Functional Genomics Enables Identification of Genes of the Arginine Transaminase Pathway in Pseudomonas aeruginosa

Abstract

Arginine utilization in Pseudomonas aeruginosa with multiple catabolic pathways represents one of the best examples of the metabolic versatility of this organism. To identify genes involved in arginine catabolism, we have employed DNA microarrays to analyze the transcriptional profiles of this organism in response to l-arginine. While most of the genes involved in arginine uptake, regulation, and metabolism have been identified as members of the ArgR (arginine-responsive regulatory protein) regulon in our previous study, they did not include any genes of the arginine dehydrogenase (ADH) pathway. In this study, 18 putative transcriptional units of 38 genes, including the two known genes of the ADH pathway, kauB and gbuA, were found to be inducible by exogenous l-arginine in the absence of ArgR. To identify the missing genes that encode enzymes for the initial steps of the ADH pathway, the potential physiological functions of those candidate genes in arginine utilization were studied by growth phenotype analysis of knockout mutants. Expression of these genes was induced by l-arginine in an aruF mutant strain devoid of a functional arginine succinyltransferase pathway, the major route of arginine utilization. Disruption of dadA, a putative catabolic alanine dehydrogenase-encoding gene, in the aruF mutant produced no growth on l-arginine, suggesting the involvement of l-alanine in arginine catabolism. This hypothesis was further supported by the detection of an l-arginine-inducible arginine:pyruvate transaminase activity in the aruF mutant. Knockout of aruH and aruI, which encode an arginine:pyruvate transaminase and a 2-ketoarginine decarboxylase in an operon, also abolished the ability of the aruF mutant to grow on l-arginine. The results of high-performance liquid chromatography analysis demonstrated consumption of 2-ketoarginine and suggested that generation of 4-guanidinobutyraldehyde occurred in the aruF mutant but not in the aruF aruI mutant. These results led us to propose the arginine transaminase pathway that removes the α-amino group of l-arginine via transamination instead of oxidative deamination by dehydrogenase or oxidase as originally proposed. In the same genetic locus, we also identified a two-component system, AruRS, for the regulation of arginine-responsive induction of the arginine transaminase pathway. This work depicted a wider network of arginine metabolism than we previously recognized.

Arginine utilization in pseudomonads can be mediated by multiple catabolic pathways (Fig. 1). The arginine deiminase pathway encoded by the arc operon (8, 9, 39) provides ATP to support slow growth under anaerobic conditions. The arginine succinyltransferase (AST) pathway encoded by the aru operon (14) and the gdhB gene (17) is the major route of arginine utilization as the carbon and nitrogen source under aerobic conditions. The arginine decarboxylase (ADC) pathway may not contribute to arginine utilization due to the lack of arginine-inducible ADC activity but rather serve to supply putrescine when arginine is abundant (15, 21, 26). Nevertheless, exogenous agmatine or putrescine, indeed, can induce all of the enzymes following its entry point into the ADC pathway and thus be utilized as the sole source of carbon and nitrogen through this pathway (11, 19, 27).

FIG. 1.

Arginine catabolic pathways in P. aeruginosa PAO1. Only relevant intermediates and genes are indicated. ADI, arginine deiminase pathway; AST, AST pathway; ADC, ADC pathway; ATA, ATA pathway; TCA, tricarboxylic acid. Also illustrated are two more routes for 2-KA synthesis: arginine racemase/d-arginine dehydrogenase in P. aeruginosa PAO1 (ADH) and l-arginine oxidase in P. putida P2 (AO). aruF, AST; aruH, ATA; aruI, 2-KA decarboxylase; gabD, 4-aminobutyraldehyde dehydrogenase; gabT, 4-aminobutyrate transaminase; gbuA, 4-guanidinobutyrase; kauB, GUBAL/4-aminobutyraldehyde dehydrogenase.

The arginine dehydrogenase (ADH) pathway was considered the second pathway for arginine utilization in Pseudomonas aeruginosa under aerobic conditions. This is supported by the observation that while an aruF mutant without a functional AST pathway grew poorly on arginine as the sole source of carbon and nitrogen, an aruF gbuA double mutant blocking both the AST and ADH pathways showed no growth on this amino acid (15). Although the kauB and gbuA genes, which encode 4-guanidinobutyraldehyde (GUBAL) dehydrogenase and 4-guanidinobutyrase of the ADH pathway (Fig. 1), have been identified or characterized (13, 16, 25), genetic information about the initial steps of this pathway was completely unavailable. While the presence of an l-arginine oxidase activity has been reported in Pseudomonas putida (2, 38), this enzymatic activity has never been demonstrated in P. aeruginosa. Instead, the presence of a d-arginine-inducible d-arginine dehydrogenase activity suggested an alternative route through this enzyme and an arginine racemase in P. aeruginosa (16). Although removal of the α-amino group from l-arginine could occur through either oxidative deamination or transamination, the results of previous studies did not favor transamination (16).

The ArgR protein, a transcriptional regulator of the AraC/XylS family (6), controls arginine-dependent induction of the arc and aru operons and the gdhB gene in P. aeruginosa (17, 18, 30). ArgR is autoinduced by exogenous arginine from the aotJQMOP-argR operon for arginine and ornithine uptake and regulation (28). The ArgR protein also serves as the repressor of the argF, argG, and carAB genes in arginine biosynthesis and the gdhA and gltBD genes in glutamate biosynthesis (12, 20, 29, 30). More than 30 genes of the ArgR regulon in P. aeruginosa have been identified by transcriptome analyses (20). The kauB and gbuA genes of the ADH pathway were not identified as members of the ArgR regulon. In this study, we conducted further transcriptome analyses to identify genes, including kauB and gbuA, that are induced by exogenous l-arginine in the absence of ArgR. Identification of genes that encode an arginine:pyruvate transaminase (ATase) and a putative 2-ketoarginine (2-KA) decarboxylase led us to propose that the α-amino group of l-arginine is removed via transamination instead of oxidative deamination by dehydrogenase or oxidase as originally proposed in the ADH pathway, and thus we proposed here to call this reformulated route the arginine transaminase (ATA) pathway. In the same genetic locus, we also identified a two-component system for the regulation of arginine-responsive induction of the ATA pathway.

