Abstract

Introduction: Rift Valley fever (RVF) represents a significant public health challenge and economic burden due to its impact on livestock and potential to affect humans. Despite attempts to develop vaccines against the Rift Valley fever virus (RVFV), existing options are limited by concerns regarding the inability to differentiate between vaccinated and infected animals, vaccine-associated viremia, and the need for booster doses. This underscores the urgent need for a novel, effective, and safe vaccine, especially for use in ruminants, which this study seeks to address.

Methods: Employing reverse vaccinology—a cutting-edge approach combining bioinformatics and reverse pharmacology—we aimed to develop a novel RVFV vaccine. We focused on the M-glycoprotein segment, identifying highly conserved and immunodominant epitopes in viral glycoprotein sequences from cattle, sheep, and goats in RVF-endemic regions of Africa. Predictions for B- and T-cell epitopes were made, followed by the design of an epitope-based vaccine incorporating ideal linkers and a Bos taurus-specific beta-defensin to enhance immunogenicity. The vaccine's secondary and tertiary structures were analyzed using SOPMA and AlphaFold2, respectively.

Results: The vaccine candidate demonstrated promising physicochemical properties, with the M-glycoprotein sequences showing high antigenicity. Structural analysis revealed a composition of 31.55% alpha helices, 44.92% random coils, 5.35% beta turns, and 18.18% extended strands. Molecular docking with Toll-like receptors 7 and 8 indicated favorable molecular binding interactions, suggesting potential efficacy in stimulating an immune response.

Conclusion: This study paves the way for the development of a novel, safe RVFV vaccine. While the results are promising, further translational research is necessary to confirm the vaccine's effectiveness in animals and its applicability for improving public health outcomes.

Introduction

Rift Valley fever (RVF) is a viral disease most commonly seen in domesticated animals. People can develop RVF through contact with the blood, body fluids, or tissues of infected animals or through bites from infected mosquitoes. Rift Valley fever virus (RVFV) can cause severe disease in newborn ruminants such as sheep, goats, camels, and cattle, with up to a 100% fatality rate. Mosquitoes are the main infection vectors, although sandflies have also shown the capability to transmit the virus1. However, adults seem less susceptible. In ruminants, the virus causes RVF, an acute hemorrhagic fever, accompanied by abortion storms in pregnant females2. Abortion rates may be up to 100% in pregnant animals3. These dire statistics indicate the need for concerted efforts to develop preventive measures against RVFV’s future resurgence.

The RVFV genome is tri-segmented, with large (L), medium (M), and small (S) segments. The L segment encodes the RNA polymerase (RNA-dependent L protein), the S segment encodes the nucleoprotein and the non-structural nucleoprotein (NSs), and the M segment encodes four proteins (two structural glycoproteins [Gn and Gc] and two non-structural glycoproteins [NSm and LGp])4.

RVFV is a vector-borne, zoonotic, negative-stranded, tri-segmented RNA virus that belongs to the genus Phlebovirus and family Bunyaviridae. Apart from affecting human and animal health, it has a negative socio-economic impact5, 6. RVFV has been reported to be endemic to the Arabian Peninsula and Africa1, 7, although it was first identified in Kenya’s Rift Valley in early 19305 (hence its name). Kenya is reported to have experienced the deadliest epizoootic of RVFV before a major outbreak in South Africa between 1950 and 1951, which led to about 500,000 abortions and 100,000 deaths among sheep; subsequent epizootics were observed in Zambia, Namibia, Mozambique, Sudan, Zimbabwe, and other East African countries2, 3. RVFV outbreaks have been associated with periods of greater than average rainfall, although this cannot easily be predicted, thereby suggesting an interplay between the environment and human and animal health3, 8.

Conventional vaccines used extensively for controlling RVF include inactivated, live attenuated vaccines DNA vaccines, viral vector-based vaccines, viral replicon vaccines, and subunit vaccines, among others9. However, these vaccines lack important properties, such as the ability to differentiate infected from vaccinated animals. Live attenuated vaccines induce long-lasting immunity and are generally inexpensive to produce but are linked to an increase in the rate of abortion in pregnant animals and can have negative effects on internal organs, especially the liver, when the attenuated virus spreads inside hepatic cells, as in a naturally occurring RVFV infection9. Other vaccines also face certain difficulties, such as the inability to distinguish between infected and vaccinated animals, viremia, and the requirement for booster shots.

Multi-epitope subunit vaccines incorporate predicted immunodominant epitope peptides from various immunogenic genes that have been identified through proteomic research10. Short immunogenic peptide fragments employed in multi-epitope vaccinations produce potent immune responses while considerably reducing the risk of allergic reactions in the host. Identifying the immunogenic epitopes obtained from viral glycoprotein or nucleocapsid sequences has greatly benefited the in silico design of peptide vaccines11. These epitopes are analyzed and approved using computational techniques based on their physicochemical characteristics, immunogenicity, and non-allergenicity. Subunit vaccines are cost-effective, non-toxic, and safer for animals than other RVFV vaccines. The aforementioned limitations of conventional vaccines and the untapped potential of multi-peptide-based vaccines have made this study necessary.

