INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pantoea agglomerans or Pantoea dispersa (20), also known as Erwinia herbicola (Löhnis) Dye are members of the Enterobacteriaceae and are ubiquitous in nature, inhabiting plants, soil, and water (16, 20, 21) and animals and humans (16, 35). Strains belonging to E. herbicola are members of the E. herbicola-Enterobacter agglomerans cluster; some have been redesignated P. agglomerans and P. dispersa, while others did not fall into either of the two species (20). P. agglomerans and P. dispersa are frequent companions of Erwinia amylovora (Burr.) Winslow et al. the causal agent of the disease fire blight of apple and pear trees (36, 38). There is current interest in P. agglomerans and P. dispersa as biological control agents for fire blight because they are harmless to apple and pear trees and are able to protect them against invasion of the pathogen (4, 29). P. agglomerans strain Eh318, isolated from a symptomless apple stem in New York State, protected immature pear fruits in the laboratory (53) and apple blossoms in controlled environment and orchard tests (5, 23, 43).

Production of antibiotics inhibitory to E. amylovora by several strains of Pantoea spp. seems important for inhibition of E. amylovora in planta (30, 45, 53). In vitro inhibition of E. amylovora by antibiotics of Pantoea spp. is well documented (24, 28, 45, 47, 48). Different strains of P. agglomerans and P. dispersa have different spectra of antimicrobial activity (15, 25) and produce different types of inhibition zones against the same indicator organism (3); both observations presumably reflect the fact that different antibiotics are produced by different strains. One strain of P. agglomerans, strain C9-1, produces three different antibiotics, which were purified and characterized preliminarily (27). The presence of an inner and an outer zone of inhibition in an E. amylovora 110-seeded agar overlay led Ishimaru and coworkers to suggest that P. agglomerans C9-1 produced more than one antibiotic (28). P. agglomerans Eh318 also forms a double halo in an overlay seeded with E. amylovora strain Ea273 (Fig. 1), which we hypothesized was due to the action of two antibiotics (S. Wright-Dobrzeniecka and S. V. Beer, Abstr. 93rd Gen. Meet. Am. Soc. Microbiol. 1993, abstr. Q420, 1993). One of these, pantocin B has been characterized chemically as (R)-N-[((S)-2-amino-propanoylamino)-methyl]-2-methanesulfonyl-succinamic acid (10); it inhibits N-acetylornithine transaminase through competitive binding with N-acetylornithine, thus interfering with the last step in the arginine biosynthetic pathway (10, 49). Antibiotics of Pantoea species frequently are grouped on the basis of the type of amino acid that, when added to the overlay, renders E. amylovora insensitive to them. Most strains of P. agglomerans and P. dispersa produce histidine-reversible or histidine- and/or leucine-reversible antibiotics (14, 50). The antibiosis of E. amylovora by P. agglomerans Eh318 is abolished in the presence of a combination of histidine and arginine, but not by either amino acid alone (48, 49).

View larger version (126K): [in this window] [in a new window]   FIG. 1.   P. agglomerans Eh318 produces a double zone of inhibition against E. amylovora Ea273 in a chloroform assay.

We have demonstrated that two distinct cosmids, pCPP702 and pCPP704, containing inserts of Eh318 DNA bestow on E. coli the ability to produce two distinct antibiotics inhibitory to E. amylovora. The observed requirement for two amino acids to abolish antibiosis by Eh318 is a consequence of the two antibiotics. They were named pantocins after the genus name of the producing organism. Histidine reversed the activity of pantocin A, and arginine reversed that of pantocin B. The distinctive antibiotic phenotypes of defined marker-exchange mutants of Eh318 that are defective in production of pantocin A and/or B provide clear genetic evidence for the production of these two antibiotics by P. agglomerans.

