The achievement is notable because of what the protein lacks. Most well-studied sweet proteins carry stabilizing features that honey truffle sweetener simply does not have. Thaumatin, mabinlin, and brazzein contain multiple disulfide bonds and mixtures of alpha-helices and beta-sheets; neoculin is reinforced by extensive disulfide crosslinks; even monellin, which is largely beta-rich, retains a short stabilizing helix. Honey truffle sweetener, by contrast, is cysteine-free and helix-free—a compact bundle of beta-strands held together by sheet hydrogen bonding and loop geometry. Comparisons of residue-contact density showed the fold is not uniformly loose; its overall packing is comparable to thaumatin and denser than brazzein. But cavity analysis revealed a larger pocket volume per residue than reference proteins, and the absence of classic stabilizing elements left the protein vulnerable to heat.

The first step was to see the molecule in atomic detail. The team crystallized the protein and determined its structure at 1.58 angstrom resolution using synchrotron radiation at the Shanghai Synchrotron Radiation Facility, solving the structure by molecular replacement with an AlphaFold-predicted model as the search template. The structure revealed a characteristic arrangement of seven major beta-strands forming a slightly twisted antiparallel sheet, flanked by three shorter strands—a so-called 7-plus-3 topology—connected by solvent-exposed loops, including an extended and apparently flexible loop between strands three and four. This high-resolution map became the common reference for every subsequent design decision, from mutation evaluation to disulfide-bond planning.

With the structure in hand, the researchers screened every one of the 121 positions for potentially stabilizing substitutions using two complementary computational methods. Rosetta, a physics-based framework, estimated the energetic effect of each mutation after side-chain repacking and constrained relaxation, while ThermoMPNN, a deep-learning model, judged whether each amino acid substitution would be compatible with the local structural environment. From the overlapping and divergent predictions, the team selected 72 single-point variants for experimental testing. The results validated the strategy: 41 of the 72 variants showed a positive shift in melting temperature, and 27 exceeded a one-degree gain. The single best mutations, Q37A and S55G, raised the melting temperature by 3.91 and 4.23 degrees respectively.

But the raw computational scores turned out to be only modestly predictive. Correlations between predicted scores and measured stability changes across the 72 variants were weak, with Pearson coefficients of just 0.206 for Rosetta and 0.248 for ThermoMPNN. That gap is precisely where the study’s most innovative element comes in: a machine-learning framework the authors call SMRC-Net, which learns from the experimental data itself. The system combines a baseline melting-temperature estimate derived from a pretrained protein language model, ESM2, with a residual correction built from molecular dynamics descriptors—flexibility changes, solvent accessibility, residue contacts, and global perturbation measures calculated from simulations of the wild-type protein and 65 mutants.

Under repeated strict nested cross-validation, the sequence-only baseline explained about 44 percent of the variance in melting temperature, with a mean absolute error of 1.35 degrees. Adding the molecular dynamics residual correction lifted the explained variance to nearly 60 percent, cut the error to 1.15 degrees, and improved the rank correlation of variants from 0.540 to 0.698. When the frozen model was tested prospectively on ten Rosetta-designed mutants it had never seen, it correctly predicted the direction of the stability change for nine of the ten—a 90 percent directional accuracy. The single miss, T115I, was predicted to stabilize the protein but produced a small experimental decrease of 0.67 degrees.

Armed with validated stabilizing mutations, the team assembled them stepwise rather than all at once, a modular strategy that kept each combination experimentally testable. Ten substitutions were grouped into four modules of two or three mutations, each verified to raise the melting temperature before further assembly. Combining the three-residue modules produced a six-site variant at 67.77 degrees, and the gains proved nearly additive—the measured improvement differed from the sum of individual contributions by only 0.12 degrees. Two parallel assembly routes then produced eight-site intermediates, and their union yielded the ten-site variant Mut10-1 at 73.03 degrees, 18.43 degrees above the wild type.

The final flourish was structural rather than chemical: an engineered disulfide bond. Screening candidate cysteine pairs by geometric criteria with Rosetta’s DisulfidizeMover identified Q37C/D46C as the most favorable linkage. Introducing that single bond added another 6.95 degrees on its own, and when combined with the ten mutations it produced the final variant, FM, with a melting temperature of 82.07 degrees. Crucially, sensory evaluation by eight trained panelists showed that both Mut10-1 and FM retained sweetness thresholds similar to the wild-type protein—evidence that the stabilizing changes did not disrupt the receptor-binding surface responsible for the protein’s intense taste.

Practical stress tests underscored the difference. After one hour at 80 degrees Celsius, the wild-type protein had largely precipitated out of solution, while FM remained predominantly soluble; even after four hours, FM retained a substantial soluble fraction. Circular dichroism spectroscopy showed that FM’s secondary structure was largely preserved through the prolonged heating, and the engineered protein also stayed soluble and structurally intact across acidic, neutral, and alkaline conditions at pH 3, 6, and 9. Molecular dynamics simulations suggested why: the mutations did not rigidify the protein wholesale but redistributed local flexibility, modestly remodeled the residue-contact network, and stabilized a single dominant folded ensemble. An experimental test supported one long-range interaction in particular—disrupting the H16-F99 contact with a double mutant lowered the melting temperature by 5.2 degrees, confirming its role in tertiary stability.