MATERIALS AND METHODS

Strains, plasmids, and growth conditions.

The major strains and plasmids used in this study are listed in Table 1. Luria-Bertani (LB) enriched medium (32) was used for strain construction with the following supplements as required: ampicillin, 50 μg/ml (for Escherichia coli); carbenicillin, 100 μg/ml (for P. aeruginosa); gentamicin, 100 μg/ml; streptomycin, 500 μg/ml; tetracycline, 50 μg/ml. Minimal medium P (MMP) (10) was used for the growth of P. aeruginosa PAO strains with supplements of specific carbon and nitrogen sources at 20 mM as indicated.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Genotype or descriptiona	Source or reference
E. coli strains
DH5α	Fφ 80dlacΔM15 Δ(lacZYA-argF)U169 deoR recA1 endA1 hsdR17(rK mK) supE44 λ thi-1 gyrA96 relA1	Bethesda Research Laboratories
SM10	thi-1 thr leu tonA lacY supE recA::RP4-2-Tc::Mu (Kmr)	7
P. aeruginosa strains
PAO1	Wild type	10
PAO501	argR::Gmr	29
PAO1-Sm	Spontaneous Smr mutant of PAO1	14
PAO4558	aruF	15
PAO4558-Sm	Spontaneous Smr mutant of PAO4558	This study
PAO4566	aruF gbuA	15
PAO5508	aruS::Tcr	This study
PAO5509	aruR::Tcr	This study
PAO5601	PA4975::TcraruF	This study
PAO5602	aruH::TcraruF	This study
PAO5603	aruI::TcraruF	This study
PAO5604	PA4978::TcraruF	This study
PAO5605	PA4979::TcraruF	This study
PAO5606	PA4980::TcraruF	This study
PAO5607	PA4981::TcraruF	This study
PAO5608	aruS::TcraruF	This study
PAO5609	aruR::TcraruF	This study
PAO5610	dadA::TcraruF	This study
PAO5018	aruS::TcrargR::Gmr	This study
PAO5019	aruR::TcrargR::Gmr	This study
Plasmids
pQF50	bla lacZ transcriptional fusion vector	3
pQF52	bla lacZ translational fusion vector	29
pRTP1	Apr Strs conjugation vector	34
pUCP18	E. coli-P. aeruginosa shuttle vector	33
pZY1	gbuA::lacZ translational fusion of pQF52	This study
pZY2	lipA::lacZ translational fusion of pQF52	This study
pZY4	PA3865::lacZ translational fusion of pQF52	This study
pZY5	PA4980::lacZ translational fusion of pQF52	This study
pZY6	dadA::lacZ translational fusion of pQF52	This study
pZY7	kauB::lacZ transcriptional fusion of pQF50	This study
pZY8	PA5313::lacZ transcriptional fusion of pQF50	This study
pZY9	aruI PA4975 clone in pUCP18	This study
pMO012222	Cosmid clone containing PA1415-PA1417 gene cluster in pLA2917; Tcr	31
pMO012227	Cosmid clone containing aruRS-aruH gene cluster in pLA2917; Tcr	31
pMO011841	Cosmid clone containing dadA gene in pLA2917; Tcr	31

^aGm^r, gentamicin resistance; Sm^r, streptomycin resistance; Tc^r, tetracycline resistance; Km^r, kanamycin resistance.

RNA isolation, generation of cDNA probes, and data analysis.

Total RNA was isolated by the hot-phenol method, followed by DNase I treatment and column RNA purification (QIAGEN). cDNA synthesis, fragmentation, labeling, and GeneChip hybridization were performed according to the protocol provided by the manufacturer (Affymetrix) as previously described (20).

The results of two independent experiments were merged for each of the following three growth conditions: wild-type strain PAO1 in glutamate MMP (1E), wild-type strain PAO1 in glutamate MMP plus l-arginine (1ER), and argR mutant strain PAO501 in glutamate MMP plus l-arginine (5ER). The merged data were used for subsequent comparisons and assessed with Microarray Suite 5.0 software (Affymetrix). Absolute expression signal values were normalized for each chip by globally scaling to a target intensity of 500. These data were imported into the GeneSpring 6.1 software (Silicon Genetics) for further analysis based on distinct expression patterns under these three conditions. Transcripts with the absence call (P > 0.04) or with a signal level (5ER) below 400 were eliminated.

Construction of lacZ fusions.

