|Home | About | Journals | Submit | Contact Us | Français|
The epiphyte Pantoea agglomerans 48b/90, which has been isolated from soybean leaves, belongs to the Enterobacteriaceae, as does the plant pathogen Erwinia amylovora, which causes fire blight on rosaceous plants such as apples and leads to severe economic losses. Since P. agglomerans efficiently antagonizes phytopathogenic bacteria, the P. agglomerans strain C9-1 is used as a biocontrol agent (BlightBan C9-1). Here we describe the bioassay-guided isolation of a peptide antibiotic that is highly active against the plant pathogen E. amylovora and pathovars of Pseudomonas syringae, and we elucidate its structure. Bioassay-guided fractionation using anion-exchange chromatography followed by hydrophobic interaction liquid chromatography yielded the bioactive, highly polar antibiotic. The compound was identified as 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine by using high-resolution electrospray ionization mass spectrometry and nuclear magnetic resonance techniques. This peptide was found to be produced by three of the nine P. agglomerans strains analyzed. Notably, the biocontrol strain P. agglomerans C9-1 also produces 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine. Previously, 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine has been characterized only from Serratia plymuthica. 2-Amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine has been shown to inhibit the growth of the human pathogen Candida albicans efficiently, but its involvement in the defense of epiphytes against phytopathogenic bacteria has not been investigated so far.
Microbial pathogens pose a major threat to many plants and can cause enormous losses in agriculture. Microorganisms that antagonize pathogens can offer a way to fight plant diseases that is more environmentally friendly than chemical treatment. Such diseases include fire blight, which is caused by Erwinia amylovora and affects many rosaceous plants, e.g., apple and pear (18, 25, 29, 38).
Suitable strains for biocontrol agents are often plant-associated microorganisms that are forced to defend their ecological niches under natural conditions and are thus adapted to competition with plant pathogens (2, 3). The species Pantoea agglomerans (formerly Erwinia herbicola) comprises many strains that are promising sources for biocontrol agents (8, 15, 30, 32, 43). P. agglomerans strains are ubiquitous in nature, inhabiting plant surfaces, water, soil, animals, and humans (9, 11). Several Pantoea isolates are known to inhibit E. amylovora efficiently in planta (39, 42). In vitro experiments have revealed some antibiotics from P. agglomerans and uncovered how they act against E. amylovora (22, 43). The known antibiotics produced by P. agglomerans strains, which belong to diverse chemical classes and affect different molecular targets, exhibit both narrow- and broad-spectrum activities (21).
For example, P. agglomerans Eh318, isolated from apple leaves, produces two peptide antibiotics, pantocin A and pantocin B; both interfere with amino acid biosynthesis. Pantocin A blocks l-histidinol phosphate aminotransferase (20), and pantocin B acts as an N-acetylornithine transaminase inhibitor (5). Consequently, their inhibitory effects can be compensated for by supplementation with l-histidine and l-arginine, respectively (43). Giddens et al. (2002) described a phenazine antibiotic and its precursors, which were produced by P. agglomerans Eh1087 (10). Andrimid, a hybrid nonribosomal peptide polyketide antibiotic from P. agglomerans Eh335, selectively blocks the carboxyl transfer reaction of prokaryotic acetyl coenzyme A carboxylase; this reaction catalyzes the first committed step of fatty acid biosynthesis (19, 26). P. agglomerans E325 sold as Bloomtime Biological (Northwest Agricultural Products, Pasco, WA) acidifies flower stigmata, thus reducing the growth of E. amylovora. Simultaneously, it produces an antibiotic that has high specificity against E. amylovora and is effective under low-phosphate and low-pH conditions (34).
P. agglomerans C9-1, which is registered as the biocontrol agent BlightBan C9-1 (Nufarm Agricultural Inc.), produces two antibiotics, herbicolin O and herbicolin I (16). Like pantocin A, herbicolin O loses its activity in the presence of histidine. However, herbicolin I does not become ineffective in the presence of amino acids (17). Although C9-1 is registered as a biocontrol agent, the chemical nature of herbicolins has remained largely unknown (13, 14).
