Atypical 16S rRNA Gene Copies in Ochrobactrum intermedium Strains Reveal a Large Genomic Rearrangement by Recombination between rrn Copies (original) (raw)

Abstract

Ochrobactrum intermedium is an opportunistic human pathogen belonging to the alpha 2 subgroup of proteobacteria. The 16S rDNA sequences of nine O. intermedium isolates from a collection of clinical and environmental isolates exhibited a 46-bp insertion at position 187, which was present in only one sequence among the 82 complete or partial 16S rDNA sequences of Ochrobactrum spp. available in data banks. Reverse transcription-PCR experiments showed that the 46-bp insertion remained in the 16S rRNA. The inserted sequence folded into a stem-loop structure, which took place in and prolonged helix H184 of the 16S rRNA molecule. Helix H184 has been described as conserved in length among eubacteria, suggesting the idiosyncratic character of the 46-bp insertion. Pulsed-field gel electrophoresis experiments showed that seven of the clinical isolates carrying the 46-bp insertion belonged to the same clone. Insertion and rrn copy numbers were determined by hybridization and I-_Ceu_I digestion. In the set of clonal isolates, the loss of two insertion copies revealed the deletion of a large genomic fragment of 150 kb, which included one rrn copy; deletion occurred during the in vivo evolution of the clone. Determination of the rrn skeleton suggested that the large genomic rearrangement occurred during events involving homologous recombination between rrn copies. The loss of insertion copies suggested a phenomenon of concerted evolution among heterogeneous rrn copies.

The genus Ochrobactrum belongs to the alpha 2 subgroup of Proteobacteria and is the closest relative to the genus Brucella. Ochrobactrum anthropi was considered the sole and type species of the genus Ochrobactrum (15) until 1998, when Ochrobactrum intermedium was proposed as a new species of this genus (50). Factors discriminating between O. anthropi and O. intermedium were their low DNA-DNA hybridization, their different 16S ribosomal DNA (rDNA) sequences, the different Western blot profiles of sodium dodecyl sulfate-polyacrylamide gel electrophoresis-separated whole-cell proteins, and the resistance of O. intermedium to colistin and polymyxin B (50). In a recent study, Lebuhn et al. (21) performed a polyphasic analysis of a large collection of Ochrobactrum sp. strains isolated from the wheat rhizoplane. They described two new species, Ochrobactrum grignonense and Ochrobactrum tritici and proposed the separation of the Brucella-Ochrobactrum group into five clades with O. intermedium belonging to clade 1 (21).

Ochrobactrum spp. are members of the soil microbiota (21), and an increasing number of studies have reported the isolation of O. anthropi and O. intermedium from clinical specimens, especially from immunocompromised patients or from patients with nosocomial infections (28, 32, 47). To date, 16S rDNA sequencing is the most reliable procedure used for identification of Ochrobactrum strains at the species level, especially in clinical specimens (32, 47).

The genomes of O. anthropi and O. intermedium have been previously described as complex with two independent circular chromosomes. Each chromosome hybridized with a 16S rDNA probe, but the rrn copy number has not been determined (17). The organization of rRNA genes as a multigene family is widespread in eubacteria. The members of an rRNA multigene family are subject to a homogenization process, allowing several gene copies to evolve in concert. In a concerted-evolution mode, a mutation occurring in one copy will be either fixed for all of them or lost for all. The recombination events involved in concerted evolution probably occur by gene conversion, a nonreciprocal recombination event in which the sequence of one copy of a gene is converted to the sequence present in another one (13, 22). Thus, rRNA sequences show low variability within species, subspecies, or genomes. However, the general extent of intraspecific variation in rDNA sequences has been observed among sequences deposited in the GenBank database (8). Furthermore, intragenomic heterogeneity in the form of nucleotide differences between 16S rDNA copies, so-called microheterogeneity, has been described in few cases. For example, microheterogeneity has been identified in Escherichia coli (7, 31), Mycobacterium terrae (34), Paenibacillus polymyxa (36), members of the class Mollicutes (14, 23, 39), and the Actinomycetales Thermomonospora chromogena (53), Thermobispora bispora (38), and Streptomyces spp. (49). Expression of two rrn operons differing in 5% of the nucleotide positions has also been described for the archeon Haloarcula marismortui (10). Microheterogeneity appeared to be more common than macroheterogeneity, involving large insertions ranging from 50 to several hundred nucleotides. Macroheterogeneity of the 16S rDNA, involving an intervening sequence (IVS) present in the gene but absent in the 16S rRNA molecule, has been observed in the archaeons Pyrobaculum aerophilum (5), Aeropyron pernix (35), and Thermoproteus spp. (16). IVSs have also been found in the 16S rDNA of some eubacteria, for instance, in the genera Campylobacter (11, 12, 25), Helicobacter (24, 45), and Clostridium (41), and in two endosymbiontic proteobacteria (42, 44). In the two eubacterial species Desulfotomaculum australicum (37) and Bacillus sporothermodurans (40), the insertion in the 16S rDNA persisted in the 16S rRNA.

