The interaction of amines with the occluded state of the Na,K-pump.

We have studied the effect of various amines on the rate of release of 86Rb from the occluded state of dog kidney Na,K-ATPase formed by pre-incubation of the enzyme with 86Rb. In the presence of MgPi, various amines act like K+ or Rb+ in blocking the release of 86Rb from one of two sites for occlusion (the "s" site). Of 38 amines tested, tetrapropylamine and various benzyl amines exhibit the highest affinity; the K1/2 for these compounds is 2-5 mM. In the presence of ATP, when 86Rb is presumably released towards the intracellular face of the pump in the normal mode of operation, 86Rb release is blocked by the presence of amine, but only if the amine is also included in a preincubation with MgPi. The data are consistent with a model in which the interaction of amine with one of the transport sites (the "f" site) prevents the E2----E1 transformation that is stimulated by ATP. When 86Rb deocclusion from the f site has occurred in the presence of amine, the lone 86Rb at the s site can be released in the presence of ATP if the amine is removed from the medium. This suggests that a single 86Rb ion at the s site can be released to the intracellular face of the membrane, and therefore that transport can occur with only one K+ site occupied. The amine that blocks release of one 86Rb ion does not itself become occluded: (a) The interaction of amine and ATP is only seen when both ligands are present in the medium; (b) the effects of amines are not "remembered" after a brief exposure to a rinse medium; (c) with the vanadate-inhibited enzyme, benzyltriethylamine and tetrapropylamine are only weakly effective in blocking 86Rb release from the s site; and (d) organic cations exhibit very low affinity in competition with 86Rb for occlusion at equilibrium. Thus the results are consistent with the idea that monofunctional amines block by binding to the f site but that, unlike K+ and Rb+, they do not become occluded. In contrast, at equilibrium ethylenediamine prevents 86Rb occlusion in a competitive manner, suggesting the possibility of occlusion of the bifunctional amine.

The Interaction of Amines with the Occluded State of the Na,K-Pump* (Received for publication, October 29, 1987) Bliss Forbush I11 With the technical assistance of Grace Jones and John T. Barberia From the Department of Cellular and Molecular Physwlogy, Yale University School of Medicine, New Haven, Connecticut 06510 We have studied the effect of various amines on the rate of release of 86Rb from the occluded state of dog kidney Na,K-ATPase formed by pre-incubation of the enzyme with S6Rb. In the presence of MgPi, various amines act like K' or Rb+ in blocking the release of s6Rb from one of two sites for occlusion (the "s" site). Of 38 amines tested, tetrapropylamine and various benzyl amines exhibit the highest affinity; the ICnh for these compounds is 2-5 mM. In the presence of ATP, when "Rb is presumably released towards the intracellular face of the pump in the normal mode of operation, "Rb release is blocked by the presence of amine, but only if the amine is also included in a preincubation with MgPi. The data are consistent with a model in which the interaction of amine with one of the transport sites (the "f" site) prevents the Ez + El transformation that is stimulated by ATP. When "Rb deocclusion from the f site has occurred in the presence of amine, the lone "Rb at the s site can be released in the presence of ATP if the amine is removed from the medium. This suggests that a single "Rb ion at the s site can be released to the intracellular face of the membrane, and therefore that transport can occur with only one K' site occupied. The amine that blocks release of one "Rb ion does not itself become occluded: (a) The interaction of amine and ATP is only seen when both ligands are present in the medium; (b) the effects of amines are not "remembered" after a brief exposure to a rinse medium; (c) with the vanadate-inhibited enzyme, benzyltriethylamine and tetrapropylamine are only weakly effective in blocking S6Rb release from the s site; and (d) organic cations exhibit very low affinity in competition with "Rb for occlusion at equilibrium. Thus the results are consistent with the idea that monofunctional amines block by binding to the f site but that, unlike K' and Rb', they do not become occluded. In contrast, at equilibrium ethylenediamine prevents "Rb occlusion in a competitive manner, suggesting the possibility of occlusion of the bifunctional amine.
It is now clear that in the K' transport steps of the Na,Kpump cycle, K' (or Rb') ions become "occluded" within the pump molecule, inaccessible for a time to the intracellular and extracellular media (Post et al., 1972;Glynn and Richards, 1982). In the normal turnover of the pump, the K' (or Rb') ions are released into the intracellular medium subsequent to * This work was supported by National Institutes of Health Grant GM-31782. 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. B SCHEME 1 the binding of ATP at a low affinity site, and accompanying an &-E2' conformational change in the enzyme; in the isolated Na,K-ATPase, this is manifested by a rapid release of %Rb on exposure to ATP (Forbush, 1987a). Alternatively, if the transport cycle is partially reversed by phosphorylation of the enzyme by Pi, the occluded ions are released (Glynn et al., 1985b;Forbush, 1987b), to the extracellular face of the pump (Kenney and Kaplan, 1987;Forbush et al., 1988). In this case it has been found that the interaction of the two 86Rb ions with the extracellular medium is an ordered process, the release of =Rb from one site ("s" site, with slow release) being blocked by occupancy of the other site ("f" site, with fast release ;Forbush, 1985,198713;Glynn et al., 1985a, 198513). The data are consistent with the simple geometric model in Scheme lA. The kinetics of interaction have led us to propose that there is a barrier to the free exchange of Rb' (or K+) between the transport sites and the extracellular medium, such that even when the enzyme is phosphorylated by Pi, a "gate" is only open briefly to allow exit of one and then the other =Rb ion (Forbush, 1987b).
