INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The plasmid biology of Erwinia amylovora, a phytopathogen that causes fire blight, is gradually emerging. The pathogen infects plants in the family Rosaceae, including apple and pear in the subfamily Pomoideae and a few other rosaceous plants such as Rubus spp. All naturally occurring strains of E. amylovora contain a nontransmissible plasmid called pEA29 or pEA28 (15, 24), and up to 25% of strains also contain a 56-kb plasmid (48). Some streptomycin-resistant strains of E. amylovora carry the tandem strA and strB genes on plasmids. Many of these strains contain the conjugative plasmid pEa34 (28), which carries strA-strB on Tn5393. Other strains carry strA and strB genes on a small mobilizable relative of the IncQ broad-host-range plasmid RSF1010 named pEa8.7 or in transposon Tn5393 on pEA29 (28, 37). The geographic distribution of pEA29 is the same as the distribution of E. amylovora, while pEa34 and pEa8.7 have been detected only in strains of E. amylovora from Michigan and California, respectively. The distribution of the 56-kb plasmid is not correlated with the geographic distribution of strains of E. amylovora. None of the plasmids found in E. amylovora have been completely sequenced, but the sequence of pEa8.7 is probably nearly identical to that of plasmid RSF1010, which has been sequenced (45).

Plasmid pEA29 has received considerable attention because its high stability in E. amylovora suggests that it may confer some traits advantageous to the bacterium. Also, PCR-based detection of pEA29 is the most commonly used method for the detection and identification of E. amylovora and is based on amplification using primers within the 1.1-kb PstI fragment (5, 25, 29). The lengths of the amplification products vary between strains due to differences in the numbers of 8-bp repeats found in the amplified fragments (21, 44), a characteristic that proved to be unstable in the host strain (44). Restriction maps of pEA29 from two isolates have been published by two groups, with some minor differences between the maps (15, 24). The plasmid is not easily eliminated by classical chemical and physical treatments; eviction by incompatibility is required for curing bacteria of pEA29 (15, 24). Plasmid-cured strains are still pathogenic but exhibit reduced virulence (16, 24) and altered growth on minimal medium without thiamine (MM  T) (16, 24). Except for the probable location of the origin of replication on the 4.4-kb PstI fragment and the sequence of a 1.1-kb PstI fragment, little is known about the genetic makeup of pEA29. No specific plasmid sequences have been associated with bacterial virulence or physiology. Increased knowledge of the genetic makeup of pEA29 may be of value in explaining why this plasmid is maintained by E. amylovora.

Sequence analysis of pEA29 was conducted to determine the genetic makeup of this indigenous plasmid. A genetic map was generated based on sequence similarities to GenBank entries for other bacteria. Physiological experiments were conducted with plasmid-cured strains to evaluate their oxidase sensitivities, chemotactic responses, and thiamine auxotrophy. Finally, based on the sequence analysis of pEA29 from strain Ea88, we were able to characterize restriction length polymorphisms found in plasmid pEA29 from Rubus strains and in a few streptomycin-resistant strains of E. amylovora from apple.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial cultures and growth media. The bacterial strains used in this study are listed in Table 1. All strains were maintained on Luria-Bertani (LB) agar (41) and grown in LB broth. Colony morphology and growth characteristics of the strains were evaluated on MM: 6 g of Na₂HPO₄, 3 g of KH₂PO₄, 0.5 g of NaCl, 1 g of NH₄Cl, 0.2 g of MgSO₄ · 7H₂O, 0.1 g of CaCl₂, 0.2 g of nicotinic acid, 0.2 g of thiamine · HCl, 20 g of sucrose, and 14 g of agar per liter (16). MM  S is MM without sucrose and with 4 g of agar per liter. Oxidative stress tests were conducted on MM supplemented with 50 µM FeSO₄ (MM + Fe) and overlaid with a MM + Fe soft agar (7 g of agar per liter) containing 5 g of sucrose per liter and inoculated with the strain to be tested. Cultures for use in oxidative stress experiments were grown in modified Ceria broth: 0.5 g of Casamino Acids, 8.5 g of KH₂PO₄, 0.1 g of MgSO₄, 1 µM MnSO₄, 1 µM FeSO₄, and 4 g of succinic acid per liter (13); the pH was adjusted to 6.0. 

