Targeted drug delivery to the thrombus by fusing streptokinase with a fibrin-binding peptide (CREKA): an in silico study

Abstract

Aim: Streptokinase has poor selectivity and provokes the immune response. In this study, we used in silico studies to design a fusion protein to achieve targeted delivery to the thrombus. Materials & methods: Streptokinase was analyzed computationally for mapping. The fusion protein modeling and quality assessment were carried out on several servers. The enzymatic activity and the stability of the fusion protein and its complex with plasminogen were assessed through molecular docking analysis and molecular dynamics simulation respectively. Results: Physicochemical properties analysis, protein quality assessments, protein–protein docking and molecular dynamics simulations predicted that the designed fusion protein is functionally active. Conclusion: Our results showed that this fusion protein might be a prospective candidate as a novel thrombolytic agent with better selectivity.

Plain language summary

Summary points.

• Thrombolytic agents are categorized into two classes based on their mechanisms: eukaryotic serine proteases and prokaryotic proteins like streptokinase (SK), which forms binary complexes with plasmin(ogen).

• Current SK versions lack selectivity, leading to non-specific activation of fibrin and systemic hemorrhage, highlighting the need for targeted drug delivery to thrombus sites.

• Peptides offer promise for fibrin targeting due to their small size, ease of manufacturing and decreased immunogenicity, with CREKA (Cys-Arg-Glu-Lys-Ala) specifically identifying fibrin–fibronectin complexes.

• In silico research presents a valuable strategy for molecular design and optimization, allowing for the evaluation of fusion proteins like CREKA-SK for targeted thrombolysis.

• This study aims to design and assess the efficacy of a fusion protein comprising CREKA and SK for targeted thrombolysis, leveraging in silico techniques for modeling and quality assessment of SK's 3D structure, docking simulations with its natural target, and molecular dynamics (MD) simulations to assess stability and flexibility.

• Extensive mapping of SK was conducted, incorporating experimental data on its structural domains and interaction sites. This mapping provided critical insights into identifying regions suitable for fusion with the fibrin-binding peptide (CREKA).

• Molecular modeling and analysis revealed the successful fusion of CREKA peptide with SK at region 302, with computational assessments confirming the stability and integrity of the fused protein.

• The development of targeted thrombolytic agents like CREKA-SK holds promise for improving clinical responses in thrombosis management, potentially reducing adverse effects associated with non-specific fibrinolysis.

Thrombosis is primarily caused by two key processes: platelet recruitment and fibrin formation. Fibrin remains on the surface as thrombi spontaneously detach or disintegrate. The primary approach in the short-term clinical management of blood clotting disorders is the breakdown of fibrin clots using thrombolytic (fibrinolytic) drugs. Blood clot lysis is brought by the activation of plasminogen (Plg) to plasmin (Pm), which dissolves the fibrin clot into soluble peptides for phagocytic elimination. Fibrinolytic agents are categorized into two classes, each with its different mechanism of action. The first, including urokinase and tissue plasminogen activator, is eukaryote serine proteases, which cleave the Plg 561–562 bond to convert it to Pm. The second group, such as streptokinase (SK) and staphylokinase, includes prokaryotic proteins with a high affinity for plasmin(ogen) to create a binary complex that causes a conformational change and a catalytically active site in the partner molecule, allowing it to convert another molecule of Plg (substrate-Plg) to plasmin [1,2].

Compared with other thrombolytic agents, streptokinase has a long half-life, which confers SK more opportunity to meet the clot. Non-specific thrombolytics, such as SK and urokinase, can produce plasmin in the bloodstream, which can degrade fibrinogen and factor VIII rather than fibrin. α2-antiplasmin or α2-macroglobulin stops this process quickly. However, SK is resistant to any kind of α2-antiplasmin among all thrombolytic agents. So, plasminemia and fibrinogenolysis are frequently associated with severe bleeding problems. As a result, in the treatment of cardiovascular disorders, plasminogen activators that are selective for fibrin are favored. Besides, as a bacterial protein, SK elicits an immunological response by provoking neutralizing antibodies [3].

