Multiple roles of Pseudomonas aeruginosa TBCF10839 PilY1 in motility, transport and infection

Polymorphonuclear neutrophils are the most important mammalian host defence cells against infections with Pseudomonas aeruginosa. Screening of a signature tagged mutagenesis library of the non-piliated P. aeruginosa strain TBCF10839 uncovered that transposon inactivation of its pilY1 gene rendered the bacterium more resistant against killing by neutrophils than the wild type and any other of the more than 3000 tested mutants. Inactivation of pilY1 led to the loss of twitching motility in twitching-proficient wild-type PA14 and PAO1 strains, predisposed to autolysis and impaired the secretion of quinolones and pyocyanin, but on the other hand promoted growth in stationary phase and bacterial survival in murine airway infection models. The PilY1 population consisted of a major full-length and a minor shorter PilY1* isoform. PilY1* was detectable in small extracellular quinolone-positive aggregates, but not in the pilus. P. aeruginosa PilY1 is not an adhesin on the pilus tip, but assists in pilus biogenesis, twitching motility, secretion of secondary metabolites and in the control of cell density in the bacterial population.


Introduction
Pseudomonas aeruginosa is a metabolically versatile Gram-negative bacterium, which inhabits terrestrial and aquatic as well as animal, human and plant host associated environments (Ramos, 2004). This opportunistic pathogen is the most dominant bacterium causing chronic infections in the cystic fibrosis (CF) lung (Lyczak et al., 2002;Gómez and Prince, 2007) and has emerged as an important causative agent of nosocomial infections, particularly in intensive care units (Trautmann et al., 2005). The first step in establishing an infection is the adherence and colonization of the epithelium, which is mediated, in part, by type IV pili (T4P) (Klausen et al., 2003;Giltner et al., 2006). These T4P are also receptors for bacteriophages and mediate a mode of surface translocation termed twitching motility (Mattick, 2002).
Pseudomonas aeruginosa T4P are polar organelles that are composed of a single protein subunit, PilA. The PilA subunits are first synthesized as prepilins. In the course of assembly in the periplasmic space, the leader peptide is cleaved and the resulting N-terminal residue is methylated (Whitchurch, 2006). The assembled pilus filament is extruded through the outer membrane via a gated channel, the multimeric PilQ secretin (Martin et al., 1993;Wolfgang et al., 2000;Collins et al., 2001). Besides the major T4P prepilin PilA, the P. aeruginosa genome contains one gene cluster (PA4549-PA4556) that encodes PilY1, PilY2 and the six minor T4P prepilins FimT, FimU, PilV, PilW, PilX and PilE (Mattick, 2002;Whitchurch, 2006). The fimU-pilVWXY1Y2E genes are cotranscribed in an operon which is under the control of the response regulator AlgR (Belete et al., 2008).
The minor prepilins possess the canonical hydrophobic N-terminal sequence signature of prepilins (Mattick, 2002;Whitchurch, 2006). They are membrane-located and are required for pilus biogenesis (Russell and Darzins, 1994;, but do not seem to be incorporated into the extracellular T4P structure. The two non-prepilin proteins that are encoded by the operon fimU-pilVWXY1Y2E, PilY1 (PA4554) and PilY2 (PA4555) are also necessary for pilus formation . In addition, PilY1 has been reported to be involved in the control of the expression of the P. aeruginosa lipase LipC (PA4813) (Martínez et al., 1999). The PilY1 protein carries a prepilin peptidase export signal and is transported across the inner membrane (Lewenza et al., 2005). PilY1 of P. aeruginosa PAO1 shares 43% homology to the C-terminal sequence of PilC2 of Neisseria meningitidis and of Neisseria gonorrhoeae. Neisserial PilC resides in the bacterial outer membrane (Rahman et al., 1997) and in the pilus tip (Rudel et al., 1995). The PilC proteins whose expression can be switched on and off by phase variation (Rytkönen et al., 2004), are major adhesins and regulate PilTmediated T4P retraction in Neisseria (Morand et al., 2004). This datum indicates that the 3.5 kb P. aeruginosa pilY1 homologue may encode a protein with multiple functions including, but not restricted to, pilus biogenesis.
Pseudomonas aeruginosa TBCF10839 (Tümmler et al., 1991) is a non-piliated CF isolate that carries an out-offrame deletion in the pilQ gene (Chang et al., 2007). Hence this strain provided the opportunity to explore functions of pilus biogenesis genes that are not directly related to their essential roles in piliation. In this study the strain TBCF10839 and mutants thereof were exploited to gain knowledge about the yet unresolved function(s) of P. aeruginosa PilY1. Besides its known roles in piliation and twitching, PilY1 was found to be involved in numerous other phenotypes, namely the secretion of phenazines and quinolines, bacterial cell lysis, virulence and mammalian host defence.

Results
Screening of a signature tagged mutagenesis library of P. aeruginosa TBCF10839 in a polymorphonuclear neutrophil phagocytosis assay Polymorphonuclear neutrophils (PMNs) are the most important mammalian host defence cells against infections with P. aeruginosa (Döring et al., 1995). A mini-Tn5 plasposon library of the CF airway isolate TBCF10839 carrying variable V 40 (V = A, G, C) oligonucleotide signature tags were screened for survival in freshly prepared human PMNs in phagocytosis assays. Batches of 48 mutants each carrying a different signature tag were exposed at a total multiplicity of infection of 10 to PMNs and viable bacteria were recovered after 2 h. Numerous transposon mutants were identified which survived at significant lower frequencies in the phagocytosis assays than the wild-type strain (L. Wiehlmann, in preparation). Unexpectedly, four mutants consistently showed higher survival rates than TBCF10839, suggesting that the inactivation of this gene by the insertion of a transposon increased resistance to killing by PMNs. Sequencing revealed that the transposon had inserted into two genes of yet unknown function (P. Salunkhe, in preparation) or into two different positions of the 3486 bp large pilus biogenesis gene pilY1 [PA4554, positions 1617 (mutant 10CB5) or 2308 (mutant 25C8)].