In this research, we use a computational design approach to develop an RVFV multi-epitope subunit vaccine, which incorporates several antigenic determinants that are conserved across different RVFV strains. The vaccine design is optimized for increased stability, immunogenicity, and manufacturability, using various adjuvants and carrier proteins and evaluated in different delivery systems. Overall, this manuscript describes a comprehensive and innovative approach to the development of an RVFV multi-epitope subunit vaccine, highlighting the potential of this approach for the development of potent vaccines against other viral diseases. Therefore, our objective in this study was to design a safe, non-allergenic, and non-toxic subunit vaccine that can elicit optimal immunity against RVF in ruminants.

Methods Retrieval of RVF Protein Sequences

The National Center for Biotechnology Information’s (NCBI) immunoinformatics investigation yielded a total of 60 RVFV amino acid sequences, which included sequences from the L, M, and S genomic segments. These sequences came from nations like Kenya, Madagascar, South Africa, Rwanda, Namibia, Egypt, Zimbabwe, and Nigeria.

Antigenicity Prediction

The VaxiJen v2.0 server was used to determine the antigenicity of the RVF proteins11. Without using alignment, the VaxiJen server estimates a protein’s antigenicity based on its physicochemical characteristics. The default threshold value was set at 0.4.

Membrane Topology of Selected Proteins

The trans-membrane topology of the protein sequences was assessed for their intra-cytoplasmic, extra-cytoplasmic, and transmembrane portions using the TMHMM server v2.012. TMHMM predicts transmembrane topology through a hidden Markov model. Proteins with large extra-cytoplasmic regions were considered for further analyses.

Multiple Sequence Alignment

The glycoprotein M sequences underwent repeated sequence alignment using the MEGA 11 program to show conserved regions across the different sequences13. The MUSCLE algorithm was specifically used to align the different protein sequences14.

Prediction of Linear B-Cells

The ABCpred, SVMTriP, and BepiPred servers were used to predict the linear B-cell epitopes. For linear B-cell prediction, ABCpred employs a partly recurrent neural network (Jordan network) with a single hidden layer. The networks comprise 35 residues in a single hidden layer and a changeable optional window length15. SVMTriP is used to locate linear epitopes based on amino acid features such as secondary structure, flexibility, antigenicity, hydrophilicity, and solvent accessibility to predict B-cells16. Using a hidden Markov model, BepiPred forecasts B-cell epitopes from antigen sequences17.

Physicochemical Property Analysis

Analyzing the physicochemical characteristics of the chosen B-cells allowed for evaluating their suitability for vaccine development. ExPASy Protparam was used for this purpose18. Additionally, utilizing VaxiJen, AllerTop, and ToxinPred, these peptides were tested for their allergenicity, antigenicity, and toxicity, respectively11, 19, 20.

Prediction of Cytotoxic T Lymphocytes (CTLs)

The glycoprotein’s (M segment) CTL epitope was predicted using NetMHCII PAN 4.121. The server employs artificial neural networks (ANNs) to forecast peptide binding to MHC alleles. The network is taught to examine peptide ligands that have been eluted by mass spectrometry and have a high binding affinity (BA). The Bovine Leukocyte Antigen (BoLA) was chosen as the reference allele set in this study. The affinity of the glycoprotein peptides for the 20 BoLA alleles BoLA-JSP.1, BoLA-HD6, BoLA-T2c, BoLA-T2b, BoLA-T2a, BoLA-T7, BoLA-D18.4, BoLA-AW10, BoLA-T5, BoLA-1:00901, and BoLA-1:00902 was assessed. The chosen peptide has a length of nine amino acids.

Prediction of Helper T Lymphocytes (HTLs)

Strongly binding MHCII HTL epitopes were predicted using NetMHCII PAN 4.021. Based on machine learning techniques trained on binding affinity or mass spectrometry (eluted ligands) techniques, this tool forecasts the likelihood that a peptide will present an antigen. NETMHCII PAN 4.0 was used to predict epitopes22. The HTL epitopes were chosen to have a default length of 15 amino acids, and the binding affinity of overlapping glycoprotein peptides to the following alleles was discovered (the following DRB1 codes are available): DRB1_0101, DRB1_0102, DRB1_0103, DRB1_0104, DRB1_0105, DRB1_0106, DRB1_0107, DRB1_0108, DRB1_0109, DRB1_0110, DRB1_0111, DRB1_0112, DRB1_0113, DRB1_0115, and DRB1_0116.

Prediction of Interleukin-4- and Interleukin-10-Inducing Epitopes

HTLs, which produce cytokines like interleukins, are crucial in coordinating the immune response. Therefore, using IL4Pred and IL10Pred, the interleukin-4- and interleukin-10-producing capacity of a subset of strongly binding epitopes was evaluated23, 24.