Direct sequencing after PCR amplification of 16S rDNA of O. intermedium isolates gave ambiguous results with double sequencing signals; this is due to an atypical insertion of 46 bp in certain 16S rDNA copies. We studied the persistence and location of the insertion in the 16S rRNA molecule. In a set of clonal isolates, the 46-bp insertion was unstable and variation in the insertion copy number revealed a large genomic deletion including one rrn operon.

MATERIALS AND METHODS

Bacterial strains, cultivation, identification, and growth curves.

The reference strain of O. intermedium, LMG 3301T (deposited as O. anthropi and now transferred to O. intermedium as the type strain [50]), was obtained from the Collection Française des Bactéries Phytopathogènes. Thirteen O. intermedium isolates were obtained from patients hospitalized in the Academic Hospital of Montpellier (Montpellier, France). Among these clinical isolates, strains ADV1 to -7 were isolated from respiratory samples of the same patient with chronic carriage over a 1-year period. O. intermedium strain PR17/sat had already been isolated from the nematode Heterorhabditis indica in Puerto Rico (3) (Personal gift from Noel Boemare, Laboratoire de Pathologie Comparée, Université de Montpellier, Montpellier, France). O. intermedium strains used in this study are listed in Table 1. Bacteria were grown on tryptic soy (TS) agar for 24 h at 37°C. The identification of the isolates as members of the species O. intermedium was based on phenotypic criteria such as Gram stain, presence of oxidase, analytical profile index (API 20E and API 20NE systems; bioMérieux, Marcy l'Etoile, France), and antibiotic susceptibility profiles (47). 16S rDNA sequencing was used to confirm the species affiliation.

TABLE 1.

Data for O. intermedium strains used in this work

Strain(s) Origin Copy no. of: 16S rDNA sequence accession no. Source and reference(s):
rrn Insertion
ADV1 Human (clinical) 4 3 AF526509 This study
ADV2 Human (clinical) 4 3 This study
ADV3 Human (clinical) 3 1 AF526516 This study
ADV4-ADV7 Human (clinical) 3 1 This study
ADV9 Human (clinical) 4 2 AF526510 and AF526517 This study
ADV10 Human (clinical) 4 0 AF526511 This study
ADV11 Human (clinical) 4 0 AF526512 This study
ADV14 Human (clinical) 4 0 AF526513 This study
ADV21 Human (clinical) 4 0 AF526514 This study
ADV24 Human (clinical) 4 0 AF526515 This study
PR17/sat Nematode 4 4 AJ245941 3 and this study
LMG 3301T Human (clinical) 4 0 U70978 15, 49, and this study

Growth curves of O. intermedium strains were performed as follows. Bacterial strains grown overnight in TS medium were diluted 50-fold in fresh TS medium and incubated at 37°C with gentle shaking. A bacterial count was performed every 15 min for 8 h by measuring the turbidity at 600 nm. Each strain was tested for growth curves in five independent experiments. For statistical analysis, the turbidity versus time was plotted. The plotted points were subjected to a nonlinear regression analysis. The comparison of the curves among strains was performed by using the one-sided Wilcoxon rank sum test. The level of the statistical significance was set at 5%.

rDNA amplification, sequencing, and analysis.

One isolated colony was suspended in 50 μl of sterile distilled water, and the DNA was rapidly extracted by a method involving boiling and freezing (43). 16S rDNAs were selectively amplified by PCR using 27f (5′-GTGCTGCAGAGAGTTTGATCCTGGCTCAG-3′; positions 8 to 36 [E. coli numbering]) as the forward primer and 1492r (5′-CACGGATCCTACGGGTACCTTGTTACGACTT-3′; positions 1478 to 1508 [E. coli numbering]) as the reverse primer. An internal primer in the 46-bp insertion, ins1 (5′-GCCCCCCTTTAAAATTTCAG-3′), was used in association with the primer 1492r for specific amplification of 16S rDNA copies carrying the insertion. Primers for 23S rDNA amplification were LS24f (5′-ATTTGGTGGATGCCTTGG-3′; positions 24 to 41 [E. coli numbering]) and LS2744r (5′-CCCGCTTAGATGCCTTCAGC-3′; positions 2744 to 2763 [E. coli numbering]). The PCRs were carried out in 50 μl of reaction mixture containing 200 nM (each) primer, 200 μM (each) deoxynucleoside triphosphate (dNTP), 1 U of Taq polymerase (Roche, Meylan, France) in the appropriate reaction buffer, and 2 μl of crude DNA extract as the template. PCR conditions were 30 cycles of 1 min at 94°C, 1 min at 65°C, and 2 min at 72°C. Amplification of a single 16S rDNA copy was performed by cutting a restriction fragment from agarose gel and using 2 μl of the melted agarose as the PCR template. Determination of fragments harboring a 16S rDNA copy was done by Southern blotting (see below). Restriction analysis of PCR products was done with 10 U of _Dra_I (New England Biolabs, Hertfordshire, United Kingdom) by following the supplier's recommendations.

