Microginins from a Microcystis sp. Bloom Material Collected from the Kishon Reservoir, Israel

During blooms, cyanobacteria produce diverse modified peptides. Among these are the microginins, which inhibit zinc-containing metalloproteases. Ten microginins, microginins KR767 (1), KR801(2), KR835 (3), KR785 (4), KR604 (5), KR638 (6), KR781 (7), KR815 (8), FR3 (9), and FR4 (10), were isolated from the extract of a bloom material of Microcystis sp. (IL-405) collected from the Kishon Reservoir, Israel in the fall of 2009. The structures of the pure compounds were elucidated using 1D and 2D NMR techniques and high-resolution mass spectrometry. The absolute configuration of the chiral centers of the amino acids were determined by Marfey’s and advance Marfey’s methods and by comparison of 1H and 13C NMR chemical shifts of the Ahda derivatives with those of known microginins. These microginins differ in sequence and absolute configuration of the chiral centers of the Ahda moieties and by N-methylation of the Ahda amine group and extent of chlorination of the Ahda terminal methyl group. The compounds were evaluated for inhibition of the zinc metalloprotease, aminopeptidase M, and exhibited low- to sub-nanomolar half maximal inhibitory concentration (IC50) values.


Introduction
Water-bloom forming cyanobacteria are prolific producers of diverse groups of highly active natural products [1]. Among the most frequently isolated are microcystins [2], micropeptins [3], aeruginosins [4], anabaenopeptins [5], microviridins [6], and microginins [7]. The microginins are linear peptides, characterized by N-terminal β-amino-α-hydroxy-decanoic or octanoic acid (in one case β-amino-decanoic) [8], which contain three to five additional amino acids or N-methylated amino acids [9]. The microginins inhibit zinc-containing metalloproteases, and their selectivity is determined by the sequence of the amino acids that are conjugated to the N-terminal β-amino-α-hydroxy-acid [10]. Thirty-one different microginins had been isolated and fully characterized by the end of 2016 (Table S1), and the use of sensitive mass spectroscopy (MS) methods, such as matrix-assisted laser desorption/ionization (MALDI) MS/MS and electrospray ionization (ESI) MS/MS, for the characterization of secondary metabolites of cyanobacterial bloom materials has enabled identified at least fifteen additional sequences of peptides belonging to the microginin group of metabolites [11,12]. These most recently identified microginins have been partially characterized with respect to their activities, the absolute configurations of the chiral centers of the amino acids, and the identities of N-methylated amino acids and aliphatic amino acids (i.e., NMe-valine versus leucine and isoleucine). The relative and absolute configurations of β-amino-α-hydroxy-decanoic (Ahda) or β-amino-α-hydroxy-octanoic acid (Ahoc) in the microginin are of great importance for biological activity [10], and three of the eight possible isomers are known to be produced by cyanobacteria. Compounds with similar planar structures but different relative and absolute configurations may not be distinguished by MS. Additional interesting aspects of the microginins are mono-or di-chlorination of the Ahda/Ahoa chain ends and N-methylation of the amino group of these acid residues. The reason that cyanobacteria produce so many derivatives of homologous peptides is unclear.

Results and Discussion
Microginins 1-10 were isolated from aqueous-methanol extract of freeze-dried Microcystis sp. bloom material (collection number IL-405) by successive separations on a reversed-phase open column, size-exclusion on a Sephadex LH-20 column, and reversed-phase HPLC.
