Unconventional Genomic Organization in the Alpha Subgroup of the Proteobacteria

Bacteria belonging to the Rhizobiaceae. In the familyRhizobiaceae are found bacteria which live in association with plant cells. The classification of Agrobacterium andRhizobium species is based on both phenotypic traits and plasmid-encoded characteristics inducing symbiosis or tumorigenesis (44), rather than chromosomal genes. Moreover, in these genera, extrachromosomal elements represent a major part of the genome (26).

(i) The genus Agrobacterium. The genusAgrobacterium consists of several genetically and phenotypically different groups or clusters (21). Differences in 16S rRNA sequences clearly separated strains of biovar 1, biovar 2, biovar 3 (Agrobacterium vitis), andAgrobacterium rubi (37).

We have previously shown that the biovar 1 strain A. tumefaciens C58 contains four replicons: two megabase-sized chromosomes (one of which is linear) and two plasmids (1). Two other strains belonging to this biovar, Agrobacterium radiobacter CFBP 2414^T and A. tumefaciens ATCC 23308^T, were tested. Two megabase-sized replicons, both hybridizing with the rRNA probe, were seen in these strains (Fig. 1 and2 and Table 1). The smaller band was more intense and more diffuse than the larger one, suggesting that, as inA. tumefaciens C58, this replicon is linear. Electrophoresis under different pulsing conditions revealed the presence of small replicons corresponding to plasmids; there were two in A. tumefaciens ATCC 23308^T but only one in A. radiobacter CFBP 2414^T, which is not a pathogen (Fig.3 and Table 1).

• Open in new tab
• Download powerpoint

Fig. 1.

PFGE of intact DNAs of bacterial species belonging to the family Rhizobiaceae: separation of large replicons. Lanes: 1, Saccharomyces pombe; 2, A. vitis CFBP 2721; 3, A. vitis CFBP 2607; 4, A. tumefaciensC58; 5, A. tumefaciens ATCC 23308^T; 6, A. radiobacter CFBP 2414^T; 7, A. rhizogenesATCC 11325^T; 8, A. rhizogenes K84; 9, A. rubi ATCC 13335^T; 10, Rhizobium fredii ATCC 35423^T; 11, Rhizobium leguminosarum bv. phaseoli ATCC 14482^T; 12, Rhizobium leguminosarum bv. trifolii ATCC 14480^T; far right, H. wingei. The positions of molecular size markers are indicated on both sides of the gel.

• Open in new tab
• Download powerpoint

Fig. 2.

Hybridization of large replicons with the 16S rRNA probe. (Upper panel) 1, A. radiobacter CFBP 2414^T; 2, A. rhizogenes K84; 3, A. tumefaciens C58; 4, A. rhizogenes ATCC 11325^T; 5, A. rubi ATCC 13335^T; 6,Rhizobium fredii ATCC 35423^T; 7, Rhizobium leguminosarum bv. trifolii ATCC 14480^T; 8,Rhizobium meliloti 2011. (Lower panel) Lanes: 1, O. anthropi ATCC 49188^T; 2, P. myrsinacearumATCC 43590^T; 3, Rhodobacter capsulatus ATCC 11166; 4, Rhizobium meliloti 2011; 5, A. tumefaciens C58. The positions of molecular size markers are shown on both sides of the two panels.

• Open in new tab
• Download powerpoint

Fig. 3.

PFGE of intact DNAs of bacterial species belonging to the family Rhizobiaceae: separation of small replicons. Lanes: 1, Saccharomyces cerevisiae; 2, A. rhizogenes K84; 3, A. vitis CFBP 2721; 4, A. vitis CFBP 2607; 5, A. tumefaciens C58; 6, H. wingei; 7, A. tumefaciens ATCC 23308^T; 8,A. radiobacter CFBP 2414^T; 9, A. rubiATCC 13335^T; 10, Rhizobium meliloti 2011; 11,Rhizobium fredii ATCC 35423^T; 12,Rhizobium leguminosarum bv. phaseoli ATCC 14482^T; 13, R. leguminosarum bv. trifolii ATCC 14480^T; 14, Saccharomyces cerevisiae. The positions of molecular size markers are shown on both sides of the gel.

The results for A. rubi ATCC 13335^T, which is one of the three strains forming this cluster (21), were similar to those for biovar 1 strains: it contained one circular and one apparently linear chromosome plus two plasmids (Fig. 1 to 3 and Table 1).

