Introduction

Strongyloides stercoralis is a significant parasite that can cause chronic and severe infections. While strongyloidiasis is primarily caused by S. stercoralis, other species such as S. fuelleborni and S. kellyi can also cause this disease, although less frequently. Recognizing its importance, the World Health Organization (WHO) has, for the first time, included S. stercoralis as a target pathogen in its 2021–2030 neglected tropical diseases (NTD) control program [1].

S. stercoralis enters the human body through the penetration of the skin by its filariform larvae present in contaminated soil. Once inside, the larvae migrate through the bloodstream to the lungs, then reach the throat, where they are swallowed and eventually become adult worms in the small intestine. Unlike many other nematodes, the eggs of this parasite hatch within the host’s body, releasing rhabditiform larvae that are excreted in the stool. These larvae can either develop into free-living adults in the environment or transform back into infectious filariform larvae, continuing the cycle of infection [2]. One of the unique characteristics of this parasite is autoinfection. Instead of being excreted, some newly produced larvae directly re-enter the bloodstream through the intestinal wall or perianal skin. This allows the parasite to sustain the infection within the host’s body without external exposure. As a result, an individual can remain infected for years or even decades without re-exposure to a contaminated environment [3].

Under normal conditions, many infected individuals may remain asymptomatic or experience mild symptoms such as abdominal pain, intermittent diarrhea, and skin rashes. However, in immunocompromised patients, especially those receiving corticosteroid therapy, the risk of hyperinfection and disseminated disease increases significantly. Other predisposing factors include infection with human T-lymphotropic virus type 1 (HTLV-1), organ transplantation, hematologic malignancies (particularly lymphoma), hypogammaglobulinemia, chronic alcohol consumption, kidney failure, severe malnutrition, diabetes, and chemotherapy [4]. In such cases, larvae spread uncontrollably throughout the body, invading organs such as the lungs, liver, brain, and genitourinary system. This condition can lead to life-threatening complications, including respiratory failure, sepsis, and septic shock. The mortality rate in these patients is extremely high and, if not diagnosed and treated in time, can exceed 70% [5].

The diagnosis of strongyloidiasis has traditionally relied on the direct observation of larvae in stool samples. However, single-sample microscopic methods have a low sensitivity of approximately 21%, primarily due to the intermittent shedding of the parasite and low infection burden (3). Polymerase chain reaction (PCR) also faces significant challenges among molecular methods. One of the main limitations of PCR is its reduced sensitivity caused by the intermittent excretion of the parasite, leading to false-negative results even in the presence of infection. Additionally, the requirement for meticulous sample processing, high costs, technical complexity, and limited accessibility in resource-poor settings are major barriers to the widespread use of this technique. Furthermore, despite PCR’s high efficacy in detecting active infections, it does not provide information about past infections or assess treatment effectiveness [6]. Serological methods, particularly enzyme-linked immunosorbent assay (ELISA), are widely used for detecting infectious agents due to their high speed, low cost, and reproducibility. However, the use of native antigens in these assays presents several challenges. The extraction and production of these antigens are complex and costly processes that can introduce impurities, leading to false-positive results. Additionally, cross-reactivity remains a fundamental issue affecting diagnostic accuracy [7].

To enhance the precision of serological methods, the design of recombinant antigens using bioinformatics approaches has emerged as a novel and promising strategy. By integrating multiple sequences from the target parasite and focusing on highly immunogenic regions, these antigens enable broader antibody detection while reducing the likelihood of false-positive results due to cross-reactivity [8]. Research on infectious disease diagnostics using ELISA has demonstrated that employing multi-epitope and fusion antigens can significantly improve both the sensitivity and specificity of the assay. These advancements play a crucial role in enhancing the accuracy and efficiency of serological diagnostic methods [9,10].

In this study, four excretory–secretory (ES) proteins of Strongyloides stercoralis SsIR, L3NieAg.01, Ss1a, and Ss3a were selected as candidates for recombinant antigen design. These proteins were chosen based on previous evidence demonstrating their strong immunogenicity, consistent expression during clinically relevant life‑cycle stages, and proven recognition by host antibodies in infected individuals. Collectively, these characteristics make them suitable components for constructing an effective fusion antigen [11,12].

