In N. spumigena, a total of 18 spumigins were detected, including nine new congeners of the compounds (Table 1, 1–4, 12, 13, 15, 17, 18). The elucidation of spumigin structures was based on the fragmentation pattern in mass spectra described by [3,12,27]. All spumigins reported so far have Hpla as the N-terminal unit [3,27]. Also, in the majority of spumigins detected in this work, Hpla¹ was present. However, in compounds 1–4, instead of an ion at m/z 208 [HOCHCO-Hty + H − CO], a fragment ion at m/z 250 was recorded (Figure 1).

The shift in mass (42 Da) was also observed in other fragment ions containing the N-terminal unit. This mass difference may indicate the presence of an acetyl group in the structure of Hpla. Such a modification was found in spumigins produced by three strains from the Gulf of Gdańsk (CCNP1401, B15a and CCNP1403). In these isolates, the modified spumigins belonged to the most abundant peptides (Table 1).

The second position in the spumigins structure was the least conserved. It was mainly occupied by Hty² (13/18), but in individual peptides, homophenylalanine Hph² (11 and 14), Leu² (15 and 16) or Tyr² (12) were also incorporated. Spumigin H (14) with Leu² and spumigin I (16) with Hph² were found only in N. spumigena bloom samples from the Baltic Sea (Table 1). In position 3 of the spumigins, the presence of Pro³ (11/18) or MePro³ (7/18) was deduced based on the enhanced ion product (EIP) spectra with immonium ions (at m/z 70 or 84) and a series of other ions containing Pro or MePro. This result confirmed a preferential activation of Pro (or MePro) by the catalytic domain of the enzyme responsible for the incorporation of the third residue during the biosynthesis of spumigins in N. spumigena CCY9414 [3].

The C-terminal position of the spumigins was occupied by Arg (1, 5, 7), its decarboxylated form called agmatine (Agm: 13), the alcohol form (Argol: 3, 6, 9, 16) or the aldehyde form (Argal: 2, 4, 8, 10–12, 14, 15). Spumigins with Argal were the most common and belonged to the most abundant congeners (Table 1). In the chromatograms, they occurred as broad, tailing and/or multiple peaks, indicating the presence of different isomers. The high prevalence of Argal in the C-terminal position confirmed the previous results on the structure of the multifunctional enzyme complex involved in the biosynthesis of spumigins in N. spumigena CCY9414. Fewer et al. [3] revealed that the module of the complex responsible for the incorporation of the C-terminal unit, apart from the Arg-activating A domain, contains a reductase domain, which leads to the production of spumigins as peptide aldehydes.

In the EIP spectrum of compound 18 with a molecular ion at m/z 457, the presence of ions at m/z 342 [Hpla-Hty + H], 439 [Hpla-Hty-Pro + H] was indicative of the spumigin structure (see Supplementary Information). This spumigin congener was found in five of the Baltic isolates. As the mass difference between the [M + H]⁺ ion of compound 18 and [Hpla-Hty-Pro + H] was calculated to be 18 Da, we proposed that the structure of compound 18 is Hpla-Hty-Pro-OH. The other tripeptide with a molecular ion at m/z 470 (17) and fragment ions at m/z 342 [Hpla-Hty + H], 453 [Hpla-Hty-MePro + H] and 84 (MePro-immonium ion), was identified as Hpla-Hty-MePro-NH₂. In our work, compound 17 was found in the cyanobacteria isolated from Lake Iznik and in a bloom sample from the Baltic Sea (Table 1). The tripeptides of general structure Hpla-Leu-Choi-NH₂, aeruginosin 298B [16] and aeruginosin EI461 [31] were identified in Microcystis aeruginosa and suggested as biosynthetic intermediates of the respective tetrapeptides.

