Against the backdrop of global food security facing the challenge of Ochratoxin A (OTA) contamination in food products, our team has conducted a systematic study on OTA biodegradative enzymes. The uniqueness of this research lies in the fact that every critical decision and directional shift has incorporated in-depth considerations of Human Practices, particularly through continuous dialogue with scientific experts, which has infused profound real-world concern into the scientific exploration of the project.
In the beginning of the project, we consulted with experts in the field of food safety with the question "How to effectively address OTA contamination in food grains?". A senior researcher clearly stated: "Current mainstream physical and chemical detoxification methods, such as ozone treatment, are effective but tend to cause nutrient loss and chemical residues and even biosorbents like zeolites exhibit limited impact. The development of efficient and specific biological degradation methods is an urgent need of the industry." This statement pointed us in the right direction, leading us to abandon traditional approaches and focus on innovative enzyme-based degradation solutions.
Why choose ADH3 among numerous candidate enzymes? This decision was also validated by both literature and expert input. An enzymology expert advised: "Selecting a high-activity 'chassis' element is half the success of the project". This prompted us to conduct extensive literature mining, ultimately identifying ADH3, the most efficient OTA hydrolase discovered to date, as our research starting point, ensuring a high-level foundation for the project.
After successfully increasing the activity of ADH3-S88E by 3.7-fold in the first cycle, we once believed that enzyme activity modification was approaching its limit. However, during a mid-term presentation, a structural biology professor raised a critical question: 'From a structural perspective, the binding affinity of ADH3's catalytic pocket to the substrate may have already reached its theoretical limit. To achieve a stepwise improvement in efficiency, it may be necessary to search for homologous enzymes with deeper catalytic pockets.'
This insightful observation became a turning point for the entire project. We immediately adjusted our direction and shifted to bioinformatics mining, eventually discovering the sequence of homologous enzyme LIADH, which shares a 73% similarity with ADH3. After systematic expression, purification, and functional verification, the results were inspiring: the activity of LIADH was twice that of ADH3, perfectly validating the expert's prediction and achieving a leap in performance.
After obtaining the high-resolution structure of LIADH, we initiated rational design work. At this stage, an engineer from a biotechnology company provided a suggestion: "An excellent industrial enzyme preparation requires not only high activity but also good stability." Therefore, while designing mutants to enhance activity (such as I326A, which increased activity by 75%), we also fully recognized the deficiencies of LIADH in terms of thermal stability and metal ion tolerance---issues that must be addressed for future application.
Looking back on the entire project, expert interviews and Human Practices (HP) are not isolated components outside scientific research, but are deeply integrated into each research cycle, serving as a "compass" to guide the project's direction. From selecting the research direction, breaking performance bottlenecks to planning application pathways, the valuable insights of experts have ensured that our work always targets real-world needs, embodying the core spirit of iGEM---'Responsible Innovation'. In the future, we will continue to listen to the voices of the industry, persist in exploring enzyme stability, heterologous expression, and practical application testing, striving to translate our research outcomes from the laboratory to farmlands and production workshops, and contributing a solid young force to safeguarding global food security.
Our design stems from addressing the challenge of Ochratoxin A (OTA) contamination in global food security. Following suggestions from expert interviews, our team took note of OTA's strong carcinogenicity, nephrotoxicity, and its widespread contamination in agricultural products such as grains, coffee, and wine. Traditional physical and chemical methods (e.g., adsorption, ozone treatment) suffer from drawbacks including low efficiency, nutrient loss, and chemical residues. Therefore, we aim to design an efficient, safe, and specific bio-enzymatic degradation method to make a modest and prudent contribution to human food safety.
To also establish a paradigm for future research on similar enzymes, our design phase integrates bioinformatics, structural biology, and rational protein engineering to create both an outstanding enzyme and a comprehensive system.
Through literature mining, we selected the amidohydrolase ADH3 from Stenotrophomonas acidaminiphila CW117 as the core biological element. It has been proven to be the most efficient OTA hydrolase------ADH3(WT) discovered to date (1.2 μg/mL enzyme completely degrades 50 μg/L OTA within 90 seconds), with an efficiency 67 times higher than that of OTase, the previously optimal enzyme. This ensures that our research starts with a high-performance "chassis" element.
