Novel Targets of the CbrAB/Crc Carbon Catabolite Control System Revealed by Transcript Abundance in Pseudomonas aeruginosa

The opportunistic human pathogen Pseudomonas aeruginosa is able to utilize a wide range of carbon and nitrogen compounds, allowing it to grow in vastly different environments. The uptake and catabolism of growth substrates are organized hierarchically by a mechanism termed catabolite repression control (Crc) whereby the Crc protein establishes translational repression of target mRNAs at CA (catabolite activity) motifs present in target mRNAs near ribosome binding sites. Poor carbon sources lead to activation of the CbrAB two-component system, which induces transcription of the small RNA (sRNA) CrcZ. This sRNA relieves Crc-mediated repression of target mRNAs. In this study, we have identified novel targets of the CbrAB/Crc system in P. aeruginosa using transcriptome analysis in combination with a search for CA motifs. We characterized four target genes involved in the uptake and utilization of less preferred carbon sources: estA (secreted esterase), acsA (acetyl-CoA synthetase), bkdR (regulator of branched-chain amino acid catabolism) and aroP2 (aromatic amino acid uptake protein). Evidence for regulation by CbrAB, CrcZ and Crc was obtained in vivo using appropriate reporter fusions, in which mutation of the CA motif resulted in loss of catabolite repression. CbrB and CrcZ were important for growth of P. aeruginosa in cystic fibrosis (CF) sputum medium, suggesting that the CbrAB/Crc system may act as an important regulator during chronic infection of the CF lung.


Introduction
Pseudomonas aeruginosa is a metabolically versatile organism, which is able to grow in a variety of different ecological niches. Its natural habitats can be nutrient-poor (e.g., aqueous environments) or nutrient-rich (e.g., the rhizosphere or infected animal tissues). The CbrAB/Crc carbon catabolite control system helps to ensure a metabolic optimization process by allowing preferential utilization of good energy sources and establishing a healthy carbon (C) to nitrogen (N) balance [1]. The CbrAB/Crc system involves a regulatory protein (Crc), which is required to establish translational repression of target mRNAs at CA (catabolite activity) motifs, usually located in the vicinity of ribosome binding sites [1][2][3][4]. The core sequence of the CA motif consists of eight nucleotides, AAnAAnAA, where n stands for any nucleotide. However, inspection of RNAs that are under Crc control suggests that n = C or U may be preferred over n = A or G in CA motifs [1,2]. A small RNA (sRNA) termed CrcZ, which possesses several conserved CA motifs and appears to bind to Crc, antagonizes translational repression mediated by the Crc protein [1]. CrcZ levels are low in the presence of a good carbon source (e.g., succinate), elevated in the presence of an intermediate carbon source (e.g., glucose) and high when a less preferred carbon compound is used as the sole energy source (e.g., mannitol). This regulation is established by the activity of the two-component system CbrAB consisting of the membrane-bound sensor CbrA and the response regulator CbrB, which are required for the expression of crcZ, together with the alternative sigma factor RpoN and IHF protein [1,5]. CbrB binds to an upstream activating sequence (UAS) in the crcZ promoter region [5]. Mutants affected in cbrB or crcZ have similar, but not identical phenotypes, indicating that CrcZ mediates most, but not all, output of the CbrAB system [1,5]. The CbrAB/Crc system is involved in the regulation of swarming, biofilm formation, cytotoxicity and antibiotic resistance in P. aeruginosa [6,7]. However, in these interactions the regulatory mechanisms have not been investigated in molecular detail.
While the effects of the CbrAB/Crc system occur essentially at a post-transcriptional level, we asked whether transcript abundance could be used to reveal targets of this system. Previous experience with the mechanistically similar GacSA/ RsmA pathway in P. aeruginosa [8][9][10] and P. fluorescens Pf-5 [11] suggests that during translational repression transcript abundance tends to be low, whereas transcript levels may be elevated during derepression, presumably reflecting effects of translational control on mRNA stability. Based on these observations, we decided to compare the transcriptome of the P. aeruginosa wild type PAO1 with that of its crc, crcZ and cbrB mutants. To find new targets of the CbrAB/Crc system, we focussed on several mRNAs having appropriately located CA motifs. The targets chosen are all involved in the utilization of relatively poor C sources. Furthermore, we found that growth of cbrB and crcZ mutants was severely impaired in an artificial cystic fibrosis (CF) sputum medium, an effect that is most likely due to repression of amino acid uptake and catabolism by the Crc protein. Thus, the CbrAB/Crc system may be an important regulatory system enabling P. aeruginosa to thrive in the CF lung.

