γ-Glutamylputrescine Synthetase in the Putrescine Utilization Pathway of Escherichia coli K-12*

Glutamate-putrescine ligase (γ-glutamylputrescine synthetase, PuuA, EC 6.3.1.11) catalyzes the γ-glutamylation of putrescine, the first step in a novel putrescine utilization pathway involving γ-glutamylated intermediates, the Puu pathway, in Escherichia coli. In this report, the character and physiological importance of PuuA are described. Purified non-tagged PuuA catalyzed the ATP-dependent γ-glutamylation of putrescine. The Km values for glutamate, ATP, and putrescine are 2.07, 2.35, and 44.6 mm, respectively. There are two putrescine utilization pathways in E. coli: the Puu pathway and the pathway without γ-glutamylation. Gene deletion experiments of puuA, however, indicated that the Puu pathway was more critical in utilizing putrescine as a sole carbon or nitrogen source. The transcription of puuA was induced by putrescine and in a puuR-deleted strain. The amino acid sequences of PuuA and glutamine synthetase (GS) show high similarity. The molecular weights of the monomers of the two enzymes are quite similar, and PuuA exists as a dodecamer as does GS. Moreover the two amino acid residues of E. coli GS that are important for the metal-catalyzed oxidation of the enzyme molecule involved in protein turnover are conserved in PuuA, and it was experimentally shown that the corresponding amino acid residues in PuuA were involved in the metal-catalyzed oxidation similarly to GS. It is suggested that the intracellular concentration of putrescine is optimized by PuuA transcriptionally and posttranslationally and that excess putrescine is converted to a nutrient source by the Puu pathway.

␥-Glutamyl linkage is an amide linkage between the ␥-position carboxyl group of glutamate and an amino group of various compounds. Compounds that have a ␥-glutamyl linkage are called ␥-glutamyl compounds, which are widely found in both prokaryotic and eukaryotic cells. For example, the peptidoglycan of Escherichia coli has a ␥-glutamyl linkage between D-glu-tamate and meso-2,6-diaminopimelic acid (1), the virulence of Bacillus anthracis is dependent on a capsule made of poly-␥glutamic acid (2), theanine (␥-glutamylethylamide) is a major "umami" component of Japanese green tea (3), and glutathione (␥-glutamylcysteinylglycine) is a very important antioxidant (4) in living cells.
Putrescine is one of the polyamines that are found in a wide range of organisms from bacteria to plants and animals; are critical for cell proliferation, differentiation, and transformation; and are involved in DNA, RNA, and protein synthesis as well as in stabilizing membrane and cytoskeletal structures (5,6). An increased concentration of polyamine is observed in cancer cells (7), there is a significantly elevated concentration (more than millimolar) of putrescine in plants under various stress conditions, and its concentration is very high (estimated to be over 30 mM) in E. coli cells (8).
PuuA was initially annotated as the putative glutamine synthetase (GS) by computer analysis based on the amino acid sequence. GS (14) catalyzes the condensing reaction of glutamate and ammonia with the aid of ATP (Scheme 2).
Schemes 1 and 2 are very similar because both reactions are ␥-glutamylation, the condensing reaction yielding an amide linkage between the ␥-carboxyl group of glutamate and the amino group of putrescine or ammonia using ATP.
In this study, we report that PuuA is an important enzyme that catalyzes the first step of the Puu pathway and is regulated by a complicated regulation system. We also discuss the similarities between GS and PuuA.