Restriction analysis of pEA29 from various strains. Plasmid pEA29 was isolated and purified from strains Ea88, Ea110, MR1, 2-95, DM22, W2, and BCN20 by alkaline lysis extraction followed by ethanol precipitation. The plasmids were digested with PstI, ClaI (Gibco BRL, Gaithersburg, Md.), and KpnI (Roche Molecular Biochemicals, Indianapolis, Ind.) according to the manufacturers' instructions. Fragment sizes were determined by comparison to the 1-kb Plus DNA ladder or  DNA-HindIII fragments (Gibco BRL) or by calculations made directly from the sequence data. For plasmids carrying insertion elements or transposons, sizes of variable region fragments were determined by sequence analysis. Absence of two PstI restriction sites in pEA29 from strain MR1 was confirmed by sequence analysis.

Plasmid DNA isolation and analysis. Plasmid DNA was isolated using a NucleoBond Plasmid Purification Midi kit (Clontech Laboratories, Palo Alto, Calif.) according to the manufacturer's instructions for low-copy-number plasmids. pEA29 from strain Ea88 was digested with PstI, and the fragments were shotgun cloned into the pGEM3zf(+) cloning vector (Promega, Madison, Wis.). The mixture was added to electroporation-competent Escherichia coli strain JM109 and pulsed at 25 µF, 200 , and 2.5 kV in a Gene Pulser II apparatus (Bio-Rad, Richmond, Calif.) using a 0.2-cm-gapped electroporation cuvette. After incubation for 1 h at 37°C in SOC medium (41), the transformants were plated on LB agar amended with 50 µg of ampicillin per ml. Plasmids from selected transformants were analyzed by PstI digestion; clones of 6 of the 10 PstI fragments of pEA29 were identified and sequenced by primer walking. The remaining sequence was filled in by sequencing directly from pEA29. The orientations of sequences from each fragment were verified by sequencing across PstI sites. Sequences from other strains were obtained by sequencing directly from pEA29. All sequencing was performed at the Michigan State University Department of Energy Plant Research Laboratory sequencing facility using ABI dye-terminator chemistry. Sequenced fragments were analyzed and assembled using the Lasergene software of DNASTAR Inc. (Madison, Wis.). The majority of the assembled sequence was derived from single sequencing runs, the sequences from which were proofread by eye; where positions were ambiguous, they were reported as "N." Sequence analysis was conducted using BLAST version 2.0 (1) at the National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov/BLAST) and the Pfam database of protein families search engine (http://pfam.wustl.edu/) (3). Open reading frames of 375 nucleotides or more were identified, and the sequences of their predicted products were compared to sequences in the GenBank protein database using the BLAST P algorithm and to sequences in the Pfam database. The algorithms BLAST N and BLAST X were also used to compare pEA29 DNA sequences to those in the GenBank database.

PCR. E. amylovora strain Ea88 was used as a positive control, and 10 µl of water was used as a negative control for all reactions. Primers AJ75 (5'-CGTATTCACGGCTTCGCAGAT) and AJ76 (5'-ACCCGCCAGGATAGTCGCATA) were used to detect pEA29 as previously described (29). Primers AJ484 (5'-CCTGTCTGGCGCATCGTATCC) and AJ485 (5'-AAAACGCGCAACCAGCATAGTGAG) were used to detect a variable region of pEA29 initially found in Rubus strain MR1. Primers AJ486 (5'-AACCGACTAAAAACGCATTGACCT) and AJ487 (5'-CTCGATCGAAGCTTGCAGAACCTC) were used to detect the msrA gene of pEA29. Amplification of the msrA gene from the chromosomes of E. coli and Erwinia chrysanthemi was conducted using primers AJ631 (5'-GATCTTCGCCATGGGTTGCTTCT) and AJ632 (5'-GGTGCTCGTCTTCGGCGTAGTAAA) with chromosomal DNA from E. chrysanthemi strain 4150 as a positive control. PCRs were carried out in a PTC-150 programmable thermal controller (MJ Research, Inc., Watertown, Mass.). The PCR mixtures (50 µl) consisted of 1× PCR buffer (10 mM Tris-HCl, pH 8.3), 1.5 mM MgCl₂, 0.1% Tween 20, 200 µM concentrations of each deoxynucleoside triphosphate, 20 pM each primer, 2.5 U of Taq DNA polymerase (Gibco BRL), and either 0.1 to 100 ng of template DNA or 1 µl of bacterial culture diluted to an optical density at 600 nm between 0.1 and 0.5. Reaction mixtures were covered with 50 µl of light mineral oil prior to placement in the thermal controller. Cycling parameters were 94°C for 4 min followed by 40 cycles of 94°C for 1 min, 52°C (variable region), 57°C (plasmid msrA), or 48°C (chromosomal msrA) for 1 min, and 72°C for 2 min, and a final incubation at 72°C for 5 min. Products were analyzed by electrophoresis in 1% (wt/vol) agarose gels in 0.5× Tris-borate-EDTA (TBE) buffer followed by ethidium bromide staining.

