Monitoring genome evolution ex vivo: reversible chromosomal integration of a 106 kb plasmid at two tRNALys gene loci in sequential Pseudomonas aeruginosa airway isolates

Claudia Kiewitz1, Karen Larbig1, Jens Klockgether1, Christian Weinel1 and Burkhard Tümmler1

Klinische Forschergruppe, Zentrum Biochemie und Zentrum Kinderheilkunde, OE 6711, Medizinische Hochschule Hannover,Carl-Neuberg-Str. 1, D-30623 Hannover, Germany1

Author for correspondence: Burkhard Tümmler. Tel: +49 511 5322920. Fax: +49 511 5326723. e-mail: tuemmler.burkhard{at}mh-hannover.de


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The genome rearrangements in sequential Pseudomonas aeruginosa clone K isolates from the airways of a patient with cystic fibrosis were determined by an integrated approach of mapping, sequencing and bioinformatics. Restriction mapping uncovered an 8·9 kb deletion of PAO sequence between phnAB and oprL in clone K, and two 106 kb insertions either adjacent to this deletion or several hundred kilobases away, close to the pilA locus. These 106 kb blocks of extra DNA also co-existed as the circular plasmid pKLK106 in several clone K isolates and were found to be closely related to plasmid pKLC102 in P. aeruginosa clone C isolates. The breakpoints of the deletion in clone K and the attB–attP sequences for the reversible integration of the plasmid in clones C and K were located within the 3’ end of the lysine tRNA structural genes (att site). pKLK106 sequentially recombined with either of the two tRNALys genes in clone K isolates. The att site of the pilA hypervariable region has been utilized by clone C to target its plasmid pKLC102 into the chromosome; the att site of the phnAB–oprL region has been employed by strain PAO to incorporate a DNA block encoding pyocin, transposases and IS elements. The use of typical phage attachment sites by conjugative genetic elements could be one of the major mechanisms used by P. aeruginosa to generate the mosaic genome structure of blocks of species-, clone- and strain-specific DNA. The example described here demonstrates the potential impact of systematic genome analysis of sequential isolates from the same habitat on our understanding of the evolution of microbial genomes.

Keywords: cystic fibrosis, genome evolution, Pseudomonas aeruginosa, recombination, tRNA gene

Abbreviations: CF, cystic fibrosis

The GenBank accession numbers for the sequences reported in this paper are AF285416AF285426.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The ubiquitous Pseudomonas aeruginosa is endowed with remarkable genome diversity. Its genetic repertoire is made up of a 5–7 Mb large circular haploid chromosome, plasmid(s), (pro)phage(s) and other mobile genetic elements (de Bruijn et al., 1998 ; Schmidt et al., 1996 ). Although the contig of common genes is conserved in almost all P. aeruginosa strains (Heuer et al., 1998a ) and its sequence diversity of about 0·3% is one order of magnitude lower than in comparable housekeeping genes of enterobacteria (Kiewitz & Tümmler, 2000 ), 1–200 kb blocks of clone- or strain-specific DNA interrupt the conserved gene contig and give rise to a mosaic genome structure (Römling et al., 1997 ). The 15–30% portion of intra- and interclonal genomic diversity is not equally distributed, but clusters in hypervariable regions of the chromosome (Römling et al., 1995 ; Heuer et al., 1998a ).

Current knowledge about genome diversity results mostly from comparative mapping and sequencing of selected isolates (Casjens, 1998 ) or propagation of bacterial populations in defined laboratory environments (Papadopoulos et al., 1999 ). These approaches inherently provide only indirect evidence of how microbial genomes evolve in complex natural habitats. For the past 15 years we have collected P. aeruginosa airway isolates taken from patients with cystic fibrosis (CF) since the onset of colonization (Römling et al., 1994 ). The P. aeruginosa infections in CF are a paradigm of how versatile environmental bacteria can conquer, adapt and persist in an atypical habitat, and successfully evade defence mechanisms and chemotherapy in a susceptible host (Tümmler & Kiewitz, 1999 ). This strain collection provides a unique opportunity to monitor genome evolution. Here we report on the aetiology of genome rearrangements in sequential isolates from a patient with CF who has been and still is harbouring the most prevalent P. aeruginosa clones C and K (Römling et al., 1995 ). A large plasmid was found to integrate and mobilize repetitively from an attB site located within two separate copies of a lysine tRNA gene. Both tRNA loci are located in hypervariable regions: one copy is located adjacent to phnAB (Heuer et al., 1998a , b ) and the other is adjacent to pilA (Römling et al., 1995 ).


