Decarboxylation in Natural Products Biosynthesis

Abstract

Decarboxylation reactions are frequently found in the biosynthesis of primary and secondary metabolites. Decarboxylase enzymes responsible for these transformations operate via diverse mechanisms and act on a large variety of substrates, making them appealing in terms of biotechnological applications. This Perspective focuses on the occurrence of decarboxylation reactions in natural product biosynthesis and provides a perspective on their applications in biocatalysis for fine chemicals and pharmaceuticals.

1. Introduction

Decarboxylation is a fundamental biochemical process involving the removal of a carboxylic acid, resulting in the release of carbon dioxide. It plays a critical role in various metabolic pathways, including the decarboxylation of α- and β-keto acids and amino acids; it is central to the biosynthesis of alkaloids, cholesterol, heme, and the catabolism of aromatic compounds.¹⁻⁴ While decarboxylation is thermodynamically favorable, the uncatalyzed reaction kinetics are slow, leading to the evolution of numerous enzymes that catalyze these reactions on diverse substrates. Given the ubiquity of decarboxylation reactions in primary metabolism, it is unsurprising that nature has repurposed many of these enzyme classes for the biosynthesis of natural products, which we review herein (Table 1).

Table 1. Summary of Types of Cofactor-Dependent and Independent Decarboxylases in Natural Product Biosynthesis.

Cofactor	Substrate	Example enzyme	Example compound
TPP	α-keto acid	Phosphonopyruvate decarboxylase	Fosfomycin
PLP	α-amino acid	Ornithine decarboxylase	Loline
	β-amino acid	Aspartate aminotransferase	Incednine
prFMN	α, β-unsaturated acid	prFMN-dependent decarboxylase	9-Methylstreptimidone
NAD	UDP-glucuronic acid	UDP-glucuronic acid decarboxylase	Everninomicin
	β-hydroxy acid	Short-chain dehydrogenase/reductase	Micrococcin
FAD	β-hydroxyphenyl acid	Vanillyl alcohol phenyl oxidase/p-cresol methyl hydroxylase	Chondrochlorens
	β-sulfhydryl acid	Homo-oligomeric flavin-containing Cys decarboxylase	Epidermin
	7-carboxy-K252c	RebC	Rebeccamycin
rSAM	Peptide C-terminal acid	rSAM-SPASM	Mycofactocin
heme	Carboxythiazoline	P450 decarboxylase	Bottromycin A2
	Fatty acid	P450 peroxygenase	Olefin
mononuclear nonheme iron	Isocyanopropanic acid	Fe(II)/α-KG decarboxylase	Ambiguine
	3-((carboxymethyl)amino)butanoic acid	Fe(II)/α-KG decarboxylase	SF2768
multinuclear nonheme iron	Carboxy thiazoline	Nonheme diiron oxygenase/decarboxylase	Barbamide
	Fatty acid	Nonheme diiron oxygenase/decarboxylase	Olefins
	Peptide C-terminal Asp	Multinuclear nonheme iron-dependent oxidative enzyme	Aminopyruvatides
FAD, hυ	Fatty acid	Fatty acid photodecarboxylase	Alkane
cofactor independent	Malonyl-ACP	Ketosynthase-like domain KSQ	FD-891
	Methylmalonyl-CoA	Bifunctional acyltransferase/decarboxylase	Curacins
	Carboxylated β-branch-ACP	ECH2 domain	Bacillaene

Among the biosynthetic pathways described here, decarboxylases can be grouped into heterolytic and homolytic C-CO₂® bond-breaking groups.⁴ In heterolytic decarboxylation, a carbanion is formed at the α-carbon center where the C-CO₂® bond is broken. Stabilization of the carbanion requires an electron sink, often in the form of a cofactor like thiamine pyrophosphate (TPP),⁵ pyridoxal 5′-phosphate (PLP),⁶ or functional groups on the substrate that can stabilize the carbanion through conjugation such as ketone, thial, or p-quinone methide. While most decarboxylases follow a heterolytic route, some enzymes break the C-CO₂® bond homolytically, resulting in radical intermediates during decarboxylation.

Numerous reviews have been published to cover the diverse mechanisms of decarboxylation.^4,7 There are also reviews that focus on specific subsets of decarboxylases based on their cofactors, such as TPP-dependent decarboxylases,^8,9 PLP-dependent decarboxylases,^10,11 P450 fatty acid decarboxylases,¹² and prFMN-dependent decarboxylases.¹³ Additionally, there are reviews that delve into the specific mechanisms of decarboxylation, including oxidative¹⁴ and nonoxidative decarboxylation.¹⁵

This Perspective focuses on decarboxylases involved in the biosynthesis of secondary metabolites. These enzymes often exhibit relaxed substrate tolerance, making these enzymes promising functional starting points to evolve for activity on novel substrates. The biosynthesis of natural products provides a diverse set of substrates with chemical features that may be absent from primary metabolism, allowing for the study of novel and unexpected enzyme chemistry. In this Perspective, we explore decarboxylases based on their mechanism and cofactor, providing specific examples of each reaction type within the context of natural product biosynthesis. Our aim is to shed light on the interplay between decarboxylases and secondary metabolite biosynthesis and highlight the impact of decarboxylation on the pharmacological properties of bioactive natural products.

2. TPP-Dependent Decarboxylases

TPP-dependent decarboxylases are enzymes within the broader group of TPP-dependent enzymes. These enzymes play a critical role in catalyzing a wide range of reactions, including the creation and breakdown of bonds adjacent to a carbonyl group. TPP-dependent decarboxylases are found in three specific superfamilies: pyruvate decarboxylase-like enzymes, sulfopyruvate decarboxylases, and phosphonopyruvate decarboxylases.⁹ Despite the difference in domain organization in these decarboxylases, a shared mechanism is observed, beginning with the formation of a ylide or C2-carbanion-TPP intermediate. This process involves the proton abstraction of the imino tautomer form of TPP by a conserved glutamic acid residue located at the active site (Figure 1A).⁸ The ylide acts as a nucleophile, adding to the α-carbonyl of an α-keto acid substrate. Decarboxylation occurs, yielding a carbanion intermediate, which undergoes a rearrangement to form an enamine intermediate with the negative charge delocalized into the thiazolium nitrogen of TPP. Finally, the enamine is protonated, leading to the release of the product as an aldehyde (Figure 1B).

Figure 1.

(A) Formation and stabilization of the ylide species. (B) General catalytic mechanism of TPP-dependent decarboxylases. Due to significant variations in domain organization among various TPP-dependent decarboxylase superfamilies, the specific placement of the catalytic Glu residue is not illustrated, and general bases are utilized in the mechanism. For a more extensive understanding of the structural and catalytic features of TPP-dependent decarboxylases, we suggest referring to comprehensive review articles.^5,8

2.1. Pyruvate Decarboxylase-like Enzymes in the Biosynthesis of Obafluorin

The pyruvate decarboxylase-like enzyme (EC 4.1.1.1) superfamily encompasses multiple subfamilies with diverse catalytic mechanisms, including decarboxylation, carboligation, and oxidation. Examples of decarboxylases within this superfamily include pyruvate, indolepyruvate, phenylpyruvate, branched chain 2-keto acid, and benzoylformate decarboxylases. These enzymes are involved in the decarboxylation of various α-keto acid substrates found in primary metabolic pathways such as yeast fermentation,¹⁶ amino acid metabolism,¹⁷ and the mandelate pathway.¹⁸ TPP-dependent decarboxylases are also involved in the production of secondary metabolites; for example, the biosynthesis of the β-lactone antibiotic obafluorin in Pseudomonas fluorescens employs the TPP-dependent decarboxylase, ObaH, to generate the key intermediate 4-nitrophenyl acetaldehyde (Figure 2A).¹⁹

Figure 2.

Biosynthetic pathway of (A) obafluorin, (B) coenzyme M, and (C) bialaphos with the decarboxylation step catalyzed by TPP-dependent enzymes highlighted.

Extensive studies have been conducted on enzymes within the superfamily of pyruvate decarboxylase-like enzymes, particularly those involved in primary metabolism, to better understand and alter substrate specificity. However, attempts to modify substrate specificity through rational design approaches, such as residue swapping, have yielded limited success.^5,8 Further research is needed to uncover the mechanisms that determine how TPP-dependent decarboxylases recognize and interact with their substrates.

