LapF, the second largest Pseudomonas putida protein, contributes to plant root colonization and determines biofilm architecture


E-mail; Tel. (+34) 958 181600; Fax (+34) 958 129600.


We have investigated the role of LapF, one of the two largest proteins encoded in the genome of Pseudomonas putida KT2440, in bacterial colonization of solid surfaces. LapF is 6310 amino acids long, and is localized on the cell surface. The C-terminal region of the protein is essential for its secretion, which presumably requires the ABC transporter encoded by an operon (lapHIJ) adjacent to the lapF gene. Although the initial attachment stages are not different between the wild type and a lapF mutant, microcolony formation and subsequent development of a mature biofilm is impaired in the mutant. This is consistent with the expression pattern of lapF; activation of its promoter takes place at late stages of growth and is regulated by the alternative sigma factor RpoS. A lapF mutant is also affected in individual and competitive plant root colonization. In these assays, mixed microcolonies formed by cells of both the wild-type and the mutant strains could be observed but microcolonies of the mutant alone were not found. These data and the localization of the protein at discrete spots in areas of contact between cells in biofilms suggest that LapF determines the establishment of cell–cell interactions during sessile growth.


Colonization of solid surfaces seems to be a general persistence strategy for microorganisms. Single- or multispecies communities associated to a surface and surrounded by an exopolymeric matrix, referred to as biofilms, are known to be formed by a wide variety of bacteria under different environmental conditions and on diverse surfaces, both biotic and abiotic. The genetic basis of biofilm development has been the subject of much attention, and a number of genes have been found in various model microorganisms to play a role in this process. It is now clear that biofilm formation is a complex phenomenon that involves the integration of environmental and cellular signals into an intrincate regulatory network, leading to the adaptation of bacteria to multicellular life on a surface (for recent reviews, see Goller and Romeo, 2008; Hengge, 2009; Landini, 2009).

Bacteria of the genus Pseudomonas– particularly Pseudomonas aeruginosa– are among the better-studied microorganisms with respect to phenotypic changes taking place throughout the process of biofilm formation and the genetic determinants involved. A general outline of the sequence of events leading from a planktonic culture to a mature biofilm has been assembled, going through the stages of initial adhesion of individual cells, followed by further colonization of the surface, growth and aggregation to form microcolonies, and production of an extracellular matrix. Genes involved in Pseudomonas attachment to surfaces and biofilm structure determinants have been identified (O'Toole and Kolter, 1998; Davey et al., 2003; Hinsa et al., 2003; Byrd et al., 2009) and variations in the gene expression profile of Pseudomonas putida and P. aeruginosa at the different stages have been reported (Sauer and Camper, 2001; Sauer et al., 2002). This model of biofilm development with a specific genetic programme following sequential steps has been the paradigm for the past years, but is being questioned (Ghigo, 2003), in part due to the incomplete information existing with respect to the potential hierarchy of genetic elements and checkpoints during biofilm formation (Monds and O'Toole, 2009). Decyphering the precise mechanisms involved in the different stages of biofilm formation will not only help to understand microbial adaptation to this particular mode of life in a wide range of environments, but it may also provide relevant information with respect to the interaction of bacteria with eukaryotic hosts.

The Gram-negative bacterium P. putida KT2440 has been extensively studied in our laboratory as a model organism in plant–bacterial interactions. Such studies have provided evidences indicating that, while a number of genetic determinants are specifically relevant for the bacterium–plant interaction, biofilm formation and root colonization share certain common mechanisms (Yousef-Coronado et al., 2008). As an example, previous work has established the importance of the surface-associated protein LapA in biofilm formation both by Pseudomonas fluorescens and P. putida (Hinsa et al., 2003; Hinsa and O'Toole, 2006). This protein was identified as a key determinant of bacterial adhesion to corn seeds (Espinosa-Urgel et al., 2000), and of root colonization by P. putida (Yousef-Coronado et al., 2008). LapA is the largest protein encoded in the genome of P. putida KT2440, and closely related proteins are present in a wide variety of microorganisms (Lasa and Penadés, 2006; Yousef and Espinosa-Urgel, 2007). Some of these large proteins containing long-amino-acid repeats have been shown to participate in biofilm formation in different organisms, as is the case with Bap in Staphylococcus aureus and Salmonella enteritidis (Cucarella et al., 2001; Latasa et al., 2005; Tormo et al., 2005), or its Acinetobacter baumanii homologue (Loehfelm et al., 2008). In the first two cases, the Bap protein is also relevant for colonization of eukayotic hosts by pathogenic strains (Cucarella et al., 2004; Latasa et al., 2005).