The authors are careful about the limits. Sweetness was tested only on untreated samples, so whether the engineered protein survives heating and pH shifts while still tasting sweet in real foods remains unverified, and sensory testing in complex food matrices is the necessary next step. The SMRC-Net predictor is also calibrated specifically to this scaffold, though the authors argue the residual-learning architecture could be retrained for other compact proteins where molecular dynamics remains computationally feasible. Still, the demonstration is striking: a cysteine-free, helix-free beta protein with no natural stabilizing armor was transformed into a heat-tolerant, sweetness-preserving molecule through a coordinated workflow of crystallography, complementary computation, machine learning, and disciplined stepwise assembly. For a protein once undone by a warm afternoon, that is a remarkable turnaround—and a template for engineering the next generation of sugar substitutes.

Subject of Research: Computation-guided protein engineering to improve the thermal stability of the fungal sweet protein honey truffle sweetener

Article Title: Integrated Computation-Guided Thermostabilization of the Fungal Sweet Protein Honey Truffle Sweetener

Article References: Wang, Z., Wang, W., Zhu, F., Zhang, Y., Li, Y., Zhu, Z., Lu, Z., Huang, A., Yu, M., & Liu, S. (2026). Integrated computation-guided thermostabilization of the fungal sweet protein Honey Truffle Sweetener. Current Research in Food Science, 13, Article 101584. https://doi.org/10.1016/j.crfs.2026.101584

Image Credits: AI Generated

DOI: 10.1016/j.crfs.2026.101584

Keywords: sweet proteins, honey truffle sweetener, protein engineering, thermostability, crystal structure, Rosetta, ThermoMPNN, molecular dynamics, machine learning, disulfide bond, sugar substitute, food science

The work, summarized in a Research Briefing by Nature Biotechnology, combines two ingredients that have rarely been brought together so systematically: high-throughput experimentation and Bayesian optimization. Bayesian optimization is a machine learning strategy designed for exactly the kind of problem the vaccine engineers faced, in which each experiment is costly, the search space of possible formulations is vast, and the relationship between the ingredients and the outcome is complex and poorly understood. Rather than testing every conceivable combination of stabilizing sugars, buffering agents, drying protocols and nanoparticle compositions, the algorithm builds a probabilistic model of the formulation landscape from the data gathered so far, then selects the next experiment where the expected gain in information or performance is highest.

This data-efficient approach matters because traditional formulation development is notoriously slow. A medicinal chemist or vaccine formulator might adjust one variable at a time, run stability assays that take days or weeks to complete, and iterate over months or years. By contrast, AGENT closes the loop: robotic high-throughput platforms prepare large batches of candidate formulations, automated assays measure their stability and bioactivity, and the Bayesian optimizer digests each round of results to propose an improved set of conditions for the next. The framework effectively takes the human out of the loop for the routine decision-making, reserving human judgment for defining the objectives and validating the final candidates.

The target of the optimization was a solid-state mRNA vaccine, meaning that the mRNA-lipid nanoparticles are dried into a stable matrix rather than suspended in liquid. Drying is a well-known route to thermal stability for biological drugs, as the absence of water dramatically slows the hydrolysis and other degradation reactions that destroy RNA. But the drying process itself is stressful: ice crystals, osmotic shocks and the removal of the hydration shell can shear nanoparticles, aggregate them, or otherwise wreck the delicate structure that allows them to fuse with cell membranes and release their payload. The art lies in choosing protective excipients, often sugars such as trehalose or sucrose, and processing conditions that let the nanoparticles pass through drying and reconstitution unscathed.

The results reported by Tian and colleagues are striking on several fronts. The optimized solid-state formulations retained full bioactivity after two months of storage at 37 degrees Celsius, a condition far beyond anything current liquid mRNA vaccines can tolerate. In immunization studies, these stored vaccines elicited immune responses that were non-inferior to those generated by fresh, never-stressed formulations, which is the critical regulatory benchmark: a thermostable vaccine is only useful if it works as well as the product it replaces. Stability at room temperature and above would allow vaccines to be shipped and stored without the elaborate cold chain of ultra-low-temperature freezers, dry ice and temperature loggers that dominates current distribution.

The team went beyond vials and syringes. The thermostable mRNA formulations were incorporated into microneedle patches, small arrays of dissolving needles that painlessly deliver vaccine into the skin. Microneedle delivery has been pursued for years as a needle-free, potentially self-administered vaccination platform, and earlier work from some of the same community demonstrated a microneedle vaccine printer for thermostable COVID-19 mRNA vaccines. The new study extends that vision by showing that the AI-optimized, cold-chain-free formulations could be delivered via microneedle patches in rodents and nonhuman primates, providing proof of concept in animals that are far closer to humans than cell cultures or mice alone.

The combination of thermostability and patch delivery points toward a fundamentally different model of vaccine distribution. A vaccine that survives months at 37 degrees Celsius and can be applied as a patch requires no freezer, no trained phlebotomist, no reconstitution step and minimal medical waste. In resource-constrained settings, where the cited literature on vaccine accessibility emphasizes that storage and distribution bottlenecks cost lives, such a product could reach communities that conventional campaigns struggle to serve. It would also simplify pandemic preparedness, since stockpiles would not depend on a continuous chain of refrigeration stretching from factory to clinic.

