDLR cells, while the 71–77/244 K and 156–157/206 K mutants exhibited increased binding to K562-VLDLR cells, although the increase was not significant (p = 0.09) for the 156–157/206 K mutant (Fig. 3F). These data suggest that mutations selected in the context of protein receptor abundance consistently increase both HS- and protein receptor-dependent infection, whereas selection for mutations in the context of HS abundance leads to increased HS-binding and HS-dependent infection, and can, but does not always, lead to increased protein receptor utilization.

To determine whether increased HS-dependent infectivity/protein receptor interaction would also lead to increased protein receptor decoy binding and neutralization, we tested the ability of VLDLR LA(1-2)-Fc to bind to and neutralize the passaging mutants on CHO-K1 and CHO-pgsA-745 cells. Both of the 71–77 passaging mutants showed high binding strength for VLDLR LA(1-2)-Fc (Fig. 3G) and comparable, limited neutralization as observed for the 71–77 parent on CHO-pgsA-745 cells (Fig. 3H). However, unlike the parental 71–77 virus, both mutants were significantly neutralized by higher concentrations of VLDLR LA(1-2)-Fc when GAGs were present (Figs. 2F and 3H). Similarly, both 156–157 passaging mutants showed high binding strength to VLDLR LA(1-2)-Fc (Fig. 3G), and no neutralization on GAG-deficient cells, but were significantly neutralized at the 100 µg/mL concentration on CHO-K1 cells (Fig. 3H). These results suggest that in the context of low-to-negative protein receptor expression, the decoy can block infectivity through blocking of HS interactions. However, whether or not the low expression of VLDLR receptors on CHO-K1 or CHO-pgsA-745 cells has a role in infection (Supplementary Fig. 3B) is not clear in these experiments. Together, the mutant data indicate that single-site, positively charged amino acid mutations selected during passage can simultaneously increase interactions with HS and protein receptors.

Mutation of HS/protein receptor binding sites attenuates EEEV disease, alters virus tropism in mice, and diminishes protection conferred by receptor decoy inhibitors

To determine whether the HS-binding site mutations lead to attenuation following the natural route of vertebrate infection, mice were inoculated with WT or mutant EEEV strain FL93-939 viruses with equal genome equivalents to 103 WT PFU subcutaneously (sc.) in the rear footpad (fp.). As with the WT virus, infection with the 84–119 mutant led to rapid mortality and limited clinical signs (Fig. 4A, B and Supplementary Fig 5A). As previously reported3,4, the 71–77 mutant led to prolonged presentation of clinical signs and a significantly extended average survival time (AST) compared to the WT. The 156–157 mutant was the most attenuated, resulting in a significantly longer AST compared to the WT and the other mutants (p < 0.0001), but it was still uniformly lethal (Fig. 4A).

Fig. 4: Mutation of HS/protein receptor binding sites attenuates EEEV disease.
figure 4

A, B CD-1 mice were infected with equal genomes of WT and mutant EEEV strains equivalent to 103 WT PFU in fp. (n = 15 from 3 independent experiments). A Survival. B Weight change. C, D CD-1 mice were inoculated ic. with equal genomes of WT and mutant SINV/EEEV virus equivalent to 104 WT PFU (n = 15 from 3 independent experiments: SINV/EEEV WT and SINV/EEEV 84–119, and n = 10 from 2 independent experiments: SINV/EEEV 71–77 and SINV/EEEV 156–157). C Survival. D Weight change. E, F CD-1 mice were inoculated in the fp. with equal genome copies of WT and mutant EEEV strains equivalent to 103 WT PFU (n = 10 from 2 independent experiments). At 2 dpi serum was collected and analyzed. E IFNα levels in serum. F Viral genomes present in the serum measured by qRT-PCR. G CD-1 mice were inoculated in both rear fp with equal genome copies of WT and mutant EEEV or VEEV nLuc expressing virus equivalent to 103 PFU of WT. At 8 hpi, PLNs were collected and replication was quantified (n = 12 for VEEV WT and n = 14 for mock from 2 independent experiments, and n = 26 from 3 independent experiments for EEEV WT, 71–77, 84–119, and 156–157). H–J CD-1 mice (n = 10 from 2 independent experiments) were inoculated in the fp. with equal genome copies of nLuc expressing virus equivalent to 103 PFU of WT. Virus dissemination was monitored daily using an in vivo imaging system (IVIS), and total flux (photons/second) was quantified. H Foot, where animals were infected. I PLNs; area roughly behind the knee on ventral view. J Head; dorsal view. Data shows (B, D) means ± SEM, (E–G) means ± SD, and (H–J) means ± 95% confidence interval (CI). Significance determined by (A, C) log-rank test, (E, G) Brown-Forsythe and Welch one-way ANOVA tests with Dunnett’s post-hoc tests, (F) one-way ANOVA with Bonferroni’s post-hoc test, and (H–J) two-way repeated measures ANOVA with Dunnett’s post-hoc tests. Exact p-values are indicated. Source data provided as a Source Data file.

