Results
Model Head

Overview

We systematically explored the core modules of "Antifreeze protein–Antifreeze chassis–Synergistic antifreeze system," achieving the following key advancements:

1. Successful construction and validation of 11 natural antifreeze proteins (AFPs) for expression and functional characterization;
2. Design and optimization of multiple non-natural antifreeze proteins, with some exhibiting activity comparable to natural proteins;
3. Demonstration of good biocompatibility, showing translational potential for applications in the cosmetics sector.
4. Development of an engineered antifreeze chassis based on EBY100 surface-display antifreeze proteins, addressing challenges in protein purification;
5. Novel construction of a "trehalose-antifreeze protein" synergistic antifreeze system, significantly enhancing freeze resistance.

Through strain freezing survival rate assays and ice recrystallization inhibition (IRI) activity assays, we characterized the performance of antifreeze proteins and antifreeze systems, demonstrating superior efficacy. Additionally, guided by feedback from Human Practices (HP), we evaluated the biocompatibility of selected antifreeze proteins. The results revealed minimal cytotoxicity, a characteristic that provides a critical foundation for a frost-resistant skincare product.

In the final phase of our research, we further explored two strategic approaches to enhance the antifreeze system:

1. Multiplexed Site-Specific Rearrangement System：We aim to identify optimal protein combinations through combinatorial screening, utilising high-throughput screening methods to evaluate the synergistic anti-freeze effects of different protein combinations.

2. Multiplexed Site-Specific inversion System：We aim to optimize the expression ratios of different proteins by controlling promoter strength.

However, the two systems related to Multiplexed Site-Specific mentioned above are significant challenges for us. Compounded by time constraints and experimental obstacles, only partial progress has been achieved thus far. Moving forward, we shall dedicate ourselves to continuous refinement and improvement.

Part 1. Antifreeze Proteins

Antifreeze proteins have been widely identified across various cold tolerant species, where they facilitate organismal adaptation to low temperature environments by lowering the freezing point of bodily fluids and reducing ice crystal size. To date, several natural antifreeze proteins have been extracted and purified, and incorporated into cream formulations as active antifreeze agents in cosmetic products. Based on this background, our laboratory has embarked on an exploration of antifreeze proteins, conducting comprehensive characterizations of their IRI activity assays and biocompatibility assays.

1.Construction and Characterization of Natural Antifreeze Proteins
1) Objective

To identify and synthesize natural antifreeze proteins with favorable functional properties, verify their antifreeze activity, and investigate in depth the relationship between antifreeze efficacy and structural characteristics of the proteins.

2) Plasmid Construction

We performed literature surveys and protein library searches to identify eleven antifreeze proteins that are representative of their respective families. After consulting with Prof. Lei Zhang, we obtained pET28a-AFPx (1-11) plasmids capable of expressing corresponding antifreeze proteins in Escherichia coli through synthetic gene technology.

Table 1. Information on Natural Antifreeze Proteins

Figure 2 IRI activity of crude enzyme (AFP4/8/9/10) Figure 2 IRI activity of crude enzyme (AFP4/8/9/10)
3) Protein Expression

To verify that the antifreeze proteins can be correctly expressed in E. coli, four engineered strains harboring antifreeze proteins with particularly prominent helical structures—namely pET28a-AFP4, pET28a-AFP8, pET28a-AFP9, and pET28a-AFP10—were selected as representative examples. They were Cultivated and induced with isopropyl-β-D thiogalactopyranoside (IPTG) (refer to protocol link).Then the target proteins were then extracted and purified. The purified proteins were analyzed by SDS PAGE.

Figure 2 IRI activity of crude enzyme (AFP4/8/9/10)
Figure 1. SDS PAGE analysis of AFP4,AFP8,AFP9,AFP10

The pET28a-AFP4 plasmid is designed to express AFP4 with a anticipated molecular weight of 27.64 kDa. The pET28a-AFP8 plasmid is designed to express AFP8 with a anticipated molecular weight of 23.5 kDa. The plasmid pET28a-AFP9 is designed to express AFP9 with an anticipated molecular mass of 25.46 kDa. The plasmid pET28a-AFP10 was is designed to express AFP10 with a anticipated molecular mass of 26.05 kDa. All target bands and expected bands are consistent.

