Differences in two Pseudomonas aeruginosa cbb3 cytochrome oxidases

Authors

  • James C. Comolli,

    1. Department of Bacteriology, University of Wisconsin-Madison, 420 Henry Mall, Madison, WI 53706, USA.
    Search for more papers by this author
    • Present address: Johnson and Johnson, Ethicon, Somerville, NJ, 08876, USA.

  • Timothy J. Donohue

    Corresponding author
    1. Department of Bacteriology, University of Wisconsin-Madison, 420 Henry Mall, Madison, WI 53706, USA.
    Search for more papers by this author

E-mail tdonohue@bact.wisc.edu ; Tel. (+1) 608 262 4663; Fax (+1) 608 262 9865.

Summary

Bacterial cytochrome cbb3 oxidases are members of the haeme-copper oxidase superfamily that are important for energy conservation by a variety of proteobacteria under oxygen-limiting conditions. The opportunistic pathogen Pseudomonas aeruginosa is unusual in possessing two operons that each potentially encode a cbb3 oxidase (cbb3-1 or cbb3-2). Our results demonstrate that, unlike typical enzymes of this class, the cbb3-1 oxidase has an important metabolic function at high oxygen tensions. In highly aerated cultures, cbb3-1 abundance and expression were greater than that of cbb3-2, and only loss of cbb3-1 influenced growth. Also, the activity of cbb3-1, not cbb3-2, inhibited expression of the alternative oxidase CioAB and thus influenced a signal transduction pathway much like that found in the α-proteobacterium Rhodobacter sphaeroides. Cbb3-2 appeared to play a more significant role under oxygen limitation by nature of its increased abundance and expression compared to highly aerated cultures, and the regulation of the cbb3-2 operon by the putative iron-sulphur protein Anr. These results indicate that each of the two P. aeruginosa cbb3 isoforms have assumed specialized energetic and regulatory roles.

Introduction

Respiration can be conceptually described as a series of redox reactions that conserve energy in the form of a transmembrane electrochemical proton gradient. Under aerobic conditions, membrane-bound cytochrome oxidases usually catalyse the terminal redox reaction of respiration, a four-electron reduction of oxygen to water. The single cytochrome oxidase of mitochondria is composed of a catalytic core bound by mitochondrially encoded polypeptides which are associated with several nuclear-encoded proteins that regulate the enzyme function (Barrientos et al., 2002). In contrast, prokaryotes often contain multiple cytochrome oxidases, each with a distinct respiratory efficiency, to modulate aerobic respiratory activity (Poole and Cook, 2000; Richardson, 2000). By altering the level of these and other respiratory enzymes, bacteria can conserve energy in widely varying environmental conditions.

The bacterial oxidases are often differentiated by their preference for reduced quinol or cytochrome as an electron donor, their subunit composition, and the metal centres at their active site (Pereira et al., 2001). Despite these differences, the catalytic core of many bacterial oxidases within the haeme-copper oxidase superfamily, including the mitochondrial cytochrome oxidase, all possess a haeme-copper metallocentre at their active site (Garcia-Horsman et al., 1994; van der Oost et al., 1994). One of the recently characterized members of this superfamily, the proteobacterial cytochrome cbb3 oxidase, was initially identified as the product of a rhizobial operon (fixNOQP) that is required for nitrogen fixation (Preisig et al., 1993; Mandon et al., 1994). Similar operons have since been characterized in many non-diazotrophic proteobacteria, where they have been given the designation of ccoNOQP (Thony-Meyer et al., 1994; Myllykallio and Liebl, 2000). The catalytic subunit of cbb3 oxidases, CcoN, contains a haeme-copper active site, but CcoN associates with mono- (CcoO) and dihaeme (CcoP) c-type cytochromes rather than the CuA-containing subunit II found in other members of the haeme-copper superfamily (Zufferey et al., 1996; Pitcher et al., 2002). The cbb3 enzymes also possess a fourth subunit, CcoQ, that may contribute to stability of the catalytic core (Zufferey et al., 1996; Oh and Kaplan, 2002).

Studies of several bacterial cbb3 oxidases have indicated that these enzymes are primarily expressed under oxygen limitation and are critical for respiration in microaerobic conditions. For instance, the fixNOQP locus was essential for rhizobia to colonize the oxygen-depleted plant rhizosphere presumably because the cbb3 oxidase is needed to generate sufficient energy for the ATP-demanding process of nitrogen fixation at the low oxygen tensions found in root nodules (Preisig et al., 1993; Delgado et al., 1998). As predicted by this notion, the Bradyrhizobium japonicum cbb3 oxidase has an oxygen affinity close to that measured in root nodules (Preisig et al., 1996), and expression of B. japonicum fixNOQP was increased at low oxygen tensions (Batut et al., 1989; Fisher, 1994). Cbb3 oxidases of other proteobacteria have also been found to have a high oxygen affinity and similar expression pattern (Mandon et al., 1994; Otten et al., 2001; Swem et al., 2001), which lends support to the importance of these enzymes under oxygen limitation. This concept has been extended to several clinically relevant bacteria that contain a cbb3 oxidase, including Vibrio cholera, Bordetella pertussis and Helicobacter pylori, which may utilize this enzyme to inhabit microaerobic niches within a host (Myllykallio and Liebl, 2000; Smith et al., 2000).