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bacterial strains.
To monitor genomic variation of closely related P. aeruginosa strains, sequential airway isolates from a patient with CF were selected: K (clone SS3 in Heuer et al. 1998a , b ), first P. aeruginosa-positive respiratory specimen; K1, K1' (SS5, SS5'), 7 months; K2 (SS20), 36 months of colonization. For comparison, P. aeruginosa strain C, which belongs to a major clone in CF patients and aquatic habitats (Schmidt et al., 1996 ), and the reference strain PAO (DSM 1707) were included.

Restriction mapping.
P. aeruginosa bacteria (5x109 P. aeruginosa cells ml-1), grown to late exponential phase were encapsulated into agarose plugs, lysed extensively with detergents and proteinase K, and the intact chromosomes cleaved with SpeI or I-CeuI as described previously (Römling et al., 1997 ). In one-dimensional PFGE, the SpeI digests were separated in a Bio-Rad CHEF-DRIII cell using three linear ramps of 5–60 s (17 h), 10–25 s (14 h) and 1–5 s (6 h) (E=6 V cm-1, 13 °C, 1% agarose gel with 10 µM thiourea in 0·5x TBE buffer).

For two-dimensional macrorestriction mapping (see Schmidt et al., 1996 ) the genome was completely digested with I-CeuI and separated in the first dimension in a Bio-Rad CHEF-DRIII cell (E=1·3 V cm-1, one linear ramp of 400–1000 s in 1 s increments, 96 h, 13 °C, 0·6% agarose gel with 10 µM thio-urea). The entire lane of each strain was cut out, digested with SpeI and separated perpendicular to the first run in the second dimension using three linear ramps from 10–50 s (24 h), 10–25 s (22 h) and 1–10 s (14 h) (E=6 V cm-1, 13 °C, 1·5% agarose gel with 10 µM thiourea).

For high-resolution Smith/Birnstiel mapping, SpeI complete/n partial double digestions (n: EcoRI, BglII, XhoI, NdeI) were carried out, the fragments separated by PFGE and hybridized with cloned SpeI fragment ends as described previously (Heuer et al., 1998b ).

SpeI maps of strains K, K1 and K2 covering the toxAhemA region were reconstructed by partial SpeI digestions of agarose-embedded chromosomes (1x1010 P. aeruginosa cells ml-1) using 0·025 U, 0·01 U or 0·005 U SpeI (New England Biolabs) for 1 h. The partial SpeI digests were separated in a Bio-Rad CHEF-DRIII cell with three linear ramps of 5–100 s (30 h), 10–40 s (10 h) and 70–100 s (10 h) (E=6 V cm-1, 13 °C, 1·5% agarose gel with 10 µM thiourea) and transferred onto nylon membranes by capillary blotting. Southern hybridizations were performed with PCR-amplified PAO sequences (http://www.pathogenesis.com/), cloned genes (Schmidt et al., 1996 ), cloned SpeI fragment ends (Heuer et al., 1998a ) and SpeI linking clones (Römling & Tümmler, 1993 ). Probes were labelled with digoxigenin-dUTP and hybridized with the pulsed-field blot as described previously (Schmidt et al., 1996 ). Hybridized fragments were detected by chemiluminescence using an alkaline phosphatase-conjugated anti-digoxigenin antibody and CDP-Star (Tropix) as substrate.

Isolation and analysis of P. aeruginosa plasmids pKLC102 and pKLK106.
High-molecular-mass plasmids were prepared on a large scale by modified alkaline lysis. Plasmid size was determined by adding up the sizes of restriction fragments after conventional or PFGE gel separation of BamHI or SpeI digests, respectively. A library of partially BamHI-restricted pKLC102 was maintained as pLAFR3-derived cosmids (Staskawicz et al., 1987 ) in Escherichia coli DH5{alpha}. For cosmid ordering, about 60 clones were digested with BamHI to completion and manually assembled by restriction fingerprint analysis. The fragment assembly was confirmed by Southern hybridization where selected cosmids, single BamHI fragments (e.g. BmQ) or the small SpeI fragment SpAQ served as probes.

PCR amplification and sequencing.
Chromosomal DNA was prepared using a rapid method for Gram-negative bacteria (Chen & Kuo, 1993 ). PCR was performed from purified DNA as described previously (Spangenberg et al., 1995 ). Primer sequences for PCR were designed to anneal to the PAO genome sequence between phnAB and oprL, the PAO sequence upstream of the pilA locus and the BamHI fragment sequence BmQ of pKLC102. The sequences are shown in Table 1. Selected PCR products were sequenced in both directions.


View this table:
[in this window]
[in a new window]
 
Table 1. Primer sequences used in this study

 
Annotation.
The phnAB and the pilA regions of the PAO genome sequence (http://www.pathogenesis.com/) were searched for homologies in public databases with the programs BLAST N, BLAST X and BLAST P. Putative ORFs were identified with GeneMark (Besemer & Borodovsky, 1999 ). Pairwise sequence comparisons and multiple alignments were generated using CLUSTAL W (Thompson et al., 1994 ). Long-range restriction maps were constructed with the in-house program MasterMap. tRNA genes were identified by the program tRNA-scan-SE (Lowe & Eddy, 1997 ).

