ABSTRACT
Dengue virus (DENV) is responsible for the most prevalent and significant arthropod-borne viral infection of humans. The leading DENV vaccines are based on tetravalent live-attenuated virus platforms. In practice, it has been challenging to induce balanced and effective responses to each of the four DENV serotypes because of differences in the replication efficiency and immunogenicity of individual vaccine components. Unlike live vaccines, tetravalent DENV envelope (E) protein subunit vaccines are likely to stimulate balanced immune responses, because immunogenicity is replication independent. However, E protein subunit vaccines have historically performed poorly, in part because the antigens utilized were mainly monomers that did not display quaternary-structure epitopes found on E dimers and higher-order structures that form the viral envelope. In this study, we compared the immunogenicity of DENV2 E homodimers and DENV2 E monomers. The stabilized DENV2 homodimers, but not monomers, were efficiently recognized by virus-specific and flavivirus cross-reactive potently neutralizing antibodies that have been mapped to quaternary-structure epitopes displayed on the viral surface. In mice, the dimers stimulated 3-fold-higher levels of virus-specific neutralizing IgG that recognized epitopes different from those recognized by lower-level neutralizing antibodies induced by monomers. The dimer induced a stronger E domain I (EDI)- and EDII-targeted response, while the monomer antigens stimulated an EDIII epitope response and induced fusion loop epitope antibodies that are known to facilitate antibody-dependent enhancement (ADE). This study shows that DENV E subunit antigens that have been designed to mimic the structural organization of the viral surface are better vaccine antigens than E protein monomers.
IMPORTANCE Dengue virus vaccine development is particularly challenging because vaccines have to provide protection against four different dengue virus stereotypes. The leading dengue virus vaccine candidates in clinical testing are all based on live-virus vaccine platforms and struggle to induce balanced immunity. Envelope subunit antigens have the potential to overcome these limitations but have historically performed poorly as vaccine antigens, because the versions tested previously were presented as monomers and not in their natural dimer configuration. This study shows that the authentic presentation of DENV2 E-based subunits has a strong impact on antibody responses, underscoring the importance of mimicking the complex protein structures that are found on DENV particle surfaces when designing subunit vaccines.
INTRODUCTION
Dengue virus (DENV) is the most prevalent arthropod-borne viral pathogen and represents a major public health threat, with approximately 400 million infections per year worldwide (1). The four DENV serotypes, transmitted by Aedes sp. mosquitoes, are members of the Flaviviridae family. DENV is endemic in >125 countries, resulting in >100 million infections annually and approximately 1 million cases of severe disease, including severe dengue fever, dengue hemorrhagic fever (DHF), and shock syndrome (1). Primary infections with one of the DENV serotypes lead to specific long-term protective immunity to the serotype of infection but leave the individual susceptible to the other three serotypes. Individuals experiencing a second DENV infection with a new serotype are more likely to develop DHF than individuals experiencing primary infections. Through a phenomenon known as antibody-dependent enhancement (ADE), serotype cross-reactive, nonneutralizing antibodies induced by the primary infection appear to facilitate virus entry through Fcγ receptor-mediated endocytosis and thereby enhance secondary infections (2, 3). DENV vaccines must induce balanced, protective immunity to all four serotypes to avoid the possibility of neutralizing some serotypes while enhancing the replication of other serotypes.
The major target of the humoral immune response against DENV infection is the envelope protein (E). The ectodomain of E is classified into three separate domains: E domain I (EDI), EDII, and EDIII (Fig. 1A). EDI is a β-barrel that connects into EDII, which contains the fusion loop (FL) peptide that facilitates membrane fusion between the virus particle and the host cell plasma membrane. EDIII harbors the presumed receptor binding sites (4–8). DENV E proteins are organized as head-to-tail homodimers on the surface of the virion. Units of three homodimers form 30 herringbone raft-like structures that lie flat on the spherical particle surface. These E oligomers form complex quaternary-structure epitopes of great importance, because these are the targets of highly neutralizing and protective human antibodies (9). Even though E is the major target for neutralizing and protective antibodies, previous recombinantly expressed soluble E (rE) or recombinant E domains have not proven to be particularly effective as vaccine antigens, since they do not induce long-lasting, high-level neutralizing antibody responses (10–13). In aqueous solution, DENV and other flavivirus rE proteins are present in a dynamic monomer-dimer equilibrium that favors the monomeric state under physiological conditions (14, 15). The absence from monomers of structural features, such as quaternary epitopes, that are present only on dimers likely explains the poor immunogenicity of flavivirus rE subunit vaccines in preclinical studies. In support of this hypothesis, recent studies have shown that the oligomeric state of the closely related Zika virus (ZIKV) rE has a strong impact on the level and specificity of neutralizing/protective antibodies (16, 17).
