Skip to main content
NCBI home page
Molecular Plant Pathology logoLink to Molecular Plant Pathology
. 2008 May 23;9(5):705–717. doi: 10.1111/j.1364-3703.2008.00482.x

Distinguishing bacterial pathogens of potato using a genome‐wide microarray approach

M AITTAMAA 1, P SOMERVUO 1, M PIRHONEN 1, L MATTINEN 1, R NISSINEN 1, P AUVINEN 2, J P T VALKONEN 1,✉
PMCID: PMC6640225  PMID: 19018999

SUMMARY

A set of 9676 probes was designed for the most harmful bacterial pathogens of potato and tested in a microarray format. Gene‐specific probes could be designed for all genes of Pectobacterium atrosepticum, c. 50% of the genes of Streptomyces scabies and c. 30% of the genes of Clavibacter michiganensis ssp. sepedonicus utilizing the whole‐genome sequence information available. For Streptomyces turgidiscabies, 226 probes were designed according to the sequences of a pathogenicity island containing important virulence genes. In addition, probes were designed for the virulence‐associated nip (necrosis‐inducing protein) genes of P. atrosepticum, P. carotovorum and Dickeya dadantii and for the intergenic spacer (IGS) sequences of the 16S–23S rRNA gene region. Ralstonia solanacearum was not included in the study, because it is a quarantine organism and is not presently found in Finland, but a few probes were also designed for this species. The probes contained on average 40 target‐specific nucleotides and were synthesized on the array in situ, organized as eight sub‐arrays with an identical set of probes which could be used for hybridization with different samples. All bacteria were readily distinguished using a single channel system for signal detection. Nearly all of the c. 1000 probes designed for C. michiganensis ssp. sepedonicus, c. 50% and 40% of the c. 4000 probes designed for the genes of S. scabies and P. atrosepticum, respectively, and over 100 probes for S. turgidiscabies showed significant signals only with the respective species. P. atrosepticum, P. carotovorum and Dickeya strains were all detected with 110 common probes. By contrast, the strains of these species were found to differ in their signal profiles. Probes targeting the IGS region and nip genes could be used to place strains of Dickeya to two groups, which correlated with differences in virulence. Taken together, the approach of using a custom‐designed, genome‐wide microarray provided a robust means for distinguishing the bacterial pathogens of potato.

INTRODUCTION

The DNA microarray technology was originally developed for parallel monitoring of gene expression. In this function, it has served for investigation of plant pathogenic fungi (e.g. Takano et al., 2003; Ulrich et al., 2006) and bacteria (e.g. Okinaka et al., 2002; de Souza et al., 2003; Zwiesler‐Vollick et al., 2002). However, microarrays are also a powerful tool for genomic comparisons and analysis of microbial populations and their diversity given that thousands of probes can be used simultaneously to describe and distinguish the microbes tested (Palmer et al., 2006). Therefore, microarrays are foreseen to provide a basis for novel and important applications in analysis of plant pathogens (reviewed in Mumford et al., 2006). There are already reports on detection and/or analysis of genomic differences between species or strains of plant pathogenic microbes (e.g. Deyong et al., 2005; Szemes et al., 2005) and some concerning nematodes, viruses and fungi pathogenic to potato (Abdullahi et al., 2005; Boonham et al., 2003, 2007; Bystricka et al., 2003; Françoisa et al., 2006). However, rather few studies have used microarrays to analyse genomic differences in plant pathogenic bacteria. Koide et al. (2004) used probes targeting 94.5% of the coding sequences of Xylella fastidiosa to compare two strains of this pathogen and successfully identified genes important for virulence. Similarly, Sarkar et al. (2006) identified individual genes and gene profiles that were associated with strains of Pseudomonas syringae isolated from different host species. Guidot et al. (2007) compared the gene content of 18 strains of Ralstonia solanacearum and found c. 50% of the genes to be conserved, including many pathogenicity‐associated genes. The results of these studies imply that microarrays can be used efficiently to detect genomic differences between bacterial strains. However, few studies have used a microarray approach to detect and distinguish a number of different bacterial species typically associated with a crop species.

Potatoes grown in temperate or cool climates are affected by many harmful bacterial pathogens that are widely distributed in most of the production areas (van der Wolf and De Boer, 2007). Pectobacterium carotovorum and P. atrosepticum (former Erwinia carotovora ssp. carotovora and E. carotovora ssp. atroseptica, respectively) (Gardan et al., 2003) cause blackleg and soft rot, as do some species in the genus Dickeya (former Erwinia chrysanthemi) (Samson et al., 2005). Streptomyces scabies (Lambert and Loria, 1989) and S. turgidiscabies (Miyajima et al., 1998) cause ‘common scab’ symptoms on potato tubers. Although S. turgidiscabies was only recently described, it may be almost as widely distributed as S. scabies (Naito et al., 2004). In northern Scandinavia, these two species often co‐occur in the same scab lesions (Lehtonen et al., 2004). The potato ring rot pathogen, Clavibacter michiganensis ssp. sepedonicus (Carlson and Vidaver, 1982), is under a special control scheme in many countries and is classified as a quarantine organism in Europe (EU Commission directive 95/4/EC and EU Council directive 98/57/EC) and as a regulated non‐quarantine pathogen in North America (NAPPO potato standard #3; http://NAPPO.org). These bacterial pathogens of potato occur also in Finland and were considered as a suitable target group for this study that aimed to develop a more generally applicable scheme for the classification of bacterial pathogens in an important crop species and, at the same time, to screen thousands of probes for their species specificity using a microarray format. However, R. solanacearum, a potato pathogen typical of tropical or sub‐tropical areas (van der Wolf and De Boer, 2007), is a quarantine pathogen that does not occur in Finland and was therefore not included in experiments, except that a few probes were designed for it for possible future purposes.

Differentiation of C. michiganensis subsp. sepedonicus, R. solanacearum, P. atrosepticum, P. carotovorum and some pectolytic Dickeya strains has been achieved using a ‘macroarray’, i.e. a nylon membrane onto which random oligonucleotides and probes targeting the 16S–23S rRNA intergenic spacer (IGS) sequences of the bacteria were spotted (Fessehaie et al., 2003). Random oligonucleotides and probes for the 16S rRNA gene and the IGS region are abundant in public databases and have been used to analyse many diverse bacteria also in the microarray format (e.g. Belosludtsev et al., 2004; Günther et al., 2006; Loy et al., 2002; Sanguin et al., 2006). However, owing to their high similarity with closely related species they alone may not always be sufficient for species identification with probes used for hybridization. When available, other genes including the topoisomerase genes (Roth et al., 2004) and genes known to affect the characteristic phenotype of target bacteria (Bodrossy et al., 2003) have been used. However, covering a broader area of the genome, or ideally the whole genome, would maximize the possibility of finding larger numbers of discriminating sequences and thus probably provide a more robust differentiation of the bacterial species.

Accumulation of whole‐genome sequence data on bacterial species and strains is accelerating owing to rapid developments in DNA sequencing technologies and bioinformatics and consequently reduced time and cost. At the same time, the latest microarray technologies allow preparation of economically more feasible arrays in which the probes are synthesized on the array in situ in any numbers required (e.g. Avarre et al., 2007; Leiske et al., 2006; Petersen et al., 2007). Hence, the sequence data available from the target organisms can be fully utilized. Array design is flexible and custom‐based. Many individual samples can be simultaneously hybridized on an array consisting of several sub‐arrays, each containing the same set of probes and surrounded by a ridge keeping the hybridization solution on the defined area. These custom‐based high‐density oligonucleotide arrays can be used for various approaches, including comparative genomic hybridization (CGH) assays (Wolber et al., 2006) and also identification of microbes from complex samples (Palmer et al., 2006, 2007).

