Interrelationships between Colonies, Biofilms, and Planktonic Cells of Pseudomonas aeruginosa

Strain and growth conditions. Pseudomonas aeruginosa PAO1 was obtained from Barbara Iglewski, University of Rochester (Rochester, NY). All cells were grown at 37°C in AGSY medium (56 mM alanine, 17 mM K₂HPO₄, 86 mM NaCl, 100 μM CaCl₂, 10 mM MgSO₄, 5 μM FeCl₂, 7.5 μM ZnCl₂, 0.5% [vol/vol] glycerol, 3 g/liter yeast extract, pH 7). Planktonic cells were grown in baffled flasks with vigorous shaking and harvested after 3 h (late exponential phase; optical density at 600 nm [OD₆₀₀], ∼1) or 9 h (stationary phase; OD₆₀₀, ∼9). Colonies were grown on 1.5% AGSY agar. After 15 or 40 h, cells were harvested by flooding the plates with ice-cold TE lysis buffer (50 mM Tris-HCl, 4 mM EDTA, pH 8.3) and scraping off the colonies with a glass spreader. After 15 h, the colonies were white and flat (0.5- to 1-mm diameter), whereas after 40 h they were pigmented and differentiated (5- to 7-mm diameter). Biofilms were inoculated with 1 ml of an overnight culture (OD₆₀₀ diluted to 1) and grown in silicone tubes (15-cm length, 6-mm internal diameter) with a continuous flow of 50 ml per hour (variation, ≤5%). This corresponds to laminar flow (Reynolds number 3.3). The tubes were placed vertically in order to avoid sedimentation of planktonic cells. Biofilms were grown for 3 or 5 days and harvested by squeezing out the accumulated material. In this growth medium, the 3-day-old biofilms represented the earliest point at which we could reliably obtain sufficient material for proteomic analysis. After this, the biofilms grew rapidly and were expected to have reached maturity by day 5. By day 6, the biofilms often began to disperse and peel off from the inside of the tube. All cells were sedimented by centrifugation (10 min, 3,000 × g, 4°C) and stored at −80°C. The biofilm pellets were larger after 5 days than after 3 days, but there was no significant difference in the numbers of CFU.

Protein extraction.Pellets were resuspended in TE lysis buffer supplemented with 0.1 mM phenylmethylsulfonyl fluoride and Complete Mini-Protease inhibitor cocktail (Roche). Cells were disrupted by sonication on ice, and debris was removed by centrifugation (30 min, 13,000 × g, 4°C). Proteins were extracted with phenol using an adaptation of the method described by Hurkman and Tanaka (13). TE-saturated phenol (1 volume) was added to the cleared cell lysates; samples were vortexed five times for 1 min each time and kept on ice for 1 min between the rounds. The layers were separated by centrifugation (10 min, 1,600 × g), and the top aqueous layer was discarded. Phenol-saturated TE (2 volumes) was added, samples were vortexed and centrifuged as before, and the aqueous layer was discarded. We found that this phenol extraction step was crucial in order to achieve consistent reproducible separation of P. aeruginosa proteins on the 2D gels. Proteins were precipitated by adding 5 volumes of 100 mM ammonium acetate in 100% methanol and leaving the samples at −20°C overnight. Pellets were washed, first with 80 mM ammonium acetate in 80% methanol and second with 80% acetone, and then air dried. Proteins were redissolved in ASB14 buffer (8 M urea, 2% [wt/vol] amido sulfobetaine 14, 5 mM magnesium acetate, 20 mM Tris-HCl, pH 8.5), quantified using the Bio-Rad DC protein assay kit by following the supplier's protocol, and adjusted to a total protein concentration of 5 mg/ml with ASB14 buffer.

