The Influence of Codon Optimization on the Immune Response and Protective Efficacy of DNA Vaccines based on the Dengue Envelope and NS1 Proteins

1. Introduction

The worldwide occurrence of dengue has significantly increased in recent years, representing a potential risk to approximately half of the global population [1]. The World Health Organization established the goal of controlling dengue by 2030, achieving a 0% fatality rate [2]. The dengue virus (DENV) comprises four genetically related serotypes (DENV1-4), and is transmitted by Aedes (Ae) aegypti or Ae. albopictus mosquitoes [3]. This enveloped virus contains a single-stranded RNA genome which encodes the structural proteins, capsid (C), envelope (E) and premembrane/membrane (prM/M), as well as the non-structural proteins NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5 [4].

The DENV infection can vary from asymptomatic to a range of clinical presentations, encompassing oligosymptomatic forms to severe cases [5]. While infection with one DENV serotype induces long-lasting protective immunity against the infecting serotype, the probability of manifesting severe dengue is significantly increased in secondary infections with heterologous serotypes [6,7,8].

The advancement of a dengue vaccine has encountered various obstacles, with the main one being the construction of a vaccine that effectively protects against the four viral serotypes, with a sustained efficacy and without the risk of vaccinated individuals manifesting severe dengue after contact with the virus [9,10]. There are currently two licensed dengue vaccines, Dengvaxia (Sanofi-Pasteur) and Qdenga (Takeda Pharmaceuticals). The Dengvaxia consists of four chimeric viruses composed of the yellow fever vaccine virus (YFV-17D) backbone and sequences encoding the prM and E proteins of the four DENV serotypes. Phase 3 trials showed that the vaccine efficacy and safety varied depending on the individual serological status, the DENV serotype and the age [11,12,13]. The implementation of this vaccine in endemic countries led to serious safety concerns since increased hospitalization risk was observed in dengue-naïve vaccines exposed to DENV. Currently, the vaccine is only recommended for individuals aged between 9 and 45 years, seropositive for dengue [13,14]. The Qdenga, which was licensed in 2022, consists of four chimeric viruses containing the attenuated DENV2 backbone in which the genes encoding the prM and E proteins were substituted by sequences from the other three DENV serotypes [15]. During the phase 3 trial conducted among children and adolescents residing in dengue-endemic regions across Asia and the Americas, the vaccine efficacy also varied depending on the recipient dengue serostatus and serotype, with greater efficacy for DENV2 [16].

All these results point out the need to develop other vaccine strategies against dengue. During the COVID-19 pandemic, nucleic acid vaccines were consolidated as an alternative to conventional vaccine platforms. RNA vaccines have been administered globally and the first DNA vaccine for use in humans against COVID-19 was recently licensed [17,18,19]. These vaccines induce broad-spectrum humoral and cellular immune responses. Although preclinical research in several animal models demonstrated the efficacy of the DNA vaccines, clinical studies revealed suboptimal immunogenicity generated by such vaccines. Thus, various approaches have been devised to increase the efficacy of DNA vaccines, including codon optimization of the specific genes according to the host species to amplify the in vivo recombinant protein production [18,20,21,22].

The E protein is the primary focus of dengue vaccines because of its role in the virus adhesion and entry into host cells, fusion of viral and cell membranes, and delivery of the virus RNA into cell cytoplasm [10,23,24,25]. This glycoprotein is structured in dimers, with each monomer comprising three domains (DI, DII and DIII), that together form the ectodomain, and a transmembrane domain, the stem and anchor region located in the C-terminal region [26,27]. The E protein plays an important role in the humoral response, mainly by inducing the production of neutralizing antibodies [25,28]. The NS1 protein elicits robust immune responses, stimulating humoral and cellular immune responses during viral infection [29,30]. This glycoprotein is found at the cellular membrane and is released as a hexamer from infected cells during the acute phase of the infection [31,32,33,34]. It plays a role in viral replication, acts in the evasion of the immune system and in some other aspects of dengue pathogenesis [35,36,37,38]. Several reports demonstrated the protective effect of NS1 and pointed it as an important antigen in the composition of a vaccine against dengue [14,39,40,41,42,43].