Microginin KR767 (1) presented a high-resolution (HR) ESI MS molecular ion adduct at m/z 768.4551, corresponding to the molecular formula C 41 H 61 N 5 O 9 , with 14 degrees of unsaturation. The 1 H NMR spectrum (Table 1) revealed the presence of two singlet phenol protons (δ H 9.25 and 9.20), five broad exchangeable protons (δ H 8.49, 8.35, 8.10, 7.92, and 6.47), four doublet aromatic signals of two para-substituted phenol rings (δ H 7.01, 7.00, 6.64, and 6.62, 2H each), eight protons next to heteroatoms (δ H 5.23, 4.86, 4.31, 4.30, 4.27, 3.40, 3.27, and 3.18), two NMe signals (δ H 2.89 brs and 2.56 brt), overlapping protons in the aliphatic region and three methyl signals (δ H 0.84 t, 0.85 d, and 0.81 d). The 13 C NMR spectrum ( Table 2) of 1 revealed, among other peaks, signals due to five carbonyls (δ C 173. 1, 171.4, 171.0, 170.1, and 168.5), an oxymethine carbon (δ C 68.5), five methine carbons (δ C 60. 3, 59.4, 54.0, 51.9, and 50.6), and one methylene carbon (δ C 46.7) attached to amines, suggesting that 1 is a pentapetide secondary metabolite. The presence of the broad triplet NMe signal (δ H 2.56, J = 4.7 Hz), which was coupled with two broad exchangeable protons at δ H 8.49 and 8.35, and the presence of a shielded carboxyl-carbon at δ C 173.1, suggest that 1 is a linear short peptide that occurs as a zwitterion of its N-and C-termini amino acids. The structures of the four proteinogenic amino acids, two Tyr, Pro, and NMeLeu, were established by a combination of COSY, HSQC, and HMBC 2D NMR experiments ( Figure 2 and Table S2). The structure elucidation of the fifth amino acid, Ahda, was more challenging. The Ahda proton spin system, 2-OH, H-2, H-3 (3-NH 2 -Me), H 2 -4, H 2 -5, and H 2 -8, H 2 -9 H 3 -10, was established through COSY correlations. However, the connection of methylene-5 to methylene-6 through methylene-8 could not be determined by the COSY or HMBC correlations, although TOCSY correlations suggest that the latter two spin systems are connected. Counting the numbers of all of the atoms engaged in the proven proteinogenic amino acids and the latter two spin systems summed to C 39 H 57 N 5 O 9 , suggesting that methylene-5 is connected to methylene-8 through two additional methylenes. This was in agreement with the number of methylene carbons in 13 C NMR and HSQC spectra. The connection of the five subunits to the planar structure of 1 was achieved through HMBC correlations observed between Ahda-CO and 1 Tyr-NH, 1 Tyr-CO and NMeLeu-H-2, and -NMe and both Pro-CO and 2 Tyr-NH, and by NOE correlations between NMeLeu-H-2, H-3, and H-3 and Pro-H-5 and H-5 ( Figure 2 and Table S2). The absolute configurations of the proteinogenic amino acids, NMeLeu, Pro, and Tyr were established by Marfey's method [14] to be of the L configuration.

Results and Discussion
Microginins 1-10 were isolated from aqueous-methanol extract of freeze-dried Microcystis sp. bloom material (collection number IL-405) by successive separations on a reversed-phase open column, size-exclusion on a Sephadex LH-20 column, and reversed-phase HPLC.
Microginin KR767 (1) presented a high-resolution (HR) ESI MS molecular ion adduct at m/z 768.4551, corresponding to the molecular formula C41H61N5O9, with 14 degrees of unsaturation. The 1 H NMR spectrum (Table 1) revealed the presence of two singlet phenol protons (δH 9.25 and 9.20), five broad exchangeable protons (δH 8.49, 8.35, 8.10, 7.92, and 6.47), four doublet aromatic signals of two para-substituted phenol rings (δH 7.01, 7.00, 6.64, and 6.62, 2H each), eight protons next to heteroatoms (δH 5.23, 4.86, 4.31, 4.30, 4.27, 3.40, 3.27, and 3.18), two NMe signals (δH 2.89 brs and 2.56 brt), overlapping protons in the aliphatic region and three methyl signals (δH 0.84 t, 0.85 d, and 0.81 d). The 13 C NMR spectrum ( Table 2) of 1 revealed, among other peaks, signals due to five carbonyls (δC 173.1, 171.4, 171.0, 170.1, and 168.5), an oxymethine carbon (δC 68.5), five methine carbons (δC 60. 