Though a primary role of cbb3 oxidases is in energy generation, these enzymes have also been implicated in the regulation of bacterial gene expression. In facultative phototrophic α- proteobacteria like Rhodobacter sphaeroides, the transcription of several loci that are normally induced under oxygen limitation was stimulated by inactivation of the cbb3 oxidase or by respiratory inhibitors that blocked cbb3 activity (O’Gara and Kaplan, 1997; Oh and Kaplan, 1999; 2000). This effect was dependent upon a two-component regulatory system that includes the sensor kinase PrrB and the response regulator PrrA (O’Gara et al., 1998), which are thought to transduce a signal from the respiratory chain to the transcription apparatus (Oh and Kaplan, 2001). The current model predicts that when oxygen becomes scarce, the corresponding change in the oxidation-reduction state of the respiratory chain stimulates PrrBA function to alter expression of several loci involved in microaerobic processes.

Completion of the genome sequence of the opportunistic pathogen Pseudomonas aeruginosa revealed that this γ-proteobacterium contained a pair of ccoNOQP operons (Fig. 1B) (Croft et al., 2000; Stover et al., 2000). Though genes encoding a cbb3 oxidase were detected in P. aeruginosa nearly a decade ago (Thony-Meyer et al., 1994), the presence of two separate loci suggested a re-evaluation of the importance of the cbb3 oxidases in the aerobic lifestyle of this bacterium (Fig. 1A). In investigating the roles of these two enzymes, we unexpectedly found that each of the two P. aeruginosa cbb3 oxidases differed in their respiratory and regulatory functions and that the cbb3-1 enzyme had an important role even in highly aerated cultures. Also, our findings indicate that the pattern of expression of the P. aeruginosa cbb3 oxidases, and not simply their subunit composition, plays a significant role in determining the contribution of each isoform to cellular energy generation.

Figure 1.

A. The predicted components of the P. aeruginosa aerobic respiratory system. In addition to cytochrome c -oxidoreductase (the cyt bc1 complex), this bacterium has five terminal oxidases, two that are predicted to use quinol as a substrate and three that use cytochrome c as an electron donor. Most dehydrogenases, including NADH dehydrogenase, reduce quinone and donate electrons to any terminal oxidase. The figure indicates that artificial electron donors, such as TMPD/ascorbate, can directly reduce cytochrome c (dotted line) and in turn cytochrome oxidases.
B. Map of the P. aeruginosa locus that contains two adjacent ccoNOQP operons. These gene products were designated according to their position on the PAO1 chromosome; open reading frames PA1552–1554 were designated ccoNOQP-1 and PA1555–1557 were named ccoNOQP-2. These gene numbers were provided in the most recent annotation of the PAO1 genome, though the small open frame ccoQ was not annotated (http://www.pseudomonas.com). The regions deleted in creation of mutant strains lacking cbb3-1 (Cco1.1), cbb3-2 (Cco2.2), or both cbb3 oxidases (CcoΔ1) are indicated (dotted lines). The map is drawn to scale (the insert depicts 500 bp); it shows the 365 bp separating the two cco loci and the location of the transcriptional start site that was mapped for each set of genes (see text).

Results

Each of the two predicted P. aeruginosa cbb3 oxidases is expressed

Each of the two adjacent P. aeruginosa ccoNOQP operons, designated ccoNOQP-1 or -2 according to their order on the PAO1 chromosome, encode the four canonical cbb3 oxidase subunits (Fig. 1B) and were therefore each capable of producing a functional oxidase. To assess the presence of individual cbb3 oxidases in P. aeruginosa wild-type cells, we used haeme peroxidase staining of cell extracts from highly aerated cultures. The two CcoP subunits had a significant enough difference in their predicted masses (34.6 kDa for CcoP1 versus 33.8 kDa for CcoP2) to allow the differentiation of individual c-type cytochrome subunits by haeme peroxidase staining after SDS–PAGE (see below). The c-type haeme of CcoO can also be detected by haeme peroxidase staining, but the predicted sizes of CcoO1 and CcoO2 are too close to allow their resolution by mobility on SDS–PAGE. However, the abundance of CcoO can be used as a reporter of cbb3 oxidase abundance as this subunit was found to be required for stability of the enzyme (Zufferey et al., 1996).

This analysis revealed that proteins with the predicted molecular weights of the two CcoP subunits were present in highly aerated wild-type cells (Fig. 2A). Examination of P. aeruginosa strains with a disruption in the locus encoding cbb3-1 (Cco1.1), cbb3-2 (Cco2.2), or both cbb3 enzymes (CcoΔ1) confirmed this notion. The c-type cytochrome of ∼34 kDa was identified as CcoP1 because of its greater predicted mass and its absence in cells lacking ccoNOQP-1 (Cco1.1), whereas the faster migrating species was attributed to CcoP2 by its loss in cells predicted to lack the cbb3-2 oxidase (Cco2.2). As expected, no CcoP or CcoO were detected in CcoΔ1, a strain with both ccoNOQP loci inactivated. In contrast, levels of CcoP1, CcoP2, and total CcoO were not significantly changed in BC1.1, a strain that lacks the ubiquinol/cytochrome c oxidoreductase (the cyt bc1 complex). However, BC1.1 was missing several c-type cytochromes, including one with a mass similar to that predicted for cytochrome c1 (PA4463).

