Phormidepistatin from the Cyanobacterium UIC 10484 – Assessing the Phylogenetic Distribution of the Statine Pharmacophore

Abstract

A new linear lipopeptide, phormidepistatin (1), containing an epi-statine amino acid was isolated from cf. Phormidium sp. strain UIC 10484. The planar structure was elucidated by 1D and 2D NMR experimentation. The relative configuration was determined by J-based configurational analysis and the absolute configuration by advanced Marfey’s analysis. Given that the statine moiety is an established pharmacophore known to inhibit aspartic proteases, phormidepistatin was evaluated against cathepsin D and displayed limited activity. With 1 containing a statine-like moiety, we sought to assess the distribution of this γ-amino acid within the phylum Cyanobacteria. In-depth MS/MS analysis identified the presence of phormidepistatin in cf. Phormidium sp. UIC 10045 and cf. Trichormus sp. UIC 10039. A structure database search identified 33 known cyanobacterial metabolites containing a statine or statine-like amino acid, and along with phormidepistatin, were grouped into 10 distinct compound classes. A phylogenetic tree was built comprising all cyanobacteria with established 16S rRNA sequences known to produce statine or statine-like-containing compound classes. This analysis suggests the incorporation of the γ-amino acid into secondary metabolites is taxonomically widespread within the phylum. Overall, it is our assessment that cyanobacteria are a potential source for statine or statine-like-containing compounds.

Graphical Abstract

The phylum Cyanobacteria is a morphologically diverse taxon of bacteria known to produce secondary metabolites with bioactivity relevant, but not limited, to cancer, microbial, and neurodegenerative disease states.^1,2 In particular, cyanobacteria from multiple taxonomic orders have been found to produce metabolites with antiproliferative activity.^3–6 The phylum has the capacity to produce nonribosomal peptides, polyketides, a hybrid of the two biosynthetic genres, ribosomally synthesized and post-translationally modified peptides, as well as other secondary metabolite types.⁷ Thus, cyanobacteria are a good source for drug lead discovery.

In our ongoing investigation to identify bioactive secondary metabolites from cyanobacteria, we obtained a new epi-statine-containing linear lipopeptide, phormidepistatin (1) from cf. Phormidium sp. UIC 10484. The original statine, (3S,4S)-4-amino-3-hydroxy-6-methylheptanoic acid (Sta), is a γ-amino acid established as the pharmacophore responsible for inhibition of aspartic proteases.^8–10 The moiety is effectively a transition state analogue of a peptide undergoing hydrolysis. In particular, it is proposed that the Sta ⍺-carbon and ß-carbon constituents mimic the scissile bond intermediate of a normal substrate.^10,11

The presence of a statine-like (Sta-like) moiety led us to evaluate phormidepistatin for inhibition against the aspartic protease cathepsin D. The proenzyme form is known to be secreted from tumorigenic breast cancer cells into the more acidic surrounding microenvironment where it is autoactivated and in turn becomes proteolytic.¹² Multiple studies have demonstrated that extracellular cathepsin D both activates pathways and inactivates inhibitory mechanisms involved in tumor cell division, angiogenesis and migration.^12–16 Overall, this aspartic protease has been associated with several cancer pathologies in addition to other disease states.^12,17–19

In addition to UIC 10484, phormidepistatin was also found in cyanobacterial strains UIC 10045 and 10339. These three strains are from two different taxonomic orders – Oscillatoriales and Nostocales.^20,21 This led us to perform a chemistry search to identify the prevalence of the statine or analogues from the pool of known cyanobacterial secondary metabolites and assess the phylogenetic distribution of the Sta/Sta-like amino acid. Our analysis suggests this pharmacophore is widely incorporated into diverse chemistry by strains throughout the phylum Cyanobacteria.

RESULTS AND DISCUSSION

The Isolation of Phormidepistatin.

Phormidepistatin (1), a white, amorphous powder, was obtained, following a multistep purification protocol as outlined in the Experimental Section. The molecular formula was calculated to be C₅₅H₈₀N₆O₁₄ based on HRESIMS (m/z 1049.5839 [M+H]⁺, 3.2 ppm error). A series of 2D NMR experiments was used to elucidate the planar structure. Phormidepistatin was dissolved in DMSO-d₆ and TFA vapor. TFA shifted the water peak to a higher frequency (from ~3.35 ppm to 4.00–5.00 ppm), which limited water’s overlap with proton signals in both 1D and 2D spectra. Additionally, it was used to improve NH resonances as it reduced proton exchange between the amide backbone and water.