3, 59.4, 54.0, 51.9, and 50.6), and one methylene carbon (δC 46.7) attached to amines, suggesting that 1 is a pentapetide secondary metabolite. The presence of the broad triplet NMe signal (δH 2.56, J = 4.7 Hz), which was coupled with two broad exchangeable protons at δH 8.49 and 8.35, and the presence of a shielded carboxyl-carbon at δC 173.1, suggest that 1 is a linear short peptide that occurs as a zwitterion of its N-and C-termini amino acids. The structures of the four proteinogenic amino acids, two Tyr, Pro, and NMeLeu, were established by a combination of COSY, HSQC, and HMBC 2D NMR experiments ( Figure 2 and Table S2). The structure elucidation of the fifth amino acid, Ahda, was more challenging. The Ahda proton spin system, 2-OH, H-2, H-3 (3-NH2-Me), H2-4, H2-5, and H2-8, H2-9 H3-10, was established through COSY correlations. However, the connection of methylene-5 to methylene-6 through methylene-8 could not be determined by the COSY or HMBC correlations, although TOCSY correlations suggest that the latter two spin systems are connected. Counting the numbers of all of the atoms engaged in the proven proteinogenic amino acids and the latter two spin systems summed to C39H57N5O9, suggesting that methylene-5 is connected to methylene-8 through two additional methylenes. This was in agreement with the number of methylene carbons in 13 C NMR and HSQC spectra. The connection of the five subunits to the planar structure of 1 was achieved through HMBC correlations observed between Ahda-CO and 1 Tyr-NH, 1 Tyr-CO and NMeLeu-H-2, and -NMe and both Pro-CO and 2 Tyr-NH, and by NOE correlations between NMeLeu-H-2, H-3, and H-3′ and Pro-H-5 and H-5′ ( Figure 2 and Table S2). The absolute configurations of the proteinogenic amino acids, NMeLeu, Pro, and Tyr were established by Marfey's method [14] to be of the L configuration.   The absolute configuration of C-3 of NMeAhda was established as R, by applying the advanced Marfey's method [15]. An attempt to establish the absolute configuration of C-2 in the unit by Mosher's method [16] failed. H-2 in 1 appears as a broad singlet overlapping with Pro-H-2. Its splitting with 2-OH (J = 6.1 Hz), suggests that the J 2-3 is small, and similar to the one measured for 4 (2.5 Hz), thus suggesting that the absolute configuration is either 2S,3S-or 2R,3R [10]. The data summarized in Table 3 for known compounds with proven absolute configurations at C-2 and C-3 of the Ahda derivatives, suggest that the proton and carbon chemical shifts are indicative of the relative configuration of these chiral centers and distinguish the 2S,3S-or 2R,3R-isomers from the 2R,3S-or 2S,3R-isomers. The chemical shifts of the latter carbons and protons of 2S,3S-Ahda and 2R,3R-Ahda of microginin 51-A and microginin 91-C [10], respectively, are almost identical (and the differences are due to dependent amino acids occupying the position next to the Ahda) and different from those of 2S,3R-Ahda in microginin [7]. The chemical shifts of H-2, C-2, H-3, and C-3 of 1 best matched those of microginin 51-B of characterized microginins, suggesting that the configuration of its chiral centers is either 2S,3S-or 2R,3R. Since the absolute configuration of C-3 was established as R, the absolute configuration of NMeAhda was assigned as 2R,3R. Based on these arguments, the structure of microginin KR767 was established to be 1 (Figure 1).  