Functional Analysis and Regulation of the Divergent spuABCDEFGH-spuI Operons for Polyamine Uptake and Utilization in Pseudomonas aeruginosa PAO1

ABSTRACT A multiple-gene locus for polyamine uptake and utilization was discovered in Pseudomonas aeruginosa PAO1. This locus contained nine genes designated spuABCDEFGHI (spu for spermidine and putrescine utilization). The physiological functions of the spu genes in utilization of two polyamines (putrescine and spermidine) were analyzed by using Tn5 transposon-mediated spu knockout mutants. Growth and uptake experiments support that the spuDEFGH genes specify components of a major ABC-type transport system for spermidine uptake, and enzymatic measurements indicated that spuC encodes putrescine aminotransferase with pyruvate as the amino group receptor. Although spuA and spuB mutants showed an apparent defect in spermidine utilization, the biochemical functions of the gene products have yet to be elucidated. Assays of lacZ fusions demonstrated the presence of agmatine-, putrescine-, and spermidine-inducible promoters for the spuABCDEFGH operon and the divergently transcribed spuI gene of unknown function. Since the observed induction effect of agmatine was abolished in an aguA mutant where conversion of agmatine into putrescine was blocked, putrescine or spermidine, but not agmatine, serves as the inducer molecule of the spuA-spuI divergent promoters. S1 nuclease mappings confirmed further the induction effects of the polyamines on transcription of the divergent promoters and localized the transcription initiation sites. Gel retardation assays with extracts from the cells grown on putrescine or spermidine demonstrated the presence of a polyamine-responsive regulatory protein interacting with the divergent promoter region. Finally, the absence of the putrescine-inducible spuA expression and putrescine aminotransferase (spuC) formation in the cbrB mutant indicated that the spu operons are regulated by the global CbrAB two-component system perhaps via the putative polyamine-responsive transcriptional activator.

Polyamines (including putrescine and spermidine) are a group of ubiquitous polycations necessary for cell growth and also serve as precursors for acetyl-and S-adenosyl-polyamines of important physiological functions in microorganisms (31,36). Biosynthesis of putrescine and spermidine has been extensively studied in Escherichia coli, and the cognate enzymes and structural genes have been established (Fig. 1). In Pseudomonas aeruginosa, the presence of all homologues of the E. coli putrescine and spermidine biosynthetic speABCDE genes (8) in the PAO1 genome (www.pseudomonas.com) suggests common pathways in these two microorganisms.
In P. aeruginosa, putrescine can be synthesized from arginine through the catabolic arginine decarboxylase (ADC) pathway (22). The ADC pathway is one of four catabolic pathways for arginine utilization in pseudomonads (11). In this pathway, exogenous arginine is converted sequentially into agmatine, putrescine, 4-aminobutyrate, and succinate before it is channeled into the tricarboxylic acid cycle (Fig. 1). Recently, we characterized the aguBA operon of P. aeruginosa PAO1 (25) that encodes agmatine deiminase (aguA) and N-carbamoylputrescine amidinohydrolase (aguB) converting agmatine into putrescine in this pathway (13) (Fig. 1). While exogenous ar-ginine can induce the synthesis of ADC catalyzing the first step of the pathway, agmatine and N-carbamoylputrescine are responsible for the subsequent induction of the aguBA operon regulated by AguR repressor (25). Thus, the regulatory mechanism of the ADC pathway is significantly different from those of other two arginine catabolic pathways, the arginine succinyltransferase (16,18,38) and the arginine deiminase pathways (6,20,23), in which exogenous arginine exerts its induction effect by the presence of the arginine-responsive ArgR regulator (27,29,30).
The catabolic route of putrescine is part of the ADC pathway. Conversion of putrescine into 4-aminobutyrate requires two enzymes: putrescine aminotransferase and 4-aminobutyraldehyde dehydrogenase (Fig. 1). While genetic evidence has indicated that the kauB gene encodes a bifunctional 4-aminobutyraldehyde/4-guanidinobutyraldehyde dehydrogenase (17), putrescine aminotransferase has not been characterized, and the corresponding gene remains to be identified in P. aeruginosa.
