Dispersal and Evolution of the Sinorhizobium meliloti Group II RmInt1 Intron in Bacteria that Interact with Plants
http://www.100md.com
分子生物学进展 2005年第6期
* Grupo de Ecología Genética, Departamento de Microbiología del Suelo y Sistemas Simbióticos, Estación Experimental del Zaidín, Consejo Superior de Investigaciones Científicas, calle Profesor Albareda 1, Granada, Spain; and Laboratorium voor Microbiologie, Vakgroep Biochemie, Fysiologie, Microbiologie, Faculteit Wetenschappen, Universiteit Gent, Gent, Belgium
Correspondence: E-mail: ntoro@eez.csic.es.
Abstract
Group II introns are both self-splicing RNAs and mobile retroelements found in bacterial and archaeal genomes and in organelles of eukaryotes. They are thought to be the ancestors of eukaryote spliceosomal introns and non–long terminal repeat retrotransposons. We show here that RmInt1, a bacterial group II intron first described in the nitrogen-fixing symbiont of alfalfa (Medicago sativa) Sinorhizobium meliloti, is also present in other Sinorhizobium and Rhizobium species. The intron-homing sites in these species are IS elements of the ISRm2011-2 group as in S. meliloti, but ectopic insertion is also observed. We present evidence that these related bacteria have acquired RmInt1 by vertical inheritance from a common ancestor and by independent horizontal transfer events. We also show that RmInt1 is mobile in related taxa of bacteria that interact with plants and tends to evolve toward an inactive form by fragmentation, with loss of the 3' terminus including the intron-encoded protein. Our results provide an overview of the evolution and dispersion of a bacterial group II intron.
Key Words: group II intron ? RmInt1 ? nitrogen fixation ? legumes ? Sinorhizobium meliloti ? Rhizobiales ? ribozyme ? reverse transcriptase
Introduction
Group II introns are self-splicing, mobile retroelements that splice via a lariat intermediate in a mechanism similar to that of spliceosomal introns (Michel and Ferat 1995). These genetic elements were first found in mitochondrial and chloroplast genomes in lower eukaryotes and plants (Michel, Umesono, and Ozeki 1989) and were later identified in eubacteria (Ferat and Michel 1993; Martínez-Abarca, Zekri, and Toro 1998; Martínez-Abarca and Toro 2000a; Zimmerly, Hausner, and Wu 2001; Dai and Zimmerly 2002a) and some archaebacterial species (Dai and Zimmerly 2003; Toro 2003). Group II introns are large catalytic RNAs displaying a conserved secondary structure with six double-helical domains (dI to dVI), one of which (dIV) may incorporate a reverse transcriptase (RT) open-reading frame (ORF) (Michel and Ferat 1995). Unlike organellar introns, most bacterial group II introns possess an intron-encoded protein (IEP; Dai and Zimmerly 2002a). Nevertheless, ORF-less group II introns have been found, e.g., in archaeabacteria (Dai and Zimmerly 2003) and the cyanobacterium, Thermosynechococcus elongatus BP-1 (Nakamura et al. 2002). Three main phylogenetic classes (IIA, IIB, and IIC) of group II introns have been described based on IEP and conserved intron RNA structures (Michel, Umesono, and Ozeki 1989; Toor, Hausner, and Zimmerly 2001; Zimmerly, Hausner, and Wu 2001; Toro, Molina-Sánchez, and Fernández-López 2002; Ferat, Le Gouar, and Michel 2003; Toro 2003). Interestingly, most bacterial introns are not located in conserved genes and most are associated with mobile DNAs (Ferat, Le Gouar, and Michel 1994; Martínez-Abarca and Toro 2000a; Granlund, Michel, and Norgren 2001; Dai and Zimmerly 2002a, 2003; Toro 2003). Over half are fragmented (Dai and Zimmerly 2002a). The IEPs of these introns have four conserved domains, including an N-terminal RT domain, domain X, a putative RNA-binding domain associated with RNA splicing or maturase activity, C-terminal DNA-binding (D) domain, and a DNA-endonuclease (En) domain. Maturase activity is required for the splicing reaction in vivo, resulting in an excised intron RNA lariat, which forms a ribonucleoprotein complex with the IEP that is involved in the mobility of the intron to intron-less alleles in a process known as retrohoming, which requires both RT and En activities (Lambowitz et al. 1999; Belfort et al. 2002). Group II introns also move (albeit at low frequency) into ectopic sites resembling the normal homing site, in a process known as retrotransposition (Yang et al. 1998; Martínez-Abarca and Toro 2000b; Dickson et al. 2001; Mu?oz, Villadas, and Toro 2001; Ichiyanagi et al. 2002; Ichiyanagi, Beauregard, and Belfort 2003), which is thought to facilitate intron dispersal.
Nuclear pre–messenger RNA introns (Michel and Ferat 1995) and non–long terminal repeat retrotransposons are both thought to descend from mobile group II introns (Eickbush 1999). According to this hypothesis, group II introns originated in bacteria and invaded the nucleus of a primitive eukaryote directly or from an organelle; they were then fragmented to form the spliceosome (Sharp 1991). The horizontal transfer of a self-splicing group II intron from a cyanobacterium to the chloroplast genome of Euglena myxocylindracea was recently shown (Sheveleva and Hallick 2004), suggesting that this process is still occurring in nature. In bacteria, the evidence for horizontal gene transfer events involving group II introns is fragmented and indirect, being based mostly on the finding that a particular class of intron is present in various host organisms and that a particular organism harbors various classes of intron. The Azotobacter vinelandii group II intron (Avi.groEL) provides an illustration of this. This intron has been inserted into the termination codon of the groEL coding sequence, and this insertion site is conserved in at least one other species of Azotobacter (Azotobacter chroococcum), suggesting intron transposition after divergence of the two taxa (Ferat, Le Gouar, and Michel 2003). Another example is the group II introns found in Methanosarcina acetivorans and Methanosarcina mazei, which were probably acquired from eubacteria (Dai and Zimmerly 2003; Toro 2003).
RmInt1 was the first group II intron described in the order Rhizobiales. It was identified in Sinorhizobium meliloti, the nitrogen-fixing symbiont of alfalfa (Medicago sativa). The RmInt1 IEP, like those of many other bacterial group II introns, lacks the DNA-En and a cognate C-terminal DNA-binding domain (Martínez-Abarca, García-Rodríguez, and Toro 2000; Zimmerly, Hausner, and Wu 2001; Dai and Zimmerly 2002a; San Filippo and Lambowitz 2002; Toro 2003). RmInt1 is nevertheless an efficient mobile element with a homing frequency (proportion of cells containing at least one homing event) approaching 100%, similar to that of fungal mitochondrial DNA introns and the lactococcal Ll.ltrB intron, inserting into 20%–45% of recipient target plasmids (Martínez-Abarca, García-Rodríguez, and Toro 2000; Jiménez-Zurdo et al. 2003). This mobile intron also transposes to ectopic sites (Martínez-Abarca and Toro 2000b; Mu?oz, Villadas, and Toro 2001). RmInt1, like other group II introns lacking the IEP En domain, may use a nascent lagging strand at DNA replication forks for priming (Martínez-Abarca et al. 2004).
This study aimed to investigate the further spread of bacterial group II introns in nature and to trace their possible evolution and involvement in horizontal transfer events. We searched for the S. meliloti RmInt1 intron and investigated its distribution in related taxa of bacteria that interact with plants. We present evidence that Sinorhizobium species inherited RmInt1 vertically from a common ancestor within the order Rhizobiales, with subsequent gains and losses in the various species and that this intron has been transferred horizontally to some Rhizobium species. We show that intron moves efficiently to homing sites in related taxa and that it tends to be inactivated by fragmentation with loss of the 3' terminus. This study sheds light on the dispersal and evolution of bacterial group II introns in natural populations.
Materials and Methods
Bacterial Strains, Media, and Growth Conditions
The bacterial species of the order Rhizobiales and the S. meliloti isolates used are listed in tables 1 and 3. Escherichia coli DH5 was routinely cultured at 37°C in Luria-Bertani medium (Sambrook, Fritsch, and Maniatis 1989), and rhizobial bacteria were grown at 28°C on tryptone yeast extract (TY) (Agrobacterium, Mesorhizobium, Rhizobium, and Sinorhizobium), yeast extract mannitol (YEM) (Bradyrhizobium) (Vincent 1970), or yeast extract bactobeet (YEB) (Azorhizobium and Xanthobacter) (Geremia et al. 1994). Ampicillin was added to the medium at a concentration of 200 μg/ml for E. coli.
Table 1 Bacterial Strains Used in this Study
Table 3 Number of Copies of RmInt1 and ISRm2011-2 in Sinorhizobium meliloti Isolatesa
DNA Hybridization and Fingerprinting
Total DNA was isolated according to standard protocols (Sambrook, Fritsch, and Maniatis 1989). DNA (2 μg) was digested with EcoRI or BamHI and subjected to electrophoresis in 0.8% Tris-acetate agarose gels. It was then blotted onto positively charged nylon membranes (Roche Diagnostic, Mannhein, Germany) according to the manufacturer's instructions. The DNA probes used for ISRm2011-2 and RmInt1 and the method used for DNA hybridization under high stringency conditions have been described elsewhere (Martínez-Abarca and Toro 2000b). The probe for ISRm10-3 was obtained with primers 5D1 (5'-ACGTGTTCTTGCCGCCGT-3') and 5D2 (5'-TTGCGCGTCTGATTGTGC-3') with a similar procedure.
DNA Sequencing and Inverse Polymerase Chain Reaction
Copies of RmInt1 were isolated from Sinorhizobium medicae DNA by BamHI digestion, electrophoresis in agarose gels, excision with the Perfectprep Gel Cleanup kit (Eppendorf AG, Hamburg, Germany), and hybridization with intron-derived probes (fig. 1A). DNA from the excised bands and DNA from strains with a single copy of RmInt1 were subjected to polymerase chain reaction (PCR) with primers GII.1 (5'-AAIAGICITGGTIGTGAGCG-3') and GII.4 (5'-TCTCGCAGAACIGTICGTGA-3') to amplify RmInt1. The resulting PCR fragments were inserted into the pGEM-T vector (Promega, Madison, Wis.) and sequenced with primers T7, SP6, 1, Int2, H3 (5'-GTATTGTTTGAAACAACTG-3'), and EB70 (5'-ATGGTGGTCAAGCAGATGA-3') (fig. 1). The internal sequence of fragmented introns from strains with a single copy was obtained by PCR with primers 1 and 18R.0, purification of the fragment with S300HR columns (Pharmacia, Little Chalfont, UK), and sequencing in an automatic laser fluorescent DNA sequencer using the same primers.
