Photoinactivation of Fluorescein Isothiocyanate-modified Na,K-ATPase by 2′(3′)-O-(2,4,6-Trinitrophenyl)8-azidoadenosine 5′-Diphosphate

The Na,K-ATPase activity of the sodium pump exhibits apparent multisite kinetics toward ATP, a feature that is inherent to the minimal enzyme unit, the αβ protomer. We have argued that this should arise from separate catalytic and noncatalytic sites on the αβ protomer as fluorescein isothiocyanate (FITC) blocks a high affinity ATP site on all α subunits and yet the modified Na,K-ATPase retains a low affinity response to nucleotides (Ward, D. G., and Cavieres, J. D. (1996) J. Biol. Chem. 271, 12317–12321). We now find that 2′(3′)-O-(2,4,6-trinitrophenyl)8-azido-adenosine 5′-diphosphate (TNP-8N3-ADP), a high affinity photoactivatable analogue of ATP, can inhibit the K+-phosphatase activity of the FITC-modified enzyme during assays in dimmed light. The inhibition occurs with aK i of 140 μm at 20 mmK+; it requires the adenine ring as 2′(3′)-O-(2,4 6-trinitrophenyl) (TNP)-UDP or TNP-uridine are less potent and 2,4,6-trinitrobenzene-sulfonate is ineffective. Under irradiation with UV light, TNP-8N3-ADP inactivates the K+-phosphatase activity of the fluorescein-enzyme and also its phosphorylation by [32P]Pi. The photoinactivation process is stimulated by Na+ or Mg2+, and is inhibited by K+ or excess TNP-ADP. In the presence of 50 mm Na+ and 1 mm Mg2+, TNP-8N3-ADP photoinactivates with a K 0.5 of 15 μm. Furthermore, TNP-8N3-ADP photoinactivates the FITC-modified, solubilized αβ protomers, even more effectively than the membrane-bound fluorescein-enzyme. These results strongly suggest that catalytic and allosteric ATP sites coexist on the αβ protomer of Na,K-ATPase.

The activity of Na,K-ATPase (EC 3.6.1.37) depends on the cytoplasmic ATP concentration in a complex nonhyperbolic manner (1)(2)(3)(4). Micromolar concentrations of ATP are sufficient to achieve maximal steady-state phosphorylation of the enzyme (K 0.5 Ͻ 1 M) and yet, in the presence of K ϩ , higher ATP concentrations further stimulate ATP hydrolysis more than 20-fold (K 0.5 Ϸ 100 -500 M). This stimulation seems to arise, at least in part, from an acceleration of K ϩ release to the cytosol (5-7) and can also be elicited by nonhydrolyzing ATP analogues (8). Low affinity ATP effects are also observed with several partial reactions of the E2 state of the enzyme, including the K ϩ -activated phosphatase activity (4,9), phosphorylation by P i (10), release of bound ouabain (11), and K ϩ -K ϩ exchange (12); in the latter case, ADP is also effective (13). Besides, a low affinity acceleration of the rate of ATP phosphorylation of Na,K-ATPase has been observed (14).
Na,K-ATPase consists of protomers containing one 112-kDa ␣, or catalytic, subunit and one 35-kDa ␤ subunit of unclear function (15)(16)(17). The ␣␤ protomers may be organized into dimers or higher oligomers in the membrane (8). It seems clear that there is one high affinity ATP binding site per ␣ subunit (18 -21). However, where ATP binds to produce the low affinity regulatory effect has been the subject of much controversy, and the issue has been further complicated by uncertainties about the oligomeric structure of the membrane-bound enzyme. We have previously used C 12 E 8 1 (22) to solubilize purified Na,K-ATPase as fully active ␣␤ protomers, and have found that these retain the dual responses to nucleotides in the absence of oligomerization (4). The implication is that the two ATP effects are intrinsic to the protomeric enzyme, i.e. that an interaction between ␣ subunits is not required for the low affinity ATP binding. Two possible mechanisms, which need not be exclusive, could account for the high and low affinity ATP effects on the protomer: 1) multiple nucleotide interactions at a single site, at different stages of the reaction cycle, and 2) the existence of an allosteric, or regulatory, nucleotide site on the ␣␤ protomer, which has not been identified so far because of the difficulties associated with its low affinity.