Sequence data.
Novel nucleotide sequences were deposited in GenBank. The accession numbers are AF285416–AF285426.


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Two-dimensional PFGE and Southern analysis indicate genome rearrangements in sequential clone K isolates from a CF patient
Genome evolution of P. aeruginosa in CF lungs was monitored by analysis of sequential isolates following the onset of colonization. One patient has been and still is chronically co-colonized with the prevalent genotypes C and K (patient 11, Römling et al., 1994 ). Genotype K and C strains are 65% similar in their SpeI band patterns (Römling et al., 1995 ), indicating that these two genotypes are more closely related to each other than the average pair of P. aeruginosa genotypes, which share only 40% of bands in their SpeI fragment patterns (Grothues & Tümmler, 1991 ).

SpeI fragment patterns of patient 11’s clone K sequential isolates changed within the first 3 years of colonization (Fig. 1). The basis for these shifts of fragment size was not that these were nucleotide substitutions in SpeI recognition sites, but gross genome rearrangements, as became evident from two-dimensional PFGE of complete I-CeuI/SpeI digestions of clone K, K1 and K2 chromosomes (Fig. 2). The intron-encoded endonuclease I-CeuI cleaves within the four ribosomal operons of P. aeruginosa (Liu et al., 1993 ). The similarity of the SpeI fragment patterns reflects the conservation of most fragment sizes in the four I-CeuI fragments CeA, CeB, CeC and CeD, but a few SpeI fragments on CeA and CeD were unique for strains K, K1 and K2, and a 29 kb fragment was either located in CeA (K) or in CeD (K1 and K2) (Fig. 2). Two of these unique fragments (strain K: 46 kb; strains K1 and K2: 169 kb; Fig. 2) were found to hybridize with strain PAO fragment SpU. We knew from previous Southern hybridizations that the phnAB hypervariable region, which maps to SpU in strain PAO, was affected by the complex genome rearrangements in the sequential clone K isolates (Heuer et al., 1998a , b ). The 1·5 Mb region flanking this hypervariable region was mapped by Southern hybridization of SpeI complete and partial digests of strain K, K1 and K2 chromosomes with PAO-derived gene probes (Fig. 3). The global gene contig turned out to be conserved between PAO and the clone K strains as had been expected from the comparative two-dimensional analysis (see Fig. 2). The SpeI fragment contig and sizes were identical within experimental error for the adjacent four fragments towards the terminus of replication, but were divergent towards the origin of replication (Fig. 3).



View larger version (58K):
[in this window]
[in a new window]
 
Fig. 1. SpeI-restriction fragment patterns of clonal variants of P. aeruginosa clone K. The genotype designation of each strain is indicated above the respective lane. For comparison, SpeI digests of reference strain PAO (DSM 1707) and strain C of the predominant clone C were included. {lambda} oligomers served as molecular mass standards. All macrorestriction fragment sizes refer to calibration with {lambda} oligomers and with SpeI-digested and PFGE-separated chromosomal DNA of PAO and C. The reader may note that PFGE inherently underestimates fragment sizes by 5–10% in the 50–1000 kb range in comparison to fragment sizes deduced from genome sequencing.

 


View larger version (64K):
[in this window]
[in a new window]
 
Fig. 2. Comparative analysis of P. aeruginosa strains PAO (DSM 1707), K, K1 and K2 by two-dimensional PFGE of sequential I-CeuI/SpeI restriction digests. After separation of the I-CeuI digested chromosomes in the first dimension, each lane of the four strains containing four I-CeuI fragments (CeA, CeB, CeC, CeD) was cut out, digested with SpeI and separated in the second dimension. In strains K, K1 and K2, a complete SpeI digest (lane termed ‘total’) is shown in the left lane. For better orientation, sizes of selected PAO SpeI fragments are indicated on the left as molecular mass standards. On the right, the differences between the three clone K strains are indicated. The 29 kb fragment co-migrates with invariant fragments of similar size as a double band either in the I-CeuI fragment CeA in strain K or in CeD in strains K1 and K2 and hence can be identified by signal intensity.

 


View larger version (18K):
[in this window]
[in a new window]
 
Fig. 3. Comparison of SpeI maps of P. aeruginosa PAO (DSM 1707), clonal variants K, K1 and K2, and strain C in the toxAhemA region. SpeI fragment designations of PAO and C were taken from Schmidt et al. (1996) . All fragment sizes (in kb) were determined by PFGE (refer to Fig. 1). The following probes were used in Southern hybridization analysis: single genes (toxA, phnA, toxR, pilA and hemA), cloned SpU fragment ends (3A1-2, 9H11), PAO-derived SpeI linking clones indicated by grey shaded boxes and rrn operons marked by black arrows. Corresponding map positions are connected by dotted lines. The grey highlighted areas indicate the integrated plasmids (pKLK106 in strains K, K1, K2 and pKLC102 in strain C). The PAO-specific SpU–SpF junction is shaded grey.