Figure 2.

Haeme peroxidase analysis of the c -type cytochromes present in membranes of P. aeruginosa wild-type or respiratory mutant strains. Strains were harvested after growth in highly aerated (A) or semi-aerobic (B) conditions and 40 µg of cell extract protein was separated by SDS–PAGE before staining for haeme-dependent peroxidase activity. Proteins with molecular weights corresponding to those of the cbb3 oxidase subunits CcoP1, CcoP2, and CcoO are indicated.

These assays also suggested that the abundance of CcoP1 was greater than that of CcoP2 in highly aerated cultures (Fig. 2A). Cell extracts from the Cco1.1 mutant contained significantly less CcoO than wild-type whereas those from Cco2.2 had only a small reduction in the level of CcoO. Though quantification from haeme-peroxidase staining is inexact, these observations suggested that cbb3-1 was more prevalent than cbb3-2 in highly aerated P. aeruginosa cultures. If this were the case, transcription of ccoNOQP-1 would be expected to be higher than ccoNOQP-2 in highly aerated cultures. To test this hypothesis, we monitored β-galactosidase activity from highly aerated cultures of wild-type cells containing a lacZ reporter plasmid that lacked a cco promoter (pLP170) or one that contained several hundred base pairs upstream of the ccoNOQP-1 (pJC753) or ccoNOQP-2 (pJC754) transcription initiation site (data not shown). This analysis showed that β-galactosidase activity in highly aerated cultures containing the ccoNOQP-1 reporter plasmid (pJC753) was roughly 1150 units, ∼sevenfold higher than from ccoNOQP-2 (∼160 units) and well above that of cells containing the control reporter plasmid that lacks a cco promoter region (∼35 units) (Fig. 3). Thus, increased transcription of the ccoNOQP-1 operon likely accounts for the higher abundance of cbb3-1 subunits that was evident by haeme-peroxidase staining of extracts from highly aerated cultures (Fig. 2A). This was confirmed by primer extension analysis of each cco operon in RNA extracted from highly aerated cultures (data not shown).

Figure 3.

β-Galactosidase levels from a reporter plasmid containing no cco promoter (pLP170), the ccoNOQP -1 promoter (pJC753), or the ccoNOQP -2 promoter (pJC754) fused to lacZ in wild-type cells or a mutant lacking Anr. Strains were grown in either highly aerated cultures (black bars) or under semi-aerobic conditions (grey bars).

These observations also indicate that the two ccoNOQP operons are independently regulated from distinct promoters, which is consistent with the lack of effect of a polar insertion into ccoNOQP-2 on cbb3-1 subunit expression (Fig. 2) and primer extension analysis of the region (see Experimental procedures). The 365 bp distance between the ccoP2 stop codon and the ccoN1 start codon and the presence of a potential transcription terminator 12 bp downstream from the ccoP2 stop codon (ACCCGCGCCC . . . GGGCGCGGGT) also support this premise.

Contribution of the cbb3 oxidases to P. aeruginosa aerobic respiration

To determine the role of each cbb3 oxidase in P. aeruginosa, the growth rates of the wild-type and mutants strains were determined. In highly aerated cultures, the growth rate of each strain was similar (Table 1), though cells lacking cbb3-1 (Cco1.1 and CcoΔ1) showed a slight, but reproducible, decrease in the number of cell divisions per hour. To analyse the respiratory activity of these strains, we examined NADH-dependent oxidase activity in wild-type and mutant cells. This assay, which measures the total respiratory activity from all the quinol and cytochrome oxidases, showed no significant difference between membranes from wild-type and mutant strains prepared from highly aerated cultures (Table 1). This suggested that the two presumed cbb3 oxidase(s) either constituted a minor part of the total activity in highly aerated cultures or that one or more of the other oxidases (Fig. 1A) compensated for the lack of the cbb3 enzymes.

Table 1. . Effects of cco mutations on growth and respiratory activity.
StrainRelevant genotypecell divisions h−1NADH oxidase activityTMPD oxidase activity
  1. The growth rates (cell divisions per hour) of wild-type P. aeruginosa and mutants lacking one (Cco1.1 and Cco2.2) or both cbb3 oxidases (CcoΔ1) were determined under highly aerated conditions. Oxidase activity (µmol of O2 min−1 mg protein−1).

