New Prenylated Aeruginosin, Microphycin, Anabaenopeptin and Micropeptin Analogues from a Microcystis Bloom Material Collected in Kibbutz Kfar Blum, Israel

Thirteen new and eighteen known natural products were isolated from a bloom material of an assembly of various Microcystis spp. collected in November, 2008, from a commercial fishpond near Kibbutz Kfar Blum, the Jordan Valley, Israel. The new natural products included the prenylated aeruginosin KB676 (1), microphycin KB921 (2), anabaenopeptins KB906 (3) and KB899 (4) and micropeptins KB928 (5), KB956 (6), KB970A (7), KB970B (8), KB984 (9), KB970C (10), KB1048 (11), KB992 (12) and KB1046 (13). Their structures were elucidated primarily by interpretation of their 1D and 2D nuclear magnetic resonance spectra and high-resolution mass spectrometry. Marfey’s and chiral-phase high performance liquid chromatography methods were used to determine the absolute configurations of their chiral centers. Aeruginosin KB676 (1) contains the rare (2S,3aS,6S,7aS)-Choi and is the first prenylated aeruginosin derivative described in the literature. Compounds 1 and 5–11 inhibited trypsin with sub-μM IC50s, while Compounds 11–13 inhibited chymotrypsin with sub-μM IC50s. The structures and biological activities of the new natural products and our procedures of dereplication are described.


Introduction
Cyanobacterial water blooms are initiated from dormant cells under appropriate environmental conditions, which develop into a stable population of different biomass intensities and toxin content, which eventually collapses, disintegrates and discharges toxins to the water environment [1]. These blooms are frequently described as a single-species phenomenon. However, recent studies suggest that many of these blooms are composed of complex populations of various strains of one or more species of the same genus, as well as associated bacteria, and the community structure varies in time and space [2,3]. The natural products that are biosynthesized by the community members most probably govern the variation of the community in space and time [4,5]. A large volume of research was devoted to the toxins that are produced by cyanobacterial blooms and to the toxic action of these toxins; microcystins, anatoxins, saxitoxins and cylindrospermopsins [6,7]. However, recent studies have suggested that microcystins are involved in quorum sensing and might be produced to manage cyanobacterial colonies in the environment [8] and not necessarily produced for their toxic effects.
The protease inhibitors that are produced by microcystin-producing genera of cyanobacteria, i.e., micropeptins, aeruginosins, anabaenopeptins, microginins and microviridins, have been shown to affect the normal life cycle of organisms that interact with cyanobacteria in the environment [9]. These secondary metabolites usually appear in the bloom in relatively high concentrations as an array of structurally-related compounds, and their ecological role is still not fully understood [10].

Results and Discussion
Thirty-one natural products were isolated from a 70% aqueous methanol extract of bloom material collected from a fishpond of the Kibbutz Kfar Blum. The compounds were separated through fractionation by reversed-phase C18 open column, size exclusion chromatography and purification on various reversed-phase high performance liquid chromatography (HPLC) columns. The fractionation process was guided by the serine protease inhibition assay. Dereplication and verification of the purity of the isolated natural products was achieved by running liquid chromatography mass spectrometry (LCMS) and nuclear magnetic resonance (NMR) spectra on each one of the isolated compounds.

Structural Elucidation of Microphycin KB921
Microphycin KB921 (2, Figure 2) was isolated as a transparent solid material presenting an HR ESI MS quasi-molecular ion ([M + Na] + ) at m/z 944.4650, which corresponded to a molecular formula of C49H63N9NaO9. The molecular formula of 2 and its NMR spectra in DMSO-d6 (Table 2) pointed to its peptidic nature. The 23 degrees of unsaturation suggested, beside the nine carbonyls, three phenyl rings and a pyrrolidine ring, a cyclic peptide structure. A full assignment of the 1 H and 13 C NMR spectra of 2 in DMSO-d6 (Supplementary Table S2) established the structures of three Phe, two Ala, one Pro, one Gln and one Leu moieties. HMBC correlations (Supplementary Table S2) allowed the construction of two short amino acid sequences: 5 Pro-6 Leu-7 Phe-8 Phe-1 Phe and 2 Ala-3 Gln. NOE correlations from a ROESY experiment (Supplementary Table S2) enabled the connection of 1 Phe-H-2 with 2 Ala-NH and 4 Ala-NH with 5 Pro-H-5 extending the amino acid sequence to: 4 Ala-5 Pro-6 Leu-7 Phe-8 Phe-1 Phe-2 Ala-3 Gln. However, the closure of the macrocyclic ring was evident only from the molecular weight of 2. To support this assumption, an MS/MS measurement was performed for 2, and some of the indicative data supporting the closure of the cycle between 3 Gln-CO and 4 Ala-NH are presented in Figure 3. Applying Marfey's method [28] to the amino acid derivatives of 2 established all amino acids as being of the L-configuration. Based on these arguments, the structure of 2 was assigned to microphycin KB921.     Microphycin KB921 (2) most probably belongs to the growing family of cyanobactins that are synthesized ribosomally. This family includes cyclic peptides, such as prenylagaramides A and B [30], planktocyclin [34] and anacyclamide [35], but their role in cyanobacteria is yet unknown.