Results
Transcriptome analysis of genes regulated by Crc -To assess the impact of the CbrAB/Crc pathway on global gene expression of P. aeruginosa, we performed a transcriptome analysis of P. aeruginosa PAO1 wild type and its crc, crcZ and cbrB deletion mutants using commercially available genome-wide DNA microarrays (Affymetrix). First, we grew cells to late exponential phase in Luria Broth (LB) as a nutrient rich medium, which mainly contains amino acids, peptides and small amounts of sugars and nucleotides [12] and which creates conditions of moderate carbon catabolite repression [3]. Second, we used a defined basal salts medium (BSM) amended with 40 mM succinate, which establishes strong catabolite repression in strain PAO1 [1]. Under both conditions, Crc mediates catabolite repression of amiE, which encodes shortchain aliphatic amidase and has previously served as a reporter gene ( [1]; Figure S1). We expected that genes whose expression is translationally repressed by the CbrAB/Crc system might show elevated transcript levels in a crc mutant, whereas down-regulated transcript levels might be seen in both cbrB and crcZ mutants, due arrest of translation.
In LB-grown cells, we found 57 transcripts to be differentially ($2.0-fold) expressed in the crc mutant, by comparison with the wild type PAO1; 19 transcripts were up-regulated and 38 were down-regulated at least two-fold (Table S1). After growth in BSM containing succinate, 95 transcripts showed altered expression; 53 transcripts were up-regulated and 42 were down-regulated (Table  S1). The overlap of genes that were differentially affected under both conditions was small ( Figure 1A). The probable reason is that transcriptional expression of catabolic genes varies greatly between the two media, making it difficult to detect the same effects of Crc on transcript abundance. Nevertheless, the combined data from both media enabled us to see a considerable number of transcripts influenced by Crc. As a control, we checked amiE transcript levels. They were elevated in the Dcrc mutant cultivated in LB (Table 1  and Table S1), which is consistent with increased amiE translation ( Figure S1) and, presumably, with concomitantly enhanced amiE mRNA stability in the absence of Crc-mediated catabolite repression. By contrast, the Dcrc mutation had no effect on amiE transcript abundance in cells grown in BSM, which lacks inducing amides and hence does not permit transcriptional expression of the amiE operon [1,13]. These results suggest that transcriptome analysis may be used to identify targets of the CbrAB/Crc system provided that sufficient transcriptional expression of the target genes takes place.
Transcriptome analysis may reveal both and indirect effects of the CbrAB/Crc system. Direct effects are assumed to involve at least one CA motif (AAnAAnAA) in the region of 250 to +50 nucleotides relative to the translational start site in target mRNAs. We found 29 Crc-regulated transcripts having an appropriately located putative CA motif (Table S1). With the exception of four transcripts (three of them in LB and one in BSM), all mRNAs harbouring a CA motif were up-regulated in the crc mutant.
Transcriptome analysis of crcZ and cbrB mutants -As mentioned above, the Crc protein and the CrcZ sRNA, which requires CbrB to be transcribed, have opposite regulatory effects; thus one might expect that transcripts that are up-regulated in a crc mutant would be down-regulated in crcZ and cbrB mutants and vice versa. To investigate this assumption, we performed transcriptome analyses with a crcZ and a cbrB mutant under the same conditions as above. In LB-grown cells, 145 transcripts were differentially expressed by a factor of $2.0 in PAO1DcrcZ, whereas 166 transcripts were differentially expressed in PAO1DcbrB, by comparison with the wild type PAO1 (Table S2). In BSM-grown cells, 157 transcripts showed differential regulation in the crcZ mutant and 60 transcripts in the cbrB mutant (Table S2). Again, the number of genes showing a similar expression pattern in both media was low ( Figure 1B and 1C). However, the crcZ and cbrB mutations had largely parallel effects; both mutations affected 101 out of 210 transcripts and 59 out of 158 transcripts in LB ( Figure 1D) and in BSM ( Figure 1E), respectively. Among the transcripts that were affected by both mutations, a majority was down-regulated (Table  S2). Some transcripts seemed to be only regulated by CbrB and not by CrcZ. However, preliminary attempts to identify CbrB binding sites upstream of the corresponding ORFs have not been successful. Apart from the crcZ promoter, known direct CbrB targets include the lipA and hutU promoters [5,14]. However, these genes did not appear in our transcriptome analysis, presumably because of lack of transcriptional inducers in the growth media. Some transcripts whose abundance appeared to be only regulated by CrcZ, but not by CbrB, were also seen, but were considered as questionable since crcZ expression strictly depends on CbrB [1]. Taken together, our transcriptome analysis is consistent with the current model stating that the sRNA CrcZ mediates most effects of the CbrAB two-component system.