Tables 1 and S2) of 2 was similar to that of 1. However, in the spectrum of 2 a triplet methylene, a triplet of a triplet methylene, and a multiplet methylene, resonating at 3.60, 1.68, and 1.33 ppm, respectively, replaced resonances at 0.84, 1.24, and 1.20 ppm, corresponding to the terminal triplet methyl group of Ahda and two adjacent methylenes, respectively, observed in the spectrum of 1. In the 13 C NMR spectrum of 2 (Tables 2 and S3), signals at 45.5, 32.2, and 26.3 ppm replaced the chemical shifts of the Ahda n-propyl chain end of 1 at 14.1, 22.2, and 31.4 ppm. These differences in the NMR data, together with the MS data, suggest that the chain terminating methyl of Ahda in 1 was substituted by a chloro-methylene in 2. The structure of the Cl-Ahda moiety and of the rest of 2 were established unequivocally by COSY, HSQC, and HMBC correlations (Table S3). The absolute configurations of the chiral centers of 2 were established by Marfey's method [14], advanced Marfey's method [15], and comparison with the chemical shifts of Ahda derivatives (Table 3). Based on the data presented, structure 2 ( Figure 1) Tables 1 and S3) of 3 differed from those of 1 and 2 only in the signals of the NMeAhda moiety; the spectrum of 3 was characterized by a triplet methine at 6.29 ppm and two distinctive methylenes at 2.13 (dt) and 1.43 (tt) ppm. The 13 C NMR spectrum (Tables 2 and S3) presented significant changes in the chemical shifts of two carbons relative to those of 1 and 2 with resonances observed at 75.0 (CH) and 43.0 (CH 2 ) ppm in the spectrum of 3. These changes in the NMR spectra of 3, relative to those of 1 and 2, and the molecular formula of 3, C 41 H 59 Cl 2 N 5 O 9 , were indicative of substitution of the NMeAhda chain end with two chlorine atoms. The structures of the Cl 2 -Ahda, two Tyr, NMeLeu, and Pro moieties and their assembly into the planar structure of 3 were established unequivocally by COSY, HSQC, HMBC, and ROESY correlations (Table S4). The absolute configurations of the chiral centers of 3 were established in the same manner as for 1. On the basis of these arguments, structure 3 ( Figure 1) was assigned for microginin KR835.
Microginin KR787 (4)  corresponded to three protons. The difference of 14 mass units between 2 and 4 and the differences in the NMR spectra suggest that the Cl-NMeAhda in 2 is replaced in 4 by a Cl-Ahda moiety. The structures of the five amino acids that compose 4-Cl-Ahda, two Tyr, NMeLeu, and Pro and their assembly into the linear structure of 4 were established by analyses of the COSY, HSQC, HMBC, and ROESY spectra ( Table S5). The absolute configuration of C-3 of Cl-Ahda was established as R by applying the advanced Marfey's method [15]. The chemical shifts of C-2 and C-3 of 2S,3S-Ahda (70.4 and 52.9 ppm, respectively, in microginin 51-A [10] and 70.6 and 53.0 ppm, respectively, in microginin 299-C [8]) and of 2R,3R-Ahda (71.0 and 52.8, respectively, in microginin 91-C [10]) were similar, but were somewhat different from those of 2S,3R-Ahda (69.5 and 52.9, respectively, in microginin [7]) ( Table 3). The chemical shifts of H-2 and H-3 in 4 matched those of microginin 51-A [10], which contains a Tyr next to the Adha moiety, but were different from those of microginin 91-C and microginin. Furthermore, the multiplicity and J-values of H-2 in 4 (δ H 4.14 dd, J = 5.4, 2.5 Hz) matched those of microginin 299-C (δ H 4.20 dd, J = 5.6, 3.0 Hz) [8] and differed from those of microginin FR3 (9) (δ H 4.10 t, J = 4.6 Hz). Thus, the absolute configuration of the Cl-Ahda chiral centers were established as 2R,3R, and microginin KR787 (4) has the structure shown in Figure 1. Microginin KR604 (5), an amorphous white solid, exhibited an HR ESI MS molecular ion at m/z 603.3764 [M − H] − corresponding to the molecular formula C 32 H 52 N 4 O 7 with nine degrees of unsaturation. The difference in the molecular formulas between 1 and 5, C 9 H 9 NO 2 , suggests that 5 lacks a terminal tyrosine unit. The 1 H and 13 C NMR spectra of 5 (Tables 1 and 2) were comparable to those of 1, with the exceptions of the regions of the signals of the terminal tyrosine unit of 1. Analyses of the COSY, HSQC, HMBC, and ROESY spectra of 5 (Table S6) established the structure of the subunits and their sequence: NMeAdha-Tyr-NMeLeu-Pro. The absolute configuration of the chiral centers of the four subunits was determined as described for 1, establishing the structure of microginin KR604 as 5 ( Figure 1).
