INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The willow (Salix spp.), together with the poplar (Populus spp.), is a deciduous tree of the family Salicaceae. Most of the Salix species thrive in a humid environment and are often used in forestry. Material from willows is used for several economic applications, e.g., the wood and twigs of willows are used to produce cricket bats (1) and basketwork (6).

The watermark disease is a vascular wilting disease that causes great losses among willow populations. The bacterium Brenneria salicis (10), formerly Erwinia salicis, is the causal agent of this disease and occurs mainly in the xylem vessels of the host plant. Infected willows show wilted, dried, brown-colored leaves and a watery, transparent color of the wood. It is commonly accepted that the spread of the watermark disease occurs by the transportation of plant material (cuttings) infected with B. salicis. Infected cuttings do not show any internal or external symptoms of the watermark disease. In order to minimize the distribution of this disease, it is of major importance to detect low concentrations of latent B. salicis bacteria in an early stage by means of a sensitive and specific molecular test. The current detection method applying immunofluorescence (enzyme-linked immunosorbent assay) often yields false positives as a result of cross-reactions but also yields false negatives because low densities of bacteria (<10⁴/ml) are not detected (5). Latent infections characterized by low densities of B. salicis require a more sensitive detection method.

PCR has proven to be successful in detecting plant-pathogenic bacteria as well as fungi (2, 7, 8, 11, 17, 18, 21, 22, 30). A PCR-based identification method for B. salicis was developed based on four specific base positions of the 16S ribosomal DNA (rDNA) sequence (9): A at Escherichia coli numbering position (3) 114, G at 217, G at 583, and A at 625. Species-specific primers complementary to these sites were developed. In the present study, the specificity of the PCR was tested with a larger collection of B. salicis strains and related plant-associated bacteria. In order to establish the genomic similarity of the B. salicis strains studied and in the hope that the origins of the various populations could be determined, amplification fragment length polymorphism (AFLP) has been used to fingerprint the genomes.

The PCR detection method for B. salicis was adjusted to be applied to the vascular fluid of willows, and a collection of willow clones was tested.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and growth. The bacterial strains used were obtained from the Belgian Coordinated Collections of Microorganisms, Laboratory of Microbiology, University of Ghent, Belgium (BCCM/LMG). Three ICMP strains were kindly provided by A. Spiers from the Horticulture and Food Research Institute, Palmerston North, New Zealand (Table 1). The following non-B. salicis phytobacteria were also included in our study: ICMP 12481^T (Brenneria alni), LMG 2694^T (Brenneria nigrifluens), LMG 2542^T (Brenneria paradisiaca), LMG 2709^T (Brenneria rubrifaciens), LMG 2724^T (Brenneria quercina), LMG 10245^T (Enterobacter nimipressuralis), LMG 2693^T (Enterobacter cancerogenus), LMG 2683^T (Enterobacter dissolvens), LMG 2715^T (Pantoea stewartii subsp. stewartii), LMG 2632^T (Pantoea stewartii subsp. indologenes), LMG 2665^T (Pantoea ananatis), LMG 2565 (Pantoea agglomerans), LMG 2660 (P. agglomerans), LMG 2906^T (Erwinia tracheiphila), LMG 7034 (Erwinia psidii), LMG 2708^T (Erwinia mallotivora), LMG 2024^T (Erwinia amylovora), LMG 11254^T (Erwinia persicinus), LMG 2688^T (Erwinia rhapontici), LMG 2404^T (Pectobacterium carotovorum subsp. carotovorum), LMG 2386^T Pectobacterium carotovorum subsp. atrosepticum), LMG 2466^T (Pectobacterium carotovorum subsp. betavasculorum), ICMP 9121^T (Pectobacterium carotovorum subsp. wasabiae), LMG 17566^T (Pectobacterium carotovorum subsp. odoriferum), LMG 17936^T (Pectobacterium cacticidum), LMG 2804^T (Pectobacterium chrysanthemi), and LMG 2657^T (Pectobacterium cypripedii).

Total bacterial DNA preparations. Genomic DNAs were prepared according to the method of Pitcher et al. (24).

PCR amplifications. PCR with primers Es1A (5'-GCGGCGGACGGGTGAGTAAA-3') and Es4B (5'-CTAGCCTGTCAGTTTTGAATGCT-3') was performed in a Genius thermal cycler (Techne, Cambridge, United Kingdom), using the following program: initial denaturation for 5 min at 95°C, followed by 35 cycles of 95°C for 25 s, 68°C for 25 s, and 72°C for 40 s, and a final extension at 72°C for 10 min. All reaction mixtures had a final volume of 20 µl and contained PCR buffer (1.5 mM MgCl₂, 50 mM KCl, 10 mM Tris [pH 8.3]), 0.2 mM (each) deoxynucleoside triphosphate, 0.45 µM (each) primer, 0.8 U of Taq polymerase (Eurotaq; Eurogentec), and 5 µl of sample. The PCR products were analyzed by running the entire reaction mixture in a Tris-acetate agarose gel (2%), staining it with ethidium bromide, and visualizing it under UV light.