Plasmid pQF52 (29), a broad-host-range lacZ translational fusion vector, was used in the construction of promoter fusions of PA1421 (gbuA), PA2862 (lipA), PA3865, PA4980, and PA5304 (dadA). DNA fragments containing the regulatory regions of interest were amplified by PCR from the genomic DNA of PAO1 with the following synthetic oligonucleotides designed to generate HindIII restriction sites on the forward primers: for PA1421 (gbuA), 5′-TTTAAGCTTCCAGTTCGCCGAGGATGCAGCAG-3′ and 5′-CTGGTGGAGATTCTTGTCCACGGGGTGGCC-3′; for PA2862 (lipA), 5′-CCTAAGCTTCGCCCTGCCCTGCCCACCTCC-3′ and 5′-CTTCTTCATGTTGTTCTCATCTCAGGTTGA-3′; for PA3865, 5′-AGCAAGCTTCCCCCATTACAGACGCGCCTC-3′ and 5′-GGTGTTCAGCGACTTGATCATGGACGACTC-3′; for PA4980, 5′-CCAAGCTTGGGCGACCCCGCCCTGGGAC G-3′ and 5′-GGGGGAAAGGTCAGTCATGCGGCC-3′; for PA5304 (dadA), 5′-AGAAAGCTTGCCGGGCAGTACGTCCAGGTC-3′ and 5′-CAGAACTCGCATTGTCGCCTCCCACGTCGC-3′. The PCR products were purified from a 1% (wt/vol) agarose gel, digested by restriction endonuclease HindIII, and ligated into the HindIII and SmaI sites of the translational fusion vector pQF52. The resulting plasmids contain the entire upstream intergenic sequences of the corresponding genes and the 5′ ends of their coding sequences fused in frame to the eighth codon of the lacZ gene in the vector.

Another broad-host-range lacZ vector, pQF50 (3), was used to produce the transcriptional fusion constructs of kauB and PA5313. Two different DNA fragments were generated by PCR employing two oligonucleotide primers designed to generate HindIII restriction sites at both the 5′ and 3′ ends of the PA5312 (kauB)-PA5313 intergenic region, 5′-CCGAAGCTTGAACTGGCGGTTGGCGGTGAA-3′ and 5′-CACAAGCTTCGCCCTGCCCTGCCCACCTCC-3′. The PCR products were purified from a 1% (wt/vol) agarose gel, digested with restriction endonuclease HindIII, and ligated into the same site of the transcriptional fusion vector pQF50. Following the cloning strategy described above, the resulting two fusion plasmids were designated pZY7 and pZY8, respectively, which represent PA5312 (kauB)::lacZ and PA5313::lacZ fusions. The nucleotide sequences of the resulting constructs were verified by nucleotide sequence determination.

Construction of mutant strains.

The EZ-Tn5 <TET-1> insertion system (Epicentre) was used for generation of knockout mutations. The 11-kb HindIII fragment covering the aruRS-aruI cluster was purified from cosmid pMO012227 (Pseudomonas Genetic Stock Center) and subcloned into the same site of the conjugation vector pRTP1 (34). The resulting 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 mixture was used to transform E. coli DH5α, and transformants were selected on LB plates with tetracycline. The insertion sites of mutant clones were mapped by HindIII 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 spontaneously streptomycin-resistant P. aeruginosa strains PAO1-Sm and PAO4558-Sm by biparental plate mating (7). After incubation at 37°C overnight, transconjugants were selected on LB plates supplemented with tetracycline and streptomycin.

PA4975, PA4976 (aruH), and PA5304 (dadA) knockout mutants were also constructed by a strategy similar to that described above. DNA fragments covering these genes were generated by PCR from the genomic DNA of PAO1 with the following synthetic oligonucleotides designed to generate HindIII restriction sites at both the 5′ and 3′ ends of the PCR products: for PA4975 and PA4976 (aruH), 5′-GTCTAAGCTTGACTGGCCTGGCGCGCGTCG-3′ and 5′-CGCAAGCTTCGGGCAGTCCGGCGTGACCCT-3′; for PA5304 (dadA), 5′-ACGGAAGCTTGGGCCGCCAGCATTTTTTA-3′ and 5′-ACCGAAGCTTTGGATATTTCCCGCTACAGC-3′. The genotypes of all the mutants constructed were verified by Southern blotting.

Complementation of aruH and aruI mutants.

The DNA fragment covering the PA4975 and aruH genes as described above was cloned into the HindIII site of pUCP18. Expression of these two genes in the resulting plasmid, pZY9, was under the control of the lac promoter. Competent cells of aruH and aruI mutants of P. aeruginosa were prepared for transformation by pZY9 by following the standard protocols.

Enzyme assays.

For the measurements of enzyme activities, cell cultures in the logarithmic phase were collected by centrifugation and the cell pellets were washed twice with 50 mM potassium phosphate buffer, pH 7.0, and resuspended in the same buffer. Cells were broken with a French pressure cell at 8,000 lb/in², and soluble cell extracts were prepared after centrifugation at 25,000 × g for 20 min. Protein concentration was determined by the method of Bradford with bovine serum albumin as the standard (1).

For the measurement of ATase activities, a two-step coupled reaction was used as described in the accompanying paper (40). Briefly, the reaction of arginine:pyruvate transamination generates 2-KA and alanine and the concentration of alanine was determined by monitoring the formation of NADH from NAD⁺ in a coupled reaction with l-alanine dehydrogenase. In these reactions, anhydrous hydrazine was used to trap the remaining substrate pyruvate and the reaction product 2-KA as the hydrazones. One unit of enzyme activity was defined as the amount of enzyme that yielded 1 nmol of l-alanine per min under the standard assay conditions. The ATA activity was linear for up to 15 min with less than 100 U of enzymes in the first reaction.

Arginine oxidase was assayed by measuring the release of ammonia from l-arginine in a coupled reaction with l-glutamate dehydrogenase under the conditions given by Miller and Rodwell (22). d-Arginine dehydrogenase was assayed with the addition of artificial electron acceptors, phenazine methosulfate, and iodonitrotetrazolium chloride, as described by Jann et al. One unit of d-arginine dehydrogenase activity was defined as the amount of enzyme that led to the reduction of 1 nmol of iodonitrotetrazolium chloride per min under the standard assay conditions. For the measurements of β-galactosidase activity, o-nitrophenyl-β-d-galactopyranoside was used as the substrate (24).