Strain and Plasmid Construction
The strains, plasmids, and oligonucleotides used in this study are listed in Table 1. The strains were derivatives of E. coli K-12. Strain SH639 (15) has a deletion of the ggt gene. P1 transduction, DNA manipulation, and transformation were performed by the standard methods (16,17). A DNA fragment of Kohara clone 257 (18) containing the puu gene cluster was used to make plasmids. The cloned regions of DNA on plasmids and the deleted regions of DNA are summarized in Fig. 1. In strain SK212, disruption of the puuR gene was carried out as follows. The 4.5-kb EcoRV-EcoRV fragment, including puuADR, was isolated from Kohara phage 257 and cloned between the XmnI site and the blunt-ended ScaI site of pACYC184 to obtain pKHG3. The 220-bp region between EcoRI and XmnI sites was replaced with the 1.2-kb HincII kanamycin resistance cassette of pUC4K. The plasmid was linearized using SalI. Homologous recombination was performed using this linearized plasmid to delete residues 65-137 of a putative 185-residue PuuR protein (Fig. 1). In FIGURE 1. The puu gene cluster and its deleted or subcloned regions in this study. The subcloned and deleted regions are indicated above or below the gene map, respectively, with respective plasmids and strain names. The puuP promoter (puuP p ) is a putative promoter. The positions of other promoters were described previously (27). It was reported previously (27) that there were FNR and ArcA (transcriptional regulators) sites in the intergenic region between puuA and puuD. FIGURE 2. Schematic presentation of putrescine degradation pathway of E. coli. PuuA, PuuB, PuuC, and PuuD comprise the Puu pathway to degrade putrescine to GABA via ␥-glutamylated intermediates. YgjG and YdcW comprise the traditional pathway to degrade putrescine to GABA without ␥-glutamylation. It was reported that PuuR (10), ArcA, and FNR (27) are transcriptional repressors of puuA and puuD. ␥-Glu-␥-aminobutyraldehyde, ␥-glutamyl-␥-aminobutyraldehyde; ␣-KG, ␣-ketoglutarate.

5Ј-CGAAGTTGTCGTTTCAGAGCGATACC-3Ј
␥-Glutamylputrescine Synthetase of E. coli K-12 SK310, the puuA gene was disrupted according to the method of Datsenko and Wanner (19) using oligonucleotides puuA-1 and puuA-2. This strain deleted the predicted ATP binding motif of PuuA ( Figs. 1 and 8). In SO58, disruptions of puuADR and puuCBE were performed by the modified method of Datsenko and Wanner (19) as follows. To disrupt puuCBE, the DNA fragment of Kohara clone 257 (18), containing the puuCBE gene region, was digested with NarI and HincII and cloned into pUC19 digested with NarI and HincII to obtain pSK169. The FRT-kan ϩ -FRT fragment was amplified by PCR using oligonucleotides pKD13-1 and pKD13-4 as primers and plasmid pKD13 as a template with KOD-plus DNA polymerase (Toyobo, Osaka, Japan). The PCR product was ligated with the 3.9-kb DraIII (blunt-ended with a Blunting kit (Takara, Kyoto, Japan)) and EcoRV fragment of pSK169, which deleted all of the puuB and most of the puuC and puuE genes (Fig. 2). The obtained plasmid was cleaved with HindIII and AatII, and the 3.0-kb fragment was used to transform TK251 by electroporation at 30°C. Kanamycin-resistant transformant SK231 was obtained. To disrupt puuADR, the FRT-cat ϩ -FRT fragment was amplified by PCR using oligonucleotides pKD13-1 and pKD3-2 as primers and plasmid pKD3 as a template with KODplus DNA polymerase. The PCR product was ligated with the 4.7-kb XmnI and HpaI fragment of pKHG3, which deleted all of the puuAD and most of the puuR genes. The obtained plasmid was cleaved with SacII and NruI, and the 2.8-kb fragment was used to transform SK231 by electroporation. Chloramphenicol-resistant transformant SO56 was obtained. Then SO58 was made by P1 transduction of the ⌬puuADR::(FRT-cat ϩ -FRT) ⌬puuCBE::(FRT-kan ϩ -FRT) alleles from SO56 to SH639. To construct pKHG8, a 2.0-kbp SmaI and HpaI fragment of pKHG3 was ligated with pBR322 digested with EcoRV and NruI. pSO105, which carries puuA H282N , was constructed with the QuikChange technique (Stratagene) using oligonucleotides puuA-3 and puuA-4 as primers and pSO97 as a template but using KOD-plus DNA polymerase (Toyobo). pSK324, which carries puuA R357Q , was similarly constructed using oligonucleotides puuA-5 and puuA-6.