PCR efficiency in vascular fluid. PCR efficiency in plant material was tested by using vascular fluid to which B. salicis cells had been added. We found that the production of the 553-bp band was inhibited by components of the vascular fluid. The following buffers and components, separate or combined, were used in an attempt to neutralize the PCR-inhibiting factors of the vascular fluid: SCP (0.01% disodium succinate, 0.01% trisodium citrate, 0.15% K₂HPO₄, 0.01% KH₂PO₄ [pH 7] [21]), SCPAP (SCP, 0.02 M sodium ascorbate, 5% polyvinylpolypyrrolidone [PVPP] [21]), extraction buffer 1 (0.01 M phosphate buffer [pH 7.2], 0.14 M NaCl, 0.1% Tween 20, 2.5% polyvinylpyrrolidone [PVP] [20]), extraction buffer 2 (extraction buffer 1 plus 2.5% PVPP [20]), bovine serum albumin, and proteinase K. None of them was able to neutralize the inhibiting factor(s). Performing a bacterial DNA extraction on the vascular fluid prior to PCR yielded the 553-bp band.

Bacterial DNA isolation from vascular fluid. Vascular fluid from willows was obtained by squeezing a cutting wiped with ethanol in a bench vise and collecting the xylem fluid in a sterile tube. One hundred microliters of this fluid was subjected to the method of Pitcher et al. (24) for DNA preparation. The DNA obtained was dissolved in 5 µl of T0.1E buffer (10 mM Tris, 0.1 mM EDTA, pH 8).

PCR sensitivity tests. To estimate the sensitivity of PCR detection in a pure B. salicis suspension, decimal B. salicis dilutions were prepared in phosphate-buffered saline. These dilutions were used for PCR and plated on GYCA medium to determine the B. salicis titer. For PCR, 1 ml of each dilution was centrifuged (10,000 × g, 10 min) and the bacterial pellet was resuspended in 1 ml of water. Five-microliter aliquots were then heated for 5 min at 95°C, cooled on ice, and tested in PCR.

AFLP. AFLP, a DNA-fingerprinting technique based on the selective amplification of genomic restriction fragments, was performed according to the method of Janssen et al. (13). The combination of EcoRI (Pharmacia, Uppsala, Sweden) and MseI (New England Biolabs) as restriction enzymes and the primer combination E01-M02 (E stands for EcoRI, and M stands for MseI; 0 stands for no selective base at the 3' end, 1 stands for an adenine, and 2 stands for a cytosine) revealed AFLP patterns composed of some 30 to 60 clearly distinguishable bands on average. The resulting profiles were further processed with the GelCompar software package (Applied Maths bvba, Kortrijk, Belgium). Concurrent amplification of large numbers of fragments typically yielded bands with unequal intensities within a single pattern. This fact together with the large numbers of bands made it particularly difficult to assign discrete band positions to the patterns, and therefore, in order to avoid subjective interpretations of band positions, the Pearson product-moment correlation coefficient was applied to measure the similarity between normalized densitometric profiles.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Four species-specific nucleotide positions on the 16S rRNA genes of B. salicis strains were used for the development of four oligonucleotide primers. They could be combined in four different pairs to amplify four specific fragments of the 16S rDNA of B. salicis strains (9). One of these primer combinations ES1a-ES4b, resulted in a 553-bp PCR amplification product. This primer pair was used in the present study to detect B. salicis strains in the vascular fluid of willows.

PCR specificity. To check the specificity of the Es1a-Es4b primer combination under the specified PCR conditions, 103 phytobacterial strains were used as targets, comprising 81 Brenneria, 6 Erwinia, 8 Pectobacterium, 5 Pantoea, and 3 Enterobacter strains. Among the bacteria from the genus Brenneria, 76 B. salicis strains were tested (Table 1). With the primers Es1a and Es4b, B. salicis strains with different geographical origins all produced a PCR band of the same length, except LMG 5216 and LMG 5217, which did not produce any PCR product at all. Phenotypic analyses by API 20E and API ZYM profiles (data not shown) and genotypic analysis by AFLP profiling (Fig. 1) revealed that LMG 5216 and LMG 5217 are not B. salicis strains. The described PCR differentiated all typical B. salicis bacteria from the other phytobacteria. None of the non-B. salicis phytobacteria produced a PCR product.