(Brief reports on these findings were made previously at scientific conferences [56; Wright-Dobrzeniecka and Beer, Abstr. 93rd Gen. Meet. Am. Soc. Microbiol 1993].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, media, and growth conditions. The bacterial strains and plasmids used in this study are listed in Table 1. All strains were cultured routinely in Luria-Bertani (LB) medium (39). The reaction of strains Eh318 and Eh252 (Table 1) in API 20E (bioMérieux Vitek, Inc., Hazelwood, Mo.) was consistent with their identification as species of Pantoea. The results of a GN2 Microlog test (Biolog, Inc., Hayward, Calif.) identified them with highest probability as belonging to P. agglomerans. E. amylovora and Pantoea spp. were cultured at 28°C, and Escherichia coli was cultured at 37°C. The following antibiotics were used at the concentrations indicated: ampicillin, 100 µg/ml; chloramphenicol, 20 µg/ml; kanamycin, 50 µg/ml; nalidixic acid, 20 µg/ml; rifampin, 25 µg/ml; spectinomycin, 50 µg/ml; tetracycline, 10 µg/ml. Antibiotic production assays were done on minimal media, either glucose-asparagine (GA) medium (52, 53) or E. coli minimal medium (EcMM), which contained per liter: 0.25 g of yeast extract (Difco Laboratories, Detroit, Mich.), 20 ml of glycerol, 4.0 g of K₂HPO₄, 1.72 g of KH₂PO₄, 0.5 g of NaCl, 2.0 g of (NH₄)₂SO₄, 0.2 g of sodium citrate, and 0.02 g of MgSO₄ · 7H₂O. Thiamine was added to GA medium and EcMM at 0.1 µg/ml for the growth of E. coli strains DH5 and JM109.

Antibiotic production assays. Antibiotic production was assayed by two methods, the live assay and the chloroform assay. In both assays, a basal layer of either GA medium or EcMM was covered with a soft-agar overlay which consisted of 0.8 ml 5× GA salts stock solution, 3.2 ml of 0.7% agar (cooled to 48 to 50°C), and 0.3 ml of bacterial indicator cells, grown to an optical density at 620 nm of 0.4, spun down, and resuspended in an equal volume of 5 mM potassium phosphate buffer, pH 6.5. Overlays seeded with E. coli DH5 were amended with thiamine to a final concentration of 0.1 µg/ml.

Gene transfer methods. The first spot-agar conjugation technique described by Steinberger and Beer (41) was employed, with the modification of 12 h of incubation of the spot and subsequent resuspension in 0.5 ml of sterile water before spreading on selective medium. The cosmids pCPP702 and pCPP704 were thus transferred from E. coli JM109 to E. coli DH5, and the pUT::mini-Tn5Cm plasmid was transferred from E. coli SM10pir to E. coli CGSC6151(pCPP719). Mobilization of the two cosmids between JM109 and DH5 required the presence of the helper E. coli HB101(pRK2013). The nalidixic acid resistance of DH5 was employed to select for the transconjugants after the matings. Routine transformations of E. coli strains with plasmids followed the procedure of V. Simanis as described by Hanahan et al. (22).

DNA isolation and manipulations. For the construction of a genomic library, the DNA from plasmid pCPP9 (2) was isolated by the large-scale alkaline lysis procedure (39) and purified further by cesium chloride-ethidium bromide equilibrium centrifugation (39). Isolation of total genomic DNA of Eh318 and its mutants, whether for library construction or to run on gels for Southern blotting, followed the procedure of Silhavy and coworkers (40). Cosmid DNA was isolated by a medium-scale alkaline plasmid preparation procedure that combined the methods described by Marko et al. (33) and Zasloff et al. (57). Plasmid DNA was routinely isolated on a small scale by an alkaline miniprep extraction procedure (7).

Construction of a genomic library. Cloning of genomic DNA fragments of Eh318 of approximately 40 kb into the cosmid vector pCPP9 (2) followed the procedure of Ish-Horowicz and Burke (26), with the inclusion of a DNA sizing step on a 10 to 40% sucrose gradient (39). Fractions containing fragments of 32 to 47 kb from a partial Sau3AI digest were dephosphorylated with calf intestinal alkaline phosphatase. The vector pCPP9 was digested separately with either EcoRI or SalI, further digested with BamHI, and ligated with the genomic Sau3AI fragments. Recombinant cosmids were packaged in vitro with the Gigapack packaging kit (Stratagene, La Jolla, Calif.), and transduced into E. coli JM109. Transductants were selected on the basis of their spectinomycin resistance.

