• Tidak ada hasil yang ditemukan

SEQUENTIAL APPROACH IN SELECTING ANTIGEN

Dalam dokumen Innovations for Healthier Aquaculture (Halaman 88-93)

A. P. Desbois

5.2 SEQUENTIAL APPROACH IN SELECTING ANTIGEN

5.2.1 reVerse VaccInologyto IdentIfy antIgenstHrougH sequence sImIlarIty

5.2.1.1 Conventional Reverse Vaccinology

The first application of bioinformatics could be dated back to the early 2000s, using RV to discover epitopes for MenB [4]. The idea of RV, which is defined as “the process of antigen discovery through the interrogation of an organism’s complete antigenic repertoire, as coded in its genomic data,” was spurred by the advancement in genomic sequencing in the late 90s. Currently, the availability of genomics and proteomics data, as well as the advancement in bioinformatics tools, has enabled high- throughput screening for vaccine candidates, which could save a considerable amount of efforts and expense for wet-lab screening steps. Overall, the strategy for RV would be different depending on the resources and target. However, a potential antigen would usually satisfy the following filters: suitable cellular localization, sequential conservation across strains, chemical properties compatible with the host’s immune system, and the process of mass-producing antigens for vaccination [5].

The first step of RV starts with identifying the potential sequences as antigens. A satisfied sequence should be an ORF that is preferably conserved so that they could represent the whole strain and have a higher protective coverage against the said pathogen. Constructing phylogenetic trees is a common choice to cluster the diverse and enormous genomes of different strains based on their genotype, thus easing the comparative steps. There are several ways to build a genotypic tree, depending on how broad the vaccine’s aim is. Multi-locus sequence typing (MLST) is the method of choice for many vaccine development projects due to its sensitivity in detecting different strains based on housekeeping genes’ allelic profiling [4, 5]. This step could either be a confirmation step for knowledge-based or experimental antigen selection or preparation for the subsequent in silico selections. While the former was already shown with the case of MenB vaccine, the latter pipeline has emerged in this decade with the help of different bioinformatic tools.

FIGURE 5.1 Schematic diagram of a combined antigen screening strategy.

68 Fish Vaccines

5.2.1.2 The Pan-Genome and Comparative Genome Analysis

Based on the premise of RV, modifications have been made to enhance the accuracy in choos- ing antigenic proteins of this approach [6]. One common derivative, pan-genome RV, utilizes the availability of multiple genomes of the same strains. A pan-genome is the full gene library of dif- ferent bacterial genomes of the same species, which comprises a core genome (common genes of all strains in the species), a dispensable genome of genes that are present in some strains only, and unique genes from each of the strain’s genome. Based on alignment and clustering ORF of these strains, the pan-, core-, dispensable, and unique genes could be divided into groups, thus giving us a clearer picture of conserved antigens that could offer broadly protective vaccines [7]. There are several pipelines available for this method, which include Panseq [8], PGAP (pan-genomes analysis pipeline) [9], and PanRV (pan-genome-reverse vaccinology) [10].

With the aim of creating a vaccine that could target pathogenic strains of the bacteria, the com- parative genome approach focuses on genes that are required for commensal strains to become virulent, which could either be original genes that are mutated or genes acquired through horizontal gene transfer. The result of this is vaccines that only target virulent strains and spare nonpathogenic bacteria [11].

5.2.2 IdentIfy antIgenstHrougH suBcellular localIzatIon

A large portion of immunogenic antigens come from the outer-membrane. Commonly, protective B-cell protein antigens are located in the outer-membrane protein (OMP) and extracellular environ- ment (secreted proteins); hence, these predicted subcellular localizations were targets for selection [12]. Also, in some cases, OMP could interact with immune cells and trigger a protective immune response (dendritic cell, T- and B-cells) through the extrusion of outer-membrane vesicles (OMV) produced by both Gram-positive and Gram-negative bacteria during growth, which further justifies the use of OMP for vaccine development [13, 14].

There have been numerous in silico methods to detect the expression localization of antigens.