P. agglomerans 48b/90 (Pa48b), an epiphyte from soybean leaves (40), attracted our attention because it strongly inhibits the growth of E. amylovora and Pseudomonas syringae pv. glycinea (27), the pathogen that causes the bacterial blight of soybean. Since the mode of action of Pa48b against plant pathogens, in particular E. amylovora, is elusive, we looked for the molecular basis for the biocontrol potential of Pa48b. Here we describe the isolation, structure elucidation, and bioactivity of a potent antibiotic against plant pathogens that is produced by several P. agglomerans strains. The properties of this antibiotic perfectly match those of the chemically unidentified herbicolin I from P. agglomerans C9-1 (BlightBan C9-1).
P. agglomerans strains and bacterial indicator strains were cultured and maintained on Standard I (St1) (Roth, Karlsruhe, Germany) agar plates. P. syringae strains were cultured and maintained on King's B agar plates (per liter, 20 g Bacto peptone, 10 g glycerol, 1.5 g K2HPO4, and 15 g Bacto agar) (23). P. agglomerans stains were cultivated in liquid PIPES medium for antibiotic production [solution I contained, per 900 ml, 34 g piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), 0.3 g KH2PO4, 1 g NH4Cl, and 1 g Na2SO4; the pH was adjusted to 6.8, solution II was added, containing, per 100 ml, 10 g glucose, 0.1 g MgCl2·6H2O, 0.01 g FeSO4·7H2O, and 0.01 g MnSO4·4H2O; the solutions were autoclaved separately] (36). The P. agglomerans strains used in this study are listed in Table Table1,1, and the indicator strains are listed in Tables Tables22 and and33.
The antimicrobial activities of Pa48b against different indicator organisms (see Tables Tables22 and and3)3) were evaluated using spot assays. Portions (0.4 ml) of overnight St1 broth cultures of the bacterial indicator strains, adjusted to an optical density at 578 nm (OD578) of 1, were added to 10 ml of melted 5b agar medium (12) and St1 agar medium at 50°C, and these cultures were poured into plates (diameter, 60 mm). Pa48b was spotted directly onto the agar surface, incubated at 28°C for 24 to 48 h, and analyzed. Alternatively, 100-μl suspensions of fungi, with a McFarland standard of 0.5, were inoculated into 34 ml of sterile melted agar medium and poured into petri dishes. Holes (diameter, 9 mm) were cut out and filled with 50 μl of the purified culture filtrates. As a reference compound, amphotericin B (10 μg/ml) was used. Malt extract agar (per liter, 40 g malt extract [Becton Dickinson], 4 g yeast extract [Serva], and 5 g Bacto agar), Bacto yeast morphology agar (BYM; Difco Becton Dickinson & Company, Sparks, MD), and potato dextrose agar (Merck, Darmstadt, Germany) were used.
The success of the purification steps and the response to treatments were monitored using E. amylovora Ea7 as an indicator strain in the agar plate diffusion assay. Two milliliters of the bacterial suspension (OD578 of 1) was added to 48 ml of 5b agar medium and poured into plates (diameter, 120 mm). Holes (diameter, 9 mm) were cut out and filled with 50 μl of the test samples. As a reference compound, spectinomycin (5 mg/ml) was used. Agar diffusion assays with the bacterial indicator organisms listed in Table Table33 were carried out in the same way.
In order to investigate the effects of N-acetylglucosamine and casein hydrolysate on antibiotic activity, 50 μl of either an N-acetylglucosamine solution (100 μg/μl), a casein hydrolysate solution (0.3%, wt/vol), or water (control) was added to 50 μl of a sterile culture supernatant. These samples were tested in agar diffusion assays with Erwinia amylovora Ea7 as the indicator strain as described above.
