Multiple phenotypic alterations caused by a c-type cytochrome maturation ccmC gene mutation in Pseudomonas aeruginosa

Type c cytochromes, involved in anaerobic and aerobic respiration, have their haem covalently linked to the apocytochrome (Thöny-Meyer, 1997). In α- and γ-proteobacteria, the biosynthesis of c-type cytochromes is mediated by cytoplasmic membrane proteins, called Ccm (cytochrome c maturation), which covalently attach haem to the CXXCH motif of apocytochromes in the periplasm (Thöny-Meyer, 1997, 2000; Cianciotto et al., 2005). The most common arrangement found in γ-proteobacteria is for all ccm genes to be in one cluster and transcribed in the same orientation, with Shewanella oneidensis as an exception (Cianciotto et al., 2005). CcmA and CcmB presumably form an ABC transporter together, and it has been proposed that CcmAB transports a reductant in the periplasm (Stevens et al., 2004). CcmC, an integral membrane protein (Ahuja & Thöny-Meyer, 2003; Gaballa et al., 1998), binds haem c in the periplasm and transfers it to the membrane-anchored periplasmic haem chaperone CcmE (Thöny-Meyer, 2003), with CcmD stabilizing their interactions (Ahuja & Thöny-Meyer, 2005). From CcmE, haem is transferred to the haem lyase, CcmF, which is associated with CcmH (Ahuja & Thöny-Meyer, 2003). CcmH keeps the cysteines in the apocytochrome reduced, while CcmG (DsbE) is a protein thiol/disulfide oxidoreductase that passes its electrons to CcmH (Edeling et al., 2004).

There are a growing number of reports that describe pleiotropic effects of ccm mutations that cannot be explained only in terms of the absence of synthesis of c-type cytochromes (reviewed by Cianciotto et al., 2005; Yurgel et al., 2007). Negative effects of mutations in some ccm genes on the production and/or uptake of siderophores in different bacteria have been described (Cianciotto et al., 2005), including the pseudomonads Pseudomonas fluorescens (Gaballa et al., 1996, 1998; Baysse et al., 2002, 2003) and Pseudomonas aeruginosa (De Chial et al., 2003). In P. fluorescens, inactivation of ccmC causes a drastic reduction in the production of the fluorescent siderophore pyoverdine (Gaballa et al., 1996, 1998; Baysse et al., 2002, 2003). The same ccmC mutant is also defective in the utilization of not only pyoverdine but also other heterologous siderophores (Baysse et al., 2003). Ferribactin, the non-fluorescent precursor of pyoverdine (Mossialos et al., 2002), is accumulated by the ccmC mutant, which fails to mature the pyoverdine chromophore (Baysse et al., 2002). In P. fluorescens ATCC 17400, a double pyoverdine-deficient/ccmC mutant also fails to produce the secondary siderophore quinolobactin (Matthijs et al., 2004; Baysse et al., 2003). However, no reduction of pyoverdine production is observed for two different ccmF mutants of Pseudomonas putida GB-1 (de Vrind et al., 1998). The ccmC mutation in P. fluorescens also causes the accumulation of the haem precursor protoporphyrin IX (PPIX) and haem depletion (Baysse et al., 2003). A hypothesis has been formulated in which the combination of haem shortage and PPIX accumulation leads to oxidative stress, since catalases are haemoproteins and PPIX generates reactive oxygen species (ROS) (Baysse et al., 2003; Cianciotto et al., 2005). One possible consequence of this continuous oxidative stress is the destruction of iron–sulphur clusters in proteins that have them in their active sites (Imlay, 2006). Interestingly, Matzanke et al. (2004) have described an [Fe–S] protein, FhuF, as being a ferrisiderophore reductase. If some ferrisiderophore reductases are [Fe–S] oxidoreductases, it could explain why a ccmC mutant is unable to use ferrisiderophores as a source of iron. The effects of different ccm mutations have not, to our knowledge, been investigated in P. aeruginosa. We therefore decided to generate mutants in different ccm genes in P. aeruginosa in order to assess their impact on the physiology of this important opportunistic pathogen that is characterized by the production of a rich array of virulence factors, among which are siderophores (Lamont et al., 2002). P. aeruginosa is a well-studied model for cell to cell communication, genomics and virulence. It also displays a range of signalling molecules and iron chelators that are not found in P. fluorescens; therefore, investigation of the impact of inactivation of ccm genes in this strain will yield significant new data to understand the pleiotropic role of some of these proteins in the physiology of the cell.

