INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Siderophores are relatively low-molecular-weight, iron-chelating agents produced by bacteria and fungi growing under low-iron stress conditions (42). These compounds bind ferric iron (Fe³⁺) with high affinity and transport the metal into the cell via dedicated uptake systems (41). Microbial siderophores vary widely in their overall structure, but most of them contain hydroxamate or catechol groups, which are involved in chelating the ferric ion (42). Apart from their role in the active transport of iron, siderophores may act as growth factors, and some show potent antimicrobial activity (41).

The capacity to utilize siderophores is important to the growth of bacteria in the rhizosphere (27) and on plant surfaces (34). Specific siderophore-producing Pseudomonas strains rapidly colonize plant roots of several crops, and this colonization can result in significant yield increases (48). Enhanced plant growth caused by these strains often is accompanied by reduction in the populations of fungi and other bacteria on the roots. These beneficial Pseudomonas strains suppress some soil-borne fungal pathogens (33), and there is convincing evidence to support a direct role of siderophore-mediated iron competition in the biocontrol ability exhibited by such isolates (33, 34). Antagonism depends on the amount of iron available in the medium: siderophore production by the biocontrol agent and sensitivity by target pathogens are expressed only under iron-limiting conditions (31, 57).

Agrobacterium rhizogenes strain K84 (formerly called A. radiobacter) is used worldwide as a commercial agent for the biocontrol of crown gall disease caused by tumorigenic Agrobacterium strains (reviewed in reference 20). Production of the antiagrobacterial antibiotic agrocin 84, which is coded for by the agrocinogenic plasmid pAgK84, is a key component in the process of biocontrol by strain K84. However, under field conditions this bacterium can protect plants against crown gall caused by pathogenic isolates resistant to agrocin 84 (11, 12, 15, 35, 36, 45, 55). Moreover, derivatives of K84 that do not produce agrocin 84 (Agr) can protect plants against agrocin 84-resistant pathogens (12, 35, 45). These Agr derivatives also control agrocin 84-susceptible pathogens, but not as effectively as does the wild-type parent (35). Thus, while production of agrocin 84 is required for optimum control of susceptible pathogens, this antibiotic is not the sole component of the biocontrol process (20). The production of antiagrobacterial substances other than agrocin 84 may play a role in the biocontrol of crown gall by strain K84 (20, 38, 55). In this regard, strain K84 produces a second antiagrobacterial substance called agrocin 434 (14). This agrocin, which most probably is a disubstituted cytidine nucleoside (17), affects only A. rhizogenes strains (previously biovar 2 strains of Agrobacterium). Recently, McClure et al. (38) suggested that agrocin 434 plays a direct role in controlling agrocin 434-susceptible pathogens.

