Adjuvant-Mediated Epitope Specificity and Enhanced Neutralizing Activity of Antibodies Targeting Dengue Virus Envelope Protein

The heat-labile toxins (LT) produced by enterotoxigenic Escherichia coli display adjuvant effects to coadministered antigens, leading to enhanced production of serum antibodies. Despite extensive knowledge of the adjuvant properties of LT derivatives, including in vitro-generated non-toxic mutant forms, little is known about the capacity of these adjuvants to modulate the epitope specificity of antibodies directed against antigens. This study characterizes the role of LT and its non-toxic B subunit (LTB) in the modulation of antibody responses to a coadministered antigen, the dengue virus (DENV) envelope glycoprotein domain III (EDIII), which binds to surface receptors and mediates virus entry into host cells. In contrast to non-adjuvanted or alum-adjuvanted formulations, antibodies induced in mice immunized with LT or LTB showed enhanced virus-neutralization effects that were not ascribed to a subclass shift or antigen affinity. Nonetheless, immunosignature analyses revealed that purified LT-adjuvanted EDIII-specific antibodies display distinct epitope-binding patterns with regard to antibodies raised in mice immunized with EDIII or the alum-adjuvanted vaccine. Notably, the analyses led to the identification of a specific EDIII epitope located in the EF to FG loop, which is involved in the entry of DENV into eukaryotic cells. The present results demonstrate that LT and LTB modulate the epitope specificity of antibodies generated after immunization with coadministered antigens that, in the case of EDIII, was associated with the induction of neutralizing antibody responses. These results open perspectives for the more rational development of vaccines with enhanced protective effects against DENV infections.

inTrODUcTiOn Adjuvants are essential components of vaccines, particularly those composed of purified antigens, which naturally have lower immunogenicity than live attenuated microorganisms. Adjuvants are known for their ability to enhance the magnitude of adaptive immune responses, particularly antibody responses, to coadmi nistered antigens; they may impact the longevity, antigen avidity, and modulation of isotype and IgG subclass switches (1,2).
To date, only a limited number of adjuvants have been incorporated into human vaccines, but several different substances, including bacterial toxins, are known for their strong and rather versatile adjuvant effects in various mammalian hosts (3,4). We are inves tigating the changes in epitope recognition elicited by adjuvants. Heatlabile toxins (LT) naturally produced by some entero toxigenic Escherichia coli strains are enterotoxins endowed with potent adjuvant effects when coadministered with purified solu ble or particulate antigens after delivery via different parenteral or mucosal routes (5)(6)(7)(8)(9). LTs, similarly to cholera toxin, are com posed of one toxic A subunit with enzymatic ADPribosylation activity and a B subunit pentamer that binds to host cell receptors (10,11). LT adjuvant effects, as well as those associated with in vitrogenerated nontoxic LT derivatives, including purified B subunits and mutated LT forms, have primarily been ascribed to the activation of dendritic cells and B and T lymphocytes, lead ing to enhanced mucosal and serum antigenspecific antibody responses (6,(12)(13)(14)(15)(16)(17)(18)(19). In addition, previous evidence indicated that LT derivative may modulate the epitope specificity of antibodies raised against coadministered antigen (12). Indeed, adjuvant mediated modulation of epitope specificity of antibodies is still a poorly understood issue, particularly in situations involving the design of vaccines capable of inducing protective immunity to various infectious diseases.
Dengue fever is caused by an arbovirus transmitted by Aedes aegypti mosquitoes. Fifty to one hundred million people are affected by the disease in tropical and subtropical regions annually, and thousands are potentially at risk to develop more serious side effects, which may lead to death caused by the illness (20,21). The dengue virus (DENV) has an envelope composed of a single protein (E) and encodes two other structural (prM and C proteins) and seven nonstructural proteins (22)(23)(24). The glycoprotein E is composed of three structural and functional domains that mediate the initial steps of the DENV infectious cycle. Receptor binding is mainly triggered by domain III [enve lope glycoprotein domain III (EDIII)], which leads to the sub sequent entry and replication of virus into susceptible host cells (22,24,25). EDIIIspecific antibodies can confer protection to DENV infection and have been considered protection correlates for antiDENV vaccines (26)(27)(28)(29)(30)(31). Nonetheless, depending on the specificity, concentration and affinity of antibodies targeting the E protein, antibodies may either block virus infection or promote enhancement of infection (the antibody-dependent enhancement effect-ADE) via cell entry mediated by Fcγ receptors (32)(33)(34)(35).
Proteomics platforms such as microarrays have been exploited as promising highthroughput assays to measure humoral immune responses, assisting our understanding of the subtle characteris tics of successful immunization, a key challenge in the rational development of effective vaccines (36)(37)(38). The immunosignature technology is a microarraybased platform that uses unnatural randomizedsequence peptides, spotted in high density, as a universal and robust assay to identify antibodybinding patterns of natural or induced responses in both humans and mice. Due to its successful applicability for both infectious and noninfectious diseases, this platform can be used as a powerful tool for system atic studies of various vaccine approaches (39)(40)(41)(42)(43)(44)(45)(46).
In this study, we compared the adjuvant effects of alum, LT and LTB in vaccine formulations containing a recombinant form of the DENV EDIII antigen. The results demonstrate that, in contrast to mice immunized with the alumadjuvanted formulation, mice parenterally immunized with EDIII admixed with LT or LTB mounted a more efficient antibody response capable of inhibit ing DENV infection. Immunosignature analysis of the humoral immune response elicited in vaccinated mice showed that LT dif ferentially modulates the epitope specificity of the EDIIIspecific IgG antibodies. In addition, the results led to the identification of an EDIIIderived peptide specifically involved in the infectivity of the virus into host cells. Together, this study emphasizes the role of adjuvants in the modulation of the epitope specificity of antibodies induced upon vaccination and, thus, may impact the development of antiDENV and other subunit vaccines.

