Both P-glycoprotein nucleotide-binding sites are catalytically active.

The technique of vanadate trapping of nucleotide was used to study catalytic sites of P-glycoprotein (Pgp) in plasma membranes from multidrug-resistant Chinese hamster ovary cells. Vanadate trapping of Mg- or Co-8-azido-nucleotide (1 mol/mol of Pgp) caused complete inhibition of Pgp ATPase activity, with reactivation rates at 37°C of 1.4 × 10-3 s−1 (t1/2 = 8 min) or 3.3 × 10−4 s−1 (t1/2 = 35 min), respectively. UV irradiation of the inhibited Pgp yielded permanent inactivation of ATPase activity and specific photolabeling of Pgp. Mild trypsin digestion showed that the two nucleotide sites were labeled in equal proportion. The results show that both nucleotide sites in Pgp are capable of nucleotide hydrolysis, that vanadate trapping of nucleotide at either site completely prevents hydrolysis at both sites, and that vanadate trapping of nucleotide in the N- or C-terminal nucleotide sites occurs non-selectively. A minimal scheme is presented to explain inhibition by vanadate trapping of nucleotide and to describe the normal catalytic pathway. The inhibited Pgp·Mg-nucleotide·vanadate complex is probably an analog of the catalytic transition state, implying that when one nucleotide site assumes the catalytic transition state conformation the other site cannot do so and suggesting that the two sites may alternate in catalysis.

P-glycoprotein (Pgp), 1 also called multidrug-resistance protein, is involved in resistance of tumors to a variety of drugs and appears to be responsible for many failures of cancer chemotherapy (reviewed in Endicott and Ling (1989), Gottesman and Pastan (1993), and Gros and Buschman (1993)). Pgp is a plasma membrane protein that reduces intracellular accumulation of drugs by transporting them out of the cell in an ATP hydrolysis-dependent process (Ruetz and Gros, 1994) (reviewed in Shapiro and Ling (1995) and Sharom (1995)). Coupling between catalytic sites for ATP hydrolysis on the cytoplasmic surface and drug-binding/transport sites in the membrane is also apparent from the drug stimulation of ATP hydrolysis seen in plasma membrane preparations (Sarkadi et al., 1992;Al-Shawi and Senior, 1993), partially purified and reconstituted Pgp (Ambudkar et al., 1992;Sharom et al., 1993), and purified, reconstituted Pgp (Shapiro and Ling, 1994;Urbatsch et al., 1994;Sharom et al., 1995).
The amino acid sequence of Pgp reveals a tandemly duplicated molecule, each half containing one cytoplasmically sided nucleotide-binding site, as diagnosed by the presence of both "homology A" and "homology B" consensus sequences (Walker et al., 1982). A major goal of current research is to determine the mode of operation of these two nucleotide-binding sites in the mechanism of action of Pgp. From the data available it has not yet been possible to determine the number of sites that are catalytically active in Pgp. There is a single relatively high K m (MgATP) around 1 mM, and ATP hydrolysis follows simple monophasic Michaelis-Menten kinetics. MgADP and MgAMP-PNP behave as classical competitive inhibitors with K i values of 0.70 and 0.35 mM, respectively (Al-Shawi and Senior, 1993;Urbatsch et al., 1994). With 8-azido-ATP, which is an excellent substrate for hydrolysis by Pgp, photoinactivation of the Pgp ATPase occurred coincidental with covalent incorporation of approximately 2 mol of 8-azido-[␣-32 P]ATP/mol of Pgp, with the incorporated analog distributed equally between N-and Cterminal halves of the molecule. These data indicated that both nucleotide-binding domains were capable of binding 8-azido-ATP (Georges et al., 1991;Al-Shawi et al., 1994). The inhibitor N-ethylmaleimide reacted at two sites, with the critical sulfhydryls located equally in N-and C-terminal halves of the Pgp, whereas NBD-Cl, another catalytic site covalent inhibitor, gave full inactivation at ϳ1 mol/mol of Pgp, with reaction occurring predominantly in the C-terminal half (Al-Shawi and Senior, 1993;Al-Shawi et al., 1994). Since MgATP afforded full protection against inhibition by both N-ethylmaleimide and NBD-Cl, it appeared that both nucleotide sites could bind MgATP and that inhibition at the C-terminal site alone gave 100% inhibition of ATPase activity.
