The pellet of the Photobacterium sp. was extracted with organic solvents (see Experimental Section 3.2) and fractionated on a diol column. Mass spectrometric analysis revealed a series of peptide-like analogues which display good ionization in ESI⁺ MS and end absorption in UV spectroscopy. The fractions containing these analogues were pooled and subjected to further purification, first on a smaller diol column and then by preparative reversed phase HPLC, which gave four fractions of ngercheumicin F, G, H, and I, respectively. Ngercheumicins A and B were also detected by LC-MS, but they were not purified in sufficient amounts for structural or biological screening studies. The four new ngercheumicins were found to inhibit transcription of the regulatory rnaIII in S. aureus, which is the effector molecule of the agr QS system (See Section 2.2 and Supplementary Information).The ngercheumicins were isolated as white solids with the respective exact masses (HR-ESI-TOF) and molecular formulae: Ngercheumicin F (m/z: [M + H]⁺ 853.5685, calculated for C₄₃H₇₇N₆O₁₁ as 853.5650), ngercheumicin G (m/z: [M + H]⁺ 855.5907, calculated for C₄₃H₇₉N₆O₁₁ as 855.5807), ngercheumicin H (m/z: [M + H]⁺ 881.6033, calculated for C₄₅H₈₁N₆O₁₁ as 881.5963), and ngercheumicin I (m/z: [M + H]⁺ 883.6255, calculated for C₄₅H₈₃N₆O₁₁ as 883.6120).

Analysis of 1D and 2D nuclear magnetic resonance (NMR) spectroscopic data obtained for the four compounds characterized the structures as cyclodepsipeptides consisting of six amino acids and a 3-hydroxy fatty acid (six NH signals and seven carbonyl resonances) (Figure 1). All four ngercheumicins were elucidated as having identical amino acid sequences, consisting of three leucines, two threonines, and one serine, as established by DQF-COSY, gHSQC, gHMBC, gH2BC, and TOCSY 2D NMR spectroscopic analyses. The closure of the macrocyclic ring through an ester linkage between the C-terminus and the hydroxyl group in one of the threonine residues was verified by the low field chemical shift of the β-proton of Thr² (H26, Table 1) and a HMBC correlation between H26 and the carbonyl (C1) of the C-terminal Leu¹ residue (for HMBC and H2BC correlations, see Supplementary Information). This formed a 16-membered macrocycle with an exocyclic chain continuing in the N-terminal direction from Thr², the chain constituting a Leu residue and a 3-hydroxy fatty acid (Figure 1). The structural difference between the four analogues was found in the length and saturation of the unbranched fatty acid chain. Ngercheumicin F and H each have one double bond in the 3-hydroxy fatty acid chain, whereas the fatty acids in ngercheumicin G and I are fully saturated, but with the same lengths as F and H, respectively. The previously isolated ngercheumicin A and B have a similar relationship, but with 12-carbon long fatty acid chains [8], whereas ngercheumicin F and H have fatty acid chains with 14 and 16 carbon atoms, respectively. The longer chains result in increasing overlap of ¹H and ¹³C resonances in the aliphatic regions on both sides of the double bond. By thorough examination of HMBC and H2BC correlations in the chains, the structures of ngercheumicin F and H were elucidated and assigned as shown in Figure 1 and Table 1, respectively. The chemical shifts of ngercheumicin H were virtually symmetrical around the double bond because of the long chain, whereas there was a slight shift in ngercheumicin F, which has a shorter alkyl chain. The position of the double bond in ngercheumicin H was tentatively assigned based on the NMR data, as the data left an ambiguity of a CH₂-group between position 41 and 47 due to the high degree of symmetry and the overlapping resonances that 2D NMR was unable to resolve. However, assuming a correct assignment, a structural pattern emerged where ngercheumicin A, F, and H had their double bonds positioned seven carbon atoms from the end of the fatty acid chain, also known as n-7 fatty acids. Counting from the peptide-end of the fatty acid chain, the double bonds in ngercheumicin F and H are thus situated further into the fatty acid chains.

The resonances originating from the double bond were very close in chemical shift, leading to severe second order effects in the ¹H multiplet patterns. Due to the second order spin systems, it was not possible to determine the size of the J coupling constant, however the total span of the multiplets at 5.31 ppm for both F and H was below 15 Hz, and assuming complete symmetry, a trans coupling seemed unlikely. The chemical shifts for the allylic carbons at 26.5–26.6 ppm are also consistent with cis configuration, as allylic carbons in trans fatty acids are about 5 ppm further downfield [13]. Therefore the double bonds in ngercheumicin F and H were assigned as cis.

