Plasticity of the P_junc Promoter of ISEc11, a New Insertion Sequence of the IS1111 Family

Abstract

We describe identification and functional characterization of ISEc11, a new insertion sequence that is widespread in enteroinvasive E. coli (EIEC), in which it is always present on the virulence plasmid (pINV) and very frequently also present on the chromosome. ISEc11 is flanked by subterminal 13-bp inverted repeats (IRs) and is bounded by 3-bp terminal sequences, and it transposes with target specificity without generating duplication of the target site. ISEc11 is characterized by an atypical transposase containing the DEDD motif of the Piv/MooV family of DNA recombinases, and it is closely related to the IS1111 family. Transposition occurs by formation of minicircles through joining of the abutted ends and results in assembly of a junction promoter (P_juncC) containing a −10 box in the interstitial sequence and a −35 box upstream of the right IR. A natural variant of ISEc11 (ISEc11p), found on EIEC pINV plasmids, contains a perfect duplication of the outermost 39 bp of the right end. Upon circularization, ISEc11p forms a junction promoter (P_juncP) which, despite carrying −10 and −35 boxes identical to those of P_juncC, exhibits 30-fold-greater strength in vivo. The discovery of only one starting point in primer extension experiments rules out the possibility that there are alternative promoter sites within the 39-bp duplication. Analysis of in vitro-generated transcripts confirmed that at limiting RNA polymerase concentrations, the activity of P_juncP is 20-fold higher than the activity of P_juncC. These observations suggest that the 39-bp duplication might host cis-acting elements that facilitate the binding of RNA polymerase to the promoter.

Bacterial insertion sequence (IS) elements are small, defined genetic entities that encode no functions other than those involved in their mobility. ISs are capable of altering gene expression and mediating a variety of DNA rearrangements through insertion into host or plasmid genomes. They have been found in a wide range of bacterial and archaeal species. The number of known IS elements is increasing, and many of them have been classified into families on the basis of conserved transposase (Tpase) domains, homology of the inverted repeat (IR) sequences, and/or other shared functional properties. Most IS elements encode a Tpase that includes the highly conserved DDE motif and are bounded by terminal IRs (13, 23).

Several mobile DNA elements of both prokaryotic and eukaryotic origin have been observed as circular intermediates after excision (23). The molecular mechanism underlying the transposition strategy has been extensively studied using two members of the IS3 family as models, IS911 (10, 36) and IS2 (19, 20). The formation of an IS minicircle has been shown to occur by a variation of the “cut-and-paste” transposition pathway (10, 20). This strategy involves generation of a “figure eight” intermediate resulting from Tpase-mediated cleavage of the IS 3′ end and from transfer of this end to a site on the same DNA strand a few nucleotides outside the other IS end. The intermediate is then converted into an IS minicircle by replication and/or by host repair activities, thus reproducing the initial replicon carrying the parental IS. The junction formed by the abutted IS ends in the minicircle is a transpositionally hyperactive substrate, which can be readily nicked by the Tpase, giving rise to a linear IS element.

A peculiarity of ISs which transpose using a reactive junction is that the abutted IRs are always separated by a short DNA linker derived from nucleotides flanking the target end in the parental insertion (22, 23, 26, 29). Another feature of several transposable elements which undergo circularization is that the junction between the left IR (IRl) and the right IR (IRr) results in assembly of a junction promoter (P_junc) capable of efficiently driving Tpase synthesis, as is the case in IS911 (36). Junction promoters are generally stronger than the indigenous IS promoters (P_IRL), which are usually located partially within the IS sequence upstream of the Tpase gene, and their formation leads to a burst of Tpase synthesis that results in efficient integration of the IRr-IRl junction (27). In IS911, as in other IS elements, the formation of a minicircle by means of a short intervening nucleotide stretch (3 to 6 bp) creates a promoter by placing a −35 element present in the terminal IRr at an optimal distance from a −10 box located in the abutted IRl (20, 36). In other cases, including the IS1111 transposon family, when the ends of the IS element are brought together to form the circular intermediate, the interstitial sequence itself represents a −10 box, which is correctly positioned with respect to the −35 region located near the right end of the IS element (26, 29).

Although integration of transposable elements often seems to occur at random, sequence-specific insertions have been observed for several elements (9). Thus, Tn7 integrates at a high frequency into a single locus (attTn7) of the Escherichia coli chromosome (21). IS1630 seems to integrate preferentially into palindromic sequences that resemble transcription terminators (4), while IS621 and ISPpu10 transpose within the repetitive extragenic palindromic sequences of E. coli and Pseudomonas putida, respectively (7, 32). In some cases, like IS117 and IS1383, which are able to generate a circular intermediate, the target sequence exhibits homology to the interstitial junction sequence, indicating that insertion may involve a site-specific recombination process (2, 26).

Here we describe identification and functional characterization of a new insertion sequence, ISEc11, which exhibits high levels of homology with elements belonging to the IS1111 family (29). ISEc11 was isolated from enteroinvasive E. coli (EIEC) strain HN280, in which it resides both on the chromosome and on the large (260-kb) F-type virulence plasmid (pINV). We show here that ISEc11 is able to induce formation of circular intermediates, is transpositionally active, and has target specificity. Minicircle formation results in the assembly of a P_junc containing a −10 box in the interstitial sequence. Moreover, we present data indicating that the plasmid copy of ISEc11 (ISEc11p) is a spontaneous variant containing an internal 39-bp duplication of the right end region. The presence of this duplication is responsible for generation of a much more efficient junction promoter upon circularization, suggesting that there are cis-acting elements in the 39-bp duplication, which facilitate the binding of RNA polymerase to the promoter.

MATERIALS AND METHODS

Bacterial strains and general procedures.

The general features of EIEC strains HN280, HN11, HN13, HN19, and HN300 (isolated in Somalia during a survey of childhood diarrheal syndromes [28]), of EIEC strains 4608 and 53638 (Walter Reed Army Institute of Research collection), and of EIEC strains 13-80 and 6-81 (Institut Pasteur Collection) have been described previously (6). Shigella flexneri strains SFZM43, SFZM46, SFZM49, SFZM50, SFZM53, M90T, and YSH6000 have also been described previously (5). E. coli K-12 strain DH10b (34) was routinely used as the recipient in transformations, while E. coli MC4100 (25) was used as a background strain in β-galactosidase assays. Bacteria were routinely grown in LB medium, and when required, ampicillin (100 μg/ml) and/or chloramphenicol (30 μg/ml) was added.