In prior studies, we developed two DNA vaccines containing the ectodomain of the E (pE1D2) sequence or the ns1 (pcTPANS1) gene from DENV2. Inoculation of these plasmids in mice elicited high protection levels against DENV2 challenge [40,44]. Considering that dengue proteins are naturally expressed in both mosquito and human cells, in the current study we aim to improve the efficacy of these DNA vaccines by increasing expression of the recombinant E and NS1 proteins in the mouse model used for testing the vaccines, as well as in humans for future clinical tests. For this purpose, we analyzed the efficiency of codon optimization of the DNA vaccines containing e (pEotmD2) and ns1 (pNS1otmD2) genes from DENV2. Human and murine cells were transfected with these new plasmids or the original vaccines, pE1D2 and pcTPANS1, to evaluate expression of the recombinant proteins in both systems. Mice were immunized with all these DNA vaccines, followed by the challenge with DENV2. In general, the codon-optimized pE1otmD2 plasmid was less efficient in mediating the envelope protein expression and eliciting neutralizing antibodies than the plasmid pE1D2. These results were also reflected in the survival and morbidity following viral challenge, in which the pEotmD2 demonstrated to be less protective. In turn, the codon optimized pNS1otmD2 plasmid mediated an increased expression of the recombinant protein NS1 and induced slightly higher antibody and protection levels compared to the pcTPANS1 plasmid.

2. Materials and Methods

2.1. Dengue Virus and Cells

The NGC DENV2 (strain New Guinea C, GenBank NCBI, M29095) was utilized in mice challenge assays, whereas for plaque reduction neutralization tests (PRNT50) we used the DENV2 44/2 [45]. For the PRNT50, Vero cells (ATCC^®, CCL-81) were maintained in medium 199 with Earle’s salts (M199, Sigma) with 5% fetal bovine serum (FBS, Invitrogen), sodium bicarbonate (25 mg/mL), in the presence of gentamicin (0.04 mg/mL, Sigma) at 37 °C with 5% CO₂. The in vitro expression evaluation of the recombinant E and NS1 proteins was performed in two different cell lines: mouse neuroblastoma, Neuro2A (ATCC^®, CCL-131) and human hepatoma, HuH7 (cells were obtained from Dr. Da Poian, Leopoldo de Meis Institute of Medical Biochemistry, Federal University of Rio de Janeiro). The Neuro2A cells were cultivated in RPMI-1640 medium (Sigma) and HuH7 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM, Sigma), both supplemented with 10% FBS at 37°C with 5% CO₂.

2.2. Original pE1D2 and pcTPANS1 Plasmids

The construction of the DNA vaccines pE1D2 and pcTPANS1 was previously described [40,44]. Briefly, the pE1D2 vaccine contains 80% of the e gene (sequence which encodes the ectodomain of the envelope protein) from DENV2, and the pcTPANS1 contains the DENV2 ns1 gene. For secretion of the recombinant proteins, the genes were fused to the human tissue plasminogen activator signal peptide (t-PA) sequence. The plasmid pcTPA, containing only the t-PA sequence, was used as a negative control.

2.3. Construction of the Codon Optimized pEotmD2 and pNS1otmD2 Plasmids

To increase expression of the envelope (the ectodomain) and NS1 DENV2 proteins, the e and ns1 genes, as well as the t-PA sequence, were codon optimized for expression in mouse and human cells (GenScript Biotech). Optimization was performed by reducing codon frequencies that are used less in mice and humans, and by increasing the frequencies of those commonly used in these organisms (Figure S1). The codon adaptation index (CAI) was calculated using the codon usage of both humans and mice. The C-G content along the sequences was also adjusted. The synthetic genes were cloned into the pcDNA3 plasmid (Invitrogen), flanked by the EcoRV and XhoI restriction enzyme sites, generating the pEotmD2 (80% of the e gene) and pNS1otmD2 (ns1 gene) constructs. Escherichia coli, DH5-α strain, were subjected to transformation with the recombinant plasmids. To identify positive clones, restriction mapping was performed, and confirmed by sequencing (ABI PRISM dye terminator, Applied Biosystems), carried out by the DNA Sequencing Genomic Platform (Fiocruz, Rio de Janeiro, Brazil).

2.4. DNA Vaccine Purification

Production of plasmids was obtained from transformed E. coli. Purification of DNA vaccines and the control pcTPA were carried out by alkaline lysis and subsequent purification on ion exchange columns, utilizing the “Qiagen EndoFree Plasmid Mega Kit” (QIAGEN), following the manufacturer’s guidelines. Plasmids were suspended in water and stored at -20°C until use. DNA quantification was performed using a Bio Photometer (Eppendorf) at 260 nm.

