Spermidine-preferential uptake system in Escherichia coli. ATP hydrolysis by PotA protein and its association with membrane.

PotA protein, one of the components of the spermidine-preferential uptake system in Escherichia coli, was purified to homogeneity, and some of its properties were examined. PotA protein showed Mg- and SH-dependent ATPase activity. The specific activity was approximately 400 nmol/min/mg of protein and the K value for ATP was 385 μM. The nature of the ATP binding site was explored by identification of the amino acid residue photoaffinity-labeled with 8-azido-ATP. It was found that 8-azido-ATP was attached to cysteine 26. In the spermidine transport-deficient mutant E. coli NH1596, valine 135 of PotA protein, which is located between two consensus amino acid sequences for nucleotide binding (50-57 and 168-173), was replaced by methionine (Kashiwagi, K., Miyamoto, S., Nukui, E., Kobayashi, H., and Igarashi, K.(1993) J. Biol. Chem. 268, 19358-19363). This mutated PotA protein could be labeled with 8-azido-ATP, but showed very low ATPase activity. To identify which cysteine is involved in the function of potA protein, cysteines 26, 54, and 276 were replaced by alanine, threonine, and alanine, respectively. Among the three mutated PotA proteins, the mutated PotA protein C54T only lost both ATPase and spermidine uptake activities. The results taken together indicate that the adenine portion of ATP interacts with a domain close to the NH-terminal end of PotA protein, and active centers of ATP hydrolysis are located both within and between the two consensus amino acid sequences for nucleotide binding. Association of PotA protein with membranes was strengthened by the existence of channel forming PotB and PotC proteins. ATPase of PotA protein was inhibited by spermidine, suggesting that uptake inhibition by spermidine may function during this process.

tained and characterized three clones of polyamine transport genes (pPT104, pPT79, and pPT71) in E. coli (6). The system encoded by pPT104 was a spermidine-preferential uptake system and that encoded by pPT79 a putrescine-specific uptake system. Furthermore, these two systems were periplasmic systems (7) consisting of four kinds of proteins: pPT104 clone encoded PotA, PotB, PotC, and PotD proteins and pPT79 clone encoded PotF, PotG, PotH, and PotI proteins, judging from the deduced amino acid sequences of the nucleotide sequences of these clones (8,9). PotD and PotF proteins were periplasmic substrate binding proteins, and PotA and PotG proteins were membrane-associated proteins having the nucleotide-binding site. PotB and PotC proteins, and PotH and PotI proteins, were transmembrane proteins probably forming channels for spermidine and putrescine, respectively. In contrast, the putrescine transport system encoded by pPT71 consisted of one membrane protein (PotE protein) having 12 transmembrane segments (10) and was active in the excretion of putrescine from cells through putrescine-ornithine antiporter (11). We also found that spermidine uptake by membrane vesicles was strongly dependent on PotD protein, and the uptake by intact cells was completely dependent on ATP through its binding to PotA protein (12).
In this study, we tried to identify the functional domain in PotA protein using the purified and several mutated PotA proteins. We found that 8-azido-ATP was attached to cysteine 26, and replacement of cysteine 54 and valine 135 by threonine and methionine, respectively, led to the loss of ATPase and spermidine uptake activities. The results also indicate that PotA protein is associated with membranes through the interaction with PotB and PotC proteins.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-A polyamine-requiring mutant, E. coli MA261 (13), generously provided by Dr. W. K. Maas, New York University School of Medicine, was grown in medium A in the absence of polyamines as described previously (14). E. coli MA261 potA::Km was prepared from E. coli MA261 by P1kc transduction as described previously (12) according to the method of Lennox (15). A proton-translocating ATPase mutant, E. coli DK8 (16), was kindly provided by Dr. M. Futai, Osaka University, and E. coli JM105 (17) was purchased from Pharmacia Biotech Inc. E. coli JM105atp Ϫ was derived from E. coli JM105 by transduction of a P1 phage-infected lysate of E. coli DK8 (⌬(atpB-atpC) ilv::Tn10) and grown in a 17-amino acid supplemented medium (18) containing 1% glucose. Appropriate antibiotics (30 g/ml chloramphenicol, 100 g/ml ampicillin, 50 g/ml kanamycin, and 15 g/ml tetracycline) were added during the culture of E. coli.
