Characterization of two alternative promoters for integrase expression in the clc genomic island of Pseudomonas sp. strain B13

Authors

  • V. Sentchilo,

    1. Process of Environmental Microbiology and Molecular Ecotoxicology, Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, Postfach 611, CH 8600 Dübendorf, Switzerland.
    Search for more papers by this author
  • A. J. B. Zehnder,

    1. Process of Environmental Microbiology and Molecular Ecotoxicology, Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, Postfach 611, CH 8600 Dübendorf, Switzerland.
    Search for more papers by this author
  • J. R. Van Der Meer

    Corresponding author
    1. Process of Environmental Microbiology and Molecular Ecotoxicology, Swiss Federal Institute for Environmental Science and Technology (EAWAG), Ueberlandstrasse 133, Postfach 611, CH 8600 Dübendorf, Switzerland.
      E-mail: vdmeer@eawag.ch; Tel. (+41) 1823 5438; Fax (+41) 1823 5547.
    Search for more papers by this author

E-mail: vdmeer@eawag.ch; Tel. (+41) 1823 5438; Fax (+41) 1823 5547.

Summary

The clc genomic island is a 105 kb integrative and conjugative element (ICE) in Pseudomonas sp. strain B13, which encodes metabolism of 3-chlorocatechol. The clc island is integrated in a tRNAGly gene, but can excise and form a circular intermediate in which both ends are connected. The integrase gene (intB13) of the clc genomic island is located at the right end, 202 bp from the junction site facing inwards. Fragments upstream of intB13 in the circular form and in the integrated form were fused to a promoterless gfp gene for Green Fluorescent Protein and introduced in monocopy onto the chromosome of strain B13. Quantitative GFP fluorescence measurements in individual cells of the different B13-derivatives revealed that the circular form fragment contained a strong constitutive promoter (Pcirc) driving intB13 expression in all cells. By using primer extension Pcirc could be mapped near the left end of the clc element and Pcirc can therefore only control intB13 expression when left and right ends are connected as in the circular form. Expression from intB13 upstream fragments from the integrated clc element was weaker than that from Pcirc and only occurred in maximally 15% of individual cells in a culture. A promoter (Pint) could be roughly mapped in this region by using reverse-transcription PCR and by successively shortening the fragment from the 5′ end. Transposon mutants in cloned left end sequences of the clc element were selected which had lost the activation potential on the Pint promoter and those which resulted in overexpression of GFP from Pint. The DNA sequence of the region of the transposon insertions pointed to a relatively well conserved area among various other genomic islands. The activator mutants mapped in an open reading frame (ORF) encoding a 175 amino acid protein without any significant similarity to functionally characterized proteins in the databases.

Introduction

The clc element is a genomic island found in Pseudomonas sp. strain B13 (Ravatn et al., 1998a, b). It has a size of approximately 105 kb and encodes among other functions the degradation of 3-chlorobenzoic acid (CBA) via chlorocatechol. The clc element is self-transmissible among bacteria from the classes ‘Beta-’ and ‘Gammaproteobacteria’ (Mokross et al., 1990; Zhou and Tiedje, 1995; Ravatn et al., 1998a, b; Springael et al., 2002) and has extensive regions of high percentage sequence identity to genomic islands in Pseudomonas aeruginosa (Larbig et al., 2002) and Xylella fastidiosa (Simpson et al., 2000; van der Meer et al., 2001), making it an interesting model to study the biology of an expanding class of mobile DNA structures referred to as ‘ecological fitness genomic islands’ (Hacker and Carniel, 2001). Genomic islands and conjugative transposons have recently been classified as members of a similar group of integrative and conjugative elements (ICE) (Burrus et al., 2002). Integrative and conjugative elements integrate by recombination between a specific site of the circular form (attP) and one or more target sites (attB, usually one for genomic islands) in the chromosome (or on large plasmids). They can excise from that location and form a circular intermediate in which both ends are connected. The circular form can transfer by conjugation to a new recipient cell in which reintegration can take place (Hacker and Carniel, 2001; Burrus et al., 2002). Genomic islands often contribute to pathogenicity of the host organism, but other factors like aromatic degradation or plant symbiosis have been found as well (for comprehensive review see Hacker and Kaper, 2002).

Comparatively little is known about the cellular factors and possible signals regulating excision and transfer of genomic islands. Like many genomic islands the clc element is integrated into a tRNA gene, in this case the tRNAGly. Integration is mediated by the IntB13 integrase, which belongs to the P4 type site-specific integrases but with moderate (35%) amino acid identity to those (van der Meer et al., 2001). The IntB13 protein is also about one-third larger in size than most other P4 type integrases (Ravatn et al., 1998c). Recombination takes place between an 18 bp sequence of the attP-site of the clc element identical to the most 3′ 18 bp from the tRNAGly (attB) which results in a duplication of the 18 bp sequence upon integration (attL and attR, Fig. 1). The intB13 integrase gene is located near the right end at a distance of 202 bp from the junction site facing inwards (Ravatn et al., 1998c). Functional studies on related integrases (Yu and Haggard-Ljungquist, 1993; Raynal et al., 1998; Marra and Scott, 1999; Cheng et al., 2000) give good reason to assume that excision of the clc element is also mediated by the IntB13 integrase, although this has not been experimentally proven and other auxilliary factors (e.g. excisionase) may be needed for that process. Because the integrase plays such an essential role in the cycle of excision and integration of bacteriophages and ICEs, regulation of integrase expression is of particular interest for understanding activity of those elements in relation to their host.

Figure 1.