To provide a measurement of neurovirulence in the absence of replication outside the CNS, we inoculated mice intracranially (ic.) with equal genomes of SINV/EEEV chimeric viruses equivalent to 104 PFU of the WT. The chimeras are attenuated in mice versus parental EEEV viruses, as their intracellular replication is controlled by SINV, which is relatively benign after ic. infection of adult mice25,34. These viruses allow detection of subtle differences in virulence conferred by the sequence of EEEV structural proteins. The WT chimera exhibits partial mortality in this model system (60%) associated with clinical signs of disease (Fig. 4C, D and Supplementary Fig. 5B). Mice inoculated with the 71–77 or 156–157 mutants did not develop lethal encephalitis and showed no or limited clinical signs of disease (Fig. 4C, D and Supplementary Fig. 5B). However, mice inoculated with the 84–119 mutant all developed clinical signs of disease and had similar weight loss as those infected with the WT (Fig. 4D and Supplementary Fig. 5B), but the mutant resulted in significantly reduced mortality (20%). Mice infected with the 71–77 or 156–157 mutants did not develop lethal encephalitis and showed no or limited clinical signs of disease (Fig. 4C, D and Supplementary Fig. 5B).

Since mice infected sc. with the 71–77, 84–119, and 156–157 mutants exhibited attenuated neurovirulence and increased duration of clinical signs versus the EEEV WT, and HS-binding efficiency has been shown to affect type I IFN induction, viremia, and draining lymph node (DLN) cell infection3,5, we investigated the impact of these mutations on each of these phenotypes following sc. infection. Consistent with previous results3,4, WT EEEV did not lead to robust induction of IFNα and exhibited the lowest level of viremia at 2 days post-infection (dpi), whereas infection with the 71–77 mutant led to significantly higher levels of IFNα and viremia (Fig. 4E, F). Despite the similarity of disease caused by the 84–119 mutant to the WT virus (Fig. 4A, B and Supplementary Fig. 5A), this mutant also elicited significantly higher levels of IFNα and viremia than the WT at 2 dpi (Fig. 4E, F). Although attenuated, infection with the 156–157 mutant resulted in significantly increased viremia versus the WT and similar IFNα levels (Fig. 4E, F).

We utilized nanoluciferase (nLuc) expressing reporter viruses to monitor replication in the site of infection (foot), draining popliteal lymph nodes (PLNs), and heads of mice following fp. inoculation with genome equivalents to 103 EEEV WT PFU. PLNs were collected from mice at 8 hpi and viral replication was quantified by in vitro luciferase assay (Fig. 4G). At this early timepoint, the 71–77 mutant exhibited significantly increased replication in the PLN (Fig. 4G). The reduction of HS/protein receptor binding by the 84–119 and 156–157 mutations did not lead to a significant increase in virus signal present in the PLN compared to WT (Fig. 4G). As expected, VEEV infection, used as a positive control, yielded a signal in the PLNs that was much higher than any EEEV virus, reflecting limited HS binding by WT VEEV and miR142-3p microRNA restriction of myeloid cell replication with EEEV5. To confirm ex vivo PLN viral load data, we quantified signal in the area of the PLN in mice using an in vivo imaging system (IVIS). Infection with the 71–77 and 84–119 mutants led to significantly higher PLN signal levels than WT virus at 2 dpi (Fig. 4H). Together, these data suggest that the 71–77 mutation increases early PLN infection, possibly due to having the most pronounced impact on HS-binding. Then, by 1 dpi, WT and the 71–77 and 84–119 mutants show similar replication in the PLNs. However, by 2 dpi, the 71–77 and 84–119 mutations increase replication compared to WT, possibly due to altered HS-binding. Whereas the 156–157 mutation does not show any PLN infection enhancement compared to the WT and exhibits a non-significant reduction at 8 hpi and 1 dpi, likely due to having the most profound impact on interactions with multiple receptors. IVIS imaging of the foot showed no significant differences between WT and the E2 mutants on 1 or 2 dpi; however, replication of the 156–157 was non-significantly decreased versus the WT on 1 dpi (Fig. 4I). Similarly, there were no significant differences in brain replication between the viruses on 3–5 dpi (Fig. 4J). However, as with other tissues, the 156–157 mutant trended toward the lowest replication, which is consistent with the prolonged survival time of mice infected with this mutant (Fig. 4A and J). Thus, reduction in HS and protein receptor binding is associated with higher IFN induction and viremia at 2 dpi, and with the 71–77 and possibly 84–119 mutants, greater DLN infection. The disruption of the known receptor contact sites with the 156–157 mutant may affect cell interactions more profoundly, leading the reduction in HS binding to not show a similar effect to that observed with the other mutants.

Pre-treatment with VLDLR LA(1-2)-Fc protects mice from lethal sc. infection of WT EEEV11. Due to the significant decreases in infection on cells over-expressing protein receptors and the significant differences noted in the ability of VLDLR LA(1-2)-Fc to bind and neutralize the mutant viruses, we tested the ability of VLDLR LA(1-2)-Fc to protect mice from infection with equal genome equivalents to 103 PFU of WT EEEV. Mice were treated intraperitoneally with 100 µg of either the VLDLR LA(1-2)-Fc or the LDLRAD3 LA1-Fc control six hours before infection. All mice treated with the LDLRAD3 LA1-Fc control before infection succumbed to infection (Fig. 5). As with previous studies11, mice pretreated with VLDLR LA(1-2)-Fc were almost completely protected from WT infection (Fig. 5A and Supplementary Fig. 6). Similar protection was observed for the 84–119 mutant (Fig. 5C). Despite the significant reduction versus WT in VLDLR-mediated infectivity and neutralization in vitro (Fig. 2A, E, F), mice treated with VLDLR LA(1-2)-Fc were also largely protected from the 71–77 mutant (Fig. 5B), which is consistent with the decoy binding to the 71–77 mutant virus in vitro (Fig. 2D). In contrast, mice infected with the 156–157 mutant were not protected by the VLDLR decoy (Fig. 5D), which is consistent with in vitro data showing an absence of binding and neutralization (Fig. 2E and Supplementary Fig. 1C).