In this work, we show that a single non-polar mutation in ccmC and, to a lesser extent, in ccmF causes multiple phenotypic changes in P. aeruginosa, including reduced pyoverdine production, reduced motility and decreased production of the phenazine pigment pyocyanin. Furthermore, we could confirm the destructive effect of the mutation on [Fe–S] clusters and the absence of catalase activity, a hypothesis formulated in a previous publication (Baysse et al., 2003).

Growth of micro-organisms.
P. aeruginosa PAO1 and its mutants (listed in Table 1) were grown at 37 °C in LB or in casamino acids (CAA) medium in the absence or presence of 50 µM FeCl₃ (Cornelis et al., 1992). To further increase iron limitation, the strong iron (III) chelator ethylenediaminedihydroxyphenylacetic acid (EDDHA; 1 mg ml^–1) was added to the CAA medium. Antibiotics were added to P. aeruginosa strains at the following concentrations: 300 µg chloramphenicol (Cm) ml^–1, 100 µg tetracycline (Tc) ml^–1, and 100 µg gentamicin (Gm) ml^–1. Escherichia coli strains were grown at 37 °C in LB with the appropriate antibiotics: 100 µg ampicillin (Amp) ml^–1, 50 µg Cm ml^–1, 25 µg spectinomycin (Sp) ml^–1 and 15 µg Tc ml^–1.

Table 1. Strains and plasmids used in this study

Growth parameters were measured using a Bioscreen apparatus (Life Technologies), with the following parameters: shaking for 20 s every 1 min; reading every 30 min; temperature 37 °C; volume of culture, 300 µl. As the inoculum, an overnight culture of PAO1 in medium was diluted to achieve a final OD₆₀₀ of 0.01. Each culture was grown in triplicate and each experiment was repeated three times. Cells were grown anaerobically (Pettigrew & Brown, 1988) by growing the cultures aerobically to a cell density of ∼30 % of that of the stationary phase, at which point NaNO₃ (5 g l^–1) was added and aeration was stopped by addition of a layer of mineral oil (Gaballa et al., 1996).

Construction of mutants.
Non-polar mutations in ccmA, ccmC, ccmE, ccmF and ccmH were generated using the pKnockout system described by Windgassen et al. (2000). Primers used for the amplification of gene fragments are listed in Table 2. At the 5' end of all forward primers there was an EcoRI site, and a SalI site was present at the 5' end of all reverse primers, except in the case of the ccmH primers (see Table 2). The amplified fragments were cloned between the EcoRI and SalI sites of the pKnockout-G vector. After conjugation and transfer of the recombinant suicide vector to P. aeruginosa, correct insertion and gene inactivation were checked by amplification using the forward primers and the Z2 primer corresponding to the lacZ gene in the pKnockout vector. Further confirmation was obtained by PCR using primers designed to amplify the full-length genes. As a consequence of the insertion of the complete plasmid following the single recombination, amplification should be negative for the mutants because of the length of the insert. The ccmC gene was also inactivated by insertion of a gentamicin cassette, as described previously (Baysse et al., 2003). A list of strains and plasmids used in this study is shown in Table 1 and a list of primers is shown in Table 2.

Table 2. Primers used in this study EcoRI and SalI sites are underlined. For ccmH, the primers contain BamHI (forward) and EcoRI (reverse) sites.

Cytochrome oxidase activity and haem staining.
Cytochrome oxidase activity was determined using the Fluka kit. In the presence of cytochrome oxidase, N,N-dimethyl-p-phenylenediamine oxalate and α-naphthol react to form indophenol blue. Protein fractions for c-type cytochrome analysis were prepared as described by Feissner et al. (2003). Cells from an overnight culture were harvested by centrifugation at 15 000 g for 15 min and resuspended in 1/100 of the original culture volume in 10 mM Tris–HCl (pH 8.0). The cell suspension was then sonicated three times for 4 min each at 4 °C. Unbroken cells were removed by centrifugation at 15 000 g for 15 min at 4 °C and discarded. Protein concentrations were determined by bicinchoninic acid (BCA) assay using BSA as a standard. Protein fractions were prepared for SDS-PAGE by mixing them with DTT-free SDS loading buffer. Samples were separated by 15 % SDS-PAGE and transferred to nitrocellulose for 60 min at 200 mA using a Bio-Rad Mini Trans-Blot cell. A 6.3 mM 3,3',5,5'-tetramethylbenzidine (TMBZ) stock solution was made in methanol immediately prior to staining, and three parts TMBZ stock solution were mixed with seven parts 0.25 M sodium acetate (pH 5.0) to produce the TMBZ staining solution. Electroblotted membranes were incubated in TMBZ staining solution for 30 min at room temperature, followed by the addition of 30 % H₂O₂ to a final concentration of 0.3 % (v/v), and incubated for another 30 min (Goodhew et al., 1990).