DENV2 rE homodimers are recognized by neutralizing antibodies with a quaternary footprint. (A) The soluble DENV2 rE monomer (rEM; aa 1 to 394) and homodimer (rED; aa 1 to 394 with A259C) were expressed in mammalian cells. (B) The oligomeric states of rEM and rED were confirmed by SDS-PAGE followed by CBB staining and Western blotting using MAb 1M7. (C) The epitope displays on rEM and rED were determined by using DENV2 serotype-specific (2D22) and DENV cross-reactive (12C1.5, 4G2, C8, and B7) MAbs. Antibodies that bind to quaternary structured epitopes are indicated by Q. Data points indicate experimental duplicates, and error bars represent standard deviations.
Here, we investigate if a disulfide-linked DENV2 E homodimer (rED) can serve as a better vaccine antigen than the rE monomer (rEM), previously tested as a vaccine antigen. We demonstrate that unlike rEM, rED is recognized by strongly neutralizing antibodies isolated from patients. rED induced higher levels of neutralizing antibodies, which targeted different antigenic sites on E protein from those targeted by antibodies induced by rEM. These results show that the oligomeric state of DENV2 E-based subunit vaccines has a strong impact on antibody responses, underscoring the importance of mimicking the DENV virion surfaces when designing subunit vaccines.
RESULTS
Expression and characterization of DENV2 rE monomers and homodimers.To evaluate if the oligomeric state of DENV2 E protein subunits influences immunogenicity, we expressed two DENV2 recombinant E variants (amino acids [aa] 1 to 394) that are monomeric (rEM) or form stable homodimers (rED) under physiological conditions. The DENV2 stable homodimers were generated by introduction of a disulfide interaction in the EDII dimer interphase (A259C) (Fig. 1A) (18, 19). The monomeric and dimeric states of the rE proteins were confirmed by SDS-PAGE run under reducing and nonreducing conditions, followed by Coomassie brilliant blue (CBB) staining and Western blotting (Fig. 1B). The protein bands that were detected by Western blotting using monoclonal antibody (MAb) 1M7 had the expected molecular weights of ∼50 kDa for the E monomer and ∼100 kDa for the E dimer. In enzyme-linked immunosorbent assay (ELISA) format binding assays, only the stable rED, not rEM, was recognized by DENV2-specific (2D22) and cross-reactive (C8, B7) potently neutralizing and protective MAbs that bind quaternary epitopes (Fig. 1C). Both the monomers and the dimers were efficiently recognized by MAbs 12C1.5 and 4G2.
DENV2 E dimers induce higher virus-specific IgG and non-cross-neutralizing antibody responses than monomers.After we confirmed the oligomeric state and proper antibody epitope display on rEM and rED subunits, we immunized BALB/c mice with 5 μg rEM, rEM plus alum, rED, or rED plus alum. The mice received boost immunizations at week 3 and week 7, and the experiment was terminated at week 13 after the first dose of vaccine (Fig. 2A). DENV2-specific IgG titers were determined over the course of the experiment (Fig. 2B). Both rEM and rED without any adjuvant induced low IgG levels after priming, but the levels increased upon boosting. In contrast, the adjuvanted groups induced high titers after priming that were only slightly elevated by boost immunizations (Fig. 2B). At the end of the study at week 13, rED induced a ∼3-fold more-potent DENV2-specific IgG response than rEM. The addition of adjuvant enhanced overall DENV2-specific IgG titers, with rED plus alum inducing the highest IgG levels (Fig. 2C). We observed a similar, but not significant, trend for neutralizing antibody titers. Over time, neutralizing antibody titers steadily increased after boost immunization and marginally dropped 5 weeks after the second boost (Fig. 2D). Overall, mice that were immunized with dimeric subunit formulations tended to have higher neutralizing antibody titers than their monomeric-subunit-immunized counterparts, although the differences were not statistically significant (Fig. 2E; see also Fig. S1 in the supplemental material).
