Present address: Department of Microbiology, Division of Clinical Integrative Biology, Sunnybrook and Women's Research Institute, 2075 Bayview Avenue, S1–26 A, Toronto, ON, Canada, M4N 3 N5.
Delineation of the recombination sites necessary for integration of pathogenicity islands II and III into the Escherichia coli 536 chromosome
Article first published online: 26 FEB 2008
© 2008 The Authors
Volume 68, Issue 1, pages 139–151, April 2008
How to Cite
Wilde, C., Mazel, D., Hochhut, B., Middendorf, B., Le Roux, F., Carniel, E., Dobrindt, U. and Hacker, J. (2008), Delineation of the recombination sites necessary for integration of pathogenicity islands II and III into the Escherichia coli 536 chromosome. Molecular Microbiology, 68: 139–151. doi: 10.1111/j.1365-2958.2008.06145.x
- Issue published online: 26 FEB 2008
- Article first published online: 26 FEB 2008
- Accepted 24 January, 2008.
- Top of page
- Experimental procedures
- Supporting Information
In uropathogenic Escherichia coli strain 536, six pathogenicity islands (PAIs) encode key virulence factors. All PAIs except PAI IV536 are flanked by direct repeats and four of them encode integrases responsible for their chromosomal excision. To study recombination sites used for the integration by PAI II536 and III536 integrases, we measured site-specific recombination between the chromosomal integration site attB, and the PAI-specific attachment site attP. We show that PAI III536 IntB, but not IntA, mediates PAI III536 integration. Studies of integrative recombination sites of both PAIs show that, when using a large cognate attP site (839 bp for PAI II536 and 268 bp for PAI III536), PAI II536 and III536 attB sites could be reduced to 16 bp and 20 bp, respectively, without affecting recombination. Further reduction to 14 bp for PAI II536 and 13 bp for PAI III536 diminished recombination efficiency. Surprisingly, attP sites could also be reduced to 14 bp (PAI II536) and 20 bp (PAI III536). The integration host factor (IHF) and the DNA-bending HU protein do not influence PAI II536 recombination, but IHF enhances PAI-III536 excision and negatively affects its integration. These data suggest that PAI intasomes differ from those of lambda and P4 integrase paradigms.
- Top of page
- Experimental procedures
- Supporting Information
The archetypal uropathogenic Escherichia coli (UPEC) strain 536 (O6:K15:H31) produces a variety of virulence factors (Schubert et al., 1998; Dobrindt and Hacker, 2001; Schneider et al., 2004; Beloin et al., 2006, reviewed in Johnson, 1991), most of which are encoded in well-characterized large chromosomal regions termed pathogenicity islands (PAIs) (defined in Blum et al., 1994, and extensively reviewed in Hacker and Kaper, 1999; 2000; Hacker, 2002; Hochhut et al. 2005).
Except for PAI IV536, all E. coli 536 islands are flanked by direct repeats (DRs). These DRs resemble the left and right end junctions (attL and attR) that result from the integration of a phage genome into the chromosome. Therefore, we speculated that PAIs were acquired by a mechanism similar to the integration mechanisms of bacteriophages (Hacker and Kaper, 2000; Dobrindt et al., 2004). PAI I, II, III and V of E. coli strain 536 can excise from the chromosome by site-specific recombination between the flanking DR structures whereas PAI IV536 and VI536 are extremely stable (Middendorf et al., 2001; 2004; Hochhut et al., 2006). A PAI VI536 homologue in UPEC strain CFT073, however, has been reported to be unstable (Antonenka et al., 2006). The reason for the different behaviour of this island in both UPEC strains, despite high overall DNA sequence similarity, is unclear. Except PAI III536, whose functional integrase gene intB is at the opposite extremity of the PAI, all PAI-borne integrase genes are adjacent to the associated tRNA-gene. The encoded integrases are responsible for the site-specific excision of their cognate PAI from the chromosome (Hochhut et al., 2006). Interestingly, the integrase from PAI II536 could excise PAI V536 by mediating site-specific recombination between PAI V536 att sites.
The PAI-encoded integrases of strain 536 belong to the tyrosine recombinase family, and are all related to the P4 coliphage integrase, except for IntB, which contains a conserved IntP4 domain and only exhibits a weak similarity (25–33%) to phage-like integrases of different bacterial species. Recombination sites and integrase structures of several members of the tyrosine recombinase family, such as lambda, HP1, L5, P2, Cre or Flp, have been characterized. They mediate a site-specific recombination reaction between the recombination site of the mobile element (attP) and the integration site on the chromosome (attB), generating two hybrid sites, attL and attR. In the case of Cre and Flp, the identical recombination sites consist of an 8 bp overlap region where the crossing over occurs, surrounded by 13 bp inverted repeats called core sites (CSs), recognized by the recombinases (van Duyne, 2001). For other tyrosine recombinases, such as lambda, the attB site consists of a short core sequence encompassing an overlap region of 6–8 bp, flanked by two CSs of 9–13 bp. The attP site is generally more complex, as the core sequence can be surrounded by accessory arm sites to which the integrase and accessory proteins involved in integration and/or excision, such as integration host factor (IHF), Xis and Fis for the λ phage (Craig and Nash, 1984; Yin et al., 1985), can bind. The recombination reaction is stepwise: the integrase and accessory cofactors, if any, recognize the attP site, forming the intasome, which then captures the attB site (Patsey and Bruist, 1995). The strand exchanges occur after formation and resolution of a Holliday junction (Hsu and Landy, 1984; Ghosh et al., 2005).
The mechanisms of the site-specific recombination mediated by PAI-associated integrases remain largely unknown. The PAIs of E. coli 536 display different stabilities (Middendorf et al., 2004). However, the overall stability of an integrative element is conditioned by both its integration and its excision frequencies, which could themselves be driven by different factors, like accessory proteins, which can affect the directionality of the reactions. To study the functional basis of this instability, we used a co-integration assay. We determined the minimal (i.e. smallest DNA region still allowing site-specific recombination) and optimal (i.e. DNA region allowing site-specific recombination with an optimal frequency) sizes of both the attB and attP recombination sites associated with PAI II536 and PAI III536. We also addressed the role of the accessory proteins IHF and HU in the excision and integration reactions. Our analyses suggest that the accessory factor IHF can affect excision and integration of PAI III536. Furthermore, the attP and attB recombination sites of PAI II536 and PAI III536 differ structurally from the defined canonical phage sites, and the intasome complex architecture in both elements differs from the P4 and lambda recombinase models.