Media and Growth of Bacteria
In all experiments, except when studying the influence of the overexpression of native PuuA on protein purification, strains were grown at 37°C with reciprocal shaking at 140 rpm in 60 ml of medium in a 300-ml Erlenmeyer flask. M9-tryptone (M9 minimal medium (16) except that 1% Bacto tryptone was used instead of 0.2% glucose) was used in the analysis of the intracellular amino acid and polyamine profiles. In the study to determine whether E. coli can grow using putrescine as the sole source of nitrogen, W-Glc-Put medium (W salts minimal medium (20) containing 0.4% glucose as the sole carbon source and 0.2% putrescine as the sole nitrogen source) was used. To determine whether E. coli can grow using putrescine as the sole source of carbon, M9-Put-AS medium (M9 minimal medium (16) except that 0.4% putrescine and 0.4% ammonium sulfate were used instead of 0.2% glucose and 0.2% ammonium chloride, respectively) was used. In growth experiments in carbonand nitrogen-limited medium, strains were precultured on an LB plate at 37°C, streaked on a nutrient-limited plate, and incu-bated at 20°C. To overexpress mutagenized PuuA, strains were grown in 60 ml of LB broth containing 100 g/ml ampicillin at 37°C with shaking at 140 rpm in a 300-ml Erlenmeyer flask. To overexpress native PuuA for protein purification, strains were grown in 200 ml of LB broth containing 100 g/ml ampicillin at 37°C with shaking at 140 rpm in a 1-liter Erlenmeyer flask. ␥-Glu-Put was enzymatically synthesized and purified as described previously (9).

Analysis of Amino Acids and Polyamines
Amino acids and polyamines in the samples were measured using an HPLC system (model LC-20AD; Shimadzu, Kyoto, Japan) equipped with a Shim-pack Amino-Na column (Shimadzu) with gradient elution at 60°C at a flow rate of 0.6 ml/min or using an HPLC system (model LC-20AD; Shimadzu) equipped with a TSKgel Polyaminepak (Tosoh, Tokyo, Japan) with gradient elution at 40°C at a flow rate of 0.4 ml/min. The running program for the HPLC system using the Shim-pack Amino-Na column was described previously (9). In the analysis using TSKgel Polyaminepak, two buffers, buffer A (18.6 mM trisodium citrate dehydrate, 400 mM sodium chloride, 20.8 mM HCl, 4% methanol, 0.0016% octanoic acid, 0.0156% Brij-35) and buffer B (93 mM trisodium citrate dehydrate, 2 M sodium chloride, 104 mM HCl, 20% methanol, 0.008% octanoic acid, 0.078% Brij-35), were used. The column was originally equilibrated with buffer A. After the sample was injected, the concentration of buffer B was kept at 0% for 5 min. Then it was increased to 100% and maintained until 30 min. Then the column was regenerated by 0.2 N NaOH from 30 to 35 min. After the regeneration step, the column was equilibrated again by buffer A from 35 to 45 min. o-Phthalaldehyde was used as the detection reagent (21) as described previously (3), and fluorescence was detected with a fluorescence detector (model RF-10AXL; Shimadzu) at an absorbance of 470 nm with excitation at 340 nm. Standard compounds were purchased from Nacalai Tesque (Kyoto, Japan) and Sigma-Aldrich except ␥-Glu-Put, which was synthesized as described previously (9). In our HPLC system equipped with the Polyaminepak, ␥-Glu-Put, putrescine, cadaverine, and spermidine were eluted at 10.5, 17.5, 22.3, and 28.1 min, respectively. In the preparation of whole-cell samples, 1 ml of A 600 ϭ 1 culture was centrifuged, and the pellet was washed with 1 ml of M9-glucose minimal medium (16). The washed pellet was resuspended in 0.2 ml of 5% trichloroacetic acid (v/v; Nacalai Tesque) and boiled in a boiling water bath for 15 min to break the cell. The suspension was centrifuged, the supernatant was applied to HPLC after filtration using Millex-LH (Millipore, Billerica, MA), and the precipitated protein was dissolved in 1 ml of 0.1 N NaOH. The protein concentration of the solution was measured by the Lowry method (22), and the polyamine concentration of the cell was calculated as nmol/mg of protein.

