Two abundant peptides with a mass of m/z 587 and m/z 589 were identified from cell extracts of Nodularia spumigena AV1 by LC-MS analysis. They were initially suspected to be new variants of spumigin based on their mass and chromatographic behavior (Figure S1 in File S1). However, fragmentation of the protonated ions m/z 587 and m/z 589 did not produce sufficient information for overall substructure elucidation and only the presence of argininal and argininol could be postulated (Figures S1–S2 in File S1). Surprisingly, MDA and DNPH derivatization of the compounds and subsequent MS² data suggested that the two peptides contained but differed by the presence of alcohol and aldehyde versions of arginine (Figure S3 in File S1). The moiety is a unique component of the aeruginosin family of linear peptides and we hypothesized that N. spumigena make members of the aeruginosin family of peptides.

A 17.6 kb aeruginosin (aer) gene cluster was subsequently identified on a single contig in the genome of N. spumigena CCY9414 (GenBank accession number CM001793) through tBLASTn searches using AerD, AerE and AerF protein sequences involved in the synthesis of the Choi moiety (Figure 2). The aer gene cluster was located 133 kb apart from the spumigin gene cluster (Figure 2), which was located just 11 kb from the nodulapeptin gene cluster (Figure 2).

The aer gene cluster encodes 8 proteins organized in a single operon (Figure 3; Table 1). The predicted substrate specificities of the AerM, AerB and AerG peptide synthetases were L-Arg, L-Tyr and Choi through comparison with other aeruginosin biosynthetic pathways (Table 2). Aeruginosin biosynthesis was predicted to start by loading short-chain fatty acids using the AerB condensation domain, as in a number of other non-ribosomal biosynthetic pathways (Figure 3). In order to test this we performed phylogenetic analyses to assign the condensation domains of AerB, AerM and AerG to different condensation domain subtypes (Figure 4). Phylogenetic analysis of the AerB condensation domain showed a close relationship between the loading condensation domains of the nostopeptolide and cyanopeptolin biosynthetic pathways (Figure 4). Genes encoding the AerD, AerE and AerF enzymes were also located in the aer gene cluster (Figure 3) as was a gene encoding a putative glycosyltransferase, AerI (Table 1).

Preliminary MS² fragmentation was not sufficient to resolve the first two sub-structural elements and suggested that the second amino acid could be Leu or Tyr while bioinformatic analyses suggested that the substrate could be L-Tyr (Table 2). In order to gain more information on the substructure of the aeruginosins an ATP-PPi exchange assay was performed which demonstrated that L-Tyr was activated by the AerB adenylation domain in vitro (Figure 5). The presence of an epimerase domain suggested that the substrate of AerB was L-Tyr, which would be epimerized to D-Tyr as found in other aeuginosins.

The main aeruginosin variant (m/z 587) was hydrolyzed in order to confirm these biochemical and bioinformatic predictions and to obtain further information on the structure of the new peptides. However, despite the presence of a full-length and apparently functional epimerase domain in AerB, the second amino acid was unambiguously determined to be L-Tyr according to chromatographic amino acid analysis of aeruginosins subjected to acid hydrolysis. The amino acids found at this position in other aeruginosins are almost exclusively D-amino acids. Reanalysis of the main aeruginosin (m/z 589) product ion spectrum also showed the presence of Tyr in position 2 (Figure S4 in File S1). The third amino acid, Choi, was in L configuration based to the configuration of Choi in aeruginosin 298-A. The N-terminal moiety could not be detected by amino acid analysis. However, GC-MS analysis of the hydrolyzed aeruginosin allowed unequivocal identification of the N-terminal moiety as hexanoic acid based on retention time and spectra (Figure 6).

The complete structure of the main aeruginosin (m/z 587) was confirmed by NMR analysis. Separation and purification of this peptide from the other aeruginosins and spumigins produced by the N. spumigena AV1 strain was hindered by the reactive aldehydic nature of the compound (Figure S1 in File S1). NaBH₄ reduction of the peptides in the methanol extract converted the aldehyde to an alcohol which made it possible to purify reduced aeruginosin by HPLC. ¹H and ¹³C NMR signals yielded four partial structures confirming the aeruginosin structure (Figure 1A). The NMR spectral data are presented in the supplementary material (Table S1 in File S1; Figures S5–S7 in File S1). Accurate mass measurement of the protonated molecules together with the ¹⁵N-labeling of the aeruginosins NAL2 and NOL3 were in full agreement with the other results.