Fig. 5: Mutation of HS/protein receptor binding sites diminishes protection conferred by a receptor decoy inhibitor.
figure 5

Survival data for mice treated intraperitoneally with 100 µg of VLDLR LA(1-2)-Fc (dashed line) or LDLRAD3 LA1-Fc (solid line) as a negative control. Six hours after treatment, mice were infected fp. with equal genome copies of WT and mutant EEEV viruses equivalent to 103 WT PFU (n = 6 mice from 2 independent experiments for LDLRAD3 LA1-Fc and n = 8 mice from 2 independent experiments for VLDLR LA(1-2)-Fc). A EEEV WT (FL93-939). B EEEV 71–77. C EEEV 84–119. D EEEV 156–157. Significance of survival for mice treated with control versus VLDLR LA(1-2)-Fc was determined by log-rank test. Exact p-values are indicated. Source data provided as a Source Data file.

Mutation of HS/protein receptor binding sites diminishes transmission to and dissemination within mosquito vectors

Finally, we examined the effects of HS-binding site mutations on the ability of EEEV to infect mosquito cells and to infect and disseminate within a mosquito vector. Cell lines from EEEV the bridge vector species35,36,37,38 Aedes albopictus (C6/36) and the potential bridge vector Ae. aegypti (Aag2) were inoculated with equal genome equivalents of WT SINV/EEEV and E2 mutants. Infectivity of the 84–119 mutant was similar to WT for both cell lines (Supplementary Fig. 7). Unexpectedly, the 71–77 mutant exhibited higher infectivity than WT with both cell lines, and infectivity of the 156–157 mutant was higher than the WT for the C6/36 cells (Supplementary Fig. 7). These data suggest that significantly reducing HS and protein receptor-binding as measured with mammalian cells, can increase mosquito cell infectivity in vitro. To determine whether the E2 mutations would increase productive infection of or dissemination within mosquitoes, measured by the detection of the virus in the body, or detection of the virus in the heads, respectively, we exposed Ae. albopictus mosquitoes with equal genome equivalents of WT or HS-binding site mutants via artificial blood meals (Table 2). All mosquitoes infected with the WT had detectable virus in their bodies and heads, indicating dissemination from the midgut. Infection with the 84–119 mutant showed a small decrease in infection and a modest yet significant decrease in dissemination, with 90% infection of bodies and 80% infection of heads. In contrast to the in vitro mosquito cell infectivity data, the 71–77 mutant exhibited a significant decrease versus the WT in both infection and dissemination, with only 20% of mosquitoes having virus in the body, and 10% in their heads (Table 2). Notably, none of the mosquitoes exposed to the 156–157 mutant had virus in their bodies or heads. Therefore, the 156–157 mutant, while inhibited in HS and protein receptor binding with mammalian cells, maintains infectivity for mosquito cells in vitro, yet is highly restricted for infection and dissemination in vivo. This suggests that the WT E2 K156-R157 residues are critical for establishing infection in the midgut epithelium of mosquitoes.

Table 2 Mutation of HS/protein receptor binding sites diminishes infection of and dissemination within mosquito vectors

Discussion

To date, EEEV remains the only alphavirus for which unpassaged strains have been shown to bind to HS efficiently3. Previous work mapped positively charged residues at positions 71, 74, and 77 of the E2 glycoprotein of EEEV, as critical for HS-binding and contribute to neurovirulence and reduction of draining lymph node infection and viral prodromal disease3,4. Cryo-EM studies with 6 kDa heparin bound to the virus further identified heparin/HS-binding sites coordinated by positively charged residues at E2 R84 and R119 and K156-R1578. Our current studies demonstrate that the newly identified E2 residues also contribute to EEEV-HS interactions that promote infection in vitro and affect pathogenesis in vivo (Table 3). Mutation of the 71–77 site, as previously reported3,4, resulted in a virus that was significantly reduced versus the WT in all assays of HS-dependence. Mutation of the 156–157 site altered infectivity dependence with increasing ionic strength, yielded variable dependence on cellular HS expression for infectivity measured with GAG/HS-deficient cells and heparinase digestion, and decreased binding to heparin. The 84–119 site appears to alter EEEV-HS interactions but was significantly different from WT only in ionic strength assays. Therefore, while differing in quantitative and qualitative aspects of the HS interaction, each site contributes to the HS-dependence of EEEV infection.

Table 3 Summary of effects of each mutation on EEEV HS or receptor interactions

Recently, the residues that mediate EEEV-VLDLR binding were mapped in cryo-EM binding studies. Two of the E2 residues that were identified as critical for engagement with the VLDLR LA1 domain were K156 and R15711,12,13. The K156 residue was also shown to be important for EEEV-ApoER2 isoform 2 interactions12. Consistent with these reports11,12,13, our data indicates that charged-to-alanine mutation at the 156–157 HS-binding site ablates EEEV-VLDLR interactions and ApoER2- and LDLR-dependent infectivity. This finding is notable as it shows that two mutations are sufficient to ablate EEEV interactions with multiple receptors, which would need to be considered for the development of infection-blocking reagents as therapeutics. Despite not being identified as VLDLR contact residues11,12,13, the 71–77 and 84–119 mutations also significantly impacted EEEV interactions with VLDLR, ApoER2, and LDLR, albeit less than the 156–157 mutation. It is possible that the 71–77 mutation impacts the ability of EEEV to engage with the LA2 domain of VLDLR, as the residues that participate in the domain A “shelf” engagement with LA2 are in proximity to the 71, 74, and 77 residues (Fig. 6)11,12,13. Although the 84–119 mutation also impacted interactions with all identified protein receptors, its effect was not as strong as the 71–77 or 156–157 mutations, implying a more limited and specific effect.