The growth curve was determined in triplicate from Pa48b cultures grown in 500-ml Erlenmeyer flasks containing 100 ml PIPES medium (36) and shaken at 200 rpm for 100 h at 10°C (Innova 4230 incubator shaker; New Brunswick Scientific, Edison, NJ), starting at an OD578 of 0.15. The cultures were inoculated with an overnight liquid culture of Pa48b (PIPES medium, 28°C, 200 rpm). The temperature dependence of antibiotic production by Pa48b was studied at 8, 10, 12, 18, and 28°C. The cells were grown to an OD578 of 5.5 and were then analyzed for their antibiotic activity. Statistical analysis was performed using Sigma Plot, version 9.01 (Systat Software, Inc., 2004).
In order to determine the influence of proteases on antibiotic activity, aliquots of the bioactive supernatant were incubated with proteinase K (final concentration, 20 mg/ml; Fermentas, St. Leon-Rot, Germany) for 12 h at room temperature or with type I Bacillus cereus β-lactamase (final concentrations, 2.5, 5, and 50 mg/ml; Sigma-Aldrich, Steinheim, Germany) for 1 h at 37°C. The heat stability of the antibiotic activity of Pa48b was analyzed by incubating the culture supernatant for 10 min at 60, 80, and 100°C and for 30 and 60 min at 100°C. The pH stability of the antibiotic activity of Pa48b was investigated by adjusting the culture supernatant to pH 3 or pH 10 for 20 min at room temperature. After the treatment, the pH was readjusted to 6.8, and the antibiotic activity against E. amylovora was tested in the agar diffusion assay.
Pa48b (a 1-liter culture) was grown in PIPES medium for 72 h at 10°C and was subsequently centrifuged, after which the supernatant was collected. The supernatant was extracted with ethyl acetate at a ratio of 1:1. The water phase was adjusted to pH 2 with HCl (36%) to precipitate PIPES. After filtration, the aqueous solution was concentrated in vacuo and the residue resuspended in water. The samples were directly applied to an anion-exchange column (Sephadex QAE, A-25; Pharmacia Fine Chemicals, Uppsala, Sweden), followed by washing with 150 ml water and 100 ml 0.2 M ammonium hydrogen carbonate. The bioactive fractions were obtained after elution with 100 ml of 30% acetic acid (AcOH) and testing of the fractions against E. amylovora Ea7 in the agar diffusion assay (5b agar medium). Active fractions were pooled and concentrated in vacuo and were subjected to high-pressure liquid chromatography (HPLC) using hydrophobic interaction liquid chromatography (HILIC) conditions (6). A Phenomenex Luna NH2 HPLC column (250 mm by 2 mm; inner diameter, 5 μm; Phenomenex, Aschaffenburg, Germany) was used with programmed gradient elution: 3 min at 0% B, with a gradient to 100% B in 27 min at a flow rate of 0.2 ml/min (where A is acetonitrile-0.1% AcOH and B is H2O-0.1% AcOH).
HPLC was performed using an HP1100 HPLC system (Agilent, Waldbronn, Germany) connected either to an LTQ electrospray ionization mass spectrometer (ESI-MS) (Thermo Fisher, Egelsbach, Germany) or to a Gilson 207 fraction collector. All fractions (fraction size, 0.5 min) were subjected to the agar diffusion assay in order to detect compounds active against E. amylovora. In order to collect enough sample for nuclear magnetic resonance (NMR) spectroscopy and biological testing, separation was performed repeatedly using a semipreparative HPLC column (Nucleodur NH2; 250 mm by 10 mm; inner diameter, 5 μm; Macherey Nagel, Düren, Germany) with programmed gradient elution (3 min at 0% B, with a gradient to 100% B in 27 min at a flow rate of 3 ml/min [where A is methyl cyanide-0.1% AcOH and B is H2O-0.1% AcOH]), a retention time of 16 min, and a yield of 1 to 2 mg from 1 liter of culture. High-resolution mass spectrometry (HR-MS) was performed by directly injecting the purified sample into an LTQ Orbitrap MS (Thermo Fisher, Egelsbach, Germany) by using a syringe pump.
