Blackwell Science, LtdOxford, UKMMIMolecular Microbiology1365-2958Blackwell Publishing Ltd, 200348615111524Original ArticleM. Klausen et al.

Roles of bacterial motility in the formation of the at P. aeruginosa biolm

Molecular Microbiology (2003) 48(6), 15111524

doi:10.1046/j.1365-2958.2003.03525.x

Biolm formation by Pseudomonas aeruginosa wild type, agella and type IV pili mutants

Mikkel Klausen,1 Arne Heydorn,2 Paula Ragas,1 Lotte Lambertsen,1 Anders Aaes-Jrgensen,1 Sren Molin1 and Tim Tolker-Nielsen1* 1 Molecular Microbial Ecology Group, BioCentrum-DTU, Building 301, Technical University of Denmark, DK-2800 Lyngby, Denmark. 2 Department of Pharmacology, Copenhagen University, Panum Institute, DK-2200 Copenhagen N, Denmark. Summary Biolm formation by Gfp-tagged Pseudomonas aeruginosa PAO1 wild type, agella and type IV pili mutants in ow chambers irrigated with citrate minimal medium was characterized by the use of confocal laser scanning microscopy and COMSTAT image analysis. Flagella and type IV pili were not necessary for P. aeruginosa initial attachment or biolm formation, but the cell appendages had roles in biolm development, as wild type, agella and type IV pili mutants formed biolms with different structures. Dynamics and selection during biolm formation were investigated by tagging the wild type and agella/type IV mutants with Yfp and Cfp and performing time-lapse confocal laser scanning microscopy in mixed colour biolms. The initial microcolony formation occurred by clonal growth, after which wild-type P. aeruginosa bacteria spread over the substratum by means of twitching motility. The wild-type biolms were dynamic compositions with extensive motility, competition and selection occurring during development. Bacterial migration prevented the formation of larger microcolonial structures in the wild-type biolms. The results are discussed in relation to the current model for P. aeruginosa biolm development. Introduction Bacterial life includes stages where the cells are associated and form a biolm on a surface (e.g. Costerton et al., 1995). The formation of these surface communities and

Accepted 18 February, 2003. *For correspondence. E-mail ttn@biocentrum.dtu.dk; Tel. (+45) 45 25 27 93; Fax (+45) 45 88 73 28.

their inherent resistance to antimicrobial agents are the cause of many persistent and chronic infections (Costerton et al., 1999). Microscopic analysis has indicated that biolm formation occurs in a sequential process of (i) transport of microbes to a surface; (ii) initial attachment; (iii) formation of microcolonies; and (iv) biolm maturation (e.g. Tolker-Nielsen et al., 2000; Sauer et al., 2002). The biolm formed by Pseudomonas aeruginosa under owthrough conditions was found to be heterogeneous with mushroom-shaped microcolonies in studies in which glucose was used as the carbon source (e.g. Stewart et al., 1993; Davies et al., 1998), and at, uniform and densely packed in studies where citrate was used as the carbon source (e.g. Heydorn et al., 2000; 2002), suggesting that P. aeruginosa biolm development is dependent on the carbon source. The use of a simple high-throughput screening assay for biolm development has greatly facilitated the analysis of the genetic elements involved in biolm formation. In this assay, bacteria grow under static conditions in the wells of microtitre plates and may form a biolm on the abiotic well surface if they possess the necessary genetic elements. Quantication of the amount of biolm formed in each well is done after removal of planktonic cells and appropriate staining of the surface-attached cells. Over the past few years, this assay has been used extensively to identify genes involved in the initial phases of biolm formation in a number of bacteria, including the opportunistic pathogens Staphylococcus epidermidis (Heilmann et al., 1996) and Pseudomonas aeruginosa (OToole and Kolter, 1998a), which have become the model organisms for Gram-positive biolm and Proteobacterial biolm. The use of ow chamber technology, uorescent reporter genes and confocal laser scanning microscopy (CLSM) has enabled very detailed studies of specic and general interactions between the members of biolm communities (e.g. Mller et al., 1998; Nielsen et al., 2000), and has permitted non-destructive studies of the dynamics and developmental steps occurring during biolm formation (e.g. Tolker-Nielsen et al., 2000). In this set-up, the bacteria grow under hydrodynamic conditions and form biolms on a glass surface. Over the past few years, this set-up has been used to study biolm formation by a number of bacteria, including P. aeruginosa (e.g. Heydorn et al., 2002; Whitchurch et al., 2002).