Recently the complete amino acid sequences of both the a and / 3 subunits of the Na,K-ATPase have been deduced from cDNA clones (Shull et al., 1985(Shull et al., , 1986Kawakami et al., 1985). To gain an understanding of how the structure of the enzyme relates to function, it will also be necessary to identify individual amino acids involved in ligand binding and catalytic activity. Suitable probes have been prepared for the ouabainbinding site (cf. Forbush, 1983;Lowndes et al., 1984), the ATP binding site (Ohta et al., 1986;Ovchinnikov et al., 1987) and the phosphorylation site ('*P itself; Albers et al., 1963;Post et al., 1965), but as yet there is no way to identify the translocation sites for Na' and K'. Two general approaches have been suggested (a) the use of group-specific reagents to label amino acids presumed to be involved in ion binding, e.g. carboxylic acids and (b) affinity or photoaffinity derivatives of organic compounds that mimic the inorganic ions. If suitable compounds can be found, the latter approach has the advantage that it will be possible to ascertain that the labeled site is indeed the cation binding site, by evaluation of the behavior of the organic molecule; it may be transported or at least occluded on a Na,K-ATPase.
It has been known for some time that the Na,K-pump is inhibited by quaternary amines such as tetraethylamine (Sachs and Conrad, 1968) and tetrapropylamine (Kropp and Sachs, 1977) in the extracellular medium. The inhibition is in competition with extracellular K', in a manner generally consistent with binding of the amines to K' transport sites (Sachs and Conrad, 1968). Tris and imidazole have also been found to interact with the pump, acting like Na' in stabilizing the E, form of the enzyme (Skou and Esmann, 1980) and in promoting phosphorylation (Norby et al., 1983;Schuurrnans Stekhoven et al., 1985), and it has been proposed that these cations bind at a distinct Tris site and at Na' sites. These interactions suggest the possibility that appropriate organic cations might be used to covalently label the transport sites. Although the apparent affinities for the organic cations are quite low (the Ki of tetrapropylamine in inhibiting transport is about 3 mM (Kropp and Sachs, 1977), the possibility that an organic cation might become "occluded" raised our hopes that a photoaffinity labeling approach would be feasible. Recently, Schuurmans Stekhoven et al. (1988) andFukushima (1987) have reported the effects of a number of amines on phosphorylation of Na,K-ATPase. Noteworthy among the compounds tested was ethylenediamine, which inhibited phosphorylation of Na,K-ATPase with K, < 0.1 mM (Schuurmans Stekhoven et al., 1988).
In this study we examine the interaction of various amines with the ffiRb-occluded state of the Na,K-pump. It is found that organic cations do indeed act like K' or Rb' at a high affinity site that is apparently the extracellular transport site; in doing so, they block both the MgPi-stimulated release of %Rb from the s site and the ATP-stimulated release of 86Rb associated with the E , + E, conformational change. However, the data indicate that the monofunctional amines do not themselves become occluded, but instead interact with the Na,K-ATPase to hold the gate open, as indicated in Scheme lB, above. On the other hand, the bifunctional compound ethylenediamine appears t o compete at the transport site with high affinity, and it may become occluded. Preliminary results of part of the work have been presented (Forbush, 1986).

EXPERIMENTAL PROCEDURES
Methods for the measurement of the rapid release of =Rb from the occluded state of Na,K-ATPase have been described previously (Forbush, 1987a(Forbush, , 1987b(Forbush, , 1988a(Forbush, , 1988b; experiments described here were carried out during the same time period as those in these references, and the same methods were used. Briefly, a sample of dog kidney Na,K-ATPase (sodium dodecyl sulfate-washed microsomes) is incubated with 0.5 mM =Rb for 30 s at 20 "C (and usually 10-60 min at 0 T ) , diluted in imidazole/EDTA, filtered, prerinsed, and transferred to a rapid filtration apparatus (Forbush, 1984) where, following exposure to a rinse solution, the release of %Rb from the sample is tained 25 mM imidazole, and 1 mM EDTA or EGTA pH 7.2 (or pH followed in a test solution as a function of time. All solutions con-6.8 for the dilution/prerinse solution), unless otherwise noted. When included, "MgP? refers to 4 mM Mg'/8 mM Pi and "MgATP" refers to 5 mM M c / 4 mM ATP. Unless noted otherwise, included monovalent cations were 100 mM.