In the present study, we report the characterization of the divergent spu operons of nine genes which are located adjacent to the aguRBA genes and are responsible for polyamine utilization. Among these genes, components of a major spermidine transport system and the putrescine aminotransferase of the ADC pathway were identified. We also demonstrated the induction effect of putrescine and spermidine on the expression of these divergent operons and a putative polyamine-respon-sive trans-acting factor interacting with the regulatory region of the divergent promoters.

MATERIALS AND METHODS
Bacterial strains, plasmids, media, and growth conditions. Bacterial strains and plasmids used in this study are listed in Table 1. The Luria-Bertani (LB) enriched medium or nutrient yeast broth (21) was used for strain construction with the following supplements as required: ampicillin, 50 g/ml (for E. coli); carbenicillin, 100 g/ml (for P. aeruginosa); streptomycin, 500 g/ml; and tetracycline, 50 g/ml. The minimal medium P (MMP) (12) was used for the growth of P. aeruginosa with supplements of carbon and nitrogen sources at 20 mM as indicated.
Construction of mutant strains. The EZ::TET-1 insertion system (Epicentre) based on the in vitro Tn5 transposition system (9) was used for generation of knockout mutations. Two EcoRI fragments covering the entire agu and spu loci were purified from cosmid pGU2 and subcloned separately into the EcoRI site of the conjugation vector pRTP1. Each resulting target plasmid DNA was incubated with the transposase and the transposon with a tetracycline resistance marker, and the in vitro transposon insertion reaction was carried out under the conditions recommended by the manufacturer. After the reaction, the reaction mixture was used to transform E. coli DH5␣, and transformants were selected on LB plates with tetracycline. The insertion sites of mutant clones were mapped by NcoI restriction endonuclease digestion and subsequently by nucleotide sequencing with a transposon-specific flanking primer. For gene replacement, the resulting transposon insertion plasmids were first introduced into E. coli SM10 and then mobilized into the spontaneous streptomycin-resistant P. aeruginosa strain PAO1-Sm by biparental plate mating (7). After incubation at 37°C overnight, transconjugants were selected on LB plates supplemented with tetracycline and streptomycin. Knockout mutants of spuC were also constructed by using an Flp recombinase target (FRT) sequence and an ⍀sp/sm cassette. The 3.3-kb EcoRI-XhoI fragment carrying spuC was first subcloned from cosmid pGU2 into suicide plasmid pEX18Ap between the EcoRI and SalI sites (14) to yield plasmid pYJ96. The ⍀sp/sm and Gm-GFP-FRT cassettes were excised as SmaI fragments from plasmids pHP45 ⍀sp/sm (4) and pSP858 (14), respectively, and subsequently inserted into the blunt-ended NotI site of spuC on pYJ96. The resultant plasmids were then mobilized into strain PAO1, as described above, to obtain strains PAO4479 (spuC::⍀sp/sm) and PAO4480 (spuC::Gm-GFP-FRT). The Gm-GFP sequence in the cassette on the spuC gene was then removed by introducing plasmid pFLP2 (carrying the Flp recombinase gene) into strain PAO4480 as described previously (14), yielding strain PAO4486 (spuC::FRT).
Construction of lacZ fusions. Plasmid pQF52, a broad-host-range lacZ translational fusion vector (30), was used in the construction of promoter fusions. For construction of pGU101 (P spuI ::lacZ) and pGU102 (P spuA ::lacZ), a DNA fragment containing the spuI-spuA intergenic region was amplified from pGU2 by PCR with two oligonucleotide primers: oligo-1 (5Ј-GCCGGTGATATCGAGG GCGAA-3Ј) and oligo-2 (5Ј-GAGACCGCCAGCGCCGTTGAC-3Ј). The PCR fragment was cloned to the SmaI site of pQF52 so that the 68th codon of spuI and the 39th codon of spuA were fused in-frame to lacZ of the vector in the resulting plasmids. The orientation of the insert on the plasmids was confirmed by nucleotide sequencing.
