Introduction

Our research group has focused on the discovery of anti-infectives from microbial metabolites.1, 2, 3 In the course of in vivo-mimic screening using silkworm larvae to discovery antibiotics that are active against methicillin-resistant Staphylococcus aureus, nosokomycins A, B, C and D (Figure 1) were isolated as active components from the culture broth of Streptomyces sp. K04-0144.4 In this study, the physico-chemical properties and structure elucidation of nosokomycins are described.

Figure 1
figure 1

Structures of nosokomycins A (1), B (2), C (3) and D (4).

Full size image

Results

Physico-chemical properties of nosokomycins

The physico-chemical properties of nosokomycins are summarized in Table 1. The strong IR absorption at 3400 and 1720 cm−1 suggested the presence of a hydroxyl group and a carbonyl group. In their UV spectra, nosokomycins showed no characteristic absorption other than end absorption, and no reliable peaks related to MW were observed for this series of compounds during FAB-MS measurements. Therefore, LC-ESI-MS (magnetic sector-type) was used to elucidate the MWs of nosokomycins (Table 1). The similarities in their physico-chemical properties strongly suggested that they are structurally related.

Table 1 Physico-chemical properties of nosokomycins A (1) to D (4)
Full size table

Structure elucidation of nosokomycin B (2)

The first structure study was carried out for nosokomycin B (2), the major active component among them. The analysis of its 1H, 13C and 2D NMR spectral data indicated that 2 is a moenomycin-like phosphoglycolipid antibiotic, which consists of five substituted sugar moieties B–F, a 2,3-dihydroxypropionic acid (H) and an unusual sesterterpenoid moiety with the formula C25H39 (I) in 2 (Figure 2). However, the end absorption of 2 in UV spectra showed a lack of the 2-hydroxy-5-oxo-1-cyclopenten-1-ly moiety (A) of moenomycin A, which had the charcteristic absoption. The molecular formula of 2 was determined by HR-ESI-MS to be C64H104N5O32P (found m/z 1484.6356 (M-H)−, calcd 1484.6324) in conjunction with its NMR data (Table 2). The presence of a phosphorus atom in 2 was clarified by the observation of the 2JC-P coupling (4.6 and 4.4 Hz) and 3JC-P couplings (9.1 and 7.2 Hz) in the 13C NMR spectra. The sesquiterpenoid moiety (I) was assigned as moenomycicol, the moiety of moenomycin A by COSY and HMBC analysis. The 3JH-H coupling constants (J1, 2=8.6 Hz, J2, 3=8.6 Hz, J3, 4=1.4 Hz and J4, 5=1.4 Hz) and the HMBC correlation between the proton signal of C-5 (δH 4.05) and the primary amide carbon signal (δC 173.2) proved the α-D-glucuronamate of sugar B. The 3JH-H coupling constants (J1, 2=8.5 Hz, J2, 3=8.6 Hz, J3, 4=8.6 Hz and J4, 5=8.6 Hz) and the HMBC correlation between the proton signal of C-2 (δH 3.82) and the acetyl methyl signal (δH 2.06) and the acetyl carbonyl carbon signal (δC 174.1) proved the 6-deoxy-N-acetyl-α-D-glucosamate of sugar C. The 3JH-H coupling constants (J1, 2=8.6 Hz, J2, 3=8.6 Hz, J3, 4=8.6 Hz and J4, 5=8.5 Hz) and the HMBC correlation between the proton signal of C-2 (δH 3.82) and the acetyl methyl signal (δH 2.06) proved the α-D-glucosamate of sugar D. The 3JH-H coupling constants (J1, 2=8.5 Hz, J2, 3=8.5 Hz, J3, 4=8.6 Hz and J4, 5=8.7 Hz) and the HMBC correlation between the proton signal of C-2 (δH 3.79) and acetyl methyl signal (δH 2.00) and the acetyl carbonyl carbon signal (δC 173.7) proved the N-acetyl-α-D-glucosaminate of sugar E. The 3JH-H coupling constants (J1, 2=1.8 Hz, J2, 3=10.1 Hz) and the NOE correlations between the proton signals of C-2 (δH 3.64) and the singlet methyl group (δH 1.23), and the HMBC correlations between the proton signal of C-3 (δH 5.10) and the ureido carbonyl carbon signal (δC 159.2) and between the proton signal of C-5 (δH 4.51) and the primary amide carbon signal (δC 174.7) proved the 3-O-carbamoyl-4-methyl-α-D-glucuronaminate of sugar F. The connectivity of sugar moeities was determined by the correlations between the anomeric proton signals and the corresponding carbon signals (C-4 of sugar C (δH 4.51) and C-1 of sugar B (δC 105.1), C-1 of sugar C (δH 4.57) and C-4 of sugar E (δC 82.4), C-1 of sugar D (δH 4.45) and C-6 of sugar E (δC 69.4), and C-1 of sugar E (δH 4.51) and C-2 of sugar F (δC 79.7)) through glycoside bonds in the HMBC spectra (Figure 3). The mode of linkage of all sugar moieties as the α-glycoside bond except sugar F, was assigned by measuring the 1H-1H coupling constants of anomeric protons. The connectivity between the sesterterpenoid moiety (I) and the 2,3-dihydroxypropionic acid (H) was determined by the correlations between the proton signal of C-1 of I (δH 4.22) and the carbon signal of C-2 of H (δC 81.2) observed in HMBC spectra (Figure 3). Finally, the linkage between the C-1 of the partial structure F and the C-3 of the partial structure H was elucidated by the correlation from the proton signal of C-1 of F (δH 5.95) and C-3 of H (δH 4.10) and the phosphorus signal of P of phosphate ester (δP-0.659) seen in 1H-31P HMBC spectra (Figure 3).

