Main

Pseudomonas spp. are ubiquitous Gram-negative bacteria that colonize and survive in numerous ecological niches including soil, water and plant surfaces. This versatility is reflected by the sizes of their genomes, which contain large sets of genes involved in carbon source utilization and adaptation. In 2001, we isolated a bacterial strain closely related to the saprophytic soil bacterium Pseudomonas putida, Pseudomonas entomophila, which triggers a systemic immune response in D. melanogaster after ingestion1. P. entomophila is highly pathogenic for both D. melanogaster larvae and adults. Its persistence in larvae leads to a massive destruction of gut cells1.

Entomopathogenic bacteria such as the Gram-negative bacteria Photorhabdus luminescens, Xenorhabdus nematophilus, Yersinia pestis, Serratia marcescens and Serratia entomophila and the Gram-positive bacterium Bacillus thuringiensis have developed different strategies to interact with and kill insects2. Some gene products derived from these bacteria as well as the bacteria themselves, have been used to generate biopesticides3. The ability of P. entomophila to orally infect and kill larvae of insect species belonging to different orders makes it a promising model for the study of host-pathogen interactions and for the development of biocontrol agents against insect pests. To unravel features contributing to P. entomophila's entomopathogenic properties, we have determined its complete genome sequence and performed a genome-wide screen for mutants affected in their ability to trigger an immune response and lethality in D. melanogaster.

Results

Genome features and comparative genomics

The P. entomophila genome is composed of a single circular chromosome of 5,888,780 base pairs (Fig. 1). Among 5,169 coding sequences identified, 3,466 genes (67%) have been assigned a predicted function (Table 1). The P. entomophila genome is smaller than the six other Pseudomonas genomes that have been published (Table 1): the human opportunistic pathogen P. aeruginosa PAO1 (ref. 4), the three P. syringae pathovars5,6,7, the plant commensal P. fluorescens Pf-5 (ref. 8) and the saprophytic soil bacterium P. putida KT2440 (ref. 9).

Figure 1: Circular representation of the P. entomophila genome.
figure 1

The outer scale indicates coordinates in base pairs (bp). Circles 1 and 2 (from outside to inside) show predicted coding regions transcribed clockwise and counterclockwise, respectively. Coding sequences are color coded by role categories: salmon, amino acid biosynthesis; light blue, biosynthesis of cofactors, prosthetic groups and carriers; light green, cell envelope; red, cellular processes; brown, central intermediary metabolism; yellow, DNA metabolism; green, energy metabolism; purple, fatty acid and phospholipid metabolism; violet, mobile and extrachromosomal element functions; pink, protein synthesis and fate; orange, purines, pyrimidines, nucleosides and nucleotides; navy blue, regulatory functions and signal transduction; lime green, secondary metabolite biosynthesis; gray, transcription; teal, transport and binding proteins; black, unknown function and hypothetical proteins. Circle 3 shows rRNA genes in salmon, tRNA genes in green and miscellaneous RNA genes in blue. Circle 4 shows transposase genes, putative prophages and gene clusters encoding secondary metabolites coded by colored symbols as follows: green arrowheads, transposases; gray, putative prophages; red, pyoverdine synthesis; light blue, cluster involved in lipopeptide II biosynthesis; violet, acinetobactin-like siderophore synthesis; light green, cluster involved in lipopeptide III biosynthesis; navy blue, cluster and isolated genes involved in lipopeptide I biosynthesis; pink, hydrogen cyanide production; brown, polyketide synthesis. Circle 5 shows the distribution of REPs. These repeats are scattered all over the genome and were found either as single elements, in paired elements or in clusters of up to six elements in alternating orientation. Circle 6 shows G+C in relation to the mean G+C in a 1,000-bp window. Circle 7 shows GC skew in a 1,000-bp window.

