TEXT

Top Abstract Text References

The integrases form a diverse family of recombinases that mediate DNA rearrangements by means of conservative site-specific recombination reactions (1, 3). To date, more than 100 members of the integrase family, including the well-studied lambda integrase protein (14), have been identified. A comparison of their sequences reveals extended areas of similarity and several types of structural difference (10, 17). Integrase-mediated recombination reactions have been studied extensively. Their complexity varies considerably both in their requirement for accessory proteins and in the size of the DNA sites involved (27). To obtain more information about site-specific recombination in gram-positive bacteria, we investigated the site-specific integration system of the bacteriophage mv4. This temperate phage infects the gram-positive bacterium Lactobacillus delbrueckii and during lysogenization, the mv4 integrase catalyzes site-specific recombination between the phage attP site and the bacterial attB site. Recombination occurs within a 17-bp core sequence, common to attP and attB. At the attB site, the core overlaps the 3' end of a tRNA^Ser gene and the integrity of the tRNA gene is preserved after recombination (9). The integration vector pMC1, which contains the mv4 elements int and attP, integrates into the chromosome in a wide variety of gram-positive bacteria (5). The highly conserved nucleotides of the pMC1 chromosomal insertion sites in these various hosts form a 20-bp consensus sequence and are thought to be essential for attB function. We have shown elsewhere that the attP site has a complex structure with a minimal size of 234 bp (4).

In this study, an attP × attB recombination assay was developed in Escherichia coli to determine the minimal functional size of the attB site and to analyze the nucleotide requirements of this sequence.

The phage mv4 site-specific recombination elements are functional in E. coli. The integrase of phage mv4 promotes recombination in a wide variety of gram-positive bacteria (5). We therefore investigated whether this recombinase was also active in a gram-negative bacterium. A plasmid recombination assay was performed with E. coli (Fig. 1A). The L. delbrueckii subsp. bulgaricus attB site and the mv4 integrative elements int and attP were inserted, respectively, into two compatible plasmids, pAM239 and pRC1 (Table 1), which have no sequences in common. The recombinant plasmids, pAMattB and pMC1 (Table 1; Fig. 1A), were introduced into recombination-deficient E. coli TG2 (Table 1). Cells resistant to erythromycin (100 µg/ml) and spectinomycin (70 µg/ml) were grown for approximately 25 generations. Plasmids were then extracted, subjected to restriction digestion, and analyzed by agarose gel electrophoresis with ethidium bromide staining. Recombination between attP and attB was detected based on the formation of a cointegrate between the two plasmids, pMC1 (5.1 kb) and pAMattB (4.6 kb). In bacteria transformed with these two plasmids, a 9.7-kb DNA molecule was detected with a restriction pattern consistent with the map of the expected cointegrate (Fig. 1A): XbaI digestion generated two fragments of 1.6 and 8.1 kb (Fig. 1B, lane 5), and the cointegrate was linearized by EcoRV digestion (Fig. 2, lane 3). The low-copy-number plasmid pAMattB was not detected in these transformants, whereas the high-copy-number plasmid pMC1 was. This suggests that almost all the pAMattB molecules had recombined with the pMC1 molecules which were in excess. The formation of the cointegrate resulted from Int-mediated recombination because no recombinant product was detected if there was no attB site present in the test (Fig. 1B, lane 6) or if the int gene was defective (Fig. 1B, lane 7). The attL and attR sites were detected in the cointegrate molecule by PCR amplification with the primer sets Ba-Pb for attL and Pa-Bb for attR (Table 1 and Fig. 1A). Amplified products of the expected length (211 and 276 bp, respectively) were obtained. If the template used in the PCR reaction was plasmid DNA from strains containing the defective int gene or lacking the attB site or a mixture of the individually purified plasmids pAMattB and pMC1, no amplification was observed, demonstrating that the amplification of attL and attR was not an artifact of the PCR experiment.

View larger version (74K): [in this window] [in a new window]   FIG. 1.   attP × attB site-specific recombination test in E. coli. (A) Schematic representation of cointegrate formation between pMC1 and pAMattB. Restriction maps of the plasmids used in the test are shown. Pa and Bb, PCR primers for attR amplification; Ba and Pb, PCR primers for attL amplification. The arrows beside the attachment sites indicate the orientation of the homologous core sequences. (B) Electrophoresis in a 0.5% agarose gel (ethidium bromide stained) of XbaI-digested plasmid DNA isolated from E. coli TG2 transformed with pMC1 (int+ attP) (lane 1), pMC2 (int attP) (lane 2), pAM239 (lane 3), pAMattB (lane 4), pMC1 and pAMattB (lane 5), pMC1 and pAM239 (lane 6), and pMC2 and pAMattB (lane 7). L, 1-kb DNA ladder (Bethesda Research Laboratories). The photograph is a negative image. Arrows indicate the 1.6- and 8.1-kb XbaI fragments of the cointegrate.

View larger version (65K): [in this window] [in a new window]   FIG. 2.   Site-specific recombination in E. coli between the attP site and truncated or mutated attB sites. The attB sequences are presented in Table 2. Electrophoresis in a 0.5% agarose gel (ethidium bromide stained) of EcoRV-digested plasmid DNA isolated from E. coli TG2 transformed with pMC1 (int+ attP) (lane 1), pAM239 (lane 2), pMC1 and pAMattB (lane 3), pMC1 and pAMattB1 (lane 4), pMC1 and pAMattB2 (lane 5), pMC1 and pAMattB3 (lane 6), and pMC1 and pAMattB4 (lane 7) is shown. Plasmids pAM239, pAMattB, and pAMattB1 to -4 have no EcoRV site. Plasmid pMC1 and the cointegrate each contain one EcoRV site. L, 1-kb DNA ladder (Bethesda Research Laboratories). The photograph is a negative image. Arrows indicate the linear cointegrate (a), linear pMC1 (b), open circular form (c), and supercoiled circular form of pAMattB (d).