- Top of page
- Experimental procedures
- Supporting Information
Characteristics of the regions surrounding PAI II536 and PAI III536 recombination sites
The largest DNA regions encompassing PAI II536 and PAI III536 recombination sites used to set up the two-plasmid system are shown in Fig. 1. Flanking DRs correspond to the left and right boundary sequences of the islands integrated into the host bacterial chromosome (similar to prophage genomes flanked by attL and attR). To determine whether obvious accessory proteins such as IHF are required for the recombination reaction, we searched for putative IHF binding sites using the online promscan software (http://molbiol-tools.ca/promscan/). Whereas only one putative IHF binding site was predicted (with a low probability) in the PAI II536 attP region (attPII), the DR on PAI III536 attP region (attPIII) is surrounded by three putative IHF binding sites. Similarly, each attB region contains only one putative IHF binding site (with a low-predicted probability), located on the right side of the DR.
Role of PAI II536- and PAI III536-encoded integrases in the integration reaction
IntII, the only PAI II536-encoded integrase, had previously been shown to be responsible for its excision (Hochhut et al., 2006). Using the two-plasmid system, we found that IntII successfully mediated site-specific recombination between a 839 bp attPII site and a 493 bp attBII site with a recombination frequency of (2.5 ± 0.7) × 10−2. IntII is thus responsible for both the excision and the integration of PAI II536 into the bacterial chromosome.
Pathogenicity island III536 has the distinctive feature of carrying two integrase genes, intA, encoding an SfX-like integrase and intB, encoding an integrase with weak similarity to integrases from other bacteria. Only IntB is required for site-specific excision of this island (Hochhut et al., 2006). Both PAI III536 integrases were tested with the two-plasmid system, using a 268 bp attPIII site and a 544 bp attBIII site. However, IntA did not mediate any detectable site-specific recombination event, whereas IntB mediated site-specific recombination with a frequency of (1.2 ± 0.84) × 10−2.
Delineation of PAI II536 recombination sites
Determination of PAI II536attB site. When the two-plasmid system was used with an 839 bp attPII and a 493 bp attBII region (Fig. 1), the recombination frequency [(2.5 ± 0.7) × 10−2] was not affected when the size of the attBII region was reduced to that of the 18 bp DR (Fig. 2). However, site-specific recombination was severely impaired when one additional base pair was removed from the right side and was completely inhibited after deletion of two base pairs (Fig. 2). On the left side, the recombination frequency was conserved when up to three base pairs were removed, dropped significantly when four base pairs were deleted, and became undetectable after a 5 bp deletion. Consequently, the minimal size for a fully functional attBII is 16 bp and its sequence is T3CGAGTCCGGCCTTCG18, although a slightly shorter 14 bp sequence (C4GAGTCCGGCCTTC17) could still mediate recombination, but with a 1000-fold lower efficiency. Symmetric sequences could not be detected in the above-defined attBII, suggesting that like attBIII, the CS and inverse core site (ICS) sequences are degenerate. It has also been published that, in contrast to the symmetry-based lambda model, asymmetry is a characteristic of the minimal form of attB for P4-like mobile elements model (Williams, 2002). The defined attBII site does not carry a predicted IHF binding site.
Analysis of the PAI II536attP site. The attPII size was determined by using the maximal attB size (493 bp) and by progressively decreasing the size of the attP region. The progressive reduction of attPII from 839 to 48 bp (DR flanked by 15 bp on each side) did not affect the frequency of site-specific recombination (Fig. 3). Further reduction of attPII to the size of the DR (18 bp) or to 16 bp (by the removal of 2 bp on the 5′ side of the DR) resulted in a moderate drop (10-fold) in the recombination frequencies, while shortening it by two more base pairs (one on each side of the 16 bp fragment) led to a 1000-fold reduction of the recombination frequency (Fig. 3). Therefore, the size of attPII necessary for an optimal recombination frequency is between 19 and 48 bp, which is larger than the 16 bp defined for attBII.
Finally, when both attPII and attBII were reduced to the sizes defined above (48 and 16 bp respectively), we observed a 100-fold reduction in the recombination frequency [(2.0 ± 1.7) × 10−4], as compared with the frequency obtained with the larger version of both att sites (attBII-493 bp × attPII-839 bp).
Comparison of the PAI II536 and PAI V536 recombination regions. IntII has previously been shown to mediate site-specific recombination not only between its cognates PAI II536 sites, but also between the PAI V536 recombination sites (Hochhut et al., 2006). To analyse this cross-talk phenomenon further and to identify features shared by both att sites, we compared the PAI II536 recombination sites defined here with the regions encompassing the PAI V536 recombination sites and with the prophage P4 att sites, as both integrases are P4-related (Fig. 4). Although PAI V536 DR contains a 1 bp deletion in its attR site, compared with the attL sequence (Schneider et al., 2004), the DR has been defined as a 23 bp sequence present in the attPV and attBV constructs used to study the cross-talk (Hochhut et al., 2006). Alignments of the attB and attP sites within each PAI show that the homology is restricted to the DR itself (Fig. 4a). Most interestingly, alignments of the attB and attP sites from the two PAIs delineated a 9 bp sequence within the DR that was completely identical in the four att sites. An identical sequence was also found in the P4 attB CS and a sequence with 1 nt difference was present in the P4 attP CS (Fig. 4b). Taking into account the variable nucleotides flanking each side of this 9 bp core, it is possible to delineate a 13 nt sequence T(C/T)CGAGTCCGG(G/C)C that might define the region recognized by IntII. This hypothesis is reinforced by the fact that this core sequence is absent from the other E. coli 536 PAIs. Furthermore, sequences surrounding IntII and Int P4 cores (the integrases of which are 91% identical), exhibited further similarities, in contrast to those surrounding PAI II536/PAI V536 and PAI V536/P4 cores (the integrases of which share 67% and 63% sequence identity respectively).