The aim of the present study was to design a microarray for distinguishing the main bacterial pathogens infecting potatoes in Finland and to utilize whole‐genome sequence data for designing probes for this purpose. In addition, detection of sequences of a ‘pathogenicity island’ containing important virulence genes in S. turgidiscabies and S. scabies (Kers et al., 2005) and the nip (necrosis‐inducing protein) genes of Dickeya and Pectobacter (Mattinen et al., 2004) associated with pathogenicity was of interest. Probes for the 16S–23S rRNA IGS sequences were added because they are commonly used (e.g. Günther et al., 2006) and available from many bacterial strains. Evaluating the large number of probes designed in this study would have been very difficult with other methods than using the microarray format. Thus, total DNA extracted from pure cultures of bacteria was used to test the probes on custom‐based microarrays.

The study also aimed to develop a scheme according to which a sample could be classified using a single channel of fluorescence (Cy3 or Cy5) rather than the ratio of fluorescence intensities between the unknown sample and the known reference pool (Palmer et al., 2006). Although in both approaches the performance of each probe must be validated with known samples, the benefit of a single‐channel system is that the number of samples which can be classified on the same array equals the number of dyes available. Furthermore, maintenance of a reference pool is not required. The results indicated that the genome‐wide approach provides a feasible and robust means for identification of many new species‐specific probes and for distinguishing between bacterial species.

RESULTS AND DISCUSSION

The microarray designed in this study contained 9676 probes. The whole‐genome sequences of P. atrosepticum (strain SCRI1043), S. scabies (strain 87.22) and C. michiganensis ssp. sepedonicus (strain ATCC 33113) were available and utilized for designing probes to the coding regions (further referred to as genes). All probes were designed taking all available sequence data of the different species to be analysed into consideration, so as to maximize the target‐specificity of the probes (see Experimental procedures for further details). The total number of gene‐specific probes was 4434 for P. atrosepticum (one for each gene), 3894 for S. scabies (for c. 50% of the genes) and 997 for C. michiganensis ssp. sepedonicus (for c. 30% of the genes) (Table 1). The sequence of the genomic pathogenicity island (PAI) of S. turgidiscabies (strain Car8) containing 31 genes (Kers et al., 2005) was used to design 226 probes for the genes and intergenic areas in this genomic region. The whole‐genome sequences of Dickeya spp. and P. carotovorum were not available but probes were designed based on the sequences of nip genes needed for virulence on potato (Mattinen et al., 2004). In addition, the 16S–23S rRNA IGS sequences were utilized for probe design (Table 1). However, the IGS sequences of P. atrosepticum and P. carotovorum were highly similar, and therefore no differentiating probes for P. atrosepticum and only two probes specific to some P. carotovorum strains could be designed based on the IGS area. These two probes were later found not to detect any bacterial strains tested in this study (data not shown). Each probe contained 40 target‐specific nucleotides on average because this probe length provided a good compromise between specificity and sensitivity, as shown in previous studies (e.g. Palmer et al., 2006). Furthermore, probes with three different lengths (30, 40 or 50 nt; ten probes of each length) were designed using the nip gene sequences of P. atrosepticum, P. carotovorum and D. dadantii. Results showed that the increasing probe length increased signal intensity, as expected (see below), but signals with all probes regardless of their length were sufficiently high for detection. Probes were designed to have a G+C content of 45–55 mol% and the melting temperature (T m) within the range of 82–90 ºC to make hybridization conditions similar to all probes. Eight sub‐arrays were synthesized on the same glass slide, which allowed us to test 16 samples simultaneously (two samples per sub‐array labelled with a different dye, Cy3 or Cy5).

Table 1.

Percentage of probes that recognized four bacterial species tested by microarray analysis.

Probe group† No. probes Bacterial species*
Pat Sca Stu Cms
Pat 4434  99.4 ± 1.3  0.4 ± 0.3   0.2 ± 0.3   0
Sca 3894   0.1 ± 0.1 90.0 ± 4.0  32.7 ± 7.0   1.5 ± 0.2
PAI Stu  226   0 53.8 ± 0.3  99.3 ± 0.9   0.2 ± 0.3
Cms  997   0.0 ± 0.1  7.0 ± 0.3   5.6 ± 2.9  99.1 ± 1.2
nip Pat   23 100.0 ± 0.0  0   0   0
nip Pca   26   0.7 ± 2.1  0   0   0
nip Dic   30   0  0   0   0
IGS Dic    6   2.1 ± 8.3  0   0   0
IGS Sca   14   0.4 ± 1.8 85.7 ± 0.0  50.0 ± 10.1   0
IGS Stu    4   0  0 100.0 ± 0.0   0
IGS Cms    8   0  0   0 100.0 ± 0.0
IGS Rso   12   0  0   0   0
Open in a new tab

The probes were designed based on sequences of different origins (probe group). The numbers (mean ± SD) indicate the percentage of those probes which gave high‐intensity signals (category III). Pooled DNA from different strains of the species was used for hybridization.

*

Pat, Pectobacterium atrosepticum (16 hybridizations); Sca, Streptomyces scabies (two hybridizations); Stu, S. turgidiscabies (two hybridizations); Cms, Clavibacter michiganensis ssp. sepedonicus (two hybridizations).

†

Pat, Sca and Cms with no further definition of the target sequence indicate the probes designed for the coding regions (genes) of these species. Other target sequences: IGS, intergenic spacer between the 16S and 23S rRNA genes; nip, necrosis‐inducing protein gene; PAI, pathogenicity island. Probes were also designed to the most conserved parts of the IGS sequences of Dickeya (Dic) strains and the IGS sequence of Ralstonia solanacearum (Rso). A total of 30 probes of different lengths were designed to the nip gene of D. dadantii (nip Dic). However, the IGS sequences of Pat and Pca were highly similar, and therefore only two probes specific to some Pca strains could be designed. They did not detect any bacterial strain tested in this study and were not included in this table.

Comparison of bacterial species using pooled DNA from different strains

Initially, equal amounts of DNA isolated from pure cultures of different strains of the same species were pooled for analysis in order to ensure that the sample would contain the majority of genes existing in the species. If a probe gave no signal, the reason could be that the probe did not work for technical reasons or the target gene was rare among the strains of the target species. This strategy also allowed us to detect non‐specific probes that gave signals not only for the aimed target species but also for other species. The use of pooled samples made the evaluation of probes more cost‐ and labour‐efficient than would have been possible by using individual strains. This approach was also motivated by the large number of probes which were designed for genes of three bacterial species that had been sequenced only from a single strain. If target sequences such as the 16S rRNA gene available from many strains and species were used (Sanguin et al., 2006), initial evaluation of probes similar to this study are no longer needed after sequence comparisons and evaluation could be started directly using single strains. Because the strains of Dickeya formerly belonged to the same species (E. chrysanthemi) they were also initially tested as a pool.

Results indicated that the quality of the microarrays was good and the contrast between the spot foreground and background was in most cases very clear (Fig. 1). Because the amount of DNA was not a limiting factor in this study, 500 ng of DNA was used for hybridization to ensure strong signals and reliable evaluation of probes. However, 50 ng of DNA was also sufficient for unambiguous detection of the signals, and many signals were detected with as little as 5 ng of DNA (Fig. 1A).

Figure 1.

