Biosynthesis of guanidine-containing natural products in cyanobacteria

Abstract

Cyanobacteria are prolific producers of structurally diverse and biologically potent natural products, a subset of which feature guanidino moieties. Introduction and modification of the guanidine group confer tuned basicity and enable extensive hydrogen bonding, cation–π, and electrostatic interactions, facilitating high-affinity binding to numerous biological targets. Although the enzymatic processes responsible for guanidine modifications in cyanobacterial pathways remain somewhat obscure, recent investigations have begun to clarify the biosynthetic machinery that mediates these distinctive transformations. In this review, we summarize these advances, with particular emphasis on the enzymatic steps responsible for guanidine installation and tailoring. These enzymatic transformations include N-prenylation, cyclization, and tricyclic guanidinium formation, representing rare or previously undescribed biosynthetic strategies in nature. This review provides new insights into the metabolic and enzymatic versatility of cyanobacteria and a foundation for future advances in enzyme engineering and therapeutic discovery.

One-Sentence Summary: This review highlights recent advances in understanding how cyanobacteria enzymatically install and modify guanidino groups to produce bioactive natural products.

Graphical Abstract

Graphical Abstract.

Introduction

Cyanobacteria are widely distributed in various water bodies, soil, and extreme environments, and their biotechnological potential in agriculture, food production, and drug discovery is increasingly recognized (Singh et al., 2016). In recent years, cyanobacteria have become an important source of bioactive natural products, including polyketides, peptides, terpenes, and alkaloids, with many featuring distinctive guanidine-containing fragments that expand the structural diversity of natural products (Baunach et al., 2024; Burja et al., 2001; D'Agostino, 2023; Dittmann et al., 2015; Singh et al., 2011; Weiss et al., 2025).

Guanidine can be toxic to cyanobacteria at elevated concentrations, leading to genomic instability and impaired growth (Itzenhäuser et al., 2025). However, certain cyanobacteria have evolved mechanisms to degrade guanidine, mitigating its harmful effects by expressing the Ni²⁺-dependent guanidine hydrolase GdmH (Funck et al., 2022; Wang et al., 2021). The discovery of GdmH underscores the ecological significance of guanidine in aquatic cyanobacteria. When introduced into natural product scaffolds, the guanidine moiety may form various interactions with proteins and other biomolecular targets through hydrogen bonding, charge pairing, and cation-π interactions (Berlinck & Romminger, 2016). These properties underpin the diverse biological activities exhibited by guanidine‐bearing natural products, including anti-tumor, anti-viral, anti-inflammatory, hypoglycemic, and neuroprotective activities (Gomes et al., 2023). For example, the sesterterpenoid scytoscalarol exhibits antimicrobial activities against a range of bacterial and fungal pathogens (Mo et al., 2009). Suomilide peptides serve as potential serine protease inhibitors (Ahmed et al., 2021; Fujii et al., 1997; Schneider et al., 2023). Saxitoxins, a family of paralytic shellfish toxins, are blockers of voltage-gated sodium channels and are being developed as neuroscience research tools and local anesthetics (Schantz et al., 1957; Thottumkara et al., 2014; Wiese et al., 2010).

Several biosynthetic gene clusters (BGCs) responsible for cyanobacterial guanidine-containing natural products have recently been discovered, revealing a diverse array of enzymes that install and functionalize the guanidine moiety. These tailoring transformations include N-prenylation, cyclization, and other specialized processes. This article reviews recent progress in understanding the biosynthesis of guanidine-containing natural products from cyanobacteria, with the compounds grouped by structural family.

Cyanobactin-Type ribosomally synthesized and post-translationally modified peptides (RiPPs)

