Exploitation of reverse vaccinology and immunoinformatics as promising platform for genome-wide screening of new effective vaccine candidates against Plasmodium falciparum

Data resources

The whole protein sequences (5548) of P. falciparum 3D7 genome were retrieved from PlasmoDB database (http://plasmodb.org) into the FASTA format, accessed on October 3, 2017 together with 5 RSKMA. Similarly, 151 non redundant EKE of P. falciparum 3D7 including 121 MHC class I and 30 MHC class II were retrieved from IEDB database (www.iedb.org) accessed on February 20, 2018 using MHC biding assay as search criteria (Additional file 1).

Methodology of the work

This investigation facilitates the screening of vaccine candidates from entire protein sequences (5548) of P. falciparum genome involved three major steps: i) prediction of probable antigens ii) prediction of T cell epitopes iii) screening of minimal epitope ensemble for maximum human population coverage. The methodology flow chart is given in Fig. 1.

Open image in new window

Prediction of antigens and physico-chemical characterization

Prediction of T cell epitopes

The subsequent step predicts the presence of immunogenic T cell epitopes in all the 5 GDPAA and 5 RSKMA by employing consensus technique of MHC class I and II tools accessible at IEDB web server. The prediction of T cell epitopes were performed for 51 HLA class I (cut off threshold IC₅₀ ≤ 500 nM) and 54 HLA class II alleles (cut off threshold ≤ 3 percentile rank) [35]. These T cell epitopes were further screened by VaxiJen2.0 tool that predicted the antigenic epitopes with threshold ≥ 0.5 [32].

Population coverage analysis of predicted T cell epitope ensemble

In the following step, PPC analysis of aforesaid combined (MHC class I and II) epitopes for all 5 GDPAA were conducted by utilizing IEDB based PPC tool against world-wide population. Furthermore, the minimal combined epitope set for each antigen was formed along with the inclusion of only those epitope which have highest PPC value and restricted by different set of MHC alleles. In case, PPC value of epitopes are equal then epitope with highest VaxiJen score was included in the minimal epitope set. Then, after employing the same protocol of minimal epitope selection, an ultimate epitope ensemble was designed with the joint screening of minimal epitope set of all the 5 GDPAA. Finally, the PPC of 5 GDPAA minimal epitope set and epitope ensemble were executed for the selected malaria endemic regions in the present study (India, South America, South-east Asia, Central Africa, East Africa, North Africa, South Africa and West Africa). So as to compare the predictive efficiency of PPC analysis for GDPAA epitope ensemble selection, the same protocol was also applied to predict epitope ensemble of 5 RSKMA and 151 EKE of P. falciparum [19, 36].

Prediction of immunogenic induction

The immunogenicity of the MHC class I epitope ensemble designed from GDPAA was estimated by CTLPred and TAPPred [34, 37, 38] while induction of IFN-γ and IL-10 by MHC class II epitope ensemble from GDPAA were predicted using tools IFNepitope and IL-10Pred, respectively [39, 40].

Prediction of epitope structure and docking studies

A structure-based docking approach was further carried out so as to improve the predictive capability of peptide-MHC binding. The combination of sequence and structure-based approaches not merely enhances the probability of MHC binding prediction but also calculates the docked epitope orientation. The complex crystal structures of HLA–A*02:01 (PDB ID: 1I4F) and HLA–DRB1*01:01 (PDB ID: 1AQD) were retrieved from the protein data bank (PDB) for MHC class I and class II, respectively. We used peptides NQMIFVSSI (C1) and LKELIKVGLPSFENL (C2) from EKE of P. falciparum as positive controls in docking studies. The structural information about the test peptides of GDPAA epitope ensemble (P1, P3 and P11) and positive control peptides were modelled using the PEP-FOLD 3 web server. The test peptides of GDPAA epitope ensemble were then docked by employing PatchDock web server to HLA–A*0201 and HLA–DRB1*0101 interacting residues as input. The best 10 HLA-peptide complexes were further refined by the FireDock web server. In order to compare the study of molecular docking, ClusPro tool was also used for the same test and control peptides along with target MHC molecule of interest [41]. The details of the docking web servers are given in the Table 1.