SK is a 47-kDa extracellular protein of 414 amino acids that is released by different strains of β-hemolytic streptococci with different structure. Still, the sole version used as a medication is acquired from group C streptococci [4]. SK is not catalytically active; instead, it binds to plasminogen and interacts with it to form an active enzyme. Due to its cost–effectiveness, SK, which is on the World Health Organization (WHO) Model List of Essential Medicines (EML), is critical in the treatment of clotting disorders especially in economically developing countries [5,6].

One of the major issues with the current SK version is non-selectivity and independent activation of fibrin, which might result in systemic hemorrhage. To improve treatment efficacy and limit adverse effects, drugs should be targeted to the site of action, in other words, the thrombus [6,7]. Review of experimental studies shows that many studies have been done on increasing the SK half-life and enhancing its specificity for fibrin. However, the necessity for improving SK is still emphasized due to its crucial role in therapy.

Targeting fibrin is a promising strategy for thrombosis monitoring and treatment. Fibrin-containing blood clots have long been used as a target for the site-specific administration of imaging agents and anticlotting medications to thrombi. Anticoagulants administration into clotting veins have been demonstrated to lessen clot formation and growth while lowering the probability of systemic side effects [8].

Peptides have become a popular alternative to antibodies regarding fibrin targeting because of their smaller size, ease of manufacturing, lower cost, decreased immunogenicity and long-term stability [9]. In recent years, the peptide Cys-Arg-Glu-Lys-Ala (CREKA), which specifically detects fibrin–fibronectin complexes, has been utilized in the development of a self-amplifying nanoparticle delivery system [10]. CREKA was selected through in vivo phage display peptide libraries targeting fibrin–fibronectin complexes in diverse pathological conditions such as cancer, atherosclerosis and microthrombosis. The receptor responsible for CREKA binding has been identified as type IV collagen within the vascular basement membrane. Despite extensive investigations into CREKA's tumor-targeting properties, knowledge regarding its molecular target remains limited to screening within in vivo phage libraries [11,12].

In the domain of molecular design and optimization, in silico studies have gained popularity for their cost-effectiveness and simplicity, bypassing the need for human or animal subjects, or cell culture. However, real-world validation remains essential. This study aims to utilize computational methods and algorithms, including protein quality assessments, protein–protein docking and molecular dynamics simulation, to develop a fusion protein of a fibrin-binding peptide (CREKA) and SK, with the goal of creating a blood clot-targeted enzyme for improved clinical outcomes.

Materials & methods

Sequence & 3D structure retrieval

The protein's amino acid sequence was elicited from UniProt Knowledgebase (UniProtKB) database at http://www.uniprot.org in FASTA format and was used as a reference sequence for further analyses. The Protein Data Bank (PDB) file of the 3D structure was downloaded from RCSB PDB webserver at https://www.rcsb.org. A comparison between the sequences from UniProt and RSCB PDB was made by Clustal Omega at https://www.ebi.ac.uk/Tools/msa/clustalo. A homology modeling was done through SWISS-MODEL at https://swissmodel.expasy.org to unify the sequences mismatches and model missing residues of the primary PDB structure. SWISS-MODEL is a server for automated comparative modeling of tertiary protein structures. The new model was refined by ModRefiner from Zhang Lab at https://zhanggroup.org/ModRefiner.

Protein mapping

Positions embracing helix, beta strand, and turn were extracted from UniProtKB. Moreover, the structure modeled in previous step was performed in PDBsum webserver at http://www.ebi.ac.uk/thornton-srv/databases/pdbsum to obtain the second structure diagram. Exposed and highly exposed residues of the protein structure were defined by Swiss-PdbViewer (SPDBV) software (available at https://www.spdbv.unil.ch). The protein–protein interactions of the 1BML structure were observed through Protein Interactions Calculator (PIC) server at https://pic.mbu.iisc.ernet.in. Also, protein secondary structure and the surface exposed residues of the streptokinase, and residues contribute to the interaction with partner and substrate plasminogen were elicited from experimental studies [3,4,13–23]. This dataset was used to locate the appropriate fusion site and, as a reference to choose the most similar model to the native SK in further modeling.

