Improved nylon polymerization using amide diads

Abstract

Nylons, a major class of synthetic polyamides, are widely used due to their excellent mechanical strength, thermal stability, and chemical resistance. Conventional nylon production relies on the polymerization of lactams or stoichiometric nylon salts. However, applying these approaches to unconventional precursors such as bio-derived glutaric acid produces polymers with low molecular weight and limited applications. To address these challenges, we demonstrated that chemically-synthesized nylon diads enable the production of higher-molecular-weight polyamides compared with traditional salts. We then identified a biosynthetic approach using amide synthetases to convert unprotected bifunctional substrates into nylon-relevant diads. Using a cofactor regeneration system, enzymatic diad synthesis was scaled to produce sufficient material for laboratory-scale characterization and solid-state polymerization. Amide synthetases demonstrated broad substrate scope, catalyzing the regioselective assembly of diverse nylon-relevant diacids, diamines, and ω-amino acids. This strategy offers a novel route to synthesize challenging nylon monomers and advances production of bioderived nylons.

---

Green foundation

1. This work describes the biosynthesis of process-advantaged amide diads for improved polymerization of nylons from renewable feedstocks

2. By increasing the molecular weights of bioderived nylons by more than 5-fold, amide diads overcome a key obstacle in developing sustainable polymers

3. In the future, amide synthetases can be applied in vivo using engineered bacteria for direct conversion of renewable feedstocks into amide diads

Introduction

Polyamides are polymers linked by amide groups and synthesized by the reaction of diamines with dicarboxylic acids, by the condensation of ω-amino acids, or by ring-opening polymerization of cyclic lactams. Nylons, a prominent category of synthetic thermoplastic polyamides, were produced at an estimated scale of 9 million tons in 2020 (Global Industry Analysts, 2020). Applications of nylons are widespread due to their exceptional mechanical strength, thermal stability and chemical resistance, making them indispensable in industries such as textiles, automotive manufacturing, and additive manufacturing. The continued demand for high-performance materials has driven research efforts aimed at improving the synthesis and functional properties of nylons, with an emphasis on achieving higher molecular weight and expanding the scope of usable precursors.^1–4

Commercial nylons are typically produced by polymerizing either a single ω-amino acid (e.g., nylon-6, also known as PA6) or a stoichiometric nylon salt formed from a diamine and a diacid (e.g., nylon-66, or PA66). Efforts to improve nylon sustainability have led to interest in incorporating bioderived components such as glutarate and cadaverine.^5–13 However, polymerization of nylon salts containing glutarate often yields polymers with low molecular weight, which has hindered their broader exploration and application.^5,14 Despite the limited studies on glutarate-based nylons, PA55 composed of cadaverine and glutarate has been reported to possess desirable properties, including a high melting temperature and excellent thermal stability.¹⁴ PA65 remains largely unexplored but shows a distinct hydrogen bonding pattern and moderate crystallinity.^5,15 More flexible synthetic routes that can readily incorporate non-conventional monomers could expand the scope of potential polyamides.

As an alternative synthesis route, PA66 diads generated during enzymatic PA66 hydrolysis have been polymerized to yield PA66 with higher molecular weight than that generated from polymerization of the corresponding nylon salt.¹⁶ However, it was unclear whether other nylon diads would similarly outperform their respective salts. Furthermore, while oligomeric products can be formed from polyamide hydrolysis, direct diad synthesis is challenging. Peptide synthesis via solid-phase methods is well-established, but it involves expensive protection and deprotection steps to control chain length, sequence and reactivity, as well as extensive use of organic solvents.¹⁷ Alternatively, enzymatic synthesis could potentially offer a more efficient and environmentally friendly route for generating these intermediates. A biosynthetic approach would be particularly useful if combined with a pathway for biosynthesis of nylon-relevant monomers, since production of zwitterionic diads would avoid the cost of separately synthesizing, neutralizing, and purifying diacids and diamines.¹⁸

Biology provides several possible routes for amide bond formation. Ribosomes can synthesize amide bonds between natural α-amino acids. However, while ribosomes can generate polyamides with high molecular weight and exquisite sequence control, expanding the monomer scope beyond canonical α-amino acids is challenging.^19–25 Additionally, non-ribosomal peptide synthetases (NRPSs) are multimodular enzymes that offer greater flexibility in monomer scope and bond formation but are difficult to express and engineer.^26–28 In addition to ribosomal and non-ribosomal peptide biosynthesis, recent work has demonstrated that enzyme promiscuity can also enable amide bond formation through distinct enzymatic mechanisms. Certain alcohol dehydrogenases have been shown to catalyze oxidative coupling of alcohols with ammonia or amines via hemiaminal intermediates, providing a redox-driven route for synthesis of primary and secondary amides.²⁹ Additionally, amide synthetases can catalyze amide bond formation in small molecules by using ATP to adenylate a carboxylic acid, which is then condensed with an amine.^30,31 These enzymes natively function in the biosynthesis of complex secondary metabolites such as antibiotics and siderophores and have been mainly repurposed for the production of pharmaceuticals.^32–36 They are relatively small (400–500 amino acids), easy to express and purify, and active in heterologous hosts.

Enzymatic amide bond formation has emerged as a sustainable alternative to traditional synthetic approaches. For example, the amide synthetase McbA from the ANL superfamily of adenylating enzymes is known for its broad substrate specificity and has been applied to the chemoenzymatic synthesis of β-carboline derivatives and other pharmaceutical-type amides.^33,37 However, McbA has been shown to have low activity with aliphatic substrates, suggesting that it may not accept nylon-relevant monomers.³² Similarly, the amide synthetase SfaB, a self-sufficient AMP ligase from the ANL superfamily that participates in an NRPS pathway, accepts a broad range of short-chain fatty acids and catalyzes amidation or thioesterification with various amine or thiol nucleophiles, but reportedly does not tolerate dicarboxylic acids such as those used in nylons.³⁸ Therefore, the application of natural or engineered amide synthetases for the synthesis of nylon building blocks remains largely unexplored.

In this study, we showed that chemically-synthesized nylon diads can be used to generate a range of polyamides with higher molecular weights compared to polymerization of traditional salts. Motivated by these findings, we then demonstrated a simplified biosynthetic strategy using amide synthetases to produce nylon diad precursors from unprotected bifunctional substrates. By coupling amide bond formation with an ATP-regeneration system, we optimized and scaled up this biocatalytic process and established it as a viable method for synthesizing nylon diads. Furthermore, we showed that amide synthetases exhibit a broad substrate scope and can catalyze the regioselective assembly of a diverse range of nylon-relevant diacids, diamines, and ω-amino acids. This approach provides a facile route for the polymerization of challenging monomers and expands the potential for tailoring nylon properties through enzymatically derived precursors.