Therefore, based on these results, we conclude that our antifreeze proteins are properly expressed in E.coli.

4) Characterization

To further validate the antifreeze activity, we evaluated IRI activity of the four antifreeze proteins (AFP4, AFP8, AFP9, and AFP10). Specifically, the four engineered bacterial strains were cultured and induced with IPTG for target protein expression. After induction, the cells were harvested by centrifugation, resuspended in PBS, and disrupted to obtain crude cell lysates. The supernatants were then subjected to IRI activity (refer to protocol link).

Figure 2 IRI activity of crude enzyme (AFP4/8/9/10)
Figure 2. Characterization of IRI activity in crude enzyme lysates containing AFP4, AFP8, AFP9, and AFP10. (a) Ice crystal images of E. coli crude enzyme lysates after freezing at −6 °C for 30 min.(b) Mean ice crystal size analysis. Sta: The wild-type E. coli crude enzyme lysate (control group).

Based on the experimental results, the crude enzyme extracts containing AFP4, AFP8, AFP9, and AFP10 reduced the average ice crystal size by approximately half after freezing compared to the control group, indicating that the obtained natural antifreeze proteins (AFPs) significantly inhibit ice crystal growth. Furthermore, combined with literature review ,our summary reveals that AFPs enriched in threonine and cysteine residues exhibit superior ice recrystallization inhibition (IRI) activity . Both α-helical and β-helical structures were observed to play critical and distinct roles in the ice-binding process. These findings establish a foundational framework for the rational design of non-natural Antifreeze Protein in subsequent studies.

2.Construction and Characterization of Non-Natural Antifreeze Proteins
2.1 Point Mutation Design
1) Objective

To validate the performance of the point mutant proteins--MUTx （x=1, 2, 3, 4）via the ESM-1v model through dry-lab simulations.

2) Plasmid construction

We obtained pET28a-MUTx （x=1, 2, 3, 4） plasmids capable of expressing corresponding antifreeze proteins in E. coli through synthetic gene technology.

Figure 3 Sequence alignments before/after mutation
Figure 3. pET28a-MUTx (x=4, 8, 9, 10) series plasmids
3) Characterization

The newly constructed E. coli strains harboring the pET28a-MUTx (x = 4, 8, 9, 10) plasmid series were subjected to induced expression. After induction, the cells were processed using the same method(refer to protocol link), and the crude cell lysates were collected for characterization of IRI activity.

Figure 4 IRI activity of AFP vs MUT series
Figure 4. Characterization of IRI activity in crude enzyme lysates containing AFP4, AFP8, AFP9, AFP10，MUT4, MUT8, MUT9, MUT10 protein. (a) Ice crystal images of E. coli crude enzyme lysates after freezing at −6 °C for 30 min.(b) Mean ice crystal size analysis. Sta:The wild-type E. coli crude enzyme lysate (control group).

MUT9, MUT10, and MUT4 proteins exhibited certain antifreeze activity. Notably, the crude lysate of MUT4 reduced the average ice crystal size by one-fourth after freezing, indicating detectable antifreeze functionality in the engineered proteins. However, none of the mutants outperformed the natural antifreeze protein, which may be attributed to the considerable challenge of achieving superior activity through limited point mutations under a restricted Number of mutations. This outcome remains within expected bounds. Given the limited number of mutable sites, identifying high-performance AFPs requires iterative rounds of screening and extensive experimental effort. However, due to the limited experimental timeline, we have decided to broaden the scope of modifications, aiming to achieve non-natural antifreeze activity that surpasses that of natural proteins.

2.2 Domain Fusion
1) Objective

To selectively fuse specific domains from antifreeze proteins--wpyx protein,x=1, 2, 3, 4) of different origins and verify whether their ice-binding activity is enhanced.

2) Construction

Flexible linkers (GGGGSGGG) and rigid linkers (EAAAKEAAAKEAAAK) were used to fuse specific domains derived from distinct antifreeze proteins.

Subsequent Dry Lab predicted that all four designed domain-fused proteins function as antifreeze proteins, with their comprehensive ability ranking as follows: wpy3 > wpy4 > wpy2 > wpy1(refer to model link). We obtained pET28a-wpyx （x=1, 2, 3, 4）plasmids in capable of expressing corresponding antifreeze proteins in E.coli through synthetic gene technology.