β-Galactosidase assays were performed with sodium dodecyl sulfate-chloroform-permeabilized cells grown in LB medium supplemented with ampicillin. β-Galactosidase activity was determined as described by Miller (25), and the results were expressed as averages of three independent experiments.

DNA manipulations.

Extraction of total and plasmid DNA, restriction digestion, electrophoresis, and purification of DNA fragments were carried out as described previously (12, 31). Southern hybridization was performed using α-³²P-labeled DNA probes essentially as described by Sambrook and Russell (34). Details concerning the primers used in this study are shown in Table S1 at http://w3.uniroma1.it/bcolonna/docs/jb06_tabs1.pdf. The IS probe was generated using oligonucleotides S1 and S2 to amplify an internal region of the ISEc11 tnpA sequence. While PCRs were routinely performed using Taq polymerase, amplicons for cloning were obtained with high-fidelity Pwo polymerase.

Plasmid construction.

Plasmids used in this study are listed in Table 1. Plasmid pGEc11, containing the ISEc11 copy inserted into the yjgZ-yjzY chromosomal locus, and plasmid pGEc11p, containing the ISEc11 plasmid variant (ISEc11p), were generated by cloning the amplicons obtained with primers CS1 and CS2 or with primers PS1 and PIS2 from genomic DNA of EIEC strain HN280 into pGEM-T (Promega). Plasmids pGJC1 and pGJP3, containing the IRl-IRr junction sequences of ISEc11 and ISEc11p, were generated by cloning into pGEM-T fragments obtained by amplifying HN280 DNA with primers PC1 and PC2 in the opposite orientations and corresponding to ISEc11 sequence positions 1062 to 1079 and 715 to 699, respectively.

TABLE 1.

Plasmids used in this study

Plasmid	Relevant features	Reference or source
pHN280	pINV of EIEC strain HN280	8
pHN281	pCryp of EIEC strain HN280	8
pRS414	pBR322 derivative suitable for lacZ translation fusion, Apr	25
pKK232-8	Vector suitable for transcriptional fusion with cat gene, Apr	Pharmacia
pGEM-T	Cloning vector, Apr	Promega
pACYC184	Cloning vector, Cmr Tcr	34
pGEc11	pGEM-T derivative containing ISEc11 element, Apr	This study
pGEc11p	pGEM-T derivative containing ISEc11p element, Apr	This study
pGJC1	pGEM-T derivative containing IRl-IRr junction of ISEc11, Apr	This study
pGJP3	pGEM-T derivative containing IRl-IRr junction of ISEc11p, Apr	This study
pRINc1	pRS414 derivative containing TnpA-LacZ fusion of ISEc11, Apr	This study
pRINp2	pRS414 derivative containing TnpA-LacZ fusion of ISEc11p, Apr	This study
pRJUc1	pRS414 derivative containing TnpA-LacZ fusion under the control of PjuncC, Apr	This study
pRJUp3	pRS414 derivative containing TnpA-LacZ fusion under the control of PjuncP, Apr	This study
pTRIS0	Entrapment vector containing ISEc11/ISEc11p target sequence, Apr	This study
pTRIS11	pTRIS0 derivative containing ISEc11 element, Apr Cmr	This study
PTRIS11p	pTRIS0 derivative containing ISEc11p element, Apr Cmr	This study
pKJEc11	pKK232 derivative containing TnpA-CAT fusion under the control of PjuncC, Apr Cmr	This study
PKJEc11p	pKK232 derivative containing TnpA-CAT fusion under the control of PjuncP, Apr Cmr	This study

Plasmid pTRIS0, a vector suitable for entrapment of ISEc11 and ISEc11p, was constructed by replacing the EcoRI-SalI pRS414 fragment, containing the lac operon, with an amplicon carrying the cat gene (Table 1). To do this, the cat gene was amplified using pACYC184 DNA (Tc^r Cm^r) as the template and primers TR1 and TR2 containing an EcoRI restriction site and a SalI restriction site, respectively. Due to the absence of a complete promoter, cat expression is silent in pTRIS0 (Ap^r Cm^s). Primer TR1 was constructed to contain the ISEc11 target site 5′-GTGAAATACTG-3′ located in the HN280 yjgY-yjgZ locus. The tetranucleotide 5′-AAAT-3′, involved in the integration process, was localized at position −10 with respect to the cat gene transcription start site (position 1).

All plasmids containing the TnpA-LacZ fusion were constructed by cloning PCR-generated fragments of the tnpA regulatory regions into the multiple cloning site of the ′lacZYA translational fusion vector pRS414. Plasmids pRINc1 and pRINp2, containing the regulatory region of the Tpase gene of the ISEc11 and ISEc11p linear copies (P_IRL), were obtained using pGEc11 and pGEc11p, respectively, as the templates and primers FL2 and FL3 as the forward and reverse primers, respectively. Plasmids pRJUc1 and pRJUp3 were obtained by amplifying fragments containing the entire IRr-IRl ISEc11 and ISEc11p junctions, using pGJC1 andpGJP3, respectively, as the templates and primers FL1 and FL3 as the forward and reverse primers, respectively. The FL1 and FL2 primers contain an EcoRI site, while reverse primer FL3 contains a BamHI site. The fragments obtained (P_IRL [428 bp] derived from pGEc11 or pGEc11p; P_juncC [668 bp] derived from pGJUc1; and P_juncP [707 bp] derived from pGJUp3 [see Fig. 3A]), which contained the N-terminal portion of the tnpA gene, were fused to the ′lacZ gene of the EcoRI-BamHI-linearized pRS414 vector.

FIG. 3.