The experiments presented in this report and in a previous publication (23) were designed to explore the second option. Our rationale was that judicious block of the high affinity ATP site in every ␣ subunit should also abolish low affinity nucleotide effects if only option 1 was applicable. Conversely, if low affinity effects of ATP, or ATP analogues, were demonstrable in Na,K-ATPase thus modified by a discriminating agent, then they should arise from binding at an allosteric site. We chose FITC as the high affinity site blocker. Under alkaline conditions FITC inactivates Na,K-ATPase from within the adenine binding subdomain of the high affinity site, forming a covalent bond with the ⑀-amino group of one of three lysines of the ␣ chain (24 -26). The FITC modification abolishes high affinity binding of formycin triphosphate (a fluorescent ATP analogue), phosphorylation from ATP, and the overall reaction cycle (27). However, the fluorescein-enzyme can bind ouabain in the presence of Mg 2ϩ , can be phosphorylated by P i , and retains its K ϩ -stimulated phosphatase activity (27).
We recently reported (23) that when FITC-modified enzyme was solubilized with C 12 E 8 under nondenaturing conditions, the resulting soluble protomers were still unable to hydrolyze ATP. The K ϩ -phosphatase activity of the FITC-modified ␣␤ protomers was similar to that of the parent membrane-bound enzyme and, crucially, it could be inhibited by TNP-ADP, a tight-binding ATP analogue. We concluded that the nucleotide had to bind at a locus different from the catalytic site (23).
In a bid to incorporate an affinity probe in the available site, we have now found that TNP-8N 3 -ADP can photoinactivate FITC-modified ␣␤ protomers after binding at equilibrium. It follows that FITC and TNP-8N 3 -ADP must be covalently attached to the same minimal enzyme unit. It then seems quite likely that TNP-8N 3 -ADP is blocking an allosteric ATP site on the protomeric fluorescein-enzyme. A preliminary communication has been presented (28).
Enzyme Purification-Na,K-ATPase was purified from pig outer renal medulla with the zonal rotor method of Jørgensen (29). The Na,K-ATPase activity ranged from 20 to 35 units/mg of protein at 37°C. (One enzyme unit ϭ 1 mol⅐min Ϫ1 ).
Na,K-ATPase Activity-Na,K-ATPase activity was measured using the coupled-enzyme spectrophotometric assay at 37°C. The reaction was run in 1 ml and followed at 340 nm and at 40-s intervals for 4 min. Conditions were: 148 mM NaCl, 25 mM histidine (pH 7.4), 15 mM KCl, 3 mM MgCl 2 , 2 mM ATP, 1 mM phosphoenolpyruvate, 150 M NADH, 10 unit each of pyruvate kinase and lactate dehydrogenase.
K ϩ -Phosphatase Activity-With either the membrane-bound or the C 12 E 8 -solubilized enzymes, the K ϩ -phosphatase activity was determined as linear time courses at 20°C rather than at 37°C, because of the limited thermal stability of the soluble protomers (22). Enzyme aliquots were mixed into an assay medium designed to give 20 mM KCl, 20 mM Tris (pH 7.5), 5 mM MgCl 2 , and 2-20 mM pNPP. Samples of the reaction mixture were quenched at intervals with 1 M NaOH, and the absorbance was determined at 410 nm. The rate of pNP release was determined from pNP calibration curves.
Phosphorylation-Phosphorylation from inorganic phosphate was carried out in a total volume of 80 l containing 10 g of Na,K-ATPase, 20 mM Tris (pH 7.5), 2 mM MgCl 2 , 1 mM EDTA, 1 mM ouabain, and 50 M [ 32 P]P i . Phosphorylation was allowed to proceed for 15 min at 20°C and was then quenched with 500 l of a chilled solution containing 7% trichloroacetic acid and 5% phosphoric acid. The acid-denatured membranes were trapped on cellulose nitrate filters (Sartorius, 25 mm ϫ 0.45 m) and washed six times with 5 ml of chilled 5% trichloroacetic acid, 5% phosphoric acid. The filters were transferred to counting vials, scintillation fluid was added, and 32 P disintegrations/min were determined. Blanks were run using boiled enzyme and deducted (less than 5% of total phosphorylation). Phosphorylation from ATP was carried out on ice by mixing 20 l of Na,K-ATPase suspension (0.5 mg/ml) into 80 l of a chilled solution providing (final concentrations): 150 mM NaCl, 20 mM Tris (pH 7.5), 2 mM MgCl 2 , 1 mM EDTA, 5 M [␥-33 P]ATP. Following a 10-s incubation at 0°C, the reaction was stopped with a chilled solution containing 7% trichloroacetic acid, 10 mM Na 4 P 2 O 7 , and 10 mM NaH 2 PO 4 . The samples were filtered and washed with chilled 5% trichloroacetic acid, 10 mM Na 4 P 2 O 7 , 10 mM NaH 2 PO 4 , 1 mM ATP.
