Cyanobactins

Cyanobacteria are one of the most promising microbial groups in the search for novel bioactive compounds. Low molecular weight peptides containing a range of proteinogenic and nonproteinogenic amino acids are a major class of bioactive compounds produced by cyanobacteria (Burja et al. 2001). A plethora of small cyclic or linear peptides with a surprisingly high level of structural variation have been reported from cyanobacteria (Welker and von Döhren 2006; Sivonen and Börner 2008). These peptides are produced by both nonribosomal and ribosomal biosynthetic pathways in cyanobacteria. The first nonribosomal pathways for cyanobacterial peptides were described in 2000 (Tillett et al. 2000; Rouhiainen et al. 2000) whereas the first ribosomal pathway was shown in 2005 for the cyanobactin patellamide (Schmidt et al. 2005).

Cyanobactin was proposed as a collective name for cyclic peptides which contain heterocyclized amino acids or isoprenoid amino acid derivatives (Donia et al. 2008a; Schmidt and Donia 2009). Cyanobactins were initially defined to contain oxazolines, thiazolines, or their oxidized derivatives oxazoles and thiazoles (Fig. 1). This definition was recently broadened to include cyclic peptides which consist solely of proteinogenic amino acids (Leikoski et al. 2010). Isoprenoid amino acid derivatives are rare but found for example in trunkamide, patellin, and anacyclamides (Tables 1 and 3, Fig. 1).

Fig. 1
figure 1

The chemical structures of a selection of cyanobactins. Trunkamide was isolated from L. patella, tenuecyclamide from N. spongiaeforme, anacyclamide from Anabaena, trichamide from T. erythraeum, and ulithiacyclamide and patellamide from Prochloron (originally from L. patella). The corresponding biosynthetic gene clusters are shown in Fig. 2

Table 1 Examples of cyclic peptides and their structural features characterized from ascidians and sponges

Occurrence and chemical diversity of cyanobactins

More than a hundred cyanobactins have been identified from symbiotic associations formed between cyanobacteria and ascidians (Table 1) or from free-living cyanobacteria (Tables 2 and 3, Fig. 1). This makes cyanobactins one of the largest classes of cyanobacterial peptides (Donia et al. 2006; Schmidt and Donia 2009). Cyanobactins identified from filter-feeding organisms, such as ascidians and sponges, usually contain from six to ten amino acids and varying numbers and combinations of oxazoles, oxazolines, thiazoles, and thiazolines. A few cyanobactins, such as comoramides, contain prenylated amino acids, and ulithiacyclamides have disulfide bridges between two cysteine amino acids (Table 1).

Table 2 Examples of cyclic peptides, their structural features reported from a diverse selection of cyanobacteria organized according to the alphabetical order of the compound name
Table 3 Examples of cyclic homodetic peptides from cyanobacteria

It is unclear if the cyanobactins are produced by the filter-feeding organisms themselves, heterotrophic bacteria, or cyanobacteria associated with these organisms (Table 1). Some cyanobactins have now been verified to be produced by cyanobacteria (Schmidt et al. 2005; Donia et al. 2006). The biosynthetic origin of most of the analogous cyclic peptides reported from cyanobacteria (Tables 2 and 3) is currently unknown. To date, only ribosomal biosynthetic pathways have been described to produce these cyclic peptides (Schmidt et al. 2005; Donia et al. 2006, 2008a; Sudek et al. 2006; Ziemert et al. 2008b; Leikoski et al. 2010). However, a nonribosomal peptide synthetase pathway could be an alternative route for the biosynthesis of these compounds.

Cyanobacterial strains produce ribosomal cyanobactins which contain heterocyclized amino acids (Table 2) and also cyclic peptides which consist solely of unmodified proteinogenic amino acids, occasionally with prenyl attachments (Table 3). Cyanobactins which are found in by cyanobacteria and contain heterocyclized amino acids range in size from six to eleven amino acids (Table 2). Oxazoles and thiazoles are common while oxazolines or thiazolines occur with a lower frequency (Table 2). The cyanobactins which lack heterocyclized amino acids vary in length from seven to 20 amino acids (Table 2). Interestingly, a feature uniting cyanobactins without heterocyclized amino acids in addition to the occasional prenyl attachment is the conserved presence of a proline residue (Table 3).

