Internal colonization pathways of potato plants by Erwinia carotovora ssp. atroseptica

Authors


Hélias To whom correspondence should be addressed (e-mail: helias@rennes.inra.fr).

Abstract

Transmission of pectinolytic Erwinia species from infected mother tubers to daughter tubers has been studied mainly through detection tests, carried out at harvest, on limited samples of tubers produced by plants grown from artificially inoculated mother tubers. However, detection has not been performed on samples collected at different stages of crop development, in order to follow the contamination progress in different organs through the plants to the progeny tubers. In this study the bacterial contamination of progeny tubers was investigated by detecting Erwinia carotovora ssp. atroseptica in different symptomless plant organs (stolons, stems, progeny tubers) and in the parts with or without symptoms of diseased stems, collected at various stages of crop development. Infection levels in below- and above-ground organs of plants of two cultivars differing in their resistance to Erwinia, infected by either vacuum infiltration or sand wounding, were monitored throughout the growing season and at harvest using DAS-ELISA and PCR. Detection tests showed that healthy organs from symptomless plants were less frequently contaminated than symptomless organs from diseased plants, and that stolons were precociously and more frequently contaminated than stems and daughter tubers, irrespective of the health of the plant. Stem infections were shown to progress latently in the stem, bacteria usually being recovered 10–15 cm past visible lesions. In many cases, typical aerial stem-rot symptoms could be related to this upward movement of bacteria from the infected mother tuber. Daughter tubers without symptoms were shown to be frequently contaminated, usually at heel ends, suggesting internal contamination from mother tuber to progeny.

Introduction

Pectinolytic Erwinia species cause soft-rot diseases of numerous crops. The Erwinia spp. associated with potato are E. carotovora ssp. atroseptica (Eca), E. carotovora ssp. carotovora (Ecc) and E. chrysanthemi (Ech). All three bacteria can cause tuber soft rot; while Eca is the predominant agent involved in blackleg under cool climates, Ech predominates in warm regions.

Studies on the epidemiology of potato blackleg caused by Eca have shown that infected seed tubers are a major source of inoculum ( Pérombelon, 1974). Transmission to daughter tubers via environmental pathways, particularly contamination occurring after rotting of the infected mother tuber, liberation of the bacteria into the soil and pollution of progeny tubers by soil water ( Elphinstone & Pérombelon, 1986), has been well established. However, studies of direct contamination from seed to progeny tubers through the aerial and subterranean plant parts are limited. Bacterial progress through the plant under field conditions has been poorly documented so far, because plants were not systematically collected during crop growth.

Such studies require the development of sensitive and specific methods to detect Eca in plant material. Dilution platings on a selective medium ( Pérombelon & Hyman, 1986) allow detection of 1–5 × 104 c.f.u. mL−1 peel extract ( Jones et al., 1994 ), but are laborious and time-consuming, and do not permit discrimination between the different subspecies of E. carotovora. Moreover, their efficacy can be seriously affected when large populations of either saprophytic bacteria or other E. carotovora subspecies (such as Ecc), able to overgrow Eca, are also present in the samples ( Jones et al., 1994 ; Pérombelon & Hyman, 1995; Van Der Wolf et al., 1996 ). Immunofluorescent colony staining (IFC) ( Jones et al., 1994 ) and ELISA after an enrichment step ( Gorris et al., 1994 ) combine isolation in a selective medium and serology, resulting in good sensitivity thresholds (1 × 102–1 × 104 c.f.u. mL−1 in peel extract); both can be applied to large samples. They often rely on monoclonal antibodies ( Vernon-Shirley & Burns, 1992; Gorris et al., 1994 ; Hyman et al., 1995 ; Alarcon et al., 1995 ), characterized by a good specificity compared with polyclonal ones, despite occasional cross-reactions ( Klopmeyer & Kelman, 1988; Van Der Wolf & Gussenhoven, 1992; Gorris et al., 1994 ; Hyman et al., 1995 ). However, their extreme specificity, often limited to one serogroup, can be a drawback for the use of these antibodies, as five serogroups have been identified to date within Eca. This problem might be overcome by pooling monoclonal antibodies raised against several serogroups, but this requires prior identification of all serogroups. PCR detection avoids this difficulty by using Eca-specific primers ( De Boer & Ward, 1995; Fréchon et al., 1995 ; Smid et al., 1995 ), but its sensitivity on plant material is limited: 1 × 105–1 × 108 cells mL−1 peel extracts ( Fraiije et al., 1996 ; Van Der Wolf et al., 1996 ; Hélias et al., 1998 ). These detection thresholds can be greatly improved by either enrichment or immunomagnetic separation of bacteria before PCR detection, which allows detection, respectively, of as little as 10 ( Fraiije et al., 1996 ) to 2000 cells mL−1 peel extract ( Van Der Wolf et al., 1996 ). PCR detection thresholds could also be lowered to 8.4 × 102 cells mL−1 using a silica-based DNA extraction ( Bertheau et al., 1998 ) combined with a colorimetric DNA detection system on microplate ( Fréchon et al., 1998 ). This rapid and sensitive technology, applicable to large numbers of samples, might prove useful for epidemiological studies.