ATP Binding-Membrane-bound Na,K-ATPase was suspended at 250 g/ml in 20 mM Tris (pH 7.5), 1 mM EDTA, 2 M [ 3 H]ATP and incubated at 20°C for 10 min. The free [ 3 H]ATP was determined in samples of the suspension obtained by suction through 0.22 M cellulose nitrate syringe-tip filters (Star, from Costar). Bound [ 3 H]ATP was calculated from the difference between the radioactive concentration of the original suspension and the radioactive concentration of the filtrate and was referred to the protein concentration.
FITC Treatment-Membrane-bound Na,K-ATPase was suspended at 1 mg of protein/ml in 150 mM NaCl, 50 mM Tris (pH 9.2), and 5 mM EDTA. FITC was prepared as a 10 mM stock solution in dimethyl sulfoxide and added to give a final concentration of 50 M. The enzyme/ FITC mixture was incubated at 20°C for 30 min in the dark, and the membranes were spun down, washed, and resuspended in fresh buffer. Control enzyme was processed in parallel, with the omission of FITC. In all cases, the FITC-treated enzyme retained less than 2% of the Na,K-ATPase activity of the parallel control. On average, 80% of the pNPPase activity was preserved. In the experiment of Fig. 2, the enzyme concentration was increased to 2.4 mg/ml, and the FITC concentration decreased to 20 M to facilitate observation of the time course of inactivation.
C 12 E 8 Solubilization-Equal volumes of detergent solution (2.4 mg of C 12 E 8 /ml of H 2 O) and enzyme suspension (800 g of protein/ml in 100 mM NaCl, 40 mM Tris (pH 7.5), 2 mM EDTA) were mixed and incubated at 20°C for 30 min. Insoluble material was removed by centrifuging at 356,000 ϫ g for 15 min (Beckman TL-100 ultracentrifuge, TLA 100.2 rotor). The supernatant was transferred to fresh tubes, stored on ice, and used the same day.
TNP Derivative Syntheses-TNP-8N 3 ADP was synthesized according to Seebregts and McIntosh (30). The reaction mixture comprised 216 l of an aqueous solution containing 90.8 mg of Na 2 CO 3 /ml and 62.9 mg of NaHCO 3 /ml, and 173 l of a 5% solution of TNBS containing 18 mg of 5,5Ј-dithiobis(2-nitrobenzoic acid). This mixture was added to 5 mg of solid 8N 3 -ADP (or UDP, uridine, or adenosine, as required) and left stirring overnight at room temperature and in the dark. The product was purified by reversed-phase high performance liquid chromatography on a C 18 column (250 ϫ 4.6 mm, Hichrom). A linear gradient (0 to 100% methanol) was delivered over 60 min using two Pharmacia-LKB 2248 high performance liquid chromatography pumps with a Pharmacia-LKB LCC 2252 controller. The optical density of the eluate was followed with a Pharmacia-LKB VWM 2141 dual wavelength monitor. TNP-8N 3 -ADP peaked at 34 min; it was collected, lyophilized, dissolved in 20 mM Tris (pH 7.5), 1 mM EDTA and stored at Ϫ80°C. The yield was approximately 20%. All TNP-derivatives had the correct UV-visible absorption spectrum, TNP-8N 3 -ADP with maxima at 280, 408, and 468 nm (30), TNP-adenosine at 259, 408, and 468 nm, and TNP-UDP and TNP-uridine at 262, 408, and 468 nm. TNP-8N 3 -ADP and TNP-UDP also had the correct A 408 /acid-labile phosphate ratio (⑀ M /P i ϭ 27.6 cm Ϫ1 ⅐ mM Ϫ2 ) (30). Each product migrated as a single peak when reapplied to the C 18 column. Acid-labile phosphate was determined by the method of Fiske and Subbarow (31), following 10 min of boiling in 1 M HCl.
Photoinactivation-The enzyme was suspended in 50 mM NaCl, 20 mM Tris (pH 7.5), 2 mM MgCl 2 and 1 mM EDTA (unless otherwise specified in the legends to the figures). The suspension, or the soluble enzyme, was pipetted into multiwell dishes to a depth of 2-3 mm and irradiated using a Flowgen VL-6MC UV lamp (6 W output, 312 nm setting) at a distance of 50 cm, inside a chamber at room temperature and constant humidity.