We now present evidences of the role played by a second large surface protein, which we have named LapF, in colonization of abiotic and biotic surfaces. While LapA determines the transition from reversible to irreversible attachment during biofilm formation (Hinsa et al., 2003), LapF is important for further development of a mature biofilm, as well as for the fitness of P. putida in the rhizosphere. Our results expose a key piece in the mechanism of biofilm formation by this plant-beneficial microorganism.


lapF (PP_0806) encodes the second largest P. putida adhesin

Strain mus-20 was originally identified in a screen for P. putida KT2440 random transposon mutants with reduced capacity to colonize corn seeds (Espinosa-Urgel et al., 2000). At the time, the incomplete information available on the genome of KT2440 prevented further characterization of this mutant. Completion of the genome sequence allowed the identification of the disrupted locus, PP_0806, and the corresponding protein, LapF (large adhesion protein). LapF is 6310 amino acids long and is related to a group of large bacterial proteins that includes LapA (Yousef-Coronado et al., 2008), a protein that has been shown to participate in bacterial adhesion to seeds and roots and biofilm formation by P. putida and P. fluorescens (Hinsa et al., 2003; Yousef and Espinosa-Urgel, 2007). As shown in Fig. 1A, the genetic organization of the chromosomal region containing lapF is similar to that of lapA, with adjacent open reading frames that encode the components of a putative type I secretion system. We have named these genes lapH (PP_0805, encoding a predicted outer membrane protein), lapI (PP_0804, a membrane-bound ATPase) and lapJ (PP_0803, a putative membrane fusion protein). The predicted start and stop codons of these three genes overlap, suggesting they are translationally coupled and form an operon. In the case of LapA, a similar transporter formed by LapB, LapC and LapE is required for translocation of this protein to the cell surface (Hinsa et al., 2003). Details on sequence similarities between the two Lap systems are described in Table S1.

Figure 1.

A. Genetic organization of the KT2440 chromosomal region containing lapF and comparison with the region comprising lapA.
B. Schematic view of the LapF protein, showing the three structural domains and significant characteristics; secretion signal and calcium binding sites (yellow), and transposon insertion (arrow) are indicated. The sequence conservation (overall stack height) and relative frequency of each residue (height of each symbol) at every position of the 64 repeats are shown. The image was created using WebLogo software (University of California, Berkeley).
C. Localization of LapF in KT2440 and mus-20 by Western blot analysis.

Although in terms of primary sequence there is significant divergence, LapF shows fundamental architectural similarity with LapA and other large surface proteins. LapF presents three domains (Fig. 1B). Domain 1, comprising the first 152 amino acids, is followed by a repetitive region (Domain 2) covering over 85% of the protein. It consists on 64 imperfect repeats of 83–91 amino acids each (Fig. 1B). Domain 3, 691 amino acids long, corresponds to the C-terminal portion of the protein. It contains a predicted calcium binding region that includes two copies of the GGXGXD motif (GGAGDD and GGSGTD), which has been proposed as a distinctive feature of proteins that are secreted via type I secretion systems (Delepelaire, 2004). The transposon insertion in mutant mus-20 is located in Domain 3, resulting in the protein being truncated 52 amino acids before the calcium binding region and secretion motifs.

LapF is localized on the cell surface and requires the C-terminal domain for translocation

In order to confirm the prediction that LapF is a surface-associated protein, polyclonal antibodies were obtained and used in a Western blot analysis of proteins obtained from the different subcellular fractions, as detailed in Experimental procedures. Results are shown in Fig. 1C. A single band of high molecular weight reacting with the antibodies could be detected in the outer membrane fraction and in the cytoplasmic fraction of KT2440, indicating that after its synthesis in the cytoplasm, the protein is translocated to the cell surface. In contrast, a band corresponding to LapF was only found in the cytoplasmic fraction in mutant mus-20 (Fig. 1C) and not in the outer membrane fraction, which indicates that the C-terminal domain disrupted by the transposon insertion is essential for translocation, as predicted, and that the role of LapF in adhesion to seeds requires its correct localization on the cell surface.

Role of LapF in plant root colonization by P. putida KT2440

Whereas the large adhesin LapA and other determinants involved in attachment to seeds are also important for further colonization of plant roots, certain seed adhesion traits have no significant impact on bacterial fitness in the rhizosphere (Yousef-Coronado et al., 2008). In order to establish the role of LapF, competitive colonization of corn roots by KT2440 and mus-20 was assayed. Seeds were inoculated with a 1:1 mix of KT2440-Sm and mus-20 before sowing in tubes containing sterile sand. Alternatively, the mix was added to the sand prior to sowing surface-sterilized seeds. Bacteria were recovered after 4, 7 and 12 days and the number of wild-type and mus-20 cells was determined. In both types of experiment, the lapF mutant showed reduced competitive fitness (Fig. 2). However, the number of both wild-type and mus-20 cells increased over time and the differences between the two strains tended to decrease, from one order of magnitude at day 4 to approximately a 3:1 proportion of KT2440 versus mus-20 after 12 days. Similar results were obtained when colonization of alfalfa roots was quantified (data not shown).

Figure 2.

Competitive colonization of corn and alfalfa rhizosphere by KT2440 and mus-20.
A. Quantitative analysis of corn root colonization. The number of cells of each strain recovered after 4, 7 and 12 days is shown. Grey bars: KT2440; white bars, mus-20.
B–D. Visualization of alfalfa root colonization by fluorescence microscopy. Images are composites of each field observed with appropriate filter sets for dsRed (KT2440, red cells) and Gfp (mus-20, green cells).
E. A field corresponding to mus-20gpf alone on alfalfa roots is shown (line: 10 µm).