3D structures prediction & validation

Whereas the SWISS-MODEL only modeled the medial missing residues, not the terminal ones, a complete 3D structure model was required. Seven models were built by using modeling servers, namely, Iterative Threading ASSEmbly Refinement (I-TASSER; at https://zhanggroup.org/I-TASSER), Protein Homology/AnalogY Recognition Engine (Phyre2; at http://www.sbg.bio.ic.ac.uk), and QUARK server (at https://zhanggroup.org/QUARK). Five models from I-TASSER and a model from Phyre2 were acquired. Also, since no templates were found for the 11 N-terminus and 42 C-terminus residues, their 3D structures were predicted separately through QUARK server which uses ab initio algorithms. The two modeled fragments were then refined by ModRefiner and assembled by ab initio Domain Assembly (AIDA) server (at https://aida.godziklab.org) with the previously modeled fragment by SWISS-MODEL. The predicted models were refined by ModRefiner.

Structure evaluation was the most important component of structure prediction. The second structure of all seven models was visualized through Chimera software and rigorously compared with data from UniProtKB and the experimental data. Models' quality assessment was done by ModFOLD8 at http://www.reading.ac.uk/bioinf/ModFOLD. Two most similar models to the native SK were chosen. The quality of the 3D structures was evaluated by ERRAT program from the SAVES online server, at https://saves.mbi.ucla.edu. Qualitative Model Energy ANalysis (QMEAN) scores were estimated by QMEANDisCo from SWISS-MODEL server at https://swissmodel.expasy.org/qmean. Finally, MolProbity, which may be found at https://www.molprobity.biochem.duke.edu, identified the Ramachandran favored residues for the anticipated structures.

Docking analysis

To perform the docking of the candidate models with plasminogen, chain A of the 1BML structure (Plg catalytic domain) was extracted by Chimera. The docking between the two SK models and the catalytic domain of plasminogen was done using ZDOCK 3.0.2 at https://www.zdock.umassmed.edu which had consistent success in the international protein–protein docking experiment, Critical Assessment of Predicted Interactions (CAPRI) through the years [24–27].

Fibrin-binding peptide fusion & validation

Protein fusion

The fibrin-binding pentapeptide (CREKA) was submitted to the QUARK server for structure prediction. A cleavage was made in the SK fusion site determined in step 2 by the Chimera software. The two SK fragments and the 3D-modeled CREKA were assembled by AIDA and eventually refined by ModRefiner to make fibrin-binding peptide fused SK (FBP-fused SK).

Validation

Several physicochemical properties of wild type (WT) and fused SK sequences were analyzed using ProtParam at https://www.web.expasy.org/protparam. Surface exposed residues of the fused SK were defined by Swiss-PdbViewer software. Second structure diagram was provided by PDBsum. The quality of the 3D structure was obtained through the ERRAT program, ModFOLD8 server, QMEANDisCo and the Ramachandran plot from PROCHECK program, provided by the SAVES webserver. Molecular docking has been carried out using ZDOCK. Interactions of docked model were observed using PIC.

Molecular dynamics simulation

Before conducting protein fusion, the preferred SK model was used as input for the molecular dynamics (MD) simulation. The next step involved performing MD simulations on the FBP-fused SK and the complex of the FBP-fused SK and Plg catalytic domain. GROMACS 2020 was used to carry out the MD simulation and the subsequent analyses. On a LINUX cluster, the Amber ff99SB-ILDN force field and SPC216 water model was used for the simulation and energy minimization. Through the steepest descent integrator, the system's energy consumption was reduced to a minimum of 50,000 cycles and a maximum of 1000.0 kJ/mol/nm. The system was further balanced using ensembles of NPT and NVT. Using Parrinello-Rahman type pressure coupling with a coupling coefficient of p = 2.0 ps and V-rescale type temperature coupling with a coupling coefficient of p = 0.1 ps, the temperature and pressure were maintained at reference values (T = 300 K, p = 1 bar). Each MD production took 100 ns to complete.

To resolve the equilibrium time range, the root mean-square deviation (RMSD) was calculated for the protein C-alpha atoms throughout the MD simulation with reference to the initial frame. Moreover, root mean-square fluctuation (RMSF) for the C-alpha atoms and the radius of gyration (Rg) of the molecule was computed.

