The ATPase activity and the functional domain of PotA, a component of the sermidine-preferential uptake system in Escherichia coli.

The ATPase activity of PotA, a component of the spermidine-preferential uptake system consisting of PotA, -B, -C, and -D, was studied using purified PotA and a PotABC complex on inside-out membrane vesicles. It was found that PotA can form a dimer by disulfide cross-linking but that each PotA molecule functions independently. When PotA was associated with the membrane proteins PotB and PotC, the K(m) value for ATP increased and PotA became much more sensitive to inhibition by spermidine. It was also shown that spermidine uptake in cells was gradually inhibited in parallel with spermidine accumulation in cells. The results suggest that spermidine functions as a feedback inhibitor of spermidine transport. The function of PotA was analyzed using PotA mutants obtained by random mutagenesis. There are two domains in PotA. The NH2-terminal domain (residues 1-250) contains the ATP binding pocket formed in part by residues Cys26, Phe27, Phe45, Cys54, Leu60, and Leu76, the active center of ATPase that includes Val135 and Asp172, and amino acid residues necessary for the interaction with a second PotA subunit (Cys26) and with PotB (Cys54). The COOH-terminal domain (residues 251-378) of PotA contains a site that regulates ATPase activity and a site involved in the spermidine inhibition of ATPase activity.

periplasmic binding proteins. The polyamine binding site lies in a cleft between the two domains as determined by crystallography and site-directed mutagenesis (9,10).
We also purified a membrane-associated ATPase (PotA) of the spermidine-preferential uptake system, and some properties of PotA, including the existence of an ATP binding site in the NH 2 terminus, were clarified (11). Recently, the structures of HisP (12) and MalK (13), membrane-associated ATPases of the histidine and maltose uptake systems, respectively, have been determined by x-ray crystallography. These proteins are likely to have functions in histidine and maltose uptake analogous to the role of PotA in polyamine uptake. The structure of HisP is very similar to that of the NH 2 -terminal domain of MalK (12,13), but HisP (258 amino acid residues) is smaller than MalK (372 amino acid residues), and MalK has an extra domain in the COOH terminus. It has been reported that the COOH terminus of MalK is critical for negative regulation of the mal regulon (14). Furthermore, it was recently found that a mutant (E306K) of the COOH terminus of MalK affects its ATPase activity, suggesting a role for this region in the ATPase activity (15). PotA (378 amino acid residues) is also expected to have an extra COOH-terminal domain, similar to that in MalK. In the present work, we studied the functions of the NH 2 -and COOH-terminal domains of PotA. We found that the NH 2terminal domain is involved in the recognition of ATP and in the interactions of PotA with a second PotA subunit and with PotB and PotC. The COOH-terminal domain was found to be involved in regulation of ATPase activity and to influence the interactions between PotA and PotB (or PotC). Furthermore, some of the characteristics of the ATPase of PotA were examined using purified PotA and a PotABC complex.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Culture Conditions-A polyaminerequiring mutant, E. coli MA261 (16), was generously provided by Dr. W. K. Maas, New York University School of Medicine. E. coli MA261 potA::Km was prepared from E. coli MA261 as described previously (17) and was grown in medium A in the absence of polyamines (18). A proton-translocating ATPase mutant, E. coli DK8 (19), was kindly supplied by Dr. M. Futai, Osaka University, and E. coli JM105 atp Ϫ was prepared by transduction of a P 1 phage-infected lysate of E. coli DK8 (⌬(atpB-atpC) ilv::Tn10) and grown in an 18-amino acid-supplemented medium (20) containing 1% glucose. Plasmids pMWpotAB and pKK-potABC were prepared as described previously (17). Transformation of E. coli cells with plasmids was carried out as described by Maniatis et al. (21). Appropriate antibiotics (100 g/ml ampicillin, 15 g/ml tetracycline, and/or 50 g/ml kanamycin) were added during the culture of E. coli having the above plasmids.
