The Role of Loop 6/7 in Folding and Functional Performance of Na,K-ATPase*

Alanine substitutions were made for 15 amino acids in the cytoplasmic loop between transmembrane helices 6 and 7 (L6/7) of the human (cid:1) 1 -subunit of Na,K-ATPase. Most mutations reduced Na,K-ATPase activity by less than 50%; however, the mutations R834A, R837A, and R848A reduced Na,K-ATPase activity by 75, 89, and 66%, respectively. Steady-state phosphoenzyme formation from ATP was reduced in mutants R834A, R837A, and R848A, and R837A also had a faster E 2 P 3 E 2 dephos- phorylation rate compared with the wild-type enzyme. Effects of L6/7 mutations on the phosphorylation domain of the protein were also demonstrated by 18 O exchange, which showed that intrinsic rate constants for P i binding and/or reaction with the protein were al- tered. Although most L6/7 mutations had no effect on the interaction of Na (cid:2) or K (cid:2) with Na,K-ATPase, the E825A, E828A, R834A, and R837A mutations reduced the apparent affinity of the enzyme for both Na (cid:2) and K

proposed that the effect of the mutations on Ca 2ϩ affinity was probably because mutations in L6/7 influenced the correct positioning of TM6, and that L6/7 was important in stabilizing the phosphorylation domain and the cation binding domain in Ca-ATPase. In the Ca-ATPase crystal structure, L6/7 is close to the P domain and the transmembrane domain, in a good position to mediate communication between the phosphorylation site and the cation binding site, and Toyoshima et al. (13) suggested that L6/7 may function as a coordinator between these sites. In the Na,K-ATPase, Shainskaya et al. (20) measured the effects of chymotrypsin digestion of L6/7 on the release of a TM5/TM6 fragment from the membrane and on Rb ϩ occlusion. Release of the fragment was prevented by occluded monovalent cations or by the competitive antagonists Ca 2ϩ , Mg 2ϩ , La 3ϩ , pxylylene bisguanidinium (pXBG 2ϩ ), m-xylylene bisguanidinium (mXBG 2ϩ ), 1-bromo-2,4,6 tris(methylisothiouronium)benzene (Br-TITU 3ϩ ), and 1,3-dibromo-2,4,6-tris (methylisothiouronium)benzene (Br 2 -TITU 3ϩ ). They proposed that L6/7 of Na,K-ATPase serves as an initial binding site for the monovalent cations and the cation antagonists, but that only the monovalent cations subsequently become occluded. In the experiments reported here, mutations to L6/7 amino acids were found to have only small effects on the apparent affinity of the Na,K-ATPase for the transported cations, and failed to prevent the binding of Br-TITU 3ϩ to the enzyme. These results are most easily understood if the binding of transported cations and cation antagonists occurs to sites on Na,K-ATPase outside of L6/7.

EXPERIMENTAL PROCEDURES
All enzymes used for molecular cloning were obtained from New England Biolabs or Roche Applied Sciences and were used as recommended by the suppliers. Pfu polymerase was obtained from Stratagene. Ouabain was purchased from Sigma-Aldrich, and the concentration of ouabain in solution was determined by spectrophotometry (21). [ 3 H]Ouabain was obtained from PerkinElmer Life Sciences. [␥-32 P]ATP was purchased from ICN Biomedicals. All other chemicals were reagent grade.
Preparation of the yeast expression plasmids for human Na,K-ATPase ␣ 1 -and ␤ 1 -subunits, designated as YhN␣1, has been described (22). Amino acid changes in L6/7 of the ␣ 1 -subunit were made using the Stratagene QuikChange TM site-directed mutagenesis kit. The presence of the mutations was confirmed by DNA sequencing by the Microchemical Core Facility at USC. Transformation of Saccharomyces cerevisiae with YhN␣1 and GhN␤1, cell growth for expression of Na,K-ATPase protein, and crude yeast membrane preparation, and SDS extraction with 0.1% (w/v) SDS were done as previously described (22,23). Protein concentration was determined by the method of Lowry et al. (24).
Na,K-ATPase Activity Measurement-Ouabain-sensitive Na,K-ATPase activity was measured using a coupled optical assay as previously described (22,25).
