A multi-stage in silico filtering approach for Leishmania major proteome for the identification of novel vaccine candidates

Abstract

Objective:

Leishmania major, the primary causative agent of cutaneous leishmaniasis, presents significant challenges for vaccine development due to antigenic diversity and the limitations of current therapeutic approaches. Therefore, this study aimed to use immunoinformatics computational approaches to identify novel vaccine candidates against leishmaniasis.

Materials and Methods:

We implemented a comprehensive multi-stage computational filtering strategy to systematically evaluate the complete L. major proteome (8,038 proteins) for membrane-associated vaccine candidates. Using integrated antigenicity prediction VaxiJen v2.0, subcellular localization analysis combining UniProt annotations and DeepLoc 2.0 predictions, dual allergenicity assessment using AllerTOP v2.0 and AllergenFP v1.0, and extensive conservation analysis were conducted across five Leishmania species (L. donovani, L. infantum, L. braziliensis, L. guyanensis, and L. mexicana) with human homology screening.

Results:

The screening achieved a 321-fold enrichment, identifying 25 nonallergenic membrane-associated proteins with high antigenic potential and optimal safety profiles. These candidates have not been previously investigated as vaccine targets, representing an expansion of the existing therapeutic repertoire. Conservation analysis revealed distinct strategic categories: seven proteins demonstrated >90% identity across multiple Leishmania species, suggesting broad-spectrum vaccine potential, while eight showed moderate conservation (>80%), indicating region-specific utility. Functional characterization revealed diverse membrane proteins, including novel adhesion molecules, transport proteins, and uncharacterized surface antigens. Critical safety assessment confirmed minimal human protein similarity (<35% identity), significantly reducing autoimmunity risks.

Conclusion:

The bioinformatics data support the substantial untapped potential within parasitic proteomes and validate systematic computational screening as a potential strategy for vaccine development. The conservation patterns identified provide clear strategic guidance for developing either pan-Leishmania vaccines for broad-spectrum protection or species-specific formulations for targeted regional use. However, additional in silico and wet lab synthesis of the vaccine has to be conducted to confirm the potential of the protein candidates.

INTRODUCTION

Leishmaniasis, a parasitic disease caused by the Leishmania genus, affects 0.7–1 million people annually in nearly 100 countries. The disease presents with a range of clinical manifestations, including visceral, mucocutaneous, and the most common, cutaneous leishmaniasis.[1,2] Despite its significant burden, leishmaniasis remains a neglected and understudied disease.[3,4] Leishmania major (L. major) is a primary cause of severe cutaneous leishmaniasis, with high prevalence in Middle Eastern countries.[5,6,7] Current treatments, such as amphotericin B, pentavalent antimonials, and miltefosine, are frequently compromised by toxicity, high cost, developing drug resistance, and poor efficacy.[8,9,10,11]

The limitations of current drugs necessitate a long-term solution, such as vaccination.[12] While various vaccine strategies have been explored,[13] traditional approaches have encountered difficulties regarding insufficient protection, complex manufacture, and safety issues.[14] Epitope-based vaccine design has emerged as a promising alternative, using immunoinformatics to identify specific antigenic determinants that can induce a tailored and safe immune response.[15] This in silico methodology has been successfully used to identify potential vaccine candidates in other protozoan parasites, such as secretory proteins in Leishmania donovani[16] and essential proteins in Toxoplasma gondii.[17] This computational technique markedly accelerates the identification and validation of new vaccine candidates,[18] a process proven effective for diseases such as gonorrhea[19] and schistosomiasis.[20]

For L. major, membrane-associated proteins are highly desirable vaccine targets due to their roles in cellular invasion and host-parasite interactions. However, current research has been limited to a small number of well-known antigens, such as GP63 and KMP-11, which restricts the discovery of novel candidates.[21,22] To address this gap, the current study employs a robust in silico pipeline to systematically screen the L. major proteome. By integrating antigenicity prediction, subcellular localization, allergenicity assessment, and cross-species conservancy analysis, we aim to expand the repertoire of potential vaccine targets by identifying novel, nonallergenic, highly antigenic, and conserved membrane proteins suitable for a next-generation, broad-spectrum vaccine against leishmaniasis.

MATERIALS AND METHODS

Proteome retrieval

Under the proteome ID UP000000542, the entire L. major proteome was downloaded in FASTA format from the UniProt database (https://www.uniprot.org/). The primary input for the forthcoming immunoinformatics analysis was this dataset.

Antigenicity prediction

The VaxiJen v2.0 server (https://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html), an alignment-independent prediction tool based on the physicochemical properties of proteins, was used to predict the antigenicity of the L. major proteome.[23] To make the analysis easier, the entire proteome was split up into several distinct FASTA files. Using the parasite model, with a default threshold value of 0.5, each file was submitted separately to VaxiJen v2.0. Only proteins with an antigenicity prediction score of 0.8 or greater were chosen for further analysis and included in the study.

Subcellular localization screening

Initially, the UniProt database annotation was used to determine the subcellular localization of antigenic proteins. For proteins lacking localization information in UniProt, the DeepLoc 2.0 server (https://services.healthtech.dtu.dk/services/DeepLoc-2.0/) was used to predict their subcellular localization.[24] For additional examination, proteins that were annotated or predicted to be in the plasma membrane or other membrane-associated compartments were chosen.

Allergenicity evaluation

Two separate servers, AllerTOP v2.1 (https://www.ddg-pharmfac.net/allertop_test/) and AllergenFP v1.1 (https://ddg-pharmfac.net/AllergenFP/), were employed to evaluate the probable allergenicity of the chosen membrane-associated antigenic proteins.[25,26] Only proteins that were determined to be nonallergenic by both AllerTOP and AllergenFP were kept for additional examination after being presented separately to both tools. These two-step methods were used to reduce false-negative predictions and improve the accuracy of allergenicity testing.

Conservancy and homolog analysis

To further investigate sequence conservation, the NCBI BLASTp tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi) was utilized to examine the antigenic and nonallergenic membrane proteins. To evaluate intra-species conservation, each chosen protein sequence was submitted to a separate BLASTp search against Leishmania (taxid: 5658) strains. To further reduce the possibility of autoimmune reactions, the sequences were further blasted against the proteome of humans (taxid: 9606) to make sure there was no apparent homology to human proteins.

RESULTS

Retrieval of Leishmania major proteome

The complete proteome of L. major, comprising 8,038 proteins, was retrieved from the UniProt database (UP000000542) and subjected to systematic computational analysis. Initial proteome characterization revealed a diverse functional landscape, with proteins ranging from 38 to 17,392 amino acids in length (mean: 542 ± 890 amino acids). The size distribution showed that 85.2% of proteins fell within the typical range of 100–1000 amino acids, consistent with other kinetoplastid proteomes.

Antigenicity profiling reveals bimodal distribution

VaxiJen v2.0 analysis of the entire proteome using the parasite model (threshold 0.5) revealed that 5,669 proteins (70.5%) demonstrated potential antigenicity, indicating a remarkably high proportion of potentially immunogenic proteins compared to other parasitic proteomes. The antigenicity score distribution exhibited a distinct bimodal pattern [Figure 1], with peaks at 0.4–0.6 and 0.8–1.0, suggesting two distinct population of proteins with different immunogenic potentials.

Figure 1.

The distribution of L. major proteome according to antigenicity scores. The total quantity of proteins is displayed in the figure as bars that correspond to various antigenicity score ranges. The total antigenic profile of L. major proteome is represented by each bar, which shows the quantity of proteins that match to a certain antigenicity score

To ensure selection of only highly antigenic candidates, we applied a stringent threshold of ≥0.8, yielding 182 proteins (2.3% of the total proteome). This represents a 97-fold enrichment compared to random selection, demonstrating the effectiveness of our filtering approach. The selected proteins showed antigenicity scores ranging from 0.8001 to 1.9007 (mean: 0.890 ± 0.122).

Functional annotation analysis of the 182 highly antigenic proteins revealed enrichment in several key biological processes: nucleic acid processing (23.1%), protein synthesis and processing (7.7%), metabolism, stress response, and structure (5.5%). While the remaining 63.7% lacked sufficient information and were annotated as “not available” including 29 membrane-associated proteins, which highlights a significant knowledge gap in the L. major proteome and suggests potential for discovering novel vaccine targets.

Subcellular localization analysis identifies membrane-associated candidate

Comprehensive subcellular localization prediction using combined UniProt annotations and DeepLoc 2.0 analysis revealed the distribution of highly antigenic proteins across cellular compartments [Table 1]. Out of the 182 analyzed proteins, 70 proteins (38.5%) lacked subcellular localization information in UniProt. Therefore, DeepLoc, a deep learning-based prediction tool, was employed. In addition, 37 (20.3%) were predicted to be membrane-associated, and 5 of them were predicted using DeepLoc, representing similar membrane content in most eukaryotic cells (~20%–30%).

Table 1.

Represent all Leishmania major proteins having antigenicity values <0.8

The membrane-associated proteins displayed higher average antigenicity scores (0.932 ± 0.20) compared to nonmembrane proteins (0.879 ± 0.08), even though the difference was not statistically significant (P = 0.113, t-test). This may further suggest that surface-exposed proteins may possess enhanced immunogenic properties in L. major.

Allergenicity assessment ensures safety profile

Dual allergenicity screening using AllerTOP v2.0 and AllergenFP v1.0 provided a robust safety assessment for the 37 membrane-associated candidates [Table 2]. The complementary nature of these tools was evident in their predictions:

Table 2.

Antigenicity and allergenicity investigation of Leishmania major membrane-associated proteins

Protein ID	Location		Score
	UniProt	DeepLoc	VaxiJen	AllerTOP	AllergenFP
E9ACK0	Membrane/Spliceosomal Complex		0.8334	Allergen	Non
Q4QBL7	Membrane		0.8126	Allergen	Allergen
Q4QHR9	Ciliary pocket/endoplasmic reticulum/plasma membrane		0.8288	Non	Allergen
Q4QJC1	Membrane		0.8041	Non	Allergen
E9ACH9	Membrane		0.8640	Non	Non
E9ACY3	Membrane		0.8814	Allergen	Non
E9ADR5	Endomembrane system/Golgi apparatus membrane		0.8726	Non	Non
E9AE22	Membrane		0.8293	Non	Non
E9AEC4	Membrane		0.8106	Non	Non
E9AEF2	Membrane		0.8949	Non	Non
E9AEH1	Membrane		0.8091	Non	Non
E9AEN0		Cell membrane	0.8865	Non	Non
E9AF05	Membrane		0.8544	Non	Non
Q4Q2N3	Membrane		0.8121	Non	Non
Q4Q4R1	Membrane		0.8250	Non	Allergen
Q4Q5U7	Membrane		0.8801	Non	Allergen
Q4Q689	Membrane		0.8239	Non	Non
Q4Q963	Membrane		0.8362	Non	Non
Q4QA86	Membrane		0.8568	Allergen	Non
Q4QAU6	Membrane		0.8376	Non	Non
Q4QCC4	Membrane/Mitochondrion		0.8413	Non	Non
Q4QF77	Membrane		0.8590	Non	Non
Q4QIJ8	NF	Cytoplasm/Cell membrane	0.8329	Non	Non
Q4QIL4	Membrane		0.8045	Non	Non
Q9NF81	Membrane		0.9227	Non	Allergen
Q4Q0T8	Membrane		0.9321	Non	Non
E9AEF5	Membrane		0.9193	Non	Non
Q4Q101	Membrane		0.9465	Non	Non
Q4Q487	NF	Cell membrane	0.9327	Non	Non
Q4QD75	Membrane		0.9570	Non	Non
Q4QIA9	Membrane		0.9120	Non	Allergen
E9ADZ2	Membrane		0.9615	Non	Non
E9ACB5	Membrane		1.0027	Non	Non
E9AG14	Membrane		1.2674	Non	Non
Q4QB56	NF	Cytoplasm/Membrane	1.2709	Non	Allergen
P90552	NF	Membrane	1.3841	Allergen	Allergen
Q4QIA8	Membrane		1.9007	Non	Non

The table contains the list of 37 membrane proteins with precise antigenicity score and allergenicity predictions obtained using the AllerTOP and AllergenFP tools. NF: Not found

• Both tools predicted nonallergenic: 25 proteins (67.6%)

• AllerTOP only predicted allergenic: 3 proteins (8.1%)

• AllergenFP only predicted allergenic: 7 proteins (18.9%)

• Both tools predicted allergenic: 2 proteins (5.4%)

The high concordance rate between the two algorithms increases confidence in allergenicity predictions. The 25 nonallergenic proteins were retained for further analysis using the Expasy ProtParam online tool (https://web.expasy.org/protparam/), which showed diverse molecular weights (4.2–225.6kDa) and isoelectric points (4.5–10.6), indicating chemical diversity suitable for different vaccine formulation strategies.

Cross-species conservation analysis reveals phylogenetic patterns

Comprehensive BLAST analysis against five Leishmania species revealed striking patterns of evolutionary conservation [Table 3]. The conservation analysis demonstrated that seven proteins with high conservancy (≥90%) in at least three species, including Q4QF77, Q4Q0T8, and Q4Q101, showed high similarity across all tested species, representing core membrane functions likely essential for parasite survival. Eight proteins displayed moderate conservation (≥80%) identity in at least three species, including E9ACH9, E9AEC4, and Q4QF77, suggesting species-specific adaptations of conserved functions. The remaining 10 proteins exhibited limited conservation with (<80%) identity. In addition, only two proteins, E9AEF2 and Q4QIA8, showed no detectable similarity to any other species. This may represent L. major-specific virulence factors or rapidly evolving surface antigens. In parallel, human homology assessment revealed three proteins with detectable similarity to human proteins, with the highest being 34% identity in E9ACH9, with no protein exceeding 35% identity threshold considered risky for autoimmunity. The phylogenetic analysis suggests that highly conserved proteins (>90% identity) likely represent essential membrane functions that could serve as broad-spectrum vaccine targets, while variable proteins might be useful for species-specific immunity.

Table 3.

Shows the final selection of 25 Leishmania major membrane proteins that are nonallergenic

Protein ID	Leishmania species					Human homolog; Identity (%)
	Leishmania donovani (%)	Leishmania infrantum (%)	Leishmania guyanensis (%)	Leishmania braziliensis (%)	Leishmania mexicana (%)
E9ACH9	92	92	78	78	88	34
E9ADR5	94	95	NS	83	94	NS
E9AE22	76	NS	NS	NS	NS	NS
E9AEC4	NS	93	81	82	93	NS
E9AEF2	NS	NS	NS	NS	NS	NS
E9AEH1	74	75	NS	NS	NS	NS
E9AEN0	28	88	61	61	70	22
E9AF05	92	93	NS	66	90	NS
Q4Q2N3	92	93	82	67	90	NS
Q4Q689	88	88	63	64	81	NS
Q4Q963	96	97	NS	88	96	NS
Q4QAU6	92	93	NS	73	87	NS
Q4QCC4	96	96	80	80	89	NS
Q4QF77	94	95	87	88	95	NS
Q4QIJ8	72	72	NS	NS	53	NS
Q4QIL4	94	94	67	68	87	NS
Q4Q0T8	97	97	86	86	95	28
E9AEF5	91	90	70	71	88	NS
Q4Q101	97	97	80	80	90	NS
Q4Q487	NS	84	54	58	71	NS
Q4QD75	89	89	NS	63	83	NS
E9ADZ2	69	65	NS	NS	NS	NS
E9ACB5	NS	60	NS	NS	74	NS
E9AG14	53	49	NS	NS	NS	NS
Q4QIA8	NS	NS	NS	NS	NS	NS

The conservation percentages against five distinct Leishmania species and human homolog to determine similarity are all included in the table. NS: No similarity

DISCUSSION

This study represents the most comprehensive computational analysis of L. major membrane proteins for vaccine development to date. Our multi-stage filtering approach successfully reduced 8,038 proteins to 25 high-priority candidates, achieving a 321-fold enrichment while maintaining functional diversity. This dramatic reduction demonstrates the power of systematic computational screening in focusing experimental resources on the most promising targets. In addition, the filtering pipeline steps displayed in the current study provide a time-saving and cost-effective method to narrow down a large set of proteins into a shortlist of potential vaccine candidates.

Antigenicity, allergenicity, and conservation are three essential features of successful vaccine candidates. Antigenicity is the ability of an antigen or substance to bind and interact with the immune system.[27] An effective vaccine should possess high antigenicity and show no allergenicity. In this study, by using VaxiJen on the whole proteome (8038 proteins), we selected 182 proteins that presented an antigenicity score greater than 0.8. This high threshold guarantees that only proteins possessing strong immunogenicity were involved.

Another significant factor for vaccine candidate selection is subcellular localization, as it determines the accessibility of the antigens to the host. Leishmania membrane-associated proteins play a critical role in the host-parasite and are often involved in immune activation, which give them priority as vaccine candidates.[28] In the current study, out of 182 proteins, 37 were identified as membrane-associated proteins by employing UniProt and DeepLoc. It was noticed that the UniProt database missed the localization information of some proteins; therefore, DeepLoc was used to ensure potential vaccine candidates were not overlooked. However, its predictions remain computational and need further experimental validation.

Allergenicity is the ability of an antigen or substance to cause an allergic reaction and overreaction of the immune system.[27] An essential consideration in vaccine development is to perform allergenicity assessment for candidate proteins, because of their ability to pose a safety concern. Out of 37 membrane-associated proteins, 25 were nonallergenic based on dual-platform evaluation by AllerTOP and AllergenFP. Notably, one of the proteins that considered for vaccine design in an earlier study HASPB (P90552) against leishmaniasis,[21] was detected as allergenic in this study. This finding raises questions about the safety profile of protein selection in future vaccine design studies.

Conserved antigens mostly represent important regions that play significant roles in parasite lifecycle and survival, which explains their sequence stability among different species. Therefore, the identification of highly conserved proteins may enhance vaccine effectiveness.[29] Highly conserved proteins (>90% identity) such as Q4QF77 and Q4Q101 represent ideal candidates for pan-Leishmania vaccines, potentially providing protection against multiple species. Moderately conserved proteins (80%–90% identity) such as Q4Q689 and Q4QCC4 could serve as components in region-specific vaccines, which account for local parasite strain variations while maintaining broad protective capacity. Species-specific proteins like Q4Q487 and E9AG14 might be valuable for targeted immunization in areas where L. major is the predominant species, potentially offering enhanced species-specific protection. The DDRGK domain-containing protein E9ACH9, showing 34% human similarity, requires careful epitope design to avoid potential cross-reactivity while maintaining immunogenicity.

The identification of 25 nonallergenic, highly antigenic membrane proteins significantly expands the repertoire of potential L. major vaccine candidates. Remarkably, none of these proteins has been previously investigated as vaccine targets in leishmaniasis research, highlighting the untapped potential of the L. major proteome. This finding contrasts sharply with the field’s historical focus on a small number of well-characterized antigens such as GP63, HASPB, LeIF and KMP-11.[16,21,30,31,32] The identification of multiple uncharacterized proteins with high antigenicity and conservation scores (e.g. E9AEF5, Q4QAU6) suggests that systematic screening can uncover novel vaccine targets that might never be identified through traditional hypothesis-driven approaches. These “hidden gems” within the proteome may possess optimal immunological properties while avoiding the immune evasion mechanisms that established antigens have evolved.

Our study introduces several methodological improvements over previous computational vaccine studies via dual allergenicity screening using complementary algorithms, significantly reduces false-negative predictions, enhancing safety assessment reliability, and comprehensive cross-species analysis against five Leishmania species provides a realistic assessment of broad-spectrum vaccine potential, moving beyond single-species studies. A key strength of the present study lies in the versatility of the suggested filtering approach, which can be applied to the proteome of other pathogens to facilitate the identification of novel vaccine candidates.

However, experimental validation remains essential. The predicted antigenicity scores, though statistically validated, require confirmation through immunological assays. Similarly, allergenicity predictions, despite dual-algorithm consensus, need experimental verification in appropriate models. This study focused on whole-protein screening; the next critical step involves detailed epitope mapping of the identified candidates. B-cell and T-cell epitope prediction, followed by HLA binding analysis, will refine vaccine design and enable personalized vaccination strategies based on population-specific HLA distributions.

In summary, this comprehensive proteome-wide analysis has successfully identified 25 high-priority membrane-associated proteins, including 15 highly conserved proteins representing novel vaccine candidates for L. major. The systematic computational approach not only expands the available target repertoire but also provides strategic insights into conservation patterns and vaccine design strategies. The high proportion of previously uncharacterized proteins among our candidates demonstrates the value of unbiased proteome screening over traditional hypothesis-driven approaches. The identified candidates offer multiple pathways for vaccine development, from individual protein vaccines to complex multi-epitope formulations. The conservation analysis provides clear guidance for developing either broad-spectrum pan-Leishmania vaccines or targeted species-specific immunizations based on epidemiological needs. However, this work is limited by its dependence on computational predictions and database annotations; experimental validation is necessary to verify the reported vaccine candidate proteins. Future research will concentrate on the experimental validation of the discovered vaccine candidates and structural modeling to develop peptide-based or multi-epitope vaccines targeting L. major.

Ethical statement

All data used in this study were downloaded from publicly accessible database (UniProt).

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.