2.2. Sulfopyruvate Decarboxylase in Coenzyme M Biosynthesis

Sulfopyruvate decarboxylase (SPDC) (EC 4.1.1.79) is part of the biosynthetic pathway for mercaptoethane sulfonate (or coenzyme M), an essential coenzyme involved in methane formation in methanoarchaea^20,21 and the metabolism of aliphatic alkenes in Syntrophoarchaeum species.²² SPDC, found in the methanogenic pathway, catalyzes the decarboxylation of 3-sulfopyruvate to produce 2-sulfoacetaldehyde. The pyrophosphate binding domain and the pyrimidine binding domain of most SPDC enzymes are encoded by separate genes, comD and comE. ComDE has a limited substrate scope and cannot utilize phosphonopyruvate or pyruvate as substrates (Figure 2B).²¹

2.3. Phosphonopyruvate Decarboxylase in the Biosynthesis of Phosphorus-Containing NPs

Phosphonopyruvate decarboxylase (PPDC) (EC 4.1.1.82) catalyzes the transformation of 3-phosphonopyruvate to 2-phosphonoacetaldehyde, a precursor for several phosphorus-containing natural products (Figure 2C).²³ The decarboxylation step is essential to the formation of carbon-phosphorus (C-P) bonds and acts as a driving force to shift the equilibrium between phosphoenolpyruvate and 3-phosphonopyruvate.²⁴ PPDC is involved in the biosynthesis of notable phosphorus-containing natural products, including the herbicide bialaphos,²⁵ the antibiotic fosfomycin (BGC0000938),²⁶ and the 2-aminoethylphosphonate unit of polysaccharide B, a component of the capsular polysaccharide complex of Bacteroides fragilis.²³ In contrast to SPDC, PPDC can accept both sulfopyruvate and pyruvate as substrates but exhibits a significantly higher turnover rate with phosphonopyruvate, suggesting the phosphono group of phosphonopyruvate serves as the binding site.²³

3. PLP-Dependent Decarboxylases

PLP-dependent decarboxylases are part of the larger family of pyridoxal 5′-phosphate (PLP) dependent enzymes, which catalyze diverse reactions such as decarboxylation, racemization, transamination, and condensation of amine and amino acid substrates.²⁷ While most reactions involve α-decarboxylation, examples of β-decarboxylation have been documented, including the notable example of IdnL3 in the incednine biosynthetic pathway,²⁸ discussed below.

The chemical mechanism of α-decarboxylation catalyzed by PLP-dependent enzymes is well established (Figure 3).¹⁰ Initially, PLP forms an internal aldimine by the formation of a Schiff-base with the α-amino group of a catalytic lysine residue within the protein. The lysine residue is subsequently replaced by the substrate amino group, forming an “external aldamine” Schiff-base intermediate, followed by decarboxylation. PLP functions as an electron sink, stabilizing the carbanions generated during decarboxylation by directing the negative charge into the π bonds system, resulting in the formation of a quinonoid intermediate. Protonation at the Cα position is facilitated by a general acid, leading to the formation of an imine product. Transaldimination with the catalytic lysine residue follows, releasing the corresponding amine product and regenerating the internal aldimine.

Figure 3.

Mechanism of α-decarboxylation by PLP-dependent decarboxylases. Note that the specific active site residue involved in protonating the quinonoid intermediate may differ between enzymes and is therefore represented generically.^6,29

Based on structural and sequence alignment, PLP-dependent enzymes are classified into 7-fold types.¹⁰ Structurally, PLP-dependent decarboxylases are classified as either fold type I or III. These fold types differ in the organization of their protein domains, the catalytic lysine residue, the PLP-interacting glycine-rich loop, and their oligomer form. Based on sequence alignment, PLP-dependent decarboxylases are classified into four groups (Table 2).

Table 2. Classification of PLP-Dependent Decarboxylases (DC) (EC 4.1.1) Based on Structure, Sequence, and Function.

	Structure
	Fold Type I			Fold Type III
Sequence	DC Group I	DC Group II	DC Group III	DC Group IV
Enzymes	Glycine dehydrogenase/ decarboxylase	Glu, His, Tyr, aromatic DC	Bacterial Orn, Lys, Arg DC	Eukaryotic Orn, Lys, Arg diaminopimelate DC, RiPP DC, NRPS DC

PLP-dependent decarboxylases participate in the biosynthesis of secondary metabolites, including alkaloids and some simple amines, a ribosomally synthesized and post-translationally modified peptide (RiPP), and several nonribosomal peptides (NRPs), which we discuss in more detail in the following sections.

3.1. α-Decarboxylation by PLP-Dependent Enzymes

3.1.1. Alkaloids

Alkaloids are a diverse category of nitrogen-containing secondary metabolites for which decarboxylation is the initial biosynthetic step. PLP-dependent decarboxylases convert amino acids to amines that later condense with an aldehyde to form an iminium intermediate that undergoes a Mannich-like reaction to achieve diverse and complex alkaloid scaffolds (Figure 4A,B).³⁰ Major alkaloid subtypes, including nicotine, tropane, lycopodium, and quinolizidine, are distinguished based on their structural scaffolds. Despite the observed structural diversity, the alkaloid amines originate from two primary sources both involving decarboxylation: (1) aliphatic polyamines, such as cadaverine, putrescine, and putrescine derivatives, are derived from Orn/Arg/Lys metabolism (Figure 4C) and (2) aromatic amines, such as tryptamine, dopamine, and tyramine, originate from Trp/Tyr metabolism (Figure 4D).

Figure 4.

(A) Biosynthesis of alkaloids. (B) Structures of some alkaloids. Structures of (C) aliphatic amines and (D) aromatic amines.

In addition to their well-known involvement in the early stage of alkaloid biosynthesis for accumulating amine precursors, PLP decarboxylases also play a significant role in the later stages of biosynthetic pathways. This is exemplified by loline (Figure 5A), a broad-spectrum anti-insect alkaloid commonly found in endophytic fungi. The loline biosynthetic gene cluster (BGC0000815) comprises 11 genes, including LolD, a PLP-dependent decarboxylase.³¹ Recent experimental verification by Gao et al. demonstrated that LolD can decarboxylate pyrrolizidine α-quaternary amino acid.³² In a subsequent study, Liu et al. demonstrated the substrate versatility of LolD, enabling rapid generation of diverse enantiopure amino-izidine motifs.³³

Figure 5.

(A) Loline and (B) muraymycin D2 biosynthetic pathways.

It is worth noting that LolD is not the only type III PLP-dependent decarboxylase that can process substrates other than common amino acids. For example, the enzyme Mur23, found in the muraymycin biosynthetic pathway, can decarboxylate nucleoside-containing substrates (Figure 5B).^34,35 Additionally, type III PLP-dependent decarboxylases can also process larger substrates such as peptides and amino acids tethered to carrier proteins in RiPP and NRP biosynthesis, respectively. These findings highlight the versatility and potential of type III PLP-dependent decarboxylases as promising biocatalysts in various biosynthetic pathways and suggests that much remains to be discovered in the substrate range of this enzyme class.

3.1.2. Ribosomally Synthesized and Post-translationally Modified Peptides

Among RiPPs, an example of PLP-dependent decarboxylation can be found in microcin C biosynthesis. Microcin C is an antibiotic found in E. coli and closely related bacteria. Inside the targeted bacterial cells, microcin C is enzymatically processed to form the mature microcin C that structurally resembles Asp-AMP (Figure 6). The mechanism by which mature microcin C exerts its action is through strong inhibition of aspartyl-tRNA synthetase, thereby interfering with translation.³⁶ Microcin C biosynthesis starts with the ribosomal synthesis of the N-formylated precursor heptapeptide (MccA) fMRTGNAN, followed by modification by the adenylyltransferase MccB, which catalyzes the ATP-dependent isomerization of the C-terminal asparagine to a succinimide followed by N-phosphorylation and ring-opening to afford a P-linked carboxamide.^37,38 Subsequently, the S-adenosyl methionine (SAM) dependent enzyme MccD transfers the 3-amino-3-carboxypropyl (ACP) group from the SAM to the phosphoamidate group of the adenylated MccA. Decarboxylation of the resulting 3-amino-3-carboxypropylated MccA is catalyzed by the N-terminal domain of the MccE enzyme (MccE^NTD), a PLP-dependent decarboxylase, leading to the generation of mature microcin C.³⁹

Figure 6.

Microcin C maturation pathway and the formation of the bioactive processed Microcin C and the structure of Asp-AMP, the native substrate of the aspartyl-tRNA synthetase.

MccE is a two-domain protein; the N-terminal domain (MccE^NTD) is homologous to PLP-dependent decarboxylases group IV, while the C-terminal domain (MccE^CTD) catalyzes the inactivation of microcin C via acetylation.^40,41 The 3-aminopropyl moiety introduced by ACP-transferase MccD and MccE has been found to increase the affinity of processed McC for AspRS.⁴² Studies have demonstrated that MccE/MccD can modify a variety of adenylated peptides with different amino acid sequences and lengths, originating from mcc operons of various bacteria, including E. coli, Lactobacillus johnsonii (MccA peptide MHRIMKN), Helicobacter pylori (MccA peptide MKLSYRN), and a 20 amino acid C-terminal fragment of Synechococcussp. CC9605 (MccA peptide LQPKRLDKVAKNQLWADMMN).^39,43

3.1.3. Polyketides/Nonribosomal Peptides

PLP-dependent decarboxylases have a substrate scope that includes amino acids tethered to an acyl carrier protein (ACP), as they participate in the early biosynthetic step of several polyketides and nonribosomal peptides, including butirosins (BGC0000694), fluvirucins, and closthioamide (BGC0001891) (Figure 7).⁴⁴⁻⁴⁶ Glutamate or aspartate, initially activated by adenylation (BtrJ, CtaD, FlvN), are ligated to an ACP via their side chain carboxyl group, forming aminoacyl-ACP (BtrI, CtaE, FlvL). The α-carboxyl group of the aminoacyl-ACP is then decarboxylated by a PLP-dependent decarboxylase (BtrK, CtaF, FlvO) to produce a γ-aminobutyrate or β-alanine.⁴⁷ In the case of butirosin, γ-aminobutyrate is further modified to the 4-amino-2-hydroxybutyryl moiety in the final aminoglycoside product. The introduction of the 4-amino-2-hydroxybutyryl moiety enhances several pharmacological properties and prevents modifications such as O-phosphorylation, O-nucleotidylylation, and N-acetylation that can deactivate the aminoglycoside antibiotic activities (Figure 7A).^48,49 In the biosynthesis of closthioamide, the decarboxylases CtaF not only decarboxylate Asp-S-CtaE but also act on the extending polypeptide-ACP substrate, where two additional Asp resides are loaded by CtaD and decarboxylated in the same manner (Figure 7B).^{44−46,50−52} In polyketide-derived macrolactams like fluvirucin, β-alanine is part of the core macrolide structure (Figure 7C).⁵⁰

Figure 7.

