The significance of the study lies in its attempt to bridge two levels of description. Molecular dynamics (MD) simulations are the workhorse of structural biology: they track every atom of a peptide and its surrounding water molecules over time, producing rich but computationally expensive movies of molecular motion. The MMFs method, by contrast, works from first principles at the level of individual non-covalent interactions. It calculates how electrostatic attractions, repulsions, hydrogen bonds, and hydrophobic contacts between neighboring side chains generate torques that twist the peptide backbone, ultimately determining the dihedral angles that define the molecule’s fold. What the new study adds is a way to fold the physics of pH into that mechanical picture.

To test the idea, the researchers selected a well-characterized model system: a beta-hairpin peptide catalogued in the Protein Data Bank under the code 1K43, originally described by Pastor and colleagues in 2002. A beta-hairpin is one of the simplest folded structures a peptide can adopt, two short strands connected by a turn, and this particular sequence contains an abundance of pH-responsive residues, including arginine, lysine, glutamate and tyrosine. That makes it an ideal proving ground for asking whether a calculation-based method can track how protonation and deprotonation events remodel the network of interactions that holds the hairpin together.

The researchers began by validating their tool against known reality. Starting from a fully unfolded model of the peptide, in which every backbone dihedral angle was set to 180 degrees, they applied the MMFs methodology to rebuild the folded structure purely from the mechanical consequences of side-chain interactions. The resulting model agreed with the experimental NMR structure with a root-mean-square deviation (RMSD) of 2.400 angstroms, a level of agreement the authors attribute mainly to the known difficulty of modeling the flexible terminal residues, arginine 1 and arginine 14. With the reference structure established, the team could then use it as a baseline for comparing pH-altered versions of the peptide.

The key methodological extension involves the pKa values of the ionizable side chains. At low pH, acidic groups such as the carboxylate of glutamate become protonated and lose their charge; at high pH, basic groups such as the amine of lysine and the phenolic groups of tyrosine shed their protons. Each of these events eliminates or creates specific non-covalent interactions, and the MMFs framework now subtracts the corresponding partial contributions to the backbone dihedral angles. At acidic pH, for example, the protonation of the terminal carboxylate of arginine 14 and the side-chain carboxyl of glutamate 12 required the removal of electrostatic attraction, repulsion and hydrogen-bond terms from the affected residues. At pH 11, deprotonation of the two tyrosines forced the team to account for a newly created electrostatic attraction between the negatively charged tyrosine 11 side chain and arginine 14, while removing the hydrogen bonds that the neutral phenols previously donated.

One of the more surprising findings concerns glycine, the smallest amino acid. The calculations revealed that glycine residues require a fundamentally different treatment under acidic conditions than under neutral or basic ones. At low pH, the phi dihedral angle of a glycine is influenced by all possible non-covalent interactions with its neighbors, apparently because the achiral nature of glycine’s central carbon permits extra local flexibility under acidic conditions. At higher pH, a simpler rule based on coplanarity with the preceding residue sufficed. To complete the framework, the authors also derived new equations for the psi dihedral angle covering combinations of secondary-structure contexts, such as left-handed alpha-helices paired with beta-strands or pi-helices, that had not appeared in their previous formulations.

To check whether these hand-calculated structures were trustworthy, the team ran 250-nanosecond molecular dynamics simulations of the peptide at pH 3, pH 7.4 and pH 11 using the YASARA engine, with careful control of protonation states, physiological salt concentration, and periodic boundary conditions. Protonation assignments were cross-validated with four independent pKa prediction tools: YASARA’s built-in method, H++, PypKa and KaML, which produced highly consistent values, for instance predicting a pKa of roughly 4.2 to 4.3 for glutamate 12 and about 10.3 to 10.8 for lysine 3. The simulations sampled conformations every 200 picoseconds, and the researchers compared the MMFs-predicted structures against conformers extracted at 25-nanosecond intervals throughout each trajectory.

The verdict was a consistent story across all three pH conditions. At pH 3, the peptide remains essentially stable: the MMFs-derived structure differed from the experimental reference by only 2.486 angstroms and matched the MD ensemble with an RMSD as low as 2.69 angstroms. At physiological pH 7.4, the hairpin partially holds together but grows more flexible, with the corresponding RMSD values rising to 3.922 angstroms against the experimental structure and a best match of 3.31 angstroms against the MD trajectory. At pH 11, the structure largely falls apart, showing an RMSD of 6.761 angstroms versus the experiment, a signature of substantial loss of native organization. Strikingly, the MMFs prediction at high pH matched the MD ensemble with an RMSD of just 2.88 angstroms, the best agreement of all, confirming that the method correctly captures the consequences of deprotonating lysine and the two tyrosines.

That agreement matters because of what it could enable. Designing peptides for drug discovery requires knowing that a candidate sequence will hold a defined shape long enough to bind its biological target, rather than flopping into an unproductive conformation. pH-responsive peptides are also the basis of so-called conformational switches, molecules that deliberately change shape in response to acidity, with applications ranging from biosensors to metal-ion chelators to drug delivery systems that release their cargo in the acidic environment of tumors or endosomes. A fast, interaction-based calculation that maps out the pH range in which a given sequence stays folded, or falls apart, could dramatically shorten the design cycle for such systems, without waiting for full-scale simulations to finish.

The authors caution, as they have in earlier work, that terminal residues remain the weak point of the MMFs approach, and the glycine behavior under acidic conditions adds a layer of system-specific complexity. But the overall message is that a mechanical, bond-by-bond accounting of protonation effects can reproduce the same pH-dependent trends that expensive atomistic simulations reveal. The researchers suggest their refined framework could inform future simulation platforms dedicated to peptide folding mechanisms, offering a complementary lens on how the invisible chemistry of a single proton reshapes the architecture of an entire molecule, one non-covalent interaction at a time.

Subject of Research: pH-dependent conformational transitions in a beta-hairpin peptide predicted by an extended Main Mechanical Forces approach and validated with molecular dynamics simulations

Article Title: Unraveling pH-dependent peptide conformational transitions via integration of main mechanical forces and side-chain interactions

Article References: Larocca, M., Floresta, G., Verderese, D., & Cilibrizzi, A. (2026). Unraveling pH-dependent peptide conformational transitions via integration of main mechanical forces and side-chain interactions. Frontiers in Chemical Biology, 5, Article 1914849. https://doi.org/10.3389/fchbi.2026.1914849

Image Credits: AI Generated

DOI: 10.3389/fchbi.2026.1914849

Keywords: peptide folding, pH-responsive peptides, Main Mechanical Forces, molecular dynamics, protonation states, non-covalent interactions, beta-hairpin, dihedral angles, pKa prediction, computational chemistry, conformational switches, drug delivery