Enzymatic peptide synthesis can draw on five enzyme families: non-ribosomal peptide synthetases, ATP-grasp ligases, α-amino acid ester acyltransferases, β-lactam acylases, and cyanophycinases. They differ in activation chemistry, ATP requirements, tolerance of unprotected amino acids, and products.…
Amide bond formation is the central chemical step in peptide synthesis, and five enzyme families can catalyze it: non-ribosomal peptide synthetases NRPSs , ATP-grasp enzymes , α-amino acid ester acyltransferases AETs , β-lactam acylases , and cyanophycinases . The families differ in how they activate the carboxyl group, whether they consume ATP, whether they accept unprotected amino acids, and what products they deliver. The evidence behind them is uneven: some routes are industrially mature, one is structurally uncharacterized, and none has been compared with the others in a controlled, quantitative study.
The underlying chemistry is the same across the five. A carboxylate is a poor electrophile, so the acyl group must be activated before an amine can attack it. Two strategies dominate. One converts the carboxylate into a mixed anhydride: an acylphosphate, as in ATP-grasp enzymes, or an aminoacyl-adenosine monophosphate, as in tRNA-dependent ligases. The other uses an ester or amide as a preformed acyl donor that a transferase or acylase moves to a nucleophile. A 2019 review by Wang and colleagues in Biomolecules 9 11 :733 covers the biosynthesis of short oligopeptides across all five families, and this article follows that survey's structure while separating established results from open questions.
NRPSs are large, modular enzymes found throughout bacteria and fungi. Their products include some of the most valuable peptide structures in medicine: penicillin, bleomycin, and cyclosporine are all NRPS products, and the range of ring sizes, modified residues, and bioactivities in this class reflects the flexibility of modular assembly. The representative enzyme shown in the review, PDB ID 2VSQ, displays the four canonical domains: adenylation A , condensation C , thioesterase Te , and peptidyl carrier protein PCP .
The review divides NRPS-type biocatalysts into ATP-dependent and ATP-independent classes by their mode of substrate activation. ATP-dependent enzymes activate amino acids as aminoacyl-adenosine monophosphate, the strategy used by tRNA-dependent ligases. ATP-independent enzymes form an aminoacyl phosphate through a transacylase. The distinction is practical as well as biochemical, because an ATP-dependent route carries a cofactor cost at production scale and an ATP-independent route does not.
What is well supported: NRPSs are widespread biosynthetic enzymes in bacteria and fungi, their products are therapeutically significant, and their domain architecture is modular. What is less supported: the review describes amide bond formation with figures rather than written catalytic detail, and the precise division of labor among the A, C, Te, and PCP domains during condensation remains an open question. For a researcher considering this family, the practical reality is that repurposing NRPSs is slow. Modules are large, non-native substrates are frequently rejected, and engineering campaigns move one iteration at a time.
A further practical point follows from the size of these enzymes. A minimal NRPS that assembles a two-residue product already requires a large multi-domain protein, and the full assembly lines for products such as cyclosporine are massive. For short oligopeptides, an NRPS is therefore a heavyweight option: its strength is the construction of complex, often modified structures that ribosomal synthesis cannot reach, not the economical production of simple dipeptides.
ATP-grasp enzymes are ATP-dependent carboxylate-amine ligases. They phosphorylate a carboxylic acid with ATP to form an acylphosphate intermediate, and the amine substrate then displaces the phosphate to form the amide bond. The family includes biotin carboxylase, D-alanine-D-alanine ligase Ddl , which assembles the D-Ala-D-Ala dipeptide, and glutathione synthetase. Its members also operate in de novo purine biosynthesis: glycinamide ribonucleotide synthetase PDB ID 2IP4 is the structural example presented in the review.
The family is structurally distinctive. Members share three conserved domains and an atypical ATP-binding fold that encloses the ATP molecule, and most require an Mg2+ ion coordinated by ATP in the active site during peptide bond formation. The fold is unrelated to standard kinase architectures, which is why ATP-grasp ligases count as a separate solution to carboxylate activation.
For preparative chemistry, the family offers ATP-dependent amide bond formation from unprotected amino acids, an attractive route to short products such as D-Ala-D-Ala or glutathione. The constraints are equally clear. Every bond consumes ATP, and the ligases are usually tuned to their natural substrates. The review documents the mechanism and structure in detail but provides no yield data, substrate scope tables, or scale-up experience for ATP-grasp enzymes as general peptide synthesis tools. Knowing how these enzymes work is not the same as knowing how to run them.
