Chimeric protein construction

The FASTA sequence for IL-24 was retrieved from the Universal Protein Knowledgebase (Accession number: Q13007), while the sequence for the lytic peptide magainin 2 was obtained from the previously reported work of Hoskin17. The 3D structure of the heterodimeric IL-24 receptor was acquired from the Protein Data Bank (PDB ID: 6DF3). To isolate the mature fragment of the IL-24 peptide, amino acids 1-56 from the amino terminus were removed. The remaining peptide was fused to magainin 2 via a previously reported rigid linker sequence (EAAAKEAAAKEAAAK)18 to engineer the final fusion construct.

Prediction of protein secondary structure

The secondary structure of the magainin 2-IL-24 fusion protein was predicted using the GOR-IV server. This method analyzes the FASTA sequence of the chimeric protein to provide insights into its secondary structural elements, including alpha-helices, random coils, beta-sheets, coiled coils, low-complexity regions, extended strands, and structurally disordered regions19,20.

3D structure modeling and refinement of magainin 2-IL-24 fusion protein

Three-dimensional (3D) structural modeling was performed using both the AlphaFold2 and I-TASSER (V5.1) online servers. Employing two distinct methodologies enhanced the accuracy and reliability of the predicted structure and facilitated the selection of the most optimal model for subsequent analyses. I-TASSER (V5.1) primarily utilizes a template-based modeling approach, initiating the process by identifying structurally similar proteins with known 3D structures from the Protein Data Bank (PDB) using threading and sequence alignment techniques21,22. AlphaFold2, developed by DeepMind, relies on deep learning techniques, employing a neural network architecture designed to predict inter-residue distances within the protein sequence. Both servers require only the FASTA sequence as an input to construct a 3D model23. The generated models were further refined to minimize energy, optimize side-chain conformations, and improve stereochemistry and bond angles using the Galaxy Refine (V2.0) online server24.

Structural validation of the fusion protein

The quality of the constructed models was evaluated using multiple computational tools. For an overall accuracy assessment, the ProSa-web tool was employed to calculate the Z-score, which provides the global quality of the 3D model and estimates its structural alignment with experimentally resolved proteins25. Next, the overall quality and reliability of the model were assessed through the ERRAT2 graph26. Lastly, the 3D model was validated for its stereochemical integrity by generating a Ramachandran plot.

Prediction of toxic effects, immunogenicity, and allergen potential

The ToxinPred server was utilized to predict regions within the submitted sequence containing toxic amino acid residues27. Allergenicity was assessed using AllerTOP, which operates based on the similarity index of protein regions with known epitopes28. The VaxiJen server was employed to compute antigenicity based on physicochemical properties29, utilizing a 0.5 threshold to differentiate between antigenic and non-antigenic properties. These predictive analyses served as a critical foundation for evaluating the safety profile of the fusion protein as a potential drug candidate.

Docking analysis and interaction studies

Blind molecular docking was performed using the ClusPro 2.0 online server16,30. The 3D structure of the heterodimeric receptor was downloaded and prepared using the PyMOL molecular graphics system, which involved the removal of endogenous IL-24, ligands, and water molecules to ensure a purified receptor structure. The PDB-formatted structures of the fusion protein and its receptor were uploaded to the ClusPro 2.0 server, and docking was performed using default parameters. The resulting docked complex was subjected to an in-depth analysis of protein-protein interactions using PDBsum and PDBePISA, enabling the identification of critical features such as binding affinity, interacting interfaces, hydrogen bonds, solvation energy (kcal/mol), and salt bridges within the docked complex31.

Molecular dynamics simulation

The top-ranked docked complex was subjected to molecular dynamics (MD) simulation using NAMD v2.14, with trajectory visualization performed in VMD v1.9.332. System preparation and topology generation were carried out using AmberTools21, employing the ff14SB force field for the protein complex33. Missing hydrogen atoms were added, and protonation states were assigned to be consistent with physiological pH. The complex was solvated in an explicit TIP3P water model within an orthorhombic periodic box extending 10 Å beyond the protein in all directions. Counter ions (Na⁺ and Cl⁻) were added to neutralize the system and achieve physiological ionic strength.

