Introduction

Transforming Growth Factor β (TGFβ) is a polypeptide secreted by cells with the capability of inducing oncogenic transformation in non-cancerous cells1. Its pivotal role in human diseases, particularly cancer, stems from its influence on tissue balance. TGFβ activation occurs during chronic inflammation and wound healing, triggered from its deposition in significant amounts within the extracellular matrix (ECM) in an inactive state2. At the cellular level, TGFβ regulates various biological processes under both normal and pathological conditions, including cell cycle regulation, apoptosis, epithelial to mesenchymal transition (EMT), and control over the extracellular matrix (ECM)3. On a broader scale, TGFβ plays a regulatory role in tissue and organ functions, impacting the differentiation and immune responses of B and T lymphocytes, thus influencing the inflammatory cascade linked to cancer progression. Additionally, it governs critical tissue interactions pivotal during embryonic organogenesis and cancer advancement2.

While animal studies and transcriptional analyses haven’t directly linked changes in TGFβ3 expression to breast cancer progression, various clinical studies have highlighted an association between increased TGFβ3 expression and breast tumorigenesis. Li et al.4 conducted a study assessing plasma TGF-β1 and -β3 levels in 80 untreated breast cancer patients, where 14 had lymph node metastases. Elevated TGFβ3 levels and TGFβ3/receptor complexes were observed in correlation with lymph node metastases. Subsequent research involving 153 invasive breast cancer tissue samples further supported this finding5, showing intense immunostaining for TGFβ3 in these samples. The upregulation of TGFβ3 significantly correlated with decreased overall survival (p = 0.0204), particularly notable when associated with lymph node metastasis. Additional evidence linking TGFβ3 to breast carcinoma metastasis comes from Amatschek et al.‘s study, which noted elevated TGFβ3 gene expression in patients with shorter survival times from breast cancer6. Soufla et al. analyzed 25 breast carcinomas and adjacent normal tissue specimens, finding significantly higher TGFβ3 transcript levels in cancer specimens than in normal tissues7. Their study involved evaluating mRNA expression profiles of Vascular Endothelial Growth Factor (VEGF), Fibroblast Growth Factor 2 (FGF2), TGF-β1, -β2, and -β3, as well as TbRI, TbRII, and TbRIII in 25 breast cancer samples, primarily infiltrating ductal carcinoma of histological grade G37, while all three TGF-β isoforms showed increased levels in cancer specimens compared to controls, only FGF2 and TGFβ3 displayed statistically significant elevations in tumors compared to adjacent normal breast tissue (p = 0.031 and p = 0.043, respectively). Extensive preclinical research has highlighted that targeting TGFβ presents a promising approach for achieving antitumor effects. Building on this premise, various classes of TGFβ inhibitors have been developed and evaluated in clinical trials, including monoclonal, neutralizing, and bifunctional antibodies, antisense oligonucleotides, TGFβ-based vaccines, and receptor kinase inhibitors. Over 15 years have passed since the initiation of the first clinical trial involving an anti-TGFβ agent. However, despite encouraging preclinical results, translating the foundational knowledge of TGFβ’s oncogenic role into clinical applications has proven to be a slow and challenging process8.