Cyanobactins are ribosomally synthesized and post-translationally modified peptides (RiPPs) produced from cyanobacteria and contain diverse oxidation, heterocyclization, and prenylation modifications (Czekster et al., 2016; Martins & Vasconcelos, 2015; Sivonen et al., 2010; Zhang et al., 2023). In recent years, a great deal of research interest has been focused on the prenylation of cyanobactins, since this modification of peptide precursors can increase membrane permeability and bioavailability, with potential applications in the development of new bioactive peptide drugs. Prenylation reactions during cyanobactin biosynthesis are catalyzed by prenyltransferases belonging to the ABBA superfamily, which was named based on the array of secondary structures containing a repeated α-β-β-α motif (Mori, 2020; Zhang et al., 2023). Depending on the amino acid residues they recognize, the enzymes are classified as O-, C-, or N-prenyltransferases (Zhang et al., 2023). Examples of O-prenyltransferases include LynF (for aestuaramides) (McIntosh et al., 2011, 2013), PagF (for prenylagarmides) (Hao et al., 2016), PirF (for piricyclamides) (Estrada et al., 2018; Morita et al., 2018), SphF (for sphaerocyclamide) (Martins et al., 2018), TruF1(for trunkamide) (Tianero et al., 2016) and TolF (for tolyamide) (Purushothaman et al., 2021), which prenylate tyrosine, serine, and threonine in their peptide substrates. The C-prenylation of tyrosine and histidine is catalyzed by KgpF and LimF, respectively, through O-prenylation followed by Claisen rearrangement (Parajuli et al., 2016; Zhang et al., 2022). The N-prenylation of the tryptophan residue in anacyclamides is accomplished by AcyF (Dalponte et al., 2018). Recent studies demonstrated that the Nω-position of an arginine residue is prenylated by AgcF and AutF in the biosynthesis of argicyclamides (1–3) and autumnalamide A (4), respectively (Clemente et al., 2022; Phan et al., 2021). AgcF performs the bis-N-prenylation of the argicyclamide precursor (Fig. 1A) (Phan et al., 2021). Structural alignment with PagF revealed that the amino acids around the substrate binding pocket of AgcF have less steric hindrance than those in PagF, enabling it to catalyze the bis-N-prenylation modification (e.g. F69/G67, H138/G133, W271/C267, and Y292/L289 in PagF and AgcF, respectively). In addition, AgcF can utilize geranyl pyrophosphate (GPP) as a non-native prenyl donor to produce trace amounts of Arg-N-geranylated products, suggesting that AgcF has some tolerance toward prenyl pyrophosphate substrates. Thus, the structure-based engineering of AgcF holds promise for expanding its substrate scope and serves as a focus for future research. The substrate specificity of AutF was explored with 24 linear and macrocyclic peptides and protected amino acids (Fig. 1B) (Clemente et al., 2022). AutF exhibited higher activity with L-arginine-containing linear substrates, compared to L- or D-arginine-containing cyclic substrates. It is worth noting that AutF can also accept non-native homoarginine-containing peptide substrates, such as 9-fluorenylmethoxycarbonyl-L-homoarginine-OH (5) (Fig. 1B), implying the potential of AutF as a powerful tool for solid phase synthesis. Recently, another Arg-Nω-bisprenyltransferase DciF from Dolicospermum circinale AWQC310F that catalyzes Arg-N^ω-bisprenylation on various cyclic/linear peptides was discovered through a gene mining approach (Fujita et al., 2025). Three key residues E49/H167/T134 that interact with the guanidine group were identified based on crystal structure analysis. Furthermore, substitution of G65 with Ile or Leu in DciF led to accumulation of the monoprenylated product, while the AutF D65G variant gained a bisprenylation activity, indicating a critical role of these residues in controlling the repeated prenylation. These studies present novel arginine-N, N′-bisprenyltransferases and offer the structural basis for guanidine prenylation in RiPP natural product biosynthesis.

Fig. 1.

N-prenylation of arginine side chains catalyzed by cyanobactin-type prenyltransferases. (A) AgcF, (B) AutF. Fmoc, 9-fluorenylmethoxycarbonyl moiety.