Strain K84 also produces a third antibiotic-like substance named ALS84, which inhibits many tumorigenic Agrobacterium strains in vitro (44). The nonspecific inhibitory activity of ALS84 is observed in mannitol-glutamate medium but not in Stonier's medium (44). These two media differ in a number of ways including their relative iron concentrations, raising the possibility that ALS84 is produced in response to low-iron conditions. Aside from agrobactin, the catechol-type siderophore secreted by A. tumefaciens strain B6 (43), there is no information available concerning the production of siderophores by members of the genus Agrobacterium. Furthermore, production of siderophores by strain K84 has not been investigated. The aim of this study was to evaluate the ability of a collection of isolates of different Agrobacterium spp. to produce iron-binding compounds under iron-limiting conditions. Because of its biocontrol properties, emphasis was placed on A. rhizogenes strain K84, with particular reference to determining whether a relationship exists between the production of siderophores and ALS84 by this bacterium.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, media, and growth conditions. The bacterial strains and plasmids used in this study are listed in Tables 1 and 2. Luria broth (Gibco-BRL, Gaithersburg, Md.) was used as the liquid medium for growing Escherichia coli. Agrobacterium strains were grown on nutrient agar (NA) (Difco, Detroit, Mich.) and in AB minimal medium (10) supplemented with 0.2% mannitol (ABM) as the carbon source. The minimal medium used for growing A. rhizogenes (biovar 2 strains) was supplemented with biotin (2 µg/ml). Antibiotics were added to the media at the following concentrations: for Agrobacterium, kanamycin, 100 µg/ml; neomycin, 100 µg/ml; gentamicin, 100 µg/ml; and tetracycline, 2 µg/ml; for E. coli, kanamycin, 50 µg/ml; and tetracycline, 20 µg/ml. Pseudomonas aeruginosa CECT110 (6), Salmonella enterica serovar Typhimurium LT2, and S. enterica serovar Typhimurium enb-1, which is a mutant of LT2 deficient in enterobactin production (46), were used as positive and negative control strains in assays for siderophore production. All strains were routinely grown in King's medium B (30) prior to analysis for siderophore production. For analysis of siderophore production, strains were grown in Stonier's medium (ST) (53), mannitol-glutamate (MG) medium (40), or CM9 medium (7). MG and CM9 media were supplemented with FeCl₃ to give a final concentration of 10 µM Fe in all of its forms, producing MGF and CM9F media, respectively. To avoid contamination with iron, glassware was cleaned with 6 M HCl (32). Casamino Acids were deferrated by extracting stock solutions with 3% (wt/wt) 8-hydroxyquinoline in chloroform (8). Water purified by reverse osmosis and by Milli-Q treatment (Millipore Corp., Bedford, Mass.) was used without exception throughout this study. When needed, the iron chelator ethylene-di(o-hydroxyphenylacetic acid) (EDDHA) was added to the medium at concentrations from 1 to 20 µM to chelate contaminating iron or to ensure iron-limited conditions. When EDDHA was used, the growth medium was stored in the cold for 48 h prior to use, to promote iron chelation. For bacterial growth under various iron conditions, cultures in MG medium at a density of ca. 10⁸ CFU per ml were diluted in the medium to be tested to about 10³ viable bacteria per ml. Cultures were incubated at 26°C and grown aerobically by shaking at 150 rpm. The numbers of bacteria in cultures at different times were determined by plating serial dilutions of samples on King's medium B. 

Detection of iron-chelating compounds and ALS84 activity. Cultures in CM9 broth were incubated at 26°C with shaking (200 rpm) to a density of ca. 10⁸ CFU per ml. Each culture was diluted to approximately 10³ cells per ml in fresh medium and incubated under the same conditions for 10 days. Throughout the incubation, the population density was monitored turbidimetrically at 660 nm to verify iron starvation, which was manifested by slower growth and a smaller final population compared to the same isolate grown under iron-sufficient conditions (CM9F medium). When the culture reached stationary phase, samples were taken, the cells were removed by centrifugation, and 500-µl volumes of the supernatant were assayed for the presence of iron chelators using the Chrome Azurol S (CAS) assay (49). This assay detects hydroxamate- and catechol-type siderophores. The synthetic iron chelator deferriferrioxamine mesylate (Desferal) was used as a control for analytical purposes, and amounts of iron chelators produced by each Agrobacterium strain were expressed as equivalents of Desferal (0.1 to 10 µM). The presence of hydroxamate and catechol compounds in the supernatants was differentiated by colorimetric assays: catechols were detected by Arnow's method (5), and hydroxamate compounds were detected by a modification of the method of Csàky (4).

DNA manipulations. Plasmid DNA was isolated from Agrobacterium spp. and E. coli by an alkaline lysis procedure as previously described (24). Total genomic DNA was prepared by the method of Glickmann et al. (22). Standard recombinant DNA techniques were used as described by Sambrook et al. (47). Plasmids were introduced into E. coli strains by transformation using CaCl₂ or by electroporation and into Agrobacterium by electroporation or by biparental cross-streak mating using E. coli S17-1 (Table 1) harboring the plasmid of interest as the conjugal donor (18, 51).