PAKwild-type1.67 ± 0.0285.5 ± 11.11264 ± 171
Cco1.1Δ cbb3-11.50 ± 0.0493.9 ± 15.3 607 ± 57.8
Cco2.2Δ cbb3-21.59 ± 0.0877.6 ± 8.6 794 ± 101
CcoΔ1Δ cbb3-1, Δ cbb3-21.50 ± 0.0295.1 ± 2.1 308 ± 54.7

In contrast, when oxidase activity was measured using an electron donor specific for cytochrome c-dependent oxidases (TMPD/ascorbate), substantial differences between mutant and wild-type membranes were found. Compared to those from wild-type cells, membranes from cells lacking cbb3-2 (Cco2.2) had 37% less TMPD-dependent activity; those from cells lacking cbb3-1 (Cco1.1) lost 48% of their activity, and membranes from cells lacking both oxidases (CcoΔ1) lost 76% of their activity. As a control we used the BC1.1 strain, which lacks the cyt bc1 complex and cannot reduce the cytochrome that is used as an electron donor for all the cytochrome c-dependent oxidases. Cells lacking the cyt bc1 complex had a level of TMPD oxidase activity similar to CcoΔ1 (Table 1) and to a mutant lacking the cbb3 oxidases and the aa3-type oxidase (data not shown), demonstrating that nearly all of the cytochrome c-dependent electron flow is blocked by loss of the cyt bc1 complex or by inactivation of both cbb3 oxidases. This led to the conclusion that nearly all of the TMPD-dependent oxidase activity in aerobically grown cells was due to the combined activity of the cbb3-1 and cbb3-2 oxidases. The decrease in TMPD oxidase activity observed in cells lacking the cyt bc1 complex is in contrast to previous studies of a B. japonicum bc1 mutant strain (Preisig et al., 1996), and could reflect instability or reduced assembly of one or more P. aeruginosa oxidases in membranes purified from BC1.1 (especially as this strain lacked several c-type cytochromes; Fig. 2). Regardless, it appeared that cbb3-1 was responsible for a greater portion of the TMPD oxidase activity in highly aerated cells, consistent with its greater expression under these conditions.

cbb3-1 is required for RoxR-dependent signal transduction

In several other proteobacteria including R. sphaeroides, homologues of the cbb3 oxidase have been implicated in signal transduction pathways that regulate the expression of respiratory enzymes by the demonstration that cells that lack this respiratory enzyme show increased transcription of loci that are normally repressed under highly aerated conditions (O’Gara and Kaplan, 1997; Oh and Kaplan, 1999; 2000). To assess if the P. aeruginosa cbb3 oxidases also have a regulatory function, the effects of mutations in individual cco operons on expression of respiratory enzymes was monitored.

To accomplish this, we compared β-galactosidase activity from a cioAB:lacZ reporter gene in highly aerated cultures of wild-type and mutant strains. We chose to monitor expression of the cioAB operon, which encodes the cyanide-insensitive bd-type oxidase (CIO), because previous studies have shown that transcription of this promoter requires the RoxR protein, a homologue of R. sphaeroides PrrA (Comolli and Donohue, 2002). This analysis showed that transcription of cioAB was ∼10-fold higher in a strain lacking either cbb3-1 (Cco1.1), both cbb3 oxidases (CcoΔ1), or the cyt bc1 complex (BC1.1) compared to the wild-type (Fig. 4). β-Galactosidase activity from the cioAB:lacZ reporter gene in cells lacking only cbb3-2 (Cco2.2) was the roughly the same as in a wild-type strain, suggesting that cbb3-2 had little or no influence on cioAB expression and that the negative influence of the cbb3 oxidases on cioAB expression depended on the presence of a functional cbb3-1 oxidase. The apparent lack of a role for cbb3-2 on cioAB expression was surprising because of the high degree of amino acid similarity of its subunits to those of cbb3-1 (the CcoN, CcoO, CcoQ, and CcoP oxidase subunits share 86.3, 87.1, 96.7 and 72.1% amino acid identity).

Figure 4.

β-Galactosidase activity in highly aerated PAK, Cco1.1, Cco2.2, CcoΔ1, or BC1.1 cells containing the promoter for the cyanide-insensitive oxidase locus ( cioAB ) fused to lacZ. Strains either carried a single mutation in the wild-type background (black bars) or were double mutants that also had the gene for the response regulator RoxR inactivated (grey bars).

To independently assess the role of each cbb3 oxidase in cioAB expression, we tested the ability of a plasmid lacking a cco operon (pTR100), or one containing ccoNOQP-1 (pJC774) or ccoNOQP-2 (pJC775) to repress cioAB transcription in cells lacking both cco operons. Haeme-peroxidase staining showed that the ccoNOQP-1 (pJC774) or ccoNOQP-2 (pJC775) containing plasmids produced cbb3-1 or cbb3-2 oxidase subunits at levels equal to or higher than that found in wild-type cells grown under high aeration (Fig. 5A). In addition, expression of cbb3-1 (pJC774) decreased β-galactosidase levels from the cioAB:lacZ reporter to ∼40% of the level found in cells containing the control plasmid, indicating that this oxidase was sufficient to reduce cioAB expression. On the contrary, expression of ccoNOQP-2 (pJC775) did not have a significant effect on β-galactosidase levels from the cioAB:lacZ reporter (Fig. 5B); further indicating that cbb3-2 did not reduce expression from this promoter.