cloning of adjuvant lTB
LTB was obtained after cloning the nucleotide sequence of the eltB gene (encoding the B subunit of the LT from H10407 strain) into the pET22b vector. The eltB sequence (450 bp) was amplified by PCR using the forward primer 5′CAGAAGGCGGAATGA CATATGAATAA3′ and reverse primer 5′TTAAAAGCATGT CTCTCGAGAGAACC3′ (the underlined sequences indicate the NdeI and XhoI restriction sites, respectively). The insert generated by the PCR reaction was digested and ligated into the corresponding NdeI and XhoI restriction sites of the pET22b expression vector (NovagenMerck Bioscience, MA, USA), generating the recombinant plasmid pET22bLTB, which was subsequently used to transform chemically competent E. coli DH5α. Transformants were analyzed by digestion with the enzymes NdeI and XhoI. The recombinant DNA fragment was sequenced and compared to data reported for the LT sequence (GenBank database GI: 408994). A chemically competent E. coli BL21 (DE3) strain was transformed with the pET22bLTB plasmid and denominated as E. coli BLLTB.

expression and Purification of adjuvants
The LT and LTB adjuvants were obtained from E. coli strains har boring the plasmid pML19 (47,48) or pET22bLTB, respectively. The purification of LT was based on a previously described method (47). The recombinant E. coli strains were cultivated in Luria Bertani (LB) medium or Terrific Broth (TB) supplemented with 100 µg/ml of ampicillin at 37°C in an orbital shaker (200 rpm). While LTB expression was obtained after 0.75 mM IPTG induc tion of the E. coli BLLTB strain cultures for 6 h, LT production was achieved by constitutive expression of the eltAB gene in overnight cultures of a pML19harboring E. coli strain. The cell pellets were suspended, and the bacterial cells were disrupted by high pres sure in a homogenizer (APLAB10 model, ARTEPEÇAS, Brazil). The LT and LTB contained in the bacterial cell homogenates were purified by affinity chromatography on immobilized dgalactose columns (Pierce, Waltham, MA, USA) previously equilibrated with TEAN buffer (50 mM Tris, 1 mM EDTA, 3 mM NaN3, and 0.2 M NaCl, pH 7.4) in an Akta FPLC (Amersham Pharmacia Biotech, Buckinghamshire, UK). The elution step was performed with TEAN supplemented with 0.3 M dgalactose. The proteins were monitored by 15% sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDSPAGE). The concentrations of the purified proteins were determined by absorbance measurements in a spec trophotometer (Gene Quant spectrophotometer GE Amersham Biosciences, Buckinghamshire, UK) as previously described (49).