PCR products were directly sequenced on an Applied Biosystems automatic sequencer (Genome Express, Meylan, France) in both directions by using forward and reverse primers. The 16S rDNA sequences were compared with sequences deposited in the GenBank, EMBL, and sequencing-genome databases by using the BLAST program (http://www.ncbi.nlm.nih.gov/blast) and with sequences deposited in the Ribosomal Database Project, version 7.0, by using SIMILARITY RANK and SUGGEST TREE (29). Prediction of RNA secondary structure by energy minimization was performed online by the MFOLD program (51).

PFGE and DNA electrophoresis.

Genomic DNAs were prepared in agarose plugs for enzymatic digestion and pulsed-field gel electrophoresis (PFGE) of intact chromosomes as previously described (17, 18). DNAs were digested with 40 U of _Spe_I or _Hin_dIII (New England Biolabs) or with 1 U of the intronic endonuclease I-_Ceu_I (New England Biolabs) (18). _Spe_I and small I-_Ceu_I fragments were separated by PFGE using a contour-clamped homogeneous electric field apparatus (CHEF-DRII; Bio-Rad, Hercules, Calif.) in a 1% agarose gel in Tris-borate-EDTA buffer (TBE; 0.5×). Pulse ramps were 5 to 35 s for 40 h at 150 V for _Spe_I fragments and 90 to 150 s for 24 h at 170 V for small I-_Ceu_I fragments. Separation of undigested DNA and large I-_Ceu_I fragments was obtained in the same PFGE apparatus in a 0.8% agarose gel in 0.5× TBE by using a pulse ramp of 90 to 300 s for 45 h at 150 V. DNAs digested with _Hin_dIII were subjected to electrophoresis for 3 h at 80 V in a 0.8% agarose gel in 0.5× TBE by using a SubCell apparatus (Bio-Rad).

Southern blotting, probes, and hybridization.

Electrophoresis gels were transferred onto a Nytran N (Schleicher & Schuell, Dassel, Germany) nylon membrane by vacuum blotting (vacuum blotter; Bio-Rad) in 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). 16S rDNA and 23S rDNA digoxigenin-labeled probes were obtained by PCR as described before using primer pairs 27f and 1492r and LS24f and LS2744r, respectively, with a dNTP mixture containing 0.1 mM digoxigenin-dUTP (Roche). The digoxigenin-labeled oligonucleotide corresponding to the 46-bp insertion was purchased from Sigma-Genosys. The hybridization of the probes was detected by a CSPD chemiluminescence system (Roche).

RNA preparation and RT-PCR experiment.

O. intermedium total RNA was isolated and purified by using the SV total-RNA isolation system (Promega, Madison, Wis.) in accordance with the supplier's protocol. Reverse transcription-PCR (RT-PCR) was performed by using the Access RT-PCR kit (Promega). The reaction volume of 50 μl contained 50 ng of the DNase-treated RNA extract, 200 nM concentrations of primers 27f or ins1 and 590r (5′-TGACTTATTGCCCGCCTACG-3′; positions 580 to 599 by E. coli numbering), 200 μM (each) dNTP, 5 U of Tfl DNA polymerase, and 5 U of avian myeloblastosis virus reverse transcriptase in the buffer recommended by the supplier. The incubation conditions were 48°C for 45 min for reverse transcription followed by 94°C for 5 min and 30 cycles of 95°C for 1 min, 60°C for 1 min, and 68°C for 2 min for PCR. Non-reverse-transcribed RNA was used as the PCR template for the negative control. _Dra_I digestion of the RT-PCR products was performed as previously described for the digestion of the PCR products.

Nucleotide sequence accession numbers.

The GenBank accession numbers for the 16S rDNA sequences of O. intermedium strains are listed in Table 1. The accession number of the 23S rDNA sequence of O. intermedium strain ADV1 was AY223505.

RESULTS

Evidence for different copies of 16S rDNA O. intermedium strains.

