Mutually Exclusive Distribution of IS1548 and GBSi1, an Active Group II Intron Identified in Human Isolates of Group B Streptococci

Identification of GBS group II intron GBSi1.All previously examined GBS isolates that harbored IS1548 had a copy of the IS element downstream of the C5a-peptidase gene (scpB) (21). This region was chosen for further examination and was sequenced in three GBS isolates of serotype III. These were 5531, an endocarditis isolate harboring IS1548, and two isolates which lacked IS1548: M732, originally isolated from a neonate with meningitis, and A5, a blood isolate from an adult. In 5531 the DNA sequence between scpB and the IS element was shown to be a noncoding region of 158 bp denominated spacer. The corresponding DNA sequence in A5 was indistinguishable from that of spacer in 5531 (Fig. 1). However, when spacer was sequenced in GBS isolate M732, it revealed an insertion of 1,857 bp. The inserted sequence was 92% identical to a sequence located downstream of dexB (α1-6 glucosidase gene) flanking the 5′ capsule region in a serotype 19F isolate ofStreptococcus pneumoniae (9). A short stretch of nucleotides in the insert had 90% identity (38 of 42 bp) to an H-repeat gene, which has features of IS elements (32) inE. coli. The G+C content of the insert in spacer was 45%.

• Open in new tab
• Download powerpoint

Fig. 1.

Genomic organization of the region downstream of the C5a-peptidase gene (scpB) in three variants of serotype III GBS isolates. Dark grey box, noncoding spacer sequence, the nucleotide sequence of which is shown. The first nucleotide after the stop codon of the C5a-peptidase gene is designated 1. Arrows, inverted repeat. Drawn approximately to scale are GBSi1 (hatched box), IS1548(light grey box), and surrounding genes (open boxes). The insertion points of group II intron GBSi1 and IS1548 in the spacer sequence are depicted by hatched and grey triangles, respectively.

The deduced amino acid sequence from the open reading frame (ORF) encompassing bp 482 to 1793 of the insert was similar to those of the maturase proteins encoded by group II introns in yeast mitochondria. The putative GBS intron designated GBSi1 was further characterized by comparison of the deduced amino acid sequence with sequences of the protein families in Pfam at the Sanger Centre. Out of a total of 436 amino acids, a domain similar to those of RTs was found to encompass amino acids 64 to 286 (Fig. 2A). These enzymes, which transcribe DNA from an RNA template, occur in a variety of mobile elements including retrotransposons, retroviruses, group II introns, and bacterial msDNAs (54). The fifth of seven domains found in the RTs contains a YxDD box, a conserved sequence present in all RTs from prokaryotes to eukaryotes. The YxDD box found in GBSi1 was RYADD at amino acids 232 to 236 (Fig. 2A). Domain X, which consists of a section of about 100 amino acids with maturase activity, is not always present in RT-containing group II introns. This region has fewer conserved features than the RTs, but domain X is usually dominated by basic amino acids (40). The corresponding region in the GBSi1 coding sequence had 22 basic and 5 acid amino acids spanning a region of 100 amino acids downstream of the RT region. An alignment between the maturase domain of the group II intron-encoded protein from the cyanobacterium Calothrix and the GBS amino acid sequence is shown in Fig. 2B. No Zn²⁺ finger-like motif, which has been associated with the C-terminal domain of intron-encoded proteins with endonuclease activity (40), was identified.

• Open in new tab
• Download powerpoint

Fig. 2.

(A) Alignment of the protein sequence deduced from group II intron GBSi1 of GBS isolate M732 with group II intron-encoded proteins of different origins. RT1 to -7, conserved domains of RTs; numbers in parentheses, accession numbers deposited in GenBank. The GBSi1 sequence is from this study (accession no. AJ292930 ). (B) Amino acid alignment of the region downstream of RT7 in the GBSi1 protein and the maturase domain of the protein encoded by a group II intron in the cyanobacterium Calothrix. The alignment was performed with BLAST at the National Center for Biotechnology Information with default settings. Numbering for the Calothrix sequence is from the database entry (accession no. CAA50529 ).