Two strains belonging to biovar 2, A. rhizogenes ATCC 11325^T and A. rhizogenes K84, were studied. In both strains, two megareplicons, both apparently circular, were separated, but only the larger one hybridized with the rRNA probe (Fig.1 and 2 and Table 1). Also present in the ATCC 11325^T and the K84 strains were two smaller replicons which correspond to the previously reported agrocine and tumor-inducing plasmids of the latter strain (21) (Fig. 3 and Table 1).

Strains of biovar 3 are tumorigenic for grape vines (21). Two of them, A. tumefaciens CFBP 2721 and CFBP 2607, were analyzed. Two megareplicons, both hybridizing with the rRNA probe and both apparently circular, could be seen, together with three plasmids for the former strain and two plasmids for the latter (Fig. 1 and 3 and Table 1).

The unusual genomic organization of A. tumefaciens C58 was first suspected because of the greater intensity and the diffuse aspect of one of the two megabase-sized bands, the one corresponding to all of the linear molecules when, under the same conditions, only some randomly linearized forms of the circular replicons enter the gel (38). Further evidence was established by insertion of a unique restriction site into each of the chromosomes that led, after enzymatic digestion, to the linearization of the circular molecules and the generation of two fragments from the linear molecules (18). In the other biovar 1 strains and A. rubi, two megabase-sized replicons were present, with the smaller molecule probably being linear. Thus, this multipartite genome with different topologies appears to be a common feature of strains of biovar 1 andA. rubi.

Biovar 2 and 3 strains also possess two megareplicons; however, their intensity in the PFGE gel suggests that they are both circular. Interestingly, while both megareplicons of biovar 3 hybridized with the rRNA probe, only the larger one of biovar 2 hybridized with it. Nevertheless, the presence of other essential housekeeping genes on the second molecule is possible, as was shown for Rhizobium meliloti (35). The sizes of the biovar 2 strain genomes (>7,200 kb) are larger than those of the biovar 1 and A. rubi strains (5,900 to 5,735 kb) as well as those of biovar 3 strains (5,500 or 4,980 kb).

(ii) The genus Rhizobium. Sobral et al. have described the presence of three megabase-sized replicons in Rhizobium meliloti 1021, one chromosome of 3.5 Mb and two megaplasmids of 1.7 and 1.3 Mb, so called because they did not hybridized with an rRNA probe (38). Nevertheless, essential housekeeping genes such as the GroEL chaperonin-encoding genes are present on these molecules, thus raising questions about the chromosomal status of these replicons (35). We tested a second strain, R. meliloti2011, and found three replicons of sizes similar to those of strain 1021, again with only the larger one hybridizing with the rRNA probe (Fig. 2 and 3, and Table 1).

For Rhizobium fredii ATCC 35423^T, two megabase-sized replicons and a small replicon were separated. Only the largest one was shown to hybridize with the rRNA probe (Fig. 1 to 3 and Table 1).

In Rhizobium leguminosarum ATCC 14480^T and ATCC 14482^T (corresponding to the biovars trifolii and phaseoli), we found two circular megareplicons plus the two large plasmids previously described for this species (29). We hybridized an rRNA probe with separated replicons of R. leguminosarum bv. trifolii. Again, only one (the largest) contains rRNA genes (Fig. 1 to 3 and Table 1).

The organization of the genome of Bradyrhizobium japonicumis different; this genome has a single chromosome that is larger (8,700 kb) than that of the other Rhizobium species (6,400 to 6,700 kb) (6, 22).

(iii)The genus Phyllobacterium. ForPhyllobacterium myrsinacearum ATCC 43590^T, five replicons were separated, with the larger (megabase-sized) one hybridizing with the rRNA probe (Fig. 2,4, and 5and Table 1). The genomic organization for the genusPhyllobacterium also seems different from that of the other genera of the Rhizobiaceae, with there being only one megareplicon (but several large plasmids) and a smaller total genome size (5,330 kb).

• Open in new tab
• Download powerpoint

Fig. 4.

PFGE of intact DNAs of bacterial species related to theRhizobiaceae: separation of large replicons. Lanes: 1,Schizosaccharomyces pombe; 2, Rhodobacter capsulatus ATCC 11166; 3, P. myrsinacearum ATCC 43590^T; 4, O. anthropi ATCC 49188^T; 5, O. anthropi LMG 3301; 6, M. dimorpha ATCC 4279^T; 7, Rhizobium meliloti2011; 8, A. tumefaciens C58; 9, H. wingei. The positions of molecular size markers are shown on both sides of the gel.

• Open in new tab
• Download powerpoint

Fig. 5.