Detection of enzymatic products of AruI by HPLC.

An assay for 2-KA decarboxylase was modified from a previous report (23) by using high-performance liquid chromatography (HPLC) to analyze the enzymatic reaction. The reaction mixture contained, in a final volume of 0.3 ml, 100 mM potassium phosphate (pH 7.5), 0.5 mM thiamine pyrophosphate (TPP), 1 mM MgSO₄, 1 mM 2-KA, and crude extract (250 μg of protein). The reaction was started by addition of crude extracts, and the mixture was incubated at 37°C for 1 h. In the negative control experiments, heat-denatured crude extracts were used to prepare the reaction mixtures. After incubation, the samples were boiled for 10 min and then filtered with an Ultrafree-0.5 PBCC centrifugal filter unit (molecular mass cutoff, 5 kDa; Millipore). After 10-fold dilution (vol/vol) with deionized water, 30-μl reaction product samples were separated on a Breeze HPLC system (Waters) equipped with a Develosil RP-Aqueous C₃₀ column (4.6 by 250 mm; Phenomenex) at a flow rate of 1 ml/min. The mobile phase was 0.1 M potassium phosphate (pH 2.0), and elution was monitored by UV detection at 205 nm. To serve as standards, authentic TPP was purchased from the vendor (Sigma) while 2-KA (16) and GUBAL (36) were synthesized by following the published protocols.

RESULTS

Identification of genes induced by l-arginine in the absence of ArgR.

To better understand arginine metabolism in P. aeruginosa, we have employed DNA microarrays to analyze the transcriptional profiles of this organism in response to l-arginine. In a previous report (20), more than 30 genes were identified as members of the ArgR regulon, in which gene expression is either repressed or activated by exogenous l-arginine in the presence of a functional ArgR. These two groups of genes exhibited distinct expression profiles, as represented by argF for the repression group and aruF for the activation group (last two lines of Table 2). In addition, ArgR binding sites have been demonstrated in these genes. To our surprise, genes of the proposed ADC and ADH pathways were not identified in the ArgR regulon. In this report, we describe a third type of expression profile in which genes were highly induced by l-arginine in argR mutant strain PAO501 but to a lesser extent in wild-type strain PAO1.

TABLE 2.

Microarray analysis of genes induced by l-arginine in the absence of ArgR in P. aeruginosa

Gene no.a	Name	Protein description	Absolute signal value
			PAO1 (E)b	PAO1 (ER)c	PAO501 (ER)
*PA0265	gabD	Succinate-semialdehyde dehydrogenase	332	707	4,227
*PA0266	gabT	4-Aminobutyrate aminotransferase	712	1,192	5,606
*PA0296	spuI	Glutamine synthetase homologue	1,018	1,301	4,505
*PA0297	spuA	Glutamine amidotransferase	186	687	2,866
*PA0298	spuB	Glutamine synthetase homologue	514	1,166	3,872
*PA0299	spuC	Putrescine aminotransferase	869	1,902	5,620
PA1415		Unknown	118	131	946
PA1416		FADd-linked oxidase	49	11	377
PA1417		Decarboxylase	68	149	531
PA1418		Transport of small molecules	152	128	1,748
PA1419		Transport of small molecules	43	72	1,015
PA1420		Unknown	47	151	1,357
PA1421	gbuA	Guanidinobutyrase	66	46	954
PA2310		Unknown	68	114	579
PA2311		Unknown	525	1,379	8,153
PA2312		Transcriptional regulator	150	339	1,424
PA2313		Unknown	70	87	190
PA2862	lipA	Lactonizing lipase precursor	760	451	3,871
PA2863	lipH	Lipase modulator protein	127	77	606
PA3690		Metal-transporting type P ATPase	162	426	8,550
PA3862		Unknown	84	245	772
PA3863		Unknown	56	265	602
PA3864		Unknown	257	418	953
PA3865		Amino acid binding protein	174	498	3,558
PA4975		NAD(P)H quinone oxidoreductase	98	225	2,635
PA4976	aruH	ATase	61	192	2,639
PA4977	aruI	2-Ketoarginine decarboxylase	59	149	3,977
PA4978		Pimeloyl-CoA synthetase	69	177	2,998
PA4979		Acyl-CoA dehydrogenase	77	17	3,435
PA4980		Enoyl-CoA hydratase/isomerase	97	687	7,154
PA4981		Transport of small molecules	57	573	12,235
*PA5302	dadX	Alanine racemase	26	85	408
*PA5303		Unknown	81	162	571
*PA5304	dadA	d-Alanine dehydrogenase	88	113	643
*PA5309		Oxidoreductase	413	651	1,878
*PA5312	kauB	4-Guanidinobutyraldehyde dehydrogenase	780	1,111	4,774
*PA5313		Pyridoxal-dependent aminotransferase	150	386	1,449
*PA5314		Unknown	123	397	1,646
PA3537	argF	Ornithine carbamoyltransferase	874	181	1,444
PA0896	aruF	AST	174	4,054	59

^aGene numbers are from the Pseudomonas Genome Project (www.pseudomonas.com). Genes also induced by agmatine, putrescine, and spermidine are marked with asterisks. The expression profiles of argF and aruF at the end of this list represent ArgR-dependent repression and activation, respectively.

^bE, cells were grown in MMP-glutamate.

^cER, cells were grown in MMP-glutamate supplemented with 20 mM l-arginine.

