Characterization of a potential catalytic residue, Asp-133, in the high affinity ATP-binding site of Escherichia coli SecA, translocation ATPase.

The high affinity ATP-binding site of SecA is located in its amino-terminal domain possessing amino acid sequences, the Walker A (GXXXXGKT) and B (ZZZZD) motifs, that are characteristic of a major class of nucleotide-binding sites (Walker, J. E., Saraste, M., Runswick, M. J., and Gay, N. J. (1982) EMBO J. 1, 945-951). Recently, we proposed that proteins possessing a typical set of Walker A and B motifs contain a conserved Glu or Asp between the two motifs. This Glu or Asp acts as a “catalytic residue” that activates a water molecule for an in-line attack on the γ-phosphate of ATP (Amano, T., Yoshida, M., Matsuo, Y., and Nishikawa, K. (1995) FEBS Lett. 359, 1-5). In the present study, the aspartate residue at position 133 in Escherichia coli SecA, which could be the “catalytic residue,” was mutated to an asparagine. The mutant SecA (SecA D133N) protein was expressed in E. coli CK4706, encoding a duplication of the secA gene, and purified to homogeneity. The in vitro protein translocation activity and membrane vesicle stimulated ATPase activity of SecA D133N were drastically reduced. Proteolytic studies indicated that the conformational changes of the mutant SecA occurring on interaction with ATP, presecretory proteins, phospholipids, and membrane vesicles, were similar to those of wild-type SecA. The mutant SecA allowed the signal peptide cleavage of proOmpA during translocation, indicating that the mutant retains the ability to bind ATP to perform the initial step of the translocation reaction. These data indicate that the carboxyl group of Asp-133 plays a role as a catalytic carboxylate, which activates a water molecule to attack γ-phosphate of ATP, and the mutant lacking this residue cannot perform the total translocation but can still perform the initial step of the protein translocation.

The high affinity ATP-binding site of SecA is located in its amino-terminal domain possessing amino acid sequences, the Walker A (GXXXXGKT) and B (ZZZZD) motifs, that are characteristic of a major class of nucleotide-binding sites ( In the present study, the aspartate residue at position 133 in Escherichia coli SecA, which could be the "catalytic residue," was mutated to an asparagine. The mutant SecA (SecA D133N) protein was expressed in E. coli CK4706, encoding a duplication of the secA gene, and purified to homogeneity. The in vitro protein translocation activity and membrane vesicle stimulated ATPase activity of SecA D133N were drastically reduced. Proteolytic studies indicated that the conformational changes of the mutant SecA occurring on interaction with ATP, presecretory proteins, phospholipids, and membrane vesicles, were similar to those of wild-type SecA. The mutant SecA allowed the signal peptide cleavage of proOmpA during translocation, indicating that the mutant retains the ability to bind ATP to perform the initial step of the translocation reaction. These data indicate that the carboxyl group of Asp-133 plays a role as a catalytic carboxylate, which activates a water molecule to attack ␥-phosphate of ATP, and the mutant lacking this residue cannot perform the total translocation but can still perform the initial step of the protein translocation.
The translocation of secretory proteins across the cytoplasmic membrane in prokaryotic cells requires several protein factors (1)(2)(3)(4). Among them, SecA is a peripheral cytoplasmic membrane protein, which plays an essential role in the translocation of secretory proteins across the cytoplasmic membrane of Escherichia coli. Biochemical studies, in addition to genetic evidence, clearly showed that SecA is involved in protein translocation (2,4). SecA interacts with both ATP and presecretory proteins (5)(6)(7). Interactions between SecA and membrane vesicles/liposomes have also been demonstrated (5,8). These interactions resulted in a conformational change of the SecA molecule (9). The possible interaction of SecA with SecY (5, 10) has also been suggested. The function of SecA was proposed to be directly related to a cycle of ATP binding and hydrolysis, which is essential for the translocation reaction (8).