Biosynthetic pathway of (A) butirosin A, (B) closthioamide, and (C) fluvirucin B2.

3.2. β-Decarboxylation by a PLP-Dependent Enzyme in Incednine Biosynthesis

The biosynthesis of incednine (BGC0000078), a macrolactam antibiotic from Streptomyces sp., involves β-decarboxylation catalyzed by IndL3 (Figure 8). IdnL3 catalyzes the key conversion of β-glutamate (which is formed from L-glutamate by the 2,3-aminomutase IdnL4) to (S)-3-aminobutyrate, the starting unit of polyketide synthase (PKS) assembly in the incednine biosynthetic pathway. Homologues of IdnL3 have also been identified in the BGC of several 3-aminobutyric acid-containing macrolactam polyketides including sipanmycin, salinilactam, micromono-lactam, and lobosamide A.^53,54 The substrate specificity of IdnL3 has been thoroughly examined, revealing that it selectively recognizes β-glutamate as its substrate.²⁸ The substrate recognition mechanism for IdnL3 remains unclear due to the absence of a crystal structure.

Figure 8.

Mechanism of β-decarboxylation of β-glutamate by the PLP-dependent enzyme IdnL3.

4. Prenylated FMN-Dependent (De)carboxylases

Prenylated flavin (prFMN)-dependent decarboxylases (also known as UbiD homologues; EC 4.1.1.98) rely on a modified flavin cofactor (Figure 9A) that enables challenging decarboxylations and carboxylations of a wide-range of aromatic and unsaturated acids (Figure 9B, C). Biosynthesis of prFMN is catalyzed by UbiX-like enzymes (EC 2.5.1.129) that perform alkylation and cyclization of reduced FMNH₂ to yield a reduced prFMN, forming a fourth nonaromatic ring between the N-5 and C-6 positions of the tricyclic heterocyclic isoalloxazine ring (Figure 9A).⁵⁵ To enable catalytic turnover, the reduced prFMN requires oxidative maturation to form the active prFMN^iminium species which is believed to occur within the UbiD active site. The sensitivity of certain UbiD enzymes to light and/or oxygen, such as Fdc (light sensitive), AroY (oxygen sensitive) and HmfF (both light and oxygen sensitive), has been assigned to the oxidative maturation and light-induced isomerization of the active prFMN^iminium species, although the underlying mechanism is not fully understood. Bridging this knowledge gap would facilitate the engineering of prFMN-dependent decarboxylases for broader industrial applications. The prFMN^iminium facilitates decarboxylation through either a 1,3-dipolar cycloaddition or an electrophilic aromatic substitution mechanism at the C4a-N5-C1’ positions. While genes encoding for UbiD and UbiX enzymes are often found in the same BGC, there are instances where UbiD enzymes can be supported by another UbiX-like enzyme encoded elsewhere in the host genome, such as in the case of naphthalene carboxylase found in sulfate-reducing culture N47.⁵⁶

Figure 9.

(A) Biosynthesis of prFMN. (B) Proposed 1,3-dipolar cycloaddition mechanism for the decarboxylation of cinnamic acid by Aspergillus niger Fdc (PDB ID: 4ZAB).⁵⁷ (C) Proposed electrophilic aromatic substitution mechanism for the decarboxylation of protocatechuic acid by Enterobacter cloacae P240 AroY(PDB ID: 5O3N).⁵⁸ In both Fdc and AroY, the catalytic Glu residue is part of an ionic network involving Glu-Arg-Glu. This network is critical for the activity of UbiD, and mutations at these residues result in the loss of activity. The current state of mechanistic studies of prFMN-dependent decarboxylases is recently reviewed.⁵⁹

The UbiD/UbiX system is widely present in microbe genomes, spanning archaea, bacteria, and fungi. It plays a crucial role in various metabolic pathways, contributing to the biosynthesis of primary metabolites like ubiquinone and secondary metabolites such as tautomycetin (BGC0000157) and 9-methylstreptimidone (BGC0000171).⁶⁰⁻⁶² These enzymes are also involved in the degradation of diverse (poly)aromatic hydrocarbons such as benzene, naphthalene, and phthalate.⁶³ Recent phylogenetic research has classified UbiD-like enzymes into three subgroups based on their native substrate preference: aromatic acids (UbiD or AroY), phenylacrylic acid (Fdc1), and α,β-unsaturated aliphatic acids (TtnD, SmdK).⁶⁰ The proposed decarboxylation mechanism for acrylic acid substrates and other α,β-unsaturated acids is the 1,3-dipolar cycloaddition mechanism shown in Figure 9B. In contrast, the electrophilic aromatic substitution mechanism is suggested for the decarboxylation of aryl acids (Figure 9C).

4.1. prFMN Decarboxylase in the Biosynthesis of Tautomycetin

Tautomycetin (TTN), derived from Streptomyces griseochromogenes, is a polyketide compound with a terminal alkene that exhibits antifungal activity by selectively inhibiting PP1, one of seven families of protein serine/threonine phosphatases.⁶⁴ TTN’s high selectivity for PP1 inhibition makes it particularly intriguing for studying the function and substrates of PP1 without interfering with other essential cellular phosphorylation processes.⁶⁵ Structural analysis of the PP1:TTN complex has revealed that the diene/alkene moiety of TTN plays a crucial role in accessing and interacting with the hydrophobic groove of the PP1 active site. The diene facilitates the formation of a covalent bond with the Cys127 of PP1, as demonstrated through crystallography, mass spectrometry, and mutation studies.⁶⁵ In the biosynthetic pathway of TTN, the UbiD/UbiX class TtnDC system catalyzes a key decarboxylation in the penultimate step, leading to the formation of the critical terminal alkene moiety (Figure 10A).^60,66

Figure 10.

Biosynthetic pathway of (A) tautomycetin and (B) 9-methylstreptimidone.

The polyketide backbone of TTN is constructed from five molecules of malonyl-CoA, four molecules of methylmalonyl-CoA, and one molecule of ethylmalonyl-CoA through the action of the type I polyketide synthases TtnAB. Subsequently, a dialkylmaleic anhydride moiety synthesized from α-ketoglutarate and propionate by TtnMOPR is incorporated into the polyketide backbone with the assistance of TtnK, resulting in a TTN F-1 intermediate. TtnF facilitates the dehydration of TTN F-1, leading to the generation of an α,β-unsaturated acid. This acid then undergoes decarboxylation, catalyzed by the TtnDC system.⁶⁰ The final oxidation step, which completes the biosynthesis of mature TTN, is catalyzed by TtnI. TtnD encodes a prFMN-dependent decarboxylase, while TtnC encodes for the prenyltransferase. The activity of TtnD has been thoroughly characterized biochemically and structurally.⁶⁰ TtnD exhibits substrate promiscuity, accepting a broad range of biosynthetic intermediates from TTN pathways, with the notable exception of TTN F-1, which lacks the double bond at the Cαβ-position of the acid substrate.⁶⁰

4.2. prFMN Decarboxylase in the Biosynthesis of 9-Methylstreptimidone

Another polyketide that has been characterized with a terminal alkene is 9-methylstreptimidone with a UbiD/UbiX pair encoded by the smdKJ genes (Figure 10B).^61,67 The activity of SmdK, a UbiD decarboxylase, was characterized by knockout experiments leading to the accumulation of the carboxylated congener of 9-methylstreptimidone, suggesting that decarboxylation to afford the terminal alkene is the final step in biosynthesis.⁶¹ 9-methylstreptimidone has demonstrated antifungal,⁶⁸ antiviral activity,⁶⁷ and induces apoptosis selectively in adult T-cell leukemia cells.⁶⁹ SAR studies have attributed its activity against leukemia T-cells to the hydrophobic terminal alkene moiety, underlining the importance of decarboxylation in the bioactivity of this natural product.⁶⁹

5. Oxidation Decarboxylation

The decarboxylation of otherwise unreactive molecules can be achieved by introducing electron sinks β to the carboxylate, such as ketone, thial, or p-quinone methide, which stabilize the carbanion-formed upon decarboxylation by conjugation.^70,71 In Nature, enzymes frequently couple oxidation and decarboxylation reactions to decarboxylate β-hydroxy, thiol, or phenol acid substrates, and it is often unclear whether the decarboxylation step itself is spontaneous or enzyme accelerated. The oxidation step in this process can occur through different mechanisms, utilizing a wide range of cofactors. NAD⁺ is used as a cofactor by short-chain dehydrogenase/reductases (SDR), such as the glucuronic acid decarboxylases or the micrococcin decarboxylase.⁷²⁻⁷⁶ A RiPP DC, recently described as belonging to the alcohol dehydrogenase family, is proposed to use NADP⁺.⁷⁷ Flavins, including flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN), are also used in the homo-oligomeric flavin-containing Cys decarboxylases (HFCD) family in RiPP and NRP natural products.^78,79