Based on these studies, our focus was directed towards TGFβ3 as a potential target for identifying inhibitors from natural sources. Computational techniques play a crucial role in the initial stages of drug discovery by leveraging modern technology to gain virtual insights into chemical systems9, thereby enhancing and supplementing experimental investigations. Among these techniques, molecular docking is a widely used in silico approach that predicts the binding interactions and modes of small molecules or macromolecules with a receptor10. Additionally, MD simulations have become increasingly significant in molecular biology and drug development11,12,13. These simulations provide detailed atomic-level insights into the dynamic behavior of proteins and other biomolecules over precise temporal scales13,14. Through molecular docking and dynamic simulations, our investigation highlights curcumin as a promising inhibitor against TGFβ3 in breast cancer. Previous research has extensively documented the properties and effects of curcumin15,16. Curcumin (1,7-bis-(4-hydroxy-3-methoxyphenyl)-hepta-1,6-diene-3,5-dione), also known as diferuloylmethane, stands as the primary constituent of turmeric, derived from the rhizome of Curcuma longa, a plant native to Southeast Asia and part of the Zingiberaceae family (ginger family)17. Turmeric contains curcuminoids, which include curcumin, demethoxycurcumin, and bisdemethoxycurcumin18. Curcumin, constituting approximately 2–5% of turmeric, not only provides the spice its characteristic yellow hue but also accounts for a significant portion of its therapeutic properties17,19. In addition to its use as a flavoring and coloring ingredient in food, turmeric has held a prominent place in Ayurvedic medicine due to its analgesic, antioxidant, antiseptic, antimalarial, and anti-inflammatory attributes20. Over the past two decades, extensive research has elucidated the significance of various functional groups in mediating the biological activities of curcumin. Key structural features contributing to its activity include the o-methoxyphenol group and methylenic hydrogen, which play pivotal roles in its antioxidant properties. These groups enable curcumin to donate electrons or hydrogen atoms to neutralize reactive oxygen species, thereby mitigating oxidative stress21. For centuries, curcumin has been a part of dietary supplements and is widely regarded as safe for consumption from a pharmacological perspective22.

Materials and methods

Receptor-ligand preparation and molecular docking

The crystal structure of TGFB3 in its three-dimensional conformation was acquired from the Protein Data Bank (https://www.rcsb.org/structure/1tgj) using the accession number 1TGJ. The TGFB3 protein structure consists of a sequence length of 112 amino acids. The three-dimensional structure of Curcumin was retrieved from the PubChem chemical database (https://pubchem.ncbi.nlm.nih.gov/compound/969516) using PubChem CID-969,516 in SDF file format. Curcumin, a natural compound found in the root of Curcuma longa, has a molecular weight of 368.4 g/mol. Moreover, for comparative interaction analysis, curcumin analogs and some FDA-approved drugs were obtained from the PubChem and Drug Bank databases, respectively (Table S1). Energy minimization of curcumin was done using Chem3D Pro 14.0 tool. The computational approach for molecular docking simulation was executed using the InstaDock standalone software23. The suitable file formats for both the protein and ligand to conduct molecular docking were prepared in PDBQT format through InstaDock, and a configuration file was generated, including the grid dimension of the protein for blind docking, having 47, 71, and 54 for X, Y, Z dimensions while − 14.922, 32.367, and 10.31 for centre X, Y, and Z respectively. The addition of non-polar hydrogen atoms was automated during the generation of PDBQT files for the receptor protein. Pymol24 and Discovery Studio visualizer25 tools were employed to visualize the docking results and analyse interactions among residues or atoms of the receptor protein and the ligand.

Additionally, the biological activity of the compounds was predicted, providing a valuable example of how chemical compounds can be studied prior to experimental validation. This was done using the freely accessible online version of PASS, which predicts the biological activity spectrum of the compounds. The biological activity spectrum of a substance consists of a list of potential bioactivity types, where the probability of activity (Pa) and the likelihood of inactivity (Pi) are both calculated26.

MD simulations

Molecular Dynamics (MD) simulation serves to analyse atomic motions within protein and protein-ligand systems27,28. The outcomes derived from the docking interaction between TGFB3, and the natural compound Curcumin were investigated using MD simulation studies. For simulating the structural coordinates of TGFB3 and its complexes with Curcumin, we employed GROMACS v 2022, a freely available MD simulation tool widely applied in discovery of drugs. The topologies of the receptor–ligand complexes were generated by parameterizing the compounds using the CHARMM General Force Field (CGenFF)29. The CGenFF server stands as a reliable web-based tool for topology generation of small compounds. Each system was positioned with a 10 Å distance from the cubic box center to the edges to allow solvation in the transferable intermolecular potential 3P (TIP3P) water model, employing the Charmm36-Jul2022 force field. Additionally, the neutralization process was done by introducing an appropriate quantity of counterions (Na + and Cl-)30. To eliminate possible steric hindrances between atoms, an energy minimization was conducted in the solvated system, utilizing 1500 steps of steepest descent followed by conjugate gradient methods. Two-step equilibration under periodic boundary settings was carried out for 1000 ps at constant volume, gradually heating from 0 to 300 K, and maintaining a pressure of 1 atm. Production simulation for all systems extended to 500 ns. Assessment of protein–ligand stability was performed by analysing the resulting data using GROMACS tools such as gmx rms, gmx rmsf, gmx gyrate, gmx sasa, gmx hbond, gmx do_dssp, gmx anaeig. Visual representations were created using qtgrace for plotting figures.