In all compounds studied, Na' was less than 1 mmol/mol relative to the amine (except 7.5 mmol/mol in tributylmethylamine), and K+ was less than 0.15 mmol/mmol (except 0.5 mmol/mol in tetrabutylamine and 0.2 mmol/mol in dipentylamine). Limited solubility precluded testing of many other available compounds (e.g. benzylbutylamine, tripentylamine, tetrahexylamine, triphenylamine). Benzyltriethylamine was obtained both as the chloride and the bromide, and identical results were obtained with each. Benzylmethylephedrinium bromide and phenacylpyridinium bromide were tested; all other compounds were prepared as chloride salts. Choline chloride was obtained from Syntex (Springfield, MO), stored frozen as a dry salt, and a fresh solution prepared before use. Decomposition is a well known difficulty with the use of choline, and we have noted that the stability of the singly occluded =Rb ion (see "Results," with respect to Fig. 6) appears to be sensitive to breakdown products or impurities in choline chloride. In direct comparisons of different batches of choline chloride from various suppliers, we have noted differences in the rate of =Rb release in solutions containing 100 mM choline and MgP,.
Data Presentation-Our methods of data analysis and presentation have been describedpreviously (Forbush, 1987a(Forbush, , 1987b(Forbush, , 1988a(Forbush, , 1988b. In Figs. 5 and 7 we have corrected for a spontaneous background rate of release applying the assumption that the amount of spontaneous release is proportional to the remaining counts; thus we have subtracted the product of the amount of remaining occluded =Rb in the sample and the instantaneous fractional background rate of release in a control sample. In practice, compared with a simple subtraction of one curve from another, this amounts to the same correction for the early time points when all of the occluded =Rb remains, but a much smaller correction late in the time course when there is no occluded =Rb left to maintain a background rate. Temperature-The rinse ("a") and test ("b") solutions in the rapid filtration apparatus were at 20 "C in all of these experiments.

Release of 86Rb in the Presence of Tetrapropylamine and
MgPi-We have previously found that when the Na,KATPase is phosphorylated by MgP,, and the K' transport sites are thereby exposed to the extracellular face of the membrane, release of ffiRb ions from the sites proceeds in an ordered fashion (Forbush, 1987~). Thus the fast release of one %Rb ion (from the f site) is unaffected by cations in the medium, but release of ffiRb from the second site is blocked by K' or Rb' in the medium. We examined a number of organic cations to see if they act like K' in blocking ffiRb release from the s site; as before, we preincubated Na,K-ATPase with =Rb to form the occluded state and studied the release of %Rb in a rapid filtration apparatus (Forbush, 1984) in media containing 4 mM Mg', 8 mM Pi, 100 mM salt, 25 mM imidazole, 1 mM EGTA. Fig. 1 compares the time course of €',-stimulated ffiRb release in choline chloride, RbCl, and tetrapropylammonium chloride. As indicated by the break in the curves in both the log (0) and integral (0) plots, tetrapropylamine (upper right) was found to act like Rb' (bottom left; or K', not shown) in blocking the release of one of the two 86Rb ions. T o confirm that theS6Rb ion whose release is blocked by tetrapropylamine is the same as the ion whose release is blocked by @Rb, i.e. the ion at the s site, we used a test solution containing RbMgPi following a rinse solution containing tetrapropylamine-MgPi.
As shown in the lower right panel of Fig. 1, in this case only a slow phase of release was observed, demonstrating that indeed the same 86Rb ion is blocked by the amine as by Rb'. A similar result was obtained when the order of RbMgPi and tetrapropylamine-MgP, was reversed (not shown). The remaining 86Rb ion rapidly dissociates in choline-MgPi (dashed lines in the same panel), indicating that the effect of tetrapro-FIG. 1. The release of -Rb from an occluded state of Na,K-ATPase in the presence of MgP, and choline, Rb+, or tetrapropylamine. Na,K-ATPase was incubated with 0.5 mM =Rb in 3 mM Mg/imidazole/EDTA for 30 s at 20 ' C to form the occluded state as described previously (Forbush, 1987a). After dilution, filtration, and prerinse in imidazole/EDTA (pH 6.8) the sample was transferred to the rapid filtration apparatus where the rinse solution contained imidazole/EDTA (except lower right) and the test solution contained MgP, and indicated cations.  Table I. [amine] (mM) 0 50 100 pylamine is readily reversible. Thus these data demonstrate that tetrapropylamine acts like Rb' (or K') and does not interfere with =Rb release from the f site, but blocks release of "jRb from the s site.