Through bioinformatics analysis, we used tools such as Phyre2 to perform sequence alignment between ADH3 and amidohydrolases with known structures. The results showed that ADH3 contains highly conserved amino acids (H83, H85, H251, H271, K210, D344) that coordinate with the binuclear metal center, classifying it into the binuclear metal amidohydrolase superfamily.
We postulated that its catalytic mechanism is similar to that of other enzymes in this family, and predict that the diversity of substrate-binding pockets is determined by the loop region. This provides clear targets and support for our subsequent mechanism exploration and molecular modification.
First, we predicted and removed the 20-amino-acid signal peptide at the N-terminus of ADH3 through bioinformatics. The optimized adh3 gene was fully synthesized and cloned into the pET46/Ek-LIC vector to form a recombinant plasmid with a 6×His tag at the N-terminus. Codon optimization was then performed, and the final recombinant plasmid was named pET46/ADH3. This enabled our constructed biological element (the ADH3 gene) to carry a standardized purification tag.
Design Objectives：
Based on the above design objectives, we systematically constructed three major types of mutants using site-directed mutagenesis technology (names were derived from subsequent experimental results):
All constructs with optimized codons were transformed into the E. coli BL21(DE3) expression system for induced expression. A well-established three-step purification process was developed:
(Figure : SDS-PAGE Image of DEAE Purification)
(Figure: ADH3 Size-Exclusion Chromatography Profile)
The above construction process demonstrates our ability to efficiently and reliably prepare large quantities of high-quality protein, which forms the basis for all subsequent tests. The final protein purity achieved was >95% (Figure 3.3), meeting the standards for structural biology research.
We established a high-performance liquid chromatography (HPLC) method to accurately quantify OTA degradation and the production of the product OTα, thereby evaluating enzyme activity.
(Figure: HPLC Chromatogram of Standard Samples)
(Figure : Verification of Inactive Mutants)
(Figure : Functional Verification of Key Sites)
This represents the most prominent technical highlight of this project. Using single-particle cryo-electron microscopy (cryo-EM) technology, we successfully resolved three sets of high-resolution structures:
(Figure : Details of OTA-Protein Interactions)
These structural data provide us with atomic-level "visual evidence", greatly enhancing the credibility of all our conclusions and serving as a bridge connecting design and learning.
Integrating all test data (mutant enzyme activity + high-resolution structures), we elucidated the complete catalytic mechanism of ADH3 for the first time:
(Figure : Schematic Diagram of the Catalytic Mechanism)
Based on the D344N/OTA complex structure, we gained a key insight: the side-chain hydroxyl group of S88 points toward the phenol hydroxyl group of the OTα moiety of the OTA molecule, but the distance is too great to form a strong interaction.
New Hypothesis：If S88 is mutated to an amino acid with a longer side chain or a charged group (e.g., Glu, Lys), stronger hydrogen bonds or electrostatic interactions may form, thereby stabilizing substrate binding more effectively and improving catalytic efficiency.
New Design：Based on this, we constructed a series of mutants including S88D, S88R, S88H, S88E, and S88K.
Iterative Test Results ：
The enzyme activity test results have perfectly verified our new hypothesis. The S88E and S88K mutants have achieved disruptive performance improvements (3.7-fold and 2.4-fold respectively), which is most likely due to the fact that the carboxyl group of Glutamic acid (Glu) and the amino group of Lysine (Lys) form a stronger hydrogen bond network with the phenolic hydroxyl group of OTα, thereby stabilizing the reaction transition state.
The test results have been directly fed back into the new design cycle and incorporated into broader project considerations.
The success of S88E has validated the effectiveness of our structure-based design approach. However, the failure to introduce bulkier aromatic amino acids at the L218 site (L218F/Y/W, i.e., Phenylalanine/Tyrosine/Tryptophan mutants) has helped us understand the precise spatial constraints of the active pocket, providing key reference information for subsequent designs.
We have developed and revised a bioinformatics structure prediction and fitting model (using AutoDock Vina software) to predict the binding affinity between Ochratoxin A (OTA) and the mutant enzymes we constructed. The predictions of this model (e.g., the lower binding energy of the S88E mutant) are well-correlated with our experimental results, laying a foundation for the bioinformatics-aided design of the next generation of mutants and creating a powerful tool for this purpose.