A. Schematic presentation of the two forms of the clc genomic island, its life cycle and reactions catalysed by the IntB13 integrase. During integration the 18 bp 3′ end of the tRNAGly (attB) recombines with an identical 18 bp sequence (filled triangles) originating from the clc element (attP). This results in a duplication of the 18 bp sequences at the other end of the integrated form. Excision results in a closed junction between left (L) and right (R) ends of the element. Promoter regions of the integrase gene (intB13) of the integrated (Pint) and the excised circular form (Pcirc) of the clc element are depicted by thin arrows showing the direction of transcription (not to proportion).
B. Integrase promoter reporter cassettes jim1 (Pint-gfp) and jim4 (Pcirc-gfp). Coding sequences for the intB13′– green fluorescent protein (gfp) and kanamycin resistance (Km-R) genes are shown as arrows (not proportional). Solid vertical bars correspond to the I- and O-ends of the Tn5 delivery system.

As far as known, several strategies can be taken by bacteriophages and ICEs to achieve differential expression of the integrase gene. An important aspect in this differential regulation is the close location and inwards directionality of the integrase gene with respect to the junction site in the circular form, allowing readthrough from promoters through the junction site in the circular form and physical separation from the promoter in the integrated form. This situation of alternative promoters occurs for integrase expression in the Yersinia pestis HPI genome island (Rakin et al., 2001). As a matter of fact, many integrase genes related to intB13 within the P4 family are also oriented inwards with respect to their integration site (van der Meer et al., 2001) and therefore may be regulated similarly. The theme of readthrough transcription in the circular excised form also takes place in the conjugative transposon Tn916, but differently. In this case the integrase gene faces outwards to the junction site and transcription starts in the middle of the transposon at the tetM gene (Celli and Trieu-Cuot, 1998; Marra and Scott, 1999). In some ICEs there is no evidence for alternating control of integrase expression. For example, although the integrase genes in the conjugative transposon CTnDOT in Bacteroides and the mobilizable unit NBU1 both face inward to their elements, their expression occurs constitutively (Shoemaker et al., 1996; Cheng et al., 2000).

In this work we characterized two alternative promoters for expression of the intB13 integrase by studying gfp expression in Pseudomonas sp. strain B13 from single-copy chromosomal transcriptional fusions. Interestingly, by quantifiying GFP expression in single cells rather than from the population as a whole, we discovered that expression of the integrase in the integrated form of the clc element was subject to regulatory control, but occurred only in a small proportion of cells. This population-dependent effect on integrase expression may have not been seen in other studies which have used reporter genes for the whole population (like gus, lacZ or luxAB). Primer extension analysis, promoter deletion studies and reverse transcriptase polymerase chain reaction were used to further locate regions in both promoters at which trans-acting factors could exert transcriptional control. Further experimental evidence for transcription factors acting on integrase expression is presented by mutational analysis of the left-end extremity of the clc element.

Results

An outward facing promoter, Pcirc, at the left end of the clc element

The clc element can occur in two different configurations: (i) integrated in the tRNAGly gene and (ii) as a closed circular form in which both ends are connected (Fig. 1). In order to study whether both configurations would lead to different expression of the intB13 gene, DNA fragments upstream of intB13 in the integrated and in the closed circular form were tested for promoter activity. A DNA fragment of 480 bp encompassing the attP region directly upstream of intB13 in the circular form of the clc genomic island was fused to the gfp gene (jim4 construction, Fig. 1). Escherichia coli strains DH5α and CC118 λpir carrying pJAMA23-jim4 and pCK218-jim4 (Experimental procedures), respectively, produced green colonies on nutrient agar and appeared brightly fluorescent in UV microscopy, suggesting that this attP fragment upstream of intB13 carried a constitutive promoter. Essentially the same result was obtained when GFP expression from the jim4 construction was studied in Pseudomonas sp. strain B13. Individual cells of Pseudomonas sp. strain B13-jim4 were brightly fluorescent both during exponential and stationary growth phase, on nutrient broth and on minimal medium with fructose or CBA as C-source. Average grey values (AGV) of individual cells, a quantitative measure for GFP fluorescence (see Experimental procedures), were normally distributed around a mean value of 200 AGV units during exponential growth and 350 AGV units during stationary phase. Four clones of Pseudomonas sp. strain B13-jim4 picked at random from the transconjugant plates all behaved alike with respect to the magnitude and distribution of GFP fluorescence values (data not shown), implying that the nature of gfp expression from this DNA fragment was not influenced by the position of insertion in the B13 chromosome. These observations showed that expression of the integrase gene in the circular form of the clc genomic island is controlled by a strong constitutive promoter (which we called Pcirc), present on the 480 bp attP fragment. Second, transcription from Pcirc did not seem to require B13- or clc-island-specific regulatory factors as it took place in two E. coli strains devoid of the clc island. Third, expression from Pcirc did not seem to require any particular growth phase or growth substrate.

To map the transcription initiation site within Pcirc, total RNA from E. coli strain DH5α (pJAMA23-jim4) was isolated and reverse transcribed using primer SV004-BamHI (Fig. 2). A single cDNA fragment was found, from which the transcription initiation site was mapped within the ‘left end’ of the clc element, i.e. 61 bp upstream of the junction between left and right ends (the junction lies 202 bp upstream of the intB13 start codon) (Fig. 2). The transcription initiation site on the sequence was preceded by strongly conserved − 35, TTGGCG, and − 10, TACAAT, hexamer boxes separated by a 17 bp spacer which is typical for σ70 dependent promoters (Harley and Reynolds, 1987; Wosten, 1998).

Figure 2.

Structure of the Pcirc integrase promoter in the free circular form of the clc genomic island.
A. Mapping of the transcription initiation site with primer extension. Lanes: RT, reverse transcription reaction with total RNA from E. coli DH5α (pJAMA23::jim4) extended with primer SV004-BamHI-R; A, C, G and T, corresponding DNA sequence obtained with the same primer. The arrow marks position of the primer extension product. Coding strand DNA sequence of the promoter area surrounding the transcription start site is shown on the right. First transcribed nucleotide marked + 1 and − 10 hexamer (−10) are shown in boldface.
B. Structure of Pcirc at the junction between left and right ends of the clc island. Boxed are the 18 nucleotides (attP) homologous to 3′ end of the tRNAGly (see also Fig. 1). Nucleotides of the transcription start (+ 1) and those matching − 10 and − 35 consensus hexamers are printed in boldface. The intB13 translation start codon is underlined and marked with the character M. Sequences of the primers SV004-BamHI-R and Jim4.1 used for RT are shown underneath the sequence. Note that the left end of the clc-island provides all DNA elements relevant to promoter functioning. Arrows points to two remarkable inverted repeats.

