Rickettsia rickettsii

Methyiobacterium extorquens Methylobacterium nodulans Bradyrhizobium japonicum

Rhodopseudomonas palustris Paracoccus denitrificans

Sinorhizobium meliloti Rhizobium leguminosarum

Agrobactenum rhizogenes

Mesorhizobium loti Azorhizobium caulinodans

Xanthobacter ay 11 is Azospinllum brasilense

Rickettsia rickettsii

Ralstonia solanacearum Alcaligenes eutrophus Ralstonia taiwanensis outgroup

Figure 1. Unrooted 16SrDNA tree of Proteobacteria. The figure shows the phylogenetic relationships between the different rhizobial genera - as represented by type species in bold - including the new Burkholderia sp. and Ralstonia sp. strains, a, P, 8, y and s represent the different subdivisions of the Proteobacteria. The tree was constructed by using the neighbor-joining method and was adapted from van Berkum and Eardly, 1998.

Nitrosovibrio tenuis

Bordetella pertussis

Rnrkhnlderia sp. ORS1827

Burkholderia vietnamensis Burkholderia sp. STM678 Burkholderia sp. STM815

Burkholderia graminis Burkholderia kurunensis

Ralstonia solanacearum Alcaligenes eutrophus Ralstonia taiwanensis


Neisseria meningitidis

Spirillum volutans Azoarcus indigens outgroup

■ 0.01 substitutions/site

Figure 1. Unrooted 16SrDNA tree of Proteobacteria. The figure shows the phylogenetic relationships between the different rhizobial genera - as represented by type species in bold - including the new Burkholderia sp. and Ralstonia sp. strains, a, P, 8, y and s represent the different subdivisions of the Proteobacteria. The tree was constructed by using the neighbor-joining method and was adapted from van Berkum and Eardly, 1998.

4. a- and P-Rhizobia Use the Same Strategy to Nodulate Legumes

The nodABC genes are responsible for the synthesis of the core structure of the Nod factors and as such are present in all a-rhizobia. P-rhizobia are not an exception, since a nod A gene could be amplified and sequenced in R. taiwanensis LMG19424 and in the three Burkholderia sp. strains STM815, STM678 and ORS1827. Sequence similarity with the different complete rhizobial NodA protein sequences available in databases ranged from 67.5% (STM678A4. caulinodans) to 77.7% (STM678/M nodulans). Further nod gene sequencing in Burkholderia sp. STM 678 and R. taiwanensis LMG14424 revealed a genetic organization of nodABC genes similar to that found in other rhizobia. In Burkholderia sp. STM678 nodAB are in the same orientation and overlapping and preceded by a nodD-dependent regulatory sequence (nod box). A nodC gene was found elsewhere in the genome. In R. taiwanensis nodB is in front of nodC, and preceded by a nod box whereas nodA was found elsewhere in the genome. Such genetic unlinkage of nodABC genes was already described (Zhang et al. 2000). A nodA mutant of Burkholderia sp. STM678, constructed by inserting a lacZ-kanamycin cassette into the nodA gene, did not nodulate M. atropurpureum, indicating that the nodAB genes are required for nodulation of this Burkholderia strain. Moreover this strain has been shown to produce Nod factors, that are N-methylated and 4,6-dicarbamoylated on the non-reducing end but not substituted on the reducing terminus (Boone et al. 1999).

Figure 2. Unrooted nodA tree showing the close phylogenetic relationship between the nodA of a- and P-rhizobia (framed names). The maximum likelihood tree is based on full-length sequences. Al., Allorhizobium, Az., Azorhizobium, B., Bradyrhizobium, Me., Methylobacterium, R., Rhizobium, S., Sinorhizobium, ter., teranga, leg., leguminosarum.

Figure 2. Unrooted nodA tree showing the close phylogenetic relationship between the nodA of a- and P-rhizobia (framed names). The maximum likelihood tree is based on full-length sequences. Al., Allorhizobium, Az., Azorhizobium, B., Bradyrhizobium, Me., Methylobacterium, R., Rhizobium, S., Sinorhizobium, ter., teranga, leg., leguminosarum.