Structural Elucidation of Anabaenopeptins KB906 and KB899
Compounds 3 and 4 ( Figure 4) were identified as new anabaenopeptins based on their mass weights and the characteristic signals of the urea bridge in the proton and carbon NMR spectra in DMSO-d6, two amide doublet signals between 6 and 7 ppm in the proton spectrum and a carbonyl signal between 157 and 158 ppm in the carbon spectrum [16].
Anabaenopeptin KB906 (3) was isolated as a white solid material that presented an HR ESI MS protonated molecular ion at m/z 907.5417 and a molecular formula of C46H71N10O9. The 1 H NMR spectrum of 3 in DMSO-d6 ( Table 2) presented in the lower field proton signals characteristic of phenol-OH, seven secondary amide protons, a phenyl and a para substituted phenol ring. Signals of six methines, two methylenes and a methyl next to electron withdrawing atoms appeared in mid-spectrum, while two doublet and two triplet methyl signals were evident among other signals in the aliphatic region of the 1 H NMR spectrum. In the 13 C NMR spectrum (Table 2), 3 presented six acid/amide carbonyl signals around 170 ppm, three quaternary carbon signals around 156 ppm and additional two quaternary and five methine carbon signals at the aromatic region, six methine carbons next to electron withdrawing groups in mid-spectrum and a handful of signals in the upper field of the spectrum. The assignment of the proton and carbon signals to the following amino acid building blocks-2 × Ile, homophenylalanine (Hph), N-methylhomotyrosine (NMeHty), Arg and ε-substituted-Lys-was achieved by interpretation of the data from the 2D NMR spectra (Supplementary Table S3). These building blocks were assembled into the planar structure of 3 through the HMBC correlations of 2 NMeHty-CO with the amide proton of 1 Ile, of 3 Hph-CO with 2 NMeHty-H-2, of 5 Lys-CO with the amide proton of 4 Ile, of 1 Ile-CO with the ε-amide proton of 4 Lys-and of the uryl-CO with the amide proton of 6 Arg and 5 Lys-H-2 and the NOE correlations of 1 Ile-NH and 2 NMeHty-H- 2, of 4 Ile-H-2 with 3 Hph-NH, of 5 Lys-ε-NH with 1 Ile-H-2 and of 5 Lys-α-NH with 6 Arg-α-NH. The application of Marfey's method [28] to the hydrolysate of 3 established Ile, Hph, NMeHty and Arg as L-configuration and Lys as D-configuration. Based on these arguments, the structure of anabaenopeptin KB906 was assigned as 3 ( Figure 4). Anabaenopeptin KB899 (4) presented comparable NMR spectra to those of 3 (Table 2) and an HR ESI MS quasi-molecular ion ([M + Na] + ) at m/z 922.4693 corresponding to the molecular formula C48H65N7NaO10. The full assignment of its NMR data (Supplementary Table S4) indicated that it shared 1 Ile, 2 Hph, 3 NMeHty and 5 Lys with 3, but contained 4 Val instead of 4 Ile and 6 Tyr instead of 6 Arg. The HMBC and NOE correlations that assisted in assembling the planar structure of 4 are presented in Figure 5. Marfey's analysis [28], similar to the one applied for 3, established the absolute configuration of Lys as D, while that of the rest of the amino acids as L, assigning structure 4 to anabaenopeptin KB899 (Figure 4).