Determination of the minimal sequence required for attB function. To determine the minimum functional size of the attB site, synthetic attB sites of various lengths with either cohesive or blunt ends were produced by annealing two complementary oligonucleotides (Eurogentec). The DNA duplexes obtained were then inserted into pAM239 at a unique restriction site (EcoRI, BamHI, SalI, or SmaI). Insertion was checked by restriction digestion and DNA sequencing. The deletion derivatives of the attB site were then tested for their ability to recombine in vivo with the attP site located on plasmid pMC1 (int⁺ attP) (Table 2 and Fig. 2). Similar amounts of recombinant product were obtained with attB sites containing at least those residues between positions 10 and +5 (pAMattB, pAMattB1, and pAMattB2) (Table 2; Fig. 2, lanes 3 to 5). Removal of the nucleotide at position 10 resulted in significantly smaller amounts of cointegrate (pAMattB3) (Table 2; Fig. 2, lane 6) and no cointegrate was detected if further deletions were made, removing the nucleotides at positions 9 or +5 (pAMattB4 and pAMattB5) (Table 2; Fig. 2, lane 7). The minimal attB fragment required for maximal activity was a 16-bp sequence (pAMattB2) (Table 2). This minimal site is included within the consensus sequence deduced from the pMC1 chromosomal insertion sites present in various hosts (5) (Table 2, bottom row) and overlaps the 5' end of the core.

Mutational analysis of the minimal attB site. Conserved nucleotides in the pMC1 insertion sites' consensus sequence (5) were mutated within the 16-bp minimal attB sequence. In the case of genetic elements carrying sequences homologous to the 3' ends of genes into which they insert, the Campbell model predicts that the strand exchange reaction occurs in the 5' portion of the core (7). Concerning the phage mv4, this prediction is consistent with previous results indicating that the first three nucleotides at the 5' extremity of the core, CCT, are important for the strand exchange reaction (5). Independent changes of two of these three nucleotides, at positions 8 and 6 (pAMattB9 and pAMattB8, respectively) (Table 2), abolished attB activity, confirming the predicted involvement of these nucleotides in attB function. Changes of other conserved nucleotides, at positions +1 and +3 (pAMattB7 and pAMattB6, respectively) (Table 2), also abolished recombination activity. These nucleotides may be involved in the recognition of the attB site by the integrase. In contrast, mutation of a nonconserved nucleotide at position +4 (pAMattB10) (Table 2) did not affect recombination.

Comparison of the mv4 attB site with other well-characterized attB sites. In all integrative recombination systems characterized, the attP and attB target sites differ greatly in length and structure (12-14, 16, 18, 19, 26). The attB site is a short DNA sequence (less than 30 bp) corresponding to the crossover region at which strand exchange takes place (7). Two imperfect inverted repeats that bind to the integrase surround a 7-bp overlap region delimited by the scattered cuts made by the recombinase. The attP site is longer (more than 200 bp) and has a more complex structure. The crossover region of attP is flanked by two arms containing additional integrase-binding sites. A higher-order structure, called the intasome, is formed on attP and is required for synapsis with the naked attB site (23). As in other site-specific integration systems, the mv4 recombination target sites are dissimilar. The attP site consists of a 234-bp DNA fragment containing five scattered putative Int-binding sites (4) whereas the attB site requires only a 16-bp sequence to be functional. The mv4 minimal attB site is the shortest attB sequence described to date. It extends beyond the 5' boundary of the core (Table 2, top row). This property is shared by four other elements that insert their DNA into a tRNA gene (12, 19, 21, 26) (Fig. 3A). It has been suggested that the frequent use of tRNA genes as insertion sites is due to the presence of dyad symmetry elements and that mobile genetic elements may use the symmetry of the anticodon stem-loop structure to form the appropriately spaced inverted integrase binding sites in attB (7, 13, 22). In phages HP1 (13), P22 (26), and L5 (19), the core sequence includes the anticodon loop sequence, which coincides with the overlap region, and this is also presumably the case for the pSAM2 integrative element (21) (Fig. 3A). This model does not apply to phage mv4 and several other elements (2, 6, 11, 20, 22, 25, 28) because the core sequence does not overlap either the anticodon loop or the TC loop (Fig. 3B). For these elements, it is unclear why the integration target is a tRNA gene.

View larger version (28K): [in this window] [in a new window]   FIG. 3.   DNA sequences of tRNA genes used as insertion sites by genetic elements. (A) Insertion sites for which the core sequence overlaps the anticodon loop sequence. The names of the genetic elements and tRNA genes into which they integrate are indicated on the left. The tRNA sequences and their 3' flanking regions are shown in uppercase and lowercase letters, respectively. The core sequences are boxed and the boundaries of the minimal sequences required for attB function are indicated by brackets. The overlap regions, when determined experimentally, are shown in boldface type. (B) Insertion sites for which the core sequence is located downstream from the anticodon loop sequence.

Nucleotide sequence accession number. The nucleotide sequence of the mv4 integration region has been deposited in GenBank under accession no. U15564.