Aeruginosins are composed of Choi³ (2-carboxy-6-hydroxyoctahydroindole) and Arg⁴ or its mimetics [1,2,16,32,33]. In the second position of aeruginosins characterized in previous studies, variable amino acids were found: Tyr, Hty, Phe, Leu or Ile [1]. In most of the published structures of the peptides, the N-terminal position was occupied by Hpla¹ [16]. Pla (phenyl lactic acid) in position 1, though rarely reported, was found in oscillarin, one of the most potent thrombine inhibitors among aeruginosin analogues [34]. Additionally, chlorinated and/or sulphated aeruginosins (at both Hpla and Choi) as well as compounds with a pentose sugar were detected [13,14,16,23,35].

In our studies, production of aeruginosins by N. spumigena isolates, regardless of their origin, was revealed for the first time. The partial reconstruction of the amino acid sequence in the four detected variants (Table 1, 19–22) was conducted with the support of the data published by [13,14,35,36], among others. Like all spumigins identified in N. spumigena cells, the peptides did not contain any chloride or sulphate groups.

As in spumigins, the C-terminal position of aeruginosins detected in this study was occupied by Arg (19), Argol (20), Argal (21) or Agm (22) (Table 1). Under the MS/MS conditions, the EIP spectra of aeruginosins were significantly affected by the unit. When Argal was incorporated in the C-terminal position, the ions at m/z 221 and 266 belonged to the most intense ions in the sequence (Figure 2).

In the case of aeruginosins with Arg, Argol and Agm, the other two ions corresponding to fragments [Choi-Arg/Argol/Agm + H] and [Choi-Arg/Argol/Agm + H − NH₂] were more abundant (see Supplementary Information). Compound 22 with Agm was found in all Australian isolates and the strains from Lake Iznik in Turkey. Similarly to Agm-containing spumigins, compound 22 was not present in either the Baltic bloom samples or the Baltic N. spumigena isolates. This finding might indicate the differences in the non-ribosomal peptide synthetases among N. spumigena populations from different regions.

In N. spumigena, two classes of cyclic NRPs were identified: the toxic pentapeptides called nodularins (23 and 24) and the hexapeptides classified as anabaenopeptins (25–47). The hepatotoxic nodularins belong to the most widely studied cyanobacterial metabolites [37]. With the exception of N. spumigena NSBL-05 and NSBL-06, the peak of NOD was one of the most intense in the obtained chromatograms. The lack of NOD and its demethylated form [D-Asp³]NOD in the two Australian strains (NSBL-05 and NSBL-06) is in agreement with the results of Bolch et al. [38] and Moffitt et al. [39], who found these strains to be non-toxic. Out of the 23 anabaenopeptins, twelve variants have been reported in this work for the first time (25, 26, 30, 37–41, 43–45, 47). Their structures were elucidated using the published data on anabaenopeptins mass fragmentation spectra [4,12,40,41,42,43]. As in all other compounds classified as anabaenopeptins, homo amino acids (Hph and Hty) were present in position 4, and N-methylated amino acids (methylhomotyrosine MeHty, methylhomophenylalanine MeHph and methylalanine MeAla) in position 5 (Figure 3).

In position 3, Met or its oxidized forms (methionine sulfoxide MetO and methionine sulfone MetO₂), were the most common residues (12/23). In other compounds the position was occupied by Val (7/23) or Ile/Leu (4/23). Position 6 was the least conserved and, as in all reported nodulapeptins, it contained Ser or its acetylated form AcSer (O-acetyl Ser) (12/23), or Met or its oxidized form MetO (8/23). Additionally, we revealed the production of three anabaenopeptins with Phe in position 6 (Figure 3 and Figure 4). In the side chain unit, either Phe¹ (Figure 3) or Ile/Leu¹ (Figure 4) was incorporated. In N. spumigena strain KAC66, production of nodulapeptins 899 (33), 901 (32) and 917 (28) with Phe in position 1 was reported by Schumacher et al. [43]. These peptides and three other nodulapeptins (25, 30 and 38), all containing Phe¹ in their backbone chain, were also detected in KAC66 in our study.