Transposon mutagenesis of strain K84 and Tn5 insertion rescue. Tn5, which codes for resistance to kanamycin and neomycin, was introduced into strain K84 via the suicide plasmid pJB4JI by conjugation with E. coli 1830 (9) as previously described (19). When plated onto medium containing kanamycin only, mutants of K84 resistant to this antibiotic appear at a high frequency (data not shown). However, when kanamycin and neomycin were used together, such spontaneous mutants did not appear at detectable levels (<10¹⁰). Consequently, transposon-induced mutants of K84 were selected on ABM supplemented with biotin, kanamycin, and neomycin (each at 100 µg per ml). Antibiotic-resistant colonies were analyzed for ALS84 activity by the antibiosis plate assay and further characterized for production of the hydroxamate siderophore by the CAS and Csàky tests.

Marker exchange mutagenesis. Tn5 insertion 8-10 in pPOR8-10 (Table 1) was marker exchanged into wild-type strain K84 using pPH1JI from E. coli 2170 (Table 1) as the eviction plasmid, all as described by Garfinkel et al. (21). Valid marker exchange events were confirmed by Southern analysis of genomic DNA isolated from the transconjugants exhibiting the correct antibiotic resistance phenotypes.

Library construction and complementation analysis. To construct a genomic library of strain K84, total DNA was partially digested with HindIII and the fragments were ligated with the cosmid vector pCP13/B as described by Hayman and Farrand (24). Ligation products were packaged in phage lambda using the Packagene kit (Stratagene, La Jolla, Calif.) as recommended by the manufacturer and recovered by transfection into E. coli LE392. Clones of interest were identified by colony hybridization using the mutated 8-10 fragment from pPOM8-10 (Table 1) as the probe.

Southern analysis. Samples of genomic DNA were digested with EcoRI, and the fragments were separated by electrophoresis on 0.7% agarose gels. DNA was transferred to a nylon membrane (Boehringer GmbH, Mannheim, Germany) as described by Sambrook et al. (47). Southern hybridizations were done using as probe a DNA fragment consisting of an internal 5.4-kb fragment of Tn5 obtained by HpaI digestion or the mutated 8-10 cloned fragment. Probes were labeled by random priming with digoxigenin-dUTP. Prehybridization, hybridization, and colorimetric detection were performed using the DNA High Prime labeling and detection kit (Boehringer) as recommended by the manufacturer.

DNA sequence analysis. DNA was sequenced from templates by the Genetic Engineering facility of the University of Illinois at Urbana-Champaign, or by BioS&T, Inc., Lachine, Quebec, Canada. In both cases the sequences of both strands were determined. DNA and translatable protein sequences were analyzed using the GCG package (version 8.1; Genetics Computer Group, Madison, Wis.), and related sequences in the databases were identified using the BLAST protocols (1, 2).

Nucleotide sequence accession number. The nucleotide sequence reported here was deposited in the GenBank database under accession number AF110469.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Production of iron-binding compounds. With the exception of A. tumefaciens strain C58 and A. vitis strain 565-50, all isolates of Agrobacterium spp. tested produced compounds that clearly react in the CAS assay, which detects ferric iron-chelating compounds (Table 2). A notable heterogeneity was observed with the CAS assay in the amount of iron chelators produced by the strains tested within each species. However, A. rhizogenes strain K84 and its derivatives, including strain K1026, in which pAgK84 has been rendered Tra, the two plasmid-deficient derivatives, K84 Agr (lacking pAgK84), and K1143 (lacking pAgK84 and pAtK84b), all produced similarly large amounts of iron chelators when grown under iron-limiting conditions (Table 2).