Fluorescence microscopy was also used to examine colonization of 7-day-old alfalfa roots by KT2440 and mus-20. For these experiments, strains were labelled with mini-Tn7 derivatives harbouring the fluorescent protein-encoding genes dsRed and gfp (Koch et al., 2001) respectively. Interestingly, microcolonies of KT2440 and mixed KT2440/mus-20 aggregates could be observed on different parts of the root (Fig. 2), but microcolonies of the mutant alone were not found. This result and the quantitative data suggest that LapF is involved in cell–cell interactions, and that the mutant can at least in part benefit from the protein produced and secreted by the wild-type strain. If this interpretation was correct, it would be expected that mus-20 also showed reduced colonization capacity in the absence of competition. To test if this was the case, corn seeds were inoculated with either strain alone, and root colonization was quantitatively assessed as before. As with the competitive assays, the number of wild-type bacteria per gram of root recovered after 4 days was significantly higher than that of mus-20 (109 ± 4.5 × 108 versus 2 × 108 ± 8.2 × 107 cfu per gram of root), a difference that was maintained after 7 and 12 days of incubation.

LapF participates in biofilm development under environmental conditions favouring prolonged sessile growth

Previous results obtained with mutant mus-20 indicated that LapF was not a relevant factor for early bacterial attachment to abiotic surfaces (Espinosa-Urgel et al., 2000), but an exhaustive follow-up of biofilm formation kinetics had not been performed. Biofilm formation by KT2440 and the lapF mutant in microtitre plates (polystyrene plastic) was monitored during growth in LB, LB diluted 1/10, and minimal medium with glucose or glucose plus casamino acids. Planktonic growth of the two strains was similar in all the media tested, with nearly identical duplication times (Table S2). The biomass attached to the surface was evaluated by crystal violet staining, as described in Experimental procedures. Significant differences were only observed in M9 with glucose at late time points; as shown in Fig. 3A, the wild-type strain continued developing a biofilm for 48 h, while the biofilm of the lapF mutant did not progress any further after 24 h. In the other media tested, mus-20 showed attachment kinetics similar to the wild type, although the maximum biomass associated to the surface was always slightly higher in the case of KT2440 (Fig. S1). As previously observed (Yousef-Coronado et al., 2008), detachment of cells from the surface began to take place after 8 h of incubation in LB and LB diluted 1/10.

Figure 3.

A. Quantification of biofilm formation by KT2440 and mus-20 grown under static conditions in microtitre plates in M9 with glucose.
B. Twenty-four and 48 h biofilms of the wild type (left tube) and mus-20 (right tube) grown in M9 with glucose on borosilicate glass tubes.

Biofilm formation was also evaluated during growth in rotating glass tubes. As with the microtitre plates, the differences were not significant in LB medium, whereas in M9 with glucose and in 1/10 LB the mutant showed reduced attachment with respect to the wild type at late time points (Fig. 3B).

These data suggested that LapF had no relevant role in the early attachment stages but could influence the development of a mature biofilm. To further explore this possibility, biofilm formation was followed on glass coverslips by fluorescence miscroscopy, using KT2440 and mus-20 labelled with gfp in single copy in the chromosome. Results are shown in Fig. 4. As expected from the previous experiments, no differences were observed in the number of cells attached to the coverslip at early time points. However, as biofilm development progressed, KT2440 showed a more organized three-dimensional structure than mus-20, and the biofilm formed after 24 h was significantly denser in the wild type than in the mutant. This phenotype of the lapF mutant is clearly different from that of a lapA mutant, which under the same conditions only showed scattered cells attached to the surface at all time points (Fig. 4).

Figure 4.

Comparative analysis of the development of KT2440gfp, mus-20gfp (lapF mutant) and mus-42gfp (lapA mutant) biofilms on glass coverslips. Cultures were grown in six-well plates with a 40 × 20 mm glass coverslip obliquely placed in each well. Coverslips were removed every 2 h and biofilm formation was followed by fluorescence microscopy (magnification: 1000×). Results shown correspond to cultures grown in M9 with glucose; similar results were obtained in 1:10 LB.

A detailed analysis of the biofilm structure formed by the wild-type and lapF- strains was then performed under flow conditions by confocal laser scanning microscopy (CLSM). As shown in Fig. 5, the differences between the wild type and mus-20 were evident; KT2440 developed a mature biofilm, whereas in the case of the lapF mutant, cells attached to the surface appeared isolated and not forming microcolonies. Analysis of biofilm parameters using the comstat software (Heydorn et al., 2000) showed a 40-fold difference in biomass and 60-fold difference in average thickness between KT2440 and mus-20 biofilms (Table 1).

Figure 5.

Confocal laser scanning microscopy images of KT2440gfp (left) and mus-20gfp (right) biofilms grown for 6 h in 1:10 LB in flow cells.