 
An island of PAO sequence in the phnAB–oprL region is absent in clone K
As indicated by high-resolution restriction maps (Fig. 4), 95 % of the PAO SpU sequence was retained in clone K isolates, but approximately 5% was deleted at the SpU/SpF junction. Taking the complete PAO genome sequence as the blueprint, oligonucleotides were designed to cover the terminal 10 kb of SpU by overlapping PCR products. Restricted clone K DNA hybridized only with those PCR products that had been amplified from PAO sequence with identical restriction fingerprints in PAO and clone K strains. These Southern data verified the absence of the terminal 7 kb of PAO SpU in clone K and narrowed the breakpoint of the conserved sequence contig to within 1 kb (Fig. 4). The breakpoint was located in a region that, according to our annotation, contains IS elements, a pyocin, integrase and orphan genes, but no housekeeping gene in strain PAO (Fig. 4).



View larger version (25K):
[in this window]
[in a new window]
 
Fig. 4. High-resolution Smith/Birnstiel map (SpeI, EcoRI, BglII, XhoI, NdeI) of the hypervariable SpU–SpF junction in P. aeruginosa PAO and of the corresponding regions in clone K. The annotation of the PAO sequence is depicted as follows: established P. aeruginosa genes are black, putative ORFs showing highly significant homologies to known genes of the same or other species are light grey, genes located on IS elements are dark grey, toxin genes are black and white patterned, and orphan genes are white. The numbers refer to putative ORFs showing pronounced amino acid identities to (1) 3-oxoacyl-[acyl carrier protein]-synthase III, (2) acyl-CoA ligase precursor, (3) DNA methyltransferase, (4) type-1 fimbrial assembly protein, (7) pyocin S of P. aeruginosa, (10) regulatory protein ExsB of Synechocystis. The ORFs 5 (transposase) and 6 are located on IS1209', which belongs to the IS3 family and shows 87% nucleotide identity to IS1209 of P. aeruginosa (Burrows et al., 1996 ). ORFs 8 and 9 are part of the second copy of IS222 in the P. aeruginosa genome (Kropinski et al., 1994 ). The 76 bp tRNALys gene and the att or att* elements (3’ terminal 45 or 21 bp of the tRNALys gene with a T overhang) are indicated by black bars. They flank the 8·9 kb insertion in PAO and the integrated 106 kb plasmid pKLK106 in strain K. The dotted line marks the integration point in strains K1 and K2. As evident from Figs 5 and 6a the plasmid recombination site is located on a 2·4 kb BamHI fragment (BmQ) forming the plasmid/chromosome junction in strain K. Triangles indicate oligonucleotide primers and asterisks depict matched pairs of complementary oligonucleotide primers.

 


View larger version (56K):
[in this window]
[in a new window]
 
Fig. 5. Comparison of the plasmids harboured by clone K and clone C. (a) Diagnostic BamHI fingerprints of the highly similar plasmids pKLK106 of clone K and pKLC102 of clone C. The conserved 2·4 kb BamHI fragment (BmQ) carrying the plasmid recombination site (attP) is indicated. (b) BamHI (outer circle) and SpeI (inner circle) maps of pKLC102. SpeI fragment designations refer to Fig. 7a of Römling et al. (1997) . BmQ is highlighted in black. Differential regions between pKLC102 and pKLK106 are shaded grey. Fragments SpAI’ and SpAQ of pKLC102 are equivalent to the diagnostic 29 kb SpeI fragment in clone K isolates (see Fig. 2).

 


View larger version (34K):
[in this window]
[in a new window]
 
Fig. 6. att sequences of P. aeruginosa. (a) Sequence alignment of the conserved bacterial and plasmid att sites and flanking regions: the tRNALys gene on SpE of PAO (pilA region, see Fig. 3), the tRNALys gene on the 169 kb SpeI fragment in strains K1 and K2 (oprLphnAB region, see Figs 3 and 4), and the plasmid recombination site (attP) of pKLC102 located on the 2·4 kb BamHI fragment BmQ (identical in pKLK106). The tRNALys gene and conserved nucleotides within this gene are highlighted in black. A tRNALys gene sequence from E. coli (GenBank accession no. K00282) is included for comparison; the anticodon is marked by asterisks. The 3’ terminal 45 bp of the tRNALys gene with an additional T make up the attP and attB sites, respectively, indicated by the line. (b, c) Sequence alignments of the left junction (b) containing the complete tRNALys gene, and the right junction (c) with the 46 bp att site underlined. The sequence alignments include both integration sites of the plasmids pKLK106 or pKLC102, either in the hemApilA region (strains K1, K2, C) or in the oprLphnAB region of PAO (strain K) (see Figs 3 and 4). (c) In PAO the conserved 46 bp att site (underlined) in the oprLphnAB region is truncated, leading to a 22 bp att* element (double underlined). The downstream sequence is conserved except for some nucleotide substitutions; one of them is an uncommon A to C transversion leading to the loss of a BglII site in strains K and K1/K2.