Aeruginosins, Aeruginoguanidines, and Microguanidines

Aeruginosin family natural products, discovered from cyanobacteria and sponges, are nonribosomal linear tetrapeptides that exhibit inhibitory activities against various serine proteases (Ersmark et al., 2008; Liu et al., 2023). The characteristic core structure of this natural product family is the 2-carboxy-6-hydroxy-octahydroindole (Choi) moiety. In addition, the C-terminus of the tetrapeptide scaffold contains a modified fragment derived from arginine, in which the guanidine moiety may be modified (Fig. 2A). Examples include the cyclic hemiaminal in aeruginosin 89-B (6) (Ishida et al., 1999) and aeruginosin 686 (7) (Ishida et al., 2009), the N-prenyl group in aeruginosin KB676 (8) (Elkobi-Peer & Carmeli, 2015), and the pyrroline ring of 1-amino-2-(N-amidino-3-Δ3-pyrrolinyl)ethane (Aeap) in suomilide (9) (Ahmed et al., 2021; Fujii et al., 1997; Schneider et al., 2023) banyaside A (10) (Pluotno & Carmeli, 2005), aeruginosin 126A (11) (Ishida et al., 2007), dysinosin B (12) (Carroll et al., 2004, 2002) and varlaxin 1046A (13) (Heinila et al., 2022).

Fig. 2.

(A) Structures of aeruginosins containing arginine derivatives. (B) Proposed formation of the hemiaminal in aeruginosin 686.

The nonribosomal peptide synthetase (NRPS) encoded by the aeruginosin 686 (7) BGC contains a C-terminal reductase domain, which is speculated to reduce a thioester intermediate. Upon peptide release, the resulting aldehyde intermediate likely undergoes cyclization to form the guanidine hemiaminal in aeruginosin 686 (7) (Fig. 2B).

The structure of Aeap features a unique pyrroline ring with two additional carbon atoms compared to the parent agmatine scaffold. The biosynthetic pathway responsible for Aeap formation was recently elucidated and found to involve Aer3, previously annotated as an unknown protein, and AerC, a Rieske oxygenase (Zhang et al., 2025). These enzymes were identified as being responsible for Aeap formation through comparative analyses of BGCs encoding aeruginosins with and without Aeap. Although Aer3 appears to lack a conserved motif based on its primary sequence, its predicted structure revealed partial resemblance to UbiA-type prenyltransferases and terpene synthases. In vitro biochemical assays demonstrated that Aer3 utilizes agmatine (14) and dimethylallyl pyrophosphate (DMAPP) as substrates to generate N⁴-isopentenylagmatine (15), highlighting a different regioselectivity of guanidine prenylation compared with AgcF and AutF (Fig. 3A). Intriguingly, a phylogenetic analysis revealed that Aer3 and cyanobactin-type prenyltransferases, which both recognize arginine moieties, form distinct evolutionary lineages (Fig. 3B) (Zhang et al., 2025). This analysis suggested that Aer3 represents a novel subgroup of prenyltransferases. Structural and sequence analyses further suggested that the conserved motifs responsible for magnesium ion coordination and prenyl pyrophosphate binding in Aer3 are DXXXE and DDD, in contrast to the canonical DDXXD and NTE motifs found in typical UbiA-type prenyltransferases.

Fig. 3.

(A) Aeap biosynthesis. (B) Phylogenetic analysis of prenyltransferases and terpene synthases. PT, prenyltransferase; TS, terpene synthase. (C) Possible mechanisms of AerC-catalyzed pyrroline formation. Fig. 3B and C are reprinted with permission from J. Am. Chem. Soc., 2025, 147(13): 10853–10858. © 2025 American Chemical Society.

In vitro assays suggested that the Rieske oxygenase AerC catalyzes the conversion of N⁴-isopentenylagmatine (15) to Aeap (16) (Fig. 3A). Isotopic labeling experiments with [1,2-¹³C₂]-DMAPP demonstrated that the carbon atoms at the C6 and C7 positions of the DMAPP-derived unit were retained in the Aeap product during the AerC-catalyzed reaction. Furthermore, experiments using [4,4,4,5,5,5-²H₆]-DMAPP revealed that the three-carbon fragment eliminated during Aeap pyrroline ring formation is released in the form of acetone. It is speculated that Aeap formation from N⁴-isopentenylagmatine (15) involves two sequential two-electron oxidation reactions (Fig. 3C) (Zhang et al., 2025). In the first step, a hydrogen atom from the C3 position of N⁴-isopentenylagmatine (15) may be abstracted by an unknown reactive iron species, generating a C3-centered radical (17). This radical may undergo intramolecular radical addition to the prenyl side chain to form a C3–C7 bond. The resulting radical intermediate (18) may then be further oxidized to yield a hypothetical hydroxylated species (19), which has not been experimentally verified. In the second oxidative step, a hydrogen atom may again be abstracted from C3 by a reactive iron species, generating another C3-centered radical intermediate (20). This species may undergo radical-mediated β-cleavage to produce Aeap (16) and acetone, following an additional one-electron oxidation. Alternatively, the C–C bond cleavage may be initiated via O–H heterolytic cleavage, as depicted in Fig. 3C. Clearly, further experiments are necessary to evaluate these and other mechanistic possibilities. Regardless of the detailed reaction mechanism, the biosynthesis of Aeap is an example of a unique guanidine modification that involves the incorporation of the isopentyl five-carbon fragment, followed by C–C bond-forming cyclization as well as C–C bond cleavage to complete the formation of the pyrroline ring. In particular, the C–C bond formation and cleavage processes are mediated by the single Rieske oxygenase AerC.

