Characterization of the Adenosine Triphosphatase Activity of the Periplasmic Histidine Permease, a Traffic ATPase (ABC Transporter)*

The superfamily of traffic ATPases (ABC transporters) includes bacterial periplasmic transport systems (permeases) and eukaryotic transporters. The histidine permease ofSalmonella typhimurium is composed of a membrane-bound complex (HisQMP2) containing four subunits, and of a soluble receptor, the histidine-binding protein (HisJ). Transport is energized by ATP. In this article the ATPase activity of HisQMP2 has been characterized, using a novel assay that is independent of transport. The assay uses Mg2+ ions to permeabilize membrane vesicles or proteoliposomes, thus allowing access of ATP to both sides of the bilayer. HisQMP2 displays a low level of intrinsic ATPase activity in the absence of HisJ; unliganded HisJ stimulates the activity and liganded HisJ stimulates to an even higher level. All three levels of activity display positive cooperativity for ATP with a Hill coefficient of 2 and aK 0.5 value of 0.6 mm. The activity has been characterized with respect to pH, salt, phospholipids, substrate, and inhibitor specificity. Free histidine has no effect. The activity is inhibited by orthovanadate, but not byN-ethylmaleimide, bafilomycin A1, or ouabain. Several nucleotide analogs, ADP, 5′-adenylyl-β,γ-imidodiphosphate, adenosine 5′-(β,γimino)triphosphate, and adenosine 5′-O-(3-thio)triphosphate, inhibit the activity. Unliganded HisJ does not compete with liganded HisJ for the stimulation of the ATPase activity of HisQMP2.

The superfamily of traffic ATPases (or ABC transporters) comprises both prokaryotic and eukaryotic transport proteins which share a conserved nucleotide-binding domain (1,2). The superfamily includes bacterial periplasmic permeases, the yeast STE6 gene product, the mammalian P-glycoprotein (multidrug resistance protein or MDR), 1 the human cystic fibrosis transmembrane conductance regulator (CFTR), and the mammalian heterodimeric transporter (TAP1/TAP2) involved in an-tigen processing (3,4). The periplasmic histidine permease in Salmonella typhimurium and the maltose permease in Escherichia coli have been extensively characterized; both are good model systems for this superfamily (5)(6)(7).
The histidine permease is composed of a soluble substratebinding receptor, HisJ (the periplasmic histidine-binding protein), and a membrane-bound complex, HisQMP 2 (8), which contains two integral membrane proteins, HisQ and HisM, and two copies of HisP which carries the ATP-binding motif (9,10). ATP hydrolysis provides the energy for the transport process (11). The histidine permease has been reconstituted into PLS that transport histidine when ATP is trapped internally and HisJ is added externally (6,12). As usually assumed in current working models, the liganded receptor sends a signal that initiates ATP hydrolysis and results in subsequent ligand translocation (13,14).
The characterization of the ATPase activity, which is likely to be common to all members of this superfamily of transporters because of the extensive sequence conservation, is crucial for understanding their mechanism of action. ATP hydrolysis has generally been ascribed to the nucleotide-binding domain. Therefore, several laboratories took the approach of purifying this domain (or component) for the purpose of studying its activity in vitro. MalK, following its solubilization with urea and purification from inclusion bodies, has ATP binding and hydrolyzing activity (15). Other nucleotide-binding components have been purified by a variety of methods and demonstrated to hydrolyze ATP (16,17). However, the activity of the purified subunits in general did not respond to stimulation by the respective soluble receptors. Thus, the interaction with receptors and substrates and the regulation of the ATPase activity must involve the integral membrane components; it is therefore necessary to characterize the properties of the intact membrane-bound complex.