Transformation of E. coli and P. aeruginosa by plasmid DNA was performed as described by Chung et al. (2) with magnesium ion for one-step preparation of competent cells. For determination of lacZ expression, the levels of ␤-galactosidase activity in logarithmically growing cells were measured by using ONPG (o-nitrophenyl-␤-D-galactopyranoside) as the substrate (24) with cell extract prepared by passing cells through a French pressure cell at 8,000 lb/in 2 . Protein concentration was determined by the method of Bradford (1). S1 nuclease mappings. RNA samples were prepared from P. aeruginosa PAO1 grown in MMP to an optical density at 600 nm (OD 600 ) of 0.5 to 0.6. A 30-ml portion of the culture was collected by centrifugation at 12,000 ϫ g at 4°C for 5 min, and RNA was purified from the suspended cell pellet by a previously described protocol (27).
Procedures for hybridization and S1 nuclease digestion were followed as described by Greene and Struhl (10). For characterization of spuI and spuA promoters, a double-stranded DNA fragment covering the 363-bp spuI-spuA intergenic region (Fig. 4b) was amplified by PCR from pGU101 with a pair of oligonucleotide primers: oligo-5 (5Ј-AACGGCACGCTGGGGTAC-3Ј) and oligo-6 (5Ј-GGGATGAAGACCGATCTGC-3Ј). One of these two primers was end labeled with [␥-32 P]ATP by T4 polynucleotide kinase before PCR, and the resulting PCR product was purified after agarose gel electrophoresis on a 1% agarose gel. For each reaction, 50 g of RNA was hybridized with the radioactive probe, and experiments were performed quantitatively to permit comparison of levels of transcripts under different growth conditions. The sizes of these transcripts were determined against a nucleotide sequencing ladder of the probe DNA.
Spermidine uptake experiments. Cultures were grown in glutamate-MMP in the absence or presence of 20 mM spermidine. Cells were harvested during logarithmic growth, washed twice with MMP, suspended at a concentration of ca. 10 8 cells/ml (OD 600 ϭ 0.1) in MMP containing chloramphenicol (250 g/ml). After incubation of the cell suspension for 5 min at 37°C, 3 H-labeled spermidine (New England Nuclear) was added to a final concentration of 20 M (44.5 mCi/mmol), and samples (0.5 ml) were withdrawn at various time intervals. Cells were collected on a cellulose membrane filter (0.22 m [pore size], type GS; Millipore) and washed with 5 ml of MMP. Incorporated radioactivity was measured in a liquid scintillation spectrometer (Beckman). In competition experiments, cold agmatine, putrescine, or spermidine (final concentration of 2 mM) was mixed with the 3 H-labeled spermidine before being added to the cell suspension.
Gel retardation assays. A 0.5-kb DNA fragment containing the spuI-spuA divergent promoter region was purified from pGU101 after digestion with Hin-dIII and BamHI restriction enzymes. The DNA probe was prepared by labeling with [␣-32 P]dATP by using the Klenow fragment. The radioactively labeled DNA probe (0.1 nM) was allowed to interact with crude cell extracts (2 g) in 20 l of a mixture containing 50 mM Tris-HCl (pH 7.5), 50 mM KCl, 1 mM EDTA, 5% (vol/vol) glycerol, and 200 ng of nonspecific sheared calf thymus DNA. Reaction mixtures were incubated for 10 min at room temperature and applied to a 5% polyacrylamide gel in Tris-acetate-EDTA running buffer. After being dried, the gel was autoradiographed by exposure to a phosphorimager plate (Fuji).