Assays for PuuA Activity
HPLC Method-␥-Glu-Put synthetase activity was determined by measuring the decrease of glutamate. A reaction mixture containing 10 mM monosodium glutamate, 10 mM putrescine dihydrochloride, 7.5 mM ATP, 30 mM MgCl 2 , and 100 mM imidazole-HCl buffer (pH 8.0) was incubated at 37°C. After

␥-Glutamylputrescine Synthetase of E. coli K-12
stopping the reaction by adding trichloroacetic acid (final concentration, 10%), the decreased glutamate by PuuA was quantitated by HPLC.
Coupled Enzymatic Method-Another simple procedure used to determine the K m value of PuuA was based on the method described previously (23). The reaction mixture containing 100 mM imidazole-HCl buffer (pH 9.0), 10 mM monosodium glutamate, 100 mM putrescine dihydrochloride, 7.5 mM ATP, 25 mM MgCl 2 , 10 mM KCl, 1 mM phosphoenolpyruvate, 0.14 mM NADH, 5 units/ml pyruvate kinase (Oriental Yeast, Tokyo, Japan), 12.6 units/ml lactate dehydrogenase (Oriental Yeast), and 1 g/ml PuuA was incubated at 25°C. The change in absorbance at 340 nm due to oxidation of NADH was followed using a UV-visible spectrometer (model UV-1600PC; Shimadzu).

Purification of PuuA
PuuA was purified from a cell-free extract prepared from a 200-ml culture of SO97 by 0 -40% ammonium sulfate precipitation and column chromatography using a HiTrap Blue column (column volume, 5 ml; GE Healthcare). During purification, the protein was basically dissolved in buffer C (20 mM imidazole-HCl (pH 8.0) and 1 mM MnCl 2 ). After ammonium sulfate fractionation, the enzyme was dissolved in buffer C and dialyzed against the same buffer. The dialyzed enzyme solution was applied to a HiTrap Blue column equilibrated previously with buffer C. PuuA was eluted with a linear gradient formed between buffer C and 20 mM ATP in buffer C. PuuA was eluted when the concentration of ATP was 8.5-11 mM. The purified PuuA appeared as a single band on the SDS-PAGE gel (Fig. 3). This purification is summarized in Table 2. One unit of PuuA was defined as the amount of enzyme required to expend 1 mol of glutamate/min under the condition of the PuuA assay using the HPLC described under "Assays for PuuA Activity." Protein concentration was measured by the Lowry method (22).

Real Time RT-PCR Analysis
Total RNA was extracted and purified using an RNA Mini kit (Qiagen, Valencia, CA) following the manufacturer's instructions. One microgram of total RNA was treated with DNase I (final concentration, 0.1 units/l; amplification grade; Invitrogen). After adding EDTA (final concentration, 2.27 mM) to inactivate DNase I, cDNA was synthesized by an iScript cDNA Synthesis kit (Bio-Rad) from 1 g of starting total RNA primed with the random primers included in the kit according to the manufacturer's instructions. Tth RNase H (final concentration, 0.52 units/l; Toyobo) was added to the reaction mixture to remove RNA. puuA-specific primers, puuA-RT1 and puuA-RT2, were designed to amplify 95-nucleotide fragments using Primer3 software. The real time PCR mixture (brought to a final volume of 10 l with deionized water) contained 5 l of iQTM SYBR Green Supermix (Bio-Rad), 3 pmol of each of the two primers, and a 2-l cDNA sample in a 96-well optical reaction plate mounted in a DNA Engine Opticon Continuous Fluorescence Detection System (Bio-Rad). The thermal cycling conditions were as follows: 95°C for 15 min and 40 cycles of 95°C for 10 s, 60°C for 20 s, and 72°C for 20 s. To ensure the absence of nonspecific PCR products, melting curve analysis and agarose gel electrophoresis were performed after each run. The number of transcripts in a sample was determined by comparing the number of cycles (C) required for the reaction to reach a common threshold (t) with a plot of Ct values against the standard pKHG8. The relative expression levels of puuA compared with controls were calculated using Opticon (Bio-Rad). The relative amount of transcripts between samples was further standardized by amplification of the gapA gene as an internal control using primers gapA-RT1 and gapA-RT2.

Sequence Alignment of Proteins
Sequence alignment between PuuA and GS from E. coli K-12 was performed by the "needle" program of EMBOSS-Align (24). The gap penalty was 10.0, and the extend penalty was 0.5.

RESULTS
PuuA Converts Putrescine to ␥-Glutamylputrescine in Vivo-As described previously (9), SK247 (⌬puuDRCBE) deletes PuuB (␥-Glu-Put oxidase), which catalyzes the next step of PuuA in the Puu pathway (Fig. 1), and this strain accumulated ␥-Glu-Put (Fig. 4A). The concentrations of putrescine in SK247 and  Table 2 showing the purification steps. "M" indicates the prestained protein marker Broad Range (New England Biolabs, Ipswich, MA). Five micrograms of protein were applied to each lane. The concentration of acrylamide in the separation gel was 12.5%.