Comparison of Various Organic Amines-In an attempt to find the most effective K+-like organic compounds and to gain insight into the structural constraints on cation interaction, we examined the concentration dependence of a number of different organic cations in blocking the release of =Rb from the s site on Na,K-ATPase. Fig. 2A presents results for three amines. Benzyltripropylamine (left) is representative of amines which interact with high affinity to block the release from the s site (like tetrapropylamine in Fig. 1); pentylamine (center) is representative of compounds with similar action but low affinity; and propylamine (right) is representative of a number of compounds which slow the release of both "jRb ions (see below). It is seen here that with increasing concentrations of benzyltripropylamine and pentylamine that, al- Values of Kapp were obtained from least squares curve fits as in Fig. 2. Where more than one determination was made, the value listed first is judged most reliable (most appropriate choice of amine concentrations, least scatter in fit rate constants), and the others are presented in order of increasing Kepp Where marked with an asterisk, there was no indication of amine-induced biphasicity in the 86Rb release time course. -, no effect of the amine was observed. $, aniline stimulated both phases of 86Rb release 1.5-fold. Phenacylpyridinium -" ". . though the initial rate of "Rb release is hardly affected, the rate of ''jRb release is decreased late in the time course; i.e. the curves are clearly biphasic. (Variation in the total amount of "jRb released in 3 s reflects error in aliquoting the sample, e.g. results with 12, 40, and 100 mM pentylamine). With propylamine the initial rate of %Rb release is lowered as well as the late phase. To estimate the degree of change in the second phase of release, first order rate constants were obtained from a least squares curve fit of the time course of =Rb release in Fig. 2 4 , utilizing only the last 40% of the total "jRb; the values are plotted in Fig. 2B (points). These data were then fit to a model of inhibitor binding to a single site: the curves in Fig. 2B express the least squares solutions. The apparent inhibitory constants obtained from these fits, and similar data for 35 other amines are summarized in Table I (17 other experiments are summarized; typically 4-6 cations/ experiment). It is seen in Table I that almost all of the amines tested slowed the second phase of ''jRb release. The affinity for the cations varies considerably: although it is difficult to formulate a general rule, it may be noted that the effectiveness of the hydrocarbon side chains increases in the order methyl < ethanol < ethyl < propyl, butyl, pentyl, benzyl. It is seen that to be effective the organic molecule must have a sufficient bulk in the side chains, with effective compounds (e.g. Kapp < 15 mM) having at least 8-9 carbon atoms; on the other hand it appears that beyond a certain size the effectiveness decreases, since among the tertiary and quaternary amines the affinity for compounds with butyl chains is less than the affinity for those with propyl chains.
Among the organic cations listed in Table I (data marked with asterisk), several compounds were similar to propylamine (Fig. 2) in that they inhibited the initial rate of =Rb dissociation as well as the late phase, generally interacting with low affinity (Kapp > 30 mM, except benzylamine). Most of these were among the smallest molecules tested. One possibility which we have not tested, is that these compounds interfere with phosphorylation of the K+-occluded form, similar to an effect of very high concentrations of NaCl (Forbush, 1987b).
The pH Dependence of Block by Amines-In a search for conditions which would increase the affinity for Na,K-ATPase for organic cation, we examined the pH dependence of the apparent affinity for benzyltriethylamine in blocking release of ''jRb from the s site. As shown in Fig. 3 (O), the apparent affinity for the amine increased about 10-fold between pH 7 and 8. Further experiments, one of which is illustrated in Fig. 3 (O), demonstrated that the dependence on pH is not unique to the interaction of amines with Na,K-

Amines and the
Occluded State of the Na,K-Pump ATPase but also applies, albeit less strongly, to the block by is higher in Rb' or benzyltriethylamine than in imidazole/ 86Rb itself. These results suggest the possibility that protons EDTA alone (upper left; the rates of release at the t = 0 time compete directly with =Rb or amines at the f site, although point are 2.1, 2.4, and 3.4 in 100 mM =Rb, 100, and 200 mM alternative explanations involving allosteric interactions can benzyltriethylamine, respectively, relative to a rate of 1.0 in not be ruled out. imidazoleDDTA). We have not further investigated as to The Interaction of Amine with the Vanadate-inhibited Na,K-ATPase-We have previously shown that the release of =Rb from the occluded state with vanadate bound VO,) is similar to release from the phosphorylated enzyme but about 25-fold slower; importantly Rbc (or K' ) blocks release from one of the two occlusion sites (Forbush, 1987b). Simple models of block predict a decreased apparent affinity for the blocking ion if the overall rate of deocclusion is decreased, as it is in the vanadate-inhibited enzyme (Forbush, 1987b, and see "Discussion" regarding Scheme 2); thus we pointed out that high apparent affinity of the vanadateinhibited enzyme for ffiRb implies that the re-release of the blocking ion involves a step in addition to simple dissociation, for instance a small conformational change that would be needed to expose an occluded ion.
