Parasitic growth of Pseudomonas aeruginosa in co-culture with the chitinolytic bacterium Aeromonas hydrophila

Authors


E-mail bodo.philipp@uni-konstanz.de; Tel. (+49) 7531 884541; Fax (+49) 7531 884047.

Summary

Polymer-degrading bacteria face exploitation by opportunistic bacteria that grow with the degradation products without investing energy into production of extracellular hydrolytic enzymes. This scenario was investigated with a co-culture of Aeromonas hydrophila and Pseudomonas aeruginosa with chitin as carbon, nitrogen and energy source. In single cultures, A. hydrophila could grow with chitin, while P. aeruginosa could not. Co-cultures with both strains had a biphasic course. In the first phase, P. aeruginosa grew along with A. hydrophila without affecting it. The second phase was initiated by a rapid inactivation of and a massive acetate release by A. hydrophila. Both processes coincided and were dependent on quorum sensing-regulated production of secondary metabolites by P. aeruginosa. Among these the redox-active phenazine compound pyocyanin caused the release of acetate by A. hydrophila by blocking the citric acid cycle through inhibition of aconitase. Thus, A. hydrophila was forced into an incomplete oxidation of chitin with acetate as end-product, which supported substantial growth of P. aeruginosa in the second phase of the co-culture. In conclusion, P. aeruginosa could profit from a substrate that was originally not bioavailable to it by influencing the metabolism and viability of A. hydrophila in a parasitic way.

Introduction

In nutrient-limited environments competition for substrates inevitably enforces interspecific interactions among bacteria. Such interactions can particularly be expected during utilization of polymeric organic compounds, which is related to the fact that their degradation starts as an extracellular process. Bacterial degradation of polymers, such as polysaccharides like chitin or cellulose, is generally initiated by extracellular hydrolytic enzymes, causing the release of oligomers and monomers, which are subsequently taken up and further metabolized by the bacterial cells (Keyhani and Roseman, 1999; Lynd et al., 2002). This common and widespread degradation strategy implies a risk for those bacteria that produce extracellular hydrolytic enzymes, because the resulting hydrolysis products are also available for bacteria that did not produce these enzymes. In consequence, the enzyme-producing bacteria (investors), which had invested energy into protein biosynthesis, may not fully benefit from their investment, while the other bacteria obtain nutrients without this energy investment (opportunists). Investors and opportunists must have strategies to secure degradation products for growth. For investor bacteria, this could be achieved by a tight coupling of polymer degradation with the uptake of oligo- and monomers. Alternatively, investor bacteria could also actively suppress growth of opportunistic bacteria by bioactive compounds. Opportunistic bacteria may acquire growth substrates by a broad metabolic flexibility and versatility to utilize any exudation from polymer-degrading investor bacteria, by very efficient uptake systems for degradation products, or by inhibition of polymer-degrading investor bacteria. The opportunistic bacteria must accomplish a trade-off of exploitation of and stable co-existence with the investor bacteria to allow it to initiate polymer degradation, because otherwise growth of the opportunists would also be suppressed.

This investor-opportunist scenario could play an important ecological role in oligotrophic aquatic systems, where polymeric organic compounds constitute a major portion of organic matter, and are, thus, an important nutrient source for heterotrophic bacteria (Unanue et al., 1999). In aquatic systems, extracellular degradation is often the rate-limiting step in the degradation of polymers (Chróst and Rai, 1993). Interactions between investors and opportunists could therefore be a relevant factor for the rates of organic matter turnover and of bacterial biomass production as well as for the composition of the community of heterotrophic bacteria in oligotrophic aquaticsystems. Despite the obvious ecological importance and the great general interest in interspecific bacterial interactions (Ryan and Dow, 2008; Haruta et al., 2009; Straight and Kolter, 2009; Hibbing et al., 2010), interactions between investors and opportunists during polymer degradation have not been subject of detailed studies. Thus, our goal was to establish an appropriate model system for studying interactions of this kind.

For this, we set up a defined co-culture with Aeromonas hydrophila strain AH-1N as the investor and Pseudomonas aeruginosa strain PAO1 as the opportunist and with chitin as the sole source of carbon, nitrogen and energy. Chitin, a linear polysaccharide of β-(1,4)-linked N-acetylglucosamine (GlcNAc) residues, is the most abundant polymer in aquatic systems (Gooday, 1990; Pruzzo et al., 2008). Aeromonads are aquatic Gammaproteobacteria, and many of them, such as A. hydrophila, are chitinolytic and employ extracellular chitinases for the breakdown of chitin (Li et al., 2007; Lan et al., 2008). In this respect, A. hydrophila is appropriate for the investor role. Pseudomonas aeruginosa is reported not to grow with chitin (Folders et al., 2001), although it produces a chitinase (ChiC; PA2300) and a chitin-binding protein (CbpD; PA0852) (Folders et al., 2000; Nouwens et al., 2003; Kay et al., 2006; Manos et al., 2009). Pseudomonas aeruginosa is metabolically versatile (Clarke, 1982; Mena and Gerba, 2009), and it is known to inhibit other bacteria by quorum sensing-regulated secondary metabolites during competition for nutrients (Norman et al., 2004; Straight and Kolter, 2009). In these respects, P. aeruginosa has the potential for an efficient opportunist in our scenario. As both bacteria occur in freshwater habitats and prefer similar environmental conditions, such as moderate temperature and neutral pH values, they are likely to encounter each other (Janda, 1985; von Graevenitz, 1987; Römling et al., 1994; van Asperen et al., 1995; Hardalo and Edberg, 1997; Mena and Gerba, 2009).

In summary, our model system with these two well-characterized bacteria has ecological relevance and offers the opportunity to investigate interspecific interactions at the physiological and molecular level. We initiated our study by investigating whether P. aeruginosa could benefit from chitin degradation by A. hydrophila.

Results

Growth in single culture

To characterize chitin degradation of A. hydrophila strain AH-1N and to verify that P. aeruginosa strain PAO1 cannot grow with chitin, both strains were incubated with chitin as a sole source of carbon, nitrogen and energy in single culture. Colony-forming units (cfu) of strain AH-1N increased concomitantly with chitin degradation reaching a final number of about 109 cfu ml−1 within 1 day. Colony-forming units of strain PAO1 increased only to a number of about 3 × 106 cfu ml−1, and chitin did not decrease (Fig. 1).

Figure 1.

Growth of A. hydrophila strain AH-1N and of P. aeruginosa strain PAO1 in single cultures with chitin. Colony-forming units (cfu) of strain AH-1N (●); cfu of strain PAO1 (▴); decrease of chitin in cultures of strain AH-1N (○) and of strain PAO1 (▵). Error bars indicate standard deviation (n = 3).

If chitin was incubated in cell-free supernatant of chitin-grown strain AH-1N containing chitinolytic enzymes, GlcNAc, (GlcNAc)2 and (GlcNAc)3 accumulated indicating that these were the primary products of chitin degradation. To investigate whether both strains could grow with these degradation products of chitin, they were incubated with GlcNac, (GlcNAc)2 and (GlcNAc)3. Additionally, we tested growth of both strains with the further plausible degradation products glucosamine and acetate. As expected strain AH-1N was able to grow with all of these compounds (data not shown). Strain PAO1 could not grow with (GlcNAc)2 and (GlcNAc)3, but was able to grow with GlcNAc, acetate and glucosamine as substrates, with the latter two only if ammonium was present in the medium. The growth rate of strain PAO1 with GlcNAc was significantly lower than the growth rate of strain AH-1N (data not shown). While strain PAO1 did not degrade chitin, we observed that more than 80% of the cells in a cell suspension attached to chitin within 30 min. In that timeframe, about 60% of the cells in a cell suspension of strain AH-1N attached to chitin.