2.5. Cell Transfection and Immunofluorescence Assay

The Neuro2A or Huh7 cells were seeded at a concentration of 2 × 10⁴ cells/well in chamber slides (LabTek, Nunc) with Opti-MEM medium (Invitrogen) and cultured overnight at 37 °C with 5% CO₂. After 24h, cells were transfected with one of the plasmids (pEotmD2, pE1D2, pNS1otmD2, pcTPANS1 or the control pcTPA), using Lipofectamine 2000 (Invitrogen), following the manufacturer’s recommendations. For Neuro2A, 0.8 µg of the recombinant plasmids were employed, whereas 1.0 µg was used for HuH7 cells. The next day, cells were fixed (4% paraformaldehyde in 0.1 M phosphate buffer, PB), then permeabilized (0.6% saponin in PB) and blocked (1% bovine serum albumin, BSA) (Sigma) and 0.2% saponin in PB). Cells were initially incubated for 1 h at 37 °C with the specified primary antibodies: monoclonal anti-DENV2 3H5 antibody (ATCC), which targets domain III of the E protein, or DENV-2 hyperimmune ascitic fluid (ATCC) that recognizes the NS1 protein. Cells were subsequently incubated for an additional hour at 37 °C with anti-mouse IgG conjugated to FITC (Southern Biotech). The visualization of cells was carried out using the Nikon Eclipse 50i fluorescence microscope (Nikon Inc).

2.6. Cell Transfection and Flow Cytometry

The Neuro2A (1 × 10⁶ cells/flask) and HuH7 (5 × 10⁵ cells/flask) were seeded in 75 cm² culture flasks, with Opti-MEM medium and cultured overnight at 37 °C with 5% CO₂. The following day, Neuro2A and HuH7cells were transfected with 8 μg or 10 μg, respectively, of one of the recombinant plasmids (pEotmD2, pE1D2, pNS1otmD2, pcTPANS1 or pcTPA) using lipofectamine 2000, following the instructions of manufacture. Following a 24-hour incubation, cells were detached using Trypsin-EDTA, fixed (4% paraformaldehyde in PB) and permeabilized (0.05% saponin in PB). For the intracellular staining of E and NS1 proteins, the same antibodies described for the immunofluorescence were used. Ten to twenty thousand events were acquired in a FACS Calibur flow cytometer (BD Bioscience) at the Flow Cytometry Platform of Oswaldo Cruz Institute (IOC-Fiocruz, Rio de Janeiro, Brazil) and data were analyzed using FlowJo v10 (BD Bioscience).

2.7. Mice Immunization and Challenge with DENV2

Experiments involving mice were conducted in compliance with ethical principles in animal experimentation outlined by the Brazilian College of Animal Experimentation and approved by the Animal Use Ethical Committee of Oswaldo Cruz Institute, Oswaldo Cruz Foundation (CEUA-IOC approval ID: L039/2015 and L022/2019). Two independent experiments were performed (5 or 10 mice per group). BALB/c mice (specific pathogens free, SPF), male, four-week-old, were acquired from the Multidisciplinary Center for Biological Investigations (CEMIB, UNICAMP, Campinas, SP, Brazil).

Animals received two doses of one of the plasmids via the intramuscular (i.m.) route: pEotmD2, pE1D2, pNS1otmD2, pcTPANS1 or pcTPA. Doses of 100 μg of DNA diluted in 100 μl of phosphate buffer saline (PBS) were administered (50 µg of DNA in each tibialis posterior muscle), two weeks apart. For evaluation of the humoral and cellular immune responses, blood and spleen were collected two weeks after the second DNA dose. Blood samples were obtained by heart puncture. Before receiving the first DNA dose, pre-immune sera were collected from the animals through retro-orbital puncture.

For protection assessment, two weeks after the last DNA vaccine dose, other animal groups were subject to an intracerebral (i.c.) challenge using 30 μL of a neuroadapted NGC DENV2 (3.5 log10 PFU/mL) diluted in serum-free E199 medium, which corresponds to approximately 40 LD50. Non-immunized animals were inoculated with DENV2, serving as a control group. Following the DENV2 challenge, animals were observed for a period of 3 weeks to assess mortality and morbidity. Morbidity was quantified using a subjective scale ranging from 0 to 4: 0 - no clinical signs; 1 - one limb paralysis or spinal column alteration; 2- severe paralysis in one limb and spinal column alteration or substantial paralysis on both hind limbs; 3 -substantial paralysis in the two hind limbs and spinal column alteration; and 4 - death. Animals were euthanized 21 days after the challenge.