Prediction and characterization of antigens

The appropriate physico-chemical properties as well as stable structure of proteins are needed to trigger immune response [33, 49]. The negative Gravy values of 5 GDPAA clearly indicate their hydrophilic nature and good water solubility except P28 (Table 2). The clustering analysis of these antigens also revealed a close similarity between CSP and P113 as well as P28 and P25 inferred by MAFFT7.0 web server (Fig. 2). A vaccine involving a little number of alleles may be enough for coverage towards naturally-circulating strains, which also supports the concept of multi-allele methodology for designing polymorphic antigens as malaria vaccines [50, 51]. The study conducted by Soulama group revealed that AMA-1 antigen of P. falciparum 3D7 showed lower allelic diversity in Central Africa compared to West Africa [52]. Thus, the limited allelic diversity of antigens screened in the present study is suitable vaccine candidates for further immunological experiment.

Open image in new window

Prediction of antigenic T cell epitopes

Population coverage analysis and selection of minimal epitope ensemble

The immunogens based on population coverage analysis enable us towards the screening of best possible minimal epitope combination, i.e., epitope ensemble so as to maximize vaccine effectiveness. In this background, the PPC strategy was followed for developing the epitope ensemble of chosen antigens towards improving the vaccine efficacy in the world-wide as well as endemic population of malaria. Primarily, the PPC of predicted T cell epitopes (MHC class I, II and combined) from 5 GDPAA with VaxiJen score ≥ 0.5 were performed for the world-wide, South East Asia, India, South-America and Central Africa, East Africa, North Africa and West Africa populations (Additional file 4). The minimum PPC value of combined epitopes (94.84%) for P25 is reported for South-America population, whereas the maximum PPC (99.99%) is forecasted for MSP1 antigen for world-wide population (Fig. 3) [36]. However, the PPC values of EKE of P. falciparum for 121 MHC class I, 30 MHC class II and 151 combined epitopes were found 96.97, 65.35 and 99.25%, correspondingly. In order to minimize the epitopes in the vaccine construct design, we have restricted epitopes number on the basis of considering minimal epitopes binding with maximum number of different MHC alleles for the entire 5 GDPAA (Additional file 5). This resulted into 5 minimal set of epitopes covering populations world-wide, South East Asia, India, South America, Central Africa, East Africa, North Africa and West Africa (Additional file 6). The PPC value (86.18%) of combined minimal epitopes is minimum for CSP antigen for South-East Asian population while the maximum coverage (99.95%) for MSP1 antigen is reported for world-wide population. Further, employing the above protocol of minimal epitope selection, an epitope ensemble was designed for each of the combined minimal epitopes of 5 GDPAA and 5 RSKMA as well as 151 EKE. Here, as detailed in Table 4, the PPC value of GDPAA epitope ensemble (99.97%) (Additional file 6) is significantly higher over the epitope ensemble of RSKMA (99.90%) (Additional file 3) and EKE of P. falciparum epitope ensemble (97.72%) for the world-wide population (Table 4). Also, it is interesting to note that the PPC value of predicted MHC class II epitope ensemble of GDPAA (93.71%) is much higher than the MHC class II epitope ensemble of EKE of P. falciparum (75.31%). However, an exception was recorded for South African population (0.43%) that could be due to limited HLA typing data as detailed in Table 5.