Growth in co-culture

To investigate whether strain PAO1 was able to benefit from chitin degradation by strain AH-1N, a co-culture with both strains was incubated with chitin as sole source of carbon, nitrogen and energy. In this co-culture chitin was degraded within 3 days, and cfu of strain AH-1N increased to numbers similar to its growth in single culture (Fig. 2). After day 2, however, cfu of strain AH-1N decreased dramatically and dropped below the detection limit of 105 cfu ml−1. Coinciding with this decrease, the culture turned dark green and foamy. At day 1 cfu of strain PAO1 had increased slightly higher than in single culture reaching numbers of about 107 cfu ml−1. After day 2, cfu of strain PAO1 increased further reaching numbers of about 109 cfu ml−1. Thus, in contrast to the single culture, strain PAO1 could grow in the co-culture, in which chitin had been provided as the sole carbon, nitrogen and energy source. This growth subsequently followed the decrease of cfu of strain AH-1N and the formation of green colour and foam. While the time point of these events varied between day 2 and day 3, the overall course of the co-culture was always reproducible. A further observation was that co-cultures, in which cfu of strain AH-1N had already decreased, contained red-coloured cells.

Figure 2.

Growth of A. hydrophila strain AH-1N and of P. aeruginosa strain PAO1 in co-cultures with chitin. Colony-forming units (cfu) of strain AH-1N (●); cfu of strain PAO1 (▴); decrease of chitin (□). Error bars indicate standard deviation (n = 3).

To investigate whether the chitinase ChiC and the chitin-binding protein CbpD as well as the sugar-binding proteins LecA and LecB were required for growth of strain PAO1 in the co-culture, respective mutants were tested. Co-cultures with these mutants did not differ from co-cultures with the wild type. Additionally, binding to chitin was not affected in strains PAO1-KO[cbpD], -KO[chiC], ΔlecA and ΔlecB.

To investigate whether the course of the co-culture with chitin was a general consequence of co-cultivation of strains AH-1N and PAO1, chitin was replaced by tryptone. In this co-culture both strains grew to numbers of about 5 × 108 cfu ml−1 (data not shown). In stationary phase cfu of strain AH-1N decreased only slightly, and the culture did not turn green. However, if the co-culture was incubated with chitin and tryptone, chitin degradation was accelerated, and within 24 h the culture had turned green, cfu of strain AH-1N were below the detection limit of 105 cfu ml−1, and strain PAO1 reached about 2 ×  109 cfu ml−1. The same number was reached in the single culture of strain PAO1 with tryptone and chitin, in which chitin was not degraded, and which did not turn green.

The course of the co-culture with chitin raised two questions. First, which substrates supported growth of strain PAO1, and, second, what caused the decrease of cfu of strain AH-1N in the co-culture?

Analysis of metabolite production in the co-culture

To address the first question we analysed supernatant samples from co-cultures of strains AH-1N and PAO1 and from single cultures of strain AH-1N growing with chitin by ion-exclusion and reversed-phase HPLC. Additionally, these samples were tested for the presence of ammonium. Ion-exclusion HPLC revealed that in single culture of strain AH-1N up to 5 mM acetate was released and subsequently taken up between 15 and 60 h of incubation. In the co-culture, the release of acetate started slowly around 30 h of incubation, increased within the next 30 h and peaked after 60 h of incubation, reaching concentrations of up to 10 mM. This peak was followed by a fast degradation of acetate within 10–20 h. Remarkably, the release of acetate coincided with the inactivation of strain AH-1N, and the degradation of acetate coincided with the increase of cfu of strain PAO1 (Fig. 3A and B). GlcNAc, (GlcNAc)2 and (GlcNAc)3 were neither detected in the single culture nor in the co-culture samples. With reversed-phase HPLC only traces of glucosamine (below 50 µM) could be detected in single and co-culture samples. In single culture of strain AH-1N up to 10 mM ammonium were released and partly taken up between 20 and 50 h of incubation (data not shown). In the co-culture about 3 mM of ammonium were detected after 20 h of incubation. This concentration level remained constant during further incubation. These results clearly showed that growth of strain PAO1 after 70 h was substantially supported by acetate as carbon and by ammonium as nitrogen source.

Figure 3.

Metabolite production in single cultures of A. hydrophila strain AH-1N and in co-cultures of A. hydrophila strain AH-1N and P. aeruginosa strain PAO1 with chitin.
A. Colony-forming units (cfu) of strain AH-1N in single culture (●), cfu of strains AH-1N (○) and PAO1 (▵) in co-culture.
B. Acetate concentration in supernatants of single cultures of strain AH-1N (□) and of co-cultures of strains AH-1N and PAO1 (inline image).
C. Pyocyanin (inline image) and HHQ (▴) concentration in supernatants of co-cultures of strains AH-1N and PAO1.
Error bars indicate standard deviation (n = 3). In (B) the three replicates for the co-culture are shown individually.

Analysis of inactivation of strain AH-1N and identification of secondary metabolites

To address the question of what caused the decrease of cfu of strain AH-1N in co-culture with strain PAO1, we investigated whether this decrease was dependent on the presence of strain PAO1 cells or not. Therefore, we incubated cell suspensions of strain AH-1N in cell-free supernatants from co-cultures, in which cfu of strain AH-1N had already decreased. This cell-free supernatant caused a more than 1000-fold decrease of cfu of strain AH-1N within 48 h, while the total cell counts remained constant over time (Fig. 4A). This indicated that the co-culture supernatant contained toxic compounds that causedinactivation but not lysis of strain AH-1N. Notably, these inactivated cells were red-coloured. Cells of strain PAO1 were not inactivated in co-culture supernatant and did not turn red. This indicated that the aforementioned red-coloured cells in the co-culture were cells of strain AH-1N. In the presence of 106 cfu ml−1 of strain PAO1, cfu of strain AH-1N decreased below the detection limit of 105 cfu ml−1 within 20 h (Fig. 4A). This indicated that cells of strain PAO1 were not necessary for but accelerated inactivation of strain AH-1N.

Figure 4.

Inactivation of cells of A. hydrophila strain AH-1N.
A. Inactivation by co-culture supernatants in the presence and absence of cells of P. aeruginosa strain PAO1. Colony-forming units (cfu) (▴) and total cell counts (TCCs) (▵) of strain AH-1N in co-culture supernatant in the absence of cells of strain PAO1; cfu of strain AH-1N in the presence of cells of strain PAO1 (106 cfu ml−1) (●); cfu (inline image) and TCCs (□) of strain AH-1N in the medium control.
B. Inactivation by pyocyanin. Colony-forming units of strain AH-1N during incubation with 0 (inline image; solvent control with methanol), 5 (▴), 10 (●) and 30 (◆) µM pyocyanin.
Error bars indicate standard deviation (n = 3).

Analysis of co-culture supernatant revealed the presence of rhamnolipids at concentrations up to 31 mg l−1, which could be a plausible cause for foam formation. Fluorescence spectroscopy revealed the presence of the siderophores pyochelin and pyoverdine, as detected by their characteristic excitation/emission wavelengths of 350/440 nm and 405/455 nm, respectively (Cox and Adams, 1985; Visca et al., 1992). Nevertheless, a co-culture with the siderophore-deficient mutant PAO1ΔpvdDΔpchEF did not differ from a co-culture with the wild type. Furthermore, elastolytic activity could be measured (data not shown), while cyanide could not be detected. In addition, compounds involved in quorum sensing could also be detected (see below).