Prior to the production simulation, the system underwent energy minimization for 10,000 steps using the conjugate gradient algorithm to resolve steric clashes. This was followed by a two-stage equilibration process: an NVT ensemble equilibration at 310 K using Langevin dynamics for temperature control, and an NPT ensemble equilibration for 1 ns at 1 atm pressure, maintained using the Nosé–Hoover Langevin piston method. The production MD simulation was then performed for 100 ns under periodic boundary conditions. Long-range electrostatic interactions were calculated using the Particle Mesh Ewald (PME) method, and covalent bonds involving hydrogen atoms were constrained using the SHAKE algorithm. Trajectory analyses were conducted using the bio3d package (v2.4-0) in R (v4.x)34, where structural stability and conformational dynamics were evaluated through root-mean-square deviation (RMSD), radius of gyration (Rg), and root-mean-square fluctuation (RMSF). Stability was assessed based on RMSD convergence and plateau behavior during the latter half of the simulation.

Expression potential assessment in E. coli

To predict the soluble expression of the fusion protein in E. coli, the SoluProt online server was utilized. This computational tool was selected based on its excellent accuracy compared to other existing tools, as demonstrated by Ghomi et al.35. By leveraging SoluProt, we were able to generate informed predictions expected to significantly enhance the efficiency and success of future in vitro protein expression experiments, ensuring optimal conditions for high-yield expression.

Engineering of magainin 2-IL 24 fusion protein

The FASTA sequences encoding IL-24 and magainin 2 were retrieved from the National Center for Biotechnology Information (NCBI) database and from a previously published study, respectively. To engineer a single polypeptide chain comprising 193 amino acids, these sequences were fused via a rigid linker incorporating the amino acid sequence EAAAKEAAAKEAAAK (Figure 1A). This rigid linker plays a crucial role in preventing aberrant disulfide bond formation between the distinct functional domains of the fusion protein. To facilitate downstream molecular cloning for in vitro expression studies, specific restriction enzyme cleavage sites were incorporated at both the N- and C-termini.

Secondary structure evaluation of chimeric proteins

Secondary structure prediction of the magainin 2-IL-24 chimeric protein, conducted using the GOR IV server, provided valuable insights into its structural composition and potential functional roles. Notably, the analysis revealed that a substantial portion of the protein (55.96%) adopts an alpha-helical conformation. This abundance of alpha helices signifies a high degree of structural stability, which is pivotal for the protein's overall functionality and therapeutic efficacy. Moreover, the presence of extended strands and beta-sheets (15.54%) adds structural diversity that may facilitate specific functional interactions and enhance adaptability to various molecular environments. Additionally, the remaining 28.50% random coil conformation provides critical flexibility, which is essential for the construct's dynamic biological functions (Figure 1B).

3D structure modeling of magainin 2-IL 24 fusion protein

To generate accurate three-dimensional representations, the chimeric protein's FASTA sequence was initially submitted to both the AlphaFold2 and I-TASSER (V5.1) online servers. AlphaFold2, recognized for its exceptional side-chain modeling accuracy, generated five models ranked by local model quality metrics; the top-ranked model was selected for subsequent analysis. I-TASSER (V5.1), employing its C-score as a confidence measure, similarly produced five candidate models. The model exhibiting the highest C-score (0.97), corresponding to a TM-score of 0.796 and an RMSD of 0.34, was chosen for comparison. The top models from both computational platforms were then compared to determine the most precise and stable configuration of the chimeric protein. A comprehensive evaluation based on structural quality, validation metrics, and Ramachandran plot scores demonstrated that the AlphaFold2-generated model was significantly superior to the I-TASSER (V5.1) model across overall scoring parameters. Consequently, the AlphaFold2 structure was selected for all downstream analyses (Figure 1C).