2003 Blackwell Publishing Ltd

1512 M. Klausen et al. Pseudomonas aeruginosa is able to swim in liquid by means of agella and to move on surfaces by means of type IV pili (Semmler et al., 1999). The surface-associated so-called twitching motility is powered by extension and retraction of type IV pili (Skerker and Berg, 2001). Using the microtitre plate assay, OToole and Kolter (1998a) showed that agella or agellum-driven motility is required for biolm formation by P. aeruginosa PA14, and that type IV pili are required for biolm and microcolony formation by this organism. It was speculated that swimming motility might enable the bacteria to overcome repulsive forces at the surfacewater interface so that they reach the surface, and that the microcolonies may be formed by twitching motility-driven cell aggregation. Using the ow chamber set-up, Heydorn et al. (2002) showed that a twitching motility-decient mutant derivative of P. aeruginosa PAO1 is capable of both biolm and microcolony formation. In fact, under the conditions used by Heydorn et al. (2002), the wild-type P. aeruginosa PAO1 formed a uniform at biolm without microcolonies, whereas the twitching motility mutant formed a biolm with numerous microcolonies. These apparently contrasting observations suggested that biolm studies are sensitive to strain and experimental differences. The effect of experimental conditions on the factors required for biolm formation was shown directly for Pseudomonas uorescens. Non-motile P. uorescens mutants were decient in biolm formation in the wells of microtitre plates when grown in minimal medium with glucose and casamino acids but, when the same mutants were grown in minimal medium with citrate, they did form biolms in the wells of the microtitre plates (OToole and Kolter, 1998b). Consequently, there may be many pathways involved in biolm development in these organisms. The different medium conditions that promote biolm formation, and the different subsets of genes required under each condition, may indicate that various environmental niches are colonized by the bacteria through biolm formation via different pathways. Despite the evidence that P. aeruginosa biolm development may be dependent on the carbon source used to support growth, and the nding that type IV pili are not always necessary for P. aeruginosa biolm or microcolony formation, there is currently only one generally accepted model for P. aeruginosa biolm development. According to this model, agella mediate transport of the bacteria to the surface, type IV pili-driven motility along the surface leads to cellular aggregation and microcolony formation, and the subsequent formation of larger sessile mushroom-shaped multicellular structures occurs via a maturation process that requires cell-to-cell signalling (e.g. Costerton et al., 1999; OToole et al., 2000; Stoodley et al., 2002). In the present study, we show that, dependent on the type of carbon source, P. aeruginosa may form a biolm with mushroom-shaped structures or a at biolm, and we carry out a characterization of the formation of the at P. aeruginosa biolm, which leads to an alternative biolm model. Results For the analysis with laser scanning microscopy, the bacteria needed to be uorescently tagged. Fluorescent tagging of the strains was done with insertion of gfp, cfp or yfp at a dened genetically neutral chromosomal locus using mini-Tn7 constructs. Initially, we tested the effect of the carbon source on the biolm structure formed by P. aeruginosa PAO1 in our ow chamber set-up. The biolms formed by Gfp-tagged P. aeruginosa PAO1 in ow chambers irrigated with minimal glucose medium were heterogeneous with mushroomshaped microcolonies (Fig. 1A), whereas the biolms formed in ow chambers irrigated with minimal citrate medium, minimal casamino acids medium or minimal benzoate medium were at and densely packed (Fig. 1BD). COMSTAT image analysis on 18 CLSM images acquired at random locations in two glucose-grown, two casamino acids-grown, two citrate-grown and two benzoate-grown P. aeruginosa biolms showed that the structural differences were highly signicant (data not shown). Although the P. aeruginosa biolm with mushroom-shaped multicellular structures has become widely accepted as the universal biolm model (e.g. Costerton et al., 1999; OToole et al., 2000; Stoodley et al., 2002), there is clearly also a need to understand the biology of the at biolm that P. aeruginosa forms under some conditions. For example, it is not known what type of P. aeruginosa biolm causes various persistent infections. The present work is an investigation of the development of the at P. aeruginosa biolm in ow chambers irrigated with citrate minimal medium. A similar analysis of the development of the heterogeneous P. aeruginosa biolm in ow chambers irrigated with glucose minimal medium is ongoing in our laboratory. Characterization of P. aeruginosa wild-type biolm development and dynamics In order to investigate P. aeruginosa biolm formation and dynamics, we inoculated a ow chamber with a 1:1 mixture of Cfp-tagged and Yfp-tagged PAO1 wild-type cells, irrigated the ow chamber with citrate minimal medium and followed the biolm development with time-lapse CLSM. As shown in Fig. 2AD, microcolonies consisting of either Cfp-tagged or Yfp-tagged cells formed initially by clonal growth. After a period of initial microcolony formation by cell proliferation at xed locations, the bacteria spread out on the substratum (Fig. 2E) and, subsequently, a at biolm covering the entire substratum was formed (Fig. 2F).
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1513

Fig. 1. CLSM micrographs of 5-day-old P. aeruginosa PAO1 biolms grown on glucose minimal medium (A), citrate minimal medium (B), casamino acids minimal medium (C) or benzoate minimal medium (D). The central pictures show top-down simulated uorescence projections, and the anking pictures show vertical sections. Bars, 20 mm.

Fig. 2. Time-lapse CLSM in a colour-coded P. aeruginosa PAO1 wild-type biolm. The biolm was initiated with a 1:1 mixture of yellow uorescent and cyan uorescent P. aeruginosa wild-type bacteria and grown on citrate minimal medium. The structural development in the biolm was followed by time-lapse CLSM. The shown CLSM side-view projections were acquired after 0 (A), 7 (B), 11 (C), 15 (D), 19 (E) and 23 (F) hours of biolm development. The complete CLSM time-lapse movie can be seen at http://www.im.dtu.dk/mol-mic-avi. Box side 230 mm, box height 44 mm. 2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

1514 M. Klausen et al. with citrate minimal medium was tested under owthrough conditions and under static conditions. As shown in Fig. 3A, the wild type and motility mutants adhered equally well in ow chambers under hydrodynamic or static conditions, indicating that agella and type IV pili do not play a role in attachment of P. aeruginosa to a glass surface in ow chambers with minimal citrate medium. As it was shown previously that agella and type IV pili mutants of P. aeruginosa PA14 were decient in biolm formation under static conditions in the wells of microtitre plates with glucosecasamino acids minimal medium (OToole and Kolter, 1998a), we tested the ability of our P. aeruginosa PAO1 wild-type, DiM, DpilA and DpilADiM strains to form biolm in the wells of microtitre plates with glucosecasamino acids minimal medium and found that the DiM, DpilA and DpilADiM mutants were decient in biolm formation in comparison with the P. aeruginosa PAO1 wild type (Fig. 3B). There is therefore no reason to assume that P. aeruginosa PAO1 and PA14 use different pathways for biolm formation. In microtitre plates with citrate minimal medium, however, the P. aeruginosa PAO1 wild type, DiM, DpilA and DpilADiM mutants all formed biolms equally well (Fig. 3B), again suggesting that agella and type IV pili do not play a role in surface attachment of P. aeruginosa in citrate minimal medium. The increased biomass in the glucose casamino acids-grown wild-type biolm in comparison with the citrate-grown biolms (Fig. 3B) correlated with a higher yield in the glucosecasamino acids medium (data not shown). Comparative analysis of biolm architectures formed by P. aeruginosa wild type and motility mutants The wild type and motility mutants were grown as biolms in ow chambers irrigated with citrate minimal medium, and development was investigated by acquiring CLSM micrographs at days 1, 4 and 7 and subjecting the digital three-dimensional images to COMSTAT image analysis. As shown on the representative CLSM micrographs in Fig. 4, and from the quantitative data in Fig. 5, the wild type and motility mutants formed biolms with very different structures. The wild type formed a at carpet, the DiM mutant formed a hilly biolm, and the DpilA and DpilADiM mutants formed biolms with irregular and protruding structures. Figure 5 shows roughness and average thickness quantication data for the four strains at days 1, 4 and 7. Each spot represents values calculated from a single CLSM image stack. The graphs therefore give both an impression of development over time and the variation within biolm quantication data. The variation within strain is highest for the double mutant. A small variation within groups is dependent on structures small enough to be repeated within the microscope eld. This can explain
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Fig. 3. Attachment to glass surface and biolm formation in the wells of microtitre plates of P. aeruginosa wild type, DiM, DpilA and DiMDpilA. A. Attachment in ow chambers with citrate minimal medium under owthrough or static conditions. Average values were calculated from image analysis on 18 CLSM images acquired in two ow chamber channels. B. Microtitre plate biolm formation in minimal medium supplemented with glucose + casamino acids (CAA) or citrate. Average values were calculated from 28 wells.