Analysis of the strength of the Tpase promoters of ISEc11 and ISEc11p. (A) The fragments containing the tnpA promoter of ISEc11 or ISEc11p in the circular form (P_juncC andP_juncP, respectively) or in the linear form (P_IRL) were obtained by PCR amplification with oligonucleotides FL1 and FL3 or with oligonucleotides FL2 and FL3. The P_IRL-containing regions of ISEc11 and ISEc11p are identical. All fragments extend up to Tpase amino acid Asp₉₂ (TnpA′). (B) β-Galactosidase activity of the TnpA-LacZ translational fusions obtained by cloning the fragments described in panel A into vector pRS414. Activity is expressed in Miller units, and the values are averages of three independent experiments (P_juncP, 3,170 ± 105 Miller units; P_juncC, 109 ± 8 Miller units; P_IRL, 5 ± 1 Miller units). (C) The affinity of RNA polymerase (RNApol) for P_juncC and P_juncP was quantified by real-time PCR as described in Materials and Methods. The fragments containing P_juncC and P_juncP described in panel A were cloned into the pKK232-8 vector upstream of a cat reporter gene. Transcripts were generated in vitro in a 60-μl reaction mixture using different amounts of RNA polymerase (0.1, 0.25, 0.5, and 1 U). The P_juncP/P_juncC affinity ratios are the averages for three independent assays.

Plasmids pKJEc11 and pKJEc11p were obtained by cloning the P_junc promoters of ISEc11 and ISEc11p into pKK232-8. To do this, we obtained two amplicons carrying P_juncC or P_juncP by using pGJC1 and pGJP3 as the templates with oligonucleotides FL1B and FL3 (containing a BamHI site). The amplicons were then cloned into BamHI-linearized pK232-8.

The sequences of PCR-generated fragments were checked by the dideoxy chain termination method (34).

Transposon mobility.

Transposition of ISEc11 and ISEc11p was tested using the pTRIS0 plasmid. pTRIS0 is an entrapment vector containing the ISEc11 insertion consensus (5′-GTGAAAATACTG-3′) upstream of a cat reporter gene (Ap^r Cm^s) which is silent because of an incomplete −10 box and because of the absence of a −35 box (Table 1). Insertion of ISEc11 or ISEc11p into the pTRIS0 target site (giving rise to plasmids pTRIS11 and pTRIS11p, respectively) was monitored by PCR analysis using primers PC1 and TR3 with total DNA from 120 HN280 Ap^r Cm^r colonies. Since chloramphenicol resistance occurs only if the IS element is inserted into its target site, the efficiency of transposition was calculated by averaging the results of three independent experiments and was expressed as the Cm^r/Ap^r ratio.

Primer extension.

Total RNA from strains DH10b(pRJUc1) and DH10b(pRJUp3) grown to an optical density at 600 nm of 0.6 was extracted by using a modification of the hot phenol method and was quantified spectrophotometrically (31). Primer PE1 was 5′ end labeled with [γ-³²P]dATP using T4 polynucleotide kinase and was hybridized with 50 μg of total RNA. Reverse transcription experiments were carried out at 42°C using cMaster RT (Eppendorf) according to the manufacturer's instructions. The resulting cDNAs were analyzed on denaturing 6% polyacrylamide gels, along with a sequencing ladder that was generated by using the same primer and pRJUp3 as the template. Sequencing reactions were performed with a T7 DNA sequencing kit (US Biochemicals) and [α-³²P]dATP. Primer extension experiments with transcripts synthesized in vitro (see below) from pRJUc1 or pRJUp3 were performed under the same conditions.

Quantitative PCR.

Plasmids pKJEc11 and pKJEc11p, derived from pKK232-8, were used as templates in in vitro transcription assays, which were performed at 37°C in 60-μl portions of buffer containing 25 mM Tris-HCl (pH 7.9), 0.1 mM dithiothreitol, 10 mM MgCl₂, 15 mM KCl, 1 mM EDTA, each ribonucleoside triphosphate at a concentration of 2.5 mM, and different amounts (0.1, 0.25, 0.5, and 1 U) of RNA polymerase (Epicenter). While 20 μl of each reaction mixture was stored without further treatment (aliquot 1), the remaining volume was digested with RNase-free RQ1 DNase (Promega). Twenty microliters of each digest was stored without further treatment (aliquot 2), and the remaining 20 μl (aliquot 3) was used as a template for a reverse transcriptase PCR (TaqMan reverse transcriptase reagents; Applied Biosystems). Aliquots 1, 2, and 3 were used as templates for quantitative PCR performed with an ABI PRISM 7300 detection system, using SYBR Green PCR Master Mix (Applied Biosystems) and primers RTC1 and RTC2, corresponding to sequences internal to the pKK232-8 cat gene. A variant of the ΔΔC_T method was used to calculate the final results. In particular, experiments were performed in triplicate, and the ΔC_T value was obtained by subtracting the average cycle threshold (C_T) value of aliquot 1 from the average C_T value of aliquot 3. Calculation of −ΔΔC_T involved subtracting pKJEc11-related ΔC_T values from pKJEc11p-related ΔC_T values. The average C_T value for aliquot 2 was used to evaluate the efficiency of the RNase-free DNase treatment. Primer pair efficiency was examined by looking at how ΔC_T varied with template dilution, according to the suggestions of Applied Biosystems.

Sequence analysis.

DNA sequence data were compared to known nucleotide and protein sequences using the BLAST Server (National Center for Biotechnology Information, Bethesda, Md.) and the IS finder database (http://www-is.biotoul.fr/is.html). Searches for Tpase functional motifs were carried out with the CD-Search program (24). Computer predictions of intrinsic bending were obtained as previously described (12, 31).

Nucleotide sequence accession numbers.

The IS element described in this paper was assigned a designation (ISEc11) by the IS finder database (http://www-is.biotoul.fr/is.html). The sequences of ISEc11 and its plasmid variant, ISEc11p, have been deposited in the GenBank database under accession numbers DQ361537 and DQ361536, respectively.

RESULTS

Identification of ISEc11.

