INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Ralstonia (formerly Pseudomonas) solanacearum (E. F. Smith) Yabuuchi et al. (47) is the causal agent of bacterial wilt, a severe and devastating plant disease in most tropical and subtropical and some warm temperate areas (22). Moreover, it can also occur in cool temperate areas (9, 33). Many economically important food crops such as potatoes, tomatoes, and bananas are affected. The disease was recorded on several hundred plant species distributed in more than 50 families (23). The species R. solanacearum is a complex taxonomic unit in which strains display an important diversity at different levels (physiological, serological, genetic characteristics, and host range). In order to describe this intraspecific variability, several systems of classification have been proposed. Thus, the species was subdivided into five races according to its host range (7, 25, 35) and into six biovars based on the utilization of three disaccharides and three hexose alcohols (21, 24, 25). Fatty acid analysis (26, 42) and protein profiling (15) were also performed but did not further clarify the relationships among R. solanacearum strains. Restriction fragment length polymorphism (RFLP) analysis (involving nine probes, seven of which encode information required for virulence and the hypersensitive response) (12-14) has provided a new classification scheme dividing the species into 46 groups in relation to geographic origin of strains and sometimes host range. The species was then separated into two major groups, the "Asiaticum" and the "Americanum" divisions, which regrouped strains from Asia and America, respectively. Further investigations comparing sequences of 16S rRNA (30, 40, 43) or using PCR amplification with tRNA consensus primers (39) supported the separation according to geographic origin.

Only a few strains originating from Africa, and only one from Reunion Island (21), were included in these previous studies. However, strains related to the three major biovars (1, 2, and 3) were isolated from various crops in Reunion (17). The aim of our study was to assess the genetic diversity within the local populations of R. solanacearum. Since we also wanted to develop molecular tools for the identification and detection of R. solanacearum biovars, we used the PCR-RFLP procedure to analyze the diversity. Recently, several authors have successfully performed PCR-RFLP analysis to assess genetic diversity among bacterial species (27, 28, 31, 44, 45). The hrp (hypersensitive reaction and pathogenicity) gene region, which is required by many phytopathogenic bacteria to produce symptoms on susceptible hosts and a hypersensitive reaction on resistant hosts or on nonhosts (1, 3, 4, 6, 18, 29), was explored for studying the variability within a collection of 120 strains isolated from different hosts over the five continents and belonging to biovars 1, 2, 3, and 4.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. Strains studied (Table 1) included diverse strains of R. solanacearum with special attention to those isolated from Africa (51 strains including 28 from Reunion Island) and strains belonging to more or less closely related species (Ralstonia pickettii, Ralstonia eutropha, Burkholderia cepacia, Pseudomonas spp., Xanthomonas spp., Erwinia chrysanthemi, and Clavibacter michiganensis subsp. michiganensis). Identification of R. solanacearum strains at the biovar level was performed by using a modification of Hayward's method (21). All strains were stored on beads in cryovials at 80°C (Microbank Pro-Lab Diagnostics). Nutrient broth cultures were grown for 24 h on a rotary shaker (150 rpm) at 28°C. Bacteria were cultivated either (R. solanacearum) on a modified Granada and Sequeira medium (19) (tryptone, 1 g/liter; peptone, 10 g/liter; agar, 18 g/liter; glycerine, 6.3 ml/liter; crystal violet, 0.002 g/liter; polymyxin sulfate, 0.01 g/liter; tyrothricin, 0.02 g/liter; chloramphenicol, 0.005 g/liter; triphenyltetrazolium chloride, 0.025 g/liter; propiconazole, 0.4 ml/liter; penicillin, 20 U/liter; pH 7.2) or (other species) on YPGA medium (yeast extract, 7 g/liter; peptone, 7 g/liter; glucose, 7 g/liter; agar, 15 g/liter; pH 7.2) and incubated for 3 days at 28°C.

DNA extraction. DNA was extracted from R. solanacearum cells grown overnight at 28°C in 30 ml of YP (yeast extract, 7 g/liter; peptone, 7 g/liter) by the hexadecyltrimethylammonium bromide method (2). DNA concentration was estimated by fluorometry (TKO 100 minifluorometer; Hoefer Scientific Instruments, San Francisco, Calif.).

DNA amplification. Pairs of primers from the nucleotide sequence of the hrp gene region of strain GMI1000 of R. solanacearum (accession no. Z14056 for EMBL-GenBank-DDBJ databases) were designed with Oligo 5.0 software (32). Eleven pairs were selected in order to explore the whole region. They delineated fragments with sizes ranging from 213 to 2,456 bp. Primers were synthesized by Genosys Biotechnologies, Cambridge, England.