Somewhat unexpectedly, the overlap between transcripts whose abundance was affected by Crc as well as by CrcZ and/or CbrB in an opposite manner was low in BSM-grown cells (17 out of 47 genes; Figures 1D and 1E, Table 1) and absent in LB-grown cells. Among the genes that did show the anticipated transcript pattern were estA (encoding a secreted esterase), the dctPMQ operon (required for C 4 -dicarboxylate transport at low concentrations), braC and the braDEFG operon (which are important for transport of branched chain amino acids) ( Table 1). Moreover, estA, dctP and braC possess CA motifs indicating that they may be targets of the CbrAB/Crc system.
Transcripts that were strongly up-regulated in LB in the absence of Crc, but not down-regulated in the absence of CrcZ or CbrB, included opdH (encoding a cis-aconitate porin) and aroP2 (encoding an aromatic amino acid transporter) ( Table 1). For opdH and aroP2 candidate CA motifs were found. Transcripts that were strongly up-regulated in BSM in the absence of Crc included acsA (encoding acetyl-CoA synthetase required for acetate catabolism), opdH, braC, braDEFG, bkdA1 (first gene of an operon involved in branched-chain amino acid assimilation), acoR and acoAB (for acetoin catabolism). CA motifs were found in all of these transcripts except for braDEFG and bkdA1 (Table 1), which provides evidence that these genes may be targets of the CbrAB/Crc system. Some striking effects of the CbrAB/Crc system on the transcriptome were not associated with CA motifs. For example, the pqsABCDE and phnAB operons were strongly down-regulated in the crcZ and cbrB mutants, especially in minimal medium with succinate (Table 1). Both operons are required for the biosynthesis of the Pseudomonas quinolone signal (PQS) and 2-heptyl-4hydroxyquinoline, which act as quorum sensing signal molecules together with the transcriptional regulator PqsR [15]. Transcript abundance of pqsR was also affected, but to a lesser extent ( Table 1). As none of these genes showed elevated transcript levels in the crc mutant, their regulation by the CbrAB/Crc system is probably indirect. Moreover, several genes and operons involved in the anaerobic denitrification pathway were affected by the CbrAB/Crc system at the transcript level (Tables S1 and S2), although the cells had been grown aerobically. However, except for narG, which has a poorly conserved CA motif ( Table 1), none of the corresponding transcripts showed a CA motif, suggesting indirect regulation by CbrAB/Crc. Finally, in cells cultivated in LB, pqqA (encoding the pyrroloquinoline quinone [PQQ] biosynthesis protein A) was one of the most strongly up-regulated transcripts in the crcZ and cbrB mutants (Table 1). P. aeruginosa when growing on ethanol uses a PQQ-dependent ethanol oxidation system (encoded by the exaA and exaBC genes), which converts ethanol to acetate. Acetyl-CoA synthetase (encoded by the acsA gene) is also essential for ethanol utilization [16]. The transcripts of the pqqABCDE operon and of the acsA and exaC genes were all differentially expressed in LB when either crcZ or cbrB was mutated (Table 1). However, as mentioned above, only the acsA transcript was also affected by a crc mutation. Thus, regulation of PQQ biosynthesis by the CbrAB/Crc system appears to be mostly indirect. The agmR gene encoding a positive regulator of PQQ biosynthesis [17] is a good candidate as it exhibits a CA motif (AUAACGACAAUAA).