Resistance could be useful for the control of soft-rot Erwinia species, as no chemical means of control are available ( Lapwood & Gans, 1984). One resistance source correlated with a high level of pectin methylation, increasing resistance of cell walls to enzyme degradation, was identified from the non-tuber-bearing Solanum brevidens and transferred to Solanum tuberosum ssp. tuberosum via somatic fusion ( Pérombelon, 1992). This showed that resistance sources from wild Solanum spp. could be utilized to increase resistance in potatoes ( Pérombelon & Salmond, 1995). Under field conditions, cultivars could be classified as resistant, susceptible and intermediate ( Lapwood & Gans, 1984; Lapwood & Read, 1986; Bain & Pérombelon, 1988; Bain et al., 1990 ).

The present work had three objectives: (i) to investigate bacterial progress through the plant by detecting bacteria in different plant organs at different dates during vegetative growth and up to harvest; (ii) to correlate the apparent health status of plants with latent infections of various organs; and (iii) to determine whether differences in resistance levels between cultivars are translated in terms of latent bacterial colonization of potato organs. To these ends various organs (stems, stolons, progeny tubers) of potato plants, grown in field plots from mother tubers inoculated with one strain of Eca, were assessed for the presence of the bacterium at different times during the growing season, using both PCR-based and DAS-ELISA methods.

Materials and methods

Field experiment

Tubers of two cultivars, Bintje (susceptible) and Désirée (moderately resistant), were inoculated with one Eca strain, chosen for its high pathogenicity, using two different methods (sand wounding or vacuum infiltration) and concentrations, and planted in a split-plot design with two repetitions on 30 April 1997 at INRA, Le Rheu. The five main plots corresponded to inoculum combinations (method × concentration), and subplots to cultivars. The five main plots were separated from one another by a single guard row of cv. Charlotte. A single guard row of cv. Charlotte also surrounded each block. Each subplot consisted of three rows of 20 plants (either Bintje or Désirée), except in the control plot where two rows of 20 plants were used for each cultivar and inoculation method. A complete description of the bacterial strain and protocols used for tuber inoculation, as well as of the disease development in the plots, is given elsewhere ( Hélias et al. 2000 ).

Collection of samples

Four symptomless and four diseased plants were collected from each subplot on 19 July. Plants were washed with tap water and dried in absorbent paper before apparently healthy organs of each plant were sorted by organ type to be tested for latent infection. Both diseased and symptomless organs were collected from diseased plants to verify the presence of Eca. Furthermore, 18 diseased stems were collected on 12 August from plants of each cultivar growing from tubers inoculated with each inoculum concentration, regardless of the inoculation method. Eight apparently healthy stems of each cultivar were collected in the control plot. All were washed separately in tap water, dried in absorbing paper and immediately prepared for Eca detection.