5. Evidence of Multiple Lateral Transfers Between a and fJ-Rhizobia

The discovery of nodulafing bacteria in the P-Proteobacteria phylum raises the question of the origin of nodulating genes, especially common nodABC genes for which hardly any homologous sequences in other organisms have been identified yet. Their occurrence in both a- and P-Proteobacteria cannot be explained in terms of descent through common ancestry, an hypothesis already rejected for the different rhizobial genera within a-Proteobacteria (Young and Haukka 1996). Phylogenetic analyzes indicate a much smaller phylogenetic distance between the nodA genes of P-rhizobia and other rhizobia than between the 16S rRNA genes of a- and P-Proteobacteria. This suggests again that the presence of nod genes in both a- and p-rhizobia occurred through horizontal gene transfer. It was however not clear whether a single transfer spread nodulation genes from one subclass to the other, or if recurrent transfers have occurred between the two subclasses.

To test these hypotheses we performed a maximum likelihood phylogenetic analysis of all available entire nodA sequences, including the four p-rhizobia, using PAUP* (Swofford 1998). The most likely tree obtained clustered the four P-rhizobial strains in 3 different clades (Figure 2). A single transfer would have given a tree with two main clades, discriminating the a- and P-groups. We tested whether such topology was statistically less likely than the obtained maximum likelihood tree. Constraining the four strains to be grouped in the same clade leads to a statistically less likely tree than the most likely tree (p=0.0372, Kishino-Hasegawa test). Constraint tree in which the three Burkholderia strains are grouped together was also rejected (p =0.0004).

These results favor recurrent transfers between the two subclasses of a- and P- over a single transfer between the two sub-classes.

6. Conclusion

The identification of rhizobia within Proteobacteria from the p-subclass shows that the ability to establish a symbiosis with legumes is more widespread in bacteria than anticipated to date. As a consequence, legumes are able to establish a symbiosis with phylogenetically distant bacteria. Such symbiosis is not a sporadic phenomenon, since Ralstonia appear to be the favorite partners of Mimosa pudica and M. diplotricha in Taiwan. We have identified nodulating Methylobacterium (Sy et al. 2001), Burkholderia and Ralstonia. Consequently the word rhizobium, originally a genus name, is now to be considered as a generic term grouping phylogenetically diverse bacteria sharing the ability to establish a legume symbiosis. Symbionts of less than 10% of the 750 legume genera being fully characterized, it is likely that further exploration of the rhizobial diversity may reveal the rhizobial nature of additional members of the P-Proteobacteria and possibly other taxonomic classes.

Our results show that the a- and p-rhizobia use the same strategy (nod genes and Nod factors) for establishing symbiosis with legumes. The spread of nod genes in a- and P-rhizobia probably originate from lateral gene transfer. This transfer may have occurred after the appearance of legumes on Earth, about 70 million years ago. Lateral transfer of symbiotic genes may occur between bacteria living in the same ecological niche, i.e. the rhizosphere. Although transfer is likely to be frequent within a bacterial population, only transfers occurring in bacteria that exhibit predisposition to the symbiosis (i.e. ability to overcome plant defenses, to infect the plant and to self-maintain in plant tissues) will be effective. The genome sequencing of phylogenetically different rhizobia should allow identifying the genetic background for plant infection and more generally the molecular basis of the preadaptation to legume symbiosis.

7. References

Boone C (1999) Carbohydr. Res. 317, 155-163 Chen et al. (2001) Inter. J. Syst. Evol. Microbiol. Denarie J et al. (1996) Annu. Rev. Biochem. 65, 503-535

Lerouge P et al. (1990) Nature 344, 781-784

Moulin L et al. (2001) Nature 411, 948-950

Perret X et al. (2000) Microbiol. Mol. Biol. Rev. 64, 180-201

Spaink HP et al. (1991) Nature 354, 125-130

Sullivan JT, Ronson CW (1998) Proc. Natl. Acad. Sci. USA 95, 5145-9 Sullivan JT et al. (1995) Proc. Natl. Acad. Sci. USA 92, 8985-9 Suominen L et al. (2001) Mol. Biol. Evol. 6, 907-16 Swofford DL (1998) PAUP Version 4, Sinauer Associates, Sunderland, MA Sy A et al (2001) J. Bacteriol. 183, 214-220

van Berkum P, Eardly BD (1998) In Spaink HP, Kondorosi A, Hooykaas PJJ (eds), The