Protein Assay-Protein was precipitated with 7% trichloroacetic acid and assayed with the bicinchoninic acid method (32) as described previously (4).
Curve Fitting and Errors-All enzyme activities were calculated from the regression coefficients of the least squares linear regression of product release time courses. Error bars represent the S.E.s on these regression coefficients and have been compounded where appropriate. Curves through photoinactivation time course data are single exponential functions with a variable infinity value, fitted using SigmaPlot 4.0. The error bars on the reciprocal data of the Dixon plot of Fig. 3 were calculated from the S.E. divided by the square of the original mean (33). The linear regressions of the reciprocal data of Fig. 3 were done with weights according to Wilkinson (33).

RESULTS
Considering the effectiveness of TNP-ADP as an inhibitor of the K ϩ -phosphatase activity of the FITC-modified enzyme (23,34), we decided to try TNP-8N 3 -ADP as a photoactivatable probe for the site of low affinity nucleotide binding in Na,K-ATPase. TNP-8N 3 -ADP has been used as a probe for the catalytic site of the Ca,Mg-ATPase of the sarcoplasmic reticulum (35) but not with Na,K-ATPase, to our knowledge. Therefore, we first observed the photoinactivation by TNP-8N 3 -ADP of the Na,K-ATPase activity of native sodium pump and the result of this experiment is shown in Fig. 1A. With 5 M TNP-8N 3 -ADP, more than 60% of the enzyme activity is lost within 20 min. On the other hand, in the absence of TNP-8N 3 -ADP, the enzyme appears to be reasonably stable in our irradiation conditions. The inactivating effect is more than half-maximal at 0.05 M TNP-8N 3 -ADP, and can be abolished by addition of 100 M TNP-ADP. The inactivation presumably occurs following reversible binding to the catalytic ATP site, due to covalent attachment of the reactive nitrene derivative generated by UV activation of the azido group (36), as is the situation with 8N 3 -ATP or 2N 3 -ATP (18,19,37,38). As it happened with other probes, the time courses of inactivation were complex; they could be best fitted as single exponential decays, but with a non-zero infinity value.
We decided on a protection strategy to discriminate in favor of a low affinity ATP site. The FITC block had earlier proved successful with the membrane-bound (34) and the protomeric (23) enzymes, and we wished to see whether other ATP site blockers could be just as effective and selective. From our earlier results (23), we should expect that low concentrations of 8N 3 -ATP be less deleterious to the K ϩ -phosphatase activity than to high affinity ATP binding (and the overall reaction). This was substantiated by the result of Fig. 1B, which shows that 8N 3 -ATP inactivates the Na,K-ATPase activity much more drastically than the K ϩ -phosphatase activity. We also con-firmed the results of Kaya et al. (39), that pyridoxal phosphate treatment leaves an intact K ϩ -phosphatase activity although it inactivates as much as 80% of the Na,K-ATPase activity, and found that either 8N 3 -ATP or pyridoxal phosphate treatment caused a 60 -70% increase in the enzyme's K 0.5 toward pNPP (not shown). However, with this protocol these two probes seem unsuitable for the present purpose as about 25% of the Na,K-ATPase survives even after long times at 20 M 8N 3 -ATP or 200 M pyridoxal phosphate, making it difficult to achieve a neat separation of nucleotide effects. We decided therefore to retain FITC as the high affinity site blocker, in the knowledge that it can cause nearly 100% inactivation of the Na,K-ATPase activity with little loss of K ϩ -phosphatase activity (23,27).