AETs matter because they can build oligopeptides from unprotected amino acids with high reported yield. The efficient method credited to Kenzo and colleagues used Empedobacter brevis ATCC 14234 and an enzyme named carboxypeptidase Y. The enzyme transfers the acyl group of an amino acid ester to an amino acid nucleophile, so neither partner requires protecting groups. That property, if general, would remove two of the main costs of chemical peptide synthesis: protection and deprotection.
The enzymes were first cloned and expressed by Isao Abe and colleagues from two bacterial strains. The two deduced amino acid sequences are 35% and 36% identical to the α-amino acid ester hydrolase from Acetobacter pasteurianus, which places them in a recognizable enzyme family while leaving their ancestry incomplete. The same protein type resurfaces in the β-lactam acylase group, since α-amino acid ester hydrolase is studied in that context as well.
The established facts are dual dipeptidyl peptidase and transferase activity, and high specificity for both acyl donors and nucleophiles. Specificity of that degree produces clean products and narrow substrate tolerance at the same time. What is missing matters more. The early report did not provide the amino acid sequence, the coding gene sequence, or a crystal structure, and no AET structure or reaction mechanism has been reported since. A researcher evaluating this route is working from activity data alone: the reaction works, but the molecular basis of donor and nucleophile recognition, the rate-limiting steps, and the prospects for broadening specificity are unknown.
β-Lactam acylases are the industrially mature members of this list. Three types dominate the literature: penicillin acylase PA , glutaryl acylase GA , and β-amino acid ester hydrolase AEH . Their natural substrates are β-lactam antibiotics, and their catalytic task is to hydrolyze an acyl side chain or transfer it to a new amine.
Penicillin acylase is the industrial anchor. It is used to produce 6-aminopenicillanic acid 6-APA from penicillin, and 6-APA is the active pharmaceutical intermediate from which semi-synthetic penicillins are made. The same enzyme can be run in synthetic mode, transferring an acyl side chain onto 6-APA to make a semi-synthetic penicillin. Beyond that product, PA has been applied to peptide synthesis, resolution of racemic mixtures, and production of chiral and achiral pharmaceutical intermediates. Glutaryl acylase handles cephalosporin processing, and AEH acts on β-amino acid esters.
The case for enzymatic β-lactam synthesis rests on process economics and environmental load. An acylase-catalyzed reaction avoids the stoichiometric reagents, low temperatures, and waste streams of chemical acylation. That claim is common in the biocatalysis literature and chemically reasonable, but the review supplies no quantitative yield or cost comparison with chemical routes, and it does not rank PA, GA, and AEH for specific semi-synthetic targets. PA is the proven choice for penicillin-class products. GA and AEH occupy narrower niches whose boundaries the published record does not sharply define.
Cyanophycin granule polypeptide CGP is an intracellular storage polymer found in most cyanobacteria. It consists of equimolar arginine and aspartic acid: each arginine is attached through its α-amino group to the β-carboxyl group of an aspartic acid residue. The 1:1 Asp:Arg stoichiometry is the polymer's analytical signature, and the β-linkage means the polymer is built without the ribosome and is not degraded by ordinary proteases, which is why dedicated cyanophycinases handle its recycling.
Synthesis is catalyzed by cyanophycin synthetase, the product of the cphA gene, a template-independent ligase rather than an NRPS. Degradation is carried out by cyanophycinases : CphB acts intracellularly, CphE acts extracellularly, and both release the dipeptide β-Asp-Arg. The biotechnological strategy combines the two activities in a recombinant host, by co-expressing CGP with a CGPase or by making the polymer and degrading it in a second step. Recombinant CGP production has been established in four hosts, summarized below.
| Host organism | Expression strategy | Reported result |
|---|---|---|
| Escherichia coli | Recombinant CGP production | CGP produced |
| Pseudomonas putida | Recombinant CGP production | CGP produced |
| Pseudomonas alcaligenes DIP1 | Recombinant CGP production | CGP produced |
| Nicotiana tabacum | CGP production alone; co-expression of CGP and CGPase | CGP stored; β-Asp-Arg dipeptides produced for storage and transport |
The tobacco result is the most complete: co-expression of the synthetase and the cyanophycinase produced a system that stores arginine and releases β-Asp-Arg dipeptides, tying synthesis to transport. Because the polymer's 1:1 stoichiometry carries over into the released dipeptide, the route delivers arginine in a defined, equimolar form. What is not established is scale. All four hosts have produced CGP or dipeptides under laboratory conditions, but the review reports no fermentation data, no volumetric productivity, and no cost analysis. Whether the CGP/CGPase route can be scaled to commercial supply of β-Asp-Arg remains an open question.