Rhizobiaceae, pp. 1-24, Kluwer Academic Publishers, Dordrecht Young JPW, Hauk KE (1996) New Phytol. 133, 87-94 Zhang XX et al. (2000) Appl. Environ. Microbiol. 66, 2988-2995


C.W. Ronson1, J.T. Sullivan1, G.S. Wijkstra1, T. Carlton1, K. Muirhead1,

J.R. Trzebiatowski2, J. Gouzy3, F.J. deBruijn2'3

'Dept of Microbiology, University of Otago, Dunedin, New Zealand

2Michigan State University, E. Lansing, MI USA

3LBMRPM, CNRS-INRA, Castanet-Tolosan, France

1. Introduction

In recent years, it has become apparent that many bacterial genomes have obtained a significant proportion of their genetic diversity through horizontal gene transfer, with acquisition presumably being counterbalanced by deletion of DNA in order to prevent excessive genome expansion (Ochman el al. 2000). Nevertheless genome size can vary widely between strains of the same species, as shown by the finding that the genome of Escherichia coli strain 0157:H7 is 810 kb larger than the genome of E. coli strain K12. Comparison of the two strains indicates that their chromosomes comprise a conserved core interrupted by many lineage-specific "islands" (Hayashi et al. 2001; Perna et al. 2001). Many gram-negative bacterial pathogens are differentiated from benign relatives by the presence of horizontally-acquired "pathogenicity islands". These chromosomally encoded regions typically contain large clusters of genes required for virulence traits, such as cell invasion, iron uptake, adhesins and hemolysin. Several encode Type III secretion systems which deliver effector proteins directly into the host cell cytoplasm (Hacker, Kaper 2000). Many pathogenicity islands are situated at tRNA or tRNA-like loci, which are common sites for the integration of foreign sequences including bacteriophage, and some contain a gene encoding a phage-like integrase of the P4 family at one end. Most pathogenicity islands are no longer transferable, indicating that they have been fixed into the genome, probably through mutation of mobility genes and/or attachment site sequences. Thus acquisition of genomic islands has effected a stable change to the ecological and pathogenic character of many bacterial species (Ochman et al. 2000).

We have shown that non-symbiotic strains of mesorhizobia can evolve into Lotus-nodulating symbionts in a single quantum leap through the acquisition of a 500-kb chromosomal element. The element integrates into a phenylalanine tRNA gene, reconstructing the gene at one end (arbitrarily defined as the left end), and producing a 17-bp direct repeat of the 3' end of the tRNA gene at the right end. Within the left end of the element, a gene intS that encodes a product with similarity to members of the phage P4 integrase subfamily is located 198 bp downstream of the tRNA gene (Sullivan, Ronson 1998; Sullivan et al. 1995). This gene is required for excision of the element as a circle as well as its integration (Sullivan et al. 2000). The element was termed a symbiosis island on the basis of its similarities to pathogenicity islands of gram-negative bacteria. Like pathogenicity islands, the symbiosis island converts an environmental strain (a soil saprophyte) into a strain capable of forming a close association with an eukaryotic host. Examples of other genomic islands for which transfer has been demonstrated include the Pseudomonas clc element and the Salmonella conjugative transposon CTnscr94. The clc element is a 105-kb transferable element that contains chlorocatechol-degradative enzymes and integrates into a glycine tRNA gene using a P4-like integrase (Ravatn et al. 1998). CTnscr94 integrates into a phe-tRNA gene and contains genes for sucrose utilization (Hochhut et al. 1997). However, although it is clear that genomic islands play a key role in microbial evolution, the extent to which they contribute to the environmental adaptation of bacteria other than pathogens is unknown.