Delineation of PAI III536 recombination sites
Determination of PAI III536attB site. The recombination frequency determined using the two-plasmid system with a 268 bp attPIII region and a 544 bp attBIII region (Fig. 1) was (1.2 ± 0.84) × 10−2 (Fig. 5). We progressively reduced the size of the region encompassing attBIII, while keeping the 268 bp attPIII region. In this way, the attBIII region could be reduced to the 48 bp DR without significantly affecting the site-specific recombination frequency. Thus, to determine within the DR the minimal region that still allowed a fully efficient recombination, the size of the DR was progressively reduced on its right side (Fig. 5). Interestingly, deletions up to 25 bp did not affect the recombination frequencies. These frequencies started to decrease when 28 base pairs were removed, and dropped dramatically after removal of 32 base pairs (Fig. 5). Thus, the right side of a fully functional attBIII site lies within the T17CGT20 sequence of the DR. The construct containing the smallest entirely functional recombination site (pDRΔ25R) was then used to determine the left-hand limit of the attBIII site. While removal of one base pair did not affect the recombination frequencies, they dropped dramatically when two or more nucleotides were removed (Fig. 5). Therefore, the fully functional attBIII site is between 16 and 19 bp long, and its sequence is T2TCGTAATGCGAAGGT(CGT20).
The identified attBIII sequence displays a symmetrical structure with two 4 bp inverted repeats (T2TCG5 and C11GAA14) flanking a central region of 5 bp, which might define the putative CS and ICS. However, the CS and ICS sequences are most likely degenerate on their external sides as, according to the literature (van Duyne, 2002), the overlapping central region is generally of 6–8 bp, and the CS and ICS, which ideally have a size of a helix turn (11–13 bp), do not form perfect inverted repeats.
Analysis of the PAI III536attP site. To delimit attPIII, the initial 268 bp attPIII region was progressively shortened on each side of the DR, while keeping the initial size of attBIII (544 bp). Recombination frequencies were not affected when the size of the region was reduced to 155 bp, with the DR centrally located, and then to the size of the 48 bp DR (Fig. 6). However, according to the minimal and ideal sizes that were determined for attBIII, reducing the attPIII region to 20 bp caused the recombination frequency to drop significantly, and it became undetectable with a 13 bp attPIII site (Fig. 6). This indicates that the size of a fully functional attPIII site is between 21 and 48 bp, and, thus, slightly larger than that of the attBIII site defined above. Furthermore, the fact that the recombination frequency of the 48 bp attP derivative, which does not carry any complete putative IHF binding site, was not reduced, suggests that IHF is not essential for PAI III536 integration.
Finally, having defined independently the smallest fully functional attBIII and attPIII sites (48 and 20 bp respectively), we determined the recombination frequency obtained when combining these two sites. Under these conditions, a 1000-fold decrease in the recombination frequency [(4.7 ± 4.6) × 10−5] was observed, as compared with the frequency obtained with the larger version of both att sites (attBIII-544 bp × attPIII-268 bp).
Integration host factor and HU do not enhance PAI II536 and PAI III536 integration reaction
Beside integrases, accessory host-encoded factors are often necessary for efficient site-specific recombination. To examine the role of the DNA-bending proteins IHF and HU in the integration reaction, the two-plasmid system was tested in E. coli MC240Pir and its himA derivative (MC251Pir) as well as in E. coli strain C600Pir and its hupAB-negative variant OPH252Pir. Plasmid constructs containing the initial large attP and attB sites of PAI II536 or PAI III536, including the predicted IHF binding sites (Fig. 1), were introduced into these strains and recombination frequencies were determined. For both PAI II536 and PAI III536, similar recombination frequencies were obtained in the presence or absence of HU (Table 1). Unexpectedly, a reproducible and significantly (P = 0.045) higher recombination frequency of PAI III536 was observed in the himA mutant. This suggests a negative effect of IHF on PAI III536 integration. To test this hypothesis further, we used the two-plasmid system with plasmid constructs lacking the putative IHF binding sites, i.e. attPII48 × attBII18 for PAI II536, and attPIII48 × attBIII23 for PAI III536. As described above, the combined small sites led to low recombination frequencies but, interestingly, these frequencies were similar in the himA+ and himA– backgrounds (Table 1). These results support the hypothesis of a negative effect of IHF on PAI III536 integration. Furthermore, to eliminate the possibility of an indirect effect of IHF on the integrase expression, we ascertained by reverse-transcription polymerase chain reaction (RT-PCR) that the integrase genes were transcribed to the same extent from the multicopy plasmid in both the himA+ and himA– backgrounds (data not shown).
|Strain||Phenotype||PAI||Size of the attP region (bp)||Size of the attB region (bp)||Recombination frequency||Standard deviation|
|MC240Pir||himA+||PAI II536||839||493||4.7 × 10−1||4.0 × 10−1|
|MC251Pir||himA–||PAI II536||839||493||8.8 × 10−1||5.8 × 10−1|
|MC240Pir||himA+||PAI III536||268||544||4.5 × 10−2||2.1 × 10−2|
|MC251Pir||himA–||PAI III536||268||544||3.3 × 10−1||2.0 × 10−1|
|MC240Pir||himA+||PAI II536||48||18||3.5 × 10−4||1.6 × 10−4|
|MC251Pir||himA–||PAI II536||48||18||1.9 × 10−4||1.3 × 10−4|
|MC240Pir||himA+||PAI III536||48||18||5.1 × 10−4||3.6 × 10−4|
|MC251Pir||himA–||PAI III536||48||23||4.5 × 10−4||2.0 × 10−4|
|C600Pir||hupAB+||PAI II536||839||493||7.8 × 10−1||1.2 × 10−1|
|OPH252Pir||hupAB–||PAI II536||839||493||8.5 × 10−1||8.2 × 10−2|
|C600Pir||hupAB+||PAI III536||268||544||4.3 × 10−2||1.4 × 10−2|
|OPH252Pir||hupAB–||PAI III536||268||544||2.1 × 10−2||1.6 × 10−2|
Integration host factor plays a role in PAI III536 but not in PAI II536 excision
We evaluated the role of IHF in PAI II536 and III536 excision using the island-probing method as described previously (Rajakumar et al., 1997; Middendorf et al., 2001; 2004) in the himD-negative strains 536-BH42 and 536-BH43 respectively (Table 3). Excision of the complete PAI II536 was detected with an unaltered frequency in the himD mutant, and in PAI III536, the overall deletion rate was almost unchanged (Table 2). However, as previously described (Middendorf et al., 2004), PAI III536 exhibits two types of deletion: 40% of the excision events are due to the integrase-mediated site-specific excision of PAI III536 (type I deletion), and 60% are due to a homologous recombination between two copies of IS100 located within the PAI (type II deletion). In the himD– background, polymerase chain reaction (PCR) analyses revealed that the majority of sucrose-resistant colonies had undergone a type II deletion, and that the slight decrease in frequency was mostly due to a threefold drop of the type I deletion frequency, as compared with the wild type (Table 2). Furthermore, when the himD mutation was introduced into a recA derivative, deletion of PAI III536 became undetectable. Interestingly, the ability to excise through a type I deletion mechanism from the chromosome was restored when intB from PAI III536 was introduced in trans on plasmid pINTB3 (Table 2). Thus, although IHF is not essential for PAI III536 excision by site-specific recombination, the presence of a functional IHF enhances the efficiency of the reaction.