Two sets of NMR signals were detected in the ¹H spectrum with a 7:3 ratio. Analyzing the NMR data (Table 1) of the major conformer, the planar structure was found to be a linear heptapeptide with a hydroxylated phenyldecanoic acid attached to the N-terminus. Of the six residues, four were standard amino acids with a γ-amino acid adjacent to an amino alcohol making up the C-terminus. The standard amino acids were identified as serine (Ser), tyrosine (Tyr), proline (Pro), and threonine (Thr) while the nonstandard residues were epi-statine (epi-Sta), a ß-hydroxylated γ-amino acid, and tyrosinol (Tol), the reduced form of tyrosine.

Table 1.

Phormidepistatin (1) NMR spectroscopic data (900 MHz, DMSO-d₆+TFA)

	position	δC, type	δH, mult (J in Hz)	COSY	HMBC
Tol	1	62.3, CH2	3.31, m	2
			3.24, m
	1-OH		nd
	2	52.7, CH	3.79, m	1, 3, NH	epi-Sta-1, 4
	3	35.7, CH2	2.68, m	2	5/6
			2.48, m
	4	129.2, C
	5/6	130.0, CH	6.97a	7/8	9
	7/8	115.0, CH	6.64a	5/6	9
	9	155.5, C
	9-OH		nd
	NH		7.55, d (8.2)	2	epi-Sta-1
epi-Sta	1	170.8, C
	2	40.2, CH2	2.16, m	3	1, 4
			2.07, dd (9.3, 14.2)
	3	70.7, CH	3.67, m	2
	3-OH		nd
	4	51.2, CH	3.68, m	5, NH
	5	38.8, CH2	1.32a	4, 6
			1.25a
	6	24.0, CH	1.56a	5, 6-Me, 7
	6-Me	21.5, CH3	0.77, d (6.5)	6
	7	24.0, CH3	0.83, d (6.7)	6
	NH		7.29, d (8.8)	4	Thr-1, 4
Thr	1	169.8, C
	2	59.0, CH	4.04, m	NH	1, 3
	3	66.5, CH	3.97, m	4	1
	3-OH		nd
	4	20.2, CH3	1.04, d (6.4)	3	2
	NH		7.69, d (8.5)	2	Pro-1
Pro	1	171.6, C
	2	59.9, CH	4.28, dd (3.4, 7.2)	3
	3	29.0, CH2	1.83, m	2, 4	1
			1.76, m
	4	24.2, CH2	1.75, m	3, 5	2
			1.59, m
	5	46.9, CH2	3.50, m	4
			3.02, m
Tyr	1	169.7, C
	2	53.0, CH	4.51, q (7.3)	NH, 3	1, Ser-1, 4
	3	36.5, CH2	2.85, dd (7.3, 13.6)	2	1, 5/6
			2.75, dd (7.3, 13.6)
	4	127.2, C
	5/6	130.4, CH	6.97a	7/8	9
	7/8	115.0, CH	6.64a	5/6	9
	9	156.1, C
	9-OH		nd
	NH		8.18, d (7.3)	2	Ser-1
Ser	1	170.2, C
	2	54.8, CH	4.31, dt (8.0, 5.8)	3, NH
	3	61.7, CH2	3.56, m	2	1
			3.52, m
	3-OH		nd
	NH		7.73, d (8.0)	2	DMPD-1
DMPD	1	175.1, C
	2	46.0, CH	2.31, quint (6.7)	17, 3	1
	3	72.0, CH	3.45, m	4
	3-OH		nd
	4	33.9, CH2	1.38a	3
			1.23a
	5	25.2, CH2	1.37a
			1.23a
	6	29.4, CH2	1.23a
	7	25.4, CH2	1.4a
			1.23a
	8	36.7, CH2	1.32a	9
			1.26a
	9	71.1, CH	3.60, m	8, 10	11
	9-OH		nd
	10	43.8, CH2	2.61, d (6.7)	9	12/13
	11	139.8, C
	12/13	129.3, CH	7.18, d (7.4)	14/15
	14/15	128.1, CH	7.25, t (7.4)	12/13, 16
	16	125.7, CH	7.15, t (7.4)	14/15
	17	14.1, CH3	0.94, d (6.7)	2	1