EIEC strains are intracellular facultative pathogens which have the same pathogenicity process as Shigella (15), which depends on the presence of a large F-type virulence plamid (pINV). E. coli HN280 is an EIEC O135 strain that was isolated during an epidemiological survey of diarrheal diseases in children in Somalia, and it harbors two large plasmids, pINV (260 kb) and pCryp (160 kb) (8). While analyzing the flanking region of the HN280 leuX gene, a potential site for pathogenicity island insertions, we identified a ca. 1.4-kb DNA segment that had structural features typical of insertion sequences. This element is inserted between the yjgY and yjgZ genes and is absent in the corresponding E. coli MG1655 region. It contains one 1,017-bp open reading frame (ORF1017), encoding a putative 339-amino-acid protein, and it is flanked by a perfect 13-bp IR, 5′-ATGAACGCATCCC-3′ (Fig. 1). As observed for a small number of other ISs (23), the IRs of this new element are not located at the IS boundaries; they are separated from the ends by 3 bases, and the right terminal sequence (TSr) (5′-ATA-3′) differs slightly from the left terminal sequence (TSl) (5′-GTA-3′). The ends of the insertion element are also flanked by a 4-bp direct repeat (5′-AAAT-3′). The new IS is designated ISEc11 (www-is.biotoul.fr), and its nucleotide sequence has been deposited in the GenBank database (accession number DQ361537).

FIG. 1.

Structure of ISEc11 and its plasmid variant, ISEc11p. ISEc11 is a 1,443-bp IS element with a perfectly conserved 13-bp subterminal inverted repeat (arrowheads), flanked by 3-bp terminal sequences (gray shading). The left TS (5′-GTA-3′) differs slightly from the right TS (5′-ATA-3′). The 4-bp sequence 5′-AAAT-3′ between the abutted ends of ISEc11/ISEc11p in the circular forms is indicated by gray letters to the left of the left TS. The open arrow indicates the transposase gene (tnpA). ISEc11p is a 1,482-bp IS element that was originally identified on the pINV plasmid of EIEC strain HN280. It differs from ISEc11 by a 39-bp duplication at the right end of the element. The duplication (ISEc11 nucleotides 1405 to 1443) contains a copy of the right IR and a copy of the right TS, as well as an upstream stretch.

In order to determine which IS family ISEc11 belongs to, we first looked for IR homologies. Closely related IRs are present in several ISs belonging to the IS1111 family (Table 2). In particular, despite slight length variations, the IRs have highly conserved sequence ends, as indicated by the presence of a beginning ATG motif, a terminal TCC (or TCCC) stretch, and a central A and GC dinucleotide. In all IS elements shown in Table 2 the IRs consist of 12 or 13 nucleotides and are flanked by terminal sequences. It is worth mentioning that in all cases the TSr trinucleotide (TAT) is conserved, while the TSl trinucleotide contains a conserved G as the outermost nucleotide.

TABLE 2.

Distinctive features of ISEc11/ISEc11p and other members of the IS1111 family

IS	Tpase		Spacer + TSlc	TSrc	IRd	Species
	DEDD motif	% Identityb
ISEc11a	D9E52D89D92	100	AAATGTA	TAT	ATGAACGCATCCC	EIEC
IS884	D9E52D89D92	60	NDe	ND	ATGAACGCGTCCC	Ralstonia eutropha
ISBcen5	D9E52D89D92	59	AATTGAA	TAT	ATGAACGCGTCCC	Burkholderia cenocepacia
ISBfun2	D9E52D89D92	60	AATTGAA	ND	ATGAACGCGTCCC	Burkholderia fungorum
ISBcen4	D9E52D89D92	61	AATGGT	TAT	ATGGACGCCTCC	Burkholderia cenocepacia
ISAfe1	D9E52D89D92	53	AGATGTA	TAT	ATGGATGCCTCC	Acidithiobacillus ferroxidans
ISAzvi3	D9E52D89D92	48	AGATGGT	TAT	ATGGACGCCTCCC	Azotobacter vinelandii
IS1383	D9E52D89D92	48	AGATGGT	TAT	ATGGACGCCTCCC	Pseudomonas putida
IS1111A	D10E63D90D93	39	CAATGAA	TAT	ATGGACCCACCC	Coxiella burnetii
PIV	D9E59D101D104	22				Moraxella bovis

^aISEc11p has the features reported for ISEc11.

^bLevel of identity to the ISEc11 Tpase.

^cSpacer sequences indicate the interstitial bases in the circular form. Terminal sequences are the nucleotide stretches that bound the IS element.

^dInverted repeats are subterminal. The left and right IRs are identical.

^eND, no data available.

Next, we compared the ISEc11 open reading frame-encoded putative protein with other IS Tpases. A computer-aided homology search revealed that the predicted ORF1017 gene product exhibits significant sequence identity (around 60%) with the Tpases of the IS884, ISBcen5, ISBfun2, and ISBcen4 elements, all of which belong to the recently described (29) IS1111 family (Table 2). On the basis of these observations, ORF1017 very likely encodes the transposase of ISEc11, and this new IS can be placed in the IS1111 family. Tpases belonging to this family constitute an atypical group of Tpases since they lack the classical catalytic DDE motif and exhibit homology with the Piv/MooV family of DNA recombinases (18, 35). Recently, it has been shown that a DEDD motif (D9, E59, D101, and D104) constitutes the active site of the Piv-specific invertase (3, 35) and that it is present, instead of a DDE motif, in Tpases belonging to the IS110/IS492 family (7, 32). Therefore, we looked for a DEDD motif in the N-terminal domain of the ISEc11 Tpase. As shown in Table 2, D residues at positions 9, 89, and 92 and an E residue at position 52 are present in ISEc11 and are perfectly conserved in several elements belonging to the IS1111 family. Interestingly, these elements also have common IRs and exhibit significant Tpase identity, suggesting that they might constitute a homogeneous group in the IS1111 family. We hypothesize that in the Tpases of these elements, analogous to other DNA recombinases, like the Piv and RuvC proteins (3), the DEDD catalytic motif may be located in an RNase H-like structural motif coordinating two divalent metal ions and directing the hydrolysis of DNA phosphodiester bonds.

In silico analysis with the CD software (24) revealed that besides a DEDD motif in the N-terminal portion, the ISEc11 Tpase exhibits, between amino acids 186 and 284, homology with a structural motif of pfam02371 (transposase family 20). Furthermore, between residues 76 and 145, there is partial homology with a domain typical of pfam1548 (transposase family 9). It has recently been reported that these two motifs are also present in the Tpase of ISPpu10, a member of the IS110/IS492 family (32).

Target specificity of ISEc11.