Figure 1

Open in a new tab

Scanned images of the signals detected on the microarray. A view of the whole microarray with eight subarrays is shown in (A), whereas areas covered by c. 800 probes (of the total of 9676 probes of one subarray) are shown at higher magnification in B and C. Total DNA extracted from pure cultures of bacteria and pooled from several strains of each species was used for hybridization. (A) Two samples were hybridized on each of the eight subarrays. The sample labelled with Cy5 (red) in all eight subarrays was Pectobacterium atrosepticum. The other samples labelled with Cy3 (illustrated as green) were (1) Streptomyces scabies, (2) Dickeya sp., (3) P. carotovorum, (4) Clavibacter michiganensis, (5) P. atrosepticum and (6–8) S. turgidiscabies. The amount of DNA per sample was 500 ng in subarrays 1–6. Signals were clear also with 50 ng of sample DNA (dilution 1 : 10, subarray 7). The image shown here was scanned using constant laser power and detector gain, and signals in subarray 8 (5 ng of DNA; dilution 1 : 100) cannot be seen. However, using increased detector gain, the most species‐specific signals (highest signal intensity) could be detected on subarray 8. (B) Magnification of a part of subarray 5: two samples of P. atrosepticum labelled each with a different dye. Intensive yellow spots (equal hybridization) correspond to probes specific to P. atrosepticum, whereas the spots with faint signal indicate non‐specific hybridization. (C) Magnification of part of the subarray 1: P. atrosepticum labelled with Cy5 and S. scabies labelled with Cy3. A ‘black spot’ (no signal) indicates no hybridization with the probe. The probes were designed to be gene‐specific, taking the whole‐genome sequence information of the species into consideration. Results indicate that most probes detect only the respective species based on which the probes were designed.

When investigating the shape of the histogram of logarithmically transformed signal intensities of all probes, three peaks could be clearly observed with most samples. 2, 3illustrate the method used to group the probes into three categories, which is the basis of identifying the probes to be used for species classification in each comparison. The general idea of grouping intensity levels based on the shape of a histogram has originally been used for binarization of medical images (Chow and Kaneko, 1972). This method has been previously applied to microarray data (Asyali et al., 2004) in a somewhat similar manner as described here. Figure 2 illustrates a case where two samples (bacterial species) do not share probes with high‐intensity signals, whereas Fig. 3 shows an example where the two species compared share several probes with high‐intensity signals. Species‐specific probes were selected based on these results. For each sample, the probes that gave no or negligible signals (peak I) were classified to category I. Probes giving signals with a low to modest intensity (peak II) were classified to category II and the probes with high signals (peak III) were classified to category III (2, 3). The threshold signal intensity defining category III was determined for each sample (hybridization) and the probes belonging to this category were used for the comparison of samples. This strategy also eliminated probes that were not working properly for some reason, e.g. because the target sequence might have had secondary structures. It was found that high signals belonging to category III were observed with the majority of probes designed for the bacterial species tested, whereas only a low proportion of the probes designed for other species gave high signal intensities (Table 1). It was also noted that including a large number of probes to different species in the array and the lower signal intensities (category II) from the probes for species different from the one tested facilitated determination of the threshold signal intensity for category III. Results were reproducible in the repeated experiments (Table 1).

Figure 2.

Figure 2

Open in a new tab

Pooled DNA of the strains of Clavibacter michiganensis ssp. sepedonicus (Cms) (labelled with Cy3) and Pectobacterium atrosepticum (Pat) (labelled with Cy5) analysed on the microarray. (A) Scatterplot shows signal intensities from each probe on the array. Signals for Cms are given on the x‐axis and those for Pat on the y‐axis. Data reveal that the samples are not detected with common probes giving high signal intensities. (B) The scatterplot presented in a logarithmic domain places the probes within four groups: (1) high signal intensities for both samples (very few probes); (2) non‐specific probes detecting both samples (relatively low signal intensities); (3) probes giving high signal intensities only for Pat; and (4) probes giving high signal intensities only for Cms. In (C) (Cms) and (D) (Pat), the histograms of the logarithmic signal intensities show three peaks (histograms smoothened by the kernel density method). A threshold value of ~10 separates the two right‐most peaks (II and III) corresponding to the non‐specific and specific probes, respectively, as shown in B. The threshold value corresponds to the raw (non‐logarithmic) intensity value of c. 1000. In (E) (Cms) and (F) (Pat) the hybridization signal intensities are indicated per groups of probes. In the boxplot, the horizontal line in the middle of the box indicates the median value of the data. The box itself shows the first and third quartile of data. Whiskers outside the box indicate the range of data up to 1.5× the box height from both ends. Data beyond these limits are shown as circles. The intensity values of all probes are shown; however, in the final classification, the probes with intensities below the threshold obtained from the intensity histogram would be eliminated. Abbreviations used in the probe group names: Pat, P. atrosepticum; Sca, S. scabies; Cms, C. michiganensis spp. sepedonicus; IGS, 16S–23S intergenic spacer; Pca, P. carotovorum; IGS Dic, probes to the IGS of Dickeya spp.; Stu, S. turgidiscabies; Rso, R. solanacearum; nip, gene for necrosis‐inducing protein; Dic Nip30‐Nip50, probes of different lengths (30–50 nt) designed for the nip gene of D. dadantii; PAI, pathogenicity island.

Figure 3.

Figure 3

Open in a new tab

Pooled DNA of the strains of Streptomyces scabies (Sca) and S. turgidiscabies (Stu) analysed on the microarray. (A) Scatterplot showing signal intensities from each probe on the array. Signals for Sca are given on the x‐axis and those for Stu on the y‐axis. (B) The scatterplot presented on a logarithmic scale places the probes within four groups: (1) high signal intensities for both samples [of the total of 3894 probes designed to target genes of Sca, 1462 probes (c. 40%) show high signal intensities also for Stu]; (2) non‐specific probes giving relatively weak signals for both samples; (3) probes giving high signals only for Stu; and (4) probes giving high signals only for Sc. In (C) (Sca) and (D) (Stu), the histograms of the logarithmic signal intensities show three peaks corresponding to the groups of probes in B, as explained in Fig. 2. In (E) (Sca) and (F) (Stu) the hybridization signal intensities are indicated per three groups of probes. Interpretation of the boxplots is as in Fig. 1. The data indicate that the probes targeting the 16S–23S intergenic spacer (IGS) can be used to distinguish the two species.

Pairwise comparisons of bacterial species using the DNA pools indicated that unrelated species, such as C. michiganensis ssp. sepedonicus and P. atrosepticum (Fig. 2), and also the related species S. scabies and S. turgidiscabies (Fig. 3) could be readily distinguished from each other using the range of probes on the array and the aforementioned strategy to categorize them according to signal intensity. Almost all probes designed for C. michiganensis ssp. sepedonicus gave high‐intensity signals (category III) for this species only, whereas c. 50 and 40% of the probes designed for the genes of S. scabies and P. atrosepticum were species‐specific, respectively. These results indicated that availability and utilization of whole‐genome sequences greatly increases the chance of obtaining specific probes. In all species, the probes designed for the IGS sequences were species‐specific (e.g. 2, 3). Similarly, the probes for the nip genes of Pectobacterium and Dickeya species appeared to be species‐specific (Fig. 2F; data not shown).

Over 100 probes including those for the IGS sequence and many probes for the pathogenicity island were specific to S. turgidiscabies. By contrast, the c. 2000 probes that were found to be specific to S. scabies were distributed over the whole bacterial genome. These probes specific to S. turgidiscabies or S. scabies excluded those c. 300 probes that were designed for the genes of S. scabies but which provided category III signals when the pooled DNA of five additional Streptomyces species of non‐potato origin were used for hybridization as a control. Based on the 16S rRNA gene sequences determined in this study (EU216727–EU216731), these additional streptomycetes were not closely related to S. scabies or S. turgidiscabies, which agreed with the aforementioned data from the microarray analysis.