In this paper we describe a detailed study of the ATPase activity of the histidine permease reconstituted in PLS, using a novel assay in which the lipid bilayer has been rendered permeable to both ATP and the receptor. It is shown that HisQMP 2 has an intrinsic low level ATPase activity in the absence of HisJ. The activity is stimulated slightly by unliganded HisJ and to a high level by liganded HisJ. Both the intrinsic and HisJ-stimulated activities display positive cooperativity for ATP. The biochemical characteristics of the activity have been defined.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Media-The following E. coli K12 strains were used: TA1889, which carries deletion unc702 (that eliminates the F 0 F 1 -ATPase) and harbors plasmid pFA17 (which contains the S. typhimurium hisQ, hisM, and hisP genes under the temperaturesensitive control of the P L promoter (18)); GA298 which is derived from TA1889 by loss of pFA17. * This work was supported by National Institutes of Health Grant DK12121 (G.F.-L. A.). 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.
ATP Hydrolysis Assay-Membrane vesicles and PLS were prepared as described earlier (6), from strain TA1889 unless specified differently. ATPase reaction mixtures contained, in a final volume of 330 l, either membrane vesicles (10 l, containing 20 mg/ml protein) or PLS (100 l, containing 0.4 and 10 mg/ml protein and phospholipid, respectively; or 20 l of PLS reconstituted from purified HisQMP 2 , containing 0.2 and 10 mg/ml protein and phospholipid, respectively), and 50 mM MOPS/K ϩ buffer, pH 7.5. Total E. coli phospholipids (3 mg/ml final concentration) were added when the assay was performed with either membrane vesicles or PLS reconstituted with purified HisQMP 2 . HisJ and histidine were added at the indicated concentrations. The reaction mixture was preincubated at 37°C for 3 min in a water bath and the assay initiated by the addition of ATP together with MgSO 4 (2 and 10 mM final, respectively, unless specified differently). At the indicated times, aliquots (100 l) were mixed with equal volumes of 12% SDS. The rate of ATP hydrolysis was determined by assaying the P i liberated (21), using a NaH 2 PO 4 calibration curve, up to 20 nmol of P i .

Permeabilization of PLS Is Necessary to Study the HisJ Stimulation of ATP Hydrolysis by HisQMP 2 -The soluble receptor
HisJ must initiate the ATPase activity of HisQMP 2 in vivo by sending a signal across the membrane bilayer, because HisJ is located in the periplasm and the ATP-binding domain of HisP is located in the cytosol. It has already been shown that if HisJ and ATP are present on opposite sides of the PLS, either with ATP trapped internally and HisJ added externally (12), or vice versa (6), the ATPase activity is stimulated by liganded HisJ. The finding that HisJ can stimulate the activity when present on either side of the PLS is due to the fact that under the conditions of reconstitution HisQMP 2 is embedded in PLS in roughly equal amounts in the right side-out and inside-out orientations (6). Because HisP has been shown to be accessible from the periplasmic side of the membrane bilayer (26), it is possible that external ATP can reach the cytoplasmic ATPhydrolyzing domain of HisP. To test this possibility, PLS containing HisQMP 2 were prepared (6) and their ATPase activity examined when both liganded HisJ and ATP were presented to the outside of PLS. Liganded HisJ does not stimulate significantly the activity (Fig. 1A, columns 1 and 2), indicating that external ATP does not have access to internal hydrolyzing site(s).
To analyze the HisJ-stimulated ATPase activity, HisJ and ATP can be placed on opposite sides of PLS, with either one being trapped internally. However, in both situations there are inherent problems. If ATP is trapped internally and HisJ is added externally, an accurate measurement of initial rates is very difficult because the internal ATP pool is rapidly exhausted 2 and the accumulated ADP inhibits the reaction (6). Alternatively, trapping HisJ inside the PLS and supplying ATP externally (6,14) presents several drawbacks: the procedure for trapping the receptor is cumbersome, a large amount of receptor protein is required, and a sizable fraction of the complexes must be reproducibly inserted in the inverted orientation; furthermore, because only one or two molecules of receptor can be trapped per PLS vesicle, 3 it would be difficult to study the kinetics of the interaction between the receptor and the complex. Therefore, a permeabilizing procedure that renders ATP and HisJ freely accessible to both sides of PLS is desirable. Total membrane solubilization would be a possible approach. We tested numerous detergents, 4 but they all inhibited the ATPase activity.