Reversed-phase HPLC analysis of co-culture supernatants revealed the production of the phenazine compounds phenazine-1-carboxylate (PCA) (not shown) and pyocyanin (Fig. 3C) after 40 h of incubation. After 80 h the concentration of pyocyanin was above 20 µm and remained constant. Pyocyanin has toxic effects on bacteria (Hassan and Fridovich, 1980; Baron and Rowe, 1981; Lau et al., 2004). Therefore, we investigated whether pyocyanin as an isolated compound could cause inactivation of strain AH-1N. Pyocyanin caused a concentration-dependent decrease of cfu in cell suspensions of strain AH-1N within 2 days (Fig. 4B).

As pyocyanin production is known to be induced by phosphate limitation (Whooley and McLoughlin, 1983; Jensen et al., 2006), we incubated the co-culture with 600 µM instead of 150 µM phosphate. In this co-culture cells of strain AH-1N were not inactivated, and the culture did not turn green (data not shown). Increasing the iron concentration from 7.5 µM to 100 µM did not change the course of the co-culture.

Impact of quorum sensing

In P. aeruginosa the production of above mentioned secondary metabolites is regulated via a hierarchical quorum-sensing system that consists of two N-acylhomoserine lactone (AHL) regulatory circuits (las and rhl) and a 2-alkyl-4-quinolone (AQ) system (Williams and Cámara, 2009). In the las and rhl circuits signalling is mediated via N-(3-oxo-dodecanoyl)-l-homoserine lactone (3-oxo-C12-HSL) and N-(butanoyl)-l-homoserine lactone (C4-HSL), respectively. Signalling in the AQ system is mediated mainly by 2-heptyl-3-hydroxy-4-quinolone, which is referred to as the pseudomonas quinolone signal (PQS) and by 2-heptyl-4-quinolone (HHQ) (Dubern and Diggle, 2008). In A. hydrophila, quorum sensing via the ahy circuit is mediated by the AHL C4-HSL (Swift et al., 1999).

Bioassays indicated the presence of C4-HSL and 3-oxo-C12-HSL in co-cultures of strains AH-1N and PAO1 during growth with chitin and the presence of C4-HSL in single cultures of strain AH-1N (data not shown). HPLC analysis of co-culture samples revealed the production of HHQ, starting after 40 h of incubation and increasing up to 5 µM within the next 30 h (Fig. 3C). While PQS could not be detected even after extracting and concentrating supernatant samples, 4-hydroxy-2-heptylquinoline-N-oxide (HQNO) was detectable in untreated supernatants.

To investigate the impact of quorum sensing by strain AH-1N on the co-culture with chitin, we co-incubated strain PAO1 with the AHL-negative mutant strain AH-1NΔahyI. The course of the co-culture did not differ from the co-culture with both wild types (data not shown). This result also showed that quorum sensing by strain AH-1N was not required for chitin degradation. In this co-culture we detected both AHLs, indicating that C4-HSL detected in the co-culture of both wild types was at least partly produced by strain PAO1. To investigate the impact of quorum sensing by strain PAO1 on the co-culture with chitin, we co-incubated strain AH-1N with the AQ-negative mutant strain PAO1ΔpqsA and the AHL-negative mutant strain PAO1(ATCC)ΔlasIΔrhlI, respectively. In these co-cultures, cfu of strains PAO1ΔpqsA (Fig. 5A) and PAO1(ATCC)ΔlasIΔrhlI (not shown) reached significantly lower numbers than the respective wild type strains. In both co-cultures, cfu of strain AH-1N decreased only slightly in stationary phase, and the cultures did not turn green and foamy. The same course was observed in co-cultures of strain AH-1N and the mutant strain PAO1ΔpqsR that lacks the transcriptional regulator PqsR, which is required for the activation of AQ biosynthesis (Gallagher et al., 2002). Strains PAO1ΔpqsA and PAO1(ATCC)ΔlasIΔrhlI could be complemented by supplying the co-culture with a mixture of C4-HSL and 3-oxo-C12-HSL (1 µM each) or with PQS (1 µM), respectively. These results showed that the inactivation of strain AH-1N in co-culture with chitin was dependent on AHL- and AQ-mediated quorum sensing by strain PAO1.

Figure 5.

Growth of A. hydrophila strain AH-1N and of P. aeruginosa mutants with defects in the AQ-mediated quorum sensing system in co-cultures with chitin.
A. Co-cultures of strain AH-1N with strains PAO1ΔpqsA and PAO1ΔpqsH, respectively. Colony-forming units (cfu) of strain AH-1N in co-cultures with strain PAO1ΔpqsA (○); cfu of strain AH-1N in co-cultures with strain PAO1ΔpqsH (●); cfu of strain PAO1ΔpqsA (▵); cfu of strain PAO1ΔpqsH (▴).
B. Co-cultures of strain AH-1N with strain PAO1ΔpqsA in the presence and absence of pyocyanin. Colony-forming units of strain AH-1N (●, ○) and of strain PAO1ΔpqsA (▴, ▵) in co-cultures with 20 µM pyocyanin (closed symbols) and with methanol (open symbols) as solvent control. The arrow indicates addition of pyocyanin and methanol, respectively.
Error bars indicate standard deviation (n = 3).

To further elucidate the impact of PQS and HHQ on the course of the co-culture, strain AH-1N was co-incubated with the mutant strain PAO1ΔpqsH, which lacks the monooxygenase for conversion of HHQ to PQS (Gallagher et al., 2002). This co-culture did not differ from the co-culture with the PAO1 wild type (Fig. 5A) and contained red cells. These results showed that PQS was not required for the inactivation and for the red colouration of cells of strain AH-1N. It was shown that the ΔpqsH mutant displays normal PqsR-dependent gene expression and virulence, but that PQS is required for full production of pyocyanin (Xiao et al., 2006). In agreement with that the pyocyanin concentration in supernatants of co-cultures of strains AH-1N and PAO1ΔpqsH was lower (up to 11 µM) than in supernatants of co-cultures with both wild types.

Effect of pyocyanin on the metabolism of strain AH-1N

Intriguingly, pyocyanin production in the co-culture coincided with the massive release of about 10 mM acetate (Fig. 3B and C). In contrast, in a co-culture of strains AH-1N and PAO1ΔpqsA, in which no pyocyanin was detectable, only about 3 mM acetate was released. These observations prompted us to investigate a possible role of pyocyanin in triggering the release of acetate.

For this, 20 µM pyocyanin was added to single cultures of strain AH-1N that had been growing with chitin for 48 h. This time point was equivalent to the onset of pyocyanin formation in the co-culture. This addition caused a release of up to 10 mM acetate within the following 48 h (data not shown), which was not observed in the respective control with methanol. As macroscopic chitin particles were mostly degraded at the time points of pyocyanin addition to single cultures and of pyocyanin production in co-cultures, the released acetate could originate from larger GlcNAc oligomers that were still present in the supernatant but could not be detected by HPLC analysis. Additionally, acetate could also originate from degradation of storage material of strain AH-1N.

To explore this effect of pyocyanin in more detail, we incubated cell suspensions of strain AH-1N with 3 mM GlcNAc as sole substrate in the presence of 20 µM pyocyanin. Within 22 h GlcNAc was completely degraded, and up to 7 mM acetate accumulated, which was not utilized further (Fig. 6A). In the control with methanol, degradation of GlcNAc was faster, and up to 3 mM acetate accumulated only transiently (Fig. 6B). Cell suspension experiments with 10 mM acetate as sole substrate revealed that acetate could not be degraded in the presence of pyocyanin, while it was completely consumed in the methanol control within 20 h (Fig. 6C). These results clearly showed that pyocyanin caused an incomplete oxidation of GlcNAc to acetate, which is based on the inhibition of the further degradation of acetate.