Figure 5.

Ability of plasmid-encoded cbb3 -1 or cbb3 -2 to repress of cioAB:lacZ activity. A plasmid containing no cco operon (pTR100), ccoNOQP -1 (pJC774), ccoNOQP -2 (pJC775), the ccoNOQP -2 coding region fused to the ccoNOQP -1 promoter (pJC789), or the ccoNOQP -1 coding region fused to the ccoNOQP -2 promoter (pJC790) was conjugated into wild type or CcoΔ1. Highly aerated cultures were analysed for their ability to produce the CcoP1 or CcoP2 by haeme peroxidase staining (A) and for β-galactosidase activity from a cioAB : lacZ reporter plasmid (B).

To test if the inability of the cbb3-2 to control cioAB expression was simply because of its low abundance in highly aerated cultures, we attempted to increase expression of cbb3-2 by placing the ccoNOQP-2 coding region under control of the ccoNOQP-1 promoter (pJC789). When this hybrid cco operon was placed in CcoΔ1, haeme peroxidase assays showed that CcoP2 and CcoO in cell extracts were present at higher levels suggesting that this strain contained more functional cbb3-2 oxidase (Fig. 5A and data not shown). However, the presence of this hybrid cco operon did not have a negative effect on cioAB transcription (Fig. 5B), which implied that the increased abundance of this enzyme was insufficient to reduce expression of the CioAB oxidase. As an additional control, we showed that cells containing a fusion of the ccoNOQP-2 promoter to ccoNOQP-1 (pJC790) had ∼30% lower β-galactosidase activity from a cioAB reporter. Presumably, this is because it produced less cbb3-1 oxidase than cells containing a plasmid with the normal cco-1 operon and even this amount of cbb3-1 was sufficient to reduce cioAB expression by a detectable amount. In summary, the results of these experiments demonstrated that even when cbb3-2 was expressed at levels approximating that of cbb3-1, this isozyme was unable to participate in the signal transduction pathway that affects cioAB transcription.

To examine if the cbb3-1-dependent reduction of cioAB expression required RoxR, we compared cioAB reporter activity in strains with or without this response regulator. Activity from the cioAB::lacZ fusion in cells lacking RoxR and either both cbb3 oxidases, the cbb3-1 enzyme, or the cyt bc1 complex was ∼80% less than in isogenic strains that contained a wild-type copy of roxR (Fig. 4). In contrast, activity from this reporter in cells lacking both RoxR and cbb3-2 was reduced only 25% relative to an isogenic strain containing a wild-type copy of roxR. This was comparable to the difference in cioAB transcription observed between wild-type cells and those containing a roxR mutation (Comolli and Donohue, 2002). Thus, inactivation of roxR was epistatic to the loss of the cbb3-1 in affecting cioAB expression, which is consistent with the RoxSR two-component signal pathway serving as a mediator of the cbb3-1-dependent signal.

Differential expression of the ccoNOQP operons in response to oxygen limitation

We were surprised to find a significant amount of cbb3-1 in extracts of highly aerated wild-type cultures because in numerous bacteria, cbb3 oxidase expression is maximal under oxygen limitation (Batut et al., 1989; Fisher, 1994; Mandon et al., 1994; Otten et al., 2001; Swem et al., 2001). When expression of the P. aeruginosa cbb3 oxidases was monitored in cells grown under semi-aerobic conditions, the abundance of CcoO and CcoP detected by haeme peroxidase staining, as well as that of other c-type cytochromes, was significantly greater than in cells grown at high aeration (Fig. 2B). In particular, increased abundance of CcoP2 suggested that more cbb3-2 was present in semi-aerobic cells compared to cells grown at high aeration. In contrast, the level of CcoP1 detectable by haeme peroxidase staining appeared comparable in extracts from each of these two cultures. Analysis of extracts from cells lacking either cbb3-1 (Cco1.1) or cbb3-2 (Cco2.2) supported this conclusion as Cco1.1 was missing detectable CcoP1 and showed little reduction in the abundance of CcoO whereas extracts from Cco2.2 lost all detectable CcoP2 and had a much less CcoO (Fig. 2B). This analysis indicated that the abundance of cbb3-1 and cbb3-2 subunits was affected differently by semiaerobic conditions; in particular it appeared that levels of cbb3-2 were increased whereas those of cbb3-1 did not change significantly.

To test this hypothesis and to assess how the ccoNOQP-1 and ccoNOQP-2 promoters responded to changes in oxygen tension, β-galactosidase activity from lacZ reporter plasmids containing no cco promoter (pLP170), the ccoNOQP-1 promoter (pJC753), or the ccoNOQP-2 promoter (pJC754) was measured in wild-type cells. There was ∼sevenfold higher activity of the ccoNOQP-1 promoter compared to that of ccoNOQP-2 in highly aerated cells while activity from the ccoNOQP-2 reporter was ∼twofold higher than ccoNOQP-1 in cells grown under semiaerobic conditions (Fig. 3). This change in the relative expression of the cco operons in semi-aerobic conditions was due to a 15-fold induction of ccoNOQP-2 without a significant change in ccoNOQP-1 transcription (Fig. 3).