Fig. 6: Location of protein receptor- and HS-binding residues on the EEEV E2 trimer.
figure 6

A ribbon model of the structure of EEEV E2 proteins as they appear in the E1/E2 trimeric spike. Side chains are displayed for residues involved in HS-binding (blue), residues identified in structural analysis as being involved in binding to one VLDLR molecule11,12,13 (red), and residues that are involved in both HS-binding and are direct contacts for VLDLR-binding (purple). Side chains for residues mutated during in vitro passage are lime green (K562-VLDLR) or cyan (BHK-21). A Top view. B Side view. Figures were made using UCSF ChimeraX53,54.

A summary of the relative effects of the HS binding site mutations (71–77, 84–119, and 156–157) on HS and protein receptor utilization (Table 3) suggests that EEEV HS- and protein receptor-binding overlap with each of the three E2 sites. The 84–119 site has a modest effect on both phenotypes, whereas the 71–77 and 156–157 sites have more profound, although phenotypically different, effects. The 156–157 site directly contacts VLDLR11,12,13, and the 156–157 mutant appears to have little to no interaction with any identified protein receptor. Further evidence for the overlap of HS- and protein receptor-binding by EEEV is provided in the phenotypes of serial passaging mutants. When EEEV was passaged on HShigh protein receptorlow cells, acquired mutations (E2 8 K and 147 K) increased EEEV-HS interactions, with the 147 K mutation also increasing LDLR-dependent infectivity. However, when passaged on cells that have a HSlow VLDLRhigh phenotype, the adaptive E2 206 K and 244 K mutations (located in domain B, which engages VLDLR)11,12,13 increased both protein receptor and HS interactions. Furthermore, our data support the observation that lysine at E2 206 results in an additional site of VLDLR engagement (as found with the PE-6 strain of EEEV)11,12,13 when the lysine at E2 156 is mutated, and may suggest that this is a cell-adaptive change.

In further support of a link between EEEV engagement with HS and protein receptors, neutralization of chimeric WT virus and passaging mutants by the VLDLR LA(1-2)-Fc receptor decoy was greater on cells that expressed HS than on GAG-deficient cells. This suggests that the strong neutralizing phenotype of the VLDLR decoy is likely also due to its ability to inhibit EEEV-HS interactions. However, binding to the VLDLR LA(1-2)-Fc alone protected mice from lethal EEEV disease11, regardless of in vitro neutralization capacity, as seen with the 71–77 mutant, suggesting the therapeutic potential of VLDLR LA(1-2)-Fc may not solely be due to blocking VLDLR (and potentially other receptor) engagement but may also be through Fc-mediated effector functions.

In reciprocal experiments using heparin to neutralize EEEV infectivity, we observed competition for infection of VLDLR, ApoER2, or LDLR over-expressing cells, even when cells were genetically altered to lack HS. Since cryo-EM studies show that related and unrelated viruses have converged to utilize the same structural approach to engage with members of the LDL-receptor family11,12, it is possible that protein receptor engagement and HS-binding sites may overlap structurally and chemically. This may explain the seemingly unlikely phenomenon of single positively charged amino acid mutations conferring efficient HS binding to cell-passaged alphaviruses7, with these mutations often appearing adjacent to or to directly overlapping identified protein receptor-binding residues (Supplementary Fig. 2). Indeed, there may be a balance during virus transmission such that satisfactory binding with protein receptors is achieved with minimal HS interactions. However, upon passage in vitro, where protein receptor abundance may be variable, single site mutations can rapidly confer increased cell attachment, thus rescuing infectivity. In vivo, selective pressures likely drive reductions in HS-binding due to enhanced clearance associated with the phenotype18,39,40. However, effects on protein receptor engagement are less predictable. It is notable that the 71–77 region is a hotspot for variation among EEEV strains, and such variants have altered engagement of HS receptors versus EEEV FL93-9394.

Despite significantly decreasing EEEV-HS interactions and ablating EEEV interactions with all four identified protein receptors, the 156–157 mutant was still uniformly fatal in mice. This result implies that loss of HS- and VLDLR/ApoER2/LDLR-binding via single-site mutation is not highly attenuating to EEEV in mammals, and it predicts the existence of additional attachment/entry receptors that have yet to be identified. Similarly, mosquito cells appear to be readily infected by this 156–157 mutant virus in vitro, yet WT residues at these positions appear to be required to establish a mosquito midgut infection after a blood meal. It is, therefore, likely that the expression patterns of as-yet-unidentified receptors are tissue dependent.

The abrogation of HS-binding diminished the neurovirulence of each mutant and enhanced neurovirulence has been attributed to the HS-binding phenotype3,4,6,25,41. Our observations that the mutations to EEEV also impact the utilization of protein receptors bring into question the role of each type of receptor in each of these phenomena. The decreased murine neurovirulence associated with diminution of HS- and VLDLR interactions will require additional investigation to parse the different receptor engagement effects. The role of HS versus protein receptor binding upon suppression of lymphoid tissue infection and viremia, previously attributed to HS binding alone3,7, will also need to be determined.

Since EEEV has evolved to circulate in nature with efficient HS-binding, it is unclear how applicable the overlap between HS- and protein receptor-binding may be to weakly HS-binding viruses or viruses that engage receptors outside of the LDL-receptor family. Testing of the interactions of WT and cell-culture adapted, HS-binding viruses (e.g., derived from VEEV18 or CHIKV22) with their cognate receptors LDLRAD319,30,31 and MXRA829,32,33 may be required to clarify this issue. Since LDL binding proteins have been identified as receptors for a number of arboviruses42,43,44,45 and in vitro acquisition of HS-binding appears to be a ubiquitous phenotype in this group7, the interplay between HS- and protein receptor-binding may be an important feature or tradeoff of the arbovirus-cell interaction.