To construct pKKpotA2BC, the 1.3-kb StyI fragment and the 5.1-kb StyI-DraIII fragment were prepared from pKKpotA and pKKpotABC, respectively. The 1.4-kb StyI-DraIII fragment prepared from pMWpotA2B was then ligated together with the 1.3-kb StyI and 5.1-kb StyI-DraIII fragments described above. Plasmid pKKpotA3BC was constructed by ligating the 0.8-kb SphI fragment of PCR product for the potA3 gene and the 7-kb SphI fragment of pKKpotABC. Plasmid pKKpotA4BC was constructed by ligating the 0.6-kb XbaI-DraIII fragment of pMWpotA4B and the 7.2-kb XbaI-DraIII fragment of pKKpo-tABC. Transformation of E. coli cells with various plasmids was carried out as described by Maniatis et al. (21).
The strains and plasmids used in this study are listed in Table I.
Purification of PotA Protein-E. coli JM105/pKKpotA was grown in 20 liters of LB medium (21) at 37°C. When cell growth reached A 600 ϭ 0.3, 0.5 mM isopropyl-␤-D-thiogalactopyranoside was added to the medium, and the culture was continued for 3 h. The 100,000 ϫ g supernatant (3.9 g of protein) was prepared as described previously (22), using Buffer A containing 0.1 M potassium phosphate buffer, pH 7.5, 10 mM EDTA, and 20 M FUT-175 (6-amino-2-naphthyl-4-guanidino-benzoate dihydrochloride), a protease inhibitor (23). The proteins (2.5 g) precipitated with 50% saturation of (NH 4 ) 2 SO 4 were dissolved in Solution A (1 mM dithiothreitol, 10 M FUT-175, and 10% glycerol) containing 0.15 M potassium phosphate buffer, pH 7.5, and were applied to a DEAE-Sephadex A-50 column (4.6 ϫ 13 cm) previously equilibrated with Solution A containing 0.15 M potassium phosphate, pH 7.5. The column was eluted with a linear gradient of 0.2-0.6 M potassium phosphate, pH 7.5, in Solution A (600 ml). The PotA protein was eluted at 0.4 M potassium phosphate, and the protein fraction containing PotA protein (110 mg) was concentrated by ultrafiltration. After the concentration of potassium phosphate was adjusted to 80 mM, the protein was applied to a Bio-Rad Econo-Pac Q Cartridge (5 ml) previously equilibrated with Solution A containing 80 mM potassium phosphate buffer, pH 7.5. The column was eluted with a linear gradient of 0.08 -0.4 M potassium phosphate, pH 7.5, in Solution A (200 ml). The fraction containing PotA protein (23.4 mg) was concentrated and chromatogra-phy with a Bio-Rad Econo-Pac Q Cartridge was repeated. Finally, 10.2 mg of PotA protein (95% purity) was obtained (Fig. 1). PotA protein was identified by Western blotting using antibody for PotA protein as described below. Mutated PotA protein (V135M) (12) was purified from E. coli JM105/pKKpotA1BC by the same method.
Assays for ATPase and Spermidine Uptake-Inside-out membrane vesicles were prepared by French press treatment of E. coli cells suspended in 0.1 M potassium phosphate buffer, pH 6.6, and 10 mM EDTA according to the method of Houng et al. (24). ATPase activity was measured by the method of Lill et al. (25), except that the reaction mixture (0.025 ml) contained 50 mM Hepes-KOH, pH 7.5, 50 mM KCl, 10 mM magnesium acetate, 1 mM [␥-32 P]ATP (specific activity, 10 -50 cpm/ pmol), and purified PotA protein or inside-out membrane vesicles. Spermidine uptake by intact cells was performed as described previously (4) using 10 M [ 14 C]spermidine as substrate.