Daughter tubers were harvested manually and individually on 9 and 10 September 1997. Diseased tubers were recorded, but discarded at harvest. Apparently healthy tubers from each plant were collected in paper bags, washed in tap water, air dried and reassessed for visual rot symptoms. Plants were assigned to one of 11 infection classes (0, 1–15, 16–25, 26–35, 36–45, 46–55, 56–65, 66–75, 76–85, 86–95 and > 95%) based on the percentage of diseased tubers in their progeny ( Hélias et al. 2000 ). For each cultivar, the progeny tubers of seven plants from each of the first two classes (0 and 1–15% visually diseased daughter tubers) were chosen among those originating from mother tubers contaminated with the lowest inoculum concentration (4.5 × 106 c.f.u. mL−1), and tested using both DAS-ELISA with and without an enrichment step, and PCR using a commercial kit (ProbeliaTM, Sanofi Diagnostic Pasteur, Marnes La Coquette, France) to detect latent infections.

Preparation of samples for Eca detection

Segments 3 cm long were collected on 19 July from the field, at soil level from each stem of each individual plant. They were cut longitudinally to facilitate crushing, and pooled into samples corresponding to individual plants. Stolons were detached from seed and daughter tubers before being grouped according to the mother plants. Tuber samples consisted of a small core collected from the stolon attachment site and a periderm strip extending from the basal heel end to the apical rose end.

Individual stems collected in August were grouped into three classes ( Fig. 1) depending on the symptom pattern. Class I included all stems with necrosis limited to the attachment site to the mother tuber, or extending from the mother tuber but remaining under ground level. Stems presenting continuous necroses from the point of attachment to the mother tuber to above ground level were grouped in class II. Class III included stems with a continuous, aerial, black necrosis, varying in length and surrounded by discontinuous necroses above and below the main necrosis. Class II corresponds to the ‘classical’ definition of blackleg, while plants in class III are typical of ‘aerial stem rot’ symptoms in the nomenclature of Pérombelon & Kelman (1987). Both cultivars were almost equally represented in classes I (five stems of Bintje and seven stems of Désirée) and II (six stems of Bintje and nine stems of Désirée), but Bintje dominated in class III (seven stems vs. one stem of Désirée). This discrepancy was due to the low proportion of stem rot symptoms in Désirée ( Hélias et al., 2000 ).

Figure 1.

Diagrammatic representation of the position of symptoms and samples in three classes of diseased potato stems used for detecting the distribution of Erwinia carotovora ssp. atroseptica.

Portions (approximately 5 cm long) were cut at intervals of 10 cm along the stems, starting from the necroses and cutting in to the apparently healthy parts ( Fig. 1). However, for stems in class I the first segment was collected at ground level, i.e. 3–5 cm above the symptoms. Samples collected from stems in class III were collected on both sides of the symptoms ( Fig. 1). On symptomless stems from the control plot, samples were collected at 10 cm intervals, starting at ground level and up to the plant apex.

Progeny tubers at harvest

Tests were applied to six samples for each inoculation method and cultivar. For each plant, five tubers were tested individually; tissue from the remaining daughter tubers was pooled to form the sixth sample. For plants grown from mother tubers inoculated by sand wounding, a seventh sample, constituted by mixing the six first samples in equal quantities, was also tested.

Samples of tubers tested individually were prepared as previously described. The collective samples prepared from the remaining tubers of the same plant consisted of cores from the heel ends, and eventually of some periderm strips collected as previously described when the amount of core material was not sufficient. Each plant sample was transferred to a plastic bag and 4 mL PBS buffer (0.8% NaCl, 0.02% KH2PO4, 0.29% Na2PO4 12H2O, 0.02% KCl pH 7.2) supplemented with 0.2% sodium diethyldithiocarbamate (DIECA) was added per g plant tissue. Samples were crushed in a custom-made bead press. Extracts were allowed to settle for 30–60 min before being subjected to the various detection methods.