2.8. Detection of Antibodies Against the E and NS1 Proteins

The ELISA assay was used to quantify specific anti-E or anti-NS1 antibody titers generated in BALB/c mice after immunization. Two independent experiments were performed with 5 animals per group. For detection of E-specific antibodies in ELISA plates, the solid-phase bound antigen was the domain III of the E protein (EDIII), which was produced in E. coli and gently provided by Dr. Luis C. Ferreira (University of São Paulo, SP, Brazil). Anti-NS1 antibodies were detected by using the recombinant NS1 protein (expressed in S2 cells), which was kindly given by Dr. Beth-Ann Coller (Merck Sharp and Dohme Corp.). The antigens EDIII or NS1 were adsorbed on ELISA MaxiSorp plates (Nunc), at the concentration of 0.2 µg and 0.1 µg for well, respectively, at 37 °C for 1 h. Subsequently, the plates were incubated with a blocking buffer (2% skim milk in 0.05% Tween-20-PBS - PBST), overnight at 4°C. The following day, the plates were washed with PBST and serum samples were added in duplicates with serial dilutions, followed by 1 h incubation at 37 °C. After another round of washing, the plates were incubated with goat anti-mouse IgG conjugated to horseradish peroxidase (Southern Biotechnology) at 37 °C. After 1 hour, plates were washed and reactions were measured at A_490nm in a microplate reader (SpectraMax 190, Molecular Devices) with OPD (ortho-phenylenediamine dihydrochloride, Sigma) and H₂O₂. The reaction was stopped with 9N H₂SO₄. The specific antibody titers were calculated as the reciprocal of serum dilution that produced an absorbance above that of the corresponding preimmune serum.

2.9. Plaque Reduction Neutralization Test (PRNT50)

Vero cells were seeded overnight in 96-well plates, 2 × 10⁴ cells per well, in E199 medium with 5% FBS and cultured at 37 °C with 5% CO₂. The next day, serum samples were initially subjected to a 56 °C heat treatment for 30 min to inactivate the complement system. Sera were two-fold serially diluted (ranging from 1:20 to 1:1260) in 60 µL of E199 medium. Subsequently, they were mixed with 60 µL of the DENV2 44/2, which contained approximately 30 PFU, and incubated at 37 °C with 5% CO₂. After 1 hour, 100 µL of the serum/virus mixture were added to Vero cell monolayers in duplicates, and plates were incubated for 1 h at 37 °C with 5% CO₂. The supernatant of the wells was removed, and 150 µL of E199 medium containing 5% FBS and 2% carboxymethylcellulose (Sigma) were added, and cell monolayers were incubated at 37 °C with 5% CO₂ for five days. The cells were fixed with 10% formalin solution and stained with 0.02% crystal violet for 30 min. Two independent experiments were performed with 5 animals per group. Neutralizing antibody titers were quantified based on a 50% plaque reduction (PRNT50).

2.10. Interferon Gamma ELISPOT Assay

The IFN-γ ELISPOT assay was carried out using the IFN-γ ELISPOT mouse set from BD Biosciences in accordance with the manufacturer’s instructions. Animals were euthanized two weeks after the second DNA vaccine dose. Splenocytes were isolated and stimulated with synthetic peptides, specific to the CD8⁺ T cell, found in the envelope ³³¹SPCKIPFEI³³⁹ or NS1 ⁶⁵AGPWHLGKL²⁷³ proteins [46,47]. In summary, 96-well ELISPOT plates were coated with anti-IFN-γ capture monoclonal antibody (100 µL/well, diluted in PBS), and incubated overnight at 4 °C. The next day, plates were washed with PBS, blocked with RPMI-1640 medium (Sigma) containing 10% FBS for 2 h at 37 °C and splenocytes were seeded in triplicates (5 × 10⁵ cells/well) in RPMI-1640 medium with 10% FBS. Non-stimulated and concanavalin A (Con A, 2 µg/well) stimulated cells were used as negative and positive controls, respectively. After 18 hours of incubation at 37 °C with 5% CO₂, cells were removed, plates were washed with distilled water and PBST, and anti-IFN-γ biotinylated detection antibody was added and incubated for 2 h at 37 °C. Following another round of washing with PBST, the plates were incubated with streptavidin-horseradish peroxidase (HRP) conjugate for 1 hour. Finally, plates were washed with PBST and the spots were revealed by adding the AEC substrate set (BD/Pharmingen) along with hydrogen peroxide (Sigma) for 20 min at room temperature. The reaction was stopped with distilled water and the spots were quantified using an automated immunospot reader (Cellular Technology Ltd.) at the ELISPOT Multi-User Platform (Fiocruz, RJ, Brazil). Two independent experiments were performed with 5 animals per group.