The GBSi1 RNA sequence could be folded (Fig.3) into a secondary structure with the well-established features of the ribozyme component of group II introns (38). The folding of GBSi1 revealed the characteristic six helical domains in addition to the sequences that form exon binding site 1 (EBS1) and intron binding site 1 (IBS1) (Fig. 3). However, the second intron-exon pairing (EBS2-IBS2) found in many group II introns could not be identified and appears to be missing. Mobile group II introns splice out from pre-mRNA as a branched molecule, a lariat. These lariats are created by a 2′-5′ bond between a bulging adenosine residue located seven to eight nucleotides upstream from the 3′ intron-exon junction in domain VI and the first nucleotide in the intron, which usually is a G (38). The first nucleotides of the GBS intron RNA would be GUACG, and an A residue is located seven nucleotides upstream of the 3′ intron-exon junction. Comparison of the GBSi1 putative secondary structure with those published for other group II introns reveals striking similarities with intron Ec.intB (17) from E. coli (Fig. 3). Some sections of domain I, which includes a potential α-α′ long-range pairing at the expected location, are reminiscent of intron P1.LSU/2 and allies (19). Since both these introns (and also theCalothrix molecule [Fig. 2B]) belong to subgroup IIB (38), GBSi1 is most likely a member of this subclass.

• Open in new tab
• Download powerpoint

Fig. 3.

Secondary-structure model of group II intron GBSi1. The structure of the ribozyme was constructed using comparative analysis (38). Arrowheads, intron-exon junctions; roman numbers, six conserved domains of group II introns. The intron ORF is located in domain IV. EBS1 designates a sequence that base pairs with IBS1 located immediately upstream of the intron insertion site. α-α′, tertiary base pairing that is generally conserved in group II introns at the locations indicated (38). Lowercase letters, base substitutions in the GBSi1-related sequence from S. pneumoniae serotype 19F (accession no. AF030367 ; note the 64-nucleotide insertion in domain I). Heavy black lines delimit sections whose nucleotides and potential secondary structure are conserved in intron Ec.IntB from E. coli (17), while grey lines indicate those segments of domain I that have clear counterparts in intron P1.LSU/2 from Pylaiella littoralismitochondria (19). n, nucleotides.

A putative target site motif for streptococcal group II introns.The insertion site of GBSi1 in spacer is preceded by an inverted repeat region, 67 bp downstream from the 3′ end ofscpB (Fig. 1). When the S. pneumoniaeserotype 4 nucleotide sequence at the ongoing sequencing project presented at the TIGR website was searched, GBSi1-like sequences were found. The homologies were consistent with two copies of the putative group II intron, one copy at position 122378 in forward orientation and the other at position 1994149 in the opposite direction. Southern blot analysis of GBS isolates M732 and A15, which according to PCR analysis harbored GBSi1, revealed two intron copies also in GBS (data not shown). When the sequence upstream of the two putative pneumococcal introns from the TIGR sequence and that of the S. pneumoniaeserotype 19F sequence were compared with the sequence upstream of the GBSi1 insert in spacer, all four sequences were found to consist of an inverted repeat region followed by a poly(T) tail (Fig.4). We have also noted the presence of this motif upstream of group II intron sequences in Bacillus halodurans, Pseudomonas alcaligenes, and Pseudomonas putida (Fig. 4). Interestingly, these introns and the S. pneumoniae 19F sequence belong to a distinct branch in the group II phylogenetic tree that was recently proposed by Martinez-Abarca and Toro (35) based on an alignment of RT domains. See Fig. 2A for comparison between the RT of GBSi1 and those of B. halodurans and P. alcaligenes. Moreover, in all members of this subgroup, domain V has only 30 nucleotides (Fig. 3) and its first 2 bp are C1:G30 and C2:G29 (rather than R:Y and A:U, as in most other group II introns).

• Open in new tab
• Download powerpoint

Fig. 4.