Figure 4 IRI activity of AFP vs MUT series
Figure 5. pET28a-wpyx (x=1, 2, 3, 4) series plasmids
3）Characterization

The newly constructed E. coli strains harboring the pET28a-wpyx (x = 1, 2, 3, 4) plasmid series were subjected to induced expression. After induction, the cells were processed using the same method(refer to protocol link), and the crude cell lysates were collected for characterization of IRI activity.

Figure 4 IRI activity of AFP vs MUT series
Figure 6. Characterization of IRI activity in crude enzyme lysates containing wpy1,wpy2,wpy3,wpy4. (a) Ice crystal images of E. coli crude enzyme lysates after freezing at −6 °C for 30 min.(b) Mean ice crystal size analysis. The wild-type E. coli crude enzyme lysate (control group).

We observed that the antifreeze efficacy of the four newly designed domain-fused antifreeze proteins (wpy1, wpy2, wpy3, wpy4) did not align with the predicted outcomes. Based on experimental validation, we conclude that the current design and prediction methodologies are flawed. Specifically, the prediction approach relied solely on primary structure superposition analysis and failed to account for interdomain interactions, rendering it inadequate for evaluating the antifreeze activity of multidomain fusion proteins. Consequently, we intend to adopt an alternative design strategy.

2.3 Inverse Folding Optimization Design
1) Objective

To validate the antifreeze activity of AFPs-IF (s=8, 9, 10) series proteins, which were computationally designed via an inverse folding strategy to optimize the natural proteins 6A8K (AFP10), 4NU2 (AFP9), and 3WP9 (AFP8).

2) Plasmid construction

We obtained pET28a-AFPs-IF (s=8,9,10) plasmids in capable of expressing corresponding antifreeze proteins in E.coli through synthetic gene technology.

Figure 5 Design of fusion proteins
Figure 7. pET28a-AFPs-IF（s=8,9,10）series plasmids
3)Characterization

Similarly, employing the identical induction and cultivation protocol, we validated the IRI (Ice Recrystallization Inhibition) activity of the proteins. (refer to protocol link).

Figure 5 Design of fusion proteins
Figure 8. Characterization of IRI activity in crude enzyme lysates containing 6A8K-IF、4NU2-IF、3WP9-IF. (a) Ice crystal images of E. coli crude enzyme lysates after freezing at −6 °C for 30 min.(b) Mean ice crystal size analysis. The wild-type E. coli crude enzyme lysate (control group).

According to the IRI activity assay, the average ice crystal size after freezing was significantly reduced in the crude enzyme lysate of 4NU2-IF (AFP9-IF), decreasing by approximately one-third compared to the control group. Its ice-recrystallization inhibition potency was comparable to that of the natural antifreeze protein AFP10, indicating that the inverse folding-based design strategy yielded functionally effective variants. However, the ice crystal size in the crude lysate of 4NU2-IF remained larger than that of AFP9 (Fig.2 and Fig.8). suggesting that further refinement of the optimization strategy is still required. Therefore, continuous refinement of the optimization strategy is still required.

2.4 De Novo Synthesis Based on Inverse Folding Optimization Design
1) Objective

To validate the antifreeze activity of unnaturally designed antifreeze proteins--De Novo x（x=1-6）generated de novo through an inverse folding-based optimization strategy.

2) Plasmid construction

We obtained pET28a-De Novo x (x=1–6) plasmids in capable of expressing corresponding antifreeze proteins in E.coli through synthetic gene technology.

Figure 8 IRI activity of FUS series
Figure 9. pET28a-De Novo x（x=1-6）series plasmids
3) Characterization

The de novo synthesized proteins were subjected to IRI activity assay under identical induction and cultivation conditions. refer to protocol link.

Figure 10 IRI activity of FUS series
Figure 10. Characterization of IRI activity in crude enzyme lysates containing De Novo x（x=1-6）. (a) Ice crystal images of E. coli crude enzyme lysates after freezing at −6 °C for 30 min.(b) Mean ice crystal size analysis. The wild-type E. coli crude enzyme lysate (control group).

The results demonstrated that approximately half of the De Novo designed antifreeze protein variants exhibited IRI activity comparable to that of natural antifreeze proteins. These experimental findings have been feedback to the dry-lab, to guide the refinement of its computational design strategy. We anticipate that further iterative optimization based on this feedback will enhance both the quality and efficacy of the designed non-native fusion proteins.