2.11. Statistical Analysis

All statistical analysis was conducted using GraphPad Prism v9.0 (GraphPad Software), with a minimum level of significance of 95%. Differences among groups were analyzed by the Kruskal–Wallis test with Dunn’s correction for the ELISA, PRNT and ELISPOT experiments. Survival and morbidity rates were estimated using the Log-Rank (Mantel-Cox) statistical test. Data were displayed as mean ± SD and the significance level for each analysis is indicated in the Figure legends.

3. Results

3.1. The Effect of Codon Optimization on E and NS1 Protein Expression in Mouse and Human Cells

In this study, we investigated the efficacy of the new DNA vaccines based on E and NS1 dengue proteins. In an attempt to increase antigen expression, we designed synthetic coding sequences for E and NS1 DENV2 proteins based on codons frequently used in both mouse and human cells. The codon optimization process increased the CAI from 0.71 to 0.87 for the envelope sequence and from 0.74 to 0.87 for the ns1 gene. In the optimized sequences, the CG content was also changed from 45.08% to 52.34% in envelope and from 44.79% to 51.17% in ns1 sequences. Peaks of %GC content in a 60 bp window were eliminated.

The codon optimized DNA vaccines were referred to as pEotmD2 and pNS1otmD2, while the previous constructs encoding the native e and ns1 genes were named pE1D2 and pcTPANS1, respectively. The antigen expression was evaluated in vitro by immunofluorescence and flow cytometry after the transfection of mouse Neuro2A and human HuH7 cells. The percentage of Envelope- or NS1-positive cells and the mean fluorescence intensity (MFI) were compared between the original (pE1D2 and pcTPANS1) and the codon-optimized plasmids (pEotmD2 and pcTPANS1). Flow cytometry gating strategy is represented in Figure S2.

The E protein was detected in both Neuro2A (Figure 1A), and HuH7 (Figure 1B) cells transfected either with pE1D2 or pEotmD2, showing that the new codon-optimized pEotmD2 construct retained its ability to mediate antigenic expression. We observed more E-positive Neuro2a cells when transfected with the plasmid pE1D2 compared to pEotmD2. This difference was also reflected in HuH7 cells, but to a lesser extent (Figure 1C). Regarding the envelope expression, differences between transfections were only detected in HuH7 cells, in which the pE1D2 plasmid led to higher protein production when compared to the codon optimized DNA pEotmD2 (Figure 1D).

The NS1 expression seemed to be slightly more effective when mediated by the codon optimized pNS1otmD2 plasmid. The protein was detected in Neuro2a (Figure 2A) and HuH7 (Figure 2B) transfected with both DNAs, pcTPANS1 and pNS1otmD2. In Neuro2A cells, the percentage of NS1-positive cells were somewhat higher when transfected with the pNS1otmD2 than with the pcTPANS1 (Figure 2C). This difference was more accentuated in human HuH7 cells (Figure 2C). However, levels of the NS1 expressed in the two cell lines were similar, as evidenced by the MFI (Figure 2D). As anticipated, cells transfected with the control pcTPA plasmid showed no positivity for the E or NS1 detection (Figure 1 and Figure 2).

3.2. The Humoral and Cellular Immune Responses Generated by Optimized or Original DNA vaccines

The effect of codon optimization on the immunogenicity of the DNA vaccines encoding the E or NS1 proteins was evaluated in immunized BALB/c mice. Two weeks after full vaccination, mouse sera and splenocytes were collected for assessment of antibody and T-cell responses (Figure 3A).

Regarding the E protein, we detected significant EDIII-specific antibody titers in animals immunized either with pE1D2 or pEotmD2, although the magnitude of this response was lower in the group inoculated with the optimized DNA when compared to mice vaccinated with the original plasmid, pE1D2 (Figure 3B). These results were also reflected in serum neutralization assays. The DENV2 neutralizing antibody titers in mice immunized with the pE1D2-DNA vaccine were statistically different compared to the pcTPA control group (Figure 3C). Mice vaccinated with the pEotmD2 also showed detectable neutralizing antibodies, although titers were lower when compared to the pE1D2-inoculated animals and not statistically different from the control group (Figure 3C).