Introduction of point mutations into either or both of the nucleotide-binding domains by site-directed mutagenesis inhibited the drug exclusion function of Pgp expressed in mammalian cells (Azzaria et al., 1989;Roninson, 1992). Loo and Clarke (1994) expressed each half of the Pgp molecule separately in Sf9 cells and found that each half-molecule displayed significant ATPase activity. This experiment showed that each nucleotide site has the potential capability for ATP hydrolysis. The state of oligomerization (homodimer? monomer?) of the expressed half-molecules of Pgp was not determined. Drug stimulation of the Pgp ATPase was apparent only in intact Pgp or when both half-molecules were expressed contemporaneously in the same cell, and Loo and Clarke concluded that coupling of ATPase activity to drug binding/transport requires interaction of both halves of the Pgp.
We have recently found that vanadate trapping of nucleotide at catalytic sites of Pgp induces stable inhibition of ATPase activity (Urbatsch et al., 1995). The nucleoside diphosphate was the trapped species whether vanadate inhibition was induced with MgATP or MgADP, and manganese or cobalt were also effective as metal cofactors. Complete inactivation of ATPase was achieved with trapping of 1 mol of MgADP/mol of Pgp, and full reactivation of ATPase correlated with release of a single mol of trapped Mg-[␣-32 P]ADP per mol of Pgp. It has previously been demonstrated in myosin that vanadate trapping of MgADP at a catalytic site produces a stably inhibited complex, which resembles the catalytic transition state (Goodno, 1979(Goodno, , 1982, and it appeared that vanadate trapping induced a similar complex in Pgp, namely Pgp⅐MgADP⅐V i .
The vanadate-trapping technique is a valuable tool for investigation of Pgp catalytic sites because it greatly increases the apparent affinity for nucleotides. Here we used vanadate trapping with the photoreactive 8-azido-ATP to investigate whether both nucleotide sites in Pgp in plasma membranes can hydrolyze nucleotide and whether vanadate trapping of nucleotide, with consequent inhibition of ATPase, occurs preferentially at one or other of the two nucleotide-binding sites. We further investigated the competition between phosphate and vanadate by measuring formation of the inhibited Pgp species. The results are interpreted in terms of mechanism of vanadateinduced inhibition of Pgp ATPase activity and the normal ATPase reaction pathway.

EXPERIMENTAL PROCEDURES
Preparation of Plasma Membranes-Plasma membranes were prepared from the multidrug-resistant Chinese hamster ovary cell line CR1R12 as described (Al-Shawi and Senior, 1993;Urbatsch et al., 1995). The membranes contained from 15 to 24% w/w of Pgp as a fraction of total membrane protein.
Assay of Pgp ATPase Activity-Pgp ATPase activity was measured as described in Urbatsch et al. (1995). The Pgp ATPase activity was around 1.3 mol of ATP hydrolyzed per min per mg of membrane protein.
Vanadate-induced Inhibition of Pgp-This was done as described in Urbatsch et al. (1995). Briefly, plasma membranes (10 g) were incubated with 200 M vanadate, 200 M nucleotide, 3 mM MgSO 4 or CoSO 4 , 10 M verapamil, 2 mM ouabain, 0.1 mM EGTA, and 40 mM Tris-HCl, pH 7.4, in a total volume of 100 l for 20 min at 37°C. (Variations of these conditions are detailed in the figures.) The incubations were started by addition of membranes and stopped by passage of the 100-l samples through centrifuge columns consisting of 1 ml of Sephadex G-50 (fine) topped with a 10-mm layer of Dowex AG1-X8 (Bio-Rad) (Penefsky, 1977;Wolodko et al., 1983) equilibrated with 0.1 mM EGTA, 40 mM Tris-HCl, pH 7.4, at 23°C. Control experiments using [␣-32 P]ATP showed that Ͻ0.002% of the applied nucleotide eluted from the columns in the absence of membranes.
UV Irradiation of Vanadate-trapped Nucleotide in Pgp-Samples eluting from the centrifuge columns after vanadate-induced inhibition were placed on ice and subjected to irradiation for 5 min ( ϭ 254 nm, 5.5 milliwatts/cm 2 ). Where required, mild trypsin digestion was carried out by the procedure of Georges et al. (1991).
