Modulating Cariogenic Interfaces Through Machine Learning-Guided Peptide Design

In 2025, dental caries remains one of the most prevalent chronic diseases worldwide. Cariostatic agents such as silver diamine fluoride (SDF) have been receiving steadily increasing interest due to their efficiency, minimal intervention requirements, short application time, and low cost. SDF is an easily applied topical solution that can be used to arrest carious decay. It first gained interest in pediatric dentistry and is now emerging as a nonsurgical treatment option for treating caries in elderly and adult populations with special needs due to their pre-existing medical conditions. While SDF treatments offer speed and ease in caries management, making it readily accessible for a wider range of patients, its utilization remains limited. Several factors affect this outcome, including long-term efficacy, impact on future restorations due to modified tissue interfaces, and black staining. Upon application, SDF undergoes a series of chemical reactions on dental tissues that lead to tissue restoration with these concerning factors, including the permanent black staining of carious lesions, i.e., enamel and dentin. Previously, our group investigated a peptide-enabled approach for biomimetic reconstruction of SDF-treated dental tissues to mitigate the undesired effects. A silver binding peptide was explored to provide an anchoring ability on SDF-treated tissues with an additional remineralization function. This thesis further explores the design of peptides capable of targeting silver compounds formed after SDF treatments while minimizing interactions with hydroxyapatite through a generative machine-learning approach with augmentation for small data sets. Silver-binding peptide sequences are selected through a process that includes tailored physiochemical analysis and in silico assessments to discover candidates with properties that target caries-related environments and functions. Additionally, candidates underwent structural mapping along a pH spectrum designed to mimic the progression from healthy to severely cariogenic environments. Novel peptide candidates were selected for experimental evaluations and confirmed for their binding on SDF-treated hydroxyapatite interfaces. Overall, the machine learning-guided peptide predictions allowed for the investigation of targeted properties that relate to peptide function and the needs of the oral cavity.

Grantor dc:publisher

University of Kansas

Year dc:date.issued

2025

Author dc:creator

• Chu, Kalea

Advisor dc:contributor.advisor

• Tamerler, Candan

• Bioengineering
• Biomedical engineering
• Nanotechnology
• biomimetic
• dental caries
• machine learning
• oral health
• peptide

Statement dc:rights

• This item is protected by copyright and unless otherwise specified the copyright of this thesis/dissertation is held by the author.

Language dc:language.iso

en

Dc Identifier Other

https://www.proquest.com/LegacyDocView/DISSNUM/32042903

OAI identifier oai:identifier

oai:kuscholarworks.ku.edu:1808/37957