The biosynthetic genes for cyanobactin production have been described in distantly related cyanobacteria Prochloron, Trichodesmium, Microcystis, Nostoc, Lyngbya, and Anabaena (Schmidt et al. 2005; Donia et al. 2006, 2008a; Sudek et al. 2006; Ziemert et al. 2008b; Leikoski et al. 2010; Fig. 2). In addition, one of the protease genes responsible for cleavage of the cyanobactin precursor peptide was shown to be common among planktonic freshwater cyanobacteria and present in 48 out of 132 strains studied (Leikoski et al. 2009). These planktonic cyanobacteria included fresh and brackish water strains from filamentous heterocystous (Anabaena, Aphanizomenon, Nodularia), filamentous (Planktothrix), as well as colony-forming (Microcystis and Snowella) cyanobacteria. The biosynthetic pathway appears to be relatively common in these strains (Leikoski et al. 2009), but detailed analysis of the gene clusters should be carried out and the structure of the compounds remains to be identified.

Fig. 2
figure 2

Cyanobactin gene clusters published from seven distantly related cyanobacteria. These gene clusters are typified by genes encoding proteases (yellow), a short precursor peptide (red), proteins involved in the maturation of the cyanobactin (black), as well as conserved and hypothetical open reading frames (white)

Biosynthesis of cyanobactins

Cyanobactins are produced through the proteolytic cleavage and head-to-tail (N–C) cyclization of precursor peptides coupled with modification of specific amino acids as in many other natural products (Oman and van der Donk 2010). The cyclic structure is formed via an amide linkage of the α-carbonyl of C-terminal amino acid and α-amino group of the N-terminal amino acid yielding a homodetic cyclic peptide. In cyanobactin biosynthesis, the precursor peptide directly encodes one or more cyanobactins flanked by the putative recognition sequences at which the precursor peptide is cleaved by two proteases (Schmidt et al. 2005; Lee et al. 2009). Cyanobactin precursor peptides encode more than one cyanobactin, and this could be a means to enhance the levels of peptide production or represent a mechanism for generating chemical diversity. The cyanobactin gene cluster encodes two proteases demonstrated to be involved in the cleavage of the precursor peptide and cyclization of the cyanobactin (Lee et al. 2009). The PatA protease encoded in the patellamide biosynthetic gene cluster was shown to cleave the precursor peptide at the N-terminal recognition sequence while the PatG protease cleaved the precursor peptide at the C-terminal recognition sequence (Lee et al. 2009). However, only the PatG protease was required for N–C cyclization of the cyanobactins (Lee et al. 2009).

The cyanobactin biosynthetic genes encoded in gene clusters are approximately 10 kb in size and contain between 7 and 12 genes (Fig. 2). The gene order is not strictly conserved (Fig. 2). However, most often, the biosynthetic genes are organized as in the patellamide pat gene cluster. All of the cyanobactin gene clusters contain two proteases, which work in tandem, a short precursor peptide as well as proteins involved in the maturation of the cyanobactins. The cyanobactin gene clusters in Prochloron are highly similar (Schmidt et al. 2005; Donia et al. 2006).

Thiazoles and oxazoles are formed through the heterocyclization and subsequent oxidation of cysteine, serine, and threonine amino acids. Cyanobactin gene clusters typically contain a gene encoding a PatD homolog which is predicted to heterocyclize cysteine, serine, and threonine to thiazolines and oxazolines (Schmidt et al. 2005). A PatF homolog is often encoded in cyanobactin gene clusters and thought to be involved in the heterocyclization and/or prenylation of cyanobactins (Schmidt and Donia 2009). The oxidase domain of the bimodular PatG protein is believed to catalyze the oxidation of thiazolines and oxazolines to thiazoles and oxazoles (Schmidt et al. 2005). Intriguingly, PatB and PatC are encoded in nearly all cyanobactin gene clusters but are nonessential, with patellamides being produced by heterologous expression of the pat gene cluster in Escherichia coli in the absence of the patB and patC genes (Donia et al. 2006, 2008a).