 
Clones K and C harbour highly similar plasmids
Several clone K isolates harboured the episomal 106 kb plasmid pKLK106. The plasmid contained a 29 kb SpeI fragment which co-migrated with, and was shown to hybridize to, a 29 kb chromosomal SpeI fragment in the clone K isolates. This 29 kb SpeI fragment was mapped close to the breakpoint in strain K, but close to pilA, 600 kb away in strains K1 and K2 (Fig. 3). This observation was the first hint that plasmid sequence could be integrated into the clone K chromosome at two distant sites. After isolation of pKLK106, its BamHI restriction fragment pattern turned out to be almost identical to that of the 102 kb clone C plasmid pKLC102 (Fig. 5), which occurs in clone C strains as an episomal plasmid and/or integrated into the clone C chromosome close to the pilA locus (Römling et al., 1997 ) (Fig. 3). Hence we concluded that the related clone C and K plasmids could integrate into the same hypervariable region.

The recombination site of pKLC102 was mapped with cosmid-derived probes to fragment BmQ (Fig. 5b). To identify the plasmid–chromosome junction and putative targeting signals, the BmQ fragment was sequenced and searched for homologies in the completed PAO genomic sequence. Perfect matches of BmQ with PAO sequence were found in the hypervariable pilA (SpE) and phnAB–oprL regions (SpU/SpF). A 45 bp sequence of BmQ was identical to the 3' region of a tRNALys gene of which single copies are encoded on SpE and SpF. The 3' ends of tRNA genes are typical targets for the integration of phages (att sites) (Campbell, 1996 ) and, correspondingly, we have designated the 45 terminal nucleotides of the tRNA gene plus an additional T as the attB site. The perfect match of chromosomal attB and plasmid attP sequences is shown in Fig. 6(a). The attB site in the BmQ fragment is followed by 295 bp non-coding sequence and a 1356 bp large ORF that encodes an integrase (int; GenBank accession no. AF285416) showing highest homology to the integrase of Actinobacillus actinomycetemcomitans (AF006830). An att–spacer–int contig is typical for phage attachment sites (Campbell, 1996 ), but unusual for a plasmid sequence.

Reversible chromosomal integration of plasmid at tRNALys loci
Combinatorial PCR was applied to test the hypothesis that the 3' sequence of the tRNALys gene had been utilized by the clone K and C strains to incorporate their respective plasmids into the chromosome. Several primers were designed downstream and upstream of the attP site in BmQ and the two tRNALys genes in the PAO genome. One plasmid-derived and one chromosome-derived primer were combined to amplify chromosomal DNA from strains C, K, K1 and K2 by PCR. Single PCR products were obtained and sequenced. In all cases, the 3' segment of the tRNALys gene was identified at the junctions between species-specific chromosomal and clone-specific plasmid sequence. In the pilA region the sequence contig in strains C, K1 and K2 reads hemA–chromosome–5' tRNALys–(BmQ–plasmid–BmQ)–att 3'–chromosome–pilA (Fig. 3). Sequence alignments of the recombination sites are shown in Fig. 6 (b, c). In strain K the plasmid is integrated into the phnABoprL region at the tRNALys gene (Figs 3 and 4), and the sequence contig correspondingly reads oprL–chromosome–5' tRNALys–(BmQ–plasmid–BmQ)–att 3'–chromosome–phnAB(Fig. 6b, c).

Interestingly, the same genomic organization also applies to the 8·9 kb block of PAO-specific DNA at the SpU/SpF border (Fig. 4). The left junction consists of the tRNALys gene and the right junction of a 21 bp segment of the 3' end of the tRNALys gene (att*). At the left junction, the PAO chromosome and the plasmid-derived DNA of strain K exhibit an extended sequence homology of about 800 bp beyond the 3' end of the tRNALys gene (Fig. 6b). In other words, the strong sequence similarity between PAO and clone K strains ends about 800 bp downstream of the 3' end of the tRNALys gene and starts again at the att* element (Fig. 4).