Bacterial migration along a surface in a biolm may be type IV pili driven (OToole and Kolter, 1998a) or agellum driven (Pratt and Kolter, 1998; Watnick and Kolter, 1999). In order to study the role of type IV pili and agella in the formation of the at P. aeruginosa biolm, we derived the mutants DpilA, DiM and DpilADiM from P. aeruginosa PAO1 by allelic displacement. Comparative analysis of adhesion of P. aeruginosa wild type and motility mutants The ability of the wild type, DpilA, DiM and DpilADiM mutants to adhere to the glass surface in ow chambers

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1515

Fig. 4. CLSM side-view projections showing the spatial structure in P. aeruginosa wild-type, DiM, DpilA and DpilADiM biolms at days 1, 4 and 7.

the fairly high variation between images of the large DpilADiM mutant biolm structures. Comparative analysis of development and dynamics in biolms of P. aeruginosa wild type and motility mutants The differences in biolm structure found for the wild-type, DpilA, DiM and DpilADiM strains suggested that bacterial motility plays a role in biolm development. Presumably, the combination of growth and non-motility created the protruding structures in the mutant biolms, whereas the combination of growth and migration resulted in the at wild-type biolm. In order to investigate this proposed role of motility in P. aeruginosa biolm development further, we performed time-lapse CLSM in two-coloured wild-type, DpilA and DiM biolms, which had all been initiated with a 1:1 mixture of Cfp-tagged and Yfp-tagged
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

cells. Microcolonies consisting of either Cfp-tagged or Yfp-tagged cells initially formed in all three biolms (Fig. 6AC). A small number of cyan uorescent cells could sometimes be observed in yellow uorescent microcolonies and vice versa, but extensively mixed microcolonies were not observed. After the initial microcolony formation, the wild-type and DiM strains spread on the substratum, whereas the cyan or yellow uorescent DpilA microcolonies grew bigger at xed locations, suggesting that expansion on the substratum by the wildtype and DiM strains was type IV pili driven. In the 4-dayold wild-type and DiM biolms, regions with Cfp-tagged and Yfp-tagged cells totally mixed occurred often (Fig. 6D and E), indicating that extensive motility occurred during biolm formation. The 4-day-old DpilA biolm did not contain mixed areas (Fig. 6F), supporting further the proposal that the majority of the motility occurring during

1516 M. Klausen et al. substratum coverage reached by the DpilA and DpilADiM mutants was about 50%. Competition and selection in P. aeruginosa biolms The very dynamic nature of the P. aeruginosa biolms investigated here suggested that competition and selection could occur extensively in the biolms. Quantication of biomass accumulation showed that the DiM and DpilADiM biolms accumulated more biomass than the wild-type and DpilA biolms (Fig. 7B). As this indicated that the DiM mutant could have a selective advantage in a dynamic biolm, we investigated competition and selection in biolms that had been initiated with a 1:1 mixture of Yfp-tagged wild type and Cfp-tagged DiM mutant. Six CLSM images were acquired at six randomly chosen positions every 30 min for the rst 24 h of biolm growth, and biomass accumulation of the yellow uorescent wild type and the cyan uorescent DiM mutant was quantied by COMSTAT analysis. As shown in Fig. 8, immediately after inoculation, the number of wild-type cells and DiM cells was roughly equal and, thereafter, the DiM mutant rapidly outcompeted the wild type. Analysis of the absolute cell numbers indicated that a substantial fraction of the wild-type cells left the substratum during the rst 2 h of biolm growth, and that the wild-type cells then grew at the substratum at the same rate as the DiM mutant for about 10 h, after which a fraction of the produced wild-type cells continuously left the biolm in the rest of the analysed time period (data not shown). The opposite colour combination was also tested and gave the same results (data not shown). Discussion Pseudomonas aeruginosa has become the model organism for proteobacterial biolms, and much work has been done in order to identify consensus in biolm formation and to formulate a general biolm model (for recent reviews, see Costerton et al., 1999; OToole et al., 2000; Stoodley et al., 2002). The present study shows that P. aeruginosa biolm formation is dependent on the carbon source used to support growth. A at P. aeruginosa biolm was formed when citrate, benzoate or casamino acids was used as carbon source, and a heterogeneous P. aeruginosa biolm with mushroom-shaped multicellular structures was formed when glucose was used as the carbon source. The generally accepted model for P. aeruginosa biolm development is based largely on studies of the heterogeneous P. aeruginosa biolm (e.g. Costerton et al., 1999; OToole et al., 2000; Stoodley et al., 2002) and does not explain the formation of the at P. aeruginosa biolm. It seemed therefore that characterization of the developmental steps leading to the formation of the at P. aeruginosa biolm was needed, and therefore the
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Fig. 5. Scatter plots showing COMSTAT-calculated average thickness and roughness in P. aeruginosa wild-type ( ), DiM ( ), DpilA ( ) and DpilADiM ( ) biolms at days 1, 4 and 7.

wild-type and DiM biolm development was type IV pili driven. Twitching motility as a spreading mechanism was also suggested by quantitative analysis of substratum coverage in wild-type, DiM, DpilA and DpilADiM biolms. As shown in Fig. 7A, the wild type and DiM mutants quickly covered the entire substratum, whereas the maximum

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1517

Fig. 6. CLSM optical sections in 1-day-old (AC) and 4-day-old (DF) P. aeruginosa wild-type (A and D), DiM (B and E) and DpilA (C and F) biolms. All biolms were initiated with a 1:1 mixture of Cfp-tagged and Yfp-tagged bacteria. Bars, 20 mm.