Southern analysis and chromosomal-DNA isolation. Genomic DNAs from E. amylovora strains MR1, Ea88, and Ea110 were isolated using a genomic DNA miniprep protocol (56). DNA (20 µg) from each strain was cut with EcoRI, loaded onto a 0.8% (wt/vol) agarose gel in TBE, electrophoresed for 1 h at 10 V/cm, and visualized by ethidium bromide staining. The DNAs were transferred to nylon membranes as previously described (40). A 425-bp probe for the msrA gene was generated by PCR with primers AJ631 and AJ632 as described above, except 200 µM (each) dATP, dCTP, and dGTP; 135 µM dTTP; and 65 µM digoxigenin-11-dUTP (Roche Molecular Biochemicals) were used. The probe was gel purified using a Quantum Prep Freeze and Squeeze gel extraction kit (Bio-Rad Laboratories). Southern hybridizations were conducted according to the instructions of the manufacturer (Roche Molecular Biochemicals) with standard hybridization buffer at 58°C and detection with CDP-Star.

Plasmid curing of pEA29. Plasmid curing of pEA29 was accomplished by electroporating plasmid pC9 (Table 1) into electroporation-competent E. amylovora strains and subculturing the resulting ampicillin-resistant transformants several times on LB agar plus ampicillin. Individual colonies were screened for the loss of pEA29 using the PCR primers AJ75 and AJ76 (29) found at nucleotide (nt) positions 25800 to 25830 and 25989 to 26064, respectively.

Pathogenicity testing. Immature Bartlett pears (Pyrus communis) were washed for 20 min in tap water and air dried. The fruits were stab inoculated with a sterile needle dipped into a single colony of a 48-h old E. amylovora culture grown on LB agar. The pears were incubated in a humidified chamber for 7 days at room temperature (16, 24).

Oxidative stress disk inhibition assay. E. amylovora strains Ea88, Ea88, MR1, MR1, 2-95, 2-95, RBA4, and RBA4 were grown to an optical density of 0.5 at 600 nm in 5 ml of modified Ceria broth cultures at 27°C. Aliquots (250 µl) of culture were added to 2.5 ml of 50°C MM + Fe soft agar and poured over the surfaces of MM + Fe plates. After 2 h, sterile filter disks impregnated with 100 µl of a 400 mM hydrogen peroxide solution were placed in the center of each plate. Plates were incubated at room temperature for 48 h before zones of inhibition were measured.

Chemotaxis experiments. Both wild-type and plasmid-cured E. amylovora strains Ea88, Ea110, and MR1 were stab inoculated into the surfaces of MM  S plates supplemented with L-aspartic acid, succinic acid, L-malic acid, fumeric acid, choline, or L-serine at a concentration of 0.1 M except for malic acid, whose concentration was 0.001 M. Media were standardized to a pH of 7.2. Plates were poured to uniform thickness. Zones of migration were measured after 9 days at room temperature.