nd – no detection

^aoverlapping signals

quint – quintet

Analysis of data from DEPTQ, HSQC, HMBC, and band-selective HMBC NMR experiments identified all carbons and confirmed the molecular formula. A combination of COSY, TOCSY, and HMBC data established the side chains of the six amino acids in addition to the extended backbone of epi-Sta (Figure 1). COSY correlations between the α-proton (or the epi-Sta γ-proton) and the adjacent NH linked each backbone NH to its corresponding side chain. A band-selective HMBC experiment was utilized to deconvolute the overlapping carbonyl signals. This experiment, combined with HMBC, established the carbonyl groups for each amino acid unit based on their correlations with the side chains’ α- and ß-protons. Both HMBC and band-selective HMBC were used to determine the phormidepistatin amino acid sequence, although no correlation was detected connecting Pro and Tyr. However, the sequence was completed by MS/MS evaluation of 1 (Figure 2). EICs from LC-MS data acquired on the hydrolysate of 1 after exposure to 6N HCl supported the order of amino acids (Figure S9, Supporting Information).

Figure 1.

Phormidepistatin (1) 2D NMR correlations.

Figure 2.

Confirmation of the phormidepistatin amino acid sequence by mass spectrometry.

The 3,9-dihydroxy-2-methyl-10-phenyldecanoic acid (DMPD) partial structure was elucidated using COSY, HMBC, and band-selective HMBC experiments. Correlations between the DMPD H-2, the methyl H₃-17, and the Ser NH protons with the carbonyl adjacent to Ser attached the moiety to the N-terminus of the heptapeptide. COSY and HMBC correlations established the connectivity between C-2, the C-2 methyl group (C-17), the secondary alcohol C-3, and the C-4 methylene. Due to overlapping resonances in the aliphatic region, a ¹H-¹³C HSQC-TOCSY experiment was required to connect C-4 through C-8. H-3 and H-9 were both found to correlate with C5–7, thus bridging the DMPD methylene chain. C-8, the hydroxylated C-9, and C-10 were assigned by COSY. HMBC correlations between H-10 and the phenyl C-12/13 connected the aromatic ring to the DMPD carbon chain.

Advanced Marfey’s method was used to determine absolute configurations of the six amino acids (Table S1). After the hydrolysis of 1 and the derivatization of the amino acids with L- or D/L-FDLA, retention times of the amino acids were compared to DLA-derivatized commercial standards. Tol, Thr, and Pro were confirmed to have the L configuration while Tyr and Ser were established as D. Epi-Sta was determined to be 3R,4S, a diastereomer of the original pepstatin A statine.²²

We attempted to determine the three stereogenic centers of the DMPD subunit. The methylated C-2 was established as erythro relative to the hydroxylated C-3 determined by a J-based configurational analysis. NMR experiments psHMBC, ¹H with a preacquisition delay set to long-range heteronuclear coupling, and HSQMBC were acquired on the full phormidepistatin structure (Figure 3; Figures S10–12). Due to the two prominent rotamers detected under NMR conditions, three of the five couplings were within the medium magnitude range. The full J-coupling magnitude pattern fit an erythro relative configuration with interchange between gauche^- and anti conformations.²³ However, attempts to determine the absolute configurations of the two DMPD secondary alcohols by Mosher ester analysis using several different conditions were not successful, even after separating DMPD from the peptidic portion of 1 by acid hydrolysis and esterifying its terminal carboxyl group (Figure S13). Thus, the absolute configuration of the DMPD subunit was not determined. The structure of phormidepistatin is [DMPD—D-Ser—D-Tyr—L-Pro—L-Thr—(3R,4S)-Sta—L-Tol].

Figure 3.

Relative configuration assessment between DMPD C2 and C3. The established coupling constant magnitudes are listed below each relative configuration/rotamer pair of Newman projections.²³ The C2-C3 relative configuration was established as erythro with the DMPD segment interchanging between gauche and anti conformations (red box).