Organic cations are very different from Rb' (or K') in their ability to block the release of one of the two =Rb ions from E(%,,). VO,. As illustrated in Fig. 4, 100 mM Rb' effectively blocks release of half of the 86Rb (bottom left; the second phase is only about half complete on this time scale; 1 mM is sufficient, see Forbush, 1987b), whereas in 100 mM benzyltriethylamine, the biphasic nature of the curve is hardly detectable (upper right). Importantly the slope of the late phase of release is less in 200 mM (lower right) than 100 mM benzyltriethylamine, indicating that the ineffectiveness of the amine is probably a result of very low apparent affinity; this is further supported by the data in Table 11, which presents the rate constants for the late phase of 86Rb release in 50,100, and 200 mM tetrapropylamine, benzyltriethylamine, and benzyltripropylamine (from the same experiment as Fig. 4). If complete block is assumed at saturating amine concentrations, then the estimated Kapp values for each of the amines in Table I1 is >170 mM. Thus the organic cations exhibit a much lower apparent affinity when the rate of 86Rb release is decreased by vanadate inhibition, than in the control enzyme; this behavior is consistent with rapidly reversible binding of the blocking cation, as contrasted to occlusion. It may be noted in Fig. 4 that the initial rate of 86Rb release The vanadate-inhibited occluded state was formed by incubation of Na,K-ATPase with 0.5 mM 86Rb, 2.7 mM Mg', 110 p~ vanadate for 30 min at 20 'C (Forbush, 1987b); after rinsing in imidazole/EDTA, ffiRb release was monitored in imidazole/EDTA or MgPi with the indicated cations. Duplicate runs are presented.
whether this is an effect of ionic strength or whether it reflects specific ion interactions. However it may be noted that in Fig.  10 of Forbush (1987b) the initial rate of 86Rb release from the vanadate-inhibited enzyme is higher in 100 mM Na' than in 100 mM Rb' . The Effect of Amines on =Rb Release in the Presence of ATP-We have previously shown that the same =Rb ions that dissociate in the presence of MgPi are rapidly released in the presence of ATP, presumably from the intracellular face of the Na,K-pump following the conformational changes involved in ion translocation (Forbush, 1987a(Forbush, , 1987b. As illustrated in the top panels of Fig. 5, the release of both "jRb ions is promoted by ATP when tetrapropylamine is present (top right), just as when %Rb (top left) or other cation (not shown) is present. Note that the rate of release is somewhat lower when tetrapropylamine replaces Rb' in the medium; this is consistent with tetrapropylamine acting as an "inert" cation like choline or N-methylglucamine and with our previous finding of a rather nonspecific stimulation of deocclusion by Rb' and other metal cations (Forbush, 1987a).
We also examined the fate of the =Rb ion that remains in the s site after exposure to RbMgP, a procedure which results in replacement of =Rb at the f site by unlabeled rubidium (Forbush, 1987b). As shown in the middle panels of Fig. 5, this ion is released in the presence of tetrapropylamine-

TABLE I1
Reduction by organic cations of the rate of "Rb release (s-') from uanudate-inhibited enzyme The data are least squares curve fits to a single exponential for the =Rb release data obtained after 9 s. Duplicate values were within & 5% of the mean (* is a single determination).

time (s)
MgATP, albeit more slowly than with RbMgATP. This result is in agreement with the above finding that both "Rb ions are released under these conditions (top panels). A remarkable finding is illustrated in the lower panels of Fig. 5: in these samples pre-release of =Rb from the f site was brought about by pre-exposure to benzyltriethylamine-MgPi (as in Fig. 1, lower right), rather than RbMgP,. In this case the ATP-stimulated release of =Rb from the s site was completely blocked by benzyltriethylamine in the medium (lower right panel; compare to lower left panel). Although other explanations will be considered below, the most plausible explanation is that continued occupancy of one cation site by amine (presumably the external f site for K') prevents the Ez + El conformational change. This would account for the strong inhibitory effect of the quaternary amines on the overall pump activity (Kropp and Sachs, 1977).
Release of @Rb from the "Singly Occupied" Transport Sites-It was just shown that in order for an organic cation to prevent the release of @jRb from the s site into an ATP-containing medium, it is necessary that the amine be present in the ATPcontaining medium as well as during preexposure to MgPi ( Fig. 5; compare upper and lower middle panels). This indicates that upon removal of amine from the medium, the amine dissociates rapidly from Na,K-ATPase; additional arguments for this conclusion will be presented below. Thus after exposure to amine-MgPi to allow release of =Rb from the f site, and then removal of amine, the transport sites must be occupied by a single =Rb ion (and possibly a proton). We found that the "spontaneous" release of this lone 86Rb ion was quite rapid compared with the rate of spontaneous release when both sites were occupied. For instance, in 100 mM choline the rate of release from the singly occupied transport sites was typically -5 s" compared with -0.2 s" in control samples (cf. Forbush, 1987a). The instability of the single ion was especially marked at low pH, as shown in Fig. 6 (O), increasing to a spontaneous deocclusion rate of -15 s" at pH 6.5. The direction of the variation with pH is also just the reverse of that when both sites are occupied (0). Note that two different scale factors are used in this figure, the rate of @jRb release from the singly occupied Na,K-ATPase being 20-1000-fold more rapid than from the fully occupied enzyme. Two uncertainties are associated with this set of observations. First, the spontaneous release of the single =Rb ion was sometimes much lower than in the experiment summarized in Fig. 6 (e.g. -1 s-l at pH 7.2). Although variation in Na,K-ATPase preparations may be part of the explanation (the experiments were conducted over a 2-year span), it is also possible that impurities in the choline chloride (see "Experimental Procedures") contribute to the high rate of spontaneous release in a Na'-like manner (see Forbush, 1986;Forbush et d . , 1988). Second, in the course of these experiments we noted that not all of the remaining =Rb was rapidly released into choline alone. Compared with the amount of =Rb release measured in solutions containing ATP (below) or into Na' (Forbush et al., 1988), only 68 f 6% of bound =Rb (14 pairs of duplicates at various pH values in three experiments) dissociated spontaneously in 2-3 s in choline with no ATP. This does not appear to be an artifact of the experimental procedure. For instance, it is not due to incomplete emptying of the f site during the rinse, which would leave some fraction of the =Rb in the doubly occupied state: this fraction would be released rapidly into amine-MgATP, but it is not seen in Fig. 5 (lower right) or Fig. 7 (lower middle). At present we have no explanation for this observation.