Detailed inspection of N. spumigena AV1 and CH307 strains allowed the identification of 11 structural variants of aeruginosins (Table 3; Table S2 in File S1). These could be divided into aeruginosins containing either an aldehyde (NAL1-NAL4) or alcohol (NOL1-NOL7) functionality. A range of short straight-chained carboxylic acids were found at the N-terminus (Table 1). Approximately 83% of the variants contained hexanoic acid in AV1 while 93% of the variants contained octanoic acid in CH307 but both strains produced small amounts of aeruginosins with shorter and longer chained carboxylic acids at this position. An O-linked pentose was identified in 4 of the 11 aeruginosins and located on the Choi moiety (Figure 1B, Figure S8 in File S1). Inspection of 16 strains of N. spumigena revealed just a single strain which lacked detectable levels of aeruginosins (Figure 7 and Table S3 in File S1). Glycosylated aeruginosins could be detected in just 3 of the 16 strains (Table S3 in File S1). Three strains, CH307, CH311 and P38, produced glycosylated aeruginosins, of which CH307 produced variants containing octanoic acid as the main fatty acid.

The presence of aeruginosins as well as the distribution of genes encoding AerM, AerB, AerG, and AerI was determined in 16 strains of N. spumigena, 4 strains of N. harveyana, and 8 strains of N. sphaerocarpa (Table S3 in File S1). All 16 strains of N. spumigena contained aeruginosin biosynthetic genes and 15 of these strains produced aeruginosins (Figure 7). Aeruginosins were not found from N. spumigena AV45, which seems to encode only a partial aer gene cluster. All strains of N. spumigena from the Baltic Sea contained aer clusters encoding the AerI putative glycosyltransferase. The gene encoding this enzyme was not detected in any of the strains isolated from Australia. Neither aeruginosins nor aeruginosin biosynthetic genes could be found in any strains of the benthic N. harveyana or N. sphaerocarpa tested (Table S3 in File S1). The majority of the 16 strains of N. spumigena produced both spumigin and aeruginosin (Figure 7). Some strains produced more aeruginosins and spumigins than others (Figure 7). About half of the 16 strains produced mPro containing spumigins while the remainder produced spumigins containing Pro. Using UV (280 nm) to quantify the amount of aeruginosins in the AV1 strain demonstrated that spumigins comprise 7 ‰ and aeruginosins 3–4 ‰ of the dry weight. All 16 strains of Nodularia spumigena produced spumigins and the nodularin toxin. Almost all of the 16 strains of Nodularia spumigena produced nodulapeptins. Suomilide was identified only in N. sphaerocarpa strains (Table S3 in File S1).

Sixteen strains of N. spumigena, 4 strains of N. harveyana and 8 strains of N. sphaerocarpa (Table S3 in File S1) were grown at a photon irradiance of 15 µmol m⁻² s⁻¹ in saline Z8 medium lacking a source of combined nitrogen for 21 days [8]. ¹⁵N-labeling of N. spumigena AV1 was performed in similar way, except that medium was buffered with 10 mM HEPES (pH 8.0). ¹⁵N-urea (98+ % ¹⁵N, ISOTEC, USA) was used as nitrogen source and nitrogen-free argon (with 20.9% O₂ and 0.45% CO₂; quality 5.7; AGA Gas Ab, Sweden) was bubbled into the medium to prevent nitrogen fixation from air.

The aer gene cluster was identified on a single 5,462,271 bp scaffold in the genome of N. spumigena CCY9414 (GenBank accession number CM001793) through BLASTp searches using AerD, AerE and AerF proteins. The genes in the aer gene cluster were predicted with Artemis using Glimmer. The starting sites were refined manually. The amino acid sequences of the genes were used to query the non-redundant database at NCBI in order to predict a function for the genes (Table 2). The substrate specificity of the activated adenylation domains in the NRPS modules was predicted by using the 10 amino acid binding pocket signature [26].