The amino terminus of SecA contains typical ATP-binding motifs, the so-called Walker A and B sequences (11), that are commonly found in many nucleotide-binding proteins. The Walker A motif (also called the P-loop) consists of a consensus sequence, GXXXXGKT (X varies), and the ␤and ␥-phosphates of ATP are liganded by Lys and Thr (12). The residues of motif A form a loop that contributes to the formation of a binding cavity for the phosphoryl groups of Mg 2ϩ -ATP (13)(14)(15)(16)(17). The Walker B motif, ZZZZD (Z is a hydrophobic residue), is responsible for the Mg 2ϩ interaction among phosphates and proteins liganded to its Asp residue. A high affinity nucleotide-binding site, located within the first 217 amino acid residues from the amino terminus of E. coli SecA (7), contains these amino acid sequences. Replacement of Lys-108, located in motif A, blocks the translocation ATPase activity of SecA and interferes with the in vitro protein translocation of proOmpA (18,19). Substitution of Asp-217 of E. coli SecA prevents the growth of cells and interferes with the translocation of precursor proteins in vivo (20).
A typical set of consensus sequence motifs A and B is likely to have a common structure, six parallel ␤ strands surrounding a central ␣ helix, and to catalyze ATP-triggered reactions. Recently, we found the existence of a conserved Glu or Asp, which may act as a general base that activates a water molecule for an in-line attack on the ␥-phosphate of ATP (21). When the amino acid sequences of the region covering motifs A and B are aligned for several protein families, it became clear that the occurrence of a Glu or Asp at 24 Ϯ 2 residues from the Lys of motif A is a common feature (22). This Glu or Asp residue is well conserved in each of the protein families. Therefore, it is likely that the carboxyl group of the Glu or Asp residue plays a functional role as a "catalytic carboxylate," which activates a water molecule that attacks the ␥-phosphate of ATP in each of the proteins, as is proposed from the x-ray structural studies on the RecA protein (13) and F 1 -ATPase (17) or biochemical studies on the F 1 -ATPase ␤ subunit (23,24).
As for SecA, such a residue is found at amino acid residue 133. To determine whether the conserved Asp acts as an essential residue or not, we constructed a site-directed mutant as to Asp-133 of E. coli SecA. Our results support the validity of the catalytic carboxylate hypothesis for the SecA function and confirm the Asp-133 residue to be the "catalytic residue" for SecA ATPase activity as in the case of F 1 -ATPase ␤ subunit. They also suggest that the ATP hydrolysis at this site is not essentially involved in the early step of protein translocation.
Site-directed Mutagenesis of the secA Gene-Site-specific mutation was performed according to the method of Kunkel (37), directed by a synthetic oligonucleotide primer corresponding to the antisense strand of the secA gene fragment subcloned into pUC118. A 0.5-kilobase BamHI-SphI fragment of pMAN400 was ligated with pUC118, which had been treated with BamHI-SphI, to create pD133N. An oligonucleotide primer, 5Ј-GCCAGGTAGTTGTTAACGGTAACTAC-3Ј, was used as a mismatch primer for the replacement of Asp-133 by Asn in E. coli SecA, the changed bases being indicated by bold letters. The underline indicates the new (HpaI) restriction site. Mutations were screened as an appearance of the restriction site and confirmed by DNA sequencing. A 0.5-kilobase BamHI-SphI fragment of this plasmid was subcloned into the BamHI-SphI-digested expression vector, pTD-SecA, to construct plasmid pTD-D133N encoding secA D133N.
Expression of the Mutant SecA-E. coli strain CK4706 (29) harboring plasmids pTD-D133N and pGP1-2 (38) was used to express the secA D133N gene. Cells harboring the plasmids were grown overnight in L broth supplemented with 100 g/ml ampicillin and 50 g/ml kanamycin at 30°C. The cells were then diluted 100-fold with fresh L broth containing ampicillin and kanamycin, and at A 600 ϭ 1.0 the temperature was raised to 42°C. After 30 min, the temperature was lowered to 37°C. After 120 min, the cells were harvested and washed with 50 mM Tris acetate, pH 7.8, 10% (w/v) sucrose, and then SecA was purified as described previously (36). About 7 mg of purified SecA D133N was obtained from 3 liters of culture.
