Alteration in Amino Acid Composition and Configuration in Cyanobacterial Peptides Affects Biological Activity

Abstract

Cyanobacterial blooms are a significant environmental concern due to their production of toxic metabolites with potential impacts on ecosystem and human health. Microcystin-LR and microcystin congeners are a major concern with respect to human exposure and intoxication, but there are hundreds of characterized cyanobacterial peptides and metabolites that are of interest for their environmental impact in a variety of classes such as cyanopeptolins/micropeptins, microviridins, microginins, and anabaenopeptins. In this study, we report the isolation and characterization of new micropeptins (1–4), a new ferintoic acid (5) a new microginin (6), and three new microviridins (7–9) from a cyanobacterial bloom sample. The new micropeptins, in particular, exhibited unprecedented modifications in their amino acid composition and configuration, which differentiate them from previously known variants. Certain alterations significantly influenced their biological activity with respect to chymotrypsin inhibition and human neutrophil elastase inhibition, highlighting the potential ecological and biomedical relevance of these compounds. We report d-amino acid incorporation into the micropeptins for the first time providing new insights into the chemical diversity of cyanobacterial natural products. These results have important implications for understanding biosynthetic flexibility, the development of bioactive agents for therapeutic applications, and highlight the need for reference materials for mass spectrometry-based metabolite annotation.

Cyanobacterial harmful algal blooms (cyanoHABs) are complex microbial systems subject to alterations in abiotic and biotic parameters. Many of these blooms are known to produce a suite of peptidic hepatotoxins called microcystins (MCs), which can cause metabolic dysfunction-associated steatotic liver disease (MASLD) and acute liver failure in humans exposed to toxins in drinking water. , The microcystins are the most well studied cyanotoxin class with respect to liver impact and public health. However, there are many other groups of cyanobacterial peptides and cyanotoxins, which are of interest to environmental scientists, toxicologists, and chemists for both their potential environmental impact and therapeutic potential (e.g., cylinderospermopsin, guanitoxin, nodularins, and aplysiatoxins), and recent research has shown the impact of non microcystin cyanobacterial metabolites and the impact of toxic mixtures on ecotoxicology. Extracts from strains of cyanobacteria that did not produce microcystins still showed strong cytotoxic potency against zebrafish with 52% of the overall mortality attributed to micropeptin K139. There are over 3,000 cyanometabolites that have been isolated from cyanobacteria including over 300 microcystins and many cyanopeptides, , and there is an important knowledge gap in the contribution of each to environmental toxicity outside of the four monitored toxins by the World Health Organization (anatoxin-a, saxitoxin, microcystin-LR, and cylindrospermopsin). For example, certain microginins (linear peptides) induced DNA damage in HepG2 liver cells. Members of the cyanopeptolins/micropeptins (cyclic depsipeptides with the unique 3-amino-6-hydroxy-2-piperidone unit, Ahp) are potent protease inhibitors. They have also shown aquatic toxicity, and the ability to reduce inflammation in microglial cells, while the microviridins (large ribosomally synthesized tricyclic depsipeptides), which also inhibit proteases, are thought to be grazing deterrent molecules. Ferintoic acids (cyclic hexapeptides) containing an ureido moiety, are structurally related to the anabaenopeptins and oscillamides, but the characterized ferintoic acids have not shown the protease inhibition activity of their structurally related congeners such as oscillamide Y. There continues to be knowledge gap in the potential environmental impact of non microcystin cyanopeptides and other cyanometabolites.

The cyanopeptide classes generally have structural consistency in composition and configuration (although with substantial variation in amino acids). For instance, the generalized microcystin structure follows: cyclo-d-Ala-X-d-MeAsp-Z-Adda-d-γ-Glu-Mdha in which the variable amino acid residues (X and Z) are l-amino acids. The microginins are linear peptides hallmarked with a 3-amino-2-hydroxyacyl residue and amino acid additions. The cyanopeptolins (aka micropeptins, aeruginopeptins, nostopeptins, etc.) are designated by the 3-amino-6-hydroxy-2-piperidone (Ahp) residue and an ester typically formed by the hydroxy group of threonine and the carboxy group of the amino acid in position 1, − although there are examples of 3-hydroxy-4-methylproline substitution for the threonine. , Exocyclic residues in the cyanopeptolins/micropeptins are typically a fatty acid linked to glutamine or other amino acid, which is linked to the threonine (all in the l-configuration), which begins the cyclic portion. Ferintoic acids and anabaenopeptins contain the ureido linkage between an exocyclic amino acid and the cyclic portion of the molecule with d-Lys and then l-amino acids comprising the full structure. Often due to the peptidic nature of these compounds and the low abundance of many of them in culture or field collections, structure characterization is completed using LC-MS or LC-HRMS. However, this approach is limited by an inability to perform configuration analysis and a lack of certified reference material to match retention times between an analyte and an authentic standard.