We carried out the experiment of Fig. 2 to verify that, in our hands, the FITC modification was indeed causing a total block of the high affinity ATP site. The results confirm and extend earlier findings (27) in that the FITC modification leaves intact the K ϩ -phosphatase activity and the phosphorylation from P i (E2 partial reactions), whereas high affinity ATP binding, phosphorylation by ATP (E1 partial reactions), and the Na,K-ATPase activity (overall reaction) can be reduced to levels below 10%. Hence, it seemed fair to assume that phosphorylation by, and hydrolysis of, ATP in the fluorescein-enzyme were inactivated because of obstructed access to the high affinity ATP site. This matter shall be taken up again under "Discussion." As the low affinity precluded direct measurements of equilibrium binding of TNP-8N 3 -ADP to the fluorescein-enzyme, we decided to evaluate the interaction from the inhibition of the phosphatase reaction. Fig. 3 shows a Dixon plot of TNP-8N 3 -ADP inhibition of the K ϩ -phosphatase activity of the fluorescein-enzyme, obtained at 20 mM K ϩ . The experiment was conducted in dimmed light, to avoid photoinactivation of the enzyme. The K i is 140 M and as the fitted lines cross each other above the abscissa, the inhibition should be of the competitive or mixed types (40). This experiment shows that TNP-8N 3 -ADP behaves toward the fluorescein-enzyme in a manner similar to that of the parent compound, TNP-ADP (23). The specificity of the effect of TNP derivatives on the fluorescein- enzyme was investigated in the experiment of Fig. 4. Trinitrobenzenesulfonate caused little or no inhibition of the K ϩ -phosphatase activity; considering also the poor effect of picrate (23), the trinitrophenyl group can be excluded as the TNP-ADP domain responsible for the inhibition. The figure also shows that TNP-ADP and TNP-adenosine are equally effective. This was rather surprising and would indicate that, at least with the fluorescein-enzyme, the phosphate groups do not play a significant role in the binding of TNP derivatives (with the native enzyme, TNP-ADP, TNP-ATP, are TNP-AMP competitively inhibit phosphorylation by P i with K i values of 1.0, 1.5, and 8.0 M) (41). TNP-UDP and TNP-uridine, again as effective as each other, were less potent than the adenine derivatives; this was to be expected as UTP is a poor substrate for the Na,K-ATPase reaction (42,43). It seems clear that a purine or pyrimidine base is essential for the interaction between the TNP-derivatives and the fluorescein-enzyme. It is reasonable, therefore, to conclude that, although the TNP group is instrumental at increasing the binding affinity (compare also Fig. 1, A and B), the fluorescein-enzyme binds TNP-ADP and TNP-8N 3 -ADP essentially because of their ATP-like properties and most probably at a site that normally accommodates ATP.
We found that UV irradiation of the fluorescein-enzyme in the presence of TNP-8N 3 -ADP resulted in irreversible loss of the K ϩ -phosphatase activity. The results presented in Fig. 5 show that the extent of the photoinactivation depends on the cation composition of the medium. There is an absolute requirement for Na ϩ or Mg 2ϩ ions, the latter being slightly more effective than the former; either will suffice, the effects being nonadditive. K ϩ ions do not support the TNP-8N 3 -ADP photoinactivation and counteract the effects of Na ϩ and Mg 2ϩ . Presumably Mg 2ϩ and Na ϩ stabilize an enzyme conformation with a high affinity for the probe, whereas K ϩ stabilizes a low affinity conformation. This interpretation seems to be supported by the experiment of Fig. 6, which shows the TNP-8N 3 -ADP concentration dependence of the inactivation process in the presence of Na ϩ and Mg 2ϩ . The experiment involved a single inactivation time, with triplicate phosphatase determinations at each TNP-8N 3 -ADP concentration; the ordinate shows the natural logarithm of the fractional K ϩ -phosphatase activity surviving, which should be proportional to the rate constant of photoinactivation, at least with ordinary single exponentials (44). This is half maximal at 15 M TNP-8N 3 -ADP and approaches 65% activity loss at saturating concentrations.
Although the method is not strict, a large discrepancy is obvious between this K 0.5 , obtained in the presence of Na ϩ and Mg 2ϩ and the K i of 140 M, obtained at 20 mM K ϩ (Fig. 3); this disparity agrees with the interpretation of the results of Fig. 5. The experiment in Fig. 7 shows that excess TNP-ADP protects against photoinactivation, as addition of 1 mM TNP-ADP largely prevents 50 M TNP-8N 3 -ADP from photoinactivating the fluorescein-enzyme. This result, taken together with those of Figs. 3 and 6, indicate that the photoinactivation does not simply occur by random collisions of the reactive nitrene derivative, but involves prior recognition of TNP-8N 3 -ADP at specific sites on the FITC-modified Na,K-ATPase.
Following illumination with UV light, TNP-8N 3 -ADP not only inactivates the pNPPase activity but also blocks the phosphorylation of the fluorescein-enzyme from inorganic phosphate (Fig. 8); in this experiment, all samples were irradiated for the maximal time, with variable exposure to TNP-8N 3 -ADP. The abolition of both partial reactions in parallel suggests that the TNP-8N 3 -ADP molecule is being incorporated at a site that is central to the functioning of the fluorescein-enzyme.