The 16S rDNA sequence of O. intermedium PR17/sat isolated from a nematode included a sequence of 46 bp inserted at position 187 (E. coli numbering). This inserted sequence was absent from the 82 complete or partial 16S rDNA sequences of Ochrobactrum spp. available in data banks in February 2003, except for one sequence obtained from an Ochrobactrum sp. isolated from the rhizosphere of rice (48). Moreover, BLAST analysis of the inserted sequence did not show significant similarity with sequences of genomes currently being sequenced. Ambiguous sequencing results consisting of double signals were obtained for strains ADV1 to -7 and strain ADV9. After several attempts an exploitable sequence for the strain ADV1 was obtained. The sequence showed the same 46-bp insertion at position 187 as that observed in the isolate PR17/sat. When the 46-bp insertion was excluded from analysis, sequences of 16S rRNA genes of strains ADV1 and PR17/sat were 99.9% identical to those of O. intermedium soil isolates OiC8a and OiC8-6 and clinical isolates 609360I and Relman 99, previously deposited in the databases. To rapidly screen Ochrobactrum isolates that carried the 46-bp insertion, a PCR anchored in the insertion was performed. Amplification products were obtained for O. intermedium strains ADV1 to -7, ADV9, and PR17/sat (data not shown). Since direct sequencing of 16S rDNA amplification products remained unsuccessful for strain ADV9, we sequenced the product obtained by a PCR anchored in the insertion. The partial sequence obtained showed 100% identity to that of strain ADV1. The 46-bp insertion was present in 9 of the 15 O. intermedium strains tested. Therefore, the insertion appeared to be widely represented though it was not a characteristic of the species. In particular, LMG 3301T did not harbor the insertion. The 46-bp insertion was absent from a collection of 10 O. anthropi isolates tested by a PCR anchored in the insertion (data not shown).

16S rDNA copy number and 46-bp insertion copy number.

The copy number of 16S rRNA gene was determined by a Southern blotting experiment, using a 16S rDNA probe, on genomic DNAs digested by _Hin_dIII (restriction site absent from the 16S rDNA in the genus Ochrobactrum). Four 16S rDNA hybridizing fragments were found in 10 (including LMG 3301T) of the 15 strains of O. intermedium (Fig. 1A and Table 1). The four fragments, named A, B, C, and D, comprised 20, 8.1, 5.9, and 4.1 kb, respectively. Surprisingly, the five remaining clinical isolates (strains ADV3 to -7) showed only three restriction fragments hybridizing with the 16S rDNA probe (Fig. 1A and Table 1). The restriction fragment C was absent from the hybridization profile. Genomic digestion with I-_Ceu_I, an intronic endonuclease that cleaved specifically a 26-bp site in eubacterial 23S rDNA (position 1909 in E. coli numbering), gave four restriction fragments (2,130, 1,850, 250, and 150 kb) for 10 (including the strain LMG 3301T) of the 15 strains but only three fragments (2,130, 1,850, and 250 kb) for O. intermedium strains ADV3 to -7 (Fig. 2; see below). These results suggested the presence of four rrn operons in a majority of O. intermedium strains including the reference strain. Operons were named rrnA, rrnB, rrnC, and rrnD, corresponding to the nomenclature for the _Hin_dIII fragments. However, O. intermedium strains ADV3 to -7 carried only three copies of the rrn operon. Operon rrnC was absent from these strains.

FIG. 1.

FIG. 1.

Copy numbers of 16S rDNA (A) and 46-bp insertion (B) in the genome of O. intermedium. Shown are Southern blots of _Hin_dIII-digested genomic DNA from strains PR17/sat (lane 1), ADV1 (lane 2), ADV3 (lane 3), and ADV9 (lane 4) and reference strain LMG 3301T (lane 5) hybridized with the 16S rDNA probe (A) and the 46-bp insertion probe (B). Strain ADV2 showed a hybridization profile identical to that of strain ADV1; strains ADV4 to -7 showed hybridization profiles identical to that of strain ADV3; strains ADV10, ADV11, ADV14, and ADV24 showed hybridization profiles identical to the strain LMG 3301T profile (data not shown). Sizes of hybridizing fragments were calculated by using λ digested by _Hin_dIII as a molecular marker.

FIG. 2.

FIG. 2.

PFGE migration of undigested and I-_Ceu_I-digested genomic DNAs of O. intermedium strains ADV1 and ADV3. (A) Migration of high-molecular-weight fragments. Lanes 1 and 2, undigested DNA from strain ADV1 (lane 1) and strain ADV3 (lane 2); lanes 3 and 4, I-_Ceu_I-digested DNA from strain ADV1 (lane 3) and strain ADV3 (lane 4); Lanes Sc and Sp, Saccharomyces cerevisiae (Sc) and Schizosaccharomyces pombe (Sp) chromosomes (Bio-Rad) as molecular size markers. (B) Migration of low-molecular-weight I-_Ceu_I fragments. Lane 1, strain ADV1; lane 2, strain ADV3. Chromosomes and I-_Ceu_I digestion patterns of O. intermedium strains ADV2 and ADV4 to -7 are identical to patterns for strains ADV1 and ADV3, respectively (data not shown). A mixture of λ digested by _Hin_dIII, the λ concatemer, and Saccharomyces cerevisiae chromosomes was used as the molecular size marker (lane λ/Sc); the bands useful for the measure were, from the bottom, 27, 50, 100, 150, 200, 225, 250, and 285 kb.