In this report, we identified and characterized four new micropeptins (1–4) three of which contained an unusual amino acid diastereomer (1) or d-amino acid (3,4), one new ferintoic acid (A variant) containing a d-amino acid other than lysine (5), one new microginin (6), and three new microviridins (7–9). Additionally, one micropeptin (2), possessed an ester linkage through serine instead of threonine. Chymotrypsin and neutrophil elastase inhibition assays showed that the presence of the serine removed inhibitory activity from the micropeptin, while the presence of l-allo-threonine significantly improved inhibitory activity. This work illustrates the vast chemical diversity in cyanoHABs and expands on the structure variations that can occur in these metabolites and provides a note of caution for LC-MS/MS characterization and emphasizes the need for more certified standards in cyanometabolite analysis.

Results and Discussion

Structure Characterization of New Cyanopeptides

As we continued to mine a set of chromatography fractions derived from a cyanobacterial harmful algal bloom from which new microcystins were previously isolated and characterized, , we found that there were certain major (Figures S1 and S2) and many minor peaks (Figures S2 and S3) in chromatograms following LC-MS analysis. Many of these analytes showed several m/z values of 1019 and 1005 (Na⁺ adducts of the previously characterized micropeptins 996 and 982, respectively), but the retention times of these analytes were different from that of the previously characterized micropeptins and we isolated and characterized these metabolites in the current study. ESI-QTOF analysis of 1 gave m/z 1005.4691 [M + Na]⁺ supporting a molecular formula of C₅₁H₆₆N₈O₁₂ which was the same formula as micropeptin 982. However, an overlay of ¹H NMR and HSQC spectra of 1 and micropeptin 982 showed several differences in chemical shift values (Figure A), We constructed the planar structure of 1 using 1D and 2D NMR (Figures S4–S10, Table ) and the structure was identical to that of micropeptin 982 including the relative configuration of the Ahp residue determined via NOE correlations. We hypothesized that the difference between 1 and micropeptin 982 was in the configuration of an amino acid, which proved to be the case as 1 contained l-allo-Thr following Marfey’s analysis (Figure B, Table S8), while the other amino acids were all of the l-configuration. We named 1 micropeptin 982 (l-allo-Thr).

1.

(A) ¹H NMR overlay of 1 and micropeptin 982 showing key differences in resonances. (B) HPLC-DAD analysis of l-FDVA derivatized amino acid standards and the derivatized hydrolysate of 1 showing a match to l-allo-Thr.

1. NMR Data for Micropeptin 982 (l-allo-Thr) (1) (500 MHz for ¹H NMR, 125 MHz for ¹³C NMR; DMSO-d ₆).

Position	δC, mult	δH, mult, J (Hz)	HMBC	ROESY
Val-1	170.8, C
2	55.8, CH	4.71, m	1
3	30.4, CH	2.04, m
4	18.9, CH3	0.86, ovlp
5	16.9, CH3	0.71, d (6.4)	4
NH		7.42, ovlp		N-MePhe-2, N-Me
N-MePhe-1	169.4, C
2	60.5, CH	5.01, ovlp	1	Val-NH
3a	33.4, CH2	3.21, m
3b		2.84, ovlp
4	137.8, C
5/9	129.1, CH	7.24, d (7.5)		Phe-2
6/8	128.3, CH	7.41, t (7.6)	4
7	126.3, CH	7.31, t (7.5)
N-Me	30.4, CH3	2.79, s		Val-NH
Phe-1	170.7, C
2	50.1, CH	4.72, m	1	N-MePhe-5, 9, Ahp-5
3a	35.2, CH2	2.84, ovlp
3b		1.67, ovlp
4	135.1, C		5, 9
5/9	129.0, CH	6.77, d (7.2)	7	Ahp-5
6/8	127.4, CH	7.17, t (7.4)
7	125.9, CH	7.14, t (7.4)	5, 9
Ahp-1	168.9
2	48.8, CH	3.59, m	1
3a	21.4, CH2	2.40, m
3b		1.60, m
4a	29.2, CH2	1.68, m
4b		1.51, m
5	73.7, CH	5.03, ovlp		phe-2, 5, 9
NH		7.09, d (9.0)		Tyr-2, Tyr-NH
OH		6.04, br
Tyr-1	170.9, C
2	53.0, CH	4.36, m	1	Ahp-NH
3a	35.0, CH2	3.12, dd (14.4, 4.0)	5, 9
3b		2.52, m
4	128.6, C
5/9	129.7, CH	6.90, d (8.4)	7
6/8	114.9, CH	6.57, d (8.4)	4, 7
7	155.8, C
NH		8.47, d (8.7)		Ahp-NH, Thr-2, 3

ESI-QTOF mass spectrometry analysis of 2 gave an m/z value of 1005.700 [M + Na]⁺, which also supported a molecular formula of C₅₁H₆₆N₈O₁₂. Analysis of 1D and 2D NMR allowed us to assign many of the same substituents found in 1 and micropeptin 982 (valine, N-methyl-phenylalanine, phenylalanine, Ahp, glutamine, and butyric acid) (Figures S11–S14). However, a TOCSY spin system of methylene protons δ_H 2.12, 1.60 and 2.46, 2.29 correlated to an α-proton δ_H 4.09 supported the assignment of a homologated tyrosine residue and deshielded methylene protons δ_H 4.95 and 4.10 correlated to δ_C 64.5 by examination of HSQC supported the assignment of a serine residue as the amino acid, instead of threonine, completing the cyclic portion of the molecule, which is an unprecedented substitution in the cyanopeptolin/micropeptin class of molecules. We named 2 micropeptin 982 (l-Ser). All amino acids in 2 were determined to be of the l-configuration following Marfey’s analysis (Table S8 and Figure S15).

ESI-QTOF mass spectrometry analysis of 3 gave an m/z value of 1005.4685 [M + Na]⁺ supporting a molecular formula of C₅₁H₆₆N₈O₁₂, which was identical to micropeptin 982. 1D and 2D NMR analysis confirmed the planar structure of 3 as identical to that of micropeptin 982 (Figures S16–S21), but the NMR data overlay with micropeptin 982 showed differences in the chemical shift of multiple signals. Marfey’s analysis explained this difference with 3 containing a d-Gln residue (converted to glutamic acid during acid hydrolysis) (Table S8, Figure S22).

ESI-QTOF mass spectrometry analysis of 4 gave an m/z value of 1019.4854 [M + Na]⁺, which supported a molecular formula of C₅₂H₆₈N₈O₁₂ identical to that of micropeptin 996. NMR analysis of primarily HQSC, TOCSY, and NOE correlations constructed the same composition and sequence of 4 to that of micropeptin 996 (Figures S23–S26). However, the ¹H NMR spectra of the two compounds did not overlap and Marfey’s analysis showed that the glutamine was of the d-configuration in 4 (Figure S27), while the remaining amino acids were of the l-configuration (Table S8). We named 4 micropeptin 996 (d-Gln).

This is the first report of amino acid diastereomers or d-amino acid incorporation into cyanopeptolins/micropeptins. In nonribosomal peptide synthetase (NRPS) biosynthetic pathways, d-amino acids are typically incorporated into peptides by embedded epimerization domains after the l-amino acid has been activated by an adenylation domain. However, there is evidence of the direct selection and activation by adenylation domains. There are multiple characterized cyanopeptolin/micropeptin gene clusters from cyanobacteria and bacteria, − and none possess an embedded epimerization domain, which provokes intriguing questions as to the substrate specificity of the A domains in these pathways, which will require further biochemically study or perhaps there are undiscovered pathways from certain cyanobacterial strains with embedded epimerization domains or possibly there are epimerase enzymes that are not in the biosynthetic clusters but act in a trans manner. Intriguingly, while small amounts of most of the new micropeptins were isolated, the amount of micropeptin 982 (d-Gln) (3) isolated (132.5 mg) was more than the amount of microcystin-LR isolated from the original cyanoHAB extract used in this study (61 mg). The composition of known micropeptins is nicely reviewed in Hasan-Amer and Carmeli (2017), and we add new information to the variability in amino acid units in this group of compounds with serine and l-allo-threonine possible in position 6 and d-Gln incorporation in position 7.

Comparison of LC-MS/MS Spectra of the 1–4 and Micropeptin 996

Compounds 1–3 all had high resolution mass spectrometry measurements that supported the same molecular formula (Figure S28). The MS/MS spectra of the [M+H–H₂O]⁺ ions of 1–3 were very similar but we could discriminate the serine-containing micropeptin from the l-Thr and l-allo-thr containing micropeptins. All metabolites had product ions corresponding to proposed fragments of Ahp-Phe-N-Me-Phe-H₂O (m/z 404) and Ahp-Phe-H₂O (m/z 243). However, m/z 282 in 1 and 3 likely corresponded to a BTA-Gln-Thr-H₂O fragment, while 2 contained a fragment 14 Da lower at m/z 268 supporting the substitution of Ser for Thr (Figure S29). Conversely, the MS/MS spectrum of 2 contained a fragment at m/z 581 likely corresponding to Htyr-Ahp-Phe-N-Me-Phe-H₂O, which was 14 Da higher than the fragment at 567 in 1 and 3, which likely corresponded to Tyr-Ahp-Phe-N-Me-Phe-H₂O. Compounds 1 and 3 showed nearly identical MS/MS fragmentation patterns (Figure S29) as did 4 and micropeptin 996 (Figure S30) for their [M+H–H₂O]⁺ precursor m/z ions (m/z 965 for 1 and 3 and 979 for 4 and micropeptin 996), which illustrates the limitations of the abilities of MS/MS to determine the identity of peptides with alterations in amino acid configuration and the need for more chromatographic standards.

HRMS analysis of 5 gave an m/z value of 867.4398 [M + H]⁺ suggesting a molecular formula of C₄₆H₅₈N₈O₉ (Figure S31). Examination of NMR data resulted in a planar structure identical to that of ferintoic acid A (Figures S31–S34). However, there were chemical shift differences assigned for the Trp residue between 5 and the published data for ferintoic acid A (δ_H 4.15 in 5; δ_H 4.42 for ferintoic acid A). Marfey’s analysis showed that the tryptophan was in the d-configuration in 5 (Figure S35) as was the Lys (consistent with ferintoic acid A) while the remaining amino acids were of the l-configuration (Table S8). We named this metabolite ferintoic acid A (d-Trp). As with the micropeptins, the available biosynthetic evidence from anaebanopeptin biosynthetic pathwys shows an epimerization domain following the A domain incorporating the Lys subunit, which explains the d-lysine incorporation in the anabaenopeptins and ferintoic acids. However, our work provokes questions into substrate specificity in other A domains in these pathways or the presence of other gene clusters with epimerization domains or trans acting enzymes.

After characterizing the major and minor metabolites in the chromatography fractions containing micropeptins, we also examined a series of fractions with higher molecular weight molecules (>1700 Da, determined by MALDI-TOF), which we speculated contained microviridins. We isolated four analytes, three of which were new microviridins (7–9) and one which was a new microginin (6). MALDI-TOF of 6 gave an m/z value of 653.2150 [M + H]⁺ suggesting a molecular formula of C₃₀H₃₈Cl₂N₄O₈. The presence of two chlorine atoms and examination of the NMR data suggested a microginin-like compound, many of which are hallmarked by dichlorination (Figures S36–S42). HSQC correlations between δ_C 71.0 and δ_H 3.88 and between δ_C 52.9 and δ_H 3.14 supported the assignment of a 3-amino-2-hydroxy acyl moiety, another structural feature of microginins. We assigned an amino acid portion of the molecule using HMBC and TOCSY correlations (Tyr¹-N-methyl-Tyr²-Pro) and the examination of a TOCSY spin system characterized a dichlorinated 3-amino-2-hydroxyhexanoic acid, and 6 proved to be an analog to the previously characterized microginin 680. The amino acids were all in the l-configuration (Table S7). We observed a small coupling between H-2 and H-3 of the 3-amino-2-hydroxyhexanoic acid moiety in 6 (3.6 Hz). This was obtained from the ¹H NMR and confirmed by examination of the DQF-COSY. We estimated that there were small ¹H–¹³C couplings between both H-3 and C-1 and between H-2 and C-4 by examining HMBC data and determining the ratio of the relative magnitudes of ¹H–¹³C cross peaks measured in the HMBC spectrum with respect to a common proton. An intense peak indicated a large ^2,3 J _CH, a weak or missing peak a small ^2,3 J _CH. We compared this assessment to the rotamers possible for vicinal methine carbons, assigning a threo relationship, which supported a relative configuration of 2S*,3R* for the 3-amino-2-hydroxyhexanoic acid moiety. We named 6 microginin 653. As shown in Figure S3, there were many other analytes isolated but we could not get enough purified material for full NMR characterization, and this illustrates that even while isolating many new compounds and minor metabolites, there are still more to characterize.

The m/z values of 7, 8, and 9 were measured at 1774.8089 [M + H]⁺, 1806.8310 [M + H]⁺, and 1782.8335 [M + H]⁺ using MALDI-TOF, which corresponded to molecular formulas of C₈₃H₁₁₁N₁₉O₂₅, C₈₄H₁₁₅N₁₉O₂₆, and C₈₂H₁₁₅N₁₉O₂₆, respectively. The NMR data of 7–9 were consistent with that of large peptide and the individual amino acids of 7 were assigned using HSQC and TOCSY correlations and amino acid connectivity was determined using HMBC and NOE correlations (Figures S43–S48) completing the structure a new tricyclic microviridin (Ac¹-Arg²-Ser³-Glu⁴-Thr⁵-Leu⁶-Lys⁷-Tyr⁸-Pro⁹-Ser¹⁰-Asp¹¹-Trp¹²-Glu¹³Glu¹⁴-Phe¹⁵) that we named microviridin 1773. Marfey’s analysis showed that all amino acids were of the l-configuration (Table S10). The difference between 7 and 8 was the presence of a methoxy signal in the HSQC spectrum (δ_C 50.8 and δ_H 3.54) in 8 which an HMBC correlation assigned to a methyl ester at the glutamate-13 residue (Figures S49–S55). We named 8 microviridin 1805 and propose, that as a likely artifact of 7, that it has the same absolute configuration. The final compound (9) was assigned using HSQC and TOCSY for the amino acid side chains and HMBC and NOE correlations for connectivity assigning a sequence of Ac¹-Arg²-Ser³-Ser⁴-Thr⁵-Leu⁶-Lys⁷-Tyr⁸-Pro⁹-Ser¹⁰-Asp¹¹-Trp¹²-Glu­(O-methyl)¹³-Glu¹⁴-Phe¹⁵ (Figures S56–S60). However, this molecule (microviridin 1781) only had one cyclic portion as the threonine ester was hydrolyzed as was the glutamate at position 13 (O-methyl). This compound also had all l-amino acids (Table S10). We speculate that 8 and 9 are isolation artifacts and that the intact natural products are tricyclic. These are the first microviridins known to the authors that contain arginine at the first amino acid position after the acetyl group.

Chymotrypsin and Neutrophil Elastase Inhibition Activity

Micropeptins have shown both trypsin and chymotrypsin inhibition with the amino acid connected to the N-terminus of the Ahp residue conferring specificity. A lipophilic residue (e.g., Htyr or Tyr) is associated with chymotrypsin inhibition. Thus, we examined whether these amino acid alterations including the incorporation of d-amino acids in these metabolites affected activity. Intriguingly, the l-allo-Thr incorporation in 1 conferred significantly improved inhibition activity compared to micropeptin 982 (d-Gln) (3) (IC₅₀ 0.85 ± 0.04 μM vs 2.5 ± 0.6 μM, p ≤ 0.0001 ANOVA with Tukey HSD Posthoc test), while the incorporation of the d-Gln in 4 did not significantly affect activity compared to micropeptin 996 (IC₅₀ 1.4 μM vs 1.5 μM, p = 0.4679). However, the incorporation of l-Ser in 2 completed reversed the inhibitory activity and no inhibition was observed (Figures A and S61A,B). In terms of structure–activity relationship, it appears that the presence of an l-allo-threonine or 3-hydroxy-4-methylproline does not decrease inhibitory activity. We speculate that there are steric effects that improve binding when these residues are incorporated, and these effects are lost when serine is incorporated. Additionally, it may be that the d-Gln, which is located on the molecule’s side chain is not heavily involved in binding interactions and altering its configuration has little effect on inhibitory activity. The chymotrypsin inhibition activity of 1–4 is similar to that of other previously described micropeptins and Ahp-containing depsipeptides from marine and freshwater cyanobacteria, but not near the most potent of these molecules tested (range 0.0025–10 μM).

2.

(A) Chymotrypsin inhibition assay and (B) neutrophil elastase assay for micropeptins 996 and 1–4.

A similar structure–activity pattern was observed when testing 1–4 and micropeptin 996 against human neutrophil elastase (Figures B and S62A,B). However, micropeptin 982 (l-allo-Thr) (1) showed much more potency compared to the other compounds tested with an IC₅₀ of 0.12 ± 0.002 μM, while the IC₅₀ values of micropeptin 996 (IC₅₀ = 0.83 ± 0.02 μM), 3 (IC₅₀ = 1.4 ± 0.15 μM), and 4 (IC₅₀ = 0.84 ± 0.12 μM) were all around 1 μM. The activity of 1 is close to that of the most potent human neutrophil elastase inhibitors discovered from marine cyanobacteria, the symplostatins and lyngbyastatins, which contain a 2-amino-2-butenoic acid subunit (Abu), and the activities of synthesized Ahp-cyclodepsipeptides. It would be intriguing to construct a micropeptin analog with both the Abu and l-allo-threonine subunits to see if it markedly improves inhibition activity.

Microviridins have shown the ability to inhibit several different serine proteases including elastase, trypsin, and chymotrypsin. Microviridin 1777 was a potent chymotrypsin inhibitor with an IC₅₀ value of 100 nM. The new microviridins that we isolated and characterized (7-9) also showed chymotrypsin inhibition activity although not the potency of microviridin 1777. Microviridins 1773 (7), 1805 (8) and 1781 (9) had an IC₅₀ values of 7.3 ± 1.3 μM, 1.5 ± 0.1 μM, and 4.0 ± 0.4 μM, respectively (Figure S63) illustrating that partial hydrolysis of the microviridins does not reduce activity. Ferintoic acid A did not show chymotrypsin inhibition, and the d-Trp-containing analog (4) was also not active up to a concentration of 10 μM.

While microcystins pose a significant threat to aquatic health, cyanopeptide production of non microcystin metabolites (e.g., cyclamides) in culture can be comparable to that of microcystins, and non microcystin metabolites have shown impacts in aquatic toxicity. Toxicological knowledge on the hundreds of cyanopeptides produced in cyanoHABs other than microcystins is relatively scant as is the ability to identify and quantify these lesser-known peptides due to lack of reference standards and methods. Characterization of the full complement of cyanometabolites and their environmental impact is important as the frequency of cyanoHABs has increased in large lakes over the past two decades, and the association with increasing temperatures portends more of these events in the future. , We add a significant new layer to structure characterization and annotation of cyanopeptides by showing for the first time the incorporation of diastereomeric amino acids and d-amino acids into the micropeptins. Additionally, in the case of 1, significant increases in biological activity were observed with l-allo-Thr incorporation, while the serine incorporation in 2 lacked any inhibitory activity. We do not yet know how widespread these congeners are and in what concentration they are found in the environment. We isolated these metabolites from a single bloom event. But future studies will improve the temporal and spatial resolution of bloom metabolomes with respect to these molecules and others. Ultimately, this information provokes intriguing questions surrounding substrate specificity and selection during biosynthesis and increases the challenges in LC-MS/MS based annotation and configuration assignment by analogy.

Experimental Section

General Experimental Procedures

NMR spectra were recorded on a Bruker 500 MHz Ascend Avance III NMR instrument and chemical shifts reported for 1–9 were referenced to the residual solvent peak of (CD₃)₂SO (δ_H 2.50 and δ_C 39.5). HRESIMS analysis was performed using a Bruker autoflex maX MALDI-TOF system. Additional LC-HRMS and LC-HRMS data were collected on an Agilent Revident Q-TOF system equipped with a Jet Stream source and 1290 Infinity II Bio LC (with multisampler and multicolumn thermostat) and MassHunter Workstation software. MS/MS data were collected in Auto MS/MS mode with a collision energy of 40 V. The MS/MS acquisition range was from m/z 100 - m/z 1700 at a rate of 3 spectra/s. Semipreparative HPLC was carried out using an Agilent 1260 system equipped with a vacuum degasser, autosampler, and diode array detector.

Isolation and Structure Elucidation of 1–9

Isolation of 1

A chromatography fraction library purchased from Biosortia Microbiomics was subjected to LC-MS and NMR analysis. This chromatography library was derived from a 2014 cyanobacterial bloom in Muskegon, MI at a holding lagoon at the Muskegon County Wastewater Management facility on Maple Island Road, Muskegon, Michigan, USA, and several new microcystins and micropeptins have been isolated and characterized from fractions from this bloom biomass in previous work. ,, The cyanoHAB itself was dominated by Microcystis sp. The library originated from 43 kg of lyophilized cyanobacterial biomass that was subjected to repeated chromatographic separations over HP-20 resin, C18, and preparative reversed-phase HPLC resulting in a chromatography fraction library. A chromatography fraction showing high abundance of m/z 1005 (micropeptin 982 + Na⁺), also showed the presence of another compound with m/z 1005 at a different retention time. The fraction was dissolved in MeOH (10 mg/mL) and subjected to separation with a YMC 5 μm Chiral ART Cellulose-SB column (250 mm × 4.6 mm) under isocratic conditions with a mobile phase of 65% H₂O and 35% CH₃CN, each modified with 0.1% formic acid, at a flow rate of 1.00 mL/min. This process yielded 2.4 mg of compound 1 (t_R 6.7 min).

Isolation of 2–5

A second fraction was examined by LC-MS showing multiple major and minor analytes with m/z 1005 and 1019. It was dissolved in MeOH (10 mg/mL) and subjected to reversed-phase semipreparative HPLC using a YMC 5 μm ODS column (250 mm × 10 mm) under isocratic conditions with a mobile phase of 65% H₂O and 35% CH₃CN, each modified with 0.1% formic acid, at a flow rate of 3.00 mL/min. Four HPLC peaks (1–4) were collected and analyzed. This purification yielded 1.3 mg of ferintoic acid A (d-Trp) (t_R 7.5 min, peak 1, 5), 2.0 mg of micropeptin 982 (l-Ser) (t_R 9.3 min, peak 2, 2), and 132.5 mg of micropeptin 982 (d-Gln) (t_R 10.1 min, peak 3, 3). Peak 4 (t_R 10.7 min, 2.8 mg) was collected and further purified under the same conditions to remove micropeptin 982 (d-Gln) (3), leading to the isolation of 1.4 mg of micropeptin 996 (d-Gln) (t_R 10.7 min, peak 4.2, 4).

Isolation of 6–9

Mass spectrometry analysis (MALDI-TOF) of certain chromatography fractions showed the presence of ions indicating a higher molecular weight (m/z > 1700). These fractions were combined and the mixture was separated on a Phenomenex Luna 5 μm Phenyl-Hexyl column (250 mm × 10 mm) with a gradient method with H₂O and CH₃CN as mobile phase solvents each modified with 0.1% formic acid and a flow rate of 3.00 mL/min. Compounds were eluted with a gradient of CH₃CN in H₂O from 25% to 30% over 30 min followed by an increase to 100% CH₃CN to 35 min and a 100% hold of CH₃CN to 40 min with a return to initial conditions at 41 min. Compounds 6–9 were isolated at t_R 11.6 min (3.0 mg), t_R 17.6 min (1.0 mg), t_R 19.6 min (1.7 mg), and t_R 22.5 min (3.7 mg), respectively.

After isolation and characterization, structures were drawn in Scifinder and searched against the available database using the Structure Match function. Additionally, SMILES codes were generated and searched against the CyanoMetDB to ensure all newly reported compounds were new.

Micropeptin 982 (l-allo-Thr) (1)

colorless oil; [α] _D = −5 (c 0.2, MeOH); UV λ_max: 203, 230, 275 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table ; HRMS (ESI-QTOF) m/z calcd for C₅₁H₆₆N₈O₁₂Na⁺: 1005.4692 [M + Na]⁺; found 1005.4691 (error −0.10 ppm).

Micropeptin 982 (l-Ser) (2)

colorless oil; UV λ_max: 228, 280 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S1; HRMS (ESI-QTOF) m/z calcd for C₅₁H₆₆N₈O₁₂Na⁺: 1005.4692 [M + Na]⁺; found 1005.4700 (error 0.80 ppm).

Micropeptin 982 (d-Gln) (3)

colorless oil; [α] _D = −26 (c 0.02, MeOH); UV λ_max: 228, 278 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S2; HRMS (ESI-QTOF) m/z calcd for C₅₁H₆₆N₈O₁₂Na⁺: 1005.4692 [M + Na]⁺; found 1005.4685 (error −0.70 ppm).

Micropeptin 996 (d-Gln) (4)

colorless oil; UV λ_max: 225, 280 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S3; HRMS (ESI-QTOF) m/z calcd for C₅₂H₆₈N₈O₁₂Na⁺: 1019.4849 [M + Na]⁺; found 1019.4854 (error 0.50 ppm).

Ferintoic Acid A (d-Trp) (5)

colorless oil; UV λ_max: 224, 280 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) δ Trp: 4.15 (CH, m), 3.12, 3.02 (CH₂, 3.12, m, 3.02, dd, J = 13.5, 4.8 Hz), 7.48 (CH, d, J = 8.2 Hz), 6.91 (CH, t, J = 8.2 Hz), 7.01 (CH, ovlp), 7.28 (CH, d, J = 8.1 Hz), 7.03 (CH, ovlp), 6.17 (NH, m), 10.73 (NH, s); Lys: 3.99 (CH, m), 1.59 (CH₂, m), 1.27, 1.12 (CH₂, m), 1.44 (CH₂, m), 3.54 (CH₂, m), 6.64 (NH, ovlp), 7.11 (NH, ovlp); Phe³: 4.35 (CH, ddd, J = 12.5, 9.0, 3.4 Hz), 3.27, 2.76 (CH₂, m), 7.04 (2CH, d, J = 7.3 Hz), 7.18 (2CH, t, J = 7.2 Hz), 7.14 (CH, d, J = 7.1 Hz), 8.68 (NH, d, J = 8.7 Hz); N-Me-Ala: 4.78 (CH, m), 1.06 (CH₃, d, J = 6.6 Hz), 1.78 (NCH₃, s); Htyr: 4.72 (CH, m), 1.88, 1.72 (CH₂, m), 2.63, 2.43 (CH₂, m), 7.01 (2CH, d, J = 8.0 Hz), 6.68 (2CH, d, J = 8.3 Hz), 8.93 (NH, d, J = 4.6 Hz); Val⁶: 3.89 (CH, t, J = 7.8 Hz), 1.94 (CH, m), 1.03 (CH₃, d, J = 6.7 Hz), 0.94 (CH₃, d, J = 6.5 Hz), 7.09 (NH, ovlp); HRMS (ESI-QTOF) m/z calcd for C₄₆H₅₉N₈O₉ ⁺: 867.4400 [M + H]⁺; found 867.4398 (error −0.23 ppm).

Microginin 653 (6)

colorless oil; UV λ_max: 228, 276 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S4; HRMS (MALDI-TOF) m/z calcd for C₃₀H₃₉Cl₂N₄O₈ ⁺: 653.2139 [M + H]⁺; found 653.2150 (error 1.68 ppm).

Microviridin 1773 (7)

colorless oil; UV λ_max: 223, 278 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S5; HRMS (MALDI-TOF) m/z calcd for C₈₃H₁₁₂N₁₉O₂₅ ⁺: 1774.8071 [M + H]⁺; found 1774.8089 (error 1.01 ppm).

Microviridin 1805 (8)

colorless oil; UV λ_max: 225, 280 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S6; HRMS (MALDI-TOF) m/z calcd for C₈₄H₁₁₆N₁₉O₂₆ ⁺: 1806.8310 [M + H]⁺; found 1806.8333 (error 1.27 ppm).