Measurements of putrescine aminotransferase activity. Cultures were grown to an OD 600 nm of 0.5 in MMP supplemented with the indicated carbon and nitrogen sources at 20 mM and harvested by centrifugation. Cell extracts were prepared by passing cells through a French pressure cell at 8,000 lb/in 2 . The assay FIG. 1. Schematic presentation of the ADC pathway and polyamine biosynthesis. Solid and broken arrows represent biosynthetic and catabolic pathways, respectively. SAM, S-adenosyl methionine; dSAM, decarboxylated SAM. The speABCDE genes, which encode biosynthetic ADC, agmatine ureidohydrolase, ornithine decarboxylase, SAM decarboxylase, and spermidine synthase, respectively (8), have been characterized in E. coli. The aguAB genes encode agmatine deiminase and N-carbamoylputrescine aminotransferase as characterized recently (25). The spuC gene as reported in this study encodes putrescine-pyruvate aminotransferase. The kauB and adcA genes represent the bifunctional 4-aminobutyraldehyde/guanidiobutyraldehyde dehydrogenase (17) and the arginine-inducible ADC (22). mixture (0.8 ml) contained 125 mM Tris-HCl (pH 10), 6.25 mM putrescine, 6.25 mM pyruvate, 125 M pyridoxal phosphate, 1.25 mM -aminobenzaldehyde, and cell extract (ca. 50 g). After 20 min of incubation at 37°C, 0.2 ml of 10% (wt/vol) trichloroacetic acid was added. The protein precipitates were removed by centrifugation, and the absorbance at 435 nm was measured. The amounts of 4-aminobutyraldehyde formed were calculated from the molar extinction coefficient (A 435 ϭ 0.17 ϫ 10 3 /M/cm) of ⌬ 1 -pyrroline (the cyclic form of 4-aminobutyraldhyde) after reaction with -aminobenzaldehyde. One unit of enzyme activity was defied as the amount of the enzyme that yielded 1 mol of product per min.

RESULTS
Construction and growth phenotype analysis of knockout mutants in the spu locus. In our previous report on characterization of the agu genes for agmatine utilization, the upstream genes ( Fig. 2) with potential biochemical functions in polyamine transport and metabolism have caught our attention. These genes encode components for a putative ABC trans-porter of polyamines (PA0295 and PA0300 to PA0304) and enzymes of unknown functions (PA0296 to PA0299). To investigate the possible involvement of these genes in polyamine uptake and metabolism, knockout mutants of these genes were generated by in vitro transposon mutagenesis, followed by biparental conjugation for homologous recombination as described in Materials and Methods. The artificial transposon cassette, EZ::TET-1, used here does not have a polar effect on expression of polycistronic transcripts since no transcriptional terminator sequence is present in the insertion cassette and as has been verified from a previous study (25). Utilization of agmatine, putrescine, and spermidine by these mutants as the sole source of carbon and nitrogen was examined in MMP, and the results of observed growth phenotypes are summarized in Fig. 2b.
In comparison to the wild-type strain PAO1, mutant strains PAO5006 to PAO5013 were defective in the utilization of supE44 hsdS20(rB-mB-) recA13 ara-14 proA2 lacY galK2 rpsL20 xyl-5 mtl-1 21 S17-1 pro thi hsdR Tp r Sm r : chromosome::RP4-  Fig. 2). Accordingly, these genes were designated spuABCDEFGH for spermidine and putrescine utilization. Although spermidine utilization was affected to different extents by any of the spu mutations, strain PAO5008 (spuC::Tc) was defective completely on putrescine and partially on agmatine utilization. The knockout mutants of the PA0295 and PA0296 genes, which are divergently transcribed from the spuABCDEFGH genes, did not show any apparent defect on the utilization of spermidine, putrescine, and agmatine. Additionally, two spuC mutants were constructed by inserting either an FRT sequence (PAO4486) or a streptomycin ⍀-loop cassette (PAO4479). Like strain PAO5008, strain PAO4486 exhibited a growth defect on both putrescine and agmatine. In contrast, strain PAO4479 showed no growth on agmatine, putrescine, and spermidine. Since the spuC function is dispensable for spermidine utilization in PAO4486 and PAO5008, the observed growth defect of PAO4479 on spermidine appeared to be due to an expected polar effect of the ⍀ cassette in spuC on expression of the downstream spuDEFGH genes and thus supports the expression of these transport genes (see below) from a promoter upstream of spuC.