cloning, expression, and Purification of antigen
The pE1D2 plasmid (50), harboring gene encoding the ectodo main of the envelope glycoprotein from the DENV2 New Guinea C (NGC) strain, was kindly provided by Dr. Ada Alves and used as template for PCR reactions. The sequences of the forward and reverse primers used in PCR reactions were 5′ACATGCGAGGATCCG GAATGTCATACTCTAT3′ (the underlined sequence indicates the BamHI restriction site) and 5′GCCTTCTACTCGAGTTAC GATAGAACTTCCTTTCTTA3′ (the underlined sequence indi cates the XhoI restriction site), respectively. The PCR product was introduced into pET28a(+) expression vector (Novagen, Germany), generating the recombinant plasmid pDEDIII. The E. coli BL21 (DE3) strain harboring the plasmid pDEDIII was cultivated in LB medium containing 50 µg/ml kanamycin. The culture at an OD600nm of 0.5 was induced by IPTG to a final concen tration of 0.5 mM and incubated for 4 h. The bacterial pellet was suspended in buffer A (0.1 M Na2PO4, 0.5 M NaCl, pH 6.8) and lysed in the APLAB10 homogenizer (ARTEPEÇAS, Brazil). The inclusion bodies were solubilized in buffer A supplemented with 8 M urea, and the recombinant protein was refolded in buffer A. The soluble protein was submitted to nickel affinity chromatogra phy using a HisTrap™ HP column (GE Healthcare Life Sciences, Buckinghamshire, UK). The purified EDIII was evaluated by 15% SDSPAGE and measured by spectrophotometry.

endotoxin removal from Purified Protein Preparations
The endotoxin was removed using a Detoxigel Endotoxin Removing Gel (Pierce, Waltham, MA, USA), and any residual endotoxin was detected by the Limulus Amebocyte Lysate (LAL) QCL 1000 kit (Lonza, Basel, Switzerland). Following treatment for LPS removal, the residual endotoxin contents in the LT, LTB, and EDIII samples were less than 0.1 EU/μg of protein, corre sponding to the levels allowed for preclinical research (51).

Biological characterization of lT and lTB adjuvants
The cytotonic activity of the LT derivatives on the adrenal Y1 cells was evaluated as previously described (6). The Y1 cells were seeded in 96well plates (5 × 10 4 cells/well) and exposed to 1 µg of LT or LTB, or to phosphatebuffered saline (PBS) as negative control, diluted in Dulbecco's modified Eagle's medium (DMEM, Vitrocell Embriolife, Brazil) supplemented with 2% fetal bovine serum (FBS, Vitrocell Embriolife, Brazil). After 8 h of incubation, the cells were observed for the cytotonic effect. The GM1ELISA was performed as previously described (52) to evaluate LT binding to the cognate ganglioside GM1. Microtiter plates (Polysorp, Thermo FisherNunc, Roskilde, Denmark) were coated with GM1 ganglioside (1 µg/ml) (SigmaAldrich, St. Louis, MO, USA) diluted in PBS (pH 7.4) and incubated overnight at room temperature in a humid chamber. Plates were blocked with PBS containing 0.1% BSA. Then, serially twofold diluted LT samples were applied in duplicate wells. To detect LT derivatives, the plates were sequentially incubated with antiLT serum (titer equal to 10 5 ) and horseradish peroxidase (HRP)conjugated antimouse IgG antibodies (SigmaAldrich) diluted to 1:1,000 and 1:3,000, respectively, in PBS containing 0.05% Tween20 and 0.1% BSA. The chromogenic reactions developed with orthophenylenediamine dihydrochloride (SigmaAldrich) and H2O2 were stopped with 1 M H2SO4 and measured at A492nm.

analysis of the Biological activity of recombinant eDiii in Vero cells
This assay was performed as previously described (53). African Green Monkey Kidney Epithelial Cells (Vero Line), which express receptors for E glycoprotein, were cultured in minimum essential medium (MEM) supplemented with 2% FBS and incubated at 37°C in an atmosphere containing 5% CO2 for 24 h. Cells were incubated with 100 µg/ml of EDIII protein in MEM at 37°C in 5% CO2 for 30 min. After washing with PBS with 2% FBS, Vero cells were fixed with 4% paraformaldehyde (SigmaAldrich, St. Louis, MO, USA) for 10 min, treated with antiEDIII sera harvested from mice immunized with EDIII plus Freund's adjuvant for 60 min and subsequently with goat antimouse IgG conjugated to FITC (Invitrogen, Paisley, UK) for further 60 min. The FITCstained Vero cells were detected by BD LSRFortessa flow cytometry (BD Bioscience, San Jose, CA, USA), and the data were analyzed using the FlowJo v10 program.