Full size image
Table 1 General features of genomes of representative Pseudomonas species
Full size table

GC skew analysis and the predicted location of the origin of replication oriC near dnaA and of the chromosome dimer resolution dif site in PSEEN2780 revealed the presence of two replichores of similar size, contrary to the unbalanced replichores found in the genomes of P. putida KT2440 (ref. 10) and P. aeruginosa PAO1 (ref. 4) (see Supplementary Fig. 1 online). BLAST comparisons of genomes from the five Pseudomonas representative species identified a set of 2,065 genes that constitutes the Pseudomonas core genome. Based on this analysis, we identified 1,002 genes unique to the P. entomophila genome. We found that, consistent with the close relatedness between P. entomophila and P. putida1, 70.2% of P. entomophila genes (3,630) have orthologs in the P. putida genome, of which more than 96% are found in synteny (see Supplementary Table 1 online). The smaller size of the P. entomophila genome compared to that of other Pseudomonas does not seem to originate from reductive evolution. Indeed the 50 genes of P. entomophila present in other Pseudomonas but absent from P. putida belong to functional classes as diverse as the 34 genes of P. putida present in other Pseudomonas but absent from P. entomophila. Furthermore, comparison of gene contents in P. entomophila and P. putida indicates that the higher number of species-specific genes in P. putida (1,774 versus 1,539) largely results from the presence of a higher number of paralogous genes (Fig. 2 and Supplementary Table 2 online). Comparison of the chromosome structures of P. entomophila and P. putida KT2440 and scatter plot analysis of syntenic regions of the two strains revealed frequent genetic inversions that reverse the genomic sequence symmetrically across oriC as observed in other bacterial genera11 (Fig. 2 and Supplementary Fig. 2 online). The same rearrangement profile was observed when comparing the P. entomophila genome with those of other Pseudomonas spp., even though the levels of orthology and of synteny were lower (see Supplementary Table 1 and Supplementary Fig. 2 online). A search for repetitive extragenic palindromic sequences (REPs) identified 943 REPs similar to those found in the genomes of P. putida KT2440 (ref. 12) and P. fluorescens Pf-5 (ref. 8). The genome of P. entomophila has been remodeled by genetic mobile elements and bacteriophage insertions considerably less than the genomes of other environmental pseudomonads such as P. putida KT2440 and P. syringae pv. tomato DC3000 (Fig. 1). Particularly notable are three clustered prophages related to FluMu phage, a pyocin-like phage and a lambdoid phage; they are inserted between recA and mutS, as observed for FluMu phage in P. fluorescens Pf-5 genome. Also of particular interest are two putative prophages inserted in genes encoding 4.5S RNA and tmRNA, respectively. The genome of P. entomophila contains only nine genes encoding transposase-like proteins including three that are remnant or inactive. Unlike the genomes of P. putida KT2440 and P. syringae pv. tomato DC3000, the genome of P. entomophila is devoid of type II introns.

Figure 2: Comparison of the P. entomophila and P. putida genomes.
figure 2

(a) Regions of significant sequence identity between the nucleotide sequence of P. entomophila (top) and P. putida KT2440 (bottom). Colinear regions are connected by red lines and inverted regions by blue lines. The display was generated using Artemis Comparison Toll (freely available at http://www.sanger.ac.uk/Software/ACT/). (b) Specific gene content comparison of the genomes of P. entomophila and P. putida KT2440. Specific genes of P. entomophila (Pe) and of P. putida KT2440 (Pp) with no ortholog in the other species are indicated in blue and green respectively, and are classified according to role categories as described in Figure 1. Two genes were considered as orthologs when their products share more than 60% identity over more than 80% of their length. Duplicated genes indicated by light colors were detected by using a constraint of 35% identity over more than 80% of the length of the protein. Aa, amino acid biosynthesis; Bc, biosynthesis of cofactors, prosthetic groups and carriers; Ce, cell envelope; Cp, cellular processes; Ci, central intermediary metabolism; Dm, DNA metabolism; Em, energy metabolism; Fam, fatty acid and phospholipid metabolism; Me, mobile and extrachromosomal element functions; P, protein synthesis and fate; Pp, purines, pyrimidines, nucleosides and nucleotides; Rf, regulatory functions and signal transduction; Sm, secondary metabolite biosynthesis; T, transcription; Tb, transport and binding proteins; Uf, unknown function and hypothetical proteins.

Full size image

Toxins against insects

We used several criteria to uncover genes that may contribute to the entomopathogenic properties of P. entomophila: specificity to the P. entomophila genome, localization within genomic islands that suggest recent lateral acquisitions (based on break of the synteny, GC content and absence of REPs) and similarity to genes associated with virulence in other systems (Table 2).