Figure 8 IRI activity of FUS series
Figure 11. Summary of IRI activity characterization results for crude enzyme extracts

Analysis reveals that through iterative exchange and refinement between dry-lab and wet-lab experiments, our design strategy has been continuously optimized. Consequently, the functional performance of the designed non-natural antifreeze proteins has progressively improved. Notably, the De Novo designed variants exhibited superior activity among all non-natural constructs, successfully meeting our intended objectives.

3.Biocompatibility Characterization

Through feedback from HP, we recognized that biocompatibility among ingredients in cosmetic formulations is particularly critical. Consequently, we conducted cytotoxicity tests using selected antifreeze proteins as representatives (Fig. 13). Specifically, we extracted and purified AFP4, AFP8, AFP9, and MUT4 proteins, followed by SDS-PAGE analysis (Fig. 1 and Fig. 12).

Figure 7 Design of de novo antifreeze proteins
Figure 12. SDS-PAGE analysis of MUT4

The plasmid pET28a-MUT4 was designed to express MUT4 with a anticipated molecular weight of 26.00 kDa. Compared to the control group, an additional band was observed in the experimental group at a position slightly above 25 kDa, indicating successful expression of the MUT4 protein. Subsequently, the purified protein solutions of AFP4, AFP9, AFP8, and MUT4 were co-cultured with L929 cells. After 48 hours of incubation, their cytotoxic effects on L929 cells were assessed(refer to protocol link).

Figure 8 Design of de novo antifreeze proteins
Figure 13. Cytotoxicity assessment of antifreeze proteins in L929 cells.(a) Summary of experimental data.(b) Cytotoxicity evaluation of antifreeze proteins against L929 cells seeded at 2,500 cells per well.(c) Cytotoxicity evaluation of antifreeze proteins against L929 cells seeded at 5,000 cells per wel

Through systematic evaluation at two distinct initial cell seeding densities, we assessed the cytotoxicity of four protein samples (AFP-4, AFP-8, MUT-4, AFP-9) against L929 murine fibroblast cells. Overall, all four proteins exhibited minimal cytotoxicity and in some cases even mild proliferative effects across all tested conditions, indicating favorable biocompatibility of the selected and engineered antifreeze proteins. These results support their potential incorporation into cream-based formulations as antifreeze-active ingredients.

Part 2. Construction and Characterization of Antifreeze Systems

We selected the S.cerevisiae EBY100 strain as our chassis organism and engineered this system to display antifreeze proteins on its surface while concurrently expressing trehalose intracellularly, thereby constructing a synergistic Antifreeze system with both extracellular and intracellular mechanisms. The resulting engineered strain was evaluated through freezing survival rate assays and ice recrystallization inhibition (IRI) activity tests.

1. Antifreeze Chassis Systems Based on Surface-Displayed Antifreeze Proteins
1.1 GFP Functionality Test
1) Objective

To validate that EBY100 yeast can successfully display proteins on its surface.

2) Plasmid Construction

We obtained the GFP fragment via PCR from a laboratory plasmid and successfully constructed the pR_DK01 plasmid through Gibson assembly into the pYD1 vector. This was subsequently transformed into EBY100 yeast cells.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 14. Construction of the pR_DK01 plasmid
Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 15. Colony PCR verification to validate the pR_DK01.

Through colony PCR, we confirmed that pR_DK01 was successfully constructed, and then we transformed it into EBY100 cells.

3) Validation

Following the successful construction of the EBY100 yeast strain, we cultured and induced EBY100-pR_DK01 (refer to protocol link), subsequently observing fluorescence intensity under a fluorescence microscope.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 16. Fluorescence micrographs of GFP yeast surface display. (a) shows the surface display of EBY100 strain containing pR_DK01 after induced expression, captured with a 200ms exposure under fluorescence. (b) the surface display of protoplasts prepared from this strain, imaged with a 1s exposure under fluorescence microscopy.
1.2 Antifreeze Chassis Systems Based on Natural Antifreeze Proteins
1) Objective

To validate the frost resistance of a surface-displayed system utilising 11 natural AFPs, and to compare its performance against yeast strains expressing AFP intracellularly.