^dFAD, flavin adenine dinucleotide.

Table 2 summarizes 38 genes in 18 putative transcriptional units falling into this group. While many of these genes have not been characterized, the following three observations are interesting. Firstly, lipAH and the PA4978-to-PA4980 genes are related to lipid or fatty acid metabolism. Secondly, the gabDT and spuIABC genes have been reported to be inducible by agmatine or putrescine, and enzymes encoded by gabDT and spuC are part of the ADC pathway. Although not shown here, these genes, as well as the dadX-PA5303-dadA, PA5309, and kauB-PA5313-PA5314 genes, were indeed induced to a high level by agmatine or putrescine in wild-type strain PAO1 in a DNA microarray analysis (data not shown). The kauB gene encodes a bifunctional dehydrogenase for 4-aminobutyraldehyde of the ADC pathway and GUBAL of the ADH pathway. Thirdly, the gbuA gene, which encodes 4-guanidinobutyrase of the ADH pathway, was also identified in this list. These results suggested that genes of the convergent ADC and ADH pathways (Fig. 1) might be regulated by a common mechanism and that the genes for the first two steps of the ADH pathway yet to be identified might also be on the same gene list.

Validation of microarray data by promoter-lacZ fusions.

Genes in Table 2 could be categorized into 18 putative transcriptional clusters. Promoters from seven of these clusters were selected for further analysis by lacZ fusions: gbuA, lipA, PA3865, PA4980, dadA, kauB, and PA5313. These promoter-lacZ fusions were constructed as described in Materials and Methods, and the effects of ArgR and exogenous l-arginine on the expression of these promoters were analyzed by measurements of β-galactosidase activities in PAO1 or the argR mutant harboring the recombinant plasmids. As shown in Table 3, all of these fusions exhibited 2- to 15-fold higher l-arginine-inducible promoter activities in the argR mutant than in wild-type PAO1. Overall, these results correlated well with microarray data on the induction pattern and supported the presence of an additional arginine-responsive regulatory mechanism.

TABLE 3.

Verification of microarray data by promoter-lacZ fusions

Gene no. (name)a and host strain (genotype)	Sp act (nmol/min/mg)b		Fold changec
	Glu	Glu + Arg
PA1421 (gbuA)
PAO1 (wild type)	9	18	2.0
PAO501 (argR)	7	216	30.9
PAO4558 (aruF)	11	177	16.1
PAO5608 (aruF aruS)	9	26	2.9
PAO5609 (aruF, aruR)	12	37	3.1
PA2862 (lipA)
PAO1 (wild type)	107	68	0.6
PAO501 (argR)	251	963	3.8
PA3865
PAO1 (wild type)	823	208	0.3
PAO501 (argR)	664	2,156	3.2
PA4980
PAO1 (wild type)	1,241	7,216	5.8
PAO501 (argR)	1,643	29,191	17.8
PAO5508 (aruS)	1,069	998	0.9
PAO5509 (aruR)	251	355	1.4
PAO5018 (argR aruS)	1,158	2,217	1.9
PAO5019 (argR aruR)	942	824	0.9
PAO4558 (aruF)	145	22,899	157.9
PA5608 (aruF aruS)	177	1,547	8.7
PAO5609 (aruF aruR)	101	635	6.3
PA5304 (dadA)
PAO1 (wild type)	30	36	1.2
PAO501 (argR)	47	121	2.6
PA5312 (kauB)
PAO1 (wild type)	111	136	1.2
PAO501 (argR)	116	266	2.3
PAO4558 (aruF)	102	234	2.3
PAO5608 (aruF aruS)	9	21	2.3
PAO5609 (aruF aruR)	4	12	3
PA5313
PAO1 (wild type)	254	335	1.3
PAO501 (argR)	178	677	3.8

^aGene numbers and names are from the Pseudomonas Genome Project (www.pseudomonas.com).

^bSpecific activities are averages of three measurements for each growth condition, with standard errors of less than 5%. Glu, cells were grown in MMP-glutamate; Glu+Arg, cells were grown in MMP-glutamate supplemented with 20 mM l-arginine.

^cFold changes in gene expression are indicated by the ratio of cells grown in glutamate MMP in the presence of arginine to cells grown in MMP-glutamate in the absence of arginine.

The gbuA locus.

The gbuA gene (PA1421) encodes a guanidinobutyrase (25) for the conversion of 4-guanidinobutyrate to 4-aminobutyrate in the proposed ADH pathway (Fig. 1). It is the first gene of a putative seven-gene operon (PA1421 to PA1415), as suggested by the correlated expression profiles of these genes from microarray data (Table 2). Among these genes, PA1416 and PA1417 were annotated to encode a flavin adenine dinucleotide-linked oxidase and a decarboxylase, respectively, which might catalyze the first two reactions of the ADH pathway. This possibility was explored by growth phenotype analysis of knockout mutants. As described in Materials and Methods, the inactivated PA1416 and PA1417 genes, carrying a tetracycline resistance cassette, were introduced by conjugation and homologous recombination into an aruF mutant devoid of the first enzyme of the AST pathway (Table 1). Growth of the aruF mutant on l-arginine as the sole source of carbon and nitrogen was retarded, as evidenced by a longer generation time than that of wild-type PAO1 (220 versus 75 min). Disruption of PA1416 or PA1417 in the aruF mutant, however, had no significant effect on l-arginine utilization, while, in contrast, disruption of gbuA abolished growth on l-arginine. These results did not support the hypothesis that PA1416 and PA1417 participate in the ADH pathway.

The dadAX locus.