In the previous engineering cycle, we learned the mechanism from tests and guided rational design. We successfully iterated an optimized component (S88E) with significantly improved performance, which has an enzyme activity 3.7 times that of the prototype ADH3. However, in our subsequent expert interviews, it was keenly pointed out that the binding degree of the catalytic site between the enzyme and substrate, which we have successfully modified, has reached or is expected to reach the limit of this enzyme. To further improve the efficiency of OTA enzymatic hydrolysis, we can only look for isozymes with deeper catalytic pockets to break through the structural bottleneck of the prototype enzyme itself.
Therefore, we plan to screen through databases to identify enzymes that share a similar structure and identical function to ADH3. We will then conduct structural analysis and functional characterization of these enzymes, with the expectation that they will exhibit higher efficiency and stronger adaptability in the degradation of OTA.
Based on the characteristics of amide metal hydrolases and the structural analysis of ADH3, we conducted data mining in the NCBI database and obtained its isozyme---amide hydrolase LlADH derived from Lysobacter luteus. We analyzed LlADH sequence using different online tools and the websites of these tools are listed. LlADH consists of a total of 442 amino acids, with approximate molecular weight 46.45 kDa and a theoretical isoelectric point of 6.20. There are two Cys residues in its sequence, and it has no transmembrane regions.
Bioinformatics Analysis Website
We performed a Sequence alignment between LlADH and ADH3 showed 73% similarity. Based on Phyre2 structure prediction, 76% structural similarity was observed, confirming that it is the chassis component with the highest comprehensive score for us.
(Figure: Sequence Alignment between ADH3 and LlADH)
The conserved amino acids involved in coordinating the binuclear metal center are marked with green circles, among which the Asp (aspartic acid) that participates in the acid/base-catalyzed protonation of the substrate is marked with blue triangles.
By using online bioinformatics tools, we predicted that the first 19 amino acids of LlADH are signal peptides. W In the constructed vector, the first 19 amino acids of LlADH were removed, and the N-terminus was fused with a 6×His purification tag. Codon optimization was performed, and the entire gene was synthesized and subcloned into the pET46/Ek-LIC vector. The plasmid was named pET46/LlADH.
The pET46/LlADH plasmid was transformed into competent cells of the expression strain BL21(DE3) for transformation and expression. Purification was performed using nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography once. The eluate was collected according to the UV curve detected by the AKTA protein purification system, followed by ultrafiltration concentration. After ultrafiltration concentration, the protein concentration was 11.7 mg/mL. The purity of the protein was verified by SDS-PAGE, and the results are shown in the figure. The position indicated by the red arrow corresponds to the concentrated LlADH. The molecular weight of the target protein is approximately at the 45 kDa band, which is consistent with the theoretical molecular weight. With a purity of over 95%, it was confirmed that the LlADH protease with high purity and suitable experimental concentration was obtained.
(Figure: SDS-PAGE analysis of concentrated LlADH
Lane 1: marker; Lane 2: LlADH)
To verify the oligomeric state and molecular weight of the recombinant LIADH protein, we conducted a systematic analysis using various biophysical methods.
The purified LIADH was analyzed using a Superose™ 6 increase 10/300 GL chromatographic column. The results showed that LIADH presented a single symmetric peak in gel filtration chromatography, and the elution position was between the 158 kDa and 440 kDa standard proteins. Based on the theoretical molecular weight of the LIADH monomer (46.45 kDa), it is speculated that the protein may exist in the form of a homooctamer.
(Figure: Size-exclusion chromatography analysis of LlADH)
To further accurately determine the molecular weight of LIADH, we used static light scattering technology for analysis. As shown in the figure, the results indicate that the molecular weight of LIADH is approximately 351.9 kDa, which is highly consistent with the theoretical molecular weight of the octamer (~371.6 kDa), confirming that LIADH exists as a homooctamer in solution.
(Figure: Static light scattering analysis of LlADH)
The morphology and distribution of LIADH were observed by negative staining electron microscopy. The results showed that the protein particles were evenly distributed with consistent morphology, with a diameter of approximately 10 nm, and obvious four-fold symmetric structures could be observed, which further supported that LIADH is a highly homogeneous octameric protein.
(Figure: Negative staining analysis of LlADH)
We analyzed the OTA hydrolysis reaction catalyzed by LIADH using HPLC. The results, as shown in the figure, indicate that as the reaction proceeds, the OTA peak gradually decreases, and a new peak appears at the retention time corresponding to the OTα standard. After 50 minutes of reaction, OTA is completely hydrolyzed into OTα. We further verified through HPLC-MS that the molecular weight and isotope distribution of the product are consistent with those of OTα, as shown in the figure, confirming that LIADH can efficiently hydrolyze OTA to produce non-toxic OTα and phenylalanine.