To verify that the TNP-8N 3 -ADP inactivation did not require  the presence of oligomeric forms, the fluorescein-enzyme was solubilized with C 12 E 8 ; this was done under conditions shown by sedimentation-velocity analysis to produce a monodisperse population of active ␣␤ protomers in which ␣-␣ interactions are absent (4,22,23). Fig. 9 shows that the soluble, FITC-modified protomers were quite stable during UV irradiation without TNP-8N 3 -ADP, or in the presence of 30 M TNP-8N 3 -ADP without UV. In contrast, simultaneous exposure to UV light and TNP-8N 3 -ADP resulted in a rapid loss of the K ϩ -phosphatase activity. Increasing the concentration of TNP-8N 3 -ADP from 10 to 30 M gave progressively higher levels of photoinactivation, to a maximum exceeding 70%. TNP-8N 3 -ADP photoinactivated the soluble fluorescein-protomers even more effectively than the membrane fluorescein-enzyme (Fig. 5). Both the FITC and TNP-8N 3 -ADP binding sites must therefore be contained within the ␣␤ protomer. DISCUSSION There have been numerous reports of low affinity nucleotide effects on the membrane-bound sodium pump (10, 14, 45-50), which cannot be easily rationalized on the basis of a single ATP site per enzyme unit; such findings have been generally interpreted on the basis of an (␣␤) 2 membrane enzyme. We have shown (4) that solubilized ␣␤ protomers retain the low affinity nucleotide responses found with the membrane-bound Na,K-ATPase (1-3). We could also demonstrate (4) that this was an inherent property of the soluble protomer and not the result of dimerization to (␣␤) 2 . The question remained as to whether the complex ATP effects arose from the existence of separate high and low affinity ATP sites on the ␣␤ protomer or simply from the changing affinity of a unique site, as has been proposed (41,(51)(52)(53).
High Affinity Site Labels-The nature of the problem made it more or less mandatory to try and mask the high affinity ATP site so as to look for a putative low affinity nucleotide site with some confidence. Fluorescein derivatives have been effective ligands for nucleotide sites in structural studies of hexokinase (54) and lactate dehydrogenase (55), where the planar fluorescein ring structure replaces for that of adenine. The incorporation of FITC in the ␣ chain of Na,K-ATPase, which can be prevented by ATP, results in covalent anchoring predominantly at Lys-501 (25), but also at Lys-480 or Lys-766 (26). As Gly-502 and Lys-480 can be labeled by 2N 3 -ATP and 8N 3 -ATP (18,19), respectively, and as FITC seems to be found in a hydrophobic environment (56), it has seemed likely that the intervening sequence is in the adenine subsite of an ATPbinding pocket or in its immediate neighborhood. We chose FITC as the blocking agent in an earlier study (23), as it was known that the modification did not hinder the K ϩ -phosphatase activity or phosphorylation by [ 32 P]P i , although it abolished high affinity binding of formycin triphosphate, phosphorylation by [␥-32 P]ATP, and the Na,K-ATPase activity (9,27,34). The experiment presented in Fig. 2 confirms those results and shows besides that high affinity [ 3 H]ATP binding is also blocked in the fluorescein-enzyme. In addition, we have reported that FITC modifies all ␣ chains in the membrane-bound Na,K-ATPase and not just one-half of a dimeric enzyme (23). In all other experiments in this report, the loss of the K ϩ -phosphatase activity following FITC treatment was at most 20%, while no more than 2% of the Na,K-ATPase activity was left in the modified enzyme. Such thorough inactivation of the overall reaction could not be obtained with 8N 3 -ATP (Fig. 1B) or pyridoxal phosphate. In the case of 8N 3 -ATP, and also of TNP-8N 3 -ADP (Figs. 1A and 5-9), this may be due to protection of the site by unreactive photolysis products or to lability of the covalent bond between enzyme and nucleotide (19); such "hitand-run" modification might render the anchoring residue unreactive toward the nitrene derivative but still permit ATP binding.