Determination of Stoichiometry of Vanadate-trapped Nucleotide in Pgp-The stoichiometry of vanadate-trapped nucleotide in Pgp was determined as described by Urbatsch et al. (1995).
Routine Procedures-SDS-gel electrophoresis, immunoblotting, and protein assay by the bicinchoninic acid method in the presence of 1% SDS were all performed as described previously (Al-Shawi and Senior, 1993).

Vanadate Trapping of 8-Azido-adenine Nucleotide in Pgp
Catalytic Sites-Vanadate-induced inhibition of the ATPase activity of Pgp in plasma membranes from multidrug-resistant Chinese hamster ovary CR1R12 cells was carried out as described under "Experimental Procedures." Briefly, the technique involved preincubation of plasma membranes with nucleotide, divalent cation, and vanadate, followed by removal of unbound ligands by centrifuge column elution and assay of Pgp ATPase activity in the eluates.
With 200 M Mg-8-azido-ATP and Co-8-azido-ATP the degree of inhibition was 85 and 95%, respectively. Similar results were seen with Mg-8-azido-ADP and Co-8-azido-ADP. Reactivation of Pgp ATPase activity occurred when the centrifuge column eluates were incubated at 37°C. Fig. 1A shows the reactivation after inhibition in the presence of 200 M Mg-8azido-ATP or Mg-8-azido-ADP. The time courses for reactivation in the two cases were identical (t 1 ⁄2 ϭ 8 min; k app ϭ 1.4 ϫ 10 Ϫ3 s Ϫ1 ). Fig. 1B shows the reactivation after inhibition in the presence of the corresponding Co-nucleotides. Here, the time courses were appreciably slower but were again the same for the two cases (t 1 ⁄2 ϭ 35 min; k app ϭ 3.3 ϫ 10 Ϫ4 s Ϫ1 ). The fact that identical reactivation rates were seen whether inhibition was induced with 8-azido-ATP or 8-azido-ADP argues strongly that the inhibited Pgp species contains vanadate-trapped 8-azido-ADP at catalytic sites. This is consistent with previous work (Urbatsch et al., 1995), which established that vanadatetrapped ADP was the inhibitory species when Pgp was preincubated with ATP and vanadate, and with the fact that 8-azido-ATP is known to be a good substrate for hydrolysis by Pgp . The reactivation of Pgp ATPase after inhibition with 8-azido-adenine nucleotide and vanadate was notably faster than was seen previously after inhibition with adenine nucleotide and vanadate. Thus the fact that reactivation was slower in the case of Co-8-azido-nucleotide proved technically advantageous.
Permanent inactivation of Pgp ATPase was obtained in samples that had been inhibited by preincubation with vanadate and 200 M Co-8-azido-ATP and then subjected to UV irradiation as described under "Experimental Procedures." Samples that had been so treated remained 91% inhibited after 4 h of incubation at 37°C. Thus, in contrast to the data in Fig. 1B, there was now essentially no reactivation, confirming that 8-azido-adenine nucleotide had been trapped in catalytic sites.
Dependence of Vanadate-induced Inhibition of Pgp ATPase Activity on Co-8-azido-ATP Concentration-The dependence of vanadate-induced inhibition of Pgp ATPase activity on the concentration of Co-8-azido-ATP was examined, and the concentration required for 50% inhibition was found to be 3 M (Fig. 2). Maximal (95%) inhibition was reached at Ն80 M. We determined the amount of radioactive nucleotide in the eluates obtained from centrifuge columns after preincubation of plasma membranes with vanadate and Co-8-azido-[␣-32 P]ATP (80 M), so as to achieve 95% inhibition of Pgp-ATPase, and the calculated stoichiometry of binding (see "Experimental Procedures") was 0.94 mol of nucleotide/mol of Pgp. Therefore, the vanadate trapping of Co-8-azido-adenine nucleotide at just one catalytic site per Pgp molecule brought about 100% inhibition of ATPase activity, consistent with what was seen earlier with Mg-adenine nucleotide (Urbatsch et al., 1995).