Detection of porphyrins.
A 3 ml overnight culture was centrifuged at 5000 g for 5 min. After discarding the supernatants, the pellet was dissolved in one-tenth of the volume (300 µl) of a mixture of acetone/0.1 M NH₄OH (9 : 1, v/v). Samples were again centrifuged and the supernatant was collected. These were analysed with a Shimadzu fluorimeter (excitation, 405 nm; emission, 630 nm).

Motility studies.
Twitching motility was assayed as described by McMichael (1992). LB agar medium (1.3 %) was stab-inoculated with 3 µl culture of the test organism so that the tip came into contact with the bottom of the plate. After 48 h of incubation at 37 °C, the agar was removed and the bottom of the Petri dish was stained with crystal violet (Merck). The swimming and swarming activities were tested in 0.4 and 1 % agar, respectively. A culture was grown to the optimal density (OD₆₀₀=0.5) and 3 µl was spotted on the surface. The plates were incubated at 37 °C for 2 days.

Detection of rhamnolipids.
Rhamnolipids were detected by the drop-collapsing method to detect the presence of wetting agents (Déziel et al., 2003) and by the more specific blue plate assay of Siegmund & Wagner (1991). This assay is based on the formation of an insoluble ion pair between cetyl-trimethylammonium bromide and methylene blue. Positive cultures are surrounded by a dark blue halo. Wells were inoculated with decreasing numbers of cells and the plates were incubated at 28 °C for 48 h, and at room temperature for another 48 h.

Detection of signal molecules.
Supernatants from overnight cultures in LB at 37 °C were extracted with dichloromethane, as described previously (Diggle et al., 2002; Yates et al., 2002) and analysed by TLC. Samples (10 µl) were spotted onto reverse-phase silica RP-18 F₂₅₄ S plates (Merck) and separated using a methanol : water (60 : 40, v/v) system. C4-homoserine lactone (C4-HSL; N-butanoyl-L-homoserine lactone) was assayed using Chromobacter violaceum CV026 as the indicator (McClean et al., 1997). For the detection of 3-oxo-C12 HSL [N-(3-oxododecanoyl)-L-homoserine lactone], samples were spotted onto RP-2 plates (Macherey-Nagel) and the TLC was resolved using a mixture of methanol and water, 45 : 55 (v/v). The Agrobacterium tumefaciens pZLR4 strain, which carries a lacZ transcriptional fusion under the control of an N-acylhomoserine lactone (AHL)-inducible promoter for longer-chain AHLs, was used to detect 3-oxo-C12-HSL (Shaw et al., 1997; Cha et al., 1998).

For both biosensors, AHLs were visualized as bright spots on a dark background when viewed with a Luminograph LB 980 (Berthold) photon video camera. For the extraction of extracellular Pseudomonas quinolone signal (PQS), cell-free supernatants were prepared from overnight cultures (10 ml) and extracted with 10 ml acidified ethyl acetate. The organic phase was dried and resuspended in 50 µl methanol. A 10 µl sample was spotted onto normal-phase silica 60 F₂₅₄ (Merck) TLC plates, pretreated by soaking in 5 % K₂HPO₄ for 30 min and activated at 100 °C for 1 h. Extracts were separated using a dichloromethane : methanol mix, 95 : 5 (v/v). PQS was visualized under UV light and identified by comparison with a synthetic standard (5 µl of a 10 mM stock). PQS and 2-heptyl-4-quinolone (HHQ) could also be visualized by using the indicator strain pqsA : : luxCDABE (Diggle et al., 2006a). Standards of C4-HSL, 3-oxo-C12-HSL and PQS were obtained from P. Williams (University of Nottingham).

Determination of iron–sulphur clusters.
Cells from an LB or CAA culture (OD₆₀₀ 0.4–0.6) were collected by centrifugation (7500 r.p.m. for 10 min at 4 °C) and washed twice with 0.85 % (w/v) NaCl and once with cell storage buffer (100 mM Tris/HCl, pH 7.3, 1 mM DTT, 1 mM PMSF). Cells were ultrasonically disrupted in the cell storage buffer at a power level of 25 % (15 min at 2 s intervals). The supernatant fraction was collected and centrifuged at 180 000 g for 60 min at 4 °C to yield membrane and soluble components. The soluble fraction was further centrifuged for 2 h at the same speed to yield a membrane-free system. Protein concentration was measured and a spectrum of the cell extracts between 350 and 550 nm was taken (Middaugh et al., 2005).