Comparison of strains of Pectobacterium and Dickeya

The results shown in Fig. 4 for single strains are representative for those typically observed with the strains of P. atrosepticum and P. carotovorum and the Dickeya strains included in this study. All probes designed according to the gene sequences of P. atrosepticum SCRI1043 provided high signal levels (category III) following hybridization with this strain (Fig. 4A). However, the number of high‐intensity signals was smaller for P. carotovorum (Fig. 4C) and even less for Dickeya spp. (Fig. 4D). Strains from P. atrosepticum and P. carotovorum shared as many as 2309 probes that gave category III signals but only 110 of these probes gave high‐intensity signals also with all strains of Dickeya spp. These microarray data pinpointing dissimilarities at several thousand loci in the bacterial genomes imply that Dickeya spp. are clearly different from the two Pectobacterium species, which is supported by previous studies based on DNA–DNA hybridization analysis (Samson et al., 2005) and phylogenetic analyses based on defined genes or parts thereof (Ma et al., 2007; Samson et al., 2005; Young and Park, 2007). Previously, all of these bacteria were classified to the genus Erwinia but the strains of former Erwinia chrysanthemi have recently been moved to a new genus, Dickeya (Gardan et al., 2003; Samson et al., 2005).

Figure 4.

Figure 4

Open in a new tab

Signal intensities representative for the strains of P. atrosepticum (A and B), P. carotovorum (C) and Dickeya spp. (D) observed with probes designed for the genes of P. atrosepticum (strain SCRI1043) (one probe per gene). The signals are presented from left to right in the order of the genes in the genome of P. atrosepticum. Five regions containing most of the between‐strain variation, as interpreted based on the signal intensities, are designated as α, β, γ, δ and ∈. (A) P. atrosepticum strain SCRI1043. (B) P. atrosepticum strain Pa s0404. (C) P. carotovorum strain Pa s0416. (D) Dickeya strain D w04K. Signal intensity distribution of all probes from an array is shown to the right. The topmost peak corresponds to the high‐intensity (category III) signals.

In general, signal intensities were relatively low for strains of Dickeya (Fig. 4D) owing to their distant relationship to P. atrosepticum and other species according to which the great majority of probes were designed. Hence, the signal intensity histograms obtained from Dickeya did not display three peaks as clearly as those obtained with other species. It was found that the number of saturated spots, i.e. spots whose intensity value exceeded the largest integer available for representing the signal intensity (in GenePix software 216 = 65 536) also affected the shape of the histogram and hence selection of the intensity cut‐off point. Regardless of these challenges, classification of signals was possible with the procedure chosen for this study (Fig. 4D).

Comparison of strain SCRI1043 (Fig. 4A) with another strain of P. atrosepticum (Fig. 4B) revealed four larger genomic regions and several shorter regions which were less similar than others in these strains, as judged based on the signal intensities. The larger dissimilar regions were designated as β (ECA1590–1679), γ (ECA2598–2637), δ (ECA2871–2921) and ɛ (ECA3695–3742) defined according to the numerical order of the genes in the genome of strain SCRI1034. Furthermore, these four genomic regions and an additional region α (genes ECA516–612) contained most of the variation between the two species, P. atrosepticum and P. carotovorum (Fig. 4C). Regions α, δ and ɛ were also the most different between these two species and Dickeya (Fig. 4D). Closer examination of the five variable areas revealed that they all consist of inserted bacteriophage and plasmid sequences.

Pairwise differences between the two strains of P. atrosepticum, four strains of P. carotovorum and five Dickeya strains tested were determined based on the number of probes which gave category III signals for one strain but less intensive, lower category signals for any other strain. Results showed that all strains could be classified correctly to the respective species using this procedure (Fig. 5).

Figure 5.

Figure 5

Open in a new tab

Differences between two strains of P. atrosepticum (Pat), four strains of P. carotovorum (Pca) and five Dickeya strains. Dissimilarity is defined by the number of probes which belong to the high signal category III in one strain but to a lower signal category in another strain. The colour bar to the right shows how the number of differing probes corresponds to each colour (dark: similar strains; white: highly dissimilar strains).

The microarray data indicated differences also between the strains belonging to the genus Dickeya. There were six probes for the most conserved parts of the IGS sequences of Dickeya on the array and a total of 30 probes with three different lengths (30, 40 or 50 nt) designed using the nip gene sequence of D. dadantii. Results revealed that the strains of Dickeya (D w04K, D s053‐3 and D s0421‐1) showing high levels of virulence in inoculated potato plants in the field (Laurila et al., 2008) were detected with all these probes. In contrast, the strains (Dw0440 and Dw054) showing low virulence on potato were detected with only four probes for the IGS region and no probe for the nip gene (Fig. 6).

Figure 6.

Figure 6

Open in a new tab

Hybridization intensities for two groups of Dickeya showing (A), high virulence or (B), low level of virulence on potato. The probes were designed based on the 16S–23S IGS region or nip gene of D. dadantii. IGS Dic1 and IGS Dic2 indicate the signal intensities from four and two probes designed for the IGS. nip Dic1 to Dic3 indicate signal intensities from probes of three lengths (30, 40 and 50 nt, respectively), ten different probes per length, targeting the nip gene. All probes were printed in triplicate on the array. In the boxplot, the horizontal line in the middle of the box indicates the median value of data. The box itself shows the first and third quartile of data. Whiskers outside the box indicate the range of data up to 1.5× the box height from both ends. Data beyond these limits are shown as circles. Note that the scale of signal intensities is logarithmic.

CONCLUSIONS

In the present study, microarray analysis allowed us to evaluate thousands of new probes in a time‐ and cost‐effective manner, which would be difficult by other means. Utilization of the whole‐genome sequence information for probe design resulted in large numbers of species‐specific probes that readily distinguished the species and strains. Classification of bacterial species and strains could be done following a hierarchical procedure as some probes detected more than one bacterial species. This procedure increased robustness and partly compensated for the lack of full‐genome information for some species of interest. For example, the soft rot enterobacteria P. atrosepticum, P. carotovorum and Dickeya spp. could be detected with a set of 110 common probes. If hybridization signals above the dynamic threshold (signal intensities belonging to category III) were obtained from these probes, the sample would be known to contain at least one of these species. Subsequently, for more detailed classification, the probes for the IGS sequences and nip genes can be used to reveal the species and, in the case of Dickeya strains, possibly to predict the level of virulence on potato. Following this scheme, the results are more reliable as compared with the classification based on only few species‐specific probes. For example, when a new deviant strain or species emerges in an existing genus, it is possible to classify the new species at multiple levels starting from the genus level and then proceeding towards the species or subspecies level. In contrast, if only highly specific probes designed for a known species were used, the presence of a new species, subspecies or a very different strain could be missed.

Taken together, all the bacterial species included in the study could be detected and differentiated with the set of probes used. This was possible even with species from which whole‐genome sequence data were not available, but in such cases the number of the specific probes was limited, as in previous studies on other bacteria (Belosludtsev et al., 2004; Günther et al., 2006; Loy et al., 2002). The results demonstrate how the rapidly increasing whole‐genome data on bacteria and other plant pathogens can be efficiently utilized through custom‐based oligonucleotide microarray technology to develop powerful tools for pathogen diagnostics and phytopathology research.