Structural Elucidation of the New Micropeptins
Micropeptin KB928 (5, Figure 6) was isolated as an amorphous white material that possessed a positive HR ESI MS quasi-molecular ion at m/z 929.5090 ([M + H] + ), corresponding to a molecular formula of C44H69N10O12 and 16 degrees of unsaturation. Its NMR data, measured in DMSO-d6 (Table 3  and Supplementary Table S5), presented signals characteristic of the micropeptin-type cyclic depsipeptide [25], i.e., five secondary amide protons (8.6-7.2 ppm), an ester oxymethine proton (5.45 ppm), an amide methyl group (2.73 ppm), the distinctive hydroxyl signal of a 3-amino-6-hydroxy-2-piperidone (Ahp) moiety (6.11 ppm Table S5) allowed the expansion of the later fragments to the entire acid residues, except for Arg, whose amide carbon was assigned based on an HMBC correlation of Ahp-NH (7.29 ppm) with the amide carbon (170.3 ppm) and the NOE of the later amide NH with Arg-H-2.  (5), KB956 (6), KB970A (7), KB970B (8) and KB984 (9).  The acid units were combined to the linear structure, 1 butyric acid ( 1 BA)-2 Asp-3 Thr-4 Arg-5 Ahp-6 Val-7 NMePhe-8 Ile, using HMBC correlations between 2 Asp-NH and 1 BA-CO, 3 Thr-NH and 2 Asp- CO,4 Arg-NH and 3 Thr-CO, 5 Ahp-NH and 4 Arg-CO, 6 Val-2 and 5 Ahp- CO,7 NMePhe-NMe and 6 Val-CO and 8 Ile-NH and 7 NMePhe-CO. The lactone bridge of 5 was established based on the HMBC correlation of 3 Thr-H-3 with 8 Ile-CO. The relative configuration of the Ahp was established as 3S*,6R* based on the chemical shifts of the protons and carbons of the Ahp moiety that were found to be identical to those of cyanopeptolin S [20] and the NOE between H-4ax (2.55 m) and the axial 6-OH (6.11 d,3.2 Hz). Marfey's analysis [28] of the amino acids derived from the hydrolysis of 5 (with and without oxidation of the aminal) revealed the presence of only L-form Arg, Asp, Glu (from Ahp), Ile, NMePhe, Thr and Val. The absolute configuration of the chiral centers of the Ahp moiety were thus established as 3S, 6R.
Marfey's analysis does not distinguish between allo-Ile and Ile and between allo-Thr and Thr. The NMR data was thus used to confirm the identity of these amino acids. We have shown in the past that the proton and carbon chemical shifts of the methyl-groups of Ile in the micropeptins are sensitive to the absolute configuration of C-3 of Ile [36]. Based on this observation and the chemical shifts of the Ile methyl groups (δC 16.0 and 11.3 for Me-5 and Me-6, respectively) in 5, it was established as L-Ile. The observed J-value (less than 1 Hz) between H-2 and H-3 of the substituted threonine in Compound 5 suggested that, as in the case of all known micropeptins, it should be threonine and not allo-threonine [37]. On the basis of these arguments, the structure of micropeptin KB928 was determined as 5.
Micropeptin KB956 (6, Figure 6) was isolated as a glassy solid, which exhibited an HR ESI MS protonated-quasi-molecular ion at m/z 957.5412, corresponding to the molecular formula C46H73N10O12 and 16 degrees of unsaturation. Its 1 H and 13 C NMR data in DMSO-d6 (Table 3) were almost identical to those of 5, except for the appearance of two additional methoxy groups (δH 3.56 s and 3.02 s and δC 51.7 and 55.5). A full assignment of the NMR data (Supplementary Table S6) by interpretation of the correlations from homo-nuclear 2D NMR experiments (COSY, TOCSY and ROESY) and hetero-nuclear 2D NMR experiments (HSQC and HMBC) revealed that 6 contained 4-OMe-Asp (δH 3.56 s and δC 51.7) and 3-amino-6-methoxy-2-piperidone (Amp, δH 3.02 s and δC 55.5), instead of Asp and Ahp that exist in 5. The assignment of the relative configuration of the Amp moiety is presented in Figure 7a, and assembly of the acid units to the cyclic structure is presented in Figure 7b. A similar analysis [28] to the one applied for 5 established the absolute configuration of the acid units in 6 as L-Arg, L-4-OMe-Asp, L-Ile, L-NMePhe, L-Thr, L-Val and 3S, 6R-Amp. Based on these arguments, the structure of micropeptin KB956 was established as 6.