Nucleotide sequence accession numbers. The following nucleotide sequences were entered into GenBank: the complete sequence for plasmid pEA29 from E. amylovora strain Ea88 (accession no. AF264948), a substitution sequence found in pEA29 from raspberry strains MR1 (accession no. AF264949) and 2-95 (accession no. AF264950), and insertion element ISEam1 found in raspberry strain 2-95 (accession no. AF264951).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Nucleotide sequence and gene organization. The complete nucleotide sequence of plasmid pEA29 from strain Ea88 consisted of 28,185 bp and had a G+C content of 50.2%. The first base of the unique BamHI site in pEA29 was designated the first nucleotide position in the sequence. The locations of the BamHI, KpnI, SalI, and PstI restriction sites were in agreement with their locations in the published restriction map of pEA29 from strain Ea7/74 (15); however, an additional PstI site which generated a 192-bp PstI fragment was located at nt positions 6033 to 6225. This fragment may have been undetected in the previous restriction analysis due to its small size. Five of the 11 predicted ClaI restriction sites (nt positions 7839, 10412, 12158, 13457, and 18064) were refractory to digestion by ClaI and overlapped a dam methylase recognition sequence. If methylated, the dam recognition sequence blocks the action of ClaI. This indicates that E. amylovora possesses a methylase similar to that encoded by the dam gene in E. coli (27).

View larger version (30K): [in this window] [in a new window]   FIG. 1.   Potential ORFs at least 375 nt in length and putative genes inferred from sequence data found in plasmid pEA29 from E. amylovora strain Ea88. The unique BamHI site and all PstI restriction sites are shown (positions in base pairs are shown in parentheses). Arrows indicate the positioning of the ORFs. A key to the ORFs is found in Table 2.

View larger version (57K): [in this window] [in a new window]   FIG. 2.   Organization of the origin of replication for plasmid pEA29 in E. amylovora Ea88. (A) Sequence upstream from the PstI site at nt position 26719, through the unique BamHI site, and into the repA gene. Relevant restriction sites within the region are indicated in bold lettering. Solid arrows indicate repeating regions. An AT-rich region is designated by the dotted line. (B) Alignment of the repeating sequences indicated in panel A. Nucleotide patterns conserved across the majority of the repeats are indicated in bold. With the exception of repeat number 4, repeats were either 22 or 11 bp long. The 11-bp repeats are truncated versions of the longer repeats.

View larger version (22K): [in this window] [in a new window]   FIG. 3.   Restriction and physical maps and sequence data for a region of pEA29 from E. amylovora Ea88 with homology to regions in Tn2501 and Tn2502. Triple lines () indicate sequenced regions of Tn2501. (A) Physical and genetic map of Tn2501 compared to those of the partial res site and resolvase (tnpR) gene found on pEA29 (nt 18084 to 18785). Positions for the transposase gene (tnpA), res site, and resolvase gene (tnpR) of Tn2501 are indicated by the hatched box, white box, and latticework box, respectively. Gene orientation is indicated by arrows. Restriction sites are as follows: BamHI (B), BglII (Bg), HindIII (H), SalI (S), SmaI (Sm), and SstI (Ss). Inverted repeats for Tn2501 are indicated by IR-L (left) and IR-R (right). (B) Sequence alignment of the res regions of Tn2501, Tn2502, and pEA29 (nt 18084 to 18188). Dyad symmetry regions consistent with Tn2501 and Tn2502 res sites I, II, and III are boxed, as is the corresponding pEA29 sequence. Areas where the sequence is in agreement with Tn2501 are indicated by asterisks.

Growth and virulence properties of the plasmid-cured strains. Several experiments confirmed that the plasmid-cured strains Ea88, Ea110, MR1, and 2-95 exhibited the expected altered growth phenotype on MM  T (16). All plasmid-cured strains exhibited translucent, raised, viscous colonies on MM  T, while the parental wild-type strains produced creamy, mucoid, fluid colonies. No morphological changes were observed when plasmid-cured strains were grown on LB broth or MM + T. Plasmid-cured E. amylovora strain Ea110 exhibited delayed virulence and a reduction in the severity of disease symptoms compared to those characteristics of the parent strain Ea110 on immature Bartlett pears, as previously reported for other cured strains of E. amylovora (16, 24). In addition, immature pears inoculated with strain Ea110 exhibited decreased amounts of viscous ooze compared to levels in pears inoculated with strain Ea110 (Fig. 4). Rubus strains are host specific; therefore, they were not tested on pear. In the oxidase stress assay, comparison of inhibition zone sizes between plasmid-cured and wild-type strains from apple, pear, and raspberry were not statistically different. Both plasmid-cured and wild-type strains of Ea88, Ea110, and MR1 were attracted to L-aspartic acid, succinic acid, L-malic acid, fumeric acid, and L-serine.