Photolabeling of the Two Pgp Nucleotide Sites by Vanadatetrapped 8-Azido-adenine Nucleotide-Pgp has two predicted nucleotide-binding sites, and it was therefore of interest to determine at which site vanadate trapping of nucleotide occurred when Pgp was maximally inhibited by a single mol of trapped nucleotide. We previously established (Urbatsch et al., 1995) that when CR1R12 plasma membranes were preincubated with vanadate and 200 M Mg-8-azido-[␣-32 P]ATP to induce inhibition of Pgp ATPase, then the eluates from centrifuge columns were subjected to UV irradiation, the Pgp was specifically photolabeled. Whereas if vanadate was omitted, no photolabeling occurred. We first confirmed here that the same result was seen when 200 M Co-8-azido-[␣-32 P]ATP was used instead of Mg-8-azido-[␣-32 P]ATP (data not shown).
We then performed experiments using a range of concentrations of Co-8-azido-[␣-32 P]ATP from 80 M down to 5 M. This concentration range produced inhibition of Pgp ATPase from 95% down to 60% (see Fig. 2), and therefore the stoichiometry of trapped nucleotide would range from ϳ1 mol/mol to around 0.6 mol/mol. Fig. 3 (lanes 1-4) shows the photolabeling of Pgp in plasma membranes after vanadate trapping with Co-8azido-[␣-32 P]ATP at concentrations of 80, 20, 10, and 5 M. Direct counting of the Pgp bands excised from SDS gels showed that 26 -54% of the trapped 32 P in the centrifuge column eluates became covalently attached to Pgp on UV irradiation and survived gel electrophoresis. For comparison, photoincorporation of vanadate-trapped azido-nucleotide analogs in myosin ranged from a few percent to 70% (Yount et al., 1992). Georges et al. (1991) showed that Pgp in Chinese hamster ovary cell plasma membranes can be fragmented into N-terminal and C-terminal "halves" by mild trypsin digestion. Each half contains one of the predicted nucleotide-binding domains. The N-terminal (glycosylated) fragment has an apparent molecular size of 100 kDa and the C-terminal fragment of 65 kDa on SDS gels. We previously confirmed, using the C219 monoclonal antibody, that this technique is reproducible with CR1R12 cell plasma membranes . Fig. 3  (lanes 5-8) shows the results obtained when the photolabeled Pgp was digested with trypsin into the two halves before running on SDS gels. It is seen that even at the lowest concentration of nucleotide used, both halves of Pgp were labeled. The distribution of the radioactivity between the two halves as determined by direct counting of bands excised from gels was: N-terminal, 53%; C-terminal, 47%. This ratio was similar at all the nucleotide concentrations studied (the variation was Ϯ9%). The fact that both nucleotide sites became labeled at all the nucleotide concentrations used showed that vanadate trapping of nucleotide in N-or C-terminal nucleotide sites occurred non-selectively.
Similar experiments to those shown in Fig. 3 were performed using Mg-8-azido-[␣-32 P]ATP, and they confirmed the above conclusions. Briefly, full inhibition of Pgp ATPase was obtained with ϳ1 mol of trapped Mg-nucleotide/mol of Pgp, and using a range of concentrations of Mg-nucleotide which yielded 36 -74% inhibition of ATPase activity by vanadate trapping, it was found after subsequent UV irradiation and mild trypsin digestion of the samples that both halves of Pgp became photolabeled (data not shown).
Competition between Phosphate and Vanadate-Further studies on the mechanism of vanadate-induced inhibition of Pgp ATPase were made possible by investigating the effects of phosphate. Fig. 4A shows the effects of phosphate on vanadateinduced inhibition with 200 M MgATP in plasma membranes. In the absence of phosphate (diamonds) the concentration of vanadate producing half-maximal inhibition was 4 M, whereas the presence of 200 mM NaP i (open circles) increased this value to 23 M. These data indicated that P i competes with V i , with K i (P i ) ϭ 38 mM. Fig. 4A (closed circles) shows data obtained in the presence of 133 mM Na 2 SO 4 , added to mimic the ionic strength provided by 200 mM NaP i . Here the concentration of vanadate producing half-maximal inhibition was 8 M.