Screening the genomic library for antibiotic-producing transductants. Transductants were screened for antibiotic production in the live assay using an overlay seeded with Ea273. The cosmid DNA of colonies that produced antibiotics was extracted, digested with several restriction enzymes, and electrophoresed to visualize restriction enzyme profiles. Antibiosis toward Ea273 in the presence and absence of arginine and histidine was evaluated by a modified chloroform assay in which the producer was allowed to grow for 2 days on GA medium before it was removed and the plate was exposed to chloroform vapors. DH5 carrying two distinct cosmids that conferred antibiosis toward Ea273 was assayed for activity against a number of different bacteria (see Table 2). The marker exchange mutants of Eh318 that were deficient in pantocin A and/or B synthesis (see below) were included to confirm that the activities of pantocin A and B were consistent in different genetic backgrounds. The results were recorded qualitatively (absence or presence of zones of inhibition) rather than quantitatively, since both the live and the chloroform tests were used. The sensitivities of Eh252 and Eh318 (control) to pantocin A and pantocin B produced by DH5(pCPP1051) and DH5(pCPP719) were tested separately. A colony of Ea273 that appeared in a zone of inhibition produced by Eh421 (deficient in pantocin A synthesis [PanA]) was propagated several times on fresh plates where Eh421 had grown, in order to select for maintenance of the antibiotic resistance phenotype. The strain Ea273R421 was subsequently seeded in an overlay that was poured over plates containing antibiotics produced by Eh421 (PanA) and Eh318.

Construction of smaller, antibiotic-encoding clones. DNA from cosmid pCPP702 and pBluescript KS(+) was digested with EcoRI, ligated, and added to competent DH5 cells. White colonies on LB agar amended with ampicillin and X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) were screened for antibiotic production in a live assay with GA medium or EcMM as the basal medium. The cloned DNA present in colonies that produced zones of inhibition against Ea273 was characterized. A smaller, antibiotic-conferring clone was designated pCPP1051. Similarly, a subclone of pCPP704 was constructed by digesting the cosmid DNA with BamHI and religating. A 32.7-kb BamHI fragment that also included the cosmid vector pCPP9 was designated pCPP719.

Transposon mutagenesis of pCPP1051 and pCPP719. The two subclones, pCPP1051 and pCPP719, were mutagenized with transposons that carried distinct antibiotic resistances (kanamycin and chloramphenicol, respectively) to allow for the construction and selection of a double marker exchange mutant of Eh318 that lacked the ability to produce either pantocin. Tn5 insertion mutagenesis of pCPP1051 was performed essentially as described by de Bruijn and Lupski (12). E. coli CC118 was the host for pCPP1051 and the ::Tn5 phage used was b221 rex::Tn5 cI857 (6). Phage stocks were propagated in strain LE392 as described (12). E. coli DH5 was transformed with plasmid DNA that had been extracted from CC118(pCPP1051) infected with ::Tn5 and plated. Resulting colonies were screened in a live assay, and insertions were mapped.

Marker exchange mutagenesis, screening, and phenotypic characterization. DNA fragments that carried transposon insertion A14 or 122 were cloned into pBR322 or pBR325, respectively, because these cloning vectors maintained themselves stably in Eh318. The new constructs were electroporated into Eh318, and cured, which allowed for marker exchange mutagenesis to occur. Successive subculturing in phosphate-limited medium, while maintaining selection for Km^r (for the Tn5 insertion) or Cm^r (for the mini-Tn5Cm insertion) (37), resulted in mutant selection. Tetracycline-sensitive colonies were screened for antibiotic production in a live assay, and colonies with reduced antibiotic production were selected for further analysis. Plasmid preparations of these confirmed the absence of the pBR325 or pBR322 constructs. The points of the insertions in putative marker exchange mutants were confirmed based on Southern blot analyses (data not shown) (55).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification and characterization of antibiotic-encoding cosmids. Five of 1,500 members of the genomic library of Eh318 in E. coli JM109 exhibited antibiosis against Ea273 on GA medium. Examination of the EcoRI and HindIII digest patterns of the five cosmids (data not shown) (55) indicated that two of the cosmids, designated pCPP702 and pCPP704, had no bands in common. The restriction enzyme patterns of two of the other cosmids (pCPP701 and pCPP703) appeared similar to those of pCPP702, whereas the pattern of the fifth cosmid (pCPP705) was distinct from all others. Figure 2 depicts the absence of common bands in pCPP702 and pCPP704 after digesting with EcoRI and a combination of EcoRI and XbaI. This initial genetic difference between these two cosmid clones was supported by detailed genetic mapping of clones, mapping of transposon insertions, and cross-hybridization tests.