␥-Glutamylputrescine Synthetase of E. coli K-12
SH639 (parental strain) were 150 and 341 nmol/mg, respectively. SO62 (pBR322/⌬puuADRCBE) did not accumulate ␥-Glu-Put, whereas strains with puuA on a plasmid (SO63) or genomic DNA (SK293) on the same genetic background accumulated ␥-Glu-Put (Fig. 4B). The concentrations of putrescine in SO62, SO63, and SK293 were 450, 122, and 210 nmol/mg, respectively. These results indicated that ␥-Glu-Put is generated by the reaction catalyzed by PuuA in vivo. PuuA Reaction In Vitro-Purified PuuA catalyzed the ␥-glutamylation of putrescine (Scheme 1). This reaction, requiring Mg 2ϩ , and Mn 2ϩ can be substituted for Mg 2ϩ . When any one of the substrates and a cofactor (putrescine, glutamate, ATP, and the metal ion) were omitted from the reaction mixture, ␥-Glu-Put was not formed.
PuuA Activity for Other Amines and Ammonia-PuuA could use several diamines, spermidine, and spermine instead of putrescine (Fig. 5). The relative activity of diamines depended on the length of the methylene chain; PuuA exhibited the highest activity for putrescine, which has four carbon atoms, and activity dropped when the methylene chain of diamine was longer or shorter than C 4 (Fig. 5). PuuA activity for spermine or spermidine was Ͻ10% of that for putrescine. PuuA could not catalyze the ␥-glutamylation of ornithine or GABA. PuuA was originally annotated as GS; however, it could not form glutamine from ammonia and glutamate.
Properties of PuuA-The optimal pH of PuuA of the ␥-Glu-Put synthesis reaction is 9. The optimal Mg 2ϩ concentration of PuuA is 25 mM. The V max of PuuA is 6.71 mol/min/mg of protein, and the calculated k cat for PuuA monomer is 5.94 s Ϫ1 . The K m is 2.07 mM for glutamate, 2.35 mM for ATP, and 44.6 mM for putrescine. PuuA was inactivated by EDTA treatment, and activity was not recovered by the addition of Mg 2ϩ or Mn 2ϩ . By gel filtration of native PuuA, the molecular weight of native PuuA was calculated to be 630,000. Because the molecular weight of PuuA monomer was 53,000 by SDS-PAGE analysis, it was suggested that PuuA forms a dodecamer as does GS.
Comparison of the Puu Pathway with the YgjG-YdcW Pathway-E. coli metabolizes putrescine to GABA by two pathways (Fig. 2): the Puu pathway (Fig. 2, left pathway) (9) and the metabolic pathway, first reported in the 1980s (Fig. 2, right pathway) (11), that is composed of YgjG (13) and YdcW (12). In the latter pathway, putrescine is metabolized to GABA without ␥-glutamylation. In contrast, in the Puu pathway, putrescine is ␥-glutamylated once and metabolized to ␥-Glu-GABA via ␥-glutamyl-␥-aminobutyraldehyde, and then the ␥-glutamyl linkage of ␥-Glu-GABA is hydrolyzed by PuuD to form GABA and Glu. To determine which pathway is more important in the utilization of putrescine as a sole nutrient source, several gene deletion mutants involved in puuA, ygjG, and ydcW were constructed, and the strains were incubated at 20°C on a W-Glc-Put plate containing putrescine as the sole source of nitrogen and an M9-Put-AS plate containing putrescine as the sole