Integrase promoter, Pint, of the integrated clc element

In the integrated form of the clc island expression of the integrase must be directed by an alternative promoter, because the left end is displaced by the gene for tRNAGly (Fig. 1). This promoter was assigned Pint. To provide an insight into the structure and functioning of Pint two approaches were employed: (i) transcription initiation site mapping by reverse transcription (RT) or reverse transcription PCR (RT-PCR) and (ii) promoter deletion mapping by fusing different promoter fragments to the gfp gene and studying GFP expression in strain B13. For primer extension total RNA was isolated from cultures of the strain B13-jim1 carrying plasmid pTCB207-6 grown on MM with 10 mM CBA until late stationary phase (64 h incubation). Strains with this plasmid were found to have the highest GFP expression from the Pint promoter (see below). Unfortunately, no cDNA transcript could be detected in three independently repeated experiments neither with the IRD-800 labelled primer SV004-BamHI nor with a radioactively labelled primer int271-Rev (Fig. 3), despite the fact that both int271-Rev and SV004-BamHI efficiently and specifically amplified DNA in a PCR with B13 chromosomal DNA as a template. The most likely explanation for the failure to detect any specific cDNA fragment is a very low abundance of the intB13 transcript in the B13 RNA sample. For this reason, we tried to determine the transcription start site at least roughly by detecting the presence or absence of an amplicon in RT-PCR using a fixed reverse primer and differently located forward primers (jim1 through jim9, Fig. 3). Reverse transcriptase-PCR products of the expected size were obtained with forward primers jim3, jim6, jim7 and jim9 but not with jim1, jim2 and jim2.6 (Fig. 3). This confined the position of the Pint transcription initiation site to the area between the primers jim2.6 and jim6 or overlapping with jim6 (Fig. 3). Minor DNA bands were occasionally also observed in the reactions containing forward primers jim2 and jim2.6, as well as with primer jim4.1, which is annealing to the left end portion of the circular form of the clc island (Fig. 2). As the amount of PCR product from these upstream primers, whenever present, was always lower than in the RT-PCR reactions containing primers annealing downstream of the proposed transcription initiation site, these minor products seem to result from intB13 RNA transcripts starting from the Pcirc promoter. In all stationary phase cultures, the circular form of the clc element can be detected by Southern hybridization at a relative amount of between 5 and 15% compared to both chromosomal copies (V. Sentchilo, unpubl. obs.), which would be a constant source for Pcirc-intB13 transcripts.

Figure 3.

Structure of the Pint integrase promoter.
A. Coding strand DNA sequence at the junction between the host chromosome and the right end of the clc-island is shown. Nucleotide numeration corresponds to the intB13 area sequence deposited in GenBank under Accession nr. AJ004950 (Ravatn et al., 1998c). The sequence of the tRNAGly gene is boxed and a vertical line marks the start of the 18 bp recombination site. The integrase gene start codon is shown in boldface and underlined. Forward and reverse primers used for constructing Pint-gfp fusions are depicted by solid arrows in 5′−3′ orientation above and below the DNA sequence. Restriction sites for cloning are indicated underneath the sequence with mutated nucleotides shown in bold. The two inverted repeats are indicated by arrows (IR1 and IR2). Approximate positions of the repressor- and activator-binding sites are depicted with dashed lines above the sequence. The beginning of the intB13 transcript originating from Pint is shown as a zig-zagged arrow.
B. RT-PCR mapping of the 5′ end of the Pint-intB13 transcript. Total RNA from stationary phase culture of B13-jim1 (pTCB207-6) grown in MM with 10 mM CBA was reverse transcribed with primer int271-R. The resulting cDNA was amplified with PCR using reverse primer int271-rev in combination with one of the forward primers (listed according to their position in the sequence) jim1, jim2, jim2.6, jim6, jim7 and jim3. Lanes: – RT, RT reaction without reverse transcriptase added; + RT, complete RT reaction; C, positive control with genomic DNA of B13-jim1 as template; M, DNA size-marker, sizes of the bands in base pairs are shown between the panels.

IntB13′::gfp expression from Pint 5′-deletion mutants

In order to confirm the location of the transcription initiation site of Pint and to map possible functional promoter and/or operator elements, six transcription fusions were constructed by cloning fragments upstream of intB13 in its integrated form subsequently deleted from the 5′ end in front of the gfp gene (Fig. 3). Each promoter-gfp fragment was inserted in single copy into the Pseudomonas sp. B13 genome by mini-Tn5 transposition. Compared to GFP expression from Pcirc the GFP fluorescence levels of individual cells of strains B13-jim1, -jim2, -jim3, -jim6, -jim7 and -jim9 were all much lower. Cells growing on liquid MM with 10 mM CBA in batch displayed maximum GFP fluorescence levels in stationary phase (after 72 h incubation). In contrast to B13 cells carrying the Pcirc-gfp fusion, expression from the Pint-gfp fusions did not lead to a homogeneous population distribution of GFP in individual cells. For example, the longest construct jim1 led to the formation of only about 15% of cells with visible fluorescence. Therefore, statistical parameters other than the population mean and average had to be used. The most optimal descriptors for fluorescence differences among the B13 derivatives carrying Pint-gfp fusions were the 95% percentile (i.e. the fluorescence value below which 95% of all individual cells in a population would fall) and the top 5% value (i.e. the mean fluorescence of those 5% of individual cells with the highest GFP fluorescence). For B13-jim1, the 95% percentile and the top 5% value were 100 and 108 AGV units respectively (Table 1).