The retention time (with a Phenomenex Luna NH2 column) was 21.7 min. HR-ESI-MS: [M+H]+ measured: 317.14514, calculated: 317.14556 (C12H21N4O6); ESI-MS1 of 317 ([M+H]+): measured: 299.13580 (100), calculated: 299.13500 (C12H19N4O5); measured: 271.13926 (2), calculated: 271.13929 (C12H17N4O4); ESI-MS2 of 299 ([M+H-H2O]+): 281 (4), 270 (3), 253 (100), 236 (7), 227 (2), 210 (2), 200 (2), 168 (1); ESI-MS3 of 253 ([M+H-HCOOH]+): 236 (100), 224 (15), 210 (22), 208 (18), 196 (12), 181 (13), 168 (17), 154 (24), 137 (14), 125 (5), 111 (26), 97 (3).
NMR spectra were recorded using a Bruker DMX 600 spectrometer (Bruker, Rheinstetten, Germany) fitted with a TXI cryoprobe. For calibration of the 1H NMR spectra, the solvent signal (D2O δ = 4.65) was used. 1H NMR (500 MHz, D2O, 300K): δ 0.90 (d, J = 6.90, 3H, H3C-4/5), 0.93 (d, J = 6.90, 3H, H3C-4/5), 1.8 to 1.9 (m, J = 1.8, 1H, CH-3), 3.68 (d, J = 1.8, 1H, CH-2"), 3.69 (d, J = 1.8, 1H, CH-3"), 3.74 (dd, J = 5.95 J = 14.70, 1H, H2C-3′), 3.82 (dd, J = 5.18 J = 14.73, 1H, H2C-3′), 4.06 (d, J = 5.70, 1H, CH-2), 4.23 (t, J = 5.53, 1H, HC-2′). 13C NMR (150 MHz, D2O, 300K): δ 18.7 (C-4/5), 20.1 (C-4/5), 31.5 (C-3), 41.1 (C-3′), 54.1 (C-2′), 54.8 (C-2"), 55.0 (C-3"), 62.9 (C-2), 168.5, 171.1, 172.3, 179.4.
In order to test for the production of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine, nine P. agglomerans strains (Table (Table1)1) were grown in 50 ml PIPES medium. One milliliter of each culture was withdrawn and centrifuged to pellet the cells, and 1 to 5 μl was analyzed by LC-ESI-MS as described above. In addition, 40 ml of the supernatants was concentrated and used for the agar diffusion assay against E. amylovora.
In the agar diffusion assay with 5b agar, Pa48b inhibited the growth of a large variety of gram-negative bacteria and several gram-positive bacteria (Table (Table22).
Changing from the synthetic medium (5b agar) to a complex medium (St1 agar) resulted in a loss of the inhibitory effect against most of the microorganisms tested (Table (Table2),2), suggesting that one or more medium constituents compensated for the antibacterial activity of Pa48b. Only the growth of Staphylococcus epidermidis was inhibited when the complex St1 agar medium was used. The inhibitory effects were not abolished by supplementing the medium with any of the 20 proteinogenic amino acids; with citrulline, homoserine, isoleucine, or ornithine (40); or with casein hydrolysate (0.15%) (data not shown). None of the fungi tested (Aspergillus nidulans, Alternaria sp., Botrytis cinerea, Geotrichum candidum, Fusarium solani, Mucor mucedo, Penicillium claviforme) except Candida albicans, which was very sensitive to Pa48b, were affected by Pa48b in agar diffusion assays with potato dextrose agar (data not shown). Cytopathic effect inhibition assays (35) showed that Pa48b had no antiviral effect on influenza A virus/Hongkong/1/68 and also failed to exhibit any cytotoxicity against Madin-Darby canine kidney cells (data not shown).
The antibiotic production of Pa48b depends greatly on the cultivation temperature. Surprisingly, the highest rate of production was reached at 8°C to 12°C and was about 34-fold the rate of production at 28°C (Fig. (Fig.1).1). Interestingly, in contrast to antibiotic production, the population size of Pa48b was not significantly influenced by the different temperatures (10°C, 18°C, 28°C) (data not shown). The growth curve and the antibiotic production of Pa48b in a shaken culture at 10°C are shown in Fig. Fig.2.2. The production of the antibiotic is strictly growth associated, starting at the early-exponential growth phase. The highest antibiotic concentration was detected when the strain reached the stationary-growth phase; thereafter, the antibiotic concentration remained stable.
