1. Introduction
Commercialized since the 1950s, synthetic plastics have significantly transformed industry and daily life due to their low cost, versatility, ease of production, and convenience of use. Yet these same traits now fuel a growing environmental crisis: Today’s global plastic waste far exceeds the capacity of collection, sorting, and disposal infrastructures [
1]. Discarded plastics accumulate on land and in water bodies, fragment into harmful micro- and nano-plastics (MNP), and ultimately threaten ecosystems and human health [
2].
Researchers have explored at least three main technological strategies to tackle the issue of plastic waste. First, the introduction of biodegradable polymers—typically polyesters sourced from renewable biomass—aids in replacing recalcitrant fossil-fuel-derived plastics, especially single-use items in the packaging sector [
3]. Despite intensive development, however, such bioplastics still account for under 1% of the market [
4], with high feedstock and processing costs restricting their broader use. Moreover, in real-world scenarios, bioplastic degradation may take years, resulting in MNP formation and additive release that present hazards comparable to those generated by petrochemical-derived plastics [
5]. Second, molecularly engineered recyclable plastics are being developed by embedding reversible chemical bonds into polymer backbones [
6]. Such designs enable depolymerization under mild, often solvent-free conditions, making it possible to recover the monomers for repolymerization. While recent studies have demonstrated elegant chemistries, such as ring-opening metathesis [
7] or acceptorless dehydrogenative polymerization [
8], the path to industrial-scale production remains uncertain, as these new materials must still demonstrate cost-competitiveness, scalability, and compatibility with existing manufacturing and recycling infrastructures. The third and most immediate route of action is to improve the recycling of conventional petrochemical plastics. Recent advancements in plastic management facilities—particularly the development and implementation of modern recycling technologies—have significantly elevated the market share of recycled plastics to 8.7% [
4]. Nevertheless, many mechanical and chemical plastic recycling processes continue to experience high energy consumption, low purity of output streams, and the production of secondary wastes along with high carbon dioxide (CO
2) emissions, indicating a need for substantial improvements [
1].
Within this context, enzyme-catalyzed depolymerization has emerged as an alternative bio-based plastic recycling method [
9]. Enzymes function effectively under mild, aqueous conditions, demonstrating high catalytic specificity for polymeric backbones containing hydrolyzable ester, carbamate, and amide bonds, without the need for harsh chemicals or high energy demands [
10,
11]. The successful industrial demonstration of enzymatic poly(ethylene terephthalate) (PET) recycling, facilitated by engineered polyester hydrolases, suggests that the primary challenge has shifted from identifying active catalysts to enhancing bioprocess scale-up and feedstock pretreatment [
11-
13] or extending upcycling opportunities to products with significant added value [
14,
15].
Challenges still exist in this research domain, including the limitation of catalytic efficiency imposed by the natural structural diversity of enzymes, which were not specifically evolved for man-made materials [
16]. Additionally, there is a need to apply insights gained from bio-based polyester recycling to more recalcitrant polymers, necessitating the continuous development of new biocatalytic approaches. This opinion article highlights several recent advancements, including the artificial intelligence (AI)-driven design [
17,
18] of plastic-degrading enzymes intended to exhibit enhanced properties beyond natural scaffolds and the use of multi-enzyme systems to address complex plastic substrates such as those with multiple hydrolyzable bonds (e.g., polyurethane, PUR) or those without functional groups (e.g., polyolefins). These developments aim to optimize the synergistic effects of various enzyme classes, thereby improving overall depolymerization performance and broadening the range of target plastics, while also addressing current technical limitations to advance toward a truly circular economy.
2. AI-assisted design of novel PET hydrolases
After two decades of extensive research and multiple engineering efforts, it appears that naturally occurring polyesterase-lipase-cutinase (PLC)-like hydrolases have reached a performance limit in PET depolymerization [
13]. Mining the sequence space for distant homologs verifies this hypothesis: As the identity to benchmark PLC scaffolds decreases, catalytic activity also diminishes, indicating that natural designs may impose inherent constraints [
19,
20]. To overcome these constraints, researchers have turned to AI-guided
de novo strategies to identify and develop entirely new biocatalysts or reshape existing motifs into more efficient architectures (
Fig. 1). Below, we discuss three representative approaches.