Table 2 Gene/gene products potentially involved in P. entomophila-D. melanogaster interaction
Full size table

Particularly striking are three genes absent from other Pseudomonas genomes that encode proteins related to insecticidal toxin complexes that have been found only in entomopathogenic enterobacteria such as Photorhabdus luminescens, Serratia entomophila, Xenorhabdus nematophilus or in Yersinia spp.13,14. Three basic types of genetic elements encode insecticidal toxin complexes: tcdA-, tcdB- and tccC-like genes. The P. entomophila genome encodes three TccC-type insecticidal toxins (PSEEN2485, PSEEN2697, PSEEN2788) (see Supplementary Fig. 3 online). In addition to these three insecticidal toxins, the P. entomophila genome, like that of P. syringae, encodes proteins more distantly related to TccC-type toxins (PSEEN701 and PSEEN702) and to TcdB-type toxins (PSEEN1172). The three P. entomophila insecticidal toxins likely play a major role in the pathogenicity of P. entomophila as TccC and TcdB proteins have been shown to have entomocidal activity15,16, even though the molecular mechanisms remain to be characterized. These findings highlight the efficient spreading of toxin-complex gene homologs in insect-interacting soil bacteria belonging to different genera.

Bacterial hemolysins are exotoxins that attack blood cell membranes and cause cell rupture by poorly defined mechanisms17. Contrary to the other Pseudomonas tested, P. entomophila secretes a strong diffusible hemolytic activity (see Supplementary Fig. 4 online) that may also be involved in pathogenicity against D. melanogaster. We identified three genes unique to P. entomophila that may be responsible for this activity (Table 2). The gene encoding PSEEN3925, a putative repeats-in-toxin (RTX) protein, is clustered with genes encoding a type I secretion system. PSEEN0968 and PSEEN3843 are proteins related to outer membrane autotransporters that have been associated with virulence in other bacteria. A number of lipases have also been shown to confer hemolytic activity. The P. entomophila genome encodes four lipases that are absent from P. putida KT2440 and that may contribute to its hemolytic activity (PSEEN709, PSEEN1065, PSEEN2195, PSEEN3432). Interestingly, the gene encoding a lysophospholipase (PSEEN709) is found in a genomic islet associated with two genes encoding proteins related to insecticidal toxins.

Proteases constitute another important group of extracellular, biologically active substances that are thought to contribute to the virulence of bacterial species. P. entomophila encodes three serine proteases (PSEEN3027, PSEEN3028, PSEEN4433) and an alkaline protease (PSEEN1550) absent from P. putida KT2440. These four genes are located at synteny break points between the genomes of P. entomophila and other Pseudomonas spp. PSEEN1550 is the homolog of the alkaline protease AprA, which has been shown to be involved in various virulence processes among different species18. AprA likely plays a key role in virulence because pathogenicity is affected in mutants defective in PrtR, the predicted transcriptional regulator of aprA (see below).

Pathogenic bacteria rely on a variety of cell surface–associated virulence factors that allow adhesion to the host surface and promote effective colonization. Filamentous hemagglutinin-like adhesins are broadly important virulence factors in both plant and animal pathogens. The genome of P. entomophila encodes three proteins (PSEEN0141, PSEEN2177, PSEEN3946) that are predicted to be involved in adhesion and cluster with genes encoding type I or two-partner secretion system proteins (Table 2). We also noticed the presence of two putative autotransporter proteins with a pertactin-type adhesion domain.

Toxins against competitors

In addition to the putative toxins described above that may be crucial for its entomopathogenic properties, P. entomophila carries a number of genes specifying diverse traits that may be required not only for interaction with insects but also for its lifestyle in soil, aquatic or rhizosphere environments (see Supplementary Fig. 5 online).

Fluorescent pseudomonads are characterized by the production of pyoverdines, a diverse class of siderophores containing a chromophore linked to a small peptide of varying length and composition synthesized by nonribosomal peptide synthases19. In P. entomophila, the two gene clusters that encode proteins required for pyoverdine biosynthesis and uptake (PSEEN1813-PSEEN1815 and PSEEN3224-3234) present a general organization similar to that found in other fluorescent pseudomonads20. We also identified a gene cluster responsible for the synthesis of a siderophore related to acinetobactin and containing a salicylamide moiety21 (Supplementary Fig. 5).

Five gene clusters that direct the production of secondary metabolites have been identified (see Supplementary Fig. 5). PSEEN5520-PSEEN5522 are responsible for hydrogen cyanide production that is involved in Caenorhabditis elegans killing by P. aeruginosa22 and in the suppression of soil-borne plant pathogens by certain Pseudomonas species23. The genome of P. entomophila contains four clusters of genes predicted to encode three different lipopeptides and a polyketide (Table 2 and Supplementary Fig. 5).