The biosynthetic origin of the N-prenylated structure in aeruginosin KB676 (8) has remained unclear, due to the lack of genomic information for the producing Microcystis species (Elkobi-Peer & Carmeli, 2015). Nevertheless, the identification of Aer3, an enzyme that accepts agmatine (14) as its substrate, implies that the N⁴-prenylated agmatine moiety in aeruginosin KB676 (8) is biosynthesized through a similar enzymatic mechanism (Zhang et al., 2025), which awaits experimental verification.

Cytotoxic aeruginoguanidines (AGDs) (21–23) and microguanindines (MGDs) (24–27) isolated from Microcystis are another class of linear nonribosomal peptides that contain guanidine fragments (Pancrace et al., 2019) (Fig. 4A). The BGCs for this class of natural products reportedly encode an arginine-recognizing NRPS and several tailoring enzymes, such as the sulfatase/sulfotransferases AgdG/AgdD, AgdP and AgdR, and the prenyltranferase AgdJ (Fig. 4A) (Pancrace et al., 2019). AgdJ was speculated to be responsible for the prenylation of the guanidine groups with DMAPP or GPP, possibly occurring after the ester/amide formation. In the above-mentioned study of Aeap (16) (Zhang et al., 2025), Aer3 was found to share sequence similarity with AgdS and AgdT, previously annotated as hypothetical proteins encoded in the MGD BGC (Fig. 4A). In vitro experiments showed that AgdS/AgdT can recognize L-arginine (28) as substrate for N-isopentenylation and N-geranylation, with a marked preference for the latter, consistent with the presence of a geranyl group in MGDs (24–27) (Fig. 4B). Whereas Aer3 favors agmatine (14) and DMAPP, AgdS/AgdT preferentially process arginine (28) and GPP. However, it is currently unclear whether the native substrate of AgdS/AgdT is L-arginine (28), because the arginine fragment of MGDs also includes a permethylated α-amino group. Thus, the order of the modification steps in MGD biosynthesis requires further verification. In addition, some MGD BGCs also contain multiple copies of AgdS/AgdT, which are named AgdS'/AgdT', with unknown biosynthetic significance. Thus, the functions of these proteins as well as AgdJ will also require experimental characterization.

Fig. 4.

(A) Aeruginoguanidine and microguanidine structures and BGCs. (B) Functions of AgdS/T demonstrated by in vitro experiments.