FIG. 1.— Schematic representation of the Sinorhizobium meliloti RmInt1 intron and the copies sequenced in other related taxa. (A) The mobile RmInt1 intron of S. meliloti strain GR4. The diagram shows ribozyme dI to dVI. The ORF is within intron dIV. Protein domains are RT and maturase (X), with a C-terminal extension indicated by an asterisk (San Filippo and Lambowitz 2002; Mu?oz-Adelantado et al. 2003). The primers used for PCR amplification and sequence analysis are indicated by arrows and their relative nucleotide positions within the intron sequence are shown in parentheses. The 5' intron probe and the 3' intron ORF-probe used for DNA hybridization are also indicated. (B) Full-length RmInt1 intron copies in various hosts from the Rhizobiaceae family. Diagrams are constructed as described above. Numbers at the beginning and at the end show the positions of the determined sequences with respect to RmInt1. Relevant features such as stop codons and deletions are also indicated. Dotted lines indicate DNA sequences that were not determined. Percentage identity to the sequence of S. meliloti RmInt1 is shown on the right. The accession numbers of the sequences are AY248839 for Sinorhizobium adhaerens 5D19, AY608908 [GenBank] for Sinorhizobium terangae ORS22, AY608905 for S. terangae ORS1009, AY608907 for Sinorhizobium medicae RMO09-1, and AY608906 for S. medicae RMO09-2. (C) Fragmented RmInt1 intron copies in various hosts from the Rhizobiaceae family. Diagrams are constructed as described above. Continuous lines indicate a lack of similarity of the sequence to sequences in the database. The abbreviated intron designation is indicated below each strain. The accession numbers of the sequences are AY608902 for Rhizobium etli Viking 1, AF176227 for R. etli CE 3, AY608903 for S. medicae RMO02, and AY608901 for S. adhaerens LMG20582
The 5' and 3' exon junctions of insertion sites were identified by inverse PCR with divergently annealing primers. Total DNA from Rhizobium etli Viking 1, Sinorhizobium terangae ORS22, and Sinorhizobium adhaerens 5D19 was first digested with EcoRI; DNA from S. medicae RMO02 was digested with BamHI; and DNA from S. adhaerens LMG20582was digested with EcoRV. The exon 3' to complete introns was amplified with primers PR1000 (5'-GCGGAAGATTGTCAAACAGC-3') and ICF (5'-CTGTTCTCTCTGGCTGACTACG-3'), whereas the 5' exon and the exons from truncated introns were amplified with primers ICI (5'-AGGATGACGAAACGGTCCT-3') and H3. The products of the amplification reactions were sequenced directly as described above. DNA sequence editing, translation and analysis, and sequence similarity searches were carried out as previously described (Martínez-Abarca and Toro 2000b).
Phylogenetic Analysis
Analyses were carried out with PHYLIP version 3.573c, using the programs Seqboot, DNAdist, Protdist, Neighbor, Protpars, DNAML, and Consense (Felsenstein 1995). Sequences were aligned by ClustalW (http://bioweb.pasteur.fr). Reference 16S ribosomal DNA (rDNA) sequences were retrieved from European Molecular Biology Laboratory and aligned using Bionumerics (Applied Math); a distance matrix was obtained with Kimura-2 correction, and a neighbor-joining tree was generated with TreeCon (version 1.3b, Van de Peer and De Wachter 1994). A bootstrap analysis of 500 replicate data sets was performed with the same program.
RmInt1 Mobility Assays
Donor (pKG2.5) and receptor pJB0.6s(+) plasmids were transferred successively from E. coli to the selected rhizobial strains by triparental mating, using the helper plasmid pRK2013 (Ditta et al. 1980). Mobility of intron copies from the genome of S. terangae ORS22 and S. adhaerens 5D19 to the receptor plasmids pJB0.6s(+) and pJB0.6as (Martínez-Abarca et al. 2004) was also tested. Homing events were analyzed as previously described (Martínez-Abarca, García-Rodríguez, and Toro 2000; Martínez-Abarca et al. 2004).
Results
Presence and Distribution of RmInt1 in the Order Rhizobiales
We evaluated the presence and distribution of RmInt1 in the order Rhizobiales. We carried out Southern blotting for 172 isolates corresponding to 29 species, 7 genera, and 4 families of bacteria, all able to interact with plants (table 1). Two different RmInt1-derived DNA probes—a 448-bp 5' intron probe and a 238-bp 3' intron ORF-probe (see fig. 1)—were used in hybridization analysis to make it possible to distinguish between full-length and fragmented introns. We focused exclusively on strong hybridization signals corresponding to sequences displaying >85% sequence identity to RmInt1 at the DNA level, as shown by subsequent analyses (see below). Within the Rhizobiaceae, our hybridization data indicated that RmInt1 was present in the genus Sinorhizobium, in S. meliloti, S. medicae, S. adhaerens, and S. terangae. It was not detected in Sinorhizobium fredii, Sinorhizobium saheli, or in the reference strains of Sinorhizobium kostiense, Sinorhizobium arboris, Sinorhizobium xinjiangensis, and Sinorhizobium morelense. In the other branches of the Rhizobiaceae tested, this intron was detected only in R. etli and Rhizobium leguminosarum bv. phaseoli (see tables 1 and 2, fig. 2A and B). No band hybridizing with the RmInt1 probes was detected in bacteria of Phyllobacteriaceae, Bradyrhizobiaceae, or Hyphomicrobiaceae, although some of the bacteria of these families (e.g., Bradyrhizobium japonicum) harbor phylogenetically related introns (Dai and Zimmerly 2002a; Toro 2003; Toro et al. 2003).
Table 2 Number of Copies of RmInt1, ISRm2011-2, and ISRm10-3 in Species of the Order Rhizobiales Carrying the Group II Intron
FIG. 2.— Presence of RmInt1 in Sinorhizobium medicae, Rhizobium etli, and Rhizobium leguminosarum bv. phaseoli. Total DNA from the strains was digested with EcoRI (R. etli and R. leguminosarum) or BamHI (S. medicae), separated on a 0.8% agarose gel, and transferred onto a nylon filter. (A) Sinorhizobium medicae RMO09 DNA was first hybridized with a 3' intron ORF-probe, then washed, and hybridized with a 5' intron probe. Asterisk indicate the excised bands used for sequencing. (B) Rhizobium etli (CE3, CFN42, and Viking 1) and R. leguminosarum bv. phaseoli (PhP17, Ro33, and Ro34) DNA hybridized with the 5' intron probe. The sizes of the molecular weight markers (in kilobits) are indicated on the left.
All the strains and field isolates of R. etli and R. leguminosarum bv. phaseoli harboring copies of RmInt1 generated positive hybridization signals with the 5' intron probe but not with the 3' intron ORF-probe. This implies that RmInt1 has lost the IEP or is generally truncated at the 3' terminus in these bacterial species (table 2). In Sinorhizobium, fragmented RmInt1 introns were also detected in S. medicae, S. adhaerens, and S. meliloti, as shown by the smaller number of hybridizing bands obtained with the RmInt1 3' intron ORF-probe than with the 5' intron probe (table 2). After S. meliloti, RmInt1 appears to be most frequent in the related species S. medicae (seven of eight strains tested), but most of the copies detected appeared to be truncated at the 3' terminus (table 2). A larger study on 19 different S. meliloti isolates (table 3) suggested that RmInt1 is present predominantly in its full-length form in this bacterial species, confirming genome sequence data for strain 1021 showing the presence of three full-length copies of RmInt1 and no fragmented copies of this intron (Galibert et al. 2001; Toro et al. 2003). Nevertheless, there is considerable diversity among S. meliloti strains in terms of the number of bands detected with the two intron-derived probes, with all strains except 5D47 having more bands hybridizing to the 5' intron probe than to the 3' intron ORF-probe, consistent with RmInt1 fragmentation occurring predominantly via truncation at the 3' terminus.
The Full-Length Rhizobiaceae RmInt1 Intron
We also studied strains from species other than S. meliloti, with the aim of characterizing the full-length copies of the RmInt1 intron and obtaining the corresponding DNA sequence. Total DNA from strains harboring more than one copy of the intron was digested with BamHI, a restriction enzyme that does not cut within the intron, and hybridized with the intron-derived probes. Southern blots suggested that S. medicae RMO09 harbored at least three full-length copies of the intron (fig. 2A). Sinorhizobium terangae strains ORS22 and ORS1009 and the S. adhaerens strain 5D19 gave single hybridizing bands of different sizes with the intron-derived probes, suggesting that these strains contained a single full-length copy of RmInt1 (data not shown). DNA corresponding to the 20- and 7.5-kb hybridizing bands from S. medicae (fig. 2A) were excised and purified from an agarose gel and subjected to PCR with the intron-derived primers GII.1 and GII.4 (fig. 1A). We also performed PCR with the same primers plus total DNA from S. terangae strains ORS22 and ORS1009 and sequenced S. adhaerens DNA with the intron-derived primers to obtain overlapping sequences (see fig. 1A). Finally, a partial DNA sequence (1,604–1,717 nt) was obtained for each RmInt1 intron copy, including the complete dIV sequence (fig. 1B). The full-length intron sequence (1,884 nt) and exon sequences were obtained for the S. adhaerens 5D19 and S. terangae ORS22 RmInt1 elements by inverse PCR (fig. 1B). DNA sequence data showed that the S. medicae RmInt1 element located in the 20-kb BamHI fragment (hereafter referred to as copy 1) and that within the 7.5-kb BamHI fragment (copy 2) were 99% identical to S. meliloti RmInt1 (strain GR4). Identities of 93% were obtained for the intron in S. adhaerens and of 89% and 85% for the RmInt1 elements in S. terangae strains ORS22 and ORS1009, respectively. Most of the nucleotide changes were silent and those resulting in amino acid changes in the IEP were mostly clustered in the interdomains or in the spacers within the RT domain. Within the ribozyme, most of the observed nucleotide changes were complementary and therefore did not alter the predicted secondary structure of the ribozyme. Nevertheless, it should be noted that S. medicae RmInt1 copy 1 has four nucleotide changes and lacks a G at position 792 that results in a frameshift mutation, producing a stop codon at position 842, resulting in a truncated IEP of 98 amino acids (fig. 1B), whereas copy 2 has only one nucleotide change. The intron in S. terangae ORS1009 has a truncated IEP of 325 amino acids due to a stop codon generated in the maturase domain at position 1540 and a deletion of 18 nt that results in the loss of six amino acid residues at the start of the RT 0 domain (fig. 1B). Thus, at least some of the full-length RmInt1 introns appear to have become inactive in their hosts. Moreover, mobility assays using target recipient plasmids did not show homing events for the S. terangae ORS22 and S. adhaerens 5D19 introns (data not shown).
Phylogenetic Analyses of the Rhizobiaceae RmInt1 IEP
We investigated the evolutionary relationships between the RmInt1 introns found in the various hosts from the Rhizobiaceae by aligning the amino acid sequences of RmInt1 IEP copies obtained and correcting sequences to obtain a full IEP when required. Phylogenetic trees were constructed by neighbor-joining (fig. 3A) and parsimony analysis. Similar branching patterns were obtained with these algorithms. Phylogenetic analysis identified two main groups well supported by bootstrapping analysis: one including the RmInt1 IEPs from S. meliloti, S. medicae, and S. adhaerens, (hereafter referred to as the S. meliloti group) and the second containing the two RmInt1 IEPs found in S. terangae strains (hereafter referred to as the S. terangae group). The RmInt1 IEP–based phylogenetic tree was consistent with that based on 16S rDNA sequences (fig. 3C). Thus, the RmInt1 intron in these Sinorhizobium species may have been vertically inherited from a common ancestor, with intron loss in several species, or by horizontal transfer among closely related strains and species.
FIG. 3.— Phylogeny of the RmInt1 group II intron. Phylogenetic analyses (A) and (B) were carried out, after ClustalW alignment, with the Neighbor, parsimony, and maximum likelihood modules of the PHYLIP package (using default settings). In both diagrams, a consensus neighbor-joining phylogram is shown, but similar results were obtained with all the methods that were used. Bootstrap values are expressed as percentages and were derived from 1,000 samplings, and those exceeding 75 are shown at the branch points. Maximum parsimony (A) and maximum likelihood (B) values are also shown in italics and parentheses. Escherichia coli IntB group II intron (X77508 [GenBank] ) was used as the out-group. (A) Phylogeny of the IEP based on the alignment of 419 amino acid residues. (B) Phylogeny of the 5' end of the ribozyme based on alignment of the 370 nt from position 180 to 550 of the RmInt1 sequence, including part of dI and extending to the start codon of the protein encoded by dIV. (C) The 16S rDNA phylogeny of the main rhizobial groups. The sequences analyzed for tree construction were obtained from the European Molecular Biology Laboratory database and aligned using Bionumerics (Applied Maths). Distances were calculated with the Kimura-2 correction, and a neighbor-joining tree was constructed with TreeCon.