Production of a hydroxamate compound and ALS84 by strain K84 depends on iron-limiting conditions. As assessed by the CAS and Csàky tests strains K84 and K84 Agr, a derivative lacking the agrocinogenic plasmid pAgK84, both produced the hydroxamate compound when grown in MG and CM9 media (Table 3). However, when grown in iron-supplemented media (MGF and CM9F media), neither strain produced detectable amounts of the compound. Similarly, the two strains did not produce the hydroxamate in ST medium, which contains nonlimiting amounts of iron (Table 3). We then examined the growth properties of strain K84 in two media, one rich in iron and the other iron limited. As shown in Fig. 1A, this strain grows well in MGF medium, which contains nonlimiting amounts of iron, ultimately reaching a final population size of greater than 10⁹ CFU per ml. Strain K84 also grows in the severely iron-deficient medium consisting of MG containing EDDHA, albeit at a lower exponential rate and to a final population size smaller by a factor of 20 (Fig. 1B). Clearly, strain K84 can scavenge the very small amounts of iron still present in MG medium supplemented with EDDHA. From these two sets of results, we tentatively conclude that the hydroxamate compound produced by strain K84 is a siderophore.

View larger version (9K): [in this window] [in a new window]   FIG. 1.   Growth of A. rhizogenes strain K84 and its M-8-10 mutant in iron-sufficient and iron-deficient media. Bacterial strains were grown in MGF medium, which contains non-limiting amounts of iron (A), and in MG medium supplemented with the chelator EDDHA, which results in severely iron-limiting conditions (B), as described in Materials and Methods. Cell growth, expressed as CFU per milliliter, was monitored by viable counts at the indicated times. , K84; , M8-10; , M8-10(pCOM8-10).

Isolation of Tn5-induced mutants of strain K84. Of a total of 417 antibiotic-resistant Tn5-generated transconjugants of strain K84 tested, 5 were identified as defective for ALS84 production using the antibiosis plate assay (Fig. 2A; Table 4). Two of the mutants, M8-10 and M12-22, did not produce a detectable zone of growth inhibition against the indicator strain, while the three others showed reduced zones of inhibition (Fig. 2A). These results were mirrored by chemical tests for the production of the hydroxamate siderophore; M8-10 and M12-22 did not produce detectable amounts of the hydroxamate siderophore as assessed by the CAS and Csàky tests, while the three other mutants produced the hydroxamate siderophore in amounts smaller than the wild-type parent (Table 4).

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Production of ALS84 activity by A. rhizogenes K84 and its mutants. (A) K84 and the five Tn5-induced mutants defective in production of the hydroxymate siderophore. (B) K84, the M8-10 mutant uncomplemented and complemented with pCOM8-10, and the marker-exchanged mutant, K84::M8-10. A. rhizogenes K84 and its mutant derivatives were spot-inoculated onto MG medium and incubated at 28°C for 48 h. Production of ALS84 was assessed by overlaying the plates with a suspension of strain NT1 as the indicator, as described by Peñalver et al. (44).

View larger version (85K): [in this window] [in a new window]   FIG. 3.   ALS84-deficient mutants of K84 contain single unique inserts of Tn5. Genomic DNA samples from strain K84 and its mutants were digested with EcoRI, and the fragments were separated by electrophoresis in a 0.8% agarose gel, denatured, and transferred to a nitrocellulose filter. The filter was hybridized with a labeled probe corresponding to the 5.4-kb internal HpaI fragment of Tn5, and hybrids were detected by a colorimetric reaction as described in Materials and Methods. Lanes contain DNA from K84 (lane 1), M8-10 (lane 2), M8-12 (lane 3), M9-22 (lane 4), M11-11 (lane 5), and M12-22 (lane 6). Numbers at the right and left margins indicate the sizes, in kilobase pairs, of the hybridizing fragments in the five mutants.