present work was carried out. Our results lead to the alternative biolm development model, which is shown as a schematic in Fig. 9 and is discussed below. Adhesion Our studies with P. aeruginosa PAO1 wild-type, DiM, DpilA and DpilADiM strains suggested that agella and type IV pili are not necessary in the adhesion process in ow chambers or microtitre plate wells with citrate minimal medium. In microtitre plate wells with glucosecasamino acids medium, however, the DiM and DpilA mutants were decient in biolm formation in agreement with the ndings of OToole and Kolter (1998a) with P. aeruginosa PA14 derivatives. Defects in biolm formation in glucose casamino acids medium, but unhindered biolm formation in citrate minimal medium, was also reported for motility mutants of P. uorescens (OToole and Kolter, 1998b). Heydorn et al. (2002) found that a P. aeruginosa DpilHIJK mutant adhered ve times better than the wild type in ow chambers with citrate. The pilHIJK genes, however, encode chemotaxis-related genes (Darzins, 1994) and, although a DpilHIJK mutant is twitching motility decient (Darzins and Russell, 1997; Heydorn et al., 2002), it has not been shown that this mutant is non-piliated. In a study by DeKievit et al. (2001), adhesion of P. aeruginosa PAO1
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

was reported to be dependent on agella and type IV pili under static conditions, whereas type IV pili and decreased agellar motility did not affect biolm formation of P. aeruginosa PAO1 in ow chambers with citrate medium or glucose medium. Another study showed that a non-agellated mutant of the P. aeruginosa PA14 strain attached poorly in comparison with the wild type in a owthrough system with glutamic acid medium (Sauer et al., 2002). A critical factor for transport of bacteria to the substratum in owthrough systems is presumably the degree of turbulence near the substratum and, as the ow in different owthrough systems may be more or less laminar/turbulent near the substratum, comparisons of bacterial adhesion in different owthrough systems may be problematic. Initial microcolony formation Based on genetic and microscopic analysis, it has been proposed that the initial microcolonies in P. aeruginosa biolms form by aggregation of bacteria via twitching motility (OToole and Kolter, 1998a), and that the initial microcolonies in Vibrio cholerae El Tor biolms form by aggregation of cells via agella-driven motility along the substratum (Watnick and Kolter, 1999). If a biolm is initiated with a 1:1 mixture of yellow and cyan uorescent

1518 M. Klausen et al. CLSM time-lapse movies of development in the wild-type and DiM biolms indicated that the shift from sessile to motile cells occurred when the initial microcolonies reached a certain size, suggesting that the shift was induced by some sort of nutrient limitation. The fact that the bacteria in the DpilA and DpilADiM biolms did not spread on the substratum but instead formed protruding microcolonies at xed locations suggested that the expansive migration in the wild-type and DiM biolms was type IV pili driven. Evidence has been presented that twitching motility may be stimulated by iron limitation (Singh et al., 2002), and that it may possibly be directed by chemical gradients (Darzins, 1994; Kearns and Shimkets, 1998; Kearns et al., 2001). However, although it is known that twitching motility in P. aeruginosa is regulated by Vfr (a homologue of the Escherichia coli cyclic AMP receptor protein, Crp; Beatson et al., 2002), by the RpoN-dependent two-component sensor regulator pair PilS/PilR (Hobbs et al., 1993; Strom and Lory, 1993) and by the FimS/AlgR sensor regulator pair (Whitchurch et al., 1996), the environmental cues that affect twitching motility in P. aeruginosa are still not well understood (Mattick, 2002). As twitching motility is powered by a mechanism involving extensionattachmentretraction of type IV pili (Skerker and Berg, 2001), it is possible that type IV pili also play a role in keeping the migrating bacteria surface associated. Yet, other cell-to-substratum and cell-to-cell connections keep the bacteria associated in the DpilA and DpilADiM biolms. The nding that the initial microcolonies in the DiM biolms did not atten completely during biolm development also suggested that agellum-driven motility played a role in the formation of the at wild-type biolm (see below).

Fig. 7. Substratum coverage (A) and biomass (B) in developing P. aeruginosa wild-type ( ), DiM ( ), DpilA ( ) and DpilADiM ( ) biolms. Average values were calculated from COMSTAT image analysis on 18 CLSM images acquired in two ow chamber channels.

bacteria, the formation of microcolonies through aggregation of the bacteria would result in mixed microcolonies containing roughly equal numbers of yellow and cyan uorescent bacteria. Here, we report that biolms that were initiated with a 1:1 mixture of yellow and cyan uorescent bacteria, after an initial growth period, consisted of small microcolonies with predominantly yellow or predominantly cyan uorescent cells. This suggests that the initial microcolonies were formed by clonal growth from single cells attached to the substratum, and that the formation of initial microcolonies through aggregation of bacteria does not play a signicant role in P. aeruginosa biolms under the conditions used in the present study. Expansive migration After the initial formation of microcolonies by proliferation of sessile cells, the bacteria spread out on the substratum.
Fig. 8. Ratio of P. aeruginosa DiM bacteria to P. aeruginosa wildtype bacteria in a developing biolm that was initiated with a 1:1 mixture of Cfp-tagged DiM bacteria and Yfp-tagged wild-type bacteria. The ratios were calculated from COMSTAT analysis on six CLSM images acquired at six randomly chosen locations every 30 min. 2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1519

Fig. 9. Formation of the at P. aeruginosa biolm. Attachment occurs independently of agella and type IV pili (A). Microcolony formation occurs by clonal growth (B). The bacteria spread on the substratum by means of twitching motility (C) and form the at mature biolm (D).

Biolm maturation and dynamics We followed biolm development by the P. aeruginosa wild type and motility mutants for 7 days. Although the accumulation of biomass in the biolms had not reached a plateau at day 7, the basic structures of the biolms (at, hilly, heterogeneous protruding) did not change in the following days and, therefore, we refer to the 7-dayold biolms as the mature biolms. The different structures of the wild-type, DiM, DpilA and DpilADiM mature biolms suggested that agella and type IV pili have roles in shaping the architecture of P. aeruginosa biolms, although they are not necessary for biolm formation. In the wild-type and DiM mature biolms, which were both initiated with a 1:1 mixture of Cfp-tagged and Yfp-tagged cells, the foci at which the initial microcolonies had formed still contained predominantly cyan uorescent or yellow uorescent cells, whereas the regions between these foci contained a mixture of cyan uorescent and yellow uorescent cells. Time-lapse CLSM suggested that production of new cells in the biolms occurred at a high rate in the foci where the initial small microcolonies had formed. The formation of larger microcolonial structures in these foci in the wild-type biolms was prevented as a result of type IV pili-driven bacterial migration, agella-driven bacterial emigration out of the biolm and probably also sloughing due to shear stress. The hilly structures in the DiM mature biolms were probably maintained in the mature biolms because production of new cells in the foci at which the initial microcolonies had formed was not entirely counterbalanced as the DiM bacteria were not able to carry out agella-driven emigration out of the biolm. Reduced emigration out of the DiM biolm is supported by the nding of higher
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