One of the most used algorithms is PSORT (Protein Subcellular Localization Prediction Tool), which is a knowledge-based method capable of multi-category sorting. The method made use of different subprograms analyzing signal peptides, transmembrane structure, amino acid composition and structure, and so on [15]. Following this lead, other localization tools have also emerged, which are listed in Table 5.2.

Despite having high immunogenicity and intense attention from researchers, OMP vaccine still has its own obstacles to overcome. The fact that OMPs would usually contain at least one transmem- brane hydrophobic domain means that the recombinant expression of these proteins will face a high risk of failure [4]. Furthermore, the variability in the surface structure of OMP among strains in

TABLE 5.2

Available Subcellular Localization Prediction Tools

Prediction Tools Main Features Site

PSORTb Prediction of cellular sub-location of whole protein

https://www.psort.org/psortb/

CELLO2GO http://cello.life.nctu.edu.tw/cello2go/

SignalP Prediction of signal peptide’s site on a protein

https://services.healthtech.dtu.dk/service.php?SignalP-5.0 TMHMM Prediction of transmembrane topology

in a protein

https://services.healthtech.dtu.dk/service.php?TMHMM-2.0 DeepTMHMM Prediction of transmembrane topology

and signal peptide

https://services.healthtech.dtu.dk/service.php?DeepTMHMM.

Phobius https://phobius.sbc.su.se/cgi-bin/predict.pl

Antigen Discovery 69

some species, which interact directly with B-cells, would be a challenge when choosing a conserved antigen for vaccine design. The appearance of hydrophobic structures in a membrane protein could lead to two scenarios: the protein was secreted to be buried in the periplasm or the outer-membrane, or the protein had a transmembrane domain. In the former case, the protein would not likely be immunogenic and will be discarded from the potential pool with the help of structural prediction tools (e.g., TMHMM for transmembrane helices prediction and PRED-TMBB for beta-barrel pro- tein) [12, 16]. On the other hand, proteins with transmembrane domains could be trimmed off and expressed only in the outer-membrane domains, thus easing the high-throughput vaccine produc- tion steps. It should be noted that with the current prevalence of artificial neural network (ANN) algorithm, an enhanced transmembrane prediction tool from the previous hidden Markov model TMHMM was developed called DeepTMHMM, which now based upon an encoder-decoder deep learning model. Even though the tool still needs time to prove its effectiveness, it has shown supe- rior performance to most traditional localization prediction tools [17].

5.3 IMMUNOINFORMATIC ASSISTANCE IN ANTIGEN DISCOVERY 5.3.1 predIctIonof t-celland B-cell epItopes

While antigenicity could be assessed in the full antigen forms, sometimes the whole protein could not be easily produced due to the hydrophobic nature of some domains, like the transmembrane/

cytosolic domains from OMPs. Thus, there is a need to “trim down” the peptide residues that could be expressed without losing the antigenic properties, and with this comes the demand for epitope prediction and mapping. Moreover, the determination of peptide fragments capable of being rec- ognized by the immune system is one of the most crucial requirements of a vaccine candidate [18].

Epitopes are regions on the surface of the antigen that interact with B- and T-cells, the main cells that orchestrate the adaptive immune response. B-cell receptors could bind to the surface of a whole antigen, which is not limited to the continuous and discontinuous protein surface, polysaccharides, nucleic acids, and other organic molecules. T-cells, in contrast, are specifically bound to small pep- tide epitopes represented on the major histocompatibility complex (MHC). There are two classes of MHC corresponding to two T-cell-mediated pathways. MHC-I representing peptides transported by the transporter associated with antigen processing (TAP) to the endoplasmic reticulum (ER). These cytosolic peptides will trigger the activity of CD8+ T-cells (cell-mediated immune system), which would trigger infected cell death. On the other hand, CD+ T-helper triggered cells by the MHC-II class with the humoral immune pathway, which would promote the secretion of neutralizing anti- bodies from plasma cells [19].