From a technical standpoint, the study is a showcase of how modern machine learning changes the economics of formulation science. Bayesian optimization is particularly suited to problems with expensive evaluations because it balances exploration and exploitation: it probes regions of the design space where the model is uncertain, while also refining candidates that already look promising. The framework’s data efficiency means that the number of physical experiments needed to reach an optimized formulation can be a fraction of what a grid search or one-variable-at-a-time approach would demand. Each experiment feeds back into a surrogate model, typically a Gaussian process or similar probabilistic regressor, which quantifies not just the predicted performance of untested formulations but the confidence in those predictions.

The subjects attached to the paper, spanning biomaterials for vaccines, biomedical engineering, DNA and RNA science, and nanoparticles, reflect the interdisciplinary nature of the achievement. The mRNA-lipid nanoparticle platform sits at the intersection of nucleic acid chemistry, colloid science, polymer physics and immunology, and no single discipline could have solved the stability problem alone. What AGENT contributes is a unifying optimization layer that treats the entire formulation-and-processing pipeline as a single search problem, allowing trade-offs to be navigated automatically. A choice of excipient that improves drying stability but slows release, for example, can be weighed against alternatives in a principled way rather than by intuition.

Caveats remain, as they always do at this stage of translation. The immunogenicity data are from rodents and nonhuman primates, and human clinical trials will be needed to confirm that the optimized solid-state vaccines match the performance of approved liquid products in people. Two months of stability at 37 degrees Celsius is a dramatic improvement, but regulators will want to see longer-term data across a range of temperatures and humidities, along with manufacturing scale-up studies. Nonetheless, the demonstration that an AI-driven, data-efficient framework can compress years of formulation development into a rapid, systematic campaign is itself a significant result. If the approach generalizes to other mRNA drugs and vaccine targets, the cold chain that has long constrained global health delivery may finally begin to loosen.

Subject of Research: AI-guided Bayesian optimization for the development of thermostable, solid-state mRNA-lipid nanoparticle vaccines

Article Title: AI-guided optimization for thermostable mRNA vaccines

Article References: AI-guided optimization for thermostable mRNA vaccines. (2026). Nature Biotechnology. https://doi.org/10.1038/s41587-026-03330-x

Image Credits: AI Generated

DOI: 10.1038/s41587-026-03330-x

Keywords: mRNA vaccines, artificial intelligence, Bayesian optimization, lipid nanoparticles, thermostability, cold chain, microneedle patches, high-throughput experimentation, vaccine formulation, global health, biotechnology, immunization

Writing in Nature Biotechnology, Jinbi Tian, Khanh T. M. Tran, and colleagues describe AGENT, short for Algorithm-Guided Experimental design for lipid Nanoparticle Thermostabilization. The platform couples high-throughput experimentation with Bayesian optimization, a machine-learning strategy that is specifically built for situations where every experiment is costly and data are scarce. Rather than screening thousands of candidate formulations blindly, AGENT extracts the maximum amount of information from each small batch of results, using Gaussian process models to predict how untested excipient combinations will behave and to decide which experiments to run next.

The formulation problem the team tackled is genuinely multidimensional. To convert liquid mRNA–lipid nanoparticles into a solid-state, water-free product, the researchers used vacuum drying and then had to identify a cocktail of excipients—sugars, buffers, and other stabilizing agents—that would preserve the nanoparticles’ structure and the mRNA’s bioactivity through drying and subsequent storage. The search space of possible excipient types, ratios, and concentrations is enormous, and the relationships between composition and stability are nonlinear and poorly understood. Traditional one-factor-at-a-time approaches, the authors note, have been constrained by narrow formulation scope and low-throughput screening, which is precisely why prior stabilization efforts have progressed slowly.

Bayesian optimization changes the economics of that search. The algorithm maintains a probabilistic model of the entire design space and, at each iteration, balances exploration of uncertain regions against exploitation of promising ones, typically using an expected-improvement acquisition function to select the next experiment. Implemented within the AutODEx platform developed at MIT’s Computer Science and Artificial Intelligence Laboratory, AGENT completed its optimization in just six iterative rounds within a single month—a pace that would be unattainable with conventional design-of-experiments workflows, which often require many more cycles and far more material to converge on an optimum.

The results are striking. The team worked with two clinically relevant lipid nanoparticle systems, one based on SM-102, the ionizable lipid in Moderna’s vaccine, and the other on ALC-0315, used by Pfizer-BioNTech. After optimization, both were converted into solid-state formulations that retained 100 percent of their bioactivity after more than two months of storage at 37 degrees Celsius—roughly human body temperature. For a class of products that normally degrades within hours or days at room temperature, that level of thermal tolerance represents a qualitative leap, and it was achieved without altering the core lipid compositions already validated in clinics.

Stability on the shelf means little if the vaccine no longer works in the body, so the researchers subjected their thermostable formulations to a demanding battery of animal tests. In rodents, solid-state SARS-CoV-2 vaccine formulations delivered by standard injection induced antigen-specific immune responses that were non-inferior to those elicited by freshly prepared liquid vaccines given intramuscularly. The team then pushed the concept further, incorporating the stabilized nanoparticles into microneedle patches—tiny arrays of dissolving needles that painlessly deliver vaccine into the skin and could, in principle, be mailed to patients or administered by minimally trained personnel.