Results and discussion

Diad polymerization and characterization

To extend the previous results showing the advantages of diad-based precursors in PA66 synthesis,¹⁶ we tested whether additional diads, particularly those incorporating glutarate, could also increase the molecular weight of the resulting polymer compared to the corresponding nylon salt (Fig. 1A). These diads were prepared from combinations of cadaverine (C) or hexamethylenediamine (M) as the diamine, and succinic acid (S), glutaric acid (G), or adipic acid (A) as the carboxylic acid (see Fig. S1 for full list of substrate codes). We previously synthesized the MA diad by coupling N-Boc-1,6-hexamethylenediamine with monomethyl adipate using EDC as the coupling agent.¹⁶ Although we initially attempted to synthesize the diads directly from unprotected diamines using the procedure reported by Madhavachary et al.,³⁹ we ultimately found this method overly cumbersome as it required complex chromatographic separation to completely remove the difunctionalized byproduct. To this end, CS, MS, CG, MG, CA, and MA (Fig. 1B) were synthesized through conventional methods to provide diads in sufficient quantity for polymerization (see SI and Fig. S2–S37 for details). These diads were polymerized by the solid-state polymerization (SSP) protocol described in our previous work.¹⁶ Number average and weight average molecular weights (M_n and M_w, respectively) of the resulting polymers were measured by size exclusion chromatography (SEC), and M_w values were compared to polyamides prepared from traditional nylon salts (Fig. 1C). These data showed diads containing adipic acid or glutaric acid moieties produced polyamides with significantly higher M_w than those synthesized from their equivalent nylon salts. Molecular weight data was not recorded for PA55 synthesized from the CG diad because the resulting polymers were insoluble in HFIP. However, analysis of polymer crystallinity by differential scanning calorimetry (DSC) showed that PA55 prepared from the CG diad provided a higher melting temperature (T_m) and lower enthalpy of melting (ΔH_m) than the equivalent polyamide produced from the PA55 salt, consistent with higher molecular weight (Table S2). We hypothesized that the differences in M_w could be attributed to multiple factors, most notably by improving the stoichiometric balance of the diamine and diacid components, minimizing evaporation of volatile hexamethylenediamine, and starting the polymerization with 50% of the potential amide bonds preformed.

Fig. 1 Nylon diads yield polyamides with high molecular weight. (A) Comparison of PA65 synthesized by solid state polymerization (SSP) from conventional nylon salt or equivalent nylon diad. (B) Chemical structures of nylon diads. CS: cadaverine-succinic acid diad; MS: hexamethylenediamine – succinic acid diad; CA: cadaverine – adipic acid diad; CG: cadaverine – glutaric acid diad; MG: hexamethylenediamine – glutaric acid diad; CA: cadaverine – adipic acid diad; MA: hexamethylenediamine – adipic acid diad. (C) Comparison of weight-averaged molecular weights (M_w) of polyamides prepared from diads and their corresponding nylon salts through solid-state polymerization. M_w data is not reported for CG because the polymers synthesized from this diad were insoluble in HFIP. Sample size is N = 3. Error bars are shown as the mean ± standard deviation. (D) Excess mass loss as determined from isothermal TGA studies. Excess mass loss is calculated as the difference between the experimentally determined mass loss and the expected mass loss assuming 100% monomer conversion.

To test these hypotheses, we used isothermal thermogravimetric analysis (TGA) to monitor the change in sample mass under simulated SSP conditions. Diad and nylon salt samples were initially conditioned at 100 °C for 1 h to remove any residual solvent and/or moisture, and subsequently polymerized by heating to 220 °C for 8 h (Fig. S40–S45). As shown in Fig. 1D, most nylon salts lose a considerably larger portion of their initial mass than their diad counterparts over the course of the polymerization, even when accounting for differences in expected mass loss caused by the evolution of water. Moreover, the majority of this mass loss occurs at relatively low monomer conversion (i.e. within approximately the first 30 min). These results imply that evaporation of diamine and/or diacid components of the nylon salts creates a stoichiometric imbalance which limits maximum X_n in accordance with Carother's equation.⁴⁰ Exceptionally high mass losses were observed during the polymerization of succinic acid-containing PA64 and PA54 samples regardless of whether the diad or nylon salt starting materials were used (Fig. S44 and S45). This mass loss was ultimately attributed to competing imidization pathways which can create volatile byproducts when imidization occurs adjacent to a polymer chain end. Imide formation was clearly visible from Fourier-transform infrared (FT-IR) spectroscopy of the PA64 and PA54 polymers, which showed an imide C=O stretch at ∼1770 cm⁻¹ (Fig. S46). Thermal imidization also explains the relatively low M_w of PA64 and PA54 samples since imidization imbalances the ratio of carboxylic acid to amine chain ends and also acts as a form of ring-chain equilibrium which further limits X_n.⁴¹

To compare the physical properties of the polyamides prepared from diads with those prepared from the corresponding nylon salt, we used thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) to measure the degradation temperatures (T_d), glass transition temperatures (T_g), melting temperatures (T_m), and enthalpies of melting (ΔH_m) (Fig. S47–S52 and Table S2). No considerable difference was observed in T_d for PA66, PA56, PA65, PA55, or PA54 prepared from either nylon salts or diads. However, a somewhat lower T_d was observed for PA64 synthesized from MS compared to its corresponding salt (Fig. S51A). Additionally, PA64 prepared from MS showed an apparent two step decomposition profile, whereas the salt-derived polyamide decomposed in a single step. Despite this difference in thermal stability, both PA64 samples display nearly identical thermal properties, consistent with the hypothesis that the two samples are chemically identical. Likewise, similar agreement in T_g, T_m, and ΔH_m for salt-derived and diad-derived polyamides were observed for PA66, PA65, and PA54. In combination, we found that diads are preferred feedstocks for nylon polymerization, since they yield polyamides that are chemically identical but have higher molecular weights compared to the corresponding salt-derived polyamides.

Amide synthetases can synthesize nylon diads

Motivated by findings that diads are promising precursors for nylon polymerization, we sought to investigate more efficient routes for diad synthesis. To overcome the need for protecting groups in chemical diad synthesis, we explored a biosynthetic route using amide synthetases to couple unprotected substrates (Fig. 2B). To evaluate the activities of amide synthetases on nylon building blocks, five enzymes (DdaG, SfaB, SuCphA1, McbA and PylC) reported to catalyze isopeptide bond formation were heterologously expressed in E. coli BL21(DE3). Among these, four enzymes (DdaG, SfaB, SuCphA1 and McbA) were successfully purified as soluble hexahistidine-tagged proteins (Fig. S53 and S54), while His-tagged PylC was insoluble. To enhance PylC solubility, a SUMO tag was added to the C-terminal of the hexahistidine tag, resulting in a soluble His-SUMO-tagged PylC (Fig. S55).⁴² We first confirmed the reported activities of DdaG to ligate 2,3-diaminopropionate with fumarate,³⁶ SfaB to ligate 5-chlorovaleric acid with cadaverine,³⁸ and SuCphA1 to ligate aspartate with arginine.⁴³ However, the His-SUMO-tagged PylC failed to exhibit its native activity of ligating lysine and ornithine.⁴⁴ Attempts to cleave the His-SUMO tag from the recombinant PylC were unsuccessful. Therefore, we next conducted in vitro biochemical assays using the soluble enzymes (DdaG, SfaB, SuCphA1, and McbA) with nylon constituents (adipic acid, hexamethylenediamine, succinic acid, cadaverine, and glutaric acid) to assess and quantify the production of the respective diads (CS, MS, CG, MG, CA and MA) by comparison to chemically-synthesized standards using direct injection immediate droplet-on-demand/open port sampling interface-mass spectrometry (IDOT/OPSI-MS) (Fig. 2C and S56).