Anr regulates ccoNOQP-2 expression

The P. aeruginosa global transcription factor Anr regulates the transcription of genes that encode enzymes required for energy conservation under anaerobic and low oxygen conditions, but it can also affect gene expression under highly aerobic conditions (Zimmermann et al., 1991; Winteler and Haas, 1996; Ray and Williams, 1997; Rompf et al., 1998). When we inspected the region surrounding the transcription initiation point for ccoNOQP-2 (determined to be 100 nucleotides upstream of the start site of the ccoN2 gene by primer extension analysis; J. C. Comolli and T. J. Donohue, unpubl.), a consensus binding site for Anr [TTGAT . . . ATCAA; (Winteler and Haas, 1996)] was found to be centred at position − 41.5. Because Anr has been reported to activate transcription from a similarly positioned site in other promoters (Zimmermann et al., 1991; Winteler and Haas, 1996), we analysed transcription of each ccoNOQP operon in a P. aeruginosa anr mutant.

Expression of ccoNOQP-2 under both growth conditions was reduced to background levels in cells lacking Anr (Fig. 3), indicating that both expression in highly aerated cultures and semi-aerobic induction of ccoNOQP-2 required this protein. This suggested that ccoNOQP-2 is regulated like the cbb3-encoding loci of many other bacteria in using an Fnr family member, in this case Anr, to control its induction under oxygen limitation (Batut et al., 1989; Preisig et al., 1993; Mandon et al., 1994; Schluter et al., 1995; 1997; Toledo-Cuevas et al., 1998; Van Spanning et al., 1997; Lopez et al., 2001; Otten et al., 2001). On the contrary, the expression of ccoNOQP-1 in the anr mutant was unchanged relative to wild-type cells (Fig. 3). Thus, the mechanisms influencing the ccoNOQP-1 promoter appear to be distinct given that changes in oxygen tension and the presence of Anr have no effect on its expression.

Discussion

Unlike mitochondria, bacteria typically have multiple respiratory oxidases with preferred electron donors and cytochrome contents that are thought to contribute to energy conservation in a given environment (Poole and Cook, 2000; Richardson, 2000). The genome sequence of the γ-proteobacterium P. aeruginosa predicted that two of the five oxidases in this opportunistic pathogen were members of the cbb3 family of haeme-copper oxidases. Our results demonstrate that both of these predicted cco operons are expressed at detectable levels in wild-type cells. Our data also suggest that the physiological function(s) of a bacterial oxidase can be determined by factors other than subunit structure or cytochrome content.

Relative roles of individual cbb3 isozymes in P. aeruginosa

Though the predicted subunits of the P. aeruginosa cbb3-1 and cbb3-2 oxidases share > 72% amino acid identity, it is possible to propose specific roles for each isozyme. In highly aerated cells, cbb3-1 subunits were more prevalent than those of cbb3-2, expression from the ccoNOQP-1 promoter was ∼sevenfold greater than expression from the ccoNOQP-2 promoter, and inactivation of ccoNOQP-1 reduced aerobic TMPD-dependent cytochrome oxidase activity to a greater extent than did inactivation of ccoNOQP-2. Given these results, we conclude that cbb3-1 plays a significant physiologic role at high oxygen tensions, suggesting that its role is distinct from most of the cbb3-type oxidases that are typically important under oxygen-limiting conditions. This suggests that cbb3-1 has an affinity for O2 that is lower than that of cbb3 oxidases from other bacteria, which would require its purification from the other P. aeruginosa oxidases to ascertain.

In contrast, the P. aeruginosa cbb3-2 oxidase appears more similar to canonical cbb3-type enzymes because, in highly aerated cultures, it was of relatively low abundance, its inactivation had a lesser impact on TMPD-dependent oxidase activity, and transcription of its promoter was reduced. However, the abundance of cbb3-2 and expression of the ccoNOQP-2 promoter increased when P. aeruginosa was grown under semi-aerobic conditions. This increased expression was dependent upon Anr, homologues of which have been shown to control expression of a number of cbb3 loci in other bacteria (Batut et al., 1989; Preisig et al., 1993; Mandon et al., 1994; Schluter et al., 1995; 1997; Van Spanning et al., 1997; Toledo-Cuevas et al., 1998; Lopez et al., 2001; Otten et al., 2001). These findings lead to the conclusion that cbb3-2 plays a significant role under oxygen limitation. Consistent with this hypothesis, strains lacking cbb3-2 had a growth defect at low oxygen tension (≤2%) whereas a strain lacking cbb3-1 grew at the same rate as wild-type cells (data not shown).

How might these cbb3 isozymes have different functions?