(Figure: HPLC analysis of OTA and OTα standards and OTA-degrading reaction mixtures catalyzed by LlADH)
(Figure: Enzymatic characterization of purified LlADH)
(A) The temperature evaluation. (B) The pH evaluation.
(C) The metal ions evaluation. (D) The thermalstability evaluation.
All assays are performed in triplicate and the results are presented as average ± SD
We tested the enzyme reaction samples of ADH3 and LlADH under exactly the same reaction conditions and systems, with the enzyme activity of ADH3 set as 100% for comparison. The results are shown in the figure. The analysis of the OTA hydrolysis reaction results indicates that the hydrolysis activity of LlADH on OTA is twice that of ADH3, making it a more efficient amide hydrolase for hydrolyzing OTA.
(Figure: Specific activity of ADH3 and LlADH)
We used a 300kV Titan Krios cryo-electron microscope to collect LIADH data, and finally reconstructed a three-dimensional structure with a resolution of 2.67 Å from 151,233 particles.
Detailed image processing and three-dimensional reconstruction are shown in the figure. Figure A is the cryo-electron microscopy image of LlADH, and the results show that the protein particles have good uniformity. Among the 520,000 protein particles initially collected from the original images, 150,000 particles were selected for two-dimensional classification after performing offset correction and contrast transfer function evaluation using cryoSPARCv3.3.1 software. Some good particles among them were used to construct the initial template. As shown in Figure B, which is part of the two-dimensional classification results of LlADH, the protein presents a C4 symmetric unit. These selected protein particles were subjected to further three-dimensional classification, and after multiple rounds of optimization, a high-resolution protein electron density map with a resolution of 2.67Å was finally reconstructed.
(Figure: Cryo-EM data processing for LlADH)
(A)Cryo-EMmicrograph. (B)2D classification.
(C)Local resolution map. (D)FSC curve.
Using the ADH3 structure (PDB: 8IHQ) as the initial model, we used UCSF Chimera software to superimpose the LlADH structure with the electron density map obtained from electron microscopy data processing. We then performed model building, refinement, and evaluation using COOT and Phenix, ultimately obtaining an atomic model where all indicators meet the standards. It can be seen from the collated data that all the overall data are within the specified values, and the structure is reliable.
(Figure: LlADH protomer in cartoon representation)
(Figure: Superposition of LlADH onto ADH3/D344N/OTA)
OTA is shown in green sticks. Residues surrounding OTA in LlADH and ADH3/D344N/OTA are shown in magentas- and green-colored lines, respectively. The carboxylated K210 and K211 functions as a bridging ligand between the two metal ions. The binuclear metal cluster (denoted as α and β) is presented as red spheres. Distances < 3.5 Å are indicated with black dashed lines in the modeled LlADH/OTA.
Although LIADH has exhibited extremely high OTA hydrolytic activity, there are still some aspects that need improvement.
Through cryo-electron microscopy structural analysis, we have initially revealed the octameric assembly mode and active center composition of LIADH. Key residues such as H86, H254, and H272 may bind to OTA through π-stacking and hydrogen bond interactions, which provides a structural basis for subsequent enzyme activity modification.
However, the failure to obtain the OTA complex structure has limited the in-depth understanding of its substrate recognition mechanism. In the future, we can try to use substrate analogs or inert mutants (such as D345N) to capture the complex structure.
The high catalytic activity and metal independence (except for Zn²⁺) of LIADH give it significant advantages in food and feed detoxification applications. Its optimal pH (8.5) is consistent with many agricultural products processing environments, indicating potential for direct application. Future research can focus on the detoxification effect and process optimization of LIADH in real matrices (such as grains and wine).
In the previous cycle of LlADH, we have, for the first time, analyzed the structure of its catalytic pocket and its catalytic mechanism using bioinformatics and structural biology (cryo-electron microscopy). At the protein level, we optimized codons using synthetic biology, thereby increasing the expression level and purity of its protein crystals. We will utilize the effective research paradigm established in the first cycle and rationally modify this protein using synthetic biology techniques.