Photoaffinity Labeling of PotA Protein with 8-Azido-ATP-Inside-out membrane vesicles (100 g of protein) were added to a buffer containing 10 mM Tris-HCl, pH 7.4, 5 mM CaCl 2 , and 4 M 8-azido-[␣-32 P]ATP (111 kBq) in a final volume of 0.1 ml and placed in a well on ice. The reaction mixture was irradiated for 3 min with UV light (12 watts) at 365 nm at a distance of 2 cm (26). The samples were then centrifuged for 20 min at 150,000 ϫ g, resuspended without boiling in Laemmli sample buffer (27) with the omission of 2-mercaptoethanol, and subjected to electrophoresis on a sodium dodecyl sulfate-12% polyacrylamide gel. Autoradiography of dried gel was performed at Ϫ80°C using an intensifying screen.
Identification of 8-Azido-ATP Photoaffinity-labeled Amino Acid in PotA Protein-Photoaffinity labeling of PotA protein was carried out as described above except that the reaction mixture (0.1 ml) contained 40 g of purified PotA protein and 50 M 8-azido-ATP. Digestion of 8-azido-ATP-labeled PotA protein with 1-tosylamino-2-phenylethyl chloromethyl ketone-treated trypsin was performed according to the method of Mimura et al. (28) using 0.8 g of the enzyme. Peptides thus obtained were separated by HPLC as described previously (29). The peptide peaks absorbing ultraviolet rays at 260 nm were analyzed by automated Edman degradation on a protein sequencer (Applied Biosystems model 477A) equipped with a phenylthiohydantoin analyzer (model 120A).
Binding of N-[ 14 C]Ethylmaleimide to PotA Protein-The reaction mixture (1.0 ml) containing 50 mM Hepes-KOH, pH 7.5, 50 mM KCl, 10 mM magnesium acetate, 0.16 mM N-[ 14 C]ethylmaleimide (37 kBq) and 50 g of protein of inside-out membrane vesicles was incubated at 37°C for 3 min. The reaction was terminated by the addition of dithiothreitol at the final concentration of 1 mM. The proteins of membrane vesicles were subjected to gel electrophoresis (27) followed by fluorography (30). Radioactivity on dried gel was quantified by Fujix imaging analyzer Fuji BAS 2000 (Fuji Photo Film Co. Ltd.).

RESULTS
Purification and Properties of PotA Protein-PotA protein was purified as described under "Experimental Procedures." The most purified preparation was found to be nearly homogeneous (Fig. 1). The molecular mass of the band was estimated to be 45 kDa, close to the deduced molecular mass (43 kDa) from the nucleotide sequence of potA gene (8). The NH 2 -terminal amino acid of the protein was probably blocked, since it could not be determined by Edman degradation.
ATPase activity of PotA protein was dependent on Mg 2ϩ , and 10 mM Mg 2ϩ was necessary to obtain the maximal activity ( Fig.  2A). The ATPase activity was strongly inhibited by spermidine, and the function of Mg 2ϩ could not be replaced by spermidine (Fig. 2B). Spermine, but not putrescine, also inhibited the ATPase activity. Since spermidine uptake was inhibited by the already accumulated spermidine, the inhibition may be in operation during this process. The specific activity was approximately 400 nmol/min/mg of protein (Table II), and the K m value for ATP was estimated to be 385 M according to the Michaelis-Menten kinetics. Since spermidine uptake was inhibited by N-ethylmaleimide (NEM) and p-chloromercuribenzoic acid (4), effect of the inhibitors on the ATPase activity was examined. As shown in Fig. 3, the ATPase activity of PotA protein was inhibited by NEM and p-chloromercuribenzoic acid, and the inhibition was restored by the addition of dithiothreitol.
Properties of PotA1 (V135M) Protein-Since mutated PotA1 (V135M) protein is expressed in E. coli NH1596, a spermidine uptake-deficient mutant (12), the properties of PotA1 protein were examined. As shown in Fig. 4, PotA1 protein on the inside-out membrane vesicles was photolabeled with 8-azido-[␣-32 P]ATP to almost the same degree as that of the normal PotA protein on the membrane vesicles (12). However, the

FIG. 2. Effect of Mg 2؉ (A) and spermidine (B) on ATPase activity of PotA protein.
Assays were performed under standard conditions except that Mg 2ϩ in the reaction mixture was changed and spermidine was added to the reaction mixture as shown in the figure. Each value is the average of three determinations. The standard deviation was within Ϯ 10% for each data point.