Based on the mass spectrometrically elucidated structures studied, it can be concluded that individual strains of N. spumigena produced anabaenopeptins with only one kind of amino acid in the exocyclic position, i.e., either Ile/Leu or Phe. This result is in agreement with the data published by Rouhiainen et al. [4] who found only one starter module in the anabaenopeptin synthetase gene cluster (apt) of N. spumigena CCY9414. The A domain of the module probably activates only Ile or only Phe. In contrast, the apt gene clusters in Anabaena encode two alternative starter modules organized in separate bimodular proteins [4]. Such organization of anabaenopeptin gene clusters or the substrate promiscuity of the A domain [44], lead to the co-production of anabaenopeptins with more than one unit in the exocyclic position 1.

According to the co linearity rule, the order and the number of the modules in the NRPS correspond to the number and sequence of the units in the biosynthesized peptide [9]. Therefore, based on the organization of the NRPS, and the substrate specificity of adenylation domains, the structure and diversity of the peptide products can be predicted. The reverse process is also possible. The analyses of the whole variety of non-ribosomal peptide congeners produced by strains belonging to the same species can be a valuable source of information about the structure of the NRPS enzyme complex. The chemical structures of spumigins and anabaenopeptins elucidated in 20 strains of N. spumigena from the Baltic Sea, coastal waters of southern Australia and freshwater Lake Iznik corresponded well with the published data on the organization of their respective gene clusters (spu and apt) and the non-ribosomal peptide synthetases enzyme complexes in N. spumigena CCY9414 [3,4].

Despite the differences in the phenotypic features of individual N. spumigena strains, there is a high percentage of similarity in their 16S rRNA gene sequences [39,45]. The genetic studies by Hayes and Barker [46] and Moffitt et al. [39] showed a lack of delineation between N. spumigena from geographically distant regions. A contradictory hypothesis was presented by Bolch et al. [38] based on the analysis of the phycocyanin intergenic spacer region (cpcBA-IGS). This highly variable region is thought to be a useful molecular marker for discrimination of cyanobacterial strains. For all Australian isolates, including the non-toxic NSBL, the cpcBA-IGS sequences were identical, but different from the Baltic populations [38]. The genetic difference between the Baltic and Australian strains was also documented by Lehtimäki et al. [45] using the RFLP (restriction fragment polymorphism) of 16S rRNA genes and other molecular techniques.

The grouping of N. spumigena strains based on the analyses of peptide profiles appears to be consistent with the genetic diversity of the cyanobacterium as determined by analysis of the nucleotide sequences of the cpcBA-IGS fragments. The principal component analyses of 50 peptides in the 20 N. spumigena strains placed all the Australian isolates in one group (Figure 5). This group included both nodularin-producing and “non-toxic” (NSBL) N. spumigena strains, and was distantly related to the strains from other geographical regions.

In addition, both the chemical analyses performed in this study and the published data on the PC-IGS sequences [38,45,47] indicated the existence of several distinct types of the Baltic N. spumigena. On the factorial plane (Figure 5), the Baltic strains were distributed widely and represented different chemotypes. Two of them, BY1 and CCNP1402, were classified to one chemotype together with N. spumigena from the freshwater Lake Iznik in Turkey (Node2, Nodg3 and Nodh2) (Figure 5). The fact that strains from such diverse environments and distinct geographical regions showed similar peptide patterns indicates that environmental conditions, or at least salinity, have no effect on the production of the metabolites.

Significant variations in the number of cyanobacterial chemotypes among different reservoirs have been reported for Planktothrix agardhii [13,48,49,50] and Microcystis aeruginosa [14,51,52]. So far, the mechanisms regulating the number and composition of cyanobacterial chemotypes in aquatic ecosystems have not been recognized. It was postulated that the diversification of oligopeptide profiles is driven by horizontal gene transfer, recombination events (especially within the adenylation domains) or gene deletion [4,32,53,54,55]. The significance of natural selection affected by environmental pressure has also been taken into account.

Despite extensive studies on cyanobacterial NRPs, the ecological function of the compounds is still ambiguous. The peptides were suggested to play some role in competition with eukaryotic algae [56], or as defense agents against grazers [18,19]. Alternatively, they might belong to signaling molecules in quorum sensing mechanisms [57]. The potential role of the compounds in cyanobacteria bloom termination through viral lysis [58] or in the interaction with chytrid fungi infecting cyanobacteria [59] was also considered.