Results

Sequence & 3D structure retrieval

There was only one tertiary structure in current databases obtained from x-ray crystallography (PDB ID: 1BML), representing SK-plasminogen interaction. The SK amino acid sequence was elicited from UniProtKB (accession No. P00779), which was identical with SK amino acid sequence from DRUGBANK (at https://go.drugbank.com) accession number DB00086. The sequence was saved in FASTA format and performed with Clustal Omega to compare with the SK from the PDB file. There were five mismatches, including residues 71, 210, 244, 253 and 303.

Part of the SK structure, i.e., the region between the protein residues 12 and 372, has been experimentally determined by x-ray crystallography (PDB ID: 1BML). There were three missing residues in the structure from 46 to 70, 175 to 181, and 252 to 262. SWISS-MODEL homology modeling server was used to build the missing residues and replace the mismatches with the sequence from UniProtKB. The model was refined by ModRefiner and will be referred to as ‘model 0’ (Figure 1).

Figure 1.

Streptokinase 3D structures visualized by the Chimera software.

(A) Streptokinase (SK) from 1BML, (B) model 0; sequence is obtained from UNIPROT and 1BML is used as template for homology modeling. Modeling has been done only in area 12 to 372, (C) model 1 (by I-TASSER server), (D) model 2 (by QUARK server).

Protein mapping & fusion site determination

From the information inferred from a combination of experimental and computational evidence, obtained from UniProtKB server, there are several loops in the regions 27–29, 256–258, 302–304 and 329–332. Model 0 secondary structure diagram acquired by PDBsum revealed hairpin loops in the regions 27–31, 112–114, 169–181, 227–232, 256–259, 302–306 and 357–362. Also, 250–loop, 170–loop, a loop in the region 88–93 and coiled coil region of 321–338 were obtained from experimental studies [3,15,20]. Since the fusion site must be surface exposed while simultaneously avoiding interaction sites and structural areas, the region 302–304 was chosen as the FBP insertion site (Supplementary Figure 1–3).

Model prediction & analysis

For SK 3D structure prediction, seven models were acquired from the previously named modeling servers. All models were visualized and analyzed by Chimera, and second structures were compared with the second structure data retrieved from UniProt and experimental studies. Two models from I-TASSER and QUARK had the most coincidence with the UniProt and experimental data, and were chosen for further analysis (Supplementary Table 1). The two models will be noted as model 1 and model 2, respectively (Figure 1).

Overall 3D structure quality was assessed for the models by ModFOLD8 demonstrating global model quality score of0.51 and 0.50 for model 1 and 2, respectively. For the validation of the selected models, they were submitted to MolProbity for structural quality assessment. The psi and phi torsion positions were predicted with the Ramachandran plot. The analyses of MolProbity revealed 87.1% residues in the favored region and 2.9% in outlier regions for model 1 and 93.9% residues in the favored region and 0.7% in outlier regions for model 2, indicating the superior quality of both models 1 and 2. SAVES server predicted ERRAT score of 75.2% for model 1 and 76.1% for model 2. QMEAN was retrieved by the QMEANDisCo method of the SWISS-MODEL server, which suggested the scores of 0.61 and 0.66 for models 1 and 2, respectively.

Protein–protein docking

PyMOL software (Schrödinger, Inc) was used to show the results of dockings between two SK models and the Plg catalytic domain (Figure 3). Results of the docking between model 1 with the catalytic domain almost coincided with the reference data. Therefore, Model 1 was selected as the reference model to conduct the fusion.

Figure 3.

The streptokinase models docking with catalytic domain of plasminogen.

(A) Docking of model 1 (blue) with plasminogen (Plg) catalytic domain (green), Conducted by ZDOCK server and visualized by PyMOL software. (B) Docking of fibrin-binding-peptide-fused streptokinase (FBP-fused SK) model with catalytic domain of Plg.

FBP-fused model validation & docking

Physicochemical properties of the WT-SK and FBP-fused SK sequence were measured by ProtParam (Table 1).

Table 1.

Physicochemical properties of the wild-type streptokinase and fused streptokinase sequence calculated by ProtParam webserver.