Regulation of virulence revealed by a genome-wide mutagenesis

To directly identify factors that modulate the interaction between P. entomophila and D. melanogaster, we generated a Tn5-derived library of variants that were individually screened for their infectious and pathogenic properties. Among the 7,500 clones, we isolated 23 mutants whose growth was not affected and that displayed attenuated infectious and/or pathogenic properties (Table 2). Identification of the mini-Tn5 insertion sites identified directly only a putative lipopeptide as a virulence factor. No other genes predicted to be virulence factors were identified, indicating a likely redundancy. By contrast, a number of insertions affected regulators that likely modulate the expression of such virulence factors. Seven independent insertions inactivated the two-component system GacS/GacA involved in the regulation of various processes, including virulence in different species, and resulted in the inability of these mutants to induce an immune response. P. entomophila gac mutants are defective in secretion of protease and hemolysin (data not shown) and do not persist in the gut of D. melanogaster1, indicating the pivotal role of GacS/GacA in modulating the entomopathogenic properties of that strain. As observed in other Pseudomonas species23, the GacS/GacA two-component system probably regulates P. entomophila virulence genes at a post-transcriptional level via the two identified small noncoding RsmY and RsmZ RNAs that alleviate post-transcriptional repression by RsmA and RsmE homologs. Three independent insertions in the prtR gene reduce the pathogenic properties of P. entomophila but retain the capacity to induce an immune response. In P. fluorescens LS107d2 (ref. 24), PrtR and PrtI regulate the transcription of the aprA-inh-aprDEF operon suggesting that P. entomophila relies on AprA protease to fully express its pathogenic properties in D. melanogaster. Two independent insertions that had the same consequences for the interaction with D. melanogaster have been found in algR. In P. aeruginosa, AlgR regulates a number of processes including fimbrial biogenesis, biofilm formation and cyanide production25,26. Altogether, genetic analysis indicates that GacA is a master regulator of the interaction and that PrtR and AlgR regulators, seem to play secondary roles in the infection process.

Metabolism, transport and regulation

The P. entomophila genome encodes most of the central metabolic pathways found in the other Pseudomonas including the pentose phosphate pathway, the Entner-Doudoroff pathway and the tricarboxylic acid cycle. Consistent with Pseudomonas metabolism, P. entomophila has an incomplete Embden-Meyerhof-Parnas pathway owing to the absence of 6-phosphofructokinase, and relies on a complete Entner-Doudoroff route for hexose utilization. The P. entomophila genome harbors several genes that encode hydrolytic activities such as chitinases, lipases and proteases as well as a set of 19 uncharacterized hydrolases, which are potentially involved in the degradation of polymers found in the soil. However, contrary to phytopathogenic strains such as P. syringae5,6,7, the genome of P. entomophila is devoid of genes encoding enzymes capable of degrading plant cell walls. This is consistent with the observation that this species is not pathogenic for plants (M. Arlat, Institut National de la Recherche Agronomique, Castanet, France, personal communication).

The P. entomophila genome also contains determinants for the catabolism of various aromatic compounds (see Supplementary Fig. 6 online) and long-chain carbohydrates. P. entomophila shares several gene clusters with P. putida27 that are involved in the degradation of various classes of aromatic compounds including benzoate and quinate, 4-hydroxybenzoate, phenylacetaldehyde and phenylalkanoate as well as phenylalanine and tyrosine. The P. entomophila genome contains two additional catabolic gene clusters present in the genome of P. aeruginosa PAO1 that encode determinants for the degradation of 3-hydroxybenzoate through gentisate28 and for the meta-cleavage of homoprotocatechuate29,30.

Consistent with the size of its genome, P. entomophila possesses more than 535 transporter-encoding genes. Remarkably, no genes encoding a type III or type IV secretion system, present in numerous Gram-negative bacterial pathogens31, were found in P. entomophila. The high numbers of transcriptional regulators (more than 300) and genes whose products are involved in signal transduction suggests that P. entomophila is able to adapt to substantial substrate variations in its habitats.