DNA extraction

DNA extractions were carried out on 1 mL aliquots of sample extracts previously stored at −20°C in Eppendorf tubes, using the protocol of Bertheau et al. (1998 ) with some modifications. Samples were centrifuged at 20 g for 5 min, followed by a second centrifugation of the supernatant at 13 226 g for 10 min. After discarding the supernatant, the pellet was resuspended in 200 μL TE buffer (0.12% Tris, 0.03% EDTA pH 8.0) containing 1% sodium dodecyl sulfate, and incubated at 55°C for 30 min. 100 μL of a 3 m potassium acetate solution in glacial acetic acid were added, and samples were decanted on ice for 10 min before centrifuging at 13 226 g for 5 min. The supernatant was transferred to a new tube, to which 700 μL of a 6 m NaI solution (89.92% NaI, 1.87% Na2SO3) and 10 μL of a 12% silica solution (pH 2) were added. Tubes were gently mixed several times during a 15 min incubation at room temperature, before being centrifuged at 13 226 g for 2 min. The supernatant was discarded before 500 μL of a wash solution (0.24% Tris, 0.03% EDTA, 0.58% NaCl pH 7.5 in an equal volume of absolute ethanol) was added. The content of the tube was mixed and centrifuged at 13 226 g for 2 min. The washing procedure was repeated. The pellet was allowed to dry in an oven at 37°C before being resuspended in 50 μL TE and incubated at 55°C for 5 min. After centrifuging at 13 226 g for 2 min, 45 μL of the supernatant was transferred to a new tube ready for DNA amplification. Tubes were kept at −20°C until used.

Detection methods

All samples were tested by PCR and amplicons were revealed on agarose gels, except for those from progeny tubers sampled after harvest, which were tested using a more sensitive commercial kit (ProbeliaTM) based on the PCR technique. The limited availability of the kit prevented its use on all samples. Furthermore, all samples were tested in DAS-ELISA with or without a previous enrichment step.

PCR protocol used for plant samples and stems

PCR was performed in a 9600 Perkin-Elmer Cetus thermal cycler using primers Y45 (5′-TCACCGGACGC CAACTGTGGCGT-3′) and Y46 (5′-TCGCC ACGTTT CAGCAGAACAAGT-3′), selected from the pectate lyase encoding gene and specific for E. carotovora ssp. atroseptica ( Fréchon et al., 1995 ). Amplification was carried out in a 50 μL reaction mixture containing 5 μL of the amplification buffer supplied with the enzyme, 5 nmol of each of the deoxynucleotides dATP, dCTP, dGTP and dTTP, 25 pmol of each primer, and 1 unit of Taq DNA polymerase (Appligene, Illkirch, France). Amplification conditions were 94°C for 5 min, followed by 40 cycles at 94°C for 30 s, 65°C for 30 s and 72°C for 45 s. Aliquots of PCR products (10 μL) were analysed by electrophoresis through 1.5% agarose (Appligene) minigels containing 0.5 μg mL−1 ethidium bromide. A standard 1 kb ladder (Gibco–BRL, Gaithersburg, USA) was included on each gel.

Probelia kit used for tests on progeny tubers

The ProbeliaTM kit prototype was used according to the protocol provided by the manufacturer (Sanofi Diagnostics Pasteur). All reagents were provided with the kit, as well as positive and negative controls. Amplification solutions were prepared by mixing 45 μL of the Eca amplification solution supplied containing the Y45–Y46 primers, deoxynucleotides and enzyme buffer with 0.2 μL of Taq polymerase and 0.5 μL of dUTP and UDG decontamination reagent per sample to be tested. PCR was carried out as follows: 45 μL of the prepared amplification mix reagent were distributed to each reaction tube (Perkin Elmer Gene Amp thin-walled reaction tubes, Branchburg, NJ, USA), followed by 5 μL test DNA material. DNA amplification was carried out using a DNA thermal cycler (9600 Perkin Elmer Cetus Corp.) as follows: one step at 50°C for 2 min (to deactivate UDG), one step at 94°C for 5 min, 40 repeats of three consecutive steps as described above, and one final step at 72°C for 5 min. The denaturing reagent (50 μL) was added to all tubes, the contents of the tubes were mixed and incubated at 20°C for 10 min, and subsequently stored at 4°C until subjected to DNA detection. DNA detection was performed using a hybridization method in microplate wells provided with the kit, according to the manufacturer's instructions. The assay consisted of four steps: (i) prehybridization; (ii) hybridization of the amplified DNA with a capture probe (5′-TGTCGCTGGCGGAAAAAGTGG- GCGATAACAT-3′) linked to wells ( Fréchon et al., 1995 ); (iii) washing; (iv) revelation using a detection probe (5′-GGCCGAGACCAACAGACAGTATGC-3′) linked to peroxidase and added to the wells together with the PCR products. Both probes are held under licence by Sanofi Diagnostics Pasteur ( Fréchon et al., 1995 ). Absorbance ratios (OD 450 : 620) greater than 0.1 were considered positive.

