INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Agrobacterium sp. is a pathogenic bacterium responsible for two plant diseases: crown gall and hairy root. As these names suggest, the visible symptoms at the infection site are the appearance of tumorous overgrowths and roots for crown gall and hairy root, respectively. Both diseases are examples of natural interkingdom genetic exchange, because the infectious process relies on the transfer of a DNA fragment(s) from the prokaryote Agrobacterium to the eukaryotic plant cells. This transferred DNA, or T-DNA, is borne on extrachromosomal bacterial replicons. These replicons are the Ti (tumor-inducing) plasmid found in bacteria responsible for crown gall disease, and the Ri (root-inducing) plasmid found in bacteria responsible for hairy root disease. Once transferred to the plant, the T-DNA integrates into the nuclear genome of the cell, where T-DNA genes are transcribed. The molecular mechanism underlying the transfer of DNA has been extensively reviewed (e.g., see references 11, 27 and 41).

Genes located on the T-DNA fall into two groups. The first one includes genes responsible for tumor or root formation (for reviews, see references 4 and 18). The second group of T-DNA genes encode enzymes catalyzing the synthesis of the low-molecular-weight compounds specific for the crown gall or hairy root cells. These compounds, termed opines, generally result from the condensation of amino acids and alpha-ketoacids, or aminoacids and sugars; they play a key role in the ecology of the plant-Agrobacterium interaction (for reviews, see references 12 and 13). The combination of opines, the synthesis and the degradation of which are due to genes borne on Ti and Ri plasmids, provides the basis for a simple classification of the pathogenic plasmids of Agrobacterium (4, 13). However, data collected from the analysis of Ti plasmids isolated from grapevine isolates strongly suggest that these plasmids are mosaic plasmids, with conserved and variable regions (30, 31, 52).

It appears that the type of disease induced by Agrobacterium depends on the type of plasmid hosted by the bacteria. In this respect, the former delineation of Agrobacterium species based on the disease symptoms, hence on traits due to plasmid-borne genes, is of little value (for a review, see reference 51). A stronger classification of Agrobacterium species has been performed using numerical taxonomy of phenotypic properties (22, 54), analysis of fatty acid methyl ester profiles (20, 44), or comparison of electrophoregrams of soluble proteins (23). These results indicate clearly that the genus Agrobacterium can be divided into three different clusters which correspond to biovars 1, 2, and 3, as termed by Keane et al. (21). Biovar 3 is now regarded as the Agrobacterium species A. vitis, which includes strains isolated from grapes (29). Similarly, biovars 1 and 2 could define different species of Agrobacterium. Further studies will be crucial to confirm or refute this hypothesis. Such studies may lead to a deep reorganization of the Rhizobium-Agrobacterium clusters within the family Rhizobiaceae, since some Agrobacterium strains have more characteristics in common with Rhizobium than with Agrobacterium (51).

Among commonly infected plants, grapevine is of major commercial importance. In France, grapevine galls have been reported in cold parts of the Rhone Valley, but also in the Bordeaux and Loire Valley regions (39). The spread has resulted from a combination of cold climatic conditions and the poor sanitary status of the cultivated material (3, 7, 8, 17, 25, 26, 28, 33, 40, 46, 48, 50, 55, 56; for a review, see reference 14). A better characterization of the Agrobacterium strains would facilitate their routine identification and subsequent control of plant sanitary conditions. To this end, we have collected 61 isolates from grapevine galls and analyzed their traits due to both plasmid-borne and chromosome-encoded genes with respect to other Agrobacterium strains, including reference strains. The results of this study are reported below.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Bacterial strains. Out of 111 Agrobacterium strains used in this study, 88 were isolated in France between 1976 and 1989, and 23 were of various origins and deposited in the French Collection of Phytopathogenic Bacteria (CFBP). Two clinical isolates were obtained from the Pasteur Institute (Paris, France) (Table 1). Agrobacterium isolates were grown on LPGA medium (38) which consisted of yeast extract (Difco Laboratories, Detroit, Mich.), 5 g/liter; Bacto Peptone (Difco), 5 g/liter; glucose, 10 g/liter; and 15 g/liter (pH adjusted to between 7 and 7.2).