Random Mutagenesis and Selection of PotA Mutants-Random mutagenesis was carried out using a PCR-based strategy (22). To obtain 1.5 kilobase pairs of mutated potA genes, PCR was performed using 5Ј-TAAGAGTCACCAAGGTGGTTAACC-3Ј (P1, sequence for Ϫ61 to Ϫ38 of potA gene) and 5Ј-CGGACGCACCTTGTGTGGCAACTT-3Ј as * This work was supported by a grant-in-aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 81-43-290-2897; Fax: 81-43-290-2900; E-mail: iga16077@p.chiba-u.ac.jp. 1 The abbreviations used are: ABC, ATP binding cassette; BMH, 1,6-bis-maleimidohexane. the 5Ј-and 3Ј-primers, respectively. The first cycle was carried out in the presence of 200 M of each of three dNTPs and 0.5 M of the fourth dNTP, with 200 M of the fourth dNTP added in the subsequent 24 cycles. Four separate reaction mixtures (with dA, dC, dG, and dT, each at low concentrations in the first cycle) were combined, purified, and digested with EcoRI. The digested fragments, which encode 24 -343 amino acid residues of PotA, were ligated with a 5.8-kilobase pair fragment obtained with pMWpotAB. E. coli MA261 potA::Km was transformed with the mutated pMWpotAB, and cells were grown on 1.5% agar plates containing medium A and 30 g/ml spermidine. Colonies thus obtained were used for the assay of spermidine transport described below.
Construction of pMWpotAC1 and pMWpotAC2-To construct pKKpo-tAC1, PCR was performed using pMWpotAB as template, P1 as described above, and 5Ј-CCCCTGCAGTTAAACACGTAAGTCTT-3Ј as primers. The PCR product thus obtained was digested with PstI and Csp45I and inserted into the same restriction sites of pKKpotABC. Site-directed mutagenesis by overlap extension using PCR (23) was performed to prepare pKKpotAC2. The template used for the first PCR was pMWpotAB. Primers used for first PCR were P1 and 5Ј-AACAAA-CAGGTTTTACGGCTCTTCGT-3Ј (complementary sequence for 732-707 except underlined base), 5Ј-ACGAAGAGCCGTAAAACCTGTTT-GTT-3Ј (sequence for 707-732 except underlined base), and 5Ј-AAGTTAACGTAGATATTACAT-3Ј (P2, complementary sequence for 845-825). A second PCR was performed using initial PCR products as templates and P1 and P2 as primers. PCR product thus obtained was digested with Csp45I and XbaI and inserted into the same restriction sites of pKKpotA (11). pMWpotAC1 was constructed by inserting the StyI-NruI fragment of pKKpotAC1 into the same restriction sites of pMWpotAB. Similarly, pMWpotAC2 was constructed by inserting the StyI-XbaI fragment of pKKpotAC2 into the same restriction sites of pMWpotAB.
Preparation of pMWpotA E297Q 2 and E297D-To construct these mutants, site-directed mutagenesis was carried out by using the QuikChange TM site-directed mutagenesis kit (Stratagene).
Photoaffinity Labeling of PotA Protein with 8-Azido-ATP-Inside-out membrane vesicles (50 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 (122 watts) at 365 nm at a distance of 2 cm (27). The samples were then centrifuged for 20 min at 150,000 ϫ g, resuspended without boiling in 2-mercaptoethanol-free SDS sample buffer (28), and subjected to 12% SDS-polyacrylamide gel electrophoresis. Binding strength of 8-azido-[␣-32 P]ATP to PotA was estimated with a Fujix Bas 2000II imaging analyzer using the dried gel.
Western Blot Analysis of PotA Protein-Antibodies for PotA protein were prepared as described previously (29) using the conjugate of the deduced NH 2 -terminal 18 amino acids (GQSKKLNKQPSSLSPLVQ) or the deduced COOH-terminal 14 amino acid residues (VESWEVV-LADEEHK) of PotA protein together with bovine thyroglobulin. Western blotting was performed according to the method of Nielsen et al. (30). Disulfide Cross-linking-Disulfide cross-linking was performed by the method of Kubo et al. (31) with some modifications. The reaction mixture (50 l) containing 10 mM Tris-HCl, pH 8.0, 28 mM KCl, 2% glycerol, 0.4 mM 2-mercaptoethanol, and 50 g of inside-out membrane vesicle protein was incubated with 1,6-bis-maleimidohexane (BMH; spacer arm, 16 Å) (32) at a final concentration of 0.5 mM at 25°C for 30 min. After the samples were mixed with 2-mercaptoethanol-free SDS sample buffer (28), samples containing 10 g of protein were subjected to 12% SDS-polyacrylamide gel electrophoresis, and subsequent Western blotting was performed as described above. pKK potAB(C69S)C, Measurement of Polyamines and Protein and Analysis of Sequence Homology of Proteins-Polyamine levels in E. coli were determined by high pressure liquid chromatography as described previously (33) after homogenization of cells and extraction of the polyamines with 5% trichloroacetic acid and centrifugation at 27,000 ϫ g for 15 min at 4°C. Protein was determined by the method of Lowry et al. (34). Sequence homologies of the proteins were analyzed according to the method of Needleman and Wunsch (35) using a DNASIS program.