We asked whether the single occupancy enzyme could undergo a conformational change and release =Rb in the presence of ATP. Fig. 7 illustrates the result of an experiment in which the effect of ATP was examined following preexposure to benzyltriethylamine-MgPi. The experiment was performed at pH 8.2 where the spontaneous release of =Rb is low (above). The first two columns in Fig. 7 present data similar to those in Fig. 5, but at higher pH and using benzyltriethylamine; as in Fig. 5, it is seen that when amine is present both during the preexposure to MgPi and during the test with ATP, 86Rb release from the s site is prevented (lower middle). When the test solution contained choline (right column), 86Rb was rapidly released both from the control FIG. 6. The pH dependence of spontaneous '"Rb release. The spontaneous release of @Rb into choline/imidazole/EGTA was measured, and the rate constant of release was obtained from the least square curve fits. 0, control, after rinsing in imidazole/EDTA. 0, release following exposure to tetrapropylamine-MgPi to cause dissociation of @Rb from the f site (from a separate experiment). Lines were drawn by eye. sample (upper right) and from the sample in which =Rb remained only in the s site. T i i s suggests that when only a single ffiRb occupies the transport sites, the translocation process involving an ATP-induced conformational change can take place.
Competition of Amines with =Rb in the Equilibrium Level of Occlusion-The results presented above demonstrate that amines can interact with Na,K-ATPase, probably at one of two =Rb transport sites. The process of ion occlusion is clearly a multistep process, involving both binding of the ions to an accessible site and a conformational change in the protein that renders the site inaccessible (Forbush,198%). If other ions interact directly with these sites, any manifestation of competition will depend on a number of parameters of the process which we still do not know (e.g. the binding and occlusion equilibrium constants and whether the singly occupied site can become occluded). However, it can be readily appreciated that in a model in which the equilibrium is displaced in the direction of occluded states from states in which the binding sites are exposed to the medium (as in the flickering gate model, in which the gate only opens briefly), an ion that interacts only with the nonoccluded states will have little effect on the apparent affinity for the overall process of occlusion of =Rb. On the other hand, an ion that interacts with both sites, and itself becomes occluded, will compete effectively with =Rb for occlusion. Therefore, we examined the effect of several amines on the apparent affinity for ffiRb at equilibrium.2 The results in Fig. 8A illustrate a decrease in the apparent affinity for ffiRb in the presence of tetrapropylamine; the data are fit by a competitive interaction of =Rb and tetrapropylamine with one set of sites on Na,K-ATPase, with Kd values 0.06 mM and 52 mM, respectively (a similar result was obtained with benzyltriethylamine, not shown). Thus, the apparent affinity for tetrapropylamine in competing with =Rb for occlusion is at least an order of magnitude lower than the apparent affinity reported above (Table I) for the effect of this same amine in blocking release of =Rb from the s site. However similar results were also obtained with N-methylglucamine (the apparent affinity for =Rb was reduced by Nmethylglucamine with Kapp = 30 mM), an inert cation with * In these experiments, 86Rb occlusion reaches an equilibrium level during the incubation period. While the determination of the amount of the occluded ion is made under disequilibrium conditions (after dilution and rinsing), the determination faithfully reflects the amount of =Rb occluded under the conditions of the incubation, since the rate of spontaneous deocclusion is very slow at 0 "C (Forbush, 1987a). for 30 s a t 20 "C; the sample was diluted in 25 ml in 100 mM K/imidazole/EDTA at 0 "C, filtered on a cellulose ester filter, and rinsed with the same medium (Forbush, 1987a). No correction was made for nonspecific binding, which probably constitutes 10-20% of the total at 1 mM @Rb (Forbush, 1987a). regard to %Rb release ( Table I), suggesting that this phenomenon may be an ionic strength effect or reflect interaction at low affinity sites other than the K' transport sites. This result contrasts sharply to the case with Rb' (or K'), where the apparent affinity for Rb' in blocking release from the s site is essentially the same as the affinity seen for "jRb in occlusion (Forbush, 1987b). The result is therefore strongly suggestive that, unlike Rb' or K', the monovalent amines do not themselves become occluded.