Validation of candidate targets -In the remainder of this study, we will focus on transcripts (estA, acsA, and aroP2) that have increased abundance in a crc mutant and all contain a putative CA motif (i.e. AAnAAnAA). In addition, we investigated the role of the CbrAB/ Crc system on bkdR and bkdA1, as the bkdA1 transcript was highly upregulated in the crc mutant and the cognate regulatory gene bkdR contains a putative CA-motif in the vicinity of its ribosome binding site. All these genes are involved in the utilization of less preferred carbon sources. Our initial concern was to validate the microarray results by an independent method. Data obtained by RT-qPCR (Table 2) were in full agreement with the microarray data (Table 1). In particular, the RT-qPCR measurements obtained in LB-grown cells proved to be more sensitive and hence possibly more accurate than the microarray data.
Regulation of uptake and degradation of poor carbon sources -Previously, only amiE and phzM mRNAs have been characterized as targets of the CbrAB/Crc system in P. aeruginosa [1,18]. In this section, we will provide evidence that estA and acsA are also targets. EstA esterase is an autotransporter that possesses lipolytic enzyme activity [19] and contributes to the hydrolysis of glycerol esters with short-chain fatty acids, which can be further used as carbon sources by P. aeruginosa [20]. EstA is located in the outer membrane [21]. In our RT-qPCR analysis, estA mRNA appeared to be downregulated in PAO1DcrcZ and PAO1DcbrB and up-regulated in PAO1Dcrc grown under both growth conditions (Table 2). A translational estA'-'lacZ fusion was assayed for b-galactosidase activity in strain PAO1 and in the cbrB, crcZ and crc mutants, after growth in LB ( Figure 2A) and BSM supplemented with 40 mM  succinate as the sole carbon source ( Figure 2B). Under both conditions, estA'-'lacZ expression was repressed in PAO1DcrcZ and PAO1DcbrB and derepressed in PAO1Dcrc. However, the estA'-'lacZ expression profiles showed quantitative differences in LB (Figure 2A) vs. BSM-succinate ( Figure 2B), in agreement with the transcript analysis (Tables 1 and 2). To confirm that estA is a target of the CbrAB/Crc system we mutated the CA motif, which is located 17 nucleotides upstream of the estA start codon (Table 1).
In the absence of the wild-type CA motif, the CbrAB/Crc system lost its ability to regulate estA completely ( Figure 2C). P. aeruginosa can use acetate as a sole carbon and energy source, by converting acetate to acetyl-CoA, which is further metabolized via the tricarboxylic acid and glyoxylic acid cycles to generate energy and cell components. The initial reaction is catalyzed by acetyl-CoA synthetase, the product of the acsA gene. Acetyl-CoA synthetase activity is induced during growth on acetate or on ethanol [16,22]. In our transcript analysis, acsA was differentially expressed in the wild type PAO1 and the crc, crcZ and cbrB mutants (Tables 1 and 2). The unexpected and unexplained finding here is that the cbrB and crcZ mutants on the one hand and the crc mutant on the other did not show opposite effects on ascA expression (Table 1). To demonstrate that the CbrAB/Crc system regulates acsA, we measured the b-galactosidase activity of a translational acsA'-'lacZ fusion in these strains. As strain PAO1 lacking crcZ or cbrB showed a severe growth defect on acetate as the sole carbon source ( Figure 3A), we grew the cells in BSM supplemented with 40 mM acetate and 5 mM succinate. As expected, ascA'-'lacZ expression was completely repressed in PAO1DcrcZ and PAO1DcbrB, and increased in PAO1Dcrc, by comparison with PAO1 ( Figure 3B). The acsA mRNA harbours an extended CA motif (AACAAAAACAA) located 25 nucleotides upstream of the acsA start codon. Mutation of the CA motif in the acsA leader of an      acsA'-'lacZ construct resulted in loss of catabolite repression mediated by Crc ( Figure 3C). Targets of the CbrAB/Crc system in the uptake and metabolism of amino acids -The CbrAB two-component system is important for maintaining the C to N balance during amino acid utilization by P. aeruginosa [23]. In a seminal study, Crc was found to be a posttranscriptional regulator of the bkd operon, which is involved in the utilization of branched-chain amino acids in P. putida and P. aeruginosa [24]. Furthermore, a recent