Comparison of the Amino Acid Sequences of PotA, MalK, and
HisP-The amino acid sequence of PotA was first compared with the sequences of MalK and HisP, whose structures have already been determined by x-ray crystallography (12,13). MalK has a longer COOH-terminal domain compared with HisP. As shown in Fig. 1, the NH 2 -terminal domain (amino acid residues 1-250) of PotA was similar to that of MalK and to the sequence of HisP. The COOH-terminal domain of PotA (amino acid residues 251-378) had only limited identity to that of MalK, in which the regulatory site of mal regulon (14) and ATPase activity (15) exists. We have previously reported that the NH 2 -terminal domain of PotA has the active site of ATPase (11,17). We hypothesized that the COOH-terminal domain of PotA has a unique function related to the spermidine uptake system of which PotA is a component.
Comparison of ATPase Activities of Purified PotA and a PotABC Complex-The ATPase activity of purified PotA and of a PotABC complex on inside-out membrane vesicles was measured in the presence of 50 mM K ϩ and 10 mM Mg 2ϩ . As shown in Fig. 2A, the K m value for ATP with purified PotA was 390 M with a Hill coefficient of 0.98. The K m value for ATP increased about 3-fold for the PotABC complex (1.49 mM) compared with PotA, and the Hill coefficient was 1.02 (Fig. 2B). The results indicate that the affinity for ATP decreases when PotA makes a complex with PotB and PotC and that there is no cooperativity for ATP during ATP hydrolysis even in the PotABC complex.
When PotA is complexed with PotB and PotC, sensitivity to stimulation by Mg 2ϩ and to inhibition by spermidine are both greatly increased (Fig. 3). The K i value for spermidine was ϳ10 M at the PotABC complex, and spermidine uncompetitively inhibited PotA activity, suggesting that spermidine binds at a site on PotA different from the ATP recognition site. When cells were incubated for 40 min in the presence of 100 M spermidine, spermidine gradually accumulated in cells, and spermidine uptake activity was reduced in parallel with spermidine accumulation (Fig. 4). Under these conditions, the level of PotA in cells did not change significantly as determined by Western blotting of PotA (data not shown). These results suggest that spermidine functions as a feedback inhibitor of spermidine uptake through inhibition of the ATPase activity of PotA.
Characteristics of PotA Mutants That Influence Spermidine Uptake-To determine which regions are involved in the function and regulation of PotA, PotA mutants that influence spermidine uptake were isolated by random mutagenesis. E. coli MA261 potA::Km was transformed with the mutant pMWpo-tAB, colonies were isolated, and the spermidine uptake of the colonies were measured. As shown in Fig. 5A, a number of PotA mutants that reduced spermidine uptake were isolated. These were PotA F27L, F45L, L60F, L76P, D172N, and E297K. Another mutant, PotA V135M, was an ATPase-deficient mutant that has been previously isolated (17). All mutants except PotA E297K were located in the NH 2 -terminal domain in which the active center of ATPase is located (Fig. 5B). To clarify the role of the COOH-terminal domain in ATPase activity, PotA mu- tants E297Q and E297D and the COOH-terminal truncated mutants C1 (301 amino acid residues) and C2 (239 amino acid residues) were constructed (Fig. 5B). As shown in Fig. 5A, the spermidine uptake activity of these mutants was greatly reduced. The ATPase activities of all mutants were also determined using inside-out membrane vesicles. The decrease in spermidine uptake was nearly parallel with the decrease in ATPase activity in all mutants except PotA L76P and C1 (Fig.  5A). For PotA L76P, the decrease in spermidine uptake was much greater than that in ATPase activity. Proline may change the structure of PotA greatly, so the interaction between the mutated PotA L76P and the channel-forming PotB and PotC may decrease. For the COOH-terminal-truncated mutant C1, the decrease in ATPase activity was much greater than that in spermidine uptake. In this mutant, an uncoupled and futile ATP hydrolysis with spermidine uptake may be preferentially inhibited because the number of molecules of ATP hydrolyzed is much higher than that of spermidine influxed with native PotA.