Southern hybridization analysis was performed to determine the exact number of copies of ISEc11 present in the EIEC strain HN280 genome. Since HN280 contains, besides the large virulence plasmid (pHN280), a large cryptic plasmid (pHN281) (Table 1) (8), we probed EcoRI digests of HN280 total DNA, as well as EcoRI digests of pHN280 and pHN281, with the ISEc11 transposase coding region. Interestingly, HN280 harbors five copies of ISEc11; four of them are located on the chromosome, and one is located on the pHN280 virulence plasmid. No hybridization was detected with pHN281 DNA.

To detect spreading of ISEc11 in EIEC strains and in the closely related pathogen S. flexneri, total DNAs from nine EIEC strains (HN280, HN11, HN13, HN280, HN300, 13-80, 6.81, 53638, and 4608) and from seven S. flexneri strains (SFZM43, SFZM46, SFZM49, SFZM50, SFZM53, M90T, and YSH6000), belonging to different serotypes and isolated in different geographic areas (5, 28), were digested with EcoRI and hybridized with the ISEc11 transposase probe. While the plasmid copy of ISEc11 is present on the same restriction fragment (5.1-kb EcoRI fragment) in all of the EIEC strains analyzed, there are differences in the number and the relative location of the chromosomal copies. In particular, while the five EIEC strains from eastern Africa (HN280, HN11, HN13, HN19, and HN300) harbor four ISEc11 copies in the same position, in the remaining four EIEC strains the number of ISEc11 chromosomal copies varies from none to four. In contrast to EIEC, none of the S. flexneri strains analyzed harbor regions homologous to the ISEc11 transposase-encoding gene (data not shown).

Alignment of the sequences adjacent to the ends of the HN280 ISEc11 copies revealed that the IS has a potential target site (5′-GTNAAAANANTG-3′) consisting of a central stretch of four A residues bounded by a GTN trinucleotide on the 5′ end and by an NANTG stretch on the 3′ end. The asymmetric structure of the target site could account for the fact that all of the ISEc11 copies analyzed were found to be inserted in the same orientation. A few members of the IS1111 family are known to have preferred target sites into which they insert in only one orientation, and often the target sites are represented by the IRs of other transposable elements (26, 29).

ISEc11p, the virulence plasmid variant of ISEc11.

Sequence analysis revealed that the single copy of ISEc11 carried by pHN280, which is 4.6 kb upstream of the ospC3 plasmid gene, contains a 39-bp additional sequence downstream of the right end of the element. This short DNA stretch is a duplication of the outermost 39 nucleotides of ISEc11 (from position 1405 to position 1443) and therefore contains another copy of the IR element (IRr) and the TSr sequence (Fig. 1). The resulting structure of this ISEc11 variant, which we designated ISEc11p (GenBank accession number DQ361536), could have originated from insertion of ISEc11 into itself, followed by a rearrangement-excision process that left only remnants of the inserted IS.

A PCR analysis was performed in order to check for the presence of ISEc11p in the pINVs of all EIEC strains used in this study (see Materials and Methods). To do this, we used primers PC1 and PIS2, corresponding to a tnpA internal sequence and a region downstream of the ISEc11p insertion site, respectively. In all pINVs an insertion appeared to have occurred at the same locus, either as ISEc11 or as ISEc11p_. In particular, the variant form ISEc11p is present only in pINVs of strains HN11, HN13, HN19, and HN300, which were isolated (like HN280) during an epidemic outbreak in Somalia (5, 28). Since there is a high level of sequence homology and since the pINVs of Shigella and EIEC are interchangeable (17), using the ShiBASE database (www.ngc.ac.cn) (38), we verified in silico that ISEc11 or remnants of ISEc11 were present in Shigella spp. pINVs. This analysis revealed that the presence of an entire ISEc11 element in pINV is a hallmark of EIEC strains, while pINV plasmids of S. flexneri, Shigella dysenteriae, and Shigella boydi contain only a fragment corresponding to nucleotides 1 to 220 of the ISEc11 sequence. The S. dysenteriae pINV also harbors a larger fragment, containing internal ISEc11 sequences (nucleotides 275 to 698). The extremely high IS content of pINV plasmids (38) and the peculiar location of ISEc11 in the pINVs of EIEC strains (i.e., an IS-rich region) could have favored a series of recombination events that left only remnants of ISEc11.

ISEc11 and ISEc11p form circular intermediates.

The sequence data strongly suggest that ISEc11 is one of the ISs which have the capacity to generate transpositionally active circular forms as a consequence of the IRl-IRr junction. Therefore, we looked for evidence of circular intermediates of ISEc11 and of its variant ISEc11p by performing a PCR analysis. Total DNA of EIEC strain HN280 was used as a template for a PCR with ISEc11 primers PC1 and PC2 oriented toward the ends of the element, so that an amplification product could occur only if the inverted repeats abutted each other as a consequence of circularization of the IS. An ca. 1.1-kb product was detected, confirming the ability of ISEc11 to generate circular forms. To ascertain whether the amplicons could have arisen from circular intermediates of ISEc11 or ISEc11p (or both), we separated the PCR products on 1.2% agarose gels. Two fragments were detected, and sequence analysis confirmed that they correspond to amplification of the circular intermediate forms of ISEc11 (1,097 bp) and ISEc11p (1,136 bp).

The two amplicons consist of the right and left ends of ISEc11 or ISEc11p joined by the abutted IRs. In particular, the right and left 13-bp IRs (IRl and IRr) were separated by a 10-bp sequence, IRr-ATAAAATGTA-IRl, that comprised the abutted terminal sequences of ISEc11 and ISEc11p (Fig. 2B). Considering the 3-bp terminal sequence (TSs) part of the IS, we concluded that they are separated by a 4-bp 5′-AAAT-3′ spacer sequence. This tetranucleotide flanks each of the IS copies on both sides. Therefore, the interstitial DNA stretch between TSl and TSr at the junction is identical to the direct repeats flanking the element in its integrated form. We hypothesize that all of the 5′-AAAT-3′ spacer could have been derived from either the left or right flanking sequence or could consist of 1, 2, or 3 bases from the left end and 3, 2, or 1 bases from the right end of the direct repeats generated by the IS upon insertion. A comparison of the 5′-AAAT-3′ spacer to the target site (5′-GTGAAATACTG-3′) (Fig. 3A) further suggested that integration of the circular IS form may have occurred by a site-specific recombination process in which the spacer tetranucleotide was used as the central recombination site.

FIG. 2.