This study aimed to design a novel recombinant fusion antigen for the serodiagnosis of strongyloidiasis using immunoinformatics approaches. The anticipated outcome of this research is the identification of a promising antigen candidate with significant potential for incorporation into ELISA-based serodiagnostic assays for strongyloidiasis. The bioinformatics findings generated in this study are expected to facilitate the development of more accurate, sensitive, and specific diagnostic tools, ultimately contributing to improved detection and control of this neglected parasitic infection.

Methods

Study design

The overall workflow of this study was shown in Fig 1, which provided a summary of the methodological steps. Four immunogenic proteins (SsIR, L3NieAg.01, Ss3a, and Ss1a) were selected for the design of the fusion antigen. The most immunogenic regions of these proteins were identified based on epitope density and minimal cross-reactivity, and they were linked, using EAAAK linkers. The designed fusion antigen was then evaluated for its physicochemical properties, solubility, antigenicity, and potential cross-reactivity. Its three-dimensional (3D) structure was predicted, and the nucleotide sequence was codon-optimized to ensure efficient expression in Escherichia coli (E. coli). Finally, the optimized sequence was in silico cloned into the pET23a(+) expression vector.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart illustrated a fusion antigen’s in-silico design stages for diagnosing strongyloidiasis.

https://doi.org/10.1371/journal.pone.0351189.g001

Antigen selection and sequence retrieval

A comprehensive literature review was conducted to identify four antigenic sequences based on their immunogenic properties as reported in previous experimental studies. Selection criteria included evidence of immunoreactivity in ELISA or immunoblot assays and the potential for diagnostic application. The selected antigens included SsIR (GenBank ID: AAB97359.1, UniProt ID: O44394) and L3NieAg.01 (GenBank ID: AAD46493.1, UniProt ID: Q9UA16), as well as Ss3a and Ss1a (Patent No: WO2017091059A1) [13]. The amino acid sequences were retrieved from the UniProt database (https://www.uniprot.org/) in FASTA format, as summarized in Table 1.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Selected antigenic proteins for the diagnosis of strongyloidiasis infection.

https://doi.org/10.1371/journal.pone.0351189.t001

Results

Prediction of signal peptide

Signal peptide prediction was performed using the SignalP server (Fig 2). Among the analyzed antigens (SsIR, L3NieAg.01, Ss3a, and Ss1a), a signal peptide was predicted only for the Ss3a antigen at its N‑terminal region. This signal peptide consisted of three distinct segments: a positively charged region (residues 1–5, shown in red), a hydrophobic region (residues 6–17, shown in orange), and a cleavage region (residues 18–20, shown in yellow). The predicted cleavage site (CS), indicated by a black dashed line, was located around position 20, suggesting the position where the signal peptide is removed to generate the mature protein.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Signal peptide in Ss3a antigen, with the black dashed line indicating the cleavage site (~position 20).

https://doi.org/10.1371/journal.pone.0351189.g002

Prediction of linear B-cell epitopes

Linear B‑cell epitopes in the four antigenic proteins were predicted using the BCPRED, ABCpred, and BepiPred servers. The predicted epitope regions are shown in Fig 3. In the protein sequences, epitopes identified by BCPRED, representing the highest‑scoring regions, are highlighted in yellow. Epitopes predicted by ABCpred are indicated by blue bars, whereas those identified by BepiPred are shown with green bars. Regions predicted as epitopes by all three servers are highlighted in red.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Predicted linear B-cell epitopes in four antigenic proteins using BCPRED (yellow), ABCpred (blue bars), and BepiPred (green bars).

Red bars indicate consensus regions identified by all three servers.

https://doi.org/10.1371/journal.pone.0351189.g003

Construction of the fusion antigen

The fusion antigen was designed by selecting protein regions with a high density of predicted epitopes and no detectable cross‑reactivity. The selected regions were SsIR (residues 1–79), L3NieAg.01 (residues 1–107), Ss3a (residues 50–124), and Ss1a (residues 213–297). These fragments were joined using the (EAAAK) amino acid linker to construct the fusion antigen. The final structure of the designed fusion antigen is shown in Fig 4.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Schematic representation of the fusion antigen, in which the selected regions (SsIR, L3NieAg.01, ss3a, ss1a) are linked using the EAAAK linker.