Mg 2ϩ as a Permeabilizing Agent for PLS-Mg 2ϩ ions, at an appropriate concentration, were found to allow hydrolysis by PLS of externally added ATP in the presence of HisJ, thus suggesting that they permeabilize PLS. This possibility was explored using CF as an indicator of PLS permeability. CF can be trapped inside PLS at high concentration, which causes quenching of its fluorescence; if PLS are permeabilized, the fluorescence would be reactivated when the CF concentration drops as a consequence of its release into the medium (27). Fig.  1B shows that 10 mM MgSO 4 induces 100% release of CF from PLS, which indicates that PLS have been permeabilized. The effect is due to Mg 2ϩ ions rather than to SO 4 2Ϫ ions, as shown by the fact that MgCl 2 permeabilizes as well as MgSO 4 (data not shown), while neither Na 2 SO 4 nor NaCl have any effect. CaCl 2 causes complete CF release at concentrations lower than those of MgSO 4 . 5 The presence of protein is not required for permeabilization because similar results are obtained with liposomes as well (data not shown). ATP and histidine are also released by exposure to 10 mM MgSO 4 , demonstrating that this effect is not limited to CF. Large molecules, such as HisJ, can be released as well from PLS by 10 mM MgSO 4 (up to 80% of the total), as determined by comparing the amounts of HisJ released from PLS before and after treatment with increasing concentrations of MgSO 4 , using SDS-PAGE. When Mg 2ϩ ions are mixed with equimolar concentrations of EDTA or ATP before addition to PLS, no permeabilization is observed, indicating that free Mg 2ϩ ions are necessary for permeabilization.
For the MgSO 4 treatment to be useful for the ATPase assay, its effect should be very rapid. Within 9 s after the addition of 10 mM MgSO 4 , CF release is complete; this is probably an overestimate because it is technically impossible to take sam-ples at earlier times. An analysis of the effect of temperature shows that permeabilization diminishes as the temperature is lowered below 37°C (Fig. 1C) and at 0°C the permeabilization effect is eliminated (for unknown reasons). All experiments in this paper were conducted at 37°C. In conclusion, although the basis of the Mg 2ϩ permeabilizing effect was not understood, this method appeared to be very useful for assaying the ATPase activity in PLS and, as shown later, also in membrane vesicles. Fig. 1D (squares) shows that the use of MgSO 4 as a permeabilizing agent allows assay of ATP hydrolysis by PLS. The rate of hydrolysis in the presence of liganded HisJ is linear (up to at least 30 min). Fig. 1E shows that the initial rate of hydrolysis is directly proportional to the volume of PLS added (with the phospholipid to protein ratio maintained constant at 50:1, w/w), and therefore, to PLS protein and phopholipid concentra-tions. This experiment also shows that a phospholipid concentration of 0.16 mg/ml is compatible with complete permeabilization. PLS were reconstituted with purified HisQMP 2 to determine the effect of varying protein concentration while maintaining a constant phospholipid concentration (3 mg/ml). Fig. 1F shows that the rate of hydrolysis is proportional to protein concentration also in this case, in which the ratio of phospholipid to protein varies from 350:1 to 7000:1 (w/w). Thus, under the standard assay conditions, the activity depends directly on the protein concentration and is not affected by the phospholipid concentration.