Phormidepistatin was evaluated for activity against ovarian adenocarcinoma OVCAR3, breast adenocarcinoma MDA-MB-231, and melanoma MDA-MB-435 cell lines. At 10 μM concentrations, no cell inhibition was detected in any of the three cancer assays. The presence of an epi-Sta moiety led us to evaluate 1 for its inhibition of aspartic protease cathepsin D. Phormidepistatin was found to have marginal inhibition activity with an IC₅₀ of 21 μM (± 2; in triplicate). One likely reason for the limited activity of 1 is the absolute configuration at C-3 of the statine group. A comparison of multiple synthetic pepstatin A analogues containing a Sta or Sta-like amino acid with either a 3S,4S or 3R,4S configuration, found the 3S,4S isomers to be 10- to 100-fold more active than the 3R,4S counterparts.²⁴

Polyphyletic Production of Phormidepistatin.

Phormidepistatin was originally obtained from UIC 10484, a strain of the order Oscillatoriales previously identified by our group to produce laxaphycin type-B compounds.²⁵ This represents the first report of co-production of a Sta or Sta-like metabolite and a laxaphycin type-B metabolite. To determine if phormidepistatin is produced by other laxaphycin type-B strains in our strain library, the metabolomes of UIC 10339 and UIC 10045 were analyzed. The phylogenetically distinct strains UIC 10339, of the order Nostocales, and UIC 10045, of the order Oscillatoriales, are both known to produce laxaphycin type-A and -B metabolites.^20,21 (Photomicrographs of the three strains are found in Figure S14.) MS/MS analysis identified phormidepistatin (1) in all three strains, although UIC 10045 and UIC 10339 appear to produce the compound in much lower titer (Figure S15). Additional phormidepistatin analogues were detected in UIC 10484 using the Global Natural Products Social Molecular Networking platform (Figures S16 and S17).²⁶

Given the close evolutionary proximity of UIC 10045 to UIC 10484 (99.4% 16S rRNA sequence identity), it was unsurprising that UIC 10045 also produces 1. However, it was unexpected to find UIC 10339 produces phormidepistatin given its taxonomic distinction to UIC 10484 and UIC 10045 (<91.5% sequence identity). While the commonly co-occurring laxaphycin type-A and -B metabolites are proposed to be produced by a shared gene cluster,²⁷ genomic analysis of the three known phormidepistatin producers would provide insight into why two distant phylogenetic clades produce three of the same compound classes.

Cyanobacteria as a Source for Sta/Sta-like-containing Metabolites.

Due to the unique structure and biomedical importance of the statine, a structure search using the Natural Products Atlas and Reaxys databases was performed to identify known cyanobacterial metabolites containing a Sta or Sta-like moiety.^28,29 This resulted in the identification of 34 secondary metabolites, including phormidepistatin. Sta-like amino acids were defined as γ-amino acids with a ß-hydroxy or derivative of the ß-hydroxy group. While the standard (3S,4S)-statine, the original moiety established as an aspartic protease inhibitor pharmacophore, was detected in multiple cyanobacterial metabolites,^30–34 the majority of the compounds contained an alternative version of it.^35–48

In addition to structural variability of the amino acid, Sta/Sta-like-containing secondary metabolites were also highly diverse. Using the MayaChemTools package,⁴⁹ a structure similarity analysis using a topological torsion fingerprint was performed on the 34 known cyanobacterial structures (Table S2). This resulted in the partitioning of the natural products into 10 clusters (Figure 4a; Figures S18–20). Of the 10 clusters, referred to as compound classes moving forward, six are produced by marine cyanobacteria and four by freshwater strains (Tables S2 and S3). Of note, compound class I metabolite malyngamide 3 and class J metabolite guamamide have overlapping structural features, but were distinct enough to be clustered separately based on the parameters used in the analysis (see Experimental Section).

Figure 4.

The 10 known cyanobacterial compound classes (A-J) containing the Sta/Sta-like amino acid and their phylogenetic distribution. (a) Compound classes comprised of multiple known metabolites only include the core overlapping scaffold. The Sta/Sta-like amino acid(s) is highlighted red in each structure. All secondary metabolites from each class are in Figures S18–20. The colored boxes representing the compound classes correspond to the colored boxes in the adjacent phylogenetic tree. Compound classes were generated using RDKitClusterMolecules through the MayaChemTools chemoinformatics software package (see Experimental Section).⁴⁹ (b) The distribution of the Sta/Sta-like containing metabolites in relation to recognized type strains covering seven of the nine established taxonomic orders.⁵⁰ Taxonomic orders are listed to the right of the tree: Chroo. = Chroococcales; Gloeo. = Gloeobacterales; Nostoc. = Nostocales; Oscil. = Oscillatoriales; Pleur. = Pleurocapsales; Spiru. = Spirulinales; Synec. = Synechococcales. The strain name/identifier color indicates the environment in which the strain was collected. Colored lines connecting strains, or a group of strains, indicate the strains produce the same Sta-like compound class and have <95% 16S rRNA sequence identity, the minimum threshold for a genus.⁵¹ The 26-strain compound class B producer clade branch was collapsed for clarity. In lieu of the clade, there is a solid black triangle. The substitutions per site scale bar is below the tree and set at 0.05. The outgroup was excluded from the figure.