PilY1 phenotype in a non-piliated host strain
PilY1 was identified in P. aeruginosa as the homologue of the gonococcal PilC which is the T4P tip-located adhesin . P. aeruginosa strain TBCF10839, however, is non-piliated due to an out-of-frame deletion in the pilQ gene (Chang et al., 2007). The dodecameric transmembrane protein PilQ plays a central role in extruding the pilus fibre through the bacterial outer membrane and thereby facilitates the presence and functionality of extracellular pili (Whitchurch, 2006). Hence, the transposon mutants in the pilY1 gene of strain TBCF10839 are double knockouts in the polycistronic operons pilM-NOPQ (PA5044-PA5040) and pilVWXY1Y2E (PA4551-PA4556). Correspondingly, neither strain TBCF10839 nor its pilY1 transposon mutants were twitching (Supporting information, Fig. S1). Complementation in trans with the functional PAO1 pilQ gene conferred twitching to TBCF10839 to levels seen in reference strain PAO1, but did not restore twitching in the pilY1 mutant (Fig. S1). Transposon insertions into the piliated and twitchingproficient PA14 strain (Rahme et al., 1995) also led to the loss of twitching (data not shown). These findings confirm the report by  that a functional PilY1 protein is necessary for twitching.
As already the TBCF10839 host strain was phenotypically silent in twitching, we hypothesized that PilY1 should have further roles in P. aeruginosa beyond pilus-mediated motility. If this is the case, the PilY1 protein should be present in the non-piliated host. The pilY1 gene of strain TBCF10839 was sequenced and highly sensitive and specific polyclonal peptide anti-PilY1 antibodies were raised against epitopes that are shared between the PilY1 orthologues of strains TBCF10839 and PAO1. PilY1immunoreactive bands of predicted size (126 kDa) and similar intensity were detected on immunoblots of PAGEseparated cell lysates from strains PAO1 and TBCF10839 ( Fig. 1) suggesting that PilY1 was present in similar amounts in the piliated and the non-piliated strain. The presence of PilY1 irrespective of the status of piliation provided the first evidence for roles of the PilY1 protein other than pilus biogenesis that should be associated with susceptibility to killing by PMNs as a surrogate function.
No PilY1-immunoreactive bands were seen in the cell lysates of the pilW mutant 14D1 (transposon insertion at position 497) and of the pilY1 mutant 25C8 with the distal insertion of the transposon at position 2308 ( Fig. 1). A faint band of about 59 kDa was recognized by the anti-PilY1 antibody in the pilY1 mutant 10CB5 with the proximal insertion at position 1617 ( Fig. 1, Fig. S2). This datum shows that the insertion of a transposon in pilW upstream of pilY1 in the fimU-pilVWXY1Y2E operon had a strong polar effect and caused a knockout of PilY1. Mutagenesis within pilY1 itself led either to a null mutant or to a leaky mutant with minute production of truncated protein.
Complementation of the null mutant in trans led to overproduction of PilY1 (Fig. 1).
The strong PilY1-immunoreactive signals of PAO1 and TBCF10839 cell lysates shown in Fig. 1 were obtained from 2 ml of culture. The corresponding extracellular protein fractions that contain secreted proteins or sheared cellular appendages like pili were immunonegative for PilY1 on blots. To improve the sensitivity, supernatants of bacterial cultures were 100-fold concentrated for proteins larger than 50 kDa by ultrafiltration and then immunoprecipitated with anti-PilY1 antibody. A PilY1 signal of predicted size of the full-length protein was still not detectable on the blots. However, an immunoreactive band of about 88 kDa was visible in the immunoprecipitate from TBCF10839 (Fig. 2). The same band became apparent on blots of whole cell lysates of TBCF10839 after longer exposure with the ECL detection reagent, whereas blots derived from cell lysates of PA14 that lack the epitopes used for anti-PilY1-antibody generation, showed no signal (Fig. S2B). This datum demonstrates that the anti-PilY1 antibody recognized PilY1-specific epitopes in the 88 kDa protein. As the immunoreactive 88 kDa signal was seen neither in cell lysates of PA14 nor in the pre-immune control of the immunoprecipitations, this 88 kDa band most likely represents a shorter isoform PilY1* of strain TBCF10839. Only PilY1*, but not the full-length PilY1, was present in the extracellular protein fraction. As the PilY1*immunoreactive signal became only visible by immunoprecipitation from concentrated supernatant, maximal 5%, but probably even less of total PilY1 resided in the extracellular fraction. No PilY1* signal was detected in the immunoblots of the TBCF10839 pilY1 transposon mutant (see Fig. 1 and Fig. S2 for the cell lysate and Fig. 2 for the immunoprecipitate of concentrated supernatant).

mRNA expression profiling
Next, evidence for the putative role(s) of PilY1 was sought by comparative mRNA expression profiling. RNA was extracted from strain TBCF10839 and its isogenic pilY1 transposon mutant 25C8 grown in Luria-Bertani (LB)  immunoprecipitates of the extracellular protein fraction that had been concentrated about a 100-fold by ultrafiltration. From left to right: PilY1-immunoreactive signals of the concentrated extracellular protein fraction of TBCF10839 pilY1::Tn5 (25C8) (lanes 1, 2, 3) or of TBCF10839 (lanes 4, 5, 6) precipitated with beads (lanes 1, 4; negative control), rabbit polyclonal anti-PilY1(TB)-antibody (lanes 2, 5) or pre-immune serum from the same rabbit (lanes 3, 6; negative control). The anti-PilY1 antibodies solely detected the shorter PilY1* isoform, but no full-length PilY1 in the supernatant of TBCF10839 cultures (lane 5). broth to late exponential phase, reverse-transcribed and hybridized onto P. aeruginosa PAO1 GeneChips. The mRNA expression profile of the transposon mutant in comparison with the wild-type strains showed the significant upregulation of 20 genes and significant downregulation of 56 genes (Supporting Information, Table S1). Genes fimUpilVWX upstream of pilY1 were the most strongly upregulated genes indicative for a feedback regulation of expression of the whole operon that was destroyed by the insertion of the transposon. Of the 71 other genes that were classified as differentially regulated, only eight genes have yet been experimentally characterized. Within the group of moderately or strongly expressed genes that were differentially regulated, conserved hypotheticals and putative enzymes of unknown function had a share of more than 80%.
Hence, the higher mRNA transcript levels of PilY1 and of the minor prepilins FimU, PilV, PilW and PilX in the pilY1 mutant (Table S1) were the only interpretable differences in the global expression profile between TBCF10839 and TBCF10839 pilY1::Tn5. This observation was verified by RT/PCR kinetics. mRNA transcript levels were measured in piliated PA14 and non-piliated TBCF10839 and their isogenic pilY1 mutants (Fig. S3). PilW mRNA transcripts were about 40-and 20-fold and mutant PilY1 mRNA transcripts were about 10-and 50-fold more expressed in the transposon mutants than in their parental TBCF10839 and PA14 strains respectively.