The Fragmented Rhizobiaceae RmInt1 Intron
The results presented above suggest that RmInt1 tends to evolve into fragmented forms by truncation at its 3' terminus, resulting in the abolition of IEP activity and intron mobility. A previous report (Bittinger et al. 2000) indicated the presence of a 3'-truncated form of RmInt1 (91% identity) in the pa plasmid of R. etli strain CE3. This intron (R.et.F1) seems to have intron RNA dI to dIII and RT domains 0–2 (Dai and Zimmerly 2002a) and is flanked by IS elements first identified in S. meliloti. We found that, in addition to R.et.F1 (lower band in fig. 2B), strain CE3 has a second fragmented copy of RmInt1 (upper band in fig. 2B). The signal corresponding to this larger copy was also detected in other R. etli strains, including Viking 1 and CFN42, and in R. leguminosarum bv. phaseoli strains (fig. 2B). We characterized the corresponding band from R. etli strain Viking 1 (hereafter referred to as R.et.F2) further by DNA sequence analysis. We also determined the DNA sequence of the single-truncated form of RmInt1 present in the S. medicae RMO02 (S.md.F1) and S. adhaerens LMG20582(S.ad.F1) strains (fig. 1C). Total DNA was digested with EcoRI and PCR was performed with primers 1 and 18R.0. The 5' end of the intron and the 5' exon sequences of S.md.F1, R.et.F2, and S.ad.F1 were then obtained by inverse PCR (see Materials and Methods). Similarly, the 3' end of the intron and sequences further downstream were obtained for the S.md.F1 intron. Sequence analyses showed 90% similarity to RmInt1 from S. meliloti strain GR4 for R.et.F2 and 93% similarity for both S.md.F1 and S.ad.F1 (fig. 1C). R.et.F2 carries a deletion of 89 nt in the RT domain coding sequence. A blast nucleotide search of databases for R.et.F2 homologs revealed the presence in R. etli strain CFN42 of an almost identical (99% identity) fragmented intron copy harbored by plasmid p42d (González et al. 2003) (fig. 1C). This R.et.F2 copy has two deletions, one of 89 and one of 112 nt, in the RT domain–coding sequence. The larger of these two deletions is located at the position from which R.et.F1 is truncated. Surprisingly, an R.et.F1 copy was also detected in the p42a plasmid harbored by the CFN42 strain (V. González, personal communication). As the corresponding hybridization signal was not detected in our CFN42 strain, as shown in figure 2B, it has probably lost either the plasmid (p42a) or a fragment containing the truncated intron copy.
Phylogenetic Analysis of the Rhizobiaceae RmInt1 Ribozyme
The ribozyme sequences of all introns (full-length and fragmented) were aligned from nucleotide positions 180 to 550 (determined for all introns in this study), including part of RmInt1 ribozyme dI and extending to the beginning of dIV, before the IEP coding region. This alignment was used to generate a phylogenetic tree using neighbor-joining (fig. 3B) and maximum likelihood algorithms. The branching of this tree was similar to that obtained with the IEP, with two main groups identified (fig. 3B). Surprisingly, R.et.F1 and R.et.F2 clustered with the S. terangae RmInt1 copies (see fig. 3B and C). Comparison of the ribozyme domain nucleotide sequences showed identical changes in the S. terangae and R. etli RmInt1 introns (fig. 4A and B). These results suggest that the truncated copies of RmInt1 in R. etli and the full-length intron of S. terangae share a common origin.
FIG. 4.— Base substitutions in the secondary structure model of dIII (A) and dIV (B) of RmInt1. Nucleotides conserved between introns are denoted by (°), substituted nucleotides are shown in bold. Identical structures were obtained for R.et.F1 and R.et.F2, which are therefore both labeled as Rhizobium etli. In dIII (A), distal substitutions were mostly complementary, preserving the structure of the RNA. The ribosome-binding site (RBS) in dIV (B) is shown. The start and stop codons of the IEP are indicated by rectangles. Relevant observations are also indicated.
Analysis of the RmInt1 Flanking Regions
In S. meliloti, RmInt1 is mainly associated with ISRm2011-2 and, to a lesser extent, with related elements (ISRm10-1, ISRm10-2) of the ISRm2011-2 group within the IS630-Tc1/IS3 insertion sequence family (Martínez-Abarca and Toro 2000b). We investigated this association in species other than S. meliloti by carrying out Southern blot hybridization using an ISRm2011-2 3' end–derived probe. At least one (S. medicae) of the 29 species tested other than S. meliloti possessed this particular IS (table 2), which appears to be also associated with RmInt1, as suggested by hybridization of the same bands to both intron and IS probes (not shown). Analysis of the DNA sequences flanking the intron showed that the RmInt1 copy of R. etli strains Viking 1 and CFN42 (R.et.F2) was inserted into a 3'-truncated form of ISRm10-1; this was also the case in R.et.F1 (Bittinger et al. 2000 and this work). The RmInt1 copy of S. terangae ORS22 was found within a full copy of ISRm10-1. These results provide further support for a common origin of the S. terangae and R. etli RmInt1 introns.
The S. adhaerens 5D19 intron is inserted into a new IS of the ISRm2011-2 group, ISRm10-3, which is present as two copies in this bacterial genome. One of the copies is intron free and the insertion of RmInt1 may therefore correspond to a genuine retrohoming event. ISRm10-3 was absent from all the other strains tested, as indicated by DNA hybridization (table 2). These results confirm previous observations in S. meliloti of a specific association between RmInt1 and a particular class of IS elements favoring the spread of this intron in related taxa of the Rhizobiaceae. In contrast, the fragmented introns of S. medicae RMO02 (S.md.F1) and S. adhaerens LMG20582(S.ad.F1) seem be located in unknown ORFs or outside of coding regions as no homologs were found in databases (fig. 5). Analysis of the nucleotides essential for RmInt1 target recognition (Jiménez-Zurdo et al. 2003) in the 5' exon of the S.md.F1 intron showed that neither the T-15 critical position in the 5' distal exon nor the IBS2 sequence and critical A-3 position in IBS1 are conserved (fig. 5). Thus, the original insertion events for both copies of the RmInt1 intron were probably ectopic transposition events rather than cases of retrohoming.
FIG. 5.— The 5' exons of the various sequenced RmInt1 introns. The sequences of seven independent target sites for RmInt1 are shown and compared with the wild-type–homing site, ISRm2011-2, from Sinorhizobium meliloti. Exon-binding sites (EBS) 1 and 2 of the intron are aligned with the targets, which are shown according to the following letter code: bold uppercase letters, conserved residues that do not affect homing efficiency; normal uppercase letters, nonconserved residues that do not affect homing efficiency; normal lowercase letters, substitutions that block intron homing according to Jiménez-Zurdo et al. (2003). Five of the seven targets had sequences resembling insertion sequences.
Mobility of RmInt1 in Rhizobial Hosts
RmInt1 invades targets in a site-specific manner, by means of an RNA intermediate (retrohoming) (Martínez-Abarca et al. 2004). The spread of RmInt1 in the Rhizobiaceae seems to have occurred by both vertical inheritance and by horizontal gene transfer events. These horizontal transfers may be related to the mobility of intron-host DNA (e.g., plasmids, IS.) rather than the independent mobility of the intron. RmInt1 mobility has been demonstrated only in S. meliloti (Martínez-Abarca, García-Rodríguez, and Toro 2000). We investigated the potential of RmInt1 to move around in other related taxa by carrying out mobility assays in various hosts of the Rhizobiaceae lacking the intron. Mobility assays using the intron donor plasmid pKG2.5 and target recipient plasmid pJB0.6+ were performed in S. medicae, S. terangae, Rhizobium tropici, R. leguminosarum bv. phaseoli, Agrobacterium rhizogenes, and Agrobacterium tumefaciens. Efficient homing was observed in all the host strains tested (fig. 6). Remarkably, although RmInt1 has never been reported in strains of Rhizobium tropici, A. rhizogenes, and A. tumefaciens, this intron was clearly mobile within these hosts. These findings provide further evidence that RmInt1 may have colonized the genome of species other than S. meliloti by independent mobility of the intron facilitated by the presence of host IS elements of the ISRm2011-2 group.
FIG. 6.— RmInt1 mobility assay in heterologous hosts. The homing assay was carried out in selected rhizobia with no copy of RmInt1 on the genome. Plasmid pools were analyzed by Southern blot hybridization with a 3'-end probe for ISRm2011-2. Strains with the donor plasmid pKG2.5 were transformed with receptor plasmid pJB0.6+ (odd lanes) or with a receptor plasmid without target site pJB129 (even lanes). Fragments detected by hybridization are indicated as follows: D, donor plasmid; R, recipient; H, homing product at the right of the gel, and molecular mass markers are indicated on the left. The following strains were used: Sinorhizobium meliloti GR4, lanes 1 and 2; Sinorhizobium medicae RMO15, lanes 3 and 4; Sinorhizobium terangae ORS19, lanes 5 and 6; Rhizobium leguminosarum bv. phaseoli Ro33, lanes 7 and 8; Rhizobium tropici B BR850, lanes 9 and 10; Agrobacterium rhizogenes 163c1, lanes 11 and 12; and Agrobacterium tumefaciens B6, lanes 13 and 14.
Discussion
The group II intron RmInt1 was first described in S. meliloti following analysis of the sequence of pRmeGR4b from GR4, a strain that contains nine copies of this element (Martínez-Abarca, Zekri, and Toro 1998). This intron was mostly found within copies of ISRm2011-2 (Martínez-Abarca, Zekri, and Toro 1998) and closely related IS elements (Martínez-Abarca and Toro 2000b). We show here that RmInt1 is also present in other Sinorhizobium and Rhizobium species. RmInt1 is present predominantly in the genus Sinorhizobium, where vertical inheritance seems to occur. However, our results also suggest that the RmInt1 intron has crossed the genus barrier and can independently colonize other bacterial genomes, including those of other rhizobia and Agrobacterium. We also present data indicating that RmInt1 tends to evolve into inactive forms by fragmentation, losing its 3' terminus.
RmInt1 is present in 90% of S. meliloti isolates (Mu?oz, Villadas, and Toro 2001) and is an active and efficient retroelement in this bacterial host. In this study, we also found copies of RmInt1 (full-length and fragmented copies) in other Sinorhizobium species (S. medicae, S. adhaerens, and S. terangae), whereas only fragmented copies were present in Rhizobium species, such as R. etli and R. leguminosarum bv. phaseoli. Both these species nodulate and fix nitrogen in a similar ecological niche, the root nodules elicited on the leguminous plant Phaseolus vulgaris. Sequence data and phylogenetic analysis suggest that RmInt1 is native to the genus Sinorhizobium, in which it appears to have been inherited vertically from a common ancestor or by horizontal transfer among closely related strains and species but possibly lost in various species. Our results also suggest that the RmInt1 intron copies in R. etli and R. leguminosarum bv. phaseoli share a common origin with the S. terangae RmInt1 intron. RmInt1 may have been present in an ancestor of the Rhizobiaceae family but are lost in most of the lineages giving rise to the various genera. Alternatively, the RmInt1 intron of R. etli and R. leguminosarum bv. phaseoli may have been acquired by lateral transfer from Sinorhizobium species. The vertical inheritance hypothesis seems to be the least plausible because it involves evolutionary convergence of the Rhizobium and S. terangae RmInt1 introns. Sinorhizobium terangae and other Sinorhizobium species have full-length introns, whereas only fragmented introns that had accumulated larger numbers of mutations were found in Rhizobium. However, most of the R. etli and R. leguminosarum bv phaseoli strains isolated from different geographical origins harbored truncated RmInt1 copies. An early transfer event before the divergence of these two rhizobial species may account for these findings.