Methods

Ethics statement for mouse studies

All mouse studies were done in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Research Council. All procedures were performed at the University of Pittsburgh under protocols (20047334, 21048822, and 23043041) approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh. Female 4-week-old CD-1 mice were purchased commercially (Charles River). Only female mice were used as they have been the primary sex used in alphavirus studies historically and in our laboratory3,4,11,18, provide a known behavioral group and, to our knowledge, no significant sex-linked aspects of alphavirus disease or protective responses have been demonstrated. Mice were housed in a facility that maintained at a temperature range of 20–26.1 °C and a 30–70% relative humidity range on a 12:12 light:dark cycle with water and food provided ad libitum. Virus inoculations were performed when mice were 4–6 weeks-old under anesthesia that was induced and maintained with isoflurane, and then mice were monitored with approved euthanasia criteria based on weight loss and morbidity, and all efforts were made to minimize animal suffering.

Cells and viruses

Baby hamster kidney (BHK-21 [ATCC CCL-10]) cells were maintained in RPMI-1640 and supplemented with 10% heat-inactivated donor bovine serum (DBS; Gibco) and 10% tryptose phosphate broth (Moltox). Chinese hamster ovary (CHO)-K1 [ATCC CCL-61], GAG-deficient mutant CHO-pgsA-745 [ATCC CRL-2242], and HS-deficient mutant CHO-pgsD-677 [ATCC CRL-2244] cells were maintained in Ham’s F-12 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS, Gibco). Human lymphoblast K562 cells expressing VLDLR, ApoER2 isoform1, ApoER2 isoform2, or empty vector (EV) controls9 were kindly provided by Jonathan Abraham (Harvard Medical School, Boston) and maintained in RPMI-1640 with 10% heat-inactivated FBS, 20 mM HEPES (Corning), and 2 µg/mL puromycin (Mirus). African green monkey Vero [ATCC CCL-81] and N2a ΔB4galt7ΔLDLR-TC LDLR (N2a ΔB4galt7-LDLR) cells10 were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated FBS (Omega Scientific), the N2as were also supplemented with 10 mM HEPES. THP-1 cells expressing LDLR or EV controls10 were maintained in RPMI-1640 with 10% heat-inactivated FBS (Omega Scientific), 10 mM HEPES, and 200 µg/mL hygromycin (Fisher). C6/36 cells [ATCC CRL-1660] were maintained in alpha MEM containing 10% heat-inactivated FBS and no antibiotics. Aag2 cells were a kind gift from Tonya Colpitts (University of South Carolina) and maintained in DMEM containing 10% heat-inactivated FBS, 1% TPB, and no L-glutamine. All media contained 2 mM L-glutamine, 100 U/mL penicillin, and 0.5 mg/mL streptomycin unless indicated otherwise.

The following viruses were used: SINV/EEEV (FL93-939), SINV/EEEV E2 71–77, SINV/EEEV E2 84–119, SINV/EEEV E2 156–157, SINV/EEEV E2 71–77/147 K, SINV/EEEV E2 71–77/244 K, SINV/EEEV E2 156–157/8 K, SINV/EEEV E2 156/157/206 K, EEEV (FL93-939), EEEV E2 71–77, EEEV E2 84–119, EEEV E2 156–157, and VEEV (TrD). EEEV strain FL93-939 was generated from a cDNA clone46 and wild-type VEEV strain Trinidad Donkey was a gift from Robert Johnston (University of North Carolina-Chapel Hill). E2 mutants EEEV 71–773, EEEV 84–1198, and EEEV 156–1578 were constructed as previously described using the QuikChange II XL mutagenesis kit (Agilent Technologies). Fragment-swapping strategies were also used to generate combinations of the E2 mutations by the corresponding restriction sites (Mlu I, EcoR I, and Not I); constructs were confirmed by Sanger sequencing carried out by Genewiz. Chimeric mutant viruses were generated by replacing the structural proteins of SINV strain TR339 with the indicated virus as previously described20,47. eGFP and nLuc TaV reporter viruses were constructed as described in Sun et al.47. For assays requiring purified viruses, viruses were passaged once in BHK-21 cells. The next day supernatant was collected, clarified, loaded onto a 20%/60% discontinuous sucrose in TNE (0.05 M Tris-HCL [pH 7.2], 0.1 M NaCl, 0.001 M EDTA) gradient, and centrifuged for 3 h at 106,803.1 × g. The virus was harvested from the interface and then pelleted by centrifugation over a 20% sucrose in TNE cushion at 106,803.1 × g for 16–20 h. The pelleted virus was then resuspended in opti-MEM aliquoted and stored at −80 °C until use. The protein concentration of purified viruses was determined by Pierce BCA protein assay (Thermo), per the manufacturer’s protocol.

