Characterization of the 2-ketogluconate utilization operon in Pseudomonas aeruginosa PAO1

Identification of a second PtxS binding site

We have shown previously that PtxS binds to a 20 bp region within the ptxS upstream region (Swanson et al., 1999). Within this 20 bp fragment, a 14 bp palindrome exists (TGAAACCGGTTTCA), which may represent the PtxS operator site (B. L. Swanson and A. N. Hamood, manuscript in preparation; Swanson et al., 1999). In this study, we tried to determine whether the palindromic sequence exists in other regions of the P. aeruginosa chromosome and if PtxS specifically binds to this sequence. The presence of such a sequence within the upstream region of other genes may indicate that the expression of these genes is regulated by PtxS. We have searched the complete nucleotide sequence of the P. aeruginosa strain PAO1 (at http://www.pseudomonas.com) for potential PtxS binding sites. The 14 bp palindrome was identified at two locations: one within the ptxS upstream region as determined previously (Swanson et al., 1999); and the other sequence immediately 3′ of ptxS (on the same DNA strand). Initial DNA binding experiments were conducted to determine whether the DNA fragment that carries this palindromic sequence specifically binds to PtxS. Analysis of the region downstream of ptxS revealed that the potential PtxS binding site is carried on a 520 bp KpnI–HindIII fragment (data not shown). The 520 bp fragment was obtained from plasmid pAH50 that we had constructed previously during the isolation and characterization of ptxR and ptxS (Hamood et al., 1996). The fragment was end-labelled using T4 polynucleotide kinase, as described previously (Swanson et al., 1999), and incubated in a typical DNA binding reaction with the cell lysate of Escherichia coli K38/pJAC17 (in which PtxS was overproduced). Incubation of the 520 bp fragment with the lysate of K38/pJAC17 produced a gel shift band (Fig. 1A, lane 5). The specificity of this binding band was confirmed by competition experiments, in which the gel shift band was eliminated upon the addition of excess unlabelled 520 bp fragment (Fig. 1A, lane 6). In another competition experiment, the binding band was not eliminated upon the addition of excess unlabelled non-specific DNA (Fig. 1B). A gel shift band that migrated at the same distance was detected when the 520 bp fragment was incubated with the lysate of the P. aeruginosa strain PAO1 (Fig. 1A, lane 2). No gel shift band was detected when the 520 bp fragment was incubated with the lysate of the ptxS isogenic mutant PAO ptxS::Ω (Fig. 1A, lane 3). To confirm these results further, we localized the binding activity to a 50 bp fragment 3′of ptxS. This fragment, which contains the 14 bp palindromic sequence, was synthesized by polymerase chain reaction (PCR). A specific gel shift band was detected when the 50 bp fragment was incubated with the lysates of K38/pJAC17 and PAO1, but not PAOptxS::Ω (data not shown). These results indicate that PtxS specifically binds to this palindromic sequence. We will refer to the PtxS operator site within the ptxS upstream region as OP1 and to the newly identified operator site as OP2.

Identification of a potential 2-ketogluconate utilization operon

OP2 may be located within the upstream region of a gene(s) that is regulated by PtxS. To examine such a possibility, we searched the available P. aeruginosa genomic sequence for a potential open reading frame(s) (ORFs) in the region 3′ of ptxS. We have identified four potential ORFs that are transcribed in the same direction (Table 1) (Fig. 2). There were no clear transcription initiation or transcription termination signals between these genes.