DENV2 E homodimers induce high neutralizing antibody responses in mice. (A) Mice were immunized with 5 μg of rEM, rED, rEM plus alum, or rED plus alum at week 0 (W0). Animals received boost immunizations with 5 μg of the same antigen at W3 and W7. Serum samples were taken at W3, W4, W8, and W13, and the experiment was terminated at W13. (B) DENV2-specific IgG titers at 1:50 serum dilutions were determined by virus capture ELISA. (C) DENV2-specific IgG titers at W13 were determined by virus capture ELISA; the area under the curve (AUC) of serially diluted immune serum was calculated using GraphPad Prism software. Each data point represents the average value of duplicates for a mouse. (D and E) DENV2-specific neutralizing antibody titers were determined using Vero-81 cells and were expressed as the serum dilution at which 50% of the virus was neutralized (Neut50). Data points represent results for individual mice, and statistical differences were determined by one-way ANOVA followed by Tukey’s test (P < 0.5).
Next, we evaluated the serotype specificity of the antibody responses by determining the binding to all four DENV serotypes and ZIKV at a 1:50 serum dilution (Fig. 3A). All groups showed the highest binding signal to DENV2 compared to the other DENV serotypes and ZIKV. Irrespective of the addition of adjuvant, mice immunized with monomers induced antibodies that cross-bind mainly to DENV1 and DENV4, with modest signals to DENV3 and ZIKV. DENV2 dimer antigens induced a milder cross-reactive response than the monomers, with moderate binding to DENV1 and low binding to DENV3, DENV4, and ZIKV. Next, we evaluated the cross-neutralizing capacity of the immune sera against all four DENV serotypes (Fig. 3B and Fig. S1). We observed low-level cross-neutralization of DENV1, without major differences between the vaccine groups. DENV3 and DENV4 were poorly neutralized by all groups.
DENV2 rED stimulates a serotype-specific neutralizing antibody response with low levels of cross-reactivity. (A) DENV cross-reactive IgG titers were determined by virus capture ELISA using 1:50-diluted immune sera. Each data point represents the average value of duplicates for a mouse, and error bars represent standard deviations. (B) DENV cross-reactive neutralizing antibody titers were determined at W13 using Vero-81 cells, and titers were expressed as Neut50 (the serum dilution that neutralizes 50% of the virus) values. Each data point represents the result for an individual mouse. (C) Serum antibody binding to DENV2 E domain III was determined by virus capture ELISA using DENV4/2EDIII chimeric viruses at a 1:50 serum dilution. Data points represent results for individual mice, and error bars represent standard deviations.
Next, we analyzed the ability of the immune serum to bind to a DENV chimera that has DENV2 EDIII transplanted into a DENV4 backbone (DENV4/2EDIII) (Fig. 3C) (20). By comparing the DENV4/2EDIII binding signals to those for DENV4 and DENV2, it is possible to estimate the relative levels of DENV2 EDIII antibodies within each group. For all groups, we observed an increase in the binding signal for DENV4/2EDIII compared to DENV4, indicating the presence of vaccine-induced antibodies requiring residues on DENV2 EDIII for binding. Interestingly, all sera from animals immunized with rED or rED plus alum, both of which were unreactive to DENV4, showed high binding signals to DENV4/2EDIII.