SpuDEFGH are components of a major uptake system of spermidine in P. aeruginosa. In accordance with the growth phenotype analysis (Fig. 2), the primary amino acid sequences of these Spu proteins exhibit the highest similarities to those of the Pot transport systems of E. coli for putrescine and spermidine uptake. Sequence analysis indicated that SpuD and SpuE are the periplasmic binding protein components, SpuF is the ATPase component, and SpuG and SpuH are the inner membrane permease components for an ABC-type transport system.
To substantiate the proposed function of Spu proteins, spermidine uptake experiments were conducted in the wild-type strain PAO1 and two spu mutants (spuD and spuE) grown in glutamate minimal medium in the presence or absence of spermidine. As shown in Fig. 3a, spermidine uptake in the wild-type strain PAO1 was induced ca. 10-fold by the presence of spermidine in the growth medium. Although the induction effect of spermidine persisted in the spuD and spuE mutants, the induction levels were significantly reduced in these mutants. The initial rate of spermidine uptake in both mutants These results indicate that both SpuD and SpuE are components of an inducible spermidine transport system that accounts for most of the spermidine uptake activities in strain PAO1 growing on spermidine. The ligand specificity of the Spu transport system was analyzed by competition tests, and the results are shown in Fig. 3b. When challenged with a 100-fold molar excess of cold spermidine, putrescine, and agmatine, the spermidine uptake in the wild-type strain PAO1 as measured by uptake of radioactive spermidine was reduced to 6, 46, and 58%, respectively. Similar levels of reduction were also observed in the spuD and spuE mutants. These results indicate that spermidine is the preferred ligand molecule of the SpuDEFGH transport system.
Abolishment of the putrescine aminotransferase activity in the spuC mutant. As shown in Fig. 2, the spuC knockout mutants (PAO4479, PAO4486, and PAO5008) lose the ability to utilize putrescine as the sole source of carbon and nitrogen. A homology search against the protein database revealed that the SpuC protein exhibits strong sequence similarities to the class III enzymes of the pyridoxal-dependent aminotransferase family, with the highest similarity (53% similarity) to an omega amino acid-pyruvate aminotransferase of P. putida (PDB ac-cession number P28269). Accordingly, the possibility of spuC encoding a putrescine-pyruvate aminotransferase (PATase) was investigated. With pyruvate rather than ␣-ketoglutarate as the amino group receptor, the results of PATase measurements in the wild-type strain PAO1 revealed that the putrescine-or agmatine-grown cells exhibited a fivefold-higher level of PATase activity than that of the glutamate-grown cells ( Table 2). The induction effects of putrescine and agmatine were reduced to 2.5-fold by the presence of glutamate in the growth medium. In contrast, in the spuC mutant, the level of PATase activity was greatly diminished and no longer inducible by putrescine. These results demonstrated that the spuC gene encodes PATase for putrescine catabolism.
Transcriptional induction of the spuA promoter by polyamines. Based on the gene organization and the results of functional analysis as described above, we hypothesized that the spuABCDEFGH genes constitute an operon which is subjected to regulation by exogenous polyamines. To test this hypothesis, the PA0296-spuA intergenic region was cloned into a lacZ fusion vector, pQF52, and the resulting spuA::lacZ fusion plasmid, pGU102, was introduced into the wild-type strain PAO1.
The recombinant PAO1 harboring pGU102 was grown in the glutamate minimal medium in the absence or presence of agmatine, putrescine, or spermidine. As measured by ␤-galactosidase activity, expression of the spuA::lacZ fusion was increased 11-, 8-, and 14-fold by exogenous agmatine, putrescine, and spermidine, respectively (Table 3). These results indicate the presence of a polyamine-responsive spuA promoter in the intergenic region.