2) Plasmid and Frost-Resistant Strain Construction

We amplified AFP (1-11) fragments via PCR from the pET28a-AFP series of plasmids. Through sequential Gibson assembly into the pYD1 vector, we successfully obtained plasmids pR_DK02 to pR_DK12. whereby the Aga2p sequence was removed from pR_DK10 (displaying AFP9) and pR_DK11 (displaying AFP10) to generate pR_DK13 to pR_DK14. These plasmids were subsequently transformed into EBY100 yeast cells.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 17. Construction of the pR_DK02-14 plasmid
Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 18. Colony PCR verification to validate the PR_DK02 to PR_DK14. (a) Lanes from left to right show junctions for plasmids pR_DK02 to pR_DK07, upper/lower junctions for pR_DK08, upper/lower junctions for pR_DK09, upper/lower junctions for pR_DK10, upper/lower junctions for pR_DK11, upper/lower junctions for pR_DK12.(b).Lanes from left to right show junctions for plasmids pR_DK13, pR_DK14.

Through colony PCR, we confirmed that pR_DK02 to pR_DK14 were successfully constructed. Subsequently, we successfully introduced these plasmids into the EBY100 yeast strain.

3) Characterisation

Following the successful construction of the EBY100 yeast strain, it was subjected to induced cultivation. The induced strain underwent IRI activity assays and freezing survival rate characterisation (refer to protocol link).

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 19. Characterisation of freeze tolerance in AFP1-11, pYD1, and GFP EBY100 strains. (a) IRI activity assay. Ice crystal images obtained from 10% glycerol PBS yeast suspensions frozen for 30 minutes at -12°C. (b) IRI activity assay – analysis of average ice crystal particle size. (c) Growth patterns following two freeze-thaw cycles at -80°C (30-minute intervals) and subsequent plate culture. (d) Survival rate post-freeze-thaw. pYD1 denotes 10% glycerol PBS yeast broth containing empty pYD1 plasmid (control group).

Compared to the pYD1 strain lacking AFPs and the chassis strain displaying GFP on its surface, our constructed antifreeze chassis—which relies on the surface display of 11 natural frost-resistant proteins—demonstrated varying degrees of frost resistance. Among these, chassis displaying AFP2 and AFP3 reduced the average ice crystal diameter to approximately half that of the control group (Fig. 19b). This indicates that our AFPs successfully conferred superior IRI performance to the chassis strains, enabling effective defence against mechanical damage caused by ice crystals under cold conditions. Concurrently, all 11 antifreeze chassis strains exhibited higher survival rates than the control group (Fig. 19d). Yeast strains displaying AFP1 and AFP8 demonstrated the highest survival rates exceeding 70%, representing 4.34-fold and 3.66-fold improvements over the control respectively. This confirms the significant cold tolerance of our freeze-resistant chassis.

Concurrently,surface-displaying strains exhibited higher IRI activity and cell survival rates than intracellularly expressing strains(Fig. 20). This validates the efficacy and superiority of our strategy to construct the antifreeze chassis by introducing a yeast surface display system.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 20. Characterisation of freeze tolerance in pYD1、GFP、AFP9、AFP10、AFP9△Aga2p、AFP10△Aga2p EBY100 strains. (a) IRI activity assay. Ice crystal images obtained from 10% glycerol PBS yeast suspensions frozen for 30 minutes at -12°C. (b) IRI activity assay – analysis of average ice crystal particle size. (c) Growth patterns following two freeze-thaw cycles at -80°C (30-minute intervals) and subsequent plate culture. (d) Survival rate post-freeze-thaw. pYD1 denotes 10% glycerol PBS yeast broth containing empty pYD1 plasmid (control group).
1.3 Antifreeze Chassis Systems Based on Non-Natural Antifreeze Proteins
1) Objective

To construct and validate the frost resistance capacity of the antifreeze chassis system derived from non-natural AFPs, obtained through point mutations and key helix substitutions, when displayed on the surface of EBY100.