Micropeptin KB970A (7, Figure 6), a glassy solid, exhibited an HR ESI MS protonated molecular ion at m/z 971.5563 corresponding to the molecular formula C47H75N10O12 and 16 degrees of unsaturation. The molecular formula of 7 exceeded that of 5 in three methylenes. The 1 H and 13 C NMR data of 3 in DMSO-d6 ( Table 3) Table S7) by interpretation of the 2D NMR experiments revealed that 7 contained a 4-OMe-Asp (like 6) and a hexanoic acid instead of Asp and butyric acid in 5. The absolute configuration of the asymmetric centers was established by a similar procedure as described for 5. Structure 7 was thus assigned to micropeptin KB970A. Micropeptin KB970B (8, Figure 6) was isolated as a glassy material with a similar HR ESI MS protonated molecular ion, m/z 971.5561, and an identical molecular formula C47H75N10O12 to that of 7. The 1 H and 13 C NMR spectra of 8 (Table 3) were almost identical to those of 7, except for the chemical shifts of the methoxy moiety (δH 3.02 s and δC 55.6) and the amino piperidone moiety. A full assignment of the NMR data of 8 (Supplementary Table S8) revealed that it differs from 7 in having the Amp moiety instead of the Ahp moiety (in 7) and Asp instead of 4-OMe-Asp. The assignment of the absolute configuration of the chiral centers was achieved in a similar way to that of 5, assigning structure 8 to micropeptin KB970B.
Micropeptin KB984 (9, Figure 6) differed in 14 mass unit from 7 and 8, exhibiting an HR ESI MS quasi-molecular ion at m/z 985.5722 and a molecular formula of C48H77N10O12 that differed in a CH2 from that of 7 and 8. A comparison of the 1 H and 13 C NMR spectra of 9 (in DMSO-d6, Table 3) with those of 7 and 8 suggested that 9 contained both 4-OMe-Asp (δH 3.56 s and δC 51.8) and Amp (δH 3.02 s and δC 55.6) moieties. The assignment of the 1D and 2D NMR data of 5 (Supplementary  Table S9) and the absolute configuration of the chiral centers, as described above for 5, established the acid units in micropeptin KB984 as L-Arg, L-4-OMe-Asp, L-Ile, L-NMePhe, L-Thr, L-Val, 3S,6R-Amp and hexanoic acid and its structure as 9.
Micropeptin KB970C (10, Figure 8) presented 1 H and 13 C NMR spectra (DMSO-d6) almost identical to those of 9 and an HR ESI MS protonated molecular ion at m/z 971.5560, in accordance with the molecular formula C47H75N10O12. The major difference between the 1 H NMR spectra of 10 and 9 (Tables 4 and 3, respectively) was at the high end of the spectrum, where 10 presented a doublet methyl (δH 0.85) instead of the triplet methyl of 9 (δH 0.84). The 14 mass units difference between 10 and 9, and the one carbon difference in the 13 C NMR spectra (Tables 3 and 4) suggested that 6 contained a valine instead of the isoleucine of 9. A full assignment of the NMR data of 10 (Supplementary Table S10) proved this assumption, while Marfey's analysis [28], similar to the one used for 5, established the absolute configuration of all of the chiral centers, as L, ascribing structure 10 to micropeptin KB970C.    Micropeptin KB1048 (11, Figure 8) was isolated as a glassy solid that exhibited a HR ESI MS complex quasi-molecular ion at m/z 1049.5448/1051.5444 (3:1, indicative of one chlorine atom) and a molecular formula of C49H78ClN10O13. The 1 H and 13 C NMR spectra (DMSO-d6) of 11 (Table 4) were similar to those of 9 (Table 3) but presented some differences in the aromatic and the aliphatic regions. In the 1 H NMR spectra, 11 presented a 1,2,4-tri-substituted phenol in the aromatic region, and three triplet and two doublet methyl signals at the higher end of the spectrum, while the rest of the spectrum seemed almost identical with that of 9.