Structural and functional analysis of the P_junc promoters of ISEc11 and ISEc11p. (A) Primer extension analysis of the transcripts generated under control of the P_junc promoters of ISEc11 (P_juncC) and ISEc11p (P_juncP). The autoradiograph is the result of a typical experiment performed with the PE1 primer and RNA extracted from strain MC4100 transformed with pRJc1 (P_juncC) or RJp3 (P_juncP). Lanes G, A, T, and C show the sequencing ladder generated with the same primer. The arrow indicates the end point of the extended product. (B) Sequence of the ISEc11 and ISEc11p minicircle junction. Inverted repeats are indicated by solid arrows and by open arrows. The spacer sequence acts like a −10 box (gray background) of the P_junc promoter. The outwardly directed −35 box upstream of the right IR contributes to generation of the P_junc promoter. The transcription start site (+1), mapped by primer extension (panel A), is located in the left IR. ISEc11p contains an additional 39-bp stretch, generated by duplication, and it also contains a second copy of the right IR (IRr*) (see the legend to Fig. 1).

Plasticity of the junction promoters.

Many indigenous promoters which drive Tpase expression are inefficient, thus limiting Tpase levels at the transcriptional stage (27). Formation of IS circles has previously been shown (36) to give rise to strong so-called junction promoters (P_junc), which allow increased Tpase synthesis. The assembly of promoters following the formation of a circle intermediate has been documented in several different IS families (23). For ISEc11 and ISEc11p, in silico analysis of the junction sequences indicated that circularization can generate a P_junc promoter arising from a −10 box located in the interstitial region and from a −35 consensus located upstream of the IRr (Fig. 2B).

To analyze the functionality of the predicted P_junc promoters and to compare their activity with that of the native Tpase promoter, we introduced into pRS414, a vector suitable for creating LacZ translational fusions (Table 1), fragments containing the upstream region of the Tpase gene when ISEc11 and ISEc11p are in the linear or circular form (Fig. 3A). Assuming that the native Tpase promoter is very likely located around the IRl, we amplified the whole left IS region using oligonucleotides FL2 and FL3 and the DNA of pGEC11 or pGEC11p (Table 1) as the template. The amplicons, which contained the same nucleotide sequences, were cloned into EcoRI-BamHI-linearized pRS414 to obtain plasmids pRINc1 and pRINp2 (Table 1). The fragments carrying the P_junc promoters, obtained by amplifying pGJC1 and pGJP3 (Table 1) with oligonucleotides FL2 and FL3, were cloned into pRS414, giving rise to plasmids pRJUc1 and pRJUp3, respectively (Table 1). All the pRS414 derivatives (pRINc1, pRINp2, pRJUc1, and pRJUp3) encode a Tpase-LacZ hybrid protein whose expression is controlled by either indigenous (P_IRL) or junction (P_juncC from ISEc11, P_juncP from ISEc11p) promoters.

The results of β-galactosidase assays (Fig. 3B) indicated that the level of expression of the TnpA-LacZ fusion under the control of P_IRL (containing the regulatory region of the tnpA gene when ISEc11 and ISEc11p were integrated) was very low (5 ± 1 Miller units). This presumably reflects very low efficiency of the Tpase promoter and is in agreement with the known inefficiency of many indigenous Tpase promoters (27). In contrast, the presence of the junction sequences activated the expression of the Tpase-LacZ fusion in pRJC1 and pRJP3, confirming that the predicted P_junc promoter is functional. Surprisingly, the β-galactosidase activity under the control of the P_juncP promoter (pRJp3) was 30-fold higher (Fig. 3B) than the β-galactosidase activity elicited by P_juncC (pRJc1).

To ascertain whether the difference was due to increased activation of P_juncP or to the use of an alternative promoter, we analyzed the transcription start position of the tnpA-lacZ mRNA. Total RNA extracted from exponentially growing cultures of strains carrying pRJc1 and pRJp3 was reverse transcribed using an end-labeled oligonucleotide primer (PE1) complementary to nucleotides 182 to 161 of ISEc11/ISEc11p. The results (Fig. 2A) show that there was only one tnpA transcript, starting downstream of the −10 box located in the interstitial sequence, at the G nucleotide (G₁₀) located in the IRl element. This indicates that transcription of the Tpase gene is dependent on the P_junc predicted in silico and rules out the possibility that there is activity of alternative promoters. Under the conditions used, the signal from pRJc1 was very low, as expected considering the large difference between P_juncC and P_juncP observed in the β-galactosidase assays (Fig. 3B).

Comparative in vitro activity assays of P_juncC and P_juncP.

The higher efficiency of P_juncP may well be explained on the basis of the upstream structural difference between the two promoters (i.e., the 39-bp duplication), which mediates either stronger binding of the RNA polymerase or better promoter accessibility for other positive regulatory factors. The latter possibility is unlikely since the higher P_juncP activity was also confirmed by primer extension assays (not shown) using in vitro transcripts originating from P_juncC (pRJUc1) and P_juncP (pRJUp3).

To determine whether the 39-bp duplication might mediate more efficient RNA polymerase binding to the promoter, we compared in vitro transcripts obtained under the control of P_juncC and P_juncP. Plasmids pKJEc11 and pKJEc11p, obtained by cloning fragments containing the P_junc promoter of ISEc11 or ISEc11p (Fig. 3A) into pKK232-8 (Table 1) upstream of a cat reporter gene, were transcribed in vitro using different amounts of RNA polymerase. The transcripts were then quantified by real-time PCR in the presence of the SYBR Green dye. As shown in Fig. 3C, at a low RNA polymerase level (0.1 U) the P_juncP activity was about 20-fold higher than the activity of P_juncC, and the difference decreased with increasing RNA polymerase levels.

These in vitro observations strongly suggest that no other regulatory factors are involved in the enhanced transcriptional activity elicited by P_juncP and that, at limiting RNA polymerase concentrations, the transcriptional stimulation strictly depends on the presence of the 39-bp duplication. Therefore, in this region we looked for cis-acting elements known for their potential to stimulate promoter activity and enhance RNA polymerase binding, intrinsic curvature of the DNA molecule (30, 37), and the presence of UP elements (11, 33).