5.1. NAD(P)⁺-Dependent Decarboxylases

5.1.1. Glucuronic Acid Decarboxylases in the Biosynthesis of Pyranose Deoxypentose-Containing NPs

Deoxygenated carbohydrate moieties, prevalent in antibiotics, antimicrobials, and therapeutic agents, play a crucial role in imparting biological functionality and enhancing receptor affinity.^80,81 Pyranose deoxypentoses, deoxy sugars found in various natural products (Figure 11), are predominantly synthesized through glucuronic acid decarboxylases, members of the short-chain dehydrogenase/reductase (SDR) enzyme family.⁷²⁻⁷⁴ As shown in Figure 11, an example deoxypentose biosynthesis involves (1) activation of α-D-glucose-1-phosphate by α-D-glucose-1-phosphate nucleotidyltransferase (EvdS4) to give NDP-α-D-glucose, (2) oxidation of NDP-α-D-glucose by NDP-α-D-glucosedehydrogenase (EvdS1) to give NDP-α- D-glucuronic acid, (3) C-4 oxidation of NDP-α-D-glucuronic acid using NAD⁺ by glucuronic acid decarboxylase (EvdS6), facilitating C-6 decarboxylation. The intermediate formed after the decarboxylation can be either 4-keto-pentose or 4-hydroxy-pentose. Further discussion will primarily focus on the differences in product formation. It is worth noting that the decarboxylation mechanism carried out by the SDR superfamily has been extensively examined elsewhere and, thus, will not be discussed further here.⁸²⁻⁸⁴

Figure 11.

Proposed biosynthetic pathway for pyranose deoxypentose precursors of deoxygenated carbohydrate natural products: everninomicin and AT2433.

To explain the observation of 4-keto versus 4-hydroxypentose products, crystal structures of glucuronic acid decarboxylases, including ArnA (PDB ID: 2BLL),⁸⁵ hUxs1 (PDB ID: 2B69),⁸⁶ and EvdS6 (PDB ID: 8SK0),⁷³ were solved. These enzymes produce 4-keto-pentose, 4-hydroxy-pentose, and both intermediates, respectively. Despite the differences in product profile between these three enzymes, the positioning of the side chains in the SDR superfamily catalytic triad Tx_nYxxxK exhibits no significant variation. The differences in product profile can instead be attributed to the relative kinetics of the oxidative decarboxylation and reduction reactions: in enzymes that favor 4-ketopentose production the release of 4-keto-pentose and NADH occurs faster than the reduction reaction. Interestingly, the product profile can be altered by extending incubation time as the formation of 4-hydroxy-pentose is observed in the case of ArnA, which was initially thought to only produce the keto product.⁸⁷ The kinetics of these reactions are believed to be controlled by a flexible loop containing Arg279 and Arg278 residues. Mutating these residues to Ala enhances loop flexibility, improving the rate of ketone product release and resulting in a significantly lower yield of the 4-hydroxy-pentose product.⁷³ Additionally, mutations in key residues involved in the hydride transfer of the reduction, such as Thr126, which coordinates the activated ketone, can cause a near-complete loss in 4-hydroxy-pentose production. To further understand the role of conformation and proposed active site interactions in catalysis, future studies using bound UDP-glucuronic acid may be necessary.

5.1.2. Short-Chain Dehydrogenase/Reductase in Micrococcin Biosynthesis

the RiPP compound micrococcin, a potent thiopeptide targeting Mycobacterium tuberculosis, features another SDR enzyme, TclP, that facilitates oxidative decarboxylation (Figure 12).^75,76,88 TclP catalyzes the conversion of the terminal Thr residue of the substrate peptide to a ketone, that is subsequently reduced to an alcohol by another SDR enzyme, TclS, resulting in the final product, micrococcin P1. Micrococcin biosynthesis entails two stages: (1) thiazole formation through cyclodehydration and dehydration of Cys residues and (2) Ser/Thr processing involving dehydration and macrocyclization, forming the characteristic pyridine ring of a thiopeptide RiPP natural product (Figure 12B).^75,76

Figure 12.

(A) Structure of micrococcin. (B) Biosynthetic pathway of micrococcin and the modification of the C-terminal threonine by TclP and TclS.

5.1.3. Iron-Type Alcohol Dehydrogenase-like Enzymes in Daptide Biosynthesis

The newly identified RiPP class, daptides, reported in 2023, features a distinctive (S)-N,N-dimethyl-1,2-propanediamine (Dmp)-modified C-terminus, achieved through the coordinated action of four enzymes, MpaBCDM, in a three-step pathway (Figure 13).⁷⁷ MpaBC, comprising the oxidative decarboxylase MpaC and the RRE-containing adapter MpaB, catalyzes the oxidative decarboxylation of the C-terminal Thr residue to yield a C-terminal ketone. Subsequently, the C-terminal ketone undergoes conversion to an amine via the PLP-dependent transaminase MpaD and then alkylation by the SAM-dependent methyltransferase MpaM. MpaC, classified as an iron-type alcohol dehydrogenase-like enzyme, is predicted to require NADP⁺ as a cofactor based on sequence similarity and to exhibit a broad substrate scope based on three canonical substrate peptides in the BGC. Mutation of the core peptide confirms that MpaC can modify a wide range of substrate peptides with a C-terminal Thr. The structural characteristics of daptides, including a hydrophobic α-helical conformation and amidated C-terminus, resemble melittin, a membrane-disrupting component of honeybee venom. Notably, daptides demonstrate hemolytic activity against bovine erythrocytes, suggesting the potential use of the Dmp-modified C-terminus as a strategy for drug design to enhance interactions with cell membranes. The MpaBC-like systems and downstream tailoring enzymes exhibit promise for enzymatic C-terminal modification of peptides.

Figure 13.

(A) Sequence alignment of different daptide substrate peptides. The region in the leader peptide that serves as a recognition sequence is underlined. (B) Biosynthetic pathway of the (S)-N₂,N₂-dimethyl-1,2-propanediamine moiety.

5.2. Flavin-Dependent Decarboxylase

In addition to the prFMN-dependent decarboxylases mentioned above, Nature employs unmodified flavin cofactors FMN and FAD as electron acceptors in oxidative decarboxylation reactions. Two flavoenzyme families with distinct mechanisms catalyze decarboxylation in secondary metabolism: (1) vanillyl alcohol phenyl oxidase/p-cresol methyl hydroxylase (VAO/PCMH) in long-chain fatty acid enol esters and PKS/NRPS biosynthesis and (2) homo-oligomeric flavin-containing Cys decarboxylase (HFCD) in RiPP and NRPS biosynthesis.^14,89,90 Additionally, there are flavin-dependent halogenases, such as Bmp2⁹¹ and Bmp5⁹² that induce decarboxylation of substrates during halogenation processes.

5.2.1. Vanillyl Alcohol Phenyl Oxidase (VAO)/p-Cresol Methyl Hydroxylase (PCMH) Family Enzymes in Fatty Acid Enol Ester and PKS/NRPS Biosynthesis

Two exemplary flavin-dependent decarboxylation reactions can be found in the biosynthesis of the polyketide-peptide antibiotic chondrochloren as well as the long-chain fatty acid enol esters, both involving decarboxylation of an N-substituted tyrosine intermediate (Figure 14AB). The key decarboxylation step is catalyzed by flavoenzymes FeeG (long-chain fatty acid enol esters) and CndG (chondrochloren).⁹³⁻⁹⁵ Both enzymes belong to the 4-phenyl oxidizing subgroup of the VAO/PCMH flavoprotein family.¹⁴ The proposed mechanism, derived from a mechanistic study of vanillyl-alcohol oxidase by Fraaije et al.,⁹⁰ involves hydride transfer from the Cα atom of the substrate to the N5 of the isoalloxazine ring, forming a quinoid species as the first intermediate. The p-quinone methide moiety serves as an electron sink, facilitating Cβ-decarboxylation, resulting in the formation of the hydroxyl-styryl-containing product. The catalytic cycle concludes with O₂-assisted oxidation of the reduced FADH₂, regenerating the oxidized FAD (Figure 14C). A crystal structure is currently unavailable for this class of enzymes.

Figure 14.

Oxidative decarboxylation steps are proposed for the biosynthesis of (A) fatty acid enol ester and (B) chondrochloren. (C) Proposed mechanism of the oxidative decarboxylation catalyzed by vanillyl alcohol phenyl oxidase/p-cresol methyl hydroxylases.

5.2.2. Homo-oligomeric Flavin-Containing Cys Decarboxylases (HFCDs)

Another class of flavoproteins catalyzing decarboxylation is the HFCDs. Enzymes of this class are found in the BGCs of holomycin (BGC0000373, a dithiopyrroles NRPS antibiotic) and in several classes of AviCys-containing RiPP, including epidermin (BGC0000508, a lanthipeptide),⁹⁶ thioviridamide (a thioamitide),⁹⁷ microbisporicin (a lanthipeptide),⁹⁸ and cypemycin (BGC0000582, a type A linaridine) (Figure 15).^99,100 The proposed reaction mechanism of HFCDs begins with the formation of a charge-transfer complex between the thiol group of Cys residue and the oxidized flavin (Figure 15A),¹⁰¹ followed by nucleophilic addition of the Cys thiol group to C4a of the oxidized flavin. Deprotonation at Cβ of Cys residue facilitates the heterolytic cleavage of the C4a–S bond to yield reduced flavin and thioaldehyde intermediate. The thioaldehyde serves as an electron sink to stabilize the decarboxylation and yield a thioenol. This thioenol later serves to facilitate the formation of the dithiopyrrole moiety in holomycins (Figure 15B) and C-terminal S-[(Z)-2-aminovinyl]-d-cysteine (AviCys) in RiPPs (Figure 15C).