Figure 6.

Effect of 20 µM pyocyanin on the degradation of GlcNAc (A and B) and of acetate (C) by cell suspensions of A. hydrophila strain AH-1N.
A. Concentration of GlcNAc (inline image) and acetate(◆) in the presence of pyocyanin.
B. Concentration of GlcNAc (inline image) and acetate (◆) in the solvent control with methanol.
C. Concentration of acetate in the presence of pyocyanin (◆) and in the solvent control with methanol (◊).
Error bars indicate standard deviation (n = 3).

Pyocyanin is known to divert the cellular electron flow by chemically and biochemically oxidizing electron carriers such as NAD(P)H and reducing oxygen to reactive oxygen species (ROS) with the superoxide radical as primary product. To investigate whether pyocyanin caused such an electron diversion in strain AH-1N, we performed oxygen consumption experiments with cell suspensions incubated with GlcNAc as described previously (Hassan and Fridovich, 1980). These experiments showed that the substrate-dependent oxygen consumption, which was inhibited by KCN, could be increased by addition of pyocyanin in a concentration-dependent manner (Fig. 7A). This increase of cyanide-insensitive oxygen consumption was a strong indication for the formation of ROS by pyocyanin, which had been reduced by electrons derived from GlcNAc oxidation. This effect of pyocyanin was not observed with suspensions of heat-inactivated cells of strain AH-1N (incubation at 80°C for 10 min) indicating that metabolically active cells were required for pyocyanin-dependent oxygen consumption. In corresponding experiments with cells of strain PAO1 pyocyanin had no influence on oxygen consumption, which is in agreement with previous studies (Hassan and Fridovich, 1980; Hassett et al., 1992).

Figure 7.

Effects of pyocyanin on A. hydrophila strain AH-1N.
A. Effect of pyocyanin on oxygen consumption rates of cell suspensions after successive addition of (i) 10 mM GlcNAc, (ii) 0.5 mM KCN and (iii) pyocyanin at the respective concentration indicated in the graph. Error bars indicate standard deviation (n = 15 for black and white bar, and n = 3 for each grey bar).
B. Effect of 10 µM pyocyanin on aconitase activity in cell extracts of strain AH-1N in the absence and presence of 30 µM NADH. Assays with methanol in the presence and absence of NADH served as controls. Error bars indicate standard deviation (n = 3).

Effect of pyocyanin on aconitase activity

It is known that the superoxide radical preferentially damages iron–sulfur cluster containing dehydratases, such as aconitase and fumarase that are both involved in acetate degradation through the citric acid cycle (Imlay, 2003). To investigate if pyocyanin would inhibit aconitase in cells of strain AH-1N, we measured aconitase activity in cell extracts prepared from cell suspensions that had been incubated with GlcNAc in the presence and absence of pyocyanin as described above. In extracts of cells, which had been incubated in the presence of pyocyanin, aconitase activity was 80–100% lower compared with the aconitase activity of 120 ± 14 mU (mg protein)−1 in extracts of cells, which had been incubated in the absence of pyocyanin. There was no difference in the activities of pyruvate dehydrogenase [249 mU (mg protein)−1 in the presence and 268 ± 64 mU (mg protein)−1 in the absence of pyocyanin] and isocitrate dehydrogenase [487 ± 11 mU (mg protein)−1 in the presence and 457 ± 28 mU (mg protein)−1 in the absence of pyocyanin] between these two cell extracts indicating that incubation with pyocyanin had no general damaging effect on enzymes.

To explore the effect of pyocyanin on aconitase activity in vitro, we prepared extracts from GlcNAc-grown cells of strain AH-1N that were harvested in the mid-exponential growth phase. In these cell extracts aconitase activity was inhibited by 50%, if the assays contained 10 µM pyocyanin and 30 µM NADH as electron donor (Fig. 7B). In the absence of NADH pyocyanin did not cause an inhibition. Thus, the inhibition of aconitase was dependent on the reduction of pyocyanin, and, in consequence, on the formation of ROS.

Growth restoration of strain PAO1ΔpqsA in a co-culture by pyocyanin

To investigate whether the release of acetate triggered by pyocyanin was sufficient to support growth of strain PAO1, we supplied a co-culture of strains AH-1N and PAO1ΔpqsA with 20 µM of pyocyanin after incubation for 48 h as described above. Strain PAO1ΔpqsA started growth within 72 h after addition of pyocyanin and reached 10-fold higher cfu numbers than in the respective co-culture control with methanol (Fig. 5B). HPLC analysis showed an increase of acetate of up to 4 mM within 24 h after addition of pyocyanin, which did not occur in the co-culture control with methanol; acetate was completely consumed when growth of strain PAO1ΔpqsA started (not shown). These results showed that the effects of pyocyanin on strain AH-1N were sufficient to restore growth of a mutant strain of PAO1 that does not produce pyocyanin itself thereby underlining the crucial role of this compound for the course of the co-culture. As expected from the cell suspension experiments (Fig. 4B), the addition of pyocyanin caused also a decrease of cfu of strain AH-1N within 3 days. This inactivation was slower than in co-cultures with the wild type indicating inactivating effects of further quorum sensing-regulated secondary metabolites. In the co-culture control with methanol cfu of strain AH-1N also decreased but to a lesser extent; this was also observed in a single culture of strain AH-1N with methanol indicating a toxic effect of this solvent.

Discussion

Bacteria investing in the production of extracellular hydrolytic enzymes for the degradation of polymers face the danger of being exploited by opportunistic bacteria that may thrive on degradation products without investing energy in the biosynthesis of these enzymes. In our study, we investigated such a scenario with a co-culture of the chitinolytic bacterium A. hydrophila strain AH-1N as investor and P. aeruginosa strain PAO1 as opportunist with chitin as the sole source of carbon, nitrogen and energy.

This co-culture had a highly reproducible biphasic course. In the first phase, strain AH-1N grew with chitin at a very similar rate as in single culture, while strain PAO1 reached slightly higher cell numbers compared with its single culture. This interaction can be described as commensal, because the opportunist grew with substrates provided by the investor, who is obviously not harmed. The second phase of the co-culture commenced with the formation of secondary metabolites by strain PAO1, whose biosynthesis was under control of quorum sensing. These metabolites caused a rapid inactivation of and the release of acetate by cells of strain AH-1N. Strain PAO1 continued growth with the released acetate and reached about 200-fold higher cell numbers than in its single culture. Thus, strain PAO1 employed its secondary metabolites to influence both the viability and the metabolism of strain AH-1N in such a way that it eventually could use the carbon, nitrogen and energy of the substrate chitin that was originally not bioavailable to it. This interaction can be described as parasitic, because the opportunist grew by forcing the investor to release substrates and by harming the investor. In conclusion, strain PAO1 progressed from a commensal to a parasitic growth strategy, and this progression was regulated by quorum sensing.