The antibiotic activity of Pa48b was not inactivated by proteinase K or β-lactamase. The antibiotic isolated from Pa48b was stable when treated with heat for 10 min at 80°C. Even after 30 min of exposure to 100°C, some antibiotic activity remained (data not shown). The antibiotic activity was also stable at extreme pH values (pH 3 and 10) (data not shown).
In order to characterize the antibiotic compound that is responsible for the observed inhibition of the growth of E. amylovora, the culture supernatant was subjected to bioassay-guided fractionation. Since the antibiotic turned out to be highly polar, it could not be extracted with organic solvents. Therefore, the supernatant was freeze-dried and then subjected to anion-exchange chromatography. The combined active fractions of the highly polar antibiotic were further purified by HILIC using an aminopropyl column (6). In LC-MS, the bioactive fraction exhibited a [M+H]+ ion of 317 (Fig. (Fig.3A).3A). HR-MS of the purified compound revealed its molecular composition to be C12H20N4O6, a compound with 5 double-bond units. The high oxygen and nitrogen contents of the compound and its chromatographic behavior pointed to a peptide-like structure. Tandem MS (MS-MS) analysis showed the loss of H2O and of HCOOH. The MS3 of the [M+H-H2O-HCOOH]+ ion gave a highly characteristic fragmentation pattern (Fig. (Fig.3A).3A). In order to elucidate the structure of the suspected peptide, the compound was purified from 1 liter of culture for detailed NMR analysis. In the 1H NMR (Fig. (Fig.3B)3B) at 0.90 and 0.93 ppm, two doublets with an integral of three were observed; these corresponded to two methyl groups [C-4 and C-5 of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine], which had a methine group at 1.8 to 1.9 ppm (C-3) as their neighbor. The doublet signal at 4.09 ppm corresponded to an amino methine moiety (C-2) that showed a cross-signal in the correlation spectroscopy (COSY) to C-3. From these data, valine was identified as a structural element of the antibiotic. A 2,3-diaminopropionic acid moiety was deduced for the signals at 3.74 (C-3′), 3.82 (C-3′), and 4.23 ppm (C-2′), each corresponding to 1 proton, which showed correlations in the COSY. The two signals at 3.68 and 3.69 exhibited a small coupling constant of J = 1.8 Hz, which is characteristic for an epoxide moiety (C-2" and C-3"). The 13C NMR and attached proton test NMR spectra (Fig. (Fig.3C)3C) showed four carbonyl signals at 168.5, 171.1, 172.3, and 179.4 ppm, which account for one carboxylic acid moiety and three amide groups, respectively. At 18.7 (C-4/5) and 20.1 (C-4/5) ppm, both methyl groups were observed, and at 31.5 ppm, the β-CH moiety (C-3) of the valine was observed. The signals at 54.1 and 62.9 ppm correspond to two amino methine moieties: C-2′ of the 2,3-diaminopropionic acid and C-2 of valine. At 41.1 ppm, a signal for an amino methylene carbon, C-3′ of the 2,3-diamino propionic acid, appeared. The remaining two signals at 54.8 and 55.0 are characteristic for CH carbons of the epoxide moiety (C-2" and C-3") of the epoxyfumaric acid amide. By following the connectivity using H,H-COSY, heteronuclear multiple-bond correlation, and heteronuclear single-quantum coherence spectra, the structure of the antibiotic was identified as 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine (Fig. (Fig.3D).3D). Since 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine has been characterized from Serratia plymuthica (37), it was possible to compare our proposed structure with the data in the literature, and they matched perfectly.