biomass accumulation in the DiM biolms in comparison with the wild-type biolms. The protruding elongated microcolonies in the DpilA and DpilADiM mature biolms most probably developed during growth because the cells were unable to spread on the substratum by means of twitching motility. CLSM time-lapse movies showing development of wild-type and DiM biolms, and illustrating the points discussed here, are accessible at the internet site http://www.im.dtu.dk/mol-mic-avi. In support of the role of twitching motility in biolm development proposed here, Singh et al. (2002) showed that the presence of lactoferrin stimulates twitching motility through iron chelation and results in a at P. aeruginosa biolm, as opposed to a heterogeneous biolm obtained under their conditions in the absence of lactoferrin. Although it was shown that the presence of lactoferrin did not inhibit growth, the at biolms found in the Singh et al. (2002) study were much thinner than the at biolms observed in the present study. Gene expression in P. aeruginosa PAO1 biolm cells and planktonic cells has been compared using DNA microarrays (Whiteley et al., 2001). The genes for synthesis of agella and type IV pili were found to be repressed in the biolm cells. The wild-type, DiM, DpilA and DpilADiM strains in the present study initially formed similar young biolms with small microcolonies, but eventually formed very different mature biolms, suggesting that agella and type IV pili had roles in the later stages of biolm development. Whiteley et al. (2001) used minimal casamino acids medium in their studies, which is shown here to give rise to a at P. aeruginosa ow chamber biolm similar to the citrate-grown ow chamber biolm. Whiteley et al. (2001), however, used a chemostat vessel with granite pebbles as the substratum for biolm forma-

1520 M. Klausen et al. tion, and it cannot be excluded that P. aeruginosa biolm development in such a set-up differs from P. aeruginosa biolm development in ow chambers. Competition and selection in biolms The nding that the P. aeruginosa wild-type and DiM biolms are dynamic compositions with extensive migration occurring during development suggested that mutants that have a selective advantage in biolms may outcompete less tted variants. Competition experiments, in which biolms were initiated with 1:1 mixtures of Yfptagged DiM mutant and Cfp-tagged wild type (and the opposite colour combination), and the biomass accumulation of each strain in the developing biolm was quantied by COMSTAT analysis, showed that the DiM mutant ousts the wild type. This suggests that biolm experiments, especially those performed in owthrough systems, should also be regarded as selection experiments. That is, the physiological state of a population in a mature biolm could be the result of a selection process instead of a differentiation process as assumed by most authors (e.g. OToole et al., 2000; Stoodley et al., 2002). P. aeruginosa isolates from the lungs of cystic brosis patients are often non-agellated (Luzar et al., 1985; Mahenthiralingam et al., 1994), indicating that selection for non-agellated variants may occur in the biolms in the lungs of cystic brosis sufferers. Conclusions A model for P. aeruginosa biolm formation, which suggests the involvement of agella in attachment, the requirement of type IV pili-driven motility in microcolony formation and the subsequent formation of larger microcolonial structures via a maturation process, has become widely accepted as a general model for P. aeruginosa biolm development (e.g. Costerton et al., 1999; OToole et al., 2000; Stoodley et al., 2002). The present investigation of citrate-grown P. aeruginosa ow chamber biolms suggests an alternative model for biolm development shown schematically in Fig. 9. In this model, agella play no role in attachment, initial microcolony formation occurs by clonal growth, and twitching motility thereafter causes spreading over the substratum and prevents the formation of larger microcolonial structures in the very dynamic at mature biolm. It appears that biolm development occurs differently under different nutritional and environmental conditions, and that biolm models with present knowledge can only be contextual. Experimental procedures Bacterial strains and media
The strains used in the study are listed in Table 1. Escherichia coli MV1190 was used for cloning, while E. coli HB101 and CC118(lpir) were used in matings. Modied FAB medium (Heydorn et al., 2000) was used supplemented with 10 mM citrate for batch overnight cultures, and with 0.1 mM citrate, 0.3 mM glucose, 0.015% casamino acids or 0.5 mM benzoate for biolm cultivation. Biolms and batch cultures were grown at 30C.

Recombinant DNA techniques

The preparation of chromosomal and plasmid DNA, restriction endonuclease digestion, ligation reactions, polymerase chain reaction (PCR) and Southern blotting were carried out using standard protocols (Ausubel et al., 1992).

Construction of plasmids used for allelic displacement

The pilA::Telr cassette was constructed by: (i) PCR amplifying the pilA gene from genomic DNA of P. aeruginosa PAO1 using the primer pair 5-ggaatcgaattcagaagtacgcggtcacctg3/5-cggagatgcctacaaagagc-3; (ii) cloning the blunted and EcoRI-digested 1.6 kb PCR fragment, into EcoRISmaIdigested pUC18-NotI; and (iii) replacing a 71 bp KpnIBstXI fragment from the coding region of the pilA gene with a tellurite resistance gene obtained from pUT-Tel (SanchezRomero et al., 1998) as a blunted BamHIHindIII fragment. The pilA knock-out plasmid, pTTN80, was constructed by cloning the pilA::Telr-containing NotI fragment in the NotI site of pCK318. The iM::Tcr cassette was constructed by: (i) PCR amplifying two separate parts of the iM gene from genomic DNA of P. aeruginosa PAO1 using the primer pair 5-ggaatcaagcttcggcctgytgctggcgatcg-3/5-ggaatcggatccattt ccagggtcggcatgcg-3 for the left 607 bp fragment and the primer pair 5-ggaatcgagctccgctgttcatcctcgacgcc-3/5-ggaat cgaattcagsgccaggttgcccttgtg-3 for the right 594 bp fragment; and (ii) cloning the PCR fragments, digested with BamHI and HindIII or SacI and EcoRI, in pTTN60 digested sequentially with the same restriction enzymes on either site of the Tcr gene. The iM knock-out plasmid, pTTN61, was constructed by cloning the iM::Tcr-containing NotI fragment in the NotI site of pCK318.