|E. coli strains||Main features||Reference|
|E. coli 536||Uropathogenic E. coli strain||Berger et al. (1982)|
|536-BH26||PAI I536::sacBΔintPAI I::cat||Hochhut et al. (2006)|
|536-BH13||PAI II536::sacBΔintPAI II::cat||Hochhut et al. (2006)|
|536-PAI II::sacB||PAI II536 tagged with sacB||Middendorf et al. (2001)|
|536-PAI III::sacB||PAI III536 tagged with sacB||Middendorf et al. (2001)|
|536-BH42||PAI II536::sacBΔhimD::cat||This study|
|536-BH43||PAI III536::sacBΔhimD::cat||This study|
|536-BH83||ΔrecA PAI III536::sacBΔhimD::cat||This study|
|DH5α||F-ϕ80ΔlacZΔM15 Δ(lacZYA-argF)U169 deoR, recA1 endA1 hsdR17(rk- mk +) phoA supE44 λ- thi-1 gyrA96 relA1||Bethesda Research Laboratories|
|β2163||(F–) RP4-2-Tc::Mu ΔdapA::(erm-pir)||Demarre et al. (2005)|
|TOP10||F–mcrAΔ(mrr-hsdRMS-mcrBC) f80lacZΔM15 ΔlacX74 deoR recA1 araD139 Δ(ara-leu)7697 galU galK rpsL (StrR) endA1 nupG||Invitrogen|
|MG 1655||E. coli K-12 wild-type||Guyer et al. (1981)|
|MC240||araΔ(lac-pro) nalA metB arg Eam rif thi supF||Gamas et al. (1986)|
|MC251||MC240 himA(Δ82::Tn10)||Gamas et al. (1986)|
|MC240Pir||MC240 ΔthyA::(erm-pir116) [EmR]||This study|
|MC521Pir||MC251 ΔthyA::(erm-pir116) [EmR]||This study|
|C600||thr leuB6 lacYI thi tonA2l supE44||Appleyard (1954)|
|OPH252||thr leuB6 lacYI thi tonA2l supE44 hupB::KmRhupA::CmR||Huisman et al. (1989)|
|C600Pir||C600 ΔthyA::(erm-pir116) [EmR]||This study|
|Strain||Relevant genotype||Deletion rate (×10−5)||PAI III536 deletion rate (%)|
|Type I||Type II|
|536 PAI II::sacB||PAI II536 tagged with sacB||2.7|
|536 PAI III::sacB||PAI III536 tagged with sacB||0.85||40||60|
|536-BH83||PAI III536::sacBΔrecAΔhimD::cat||< 0.1||0||0|
|536-BH83/pINTB3||PAI III536::sacBΔrecAΔhimD::cat/intBPAI III+||0.6||100||0|
- Top of page
- Experimental procedures
- Supporting Information
In this study, we characterized the recombination sites of E. coli 536 PAI II and PAI III in order to better understand the mechanisms of PAI integration. Together with our previous work (Hochhut et al., 2006), the data show that PAIs encode specific integrases that mediate both excision and integration of their cognate PAI by site-specific recombination.
Although genetic instability has been described for a number of PAIs, very little is known about their integration/excision mechanisms and their recombination sites. Excision from the chromosome occurs in the form of a transient circular intermediate, a mechanism resembling that of phage lambda (van Duyne, 2005). Therefore, we hypothesized that PAI attP sites could be larger than the attB sites and contain sites recognized by accessory proteins, such as IHF and HU. We demonstrated that an attBIII site of 20 bp is fully functional with a large 268 bp attPIII site, and that a minimal size of 13 bp could still allow some recombination. attBII could be reduced to 16 bp when used with a large 839 bp attPII site, without affecting the site-specific recombination frequency, and it kept some functionality when it was reduced to 14 bp. IntII can mediate deletion of both PAI II536 and PAI V536 and, interestingly, the comparison of the attII and attV sites reveals a 13 bp conserved sequence between the two sites that is absent from the other E. coli 536 PAIs and could be a consensus sequence recognized by PAI II536 integrase. Furthermore, as the PAI II536 att sites are very similar to their P4 phage homologues (Fig. 4), it would be interesting to determine whether the P4 integrase can mediate attPII × attBII recombination. Clear symmetry is not visible in the different sites, suggesting either that the CS and ICS used for the integrase binding and to assemble the recombinase tetramers complex are degenerate, or that other, still not identified, recognition structures serve for the assembly of the intasome. Interestingly, these results also show that DR-III size is actually much larger than the attBIII site itself. However, PAI III536 is inserted into the thrW gene and the 48 bp DR-III encompasses the complete 3′ end of the thrW gene (Dobrindt et al., 2002). Accordingly, this long length of DR-III is only important for the reconstitution of the thrW tRNA gene after PAI III536 excision and integration.