Secretion of secondary metabolites
Pseudomonas aeruginosa secretes numerous coloured secondary metabolites (Ramos, 2004). Both TBCF10839 and PA14 are strong producers of these compounds, but their isogenic pilY1 mutants were colourless. Notably none of the known genes involved in the biosynthesis of secondary metabolites was differentially regulated in the TBCF10839 pilY1::Tn5 (25C8) transcriptome. Thus a process at the post-transcriptional level probably affected the secretion of these compounds in the pilY1 mutants.
To characterize this phenotype in more depth, the time-course of growth and of the secretion of the phenazine pyocyanin and of the siderophore pyoverdine were monitored for strains TBCF10839, PA14 and their pilY1 mutants 25C8, 25263 and 25563 in King A and King B medium respectively. In both genetic backgrounds the growth curves uncovered two remarkable features of the mutant strains (Fig. 3). First, the pilY1 mutants were growing to higher bacterial densities (panels A and B in Fig. 3), and second, they secreted less pyocyanin at Values were normalized to the pyocyanin secretion of the TBCF10839 and PA14 wild-type strains respectively. 1, strain PAO1; 2, TBCF10839; 3, TBCF10839 complemented with pME6010 (vector control); 4, TBCF10839 pilY1::Tn5 (25C8); 5, TBCF10839 pilY1::Tn5 (25C8) complemented with pME6010; 6, TBCF10839 pilY1::Tn5 (25C8) complemented with pME6010::TBpilY1; 7, PA14; 8, PA14 MAR2xT7 pilY1 mutant ID25263; 9, PA14 MAR2xT7 pilY1 mutant ID25563. Means of absolute values were examined for significant differences between two strains by two-tailed t-test and subsequent Bonferroni correction for multiple testing (*P < 0.05; **P < 0.01; ***P < 0.001). all time points (panels C and D). Complementation of TBCF10839 pilY1::Tn5 restored the secretion of pyocyanin (Fig. 3E). The production of pyoverdine was not affected by PilY1 deficiency (data not shown) which was consistent with the finding that the global production of siderophores on chromazurol iron indicator plates was similar for wild-type strains, pilW and pilY1 mutants and their complemented revertants (Fig. S4B).
Pseudomonas aeruginosa possesses an extended network of N-acylhomoserine lactone (AHL) quorumsensing (QS) systems that upregulate large, overlapping sets of genes including the pyocyanin biosynthesis gene cluster (Whiteley et al., 1999;Juhas et al., 2004;Schuster and Greenberg, 2006). Defective QS may improve fitness and growth in stationary phase (Heurlier et al., 2006). The stronger growth in stationary phase and the impaired secretion of pyocyanin of the PA14 and TBCF10839 pilY1 mutants (see Fig. 3) thus mimics aspects of a QS-defective phenotype. Hence we checked AHL production in our strain panel. However, as shown in Fig. S5, all strains were similarly proficient in the production of short-chain and long-chain AHLs. In other words, inactivation of pilY1 neither directly nor indirectly affected secretion of AHLs.
A major readout of AHL-mediated cell-to-cell communication is the stimulation of type II secretion virulence effectors such as elastases. Protease secretion can be screened on agar plates that contain casein as sole carbon source (Cowell et al., 2003). Figure S4A demonstrates that PA14, TBCF10839 wild type and its pilW and pilY1 transposon mutants were proficient in casein degradation. In summary, there is no evidence that the AHL-dependent QS systems are perturbed in the pilY1 mutants that could account for the impaired secretion of pyocyanin.
Besides phenazines and siderophores, the third major class of coloured secondary metabolites in P. aeruginosa that emit visible light by fluorescence are the 4-hydroxy-2-alkylquinolines (HAQs) which include antimicrobial N oxides and the cell-to-cell communication molecules 3,4-hydroxy-2-heptylquinoline [pseudomonas quinolone signal (PQS)] and 4-hydroxy-2-heptylquinoline (HHQ) (Pesci et al., 1999;Lépine et al., 2004;Xiao et al., 2006). Cultures of pilY1 and pilW transposon mutants of TBCF10839 (Fig. 4B), but not cultures of pilY1 mutants of PA14 (Fig. S6) grown for 15 h to late exponential phase contained lower amounts of HAQs than cultures of the wild-type strain. The more hydrophobic compounds including PQS were underrepresented. Supply of the pilY1 or pilW gene in plasmid pME6010, but not plasmid alone, restored the composition of HAQs in TBCF10839 mutants to wild-type levels (Fig. 4).
In summary, the reduced secretion of pyocyanin was the major cause for colourless cultures of the PA14 and TBCF10839 pilY1 mutants. The latter mutants also produced and secreted less HAQs.
Autolysis has been described to be induced either by bacteriophage (see Fig. 6, no. 14) or by the uncontrolled overproduction of HAQs which can be mimicked by the inactivation of the pqsL (PA4190) gene (Fig. 6, no. 13) (D'Argenio et al., 2002). The morphotype of pilY1 and pilW mutants did not resemble that of phage-induced lysis. Electron microscopic examination of colonies of the mutants (see below) did not observe any phage, whereas numerous phages were seen after transduction of strain PAO1 with phage F116 as positive control (Fig. S8). The pilY1 and pilW mutant morphotype was also phenotypically different from that of a HAQ overproducer. The pqsL transposon mutant of TBCF10839 exhibited a patchy morphotype with lytic and non-lytic zones side-by-side throughout the colony (Fig. 6, no. 13). Colonies were glossy in the HAQ-overexpressing pqsL mutant.

PilY1 is associated with small extracellular aggregates in electron microscopy
To seek for extracellular structures that are present in PilY1-proficient, but absent in PilY1-deficient cells, cultures of wild-type strains, pilW or pilY1 mutant and complemented strains were compared by anti-PilY1 immunogold electron microscopy (Fig. 7). Negative staining visualized various extracellular structures, namely cellular appendages (pili, flagella), vesicular structures and small aggregates, the latter of which was decorated with the anti-PilY1 antibody.
TBCF10839 bacteria were surrounded by 30-80 nm round aggregates (Fig. 7B). Isolated smaller irregularly formed extracellular products were decorated with the anti-PilY1 antibodies. PAO1 bacteria were surrounded by only very small amounts of vesicular structures. The PilY1 antibody exclusively decorated the small extracellular aggregates (Fig. 7A). Examination of 70 mm of pili did not reveal a single gold particle along or at the end of the pili in contrast to 130 particles that were counted in a 70 mm stretch of nearest neighbour aggregates. This datum demonstrates that in P. aeruginosa the PilY1 protein is not associated with any extracellular pilus structure (P = 10 -25 , Fisher's exact test), but with small aggregates. In piliated TBCF10839 complemented with the PAO1 pilQ gene the small aggregates were decorated with the PilY1 antibody gold particles as seen for the wild-type strain (Fig. 7E), but no immunoreactive signal was detected in association with the pili. Cultures with PilY1-deficient mutants exhibited collapsed vesicles 50-80 nm in diameter, but no small aggregates and no T4P structures ( Fig. 7F and H). The PilY1 mutants complemented with a pilY1 containing plasmid were surrounded by 30-50 nm thick round vesicles and collapsed structures and by very small aggregates that were decorated with the PilY1 antibody ( Fig. 7G and I). The PilW mutant showed a few PilY1 antibody-positive aggregates (Fig. 7J), but on the other hand the complementation in trans with recombinant pilW plasmid led to fewer antibody-positive aggregates than the complementation of pilY1 mutants with pilY1 (compare Fig. 7K with Fig. 7G and I).
In summary, cells of all investigated P. aeruginosa strains were found to be surrounded by vesicular structures smaller than 100 nm in size. Pili were only detected in strain PAO1 and strain TBCF10839 complemented with the pilQ gene of strain PAO1. Cultures of PilY1-proficient cells, but not those of PilY1-deficient cells, contained very small extracellular aggregates ('nanoparticles') that were decorated by PilY1 antibody. As the immunoprecipitation experiments detected only the~88 kDa isoform PilY1*, but not full-length PilY1 in the concentrated extracellular protein fraction, the PilY1-immunoreactive signals in the electron microscopic examination should represent the localization of PilY1* in the extracellular medium.
The exact composition of the PilY1*-positive nanoparticles is not known, but according to our data they contain -besides probably further, yet uncharacterized components -PilY1* and the spectrum of TLC-separated hydrophobic secondary metabolites that were extracted from the immunoprecipitates of the > 50 kDa fraction of spent supernatants. According to the R f-values, this fraction contained HAQs including HHQ, but was free of pyocyanin and partially depleted of PQS.