Attempts to obtain absolute stereochemical assignment of the amino acids were done by Marfey’s method, using existing methods for acid hydrolysis [14] and derivatization with Marfey’s reagent [15]. Complete stereochemical assignments were not obtained as Marfey’s method revealed the presence of both l- and d-Leu, and unambiguous resolution of the hydroxyamino acids Ser and Thr is known to be challenging [16]. Pure enantiomers of Leu, Ser and Thr were used to synthesise single diastereomers with Marfey’s reagent for comparison with the Marfey’s derivatives of the hydrolysed ngercheumicins. This also included allo-Thr. By comparison to the pure amino acid derivatives, l-Ser, d-Thr and d-allo-Thr were dismissed. Therefore the configuration of the Ser residue was firmly assigned as d-Ser. Although the peaks of l-Thr, d-Ser, and l-allo-Thr eluted within a narrow spectral window, the elution order together with MS detection verified the presence of l-allo-Thr and no l-Thr. This was supported by the size of the J coupling constant from Thr² H26 to H25 and H27. The absolute configuration of the three Leu residues was ambiguous, however both l- and d-Leu were present. Due to the minute amounts available, (0.5–1.1 mg of each analogue) the configuration of the 3-hydroxy fatty acid was not determined.

Organic synthesis could be a solution to the supply problem of these peptides, as synthesis of cyclic peptides is often a relatively straightforward procedure. Many cyclic peptides from marine microbial sources contain non-proteinogenic and d-amino acids as well as polyketide-derived structural motifs or fatty acids [17], which is also the case for the solonamides and the ngercheumicins. Undoubtedly, these traits complicate synthesis of both natural products and analogues. However, a study by Molhoek et al. showed that cyclization of a peptide antibiotic and substitution of l- with d-amino acids improved stability towards bacterial proteases (including those in S. aureus) and decreased cytotoxicity while retaining the antibacterial activity [18]. The ngercheumicin macrocycle is closed through an ester bond between the C-terminus and the side chain of a Thr residue. This is a common trait in depsipeptides, and e.g., the cyclodepsipeptide plitidepsin [2], which is undergoing clinical trials for treatment of several cancers, has this feature in common with the ngercheumicins.

Ngercheumicin F, G, H, and I were examined in the S. aureus lacZ reporter assays described by Nielsen et al. [19] monitoring transcriptional activity of the hla, spa, and P3 (rnaIII) promotors. All four ngercheumicins increased transcription of spa and reduced expression of hla and rnaIII, compared to a solvent control (see Supplementary Information). The inverse effect of the ngercheumicins on spa expression compared to rnaIII and hla indicates that the ngercheumicins interfere with agr activation.

To confirm these initial results, cultures of a highly virulent, community-acquired strain of S. aureus USA300 were exposed to ngercheumicins F–I and RNAIII expression was monitored in exponential and stationary growth phase by Northern blot analysis. Data confirmed that ngercheumicins reduced expression of rnaIII and thus interfere with agr in CA-MRSA strain USA300. At the ngercheumicin concentrations used in the culture samples for Northern blotting, there was a minor down regulation of rnaIII by ngercheumicin G and H at 5 µg mL⁻¹ (see Supplementary Information). At 20 µg mL⁻¹ (23 μM, see Figure 2), the rnaIII inhibiting effect was more pronounced for ngercheumicin G and H, but also F showed activity here and appeared to have the highest rnaIII inhibiting activity. Ngercheumicin I did not show any significant effect at either 5 or 20 µg mL⁻¹.

The ngercheumicins share structural traits with the AIPs of S. aureus. S. aureus strains express one of a least four variants of AIPs and each variant induces agr expression in strains of the same type but repress agr expression in strains of other types [20]. Interestingly, the ngercheumicins resemble the type II and III AIPs, whereas the solonamides resemble type I and II the most [21]. All have 16-membered macrocyclic rings and flexible exocyclic chains (Figure 3). The AIPs have purely peptidic exocyclic chains, whereas the depsipeptides all contain fatty acid chains, but both the AIPs and the cyclodepsipeptides have variations in the length of the exocyclic chains, which perhaps relate to receptor specificity. Looking at the amino acid sequence of the ngercheumicins (Figure 3) starting at the C-terminus, the two Leu residues are also found in AIP-III, and the Thr-Ser sequence closely resembles the Ser-Ser sequence in AIP-II. It should be noted that this comparison is made without taking 3D configuration of the side chains into account, because only the 2D structures are known. It is however clear that part of the backbone structure is highly homologous to the native AIPs of S. aureus.