Table 1. comstat analysis of biofilm parameters in KT2440 and the lapF mutant mus-20.
Biomass (µm3 µm−2)2.290.06
Average thickness (µm)8.270.14
Maximum diffusion distance (µm)3.221.24
Maximum thickness (µm)2910

Biofilm formation on glass coverslips was also examined in co-cultures of the wild type and the mutant tagged with fluorescent proteins. As in the pure culture experiments, although cells of both strains initiated attachment, mus-20 was unable to form microcolonies (Fig. 6), and only individual cells of the mutant embedded in microcolonies of the wild-type strain could be observed at late time points. This result supports a role for LapF in cell–cell interactions and suggests that cells unable to localize the protein on their surface cannot develop a mature biofilm by themselves. Differential effects due to the specific fluorescent protein used in each strain were discarded by switching the tags (Fig. 6).

Figure 6.

Top row: Mixed culture biofilms of KT2440gfp and mus-20dsRed on glass coverslips, followed at different times. Magnification: 400×. Bar: 10 µM. Bottom row: Mixed culture biofilms of KT2440dsRed (A) and mus-20gfp (B) after 24 h of growth in 1:10 LB. (C) is a combined image of (A) and (B). Magnification: 1000×. Micrographs were taken with appropriate filter sets for each fluorescent protein.

Transcription of lapF is controlled by the alternative sigma factor σS

In order to analyse the expression pattern of lapF, a 411 bp fragment containing the upstream region and the first 32 codons of lapF was PCR-amplified and cloned in plasmid pMP220, as described in Experimental procedures, to generate a transcriptional fusion with the reporter gene lacZ devoid of its own promoter. The resulting plasmid, pMMG1, was electroporated into KT2440, and β-galactosidase activity was followed during growth in liquid LB medium. As shown in Fig. 7, expression of the lapF::lacZ fusion was low in exponential phase and increased significantly upon entry in stationary phase. This prompted us to test if expression from the lapF promoter was dependent on the alternative RNA polymerase sigma factor σS (RpoS), which controls the transcription of a number of genes upon entry in stationary phase and in response to environmental stresses (Hengge-Aronis, 2002; Potvin et al., 2008). Plasmid pMMG1 was introduced in C1R1, a rpoS derivative of KT2440 (Ramos-González and Molin, 1998). Expression of lapF::lacZ was almost completely abolished in this mutant (Fig. 7), indicating that the lapF promoter is directly or indirectly under the control of σS. Similar results were obtained with cultures grown in minimal medium with citrate or glucose as carbon sources, although the maximal level of β-galactosidase activity observed in the wild type was slightly higher than that obtained in LB (Table S3). Strain R6C1, a merodiploid derivative of KT2440 that carries a mutated and a wild-type copy of rpoS in the chromosome (Ramos-González and Molin, 1998), was used to confirm the influence of σS upon lapF::lacZ expression. R6C1 harbouring pMMG1 showed β-galactosidase activity kinetics similar to the wild type (Fig. 7).

Figure 7.

Expression of lapF::lacZ during growth in liquid medium. KT2440 (closed circles), its rpoS derivative C1R1 (open squares), and the merodiploid strain R6C1 (closed squares) harbouring pMMG1, were grown in LB and β-galactosidase activity was followed over time. KT2440 harbouring pMP220 (open circles) is shown as a negative control.

Given the role of LapF in the interaction with plants, the effect of root exudates on expression of the lapF::lacZ fusion was also tested. Corn root exudates were obtained from 4- and 7-day-old plants and added at different concentrations to cultures of KT2440(pMMG1). No significant influence on β-galactosidase activity could be detected with any of the concentrations tested (Table S3).

Expression of lapF in biofilms and the rhizosphere

As a tool to investigate lapF expression in situ, a transcriptional fusion was constructed by cloning the 411 bp fragment mentioned above in pRU1097, which harbours a promoterless gfp gene (Karunakaran et al., 2005). The resulting plasmid, pMMG2, was introduced in KT2440 and expression of the fusion was confirmed in stationary-phase cultures by fluorescence microscopy, whereas no expression was observed in exponentially growing cells (Fig. S2).

Expression of the lapF::gfp fusion was followed in biofilms under static conditions on glass coverslips by fluorescence microscopy. As with the liquid cultures, fluorescence of surface-associated cells was barely noticeable during the first 4 h, afterwards increasing as the biofilm developed (Fig. 8). Interestingly, not all the cells in the population expressed the lapF::gfp fusion, fluorescence concentrating for the most part in densely colonized areas. Similar results were obtained in biofilms grown in flow cells: expression of the lapF::gfp fusion was observed in the more densely colonized areas, particularly where microcolonies had already started to form, while no fluorescence was detected where only a few bacteria were attached to the surface (not shown).

Figure 8.

Expression of lapF::gfp in biofilms of KT2440 grown on glass coverslips under static conditions, after 4, 6 and 24 h. Phase-contrast (left panels) and fluorescence (right panels) microscopy were used to examine the same field. Magnification: 400×.