proteomic analysis [7] revealed that Crc is involved in the regulation of several transporters required for the uptake of amino acids in P. aeruginosa. Our transcriptome analysis revealed a number of CbrAB/Crcregulated genes and operons (e.g., bkdA1A2B-lpdV, braBCDEDG, aroP2, aruBEFG, aotP, arcBC) that are involved in the utilization and transport of amino acids. In agreement with the original observations of Hester et al. [24], we found that the CbrAB/Crc system is involved in the regulation of branched-chain amino acid utilization in P. aeruginosa. In BSM supplemented with valine, isoleucine and leucine as the sole carbon sources, strains PAO1DcrcZ and PAO1DcbrB were not able to grow, whereas strains PAO1 and PAO1Dcrc did grow ( Figure 4A). Branchedchain amino acid assimilation requires an initial deamination step performed by the branched-chain amino acid transaminase (encoded by ilvE), followed by decarboxylation by the branchedchain keto acid dehydrogenase (encoded by the bkdA1A2B-lpdV operon). The bkd operon is induced by branched-chain amino acids as well as branched-chain keto acids [25]. In P. putida, all five genes are repressed by Crc [26], whereas in P. aeruginosa only the bkd operon, but not the ilvE gene, seems to be under catabolite repression control by Crc (Tables 1 and 2). To verify the results from our transcriptome data, we analyzed the expression of a translational bkdA1'-'lacZ fusion in LB supplemented with 23 mM leucine, 25.6 mM valine and 7.6 mM isoleucine. As shown in Figure 4B, bkdA1'-'lacZ expression was slightly increased in PAO1Dcrc by comparison with that in PAO1, and strongly repressed in PAO1DcrcZ and PAO1DcbrB. In P. putida, a CA motif (AAAAACAA) is located 24 nucleotides upstream of the bkdA1 start codon. By contrast, in P. aeruginosa a putative CA motif (AAAAACAAAA) occurs 187 nucleotides upstream of the bkdA1 start codon and it is questionable whether this sequence would affect translation initiation of bkdA1. A more likely scenario pictures the bkdR gene, which lies upstream of the bkd operon, as a target of the CbrAB/Crc system. In P. putida, the transcriptional regulator BkdR positively controls expression of the bkd operon [27]. The BkdR protein content was elevated in a crc mutant while the bkdR mRNA levels were unchanged, whereas bkdA1 expression was affected at both the mRNA and the protein level [28].
In P. aeruginosa, two putative CA motifs can be found in the untranslated leader region of bkdR (AACUACAAGAA at 236 to 226 and AACAAGAGAAACAA at 220 to 27). To see whether the CbrAB/Crc system regulates the expression of bkdR, we performed b-galactosidase assays of a translational bkdR'-'lacZ fusion in LB supplemented with 23 mM leucine, 25.6 mM valine and 7.6 mM isoleucine. The activity of this fusion was 1.5-to 2.0fold increased in PAO1Dcrc compared to PAO1 and was practically undetectable in PAO1DcrcZ and PAO1DcbrB ( Figure 4C), confirming that bkdR is a target of the CbrAB/Crc system in P. aeruginosa.
Another good candidate for interaction with the CbrAB/Crc system was the aroP2 gene, which encodes a transporter for aromatic amino acids. A growth experiment was performed with PAO1 and its crc, crcZ, and cbrB deletion mutants in BSM supplemented with 4 mM tyrosine as the sole carbon source. The crcZ and cbrB mutants did not grow, whereas PAO1 and PAO1Dcrc reached an OD 600 of 1 after 9 h; PAO1Dcrc even grew slightly faster than the wild type strain ( Figure 5A). The expression of a translational aroP2'-'lacZ fusion was very low in LB and in BSM amended with tyrosine (data not shown), as would be expected for a gene specifying an inner membrane protein. It has been shown previously that aroP2 expression is up-regulated in a lyophilized CF sputum medium, which is rich in amino acids [29]. Therefore, we used a synthetic CF sputum medium (SCFM; [30]) for measuring the b-galactosidase activities conferred by the aroP2'-'lacZ fusion. Although expression was still very low under these conditions, a 2to 4-fold higher expression of the aroP2'-'lacZ fusion was measured in PAO1Dcrc, compared with PAO1; in the crcZ and cbrB mutants aroP2 was fully repressed ( Figure 5B). The deletion of the CA motif in the fusion construct resulted in loss of regulation by Crc ( Figure 5C) and confirmed that aroP2 is a target of the CbrAB/Crc system.