RmInt1 is currently very successful in S. meliloti because of the presence of host mobile DNA of the ISRm2011-2 group, which provides a site-specific target and facilitates the survival and spread of the intron. Studies on natural field populations of S. meliloti have shown that the RmInt1 intron spread not only by transposition of the host mobile DNA but also by independent movement of the intron (Mu?oz, Villadas, and Toro 2001). Similar conclusions have been drawn concerning the independent mobility of group II introns in E. coli populations, in studies of group II introns in the ECOR collection (Dai and Zimmerly 2002b), and experimental demonstrations of conjugation-mediated transfer of the LltrB group II intron between bacterial species (Belhocine, Plante, and Cousineau 2004). RmInt1 spreads by retrohoming or ectopic transposition, a conclusion also supported by this work. An analysis of flanking exon sequences suggested that in some cases intron insertion may have involved an ectopic transposition event, and the mobility assay data indicated that RmInt1 displays retrohoming in bacterial species and genera other than S. meliloti.
It has been reported that bacterial group II introns degenerate mainly by fragmentation (Dai and Zimmerly 2002a). It has been reported that 38% of the described bacterial intron fragments are 5' truncated (Dai and Zimmerly 2002a), possibly due to incomplete reverse transcription after intron RNA insertion. Interestingly, we found that RmInt1 tended to degenerate by means of truncation at its 3' terminus. In at least one case (R.et.F1), fragmentation may have occurred by insertion of an IS (ISRm3) with further rearrangements and deletion. In other cases (R.et.F2), the intron displayed internal deletions. Similar data have been reported for other bacterial group II introns, such as E. coli intron E.c.I4 (Dai and Zimmerly 2002b), and this may be a general mechanism for the generation of 3'-truncated fragmented introns.
Our results provide further evidence that bacterial group II introns are able to propagate by vertical and lateral transfer, retrohoming, or ectopic transposition and show that they tend to evolve into fragmented introns, mechanisms underlying the origin of eukaryotic spliceosomal introns.
Acknowledgements
The authors thank all the donors of strains and isolates for their invaluable generosity. This work was supported as part of research projects BIO99-0905 and BIO2002-02579 by the Ministerio de Ciencia y Tecnología. E.M.-A. was funded by Junta de Andalucía. M.F.-L. received postdoctoral grants from the Ministerio de Ciencia y Tecnología. A.W. would like to thank the Fund for Scientific Research–Flanders for her postdoctoral research fellowship. The R. etlim CFN42 was provided by Dr. V. González.
References
Belfort, M., V. Derbyshire, M. M. Parker, B. Cousineau, and A. M. Lambowitz. 2002. Mobile introns: pathways and proteins. Pp. 761–783 in N. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds. Mobile DNA. 2nd edition. ASM Press, Washington D.C.
Belhocine, K., I. Plante, and B. Cousineau. 2004. Conjugation mediates transfer of the Ll. LtrB group II intron between different bacterial species. Mol. Microbiol. 51:1459–1469.
Bittinger, M. A., J. A. Gross, J. Widom, J. Clardy, and J. Handelsman. 2000. Rhizobium etli CE3·carries vir genes homologs on a self-transmissible plasmid. Mol. Plant Microbe Interact. 13:1019–1021.
Dai, L., and S. Zimmerly. 2002a. Compilation and analysis of group II intron insertions in bacterial genomes: evidence for retroelement behaviour. Nucleic Acids Res. 30:1091–1102.
———. 2002b. The dispersal of five group II introns among natural populations of Escherichia coli. RNA 8:1294–1307.
———. 2003. ORF-less and RT-encoding group II introns in archaebacteria, with a pattern of homing into related group II intron ORFs. RNA 9:14–19.
Dickson, L., H. R. Huang, L. Liu, M. Matsuura, A. M. Lambowitz, and P. S. Perlman. 2001. Retrotransposition of a yeast group II intron occurs by reverse splicing directly into ectopic DNA sites. Proc. Natl. Acad. Sci. USA 98:13207–13212.
Ditta, G., S. Stanfield, D. Corbin, and D. R. Helinski. 1980. Broad host range DNA cloning system from gram-negative bacteria. Construction of a gene bank of Rhizobium meliloti. Proc. Natl. Acad. Sci. USA 77:7347–7351.
Eickbush, T. H. 1999. Mobile introns: retrohoming by complete reverse splicing. Curr. Biol. 9:11–14.
Felsenstein, J. 1995. PHYLIP (phylogeny inference package). Version 3.55c. Distributed by the author, Department of Genetics, University of Washington, Seattle. (http://evolution.genetics.washington.edu/phylip/getme.html).
Ferat, J. L., M. Le Gouar, and F. Michel. 1994. Multiple group II self-splicing introns in mobile DNA from Escherichia coli. C. R. Acad. Sci. Ser III Life Sci. 317:141–148.
Ferat, J. L.,. 2003. A group II intron has invaded the genus Azotobacter and is inserted within the termination codon of essential groEL gene. Mol. Microbiol. 49:1407–1423.
Ferat, J. L., and F. Michel. 1993. Group II self-splicing introns in bacteria. Nature 364:358–361.
Galibert, F., T. M. Finan, S. R. Long et al. (56 co-authors). 2001. The composite genome of the legume symbiont Sinorhizobium meliloti. Science 293:668–672.
Geremia, R. A., P. Mergaert, D. Geelen, M. Van Montagu, and M. Holsters. 1994. The NodC protein of Azorhizobium caulinodans is an N-acetylglucosaminyltransferase. Proc. Natl. Acad. Sci. USA 91:2669–2673.
González, V., P. Bustos, M. A. Ramírez-Romero et al. (16 co-authors). 2003. The mosaic structure of the symbiotic plasmid of Rhizobium etli CFN42 and its relation to other symbiotic genome compartments. Genome Biol. 4:R36.
Granlund, M., F. Michel, and M. Norgren. 2001. Mutually exclusive distribution of IS1548 and GBSi1, an active group II intron identified in human isolates of group B streptococci. J. Bacteriol. 183:2560–2569.
Ichiyanagi, K., A. Beauregard, and M. Belfort. 2003. A bacterial group II intron favors retrotransposition into plasmid targets. Proc. Natl. Acad. Sci. USA 100:15742–15747.
Ichiyanagi, K., A. Beauregard, S. Lawrence, D. Smith, B. Cousineau, and M. Belfort. 2002. Multiple pathways for retrotransposition of the Ll.ltrB group II intron include a potential relationship to DNA replication. Mol. Microbiol. 46:1259–1272.
Jiménez-Zurdo, J. I., F. M. García-Rodríguez, A. Barrientos-Durán, and N. Toro. 2003. DNA target site requirements for homing in vivo of a bacterial group II intron encoding a protein lacking the DNA endonuclease domain. J. Mol. Biol. 326:413–423.
Lambowitz, A. M., M. G. Caprara, S. Zimmerly, and P. S. Perlman. 1999. Group I and group II ribozymes as RNPs: clues to the past and guides to the future. Pp. 451–485 in T. R. Cech, and J. F. Atkins, eds. The RNA world. 2nd edition. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Martínez-Abarca, F., A. Barrientos-Durán, M. Fernández-López, and N. Toro. 2004. The RmInt1 group II intron has two different pathways for mobility using predominantly the nascent lagging strand at DNA replication forks for priming. Nucleic Acids Res. 32:2880–2888.
Martínez-Abarca, F., F. M. García-Rodríguez, and N. Toro. 2000. Homing of a bacterial group II intron with an intron-encoded protein lacking a recognizable endonuclease domain. Mol. Microbiol. 35:1405–1412.
Martínez-Abarca, F., and N. Toro. 2000a. Group II introns in the bacterial world. Mol. Microbiol. 38:917–926.
———. 2000b. RecA-independent ectopic transposition in vivo of a bacterial group II intron. Nucleic Acids Res. 28:4397–4402.
Martínez-Abarca, F., S. Zekri, and N. Toro. 1998. Characterization and splicing in vivo of a Sinorhizobium meliloti group II intron associated with particular insertion sequences of the IS630-Tc1/IS3·retroposon superfamily. Mol. Microbiol. 28:1295–1306.
Michel, F., and J. L. Ferat. 1995. Structure and activities of group II introns. Annu. Rev. Biochem. 64:435–461.
Michel, F., K. Umesono, and H. Ozeki. 1989. Comparative and functional anatomy of group II catalytic introns—a review. Gene 82:5–30.
Mu?oz, E., P. J. Villadas, and N. Toro. 2001. Ectopic transposition of a group II intron in natural bacterial populations. Mol. Microbiol. 41:645–652.
Mu?oz-Adelantado, E., J. San Filippo, F. Martínez-Abarca, F. M. García-Rodríguez, A. M. Lambowitz, and N. Toro. 2003. Mobility of the Sinorhizobium meliloti group II intron RmInt1 occurs by reverse splicing into DNA, but requires an unknown reverse transcriptase priming mechanism. J. Mol. Biol. 327:931–943.
Nakamura, Y., T. Kaneko, S. Sato et al. (21 co-authors). 2002. Complete genome structure of the thermophilic cyanobacterium Thermosynechococcus elongatus BP-1. DNA Res. 9:123–130.
Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
San Filippo, J., and A. M. Lambowitz. 2002. Characterization of the C-terminal DNA-binding/DNA-endonuclease region of a group II intron-encoded reverse transcriptase. J. Mol. Biol. 324:933–951.
Sharp, P. A. 1991. Five easy pieces. Science 254:663.
Sheveleva, E. V., and R. B. Hallick. 2004. Recent horizontal intron transfer to a chloroplast genome. Nucleic Acids Res. 32:803–810.
Toro, N. 2003. Bacteria and archaea group II introns: additional mobile genetic elements in the environment. Environ. Microbiol. 5:143–151.
Toor, N., G. Hausner, and S. Zimmerly. 2001. Coevolution of group II intron RNA structures with their intron-encoded reverse transcriptase. RNA 7:1142–1152.
Toro, N., F. Martínez-Abarca, M. Fernández-López, and E. Mu?oz-Adelantado. 2003. Diversity of group II introns in the genome of Sinorhizobium meliloti strain 1021: splicing and mobility of RmInt1. Mol. Genet. Genomics 268:628–636.
Toro, N., M. D. Molina-Sánchez, and M. Fernández-López. 2002. Identification and characterization of bacterial class E group II introns. Gene 299:245–250.
Van de Peer, Y., and R. De Wachter. 1994. TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Comput. Appl. Biosci. 10:569–570.
Vincent, J. M. 1970. A manual for the practical study of the root-nodule bacteria. IBP handbook No. 15. Blackwell Scientific, Oxford.
Yang, J., G. Mohr, P. S. Perlman, and A. M. Lambowitz. 1998. Group II intron mobility in yeast mitochondria: target DNA-primed reverse transcription activity of al1 and reverse splicing into DNA transposition sites in vitro. J. Mol. Biol. 282:505–523.