Fig. 2 Diad quantification. (A) Chemical synthesis of the hexamethylene diamine – succinic acid (MS) diad. (B) Enzymatic synthesis of the MS diad. (C) Small-scale quantification of enzymatically synthesized nylon diads. In vitro biochemical assays were conducted by incubating 5 mM different diacids (i.e., succinic acid, glutaric acid, or adipic acid) with 5 mM diamines (i.e., cadaverine or hexamethylenediamine) in the presence of 10 μM enzymes (i.e., DdaG, SfaB, SuCphA1 or McbA) at 30 °C for 16 h with shaking at 200 rpm. Products were quantified by comparing to the chemically-synthesized standards using I.DOT/OPSI-MS. Sample size is N = 3. Error bars are shown as the mean ± standard deviation. (D) ¹H NMR spectra of enzymatic-synthesized and chemically-synthesized MA diad.

Encouragingly, DdaG and SfaB showed significant activities and distinct specificities for nylon diad synthesis. The molar yields after overnight incubation ranged from 3% to 80%, with DdaG demonstrating the highest activity in synthesizing MS (∼80%) and CS (∼60%) but low production of MG and MA. In contrast, SfaB exhibited lower activity than DdaG for production of MS and CS but produced higher amounts of MG (∼20%) and MA (∼3%). No detectable formation of triads, such as diacid triads (e.g., SMS and AMA) and diamine triads (e.g., MSM and MAM) or higher-order oligoamides were observed. This result indicated that those amide synthetases actively catalyze diad formation but do not extend to higher-order oligoamides. We next scaled up MA production with SfaB. As an additional safeguard, we sought to confirm the structure of enzymatically-produced MA. To this end, we developed a purification protocol wherein the crude reaction mixture was reacted with 2-acetyldimedone (dde-OH) to provide a dde-functionalized adduct that is easily isolated chromatographically, and subsequently deprotected via hydrazinolysis (Scheme S1 and Fig. S57 and S58). This allowed us to confirm the structure of enzymatically synthesized MA by comparing the ¹H NMR spectra to the chemically synthesized standard (Fig. 2D). Scale-up of this process would require the development of a more efficient purification route.

We did not observe detectable product formation in reactions with SuCphA1 or McbA, which is consistent with findings of a previous report that McbA does not accept aliphatic substrates.³² Interestingly, previous studies showed that SfaB cannot adenylate diacids for subsequent condensation with hydroxylamine, based on colorimetric detection of pyrophosphate released during adenylation.³⁸ However, our observations reveal that SfaB can act on a range of diacids (i.e., succinic acid, glutaric acid and adipic acid) in combination with cadaverine and hexamethylenediamine. These differing results suggest that SfaB also exhibits specificity towards the amine acceptor.

Scale up and optimization of diad biosynthesis

To demonstrate the utility of enzymatic synthesis for diad production, we next focused on optimizing conditions for scale-up and isolation, using synthesis of MS by DdaG as a model reaction. We increased the concentration of each substrate to 50 mM and increased the reaction volume to 50 mL, while keeping the concentration of DdaG at 10 μM. Based on OPSI-MS analysis, the reaction yield was 44%. The crude reaction mixture was then purified following the protocol previously described for MA to provide 60 mg (∼11%) of spectroscopically pure MS (Fig. S59 and S60).

Since the reactions did not fully convert the substrates to products, we next investigated factors that might limit yield. To assess the potential for product inhibition, we conducted reactions in which we incubated DdaG with 5 mM succinate and 5 mM hexamethylenediamine, with or without the addition of 2.5 mM of the purified MS diad. The reactions produced similar increases in MS concentration (Fig. S95A), indicating that product inhibition was not significantly limiting the yield. Next, we tested for diad hydrolysis by incubating purified DdaG with the MS diad. Diad concentrations remained unchanged relative to the no-enzyme control (Fig. S95B), suggesting negligible hydrolysis. Finally, we incubated reactions for 24 h, measured the MS concentration, and then added an additional 5 μM of DdaG and let the reaction continue for an additional 24 h. The addition of fresh enzyme approximately doubled the final MS concentration (Fig. S95C), indicating the enzyme deactivation likely limits conversion. Together, these results indicate that the limited yield is unlikely to arise from strong product inhibition or hydrolytic equilibrium and, instead, is associated with additional factors such as enzyme deactivation that could be addressed through enzyme engineering.

DdaG and SfaB are ATP-dependent amide synthetases, and providing super-stoichiometric ATP to ensure effective conversion is cost-prohibitive. As an alternative, we implemented an ATP recycling system using a purified type 2-III polyphosphate kinase (PPK12).^34,45,46 In addition, we included inorganic pyrophosphatase from E. coli (ecPPase) to prevent potential inhibition by pyrophosphate. We then performed reaction optimization using MS production by DdaG. We iteratively optimized the concentrations of the substrates, polyphosphate, magnesium, AMP, and pH (Fig. S61–65). Based on these results, the finalized optimal conditions for MS production were 50 mM succinic acid, 100 mM hexamethylenediamine, 100 mM MgCl₂, 10 μM DdaG, 1 mg mL⁻¹ PPK12, 5 mM AMP, 0.2 U ecPPase and 100 mM HEPES (pH 8). With these optimal conditions, we subsequently scaled up our reaction system with ATP regeneration from 50 μL to a 5 mL reaction volume. After 24 h incubation, a small volume (50 μL) of sample was analyzed by I.DOT/OPSI-MS, indicating a promising yield of ∼50% (∼5 g L⁻¹) and demonstrating a more than 3-fold improvement from the initial conditions (Fig. 3). Although these experiments were performed with a single reaction, they demonstrate proof of principle that enzymatic catalysis using amide synthetases for nylon diad synthesis is both scalable and amenable to optimization. Future work should focus on further scale-up and process optimization to improve productivity, robustness, and applicability to specific reactions. In addition, we calculated the E-factor, a key metric in green chemistry, to evaluate and compare the environmental impact of enzymatic and chemical diad synthesis. The ATP-regenerated enzymatic approach exhibits a substantially lower E-factor (i.e., 4.3–8.7; excluding or including HEPES buffer, respectively) than the conventional chemical synthesis route (i.e., 63), highlighting the potential of biocatalysis as a greener alternative.

Fig. 3 Diad scale up and optimization. (A) ATP-recycled scale-up of MS production. A 5 mL reaction with increased substrate loading and an ATP regeneration system was performed, and the yield was quantified by mass spectrometry. (B) Comparison of MS production under different approaches. The 50 mL reaction (left) was performed with initial reaction conditions: 50 mM succinic acid, 50 mM hexamethylenediamine, 10 mM MgCl₂, 10 μM DdaG, 50 mM ATP and 100 mM HEPES (pH 8). For the 5 mL non-optimized ATP-recycled reaction (middle), 50 mM ATP was replaced with 1 mg mL⁻¹ PPK12, 1 mM ATP, 1 mM AMP, 25 mM SHMP and 0.2 U ecPPase with other components unchanged. For the 5 mL optimized ATP-recycled reaction, 100 mM hexamethylenediamine, 100 mM MgCl₂ and 5 mM AMP were used with other components unchanged from the non-optimized 5 mL reaction. All the reactions were incubated for 24 hours at 30 °C with shaking at 200 rpm. These experiments were conducted in N = 1.

In vitro synthesis is well-suited for rapid lab-scale oligomer production to enable polymer synthesis and characterization. Longer term, in vivo strategies offer significant advantages for scale-up. Through metabolic engineering, intracellular ATP availability can be modulated to support ATP-dependent biosynthetic reactions without the need for external supplementation.⁴⁷ Moreover, the microbial biosynthesis of nylon-relevant monomers has been achieved. For example, engineered Escherichia coli strains have been developed for the production of adipic acid at high titers.⁴⁸ Similarly, de novo microbial synthesis of hexamethylenediamine and related diamines from glucose has been demonstrated.⁴⁹ Collectively, these advances highlight the potential for future work of developing in vivo platforms for the synthesis of nylon-relevant precursors, in which heterologous expression of amide synthetases could enable intracellular diad formation from endogenously produced substrates.