View larger version (100K): [in this window] [in a new window]   FIG. 2.   Restriction enzyme patterns of cosmids pCPP704 (lanes 3, 7, and 11) and pCPP702 (lanes 4, 8, and 12) and a subclone of pCPP702, pCPP1051 (lanes 2, 6, and 10). Lane 1 has Eh318 genomic DNA; lanes 5, 9, and 13 have  digested with HindIII as molecular weight markers. The DNA was digested to completion with EcoRI and XbaI (lanes 1 through 4), EcoRV (lanes 6 through 8), and EcoRI (lanes 10 through 12). The arrows indicate the fragments that hybridized with a 3.9-kb XbaI-HindIII fragment of pCPP1051 (see Fig. 3) that covered a region considered important for synthesis of pantocin A. The composite photograph was created with Adobe Photoshop 5.5.

Construction of pCPP1051 and pCPP719. Digestion of pCPP702 with EcoRI resulted in seven fragments, whose sizes11.8, 10.8, 8.4, 4, 3, 2.3, and 1.8 kb (Fig. 2)total 42.1 kb, a number which includes the cosmid vector pCPP9 (5.3 kb). Recombinant plasmid pCPP1051 consisted of an 11.8- and an 8.4-kb EcoRI fragment from pCPP702 cloned into pBluescript. DH5(pCPP1051) inhibited Ea273 when grown on EcMM. A restriction enzyme map of the 20.2-kb insert of pCPP1051 was generated (Fig. 3). The genetic region involved in the biosynthesis of pantocin B (10) was cloned from pCPP704 into pCPP719. This clone conferred upon DH5 the ability to inhibit the growth of Ea273. A detailed restriction enzyme map of pCPP719 was generated (Fig. 4).

View larger version (9K): [in this window] [in a new window]   FIG. 3.   Physical map of the 20.2-kb insert (top) from pCPP702 in pCPP1051. The 11.8-kb EcoRI fragment containing the Tn5 insertion 122 (middle) that was cloned into pBR325 to create pCPP726. This clone was used for marker exchange of insertion 122 into Eh318 to generate Eh421 (PanAl). The positions of Tn5 insertion 122, which abolishes pantocin A production (filled lollipop), and insertion 326, which does not affect production (open lollipop), are shown at the bottom. The XbaI-HindIII fragment (solid gray bar, below map) was radioactively labeled for use as a probe in Southern blots. Abbreviations: A, ApaI; B, BamHI; C, ClaI; E, EcoRI; R, EcoRV; S, SaII; Sm, SmaI; X, XbaI. The figure was generated from Innovative Data Design MacDraft, converted to MacroMedia FreeHand 8, and printed from Adobe Photoshop 5.5.

View larger version (11K): [in this window] [in a new window]   FIG. 4.   Physical map of pCPP719 (top), with the DNA of the pCPP9 vector indicated as a thick bar. The ClaI fragment resulting from a partial digest of pCPP810, containing the mini-Tn5Cm insertion A14 (bottom) was cloned into pBR322 to create pCPP723. This clone was used for marker exchange of insertion A14 into Eh318 and Eh421 (PanA) to generate Eh439 (PanB) and Eh440 (PanAB), respectively. At the bottom is a map of the locations of mini-Tn5Cm insertions (filled lollipops), all of which resulted in abolished pantocin B production. The 23-kb SmaI fragment that was radioactively labeled for use in Southern blot experiments is indicated as two solid gray bars (below map). The dashed lines indicate the points of continuation in the circular pCPP719, and tick marks indicate the points of SmaI cleavage. Abbreviations: B, BamHI; C, ClaI; E, EcoRI; S, SalII; Sm, SmaI; X, XbaI. The figure was generated from Innovative Data Design MacDraft, converted to MacroMedia FreeHand 8, and printed from Adobe Photoshop 5.5.