Zimmerly, S., G. Hausner, and X. Wu. 2001. Phylogenetic relationships among group II intron ORFs. Nucleic Acids Res. 29:1238–1250.(Manuel Fernández-López*, )
Correspondence: E-mail: ntoro@eez.csic.es.
Abstract
Group II introns are both self-splicing RNAs and mobile retroelements found in bacterial and archaeal genomes and in organelles of eukaryotes. They are thought to be the ancestors of eukaryote spliceosomal introns and non–long terminal repeat retrotransposons. We show here that RmInt1, a bacterial group II intron first described in the nitrogen-fixing symbiont of alfalfa (Medicago sativa) Sinorhizobium meliloti, is also present in other Sinorhizobium and Rhizobium species. The intron-homing sites in these species are IS elements of the ISRm2011-2 group as in S. meliloti, but ectopic insertion is also observed. We present evidence that these related bacteria have acquired RmInt1 by vertical inheritance from a common ancestor and by independent horizontal transfer events. We also show that RmInt1 is mobile in related taxa of bacteria that interact with plants and tends to evolve toward an inactive form by fragmentation, with loss of the 3' terminus including the intron-encoded protein. Our results provide an overview of the evolution and dispersion of a bacterial group II intron.
Key Words: group II intron ? RmInt1 ? nitrogen fixation ? legumes ? Sinorhizobium meliloti ? Rhizobiales ? ribozyme ? reverse transcriptase
Introduction
Group II introns are self-splicing, mobile retroelements that splice via a lariat intermediate in a mechanism similar to that of spliceosomal introns (Michel and Ferat 1995). These genetic elements were first found in mitochondrial and chloroplast genomes in lower eukaryotes and plants (Michel, Umesono, and Ozeki 1989) and were later identified in eubacteria (Ferat and Michel 1993; Martínez-Abarca, Zekri, and Toro 1998; Martínez-Abarca and Toro 2000a; Zimmerly, Hausner, and Wu 2001; Dai and Zimmerly 2002a) and some archaebacterial species (Dai and Zimmerly 2003; Toro 2003). Group II introns are large catalytic RNAs displaying a conserved secondary structure with six double-helical domains (dI to dVI), one of which (dIV) may incorporate a reverse transcriptase (RT) open-reading frame (ORF) (Michel and Ferat 1995). Unlike organellar introns, most bacterial group II introns possess an intron-encoded protein (IEP; Dai and Zimmerly 2002a). Nevertheless, ORF-less group II introns have been found, e.g., in archaeabacteria (Dai and Zimmerly 2003) and the cyanobacterium, Thermosynechococcus elongatus BP-1 (Nakamura et al. 2002). Three main phylogenetic classes (IIA, IIB, and IIC) of group II introns have been described based on IEP and conserved intron RNA structures (Michel, Umesono, and Ozeki 1989; Toor, Hausner, and Zimmerly 2001; Zimmerly, Hausner, and Wu 2001; Toro, Molina-Sánchez, and Fernández-López 2002; Ferat, Le Gouar, and Michel 2003; Toro 2003). Interestingly, most bacterial introns are not located in conserved genes and most are associated with mobile DNAs (Ferat, Le Gouar, and Michel 1994; Martínez-Abarca and Toro 2000a; Granlund, Michel, and Norgren 2001; Dai and Zimmerly 2002a, 2003; Toro 2003). Over half are fragmented (Dai and Zimmerly 2002a). The IEPs of these introns have four conserved domains, including an N-terminal RT domain, domain X, a putative RNA-binding domain associated with RNA splicing or maturase activity, C-terminal DNA-binding (D) domain, and a DNA-endonuclease (En) domain. Maturase activity is required for the splicing reaction in vivo, resulting in an excised intron RNA lariat, which forms a ribonucleoprotein complex with the IEP that is involved in the mobility of the intron to intron-less alleles in a process known as retrohoming, which requires both RT and En activities (Lambowitz et al. 1999; Belfort et al. 2002). Group II introns also move (albeit at low frequency) into ectopic sites resembling the normal homing site, in a process known as retrotransposition (Yang et al. 1998; Martínez-Abarca and Toro 2000b; Dickson et al. 2001; Mu?oz, Villadas, and Toro 2001; Ichiyanagi et al. 2002; Ichiyanagi, Beauregard, and Belfort 2003), which is thought to facilitate intron dispersal.
Nuclear pre–messenger RNA introns (Michel and Ferat 1995) and non–long terminal repeat retrotransposons are both thought to descend from mobile group II introns (Eickbush 1999). According to this hypothesis, group II introns originated in bacteria and invaded the nucleus of a primitive eukaryote directly or from an organelle; they were then fragmented to form the spliceosome (Sharp 1991). The horizontal transfer of a self-splicing group II intron from a cyanobacterium to the chloroplast genome of Euglena myxocylindracea was recently shown (Sheveleva and Hallick 2004), suggesting that this process is still occurring in nature. In bacteria, the evidence for horizontal gene transfer events involving group II introns is fragmented and indirect, being based mostly on the finding that a particular class of intron is present in various host organisms and that a particular organism harbors various classes of intron. The Azotobacter vinelandii group II intron (Avi.groEL) provides an illustration of this. This intron has been inserted into the termination codon of the groEL coding sequence, and this insertion site is conserved in at least one other species of Azotobacter (Azotobacter chroococcum), suggesting intron transposition after divergence of the two taxa (Ferat, Le Gouar, and Michel 2003). Another example is the group II introns found in Methanosarcina acetivorans and Methanosarcina mazei, which were probably acquired from eubacteria (Dai and Zimmerly 2003; Toro 2003).
RmInt1 was the first group II intron described in the order Rhizobiales. It was identified in Sinorhizobium meliloti, the nitrogen-fixing symbiont of alfalfa (Medicago sativa). The RmInt1 IEP, like those of many other bacterial group II introns, lacks the DNA-En and a cognate C-terminal DNA-binding domain (Martínez-Abarca, García-Rodríguez, and Toro 2000; Zimmerly, Hausner, and Wu 2001; Dai and Zimmerly 2002a; San Filippo and Lambowitz 2002; Toro 2003). RmInt1 is nevertheless an efficient mobile element with a homing frequency (proportion of cells containing at least one homing event) approaching 100%, similar to that of fungal mitochondrial DNA introns and the lactococcal Ll.ltrB intron, inserting into 20%–45% of recipient target plasmids (Martínez-Abarca, García-Rodríguez, and Toro 2000; Jiménez-Zurdo et al. 2003). This mobile intron also transposes to ectopic sites (Martínez-Abarca and Toro 2000b; Mu?oz, Villadas, and Toro 2001). RmInt1, like other group II introns lacking the IEP En domain, may use a nascent lagging strand at DNA replication forks for priming (Martínez-Abarca et al. 2004).
This study aimed to investigate the further spread of bacterial group II introns in nature and to trace their possible evolution and involvement in horizontal transfer events. We searched for the S. meliloti RmInt1 intron and investigated its distribution in related taxa of bacteria that interact with plants. We present evidence that Sinorhizobium species inherited RmInt1 vertically from a common ancestor within the order Rhizobiales, with subsequent gains and losses in the various species and that this intron has been transferred horizontally to some Rhizobium species. We show that intron moves efficiently to homing sites in related taxa and that it tends to be inactivated by fragmentation with loss of the 3' terminus. This study sheds light on the dispersal and evolution of bacterial group II introns in natural populations.
Materials and Methods
Bacterial Strains, Media, and Growth Conditions
The bacterial species of the order Rhizobiales and the S. meliloti isolates used are listed in tables 1 and 3. Escherichia coli DH5 was routinely cultured at 37°C in Luria-Bertani medium (Sambrook, Fritsch, and Maniatis 1989), and rhizobial bacteria were grown at 28°C on tryptone yeast extract (TY) (Agrobacterium, Mesorhizobium, Rhizobium, and Sinorhizobium), yeast extract mannitol (YEM) (Bradyrhizobium) (Vincent 1970), or yeast extract bactobeet (YEB) (Azorhizobium and Xanthobacter) (Geremia et al. 1994). Ampicillin was added to the medium at a concentration of 200 μg/ml for E. coli.
Table 1 Bacterial Strains Used in this Study
Table 3 Number of Copies of RmInt1 and ISRm2011-2 in Sinorhizobium meliloti Isolatesa
DNA Hybridization and Fingerprinting
Total DNA was isolated according to standard protocols (Sambrook, Fritsch, and Maniatis 1989). DNA (2 μg) was digested with EcoRI or BamHI and subjected to electrophoresis in 0.8% Tris-acetate agarose gels. It was then blotted onto positively charged nylon membranes (Roche Diagnostic, Mannhein, Germany) according to the manufacturer's instructions. The DNA probes used for ISRm2011-2 and RmInt1 and the method used for DNA hybridization under high stringency conditions have been described elsewhere (Martínez-Abarca and Toro 2000b). The probe for ISRm10-3 was obtained with primers 5D1 (5'-ACGTGTTCTTGCCGCCGT-3') and 5D2 (5'-TTGCGCGTCTGATTGTGC-3') with a similar procedure.
DNA Sequencing and Inverse Polymerase Chain Reaction
Copies of RmInt1 were isolated from Sinorhizobium medicae DNA by BamHI digestion, electrophoresis in agarose gels, excision with the Perfectprep Gel Cleanup kit (Eppendorf AG, Hamburg, Germany), and hybridization with intron-derived probes (fig. 1A). DNA from the excised bands and DNA from strains with a single copy of RmInt1 were subjected to polymerase chain reaction (PCR) with primers GII.1 (5'-AAIAGICITGGTIGTGAGCG-3') and GII.4 (5'-TCTCGCAGAACIGTICGTGA-3') to amplify RmInt1. The resulting PCR fragments were inserted into the pGEM-T vector (Promega, Madison, Wis.) and sequenced with primers T7, SP6, 1, Int2, H3 (5'-GTATTGTTTGAAACAACTG-3'), and EB70 (5'-ATGGTGGTCAAGCAGATGA-3') (fig. 1). The internal sequence of fragmented introns from strains with a single copy was obtained by PCR with primers 1 and 18R.0, purification of the fragment with S300HR columns (Pharmacia, Little Chalfont, UK), and sequencing in an automatic laser fluorescent DNA sequencer using the same primers.