␥-Glutamylputrescine Synthetase of E. coli K-12
source of carbon. The deletion mutants of puuA could not grow on the W-Glc-Put plate (Fig. 6A) or M9-Put-AS plate (Fig. 6B); however, the deletion of ygjG and/or ydcW had no influence on the utilization of putrescine as the sole source of nitrogen or carbon (Fig. 6, A and B). When incubated at 37°C, all strains, including the wild-type strain, could not grow on the M9-Put-AS plate.
Physiological Role of PuuA-To confirm the importance of PuuA to utilize putrescine as a nutrient source in E. coli, a complementation experiment was performed. SK308 (pBelobac11/puuA ϩ ), SO115 (pBelobac11-puuA/⌬puuA), and SO116 (pBelobac11/⌬puuA) were incubated at 20°C on plates with putrescine as the sole source of nitrogen or carbon (grown on W-Glc-Put and M9-Put-AS plates). Strains, which have puuA ϩ on genomic DNA (SK308) or on the single copy plasmid pBelobac11 (SO115), could grow on a plate with putrescine as the sole source of nitrogen (Fig. 6C) or carbon (Fig. 6D), respectively, whereas the ⌬puuA strain (SO116) could not grow. When incubated at 37°C, SO115 (pBelobac11-puuA/⌬puuA) formed mucoid colonies on the W-Glc-Put plate, and none of the strains could grow on the W-Put-AS plate (data not shown). The results indicated that the ␥-glutamylation of putrescine catalyzed by PuuA is absolutely essential for the utilization of putrescine as a sole nitrogen or carbon source in E. coli.
Expression Pattern Analysis of puuA by Real Time RT-PCR-Real time RT-PCR analysis revealed that the expression of puuA was induced by putrescine (Fig. 7A), one of the substrates of the PuuA reaction. The ⌬puuR strain overexpressed puuA (Fig. 7B).
Site-directed Mutagenesis of the Catalytic Site of PuuA-PuuA was originally annotated as GS on the basis of the amino acid sequence. The active amino acid residues of GS have been well studied, and 19 amino acid residues of GS have been identified as active amino acid residues (25). PuuA shares eight of the 19 important active amino acid residues of GS. Two amino acid residues of PuuA (Fig. 8), His-282 (His-269 in GS) and Arg-357 (Arg-344 in GS), were chosen from the amino acid residues conserved by GS and PuuA in their catalytic site. Sitedirected mutagenesis of PuuA H282N and PuuA R357Q was performed to confirm that the amino acid residues conserved by GS and PuuA are important in the enzyme activity of PuuA.

␥-Glutamylputrescine Synthetase of E. coli K-12
PuuA activities of the cell-free extract of strains, which overexpressed mutagenized PuuAs, were measured. PuuA activity was impaired to 9% of wild-type PuuA (measured using the cell-free extract of SO97) by the H282N mutation (SO106), to 3% by R257Q (SK326), and to 0.05% by the double mutation of H282N and R357Q (SK327). These results are consistent with previous results that GS H269N exhibits 3% and GS H269N/R344Q exhibits only 0.1% activity of GS ∧ WT (26). All mutated PuuAs were very rapidly inactivated (after Ͻ3 h of incubation in buffer C at 4°C).
Sequence Alignment between PuuA and GS-The result of sequence alignment between PuuA and GS of E. coli is shown in Fig. 8. The identity is 23.7%, similarity is 38.1%, and the gap is 21.8%. GS has two divalent cation binding sites, n1 and n2, and has three sites that are involved in binding with three substrates, ammonium ion, glutamate, and ATP. Under this alignment, the amino acid residues that were reported previously (25) to be involved in binding with n1, n2, ammonium ion, glutamate, and ATP in GS are indicated, respectively. At the n2 and ATP binding sites, 100% (three of three residues and two of two residues, respectively) of active or catalytic amino acid residues in GS are conserved in PuuA. Glutamate and n1 sites are relatively well conserved (three of four residues and two of three residues, respectively) between GS and PuuA; however, none of three ammonium ion sites in GS is conserved in PuuA.