Susceptibility to absence of GAGs and high ionic conditions

Viruses were serially diluted in virus diluent (VD; PBS containing cations supplemented with 1% DBS, 100 U/mL penicillin, and 0.5 mg/mL streptomycin) and titrated in triplicate on control CHO-K1, CHO-pgsA-745, and CHO-pgsD-677 cells for 1 h at 37 °C. After infection, cells were overlayed with immunodiffusion-grade agarose (MP Biomedical) and incubated at 37 °C 5% CO2 for 18–20 hpi, and then GFP-expressing plaques were counted by fluorescence microscopy. Titers were determined on all cell types, and percent infectivity compared to the CHO-K1 controls was determined for each virus. To test whether mutant viruses are susceptible to increasing ionic conditions, viruses were serially diluted in RPMI-1640 with increasing concentrations of NaCl. Diluted viruses were then used to infect BHK-21 cells for 1 h at 37 °C in triplicate before being overlayed with immunodiffusion-grade agarose and incubated at 37 °C 5% CO2 for two days. GFP-expressing plaques were counted by fluorescence microscopy and percent infectivity compared to RPMI-1640 only control, which has a salt concentration of approximately 102 mM, for each virus.

Heparinase sensitivity assays

Heparinase treatments for BHK-21 cells were done as previously described4,22. Cells were washed once with PBS containing cations, then treated with 0, 0.5, or 1 U/mL heparinase I, II, or III (New England BioLabs) in PBS for 1.5 h at 37 °C. Cells were washed twice with PBS, infected with chimeric eGFP-expressing virus in triplicate for 1 h at 37 °C, and then overlayed with immunodiffusion-grade agarose. At 18–20 hpi, eGFP-expressing plaques were visualized by fluorescence microscopy. Titers for each treatment were determined, and percent infectivity compared to no treatment was determined for each condition.

For K562-EV and K562-VLDLR, 105 cells were washed with PBS and then treated with 0, 1, 2, or 4 U/mL heparinase II in PBS for 1 h at 37 °C. Cells were washed twice with PBS, infected with chimeric eGFP-expressing WT virus in triplicate for 1 h at 37 °C, and then overlayed with media. At 20 hpi, cells were washed and fixed in 4% PFA and analyzed using BD LSRFortessa and FlowJo 10.8 software (Supplementary Fig. 8A). Percent infectivity was determined for every condition on both cell types.

Indirect heparin binding assay

Viruses were serially diluted to approximately 106 PFU/mL, then 50 µL of virus was added in triplicate to collagen- or heparin-agarose (Sigma), which had been washed with RPMI-1640 and incubated on ice for 30 min. Beads were pelleted, and the unbound virus in the supernatant was titrated in duplicate on BHK-21. Virus was allowed to infect the cells for 1 h at 37 °C, and then cells were overlayed with immunodiffusion-grade agarose. At 18–20 hpi, eGFP-expressing plaques were visualized by fluorescence microscopy. Titers for each virus in each condition were recorded, and the percent infectivity of the virus incubated with heparin beads compared to collagen beads was determined.

Cellular binding assays

For CHO-K1 and CHO-pgsA-745, cells were seeded into a 96-well plate the previous day, and for K562-EV and K562-VLDLR cells 105 cells were transferred to a V-bottom plate. Cells were chilled on ice and inoculated with 5 × 109 genomes of the WT and mutant chimeric eGFP reporter viruses for 75 min on ice. Cell monolayers were then washed with chilled PBS before lysis by Tri-Reagent. RNA was isolated using 1-bromo-3-chloropropane (BCP) and isopropanol, and cDNA were generated using an M-MLV kit (Invitrogen). Binding to CHO cells was determined by digital droplet PCR (ddPCR) using the 2× ddPCR Supermix for Probes (No dUTP) kit (Bio-Rad) with primers targeting SINV nsp2 and Mmadhc48 (Supplementary Table 1), and analyzed using QX200 ddPCR system running Standard Edition software 2.0.0. Fold-change was directly quantified, and then percent binding compared to CHO-K1 was determined for each virus. Binding to K562 cells was determined by qPCR using TaqMan Fast Universal PCR Master Mix (2×), no AmpErase (Applied Biosystems), using primers targeting SINV nsp2 and GAPDH (Supplementary Table 1) with reactions run on a QuantStudio Flex-6 or QuantStudio 3 using Real-Time PCR software v1.7.2.

Enzyme-linked immunosorbent assays (ELISAs)

To determine KD apparent values and Hill slope, we performed a modified method of ELISAs as described in Syedbasha et al.49. Briefly, Maxisorp plates (Thermo) were plated with either 10 mg/mL of heparin sodium salt (Sigma) or 1 µg/mL of VLDLR LA(1-2)-Fc in PBS and incubated at 4 °C overnight. The plates were then washed three times with TBS-T (25 mM Tris, 0.15 M NaCl, 0.05% Tween-20 at pH 7.5; Thermo), and blocked with 5% bovine serum albumin (BSA, Thermo) in PBS for 2 h at room temperature (RT). Plates were washed once, then purified viruses that were twofold serially diluted in PBS containing 2% BSA (BSA/PBS) were added in triplicate along with a control containing no virus, and incubated for 2 h at RT. Plates were then washed three times and incubated with 1:11,000 of αEEEV ascites (ATCC) diluted in BSA/PBS for 1 h, then washed again and incubated with a 1:5000 dilution of goat anti-mouse HRP (Thermo) in BSA/PBS for 45 min at RT. Plates were then washed a final time and incubated with TMB substrate (Thermo) at RT for 10 min, and then the reaction was stopped with 2 M H2SO4. Plates were then read at 450 nm (Supplementary Figs. 1C and F and 4B, C) and analyzed by subtracting the average of the no virus control to all the test samples, then determining % binding for each virus concentration, and analyzed as described in Syedbasha et al.49.