ORF1, ORF2, ORF3 and ORF4

The 783 bp ORF1, which is located 72 bp downstream of ptxS and has a G+C content of 67%, encodes a 261-amino-acid protein with a predicted molecular weight of 28.9 kDa (Table 1; Fig. 2). Although the amino acid sequence of this predicted protein showed no significant similarity to other proteins in the database, it contains all the conserved amino acids within the aldose 1-epimerase proteins motif (fy) p… (ilv)… (ilv).y. (ilv) (at position 219–231) (Bouffard et al., 1994; Blattner et al., 1997; data not shown). A putative Shine–Dalgarno (SD) sequence was identified 12 bp from the predicted ATG start codon (data not shown). However, based on comparisons with known E. coli promoters, no −10 or −35 sites were found upstream of this start codon. ORF2 is located 28 bp 3′ of ORF1 and has 912 nucleotides and a 67.8% G+C content (Table 1). A putative SD sequence was identified 12 bp from the predicted ATG start codon (data not shown). The 32.6 kDa protein encoded by ORF2 showed a high degree of homology to several carbohydrate kinases. The most significant homology was detected with the 2-keto-3-deoxygluconate kinase of Bacillus subtilis, KdgK (46% identity and 59% similarity; Table 1). The protein belongs to the PfkB family of carbohydrate kinases (Wu et al., 1991). Both signature 1 and 2 of this family were detected at the amino- and carboxy-terminus regions of the predicted protein (Fig. 3). In addition, the protein contains a putative ATP-binding motif (amino acids 256 to 277) (Fig. 3). ORF3 is located 62 bp downstream of ORF2. It contains 1308 nucleotides, a 66.3% G+C content and a potential SD sequence 12 bp upstream of the ATG start codon (Table 1). Although potential −10 and −35 sites that resemble those of the E. coliσ⁷⁰ were identified, these sequences overlap the SD site and are less likely to function in the transcription of ORF3. Hydropathy analysis of the 48 kDa predicted protein encoded by ORF3 showed several hydrophobic regions, which suggests that the protein is an integral membrane protein (data not shown). In addition, the amino acid sequence of the protein showed similarity to several transporter proteins (data not shown). The highest degree of similarity was to the putative tartrate transporter of Agrobacterium vitis (50% similarity and 32% identity). The protein is likely to belong to the anion:cation symporter (ACS) family of transporters (Pao et al., 1998). It contains all but one of the amino acids within the conserved G(X)₂E(X)₄P(X)₈W(X)P(X)₂ER motif (data not shown; Pao et al., 1998). ORF4 is located 18 bp 3′ of ORF3. It consists of 987 nucleotides, has a 70.8% G+C content and a potential SD sequence 10 bp upstream of the ATG start codon (Table 1). The amino acid sequence of the 35.59 kDa predicted protein encoded by ORF4 showed a high degree of homology to several known dehydrogenases (data not shown). The most significant resemblance (57% identity and 68% similarity) was to the E. coli putative 2-hydroxyacid dehydrogenase, YiaE (Fig. 4). The protein contains most of the conserved amino acids within the 2-hydroxyacid dehydrogenase signal (amino acids 157 to 185 and 207 to 229; Fig. 4). In addition, other potential motifs (such as the ATP/GTP and cyclic nucleotide-binding motifs) were identified (Fig. 4).

Expression of ORF1 and ORF2 in E. coli

To determine the translational products encoded by the potential ORFs, we expressed the genes from the T7 promoter and selectively labelled their products using the T7 expression system (Tabor and Richardson, 1985). The DNA fragment that carries an intact ptxS, ORF1, ORF2 and ORF3 was obtained from the original cosmid (pAH50) that contains ptxS and ptxR (Hamood et al., 1996). Based on the available nucleotide sequences of the three ORFs, we isolated a 5.7 kb BamHI fragment that carries these ORFs from pAH50 and cloned it into the BamHI site of the pT7-6 expression vector generating pBS50. The orientation of the genes carried on the fragment was confirmed using available restriction sites. The cloned genes were expressed from the T7 promoter, and their products were selectively radiolabelled with [³⁵S]-methionine as described previously (Tabor and Richardson, 1985). The E. coli K-38 strain containing the plasmid vector pT7-6 was used as a negative control. As shown in Fig. 5, lane 2, three bands that represent proteins encoded by the ORFs carried on the BamHI fragment were detected. The band at approximately 37 kDa most probably represents PtxS (based on previous analyses) (Colmer and Hamood, 1998). The 32 kDa and 29 kDa polypeptides most probably represent the products encoded by ORF2 and ORF1 respectively (based on the comparison with the size of the predicted proteins; Table 1). No polypeptide was detected that represents the predicted 48 kDa product of ORF3. The reason for our failure to detect this product is not known at this time.