LapF expression was also tested in the rhizosphere. Aggregates of cells expressing the lapF::gfp fusion could be detected in different colonization sites along the root and on root hairs (Fig. 9). This is consistent with previous data indicating that transcription of the rpoS gene is induced in the rhizosphere with respect to exponentially growing cultures (Matilla et al., 2007).

Figure 9.

Expression of lapF::gfp in KT2440 colonizing alfalfa roots, analysed by fluorescence microscopy. Two root areas are shown: a zone of epidermal cells along the root (left) and a root hair (right). Magnification: 1000×. Bar: 10 µM.

In situ immunodetection reveals specific localization of LapF in P. putida biofilm cells

Taking advantage of the anti-LapF polyclonal antibodies, we decided to observe the localization of the protein in KT2440 biofilms, by immunofluorescence and microscopy. For this purpose, biofilms grown on glass coverslips for 24 h were fixed and incubated with the primary antibodies, followed by incubation with a secondary goat anti-rabbit antibody conjugated with Alexa Fluor 488 fluorophore. Cells were then observed by fluorescence microscopy. As shown in Fig. 10, fluorescence was detected mainly in cells forming part of microcolonies, and not in isolated cells, a result that is consistent with the lapF::gfp expression data. Interestingly, fluorescence staining in KT2440 showed a highly punctuate appearance, with one or two spots per cell, and always localized between cells, so that the outer layer of cells in the microcolonies did not show staining. In agreement with the results obtained by Western blot, very faint fluorescence was detected in the mus-20 mutant after long exposure times (Fig. 10). However, when bacteria were permeabilized before incubation with the anti-LapF antibodies, fluorescence could be observed in mus-20 cells (Fig. 10), confirming that the mutant is unable to transport the protein outside the cell. All these data support a role of LapF in cell–cell interactions and suggest that the protein is concentrated at specific sites on the bacterial surface.

Figure 10.

In situ localization of LapF in KT2440 biofilms by immunofluorescence.
A–F. Biofilms were grown for 24 h on glass coverslips before being fixed and incubated with anti-LapF polyclonal antibodies, followed by incubation with Alexa Fluor 488-conjugated goat anti-rabbit secondary antibodies, and visualization by phase-contrast (A and D) and fluorescence (B and E) microscopy (magnification: ×1000). Images (C) and (F) are composites of (A) and (B), and (D) and (E) respectively.
G and H. Enlarged images showing details of cell clusters and LapF localization in KT2440 (composite images).
I. Detail of a KT2440 cell permeabilized with Triton X-100 before incubation with antibodies.
J and K. Fluorescence images corresponding to mus-20 after identical incubations with anti-LapF and secondary antibody, with (J) or without (K) previous cell permeabilization. Exposure time for fluorescence in (K) was two times longer than that used in all other images.

Anti-LapF antibodies hamper biofilm development

To obtain further evidences of the role of LapF, biofilm formation assays were performed on glass coverslips in the presence of a 1:100 dilution of anti-LapF antibodies. As shown in Fig. 11, addition of the antibodies did not influence initial attachment of KT2440 to the glass surface. However, they obstructed the appearance of three-dimensional microcolonies, which were evident after 5 h of growth in the absence of the antibodies; in their presence, small clusters of cells were only observed after 8 h of growth (Fig. 11), when a dense biofilm was already formed under normal conditions. Thus, the behaviour of the wild-type strain in the presence of anti-LapF antibodies was to a certain extent similar to that of the lapF mutant, adding weight to the involvement of LapF in cell–cell interactions during biofilm development.

Figure 11.

Anti-LapF antibodies hamper microcolony development by P. putida KT2440. Biofilm formation on glass coverslips was followed by phase contrast microscopy, in the absence (left, −Ab) or presence (right, +Ab) of a 1:100 dilution of anti-LapF antiserum. Magnification: 400×.


In this work we have characterized the role of LapF, a large repetitive protein that we have shown is localized on the surface of P. putida cells. As its ‘bigger sister’ LapA, LapF participates in the process of colonization of both abiotic and plant surfaces. However, while LapA determines the progression of the biofilm beyond the initial attachment phase (Hinsa et al., 2003), LapF seems to be required at later stages. Thus, a lapF mutant, while retaining its ability to go into a sessile mode of life, is unable to form a structured biofilm. This role had been previously missed (Espinosa-Urgel et al., 2000) because attachment to abiotic surfaces had only been tested at early time points, when the contribution of LapF appears not to be determinant, or at least is not evident under any of the experimental conditions tested. For this reason, LapF was assumed to participate specifically in bacterial adhesion to plant seeds. We have now presented evidences of its importance for colonization of both abiotic and plant surfaces. This adds a new element to the previously identified functions required for the establishment of P. putida sessile populations regardless of the surface to be colonized, such as the above mentioned adhesin LapA, or HemN-2 a protein involved in haem biosynthesis (Hinsa et al., 2003; Yousef-Coronado et al., 2008). Thus, while bacterial populations associated to plant roots may not always be considered biofilms sensu stricto, colonization of abiotic and plant surfaces have essential mechanistic similarities.