PCA and essential dynamics

The application of principal component analysis (PCA) and essential dynamics helps in reducing data dimensionality while preserving the dataset’s variability, employing a well-established mathematical algorithm31. PCA identifies principal components (PCs) representing directions with maximal data variance, condensing numerous variables into a few components, effectively portraying each sample32. This condensed representation allows for visual assessments of sample similarities, differences, and potential groupings. Additionally, PCA illuminates high-amplitude motion within MD trajectories. In our study, PCA and free energy landscape (FEL) analyses were utilized to explore conformational sampling, atomic motions, and stability within MD trajectories of both TGFB3 and the TGFB3-curcumin complex. The covariance matrix of atomic fluctuations was calculated, focusing on Cα atoms to ensure the analysis captured functionally relevant motions. The first few PCs that cumulatively captured major variance were considered for further analysis, therefore first two principal components were selected.

MM-PBSA analysis

MM-PBSA calculations facilitate the integration of high-throughput molecular dynamics (MD) simulations with the estimation of binding free energies for protein-ligand interactions. These calculations include contributions from interaction energies such as van der Waals, electrostatics, polar solvation, and overall binding energy. In this study, we utilized the gmx_mmpbsa33. The gmx_mmpbsa tool was executed in a using MD trajectory file of the protein-ligand complex alongside the required parameter file. The binding free energy components, including van der Waals, electrostatics, polar solvation, and SASA contributions, were determined using the mmpbsa analyser tool.

Limitations

In silico tools like molecular docking are extensively utilized for investigating structure-activity relationships of proteins and peptides, as well as for virtual screening to discover novel bioactive molecules. Despite their utility, molecular docking has limitations, such as restricted sampling of ligand and receptor conformations and the approximate scoring functions, which can sometimes give results that poorly correlate with experimental binding affinities. MDS on the other hand, offers a computational method to analyze the motion and equilibrium of molecules in various scientific disciplines. MDS provides detailed insights into the behavior of protein-ligand complexes at the atomic level with high temporal resolution. While MDS has proven effective in addressing biochemical challenges, it faces two significant limitations. First, the force fields (FFs) used in simulations are continuously refined to improve accuracy. Second, the high computational requirements of MDS often limit routine simulations to relatively short timescales, typically insufficient to achieve adequate sampling of conformational states.