Suggested target motif for streptococcal group II introns. The first nucleotides of the putative group II introns and the sequence immediately upstream of the intron for GBS (GBSi1) andS. pneumoniae (Pn) are shown. For comparison, similar sequences upstream of group II introns in B. halodurans(B.hd.I and -II), P. alcaligenes (Ps.alc.), and P. putida (Ps.put.) are shown. Forty-three nucleotides are left out in the sequences from the Pseudomonas species. The sequences from S. pneumoniae serotype 4 are from the TIGR website, (http://www.tigr.org .; 7.25.2000) and are designated Pn4. Pn19F,S. pneumoniae of serotype 19F. Figures in parentheses (left) indicate the nucleotide positions of GBSi1 (this study) and of potential introns in the TIGR and other sequences. The accession number of the Pn19F sequence is AF030367 , and those of B. halodurans I and II, P. alcaligenes, and P. putida are AB031210 , AP001507 , U77945 , and X91654 , respectively. Arrows, locations of the inverted repeat sequences forming potential RNA helices. IBS1 of GBSi1 is boxed.

Distribution of IS1548 and GBSi1 in GBS isolates.Seventeen of 67 GBS isolates were previously shown to harbor IS element IS1548 (21). The 67 isolates were examined by PCR to reveal the distribution of the group II intron. The GBSi1 intron was found in 26 isolates of serotypes II, III, and V (Table2). None of the isolates that harbored GBSi1 was shown to contain IS1548. In order to test if this mutually exclusive distribution extended to other insertion sequences, the presence of IS861 among the isolates was determined. A PCR product of about 1,200 bp indicated the presence of IS861 in 34 isolates, distributed among all serotypes except the single serotype IV strain of this study (Table 2). All isolates containing IS1548 also harbored IS861, and IS861 was found in several isolates where the intron was present. In 15 of the 67 isolates none of the three elements could be identified. The mutually exclusive distribution of IS1548and GBSi1 was observed in both the Swedish and the American isolates. The latter included three ET1 and ET12 serotype III isolates, defined by Musser et al. to belong to high-virulence and low-virulence subpopulations respectively (46). The high-virulence subpopulation contained GBSi1, whereas the low-virulence group of isolates harbored IS1548 (Table 2).

DNA sequence downstream of a copy of IS1548 located in the scpB region.The region downstream of the copy of IS1548 located in the scpB region in endocarditis isolate 5531 was sequenced. The DNA sequence revealed the presence of lipoprotein-encoding gene lmb and a gene with unknown function, called orfY, located downstream of lmb(53). The amino acid sequence deduced from lmbhad similarities with AdcA in S. pneumoniae (35% identity over 303 amino acids) in addition to other members of a solute binding family of putative adhesive proteins represented by, for example, FimA in Streptococcus parasanguis, PsaA in S. pneumoniae, and ScaA in Streptococcus gordonii(53). Furthermore, the lmb nucleotide sequence was almost identical (96%) to the incomplete sequence of 450 bp called orf3 located downstream of scpA in group A streptococcus M-type 49 (47). The lmbgene was present in all isolates examined including GBS isolates M732, harboring GBSi1 but not IS1548, and A5, which lacked both GBSi1 and IS1548 (Fig. 1). The DNA sequence of GBS 5531 revealed that in this endocarditis isolate IS1548 was inserted 9 bp from the putative −35 promoter of lmb. The GBSi1 insertion in spacer in the M732 isolate from a neonate with meningitis was located 88 bp further upstream (Fig. 1).