Impact of PilY1 on virulence in worms and mice
The in vitro analyses demonstrated that the pilY1 mutants were growing better in stationary phase, were prone to autolysis and despite normal AHL production deficient in the secretion of pyocyanin. We wanted to test the impact of this phenotype on the virulence of the TBCF10839 strain in vivo.
The 'fast-killing' of Caenorhabditis elegans has been ascribed to the pyocyanin-mediated production of reactive oxygen species (Mahajan-Miklos et al., 1999;Tan et al., 1999). In accordance with the literature reports, the pyocyanin-deficient TBCF10839 pilY1 mutant was attenuated in the killing of the nematodes (Fig. 8). On the other hand, of more than 3500 screened TBCF10839 transposon mutants, the two pilY1 mutants were most robust against killing by PMNs (see above). As PMNs represent the major recruited mammalian cellular host defence against P. aeruginosa, we hypothesized that the pilY1 mutants should be endowed with higher fitness to colo- nize and persist in a mammalian host. To test this hypothesis, one pilY1 and one pilW mutant and 23 other differentially tagged non-auxotrophic signature tagged mutagenesis (STM) TBCF10839 mutants were inoculated into murine airways by intratracheal instillation. Recovery of the surviving bacteria 48 h later demonstrated that the pilW and pilY1 mutants were obtained in higher yields from the infected mouse lungs than from pools of the same mutants that had been cultured in parallel in LB broth in vitro (Fig. S9A). Competition experiments with two other sets of isogenic STM mutants showed again that the pilW and pilY1 mutants had the highest fitness to persist and grow in murine airways ( Fig. S9B and C). In parallel to the higher survival of the pilY1 mutant in the murine habitat, the immigration of PMNs into the lungs was evoked in the mouse. A dose of 8 ¥ 10 5 alginate-embedded TBCF10839 bacteria or of its isogenic pilY1 mutant 25C8 was instilled into murine airways. The bronchoalveolar lavage fluid collected 3 or 4 days after infection with the pilY1 mutant contained higher number of white blood cells that could be attributed to the significant increase of PMNs (P < 0.05) (Fig. 9). The lack of PilY1 led to a stronger emigration of PMNs into the airways. The stronger attraction of PMNs to the infected site, however, was not associated with increased morbidity. When mice were challenged with the same doses of planktonic colony-forming units (cfu) of TBCF10839 wild type, TBCF10839 pilW::Tn5 or TBCF10839 pilY1::Tn5, the signs of acute illness as indicated by the body condition score (Munder et al., 2005) and the intermittent drop of body weight and rectal temperature were lower upon exposure to the pilW or pilY1 transposon mutant (Fig. S10). By day 2 the wild-type strain had caused a profound and the mutants a moderate pneumonia (see histopathology in Fig. 10). Although the transposon insertion into pilY1 rendered the P. aeruginosa cell resistant to killing by the most important professional phagocyte, the global virulence of the strain was attenuated.