Microginin KR638 (6) (Tables 1 and 2) and molecular formula of 6 with those of 1, 2, and 5 suggested that 6 is the Cl-NMeAhda derivative of 5. The sequence of 6, Cl-NMeAdha-Tyr-NMeLeu-Pro, and the structures of the four substructures that compose 6 were established by analyses of COSY, HSQC, HMBC, and ROESY (Table S7). The absolute configuration of the chiral centers of microginin KR638 were determined by the same procedures described above, establishing structure 6 ( Figure 1).
Microginins KR781 (7) and KR815 (8) Figure 1). Microginins KR787 (7) and KR815 (8) are methyl ester derivatives of microginins KR767 (1) and KR801 (2), respectively. Microginin FR3 (9) was isolated as an amorphous white solid. It presented an HR ESI MS molecular adduct ion [M + H] + at m/z 728.3874 corresponding to the molecular formula C 37 H 53 N 5 O 10 with 14 degrees of unsaturation. The NMR spectra of 9 (Tables 1, 2 and S10) indicated the presence of an aliphatic chain and two para-substituted phenol rings, suggesting that 9 belongs to the microginin type of metabolites. Comparison of the NMR spectra of 9 and 1 suggested that these two compounds have different amino acid compositions (Table 1). In the 1 H NMR spectrum of 9, there was no indication of aliphatic methyls or N-methyls, which were found in 1. Analysis of the NMR data ( 1 H, 13 C, HSQC, HMBC, COSY, TOCSY, and ROESY) of 9 established the structures of five subunits and their sequence: Ahda-Thr-Pro-1 Tyr-2 Tyr (Table S10 and Figure 3). The sequence of 9 is identical to the one established by MS/MS for microginin FR3 collected from bloom materials in Germany [11]. The absolute configurations of the proteinogenic amino acids Thr, Pro, and Tyr were established by Marfey's method [14] to be of L configurations. The absolute configuration of C-3 of Ahda was established as R by applying the advanced Marfey's method [15]. As for 1-8, the absolute configuration of the Ahda-2-chiral center was proven by comparison of the chemical shifts of H-2, C-2, H-3, and C-3 of 9 with those of microginins for which the absolute configuration has been established unequivocally ( Table 3). The chemical shifts of carbons and protons of Ahda in 9 matched those of 2S,3R-Ahda in a previously reported microginin [7]. Interestingly, the absolute configurations of C-2 and C-3 in 9 were identical to those of microginin FR1, which were established by X-ray diffraction analysis [18]. Based on this evidence, the structure of 9 was established ( Figure 1) and shown to be that of microginin FR3.
Mar. Drugs 2018, 16,x FOR PEER REVIEW 8 of 13 microginin type of metabolites. Comparison of the NMR spectra of 9 and 1 suggested that these two compounds have different amino acid compositions (Table 1). In the 1 H NMR spectrum of 9, there was no indication of aliphatic methyls or N-methyls, which were found in 1. Analysis of the NMR data ( 1 H, 13 C, HSQC, HMBC, COSY, TOCSY, and ROESY) of 9 established the structures of five subunits and their sequence: Ahda-Thr-Pro-1 Tyr-2 Tyr (Table S10 and Figure 3). The sequence of 9 is identical to the one established by MS/MS for microginin FR3 collected from bloom materials in Germany [11]. The absolute configurations of the proteinogenic amino acids Thr, Pro, and Tyr were established by Marfey's method [14] to be of L configurations. The absolute configuration of C-3 of Ahda was established as R by applying the advanced Marfey's method [15]. As for 1-8, the absolute configuration of the Ahda-2-chiral center was proven by comparison of the chemical shifts of H-2, C-2, H-3, and C-3 of 9 with those of microginins for which the absolute configuration has been established unequivocally ( Table 3). The chemical shifts of carbons and protons of Ahda in 9 matched those of 2S,3R-Ahda in a previously reported microginin [7]. Interestingly, the absolute configurations