The effect of varying the Mg 2ϩ concentration on the rate of hydrolysis is shown in Fig. 1A. When Mg 2ϩ is in excess over ATP and in the absence of HisJ, the hydrolysis rate is 3-fold higher than when they are equimolar (compare columns 1 and To determine the total fluorescence, at completion of the experiment, Triton X-100 was added to the diluted PLS to a final concentration of 0.2% (by volume), thus releasing all of the CF. The percent release of CF that is induced by salts was calculated by comparing the fluorescence intensity before and after the addition of Triton X-100. C, MgSO 4 (10 mM final) was added to CF-containing PLS; after incubation for 3 min at the indicated temperatures, PLS were diluted and the fluorescence measured as in B. D, ATP hydrolysis by PLS in the presence of 2 mM ATP and 10 mM MgSO 4 was determined by the release of P i at various times (squares). In a parallel experiment (open diamonds), EDTA (7.5 mM) was added 2 min after initiation of the reaction. E, various volumes of PLS (0.4 and 10 mg/ml, protein and phospholipid, respectively) were diluted into 50 mM MOPS/K ϩ , pH 7.5, 12 M HisJ, 100 M L-histidine, in a final volume of 320 l, and the initial linear rate of ATP hydrolysis was assayed. F, the experiment is the same as in E, except that the PLS were prepared with purified HisQMP 2 and E. coli phospholipids were added to the final reaction mixture to a final concentration of 3.0 mg/ml.

3).
In the presence of liganded HisJ, the relative increase is 15-fold (compare columns 2 and 4). In both cases, the increase in activity is due to the exposure of ATP to internal ATPhydrolyzing sites by the permeabilization procedure. The larger increase in the presence of liganded HisJ is due to the fact that both internal ATP-hydrolyzing sites and internal HisJ contact sites become accessible to ATP (for hydrolysis) and HisJ (for signaling); thus, the stimulatory effect of liganded HisJ is the factor being measured. The increase in activity in the presence of MgSO 4 is not due to P i being released from PLS because the activity is dependent on the presence of ATP (data not shown). Fig. 1D indicates that ATP is continuously replenished inside PLS, i.e. PLS are permeable throughout the length of the assay. If permeabilization were transient, the activity would rapidly decrease in time because of ATP exhaustion and accumulation of inhibitory ADP (as is the case in non-permeabilized PLS (6)). Alternatively, permeabilization might be transient, while giant PLS might be formed which trap sufficient ATP for sustained hydrolysis. If this were true, chelation of Mg 2ϩ by EDTA should have no effect. However, ATP hydrolysis stops promptly upon addition of EDTA (Fig. 1D, diamonds). The latter experiment also indicates that permeabilization can be reversed.
Multiple Levels of Activity of HisQMP 2 -Three levels of ATPase activity are detected in PLS: 50, 100, and 370 nmol/ min/mg protein in the absence of HisJ, and in the presence of HisJ and liganded HisJ, respectively (Fig. 2). The low level ATP hydrolysis observed in the absence of HisJ may be due either to contaminating ATPases or to an intrinsic activity of HisQMP 2 , or a combination of both. To differentiate between these possibilities, PLS containing various amounts of HisQMP 2 in the presence of identical amounts of the other proteins were prepared. A culture of TA1889 was heat induced for different lengths of time and the concentration of HisQMP 2 in PLS present at each of the induction times was determined by SDS-PAGE and immunoblotting using antibodies raised against HisP and HisQ. Fig. 3 (A and B) shows that, as expected, the amount of HisQMP 2 in PLS increases as the heat induction time increases (lanes 5-8). The ATPase activity in these PLS tested in the absence of HisJ increases in parallel (Fig. 3C, open squares). To