https://doi.org/10.1371/journal.pone.0351189.g004

Antigenicity and cross-reactivity analysis

The predicted cross‑reactivity profiles of the full‑length antigens, their selected truncated regions, and the final fusion antigen were evaluated using BLASTp against proteins from several parasitic pathogens commonly associated with serological cross‑reactions. As shown in Table 2, the full‑length antigens exhibited varying degrees of sequence similarity with proteins from other parasites. Among them, SsIR demonstrated the highest level of potential cross‑reactivity, showing similarity with proteins from eight infections, including Toxocara, Echinococcus, Plasmodium, Hymenolepis, Leishmania, Cryptosporidium, Trichostrongylus, and Necator americanus. The L3NieAg.01 antigen also showed predicted similarity with seven infections, including Toxocara, Fasciola, Echinococcus, Hymenolepis, Toxoplasma, Trichostrongylus, and Necator americanus. In contrast, Ss3a did not show detectable similarity with any of the evaluated parasites, suggesting a potentially higher level of specificity. The Ss1a antigen displayed limited cross‑reactivity, with sequence similarity observed only with Toxocara, Plasmodium, and Cryptosporidium. Importantly, none of the selected truncated fragments used for constructing the fusion antigen showed detectable similarity with the evaluated parasitic pathogens in the BLASTp analysis. Consequently, the final fusion antigen also showed no predicted cross‑reactivity with any of the investigated parasites (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Predicted cross‑reactivity of full‑length antigens, selected truncated regions used in this study, and the designed fusion antigen with various parasitic pathogens based on BLASTp homology analysis.

https://doi.org/10.1371/journal.pone.0351189.t002

The antigenicity of the full‑length antigens, the selected truncated regions used in this study, and the final fusion construct was evaluated using the VaxiJen server with a threshold of 0.5. As summarized in Table 3, all analyzed sequences were predicted to be antigenic. Among the full‑length antigens, SsIR showed the highest antigenicity score (0.8672), followed by Ss3a (0.8616), L3NieAg.01 (0.6152), and Ss1a (0.5116). Notably, the selected fragment of SsIR (residues 1–79) exhibited the highest predicted antigenicity score (0.9873). The selected fragments from L3NieAg.01, Ss3a, and Ss1a also maintained antigenicity scores above the defined threshold. The designed fusion antigen showed a VaxiJen score of 0.6949, indicating that the antigenic potential was preserved after combining the selected fragments.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Predicted antigenicity scores of full-length antigens, selected truncated regions, and the final designed fusion antigen.

https://doi.org/10.1371/journal.pone.0351189.t003

Physicochemical properties

The physicochemical properties of the fusion antigen were analyzed using the ProtParam server. The results indicated that the protein is stable, with an instability index of 39.68. The molecular weight was estimated to be approximately 50.81 kDa, and the predicted isoelectric point (pI) was 7.12. The aliphatic index was calculated as 57.83. In addition, SolPro analysis predicted that the fusion antigen is soluble, with a solubility score of 0.814387 (Table 4).

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Evaluation of a fusion antigen’s physicochemical properties, antigenicity, and cross-reactivity.

https://doi.org/10.1371/journal.pone.0351189.t004

Prediction of secondary and tertiary structures and validation of the 3D model

The secondary structure of the fusion antigen was predicted using the PSIPRED server (Fig 5a). Tertiary structure prediction was performed with the GalaxyTBM server, which generated five candidate models. These models were assessed using Ramachandran plot analysis, and model 2 was selected as the optimal 3D structure (Fig 5b). In the selected model, 92.8% of the residues were located within the allowed regions (Fig 5c).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Prediction and evaluation of the fusion antigen structure.

(a) Prediction of the secondary structure using the PSIPRED server. (b) The optimal 3D structural model generated by the GalaxyTBM server. (c) Ramachandran plot for structural quality assessment shows that 92.8% of the residues fall within the allowed regions.

https://doi.org/10.1371/journal.pone.0351189.g005

Prediction of conformational B-cell epitopes

Conformational B-cell epitopes were predicted using the ElliPro server, which identified ten discontinuous epitopes comprising a total of 176 residues. The prediction scores ranged from 0.843 to 0.503, reflecting different confidence levels. The predicted epitopes varied in size from 6 to 66 amino acids. Detailed results are summarized in Table 5 and visualized in Fig 6.