Genomic DNA was extracted from the cultivated strains as previously described [8]. We amplified four genes from the aer gene cluster, aerM, aerB, aerG and aerI, by PCR using oligonucleotide primers designed from the N. spumigena CCY9414 genome sequence (Table S4 in File S1). The PCR reactions were performed in a 20 µl final volume containing 1 µl of DNA, 1× DyNAzyme II PCR buffer, 100 mM of each deoxynucleotide, 0.4 mM of each oligonucleotide primer, and 0.4 units of DyNAzyme II DNA polymerase (Finnzymes, Espoo, Finland). The following protocol was used: 94°C, 3 min; 25 cycles of 94°C, 30 s; 63°C, 30 s; 72°C, 1 min; and 72°C, 10 min. PCR to confirm the deletion of the aerI gene was performed as before but with an annealing temperature of 58°C. PCR products were visualized on 1.5% agarose gels containing 0.5× TAE run at 120 V for 20–25 min and scored for the presence or absence of PCR products of the expected length. The 16S rRNA gene was amplified and sequenced from N. spumigena CH307, P38 and AV45 as previously described [27] and the sequence data was deposited in GenBank (KF360086-KF360088). An alignment of 16 strains of N. spumigena, 8 strains of N. sphaerocarpa and 4 strains of N. harveyana was made using Bioedit. Gaps and ambiguous regions were excluded and a total of 1340 bp of sequence was considered for phylogenetic analysis. A neighbor-joining tree was constructed using DNADIST and NEIGHBOR as implemented in the PHYLIP package. The tree was midpoint rooted using RETREE. 1000 bootstrap replicates were constructed using SEQBOOT, DNADIST, NEIGHBOR and CONSENSE in the PHYLIP package. The production of aeruginosin, spumigin, nodularin, nodulaopeptin and suomilide was mapped to this tree.

LC-MS analyses were performed with an Agilent 1100 Series LC/MSD Ion Trap XCT Plus System (Agilent Technologies, Palo Alto, CA, USA) using a Phenomenex Luna C8 (150×2.0 mm, 5 µm, Phenomenex, Torrance, CA, USA) LC-column. Between 7 and 31 mg of freeze-dried cells of Nodularia strains were extracted for 20 s with 1 ml of methanol in 2 ml plastic tubes containing approximately 200 µl of 0.5 mm glass beads (Scientific Industries, New York) using Fast Prep homogenizer (FP120, Bio 101, Savant) at speed value of 6 m s⁻¹. Extracts were centrifuged for 5 min at 10 000 g prior to LC-MS analysis. High accuracy mass of the aeruginosins of N. spumigena AV1 was measured by UPLC-ESI-QTOF mass spectrometry performed on Synapt G2 HDMS (Waters, MA, USA) in high resolution mode and m/z 500–850 mass range. In MS/MS analysis the mass range was m/z 50–531.

Aeruginosins were derivatized with malondialdehyde (MDA) using 100 µl of methanol extract from N. spumigena AV1 evaporated to dryness in vacuum centrifuge. The resultant residue was dissolved in 100 µl of 12 M H₃PO₄ and 2.4 µl of 1,1,3,3-tetraethoxypropane (Sigma) was added. The sample was evaporated in a vacuum centrifuge after 1 h at room temperature and dissolved in 100 µl of methanol. DNPH derivatives were prepared as previously described [8].

Isolated aeruginosin NOL3 (100 µg) were dried in a 300-µl glass vial which was then transferred to a 4-ml glass vial containing 1 ml of 6 M HCl. The vial was flushed with argon prior to being closed. Acid hydrolysis was performed by incubating overnight at 110°C. After hydrolysis, the inner vial was dried for 30 min with a vacuum centrifuge. Hydrolyzed and reference amino acids were derivatized by the Marfey method using L-FDAA (1-fluoro-2,4-dinitrophenyl-5)-L-alaninamide) reagent. Reaction mixtures were analyzed with a Luna C18(2) column (150×2, 5 µm; Phenomenex) eluted with an acetonitrile (solvent B) and 0.1% aqueous formic acid (solvent A) gradient (20% B to 75% B in 45 min and then 5 min at the reached level) at a flow rate of 0.2 ml min⁻¹ at a temperature of 40°C. The detection wavelength for the Marfey derivatives was 340 nm. Acid hydrolysate of aeruginosin 298-A from Microcystis aeruginosa NIES-298 was used as a reference for L-Choi [28].

