Engineered thermo-stable, low-ph, low-allergen thaumatin sweetener variants

Overview

This research initiative aims to engineer next-generation Thaumatin sweetener variants that maintain sweetness under challenging conditions, including high temperatures and low pH environments, while reducing allergenicity. Thaumatin is a naturally sweet protein prized for its intense sweetness, but its sensitivity to heat and pH, along with allergenic potential, has limited broader commercial adoption in food and beverage products.

The project leverages an iterative molecular breeding approach combined with machine learning to systematically screen and optimize protein variants. By integrating rational design based on Thaumatin's crystal structure with high-throughput functional screening, the platform can identify variants that preserve sweetness across diverse physicochemical conditions. This enables food manufacturers to use Thaumatin-based sweeteners in a wider range of products, including baked goods, acidic beverages, and processed foods where conventional Thaumatin would lose functionality.

Technical specifications

Key features:

• Rational variant design guided by Thaumatin crystal structure data, targeting regions sensitive to temperature and pH degradation
• Multi-well in-vitro protein expression system enabling parallel screening of numerous variants
• HEK cell-based reporter system with stable expression of sweet taste receptors T1R2 and T1R3 via lentiviral transduction
• cAMP-responsive luciferase reporter providing objective, quantifiable measurement of sweetness activation
• TensorFlow-based machine learning models that analyze relationships between sequence variations, treatment conditions, and receptor activation to predict optimal sequences
• Iterative design-test-learn cycles (2 to 3 rounds) to progressively refine variant performance
• Yeast expression system for industrial-scale production of selected variants

The workflow begins with DNA synthesis encoding designed variants, followed by in-vitro protein production. Variants are exposed to various temperatures, pH levels, and incubation times, then functionally assessed using the sweet receptor reporter system. Sequence and performance data feed into machine learning models that predict the next generation of improved variants.

Technology readiness level

This research is at an early stage (TRL 2-3). The hypothesis has not yet been validated, and experimental work is contingent upon securing funding. Once initiated, the team will construct the HEK reporter cell line, design initial variant libraries, and execute the iterative screening and machine learning cycles. Successful validation will lead to yeast-based expression for industrial scale-up, positioning the technology for potential partnership with food ingredient companies seeking novel sweetener solutions.