Antigenicity, Toxicity, and Allergenicity Prediction

The toxicity, antigenicity, and allergenicity of strong CTL peptides and HTL epitopes that induce IL-4 and IL-10 were examined. The ToxinPred server, according to Gupta et al.19, predicts peptide toxicity along with its physical properties, such as molecular weight, charge, hydrophobicity, and amphipathicity. The AllerTop v2.0 server reportedly uses machine learning methods including auto- and cross-variance transformation and amino acid E-descriptors and predicts the allergenicity of proteins and peptides, according to Dimitrov et al.20. Doytchinova et al.11 claim that the categorization of antigens by VaxiJen v2.0 is solely based on the physicochemical characteristics of the proteins involved.

Epitope Conservation Determination

B-cells, HTLs, and CTLs were compared via multiple sequence alignment to determine whether they fall within conserved regions. For the final vaccine design, only completely conserved epitopes were considered. For the vaccine to be efficacious across a broad spectrum, highly conserved epitopes are required25. Furthermore, epitopes in the extra-cytoplasmic portion of the glycoprotein sequence were prioritized for selection.

Primary Construct Assembly

The chosen HTL, CTL, and B-cell epitopes served as the building blocks for the vaccine. To increase the final construct’s immunogenicity, a stiff EAAAK linker was used to introduce a Bos taurus-specific beta defense adjuvant to the N-terminus of the vaccine. AAY linkers were used to connect additional CTL epitopes. The HTL and B-cell epitopes were joined together using GPGPG linkers. According to Kavoosi et al.26, the glycine-rich linker GPGPG serves to increase solubility and promote free movement across adjacent domains. To boost the immunogenicity of the vaccine design using the EAAAK linker, a Bos taurus-specific beta defensin adjuvant was also inserted at the N-terminus27.

The Proposed Vaccine’s Allergenicity, Solubility, and Antigenicity

The allergenicity of the vaccine construct was assessed using the AllerTop v.2.0 server, while the antigenicity and solubility of the vaccine construct were assessed using VaxiJen v2.0 and Protein Sol, respectively11, 20, 28.

Analysis of the Vaccine Construct’s Physicochemical Characteristics Secondary and Tertiary Structure Prediction

The secondary structure was predicted using the SOPMA server. The server predicts protein secondary structure from amino acid sequences using the self-optimized prediction approach29.

The AlphaFold server was used to predict the tertiary structure of the vaccine construct30. AlphaFold uses innovative neural networks in combination with the primary amino acid sequence and alignment sequence of homologs to predict the tertiary structure of a particular protein.

Conformational Epitope Prediction from 3D Vaccines

Thornton’s approach and a residue clustering algorithm on the IEDB database are both used by the ElliPro server to predict conformational epitopes on the 3D structure. According to Ponomarenko et al.31, ElliPro offers a framework for predicting tertiary structure models from amino sequences.

Refinement and Validation of the Tertiary Structure

The quality of the 3D structure of the chosen model, protein structure refinement, protein interaction prediction, GPCR applications, and water position prediction were all improved using the GalaxyWEB refine server32. By recreating and repackaging protein side chains using dynamic simulation techniques, this server improves 3D protein architectures, facilitating the production of proteins with appropriate structural relaxation.

Thereafter, structural analysis using Ramachandran and Z-score analysis was used to validate the vaccine’s structure33, 34. PROCHECK and ProSA-web were used. The Ramachandran plot, which displays the proportion of the amino acids present in the preferred, permitted, and forbidden zones, is produced by the PROCHECK server. ProSA-web performs mathematical analysis on the entire protein structure quality score and presents values and energy.

Codon Optimization and In Silico Cloning

To adapt the vaccine for expression in the Escherichia coli strain K-12 sub-strain MG1655 chosen as the expression host, codon optimization was performed using JCAT35. This is crucial for maximizing protein expression during the recombinant vaccine’s cloning process. Using the SnapGene v5.1.7 program, the chimeric model was virtually cloned into an appropriate expression vector, E. coli K12 pET-28a (+)36.

Molecular Docking

A computational procedure called molecular docking is used to predict the affinities of the ligand and receptor to create stable complexes. Using the HDOCK server, the final vaccine design was docked against toll-like receptor (TLR) 7 and 837.

Table 1.

Antigenicity properties of selected proteins

Genome Segment Protein Antigenicity L Polymerase Non-antigen M Glycoprotein Antigen Nucleocapsid Non-antigen

Table 2.

Transmembrane topology of selected protein highlighted in bold

Genome Segment Protein Protein Segment Position L Polymerase 1-2091 Inside M Glycoprotein 1-16 Signal peptide 17-584 Outside 585-603 TMHelix 604-675 Inside 676-685 TMHelix 686-1159 Outside 1160-1178 TMHelix 1179-1197 Inside Nonstructural 1-265 Inside Nucleocapsid 1-245 Inside

Table 3.

Selected predicted linear B-cell-based epitopes

Peptide Antigenicity Allergen city