Site-specific recombination events were still detected when each attP site was reduced to a fragment as small as 20 bp for attPIII and 14 bp for attPII, and therefore devoid of IHF binding sites. As the accessory factor IHF and the related DNA-bending protein HU may be involved in site-specific recombination, we investigated their role in site-specific recombination reactions of PAI II536 and III536. HU has no effect on PAI II536 and PAI III536 integration. IHF, however, plays a role in the excision of PAI III536 (and PAI I536, data not shown), but not of PAI II536, indicating that the E. coli 536 PAIs differ in their requirement for IHF as an excision cofactor. This role is probably indirect and could be linked to a positive control of the intB promoter, only visible with intB in single copy in its original context, while overexpression from a multicopy plasmid relieves the IHF control because IntB is no longer limiting (Table 2). However, intB overexpression only restores the basal excision level (10−5) observed in the IHF+ background from the expression of the unique intB chromosomal copy. An excisionase may be involved in the excision, as such a requirement could explain the low excision rate measured in its absence. Indeed, PAI II536 and P4 integrases are 91% identical, and P4 has been shown to require the Vis excisionase (Cali et al., 2004; Piazzolla et al., 2006). Furthermore, excision of other PAIs can be stimulated by excisionases (Lesic et al., 2004; Luck et al., 2004). As excisionases, like recombination directionality factors, contain few, if any, highly conserved residues, it is difficult to identify them by sequence comparison. There is only one putative excisionase gene in the E. coli 536 genome, adjacent to intA on PAI III536. However, the deletion of this gene did not affect PAI III536 excision (B. Hochhut and G. Balling, unpubl. data). More interestingly, IHF is not required for the integration of PAI II536 but, unexpectedly, it impairs PAI III536 (and also PAI I536, data not shown) integration. Several studies have shown that IHF can sometimes enhance (Leong et al., 1984; Lee et al., 1991; van Rijn et al., 1991; Frumerie et al., 2005) but not always improve (Cho et al., 1999) integration. To our knowledge, we present the first example of a negative effect of IHF on the integration. We hypothesize that increasing PAI III536 (and PAI I536) excision and preventing its integration might play a role during the infection or adaptation to environmental changes.
The highest integration frequencies exhibited by both PAI II536 and III536 were about 10−2. However, intriguingly, the recombination frequencies differed depending on whether we reduced both attP and attB sites together, or we kept one large att site while progressively reducing the other site. This suggests that both recombination sites probably carry protein binding sites recognized either by the integrase itself or by accessory proteins. Thus, to have a fully functional intasome leading to optimal recombination, one of the two recombination sites must be complete. It is also possible that the topology of the DNA interferes with the recombination reaction, or that a cofactor playing a role in the integration reaction recognizes the att sites indifferently. Another puzzling observation is that even when intB is overexpressed from a multicopy plasmid, the PAI III536 deletion frequency is only of 10−5 (Table 2). This could be an indication that another partner involved in the site-specific recombination is lacking.
Thus, in contrast to the lambda paradigm, PAIs have much smaller attP sites (of a size similar to the attB sites), suggesting that their site-specific recombination complexes differ from that of lambda. This is also supported by a higher similarity of the IntII and IntB domain structures relative to that of the lambda integrase (data not shown). Moreover, IHF, which is not essential but increases PAI III536 excision, has a surprising negative role in PAI III536 integration. Therefore, the E. coli 536 PAI recombination reactions seem simpler than that of lambda, and closer to the P4 recombination reaction, although no excisionase has been identified. The absence of specific accessory proteins might have been selected to facilitate the dissemination of these ‘genetic parasites’ among only remotely related bacterial species.
- Top of page
- Experimental procedures
- Supporting Information
Media and bacterial strains
Luria broth (LB), used to grow all strains, was supplemented when necessary with diaminopimelic acid to a final concentration of 0.3 mM. For solid media, agar was added to a final concentration of 1.5% (wt/vol). Antibiotics were used at the following concentrations: ampicillin (Ap) 100 μg ml−1; chloramphenicol (Cm) 20 μg ml−1; kanamycin (Km) 50 μg ml−1; spectinomycin (Sp) 100 μg ml−1. Counter selective media contained 7% (wt/vol) sucrose. Strains used and constructed for this study are listed in Table 3.
Plasmids constructed for this study are listed in Table 4. All constructs were checked by sequencing.