Patellamides, lissoclinamides, ulithiacyclamides, and patellins

Biologically active cyclic peptides have been reported from Lissoclinum patella ascidians, but the biosynthesis of the compounds is often attributed to the small photosynthetic obligate symbionts of these ascidians (Sings and Rinehart 1996). The findings of Schmidt and coworkers demonstrated that the cyanobactin pathway was responsible for the production of patellamide in symbiotic uncultured Prochloron spp. as well as lissoclinamides, ulithiacyclamides, patellin, and the anticancer drug candidate trunkamide (Schmidt et al. 2005; Donia et al. 2006; Donia et al. 2008a). The cyanobactins which are produced by Prochloron spp. are structurally diverse and range in size from six to eight amino acids (Table 1, Schmidt et al. 2005; Donia et al. 2006, 2008a).

Ascidians harbor a mixture of symbiotic Prochloron strains. The portion of the precursor peptide gene encoding the core peptide in the cyanobactin gene clusters from Prochloron spp. is hypervariable. Hence, a library of cyanobactins is synthesized when multiple uncultivated Prochloron strains are present in the tunicate (Donia et al. 2006).

Patellins and trunkamides produced by uncultured symbiotic Prochloron spp. contain prenylated amino acids (Donia et al. 2008a). The prenylated cyanobactins do not have oxazoles or thiazoles, only unoxidized forms of these heterocyclized amino acids occur, and the oxidase domain of PatG is missing in the patellin and trunkamide biosynthetic gene cluster (Schmidt and Donia 2009). In addition, the prenylating pathway encodes two PatF proteins whose roles are unclear, and the prenylation is not fully understood at present (Schmidt and Donia 2009). Ulithiacyclamides contain four cysteines, two of which form a disulfide bridge and the other two cysteines are heterocyclized to thiazoles (Table 1). Among the more than hundred cyanobactins, disulfide bridges are only found in ulithiacyclamides (Tables 1, 2, and 3).

Tenuecyclamides

Tenuecyclamides are hexapeptides containing two thiazoles and an oxazole (Banker and Carmeli 1998, Table 2). A cyanobactin pathway for the biosynthesis of tenuecyclamide was characterized in an epilithic Nostoc spongiaeforme strain (Fig. 2, Donia et al. 2008a). The tenuecyclamide ten gene cluster has an identical gene organization to pat gene cluster as the genes are transcribed to the same direction, and the gene order is conserved. The TenE precursor peptide encodes four copies of the core peptide amino acid sequences which form tenuecyclamides A and C (Donia et al. 2008a). Tenuecyclamides A and B differ only in stereochemistry (Table 2).

Trichamide

Trichamide is cyclic 11 amino acid cyanobactin which contains two thiazoles (Sudek et al. 2006). This cyanobactin was discovered through genome mining of Trichodesmium erythraeum IMS101, a free-living, nitrogen-fixing filamentous marine cyanobacterium (Schmidt et al. 2005; Sudek et al. 2006). The tri gene cluster is 12.5 kb and encodes 11 open reading frames (ORFs) with a bidirectional gene order for which only six genes have an assigned function (Fig. 2, Sudek et al. 2006). The trichamide precursor peptide encodes a single copy of the trichamide core peptide flanked by the putative recognition sequences. In contrast to Pat proteins, the oxidase and the protease domains in Trichodesmium are encoded by separate genes (Sudek et al. 2006).

Lyngbyabactins

A cyanobactin gene cluster has been also identified from the genome of the marine Lyngbya aestuarii CCY9616 (Fig. 2, Donia et al. 2008a). The products of this gene cluster, lyngbyabactins A and B, were predicted from the precursor peptide LynE (Donia et al. 2008a). The peptides are predicted to include isoprenoid amino acid derivatives (Donia et al. 2008a). However, lyngbyabactins A and B have not yet been detected from L. aestuarii CCY9616. The lyngbyabactin biosynthetic gene cluster has homologs of all genes present in the pat gene cluster but it also contains five other open reading frames which have no function assigned (Donia et al. 2008a).