The microneedle experiments extended into non-human primates, a critical translational step. Patch delivery of the thermostable solid-state vaccine in cynomolgus monkeys again produced antigen-specific immune responses comparable to conventional intramuscular injection of freshly prepared vaccine. Because microneedle patches eliminate the need for needles, syringes, and highly trained injectors, and because the solid-state formulation removes the need for refrigeration, the combination addresses two of the most persistent bottlenecks in global immunization campaigns simultaneously.

The significance of the work extends beyond any single vaccine. The mRNA–LNP platform is now being pursued for cancer immunotherapies, gene editing, and treatments for rare genetic diseases, and all of these applications inherit the same storage fragility. Solid-state, water-free formulations open the door to integrating mRNA payloads into emerging delivery modalities, from microneedle patches to other advanced systems, that would be incompatible with a liquid product requiring ultra-cold handling. The authors also point out that the cold chain carries substantial economic and environmental costs, with estimates suggesting that vaccine distribution inefficiencies contribute significantly to global inequity in immunization coverage.

Methodologically, the study adds to a growing body of evidence that Bayesian optimization can compress discovery timelines in materials science and pharmaceutical development, where experiments are expensive and the parameter space is combinatorially large. Similar approaches have accelerated the identification of 3D printing materials and complex heterostructures, and the MIT team has made both its datasets, deposited in a public Zenodo repository, and its source code openly available, lowering the barrier for other laboratories to adopt the workflow. The optimization framework treats excipient selection as a multi-objective problem, allowing researchers to specify what matters—potency retention, storage stability, manufacturability—and let the algorithm navigate trade-offs automatically.

Caveats remain before thermostable mRNA vaccines reach clinics. The reported storage data cover months rather than years, and regulatory agencies will require extensive documentation of long-term stability, manufacturing consistency, and safety under Good Manufacturing Practice conditions. Human clinical trials of solid-state formulations and microneedle delivery lie ahead. Yet the demonstration that AI-guided experimentation can find stabilizing excipient compositions in six iterations and one month suggests that the development bottleneck may no longer be the search itself. If the approach generalizes across payloads and lipid systems, the freezer—and the costly, fragile supply chain it demands—could finally become optional for mRNA medicine.

Subject of Research: AI-guided optimization of thermostable solid-state mRNA–lipid nanoparticle vaccine formulations

Article Title: Accelerated discovery of thermostable mRNA–lipid nanoparticle vaccines using data-efficient AI

Article References: Tian, J., Tran, K. T. M., Pogostin, B. H., Sheridan, O., Mursalova, S., Lee, A. H., Liu, S., Hamkins, J., Antov, D., Power, A. L., Dash, Z. S., Yun, D., Konaković Luković, M., Langer, R. S., & Jaklenec, A. (2026). Accelerated discovery of thermostable mRNA–lipid nanoparticle vaccines using data-efficient AI. Nature Biotechnology. https://doi.org/10.1038/s41587-026-03331-w

Image Credits: AI Generated

DOI: 10.1038/s41587-026-03331-w

Keywords: mRNA vaccines, lipid nanoparticles, Bayesian optimization, thermostability, cold chain, microneedle patches, vacuum drying, excipients, SARS-CoV-2, drug delivery, machine learning, Nature Biotechnology

The significance of the result lies in what it eliminates. Conventional enzymatic PET recycling depends on thermostable enzymes operating near the glass transition temperature of PET, where polymer chains become flexible enough for the catalyst to access them. That strategy works, but it demands substantial energy input to maintain elevated temperatures, relies on freshwater-based buffer systems, and creates a downstream bottleneck: the monomers released by hydrolysis must then be cooled to 30 to 40 degrees Celsius before bacteria or engineered chassis organisms can assimilate them, adding an energy-intensive cooling step. Running depolymerization in seawater at moderate temperature collapses this entire pipeline into a single, low-energy process, and seawater is free, abundant, and sterile enough in its salinity to support non-sterile bioprocessing.

The team began by screening eight representative PET hydrolases in artificial seawater prepared according to the ASTM D1141-98 standard, at 30 degrees Celsius. Among the candidates were esterases from Bacillus subtilis and Thermobifida fusca, the marine-derived PE-H from Pseudomonas aestusnigri, the polyesterase Cut190, the fungal cutinase FsC, the metagenomic leaf-branch compost cutinase LCC, and IsPETase, the enzyme originally discovered in the bacterium Ideonella sakaiensis in 2016. IsPETase proved the clear winner, outperforming the second-best enzyme by more than 4.5-fold. The catch was fragility: wild-type IsPETase has a half-life of only about one day at 30 degrees Celsius, far too unstable for sustained industrial use.

To fix that, the researchers turned to a semi-rational protein engineering strategy focused on rigidifying flexible sites. They analyzed seven published crystal structures of IsPETase with the program B-FITTER, ranking residues by B-factors to identify the 45 most mobile positions, and added 11 positions where literature had already reported stabilizing mutations. The overlap yielded a final list of 53 target residues. At each site, they built saturation mutagenesis libraries using degenerate NNK codons, which encode all 20 amino acids, and screened the libraries with a clever Petri-dish assay: after a heat shock at 75 degrees Celsius that lyses the host E. coli cells, surviving enzyme converts a chromogenic substrate into a visible blue product, allowing positive mutants to be picked by eye. The screen produced 22 beneficial mutations from 17 positions, raising the enzyme’s T50 inactivation temperature by up to 9.8 degrees in single mutants.