Result and discussion

Molecular docking and interaction analysis

The 3D crystal structure of TGFβ3 was retrieved from the PDB database using the PDB ID 1TGJ. The three-dimensional structure of Curcumin was retrieved from the PubChem chemical database, utilizing PubChem CID-969,516 in SDF file format. Subsequently, molecular docking was conducted for both the receptor and ligand using InstaDock software. InstaDock is a freely available Graphical User Interface (GUI) based program specifically designed for efficient molecular docking. The binding of curcumin to TGFB3 resulted in a binding energy of -6.3 kcal/mol (Table S1). Moreover, the binding energy standard error (SE) was calculated as 0.0455 kcal/mol using a Python script, which provides a measure of the reliability of the binding affinity. The input for the calculation included the binding energies of all poses of curcumin with TGFB3 -6.3, -6.2, -6.2, -6.1, -6.1, -6.1, -6.0, -5.9, and − 5.9 kcal/mol. Interactions of all nine conformers were assessed (Figure S2), and the one position indicating a significant interaction between curcumin and TGFB3 was selected (Fig. 1A). For comparative interaction analysis, curcumin analogs and some FDA-approved drugs were obtained from the PubChem and Drug Bank databases, respectively. Among them, curcumin exhibited higher binding energy (Table S1) and stronger interactions (Figure S3). Furthermore, interaction analysis was performed using Discovery Studio Visualizer and PyMol software. Curcumin demonstrated crucial interactions, including hydrogen bonds with residues such as ASP323, ARG325, VAL333, HIS334, PRO336, LYS337, GLY393, and ARG394 (Fig. 1B & C). Additionally, other interactions such as Pi-Cation bonds involving residue LYS337, Alkyl, and Pi-Alkyl bonds involving residues PHE324, ARG325, and TYR391, as well as van der Waals interactions involving residues GLN326, LY3S31, TRP332, and VAL392 (Fig. S1B), were predicted. Notably, ARG325 and LYS337 each residue creating two bonds (Fig. S1). These findings strongly suggest that curcumin may be considered a potential inhibitor for TGFB3. In addition, structural-activity predictions were performed using the PASS online tool, which provides the probability of activity (Pa) and the probability of inactivity (Pi) values. The results, presented in Table S2, indicate that curcumin exhibits anticancer activity, as evidenced by its higher Pa value. These predicted results further highlight the potential of curcumin as an effective anticancer agent.

Fig. 1
figure 1

Representation of molecular docking result. (A) A cartoon view of TGFB3 protein in complex with curcumin. (B) Three-dimensional representation of TGFB3-curcumin complex showing interacting residues. (C) Conserved residues of TGFB3 protein in complex with curcumin in ball stick form.

Full size image

MD simulation

Molecular dynamics (MD) simulation is a prevalent method employed to explore the structural intricacies and dynamic behaviour of protein–ligand interactions27. This study focused on two systems, TGFB3-curcumin and apo TGFB3, subjected to 500 ns MD simulations. The analysis delved into the stability and dynamics of TGFB3 when bound to curcumin, assessing diverse systematic and structural parameters. To validate system equilibration before analysis, the potential energy of free TGFB3 and the TGFB3-curcumin complex was tracked. The observed consistent temperature fluctuations at 300 K across both systems affirm the stability and reliability of the MD simulations conducted. Notably, the average potential energy for free TGFB3 and the TGFB3-curcumin complex was measured such as -844,629 kJ/mol and − 590,782 kJ/mol, respectively, signifying the enhanced stability of the TGFB3-curcumin complex compared to the unbound TGFB3. Moreover, snapshot of both apo TGFB3 protein and TGFB3-curcumin complex at different time period was taken such as 0ns, 100ns, 200ns, 300ns, 400ns and 500ns (Figure S4&5). The result suggest that TGFB3-curcumin complex was at stable state at the end of MD simulation.

Structural dynamics and compactness

To delve deeper into characterizing the induced structural and conformational changes due to binding, a 500ns MD simulation was conducted. The chosen docking conformations of TGFB3 in complex with curcumin were investigated through a 500ns MD simulation, shedding light on the dynamic behaviour of the complex. This was assessed by calculating the backbone RMSD values of the protein over the simulation time (Fig. 2). The average RMSD for TGFB3 in its unbound state was found to be 0.35 nm, whereas in complex with curcumin, it was reduced to 0.33 nm (Table 1). The RMSD plot in Fig. 2A indicates that the TGFB3-curcumin complex trajectory remained low and stable up to 200 ns of simulation. A minor drift was observed between 200 ns and 300 ns. Meanwhile, the graph representing TGFB3 in its unbound state also exhibited a shift after 250 ns. Moreover, the distribution plot, displayed as a Probability Density Function (PDF), indicated bit higher RMSD values for TGFB3-curcumin complex (Fig. 2, lower panel). These RMSD results imply that the TGFB3-curcumin complex demonstrated greater stability throughout the 500 ns MD simulation compared to the unbound TGFB3 protein.

