Infectious diseases caused by drug-resistant bacteria are one of the most serious problems currently facing the medical field, as warned by the CDC1 and the WHO.2, 3 Among these drug-resistant bacteria, methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococcus faecalis/faecium (VRE) are two of the most intractable nosocomial pathogens that show broad-range drug resistance. Vancomycin has been used as the last resort for the treatment of infectious diseases caused by S. aureus and enterococci; however, since 1986, VRE strains have emerged and become widespread.4, 5 Since the beginning of this century, vancomycin intermediate-resistant S. aureus have been frequently isolated6 and 33 strains of vancomycin high-level resistant S. aureus have also emerged in Asian countries and in the United States.7 Effective drugs against these pathogens are urgently needed.

During the course of our screening for antibiotics active against both methicillin-resistant S. aureus and VRE, pargamicin A (PRG-A, Figure 1) was discovered from the fermentation broth of soil actinomycete strain Amycolatopsis sp. ML1-hF4.8 PRG-A is a structurally unique cyclic peptide consisting of N-methyl-3-hydroxy valine, 4-hydroxy piperazic acid (4-OH-Pip), sarcosine, phenylalanine, N-hydroxy isoleucine (NOH-Ile) and piperazic acid (Pip), which shows excellent in vitro antibacterial activity that is either comparable to, or more potent than, that of other currently available drugs, including vancomycin.8 Our previous studies revealed that PRG-A exerted rapid bactericidal activity against staphylococci and enterococci via a mechanism that involved disruption of the membrane integrity of the target cells,9 a mode of action distinct from that of daptomycin.9, 10 During the process of optimization of PRG-A production, we discovered new active components PRG-B, -C and -D (Figure 1) in the culture broth of the PRG-A-producing strain Amycolatopsis sp. ML1-hF4. This present study investigates the fermentation, isolation, structural elucidation and antimicrobial activities of these new PRG-A-analogs and discusses the structure–activity relationship of PRGs.

Figure 1
figure 1

Structure of PRGs A, B, C and D.

Full size image

The PRG-A producing strain was cultured for 3 days at 30 °C in liquid media consisting of 2% galactose, 2% dextrin hydrate, 1% soytone, 0.5% corn steep liquor, 0.2% ammonium sulfate and 1.0% calcium carbonate (pH 7.0 before sterilization). This seed culture was inoculated (3%) into media consisting of 0.33% galactose, 0.33% dextrin hydrate, 0.17% glycerol, 0.17% soytone, 0.083% corn steep liquor, 0.033% ammonium sulfate and 0.2% calcium carbonate (pH 7.0 before sterilization), and cultivated for 7 days at 27 °C with shaking (180 r.p.m.). This media was sixfold diluted compared with that reported at the time of PRG-A discovery.8 The resulting 2.5 l culture broth was then centrifuged (3000 g, 10 °C, 15 min) and the supernatant was applied to a Diaion HP-20 column (Mitsubishi Chemical Co., Tokyo, Japan, 300 ml wet volume). The column was washed with 1 liter each of distilled water, 40% aqueous methanol and was eluted with 90% aqueous methanol. The eluted fraction was dried on a rotary evaporator (733.2 mg) and then purified by reverse-phase HPLC. The HPLC conditions were as follows: column, Shiseido CAPCELL PAK UG-120 (Tokyo, Japan, 30 × 250 mm); flow rate, 10 ml min−1; solvent, 38–47% CH3CN aq+0.01% trifluoro acetic acid, linear gradient; column oven, 48 °C; detection, 210 nm. PRG-B (5.1 mg), PRG-C (24.9 mg), PRG-D (50.1 mg) and PRG-A (26.4 mg) were eluted at 25–26, 44–45, 47–48 and 70–71 min, respectively.