Marker exchange mutagenesis and complementation of the M8-10 mutant. A 9.1-kb EcoRI fragment containing the Tn5 insertion in M8-10 was recovered by cloning an EcoRI digest of genomic DNA into pUC19. The fragment, present on pPOM8-10 (Table 1), was recloned into pRK415 to yield pPOR8-10, and this clone was used to marker exchange the insertion mutation into wild-type strain K84 as described in Materials and Methods. Correct homogenotes were identified based on antibiotic resistance profiles and were confirmed by Southern analysis using the Tn5 probe (data not shown). All proper homogenotes tested failed to produce either ALS84 (Fig. 2B and data not shown) or the hydroxamate siderophore (data not shown). These results confirm that both phenotypes are due to a single transposon insertion. One such homogenote, K84::8-10, was retained for further studies (Fig. 2B; Table 4).

The M8-10 mutant is defective for growth under low-iron conditions. Strain M8-10 grew at exponential rates and to final yields indistinguishable from those of its parent in MG medium supplemented with iron (Fig. 1A). However, the mutant failed to sustain growth in MG medium containing EDDHA (Fig. 1B). Introducing pCOM8-10 into M8-10 restored growth of the mutant to parental levels in the severely iron-depleted medium (Fig. 1B). Thus, the mutation in M8-10 abolishes production of the hydroxamate compound and also the ability to grow in iron-deficient medium, and both phenotypes are restored to normal by pCOM8-10.

Sequence analysis. We subcloned the 3.5-kb EcoRI fragment from pCOM8-10 into pUC19 to produce pPO8-10 (Fig. 4) and determined the complete double-strand sequence of the insert. The segment is 3,500 bp long and contains a single uninterrupted open reading frame (ORF) (Fig. 4 and 5). The deduced amino acid sequence of this ORF indicates that it could encode a portion of a protein that is related to peptide synthetases of the gramicidin S synthetase repeat superfamily found in a range of bacterial and fungal species (Fig. 5A). The product of this ORF is 30% identical and 46% similar at the amino acid sequence level to two segments of the product of the pvdD gene from P. aeruginosa (Fig. 5B and C). PvdD, which is composed of two long amino acid repeats, is one of the four identified proteins required in the pathway for biosynthesis of the hydroxamate siderophore pyoverdine by this bacterium (39). We conclude that the 3.5-kb ORF is an internal portion of a longer gene required for biosynthesis of ALS84 and the hydroxamate siderophore in strain K84.

View larger version (4K): [in this window] [in a new window]   FIG. 4.   Genetic structure of the siderophore biosynthesis locus of A. rhizogenes K84. The upper line shows the restriction map of a portion of the ca. 30-kb insert of genomic DNA in pCOM8-10, which complements the ALS84- and siderophore-defective phenotypes of the M8-10 mutant of A. rhizogenes K84. pPO8-10 represents the 3.5-kb EcoRI fragment from pCOM8-10, which was sequenced. The product coded for by the ORF in this fragment is depicted as the open bar. The inverted triangle indicates the site of the Tn5 insertion within the 3.5-kb EcoRI fragment in the M8-10 mutant.

View larger version (71K): [in this window] [in a new window]   FIG. 5.   Repeat structures of members of the gramicidin S synthetase gene superfamily and their relatedness to the translation product of orf1 from strain K84. (A) Repeating structures of several members of the gramicidin S gene family including ORF1 from A. rhizogenes K84. SyrE is from P. syringae pv. syringae and contributes to the synthesis of syringomycin (23). SnbDE is from Streptomyces pristinaespiralis and is required for synthesis of pristinamycin (13). (B and C) Relatedness of the protein product of ORF1 to the first (B) and second (C) repeats of PvdD. The predicted amino acid sequence of ORF1 was aligned with that of PvdD using the Pile-Up routine of the GCG program package.