Table 1 Average parameters calculated post 200 ns MD simulation.
Full size table

The vibrations around the equilibrium aren’t random; they rely on the local structural flexibility. To assess the average residual fluctuation during the simulation, the root-mean-square fluctuation (RMSF) was plotted as a function of residue number for both the apo TGFB3 and the TGFB3-curcumin complex (Fig. 2B). The average RMSF values for the apo TGFB3 and TGFB3-curcumin complex were measured at 0.15 nm and 0.60 nm, respectively (Table 1). This prediction indicates that TGFB3 displayed an increase in residual fluctuation after curcumin binding, suggesting minimal stability in the complex. Additionally, the maximum RMSF values for the apo TGFB3 and TGFB3-curcumin complex were calculated at 0.55 nm and 1.27 nm, respectively, demonstrating an increment in RMSF values after curcumin interaction. Furthermore, the distribution plot of RMSF as a Probability Density Function (PDF) aligns with the fluctuation results (Fig. 2, Lower panel). Both the RMSF plot and the calculated values suggest that the TGFB3-curcumin complex shown less stability throughout the entire 500 ns simulation.

Fig. 2
figure 2

Graphical representation of MD simulation result. (A) RMSD plot indicating deviation of TGFB3 protein and TGFB3-curcumin complex. (B) RMSF plot indicating residual fluctuation of TGFB3 protein and TGFB3-curcumin complex. Lower panel indicating distribution plot of (C) RMSD and (D) RMSF.

Full size image

The Radius of Gyration (Rg) serves as a parameter associated with a protein’s tertiary structural volume, providing valuable insights into its stability within a biological system34. Typically, a protein with a higher Rg indicates less tightly packed structure. The average Rg value measured reflected stability in the TGFB3-curcumin complex. Comparing the average Rg values, the apo TGFB3 exhibited a value of 1.89 nm, whereas the TGFB3-curcumin complex displayed a slightly lower value of 1.86 nm (Table 1). The plot depicted in the Fig. 3A demonstrates a similar pattern between TGFB3 and the TGFB3-curcumin complex up to 200 ns. However, after that point, the graph representing the TGFB3-curcumin complex showed a noticeable shift and remained fluctuating until the end of the 300 ns simulation. After 300 ns time the complex plot remained lower than TGFB3 protein at the end of simulation. Moreover, the maximum Rg values recorded for the apo TGFB3 and TGFB3-curcumin complex were 1.97 nm and 2.04 nm, respectively. Additionally, the distribution plot of Rg represented as a Probability Density Function (PDF) (Fig. 3, Lower panel) further supports the prediction of a lower Rg in the TGFB3-curcumin complex. These Rg results suggest that the TGFB3-curcumin complex achieves enhanced stability and compactness compared to the apo TGFB3.

The Solvent Accessible Surface Area (SASA) represents the part of a protein that accessible for surrounding solvent molecules. SASA serves as a crucial determinant of a protein’s accessibility to the solvent, thereby indicating the protein’s unfolding during MD simulation35. The SASA plot depicted in the Fig. 3B illustrates the unfolding of the TGFB3 protein occurring around 50 ns. After 300 ns, the TGFB3 SASA showed an increase, signifying reduced stability, while the TGFB3-curcumin complex maintained a consistently steady SASA graph throughout the simulation except small drift between 200 ns to 300 ns, indicating higher stability. Comparing the average SASA values, the apo TGFB3 exhibited an average SASA value of 80.18 nm², whereas the TGFB3-curcumin complex displayed a lower value of 79.41 nm². Additionally, the maximum SASA values calculated for the apo TGFB3 and TGFB3-curcumin complex were 86.48 nm² and 92.89 nm², respectively. Furthermore, the distribution plot shown as a Probability Density Function (PDF) in the Fig. 3 also demonstrates that the SASA values of the TGFB3-curcumin complex consistently remained stable than those of the TGFB3 apo form. This result strongly suggests that the TGFB3-curcumin complex maintains a folded and stable conformation.

Fig. 3
figure 3

Graphical representation of MD simulation result. (A) Rg plot indicating stability and compactness of TGFB3 protein and TGFB3-curcumin complex. (B) SASA plot indicating solvent accessible surface part of TGFB3 protein and TGFB3-curcumin complex form. Lower panel indicating distribution plot of (C) Rg and (D) SASA.