Enzymatic synthesis of novel nylon diads

Since DdaG and SfaB exhibit activities in producing nylon-relevant diads, we next sought to explore their substrate range, which could provide opportunities to develop novel nylons. Therefore, we conducted in vitro biochemical assays using DdaG and SfaB with a diverse range of substrates, including traditional nylon constituents (adipic acid A, hexamethylenediamine M, and terephthalic acid T) and novel compounds varying in chain length, functional groups (e.g., thiol, ketone, and hydroxyl groups), aromaticity, and cyclicity, including dicarboxylic acids, diamines, and ω-amino acids (Fig. S1).

We initially detected product formation using direct injection IDOT/OPSI-MS for untargeted characterization of enzymatic reactions.⁵⁰ Using this approach, 41 out of 96 reaction combinations generated detectable products above the control using either DdaG or SfaB (Fig. 4 and S66). To more accurately identify these products, we confirmed 24 structures through analysis of fragmentation patterns by MS² (Fig. S67–S90). Both enzymes displayed activities in ligating the carboxylic groups of the diacids with a wide spectrum of amine acceptors including linear diamines (cadaverine C, hexamethylenediamine M), an aromatic diamine (p-xylylenediamine X), a cyclic diamine (cis-1,4-cyclohexanediamine N), and the amine groups of ω-amino acids (6-aminohexanoic acid H).

Fig. 4 Amide synthetases have broad substrate range for potential nylon precursors. In vitro biochemical assays were conducted by incubating different diacids (i.e., succinic acid, glutaric acid, adipic acid and thiomalic acid) with diamines or ω-amino acids in the presence of enzymes (i.e., DdaG or SfaB) or no-enzyme control. Products were assayed using OPSI-MS. The log₁₀MS signal intensity was cut off at 5.0 to filter out noise. C: cadaverine; M: hexamethylenediamine; X: p-xylylenediamine; N: cis-1,4-cyclohexanediamine; B: 4-aminobutyrate; V: 5-aminovalerate; H: 6-aminohexanoic acid; E: 4-amino-2-hydroxybutanoic acid; F: 4-amino-3-hydroxybutanoic acid. Sample size is N = 3. Data for CS, MS, CG, MG, CA, and MA shown in Fig. 2B are replotted here as unquantified MS intensities for comparison.

Notable differences between DdaG and SfaB were observed in their preference for carboxylic acid donors, consistent with our quantification of their activity in producing more conventional nylon-relevant diads. DdaG exhibited excellent catalytic performance in ligating succinic acid (S) with a wide range of amine acceptors but showed lower activities with glutaric acid (G) and adipic acid (A) compared to SfaB. In contrast, SfaB was better using G and A as carboxylic donors with diverse amine acceptors. In addition, DdaG can also catalyze reactions involving terephthalic acid (T) and α-ketoglutaric acid (K) with multiple diamines, while SfaB was inactive in these reactions. Alternatively, SfaB was capable of activating 1,4-cyclohexanedicarboxylic acid (O) with specific diamines (i.e., M and X), whereas DdaG showed negligible activity (Fig. S66). It is also notable that both enzymes were unable to activate the carboxylic groups of ω-amino acids. Interestingly, we noticed both enzymes could activate diacids bearing additional functional groups such as thiomalic acid (P). The broad substrate scope of DdaG and SfaB suggested the potential of using amide synthetases to produce novel value-added oligoamides.

To obtain mechanistic insights into whether substrate binding interactions could contribute to the distinct diacid preferences of DdaG and SfaB, we used Boltz-2^51,52 to model representative enzyme–substrate complexes (i.e., DdaG-succinate/glutarate and SfaB-succinate/glutarate) and analyzed predicted affinity values across 100 models. For each enzyme–substrate pair, all 100 models had high Boltz confidence values (DdaG-glutarate > 0.89, DdaG-succinate > 0.89, SfaB-succinate > 0.91; SfaB-glutarate > 0.93). The predicted affinity trends were consistent with the experimental preferences. For SfaB, glutarate was favored over succinate in 82 of 100 paired models, with median predicted IC₅₀ values of 626 and 851 μM, respectively. For DdaG, succinate was favored over glutarate in 99 of 100 paired models, with median predicted IC₅₀ values of 265 and 519 μM, respectively (Table S3). We then used protein–ligand interaction profiler (PLIP)⁵³ to analyze the highest-affinity model for each enzyme–substrate pair to identify favorable noncovalent interactions (Fig. S96). In both enzymes, the two substrates showed similar hydrophobic and salt-bridge interactions, whereas the preferred substrates showed additional hydrogen-bonding interactions (Tables S4 and S5). In DdaG, the succinate-bound model contained two more hydrogen bonds than the glutarate-bound model. In SfaB, the glutarate-bound model contained two hydrogen bonds that were absent in the succinate-bound model. Together, the predicted affinity values from Boltz-2 and subsequent PLIP analysis suggest that differences in substrate binding interactions may contribute to the observed diacid preferences of DdaG and SfaB. Future crystal structures of specific enzyme–substrate complexes would provide more definitive molecular insights for understanding the substrate specificities.

We also determined that DdaG regioselectively ligated S with the asymmetric polyamide spermidine. Of the two possible primary amines in spermidine, DdaG selectively coupled S with the primary amine nearer the middle amine, producing only a single diad product. This regioselectivity highlights the potential of amide synthetases to synthesize ordered polymers from asymmetric substrates, which is challenging for traditional chemical synthesis methods (Fig. S91).⁵⁴

Given the distinct activities of these enzymes with diacids and their considerable tolerance for various diamines, we investigated the effect of diacid and diamine chain length on enzyme activity. We tested linear diacids and diamines with carbon lengths of 7 to 10, excluding C10 diamines due to solubility issues (Fig. S92). The results for DdaG were consistent with our previous findings, showing a preference for S as the diacid and comparable activity across the full range of diamine lengths. In contrast, SfaB, although less active in ligating S with longer diamines, showed activity with C7–C10 diacids and M.

To further characterize DdaG performance, we conducted kinetic studies using succinic acid and hexamethylenediamine as substrates. The k_cat and K_m values for DdaG with these substrates were determined to be 3.2 ± 0.2 min⁻¹ and 8.8 ± 1.8 mM (Table 1 and Fig. S93), respectively. Compared to its native reaction with fumaric acid and 2,3-diaminopropionic acid,³⁶ the k_cat with non-native substrates is 4-fold lower and K_m is 15-fold higher.