Transposon mutagenesis and construction of clones for mutagenesis of Eh318. The two subclones, pCPP1051 and pCPP719, were subsequently mutagenized with Tn5 and mini-Tn5Cm, respectively, to generate constructs potentially useful for marker exchange mutagenesis of Eh318 and to determine the approximate locations and sizes of pantocin-encoding regions in the subclones. Colonies of DH5(pCPP1051::Tn5) and DH5(pCPP719::miniTn5Cm) that had lost the ability to inhibit growth of E. amylovora were selected, and the transposon insertions were mapped. The Tn5 insertion 122 abolished pantocin A production by DH5(pCPP1051) while insertion 326 did not (Fig. 3). The 29 mini-Tn5Cm-insertions all abolished pantocin B production by DH5(pCPP719) (Fig. 4). Insertion A14 was chosen for marker exchange mutagenesis. The 11.8-kb EcoRI fragment of pCPP1051 containing insertion 122 was cloned from pCPP745 (a pantocin A-deficient Tn5 mutant of pCPP1051) into pBR325 to generate pCPP726 (Fig. 3). Similarly, a 10.3-kb ClaI fragment of pCPP719 containing insertion A14 was cloned from pCPP810 (the mini-Tn5Cm mutant of pCPP719) into pBR322 to generate pCPP723 (Fig. 4).

Marker exchange mutagenesis. Plasmids pCPP726 and pCPP723 were separately introduced into Eh318 and subsequently cured, allowing for homologous recombination of the transposon insertions into the genome of Eh318. A marker exchange mutant originating from Eh318(pCPP726) was designated Eh421 (PanA), and a mutant originating from Eh318(pCPP723) was designated Eh439 (PanB). Eh421 (PanA) was mutagenized using pCPP723 to generate a mutant of Eh318 that carried both transposons, designated Eh440 (PanAB). Eh440 (PanAB) was identified by its complete lack of antibiosis toward Ea273 in a live test. The genomic DNA of Eh421 (PanA) was hybridized to a radioactive probe of a 3.9-kb XbaI-HindIII fragment of pCPP1051 (Fig. 3) and that of Eh439 (PanB) and Eh440 (PanAB) to a probe of the 23-kb SmaI fragment of pCPP719 (Fig. 4), respectively. Since Eh440 (PanAB) was derived from Eh421 (PanA), it was not necessary to confirm the location of the Tn5 insertion. In all cases, the mutants were true marker exchange mutants, based on analysis of the Southern blots (see details of analysis below) (55). DNAs of Eh421 (PanA) and Eh318 were digested with NotI and XbaI, HindIII, BglII, and EcoRI and probed with the 3.9-kb insert DNA of pCPP717. The Tn5 insertion was detected in the expected position based on the analysis of several enzyme digests (data not shown) (55). The 12-kb NotI-XbaI fragment of Eh318 was replaced by a 10- and a 3.2-kb fragment in Eh421 (PanA). The 1.45-kb HindIII fragment of Eh318, which had been mutated by the inserted transposon, was absent in Eh421 (PanA) as expected.

Cross-hybridization data. EcoRI-XbaI-, EcoRV-, and EcoRI- digested DNA of Eh318, pCPP702, pCPP704, and pCPP1051 was hybridized to a 3.9-kb XbaI-HindIII fragment of pCPP1051 that encompassed the DNA region of the Tn5 insertion site 122 in the pantocin A-deficient clone pCPP726 (Fig. 3). The only hybridizing fragments were those of Eh318, pCPP702 or pCPP1051 origin, as indicated in Fig. 2: a 9.9-kb EcoRI-XbaI (lane 1), a 7.2-kb EcoRI-XbaI (lanes 2 and 4), a 9.9-kb EcoRV (lane 9), and a 12-kb EcoRI (lane 16 and 18) band (data not shown).

Effect of amino acid supplementation on activity of pantocins. The antibiotic activity of DH5(pCPP702) to E. amylovora was inhibited by the presence of histidine but not arginine; however, arginine but not histidine inhibited the activity of DH5(pCPP704). However, the zone of inhibition produced by Eh318 in an overlay seeded with E. amylovora was not affected by the presence of either amino acid, when added separately or together.