FIG. 1.— Schematic representation of the Sinorhizobium meliloti RmInt1 intron and the copies sequenced in other related taxa. (A) The mobile RmInt1 intron of S. meliloti strain GR4. The diagram shows ribozyme dI to dVI. The ORF is within intron dIV. Protein domains are RT and maturase (X), with a C-terminal extension indicated by an asterisk (San Filippo and Lambowitz 2002; Mu?oz-Adelantado et al. 2003). The primers used for PCR amplification and sequence analysis are indicated by arrows and their relative nucleotide positions within the intron sequence are shown in parentheses. The 5' intron probe and the 3' intron ORF-probe used for DNA hybridization are also indicated. (B) Full-length RmInt1 intron copies in various hosts from the Rhizobiaceae family. Diagrams are constructed as described above. Numbers at the beginning and at the end show the positions of the determined sequences with respect to RmInt1. Relevant features such as stop codons and deletions are also indicated. Dotted lines indicate DNA sequences that were not determined. Percentage identity to the sequence of S. meliloti RmInt1 is shown on the right. The accession numbers of the sequences are AY248839 for Sinorhizobium adhaerens 5D19, AY608908 [GenBank] for Sinorhizobium terangae ORS22, AY608905 for S. terangae ORS1009, AY608907 for Sinorhizobium medicae RMO09-1, and AY608906 for S. medicae RMO09-2. (C) Fragmented RmInt1 intron copies in various hosts from the Rhizobiaceae family. Diagrams are constructed as described above. Continuous lines indicate a lack of similarity of the sequence to sequences in the database. The abbreviated intron designation is indicated below each strain. The accession numbers of the sequences are AY608902 for Rhizobium etli Viking 1, AF176227 for R. etli CE 3, AY608903 for S. medicae RMO02, and AY608901 for S. adhaerens LMG20582
The 5' and 3' exon junctions of insertion sites were identified by inverse PCR with divergently annealing primers. Total DNA from Rhizobium etli Viking 1, Sinorhizobium terangae ORS22, and Sinorhizobium adhaerens 5D19 was first digested with EcoRI; DNA from S. medicae RMO02 was digested with BamHI; and DNA from S. adhaerens LMG20582was digested with EcoRV. The exon 3' to complete introns was amplified with primers PR1000 (5'-GCGGAAGATTGTCAAACAGC-3') and ICF (5'-CTGTTCTCTCTGGCTGACTACG-3'), whereas the 5' exon and the exons from truncated introns were amplified with primers ICI (5'-AGGATGACGAAACGGTCCT-3') and H3. The products of the amplification reactions were sequenced directly as described above. DNA sequence editing, translation and analysis, and sequence similarity searches were carried out as previously described (Martínez-Abarca and Toro 2000b).
Phylogenetic Analysis
Analyses were carried out with PHYLIP version 3.573c, using the programs Seqboot, DNAdist, Protdist, Neighbor, Protpars, DNAML, and Consense (Felsenstein 1995). Sequences were aligned by ClustalW (http://bioweb.pasteur.fr). Reference 16S ribosomal DNA (rDNA) sequences were retrieved from European Molecular Biology Laboratory and aligned using Bionumerics (Applied Math); a distance matrix was obtained with Kimura-2 correction, and a neighbor-joining tree was generated with TreeCon (version 1.3b, Van de Peer and De Wachter 1994). A bootstrap analysis of 500 replicate data sets was performed with the same program.
RmInt1 Mobility Assays
Donor (pKG2.5) and receptor pJB0.6s(+) plasmids were transferred successively from E. coli to the selected rhizobial strains by triparental mating, using the helper plasmid pRK2013 (Ditta et al. 1980). Mobility of intron copies from the genome of S. terangae ORS22 and S. adhaerens 5D19 to the receptor plasmids pJB0.6s(+) and pJB0.6as (Martínez-Abarca et al. 2004) was also tested. Homing events were analyzed as previously described (Martínez-Abarca, García-Rodríguez, and Toro 2000; Martínez-Abarca et al. 2004).
Results
Presence and Distribution of RmInt1 in the Order Rhizobiales
We evaluated the presence and distribution of RmInt1 in the order Rhizobiales. We carried out Southern blotting for 172 isolates corresponding to 29 species, 7 genera, and 4 families of bacteria, all able to interact with plants (table 1). Two different RmInt1-derived DNA probes—a 448-bp 5' intron probe and a 238-bp 3' intron ORF-probe (see fig. 1)—were used in hybridization analysis to make it possible to distinguish between full-length and fragmented introns. We focused exclusively on strong hybridization signals corresponding to sequences displaying >85% sequence identity to RmInt1 at the DNA level, as shown by subsequent analyses (see below). Within the Rhizobiaceae, our hybridization data indicated that RmInt1 was present in the genus Sinorhizobium, in S. meliloti, S. medicae, S. adhaerens, and S. terangae. It was not detected in Sinorhizobium fredii, Sinorhizobium saheli, or in the reference strains of Sinorhizobium kostiense, Sinorhizobium arboris, Sinorhizobium xinjiangensis, and Sinorhizobium morelense. In the other branches of the Rhizobiaceae tested, this intron was detected only in R. etli and Rhizobium leguminosarum bv. phaseoli (see tables 1 and 2, fig. 2A and B). No band hybridizing with the RmInt1 probes was detected in bacteria of Phyllobacteriaceae, Bradyrhizobiaceae, or Hyphomicrobiaceae, although some of the bacteria of these families (e.g., Bradyrhizobium japonicum) harbor phylogenetically related introns (Dai and Zimmerly 2002a; Toro 2003; Toro et al. 2003).
Table 2 Number of Copies of RmInt1, ISRm2011-2, and ISRm10-3 in Species of the Order Rhizobiales Carrying the Group II Intron
FIG. 2.— Presence of RmInt1 in Sinorhizobium medicae, Rhizobium etli, and Rhizobium leguminosarum bv. phaseoli. Total DNA from the strains was digested with EcoRI (R. etli and R. leguminosarum) or BamHI (S. medicae), separated on a 0.8% agarose gel, and transferred onto a nylon filter. (A) Sinorhizobium medicae RMO09 DNA was first hybridized with a 3' intron ORF-probe, then washed, and hybridized with a 5' intron probe. Asterisk indicate the excised bands used for sequencing. (B) Rhizobium etli (CE3, CFN42, and Viking 1) and R. leguminosarum bv. phaseoli (PhP17, Ro33, and Ro34) DNA hybridized with the 5' intron probe. The sizes of the molecular weight markers (in kilobits) are indicated on the left.
All the strains and field isolates of R. etli and R. leguminosarum bv. phaseoli harboring copies of RmInt1 generated positive hybridization signals with the 5' intron probe but not with the 3' intron ORF-probe. This implies that RmInt1 has lost the IEP or is generally truncated at the 3' terminus in these bacterial species (table 2). In Sinorhizobium, fragmented RmInt1 introns were also detected in S. medicae, S. adhaerens, and S. meliloti, as shown by the smaller number of hybridizing bands obtained with the RmInt1 3' intron ORF-probe than with the 5' intron probe (table 2). After S. meliloti, RmInt1 appears to be most frequent in the related species S. medicae (seven of eight strains tested), but most of the copies detected appeared to be truncated at the 3' terminus (table 2). A larger study on 19 different S. meliloti isolates (table 3) suggested that RmInt1 is present predominantly in its full-length form in this bacterial species, confirming genome sequence data for strain 1021 showing the presence of three full-length copies of RmInt1 and no fragmented copies of this intron (Galibert et al. 2001; Toro et al. 2003). Nevertheless, there is considerable diversity among S. meliloti strains in terms of the number of bands detected with the two intron-derived probes, with all strains except 5D47 having more bands hybridizing to the 5' intron probe than to the 3' intron ORF-probe, consistent with RmInt1 fragmentation occurring predominantly via truncation at the 3' terminus.
The Full-Length Rhizobiaceae RmInt1 Intron
We also studied strains from species other than S. meliloti, with the aim of characterizing the full-length copies of the RmInt1 intron and obtaining the corresponding DNA sequence. Total DNA from strains harboring more than one copy of the intron was digested with BamHI, a restriction enzyme that does not cut within the intron, and hybridized with the intron-derived probes. Southern blots suggested that S. medicae RMO09 harbored at least three full-length copies of the intron (fig. 2A). Sinorhizobium terangae strains ORS22 and ORS1009 and the S. adhaerens strain 5D19 gave single hybridizing bands of different sizes with the intron-derived probes, suggesting that these strains contained a single full-length copy of RmInt1 (data not shown). DNA corresponding to the 20- and 7.5-kb hybridizing bands from S. medicae (fig. 2A) were excised and purified from an agarose gel and subjected to PCR with the intron-derived primers GII.1 and GII.4 (fig. 1A). We also performed PCR with the same primers plus total DNA from S. terangae strains ORS22 and ORS1009 and sequenced S. adhaerens DNA with the intron-derived primers to obtain overlapping sequences (see fig. 1A). Finally, a partial DNA sequence (1,604–1,717 nt) was obtained for each RmInt1 intron copy, including the complete dIV sequence (fig. 1B). The full-length intron sequence (1,884 nt) and exon sequences were obtained for the S. adhaerens 5D19 and S. terangae ORS22 RmInt1 elements by inverse PCR (fig. 1B). DNA sequence data showed that the S. medicae RmInt1 element located in the 20-kb BamHI fragment (hereafter referred to as copy 1) and that within the 7.5-kb BamHI fragment (copy 2) were 99% identical to S. meliloti RmInt1 (strain GR4). Identities of 93% were obtained for the intron in S. adhaerens and of 89% and 85% for the RmInt1 elements in S. terangae strains ORS22 and ORS1009, respectively. Most of the nucleotide changes were silent and those resulting in amino acid changes in the IEP were mostly clustered in the interdomains or in the spacers within the RT domain. Within the ribozyme, most of the observed nucleotide changes were complementary and therefore did not alter the predicted secondary structure of the ribozyme. Nevertheless, it should be noted that S. medicae RmInt1 copy 1 has four nucleotide changes and lacks a G at position 792 that results in a frameshift mutation, producing a stop codon at position 842, resulting in a truncated IEP of 98 amino acids (fig. 1B), whereas copy 2 has only one nucleotide change. The intron in S. terangae ORS1009 has a truncated IEP of 325 amino acids due to a stop codon generated in the maturase domain at position 1540 and a deletion of 18 nt that results in the loss of six amino acid residues at the start of the RT 0 domain (fig. 1B). Thus, at least some of the full-length RmInt1 introns appear to have become inactive in their hosts. Moreover, mobility assays using target recipient plasmids did not show homing events for the S. terangae ORS22 and S. adhaerens 5D19 introns (data not shown).
Phylogenetic Analyses of the Rhizobiaceae RmInt1 IEP
We investigated the evolutionary relationships between the RmInt1 introns found in the various hosts from the Rhizobiaceae by aligning the amino acid sequences of RmInt1 IEP copies obtained and correcting sequences to obtain a full IEP when required. Phylogenetic trees were constructed by neighbor-joining (fig. 3A) and parsimony analysis. Similar branching patterns were obtained with these algorithms. Phylogenetic analysis identified two main groups well supported by bootstrapping analysis: one including the RmInt1 IEPs from S. meliloti, S. medicae, and S. adhaerens, (hereafter referred to as the S. meliloti group) and the second containing the two RmInt1 IEPs found in S. terangae strains (hereafter referred to as the S. terangae group). The RmInt1 IEP–based phylogenetic tree was consistent with that based on 16S rDNA sequences (fig. 3C). Thus, the RmInt1 intron in these Sinorhizobium species may have been vertically inherited from a common ancestor, with intron loss in several species, or by horizontal transfer among closely related strains and species.
FIG. 3.— Phylogeny of the RmInt1 group II intron. Phylogenetic analyses (A) and (B) were carried out, after ClustalW alignment, with the Neighbor, parsimony, and maximum likelihood modules of the PHYLIP package (using default settings). In both diagrams, a consensus neighbor-joining phylogram is shown, but similar results were obtained with all the methods that were used. Bootstrap values are expressed as percentages and were derived from 1,000 samplings, and those exceeding 75 are shown at the branch points. Maximum parsimony (A) and maximum likelihood (B) values are also shown in italics and parentheses. Escherichia coli IntB group II intron (X77508 [GenBank] ) was used as the out-group. (A) Phylogeny of the IEP based on the alignment of 419 amino acid residues. (B) Phylogeny of the 5' end of the ribozyme based on alignment of the 370 nt from position 180 to 550 of the RmInt1 sequence, including part of dI and extending to the start codon of the protein encoded by dIV. (C) The 16S rDNA phylogeny of the main rhizobial groups. The sequences analyzed for tree construction were obtained from the European Molecular Biology Laboratory database and aligned using Bionumerics (Applied Maths). Distances were calculated with the Kimura-2 correction, and a neighbor-joining tree was constructed with TreeCon.