Table 1 Kinetic parameters of DdaG with nylon-relevant substrates (i.e., succinic acid and hexamethylenediamine) compared to native substrates (i.e., fumaric acid and 2,3-diaminopropionic acid)

Substrates	kcat (min^−1)	Km (mM)	Note
Succinic acid and hexamethylenediamine	3.2 ± 0.2	8.8 ± 1.8	This study
Fumaric acid and 2,3-diaminopropionic acid	11.6 ± 0.4	0.55 ± 0.07	Hollenhorst et al.^36

---

Finally, we tested amide synthetase activity using cell-free enzyme expression. When expressed in a cell-free system, DdaG showed high activity ligating its native substrates but low production of the PA64 diad (Fig. S94). Similarly, SfaB exhibited higher activity with a short chain fatty acid (i.e., 5-chlorovalerate) than glutarate. These results suggest that amide synthetases have substantial engineering potential to enhance their activity with non-native or non-preferred substrates, and a cell-free system could serve as an efficient platform for high-throughput enzyme engineering.³²

Conclusion

In this work, we demonstrated that nylon diads can be polymerized to yield polymers with higher molecular weight than the corresponding traditional salts, particularly for Nylon 65 incorporating the bioderived monomer glutaric acid. Soluble amide synthetases provide a simple route to regioselectively couple unprotected, nylon-relevant diacids, diamines, and ω-amino acids into amide diads. Reaction molar yields reached up to 80%, while reaction kinetics demonstrated substantial room for further improvement through enzyme engineering. Reaction scale-up and optimization highlighted both the potential for simple and efficient enzymatic synthesis of sufficient quantities of oligoamides for further characterization, and the need to develop isolation protocols that simplify their isolation. Moreover, the broad substrate scope of these amide synthetases allows for the facile synthesis of diverse novel nylon oligomers, thereby facilitating the development of next-generation polyamides with enhanced or specialized properties. Notably, this approach could be extended from in vitro to in vivo systems by integrating enzyme-catalyzed amide bond formation with microbial monomer synthesis via engineered metabolic pathways. By reducing the need for pH control and product separation, simultaneous biosynthesis and ligation would offer substantial process advantages for nylon synthesis from renewable feedstocks.

Methods

Materials

LB agar, LB medium and Terrific Broth were purchased from BD Difco. All chemicals were purchased from ThermoFisher, Alfa Aesar, or Tokyo Chemical Industry (TCI). Pyrophosphatase was purchased from New England Biolabs.

Synthesis and expression of enzymes

Genes encoding DdaG³⁶ (GenBank: ADN39487.1), SfaB⁵⁵ (GenBank: ATY72527.1), McbA³⁵ (GenBank: AGL76720.1), and PylC⁴⁴ (GenBank: AAZ69813.1) were synthesized by GenScript (Piscataway, NJ) with a N-terminal His₆ tag for purification and cloned into pET28a(+). The gene encoding SuCphA1⁴³ (GenBank: BAA17890.1) was synthesized by GenScript (Piscataway, NJ) with a N-terminal His6 tag for purification and cloned into pETDuet-1. For the His-SUMO-PylC construct, PCR amplified PylC was inserted into Plasmid #112793 purchased from Addgene and replaced the location of EfaCas1 using Gibson Assembly. Plasmids containing His-DdaG, His-SfaB, His-McbA, His-PylC, His-SUMO-PylC, His-SuCphA1, His-PPK12 were transformed into chemically competent BL21(DE3) Escherichia coli cells (Sigma-Aldrich, St Louis, MO) according to the manufacturer's specifications.

His-PPK12 was produced as described previously.⁴⁵ Expression and purification of His-DdaG, His-SfaB, His-McbA, His-PylC, His-SUMO-PylC and His-SuCphA1 were conducted following previously published methods. A colony or glycerol stock of BL21(DE3) E. coli containing plasmid DNA was used to inoculate 10 mL of LB medium supplemented with appropriate antibiotics. The culture was incubated at 37 °C and 250 rpm overnight. The overnight culture was then diluted 100-fold into 200 mL of Terrific Broth. The subculture was incubated at 37 °C at 250 rpm until an OD of 0.4–0.6 was reached. Protein expression was induced by adding IPTG to a final concentration of 0.1 mM. The temperature was decreased to 16 °C and the culture was grown with shaking overnight. The cells were harvested by centrifugation (4000 rpm, 40 min, 4 °C), resuspended in 5 mL of lysis buffer (50 mM HEPES, 300 mM NaCl, 10 mM imidazole, 10 mM MgCl₂, 10% glycerol, pH 8.0), and sonicated on ice. Cell debris was removed by centrifugation (10 000 rpm, 30 min, 4 °C). The supernatant was then filtered with an MCE filter (0.22 μm) and loaded at 3 mL min⁻¹ onto an ÄKTA Start FPLC (Cytiva Marlborough, MA) equipped with a 5 mL His-Trap™ column (Cytiva). The column was washed with wash buffer (50 mM HEPES, 300 mM NaCl, 30 mM imidazole, 10 mM MgCl₂, 10% glycerol, pH 8.0) before eluting with elution buffer (50 mM HEPES, 300 mM NaCl, 250 mM imidazole, 10 mM MgCl₂, 10% glycerol, pH 8.0) at 3 mL min⁻¹. Purified proteins were concentrated with an Amicon® Ultra centrifugal filter at 10 kDa cutoff, and buffer exchanged into exchange buffer (50 mM HEPES, pH 8.0, 100 mM NaCl, 10 mM MgCl₂, 10% glycerol). The final purified product was analyzed by SDS-PAGE and enzyme concentration was measured using a NanoDrop™ 1000 Spectrophotometer (Thermo Scientific) with exchange buffer as a blank, and the extinction coefficient of each protein was calculated using ProtParam from the ExPASy Proteomics Server. The molecular weight of each protein was calculated using ProtParam from the ExPASy Proteomics Server, and further confirmed by the SDS-PAGE method. Pure enzymes were stored at −80 °C for subsequent activity assays.

In vitro enzyme activity assays

A typical enzymatic reaction was carried in triplicate at 50 μL containing 100 mM HEPES (pH 8.0), 10 mM ATP, 10 mM MgCl₂, 10 μM enzyme, 5 mM carboxyl group-containing compounds and 5 mM amine group-containing compounds, and was incubated at 30 °C for 16 h with shaking at 200 rpm. The reaction was quenched by adding 50 μL methanol. The samples were centrifuged at 4000 rpm for 10 min, and the supernatant was subjected to MS analysis to determine the formation of corresponding products. For the 50 ml reaction, 100 mM HEPES (pH 8.0), 50 mM ATP, 50 mM MgCl₂, 10 μM enzyme, 50 mM disodium succinic acid and 50 mM hexamethylenediamine were incubated at 30 °C for 16 h with shaking at 200 rpm. The reaction was quenched by adding 50 mL methanol. The samples were centrifuged at 4000 rpm for 20 min, and the supernatant was subjected to product isolation.

Enzymatic reaction optimization coupling with ATP regeneration system

Reactions with ATP regeneration were performed in 100 mM HEPES pH 8, 1 mM ATP, 1 or 5 mM AMP, 10–100 mM MgCl₂, 5–25 mM SHMP, 50 mM disodium succinic acid, 50–250 mM hexamethylene diamine,10 μM DdaG, 0.2 U ecPPase, and 1 mg mL⁻¹ PPK12. These reactions were then incubated at 30 °C with shaking at 200 rpm. After 24 h, reaction mix was quenched with 1 volume of methanol and samples were collected after centrifugation to clear insoluble precipitates.