Ouabain Binding Experiments-Equilibrium ouabain binding, inhibition of ouabain binding by KCl, inhibition of ouabain binding by NaCl, and NaCl concentration-dependent ouabain binding were performed as previously described (22). The data were fit to a modified Hill equation (11,26) from which were extracted values for K d , K 0.5 , and IC 50 .
Phosphate Concentration Dependence of the Mg 2ϩ and P i -dependent Ouabain Binding-The apparent affinity of the Na,K-ATPase for P i was measured by equilibrium P i -and Mg 2ϩ -dependent ouabain binding in 20 nM [ 3 H] ouabain, 4 mM MgCl 2 , 50 mM Tris/HCl (pH 7.4) with 0 -10 mM P i . Data were fitted to a rectangular hyperbola to obtain the K 0.5 for P i . 18 O Oxygen Exchange Measurements- 18 O exchange between P i and water was measured as previously described (25) using yeast membranes that had been extracted with 0.1% SDS. The exchange rate and the partition coefficient (P c ) were estimated from the isotopomer distribution at a single time point by non-linear least squares. Parameters relating exchange to metal cofactor and substrate binding and to the equilibrium between non-covalently and covalently bound P i were estimated by globally fitting the rate equation for ordered binding of Mg 2ϩ before P i to plots of the exchange rate as a function of both Mg 2ϩ and P i concentration (25).
Heat Inactivation of Ouabain Binding-1 mg of crude yeast membrane protein was suspended in 0.4 ml of 25 mM imidazole/HCl, 1 mM EDTA (pH 7.4) and was heated at 50°C for 90 s. The ouabain binding capacity of the sample was measured at 37°after heating and was compared with that of the control sample without heating. The experiments were done in triplicate.
Steady-state Phosphorylation from [␥-32 P]ATP-Steady-state phosphoenzyme formation was carried out in 200 l of 20 mM Tris/HCl (pH 7.4), 3 mM MgCl 2 , 100 mM NaCl, and 2 M [␥-32 P]ATP. The reaction was initiated by adding the [␥-32 P]ATP, and was continued for 10 s before being quenched by 0.8 ml of 1 M H 3 PO 4 (pH 2.4). After quenching, the tubes were immersed in ice for 10 min. After adding 100 g of bovine serum albumin, the samples were pelleted in a microcentrifuge at 13,000 ϫ g for 20 min at 4°C. The pellets were washed three times by centrifugation with 1.1 ml of cold washing buffer (0.25 M H 3 PO 4, pH 2.4), once with 1.1 ml of cold distilled water, and were dissolved in 35 l of sample loading buffer (prepared by adding 0.2 ml of 50% glycerol containing 1 mg/ml bromphenol blue and 10 l of 2-mercaptoethanol to 0.8 ml of 10 mM Na 2 HPO 4 , 1% SDS) (18). The entire sample was applied to a 4% SDS-polyacrylamide gel, pH 6.2 (18,27). After running the gel at 100 mA for 3 h at 4°C, the gels were dried overnight in a fume hood. Radioactive phosphoenzyme was detected by autoradiography, and the quantitation of the phosphoenzyme specific to Na,K-ATPase was obtained by liquid scintillation counting of slices corresponding to the ␣-subunit bands of the dried gel. Reactions carried out with 100 mM KCl replacing NaCl served as control for background phosphorylation.