Our conclusion that the two tRNALys genes are hotspots for the integration and excision of DNA was supported by the subculturing of single colonies of clone K isolates in vitro. Numerous descendants of a K1 ancestor carried pKLK106 DNA in the phnAB instead of the pilA region (for example strain K1', Fig. 1), implying that the plasmid sequence was rapidly mobilized from one att site to the other. In summary, the genome rearrangements observed in sequential clone K isolates were based on the reversible integration of pKLK106 in either one or the other chromosomal tRNALys gene. One site has also been employed by strain PAO to incorporate a DNA block encoding pyocin, transposases and IS elements, the other site by clone C to target its plasmid pKLC102 into the chromosome.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The genome rearrangements in sequential clone K isolates from the airways of a CF patient who is homozygous for the major CFTR mutation {Delta}F508 (Tümmler & Kiewitz, 1999 ) were established by an integrated approach of mapping, sequencing and bioinformatics. The diverse genomic organization originated from the reversible integration of a 106 kb plasmid pKLK106 in either of two identical chromosomal recognition sites (attB). The target sites were identified within the 3' end of lysine tRNA structural genes which are located in the phnAB and pilA regions. The attB site close to pilA was also employed by strain C to integrate a highly similar plasmid, pKLC102.

An 8·9 kb island was integrated into the PAO chromosome at the same tRNA recognition site in the phnAB–oprL region as pKLK106, albeit the discernible att 3' sequence was truncated (att*). The annotation of the 8·9 kb insertion provides evidence that the DNA block was acquired by horizontal gene transfer (Fig. 4). Two different IS elements are located close to the borders of the element and encode transposases. One large ORF encodes a bactericidal S-type pyocin. Otherwise, no function could be ascribed to the ORFs of the insertion. Twelve out of 17 identified ORFs upstream of and within the DNA block are orphan genes with no significant homology to any entry in the public sequence databases (Fig. 4), supporting the notion that a hypervariable region should be devoid of (essential) housekeeping genes.

The 8·9 kb insertion is stably maintained in strain PAO and hence behaves like a pathogenicity island (Hacker et al., 1997 ; Hou, 1999 ). tRNAs play a role in the horizontal transfer of virulence gene clusters between different pathogens. An example is the tRNASel locus of E. coli that has served as the site of integration of two distinct pathogenicity islands that are responsible for converting benign strains into uro- and enteropathogens (Blanc-Potard & Groisman, 1997 ).

tRNAs play versatile roles in prokaryotes and eukaryotes. They are central components of the translational machinery and are essential for replication of retroviruses because the tRNAs bind to viral genomes through their 3'-end sequences and act as primers for initiation of viral replication (Hou, 1999 ). Many temperate bacteriophages integrate into the bacterial chromosome via site-specific integration at an attB site that is typically within, or overlaps with, the 3' end of a tRNA gene (Campbell, 1996 ). In P. aeruginosa, the cytotoxin-converting temperate phage {phi}CTX was found to integrate at a chromosomal serine tRNA gene (Hayashi et al., 1993 ; Nakayama et al., 1999 ).

Whereas the integration of phage genomes into their host chromosome at the 3' end of tRNA genes has been demonstrated for numerous taxospecies (Campbell, 1996 ), reports that plasmids can recombine with chromosomal tRNA genes are rare. Examples are the elements pSAM2 from Streptomyces ambofaciens (Raynal et al., 1998 ) and pSE101 from Saccharopolyspora erythraea (Brown et al., 1994 ), which integrate into the chromosome at the 3' ends of the tRNAPro (pSAM2) and tRNAThr (pSE101) genes. Interestingly, pSAM2, pSE101 and our P. aeruginosa plasmids pKLC102 and pKLK106 all encode an integrase gene (int) in close vicinity to the attP end.

We compared the att sequences, their localization within the tRNA genes and the organization of the att–spacer–int sequence contig of pKLK106 with that of conjugative mobile genetic elements pSAM2 (Raynal et al., 1998 ), pSE101 (Brown et al., 1994 ) and that of the temperate phages mv4 (Auvray et al., 1997 ; 1999 ), Sfi21 (Bruttin et al., 1997 ) and VWB (Van Mellaert et al., 1998 ) and the classical paradigm phage {lambda} (Campbell, 1996 ). Amongst this set of selected examples, pKLK106 and pKLC102 are conspicuous by containing the largest spacer sequence of 295 bp between att and int. The attint sequence contigs of the plasmid-like mobile genetic elements are no more closely related to each other than they are to those of the phages. No general rules were seen with respect to the overlap of tRNA and att gene sequences, primary sequences and length of the att site, GC-contents and length of the spacer, and the coding sequence of int and its orientation with respect to att. All investigated examples share the localization of att at a tRNA gene locus and an adjacent int gene, but otherwise each analysed phage or conjugative genetic element is characterized by individual features of its primary sequence at the att locus.