The 46-bp insertion introduced a unique _Dra_I restriction site in 16S rDNA PCR products. _Dra_I digestion did not cut amplicons obtained for O. intermedium strains LMG 3301T, ADV10, ADV11, ADV14, ADV21, and ADV24 and totally digested amplification products from O. intermedium strain PR17/sat. In contrast, _Dra_I partially digested the 16S rDNA PCR products obtained from strains ADV1 to -7 and ADV9, suggesting that not all their 16S rDNA copies harbored the 46-bp insertion (data not shown). The copy number of the insertion was determined by a Southern blotting experiment using the 46-bp probe on the products of _Hin_dIII genomic digestion (Fig. 1B and Table 1). Among the fragments not carrying 16S rDNA, no hybridizing fragments were observed. The four 16S rDNAs of strain PR17/sat possessed the insertion. There were three copies of the insertion for strains ADV1 and -2 in rrnA, rrnC, and rrnD. Operons rrnA and rrnB of strain ADV9 harbored the 46-bp insertion, whereas it was observed only in rrnD for strains ADV3 to -7.

Instability of the 46-bp insertion and of a large genomic deletion in clonal isolates.

O. intermedium strains ADV1 to -7 were chronologically isolated from the same patient over a 1-year period of chronic carriage. The clonality of the seven strains was studied by PFGE after _Spe_I macrorestriction. Identical PFGE patterns were obtained on one hand for strains ADV1 and -2 and on the other hand for strains ADV3 to -7 (Fig. 3). These two patterns differed slightly, with one additional band of about 150 kb present only in the former pattern. On the basis of the criteria of Tenover et al. (46) and the high degree of PFGE pattern polymorphism among unrelated O. intermedium strains (47), the isolates ADV1 to -7 should be considered a clone. Seven isolates of a single clone provided the opportunity to study the stability of the 46-bp insertion in “field” conditions. In these clonal isolates, the copy number of the insertion decreased from three (strains ADV1 and -2) to one (strains ADV3 to -7) whereas the rrn copy number decreased from four to three for the same strains (Fig. 1). Hybridization of the 16S rDNA probe on _Spe_I patterns confirmed the loss of one rrn copy in strains ADV3 to -7. This copy was carried on the 150-kb fragment present in the genomes of strains ADV1 and -2 but was absent from strains ADV3 to -7 (data not shown).

FIG. 3.

FIG. 3.

PFGE of _Spe_I-digested genomic DNA from O. intermedium strains. Lanes 1 to 7, strains ADV1 to -7, respectively; lane λL, λ ladder, PFGE Marker I (Roche), as a molecular size marker. Arrowhead, 150-kb band present in lanes 1 and 2.

PFGE migration of undigested DNAs of strains ADV1 to -7 allowed the visualization of two chromosomes (Fig. 2A, lanes 1 and 2) as previously observed for the type strain (17) and for all the isolates belonging to the species O. intermedium tested in this work (data not shown). The two chromosomes migrated in PFGE as faint bands, suggesting their circular topology. Strains ADV1 and -2 harbored two chromosomes of 2.38 and 2 Mb, whereas in strains ADV3 to -7 the size of the large chromosome was 2.38 Mb and that of the small one was 1.85 Mb. Owing to the sizes of I-_Ceu_I fragments (Fig. 2A, lanes 3 and 4, and B, lanes 1 and 2) and the sizes of undigested chromosomes, we deduced that each chromosome of strains ADV1 and -2 carried two rrn copies separated by 250 kb on the large chromosome and by 150 kb on the small one. Moreover, I-_Ceu_I digestion indicated the loss of the 150-kb fragment in strains ADV3 to -7 (Fig. 2B). These results suggested that a large deletion of 150 kb occurred in the small chromosome and that it was related to the loss of one rrn copy in strains ADV3 to -7. The sequencing of the 23S rRNA gene of strain ADV1 showed the presence of a _Spe_I site in position 932 (E. coli numbering), about 1 kb before the I-_Ceu_I site. The location of this _Spe_I site explained why DNA deleted by genomic rearrangement corresponded to _Spe_I and I-_Ceu_I restriction fragments of nearly the same size in PFGE experiments.

To determine the orientation of rrn operons on the two chromosomes, we performed hybridization experiments using 16S rDNA and 23S rDNA probes on I-_Ceu_I digestion products (data not shown). Then, insertion copies were located using the 46-bp insertion probes on I-_Ceu_I digestion products. An I-_Ceu_I map and rrn skeleton of the two chromosomes of strains ADV1 and ADV3 are schematically represented in Fig. 4. We observed in the small chromosomes of strains ADV3 to -7 a large deletion of 150 kb, which included rrnC. Moreover, rrnA remained on the small chromosome but lost the 46-bp insertion. Direct sequencing of 16S rDNA amplification product obtained with _Hin_dIII fragment A of strain ADV3 as the template gave a partial sequence of rrsA (accession no. AF526516), which was strictly identical to those of insertion-free rrs copies of O. intermedium.

FIG. 4.

FIG. 4.

Schematic representation of I-_Ceu_I map and rrn skeleton of the two chromosomes of strains ADV1 and -2 (A) and ADV3 to -7 (B). I, I-_Ceu_I restriction site. Sizes of chromosomes and I-_Ceu_I restriction fragments are indicated in megabases and kilobases, respectively. Arrows indicate the orientation of rrn operons. Black circles, 46-bp insertion.