EXPERIMENTAL PROCEDURES

Sequence data and probe design

The whole‐genome sequences of P. atrosepticum (strain SCRI1043; accession no. BX950851) (Bell et al., 2004), S. scabies (strain 87.22) (The Sanger Institute, UK, and R. Loria, Cornell University, USA) and C. michiganensis ssp. sepedonicus strain ATCC 33113 (Ishimaru et al., 2004; The Sanger Institute, UK, and C. A. Ishimaru et al.; http://www.sanger.ac.uk/Projects/C_michiganensis/) and the PAI of S. turgidiscabies (accession nos AY707079, AY707080, AY707081) (Kers et al., 2005) were obtained from the respective database. The sequences of S. scabies and C. michiganensis ssp. sepedonicus were not yet fully annotated but they were available from the genome project database of the Sanger Institute, UK. The 16S–23S IGS sequences were obtained from Genbank: Dickeya spp. (nine sequences: AF232681–3, AF373161, AF373199–203), S. scabies (100 sequences: AB026199–216, AB041088–127, AB041136, AB042771–82, AF363485, AY296914, AY296921–26, AY296969–70, A296973, AY296997–7020, AY845136), S. turgidiscabies (19 sequences: AB026221, AB041138–48, AB042785–6, AF363486, AY296927–30), C. michiganensis ssp. sepedonicus (U09382) and R. solanacearum (20 sequences: AJ277767–77, AJ277848–56). Sequences of the nip genes were available for P. carotovorum (Mattinen et al., 2004), P. atrosepticum (BX950851) and D. dadantii (Genome Center of Wisconsin, USA; http://asap.ahabs.wisc.edu/asap/home.php).

Probes were designed using the OligoArray 2.1 software (Rouillard et al., 2003) using the following criteria. One aim was to design a specific probe with 40 target‐specific nucleotides for each gene of the three bacterial species of which the whole‐genome sequence was available. Specificity of probes was estimated based on BLASTN searches (Altschul et al., 1997), in which the user defines the set of sequences against which the probe candidates are compared. For each gene of P. atrosepticum (Pat), this set consisted of other genes of Pat and all genes of Streptomyces scabies (Sca) and Clavibacter michiganensis (Cms). For each gene of Cms, the set consisted of other genes of Cms and all genes of Pat and Sca, and for each gene of Sca, the set consisted of other genes of Sca and all genes of Pat and Cms. For each open reading frame (ORF) in PAI characterized from S. turgidiscabies (Stu), the set consisted of other ORFs in PAI and all genes of Pat and Cms. The reason why Sca was not included in this comparison was that Sca and Stu share at least some regions of the PAI and the aim was to have probes covering the entire PAI rather than just probes which might distinguish these two species. The melting temperature (T m) of the probes had to fall in the range 82–90 °C and the G+C content had to be 45–55 mol%. Homopolymers (stretches of the same base) longer than 5 nt were not permitted. These criteria and conditions limited the number of probes that could be designed. The criteria for the probes designed for the 16S–23S IGS sequences were as above, except that the probe length was allowed to vary to fit the probes in the aforementioned T m range. Probes for the nip genes were designed to have defined lengths of 30, 40 or 50 nt in order to investigate the effect of probe length on detection specificity and signal intensity. As the last step, the necessary number of thymine bases [poly(T)] was added to the 3′‐end to make all probes 60 nt in length, which was the maximum length for probes to be synthesized directly on the array in situ (Agilent 8*15K custom arrays; Agilent, Santa Clara, CA). Each of the eight sub‐arrays synthesized on the same array contained the same set of 9676 unique probes. Probes designed for the nip genes, 16S–23S IGS and PAI were replicated on the sub‐array 3–10 times.

Bacteria

The bacteria used in this work are listed in Table 2. Most are pathogenic strains isolated from potatoes. The five Finnish strains of Dickeya were known to have different levels of virulence on potato. Strain D s0432‐1 and the D. dianthicola‐like strains D w04K and D s053‐3 express high levels of virulence, whereas strains D w0440 and D w054 are less virulent on potato, as shown in field experiments (Laurila et al., 2008). Species identity of the strains of the two Streptomyces species was reconfirmed by PCR using species‐specific primers (Lehtonen et al, 2004). Additional, non‐pathogenic streptomycetes were obtained from the collection of the Department of Applied Biology, University of Helsinki, and included as controls. A partial sequence of the 16S rRNA gene of the five streptomycetes strains with non‐potato origin was determined in this study. A fragment (c. 1.5 kb) of the gene was amplified by PCR using the primers pA and pH′ described by Edwards et al. (1989) and sequenced using the same primers and two additional, nested primers, 16Sf (5′‐AGCGTTGTCCGGAATTATTG‐3′) and 16Sr (5′‐TTGCGGGACTTAACCCAACAT‐3′). The sequences were deposited with GenBank under accession numbers EU216727–EU216731.

Table 2.

Bacterial strains used in this study.

Species Strain Origin Reference or donator
P. carotovorum Pc s0429‐6 potato, Finland Laurila et al., 2008
P. carotovorum Pc s0416 potato, Finland Laurila et al., 2008
P. carotovorum Pc t0436‐2 potato, Finland Laurila et al., 2008
P. carotovorum Pc t0438‐1 potato, Finland Laurila et al., 2008
P. carotovorum SSC1 potato, Finland Pirhonen and Palva, 1988
P. atrosepticum Pa s0404 potato, Finland Laurila et al., 2008
P. atrosepticum Pa t0436‐1 potato, Finland Laurila et al., 2008
P. atrosepticum SCA12 potato, Finland Pirhonen and Palva, 1988
P. atrosepticum SCRI1043 potato, Scotland Hinton et al., 1985
Dickeya sp. D s0432‐1 potato, Finland Laurila et al., 2008
Dickeya sp. D w04K river water, Finland Laurila et al., 2008
Dickeya sp. D s053‐3 potato, Finland Laurila et al., 2008
Dickeya sp. D w0440 river water, Finland Laurila et al., 2008
Dickeya sp. D w054 river water, Finland Laurila et al., 2008
C. michiganensis ssp. sepedonicus p 45 potato, Canada De Boer et al., 1994
C. michiganensis ssp. sepedonicus 3 NM potato, Canada De Boer et al., 1994
C. michiganensis ssp. sepedonicus 3 M potato, Canada De Boer et al., 1994
C. michiganensis ssp. sepedonicus 1438 Finland A. Saano
S. scabies 14 potato, Sweden H. Bång
S. scabies 208 potato, Sweden H. Bång
S. scabies 267 potato, Finland Lindholm et al., 1997
S. scabies 289 potato, Finland Lindholm et al., 1997
S. scabies 364 potato, Finland Lindholm et al., 1997
S. scabies ATCC 49173 potato, U.S.A Lambert and Loria, 1989
S. turgidiscabies 65 potato, Sweden H. Bång
S. turgidiscabies 261 potato, Finland Lindholm et al., 1997
S. turgidiscabies 287 potato, Finland Lindholm et al., 1997
S. turgidiscabies 300 potato, Finland Lindholm et al., 1997
S. turgidiscabies 323 potato, Finland Lindholm et al., 1997
S. griseoviridis K61 peat soil, Finland Tahvonen, 1982
Streptomyces sp. 6I forest soil, Finland H. Kortemaa
Streptomyces sp. 16II forest soil, Finland H. Kortemaa
Streptomyces sp. SscPO carrot, Finland A. Pohto
Streptomyces sp. Antipin unknown J. Kankila
Streptomyces sp. 650 unknown A. Korhonen
Open in a new tab

In the names of Pectobacterium and Dickeya strains, the letters s, t and w indicate isolation from potato stem, potato tuber or river water, respectively. All strains of S. scabies and S. turgidiscabies were originally isolated from scab lesions on potato tubers.