Figure 2
figure 2

Partial structures of nosokomycin B (2).

Full size image
Table 2 NMR spectral data of nosokomycins A (1) to D (4)
Full size table
Figure 3
figure 3

The linkages between the partial structures in nosokomycin B (2).

Full size image

From these results, the structure of 2 was elucidated as shown in Figure 1, which was a moenomycin A analog lacking moiety A. 2 was identified with the semisynthetic moenomycin derivative.6 Very recently, a biosynthetic intermediate was found by mass and MS2 fragmentation data.7

Structure elucidation of nosokomycin A (1)

The molecular formula of 1 was determined by HR-ESI-MS to be C64H103N4O33P (found m/z 1485.6206 (M-H)−, calcd 1485.6164) and the results indicated the replacement of NH in 2 with O in 1. As the 1H and 13C NMR spectra of 1 (Table 2) resembled those of 2, it was difficult to elucidate the differences in structures by only measuring the chemical shifts among candidates (the primary amide group at C-5 in sugar B and at C-5 in sugar F). To observe the correlation between the proton signal at δH 4.05 (C-5 of sugar B) and the nitrogen signal at δN 269.4 (N of the primary amide group of sugar B), 1H-15N HMBC experiments were carried out (Figure 3). However, these correlation results were not seen on 1, indicating that the primary amino group belonging to the C-5 of sugar B in 2 is replaced with a hydroxyl group in 1.

The presence of this compound in the flavomycin complex was predicted by LC-ESI-IT-MS analysis.8 Very recently, 1 was reported to be an intermediate of moenomycin biosynthesis, the structure of which was deduced by exact mass and MS2 fragmentation data.7 However, we first isolated 1 as the metabolite from the culture broth of wild-type actinomycete and obtained its full spectral data.

Structure elucidation of nosokomycin C (3)

The molecular formula of 3 was determined by HR-ESI-MS to be C58H93N4O28P (found m/z 1323.5656 (M-H)−, calcd 1323.5636), indicating that 3 lacked C6H10O5 (one sugar unit) compared with 1. From a comparison of the 1H and 13C NMR spectra of 3 and 1 (Table 2), the signals of sugar D seen in 1 were absent in 3, and the chemical shift of C-6 in sugar E in 3 was shifted to a higher field in the 13C NMR spectra (δC 69.4 in 1, δC 60.8 in 3). Therefore, the structure of 3 was elucidated to be the nosokomycin A analog lacking moiety D, as shown in Figure 1.

Structure elucidation of nosokomycin D (4)

The molecular formula of 4 was determined by HR-ESI-MS to be C58H94N5O27P (found m/z 1322.5859 (M-H)−, calcd 1322.5795), which was smaller than that of 2 by C6H10O5. From a comparison of the 1H and 13C NMR spectra of 4 and 2 (Table 2), the signals of sugar D in 2 were found to be absent in 4, and the chemical shift of C-6 seen in sugar E in 4 was shifted to a higher field in 13C NMR spectra (δC 70.1 in 2, δC 61.0 in 4). Therefore, the structure of 4 was elucidated to be the nosokomycin B analog lacking moiety D, as shown in Figure 1.