Discussion
Signature tagged mutagenesis has been originally developed by David Holden and colleagues to identify new virulence genes of Salmonella typhimurium in animal infection models (Hensel et al., 1995), but during the last years the method has been adopted to genome wide scans of other bacteria in disease habitats and the environment (Shea et al., 2000;Mazurkiewicz et al., 2006). The beauty of the technology is an in vivo selection process done by the habitat among a mixed population of mutants (Levesque, 2006). STM has the potential to discover novel and unexpected targets and pathways. Such a surprising finding was the starting point of this work. Screening of a STM library in a non-piliated P. aeruginosa strain uncovered that transposon inactivation of its pilY1 gene rendered the bacterium more resistant against killing by PMNs than the wild type and any other of the more than 3500 tested mutants. PMN-mediated phagocytosis and killing is the most important mammalian host defence mechanism against P. aeruginosa (Döring et al., 1995) and hence we wanted to resolve the phenotypic commonalities and differences between PilY1-proficient and PilY1-deficient isogenic strains.
The fimU-pilVWXY1Y2E operon encodes five minor prepilins and two non-prepilin proteins, namely PilY1 and PilY2 (Belete et al., 2008). The prepilins are involved in export, maturation and presentation of the major fimbrial subunit, when these genes are expressed in the correct stoichiometric ratios (Alm and Mattick, 1995;. The prepilins, but also PilY1 (see Fig. S1) are necessary to confer twitching motility. However, to date there are no  reports that show the incorporation of the proteins in the T4P structure. This fits with our observation that the PilY1 protein was present in similar amounts in the piliated PAO1 and the non-piliated pilQ-defective TBCF10839 strains (Fig. 1).
The pilY1 gene has been annotated to encode the tip-associated adhesin (Pseudomonas database http:// www.pseudomonas.com; Stover et al., 2000) like its homologue in N. gonorrhoeae PilC (Rudel et al., 1995). Figure S11 shows a dendrogram of the closest currently known homologues of PilY1 based on CLUSTALX protein sequence alignments. The neisserial PilC and the P. aeruginosa PilY1 proteins form separate branches. The substantial sequence diversity may result in different functions, and indeed no PilY1-immunoreactive signals were associated with the extracellular parts of PAO1 pili in electron microscopy (Fig. 7). Hence, the claim in the database is not supported that PilY1 in P. aeruginosa is a tip-associated adhesin.
Neisseria gonorrhoeae PilC1 and PilC2 are multifunctional proteins in correspondence with their dual localization in cellular appendages and the bacterial outer membrane (Rahman et al., 1997), i.e. they promote adhesion, piliation and transformation competence that are executed by different modules (Morand et al., 2001). This study demonstrates that P. aeruginosa PilY1 is also a multifunctional protein. First, as shown in this and previous work , PilY1 is indeed essential for piliation. Second, and this is new, PilY1 of P. aeruginosa is involved in the release of secondary metabolites into the extracellular space and is necessary to prevent autolysis in a bacterial colony. The latter two functions of PilY1 were observed in both piliated and non-piliated strains and hence are independent of pilus assembly.
The anti-PilY1 antibody recognized, besides the fulllength PilY1, a shorter isoform PilY1*. The specificity of the PilY1* signal was corroborated by the immunonegative blots of isogenic PilY1 knockouts and of strain PA14 whose PilY1 protein does not contain the epitopes against which the antibody was raised ( Fig. 1 and Fig. S2). PilY1* was the only PilY1 isoform in immunoprecipitates of the extracellular protein fraction of TBCF10839. Hence PilY1* should be the PilY1 isoform that was visualized in the extracellular nanoparticles by immunogold electron microscopy. The particles were absent in pilY1 knockout mutants which points to a functional association between PilY1* and the make-up of these structures. The PilY1 immunoprecipitates of the concentrated > 50 000 molecular weight extracellular fraction contained numerous hydrophobic compounds including HAQs. As these compounds have a molecular weight below 1000, they coprecipitated as polydisperse aggregates and thus most likely reflect the secondary metabolite components of the PilY1*-positive nanoparticles. Pyocyanin was completely and PQS was largely lacking from the particles. The signal molecule PQS is known to be extruded by vesicles (Mashburn and Whiteley, 2005), but the other major HAQs are translocated by other means across the bacterial membranes. Hence we conclude that PilY1 and/or PilY1* is involved in the discharge of these secondary metabolites. The transport was accomplished by both the piliated PAO1 and the non-piliated pilQ-defective TBCF10839 strains, suggesting that this function does not require a properly elongating T4P. Haematoxilin-eosin staining; original magnification ¥100, scale bars 100 mm. TBCF10839 wild type (A) induces a profound purulent pneumonia with numerous foci of inflammation (infiltration of neutrophils and alveolar macrophages) up to tissue necrosis. In contrast the lungs of both isogenic mutants (B, C) show a moderate purulent alveolar pneumonia (neutrophils and alveolar macrophages in the alveoli but not within the bronchi). (D) Normal lung parenchyma after PBS application. The in trans complemented mutants were not investigated because the strains lose their episomal plasmids in murine lungs over time.
PilY1 is transported to the periplasm in a Secdependent manner (Goldberg et al., 2008) in accordance with their presence in the membrane fraction of a cell lysate. To be released into the extracellular medium, a small portion of the PilY1 pool may be carried by the T4P all the way to the external medium or may be expelled by the brief elongation of the type II pseudopilus as it is the case of type II-dependent exoproteins. As the nanoparticles are present in the extracellular medium of piliated and non-piliated P. aeruginosa (Fig. 7), the latter rather than the former mechanism should account for the release of PilY1* and secondary metabolites into the extracellular space. Although we have yet no experimental evidence for this, it is tempting to assume that prior to or during this process PilY1 is cleaved to the extracellular isoform PilY1*.
Transposon insertions in TBCF10839 pilW and pilY1 caused similar phenotypes. The insertions may have affected the expression of the individual genes within the fimU-pilVWXY1Y2E operon to differential extent, but with respect to PilY1 protein the pilW and pilY1 mutants caused the same phenotype as judged by immunoblot analysis (Fig. 1). Both mutants were PilY1 knockouts. Likewise, macroscopic phenotypes such as the autolytic morphotype (Fig. 6), pyocyanin and HAQ deficiencies (Fig. 4) and persistence in (Fig. S9) and inflammation of murine airways (Fig. 10) were indistinguishable between the examined pilW and pilY1 mutants. Subtle differences were visualized in the electron micrographs (Fig. 7). Some extracellular PilY1* was seen in the pilW mutant at far below wild-type levels (Fig. 7J).
PA14 pilY1 and TBCF10839 pilY1 transposon mutants were growing to higher cell densities than their parental strains. PA14 mutants appeared colourless, but their HAQ and pyocyanin deficiencies at 37°C were less pronounced (Fig. S6) than those of the TBCF10839 mutants (Fig. 4). Interestingly, the PA14 mutants showed autolysis at 25°C, but not 37°C, whereas the TBCF10839 mutants showed autolysis at 37°C, but not at 42°C. This temperature dependence points to a delicate interplay between growth and secretion of secondary metabolites that is modulated by the genetic background with PilY1 being a major modifier to prevent cell lysis. According to our data PilY1 seems to be involved in the control of both cell density (Fig. 3) and secretion (Figs 3-5, S6, S7). Cell density is monitored by the QS network (Juhas et al., 2004). AHL levels were normal in the pilY1 mutants (Fig. S5) and thus signalling via the las and rhl networks probably remained unaffected in PilY1-deficient cells. The levels of PQS and pyocyanin, however, were lower in the mutants so that the PQS and PYO regulons operating in the late exponential and early stationary phase (Dietrich et al., 2006) could principally be affected in the pilY1 mutants. The transcriptome, however, provided no evidence for any direct or indirect perturbation of the QS regulons in the TBCF10839 pilY1 transposon mutant (Table S1). Hence, the inactivation of pilY1 exerts its effects at the posttranscriptional level. The experimental data point to direct roles of PilY1 at the interface between cell and environment. Besides the assistance in secretion and pilus biogenesis, the annotated role of PilY1 as an adhesin may be materialized by the control of growth during stationary phase. The cell-associated PilY1 and the extracellular nanoparticle-associated PilY1* could scan cell density by scoring cell-cell, cell-matrix or matrix-matrix contacts mediated by PilY1-PilY1, PilY1-PilY1* or PilY1*-PilY1* interactions respectively. These contacts between PilY1/PilY1* molecules could confine the density of the stationary bacterial population. This scenario is similar to the dual roles of numerous proteins involved in territorial stability in multicellular organisms that either exist in the membrane or are processed by proteolytic cleavage and then released to become part of the extracellular matrix. The inactivation of PilY1 will lead to the observed stronger growth in stationary phase and finally autolysis at high cell densities, but on the other hand it may facilitate persistence in hostile habitats like the murine lung or lysosomal compartments of a PMN.
In the latter case, the intracellular retention of HAQs like PQS (Häussler and Becker, 2008) may help to counterbalance the oxidative stress exerted by the neutrophil. When the phagocytosed mutant cells are delivered to the phagolysosome, the lysosomal enzymes will lyse the bacterial cells. HAQs are released into the lysosomal compartment. The noxious bacterial compounds will damage the PMN and thereby impede bacterial cell killing. Consequently PilY1-defective P. aeruginosa cells will survive better during a standardized PMN phagocytosis assay in vitro. On the other hand, if the toxic effect is exerted by viable bacteria via the release of pyocyanin, the eukaryotic host will be less harmed by an attack of pyocyanin-deficient PilY1 mutants. An example is the C. elegans fast-killing infection model where the secretion of pyocyanin by viable bacteria is instrumental for virulence ) (see Fig. 8).
In summary, P. aeruginosa PilY1 has a role in the control of cell density in the bacterial population and assists in pilus biogenesis, twitching motility and the release of secondary metabolites into the extracellular space. PilY1 shows substantial amino acid sequence diversity particularly in the N-terminal regions (Fig. S11). In conjunction with divergent genetic backgrounds, this may modulate phenotype and fitness traits of clones in the P. aeruginosa population (Wiehlmann et al., 2007a).
Inactivation of pilY1 in the CF isolate TBCF10839 rendered the bacterial cells more resistant to killing by PMNs and improved their persistence in murine airways. Future studies should reveal whether pilY1 or the other genes in the operon are targets for mutagenesis in natural infections with P. aeruginosa and may confer an advantage or disadvantage in fitness to colonize and to persist in animate habitats and to breach epithelial barriers.

Strains, plasmids and growth conditions
Plasmids, strains and culture conditions are listed in Table S2. The P. aeruginosa PA14 mutants ID 25563 and 25263 with transposon insertions at different positions of the pilY1 gene were recovered from the PA14 transposon library . The P. aeruginosa TBCF10839 STM transposon library was constructed with the plasposon pTnModOGm (Dennis and Zylstra, 1998) carrying variable signature tags as described previously (Wiehlmann et al., 2007b;Rakhimova et al., 2008). P. aeruginosa or Escherichia coli strains were routinely grown overnight as shaken cultures (230 r.p.m.) in LB broth medium at 37°C. Five millilitre LB broth was inoculated with a toothpick of frozen bacterial stock solution and incubated for 16-48 h. Recombinant E. coli DH5a strains transfected with pME6010 (Heeb et al., 2000) were growing in the presence of 50 mg ml -1 tetracycline and recombinant P. aeruginosa were cultured in the presence of 200 mg ml -1 tetracycline. P. aeruginosa PAO1 was transduced with phage F116 as follows (Maillard et al., 1995;Byrne and Kropinski, 2005): Phage F116 and 200 ml of an overnight culture of P. aeruginosa PAO1 were inoculated into 20 ml LB broth in a 100 ml flask and incubated for 2 h at 37°C with shaking (200 r.p.m.). After the addition of 40 mg mitomycin, the incubation was continued for another 30 min. The cells were pelleted by centrifugation at 4°C. The pellet was dissolved in 20 ml fresh LB broth and incubated for a further 2 h (37°C, 200 r.p.m.). Cells were pelleted by centrifugation at 4°C, and the supernatant was filtrated through a sterile 0.2 mm filter to harvest the phages. For electron microscopy, 10 ml each of this phage stock solution was dropped on single PAO1 colonies that had been grown overnight on agar plates.