Biochemical characters for presumptive diagnosis of Agrobacterium. Gram strain response was determined using the aminopeptidase test from Merck (Darmstadt, Germany). The following conventional biochemical characteristics were assessed according to the method of Popoff et al. (36): presence of esculin--glucosidase, urease (in urea-indol medium; Diagnostics Pasteur, Marne-la-Coquette, France), orthonitro-phenyl--D-galactopyranoside (ONPG) -galactosidase, gelatinase, Tween 80 esterase, DNase on DNA agar (Diagnostics Pasteur). 3-Ketolactose production (according to Bernaerts and De Ley [2]) and phenylalanine desaminase (PAD) activity were also assayed. PAD detection was carried out on phenylalanine agar, which was made of DL-phenylalanine, 2 g/liter; yeast extract (Difco), 3 g/liter; NaCl, 5 g/liter; K₂HPO₄, 1 g/liter; and agar, 12 g/liter. Agrobacterium strains were streaked on this medium to a high density and kept at 26 to 27°C. After 40 to 48 h, the culture was covered with a few drops of FeCl₃ (density, 1.26) diluted 1/3 (vol/vol) with distilled water. A positive assay is indicated by an olive-green coloration appearing rapidly and remaining stable for 1 to several hours. Characteristics presumptive for Agrobacterium species were confirmed for all assayed strains using the identification system for Pseudomonas and related bacteria (Diagnostics Pasteur). This system also gave data on nitrate and arginine metabolism.

Nutritive characteristics. Utilization (with acid formation) of melizitose, dulcitol, erythritol, and ethanol and utilization (with alkali formation) of L-(+)-tartrate and malonate were assayed. These compounds were added at 1% (vol/vol or wt/vol) to the minimal medium, which consisted of NH₄H₂PO₄, 1 g/liter; KCl, 2 g/liter; MgSO₄ · 7H₂O 0.2 g/liter; yeast extract (Difco), 0.1 g/liter; and bromothymol blue, 0.08 g/liter (pH 7.2) (1). Five milliliters of this medium inoculated with Agrobacterium strains using 48-h precultures performed on LPGA medium (38), and incubated in a shaker (120 rpm) at 27°C. Growth and acid production were generally stopped after 72 h of incubation but for some strains were stopped after 5 days of incubation.

Digital-numerical taxonomy. A total of 167 characteristics (based on 20 biochemical and physiological tests plus assimilation of carbon sources) were included in the digital-numerical taxonomy analysis. A distance matrix was calculated using the Jaccard coefficient (47). Cluster analysis was done by using the unweighted pair group method of average with arithmetic mean (47).

Pathogenicity assays. Three plant species were used: Kalanchoe tubiflora, Datura stramonium, and Lycopersicon esculentum (var. Montfavet 63/5). These were kept in a growth chamber at a day temperature of 23°C and a night temperature of 18°C, with a 16-h light, 8-h dark photoperiod and a relative humidity of 80 to 85%. Suspensions of the bacteria to be assayed were made in sterile water and adjusted to ca. 10⁸ CFU/ml. Of these suspensions, 50-µl aliquots were used to inoculate the plants wounded at the second, fifth, and sixth internodes starting from the apex (K. tubiflora) or at the second and fourth internodes (D. stramonium and L. esculentum) at the stage when four leaves had expanded. The reactivity to inoculation was estimated after 40 days to differentiate the various types of reaction, particularly on K. tubiflora. Appearance of tumorous outgrowths was assessed by visual inspection of the inoculated plants.

Opine detection in the tumors and opine utilization by the bacteria. Detection of opines in tumorous tissues and their utilization by the inducing bacteria were performed by using high-voltage paper electrophoresis, as reviewed by Dessaux et al. (12).

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Analysis of phenotypic, chromosome-encoded characteristics. (i) Identification of Agrobacterium strains. All assayed strains (n = 111) exhibited ONPG-hydrolase (-galactosidase) and urease activities, and were able to degrade esculin. This confirmed that these strains belonged to the genus Agrobacterium (24). Additionally, the assayed strains were not able to degrade gelatin or to reduce tetrathionate. It is noteworthy that a negative response for the DNase and Tween esterase assays cannot be used as an orientation test for identifying Agrobacterium strains because, out of 111 assayed strains, 12 exhibited DNase activity while 6 produced a Tween esterase.