Saxitoxin

Saxitoxin (STX, 32) and its analogs are a group of natural neurotoxic guanidine-containing alkaloids from marine dinoflagellates and freshwater cyanobacteria (Moustafa et al., 2009; Schantz et al., 1957; Thottumkara et al., 2014) (Fig. 5). They are potent neurotoxins that block voltage-gated sodium channels, leading to paralysis and causing environmental problems associated with their accumulation in shellfish like clams, mussels, oysters, and scallops (Kellmann et al., 2008; Wiese et al., 2010). Saxitoxin (32) has a unique guanidinium skeleton comprising a perhydropurine-like tricyclic core (6-5-6 ring system), which is critical in cation–π and hydrogen bonding interactions with sodium channels (Thottumkara et al., 2014). The BGCs responsible for saxitoxin biosynthesis have been identified in various cyanobacterial strains (Hackett et al., 2013; Kellmann et al., 2008; Mihali et al., 2009; Moustafa et al., 2009). The polyketide synthase-like enzyme SxtA, containing an S-adenosylmethionine-dependent methyltransferase, as well as general control non-derepressible 5-related N-acetyltransferase, acyl carrier protein, and 8-amino-7-oxononanoate synthase domains, initiates saxitoxin biosynthesis by extending L-arginine (28) to form 4-amino-3-oxo-guanidinoheptane (29). The amidinotransferase SxtG then catalyzes the amidino transfer reaction from arginine onto the amino group of 4-amino-3-oxo-guanidinoheptane (30). The resulting product (30) may undergo non-enzymatic cyclization to afford one of the cyclic guanidinium units (in 31) of saxitoxin (32) (Fig. 5) (Lukowski et al., 2020). Alternatively, based on in vitro analysis, the Rieske oxygenase SxtH is proposed to generate 33 by hydroxylating 4-amino-3-oxo-guanidinoheptane (29), which is reportedly a substrate of SxtG (Lukowski et al., 2018). Thus, further investigation will be needed to clarify the definitive biosynthetic functions of SxtG and SxtH. The hydroxyl group at C12 in saxitoxin may be installed by the Rieske oxygenase SxtT (Liu et al., 2022; Lukowski et al., 2018) during the late-stage biosynthesis. Nevertheless, little is known about how the two additional C–N bonds are formed to construct the characteristic tricycle of saxitoxins.

Fig. 5.

Proposed biosynthetic pathway of saxitoxin.

Among the saxitoxin biosynthetic enzymes, the crystal structure of the amidinotransferase SxtG has been solved (Lukowski et al., 2020). Structural alignment of SxtG with the polyamine amidinotransferase HsvA (Shanker et al., 2017) showed that the substrate binding pocket of SxtG is more solvent-exposed and polar than that of HsvA. The conserved C362 likely serves as a catalytic residue that forms a covalent link with the amidino group donated by l-arginine and then transfers it onto the amine substrate. This process may be facilitated by a hydrogen bonding network involving D209, D131, and H258. The negatively charged binding pocket, enriched with aspartic acid and glutamic acid residues such as E58, E59, E255, and E276, and the aspartic acid-rich active site within the substrate binding pocket, likely facilitate the binding of positively charged substrates through electrostatic interactions. This structural variation relative to other structurally characterized amidinotransferases is a key determinant of the substrate specificity, enabling SxtG to accommodate a diverse array of amine substrates including α-aminoketones, α-amino methyl esters, and other primary amines.

Cylindrospermopsins

Cylindrospermopsins (CYN) belong to a class of alkaloid cyanotoxins that includes structurally related compounds such as 7-epi-cylindrospermopsin (34), 7-deoxy-cylindrospermopsin (35), and 7-deoxy-desulfo-cylindrospermopsin (36), produced by cyanobacteria (Fig. 6A) (Kinnear, 2010). CYN contains a guanidine-containing tricyclic structure linked with a uracil-like ring fused with arginine, which makes CYNs unique among cyanobacteria-derived alkaloids. The BGC responsible for the biosynthesis of CYN was first identified and sequenced in Cylindrospermopsis raciborskii AWT205, and subsequently found in other cyanobacteria (Mihali Troco et al., 2008) (Mazmouz et al., 2010). Feeding experiments revealed that guanidinoacetic acid can be a source of the N atom in the guanidino fragment of CYN (Burgoyne et al., 2000). It is proposed that the biosynthetic pathway of CYN is initiated by CyrA, an L-arginine: glycine amidinotransferase that catalyzes the formation of a guanidine unit on the glycine precursor (37), using l-arginine as a co-substrate, to form guanidinoacetate (38) (Fig. 6B) (Muenchhoff et al., 2012, 2010). CyrB then performs a polyketide extension process and guanidinium cyclization through a Michael-type nucleophilic mechanism. After additional chain extension catalyzed by CyrC, CyrD prepares an α,β-unsaturated thioester that is attacked by the guanidine unit to form a six-membered ring. In a similar mechanism, CyrE constructs the third ring involving guanidine. Arg-CYN (39) is likely formed by the condensation of L-arginine with the β-keto thioester intermediate linked to CyrF, which may be catalyzed by the N-acetyltransferase (NAT)-like domain identified at the C-terminal end of CyrF (Méjean et al., 2022). CyrG may subsequently catalyze the hydrolysis of Arg-CYN (39) to give 7-deoxy-desulfo-CYN (36) and ornithine. While the proposed condensation and hydrolysis mechanisms involving the arginine side chain are intriguing, the functions of the hypothesized enzymes have not been fully confirmed.