Chrome azurol (CAS) assay for the detection of siderophores.
The presence of siderophores was determined by a universal chemical method using the high affinity of CAS for iron (III) (Schwyn & Neilands, 1987). Agar plates were made containing CAA with the ternary complex chrome azurol S/iron (III)/hexadecyl trimethylammonium bromide as the indicator. A 5 µl sample of a CAA culture (OD₆₀₀ 0.5), was spotted onto CAS plates and incubated overnight. The diameter of the orange halo was measured the next day.

Detection of pyoverdine by IEF and purification of pyoverdine.
Pyoverdine production was assessed by measuring the OD₄₀₀ of 24 h CAA cultures normalized to the OD₆₀₀ (Höfte et al., 1993). Pyoverdine isoforms were detected by IEF of the supernatants of the same CAA liquid cultures, followed by CAS overlay, as described previously (Koedam et al., 1994).

Detection of pyochelin.
Pyochelin was extracted from overnight CAA culture supernatants and detected using the method of Farmer & Thomas (2004). Aliquots of culture (30 ml) were extracted with 12 ml acidified ethyl acetate. The organic phase was transferred to a fresh tube and dried to completion under a stream of helium gas. The solute was resuspended in 50 µl methanol for further analysis. A spectrum was taken between 200 and 450 nm. A 10 µl sample was spotted onto a normal-phase silica 60F₂₅₄ (Merck) TLC plate. Extracts were separated using a chloroform/acetic acid/ethanol mix (90 : 5 : 2.5). The plate was visualized using a UV transilluminator and photographed. The pyochelin levels in culture supernatants were estimated by comparing them with ethyl acetate extracts of other strains with mutations in different pyochelin biosynthesis genes (Table 1). Stereoisomeric forms of pyochelin were detected under UV and by measuring the R_F value after spraying the plate with a solution of FeCl₃ (0.3 M in 0.1 M HCl). LC/MS analyses were performed on a Kontron 325 system, coupled to the mass spectrometer and equipped with a UV detector (model 322), an automatic injector (model 465) and LC-6A type pumps. The column used was a Vydac 218TP54 RP column (C18, 5 µm, diameter 0.46 cm, length 25 cm) and a flow rate of 1 ml min^–1 was maintained. MS data were recorded on a VG Quattro II spectrometer (electrospray ionization, cone voltage 70 V, capillary voltage 3.5 kV, source temperature 80 °C). Data collection was done using the Masslynx software.

Assay for pyocyanin production.
Pyocyanin was visualized by growing bacteria for 48 h on P agar (Difco). Pyocyanin was extracted as described by Mavrodi et al. (2001). Briefly, the agar was collected in a Falcon tube and 10 ml chloroform was added per 12.5 g agar medium. The phenazine pigment was extracted for 2 h at 37 °C, after which 2 ml 0.5 M HCl was added and the mixture shaken vigorously. The pink top layer was removed and its A₅₂₀ was measured.

Catalase detection.
Protein extraction was performed as described in Feissner et al. (2003), and the resulting protein extracts were analysed by 7.5 % SDS-PAGE with a DTT-free loading buffer. Catalase activity was detected as described by Wayne & Diaz (1986).

Construction of P. aeruginosa mutants in different ccm genes
Using the pKnockout vector delivery system for rapid gene inactivation (Windgassen et al., 2000), we obtained mutants in the ccmA, ccmC, ccmE, ccmF and ccmH genes. No mutant could be obtained in ccmG. A mutant initially identified as being in ccmB had the vector inserted in an xcp gene and was therefore eliminated. All other mutants were verified by analysis of the sequence of the fragment amplified with the corresponding forward and Z2 primers (Table 2). Using primers corresponding to the 5' and 3' ends of the different genes, we could obtain amplification of ccmA, ccmC, ccmE, ccmF and ccmH for the wild-type, but not in the case of the corresponding mutants. As control for the quality of DNA we used primers for oprL, and this gene was amplified in the wild-type and the mutants (results not shown). This is due to the presence of the large pKnockout insert in the interrupted gene, giving a fragment that cannot be amplified by Taq polymerase (results not shown). In order to make comparisons with the previously described ccmC gene mutant in P. fluorescens ATCC 17400, which was obtained by the insertion of a gentamicin cassette, we also inactivated the P. aeruginosa ccmC gene via insertion of a gentamicin cassette, as previously described (Baysse et al., 2003). This last mutant was termed ccmC1 and had exactly the same phenotype as the pKnockout ccmC mutant (see below). Unless specifically mentioned, the results are given for the pKnockout ccmC mutant.