While only the compound class C metabolites contain the original statine moiety [(3S,4S)-4-amino-3-hydroxy-6-methylheptanoic acid],²² these data suggest cyanobacteria broadly incorporate the general ß-hydroxy γ-amino acid core into a diversity of chemical structures. The 10 compound classes are comprised of both linear and cyclic structures containing diverse peptidic and polyketide subunits. All compound classes except class F (hapalosin) include at least one analogue with a linear base structure. Of these, compound class B metabolites (dolastatin 10 and analogues) contain two consecutive Sta-like amino acids. Compound class F, class G (hoiamides A and B), and class H (multiple lyngbyabellins) all contain macrocycles. For these, the Sta-like amino acid is found either within the macrocycle (classes F and G) or as part of the linear extension (class H). Overall, the 10 known Sta/Sta-like-containing compound classes occupy broad chemical space with variable positioning of the pharmacophore or its derivatives.

To improve the understanding of the phylogenetic distribution of Sta/Sta-like-containing compound classes in cyanobacteria, a phylogenetic analysis was performed with known Sta/Sta-like-containing producers and the three phormidepistatin-producing UIC strains. At least one strain known to produce eight of the 10 compound classes (excludes class I and J producers) has an associated published 16S rRNA sequence. A phylogenetic tree was built comprised of these strains in addition to type strains accepted in the CyanoDB 2.0 database⁵⁰ to add taxonomic perspective (Figure 4b). In total, there are 40 known Sta/Sta-like producers with an established 16S rRNA sequence, of which 26 are known compound class B producers.

From this analysis, it was found that secondary metabolites containing a Sta or Sta-like moiety are produced by strains distributed throughout multiple phylogenetic groups comprising several taxonomic orders with strain morphologies ranging from coccoidal, non-heterocyst-forming filamentous, heterocyst-forming filamentous, and true branching filamentous. These phylogenetically diverse strains have been collected from both freshwater and marine environments (Table S3). Three of the 10 compound classes have been identified in more than one genus (compound classes A, B, and G), including the phormidepistatin series (class A).

Overall, the Sta/Sta-like-containing cyanobacterial compound classes have demonstrated activity in cancer cell and target-based assays.^{5,34,36,37,40–45,47,52,53} More impressively, a semisynthetic analogue of compound class B metabolite dolastatin 10 is the drug payload in three different antibody-drug conjugates used to treat cancer.^54–56 However, apart from compound class A phormidepistatin reported here, only class C metabolites have been evaluated against aspartic proteases. Class C, largely comprised of the standard 3S,4S-statine, has displayed high pM to low μM IC₅₀ values against cathepsins D and E in addition to BACE1 (Table S4).

With a dataset limited to only the known structure pool and thus a small representation of the full metabolic potential of cyanobacteria, we identified a large number of Sta/Sta-like-containing metabolites. Our data suggest the Sta/Sta-like amino acid is broadly incorporated into diverse chemistry produced by both freshwater and marine strains throughout the phylum. Given the biomedical potential in targeting aspartic proteases, there is potentially an untapped layer of cyanobacterial chemistry not yet known or even surveyed for drug discovery purposes.

EXPERIMENTAL SECTION

General Experimental Procedures.