N. spumigena appeared to be an effective producer of NRPs. Several classes of the peptides, including those analyzed in our study, inhibit the activity of important metabolic enzymes and are considered to be potential therapeutic agents. Fewer et al. [3] revealed a strong (nM) trypsin inhibition of spumigin E (8). Aeruginosins were revealed to exhibit high in vitro inhibitory activity against blood coagulation factors (thrombin, trypsin, Factor VIIa and Factor IXa) [16,21,22]. The activity of anabaenopeptins towards protein phosphatases [60] and proteolytic enzymes such as trypsin, chymotrypsin, elastase and carboxypeptidase [61,62,63] was also demonstrated.

Twenty strains of N. spumigena were analyzed. Seven of them were isolated from the Gulf of Gdańsk, southern Baltic, during the summer bloom of cyanobacteria: CCNP1401, CCNP1403 and B15a in 1997, CCNP1402 in 2005 and CCNP1423, CCNP1424 and CCNP1425 in 2011. Two other Baltic strains were obtained from culture collections: KAC66 isolated from Landsort Deep in 1996 (Kalmar Algae Collection, University of Kalmar, Kalmar, Sweden) and BY1 isolated from Arkona Sea in 1986 (Department of Applied Chemistry and Microbiology, Helsinki University, Helsinki, Finland). Three N. spumigena strains, Node2, Nodg3 and Nodh2, were isolated in 2005 from the freshwater Lake Iznik located in the southeast of the Marmara region in Turkey. Eight Australian strains were provided by the CSIRO Marine Laboratories (Hobart, Tasmania, Australia): NSBL-05 and NSBL-06 (Lake Bullenmerri, Victoria, Australia, 1993), NSGL-01 (Gippsland Lake, Victoria, Australia,1993), NSLA-01 (Lake Alexandrina, South Australia, Australia, 1990), NSOR-01 (Orielton Lagoon, Tasmania, Australia, 1993), NSKR-07 (Kalgan River Estuary, Western Australia, Australia, 1995), NSBR-02 (Blackwood River, Western Australia, Australia, 1994), and NSPH-02 (Peel-Harvey Inlet, Western Australia, Australia, 1992).

Cyanobacteria were grown in a Z8 medium (without nitrogen) prepared using MilliQ water (PF system, Millipore, Molsheim, France) with NaCl added to produce a final salinity of 7 psu. Cultures were incubated for approximately four weeks in growth chambers at 22 °C and irradiance of 20 μmol photons m⁻² s⁻¹ with a 16:8 light:dark cycle. Cyanobacteria were harvested when the stationary phase of growth was reached and the optical density of cell suspension was about 0.230 (at 750 nm). Additionally, field samples were collected during blooms of N. spumigena in the Gulf of Gdańsk on 30 June 2011 and 3 July 2012. The analyzed N. spumigena strains possessed discoid vegetative cells (mean width 5.22–7.05 μm; mean length 2.87–4.68 μm), with gas vesicles. Heterocytes were more spherical (mean width 6.38–8.12 μm; mean length 4.40–7.67 μm), and occurred predominantly in intercalary positions in straight or curved filaments.

N. spumigena cultures (30 mL) or bloom samples (100 mL) were filtered onto GF/C glass-fiber filters (Whatman). The material was extracted with 2.0 mL of 5% acetic acid in MilliQ water by a 1 min probe sonication with an ultrasonic disrupter (HD 2070 Sonopuls, Bandeline, Berlin, Germany) followed by a 15 min bath sonication (Sonorex, Bandeline, Berlin, Germany). The samples were then centrifuged at 10,000 g for 15 min and the obtained supernatants were subjected to LC-MS/MS analysis.