View larger version (129K): [in this window] [in a new window]   FIG. 1.   AFLP patterns of B. salicis strains and related Brenneria, Pectobacterium (Pectob.), and Erwinia strains. Aberrant bands within the B. salicis profiles are indicated with arrows. Pectob. c., Pectobacterium carotovorum subspecies. Braces around species names indicate a doubtful taxonomic position of the strain.

Detection sensitivity in pure culture. We were able to observe the production of a 553-bp band on agarose gel with dilutions of a pure B. salicis culture containing 40 CFU ml¹. This sensitivity corresponded with an average detection limit of 0.2 CFU of B. salicis per PCR. It was possible to confirm that the PCR product was from amplification of the target DNA and not of any other fragment of the genome by sequencing the amplification product. The nucleotide sequence of the amplification product was identical to the nucleotide sequence of the corresponding part of the 16S rRNA gene of B. salicis.

AFLP genotyping of B. salicis and related Brenneria, Pectobacterium, and Erwinia strains. Figure 1 shows normalized AFLP patterns for 86 strains sorted according to an unweighted pair group method with averages dendrogram based upon the product-moment correlations between entire densitometric profiles.

PCR detection in vascular fluid of willows. We selected three willow clones (89/036.73, 89/036.68, and 89/036.62) which had been artificially infected with B. salicis in October 1993 by stem incision and from which the bacterium B. salicis could be isolated from the vascular fluid by plating in 1997. These plants showed external symptoms of the watermark disease at the moment that vascular fluid was taken. With the described procedure, i.e., DNA extraction followed by PCR amplification, positive PCR products were detected in the vascular fluid from these willow clones (Fig. 2). In none of these plant tests was a nonspecific PCR signal produced by plant material or other plant-associated organisms. The vascular fluid of willow clone 89/036.62 was filtered through a sterile bacterial filter (0.22-µm pore size) in order to generate a negative control. This filtrate did not yield any PCR product after the molecular test (Fig. 3).

View larger version (72K): [in this window] [in a new window]   FIG. 2.   PCR product, formed with primers Es1a and Es4b, of vascular fluid of infected willow clones 89/036.73 (lane 1), 89/036.68 (lane 2), and 89/036.62 (lane 3). Lane 4, DNA molecular size marker VIII (Boehringer Mannheim).

View larger version (59K): [in this window] [in a new window]   FIG. 3.   PCR product, formed with primers ES1a and ES4b, of vascular fluid of the infected willow clone 89/036.62 (lane 2) and a filtrate (filter pore size, 0.22 µm) of the same extract (lane 3). Lane 1, DNA molecular size marker VIII (Boehringer Mannheim).

Detection limit in vascular fluid of willows. We were able to observe the production of a 553-bp band on agarose gel with dilutions of vascular fluid containing 20 CFU ml¹ (Fig. 4). This sensitivity corresponded to detection of an average of 0.1 B. salicis cells per PCR. The vascular fluid samples of the willows were not fresh but had been collected and kept frozen for a period of time; therefore, an unknown portion of the B. salicis population might have died, resulting in underestimated CFU values in the sensitivity test. Positive PCRs found in a few dilutions where no colonies were recovered may have resulted from DNA templates of dead B. salicis cells. Nevertheless, the detection limits of the proposed procedure for B. salicis in willows are lower than those found for other bacteria in similar studies (21, 22, 30).

View larger version (40K): [in this window] [in a new window]   FIG. 4.   Sensitivity test of detection on gel of the PCR products obtained with the primer pair ES1a-ES4b and with diluted vascular fluid of willows containing approximately 105, 104, 2.8 × 103, 3.8 × 102, 80, 20, 2, and <1 CFU of B. salicis ml1 (lanes 2 through 12, respectively). Lanes 1 and 13, DNA molecular size marker VIII (Boehringer Mannheim).

Field study. In addition to the three clones tested earlier, the vascular fluid of 31 artificially infected and 13 nonartificially infected willow clones and three poplar clones were tested with the proposed molecular detection method (Tables 2 and 3). All artificially infected willow clones were found to contain B. salicis bacteria in their vascular fluid. Of the nonartificially infected clones, two were found to contain B. salicis, although they did not show any external or internal symptoms of the watermark disease. One of the three poplar clones was also found to contain B. salicis cells in its vascular fluid.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Separately, the nucleotide sequences of the primers Es1a and Es4b showed similarities up to 100% with known sequences in EMBL and GenBank data banks. However, the nucleotide sequences of the combined primers did not match with any sequence in the data banks other than that of B. salicis, indicating that this primer pair is specific for this species. In most bacteria, the rDNA operon is present in several copies per genome (28), implying a relatively high and constant sensitivity of the rDNA PCR (18, 27). The expected 553-bp fragment was amplified only from B. salicis strains. No DNA fragments were amplified from the 27 related phytobacterial strains (Table 1).