Antibody mapping using yeast display of DENV2 envelope domains.To further characterize the specificity of antibodies in mice immunized with rEM and rED subunits, we used transformed Saccharomyces cerevisiae cells that display either DENV2 EDI, both EDI and EDII (EDI/EDII), EDI/EDII with a W101K substitution to disrupt the binding of fusion loop antibodies (EDI/EDIIFLmut), or EDIII on their surfaces (Fig. 4A). All antigens were equipped with a C-terminal Myc (cMyc) tag, which was used to determine the level of protein expression on yeast cell surfaces. The correct folding of all DENV2 E domains displayed was validated using MAbs 3F9 (EDI, EDI/EDII, and EDI/EDIIFLmut), 3H5 (EDIII), 1C19 (EDI/EDII and EDI/EDIIFLmut), and 1M7 (EDI/EDII and EDI/EDIIFLmut) (Fig. 4B). Serially diluted immune sera were incubated with induced yeast cells, and the mean fluorescence intensities (MFIs) of antibodies bound to DENV2 peptides in cMyc-positive yeast cells were calculated (Fig. 4C).
Yeast cell surface display of DENV2 domains reveals that rEM- and rED-induced immune sera differ in E-domain focus. (A) Diagram of N-terminal yeast surface display of DENV2 EDI/EDII via fusion with yeast mating protein Aga2p. Each yeast-displayed DENV2 E domain contains a C-terminal Myc (cMyc) tag for independent detection of expression through binding of anti-cMyc antibodies. Epitope presentation and proper folding of the DENV2 E domain are detected through binding of known DENV2 E-domain-specific antibodies (Table 1). (B) Representative flow cytometry data and gating strategy for analysis of mouse serum antibody binding to DENV2 E domains. Yeast surface expression of DENV2 E domains is measured through chicken anti-cMyc binding (brown), detected by a fluorescently labeled (Alexa Fluor 633) secondary anti-chicken antibody. The double-staining strategy for the DENV2 E domain displayed on yeast cells is represented by yeast cells displaying EDI/EDII that were stained with anti-cMyc and anti-DENV EDII antibody 4G2 (green). This strategy eliminates the background signal from non-E-domain-expressing cells (Q1 and Q2) for analysis of mouse serum antibody binding to cMyc-positive cells. The geometric mean fluorescence intensity (G-MFI) of gated cMyc-positive cells is recorded (Q3 and Q4) and used for AUC analysis. (C) Binding of known DENV2 E-domain-specific antibodies 1C19 (EDII), 1M7 (EDII/FL), 4G2 (EDII/FL), 3F9 (EDI), and 3H5 (EDIII) to EDI/EDII (green), EDI/EDIIFLmut (purple), EDI (red), and EDIII (blue) for validation of proper domain folding and epitope presentation. Positive binding events were compared to the unstained fluorescence signal and E-domain-displaying cells stained with an alternate E-domain antibody as a negative control.
In accordance with the observation that rEM and rED immune sera can bind to the DENV4/2EDIII chimera (Fig. 3C), we observed that both rEM and rED subunits induced antibodies that recognized EDIII displayed on yeast cells. Moreover, the yeast display method of antibody mapping revealed that rEM antigens induce higher levels of EDIII binding antibodies than rED, and this trend was sustained with the addition of adjuvant (Fig. 5A). We observed a seemingly opposite trend for EDI binding. Although the difference is not significant, rED subunits appear to induce slightly higher levels of EDI-targeting antibodies (Fig. 5B). Significant differences in binding were observed for EDI/EDII display. Irrespective of adjuvant coformulation, rED-induced immune sera contained higher levels of EDI/EDII binding antibodies than rEM-induced sera (Fig. 5C). This trend did not change with the FL mutation that was introduced in the EDI/EDIIFLmut construct (Fig. 5D). However, comparing the binding signal of each immune serum to EDI/EDII and EDI/EDIIFLmut enabled us to get an indication of the relative levels of FL binding antibodies. The FL mutation resulted in an ∼80% reduction of EDI/EDII binding for immune sera with rEM, ∼40% for those with rEM plus alum, ∼20% for those with rED, and ∼18% for those with rED plus alum (Fig. 5E).