Table 1. . Effect of 5′ deletions on the expression of Pint-gfp reporter gene fusions.
Reporter strainaSize ofdeletion, bpGFP fluorescence95% percentilecTop 5% meand
  • a

    . Reporter strains Pseudomonas sp. B13-jim1, -jim2, -jim6, -jim7, -jim9 and -jim3 (listed according to the size of the promoter region deleted) were grown in liquid MM with 10 mM CBA until stationary phase (72 h).

  • b. The length of the promoter region is 232 bp counting from the intB13 start codon (Fig. 3).

  • c

    . GFP fluorescence intensity is given as the 95% percentile average grey value of all cells in the population. Values in parenthesis represent the 95% confidence intervals calculated with the R-software (see Experimental procedures).

  • d

    . Values given are the arithmetic mean of the cellular average grey values for those 5% of the population with highest fluorescence intensities.

B13-jim1  0b100 (99–101)108
B13-jim2 56161 (158–164)180
B13-jim6114120 (118–122)134
B13-jim7143 83.5 (83.3–83.8) 86.6
B13-jim9178 85.2 (85.1–85.3) 85.9
B13-jim3198 84.2 (84.1–84.2) 85.2
B13  79 79
Detection limit  79 

Shortening the promoter fragment as in the construct jim2 (Fig. 3), significantly increased fluorescence values of single cells compared to jim1 (i.e. 161 versus 100, respectively) and increased the proportion of fluorescent cells to 18% (Table 1). Subsequent shortening to the size encompassed by the jim6-gfp fusion resulted in a decreased GFP expression, but remaining significantly higher than in jim1 (13% induced cells in stationary phase, 95% percentile of 120 AGV units) (Table 1). Further shortening of the promoter region beyond the primer jim6 abolished gfp transcription completely as virtually no GFP was produced in B13 strains –jim7, -jim9 and –jim3 (Table 1). For every promoter fragment, four independently selected clones of strain B13 were assayed, which all behaved alike with respect to GFP expression (not shown). This indicated that the position of the promoter-gfp fusion in the B13 genome played no role in the observed GFP expression, but that variations in GFP expression were due to deletions of essential promoter/operator elements. These observations, first, confirmed the approximate location of the transcription initiation site obtained by RT-PCR analysis, i.e. lying within the area adjacent to primer jim6 (Fig. 3), as shortening the promoter fragment to beyond jim6 abolished GFP expression. Second, they pointed at the possible involvement of repression and activation mechanisms regulating expression from Pint, as shortening the fragment from jim1 to jim2 actually caused an increase in GFP expression (as if a repressor binding site was removed), but further shortening to jim6 again caused a decrease (as if an activator binding site was subsequently affected).

Cloning and mapping of genes involved in regulation of the integrase expression

Preliminary sequence data of a 9 kb region at the left end of the clc element, containing gene homologues of phage- and plasmid-type transcription regulators (see below) led us to the idea that some regulatory factors for intB13 expression were encoded in this area. To test this hypothesis, two adjacent EcoRI fragments spanning together the 9 kb left end region were separately brought into strain B13-jim1 on the broad-host-range hybrid vector pKT230-pUC28 (pTCB207 and pTCB208, Fig. 4) and mutagenized by transposon insertions.

Figure 4.

Genetic organization of the 6 383 bp EcoRI fragment present in pTCB207 and location of the < TET-1 > transposon insertions.
A. Junction of the left end (attL) of the clc island (depicted by a solid line) with the chromosome (dashed line), with the location of the 4.2 and 6.4 kb EcoRI fragments (Ravatn et al., 1998b).
B. Genetic map of the insert of pTCB207. Relevant features of pKT230 and pUC28 adjacent to the 6.4 kb EcoRI insert are shown. Direction of transcription from the promoter of the kanamycin-resistance gene is indicated. Open arrows (drawn to proportion) mark the positions of major open reading frames (orf1–7) identified with DNASTAR. A thick outline marks the gene for the putative activator of the integrase (inrR). Location of the < TET-1> transposon insertions is depicted by triangles and numbers of the corresponding pTCB207 mutant plasmids are shown underneath. Filled triangles indicate insertions that caused ‘up’ or ‘down’ regulation of Pint-gfp expression. Open triangles points to insertions without measurable effect on GFP expression.

Introduction of pTCB208 carrying the most exterior 4.2 kb EcoRI fragment (Fig. 4) into strain B13-jim1 exerted no effects on the magnitude of GFP expression per cell nor on the relative fraction of induced cells in the whole population (not shown). In contrast, introduction of the plasmid pTCB207, carrying the 6.4 kb EcoRI fragment located further inwards from the left end (Fig. 4) caused measurably increased GFP production compared to B13-jim1 devoid of any plasmid (Table 2). Plasmid pTCB207 was then in vitro mutagenized with the tetracycline resistance insertion cassette < TET-1 > in order to produce interruptions of any relevant gene functions. Twelve pTCB207 transposon mutants (Fig. 4) were again introduced via conjugation into B13-jim1 and the effects on GFP expression were studied in late stationary phase after growing in MM with 10 mM CBA as C-source. Two clones of B13-jim1 carrying plasmids with transposon insertions in the right proximity of the 6.4 kb insert (pTCB207–41and − 46; Fig. 4) displayed significantly less GFP both in fluorescence of individual cells and in the fraction of induced cells in the population compared to B13-jim1 (pTCB207) (Table 2). Conversely, three other transposon insertions all clustering between co-ordinates 2 and 3 kb, as in pTCB207-6, − 52 and − 64 (Fig. 4), caused a significant increase of both the fraction of induced cells in the population and the levels of GFP fluorescence in individual cells compared to B13-jim1 (pTCB207) (Table 2). The seven remaining transposon insertions in pTCB207 did not measurably change the character of Pint expression as exemplified by pTCB207-7, − 48 and − 62 (Table 2).