The assignment of the NMR data of 11 (Supplementary Table S11), by interpretation of the correlations from the homo-and hetero-nuclear 2D NMR experiments (COSY, TOCSY, ROESY, HSQC and HMBC), revealed that 11 contained hexanoyl ,arginyl,Amp,N, and isoleucyl moieties. These moieties were assembled to planar structure 11 through HMBC and ROESY correlations. Appling Marfey's methodology [28] established the absolute configuration of all of the amino acids as of the L-configuration. The observed J-value between H-2 and H-3 (0-1 Hz) of the threonine unit of 11 suggested that it should be L-threonine and not L-allo-threonine [37]. The 3S,6R absolute configuration of the Amp moiety was inferred (similar to the one presented in Figure 7a) from the NOEs of axial H-3 with axial H-5 (δH 1.70), axial H-4 (δH 2.40 brq,12.8 Hz) with equatorial H-5 (δH 2.06 brd,13.4 Hz) and equatorial H-6 (δH 4.46) with axial H-5 and equatorial H-5, based on the S absolute configuration of Glu that was established by Marfey's analysis. In the case of the N,N-disubstituted-Ile, the carbon chemical shifts measured for Compound 11, 10.4 (C-5) and 13.7 (C-6) ppm, were found to be similar to those measured for L-Ile, 10.3 (C-5) and 13.9 (C-6) ppm (established for nostopeptins A and B by chiral-GCMS [38]), and different from those measured for allo- Ile, and 14.1 (C-6) ppm (established for micropeptin KT1042 [27]). For the Ile moiety at the carboxylic end of the peptide, the measured carbon chemical shifts were 11.6 (C-5) and 14.1 (C-6) ppm, matching those of L-allo-Ile (established for oscillapeptin J by chiral-GCMS [39]), 11.4 (C-5) and 14.3 (C-6) ppm and differing from those of L-Ile (established for nostopeptins A and B by chiral-GCMS [38]), 11.3 for (C-5) and 16.1 (C-6) ppm. Based on these arguments, structure 11 was assigned to micropeptin KB1048.
Micropeptin KB992 (12, Figure 8) was isolated as a transparent solid that presented an HR ESI MS quasi-molecular ion ([M + Na] + ) at m/z 1015.5121 corresponding to the molecular formula C50H72N8NaO13 and 19 degrees of unsaturation. Its 1 H NMR spectrum (DMSO-d6, Table 4) presented signals corresponding to a mono-substituted phenyl ring, a para-substituted phenol, three doublet secondary amide protons (a forth signal was shown to be buried under the signals of the phenyl ring by COSY correlation), two singlet primary-amide protons, eleven protons next to electron withdrawing groups, two singlet methyl groups attached to electronegative atoms and, among others, six doublet methyl groups and one triplet methyl group in the aliphatic region. Among other signals, the 13 C NMR spectrum ( Table 4) presented signals of eight carbonyls, eight aromatic signals in accordance with one mono-substituted phenyl and one para-substituted phenol and eleven carbons next to electron withdrawing groups.
The interpretation of the 1D and 2D NMR spectra of 12 (Supplementary Table S12) revealed the planar structures of p-hydroxyphenyl lactyl (Hpla), aspaginyl, O-substituted threonyl, leucyl, Amp, N,N-disubstituted isoleucyl, NMe-phenylalanyl and isoleucyl, which closed a lactone ring with the oxygen of the threonyl moiety. The absolute configurations of the amino acids (all L) and the Amp moiety (3S,6R) were assigned for 11 as described above. The absolute configuration of Hpla was assigned as D by chiral-HPLC. On the basis of these arguments, structure 12 was assigned to micropeptin KB992.