Since agmatine can serve as the precursor for polyamines synthesis via the ADC pathway (Fig. 1), the induction effect of agmatine could be due to the elevated intracellular pools of putrescine and/or spermidine. Along this line, a genetic block on conversion of agmatine to putrescine is expected to abolish the induction effect of agmatine. To test this hypothesis, plasmid pGU102 was introduced into an aguA mutant PAO5001, and the expression profile of the spuA::lacZ fusion was determined. As shown in Table 3, the induction effect of agmatine  The spuI gene is also subjected to transcriptional induction by polyamines. As shown in Fig. 1, sequence analysis predicted that the PA0296 gene is divergently transcribed from the spuABCDEFGH genes. The amino acid sequences of PA0296 and SpuB possess 34% identity to each other and exhibit 26 and 28% identity, respectively, to the amino acid sequence of glutamine synthetase (GlnA) of P. aeruginosa. However, while the spuB gene is required for spermidine utilization, the knockout mutant of the PA0296 gene, as in strain PAO5005 (Fig. 2), did not show any growth defect on agmatine, putrescine, and spermidine. Nevertheless, the results as described below indicate that the expression of PA0296 is inducible by exogenous polyamines. Therefore, the PA0296 gene was tentatively designated spuI.
The effects of agmatine, putrescine, and spermidine on the spuI promoter activity were analyzed by measurements of ␤-galactosidase activity from the spuI::lacZ fusion on plasmid pGU101, which contains the same spuI-spuA intergenic region as pGU102 but in a reverse orientation on the vector pQF52. As shown in Table 3, exogenous agmatine, putrescine and spermidine exerted eight-, seven-, and fivefold induction effects, respectively, on the spuI expression in the wild-type strain PAO1. In the aguA mutant, an induction effect by agmatine was abolished, whereas the effects of putrescine and spermidine persisted. These results indicate that the spuI promoter is also inducible by polyamines, a finding similar to that seen with the divergently transcribed spuA promoter.
Nucleotide sequence analysis revealed the presence of a strong transcriptional terminator structure at the end of the spuI coding sequence. In fact, this transcriptional terminator abolished the extension process by Taq DNA polymerase during cycle sequencing reactions at 55°C (data not shown). Therefore, the spuI gene likely stands as a single-gene operon. S1 nuclease mappings of the divergent spuA and spuI promoters. The expression patterns of spuA and spuI promoters in the wild-type strain of P. aeruginosa PAO1 were analyzed by S1 nuclease mappings. As shown in Fig. 4a, polyamine-inducible transcripts were detected for the divergent spuA and spuI pro- 4. (a) S1 nuclease mapping of P spuI and P spuA transcripts. RNA samples and the radioactive probe for S1 nuclease mapping were prepared as described in Materials and Methods. RNA samples were purified from cells grown in MMP with 20 mM concentrations of the supplements as indicated: glu, glutamate; arg, arginine; agm, agmatine; put, putrescine; and spd, spermidine. (b) Nucleotide sequence of the spuI-spuA divergent promoter region. The 5Ј ends of the spuI and spuA transcripts are indicated as arrows, and the possible Ϫ10 and Ϫ35 sequences for the corresponding promoters are underlined. a Cells were grown in MMP with supplements as indicated at 20 mM. Glu, glutamate; Agm, agmatine; Put, putrescine; Spd, spermidine. ND, not determined. Values are the averages of two measurements, with the standard errors (not shown) all below 5% of the corresponding average. moters, a finding consistent with the results of the lacZ fusion studies as described above. By running against a nucleotide sequencing ladder (data not shown), the 5Ј ends of the transcripts from these two promoters were determined to be 45 and 34 bp upstream from the proposed ATG initiation codons of spuI and spuA, respectively. Sequences resembling the consensus Ϫ35 and Ϫ10 regions of 70 promoters were found in the appropriate distance from the determined 5Ј end of each transcript (Fig. 4b). Since all S1 mappings were done with the same amount of total RNA, higher transcript levels of spuI than spuA suggested a stronger promoter activity of spuI. These results are consistent with the measurements of lacZ fusions (Table 3).
Evidence for the presence of trans-acting factors in control of the divergent spuA and spuI promoters. The aguR gene immediately upstream of the spu genes (Fig. 2a) regulates the aguBA operon for agmatine catabolism (25). To analyze possible control of the spu expression by aguR, the spuA and spuI expression profiles were determined in the aguR mutant PAO5003. As shown in Table 3, both spuA and spuI promoters remained inducible by polyamines in the aguR mutant. In fact, the induced levels of spuA and spuI promoters in the aguR mutant were ca. 50% higher than those in the wild-type PAO1. However, the AguR protein did not interact with the spuA-spuI intergenic region by gel retardation assays (data not shown), excluding the possibility of AguR as the transcriptional regulator of the spuA and spuI promoters.