The loss of one 16S rDNA copy and/or the genomic deletion that occurred seemed to affect the phenotypes of the strains. Indeed, the colonies of strains ADV1 and -2 were mucoid and quickly became confluent during incubation at 37°C, as described for the type strain (50) and for clinical isolates (47) of O. intermedium, whereas the colonies of strains ADV3 to -7 were smooth and nonmucoid. Moreover, strains ADV3 to ADV7 showed a growth rate about 30% lower than that for strains ADV1 and ADV 2 (data not shown). Statistical analysis showed that the difference in growth rates was significant (P = 0.0143).

Expression and two-dimensional structure of the 46-bp insertion.

We selectively amplified 16S rRNA molecules carrying the 46-bp insertion by RT-PCR using the insertion-specific primer ins1 and the consensual primer 590r as forward and reverse primers on total-RNA extracts. RT-PCR products of about 380 bp were obtained from O. intermedium strains ADV1, ADV3, ADV9, and PR17/sat (Fig. 5A) and strains ADV2 and ADV4 to -7 (data not shown). The size of a fragment was in accordance with the position of the 46-bp insertion in the 16S rRNA gene. No amplification was obtained for strain LMG 3301T (Fig. 5A) and the other strains (data not shown). The RT-PCR performed using the consensual primers 27f and 590r on RNA extracts of strains ADV1, ADV3, ADV9, PR17/sat, and LMG 3301T gave positive controls. PCR amplification of RNA samples without reverse transcription was negative in all experiments. We concluded that the 46-bp insertion was expressed in the 16S rRNA.

FIG. 5.

FIG. 5.

Expression of the 46-bp insertion and detection of the two types of 16S rRNA by RT-PCR. The strains analyzed are indicated at the top. (A) Lanes +, use of consensual primer pair 27f and 590r; lanes ins, use of an insertion-specific pair of primers, ins1 and 590r. (B) Lanes +, RT-PCR products obtained with the primers 27f and 590r; lanes _Dra_I, RT-PCR products obtained in + lanes digested by _Dra_I. Lanes − (A and B), negative controls performed on the RNA of strain PR17/sat without reverse transcription and with the primers 27f and 590r (negative controls were also obtained for other strains and for the primers ins1 and 590r but are not shown); lane M, 100-bp DNA ladder (Promega) as a molecular weight marker.

RT-PCR products obtained by using the consensual primers 27f and 590r on RNA extracts of strains LMG 3301T, ADV1, ADV3, ADV9, and PR17/sat were then digested by the endonuclease _Dra_I. As seen before, the 46-bp insertion introduced a unique _Dra_I restriction site in the 16S rDNA of O. intermedium, and then _Dra_I digestion could be used to visualize the amplification of both 16S rRNA types. _Dra_I partially digested the RT-PCR products obtained from strains ADV1, ADV3, and ADV9, confirming the expression of both types of 16S rRNA in the strains harboring heterogeneous 16S rDNA copies (Fig. 5B). The sizes of the digested fragments (380 and 170 bp) were in accordance with the position of the 46-bp insertion in the 16S rDNA. The intensity of the undigested band of 550 bp, which corresponded to the 16S rRNA without insertion, seemed to decrease slightly as the number of insertion copies in 16S rDNA increased (Fig. 5B). This result was obtained in three independent experiments. The insertion-free strain LMG 3301T and strain PR17/sat, which carried the 46-bp insertion on all the 16S rDNA copies, were, respectively, used as negative and positive controls for the _Dra_I digestion of RT-PCR products (Fig. 5B).

When analyzed with the MFOLD program, the sequence folded into a stem-loop structure with a predicted free energy −36.4 kcal/mol at 37°C. This secondary structure was placed on the secondary structure model of the 16S rRNA molecule of Brucella suis described by Gutell et al. (http://www.rna.icmb.utexas.edu/). The 46-bp insertion took place in, and prolonged helix H184 of, the 16S rRNA (Fig. 6A and D). Helix H184 was variable in sequence among alpha proteobacteria and carried a tetranucleotide, allowing distinction between O. anthropi (TTCG) and O. intermedium (GAAA) (Fig. 6B and C). However, in 2,184 proteobacteria analyzed in the Gutell laboratory database (http://www.rna.icmb.utexas.edu/), this stem-loop is formed by 10 bp and neither less nor more nucleotides are known to exist in this region, indicating the very unconventional character of the 46-bp insertion.

FIG. 6.

FIG. 6.

Secondary structure of the 46-bp insertion. (A) Secondary structure of the insertion (boldface) replaced in O. intermedium 16S rRNA partial two-dimensional structure (positions 139 to 204; E. coli numbering). (B and C) Structures of helix H184 in 16S rRNA of O. anthropi and O. intermedium, respectively. (D) Location of the region carrying helix H184 on the E. coli 16S rRNA secondary-structure model.