Colony morphology
Colony morphology was visualized on Congo red (40 mg ml -1 ) agar plates (Freeman et al., 1989). Three microlitre P. aeruginosa cultures from late stationary phase were inoculated onto the plate and incubated at 25°C or 37°C for 24-48 h. Autolysis was documented after 24-48 h.

Twitching motility assay
Twitching motility was assayed according to the protocol by . Briefly, bacteria from liquid culture were subsurfacely inoculated on LB agar allowing bacterial locomotion at 37°C. After 24 h the twitching zone on the Petri dish was visualized by Coomassie staining.

DNA preparation
Genomic DNA from P. aeruginosa was isolated according to a protocol optimized for Gram-negative bacteria (Chen and Kuo, 1993).

Complementation of pilW and pilY1 transposon mutants
DNA fragments containing the pilW (PA4552) or pilY1 (PA4554) genes were amplified from the genomic DNA of P. aeruginosa TBCF10839 and ligated into the BglII-and EcoRI-site of the shuttle vector pME6010 (Heeb et al., 2000). Introduction of the plasmid construct into strains TBCF10839 PA4554::Tn5 or TBCF10839 PA4552::Tn5 was carried out by electroporation. Revertants were selected on LB agar containing 200 mg ml -1 tetracycline. Sequencing of the PCR products was performed by Qiagen (Hilden).

Fractionation of extracellular, intracellular and surface proteins
Whole cell lysates of P. aeruginosa were prepared by resuspending the bacterial pellet of a 2 ml culture in SDS-PAGE loading buffer followed by denaturation at 97°C for 4 min.
The extracellular protein fraction was prepared by two different procedures. Either the bacterial cell free supernatant of a 2 ml culture was precipitated by 70% EtOH or 10% (NH4)2SO4 at -21°C or 4°C respectively (Nilsson et al., 2000;Riedel et al., 2006). The protein pellets were collected by centrifugation (20 000 g, 30 min) and washed once with ice-cold ethanol. Air-dried pellets were resuspended in SDS-PAGE loading buffer for further immunoblot analysis. Alternatively, P. aeruginosa strains were cultured in 0.4 l tryptone/ NaCl broth at 37°C with 150 r.p.m. until early stationary phase (OD578~1.5-1.7). The bacteria were repetitively pelleted (first, 6000 g, 10 min, 4°C; then three times 8000 g, 15 min, 4°C). The supernatant was collected and then its macromolecular contents were concentrated by about 100fold by centrifugation (2000 g) through a membrane filter (Vivaspin 20, cut-off 50 kDa, Sartorius). Aliquots of the retentate were analysed for its contents of HAQ by TLC (see below) or of PilY1 by immunoprecipitation. Ten microlitre affinity-purified anti-PilY1 rabbit polyclonal antiserum, 25 ml protein A agarose (Santa Cruz Biotechnologies) and 10 ml protein G agarose (Santa Cruz Biotechnologies) were sequentially added to 500 ml retentate and incubated overnight at 4°C. In parallel, further 500 ml aliquots of retentate were processed with either the pre-immune serum of the anti-PilY1 antiserum (1:1300 dilution) or beads only (negative controls). After centrifugation (5 min, 1000 g, 4°C) the beads were washed with PBS. After addition of 18 ml 3¥ Laemmli buffer (7% SDS (w/v)), the bead solution was incubated for 5 min at 95°C and then applied to 6% SDS-PAGE. The gel-separated immunoprecipitated proteins were then probed for anti-PilY1-immunoreactive signals by immunoblot as described below.

PilY1 immunoblot analysis
Equal amounts of proteins from whole cell lysates or extracellular protein supernatant precipitates were separated on 6% SDS-PAGE and blotted onto nitrocellulose membrane. Detection of PilY1 with an affinity-purified anti-PilY1 rabbit polyclonal antiserum (1:1000) that had been raised against linear epitopes of PilY1 in PAO1 and in TB (pos. 338-352 CLPDGKSYSSQTPYRD) (Eurogentec) was carried out using the ECL Western Blot detection kit according to the manufacturer's protocol (Amersham). Pre-immune serum (1:500) did not show any immunoreactive signal against 100 ng peptide epitope or less. The sensitivity threshold of the used antiserum to the pilY1 peptide was 1 ng (Fig. 1).

Immunogold electron microscopy
Bacteria were grown on LB agar plates. After 7.5 h microcolonies were collected on grids with carbon-coated formavar films. For negative staining the grids were floated on water and then on 1% aqueous uranyl acetate (Merck, Darmstadt, Germany) for 1 min respectively. For immunostaining the blocking reaction was performed in 3% (w/v) bovine serum albumin in phosphate buffered saline (PBS). The incubation on ice started with an affinity-purified rabbit antibody against PilY1-peptide diluted 1:30 in 3% bovine serum albumin in PBS for 40 min. After washing three times with PBS the grids were incubated with 12 nm colloidal gold-affinipure goat antirabbit IgG (Jackson Immunoresearch, West Grove, PA, USA) for 40 min followed by the negative staining as described before. The grids were examined in a Philips EM 301 electron microscope. The electron micrographs were selected, digitized and processed using Adobe photoshop 6.0.

mRNA expression analyses
Whole genome mRNA expression profiling: RNA extraction, preparation of samples, hybridization of the P. aeruginosa PAO1 microarray and evaluation of data were performed according to the recommendations of the manufacturer and in-house protocols (von Götz et al., 2004;Salunkhe et al., 2005a,b). For semiquatitative RT/PCR assays, RNA was extracted from 6 ¥ 10 8 cfu of bacteria grown to late exponential phase (OD578~2.5) using the Qiagen RNeasy Kit according to the manufacturer's instructions. DNA contaminations were removed by treatment with RNAse-free DNAse. cDNA was synthesized from 500 ng RNA by reverse transcription with random primers (AffinityScript QPCR cDNA Synthesis Kit, Stratagene). For PCR kinetics, a 1/400 aliquot of the resulting cDNA was amplified in a 50 ml reaction mixture (5 ml 10¥ reaction buffer (Eurogentec), 3.3 ml 25 mM MgCl2, 1 ml DMSO, 10 ml primer solution (5 mM each), 3 ml dNTPs (2 mM each). 1 U Goldstar-DNA-Polymerase (Eurogentec)). Primers are listed in Table S2. Aliquots were withdrawn every two cycles, separated by electrophoresis and stained with ethidium bromide. The relative amounts Ni and Nj of the template cDNA sequences i and j in the reaction mixture were determined from the titration for the first reaction cycle n when the PCR products became visible by ethidium bromide fluorescence during late exponential phase of PCR according to the equation whereby the efficiency R of the used thermocycler during the exponential phase of PCR was determined to be R = 0.78 Ϯ 0.02 for PCR products of 100-800 bp in length within the interval of reaction cycles 10 < n < 35 (Bremer et al., 1992).