Ethylenediamine has been reported to be an effective inhibitor of Na,K-ATPase activity, with a K j < 0.1 mM (Schuurmans Stekhoven et al., 1988); as shown in Fig, 8B, ethylenediamine competes with ''jRb with high affinity (0.03 mM in this experiment, 0.014, 0.07, 0.12 mM in three others; the reason for the wide spread is unclear). Thus, although ethylenediamine interacts weakly if at all with the f site to block %Rb from the s site (Table I), it can effectively compete with %Rb in occlusion, possibly by binding to both transport sites and becoming occluded.

DISCUSSION
We have found that many organic amines interact with the Na,K-ATPase, in a manner very similar to '%b or K'. The action of these compounds on the occluded state of the enzyme is to block the release of ffiRb from one of two transport sites, without affecting the release of the other 86Rb ion. It is release of 86Rb from the s site that is blocked by amines, just as it is by K' or Rb' (Fig. l), so it is reasonable to propose that the organic molecule binds to the same site as K' or Rb', that is to the other transport site (the f site).
Since the benzyl amines are among the most effective in blocking release of @Rb from Na,K-ATPase, and since the specificity is very broad, it is likely that suitable photoaffinity reagents (e.g. nitroazidobenzyl amines) could be designed which would interact with the Na,K-ATPase in the same way. As shown here ( Fig. 2 and Table I), the apparent affinity for the organic cations is quite low (Kap,, > 1 mM), and there is no indication from the data that a monofunctional amine can be found with a substantially greater affinity for the transport site. If this reflects the true K d for the amines (see below) it is well out of the range in which a photoaffinity approach would be useful, because of nonspecific binding and photoincorporation. However, if it were found that the organic cation became "occluded," it might be possible to wash away most of the nonspecifically bound compound prior to photolabeling.
Unfortunately, several lines of evidence demonstrate that the monovalent amines do not become occluded. 1) The inhibition of %Rb dissociation from the s site in the presence of ATP requires the continued presence of amine in solution ( Figs. 5 and 7). When the overall process of deocclusion is slowed by vanadate inhibition, the apparent affinity for amine decreases by more than an order of magnitude compared with the uninhibited enzyme ( Fig. 4 and Table 11) as predicted for a rapidly reversible inhibitor (see discussion of Scheme 2, below), in contrast to an increase in apparent affinity for Rb' in similar experiments (Forbush, 1987b). 3) The monovalent amines compete poorly with %Rb for formation of the OCcluded state (Fig. 8 A ) . 4) In experiments to be presented elsewhere (Forbush et al., 1988; see also Forbush, 19861, we have found a dramatic effect of Na' in driving the release of the lone %Rb ion from the s site; quaternary amines are fully competitive with Na' and stabilize the "Rb ion, but the amine is effective only as long as it is kept in the medium. 5 ) We have performed a number of experiments in search of direct evidence that tetrapropylammonium ion or benzyltriethylammonium ions can themselves become occluded on Na,K-ATPase in a manner similar to K' or Rb' (results not shown here). Generally, we looked for continued stabilization of 86Rb at the s site after removing tetrapropylamine from the medium, in the presence or absence of MgZ' and Pi. In a typical experiment, "Rb was bound to Na,K-ATPase, exposed to benzyltriethylamine-MgPi during the dilution-filtration-prerinse procedure prior to transfer to the rapid filtration apparatus, rinsed with imidazole/histidine in the first solution, and tested with amine-ATP (or Na'; see Forbush et al., 1988) in the second solution. In none of these experiments did we obtain any evidence that the effect of the organic cation could be "remembered" after exposure to the rinse solution. Thus, we conclude that the organic cation that blocks the release of one "jRb ion cannot itself become occluded.
We have found it convenient to think of the action of Rb', its congeners, and organic cations in terms of a geometrical model indicated in Scheme 1 in the Introduction (Forbush, 1985;Glynn et al., 1985a;Forbush, 1987b). It is proposed that release to the extracellular medium of %Rb from the innermost site (5) is blocked by occupancy of the f site by a K' or Rb' ion. Importantly, interaction of extracellular ions with the f site is limited by some slow step, as would be expected of a small conformational change associated with opening of a gate (Forbush, 1987b). In this context, the organic cations appear to be able to interact like Rb' or K' (hypothetically at the f site), but somehow, possibly due to steric hindrance, it appears that the gate is unable to close on the organic cations (Scheme 1B). However alluring this notion, it should be stressed that kinetic data cannot prove such models. We have previously noted that, although the data are fairly compelling that the blocking Rb' ion is bound at the occlusion site, it is not possible to prove by kinetic evidence alone that the constraint on release from the s site is simple steric hindrance, as opposed to an allosteric interaction. With regard to the organic cations, the idea that these bind to the transport sites is supported by analogy to the blocking action of K' or Rb', but because the monovalent amines do not become occluded, we lack direct evidence to prove this point. Although it appears unlikely, it is possible that amines bind to another site (not the f transport site) and by an allosteric effect prevent release from only one of the two K' transport sites.