Growth characteristics
Delayed growth in LB medium was observed for the ccmC and ccmF mutants, while the other mutants grew like the wild-type (see Fig. 1a for wild-type, ccmC and ccmF). The wild-type strain reached stationary phase after 14 h, the ccmF mutant after 24 h, and the ccmC mutant after 48 h of growth. In the iron-limiting CAA medium and in the presence of the strong iron (III) chelator EDDHA, the growth of the ccmF mutant was further retarded, but the growth of the ccmC mutant was almost completely abolished (Fig. 1b). Addition of pyoverdine to the CAA plus EDDHA medium partially restored the growth of the ccmF mutant, but had only a small effect on the growth of the ccmC mutant (Fig. 1c). From these results we can say that the two ccmC mutants obtained via different gene inactivation procedures behave very similarly, and that the ccmF and ccmC mutations affect growth in all media tested, including CAA and CAA plus iron (results not shown). Furthermore, pyoverdine had only a small stimulatory effect on the growth of the ccmC mutant and only partially restored the growth of the ccmF mutant.

(24K): Fig. 1. Growth of wild-type (•), ccmC () and ccmF () in LB medium (a), CAA plus 1 mg EDDHA ml–1 (b), and CAA plus 1 mg EDDHA ml–1 and 50 µM pyoverdine (c).

Cytochrome c and porphyrin production
All mutants, except ccmC and ccmF, were still oxidase positive, suggesting that they were still able to produce c-type cytochromes (results not shown). Accumulation of porphyrins was analysed by fluorimetry, and was most strongly evidenced for the ccmC and ccmF mutants (Fig. 2a). The ccmC and ccmF mutants accumulated a porphyrin with a peak of emission at 637 nm (Fig. 2a, results shown only for ccmC). When the P. aeruginosa ccmC mutant was complemented with the P. fluorescens ATCC 17400 ccmC gene (Baysse et al., 2003), the porphyrin peak was strongly reduced (Fig. 2a). The combination of TMBZ and H₂O₂ is used as a stain for the peroxidase activity of cytochrome c in SDS polyacrylamide gels. This reagent can be used to detect very low levels of haem-associated peroxidase activity (Goodhew et al., 1990). Haem staining after SDS-PAGE confirmed the presence of c-type cytochromes in all mutants, except in the ccmC mutant, while the intensity of stained c-type cytochrome bands was strongly reduced for the ccmF mutant (Fig. 2b).

(52K): Fig. 2. (a) Detection of porphyrins in total cell extracts of wild-type (•), ccmC (), and ccmC complemented by the ccmC gene from pPyov35 (+). (b) Haem staining of c-type cytochromes after SDS-PAGE. Lanes: 1, pre-stained molecular mass markers; 2, total extract from wild-type; 3, total extract from ccmC mutant; 4, total extract from ccmF mutant.

Production of siderophores
All mutants produced pyoverdine to the same extent as the wild-type, except for the two ccmC mutants, for which pyoverdine production was found to be drastically reduced, as evidenced by the colour of the CAA culture supernatant and the absence of fluorescence under UV (Fig. 3a, results shown only for the wild-type and the ccmC mutant). Quantitative pyoverdine measurements gave an OD₄₀₀ between 3 and 3.5 for stationary-phase cultures for all tested strains, except in the case of the two ccmC mutants, for which this value never exceeded 0.05. On CAS plates, however, high levels of production of siderophores by the ccmC mutant were evident (Fig. 3b), and the halo produced around the inoculated bacteria was similar in size and colour to the one produced by a pvdD pyoverdine-negative mutant which still produces the second siderophore pyochelin. Detection of siderophore activity by IEF followed by CAS overlay showed that the ccmC and ccmF mutant siderophore patterns were different from those of the other strains and the wild-type (Fig. 3c). To investigate whether the ccmC mutant siderophore activity seen on CAS agar plates is due to pyochelin, this siderophore was detected by LC/MS. Two peaks were detected by HPLC, one eluting at 19.8 min and the other at 21.1 min, for both wild-type and the ccmC mutant (results not shown). The compound eluting at 19.8 min had a mass of 325 Da and the 21.1 min compound had a mass of 326 Da, close to the expected mass for the two interconvertible pyochelin isomers (324.1 Da) which account for the two different peaks in HPLC (Schlegel et al., 2004). The ccmC mutant produced about twice the amount of pyochelin compared to the wild-type (results not shown).