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. List of discontinuous B-cell epitopes in fusion antigen predicted by the ElliPro server.

https://doi.org/10.1371/journal.pone.0351189.t005

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Prediction of conformational B-cell epitopes using the ElliPro server.

The fusion antigen construct is represented as gray sticks, with the predicted conformational B-cell epitopes highlighted in yellow.

https://doi.org/10.1371/journal.pone.0351189.g006

Quality assessment of the optimized sequence and prediction of mRNA secondary structure

Codon optimization was performed to improve the expression of the fusion antigen in Escherichia coli. After optimization, the codon adaptation index (CAI) of the original sequence decreased from 0.96 to 0.92, which remains within the recommended range for efficient expression (Fig 7a). The frequency of optimal codons (FOP) analysis showed that the original sequence contained a lower proportion of preferred codons, whereas the optimized sequence exhibited a substantial increase in the use of frequently employed codons, supporting improved translational efficiency (Fig 7b). The GC content was also adjusted to fall within the optimal range, decreasing from 52% to 47% (Fig 7c). The initial sequence contained one CIS element and six repetitive sequences, which were successfully eliminated following codon optimization (Table 6).

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Evaluation of negative CIS elements and repetitive sequences before and after codon optimization.

https://doi.org/10.1371/journal.pone.0351189.t006

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Evaluation of the optimized sequence and mRNA secondary structure for the designed fusion antigen.

(a) Codon adaptation index (CAI) before (0.96) and after (0.92) codon adjustment. (b) Frequency of optimal codons (FOP) before and after optimization. (c) GC content before (52%) and after (47%) optimization. (d) Predicted mRNA secondary structure with a minimum free energy (MFE) −304.12 [Initially −391.00] kcal/mol.

https://doi.org/10.1371/journal.pone.0351189.g007

Furthermore, prediction of the mRNA secondary structure indicated that the minimum free energy (MFE) of the optimized sequence was −304.12 kcal/mol, a decrease from the original sequence’s MFE of −391.00 kcal/mol (Fig 7d).

Recombinant vector design

The recombinant pET23a(+) vector, harboring the fusion antigen sequence, was designed using SnapGene software. In this construct, the NdeI restriction site was positioned upstream, while the XhoI restriction site was located downstream of the target sequence. The final vector construct is shown in Fig 8.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. In-silico design of the recombinant pET-23a(+) vector containing the fusion antigen.

The codon-optimized sequence (highlighted in pink) was inserted between the NdeI and XhoI restriction sites within the pET-23a(+) expression vector, representing a black circular structure.

https://doi.org/10.1371/journal.pone.0351189.g008

Discussion

Bioinformatics analyses conducted in this study indicated that the designed fusion antigen possesses favorable immunological characteristics, supporting its potential application in ELISA‑based diagnostic assays for human strongyloidiasis.

For the construction of the fusion antigen, four protein sequences SsIR, L3NieAg.01, Ss3a, and Ss1a were selected based on their reported antigenic properties and relevance in previous studies [11,12]. SsIR is one of the major excretory‑secretory (ES) proteins expressed during both the invasive larval and adult stages of S. stercoralis. It is secreted through the parasite’s ES system and plays a significant role in modulating host immunity [29]. Earlier investigations have shown that SsIR can impair the function of dendritic cells and macrophages, leading to reduced antigen presentation to T lymphocytes and weakening downstream immune activation. This immunomodulatory effect contributes to parasite survival by enabling evasion of host immune defenses [30]. L3NieAg.01, another ES protein expressed by adult parasites, is known to induce robust immune responses. In addition to its involvement in humoral immunity, recent evidence suggests that this antigen may promote parasite attachment to intestinal epithelial cells, potentially supporting tissue colonization and persistent infection [12,31]. Ss1a and Ss3a are ES antigens secreted during the invasive larval stage and are also recognized for their immunogenicity. These proteins stimulate host immune responses and are considered valuable targets for diagnostic and vaccine development [11].

To construct the fusion antigen, selected regions from the four target antigens were assembled into a single continuous polypeptide sequence. The fragments were connected using the EAAAK linker (Fig 4). This linker forms a rigid α‑helical structure that preserves adequate spatial separation between the fused sequences, minimizes steric interference, and enhances the accessibility of epitopes for antibody binding [32]. The EAAAK linker has been widely applied in recombinant multi‑epitope antigen design. For example, it has been used successfully to connect epitopes from five tegument proteins in the development of a multi‑epitope antigen for human cytomegalovirus (HCMV) detection [32].