Construction of DiM, DpilA and DiMDpilA mutant derivatives of P. aeruginosa PAO1 by allelic displacement
The P. aeruginosa DiM, DpilA and DiMDpilA mutants were constructed by allelic displacement using triparental mating as described previously (Andersen et al., 1998) with E. coli HB101/RK600, P. aeruginosa PAO1 and E. coli CC118(lpir) containing delivery vector pTTN61 or pTTN80. Exconjugants with knock-out constructs inserted in the chromosome were selected on AB plates supplemented with citrate (10 mM) and tellurite (150 mg ml-1) or tetracycline (80 mg ml-1). Subsequent sucrose-based screening for double cross-over mutants was performed as described by Schweizer and Hoang (1995). Allelic displacement was conrmed by Southern analysis and PCR. As opposed to the wild type, the DiM and DiMDpilA mutants did not spread in LB medium solidied with 0.3% agar, and did not show swimming motility under microscopic inspection. The presence of agella on the wild type and the absence of agella on the DiM and DiMDpilA mutants were conrmed using Leifsons agella stain 2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1521

Table 1. Strains and plasmids used in the study. Strain or plasmid E. coli MV1190 HB101 CC118(lpir) P. aeruginosa PAO1 PAO1GFP PAO1CFP PAO1YFP DiMGFP DiMCFP DiMYFP DpilAGFP DpilACFP DpilAYFP DpilADiMGFP DpilADiMCFP DpilADiMYFP Plasmids pUT-Tc pUT-Tel pUC18NotI pTTN50 pTTN60 pCK318 Relevant characteristics D(lac-proAB) thi supE D(srl-recA)306::tn10(F:traD36 proAB lacIq D(lacZ)M15) Smr, recA thi pro leu hsdRM+ D(ara-leu) araD DlacX74 galK phoA20 thi-1 rpsE rpoB argE(Am) recA; lysogenized with lpir phage Wild type PAO1 tagged with eGFP in a mini-Tn7 construct, Gmr PAO1 tagged with eCFP in a mini-Tn7 construct, Gmr PAO1 tagged with eYFP in a mini-Tn7 construct, Gmr iM inactivated by allelic displacement; tagged with eGFP in a mini-Tn7 construct, Tcr, Gmr iM inactivated by allelic displacement; tagged with eCFP in a mini-Tn7 construct, Tcr, Gmr iM inactivated by allelic displacement; tagged with eYFP in a mini-Tn7 construct, Tcr, Gmr pilA inactivated by allelic displacement; tagged with eGFP in a mini-Tn7 construct, Telr, Gmr pilA inactivated by allelic displacement; tagged with eCFP in a mini-Tn7 construct, Telr, Gmr pilA inactivated by allelic displacement; tagged with eYFP in a mini-Tn7 construct, Telr, Gmr pilA and iM inactivated by allelic displacement; tagged with eGFP in a mini-Tn7 construct, Telr, Tcr, Gmr pilA and iM inactivated by allelic displacement; tagged with eCFP in a mini-Tn7 construct, Telr, Tcr, Gmr pilA and iM inactivated by allelic displacement; tagged with eYFP in a mini-Tn7 construct, Telr, Tcr, Gmr Source of Tcr cartridge Source of Telr cartridge Cloning vector pUC18Not:: PA1/04/03-dsRed; used in the construction of miniTn7(Gm)PA1/04/03-ecfp-a and miniTn7(Gm)PA1/04/03-eyfp-a pUC18Not with a Tcr gene cloned in the SmaI site; used in the construction of knock-out cassettes Apr sacB ori-R6K RP4-mob+ Source or reference Vieira and Messing (1987) Kessler et al. (1992) Herrero et al. (1990)

Holloway and Morgan (1986) This study This study This study This study This study This study This study This study This study This study This study This study

De Lorenzo et al. (1990) Sanchez-Romero et al. (1998) Herrero et al. (1990) Tolker-Nielsen et al. (2000) This study M. Gjermansen, P. Ragas, C. Sternberg, S. Molin, and T. Tolker-Nielsen (unpubl.) This study This study Kessler et al. (1992) Bao et al. (1991) Koch et al. (2001) Koch et al. (2001) This study This study

pTTN61 pTTN80 RK600 pUX-BF13 pBK-mini-Tn7-gfp2 pBK-miniTn7-WGm miniTn7(Gm)PA1/04/03-ecfp-a miniTn7(Gm)PA1/04/03-eyfp-a

pCK318 with iM::Tcr cassette cloned in the NotI site; used for allelic displacement pCK318 with pilA::Telr cassette cloned in the NotI site; used for allelic displacement Cmr ori-ColE1 RK2-mob+RK2-tra+ helper plasmid in matings Apr mob+ ori-R6K; helper plasmid; providing the tn7 transposition functions in trans Gmr Apr mob+; delivery plasmid for mini-Tn7-Gmr-PA1/04/03-egfp Used in the construction of mini-Tn7 delivery plasmids Gmr Apr mob+; delivery plasmid for mini-Tn7-Gmr-PA1/04/03-ecfp Gmr Apr mob+; delivery plasmid for mini-Tn7-Gmr-PA1/04/03-eyfp

ing method (Meynell and Meynell, 1970). As opposed to the wild type, the DpilA and DiMDpilA mutants did not spread by twitching motility on the interstitial surface between 1.5% agar and the Petri dish plastic surface. P. aeruginosa mutants inactivated in the major pilin subunit gene, pilA, are reported to be non-piliated (Mattick, 2002).

the templates pECFP and pEYFP (Clontech) using the primer pair 5-atatagcatgctgagcaagggcgaggagctg-3/5-ctctc aagcttattacttgtacagctcgtccatgcc-3; (ii) cloning SphIHindIIIdigested PCR fragments into the SphIHindIII site of pTTN50; and (iii) cloning the 2000 bp NotI fragments containing the PA1/04/03-ecfp and PA1/04/03-eyfp fusions into the NotI site of pBK-miniTn7-WGm.

Construction of plasmids used for uorescent tagging

The delivery plasmids miniTn7(Gm)PA1/04/03-ecfp-a and miniTn7(Gm)PA1/04/03-eyfp-a were constructed by: (i) PCR amplifying the ecfp and eyfp genes as 740 bp fragments from
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Insertion of cfp, gfp and yfp into the chromosome of P. aeruginosa PAO1, DiM, DpilA and DiMDpilA
In order to insert ecfp, egfp or eyfp into a neutral inter-

1522 M. Klausen et al.

genic region downstream of the glmS gene in the P. aeruginosa genome, we used the mini-Tn7 system described by Koch et al. (2001). Four-parental mating was carried out with the relevant P. aeruginosa strain, E. coli HB101/ RK600, E. coli HB101/pUX-BF13 and E. coli HB101 containing pBK-mini-Tn7-gfp2, miniTn7(Gm)PA1/04/03-ecfp-a or miniTn7(Gm)PA1/04/03-eyfp-a. The egfp, ecfp and eyfp genes are fused to the PA1/04/03 promoter (Lanzer and Bujard, 1988), which acts as a constitutive promoter in Pseudomonas spp. (Andersen et al., 1998). Fluorescent exconjugants with mini-Tn7 cassettes inserted in the chromosome were selected on AB plates supplemented with citrate (10 mM) and gentamicin (30 mg ml-1). The correct insertion of the mini-Tn7 cassettes in the intergenic region immediately downstream of the glmS gene was conrmed by Southern analysis and PCR. The uorescently tagged strains showed no phenotypic changes compared with the parental strains, when tested in liquid medium or ow chamber biolms. laboratory automation workstation from Beckman Coulter, which (i) washed the wells twice with 0.9% NaCl; (ii) stained with 0.1% crystal violet; (iii) washed twice with 0.9% NaCl; and (iv) resuspended in 96% ethanol. Biolm cell-associated dye was measured at OD600.