Sequences homologous to ORF1 are present in some but not all isolates of Agrobacterium spp. Using the mutated 8-10 EcoRI cloned fragment as probe, we examined the genomes of other isolates of Agrobacterium spp. for sequences related to the 3.5-kb EcoRI fragment associated with the synthesis of the hydroxamate siderophore. With the exception of strain 619, the probe hybridized with the genomic DNA from all A. rhizogenes isolates tested (Table 2). All hybridizing isolates produced detectable amounts of hydroxamate compounds and inhibited the growth of strain NT1, but only under iron-limiting conditions (Table 2). A. rhizogenes 619, the genomic DNA of which did not hybridize with the biosynthetic gene, produced detectable amounts of hydroxamate compounds, but this strain did not produce any substances that inhibited the growth of strain NT1. The probe did not detectably hybridize with genomic DNA from A. tumefaciens or A. vitis isolates, even from strains that produced hydroxamate compounds and also inhibited strain NT1 under iron-limiting conditions (Table 2). Genomic DNA from P. aeruginosa, which produces hydroxamate siderophores, and from S. enterica serovar Typhimurium strain LT2, which produces a catechol siderophore, did not hybridize detectably with the probe containing the biosynthetic gene from K84 (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We conclude from these studies that with the exception of strain C58 and possibly strain 565-50, all isolates of Agrobacterium spp. tested produce detectable levels of iron-chelating compounds when grown under iron-limiting conditions. As assessed by the CAS assay, there is a notable heterogeneity in the amount of iron chelators produced in culture by various strains within each species. Moreover, the hydroxamate and catechol compounds produced by the members of the genus vary from one isolate to another, with some producing only catechols and others producing only hydroxamate compounds. Nine isolates distributed among the three studied species gave positive reactions in both the Arnow and Csàky tests. These strains may produce compounds of both classes; simultaneous production of catechol and hydroxamate siderophores has been reported for some members of the Enterobacteriaceae (3, 7) and for some Pseudomonas spp. (6). Alternatively, these isolates may produce a catechol-hydroxamate-type compound similar to the pyoverdines produced by Pseudomonas spp., which react in both chemical tests.

A. rhizogenes strain K84 produced large amounts of a ferric iron-chelating compound that reacts in the Csàky test for hydroxamate groups. Two lines of evidence from physiological studies support our hypothesis that this hydroxamate compound is a true siderophore. First, the compound is produced in large amounts only when the cells are grown in iron-limited medium. Second, production of the compound is required for growth under such conditions; while K84 can grow in low-iron medium, strain M8-10, a mutant of K84 that does not produce the activity, exhibits a significant growth defect when cultured in iron-deficient media. On the other hand, the mutant and its parent show virtually identical growth properties when cultured in iron-sufficient medium.

Our genetic and biochemical results indicate that production of the previously reported inhibitory substance ALS84 is related to synthesis of the hydroxamate siderophore in strain K84. Production of both ALS84 and the siderophore requires identical low-iron conditions. In addition, the degree of the defect in production of ALS84 shown by the five Tn5-derived mutants of strain K84 was mirrored by the degree of the defect of those mutants in production of the siderophore. Third, a single complementable mutation in strain K84 abolished production of both activities. While it was possible that this mutation disrupted some positive control function, sequence analysis of the mutated ORF in strain M8-10 showed that the transposon is located in a gene related to genes required for hydroxamate-type siderophore biosynthesis in other bacteria. The deduced amino acid sequence of the product of this ORF is related to peptide synthetases of the gramicidin S synthetase repeat superfamily including pyoverdine synthase D, an enzyme involved in the biosynthesis of the siderophore pyoverdine in P. aeruginosa (39). Members of this family are large genes ranging in size from 6 to 12 kb (13, 23, 54) and characteristically contain two or more highly similar repeating units (Fig. 5A). For PvdD, the protein contains two similar domains of about 1,000 amino acids each and the corresponding DNA repeat sequences span about 3 kb each (39). Thus, the fragment that we sequenced, which corresponds to the site at which Tn5 is inserted in the M8-10 mutant, probably represents an internal portion of a pvdD-like gene. This fragment is only a part of the gene itself and, no doubt, a small portion of the locus required for production of ALS84 and the siderophore produced by strain K84. A region of at least 78 kb of the chromosome is required for pyoverdine biosynthesis in P. aeruginosa PAO (54). Given that pCOM8-10, which contains a ca. 30-kb insert, does not confer siderophore production on C58, we suspect that a similarly large locus is required for biosynthesis of the chelator and ALS84 activities in K84 (data not shown).