determine whether the heat induc-tion of contaminating ATPases contributes to this activity, PLS were prepared from strain GA298 (which contains no HisQMP 2 ) under the same conditions as for TA1889 ( Fig. 3A  and B, lanes 1-4); they were shown to have a constant activity at all heat induction times (Fig. 3C, open circles). Thus, the activity in TA1889 PLS in the absence of HisJ is due to the sum of an activity intrinsic to HisQMP 2 (about 25 nmol/min/mg at 70 min induction time) and that of contaminating ATPases. That the complex has an intrinsic activity was further established by purifying it and showing that, upon reconstitution into PLS, it hydrolyzes ATP (140 nmol/min/mg) in the absence of HisJ (Fig. 3D, squares) That the activity stimulated by liganded HisJ is due to HisQMP 2 (as opposed to a contaminating activity which is stimulated by HisJ) is shown by its dependence on the HisQMP 2 content in PLS (Fig. 3C, solid squares). Purified HisQMP 2 gave the same result (Fig. 3D, diamonds), displaying an activity of 1.05 mol/min/mg protein (an activity of 1.94 mol/min/mg protein was obtained in the presence of a saturating concentration of liganded HisJ (50 M)). The stimulation is specific for HisJ because a number of HisJ mutant proteins were found that are unable to stimulate the ATPase activity. 7 HisQMP 2 also interacts with another soluble receptor, LAO (the periplasmic lysine-, arginine-, ornithine-binding protein) (28), which, therefore, would also be expected to stimulate the ATPase activity. Indeed, both unliganded and L-arginine-liganded LAO stimulate the activity. The affinity of LAO for HisQMP 2 and the maximum level of stimulated activity (purified HisQMP 2 ) were determined to be 5.3 M and 1.75 mol/ min/mg, respectively. These values are very similar to those obtained for HisJ (13).
Positive Cooperativity of ATPase Activity-Purified HisQMP 2 in PLS was used to accurately determine the affinity of HisQMP 2 for ATP, both for the intrinsic activity and the activity stimulated by liganded HisJ. In both cases, the curves are sigmoidal with respect to ATP concentration (Fig. 4), implying a positive cooperativity between ATP-binding sites in HisQMP 2 , with identical Hill coefficients: n app ϭ 1.9 Ϯ 0.05. Thus, two binding sites in HisQMP 2 are involved. The K 0.5 values for ATP are 0.5 and 0.6 mM, respectively, for the activity stimulated by liganded HisJ (8 M) and for the intrinsic activity. The V max values are 1.9 and 0.14 mol/min/mg, respectively, for the activity stimulated by liganded HisJ (11 M) and for the intrinsic activity. To exclude the possibility that the cooperativity is due to the presence of excess Mg 2ϩ , PLS containing internally trapped liganded HisJ were assayed in the presence of external ATP; the activity showed positive cooperativity with a Hill coefficient and K 0.5 value of 2.0 and 1.7 mM, respectively.
Nucleotide Specificity and Inhibitors-Nucleotides other than ATP have been shown to support histidine transport and to bind to HisP (18,29), indicating that HisQMP 2 has a relatively broad specificity for the hydrolysis of nucleotides and for their utilization in energizing transport. The relative rates of hydrolysis of nucleotides at concentrations between 0.2 and 2 mM are in the following order: ATP Ͼ GTP Ͼ UTP Ͼ CTP (Table I). This order of preference is the same as that obtained when nucleotides were tested for the energization of transport (12,29). The ATP analogues, ADP, AMP-PNP, AMP-PCP, and ATP␥S, inhibit the activity (Table I). Vanadate, a potent inhibitor of P-type ATPases (30) and histidine transport (24), inhibits the activity of HisQMP 2 , giving 50% inhibition at 6.5 M. Bafilomycin A 1 (100 M), oubain (up to 3 mM), and NaN 3 (10 mM) do not inhibit. Incubation of PLS for 15 min at 37°C before the assay with 10 mM N-ethylmaleimide has no effect.