Figure 15.

(A) Mechanism of homo-oligomeric flavin-containing cysteine decarboxylases is illustrated using EpiD as an example (PDB ID: 1G63). Roles of HFCDs in the biosynthesis of (B) holomycin and (C) AviCys-containing RiPPs.

AviCys-modified peptides, which display potent antimicrobial or anticancer activity, possess an intricate C-terminal ring that is critical for desirable drug-like properties such as heat and pH stability, high target specificity, and resistance toward proteases and that locks the peptide in an active conformation for target binding. Recently, there has been a detailed review on the participation of HFCDs in the biosynthesis of AviCys-containing peptides.¹⁰⁰ The enzymes responsible for catalyzing the initial decarboxylation step in AviCys formation have been extensively studied, and successful reconstitution of several biosynthetic pathways has been achieved.^98,100,102 However, the subsequent step of thioether cyclization, which completes the AviCys formation, is not fully understood. It is currently proposed to be either an enzyme-catalyzed or a spontaneous process.¹⁰⁰ Notably, the AviCys decarboxylase CypD exhibits a broad substrate scope, suggesting its potential utility as a biocatalyst for AviCys synthesis in modified peptides.¹⁰⁰ The role of HFCDs in AviCys-containing peptide biosynthesis has recently been reviewed in detail.¹⁰⁰

5.2.3. FAD-Dependent Decarboxylase in Indolocarbazole Biosynthesis

Indolocarbazoles¹⁰³ comprise a diverse array of alkaloid natural products that exhibit multifaceted pharmaceutical activities, including the inhibition of protein kinase,¹⁰⁴ DNA replication,¹⁰³ and anticancer effects.^105,106 The diversity of indolocarbazoles partially stems from variations in the oxidation states of the pyrroline ring moiety, products of oxidative decarboxylation catalyzed by FAD-dependent enzymes. In the biosynthesis of rebeccamycin (BGC0000821) and staurosporine (BGC0000825), this key step is catalyzed by RebC and StaC, respectively. The biosynthesis of indolocarbazoles involves the formation of the chromopyrrolic acid (CPA) intermediate from two tryptophan molecules. As shown in Figure 16A, CPA undergoes aryl-aryl coupling, followed by decarboxylation catalyzed by P450 enzymes, StaP or RebP, to yield 7-carboxy-K252c. This intermediate is then decarboxylated by StaC to yield K252c or is decarboxylated and oxidized by RebC to yield arcyriaflavin A. In the absence of StaC and RebC, the product of the StaP/RebP-catalyzed transformation of CPA yields multiple products, including K252c, 7-hydroxy-K252c, and arcyriaflavin A, with 7-hydroxy-K252c being the primary product, highlighting the importance of StaC and RebC in controlling the formation of the correct indolocarbazole cores while preventing site reactions.

Figure 16.

(A) Formation of the crucial intermediate 7-carboxy-K252c from chromopyrrolic acid (CPA) through aryl–aryl coupling, catalyzed by the P450 enzyme StaP/RebP. The subsequent step involves the decarboxylation of 7-carboxy-K252c, yielding different products mediated by the FAD-dependent enzymes StaC and RebC. (B) Mechanism of decarboxylation catalyzed by StaC/RebC-10X and RebC yielding different products. Arg residue is numbered based on crystal structure of RebC-10X (PDB IB: 4EIQ).

While StaC and RebC share a 65% sequence similarity, they are distinct in FAD-binding affinity and product profile. Interestingly, introducing two mutations to RebC (F216V and R239N) allows the reaction mechanism to switch to that of StaC.^107,108 The cocrystal structures of the 7-carboxy-K252c substrate with RebC and RebC-10X (a mutated RebC with StaC-like activity) reveal that RebC-10X selectively binds the nonplanar keto tautomer, while RebC binds the planar enol tautomer of the substrate (Figure 16B).¹⁰⁹ In the keto tautomer, spontaneous decarboxylation is facilitated due to conjugation with Cβ, serving as an electron sink. In contrast, in the enol form, C-7 is at sp² hybridization and does not facilitate spontaneous decarboxylation. FAD-assisted hydroxylation at C-7 is proposed to change the hybridization of C-7 from sp² to sp³, facilitating decarboxylation.

6. Homolytic Decarboxylation

In contrast to the above examples of heterolytic (2-electron) decarboxylation, here we detail examples of homolytic decarboxylation, which leverages various natural oxidants to induce decarboxylation. Most often β-oxidation is invoked in proposed mechanisms, though in at least one case, the mechanism is proposed to proceed through a carboxyl radical.

6.1. Radical SAM-Dependent Decarboxylase in the Biosynthesis of Mycofactocin

MftC is an enzyme that belongs to the radical SAM (rSAM) enzyme superfamily, which is the largest known protein superfamily utilizing a [4Fe-4S] cluster for the reductive cleavage of SAM to form L-methionine and a strongly oxidizing 5-deoxyadenosine radical (dAdo^•).¹¹⁰ Once formed, the dAdo^• radical can participate in various transformations, including cyclization, methylation, and dehydration. Among the members of this large protein family, MftC belongs to a specific rSAM subfamily that is characterized by a 100-amino acid C-terminal domain called SPASM shared by rSAM enzymes involved in the biosynthesis of subtilosin A, pyrroloquinoline quinone, anaerobic sulfatase, and mycofactocin.¹¹¹ While rSAM-SPASM proteins typically catalyze oxidative bond-forming reactions, MftC is an exception. Its catalytic activity involves a decarboxylation reaction and the formation of a C–C bond.¹¹² MftC uses two equivalents of SAM: one for decarboxylation and the second to forge a C-C bond between Cα of a Tyr residue and Cβ of a Val residue, both found in the C-terminus of the substrate peptide MftA.¹¹³ MftC features three [4Fe-4S] clusters: the rSAM cluster, where dAdo^• formation happens, and the two auxiliary clusters found within the SPASM domain. All these clusters are crucial for MftC activity.¹¹³ The first auxiliary cluster, AuxI, is proposed to play a role in substrate binding and electron shuttling between the active site, and the second auxiliary cluster AuxII, which serves as electron storage.¹¹⁴ It is noteworthy that the MftC reaction exclusively takes place in the presence of the peptide chaperone MftB.¹¹⁵

The proposed mechanism for the decarboxylation and ligation processes catalyzed by MftC unfolds as follows (Figure 17): MftC initiates the process by reductively cleaving SAM, generating a 5-deoxyadenosine radical (dAdo^•). In the first phase of decarboxylation, the dAdo^• extracts a hydrogen from the substrate peptide, producing an alkyl radical at Cβ-Tyr30. Subsequently, 1-electron oxidation occurs at the alkyl radical, coupled with the deprotonation of the phenolic proton, to form a quinone methide intermediate. This quinone methide serves as an electron sink, facilitating the decarboxylation process and forming a vinyl phenol-containing intermediate, which serves as the substrate for the second stage of ligation. In this stage, a second dAdo^• extracts a hydrogen from the Cβ-of the penultimate Val29. The newly formed alkyl radical then attacks the neighboring C-C unsaturated bond at the C position, resulting in the formation of a 5-membered ring. The remaining electron from the double bond has the potential to form a radical on Cβ or the phenol ring of Tyr30. This radical is subsequently quenched to yield the final product. It is important to note that the identity of the electron, hydrogen acceptor, and donor in the proposed mechanism are unknown.

Figure 17.

Proposed mechanism of the decarboxylation and ligation catalyzed by MftC consuming two equivalents of SAM to modify one equivalent of substrate peptide.

The activity of MftC relies on the presence of a hydrogen-donating group in the phenol, demonstrated by the lack of activity observed in the mutated substrate Y30F. The MftC-catalyzed decarboxylation resembles the mechanism observed in the 4-phenyl oxidizing subgroup of vanillyl alcohol phenyl oxidase/p-cresol methyl hydroxylase in N-substituted tyrosine decarboxylase (refer to Section 5.2.1). Both mechanisms require the formation of a p-quinone intermediate, emphasizing a shared step in these enzymatic processes.

6.2. Fe-Dependent Decarboxylases

Decarboxylation reactions catalyzed by iron-dependent enzymes have been observed in both heme (P450) and nonheme iron enzymes. While most iron-dependent enzymes catalyze a two-electron oxidative decarboxylation, a few enzymes, such as the recently discovered multinuclear nonheme iron-dependent (MNIO) enzyme ApyH or Fe(II) and α-ketoglutarate (Fe(II)/α-KG)-dependent ScoE, can perform four-electron oxidative decarboxylation.^116−118

The mechanisms of iron-dependent decarboxylases involve several shared steps. First, the iron cofactor activates dioxygen or hydrogen peroxide to form reactive species such as the high valent iron-oxo intermediate Fe^IV=O (Figure 18). The reactive species abstracts a hydrogen at the Cβ of the carboxylic acid, yielding an alkyl radical. In two-electron oxidative decarboxylation, three possible pathways are proposed for the decarboxylation of alkyl radical (Figure 18). The first pathway is an OH-rebound pathway involving radical hydroxylation followed by dehydration and decarboxylation. The second pathway is a single-electron transfer pathway in which the unpaired electron is transferred from the radical to the Fe^III–OH species to produce a substrate cation that subsequently undergoes decarboxylation. The third pathway involves diradical intermediate formation to support homolytic decarboxylation. While all three pathways have been proposed, studies on two different enzymes suggest that the OH-rebound pathway (i) is less favored. Rapid kinetic studies of the P450 peroxygenase OleT_Je (Section 6.2.1.2) have shown an unusually stable compound II, whose lifetime exceeds that of the radical species.¹¹⁹ Additionally, investigations on the Fe(II)/α-KG decarboxylase IsnB (Section 6.2.2.1) incubated with β-hydroxy acid did not observe a decarboxylated product.¹²⁰ Further research is needed to gain a deeper understanding of the mechanism of two-electron oxidative decarboxylation. The four-electron oxidative decarboxylation catalyzed by ApyH and ScoE has some features similar to the above, which will be discussed separately below.