For DNA isolation, C. michiganensis ssp. sepedonicus strains were grown on solid YGM (De Boer and Copeman, 1980) for 5 days at 26 °C. The Streptomyces strains were grown in potato dextrose broth (Sigma, Steinheim, Germany) at 28 °C for 7 days and the other bacteria in Luria–Bertani broth (Becton Dickinson, Sparks, MD) at 37 °C for 1–2 days with shaking. The cells were collected, resuspended into 600 µl TE buffer (50 mm Tris‐HCl, pH 8; 20 mm EDTA), and 200 µl 5% sodium dodecyl sulphate (SDS; VWR, Poole, UK) and 200 µl of proteinase K (1 mg/mL) (Finnzymes, Espoo, Finland) were added. The soft rot enterobacteria were incubated for 30 min and the Streptomyces and the Clavibacter strains were incubated for up to 4 h at 37 ºC until lysed. After 10 min of centrifugation at 9500 g the supernatant was extracted with phenol and chloroform and the DNA precipitated with ethanol, washed with 70% ethanol, dried and resuspended in deionized sterile water.

Hybridization

DNA samples (500 ng, or in a few cases 50 or 5 ng; Fig. 1A) were labelled with Cy3 dCTP (GE Healthcare, Bucks., UK) and purified using BioPrime labelling kit (Invitrogen, Carlsbad, CA). The concentration of DNA and the incorporation of the dye were checked with Nanodrop analysator (Nanodrop Technologies, Wilmington, MA) before and after labelling. Hybridization was carried out at 65 °C for 18–20 h using the microarray manufacturer's reagents and 8 × 15K CGH protocol (Agilent publication number G4427‐90010; http://www.chem.agilent.com/scripts/LiteraturePDF.asp?iWHID=49264).

For experiments using pooled DNA of the different strains of a bacterial species, the DNA was first pooled and then labelled. Experiments were repeated with newly extracted DNA swapping the dyes so that the reference P. atrosepticum pool was labelled with Cy3 and the other samples with Cy5. The DNA of strains Pc s0416, Pc t0436‐2, Pc t0438‐1 and SSC1 of P. carotovorum, strains Pa s0404 and SCRI1043 of P. atrosepticum and all strains of Dickeya were also labelled and hybridized as individual samples, and the experiments were repeated with new samples of DNA.

Image and data analysis

Microarray slides were scanned with a GenePix 4200 AL scanner (Axon Instruments, Foster City, CA) using a pixel resolution of 5 µm. Image analysis and spot segmentation were done with GenePix Pro 6.0 software. Each spot was characterized by the difference between the mean of the foreground and the median of the local background pixel intensities. In the case of replicated probes, the median value was used.

Spot intensity values were investigated using the kernel density method (Silverman, 1986), which provides a smoothened representation of the spot intensity distribution. The benefit over a standard histogram is that specification of the number of histogram bins is not needed. Each data point acts as a mean of the kernel and the distribution of the entire data set is obtained as a sum of all kernels. Smoothness of the representation is controlled by the kernel bandwidth. The R program was used for computation (R Development Core Team, 2004). The Gaussian shape was chosen for the kernels and the bandwidth was determined by the default method ‘bw.nrd0’ of the R package ‘stats’.

Intensity distribution was divided into three regions based on its shape. The regions corresponded to the probes with no signal (category I), non‐specific hybridization (category II) and specific hybridization (category III). The histogram‐based threshold selection is a common practise in image processing (Gonzalez and Woods, 1992). Our approach was semi‐automatic. The user indicates the approximate locations of the two peaks in the histogram and the computer program subsequently determines the minimum of the ‘valley’ between the peaks. The intensity value corresponding to the determined minimum was used as a threshold in further analyses of data. In most cases three peaks were observed, among which one peak (I) corresponded to probes giving no signal in the hybridization. However, the aim of the analysis was to determine the threshold between the two peaks (II and III) of signal intensities corresponding to the background level (non‐specific) and high‐intensity (specific) signals, respectively. In exceptional cases only a single peak was observed in the intensity histogram and the threshold was set manually. This situation occurred when nearly all probes provided only signal intensities corresponding to background level.

In the present study, the intensity histogram was modelled by a Gaussian‐kernel density where the number of the Gaussians corresponded to the number of the probes. An alternative way to separate two intensity peaks would be to fit only two Gaussians to the histogram, one corresponding to non‐specific hybridization and another corresponding to specific hybridization. After estimating the parameters of this two‐component mixture, the optimal threshold can be readily calculated (Asyali et al., 2004; Chow and Kaneko, 1972).

ACKNOWLEDGEMENTS

We thank Dr Solke De Boer for information concerning the status of Cms in North America. Financial support from the Ministry of Agriculture and Forestry (grants 3326/501/2002 and 4541/501/2006), the Finnish Funding Agency for Technology and Innovation (TEKES) (grant 2431/31/04) and Mobidiag Oy is gratefully acknowledged.