We tested a few phenotypical traits of the strains LMG 3301T, ADV9, and ADV1. These strains displayed the same overall genomic organization, as judged by I-_Ceu_I digestion profiles (data not shown), but differed in the copy number of the insertion (0, 2, and 3, respectively). We saw no difference in colony aspect, morphology after Gram stain, biochemical traits on API 20E and API 20NE systems, and antibiotic susceptibility profiles. Moreover, strains LMG 3301T, ADV9, and ADV1 showed similar growth curve profiles. Thus, the few characters we tested did not allow us to relate the expression of the insertion to a particular phenotype.

DISCUSSION

We found in the O. intermedium genome four rrn copies. Two copies were located on the large chromosome, and the two others were located on the small one. The two copies were in the same orientation on each chromosome. While there are limited examples where multiple copies have been characterized for the same alpha proteobacterium, 16S rDNA copies appeared to be homogeneous in this subphylum. Thus, alignment of rrs sequences obtained from totally sequenced alpha proteobacterial genomes by using the BLAST program showed that the rrs copies in Brucella melitensis, Brucella suis, Agrobacterium tumefaciens, Sinorhizobium meliloti, Mesorhizobium loti, and Caulobacter vibrioides genomes were strictly identical. In contrast, we observed different copies of 16S rDNA in the genome of O. intermedium, a species closely related to the genus Brucella. This is the first described case of rrs heterogeneity in the alpha subgroup of proteobacteria. We found atypical copies of 16S rRNA genes with an additional sequence of 46 bp at position 187 (E. coli numbering) in several isolates of O. intermedium. Since this insertion was also present in 16S rRNA, it could not be related to IVSs that have been described for a few eubacterial species (11, 12, 24, 25, 41, 42, 44, 45). Up to this date, such a large insertion remaining in the 16S rRNA had only been described for D. australicum (37) and Bacillus sporothermodurans (40), two species belonging to the phylum comprising Bacillus and Clostridium. We present here the first description of such an insertion in the Proteobacteria phylum. The 46-bp insertion was found in seven clinical isolates belonging to the same clone and in another clinical strain epidemiologically unrelated. Moreover, we found the insertion in an O. intermedium strain isolated from the nematode Heterorhabditis indica in association with the bacterial symbiont Photorhabdus luminescens (3). Clinical and nematode isolates were geographically unrelated since the former were isolated in France and the latter were isolated in Puerto Rico. Thus, the atypical insertion in 16S rRNA of O. intermedium is not an idiosyncrasy of a particular strain but is on the contrary widely represented among natural isolates. However, the 46-bp insertion was absent from O. intermedium 16S rDNA sequences previously deposited in the databases. 16S rRNA gene sequences obtained by direct sequencing of PCR products leading to preferential amplification of one copy or by cloning one isolated copy could lead to underestimation of the occurrence of the insertion.

The question about the origin of the 46-bp insertion in the 16S rRNA genes can be discussed only hypothetically. The GC content of the insertion (47%) was compared to the GC content of the entire 16S rDNA of O. intermedium (55%) and to that of the region surrounding the atypical insertion, from position 139 to 204 (45%). The similarity in GC content neither suggested nor ruled out the hypothesis of acquisition by lateral transfer. Moreover, the sequence of the 46-bp insertion did not match any other sequences of other species. Thus, arguments for lateral transfer were quite inconclusive.

When it was included in the secondary-structure model of the 16S rRNA molecule of Brucella suis and E. coli, the 46-bp insertion prolonged the helix H184. This stem-loop has been described as conserved in length among the proteobacteria (http://www.rna.icmb.utexas.edu/) suggesting the atypical character of the 46-bp insertion. The Database of Ribosomal Cross-Links (4) indicated that helix H184, precisely from position 189 to 191, was involved in the cross-link between 16S rRNA and ribosomal protein S13. Ribosomal protein S13 has been described as important for both translation initiation and elongation because it is cross-linked to the three translation initiation factors and to tRNA in the P site (6). RT-PCR experiments showed that copies with and without the insertion were expressed. Although these experiments did not quantify the expression of each type of 16S rRNA gene, the amounts of the two types of 16S rRNA appeared to correlate with the proportion of both types of 16S rDNA, as judged by the intensity of the RT-PCR-amplified fragments. However, the problem of a potential phenotypical impact of the 46-bp insertion could not be solved in this work.

We analyzed clonal strains chronologically isolated from respiratory samples of a patient with chronic O. intermedium carriage over a 1-year period. The loss of two insertion copies and a genomic rearrangement occurred between the second (ADV2) and the third (ADV3) isolates obtained. In vitro, a variation in colony aspect and growth rate between strains ADV2 and ADV3 was observed. The elongation of doubling time could be related to the loss of one rrn in strain ADV3, as previously described for E. coli after inactivation of a variable number of rrn copies (9).