Table 2. . Regulatory effect of pTCB207 and its selected transposon insertion mutants on GFP expression from different reporter constructs in Pseudomonas sp. strain B13.
StrainPlasmidGFP fluorescenceb 95% percentileTop 5% mean
  • a

    . Reporter Pseudomonas sp. strains B13-jim1, -jim2 and -jim6 (with or without plasmid to test) were grown in MM with 10 mM CBA until late stationary phase (7 days).

  • b. As for Table 2.

  • c. Highlighted with bold face are pTCB207 derivatives where < TET-1> transposon insertions resulted in up- or downregulation of the Pint-gfp fusion.

B13-jim1aNone 97 (93–99)c111
pTCB207107 (105–109)118
pTCB207–6c122 (119–125)140
pTCB207-7108 (107–110)122
pTCB207-4197 (95–99)111
pTCB207-48106 (103–108)119
pTCB207-62105 (103–107)115
   
B13-jim2None137 (125–146)200
pTCB207180 (166–191)238
pTCB207-6223 (216–232)286
pTCB207-41149 (140–161)213
   
B13-jim6None121 (119–123)134
pTCB207-6166 (163–169)198

Repressor and activator binding sites

To further investigate the nature of the ‘up’ and ‘down’ transposon insertions, GFP expression was assayed in two B13 derivative strains with the Pint promoter deletions fused to gfp in the presence of pTCB207, pTCB207-41 (with the ‘down’ mutation) or pTCB207-6 (with the ‘up’ mutation). Like in B13-jim1, introducing the plasmid pTCB207 into B13-jim2 (Fig. 3) resulted in an enhanced GFP production, i.e. the 95% percentile value increased with 43 and the top 5% mean value with 38 AGV units (Table 2). With pTCB207-41, the stimulation was weakened (Table 2). Perhaps unexpectedly, introduction of the plasmid pTCB207-6 into B13-jim2 resulted in an ‘up’ phenotype of GFP expression, and even in B13-jim6 an ‘up’ phenotype was observed (Table 2).

DNA sequencing and sequence analysis.

The sequence of the DNA region present in pTCB207 was determined. This region is 6383 bp long with an average G + C content of 64.3%. A blastn search (Altschul et al., 1997) revealed that the nucleotide sequence of this 6.4 kb area overall has a high level of identity to contiguous DNA regions in the several other microbial genomes (Table 3). Protein coding sequences of seven major open reading frames identified in pTCB207 were assigned orf1 through 7 (Table 3, Fig. 4). The amino acid sequences predicted from the seven ORFs also matched quite well to the predicted ORFs in the genomes mentioned above (Table 3). However, most of predicted proteins carried no known or hypothetical functions, except for the ORF1 and ORF3 proteins, which beared conserved domains of the ParA and ParB protein families respectively (Table 3). Two transposon insertions with lower GFP production from Pint (nrs. 46 and 41) mapped in orf7. The three upregulating < TET-1> transposon insertions were mapped in orf3, but two other neutral insertions as well (i.e. pTCB207-10 and − 48, Fig. 4).

Table 3. . Location, proposed function and homologies of peptides predicted from ORFs within the 6.4 kbp EcoRI fragment at the left end of the clc genomic island.
ORFORF locationaLength,nucleotides/amino acidsProposed functionPercentage of amino acids identity to ORFs in
BfubXfaRmeXaxPagi-2Pagi-3SenPfl
  • a. First and last nucleotides as in the sequence deposited at GenBank under Accession number AJ536665.

  • b

    .Bfu, Burkholderia fungorum, unfinished genome (NZ_AAAC01000104); Xfa, Xylella fastidiosa, complete genome (AE004000); Rme, Ralstonia metallidurans, unfinished genome (NZ_AAAI01000352); Pagi-2, Pseudomonas aeruginosa genomic island PAGI-2 (AF440523); Pagi-3, P. aeruginosa genomic island PAGI-3 (AF440524); Xax, Xanthomonas axonopodis, complete genome (AE011858); Sen, Salmonella enterica, complete genome (AL513382); Pfl, P. fluorescens, unfinished genome (NZ_AABA01000162).

  • c. Similarity is not detected with blastP (Altschul et al., 1997).

  • d

    . A region between amino acid residues 104 and 194 aligns with the protein domain conserved among ParA-type ATPases (position 22–106) [pfam0991, E-value 0.007].

  • e

    . Protein region between amino acid residues 39 and 102 aligns with ParBc, ParB-like nuclease domain (position 3–93) [pfam02195; E-value 8E-05].

  • f

    . ORF3 corresponds to two consecutive ORFs in X. fastidiosa (XF1784 and XF1783).

ORF130–905876/291Conserved hypothetical protein
Similarity to members of ParA familyd
100919376937139– c
ORF2889–1146258/85Conserved hypothetical protein7659
ORF31139–27911653/550Conserved hypothetical protein
Similarity to members of ParB familye
10076, 91f815886614242
ORF42807–3367561/186Conserved hypothetical protein10093907390683441
ORF53243–46161374/457Conserved hypothetical protein100807562826331
ORF64947–5726780/259Conserved hypothetical protein10091867790802852
ORF7(inrR)5723–6250528/175Conserved hypothetical protein
Activator of integrase expression
100858665866745

Discussion

The work presented here firmly established that transcription of the intB13 integrase gene in the clc genomic island occurs from two alternative promoters, Pcirc and Pint, depending on the element's state in the cell. In the closed circular form intB13 is transcribed from the constitutive Pcirc promoter. This promoter directs transcription from the left end through the junction into the right end and the integrase gene. In the integrated clc element the arrangement of the integrase promoter is different, as a result of displacement of the left end promoter by the tRNAGly gene.