|Plasmid name||Main features||Reference|
|pGEM-T||bla, T/A cloning vector||Promega|
|pGEM-T easy||bla, T/A cloning vector||Promega|
|pSW23T||cat, R6K origin of replication, oriTRP4||(Demarre et al., 2005)|
|pPAI III-CI||pGEM-T easy + attPIII||(Middendorf et al., 2004)|
|pPAI II-CI||pGEM-T easy + attPII||(Middendorf et al., 2004)|
|pINTB3||pASK75 + intB||This study|
|pSW25T||spec, R6K origin of replication, oriTRP4||(Demarre et al., 2005)|
|Plasmid name||Size of the fragment containing the att site (bp)||Primers used for att amplification and for the integrase amplification|
|pattBIII-544||544||ΔPAI III up and ΔPAI III lp|
|pattBIIISp||544||ΔPAI III up and ΔPAI III lp|
|pattBIII-343||343||III-attB1 and III-attB2|
|pDR3||48||DR-III+ and DR-III-new|
|pDR3Δ5R||43||DR3delta5R-lacZ and lacZ-2|
|pDR3Δ10R||38||DR3delta10R-lacZ and lacZ-2|
|pDR3Δ15R||33||DR3delta15R-lacZ and lacZ-2|
|pDR3Δ20R||28||DR3delta20R-lacZ and lacZ-2|
|pDR3Δ25R||23||DR3delta25R-lacZ and lacZ-2|
|pDR3Δ28R||20||DR3delta28R-lacZ and lacZ-2|
|pDR3Δ32R||16||DR3delta32R-lacZ and lacZ-2|
|pDR3Δ33R||15||DR3delta33R-lacZ and lacZ-2|
|pDR3Δ34R||14||DR3delta34R-lacZ and lacZ-2|
|pDR3Δ35R||13||DR3delta35R-lacZ and lacZ-2|
|pDR3Δ5 l||43||DR3delta5L-lacZ and lacZ-2|
|pDR3Δ25R4L||19||DR3delta25R4L-lacZ and lacZ-2|
|pDR3Δ25R3L||20||DR3delta25R3L-lacZ and lacZ-2|
|pDR3Δ25R2L||21||DR3delta25R2L-lacZ and lacZ-2|
|pDR3Δ25R1L||22||DR3delta25R1L-lacZ and lacZ-2|
|pDR3Δ34R1L||13||DR3delta34R1L-lacZ and lacZ-2|
|pattBII||493||M803b and M805c|
|pattBIISp||493||M803b and M805c|
|pDR2||18||DR2-lacZ and lacZ-2|
|pDR2Δ3R||15||DR2delta3R-lacZbis and lacZ-2|
|pDR2Δ2R||16||DR2delta2R-lacZ and lacZ-2|
|pDR2Δ1R||17||DR2delta1R-lacZ and lacZ-2|
|pDR2Δ5 l||13||DR2delta5LA-lacZ and lacZ-2|
|pDR2Δ4 l||14||DR2delta4L-lacZ and lacZ-2|
|pDR2Δ3 l||15||DR2delta3LA-lacZ and lacZ-2|
|pDR2Δ2 l||16||DR2delta2LA-lacZ and lacZ-2|
|pDR2Δ1 l||17||DR2delta1LA-lacZ and lacZ-2|
|pDR2Δ3L1R||14||DR2delta3L1RA and lacZ-2|
|pintB||–||IntPAI3-F and IntPAI3-R|
|pKintB attPIII-268||268||attP1 and attP2|
|pKintB attPIII-155||155||III-attP5 and attP2|
|pKintB attPIII-48||48||lacZ-DR3-SacII and lacZ-2SphI|
|pKintB attPIII-20||20||attP3-20-Kmfw-SacII and Kmrev-SacII|
|pKintB attPIII-13||13||attP3-13-Kmfw-SacII and Kmrev-SacII|
|pintA||–||Int-sfx and xis3|
|pKintA attPIII-268||268||attP1 and attP2|
|pintII||–||II-intF and II-intR|
|pKintII attPII||839||II-attPup and IIattPlp|
|pKintII attPII-253||253||attP2-up and attP2-down|
|pKintII attPII-48||48||lacZ-attP2-48-SacII and lacZ-2SphI|
|pKintII attP-DR||18||attP2-DR-Kmfw-SacII and Kmrev-SacII|
|pKintII attP-16||16||attP2-16-Kmfw-SacII and Kmrev-SacII|
|pKintII attP-14||14||attP2-14-Kmfw-SacII and Kmrev-SacII|
Constructions of the plasmids containing the attB sites. All fragments containing the attB sites of different sizes were cloned into the EcoRI site of pSW23T (Demarre et al., 2005). Plasmid pattBII has been previously described (Hochhut et al., 2006). To obtain the non-interrupted attBIII site on the chromosome of E. coli 536, deletion of PAI III536 was performed as previously described: 100 μl of 10−2 and 10−3 dilutions of an overnight culture of E. coli 536 ΔrecA PAI III::sacB (Middendorf et al., 2004) were plated on LB+ sucrose plates and grown for 48 h at 20°C. A 544 bp chromosomal DNA fragment centred on PAI III536 DR was amplified by PCR (primer pair ΔPAI III up/ΔPAI III lp) on a boiled sucrose-resistant colony. This attBIII-544 fragment was then TA-cloned into the pGEM-T easy vector (Promega), and the EcoRI fragment containing attBIII-544 was released and subsequently cloned into pSW23T, creating pattBIII-544 (Table 4). To reduce the size of the attBIII site, a fragment of 343 bp surrounding PAI III536 DR was first amplified by PCR with the primers III-attB1/III-attB2, using pattBIII-544 as a template and then cloned into pGEM-T easy vector. The resulting EcoRI fragment containing attBIII-343 was finally cloned into pSW23T, generating pattBIII-343 (Table 4). To clone PAI III536 DR into pSW23T, 500 pmol of primers DR-III+ and DR-III-new, which consist in the DR extended by the 5′ overhanging sequence corresponding to EcoRI digestion and 3′ phosphorylated, were annealed and cloned into the EcoRI site of pSW23T, creating pDR3. All attBIII fragments smaller than the DR and cloning of intact or partially deleted PAI II536 DR resulted from a fusion with a 247 bp internal fragment of lacZ. PCRs were performed with lacZ-2 primer and the primers corresponding to the different deletions inside the DR, as indicated in Table 4: numbers in the primer and derived plasmid names indicate the number of deleted base pairs in the DR. R indicates a deletion from the right-hand extremity of the DR, or L from its left hand. All PCR products were first TA-cloned into pGEM-T easy vector, and subsequently into pSW23T. The attBII-493 and attBIII-544 sites from pattBII and pattBIII were cloned into the EcoRI site of pSW25T (Demarre et al., 2005), creating plasmids pattBIISp and pattBIIISp respectively.