The redox active compound pyocyanin played the key role in this growth strategy of strain PAO1 and exhibited two functions as shown by using it as a pure compound. First, pyocyanin triggered the release of acetate by strain AH-1N, and, second, caused an inactivation of strain AH-1N. The oxygen consumption experiments provided strong evidence that pyocyanin caused the formation of the superoxide radical and further ROS in strain AH-1N. This is in agreement with the inhibition of the superoxide radical-sensitive aconitase in cells of strain AH-1N that had been incubated with pyocyanin. This effect of pyocyanin on aconitase has also been shown in human lung epithelial cells (O'Malley et al., 2003). Inhibition of aconitase interrupts the citric acid cycle already at the second step. Therefore, inhibition of aconitase sufficiently explained the observed complete block of acetate degradation in the presence of pyocyanin. In contrast, pyocyanin had only a slight inhibiting effect on GlcNAc degradation and did not affect pyruvate dehydrogenase activity. In the co-culture, continued degradation of GlcNAc oligomers by strain AH-1N in the presence of pyocyanin would therefore lead to an accumulation of acetyl-CoA, which could plausibly explain the release of acetate into the medium. An alternative explanation for the release of acetate could be that the reduction of oxygen by pyocyanin might lower the oxygen partial pressure to a degree that fermentative metabolism of strain AH-1N is induced, even though oxic conditions still prevail. Strain AH-1N was able to ferment GlcNAc to acetate, lactate, succinate, acetoin, formate and ethanol (N. Jagmann and B. Philipp, unpublished). Theoretically, pyocyanin could transfer electrons derived from the oxidative branch of GlcNAc fermentation from NADH to oxygen, thereby avoiding the formation of reduced fermentation products as electron sinks. In consequence, acetate would be the only fermentation product of GlcNAc. However, it was unlikely that pyocyanin caused a switch to anaerobic metabolism, because in extracts of cells grown in the presence of pyocyanin a high pyruvate dehydrogenase activity was detected, which was absent in extracts of GlcNAc-fermenting cells (N. Jagmann and B. Philipp, unpublished). Therefore, we conclude that pyocyanin forced strain AH-1N into incomplete oxidation of the available substrates to acetate as a dead-end metabolite by inhibition of acetate degradation through the citric acid cycle. Based on this inhibiting effect pyocyanin could also complement the phenotype of a ΔpqsA mutant of strain PAO1 by restoring growth in the co-culture. To our knowledge, this effect of pyocyanin has never been reported. This mechanism might as well be applicable for exploiting bacteria degrading cellulose, which is also not bioavailable for strain PAO1 according to its genome sequence.

In addition, pyocyanin caused a long-term inactivation of strain AH-1N, showing that the cells had been irreversibly damaged by ROS. Pyocyanin-dependent inactivation has also been shown with Escherichia coli (Hassan and Fridovich, 1980; Hassett et al., 1992), Staphylococcus aureus (Biswas et al., 2009) as well as with several other bacteria (Baron and Rowe, 1981; Norman et al., 2004). In the co-culture, other secondary metabolites, such as rhamnolipids (Haba et al., 2003) and HQNO (Hoffman et al., 2006), had certainly contributed to inactivating strain AH-1N because the rate of inactivation was lower in the co-culture with the ΔpqsA mutant after addition of pyocyanin only. In cell suspension experiments, inactivation was strongly accelerated by the presence of cells of strain PAO1, which might be caused by cell-bound factors of strain PAO1. Additionally, strain PAO1 might scavenge nutrients that could otherwise extend the viability span of strain AH-1N. The long-term advantage of inactivating strain AH-1N could be the suppression of a competitor for potential future substrates that strain PAO1 can utilize on its own.

Interestingly, the inactivation of strain AH-1N was accompanied by a red colouration of cells, which was reminiscent of two phenomena, the so-called red death of Caenorhabditis elegans and the inactivation of Candida albicans, that are both caused by P. aeruginosa (Gibson et al., 2009; Zaborin et al., 2009). Red death of C. elegans is dependent on a PQS/Fe3+ complex that causes the same symptoms in the animal as did cells of the P. aeruginosa wild type. In our system, however, PQS was not required because a ΔpqsH mutant that produces HHQ but no PQS also caused the formation of inactivated red cells in the co-culture. As a further difference, the siderophore-deficient mutant PAO1ΔpvdDΔpchEF, which is not lethal to C. elegans (Zaborin et al., 2009), did not differ from the wild type in our system. These findings suggest that the red colouration of inactivated cells of strain AH-1N was caused by a mechanism different from that observed with C. elegans. The inactivation of C. albicans is also accompanied by a red colouration of the fungal cells, which is dependent on 5-methyl-phenazinium-1-carboxylate (5MPCA), the precursor of pyocyanin (Gibson et al., 2009). As in our system, pyocyanin as a pure compound did not cause red colouration, the red coloration disappeared after addition of the reducing agent dithionate, and the pigment appears to be strongly associated with the cells (N. Jagmann and B. Philipp, unpublished). These features suggest a similar mechanism for the red coloration in strain AH-1N and C. albicans. However, inactivation by 5MPCA requires cell–cell contact of P. aeruginosa and C. albicans on solid media, and, unlike PCA and pyocyanin, 5MPCA does not accumulate in culture supernatants and is not stable at neutral pH (Gibson et al., 2009). These features suggest some differences to our system, because red coloration of cells of strain AH-1N also occurred upon incubation with cell-free supernatants from co-cultures.

The timing for the production of bioactive secondary metabolites by strain PAO1 in the co-culture was crucial, because their premature formation would suppress chitin degradation by strain AH-1N before strain PAO1 could profit from it. Quorum sensing is a means for linking the formation of bioactive secondary metabolites to the environmental situation, population density and the physiological status (Williams and Cámara, 2009). The prerequisite for the formation of signalling molecules and secondary metabolites is the availability of substrates. This raised the question, which substrates, apart from possible storage material, strain PAO1 could use in the first phase of the co-culture. The fact that no primary chitin degradation products and only traces of glucosamine were detectable in single cultures of strain AH-1N revealed that there was a tight coupling of chitin hydrolysis and uptake of the degradation products by this strain. Therefore, it was unlikely that GlcNAc or glucosamine were available as substrates for strain PAO1 in the first phase of the co-culture. It was more probable that strain PAO1 took advantage of the transient acetate release that accompanied chitin degradation by strain AH-1N. Such a transient release of acetate has also been shown for other Aeromonads growing aerobically with carbohydrates (Namdari and Cabelli, 1990). In addition, it was possible that strain PAO1 could grow with other minor exudates of strain AH-1N, as it is known that P. aeruginosa can grow in media with very low concentrations of organic compounds (Favero et al., 1971; van der Kooij et al., 1982). Growth of strain PAO1 in the first phase of the co-culture was therefore likely based on utilizing compounds released from strain AH-1N in any case during chitin degradation, including ammonium as a nitrogen source. From this rather precarious metabolic situation strain PAO1 progressed into the second phase of the co-culture, in which it actively enforced the release of acetate by strain AH-1N by employing quorum sensing-regulated properties.

Quorum sensing is defined to be cell density-dependent, but it has been shown that quorum sensing of P. aeruginosa can also be regulated by environmental factors independent of cell density (Fuqua et al., 2001; Wagner et al., 2003; Duan and Surette, 2007; Williams and Cámara, 2009). The fact that the production of quorum sensing-regulated secondary metabolites by strain PAO1 in the co-culture occurred at a relatively low cell density of about 107 cfu ml−1, corresponding to an optical density at 600 nm (OD600) of 0.01, indicated a strong impact of environmental factors. The precarious metabolic situation in the first phase was likely to result in nutrient starvation, which could lead to the induction of quorum sensing via the stringent response (van Delden et al., 2001). Quorum sensing can also be induced upon iron (Bollinger et al., 2001; Kim et al., 2005; Jensen et al., 2006; Duan and Surette, 2007) and phosphate limitation (Bazire et al., 2005; Jensen et al., 2006). While increasing the iron concentration had no effect, increasing the phosphate concentration suppressed the production of bioactive secondary metabolites. This clearly showed that phosphate limitation was an important trigger for quorum sensing in the co-culture. In addition to nutritional factors specific signals from strain AH-1N might also be involved in inducing quorum sensing under these culture conditions. It was unlikely that C4-HSL produced by strain AH-1N played such a role, because co-cultures with the ΔahyI mutant of strain AH-1N did not differ from those with the wild type. Finally, chitin-specific cues could also contribute to triggering quorum sensing in strain PAO1. Studies for identifying molecular mechanisms underlying the course of the co-culture are currently on the way in our laboratory.