The optical rotation was measured on a PerkinElmer 241 polarimeter. The UV spectrum was attained using a Varian Cary 5000 spectrophotometer while the IR spectrum was acquired using a Thermo-Nicolet 6700 FT-IR equipped with a Smart iTR ATR sampling accessory. NMR spectra, both 1D and 2D, were collected on a Bruker AVII 900 MHz NMR spectrometer with a 5 mm CPTCI RF probe. The ¹H and ¹³C spectra were calibrated using the DMSO-d₆ solvent resonance signals (δ_H 2.50 and δ_C 39.50 ppm). TFA vapor was added to the NMR samples by dispensing a plume of TFA residue through a 200 μL pipette tip with a severed end. The NMR tubes were vortexed to mix the acid with the sample and stored in a desiccator for 24 h prior to experimentation. A Shimadzu Nexera X2 multimodule UHPLC system with a reversed-phase C₁₈ UHPLC column (50×2.1mm, Phenomenex Kinetex) connected by capillary to a Bruker Impact II UHR Qq-TOF was used to obtain HR-LC-MS/MS data. Collision energies were set to 45, 52, and 60 eV for the phormidepistatin MS/MS analysis with the precursor mass window set between 1,000 – 1,500 m/z. Detection was only in positive mode. Both an Agilent 1100 HPLC system outfitted with a diode array detector and a Waters 996/515 modular HPLC system were used for isolation and analytical experiments.

Biological Material.

Cyanobacterial strains UIC 10339 and 10045 were collected and isolated as previously reported.^20,21 UIC 10484 was isolated using a standardized micropipette technique⁵⁷ from a freshwater sample collected from Wall Lake in Indiana, United States (latitude: 41.72572; longitude: −85.20015) in August 2014. It was subsequently grown in 150 mL BG-12 medium.⁵⁸ All cultures were grown in a room set to 22 °C and exposed to direct 1.03 klx fluorescent lighting 18 h per day. For large-scale growth, 10484 was cultured in both 2 L (2.8 L Fernbach flasks) and 10 L (13 L glass carboys) of medium exposed to 1.7 L/min of filtered ambient air. Large scale batches were cultured for six weeks. UIC 10484 is preserved in the Orjala lab culture library at the University of Illinois at Chicago.

Morphological, Taxonomic, and Phylogenetic Assessment of UIC Strains.

The morphological and 16S rRNA taxonomic assessments established UIC 10484 as cf. Phormidium sp. Using a Zeiss Axiostar Plus light microscope set to brightfield illumination at 400x, UIC 10484 was observed to be composed of unbranched isopolar filaments lacking heterocysts, which are typical Oscillatoriales features (Figure S14). DNA extraction, 16S rRNA sequencing, and the phylogenetic assessment were performed previously.²⁵ From this assessment, UIC 10484 (16S acc. no. MN453282) was found to be closely related to Phormidium irrigium provisional type strain CCALA759. UIC 10045 (acc. no. KF444211) was originally classified as cf. Oscillatoria sp. but re-classified as a cf. Phormidium sp.²⁵ UIC 10339 (acc. no. KF444210) was determined to be a cf. Trichormus sp.²¹ Photomicrographs of the three UIC strains are in Figure S14.

Extraction and Isolation of Phormidepistatin.

UIC 10484 was harvested from 136 L of BG-12 medium and lyophilized resulting in 33.5 g of total dry cell mass. The cell mass was soaked in CH₂Cl₂:MeOH (1:1) extraction solvent for 24 h and then filtered. This effort resulted in 8.81 g of extract, which was fractionated using Diaion HP-20SS resin. A step gradient was utilized with 0, 20, 40, 70, 90, and 100% IPA (fractions 1 to 6, in the same order). The compound was most abundant in fractions 3 and 4.

A semi-preparative C₁₈ reversed-phase column (Agilent, 250×10 mm) was used to separate the components in the fractions using a Waters 515 HPLC pump system connected to a Waters 996 Photodiode Array Detector. The HPLC method utilized was as follows: an initial 5 min isocratic 20% CH₃CN step followed by a 10 min gradient from 20–40% CH₃CN at 4.0 mL/min and a 6 min wash with 100% CH₃CN at 6.0 mL/min. Impure 1 was collected at 17.5 min and then purified with a second HPLC step using a C₁₈ reversed-phase column (Phenomenox, 250×4.6 mm). A preliminary 35% CH₃CN 5 min isocratic interval followed by a gradient from 35–60% CH₃CN over 5 min eluted phormidepistatin at 6.5 min. A total of 6.0 mg (0.02% of dry weight) was ultimately isolated from the 8.81 g of extract.