Kinetic analysis of DdaG on reaction of succinate and hexamethylenediamine

The ligation of succinate and hexamethylenediamine by purified DdaG was monitored by observation of the product MS formation according to OPSI-MS. Kinetic assays were performed in triplicate with a reaction mixture containing 0.5 μM DdaG, 10 mM ATP, 100 mM HEPES (pH 8), 10 mM MgCl₂, 0.5 mM to 50 mM disodium succinate, 50 mM hexamethylenediamine in total volume of 50 μl reaction. Samples were taken after incubation at 30 °C for 5 min, 15 min, 25 min, and 35 min. The reaction was quenched by adding 50 μL methanol. The samples were centrifuged at 4000 rpm for 10 min, and the supernatant was subjected to OPSI-MS analysis to determine the formation of corresponding products. Initial velocities were determined by linear regression analysis of the data. The kinetic parameters k_cat, and K_m and standard error were calculated by Graphpad Prism 9.

I.DOT/OPSI-MS analysis of enzyme activity

The immediate drop-on-demand technology (Dispendix GmbH, Stuttgart, Germany) coupled with open port sampling interface-mass spectrometry (I.DOT/OPSI-MS) was used to analyze enzymatic reactions as previously described in detail.⁵⁰ Briefly, single droplets were mixed with solvent in an on-line fashion, enabling sampling throughputs of up to 2 s per sample. Here, enzymatic reactions were diluted 1 : 100 (v/v) in high performance liquid chromatography (HPLC) grade water with 500 nM propranolol acting as an internal standard for droplet capture. 40 µL of diluted reactions were transferred to a I.DOT S.100 96-well plate and analyzed without further separations. The I.DOT system was used to eject 20 nL of sample into the OPSI, into a flow of 75/25/0.1 (v/v/v) acetonitrile/water/formic acid and transported the sample to the electrospray ion source of the mass spectrometer. Dispensing throughput was 4 s per sample and each sample was measured in triplicate. Peak widths were ∼1.2 s wide.

For identification of products by high mass resolution, a Thermo Q-Exactive HF mass spectrometer (ThermoFisher Scientific) in positive ion mode was used for characterization. Scan settings were sheath gas = 80, auxiliary gas = 40, electrospray voltage = 4 kV, ion injection time = 50 ms, automatic gain control = 3e⁶, capillary temperature = 200 °C, mass/charge (m/z) range = 100–750 m/z, and OPSI solvent flow of 250 µL min⁻¹. A 60 000 mass resolution was used for quantitative analyses while 240 000 resolution was used for compound identification. Tandem MS data were acquired using the Q-Exactive HF, but water adduct formation was commonly observed in the ion trap. To avoid water adducts in tandem mass spectra, tandem mass spectrometry data was collected using a Sciex 5600 TripleTOF time of flight mass spectrometer (Sciex) with the following settings: GS1 = 75, GS2 = 35, electrospray voltage = 5.5 kV, capillary temperature = 200 °C, and OPSI solvent flow of 150 µL min⁻¹. Collision energies across 10–50 eV were evaluated for each ion.

Control of I.DOT settings (positioning, well selection, timing and droplet dispensing cycles), OPSI peak finding, data extraction from vendor file formats, and signal integration were enabled through custom software packages written in Delphi, Python, and C#, available upon request. For each droplet sampling event, the average mass spectrum from the resulting mass spectral peak was normalized to propranolol internal standard signal and background subtracted. Log–log matrix matched calibration curves were used for quantitation. Mass spectrometric signals typically spanned several orders of magnitude, thus, log–log calibrations resulted in reduced error across the full concentration range.

Tandem mass spectra

I.DOT/OPSI-MS² data were acquired for the possible observed compounds shown in Fig. 2 and 3 in the main text. MS² data were acquired on a Sciex 5600 TripleTOF + instrument using the same OPSI-MS probe as described previously. A time-of-flight system was preferred as in trap water adducts were observed for some compounds when fragmented with the orbitrap system. Putative fragment identifications are given for each compound using hypothesized fragmented pathways. All tandem mass spectra shown used a collision energy of 20 eV and a declustering potential of 100. Additional collision energies were used to help with fragment identifications as necessary (data not shown). For brevity, a single tandem mass spectrum is shown for each possible compound. For compounds without a tandem mass spectrum, signals for those compounds were either too weak to generate clear tandem MS or were obfuscated by overlapping background ions in the 1 m/z isolation window.

Nuclear magnetic resonance (NMR) spectroscopy

¹H and ¹³C nuclear magnetic resonance NMR spectroscopy was performed on a Bruker Avance III 400 NMR spectrometer operating at 400 MHz (¹H) or 100 MHz (¹³C). Chemical shifts are reported in parts per million (ppm) relative to residual protonated solvent.

Size exclusion chromatography (SEC)

SEC analysis was performed on an Agilent 1260 Infinity II LC system equipped with an Agilent PL HFIPgel guard column (9 μm, 50 × 4.6 mm) and two Agilent PL HFIPgel columns (9 μm, 250 × 4.6 mm). A 0.02 M CF₃COONa HFIP solution was used as the mobile phase at 40 °C and a flow rate of 0.300 mL min⁻¹. Elution time was monitored using a differential refractive index (dRI) detector, a light scattering detector operating with two angles at 90° and 15°, and a differential viscometer. Number-average molecular weights (M_n), weight-average molecular weights (M_w) and dispersities (Đ = M_w/M_n) were calculated based on calibrations with PMMA standards using the Agilent GPC/SEC software.

Thermogravimetric analysis (TGA)

TGA was performed on a TA Instruments Discovery TGA 55 under a nitrogen atmosphere. Sample mass was recorded as a function of temperature while the sample was heated from room temperature to 600 °C at a heating rate of 20 °C min⁻¹. Degradation temperature (T_d) was taken as the temperature at which the sample had lost 5% of its initial mass.

Isothermal TGA studies were performed on a TA Instruments Discovery TGA 55 using a platinum pan under a nitrogen atmosphere. Samples were initially conditioned at 100 °C for 1 h to remove any residual moisture or solvent, and subsequently heated to 220 °C for 8 h to simulate SSP conditions. Theoretical mass loss (Δm_th) was calculated from the following equation:

where MW_RU is the molecular weight of the polymer repeat unit, and MW_SM is the molecular weight of the salt or diad starting material. Excess mass loss was then calculated as the difference between the experimentally measured mass loss (Δm) and Δm_th.

Differential scanning calorimetry (DSC)

DSC was performed on a TA Instruments Discovery DSC 250. Measurements were carried out in standard aluminum pans using a heat-cool-heat cycle with 10 and 5 °C heating and cooling rates, respectively. Melting (T_m) temperature and enthalpy of melting were measured from the first heat cycle while glass transition (T_g) temperatures were measured from the second heating cycle.

Fourier-transform Infrared (FTIR) spectroscopy

Attenuated total reflectance (ATR) Fourier-transform infrared spectroscopy was performed using a PerkinElmer Spectrum Two FTIR spectrometer. Transmission spectra were collected from 4000 to 500 cm⁻¹.

Boltz-2 modeling and analysis

Boltz-2^51,52 was used to generate 100 models of each enzyme–substrate complex from their respective protein sequences and ligand representations (i.e., DdaG-succinate; DdaG-glutarate; SfaB-succinate; SfaB-glutarate). The adenosine monophosphate cofactor was included in all models using its Chemical Component Dictionary ID (AMP). SMILES strings used for succinate and glutarate were C(CC(=O)[O–])C(=O)[O–] and O=C([O–])CCCC(=O)[O–], respectively. Multiple sequence alignments required by Boltz-2 were generated with the ColabFold server.⁵⁶ A single contact restraint was applied during co-folding to ensure substrate binding in the substrate binding pocket. We applied a restraint with a maximum distance of 6 Å between a pocket residue (Thr328 for SfaB and Arg93 for DdaG) and one of the chemically equivalent carboxylate carbon atoms of each substrate.