After identifying the bioactive compound from Pa48b, we analyzed seven additional P. agglomerans isolates and P. agglomerans C9-1 (biocontrol agent BlightBan C9-1) for their potential abilities to produce 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine. Like Pa48b, C9-1 and 39b/90 also produced 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine (data not shown).
A secondary bioactivity screening (Table (Table3)3) with pure 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine led to results similar to those of the primary screening with the extract of Pa48b (Table (Table2).2). 2-Amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine exhibits broad antimicrobial potential and inhibited Erwinia amylovora, Agrobacterium tumefaciens, Escherichia coli, several Pseudomonas syringae pathovars, Serratia marcescens, Bacillus subtilis, different Candida albicans strains, and the yeast Yarrowia lipolytica (Table (Table33).
Because of the epiphytic origin of P. agglomerans and its use as a biocontrol organism against plant diseases, we concentrated mainly on testing the potential of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine against bacterial phytopathogens. 2-Amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine strongly inhibited all tested E. amylovora strains that cause the economically important fire blight disease (Fig. (Fig.3E;3E; Table Table2).2). The minimal concentration of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine necessary to inhibit E. amylovora Ea7 and P. syringae pv. morsprunorum in 5b agar was less than 20 μg/ml. A minimum of 200 μg/ml of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine inhibited several plant-pathogenic P. syringae pathovars (Table (Table33).
However, 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine did not prevent the growth of either of these plant pathogens if the agar diffusion assays were performed in a complex medium such as St1 agar. Because the inhibitory effect of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine can be compensated for by adding N-acetylglucosamine to 5b medium (Fig. (Fig.4),4), it is likely that a sufficient supply of N-acetylglucosamine in the complex medium (St1) prevents Pa48b from inhibiting the growth of the microorganisms tested (Table (Table22).
In order to inhibit the growth of the human-pathogenic fungus C. albicans by 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine, concentrations as low as 1.5 μg/ml were sufficient. Thus, 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine was active against C. albicans at a level comparable to that of the established antifungal amphotericin B. However, after 4 days of incubation, a few presumably resistant or adapted colonies of C. albicans arose within the inhibition zones of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine but not within the inhibition zones of amphotericin B (data not shown).
The species P. agglomerans is known to colonize a variety of habitats, for example, plant surfaces (33). The epiphyte Pa48b has been isolated from soybean leaves and found to be well adapted to its niche (40). Besides coping with low nutrient availability, Pa48b efficiently suppresses competitors, as our in vitro screening of bacterial phytopathogens for growth inhibition revealed (Table (Table2).2). Pa48b produces an antibiotic with broad activity against the gram-negative bacteria tested and the gram-positive bacterium Bacillus subtilis (Table (Table3).3). Thus, Pa48b is a promising biocontrol agent against various microbial plant diseases.
In order to characterize the compound(s) with high activity against plant pathogens, in particular against E. amylovora, a bioassay-guided isolation approach was used. A highly polar antibiotic was obtained after anion-exchange chromatography and HILIC-HPLC purification (6). The purified antibiotic turned out to be stable at extreme pHs; in addition, it was resistant to heat and treatment with proteinase K and β-lactamase.
Using HR-ESI-MS and NMR experiments, the structure of the compound was identified as 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine; this compound has already been isolated by Shoji et al. from Serratia plymuthica CB-25 (37). In addition, there are two closely related peptides, A19009 and Sch37137, from Streptomyces collinus and Micromonospora sp., respectively (7, 28). However, neither 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine nor the related peptides A19009 and Sch37137 had been isolated from P. agglomerans or characterized as highly active against plant pathogens such as E. amylovora and P. syringae pathovars. But Shoji et al. characterized 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine as an excellent inhibitor of the human pathogen C. albicans (37), which we also observed in our biological screening.