DISCUSSION
In our previous report (9), MalEfused PuuA was used to show that PuuA catalyzes the ␥-glutamylation of putrescine in vitro and in vivo because we could not express and purify non-tagged PuuA efficiently. Accordingly the physiological role and kinetic parameters of PuuA could not be evaluated meaningfully. Also regulation of the expression of puuA was not fully reported except that the regulation of puuA occurring in response to O 2 was mediated by ArcA and FNR (27). Furthermore experimental comparison of two putrescine degradation pathways in E. coli (9, 10, 12, 13) ( Fig. 2) has not been performed, and the similarity between GS and PuuA has not been discussed. The following discussion will address these issues.
Importance of PuuA Reaction in E. coli Cells-Because polyamines, including putrescine, are highly reactive molecules, they play important roles in cell growth and proliferation (5,6,28). Abnormal accumulation of polyamine, however, can lead to inhibition of cell growth and protein synthesis (29); therefore, polyamine concentration should be regulated strictly and promptly to maintain optimum cell growth. It was reported previously that N-acetylpolyamine had less effect on growth and protein synthesis than non-acetylated polyamine (30), and a pathway to synthesize N-acetylpolyamine to decrease the toxicity of excess polyamine exists in both prokaryotes and eukaryotes (6,31). Because the ␥-glutamylation of putrescine by PuuA is similar to the acetylation of polyamines, it is predicted that PuuA also plays a role in detoxifying excess polyamine. A noteworthy property of PuuA is the extraordinarily high K m value for putrescine (44.6 mM). This high K m value is consistent with the high K m value (9.2 mM) for putrescine of His-tagged YgjG (putrescine:␣-ketoglutarate aminotransferase) of E. coli (13). The intracellular concentration of free putrescine in E. coli was reported previously to be significantly high (12 mM) (32). Around this concentration, the reaction Identical and similar amino acid residues are indicated by black and gray shading, respectively. Under the alignment, the amino acid residues involved in the n1, n2, ammonium ion, glutamate, and ATP binding sites in GS are indicated. Above the alignment, two mutated amino acid residues in PuuA and a putative ATP binding motif deduced in PROSITE are indicated. Gaps are indicated by dashes. Alignment was performed by the needle program of EMBOSS-Align, and shading was performed using the BOXSHADE program.