DAS-ELISA

The ELISA tests were carried out using commercial REALISATM kits (R700, Durviz C.E., Valencia, Spain). The protocol used was an indirect double-antibody sandwich procedure, using a polyclonal antibody to coat ELISA plates and the monoclonal antibody 4G4, specific for Eca serogroup I ( Gorris et al., 1994 ) as the enzyme conjugate. The assay included (or not) an enrichment step as previously described ( Gorris et al., 1994 ). All reagents were provided in the kit. The kits were used according to the manufacturer's instructions, with the following modifications. The extract from each sample was tested in triplicate on both microplates with and without enrichment. Tests were applied on 50 μL plant extract supplemented with 50 μL sample buffer (PBS plus 0.2% DIECA). When enrichment was used, 100 μL of the double polypectate enrichment medium (DPEM; Gorris et al., 1994 ) was added to each well before plates were incubated under anaerobic conditions at 27°C for 40 h. 100 μL PBS–DIECA replaced DPEM for the plates without enrichment. Skimmed milk powder (0.5%) replaced 0.5% bovine serum albumin as blocking agent in the PBS buffer for the dilution of the Eca specific monoclonal antibody, because of its ability to improve antiserum specificity. Optical densities (read at 405 nm after 30 and 60 min) greater than twice those of the negative control were considered positive.

Data analysis

Data obtained from detection on progeny tubers, stems of plants collected during cropping, and harvested tubers were subjected to anova. Percentages of positive results obtained on tubers and stems using DAS-ELISA were used in the case of tests on plant organs, with plants as replicates. Variables used were cultivars, inoculum concentration and plant status (diseased vs. symptomless). Interactions between cultivars and concentrations, as well as between cultivars and plant status, were also tested. For tubers collected at harvest, the percentage of positive results obtained over five and six samples per plant were analysed by anova, with plants as replicates. Variables tested comprised cultivars, inoculation methods and infection classes (0 vs. 1–15%).

Results

Contamination of stolons, tubers and stems during vegetative growth

Comparison of detection methods

PCR applied on DNA extracted from organs of symptomless plants did not show any amplification after migration on agarose gels, irrespective of cultivar or inoculum concentration. PCR tests applied on samples from diseased organs to verify the presence of Eca gave 75 and 100% of positive results for Bintje stems from plants inoculated with 4.5 × 106 and 4.05 × 108 c.f.u. mL−1, respectively, and 100% for Désirée stems at both concentrations. When PCR was applied to diseased tuber samples, only 12.5% of the Bintje samples were positive for the 4.5 × 106 c.f.u. mL−1 inoculum concentration, and none at 4.5 × 108 c.f.u. mL−1. None of the diseased Désirée tubers gave a positive signal by PCR, irrespective of the initial inoculum concentration. Finally, none of the DNA samples extracted from diseased stolon samples yielded an amplification product, irrespective of cultivar or inoculum concentration. Because of its poor performance, PCR detection was not attempted on symptomless parts of diseased plants.

Early infection of stolons and daughter tubers

DAS-ELISA tests applied on symptomless organs showed that stolons were frequently infected, whatever the disease status of the plant, the inoculum concentration and the cultivar. Stolons were also the organs most frequently infected in control plants of both cultivars ( Table 1).

Table 1.  Proportion of apparently healthy organs from plants of two potato cultivars, grown from seed tubers artificially infected by infiltration, and latently infected with Erwinia carotovora ssp. atrosepticaaOrgans were collected from field plots shortly after tuber induction.bNo, no symptoms seen on any part of the plant; Yes, symptoms present on the plant but not on the organs sampled.cResults show the frequency of positive tests following detection with two serological methods for eight different plants per cultivar, initial inoculum concentration and plant status combination. Thumbnail image of

No significant effects of cultivar (F = 0.85, P = 0.378), inoculum concentration (F = 2.36, P = 0.156), or interaction between these two variables (F = 0.09, P = 0.765) on the percentage of latent infections of young daughter tubers were detected by anova, and no interaction was detected between cultivar and diseased status of the plant (F = 0.09, P = 0.765). However, symptomless tubers from diseased plants showed a significantly higher proportion of latent infections than those from symptomless plants (F = 34.06, P < 0.001), irrespective of the cultivar considered (F = 0.09, P = 0.765).