Microscopy and image acquisition

All microscopic observations and image acquisitions were performed on a Zeiss LSM 510 confocal laser scanning microscope equipped with detectors and lter sets for monitoring of Cfp, Gfp and Yfp uorescence. Images were obtained with a 63 1.4 objective or a 40 1.3 objective. Simulated three-dimensional images, shadow projections and sections were generated using the IMARIS software package (Bitplane).

Image analysis
STAT

Cultivation of biolms
Biolms were grown at 30C in ow chambers with individual channel dimensions of 1 4 40 mm. The ow system was assembled and prepared as described previously (Mller et al., 1998). The ow chambers were inoculated by injecting 350 ml of overnight culture diluted to an OD600 of 0.001 into each ow channel with a small syringe. After inoculation, ow channels were left without ow for 1 h, after which medium ow was started using a Watson Marlow 205S peristaltic pump. Mean ow velocity in the ow cells was 0.2 mm s-1, corresponding to laminar ow with a Reynolds number of 0.02.

CLSM images were analysed by the computer program COM(Heydorn et al., 2000). A xed threshold value and connected volume ltration was used for all image stacks. A script written in MATLAB (MathWorks) was used to count attached cells from CLSM images.

Acknowledgements
This work was supported by the Danish Biotechnological Research Program. We thank Anne Nielsen for technical assistance with the construction of miniTn7(Gm)PA1/04/03-ecfpa and miniTn7(Gm)PA1/04/03-eyfp-a.

Attachment assays
To quantify attachment under owthrough conditions, 350 ml of overnight cultures (wild type, DiM, DpilA and DiMDpilA) diluted to OD600 = 0.1 was injected into two channels with the medium ow running (0.2 mm s-1). After 5 min with medium ow running, nine CLSM images were captured at random positions in each of the ow channels. To quantify attachment under static conditions, 350 ml of overnight cultures (wild type, DiM, DpilA and DiMDpilA) diluted to OD600 = 0.001 was injected into ow chambers without ow. After 1 h, the ow was started (0.2 mm s-1), and nine CLSM images were captured at random positions in each ow channel. The number of adherent bacteria was subsequently counted by subjecting the CLSM images to image analysis.

References
Andersen, J.B., Sternberg, C., Poulsen, L.K., Bjrn, S.P., Givskov, M., and Molin, S. (1998) New unstable variants of green uorescent protein for studies of transient gene expression in bacteria. Appl Environ Microbiol 64: 2240 2246. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., and Struhl, K. (1992) Short Protocols in Molecular Biology. New York: Greene Publishing and John Wiley & Sons. Bao, Y., Lies, D.P., Fu, H., and Roberts, G.P. (1991) An improved Tn7-based system for the single-copy insertion of cloned genes into chromosomes of gram-negative bacteria. Gene 109: 167168. Beatson, S.A., Whitchurch, C.B., Sargent, J.L., Levesque, R.C., and Mattick, J.S. (2002) Differential regulation of twitching motility and elastase production by Vfr in Pseudomonas aeruginosa. J Bacteriol 184: 36053613. Costerton, J.W., Lewandowski, Z., Caldwell, D.E., Korber, D.R., and Lappin-Scott, H.M. (1995) Microbial biolms. Annu Rev Microbiol 49: 711745. Costerton, J.W., Stewart, P.S., and Greenberg, E.P. (1999) Bacterial biolms: a common cause of persistent infections. Science 284: 13181322. Darzins, A. (1994) Characterization of a Pseudomonas aeruginosa gene cluster involved in pilus biosynthesis and
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

Microtitre plate biolm formation assay

The microtitre plate biolm formation assay was performed according to the method of OToole and Kolter (1998a) with modications. Overnight cultures were grown in modied FAB medium supplemented with 0.2% glucose + 0.5% casamino acids, or 10 mM citrate, and diluted to OD600 = 0.001 with fresh medium. The diluted cultures (100 ml) were transferred to wells of polystyrene microtitre plates and incubated for 48 h at 30C. Biolm cells were stained using a Biomek 2000

Roles of bacterial motility in the formation of the at P. aeruginosa biolm 1523

twitching motility: sequence similarity to the chemotaxis proteins of enterics and the gliding bacterium Myxococcus xanthus. Mol Microbiol 11: 137153. Darzins, A., and Russell, M.A. (1997) Molecular genetic analysis of type-4 pilus biogenesis and twitching motility using Pseudomonas aeruginosa as a model system a review. Gene 192: 109115. Davies, D.G., Parsek, M.R., Pearson, J.P., Iglewski, B.H., Costerton, J.W., and Greenberg, E.P. (1998) The involvement of cell-to-cell signals in the development of a bacterial biolm. Science 280: 295298. De Lorenzo, V., Herrero, M., Jacubzik, U., and Timmis, K.N. (1990) Mini-Tn5 transposon derivatives for insertion mutagenesis promoter probing, and chromosomal insertion of cloned DNA in Gram-negative eubacteria. J Bacteriol 172: 65686572. DeKievit, T.R., Gillis, R., Marx, S., Brown, C., and Iglewski, B.H. (2001) Quorum-sensing genes in Pseudomonas aeruginosa biolms: their role and expression patterns. Appl Environ Microbiol 67: 18651873. Heilmann, C., Gerke, C., Perdreau-Remington, F., and Gtz, F. (1996) Characterization of Tn917 insertion mutants of Staphylococcus epidermidis affected in biolm formation. Infect Immun 64: 277282. Herrero, M.V., de Lorenzo, V., and Timmis, K.N. (1990) Transposon vectors containing non-antibiotic resistance markers for cloning and stable chromosomal insertion of foreign genes in Gram negative bacteria. J Bacteriol 172: 65576567. Heydorn, A., Nielsen, A.T., Hentzer, M., Sternberg, C., Givskov, M., Ersbll, B.K., and Molin, S. (2000) Quantication of biolm structures by the novel computer program COMSTAT. Microbiology 146: 23952407. Heydorn, A., Ersbll, B., Kato, J., Hentzer, M., Parsek, M., Tolker-Nielsen, T., et al. (2002) Statistical analysis of Pseudomonas aeruginosa biolm development: Impact of mutations in genes involved in twitching motility, cell-to-cell signaling, and stationary-phase sigma factor expression. Appl Environ Microbiol 68: 20082017. Hobbs, M., Collie, E.S., Free, P.D., Livingston, S.P., and Mattick, J.S. (1993) PilS and PilR, a two-component transcriptional regulatory system controlling expression of type 4 mbria in Pseudomonas aeruginosa. Mol Microbiol 7: 669682. Holloway, B.W., and Morgan, A.F. (1986) Genome organization in Pseudomonas. Annu Rev Microbiol 40: 79105. Kearns, D.B., and Shimkets, L.J. (1998) Chemotaxis in a gliding bacterium. Proc Natl Acad Sci USA 95: 11957 11962. Kearns, D.B., Robinson, J., and Shimkets, L.J. (2001) Pseudomonas aeruginosa exhibits directed twitching motility up phosphatidylethanolamine gradients. J Bacteriol 183: 763767. Kessler, B., de Lorenzo, V., and Timmis, K.N. (1992) A general system to integrate lacZ fusions into the chromosomes of gram negative eubacteria: regulation of Pm promoter in the TOL plasmid studied with all controlling elements in monocopy. Mol Gen Genet 233: 293301. Koch, B., Jensen, L.E., and Nybroe, O. (2001) A panel of Tn7-based vectors for insertion of the gfp gene or for delivery of cloned DNA into Gram-negative bacteria at a
2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524