View larger version (100K): [in this window] [in a new window]   FIG. 4.   Surface and cross section of immature pears stab inoculated with wild-type and plasmid-cured E. amylovora strain Ea110 and incubated for 7 days at room temperature under high humidity. (A and C) Pears inoculated with strain Ea110; (B and D) pears inoculated with strain Ea110. Pears inoculated with plasmid-cured strains exhibited less water soaking and ooze production, and symptom expression was delayed and remained localized around the site of inoculation. Pears inoculated with wild-type strains exhibited extensive water soaking and a proliferation of ooze droplets (arrow), and necrosis developed rapidly and extended beyond the inoculation site.

Variation in pEA29. Restriction analysis with enzymes PstI and KpnI was used to analyze pEA29 from strains isolated from tree fruit, ornamental, and Rubus hosts, including three strains with the streptomycin resistance transposon Tn5393 (Table 1). Restriction digests of pEA29 from pear strain Ea88 and from strains of E. amylovora from Indian hawthorne and apple were nearly identical. However, digests of pEA29 from Rubus strain MR1 were distinct from those for Ea88 and the other tree fruit strains (Fig. 5A and 6). PstI digests of pEA29 from all raspberry strains yielded a 5-kb fragment, rather than the expected 4.11- and 0.82-kb PstI fragments, and a 13.8-kb PstI fragment, rather than the 10.86- and 3.59-kb fragments found in Ea88 (Fig. 5B). Sequence analysis of the 5-kb fragment from strain MR1 revealed a point mutation at nt position 2645 that eliminated the PstI restriction site. KpnI restriction analysis of pEA29 from each raspberry strain of E. amylovora revealed a fragment with 629 fewer nucleotides than were in the corresponding 11.73-kb fragment found in Ea88 (Fig. 5B and 6, lanes 2 to 4). Sequence analysis revealed that a 1,890-bp region in pEA29 from strain Ea88 (nt positions 9635 to 11525), which contained a PstI site and ORF 7, was missing in pEA29 from strains MR1 and 2-95 (Fig. 5C). A different 1,264-bp sequence was present in the two raspberry strains. This sequence variation accounted for the 13.8-kb PstI fragment and for the KpnI restriction fragment length polymorphism detected in raspberry strain MR1. It was also observed that the first 51 nucleotides in the 1,264-bp region in MR1 were identical to 51 nucleotides on the reverse strand of pEA29 in Ea88. No ORFs greater than 375 nt were detected in the new sequence, and no homology to existing GenBank entries was detected for several smaller ORFs of 150 to 225 nt (data not shown).

View larger version (27K): [in this window] [in a new window]   FIG. 5.   (A) Comparison of pEA29 PstI and KpnI restriction enzyme maps for E. amylovora strains Ea88 and MR1 from pear and raspberry, respectively. The maps were normalized to the unique BamHI restriction site. (B) Sizes in kilobases of the restriction fragments shown in the restriction maps above. (C) Map of a variable region of pEA29 found in two raspberry strains of E. amylovora. The locations of the 1,264-bp sequence of pEA29 found in raspberry strains MR1 and 2-95 and the corresponding 1,890-bp sequence found in pear strain Ea88 are shown. Shaded bars indicate that the sequences flanking the variable region are identical in the three strains. KpnI sites (K) (nt positions 2011 and 13642) used to confirm the fragment size variation are shown. Dotted arrows indicate the locations of primers AJ484 and AJ485, used to detect length variation in this region. Plasmid pEA29 in strain MR1 is 629 bp smaller than the 28,185-bp plasmid found in Ea88. The substituted sequence starts at bp 9635 and returns to Ea88 (pEA29) homology at nt 11525. A hypothetical ORF (ORF 7; solid arrow) and the restriction sites for PstI (P), SalI (S), and methylated ClaI (Cm) are missing in pEA29 of strains MR1 and 2-95.