By gel retardation assays with cell extracts of the wild-type strain PAO1, we could detect the presence of a DNA-binding protein interacting specifically with the spuA-spuI intergenic region. As shown in Fig. 5, the formation of a DNA-protein complex was increased in the crude extracts of cells grown in the presence of agmatine, putrescine, or spermidine. These results suggest a possible role of this putative DNA-binding protein as a trans-acting factor in the regulation of the divergent spuA and spuI promoters.
Another possible candidate for the transcription regulator of spu genes could be the CbrAB two-component system (26). The cbrAB mutants are unable to utilize agmatine, putrescine, and spermidine, as well as a variety of other compounds (26). In the cbrB mutant PAO4455, the activity of agmatine deiminase (encoded by aguA) remained inducible by exogenous agmatine to the levels comparable to that of the wild-type strain (Table 2). Therefore, it is conceivable that the defect of agmatine utilization in PAO4455 is the result of a block on putrescine catabolism. To investigate whether the spuABC-DEFGH operon is controlled by cbrB, we measured putrescine aminotransferase activity (encoded by spuC) and ␤-galactosidase activity from the spuA::lacZ fusion in the cbr mutants. When grown in the glutamate minimal medium, the level of PATase in PAO4455 (cbrB) was significantly lower than that of the wild-type strain PAO1 (Table 2). Furthermore, this enzyme was no longer induced by exogenous putrescine. Consistent with these results, the spuA::lacZ expression of pGU102 in the cbrAB mutant (PAO5100) also became noninducible by agmatine and putrescine (Table 3).
Gel retardation assays were performed to analyze the effect of cbrAB on the expression of the putative polyamine-responsive regulatory protein for the spuI-spuA divergent promoters. As shown in Fig. 5, in comparison to the wild-type PAO1, the induction effect of agmatine, putrescine, and spermidine on formation of the putative polyamine-responsive regulatory protein was apparently reduced in the cbrAB deletion mutant PAO5100.

DISCUSSION
In the present study, we have identified and characterized the divergent spuI-spuABCDEFGH operons for spermidine uptake and spermidine-putrescine utilization. Several lines of evidence supported that the ABC transporter system encoded by the spuDEFGH genes is the major spermidine uptake system in P. aeruginosa. First, knockout mutants of any of these genes abolish specifically spermidine utilization as the sole source of carbon and nitrogen in P. aeruginosa (Fig. 2). Second, spermidine uptake is inducible in spermidine-grown cells and is greatly reduced in either spuD or spuE mutant (Fig. 3a). Third, spermidine, rather than putrescine (or agmatine), is the preferred ligand molecule for SpuD and SpuE (Fig. 3b). The primary amino acid sequences of SpuD and SpuE are 57% identical to each other, and both exhibit an average of 54% identity to PotF and 35% identity to PotD of E. coli. The PotF and PotD proteins of E. coli are the preferential periplasmic binding proteins for putrescine and spermidine, respectively (15,19). Contrary to such sequence similarities, SpuD and SpuE prefer spermidine rather than putrescine as the transport ligand as determined empirically in this report. Furthermore, unlike the E. coli Pot transport systems in which a single binding protein is sufficient for the transport activity, both SpuD and SpuE are required for the optimal uptake of spermidine by the SpuDEFGH system. As shown in Fig. 2a, residual spermidine uptake can still be observed in either spuD or the spuE mutant. The possibility of another polyamine uptake system contributing to spermidine uptake in P. aeruginosa cannot be ruled out from this study.