Involvement of ORFs 1–4 in 2-ketogluconate utilization in P. aeruginosa

The above results suggest that ORFs 1–4 may constitute a potential operon that is involved in carbohydrate metabolism in P. aeruginosa. To examine this possibility, the ability of an isogenic mutant in this operon to grow in minimal medium containing either glucose or other glucose metabolites as a sole carbon source was examined. An isogenic mutant that carries a deletion within ORF1 was constructed in the P. aeruginosa strain PAO1 as described previously (Stibitz et al., 1986; Colmer and Hamood, 1998). In this mutant (PAO1kgu::Ω), a 276 bp StyI internal fragment of ORF1 was deleted and replaced by the 2 kb omega fragment, which carries transcription termination and translation stop codons in both orientations (Prentik and Kirsch, 1984; Fig. 2). The construction of the mutant was confirmed by Southern blot hybridization experiments (data not shown). If ORFs 1–4 constitute an operon, the mutation in ORF1 should have a polar effect on the transcription of the other genes within the potential operon. Both PAO1 and PAO1kgu::Ω were grown in minimal media containing a 10 mM concentration of glucose, 2-ketogluconate, gluconate, glucose-6-phosphate or 6-phosphogluconate (products of the P. aeruginosa glucolysis cycle) as a sole carbon source. In comparison with PAO1, PAO1kgu::Ω grew in all media tested except the one containing 2-ketogluconate as the sole carbon source (Table 2).

Further complementation experiments, using the previously described P. aeruginosa 2-ketogluconate-negative mutant PRP621 (Temple et al., 1998), were conducted to confirm that ORFs 1–4 are involved in 2-ketogluconate utilization. This mutant, which mapped at minutes 46–48 of the P. aeruginosa chromosome, does not grow in a minimal medium in which 2-ketogluconate is the sole carbon source (Temple et al., 1998; P. Phibbs, personal communication). Thus, plasmid pAH50 (which carries ptxS and ORFs 1–4) was introduced into the PRP621 strain. In comparison with PRP621, which carries a cloning vector (pLAFR1), PRP621/pAH50 grew in M-9 minimal medium supplemented with 2-ketogluconate as the sole carbon source (Table 2). These results suggest that ORFs 1–4 constitute a potential operon that is involved in the metabolization of 2-ketogluconate in P. aeruginosa. We propose to call this operon the kgu operon.

Depending on the growth conditions, P. aeruginosa metabolizes glucose to the intermediate 6-phosphogluconate by three possible pathways (Temple et al., 1998). Under anaerobic and oxygen-limited conditions, glucose is actively transported into the periplasm by a specific glucose binding protein and then converted into 6-phosphogluconate (Hunt and Phibbs, 1983; Temple et al., 1998). Under aerobic conditions, glucose is first oxidized in the periplasm into gluconate or 2-ketogluconate (Roberts et al., 1973; Temple et al., 1998). The gluconate is imported into the periplasm by a specific permease (encoded by gnuT gene) and assimilated into 6-phosphogluconate by a glucokinase (encoded by the gnuK gene) (Temple et al., 1998). Temple et al. (1998) suggested previously that 2-ketogluconate is actively transported across the cytoplasmic membrane of P. aeruginosa by an active transport system and converted into 6-phosphogluconate by an ATP-dependent kinase (KguK) and an NAD(P)H-dependent reductase (KguR). Although the products of ORFs 1–4 have not been biochemically characterized, their amino acid homology to other previously characterized proteins (Table 1) suggests that they are required for different aspects of 2-ketogluconate utilization. For example, ORF1 may code for an aldose-epimerase. In addition, similar to gnuT, ORF3 may code for the 2-ketogluconate permease. Furthermore, ORFs 2 and 4 may code for KguK and KguR respectively. Therefore, we propose the names kguE, kguK, kguT and kguD(dehydrogenase) for ORFS 1–4 respectively (Fig. 2). The location of these genes (and their products) in the 2-ketogluconate utilization pathway is illustrated in Fig. 6.

Identification of the PtxS molecular effector (kgu molecular inducer)

Binding of PtxS to the upstream region of ORF1 suggests that the putative kgu operon is regulated by PtxS (Fig. 1). Similar to other GalR–LacI proteins, this binding may be modulated by an effector molecule (one of the metabolites of the 2-ketogluconate utilization pathway). Therefore, to identify the PtxS molecular effector, the effect of different metabolites on the in vitro binding of PtxS to the 520 bp KpnI–HindIII fragment, which carries OP2, was examined. In a typical protein–DNA binding reaction, the lysate of the E. coli strain K38/pJAC17 was incubated with the 520 bp fragment in the presence of 10 mM glucose, 2-ketogluconate, gluconate, glucose-6-phosphate or 6-phosphogluconate. As shown in Fig. 7A, the binding of PtxS to the 520 bp fragment was inhibited in the presence of 2-ketogluconate. None of the other sugars tested caused any detectable change (Fig. 7A).