Concerning the cellular immune response, we observed significant IFN-γ production by T-cells collected from animals vaccinated either with the pE1D2 or the pEotmD2 plasmid. Furthermore, unlike the antibody response, the pEotmD2-immunized group showed a slightly higher number of E-specific T-cells than pE1D2-inoculated mice (Figure 3D).

The pcTPANS1 and pNS1otmD2 DNA vaccines, in turn, induced similar anti-NS1 antibody levels, statistically higher than the control pcTPA group (Figure 3E). Both vaccines also activated specific IFN-γ-producing T cell response upon stimulation with the NS1 peptide. Although not statistical, the number of NS1-specific T cells was slightly higher after vaccination with the pcTPANS1 in comparison with the codon-optimized pNS1otmD2 (Figure 3F).

3.3. Codon Optimization Failed to Improve the Efficacy the NS1- or E-Based DNA Vaccines

To assess the impact of codon optimization on the protection generated by the DNA vaccines, immunized BALB/c mice were challenged with DENV2, two weeks after full vaccination (Figure 4A). The efficacy of vaccination was assessed daily by survival and morbidity.

Immunization with the pE1D2 DNA vaccine generated 100% survival against the DENV2 challenge. However, the group immunized with the optimized plasmid pEotmD2 showed 60% survival, with deaths starting on the 7th day after challenge. In contrast, only 35% of non-immunized DENV2-challenged mice survived, with peak death between the 7th and 15th days post-infection (Figure 4B). Morbidity was also observed, regarding the percentage of affected mice as well as according to an arbitrary pre-defined score scale of clinical signs of infection. The pE1D2 DNA vaccine provided full protection against DENV2, since no mice presented any infection sign (Figure 4C,D). Conversely, around 70% of pEotmD2-immunized mice showed morbidity (Figure 4C), with a high number of animals reaching the 3 or 4 morbidity degrees (Figure 4D). As expected, most non-immunized mice (85%) showed clinical signs of infection (Figure 4C,D). Furthermore, no statistical differences were observed between the pEotmD2 and non-immunized groups concerning the survival or morbidity parameters, thus indicating that codon optimization impaired protection generated by the E-based DNA vaccine.

The pcTPANS1 and the codon-optimized pNS1otmD2 generated high percentages of protection, with 90% and 100%, survival rates, respectively (Figure 5A). Only 10% of the pcTPANS1-immunized mouse group showed the highest degree of morbidity between days 7 and 10 post-infection (Figure 5B,C). Although all pNS1otmD2-vaccinated animals survived virus challenge, 15% of them showed morbidity between the 13th and 15th days post-infection (Figure 5D) with degrees varying between 1, and 3 (Figure 5C). Thus, the two NS1 DNA vaccines generated significant protection compared to the non-immunized group, either in terms of survival or morbidity (Figure 5). Although there were no statistical differences between groups immunized with the pcTPANS1 compared to the codon-optimized pNS1otmD2 plasmid, the 100% survival rate from the pNS1otmD2 group reveals a slight improvement in vaccination-mediated protection.

4. Discussion

Considerable efforts have been made to improve the immunogenicity and efficacy of the dengue DNA vaccine, such as exploring alternative administration routes, employing immunostimulatory CpG motifs, optimizing codons, incorporating genetic adjuvants, and implementing heterologous prime/boost regimens [10,21,48].

Within this context, in previous studies our group assessed a range of DNA vaccine constructs focused on the DENV2 E and NS1 antigens. The objective was to identify the most effective design for eliciting robust immunogenicity and providing protection. Four DENV2 NS1 DNA vaccines were developed, each with unique designs. Two constructs, pcTPANS1ANC and pcENS1ANC, utilized a hydrophobic segment from NS2a to direct NS1 to cell surfaces, while the other two, pcTPANS1 and pcENS1, were designed for protein secretion. These differences affected NS1 expression mammalian cells, and consequently the resulting immunogenicity and protection in vivo. Among these different DNA vaccines, the pcTPANS1 and pcENS1 triggered the most robust antibody responses and protection, with the pcTPANS1 eliciting a slightly superior level of protection [34]. Regarding the envelope protein, two vaccines were constructed: the pE1D2, containing the ectodomain (DI, DII and DIII) of the DENV2 E sequence and the pE2D2, containing only the DIII sequence, both fused to t-PA sequence. Both vaccines elicited neutralizing antibodies, but pE1D2 provided greater protection compared to pE2D2 [44]. Additionally, pcTPANS1 and pE1D2 were effective in inducing a cellular immune response in mice [49].