Having two functionally distinct promoters for integrase expression seems to be more common in ICEs and bacteriophages. Perhaps its significance lies in the necessity for these elements to achieve site-specific integration of the circular intermediate after conjugation or infection. This can be ensured by having integrase expression in the circular form under control of a strong constitutive promoter, such as Pcirc, and producing temporarily larger amounts of integrase enzyme. Such strategies occur in phage P4 (Pierson and Kahn, 1987), in the HPI pathogenicity island of Yersinia (Rakin et al., 2001) and the Streptomyces bacteriophage φC31 (Kuhstoss and Rao, 1991). Perhaps exceptionally, this promoter metamorphosis and high integrase expression can lead to very unstable chromosomal structures (Ravatn et al., 1998b). When the clc island was conjugated into P. putida F1 multiple tandem copies of the clc element were formed upon integration (Ravatn et al., 1998b). In such tandem copies the left end and right end are again connected with the result that intB13 expression is brought back under control of the Pcirc promoter and much more frequent excision and recircularization occurs.

IntB13 expression from Pint had a completely different behaviour than from Pcirc. The weak transcriptional activity from Pint is probably reflecting the true nature of the genomic island, namely favouring the integrated state. Very unusually and not described before, induction from Pint was not only weak but confined to a small subpopulation of cells. This may be much more common for other ICEs, which, however, have not been studied with promoter gfp fusions and where one could not have detected population dependent gene expression. Despite the population-dependent gene expression effect, the results obtained here were largely consistent with the hypothesis that expression from Pint is subject to positive and negative transcription control, similar to dual control of phage integrase expression (Saha et al., 1987a, b; Friedman, 1992; Liu et al., 1997; Eriksson et al., 2000). All data so far confirm that the product of orf7 would encode a transcriptional activator for integrase expression. Therefore, we propose to name this ORF inrR, for integrase regulator. Mutations in inrR abolished the activation effect and deletion of the promoter region to beyond jim6 abolished activation potential as well. The results of transposon mutagenesis were not so clear for the proposed repressor function. A repressor function was postulated on the basis of increased GFP expression when deleting part of the promoter/operator region (as in the jim2-gfp fusion). However, although an ‘up’ phenotype was observed among the transposon mutants in plasmid pTCB207 which all mapped in orf3, it is unlikely that the orf3 gene codes for the actual repressor. First, because two other transposon insertions in the same ORF had no effect on GFP expression levels. Second, because the same transposon mutants that caused the ‘up’ phenotype in B13-jim1, also caused an ‘up’ phenotype in B13-jim2, in which the presumed repressor binding site was removed. This suggests that the orf3 product indirectly affects intB13 expression and that yet another protein binds to the ‘repressor binding site’ upstream of intB13 (Fig. 3).

It was only partially possible to map Pint, which, however, still allows some conclusions on possible involved sigma factors to be drawn. The 5′ end of the intB13 transcript in the integrated form could be confined to a region between primer 2.6 and the 5′-half of jim6 (Fig. 3). There are conserved − 35-(TTGAAA) and − 10-like (TTTTTT) boxes in this region, 17 base pairs apart, but these lie too close to the start of the intB13 transcript identified by RT-PCR (e.g. the TTTTTT box lies completely within the jim6 primer region). This suggests that other sigma factors than σ70 might be involved in recognizing Pint, which is not uncommon for integrase promoters from other elements. For example, the phage λ integrase promoter Pi has only a − 35-like element (Davies, 1980), whereas in phage 186 the Pe promoter for immunity repressor and the integrase only has a − 10-like element (Shearwin and Egan, 2000). Activation of these promoters is achieved by recruitment of RNA polymerase with CII transcriptional activators (Ho et al., 1983; Shearwin and Egan, 2000). Similarly, the promoter of the repSA-xis-int operon in the pSAM2 integrative and conjugative element in Streptomyces is devoid of both −10 and −35 boxes but requires the Pra protein for transcription activation (Sezonov et al., 1998, 2000).

It appeared that the gene cluster at the left end of the clc island is well conserved also in other bacterial genomes (Table 3). At least four of these clusters occur on genomic islands, i.e. PAGI-2 and PAGI-3 in P. aeruginosa (Larbig et al., 2002); one in R. metallidurans CH34 (Larbig et al., 2002); and one in X. fastidiosa (van der Meer et al., 2001). All these elements also carry integrase genes downstream of tRNAGly genes and strongly related to intB13. As the overall DNA sequences contained by the genomic islands of X. fastidiosa and P. aeruginosa are very different (Larbig et al., 2002), the conserved left end region may be especially important for regulation of the integrase expression, as shown here for the InrR protein, but also for excision or conjugative transfer. Although we could not firmly establish the role of the ORF3 peptide, its homology to the ParB family of transcriptional repressors was significant and may indicate a role in regulating conjugative transfer. Upstream of orf3 and in the same direction was located an ORF (orf1), which can encode a protein with an amino acid sequence motif conserved for ParA. The parA parB pair of genes is a well known couple of chromosome- and plasmid-partitioning determinants but have other roles as well (Gerdes et al., 2000). For example, the korA/incC-korB pair in the central control operon of IncP-1 plasmids (Pansegrau et al., 1994) determine plasmid partitioning but also exert global transcription control over plasmid replication and conjugative transfer (Bechhofer et al., 1986; Jagura-Burdzy and Thomas, 1994; Zatyka et al., 1997; Jagura-Burdzy et al., 1999). A ParB-like transcription regulator was also described in Shigella spp. controlling not partitioning but pathogenicity (Watanabe et al., 1990; Beloin et al., 2002).

Which environmental or physiological conditions exactly cause the integrated Pint promoter to become activated is not yet very clear, although we observed that GFP expression was highest in stationary phase cultures. Perhaps induction of the integrase of the clc genomic island is part of a cascade aimed to rescue the island to another recipient under unfavourable environmental conditions. Unraveling the regulatory network that links the host physiology to the activity of the clc genomic island may turn out to be exemplary for a number of other genomic islands, which seem to be more prevalent than previously thought.