of C-2 and C-3 in 9 were identical to those of microginin FR1, which were established by X-ray diffraction analysis [18]. Based on this evidence, the structure of 9 was established ( Figure  1) and shown to be that of microginin FR3. Microginin FR4 (10), an amorphous white solid, had an HR ESI MS molecular cluster ion at m/z 742.4031, corresponding to the molecular formula C38H55N5O10 with 14 degrees of unsaturation. Comparison of the 1 H and 13 C NMR spectra (Tables 1 and 2) of 10 with those of 9 revealed signals of an additional NMe (δH 2.50 brs, δC 31.3 CH3) and differences in the chemical shifts of the ammonium signals (δH 8.36 brs and 8.17 brs) and a methine (δH 3.30 m, δC 60.1 CH). The differences between 9 and 10 in the NMR spectra and in molecular formulas suggest that the Adha ammonium group in 9 is substituted by a methyl ammonium group in 10. Full analysis of the NMR data ( 1 H, 13 C, HSQC, HMBC, COSY, TOCSY, and ROESY) of 10 established the structures of the five subunits and the sequence, NMeAdha-Thr-Pro-1 Tyr-2 Tyr (Table S11). The absolute configurations of the proteinogenic amino acids, Thr, Pro, and Tyr, and of C-3 of NMe-Ahda were established by the same procedures described for 9. In 10, the proteinogenic amino acids are in the L configuration and the NMeAhda is 3R. Unfortunately, the chemical shifts of NMeAhda C-2, C-3, H-2, and H-3 did not match those of the Ahda moieties of microginins with proven absolute configurations ( Table 3). The structural similarities of 9 and 10 and the similar splitting of the Ahda H-2 (δH 4.10 t, J = 4.6 Hz in 9; δH 4.18 t, J = 4.4 Hz in 10), which differed from those of 2 and 4 (δH 4.30 brd, J = 5.6 Hz in 2; δH 4.14 dd, J = 5.4, 2.5 Hz in 4), suggest that Ahda moieties of 9 and 10 share the same absolute configuration. Furthermore, comparison of the changes in the chemical shifts of C-2, C-3, H-2, and H-3 in pairs of compounds with Ahda/NMeAhda and defined absolute configurations of 2S,3S (microginin 51-A and microginin 51-B [10], ∆δC-2 + 2.3, ∆δC-3 − 6.9, ∆δH-2 − 0.14, ∆δH-3 + 0.02 ppm) and 2R,3R (2 and 4, ∆δC-2 + 2.1, ∆δC-3 − 7.1, ∆δH-2 − 0.15, ∆δH-3 0 ppm) with those of 9 and 10 (∆δC-2 + 1.4, ∆δC-3 − 6.9, ∆δH-2 − 0.08, ∆δH-3 − 0.08 ppm) suggest that the absolute configurations of the chiral centers of Ahda moieties in 9 and 10 are different from those of 1-8 and from those of microginins 51-A and 51-B. These findings, together with the 3R configuration established for the NMeAhda, lead us to suggest the 2S,3R absolute Microginin FR4 (10), an amorphous white solid, had an HR ESI MS molecular cluster ion at m/z 742.4031, corresponding to the molecular formula C 38 H 55 N 5 O 10 with 14 degrees of unsaturation. Comparison of the 1 H and 13 C NMR spectra (Tables 1 and 2) of 10 with those of 9 revealed signals of an additional NMe (δ H 2.50 brs, δ C 31.3 CH 3 ) and differences in the chemical shifts of the ammonium signals (δ H 8.36 brs and 8.17 brs) and a methine (δ H 3.30 m, δ C 60.1 CH). The differences between 9 and 10 in the NMR spectra and in molecular formulas suggest that the Adha ammonium group in 9 is substituted by a methyl ammonium group in 10. Full analysis of the NMR data ( 1 H, 13 C, HSQC, HMBC, COSY, TOCSY, and ROESY) of 10 established the structures of the five subunits and the sequence, NMeAdha-Thr-Pro-1 Tyr-2 Tyr (Table S11). The absolute configurations of the proteinogenic amino acids, Thr, Pro, and Tyr, and of C-3 of NMe-Ahda were established by the same procedures described for 9. In 10, the proteinogenic amino acids are in the L configuration and the NMeAhda is 3R. Unfortunately, the chemical shifts of NMeAhda C-2, C-3, H-2, and H-3 did not match those of the Ahda moieties of microginins with proven absolute configurations ( Table 3). The structural similarities of 9 and 10 and the similar splitting of the Ahda H-2 (δ H 4.10 