animal immunizations
All experiments involving mice were approved by the Committee on the Ethical Use of Laboratory Animals of the Institute of Biomedical Sciences (CEUA 198) at the University of São Paulo (USP), in accordance with the guidelines for the care and use of laboratory animals adopted by the National Council of Animal Experimentation (CONCEA). The mice were obtained from the Isogenic Mouse Breeding Facility of Department of Parasitology, Institute of Biomedical Sciences-USP. Female BALB/c mice (6-8 weeks old) were immunized via the subcutaneous (s.c.) route at days 0, 14, and 28 of the immunization protocol with the vaccine formulations. Groups of five BALB/c mice were treated with 100 µl of PBS alone as a negative control or PBS containing 10 µg of EDIII with or without one of the following adjuvants: 1.0 µg of LT, 3.2 µg of LTB, or 12.5 µg of Al(OH)3 (Rehydragel, Reheis, NJ, USA). Serum samples were harvested 2 weeks after each vaccine dose and stored at −20°C until immunological analyses were conducted. elisa ELISA for antiEDIII antibodies was performed as previously described (53). Microtiter plates (Maxisorp Thermo Fisher Nunc, Roskilde, Denmark) were coated with EDIII (1 µg/ml) diluted in PBS pH 7.4 and incubated overnight at 4°C. For ELISA using denaturing conditions, the EDIII protein was submitted to heat treatment at 100°C for 15 min and used to coat the plates. Plates were blocked with PBS containing 3% gelatin for 2 h at 37°C. The serum samples harvested from mice immunized with EDIII alone or coadministered with each adjuvant were applied to duplicate wells and serially twofold diluted. The serum IgGs were measured with HRPconjugated antimouse IgG (1:3,000), IgG1 (1:10,000), or IgG2a (1:3,000) antibodies (SigmaAldrich, St. Louis, MO, USA). Absorbance at 492 nm was obtained as previously described and used to calculate the antibody titers, defined as the highest sample dilution able to generate an A492nm of 0.2 above the preimmune sera.

Purification of serum igg Fractions
AntiEDIII IgG antibody fractions were obtained from serum samples harvested from mice submitted to EDIIIbased immuni zation regimens. First, the recombinant 6×Histagged EDIII was immobilized in a column filled with a nickelcoupled resin. After serum delipidation, as previously described (54), the sera were submitted to affinity chromatography on EDIIIimmobilized resin using 0.05 M PBS, pH 7.4. The antiEDIII antibodies were eluted in 0.1 M glycine buffer (pH 4), neutralized with 1 M Tris-HCl (pH 9.0), and dialyzed in PBS (pH 7.4). Subsequently, the antiEDIII antibodies were subjected to affinity chromatog raphy to protein G sepharose according to the manufacturer's instructions (GE Healthcare, Buckinghamshire, UK), and the IgG fraction was obtained following the elution step described above. The purified EDIIIspecific IgG antibodies were measured by BCA assay (Pierce, Waltham, MA, USA) and monitored by 12.5% SDSPAGE.

affinity assay for anti-eDiii igg antibodies
The antibody affinity index was measured using a modified ELISA protocol with ammonium thiocyanate (47). Briefly, the plates were coated with EDIII as described above. The sera collected from mice from different immunization groups were tested at dilutions corresponding to an OD492nm of 0.8, while purified antiEDIII IgG antibodies were evaluated at an OD492nm of 0.5. After incubation with sera or purified antibodies, different concentrations of sodium thiocyanate were added to wells and incubated for 15 min. The plates were washed and incubated with HRPconjugated antimouse IgG antibody. The percentage of antibodies bound to EDIII was determined: OD492nm in the presence of ammonium thiocyanate × 100/OD492nm in the absence of ammonium thio cyanate. The values obtained with the serum samples represent the antibody avidity, which under our conditions depends on the affinity and valence of the immunoglobulins since the parameter of structural arrangement was standardized by the use of the antigen immobilized on the solid support. The analyses of the purified IgG antibodies showing constant Ig valence generated average affinity values for each immunization group.

Plaque reduction neutralization Test (PrnT)
To perform the PRNT, the DENV2 NGC strain, kindly provided by Laura Helen Vera Gonzales Gil (Fiocruz, Recife, Brazil), was purified, and the Vero cells were cultivated as previously described (35). Briefly, the NGC strain was propagated in the Aedes albopictus cell line C6/36 cultured in Leibovitz medium containing 5% FBS and grown at 28°C for 7 days. The superna tant of C6/36 infected with the NGC strain was collected and titrated for the determination of PFU/ml. For the neutralization test, Vero cells were maintained in DMEM supplemented with 10% FBS and seeded in 24well plates (1 × 10 5 cells/well) 24 h before infection. Serum samples from immunized mice were inactivated for complement proteins for 30 min at 56°C and serially twofold diluted. Purified EDIIIspecific IgGs were tested at different concentrations. C6/36 cell supernatants contain ing 500 PFU of DENV/ml were incubated with each dilution of antiEDIII sera or purified IgG antibodies for 1 h at 37°C. Vero cells were washed with serumfree DMEM and infected in duplicate wells with 200 µl of the neutralization mixture for 1 h at 37°C. The viral suspension was removed, and the cells were overlaid with 1 ml of complete DMEM [2.5% FBS and 1% carboxymethylcellulose (Synth, São Paulo, Brazil)]. Following incubation of the plates at 37°C for 7 days, the cells were fixed for 15 min with 4% paraformaldehyde and stained with crystal violet for 10 min. The percentage of plaque reduction compared to the positive control (DENV not exposed to antibodies) was calculated. Neutralizing antibody titers were expressed as the serum dilution, or IgG concentration, yielding 50% plaque number reduction (PRNT50).