View larger version (112K): [in this window] [in a new window]   FIG. 6.   Fragment length polymorphism in KpnI digests of pEA29 from seven strains of E. amylovora. Lanes 1 to 9 contain the 1-kb Ladder Plus, strains Ea88, Ea110, MR1, 2-95, W2, DM22, BCN20, and a  HindIII ladder, respectively. The 11.7-kb fragment found in strains Ea88 and Ea110 (lanes 2 to 3) is slightly smaller in strains MR1 and 2-95 (lanes 4 to 5). The 4.1-kb KpnI fragment in strains Ea88, Ea110, and MR1 (lanes 2 to 4) is a 5.4-kb fragment in strain 2-95 (lane 5) and a 10.8-kb fragment in strains W2 and DM22 (lanes 6 to 7). The 11.7-kb KpnI fragment found in the strains shown in lanes 2 to 7 is an 18.5-kb fragment in strain BCN20 (lane 8). Digests were analyzed on a 1% (wt/vol) agarose gel in 0.5× TBE buffer stained with ethidium bromide.

View larger version (46K): [in this window] [in a new window]   FIG. 7.   Map of ISEam1 found in plasmid pEA29 of E. amylovora strain 2-95. (A) Genetic and physical map of ISEam1 and its location within pEA29. Boxed elements indicate a 100-bp direct repeat found prior to the insertion site (nt position 15692) and at the 3' end of ISEam1. Genes for transposase and a hypothetical (hyp.) ORF are indicated with arrows. Restriction sites are as follows: SphI (Sp), SmaI (Sm), EcoRV (E), PvuII (P), ClaI (C), and DraI (D). (B) Nucleotide sequence of ISEam1. Underlining indicates the position and nucleotide sequence of the transposase gene tnpR. A dotted line indicates the 100-bp direct repeat region. (C) Alignment of the 100-bp direct repeat located in the pEA29 sequence upstream from the insertion site for ISEam1 and the repeat found in ISEam1 at the 3' end.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Plasmid-cured strains of E. amylovora have consistently exhibited reduced virulence (references 16 and 24 and this study). Knowledge of the nucleotide sequence of pEA29 has provided insight into the genetic organization of the plasmid. It is possible that several of the genes found in pEA29 interact such that plasmid-free strains are less adaptive for survival in plant tissues. Another possibility is that one or more of the genes identified in pEA29 are virulence determinants that may contribute to the ability of the bacterium to cause fire blight. The peptide methionine sulfoxide reductase (msrA) is a virulence determinant in the phytobacterium E. chrysanthemi (13). Like msrA mutants of E. chrysanthemi, msrA-minus plasmid-free strains of E. amylovora exhibited reduced virulence on pear slices and reduced systemic invasion of apple and pear seedlings (16, 24). However, msrA-minus strains of E. amylovora did not exhibit increased sensitivity to H₂O₂ as reported for msrA mutants of E. chrysanthemi (13) and E. coli (33). The putative MCP may enable E. amylovora to detect an attractant in the cell's environment, providing a chemotaxis stimulus as reported for E. coli (8), or the choline transport protein (BetT) may help protect the pathogen against osmotic stress (22). The betT gene is typically associated with the glycine betaine pathway. While choline alone provides no osmoprotective properties (49), glycine betaine, the final product of the pathway, has been shown to protect E. coli from osmotic stress (49), and the genes coding for the rest of the glycine betaine pathway may be located on the bacterial chromosome. If present, the pathway could explain the ability of E. amylovora to survive the osmotically stressful environment present in the blossom (38), the typical route of colonization in new fire blight infections. Finally, altered exopolysaccharide synthesis, as observed on MM  T, may explain the delay in symptom expression observed with plasmid-free strains (references 24 and 16 and this study). Although we have identified a few genes that may be involved in virulence, a functional analysis of each gene is needed to establish which are virulence factors in E. amylovora.