Pyocyanin and pyoverdine secretion
Pseudomonas aeruginosa strains were grown with shaking at 37°C in King A or King B medium (King et al., 1954) that preferentially stimulate the production of either pyocyanin (King A) or pyoverdine (King B). Cells were sedimented by centrifugation at 7000 g for 10 min. The absorption of the supernatant was measured spectrophotometrically at 695 nm (pyocyanin) or at 380 nm (pyoverdine).

Plate assays of siderophore and protease secretion
Siderophore secretion was assessed by an orange halo around colonies grown for 24 h at 37°C on chromazurol plates as described by Schwyn and Neilands (1987). Secretion of casein-degrading proteases was examined by growing the analysed P. aeruginosa strains on M9 agar plates supplemented with 0.8% (w/v) casein (Cowell et al., 2003).

Detection and characterization of AHLs
Bacteria were grown in LB broth to late exponential phase (OD578~2.5). The AHL molecules were extracted with dichlormethane from spent culture supernatants of the strains. The extract was dried and redissolved in 0.2 ml ethyl acetate. An aliquot of 10 ml was separated by TLC and visualized by overlaying the TLC plates with soft agar seeded with the luxAB based AHL biosensor E. coli MT102 (pSB403) (detects long-chain and short-chain AHLs; Winson et al., 1998), biosensor Chromobacterium violaceum CV026 (high sensitivity for short-chain AHLs; McClean et al., 1997) or biosensor Pseudomonas putida F117 pKR-C12 (high sensitivity for long-chain AHLs; Steidle et al., 2001). Bioluminescent spots indicating AHL were detected by exposure of a X-ray film. On the basis of their mobilities (Rf-values) and by including appropriate reference compounds a tentative identification of AHLs present in the culture extracts was possible.

HAQ detection
Pseudomonas aeruginosa strains were grown in a 50 ml flask in 10 ml LB broth with constant shaking at 37°C for 15 h or up to an optical density of OD578 = 2.5. One millilitre of the culture was extracted with 2 ml dichlormethane by vigorous shaking and the liquid phases were separated by centrifugation at 5000 g for 10 min. One millilitre of the organic phase containing HAQs and pyocyanin as major components was dried by evaporation. The pellet was resuspended in 50 ml methanol. Eight microlitres thereof was separated on Silica Gel 60 F254 TLC plates (that had been pre-soaked for 30 min in 5% (w/v) aqueous KH2PO4 solution and then dried for 60 min at 80-90°C prior to use) with 5% methanol/95% dichlormethane as the mobile phase. Fluorescent spots were visualized under UV light and photographed. PQS was synthesized from HHQ by the procedure described by Pesci et al. (1999) and used as standard.

Phagocytosis assay in granulocytes
Bacteria were grown in LB medium at 37°C for 16 h. Granulocytes were prepared from the blood of healthy donors.
Approximately 10 ml of fresh blood (with 100 I.U. heparin) was used for each experiment. The blood was mixed with 5 ml 10% hydroxyethyl starch and most of the erythrocytes were separated by sedimentation (45 min, room temperature). The granulocytes were then separated by centrifugation (3000 g, 15 min) using a lymphocyte separation medium (Lymphoprep, Axis-Shield, Oslo, Norway). The cell pellet was dissolved in 0.2 ml distilled water for 10 s to lyse residual erythrocytes. After mixing with 2 ml 1.1¥ PBS the cells were pelleted again by centrifugation. The sedimented granulocytes were resuspended in 2 ml RPMI1640 medium and stored on ice. Granulocytes were counted in a Neubauer chamber. 0.2 ml AB-blood serum and a 10-fold excess of bacteria (determined by photometry 0.6 OD578 = 10 9 cfu ml -1 ) were added to the granulocyte suspension. This assay was incubated for 120 min (37°C, 200 r.p.m.). After incubation, the granulocytes with the internalized bacteria were separated from extracellular bacteria by centrifugation at 800 g for 10 min, resuspension in 0.3 ml RPMI1640, filtration (nitrocellulose, pore size 2 mm, Sartorius), and washing with PBS. The filters with the adhered granulocytes were transferred into distilled water for lysis and mixed vigorously for 5 min. The bacteria were transferred to new tubes and centrifuged (4000 g, 10 min). The supernatant was discarded and the pelleted bacteria were plated on LB agar. Survival of different P. aeruginosa strains was determined by cfu counts.

Screening of pools of STM mutants for survival in the PMN phagocytosis assay or in the acute airway infection model in mice
Transposon mutants with different signature tags were separately grown in LB (37°C) overnight and pooled directly before the experiment. In case of phagocytosis assays, aliquots of a 10-fold excess of bacteria to granulocytes were tested in parallel in phagocytosis assays in the absence (control I) or presence of PMNs (experiment type I). After 2 h of incubation, extracellular bacteria were removed by filtration through 5 mm nitrocellulose filters. Intracellular bacteria were released from the retained PMNs on the filter by the addition of 3 ml distilled water and vortexing for 5 min. The bacterial suspension was pelleted (4000 g, 10 min), resuspended in 0.1 ml PBS and plated on LB agar. Control I bacterial suspension was plated in parallel. Bacteria were collected after overnight incubation at 37°C.
In the mouse infection experiments, 100 ml of bacterial suspension was cultivated on LB agar or liquid for 48 h at 37°C (control II). Thirty microlitres (7.5 ¥ 10 6 cfu) was used for the intratracheal mice infection (experiment type II). Inoculation of bacteria and maintenance of mice are described in Murine infection experiments. After 48 h of infection, mice were sacrificed and the lungs were homogenized. Bacteria from the homogenised lungs were recovered on LB and LB agar at 37°C overnight. In parallel, bacteria from the control plates were collected and incubated on LB and LB agar at 37°C overnight in the same incubator.
Genomic DNA was prepared from both control and experiment and a PCR was performed to amplify the signature tags [primers P1 and P2 (Table S2; 35 cycles; 20 s, 58°C; 20 s, 72°C; 30 s, 94°C; 10 s ramp between each step)]. The PCR products were digested for 16 h with HindIII and the specific 40 bp sequence tags were separated from the common flanking 20 bp sequences by PAGE [10% gel (19+1 acrylamide/ bis-acrylamide) in TBE buffer]. The 40 bp sequence tags were cut out from the gel and purified (QIAGEN). The 40 bp sequences were labelled with DIG-ddUTP using a terminal transferase (Roche) and hybridized onto dot blots. The dot blots were prepared as follows: 80 ml of the PCR products amplified from each single mutant was mixed with 40 ml of 3 M NaOH and 280 ml of TE buffer and denatured at 65°C for 30 min. After transfer to ice, 400 ml of 2 M CH3COONH4 was added and after short incubation, an aliquot of 95 ml of the DNA solution was applied to a Minifold-DOT-vacuum-blot device (Schleicher and Schuell) to immobilize the DNA on a Hybond N + membrane. Dried and cross-linked membranes were pre-hybridized with 10 ml hybridization buffer [0.5 M NaH2PO4 ¥ 2H2O, 7% SDS, 1 mM EDTA, 0.5% blocking reagent (Roche), pH 7.2] for 2 h at 58°C and afterwards hybridized with 40 ml of hybridization buffer containing denatured DIG-labelled probes for 16-24 h at 58°C in a hybridization oven. Non-specifically bound probe solution was removed by several washing steps and hybridization signals were detected by washing the membrane with anti-DIGalkaline phosphatase (Roche) and CDP Star. Chemoluminescence signals were detected on X-ray films and quantified by PCBAS, version 2.09f (raytest Isotopenmeßgeräte GmbH). Signal intensity of each dot was compared with that of the corresponding signal of the probe prepared from pooled bacteria grown in parallel on LB agar without in vivo selection. Hybridization signals out of the 95% confidence interval of the mean were interpreted to be significantly different from the average signal.