Full size image

Dynamics of hydrogen bonds

Hydrogen bonds play a pivotal role in protein folding, structure, and molecular recognition by providing directional interactions35. The core of many protein structures comprises secondary structures like alpha helices and beta sheets, where the hydrogen-bonding potential between the main chain carbonyl oxygen and amide nitrogen is fulfilled, typically found in the protein’s hydrophobic core. These hydrogen bonds between proteins and their ligands offer specificity and directionality in interactions, a fundamental way of molecular recognition36. For a stable protein-ligand complex, the formation of hydrogen bonds is crucial. To assess the consistency of intramolecular bonding in TGFB3 before and after curcumin binding, the time evolution of hydrogen bonds was analyzed. The generated plot indicated a substantial increase in the number of hydrogen bonds within TGFB3 when curcumin bound till 100 ns, minor decrement was observed between 200 ns to 300 ns. Specifically, intramolecularly, the average number of H-bonds formed prior and post curcumin binding in TGFB3 were evaluated at 70 and 68, respectively (Fig. 4A; Table 1). Furthermore, the maximum number of intramolecular hydrogen bonds calculated prior and post curcumin interaction were 87 and 90. The plot indicated a higher number of hydrogen bonds formed in TGFB3 when complexed with curcumin, particularly during the period from 300 ns to 400 ns and after 450 ns. Additionally, the distribution plot of intramolecular hydrogen bonds (see Fig. 4, Lower panel) showed an increasing trend. Interestingly, the intramolecular hydrogen bonds in TGFB3 demonstrated stability through entire simulation, even after the interaction of curcumin. Furthermore, an evaluation of intermolecular hydrogen bonds was conducted to assess their constancy between curcumin and TGFB3. Within the TGFB3-curcumin complex, the maximum hydrogen bonds formed were estimated to be 5 (Fig. 4B). The Probability Density Function (PDF) suggested a fair consistency in the number of intermolecular hydrogen bonds in the curcumin-TGFB3 complex, with a higher PDF value indicating a number of hydrogen bonds (Fig. 4).

Fig. 4
figure 4

Dynamic hydrogen bonds. (A) Intramolecular hydrogen bonds plot indicating hydrogen bonds formed within TGFB3 before and after curcumin interaction. (B) Intermolecular hydrogen bonds plot indicating bonds formed between TGFB3-curcumin complex. Lower panel indicating distribution plot of (C) Intramolecular and (D) Intermolecular hydrogen bonds formation.

Full size image

Secondary structure dynamics

To comprehend the alterations in secondary structure induced by ligand binding, the time-evolution profiles of secondary structures calculated through DSSP are presented in Fig. 5. This analysis aims to assess the content of protein’s secondary structure over time. Secondary structure assignments like α-helix, β-sheet, and turns were dissected into individual residues for each time step. The average residues engaged in the formation of secondary structure slightly increased in the complexes. This increases due to from a higher percentage of β-sheet and bend and turn formation compared to TGFB3 alone (Table 2). Conversely, there was a minor reduction in the percentage of coil, A-helix and 5-helix formation observed in the TGFB3-curcumin complex (Fig. 5).

Table 2 Secondary structure elements composition calculated after 500 ns MD simulations.
Full size table
Fig. 5
figure 5

Secondary structure plot. The analysis shows the structural elements present in (A) TGFB3 and (B) TGFB3-curcumin complex.

Full size image

PCA and FEL

Principal Component Analysis (PCA) or Essential Dynamics (ED) is employed to assess the overall expansion of a protein throughout various simulations32. Using the gmx covar module, the dynamics of TGFB3 were computed concerning the protein’s alpha carbon atom. PCA identifies large-scale average movements in a protein, revealing the structures underlying atomic fluctuations37. The conformational sampling of TGFB3 systems within the essential subspace is depicted in the Fig. 6, demonstrating global motions along PC1 and PC2 projected by the alpha carbon atom.