Experimental procedures

Strains and plasmids

Escherichia coli DH5α (Gibco BRL, Life Technologies) was used routinely for plasmid propagation and transformation. Escherichia coli HB101 (Promega) was used for the propagation of the helper plasmid pRK2013 (Ditta et al., 1980), E. coli CC118 λpir (Kristensen et al., 1995) was used to maintain plasmid pCK218 (Kristensen et al., 1995) and its derivatives. Escherichia coli S17-1 (Simon et al., 1983) was used as donor for plasmids derived from pKT230 (Bagdasarian et al., 1981) Pseudomonas sp. strain B13 (Dorn et al., 1974) is the original host of the clc element.

Promoter reporter construction

The approach used to construct integrase promoter reporters was as follows. Briefly, six DNA fragments in the range from 70 to 274 bp covering the integrase promoter region at attR were amplified by using the PCR with plasmid pRR108 (Ravatn et al., 1998c) (see Fig. 3). Primers for PCR amplification were designed such as to introduce SphI, PstI or BamHI restriction sites (Figs 2 and 3) in order to facilitate subsequent cloning of the amplified fragments to a promoterless gfp gene. The gfp gene used here coded for the F64L,S65T enhanced green fluorescent mutant protein (Miller and Lindow, 1997) and was subcloned into plasmid pJAMA23 (Jaspers et al., 2001). Inserts covering parts of the intB13 upstream region were sequenced to detect possible changes introduced by PCR amplification. A 480 bp EcoRI fragment containing the region upstream of the intB13 gene in the circular form of the clc element was retrieved from plasmid pRR146 (Ravatn et al., 1998c) and cloned into pJAMA23 in front of the gfp gene. All resulting intB13 promoter-gfp fusions were retrieved as single NotI fragments and cloned into the mini-Tn5 delivery plasmid pCK218 (Kristensen et al., 1995), thereby replacing the luxAB genes.

Subsequent random insertion of the intB13 promoter-gfp fragments into the genome of Pseudomonas sp. strain B13 was achieved through mobilization of the pCK218 derivative plasmids from E. coli CC118λpir by helper plasmid pRK2013 in a triparental filter mating as described by Herrero et al. (1990). The presence and sizes of the intB13 promoter-gfp fragment and the absence of the plasmid backbone in B13 transconjugants were verified by antibiotic resistance profiling and by using the PCR with two construct-specific primers (i.e. Ter-Fw 5′-CAGGAATTTCGAGGCATGC-3′ and gfp-R, 5′-GTATGTTGCATCACCTTCACC-3′). Clones with correct insertions were assigned Pseudomonas sp. strain B13 (jim1 through 9) (see Fig. 3 for details).

Media and growth conditions

Luria–Bertani broth (LB) was routinely used for growing E. coli and Pseudomonas strains. As a defined mineral medium (MM) the type 21C mineral medium (Gerhardt et al., 1981) was used, supplemented either with 10 mM CBA or with 10 mM fructose. When required 50 µg of ampicillin, kanamycin, rifampin, streptomycin and/or 5 µg tetracycline per ml were added. Strains of Pseudomonas were grown at 30°C, those of E. coli at 37°C.

Cultivation of the reporter strains

Constructed reporter strains of Pseudomonas sp. B13 were stored in 20% glycerol stocks at − 80°C and whenever necessary plated on MM agar with the appropriate antibiotics and 5 mM CBA as sole carbon source. To obtain a liquid culture for microscopy a single colony was inoculated into 5 ml of MM supplemented with 10 mM CBA or fructose in a 35 ml glass tube and shaken at 200 r.p.m. Typically, exponentially growing cultures were obtained within 16 h, and after between 24 and 48 h of cultivation the cultures reached stationary phase.

Fluorescence microscopy

The GFP fluorescence intensities of individual cells of Pseudomonas sp. strain B13 derivatives were examined with an Olympus BX50 epifluorescence microscope. Cells from liquid cultures were tenfold concentrated by centrifugation and resuspension, applied to glass slides, covered with a cover slip and directly studied under the microscope. Cells attached to the cover slip were focused. Images of at least 200 cells per field were taken with a cooled CCD camera (Photometrics SenSys:1401E, Roper Scientific, USA), a 100 ×/1.30 oil immersion lens (UPIanF1, Olympus, Japan) and an exposure time of 300 ms. The filter used for GFP fluorescence was HQ-EGFP (F41-017, AF Analysentechnik, Tübingen, Germany) with emission filter HQ 525/50, beamsplitter Q 495 LP and excitation filter HQ 470/40. Both phase-contrast and GFP fluorescence images of 6–12 randomly chosen fields per sample were recorded with the acquire function of the metamorph Software (Version 4.6, Universal Imaging). Samples to be directly compared were analysed on the same day to avoid possible UV light source intensity variations.

Image analysis

Quantification of fluorescence intensities of individual cells in an image was performed in an automatic subroutine within the metamorph software. To account for background variation, a low pass function was generated, which was subtracted from the original image. Then, a lower threshold was defined to remove background and to allow automatic object identification. The average grey value (AGV), the total grey value of all pixels per cell and the total area were recorded automatically for each individual object (cell) as well as summed up for all objects of an image. Average grey values were then pooled together for all images of each particular culture sample or condition. In each culture sample a minimum of 2000 cells was analysed. A value of 79 AGV corresponds to the background level of the camera system (i.e. cells are dark and have no detectable fluorescence).

Population analysis

Differences in cellular average grey values among samples were analysed by determining the distribution of AGVs for all cells in a population. For this purpose, the AGVs of all cells were sorted, ranked and plotted against their position in the cumulative distribution curve (i.e. the ranking number divided by the total number of cells in the population, multiplied by 100). A statistical subroutine of the program r (Ihaka and Gentleman, 1996) (http://www.cran.r-project.org) was written in order to analyse differences in the cumulative distribution curves based on non-distribution type functions. The most prominent parameters for comparing statistical differences among samples were found to be the AGV (or GFP fluorescence value) corresponding to the 95% percentile of the sample population and the arithmetic mean of the cellular AGV values for those 5% of the population with the highest single AGVs (indicated as ‘Top 5% Mean’). Intervals for 95% confidence were calculated for each of the derived 95% percentile fluorescence values based on 200 bootstraping cycles.