In vitro self-splicing activity of the GBSi1 intron.Since the self-splicing activity of in vitro-synthesized group II intron transcripts is generally unaffected, or even increased, by deletion of the terminal loop of domain IV (reviewed by Michel and Ferat [37]), the size of this loop was reduced from 1403 to 8 residues (resulting in a 462-nucleotide intron) in the construct we used to test for in vitro self-splicing (see Materials and Methods). When a purified intron-containing transcript was incubated at 45°C in the presence of elevated concentrations of ammonium and magnesium ions, two major products with the expected electrophoretic mobilities for the ligated exons and linear intron were generated by a rather slow reaction which consumed about half of the precursor molecules in 2 h (Fig.5A; experimental conditions were about optimal in terms of yield of the major products). The identity of the gel-eluted products was confirmed by sequencing them with an RT. The sequence of the excised intron (Fig. 5B) could be read almost all the way to the final stop, which coincided with the intron-5′ exon junction. Despite repeated attempts at reverse transcription with a primer complementary to the 3′ exon, the ligated exons yielded only rather low-quality sequencing lanes (Fig. 5C). As expected, the sequence remained compatible with that of the precursor transcript all the way to the 3′ splice site and then diverged from it, but reading was not possible over the entire length of the molecule. Nevertheless, elongation could be shown to stop at precisely the expected site, 21 nucleotides upstream of the ligation junction (Fig. 5C).

• Open in new tab
• Download powerpoint

Fig. 5.

In vitro self-splicing activity of the GBSi1 molecule. (A) Schematic presentation of GBSi1, the intron precursor used in the experiment, and the excised intron. (B) Time course of a self-splicing reaction. Shown is an autoradiograph of a gel electrophoresis of reaction products produced under optimal conditions (see Materials and Methods). The numbers indicate expected lengths in nucleotides. Lane MW, a sample reaction of a construct derived from group II intron P1.LSU/2 (10) electrophoresed on the same gel as GBSi1. The reaction products include two lariats composed of a 572-nucleotide circle and 7- and 114-nucleotide tails. (C) Reverse transcription of the gel-eluted precursor transcript and excised intron with primer GBS1 complementary to a sequence located downstream of the 5′ intron-exon junction. Shown is an autoradiograph of a sequencing gel. Lanes are designated with the base complementary to the dideoxynucleotide used for sequencing; −, no dideoxy. The sequence shown at right is the one inferred for the precursor transcript (arrow, expected intron-5′ exon junction). (D) Reverse transcription of the gel-eluted precursor transcript and ligated exons with primer GBS2 complementary to the 3′ exon. Labeling is as in panel C. The sequences shown at left and right are the ones inferred for the ligated exons and precursor transcript (arrows, expected site of ligation and the intron-3′ exon junction, respectively).

In addition to the excised intron and ligated exons, a minor product with the expected mobility for the 3′ exon became apparent at long reaction times; it is likely to be generated by the “spliced-exon reopening” reaction described by Jarrell et al. (26). Small amounts of another product which migrated just beyond the precursor and which could correspond to the linear intron-3′ exon reaction intermediate were also observed, especially at high temperatures, which result in a weakened intron-exon interaction. Much more surprising was our complete inability to detect a branched (lariat) molecule, whose highly retarded migration in polyacrylamide gels (Fig. 5A, lane MW) is diagnostic. Under all conditions tested, i.e., temperature between 40 and 55°C, pH between 6.2 and 7.6, ammonium between 100 mM and 1 M, and magnesium between 20 and 100 mM, splicing appears to be initiated exclusively by hydrolysis at the 5′ splice site (see discussion in reference 37). We also checked that lengthening the 5′ exon so as to include the inverted repeat sequence which extends from positions −39 to −9 (see above and Discussion) does not improve the efficiency of the splicing reaction.

Group II intron GBSi1 identified in GBS.This work presents a previously undescribed group II intron in GBS. There has been a substantial increase in the number of group II introns from different bacterial genera reported since the first description of group II introns in bacteria in 1993 (17, 18, 28, 34, 39, 45, 49, 52, 56). However, information about their distribution among bacterial populations, the impact of the bacterial introns on the functionality of the genome, and the insertion sites of the introns is limited. Among the few gram-positive species in which group II introns have been described are Clostridium difficile, L. lactis, and Bacillus megaterium (25, 39, 45, 52). The identification of GBSi1 and the related putative group II introns inS. pneumoniae extends the presence of group II introns to the clinically important genus Streptococcus.

