Two new 2-alkylquinolones, inhibitory to the fish skin ulcer pathogen Tenacibaculum maritimum, produced by a rhizobacterium of the genus Burkholderia sp.

Exploration of rhizobacteria of the genus Burkholderia as an under-tapped resource of bioactive molecules resulted in the isolation of two new antimicrobial 2-alkyl-4-quinolones. (E)-2-(Hept-2-en-1-yl)quinolin-4(1H)-one (1) and (E)-2-(non-2-en-1-yl)quinolin-4(1H)-one (3) were isolated from the culture broth of strain MBAF1239 together with four known alkylquinolones (2 and 4–6), pyrrolnitrin (7), and BN-227 (8). The structures of 1 and 3 were unambiguously characterized using NMR spectroscopy and mass spectrometry. Compounds 1–8 inhibited the growth of the marine bacterium Tenacibaculum maritimum, an etiological agent of skin ulcers in marine fish, offering new opportunities to develop antibacterial drugs for fish farming.

that Burkholderia produce many more secondary metabolites than reported, as this group was previously classified into the genus Pseudomonas [8]. In fact, the high capacity of Burkholderia in secondary metabolism is demonstrated by the presence of unique functionalities, such as monocyclic 3-pyrazolone [9], α-aminoacrylonitrile, and thioimidazolinone [10], all of which are not preceded in metabolites from other taxa.
As part of our program to further explore this unique pharmacological resource, rhizobacteria of the genus Burkholderia were collected and tested for the production of antimicrobial metabolites against a panel of plant and animal pathogens consisting of 4 bacteria, 1 yeast, and 4 fungi. The result of this screening prompted the detailed chemical study of a strain coded as MBAF1239, which resulted in the isolation of eight antibacterial metabolites, including the two new 2-alkylquinolones 1 and 3 ( Figure 1).
Strain MBAF1239 was seed-cultured in V22 medium and subsequently transferred into IMM-HS medium, which was designed for metabolite production based on the composition of IMM [11] and HS media [12]. The fermented culture was extracted with 1-BuOH and the extract was fractionated by solvent-partitioning to yield n-hexane-, 90% aqueous MeOH-, and 60% aqueous MeOH-soluble fractions. Antimicrobial testings against Rhizopus oryzae (the fungal pathogen of rice seedling blight), Trichophyton rubrum (dermatophytosis pathogen), and Tenacibaculum maritimum (the causative organism for skin ulcers in marine fish) revealed that the second fraction was the most active. The second fraction was then further fractionated by ODS flash chromatography and purified by HPLC to yield the eight metabolites 1-8.
The molecular formula of 3 was established as C 18  , and a methyl (δ H /δ C 0.89/14.1) group, leaving one carbonyl (δ C 179.0) and three aromatic resonances (δ C 149.9, 139.4, and 125.3) as quaternary carbons (Table 1). Because these structural elements accounted for six out of eight degrees of unsaturation, the remaining two degrees correspond to two rings, which constitutes a fused bicyclic structure as suggested by the number of available aromatic carbons (eleven). A 4-quinolone substructure was indicated by a peaksplitting at the 340-320 nm region in the UV spectrum (328 and 322 nm) [13]. Indeed, 1 H NMR resonances at the down field region was superimposable on those of known 2-heptyl-4(1H)quinolone (2, Supporting Information File 1, Figure S11). The COSY and HMBC correlations also supported this assignment ( Figure 2; Supporting Information File 1, Figures S8 and S10). An extension of a 2-nonenyl group (C9-C17) at C2 was supported by an HMBC correlation from H3 to C9 ( Figure 2).
The obtuse 1 H NMR signal shapes, presumably due to limited solubility of 3 in CDCl 3 , hampered unequivocal determination of the C10-geometry based on a coupling constant between the olefinic protons H10 and H11. Instead, a lack of NOESY correlation between these protons was indicative of an E-geometry (Supporting Information File 1, Figure S12). This was finally supported by a chemical shift value for the C12 allylic carbon at δ 32.5, which is closer to that of an E-isomer, burkholone (δ 32.5) [14], than that of an Z-isomer, haplacutine F (δ 27.7) Figure 3: Referential 13 C chemical shifts of an allylic carbon in burkholone [14] and haplacutine F [15].  [15] ( Figure 3). Thus, the structure of 3 was concluded to be (E)-2-(non-2-en-1-yl)quinolin-4(1H)-one.
The molecular ions of 1 were observed at m/z 242 and m/z 240 in the positive and negative modes, respectively, revealing a 28 Da smaller molecular weight for 1 relative to 3. The 1 H NMR spectra of both compounds (Supporting Information File 1, Figures S6 and S1) are similar except for the integral of the methylene resonances between 1.48 and 1.23 ppm. While the resonance amounted to 8H-equivalent in 3, the resonance in 1 was equivalent to 4H, suggesting that 1 is a congener of 3 with a two-methylene shorter appendage. This was later confirmed by the interpretation of COSY, HSQC, and HMBC spectra (Supporting Information File 1, Figures S3-S5), allowing the full assignment of 1 H and 13 C NMR resonances (see Experimental). Thus, 1 was determined to be (E)-2-(hept-2-en-1yl)quinolin-4(1H)-one.
A recent metabolomic analysis using LC-MS verified the presence of more than 50 2-alkyl-4-quinolones [35], most of which remain chemically and biologically uncharacterized. Compounds 1 and 3 are among these uncharacterized analogs, with their (presumable) detection by mass spectrometry reported twice [29,36] and three times [36][37][38], respectively. In this study, we have isolated both compounds for the first time, which enabled rigorous structure characterization, including the position and geometry of unsaturation in the side chains, as well as evaluation of their bioactivity (see below).
Compounds 1-8 at 10 μg/disc inhibited the growth of a bacterium of the phylum Bacteroidetes, T. maritimum (Table 2). Overall, alkylquinolones 1-6 were more potent than 7 and 8. Among 1-6, the 2-heptenyl-3-methyl congener 4 was the most active. Compounds 3-6 also inhibited the growth of the fungi R. oryzae and T. rubrum, while 1 and 2 did not. This may to some extent attributable to the global lipophilicity of molecules, as 1and 2 are among the fastest eluting congeners during the reversed-phase separation. Because T. maritimum is one of the major etiologies for fatal skin ulcers in marine fish [39], 1-8 could offer novel scaffolds to develop new therapeutic modalities for this economically devastating epizootic.

Fermentation, extraction, and isolation
Burkholderia sp. MBAF1239 was seed-cultured in 500 mL K-1 flasks each containing 100 mL of medium For the extraction of secondary metabolites, 100 mL of 1-butanol was added to each flask, and they were allowed to shake for 1 h. The resulting suspension was centrifuged at 6000 rpm for 10 min to separate organic and aqueous layers, the former of which was concentrated in vacuo to give a 5.35 g extract from a 2 L culture. The crude extract was successively partitioned between 60% MeOH (250 mL) and CH 2 Cl 2 (250 mL × 3) and the latter between 90% aqueous MeOH (150 mL) and n-hexane (150 mL × 3

Evaluation of antimicrobial activity
The antibacterial and antifungal activity of 1-8 was evaluated by a paper-disc agar diffusion method described in our previous study [40]. Flexibacter maritimus medium (0.5% peptone and 0.05% yeast extract in sea water) solidified with 10% agar was used to test against T. maritimum.

Supporting Information
Supporting Information File 1 1 H and 13 C NMR, COSY, HSQC, and HMBC spectra for compounds 1 and 3.