Little is known about peptide signaling in the marine environment and whether other groups of peptide signal molecules exist there. It can be speculated that these depsipeptides interfere with QS pathways present in the marine environment or even act as alternative quorum sensing molecules. Existing QS systems include the N-acyl homoserine lactones in Gram-negative bacteria and the auto inducing peptides in the Gram-positive staphylococci [22]. That ngercheumicins F–I were isolated in low yields as intracellular metabolites, unlike the solonamides that were excreted in large amounts, could mean that they do not act as intercellular signaling molecules in the natural environment and therefore there could be an unexplored biological role for the ngercheumicins. Analyses of co-cultures of this Photobacterium with naturally co-occurring or competitive strains could provide further insight into the biological function of these compounds.

Bacterial strain s2753 was isolated from a mussel surface collected in the tropical Pacific (9.1° S 156.8° E) during the Danish Galathea 3 expedition [3]. S2753 was assigned to the Vibrionaceae and sequence similarity of household genes identified it as a Photobacterium halotolerans as previously described [3,4,5].

S2753 was cultured in 10 L glass fermentors in 5 × 4 L SSS containing 0.4% glucose and 0.3% casamino acids in (25 °C, 72 h, 100 rpm). The liquid culture was centrifuged (15 min, 3500 × g) to isolate the pellet from the broth and Diaion HP20SS (12 g L⁻¹) which was used to extract the bioactive compounds holomycin and solonamides A–B as previously described [4,5]. The pellet was extracted with 1 L 1:9 (v/v) MeOH/EtOAc (25 °C, 24 h, 100 rpm) and filtered off through a Watman 1 filter. The pellet extract was concentrated on a rotary evaporator and absorbed onto 5 g Isolute diol (Biotage, Uppsala, Sweden) for dry loading onto a 50 g SNAP column packed with Isolute diol and eluted on an Isolera automated flash system (Biotage, Uppsala, Sweden) using solvents ranging from heptane, dichloromethane, EtOAc to pure MeOH (30 mL min⁻¹, 72 min). A total of 33 fractions were collected and subjected to LC-UV/MS. Fractions 7 to 10 (dichloromethane/EtOAc) were pooled and absorbed onto 1.5 g Isolute diol and further fractionated on a 10 g diol column run by gravity with heptanes, dichloromethane, EtOAc and MeOH as above. This yielded 10 fractions (again subjected to LC-UV/MS) of which fractions 6 and 7 (10%–30% MeOH in EtOAc) were purified on a Luna II column (5 μm C₁₈, 250 × 10 mm ID, Phenomenex) in a Gilson 322 liquid chromatograph with a 215 liquid handler/injector (BioLab, Risskov, Denmark) going from 70% to 100% aqueous MeCN, 20 mM formic acid, over 10 min followed by 6 min isocratic elution. This yielded 12 fractions of which pure compounds were obtained directly: Ngercheumicin F (0.5 mg), Ngercheumicin G (1.0 mg), Ngercheumicin H (0.5 mg), and Ngercheumicin I (1.1 mg). Selected chromatograms are available in the Supplementary Information.

LC-UV/MS analyses were performed on an Agilent 1100 HPLC system with a diode array detector coupled to an LCT TOF mass spectrometer (Micromass, Manchester, UK) using a Z-spray ESI source. Separation was performed at 40 °C with a Luna II C₁₈ column (50 × 2 mm ID, 3 μm, Phenomenex, Torrance, CA, USA), applying a linear gradient of 15%–100% aqueous MeCN, 20 mM formic acid (LC-MS-grade), over 20 minutes at a flow rate of 0.3 mL/min. MS experiments were performed in ESI⁺ with a data acquisition range of m/z 100–2000. Accurate masses of ammonium adducts were measured for Ngercheumicin F (m/z: [M + NH₄]⁺ 870.5979, calculated for C₄₃H₈₀N₇O₁₁ as 870.5916), ngercheumicin G (m/z: [M + NH₄]⁺ 872.6206, calculated for C₄₃H₈₂N₇O₁₁ as 872.6072), ngercheumicin H (m/z: [M + NH₄]⁺ 898.6359, calculated for C₄₅H₈₄N₇O₁₁ as 898.6229), and ngercheumicin I (m/z: [M + NH₄]⁺ 900.6450, calculated for C₄₅H₈₆N₇O₁₁ as 900.6385). The [M + H]⁺ adducts were reported in Section 2.1.