In conclusion, by the analysis of sequential clone K isolates from CF airways, we have detected the reversible integration of a 106 kb plasmid at two identical att sites within the P. aeruginosa chromosome. The utilization of typical phage attachment sites by conjugative genetic elements could be one of the major mechanisms that allows P. aeruginosa to generate the mosaic genome structure of blocks of species-, clone- and strain-specific DNA. Our example demonstrates the potential impact of systematic genome analysis of sequential isolates from the same habitat on our understanding of the evolution of microbial genomes. The comparative genomics of strains that are collected from natural habitats over an extended period of time under well-documented sampling conditions provides robust primary data about genomic evolution in natural bacterial populations that are not biased by any heuristic assumptions or models and hence should withstand the test of time.


   ACKNOWLEDGEMENTS
 
Claudia Kiewitz and Karen Larbig contributed equally to this work. This work was supported by the Deutsche Forschungsgemeinschaft (Schwerpunktprogramm ‘Mikrobielle Ökologie’). Short-term fellowships to J.K. and C.W. by the Christiane-Herzog Stiftung are gratefully acknowledged. We cordially thank the University of Washington Genome Center (Drs X.-Q. Pham and Maynard Olson), PathoGenesis Corporation (Drs Richard Garber and Ken Stover) and the Pseudomonas aeruginosa Genome Analysis Committee (coordinators of PAGAC, Drs Fiona Brinkman and Robert Hancock) for the access to the PAO genome sequence data throughout the on-going genome project. This information was an invaluable aid to speeding up our comparative genome analysis.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Auvray, F., Coddeville, M., Ritzenthaler, P. & Dupont, L. (1997). Plasmid integration in a wide range of bacteria mediated by the integrase of Lactobacillus delbrueckii bacteriophage mv4. J Bacteriol 179, 1837-1845.[Abstract]

Auvray, F., Coddeville, M., Ordonez, R. C. & Ritzenthaler, P. (1999). Unusual structure of the attB site of the site-specific recombination system of Lactobacillus delbrueckii bacteriophage mv4. J Bacteriol 181, 7385-7389.[Abstract/Free Full Text]

Besemer, J. & Borodovsky, M. (1999). Heuristic approach to deriving models for gene finding. Nucleic Acids Res 27, 3911-3920.[Abstract/Free Full Text]

Blanc-Potard, A. B. & Groisman, E. A. (1997). The Salmonella selC locus contains a pathogenicity island mediating intramacrophage survival. EMBO J 16, 5376-5385.[Abstract/Free Full Text]

Brown, D. P., Idler, K. B., Backer, D. M., Donadio, S. & Katz, L. (1994). Characterization of the genes and attachment sites for site-specific integration of plasmid pSE101 in Saccharopolyspora erythraea and Streptomyces lividans. Mol Gen Genet 242, 185-193.[Medline]

de Bruijn, F. J., Lupski, J. R. & Weinstock, G. M. (1998). Bacterial Genomes: Physical Structure and Analysis. New York: Chapman & Hall.

Bruttin, A., Foley, S. & Brussow, H. (1997). The site-specific integration system of the temperate Streptococcus thermophilus bacteriophage {phi}Sfi21. Virology 237, 148-158.[Medline]

Burrows, L. L., Charter, D. F. & Lam, J. S. (1996). Molecular characterization of the Pseudomonas aeruginosa serotype O5 (PAO1) B-band lipopolysaccharide gene cluster. Mol Microbiol 22, 481-495.[Medline]

Campbell, A. M. (1996). Bacteriophages. In Escherichia coli and Salmonella: Cellular and Molecular Biology, 2nd edn, pp. 2325–2338. Edited by F. C. Neidhardt & others. Washington, DC: American Society for Microbiology.

Casjens, S. (1998). The diverse and dynamic structure of bacterial genomes. Annu Rev Genet 32, 339-377.[Medline]

Chen, W. P. & Kuo, T. T. (1993). A simple and rapid method for the preparation of gram-negative bacterial genomic DNA. Nucleic Acids Res 21, 2260.[Medline]

Grothues, D. & Tümmler, B. (1991). New approaches in genome analysis by pulsed-field gel electrophoresis: application to the analysis of Pseudomonas species. Mol Microbiol 5, 2763-2776.[Medline]

Hacker, J., Blum-Oehler, G., Mühldorfer, I. & Tschäpe, H. (1997). Pathogenicity islands of virulent bacteria: structure, function and impact on microbial evolution. Mol Microbiol 23, 1089-1097.[Medline]

Hayashi, T., Matsumoto, H., Ohnishi, M. & Terawaki, Y. (1993). Molecular analysis of a cytotoxin-converting phage, {phi}CTX, of Pseudomonas aeruginosa: structure of the attP-cos-ctx region and integration into the serine tRNA gene. Mol Microbiol 7, 657-667.[Medline]

Heuer, T., Bürger, C. & Tümmler, B. (1998a). Smith/Birnstiel mapping of genome rearrangements in Pseudomonas aeruginosa. Electrophoresis 19, 495-499.[Medline]