Combining mutations proved the delicate part. Two multi-mutants, M7 and M15, pushed the T50 up by 24.9 and 21.7 degrees respectively, but their hydrolysis activities collapsed to just 29 and 25 percent of the wild-type level, a classic example of the trade-offs and epistatic interactions that plague multiparameter enzyme engineering. Structural analysis suggested that the S207E mutation, sitting adjacent to the catalytic residue D206, was sabotaging activity, so the team removed it to create M6, which restored activity to 92 percent of wild type. Adding the R280A mutation, previously reported to aid substrate binding, lifted activity to 117 percent of wild type, and introducing an engineered disulfide bridge between positions 233 and 282, inspired by stabilizing strategies in other PET hydrolases, boosted the T50 by a further 8.9 degrees without sacrificing activity.

The final variant, M8, is a tour de force of balanced engineering. Its melting temperature rose by 27.3 degrees Celsius over wild type, its catalytic activity increased 1.14-fold, and, unexpectedly, its soluble expression yield in E. coli jumped 14.3-fold. The expression boost traces to the P181V mutation, which appears to promote formation of a more extensive beta-sheet structure that folds more readily in the bacterial cytoplasm. Because enzyme production cost is a major barrier to industrial biocatalysis, a variant that is simultaneously more stable, more active, and far cheaper to manufacture represents a rare triple win. Benchmarked against the leading thermostable enzymes DuraPETase and LCC-ICCG under identical conditions, M8’s overall depolymerization efficiency, combining yield and activity, exceeded them by 32.2-fold and 10.4-fold respectively.

To understand why M8 works so well, the team solved its crystal structure at the Shanghai Synchrotron Radiation Facility and ran extensive molecular dynamics simulations in artificial ocean water. The overall fold was preserved, with a root mean square deviation of only 0.178 angstroms from wild type, but individual mutations each contributed stabilizing interactions: K95N forms a new hydrogen bond with S242, I168R creates a salt bridge with D186, P181V strengthens hydrophobic packing with L167, S214V places a hydrophobic contact near W185, and A248D forms a transient salt bridge with a rotated R100 that appears in nearly 80 percent of simulation frames. The flexible loops spanning residues 202 to 218 and 231 to 240 became markedly more rigid. Meanwhile, the R280A mutation removes a spatial clash between arginine 280 and the terminal phenyl ring of the substrate, and simulations showed the distance between catalytic residues S160 and H237 shortened dramatically in M8, facilitating the proton shuttling that drives catalysis.

The decisive test came in natural seawater collected from the South China Sea, using post-consumer PET powder ground from fruit packaging trays. The team optimized substrate loading from 5 to 15 percent by weight and enzyme concentrations from 500 to 2000 nanomolar, maintaining the pH at 8.2 with periodic sodium hydroxide additions. Over five days of continuous operation, product release remained essentially linear at both 30 and 37 degrees Celsius. At 37 degrees, total soluble products reached 76.8 millimolar, roughly two-thirds of it terephthalic acid, corresponding to a monomer production rate of 15.4 millimolar per day. At 30 degrees the rate was 5.2 millimolar per day. Crucially, 15.4 millimolar per day falls squarely within the 3 to 28.5 millimolar per day uptake rates reported for monomer-assimilating microbes, meaning the enzyme’s output is sufficient to feed a downstream fermentation without any buffer or dilution step.

That compatibility underpins the team’s larger vision: a simultaneous enzymatic depolymerization and fermentation process, or SEDF, modeled on the simultaneous saccharification and fermentation framework used in lignocellulosic biofuel production, but conducted entirely in seawater. The missing piece is a microbial chassis that can consume terephthalic acid and ethylene glycol under high-salinity conditions; existing PET-metabolizing strains such as Pseudomonas and Rhodococcus pyridinivorans have only been demonstrated in freshwater. The authors point to Halomonas bluephagenesis, a salt-loving bacterium already scaled to commercial fermenters producing tens of thousands of tons of bioplastic annually, as the most promising candidate for engineering a complete seawater-based recycling loop.

The enzyme’s robustness extends beyond the reactor. M8 retained more than 80 percent of its activity after four months of incubation under high-salinity, low-temperature conditions, suggesting potential for treating PET microplastic contamination in industrial wastewater or even in-situ marine remediation, though the authors caution that open-environment deployment would require engineered microbial carriers, rigorous biocontainment strategies, and thorough ecological risk assessment. They are equally candid about the work’s limitations: the experiments were milliliter-scale, and only a techno-economic analysis at pilot or cubic-meter scale, weighing slower reaction kinetics against eliminated heating and freshwater costs, can establish true industrial competitiveness. Computational tools including FoldX, Pythia, ThermoMPNN and CataPro failed to predict further improvements over M8, underscoring that experimental screening remains indispensable. Still, with a single engineered enzyme now capable of chewing through real plastic waste in real seawater at body temperature, the prospect of freshwater-free, low-energy plastic bio-recycling has moved from concept to demonstrable reality.