Microcyclamides and aerucyclamides

Microcyclamide is a cytotoxic cyclic hexapeptide reported from the freshwater bloom-forming cyanobacterium Microcystis aeruginosa NIES-298 (Ishida et al. 2000). The microcyclamides 7806A and B of M. aeruginosa PCC 7806 were later renamed as aerucyclamides, and the structures were revised (Portmann et al. 2008a, b). Microcyclamides and aerucyclamides are assembled via the cyanobactin biosynthetic pathway in two strains of M. aeruginosa NIES 298 and PCC 7806 (Ziemert et al. 2008b; Portmann et al. 2008a, b). The mca gene clusters have the same gene order as the pat genes except for two additional open reading frames for which no functions could be assigned (Fig. 2, Ziemert et al. 2008b). The McaE precursor peptide encodes two identical copies of the microcyclamide core peptides in M. aeruginosa NIES-298 (Ziemert et al. 2008b).

Anacyclamides

A variety of anacyclamides have been identified from strains of the genus Anabaena (Leikoski et al. 2010, Table 3). Anacyclamides consist of proteinogenic amino acids, and some contain prenyl or geranyl groups through the posttranslational modifications of specific amino acids. The acy gene cluster in Anabaena sp. 90 encodes 11 ORFs, and it is arranged in an eleven kb operon which is bidirectionally transcribed as the trichamide gene cluster (Fig. 2). The gene cluster differs from pat gene cluster in that there were no homologs for all pat genes. Additionally, there were hypothetical ORFs present in the acy gene cluster that were absent in the pat gene cluster (Leikoski et al. 2010). The precursor peptide AcyE encodes single copy of the anacyclamide A10 (Fig. 1) flanked by putative recognition sequences which differ substantially from other cyanobactin precursors (Leikoski et al. 2010). The anacyclamide gene cluster lacks a PatD homolog which agrees well with the anacyclamide structure which has no posttranslationally heterocyclized amino acids (Leikoski et al. 2010). The AcyG protein lacks an oxidase domain which is consistent with the absence of heterocyclized amino acids (Leikoski et al. 2010). In anacyclamides, the length of the peptides varies greatly which is achieved by expansion of the AcyE precursor protein. Anacyclamides showed great amino acid variation since only one proline was conserved (Table 3).

Bioactivities

The cyanobactins and cyclic peptides with analogous structures have various reported bioactivities (in detail, see Tables 1, 2, and 3). The diverse bioactivities are derived from versatile structures, but all cyanobactins do not exhibit bioactivities in the tests used or those have not been studied yet (Tables 1, 2, and 3). Several of the cyanobacterial metabolites have been found to be anticancer compounds, e.g., trunkamide (Salvatella et al. 2003); some have multidrug reversing activities (Ogino et al. 1996) as well as activities against tropical parasites such as malaria-causing Plasmodium falciparum (Linington et al. 2007; Portmann et al. 2008b). In bacteria, many of the ribosomally produced peptides are antibiotics or bactericides produced to kill or inhibit growth of competing microbes (Nolan and Walsh 2009). Cyanobacteria are photosynthetic autotrophic organisms grazed by eukaryotic organisms. They are believed to form mutualistic association with heterotrophic bacteria, but this may not be always the case (Manage et al. 2000; Berg et al. 2009). Nostocyclamides for example have been shown to contain anticyanobacterial activity (Todorova et al. 1995; Jüttner et al. 2001). There are compounds isolated from cyanobacteria with antibiotic (Ishida et al. 1997) or antiviral (Boyd et al. 1997; Bokesch et al. 2003) effects, leaving open the option that some of the cyanobactins may prove to be antibiotic or antiviral compounds.

Analogous pathways

In addition to cyanobactins, another cyanobacterial peptide class, microviridins, was recently shown to be ribosomally produced in M. aeruginosa and Planktothrix agardhii, but their biosynthetic machinery differs from that of cyanobactins (Ziemert et al. 2008a; Philmus et al. 2008). These compounds were originally thought to be products of nonribosomal peptide biosynthesis. However, microviridins are synthesized from precursor peptides that are converted into tricyclic depsipeptides through the action of ATP grasp ligases and a transporter peptidase (Ziemert et al. 2008a; Philmus et al. 2008). The work of Philmus et al. (2008) reported similar gene clusters in the genomes of Anabaena variabilis, Nostoc punctiforme, and Nodularia spumigena as well as in genomes of other bacteria.