The soil and entomopathogenic lifestyle of P. entomophila

The metabolic properties of P. entomophila predicted from its genome suggest that this strain is a ubiquitous, metabolically versatile bacterium that may colonize diverse habitats including soil, rhizosphere and aquatic systems as shown for P. putida KT2440. However, in contrast to P. putida, P. entomophila contains a number of genes that are predicted, or have been shown, to be important for virulence. The expression of these factors is under the control of the major regulator GacA and presumably allows this strain to exploit new niches and interact with various insects, particularly D. melanogaster (Fig. 3).

Figure 3: Steps in the interaction between P. entomophila and D. melanogaster.
figure 3

Five different steps are shown: 1. ingestion of P. entomophila through the esophagus; 2. resistance to oxidative stress in response to a oxidative burst in the gut; 3. persistence of P. entomophila in the gut; 4. escape from immune response effectors; 5. pathogenicity and lethal outcome of the interaction after important modifications of the midgut physiology including microvilli disruption, cell destruction (indicated by a *) and in some cases peritrophic matrix disorganization (indicated by a dashed line). Red indicates important steps in the infection process. Blue indicates newly identified proteins that could be involved at these steps in the process. Time scale is indicated in brackets. Ep, epithelial cell; mv, microvilli; PM, peritrophic matrix; gc, gastric cecum.

Full size image

In D. melanogaster, an environment hostile for microbial colonization is maintained in the gut by secretion of antimicrobial factors such as lysozymes32,33 and other digestive enzymes. Recently, it has been shown that a unique epithelial oxidative burst limits microbial proliferation in the gut34; resistance to oxidative stress might therefore be a prerequisite for D. melanogaster gut colonization. The P. entomophila genome encodes 40 proteins that are predicted to be involved in resistance to oxidative stress including four catalases, two superoxide dismutases, three hydroperoxide reductases and eleven glutathione-S-transferases. It is noteworthy that resistance to oxidative stress is probably not sufficient for colonization as other Pseudomonas species that possess a large repertoire of oxidant detoxifying proteins are not able to persist in the gut of D. melanogaster1. This assumption is further reinforced by the observation that P. entomophila gacA mutants were not less resistant to peroxide, hypochlorite or paraquat (data not shown). As the P. entomophila-D. melanogaster interaction is specific, P. entomophila infectivity likely involves the expression of a specific gene enabling this strain to persist in the D. melanogaster gut, as shown for the Erwinia carotovora Evf factor35. Because P. entomophila does not contain any evf-related genes, we cannot predict candidates for this putative persistence promoting factor (ppf in Fig. 3). Nonetheless, this gene is likely regulated by the GacS/GacA two-component system: the gacA::Tn5 or gacS::Tn5 mutants do not persist in the gut and P. entomophila cells are infectious only at stationary phase, concomitant with Gac activation of virulence genes (data not shown). It is striking to note that in both P. entomophila and E. carotovora35, genes required to interact with D. melanogaster are under the control of global regulators, that is, Hor and GacA, respectively, revealing the branching of virulence genes in a complex regulatory network.

Infection of D. melanogaster by P. entomophila is accompanied by blockage of food-uptake1. This phenomenon is also observed in the interaction between Serratia entomophila and the grass grub Costelytra zealandica or between Yersinia pestis and the flea. The processes used to effect food blockage seem to be different in the two systems; Y. pestis relies on phospholipase synthesis and biofilm formation36,37 whereas the mechanism used by S. entomophila remains unknown38. Genes responsible for the anti-feeding determinants of S. entomophila have a prophage origin and no related genes were identified in the genome of P. entomophila. Since algR mutants still provoke food-uptake blockage, biofilm formation is probably not essential for D. melanogaster infection by P. entomophila.

The persistence of P. entomophila in the larval gut triggers both a local and systemic immune response1. The P. entomophila level remains high in wild-type larvae, similar to that observed in a relish mutant unable to induce an immune response1, suggesting that this strain is able to escape the D. melanogaster immune response. Biofilm formation might protect P. entomophila cells from immune effectors or persistence of bacteria might result from the degradation of effectors. The defects observed with prtR mutants indicated that AprA may degrade antimicrobial peptides, as indicated by recent in vivo studies39, and consequently disable the immune response.

Twelve hours after D. melanogaster ingests the bacteria, physiological modifications to the fly caused by P. entomophila are dramatic and the expression of 205 D. melanogaster genes is modified1 (Fig. 3). These changes probably result from the action of virulence factors such as proteases, hemolysins, insecticidal toxin-like proteins, secondary metabolites or hydrogen cyanide. However, lethality starts to be apparent after 16 h, indicating that this late gene expression will have no effect on the fatal outcome of the interaction.