Property	WT-SK	Fused SK
Molecular weight	47286.76	47874.45
Theoretical pI	5.12	5.19
Extinction coefficients	38280	38280
Estimated half-life (mammalian reticulocytes, in vitro)	20 h	20 h
Instability index	23.61	23.60
Aliphatic index	79.59	78.88
GRAVY	-0.728	-0.738

There is no statically significant difference observed between the two groups. The instability index provides a method to estimate the stability of the protein in a test tube. A protein with an instability score less than 40 is predicted to be stable. The aliphatic index of a protein is described as the relative volume occupied by amino acids which have an aliphatic side chain in their structure.

SK: Streptokinase; WT: Wild-type.

The stereochemical quality and accuracy of the predicted fusion model were analyzed using ERRAT by SAVES, QMEAN by SWISS-MODEL, and Ramachandran plot generated by PROCHECK. Second structure was compared with the reference data, which showed a good coincidence. Ramachandran's analysis estimated a score of 80% in most favored regions and 1.6% in forbidden region. The global model quality score of 0.43 was predicted by ModFOLD8. ERRAT showed an overall quality score of 73.23%. Moreover, SWISS-MODEL calculated a QMEAN score of 0.62 by the QMEANDisCo algorithm. In the final step, docking was done between the fused model and catalytic domain of Plg. Compared with the 1BML and experimental studies, the result was acceptable, and the critical interaction sites between SK and partner plasminogen were conserved.

Molecular dynamics simulation

RMSD analyses of the backbone atoms were calculated separately for the WT-SK, fused SK and the complex of FBP-fused SK and Plg catalytic domain. The complex RMSD value rose quickly from zero for the initial frame, and subsequently reached an approximately steady state after nearly 10 ns (Figure 4). The RMSF analysis also allowed us to identify the more flexible parts of the proteins and to compare the flexibility of different parts of the system relative to each other.

Figure 4.

C-alpha Root Mean-Square Deviation plot of wild-type streptokinase (blue), fibrin-binding-peptide-fused streptokinase (red) and complex of fibrin-binding-peptide-fused streptokinase and plasminogen catalytic domain (black).

The graphs are produced by GROMACS and represent the Root Mean-Square Deviation of the computational modeled structures, validating their structural stability.

SK: Streptokinase; WT: Wild-type.

Discussion

According to previous studies, CREKA peptide can be a good candidate for fusion due to its fibrin affinity [10]. The CREKA peptide, which has been produced using in vivo phage display approach, exhibited interactions with fibrin clots with a favorable targeting ability to fibrin [28,29]. In this regard, SK-CREKA fusion protein can offer an opportunity for improving the SK safety and therapeutic efficacy. Moreover, the effective dose and administration frequency may be lowered as a result of SK-CREKA fusion. The CREKA was fused from region 302 of the SK.

SK is composed of three domains: alpha (amino acid region 1–146), beta (147–290) and gamma (291–414) [13]. To locate the FBP fusion site, the exposed non-structural areas were selected. According to experimental studies, amino acids involved in protein–protein interactions are among these areas: 1 to 59 [3,15,19], 88 to 97 [3], 158 to 219, 234 to 293 [3,15,18], 314 to 387 [17,21,22], and 414 [3,23], and according to the information obtained from 1BML, in addition to the mentioned residues, amino acids number 87, 112, 134, 220 and 311 are also involved in interactions (Supplementary Figure 3). Accordingly, exposure areas not involved in the interactions or the structural domains are fragments 60–66, 118–122, and 147–157 [4,13,30]. In addition, important loops of the molecule are fragments 88–97, 170–182, 256–258, 302–304 and 329–332 [3,20,31]. Among these, the loop located in the 302–304 region does not interact with plasminogen (partner or substrate). Considering all these aspects, the fibrin-binding peptide was fused from region 302. Examining the WT SK sequence and the fused one by ProtParam server, there was not much difference between important physicochemical characteristics (Table 1). Regarding GRAVY, negative values indicate hydrophilicity, which is necessary for the molecule to spread and move in the bloodstream [32–34]. The conformational space phi and psi of amino acid residues obtained through the Ramachandran plots determine the integrity and validity of a protein 3D structure [35]. There were only a few outliers in the Ramachandran plot for model 1 and the fused-SK model, indicating that the quality of predicted structures is remarkably acceptable. ERRAT verifies the structure of the protein by detecting local errors based on the statistics of non-bonded atomic interactions and comparing them with highly refined structures to suggest an overall quality factor. ERRAT scores for model 1 and the fused-SK were 75.2 and 73.2%, respectively. A model with an score greater than 50% is considered as a high-quality model [36]. It is important that the ERRAT plot did not change significantly after the fusion. The global model quality score of both model 1 and the fused-SK model calculated by the ModFOLD8 were in the acceptable range presenting high similarity to the native structure [37]. There was also no significant difference in global QMEAN scores between the model 1 and the fused-SK model, which were estimated by QMEANDisCo method in the SWISS-MODEL web server. Importantly, there was no need to add a linker because there was no noticeable change in the final conformation of the molecule after the addition of the CREKA to the SK sequence, and the validation results of the fused-SK were likewise acceptable.