DENV2 rED-induced antibodies focus on epitopes present in EDI and EDII with minimal induction of FL-targeting antibodies. (A to D) The levels of serum antibodies that bind to DENV2 EDIII (A), EDI (B), EDI/EDII (C), and EDI/EDIIFLmut (D) were determined using yeast cell display of the appropriate DENV2 domains. EDI/EDIIFLmut was generated by a W101K substitution to eliminate the fusion loop epitope. (E) The relative loss of binding to the fusion loop epitope was calculated by comparing the serum antibody binding signals of EDI/EDII and EDI/EDIIFLmut. Each data point represents a result for an individual animal. Statistical differences were determined by one-way ANOVA followed by Tukey’s test (P < 0.5) (F) Heat map of the DENV2 rE proteins indicating which domains on the proteins are targeted by monomer (green)- or dimer (red)-elicited serum antibodies.
DISCUSSION
There is a dire need for a safe and efficacious DENV vaccine that can be administered to all groups at risk of infection. With leading tetravalent live-attenuated DENV vaccines, it has been challenging to formulate vaccines that induce balanced protective responses against each of the DENV serotypes. In-depth characterization of the vaccine-induced immune responses indicated that antibody responses are largely dominated by one of the DENV serotypes, most likely due to differential replication of individual attenuated vaccine strains (21). In DENV-naïve individuals, a poorly balanced vaccine dominated by the replication of one vaccine component can stimulate immune responses that mimic primary DENV immunity and increase the risk of severe dengue disease upon a subsequent DENV infection. Dengvaxia is contraindicated in naïve individuals, because this vaccine increased the risk of severe dengue disease in DENV-naïve children who were vaccinated (22).
Subunit vaccines have the potential to reliably stimulate balanced immune responses to each vaccine antigen, because immunogenicity is not dependent on viral replication. The subunit that structurally resembles the native virus particle most accurately would be a virus-like particle (VLP). DENV VLPs are produced by the coexpression of prM and E, which naturally assemble as replication-defective enveloped particles. Both cross-reactive and serotype-specific dimer-dependent neutralizing antibodies are able to bind to DENV VLP surfaces, indicating that many important epitopes are conserved (23). One major safety advantage of using rE subunits over VLPs is that DENV VLPs are difficult to express and include prM, which has been shown to produce nonneutralizing antibodies that are prone to ADE. Unfortunately, flavivirus E-based subunits have performed poorly as vaccine antigens in preclinical and clinical studies (13). Enhancing the immunogenic potency of E subunits has therefore been a focus of our research (15, 16, 24, 25). Strategies including nanoparticle display of DENV E antigens and the use of potent adjuvants have previously been shown to increase serotype-specific neutralizing antibody responses (24–27). Studying the biochemical properties of flavivirus rE proteins has offered possible explanations for the poor performance of E subunits and opened new avenues for developing better subunit vaccines (14, 28). Recombinantly expressed soluble flavivirus E proteins reside in a dynamic monomer-dimer equilibrium that favors the monomeric state under physiological conditions, leading to the poor display of critical quaternary epitopes that are recognized by neutralizing and protective human antibodies (14, 15). Previous studies with the closely related ZIKV have shown that presenting the E subunit as stable homodimers significantly impacts the functionality and domain focus of protective humoral immune responses (16, 17).
Here. we demonstrate that DENV2 rEM and rED subunit antigens similarly induce distinct antibody responses. DENV2 rED antigens were recognized by serotype-specific and cross-reactive MAbs that bind to epitopes of varying complexity. In comparison to DENV2 rEM, rED stimulated higher levels of virus-specific antibodies. In addition, moderate cross-binding to other DENV serotypes tracked with cross-neutralization in rED immune sera. Conversely, cross-binding antibodies induced by rEM subunits were nonneutralizing. This group of cross-reactive but nonneutralizing antibodies might potentially have negative side effects and could facilitate ADE. Further studies are needed to investigate this observation. Antibody mapping by yeast display showed that DENV2 rEM stimulated an EDIII- and FL-biased response, whereas rED-induced antibodies focused on epitopes present in EDI and EDII (Fig. 5F). In line with previous ZIKV studies, DENV2 rED stimulated responses with minimal induction of FL-targeting antibodies (17). Our observed differences in antibody focus between rEM and rED immune sera are seemingly not influenced by the presence or absence of the alum adjuvant. The adjuvant effect was most prominent after the first immunization, indicating an important role in immune response initiation. It is known that alum acts as an antigen depot and that it recruits leukocytes to the site of infection. Even though DENV2-specific IgG titers were generally higher in the adjuvanted groups, our data do not support the conclusion that the adjuvanted subunits are better vaccine candidates than rE subunits alone. Adjuvants will likely have a stimulating effect on subunit vaccine formulation, but further studies are needed to elucidate this effect.