Our study is a further example that the quorum sensing-regulated secondary metabolites of P. aeruginosa, which are mainly viewed as its virulence factors in human infections, do also play an important ecological role in competition with other microbes (Hogan and Kolter, 2002; Norman et al., 2004). The parasitic exploitation of chitin-degrading bacteria could contribute to the occasional massive developments of P. aeruginosa during the summer season in oligotrophic freshwater systems, which usually contain only very low numbers of this opportunistic pathogen (van Asperen et al., 1995; Mena and Gerba, 2009).

Experimental procedures

Bacterial strains and growth media

Bacterial strains and plasmids used in this study are listed in Table 1. For cultivation of P. aeruginosa and A. hydrophila strains in liquid culture a mineral medium designated medium B was used. Medium B contained the following components: 50 mM Hepes (pH 7), 0.5 mM MgSO4, 14 mM KCl, 7.2 mM NaCl, 5 mM NH4Cl. After autoclaving, the medium was complemented with 0.01 mM CaCl2, 0.15 mM Na-K-phosphate buffer (pH 7; 0.105 mM K2HPO4, 0.045 mM NaH2PO4) and the trace element solution SL10 (Widdel et al., 1983). Acetate and tryptone were used as carbon and energy sources. If Chitin, N-acetylglucosamine (GlcNAc), N,N′-diacetylchitobiose [(GlcNAc)2, Sigma], N,N′,N″-triacetylchitotriose [(GlcNAc)3, Sigma] and glucosamine served as carbon, energy and nitrogen sources, ammonium was omitted from the medium.

Table 1.  Strains and plasmids used in this study.
Strains and plasmidsRelevant characteristicsSource or reference
Pseudomonas aeruginosa  
 PAO1PAO1 Nottingham wild typeHolloway collection
 PAO1(ATCC)PAO1 wild type ATCC 15692ATCC
 PAO1-Tn7-cfpPAO1 with Tn7 chromosomal insertion of cfpThis study
 PAO1(ATCC)ΔlasIΔrhlIPAO1(ATCC) with Tc cartridge inserted into unique EcoRI site of rhlI and with Gm cartridge inserted into unique EcoRI site of lasIBeatson et al. (2002)
 PAO1ΔpqsApqsA deletion mutant, PQS negativeAendekerk et al. (2005)
 PAO1ΔpqsHpqsH deletion mutantFletcher et al. (2007)
 PAO1ΔpqsRpqsR deletion mutantUniversity of Nottingham
 PAO1ΔpvdDΔpchEFpvdDpchEF deletion mutant; pyoverdin and pyochelin negativeGhysels et al. (2004)
 PAO1ΔlecAlecA deletion mutantUniversity of Nottingham
 PAO1ΔlecBlecB deletion mutantUniversity of Nottingham
 PAO1-KO[chiC]PAO1, chiC::pKO[chiC]This study
 PAO1-KO[cbpD]PAO1, cbpD::pKO[cbpD]This study
Aeromonas hydrophila  
 AH-1NAH-1N wild typeSwift et al. (1999 )
 AH-1NΔahyIahyI deletion mutant, AHL negativeLynch et al. (2002)
Escherichia coli  
 S17-1thi pro hsdR hsdM+recA RP4-2-Tc::Mu-Km::Tn7Simon et al. (1983)
Plasmids  
 pKnockout-GSuicide vector used for gene inactivation (Apr, Gmr)Windgassen et al. (2000)
 pKO[chiC]pKnockout-G harbouring an internal PstI fragment (798 bp) of chiCThis study
 pKO[cbpD]pKnockout-G harbouring an internal HincII fragment (606 bp) of cbpDThis study

Pseudomonas aeruginosa strain PAO1, A. hydrophila strain AH-1N and unmarked deletion mutants of strains PAO1 and AH-1N were maintained on solid (1.5% w/v agar) medium B plates containing 1% tryptone. Pseudomonas aeruginosa strain PAO1-Tn7-cfp was maintained on solid medium B plates containing 1% w/v tryptone and 30 µg ml−1 chloramphenicol. Plasmid-harbouring E. coli strain S17-1 was selected and maintained on LB plates containing 100 µg ml−1 ampicillin. Insertional mutants of P. aeruginosa were selected on Pseudomonas isolation agar (PIA; Difco) containing 500 µg ml−1 carbenicillin. For cfu counts, LB plates were used. Additionally, unmarked P. aeruginosa strains were selected on PIA plates containing 20 µg ml−1 tetracycline, and P. aeruginosa strain PAO1-Tn7-cfp was selected on LB plates containing 30 µg ml−1 chloramphenicol.

Preparation of suspended chitin

Suspended chitin was prepared according to a modified protocol (Reichenbach, 2006). Twenty grams of commercial chitin flakes (practical grade; Sigma) were added to 400 ml of HCl (37%) under stirring until a homogenous slurry was obtained. After stirring for 20 min the suspension was poured into 5 l of ice-cold deionized water. After stirring for 10 min the suspension was filtered through a cellulose coffee filter, and the chitin was washed repeatedly with deionized water until a pH value of about 4 was reached. After resuspension of the chitin in deionized water and adjusting the pH to 7 with NaOH, the suspension was again filtered through a cellulose filter, and the chitin was resuspended in 1 l of medium B to obtain a final concentration of 2% (w/v). The chitin suspension was transferred into bottles and autoclaved. For cultivation a final chitin concentration of 0.5% (w/v) was used.

Growth experiments

All growth experiments were performed at 30°C. Pre-cultures of strains AH-1N and PAO1 were incubated in 4 ml of medium B containing 0.1% tryptone in 15 ml test tubes on a rotary shaker (innova 4000 incubator shaker; New Brunswick or KS4000i control; IKA) at 200 r.p.m. for 13–16 h at 30°C. Growth of pre-cultures was measured as OD600 with a spectrophotometer. Pre-cultures were harvested by centrifugation at 9300 g for 5 min, washed with medium B without ammonium and complementing solutions, and were used to inoculate main cultures at OD600 = 0.001 for both strains, which equals 106 cells ml−1. Main cultures were incubated on a rotary shaker at 200 r.p.m., unless indicated otherwise.

For growth experiments with chitin, main cultures were incubated in 4 ml of medium B in 15 ml test tubes or in 100 ml of medium B in a 500 ml Erlenmeyer flask without baffles. For growth experiments with GlcNAc (5 mM), glucosamine (5 mM), tryptone (0.1% or 0.5%) and acetate (10 mM) main cultures were incubated in 4 ml of medium B in 15 ml test tubes. For growth experiments with (GlcNAc)2 (2.5 mM) and (GlcNAc)3 (1.7 mM) main cultures were incubated in 400 µl of medium B in 48-well microtitre plates (Nunc) on a rotary shaker at 180 r.p.m.

Bacterial growth in single and co-cultures with chitin and/or tryptone was measured by determination of cfu. An aliquot of 20 µl of each culture was diluted in a decimal series in 180 µl of medium B without ammonium and complementing solutions. From appropriate dilution steps, three aliquots of 10 µl were used for cfu counts by the drop plate method (Hoben and Somasegaran, 1982).