Reviewing experimental studies, the areas associated with partner plasminogen binding to SK were identified as residues 1 [3], 17–26 [13], 37–51 [15,16], 88–97, 158–169, 183–219, 234–253, 274–293, 314–387 [15,21,22] and 414 [3,23] (Supplementary Figure 3). Active plasmin is generated after the cleavage of the 561–562 bond in the Plg. But in the SK.PLG complex, a conformational change instead of this cleavage results in the production of the active molecule. Two assumptions have been proposed to justify this phenomenon. First, binding of residues 314–342 SK gamma domain to the ‘autolysis loop’ region of Plg (692–695) occurs near the activation cleavage site, which causes a conformational change and a salt bridge between K698 and D740, leading to active site formation. Second, the insertion of I1 into the active site of Plg, which forms a crucial salt bridge with D740, results in the active conformation of Plg [14]. Due to the absence of residues 1–11 in 1BML, any interaction between 1 and 740 cannot be seen. Nevertheless, other interactions, including 314–342 to the Plg autolysis loop, are entirely consistent with the experimental data. Among these, the interaction of residues 314 with 626, 315 with 625, 324 with 694 and 622, 326 with 693, 332 with 623, 333 with 625, 332 with 623, 333 with 625 and 337 with 622 can be mentioned. The mentioned bonds and many other interaction sites between SK and partner Plg were used as reference data in the docking analysis of SK models 1 and 2 with Plg catalytic domain and then fused SK with the Plg catalytic domain. In the presence of SK, Plg changes to the extend form where all the kringles and the catalytic domain are exposed and allows the three domains of SK to associate with the catalytic domain of Plg. Affinity toward Plg is enhanced by the C-terminal residue 414, which interacts with the Kringle 4 (K4) of the Plg. The next step is forming an active site in Plg, which makes the whole complex enzymatically active. Since only the Plg catalytic domain was used for docking, the interaction between 414 and K4 Plg did not occur because of the absence of the K4. Also, the interactions of initial residues did not occur, possibly due to a change in the spatial orientation of the two molecules. Another possibility can be attributed to the difference in the predicted structure of N-terminal amino acids with the WT-SK model. However, due to complete coincidence of other interactions, especially the 314–342 area with the findings of 1BML and experimental studies, the docking result is considered reliable (Figure 6).

Figure 6.

Left: chains A (catalytic domain of plasminogen) and B (fused streptokinase) residues interactions across interface; depicted by PDBsum webserver. Right: schematic diagram of interactions between protein chains (red: salt bridges, blue: hydrogen bonds, orange: non-bonded contacts).

The area of each circle is proportional to the surface area of the corresponding protein chain. The extent of the interface region on each chain is represented by the black wedge whose size implies the interface surface area.