Preparation of Urea-treated Inverted Membrane Vesicles-Inverted membrane vesicles were prepared from E. coli K003 (⌬uncB-C) or W3110 M25 harboring both pMAN809 and pMAN510 (39). The membrane vesicles were treated with 6 M urea, 50 mM potassium phosphate, pH 7.5, for 1 h on ice, recovered by centrifugation, and then suspended in 50 mM potassium phosphate, pH 7.5.
Determination of Endogenous, Membrane, and Translocation ATPase Activities-A coupled spectrophotometric assay involving pyruvate kinase and lactate dehydrogenase was performed as described (40). Oxidation of NADH was continuously monitored at 340 nm with a Shimadzu UV-3000 spectrophotometer, and the amount of ATP hydrolyzed was calculated from that of NADH oxidized.
In Vitro Transcription and Translation-In vitro transcription of genes encoding presecretory proteins was performed with SP6 RNA polymerase, as described (41). The translation reaction was carried out in the presence of [ 35 S]methionine as described (42). 35 S-Labeled proOmpA was partially purified as described previously (31).
Translocation Reaction-The method of translocation of precursor proteins into everted membrane vesicles was principally the same as that described previously (43). In the case of proOmpA, the translocation mixture (100 l) comprised 35 S-labeled proOmpA, membrane vesicles (100 g of protein), 2 mM ATP, 5 mM MgSO 4 , 50 mM potassium phosphate, pH 7.5, 0.5 mg/ml bovine serum albumin, 2 mM dithiothreitol, 15 g/ml SecB, and 40 g/ml SecA. The ATP was regenerated in the presence of 10 mM creatine phosphate and 10 g/ml creatine kinase. 35 S-Labeled proOmpA was diluted into the reaction mixture from the solution comprising 6 M urea, 50 mM potassium phosphate, pH 7.5, and 1 mM dithiothreitol. Translocation reactions were started by the addition of the indicated concentrations of ATP, followed by incubation for the indicated times at 37°C. Then samples were treated with proteinase K (200 g/ml) for 20 min at 0°C, subjected to trichloroacetic acid precipitation, and analyzed by SDS-polyacrylamide gel electrophoresis and fluorography.
Proteolytic Digestion of SecA-Unless otherwise stated, all experiments were performed as described previously (9) using Staphylococcus aureus V8 protease.
SecA Membrane Binding Assay-14 C-Labeled SecA was prepared by reductive methylation of the ␣and ⑀-amino groups (44) as described (18). Binding of 14 C-labeled SecA to urea-washed E. coli K003 everted membrane vesicles was performed in 100-l reaction mixtures comprising 200 g of membrane vesicles in buffer (50 mM HEPES-KOH, pH 7.5, 30 mM KCl, 2 mM Mg(OAc) 2 , 0.5 mg/ml bovine serum albumin, and 2 mM dithiothreitol), as described previously (45). 14 C-Labeled SecA (10 -800 nM) was added, and after incubation for 30 min at 4°C, the membranes were sedimented by ultracentrifugation, and the amounts of 14 C-labeled SecA in the pellets, supernatants, and total reaction mixture were determined by quantifying radioactivity with a liquid scintillation counter. Data were corrected for background radioactivity and analyzed according to Scatchard (46).