Subject of Research: Protein engineering of the PET-degrading enzyme IsPETase for seawater-based plastic depolymerization at ambient temperature

Article Title: Engineering a robust IsPETase for energy-efficient PET depolymerization in natural seawater at ambient temperatures

Article References: Huang, X., Jia, Q., Li, G., Yang, X., Xu, S., Liu, J., Li, W., Liu, Y., Xie, W., & Cao, L. (2026). Engineering a robust IsPETase for energy-efficient PET depolymerization in natural seawater at ambient temperatures. Advanced Biotechnology, 4(2), Article 14. https://doi.org/10.1007/s44307-026-00104-z

Image Credits: AI Generated

DOI: 10.1007/s44307-026-00104-z

Keywords: PET recycling, IsPETase, enzyme engineering, plastic pollution, seawater biocatalysis, thermostability, protein engineering, biodegradation, molecular dynamics, industrial biotechnology, plastic upcycling, Halomonas

The enzyme at the heart of the work is phenylalanine dehydrogenase from Natranaerobius thermophiles, a microorganism originally isolated from marine environments that thrives at elevated temperatures and in extremely salty, alkaline conditions. Enzymes from such extremophiles are prized in industry precisely because they tolerate conditions that would destroy ordinary proteins. Yet even these rugged biocatalysts can benefit from a supportive scaffold. Immobilizing enzymes on solid carriers is a classic strategy in biotechnology: it allows the catalyst to be recovered and reused, protects its three-dimensional structure, and simplifies product purification. The challenge has always been finding carriers that are cheap, biocompatible, and mechanically useful at the same time.

The Xiamen team, led by Shizhen Wang, turned to wool keratin extracted through a reduction-dissolution process using sodium sulfide, urea, and sodium dodecyl sulfate. Keratin is no ordinary protein. Its three-dimensional network is stabilized by disulfide bonds, hydrogen bonds, and ionic interactions, and it is studded with functional groups including amino, carboxyl, and sulfhydryl moieties. These groups make keratin inherently biocompatible and give it a rich chemical surface for binding other molecules. Graphene oxide, meanwhile, is a two-dimensional carbon sheet decorated with oxygen-containing groups, widely used in biosensors and drug delivery because of its heterogeneous surface chemistry and enormous surface area.

The assembly strategy was elegantly simple. The researchers first mixed the purified enzyme with keratin solution, then added graphene oxide and stirred the mixture at low temperature. The resulting GO-WK-NT composite self-organized into a layer-by-layer sandwich structure, with keratin and enzyme complexes interleaved between graphene oxide sheets. Spectroscopic and microscopic evidence supported this architecture: Fourier transform infrared spectroscopy revealed a new peak at 500 cm−1 indicating fresh interactions among the three components, while X-ray diffraction showed distinctive peaks at 31.82 and 41.2 degrees, reflecting keratin chains rearranging in parallel or folded configurations between the oxide lamellae. Scanning electron microscopy confirmed that graphene oxide was uniformly dispersed throughout the keratin-enzyme matrix rather than clumping, as it tends to do on its own.

The binding forces holding this sandwich together are multiple and cooperative. Hydrogen bonds and electrostatic interactions link the charged groups of keratin to both the enzyme surface and the oxidized carbon sheets, while hydrophobic and π-π interactions contribute additional grip. Zeta potential measurements tracked the progressive accumulation of negative surface charge, from −2.18 millivolts for the free enzyme to −7.28 millivolts for the full composite, consistent with stepwise layering. This web of weak, redundant interactions is precisely what makes the material soft yet stabilizing: it cradles the enzyme without covalent modification that might distort its active site.

Performance testing showed that the composite outperformed both the free enzyme and a simpler graphene-oxide-only formulation. While the free enzyme displayed an activity of 4.48 units per milligram, the graphene oxide immobilizate reached 5.22 and the keratin hybrid reached 6.36 units per milligram. Reusability, a key economic criterion for industrial biocatalysts, was equally impressive: after five cycles of recovery and reuse, the hybrid retained 73.5 percent of its initial activity, slightly ahead of the 72.1 percent retained by the simpler composite, whose performance decays as enzyme desorbs from bare oxide surfaces.

Thermal resilience told an even clearer story. After two and a half hours of incubation at 70 degrees Celsius and pH 9.0, the free enzyme retained only 46.1 percent of its activity, whereas the keratin-graphene hybrid kept 69.4 percent, with a measured half-life of 4.7 hours. The authors attribute this to the keratin’s functional groups forming multi-site bonds that lock the enzyme’s structure in place, while its hydrophilic groups preserve the crucial hydration shell that keeps proteins folded. The keratin matrix also acts as a local buffer: at pH 11.0, the immobilized enzyme retained 57.1 percent activity compared with just 34.8 percent for the free enzyme, because the amino, carboxyl, and sulfhydryl groups moderate the pH microenvironment surrounding the active site.

Kinetic analysis using Hanes-Woolf plots revealed that immobilization increased both the maximum reaction velocity and the apparent Michaelis constant, the latter reflecting added diffusion resistance through the layered structure. Nevertheless, the catalytic efficiency, expressed as kcat/Km, rose from 2.72 to 3.32 mM−1·s−1 for the hybrid enzyme. One striking finding emerged from solvent testing: in 30 percent cyclohexane, the immobilized enzyme actually reached 130.6 percent of its baseline activity, while the free enzyme dropped to 48.6 percent. The researchers explain that cyclohexane prevents agglomeration of graphene oxide and dissolves the hydrophobic substrate, ethyl 2-oxo-4-phenylbutyrate, more effectively, while the keratin coating shields the enzyme from direct solvent attack.

