Reference #: 1759
The University of South Carolina is offering licensing opportunities for Engineered peptides by machine learning for cell osteogenic differentiation.
Background:
Resorbable materials loaded with bone morphogenetic proteins (BMPs) are used clinically as a regenerative alternative to autologous bone transplantation for treating skeletal defects to restore continuity. The BMP induces migration of progenitor cells from the surrounding tissue to the injury site followed by their differentiation to pre-osteoblasts and bone morphogenesis. The resorption of the graft concurrent with osteogenesis provides volume for the apposition of new bone. However, the short half-life of the recombinant human bone morphogenetic protein-2 necessitates administering 3-4 orders of magnitude higher doses than the endogenous amount for bone formation and healing. The high doses combined with BMPs’ role in the development of a wide range of tissues from embryonic to adulthood cause undesired side effects such as bone overgrowth, immune response, tumorigenesis, and neurological complications. Consequently, the FDA has issued warning to physicians against off-label use of rhBMP-2 in clinical procedures which has limited the wide spread use of BMPs in orthopedics.
An alternative approach is to use the bioactive knuckle epitope domain of BMP-2 (BMP2-KEP) with an open-arm structure as part of the protein for engineering skeletal tissues. However, the osteogenic activity of free peptide is orders of magnitude lower than the native peptide which is attributed to the closed-arm structure of the free peptide.
Invention Description:
This innovation details a Quantitative Structure Activity Relationship (QSAR) using different machine learning (ML) models to correlate 20-mer sequences of modified BMP-2 KEP to their configurational properties. As the existing structure-property data for osteogenic peptides are insufficient for training ML models, the SIMFIM mesoscale simulation model was used to obtain structural properties of the modified BMP2-KEP sequences to create a database. As the features of the database, the residues in the sequences were represented by different amino acid descriptor (AAD) scales. Grid search was used to fine tune the hyperparameters of different ML models. The performances of all the models were compared using the R2 performance metric. Feature importance and SHAP interaction analysis were done to determine which residue positions and properties had a greater contribution to the structural properties of the sequences. These studies led to a trained and tested QSAR model for predicting the structural properties of any modified BMP2-KEP sequence for the purpose of discovering novel 20-mer sequences with open-arm structures. The model predicted new peptides with higher binding affinity to its receptor, hence higher osteogenic activity, than the unmodified BMP2-KEP peptide in the free state.
Potential Applications:
Medical field, specifically with the treatment of osteoporosis
Advantages and Benefits:
This innovation is cost-effective with less side-effects than current treatments available on the market.