Preface

The deep blue color of blueberries comes from a class of key molecules—anthocyanin glycosides. These natural pigments are not only the “color artists” of the plant world but also star molecules in functional foods and drug development due to their antioxidant and anti-inflammatory activities. However, the final structure and function of these glycosides are determined by a group of invisible “sugar artisans”—glycosyltransferases (UGTs).

A team led by Professor Wei Chen at Zhejiang University recently published a study in Science Advances, where they identified a special glycosyltransferase from blueberry, UGT75AJ2. This enzyme naturally possesses a rare dual glycosylation ability at the 3′-O and 7-O positions, making it highly valuable for synthesizing rare anthocyanin glycosides and natural blue pigments. Interestingly, although phylogenetically classified within the 5-O-glycosyltransferase clade, experiments confirmed that it mainly catalyzes glycosylation at the 3′-OH and 7-OH positions. This mismatch between its classification and function suggested a unique catalytic logic.

This led researchers to a critical question: Can this “abnormal logic” be deciphered and converted into programmable, controllable glycosylation?

However, this “artisan” comes with three natural shortcomings—extremely low activity, a single donor preference (only UDP-Glc), and poor regioselectivity—greatly limiting its application. These challenges are widespread across the UGT family. Traditional enzyme engineering is like “searching for a needle in a haystack”—inefficient and prone to local optimization, making it far from capable of decoding or reshaping catalytic logic with high precision.

How can one tame this unpredictable “craftsman,” turning its initial boundary-breaking behavior into precisely programmable glycosylation?

The research team’s answer is FRISM (Focused Rational Iterative Site-specific Mutagenesis). This is not just “structure-guided engineering” but a reproducible engineering loop that deeply integrates structural prediction, energy analysis, and iterative optimization, enabling systematic decoding and remodeling of enzyme catalytic logic.

FRISM Step 1: “Precision Targeting”—Moving From Blind Search to Guided Engineering

The team first used multiple structural models and molecular docking to identify about 70 potential key residues near the active pocket. These residues were then subjected to virtual saturation mutagenesis, followed by calculations of mutational free-energy changes (ΔΔG_mut) and statistical filtering (FDR < 0.05) to pinpoint destabilizing residues (ΔΔG_mut > 0) that could “perturb” the enzyme structure. Evolutionary conservation analysis was then applied to eliminate immutable functional residues, ultimately narrowing the list to 34 key residues located in the donor and acceptor channels as core mutation hotspots.

Unlike traditional rational design that seeks “stronger binding” (ΔΔG_mut < 0), this study uniquely focused on mutations that “weaken” binding (ΔΔG_mut > 0). The underlying logic is that controlled, moderate destabilization can force substrates to adopt new conformations within the pocket, making them more likely to reach a catalytically favorable transition state instead of being overly stabilized in nonproductive states. This counterintuitive strategy laid the theoretical foundation for the dramatic enhancement in catalytic performance.

Step 2: “Iterative Evolution”—Four Rounds of Optimization, Achieving a 128-Fold Activity Boost

After identifying mutation hotspots, FRISM’s second phase is modular, stepwise optimization. The team divided energy hotspots into functional modules. Starting with single mutants, they iteratively screened for the most improved combinations. After four rounds, they constructed a quadruple mutant Mut4-1 (S367A/V274A/F82V/I132T), which increased conversion from 0.17% in the wild type to 22.4%. Catalytic efficiency (kcat/Km) improved 73-fold, and overall activity increased 128-fold.

Back in 2020, the same research group addressed a similar challenge using PcHNL5, an FAD-dependent enzyme from sweet almond (Prunus communis). By analyzing the crystal structure of PcHNL5 bound to its natural substrate benzaldehyde and rationally mutating residues in the substrate tunnel, they discovered that a single mutation, L331A, expanded the tunnel, allowing the enzyme to process bulky non-natural substrates such as 1,3-dioxane-derived benzaldehydes (1d). However, the catalytic efficiency of PcHNL5_L331A was only 1.8 s⁻¹·mM⁻¹, far lower than the 76 s⁻¹·mM⁻¹ of the wild type toward benzaldehyde.

Breakthrough 1: Reshaping the Donor Channel and Breaking the “Single-Sugar” Limitation

UGTs typically suffer from a fundamental limitation—they almost exclusively accept UDP-Glc as the sugar donor. By analyzing the electrostatic and hydrogen-bonding networks in the donor-binding pocket, the team identified F366 and S367 as “gatekeeper” residues. They designed a triple mutant (S14G/F366H/S367G) that enlarged and increased the flexibility of the donor pocket while preserving catalytic-site architecture. The engineered enzyme successfully broke its natural restriction, efficiently accepting four donors: UDP-Glc, UDP-Gal, UDP-Xyl, and UDP-GlcNAc—achieving a leap from mono-donor to multi-donor capability.

Breakthrough 2: Programmable Regioselectivity, Enabling Position-Controlled Glycosylation

The team further demonstrated that altering the spatial orientation of residues such as I132 and S367 can program glycosylation at different hydroxyl positions on flavonoids:

• I132E favors 3′-O glycosylation
• S367W favors 7-O glycosylation
• S367F enables the rare 5-O modification

Through combinatorial optimization, they developed two high-performance triple mutants—Y75E/F185Y/I132T and L112F/M273E/S367D. These achieved:

• 99% selectivity for 3′-O glycosylation (93% conversion)
• 85% selectivity for 7-O glycosylation (80% conversion)

These results push controllable and programmable enzyme design to new heights.

Conclusion

The FRISM strategy establishes a standardized route for targeted UGT engineering through precise hotspot identification and a unique “energy perturbation” design philosophy. This strategy not only enhanced UGT75AJ2 activity by 128-fold but also broke through challenges in donor flexibility and regioselectivity control. It provides a valuable tool for green manufacturing of natural products and drug development, and lays a so