Is the FITC Block Absolute?-For our purpose, the essential requirement is that the block of the high affinity site be complete and reliable; it should not matter which of the nearby lysines forms the thiourea linkage provided the fluorescein fully prevents ATP access to the site. Therefore, the possibility must be carefully considered that fluorescein causes only partial steric hindrance at a unique binding pocket, or even that the fluorescein sits outside the ATP site altogether. In the former case, the inactivation of E1 reactions would be formally analogous to fully competitive inhibition by an untethered ligand with a very high affinity. In the second case, the inactivation should also result from a decreased ATP affinity, only that caused by long range effects of fluorescein binding; the situation would be akin to that of partially competitive inhibition (40). We may assume, for the sake of argument, that all the inactivating fluorescein does is to cause a very large decrease in ATP affinity. The test of choice when a competitive component is suspected (40) is to raise the substrate concentration to look for an increase in the enzymic rate. This possibility can be excluded, as absolutely no relief of inactivation could be detected in determinations of Na,K-ATPase activity with the fluorescein-enzyme when the ATP concentration was raised from 2 mM, to 8 (34) or 20 mM (23). We may choose to compound our objections and assume that, besides decreasing the ATP affinity, the FITC modification also interferes with other properties of the enzyme to cause the complete loss of Na,K-ATPase activity. Of course, it is not possible to exclude this possibility but the odds seem stacked against it, as the fluorescein-enzyme can: (i) be phosphorylated by P i , acetyl phosphate, and pNPP (27,57,58); (ii) hydrolyze pNPP, 3-O-methylfluorescein phosphate, and acetyl phosphate in the presence of K ϩ (9,23,27,34,57); (iii) occlude Rb ϩ (27); (iv) bind ouabain (27); and (v) experience conformational changes in response to Na ϩ and K ϩ (27). Moreover, it has been suggested (9) that FITC might block a unique ATP site when the enzyme is in its Na ϩ form but that the tethered fluorescein could pivot to one side when the enzyme is in the K ϩ form, allowing access to phosphatase substrates and nucleotides to the site. Our results make this proposal quite unlikely, as TNP-8N 3 -ADP inactivates the fluorescein-enzyme much more efficiently in the presence of 50 mM Na ϩ than in 50 mM K ϩ (Fig. 5), conditions under which the modified enzyme adopts Na ϩ and K ϩ conformations, respectively (27). All things considered, the most likely explanation for the FITC inactivation of E1 reactions is a comprehensive block of the adenine-binding region of a high affinity ATP site. A tight fluorescein attachment seems to be borne out by fluorescence polarization anisotropy studies (56) which indicate that the fluorescein is rigidly held at its binding site and unable to rotate about the tethering point.
TNP-8N 3 -ADP as an Affinity Label-As expected for an affinity label (59), in the absence of actinic light TNP-8N 3 -ADP binds to the fluorescein-enzyme, as shown by its inhibition of the K ϩ -phosphatase activity. The binding takes place with a low affinity (Fig. 3), just as it happens with ATP and other nucleotides (9), including TNP-ADP (23,34). The inhibition could not result from binding at the blocked high affinity ATP site, inasmuch as TNP-8N 3 -ADP and TNP-ADP are bulkier than ATP, which is excluded (Fig. 2). There must be an alternative pocket where TNP-8N 3 -ADP binds to the fluoresceinenzyme, first reversibly, then covalently following photoactivation of the azido group. The TNP derivative binding depends on the presence of a purine or pyrimidine moiety (Fig. 4), the former being more effective than the latter.
The notion that the photoinactivation involves specific sites on the FITC-modified enzyme is reinforced by the protection afforded by excess TNP-ADP, probably by competition (Fig. 7), and by the large increase in TNP-8N 3 -ADP effectiveness when the enzyme is in a Na ϩ conformation or in the presence of Mg 2ϩ (Figs. 3, 5, and 6). TNP-ADP also protects the native enzyme from TNP-8N 3 -ADP inactivation (Fig. 1A). The K 0.5 obtained in Fig. 6 represents only an approximation, as it is based on single time points in photoinactivation time courses that cannot normally be fitted by single exponentials (cf. Refs. 18 and 19). However, it seems clear that much lower TNP-8N 3 -ADP concentrations are required to inactivate the Na,K-ATPase activity of the native enzyme (Fig. 1A) than the K ϩ -phosphatase activity of the fluorescein-enzyme, even considering the enabling effects of Na ϩ plus Mg 2ϩ (Fig. 6). Finally, the phosphorylation of the enzyme from P i is photoinactivated by TNP-8N 3 -ADP concurrently with the loss of K ϩ -phosphatase activity (Fig. 8). Phosphorylation by P i is an essential step in the reversal of the Na,K-ATPase reaction and involves access to the center of the catalytic enzyme (60). Therefore, the inactivation of the fluorescein-enzyme seems to follow reversible and specific TNP-8N 3 -ADP binding at a main site on each ␣␤ protomer (Fig. 9), with Na ϩ and Mg 2ϩ sites playing an essential role in the process.