Constructions of the plasmids containing the attP sites. pintII and pKintII-attPII have already been described (Hochhut et al., 2006). A boiled colony of E. coli strain 536 was used as template to amplify the intB and intA integrase genes of PAI III536 with primers intPAI3B-F/intPAI3B-R, and int-sfx/xis3 respectively. PCR products were cloned into pGEM-T easy vector, resulting in the pintB and pintA plasmids. The various attPII sites of plasmids pKintII attPII-253 and pKintII attPII-48 were constructed as follows: PCRs to amplify the different attPII sites were performed on pPAI II-CI plasmid (Middendorf et al., 2004) with primer pairs attP2up/attP2down, and lacZ-attP2-48-SacII/lacZ-2SphI, and cloned into pGEM-T easy vector. The SphI-SacII fragments containing attPII were subsequently cloned into pintII, and finally, a neo cassette, first amplified on pCRII-TOPO (bla, neo, T/A cloning vector, Invitrogen) with primers KmFw/KmRev, cloned into pGEM-T easy and finally released by a NotI digestion, was cloned into the NotI site of these plasmid intermediates containing intII and the different attPII fragments, generating the final plasmids pKintII attPII-253 and pKintII attPII-48 respectively. Similarly, PCRs to amplify the different fragments used for the attPIII sites delineation were performed on plasmid pPAI III-CI (Middendorf et al., 2004) with primer pairs attP1/attP2, and III-attP5/attP2 for plasmids pKintB attPIII-268, pKintA attPIII-268 and pKintB attPIII-155, or on a boiled colony of E. coli MG1655 with lacZ-DR3SacII/lacZ-2SphI for plasmid pKintB attPIII-48, and cloned into pGEM-T easy vector. The SphI-SacII fragments containing the different attPIII sites were cloned into pintB, and the attPIII-268 fragment was cloned into pintA. Finally, the neo cassette was cloned into the NotI site of pintB and pintA derivatives, generating the final plasmids pKintA attPIII-268, pKintB attPIII-268, pKintB attPIII-155 and pKintB attPIII-48. For plasmids pKintII attPII-14, pKintII attPIII-16, pKintII attPII-DR, pKintB attPIII-13, and pKintB attPIII-20, the sequence of the attP site of interest was added at the 5′ extremity of the primer amplifying the upper part of the neo cassette. The attP-kanamycin fragment generated by PCR on pCRII-TOPO was first cloned into the pGEM-T vector, and subsequently into the SacII restriction site of pintB or pintII.
Restriction enzymes and DNA modifying enzymes were purchased from New England Biolabs (Frankfurt) and used according to the manufacturer's instructions. Plasmid DNA was isolated following standard protocols (Sambrook et al., 1989). DNA fragments were purified using Qiagen (Hilden) products. DNA sequencing was done with the BigDye system (PE Biosystems) and ABI-377 automated DNA sequencers (Applied Biosystems, Weiterstadt, Germany).
Oligonucleotides were purchased from Sigma-Genosys (Taufkirchen, Germany). A complete list of primers used in this study is available as supplemental material (Table S1).
Construction of strains to study the role of IHF and HU in PAI integration
The roles of IHF and HU in the integration mechanism were evaluated by comparison of the recombination frequencies in strain MC240Pir and in its himA derivative, MC251Pir, as well as C600Pir and its hupAB derivative, OPH252. These four strains (listed in Table 3) were constructed by P1 transduction of the chromosomal ΔthyA::(erm-pir116) allele from strain Π1 (Demarre et al., 2005) to insert the pir gene into the chromosome.
Conjugation and transformation assays
Pir expressing strains β2163, MC240Pir or MC251Pir were co-transformed with pSW23T containing the attB site of interest and with a pGEM-T derivative containing the integrase and attP site of interest. Conjugation assays were performed as described in Hochhut et al. (2006). Alternatively, for transformation assays, plasmid extracts from overnight cultures of β2163, MC240Pir or MC251Pir containing the two plasmids and co-integrates were used to transform TOP10 cells. After 1 h of expression at 37°C, 100 μl of 10−1−10−4 serial dilutions of the bacterial suspension were plated on LB plates containing Ap, Km and Cm to select for the co-integrates, and LB plates containing Ap and Km to select for the initial pGEM-T derivative. To test the influence of the HU protein, the bacterial suspensions were plated on LB plates containing Ap, Km and Sp. The frequency of recombination (number of colonies containing the co-integrates divided by the number of colonies containing the pGEM-T derivative) was calculated from the data of at least four independent experiments. The site-specific recombination event was checked by sequencing six to 10 random co-integrates with primers pSW-cl-up and pSW-cl-down.
Strategy used to delineate the attP and attB recombination sites
A two-plasmid system (Demarre et al., 2005) was used to study the specificity of the recombination reaction (Fig. 7). If the integrase mediates site-specific recombination between the tested attP and attB site derivatives, a co-integrate of the two compatible plasmids will be formed. Co-integrates were recovered either from the Pir– recipient strain DH5α after conjugation, or from TOP10 cells after transformation. In the conjugation assay (Demarre et al., 2005), the recombination frequency was calculated as the ratio between the number of transconjugants and the number of total bacteria. In the transformation assay, a plasmid extract from an overnight culture (containing a mixture of the two original plasmids and their co-integrates) was used to transform TOP10 cells (Pir-), and the recombination frequency was calculated as the number of transformants containing co-integrates divided by the number of transformants containing either the co-integrates or the initial pGEM-T constructs. Finally, in both assays, the recombination sites corresponding to the co-integrate junctions were analysed by sequencing between six and 10 randomly chosen clones per experiment.
Pathogenicity islands deletion assays to study the role of IHF in PAI II536 and III536 excision
Polymerase chain reaction assays for detection of PAIs deletion and determination of deletion rates of sacB-labelled PAIs were done as previously described (Hochhut et al., 2006).
- Top of page
- Experimental procedures
- Supporting Information
We thank Dr J.M. Clément (Paris) for helpful discussion, and M. Strengert (Würzburg) for technical assistance. C.W. was a postdoctoral fellow of the Bayerische Forschungsstiftung. The work of the Würzburg group was supported by the DFG (Sonderforschungsbereich 479, TP A1). Work in the E.C., J.H. and D.M. laboratories is carried out in the frame of the European virtual institute for functional Genomics of bacterial pathogens (CEE LSHB-CT-2005-512061).