Through structural comparison, we found that the active site residues of ADH3, such as G88, G130, H16, etc., as well as six metal-iron coordinating residues, correspond to similar sites in LlADH like G89, G131, H164, etc., which also have a similar distribution in the LlADH/OTA structure. This indicates that these residues should also play an important role in LlADH catalysis. This further verifies the high conservation of the ADH family in catalytic mechanisms, especially in terms of coenzyme binding and metal ion stabilization. The spatial arrangement of key residues determines substrate specificity and reaction efficiency.
(Figure: Superposition of LlADH onto ADH3/D344N/OTA)
OTA is shown in green sticks. Residues surrounding OTA in LlADH and ADH3/D344N/OTA are shown in magentas- and green-colored lines, respectively. The carboxylated K210 and K211 functions as a bridging ligand between the two metal ions. The binuclear metal cluster (denoted as α and β) is presented as red spheres. Distances < 3.5 Å are indicated with black dashed lines in the modeled LlADH/OTA.
In the modeling structure, the six amino acid residues L219, G131, V348, G89, G91, and I326 are farther from OTA than other OTA-binding amino acid residues. Therefore, it is speculated that replacing these residues to introduce additional substrate-enzyme interactions may increase the enzymatic activity of LlADH.
Using the pET46/LlADH plasmid as a template, site-directed mutagenesis of LlADH was performed via polymerase chain reaction with the primers designed above to construct a series of mutant plasmids.
After correct sequencing, expression and purification are performed. The expression and purification process of the mutant protein is the same as that of the LlADH wild type.
Finally, S89D, S89F, S89H, S89K, S89R, S91R, G131F, L219H, L219K, L219R, I326A, V348F, V348H, and V348S were obtained. The purified proteins were verified by SDS-PAGE. The concentration results of LlADH, S91R, L219H, L219K, L219R, I326A, V348F, V348H, and V348S are shown in the figure, and the concentration results of S89D, S89F, S89H, S89K, and S89R are shown in the figure. The size of all concentrated mutant proteins is around 45 kDa, and the protein purity is suitable for subsequent enzyme activity experiments.
(Figure: SDS-PAGE analysis of LlADH variants)
A：1：marker；2：WT； 3-10：S91R、L219H、L219K、L219R、I326A、V348F、V348H、V348S B：1：marker；2-6：S89D、S89F、S89H、S89K、S89R；7：WT
Based on the above analysis results, using the established HPLC detection system, the enzyme activities of LlADH and its mutants were analyzed, and the results are shown in the figure:
(Figure: LlADH engineering)
The enzyme activity of variants was conducted in triplicate and the average values with the error bars representing standard deviations are shown. ND, not detectabl
In this round, based on the structural analysis of LlADH, potential sites were predicted through structural stacking analogy, and multi-directional mutations were formulated. After a complete design-build-test process, mutation sites that can significantly improve enzyme activity were identified, and the enzyme activity test was successfully passed. The next step will focus on combinatorial mutations of key residues to further optimize the spatial configuration and charge distribution of the substrate-binding pocket, aiming to synergistically enhance catalytic efficiency and substrate specificity. Meanwhile, the stability of the mutants under different reaction conditions will be verified to provide data support for subsequent large-scale applications. Virtual experiment verification will combine molecular dynamics simulations and free energy calculations. Additionally, efforts will be made to explore the high-efficiency expression process of LlADH in food-safe industrial strains such as Pichia pastoris and optimize fermentation conditions to improve the soluble expression level and catalytic activity.
Additionally, it is exciting that our attempts to express the engineered protein in Pichia pastoris have shown relatively stable and effective relative enzymatic activity efficiency.The relative enzymatic activity tests demonstrated that our engineered protein could be efficiently expressed in both eukaryotic and prokaryotic model organisms, with no significant differences observed.
(Figure: Relative enzyme activity result)
According to literature and team enzyme activity testing, Fpro enzyme has been proven to efficiently degrade AFB1, OTA, and ZEN, three fungal toxins, with relatively low degradation rates. We found in subsequent experiments that the enzyme also has the ability to degrade Patulin, achieving a preliminary functional leap of "one enzyme, four effects".
Through electron microscopy analysis, we obtained the empty structure of Fpro, as shown in the schematic diagram below.
The team has conducted rational design mutations for Fpro, but the mutant enzyme activity test results have the following issues: