Fatty Acid Substitutions Modulate the Cytotoxicity of Puwainaphycins/Minutissamides Isolated from the Baltic Sea Cyanobacterium Nodularia harveyana UHCC-0300.

Puwainaphycins (PUW) and minutissamides (MIN) are structurally homologous cyclic lipopeptides that exhibit high structural variability and possess antifungal and cytotoxic activities. While only a minor variation can be found in the amino acid composition of the peptide cycle, the fatty acid (FA) moiety varies largely. The effect of FA functionalization on the bioactivity of PUW/MIN chemical variants is poorly understood. A rapid and selective liquid chromatography-mass spectrometry-based method led us to identify 13 PUW/MIN (1-13) chemical variants from the benthic cyanobacterium Nodularia harveyana strain UHCC-0300 from the Baltic Sea. Five new variants identified were designated as PUW H (1), PUW I (2), PUW J (4), PUW K (10), and PUW L (13) and varied slightly in the peptidic core composition, but a larger variation was observed in the oxo-, chloro-, and hydroxy-substitutions on the FA moiety. To address the effect of FA substitution on the cytotoxic effect, the major variants (3 and 5-11) together with four other PUW/MIN variants (14-17) previously isolated were included in the study. The data obtained showed that hydroxylation of the FA moiety abolishes the cytotoxicity or significantly reduces it when compared with the oxo-substituted C18-FA (compounds 5-8). The oxo-substitution had only a minor effect on the cytotoxicity of the compound when compared to variants bearing no substitution. The activity of PUW/MIN variants with chlorinated FA moieties varied depending on the position of the chlorine atom on the FA chain. This study also shows that variation in the amino acids distant from the FA moiety (position 4-8 of the peptide cycle) does not play an important role in determining the cytotoxicity of the compound. These findings confirmed that the lipophilicity of FA is essential to maintain the cytotoxicity of PUW/MIN lipopeptides. Further, a 63 kb puwainaphycin biosynthetic gene cluster from a draft genome of the N. harveyana strain UHCC-0300 was identified. This pathway encoded two specific lipoinitiation mechanisms as well as enzymes needed for the modification of the FA moiety. Examination on biosynthetic gene clusters and the structural variability of the produced PUW/MIN suggested different mechanisms of fatty-acyl-AMP ligase cooperation with accessory enzymes leading to a new set of PUW/MIN variants bearing differently substituted FA.

Introduction

Cyanobacteria are prolific producers of a large array of natural products, including peptides, polyketides, alkaloids, lipids, polyketones, and heterocyclic compounds.[1] They are also known to produce cyanotoxins during periodic blooms that have largely occurred in the last 20 years.[2−4] Among peptides, cyclic lipopeptides (CLPs) have received considerable attention for their wide range of bioactivities.[5−8] CLPs consist of a peptidic backbone with a diverse array of proteinogenic and non-proteinogenic amino acids attached to a fatty acid (FA) tail forming a cyclic ring structure.[9] The structure of CLPs substantially differs on amino acid composition and substitutions on the FA moiety. This unique structure and the resulting amphipathic molecular nature of CLPs promote integration into the membrane of the target organism, making them membrane-active compounds, such as surfactin, fengycin, iturins, and daptomycins.[10] CLPs are reported to possess significant antifungal[11] and cytotoxic activities[12−14] and also often used as biocontrol agents due to their antagonistic activity against a wide range of potential phytopathogens.[15−17]

The structure and the FA moiety of cyanobacterial CLPs differ substantially ranging from fully saturated FA chains, including puwainaphycins (PUW),[18] minutissamides (MIN),[19,20] muscotoxins,[21] and laxaphycins,[22] to polyunsaturated FAs such as in anabaenolysins.[12] PUW and MIN are structurally homologous amphipathic CLPs featuring a 10-membered peptide ring cyclized to form a lactam ring between an amino acid and an amino group bearing the FA moiety (β-amino FA) forming a lipid tail.[18−20,23] PUW/MIN has been reported from the genera Cylindrospermum, Symplocastrum, and Anabaena.[24] Twenty-one variants have been reported so far with the structural diversity arising from the differences in their peptide core as well as the FA substitution of different lengths (C10–C18) (Figure ). A wide range of bioactivities have been reported for PUW/MIN variants, including cardiovascular activity, anti-proliferative activity, and antifungal activity.[11,18−20,25,26] PUW/MIN chemical variants are synthesized by a hybrid non-ribosomal peptide synthetase (NRPS) and polyketide synthase (PKS) pathway that is accompanied by a set of tailoring enzymes.[23] PUW/MIN biosynthesis begins with fatty-acyl-AMP ligases (FAAL) that are responsible for the activation of FAs, which are subsequently elongated by type I PKS modules encoded by PuwB and PuwE proteins.[24] The variability of the FA length is achieved by the broad substrate specificity of the PUW/MIN FAAL enzymes. However, some of the PUW/MIN biosynthetic gene clusters encode two alternative starter modules, each of which activates a different set of FAs.[24] The variability of the FA is further increased through substitution catalyzed by the PuwK halogenase, the PuwJ oxygenase, and the PuwL O-acetyltransferase.[24]

Figure 1

General structure of puwainaphycin and minutissamides. Conserved amino acids are depicted in red, and the variation among amino acid compositions is depicted in gray. FA[1] is the position for FA elongation, whereas R1–R5 are the variably functionalized alkyl substitutions. Dhb = dehydrobutyrine.

We report here the discovery of eight known (3, 5–9, 11, and 12) and five new variants (1, 2, 4, 10, and 13) of PUW/MIN from the Nodularia harveyana strain UHCC-0300 using a previously developed rapid and selective mass spectrometry-based method for β-amino FA lipopeptide analysis[27] (Table ). This method is based on high-performance liquid chromatography connected to tandem mass spectrometry with high-resolution mass spectrometry (HPLC–HRMS/MS) and allows the sensitive and efficient detection of PUW/MIN. The variability observed among the five new variants was mainly due to the difference in their amino acid composition. Finally, we study and report the effects of length and substitution (oxo-, hydroxy-, and chloro-) on the FA moiety as well as the amino acid position toward the change in the cytotoxicity against the human epithelioid cervical cancer cell line (HeLa).

Table 1

High-Performance Liquid Chromatography–High-Resolution Mass Spectrometry (HPLC–HRMS) Data of the N. harveyana Strain UHCC-0300 Crude Extractsa

RT (min)	[M + H]+	formula	error (ppm)	FA-diagnostic ion (m/z)	formula	error (ppm)	FA chain	sequence of amino acids	compounds
11.2	1209.7116	C56H97N12O17+	2.2	224.2361	C15H30N+	5.3	C16 (OH)	FA-Val-Dhb-Thr-Thr-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW H (1)
				242.2473	C15H32NO+	2.2
11.4	1223.7273	C57H99N12O17+	2.2	224.2366	C15H30N+	2.9	C16 (OH)	FA-Val-Dhb-Thr-Thr-Gln-Ala-methoxyThr- N-MeAsn-Pro	PUW I (2)
				242.2496	C15H32NO+	7.2
12.6	1237.7433	C58H101N12O17+	2.5	252.2683	C17H34N+	0.9	C18 (OH)	FA-Val-Dhb-Thr-Thr-Gln-Gly-methoxyThr-N-MeAsn-Pro	MIN K (3)
				270.2798	C17H36NO+	2.4
12.7	1221.7115	C57H97N12O17+	2.1	268.2626	C17H34NO+	3.2	C18 (O)	FA-Val-Dhb-Thr-Ser-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW J (4)
13.0	1235.7260	C58H99N12O17+	1.2	268.2627	C17H34NO+	2.8	C18 (O)	FA-Val-Dhb-Thr-Thr-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW A (5)
13.2	1249.7429	C59H101N12O17+	2.2	268.2621	C17H34NO+	5.2	C18 (O)	FA-Val-Dhb-Thr-Thr-Gln-Ala-methoxyThr-N-MeAsn-Pro	MIN E (6)
13.5	1233.7471	C59H101N12O16+	1.4	268.2645	C17H34NO+	3.7	C18 (O)	FA-Val-Dhb-Thr-Val-Gln-Gly-methoxyThr-N-MeAsn-Pro	MIN L/PUW B (7)
13.7	1247.7621	C60H103N12O16+	0.9	268.2638	C17H34NO+	1.0	C18 (O)	FA-Val-Dhb-Thr-Val-Gln-Ala-methoxyThr-N-MeAsn-Pro	MIN H (8)
14.2	1227.6791	C56H96ClN12O16+	3.3	224.2372	C15H30N+	0.4	C16 (Cl)	FA-Val-Dhb-Thr-Thr-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW C (9)
				260.2149	C15H31ClN+	3.7
14.4	1241.6948	C57H98ClN12O16+	3.3	224.2368	C15H30N+	1.9	C16 (Cl)	FA-Val-Dhb-Thr-Thr-Gln-Ala-methoxyThr-N-MeAsn-Pro	PUW K (10)
				260.2144	C15H31ClN+	1.7
14.8	1225.7011	C57H98ClN12O15+	4.4	224.2376	C15H30N+	1.4	C16 (Cl)	FA-Val-Dhb-Thr-Val-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW D (11)
				260.2121	C15H31ClN+	7.2
15.1	1193.7143	C56H97N12O16+	0.2	226.2529	C15H32N+	0.1	C16	FA-Val-Dhb-Thr-Thr-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW E (12)
15.7	1191.7376	C57H99N12O15+	2.4	226.2533	C15H32N+	1.8	C16	FA-Val-Dhb-Thr-Val-Gln-Gly-methoxyThr-N-MeAsn-Pro	PUW L (13)

The RTs, protonated molecules ([M + H]+) and molecular formula provided for the experimental m/z, FA-diagnostic ion (m/z), sequence of amino acids, and error in ppm for the compounds detected. Dhb = dehydrobutyrine. Variable amino acid positions (5 and 7) are presented in bold.

Results and Discussion

HRMS/MS Analysis of the Crude Extract

Targeted analysis of the crude extract of N. harveyana UHCC-0300 resulted in the discovery of 13 PUW/MINs (1–13), including 8 known and 5 unknown variants (Table ). The fragmentation spectra obtained at a collision energy of 100 eV revealed the presence of diagnostic immonium ions CH(2N+ resulting from the cleavage of the modified FAs. This allowed us to assign FA chain lengths in the case of compounds m/z 1193.7143 (PUW E, 12) and m/z 1191.7376 (PUW L, 13) which possess unsubstituted C16 FA. Other compounds with either hydroxylated or chlorinated substitution on the FA chain generated the diagnostic ion fragment with the general molecular formula CH2N+, which is usually the base peak of the MS/MS spectrum at high energy. This ion was followed by the product ions CH2NO+ and CH2NCl+ for variants bearing hydroxylated and chlorinated FAs, respectively. PUW/MIN chemical variants containing an oxo group bound to a FA chain have their diagnostic immonium ion with a general formula of CH2NO+ (Table ). Further, the amino acid sequences were attributed by the fragmentation spectra generated from a lower collision energy at 35–70 eV (Tables –4). The typical fragmentation pattern of methoxy-Thr (OMe-Thr) containing PUW is the opening of the amino acid cycle between Pro and N-MeAsn, followed by the loss of CH4O from OMe-Thr. Subsequently, the consecutive losses of NMeAsn, Dhb (dehydrobutyrine), Gly/Ala, Gln, Val/Thr/Ser, Thr, Dhb, and Val were observed. Alteration of amino acids in the peptide ring, for example, Gly to Ala or Val to Ser to Thr, has almost no effect on the retention time (RT) of the compound (Table ). However, substitution on the FA side chain led to significant shifts in the RT on a reversed-phase column as observed previously.[27] PUW/MIN variants containing hydroxylated FA chains were eluted first, followed by variants with oxo and chloro substitutions, and then finally variants with no substitution on the FA chain were eluted last, even though the chain lengths were two carbons less than the oxo-substituted compounds (Table ).

Table 2

Precursor Ion (m/z) Data for Compounds 1, 3, 5, 9, and 12

fragment ion assignment	PUW H (1)	MIN K (3)	PUW A (5)	PUW C (9)	PUW E (12)
[M + H]+	1209.7105	1237.7433	1235.7260	1227.6791	1193.7143
[M–CH4O + H]+	1177.68	1205.71	1203.70	1195.65	1161.70
[M–CH4O-N-MeAsn + H]+	1049.62	1077.65	1075.64	1067.59	1033.62
[M–CH4O–N-MeAsn-Dhb + H]+	966.58	994.61	992.60	984.55	951.61
[M–CH4O–N-MeAsn-Dhb-Gly + H]+	909.57	937.57	935.58	927.54	893.57
[M–CH4O–N-MeAsn-Dhb-Gly-Gln + H]+	781.51	809.55	807.52	799.47	765.50
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Thr + H]+	680.46	708.48	706.48	698.43	664.46
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Thr-Thr + H]+	579.42	607.45	605.43	597.38	563.42
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Thr-Thr-Dhb + H]+	496.37	524.41	522.39	514.34	480.38
FA imm ion + CH2O + Pro		379.33	395.33	351.30	353.32
FA imm ion + H2O + CH2O + Pro	369.27	397.34
FA imm ion + CH2O		282.28	298.27	254.25	256.26
FA imm ion + H2O + CH2O	272.22	300.29
FA imm ion + H2O	242.21	270.28
FA imm ion + HCl				260.21
FA imm ion	224.20	252.27	268.26	224.24	226.25

Table 4

Precursor Ion (m/z) Data for Compounds 4, 7, 8, 11, and 13

fragment ion assignment	MinL/PUW B (7)	PUW D (11)	PUW L (13)	MIN H (8)	PUW J (4)
[M + H]+	1233.7471	1225.7011	1191.7376	1247.7621	1221.7115
[M–CH4O + H]+	1201.72	1193.66	1159.71	1215.73	1189.67
[M–CH4O–N-MeAsn + H]+	1073.66	1065.62	1031.67	1087.67	1061.62
[M–CH4O–N-MeAsn-Dhb + H]+	990.62	982.56	948.60	1004.64	978.58
[M–CH4O–N-MeAsn-Dhb-Gly + H]+	933.60	925.54	891.59		921.58
[M–CH4O–N-MeAsn-Dhb-Ala + H]+				933.60
[M–CH4O–N-MeAsn-Dhb-Gly-Gln + H]+	805.54	797.49	763.53		793.50
[M–CH4O–N-MeAsn-Dhb-Ala-Gln + H]+				805.54
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Val + H]+	706.47	698.43	664.47
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Ser + H]+					706.47
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Val + H]+				706.48
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Val-Thr + H]+	605.43	597.38	563.42
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Ser-Thr + H]+					605.43
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Val-Thr + H]+				605.43
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Val-Thr-Dhb + H]+	522.39	515.35	480.38
[M–CH4O–N-MeAsn-Dhb-Gly-Gln-Ser-Thr-Dhb + H]+					522.38
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Val-Thr-Dhb + H]+				522.39
FA imm ion + CH2O + Pro	395.33	351.30	353.32	395.33	395.33
FA imm ion + CH2O	298.27	254.25	256.26	298.27	298.27
FA imm ion + H2O
FA imm ion + HCl		260.21
FA imm ion	268.26	224.24	226.25	268.26	268.26

Isolation and Structural Elucidation of 1–13

The benthic N. harveyana strain UHCC-0300 was isolated from the coastal area of the Baltic Sea in 2001[28] and grown in 100 L of Z8xS growth medium under the continuous illumination of 3.2–3.7  μmol photons m–2  s–1.[29] The freeze-dried biomass (20 g) was extracted with 70% MeOH in water and dried under vacuum. The crude extract was subsequently fractionated using reversed-phase flash column chromatography, eluting with a mixture of H2O/CH3CN (from 0 to 100%) and then with 100% of MeOH, to afford 12 fractions (F1–F12). HRMS/MS analysis of all fractions indicated the presence of lipopeptides in the fractions F7–F9. Subsequent chromatography of these fractions using a Sephadex LH-20 followed by a semipreparative HPLC yielded compounds 3 (0.60 mg), 5 (9.70 mg), 6 (3.70 mg), 7 (1.20 mg), 8 (1.20 mg), 9 (0.57 mg), 10 (0.51 mg), and 11 (0.52 mg). The structures of compounds 5 and 6 were elucidated by HRMS/MS combined with nuclear magnetic resonance (NMR) data [one-dimensional (1D) and two-dimensional (2D) NMR]. The comparison of the NMR data with the data reported in the literature (Table S1) allowed the assignment of compound 5 as PUW A and compound 6 as MIN E.[18−20] Compounds 1–4 and 7–11 were identified by the HRMS fragmentation data (the amino acid sequence as well as the FA length). Data allowing fragment identification are reported in Tables –4. Due to the low amount of the isolated compounds, the position of substituents on the FA chain remains unresolved. Compound 12 was assigned as PUW E, with the C16-FA chain. A new variant, PUW L (13) (FA-Val-Dhb-Thr-Val-Gln-Gly-methoxyThr-N-MeAsn-Pro), differs from PUW E (12) by a Val unit instead of the Thr unit. The other new variants detected were designated as PUW H (1), PUW I (2), PUW J (4), and PUW K (10). PUW H (1) and PUW I (2) bearing hydroxylated C16-FA only differs from each other by a Gly/Ala unit, whereas PUW K (10) containing the same amino acid sequence as PUW I (2) differs by bearing a chlorinated C16-FA. PUW J (4) and PUW A (5) bearing oxo-C18-FA differs from each other by a Ser/Thr unit.

Elucidation of the Puw/Min Biosynthetic Gene Cluster in N. harveyana UHCC-0300

PUW/MINs were previously reported to be synthesized via a hybrid NRPS/PKS machinery encoded in 60–64 kb biosynthetic gene clusters.[24,30] In this study, we identified the 63 kb puw biosynthetic gene cluster using tBLASTn searches against the standalone database based on a draft genome sequence for N. harveyana UHCC-0300. These searches were conducted using the biosynthetic enzymes reported previously (puw biosynthetic gene clusters).[18,23] The puw biosynthetic gene cluster from the N. harveyana strain UHCC-0300 encoded 13 biosynthetic proteins organized in a bidirectional operon (Figure A). This puw biosynthetic gene cluster encoded two alternative loading modules, PuwI and PuwC, for the activation of different FA starter units as reported previously in the case of Anabaena sp. strains.[23] The FA moiety is then elongated by the PuwB and PuwE PKSs identically as in other puw biosynthetic gene clusters.[18,23] Biosynthetic genes encoding the PuwK accessory halogenase and the PuwJ monooxygenase were also present (Figure A). The presence of two alternatives PuwI and PuwC as FAAL starters was reported to result in the synthesis of lipopeptides with FA moieties of differing lengths and functionalization in Cylindrospermum and Anabaena strains.[8,23] However, in N. harveyana strain UHCC-0300, the FA moieties activated by the individual starter modules appear differently (Figure A). PuwC was previously corroborated to activate FA chains resulting in the production of variants with a shorter FA moiety (C10–C14) in Cylindrospermum and Anabaena strains,[24,30] while PuwI tends to activate longer FA residues resulting in the production of variants with C16–C18 FA in Symplocastrum and Anabaena strains.[24] It is notable that in Anabaena minutissima UTEX B1613, which encodes both PuwC and PuwI together with the accessory halogenase PuwK and oxygenase PuwJ, the substitution pattern differs for FA with different lengths. Puw/Min variants with short FA chain (activated by PuwC) are halogenated, while the long-tail FA PUW/MIN variants (activated by PuwI) contain only hydroxyl- or oxo-substitution. Among PUW/MINs produced by N. harveyana UHCC-0300 (containing PuwC, PuwI, PuwK, and PuwJ), variants with only C16 and C18 FA chains were detected. While both the variants (C16 and C18) contain hydroxyl- or oxo-substitution (compounds 1–8), only the C16 variants were found with chlorine substitution (compounds 9–11). Based on our data, we assume that the PuwC enzyme cooperates with PuwK to generate halogenated variants and PuwI/PuwJ activity results in hydroxyl-/oxo-substituted variants, which was in accordance with the previously published report. Moreover, in N. harveyana UHCC-0300, PuwC has apparently altered substrate specificity (activation of C12 FA resulting in PUW/MIN with C16 FA chain after elongation by PuwB/PuwE) compared to PuwC in Cylindrospermum/Anabaena reported previously (activation of C6–C10 resulting in PUW/MIN with C10–C14 FA chain after PuwB/PuwE elongation).

Figure 2

Structure of the puwainaphycin (puw) biosynthetic gene cluster organization and functional annotation of puwA-J genes. (a) Gene arrangement of the puw biosynthetic gene cluster in the N. harveyana UHCC-0300 strain and the proposed biosynthetic scheme and (b) comparison of A-domains of 10 puw biosynthetic gene clusters identified from public databases.

The PuwA, PuwE, PuwF, PuwG, and PuwH enzymes catalyze the incorporation of nine amino acids into the growing peptide chain (Figure ). Minor variants were observed involving substitution of amino acids similar in structure and hydrophobicity, including Thr to Val (compounds 7, 8, 11, and 13) or Thr to Ser (compound 4) at position 5 and Gly to Ala at position 7 (compounds 2, 6, 8, and 10) (Table ). This suggests probable substrate promiscuity of the PuwG A4 and PuwG A6 adenylation domains. The stereochemistry of amino acid residues present in PUW/MIN has been widely studied. All the residues forming the peptide cycle were proved to be present in the L-form. The only exception is the presence of D-Ala at position 7, which was due to the presence of an epimerase domain in all the known Puw/Min biosynthetic gene clusters published previously. Similarly, in the proposed biosynthetic scheme, an epimerase domain was found with PuwG, suggesting the presence of D-Ala. Based on this, we concluded that the stereochemistry of the obtained PUW/MIN is identical as proved previously.[20] We conducted BLASTp searches using conserved enzymes from the N. harveyana UHCC-0300 puw biosynthetic gene cluster against the nonredundant database at NCBI and identified 10 complete puw biosynthetic gene clusters (Figure B). The 10 puw biosynthetic gene clusters shared a conserved gene order and encoded 12 puw biosynthetic enzymes (Figure B). This finding suggests that while rare, the puw biosynthetic gene clusters encoding alternative lipoinitiation mechanisms are widespread in the Nostocales.

Effect of FA Substitution on Bioactivity

The anti-proliferative activity of PUW F, MIN A-D, and MIN E-L has been reported in the literature, with different experimental settings and on different human cell lines (HeLa, colon carcinoma HT-29, and melanoma cell MDA-MB 435, respectively).[19,20] Recently, the cytotoxic activity of PUW/MIN variants with respect to the length of the FA moiety was studied by our team, and it was concluded that in a certain FA length span (C12–C14), the cytotoxicity increases with the FA length but reduces further with FA extension.[26] We used identical setting of the experiment and performed the cytotoxicity assays on human HeLa cells. To obtain a broader picture, we included four PUW/MIN variants (14–17) previously isolated, differing in their amino acid positions 4–8 from the PUW/MIN variants isolated from N. harveyana strain UHCC-0300 (Figure ). The four variants are as follows: MIN C (14) bearing an oxo group, MIN D with the hydroxylated FA moiety (15) isolated from Anabaena sp. UHCC-0399, and the chlorinated variants PUW F possessing C14–Cl FA (16) and MIN A possessing C12–Cl FA (17) isolated from Cylindrospermum alatosporum CCALA 988[24,31] (Figure ).

Figure 3

Structures of 5, 6, and 14–17.

Compounds 5–8, bearing the oxo substitution on the C18 FA chain, manifested a moderate cytotoxic effect with IC50 values of 3.8 ± 0.8, 2.4 ± 0.5, 2.7 ± 0.5, and 4.5 ± 0.4 μM, respectively, on HeLa cell lines. A similar IC50 value (3.8 ± 0.5 μM) was observed for compound 14 bearing the oxo group on the C16 FA chain, differing from compounds 5–8 in the peptide core sequence (Table ). A lower potency (IC50 of 11.8 μM) was recorded for compound 14 in human colorectal HT-29 cells.[19] Concerning the hydroxylated variant, compound 3 showed a weaker potency with an IC50 value of 9.3 ± 3.2 μM. Interestingly, compound 15 (C16, OH) was not cytotoxic up to the highest tested concentration of 40 μM, whereas previously, it was observed to possess a weak cytotoxicity on HT-29 cells (IC50 = 22.7 μM).[14,19] Finally, variants with chlorine substitution on C16 FA (compounds 9–11) displayed IC50 values of 2.3 ± 0.9, 5.2 ± 1.4, and 3.4 ± 1.3 μM, respectively. Chlorinated compounds 16 and 17 with C14 FA manifested very variable IC50 values of 3.09 ± 0.3 and 33.89 ± 6.3 μM, respectively, which is likely the result from different positions of the Cl substituent of the FA (Figure ). The chlorinated variants with C14 FA were previously shown to possess IC50 values of 1.2 and 2.6 μM against MDA-MB-435 cells.[20] Although non-substituted PUW/MIN variants were not subjected to the cytotoxicity test in this study, it is worthy to mention that PUW F bearing unsubstituted C14 FA showed moderate IC50 values against HeLa with an IC50 value of 2.2 μM,[25] and MIN A with C12 FA was reported to possess a similar potency against HT-29 and HeLa cells with IC50 values of 2.0 μM[19] and 2.8 μM, respectively.[26]

Figure 4

Cell viability was assessed by the MTT assay at a 48 h exposure time. The cell viability was calculated as the percentage of viable cells in compound-treated cells relative to control. All the experiments were performed with at least three independent biological replicates. Data from repeated measurements were shown as the mean ± SE. IC50 values were calculated using GraphPad Prism 5.

Conclusions

The current study presents the identification of 13 PUW/MIN chemical variants produced by the N. harveyana strain UHCC-0300 and proposes a biosynthetic route for their production. Searches against public sequence repositories suggested the widespread occurrence of PUW/MIN biosynthetic gene clusters encoding strains in the environment. Moreover, this is the first comprehensive study reporting the anti-proliferative activity of PUW/MIN variants on human cells in vitro. Based on our findings, the following conclusions can be reported: PUW/MIN bearing unsubstituted and an oxo-substituted FA chain (C14–C18 length) possesses moderate cytotoxicity with a comparable IC50 value, demonstrating that the oxo-substitution does not affect the bioactivity. In addition, our findings suggest that the differences in the PUW/MIN peptide ring at positions 4–8 have only a minimal effect on cytotoxicity. The differences between the IC50 values of oxo- and hydroxyl-substituted variants clearly show that the lipophilicity of the FA residue is essential for the compound’s interaction with the plasma membrane.[21,25] Finally, the chlorine substitution and the position of the substitution affected largely the compound’s cytotoxicity potential.

Experimental Section

Cultivation of Cyanobacterial Biomass and Crude Extract Preparation

Crude extracts were prepared following the pre-established protocol.[33,34] Briefly, freeze-dried biomass was ground with the sea sand and extracted three times with 75% MeOH in water, followed by bath sonication. The extracts were evaporated under vacuum using a rotary vacuum evaporator (Heidolph, Germany) and dissolved with 75% MeOH in water to get a final concentration of 4 mg/mL for LC–MS/MS analysis.

HPLC–MS/MS Analysis

A Thermo Scientific DionexUltiMate 3000 UHPLC (Thermo Scientific) equipped with a diode array detector was used for the analysis of the crude extract. HPLC separation was performed on a reversed-phase Kinetex Phenomenex C18 column (150 × 4.6 mm, 2.6 μm) with H2O/CH3CN containing 0.1% HCOOH as a mobile phase. The flow rate during analysis was 0.6 mL/min. The gradient is as follows: H2O/CH3CN 85/15 (0 min), 85/15 (in 1 min), 0/100 (in 20 min), 0/100 (in 25 min), and 85/15 (in 30 min). HPLC was connected to a high-resolution mass spectrometer with an electrospray ionization source (impact HD mass spectrometer, Bruker). The mass spectrometer settings are as follows: dry temperature 200 °C; drying gas flow 12 L/min; nebulizer 3 bar; capillary voltage 4500 V; and endplate offset 500 V. The spectra were collected in the range 20–2000 m/z with the spectra rate 4 Hz. The collision energy alternated from 35 eV or 50 to 100 eV. Calibration was determined using a LockMass 622 as an internal calibration solution and CH3COONa clusters at the beginning of each analysis. The extract was analyzed for the characteristic β-amino FA immonium ion following the previously described protocol.[27]

Genome Sequencing, Assembling, Annotation, and Mining for the Identification of the Puw/Min Gene Cluster

N. harveyana UHCC-0300 was first isolated on 28/10/2001 from a plant surface at the littoral zone in Vartiokylänlahti, Helsinki, Finland. The N. harveyana UHCC-0300 strain was grown in a photon irradiance of 5 μmol m–2 s–1 in Z8xS medium that lacked a source of combined nitrogen. The strain was grown for 21 days at 18 °C and was harvested by centrifugation at 10,000g for 7 min (Sorvall LYNX 6000 Superspeed Centrifuge, Thermo Fisher Scientific). The genomic DNA of the strains was extracted from 89 mg of wet cells that were lysed with a heat-shock treatment consisting of repeated (×15) liquid nitrogen immersion and thawing at a 55 °C water bath. DNA extraction was then carried out using a commercial DNA extraction kit (E.Z.N.A. SP Plant DNA Mini Kit Protocol—Fresh/Frozen Samples, Omega Bio-Tek). The DNA yield and quality were verified by NanoDrop (NanoDrop 1000 Spectrophotometer, Thermo Fisher Scientific). The DNA size and quality were further assessed by gel electrophoresis at room temperature (100 V, 400 mA, 30 min, 0.9% agarose) in 0.5 × TAE-buffer (20 mM Tris, 10 mM acetic acid, 0.5 mM ethylenediaminetetraacetic acid, pH 8.3). Libraries were prepared with a Nextera DNA flex library prep kit (recently renamed to Illumina DNA Prep), and Illumina MiSeq sequencing was carried out using the MiSeq Reagent Kit v3 (600 cycle). The obtained sequences were trimmed to remove adapters using Cutadapt-1.9.1 with options -q 25 -m 50,[35] and the assemblies were prepared from the trimmed fastaq files using SPAdes v3.12.0 with the careful option.[36] The resulting assembly of 5.3 Mb and 113 scaffolds was then further processed for taxonomic classification using Kraken v2,[37] and contaminating scaffolds were removed with ZEUSS v1.0.2.[38]

Identification and Annotation of the puw Biosynthetic Gene Cluster

A 63 kb puw biosynthetic gene cluster was identified through tBLASTn searches using PuwA-K as query sequences against a standalone BLAST database of the N. harveyana UHCC-0300 genome. There was a single 127-bp gap in the puw biosynthetic gene cluster of N. harveyana UHCC-0300. This gap was closed by PCR and Sanger sequencing. The fragment containing the gap was amplified with the oligonucleotide pair puwIF (5′ TTATTCATGACTTTGGGATGATCC-3′) and gap1R (5′-TACTGGAAAATGCCCTCACCAGTTGG-3′) and then purified using the PCR clean-up kit (Macherey-Nagel NucleoSpin Gel and PCR Clean-up Kit, Fisher Scientific). Sanger sequencing of the gap was done using the primer pair gap1F (5′-GCTTTCGAGAGCGTGATTTAGGCAAAG-3′) and gap1R (5′-TACTGGAAAATGCCCTCACCAGTTGG-3′) at the Eurofins Genomics facilities. The genes encoded in the puw biosynthetic gene cluster were predicted using GLIMMER. Start sites were predicted, and proteins were annotated manually using a combination of searches against the Conserved Domain Database and protein classification resources at NCBI and InterProScan searches and BLASTp searches against the non-redundant database at NCBI. The annotated sequence of the puw gene cluster from N. harveyana UHCC-0300 was deposited in GenBank under the accession number OK416066. We conducted BLASTp searches using conserved enzymes from the N. harveyana UHCC-0300 puw biosynthetic gene cluster against the non-redundant database at NCBI in order to identify complete puw biosynthetic gene clusters in public databases.

Isolation and Structural Elucidation of Compounds 1–13

Large-scale cultivation (100 L) of the strain yielded 20 g of the dried biomass. The crude extract was prepared as described above. The crude extract (5 g) obtained was fractionated using reversed-phase flash column chromatography, eluting with a mixture of H2O/CH3CN (from 0 to 100%) and then with 100% of MeOH, to afford 12 fractions. Three fractions (F7, F8, and F9) were subjected to a Sephadex LH-20 gel chromatography column eluting with CHCl3/MeOH (1:1) and subsequently purified using semipreparative reversed-phase column chromatography (Phenomenex Kinetex 5 μm EVO C18 100 Å, 200 × 10.0 mm) eluted with H2O (A)/CH3CN (B) both containing 0.1% HCOOH at a flow rate of 3 mL/min using the following gradient: A/B 68/32 (0 min) and 60/40 (in 50 min). The compound elution was monitored on the MWD detector set to 220 nm to obtain compounds 3 (0.6 mg), 5 (9.7 mg), 6 (3.70 mg), 7 (1.20 mg), 8 (1.02 mg), 9 (0.20 mg), 10 (0.15 mg), and 11 (0.12 mg). The NMR spectra were acquired at 25 °C on a Bruker AVANCE Neo 700 MHz spectrometer (Billerica, MA, US) equipped with a triple resonance CHN cryoprobe using DMSO-d6 (Sigma-Aldrich, Milan, Italy) as solvents and the 1D and 2D standard pulse sequences provided by the manufacturer. The 1H chemical shifts were referenced to the residual solvent’s protons resonating at 2.50 (CHD2SOCD3) ppm. All 13C-NMR chemical shifts were assigned using the 2D spectra; therefore, monodimensional 13C-NMR spectra were not recorded and were referenced to the solvents’ methyl carbons resonating at 39.51 ppm (DMSO-d6). Abbreviations for signal couplings are as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, and b = broad. The HR-MS/MS data (Figure S1) and the comparison of the NMR data with the data reported in the literature (Table S1) allowed the assignment of compound 5 as PUW A (Figures S2–S8) and compound 6 as MIN E[18−20] (Figures S9–S15). The HR-MS/MS fragmentation data for all the new compounds, PUW H–PUW K (1, 2, 4, 10, and 13), isolated are provided in Figure S16. The purity of all the other compounds proceeded further for cytotoxicity was evaluated on the basis of 1H-NMR spectra (for compounds 3, 7, and 8, Figures S17–S19) and UV–vis chromatogram (for compounds 8–11, Figure S20).

Compound (5): δ 9.10 (1H, s), δ 8.75 (1H,br s), δ 8.35 (1H, d, J = 9.2), δ 7.96 (1H, t, J = 6.2), δ 7.52 (1H, s), δ 7.38 (1H, s), δ 7.30 (1H, s) δ 6.87 (1H, d, J = 8.7) δ 6.84 (1H, d, J = 10.4), δ 6.81 (2H, ovl), δ 5.97 (1H, s), δ 5.56 (1H, dd, J = 11.6,3.0), δ 5.53 (1H, d, J = 4.5), δ 5.36 (1H, q, J = 7.5), δ 5.20 (1H, d, J = 5.2), δ 4.90 (1H, dd, J = 9.3, 2.4), δ 4.72 (1H, dd, J = 9.1, 2.6), δ 4.58 (1H, m), δ 4.30 (1H, dd, J = 9.1, 6.5), δ 4.27(1H, dd, J = 8.3, 1.8), δ 4.17 (3H, ovl), δ 4.10 (2H, ovl), δ 3.99 (1H, dd, J = 16.9, 7.3), δ 3.94 (1H, ddd, J = 11.0, 10.5, 5.1), δ 3.83 (1H, t, J = 3.8) δ 3.73 (1H, m) δ 3.22 (1H, dd, J = 17.1, 6.5), δ 3.18 (3H, s), δ 3.17 (1H, m), δ 3.03 (1H, dd, J = 16.7, 11.9), δ 2.96 (3H, s), δ 2.40 (4H, ddd, J = 7.7,7.0,6.0), δ 2.17 (2H, m), δ 1.98 (4H, ovl), δ 1.82 (3H, ovl), δ 1.74 (3H, d, J = 7.4), δ 1.70 (1H, m), δ 1.61 (1H, m), δ 1.45 (4H, ovl), δ 1.25 (21H, ovl), δ 1.00 (3H, d, J = 6.1), δ 0.89 (3H, d, J = 6.9), δ 0.84 (6H, ovl), δ 0.57 (3H, d, J = 6.7)

Compound (6): δ 9.10 (1H, s), δ 8.84 (1H, d, J = 3.5), δ 8.39 (1H, d, J = 9.7), δ 7.59 (1H, d, J = 7.5), δ 7.51 (1H, s), δ 7.28 (1H, s), δ 7.26 (1H, d, J = 7.9) δ 6.85 (1H, d, J = 8.8) δ 6.82 (1H, s), δ 6.78 (1H, d, J = 10.1), δ 6.75 (1H, d, J = 9.3), δ 6.02 (1H, s), δ 5.57 (1H, d, J = 4.4), δ 5.53 (1H, dd, J = 11.6,3.0), δ 5.39 (1H, q, J = 7.4), δ 5.29 (1H, d, J = 4.7), δ 5.02 (1H, dd, J = 9.8, 2.2), δ 4.79 (1H, dd, J = 9.5, 1.7), δ 4.59 (1H, m), δ 4.32 (1H, dd, J = 9.2, 6.5), δ 4.26(1H, dd, J = 8.2, 1.9), δ 4.19 (4H, ovl), δ 4.09 (1H, ddd, J = 10.3, 8.9, 4.6), δ 3.94 (1H, ddd, J = 11.6, 10.8, 5.1), δ 3.91(2H, ovl), δ 3.72 (1H, m), δ 3.14 (3H, s), δ 3.12 (1H, m), δ 3.00 (1H, dd, J = 15.3, 12.0), δ 2.93 (3H, s), δ 2.38 (4H, q, J = 6.8), δ 2.15(2H, m), δ 1.98 (4H, ovl), δ 1.81 (3H, ovl), δ 1.75 (3H, d, J = 7.3), δ 1.70 (1H, m), δ 1.61 (1H, m), δ 1.44 (4H, ovl), δ 1.29 (3H, d, J = 7.4), δ 1.25 (21H, ovl), δ 0.96 (3H, d, J = 6.1), δ 0.89 (3H, d, J = 6.6), δ 0.84 (6H, ovl), δ 0.58 (3H, d, J = 6.6)

Cytotoxic Activity

Besides compounds 3 and 5–11 isolated from N. harveyana UHCC-0300, the present study has included four additional variants, MIN C (14) and MIN D (15), previously isolated from the Anabaena sp. strain UHCC-0399, and MIN A (17) and 11-chloro-4-methyl-Ahdoa-PUW F (16) from the strain Cylindrospermum alatosporum CCALA 988.[24,39] A total of 12 compounds (3, 5–11, and 14–17) were tested for cytotoxicity against the human epithelioid cervical cancer cell line (HeLa). The cells were cultivated in RPMI cultivation media supplemented with 1% antibiotic–antimycotic solution, 1% l-glutamine, and 5% fetal bovine serum. The cells were seeded in a density of 10,000 cells per well 1 day prior to the experiment. On the next day, the cultivation medium was replaced with the cultivation medium containing desired concentrations (20, 10, 5, 2.5, 1.25, 0.63, and 0.31 μM) of the tested compounds. The vehicle dimethyl sulfoxide (DMSO) concentration did not exceed 0.5%. The cell viability after the 48 h exposure to the compounds was assessed using the MTT assay as reported previously.[40] The absorbance of the compound-treated cells was measured at 590 nm (reference wavelength at 640 nm) was divided by the values obtained for the control cells and expressed in percent. All experiments were performed in biological triplicates, each including a technical triplicate of each condition. The IC5O values were calculated using a variable slope (four-parameter) function with Hill’s slope in GraphPad Prism software.

Data Deposition

The strain N. harveyana has been deposited to culture collection of UHCC under the strain number UHCC-0300. The Puw/Min biosynthetic gene cluster from N. harveyana UHCC-0300 is available under the accession number OK416066.

Table 3

Precursor Ion (m/z) Data for Compounds 2, 6, and 10

fragment ion assignment	PUW I (2)	MIN E (6)	PUW K (10)
[M + H]+	1223.7273	1249.7429	1241.6948
[M–CH4O + H]+	1191.69	1217.71	1209.66
[M–CH4O–N-MeAsn + H]+	1063.46	1089.65	1081.60
[M–CH4O–N-MeAsn-Dhb + H]+	980.59	1006.62	998.57
[M–CH4O–N-MeAsn-Dhb-Ala + H]+	909.56	935.58	927.53
[M–CH4O–N-MeAsn-Dhb-Ala-Gln + H]+	781.50	807.53	799.48
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Thr + H]+	680.47	706.48	698.43
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Thr-Thr + H]+	579.41	605.43	597.38
[M–CH4O–N-MeAsn-Dhb-Ala-Gln-Thr-Thr-Dhb + H]+	496.38	522.39	514.34
FA imm ion + CH2O + Pro	351.30	395.33	351.30
FA imm ion + CH2O	254.25	298.27	254.25
FA imm ion + H2O	242.25
FA imm ion + HCl			260.21
FA imm ion	224.24	268.26	224.24