Effect of Histidine-It has been frequently hypothesized that the membrane-bound complex of periplasmic permeases carries a substrate-binding site (6,(31)(32)(33)(34). The existence of such a site in HisQMP 2 has been suggested by the properties of receptor-independent mutants (32,35), the inhibition of in vitro transport by histidine trapped inside PLS (6), and the altered substrate specificity of several strains carrying mutations in either HisQ or HisM (31). The function of this postulated site might be to bind the histidine released from liganded HisJ and initiate a series of conformational changes required for trans-7 C. E. Liu, A. Wolf, and G. F-L. Ames, unpublished data.  Ϫ 10 mM N-Ethylmaleimide Ϫ 10 mM a The activity was measured in PLS in the presence of 8 M liganded HisJ as a function of nucleotide concentration. The K 0.5 value is the concentration of nucleotide which results in half the activity obtained in the presence of 2 mM nucleotide. The represents the activity in the presence of 2 mM nucleotide. b IC 50 is the concentration which gives 50% inhibition in the presence of 2 mM ATP. The maximum concentration is the highest concentration tested, which does not inhibit activity. Bafilomycin A 1 was dissolved in ethanol, and the final ethanol concentration in the assay did not exceed 1%.
location, or perhaps to bind directly free histidine to prime HisQMP 2 for interaction with HisJ. The effect of histidine on the rate of hydrolysis was determined as a function of histidine concentration in the presence of a fixed amount of 20 M unliganded HisJ (Fig. 5). The hydrolysis rate increases in parallel with increasing histidine concentrations up to 20 M, after which the activity remains constant. Thus, free histidine has no effect on the ATPase activity (Fig. 5). These results suggest that the substrate-binding site on HisQMP 2 is not involved in priming HisQMP 2 for transport in the presence of HisJ. Free histidine, up to 20 mM, has no effect on the intrinsic activity either (Fig. 5, inset), indicating that free histidine in the medium is unlikely to initiate transport in the absence of HisJ. In conclusion, the ATPase activity of HisQMP 2 (either stimulated by liganded HisJ or the intrinsic activity) is not affected by free histidine.
Effect of NaCl, MgSO 4 , and pH-NaCl is not required for activity and at concentrations higher than 100 mM it inhibits activity, with 50% inhibition at 200 mM NaCl (Fig. 6A). The inhibition profile is similar to that observed for histidine transport (6). MgSO 4 at concentrations between 5 and 10 mM gives the highest activity, with concentrations higher than 10 mM being inhibitory (Fig. 6B). The pH optimum for the activity is broad, between 7.5 and 8.5 (Fig. 6C). This optimum may represent an average between the optimum pH values for the HisJ-HisQMP 2 interaction and for ATP hydrolysis, because the pH optimum for the ATPase activities of several HisJ-independent HisQMP 2 mutant complexes is 8.5 8 and the pH optimum for histidine transport is about 7.5 (6,24,36).
Effect of Phospholipids-The effect of various phospholipids on the activity of HisQMP 2 was tested. Membrane vesicles were used in this case since they contain relatively low levels of endogenous phospholipids, thus minimizing interference by endogenous lipids. Membrane vesicles are readily permeabilized by our procedure. Fig. 7 shows that the activity increases with the addition of total E. coli phospholipids, reaching a maximum at 2 mg/ml, with a phospholipids to protein ratio of about 50:1 (w/w). Table II shows that, except for PE, individual phospholipids are inhibitory at high concentrations, with the strongest inhibition being given by lysophosphatidic acid. Their inhibitory effect is relieved (with the exception of high concentrations of cardiolipin, lysophosphatidylethanolamine, and lysophosphatidic acid) when tested in combination with PE. None of the combinations results in an activity higher than that in the presence of total E. coli phospholipids.