In a striking departure from canonical hydrolases, one team repurposed fragaceatoxin C (FraC), a pore-forming protein from
Actinia fragacea, into a PET-hydrolyzing nanoreactor [
21]. Employing the Protein Energy Landscape Exploration (PELE) software, the researchers identified three potential binding sites for small ester substrates and incorporated catalytic triad residues along with oxyanion-hole geometries through site-directed mutagenesis. The assembly of the modified FraC into an octamer resulted in the formation of a "catalytic nanopore" featuring up to eight active sites. Biochemical characterization of PET nanoparticles at 40 °C demonstrated a rate constant (
kcat) greater than 8 s
-1, significantly surpassing the rates of most globular PET hydrolases (0.5-2.0 s
-1) [
13]. This improvement may be due to the presence of multiple catalytic triads within the engineered nanopore [
22]. However, the ability to hydrolyze larger PET substrates is constrained by the limited access to the buried active sites, dictated by the size of the nanopore opening, thereby illustrating a trade-off between turnover and substrate accessibility.
Recently, Lauko et al. [
23] proposed a workflow for designing serine hydrolases from scratch, utilizing a suite of computational tools developed by Baker’s team at the University of Washington. Starting with transition-state geometries derived from canonical ester hydrolysis mechanisms, the researchers effectively positioned side-chain coordinates, constructed backbones surrounding the catalytic triad, and optimized sequences before subjecting the designs to AlphaFold2 for structural predictions. After several cycles of iterative computation and evaluation of the new design’s active-site pre-organization using a deep neural network named PLACER (stands for protein-ligand atomistic conformational ensemble resolver), which is essential for creating hydrolytic activity, selected designs were recombinantly produced and biochemically characterized. One of the most active designs, referred to as "momi120-103" (
Fig. 1(a)), has been optimized with a ligand composed of two PET repeating units and demonstrates the ability to hydrolyze the model substrate 4-methylumbelliferone phenylacetate at a
kcat of 0.057 s
-1. This rate constant is significantly lower than that of established PET hydrolases on comparable substrates [
13], indicating that creating an efficient true
de novo PET hydrolase remains a distant objective. In comparison with other naturally evolved PET hydrolases (
Figs. 1(c) and
(d)), the catalytic site’s degree of burial may again hinder its access to the polymeric substrate.
A third strategy involves deconstructing the leaf-branch compost cutinase (LCC) into its essential functional motifs, including catalytic residues and neighboring secondary structures, and subsequently re-scaffolding them onto
de novo backbones [
24]. Following the identification of active-site segments via molecular docking, Ding et al. [
24] utilized the joint RoseTTAFold™ network (RF
joint)-guided inpainting [
25] to complete the surrounding sequence, thereby generating virtual enzymes with low root mean square deviation of distances between backbone C
α atoms (C
α-RMSD) relative to the original LCC motif. Experimental hits with low expression levels were further improved through iterative computation employing tools to refine structures and sequences. The optimal redesigned (Rs)PETase1 (
Fig. 1(b)) variant is 30% shorter than LCC and exhibits only 34% sequence identity, while maintaining comparable catalytic efficiency on amorphous PET substrates. However, it has significantly lower thermostability, as evidenced by a melting point of 56 °C compared with that of wild-type LCC at around 85 °C [
26]. The notable structural similarity to LCC (C
α-RMSD of 1.66,
Fig. 1(c)) highlights a relationship between the natural PET hydrolase scaffold and its enhanced depolymerization capability, as previously suggested [
19]. Given that other naturally evolved PET hydrolases from a fungal origin with similar molecular weights also exist (
Fig. 1(d)), the advantages of this structurally "truncated LCC" require further evaluation to substantiate its benefits for future applications.
Together, these pioneering efforts demonstrate that AI-driven enzyme design can reconfigure the catalytic landscape established by natural PET hydrolases. Nonetheless, high catalytic performance appears to be significantly dependent on a structural resemblance to the naturally evolved PLC-like PET hydrolases rather than new structural features. In future design efforts, challenges such as the accessibility of polymeric substrates and recombinant expression yields must be systematically addressed alongside the introduction of hydrolytic activity into new protein scaffolds.
3. Designing multi-enzyme systems for accelerated plastic depolymerization
The enzymatic depolymerization of plastics poses distinct challenges when compared with biotransformation involving small molecules. Insoluble, high-molecular-weight polymer substrates dictate the characteristics of this reaction as interfacial catalysis [
27], where the accessibility of the enzyme to each scissile bond within a single polymer chain can vary significantly. Multi-enzyme systems, in which each catalyst targets a specific bond or sequentially facilitates the formation of cleavable bonds, provide promising strategies to address the inherent degradability limitations of various types of plastics [
11].