COSY and TOCSY correlations (Supplementary Table S13) were used to assign the sequence of the proton signals of HcAla: α-NH (δH 8.47) through H-9pax and H-9peq and of the latter two with H-4, while the HSQC correlations assigned the carbons of this moiety. The carboxyl of HcAla was assigned through HMBC correlation of Ahp-α-NH with the amide carbonyl that resonated at δC 170.7 and NOE correlations of Ahp-α-NH with HcAla-α-NH and H-2. Assuming a twisted boat conformation for the cyclohexenyl moiety, the pseudoaxial H-8 (H-8pax) and H-9 (H-9pax) were identified by their shift to a higher field. The NOE correlation of H-9pax with H-7 and of H-8pax with H-4, as well as the rest of the NOEs of this spin system shown in Figure 9 established both as pseudoaxial and the relative configuration of the hydroxyl cyclohexenyl moiety as 4S*,7R*. Although the NOE pattern of H-2, 3a, 3b and 4 pointed to a restricted rotation (Figure 9), it was not possible to assume the relative configuration of Positions 2 and 4. The absolute configuration of C-2 of HcAla was determined by the advanced Marfey method [41]. The hydrolysate of 13 was reacted with L-and D-FDAA, and the retention times of the HcAla-DAA derivatives were obtained using HPLC-MS. The retention time of the HcAla-L-DAA (46.9 min) was shorter than that of HcAla-D-DAA (50.2 min), suggesting that HcAla-C-2 is of the L-configuration [41]. The absolute configuration of the other amino acids of 13 was elucidated by a combination of Marfey's method [28] and NMR data, as described above for 5-12. Based on these arguments, structure 13 was assigned to micropeptin KB1046. Micropeptins 5-13 could be divided into two separate groups; those that contain a short fatty acid in the side chain and arginine at the fourth position from the amino termini of the peptide and those that contain an Hpla residue at the side chain and lipophilic acid at the fourth position. However, the known micropeptins that were isolated from the bloom material also contained combinations of short fatty acids in the side chains and lipophilic residues at Position 4, and hydroxy acids at the side chain aside from arginine at Position 4. This was more in accordance with the regular varieties observed in bloom-forming cyanobacteria.

Methods of Dereplication
Over the past several years, we have examined several methods to avoid the isolation and structural elucidation of known metabolites from cyanobacterial bloom material. Among the methods we applied were MALDI-TOF MS of on-plate extracts of small amounts of cyanobacteria colonies that gave reproducible results for the major components isolated from the large-scale extract of the bloom material, but failed to give reproducible results for the minor isolated components. In some cases, it seemed as if easily-ionized natural products hindered the ionization of some types of natural products. The yields of the bioactive natural products from cyanobacterial blooms are usually low (3 × 10 −3 % to 1 × 10 −4 %), and these compounds are accompanied by large amounts of fatty-acids and related metabolites that interfere with their chromatography and ionization when LCMS is used for dereplication. In our attempts to use the ESI LCMS methodology for dereplication of cyanobacterial bloom extracts, we found that we are usually able to detect the major bioactive metabolites that are available at yields of about 1 × 10 −3 % of the crude extract, but fail to detect the minor ones. The methodology we currently use combines bioactivity (usually protease inhibition)-guided isolation and purity verification by ESI LCMS and NMR, which allows us to identify the previously known compounds before we start the structural elucidation. This methodology allowed us to identify the stereoisomers of the Choi unit in the aeruginosins [15,27,35] and allo-Ile-versus Ile-containing anabaenopeptins [15]. aeruginosin 298B (4.4 mg, rt 17.0 min, 4.6 × 10 − 4 % yield), anabaenopeptin 908 (5.8 mg, rt 19.5 min, 6.1 × 10 − 4 % yield) and anabaenopeptin H (4.0 mg, rt 21.0 min, 4.2 × 10 − 4 % yield).

Protease Inhibition Assays
The procedures used to determine the inhibitory activity of the new compounds on trypsin and chymotrypsin were described in a previous paper [19]. The range of concentrations used in the assays was between 45.5 μM and 0.011 μM.

Conclusions
Microcystis blooms usually produce a large variety of short peptides, of which the micropeptins are dominant in quantity and diversity. In the current research, the 7:3 MeOH:H2O extract yielded 31 different peptides of six structural groups. As in the case of many other Microcystis blooms that we have investigated [18,19,24,25], the diversity of the micropeptins (144 isolated variants excluding this publication) isolated was the greatest (15 compounds). The number of the anabaenopeptins (57 isolated variants excluding this publication) was second to the micropeptins, with nine closely-related analogs, while the aeruginosins (the second most diverse group of protease inhibitors in cyanobacterial blooms, 69 isolated variants excluding this publication) were represented by only three analogs, one of which was the prenylated aeruginosin KB676 (1). These three groups of modified peptides are biosynthesized by non-ribosomal peptide synthetases that are flexible and capable of synthesizing series of analogues metabolites. The reason(s) for the biosynthesis of these metabolites and their high variability in cyanobacterial water blooms are intriguing, and our current research is aimed at revealing the purpose for the biosynthesis of these metabolites.

Supplementary Information
1D ( 1 H, 13 C) and 2D NMR (HSQC, HMBC, COSY, ROESY) spectra and HR MS data of Compounds 1-13, tables of full NMR data of 1-13 and two figures with the structures of the known metabolites isolated in this study are available.