Comparison of the "plus Na 2 SO 4 " data with the "plus NaP i " data yielded a K i (P i ) of 100 mM. Fig. 4B shows a parallel experiment using 200 M MgADP. Here, the concentration of vanadate producing half-maximal inhibition was 9 M (no P i , squares) or 100 M (plus 200 mM NaP i , open inverted triangles). The calculated K i (P i ) from these data was 22 mM, and inhibition was again competitive. The closed inverted triangles show similar data obtained in the presence of 133 mM Na 2 SO 4 . Here the concentration of vana-

FIG. 2. Dependence of vanadate-induced inhibition of Pgp
ATPase activity in plasma membranes on Co-8-azido-ATP concentration. Inhibition was induced as described under "Experimental Procedures" using 200 M vanadate, 3 mM CoSO 4 , and 8-azido-ATP concentration as indicated. The control (uninhibited) activity was that of a sample preincubated without nucleotide. date producing half-maximal inhibition was 26 M, corresponding to a K i (P i ) of 70 mM.
Further experiments were performed in which V i concentration was held constant at 200 M, and MgATP or MgADP concentration was varied from zero to 200 M in the presence or absence of 200 mM P i . There was no significant effect of P i on the observed inhibition at different nucleotide concentrations. Fig. 5 shows one typical experiment performed to determine the effect of 200 mM NaP i on ATPase activity of Pgp in plasma membranes at varying concentrations of MgATP. In order to control for ionic strength, Na 2 SO 4 (133 mM) was included in the assays when P i was not present. In separate experiments it was confirmed that Na 2 SO 4 at this concentration had no significant effect on V max or K m (MgATP) of Pgp in plasma membranes. The average data of three experiments showed that 200 mM P i reduced V max from 1.2 to 0.6 mol of ATP hydrolyzed per min per mg of membrane protein, while K m (MgATP) was reduced from 1.5 mM to 1.2 mM. These experiments showed that P i is a weak, mixed-type inhibitor of ATP hydrolysis and that the concentration of P i needed to reduce V max by 50% is approximately 200 mM. DISCUSSION We used the technique of vanadate-induced inhibition by trapped nucleotide to study catalytic properties of P-glycoprotein in plasma membranes from multidrug-resistant CR1R12 Chinese hamster ovary cells. It was demonstrated that in presence of vanadate plus 8-azido-[␣-32 P]ATP and either Mg 2ϩ or Co 2ϩ , 8-azido-adenine nucleotide became stably trapped in Pgp catalytic sites. Trapping of 1 mol of nucleotide per mol of Pgp caused complete inhibition of ATPase. The evidence presented here and previously (Urbatsch et al., 1995) indicates that nucleoside diphosphate is the trapped inhibitory species. UV irradiation of inhibited Pgp caused permanent inactivation and revealed that both halves of Pgp became photolabeled. Vanadate trapping of nucleotide in the N-or C-terminal nucleotide sites appeared to occur non-selectively. The true distribution of vanadate-trapped nucleotide in the two sites might be obscured by differential covalent attachment upon UV irradiation or loss of covalent label on SDS-gel electrophoresis. Nevertheless, the presence of radioactive label in both halves of Pgp provides convincing evidence that both nucleotide sites in intact Pgp are capable of hydrolyzing 8-azido-ATP to 8-azido-ADP and thus forming the stably inhibited Pgp⅐Me-8-azido-ADP⅐V i complex. Furthermore, vanadate trapping of nucleotide at either of the  1-4, 80, 20, 10, and 5 M nucleotide; lanes 5-8, same as lanes 1-4 except that the samples were subjected to mild trypsin digestion to fragment the Pgp into N-and C-terminal halves before being applied to SDS gels. The position of N-and C-terminal halves of Pgp on SDS gels was confirmed using immunoblotting with C219 anti-Pgp monoclonal antibody. two sites must be able to completely prevent hydrolysis at both sites in order to give complete inhibition of ATPase activity. Therefore, both nucleotide sites in Pgp are potentially catalytically active, but they are not independent; indeed there is very strong cooperative interaction between them. One likely possibility is that the two sites alternate in catalysis.

Effects of High Phosphate Concentrations on MgATP Hydrolysis-
The data presented here are consistent with previous results , which indicated that ATP and 8-azido-ATP bind to both of the two nucleotide-binding sites. Much higher concentrations of 8-azido-ATP were used previously than here because vanadate was not used, and this explains why labeling stoichiometries of up to 2 mol/mol of Pgp were seen in the earlier work. It was also seen previously that NBD-Cl gave full inhibition of ATPase at a labeling stoichiometry of 1 mol/mol of Pgp. This also is consistent with the results obtained here. Interestingly, NBD-Cl reacted predominantly in the C-terminal half, this preference being presumably due to a specific reactive residue in the C-terminal nucleotide site, which is either absent or less reactive in the N-terminal nucleotide site. N-Ethylmaleimide, in contrast, appeared to react in both nucleotide sites.