Colonies of strains PAO1 and AH-1N were distinguished by colony morphology, and colonies of PAO1 were additionally selected using antibiotics (see above). The detection limit of strain AH-1N in co-cultures was 105 cfu ml−1, because the respective dilution still allowed the unambiguous detection of colonies of strain AH-1N in the presence of higher colony numbers of strain PAO1. Growth of single cultures with GlcNAc, glucosamine or acetate as carbon sources was measured as OD600 with a spectrophotometer (model M107 with test-tube holder; Camspec). Growth of single cultures with (GlcNAc)2 and (GlcNAc)3 was measured as OD595 in a microplate reader (Genios, Tecan). Total cell counts were determined as described previously (Styp von Rekowski et al., 2008).

Construction of plasmids and insertional mutants

Genomic DNA of strain PAO1 was purified with the Puregene Tissue Core Kit B (Qiagen). PA2300 (chiC) was amplified using the primer chiC-F (5′-TGGTAGACGCTCGCGCCTGTTTTT-3′) and chiC-R (5′-GCTCTCGCCGGCCAAAGGAC-3′). PA0852 (cbpD) was amplified using the primer cbpD-F (5′-CCGTCACATTTGGTAGGGAC-3′) and cbpD-R (5′-GCTTGAACAGGCACACGTAG-3′). To construct the plasmids pKO[chiC] and pKO[cbpD] an internal PstI fragment of chiC and an internal HincII fragment of cbpD, respectively, were cloned into the respective restriction sites of the suicide vector pKnockout-G (Windgassen et al., 2000). The resulting plasmids were transferred into strain PAO1 by bi-parental mating with E. coli strain S17-1 as donor. Escherichia coli strain S17-1 (donor) was grown in LB medium with 100 µg ml−1 ampicillin at 200 r.p.m. at 30°C, while strain PAO1 (recipient) was grown in LB medium at 50 r.p.m. at 42°C. After incubation overnight, 5 × 108 cells of the donor and 1 × 109 cells of the recipient were harvested by centrifugation at 9300 g for 1.5 min, washed with 500 µl of pre-warmed LB medium, and finally resuspended in 50 µl of LB medium. Donor and recipient were carefully mixed by pipetting and spread onto sterile membrane filters (Durapore membrane filters, 0.22 µM GV; Millipore) that were placed on pre-warmed LB plates. After incubation for 6 h at 37°C, the filters were transferred to a 12 ml plastic tube containing 2 ml of 0.9% NaCl. After vortexing, the cell suspensions were transferred to a 2 ml plastic tube, centrifuged at 9300 g for 2 min and resuspended in 600 µl of 0.9% NaCl. Aliquots of the cell suspensions were spread on PIA plates containing 500 µg ml−1 carbenicillin to select for insertional mutants of PAO1. Correct chromosomal insertion of the plasmids was confirmed by PCR with the primers pKO-G (5′-GCGCGTTGGCCGATTCATTA-3′) and chiC-check-R (5′-GTGAAGGCTACCGGCGGC-3′) or cbpD-check-R (5′-GCTGACCGCCCCGTAGG-3′). Strain PAO1-Tn7-cfp was constructed as described previously (Klebensberger et al., 2007).

Cells suspension experiments with strain AH-1N

Pre-cultures of strain AH-1N were incubated for 7–9 h as described above. Main cultures were incubated for 16 h as described above, harvested by centrifugation at 8300 g for 5 min and washed with medium B without ammonium and complementing solutions before resuspension in the appropriate medium or sterile culture supernatant to an OD600 of 1. Experiments were performed with 4 ml cell suspension in 15 ml test tubes or with 20 ml cell suspension in 100 ml Erlenmeyer flasks without baffles at 30°C in rotatory shakers at 200 r.p.m.

For cell suspension experiments with supernatant of chitin-grown co-cultures, main cultures of strain AH-1N were grown with 0.5% tryptone. Co-culture supernatant was obtained by incubating the co-cultures for 5 days as described above before two centrifugation steps at maximum speed for 15 min at 15°C and repeated filter-sterilization of the supernatant. For cell suspension experiments with pyocyanin, main cultures of strain AH-1N were grown with 0.5% tryptone. Cell suspensions were prepared by resuspending the cells in medium B without ammonium with the addition of 5–20 µM pyocyanin (Cayman; 10 mM stock solution in methanol). As a control, only methanol was added. For cell suspension experiments for monitoring the degradation of GlcNAc (3 mM) or acetate (10 mM) in the presence or absence of pyocyanin, main cultures of strain AH-1N were incubated with 10 mM GlcNAc. Cell suspensions were prepared by resuspending the cells in medium B without ammonium and complementing solutions with the addition of 20 µM pyocyanin. For cell suspensions with acetate, ammonium was added to the medium. As a control, only methanol was added.

Chitin binding assay

Pre-cultures of strains AH-1N, PAO1, PAO1-KO[cbpD], PAO1ΔlecA and PAO1ΔlecB were incubated with 0.1% tryptone for 7 h as described above. Main cultures of these strains were incubated with 0.5% tryptone for 16 h as described above, harvested by centrifugation at 8300 g for 5 min and washed with medium B without ammonium and complementing solutions before resuspension in medium B with and without 0.5% chitin to an OD600 of 1 in 15 ml test tubes. After shaking of the tubes, chitin was allowed to sediment for 30 min before cfu from the chitin-free supernatant were determined. Colony-forming unit numbers in cell suspensions without chitin were set as 100%.

Quantification of substrates and degradation products

Suspended chitin in test tubes was quantified by measuring its filling level. After shaking of the tubes to obtain a homogenous distribution, chitin was allowed to sediment for 30 min before its filling level was measured.

Samples for measurements of degradation products were centrifuged in 1.5 ml plastic tubes at maximum speed for 15 min at room temperature. Supernatants were transferred into new plastic tubes and stored at −20°C until further analysis. Acetate, GlcNAc, (GlcNAc)2 and (GlcNAc)3 were determined by ion-exclusion HPLC as described previously (Klebensberger et al., 2006). Glucosamine was determined by reversed-phase HPLC as described previously (Zhu et al., 2005). Derivatization was performed in 2 ml plastic tubes. Two hundred microlitres of borate buffer (200 mM, pH 7), 200 µl of 129.35 mg l−1 Fmoc-Cl in acetonitrile and 20 µl of the sample or standard solution [0–1 mM d-glucosamine hydrochloride (Sigma) in medium B without ammonium] were mixed and incubated at 20°C for 30 min. Analysis was performed with a HPLC system (Prominence liquid chromatograph; Shimadzu) equipped with a diode array detector (SPD M20A; Schimadzu) and a reversed-phase column (EC 150/4.6 Nucleodur 100-3 C18ec; Macherey and Nagel) using the previously described gradient (Zhu et al., 2005) at a flow rate of 0.8 ml min−1. Fifty microlitres were injected onto the column.

Ammonium was determined enzymatically with glutamate dehydrogenase. 100 mM TEA buffer (pH 8.6), 5 mM α-ketoglutarate, 150 µM NADPH and 0.96 U glutamate dehydrogenase (Sigma) were mixed in wells of a 96-well microtitre plate (Nunc) to a final volume of 190 µl. The assay was started by adding 10 µl of the sample or NH4Cl standard (0–2 mM) and incubated at 30°C. The absorption at 340 nm was measured at time zero and after 20 min in a microplate reader (Genios, Tecan).

Determination of phenazines and alkylquinolones

The phenazines pyocyanin and PCA and the alkylquinolones HHQ, HQNO and 2-heptyl-3-hydroxy-4-quinolone (PQS) in co-culture supernatants were determined by reversed-phase HPLC (see above) with a flow rate of 0.8 ml min−1. A gradient method was applied starting with 10% acetonitrile for 2 min, rising to 70% acetonitrile within 10 min, staying at 70% acetonitrile for 3 min, and returning to 10% acetonitrile within 1 min, followed by an equilibration of 5 min. Fifty microlitres were injected onto the column. Compounds were identified by co-elution with and by comparison of UV/VIS-spectra of reference compounds.