2-Amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine formation is associated with growth. The production of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine is temperature dependent: its rate of production is optimal between 8°C and 12°C. Interestingly, not only is the rate of antibiotic production highest at low temperatures, but after a short adaptation phase, Pa48b grew as well at 10°C as at 28°C. Secondary metabolite biosynthesis at low temperatures has been observed for phytotoxins such as persicomycin (1), phaseolotoxin, or coronatin (4), which play important roles in the progression of plant diseases in different P. syringae pathovars. In the case of Pa48b, the epiphyte might benefit from its substantial growth and the high rate of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine production at low temperatures in springtime to outcompete other epiphytic and phytopathogenic microorganisms.
In contrast to the impact of many antibiotics from P. agglomerans, such as pantocins A and B (43) or herbicolin O (17), the impact of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine cannot be compensated for by supplementing the medium with amino acids or casein hydrolysate. Therefore, this compound is different from most other antibiotics from P. agglomerans strains. More than 90% of the 90 strains tested did not inhibit E. amylovora in the presence of different amino acids (41). Still, the growth-inhibiting effect of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine is compensated for if E. amylovora is grown on agar plates containing complex medium. In this respect, it was interesting to see how 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine acted. Since the related peptides Sch37137 and A19009 from Micromonospora sp. and Streptomyces collinus, respectively, inhibit the glucosamine-6-phosphate synthase (GlmS; EC 126.96.36.199) (7, 28), we suspected a similar mode of action for 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine. GlmS catalyzes the formation of glucosamine-6-phosphate (31), which is a key substrate of bacterial peptidoglycan and fungal chitin (24).
Supplementation of our synthetic medium with N-acetylglucosamine caused 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine to become inactive against E. amylovora, suggesting that 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine acts as a GlmS inhibitor, as do the peptides Sch37137 and A19009 (Fig. (Fig.4).4). Thus, as for other antibiotics from P. agglomerans strains, the ability of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine to antagonize phytopathogens such as E. amylovora depends strongly on nutrient availability.
In order to study the distribution of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine among P. agglomerans strains, we analyzed seven additional selected Pantoea isolates (all from Dornburg, Germany [Table [Table1])1]) and the P. agglomerans strain C9-1. The active compound was found in the culture supernatants of Pa48b, Pa39b, and the biocontrol strain P. agglomerans C9-1 (BlightBan). Comparing the properties of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine with the information available (molecular formula, chemical and biological properties) for the antimicrobial herbicolin I from P. agglomerans C9-1, we conclude that the two compounds are identical (16, 17). In order to characterize the role of each herbicolin, Ishimaru et al. (1988) used E. amylovora strains resistant to herbicolin O and herbicolin I in immature pear fruit assays with P. agglomerans C9-1 (17). Herbicolin O-resistant strains were sensitive only to herbicolin I. Disease symptoms were suppressed half as efficiently as those due to wild-type strains when herbicolin I-resistant strains were used in these assays. Therefore, herbicolin I was found to have a strong impact on the biological control of fire blight disease. A comparison of our agar diffusion assays with the assays used for E. amylovora and Pseudomonas syringae pv. glycinea suggests that 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine is more important in suppressing E. amylovora than in suppressing P. syringae pathovars (Table (Table33).
In summary, the epiphyte Pa48b produces 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine, which is highly active against bacterial plant pathogens, in particular E. amylovora, in vitro. To date, 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine has been known only from S. plymuthica CB-25 and not from Pantoea agglomerans strains. The antibiotic herbicolin I from P. agglomerans C9-1, the structure of which has not been elucidated beyond its molecular formula (16), appears to be identical to 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine. Future in planta experiments will reveal the potential of 2-amino-3-(oxirane-2,3-dicarboxamido)-propanoyl-valine as an agent against plant diseases and will allow us to understand its role in the antagonism of Pa48b against plant pathogens.
We thank Marc Lamshöft for HR-ESI-MS measurements.
U.F.S. is grateful for a Ph.D. fellowship from the Deutsche Bundesstiftung Umwelt. D.S. and P.S. gratefully acknowledge funding from the Deutsche Forschungsgemeinschaft (Emmy Noether fellowships SP 1106/3-1 and PS 718/1-2) and financial support from the Verband der Chemischen Industrie.
Published ahead of print on 9 October 2009.