Evidence against a Second ATP Site-According to one report (51), ATP did not bind at a high affinity site on the vanadate-inhibited Na,K-ATPase, nor did it promote vanadate release with a low affinity; these findings were interpreted as evidence for a single ATP site on the enzyme. In the presence of Na ϩ , however, ATP and ␤␥-methylene-ATP were later found to accelerate vanadate release from the enzyme, with a typically low affinity (50). Early results with TNP-nucleotides (41) were also interpreted as if supporting the idea of a single ATP site. Essentially, it was found that TNP-ATP inhibited the Na,K-ATPase activity in a competitive fashion, and with similar apparent affinities (K i ϭ 0.1-0.4 M at 37°C), at high as well as at low ATP concentrations. Irrespective of having separate sites or not, one possible explanation is that, at fixed TNP-ATP, a rise in ATP concentration displaces inhibitor not only at the low affinity, but also at the high affinity, binding stage (or site). This should result in recruiting more enzyme units to the reaction, i.e. more elements amenable to low affinity regulation. The low affinity component of the ATP concentration dependence will then become progressively elevated as the ATP concentration increases, and this will look like an apparent reduction in ATP affinity (i.e. competitive inhibition), the higher the fixed TNP-ATP concentration the more marked the effect. Relief of TNP-ATP inhibition at the catalytic site could then be the reason why both regulatory and high affinity ATP ranges appear to respond to the same submicromolar TNP-ATP concentrations. In fact, mixed inhibition, and parabolic Dixon plots, were observed (41); the latter normally imply more than one inhibitor binding site.
Recent evidence against a separate low affinity ATP site arises from experiments (52) with N-(2-nitro-4-isothiocyanophenyl)-imidazole (NIPI), a probe that also binds at Lys-501 (53). Modification of the sodium pump with NIPI causes the loss of high affinity ATP binding and Na,K-ATPase activity, but not phosphorylation by P i or Rb ϩ occlusion. Crucially, the occluded Rb ϩ cannot be displaced by 4 mM ATP, and this should denote absence of a low affinity ATP site (52). It may be that occupation of the catalytic site by NIPI results in a much reduced affinity at the alternative site; such is our experience with ATP in relation to the K ϩ -phosphatase activity of the FITC-modified enzyme, in which case only the tight-binding TNP nucleotides are effective (Figs. 3 and 4) (23). On the other hand, it has since been reported that ATP can accelerate both occlusion and deocclusion reactions in the native enzyme, with similar (low) ATP affinities (61). It is just possible that in the case of the NIPI-modified enzyme, the ATP effects on Rb ϩ binding and release exactly cancel each other.
Recent supporting evidence-Extensive trypsinolysis of Na, K-ATPase produces membranes containing a small number of terminal ␣-chain peptides: this preparation can still occlude Rb ϩ ions (62). It has now been reported (63) that ATP can increase Rb ϩ occlusion and release rates in the trypsinized membranes, with intermediate or low binding affinity. The fragment Asn-353-Lys-736, which encompasses all residues involved in high affinity ATP binding known so far, is missing; therefore, there must be an alternative ATP binding pocket formed by the short terminal cytoplasmic peptides. A recent study of Na,K-ATPase whose high-affinity ATP site has been blocked by FITC or pyridoxal 5Ј-diphospho-5Ј-adenosine (64) has reported low-affinity [ 3 H]-ATP binding to the modified enzymes in the presence of 15 mM Na ϩ . At 300 M ATP, the estimates are around 0.5 mol ATP bound per mol of modified ␣ chain.
Summary-The results described above show that TNP-8N 3 -ADP photoinactivates FITC-modified Na,K-ATPase, either membrane-bound or as soluble protomers, and that this process is preceded by probe binding which is both reversible and specific. These findings confirm our earlier observation that each ␣␤ unit can simultaneously bind fluorescein covalently and TNP-ADP at equilibrium (23). The inference is that each ␣␤ protomer should present one catalytic ATP site (which is here blocked by FITC) and one allosteric site, which is available for TNP-8N 3 -ADP. This second center of nucleotide interaction in the ␣␤ protomer is probably the site where ATP binds as a regulator during the normal operation of the native enzyme. We are now using radioactive TNP-8N 3 -ADP to map its location in the sodium pump structure; preliminary results indicate that about 1 mol of TNP-8N 3 -ADP can be incorporated per mol of FITC-modified ␣ subunit.