Discussion

According to the structures of the nosokomycins elucidated in this study, they belong to the moenomycin family of antibiotics. Nosokomycin B (2), which lacks the chromophoric cyclopentenone moiety (A), was identical to the semisynthetic moenomycin A derivative.6 However, 2 was isolated as a main product from Streptomyces sp. K04-0144. More than 25 members of the moenomycin family have been reported from microbial metabolites.8 However, the structures of only a few members have been fully elucidated.5

From the structure–activity relationship of moenomycin derivatives reported previously, it has been found that moenomycin trisaccharides containing units C, E, F, G, H and I are the smallest cores to show antibiotic activity in vivo,6 whereas moenomycin disaccharides containing units E–I still function as transglycosylase inhibitors in vitro.9 Nosokomycins are larger than the smallest cores of in vivo-active moenomycin derivatives and are also smaller than moenomycin A itself. Therefore, it is reasonable that nosokomycins retain activity.

Recently, Walker and coworkers10 proposed a biosynthetic pathway for the cyclopentenone (A ring) moiety and the pentasaccharide section of moenomycin A using a whole-genome scanning approach with the producing strain Streptomyces ghanaensis ATCC 146723 in combination with gene knockout and complementation experiments.10, 11 They described that MoeA4 functions as an acyl-CoA ligase that cyclizes aminolevulinate to form the cyclopentenone (A ring) moiety and that MoeB4 functions as an amide synthetase that couples the A ring moiety to the C-6 of the B ring of moenomycin A. As described above, all nosokomycins lack the chromophoric cyclopentenone moiety, and no members of the moenomycin family were detected in the culture broth of the nosokomycin-producing strain, Streptomyces sp. K04-0144 strain. Therefore, we speculate that moeA4 and/or moeB4 genes do/does not work or lack(s) the nosokomycin-producing strain.

Methods

General experiments

NMR spectra were measured on a Varian XL-400 spectrometer (Varian, Palo Alto, CA, USA) with 1H NMR measured at 400 MHz and 13C NMR measured at 100 MHz in methanol-d4. Chemical shifts are expressed in δ values (p.p.m.) with methanol-d4 (δc 49.0) used as an internal reference for 13C NMR spectra and methanol-d4 (δH 3.30) used an internal reference for 1H NMR spectra. IR spectra were measured on a Horiba FT IR-710 spectrometer (Horiba, Kyoto, Japan), and UV spectra were measured on a Beckman DU-640 spectrophotometer (Beckman Coulter, Fullerton, CA, USA). Optical rotation was measured on a JASCO DIP-370 digital polarimeter (JASCO, Hachioji, Japan).

LC-ESI-MS experiments

LC-ESI-MS spectra were measured on a JEOL JMS-700 magnetic sector-type mass spectrometer (JEOL, Akishima, Japan) coupled with an Agilent 1100 G1310A liquid chromatography pump (Agilent Technologies, Santa Clara, CA, USA).12 The LC conditions were as follows: column, PEGASIL ODS (2.0ø × 50 mm, Senshu Scientific, Tokyo, Japan); mobile phase, MeOH; and flow rate, 0.2 ml min−1. The ESI source was operated in negative ion mode with a ring voltage of 100 V. The standard substance YOKUDELNA (JEOL) was used for mass calibration with a scan range of m/z 100–2000.

31P NMR and 1H-31P HMBC experiments

31P NMR spectra were measured on a Mercury-300 spectrometer (Varian) at 300 MHz in methanol-d4 at room temperature. Chemical shifts are expressed in δ values (p.p.m.), and triphenylphospine (δP 0.0) was used as an internal reference. 1H-31P HMBC spectra were measured under following conditions: sample 10 mg, f1 × f2=2048 × 512 points, nt=16, ni=160. The pulse sequence was described in a previous study.13

1H-15N HMBC experiments

15N NMR spectra were measured on a Varian Mercury-300 spectrometer at 300 MHz in methanol-d4 at room temperature. Chemical shifts are expressed in δ values (p.p.m.) and benzamide (δN 105.4) was used as an internal reference. 1H-15N HMBC spectra were measured under the following conditions: sample 10 mg, f1 × f2=2048 × 2048 points, nt=4000, ni=90. The pulse sequence was described in a previous paper.14