The expression pattern of lapF is consistent with its observed participation in later stages of biofilm development, a role that suggests that the protein is involved in cell–cell rather than in cell–surface interactions. This is supported by the fact that, while the lapF mutant does not form three-dimensional microcolonies on abiotic or biotic surfaces, mixed microcolonies of the wild-type and the mutant strains are observed in competitive colonization assays in the rhizosphere. Microcolonies of the mutant alone are not found in these experiments. Similarly, in mixed culture biofilms the mutant does not form microcolonies, but isolated cells of mus-20 appear interspersed in wild-type microcolonies. LapF could therefore be one of the protein components of the extracellular matrix of P. putida biofilms, helping bacteria to anchor to each other. Protease treatment is known to cause biofilm disruption even in the case of mutants showing defects in biofilm dispersal (Gjermansen et al., 2010; F. Yousef-Coronado and M. Espinosa-Urgel, unpublished), and it also prevents colonization of seeds by P. putida (Espinosa-Urgel et al., 2000). The participation of the LapA protein in the composition of the matrix, with an exopolysaccharide-affixing role, has been recently postulated (Gjermansen et al., 2010), although direct evidence of this function is still lacking. Based on our results and those published previously (Hinsa et al., 2003), we propose that LapA is first required for surface colonization and then LapF, and probably LapA as well, participate in the build-up of a three-dimensional, structured biofilm. The effect of antibodies against LapF on biofilm development, and the results obtained in the immunolocalization assays support this function of LapF as mediator of intercellular adherence. We believe the specific localization of the protein at certain sites on the cell surface is an indication of its architectural role. To our knowledge, this is the first time an element participating in biofilm build-up is thus charted.

The finding that lapF transcription is dependent on the stationary-phase and general stress response sigma factor RpoS was somewhat surprising, since previous evidences indicated that starvation induces dispersal of P. putida biofilms (Gjermansen et al., 2005), and the rhizosphere, where expression of lapF is also observed, is presumed to be a relatively nutrient-rich environment. However, global gene expression analysis of P. putida in the rhizosphere has revealed that plant roots not only provide nutrients but also impose stress on colonizing bacteria, and rpoS transcription is increased in the rhizosphere compared with exponentially growing cultures or cells growing in microcosms in the absence of the plant (Matilla et al., 2007). The role of RpoS in biofilm formation has been studied mainly in Escherichia coli and remains controversial due to conflicting data with respect to the effect of rpoS mutations on biofilm structure and biomass. A recent study (Collet et al., 2008) has shown that a rpoS mutant of E. coli forms biofilms with altered architecture, and that a set of proteins expressed in biofilms are under the control of σS. The contribution of a number of RpoS-regulated genes to the biofilm lifestyle of P. aeruginosa has also been deduced from transcriptomic data (Waite et al., 2006). These and other results (Landini, 2009) reveal that there is an overlap between environmental stress responses mediated by σS and traits important for biofilm formation. In E. coli, this overlap involves the regulatory protein CsgD – which controls the synthesis of the fibre-like appendages known as curli under low temperature conditions – and several genes encoding GGDEF family proteins. These proteins are responsible for the synthesis of the secondary messenger cyclic-di-GMP, the levels of which are known to modulate motility, virulence and the transition between the planktonic state and biofilm formation (recently reviewed by Römling and Simm, 2009, and by Hengge, 2009) in different bacteria. Thus, a complex regulatory network modulates bacterial commitment to sessile life, and is likely to lie also underneath the physiological heterogeneity observed in biofilm cells (Spormann, 2008). In this respect, it is worth noting that differences in expression of a lapF::gfp fusion were observed among cells in different areas of P. putida biofilms, while more or less uniform expression could be detected in stationary-phase liquid cultures (Fig. S2). Since transcription from the lapF promoter seems to be strictly dependent on RpoS, one possibility is that RpoS is itself differentially regulated in various microenvironments within the biofilm, or at different developmental stages. Alternatively, an additional, biofilm-specific factor could also participate in the regulation of lapF transcription, or in the localization of the protein. In S. aureus, a phase variation mechanism appears to contribute, besides other regulatory elements, to control expression of the large repetitive adhesin Bap (Tormo et al., 2007). It will be interesting to check if a similar phenomenon could be taking place in P. putida biofilms.