2. The Mesorhizobial Population Harboring the Symbiosis Island is Diverse

The symbiosis island was discovered during a study undertaken to examine generation of genetic diversity in a rhizobial population which developed under a stand of Lotus corniculatus established with a single inoculant strain in a region where there were no pre-existing rhizobia capable of nodulating the plant. Populations of indigenous rhizobia can rapidly supplant inoculant strains, presumably due to their superior environment-specific adaptive traits, even when the initial population is small or undetectable. Gaining an understanding of the ecology of indigenous rhizobial populations is a crucial step towards developing effective strategies to increase symbiotic nitrogen fixation through the addition of selected inoculant strains. The site examined was established using strain ICMP3153 as inoculant seven years prior to sampling. Differences in growth rate amongst strains isolated from nodules were noted, and RFLP profiling confirmed that considerable genetic diversity existed within the population. Only 20% of the nodule isolates were the same as ICMP3153, including strain R7A which has been used for subsequent studies. Subsequent molecular studies showed that the diverse strains were derivatives of indigenous non-symbiotic mesorhizobia that had acquired the symbiosis island from ICMP3153 (Sullivan et al. 1995).

The diverse symbiotic strains, together with seven non-symbiotic strains (strains CJ1-CJ7) that were isolated from the same site, were characterized by RFLP and multilocus enzyme electrophoresis (MLEE) analysis, full length 16S rRNA gene sequencing and total DNA:DNA hybridization analysis. The results showed that four non-symbiotic strains belonged to the same species as the diverse symbiotic strains, whereas the other three non-symbionts and the original inoculant strain represented further genomic species of mesorhizobia. Even within the same genomic species, the field isolates showed substantial genetic diversity (Sullivan et al. 1996). This diversity was further indicated by comparing the DNA sequence from six strains surrounding the phe-tRNA gene - the sequence immediately upstream from the tRNA gene was highly conserved, whereas the strains fell into three groups on the basis of sequence similarity downstream of the inserted symbiosis island (Sullivan, Ronson 1998). It was proposed that this diversity might represent further genomic islands integrated at the same tRNA locus that may adapt the indigenous strains to the local environment.

To learn more about the contribution of genomic islands to the evolution and niche adaptation of mesorhizobia, we have sequenced the symbiosis island from strain R7A and further characterized the DNA regions downstream of the phe-tRNA locus in several diverse strains. We also compared the sequences to the genome sequence of a Japanese isolate of M. loti, strain MAFF303099. The 7.6 Mb MAFF303099 genome consists of a chromosome and two plasmids, pMla and pMlb (Kaneko et al. 2000). Here we highlight the genetic diversity uncovered by a comparative analysis of the R7A and MAFF303099 symbiosis islands, and report on the identification of further genomic islands that may contribute to the diversity and adaptation of the bacteria.

3. Comparative Analysis the R7A and MAFF303099 Symbiosis Islands

Comparison with R7A indicates that the MAFF303099 chromosome contains a 610,975-bp symbiosis island integrated adjacent to the phe-tRNA gene (Kaneko et al. 2000). The R7A island at 501,801 bp in size is 109 kb smaller than the M. loti MAFF303099 island and encodes 416 potential genes. Comparisons of the two M. loti symbiosis islands indicate that they have similar metabolic and symbiotic potential. The two islands share a conserved backbone sequence of 248 kb with about 98% DNA sequence identity, indicating that the two islands evolved from a common ancestral source. The backbone contains the key symbiotic gene complement including all the genes required for Nod factor synthesis. It is interrupted by a series of strain-specific "islets" that represent DNA either lost or gained by each strain and range in size from a few base pairs up to 168 kb. The few non-syntenous regions that encode similar proteins show less than 90% nucleotide identity, suggesting that most were separately acquired by each island rather than arising through translocation. About 8% of the R7A island consists of insertion sequences (six identifiable intact genes) or fragments thereof, compared to 19% for MAFF303099 (Kaneko et al. 2000), which accounts for a significant portion of the size difference between the two islands. Analysis of the strain-specific segments of both islands reveals that in addition to IS genes, they contain mainly hypothetical genes, metabolic genes and ABC transporters. One significant difference is that the R7A island has a gene cluster with strong similarity to those vir genes from Agrobacterium tumefaciens that encode the Type IV pilus through which T-DNA is transferred to the plant. This cluster is missing from MAFF303099, which in turn has a gene cluster with strong similarity to the cluster encoding a type III secretion system in Rhizobium strain NGR234 (Viprey et al. 1998) that is missing from R7A. Another interesting feature is that of the 114 hypothetical genes detected in R7A that have no database matches in other bacteria, 102 are not present in M. loti MAFF303099 indicating that they are strain- rather than species-specific.