Restriction fragment analysis. The amplified DNA fragments considered to be specific to R. solanacearum were digested with restriction endonucleases according to the manufacturer's directions (GIBCO BRL; Boehringer Mannheim). Enzymes were chosen on the basis of the nucleotide sequence of the hrp gene region of strain GMI1000 by using Oligo 5.0 software (32). Among all considered enzymes, only 13 available in the laboratory were retained: AvaI, BglII, BssHII, EclXI, EcoRI, HaeII, HindII, NotI, PstI, PvuI, PvuII, SacI, and SmaI. Restriction fragments were separated by electrophoresis and visualized as described previously (37).

Data analysis. Data derived from the different RFLP patterns exhibited by the tested strains (presence or absence of bands) were used for a hierarchical cluster analysis (HCA). With Statlab software (41), clustering was based on the Euclidean distance between strains (Ward's method [46]). The truncation level in the resulting dendrogram was thus determined to be that which provided the smallest number of clusters for which the variance within clusters was significantly (P = 0.05) different from the variance between clusters.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Specificity of primers to the hrp gene region of R. solanacearum. A suitable amplification pattern was obtained with only 6 of the 11 pairs of primers, giving fragments ranging from 213 to 1,993 bp (according to the sequence of the reference strain GMI1000). The amplified fragments were distributed along the hrp gene region and cover a variable number of genes (two to four) which have been previously defined: RS20-RS201, 1,452 bp over hrpV-hrpU-hrpT-hrpQ; RS30-RS31, 1,993 bp over hrpO-hrpN; RS50-RS501, 1,200 bp over hrpN-hrpK-hrpJ-hrpI; RS600-RS61, 905 bp, and RS80-RS81, 1,537 bp, both over hrpC-hrpB (Fig. 1); and RS90-RS91, 213 bp over hrpB-hrpA.

View larger version (21K): [in this window] [in a new window]   FIG. 1.   Location of the five selected pairs of primers (RS20-RS201, RS30-RS31, RS50-RS501, RS600-RS61, and RS80-RS81) within the hrp genes of R. solanacearum (strain GMI1000). (a), number of the base on the 5'-3' DNA sequence; (b), number of the base at the 5' end of the primer.

Restriction endonuclease analysis of specific amplified hrp sequences. The six amplified DNA fragments, which were considered to be specific to R. solanacearum, were digested with the 13 selected restriction endonucleases. Different restriction patterns among the 120 strains of R. solanacearum were observed with different restriction endonucleases: AvaI (four patterns) and PvuII (four patterns) for the RS20-RS201 sequence; HindII (six patterns) for the RS30-RS31 sequence; SacI (three patterns) for the RS50-RS501 sequence; HaeII (six patterns) for the RS600-RS61 sequence; BssHII (six patterns), NotI (three patterns), and PstI (four patterns) for the RS80-RS81 sequence (Fig. 2).

View larger version (70K): [in this window] [in a new window]   FIG. 2.   Restriction patterns (see explanation in Table 2) of the five amplified fragments of the hrp gene region of R. solanacearum when digested by the designated enzymes. (A) RS20-RS201 AvaI (left side) and PvuII (right side); (B) RS30-RS31 HindII; (C) RS50-RS501 SacI; (D) RS600-RS61 HaeII; (E) RS80-RS81 BssHII (left side), NotI (right side), and PstI (right side). M, molecular size markers (100-bp ladder; GIBCO BRL). The size (in base pairs) of the bands was estimated from the sequence of the hrp gene region of the GMI1000 strain.

Clustering of the PCR-RFLP profiles. Among the 120 strains of R. solanacearum, 15 different profiles could be distinguished (Table 2). A profile was the result of the combination of the RFLP patterns given by the eight restriction endonucleases which generated polymorphism and which were selected for the data analysis. The HCA resulted in a dendrogram showing the genetic relatedness between strains (Fig. 3). The truncation level allowed separation of eight PCR-RFLP clusters designated clusters I to VIII. While five clusters (II, III, IV, V, and VIII) contained a unique profile, the three remaining contained two (cluster VII), three (cluster I), and five (cluster VI) different profiles. The number of strains in each cluster ranged from 1 to 40, but each cluster contained strains belonging to the same biovar, with the exception of cluster VI, which included strains of biovars 3 and 4. Biovar 1 strains were distributed over five clusters (I, II, III, V, and VII), biovar 3 was grouped into both cluster VI and cluster VIII (only one strain within the latter), and the 36 strains belonging to biovar 2 were grouped together in cluster IV.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Dendrogram resulting from an HCA based on the restriction patterns of the five amplified fragments within the hrp gene region of 120 strains of R. solanacearum. a, the relative distance between the farthest clusters was assumed to be 100.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   Location of the 36 restriction sites identified in the five amplified fragments of the hrp gene region of R. solanacearum when digested by AvaI and PvuII for RS20-RS201 (A); by HindII for RS30-RS31 (B); by SacI for RS50-RS501 (C); by HaeII for RS600-RS61 (D); and by BssHII, NotI, and PstI for RS80-RS81 (E), as estimated from the DNA sequence of strain GMI1000 and the size of the bands of the restriction patterns. a, number of the base at the 5' end of the primer; b, number of the base at left of the restriction site; c, the size of the amplified fragment was higher for the four strains of cluster V.