PFGE of intact DNAs of bacterial species related to theRhizobiaceae: separation of small replicons. Lanes: 1, lambda DNA ladder; 2, H. wingei; 3, P. myrsinacearum ATCC 43590^T; 4, M. dimorphaATCC 4279^T; 5, Rhodobacter capsulatus ATCC 11166; 6, H. wingei; 7, O. anthropi ATCC 49188^T; 8, O. anthropi LMG 3301; 9,Saccharomyces cerevisiae; 10, lambda DNA ladder. The positions of molecular size markers are indicated to the left and right.

Related bacteria belonging to others genera and families.Members of the Rhizobiaceae are also related to taxonomically different organisms (9, 43). Rhodobacter sphaeroides is found on a distant branch of rRNA superfamily IV, which comprises Agrobacterium species, Rhizobiumspecies, and Brucella abortus (9). Other bacteria that are closely related to rRNA superfamily IV are Mycoplana dimorpha, Ochrobactrum anthropi, and Bartonella quintana (37, 40, 43). These related bacteria represent a heterogeneous group whose members have few common features;Rhodobacter sphaeroides is a facultative photosynthetic bacterium (39), M. dimorpha is a soil-living organism (43), Brucella and Bartonellaspecies are animal pathogens (42, 43), and O. anthropi is a rare opportunistic pathogen of immunocompromised patients (2).

(i)The genus Rhodobacter. Suwanto and Kaplan have shown that Rhodobacter spheroides 2.4.1 possesses two chromosomes, one of 3,000 kb and the other of 900 kb (39). However, the chromosomal structure of Rhodobacter capsulatus SB1003 is quite different, consisting of a unique 3,800-kb chromosome and a 134-kb plasmid (12). We investigated another strain ofRhodobacter capsulatus (ATCC 11166) and found only one megabase-sized replicon, which hybridized to the rRNA probe, and a small replicon (Fig. 2, 4, and 5 and Table 2).

(ii) The genera Ochrobactrum and Brucella. Two strains of O. anthropi were studied. Two megareplicons of similar sizes were found in both strains ATCC 49188^T and LMB 3301, and two small replicons of different sizes were found in each of these strains (Fig. 2, 4, and 5 and Table 2). Only the two larger bands hybridized with the rRNA probe. The physical map ofBrucella melitensis has been constructed, and it demonstrated the presence of two circular chromosomes (27). These two chromosomes are also present in the other species of this genus (28), with the exception of one biovar (see below).

(iii) The genera Mycoplana and Bartonella. The type strain M. dimorpha ATCC 4279^T had only one (megabase-sized) replicon and two plasmids (Fig. 4 and 5 and Table2). Bartonella quintana was shown to possess only one chromosome (reference 32 and unpublished data).

Does genomic organization have phylogenetic significance?Using highly conserved sequences such as the rRNA genes or housekeeping proteins such as the GroEL chaperonin and RecA, phylogenetic trees have been constructed which have allowed the definition of the alpha subgroup of the Proteobacteria (9, 10, 40-42). The genomic organization of bacteria belonging to this group has been studied to see if a correlation with the phylogeny could be demonstrated.

Genome size differences, increasing with the evolutionary genetic distance between lineages, were shown to exist for the major subgroups of Escherichia coli, which suggests that there is a phylogenic component to this variation (4). The genome ofA. rhizogenes K84 (7,265 kb) is 1.45 times larger than that of A. vitis CFBP 2607 (4,980 kb). This degree of variation is comparable to that seen for different strains of Burkholderia cepacia (13). Strains of Agrobacteriumbiovars 1 and 2 exhibit only 15% DNA homology (21). Our results show that their genome sizes and organizations are also very different, thus providing further evidence that they are genetically distinct. Sawada et al. place Agrobacterium biovar 2 closer to Rhizobium fredii, and this is again supported by the genomic organization (37).

Most of the organisms possessing several megabase-sized replicons belong to the alpha subgroup of the Proteobacteria (Fig.6); the exceptions are Burkholderia (Pseudomonas) cepacia, which is classified in the β2 subgroup (42), and L. interrogans, which is a spirochete (45). The members of subgroup α2 form a tight cluster, while the β subgroup constitutes a quite phylogenetically diverse class (20, 42). The existence of a more complex genomic architecture (with two or three chromosomes) may have phylogenetic significance if this trait is also found in other organisms of the same lineage. Among the alpha-subgroup genera that we have investigated, this feature is present in all of the species ofAgrobacterium, Rhizobium, Brucella(except one [see below]), and Ochrobactrum. On the contrary, Bradyrhizobium, Phyllobacterium,Mycoplana, and Bartonella species have only one chromosome. Finally, in the genus Rhodobacter, R. sphaeroides has two chromosomes while R. capsulatus has only one.