t, J = 4.6 Hz in 9; δ H 4.18 t, J = 4.4 Hz in 10), which differed from those of 2 and 4 (δ H 4.30 brd, J = 5.6 Hz in 2; δ H 4.14 dd, J = 5.4, 2.5 Hz in 4), suggest that Ahda moieties of 9 and 10 share the same absolute configuration. Furthermore, comparison of the changes in the chemical shifts of C-2, C-3, H-2, and H-3 in pairs of compounds with Ahda/NMeAhda and defined absolute configurations of 2S,3S (microginin 51-A and microginin 51-B [10], ∆δ C-2 + 2.3, ∆δ C-3 − 6.9, ∆δ H-2 − 0.14, ∆δ H-3 + 0.02 ppm) and 2R,3R (2 and 4, ∆δ C-2 + 2.1, ∆δ C-3 − 7.1, ∆δ H-2 − 0.15, ∆δ H-3 0 ppm) with those of 9 and 10 (∆δ C-2 + 1.4, ∆δ C-3 − 6.9, ∆δ H-2 − 0.08, ∆δ H-3 − 0.08 ppm) suggest that the absolute configurations of the chiral centers of Ahda moieties in 9 and 10 are different from those of 1-8 and from those of microginins 51-A and 51-B. These findings, together with the 3R configuration established for the NMeAhda, lead us to suggest the 2S,3R absolute configuration for NMeAhda in 10. Based on these arguments, structure 10 ( Figure 1) is proposed to be that of the previously identified microginin, FR4 [11].

General Experimental Procedures
Optical rotation values were obtained on a Jasco P-1010 polarimeter (Jasco, Oklahoma City, OK, USA) at the sodium D line (589 nm). UV spectra were recorded on an Agilent 8453 spectrophotometer (Agilent, Santa Clara, CA, USA). IR spectra were recorded on a Bruker Tensor 27 FT-IR instrument (Bruker, Billerica, MA, USA). NMR spectra were recorded on Bruker Avance and Avance III spectrometers (Bruker, Karlsruhe, Germany) at 500.13 MHz for 1 H and 125.76 MHz for 13 C. DEPT, COSY-45, gTOCSY, gROESY, gHSQC, gHMQC, and gHMBC spectra were recorded using standard Bruker pulse sequences (Bruker, Karlsruhe, Germany). NMR chemical shifts were referenced to TMS δ H and δ C = 0 ppm. HR ESI MS were recorded on a Waters MaldiSynapt instrument (Waters, Milford, MA, USA), and LC ESI MS spectra were recorded on a Waters Xevo TQD instrument (Waters, Milford, MA, USA). HPLC separations were performed on a Merck Hitachi HPLC system (L-6200 Intelligent pump and L-4200 UV-VIS detector, Hitachi, Tokyo, Japan), a JASCO P4-2080 plus HPLC system with a multiwavelength detector (Jasco International, Tokyo, Japan), and an Agilent 1100 Series HPLC system (Agilent, Santa Clara, CA, USA).

Biological Material
Microcystis sp. (collection No. IL-405) was collected from the Kishon reservoir in November 2009. The sample was identified by microscopic observation as a Microcystis sp. A lyophilized voucher sample (IL-405) was deposited in the culture collection of Tel Aviv University (Tel Aviv, Israel).

Determination of the Absolute Configuration of the Amino Acids by Marfey's Method
Compounds 1-10 (0.5 mg each) were hydrolyzed in 6 N HCl (1 mL). The reaction mixture was maintained in a sealed glass bomb at 104 • C for 18 h. The acid was removed in vacuo, and the residue was suspended in 250 µL of H 2 O. A solution of 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide (FDAA) [14] in acetone (0.03 M, 20 µL per each amino acid in the peptide) and NaHCO 3 (1 M, 20 µL per each amino acid) were added to the reaction vessel. The reaction mixture was stirred at 40 • C for 2.5 h in the dark. HCl (2 M, 10 µL per each amino acid) was added to the reaction vessel, and the solution was evaporated in vacuo. The FDAA-amino acid derivatives from the hydrolysate were dissolved in 1 mL CH 3 CN and compared with standard FDAA-amino acids by HPLC analysis: LiChroCART RP-18 column (5 µm, 250 mm × 4.6 mm), flow rate 1 mL/min, UV detection at 340 nm, linear gradient elution from 0.1% aq. TFA buffer (pH 3) to 6:4 MeCN:0.1% aq. TFA buffer (pH 3) within 60 min. The absolute configuration of each amino acid was determined by spiking the derivatized hydrolysates with a D,L-mixture of the standard derivatized amino acids.