REFERENCES

  1. Abdullahi, I. , Koerbler, M. , Stachewicz, H. and Winter, S. (2005) The 18S rDNA sequence of Synchytrium endobioticum and its utility in microarrays for the simultaneous detection of fungal and viral pathogens of potato. Appl. Microbiol. Biotechnol. 68, 368–375. [DOI] [PubMed] [Google Scholar]
  2. Altschul, S.F. , Madden, T.L. , Schiffer, A.A. , Zhang, J. , Zhang, Z. , Miller, W. and Lipman, D.J. (1997) Gapped BLAST and PSI‐BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389–3402. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Asyali, M. , Shoukri, M. , Demirkaya, O. and Khabar, K. (2004) Assessment of reliability of microarray data and estimation of signal thresholds using mixture modeling. Nucleic Acids Res. 32, 2323–2335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Avarre, JC. , De Lajudie, P. and Béna, G. (2007) Hybridization of genomic DNA to microarrays: a challenge for the analysis of environmental samples. J. Microbiol. Methods, 57, 269–278. [DOI] [PubMed] [Google Scholar]
  5. Bell, K.S. , Sebaihia, M. , Pritchard, L. , Holden, M.T. , Hyman, L.J. , Holeva, M.C. , Thomson, N.R. , Bentley, S.D. , Churcher, L.J. , Mungall, K. , Atkin, R. , Bason, N. , Brooks, K. , Chillingworth, T. , Clark, K. , Doggett, J. , Fraser, A. , Hance, Z. , Hauser, H. , Jagels, K. , Moule, S. , Norbertczak, H. , Ormond, D. , Price, C. , Quail, M.A. , Sanders, M. , Walker, D. , Whitehead, S. , Salmond, G.P.C. , Birch J., P.R.J., Parkhill, J. and Toth, I.K. (2004) Genome sequence of the enterobacterial phytopathogen Erwinia carotovora subsp. atroseptica and characterization of virulence factors Proc. Natl Acad. Sci. USA, 101, 11105–11110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Belosludtsev, Y.Y. , Bowerman, D. , Weil, R. , Marthandan, N. , Balog, R. , Luebke, K. , Lawson, J. , Johnston, S.A. , Lyons, C.R. , Obrien, K. , Garner, H.R. and Powdrill, T.F. (2004) Organism identification using a genome sequence‐independent universal microarray probe set. BioTechniques, 37, 654–660. [DOI] [PubMed] [Google Scholar]
  7. Bodrossy, L. , Stralis‐Pavese, N. , Murrell, J.C. , Radajewski, S. , Weilharter, A. and Sessitsch, A. (2003) Development and validation of a diagnostic microbial microarray for methanotrophs. Environ. Microbiol. 5, 566–582. [DOI] [PubMed] [Google Scholar]
  8. Boonham, N. , Walsh, K. , Smith, P. , Madagan, K. , Graham, I. and Barker, I. (2003) Detection of potato viruses using microarray technology: towards a generic method for plant viral disease diagnosis. J. Virol. Meth. 108, 181–187. [DOI] [PubMed] [Google Scholar]
  9. Boonham, N. , Tomlinson, J. and Mumford, R. (2007) Microarrays for rapid identification of plant viruses. Annu. Rev. Phytopathol. 45, 307–328. [DOI] [PubMed] [Google Scholar]
  10. Bystricka, D. , Lenz, O. , Mraz, I. , Dedic, P. and Sip, M. (2003) DNA microarray: parallel detection of potato viruses. Acta Virol. 47, 41–44. [PubMed] [Google Scholar]
  11. Carlson, R.R. and Vidaver, A.K. (1982) Taxonomy of Corynebacterium plant pathogens, including a new pathogen of wheat, based on polyacrylamide gel electrophoresis of cellular proteins. Int. J. Syst. Bacteriol. 32, 315–326. [Google Scholar]
  12. Chow, C. and Kaneko, T. (1972) Automatic boundary detection of the left ventricle from cineangiograms. Comp. Biomed. Res. 5, 388–410. [DOI] [PubMed] [Google Scholar]
  13. De Boer, S. , Stead, D. , Alivizatos, A. , Jansen, J. , Van Vaerenbergh, J. , De Haan, T. and Mawhinney, J. (1994) Evaluation of serological tests for detection of Clavibacter michiganensis subsp. sepedonicus in composite potato stem and tuber samples. Plant Dis. 78, 725–729. [Google Scholar]
  14. De Boer, S.H. and Copeman, R.J. (1980) Bacterial ring rot testing with the direct fluorescent antibody staining procedure. Am. Potato J. 57, 457–465. [Google Scholar]
  15. Deyong, Z. , Willingmann, P. , Heinze, C. , Adam, G. , Pfunder, M. , Frey, B. and Frey, J.E. (2005). Differentiation of cucumber mosaic virus isolates by hybridization to oligonucleotides in a microarray format. J. Virol. Methods, 123, 101–108. [DOI] [PubMed] [Google Scholar]
  16. Edwards, U. , Rogall, T. , Blöcker, H. , Emde, M. and Böttger, E.C. (1989). Isolation and direct complete nucleotide determination of entire genes. Characterization of a gene coding for 16S ribosomal RNA. Nucleic Acids Res. 17, 7843–7853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Fessehaie, A. , De Boer, S.H. and Lévesque, C.A. (2003) An oligonucleotide array for the identification and differentiation of bacteria pathogenic on potato. Phytopathology, 93, 262–269. [DOI] [PubMed] [Google Scholar]
  18. Françoisa, C. , Kebdania, N. , Barkerb, I. , Tomlinsonb, J. , Boonhamb, N. and Castagnone‐Sereno, P. (2006) Towards specific diagnosis of plant‐parasitic nematodes using DNA oligonucleotide microarray technology: a case study with the quarantine species Meloidogyne chitwoodi . Mol. Cell. Probes, 20, 64–69. [DOI] [PubMed] [Google Scholar]
  19. Gardan, L. , Gouy, C. , Christen, R. and Samson, R. (2003) Elevation of three subspecies of Pectobacterium carotovorum to species level: Pectobacterium atrosepticum sp. nov., Pectobacterium betavasculorum sp. nov. and Pectobacterium wasabiae sp. nov. Int. J. Syst. Evol. Microbiol. 53, 381–391. [DOI] [PubMed] [Google Scholar]
  20. Gonzalez, R. and Woods, R. (1992) Digital Image Processing. New York: Addison‐Wesley. [Google Scholar]
  21. Guidot, A. , Prior, P. , Schoenfeld, J. , Carrere, S. , Genin, S. and Boucher, C. (2007) Genomic structure and phylogeny of the plant pathogen Ralstonia solanacearum inferred from gene distribution analysis. J. Bacteriol. 189, 377–387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Günther, S. , Groth, I. , Grabley, S. and Munder, T. (2006) Design and evaluation of an oligonucleotide‐microarray for the detection of different species of the genus Kitasatospora . J. Microbiol. Methods, 65, 226–236. [DOI] [PubMed] [Google Scholar]
  23. Hinton, J.C.D. , Perombelon, M.C.M. and Salmond, G.P.C. (1985) Efficient transformation of Erwinia carotovora subsp. carotovora and E. carotovora subsp. atroseptica . J. Bacteriol. 161, 786–788. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ishimaru, C.A. , Knudson, D.L. , Brown, S.E , Francis, D.M. and Parkhill, J. (2004) Genome sequencing of Clavibacter michiganensis subsp. sepedonicus . Phytopathology, 94, S123. [Google Scholar]
  25. Kers, J.A. , Cameron, C.D. , Joshi, M.V. , Burkhalid, R.A. , Morello, J.E. , Wach, M.J. , Gibson, D.M. and Loria, R. (2005) A large, mobile pathogenicity island confers plant pathogenicity on Streptomyces species. Mol. Microbiol. 55, 1025–1033. [DOI] [PubMed] [Google Scholar]
  26. Koide, T. , Zaini, P.A. , Moreira, L.M. , Vencio, R.Z.N. , Matsukuma, A.Y. , Durham, A.M. , Teixeira, D.C. , El‐Dorry, H. , Monteiro, P.B. , Da Silva, A.C.R. , Verjovski‐Almeida, S. , Da Silva, A.M. and Gomes, S.L. (2004) DNA microarray‐based genome comparison of a pathogenic and a nonpathogenic strain of Xylella fastidiosa delineates genes important for bacterial virulence. J. Bacteriol. 186, 5442–5449. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lambert, D.H. and Loria, H. (1989). Streptomyces scabies sp. nov., nom. rev. Int. J. Syst. Bacteriol. 39, 387–392. [Google Scholar]
  28. Laurila, J. , Ahola, V. , Lahtinen, A. , Joutsjoki, T. , Hannukkala, A , Rahkonen, A. and Pirhonen, M. (2008) Characterization of Dickeya strains isolated from potato and river water samples in Finland. Eur. J. Plant Pathol . in press.
  29. Lehtonen, M.J. , Rantala, H. , Kreutze, J.F. , Bång, H. , Kuisma, L. , Koski, P. , Virtanen, E. , Vihlman, K. and Valkonen, J.P.T. (2004) Occurrence and survival of potato scab pathogens (Streptomyces species) on tuber lesions: quick diagnosis based on a PCR‐based assay. Plant Pathol. 53, 280–287. [Google Scholar]