The figure illustrates that the bound TGFB3 exhibits stable clusters compared to the unbound TGFB3 as it covers lesser subspace, especially in PC1. Conversely, the TGFB3-curcumin complex displays a distinct cluster than TGFB3 alone (Fig. 6). Overall, PCA results suggest that the TGFB3-curcumin complex induces more motion, indicating a different stable conformation in complex. These decreasing conformational dynamics could stem from a highly stable and optimized interaction network, effectively constraining the alpha carbon atom dynamics of apo form of protein and complex in form. Moreover, time-based eigenvector analysis was also done and plot displayed in Fig. 6B. As figure displayed except minor fluctuations between 200 ns to 300 ns at both vectors the complex shown stable conformations.

Fig. 6
figure 6

PCA analysis. (A) Two-dimensional conformational projections of TGFB3 and TGFB3-curcumin complex. (B) Time dependent projections of trajectories on eigenvectors.

Full size image

Additionally, the Gibbs free energy landscape, measured by using gmx covar, gmx anaeig, and gmx sham with the projections of their respective first (PC1) and second (PC2) eigenvectors38, is depicted in the Fig. 7. This landscape examines the direction of fluctuation in both systems for all Cα atoms of the free TGFB3 and TGFB3-curcumin complex structures from trajectories. Dark blue color shown lower energy. Notably, the main well showing free energy in the global free energy minima region underwent complete change after curcumin binding. The two energy wells represent less stable conformational states of TGFB3. Curcumin binding led to two small global minima of TGFB3 during the 500 ns MD simulations. Two clearly distinguishable minima appear in the energy landscape, representing metastable conformational states with minor energy barriers in TGFB3 (Fig. 7). The most stable conformational state was seen in the TGFB3-curcumin complex, where local minima shrink within the energy landscape. Comparatively, the Free TGFB3 and TGFB3-curcumin formed two wider and two small minima associated to metastable conformations, respectively, throughout the population of trajectories (Fig. 7). The structural presentation from the free-energy landscapes (FELs) indicated that TGFB3 maintained a stable conformation throughout the simulation, even in the presence of curcumin (Fig. 7, lower panels). This result suggests that curcumin might contribute to stabilizing a specific conformation of TGFB3. Additionally, this study underscores the efficacy of FEL analysis in elucidating protein folding dynamics and conformational alterations induced by ligand interactions. Deeper blue shading in Free Energy Landscapes (FELs) signifies protein conformations with lower energy near-native states. TGFB3-curcumin appears to have a two small overall global minimum, confined within a blue basin (Fig. 7).

Fig. 7
figure 7

Free energy landscape of (A) TGFB3 and (B) TGFB3-curcumin complex. Lower panels indicate protein and protein-ligand complex fetched from the free-energy wells in FELs.

Full size image

MM-PBSA analysis

To validate our study’s findings, MM-PBSA analysis was performed to estimate the binding free energy of TGFB3-curcumin complex. Notably, existing literature supports the reliability of MM-PBSA-derived binding free energy (ΔGbind) values, often showing strong correlation with experimental data. Including individual energy components in the calculations provides valuable insights into the ligand binding mechanism. The MM-PBSA results for the curcumin are presented in Table 3. The data reveal that the overall binding free energy of the hit molecule was − 14.9 kcal/mol, which is significant, indicating stronger binding affinity. Key energy terms, including van der Waals energy, electrostatic energy, SASA energy, and polar solvation energy, were identified as critical contributors to the total binding free energy. In this study, van der Waals and GGAS energies emerged as the primary driving forces behind the binding interactions of curcumin, significantly influencing their overall binding free energy.

Table 3 The binding free energy of the TGFB3-curcumin complex calculated using MM-PBSA analysis.
Full size table

Conclusion

Natural compounds have long been valued for their medicinal benefits, and curcumin, renowned for its therapeutic properties and culinary use, has garnered significant attention. This investigation employed curcumin against the TGFB3 target, employing various computational methodologies. The outcomes revealed compelling results from molecular dynamic simulations, showcasing stable systems through deviations, compactness, and gyration and PCA along with FEL findings throughout the simulation processes. These in silico findings collectively suggest curcumin’s potential as a credible inhibitor for TGFB3 in breast cancer.