There is currently an ongoing debate regarding whether biofilm formation can be considered a true multicellular developmental process (Monds and O'Toole, 2009), a model that has been generally favoured in the past decade and is now questioned by some authors. The sequential requirement of LapA and LapF at different stages of biofilm formation by P. putida would be in agreement with a developmental model, resembling the chronological checkpoints that exist in established examples of microbial development, such as fruiting body formation by Myxococcus or sporulation in Bacillus (Kroos, 2007). On the other hand, the fact that expression of lapF is under the control of the global transcriptional regulator σS, an alternative sigma factor involved in a variety of starvation- and stress-related processes, seems to speak against lapF being part of a dedicated biofilm regulatory programme. Yet, considering that LapF appears to mediate cell–cell interactions, its σS-dependent expression in stationary phase outside of a biofilm might be expected to result in the formation of bacterial aggregates in the liquid medium, a phenomenon that is not normally observed under standard culture conditions despite the presence of the protein on the bacterial surface. It is possible that post-translational modifications of LapF or its interaction with other extracellular components determine the ultimate function of the protein in different situations. Thus, it could be argued that for LapF to exert its intercellular linkage role, bacteria must first be associated to a surface and hence have gone through the previous stages of biofilm formation. Such history-dependent operation is compatible with a developmental process, although other possibilities cannot be excluded. An in-depth analysis of the temporal and/or environmental control of LapF transport and functionality, and how these are integrated in the whole process of biofilm formation will contribute further to a mechanistic portrayal of the sessile lifestyle of P. putida.

Experimental procedures

Strains, plasmids and growth conditions

Pseudomonas putida KT2440 is a plasmid-free derivative of P. putida mt-2, originally isolated from a vegetable orchard in Japan (Nakazawa, 2002). Strains mus-20 (lapF mutant) and mus-42 (lapA mutant) were obtained by random transposon mutagenesis with mini-Tn5[Km1] and identified as defective in attachment to corn seeds (Espinosa-Urgel et al., 2000; Yousef-Coronado et al., 2008). C1R1 is a rpoS null derivative of KT2440, while R6C1 is a merodiploid carrying an intact and a mutated copy of rpoS in the chromosome (Ramos-González and Molin, 1998). KT2440-Sm, a streptomycin-resistant derivative of KT2440 obtained by site-specific insertion of mini-Tn7ΩSm, has been described elsewhere (Yousef-Coronado et al., 2008). Other mini-Tn7 derivatives (Koch et al., 2001) were used to obtain fluorescently labelled strains KT2440dsRed[Gm], KT2440gfp[Km] and mus20gfp[Gm] by conjugation. E. coli DH5α was routinely used in cloning experiments. Expression vectors pMP220 (Spaink et al., 1987) and pRU1097 (Karunakaran et al., 2005) were used to construct pMMG1 (lapF::lacZ) and pMMG2 (lapF::gfp) respectively (see below). E. coli was routinely grown at 37°C in LB. P. putida strains were grown at 30°C either in LB or in M9 minimal medium with MgSO4, Fe-citrate and trace metals (Yousef-Coronado et al., 2008), and glucose (20 mM) or citrate (15 mM) as carbon and energy source. When appropriate, antibiotics were added at the following concentrations: kanamycin (Km) 25 µg ml−1; tetracycline (Tc) 15 µg ml−1; streptomycin 50 µg ml−1; gentamicin (Gm) 10 or 100 µg ml−1.

Molecular biology techniques

Plasmid DNA isolation, PCR amplification, digestion with restriction enzymes, gel electrophoresis and DNA ligation were done using standard protocols (Ausubel et al., 1987). Plasmids pMMG1 and pMMG2 were constructed by PCR amplification of a 517 bp fragment that includes the intergenic region between PP_0807 and lapF, using primers LAPF1 (5′-GCGAATTCCAGCGACAGGTGATCGAAG-3′) and LAPF2 (5′-GTCCACGGCGAAGAAGTTAC-3′). The PCR product was digested with EcoRI (underlined in LAPF1) and KpnI, and the resulting 411 bp fragment was cloned in pMP220 and pRU1097. The absence of mutations in the cloned fragment was confirmed by sequencing.

To obtain antibodies against LapF, the Abie Pro 3.0: Peptide Antibody Design software (Chang Bioscience) was used to select an antigenic peptide (GRGEAGATVEVRNDQG), which corresponds to the best conserved region common to the 64 imperfect repeats of LapF. This peptide was synthesized and conjugated to Klh, and used to immunize SPF New Zealand rabbits. Immunization and antisera recovery were performed at the Center for Animal Production of the Universidad de Sevilla.

For subcellular fractionation cells were grown overnight at 30°C and B-PER Bacterial Protein Extraction Reagent (Pierce) was used for separating the soluble (cytosolic) fraction from 1.5 ml of bacterial culture at an OD600 of 2. We followed a modified procedure developed by Cheng et al. (1973) to obtain outer membrane proteins. Two millilitres of overnight cultures were centrifuged and resuspended in a SL solution (20% sucrose; 1 mg ml−1 lysozyme; 0.01 M Tris 8.4) and incubated at 25°C for 1 h. After 10 min at 8000 r.p.m. cells were resuspended with a Tris-Mg solution (0.01 M MgCl2; 0.01 M Tris pH 8.4), incubated at 25°C for 20 min and centrifuged. The supernatant containing surface proteins was then collected. Cell fractions were quantified and detected after 5% SDS-PAGE. Western blot analysis of proteins was performed following standard techniques (Ausubel et al., 1987), after gel transfer onto PVDF membranes. Membranes were incubated with LapF-specific polyclonal antiserum (1:4000) and detection was done with anti-rabbit horseradish peroxidase-conjugated secondary antibody (Bio-Rad), using a chemiluminescence-based detection system (Bio-Rad). A control experiment where incubation with the anti-LapF antibodies was omitted was also carried out to ensure the absence of non-specific binding of the secondary antibodies to the samples.