At the least, these results show that ALS84 and the hydroxamate siderophore produced by strain K84 have a common biosynthetic route. However, we consider it likely that the two activities are one and the same compound. Proof of this hypothesis awaits chemical analysis. However, the inhibitory activity of ALS84 is detectable only when the sensitive strain is tested under iron-limiting conditions, suggesting that the antagonism exerted by the compound operates through the iron acquisition system. This dependence on low-iron conditions indicates that ALS84 may deplete the medium of available iron, thereby inhibiting the growth of the indicator strain. This hypothesis, coupled with the fact that a mutation in a single gene abolishes the production of the hydroxamate iron chelator and ALS84, strongly suggests that ALS84 is the siderophore.

Genetic and hybridization analyses indicated that the ORF1 gene required for biosynthesis of ALS84 and the siderophore is not located on any of the three plasmids resident in strain K84. This observation agrees with a previous report showing that ALS84 synthesis is not coded for by pAgK84 or pAtK84b (44). Thus, we conclude that production of ALS84 and of the hydroxamate siderophore are encoded by the chromosomes of strain K84.

Based on chemical and biological assays and on Southern analyses, production of hydroxamate compounds similar or identical to the siderophore produced by strain K84 is common among isolates of A. rhizogenes. However, while the presence of fragments in A. rhizogenes that hybridize with the 8-10 probe was correlated with production of hydroxamate-type iron chelators and inhibition of strain NT1 under iron-limiting conditions, the production of hydroxamate compounds is not necessarily correlated with the presence of a hybridizing fragment. For example, genomic DNA from A. rhizogenes strain 619 did not hybridize detectably with the 8-10 probe, but this isolate produced detectable amounts of one or more hydroxamate compounds. Similarly, production of these hydroxamates was not always associated with inhibition of strain NT1, suggesting a heterogeneity with respect to the iron-binding compounds produced by different isolates of A. rhizogenes. Consistent with this interpretation, substantial restriction length polymorphisms exist within the biosynthetic locus among the five independent isolates of A. rhizogenes examined (data not shown). Interestingly, three isolates of A. vitis produced hydroxamate compounds, and these strains inhibited the growth of strain NT1, but again only under iron-limiting conditions. However, the 8-10 probe did not detectably hybridize with genomic DNA from these three strains. Taken together, these results suggest that members of the genus Agrobacterium can produce different types of hydroxamate iron-binding compounds. Alternatively, it is possible that the genes in A. rhizogenes and A. vitis responsible for production of iron chelators are of the same phylogenetic lineage but have diverged significantly over the course of the evolution of these two species.

Strain K84 efficiently colonizes the root systems of several plant hosts (16, 37, 45, 50, 52, 55). It is reasonable to hypothesize that the capacity of this bacterium to colonize the roots of treated plants is an important factor in the successful biological control of crown gall disease. The ability of soil-borne bacteria to produce and utilize siderophores confers an ecological advantage in colonizing the rhizosphere (27). Moreover, pyoverdine-type siderophores produced by certain Pseudomonas spp. are involved directly in the ability of these strains to colonize the rhizosphere in presence of other soil-borne bacteria. Production of this hydroxamate siderophore by strain K84 could represent a trait important to the successful interaction between this microbe and its plant host. In addition, siderophores produced by certain isolates of Pseudomonas spp. play a direct role in biocontrol of some plant pathogens (33, 34). Production of ALS84 and the siderophore may, in a similar fashion, contribute to the biocontrol of crown gall by strain K84, especially under conditions of iron limitation.