We also examined the effect of 2-ketogluconate on PtxS binding to OP1. Previous studies have shown that effector molecules interfere with the binding of several of the autoregulated proteins of the GalR–LacI family to the upstream region of their genes (Gerlach et al., 1990; Rolfes and Zalkin, 1990; Weickert and Adhya, 1993a). The lysate of E. coli strain K38/pJAC17 was incubated with the 52 bp fragment of the ptxS upstream region (which carries OP1) in the presence of 10 mM 2-ketogluconate. As shown in Fig. 7B, 2-ketogluconate interfered with the PtxS binding to the 52 bp fragment. Other tested metabolites produced no detectable effect (Fig. 7B). These results suggest that 2-ketogluconate is the molecular inducer that affects both PtxS autoregulation and the regulation of the kgu operon by PtxS.

Examination of Fig. 7 indicates that 10 mM ketogluconate has a more pronounced effect on PtxS binding to OP1 than on its binding to OP2. As all experiments were conducted using the same amount of proteins and DNA probe, it is possible, therefore, that 2-ketogluconate inhibits PtxS binding to OP1 more efficiently than PtxS binding to OP2. To examine such a possibility, we determined the effect of two different concentrations of 2-ketogluconate on PtxS binding to OP1 and OP2. At 5 and 10 mM concentrations of 2-ketogluconate, PtxS binding to OP2 was significantly reduced, whereas PtxS binding to OP1 was eliminated (data not shown). Additional titration experiments were conducted to determine the minimum concentration of 2-ketogluconate that is required to interfere with PtxS binding to OP1. PtxS binding to OP1 was inhibited at 3 mM concentration and reduced at 1 mM concentration of 2-ketogluconate (Fig. 8). However, PtxS binding to OP1 was not affected at either 100 µM or 10 µM 2-ketogluconate (Fig. 8).

Determining whether ptxS is part of the kgu operon

The presence of the14 bp palindromic sequence within the upstream region of kguE (Fig. 2) suggests that the kgu operon is regulated by the adjacent, but separate, ptxS gene. However, it is also possible that ptxS, which is separated from ORF1 by 72 bp and is transcribed in the same direction as ORF1 (Fig. 2), is part of the kgu operon. Thus, to determine whether kguE–kguD constitutes a separate operon, we investigated the presence of a potential promoter within the 520 bp KpnI–BamHI fragment that carries the ptxS–ORF1 intergenic region (Fig. 2). The 520 bp fragment was cloned in the lacZ broad-host-range vector pMP190 (Spaink et al., 1987), and the resulting plasmid (pBS54), in which the lacZ gene is transcribed in the same direction as ORF1, was introduced into PAO1. Cells were grown in LB broth containing 2% glucose (to induce the potential kgu operon). The presence of high levels of glucose in the growth media has been shown to induce the P. aeruginosa 2-ketogluconate pathway (Lessie and Phibbs, 1984). However, the level of β-galactosidase activity produced by pBS54 was similar to that produced by the control strain (PAO1 carrying the cloning vector pMP190; data not shown). This suggests that ptxS together with the other four genes constitute an operon.

To support this possibility further, we examined the ability of the previously constructed ptxS isogenic mutant (PAO1ptxS::Ω) (Colmer and Hamood, 1998) to grow in a minimal medium containing 2-ketogluconate as the sole carbon source. If ptxS is a separate gene, PAO1ptxS::Ω would be capable of using 2-ketogluconate (the negative regulation of the kgu operon is eliminated). Alternatively, if ptxS is part of the kgu operon, the PAO1 ptxS::Ω mutant would not grow in this medium (similar to the kguE deletion mutant described above). The M-9 minimal medium containing 2-ketogluconate as the sole carbon source supported the growth of PAO1 but not PAO1ptxS::Ω (Table 2). These results suggest that ptxS is the first gene in the kgu operon.

Bacterial strains, plasmids, media and growth conditions

The bacterial strains and plasmids used in this study are listed in Table 3. E. coli strains were grown in Luria–Bertani medium [1% Bacto tryptone (Difco Laboratories), 0.5% yeast extract, 1% NaCl] (Miller, 1972). P. aeruginosa strain PAO1 and its ptxS isogenic mutants were grown in M-9 media (Miller, 1972) with the addition of 10 mM carbon source (Sigma-Aldrich Chemicals) as indicated. A calcium-free solution of 100 mM 2-ketogluconate was prepared as follows: 1.45 g of 2-ketogluconic acid salt (Sigma; H20 content 1.5 mol mol⁻¹) was dissolved in 48 ml of distilled water, heated to 70°C for 10 min and 12 ml of 10.5 M potassium phosphate buffer (pH 7.0) was added. After centrifugation for 15 min at 6.0 g, the resulting 100 mM 2-ketogluconate solution was decanted from the insoluble calcium phosphate and filter sterilized. Cultures were grown at 37°C with vigorous aeration. Antibiotics were used at the following concentrations: ampicillin, 75 µg ml⁻¹ (E. coli); streptomycin, 400 µg ml⁻¹ (P. aeruginosa); rifampin, 80 µg ml⁻¹ (P. aeruginosa).