The Fragmented Rhizobiaceae RmInt1 Intron
The results presented above suggest that RmInt1 tends to evolve into fragmented forms by truncation at its 3' terminus, resulting in the abolition of IEP activity and intron mobility. A previous report (Bittinger et al. 2000) indicated the presence of a 3'-truncated form of RmInt1 (91% identity) in the pa plasmid of R. etli strain CE3. This intron (R.et.F1) seems to have intron RNA dI to dIII and RT domains 0–2 (Dai and Zimmerly 2002a) and is flanked by IS elements first identified in S. meliloti. We found that, in addition to R.et.F1 (lower band in fig. 2B), strain CE3 has a second fragmented copy of RmInt1 (upper band in fig. 2B). The signal corresponding to this larger copy was also detected in other R. etli strains, including Viking 1 and CFN42, and in R. leguminosarum bv. phaseoli strains (fig. 2B). We characterized the corresponding band from R. etli strain Viking 1 (hereafter referred to as R.et.F2) further by DNA sequence analysis. We also determined the DNA sequence of the single-truncated form of RmInt1 present in the S. medicae RMO02 (S.md.F1) and S. adhaerens LMG20582(S.ad.F1) strains (fig. 1C). Total DNA was digested with EcoRI and PCR was performed with primers 1 and 18R.0. The 5' end of the intron and the 5' exon sequences of S.md.F1, R.et.F2, and S.ad.F1 were then obtained by inverse PCR (see Materials and Methods). Similarly, the 3' end of the intron and sequences further downstream were obtained for the S.md.F1 intron. Sequence analyses showed 90% similarity to RmInt1 from S. meliloti strain GR4 for R.et.F2 and 93% similarity for both S.md.F1 and S.ad.F1 (fig. 1C). R.et.F2 carries a deletion of 89 nt in the RT domain coding sequence. A blast nucleotide search of databases for R.et.F2 homologs revealed the presence in R. etli strain CFN42 of an almost identical (99% identity) fragmented intron copy harbored by plasmid p42d (González et al. 2003) (fig. 1C). This R.et.F2 copy has two deletions, one of 89 and one of 112 nt, in the RT domain–coding sequence. The larger of these two deletions is located at the position from which R.et.F1 is truncated. Surprisingly, an R.et.F1 copy was also detected in the p42a plasmid harbored by the CFN42 strain (V. González, personal communication). As the corresponding hybridization signal was not detected in our CFN42 strain, as shown in figure 2B, it has probably lost either the plasmid (p42a) or a fragment containing the truncated intron copy.
Phylogenetic Analysis of the Rhizobiaceae RmInt1 Ribozyme
The ribozyme sequences of all introns (full-length and fragmented) were aligned from nucleotide positions 180 to 550 (determined for all introns in this study), including part of RmInt1 ribozyme dI and extending to the beginning of dIV, before the IEP coding region. This alignment was used to generate a phylogenetic tree using neighbor-joining (fig. 3B) and maximum likelihood algorithms. The branching of this tree was similar to that obtained with the IEP, with two main groups identified (fig. 3B). Surprisingly, R.et.F1 and R.et.F2 clustered with the S. terangae RmInt1 copies (see fig. 3B and C). Comparison of the ribozyme domain nucleotide sequences showed identical changes in the S. terangae and R. etli RmInt1 introns (fig. 4A and B). These results suggest that the truncated copies of RmInt1 in R. etli and the full-length intron of S. terangae share a common origin.
FIG. 4.— Base substitutions in the secondary structure model of dIII (A) and dIV (B) of RmInt1. Nucleotides conserved between introns are denoted by (°), substituted nucleotides are shown in bold. Identical structures were obtained for R.et.F1 and R.et.F2, which are therefore both labeled as Rhizobium etli. In dIII (A), distal substitutions were mostly complementary, preserving the structure of the RNA. The ribosome-binding site (RBS) in dIV (B) is shown. The start and stop codons of the IEP are indicated by rectangles. Relevant observations are also indicated.
Analysis of the RmInt1 Flanking Regions
In S. meliloti, RmInt1 is mainly associated with ISRm2011-2 and, to a lesser extent, with related elements (ISRm10-1, ISRm10-2) of the ISRm2011-2 group within the IS630-Tc1/IS3 insertion sequence family (Martínez-Abarca and Toro 2000b). We investigated this association in species other than S. meliloti by carrying out Southern blot hybridization using an ISRm2011-2 3' end–derived probe. At least one (S. medicae) of the 29 species tested other than S. meliloti possessed this particular IS (table 2), which appears to be also associated with RmInt1, as suggested by hybridization of the same bands to both intron and IS probes (not shown). Analysis of the DNA sequences flanking the intron showed that the RmInt1 copy of R. etli strains Viking 1 and CFN42 (R.et.F2) was inserted into a 3'-truncated form of ISRm10-1; this was also the case in R.et.F1 (Bittinger et al. 2000 and this work). The RmInt1 copy of S. terangae ORS22 was found within a full copy of ISRm10-1. These results provide further support for a common origin of the S. terangae and R. etli RmInt1 introns.
The S. adhaerens 5D19 intron is inserted into a new IS of the ISRm2011-2 group, ISRm10-3, which is present as two copies in this bacterial genome. One of the copies is intron free and the insertion of RmInt1 may therefore correspond to a genuine retrohoming event. ISRm10-3 was absent from all the other strains tested, as indicated by DNA hybridization (table 2). These results confirm previous observations in S. meliloti of a specific association between RmInt1 and a particular class of IS elements favoring the spread of this intron in related taxa of the Rhizobiaceae. In contrast, the fragmented introns of S. medicae RMO02 (S.md.F1) and S. adhaerens LMG20582(S.ad.F1) seem be located in unknown ORFs or outside of coding regions as no homologs were found in databases (fig. 5). Analysis of the nucleotides essential for RmInt1 target recognition (Jiménez-Zurdo et al. 2003) in the 5' exon of the S.md.F1 intron showed that neither the T-15 critical position in the 5' distal exon nor the IBS2 sequence and critical A-3 position in IBS1 are conserved (fig. 5). Thus, the original insertion events for both copies of the RmInt1 intron were probably ectopic transposition events rather than cases of retrohoming.
FIG. 5.— The 5' exons of the various sequenced RmInt1 introns. The sequences of seven independent target sites for RmInt1 are shown and compared with the wild-type–homing site, ISRm2011-2, from Sinorhizobium meliloti. Exon-binding sites (EBS) 1 and 2 of the intron are aligned with the targets, which are shown according to the following letter code: bold uppercase letters, conserved residues that do not affect homing efficiency; normal uppercase letters, nonconserved residues that do not affect homing efficiency; normal lowercase letters, substitutions that block intron homing according to Jiménez-Zurdo et al. (2003). Five of the seven targets had sequences resembling insertion sequences.
Mobility of RmInt1 in Rhizobial Hosts
RmInt1 invades targets in a site-specific manner, by means of an RNA intermediate (retrohoming) (Martínez-Abarca et al. 2004). The spread of RmInt1 in the Rhizobiaceae seems to have occurred by both vertical inheritance and by horizontal gene transfer events. These horizontal transfers may be related to the mobility of intron-host DNA (e.g., plasmids, IS.) rather than the independent mobility of the intron. RmInt1 mobility has been demonstrated only in S. meliloti (Martínez-Abarca, García-Rodríguez, and Toro 2000). We investigated the potential of RmInt1 to move around in other related taxa by carrying out mobility assays in various hosts of the Rhizobiaceae lacking the intron. Mobility assays using the intron donor plasmid pKG2.5 and target recipient plasmid pJB0.6+ were performed in S. medicae, S. terangae, Rhizobium tropici, R. leguminosarum bv. phaseoli, Agrobacterium rhizogenes, and Agrobacterium tumefaciens. Efficient homing was observed in all the host strains tested (fig. 6). Remarkably, although RmInt1 has never been reported in strains of Rhizobium tropici, A. rhizogenes, and A. tumefaciens, this intron was clearly mobile within these hosts. These findings provide further evidence that RmInt1 may have colonized the genome of species other than S. meliloti by independent mobility of the intron facilitated by the presence of host IS elements of the ISRm2011-2 group.
FIG. 6.— RmInt1 mobility assay in heterologous hosts. The homing assay was carried out in selected rhizobia with no copy of RmInt1 on the genome. Plasmid pools were analyzed by Southern blot hybridization with a 3'-end probe for ISRm2011-2. Strains with the donor plasmid pKG2.5 were transformed with receptor plasmid pJB0.6+ (odd lanes) or with a receptor plasmid without target site pJB129 (even lanes). Fragments detected by hybridization are indicated as follows: D, donor plasmid; R, recipient; H, homing product at the right of the gel, and molecular mass markers are indicated on the left. The following strains were used: Sinorhizobium meliloti GR4, lanes 1 and 2; Sinorhizobium medicae RMO15, lanes 3 and 4; Sinorhizobium terangae ORS19, lanes 5 and 6; Rhizobium leguminosarum bv. phaseoli Ro33, lanes 7 and 8; Rhizobium tropici B BR850, lanes 9 and 10; Agrobacterium rhizogenes 163c1, lanes 11 and 12; and Agrobacterium tumefaciens B6, lanes 13 and 14.
Discussion
The group II intron RmInt1 was first described in S. meliloti following analysis of the sequence of pRmeGR4b from GR4, a strain that contains nine copies of this element (Martínez-Abarca, Zekri, and Toro 1998). This intron was mostly found within copies of ISRm2011-2 (Martínez-Abarca, Zekri, and Toro 1998) and closely related IS elements (Martínez-Abarca and Toro 2000b). We show here that RmInt1 is also present in other Sinorhizobium and Rhizobium species. RmInt1 is present predominantly in the genus Sinorhizobium, where vertical inheritance seems to occur. However, our results also suggest that the RmInt1 intron has crossed the genus barrier and can independently colonize other bacterial genomes, including those of other rhizobia and Agrobacterium. We also present data indicating that RmInt1 tends to evolve into inactive forms by fragmentation, losing its 3' terminus.
RmInt1 is present in 90% of S. meliloti isolates (Mu?oz, Villadas, and Toro 2001) and is an active and efficient retroelement in this bacterial host. In this study, we also found copies of RmInt1 (full-length and fragmented copies) in other Sinorhizobium species (S. medicae, S. adhaerens, and S. terangae), whereas only fragmented copies were present in Rhizobium species, such as R. etli and R. leguminosarum bv. phaseoli. Both these species nodulate and fix nitrogen in a similar ecological niche, the root nodules elicited on the leguminous plant Phaseolus vulgaris. Sequence data and phylogenetic analysis suggest that RmInt1 is native to the genus Sinorhizobium, in which it appears to have been inherited vertically from a common ancestor or by horizontal transfer among closely related strains and species but possibly lost in various species. Our results also suggest that the RmInt1 intron copies in R. etli and R. leguminosarum bv. phaseoli share a common origin with the S. terangae RmInt1 intron. RmInt1 may have been present in an ancestor of the Rhizobiaceae family but are lost in most of the lineages giving rise to the various genera. Alternatively, the RmInt1 intron of R. etli and R. leguminosarum bv. phaseoli may have been acquired by lateral transfer from Sinorhizobium species. The vertical inheritance hypothesis seems to be the least plausible because it involves evolutionary convergence of the Rhizobium and S. terangae RmInt1 introns. Sinorhizobium terangae and other Sinorhizobium species have full-length introns, whereas only fragmented introns that had accumulated larger numbers of mutations were found in Rhizobium. However, most of the R. etli and R. leguminosarum bv phaseoli strains isolated from different geographical origins harbored truncated RmInt1 copies. An early transfer event before the divergence of these two rhizobial species may account for these findings.