AFLP genotyping of B. salicis strains revealed that the bacterial genomes of this pathogen in Northwest Europe are very similar. The very similar AFLP patterns of the B. salicis strains investigated indicate that the spread of the watermark disease in Belgium, the United Kingdom, and The Netherlands is most probably due to the import and export of infected cuttings. The banding pattern of one European strain, LMG 6381, is slightly aberrant, which is probably related to its origin, as this strain was not isolated from Salix spp. but from Populus robusta. The three New Zealand isolates included in our study were identified as B. salicis by PCR analysis with the primer pair Es1a-Es4b. Their AFLP pattern, however, differed from those of the European isolates, suggesting that these strains were not imported from Europe as was initially thought. Of the 71 European strains, 66 had very similar AFLP patterns, with the exception of five possible polymorphic bands (Fig. 1). Five of the 15 isolates from the United Kingdom, 10 of the 16 Dutch isolates, and 4 of the 35 Belgian isolates showed at least one of these polymorphic bands; the geographical origins of the strains could not be correlated with one or more of the polymorphic bands. In an attempt to correlate AFLP fingerprints with the virulence of B. salicis, we were unable to correlate the nonconforming bands with the outcomes of hypersensitivity tests performed on tobacco leaves (data not shown). This lack of correlation can be explained by the difference in composition of tobacco leaves compared to the xylem vessels of a willow. Infection trials on willows are more suitable for testing virulence, but they are more time consuming, as it can take 5 to 10 years before disease symptoms are detectable. Such infection tests are currently being carried out, but results are not yet available.

PCR efficiency in vascular fluid could only be obtained after a DNA extraction. This is probably due to compounds present in the vascular fluid that inhibit the PCR. Similar problems have been encountered in other plant-associated PCR studies (4, 15, 21, 31). The inhibition might be due to tannins, humic acids, polysaccharides, or phenolic compounds, which are suspected to affect enzymatic activities and to bind to RNA and DNA upon cell lysis (14, 29). Several procedures have been proposed to overcome these problems. The use of PVP and PVPP improved the PCR results in other studies (14, 20, 23, 32). Other buffers proved to be useful as well (16, 21). More recently, a combined biological and enzymatic amplification (BIO-PCR [26]), magnetic capture-hybridization-PCR (12), and immunocapture-PCR (19) were developed to circumvent inhibiting compounds. After having tested several buffers, we found that a very sensitive PCR detection could be achieved after a DNA extraction, a procedure that has also been successful in other PCR detection methods for plant pathogens (4, 18, 25).

Five strains showed a completely different AFLP band pattern and could be excluded from the species B. salicis: LMG 2947, LMG 6151, LMG 18287, LMG 5217, and LMG 5216 (Fig. 1). We mentioned before that the last two strains reacted negatively in our PCR test for B. salicis. The AFLP pattern of LMG 5217 correlated very well (82.6%) with that of the type strain of E. amylovora. Strain LMG 2947 correlated at about the same level (82.2%) with the type strain of B. rubrifaciens. Strains LMG 5217 and LMG 2947 can therefore probably be identified as E. amylovora and B. rubrifaciens, respectively. Among the other three strains that are aberrant in AFLP, strain LMG 6151 showed API 20E and API ZYM profiles different from those of the majority of the B. salicis strains (data not shown). Although our results indicate that these three strains do not belong to B. salicis either, the available data do not allow us to determine their exact species identities.

When applied to the vascular fluid of 34 willow clones, which had all been artificially infected with B. salicis a few years ago (Table 2), our detection method revealed the presence of B. salicis cells in all 34 samples. However, only 13 of the 34 clones showed external and/or internal symptoms of the watermark disease at the time of sampling. Thirteen nonartificially infected willow clones were tested as well. Two (82.237 and 82.276) of these clones were found to be positive for the presence of B. salicis (Table 3), which indicates a natural infection before or after planting. Clones 82.237 and 82.276, therefore, should not be used for multiplication. B. salicis was also found in the vascular fluid of one of three poplars that were planted near infected willows. The infected poplar did not show any symptoms of disease. As Populus spp. also belong to the family Salicaceae, this bacterium might well occur in the vascular fluid of poplars without exhibiting any form of pathogenicity.

The PCR assay described here allows the detection of the watermark disease pathogen in the vascular fluid of willows in an early stage and opens new perspectives for breeding and epidemiology.