Rhizosphere colonization assays

Competitive colonization assays were performed as previously described (Yousef-Coronado et al., 2008), using 1:1 mixes of strains KT2440-Sm and mus-20, or each strain individually, for quantitative assays, and KT2440dsRed and mus-20gfp for visualization by fluorescence microscopy. For quantitative analysis of colonization, roots were cut, weighed and introduced in tubes with 20 ml of M9 basal medium and 4 g of glass beads (diameter 3 mm). Bacteria attached to the roots were recovered by vortexing for 2 min, and their numbers were calculated by plating serial dilutions of the resulting suspension on selective medium (M9 with citrate and Sm or Km). Alternatively, inspection of KT2440dsRed and mus-20gfp cells colonizing the root surface was performed on a Zeiss Axioscope fluorescence microscope with appropriate filter sets, coupled to a Nikon DS5-Mc CCD camera.

Biofilm formation analysis

Biofilm formation was examined during growth in polystyrene microtitre plates (Sterilin) or in borosilicate glass tubes, as described previously (Yousef-Coronado et al., 2008). Biomass attached to the surface was visually inspected by staining with crystal violet, and quantified after solubilizing the dye with 70% ethanol and measuring absorbance at 590 nm (O'Toole and Kolter, 1998).

Microscopy analysis of biofilm formation on glass was analysed by obliquely placing 40 mm glass coverslips into the wells of a six-well plate where cultures were allowed to grow in LB or minimal medium at 30°C. Visualization was performed as described in the previous section. A modification of the method described by Gjermansen et al. (2005) was used for biofilm formation under flow conditions, using LB diluted 1:10 as growth medium. Biofilms were grown at 30°C in three-channel flow chambers (BioCentrum-DTU, Technical University of Denmark), using a Watson-Marlow 205S peristaltic pump (Watson-Marlow, Wilmington, MA). Fresh overnight cultures were diluted to an OD600 = 0.5 and 300 µl were injected in the flow chamber. During the first hour, the flow was turned off in order to allow cells attach to the chamber; then, the flow was turned on and kept at a constant rate of 3 ml h−1 (laminar flow conditions). Biofilm structures were visualized with a Nikon C1 confocal laser scanning microscope. Images were analysed with Imaris software (Bitplane), and biofilm parameters (biovolume, surface coverage, average thickness and maximum thickness) were calculated using comstat (Heydorn et al., 2000).

Collection of corn root exudates

Corn seeds were sterilized as above and hydrated for 48 h at 30°C in Phytagel (Sigma). After incubation, seeds were placed on a plastic floating structure in contact with 400 ml of sterile deionized water and Fe-EDTA, within a transparent container, where plants grew hydroponically in a sterile environment. Exudates released to the liquid were collected after 4 or 7 days, filtered into glass beakers and freeze-dried. Exudates were kept at −80°C, and dissolved in 50% methanol immediately before use.

Measurement of β-galactosidase activity

β-Galactosidase activity in P. putida cultures harbouring pMP220 or pMMG1 was assayed as previously described (Espinosa-Urgel and Ramos, 2004). Experiments were repeated three times and results shown correspond to a representative experiment. Activity values are given in Miller units (Miller, 1972).


LapF immunodetection was performed as described by Schmidt-Eisenlohr et al. (2001) with some modifications. Cells from static biofilms formed after 24 h of growth in 1:10 LB were first washed with PBS and fixed with cold paraformaldehyde (4% v/v) for 30 min at 4°C. After three subsequent washes in PBS, samples were incubated for blocking 30 min at room temperature with 1% BSA (w/v). Alternatively, cells were permeabilized with Triton X-100 prior to the blocking step. After three more washes in PBS, cells were incubated at 4°C overnight with primary anti-LapF antiserum diluted 1:100. Samples were then washed with PBS and incubated with goat anti-rabbit Alexa Fluor 488 secondary antibodies (Sigma) for 3 h at room temperature, washed three times with PBS and treated with antifade reagent (Citifluor, Sigma) before visulization by fluorescence microscopy. A control omitting anti-LapF antibodies was also carried out to ensure the absence of non-specific binding of the secondary antibodies to the samples.


We thank S. Molin and M.I. Ramos-González for providing strains and plasmids for fluorescence tagging and expression analysis; J.D. Alché for help with confocal microscopy; G. O'Toole and M.I. Ramos-González for critically reading the manuscript; and two anonymous reviewers for thorough revision and suggestions for improvement of the article. This work was funded by Grant BFU2007-64270 from the Plan Nacional de I+D+i and FEDER. M.M.-G. and F.Y.-C. are recipients of FPI fellowships from Ministerio de Ciencia e Innovación.