Therefore, given the protective efficacy of these two vaccines, in the present investigation, the nucleotide sequences encoding the envelope ectodomain and NS1 proteins were codon-optimized for expression in both murine and human cells. The objective of this work was to further increase the protective capacity of these DNA vaccines either in murine experimental models or in future clinical trials. Codon optimization is an important approach for enhancing protein expression in heterologous systems, such as DNA vaccines. Such approach, which exploits the degenerate nature of the genetic code, alters several critical parameters for expression efficiency, such as codon usage bias, GC content, mRNA secondary structure, cryptic splicing sites, premature PolyA sites, among others [50,51,52]. Several studies have evidenced that codon optimization leads to an increase not only in protein expression, but also in the immunogenicity and protection of several DNA vaccines for different pathogens [53,54,55,56,57,58,59].

Initially, in vitro tests were carried out to compare the ability of the plasmids containing the native genes, pE1D2 and pcTPANS1, with the codon-optimized constructs, pEotmD2 and pcNS1otmD2, in mediating E and NS1 proteins expression in both murine and human cell lines. Assessment of transfection efficiency revealed that the optimized pEotmD2 plasmid exhibited slightly lower performance compared to the pE1D2, while the codon-optimized plasmid pNS1otmD2 demonstrated slightly superior efficiency in comparison to the pcTANS1.

Analysis of the humoral immune response generated by plasmids encoding the E protein aligns with the lower protein production observed in vitro. Overall, immunization with the original pE1D2 plasmid elicited slightly higher levels of E-specific and neutralizing antibodies compared to immunization with the pEotmD2, though these differences were not statistical. In contrast, alteration in the E protein codons had an inverse effect on the cellular immune response, in which the pEotmD2 was slightly more effective in stimulating the IFN-γ production. The protection evaluation against a neuroadapted DENV2 showed that immunization with the pE1D2 was statistically more effective when compared to the pEotmD2. Indeed, the pE1D2 exhibited robust protection, consistent with earlier observations [44,60,61]. In contrast, protection conferred by the plasmid harboring the optimized gene, pEotmD2, was relatively modest and did not exhibit statistical differentiation from the control group. Thus, although no major difference was observed in the envelope expression level nor in the humoral immune response, the codon changes of the E protein substantially affected the protective capacity of the DNA vaccine.

Concerning the plasmids encoding the NS1 protein, administration of the pNS1otmD2 induced marginally higher antibody levels than the pcTPANS1-immunized group. Similarly, the pNS1otmD2 provided slightly superior levels of protection compared to the pcTPANS1, yet without significant statistical differences. The analysis of the humoral immune response and protection correlates with the findings of in vitro experiments, in which the pNS1otmD2 plasmid demonstrated greater efficiency compared to the pcTPANS1. Considering that optimization efficiency can be influenced by the host system and given that the enhancement in the in vitro NS1 expression mediated by pNS1otmD2 was higher in human than in murine cells, there is a possibility that this codon-optimized vaccine could induce a more robust protective response in primate models or in humans.

Overall, codon optimization is a useful strategy for the designing of DNA vaccines, but it does not guarantee an increase in recombinant protein expression or in the protective immunity induced by the vaccine [62]. In the present investigation, although the codon optimization of E and NS1 sequences for human and mouse cells did lead to an increase in the CAI and an adjustment in GC content, this process had divergent effects on the expression of these two proteins and on the immunogenicity and protective capabilities of the two vaccines. Surprisingly, the envelope sequence optimization yielded an opposite effect to what was expected, with a decrease in its performance regarding the expression of the recombinant protein and the induced immunity. Although it is not possible to establish a direct correlation, the reduced efficacy of pEotmD2 in the in vivo experiments probably reflects the lower efficiency of this plasmid in mediating the envelope protein in vivo expression. Interestingly, the pEotmD2 plasmid elicited slightly higher levels of cellular response than the pE1D2 plasmid.