Mobility of group II introns was first shown in introns of the yeastSaccharomyces cerevisiae mitochondrial cytochrome oxidase genes (cox) (31, 43). The two cox1introns ail and ai2 are the best-characterized mobile group II introns. They have a long ORF encoding a protein with RT (27) and maturase characteristics (42). The organization of group II introns is conserved throughout the eukaryotic and prokaryotic kingdoms. Regardless of their origin, their RNA can be folded into a conserved structure composed of six secondary structure domains radiating from a central wheel. When an ORF encoding a maturase and/or is present, it is located in domain IV (38). These characteristics are present in GBSi1, which has many other features of a bona fide group II intron, such as sequences nearly identical to the consensus at the 5′ and 3′ extremities, a bulging A on the 3′ side of domain VI, and EBS1-IBS1 and α-α′ potential long-range pairings (Fig. 3). At the same time, the GBSi1 ribozyme is somewhat divergent from most of its group II counterparts: the EBS2-IBS2 pairing appears to be missing and the highly conserved domain V is lacking 2 bp. Because one trivial explanation for those unusual features would be that the GBSi1 insert is a defective copy of a previously active transposable element, we attempted to rule out this interpretation by investigating the capacity of GBSi1-containing transcripts to self-splice in vitro.

While the ability to perform self-splicing is necessary both for the excision and transposition of group II introns, the splicing reaction has been shown to be facilitated by the intron-encoded protein (36); in the absence of this product, most “active” group II introns perform rather sluggishly in vitro, even at elevated magnesium and monovalent ion concentrations (37). The rather slow kinetics of the reaction shown in Fig. 5 thus provide no indication that the GBSi1 molecule has become freed from selective pressure for optimal function in vivo. On the other hand, the absence of a lariat product, which is so far unique for a naturally occurring member of group II, is all the more surprising in that the bulging A, whose 2′-OH group is responsible for branch formation in other group II introns, is definitely present at the expected location on the 3′ side of domain VI (Fig. 3). One possibility worth investigating would be that, for GBSi1, the intron-encoded protein is necessary to initiate splicing by transesterification.

Identification of a putative group II intron target site motif.A model for the retrohoming process in bacteria has been suggested by Cousineau et al. based on the mobility of L. lactis group II intron LlLtrB (11). The intron RNA and the intron-encoded protein form a combined RNA-protein complex acting in consert to cleave the recipient DNA. The RNA catalyzes the cleavage of the sense strand at the intron insertion site, and the protein cleaves the antisense strand at position +9 from the insertion site (36). A cDNA copy generated by reverse transcription is then incorporated into the recipient DNA (11). The retrohoming in L. lactis is independent of extensive homologous recombination and of recA. However, about 40 to 50 nucleotides are required for the ribonucleoprotein recognition of the insertion site, and efficient retrotransposition requires at least 25 nucleotides upstream and downstream of the insertion site (11). The intron RNA base pairs with nucleotides −13 to +1 at the intron binding sites, (IBSs), whereas the protein recognizes the nucleotide sequence from −25 to −13 and +2 to +25 counting from the intron insertion site. The sequences upstream of the insertion site of GBSi1, of the putative group II introns in S. pneumoniae, and of the group II introns of branch 3 as described by Martı́nez-Abarca and Toro (35) were found to have a common theme. Each sequence contains a region of 29 to 39 nucleotides, upstream of the first guanidine of the intron, composed of an inverted repeat sequence followed by a poly(T) tail, a combination that is common in bacterial transcription termination sites. High target specificity is important for the retrohoming of group II introns (30, 37). It seems likely that the inverted repeat region serves as a motif that, by adding to the specificity of recognition between the streptococcal group II introns and their target sites, compensates for the absence of the EBS2-IBS2 pairing. A similar recognition pattern for the integrases of other mobile elements has been described. In phage lambda, the integrase recognizes approximately 30 bases made up of a pair of imperfect inverted repeats in the recombination sites (6).