Fig. 6.

(A) Structure of cylindrospermopsins. (B) Proposed biosynthesis pathway.

Guanitoxin

Guanitoxin (anatoxin-a(s)) (43) is an irreversible inhibitor of acetylcholinesterase, acting through a mechanism similar to that of organophosphates, such as the chemical warfare agent sarin and the banned pesticide parathion (Fernandes et al., 2024; Fiore et al., 2020). It consists of a four-carbon backbone modified by a guanidino group and a phosphate moiety attached to the oxygenated N1 position. Moore's group identified the guanitoxin-producing cyanobacterial strain Sphaerospermopsis torques-reginae ITEP-024 and described the complete guanitoxin biosynthetic pathway (Fig. 7A) (Lima et al., 2022). The pathway begins with the hydroxylation of L-arginine (28) at the C5 position to generate 4-hydroxy-L-arginine (40), which is catalyzed by the α-ketoglutarate (αKG)-dependent nonheme iron enzyme GntB. The PLP-dependent enzyme GntC performs the C–N bond-forming cyclization of 4-hydroxy-L-arginine (40) to produce the key cyclic guanidino product (41) (Fig. 7B). Subsequent oxidation, C–C cleavage, amination, and N-methylation steps convert the GntC product (41) to pre-guanitoxin (42), in which the guanidine nitrogen is hydroxylated by the N-hydroxylase GntA and subsequently phosphorylated by the kinase GntI. The GntJ-catalyzed methylation of one of the oxygen atoms then generates guanitoxin (43).

Fig. 7.

(A) Proposed biosynthesis pathway of guanitoxin. (B) Proposed mechanism of GntC.

The crystal structure of the key PLP-dependent cyclase GntC has been solved, and the mechanism of GntC was proposed based on in vitro assays with enzyme variants (Cordoza et al., 2023). The GntC reaction begins with external aldimine formation between PLP and 4-hydroxy-L-arginine (40), followed by α-deprotonation to yield a quinonoid intermediate (44). Reprotonation at C5 forms a ketimine (45), enabling β-deprotonation and enamine (46) formation. A water-mediated hydrogen bonding interaction between the γ-hydroxy group at the C4 position and N52 is proposed to activate the intermediate for dehydration, generating an α,β-unsaturated iminium intermediate (47). The terminal guanidine moiety then undergoes a Michael-type addition to form the five-membered cyclic guanidine ring in 48. Tautomerization of 48, followed by imine exchange with K219, produces the GntC product (41). This PLP-mediated intramolecular cyclization is a rare example of how nature divergently produces cyclic arginine noncanonical amino acids.

Conclusions

Guanidine-modified natural products embody the remarkable structural and functional diversity of cyanobacterial secondary metabolites. Elucidations of their BGCs and associated enzymes have revealed the specialized mechanisms for modifications of guanidine groups through various prenylations and cyclizations. These chemical transformations play key roles in constructing the chemical architectures critical for their biological activities and provide unique enzymatic tools for synthetic biology and drug development. With the deepening understanding of these pathways, the potential for designing novel biocatalysts and engineered guanidine pharmacophores will continue to expand. In addition, the recognition of the ecological and toxicological impacts of these metabolites has strengthened the need for continued monitoring and research of cyanobacterial blooms. In summary, the studies of guanidine-modified natural products highlight the biosynthetic pathways of cyanobacteria and their value as sources of bioactive molecular frameworks.

Funding

This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (Japan Society for the Promotion of Science KAKENHI Grant Number JP22H05123, JP24H01309, JP25H02006, JP25K02417) and Japan Science and Technology Agency (FOREST Grant Number JPMJFR2305). W.Z. is a recipient of the JSPS Postdoctoral Fellowship for Foreign Researchers.

Conflict of interest

The authors declare no conflict of interest.

Data availability

No new data were generated or analyzed in this review.