Immunoinformatics is a subfield of bioinformatics that deals with algorithms for modeling the immune systems, subsequently enabling the mapping of B-cell and T-cell epitopes. From the basis of B-cell and T-cell recognition, there were four possible classes of epitopes that should be con- sidered: MHC-I epitopes, MHC-II epitopes, B-cell continuous epitopes, and B-cell discontinuous epitopes. The prediction tools for each of the epitope classes are listed in Table 5.3.

It should be noted that epitopes should also be evaluated for other qualifications apart from potential binding affinities with immune receptors, namely, Kolaskar–Tongaonkar antigenicity, Emini surface accessibility prediction, Chou–Fasman beta-turn prediction, prediction of floppy- prediction, and Parker hydrophilicity prediction [20].

5.3.2 molecular dockIng

5.3.2.1 Obtaining Molecular Structures

To perform molecular docking experiments, receptor and ligand structures must be available. The Protein Data Bank (PDB) is the structural data database of biological macromolecules solved by X-ray crystallography, nuclear magnetic resonance (NMR) spectroscopy, cryo-electrical microscopy, and

70 Fish Vaccines

theoretical modeling. However, only a fraction of discovered proteins got their structures predicted with satisfying resolutions for docking. Thus, efforts have been made in the field of computational structural predictions, introducing several highly accurate and easy-to-operate prediction tools.

Homology modeling has been the best strategy for protein tertiary structure predictions, which was implemented in both online tools (e.g., SWISS-MODEL) and computer programs (e.g., Modeller and Deepview). However, the approach is limited by the requirement of protein template structures, which could affect the accuracy of the predicted structures. Fortunately, the rapid development of ANN algorithms has enabled an ANN-based model, Alphafold, to solve the dilemma that has existed for decades. The model is continued to be optimized and be easily accessed online through Google Colaboratory.

5.3.2.2 Molecular Docking Tools

Molecular docking was originally the most preferred structured-based method to design drug molecules for pharmaceutical research. The method simulates the receptor–ligand complex by predicting how different conformations of a molecule could bind to a “pocket” in the target pro- tein [21]. Its applications to epitope mapping include predicting possible interaction and binding affinity between the epitopes and the cavities of antibodies, B- and T-cell receptors, and MHC molecules. Available tools for molecule docking, namely, Autodock Vina, ZDock, autodock 4, Patch dock, Molegro Virtual Docker, Mti auto dock, Cluspro 2.0, and PyRx were all applied in previous vaccine discovery studies. In addition, supplementary tools might be useful for analyz- ing molecular interactions (Schrodinger Academic Desmond MDS suite and/or PyMol and Jmol visualizing systems) [22, 23].

TABLE 5.3

Antigen Mapping Software Classified by Epitope Types

Epitope Class In Silico Tools

MHC-I epitope binding prediction SVMHC

TepiTope ProPred-I NetMHCpan CTLpred RANKPEP NetCTL

MHC-II epitope binding prediction SVMHC

TepiTope ProPred-I NetMHCpan CTLpred RANKPEP nHLAPred MHCpred Discontinuous B-cell epitope prediction Discotope

ElliPro CBtope Continuous B-cell epitope prediction BCpred ABCpred Bepipred IEDB Ellipro

Antigen Discovery 71

5.4 ANTIGEN VALIDATION BY EXPERIMENTAL APPROACHES 5.4.1 a classIc pIpelIneof antIgen dIscoVery ValIdatIon

The traditional approach to discover a potential antigen would be to fractionate the pathogen and examine the fractions for immunoreactivity or be characterized by chromatography methods (LC- MS/MS) [24, 25]. With the current development of bioinformatics tools, the process could be semi- computational with the use of subcellular localization prediction tools, as well as other protein signal/protein secretion prediction tools, thus saving time and money [24]. However, it should be noted that the predictions of computational tools are not always aligned with practical applications, which could be attributed to the lack of data for training prediction tools. Hence, most often, the

“candidate antigens” must go through other rounds of validation by different experimental steps.