We further analyzed these mutants by measuring the K m and V max values for ATPase activity and the affinity of binding of ATP using 8-azido-ATP binding to Cys 26 (11) with inside-out membrane vesicles (Table I). In PotA mutants F27L, F45L and L60F, the K m value for ATP increased and the strength of ATP binding decreased, suggesting that these residues contribute to the ATP binding site. In the D172L mutant, which is located in Walker motif B (Fig. 5B), ATPase activity was abolished, similar to effects seen with a V135M mutant (11), although ATP binding was observed at both the D172L and V135M (11) mutants, suggesting that Asp 172 and Val 135 are part of the active site of the ATPase. In the L76P mutant, ATPase activity was reduced due to a decrease in V max . Although the K m value for ATP was decreased with PotA L76P, the strength of ATP binding was also reduced. Interaction of 8-azido-ATP with Cys 26 on PotA L76P may be reduced due to the structural change of PotA L76P.
The COOH-terminal mutants E297K, E297Q, and E297D had ATP binding affinities and an intact ATP binding site similar to native PotA, judging from the K m values and binding of 8-azido-ATP to Cys 26 (Table I). However, the ATPase activity of these mutants was greatly reduced, indicating that the COOH-terminal domain is involved in regulation of ATPase activity. Similar results were obtained with the COOH-terminal-truncated mutants C1 and C2, but the apparent decrease in ATP binding at these mutants (0.64 and 0.70 in Table I) may be due to a decrease in the affinity of the C1 and C2 mutants for PotB and PotC. To test this idea, the total amount of native and mutant PotA in cells and the amount in the cell membranes were measured by Western blotting. Although comparable amounts of native PotA and C1 and C2 mutants were produced in cells (Fig. 6A), the amount of C1 and C2 associated with membrane was less than that of the native PotA (Fig. 6B). These results are consistent with the idea that the COOHterminal domain stimulates the association between PotA and the membrane-bound PotB and PotC subunits.
The COOH-terminal domain of PotA is involved in the inhibition of ATPase activity by spermidine. As shown in Fig. 7A, mutants in the NH 2 -terminal domain did not affect inhibition by spermidine, whereas mutants in the COOH-terminal domain greatly reduced inhibition by spermidine (Fig. 7B). This suggests that the COOH-terminal domain of PotA has a spermidine binding site involved in the regulation of its ATPase activity.
Interaction of PotA with a Second PotA Subunit and with PotB and PotC-To examine how PotA interacts with a second PotA subunit and with PotB and PotC, cross-linking of cysteine residues between subunits was studied using BMH (spacer arm, 16 Å). In PotA, there are three cysteine residues (Cys 26 , Cys 54 , and Cys 276 , see Fig. 5B). There are two cysteine residues in PotB (Cys 69 on the second transmembrane segment and Cys 201 on the fifth transmembrane segment) and one cysteine residue in PotC (Cys 139 on the fourth transmembrane segment) (29).
When native PotA was used for the cross-linking experiment, a PotA dimer as well as a PotA-PotB complex were formed (Fig.  8A, lane 2). Similar results were obtained with PotA V135M ,  F27L, F45L, L60F, D172N, E297Q, and E297D (Fig. 8A). However, no PotA-PotB complex was obtained with PotA L76P (Fig.  8A, lane 7), and a PotA-PotC complex was obtained instead of TABLE I K m and V max values of ATPase activity and strength of ATP binding to PotA mutants on inside-out membrane vesicles ATPase activities of PotA mutants were measured by changing the concentration of ATP, and the K m for ATP and V max were measured. ATP binding was measured by photoaffinity labeling of PotA mutants with 8-azido-ATP as described under "Experimental Procedures." Each value is the average of duplicate determinations. ND, not determined due to the low ATPase activity.  PotA-PotB complex with PotA E297K (Fig. 8A, lane 9).
PotA Cys 26 was involved in the formation of PotA dimers because there was no dimer formation with PotA C26A (Fig.   8B, lane 2), and PotA Cys 54 was involved in the formation of a PotA-PotB complex because this complex was not formed using PotA C54T (Fig. 8B, lane 3). The formation of a PotA-PotB complex was confirmed by the finding that its formation was reduced when a cysteine-mutated version of PotB (C69S and C201S) was used (Fig. 8B, lanes 5 and 6). The PotB C69S mutant had a larger effect on formation of the PotA-PotB complex than did the PotB C201S mutant. The PotA-PotB complex was not observed with PotB C69S/C201S double mutant (data not shown). The formation of a PotA E297K-PotC complex was confirmed by the finding that a PotA E297K-PotB complex (rather than PotA E297K-PotC) was formed when Cys 139 in PotC was mutated to serine (Fig. 8B, lane 9). Furthermore, no cross-linking between PotA E297K and PotC (or PotB) was obtained when Cys 69 and Cys 201 in PotB and Cys 139 in PotC were mutated (Fig. 8B, lane 10). These results suggest that marked structural changes occurred in the PotA L76P and E297K mutants.