Four putative enzymes are encoded by the spuABC and spuI genes. While mutations on spuI did not show any growth defect on agmatine and polyamines, the contiguous spuABC genes are all related to polyamine catabolism. Except for SpuC, the FIG. 5. Gel retardation assays of a polyamine-inducible trans-acting factor and the spuI-spuA regulatory region. The crude cell extracts were prepared from cells grown in MMP containing 20 mM glutamate (glu), agmatine (agm), putrescine (put), or spermidine (spd). In the binding reactions, the radioactive DNA probe was incubated with the crude extract as indicated. Lane C is a control without cell extract. F, free probe; B, bound probe. Based on this new assignment of spuC in the ADC pathway, one would expect a growth defect of spuC mutants on agmatine. Indeed, the spuC mutants grow poorly on agmatine. Similar basal growth on agmatine has been observed with the aguA and aguB mutants, presumably due to the presence of a second route for agmatine utilization, as has been reported for P. cepacia (37); in this organism, agmatine can be converted by agmatine dehydrogenase into guanidinobutyraldehyde, a precursor of guanidinobutyrate in the arginine dehydrogenase pathway.
From the results of lacZ fusion assays and S1 nuclease mappings, we demonstrated that the divergent spuA and spuI promoters are inducible by exogenous agmatine, putrescine, and spermidine. The abolishment of agmatine effect in the aguA mutant supports the notion that putrescine and/or spermidine is the inducer molecule(s) for activation of these promoters. We were not able to differentiate further the effects of putrescine and spermidine due to the fact that putrescine is the precursor as well as the catabolic product of spermidine and that mutants defective in spermidine synthetase and spermidine catabolic enzyme were not available. Little is known about the catabolism of spermidine in pseudomonads. It has been reported that exogenous spermidine can be cleaved by spermidine dehydrogenase into putrescine and 3-aminopropionaldehyde in a strain of Pseudomonas sp. (28).
The aguR gene, which is located immediately upstream of the spu genes and encodes a repressor protein in regulation of the aguBA operon (25), was excluded from the present study as the regulatory gene of spu operons. In search of a trans-acting factor in control of the spu promoters, the results of gel retardation assays (Fig. 5) revealed the presence of a polyamineresponsive DNA-binding protein forming a nucleoprotein complex with the spuA-spuI regulatory region. With respect to the control by polyamines, the patterns of this nucleoprotein complex formation correlate well with the induction profiles of the spuA and spuI promoters, suggesting a transcriptional activator feature of this DNA-binding protein that has yet to be identified.
Another regulatory element in control of the spu operons is the CbrAB two-component system of P. aeruginosa. Although the physiological signal activating the CbrAB system is not known, the importance of this system in a global control of catabolic pathways in P. aeruginosa has been reported (26). The present results indicate that CbrB participates in polyamine induction of the spuA promoter (Table 3) and thereby the spuC gene encoding PATase (Table 2) but plays no role in the agmatine-inducible expression of aguA encoding agmatine deiminase ( Table 2). These results led us to conclude that the previously reported growth defects of cbrAB mutants on agmatine, putrescine, and spermidine are a consequence of abolished induction of spu genes by these compounds. While CbrB, an NtrC-type transcriptional activator, would be involved in activation of the 54 promoters, the spuA promoter appears to have the consensus Ϫ10 and Ϫ35 sequences for 70 -RNA polymerase holoenzyme (Fig. 4b). The CbrB response regulator might regulate the expression of the putative polyamineinducible transcription factor gene as described above.
One interesting feature of these spu genes is the presence of two putative glutamine synthetase homologues encoded by spuB and spuI. The SpuB and SpuI proteins exhibit significant identity (42 and 35%, respectively) to another glutamine synthetase homologue encoded by the ycjK gene of unknown function in E. coli. The ycjK gene of E. coli is also divergently transcribed from an operon of five genes: ycjL-ycjC-aldH-ordL-goaG. Interestingly, YcjL and GoaG show 44 and 30% identity, respectively, to SpuA and SpuC at the amino acid sequences. While SpuC was identified in this report as putrescine or pyruvate aminotransferase, the results of sequence comparison suggested an amidotransferase activity to SpuA. The intriguing similarities of these operons in gene arrangement and protein sequences between E. coli and P. aeruginosa suggest that they might be evolutionarily conserved due to a common or related physiological function in polyamine metabolism that has yet to be elucidated.