FIG. 3. Effect of N-ethylmaleimide (A) and p-chloromercuribenzoic acid (B) on ATPase activity of PotA protein.
Assays were performed with (E) or without (q) 2 mM dithiothreitol. Each value is the average of three determinations. The standard deviation was within Ϯ 10% for each data point.   membrane vesicles did not have any ATPase activity (Table II). The inability of PotA1 protein to hydrolyze ATP was confirmed with purified PotA1 protein. These results indicate that PotA1 protein is a mutant of ATPase activity.
Identification of 8-Azido-ATP Photoaffinity-labeled Amino Acid-Purified PotA protein was photoaffinity-labeled with 8-azido-ATP in the presence of 5 mM Ca 2ϩ , in which the ATPase activity of PotA protein was inhibited. Then, the PotA protein was digested with trypsin and subjected to HPLC to separate the hydrolyzed products. Fig. 5 shows the peptide elution profile obtained by HPLC. There were two peaks showing an absorbance at 260 nm: one (peak A) was eluted at 11.3 min and the other (peak B) at 24.5 min. Since peptide B also showed absorbance at 275 nm and A 275 was higher than A 260 , it was expected that peptide B contained tryptophan. The sequence of peptide B was MAINWVESWEVVLADEEHK, corresponding to the carboxyl-terminal tryptic peptide (360 -378) of PotA protein (Fig. 6A). However, peptide A did not show strong absorbance at 275 nm, indicating that it contained the 8-azido-ATP photoaffinity-labeled amino acid. The sequence of peptide A was XFDGK (Fig. 5). The first amino acid was not identified and the others corresponded to the 27-30 amino acids of PotA protein (Fig. 6A). Thus, it was concluded that cysteine 26 was photoaffinity-labeled with 8-azido-ATP.
Determination of Cysteine Residue Involved in ATPase Activity of PotA Protein-PotA protein contains three cysteine residues (Cys-26, Cys-54, and Cys-276). To identify which cysteine is involved in its ATPase activity, they were converted to alanine, threonine, and alanine, respectively, using site-directed mutagenesis on the potA gene. ATPase activity was measured using inside-out membrane vesicles prepared from E. coli JM105atp Ϫ /pKKpotABC, in which the potA gene was modified by site-directed mutagenesis. As shown in Table III, only PotA3 (C54T) protein did not have significant ATPase activity. The other proteins A2 (C26A) and A4 (C276A) showed the ATPase activity. Since the amount of PotA4 protein on the vesicles was small (Figs. 7, C and D), the ATPase activity of the vesicles with PotA4 protein was lower than with PotA2 protein. When another PotA3* (C54A) protein was made, the vesicles containing the protein did not show any ATPase activity, in spite of its small amount (rapidly degraded) (data not shown). Spermidine uptake activity was measured using E. coli MA261 potA:: Km/pMWpotAB, in which potA gene was modified by site-directed mutagenesis. The uptake activity paralleled the activity of ATPase. E. coli MA261 potA::Km/pMWpotA3B showed very low spermidine uptake activity (Table III).
Next, the binding of [ 14 C]NEM to PotA2, PotA3, and PotA4 proteins was examined. As shown in Fig. 7B, [ 14 C]NEM could bind to the three mutated PotA proteins. When the amount of [ 14 C]NEM bound to the mutated PotA proteins was calculated by measuring radioactivity and density of band from the data of Fig. 7, B and D, it was approximately 60 -70% of that to the normal PotA protein. The results suggest that NEM can bind to all three cysteines (Cys-26, Cys-54, and Cys-276). The secondary structure of PotA protein was analyzed by the method of Chou and Fasman (32). As shown in Fig. 6B, the three cysteines are located at the turning region of the protein.
Stimulation of Association of PotA Protein with Membranes by PotB and PotC Proteins-It was tested whether membrane association of PotA protein is influenced by the channel-forming PotB and PotC proteins. Inside-out membrane vesicles were prepared from E. coli JM105/pKKpotA and JM105/ pKKpotABC cultured in the presence of isopropyl-␤-D-thiogalactopyranoside. Although the total amount of PotA protein was almost the same in both cells (Fig. 8A), the association of PotA protein with membranes was greatly stimulated by PotB and PotC proteins (Fig. 8B). The results indicate that PotA protein is associated with membranes through the interaction with PotB and PotC proteins.