Phormidepistatin (1):

White, amorphous powder; [a]²²_D −16 (c 0.1, DMSO); UV (MeOH) λ_max (log ε) 274 (3.10); IR (neat) ν_max 3288 (br), 2922, 1680, 1639, 1551, 1517, 1450, 1208, 1138 cm⁻¹; ¹H and ¹³C NMR (DMSO-d₆) Table 1; HRESIMS m/z 1049.5839 [M+H]⁺ (calcd for C₅₅H₈₁N₆O₁₄, 1049.5805).

Absolute Configuration.

The absolute configurations of the amino acids were determined using the advanced Marfey’s method. An initial 0.1 mg aliquot of 1 was hydrolyzed in 1 mL of 6N HCl at 110 °C under vacuum. After 24 h, the hydrolysate was split into two equal amounts and dried under a stream of ambient air to remove HCl. To evaporate off residual HCl, 500 μL of H₂O was added to each vial, dried under an air stream, and repeated. The hydrolysate samples were derivatized with 1-fluoro-2,4-dinitrophenyl-5-leucinamide (L-FDLA or D/L-FDLA) following an established protocol.^21,59–61

Under the described hydrolytic conditions, the phormidepistatin epi-Sta racemized resulting in two statine diastereomers, an issue reported when hydrolyzing other compounds with the γ-amino acid.³⁰ Previous work has found epimerization during acid hydrolysis to occur at a higher rate with the peptide intact compared to when the peptide has been broken down into its free amino acids.⁶² To reduce epimerization, 0.1 mg of phormidepistatin was exposed to the same conditions but the reaction time was reduced from 24 h to 4 h. This reduction in time resulted in the one Sta diastereomer.

The absolute configurations of Tol and Ser were determined by HPLC co-injection with derivatized commercial standards. With a reversed-phase analytical C₁₈ column (Phenomenex, 250×4.6mm) housed in an Agilent 1100 HPLC system, the two amino acids were evaluated using the following method: an initial isocratic 1.0 mL/min flow of 15% CH₃CN (+0.1% formic acid) for 5 min, followed by a 30 min gradient from 15% to 35% CH₃CN. Due to peak overlap, Sta, Thr, Pro, and Tyr absolute configurations were assessed using a reversed-phase C₁₈ UHPLC column (Phenomenex Kinetex, 50×2.1mm) coupled to a Bruker Impact II UHR Qq-TOF mass spectrometer. Derivatized amino acid retention times were compared to DLA-derivatized commercial standard retention times with a gradient of 25% to 65% CH₃CN (+0.1% formic acid) over 9 min. The full advanced Marfey’s dataset is in Table S1. In addition, details regarding the Mosher ester analysis attempts are found in Figure S13.

Biological Activity.

Phormidepistatin was evaluated against human melanoma MDA-MB-435, breast adenocarcinoma MDA-MB-231, and ovarian adenocarcinoma OVCAR3 cell lines, which were purchased from American Type Culture Collection. Cells were grown and inhibition was assessed using a previously described protocol.⁶³ Paclitaxel was used as the positive control.

The target-based cathepsin D enzyme assay was carried out in triplicate by Reaction Biology Corporation. Pepstatin A was used as the positive control. Activity was assessed over 120 min measuring fluorescence induced by cleavage of MCA-PLGL-Dap(Dnp)-AR-NH₂ (328/420 nm excitation/emission). The reaction buffer for the assay was 100 mM CH₃COONa, 200 mM NaCl, 0.02% Brij 35, and 1% DMSO with a final pH of 3.5.

Culture and Extraction of UIC 10484, 10339, and 10045.

Strains UIC 10484, 10339, and 10045 were all grown in both Z and BG-12 media.^58,64 UIC 10484 was originally grown in BG-12 while both 10339 and 10045 were originally grown in Z.^20,21,25 The three strains were initially cultured in 100 mL of each medium for 15 days at 22 °C under direct 1.03 klx fluorescent light. After the 15-day exposure period, 50–55 mg of wet cell mass was transferred from the old media to 100 mL of fresh media of the same type and grown for three weeks under the same conditions. After three weeks, the cell mass was harvested and the media removed by centrifugation resulting in <0.5 mL residual media. The wet cell mass was transferred to 100 mL of CH₂Cl₂:MeOH (1:1) extract solvent and soaked for 24 h. The extract was then separated from the remaining cellular material using a separation funnel and dried using a rotary evaporator.