The physicochemical properties of PRG-A, B, C and D are summarized in the Supplementary Material (Supplementary Table S1). The molecular formulae of PRG-A, B, C and D were C34H52N8O9, C33H50N8O9, C34H52N8O10 and C34H50N8O10, respectively, as determined by HR–ESI–MS and NMR spectra (1H and 13C NMR spectra, Supplementary Figures S1–S6; 2D-COSY spectra, Supplementary Figures S7–S9; and HR–ESI–MS, Supplementary Figures S10–S12). NMR data are summarized in Table 1. The molecular formulae, IR spectra, optical rotations and NMR spectral data of the PRGs indicated that they are closely related in structure, except for the Pip or NOH-Ile moieties. The structures of PRG-B, C and D were determined by comparison with the NMR data for PRG-A and the structural differences are shown in Figure 2.

Table 1 NMR data of PRGs A, B, C and D
Full size table
Figure 2
figure 2

Structural difference of PRGs by NMR analyses.

Full size image

The molecular formula of PRG-B was C33H50N8O9, which was smaller by one CH2 unit than PRG-A. The 1H correlations of PRG-B changed the signals at δ 5.37, 2.41, 0.64 and 0.96 compared with those in PRG-A at δ 5.38, 2.11, 0.59, 1.33 and 0.80, as determined by the analysis of the 1H-1H COSY spectrum for PRG-B. These findings indicated that PRG-B possessed a N-hydroxyvaline (NOH-Val) moiety instead of the NOH-Ile moiety present in PRG-A. The order of connectivity of these Pip and NOH-Val moieties was determined to be the same as that observed for PRG-A by analysis of the HMBC spectrum (from δ 5.37 (NOH-Val, H-α) and δ 5.12 (Pip, NH) to 171.9 (NOH-Val, CO); 5.65 (Pip, H-α) and 3.24 (N-methyl-3-hydroxy valine, NMe) to 171.4 (Pip, CO)). The other NMR data and the molecular formula (C33H50N8O9) were all consistent with this change. Thus, the structure of PRG-B was determined as shown in Figure 1.

The molecular formula of PRG-C was C34H52N8O10, which was larger by one oxygen atom than PRG-A. Analysis of the 1H-1H COSY spectrum for PRG-C revealed signals at δ 5.71, 1.94, 2.13, 3.72, 2.84, 3.11, 5.02 and 6.48, compared with signals at δ 5.63, 1.85, 1.47, 2.38, 2.62, 3.11 and 5.12 for PRG-A. This indicated that PRG-C possessed a 4-OH-Pip moiety instead of the Pip moiety observed in PRG-A. The order of connectivity of these 4-OH-Pip and NOH-Ile moieties was determined to be the same as that seen for PRG-A by analysis of the HMBC spectrum (from δ 5.71 (4-OH-Pip, H-α) and 3.33 (N-methyl-3-hydroxy valine, NMe) to 172.4 (4-OH-Pip, C-α); 5.46 (NOH-Ile, H-α) and δ 5.02 (4-OH-Pip, NH) to 172.1 (NOH-Ile, CO)). The other NMR data and the molecular formula (C34H52N8O10) were all consistent with this change. Thus, the structure of PRG-C was determined as shown in Figure 1.

The molecular formula of PRG-D was C34H50N8O10, which was smaller by one H2 unit than PRG-C. Analysis of the 1H-1H COSY spectrum revealed signals at δ 6.30, 2.49, 3.40, 3.46 and 5.08 for PRG-D, compared with signals at δ 5.71, 1.94, 2.13, 3.72, 2.84, 3.11, 5.02 and 6.46 for PRG-C. The connectivity of the moiety determined by HMBC and the cross-peaks observed from δ 3.40 and 2.49 to the carbonyl carbon at δ 202.2 indicated that PRG-D possessed a 4-oxo-piperazine (O-Pip) moiety instead of the 4-OH-Pip moiety observed in PRG-C. The order of connectivity of these O-Pip and NOH-Ile moieties was determined to be the same as in PRG-C, as determined by analysis of the HMBC spectrum (from δ 6.30 (O-Pip, H-α) and 3.21 (N-methyl-3-hydroxy valine, NMe) to 169.0 (O-Pip, CO); 5.69 (NOH-Ile, H-α) and 5.08 (O-Pip, NH) to 172.6 (NOH-Ile, CO)). The other NMR data and the molecular formula (C34H50N8O10) were all consistent with this change. Thus, the structure of PRG-D was determined as shown in Figure 1. Taken together, these results revealed that PRG-B, C and D are new cyclic peptide metabolites.