Overall the comparative analysis of the islands emphasizes that they are dynamic mosaics shaped by multiple recombination events and in particular acquisition and deletion of DNA segments. As well as strain-specific regions, variable G+C content and insertion sequences, there are several gene fragments or pseudogenes, some of which are in differing stages of decay in the two islands. Some have intact orthologs on the R7A island whereas others do not. In addition, a number of gene clusters found on the R7A island are also present on pMLa in MAFF303099 and some of these are absent from the MAFF303099 island.

4. Sequence Diversity Downstream of the phe-tRNA Locus or Symbiosis Island

DNA regions of about 1 kb downstream of the symbiosis island were amplified by inverse PGR and sequenced from 35 strains to determine whether the diverse mesorhizobial population contains additional acquired elements inserted adjacent to the phe-tRNA locus. In addition, cosmid clones of the DNA regions were isolated and partially sequenced for strains CJ3, CJ4, R7A and R88b. The sequences aligned into 15 similarity groups, with two groups found multiple times (Table 1).

Group A contains five strains with sequence similarity to a P4 integrase gene directly downstream of the 17-bp repeat of the phe-tRNA gene that demarcates the border of the symbiosis island. Strain MAFF303099 also contains a P4 integrase downstream of its symbiosis island (Kaneko et al. 2000). The gene products of that integrase, intR88B and intCJ4 have very high nucleotide sequence identity (94.5%) with each other. They do not share significant nucleotide sequence identity with the symbiosis island integrase, and show only 50% amino-acid identity with it. However DNA downstream of the integrase genes diverged sharply in each of the three strains, suggesting that the three genes are associated with different acquired DNA regions.

Table 1. Grouping of mesorhizobial strains according to sequence similarity directly downstream of the symbiosis island.


Seq. Similarity

Nt ident. to MAFF303099

No similarities No similarities

Transposase Aldehyde DH Insertion seq

P4 integrase FhuBD

No No No No Yes

500 bp only of 6 kb

The sequence of an 8-kb region for strain CJ3 was completed and revealed genes with strong similarity to fhuDB genes from a number of bacteria that are required for the transport of the ferric hydroxamate siderophore ferrichrome (Guerinot 1994). Hybridization analysis indicated that these genes were present in only a subset of strains, indicating that they are acquired. They are absent from MAFF303099. A gene ggt encoding y-glutamyl transpeptidase was present directly upstream of the fhuBD genes, ggt is involved in glutathione metabolism and is expected to be a core chromosomal gene. Hybridization of the ggt gene to diverse strains showed that it was present in single copy in all strains. It is also present in MAFF303099. Hence the junction between the 6.6 kb acquired element in CJ3 and the core chromosome lies between the fhuD and ggt genes (Figure 1).

Strain R88B contains a P4 integrase and also the fhuBD genes. The fhuDB genes are adjacent to ggt and contiguity with MAFF303099 was found for ggt and several genes upstream of ggt, confirming that the junction between the core chromosome and acquired elements lies between ggt and fhuD. A cosmid containing the intRnB gene was also isolated and partially characterized. The region downstream of the int gene has a complex repeat structure and shows similarity to a gene encoding an outer membrane adhesin from Pseudomonas putida that has recently been identified as essential for seed colonization (Espinosa-Urgel et al. 2000). The region is absent from MAFF303099 and most other strains tested. The fliuDB and mus-20 regions are yet to be linked (Figure 1).



CD 00 00 00 00

Was this article helpful?

0 0

Post a comment