MD simulation results confirmed the stability of the fused model structure. The root mean square deviation (RMSD) analysis of the wild-type (WT) SK, fused SK, and the complex revealed differing timescales for attaining stability, with the systems requiring approximately 50, 25 and 12 ns, respectively, to approach stable RMSD values (Figure 4). Notably, the Complex exhibited more consistent RMSD values, fluctuating within a narrower range of 0.4 to 0.6 ns, indicative of a higher degree of stability. This observation suggests successful attachment of the catalytic domain of Plg to the fused SK, thereby enhancing the stability of the resultant complex. Also, the MD simulations indicated that the fused CREKA has a good fluctuation in the whole protein structure, which approves the random coil structure of the fragment. As indicated in Figures 2 & 5, lower RMS fluctuation values for amino acid fragments comprising helix or beta-strand indicate rigidity in those locations, including 17–26, 134–145, 196–210 and 267–278. In contrast, a higher RMSF value is predicted for more flexible fragments consisting random coil structures, which are commonly surface exposed and involved in protein–protein interactions, e.g., fragments 44–59, 254–261, 304–308 and 326–342 (Supplementary Figures 2 & 3) [38,39]. The number of residues participating in α-helix, β-sheet and coil structures was almost constant during the period of MD simulation. As shown in Figure 5, the regions with high RMSF value in fused SK graph seem to be more static and less flexible in the complex graph, indicating that those areas are associated with the partner Plg interaction [40].

Figure 2.

Fibrin-binding-peptide-fused streptokinase secondary structure provided by PDBsum (Helices labelled H1, H2, … and strands by their sheets A, B, …; β: beta turn; γ: gamma turn; ﬤ: beta hairpin).

Figure 5.

Comparison of the fused streptokinase and the complex of fused streptokinase and plasminogen catalytic domain root-mean-square fluctuation graphs.

Regions exhibiting elevated root-mean-square fluctuation values within the fused streptokinase graph appear to demonstrate reduced flexibility and increased stability within the complex graph, suggesting a correlation with partner plasminogen interaction.

The stability and dynamics observed in MD simulations support our choice of region 302 as the fusion site, which is non-structural and does not compromise SK interaction sites. These findings align with regions implicated in partner Plg binding from experimental studies, suggesting the potential of the fusion protein for improved therapeutic outcomes, including enhanced clot targeting and fibrinolysis, with reduced risk of off-target effects. These results advance our understanding of fibrinolytic activity mechanisms and offer promise for safer fibrinolytic therapy development.

Several limitations were encountered during the course of this study. First, the structural model of SK derived from the 1BML structure exhibited significant gaps, with 96 missing residues and numerous mismatches with the sequence obtained from the UniProt database. As a result, structural modeling of SK was necessitated using prediction software, which may have introduced inaccuracies into the model. Second, resource constraints limited the duration of MD simulations for each system to 100 ns, potentially limiting the thorough exploration of dynamics and restricting the ability to repeat simulations for result validation. Additionally, the exact binding site of CREKA to its target has not yet been determined, precluding the accurate estimation of the affinity of fused SK to fibrin, and simulation of the docking of fused SK with fibrin. These limitations underscore the need for further research to address these challenges and refine our understanding of the therapeutic potential of the fused protein.

Furthermore, to provide a better understanding of the fusion protein's potential benefits and position within the current therapeutic landscape, comparative analysis of the fusion protein's efficacy and selectivity against other existing thrombolytic agents in a similar in-silico environment could be a potential direction for future research.

Conclusion

Despite some benefits of streptokinase, poor selectivity and high immunogenicity can be mentioned as its disadvantages. CREKA was fused to streptokinase to alleviate these problems via in silico studies. The results of physicochemical properties analysis, protein quality assessment, protein–protein docking and molecular dynamics simulations showed that the designed fusion protein could be a proper candidate as a novel thrombolytic agent with better clinical responses. Further in vitro and in vivo studies should be done to confirm this claim.

Funding Statement

This work was financially supported by the Shiraz University of Medical Sciences (Grant no. 25161).

Author contributions

MS Hajizade: conceptualization, data curation, formal analysis, writing (original draft). MJ Raee: Project administration. SN Faraji: Methodology. F Farvadi: Investigation. M Kabiri: Visualization, Writing (review and editing). S Eskandari: Software. AM Tamaddon: Validation.

Financial disclosure

This work was financially supported by the Shiraz University of Medical Sciences (Grant no. 25161). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, stock ownership or options and expert testimony.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.