Consequences of catabolite repression for environmental adaptation of P. aeruginosa -As we have shown that the CbrAB/Crc system strongly affects the utilization and transport of various less preferred carbon and nitrogen sources, the question arises as to what benefits P. aeruginosa may obtain from this regulation in natural environments, e.g. in the CF lung where amino acids are important C sources [29,30]. To investigate the impact of catabolite repression under such conditions, we cultivated PAO1, PAO1Dcrc, PAO1DcrcZ and PAO1DcbrB aerobically in the artificial sputum medium (SCFM). The crcZ and cbrB deletion mutants grew more slowly than the PAO1 wild type, whereas the crc deletion strain grew similarly to PAO1 ( Figure 6). In this sputum medium, tyrosine and phenylalanine are important C sources and degradation of both amino acids is induced in P. aeruginosa [31]. Moreover, leucine (1.6 mM), isoleucine (1.1 mM) and valine (1.1 mM) are present at high concentrations in this medium [30]. As we have shown above, tyrosine uptake ( Figure 4) and utilization of branched-chain amino acids ( Figure 5) are tightly controlled by the CbrAB/Crc system. The growth patterns seen in SCFM ( Figure 6) are fully consistent with this regulation. As during chronic infection P. aeruginosa forms biofilms [32], we also measured biofilm formation in PAO1, PAO1Dcrc, PAO1DcrcZ and PAO1DcbrB in SCFM using a static biofilm assay. The crcZ and cbrB mutants were impaired in biofilm formation whereas the biofilm formed by PAO1Dcrc was increased ,5-fold, by compar- ison with the wild type ( Figure S2). In BSM-succinate medium we observed the same pattern as that already seen in SCFM ( Figure  S2). However, when we used BSM medium supplemented with an intermediate (glucose) or non-preferred carbon source (mannitol), the repressing effect of Crc on biofilm formation was lost ( Figure  S2). In conclusion, global regulation by the CbrAB/Crc system may give the wild type P. aeruginosa a selective advantage over crcZ or cbrB mutants in the CF lung and may help the bacterium to persist in CF patients.

Discussion
In our present transcriptome analysis, the expression of $380 genes was affected by mutations in the CbrAB/Crc system in P. aeruginosa when we take into account data obtained in LB and in the defined medium BSM-succinate. Thus, in spite of a major output at the level of translation, the CbrAB/Crc signal transduction pathway clearly has a global impact on transcript abundance. In agreement with earlier phenotypic data [1], the present transcriptomic data indicate that most CbrB activity is linked to CrcZ activity ( Figure 1) in that transcriptomic responses were similar, but not identical in cbrB and crcZ mutants (Tables S1 and S2). It is therefore possible that CbrB could act as a transcriptional regulator of genes other than the previously recognized crcZ, lipA and hutU genes [5]. However, such novel CbrB target genes remain to be identified. Whether CrcZ regulates mRNA expression independently of Crc is another question that cannot be answered by the experiments of this study.
The overlap between Crc-and CrcZ-regulated transcripts was unexpectedly low (Figure 1). There may be several reasons for this finding. In general, mRNAs are more stable when they are translated [33]. Nevertheless, in a crc mutant, derepressed translation may not always result in more stable, hence more abundant transcripts, whereas in a crcZ mutant, in which translational initiation of target mRNAs is repressed, the (negative) effects on transcript abundance may be more obvious. However, under both growth conditions used here, expression of crcZ is low and thus Crc-mediated repression of target mRNAs by Crc is strong [1]. Therefore, a crcZ mutation may result only in minor and possibly undetectable changes of the expression of some genes. As shown in Table 2, RT-qPCR (as a more sensitive technique than the microarray analysis) established that the estA, acsA, bkdA1 and aroP2 transcripts were regulated by Crc and CrcZ/CbrB in an opposite manner. However, there are also limitations. For example, bkdR did not reveal changes of transcript levels, whereas a bkdR'-'lacZ reporter fusion was regulated (Table 2). Finally, many genes were not induced at the transcriptional level under the growth conditions used, making changes due to catabolite repression difficult to detect. Despite such limitations, we have been able to use transcriptome analysis as a tool to identify several novel targets of the CbrAB/Crc system.