Intrinsically curved (bent) DNA molecules most frequently appear when recurrent short sequences of A residues occur in phase with the B-DNA helical repeat (37). A stimulatory effect of bent DNA on promoter activity has been demonstrated in a number of cases (30). It appears unlikely that DNA geometry might appreciably influence the efficiency of the P_junc promoters of ISEc11/ISEc11p, since computer-generated bending predictions, obtained as described previously (31), indicated that the overall intrinsic curvature of a region encompassing the P_juncC and P_juncP promoters (from 300 bp upstream to 200 bp downstream of the transcription start) is weak and is not significantly affected by the presence of the 39-bp duplication (data not shown).

In many bacterial and phage promoters significant stimulation of the transcriptional activity has been shown to depend on the binding of the carboxy-terminal domain of the RNA polymerase α subunit to upstream sequences. These so-called UP elements are usually located between position −38 and position −59 with respect to the transcription start site and consist almost exclusively of A residues (11, 33). In terms of both nucleotide sequence and relative position, the presence of an A-rich stretch in the 39-bp duplication of P_juncP and its location (nucleotides −41 to −51) (Fig. 2B) correspond well to a potential role as an UP element.

Transposition of ISEc11 and ISEc11p into the target sequence.

As previously shown, analysis of the sequences flanking the ISEc11 element revealed the presence of a preferential target site. To demonstrate transposition of ISEc11 and ISEc11p, we constructed a specific entrapment vector, pTRIS0 (Ap^r) (Table 1), carrying the target sequence (5′-GTGAAATACTG-3′) and a cat reporter gene which cannot be expressed, since its promoter is incomplete (Fig. 4A) due to the lack of the −35 consensus and to replacement of the −10 box with the target sequence. Hence, the Cm^r phenotype can occur only if the regulatory region is restored by supplying a properly spaced −35 box together with a functional −10 box. On the basis of the sequences of ISEc11 and ISEc11p, insertion of one of these ISs into the pTRIS0 target site is expected to supply a −35 region and to create a functional −10 box, thus inducing expression of the cat gene (Fig. 4B).

FIG. 4.

Structure of entrapment vector pTRIS0. (A) Plasmid pTRIS0 is a pRS414 derivative that was created to analyze the transposition of ISEc11 and ISEc11p. The 5′-GTGAAAATACTG-3′ nucleotide stretch (gray background) corresponding to the target site in the yjgY-yjgZ locus was used as a substrate for ISEc11 or ISEc11p insertion. The target site was inserted upstream of a defective promoter of the vector-encoded cat gene. The cat gene was silenced due to the lack of a complete promoter region (+1 indicates the potential transcription start). (B) Integration of ISEc11 or ISEc11p created a functional promoter by supplying a −35 box, located just upstream of the right IR, and by completing a −10 box by insertion of right end terminal sequences. Integration of ISEc11 or ISEc11p restored expression of the cat gene.

We first introduced pTRIS0 into HN280, selecting for Ap^r. Next, three independent HN280(pTRIS0) transformants were grown overnight and plated on LB medium plates containing ampicillin and on LB medium plates containing ampicillin and chloramphenicol. We obtained Ap^r Cm^r colonies at a frequency of 3 × 10⁻⁶. To demonstrate that the emergence of the Cm^r phenotype was dependent on the presence of the IS in pTRIS0, DNAs from 120 HN280 Ap^r Cm^r colonies, derived from three different transformation experiments, were used as PCR templates with the PC1 and TR3 primers located in the IS element and in the cat coding sequence, respectively. All samples were found to contain an IS element at the target site. In particular, 42 of the 120 isolates contained ISEc11 sequences, while the remaining 78 isolates harbored ISEc11p sequences. These results indicate that Cm^r expression in all cases was induced by insertion of ISEc11 or ISEc11p into the target sequence, with the latter exhibiting higher efficiency despite its lower copy number. The data also demonstrated that the −35 region located upstream of the IRr sequence is functional, provided it is correctly spaced with respect to the −10 box.

DISCUSSION

Transposition of bacterial insertion sequences is generally tightly regulated in order to limit potentially detrimental effects of excessive genome rearrangements (27). At the transcriptional level Tpase synthesis is severely limited mainly by the presence of very weak indigenous promoters, generally located in the left end of the IS element. During the transposition process many ISs are able to assemble a new transient promoter, P_junc, which is able to stimulate Tpase expression significantly. The generation of P_junc depends on the formation of circular copies of the IS (36). It has been suggested that the greater strength of P_junc promoters provides a burst of Tpase gene expression, improving the chance of reintegrating the excised IS before the nonreplicating minicircle gets diluted in dividing cells (27). In this work, by analyzing the new insertion sequence ISEc11 and its natural variant ISEc11p, we found that the native great strength of a P_junc promoter is further increased by the presence of additional cis-acting upstream elements.

ISEc11 is a new IS element that is widespread in enteroinvasive E. coli strains. We found that in these strains, which together with Shigella strains are the etiological agents of bacillary dysentery (15), the ISEc11 element is always present on the pINV plasmid and also is very frequently present on the chromosome. pINVs of Shigella and EIEC strains are large (220-kb to 260-kb) F-type plasmids that contain all the genes required for invasion and for intra- and intercellular spread, including their positive activators (8, 12, 31). Despite the fact that pINVs are functionally interchangeable between Shigella and EIEC, the presence of a complete ISEc11 on the pINV plasmids is limited to EIEC, whereas Shigella pINVs harbor only remnants of this IS element. While the pINVs of four of nine EIEC strains (4608, 53638, 13.80, and 6.81) (6) used in this study harbor a native copy of ISEc11, in the remaining five strains (HN280, HN11, HN13, HN19, and HN300) (28) the pINV-located ISEc11 is represented by a natural variant, ISEc11p. Both ISs are integrated into the same plasmid locus. Compared to ISEc11, ISEc11p contains a 39-bp duplication, probably generated by insertion of the IS element into itself, followed by an excision event leaving only part of the right end of the excised IS (Fig. 1). Indeed, the duplication contains a perfect copy of the rightmost 39 bp of ISEc11, including the IRr and TSr elements.