N. spumigena cell extracts were first screened for peptides using a nano-LC (nanoAquity, Waters, Milford, MA, USA) coupled to a QStar Elite hybrid quadrupole-time-of-flight tandem mass spectrometry system (Q-TOF MS/MS, Applied Biosystems MDS Sciex, Concord, ON, Canada). The analyses were conducted using a NanoSpray II source and Heated Interface (150 °C) on the QSTAR Elite LC-MS/MS. Chromatography was performed with 5% acetonitrile in MilliQ water plus 0.1% formic acid (solvent A) and 0.1% formic acid in acetonitrile (solvent B). Samples (0.5 μL) were injected on a nanoAquity UPLC Symmetry C18 trapping column (180 μm × 20 mm, 5 μm, Waters, Milford, MA, USA) and washed with solvent A at a flow rate of 15 μL min⁻¹ for 2 min. Peptides were separated on a nanoAquity UPLC BEH130 C18 column (75 μm × 100 mm, 1.7 μm, Waters, Milford, MA, USA). The gradient started at 5% B and went to 30% B within 30 min at a flow rate of 0.4 μL min⁻¹. The content of phase B was then increased to 99% within the next 15 min and kept at that level for 10 min before returning to the starting conditions. Compounds eluting from the column were ionized using a PicoTip emitter (10 μm tip I.D.; New Objective, Woburn, MA, USA). MS spectra were acquired over the range 350–2000 Da. The instrument was operated in the positive mode. Ionspray voltage was 2.4 kV, with the nebulizing gas nitrogen pressure and curtain gas nitrogen pressures set at 25 p.s.i and 20 p.s.i., respectively (1 p.s.i. = 6894.76 Pa). When the ion product (IP) mode was used, the collision energy (CE) was 55 eV and mass range 50–1000 Da. The system was calibrated with a solution containing synthetic peptide ALILTLVS and CsJ (5 × 10⁻⁷ M) in 50% aqueous methanol plus 0.1% formic acid. Data acquisition and processing were accomplished with the Analyst^® QS 2.0 software.

Structural analyses of selected cyanobacterial peptides were also performed using Agilent 1200 (Agilent Technologies, Waldboronn, Germany) coupled online to a hybrid triple quadrupole/linear ion trap mass spectrometer (QTRAP5500, Applied Biosystems, Sciex; Concorde, ON, Canada). As a mobile phase a mixture of A (5% acetonitrile in MilliQ water plus 0.1% formic acid) and B (0.1% formic acid in acetonitrile) was used. Separation was performed on a Zorbax Eclipse XDB-C18 column (4.6 × 150 mm; 5 μm) (Agilent Technologies, Santa Clara, CA, USA). Phase B was linearly increased from 15% to 75% in 5 min and then to 90% in the next 5 min. This composition of the mobile phase was held for 5 min and brought back to 15% B in 1 min. The column oven temperature was 35 °C, the flow rate was 0.6 mL·min⁻¹ and the injection volume was 5 μL. Turbo ion spray (550 °C) voltage was 5.5 kV, with the nebulizer gas pressure and curtain gas pressures set at 60 p.s.i. and 20 p.s.i., respectively.

To characterize the structure of cyanobacterial peptides with the QTRAP LC-MS/MS system, the experiments were run using the information dependent acquisition method (IDA) and in enhanced ion product mode (EIP). In EIP mode, the ions fragmented in the collision cell (Q2) were captured in the ion trap and then scanned. In the IDA method, Q3 survey scans were used to automatically trigger an EIP scan if the signal was above a threshold of 100,000 cps. EPI spectra were acquired from 50 to 1000 Da with a scan speed of 2000 Da s⁻¹ and a collision energy (CE) of 45 V with collision energy spread (CES) of 20 V. The linear ion trap fill time was 50 msec. The dynamic exclusion was activated to minimize the risk of missing the co-eluting compounds. Data acquisition and processing were accomplished using Analyst QS^® 1.5.1 software.

LC-MS/MS analyses were performed for each strain at least five times using extracts obtained from different culture experiments conducted during two years of the study.

The principal component analysis (PCA) ordination was performed to study the differences in peptide pattern among 20 N. spumigena strains from various geographical regions. PCA was generated using the multivariate data analysis software STATISTICA v.9 (StatSoft, Tulsa, OK, USA). In the analyses, all detected peptides (n = 50) and the relative intensities of respective mass signals were included.