While the presence of an l-arginine oxidase activity in P. putida was reported, this enzymatic activity has never been demonstrated in P. aeruginosa. Instead, a specific d-arginine-inducible d-arginine dehydrogenase was reported in P. aeruginosa PAO1 (16), leading Jann et al. to propose an alternative route for the ADH pathway involving this enzyme and an arginine racemase (Fig. 1). The current genome annotations (www.pseudomonas.com) designated PA5304 and PA5302 dadA and dadX (Table 2) on the basis of the amino acid sequence similarity of the encoded proteins to the catabolic d-alanine dehydrogenase and the alanine racemase of E. coli, respectively. Since these two genes were induced by l-arginine, experiments were conducted to test the hypothesis that they might encode the d-arginine dehydrogenase and arginine racemase.

The effects of l-alanine and arginine on dadAX expression were studied by measurement of β-galactosidase activity from wild-type strain PAO1 harboring a dadA::lacZ fusion plasmid, pZY6. It was found that the promoter activity of dadA on plasmid pZY6 was induced by l-alanine (950 nmol/min/mg) rather than by d-arginine (52 nmol/min/mg) or by l-arginine (36 nmol/min/mg). l-Arginine, however, did exert a moderate induction effect on the dadA promoter in the argR (121 nmol/min/mg) or aruF (212 nmol/min/mg) mutant.

To investigate the role of DadA in arginine catabolism, dadA knockout mutants of wild-type strain PAO1 or its aruF mutant were constructed. Disruption of dadA only resulted in a negligible decrease in d-arginine dehydrogenase activity (data not shown). However, the dadA aruF double mutant did not show growth, while the dadA mutant grew well, on l-arginine as the sole source of carbon and nitrogen.

These results would be expected if, according to the gene annotations, the dadAX genes encode a dehydrogenase and a racemase for alanine, but not for arginine. A moderate induction effect by l-arginine to dadAX in an aruF mutant and the abolishment of arginine utilization in a dadA aruF mutant would suggest the possibility that the intracellular alanine pool was increased when l-arginine was used to supplement the aruF mutant. Alanine could be a side product of arginine utilization via an alternative pathway when the AST pathway is not available. However, none of the proposed reactions in the ADH pathway would make alanine. This led us to propose that an ATase, instead of arginine dehydrogenase or oxidase, catalyzes the removal of the α-amino group from l-arginine and makes alanine with pyruvate as the amino group acceptor.

The aruIH locus.

As shown in Table 2, the contiguous PA4981-to-PA4975 genes were coordinately induced to a high level by l-arginine in the argR mutant. Among these genes, PA4976 and PA4977 encode a putative class I transaminase and thiamine pyrophosphate-dependent decarboxylase, respectively. They could serve to catalyze the first two steps of our newly proposed ATA pathway. Mutants in which these seven genes were knocked out were constructed in an aruF mutant as described in Materials and Methods, and the growth phenotype of these mutants on l-arginine was determined. The results show that disruption of PA4976 or PA4977 in the aruF mutant abolished growth on l-arginine as the sole source of carbon and nitrogen, and it showed no effect on glutamate utilization (data not shown). The effect of PA4976 and PA4977 on arginine utilization was only observed in the aruF mutant but not in wild-type strain PAO1 (data not shown). PA4976 and PA4977 were designated aruH and aruI (arginine utilization), in accordance with their possible physiological functions.

Other genes in this locus encode proteins for energy metabolism (PA4975), lipid metabolism (PA4980 to PA4978), and amino acid transport (PA4981), as suggested by genome annotations (Table 2). Except for PA4979, which encodes a putative acyl coenzyme A (CoA) dehydrogenase, the mutants lacking these genes in this locus grew on l-arginine identically to the parent strain (data not shown). It was noted that the lipAH operon, which encodes a lipase and a lipase modulator, was also induced by l-arginine in the argR mutant (Tables 2 and 3). These results suggested an interesting link between l-arginine and lipid metabolism with unknown physiological significance.

Expression of the aruIH genes could be from an l-arginine-inducible promoter immediately upstream of PA4980, as demonstrated in Table 3. With no apparent rho-independent terminator structure in the nucleotide sequence of this region, it is likely that this detected promoter of PA4980 could serve to make a polycistronic transcript for PA4980 to PA4975. In contrast, a rho-independent terminator was identified immediately after the 3′ end of the PA4981 coding sequence in the PA4981-PA4980 intergenic region. This suggested the presence of another promoter for PA4981 that has yet to be identified.

The transposon used in the construction of these mutants does not have any transcriptional terminator structures in the flanking regions. Insertion of such a transposon was not expected to have a polar effect on the expression of downstream genes. Complementation by pZY9 carrying aruH only restored growth on l-arginine in the aruF aruH mutant and not in the aruF aruI mutant. These results support the notions that transposon insertion at aruI does not exert a polar effect on the downstream aruH gene and that both aruH and aruI are related to l-arginine utilization.

AruH is an l-ATase.

Enzymatic measurements were conducted as described in Materials and Methods to demonstrate the proposed l-ATase activity of AruH. As shown in Table 4, an l-arginine-inducible ATase activity was detected in the aruF mutant when pyruvate was used as the amino group acceptor in the reaction. This ATase activity was abolished when an aruH knockout was introduced into the strain. While the ATase activity was induced at least 76-fold by l-arginine in the aruF mutant, no arginine induction of this enzyme could be observed in wild-type strain PAO1. These results supported the proposed biochemical function of AruH and its potential role in the ATA pathway as the second arginine catabolic pathway under aerobic conditions. A detailed characterization of AruH is given in the accompanying paper (40).

TABLE 4.

