Insights into Chemical Structure-Based Modeling for New Sweetener Discovery

Area of Science:

• Computational chemistry
• Molecular modeling
• Sweetener research

Background:

• The development of novel, high-sweetness, low-calorie sweeteners is an ongoing challenge.
• Understanding the molecular mechanisms of sweetness perception is crucial for designing new sweeteners.

Purpose of the Study:

• To investigate structure-based machine learning models for predicting sweetness.
• To elucidate the sweetness recognition mechanism of sweet compounds.

Main Methods:

• Utilized structure-based machine learning models to analyze molecular properties.
• Investigated the relationship between molecular structure and sweetness.
• Analyzed the binding interactions of sweet compounds with the T1R2-T1R3 receptor.

Main Results:

• Sweetness depends on molecular connectivity, composition, efficiency, complexity, and shape.
• Relative sweetness is influenced by molecular properties, complexity, and composition.
• Machine learning models accurately classified sweet/non-sweet compounds and predicted relative sweetness.
• Sweet compounds bind to the VFT domain of T1R2-T1R3 via hydrogen bonds.
• Higher sweetness correlates with stronger binding to the VFT domain.

Conclusions:

• Molecular features can predict a compound's sweetness and relative sweetness.
• The VFT domain of T1R2-T1R3 is a key binding site for sweet compounds.
• This research provides valuable insights for designing new sweeteners.