For each enzyme–substrate pair, all 100 Boltz-2 predicted affinities were analyzed as an ensemble. Because affinity_pred_value is reported as log₁₀(IC₅₀), with IC₅₀ in μM, affinity predictions were characterized on the log₁₀-transformed scale. For each complex, we calculated the mean, median, standard deviation, interquartile range, minimum, and maximum of the log₁₀(IC₅₀ μM) values. Nonparametric bootstrap sampling was used to estimate 95% confidence intervals for the mean and median values. For each enzyme–substrate pair, 10 000 bootstrap datasets were generated by sampling 100 affinity values with replacement from the original set of 100 Boltz-2 affinity values. The mean and median were calculated for each bootstrap dataset, and the 2.5^th and 97.5^th percentiles of the resulting bootstrap distributions were used as the lower and upper bounds of the 95% confidence intervals, respectively. Back-transformed IC₅₀ values are provided in the text for interpretability. Predicted IC₅₀ values in μM were obtained by backtransforming the logarithmic values using 10^x. Substrate preferences for each enzyme were assessed by a paired comparison of matched model replicates, with the substrate having the lower affinity value considered the more strongly bound substrate.

For simplicity, we selected the model of each enzyme–substrate complex with the most favorable ligand binding affinity for structural analysis with the protein ligand interaction profiler (PLIP).⁵³ However, the affinity conclusions are based on the full ensembles.

Author contributions

L. Q. performed the majority of the enzymatic characterization and optimization. I. T. D.: oligomer synthesis, polymerization, and characterization. D. L. C: analytical characterization. J. M. P.: modeling and analysis. V. K.: analytical method development. N. T. Z.: characterization of long-chain substrates. N. D. O.: polymer thermal characterization. N. A. T.: oligomer synthesis. J. F. C.: analytical methods development and analysis, supervision of analytical research. J. C. F.: conceptualization, supervision of chemistry research. J. K. M: conceptualization; funding acquisition; supervision of biochemistry research.

Conflicts of interest

LQ, ITD, JCF, and JKM are inventors on a patent application related to this work.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6gc00885b.

Acknowledgements

Research was primarily sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725, for the U.S. Department of Energy. Cell-free expression was supported by the Center for Bioenergy Innovation (CBI), U.S. Department of Energy, Office of Science, Biological and Environmental Research Program under Award Number ERKP886. Work by N. T. Z. was supported by the U.S. Department of Energy, Office of Science, Office of Workforce Development for Teachers and Scientists (WDTS) under the Science Undergraduate Laboratory Internships (SULI) program. We thank Madan Gopal and Wilfred Chen (University of Delaware) for the PPK12 expression plasmid.

This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (https://energy.gov/downloads/doe-public-access-plan).