immunosignature assay
For the peptide microarray assay, slides were manufactured using in situ synthesis of 330,000 (CIM330K) sequences of nonnatural randomized peptides as previously described (40). Statistical analyses of microarray data were performed using JMP 12 software (Statistical Discovery Software from SAS) by importing data analysis from Excel and imageprocessed data from GenePix Pro6.0 (Molecular Devices, Sunnyvale, CA, USA). Raw intensities were normalized to each slide by divid ing all values per array by the median of that array (median normalization). Poor quality spots were excluded from analysis by flagging them as "absent" upon visual inspection. Values less than 0.01 were set to 0.01, and values from duplicate arrays were averaged and used in the analysis. Important peptides were determined by Student's twotailed ttest, with Benjamini and Hochberg multiple test correction applied to a p value < 0.05. For principal component analysis, we used significant peptides for each vaccine approach and an analysis using JMP 12 soft ware (39). The prediction of recognized epitopes was based on lists of informative peptides compared to the EDIII sequence from the DENV2 strain NGC (AHG97599.1) using GuiTope (55). The amino acid substitution matrix was provided by the peptide library and protein sequence. An inversion weight of 1 was selected. Library subtracted scores were returned using the mean of 10 random samplings of the peptide library and a minimum score cutoff of 8.0. The results were graphed using GraphPad Prism v5 using a moving average of 15. competition assay for DenV2 Using Peptides This assay was adapted from a protocol previously described (56,57). Vero cells (1 × 10 5 cells/well, on 96well plates) were cultured in MEM supplemented with 3% FBS and incubated at 37°C in an atmosphere containing 5% CO2 for 18 h. Intact or heatdenatured EDIII pro tein, as well as peptide 47 (369AEPPFGDSYIIIGVEPGQLK388), custommade by GeneScript (Piscataway, NJ, USA), at different molar concentrations, was incubated with the DENV2 strain NGC (multiplicity of infection-equal to 1.0). The mixtures were added to Vero cells at 37°C in 5% CO2 for 1 h. The mixtures were removed, and MEM supplemented with 2% FBS was added to each well. Following 18 h of incubation, the culture plates were washed with PBS and treated with 50 µl/well of trypsin (2.5 mg/ml) plus EDTA at 37°C in 5% CO2 for 10 min to remove the cells. Vero cells were resuspended with PBS containing 10% FBS and fixed and permeabilized with Cytofix/Citoperm kit (BD Bioscience, San Jose, CA, USA) for 10 min. The cells were incubated for 1 h with mAb 4G2, which recognizes the fusion loop at the extremity of the domain II of the envelope protein (ATCCHB112), and subsequently labeled with antimouse IgG conjugated with Alexa 488 (BD Biosciences) for 30 min. The stained cells were measured by flow cytometry using a BD LSRFortessa (BD Bioscience, San Jose, CA, USA) instrument, and the data were analyzed with FlowJo v10 software to determine the amount of DENVpositive Vero cells. Viral replication analysis was carried out at 24 h post infection based on a protocol previously described (58). RNA was extracted from 100 µl of infected Vero cell, using a guanidine isothiocyanate phenol method (Trizol LS, Invitrogen), according to the manufacturer's instructions. The quantitative realtime polymerasechainreaction was carried out with a set of primers and probes with FAM as dye reporter for the probe. Primers/ probes used in assay for DENV were previously described (58). The assay was performed using the AgPathIDTM OneStep RTPCR reagents (Applied Biosystems). Succinctly, we used 5 µl of the extracted RNA in 1 µl of the mix from primers/probe (10 pM/μl) and 19 µl of the reagent mix from AgPathIDTM One Step RTPCR kit following the manufacturer's instructions.

structural analyses
The amino acid sequence composing the main predicted epitope was indicated in a previously published structural model of the quaternary DENV2 envelope glycoprotein (PDB accession num ber 1OKE) using PyMol v1.8.4.0.