Variability in restriction sites in the hrp gene region of R. solanacearum. Thirty-six restriction sites were identified within the five amplified fragments when the eight selected enzymes were used (Fig. 4). Eight sites appeared to be common to all 120 strains (A5, A7, B4, B5, E12, E13, E14, and E15). The occurrence of the 28 remaining sites was variable according to the RFLP clustering of the strains. However, 14 among them appeared to be particularly remarkable since they could separate the biovars and/or the geographic or botanical origins (Table 4).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The exploration of the hrp gene region with 11 selected pairs of primers gave six amplicons which were confirmed to be specific to R. solanacearum. Indeed, no amplification was observed with DNA from strains belonging to other bacterial species and even from such closely related species as R. pickettii, R. eutropha, and B. cepacia. Consequently, the hrp region seems to be useful for the identification and specific detection of strains of R. solanacearum. However, since we did not succeed in getting strains of Pseudomonas celebense and Pseudomonas syzygii from laboratory collections, which are also species close to R. solanacearum, the amplification within their hrp region when the selected primers were used was not checked. Nevertheless, since our main objective was to develop molecular tools for the detection of populations of R. solanacearum on Reunion Island and since these particular pathogenic species (P. celebense and P. syzygii) were recorded only in Indonesia on bananas and cloves, respectively, a lack of specificity in that case would be of no consequence.

Although we concentrated mainly on the strains originating from Reunion Island (28 strains), the remaining 92 isolates were chosen to represent the broad host range, wide geographic distribution, and metabolic diversity (biovars) of R. solanacearum. Although our analysis gave a lower resolution level than that seen after genomewide RFLP analysis (12-14), we found that it gave reliable estimates of phylogenetic relationships among strains of R. solanacearum. Whereas the 46 described RFLP profiles were correlated with geographic origin, biochemical typing, and host origin, we identified 15 PCR-RFLP profiles distributed into eight clusters, these clusters being correlated with biochemical typing and to a lesser degree with geographic origin. Since some PCR-RFLP patterns correlated well with biovar typing, the PCR-RFLP procedure provides a complementary or alternative method for biovar determination.

The PCR-RFLP analysis confirmed the great variability within R. solanacearum. Biovar 1 strains showed the greatest diversity since they were distributed into five of the eight clusters. Six biovar 1 strains originating from the Musaceae family were distributed into two specific clusters, one (cluster V) including four of these strains and the other (cluster III) comprising the two remaining strains together with one strain isolated from potato. Among the 36 restriction sites identified on the five amplified fragments, only 25 were common to the musaceous strains, whereas there were 28 sites common to biovar 2 and biovar 3 strains. Moreover, among the 12 discriminating restriction sites located on the fragment delineated by the RS80-RS81 primer pair, only three were common to both clusters III and V while all were shared by clusters III and IV. These features suggest that there are important differences between the musaceous strains distributed in two separate clusters. All of these strains came from Central America or northern South America, but nothing was known of their pathogenicity, and no clear indication of the race to which they belonged was reported. The strains of cluster V could belong to race 2, whereas those of cluster III might be associated with race 1, which could explain the presence of the Colombian strain isolated from potato within the cluster. One particular strain isolated from Musa sp. and characterized as belonging to biovar 3 (36) fell into cluster VI, as most of the strains were related to the same biovar. Further studies incorporating more strains isolated from Musaceae and belonging either to race 1 or to race 2 are required.

Clusters I and II included all biovar 1 strains originating from the Americas, more specifically, either from North America for cluster I or from Central America for cluster II. The fact that four African isolates fell into these clusters suggests that they could have been introduced from the Americas. Most biovar 1 strains isolated from African countries, however, were included in cluster VII. All of these strains originated from southern Africa, including Reunion Island, Madagascar, Zimbabwe, and Angola, whereas the African isolates from clusters I and II came from the northern part of Africa (Burkina Faso and Kenya). Thus, Africa may have two different biovar 1 populations, either endemic and commonly isolated in southern countries or introduced from the Americas through direct or indirect commercial exchanges. Although both populations belonged to the same biovar, there was no indication that they have similar host ranges (and/or similar virulence).