Figure 18.

General mechanisms of homolytic decarboxylation catalyzed by iron-dependent enzymes.

6.2.1. Heme-Dependent Decarboxylases

Cytochrome P450 enzymes (P450s or CYPs) are a diverse group of heme-thiolate proteins capable of catalyzing numerous reactions, including hydroxylation, epoxidation, dehalogenation, and decarboxylation. These reactions are mediated by the reactive intermediate called Compound I, which is generated through two different pathways. Enzymes such as P450 monooxygenases utilize O₂ and NAD(P)H as the source of oxygen and electron sources, while enzymes like P450 peroxygenases directly employ hydrogen peroxide via the peroxide shunt pathway to produce Compound I (Figure 19A). P450 decarboxylases, which can utilize either of these pathways, will be discussed further in the following section.

Figure 19.

(A) Catalytic cycle of cytochrome P450 monooxygenases and peroxygenases: P450 peroxygenases utilize the peroxide shunt pathway to directly convert the substrate-bound P450 to the ferric-hydroperoxo species using H₂O₂. Biosynthetic pathways of (B) bottromycin A2, (C) camalexin, and (D) terminal olefins.

6.2.1.1. P450 Enzyme in Bottromycin A2 Biosynthesis

In the penultimate biosynthetic step en route to the potent antibiotic bottromycin A2 (BGC0000468),¹²¹ BotJ (also known as Bot^CYP), a cytochrome P450 decarboxylase enzyme,¹²² catalyzes the oxidative decarboxylation of the C-terminal thiazoline to a des-carboxy thiazole (Figure 19B). A similar modification, catalyzed by another P450 enzyme, is observed in the biosynthesis of the plant alkaloid camalexin (Figure 19C). The activity of Bot^CYP has been successfully reconstituted in vitro using the ferredoxin-ferredoxin reductase system.¹²²In vitro reconstitution also shows that Bot^CYP selectively acts on a D-Asp-containing intermediate.¹²³ The origin of this selectivity remains to be determined since the crystal structure of a close homologue of Bot^CYP (Sal^CYP from Salinispora tropica PDB: 7ABB) did not provide an adequate explanation.¹²²

6.2.1.2. P450 Peroxygenase in Olefin Biosynthesis

OleT_JE (Figure 19D) is a member of the peroxygenase family, a subfamily of P450-containing proteins.¹²⁴ This enzyme is responsible for catalyzing the oxidative decarboxylation of fatty acids, resulting in the production of corresponding terminal olefins. Initially discovered in the genome of the bacteria Jeotgalicoccus spp., OleT_JE has been found to generate terminal olefins such as 18-methyl-1-nonadecene and 17-methyl-1-nonadecene.¹²⁴ Noncanonical substrates, such as short-chain aliphatic or phenyl-substituted fatty acids, yield Cα and Cβ hydroxyl products.^125−127 The regio- and chemoselectivity of the reaction depend on multiple factors, including the structure of the substrate, the positioning of the substrate within the active site, and the accessibility of the oxidizing species Fe^IV=O to the substrate.

Recently, OleTP_RN, a homologue of OleT_JE, was discovered in the Rothia genus. OleTP_RN generates alkenes from a broad range of saturated and unsaturated fatty acids, including oleic and linoleic acids, the most naturally abundant fatty acids.¹²⁸ Two structural features differentiate the chemo-selectivity between OleT_JE and OleTP_RN (PDB ID: 5M0N and 8D8P, respectively). First, OleTP_RN lacks a long F-G loop, which anchors the fatty acids through residue Leu176 in OleT_JE. Second, OleTP_RN possesses a hydrophobic cradle in the distal region of the substrate-binding pocket. This cradle plays a crucial role in the activity of OleTP_RN on long-chain fatty acids and facilitates the release of products derived from short-chain fatty acids.

6.2.2. Mononuclear Nonheme Iron Decarboxylases

6.2.2.1. Fe(II)/α-KG Decarboxylases in the Biosynthesis of Isonitrile-Containing Natural Products

Many natural products contain isonitrile moieties. Two biosynthetic pathways to isonitriles have been characterized, both involve oxidative decarboxylations catalyzed by Fe(II)/α-KG oxygenases. These enzymes utilize iron, oxygen, and α-ketoglutarate as cosubstrates to generate a reactive Fe^IV=O species (Figure 20A). This species then oxidizes the substrates through a radical mechanism, allowing for a wide range of transformations such as hydroxylation, epimerization, halogenation, ring closure and expansion, in addition to decarboxylation. The first pathway proceeds through isonitrile synthases and Fe(II)/α-KG decarboxylases to produce vinyl isonitriles, including hapalindoles, rhabduscins, and byelyankacins (Figure 20B).^129−131 In the second pathway, a dual-function thioesterase and a Fe(II)/α-KG decarboxylase are responsible for the production of isonitrile lipopeptides such as SF2768, aerocyanidin, and amycomicin (Figure 20C).¹¹⁸

Figure 20.

Fe(II)/α-KG decarboxylases in the biosynthesis of isonitrile-containing natural products and their mechanisms. (A) Biosynthetic pathway of vinyl isonitrile-containing natural products involving isonitrile synthase followed by mononuclear iron Fe(II)/α-KG decarboxylase IsnB or AmbI3. (B) Biosynthetic pathway of the isonitrile lipopeptide SF2768 with the four-electron oxidative decarboxylation catalyzed by ScoE. (C) Mechanism of forming the ferryl species in Fe(II)/α-KG enzymes. (D) Mechanism of Fe(II)/α-KG dioxygenases catalyze the oxidative decarboxylation of N-alkyl glycine.

In the vinyl isonitrile biosynthetic pathway, the isonitrile synthase converts a Trp or a Tyr and a ribulose-5-phosphate to form isocyanopropanic acids. The isocyanopropanic acid subsequently undergoes an oxidative decarboxylation catalyzed by a Fe(II)/α-KG decarboxylase to yield indole/phenol vinyl isonitriles, which undergo further transformations to the final natural products (Figure 20A).^132,133 Extensive research has been conducted on the Fe(II)/α-KG decarboxylases involved in the biosynthesis pathway of vinyl isonitrile, specifically the enzymes AmbI3 and IsnB.^134,135 AmbI3, involved in ambiguine biosynthesis, exclusively synthesizes the Z isomer. Its homologue IsnB, strictly produces E isomer vinyl isonitriles.¹³⁶ The enzyme demonstrates versatility with different indolic substrates. A substrate scope study has revealed that the size of the substitution groups on the indole ring has a more significant impact on the biosynthesis of ambiguine derivatives than their position.¹³⁷

In the biosynthesis of isonitrile lipopeptides, the dual-function thioesterase facilitates the Michael addition of glycine to an α,β-unsaturated thioester, forming an N-alkyl glycine moiety. This is followed by a four-electron oxidative decarboxylation catalyzed by a Fe(II)/α-KG decarboxylase to generate the key isonitrile moiety. This pathway has been extensively investigated in the production of antibiotic isonitrile lipopeptides SF2369 and SF2768, derived from Actinomycetes and Streptomyces sp., respectively.¹³⁸ In SF2369 biosynthesis, Fe(II)/α-KG decarboxylase ScoE converts (R)-3-((carboxymethyl)amino)butanoic acid into (R)-3-isocyanobutanoic acid. The crystal structures of ScoE and its homologue SfaA have been reported, including the cocrystal of an N-alkyl glycine substrate and ScoE.^138,139 The structure revealed that one of the hydrogen atoms attached to C5 of the N-alkyl glycine substrate points toward the iron center at a distance shorter than 4.9 Å, suggesting its potential site of oxidation. It is important to note that ScoE utilizes two equivalents of α-KG in the decarboxylation process to produce two equivalents of the ferryl species, which ultimately leads to the formation of one equivalent of the isonitrile compound. The proposed mechanism for isonitrile formation by ScoE-type enzymes is illustrated in Figure 20D. In this mechanism, the ferryl species initiates dehydration by abstracting a hydrogen atom, followed by hydroxylation at the C5 position. This hydroxylated intermediate then undergoes dehydration, facilitated by Tyr96 (ScoE numbering) acting as a base, resulting in the formation of an imine intermediate. In the second oxidation step, the regenerated Fe^IV=O species abstracts a hydrogen atom at the C5 position within the imine intermediate, leading to the formation of a radical. This radical undergoes single electron transfer to the Fe^III-OH species, followed by decarboxylation, ultimately leading to the production of the desired isonitrile product.