Protein Determination-Protein concentrations were determined by the method of Lowry et al. (47) with bovine serum albumin as a standard. Fig. 1, when the amino acid sequences of the region covering Walker A and B motifs were aligned for several SecA proteins, it became clear that the occurrence of Glu or Asp at a position 24 Ϯ 2 residues away from the Lys of motif A is a common feature. To reveal the functions of these carboxyl groups and ATP hydrolysis in E. coli, Asp-133 of E. coli SecA was mutated to the cognate amide residue, asparagine. The mutant SecA protein thus produced was difficult to separate from the wild type. To overcome this problem, we used E. coli CK4706 as a host strain, which has duplicated SecA coding sequences and which expresses functional fusion SecA (29). As a result, the SecA D133N protein was purified to homogeneity (data not shown).

Construction and Expression of the Mutant secA Gene-As shown in
SecA D133N Is Impaired in ATPase Activity-SecA exhibits low endogenous ATPase activity, which is significantly enhanced in the presence of both everted membrane vesicles (membrane ATPase) and presecretory proteins (translocation ATPase) (48). The endogenous, membrane, and translocation ATPase activities of the purified wild-type and mutant SecA were assayed (Fig. 2). The ATPase activities of the wild-type SecA protein were similar to those reported previously (40). SecA D133N showed significantly lower ATPase activities, especially translocation ATPase activity. These results indicate that Asp-133 plays an important role in SecA ATPase activity in relation to translocation.
SecA D133N Activity in in Vitro Protein Translocation-The SecA D133N activity in in vitro protein translocation was determined using three radiolabeled model preproteins, proOmpF-Lpp, wild-type proOmpA, and proOmpA-D26 (Fig.  3). Urea-treated everted membrane vesicles and translated precursor proteins, which both exhibited no in vitro protein translocation activity in the absence of wild-type SecA, were used (40). The SecA D133N protein exhibited no translocation activity. These results demonstrate that the carboxyl group of Asp-133 of E. coli SecA is an essential residue for SecA-dependent protein translocation activity. Consistent with the in vitro results, SecA D133N protein could not complement the secretion defect of E. coli MM66 at the nonpermissive temperature (data not shown), suggesting that the Asp-133 is critical for SecA function in vivo.
SecA D133N Can Interact with ATP, Presecretory Protein, Everted Membrane Vesicles, and Phospholipids-The sensitivity of SecA to protease V8 of S. aureus was markedly affected in the presence of either ATP, presecretory proteins, membrane vesicles, or phospholipids (9). A similar experiment was performed with SecA D133N (Fig. 4). The sensitivities of the wild-type SecA and SecA D133N proteins to V8 were the same as each other (Fig. 4A). ATP, ADP, and ATP␥S render the amino-terminal 95-kDa portion of the wild-type SecA highly resistant to V8. This is consistent with the previous finding that ATP interacts with the amino-terminal 25-kDa portion, the high affinity nucleotide binding site, of the SecA molecule (7). In the presence of ATP, in analogy with the wild-type SecA, the mutant SecA became resistant to V8 digestion over a wide concentration range of ATP, as in the case of the wild-type SecA (Fig. 4, B and C). ATP␥S and ADP were as active as ATP in rendering the mutant resistant to V8 digestion (data not shown). In the presence of proOmpA or E. coli phospholipids, on the other hand, SecA became more sensitive to V8 digestion. SecA D133N was also digested as rapidly as the wild-type SecA was digested (Fig. 4, D and E). Therefore, the conformational changes of the mutant SecA induced by these nucleotides, preproteins, everted membrane vesicles, or phospholipids were essentially the same as those of the wild-type SecA.
The direct interaction of SecA D133N with secretory proteins was also demonstrated by means of chemical cross-linking with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide using proOmpF-Lpp (49). Cross-linking with the mutant SecA to the same extent as that with the wild-type SecA was observed (data not shown). Our previous report (9) demonstrated that the V8 digestion profile after successive treatment with ATP and proOmpA was essentially the same as that with proOmpA alone, and we concluded that this is due to the release of ADP/ATP from the SecA molecule upon interaction with proOmpA. SecA D133N behaved in a similar manner (Fig. 4F).