Reliability and validation-based selection of a chimeric protein model

The comparison of homology models generated by AlphaFold2 and I-TASSER (V5.1) involved the evaluation of several critical parameters to assess structural quality and reliability. The Ramachandran plot, a crucial tool for assessing stereochemical quality, revealed that both models possessed a high proportion of amino acids in the favorable region (97% for AlphaFold2 and 92% for I-TASSER), indicating energetically favorable phi and psi conformations. However, AlphaFold2 exhibited superior stereochemistry (Figure 2A). ERRAT2 analysis, which evaluates non-bonded atomic interactions, yielded an ERRAT score of 95 for the AlphaFold2 model, compared to a slightly lower score of 94% for the I-TASSER (V 5.1) model, suggesting better agreement with high-resolution structural data (Figure 2B). The Z-score, which evaluates overall model quality relative to experimentally determined structures, was negative for both models, reflecting favorable native-like folding. The AlphaFold2 model yielded a Z-score of -5.84 compared to the I-TASSER (V 5.1) model score of -6.02 (Figure 2C). Collectively, based on Ramachandran stereochemistry, ERRAT scores, and global Z-scores, the AlphaFold2 model proved to be more reliable and accurate, justifying its selection for further analysis and applications involving the fusion protein structure.

Physicochemical characteristics and solubility prediction

The physicochemical properties of the chimeric fusion protein provide valuable insights into its potential as a therapeutic agent. These properties were estimated using the ProtParam server (Table 1). The fusion protein consists of 193 amino acids with a molecular weight of 22.02 kDa, indicating its substantial size and complexity. The isoelectric point (pI) of 9.19 suggests that the protein is relatively basic. Additionally, a high extinction coefficient of 22,585 M⁻¹ cm⁻¹ indicates that it can be effectively quantified via spectrophotometric assays, facilitating further analysis and development. Notably, the protein contains 20 negatively charged residues and 25 positively charged residues, which may influence electrostatic interactions during physiological processes. Moreover, the estimated half-life of over 10 hours in E. coli, along with an instability index of 36.78, suggests that the protein is thermodynamically stable and persistent within a biological system. A high aliphatic index (88.50) and a negative Grand Average of Hydropathicity (GRAVY) score (-0.147) indicate an optimal balance of hydrophilicity and hydrophobicity, facilitating interaction with cellular components. On the Protein-sol online server, the selected model exhibited a scaled solubility value of 0.478, which exceeds the threshold of 0.45, indicating that the fusion protein is likely soluble—a critical factor in drug formulation and delivery. Adequate solubility ensures the protein is readily available for interactions with target cancer cells and can be administered through various pharmaceutical routes. These properties collectively emphasize the structural viability of the magainin 2-IL-24 fusion protein as a stable, soluble candidate for anti-cancer therapeutics, highlighting its potential for targeted interactions and stability within biological systems, making it a promising candidate for further investigation and therapeutic development.

Evaluation of allergenicity, antigenicity, and toxicity

The comprehensive assessment of allergenicity, antigenicity, and toxicity is an essential step in evaluating the clinical safety of therapeutic proteins. In this study, AllerTOP predicted the fusion construct to be non-allergenic, indicating a low probability of inducing hypersensitivity reactions upon systemic administration. VaxiJen analysis, utilizing a threshold of 0.5, classified the protein as non-antigenic, suggesting it is unlikely to provoke adverse immune responses—a characteristic that is highly desirable for therapeutic applications. Furthermore, ToxinPred evaluations confirmed its non-toxic nature, substantively reinforcing the overall safety profile of the chimeric protein. Collectively, these in silico predictions support a highly favorable safety profile, providing a robust rationale for future experimental and in vivo studies to validate the therapeutic applicability of this chimeric construct.

Expression prediction in E.coli

Computational analysis utilizing SoluProt 1.0 predicted a solubility score of 0.578. By exceeding the predefined threshold value of 0.5, this result strongly suggests a favorable probability of soluble protein expression within an E. coli host system. While this in silico prediction indicates high feasibility for successful expression, it does not definitively confirm soluble production yields under in vitro experimental conditions. Consequently, rigorous empirical validation remains imperative to confirm the actual expression levels and solubility profile of the construct within a bacterial system.