Discussion

The complete genome sequence of P. entomophila provides insight into this organism's entomopathogenic lifestyle. Combined with a genetic approach, it has revealed potential virulence factors along with regulators that modulate their expression. This study also revealed that P. entomophila is the first Pseudomonas strain to be pathogenic in a multicellular organism and at the same time to be devoid of a type III secretion system. Its potential to use various plant-derived compounds including aromatic molecules, and its antibiotic- and oxidative stress-resistance capacities suggest that P. entomophila is a commensal bacterium. As this strain is not a plant pathogen, it may have potential to control insects. Unexpectedly for an environmental isolate, P. entomophila has a genome that contains a limited number of bacteriophages and transposons. This may contribute to its relatively small size compared to other Pseudomonas genomes. Finally, the complete genome sequence of P. entomophila provides a framework for further studies to characterize its pathogenic properties and for a host-pathogen system in which both organisms are amenable to genetic and genomic analysis.

Methods

Genome sequencing, assembly and annotation.

The complete genome sequence of P. entomophila L48 was determined using the whole-genome shotgun method (10 × coverage, using two plasmid libraries and one BAC library to order contigs). Finishing was performed by PCR amplification from contigs extremities. After a first round of annotation, regions of lower quality as well as regions with putative frameshifts were resequenced from PCR amplification of the dubious regions. Using the AMIGene software (annotation of microbial genes)40, a total of 5,279 CoDing Sequences were predicted (and assigned a unique identifier prefixed with “PSEEN”), and submitted to automatic functional annotation: exhaustive BLAST searches against the UniProt databank were performed to determine significant homology. Protein motifs and domains were documented using the InterPro databank. In parallel, genes coding for enzymes were classified using the PRIAM software41. TMHMM vs2.0 was used to identify transmembrane domains42, and SignalP 3.0 was used to predict signal peptide regions43. Finally, tRNAs were identified using tRNAscan-SE44. Sequence data for comparative analyses were obtained from the NCBI databank (RefSeq section). Putative orthologs and synteny groups (that is, conservation of the chromosomal colocalization between pairs of orthologous genes from different genomes) were computed between P. entomophila and all the other complete genomes as previously described45. Manual validation of the automatic annotation was performed using the MaGe (Magnifying Genomes) interface, which allows graphic visualization of the P. entomophila annotations enhanced by a synchronized representation of synteny groups in other genomes chosen for comparisons45. All the data (that is, syntactic and functional annotations, and results of comparative analysis) were stored in a relational database, called EntomoScope. This database is publicly available via the MaGe interface at http://www.genoscope.cns.fr/agc/mage/.

Bacterial mutagenesis and screening.

Random mutagenesis was performed by biparental mating using P. entomophila1 and Escherichia coli S17.1-λpir46 carrying the pUT-Tn5-Tc suicide plasmid as previously described47. A total of 7,500 TcR colonies obtained from several independent conjugations were screened individually as previously described35. Transconjugants that displayed attenuated virulence were subjected to several secondary screenings by natural infection as previously described1. Insertion sites were determined using two different methods. First, genomic DNA was digested by PstI or NotI/PstI and ligated into pUC18 and pBlueScript, respectively. Clones that contained the mini-transposon and its flanking sequences were selected by plating the E. coli BW25142 transformants on tetracycline (10 μg/ml). One flanking region was sequenced from the Tc gene using the oligonucleotide (Tc-F) 5′-TCGTCGACAAGCTTCGG-3′. Some insertion sites were determined by reverse PCR method. Genomic DNA was digested by either PstI or EagI, self-ligated and amplified using the oligonucleotides Tc-F and 5′-AGATCTGATCAAGAGACAT-3′ for PstI-digested DNA or 5′-GGCGGCCCTATACCTTGTCTG-3′ (Tet-end) and 5′-CATAATGGGGAAGGCCAT-3′ for EagI-digested DNA, respectively. One flanking region was sequenced using the oligonucleotides Tc-F or Tet-end. Insertion sites were confirmed by amplifying the region overlapping the insertion site. Southern blot analysis was carried out to verify that the selected clones only carried a single copy of the transposon.

Accession numbers.

The P. entomophila nucleotide sequence and annotation data have been deposited in the EMBL databank under accession number CT573326.

Note: Supplementary information is available on the Nature Biotechnology website.