In conclusion, our previous work with covalently linked ZIKV rED and the current studies with DENV2 rED demonstrate that stabilized dimers stimulate high levels of antibodies directed to complex epitopes that are also the targets of strongly neutralizing antibodies that develop in people exposed to wild-type (WT) flavivirus infections (16). In contrast, rEM, which has not shown much promise as a vaccine antigen, stimulates antibodies mainly directed to simple epitopes contained within each monomer as well as antibodies to the highly conserved fusion loop, which has been implicated as a target of poorly neutralizing, disease-enhancing antibodies. Our results provide a foundation and strong rationale for further developing flavivirus E dimers and higher-order oligomers as subunit vaccines to protect people from endemic and emerging flaviviral infections.
MATERIALS AND METHODS
Cells and viruses.Vero-81 cells were maintained at 37°C under 5% CO2 as a monolayer culture in Dulbecco’s modified Eagle medium (DMEM; Gibco) plus 1% nonessential amino acids and 5% fetal bovine serum, with 100 μg/ml streptomycin and 100 U/ml penicillin. Suspension culture-adapted EXPI293 cells (Thermo Fisher) were maintained in EXPI293 expression medium (Life Technologies). To determine (neutralizing) antibody titers, DENV1 WestPac-74, DENV2 S-16803, DENV3 CH53489, and DENV4 TVP-376 were used. Chimeric DENVs (DENV4/2EDIII and DENV4/23F9) were propagated in C6/36 insect cell lines.
DENV2 recombinant E antigen production and purification.The DENV2 recombinant E (S-16803; aa 1 to 394) monomers (rEM) were expressed using the EXPI293 transient expression system (Thermo Fisher) following the manufacturer’s protocol. The DENV2 stable E homodimers (rED A259C) were expressed in a CHO-DG44 (Thermo Fisher) stable cell line, according to the manufacturer’s protocol. All proteins were equipped with a C-terminal 6×His tag for detection and purification purposes. Antigens were affinity purified by Co2+ Talon (TaKaRa Bio) chromatography columns, followed by size exclusion chromatography. All purified proteins were flash frozen in 1× phosphate-buffered saline (PBS) (pH 7.4) plus 10% glycerol and stored at –80°C.
Protein analysis.Purified rEM and rED antigens were evaluated by ELISA and SDS-PAGE under denaturing and reducing conditions, followed by Coomassie brilliant blue (CBB) staining or Western blotting. For SDS-PAGE, 1-μg portions of rEM and rED proteins were incubated in denaturing loading buffer with or without 4% β-mercaptoethanol for 10 min at 95°C. Loading samples were centrifuged, and proteins were separated by SDS-PAGE followed by CBB staining (0.1% Coomassie blue, 10% acetic acid, and 50% methanol). For Western blotting, proteins were transferred to nitrocellulose membranes and blocked with 3% skim milk in Tris-buffered saline (TBS) plus 0.02% Tween 20 for 1 h at 37°C. Next, the membrane was incubated with 0.5 μg/ml MAb 1M7 in blocking buffer for 1 h at 37°C. Membranes were washed and developed using NBT/BCIP (nitroblue tetrazolium/5-bromo-4-chloro-3-indolylphosphate) substrate (Thermo Fisher), and the reaction was stopped in Milli-Q water.
Mouse immunizations.Female BALB/c mice were purchased from The Jackson Laboratory and were inoculated at 6 to 12 weeks of age. Each mouse was immunized subcutaneously with 5 μg of either DENV2 rEM (n = 5), rEM plus 500 μg alum (Alhydrogel; Invivogen) (n = 5), rED (n = 5), or rED plus alum (n = 5). All groups received three immunizations (days 0, 21, and 49), and serum samples were collected on days 21, 28, 56, and 91 by submandibular bleeding.