DNA manipulations

DNA manipulations and other molecular biology techniques were performed essentially as described previously (Ausubel et al., 1988). Plasmids were introduced into P. aeruginosa by electroporation as described previously (Smith and Iglewski, 1989). P. aeruginosa chromosomal DNA was extracted according to the methods of Goldberg and Ohman (1984). PCR amplification of specific DNA fragments was carried out as described previously (White, 1993). Plasmid pBS54 (Table 3) was constructed by cloning the 520 bp fragment that carries the ptxS–ORF1 intergenic region (Fig. 2) in the lacZ broad-host-range vector pMP190 (Spaink et al., 1987). The resulting plasmid (pBS54), in which the lacZ gene is transcribed in the same direction as ORF1, was introduced into PAO1.

Gel shift assay

DNA fragments were obtained by either restriction digestion or PCR. The fragments were labelled with [γ³²P]-ATP (Amersham-Pharmacia Biotech) by the end-labelling technique using T4 polynucleotide kinase (Ausubel et al., 1988). The binding experiments were performed as described previously (Swanson et al., 1999) with the addition of 10 mM carbon source to the binding reaction as indicated.

Expression experiments

PtxS was overproduced in the E. coli strain K-38 for use in the gel shift assay by the T7 expression system as described previously (Tabor and Richardson, 1985; Colmer and Hamood, 1998). PtxS and its downstream ORFs were expressed in the T7 expression system with the addition of Tran[³⁵S]-methionine/cysteine (NEN Biochemicals) as indicated to label the expressed proteins selectively.

Nucleotide sequence analysis and protein sequence comparisons

The nucleotide sequence was analysed using the dna strider program. The protein sequence comparisons were performed using the GCG software package (Devereux et al., 1984). The Kyte and Doolittle (1982) hydropathy plot analysis was used to examine the amino acid sequence of the predicted proteins for potential hydrophobic regions. DNA sequences downstream of ptxS were obtained by searching the chromosomal database at the P. aeruginosa genome project website (http://www.pseudomonasgenome.com).

β-Galactosidase assay

A 1 ml pellet of the cell cultures was washed and resuspended in 100 µl of distilled water. The cells were disrupted by sonication, and the levels of β-galactosidase activity were determined as described previously (Miller, 1972).

Construction of a kgu isogenic mutant in P. aeruginosa strain PAO1

An isogenic mutant of the P. aeruginosa strain PAO1 that carries a deletion in ORF1 was constructed by the gene replacement technique described previously (Stibitz et al., 1986; Hamood et al., 1996). Plasmid pBS47, which is a pUC18 recombinant plasmid that carries a 5.7 kb BamHI fragment of the PAO1 chromosome, was used for these experiments. The 5.7 kb BamHI fragment contains ptxS, ORF1, ORF2, ORF3 and part of ORF4 (Fig. 2). A 2.3 kb SmaI fragment, which carries ORFs 1 and 2, was isolated from pBS47 and cloned into the SmaI site of pUC18 generating plasmid pBS55. Plasmid pBS55 was digested with StyI, and a 276 bp ORF1 internal fragment was deleted. The 5′ protruding ends were converted to blunt ends, and the 2.0 kb SmaI Ω fragment was inserted into the deleted region, generating plasmid pBS56. The Ω fragment carries the T4 transcriptional termination signal and translational stop codon in both orientations (Prentik and Kirsch, 1984). Plasmid pBS56 was introduced into PAO1 by electroporation, and transformants were recovered on LB plates containing Sm. The resulting Sm^r colonies were then screened for Sm^r, Cb^s colonies. These clones represent cells in which a double cross-over event occurred, resulting in the loss of the cloning vector and the wild-type copy of ORF1 and the retention of the Ω-interrupted copy of ORF1. This insertion should create a polar effect and obliterate the transcription of the entire 2-ketogluconate utilization operon. Construction of the isogenic mutant was confirmed by Southern blot hybridization experiments as described previously (Ausubel et al., 1988).