(42K): Fig. 3. Production of siderophores. (a) Culture supernatants of wild-type (left) and ccmC mutant (right) grown in CAA for 24 h. The picture on the right shows the fluorescence of the same tubes when exposed to UV light. (b) Comparison of the discoloration haloes around inoculated wild-type (top), pvdD pyoverdine-negative mutant (middle right), a pyochelin-negative pchEF mutant (centre), a double pchEF pvdD mutant (bottom), and the ccmC mutant (middle left). The plate was incubated for 48 h. (c) IEF separation of semi-purified and concentrated pyoverdines from ccmC (lane 1), wild-type (wt; lane 2) and ccmF (lane 3). IEF was done on Ampholine PAG plates with a pH 3–10 gradient, and the bands were visualized after CAS overlay (see text for details).

Production of pyocyanin
During the course of our experiments we quickly realized that both ccmC mutants also showed a strongly reduced production of the blue-green phenazine pigment pyocyanin on LB or on P-agar medium (Fig. 4). Complementation of the ccmC mutant by the P. fluorescens ATCC 17400 ccmC gene in trans restored the capacity to produce pyocyanin (Fig. 4). Quantification of pyocyanin in P-agar medium showed that the production of this pigment is significantly reduced in the ccmC mutants and the ccmF mutant compared to wild-type (10–12 % and 30–40 % of wild-type levels, respectively; results not shown). The ccmC mutant produced colonies of the same size as the wild-type on P agar (Fig. 4), and even prolonged incubation times (more than 3 days) did not induce pyocyanin production in the ccmC mutant.

(55K): Fig. 4. Production of pyocyanin on P-agar plates incubated for 2 days by wild-type (1), ccmC (2), and ccmC complemented by the ccmC gene from pPyov35 (from P. fluorescens ATCC 17400; 3).

Swarming, swimming and twitching motilities
Since pyocyanin production is a trait known to be controlled by quorum sensing in P. aeruginosa (see Discussion), we decided to look at another quorum sensing-controlled phenotype, motility. As shown in Fig. 5(a), the ccmC mutant, and to a lesser extent the ccmF mutant (results not shown), was much less motile in LB medium (swarming, swimming and twitching). Supplementary Fig. S2a, b shows the motility of wild-type and ccmC mutant cells observed under the phase-contrast microscope.

(65K): Fig. 5. (a) Top pictures: swarming motility of wild-type and ccmC mutant strains grown on 0.4 % agar LB medium; middle pictures: swimming motility of wild-type and ccmC mutant grown on 1 % agar LB medium; bottom pictures: twitching motility of wild-type and ccmC mutant. The strains were grown for 24 h. In all three pictures wild-type and ccmC cultures are on the left- and right-hand plates, respectively. (b) Detection of rhamnolipid production. Plates were incubated at 28 °C for 48 h and at room temperature for another 48 h. Wells were inoculated with the number of cells indicated at the top of the figure. Top, wild-type (WT); middle, ccmC; bottom, ccmF. See text for details.

Detection of rhamnolipids
Rhamnolipids are glycolipids known to be involved in the swarming motility of P. aeruginosa (Caiazza et al., 2005). The drop-collapsing assay was used to measure the wetting capacity of a P. aeruginosa culture supernatant (Déziel et al., 2003). The ccmC mutants produced less rhamnolipid than the other mutants, as judged by the absence of drop-collapsing activity, a phenotype also observed for the ccmF mutant (results not shown). In order to measure more accurately the production of rhamnolipids, we used the blue plate detection method of Siegmund & Wagner (1991). As can be seen in Fig. 5(b), there was a clear reduction of the diameter of the halo around ccmC and ccmF inocula compared to the wild-type. At a cell density of 1.25x10⁸ per well there was no detectable production of rhamnolipids by the ccmC mutant.

Production of signal molecules
Since the phenotypes described above are all quorum sensing-dependent, we also looked at the production of the three signal molecules known to be produced by P. aeruginosa, 3-oxo-C12-HSL, C4-HSL, and PQS. For the detection of C4-HSL, the Chromobacter violaceum CV026 strain was used (McClean et al., 1997), while the Agrobacterium tumefaciens pZLR4 strain, which carries a lacZ trancriptional fusion under the control of an AHL-inducible promoter for longer-chain AHLs, was used to detect 3-oxo-C12-HSL (Shaw et al., 1997; Cha et al., 1998). The results show that the two ccmC mutants produce wild-type levels of the two AHLs. The levels of these two AHLs were also unchanged in all the other strains tested (Supplementary Fig. S1).

For the detection of PQS, extracts from spent medium were separated by TLC and fluorescent spots were detected by UV. PQS could be detected in all extracts. Overlay of the TLC plate by the P. aeruginosa pqsA : : lux indicator strain confirmed that PQS was produced by all strains (results not shown).