Despite the high degree of amino acid similarity between the cbb3-1 and cbb3-2 subunits only cbb3-1 repressed expression of the cyanide insensitive oxidase (CIO). Because cbb3-2 was ineffective in repressing cioAB when cells contained levels of CcoP2 that were comparable to or exceeded those of CcoP1 in a wild-type strain, it is likely that an inherent feature of one of the cbb3-1 oxidase subunits enables it to communicate with the RoxR-dependent signalling pathway. There are limited differences in the CcoN, CcoO and CcoQ subunits of cbb3-1 and cbb3-2 scattered throughout the proteins, but localized regions of high divergence in the CcoP subunits could define domains that are required for productive interaction with the sensor kinase RoxS, or another signalling partner. From a bioenergetic standpoint, differences in the affinities of the two cbb3 isozymes for oxygen, their electron donors or in their ability to pump protons, might explain the physiological functions suggested by our results.

Signal transduction by cbb3-1

The influence of P. aeruginosa cbb3-1 on cioAB expression recapitulates the signal transduction pathway present R. sphaeroides, where the cbb3 oxidase negatively influences the aerobic expression of loci involved in anaerobic metabolism (O’Gara and Kaplan, 1997; O’Gara et al., 1998; Oh and Kaplan, 1999; 2000; 2002). The cbb3-1/RoxSR signalling pathway affects loci encoding energy-generating complexes other than the CIO (data not shown), so this system could act as a homeostatic regulator that helps fine tune the respiratory system of P. aeruginosa. The stimulus sensed by cbb3-1 is unknown, but conditions that change activity of this oxidase would ostensibly affect the signalling pathway. This could be oxygen, but other compounds that alter electron flow, such as cyanide, also appear to be effective (Comolli and Donohue, 2002; Cooper et al., 2003). In fact, cyanide is produced by P. aeruginosa in sufficient quantities to inhibit the function of its haeme-copper oxidases (Pudek and Bragg, 1974; Castric, 1975; Solomonson, 1981; Blumer and Haas, 2000). Because the cyanide-insensitive oxidase (CIO) does not have a haeme-copper catalytic centre and can still function in the presence of cyanide, cyanide-dependent induction of the CIO (Comolli and Donohue, 2002) via cbb3-1/RoxSR would enable P. aeruginosa to respire in situations where the compound accumulates.

Do other proteobacteria contain multiple cbb3 oxidases?

Tandem ccoNOQP operons are predicted to exist in the genome sequences of the other free-living Pseudomonads P. fluorescens Pf-5 and P. putida, but the plant-pathogenic P. syringe contains only a single operon (http://www.tigr.org). It is possible that the presence of two cbb3 oxidases may provide the respiratory flexibility that certain free-living Pseudomonads need to survive in the diverse environments. Two fixNOQP loci have also been seen in several rhizobial species, though in these cases the loci were split on the chromosome and a symbiosis plasmid, and the functions of the oxidases appear redundant (Patschkowski et al., 1996; Schluter et al., 1997).

In summary, these experiments provide evidence for the roles of the two P. aeruginosa cbb3-1 and cbb3-2 oxidases. Both cbb3 oxidases are predicted to make some contribution to energy conservation at high aeration since inactivation of both cbb3-1 and cbb3-2 caused a greater reduction in the TMPD-dependent oxidase activity than did either inactivation of either enzyme alone. Combined, the other three predicted P. aeruginosa oxidases also appear to effectively conserve energy as the NADH-dependent oxidase activity and the growth rate was similar in cells lacking both cbb3 oxidases and in wild-type cells. This is consistent with inactivation of the cbb3 oxidases being compensated by the increased activity of other P. aeruginosa oxidases. Because the cbb3-1/RoxSR system is known to stimulate expression of at least one alternative oxidase, CioAB, it could provide one way in which P. aeruginosa compensates for respiratory limitations and continues to effectively generate energy in a number of different environments.

Experimental procedures

Strains and growth conditions

Escherichia coli strains were grown in Luria–Bertani broth (LB) containing antibiotics when appropriate. Pseudomonas aeruginosa strain PAK was grown in LB or on Vogel-Bonner agar ( Vogel and Bonner, 1956 ). Highly aerated culture conditions were maintained by diluting P. aeruginosa overnight cultures into a small volume (8–10 ml) of growth medium in a 125 ml flask shaken at 260 r.p.m. at 37°C. Cells were harvested or assayed in logarithmic growth phase at an OD 600 of 0.4–0.5. For semi-aerobic conditions, P. aeruginosa was diluted into 60 ml medium in a 125 ml flask that was shaken at 90 r.p.m. at 37°C. The rate of cell division for individual strains was calculated as described previously ( Comolli and Donohue, 2002 ).

