Human bone morphogenetic proteins (BMPs) are used in certain clinical applications such as spine fusion and non-unions to accelerate bone regeneration and healing. However, due to their short-half-life and diffusion, doses much higher than the endogenous amount are used clinically to stimulate bone formation. These high doses cause undesired side effects such as bone overgrowth and tumorigenesis, which has limited their widespread use in orthopedic surgery. An alternative approach is to use peptides derived from the bioactive domains of BMPs to reduce side effects[1, 2]. However, the osteogenic activity of these peptides is much lower than that of BMPs which is attributed to conformational and solubility differences between the native and free peptide. In this regard, there is a need to design and discover novel peptides that mimic the native conformation and solubility of the epitope sequence on BMPs that interact with cell surface receptors. The bioactive domain of the BMP-2 peptide known as the knuckle epitope has an open arm structure when it is a part of the protein but behaves differently in its free form indirectly affecting its osteogenecity[3, 4]. Hence there is a need to come up with new peptide sequences which mimic the knuckle epitope sequence and has biological activity comparable to that of the BMP-2 protein. We hypothesize that such biomimetic peptides can be engineered using the power of mesoscale simulation and machine learning. We developed a uniquely parameterized mesoscale simulation method or Dissipative particle dynamics method, in which the macromolecules are coarse-grained into beads, to determine the biophysical properties of the peptides. The forcefield for this method included evaluating the conservative potential term by considering specific interactions between the different beads and modelling the bonded interaction potentials such as the bond stretching as harmonic bonds, bond angles assume cosine harmonic functional forms and dihedral angle energies are modelled through a shifted dihedral potential[5]. The electrostatic interactions were determined using a function that allows to distribute or smear charges on the bead. We examined this method at different coarse-graining levels or mappings for a known 12-mer peptide TrpZip2 (PDB ID: 1LE1). A mesostructure was created with one peptide in a box of water with periodic boundary conditions and simulated under NVT conditions. We compared the radius of gyration values of the simulated backbone with the actual structure. The radius of gyration of the 20 structures in the PDB database was in the range 5.6 A - 6.2 A. The average of the radius of gyration of the simulated peptide structure was within the range of the radius of gyration values of the actual structure. This method can be used to predict the structural properties of peptides in a quick and reliable way. Using this method and existing experimental characterization techniques, the structural properties and biological activity of several modified knuckle epitope sequences will be determined to create a database. This database will be used to train and build a Quantitative Structure-Activity relationship model through machine learning algorithms which can then generate potential osteogenic peptide sequences. The discovery of highly active peptides could potentially expand their use in orthopedic applications to improve the quality of life of patients with skeletal injuries.
References
[1] R. Visser, G.A. Rico-Llanos, H. Pulkkinen, J. Becerra, Peptides for bone tissue engineering, Journal of Controlled Release 244 (2016) 122-135.
[2] G. Tang, Z. Liu, Y. Liu, J. Yu, X. Wang, Z. Tan, X. Ye, Recent Trends in the Development of Bone Regenerative Biomaterials, Frontiers in Cell and Developmental Biology 9 (2021) 665813.
[3] G. Zhao, L. Zhang, L. Che, H. Li, Y. Liu, J. Fang, Revisiting bone morphogenetic proteinâ2 knuckle epitope and redesigning the epitopeâderived peptides, Journal of Peptide Science 27(6) (2021).
[4] S. Moeinzadeh, D. Barati, S.K. Sarvestani, T. Karimi, E. Jabbari, Experimental and Computational Investigation of the Effect of Hydrophobicity on Aggregation and Osteoinductive Potential of BMP-2-Derived Peptide in a Hydrogel Matrix, Tissue Engineering Part A 21(1-2) (2015) 134-146.
[5] R. Vaiwala, K.G. Ayappa, A generic force field for simulating native protein structures using dissipative particle dynamics, Soft Matter 17(42) (2021) 9772-9785.
![](https://www.aiche.org/sites/default/files/aiche-proceedings/3382596/papers/661438/Paper_661438_abstract_196997_0.png)