Resistance to oxidative stress and [Fe–S] content
Both ccmC and ccmF mutants were also found to be more sensitive to H₂O₂ when grown in LB or in CAA medium with or without iron. The diameter of growth inhibition in LB was 3.57 cm (SD±0.06 cm) for the wild-type and 4.07 cm (SD±0.15 cm) and 6.17 cm (SD±0.06 cm) for the ccmC and ccmF mutants, respectively. In CAA medium, the effect was even more pronounced since the following growth inhibition diameters were observed: 3.17 cm (SD±0.15 cm) for the wild-type, 7.77 cm (SD±0.15 cm) for ccmC, and 8.5 cm (SD±0.06 cm) for ccmF. After SDS-PAGE, no catalase activity could be detected for the ccmC mutant, although a clear band was observed for the cellular and extracellular extracts of the wild-type and the other mutants; in the case of the ccmF mutant the catalase band was present, but very faint (Fig. 6).

(66K): Fig. 6. Detection of catalase activity after SDS-PAGE. Lanes: 1, molecular markers; 2, total cellular extract of wild-type; 3, total cellular extract of ccmC; 4, total cellular extract of ccmC1; 5, total cellular extract of ccmF.

Spectrophotometric measurement revealed that the peak corresponding to [Fe–S] clusters was strongly reduced in ccmC and ccmF (Fig. 7).

(13K): Fig. 7. Detection of [Fe–S] clusters in cell extracts (LB) from wild-type (•), ccmC () and ccmF (). The spectrum was taken between 350 and 550 nm.

Our results confirm the pleiotropic nature of the ccmC mutation (and to a lesser extent of the ccmF mutation) in P. aeruginosa, a feature already observed in other bacteria, in which it has been found that CcmC in particular is important for the utilization of ferrisiderophores as a source of iron (reviewed by Cianciotto et al., 2005). However, to our knowledge, no systematic comparisons of the phenotypic effects of mutations in different ccm genes have been performed to date. Also, it was interesting to find out the effects of these mutations on the opportunistic pathogen P. aeruginosa. We successfully obtained knockouts in the ccmA, ccmC, ccmE, ccmF and ccmH genes of P. aeruginosa PAO1. Unexpectedly, cytochrome c production was not affected in the ccmA, ccmE and ccmH mutants, while it was strongly reduced in the ccmF mutant and completely abolished in the ccmC mutants. This is maybe due to the method of inactivation that we chose. Indeed, the pKnockout system is designed to generate non-polar insertions of the recombinant plasmid after a single recombination event. This means that the genes downstream are transcribed, but also that the 5' end of the gene is transcribed, explaining why the effects are less drastic compared to what has been reported for other mutants (produced by transposon or antibiotic cassette insertion) described in the literature. In agreement with our previous results in P. fluorescens ATCC 17400, the P. aeruginosa ccmC mutant was defective for the production of pyoverdine (Gaballa et al., 1996, 1998; Baysse et al., 2002, 2003). Production of pyoverdine by the ccmC mutant of P. aeruginosa could be restored by complementation with the P. fluorescens ATCC 17400 ccmC gene in trans (Baysse et al., 2003; and results not shown). Interestingly, production of pyochelin was not affected by the ccmC mutation and was even increased. In stark contrast, production of the secondary siderophore quinolobactin could not be detected in the P. fluorescens ATCC17400 ccmC mutant (Baysse et al., 2003). It has been demonstrated previously that haem is necessary for the production of both pyoverdine and quinolobactin (Baysse et al., 2001, 2003). In line with this, the first enzyme in the pathway for quinolobactin synthesis, tryptophan 2,3-dioxygenase, is a haemoprotein, explaining why the ccmC mutation can affect the biosynthesis of this siderophore since it results in a haem shortage (Baysse et al., 2003; Matthijs et al., 2004, 2007). Recently, Matthijs et al. (2007) showed that synthesis of quinolobactin and its precursor thioquinolobactin is strongly impaired in a hemA haem biosynthesis mutant. By IEF and CAS overlay we also detected non-fluorescent isoforms of pyoverdine in our ccmC mutants similar to what was observed by Baysse et al. (2002) in the P. fluorescens ATCC 17400 ccmC knockout strain. This result supports the hypothesis that the maturation of the pyoverdine chromophore from the precursor ferribactin, which takes place in the periplasm, does not occur in the ccmC mutant, because of a lack of oxidative power (Baysse et al., 2002). Recently, it has been shown that at least one pyoverdine biosynthesis enzyme, PvdN, is periplasmic and is exported via the Tat secretion pathway (Voulhoux et al., 2006). In their model, Baysse et al. (2003) also predicted that the observed accumulation of the direct haem precursor, protoporphyrin IX, could generate an oxidative stress, which in turn could result in the destruction of [Fe–S] clusters in proteins having them in their catalytic centres (Imlay, 2006). In this context it is interesting to mention that one siderophore reductase, FhuF, is a [Fe–S] protein (Matzanke et al., 2004). Here we could demonstrate that indeed the ccmC and ccmF mutations cause an accumulation of porphyrins and a drastic reduction of the amount of [Fe–S] clusters. The previously described haem depletion in the P. fluorescens ATCC 17400 ccmC mutant (Baysse et al., 2003) is also confirmed in our research by the absence of the haemoprotein catalase activity (Fig. 6).