To solve the absolute configuration of the amino acids Marfey’s method was applied: 100 μg of each cyclodepsipeptide was subjected to acid hydrolysis (200 μL 6 M HCl, 110 °C, 20 h), redissolved in 50 μL water and added 20 μL 1 M aqueous NaHCO₃, then derivatised with 100 μL 1% w/v Marfey’s reagent in acetone (1-fluoro-2,4-dinitrophenyl-5-l-alanine amide, FDAA, Sigma-Aldrich, St. Louis, MO, USA) at 40 °C for 1 h as described by Bonnard et al. [15]. The reaction mixtures were neutralised with 10 μL 2 M aqueous HCl and diluted with 820 μL MeOH. Pure amino acid standards were derivatised by the same procedure using 50 μL 50 mM aqueous amino acid solution. Ultra-high performance liquid chromatography-diode array (UHPLC-DAD) separation and detection of the amino acid derivatives was done on a Dionex RSLC Ultimate 3000 (Dionex, Sunnyvale, CA, USA) equipped with a diode array detector. The separation was done in a Kinetex C₁₈ column (150 × 2.10 mm, 2.6 μm, Phenomenex) at 60 °C with a flow rate of 0.8 mL min⁻¹ using two different linear gradient methods. Method A included all l- and d-amino acids in the structure, whereas method B was run with shallow gradient and included l- and d-allo-Thr but not l- and d-Leu. This was an attempt to distinguish derivatives of d-Ser, l-Thr and l-allo-Thr which had very similar chromatographic properties on the column.

Method A: From 8% to 15% aqueous MeCN, 0.65 mM TFA, over 22 min followed by an increase from 15% to 100% for 8.5 min. Retention times for the FDAA-amino acid derivatives were: l-Ser (4.21 min), l-Thr (5.46 min), d-Ser (5.56 min), d-Thr (12.38 min), l-Leu (25.23 min), d-Leu (26.00 min). Unreacted FDAA eluted at 11.0 min.

Method B: From 8% to 10% aqueous MeCN, 0.65 mM TFA, over 25 min followed by an increase from 10% to 100% over 5.5 min. Retention times for the FDAA-amino acid derivatives were: l-Ser (4.46 min), l-Thr (5.90 min), d-Ser (5.97 min), l-allo-Thr (6.15 min), d-allo-Thr (10.28 min), d-Thr (15.82 min). Unreacted FDAA eluted at 10.4 min.

NMR spectra were recorded on a Bruker Avance 800 MHz spectrometer equipped with a 5 mm TCI Cryoprobe using standard pulse sequences. The NMR data used for the structural assignments were acquired in DMSO-d₆ (δ_H 2.49 ppm and δ_C 39.5 ppm). ¹H and ¹³C NMR spectra are available in the Supplementary Information.

The S. aureus lacZ reporter assays were performed as described by Nielsen et al. [19]. Ngercheumicin F, G, H, and I were dissolved in DMSO, and DMSO and H₂O was included as negative controls in the assay. Pictures were taken after 11, 13 and 34 h of incubation for the rnaIII-, hla- and spa-reporter strains respectively (Supplementary Information).

RNA for Northern blotting was purified from USA300 (FPR3757) samples from cultures grown in 100 mL Erlenmeyer flasks containing 10 mL Tryptone Soya Broth (TSB, Oxoid, Greve, Denmark) shaking vigorously (200 rpm) in a water bath at 37 °C. Ngercheumicins (5 µg mL⁻¹ and 20 µg mL⁻¹) and DMSO (control) were added at OD₆₀₀ = 0.4, and samples were taken at OD₆₀₀ = 0.8 and 3.0. Northern blotting was performed as previously described [5] using an RNAIII-probe constructed using primer rnaIII forward (5′-GGG GAT CAC AGA GAT GTG ATG-3′), and rnaIII reverse (5′-GGG CAT AGC ACT GAG TCC AAG G-3′)(TAG Copenhagen A/S, Frederiksberg, Denmark).