100 µg of aeruginosin NOL3 was incubated with 100 µl of 5 M NaOH in a closed vial for 24 h at 110°C. The entire 100 µl solution was transferred to a 20 ml brown vial with 18 mm magnetic caps with silicon/Teflon disks (VWR International, USA) containing 4 ml of MilliQ water, 1.5 g of NaCl and 50 µl of 17.5 M H₃PO_4, and the vial was immediately sealed. The vial was agitated for 5 min at 70°C and 500 rpm in a GC-MS autosampler (combiPAL, CTC Analytics). A SPME (Solid Phase Micro Extraction) needle penetrated 12 mm through the vial cap, exposing 10 mm of the fiber (DVB/CAR/PDMS, Supelco, Sigma-Aldrich Co., USA) inside the vial. After a 30 min extraction the fiber was retracted and sample was injected in the GC column with an injection needle penetration of 32 mm and the fiber exposure of 10 mm. After 10 min of desorption time, the fiber was removed from the injection port and the sample was injected with a split ratio of 5∶1. HP 6890 gas chromatograph with an Agilent 5973 Network mass selective detector (Agilent Technologies, Wilmington, DE, USA) with a split/splitless injector and a SPB™-624 capillary column (30 m×0.25 mm, 1.4 um; Supelco, Sigma-Aldrich Co., USA) were used as follows: Injection and detector at 250°C, oven started at 150°C for 2 min from which temperature increased 5°C min⁻¹ during 10 min. The final temperature was 220°C after a total run time of 26 min. Helium was used as carrier gas with a flow rate of 1 ml min⁻¹. A standard 0.14 mM hexanoic acid (Sigma-Aldrich Co., USA) solution was prepared. The full scan electron impact mass spectra were obtained at a range of 50–200 m/z.

Aldehydes were converted to alcohols in the methanol extract using NaBH₄ reduction which made it possible to purify aeruginosin NOL1 by HPLC. One gram of freeze dried AV1 cells was extracted with 70 ml of methanol using a tip homogenizer (SilentCrusher M, Heidolph, Germany) in three 30 sec cycles at ambient temperature with a speed of 16000 rpm 3×30 sec. The suspension was centrifuged (10000 g, 5 min) and dichloromethane and water was added to the supernatant in volume ratio of 1∶1:1. Phases were separated by centrifugation (5000 g, 5 min). The upper water/methanol phase was collected and vacuum evaporated to dryness. The residue was dissolved in 4 ml of methanol, 50 mg of NaBH₄ was added and after 5 min reaction time the solution was vacuum evaporated to dryness. The residue was dissolved in 1 ml of 15% acetonitrile. 100 µl portions of the solution were injected 10 times into a Luna C8 (2) (10×150 mm, 5 µm, 100 Å, Phenomenex) column which was eluted isocratically with 0.1% TFA in 15% acetonitrile. Pooled fractions containing aeruginosin NOL1 were evaporated in a vacuum and dissolved in CD₃OD for NMR. ¹H and ¹³C NMR spectra were obtained with a Varian Unity Inova 600 MHz NMR spectrometer equipped with cryogenically cooled triple-resonance ¹H, ¹³C, ¹⁵N probe head and actively shielded z-gradient system. DQF-COSY, TOCSY (120 ms mixing time) experiments were collected using 2048 and 512 data points in F₁ and F₂ dimensions, corresponding to acquisition times of 0.34 and 0.085 s, respectively. The corresponding acquisition times in ¹³C HSQC and ¹³C HMBC experiments were 0.02 (¹³C dimension) and 0.17 (¹H dimension), and 0.014 (¹³C dimension) and 0.34 (¹H dimension), respectively. The average one- and three-bond ¹H-¹³C couplings were estimated to be 140 Hz and 8 Hz, and ¹H-¹³C transfer delays for HSQC and HMBC were set to 3.57 and 62.5 ms, respectively. All spectra were collected at 25°C. Spectra were processed and analyzed using VNMRJ 2.1 version B and ACD/SpecManager version 11.03 software packages.

The region of the aerB gene encoding the adenylation domain was amplified by PCR from N. spumigena CCY9414 using oligonucleotide primers designed to anneal to the substrate-conferring portion of each adenylation domain. Primer design and PCR reactions were performed as described previously [8]. PCR products were digested with NcoI and PmeI restriction enzymes, gel excised and ligated to pFN18A (HaloTag® 7) T7 Flexi® vector (Promega, WI, USA) opened with the same enzymes. Ligation mix was transformed into Escherichia coli (KRX) competent cells following the manufacturer’s instructions. Colonies were grown in shaker (160 rpm) at 37°C overnight in 3 ml of LB medium supplemented with 100 mg ml⁻¹ ampicillin. In the following day, 400 µl was used to inoculate 20 ml of TB medium containing 50 mg ml⁻¹ carbenicillin and incubated with shaking at 37°C (160 rpm) for 1.5 h and then induced by the addition of 0.1% of rhamnose and the culture was grown overnight (16–18 h) in shaker (100 rpm) at 24°C. E. coli cells were collected and sonicated as described previously [8]. The expression of soluble protein was observed in 10% SDS PAGE gel. The soluble adenylation domains were purified using HaloTag® Protein Purification System (Promega). Protein concentration of the preparations was measured with the BCA protein assay kit (Pierce). ATP-pyrophosphate exchange assay was performed as described previously [9].