The stereochemistry of hydroxyl group of 4-OH-Pip was determined by 2D ROESY spectrum (Supplementary Figure S13) of PRG-C. ROEs from H-βax (2.13 p.p.m.) to H-δax (3.07 p.p.m.) in 4-OH-Pip indicated a 1,3-diaxial orientation, suggesting 4-OH-Pip is a chair formation and the correlation from H-α (5.26 p.p.m.) to H-β (2.13, 2.51 p.p.m.) revealed the H-α was equatorial. Furthermore, ROESY correlations from H-γ (3.95 p.p.m.) to both axial and equatorial of H-β and H−δ (2.94 and 3.07 p.p.m.) suggested that H-γ was equatorial and the γ–hydroxyl group was axial established the relative stereochemistry of 4-OH-Pip is R*,R*. The 4-OH-Pip2 in PRG-C was determined by the same way as R*,R*.

The stereochemistry of constituent amino acids of PRG A, B, C and D were determined by advanced Marfey’s method using 1-fluoro-2,4-dinitrophenyl-5-D- and L-leucineamide11 (see Supplementary Materials, Supplementary Figures S14–S17). The acid hydrolysates of PRGs were derivatized by treating with 1-fluoro-2,4-dinitrophenyl-5-D- and L-leucineamide, and analyzed using extracted ion chromatogram by LC–HR–ESI–MS. We referred the report of dentigerumycin to analyze the elution pattern of both D-Pip and (2S, 4S)-4-OH-Pip.12 It was revealed that PRG-A contained D-Pip, D-Phe, (2R, 4R)-4-OH-Pip and NMe-3-OH-L-Val, PRG-B contained D-Pip, D-Phe and NMe-3-OH-L-Val, and PRG-C and D contained D-Phe, (2R, 4R)-4-OH-Pip and NMe-3-OH-L-Val, respectively. NOH-Val and NOH-Ile might be epimerized during acid hydrolysis. The 1-fluoro-2,4-dinitrophenyl-5-D- and L-leucineamide derivatives of 4-OH-Pip in PRG-B and O-Pip in PRG-D were not detectable. Collectively, the absolute structures of PRGs are shown in Figure 1.

The antimicrobial activities of the PRGs are summarized together with that of vancomycin in Table 2. MICs were determined as previously reported.13 PRG-A and PRG-C exhibited potent antibacterial activity against Gram-positive bacteria including methicillin-resistant S. aureus and VRE; the antibacterial activity of PRG-B and PRG-D against these bacteria was weaker. PRG-C and PRG-D, which possess a polar group in the northern region of Pip, exhibited four- to eightfold weaker antibacterial activity against staphylococci than against enterococci, whereas PRG-A and PRG-B showed equal activity against staphylococci and enterococci. This suggested that the presence of polar groups in the northern region of Pip might be disadvantage to the interaction with staphylococcal membrane. VRE were as sensitive to all of the PRGs as vancomycin-sensitive enterococci and all of the tested compounds, including vancomycin, showed no activity against Gram-negative bacteria.

Table 2 Antimicrobial activities of pargamicins
Full size table

The only structural differences between the PRGs are in the Pip-(NOH-Ile) moiety of PRG-A. However, despite this structural similarity, the antimicrobial activities of the PRGs showed remarkable differences. This suggested that the side chain of isoleucine residues and the northern Pip structures might have an important role in the interaction with the bacterial membrane and the localization of PRGs on the bacterial surface. Further studies into Pip containing cyclic peptides, such as the PRGs, may therefore result in the discovery of new types of antibacterial drugs that selectively disturb the bacterial membrane.

Dedication

The authors dedicate this work to Professor Satoshi Ōmura, a distinguished Novel Prize awardee in Physiology or Medicine 2015.