The 36 strains of biovar 2 displayed a similar profile which was characterized by the specific restriction pattern DHa2 and were included in cluster IV, close to those encompassing the biovar 1 strains of American origin. The consistent homogeneity of biovar 2 strains, although they were collected from 20 countries distributed worldwide, could be attributed to their narrow host range, including only potato and tomato plants. The result agrees with the commonly accepted hypothesis of a common origin for all the biovar 2 strains. South America is the presumed origin, and the wide distribution of these strains is probably due to the dissemination of latently infected plant material (particularly potato tubers) by humans (8, 10, 22). The biovar 2 strains (cluster IV) are closely related to cluster III strains, since 34 of the 36 restriction enzyme sites were common to both clusters, underlining the proximity of some musaceous isolates to race 3 strains.

Compared to biovar 1 strains, biovar 3 strains showed rather modest genetic diversity since they could be assigned to one major cluster. An additional cluster with a unique strain originating from Japan was also described. The few biovar 4 strains fell into the same cluster as most biovar 3 strains, indicating that there were only slight differences between these biovars. However, six different profiles more or less correlated with geographic origin (Asia, cluster VIa; Reunion Island, cluster VIb; America, cluster VIc) were identified.

The dendrogram resulting from an HCA revealed the separation of R. solanacearum into two major divisions. This result confirmed the conclusion of many previous studies on DNA homologies and physiological characterization of strains (20, 34) and more recently of RFLP analysis (12-14), of 16S rRNA sequencing (30, 40, 43), or of PCR amplification with tRNA consensus primers (39). The first division, Americanum sensu Cook et al. (12), contains biovar 1 and 2 strains, and the second division, Asiaticum sensu Cook et al. (12), includes biovar 3 and 4 strains. Thus, compared to other genomic regions (16S rRNA and tRNA), the hrp gene region, which is involved in host-pathogen interactions, revealed the same major trend of diversity, suggesting that hrp genes have evolved in parallel with 16S rRNA and tRNA.

The amplified fragment delineated by the RS80-RS81 pair of primers provided much more polymorphism than all the others: 12 discriminating restriction sites were identified and permitted separation of certain groups of strains. All of these polymorphisms were located within the hrpB regulatory gene (16). This observation confirms that regulatory systems of bacteria seem to be less conserved than those genes whose function they govern (5). Furthermore, this result suggests that the hrp regulatory gene may have other metabolic functions besides its role in pathogenic diversity of R. solanacearum. More precise analysis of the hrpB gene in different strains in the future might provide a useful way of relating pathogenicity gene function to genetic diversity.

Although the dendrogram confirms the separation of R. solanacearum into two groups, the distribution of biovar 1 strains, which displayed a rather wide variability, did not agree completely with the scheme proposed by Cook et al. (12). Clusters I, II, III, and V (American biovar 1 strains) were close to cluster IV, which included biovar 2 strains, and would thus belong to the Americanum division, whereas cluster VII (African biovar 1 strains) near cluster VI (biovar 3 and 4 strains) would be separated and connected rather to the Asiaticum division. The African strains included in cluster VII could have evolved separately as a result of geographic isolation and thereby have contributed to increasing the diversity of the species. Clearly, further analysis with other techniques such as DNA probes for RFLP analysis (12-14) and/or 16S rRNA sequencing (30, 40, 43) to confirm other characteristic features of these strains would be of interest. Preliminary results obtained with the R. solanacearum-specific primer pair PS96-H and PS96-I (38) support the hypothesis of separate evolution of these strains, since these primers never led to amplification of any biovar 1 strain originating from Reunion Island, Madagascar, Zimbabwe, or Angola (data not shown). Whatever the explanation, these African biovar 1 strains shared more sites with biovar 3 strains (23 sites) than with the American biovar 1 strains (13 to 19 sites according to the cluster), and the conclusion is that American and African biovar 1 strains are phylogenetically distinct, the latter being more closely related to Asiatic (biovar 3 and 4) strains.

Our study of the genetic diversity of the hrp gene region of R. solanacearum thus provides discriminating tools which besides being useful for fundamental research can also be used for diagnostic purposes. For example, biovars 1 and 2 and the combination of biovars 3 and 4 can easily be distinguished from each other by the restriction pattern generated after amplification with the RS600 and RS61 primers when digested by HaeII: DHa3 for biovars 3 and 4; DHa2 for biovar 2; and DHa1, DHa4, DHa5, or DHa6 for biovar 1. Moreover, the restriction pattern could give useful information about the geographic origin of the biovar 1 strain. As PCR amplification is known to be a very sensitive technique, such primers could be used to detect the populations of R. solanacearum in plant, irrigation water, or soil extracts. They could also be employed to clarify some aspects of the epidemiology of bacterial wilt regarding, for instance, seed as a vehicle of disease spread or some weeds or resistant plants as possible carriers of low levels of infectious populations.