References

1. L. P. Manker, M. A. Hedou, C. Broggi, M. J. Jones, K. Kortsen, K. Puvanenthiran, Y. Kupper, H. Frauenrath, F. Marechal, V. Michaud, R. Marti, M. P. Shaver and J. S. Luterbacher, Nat. Sustain., 2024, 7, 640–651.
2. P. N. Stockmann, D. Van Opdenbosch, A. Poethig, D. L. Pastoetter, M. Hoehenberger, S. Lessig, J. Raab, M. Woelbing, C. Falcke, M. Winnacker, C. Zollfrank, H. Strittmatter and V. Sieber, Nat. Commun., 2020, 11, 509.
3. B. Yang, M. Yang, W. Li, H. Yan, W. Chen and Z. Liu, Ind. Eng. Chem. Res., 2025, 64, 14306–14317.
4. X.-T. Tang, J.-A.-Q. Zhou, Y.-M. Tu, H.-Z. Fan, M.-Y. Wang, Q. Cao, Z. Cai and J.-B. Zhu, Angew. Chem., 2025, 137, e202505310.
5. C. M. Rohles, L. Gläser, M. Kohlstedt, G. Gießelmann, S. Pearson, A. del Campo, J. Becker and C. Wittmann, Green Chem., 2018, 20, 4662–4674.
6. T. Han, G. B. Kim and S. Y. Lee, Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 30328–30334.
7. S. J. Park, E. Y. Kim, W. Noh, H. M. Park, Y. H. Oh, S. H. Lee and S. Lee, Metab. Eng., 2013, 16, 42–47.
8. C. M. Rohles, G. Gießelmann, M. Kohlstedt, C. Wittmann and J. Becker, Microb. Cell Fact., 2016, 15, 154.
9. M. Zhang, C. Gao, X. Guo, S. Guo, Z. Kang, D. Xiao, J. Yan, F. Tao, W. Zhang, W. Dong, P. Liu, C. Yang, C. Ma and P. Xu, Nat. Commun., 2018, 9, 2114.
10. W. Li, L. Ma, X. Shen, J. Wang, Q. Feng, L. Liu, G. Zheng, Y. Yan, X. Sun and Q. Yuan, Nat. Commun., 2019, 10, 3337.
11. H. T. Kim, K.-A. Baritugo, Y. H. Oh, S. M. Hyun, T. U. Khang, K. H. Kang, S. H. Jung, B. K. Song, K. Park, I.-K. Kim, M. O. Lee, Y. Kam, Y. T. Hwang, S. J. Park and J. C. Joo, ACS Sustainable Chem. Eng., 2018, 6, 5296–5305.
12. T. Mimitsuka, H. Sawai, M. Hatsu and K. Yamada, Biosci. Biotechnol. Biochem., 2007, 71, 2130–2135.
13. Z.-G. Qian, X.-X. Xia and S. Y. Lee, Biotechnol. Bioeng., 2011, 108, 93–103.
14. L. Zhou, K. Cheng, T. Liu, N. Li, H. Zhang and Y. He, Eur. Polym. J., 2022, 180, 111618.
15. E. Navarro, L. Franco, J. A. Subirana and J. Puiggali, Macromolecules, 1995, 28, 8742–8750.
16. E. E. Drufva, J. F. Cahill, P. M. B. Saint-Vincent, A. N. Williams, V. Bocharova, N. Capra, F. Meilleur, D. L. Carper, C. Bourgery, K. Miyazaki, M. Yonemura, Y. Shiraishi, J. M. Parks, M. Zhou, I. T. Dishner, J. C. Foster, S. J. Koehler, H. R. Valentino, A. Sedova, V. Kertesz, D. P. Vasileva, L. H. Hochanadel, C. A. Figg, S. Negoro, D.-I. Kato, S. H. Chen and J. K. Michener, Chem Catal., 2025, 101418.
17. I. Coin, M. Beyermann and M. Bienert, Nat. Protoc., 2007, 2, 3247–3256.
18. S. C. Mokwatlo, B. C. Klein, P. T. Benavides, E. C. D. Tan, C. M. Kneucker, C. Ling, C. A. Singer, R. Lyons, V. Sànchez i Nogué, K. V. Hestmark, M. A. Ingraham, K. J. Ramirez, C. W. Johnson, G. T. Beckham and D. Salvachúa, Green Chem., 2024, 26, 10152–10167.
19. T. Katoh, T. Sengoku, K. Hirata, K. Ogata and H. Suga, Nat. Chem., 2020, 12, 1081–1088.
20. L. Wang, A. Brock, B. Herberich and P. G. Schultz, Science, 2001, 292, 498–500.
21. R. Fricke, C. V. Swenson, L. T. Roe, N. X. Hamlish, B. Shah, Z. Zhang, E. Ficaretta, O. Ad, S. Smaga, C. L. Gee, A. Chatterjee and A. Schepartz, Nat. Chem., 2023, 15, 960–971.
22. Z. L. Watson, I. J. Knudson, F. R. Ward, S. J. Miller, J. H. D. Cate, A. Schepartz and A. M. Abramyan, Nat. Chem., 2023, 15, 913–921.
23. J. Lee, K. J. Schwarz, D. S. Kim, J. S. Moore and M. C. Jewett, Nat. Commun., 2020, 11, 4304.
24. M. Spinck, C. Piedrafita, W. E. Robertson, T. S. Elliott, D. Cervettini, D. de la Torre and J. W. Chin, Nat. Chem., 2023, 15, 61–69.
25. D. L. Dunkelmann, C. Piedrafita, A. Dickson, K. C. Liu, T. S. Elliott, M. Fiedler, D. Bellini, A. Zhou, D. Cervettini and J. W. Chin, Nature, 2024, 625, 603–610.
26. I. B. Folger, N. F. Frota, A. Pistofidis, D. L. Niquille, D. A. Hansen, T. M. Schmeing and D. Hilvert, Nat. Chem. Biol., 2024, 20, 761–769.
27. K. A. J. Bozhüyük, L. Präve, C. Kegler, L. Schenk, S. Kaiser, C. Schelhas, Y.-N. Shi, W. Kuttenlochner, M. Schreiber, J. Kandler, M. Alanjary, T. M. Mohiuddin, M. Groll, G. K. A. Hochberg and H. B. Bode, Science, 2024, 383, eadg4320.
28. K. A. J. Bozhüyük, F. Fleischhacker, A. Linck, F. Wesche, A. Tietze, C.-P. Niesert and H. B. Bode, Nat. Chem., 2018, 10, 275–281.
29. M. Damian, V. Tseliou, P. Peters, T. Knaus and F. G. Mutti, Angew. Chem., 2026, 138, e202515469.
30. M. Lubberink, W. Finnigan and S. L. Flitsch, Green Chem., 2023, 25, 2958–2970.
31. M. Winn, S. M. Richardson, D. J. Campopiano and J. Micklefield, Curr. Opin. Chem. Biol., 2020, 55, 77–85.
32. G. M. Landwehr, J. W. Bogart, C. Magalhaes, E. G. Hammarlund, A. S. Karim and M. C. Jewett, Nat. Commun., 2025, 16, 865.
33. M. Petchey, A. Cuetos, B. Rowlinson, S. Dannevald, A. Frese, P. W. Sutton, S. Lovelock, R. C. Lloyd, I. J. S. Fairlamb and G. Grogan, Angew. Chem., Int. Ed., 2018, 57, 11584–11588.
34. M. R. Petchey, B. Rowlinson, R. C. Lloyd, I. J. S. Fairlamb and G. Grogan, ACS Catal., 2020, 10, 4659–4663.
35. Q. Tang, M. Petchey, B. Rowlinson, T. J. Burden, I. J. S. Fairlamb and G. Grogan, ACS Catal., 2024, 14, 1021–1029.
36. M. A. Hollenhorst, J. Clardy and C. T. Walsh, Biochemistry, 2009, 48, 10467–10472.
37. C. Ji, Q. Chen, Q. Li, H. Huang, Y. Song, J. Ma and J. Ju, Tetrahedron Lett., 2014, 55, 4901–4904.
38. M. Zhu, L. Wang and J. He, Angew. Chem., Int. Ed., 2021, 60, 2030–2035.
39. R. Madhavachary, E. M. M. Abdelraheem, A. Rossetti, A. Twarda-Clapa, B. Musielak, K. Kurpiewska, J. Kalinowska-Tłuścik, T. A. Holak and A. Dömling, Angew. Chem., Int. Ed., 2017, 56, 10725–10729.
40. G. Odian, Principles of polymerization, John Wiley & Sons, Nashville, TN, 4th edn, 2004.
41. H. Li, Y. Shang, W. Huang, B. Xue, X. Zhang, Z. Cui, P. Fu, X. Pang, Q. Zhao and M. Liu, J. Appl. Polym. Sci., 2021, 138, 51017.
42. F. Quitterer, A. List, P. Beck, A. Bacher and M. Groll, J. Mol. Biol., 2012, 424, 270–282.
43. I. Sharon, S. Pinus, M. Grogg, N. Moitessier, D. Hilvert and T. M. Schmeing, Nat. Commun., 2022, 13, 3923.
44. M. A. Gaston, L. Zhang, K. B. Green-Church and J. A. Krzycki, Nature, 2011, 471, 647–650.
45. M. R. Gopal, R. M. Dickey, N. D. Butler, M. R. Talley, D. T. Nakamura, A. Mohapatra, M. P. Watson, W. Chen and A. M. Kunjapur, ACS Catal., 2023, 13, 4778–4789.
46. M. Tavanti, J. Hosford, R. C. Lloyd and M. J. B. Brown, Green Chem., 2021, 23, 828–837.
47. K. Y. Hara and A. Kondo, Microb. Cell Fact., 2015, 14, 198.
48. M. Zhao, D. Huang, X. Zhang, M. A. G. Koffas, J. Zhou and Y. Deng, Metab. Eng., 2018, 47, 254–262.
49. N. Qin, F. Zhu, Y. Liu, D. Liu and Z. Chen, ACS Synth. Biol., 2025, 14, 598–608.
50. J. F. Cahill, V. Kertesz, P. Saint-Vincent, H. Valentino, E. Drufva, N. Thiele and J. K. Michener, J. Am. Soc. Mass Spectrom., 2023, 34, 1383–1391.
51. J. Wohlwend, G. Corso, S. Passaro, N. Getz, M. Reveiz, K. Leidal, W. Swiderski, L. Atkinson, T. Portnoi, I. Chinn, J. Silterra, T. Jaakkola and R. Barzilay, bioRxivorg, 2025 preprint,  DOI:10.1101/2024.11.19.624167.
52. S. Passaro, G. Corso, J. Wohlwend, M. Reveiz, S. Thaler, V. R. Somnath, N. Getz, T. Portnoi, J. Roy, H. Stark, D. Kwabi-Addo, D. Beaini, T. Jaakkola and R. Barzilay, bioRxivorg, 2025 preprint,  DOI:10.1101/2025.06.14.659707.
53. P. Schake, S. N. Bolz, K. Linnemann and M. Schroeder, Nucleic Acids Res., 2025, 53, W463–W465.
54. N. A. Rorrer, S. F. Notonier, B. C. Knott, B. A. Black, A. Singh, S. R. Nicholson, C. P. Kinchin, G. P. Schmidt, A. C. Carpenter, K. J. Ramirez, C. W. Johnson, D. Salvachúa, M. F. Crowley and G. T. Beckham, Cell Rep. Phys. Sci., 2022, 3, 100840.
55. L. Wang, M. Zhu, Q. Zhang, X. Zhang, P. Yang, Z. Liu, Y. Deng, Y. Zhu, X. Huang, L. Han, S. Li and J. He, ACS Chem. Biol., 2017, 12, 3067–3075.
56. M. Mirdita, K. Schütze, Y. Moriwaki, L. Heo, S. Ovchinnikov and M. Steinegger, Nat. Methods, 2022, 19, 679–682.

---

Footnote

† These authors contributed equally.

---