Transcription initiation start mapping using reverse transcription

Total cellular RNA was isolated using the acid phenol extraction protocol (Aiba et al., 1981) followed by DNAse I (Amersham) treatment, phenol/chloroform purification and precipitation with 70% ethanol and 0.3 M sodium acetate (pH 5.2). The reverse transcriptase (RT) reaction was done with 1 µg total RNA by using the Enhanced Avian RT First Strand Synthesis kit (Sigma) according to the guidelines of the supplier (Sigma). The reverse transcriptase reaction was initiated from primer SV004-BamHI-R (5′-GAGGATCCTAAG TAATGAC-3′), which was labelled at its 5′ end with the fluorescent dye IRD-800 (purchased from MWG Biotech, Munich, Germany). The resulting RT products were directly analysed on an automated sequencer (model 4200IR2; LI-COR, Lincoln, NB). Parallel reference sequencing reactions were produced with the same primer on plasmid pJAMA23-jim4 by using the Thermo Sequenase fluorescence labelled primer cycle kit with 7-deaza-dGTP (Amersham). To approximately map the transcription initiation start in those RNA samples with low abundance of intB13 specific RNA transcripts we carried out RT-PCR. cDNA obtained by the RT reaction on total RNA with primer int271-R (5′-GACTCGC CCTTGGTACTGCTCGT-3′) was amplified by PCR as follows. 50 µl RT reaction mixture was treated for 1 h with 1 unit of DNAse-free RNAse (Boehringer Mannheim) and 5 µl used in the PCR with either one of the forward primers jim1, jim2, jim2.6, jim3, jim4.1, jim6, jim7 and jim9 (Figs 2 and 3) in combination with the reverse primer int271-R. Each PCR reaction mixture contained Sigma PCR buffer, 250 µM of each deoxynucleoside triphosphate (Sigma), 1 unit of TAQ polymerase (Sigma), and 1 pM of each primer in 50 µl final volume. The following cycling parameters were implemented: 2 min hot-start at 95°C followed by 35 cycles of 30 s denaturation at 93.5°C, 30 s of annealing at 50°C and 45 s of extension at 72°C, with 4 min of final extension at 72°C. As negative control for the RT-PCR reaction total RNA in RT reaction mix but without reverse transcriptase enzyme was used. As positive control, purified genomic DNA of strain B13-jim1 was amplified in the PCR, without RT-step.

Cloning and mapping of genes involved in Pint regulation

Two EcoRI fragments of 4.2 and 6.4 kb size, originating from the left end of the clc element (see Fig. 4) were cloned from the cosmid 3G3 (Ravatn et al., 1998b) into pUC28, which resulted in plasmids pTCB172 and pTCB177 respectively. Plasmids pTCB172 and 177 were cleaved with HindIII and ligated with pKT230 (Bagdasarian et al., 1981) giving rise to pTCB208 and pTCB207 respectively. Both plasmids contained stable coreplicons and were maintained in E. coli S17-1 under ampicillin and streptomycin selection. To assess if these two DNA fragments interfered with expression of the Pint-gfp fusion, plasmids pTCB207 and pTCB208 were transferred by filter mating from E. coli S17-1 to Pseudomonas sp. B13-jim1. Expression of GFP after 7 days of growth in liquid MM with 10 mM CBA in Pseudomonas sp. strain B13-jim1 (pCBA207) or (pCBA208) was compared to that in strain B13-jim1 without any plasmid and strain B13-jim1 (pKT230). Further functional mapping was performed by mutagenizing purified plasmid pTCB207 in vitro by using the EZ::TnTM < TET-1 > Insertion Kit (Epicentre technologies) according to the manufacturer's recommendations. After transformation into E. coli S17-1 mutant plasmids carrying the randomly inserted tetracycline resistance transposon < TET-1 > were isolated and screened with EcoRI restriction analysis for insertions in the 6.4 kb EcoRI fragment. DNA sequences flanking the < TET-1 > insertion in 12 selected plasmids were determined by DNA sequencing using primers internal to < TET-1 > facing outwards (Epicentre). These 12 plasmids, assigned pTCB207-6, − 7, − 10, − 41, − 46, − 48, − 52, − 53, − 62, − 64, − 66 and − 68 (Fig. 4) were subsequently introduced into Pseudomonas sp. strain B13-jim1 for testing the effect of < TET-1 > insertions on GFP production. Plasmid pTCB207, pTCB207-6 and −41 were transferred to B13-jim2, and pTCB207-6 to B13-jim6 to assess their influence on transcription from the jim2 and jim6 shortened integrase promoters.

DNA sequencing and analysis

DNA sequencing was performed on double stranded DNA templates with a Thermo Sequenase cycle sequencing kit with 7-deaza-dGTP (Amersham). Sequencing reactions were analysed on an automated DNA sequencer model 4200 IR2 (LI-COR, Lincoln, NE, USA). Primers for sequencing were labelled with fluorescent dyes IRD-800 and IRD-700 at their 5′ end and were purchased from MWG Biotech (Ebersberg, Germany). Sequence assembly and computer analysis was done with the dnastar software (DNASTAR, Madison, WI, USA). The strategy to sequence the insert of pTCB207 consisted of constructing overlapping subclones plus sequencing outwards from < TET-1 > insertions. Sequences were aligned using the program seqman within the dnastar package. Multiple (5–19) single sequences read in both directions covered the complete region. A few conflict positions in the program-generated consensus sequence were manually corrected by checking raw sequencing data files. The resulting DNA sequence is deposited in GenBank under Accession number AJ536665. Open reading frames were identified with the dnastar application MapDraw and compared with sequences in GenBank and EMBL using the blast search tools (Altschul et al., 1997).

Ancillary