Spectrum of activity. The antibacterial spectra of activity of Eh318, DH5(pCPP702), Eh439 (PanB), DH5(pCPP704), Eh421 (PanA), and Eh440 (PanAB) are summarized in Table 2. The antibacterial spectra of the two antibiotics produced by Eh318 that are inhibitory to E. amylovora differ somewhat but overall are highly similar. Typically, the antibiotics inhibit close relatives of Eh318, such as species of Pantoea, Erwinia, Enterobacter, and Serratia. Eh252 was resistant to pantocin A but not pantocin B (Fig. 5; Table 2). In addition, we can conclude from the data that a third antibiotic of Eh318 inhibits some nonenterics, judging by the spectrum of activity of Eh440 (PanAB), and this antibiotic does not inhibit E. amylovora.

View larger version (66K): [in this window] [in a new window]   FIG. 5.   Sensitivity of Eh252 (A) and Eh318 (B), to antibiotics produced by Eh318, Eh252, DH5(pCPP1051) (denoted 1051), and DH5(pCPP719) (denoted 719), in a live assay.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

P. agglomerans strain Eh318 produces two antibiotics that are active against E. amylovora Ea273 based on genetic, biological, and chemical evidence. We have proposed to name them pantocin A, whose biosynthetic genes are present in pCPP702 and pCPP1051, and pantocin B (10), whose biosynthetic genes are present in pCPP704 and pCPP719. The DNA regions responsible for the synthesis of pantocin A and B are distinct based on size, restriction maps (Fig. 3 and 4), and lack of hybridization of DNA for biosynthesis of pantocin A to that for biosynthesis of pantocin B (Fig. 2). The activities of the two antibiotics also are clearly distinct. The activity of DH5(pCPP702) is lost in the presence of histidine, while that of DH5(pCPP704) is lost in the presence of arginine. The spectra of activity for the antibiotics produced by strains that synthesize one or both antibiotics are distinct (Table 2). The double zone of inhibition produced by Eh318 in an overlay seeded with Ea273 (Fig. 1) likely is due to the presence of two antibiotics, of which one diffuses further than the other. Mutants that produce only one of the pantocins, i.e., Eh421 (PanA) and Eh439 (PanB), produce single, discrete zones in overlays of Ea273. A variant of Ea273 with spontaneous resistance to the antibiotic produced by Eh421 (PanA), i.e., to pantocin B is sensitive to Eh318.

Chemical data also suggest that the two antibiotics are distinct. Pantocin B was recently identified as (R)-N-[((S)-2-amino-propanoylamino )-methyl ] - 2 - methanesulfonyl-succinamic acid, a peptide of 296 Da (10). It is sufficiently stable to allow for its isolation and characterization from culture supernatants of DH5(pCPP719). In contrast, pantocin A is labile to extremes of pH, and therefore it has been recalcitrant to isolation and structural characterization using similar procedures as employed for pantocin B (M. Jin, personal communication).

Subcloning and transposon mutagenesis data suggest that the genetic region involved in the biosynthesis of pantocin A is at most 7.5 kb, while that of pantocin B is at least 18.5 kb. The Tn5 insertion 326 at map position 4.3 (Fig. 3) does not abolish the antibiotic activity of DH5(pCPP1051). A subclone of pCPP1051 that carries only the 11.8-kb EcoRI fragment confers pantocin A production on DH5 (55). Hence, the genes for pantocin A biosynthesis lie within a 7.5-kb region. Based on the mapping of 29 mini-Tn5 insertions in pCPP719 (Fig. 4), the biosynthetic region for pantocin B is at least 18.5 kb. The genetic regions involved in the synthesis of antibiotics of other strains of P. agglomerans-P. dispersa are under investigation elsewhere. For Eh1087, a New Zealand strain, a 2.2-kb region was found to be essential (31). In Eh252, another New York strain, deletion and complementation analysis of transposon-bearing clones delimited the mccEh252 biosynthetic genes to a 2.4-kb region (44). In C9-1, a Michigan strain, a cosmid clone, AA818, was identified from a genomic DNA library that confers on DH5 the ability to synthesize herbicolin O (11).