(ii) Numerical taxonomy. The dendrogram displaying the distance relationships amongst the 111 strains included in this study is shown in Fig. 1. At a phenotypic distance of 0.3, three major and one minor phena were delineated. The major phena 1, 2, and 3 precisely group strains of the three biovars, 1, 2, and 3, respectively. Phenon 4 included three strains, CFBP 2724, 2725, and 2771. Although these strains clustered with biovar 2 strains at a distance of 0.354, they must be regarded as atypical since they exhibit many characteristics which are not common to those of biovar 2 strains (Table 2). Whether the three above-mentioned strains are related to those described by Bouzar et al. (6) remains to determined. At a shorter distance (0.254), phenon 3 divided into two subphena (3a and 3b) which comprised, respectively, 36 and 9 strains, leaving 2 isolated strains (CFBP 2617 and CFBP 2678). At the same distance (0.254), phenon 2 divided into two subphena (2a and 2b) which comprised, respectively, 31 and 2 strains. Strains isolated from grapevines clustered as follows: 10 strains in phenon 1 (which includes 28 strains), 1 strain in phenon 2 (which includes 33 strains), 47 strains in phenon 3 (which includes 47 strains), and 3 strains in phenon 4 (which includes 3 strains). Overall, and except for the three strains CFBP 2724, 2725, and 2771, biovar determination yields clear-cut results. The perfect correspondence between phena 1 and 2 and biovars 1 and 2, respectively, strongly suggests that biovars 1 and 2 could correspond to two distinct species. Bouzar (5) and Sawada et al. (45) previously made this proposal.

View larger version (33K): [in this window] [in a new window]   FIG. 1.   Phenotypic analysis of 111 Agrobacterium strains. Results are presented as a dendrogram based on phenotypic distance value calculation using the Jaccard coefficient and the unweighted pair group method of average with arithmetic mean method. V, from grapevine.

(iii) Differential characteristics. The characteristics that differentiate the three phena and the three strains of phenon 4 are shown in Table 2. Ten assays (3-ketolactose production, presence of oxidase, presence of PAD, and utilization of dulcitol, melezitose, L-rhamnose, malonate, propionate, citrate, and L-ornithine) have been used previously by Kersters and De Ley (24) to differentiate among biovars of Agrobacterium. As shown in Table 2, 12 additional characteristics could be used to identify the biovars and the three isolated strains. Interestingly, our results confirm the validity of using the 3-ketolactose criterion to identify biovar 1 strains, since all these strains produced this lactose derivative (Table 2).

Analysis of traits due to plasmid-borne genes. (i) Utilization of L-(+)-tartrate. Utilization of L-(+)-tartrate yielded positive results for all biovar 2 and 3 strains and only for the biovar 1 strains isolated from grapevine tumors. Though it has been demonstrated that utilization of L-(+)-tartrate is characteristic of many plant-pathogenic bacteria, Szegedi (49) suggested that the degradation of this compound by A. vitis (biovar 3) strains might be due to their adaptation to grapevines. Indeed, tartaric acid is a major chemical component of grapevines (37, 42). Our results are consistent with Szegedi's hypothesis.

(ii) Pathogenicity assays. The results of the pathogenicity assays are summarized in Table 1. Only ca. 60% of the biovar 1 strains induced tumors upon inoculation of tomato plants and daturas. Among the strains which were nonpathogenic on tomato plants, seven were isolated from grapevines. Out of these seven strains, three induced overgrowths on daturas, suggesting a possible host range limitation. On the other hand, biovar 1 strains isolated from other host plants (Prunoideae, Pomoideae, Populus sp. and Chrysanthemum sp.) were pathogenic on most if not all test plants, with the exception of strain CFBP 2741 isolated from Prunus rubiera tumors.

(iii) Production and utilization of opines. Opines synthesized in the tumors and opines degraded by Agrobacterium strains were analyzed, and results are summarized in Table 1. Four opine groups can be defined from the analysis of tumors induced by biovar 1 strains: octopine, nopaline, mannopine-agropine, and cucumopine-octopine. However, two opine degradation patterns were unusual. Firstly, some cucumopine-octopine grapevine strains degraded both opines, while others degraded cucumopine, octopine, and nopaline. Two of these strains (CFBP 2732 and 2682) remained nonpathogenic on the three test plants. Though not formally demonstrated, their opine degradation capability suggests that they do, however, harbor a Ti plasmid. The second unusual degradation pattern was detected in strains that induced mannopine-agropine-type tumors (CFBP 2712, 2713, and 2714): these degraded only mannopine. If this result is not artifactual, it could be attributed either to a mutation, as reported for mannopinic acid utilization in strain 89.10 (16), or to plasmid dissociation (34) or cointegration (53).