Gene inactivation

Mutant strains were generated using homologous recombination to inactivate the appropriate chromosomal locus. Regions of interest were amplified from PAK genomic DNA and ligated into a cloning vector. A section of the coding sequence was then deleted (and replaced by an antibiotic resistance cassette in some cases) and this inactivated operon was then cloned into the suicide plasmid pNHG1 (Jeffke et al., 1999). To inactivate ccoNOQP-1, a 1554 bp BstBI to EcoRI fragment was replaced by a ΩGmr cassette to create strain Cco1.1 (Fig. 1B). A mutant strain with a ccoNOQP-2 lesion, Cco2.2, was generated by substitution of a 2210 bp AccI to BglII fragment with a ΩGmr cassette. Inactivation of both ccoNOQP operons to produce CcoΔ1 was accomplished by a 5300 bp XhoI to EcoRI deletion. A strain lacking the bc1 complex (BC1.1) was constructed by replacing a 1081 bp BspEI to KpnI region of the locus including PA4429 and PA4430 with a ΩGmr cassette. A 335 bp PvuII fragment of anr was deleted to create strain Anr1, which was unable to use nitrate in the absence of oxygen as expected (Sawers, 1991).

Suicide plasmids were transferred to strain PAK from E. coli S17-1 and exconjugates were selected on VBM with or without gentamicin plus 5% sucrose. Genetic alteration in the mutant strains was confirmed by amplification of the appropriate region and diagnostic restriction enzyme digestion. To create strains lacking both RoxR and a particular respiratory enzyme, a suicide plasmid containing the roxR gene with 533 bp replaced by a ΩTcr cassette (Fellay et al., 1987) was transferred to a strain containing the appropriate oxidase mutation.

Detection of c-type cytochromes

Mid-log phase cells were harvested, washed with 50 mM KPO4 pH 7.1, then suspended in 0.4 ml 50 mM KPO4 pH 7.1, 5 mM MgCl2, 100 µg ml−1 DNAse and 0.2 mg ml−1 lysozyme. After 15 min on ice, cells were lysed by sonication, the extract was diluted into SDS–PAGE sample buffer, and for each sample 40.5 µg of protein was loaded onto a 4–20% gradient gel. To detect c-type cytochromes, gels were stained with 3,3′,5,5′-tetramethylbenzidine and hydrogen peroxide as described (Thomas et al., 1976).

Construction of reporter plasmids and β-galactosidase assays

Potential promoter regions of ccoNOQP-1 or ccoNOQP-2 were amplified from genomic DNA, digested with XbaI/BamHI or StuI/BamHI, respectively, and cloned into corresponding sites in the lacZ reporter plasmid pLP170 (Preston et al., 1997). The ccoN-1:lacZ reporter plasmid (pJC753) included 552 bp upstream of the transcriptional start site (45 bp from the ccoN-1 start codon; J. C. Comolli and T. J. Donohue, unpubl.), whereas the ccoN-2:lacZ reporter plasmid (pJC754) included 617 bp upstream of the transcriptional start site (100 bp from the ccoN-2 start codon; J. C. Comolli and T. J. Donohue, unpubl.).

The cioAB:lacZ reporter plasmid (pJC701) (Comolli and Donohue, 2002) was used in strains lacking a single oxidase. For use in the complementation analysis, a cioAB:lacZ reporter was created by cloning the promoter fragment from pJC701 into pHRP309 (Parales and Harwood, 1993) to create pJC767. Plasmids were introduced into the appropriate P. aeruginosa strain by electroporation or conjugation and activity assays were performed as described (Comolli and Donohue, 2002). Assays were performed in triplicate and the error bars represent standard deviation from the mean. Although the reporters contain identical promoters, the background activity produced from pJC767 was much higher than from pJC701 so the activity from the promoterless pHRP309 was subtracted from that obtained with pJC767.

Oxidase activity

The rate of oxygen consumption in membranes from cells grown under aerobic conditions was determined as described previously (Comolli and Donohue, 2002). NADH-dependent oxidase activity was assayed using 0.5 mM NADH as a substrate; TMPD-dependent oxidase activity with 0.2 mM TMPD and 10 mM ascorbate. Values presented are the average of three independent measurements and the error is the standard deviation.

Complementation of oxidase mutants

Plasmids for complementation of cbb3 oxidase mutants were constructed in pTR100 (Weinstein et al., 1992). A 4.1 kb fragment containing the ccoNOQP-1 promoter and coding region or a 3.9 kb fragment containing the ccoNOQP-2 promoter and coding region were cloned with XbaI and SacI into pTR100 to create pJC774 and pJC775. Plasmids containing each ccoNOQP coding region fused to the promoter of the other operon (pJC789 and pJC790) were generated using complementary SapI sites created in a conserved 12 base pair region (− 4 to + 5 base pairs) around the start codons of ccoN1 and ccoN2. Each promoter region was amplified with primers containing an XbaI or SapI site whereas coding regions were amplified using primers containing a SapI or SacI site. Amplification products were cloned into pTR100 digested with XbaI and SacI and these plasmids were transferred to the appropriate P. aeruginosa strain using E. coli S17-1.

Acknowledgements

This work was supported by NIH grant GM37509 to TJD. JCC was supported by NRSA fellowship F32 G20344 from NIGMS. We would like to thank P. Agron, C. Harwood, L. Passador, J.-I. Oh and H. Schweizer for contributing plasmids that were used in this study. The Pseudomonas Genome Project (http://www.pseudomonas.com) and The Institute for Genomic Research (http://www.tigr.org) are also greatly appreciated for their Pseudomonas genomic information.

Ancillary