• Open in new tab
• Download powerpoint

Fig. 6.

Phylogenetic tree showing the genomic organization of organisms belonging to the α2 subgroup of theProteobacteria (plus Rhodobacter species, which belong to the α3 subgroup) (redrawn from reference 43). Organisms with complex genomes are indicated in boldface and underlined.

The Rhizobiaceae can be divided into two groups. The fast-growing strains (Rhizobium meliloti, Rhizobium fredii, and Rhizobium leguminosarum) all have complex genomes, while the slow-growing species Bradyrhizobium japonicum has a single, very large chromosome (22). The genus Bradyrhizobium, however, is only remotely related to the other genera of the Rhizobiaceae (41). The deeper branching found for Bradyrhizobium japonicum with both the 16S rRNA and the GroEL sequences (9, 40, 43) could mean that the origin of this lineage is close to the single-chromosome ancestor. This taxonomically different genus (17) represents a separate line of descent in the α2 subgroup of theProteobacteria (41), one which is also remote from the Agrobacterium rRNA branch in rRNA superfamily IV (16). In contrast, Rhodobacter capsulatus andRhodobacter sphaeroides, with one and two chromosomes, respectively, branch together on the phylogenetic tree (10). In this case of two species belonging to the same lineage, it is difficult to explain how two organisms with such different genomic organizations could have a common ancestor unless this feature is not linked with the phylogeny. Moreover, within the same species—Brucella suis—the genome of the biovar 3 reference strain is composed of a single chromosome of 3.2 Mb while biovar 1 members each possess two chromosomes, of 2.1 and 1.15 Mb, and biovar 2 and 4 members each have two chromosomes, of 1.85 and 1.35 Mb. The four biovars are phenotypically very similar, and the restriction maps of their genomes are also very similar except for the distribution of the same sequences on different linkage groups (19). Other evidence is from outside of the α-proteobacteria, for the gram-positive bacterium Bacillus cereus, whose different strains vary with respect to their chromosome sizes and genome organizations. Within this species, the genome may exist either as one large chromosome with small plasmids or as a small chromosome with stably maintained large extrachromosomal elements which may be considered as fragments of a secondary chromosome (5).

Thus, the presence of multiple chromosomes in α-proteobacterial genomes is not related to a common phylogeny, since it is not always shared either by all of the members of a same clade (e.g.,Rhodobacter genus) or even by all of the strains of the same species (e.g., Brucella suis). This trait, found inconstantly among different bacterial lineages, rather seems to have been acquired independently. Where does this complex organization originate?

The classical model of genome evolution involves gene duplication followed by divergence. This can occur via a tandem duplication in the genome, achieved by recombination between repeated sequences (34). Such repeats could be rRNA operons. Following this, a second intrachromosomal recombination event, occurring anywhere within the duplicated region, will result in the formation of two stable replicons if both molecules have an origin of replication; alternatively, the second origin of replication could be acquired by lateral transfer from a different organism. A comparison of the sequences of these molecules will distinguish between these two possibilities. Nevertheless, there is no known environment shared by these different organisms which could explain their “infection” by a new origin. In the case of the genus Brucella, we have shown that the different species exhibit differences in genomic organization. The differences in chromosome size and number can be explained by the occurrence of rearrangements at chromosomal regions containing the three rrn genes. The location and orientation of these genes confirmed that these rearrangements are due to homologous recombination at the rrn loci (19). This phenomenon occurred naturally in the genus Brucella; however, recently the 4,188-kb circular genome of Bacillus subtilis was artificially dissected into two stable circular chromosomes in vivo by such a mechanism (15).

The coexistence of linear and circular chromosomes in the same bacterial cell raises another question. The chromosome ofStreptomyces lividans probably oscillates between linear and circular forms, and this may also occur in other bacteria (31). It has been suggested by Hinnebusch and Tilly (14) that one of the origins of linear DNA in bacteria could be genetic exchange between procaryotes and eucaryotes. These authors also add that the most evident example of this gene exchange is the transfer of DNA from the phytopathogen A. tumefaciens into a plant cell to induce the formation of a crown gall tumor. It is perhaps not a coincidence that in this species the chromosomes exhibit both types of structures.

The reason for the presence of a complex genomic organization in many members of the α-proteobacteria remains to be determined. While we have shown that the possession of a complex genome does not have a clear phylogenetic significance, we can speculate that there are structures in or functions of the genome of the alpha subgroup which favor their appearance.