The practical target of all this engineering is L-homophenylalanine, an unnatural amino acid used as a chiral building block in pharmaceutical synthesis. The enzyme catalyzes the reductive amination of the keto-acid ester substrate using NADH as a cofactor, a green route that avoids the metal catalysts and harsh conditions of traditional chemical synthesis. An immobilized, reusable, solvent-tolerant version of this catalyst could make such bioprocesses considerably more economical, and the fact that one of its key ingredients is textile waste adds a circular-economy dimension that is increasingly demanded by both regulators and consumers.

Looking beyond chemical manufacturing, the authors point toward biosensing and wearable technology. Graphene oxide is an excellent electrode material, and keratin-based graphene composites have already been spun into flexible strain-sensor inks. A soft, biocompatible enzyme-carrier that functions at 70 degrees Celsius and in high salt could underpin enzyme electrodes for extreme environments or conformal health-monitoring devices. The work, funded by the National Natural Science Foundation of China, thus transforms two humbling feedstocks, landfill-bound wool and a sheet of oxidized carbon, into a precisely layered biological machine, showing how materials science and extremophile biology can converge on chemistry that is simultaneously tougher, greener, and cheaper.

Subject of Research: Immobilization of an extremophile phenylalanine dehydrogenase on a wool keratin and graphene oxide hybrid biomaterial for enzyme stabilization and green biosynthesis

Article Title: Hybrid soft biomaterial of wool keratin and graphene oxide for immobilization of thermophilic and halophilic dehydrogenase from extremophile

Article References: Zeng, X., Lei, H., Zhang, J., Xie, B., Jiang, L., & Wang, S. (2025). Hybrid soft biomaterial of wool keratin and graphene oxide for immobilization of thermophilic and halophilic dehydrogenase from extremophile. Blue Biotechnology, 2(1), Article 25. https://doi.org/10.1186/s44315-025-00049-3

Image Credits: AI Generated

DOI: 10.1186/s44315-025-00049-3

Keywords: wool keratin, graphene oxide, enzyme immobilization, extremophile, phenylalanine dehydrogenase, Natraanaerobius thermophilus, L-homophenylalanine, biocatalysis, thermostability, biomaterials, waste valorization, blue biotechnology

Unspecific peroxygenases, or UPOs, are heme-thiolate enzymes that use hydrogen peroxide to insert oxygen atoms into carbon-hydrogen bonds and double bonds under remarkably mild conditions. That ability makes them attractive alternatives to metal-based oxidation catalysts, which often require harsh reagents, elevated temperatures, and elaborate protecting-group strategies. Yet industrial adoption has lagged because most UPOs are difficult to produce, poorly stable, or hard to engineer. CglUPO, the enzyme from Chaetomium globosum discovered in 2017, exemplified all three problems. It oxidizes an unusually broad range of substrates, including steroids, resveratrol analogs, isophorone, fatty acids, and short-chain olefins, but the wild-type enzyme, when produced in its native fungus, loses significant activity within days even at 4 degrees Celsius.

The first hurdle was expression. When the CglUPO gene was previously expressed in Saccharomyces cerevisiae, secretion levels of roughly 0.6 milligrams per liter were far too low to support laboratory evolution, which requires screening thousands of variants, each producing measurable enzyme. The team, led by Miguel Alcalde’s group, tested signal peptides from several peroxygenase-producing fungi to drive secretion, finding that CglUPO’s own native leader sequence performed best, yielding about 3 units per liter of activity in 96-well microfermentations. That modest baseline was enough to launch a directed evolution campaign combining rounds of error-prone PCR, saturation mutagenesis, and in vivo DNA recombination, all screened with a high-throughput assay based on oxidation of 2,6-dimethoxyphenol.

An unexpected bonus appeared immediately: simply switching the expression host from the native fungus to yeast raised the enzyme’s kinetic thermostability, measured as the T50 value at which half the activity survives a 10-minute incubation, by 4 degrees, from 52 to 56 degrees Celsius. The researchers attribute this likely to yeast glycosylation, which can stabilize the secreted protein. Subsequent evolution rounds delivered mutants 9G3 and 11G4 bearing substitutions at position 210, a saturation library at that position produced variant 4B10 with a 16-fold total activity improvement thanks to an M210R mutation, and recombination of the best parents yielded the AckBar mutant, carrying F154L and M210V, with roughly 10-fold improved functional expression and secretion levels of about 6 milligrams per liter in yeast.

To move from laboratory evolution to industrial relevance, the team exploited a tandem-yeast strategy: Saccharomyces cerevisiae for building and screening mutant libraries, and Pichia pastoris, recently renamed Komagataella phaffii, for large-scale production. In a 30-liter fed-batch bioreactor, AckBar reached approximately 90 milligrams per liter, tripling the yield of the parental enzyme in the same host without any process optimization. The purified enzymes, roughly 37 kilodaltons with about 24 percent glycosylation, showed T50 values near 60 degrees Celsius, placing them among the most stable recombinant UPOs reported to date. Crucially, while the wild-type enzyme died within days at refrigeration, AckBar retained nearly full activity for months in storage. Both variants were active across a broad pH range from 5.0 to 9.0.