Strains of E. amylovora cured of pEA29 have been reported to exhibit thiamine auxotrophy (24) and altered exopolysaccharide production on MM  T (16). In the present study we report the detection of a cluster of three putative genes (thiO, thiG, and thiF) involved in thiamine biosynthesis. In E. coli, thiFGHSI and dxs are necessary for thiazole biosynthesis and thiE is necessary for the biosynthesis of thiamine phosphate from thiazole and pyrimidine (4). It was suggested that thiO has amino acid oxidase activity important in the biosynthesis of thiazole from cysteine (32), thiG is involved in the synthesis of 4-methyl-5-(beta-hydroxyethyl) thiazole, and thiF catalyzes both the adenylation of ThiS and the transfer of sulfur from S-cysteine to an intermediate in thiazole synthesis. In the soil bacterium R. etli four thiamine biosynthetic genes (thiC, thiO, thiG, and thiE) were clustered in plasmid pb (32), and in E. coli a five-gene operon (thiC, thiE, thiF, thiG, and thiH) was detected in the chromosome (52). In both E. coli and Salmonella enterica serovar Typhimurium the genes involved in thiamine biosynthesis, salvage, and transport were scattered throughout the chromosome in three separate operons and four single-gene loci (6, 11, 20, 50, 52, 53). Two thiamine biosynthesis operons and a single gene locus in B. subtilis have been characterized (57). The thiazole biosynthesis cluster in B. subtilis does not contain thiH; instead, B. subtilis contains a gene with similarity to the thiO gene from R. etli (4). To date, 12 genes involved in thiamine biosynthesis in prokaryotes have been identified. Other genes involved in thiamine biosynthesis, including thiC, thiE, and thiS, may be located on the E. amylovora chromosome. It has been suggested that genes for thiamine biosynthesis are located in a 4.4-kb PstI fragment of pEA28 in strain CFBP1430 (24). When we examined two CFBP1430 derivatives by PCR, thiO, thiG, and thiF were detected in the same locations on the plasmid as in strain Ea88 (data not shown). Since Laurent et al. (24) were able to restore thiamine prototrophy to a Thi strain of E. coli with a clone containing the 4.4-kb PstI fragment of pEA28, there is a possibility that pEA29 in strain CFBP1430 contains additional genes for thiamine biosyntheses.

We confirmed earlier reports that the replication region for plasmid pEA29 was located on the 4.4-kb PstI fragment and that introducing a plasmid carrying this fragment into E. amylovora resulted in the loss of pEA29 from E. amylovora by incompatibility (references 16 and 24 and this study). Sequence analysis of the replication region of pEA29 revealed a cluster of direct repeats termed iterons (12); iterons are binding sites for RepA proteins and play a role in the control of plasmid copy number (18, 34). RepA was identified based on similarity to replication proteins from other plasmids and its location adjacent to the iterons. The configuration of the direct and partial repeats and the AT-rich region is consistent with a theta-type plasmid replication mechanism (12). However, the parA and parB loci were located some distance from the repA locus. Plasmid pEA29 is extremely stable in E. amylovora, and the parA and parB genes likely contribute to the ubiquitous occurrence of this plasmid by providing active DNA partitioning functions during cell division. As expected, because pEA29 is nontransmissible, the nucleotide sequence was devoid of genes involved in conjugal transfer and plasmid mobilization.

Other putative genes found on pEA29 may involve regulation of gene transcription or substrate utilization. The H-NS DNA binding protein coded for by the hns gene has been implicated in several functions. H-NS can act as either a transcription activator or a repressor and may be important for DNA compaction (46, 47) or alteration of DNA topology by generating negative supercoiling (36, 51). The protein preferentially binds to curved strands of DNA and has no known sequence specificity. Among the genes regulated by H-NS proteins are the proV, proW, and proX osmoregulation genes found in the proU operon of E. coli (26) and the virF virulence gene in Shigella spp. and E. coli (14). A characteristic shared by genes dependent on regulation by H-NS is their induction by extreme environmental stresses such as osmotic or cold shock. It is unknown at this time which genes are regulated by the H-NS protein in E. amylovora or whether the protein acts as an activator or repressor for that gene. A second transcriptional regulator found on pEA29 is the LysR-type transcriptional regulator coded for by trlA. LysR-type transcriptional regulators are one of the most common types of autoregulated positive transcriptional regulators and are typically divergently transcribed from a promoter that is very close to and often overlaps the promoter of their target gene (42). The closest divergently transcribed gene relative to trlA on pEA29 is the aldD gene, which codes for a putative aldehyde dehydrogenase.

