Throughout human history, the pursuit of protection from cold has evolved from a fundamental need for survival assurance to a refined quest for quality—progressing from basic heating sources and insulating clothing toward modern lifestyles that integrate functionality with aesthetic appeal. However, amid today’s intensifying climate change, the increasing frequency of extreme cold events has made frostbite and cold-induced skin conditions a growing concern that impairs daily life quality . Against this backdrop, conventional cold-protection methods have become increasingly inadequate to meet society’s heightened expectations: what is urgently called for today is a holistic solution that unites protection efficacy, user experience, and aesthetic appeal.

Figure 1. Global snow cover extent, derived from MODIS on NASA's Terra satellite

Figure 2. Prevalence of skin conditions across climate zones. (Data from our human practice survey)

In the fashion and cosmetics industry, effective protection under low-temperature conditions has remained a persistent challenge. Current products face three major limitations: first, functional constraints, where conventional formulations not only exhibit reduced efficacy in extreme cold but also suffer from physical instability; second, experiential compromises, as over-simplified formulations designed for cold resistance fail to meet dual consumer expectations for texture and aesthetic appeal; and third, a deeply ingrained industry mindset that often prioritizes style over comfort. At the core of these challenges lies the unresolved issue of how to achieve organic integration of protective performance, user experience, and aesthetic appeal in low-temperature contexts .

In recent years, enterprises and R&D professionals continue to explore pathways for enhanced cold resistance.

On the ingredient and formulation front, most brands utilize components such as ceramides, cholesterol, and free fatty acids to mimic the skin’s intercellular matrix . While this approach partially reinforces the physical barrier, it fails to address the fundamental issues of cellular dehydration and ice crystal damage under cold conditions. Existing formulations largely remain at the stage of passive repair, lacking active antifreeze mechanisms, and their protective efficacy diminishes rapidly in extreme cold environments.

Regarding functional coatings for specific scenarios (such as automotive hidden door handles and leather goods), current solutions predominantly rely on chemical antifreeze agents or silicone-based protective films. However, these conventional approaches face clear limitations: chemical agents may cause corrosion or environmental concerns, while passive protective layers cannot actively inhibit ice formation. Particularly under persistent low temperatures, their effectiveness declines significantly, failing to meet the demands of sustainable and high-performance protection .

Initially, we conducted systematic literature reviews and computational screening to identify antifreeze proteins (AFPs) with excellent performance from diverse biological sources and constructed an endogenous expression pathway. However, after discussions with R&D departments in the cosmetics industry, we recognized the bottlenecks of high cost and low yield associated with traditional protein purification processes. This insight prompted us to pivot toward designing a yeast surface display system. We successfully achieved the surface display of natural AFPs on yeast and validated its effectiveness through characterization experiments.

To enhance the performance of antifreeze proteins, this project established a “design-build-simulate” framework based on computational simulations and artificial intelligence. During the protein design phase, a three-track parallel strategy was implemented. We employed the ESM-1v model to analyze natural antifreeze protein sequences, identifying variable regions to guide precise site-directed mutagenesis. Concurrently, we utilized the inverse folding algorithms of ProteinMPNN to systematically explore the sequence space of natural templates, generating diverse candidate sequences that underwent high-confidence structural screening using AlphaFold2. For de novo design, we integrated three innovative strategies (spiral periodic function array, ice lattice-matching function arrays, and AI-generated backbones) to tailor protein structures for specific antifreeze requirements, followed by inverse folding optimization that successfully produced novel amino acid sequences with ideal antifreeze functionality.

Building upon this design framework, we developed the specialized prediction model AFP-igemTJ2025, which enables direct functional prediction from sequence data and facilitates efficient, accurate screening of newly designed proteins. Finally, through molecular dynamics simulations, we systematically analyzed the structural stability, dynamic behavior, and ice growth inhibition mechanisms of all selected candidates in ice-water systems, providing structural-level theoretical guidance for subsequent rational design iterations.

In response to industrial requirements for low-molecular-weight antifreeze components identified through human practices feedback, we conducted systematic development of antifreeze short peptides. Through repetitive motif analysis of small-molecular-weight peptides with sequence similarity in natural protein sequences, we identified key functional regions influencing antifreeze activity. By constructing phylogenetic trees to resolve evolutionary relationships, we ultimately selected candidate peptides exhibiting structural stability, synthetic feasibility, and evolutionary representativeness, establishing a foundation for subsequent wet lab validation and functional development.

Additionally, we designed an online chatbot to serve as an ongoing engagement platform connecting the project with the public. Beyond its role in answering questions and promoting synthetic biology literacy, the system facilitates real-time synchronization with societal demands, thereby informing iterative improvements in our project design.

Figure 3 Integration of Wet-Lab, Computational, and Human Practices Workstreams Across Project Phases (Early, Mid, and Late)

Following the rational design and engineering of AFPs, we established a regulatory system based on the galactose-inducible promoter pGAL1 to ensure efficient and controllable expression of AFPs in the yeast surface display system. Ultimately, we completed plasmid construction and relevant characterization through wet lab experiments, confirming that all engineered proteins exhibited significant antifreeze activity.

Figure 4 Generalized genetic circuit for inducible AFP surface display. This schematic illustrates the core design for yeast-based antifreeze protein presentation. (The AFP Expression Cassette generically represents all engineered AFP variants (including point mutants and chimeric proteins)

While delving deeper into protein engineering pathways, we recognized that single-component antifreeze strategies might have performance ceilings. Interestingly, during investigations into extreme condition tolerance in yeast models, we serendipitously discovered a crucial metabolic pathway—through SCRaMbLE system screening and mass spectrometry analysis, we identified a gene cluster containing five key coding regions, which was confirmed to be directly involved in endogenous trehalose synthesis in yeast.

Based on this discovery, we began exploring the synergistic antifreeze mechanism between trehalose and engineered AFPs. Trehalose, a known natural protectant, stabilizes macromolecular structures through the "water replacement" effect. Our engineered AFPs are specifically optimized for interaction with ice crystal interfaces. Their combination creates a unique synergistic effect: AFPs inhibit the formation and growth of ice crystals, while trehalose further enhances the overall system's antifreeze performance by maintaining the stability of the local water molecular network.

Figure 5 Schematic Diagram of the Antifreeze Protein-Trehalose Synergistic System

In summary, Project KUNPENG has developed an efficient and practical novel yeast-based antifreeze system, providing a new biological solution to extreme environmental challenges and inaugurating a new chapter for synthetic biology in the field of low-temperature protection.

This product breaks from traditional moisturizing concepts by leveraging the synergistic action of AFPs and trehalose to form a protective layer on the skin surface, directly inhibiting ice crystal growth and reducing physical damage caused by low temperatures. This innovative formulation will provide a new active protection solution for outdoor workers and athletes in severe cold environments.

To address the winter freezing issue of hidden door handles in vehicles, we intend to develop a bio-based antifreeze coating utilizing AFPs. This product works by inhibiting strong ice adhesion, making the ice layer easy to remove by normal actuation force, ensuring reliable operation of door handles in low temperatures, and providing an eco-friendly, efficient antifreeze guarantee for smart mobility.

Aimed at the problem of leather products becoming stiff and cracking in low-temperature environments, we have developed a bio-based leather antifreeze care agent. This product uses AFPs to inhibit ice crystal formation between leather fibers, combined with trehalose to maintain collagen structural integrity, thereby extending the service life of leather goods under harsh cold conditions and offering a professional low-temperature protection solution for high-end leather goods.