6.2.3. Multinuclear Nonheme Iron Decarboxylases

6.2.3.1. Nonheme Diiron Oxygenase/Decarboxylases in the Biosynthesis of Lyngbyapeptins and Barbamide

Another mode of carboxythiazoline decarboxylation to yield thiazoles is showcased in the biosynthesis of lyngbyapeptins and barbamide. By contrast with P450 catalyzed decarboxylation for bottromycin described above (section 6.2.1.1), thiazoles in lyngbyapeptidns and barbamide are formed via nonheme diiron decarboxylase-catalyzed decarboxylation of carboxy thiazolines (Figure 21).^140,141 In a recent investigation led by Kudo et al., decarboxylase LynB7 was found to decarboxylate non-native truncated substrates (Figure 21, R = H, Ph), highlighting its synthetic versatility.¹⁴⁰ A proposed mechanism for LynB7-catalyzed decarboxylation, based on a kinetic isotope effect study at C5 of the thiazoline ring, suggests that the bis(μ-oxo)diiron^IV species (intermediate Q) abstracts a hydrogen atom at C5, yielding a radical intermediate.^140,142 This radical species then undergoes hydroxylation followed by decarboxylation (pathway i) or may undergo a further 1e- oxidation to induce decarboxylation to yield the thiazole product (pathway ii).

Figure 21.

Structure of lyngbyapeptin and barbamide and the proposed mechanism of the decarboxylation catalyzed by diiron decarboxylase LynB7.

6.2.3.2. Multinuclear Nonheme Iron-Dependent Oxidative (MNIO) Enzymes in RiPP Biosynthesis

According to crystal structure analysis, MNIO enzymes, a recently discovered enzyme class, contain two or three iron ions in their active sites.^143−145 Despite being a relatively new enzyme class, MNIO enzymes have displayed remarkable capabilities in catalyzing intricate transformations of peptide substrates, including rearrangement, deformylation, and N-Cα bond cleavage.^143,146,147 In a recent study, the reactivity of MNIO enzymes has been expanded to include the first reported example of oxidative decarboxylation within this enzyme family.¹¹⁶ This reaction is catalyzed by the oxidase ApyH, which works in conjunction with its partner protein ApyI (Figure 22). ApyH catalyzes the decarboxylation of the C-terminal Asp residue of the substrate peptide to yield a C-terminal aminopyruvic acid. In the proposed mechanism, the Fe^II ion initiates the reaction by abstracting hydrogen and hydroxylating the C-terminal Asp at Cβ. Subsequently, the enzyme abstracts another hydrogen at Cβ. The resulting radical intermediate can then undergo proton-coupled electron transfer, forming a β-keto acid that undergoes decarboxylation. Alternatively, the radical may stimulate decarboxylation, leading to the formation of an enol that tautomerizes into a ketone at Cβ.

Figure 22.

Mechanism of the four-electron oxidative decarboxylation of N-alkyl glycine by multinuclear nonheme iron-dependent enzyme ApyH.

6.2.3.3. Binuclear Nonheme Iron(II) Enzyme in Olefin Biosynthesis

It is not uncommon to observe different enzyme classes catalyzing similar reactions, particularly among those capable of catalyzing a broad range of transformations, as exemplified by carboxythiazoline decarboxylation (Section 6.2.1.1 and Section 6.2.3.1). Similarly, fatty acid decarboxylation to form olefins can be achieved through the action of enzymes such as the P450 peroxygenase OleT_JE (described in section 6.2.1.2), or the binuclear nonheme iron(II) enzyme UndA. UndA is responsible for the oxidative decarboxylation of lauric acid to afford 1-undecene, a prevalent hydrocarbon semivolatile metabolite in Pseudomonas (Figure 23).^148−150 The crystal structure of UndA (PDB ID: 4WWZ) reveals a substrate-binding site comprising a profound hydrophobic tunnel extending from the surface to the active site. This tunnel’s length governs UndA’s substrate scope, encompassing medium-chain fatty acids within the range of C10 to C14.¹⁴⁸ Examining the crystal structure further reveals the presence of two iron binding clusters in UndA of 4.9 Å between the two iron atoms. This distance significantly exceeds the average of 3.3 Å observed in soluble methane monooxygenases (sMMO). Moreover, the critical intermediate Q (Figure 21), essential for the sMMO-like mechanism, remains elusive in the case of UndA based on the Mössbauer spectroscopic characterization. This observation, coupled with chemical models derived from X-ray crystal structures and density functional calculations, leads to a proposed mechanism of forming the active iron radical species that diverge from the traditional sMMO mechanism. This uncommon mechanism of UndA has been described in detail in the work of Wang et al.¹⁵¹

Figure 23.

Fatty acid decarboxylation catalyzed by (A) P450 peroxygenase OleT_JE (Section 6.2.1.2), (B) binuclear nonheme iron(II) enzyme UndA (Section 6.2.3.3), and (C) photodecarboxylase CvFAP (Section 6.3).

6.3. Photodecarboxylase in the Biosynthesis of Alkanes

Fatty acid decarboxylation can be achieved through yet another mechanism by the fatty acid photodecarboxylase CvFAP, leading to the production of alkanes rather than alkenes as described above (Figure 23). CvFAP, discovered in the microalgae Chlorella variabilis NC64A,¹⁵² represents a unique algal-specific branch within the glucose methanol choline oxidoreductase family. Activated by blue light (λ = 400–520 nm), CvFAP efficiently catalyzes the decarboxylation of fatty acids, yielding CO₂ and the corresponding alkanes (Figure 23C).

The crystal structure of CvFAP (PDB: 6YRU) reveals a narrow hydrophobic tunnel serving as the substrate binding site. Within this tunnel, the hydrophobic tail of the fatty acid substrate interacts with Tyr466, while the carboxylate part interacts with the side chains of Arg451, Cys432, and Gln486. The FAD cofactor is positioned at the closed end of the tunnel, with its tricyclic ring oriented toward the carboxylic group of the substrate.¹⁵² This organization suggests that FAD, a light-absorbing chromophore, can serve as a medium to transfer energy from blue light to the carboxylic group of the substrate. This observation allows Sorigué et al. to lay the foundation for the mechanistic study of CvFAP, further explored by Scrutton et al., who described a red-shifted flavin intermediate (Figure 24).^152−155 In the proposed mechanism, blue light triggers the excitation of the FAD cofactor; the photoexcited cofactor (FAD*) seizes one electron from the anionic acid RCOO–, resulting in the formation of an anionic semiquinone FAD^·– and a carbonyloxy radical RCOO^·. The homolytic decarboxylation of the carbonyloxy radical RCOO·, giving rise to an alkyl radical R· and a CO₂ molecule, outpaces the slow back electron transfer, establishing decarboxylation as the predominant pathway. Subsequently, the alkyl radical undergoes a transformation into the alkane product RH through the return of electrons from FAD^•– and proton transfer from an unidentified donor.

Figure 24.

Proposed mechanism of the decarboxylation catalyzed by CvFAP.

The unique properties of CvFAP have positioned it as a potential biocatalyst for the conversion of low-cost microbial biomass into economically viable biofuels.¹⁵² Under blue light illumination, CvFAP has demonstrated near-quantitative yields of alkanes, with turnover numbers reaching up to 8000.¹⁵⁶ Significant efforts have been directed toward expanding substrate selectivity,^157−159 and improving catalytic efficiency.^160,161 Proof-of-concept studies have demonstrated integration of CvFAP into catalytic cascades,^162−164 these and other applications have been recently reviewed.¹⁶⁵

7. Cofactor-Independent Decarboxylases Utilizing Acid–Base Catalysis

Previously mentioned decarboxylases employ cofactors like thiamine pyrophosphate, pyridoxal phosphate, flavin, nicotinamide, and metals to facilitate the process. Nature has also evolved cofactor-independent decarboxylases, which predominantly operate through acid–base mechanisms, heavily relying on active site microenvironments and substrate electronics. In secondary metabolite biosynthesis, these cofactor-independent decarboxylases participate in synthesizing β-branching polyketides (Figure 25). The biosynthetic process encompasses three steps, with two involving decarboxylation. The initial step forms acetyl- or propionyl-S-ACP_donor through two distinct pathways, either ketosynthase (KS)-like domain catalyzed or acyltransferase/decarboxylase (AT/DC) catalyzed, both of which involve decarboxylation. The second step in β-branching, catalyzed by a multiprotein 3-hydroxy-3-methylglutaryl synthase (HMGS) cassette, entails an aldol addition resulting in a 3-hydroxy-3-methylglutaryl intermediate. The third step in β-branching comprises a hydration and an optional decarboxylation, catalyzed by an enoyl-CoA hydratase domain (ECH1 and ECH2, respectively) leading to carboxy or des-carboxy β-branching products, the latter of which may be β,γ or α,β-unsaturated.

Figure 25.

(A) Biosynthetic pathway and (B) structure of β-branching polyketides.