Measurements of arginine transaminase activities in P. aeruginosa PAO1 and its mutant strains

Strain (genotype)	Sp acta (nmol/min/mg)
	Glu	Glu + Arg
PAO1 (wild type)	<0.5	<0.5
PAO4558 (aruF)	<0.5	38
PA5602 (aruH aruF)	<0.5	<0.5
PA5608 (aruS aruF)	<0.5	3
PA5609 (aruR aruF)	<0.5	<0.5

^aCells were cultured to mid-log phase (optical density at 600 nm = 0.5) in MMP-glutamate medium with or without l-arginine. Cell extract preparation and enzyme assays are described in Materials and Methods.

AruI encodes a 2-KA decarboxylase.

On the basis of the results of sequence and growth phenotype analyses, we proposed that AruI encodes a 2-KA decarboxylase catalyzing the second step of the ATA pathway. To test this hypothesis, we set up the proposed decarboxylation reaction as described in Materials and Methods with crude extracts of the aruF and aruF aruI mutants grown in minimal medium with glutamate and arginine as the carbon and nitrogen sources, and the reaction components were analyzed by HPLC. As demonstrated in Fig. 2, with the crude extract of the aruF mutant, consumption of 2-KA was accompanied by generation of a new compound, presumably GUBAL; more than 85% of the 2-KA was converted into GUBAL in the reaction. In contrast, consumption of only about 15% of the substrate was observed under the same assay conditions with cell extract of the aruF aruI mutant. These results support our hypothesis that AruI encodes a 2-KA decarboxylase.

FIG. 2.

HPLC analysis of the AruI reaction products. Chromatograms of components of the reaction mixture at 205 nm with either heat-inactivated crude extracts (aruF mutant, broken gray line; aruF aruI mutant, solid gray line) or active crude extracts (aruF mutant, solid black line; aruF aruI mutant, broken black line) were recorded as described in Materials and Methods. Signal peaks for TPP, 2-KA, and GUBAL were individually identified by comparison to the retention times of the authentic reference compounds. In this system, the retention times of TPP, 2-KA, and GUBAL were 4.0, 5.9, and 7.5 min, respectively. Traces are offset vertically, and A₂₀₅ values are in arbitrary units.

Regulation of the ATA pathway by the AruRS two-component system.

The observation that the genes listed in Table 2 were induced by exogenous l-arginine in the argR mutant strongly suggested the presence of another arginine-responsive regulatory system. Genes PA4982 and PA4983, which encode a sensor and a response regulator of a putative two-component system, were located immediately upstream of the aruIH locus. To investigate the potential role of this two-component system in the regulation of genes in the aruIH locus, PA4982 and PA4983 knockout mutants were constructed and expression of the PA4980::lacZ fusion in the resulting mutants was analyzed by measurements of β-galactosidase activity. As shown in Table 3, while l-arginine exerted 5.8- and 17.8-fold induction of the PA4980 promoter in the wild-type strain and the argR mutant, respectively, this induction effect was abolished when either PA4983 or PA4982 was disrupted, regardless of ArgR. We therefore designated these two genes aruS (PA4982) and aruR (PA4983).

The effect of aruRS on arginine utilization was also studied by growth phenotype analysis. It was found that disruption of either aruS or aruR in the aruF mutant prevented growth on l-arginine. Furthermore, the ATA activities showed no induction by l-arginine in the aruR mutant and marginal induction in the aruS mutant (Table 4). Similarly, the levels of kauB and gbuA promoter activities in these two mutants were significantly reduced in the presence of l-arginine (Table 3). These results demonstrated the essential role of AruRS in the regulation of ATA synthesis, hence, the ATA pathway, in response to exogenous l-arginine.

DISCUSSION

Three lines of evidence that arose from this study led us to conclude that the ATA pathway, instead of the ADH pathway, is the second pathway for l-arginine utilization in P. aeruginosa under aerobic conditions. Firstly, an l-arginine-inducible ATA was encoded by the newly identified aruH gene. Secondly, disruption of aruH in a strain devoid of a functional AST pathway abolished the ability of the resulting strain to utilize l-arginine as the sole source of carbon and nitrogen. Thirdly, AruH utilizes pyruvate as the amino group acceptor, and hence makes alanine, in the transamination reaction. Recycling of pyruvate through the catabolic alanine dehydrogenase DadA is essential for arginine utilization in the ATA pathway, as supported by the observation that the aruF dadA mutant cannot grow on l-arginine as the sole carbon source. It is likely that in this mutant, accumulation of alanine without recycling of pyruvate would drain the intracellular pool of pyruvate and thus the carbon source provided by l-arginine through the ATA pathway.

While no l-arginine dehydrogenase or oxidase activity has ever been demonstrated in P. aeruginosa, the results of a previous study also did not favor the presence of an ATA. It was based on the observation that arginine auxotrophs were not complemented by 2-KA in the presence of l-alanine, assuming that l-arginine would be generated by these two substrates in the reverse reaction of arginine transamination. It was very likely that the proposed reverse reaction never happened, for the following reasons. The ATA may not be expressed in the arginine auxotrophs in the absence of arginine. Even if the enzyme were expressed, 2-KA and l-alanine might not be available at a concentration high enough to drive the reverse reaction. Alanine could induce the expression of DadX and DadA and thus be converted into pyruvate. Also, it has been reported in the same study that 2-KA induced GbuA, suggesting catabolism of 2-KA into 4-guanidinobutyrate (25). In addition, as reported in the accompanying paper (40), no l-arginine synthesis can be detected by HPLC in the reverse reaction with purified AruH under the assay conditions used.