In the classic example of vaccine screening for N. meningitidis, 570 potential antigen genes were cloned to E. coli for recombinant production of antigens, of which 350 were expressed successfully.

After mouse immunization, FACS (fluorescence-activated cell sorting) analysis revealed that only 26% of the total injected vaccine showed the ability to induce humoral responses. Furthermore, by assessment of antibody titers for neutralizing antibodies against the target pathogens, only one-third of the candidates remained for the development of MenB vaccine [4]. The process has highlighted the importance of a rigorous screening for candidate antigens after in silico prediction steps.

5.4.2 dIfferent metHodsfor antIgenIc eValuatIonof predIcted antIgens

5.4.2.1 Antibody Analysis

Generally, since the above approaches are mainly for subunit vaccines, the candidate protein must first be expressed in recombinant hosts. This also becomes the technical requirement for the anti- gens, as they should be preferably soluble when being produced in the host. Then, like other types of vaccines, these antigens would be tested for their immunostimulatory effect in animal models. The antisera from these models would go through numerous assays to confirm the presence of targeted antibodies, as well as bactericidal activity in the sera.

Active immunization is the state in which the host body can produce specific antibodies against a pathogen after exposure to antigens of the said pathogen, and this is also the major target of vac- cines. Therefore, evaluating antibodies level against the vaccine is the first step in confirming the validity of chosen antigens. Enzyme-linked immunosorbent assay (ELISA) and/or Western Blot are standard methods applied to measure the quantity of antibody titer in the antisera and could even quantify the immunoreactivity of the antibodies [26].

5.4.2.2 Lymphocytes Analysis

Because the first stage in an immune response generates different classes of effector lymphocytes, one could predict that the immune pathway corresponded to an invading agent through the number of lymphocytes presenting in lymph nodes and organs [27]. Using lymphocyte proliferation assays, the rate of adaptive lymphocytes (T- and B-cells) differentiated from mononuclear cells at a specific site like the spleen could be enumerated, and if this rate is elevated in the challenge experiments, then the vaccine is proved to be effective in triggering long-term protection from lymphocytes, and could even have the potential to kick-start the cell-mediated immune response [28]. Apart from the former in vitro method, there are also in situ methods to count lymphoid cells, applying immuno- fluorescence and immunohistochemistry techniques, namely, microscopy and flow cytometry [29].

5.4.2.3 Cytokine Profile

Besides the ability to trigger the humoral immune response, the vaccine should also be able to induce T-cell responses. While the role of T-cells in assisting vaccine protection is still unclear, their importance in maintaining the memory response after the initial immunization has been observed.

72 Fish Vaccines

T-cells are traditionally divided into cytotoxic CD8+ cells (CTL) and CD4+ T-helper (TH) cells.

While CTLs are effective in dealing with infected cells and tumor cells, Th cells are mainly in charge of modulating the “mode” of immune response to eliminate different pathogens efficiently.

Therefore, CD4+ cells are further classified into Type 1 helper cell (Th1 cell), Type 2 helper cell (Th2 cell), and IL-17 secreting helper cell (Th17 cell). Particularly, Th1 cells are key players in monitoring defense against intracellular pathogens, including protozoa, bacteria, and viruses [30], while Th2 cells are responsible for host response against parasite infections, venom, and allergens.

Finally, the third subtype Th17 cell will mount a defense against extracellular pathogens and fungi, ones that the Th1 and Th2 are not suited against.

Each type of T-helper cell can manifest a distinct pathway through the secretion of different cytokines set to deal, whereas the Th cells also proliferate under the regulation of cytokines and are guided by chemokines. Hence, cytokine and chemokine assays are also fundamental analyses of vaccine efficacy. Table 5.4 shows the main cytokines promoting Th1, Th2, and Th17 cells [30].

The assays could be carried out using high-performance liquid chromatography, bioassays, protein assays, ELISA, and other immunoassays [31].

Dalam dokumen Innovations for Healthier Aquaculture (Halaman 88-93)