Genome to PFU determination

WT and mutant viruses were titrated in triplicate by standard plaque assay on BHK-21 cells. To determine genome numbers 200 µL virus stock was added to 800 µL Trizol (Life Technologies). RNA was extracted with 1-bromo-3-chloropropane (BCP) and isopropanol. RNA was then reverse transcribed using the First-strand cDNA Synthesis M-MLV Reverse Transcriptase kit (Invitrogen) using a primer targeting EEEV nonstructural protein 2 (NSP2) coding region containing a T7 promoter tag (Supplementary Table 1). Genomic vRNA was quantified by qRT-PCR performed using the manufacturer’s instructions for the 2× Fast TaqMan Universal PCR Master Mix, No AmpErase UNG (Applied Biosystems), with the forward primer and probe EEEV NSP2 (Supplementary Table 1) and the reverse primer targeting the introduced T7 promoter (Supplementary Table 1), on a QuantStudio Flex-6 using Real-Time PCR software v1.7.2. Quantification of the amount of genomes present was determined using a standard curve from in vitro transcribed EEEV RNA.

Replication curves

WT and mutant viruses were diluted in VD to genome equivalents corresponding to WT EEEV MOI 1 and used to infect BHK-21 cells in triplicate at 37 °C for 1 h. Then, media was added to the infected cells. At 0, 6, 12, 24, and 48 hpi, one-tenth of the media was collected, aliquoted, and stored at −80 °C. The media on the cells was replenished with an equal volume of fresh media. Titers of collected samples were determined by plaque assay on BHK-21 cells.

Infection of K562 and THP-1 cells

K562 cells expressing EV, VLDLR, ApoER2 isoform 1, or ApoER2 isoform 2 were infected with genome equivalents of WT or mutant SINV/EEEV viruses containing eGFP TaV at MOI 2.4 for WT. THP-1 cells expressing EV or LDLR were infected with 1:2 dilutions of stock eGFP TaV expressing SINV/EEEV WT or mutant viruses. Resuspended cells were spread across three wells and incubated at 37 °C 5% CO2 for 1 h, then 100 µL media was added to wells. At 16–19 hpi, cells were washed and fixed in 4% PFA and analyzed using a BD LSRFortessa and FlowJo 10.8 software (Supplementary Fig. 8A). Percent infectivity was determined for each virus on each cell type.

Infection of C6/36 and Aag2 cells

C6/36 were plated in a 24-well plate. When cells reached confluence, they were counted and infected in triplicate with genome equivalents of WT or mutant SINV/EEEV viruses containing eGFP TaV at MOI 1 for WT. Aag2 cells were plated in a 48-well plate, and confluent cells were counted and then infected in triplicate with genome equivalents of WT or mutant SINV/EEEV viruses containing eGFP TaV at MOI 5 for WT. Cells were incubated at 28 °C 5% CO2 for 1 h, then 1 mL media was added to wells. At 16–19 hpi, cells were washed and fixed in 4% PFA and analyzed using a BD LSRFortessa and FlowJo 10.8 software (Supplementary Fig. 8A). Percent infectivity was determined for each virus on each cell type.

VLDLR decoy inhibition assays

Chimeric eGFP-expressing WT or HS mutant viruses were diluted in VD such that a target of 50 foci for each well would be counted and incubated with dilutions of VLDLR LA(1-2)-Fc or LDLRAD3 LA1-Fc as a control for 1 h at 37 °C. Complexes were used to infect duplicate wells of 24-well plates seeded with Vero, CHO-K1, or CHO-pgsA-745 cells for 1 h at 37 °C 5% CO2 before being overlayed with immunodiffusion-grade agarose. At 16–20 hpi, eGFP-expressing plaques were counted, and percent neutralization for each condition was determined for each virus.

Heparin inhibition assays

Chimeric eGFP-expressing WT or 71–77 mutant viruses were diluted 1:3 in VD and incubated with dilutions of Heparin or BSA as a control for. one hour at 37 °C. Complexes were used to infect duplicate wells of N2a ΔB4galt7ΔLDLR-TC LDLR cells or 106 cells of K562-VLDLR, K562-ApoER2, or THP-1 LDLR for one hour at 37 °C 5 5 5% CO2 before being overlayed with immunodiffusion-grade agarose. At 16–20 hpi, N2A cells were detached with TrypLE, and all cells were washed and fixed in 4% PFA and analyzed using a BD LSRFortessa and FlowJo 10.8 software (Supplementary Fig. 8A). Percent infectivity was determined for each condition, then relative infectivity to control for each condition was determined for each cell type.

Cell surface staining of receptors

For cellular surface staining, adherent cells were washed with PBS, detached with TrypLE (Thermo Fisher), and counted; suspension cells (K562 cells) were counted, and then all cells were transferred to a V-bottom 96-well plate. To determine protein receptor expression, the 106 cells were blocked in 100 μL of PBS containing 5% goat serum for 30 min at 4 °C. The samples were then incubated with primary antibodies (10 μg/mL in PBS containing 2% goat serum) for 45 min at RT. Primary antibodies used were anti-ApoER2 (Sigma, WH0007804M1-100UG, clone 3H2), anti-LDLR (R&D Systems, MAB2148-100, clone 472413), anti-VLDLR (GeneTex, GTX79552, clone 1H10), or isotype control (Thermo Fisher 14-4714-82, clone P3.6.2.8.1). Cells were washed three times in PBS containing 2% goat serum and then incubated with an Alexa Fluor 488 conjugated goat anti-mouse antibody (Thermo Fisher A-11001) diluted at 1:200 in 2% goat serum at RT for 45 min. Following this, cells were washed twice in 2% goat serum and resuspended in FACS buffer (PBS with 3% FBS and 1 mM EDTA) for further analysis. To determine cellular HS expression, 5 × 104 cells were washed in FACS buffer before incubating in blocking buffer, either PBS with a 1:20 dilution of Human TruStain FcX (Biolegend, Cat. No. 422301) for K562 cells or PBS with 5% goat serum for CHO, Vero, or BHK-21 cell lines. After this, cells were then incubated in a 1:200 dilution of anti-HS (AMSBIO 370255-1, clone F58-10E4) or isotype control (BD Biosciences 553472, clone G155-228) diluted in FACS buffer for 1 h at 4 °C. Cells were washed twice and incubated with a 1:500 dilution of Alexa Flour 488 conjugated goat anti-mouse IgM (Thermo Fisher A-21042) for 1 h at 4 °C. Then, cells were washed twice in FACS buffer for further analysis. Samples were analyzed using a BD LSRFortessa and FlowJo 10.8 software (Supplementary Fig. 8B).