The presence of GBSi1 and IS1548 in GBS isolates is mutually exclusive.When 40 GBS isolates of serotype III were analyzed, it was found that 20 isolates harbored GBSi1 and 17 contained IS1548 but that in none of the isolates were both elements present. Several studies of GBS populations support a basically clonal distribution of isolates (24, 46, 48). Hauge et al. revealed six major lineages in the GBS population by combining electrophoretic analysis of multilocus enzymes with several other tests, among which was an assay for hyaluronidase activity (24). This and other studies have demonstrated that GBS type III isolates belong to two major, distantly related evolutionary lineages (24, 46, 48). One of the two major lineages among type III isolates of GBS found by Hauge et al. exhibited a hyaluronidase-negative phenotype (24). This genetic lineage of GBS included the three ET12 isolates examined in this study, originally defined by Musser et al. as a low-virulence subpopulation of GBS (46). Serotype III isolates that harbor IS1548, including these three ET12 strains, have an insertional mutation in the hyaluronidase gene and lack hyaluronidase activity (21). GBSi1 could possibly constitute a marker for the other major lineage of serotype III isolates. In favor of this hypothesis is the fact that, besides the mutual exclusiveness of IS1548 and GBSi1 among the isolates, the three ET1 isolates that previously had been assigned to the second major lineage of GBS type III (24, 46) all contained GBSi1.

No association between the existence of GBSi1 and another putative mobile element was found in the GBS isolates. This is in contrast to many other bacterial group II introns, which reside in transposable elements. In C. difficile the group II intron is located in conjugative transposon Tn5397 of the Tn916 family (45). The group II introns in E. coli andShigella flexneri reside in IS or IS-like elements, and one group II intron from L. lactis is found on a conjugative plasmid (17, 39, 49). It has been suggested that other mobile genetic elements may act as carriers of bacterial group II introns (17, 34). Lactococcal group II intron Ll.LtrB has successfully been transferred via a plasmid both to L. lactis and to E. coli (11). However, there are non-long terminal repeat retrotransposable elements other than the group II introns that seem to be acquired only by virtue of vertical descent and thus are ancient participants in the host genome (33). Whether the group II introns in GBS are subject to horizontal transfer or mainly are vertically transmitted is not known. One possibility is that the genetic background of the strains interferes with the ability to harbor both IS1548 and GBSi1 in the same isolate. Interference between the target sites of the two elements would constitute a plausible explanation for mutual exclusion. IS1548 and GBSi1 both have a target site in the spacer region downstream of scpB. However, their sites of insertion are located 88 bp apart, and the presence of either element in the spacer sequence causes no apparent alteration to the presumed target sequence of the other element. To ascertain whether such interactions exist, further experimental work is needed. Another conceivable interaction between the elements could result from GBSi1 and IS1548 being harbored by yet-unidentified bacteriophages or transposons with immunity against the presence of the other element in the genome.

Different positioning of IS1548 and GBSi1 in relation to the downstream-located lmb gene.Downstream of the spacer region among GBS isolates harboring either GBSi1, IS1548, or neither of the elements was gene lmb, previously defined as encoding a laminin binding protein (53). Interestingly in view of the suggested association of IS1548-harboring isolates with endocarditis,lmb is similar to a family of streptococcal lipoprotein adhesins including FimA and PsaA. FimA of Streptococcus parasanguis (16) has been associated with the colonization of damaged heart tissue in an endocarditis rat model (5). S. pneumoniae psaA mutants have been shown to display both reduced adherence to the A-549 lung epithelial cell line and reduced virulence in mice (4). The proteins in this family appear to be both adhesins and part of ATP-binding cassette (ABC) transporters for metal cations (13, 29). Recent data have also linked these ABC transporters to competence for genetic transformation in S. pneumoniae and S. gordonii(13, 14, 29).

It remains to show whether Lmb has any additional substrate-binding capacities besides the ability to bind laminin and to elucidate the possible implications of this protein for the pathogenesis of GBS endocarditis. In this context it will also be important to investigate whether the different positionings of IS1548 and GBSi1 with respect to the tentative lmb promoter affect transcription and hence possibly also the pathogenicity of the isolates.