Sequencing of transposon flanking genes
Transposon-flanking sequences were amplified by the Y-linker method (Kwon and Ricke, 2000). Oligonucleotide sequences (Y-linker, Tn5MOD) are listed in Table S2. The Y-linker was prepared by the annealing of two oligonucleotides, Y1 and Y2. Genomic DNA of the mutants in aliquots of 1 mg was cut with 5 U NleIII or SphI (New England Biolabs) for 3 h and ligated to the phosphorylated and annealed Y-linker with T4-DNA ligase at 25°C for 2 h. The resulting product was used as a template in a PCR with the Y-and pTnMOD-specific primers. The resulting PCR products were purified by agarose gel electrophoresis and extracted using a Qiagen Gel extraction Kit. Sequencing was done by QIAGEN using the pTnMOD-specific primer. The raw sequences were analysed by BLASTn search against the sequences of the predicted genes as well as the complete genome sequence of P. aeruginosa PAO1 (Stover et al., 2000) or PA14 . 1.7 (w/v)% agar) and incubated overnight at 37°C. Fifty synchronized C. elegans nematodes (stage N4) were added onto the plates and the number of dead worms was counted after 1, 3 and 6 h. Plates inoculated with E. coli OP50 served as negative controls.

Murine infection experiments
Acute infection with planktonic P. aeruginosa Bacteria were grown in LB broth overnight at 37°C (230 r.p.m.) to stationary phase. The bacteria were pelleted by centrifugation (5000 g, 10 min), washed twice with sterile PBS and the optical density of the bacterial suspension was adjusted by spectrophotometry at 578 nm. The intended number of cfu was extrapolated from a standard growth curve, and appropriate dilutions with sterile PBS were made to prepare the inoculum for the mice. To verify the correct dilution, an aliquot was serially diluted on LB agar plates. Ten-to twelve-week-old female mice of the inbred strain C3H/HeN (Charles River, Sulzfeld, Germany) were inoculated with 30 ml of the bacterial suspension via view controlled intratracheal instillation. This non-invasive application technique via catheter allows controlled delivery of the bacteria to the lungs (Munder et al., 2002). During the experiments mice were maintained in microisolator cages with filter top lids at 21 Ϯ 2°C, 50 Ϯ 5% humidity and 12 h light-dark-cycle. They were supplied with autoclaved, acidulated water and fed ad libitum with autoclaved standard diet. Prior to the start of the experiments animals were acclimatized for at least seven days. In case of 14-day infection experiments, the weight and rectal temperature of the mice were measured daily and their body condition was determined using a self-developed score (Munder et al., 2005). Murine behaviour was scored for the parameters vocalization, piloerection, attitude, locomotion, breathing, curiosity, nasal secretion, grooming and dehydration.
All animal procedures were approved by the local District Governments and carried out according to the guidelines of the German law for the protection of animal life.

Murine inflammatory response to airway infection with alginate-embedded P. aeruginosa
Immobilization of bacteria in seaweed alginate beads. Immobilization of P. aeruginosa in seaweed alginate beads was performed as previously described (Pedersen et al., 1990). In brief, the bacteria were isolated from 100 ml of overnight culture grown in filtered oxbroth LB by centrifugation (12 000 g, 10 min, 4°C) and resuspension of the pellet in 5 ml sterile serum bouillon LB. One millilitre of bacterial suspension was mixed with 9 ml of sterile seaweed alginate (Protanal 10/60; Protan, Drammen, Norway). The mixture was forced with air through a channel into a solution of 0.1 M CaCl2 in 0.1 M Tris/HCl buffer (pH 7.0). The content of embedded bacteria in terms of cfu was estimated by serial dilutions and the challenge suspension of P. aeruginosa was adjusted to yield 8 ¥ 10 4 cfu ml -1 .
Murine airway infections. Ten-to twelve-week-old female C3H/HeN mice were anaesthetized by subcutaneous injection of 0.2 ml Hyp/Mid [2.5 ml mg -1 Hypnorm (Janssen, Birkerød, Denmark) and 1.25 mg ml -1 Midazolam (Roche, Basel, Switzerland)] in sterile water prior to challenge. The trachea was exposed and penetrated with an 18G needle. Intratracheal challenge was performed by installing 40 ml (8 ¥ 10 4 cfu ml -1 ) of alginate-embedded bacteria in the left lung~11 mm from the tracheal penetration site with a beadtipped curved needle (Moser et al., 1997). The incision was sutured and healed without any complications. The mice were sacrificed on day 4 by intraperitoneal injection of 0.05 ml 20% pentobarbital (KVL, Copenhagen, Denmark).
Blood leucocytes. The thoracic cavity was exposed and blood was collected from the heart in heparinized syringes and immediately put on ice.
Bronchoalveolar lavage. The trachea was exposed and canulated with a size 22G catheter (OPTIVA* 2, Johnson and Johnson Medical, Brussels, Belgium). A lavage was performed six times by repeated flushing with 1.5 ml ice-cold PBS and the BAL fluid was stored on ice.
Measurement of leukocytes from blood and BAL-fluid. The fraction of PMNs was estimated by adding 75 ml peripheral blood or 100 ml BAL-fluid to 4 ml cold 0.87% (w/v) NH4Cl buffer in order to lyse the erythrocytes. The cell pellet obtained after centrifugation was washed in PBS before adding the antibodies listed in Table S3. The cells were incubated on ice in the dark for 30 min and washed once in cold PBS. The pellet was resuspended in 200 ml before flow cytometry. The concentration of PMNs was measured by TruCount (340334, Becton Dickinson). Briefly, 200 ml BAL fluid or 50 ml blood was added to a TruCount Tube. Cells were fixed and the DNA was stained by adding 450 ml of 10% (v/v) FACS lysing solution (349202, Becton Dickinson) and 100 mg ml -1 propidium iodide (P-4170, Sigma) in MilliQ water. The samples were kept on ice in the dark for at least 10 min before analysing by flow cytometry.
The endobronchial content of PMNs was estimated by multiplying the concentration of PMNs with the volume of recovered BAL fluid.
Flow cytometry. The samples were analysed using a FACSort (Becton Dickinson) equipped with a 15 mW argonion laser tuned at 488 nm for excitation and a red diode laser for excitation at 635 nm. Light scatter and logarithmically amplified fluorescence parameters from at least 10 000 events, when possible, were recorded in list mode after gating on light scatter to avoid debris, cell aggregates, and bacteria. The instrument was calibrated using Calibrite (Becton Dickinson). Leucocytes were identified according to their morphology, by their expression of CD45, and their content of DNA.
of the International German-Danish Research Training Group 'Pseudomonas: Pathogenicity and Biotechnology' (IRTG 653). SH received a Georg-Christoph-Lichtenberg stipend from the government of Lower Saxony and was a member of MHH′s international PhD Program 'Infection Biology'.