Proteomics and data analysis.We analyzed four biological replicates for each of the six conditions examined, i.e., 24 samples in all (14). Proteins were minimally labeled using charge neutral CyDye DIGE Fluor minimal dyes (GE Healthcare). Two samples from each of the growth conditions were labeled with Cy3 and two with Cy5 in order to avoid potential dye-related artifacts. A pooled internal standard consisting of equal amounts of all 24 protein samples was generated and adjusted to a final protein concentration of 5 mg/ml with ASB14 buffer. The standard was labeled with Cy2 and included in all gels. This ensured that all proteins were represented on every gel, allowing direct inter- and intragel comparison of protein spots. Furthermore, this procedure also made it possible to distinguish between spots that were absent because of differences in the biological samples or because of technical differences between the gels. Fifty micrograms of each sample or pooled internal standard was labeled with Cy3, Cy5, or Cy2 (200 pmol) for 30 min, and the labeling reaction was quenched by adding lysine (50-fold excess). Following this, samples were combined (details available upon request) so that each gel contained two samples (one labeled with Cy3 and one with Cy5) and a Cy2-labeled standard. The first dimension was run on 24-cm isoelectric focusing DryStrips (pH 4 to 7; GE Healthcare), and the second dimension was run on Ettan DALT 12% sodium dodecyl sulfate-polyacrylamide gels. Gels were scanned at the appropriate wavelengths for excitation of Cy3, Cy5, and Cy2 using a Typhoon 9400 scanner, and the resulting 36 images (3 for each of the 12 gels) were analyzed using the DeCyder software package (GE Healthcare). A typical image of a Cy2-labeled internal standard was chosen as a master gel to which spots from the remaining 11 Cy2-labeled images were matched using DeCyder BVA 5.02. Following this, a univariate BVA was performed as previously described (1, 33) using Student's t test (P ≤ 0.01). Based on data generated by DeCyder BVA, a principal-component analysis (PCA) was performed using SimcaP 10.0 (Umetrics). PCA utilizes the BVA output data. However, unfiltered BVA data may contain noise that can easily be excluded by visual inspection in the univariate analysis but which, if not removed, can interfere with multivariate analysis. Therefore, the data were filtered prior to PCA such that only those spots which were resolved in at least 75% of all gels and had a pixel intensity of ≥1 × 10⁴ were included. The filtered data set was also subjected to a hierarchical cluster analysis (HCA) using Pearson's correlation in Java TreeView 1.0.12 (jtreeview.sourceforge.net). Preparative gels were loaded with 200 μg of unlabeled pooled internal standard and then fixed (45% methanol, 1% acetic acid) and stained with Coomassie blue. Proteins of interest were excised and subjected to tryptic digestion and analyzed by liquid chromatography (LC)-tandem mass spectrometry (MS-MS). Peptides were separated by reverse-phase nano-high-performance LC (CapLC pump from Waters; 15-cm PepMap C₁₈ column with a 75-μm inside diameter from LC Packings) on a capillary LC system attached to a model QTof2 mass spectrometer (Waters). Peptides were eluted with a flow rate of 2.95 μl/min in a gradient of 5 to 55% acetonitrile over 33 min and 55 to 95% over 5 min. The MS-MS fragmentation data were used to search the NCBI database (www.ncbi.nlm.nih.gov ) using the MASCOT search engine (www.matrixscience.com ).

Biological variance analysis and Student's t test.Between 1,279 and 1,606 protein spots in the Cy2-labeled standards were matched to the master gel, an image of which is shown in Fig. S1 in the supplemental material. Following gel-to-gel matching, each protein spot (with a minimum of three data points for each of the relevant conditions) was analyzed for significant differences between the six groups of samples using Student's t test with a threshold set at a P of ≤0.01 (Table 1). The highest number of significant changes (358 spots) was found in a comparison of PE and PS. The smallest number of differences was found between B3d and B5d (46 spots), followed by PE and C15h (106 spots). Furthermore, PS samples were more similar to C40h than to any other group of samples. Biofilms, irrespective of age, were more similar to PE and C15h than to PS and C40h. This is in contrast to previous studies that have suggested a high degree of similarity between biofilms and stationary-phase planktonic cells (12, 35).

PCA and HCA.In order to validate the results from the univariate analysis, we employed two multivariate analytical techniques. PCA is frequently used to identify patterns in data of high dimension in situations where graphical representation is not readily available, whereas HCA sorts samples into groups based on degree of similarity. The analyses were unsupervised, meaning that the calculations were performed without any information on the nature of the samples. The filtered data set used for multivariate analyses (see Materials and Methods) contained 1,072 variables, with 24 observations for each.

First, PCA was used to identify the largest variations in the data set. This analysis is based upon the generation of a covariance matrix describing how each spot varies with respect to every other spot in the data set. Eigenvectors, along with their corresponding eigenvalues, can be calculated for the covariance matrix, describing correlated variation within the data set. The eigenvector with the highest eigenvalue represents the largest variation in the data set and is the first principal component (PC1). Similarly, the eigenvector with the second-highest eigenvalue is PC2 and so on. In this way, the complexity inherent in this high-dimensional data set can be reduced to a relatively small number of ranked orthogonal PCs (29). A score plot of PC1 versus PC2, which accounted for 48% of the total variation in the data set (29% for PC1 and 19% for PC2), clustered samples from the six different groups (Fig. 1). PC1 separated PS and C40h from the other samples. The effect of the incubation time in planktonic cells and colonies therefore appeared to account for the largest variation in the conditions studied. PC2 separated biofilms from both planktonic cells and colonies, indicating a different protein profile in these samples. Interestingly, the biofilm samples were closer to the actively growing PE and C15h than to PS and C40h samples.