␥-Glutamylputrescine Synthetase of E. coli K-12
velocity of PuuA is not saturated and is almost proportional to putrescine concentration because of its high K m value. This indicates that PuuA works efficiently when putrescine concentration in cells is so high that intracellular putrescine has a detrimental effect on E. coli. As a result, E. coli cells can regulate the intracellular concentration of putrescine by PuuA without changing the expression level of puuA. In contrast, several puuA overexpression strains exhibited severe growth defects (data not shown) probably because putrescine concentration decreases abnormally as a result of an excess amount of PuuA in the cell. It is predicted that the PuuA level, which influences the concentration of putrescine, must be strictly regulated.
Expression of puuA-As shown in Fig. 8, the transcription level of puuA was up-regulated by the addition of putrescine to the medium and overexpressed by the deletion of puuR. These results are consistent with the expression pattern of puuD (10). The observed expression regulation of puuA and the fact that the predicted puuA and puuD promoters are overlapped and in reverse directions (Fig. 1) suggest that all puu genes, including puuA and puuD, are synchronically regulated by PuuR and/or other regulators. For example, it was reported recently that ArcA and FNR repressed the expression of puuA when E. coli was grown under anaerobic conditions and that there are putative binding motifs of ArcA and FNR upstream of the open reading frame of puuA (27). These results indicated that excess putrescine is appropriately catabolized by products of the puu gene cluster that are regulated by ArcA, FNR, and PuuR to decrease the toxicity of putrescine and to utilize putrescine as both nitrogen and carbon sources.
Two Putrescine Degradation Pathways to Yield GABA-There are two putrescine degradation pathways in E. coli: the Puu pathway (9) in which putrescine is first ␥-glutamylated and metabolized to GABA and the YgjG-YdcW pathway (11)(12)(13) in which putrescine is metabolized to GABA without ␥-glutamylation (Fig. 2). It was reported that the ⌬ydcW strain on the MG1655 background could grow on M9 minimal medium supplemented with 0.4% putrescine as a sole carbon source and 0.2% ammonium sulfate as a sole nitrogen source or supplemented with 0.4% glucose as a sole carbon source and 0.2% putrescine as a sole nitrogen source (12). This result indicated that another pathway besides the YgjG-YdcW pathway plays an important role in utilizing putrescine as a nutrient source. As shown in Fig. 6, the ⌬puuA strain did not grow, whereas ⌬ygjG and ⌬ydcW strains grew on media supplemented with putrescine as a sole nitrogen or carbon source (Fig. 6, A and B). The result strongly indicates that the more important pathway for putrescine catabolism is the Puu pathway. In the Puu pathway, putrescine is first ␥-glutamylated and then metabolized further. Because the ␥-glutamylation of putrescine by PuuA requires ATP hydrolysis, the YgjG-YdcW pathway, which does not require ATP, seems more advantageous. Nevertheless our data indicate that the Puu pathway functions more effectively even in nutrient-limited media in which E. coli must save energy. If putrescine is metabolized by the YgjG-YdcW pathway, E. coli could apparently save energy; however, ␥-aminobutyraldehyde, an intermediate of the YgjG-YdcW pathway, is unstable and spontaneously cyclized to ⌬ 1 -pyrroline. ␥-Glutamylation of putrescine is suggested to prevent ␥-aminobutyraldehyde cycli-zation (9). The result in Fig. 6 shows that the stabilization effect of the intermediate by ␥-glutamylation is definitely effective and exceeds the loss of energy by ␥-glutamylation. ␥-Aminobutyraldehyde cyclizes to form a five-membered ring, whereas ␥-glutamyl-␥-aminobutyraldehyde would form a 10-membered ring if cyclization were to take place at both ends of the molecule. It was reported previously (33) that a 10-membered ring is 10 Ϫ8 times less likely to be formed than a five-membered ring. From the above consideration and results, it is very reasonable that the ␥-glutamylation of putrescine by PuuA, forming an amide bond between the amino group of putrescine and the carboxyl group of glutamate, prevents the cyclization reaction because the formation of a 10-membered ring of ␥-glutamyl-␥aminobutyraldehyde is difficult.
The reason for the mucoid growth of SO115 (pBelobac11-puuA/⌬puuA) incubated at 37°C on the W-Glc-Put plate is not known. It is possible that the puuA gene on this plasmid is expressed abnormally because the putative operator region of puuA was deleted. The reason why all strains, including SH639, could not grow on the M9-Put-AS plate at 37°C but grew on the same plate at 20°C also remains to be elucidated.
Similarity of PuuA and GS of E. coli-PuuA exhibits significant amino acid sequence similarity to GS (Fig. 8). The reactions of GS and PuuA are very similar in terms of the formation of an amide bond between the ␥-position carboxyl group of Glu and ammonia (in GS) and between the ␥-position carboxyl group of Glu and the amino group of putrescine (in PuuA), respectively. Furthermore gel filtration analysis revealed that PuuA exists as a dodecamer in the native state, and its activity requires Mg 2ϩ or Mn 2ϩ just like GS (25).
There are two divalent metal ion binding sites in GS, termed n1 and n2. The n1 site, which consists of Glu-131, Glu-212, and Glu-220 in GS, is involved in the conformational change of GS that is responsible for enzyme activity (25). Judging from the amino acid sequence, two of three amino acid residues, Glu-131 and Glu-220 in GS, are conserved in PuuA, but the region around Glu-212 in GS is significantly different from that in PuuA (Fig. 8). Moreover the binding site of the ammonium ion, which consists of Glu-50, Tyr-179, and Glu-212 in GS, is adjacent to the n1 site, and none of these three amino acid residues are conserved in PuuA (Fig. 8) probably because the substrate of PuuA is putrescine. As a result, it is possible that the conformation around the n1 site of PuuA is significantly different from that of GS.
The other metal binding site, the n2 site, composed of Glu-129, His-269, and Glu-357 in GS, is overlapped in the ATP binding site (25). All three amino acid residues that comprise the n2 site in GS are conserved in PuuA (Fig. 8). It was reported that His-269 and Arg-344 of GS were involved in forming both the ATP binding site and the n2 site (25). The two amino acid residues of PuuA, His-282 and Arg-357, which correspond to His-269 and Arg-344 of E. coli GS (Fig. 8), were chosen in a site-directed mutagenesis experiment. The significant decrease of enzymatic activity of PuuA H282N and PuuA R359Q mutations suggested that these amino acid residues form the same type of ATP binding site as in GS.
The n2 site is also involved in metal-catalyzed oxidative modification in GS (26). This modification is the initial and impor-␥-Glutamylputrescine Synthetase of E. coli K-12 tant step in the proteolytic degradation of GS. The oxidative modification of His-269 to Asn and Arg-344 to Gln induces the loss of activity of GS followed by increased susceptibility to proteolytic degradation (26). This mechanism is thought to be one of the regulation systems of the concentration of GS protein in the cell (34). As in GS, site-directed mutagenesis of the corresponding two amino acids of PuuA decreased its enzyme activity, and mutated PuuAs were rapidly inactivated. This suggests that the intracellular concentration of PuuA protein is also regulated posttranslationally by oxidative modification and proteolysis in response to the cellular concentration of putrescine.