The new genomic organization and the new phenotype were maintained in four subsequent isolates obtained over a 4-month period. We considered that genomic modifications appeared naturally but not in vitro, because isolates were analyzed with no more than one subculture event. This suggests that the new genomic organization gave a selective advantage to the strain in vivo. We never observed a mixture of mucoid and nonmucoid phenotypes in the seven clinical samples obtained from the patient. Moreover, the PFGE patterns did not suggest a mixture of two different genomic structures in the same DNA preparation. As a consequence, both genotypes and both phenotypes did not coexist in the samples tested. This did not rule out the possibility of coexistence of the two types of strain in the patient at an unexplored stage of the carriage.

The relation between host-restricted life style and a small genome size is patent in bacteria, particularly in alpha Proteobacteria (33). O. intermedium is a free-living bacterium and has a larger genome than its phylogenetic neighbors with an intracellular life style, such as Brucella and Bartonella (17, 33). It might be suggested that the 1-year restriction of isolates ADV1 to -7 to a very narrow ecological niche, i.e., the human respiratory system, led to reductive evolution (2, 33).

The 16S rDNA copy that lost the insertion (rrsA in strains ADV3 to -7) exhibited exactly the same sequence as the insertion-free copy (rrsB). When present, the 46-bp insertion replaced the tetranucleotide GAAA classically observed in O. intermedium 16S rDNA. Excision of an inserted element by a site-specific recombination event did not lead to the reformation of the GAAA tetranucleotide (Fig. 6). Gene conversion has been previously proposed as the mechanism of homogenization among rrn copies in bacteria (13, 22). Thus, a homogenization process due to gene conversion among rrn copies could explain the loss of the insertion. Physical mapping of the two chromosomes showed that rrnA and rrnB were on different chromosomes. Thus, we describe here a case of genetic recombination between two independent bacterial chromosomes.

A second insertion copy was lost due to the deletion of one rrn copy and 150 kb of the small chromosome. The two rrn copies flanking the 150-kb fragment on the small chromosome of strain ADV1 were in the same orientation, suggesting that a deletion event occurred by homologous recombination between the two rrn copies of the small chromosome (2, 18). Variations in genome structure by rearrangements at rrn loci have previously been described in host-specialized Salmonella serovars (S. enterica serovars Typhi, Paratyphi C, Gallinarum, and Pullorum) (26), in Vibrio cholerae (20), and in the genus Pasteurella (27). In the alpha proteobacterium Brucella suis, the differences in chromosome size and number have been explained by rearrangements at chromosomal regions containing the three rrn genes. The location and orientation of these genes suggested that these rearrangements were due to homologous recombination at the rrn loci (18). In the case described here, homologous recombination between rrn copies in the same orientation led to a deletion of a 150-kb genomic region flanked by the two copies. One of the two rrn copies was also deleted in this rearrangement process. Although deletion or duplication of rrn was observed a long time ago in Salmonella enterica serovar Typhimurium (1) and Bacillus subtilis (52) during laboratory maintenance, the first descriptions of natural isolates of V. cholerae were done recently (19, 20). The presence of tandemly repeated operons and the occasional deletion of one of the pairs are similar to the situation for a Bacillus subtilis laboratory strain (52). Deletion of an rrn operon is observed to occur quite often, presumably by intrachromosomal recombination within the tandemly repeated sets. In contrast, a recombination event in the small chromosome of O. intermedium strain ADV1 occurred between two distant rrn copies separated by 150 kb.

The alpha proteobacterial phyletic group is characterized by a high occurrence of complex genomes and often-important genome structure variations among bacteria belonging to the same species or the same genus. This is the case in the species Brucella suis (18) and in the genera Azospirillum (30), Rhizobium, Agrobacterium, and Rhodobacter (17). A high frequency of rearrangement processes involving homologous recombination between rrn copies could explain some of the variations observed in the genome structure (18). Moreover, as we showed before, a 100% sequence identity among 16S rDNA copies has been observed in all alpha proteobacterial genomes sequenced so far. In contrast, a low level of heterogeneity among copies was observed in the majority of totally sequenced genomes of bacteria belonging to other phyla. The homogeneity in rrn copy sequences observed in alpha proteobacteria might reflect very efficient concerted evolution among members of the rrn multigenic family in this bacterial phylum. The atypical insertion we observed in some strains of O. intermedium allowed us to monitor the genetic exchange among rrn copies in a set of clonal isolates. The results gave a new illustration of genomic rearrangement among rrn copies in alpha proteobacteria and an additional argument in favor of a particularly high frequency of genetic exchanges among rrn copies in this bacterial subphylum.

Acknowledgments

We gratefully acknowledge Hélène Darbas, Hélène Jean-Pierre, and Noël Boémare for providing bacterial strains. We thank Jean-Luc Jeannot, Agnès Masnou, and Isabelle Diego for technical assistance and Claudine Gras for her contribution. We also thank Gregory Nickson for reading the manuscript.

This work was supported by the association ADEREMPHA, Montpellier, France.

REFERENCES