Docking analysis

The docking study between the magainin 2-IL-24 fusion protein and its heterodimeric receptor provided critical insights into their binding affinity and structural interaction. By employing ClusPro 2.0 for protein-protein docking, the mechanism of interaction between the fusion protein and its receptor was elucidated, yielding a rigorous assessment of their binding affinity. The selection of the top docked structure was predicated on scoring criteria, including electrostatic energies and van der Waals attractions (Figure 3). Notably, the major interactions observed during protein-protein docking were driven by electrostatic and van der Waals forces, which yielded promising and reliable results36. The robust docking methodology inherent to ClusPro 2.0 ensures high accuracy in the binding affinity evaluation. The generation of ten distinct dock models, each with unique weighted scores, allowed for a comprehensive assessment of binding energetics. The optimized balanced model exhibited a center-weighted score of -868.9 and a minimum energy value of -898.2 kcal/mol, reflecting a highly stable and strong binding interaction (Table 2). These results signify the robust potential of the magainin 2-IL-24 fusion protein to interact effectively with the IL-24 heterodimeric receptor, which is a pivotal step in its development as a candidate for anti-cancer therapeutics. In this analysis, the center score represents the average energy of surrounding structures, while the weighted lowest-energy score reflects the most stable conformation within that specific cluster.

Residue interactions analysis

In recent years, advancements in molecular modeling software and computational tools have greatly accelerated drug discovery efforts, particularly in elucidating the complex protein-protein interactions (PPIs) that are relevant to various diseases, including cancer. Historically, PPIs were often considered "undruggable" due to their challenging binding interfaces; thus, the investigation of such interactions represents a critical frontier in modern pharmacological research37. Computational analysis of the docked complex, facilitated by the PDBePISA and PDBsum servers, provided valuable insights into the binding interface, which was characterized by the presence of 17 intermolecular interactions, encompassing 6 salt bridges and 11 hydrogen bonds. Specifically, regarding the fusion protein's interaction with the IL-22R1 subunit, 11 distinct bonds were predicted, comprising 3 salt bridges and 8 hydrogen bonds. This interaction is particularly robust, with interfacial residues spanning an area of 695.2 Ų and contributing to a stabilization energy of -6.5 kcal/mol. Conversely, the interaction between the fusion protein and the IL-20R2 subunit resulted in 6 bonds, consisting of 3 salt bridges and 3 hydrogen bonds. This interaction spans an interface of 679.8 Ų and is associated with a stabilization energy of -0.8 kcal/mol (Figures 3 and 4). Moreover, Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) analysis revealed a substantial binding free energy of -83.28 kcal/mol, indicative of a highly favorable and stable binding affinity between the fusion protein and the receptor subunits. Table 3 illustrates the estimated per-residue energy contributions that are crucial for complex formation.

Molecular dynamics simulation

The The MD simulation graphs for the docked complex provide critical insights into the stability and dynamic behavior of the protein-receptor complex over a 100 ns trajectory. The RMSD graph (Figure 5A) illustrates the structural stability of the protein during the simulation. Initially, a sharp increase in RMSD is observed, which stabilizes around 20 ns and remains relatively constant, exhibiting only slight fluctuations around 4-5 Å for the remainder of the simulation. This suggests that the protein has reached a stable conformation following the initial equilibration phase, indicating a robust interaction with the receptor.

The Rg (radius of gyration) graph (Figure 5B) delineates the compactness of the protein structure over time. The Rg values demonstrate minor fluctuations around 29 Å throughout the simulation, indicating that the fusion protein maintains a consistent and compact conformation. The lack of significant changes in Rg suggests that the protein does not undergo major conformational shifts and remains stable throughout the simulation.

The RMSF (Root Mean Square Fluctuation) graph (Figure 5C) characterizes the flexibility of individual residues within the protein. Most residues exhibit relatively low fluctuations (approximately 2 Å), indicating that the majority of the protein remains stable. However, a noticeable peak is evident around residue 400, suggesting a region of increased flexibility. This could indicate the presence of a loop or unstructured region within the protein that contributes to localized flexibility, which might be functionally significant for the protein’s interaction with the receptor.

Overall, the MD simulation results indicate that the magainin 2-IL24 fusion protein forms a stable complex with its cognate receptor, maintaining structural integrity and consistent interactions throughout the simulation period. The observed stability and minor fluctuations reinforce the potential efficacy of this fusion protein as a therapeutic agent for targeting breast cancer cells.