Pantocin A and pantocin B have similar but distinct spectra of activity. However, only pantocin B inhibited P. agglomerans Eh252, Xanthomonas campestris pv. pelargonii and E. coli DH5 (Table 2). The two pantocins produced by DH5 carrying the cosmids mainly inhibited enteric strains of bacteria. They are together solely responsible for the inhibition of Erwinia stewartii, Erwinia chrysanthemi, Erwinia carotovora subsp. carotovora, and E. amylovora by Eh318, judging by the absence of inhibition of these strains by Eh440 (PanAB). This result is consistent with those of El-Goorani and coworkers, who found that the antimicrobial activity of Eh318 primarily affects enterics, with the exception of Rhodococcus fascians (15), and the same was true for several other strains of Pantoea spp. (15, 25, 28). The antibiotics of several Pantoea strains were initially designated bacteriocins due to their inhibition primarily of closely related species (3). Based on the inhibition by Eh440 (PanAB) of Streptococcus faecalis, X. campestris pv. pelargonii, Pseudomonas syringae pv. tomato, and Klebsiella pneumoniae, Eh318 likely produces a third antibiotic compound that is ineffective on erwinias.

Earlier studies have found that the zone(s) of inhibition produced by Eh318 in an overlay seeded with E. amylovora is abolished or reduced in diameter to 50% or less in the presence of a combination of histidine and arginine (48-50). However, in our experiments, using as much as 10 mg of the two amino acids per ml in the overlay did not prevent the formation of zones. Perhaps numbers of Eh318 cells used or the time they were allowed to produce antibiotic overcame the arginine-histidine supplementation effect. The arginine effect on pantocin B activity likely is due to the redundancy of the arginine biosynthetic pathway, the target of pantocin B (10), when arginine is supplied exogenously.

Pantocin A is inactive in the presence of histidine, which is a characteristic of most antibiotics produced by Pantoea species that have been tested (14, 50, 51). MccEh252, the antibiotic produced by P. agglomerans strain Eh252 (J. L. Vanneste, J. Yu, D. C. Cornish, and M. D. Voyle, 7th Int. Congr. Plant Pathol, paper 3.5.4, 1998 [www.bspp.org.uk/icpp98/abstracts /3.5/4.html]), also is antagonized by histidine. Interestingly, Eh252 was unaffected by pantocin A, but it was antagonized by pantocin B (Fig. 5; Table 2). Strains Eh252 and Eh318 both were isolated from apple tissue in the same fruit growing area in New York state. The two histidine-type antibiotics, pantocin A and mccEh252, have low molecular masses (<3,000 Da) (Jin, personal communication; Vanneste et al., 7th Int. Congr. Plant Pathol.). Although their structures are not known, mccEh252 has been proposed to be a peptide and a microcin (46; Vanneste et al., 7th Int. Congr. Plant Pathol.) based on its protease sensitivity (45). Pantocin A is water soluble (Jin, personal communication) and also is probably a small peptide (55). A DNA fragment of Eh252 hybridized to a similar sized fragment (2.5 kb, XbaI-HindIII) in Eh318 when probed with radioactively labeled DNA of pCPP1051 (S. A. I. Wright and S. V. Beer, unpublished data). This suggests that pantocin A and mccEh252 have one or more homologous genes that are involved in biosynthesis. It is unlikely, however, that pantocin A and mccEh252 are identical, since Eh252, but not Eh318, is active against A. tumefaciens (15), and pantocin A but not mccEh252 is active against Pantoea stewartii and Serratia marcescens (Table 2) (45; Wright and Beer, unpublished data).

The only Pantoea histidine-type antibiotic for which there exists some structural information is herbicolin O, a -lactam antibiotic (27). Herbicolin O is similar to pantocin A in that its molecular weight is less than 3,500 and its activity is labile to acid (pH 3.5) and base (pH 10) (28). Although their antimicrobial spectra have not been compared in the same assay at the same time, they are both active against Eh112Y, E. amylovora, Enterobacter aerogenes, S. marcescens, E. carotovora subsp. carotovora, and inactive against several pseudomonads, Bacillus megaterium, and K. pneumoniae (Table 2), (15, 28). Pantocin A, mccEh252, and herbicolin O may fall into a family of structurally related compounds. Interestingly, the genes responsible for their synthesis do not reside on native plasmids (11, 45, 55), whereas the biosynthetic genes for several other antibiotics of Pantoea sp. are plasmid borne (19, 30, 42).

We have demonstrated the value of using a genomic library to identify and isolate clones corresponding to distinct biosynthetic regions of two antibiotics. The differences in their inhibitory activities and sensitivities to extremes of pH, in addition to the genetic difference, clearly indicate that pantocin A and pantocin B are two distinct compounds, which are produced by one strain of P. agglomerans.