The PHLDA1-encoded protein (PEP) is a pivotal regulator of cardiomyocyte apoptosis and a promising therapeutic target for cardiovascular diseases. Leveraging in silico approaches, we characterized PEP’s structure using AlphaFold3 (AF3) and refined it via molecular dynamics (MD) simulations with various force fields. Kinetic and thermodynamics analysis further confirmed revealed that the Amber99 force field induced the most significant structural change, with the most significant decrease in radius of gyration and lowest potential energy. High-throughput screening of 20 phenylalanine-based dipeptides (FA-FY) revealed sequence-dependent binding to PEP, with FF, FH, FP, and FY exhibiting the maximum count of binders (11, 7, 6, and 6, respectively). Key residues (e.g., L55, C60, E142, E89) formed a continuous binding groove, while FF dipeptides demonstrated cooperative aggregation-enhanced binding. Notably, low-pLDDT regions (< 50) frequently participated in ligand interactions, challenging the rigid “lock-and-key” paradigm and underscoring the role of intrinsic disorder in binding. this work establishes PEP as a tractable target for dipeptide inhibitors and provides a computational framework for designing peptide-based therapeutics against cardiac injury.
Cardiovascular disease (CVD) remains the leading cause of death worldwide, accounting for approximately 17.9 million deaths annually. Myocardial ischemia, myocardial infarction, hypertrophy, and heart failure are major contributors to CVD-related mortality, highlighting the critical need for timely therapeutic interventions to preserve cardiac function. Cardiomyocytes are the basic functional units of the heart, responsible for the contraction and pumping functions, which are essential for maintaining systemic circulation. Cardiomyocytes are also capable of generating and transmitting electrical impulses, which are the triggering signals for heart contraction and diastole, and are tightly connected to each other by transverse, intermediate and gap junctions, ensuring that the heart as a whole pumping blood effectively. Damage or dysfunction of cardiomyocytes impairs the heart’s pumping capacity, often leading to heart failure and life-threatening complications. Consequently, research targeting cardiomyocyte-specific proteins with pharmacological potential is crucial for developing novel therapeutic strategies.
Peptide drugs, composed of 2 to 50 amino acid residues linked by peptide bonds, represent a promising class of therapeutics with broad applications in medicine. These drugs are particularly effective in treating complex diseases such as cancers, immune disorders, and metabolic dysregulation due to their high specificity (i.e., minimal off-target effects). Compared to conventional small-molecule drugs, peptide therapeutics exhibit superior selectivity, rapid systemic clearance, and reduced risk of toxicity or accumulation. Advantages of peptide drugs also include their low immunogenicity (unlike larger biologics such as monoclonal antibodies), lower required dosages, and higher potency per unit mass. Their clinical utility extends to diverse conditions, including oncology, infectious diseases (e.g., hepatitis, HIV/AIDS), and metabolic disorders (e.g., diabetes). Given these properties, the identification and screening of peptide ligands capable of selectively targeting disease-relevant proteins are critical for advancing biomedical research and drug development.
The Pleckstrin Homology-Like Domain Family A Member 1 encoded-protein (PHLDA1-encoded protein, abbreviated as PEP) is a regulatory protein associated with cell growth, survival and apoptosis. In cardiomyocytes, the expression and functional role of PEP have garnered significant research interest. Studies have demonstrated that PEP overexpression in H9c2 cardiomyocytes leads to decreased cell viability. Mechanistically, PEP modulates the stability of Bax, a critical mediator of the mitochondrial apoptosis pathway, through protease-dependent regulation, and PHLDA1 knockout attenuates Bax upregulation during myocardial ischemia-reperfusion injury. Targeted screening of short peptide ligands (e.g., dipeptides) capable of binding to PEP may yield novel therapeutic candidates. By selectively modulating PEP’s function, such peptides could offer a viable strategy to mitigate myocardial ischemia and related cardiovascular conditions.
Notably, while PEP has been implicated in cardiomyocyte apoptosis, no prior studies have systematically investigated its structure-function relationships or explored its potential as a drug target. Our in silico approach combining molecular dynamics (MD) simulations with biophysical analyses represents the first comprehensive effort to: (1) map PEP’s druggable binding sites, and (2) identify optimal dipeptide ligands capable of modulating its pathological interactions. MD simulation is a powerful computational methodology rooted in Newtonian mechanics, widely employed in chemistry, physics, biology, materials science, and pharmaceutical research. Its ability to efficiently explore molecular motion trajectories, interactions, and structural dynamics offers a time- and cost-effective alternative to experimental approaches, while providing atomic-level insights that complement empirical studies. Our strategy combines computational efficiency with mechanistic precision, offering a foundation for developing novel peptide-based therapies against PEP-driven cardiovascular pathologies.