2) Plasmid and Frost-Resistant Strain Construction

We amplified the MUT4 fragment from pET28a-MUT4 via PCR and the gene fragments were successfully ligated onto the pYD1 vector through Gibson assembly, thereby constructing a complete plasmid pR_DK15. Through key helix substitutions: natural helix sequences (derived from AFP9 and AFP10) were substituted into the AFP11 protein structure to yield pR_DK16 and pR_DK17; non-natural helix sequences were substituted into the AFP11 structure to yield pR_DK18 and pR_DK19. These constructs were subsequently transformed into EBY100 yeast cells.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 21. Construction of the pR_DK15-19 plasmid
Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 22. Colony PCR verification to validate the PR_DK15 to PR_DK19. (a) Verification band for pR_DK15 plasmid.(b) Plasmid verification gel image for plasmid pR_DK16. (c) Plasmid verification gel image for plasmid pR_DK17. (d) Plasmid verification gel image for plasmid pR_DK18 and pR_DK19.

Through colony PCR, we confirmed that pR_DK15 to pR_DK19 were successfully constructed, and then we transformed them into EBY100 cells.

3) Characterisation

Following the successful construction of the EBY100 yeast strain, it was subjected to induced cultivation. The induced strain underwent IRI activity assays and freezing survival rate characterisation (refer to protocol link).

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 23. Characterisation of freeze tolerance in pYD1、AFP4、MUT4、AFP11、NEW11-9、NEW11-10、NEW11-A、NEW11-B EBY100 strains. (a) IRI activity assay. Ice crystal images obtained from 10% glycerol PBS yeast suspensions frozen for 30 minutes at -12°C. (b) IRI activity assay – analysis of average ice crystal particle size. (c) Growth patterns following two freeze-thaw cycles at -80°C (30-minute intervals) and subsequent plate culture. (d) Survival rate post-freeze-thaw. pYD1 denotes 10% glycerol PBS yeast broth containing empty pYD1 plasmid (control group).

By displaying point mutations and the non-natural proteins with key helix substitutions demonstrated improved IRI activity and cellular viability compared to the control group. This indicates we have successfully enhanced the cold tolerance of the antifreeze chassis, achieving an optimised upgrade of the fundamental antifreeze system.

2. Synergistic Antifreeze Systems
2.1 Enhancement of the trehalose metabolic pathway
1) Objective

To validate the freeze resistance of EBY100 yeast with an enhanced trehalose metabolic pathway.

2) Plasmid Construction

Employing a narrow-down strategy within yeast cells, a plasmid containing the trehalose metabolic pathway was screened and constructed. Relevant fragments (key Gene1-key Gene2-key Gene3, key Gene4-key Gene5) were obtained from the yeast genome via PCR and Overlap Extension PCR. The resulting plasmid pRS413-trehalose was then transformed into EBY100. Meanwhile, the empty pRS413 plasmid was transformed into the EBY100 strain to serve as the control.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 24. Construction of the pRS413-trehalose
Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 25. PCR results for fragments associated with the construction of plasmid pRS413-trehalose. From left to right:fragment key Gene1-key Gene2-key Gene3 (4885 bp) and linearised pRS413 vector backbone (4762 bp), key Gene4-key Gene5 fragment (4002 bp)

As can be seen from the figure, we have successfully obtained the target fragment. Subsequently, we constructed the complete plasmid through Gibson assembly.

Figure 1 Genetic circuit schematic of plasmid pR_DK01
Figure 26. Colony PCR verification to validate the pRS413-trehalose and pRS413. From left to right: the junctions of plasmid pRS413 and plasmid pRS413-trehalose , each with a target length of approximately 1000 bp

Through colony PCR, we confirmed that pRS413-trehalose plasmid was successfully constructed, and then we transformed it into EBY100 cells.

3) Expression validation

We shall inoculate the successful strain into SC-His medium, conduct induction cultivation in accordance with standard operating procedures for microbial fermentation engineering, extract trehalose using non-targeted metabolomics extraction methods (refer to protocol link), and subsequently analyse and identify the product via LC-MS/MS.

pR_DK15 plasmid genetic circuit schematic
Figure 27. LC MS analysis of trehalose in different yeast strains. (a) LC MS chromatograms of the trehalose standard. (b) LC MS chromatograms of the EBY100- pRS413 strain. (c) LC MS chromatograms of the engineered EBY100 pRS413 trehalose strain. (d) Quantitative comparison of trehalose content between EBY100- pRS413 and EBY100 pRS413 trehalose strains.

