The transposable DNA elements known as insertion sequences (ISs) are a potent force in the evolution of bacteria. Through their activity, these elements contribute to the raw material of genetic diversity upon which forces of natural selection act to propel evolution. Among the genotypic changes catalyzed by ISs are gene inactivation, gene activation through the juxtaposition of an IS-contained promoter, deletions and inversions of adjacent DNA sequences, and mobilization of flanked DNA sequences by transposition. In addition, by serving as portable regions of sequence similarity, ISs can foster rearrangements catalyzed by host homologous recombination systems. These elements are ubiquitous in bacterial genomes. Upon insertion into a novel site, nearly all ISs have been found to create a duplication of from 2 to 14 bp, originally present at the target sequence, which subsequently flanks the inserted IS (3). The boundaries of insertion sequences are typically defined by short perfect or nearly perfect inverted repeat sequences of from 10 to 40 bp (3).

In the human pathogen Bordetella pertussis, two insertion sequences named IS481 and IS1002 have been characterized (4, 7). These elements can be considered members of the IS3 family of insertion sequences based on the sequence similarity of their predicted transposase proteins to the orfB product of the IS3 family transposases (3). Approximately 100 copies of IS481 are present in B. pertussis strains (2). The copy number of IS1002 has been reported to range from 4 to 8 (8). IS481 and IS1002 are similar in length—1,053 and 1,040 bp, respectively—and show 61.5% nucleotide sequence identity (4, 7). The terminal, imperfect, inverted repeats defining the ends of these elements have been reported to be 28 and 29 bp in length, respectively (4, 7). In both cases the precise limits of these elements were defined as the sequence bounded by terminal inverted repeat sequences after comparison of the DNA sequence of fragments containing IS insertions at different chromosomal locations. Based on this analysis, it was concluded that, unlike other ISs, IS481 and IS1002 do not create a short duplication upon insertion (4, 7). Although a number of other examples of IS481 and IS1002 have been sequenced, observation of transposition into a position where the DNA sequence was previously known has not been reported.

Recently, such transposition events have been observed. It was found that short C-terminal deletions of the bvgA gene, a global regulator of virulence gene expression in B. pertussis, when present on an autonomously replicating plasmid in the presence of a functional chromosomal bvgAS locus, conferred a profound growth-inhibitory phenotype (6). Among mutations which relieved this growth inhibition were insertions of IS481 in the plasmid copy of bvgA and insertions of IS481 or IS1002 in the chromosomal bvgAS locus. These results are summarized in Fig. ​Fig.1.1.

Open in a separate window

FIG. 1

DNA sequence of the bvg locus prior to and following insertion of IS481 or IS1002. The isolation of these bvg alleles has been reported elsewhere (6). The identification of these elements as IS481 and IS1002 was based on DNA sequence analysis of approximately 150 bp of the ends, which shows them to be minor variants of the reported IS481 and IS1002 sequences (data not shown). The target site duplicated upon insertion is boxed. Brackets show the extent of the IS element, as interpreted here (solid lines) and as previously interpreted (dashed lines) (4, 7).

DNA sequence analysis of the sites of IS insertion in these mutants led to an unexpected observation. A 6-bp sequence originally present at the site of insertion in the bvg locus was duplicated at the junction points of bvg DNA and the DNA sequence of the IS. Since this sequence is precisely the sequence previously present at the site of insertion, the most parsimonious interpretation of these facts is that IS481 and IS1002 do create a duplication of 6 bp upon insertion. These sequences have previously been considered to be part of the IS elements themselves.

In light of these results, the DNA sequences at the junction point of previously reported examples of IS481 and IS1002 were reexamined. These sequences, presented in Fig. ​Fig.2,2, are consistent with a duplication upon insertion, in that the terminal 6 bp previously interpreted as part of the IS sequences form a perfect direct repeat in the majority of examples. The absence of a perfect repeat in several instances does not invalidate the hypothesis of duplication upon insertion because events subsequent to insertion could lead to a different 6-bp sequence flanking the two ends. Either deletion formation, which is a well-documented activity catalyzed by IS elements, or homologous recombination with a copy of the IS at a different location would be predicted to result in nonidentical sequences flanking the element.

Open in a separate window

FIG. 2

Alignment of the DNA sequence at the ends of previously reported occurrences of IS481 and IS1002. All sequences are presented with the IS element from left (L) to right (R), with respect to the originally reported sequences (4, 7), even if not published or submitted in this format to the GenBank database. In cases where the sequence of only one end of the IS has been reported, a prime denotes truncation of the sequence. A question mark denotes a lack of reported sequence information on the outermost nucleotide of the flanking hexanucleotide. Underlined bases denote differences between the flanking repeats. Note that insertions in the bvg locus are inverted relative to their representation in Fig. ​Fig.1.1. For consideration of the relative frequencies of site usage, flanking sequences were counted once if they were perfect repeats and separately if not. By this method, 22 flanking sequences were catalogued. Of these, 21 had the sequence NCTANN (NNTAGN on the opposite strand), 16 had the sequence NCTAGN, and 1 matched the consensus at only two nucleotides (NNTANN).

Based on a comparison of the 6-bp direct repeats flanking these insertion sites, the consensus sequence NCTANN (NNTAGN on the opposite strand) occurs 21 of 22 times. However, it is noteworthy that in 16 of 21 of these, the repeat sequence is the nearly palindromic NCTAGN, which matches this consensus both forwards and backwards. NCTAGN may be a preferred target site for insertion, which would explain its preponderance in this sampling. Based on a G+C composition of 67%, only 7 of 21 sequences matching the consensus NCTANN would be predicted to have the sequence NCTAGN. The near-palindromic nature of the duplicated insertion sites means that they are also nearly perfect inverted repeats; this explains the earlier inclusion of five of six nucleotides of the target site in the terminal inverted repeats of these IS elements (4, 7). The interpretation proposed here dictates that the reported size of the original examples of IS481 and IS1002 should be revised to be 1,043 and 1,030 bp, respectively, with flanking direct repeats of 23 and 24 bp, respectively. As a result, the terminal nucleotides defining the ends of these elements can be represented by 5′-TG…CA-3′, similar to other members of the IS3 family and consistent with their inclusion in this family based on the sequence of their predicted transposase proteins (3). These findings suggest that the mechanism of transposition of these elements involves a staggered, rather than a blunt, cleavage of the DNA at the target site.

The findings reported here also have implications which extend to issues of genomic variability. The restriction enzymes SpeI and XbaI (recognition sequences ACTAGT and TCTAGA, respectively) have been found to be useful in the analysis of genomic structure and organization in B. pertussis (1) because in B. pertussis, as in other organisms, these sites are much rarer than would be predicted based on the expected occurrence of a particular hexanucleotide in random nucleotide sequence. This is due to the fact that the CTAG tetranucleotide is dramatically underrepresented in bacterial DNA. This tetranucleotide is often found as the core of the consensus sequence for insertion as proposed above, which suggests that SpeI and XbaI sites would be preferred sites for insertion. This could affect the variability of SpeI or XbaI digestion patterns resulting from the duplication, or the deletion, of SpeI and XbaI sites resulting from IS481- and IS1002-catalyzed insertion or deletion formation. Indeed, it has previously been reported that isolates of B. pertussis from a whooping cough outbreak in Alberta, Canada, demonstrated remarkable variability in their XbaI digestion patterns as assessed by pulsed-field gel electrophoresis (1).