The absolute configurations of C-3 of the Ahda derivatives of compounds 1-10 were analyzed by the advanced Marfey's method [15]. A 0.5-mg portion of each compound was hydrolyzed, as described above, divided into two portions and derivatized, one with L-FDAA and the other with D-FDAA. The two samples of L-and D-FDAA derivatives were analyzed by ESI LC MS. The analysis was performed on a Waters Acquity UPLC coupled with a UV detector (Waters Acquity-TUV detector) and mass spectrometer (Waters Xevo TQD) on a C18 (1.7 µm, 2.1 Å,~100 mm) column (Waters). The mobile phase compositions were (A) 95:5 H 2 O/MeCN, 0.1% formic acid and (B) MeCN, 0.1% formic acid. The elution gradient was as follows: 1 min of 100% A, linear gradient to 40% B over 25 min, and hold for 4 min. Samples of 10 µL were injected, and the flow rate was 0.5 mL/min. The UV detector was set to 340 nm, and the mass spectrometer was operated in both negative and positive ion modes, scanning between 200 and 650 mass units. The interpretation of the data was conducted after the run on both positive and negative ion modes using Waters MassLynx software (v4.1, Waters, Milford, MA, USA).

Aminopeptidase M Inhibitory Assay
Aminopeptidase M inhibitory activity was determined according to the procedure described by Ishida et al. [10]. A reaction mixture composed of 20 µL of L-leucine-p-nitroanilide (2 mM in 0.1 M Tris-HCl buffer at pH 7.0), 50 µL of 0.1 M Tris-HCl buffer (pH 7.0) and 20 µL of test solution was added to each well of a 96-well microtiter plate. The plate was incubated at 37 • C for 5 min; thereafter, 10 µL of porcine kidney microsomal aminopeptidase M [EC Number 232-942-5, Sigma-Aldrich (St. Louis, MO, USA), 5.5 µL in 3.5 M (NH 4 ) 2 SO 4 ] was added to each well, and the absorbance at 405 nm was measured for 30 min. The microginin samples were dissolved in DMSO at a concentration of 1 mg/mL, and the IC 50 values were determined by analysis of a series of dilutions (from 45.5 µM to 0.00069 µM). A sigmoidal curve of the enzyme inhibition versus the concentration of the inhibitor was observed that was fit to a 4-parameter logistic model [19].

Conclusions
In the current research, we isolated ten microginins produced by a Microcystis sp. collected from the Kishon Reservoir in Israel. These microginins appear to be encoded by two distinct gene clusters that yield different sequences and absolute configurations of the chiral centers of the Ahda moiety. Compounds 1 to 6 share the same sequence: (2R,3R)-Ahda-L-Tyr-L-NMeLeu-L-Pro-X (X=H or L-Tyr). Compounds 9 and 10 share a different sequence: (2S,3R)-Ahda-L-Thr-L-Pro-L-Tyr-L-Tyr. Furthermore, these linear peptides are additionally altered by N-methylation of Ahda-amine group and, in the case of 1-6, by mono-or dichlorination of Ahda terminal methyl group. Although inhibitory activity depends on the sequences of the microginins [10] and Ahda-N-methylation, chlorination of the Ahda terminal methyl group did not influence the extent of the aminopeptidase M inhibition. The augmentation of the diversity of the microginins by halogenation is common to cyanobacteria metabolites, including other protease inhibitors, such as the aeruginosins and micropeptins that are produced by Microcystis spp. [1,20,21]. The reason for augmentation of the diversity of these groups of metabolites by mono-, di-, and tri-halogenation and hetero-halogenation is not clear. It might reflect ecological condition or stress, be a signal for managing the cyanobacteria colony, or function in quorum sensing. Our current research is aimed at clarifying this misunderstood phenomenon.