Heuer, T., Bürger, C., Maaß, G. & Tümmler, B. (1998b). Cloning of prokaryotic genomes in yeast artificial chromosomes: application to the population genetics of Pseudomonas aeruginosa. Electrophoresis 19, 486-494.[Medline]

Hou, Y. M. (1999). Transfer RNAs and pathogenicity islands. Trends Biochem Sci 24, 295-298.[Medline]

Kiewitz, C. & Tümmler, B. (2000). Sequence diversity of Pseudomonas aeruginosa: impact on population structure and genome evolution. J Bacteriol 182, 3125-3135.[Abstract/Free Full Text]

Kropinski, A. M., Farinha, M. A. & Jansons, I. (1994). Nucleotide sequence of the Pseudomonas aeruginosa insertion sequence IS222: another member of the IS3 family. Plasmid 31, 222-228.[Medline]

Liu, S. L., Hessel, A. & Sanderson, K. E. (1993). Genomic mapping with I-CeuI, an intron-encoded endonuclease specific for genes for ribosomal RNA, in Salmonella spp., Escherichia coli, and other bacteria. Proc Natl Acad Sci U S A 90, 6874-6878.[Abstract]

Lowe, T. M. & Eddy, S. R. (1997). tRNAscan-SE: a program for improved detection of transfer RNA genes in genomic sequence. Nucleic Acids Res 25, 955-964.[Abstract/Free Full Text]

Nakayama, K., Kanaya, S., Ohnishi, M., Terawaki, Y. & Hayashi, T. (1999). The complete nucleotide sequence of {phi}CTX, a cytotoxin-converting phage of Pseudomonas aeruginosa: implications for phage evolution and horizontal gene transfer via bacteriophages. Mol Microbiol 31, 399-419.[Medline]

Papadopoulos, D., Schneider, D., Meier-Eiss, J., Arber, W., Lenski, R. E. & Blot, M. (1999). Genomic evolution during a 10,000-generation experiment with bacteria. Proc Natl Acad Sci U S A 96, 3807-3812.[Abstract/Free Full Text]

Raynal, A., Tuphile, K., Gerbaud, C., Luther, T., Guerineau, M. & Pernodet, J. L. (1998). Structure of the chromosomal insertion site for pSAM2: functional analysis in Escherichia coli. Mol Microbiol 28, 333-342.[Medline]

Römling, U. & Tümmler, B. (1993). Comparative mapping of the Pseudomonas aeruginosa PAO genome with rare-cutter linking clones or two-dimensional pulsed-field gel electrophoresis protocols. Electrophoresis 14, 283-289.[Medline]

Römling, U., Fiedler, B., Boßhammer, J., Grothues, D., Greipel, J., von der Hardt, H. & Tümmler, B. (1994). Epidemiology of chronic Pseudomonas aeruginosa infections in cystic fibrosis. J Infect Dis 170, 1616-1621.[Medline]

Römling, U., Greipel, J. & Tümmler, B. (1995). Gradient of genomic diversity in the Pseudomonas aeruginosa chromosome. Mol Microbiol 17, 323-332.[Medline]

Römling, U., Schmidt, K. D. & Tümmler, B. (1997). Large genome rearrangements discovered by the detailed analysis of 21 Pseudomonas aeruginosa clone C isolates found in environment and disease habitats. J Mol Biol 271, 386-404.[Medline]

Schmidt, K. D., Tümmler, B. & Römling, U. (1996). Comparative genome mapping of Pseudomonas aeruginosa PAO with P. aeruginosa C, which belongs to a major clone in cystic fibrosis patients and aquatic habitats. J Bacteriol 178, 85-93.[Abstract]

Spangenberg, C., Fislage, R., Sierralta, W., Tümmler, B. & Römling, U. (1995). Comparison of type IV-pilin genes of Pseudomonas aeruginosa of various habitats has uncovered a novel unusual sequence. FEMS Microbiol Lett 125, 265-273.[Medline]

Staskawicz, B., Dahlbeck, D., Keen, N. & Napoli, C. (1987). Molecular characterization of cloned avirulence genes from race 0 and race 1 of Pseudomonas syringae pv. glycinea. J Bacteriol 169, 5789-5794.[Medline]

Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994). CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Res 22, 4673-4680.[Abstract]

Tümmler, B. & Kiewitz, C. (1999). Cystic fibrosis: an inherited susceptibility to bacterial respiratory infections. Mol Med Today 5, 351-358.[Medline]

Van Mellaert, L., Mei, L., Lammertyn, E., Schacht, S. & Anné, J. (1998). Site-specific integration of bacteriophage VWB genome into Streptomyces venezuelae and construction of a VWB-based integrative vector. Microbiology 144, 3351-3358.[Abstract]

Received 30 March 2000; revised 17 July 2000; accepted 21 July 2000.