Open image in new window

Table 4

The world-wide PPC coverage of epitope ensemble predicted from all 5 genome derived predicted antigenic adhesins (GDPAA) and 5 randomly selected known malarial adhesins (RSKMA) as well as 151 experimentally known epitopes (EKE) of P. falciparum

S.No.	Epitope ensemble	PPC for world-wide population (%)
		MHC class I	MHC class II	Combined
1.	GDPAA MHC class I: FAMSNALLV, ILMLILYSF, YTLTAGVCV, FSSSNNSVY, YEMKFNNNF, LIVCSIFIK, YFNDDIKQF, LLKSYKYIK, SRLKKRKYF, KGMSSSQEM MHC class II: IPFFILHILLLQFLL, FIQLYITLNKARVTE, TCGNGIQVRIKPGSA	99.53	93.71	99.97
2.	RSKMA MHC class I: YFFASFFVL, FTYDSEEYY, FAFPPTEPL, FMPPRRQHF, FRDEWWKVI, RIYDKNLLM, KLYFPTPAL, YAFSEECPY, ISFQNYTYL MHC class II: DKMKIIIASSAAVAV, YKYAASFTLAAILFL	98.67	92.31	99.90
3.	EKE of P. falciparum MHC class I: RPRGDNFAV, TPYAGEPAPF, YLINKHWQR, NQMIFVSSI, KVSDEIWNY MHC class II: DAEVAGTQYRLPSGKCPVFG, LKELIKVGLPSFENL, ALLIIPPKIHISIEL	90.78	75.31	97.72

Table 5

The PPC value of epitope ensemble vaccine candidate of 5 genome derived predicted antigenic adhesins (GDPAA) and 5 randomly selected known malarial adhesins (RSKMA) as well as 151 experimentally known epitopes (EKE) of P. falciparum for malaria endemic regions

Population	GDPAA (%)	RSKMA (%)	EKE of P. falciparum (%)
North Africa	99.93	99.84	95.34
India	99.54	99.17	93.01
East Africa	99.68	99.41	92.17
Southeast Asia	99.39	99.07	81.71
South America	98.94	97.96	87.66
Central Africa	99.31	98.78	90.47
West Africa	99.84	99.19	92.88
South Africa	0.43	0.43	0.40

The GDPAA epitope ensemble as a vaccine candidate includes 10 MHC class I epitopes (P1:FAMSNALLV, P2:ILMLILYSF, P3:YTLTAGVCV, P4:FSSSNNSVY, P5:YEMKFNNNF, P6:LIVCSIFIK, P7:YFNDDIKQF, P8:LLKSYKYIK, P9:SRLKKRKYF, P10:KGMSSSQEM) and 3 MHC class II epitopes (P11:IPFFILHILLLQFLL, P12:FIQLYITLNKARVTE, P13:TCGNGIQVRIKPGSA) as presented in Table 4 (Additional file 7). Out of 10 MHC class I epitopes, 4 epitopes (P1, P2, P5 and P9) were confirmed as cytotoxic T-lymphocyte (CTL) epitopes by using CTLPred, while 7 epitopes (P1, P2, P3, P5, P7, P8, and P9) were confirmed as TAP binders by TAPPred. Also, the epitope P11 was found to be inducer of both IFN-γ and IL-10 cytokines predicted by IFN-epitope and IL-10Pred, respectively while epitope P13 induced only IFN-γ. The GDPAA epitope ensemble showed 99.54% of PPC value for Indian population (Table 5). In the West African population, HLA–B*5301 and HLA–DRB1*1302 alleles are linked with the reduction in life-threatening malaria [56]. In our study, the GDPAA epitope ensemble P1: FAMSNALLV binds to the HLA–B*5301 and P12: FIQLYITLNKARVTE also binds to the HLA–DRB1*1302. Moreover, the epitope P4: FSSSNNSVY and P8: LLKSYKYIK of GDPAA ensemble binds to HLA–A*30:02 and HLA–A*30:01, correspondingly, which are also linked with cerebral malaria [57]. In addition, the MHC class II epitopes P12: FIQLYITLNKARVTE and P11: IPFFILHILLLQFLL binds to the HLA–DRB1*04:01, which is accompanying with severe malaria in Northern Ghana [58]. There are 5 most frequently HLA alleles (HLA–B*15:03, HLA–B*42:01, HLA–B*53:01, HLA–B*58:02 and HLA–B*57:03) found in African regions [59]. The peptide P10: KGMSSSQEM found in the GDPAA epitope ensemble also binds to HLA–B*5801, while peptide P1: FAMSNALLV binds to HLA–B*5301.