Microviridin 1781 (9)

colorless oil; UV λ_max: 225, 280 (from HPLC-DAD); ¹H NMR (500 MHz, DMSO-d ₆) and ¹³C NMR (125 MHz, DMSO-d ₆), see Table S7; HRMS (MALDI-TOF) m/z calcd for C₈₂H₁₁₆N₁₉O₂₆ ⁺: 1782.8333 [M + H]⁺; found 1782.8335 (error 0.11 ppm).

Configuration Analysis of 1–6

To determine the absolute configuration of the α-amino acids in compound 1 and 6, 0.3 mg of each compound was hydrolyzed in 0.5 mL of 6 N HCl and heated at 90 °C for 16 h. The hydrolysate was dried, reconstituted in 100 μL of 1 M NaHCO₃ solution, and derivatized with 500 μL of a 1% solution of N-α-(2,4-dinitro-5-fluorophenyl)-l-alanine amide (l-FDAA) in acetone, followed by heating at 40 °C for 1 h. The reaction mixture was then cooled to room temperature and quenched with 50 μL of 2 N HCl. The hydrolysate was subsequently diluted 1:10 with 1:1 H₂O/CH₃CN to a final volume of 1 mL and analyzed by HPLC-DAD using a Kinetex C18 column (150 mm × 4.6 mm). The separation was performed with a gradient of CH₃CN in H₂O from 20% to 50% over 30 min and a return to initial conditions from 31 to 36 min, with a flow rate of 0.6 mL/min. The absolute configuration of each amino acid was determined by comparing the retention times of the hydrolysate with those of standard amino acids derivatized in the same way as the hydrolysate (Table S8).

Due to limited resolution in differentiating l-allo-Thr and l-Thr in 1, N-α-(2,4-dinitro-5-fluorophenyl)-l-valinamide (l-FDVA) was used in place of l-FDAA. Compounds 1-5 were analyzed with the same hydrolysis procedure but with l-FDVA. Compounds 7 and 9 were derivatized in a similar manner with some small changes (0.4 mg of 7 and 9 were heated at 130 °C in 0.8 mL of HCl for 24 h). The hydrolysates and the l-FDVA derivatized amino acid standards were subjected to LC-MS analysis using the same gradient method with a Luna 5 μm C18 column (150 mm × 2 mm) at a flow rate of 0.4 mL/min again comparing retention times of the derivatized hydrolysate to those of l-FDVA derivatized l- and d-amino acid standards (Table S8–S10).

Chymotrypsin and Neutrophil Elastase Inhibition Assays

The chymotrypsin inhibition assay was performed using the Chymotrypsin Inhibitor Assay Kit (BioAssay Systems, Hayward, CA, USA) with α-chymotrypsin from Worthington Biochemical Corporation (Lakewood, NJ, USA). For each assay, 30 μL of enzyme solution (10 U/mL in assay buffer) was mixed with 12 μL of micropeptin 996, micropeptin 982, 1-5 and 7–9 in a dilution series (ranging from 10 μM to 1 nM in DMSO) in a 96-well plate and micropeptin 996 was used as a positive control. The plate was gently tapped to mix and incubated at room temperature for 15 min. Following incubation, 78 μL of working reagent, consisting of 1% substrate and 99% assay buffer, was added to each well. Fluorescence at λ_ex/em = 485/530 nm was measured every 30 s for 30 min using a Tecan Spark multimode plate reader. The elastase inhibition assay was conducted using a Neutrophil Elastase Inhibitor Screening Kit (Abcam, Cambridge, UK). The assay was conducted in accordance with the manufacturer’s instructions. Micropeptin 996 and 1–4 were solubilized in DMSO in a dilution series to yield final reaction mixtures from 10 μM to 1 nM. λ_ex/em = 400/505 nm was measured every 60 s for 10 min using a Tecan Spark multimode plate reader. Data analysis and IC₅₀ calculations for both assays were performed using Prism (GraphPad, San Diego, CA, USA) and the online IC₅₀ calculator available from AAT Bioquest.

The NMR data for all characterized compounds has been uploaded to NP-MRD [ID#s: NP0350757 (1), NP0350758 (2), NP0351036 (3), NP0350759 (4), NP0350760 (5), NP0350762 (6), NP0350763 (7), NP0350764 (8), NP0350765 (9)]. These data will be made publicly available after publication and can be found at https://np-mrd.org/.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jnatprod.5c00619.

• NMR and mass spectrometry data, Marfey’s analysis, and bioassay data (PDF)

M. J. B. conceived the study. R. X., H. X., and A. M. S. M. performed experimental work, data acquisition, and data analysis. The first draft of the manuscript was written by M. J. B. with editing by all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.