Unliganded HisJ Does Not Compete with Liganded HisJ in the Stimulation of ATPase Activity-It was reported previously that unliganded HisJ interacts with HisQMP 2 with an affinity similar to that of liganded HisJ, as measured by cross-linking experiments, stimulation of ATPase activity (13), and inhibition of histidine transport (24). Therefore, it was expected that unliganded HisJ would compete with liganded HisJ also in the 8 A. Wolf, personal communication. stimulation of ATPase activity. It was predicted that addition of unliganded HisJ would inhibit significantly the rate of ATP hydrolysis as stimulated by liganded HisJ, because the former stimulates a much lower level of activity. Fig. 8 shows that the initial rate increases with increasing concentrations of unliganded HisJ, when either none, 5 M, or 10 M histidine are present. In the absence of histidine the initial rate increase is small, as expected for the stimulation by unliganded HisJ (Fig.  2) (13). In the presence of histidine a rapid increase is observed until the HisJ concentration equals that of histidine, after which the initial rate increases moderately, reaching values that are the sum of the rates as stimulated by the liganded and unliganded HisJ present. Because the amount of HisQMP 2 is limiting, it appears that there is no competition between unliganded and liganded HisJ in the stimulation of ATP hydrolysis. No competition was observed when the concentration of unliganded HisJ was increased up to 180 M.

DISCUSSION
In this study we show that Mg 2ϩ ions allow the simultaneous presentation of molecules of a variety of sizes to the inner and outer surfaces of membrane vesicles and PLS. Divalent cations induce liposome fusion and the loss of encapsulated material into the medium (37). This leakage has been ascribed to the formation of the hexagonal (H II ) phase in the phospholipids (38). Although Ca 2ϩ is generally more effective than Mg 2ϩ in inducing fusion, Mg 2ϩ is quite effective in liposomes containing PE (37). This explains the results presented here using PLS formed from E. coli phospholipids which contain a high percentage of PE (39). We observed that PLS suspensions become cloudy upon addition of 10 mM MgSO 4 , indicating aggregation, as has been reported before in fusion experiments (38). However, a potent liposome aggregation agent, protamine (40,41), induced negligible ATP release from PLS (data not shown), indicating that the Mg 2ϩ -induced leakage has specific properties. An alternative explanation for the Mg 2ϩ -induced permeability is that PE is induced to organize itself in a lipidic cubic phase, which allows the penetration of polar molecules (42).
The method we present here overcomes many of the technical difficulties inherent in the study of the membrane-bound ATPases of periplasmic permeases. In addition to the already mentioned advantage of making all ATP-binding sites, extraand intravesicular, accessible to both ATP and HisJ, the permeabilization is likely to eliminate the multilamellarity of PLS; the overall effect is that of increasing the number of available functional sites. The accessibility of the signaling sites to the total HisJ pool, rather than to a low number of trapped molecules, 3 makes it possible to study the kinetic aspects of the stimulation. The method has numerous additional advantages. (i) It is simple and rapid. (ii) It gives very reproducible results. (iii) It can be used for both membrane vesicles and PLS, thus allowing the activity to be followed through an entire purification scheme.
HisQMP 2 displays three levels of ATPase activity, depending on the presence and ligation state of the receptor, HisJ. The fully stimulated activity of HisQMP 2 is about 10-fold higher than the intrinsic activity. The possibility that the intrinsic activity reflects an artifact caused by the solubilization and reconstitution procedures is excluded because HisQMP 2 in native membrane vesicles also displays intrinsic activity. The regulatory/signaling action of HisJ may operate through HisQ and/or HisM by one of two basically different mechanisms: either by relieving a repressed catalytic state of HisP or by   positively inducing its activity. The first of these alternatives is more likely because HisP isolated in the absence of HisQ and HisM has an activity considerably higher than the intrinsic activity of HisQMP 2 , 9 suggesting that HisQ and/or HisM may be exerting the repression.