Vaccine-induced antibody evaluation.Antibody responses specific to the four DENV serotypes or the DENV4/2 chimeras were evaluated by antigen capture ELISA, and immune responses to each serotype were measured separately. Briefly, DENV was captured on 1M7-coated ELISA plates and incubated with serially diluted mouse immune sera for 1 h at 37°C. Bound antibodies were detected by alkaline phosphatase (AP)-conjugated anti-mouse IgG Abs (1:2,500; Sigma); plates were developed using alkaline phosphatase substrate; and absorbance was measured at 405 nm. IgG titers were quantified by determining the area under the dilution curve (AUC) using GraphPad Prism software.
Evaluation of neutralizing antibody responses.To evaluate the neutralizing capacity of the mouse immune sera, Vero-81 cells were seeded (25,000 cells per well) and incubated overnight at 37°C. Immune sera were serially diluted in Opti-MEM (Gibco) plus 2% fetal bovine serum and were incubated for 45 min at 37°C with the amount of DENV that had been previously determined to infect 15% of the Vero cells. The cells were washed with Opti-MEM, and the medium was aspirated. The virus-serum mixture was added to the cells and incubated for 2 h at 37°C. Following incubation, the cells were washed with DMEM plus 2% fetal bovine serum and were incubated overnight at 37°C. The next day, cells were washed with PBS and harvested from the plate by trypsin (Gibco). Cells were fixed with 4% paraformaldehyde, washed with permeabilization buffer, and blocked with 1% normal mouse serum in permeabilization buffer for 30 min at room temperature. Next, the infected cells were stained with Alexa Fluor 488-conjugated MAb 1M7 (5 μg/μl) for 1 h at 37°C. The cells were subsequently washed in permeabilization buffer and were resuspended in 200 μl fluorescence-activated cell sorter (FACS) buffer. The percentage of infected cells was evaluated using the Guava Flow cytometer (EMD Millipore), and the neutralizing capacity of serum was expressed as Neut50 (the serum dilution at which 50% of the virus was neutralized). The Neut50 value was calculated by GraphPad Prism software.
Construction and yeast display of DENV2 E-domain peptides.The pETCON yeast display vector and the Saccharomyces cerevisiae yeast strain EBY100, gifts from the Baker Laboratory (University of Washington, Seattle, WA, USA), were used for all yeast display experiments. The DENV2 16681 strain E amino acid sequence (GenBank accession no. NC_001474.2) was used to design the EDI (aa 1 to 49, 136 to 190, 280 to 296) EDI/EDII (aa 1 to 296), and EDIII (aa 296 to 394) gene fragments. To eliminate the fusion loop epitope (FLE), a W101K mutation was introduced to EDI/EDII (EDI/EDIIFLmut). For the EDI construct, the following changes were introduced to connect the β-turns: between aa 49 and aa 136, an Asn-Gly linker was used, whereas between aa 190 and aa 280, the native Thr-Gly-Thr-Gly sequence was used. Codon-optimized gene fragments were synthesized with a 50-bp homologous recombination sequence (HR-5′ [TAGTGGTGGAGGAGGCTCTGGTGGAGGCGGTAGCGGAGGCGGAGGGTCG] and HR-3′ [GAACAAAAGCTTATTTCTGAAGAGGACTTGTAATAGCTCGAGATCTGATA]) by Twist Bioscience. Each gene fragment was PCR amplified with primers that anneal to 5′ HR (forward) and 3′-HR (reverse), and the pETCON vector was linearized using NheI, BamHI, and SalI restriction endonucleases (Thermo Fisher). Next, 1 μg of digested pETCON and 3 to 5 μg of an appropriate gene insert (vector/insert ratio, 1:3 to 5) were transformed into electrocompetent EBY100 yeast cells. The yeast was recovered for 1 h at 30°C in YPD medium (10 g/liter yeast extract, 20 g/liter peptone, and 20 g/liter dextrose) and was subsequently pelleted