- Top of page
- Experimental procedures
- Supporting Information
- 2006) Independent acquisition of site-specific recombination factors by asn tRNA gene-targeting genomic islands. Int J Med Microbiol 296: 341–352. , , , and (
- 1954) Segregation of lambda lysogenicity during bacterial recombination in Escherichia coli K12. Genetics 39: 429–439. (
- 2006) The transcriptional antiterminator RfaH represses biofilm formation in Escherichia coli. J Bacteriol 188: 1316–1331. , , , , , , and (
- 1982) Cloning of the chromosomal determinants encoding haemolysin production and mannose resistant haemagglutination in Escherichia coli. J Bacteriol 152: 1241–1247. , , , , and (
- 1994) Excision of large DNA regions termed pathogenicity islands from tRNA-specific loci in the chromosome of an Escherichia coli wild-type pathogen. Infect Immun 62: 606–614. , , , , , , and (
- 2004) Bacteriophage P4 Vis protein is needed for prophage excision. Virology 322: 82–92. , , , , , , and (
- 1999) Site-specific recombination of bacteriophage P22 does not require integration host factor. J Bacteriol 181: 4245–4249. , , , and (
- 1984) E. coli integration host factor binds to specific sites in DNA. Cell 39: 707–716. , and (
- 2005) A new family of mobilizable suicide plasmids based on broad host range R388 plasmid (IncW) and RP4 plasmid (IncPalpha) conjugative machineries and their cognate Escherichia coli host strains. Res Microbiol 156: 245–255. , , , , , and (
- 2001) Whole genome plasticity in pathogenic bacteria. Curr Opin Microbiol 4: 550–557. , and (
- 2002) Genetic structure and distribution of four pathogenicity islands (PAI I536-PAI IV536) of uropathogenic Escherichia coli strain 536. Infect Immun 70: 6365–6372. , , , , , , and (
- 2004) Genomic islands in pathogenic and environmental microorganisms. Nat Rev Microbiol 2: 414–424. , , , and (
- 2001) A structural view of Cre-loxP site-specific recombination. Annu Rev Biophys Biomol Struct 30: 87–104. (
- 2002) A Structural View of Tyrosine Recombinase Site-Specific Recombination. Washington DC: American Society for Microbiology Press. (
- 2005) Lambda integrase: armed for recombination. Curr Biol 15: R658–R660. (
- 2005) Cooperative interactions between bacteriophage P2 integrase and its accessory factors IHF and Cox. Virology 332: 284–294. , , , and (
- 1986) Replication of pSC101: effects of mutations in the E. coli DNA binding protein IHF. Mol Gen Genet 204: 85–89. , , , , , and (
- 2005) Peptide trapping of the Holliday junction intermediate in Cre-loxP site-specific recombination. J Biol Chem 280: 8290–8299. , , , , and (
- 1981) Identification of a sex-factor-affinity site in E. coli as gamma delta. Cold Spring Harb Symp Quant Biol 45 Part 1: 135–140. , , , and (
- 1999) The concept of pathogenicity islands. In: Pathogenicity Islands and Other Mobile Virulence Elements. Kaper, J.B and Hacker, J. eds. Washington, DC: American Society for Microbiology Press, pp. 1–11. , and (
- 2000) Pathogenicity islands and the evolution of microbes. Annu Rev Microbiol 54: 641–679. , and (
- 2002) Pathogenicity Islands and the Evolution of Pathogenic Microbes. Berlin. Heidelberg, New York: Springer. , and (ed.) (
- 2005) Pathogenicity islands and their role in bacterial virulence and survival. Contrib Microbiol 12: 234–254. , , and (
- 2006) Role of pathogenicity island-associated integrases in the genome plasticity of uropathogenic Escherichia coli strain 536. Mol Microbiol 61: 584–595. , , , , , , et al. (
- 1984) Resolution of synthetic att-site Holliday structures by the integrase protein of bacteriophage lambda. Nature 311: 721–726. , and (
- 1989) Multiple defects in Escherichia coli mutants lacking HU protein. J Bacteriol 171: 3704–3712. , , , , , and (
- 1991) Virulence factors in Escherichia coli urinary tract infection. Clin Microbiol Rev 4: 80–128. (
- 1991) Genetic analysis of Escherichia coli integration host factor interactions with its bacteriophage lambda H' recognition site. J Bacteriol 173: 609–617. , , , and (
- 1984) The phi 80 and P22 attachment sites. Primary structure and interaction with Escherichia coli integration host factor. J Biol Chem 260: 4468–4477. , , , , , and (
- 2004) Excision of the high-pathogenicity island of Yersinia pseudotuberculosis requires the combined actions of its cognate integrase and Hef, a new recombination directionality factor. Mol Microbiol 52: 1337–1348. , , , , , and (
- 2004) Excision of the Shigella resisistance locus pathogenicity island in Shigella flexneri is stimulated by a member of a new subgroup of recombination directionality factors. J Bacteriol 186: 5551–5554. , , , , and (
- 2001) The pathogenicity island (PAIs) of the uropathogenic Escherichia coli strain 536: island probing of PAI II536. J Infect Dis 183: S17–S20. , , , , , and (
- 2004) Instability of pathogenicity islands in uropathogenic Escherichia coli 536. J Bacteriol 186: 3086–3096. , , , , , and (
- 1995) Characterization of the interaction between the lambda intasome and. Attb J Mol Biol 252: 47–58. , and (
- 2006) Expression of phage P4 integrase is regulated negatively by both Int and Vis. J Gen Virol 87: 2423–2431. , , , , , , et al. (
- 1997) Use of a novel approach, termed island probing, identifies the Shigella flexneri shepathogenicity island which encodes a homolog of the immunoglobulin A protease-like family of proteins. Infect Immun 65: 4606–4614. , , and (
- 1991) Analysis of the IHF binding site in the regulatory region of bacteriophage Mu. Nucleic Acids Res 19: 2825–2834. , , and (
- 1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory. , , and (
- 2004) The pathogenicity island-associated K15 capsule determinant exhibits a novel genetic structure and correlates with virulence in uropathogenic Escherichia coli strain 536. Infect Immun 72: 5993–6001. , , , , , , et al. (
- 1998) Prevalence of the ‘high-pathogenicity island’ of Yersinia species among Escherichia coli strains that are pathogenic to humans. Infect Immun 66: 480–485. , , , , and (
- 2002) Integration sites for genetic elements in prokaryotic tRNA and tmRNA genes: sublocation preference of integrase subfamilies. Nucleic Acids Res 30: 866–875. (
- 1985) Interaction of the lambda site-specific recombination protein Xis with attachment site DNA. Proc Natl Acad Sci USA 82: 1040–1044. , , and (
- Top of page
- Experimental procedures
- Supporting Information
|MMI_6145_sm_Table_S1.pdf||224K||Supporting info item|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.