In microbes, aldehyde dehydrogenases have narrow inducer and substrate specificities and enhance the flexibility of a microbe's degradative, fermentative, and biosynthetic pathways. In some instances, aldehyde dehydrogenases may provide a key oxidative link between primary and secondary metabolites. In the glycine betaine osmoregulation pathway, the soluble NAD-dependent betaine aldehyde dehydrogenase found in the pathway converts glycinebetaine aldehyde to glycine betaine in the final step (23). Although not the strongest match to the aldD gene in E. amylovora, the 490-amino-acid betaine aldehyde dehydrogenase protein found in E. coli (accession no. M77739) did share 31% identity (140 of 448 amino acids) and 48% similarity (220 of 448 amino acids) with aldD. Further examination of the aldD gene may reveal the target substrate of the predicted product. The predicted product of ORF 7 has no matches in GenBank. However, because of the large size of the predicted product, some comment is warranted. ORF 7 was found in all apple, pear, and Indian hawthorn strains examined in this study, but it was not detected in any of the Rubus strains. It is possible that the predicted product of ORF 7 is necessary for infection of tree fruit and ornamental hosts. This might explain why Rubus strains are not able to infect pears in pathogenicity testing. Alternatively, it is possible that the product of ORF 7 is not conserved in Rubus strains because it has a minor function. Further study is needed to elucidate the function of the hypothetical protein.

A link was observed between the organization of pEA29 and the strain's original host plant. Derivatives of pEA29 found in strains isolated from apple, pear, and Indian hawthorn were homogenous; however, significant sequence variation was found to exist between strains from plants in the subfamily Pomoideae and those from Rubus plants. Derivatives of pEA29 found in strains isolated from Rubus exhibited restriction fragment length polymorphism when they were digested with KpnI and PstI. All Rubus derivatives lacked two PstI sites and were missing 1,890 bp of sequence found in tree fruit strains. Rubus strains contained an entirely different 1,264-bp region. Since the Rubus strains that we examined came from different geographic regions, Michigan in the United States and Nova Scotia in Canada, this variation in sequence may be a common feature of pEA29 found in strains of E. amylovora from Rubus.

The detection of transposon Tn5393 on pEA29 is interesting with respect to circulation of antibiotic resistance genes in the environment. Tn5393 is found on the conjugative plasmid pEA34 in E. amylovora and Pantoea agglomerans and carries the strA and strB genes for streptomycin resistance (10, 19). The detection of Tn5393 on pEA29 is a recent event (19, 28) and represents further evolution of streptomycin resistance in E. amylovora in Michigan apple orchards. The insertion of Tn5393 into different sites on pEA29 in field strains indicates that resistance in these strains involved separate insertion events. Since pEA29 is nontransferable, strains containing pEA29::Tn5393 would not be an important source for further dissemination of the strA-strB gene throughout microbial communities.

There were remnants of several insertion sequences detected on pEA29 during sequencing that resembled insertion elements identified previously in other unrelated bacteria. A remnant of Tn2501 and direct repeats found in IS911 were detected in all derivatives of pEA29. Residual traces of several genes, including a remnant from a parA gene located downstream from the repeats located in the 1.1-kb PstI fragment, were found. Kim and Geider (21) reported finding a 108-amino-acid ORF with similarity to ParA from Agrobacterium tumefaciens at this location on the plasmid. We located a 65-amino-acid ORF at bp 26,241 to 26,438 exhibiting the highest similarity to the Mycobacterium celatum parA gene (accession no. AAD42964.1) and had no BLAST hit to the 222-amino-acid ParA protein from Agrobacterium tumefaciens (accession no. S07280). The presence of a partial parA gene may explain the occurrence of the 8-bp repeats found consistently in this region. Remnants of a single gene which once coded for an outer membrane protein belonging to the peptidase family S18 (omptin family) were identified in gaps between some of the large ORFs on pEA29 (data not shown). The closest GenBank match by BLAST X homology was the plasminogen activator gene in Yersinia pestis (accession no. P17811) coding for a coagulase or fibrinolysin precursor. These remnants may provide important clues to the evolution of pEA29 and E. amylovora. ISEam1 was found adjacent to an IS911 direct repeat in pEA29. This region of the plasmid may be highly variable because it is an attractive site for insertion. Very few insertion sequence-like elements have been detected in E. amylovora previously, and they may be important for tracing local outbreaks of disease.