In the presence of urea-treated membrane vesicles, the digestion patterns of the wild-type SecA and SecA D133N were about the same as each other (Fig. 4G). Taken together, these data suggest that SecA D133N can interact with ATP, pres- ecretory proteins, everted membrane vesicles, and phospholipids, in the same fashion as the wild-type SecA.
Binding of SecA D133N and Wild-type SecA to Cytoplasmic Membranes-SecA binds to E. coli inverted membrane vesicles at the site comprising SecY/E/G, the membrane-embedded portion of the translocation machinery, and acidic phospholipids (45,50). SecA also binds to membrane phospholipids. This may be due to the low affinity association with phospholipids (8). Using 14 C-labeled wild-type SecA, which is fully active (51), and SecA D133N, the binding of SecA to urea-treated everted membrane vesicles of E. coli was assayed. The mutant and wild-type SecAs showed similar binding affinities to the mem-brane (Fig. 5). These data show that SecA was not affected in its binding ability as to SecY/E/G by the replacement of Asp-133 with Asn.

SecA D133N Is Functional in the Initial
Step of Translocation, Including Processing of the Signal Peptide-SecA undergoes ATP-modulated cycles of membrane insertion and deinsertion (52). In the early stage of presecretory protein translocation, SecA is assumed to use the energy of ATP binding to insert a precursor protein into the membrane so as to expose the signal cleavage site to the signal peptidase (53). 35 S-Labeled proOmpA was mixed with inverted membrane vesicles in the presence or absence of SecA and various nucleotides (Fig. 6). Proteolytic processing of proOmpA to OmpA due to the signal peptidase took place in the presence of AMP-PNP as well as ATP, demonstrating that at least a small N-terminal part of the proOmpA molecule can be translocated across the membrane, thereby becoming accessible to the signal peptidase, without using the energy of ATP hydrolysis, as reported previously (Fig. 6A). This view is consistent with that previously reported (53). When samples were further incubated with proteinase K at 0°C to digest polypeptide domains that had not been translocated into the membrane vesicles, translocation intermediates of proOmpA were detected only in the presence of ATP but not in that of a nonhydrolyzable analog (53) (Fig.  6B). Similar experiments were performed with SecA D133N. The processing of proOmpA by the signal peptidase is similar to that with the wild-type SecA (Fig. 6A). On the other hand, neither ATP nor its analog could support the translocation (Fig.  6B). These data clearly demonstrate that SecA D133N is able to bind ATP and thereby support the initial step of translocation to facilitate the processing of proOmpA by the signal peptidase. However, it retains no ability to promote the translocation reaction further.
SecA D133N Interferes with in Vitro Protein Transloca- tion-To determine the functional impact of SecA D133N on the membrane during protein translocation, the effect of SecA D133N on the SecA-dependent translocation of 35 S-labeled proOmpA into everted membrane vesicles was examined in vitro in the presence of 100 M (Fig. 7A) or 10 M ATP (Fig. 7B). The translocation of proOmpA was efficiently inhibited as the amount of mutant SecA was increased (Fig. 7, A and B). It seems likely that the mutant SecA competes with the wild-type SecA for binding to the SecY/E/G translocation complex.
When a higher concentration of ATP (1 mM) was added to the reaction mixture, the interference described above was not observed at all (Fig. 7C).

DISCUSSION
The N-terminal region of SecA contains two nucleotide-binding sequence motifs that are found in a wide range of nucleotide-binding proteins; they are motif A (GXXXXGKT) and motif B (ZZZZD). Between the two motifs, there is a conserved Glu or Asp, which we proposed as a "catalytic residue" (22). We demonstrated here that the replacement of Asp-133, the putative catalytic residue of E. coli SecA, with Asn resulted in nearly complete inactivation of the in vitro presecretory protein translocation reaction (Fig. 3). This indicates the essentiality of the carboxyl group of SecA Asp-133 in translocation. SecA D133N exhibited about 50% of the endogenous and membrane ATPase activities, respectively, and about 20% of the translocation ATPase activity (Fig. 2). SecA has been shown to possess three (48) or two (19) ATP-binding sites. The remaining ATPase activity may be due to the hydrolytic reaction catalyzed by other ATP binding site(s). However, the large decrease in the translocation ATPase activity suggests that the amino-terminal ATP-binding site, compared with other ATP-binding sites, is more closely connected with the translocation reaction.