The interaction between the peptide and MHC is an important characteristic of T cell epitope and the accessibility of the crystal structures facilitates us to determine the mode of interaction by computational methodology [60]. The docking data obtained from the freely available online tools PatchDock and ClusPro depicted that the anchor residues of majority of the epitope ensemble associate with the active site residues of HLA–A*0201 and HLA–DRB1*0101 that specify the superior mode of binding of screened epitopes with the T cell receptor. For HLA–A*0201, the geometric shape complementarities scores for test peptides P1 and P3 were 7284 and 7586, respectively, which were very close to the score of the positive control peptide C1 (7604) and global energy for test peptides P1and P3 were − 11.93 and − 15.72, correspondingly that were less than to the positive control peptide C1 (− 7.99). Similarly for HLA–DRB1*0101, the geometric shape complementarity scores for P11 was 11,870, which was very close to the score of the positive control C2 peptide (10806) and global energy for P11 was − 61.82, which was less than the positive control C2 peptide (− 55.11). The ClusPro docking tool was also used to predict the HLA–epitope complex molecular model and calculate the clustering score. The test epitopes (P1, P3 and P11), target MHC proteins and control peptides were same as selected in the PatchDock tool to compare the docking results. For HLA–A*0201, the lowest clustering scores for P1 and P3 peptides were − 694.4 and − 758.3, respectively, which were lower to the score of the positive control C1 peptide (− 647.1). Similarly, for HLA-DRB1*0101, the lowest clustering score for P11 peptides was − 1159.8 which was very much lower than the score of the positive control C2 peptide (− 715.7). The docking models for HLA–A*0201-C1, HLA–A*0201-P1, HLA–A*0201-P3, HLA–DRB1*0101-C2 and HLA–DRB1*0101-P11 are shown in the Fig. 4.

Open image in new window

The reverse vaccinology and immunoinformatics strategies are still under progressive phase, however reasonable triumphs may result in substantial advancement on epitope-based ensemble vaccine efficacy against malaria pathogens for example, by enhancing coverage in the target populations through reasonably bearing in mind the specificity as well as occurrence of the HLA molecules [61, 62]. Therefore, epitope ensemble provided in the present study provides the basis for effective malaria vaccine design. A well-known drawback of epitope ensemble vaccine is poor immunogenicity, usually necessitating the use of suitable adjuvants [63]. Therefore, these predicted and experimentally validated epitopes ensemble could be tested for further studies like nanovaccine formulation and evaluation in the experimental animal model for actual efficacy of nano sized malaria vaccine. Numerous investigations found to depict that malarial antigens showed more immunogenicity and superior correlated with protection when presented on nanoparticles based carrier systems like self-assembling protein nanoparticle (SAPN). The SAPN depend on coiled-coil domains of proteins to form stable nanoparticles [64]. Recently, Burkhard and Lanar developed protein based nano vaccines, which provide robust immunity against malaria [65]. This SAPN contains HLA-supertypes-restricted CD8⁺ T cell epitopes (separated by N/KAAA spacers and optimized for proteasomal cleavage) from antigens expressed during malaria pathogen life-cycle, the universal CD4⁺ T cell epitope, and flagellin as a scaffold and TLR5 agonist. On the other hand, used de novo designed amino acid domains to fuel the development of the coiled-coil scaffolds that present the antigenic epitopes on the nanoparticles surface [66]. As the surface area of the nanoparticles increases with the reduction of particles size, therefore there is a great need to develop more in silico strategy for effective nanovaccine designs that fulfil the vaccine requirement of needy human being [67].