The most hydrophilic and hydrophobic extract components were minimized in all six samples (three strains grown in both media) using an SPE C₁₈ cartridge (100mg/1mL; Agela Technologies). Each sample was dissolved/suspended in 500 μL of 5%MeOH/95%H₂O, transferred to the cartridge, and then vacuumed through it. An additional 500 μL of the solvent mix was passed through the resin. A total of 5 mL MeOH was then passed through the cartridge, dried down, and prepared for MS/MS analysis.

A reversed-phase C₁₈ UHPLC column (Phenomenex Kinetex, 50×2.1mm) in a Shimadzu Nexera X2 series UHPLC system connected to a Bruker Impact II mass spectrometer was used for the analysis. Concentrations were set to 1.0 mg/mL with 3 μL injections. The sample constituents were separated with an initial 15% CH₃CN (+0.1% formic acid) isocratic step for 0.5 min followed by a gradient over 6 min from 15% CH₃CN to 100% CH₃CN at 0.5 mL/min. Presence of phormidepistatin was established by ion detection, retention time, and MS/MS fragmentation. To ensure the fragmentation of the [M+H]⁺ ions of phormidepistatin, the parent mass range for fragmentation was set between both 50–1500 and 1000–1500 m/z. The scan range was between 50–1500 m/z with a spectra rate set to 6.0 Hz. Tandem data were collected on the five most abundant precursor ions. Stepping mode was applied in which ions were exposed to collision energies between 35–70 eV. A target intensity of 20,000 counts was set with a maximum MS/MS spectra acquisition at 20.00 Hz and minimum of 3.00 Hz. The experiments were only performed in positive mode.

Natural Products Atlas Search Strategy.

A substructure search was performed using the Natural Products Atlas database.²⁸ A generic ß-hydroxylated γ-amino acid with the SMILES notation NCC(O)CC=O was used for the search. This resulted in 588 natural products containing an iteration of the generic Sta-like moiety. Cyanobacterial metabolites with variable side chains were accepted for the structure similarity analysis. Minor modifications to the ⍺- and ß-group (i.e., methylation, acetylation) were accepted as well. Dolastatin H and isodolastatin H were excluded because they have yet to be confirmed as independent cyanobacterial natural products.⁶⁵ The search was performed October 2020.

Reaxys Search Strategy.

Using the Reaxys chemistry search engine,²⁹ a query was run with the same substructure described in the previous section. To reduce the pool of chemistry, the ‘Isolated from Natural Source’ search field was selected. A ‘contains’ boolean operator phrase ‘cyanobacter’ was used to narrow the query to only cyanobacterial secondary metabolites. These criteria resulted in 34 compounds. The search was performed October 2020.

Structure Similarity Clustering.

The 34 known Sta/Sta-like-containing secondary metabolites, which include phormidepistatin and the 33 structures obtained from the two searches, were partitioned into similarity clusters using the RDKitClusterMolecules python script through the MayaChemTools software package (Table S2).⁴⁹ A topological torsions fingerprint with an fpSize of 16,384 was applied to characterize the structures. A tanimoto similarity function and a non-hierarchical Butina clustering algorithm with the cutoff set to 0.6 were used to cluster the compounds.

Phylogenetic Analysis of Sta/Sta-like-containing Secondary Metabolite Producers.

A total of 40 strains known to produce eight of the 10 compound classes had a published 16S rRNA sequence of sufficient length (>900 bp). Of those, 26 are established compound class B producers. The 40 cyanobacterial producing strains, 28 type strains from CyanoDB,⁵⁰ and a Melainabacteria outgroup (Vampirovibrio sp.; acc. no. HM038000) totaling 69 sequences were phylogenetically assessed. The type strains included were collected from marine, freshwater, hypersaline, and saline-alkaline environments. The Sta/Sta-like-producing strains with their associated accession numbers are listed in Table S3.

The Geneious Prime 2020.0.5 bioinformatics platform was used to perform tree-building tasks.⁶⁶ The 16S rRNA sequences for the 69 strains were aligned using MUSCLE.⁶⁷ After alignment, the sequence lengths were reduced to 930 bp. A maximum-likelihood method via a RaxML plug-in was used to build the tree.⁶⁸ The rapid bootstrapping algorithm was utilized running 1,000 total bootstraps for the analysis. Based on a jModelTest2 evaluation of the dataset, GTR+I+G was selected as the optimal nucleotide substitution model available in RaxML.⁶⁹