The design of fusion antigens using bioinformatics approaches can significantly enhance the accuracy of serological diagnostic methods. By integrating multiple antigenic sequences from the target pathogen and focusing on regions with high immunogenic potential, such constructs allow for broader antibody recognition while minimizing the risk of false-positive results due to cross-reactivity [33].

Although the proposed construct contains four antigenic components (compared with one or two components in previously reported systems), this design was implemented intentionally. By integrating immunodominant regions derived from multiple antigens, the fusion construct was designed to provide broader epitope coverage while simultaneously eliminating sequence regions associated with potential cross‑reactivity. Therefore, the increased complexity of this construct reflects a rational design strategy aimed at achieving an optimal balance between sensitivity and specificity.

The predicted cross‑reactivity profiles of the full‑length antigens, their selected truncated regions, and the final fusion antigen were evaluated using BLASTp against proteins from several parasitic pathogens that are commonly associated with serological cross‑reactions. The results indicated that the full‑length antigens exhibited varying degrees of sequence similarity with proteins from other parasites. Among them, SsIR showed the highest potential for cross‑reactivity, while L3NieAg.01 also demonstrated predicted similarities with several parasitic proteins. In contrast, Ss3a showed no detectable similarity with any of the evaluated parasites, suggesting a potentially higher level of specificity. Ss1a exhibited only limited sequence similarity with a few parasitic organisms. Importantly, none of the selected truncated fragments used for the construction of the fusion antigen showed detectable similarity to the evaluated parasitic pathogens in the BLASTp analysis. Consequently, the final fusion antigen also showed no predicted cross‑reactivity with any of the examined parasites (Table 2).

Assessing the potential for cross‑reactivity represents a critical step in the rational design of fusion antigens, as the removal of sequence regions that may interact with antibodies generated against unrelated pathogens can significantly improve the specificity of serological assays [9,32].

However, it should be noted that BLASTp‑based analyses provide only an initial level of screening. Sequence similarity alone cannot fully capture the complexity of epitope–antibody interactions. In some cases, antibody cross‑reactivity may occur even when overall sequence similarity is relatively low. Moreover, sequence‑based comparison methods are unable to detect structural mimicry or physicochemical similarities between epitopes. Therefore, the in silico analyses performed in this study should be considered a preliminary assessment of antigen specificity. Experimental validation using sera from patients with other parasitic infections will be necessary to definitively evaluate the diagnostic specificity of the antigen and to confirm the absence of cross‑reactivity under real diagnostic conditions.

These findings are consistent with the study by Omidian et al., who developed a fusion antigen (SsIR–Ss1a) for the serological detection of strongyloidiasis. By aligning the target antigenic sequences with those of other infectious and parasitic agents, the researchers identified immunodominant regions that lacked cross‑reactivity. In their analysis, only three of the 52 serum samples from patients with other parasitic and infectious diseases showed cross‑reactive responses. The SsIR–Ss1a antigen demonstrated a sensitivity of 93.94% and a specificity of 97.22%, indicating strong diagnostic performance for strongyloidiasis [9].

Multi‑epitope and fusion antigens have been widely proposed as effective strategies for improving the specificity of serological assays. However, inappropriate or non‑specific selection of target sequences can reduce diagnostic accuracy and lead to false‑positive results. For example, a study evaluating a chimeric protein (OvCB_OvAEP_OvCF) for the diagnosis of opisthorchiasis reported strong IgG reactivity and achieved 100% sensitivity. However, cross‑reactivity with sera from patients infected with other helminths and protozoa reduced the specificity to 78.4% [34]. Similarly, another study assessed a multi‑epitope antigen for the diagnosis of visceral leishmaniasis using ELISA and Western blot assays. Although the antigen showed high sensitivity (93.1%), its specificity for correctly identifying negative cases was only 77.4%, indicating the need for further optimization [35].