7.1. Ketosynthase-like Domain (KS^Q) Pathway to Acetyl-S-ACP_donor

In the initial step of β-branching polyketide biosynthesis, distinct pathways produce acetyl-S-ACP_donor and propionyl-S-ACP_donor. Acetyl-S- ACP_donor biosynthesis involves a tridomain loading module, widely distributed in various polyketides, including β-branching polyketides. The module comprises ketosynthase-like (KS^Q), acyltransferase (AT), and acyl carrier protein (ACP) domain. The AT domain loads malonyl moiety from malonyl-CoA to the ACP domain, followed by malonyl-ACP decarboxylation catalyzed by the KS^Q domain to yield acetyl starter unit.^166,167 Structural insights into the decarboxylation process are detailed in the work by Chisuga, et al.,¹⁶⁸ for the KS^Q domains GfsA KS^Q and CmiP4 KS^Q, found in the biosynthesis of macrolide antibiotics FD-891 and cremimycin, respectively.^169,170 Analyzing the crystal structure of GfsA-KS^Q with a malonyl thioester substrate analog enables the identification of essential residues involved in the decarboxylation mechanism. This mechanism resembles the decarboxylation process observed in β-keto acids (Figure 26). GfsA-KS^Q initiates decarboxylation by precisely positioning the substrate, facilitating the stabilization of the enolate intermediate through two conserved threonine residues in the active site of GsfA-KS^Q.

Figure 26.

Acid–base catalysis in the mechanism of the decarboxylation catalyzed by KS^Q.

7.2. Bifunctional Acyltransferase/Decarboxylase Pathway to Propionyl-S-ACP_donor

The second pathway that yields propionyl-S-ACP_donor involves a decarboxylation followed by condensation of methylmalonyl-CoA catalyzed by a bifunctional AT/DC.¹⁷¹ This pathway is observed in the biosynthesis of leinamycin,¹⁷² myxovirescin,¹⁷³ largimycin,¹⁷¹ and fogacin C.¹⁷⁴ Lohman et al. reported the cocrystal structure of the bifunctional AT/DC LnmK from leinamycin biosynthesis with methylmalonyl-CoA, revealing a shared active site for decarboxylase and acyltransferase activities.¹⁷⁵ Importantly, the sequence of events dictates that DC activity precedes AT activity. The proposed mechanism for the bifunctional enzyme LnmK unfolds as follows: (2R)-methylmalonyl-CoA undergoes decarboxylation, forming an enolate intermediate stabilized by Phe65 (Figure 27). The enolate then deprotonates the hydroxy group of Tyr62, giving rise to a phenolate. This phenolate, acting as a nucleophile, facilitates acyl transfer from propionyl-CoA to the enzyme, forming the pivotal intermediate propionyl-O-LnmK. Subsequently, the propionyl moiety is transferred to LnmL (ACP_donor), ultimately yielding propionyl-S-ACP_donor.¹⁷⁶

Figure 27.

Mechanism of the decarboxylation catalyzed by the bifunctional AT/DC LnmK.

7.3. ECH2 Domain-Catalyzed Decarboxylation to Yield β,γ- or α,β-Unsaturated Polyketides

The final step in some β-branched natural products is a decarboxylation step, catalyzed by the ECH2 domain, and is pivotal to shaping the diversity of β-branching polyketides. This step results in the formation of β,γ- or α,β-unsaturated products, which can undergo further modifications at the unsaturated moiety to yield even more diverse polyketide products (Figure 28). However, the factors that determine the formation of β,γ- or α,β-unsaturated products in this step are still unknown. Understanding these factors is essential for engineering the ECH2 domain to enable the biosynthesis of β-branched polyketides. The crystal structure of the ECH2 domain has been reported for CurF (PDB ID: 2Q2X) in curacins biosynthesis¹⁷⁷ and PksI (PDB ID: 4Q1G) in bacillaene biosynthesis (BGC0001089).¹⁷⁸ The mechanism closely resembles that of other cofactor-independent decarboxylases, commencing with the deprotonation of the carboxylic acid by an active site histidine (Figure 28). Following decarboxylation, the enolate is stabilized within an oxyanion hole formed by the backbone NHs of alanine and glycine residues at the active site. Finally, either another lysine (CurF) or the catalytic histidine (PksI) is proposed to protonate the intermediate. The ACP-bound substrate is found to be preferred over the CoA-bound substrate.^177,178

Figure 28.

Mechanism of decarboxylation catalyzed by the enoyl-CoA hydratase domain ECH2 PksI.

8. Conclusion

Living organisms extract energy from a variety of carbon substrates through multistep oxidation cascades and frequently utilize decarboxylation reactions to extract additional reducing equivalents from oxidizable substrates. The stepwise oxidation reactions that Nature relies on for energy lead to the biosynthesis of numerous metabolites that bear carboxylic acid functional groups. Natural products are often synthesized by utilizing these primary metabolites as building blocks and frequently repurpose primary metabolic enzymes for their biosynthesis. In order to build complex, bioactive structures it is often necessary to remove the carboxylic acid groups inherited from these primary metabolic building blocks. For instance, in alkaloid and carbohydrate biosynthesis, decarboxylation takes place at the monomer stage prior to subsequent elaboration into more complex structures and has repurposed enzymes already present in primary metabolic pathways such as PLP- and TPP-dependent decarboxylases. In other instances, decarboxylation tailors natural products to achieve bioactive, drug-like structures, for example, in the C-terminal modification found in daptides, AviCys-containing peptides, and carboxythiazoline decarboxylation.

Given the importance of decarboxylation to natural product biosynthesis and the broad array of substrates that undergo decarboxylation, it is unsurprising that nature has evolved a diverse range of enzymatic decarboxylative mechanisms tailored to match the electronic requirements of the substrate. Decarboxylation can occur via two-electron mechanisms, utilizing cofactors such as TPP and PLP to stabilize the carbanion formed during carbon-carboxylate bond cleavage. These cofactors establish a covalent bond with the substrate via carbonyl or amine and serve as electron sinks during decarboxylation. Substrates with substituent groups (such as -OH, -SH, or hydroxyphenyl) at the Cβ position can undergo oxidation to create an electronic sink within the substrate itself. In contrast, those lacking substituent groups at both the Cα and Cβ may require a radical species to activate the Cβ-H bond necessary for decarboxylation.

Due to the widespread availability of carboxylic acid substrates for commercial synthesis, there is significant interest in pursuing synthetic methods which proceed via decarboxylation. Decarboxylases show significant potential as biocatalysts for use in the synthesis of such chemicals. TPP- and PLP-dependent dependent enzymes have been successfully engineered to produce numerous commodity chemicals. TPP-dependent decarboxylases are widely used in constructing artificial biosynthetic pathways for the production of alcohols such as isobutanol, n-pentanol, and 1,2,4-butanetriol,^179−181 while PLP-dependent decarboxylases have been exploited to produce commodity chemicals and important precursors such as γ-aminobutyric acid, cadaverine (a biopolymer precursor), and dopamine.^182−184 Additionally, given that numerous decarboxylases are involved in the production of alkene and alkanes from abundant fatty acids, there is significant potential for their use in biofuel production.

Given the microscopic reversibility of enzymatic reactions, decarboxylases can also catalyze carboxylation despite the uphill thermodynamics.¹⁸⁵ Indeed, carboxylation reactions have been demonstrated with both TPP and prFMN-dependent enzymes. Recently, TPP-dependent enzymes have been employed in amino acid production from aldehydes by coupling the decarboxylase with a transaminase or amino acid dehydrogenase.¹⁸⁶ On the other hand, prFMN enzymes display carboxylation activity toward various substrates, including diverse aromatic compounds, and have been the subject of investigation for potential applications, although further research is necessary.^187−191

The rational engineering of novel metabolites or therapeutics combining enzymes from different biosynthetic pathways is a major goal of synthetic biology. Decarboxylases play a crucial role in this endeavor. For example, Fe(II)/α-KG enzymes can introduce clickable moieties in isonitrile-containing natural products, facilitating simplified synthesis of derivative molecules with increasing structural complexity or modification of their pharmacokinetic properties. In peptide-based natural products, decarboxylation of the C-terminal carboxylic acid group has been observed in various pathways. While further structure–activity studies are necessary to fully define the significance of C-terminal decarboxylation in terms of stability and activity, there are obvious benefits to this reaction for peptides. For instance, removing the C-terminal carboxylic acid group can reduce degradation by carboxypeptidases due to the crucial role of the carboxylic acid functionality in carboxypeptidase substrate binding.¹⁹² Decarboxylation improves the membrane permeability of peptides,¹⁹³ and facilitates cyclization processes, as evidenced by mycofactocin and Avi(Me)Cys-containing natural products.^100,115 With an eye toward applications, decarboxylases present in RiPP BGCs may also find utility in surface display methods that allow free C-termini as in certain subtypes of phage and yeast display.^194,195

With the rapid progress in genome sequencing and genome mining techniques, an increasing number of biosynthetic pathways are being elucidated, shedding light on the catalytic role of decarboxylation by novel classes of enzymes. Recent developments have unveiled fascinating examples, such as the iron-type alcohol dehydrogenase-like enzyme MpaBC involved in daptides biosynthesis,⁷⁷ as well as multinuclear nonheme iron-dependent oxidative enzymes capable of catalyzing a four-electron oxidative decarboxylation.¹¹⁶ These discoveries highlight the potential for the future discovery of novel decarboxylases and their diverse substrates. Since decarboxylation is by its very nature a ‘traceless’ reaction, we anticipate that numerous new natural product decarboxylases remain to be uncovered. This exploration will invariably lead to the discovery of novel enzyme classes, expanding our current understanding and offering possibilities for harnessing decarboxylases to produce valuable compounds, including biofuels, fine chemicals, and medicines.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. CRediT: N.A.N.: writing-original draft, writing-review & editing; J.H.F.: writing-review & editing; J.A.M.: conceptualization, writing-review & editing. CRediT: J.H.F.: writing-review & editing; J.A.M.: conceptualization, writing-review & editing.

The authors declare no competing financial interest.