• Open in new tab
• Download powerpoint

FIG. 1.

PCA score plot of PC1 versus PC2. PC1 accounted for 29% of the variation in the data set, whereas PC2 accounted for another 19%. Shown are results for planktonic cells in exponential phase (⋄) or in stationary phase (♦), colonies incubated for 15 h (▵) or 40 h (▴), and biofilms incubated for 3 days (□) or 5 days (▪).

In order to confirm the clustering of samples, especially regarding those that appeared to be closely related, the same data set was subjected to an HCA (Fig. 2A). This analysis confirmed that the major split in the dendrogram was between biofilms, PE and C15h on one hand and PS and C40h on the other hand. The analysis also showed that the biofilms had expression profiles that were different from those of both colonies and planktonic cells and that these were more closely related to PE and C15h than to PS and C40h.

• Open in new tab
• Download powerpoint

FIG. 2.

(A) HCA of data from DeCyder BVA based on the complete filtered data set of 1,072 discrete variables. (B) Proteins identified as modulated between selected conditions. All proteins identified as “single hits” are displayed. The master number is assigned by the DeCyder software and refers to the position of the protein spot on the master gel (Fig. S1 in the supplemental material). An overview of the identified proteins and the corresponding MS-MASCOT data are shown in Table S2 in the supplemental material. The n-fold change refers to the first condition compared with the one in brackets; e.g., negative values in the column headed by “PS(PE)” refer to a down-regulation in PS compared with expression in PE. Proteins that are among the 15 spots with the highest change (n-fold) in the conditions compared are marked with an asterisk. In all cases, the changes are significant, with a P of ≤0.01. Pearson's correlation coefficient indicates the correlation between variables or groups of variables and is shown for the most similar expression profiles. A hierarchical cluster analysis of protein spots in the full data set divided them into three categories, which are separated by bold lines. (C) Heat plot of the identified proteins showing the variation in expression profiles across all 24 samples.

Protein identification.Between PE and PS, approximately one-third of the reliably detected proteins were found to be modulated (358 spots out of 1,072 detected in ≥75% of the samples [Table 1]). Furthermore, the PCA plot (Fig. 1) indicated that the effects of the incubation time were similar in planktonic cells and in colonies. Indeed, of the 209 proteins that were modulated between C15h and C40h, 122 were also found to be modulated between PE and PS, and 118 of these changed in the same direction. However, PC2 indicated that there were differences between B3d, PE, and C15h other than those related to incubation time. Between these three groups, as well as between PE and PS, we attempted to identify those proteins which exhibited the highest change (n-fold). Some protein spots yielded a list of peptides that matched more than one protein in the P. aeruginosa database. This could be due to spot comigration or incomplete separation of proteins on the gels. In the case of such mixed hits, we were unable to account for which of the two or more proteins was responsible for the observed modulation, so these have been excluded from further discussion. Single hits, along with their level of change (n-fold) between selected conditions are presented in Fig. 2B. Furthermore, a selection of modulated proteins had been identified in a previous study under the same experimental conditions, and these are also shown in Fig. 2B. All single hits are presented in Fig. 2B, and a list of these proteins with the corresponding MS-MASCOT data are shown in Table S2 in the supplemental material.

In order to group the identified proteins into categories, an HCA was performed on the filtered data set of 1,072 protein spots. This analysis segregated the proteins into three main categories (Fig. 2B). Pearson's correlation coefficient is given for the most similar expression profiles, and expression patterns of the identified proteins across all 24 samples are shown in a heat plot (Fig. 2C). The three main categories were characterized by (i) proteins that were highly expressed in C15h and PE, (ii) proteins that were highly expressed in biofilms, or (iii) proteins that were highly expressed in PS and C40h. In category 1, the majority of the proteins were related to metabolism, transcription, and translation. Category 3 consisted mainly of proteins previously identified to be regulated by quorum sensing, RpoS, or both. Category 2 comprised proteins involved in diverse physiological processes, ranging from metabolism, adaptation/protection, and attachment through to secreted factors. Furthermore, two Usp-type stress proteins, PA4352 and PA3309, which have been previously shown to be induced under anaerobic conditions (4, 26), were highly expressed in biofilms and C40h. This could indicate an anaerobic environment in these growth modes.