ISEc11 and ISEc11p are able to form minicircles and contain subterminal 13-bp IR elements flanked by slightly different (3 bp) terminal sequences (Fig. 1 and Table 2). Upon circularization both elements give rise to a functional P_junc promoter consisting of a −10 box (5′-ATAAAA-3′), generated by joining of the TS through a 4-bp spacer (5′-AAAT-3′), and a −35 box (5′-TTGCAAA-3′) located in the IRr element. The AAAT sequence between the abutted TS elements is derived from the tetranucleotide direct repeats flanking the IS ends in the integrated form (Fig. 1). Although upon circularization ISEc11 and ISEc11p form a junction promoter carrying identical −10 and −35 boxes, β-galactosidase assays (Fig. 3B) revealed that the junction promoter of ISEc11p (P_juncP) is able to induce a 30-fold-higher level of TnpA-LacZ synthesis than the ISEc11 junction promoter (P_juncC) induces. The presence of only one starting point in primer extension experiments (Fig. 2A) rules out the possibility that the 39-bp duplication in P_juncP contains an alternative promoter.

To investigate the dependence of the enhanced P_juncP activity on trans-acting regulatory factors or on additional cis-acting elements, we analyzed in vitro transcripts generated by P_juncP and P_juncC. As shown in Fig. 3C, real-time PCR assays confirmed that P_juncP exhibits promoter activity that is up to about 20-fold greater than the promoter activity exhibited by P_juncC. Taking into account the fact that the difference is particularly significant at a low RNA polymerase concentration, it is likely that the presence of the 39-bp duplication strongly increases the affinity of P_juncP for the RNA polymerase. Recently, it has become increasingly evident that besides the classical −35 and −10 boxes, many promoters contain one or two recognition boxes for the carboxy-terminal domain of the RNA polymerase α subunit. These so-called UP elements are characterized by a high A content and are known to stimulate transcription up to 100-fold by facilitating the initial binding of RNA polymerase and the subsequent transcription initiation steps (11). In silico analysis revealed that the 39-bp duplication contains A stretches (from position −41 to position −51) which closely mimic the features of UP elements (Fig. 2B) and therefore could have the potential to act as transcription-stimulating cis elements.

Helically phased upstream A-rich regions have also been shown to play a critical role in altering the architecture of a promoter through creation of an intrinsically curved DNA structure that is potentially able to affect transcription (30, 37). Computer-based DNA curvature predictions (31) for a 500-bp region encompassing the junction promoters of ISEc11 and ISEc11p did not reveal significant overall curvature, irrespective of the presence of the 39-bp duplication (data not shown), making it unlikely that DNA bending contributes significantly to the efficiency of the P_junc promoters of ISEc11 and ISEc11p.

The role of the −35 box (5′-TTGCAA-3′) located near the IRr of ISEc11 is quite interesting in the case of the integrated IS. By means of transposition experiments we showed that ISEc11 and ISEc11p can transpose into a specific entrapment vector containing the target sequence. Insertion of these IS elements promotes the transcription of a reporter gene through the generation of a new promoter. This is determined both by the presence of an available −35 box in the IRr and by the creation of a new −10 region resulting from the addition of a short AT stretch to the target site (Fig. 4B). It is plausible that the promoter-promoting potential of ISEc11/ISEc11p might have been exploited during evolution to shape the transcription profiles of EIEC strains, enhancing their adaptability to new environmental conditions. By using in silico analysis we found that ISEc11 is homologous to putative IS elements present on the chromosome of pathogenic strains related to E. coli, like Shigella strains (38). In particular, Shigella sonnei harbors three complete copies of ISEc11 (one of them is interrupted by an IS1 element), while two complete copies are present in both S. dysenteriae and S. boydi. The presence of remnants of ISEc11 in the pINV plasmids of all Shigella spp. strains strongly suggests that this IS has played an active role in the evolution of pINVs. Such a strong association between an IS element and its host has been reported previously for the IS605 and IS607 elements and Helicobacter pylori, leading to the hypothesis that there is a dynamic coevolutionary process involving these ISs and their hosts (14, 16).

Several features indicate that ISEc11 and its variant, ISEc11p, belong to the IS1111 family; besides the presence of subterminal IRs and the homology of these IRs with those of other IS1111-type elements (Table 2), a relevant trait is the existence of a specific target site into which the IRs insert in a preferred orientation (26, 29). Another strong common feature is the presence of a Tpase lacking the classical DDE motif and exhibiting relevant similarities to the Piv protein, a site-specific invertase of Moraxella lacunata (18, 29). Recently, it has been shown that the catalytic domain of the Piv invertase contains a DEDD motif and that the four residues are required for inversion catalysis and for intramolecular recombination (3). Our in silico studies (Table 2) revealed that a DEDD motif is also present in the ISEc11/ISEc11p Tpase and that it is highly conserved (D₉E₅₂D₈₉D₉₂) in several members of the IS1111 family. The similarity of this Tpase to DNA recombinases belonging to the Piv/MooV family and the fact the 5′-AAAT-3′ tetranucleotide that acts as a spacer between the IS ends in the circular form is also present on both sides of the element in its integrated form strongly suggest that integration could occur by site-specific recombination rather than by transposition.

The high number of transposable elements identified so far and the availability of in vitro transposition assays have highlighted the variety of transposition strategies (1, 10, 13, 23, 32). A transposition mechanism involving site-specific recombination at a 3-bp core site, located within the spacer separating the IS ends in the circular form, has been described for IS1383 and IS117, two other elements that form circular intermediates (2, 26). To explain the presence of a 4-bp direct duplication in ISEc11/ISEc11p, it is possible to envisage that the circular form is generated by a cut at the end of the target site, leaving short single-stranded sticky ends which could be joined together to form a circular intermediate. Following the circularization event integration into a new target site may occur around a central recombination site consisting of the homologous tetranucleotides located within the target site, as well as within the spacer separating the TSs in the circular form.

Integration at specific target sites and the sophisticated complexity of the regulatory networks underlying Tpase expression stress the importance of the compromise evolutionary strategies used by transposable elements to cope with conflicting requirements, like the management of a potentially damaging transposition activity and the maintenance of the functional integrity of the host genome. Although the evolutionary advantage conferred by the increased Tpase levels associated with the presence of the ISEc11p is not clear yet, the structure of P_juncP nicely exemplifies how a highly efficient promoter can be generated by a comparatively simple event, such as the joining of small DNA consensus elements at suitable reciprocal distances.