Stem infections

Latent infections were significantly more frequent in basal parts of the stems of cv. Bintje than in those of cv. Désirée (F = 6.92, P = 0.025), and in symptomless parts of diseased plants (F = 93.08, P < 0.001); inoculum concentration did not significantly affect the proportion of latent infections (F = 0.77, P = 0.401) ( Table 1). No interactions were detected between cultivar and inoculum concentration (F = 3.08, P = 0.110), nor between cultivar and disease status (F = 3.08, P = 0.1099). Including control plants in the anova did not affect these conclusions (data not shown).

The highest proportion of latent infections on individual diseased stems was detected using DAS-ELISA with an enrichment step. Sensitivity of PCR and standard DAS-ELISA on these samples was comparable ( Fig. 2). In both cultivars the bacteria were always detected within the visibly diseased tissue in classes II and III. For stems of class I, bacteria were always detected at ground level (3–5 cm above the necrotic tissue) in Bintje, but only in 57.1% of Désirée stems. The bacterium was detected in over 50% of the samples collected 10–15 cm above visible symptoms in both cultivars. This proportion was higher in Désirée than in Bintje for plants in classes II and III ( Fig. 2), although the presence of a single Désirée plant in class III should lead to caution in interpreting this observation. Bacteria could be detected further from the symptoms in a lower proportion of the samples. Eca was detected in all samples taken below the visible symptoms in the eight stems of class III. No infection was detected on samples taken from control plants, whatever the detection technique used.

Figure 2.

Proportion of samples collected at different heights along individual stems with symptoms, showing latent infections by Erwinia carotovora ssp. atroseptica, according to cultivar, symptom pattern and detection method. The position of samples along the stems is coded according to the diagrams in Fig. 1. The number of samples tested is indicated in brackets. White bars, cv. Bintje; grey bars, cv. Désirée.

Latent infection of progeny tubers at harvest

The highest proportions of latent infections in all groups of plants were detected using E-DAS-ELISA ( Table 2). At least one of the 35 tubers tested individually in each group gave a positive detection with one or other method. ANOVA revealed no significant effects of cultivar (F = 0.29, P = 0.5941) or of inoculation mode (F = 0.29, P = 0.5941) on the proportion of samples testing positive, but symptomless tubers from plants with diseased progeny (infection class 1–15%) were latently infected in significantly higher proportions than were symptomless tubers from apparently healthy progenies (infection class 0) (F = 12.16, P < 0.001). Very high proportions of latent infections were recorded on the collective samples prepared from the progeny tubers not tested individually ( Table 2).

Table 2.  Proportion of apparently healthy tubers, harvested from field plots, from plants of two potato cultivars grown from seed tubers artificially infected, and latently infected by Erwinia carotovora ssp. atrosepticaaStandard deviation of means.bData represent the frequency of positive tests among five individual tubers per plant for seven plants per cultivar, inoculation method and infection class.cData represent the frequency of positive tests calculated over seven different plants per cultivar, inoculation method and infection class. Thumbnail image of

Discussion

Early infection of all vegetative organs of potato plants grown under field conditions from seed tubers artificially contaminated with an Eca strain were shown to occur during the growing season. This is explained by the persistence in the field of conditions favouring the development of the disease, through frequent irrigation. Stolons were always contaminated, and tuber tissue close to stolon attachment sites frequently harboured detectable populations of the bacterium at harvest, showing that internal contamination of progeny tubers via stolons is an important contamination pathway. The high proportion of latent infections in collective tuber samples at harvest is probably explained by the fact that collective samples have a higher probability of containing at least one contaminated tuber, but may also reflect the higher contamination rates of small (young) tubers.