The main targets of the CbrAB/Crc system seem to be metabolic transcripts, which is expected for a catabolite repression control system. Moreover, as noted before [6,7,34], the CbrAB/ Crc system not only regulates genes for the utilization of less preferred C sources, but also dramatically influences motility and biofilm development, traits that are involved in the pathogenicity of P. aeruginosa and that depend on nutritional conditions. For example, a defect in CbrAB-mediated regulation of aromatic and branched-chain amino acid utilization affected growth and biofilm development of P. aeruginosa in sputum medium (Figures 6 and S2). In addition to aroP2 and bkdR (Figures 4 and 5), the braC gene and the braDEFG operon (encoding transport proteins with high affinity for the branched-chain amino acid leucine, isoleucine and valine and with low affinity for alanine and threonine; [35,36]), were down-regulated in the crcZ mutant in both media and upregulated in the crc mutant in BSM (Table 1). An incomplete CA motif (UACAACAA at +23 to +30) in the coding region of braD as well as an extended, but imperfect CA motif (ACAACAAUGA-CAACA at 213 to 26) upstream of the braC start codon suggests that the CbrAB/Crc system also controls these genes. However, further investigation is required to confirm this regulation.
In sputum medium a crc mutant had no growth advantage over the wild type ( Figure 6) and there is currently no evidence that colonization of the CF lung by P. aeruginosa would select for crc mutants. However, both the crcZ and the cbrB mutant were partially deficient for growth in SCFM and may therefore be affected in virulence.
It has been known for some time that nutritional conditions influence biofilm development [37,38] and that the CbrAB/Crc system is an important regulator for biofilm formation, although some published results appear to be strain-specific [6,7,34,39]. For example, a PA14 Dcrc mutant produced less biofilm in minimal media containing glucose [6,34] whereas a PAO1 Dcrc strain showed enhanced biofilm formation in LB compared with the wild type strain [7]. Our data indicate that the impact of the CbrAB/ Crc system on biofilm development depends on media and carbon sources ( Figure S2). In BSM medium, different carbon sources clearly affected the formation of static biofilms ( Figure S2), which is consistent with different carbon sources having different regulatory effects on targets of the CbrAB/Crc system. By contrast, in LB medium the typical pattern of upregulation in the crc mutant and strong repression in the crcZ and cbrB deletion strains was not observed. In conclusion, the CbrAB/Crc system acts on a variety of different transcripts involved in biofilm development. Finding out more about the relationships between nutrient availability, catabolite repression control and regulation of virulence and biofilm development in P. aeruginosa may ultimately lead to a better understanding of the bacterium's pathogenicity and intrinsic resistance to antimicrobial agents.

Methods
Bacterial strains and growth conditions -The strains and plasmids used in this study are listed in Table S3. Unless indicated otherwise, cells were grown in Luria Broth (LB) [12,40] or in a minimal medium (BSM) [41] supplemented with 40 mM succinate [1]. For the investigation of special targets and growth behavior we used BSM amended with either 4 mM tyrosine, 5 mM succinate +40 mM acetate, 40 mM acetate or 23 mM leucine +25.6 mM valine +7.6 mM isoleucine as the sole carbon sources and LB supplemented with 23 mM leucine +25.6 mM valine +7.6 mM isoleucine. The synthetic sputum medium SCFM has been described by Palmer et al. [30]. When required, antibiotics were added to the media at the following concentrations: 100 mg ml 21 ampicillin for Escherichia coli and 125 mg ml 21 tetracycline for P. aeruginosa.
Transcriptome analysis -Overnight cultures of P. aeruginosa PAO1, the DcbrB mutant PAO6711, the Dcrc mutant PAO6673 and the DcrcZ mutant PAO6679 were diluted to an initial OD 600 of 0.05 in 20 ml of LB or BSM amended with 40 mM succinate. The cultures were grown at 37uC with vigorous shaking until they reached an OD 600 of 1.6. RNA purification, cDNA synthesis and cDNA hybridization were performed as described previously [42]. Processing of the P. aeruginosa GeneChip Array (Affymetrix) was performed at the University of Lausanne Center for Integrative Genomics. For each condition, cultures were grown in triplicate, and RNAs from these cultures were pooled before proceeding to