statistical analyses
The data are represented as arithmetic means ± SD and analyzed for variance (ANOVA) with a subsequent Bonferroni's multiple comparison test using GraphPad Prism v5 software. For the data generated by the PRNT and affinity assay, the statistical analyses were calculated from two independent experiments, while all other results were analyzed from three independent experiments. resUlTs lT and lTB enhance eDiii-specific antibody responses The recombinant forms of LT, LTB and type 2 DENV EDIII were purified by affinity chromatography (Figures S1 and S2 in Supplementary Material). Both LT and LTB bind to host cell receptors, but as expected, only LT had cytotonic effects to Y1 cells (Figures S1C,D in Supplementary Material). The recombi nant EDIII also preserved the receptorbinding function of the native viral protein ( Figure S2B in Supplementary Material). Mouse groups were immunized with purified EDIII and vaccine formulations containing EDIII admixed with alum, LT, or LTB. Mice received three s.c. doses at intervals of 2 weeks and were bled one day before each dose and 2 weeks after the last vaccine dose ( Figure 1A). Significant increases in the EDIIIspecific IgG antibody levels were detected in serum samples of mice immu nized with vaccines containing LT and LTB, but not with alum compared to EDIII alone (Figures 1B,C). The antigenspecific IgG1/IgG2a ratios in vaccinated mice were similar, indicating that the adjuvants were not able to modulate the IgG subclass responses induced by the antigen alone. Although antiEDIII IgG1 and IgG2a titers detected in mice immunized with LT and LTB enhanced in comparison with the nonadjuvanted EDIII, significant differences were only found regarding to IgG1 levels ( Figure 1D). In addition, EDIIIspecific antibodies in serum samples collected from vaccinated mice were capable of recogniz ing both conformational and linear epitopes of the antigen, as demonstrated by reactivity with both intact and heatdenatured antigen. Serum samples from mice immunized with LT or LTB showed significantly higher reactivity with both conformational and linear epitopes than those collected from mice immunized with nonadjuvanted EDIII or EDIII admixed with alum ( Figures 1E,F).

antigen reactivity and DenV-neutralization Properties of Purified eDiii-specific antibodies
To evaluate the role of adjuvants in the modulation of the effec tor activities of antigendirected antibodies induced by subunit vaccines, EDIIIspecific antibodies raised in different mouse groups were purified by affinity chromatography (Figure S3 in Supplementary Material). The purified antiEDIII antibodies reacted with the native antigen but did not recognize the heatdenatured antigen, suggesting that the purified antibod ies are specifically directed toward conformational epitopes (Figure 2A; data not shown). No difference in antigen affinity was detected in EDIIIspecific antibodies raised in the different immunization groups (Figure 2B). AntiEDIII antibodies, either in serum or purified fractions from vaccinated mice showed virusneutralization activity in vitro; however, antiEDIII anti bodies elicited with LT or LTB were more efficient than those (e,F) Reactivity of serum antibodies carried out by ELISA with the native purified EDIII antigen (e) or the same antigen submitted to a heat denaturation treatment (F) used as solid phase bound antigens. Anti-EDIII titers were determined in three independent experiments with at least five animals per immunization group. Values represent means ± SD of the IgG titers. ***p < 0.001, **p < 0.01, and *p < 0.05, comparing adjuvanted EDIII-immunized mice to non-adjuvanted EDIII-immunized mice or alum-adjuvanted mice (ANOVA with Bonferroni post hoc test).  Table S1 in Supplementary Material). The mice immunized with LTB coadministered to EDIII elicited antigenspecific sera with twofold higher neutralizing capability than LTadjuvanted vaccination ( Figure 2C; Table S1 in Supplementary Material). The purified antiEDIII IgG from both LT and LTBadjuvanted mice yielded neutralization titers (PRNT50) of 1.6 ng/ml and thereby revealed strong virusneutralizing effects ( Figure 2D).

immunosignatures of anti-eDiii antibodies
To gain a better understanding of the enhanced virus neutralization activity of the antiEDIII antibodies raised in mice immunized with different vaccine formulations, we deter mined the immunosignatures of purified EDIIIspecific anti bodies raised in different immunization groups (Figure 3). By profiling the antibodies with random linear peptide sequences spotted on microarray slides, we observed that antibodies from mice immunized with EDIII and LT or LTB had distinct immunological signatures for those raised in mice immunized without adjuvant or with alum (Figures 3A,B). In addition, fewer peptides were recognized with high binding intensity by EDIIIspecific IgGs adjuvanted with LT or LTB, suggesting an increased specificity of these antibodies (Figures 3A,C). Analysis of peptidebinding profiles indicated that antibodies raised in mice immunized with LT and LTB reacted with 119 peptides sequences each, while antibodies raised in mice immu nized only with antigen, or with alum admixed with the antigen, reacted with 605 and 603 peptides, respectively. Among these peptides, only 10 were solely recognized by antibodies raised in mice immunized with LT or LTB and 88 peptides among all groups ( Figure 3C).