To our surprise, the ccmC mutant was also deeply affected in some quorum-sensing-dependent phenotypes since it failed to produce pyocyanin and was non-motile (swarming, swimming and twitching). Interestingly, a recent report described that some mutations in some c-type cytochrome biogenesis genes, including ccmC, in Sinorhizobium meliloti also affect motility (Yurgel et al., 2007). In P. aeruginosa, most virulence factors are regulated via cell-to-cell communication or quorum sensing (Swift et al., 2001; Withers et al., 2001; Bassler, 2002; Cámara et al., 2002), and in a growth phase-dependent manner (Diggle et al. 2002, 2003). P. aeruginosa has two AHL-dependent quorum-sensing systems, the las and rhl systems (Winzer & Williams, 2001; Lazdunski et al., 2004). LasI is the 3-oxo-C12-HSL synthase, and LasR is its cognate response regulator, while the rhl system involves the RhlI C4-HSL synthase and the response regulator RhlR. Separately or together, the las and rhl systems regulate the production of extracellular virulence determinants such as elastase, the LasA protease, alkaline protease and exotoxin A, and the production of rhamnolipids, hydrogen cyanide, pyocyanin, siderophores and the cytotoxic lectins PA-IL and PA-IIL (Diggle et al., 2002; Winson et al., 1995; Lazdunski et al., 2004). The las system exerts transcriptional control over the rhl system (Lazdunski et al., 2004), but the rhl system can also function independently of the las system (Diggle et al., 2003). A third signal molecule is characterized as 2-heptyl-3-hydroxy-4-quinolone and termed PQS (reviewed by Diggle et al., 2006b). Synthesis of PQS depends on the pqsABCDE locus, which is responsible for generating multiple 4-hydroxyquinolones, including HHQ, the immediate PQS precursor, which is converted into PQS by PqsH, a putative periplasmic monoxygenase (Déziel et al., 2004). Transcription of the pqsH gene is under the control of the las system (Gallagher et al., 2002), which suggests that PQS synthesis depends on the production of 3-oxo-C12-HSL. However, LasR-independent production of PQS has also been demonstrated (Diggle et al., 2003). On the other hand, McGrath et al. (2004) have demonstrated that the rhl system represses PQS production. It therefore seems that the PQS regulon is intertwined in the AHL quorum-sensing systems of P. aeruginosa. PQS is required for the production of pyocyanin in P. aeruginosa (Gallagher et al., 2002; Diggle et al., 2002, 2003; Dietrich et al., 2006; Price-Whelan et al., 2006).

Nevertheless, in our experiments the production of signal molecules was not really affected by any of the mutations. The ccm genes are not themselves subject to control by the quorum-sensing signal molecules AHL and PQS (Wagner et al., 2003; Schuster et al., 2003; Hentzer et al., 2003; Bredenbruch et al., 2006). The question remains why CcmC, and to a lesser extent CcmF, are important for functions other than c-type cytochrome biogenesis, and why phenotypes such as motility and pyocyanin production are affected. One possible explanation is that pyoverdine, which is a signal molecule as well (Lamont et al., 2002; Visca et al., 2002, 2007), also controls traits such as motility and phenazine pigment production. Others have made the interesting observation that pyoverdine-negative mutants are unable to form mature biofilms, suggesting that pyoverdine acts not only as a siderophore but also as an important signal (Banin et al., 2005). Accordingly, a very recent report from Matilla et al. (2007) has described the importance of pyoverdine for the motility of P. putida KT2440. Another hypothesis, not excluding the first one, is that CcmC is required for the oxidative power of the periplasm, as suggested by Baysse et al. (2002). Pyocyanin and 4-quinolones are also redox-active molecules which could play a role in the electron transport chain under aerobic conditions (reviewed by Price-Whelan et al., 2006). It seems likely therefore that CcmC is a key component for the normal functioning of aerobic respiration and that its absence negatively influences many pathways in P. aeruginosa, which makes it an interesting target for future drug development.

We thank Professor G. Laus from the Vrije Universiteit Brussel for the LC/MS analysis of pyochelin.

Edited by: W. Bitter