Although SecA D133N was defective in the overall protein translocation reaction, SecA D133N retained partial functionality and was able to interact with the membrane components of the translocation machinery as follows. (i) Conformational changes of the mutant SecA caused by interactions with ATP, preproteins, or everted membrane vesicles were the same as that of the wild type, as revealed by the V8 protease digestion pattern (Fig. 4). (ii) The mutant SecA retained the ability to bind ATP significantly and performed the initial stage of translocation so as to carry preproteins to a site accessible to signal peptidase (Fig. 6, also see Ref. 49), whereas translocated species were not observed (Fig. 6B). (iii) The purified SecA D133N blocked in vitro translocation by occupying SecY/E/G (Figs. 5 and 7) with low ATP concentrations. Taking the results together, we conclude that SecA D133N, which has a very low translocation ATPase, is able to perform the initial step of the FIG. 6. ATP binding to the wild-type SecA and SecA D133N permits the proteolytic processing of proOmpA. 35 S-Labeled proOmpA-SecB, membrane vesicles, and the wild-type (wt) or SecA D133N were preincubated for 2.5 min at 37°C. Samples were further incubated for 10 min with 10 M ATP with the addition of an ATPregenerating system, 5 mM AMP-PNP, 5 mM ATP, or nothing. A, half of each sample was analyzed by SDS-polyacrylamide gel electrophoresis, followed by fluorography. B, the remaining sample was treated with proteinase K (final, 200 g/ml) for 20 min at 0°C and then analyzed. Translocation intermediates are indicated by asterisks. FIG. 7. SecA D133N interferes with in vitro translocation of proOmpA. In vitro translocation reactions were performed in the presence of 40 g/ml of the wild-type SecA and the indicated amounts of SecA D133N. The concentrations of ATP are indicated. After 30 min, samples were treated with proteinase K (final, 200 g/ml) for 10 min on ice, recovered by trichloroacetic acid precipitation, and then analyzed by SDSpolyacrylamide gel electrophoresis. Translocation intermediates are indicated by asterisks. translocation until the signal peptide region reaches a site accessible to signal peptidase. This idea is consistent with that proposed by Wickner (3), who studied with nonhydrolyzable ATP analogs.
When the ATP concentration was lower (10 -100 M), wildtype SecA-driven translocation was inhibited by SecA D133N (Fig. 7). This suggests that with low concentrations of ATP, SecA D133N can mediate the translocation pathway to the signal peptide processing step. However, under the standard conditions (ATP concentration, 1 mM) even an excess amount of SecA D133N did not interfere with the wild-type SecA driven proOmpA translocation. The latter situation probably occurs in vivo, because the SecA D133N protein was expressed in vivo without CK4706 cell growth being affected (data not shown). At this stage of our studies, the reason why the translocation inhibition only takes place with a low concentration of ATP is not clear. SecA D133N may be somehow displace by wild-type SecA efficiently at the SecY/E/G site when there is a high concentration of ATP.
In conclusion, the conserved Asp residue in E. coli SecA (Glu in SecA from some species) is involved in the catalytic function of ATP hydrolysis and translocation activity. The marked decrease in ATPase activity by the replacement of the acidic aspartate residue with the cognate amine, asparagine, at position 133 is consistent with our idea that this carboxyl group activates a water molecule which then attacks the ␥-phosphate of ATP in SecA and other nucleotide-binding proteins, such as the RecA protein and F 1 -ATPase ␤ subunit.