epitopes Predicted by immunosignatures are critical for DenV infection
Prediction of epitopes present in the EDIII antigen and recog nized by the different set of purified IgG antibodies showed that mice immunized with both LT derivatives reacted with a specific sequence comprising the EF loop, F betasheet, and FG loop between amino acid residues 371-384 (Figures 3D and 4A,B).
To further investigate the biological relevance of this finding, we tested a 21mer synthetic peptide encompassing the complete predicted epitope of the EDIII antigen ( Figure 4B). As shown in Figure 4C and Figure S4 in Supplementary Material, the peptide (EDIII369-388) efficiently blocked the infectivity of DENV2 NGC FigUre 3 | Immunosignature analyses of purified envelope glycoprotein domain III (EDIII)-specific IgG antibodies. (a) Heat map demonstrating the immunosignatures detected with EDIII-specific IgG antibodies purified from serum samples collected from mice submitted to the different immunization regimens and a mock mouse group (control). A Student's t-test (p < 3.33 × 10 −6 ) between vaccine groups was used to select the informative peptides. Hierarchical clustering using EUCLIDEAN distance was used as a measure of similarity to cluster the selected peptides (x axis) and vaccine groups (y axis). Peptides' intensity is colored where blue corresponding to low intensity and red to high intensity. (D) Antigenic epitope prediction for vaccine groups using random peptide arrays. Significant peptides sequences recognized by antibodies of each vaccine group were submitted to epitope prediction with the GuiTope software and using the EDIII protein sequence from DENV2 New Guinea C (AHG97599.1) as reference. Each line graph represents the GuiTope prediction score for each vaccine group.