In comparison, the AST pathway is more efficient than the ATA pathway in many aspects. First, it is more efficient in terms of l-arginine affinity. The K_m of l-arginine is 0.5 mM to AST AruF (37) and 14 mM to ATA AruH (40). Second, it is more efficient in terms of gene organization and regulation. The genes that encode enzymes of the AST pathway are organized into the aru operon and the gdhB gene, and they are controlled by a single regulator, ArgR. Genes of the ATA pathway are most likely grouped into several regulatory modules in response to different intermediate compounds that were also involved in polyamine catabolism (9). We found in this study that the AruRS two-component regulatory system is essential for aruHI induction by l-arginine; however, whether l-arginine per se is the induction signal of AruRS remained to be elucidated. Without a functional ArgR protein, it is known that the AST pathway and l-arginine uptake would be severely hampered. It has been reported that blocking of the AST pathway alone is sufficient to render enzymes of the ATA pathway inducible by l-arginine (15, 16). Perhaps N²-succinyl-l-arginine or its derivatives of the AST pathway might serve as an antagonistic signal to block the AruRS system, and subsequently expression of the less efficient ATA pathway, when the AST pathway is available.

The kauB and gbuA genes of the ATA pathway could be controlled by a regulatory module different from ArgR and AruRS. The gbuA gene has been reported to be regulated by GbuR, a transcriptional activator of the LysR family, in the presence of 4-guanidinobutyrate (26). It was still not clear how kauB was regulated, as this gene encodes a bifunctional enzyme in the ADC and ATA pathways (16). As mentioned earlier, putrescine and spermidine exerted a stronger induction effect on the expression of kauB than l-arginine (data not shown). It was possible that a common regulatory mechanism could be shared by kauB and other genes involved in polyamine metabolism.

The gene organization of the aruHI locus in P. aeruginosa PAO1 was found to be highly conserved in P. putida KT2440 and Pseudomonas fluorescens Pf-1, as shown in Fig. 3. Given the proven functions of AruIH in the ATA pathway of P. aeruginosa and the conserved gene organization, it is very likely that the ATA pathway also contributes to l-arginine utilization in P. fluorescens and P. putida. Although an l-arginine oxidase activity has been reported in P. putida P2 (2, 38), we were unable to detect this enzyme activity in P. putida KT2440 (data not shown). However, it was noted that the reading frame of P. putida AruI was interrupted by a termination codon in its coding sequence. Furthermore, PP3722, which encodes a putative amino acid racemase, was located between aruI and aruH in this organism. More work is required to answer the question of 2-KA synthesis in P. putida.

FIG. 3.

Conserved gene organization of the aruRS-aruIH locus in pseudomonads. Functions of the aru genes in this locus of P. aeruginosa are reported in this study, and the same gene designations were given to their counterparts in P. putida and P. fluorescens on the basis of sequence similarities. Other genes in the locus were numbered according to the individual genome annotation projects. Regulatory genes are marked with filled arrows; the differences of two genes in P. aeruginosa and three genes in the other two species were most likely due to reading frameshifts caused by point mutations in the nucleotide sequences. Additional gene 3722 of P. putida, which encodes a putative amino acid racemase, was unique to this organism. Also presented are locations of transposons carrying a tetracycline resistance cassette (Tet) in the mutants of P. aeruginosa as described in the text. Open and filled triangles represent Tet in the same and reverse orientations of the inserted genes, respectively.

As evidenced in the accompanying paper (40), we concluded that AruH is not the homologue of E. coli AspC (4), an aspartate transaminase for aspartate biosynthesis, as was initially described in the Pseudomonas genome annotation. With E. coli AspC as the template to search against the PAO1 genome, the results of a BLAST search showed that PA3139 and PA0870 are the two most prominent candidates. However, PA0870 was annotated as PhhC, which has been reported as an essential transaminase for tyrosine and phenylalanine catabolism (35). Therefore, it is likely that PA3139 is the biosynthetic aspartate transaminase. Similarly, the AruI protein was demonstrated by HPLC analysis in this study to catalyze the decarboxylation reaction of 2-KA in the ATA pathway and is not likely the homologue of E. coli acetolactate synthase IlvB (5), as suggested by genome annotation. Instead, PA4696 is the most likely candidate on the basis of its 77% sequence similarity to E. coli IlvB. Moreover, it was noted that AruI exhibited 57% similarity to another putative TPP-dependent decarboxylase encoded by PA1417, which was also identified in this study as a gene inducible by l-arginine in the absence of ArgR. While the physiological function of PA1417 remained unknown, it was intriguing to speculate that it might be responsible for the observed residual 2-KA decarboxylase activity in the aruF aruI mutant (Fig. 2).

In this study, we have ruled out the significance of d-arginine-inducible d-arginine dehydrogenase (16) in l-arginine utilization; however, this yet-to-be-identified enzyme could still provide an alternative route for d-arginine utilization. We have observed that d-arginine utilization was partially retarded in the aruF mutant devoid of the AST pathway, and blocking of the ATA pathway in this mutant by either aruH or aruI did not result in further inhibition of growth on d-arginine (data not shown). To be utilized by the AST pathway, d-arginine is supposed to be converted into l-arginine by a racemase, as d-arginine is a competitive inhibitor of AST catalyzing the first reaction of the AST pathway (37). The physiological functions of arginine racemase and d-arginine dehydrogenase in arginine metabolism warrant further investigation.

In conclusion, the complexity of arginine metabolism in P. aeruginosa was fully illustrated in our previous report (20) and in this study employing DNA microarray analyses. While some of these discoveries have apparent linkages to arginine metabolism, many of the newly identified avenues remained interesting topics to explore.