The determination of an accurate protein structure serves as the critical foundation for rational drug design. In this study, we began by retrieving the complete amino acid sequence of the PEP from the UniProt database (see Method for specific website). This sequence served as input for AF3, which generated a three-dimensional structural prediction of PEP as shown in Figure 1a. To assess the reliability of this predicted structure, three complementary metrics are employed: the per-residue pLDDT (predicted Local Distance Difference Test) scores, ipTM (interface predicted Temperate Modelling score), and PAE (Predicted Aligned Error), with detailed results presented in Figure 1(a).
Figure 1: Structural comparison between AlphaFold3 (AF3)-predicted and molecular dynamics (MD)-simulated conformations of PHLDA1-encoded protein (PEP). (a) AF3-predicted PEP structure color-coded by predicted Local Distance Difference Test (pLDDT) confidence scores, with ipTM (interface predicted Temperate Modelling score) and PAE (Predicted Aligned Error) prediction metrics indicated. (b-d) Structural alignments between the initial AF3 prediction (cyan) and MD-relaxed conformations (t = 400 ns under NPT ensemble) simulated using: (b) Amber03-green, (c) CHARMM27-purple, and (d) Amber99 force fields-red. Alignment was performed using PyMOL with root-mean-square deviation (RMSD) values shown for each force field comparison.
The predicted model is visually annotated using a color gradient based on pLDDT confidence values. Regions exhibiting pLDDT scores above 90, depicted in dark blue, represent the most reliable portions of the prediction with near-atomic accuracy. Areas scoring between 70 - 90, shown in light blue, indicate moderately confident predictions that likely represent genuine structural elements but may contain some local flexibility. Yellow-colored segments (50 - 70) correspond to lower-confidence regions where the predicted conformation should be interpreted with caution. Finally, orange portions (pLDDT < 50) signify very low confidence predictions, typically corresponding to intrinsically disordered regions or predictions with low confidence. It can be observed that only a small fraction of the sequence exhibiting α-helix conformation (in dark blue) has very high level of prediction confidence (pLDDT > 70). Notably, the core functional domains of PEP appear predominantly in the high-confidence regions, supporting their validity for our subsequent interaction studies with potential dipeptide ligands. However, the low ipTM score (= 0.48) and PAE (a large fraction of predicted distance error for a pair of residues is larger than 25 Å) suggests overall limited confidence in the predicted structure, implying that the initial AF3-predicted structure may require optimization, and the low-confidence regions are highly dynamic and prone to conformational changes during MD simulations.
The structural comparison between the AF3-predicted PEP conformation and the MD-simulated structures reveals several important insights into the protein’s dynamic behavior. Through PyMOL-based structural alignment, we observed significant conformational changes following MD simulation, particularly in regions that initially exhibited low prediction confidence (pLDDT < 70) in the AF3 model. These flexible regions, often corresponding to loop domains or solvent-exposed surfaces, demonstrate the expected structural plasticity during simulation, while the high-confidence core domains (pLDDT > 70) maintain greater stability across all force fields tested. The quantitative RMSD analysis provides compelling evidence for force field-dependent behavior of PEP. The Charmm27 force field yielded the most stable simulation (RMSD = 31.178 Å), suggesting its parameters may be well-suited for modeling PEP’s structural dynamics. In contrast, the larger deviations observed with Amber03 (35.926 Å) and especially Amber99 (41.435 Å) indicate these force fields may introduce excessive conformational sampling or insufficient restraints for this specific protein system.
These findings have important implications for both computational and experimental studies of PEP. The force field-dependent structural variations highlight the need for careful parameter selection when studying PEP’s conformational landscape, particularly for molecular docking or binding studies where accurate representation of binding site geometry is crucial. The observed stability of high-confidence regions across simulations supports their use in initial binding pocket identification, while the dynamic nature of low-confidence regions suggests they may require enhanced sampling techniques or experimental constraints (such as Nuclear Magnetic Resonance Spectroscopy, NMR data) for reliable modeling. Furthermore, the superior performance of Charmm27 in maintaining structural integrity suggests it should be the force field of choice for future studies of PEP’s interactions with potential therapeutic peptides. The significant structural changes observed in low-confidence regions also provide valuable insights into PEP’s potential functional mechanisms. These flexible domains may represent important regulatory regions that undergo conformational switching during biological activity or binding events. The force field-dependent variations in these regions could reflect genuine structural plasticity that might be functionally relevant, rather than simply being artifacts of the simulation parameters. This interpretation is supported by the known role of PEP in mediating cellular stress responses, where structural flexibility could be crucial for its function in apoptosis regulation.
These results collectively demonstrate that while AF3 provides a valuable starting point for PEP structural studies, subsequent MD refinement is essential, particularly for t