Then came the surprises. Screening the evolved variants against model substrates revealed that AckBar dramatically inverted the enantioselectivity of tetrahydronaphthalene oxidation, producing the S-enantiomer of the benzylic alcohol with 86 percent enantiomeric excess, whereas the parental enzyme favored the R-enantiomer. Even more striking, both the parental and evolved enzymes catalyzed the anti-Markovnikov Wacker-Tsuji oxidation of styrene to phenylacetaldehyde, a long-pursued synthetic transformation that had only very recently been reported in other peroxygenases and in engineered cytochrome P450 monooxygenases, but never before for CglUPO. In such oxidations, the oxygen lands on the terminal carbon of the alkene to yield an aldehyde rather than the more conventional epoxide, giving direct access to aldehyde building blocks under mild aqueous conditions.

Control experiments confirmed that the aldehyde forms directly from the alkene and not through rearrangement of a preformed epoxide: when styrene oxide was offered as a substrate, no phenylacetaldehyde appeared whether or not hydrogen peroxide was present. With 4-chlorostyrene, the aldehyde-to-epoxide ratio climbed as high as 0.79 for AckBar, and total turnover numbers exceeded 5,000 for epoxidation with low enzyme loadings of 0.1 micromolar. To enable future optimization of this reactivity, the team developed a high-throughput screening assay using Purpald, an aldehyde-specific colorimetric reagent previously applied to evolving anti-Markovnikov P450s. In microfermentations of yeast secreting AckBar, the assay gave a stable, quantifiable response, opening the door to dedicated evolution campaigns aimed at pushing the moderate chemoselectivity toward the levels industry demands.

The reactivity profile extended well beyond styrenes. AckBar oxidized ethylbenzene with 79 percent enantiomeric excess toward the R-alcohol, compared with 31 percent for the parental enzyme, and overoxidized alcohols to ketones to a much greater degree, a property that is useful in cases like isophorone, where the overoxidized product 4-ketoisophorone normally requires cumbersome multi-enzyme cascades. With the bicyclic terpene pinene, AckBar performed notably higher levels of direct carbon-hydrogen oxidation to verbenol and verbenone, exceeding 800 total turnover numbers, whereas limonene yielded exclusively epoxide products, with the cis-to-trans ratio flipping depending on which limonene enantiomer was supplied. This preference for epoxidation over allylic hydroxylation with limonene matches the behavior of other UPOs and underscores how substrate structure governs the reaction channel.

To explain these altered selectivities, the researchers turned to long molecular dynamics simulations, since no crystal structure of CglUPO exists. They built a structural model with AlphaFold2, aligned it to the crystal structure of a related peroxygenase from Hypoxylon sp. EC38, and docked the heme cofactor and substrates before running multiple 0.5-microsecond trajectories for each enzyme-substrate pair. For tetrahydronaphthalene, the simulations showed that the F154L and M210V mutations, both located in the heme access channel, enlarge and reshape the active-site cavity, allowing the substrate to penetrate deeper and lock into catalytically competent poses stabilized by interactions with residues His88 and Glu158. In the parental enzyme, the substrate samples non-reactive positions; in AckBar, the hydrogen atom that yields the S-alcohol sits preferentially close to the iron-oxo of Compound I, reproducing the experimentally observed inversion of enantioselectivity. For styrene, the simulations linked chemo- and enantioselectivity to the geometry of the oxo attack on the terminal versus internal carbon of the double bond, predicting predominant epoxidation with a slight S-preference consistent with the experimental data.

The authors are candid about the study’s limits: all biotransformations were analytical-scale, some products could not be resolved on chiral columns, and every structural interpretation rests on a model whose C-terminal region was predicted with low confidence and equilibrated separately. Even so, the achievement is substantial. A promiscuous but fragile enzyme from a fungus that grows on wallpaper has been re-engineered into a secretable, evolvable, bioreactor-ready catalyst that already performs one of synthetic chemistry’s most coveted reactions. With ongoing evolution of AckBar’s anti-Markovnikov activity, deliberate engineering of the signal peptide, and complementary strategies such as ancestral sequence reconstruction, the team believes the full breadth of this enzyme’s capabilities is now within reach, offering a sustainable enzymatic answer to selective oxygen insertion that chemical catalysis has struggled to provide.

Subject of Research: Directed evolution of the fungal unspecific peroxygenase CglUPO for improved expression, stability, and selective oxyfunctionalization chemistry

Article Title: Promiscuous fungal peroxygenase from Chaetomium globosum by directed evolution

Article References: Upp, D. M., Mateljak, I., Mendez-Sanchez, D., Gomez de Santos, P., Sanchez-Moreno, I., Valles, M., Świderek, K., Moliner, V., Scheibner, K., Hofrichter, M., Gonzalez-Perez, D., & Alcalde, M. (2026). Promiscuous fungal peroxygenase from Chaetomium globosum by directed evolution. iScience, 29(10), Article 117534. https://doi.org/10.1016/j.isci.2026.117534

Image Credits: AI Generated

DOI: 10.1016/j.isci.2026.117534

Keywords: directed evolution, unspecific peroxygenase, Chaetomium globosum, biocatalysis, anti-Markovnikov oxidation, enzyme engineering, Saccharomyces cerevisiae, Pichia pastoris, molecular dynamics, oxyfunctionalization, thermostability, heme-thiolate enzyme