Other research groups also observed a reduction in the recombinant protein production by employing a codon optimization strategy. The sequence modification of the gene that codes a human potassium channel (hERG) by employing synonymous codon substitutions to adjust GC content, reduced the incidence of rare codons and minimize mRNA secondary structure, resulting in protein expression decreased in HEK293T cells [63,64]. Agashe et al. [65] observed that creation of synonymous variants for a critical gene coding an enzyme in Methylobacterium extorquens resulted in substantially reduced gene expression and enzymatic activity compared to the wild-type version. Notably, using solely rare codons in the gene led to a 40% decrease in fitness, while a sequence entirely comprised of frequent codons caused a more dramatic reduction in fitness, exceeding 90%.

Synonymous modifications in the coding region can influence both mRNA stability and protein properties, including conformation, stability, post-translational modification sites, and even its functionality [51,62,64]. One possible explanation for the reduced efficiency of pEotmD2 in comparison to pE1D2 could be attributed to codon substitutions influencing protein folding. Thus, even if the amino acid sequence remains unchanged, variations in the translation speed may result in the generation of distinct protein structures. Such structural variances could influence E epitopes and, as a result, impact the quality and protective potential of the immune response elicited against this protein. Alterations in the E protein folding may also have increased its transportation for processing and subsequent presentation to T cells. Several reports have demonstrated the influence of synonymous codon substitution on the folding of the recombinant protein [66,67,68,69,70]. Hu and colleagues [69] studied the effects of synonymous codons on the properties of the anti-human IgE antibody produced in Escherichia coli. Authors generated 342 codon variants and observed that expression of the recombinant protein in bacteria resulted in antibodies that differed significantly in solubility and functionality, with antigen-binding activity ranging from no binding to a high binding affinity. The circular dichroism (CD) analysis of enhanced-affinity variants revealed noticeable structural and conformational alterations. In another study focused on the fluorescent protein YKB (yellow-black-blue), authors examined whether altering synonymous codons could result in variations in its conformation and, consequently, its fluorescent properties. The findings of this investigation underscored that even minor changes in codon usage can result in significant alterations in protein folding [68]. In a report on Neurospora cell-free translation system, it was noted that, despite a codon-optimized luciferase mRNA completing translation elongation more quickly than the wild-type mRNA and yielding a greater quantity of protein, a substantial portion of the protein turned out to be non-functional. These results indicated that codon usage affects translation elongation speed, consequently impacting on protein functionality by regulating co-translational folding [70].

The impact of codon optimization is subject to variation depending on the specific antigen, pathogen, and host system chosen. In most studies, the increased expression of recombinant proteins mediated by codon-optimized DNA vaccines correlates with enhanced immune responses and superior protection when compared to vaccines containing native genes [54,55,56,57,58,59]. Nevertheless, some reports have shown that the increase in expression mediated by plasmids containing codon-optimized genes does not consistently result in enhanced immunogenicity or increased protection [71,72,73]. In the evaluation of a DNA vaccine based on FimH, a protein involved in the adhesion of uropathogenic Escherichia coli (UPEC) to epithelial cells, authors observed that, although the DNA vaccine containing mammalian codon-optimized fimH gene exhibited higher expression levels compared to the DNA vaccine with native gene, both vaccines were equally effective in conferring protection in mice, with no significant difference between them [73]. In another study, the impact of mammalian codon optimization on the immunogenicity and protective potential of DNA vaccines encoding antigens from the pre-erythrocytic stage of Plasmodium falciparum and Plasmodium yoelii was assessed in mice. While codon optimization notably enhanced in vitro antigen expression, it did not lead to an improvement in antigen-specific T-cell responses or protection. [72]. Döşkaya and colleagues [71], in their analysis of a codon-optimized DNA vaccine against T. gondii, observed that the immune response and protective efficacy of the codon-optimized GRA1 DNA vaccine were slightly superior to those of the wild-type GRA1 DNA vaccine. These results resemble those observed in in vivo assessments of the pNS1otmD2 plasmid, which yielded slightly greater protection levels than the pcTPANS1.

Our study revealed contrasting results for codon optimization of the two dengue DNA vaccines. While the vaccine encoding the envelope protein demonstrated reduced efficacy, the vaccine containing the ns1 gene exhibited slightly enhanced protection in the murine model. An important consideration in our study is that codon optimization was targeted to two systems: mouse cells and human cells. This adaptation may have affected the optimal protein expression pattern in each of these systems and, consequently, influenced the observed results.

The outcome of codon optimization may differ in each unique case, and accurately predicting the impact of synonymous codon usage on the properties of an antigen or the immunogenicity of a vaccine is a challenge. Hence, although codon optimization is a highly valuable tool in the development of DNA vaccines, its application demands thorough evaluation.