  30. Leiske, D.L. , Karimpour‐Fard, A. , Hume, P.S. , Fairbanks, B.D. and Gill, R.T. (2006) A comparison of alternative 60‐mer probe designs in an in‐situ synthesized oligonucleotide microarray. BMC Genomics, 7, Art. No. 72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lindholm, P. , Kortemaa, H. , Kokkola, M. , Haahtela, K. , Salkinoja‐Salonen, M. and Valkonen, J.P.T. (1997) Streptomyces spp. isolated from potato scab lesions under Nordic conditions in Finland. Plant Dis. 81, 1317–1322. [DOI] [PubMed] [Google Scholar]
  32. Loy, A. , Lehner, A. , Lee, N. , Adamczyk, J. , Meier, H. , Ernst, J. , Schleifer, K.‐H. and Wagner, M. (2002) Oligonucleotide microarray for 16S rRNA gene‐based detection of all recognized lineages of sulfate‐reducing prokaryotes in the environment. Appl. Environ. Microbiol. 68, 5064–5081. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Ma, B. , Hibbing, M.E. , Kim, H.S. , Reedy, R.M. , Yedidia, I. , Breuer, J. , Breuer, J. , Glasner, J.D. , Perna, N.T. , Kelman, A. and Charkowski, A.O. (2007) Host range and molecular phylogenies of the soft rot enterobacterial genera Pectobacterium and Dickeya . Phytopathology, 97, 1150–1163. [DOI] [PubMed] [Google Scholar]
  34. Mattinen, L. , Tshuikina, M. , Mäe, A. and Pirhonen, M. (2004) Identification and characterization of Nip, necrosis‐inducing virulence protein of Erwinia carotovora subsp. carotovora . Mol. Plant–Microbe Interact. 17, 1366–1375. [DOI] [PubMed] [Google Scholar]
  35. Miyajima, K. , Tanaka, F. , Takeuchi, T. and Kuninaga, S. (1998) Streptomyces turgidiscabies sp. nov. Int. J. Syst. Bacteriol. 48, 495–502. [DOI] [PubMed] [Google Scholar]
  36. Mumford, R.A. , Boonham, N. , Tomlinson, J. and Barker, I. (2006) Advances in molecular phytodiagnostics – new solutions for old problems. Eur. J. Plant Pathol. 116, 1–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Naito, S. , Kondo, N. , Akino, S. , Ogoshi, A. and Tanaka, F. (eds) (2004) Novel Approaches to Control of Potato Scab. Proc. Int. Potato Scab Symp. (IPSS 2004). Sapporo, Japan: Hokkaido University. [Google Scholar]
  38. Okinaka, Y. , Yang, C.H. , Perna, N.T. and Keen, N.T. (2002) Microarray profiling of Erwinia chrysanthemi 3937 genes that are regulated during plant infection. Mol. Plant–Microbe Interact. 15, 619–629. [DOI] [PubMed] [Google Scholar]
  39. Palmer, C. , Bik, E. , Eisen, M. , Eckburg, P. , Sana, T. , Wolber, P. , Relman, D. and Brown, P. (2006) Rapid quantitative profiling of complex microbial populations. Nucleic Acids Res. 34, e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Palmer, C. , Bik, E.M. , DiGiulio, D.B. , Relman, D.A. and Brown, P.O. (2007) Development of the human infant intestinal microbiota. PLoS Biol. 5, 1556–1573. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Petersen, K. , Øyan, A.M. , Rostad, K. , Olsen, S. , Bø, T.H. , Salvesen, H.B. , Gjertsen, B.T. , Bruserud, Ø. , Halvorsen, O.H. , Akslen, L.A. , Steen, V.M. , Jonassen, I. and Kalland, K.H. (2007) Comparison of nucleic acid targets prepared from total RNA or poly(A) RNA for DNA oligonucleotide microarray hybridization. Anal. Biochem. 366, 46–58. [DOI] [PubMed] [Google Scholar]
  42. Pirhonen, M. and Palva, E.T. (1988) Occurrence of bacteriophage t4 receptor in Erwinia carotovora . Mol. Gen. Genet. 214, 170–172. [Google Scholar]
  43. R Development Core Team (2004) R: a Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing; (http://www.R‐project.org). [Google Scholar]
  44. Roth, S.B. , Jalava, J. , Ruuskanen, O. , Ruohola, A. and Nikkari, S. (2004) Use of an oligonucleotide array for laboratory diagnosis of bacteria responsible for acute upper respiratory infections. J. Clin. Microbiol. 42, 4268–4274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Rouillard, J.M. , Zuker, M. and Gulari, E. (2003) OligoArray 2.0: design of oligonucleotide probes for DNA microarrays using a thermodynamic approach. Nucleic Acids Res. 31, 3057–3062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Samson, R. , Legendre, J.B. , Christen, R. , Fischer‐Le Saux, M. , Achouak, W. and Gardan, L. (2005) Transfer of Pectobacterium chrysanthemi (Burkholder et al. 1953) Brenner et al. 1973 and Brenneria paradisiaca to the genus Dickeya gen. nov. as Dickeya chrysanthemi comb. nov. and Dickeya paradisiaca comb. nov. and delineation of four novel species, Dickeya dadantii sp. nov., Dickeya dianthicola sp. nov., Dickeya dieffenbachiae sp. nov., and Dickeya zeae sp. nov. Int. J. Syst. Evol. Microbiol. 55, 1415–1427. [DOI] [PubMed] [Google Scholar]
  47. Sanguin, H. , Herrera, A. , Oer‐Desfeux, C. , Deschesne, A. , Simonet, M. , Navarro, E. , Vogel, T.M. , Moënne‐Loccoz, Y. , Nesme, X. and Grundmann, G.L. (2006) Development and validation of a prototype 16S rRNA‐DSA’. Environ. Microbiol. 8, 289–307. [DOI] [PubMed] [Google Scholar]
  48. Sarkar, S.F. , Gordon, J.S. , Martin, G.B. and Guttman, D.S. (2006) Comparative genomics of host‐specific virulence in Pseudomonas syringae . Genetics, 174, 1041–1056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Silverman, B.W. (1986) Density Estimation. London: Chapman and Hall. [Google Scholar]
  50. De Souza, A.A. , Takita, M.A. , Coletta, H.D. , Caldana, C. , Goldman, G.H. , Yanai, G.M. , Muto, N.H. , De Oliveira, R.C. , Nunes, L.R. and Machado, M.A. (2003) Analysis of gene expression in two growth states of Xylella fastidiosa and its relationship with pathogenicity. Mol. Plant–Microbe Interact. 16, 867–875. [DOI] [PubMed] [Google Scholar]
  51. Szemes, M. , Bonants, P. , DeWeerd, M. , Bane, J. , Landegre, U. and Schoen, C.D. (2005). Diagnostic application of padlock probes‐multiplex detection of plant pathogens using universal microarrays. Nucleic Acids Res. 33, e70. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Tahvonen, R. (1982) Preliminary experiments into the use of Streptomyces spp. isolated from peat in the biological control of soil and seed‐borne diseases in peat culture. J. Sci. Agric. Soc. Finl. 54, 357–369. [Google Scholar]
  53. Takano, Y. , Choi, W.B. , Mitchell, T.K. , Okuno, T. and Dean, R.A. (2003) Large scale parallel analysis of gene expression during infection‐related morphogenesis of Magnaporthe grisea . Mol. Plant Pathol. 4, 337–346. [DOI] [PubMed] [Google Scholar]
  54. Ulrich, G. , Seong, K.Y. , Boddu, J. , Cho, S.H. , Trail, F. , Xu, J.R. , Adam, G. , Mewes, H.W. , Muehlbauer, G.J. and Kistler, H.C. (2006) Development of a Fusarium graminearum Affymetrix GeneChip for profiling fungal gene expression in vitro and in planta . Fungal Genet. Biol. 43, 316–325. [DOI] [PubMed] [Google Scholar]
  55. Van der Wolf, J.M. and De Boer, S.H. (2007) Bacterial pathogens of potato In: Potato Biology and Biotechnology (Vreugdenhil D., Bradshaw J., Gebhardt C., Govers F., MacKerron D.K.L., Taylor M.A. and Ross H.A., eds), pp. 619–641. Amsterdam: Elsevier. [Google Scholar]
  56. Wolber, P.K. , Collins, P.J. , Lucas, A.B. , De Witte, A. and Shannon, K.W. (2006) The Agilent in situ‐synthesized microarray platform. Methods Enzymol. 410, 28–57. [DOI] [PubMed] [Google Scholar]
  57. Young, J.M. and Park, D.C. (2007) Relationships of plant pathogenic enterobacteria based on partial atpD, carA, and recA as individual and concatenated nucleotide and peptide sequences. Syst. Appl. Microbiol. 30, 343–354. [DOI] [PubMed] [Google Scholar]
  58. Zwiesler‐Vollick, J. , Plovanich‐Jones, A.E. , Nomura, K. , Bandyopadhyay, S. , Joardar, V. , Kunkel, B.N. and He, S.Y. (2002) Identification of novel hrp‐regulated genes through functional genomic analysis of the Pseudomonas syringae pv. tomato DC3000 genome. Mol. Microbiol. 45, 1207–1218. [DOI] [PubMed] [Google Scholar]

Articles from Molecular Plant Pathology are provided here courtesy of Wiley

RESOURCES