neutral chromosomal site. J Microbiol Methods 45: 187 195. Lanzer, M., and Bujard, H. (1988) Promoters largely determine the efciency of repressor action. Proc Natl Acad Sci USA 85: 89738977. Luzar, M.A., Thomassen, M.J., and Montie, T.C. (1985) Flagella and motility alterations in Pseudomonas aeruginosa strains from patients with cystic brosis: relationship to patient clinical condition. Infect Immun 50: 577582. Mahenthiralingam, E., Campbell, M.E., and Speert, D.P. (1994) Nonmotility and phagocytic resistance of Pseudomonas aeruginosa isolates from chronically colonized patients with cystic brosis. Infect Immun 62: 596605. Mattick, J.S. (2002) Type IV pili and twitching motility. Annu Rev Microbiol 56: 289314. Meynell, G.G., and Meynell, E. (1970) Theory and Practice in Experimental Bacteriology, 2nd edn. London: Cambridge University Press, pp. 167168. Mller, S., Sternberg, C., Andersen, J.B., Christensen, B.B., and Molin, S. (1998) In situ gene expression in mixedculture biolms: evidence of metabolic interactions between community members. Appl Environ Microbiol 64: 721732. Nielsen, A.T., Tolker-Nielsen, T., Barken, K.B., and Molin, S. (2000) Role of commensal relationships on the spatial structure of a surface-attached microbial consortium. Environ Microbiol 2: 5968. OToole, G.A., and Kolter, R. (1998a) Flagellar and twitching motility are necessary for Pseudomonas aeruginosa biolm development. Mol Microbiol 30: 295304. OToole, G.A., and Kolter, R. (1998b) Initiation of biolm formation in Pseudomonas uorescens WCS365 proceeds via multiple, convergent signalling pathways: a genetic analysis. Mol Microbiol 28: 449461. OToole, G., Kaplan, H.B., and Kolter, R. (2000) Biolm formation as microbial development. Annu Rev Microbiol 54: 4979. Pratt, L.A., and Kolter, R. (1998) Genetic analysis of Escherichia coli biolm formation: roles of agella, motility, chemotaxis and type I pili. Mol Microbiol 30: 285293. Sanchez-Romero, J.M., Diaz-Orejas, R., and de Lorenzo, V. (1998) Resistance to tellurite as a selection marker for genetic manipulations of Pseudomonas strains. Appl Environ Microbiol 64: 40404046. Sauer, K., Camper, A.K., Ehrlich, G.D., Costerton, J.W., and Davies, D.G. (2002) Pseudomonas aeruginosa displays multiple phenotypes during development as a biolm. J Bacteriol 184: 11401154. Schweizer, H.P., and Hoang, T.T. (1995) An improved system for replacement and xylE fusion analysis in Pseudomonas aeruginosa. Gene 158: 1522. Semmler, A.B., Whitchurch, C.B., and Mattick, J.S. (1999) A re-examination of twitching motility in Pseudomonas aeruginosa. Microbiology 145: 28632873. Singh, P.K., Parsek, M.R., Greenberg, E.P., and Welsh, M.J. (2002) A component of innate immunity prevents bacterial biolm development. Nature 417: 552555. Skerker, J.M., and Berg, H.C. (2001) Direct observation of extension and retraction of type IV pili. Proc Natl Acad Sci USA 98: 69016904. Stewart, P.S., Peyton, B.M., Drury, W.J., and Murga, R.

1524 M. Klausen et al.

(1993) Quantitative observations of heterogeneities in Pseudomonas aeruginosa biolms. Appl Environ Microbiol 59: 327329. Stoodley, P., Sauer, K., Davies, D.G., and Costerton, J.W. (2002) Biolms as complex differentiated communities. Annu Rev Microbiol 56: 187209. Strom, M.S., and Lory, S. (1993) Structurefunction and biogenesis of the type IV pili. Annu Rev Microbiol 47: 565596. Tolker-Nielsen, T., Brinch, U.C., Ragas, P.C., Andersen, J.B., Jacobsen, C.S., and Molin, S. (2000) Development and dynamics of Pseudomonas sp. biolms. J Bacteriol 182: 64826489. Vieira, J., and Messing, J. (1987) Production of single stranded plasmid DNA. Methods Enzymol 153: 311. Watnick, P.I., and Kolter, R. (1999) Steps in the development of a Vibrio cholerae El Tor biolm. Mol Microbiol 34: 586 595. Whitchurch, C.B., Alm, R.A., and Mattick, J.S. (1996) The alginate regulator AlgR and an associated sensor FimS are required for twitching motility in Pseudomonas aeruginosa. Proc Natl Acad Sci USA 93: 98399843. Whitchurch, C.B., Tolker-Nielsen, T., Ragas, P.C., and Mattick, J.S. (2002) Extracellular DNA is required for bacterial biolm formation. Science 295: 1487. Whiteley, M., Bangera, M.G., Bumgarner, R.E., Parsek, M.R., Teitzel, G.M., Lory, S., and Greenberg, E.P. (2001) Gene expression in Pseudomonas aeruginosa biolms. Nature 413: 860864.

2003 Blackwell Publishing Ltd, Molecular Microbiology, 48, 15111524