Finally, cysteines were inserted in PotC at positions similar to those where Cys is found in PotB (Cys 74 on the second transmembrane segment and Cys 192 on the fifth transmembrane segment), and cross-linking of this mutated PotC with PotA was studied. As shown in Fig. 8B, lanes 11 and 12, there was no PotA-PotC complex formed, only a PotA-PotB complex. The results suggest that PotA preferentially and predominantly interacts with PotB rather than PotC.

DISCUSSION
In this study, we have investigated the ATPase activity of PotA, a component of the spermidine-preferential uptake system in E. coli. Most ABC transporters are thought to utilize two ATPase subunits in the transport process (36), and positive cooperativity for ATP during ATP hydrolysis was reported in HisP (Hill coefficient ϭ 2) (37) and MalK (Hill coefficient ϭ 1.3) (38). However, it has been reported recently that there is no cooperativity for ATP using purified HisP (39) and that one intact HisP in the heterodimers between the wild type and mutant HisP can catalyze ATP hydrolysis and histidine transport (40). For PotA, no cooperativity for ATP was observed (Hill coefficient ϭ 1.0). We have also observed that purified PotA exists predominantly as a monomer as determined by gel filtration (data not shown). Our results indicate that each PotA subunit functions independently, although PotA is able to form a dimer by disulfide cross-linking.
In the spermidine uptake system, ATPase activity is always greater than spermidine uptake activity. For example, when both activities were measured with E. coli JM105 atp Ϫ /pMW-potABC, the ratio of ATPase and spermidine uptake activities was estimated to be 3-10. We reported that spermidine strongly inhibits ATPase activity through its interaction with the COOH-terminal domain of PotA. Spermidine not only functions as a feedback inhibitor of spermidine uptake through the inhibition of ATPase but also may function as an inhibitory factor of the uncoupled and futile ATP consumption with the uptake. Feedback inhibition of spermidine uptake by spermidine accumulation also occurs at the transcriptional level, in which spermidine enhances the inhibition of transcription caused by PotD of the spermidine uptake operon (41).
Because the molecular size of PotA is similar to that of MalK, the amino acid residues involved in the ATPase activity of PotA were compared with those in MalK (15,42,43). For PotA, important amino acid residues (Phe 27 , Phe 45 , Leu 60 , Leu 76 , Met 135 , and Asp 172 ) were located in or close to the Walker motif A and B, the ABC signature region, and the switch region. The exception was Glu 297 . For MalK, important amino acid residues (Lys 42 , Leu 86 , Gly 137 , Gln 140 , Asp 165 , and His 192 ) were located in positions similar to the key residues in PotA except Glu 306 . It has been reported that Glu 306 is important for ATPase activity of MalK (15). Similarly, Glu 297 in PotA was involved in the ATPase activity and the inhibition by spermi-dine. It should be noted that both Glu 297 in PotA and Glu 306 in MalK are located in the middle of the same amino acid motif (RPEDL) (see Fig. 1), which overlaps the regulatory domain motif 2 (13,44). These results suggest that the COOH-terminal domain, which is present in PotA and MalK but not in HisP, plays important roles in ATPase activity, although both the ATP binding pocket and the active site are found in the NH 2terminal domain. In addition, the COOH-terminal domain of PotA was involved in the interaction with the channel-forming protein PotB (or PotC).
In the tertiary structure of MalK (13), Glu 306 in the RPEDL motif of the COOH-terminal domain, which is important for ATPase activity, is located close to the Walker motif B in the NH 2 -terminal domain, which is the active center of ATPase. Thus, the RPEDL may regulate the ATPase activity through its interaction with the Walker motif B. We expect that the RPEDL motif in PotA has a similar function. If this is the case, spermidine may interact with Glu 297 in the RPEDL motif and inhibit ATPase activity. When PotA makes a complex with PotB and PotC, the cleft between the NH 2 -and COOH-terminal domains may become a suitable structure to interact with spermidine. Experiments are now in progress to clarify how spermidine and the COOH-terminal domain affect ATPase activity.