RmInt1 is currently very successful in S. meliloti because of the presence of host mobile DNA of the ISRm2011-2 group, which provides a site-specific target and facilitates the survival and spread of the intron. Studies on natural field populations of S. meliloti have shown that the RmInt1 intron spread not only by transposition of the host mobile DNA but also by independent movement of the intron (Mu?oz, Villadas, and Toro 2001). Similar conclusions have been drawn concerning the independent mobility of group II introns in E. coli populations, in studies of group II introns in the ECOR collection (Dai and Zimmerly 2002b), and experimental demonstrations of conjugation-mediated transfer of the LltrB group II intron between bacterial species (Belhocine, Plante, and Cousineau 2004). RmInt1 spreads by retrohoming or ectopic transposition, a conclusion also supported by this work. An analysis of flanking exon sequences suggested that in some cases intron insertion may have involved an ectopic transposition event, and the mobility assay data indicated that RmInt1 displays retrohoming in bacterial species and genera other than S. meliloti.
It has been reported that bacterial group II introns degenerate mainly by fragmentation (Dai and Zimmerly 2002a). It has been reported that 38% of the described bacterial intron fragments are 5' truncated (Dai and Zimmerly 2002a), possibly due to incomplete reverse transcription after intron RNA insertion. Interestingly, we found that RmInt1 tended to degenerate by means of truncation at its 3' terminus. In at least one case (R.et.F1), fragmentation may have occurred by insertion of an IS (ISRm3) with further rearrangements and deletion. In other cases (R.et.F2), the intron displayed internal deletions. Similar data have been reported for other bacterial group II introns, such as E. coli intron E.c.I4 (Dai and Zimmerly 2002b), and this may be a general mechanism for the generation of 3'-truncated fragmented introns.
Our results provide further evidence that bacterial group II introns are able to propagate by vertical and lateral transfer, retrohoming, or ectopic transposition and show that they tend to evolve into fragmented introns, mechanisms underlying the origin of eukaryotic spliceosomal introns.
Acknowledgements
The authors thank all the donors of strains and isolates for their invaluable generosity. This work was supported as part of research projects BIO99-0905 and BIO2002-02579 by the Ministerio de Ciencia y Tecnología. E.M.-A. was funded by Junta de Andalucía. M.F.-L. received postdoctoral grants from the Ministerio de Ciencia y Tecnología. A.W. would like to thank the Fund for Scientific Research–Flanders for her postdoctoral research fellowship. The R. etlim CFN42 was provided by Dr. V. González.
References
Belfort, M., V. Derbyshire, M. M. Parker, B. Cousineau, and A. M. Lambowitz. 2002. Mobile introns: pathways and proteins. Pp. 761–783 in N. Craig, R. Craigie, M. Gellert, and A. M. Lambowitz, eds. Mobile DNA. 2nd edition. ASM Press, Washington D.C.
Belhocine, K., I. Plante, and B. Cousineau. 2004. Conjugation mediates transfer of the Ll. LtrB group II intron between different bacterial species. Mol. Microbiol. 51:1459–1469.
Bittinger, M. A., J. A. Gross, J. Widom, J. Clardy, and J. Handelsman. 2000. Rhizobium etli CE3·carries vir genes homologs on a self-transmissible plasmid. Mol. Plant Microbe Interact. 13:1019–1021.
Dai, L., and S. Zimmerly. 2002a. Compilation and analysis of group II intron insertions in bacterial genomes: evidence for retroelement behaviour. Nucleic Acids Res. 30:1091–1102.
———. 2002b. The dispersal of five group II introns among natural populations of Escherichia coli. RNA 8:1294–1307.
———. 2003. ORF-less and RT-encoding group II introns in archaebacteria, with a pattern of homing into related group II intron ORFs. RNA 9:14–19.
Dickson, L., H. R. Huang, L. Liu, M. Matsuura, A. M. Lambowitz, and P. S. Perlman. 2001. Retrotransposition of a yeast group II intron occurs by reverse splicing directly into ectopic DNA sites. Proc. Natl. Acad. Sci. USA 98:13207–13212.
Ditta, G., S. Stanfield, D. Corbin, and D. R. Helinski. 1980. Broad host range DNA cloning system from gram-negative bacteria. Construction of a gene bank of Rhizobium meliloti. Proc. Natl. Acad. Sci. USA 77:7347–7351.
Eickbush, T. H. 1999. Mobile introns: retrohoming by complete reverse splicing. Curr. Biol. 9:11–14.
Felsenstein, J. 1995. PHYLIP (phylogeny inference package). Version 3.55c. Distributed by the author, Department of Genetics, University of Washington, Seattle. (http://evolution.genetics.washington.edu/phylip/getme.html).
Ferat, J. L., M. Le Gouar, and F. Michel. 1994. Multiple group II self-splicing introns in mobile DNA from Escherichia coli. C. R. Acad. Sci. Ser III Life Sci. 317:141–148.
Ferat, J. L.,. 2003. A group II intron has invaded the genus Azotobacter and is inserted within the termination codon of essential groEL gene. Mol. Microbiol. 49:1407–1423.
Ferat, J. L., and F. Michel. 1993. Group II self-splicing introns in bacteria. Nature 364:358–361.
Galibert, F., T. M. Finan, S. R. Long et al. (56 co-authors). 2001. The composite genome of the legume symbiont Sinorhizobium meliloti. Science 293:668–672.
Geremia, R. A., P. Mergaert, D. Geelen, M. Van Montagu, and M. Holsters. 1994. The NodC protein of Azorhizobium caulinodans is an N-acetylglucosaminyltransferase. Proc. Natl. Acad. Sci. USA 91:2669–2673.
González, V., P. Bustos, M. A. Ramírez-Romero et al. (16 co-authors). 2003. The mosaic structure of the symbiotic plasmid of Rhizobium etli CFN42 and its relation to other symbiotic genome compartments. Genome Biol. 4:R36.
Granlund, M., F. Michel, and M. Norgren. 2001. Mutually exclusive distribution of IS1548 and GBSi1, an active group II intron identified in human isolates of group B streptococci. J. Bacteriol. 183:2560–2569.
Ichiyanagi, K., A. Beauregard, and M. Belfort. 2003. A bacterial group II intron favors retrotransposition into plasmid targets. Proc. Natl. Acad. Sci. USA 100:15742–15747.
Ichiyanagi, K., A. Beauregard, S. Lawrence, D. Smith, B. Cousineau, and M. Belfort. 2002. Multiple pathways for retrotransposition of the Ll.ltrB group II intron include a potential relationship to DNA replication. Mol. Microbiol. 46:1259–1272.
Jiménez-Zurdo, J. I., F. M. García-Rodríguez, A. Barrientos-Durán, and N. Toro. 2003. DNA target site requirements for homing in vivo of a bacterial group II intron encoding a protein lacking the DNA endonuclease domain. J. Mol. Biol. 326:413–423.
Lambowitz, A. M., M. G. Caprara, S. Zimmerly, and P. S. Perlman. 1999. Group I and group II ribozymes as RNPs: clues to the past and guides to the future. Pp. 451–485 in T. R. Cech, and J. F. Atkins, eds. The RNA world. 2nd edition. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Martínez-Abarca, F., A. Barrientos-Durán, M. Fernández-López, and N. Toro. 2004. The RmInt1 group II intron has two different pathways for mobility using predominantly the nascent lagging strand at DNA replication forks for priming. Nucleic Acids Res. 32:2880–2888.
Martínez-Abarca, F., F. M. García-Rodríguez, and N. Toro. 2000. Homing of a bacterial group II intron with an intron-encoded protein lacking a recognizable endonuclease domain. Mol. Microbiol. 35:1405–1412.
Martínez-Abarca, F., and N. Toro. 2000a. Group II introns in the bacterial world. Mol. Microbiol. 38:917–926.
———. 2000b. RecA-independent ectopic transposition in vivo of a bacterial group II intron. Nucleic Acids Res. 28:4397–4402.
Martínez-Abarca, F., S. Zekri, and N. Toro. 1998. Characterization and splicing in vivo of a Sinorhizobium meliloti group II intron associated with particular insertion sequences of the IS630-Tc1/IS3·retroposon superfamily. Mol. Microbiol. 28:1295–1306.
Michel, F., and J. L. Ferat. 1995. Structure and activities of group II introns. Annu. Rev. Biochem. 64:435–461.
Michel, F., K. Umesono, and H. Ozeki. 1989. Comparative and functional anatomy of group II catalytic introns—a review. Gene 82:5–30.
Mu?oz, E., P. J. Villadas, and N. Toro. 2001. Ectopic transposition of a group II intron in natural bacterial populations. Mol. Microbiol. 41:645–652.
Mu?oz-Adelantado, E., J. San Filippo, F. Martínez-Abarca, F. M. García-Rodríguez, A. M. Lambowitz, and N. Toro. 2003. Mobility of the Sinorhizobium meliloti group II intron RmInt1 occurs by reverse splicing into DNA, but requires an unknown reverse transcriptase priming mechanism. J. Mol. Biol. 327:931–943.
Nakamura, Y., T. Kaneko, S. Sato et al. (21 co-authors). 2002. Complete genome structure of the thermophilic cyanobacterium Thermosynechococcus elongatus BP-1. DNA Res. 9:123–130.
Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
San Filippo, J., and A. M. Lambowitz. 2002. Characterization of the C-terminal DNA-binding/DNA-endonuclease region of a group II intron-encoded reverse transcriptase. J. Mol. Biol. 324:933–951.
Sharp, P. A. 1991. Five easy pieces. Science 254:663.
Sheveleva, E. V., and R. B. Hallick. 2004. Recent horizontal intron transfer to a chloroplast genome. Nucleic Acids Res. 32:803–810.
Toro, N. 2003. Bacteria and archaea group II introns: additional mobile genetic elements in the environment. Environ. Microbiol. 5:143–151.
Toor, N., G. Hausner, and S. Zimmerly. 2001. Coevolution of group II intron RNA structures with their intron-encoded reverse transcriptase. RNA 7:1142–1152.
Toro, N., F. Martínez-Abarca, M. Fernández-López, and E. Mu?oz-Adelantado. 2003. Diversity of group II introns in the genome of Sinorhizobium meliloti strain 1021: splicing and mobility of RmInt1. Mol. Genet. Genomics 268:628–636.
Toro, N., M. D. Molina-Sánchez, and M. Fernández-López. 2002. Identification and characterization of bacterial class E group II introns. Gene 299:245–250.
Van de Peer, Y., and R. De Wachter. 1994. TREECON for Windows: a software package for the construction and drawing of evolutionary trees for the Microsoft Windows environment. Comput. Appl. Biosci. 10:569–570.
Vincent, J. M. 1970. A manual for the practical study of the root-nodule bacteria. IBP handbook No. 15. Blackwell Scientific, Oxford.
Yang, J., G. Mohr, P. S. Perlman, and A. M. Lambowitz. 1998. Group II intron mobility in yeast mitochondria: target DNA-primed reverse transcription activity of al1 and reverse splicing into DNA transposition sites in vitro. J. Mol. Biol. 282:505–523.
Zimmerly, S., G. Hausner, and X. Wu. 2001. Phylogenetic relationships among group II intron ORFs. Nucleic Acids Res. 29:1238–1250.(Manuel Fernández-López*, )