The biosynthetic gene clusters encoding the production pathway of ribosomal peptides with oxazoles and thiazoles are present in a broad range of bacteria (Lee et al. 2008). Bacteria distantly related to cyanobacteria are known to produce bacteriocins by the posttranslational modification of gene-encoded precursor peptides, including microcins and the lanthionine-containing lantibiotics of Gram-positive bacteria (Jack et al. 1995; Jack and Jung 2000; Nolan and Walsh 2009). Cyanobactin biosynthesis is analogous in many ways to the biosynthesis of bacteriocins (Franz et al. 2007; Nolan and Walsh 2009). The leader-peptide-guided biosynthesis is common in many ribosomally synthesized natural products where the precursor peptide is synthesized and cleaved, and in some cases the core peptide is posttranslationally modified (Oman and van der Donk 2010). Bacteriocins can be also circularized as cyanobactins, and some bacteriocins have similar posttranslational modifications as cyanobactins for example thiazoles and oxazoles (Jack and Jung 2000; Maqueda et al. 2008; Martin-Visscher et al. 2009).

Many of cyanobacterial bioactive compound classes are synthesized on nonribosomal peptide synthetases (NRPS) or combined NRPS and polyketide synthases (Welker and von Döhren 2006; Sivonen and Börner 2008). However, the cyanobactins have been shown to be produced by the posttranslational modification of the gene-encoded precursor peptides (e.g., Schmidt et al. 2005; Donia et al. 2008a). The gene clusters responsible for ribosomal peptide production are small compared to the large nonribosomal peptide synthetase gene clusters. In NRPS, variation in the chemical structure of the peptide is achieved by utilization of more than 200 nonproteinogenic amino acids (Nolan and Walsh 2009) whereas ribosomal peptides are restricted to 20 proteinogenic amino acids which may be posttranslationally modified. In NRPS, the enzymes seem to have relaxed substrate specificity and thus allow simultaneous production of a number of structural variants in the same strain of a cyanobacterium (Welker and von Döhren 2006).

Biotechnological aspects

The biotechnological exploitation of cyanobactins will require detailed studies on the enzymes involved in the biosynthesis as well as mechanisms of action of these peptides. It is possible to express the cyanobactin gene clusters in heterologous hosts (Schmidt et al. 2005; Donia et al. 2006, 2008a; Leikoski et al. 2010). The small size of the cyanobactin gene clusters and the expression of the entire clusters in heterologous hosts will provide new possibilities to create compound libraries and novel compounds (Donia et al. 2006, 2008a). In the cyanobactin pathway, heterologous expression gives options to study the role of individual genes in biosynthesis as well as produce novel peptides. The cyanobactin pathway was utilized in E. coli to synthesize an engineered peptide eptidemnamide, a cyclic peptide similar to an anticoagulant in clinical use (Donia et al. 2006). This approach demonstrates a means to exploit cyanobacterial pathways and produce novel compounds by the rational design of peptides. The peptide-precursor-directed synthesis allows manipulations directly to the precursor gene and enables production of engineered peptides in heterologous hosts (Oman and van der Donk 2010). In addition, the enzymes in the cyanobactin pathways could be used as catalysts in aiding chemical synthesis of the desired compounds. The work by Lee et al. (2009) not only clarified the role of proteases in the cyanobactin biosynthesis but demonstrated the potential of the enzyme as general catalysts for cyclization of peptides. The technological advantages of the PatG protease were that no energy is required for the cleavage and cyclization, and also the protease was proven to be tolerant of different substrate lengths and sequences as long as the C-terminal recognition sequence was present (Lee et al. 2009). This is important as in synthetic peptide manufacture the head-to-tail cyclization step restricts peptide production in bulk amounts.

It should be noted that whole-genome information has already led to the discovery of cyanobactin biosynthesis as well as several new compounds and compound classes, e.g., patellamides (Schmidt et al. 2005), trichamide (Sudek et al. 2006), and anacyclamides (Leikoski et al. 2010). The increasing number of genome projects on cyanobacteria and metagenomic studies (Schmidt and Donia 2009) applied to various environments are likely to yield new discoveries including diverse ribosomal pathways and novel cyanobactins in the future.