This concept can easily be exemplified with the chemically simplest polyester substrate. Dual-enzyme approaches have demonstrated great advantages over the use of a single PET hydrolase [
28-
30]. The soluble degradation intermediates, comprising short-chain aromatic esters, can competitively inhibit PET hydrolases. However, the inclusion of a secondary "helper" enzyme that specifically hydrolyzes these inhibitory small molecules [
31,
32] can mitigate product inhibition in a tandem action, thereby significantly enhancing the overall conversion.
The depolymerization of PURs—particularly those containing polyester-based polyols [
33] —can benefit from a similar strategy employing various enzyme classes. Polyester hydrolases specifically target the ester bonds in soft polyol segments [
34,
35], while carbamate hydrolases cleave urethane bonds within hard segments [
36-
39]. Collectively, the depolymerization levels achieved were significantly greater than those reported for any single enzyme class acting independently. A similar dual enzyme system demonstrated promising efficacy in the depolymerization of mixed plastics containing both PUR and various polyesters (
Fig. 2(a)) [
40]. This scenario closely resembles the complex challenges of waste plastic disposal in the real world.
To apply multi-enzyme concepts to non-hydrolyzable polymers such as polyolefins, it is necessary to implement pre-processing steps involving harsh chemicals, ultrasonication, and specific enzymatic oxidation in order to incorporate labile functionalities into the inert polymeric backbones [
41]. The subsequent depolymerization of low-molecular-weight polyethylene (PE) or polyvinyl alcohols (PVAs) via an enzyme cascade begins with an alcohol dehydrogenase (ADH) that converts hydroxylated side chains into ketones, which are subsequently converted into backbone esters by a Baeyer-Villiger monooxygenase (BVMO;
Fig. 2(b)).
In situ cofactor recycling enables the formation of multiple ester bonds without the need for a continuous supply of the nicotinamide adenine dinucleotide phosphate cofactors (NADP
+/NADPH). This cascade demonstrated significantly higher transformation efficacy for the water-soluble vinyl polymer PVA [
42], compared with insoluble functionalized PE, which has considerably fewer side-chain hydroxyl groups. The polyester-like main chains can be effectively depolymerized by various esterases, resulting in the production of, for example, ω-hydroxyacids or α,ω-dicarboxylic acids. Obviously, this biocatalytic approach cannot revert the polymer substrate to its original educts (i.e., ethylene or vinyl acetate) for initial synthesis. On the other hand, an upcycling strategy that employs these degradation products for microbial growth and the biosynthesis of value-added chemicals is now feasible [
43]. While these enzyme cascades demonstrate significant advantages over numerous previously reported studies utilizing single oxidative enzymes, which often suffer from low reproducibility and uncertain mechanisms [
44,
45], the performance of current multi-enzyme methods still falls short of industrial benchmarks. Rate-limiting steps include low turnover of oxidative enzymes on solid substrates, mass-transfer barriers at the interface, and cofactor regeneration in large reactors. Future optimization must focus on engineering enzyme modules for greater substrate affinity, integrating downstream microbial conversion in consolidated bioprocesses, and coupling with pretreatment technologies (e.g., mild pyrolysis or mechano-chemical activation) to improve polymer accessibility [
46].
4. Conclusions and outlook
The ongoing issue of plastic pollution highlights the critical need for sustainable solutions throughout the entire life cycle: developing new polymers that incorporate recyclability and advancing recycling technologies to tackle present waste materials. Despite the long-standing recognized risks of plastic pollution, recent unsuccessful negotiations for the Global Plastic Treaty underscore the considerable difficulties in aligning scientists, industries, and policymakers across nations to effectively address this global challenge through binding agreements and commitments [
47].
As biochemists, we assert that bio-based innovations will be crucial for future advancements in plastic recycling. Insights from the successful biological PET recycling sector suggest that any prospective laboratory advancements must be evaluated immediately for their technical scalability, cost-effectiveness, and compatibility with existing recycling infrastructures. Alongside comprehensive insights into future research directions in this evolving domain [
11,
13], the increasing demand for novel (non-natural) biocatalysts and their combinatorial and synergistic capabilities, as emphasized in this article, requires extensive further investigations. In any case, aligning scientific innovation with industrial feasibility at the earliest stages is essential for closing the loop on polymer use and promoting a real green transition to a circular plastic economy.
PDF
(1044KB)