EBY100-pRS413 exhibited the single expected peak in the m/z 360>85 channel, whilst an additional extraneous peak appeared in the m/z 360>163 channel. The prominence of this non-target peak is likely attributable to the low trehalose production in the control strain, resulting in a low main peak that makes other signals appear relatively significant (Fig. 27b). In contrast, the engineered strain EBY100-pRS413-trehalose showed a single peak in both channels, with a peak shape closely matching that of the standard, indicating successful and relatively pure trehalose expression in this strain (Fig. 27c). Further analysis (Fig. 27d) confirmed a significantly higher trehalose content in the engineered strain compared to the control. These results collectively demonstrate that introducing the pRS413-trehalose plasmid into EBY100 cells effectively enhanced trehalose yield.

4) Effect validation

We concurrently cultured both EBY100-pRS413 yeast and the EBY100-pRS413-trehalose yeast. Post-cultivation strains underwent IRI activity testing and freezing survival rate characterisation (refer to protocol link).

Colony PCR gel electrophoresis image. Verification band for pR_DK15 plasmid
Figure 28. Characterisation of freeze tolerance in EBY100- pRS413，and EBY100-pRS413-trehalose yeast strains. (a) IRI activity assay. Ice crystal images obtained from 10% glycerol PBS yeast suspensions frozen for 30 minutes at -24℃. (b) IRI activity assay – analysis of average ice crystal particle size. (c) Yeast strain viability testing at -24℃.

EBY100-pRS413-trehalose yeast strain, which additionally introduced the pRS413 trehalose plasmid, exhibited higher IRI activity and survival rates than the control group. This indicates that enhanced trehalose metabolism in the yeast strain significantly improves the freeze tolerance of the chassis organism, with results meeting expectations.

2.2 Synergistic Antifreeze System Combining Trehalose and Antifreeze Proteins
1) Objective:

To validate the freeze-resistant capacity of the trehalose-antifreeze protein synergistic antifreeze system.

2) Strain Construction:

The pRS413-trehalose plasmid shall be transformed into EBY100 yeast strains displaying AFP4, AFP8, AFP9, and AFP10.

Plasmid transformation verification gel image for EBY100-pRS413, EBY100-pRS413-trehalose
Figure 29. Plasmid transformation verification gel image

Through colony PCR, we confirmed that pRS413-trehalose plasmid was successfully transformed into EBY100 yeast displaying the antifreeze protein on its surface.

3) Characterisation:

Following the successful construction of the EBY100-AFP- pRS413-trehalose yeast strain, it was subjected to induced cultivation. The induced strain underwent IRI activity assays and freezing survival rate characterisation (refer to protocol link).

LC-MS analysis of trehalose in different yeast strains
Figure 30. Performance characterization diagram of four synergistic antifreeze chassis. (a) IRI activity test: ice crystal images of yeast suspensions in 10% glycerol PBS after freezing at -12°C for 30 minutes. (b) Bar graph of ice crystal average particle size analysis.

The experimental results clearly demonstrate that synergistic antifreeze system combining trehalose and antifreeze proteins displays significantly higher antifreeze activity than strains expressing trehalose alone or displaying antifreeze proteins solely on their surfaces. These findings are consistent with our expectations.

Part 3. In-depth Optimisation and Construction of the Antifreeze System

We have attempted two strategies: Multiplexed Site-Specific Rearrangement System and Multiplexed Site-Specific inversion System, to conduct in-depth optimisation and upgrading of the antifreeze system.

3.1 Construction of Multiplexed Site-Specific Rearrangement System
1) Objective

To screen for the optimal protein combination.

2) Plasmid Construction

We obtained the AFP gene fragments by PCR amplification from the pET28a-AFP series plasmids, and the linearized pTJ25-CL vector was obtained by PCR amplification from BioBrick. Then, multiple rearrangement units are obtained through OE-PCR amplification, and finally the complete plasmid is obtained through Gibson assembly (refer to protocol link).

LC-MS analysis of trehalose in different yeast strains
Figure 31. Construction of pTJ25-CL for the Multiplexed Site-Specific Rearrangement System
3.2 Construction of Multiplexed Site-Specific inversion System
1) Objective

To control promoter strength and optimise the expression ratio of different proteins.