In vitro virus passaging

Stocks of eGFP-expressing chimeric single mutant viruses were serially passaged ten times in duplicate on BHK-2, K562-VLDLR, and K562-EV cells. For BHK cells, stock or supernatant from previous passage collected every 24 hpi was diluted 1:100 for passages 1 and 2, then 1:1000 for passages 3–10. For K562 cells, stock or supernatant from the previous passage collected every 2 dpi was diluted 1:100 or undiluted for all passages. Increases in infectivity was measured by titration of collected supernatant by standard plaque assay on BHK-21 cells (Supplementary Fig. 3E). Viral RNA from supernatant collected at passages 5 and 10 from all replicates on both cell lines was added to Tri Reagent containing 5 µg of tRNA carrier, and then RNA was extracted per manufacturer protocol. Two-step RT-PCR was then performed using the First-strand cDNA Synthesis M-MLV Reverse Transcriptase kit (Invitrogen) and a primer targeting EEEV 6 K (Supplementary Table 1). The E2 gene was then amplified with PfuUltra-HF per the manufacturer’s protocols using primers targeting EEEV 6 K and EEEV E3 (Supplementary Table 1). Sequencing was then carried out by Genewiz and analyzed using SeqMan Ultra DNASTAR 17.

Mosquito infections

All mosquito procedures were performed at the University of Texas Medical Branch in ACL3 facilities. Aedes albopictus (Salvador, Brazil) mosquitoes were reared and maintained at 28 °C and 80% relative humidity with a 12:12 light:dark cycle with water and sucrose provided ad libitum.

Adult female Ae. albopictus mosquitoes were infected with equivalent genomes of EEEV WT, 71–77, 84–119, and 156–157 corresponding to 8.2 log10 WT PFU/mL in artificial bloodmeals. Engorged females were incubated at 27 °C for 10 days under 12-h light/12-h dark circadian lighting conditions. Subsequently, mosquitoes were euthanized, and bodies and heads were manually separated and assayed for infection by inoculation of triturated heads (containing salivary glands) or bodies (carcasses and legs) onto Vero cell monolayers and observation for cytopathic effects as previously described5,50.

Mouse studies

For pathogenesis studies 4–6-week-old female CD-1 mice (Charles River) were infected sc. in the rear fp. with WT or mutant EEEV viruses with genome equivalents corresponding to 103 WT PFU. CD-1 mice were also infected ic. with SINV/EEEV WT or mutant viruses with genome equivalents corresponding to 104 WT PFU.

For VLDLR LA(1-2)-Fc decoy studies 4–6 week-old female CD-1 mice were treated with 100 µg of VLDLR LA(1-2)-Fc or LDLRAD3 D1-Fc six hours before sc. infection in the rear footpad with WT or mutant EEEV viruses with genomes equivalent to 103 WT PFU. Mice were monitored for weight loss and clinical signs of illness for 14 dpi.

Quantification of virus-infected cells from PLNs

Quantification of virus-infected cells in the PLNs was performed as described in Trobaugh DW et al.5. CD-1 mice were inoculated s.c. in both rear footpads with equal genomes of EEEV nLuc TaV WT, single mutant viruses, and VEEV WT nLuc TaV at 103 PFU WT. At 8 hpi, PLNs were harvested, flash-frozen on dry ice, homogenized, and then resuspended in 200 µL of reporter lysis buffer before being stored. Expression of nLuc was analyzed using the Nano-Glow luciferase assay system (Promega).

In vivo imaging and bioluminescence analysis

Mice were inoculated s.c. in the rear footpad with equal genome copies of WT EEEV or mutant viruses expressing nLuc. On days 1, 2, 3, 4, and 5 post-infection, mice were injected s.c. with 10 µg of Nano-glow in PBS as previously described6,51. Mice were visualized 4 min after injection using an IVIS Spectrum CT Instrument (PerkinElmer) (Supplementary Fig. 9). The total flux (photons per second) in the footpad, head, or lower body (to capture luminescence in the PLNs) for each animal was determined using Living Image Software 4.5.1.

Mouse serum analysis

Viremia was determined from sera collected during the pathogenesis studies by qRT-PCR as described in Albe et al.52. Briefly, RNA was harvested from sera as described in genome quantification when determining genome to PFU ratios. Levels of other cytokines and chemokines were measured by a mouse cytokine and chemokine multiplex bead array (Millipore) and read on a Bioplex 200 plate reader (Luminex) per manufacturer’s protocol.

Statistical analysis

Statistical significance for all experiments was determined using Prism version 10.5.0 (GraphPad Software) and is indicated in each figure legend. Cell culture experiments were analyzed by one- or two-way ANOVA or repeated measures ANOVA with post-hoc tests. Mouse studies were performed at least twice with similar results, and the significance of survival differences was determined by the Mantel-Cox log-rank test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.