The practical question is which family a researcher or buyer should use. The answer starts with the product, not the enzyme: the five routes are not interchangeable, and each family's strongest claim to attention follows from what it can make. The table below condenses the evidence for each.
| Enzyme family | Activation strategy | Representative products | Main gap in evidence |
|---|---|---|---|
| NRPS | Aminoacyl-adenylate or aminoacyl phosphate | Penicillin, bleomycin, cyclosporine | Domain-level mechanism; engineering reliability |
| ATP-grasp | Acylphosphate from ATP | D-Ala-D-Ala, glutathione, purine precursors | Preparative scope and measured yields |
| AET | Acyl transfer from an amino acid ester | Short oligopeptides from unprotected amino acids | No structure or reaction mechanism |
| β-Lactam acylase | Acyl transfer from a β-lactam amide or ester | 6-APA, semi-synthetic penicillins | Which acylase for which product |
| Cyanophycinase | Hydrolysis of the CGP polymer | β-Asp-Arg dipeptides | Commercial scalability |
The table condenses what the literature actually supports. NRPSs and ATP-grasp enzymes offer ATP-dependent strategies for forming amide bonds from amino acids. AETs and β-lactam acylases work from pre-activated donors and avoid ATP. Cyanophycinases are hydrolases that generate a defined dipeptide from a polymer rather than building one bond at a time. Matching a family to a product is straightforward when the product defines the route. 6-APA can realistically come only from penicillin acylase. β-Asp-Arg is available only through the cyanophycin system. For a general protected-group-free short peptide, AETs are the candidate, but the missing structure and mechanism make them a research bet rather than a process decision. Only penicillin acylase has a documented role in racemic resolution among the five families.
Process design follows the same logic. ATP-grasp and NRPS routes must budget for the cost of ATP and its regeneration. AET and β-lactam acylase routes need a stoichiometric acyl donor, such as an amino acid ester or a β-lactam, and must separate the product from the released leaving group. Cyanophycin routes avoid both problems by hydrolyzing a polymer, but they deliver only the dipeptide encoded by the polymer structure. None of the families is described in the review with the kinetic parameters, space-time yields, or isolation procedures that a process chemist would need to choose among them.
The limits of the evidence matter here. Neither the review nor the underlying reports provide quantitative yield or cost comparisons among the five families, and benchmarks against chemical peptide synthesis are absent as well. The environmental claims attached to enzymatic routes are asserted rather than measured. The early AET method included no sequence or structural data, and none has appeared since. The mechanistic descriptions should be weighted accordingly. The industrial record of penicillin acylase is strong, the modular logic of NRPSs is well established if hard to engineer, and the remaining families are better understood as active research areas than as turnkey production platforms.
Five questions define the near-term agenda. The most pressing is the structure and catalytic mechanism of AETs, because no structure exists and the mechanism is unknown. Whether AET substrate specificity can be broadened for industrial oligopeptide synthesis depends on that structural work. The scalability of the CGP/CGPase co-expression system for β-Asp-Arg production has not been demonstrated at commercial scale. Which β-lactam acylase is most effective for a given semi-synthetic β-lactam antibiotic remains unresolved in the comparative sense, despite the maturity of penicillin acylase itself. And the precise catalytic roles of the NRPS A, C, Te, and PCP domains in amide bond formation are more consensus than demonstrated, since the review describes them with figures rather than written mechanism.
Each of these questions has a concrete experimental route. A crystal structure of an AET with an acyl donor bound would settle the specificity question. A fed-batch comparison of the tobacco and Pseudomonas systems would test CGP scalability. A single semi-synthetic penicillin made with PA, GA, and AEH in parallel would rank the acylases. Domain deletion or single-turnover experiments on NRPS modules would assign the catalytic steps. None of these experiments appears in the review.
For researchers and buyers, the actionable conclusion is simpler than the research agenda. Choose penicillin acylase when the target is a penicillin-class product. Choose the cyanophycin system when the target is β-Asp-Arg. For everything else, treat the families as options with distinct activation chemistry, and plan to generate the missing data yourself.
Peptides referenced: Glutathione.
Related reading: Reversible Double Linkers Reduce Amyloid Peptide Aggregation, Peptide Antigen Design: Key Parameters and Practical Guidelines, How Flexible Cyclic Peptides Enter and Cross Cell Membranes, Condensation Agents in SPPS: How to Choose the Right One.