and resuspended in 1 ml of SDCAA medium (20 g/liter dextrose, 5 g/liter Casamino Acids, 6.7 g/liter yeast nitrogen base, 5.40 g/liter Na2HPO4, and 7.45 g/liter NaH2PO4). The transformed yeast was serially diluted on SDCAA agar plates and grown for 48 to 72 h at 30°C for the selection of successful stable clones. Positive clones were expanded into 5 ml of SDCAA plus 100 μg/ml streptomycin and 100 U/ml penicillin for 20 to 24 h at 250 rpm and induced in SGCAA (20 g/liter galactose, 2 g/liter dextrose, 5 g/liter Casamino Acids, 6.7 g/liter yeast nitrogen base, 5.40 g/liter Na2HPO4, and 7.45 g/liter NaH2PO4) at 20°C. Surface expression of the DENV2 peptides was validated with flow cytometry (Guava; Millipore) by surface staining of a chicken anti-Myc primary antibody (Thermo Fisher) at 4 μg/ml and an Alexa Fluor 488-conjugated anti-chicken secondary antibody at 8 μg/ml. Correct folding of the DENV2 peptides was validated by using DENV MAbs 1M7, 4G2, 1C19 (against EDI/EDII and EDI/EDIIFLmut; 1 μg/ml), 3H5 (against EDIII; 1 μg/ml), and 3F9 (against EDI/EDII, EDI/EDIIFLmut, and EDI; 500 μg/ml) (Table 1) (20, 29–35). Confirmed stable clones were selected for all future experiments, and inserted gene sequences were confirmed by Sanger sequencing (Eurofins Genomics).
Dengue virus-specific monoclonal antibodiesa
Binding of mouse serum antibodies to DENV2 EDI, EDIII, EDI/EDII, and EDI/EDIIFLmut on yeast cell surfaces.For binding analysis, we used freshly passaged yeast cells expressing the appropriate DENV peptide or domains. Cells were washed three times with ice-cold 1× PBS (pH 7.4) plus 0.1% bovine serum albumin (BSA) by centrifugation at 4,000 rpm for 1 min at 4°C. Per mouse serum sample, 2 × 106 yeast cells were incubated with a chicken anti-Myc MAb (4 μg/ml) and diluted mouse immune serum for 1 h at room temperature on a rotator. After incubation, cells were washed three times and incubated with a peridinin chlorophyll protein (PerCP)-conjugated anti-mouse antibody (at 2 μg/ml; BioLegend) and an Alexa Fluor 488-conjugated anti-chicken antibody (8 μg/ml) for 1 h at room temperature. Next, the cells were washed three times and resuspended in PBS plus 0.1% BSA. Binding of serum antibodies to the DENV2 peptides displayed on yeast cells was determined using flow cytometry (Guava; Millipore). Myc-positive cells were gated, and the mean fluorescence intensity (MFI) was used as a measure of serum antibody binding. The area under the curve (AUC) of diluted serum incubated with yeast cells was calculated using GraphPad Prism software, and significant differences between groups were determined by one-way ANOVA followed by Tukey’s test (P < 0.5).
Ethics statement.All mouse experiments were performed under protocols approved by the Institutional Animal Care and Use Committee of the University of North Carolina. All experiments followed ethical guidelines and federal regulations according to the Public Health Service Policy on Humane Care and Use of Laboratory Animals, the Animal Welfare Act, and the Guide for the Care and Use of Laboratory Animals (36).
ACKNOWLEDGMENTS
We thank the Baker Laboratory at the University of Washington for kindly providing the yeast cells and expression plasmids. We also thank the Baric Laboratory at the University of North Carolina for providing the chimeric DENV viruses.
This research was supported by U.S. Department of Defense grant PR171402P1 (principal investigators, B. Kuhlman and A. M. de Silva) and U19 grant AI109784-01 (principal investigator, J. Ting).
FOOTNOTES
- Received 22 April 2020.
- Accepted 23 June 2020.
- Accepted manuscript posted online 1 July 2020.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