DISCUSSION
The spermidine-preferential uptake system belongs to periplasmic active transport systems (permeases), which consist of one periplasmic substrate-binding protein and three membrane-bound components (7). One of the membrane components is a nucleotide-binding protein involved in energy supply. A model for the structure of the nucleotide-binding protein (HisP) in the histidine transport system was proposed by analogy to the adenylate kinase structure (33). It was shown that site A (Gly-X-X-Gly-X-Gly-Lys) of consensus amino acid sequences for nucleotide binding is important for ATP hydrolysis. Furthermore, MalK protein, the nucleotide-binding protein in the maltose transport system, has been purified to homogeneity, and it was reported that the specific activity of MalK protein as ATPase was approximately 130 nmol/min/mg of protein (34). The specific activity of PotA protein as ATPase was 396 nmol/min/mg of protein, higher than that of MalK protein, and the activity was inhibited by NEM. We found that cysteine 54, located in site A (GPSGCGKT), is involved in ATP hydrolysis. This supports the idea that site A is important for ATP hydrolysis (33). We found that valine 135 is also located in the active center of ATPase in PotA protein. Replacement of proline 172 by threonine in site B (Pro-X-Val(Leu)-Leu-Leu-X-Asp-Glu) of the consensus amino acid sequences for nucleotide binding in HisP protein-stimulated ATPase activity (35,36). These results indicate that the active sites of ATP hydrolysis of the nucleotide-binding protein in periplasmic active transport systems are located both within and between the two consensus amino acid sequences for nucleotide binding.
In MalK protein, cysteine is located at the equivalent position of cysteine 54 of PotA protein (37). When this cysteine was replaced by glycine, the transport activity did not change significantly (38). This suggests that cysteine may not necessarily be essential, but still be important for the ATPase and transport activities. When ATPase activities of HisP and MalK were measured using proteoliposomes equivalent to rightside-out membrane vesicles, the existence of substrate and substratebinding protein in proteoliposomes was essential for the ATPase activity (39,40). Structure of the nucleotide-binding protein in inside-out membrane vesicles or in free form may be different from that in proteoliposomes. We also found that cysteine 26 was photoaffinity-labeled by 8-azido-ATP. Cysteine 26 is the 24th amino acid from site A (GPSGCGKT) (Fig. 6A). It has been reported that histidine 19 of HisP protein is photoaffinity-labeled by 8-azido-ATP (28) and is the 21st amino acid from site A. The results suggest that the adenine portion of ATP interacts with a domain close to the NH 2 -terminal end of the nucleotide-binding protein in periplasmic active transport systems. When cysteine 26 of PotA protein was replaced by alanine, the mutated PotA protein was still photoaffinity-labeled with 8-azido-ATP. Threonine 39 in the mutated PotA protein might be photoaffinity-labeled by 8-azido-ATP instead of cysteine 26, since it has been reported that serine 41 in HisP protein was also photoaffinity-labeled in addition to histidine 19 in the presence of 5 mM Ca 2ϩ and 10 mM Mg 2ϩ (28). Thus, the pocket for ATP on the nucleotide-binding protein in periplasmic active transport systems may be wide. The results taken together indicate that the center of ATPase activity is located in the NH 2 -terminal portion in PotA protein.
In fact, the NH 2 -terminal peptide consisting of 239 amino acids of PotA protein showed ATPase activity. 2 It has been reported that the interaction of HisP protein with membranes was enhanced by HisQ and HisM membrane proteins (41). We also found that the association of PotA protein with membranes was strengthened by the existence of PotB and PotC channel forming proteins. When the secondary structures of PotB and PotC proteins were compared, common amino acid sequences, LEAAR(K)DLGAS, were observed in the hydrophilic region (8). These sequences may be involved in the interaction with PotA protein. It has been reported that a similar amino acid sequence has been found in a hydrophilic loop of channel forming membrane proteins in periplasmic active transport systems (42).