The constant detection of Eca in stem sections below visible aerial symptoms shows that stems were infected from the mother tubers and that bacteria moved internally up the stems. Visual observations underestimate the proportion of contaminated plants, as latent infections of symptomless tubers or stems were frequently detected. These observations support and extend those of Allefs et al. (1996 ), who demonstrated that Eca and E. chrysanthemi could be detected in stem tissue, 1 cm up from the mother tuber, 3 days after mother tubers were inoculated. Bacteria were also detected in apparently healthy parts of diseased stems. Detection rates obtained for sections of stems belonging to class I were always lower than those of the corresponding sections of stems from the other classes, possibly because of the 3–5 cm interval between the sections collected and visible symptoms.

Several investigations have demonstrated the possibility of infection of aerial plant parts by insect transmission from diseased foliage of other plants during the cropping season ( Harrison et al., 1977 ; Kloepper et al., 1981 ), or by or wind-dispersed aerosols ( Graham et al., 1977 ; Quinn et al., 1980 ). These studies led to the classification of stem rot symptoms as either ‘typical blackleg’, originating predominantly from tuber-borne inoculum, or ‘aerial stem rot’, thought to result primarily from airborne infections ( Pérombelon & Kelman, 1987). However, the presence of the bacterium in all stem portions below aerial symptoms in plants of class III, corresponding to the aerial stem rot type of symptom distribution, supports the possibility that both symptom types could derive from tuber-borne infections. Observations showed that symptoms were predominantly located at weak points of the stems, such as those induced by natural or artificial wounding during scoring operations.

No differences were observed between the two cultivars Bintje and Désirée in levels of latent infection of tubers collected either during the growing season or at harvest, despite the much lower development of the disease in cv. Désirée ( Hélias et al., 2000 ). This observation suggests that different levels of bacterial populations are required to incite similar symptoms on cultivars with different resistance levels. This is in accordance with the data of Bain et al. (1990 ), who determined that the mean number of Eca present on seed necessary to induce blackleg varies with cultivars. However, significant cultivar-related differences were noticed for stem contaminations. This observation can be related to the differences in resistance expression of the cultivar depending on the organ tested. Several authors noted that one of the difficulties in screening cultivars for resistance to pectinolytic Erwinia was related to the lack of correlation between stem and tuber resistance ( Allefs et al., 1995 ), and to the different performance of inoculation methods, leading to different rankings of potato cultivars ( Bain & Pérombelon, 1988).

The difficulty of detecting the bacterium on diseased organs (stolons and tubers) by PCR could be explained by inappropriate storage of the samples. These samples were stored for longer than those from diseased stems before DNA was extracted. The absence of glycerol in samples as a thermoprotectant for bacteria could have favoured the disintegration of bacterial cells and degradation of DNA during storage of crushed samples. Blackening of crushed samples was frequently observed, especially in diseased organs; it was probably related to the oxidation of plant phenolics, which may explain the poor performance of PCR detection. This phenomenon has been described previously, particularly with cv. Désirée, and resulted in the failure to detect Ralstonia solanacearum by PCR ( Elphinstone & Stanford, 1998). Moreover, De Boer & Ward (1995) showed that few bacteria remain intact in severely infected tissue, which could also explain the poor performance of PCR. Conversely, better performance of DAS-ELISA related to PCR is explained by the same authors by a better stability of the LPS (detected by anti-Eca antibodies) compared to DNA after the death of cells.

The most discriminating results were not always obtained with the same detection method. This probably reflects the fact that all current methods are basically threshold-based, allowing the detection of a number of bacteria corresponding to their sensitivity threshold. However, the most important criterion for epidemiological studies is the actual number of bacteria present at a given site. Since the size of epidemiologically significant populations probably depends on the organ, and most certainly on the cultivar considered, it is important to combine methods with different sensitivity thresholds to obtain a comprehensive view of the epidemiological relationships between Eca and potato plants. For instance, the number of bacteria estimated by IFC was less than 1 × 103 c.f.u. in symptomless stem samples, compared with at least 1 × 106 c.f.u. in stems showing blackleg ( Allefs et al., 1996 ), which is consistent with our own data obtained with E-DAS-ELISA on symptomless and diseased stems. However, a better investigation of these relationships would require the development of quantitative detection methods, allowing direct assessment of bacterial numbers on a given plant part.

Acknowledgements

We thank C. Guérin, I. Glais and S. Guilloteux for technical assistance.

Ancillary