2) Plasmid construction

We obtained the AFP fragments from the pET28a-AFP series plasmids by PCR amplification, and then obtained the promoter library-related fragments through the same method from the plasmid obtained from the laboratory. Finally, we obtained multiple inverted units through OE-PCR, and through Gibson assembly, we obtained the complete plasmid pTJ25-MI.

LC-MS analysis of trehalose in different yeast strains
Figure 32. Construction of pTJ25-MI for the Multiplexed Site-Specific inversion System
LC-MS analysis of trehalose in different yeast strains
Figure 33. PCR results for target fragments in the Multiplexed Site-Specific Rearrangement System and the Multiplexed Site-Specific inversion System (a) The Multiplexed Site-Specific Rearrangement System: From left to right: vector backbone (4000 bp), AFP1 + T7 terminator + pBAD promoter + AFP2 (1218 bp), T7 terminator + pBAD promoter + AFP3 (945 bp), T7 terminator + pBAD promoter + AFP4 (939 bp), T7 terminator + pBAD promoter + AFP5 (924 bp), T7 terminator + pBAD promoter + AFP6 (804 bp), T7 terminator + pBAD promoter + AFP7 (1527 bp), T7 terminator + pBAD promoter + AFP8 (1239 bp), T7 terminator + pBAD promoter + AFP9 (1305 bp), T7 terminator + pBAD promoter + AFP9 + T7 terminator (1523 bp). (b) The Multiplexed Site-Specific inversion System: From left to right: Vector backbone (3464 bp), Promoter library (7658 bp), T12 + PHXT7 (681 bp), TDH2t + PTDH1 (748 bp), pBIO2-T1 (787 bp), T12 + PHXT7 + AFP9 (1480 bp), pBIO2-T1 + AFP8 (1520 bp), TDH2t + PTDH1 + AFP10 (1574 bp)

For both systems, we have now completed construction of the primary rearrangement and inversion units (Fig. 33). However, this undertaking presents considerable challenges and remains incomplete due to time constraints and experimental obstacles. Nevertheless, it holds significant importance for both antifreeze protein library construction and enhancing the moisturising and freeze-protection efficacy of cosmetics under cold conditions. We are confident in supplementing and refining this work in the future. Moving forward, we shall employ Overlap Extension PCR and Gibson assembly to link existing core modules into pTJ25-CL and pTJ25-MI constructs. This will enable characterisation and screening of the completed 'random rearrangement system' and 'multiple inversion system', facilitating intelligent combination and proportional regulation of diverse antifreeze proteins. Our ultimate goal is to establish a multifunctional, programmable antifreeze protein library.

Discussion and Conclusion

Our research systematically progresses from natural antifreeze proteins (AFPs) to non-natural protein design, chassis construction, and synergistic mechanism exploration, forming a comprehensive, tiered 'protein–system–application' research chain.

We successfully expressed 11 natural antifreeze proteins, with IRI experiments confirming their significant ice crystal growth inhibition capacity. This provides foundational structural and functional templates for subsequent protein design. Through four strategies—point mutation, domain fusion, inverse folding strategy and De Novo design—we explored pathways for constructing non-natural antifreeze proteins. Among these, the inverse folding strategy and De Novo design demonstrated the greatest potential, with some protein activities approaching those of natural counterparts. This highlights the powerful application prospects of computationally-aided design in protein engineering. However, the domain fusion strategy failed to achieve the anticipated results, indicating that the fusion approach, linker selection, and host expression system significantly influence protein correct folding and function.

Compared to intracellular expression, frost-resistant strains displaying antifreeze proteins on their surfaces demonstrated superior IRI activity and cryopreservation survival rates. This approach not only enhances functional efficiency but also circumvents complex purification procedures, offering a viable pathway for industrial manufacture.

We pioneered and implemented the 'trehalose-antifreeze protein synergistic antifreeze strategy', significantly enhancing the freeze tolerance of chassis strains by intensifying intracellular trehalose metabolism in tandem with surface-displayed antifreeze proteins. Experimental results unequivocally demonstrate that the synergistic system exhibits superior antifreeze activity compared to single mechanisms, providing both theoretical foundations and technical support for subsequent multifunctional chassis design.

Finally, all tested proteins exhibited minimal toxicity in L929 cells, with some even promoting cell proliferation, demonstrating favorable biocompatibility. This lays the groundwork for future applications in cosmetics, pharmaceuticals, and related fields.