The high diagnostic performance of assays based on NIE and NIE/SsIR antigens has been well documented, with most studies reporting sensitivities and specificities exceeding 90% [36]. Nevertheless, even these widely used diagnostic systems have demonstrated measurable cross‑reactivity. These observations suggest that, despite their strong overall diagnostic performance, the specificity of the currently used reference antigens may still be affected by shared antigenic determinants among phylogenetically related parasites. Furthermore, the number and diversity of heterologous serum samples included in cross‑reactivity panels can also influence the reported specificity values, as the use of broader and more diverse serum panels increases the likelihood of detecting potential cross‑reactive responses [37].

The antigenicity of the full‑length antigens, the selected truncated regions used in this study, and the final fusion construct was evaluated using the VaxiJen server with a threshold of 0.5. As shown in Table 3, all analyzed sequences were predicted to be antigenic. Among the full‑length antigens, SsIR exhibited the highest antigenicity score (0.8672), followed by Ss3a (0.8616), L3NieAg.01 (0.6152), and Ss1a (0.5116). Notably, the selected fragment of SsIR (residues 1–79) showed the highest predicted antigenicity score (0.9873). The selected fragments derived from L3NieAg.01, Ss3a, and Ss1a also retained antigenicity scores above the defined threshold. Furthermore, the designed fusion antigen obtained a VaxiJen score of 0.6949, indicating that the overall antigenic potential was preserved after integrating the selected fragments into a single construct.

It should be noted that the removal of certain sequence regions from individual antigens performed to eliminate segments with potential cross‑reactivity identified through BLASTp analysis may partially reduce the intrinsic antigenic strength of those proteins. In this context, integrating multiple antigenic components derived from the target pathogen into a single fusion construct represents a rational strategy to compensate for this potential reduction. Incorporating several epitope‑rich regions originating from different antigens within a single recombinant construct may broaden the range of antibody recognition across heterogeneous host immune responses and thereby potentially enhance the overall sensitivity of serological diagnostic assays [38,39].

In a related study, the identification of linear B-cell epitopes within the SjSAP4 protein sequence and their diagnostic potential for Schistosoma japonicum infection responsible for schistosomiasis was evaluated. Peptides derived from the SjSAP4 protein were expressed as fusion proteins with a GST tag and subsequently assessed using ELISA. The results demonstrated that a dual-peptide ELISA combining SjSAP4 and SjSP-13V2 peptides achieved a sensitivity of 84% and specificity of 100%, outperforming single-peptide ELISA formats. These findings underscore the diagnostic advantages of combining antigenic epitopes from different proteins to develop more effective tools for detecting schistosomiasis infections [40].

Another study compared the diagnostic performance of individual proteins BP26, Omp25, and Omp31 with that of a multi-epitope fusion protein for the serological detection of brucellosis. The multi-epitope construct demonstrated superior performance, with a positive predictive value (PPV) of 100% and a negative predictive value (NPV) of 98.41%, indicating its high accuracy in differentiating between positive and negative sera compared to single-antigen assays [41].

The physicochemical properties of the designed fusion antigen were assessed using ProtParam, focusing on instability and aliphatic indices. These properties are essential for recombinant protein expression. The instability index was calculated as 39.68, below the threshold of 40, indicating likely stability under physiological conditions [42]. The aliphatic index was 57.83, reflecting the proportion of aliphatic side chains (alanine, valine, isoleucine, and leucine) and suggesting good thermostability [43]. SolPro was used to predict solubility, yielding a probability score of 0.81. This result indicates that the fusion antigen is likely to be soluble, which is important for selecting an expression system and designing extraction and purification strategies [44].

The 3D structural validation of the selected model showed that approximately 90.2% of the predicted amino acid residues were located within the favored regions of the Ramachandran plot, while only a small fraction appeared in disallowed region. The aim of the three‑dimensional modeling in the present study was not to determine the exact atomic‑level structure of the fusion protein, but rather to provide a topological approximation of the spatial arrangement of the selected epitopes. This level of modeling is sufficient for the diagnostic purpose of the study, which focuses on the design of the fusion construct and the evaluation of epitope accessibility, and it can effectively guide subsequent experimental steps, particularly during protein extraction and purification.Since the structural model suggests that certain epitopes may be partially buried within the protein core, such information supports the use of denaturing purification conditions to ensure exposure of these regions. Accordingly, in the experimental phase, it is recommended that the fusion antigen be denatured using 8 M urea and subsequently immobilized in a linear conformation onto the ELISA plate surface. Under these conditions, the proper or improper folding of the protein does not play a decisive role in antigen performance in the ELISA assay. Therefore, the three‑dimensional model is primarily intended to guide the optimization of extraction and purification procedures, rather than to define the precise structural conformation of the fusion protein [32].

Conformational B‑cell epitopes in the final 3D model were predicted using the ElliPro tool. This analysis identified ten discontinuous epitopes. However, it should be noted that the fusion antigen was constructed by linking four distinct sequences with linkers; therefore, the native conformational epitopes present in the parental antigens may not be fully preserved in the chimeric structure. For this reason, the primary focus of the construct design was on linear B‑cell epitopes. Nevertheless, conformational epitope prediction was performed to assess the potential formation of new structural patches within the fusion construct and to provide a more comprehensive overview of its antigenic features [32].

To optimize the expression of the recombinant protein and mitigate challenges associated with codon usage bias, the sequence was optimized. This process involved evaluating several key parameters: CAI, FOP, GC content, and the presence of cis-acting regulatory elements, and repetitive sequences.

Following optimization, the CAI reached 0.92 (Fig 7a) [32,45]. This high value indicates strong compatibility between the optimized codons and the host organism’s translation machinery, strongly suggesting enhanced translational performance, as values closer to 1.0 signify preferential use of host-favored codons [32,45]. The FOP analysis showed that the original sequence had a lower proportion of preferred codons (Fig 7b) [32,45]. A higher FOP, achieved in the optimized sequence, implies a greater abundance of codons efficiently recognized by host ribosomes, which can significantly improve protein yield [32,45]. The GC content of the optimized construct was successfully adjusted to 47% (Fig 7c) [32,45]. This value falls within the optimal range (30–70%), which is critical for DNA duplex stability and directly impacts transcription and translation efficiency [32,45]. Crucially, the initial sequence contained one cis-regulatory element and six repetitive sequences, elements known to negatively affect expression. These problematic elements were effectively removed during optimization (Table 6), minimizing the risk of premature translation termination and maximizing the expression potential of the target antigen.

The mFold server predicted an mRNA minimum free energy (MFE) of –304.12 kcal/mol after optimization, compared with –391.00 kcal/mol before optimization (Fig 7d). Although lower free energy generally indicates higher structural stability, excessively stable mRNA structures may negatively affect translation efficiency. Highly stable structures can promote the formation of strong hairpin loops that may mask the ribosome‑binding site or impede ribosome progression along the mRNA. Therefore, the reduced structural stability observed after optimization may help decrease rigid secondary structures and improve ribosomal accessibility during translation [46].

The recombinant pET23a(+) vector carrying the fusion antigen was designed using SnapGene software. This vector contains a C‑terminal histidine tag located within the multiple cloning site (Fig 8), which facilitates the purification of the recombinant protein. The choice of affinity tags is an important consideration in diagnostic applications because certain tags may contribute to cross‑reactivity in detection assays. For instance, commonly used expression vectors such as pET32a and pGEX4T1 contain thioredoxin (trx) and glutathione S‑transferase (GST) tags, respectively. These tags share notable sequence homology with several parasitic proteins, which may increase the likelihood of cross‑reactive signals. Moreover, if these tags are not removed after protein expression and remain fused to the antigen, they may act as additional epitopes and further contribute to cross‑reactivity. Such cross‑reactivity could potentially reduce the specificity and reliability of serological tests for the diagnosis of Strongyloides infection [32].

Conclusion

In this study, we utilized bioinformatics tools to design a novel fusion antigen by selecting regions with high immunogenic potential. This approach was intended to improve antibody recognition while minimizing the risk of false-positive results arising from cross-reactivity. In silico analyses demonstrated that the designed fusion construct exhibits promising immunological features, supporting its potential application in ELISA-based diagnostic assays for human strongyloidiasis. However, a major limitation of this study is the lack of experimental validation. All current findings are only based on computational predictions. Therefore, further experimental studies are crucial to evaluate the construct’s diagnostic accuracy, specificity, and feasibility for clinical implementation.

Supporting information

Acknowledgments

The authors gratefully acknowledge the colleagues who provided valuable scientific insights and technical assistance during the completion of this research.