It is likely that the intrinsic affinities for monovalent amines interacting at the f site are much higher than the apparent affinities indicated by the ICapp values in Table I. This is a multistep reaction sequence, and in such cases the apparent dissociation constant for a substrate (K,) or for an inhibitor (Ki) is generally different than the intrinsic dissociation constant characteristic of the binding site (cf. Stein, 1986). In the proposed model for block by organic cations depicted in Scheme 2, amines bind to the f site only when the gate is open. We measure the rate of the release of the second of the two occluded "Rb ions. In our flickering gate model (see Forbush, 198713, 1988~) the sequence of events involves opening of the gate (slow) and dissociation of the first ion (very fast; kA) followed by closing of the gate ( k B ; fast) and finally jump of the ion within the pocket, reopening of the gate (slow), and dissociation of ''jRb (kc). In the absence of amine, the rate determining step in release of the second =Rb ion is the reopening of the gate (kB >> kc). Amine reduces the concentration of the intermediate B, and thereby reduces the rate of B + C ; the overall rate of B 4 C "+ D is reduced by half (Kapp = [amine]) when B + C is as slow as C + D, that is when (B/(B + B'))*lzB = kc. On simplification (assuming that amine binding is in rapid equilibrium, B/B' = K d / [amine]) we have K,, = Kd*(kB -JZc)/kc, or since kB >> kc, Restated, the apparent dissociation constant for binding of amine will be larger than the intrinsic dissocia-tion constant by the ratio of the rate constants3 kB and kc.
The above modeling is supported by the observation that when is reduced by vandate inhibition, the Kapp for the amines is greatly increased (Fig. 4 and Table 11). If ks were known, we could calculate the true Kd for monovalent amines at the f site. We do not yet have an estimate of the value of kg, but for this type of model to fit available data, under the conditions in Fig. 2, kB must be at least %fold greater than &, and possibly much more. Thus, we estimate that the Kd for monovalent amines such as tetrapropylamine or benzyltriethylamine is 4 mM (possibly CO.1 mM). Even so, since the intermediate to which the amines bind is present only transiently, it will be difficult to exploit photoaffinity derivatives of these compounds.
We have found that when organic cations occupy the f site, the release of ffiRb from the s site is not stimulated by ATP ( Figs. 2 and 7), indicating that Na,K-ATPase is probably unable to undergo the Ez --* E, conformational change under this condition. Since a distinguishing feature of the organic cation interaction is that these compounds do not become occluded, we propose (a) that the E, E2 conformational change cannot take place as long as the gate regulating exposure of K' sites to the extracellular medium remains open. There are other alternative possibilities: ( b ) We have found that the rate of deocclusion in the presence of ATP depends upon the nature of the cation sharing the pocket with a single ffiRb ion (Forbush, 1987b); it may be that organic cations are an extreme case, in which the conformational change is prohibited altogether. This is subtly different from the proposal in (a), in that it is the nature of the ion in the pocket rather than the state of the gate which affects deocclusion. (c) The action of organic cation is blocking deocclusion with ATP might be mediated at another site altogether.
However this seems very unlikely, since the inhibition only occurs in the case when the enzyme has also been preexposed to the organic cation under conditions in which the f site is presumed to become occupied by that cation.
We have shown here that ATP stimulates deocclusion when only one %Rb ion occupies the occlusion sites (Fig. 7). The simplest explanation is that this ion is released to the intracellular face of the membrane, and therefore that a single 86Rb ion could be transported in one pump cycle. However, for this to take place during transport would also require that the singly occupied occluded state be formed in preference to the doubly occupied state at some concentration of extracellular K' (or %Rb), and we do not yet have sufficient infor- The model treated here (Scheme 2) is a simplification. In fact the reaction B * C is probably slowly reversible (with k+, = kc), as are the reactions in A + D and C + D (except ion dissociation into an infinitely dilute medium). Computer modeling of more complex cases using an iterative approach, as well as explicit evaluation of a multiple state equilihrium preceding a slow step (not presented), has lead-to the same conclusion reached here: the apparent dissociation constant for amine is greater than the intrinsic dissociation constant by a factor given approximately by the ratio of kB and (for = kc). mation to know whether this occurs. Furthermore, it should be noted that we cannot be certain that the ATP-stimulated ffiRb release is in the intracellular direction, since we have presented arguments that ATP may also modulate release in the extracellular direction in the presence of vanadate or Pi (Forbush, 1987b).
Ethylenediamine does not interact strongly with Na,K-ATPase to slow dissociation of either of the =Rb ions from the occluded state (Table I, KaPp > 100 mM). On the other hand, this bifunctional amine competes effectively with =Rb in preventing formation of the occluded state, indicating that ethylenediamine complexes with the Na,K-ATPase in a state parallel to the =Rb-occluded form. This conclusion is also supported by the finding that ethylenediamine inhibits phosphorylation (Schuurmans Stekhoven et al., 1987). Therefore it seems likely that ethylenediamine acts like K+ or Rb' in that it can occupy K' transport sites and become occluded; if so, suitable derivatives may be useful in labeling the ion translocation sites on Na,K-ATPase.