Determination of rhamnolipids

The concentration of rhamnolipids in co-culture supernatants was determined after modified protocols (Ramana and Karanth, 1989; Chayabutra et al., 2001; Pinzon and Ju, 2009). Ten millilitres of co-culture supernatant was acidified (pH 3–4) with 4.5 M sulfuric acid and extracted twice with 30 ml of dichloromethane. After dichloromethane was evaporated, the solid matter was dissolved in a small volume of methanol. After evaporating methanol under a stream of N2, the solid matter was dissolved in 400 µl of 0.05 M NaHCO3 and mixed with 600 µl of anthron solution (0.005 g l−1 in sulfuric acid) carefully releasing the overpressure. The mixture was incubated at 100°C for 30 min, and the absorption at 460 nm was measured with sulfuric acid as reference. For a standard curve, 400 µl of rhamnose (0–1 mM) was mixed with 600 µl of anthron solution and treated as described above. Rhamnolipid concentration was calculated based on the assumption of a 2.5:1 rhamnolipid-to-rhamnose mass ratio (Pinzon and Ju, 2009).

Determination of elastolytic activity, cyanide measurements and detection of siderophores

Elastolytic activity of co-culture supernatants was determined with the elastin-congo red (ECR; Sigma) assay (Ohman et al., 1980). A 100 µl aliquot of co-culture supernatant was mixed with 900 µl of ECR buffer (100 mM TRIS/HCl, 1 mM CaCl2, pH 7) containing 8 mg of ECR and incubated with shaking for 21 h at 37°C. The remaining insoluble ECR was removed by centrifugation, and the absorption of the supernatant was measured at 495 nm. ECR buffer was used for the negative control. Cyanide levels were determined using Merckoquant cyanide test strips (Merck). Pyoverdine and pyochelin were qualitatively detected in co-culture supernatants using a fluorescence spectrophotometer (LS50B; Perkin Elmer).

Bioassay for AHL production

Chitin-grown co-cultures of strains AH-1N and PAO1 and chitin-grown single cultures of strain AH-1N were tested for AHL production in bioassays with the bioluminescent E. coli strains pSB401, pSB536 and pSB1075 (Winson et al., 1998; Yates et al., 2002) as described previously (Styp von Rekowski et al., 2008). At each sampling time aliquots of the co-cultures were centrifuged at maximum speed for 15 min and 50 µl of the supernatant was used in the assay.

Oxygen consumption measurements

Cells of strains AH-1N and PAO1 were grown in medium B with 10 mM GlcNAc as described above and harvested in late exponential phase by centrifugation at 8300 g for 10 min at 20°C. Cells were washed once with 50 mM potassium phosphate buffer (pH 7) and finally resuspended in the same buffer to obtain an OD600 of about 20. For oxygen consumption measurements, cell suspensions were diluted with pre-warmed phosphate buffer to an OD600 of 1 (equivalent to a dry weight of about 0.25 mg) in the reaction chamber of a Clark-type electrode (Rank Brothers). After a constant basal oxygen uptake rate was observed, 10 mM GlcNAc, 0.5 mM KCN (50 mM stock solution in 20 mM NaOH) and pyocyanin at different concentrations were successively added to a final volume of 1 ml. Each addition was made after constant oxygen uptake rates had established.

Enzyme assays and protein determination

Activity of aconitase (EC 4.2.1.3) was measured in cell extracts of strain AH-1N by monitoring the formation of cis-aconitate from isocitrate in a spectrophotometer at 240 nm as described earlier (Hausladen and Fridovich, 1996). For preparing cell extracts, cells of strain AH-1N were grown in medium B with 10 mM GlcNAc as described above and harvested in the mid-exponential phase. For this, cultures were poured into pre-cooled plastic tubes and supplied with 1 mM dithiothreitol (DTT) and 0.5 mM MnCl2 and centrifuged at 8300 g for 10 min at 4°C. Cells were washed with ice-cold 90 mM TRIS/HCl buffer (pH 8) containing 1 mM DTT and 0.5 mM MnCl2 and finally resuspended in the same buffer to obtain an OD600 of about 20. These cell suspensions were transferred into serum bottles with butyl rubber septa. After exchanging the headspace with nitrogen gas, cells were broken in a pre-cooled French press under anoxic conditions as described earlier (Philipp and Schink, 1998). Cell extracts were separated from cell debris by centrifugation at 20 000 g for 20 min at 4°C and dispensed into glass vials sealed with butyl rubber septa under anoxic conditions. Anoxic cell extracts were constantly kept on ice and immediately used for determination of aconitase activity. Assay mixtures contained anoxic 90 mM TRIS/HCL buffer (pH 8), 25–50 µl of cell extract (approximately 0.1 mg of protein) and were started by the addition of 20 mM dl-isocitrate. To investigate the influence of pyocyanin on aconitase activity, assay mixtures were supplied with 30 µM NADH and 10 µM pyocyanin or the respective amount of methanol as a solvent control before starting the reaction with isocitrate. Extracts from cells that were used for monitoring the degradation of GlcNAc in the absence and presence of pyocyanin (see above) were prepared in the same way. In these extracts, also the activity of isocitrate dehydrogenase (EC 1.1.1.42) was measured by monitoring the isocitrate-dependent formation of NADPH in a spectrophotometer at 365 nm (ε = 3.4 mM−1 cm−1). Assay mixtures contained anoxic 90 mM TRIS/HCL buffer (pH 8), 1 mM NADP+, 25–50 µl of cell extract (approximately 0.1 mg of protein) and were started by the addition of 10 mM dl-isocitrate.

Activity of pyruvate dehydrogenase (EC 1.2.4.1) was determined by monitoring the CoA- and pyruvate-dependent formation of NADH in a spectrophotometer at 365 nm (ε = 3.4 mM−1 cm−1). Extracts were prepared from cells that were used to monitor the degradation of GlcNAc in the presence and absence of pyocyanin in the same way as described above but with the following differences: cells were washed with 50 mM TRIS/HCl buffer (pH 7.6) and disruption by French press as well as further processing was performed under oxic conditions. Assay mixtures contained 50 mM TRIS/HCl buffer (pH 7.6), 2 mM MgCl2, 2.5 mM DTT, 0.4 mM thiamine pyrophosphate, 0.1 mM CoA, 5 mM NAD+, 50 µl of cell extract (approximately 0.1 mg of protein) and were started by the addition of 5 mM pyruvate.

Protein concentrations in cell extracts were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific).

Acknowledgements

The authors like to thank Paul Williams, Steve Diggle and Miguel Cámara (Nottingham) for the gift of the P. aeruginosa strains ΔpqsA, ΔpqsH, ΔpqsR, ΔlecA, ΔlecB as well as for helpful discussions. Furthermore, the authors like to thank Pierre Cornelis (Brussels) for the gift of the ΔpvdpchEF mutant, Susanne Fetzner (Münster) for the gift of PQS and HHQ, and David Schleheck (Konstanz) for the gift of the P. aeruginosa strains PAO1(ATCC) and PAO1(ATCC)ΔlasIΔrhlI. Katharina Styp von Rekowski is acknowledged for initial experiments on this project. Technical support from Kathrin Happle and continuous support from Bernhard Schink is acknowledged. This work was funded by a grant of the Deutsche Forschungsgemeinschaft (project B9 in SFB 454) to B.P.

Ancillary