The stimulated ATPase activity determined here for PLS is 40-fold higher than previously reported: 370 versus 9 nmol/ min/mg of protein (12). The higher hydrolysis rate is presumably due to better accessibility of ATP, as compared with assay systems in which ATP is trapped internally, in combination with improved handling of the preparations. This higher value may represent the full hydrolyzing capability of HisQMP 2 since it is comparable to the highest values measured for hisP mutants that have signal-independent activities (32,43). Similar levels of stimulated activity have been reported for the maltose transport complex (reconstituted from solubilized membrane vesicles in PLS): 390 nmol/min/mg of protein (14). Purified HisQMP 2 hydrolyzes ATP with a V max of 2100 nmol/min/mg of protein (Fig. 4), which is calculated to correspond to a turnover rate of 8 s Ϫ1 in the presence of saturating liganded HisJ. PLS transport histidine at a rate of 8 nmol/min/mg of protein, which can be corrected to 16 nmol/min/mg, since half of the complexes are inserted in the inverted orientation; in addition, the transport rate is improved 4-fold (64 nmol/min/mg) if an ATP-regenerating system is included in the PLS (6). Therefore, a potential ratio of ATP hydrolyzed to histidine transported can be calculated to be 6. However, since a transport rate of 40 nmol/ min/mg has been observed in the absence of an ATP-regenerating system, PLS have the potential to reach a transport rate of 160 nmol/min/mg in the presence of an ATP-regenerating system. The latter rate gives a ratio of ATP hydrolyzed to histidine transported of 2.3, which may be a reasonable estimate of the actual stoichiometry.
Both the stimulated and intrinsic activities display positive cooperativity for ATP, suggesting an interaction between the two identical HisP subunits; the interaction, therefore, must be independent of HisJ signaling. The two subunits may both hydrolyze ATP, or only one may do so, with the other one only binding ATP and enhancing the activity. The notion that the two subunits of HisP interact with each other within the complex is consistent with recent evidence that the active form of soluble HisP is a dimer. Since the two subunits of HisP in HisQMP 2 have identical sequences and no post-translational modification has been observed, it is possible that either subunit has the potential to bind ATP first and facilitate binding and/or hydrolysis by the other subunit. The K 0.5 value for HisQMP 2 of 0.6 mM for both the intrinsic and stimulated activities is compatible with the value estimated from in vivo histidine transport assays (36) and with that obtained for the maltose permease (44) and MDR (45,46).
The finding that exogenously added phospholipids enhance the activity as assayed in membrane vesicles (Fig. 7) may reflect a more efficient permeabilization process, by increasing the phospholipid/protein ratio. The inhibition of activity by several individual phospholipids may reflect a decrease in Mg 2ϩ -mediated vesicle fusion due to a reduced relative concentration of PE in the bilayer; alternatively, it reflects a loss in activity which might require a particular combination of phospholipids. Since PE is the dominant phospholipid in E. coli (39), its addition causes a relatively small change in phospholipid ratios and does not result in inhibition. The strong inhibition by lysophospholipids may reflect an inhibitory detergent-like action, a sensitivity common with that of other traffic ATPases (MDR (46), HlyB (16), and the maltose permease (44)).
The lack of competition by unliganded HisJ for liganded HisJ-stimulated ATP hydrolysis is unexpected. A possible explanation may be that, even though the unliganded form replaces liganded HisJ on HisQMP 2 , it has a chance of capturing a histidine molecule while still interacting with the complex and, consequently, of stimulating ATP hydrolysis. This may be possible if the HisJ-HisQMP 2 complex is long-lived as compared with the rate of redistribution of histidine molecules among the HisJ molecules and the signaling process is very rapid.
The ATPase activity of HisQMP 2 shares with other members of the superfamily several features, which, therefore, may be characteristic of the superfamily. An intrinsic activity has also been shown for MDR (47). Cooperativity with respect to ATP has been demonstrated also for the maltose permease (44) and possibly for MDR (as the early points in Fig. 3 in Ref. 46 might suggest). The activity is sensitive to vanadate (a P-type ATPase inhibitor) in the case of HisQMP 2 (this paper and Ref. 24), the maltose permease (44), and MDR (45,46); HisQMP 2 and MDR are insensitive to ouabain (a Na ϩ K ϩ -ATPase inhibitor), azide (a F 0 F 1 -ATPase inhibitor) (47,45). 10