The data presented here are also consistent with previous mutagenic analyses of Pgp, which showed that introduction of point mutations into either or both nucleotide sites rendered the Pgp completely inactive (Azzaria et al., 1989;Roninson, 1992), and also with the data of Loo and Clarke (1994), which showed that either nucleotide site is potentially capable of ATP hydrolysis. It seems likely that ATPase activity in the experiments of Loo and Clarke (1994) resulted from formation of homodimers of the expressed "half-molecules." In Scheme 1 we propose a minimal reaction pathway for vanadate-induced inhibition of Pgp. This scheme draws on work presented here and previously (Urbatsch et al., 1995) and parallels schemes for vanadate-induced inhibition of myosin ATPase (Goodno, 1979(Goodno, , 1982Goodno and Taylor, 1982) and dynein ATPase (Shimizu and Johnson, 1983). Vanadate-induced inhibition of Pgp is seen when either MgATP or MgADP is incubated with vanadate (V i ), but MgADP is the trapped ligand in both cases. We propose that the Pgp⅐MgADP species forms by either pathway k ϩ1 ⅐k ϩ2 ⅐k ϩ3 or by pathway k Ϫ4 . V i can then bind to Pgp⅐MgADP (step k ϩ5 ) to form the inhibited Pgp⅐MgADP⅐V i complex. Reactivation of ATPase is slow (steps k Ϫ5 ⅐k ϩ4 ) (k app ϭ 1.4 ϫ 10 Ϫ4 s Ϫ1 at 37°C) (Urbatsch et al., 1995) and must be limited by rate k Ϫ5 , since overall ATPase turnover in Pgp is ϳ20 s Ϫ1 at 37°C.
The finding that phosphate competes with vanadate (Fig. 4), but not with nucleotide, during formation of the inhibited Pgp species indicated that the stable inhibited complex cannot form when the catalytic site of Pgp is occupied by MgADP plus P i . Thus P i appeared to compete with V i at step k ϩ5 but was clearly a weakly binding ligand. V i was shown previously to be a P i analog, which became bound to P i sites in both myosin and dynein (Goodno and Taylor, 1982;Werber et al., 1992;Shimizu and Johnson, 1983), and it appears that V i occupies essentially the same position in the Pgp catalytic site as that of P i derived from ATP during hydrolysis. The weak binding of P i by Pgp implies that a large free energy decrease occurs on P i release (step k ϩ3 ) and that this step may normally be coupled to transport of drugs. P i at a very high concentration did inhibit ATPase activity (Fig. 5). An explanation that is consistent with Scheme 1 would be that P i , when present at a high concentration in the medium, can bind to the Pgp⅐MgADP species, which has formed by catalytic steps k ϩ1 ⅐k ϩ2 ⅐k ϩ3 , and that this prevents MgADP release at step k ϩ4 . This would imply that P i release must precede MgADP release in catalysis and that MgADP release (step k ϩ4 ) is rate-limiting. Shimizu and Johnson (1983) analyzed presteady state kinetics of MgATP binding and hydrolysis by dynein ATPase and found that in the presence of vanadate the very first cycle of MgATP hydrolysis as measured by ␥-32 P release occurred in a normal manner, but the ATPase activity became inhibited after just a single turnover. In Pgp, as we reported earlier, vanadate inhibition induced with MgATP was rapid, whereas that induced with MgADP was much slower. We speculate that the first catalytic turnover in Pgp could rapidly produce an intermediate that favors occurrence of inhibition and that the binding of vanadate to this intermediate is also rapid.
Overall, it appears reasonable to assume that the inhibitory Pgp⅐MgADP⅐V i complex is a stable analog of the Pgp⅐MgADP⅐P i transition state complex and that the normal catalytic pathway in Pgp follows the route (k ϩ1 ⅐k ϩ2 ⅐k ϩ3 ⅐k ϩ4 ). Understanding how the two catalytic sites interact such that presence of the Pgp⅐MgADP⅐V i complex at either site completely blocks ATP hydrolysis at both sites and whether this interaction is a reflection of normal catalytic behavior remains to be clarified. The present data suggest that once one site assumes the catalytic transition state conformation, the other site cannot do so.