DiscUssiOn
A considerable amount of information regarding the use of bacterial toxins such as LT as vaccine adjuvants has accumulated during the last four decades. Nonetheless, perspectives on the clinical use of nontoxic LT derivatives have been overshadowed by the induction of undesirable side effects (transient facial paralysis) observed in patients submitted to intranasal immu nization trials (59). However, recent observations describing the successful use of LT derivatives as parenterally administered adjuvants rekindled interest in these adjuvants under both experimental and clinical conditions (8,9,(60)(61)(62). In this study, we evaluated the adjuvant properties of LT and a nontoxic derivative composed of the B subunit pentamer as parenterally delivered adjuvants admixed with a recombinant form of the DENV2 EDIII. Our results demonstrated that, unlike alum, LT and LTB enhanced antibody responses to EDIII and induced the generation of antibodies with increased capacity to neutralize DENV2 particles. In addition, we demonstrated that both LT and LTB drastically changed the epitopebinding profiles of EDIIIspecific IgG antibodies. Immunosignature analyses per formed with peptide microarrays led us to identify an epitope located between the EDIII EF and FG loops, which is specifically recognized by antibodies raised in mice immunized with LT or LTB but not by antibodies raised in mice immunized with alumadjuvanted EDIII or nonadjuvanted antigen. Remarkably, a synthetic peptide encompassing the epitope identified by the immunosignature analyses interfered with the infectivity of DENV2 in mammalian cells, opening perspectives for therapeu tic and preventive interventions. Collectively, this study demon strated that LT derivatives can modulate the epitope specificity of antibodies after parenteral immunization with purified proteins and thus may contribute to the rational design of subunit vac cines for DENV and other infectious agents. The evidence presented here demonstrates that LT used as a parenteral adjuvant promotes strong neutralizing antibody responses to DENV infection. Antibody neutralization titers achieved in mice immunized with EDIII and LT or LTB were significantly higher than those elicited in mice immunized with alum. In accordance with these findings, the EDIIIspecific IgG fraction purified from mice immunized with LT or LTB showed greater virusneutralizing activity than antibodies purified from mice immunized with alum. Furthermore, purified IgG col lected from mice immunized with LT or LTB revealed a strong virusneutralizing effect compared to the previously described antiEDIII mAbs, such as 2H12, 9D12, and 1A1D2 that block DENV infection under in vitro conditions with different intensity (26,32,63,64). The results emphasize the fact that parenterally administered LT represents a promising adjuvant alternative comparable to the most potent adjuvants presently in use under both experimental and clinical conditions. In addition, the obser vation that purified LT B subunits exert similar adjuvant effects as LT indicated that the local inflammatory reactions induced by the enzymatically active toxin can be abrogated without reducing the immunomodulatory effects to coadministered antigens.
Our results obtained with purified IgGs demonstrated that the specificity, and not the magnitude, of the antibody response is critical in virus neutralization. In fact, antibodymediated virusneutralization activity has been strongly correlated with antibody/antigen interaction force and/or epitope recognition patterns (12,26,47). In our case, the distinct neutralization effects of the EDIIIspecific polyclonal antibodies were not associated with antigen affinity. In accordance with these results, several studies have reported that the virusneutralizing activity of anti EDIII mAbs or sera does not necessarily correlate with antigen affinity (33,35,(64)(65)(66)(67). However, epitope recognition patterns of EDIIIspecific mAbs showed a positive correlation with virus infection blockage activity (26,(32)(33)(34)65). In this context, we demonstrated that the immunosignature patterns of antibodies raised in mice immunized with LT derivatives differed drastically from those raised in mice immunized with nonadjuvanted EDIII or alumadjuvanted antigen. More importantly, such differences correlated with an enhanced virusneutralizing activity of the IgG antibodies induced by LT forms. Although corroborative evidence remains to be gathered, this study clearly demonstrated that adjuvantmediated modulation of epitope recognition patterns by the antibodies raised is a critical aspect of vaccineinduced immune responses, particularly with regard to the induction of immunity to DENV.
In this report, the immunosignature analyses of EDIII specific polyclonal antibodies disclosed that, in contrast to alum, incorporation of LT or LTB to vaccine formulation resulted in antibodies with increased specificity to the epitopes located at the EDIII EF and FG loops in addition to the F βstrand. In support of this concept, our study demonstrated that a synthetic peptide comprising the predicted epitope interfered with the in vitro infectivity of DENV, as measured with Vero cells. In addition, the lateral FG loop has been implicated in DENV infectivity both in mosquito and mammalian cells (57,68,69). More importantly, part of the EDIII epitope unveiled by the immunosignature analyses, comprising the F βstrand and FG loops, has also been identified as a target epitope for antiDENV and antiZIKV mAbs isolated from high responders in clinical studies with ZIKVinfected individuals (70). Together, these findings emphasize the relevance of the EDIII lateral ridge region as a target for antibodies capable of conferring protection from flaviviruses.
The role of LT derivatives, in the context of active immuniza tion practices, on the modulation of epitope recognition profiles of antibodies has been scarcely investigated. Here, and in previ ous reports, we have demonstrated that LT derivatives shape the epitope recognition profile of antibodies to different pathogens [(12), unpublished data]. Previous studies have demonstrated that LT derivatives improve the expression of costimulatory mol ecules on B lymphocytes and dendritic cells and, consequently, on prime CD4 + T cells, leading to modulation of Th1/Th2/Th17 cytokine production patterns (12,17,19,(71)(72)(73). Although future studies will have to be performed, the capacity to program B cells using cognate T helper lymphocytes, concomitant with direct activation effects, may represent possible alternatives to explain the action of LT derivatives on the modulation of the epitope recognition pattern of antibodies.
Synthetic peptides have been largely exploited in therapeutic approaches against viral diseases such as hepatitis, acquired immunodeficiency syndrome and influenza (74). However, little effort has investigated the use of peptides for treatment of arbo viroses. Antidengue peptides under preclinical evaluation were designed after in silico analyses and target the stem region, hydro phobic pocket or FG loop of the E glycoprotein (57,69,75,76). These peptides showed halfmaximal inhibitory concentration (IC50) for virus infection or binding to host cells at micromolar levels (57,75,76). Here, based on a different experimental approach, we report the identification of a peptide that affects DENV infectivity with an IC50 in nanomolar concentration range. In contrast to previous approaches, our results demonstrate that, besides having a direct effect on the host cell/virus interaction, such peptides may be amenable to generated protective antibody responses and therefore find both prophylactic and therapeutic applications.
Epitope mapping based on linear peptides has been a routine technique for the characterization of both monoclonal and specific polyclonal antibodies. The antibody immunosignature approach relies on the use of random peptides microarrays that broader the epitope screening inasmuch not specific to a patho gen or a vaccine, resulting in reduced cost and time. In addition, random peptides can be reactive with antibodies directed against conformational epitopes, which are not possible with conven tional peptide microarray (40,76). In fact, our data showed that even without the specific EDIII sequences on the array it was possible to detect DENV neutralizing peptide demonstrating the applicability of this technique.
In conclusion, our results demonstrated that LT and LTB enhance both the magnitude and the specificity of the antibody responses elicited in mice immunized with an antiDENV subu nit vaccine, promoting changes to the epitopebinding profiles that, in turn, lead to enhanced virusneutralization effects with regard to alum. Such findings bring relevant insights into the design of vaccine formulations based on purified proteins and the Adjuvant-Mediated Anti-Dengue Antibody Epitope Specificity