Time course of E 2 P 3 E 2 Dephosphorylation-Phosphorylation at low concentration of Na ϩ in the absence of added K ϩ will accumulate E 2 P (28). The time course of E 2 P 3 E 2 can be obtained by measuring the disappearance of the phosphoenzyme after adding K ϩ to the enzyme (E 1 ϩ Na ϩ ϩ ATP % E 1 P % E 2 P ϩ K ϩ % E 2 ). Enzyme was phosphorylated in 50 l of 20 mM NaCl, 20 mM Tris/HCl (pH 7.4), 3 mM MgCl 2 , and 1 M [␥-32 P]ATP on ice for 15 s. Dephosphorylation was initiated by adding 10 l of a chase solution to give a final concentration of 0.1 mM KCl and 10 mM EDTA, followed by adding 100 l of quench solution (2 M H 3 PO 4 , pH 2.4) at serial time points. The reaction was carried out in 2-ml tubes with magnetic stirring. After adding an additional 0.8 ml of 1 M H 3 PO 4 (pH 2.4) containing 100 g of bovine serum albumin, the reaction solutions were transferred to 1.5-ml tubes and were centrifuged at 13,000 ϫ g for 20 min to pellet the protein. Reactions carried out in the presence of 20 mM KCl replacing NaCl served as control for background phosphorylation. Pellets were washed, proteins were separated by gel electrophoresis, and radioactive phosphoenzyme was quantified as described above. The data were fitted to a monoexponential equation: y ϭ ae -kt ϩ y 0 , where y is the amount of phosphorylated Na,K-ATPase, a is the maximum phosphorylation, y 0 is phosphoenzyme present at long times, k is the rate constant, and t is the chase time.
Ouabain-sensitive K ϩ -stimulated Phosphatase Activity using p-Nitrophenylphosphate (PNPP) as Substrate-The assay was carried out in duplicate in a total volume of 1 ml of 20 mM HEPES/triethylamine (pH 7.4), 3 mM MgCl 2 , 0.4 mM EDTA, 25 mM KCl, 50 M dithiothreitol, 2 mM p-nitrophenylphosphate, plus or minus 0.1 mM ouabain. The reaction was started by adding 100 g of SDS-extracted yeast membrane protein to each tube at 30-s intervals. After a brief vortex, the samples were incubated at 37°C for 30 -60 min. The reactions were stopped by adding 2 ml of 1 N NaOH. After brief centrifugation, the absorbance at 410 nM was measured for the clear supernatants. The ouabain-sensitive K ϩ -stimulated phosphatase activity was calculated as the rate of pnitrophenylphosphate hydrolysis in the absence of ouabain minus that in the presence of ouabain, normalized to the ouabain binding capacity of each sample.
Western Blot-100 g of SDS-extracted yeast membrane proteins were loaded onto 10% SDS-polyacrylamide gels. 100 ng of SDS-purified dog kidney Na,K-ATPase was used as positive control and membranes from non-transformed yeast served as a negative control. After electrophoresis samples were transferred to Immobilon P membranes and incubated with monoclonal antibody 5 (D. Fambrough, Johns Hopkins), which is specific for the ␣-subunit of Na,K-ATPase. Image detection was done using an Odyssey Infrared Imaging System.
Calculations and Data Analysis-Origin and Microsoft Excel were used to analyze the data. Best fits for non-linear curve fitting are shown as lines in the figures, and the extracted parameters are given in the tables and figure legends. Standard deviations or standard errors are indicated for the means and for parameters extracted by non-linear curve fitting. Student's t test was used for most statistical analysis. Free P i and Mg 2ϩ concentrations were calculated using Winmaxc v2.00, www.stanford.edu/ϳcpatton/maxc.html.

RESULTS
Expression of Na,K-ATPase in Yeast-The amino acid sequences of L6/7 in human Na,K-ATPase ␣ 1 -subunit and in rabbit sarcoplasmic reticulum Ca-ATPase are shown in Fig. 1. The effects of alanine substitutions of charged residues, prolines, and conserved residues in L6/7 of the human Na,K-ATPase ␣ 1 -subunit were examined after expression of the protein in yeast cells. As shown in Fig. 2, all mutants were detected in membranes prepared from transformed yeast cells, although the triple mutant E825A/E828A/D830A and the E825A, E828A, and M832A mutants were expressed at lower levels than wild type and most other mutants. All mutants except for the E825A/E828A/D830A mutant bound ouabain in an Mg 2ϩ -and P i -dependent equilibrium ouabain binding assay. Unlike the other mutants and wild-type enzyme, the E825A and E828A mutants lost the ability to bind ouabain after SDS extraction of crude microsomal membranes. The ouabain binding capacity (B max ) of the mutants in SDS-extracted yeast membranes varies, ranging from 3 to 29 pmol/mg protein (Table I). The dissociation constant for ouabain (K d ) for most mutants was similar to the wild-type value, ranging from 9 to 31 nM, except for R837A, which had a K d of 218 nM, 12-fold higher than that of wild-type Na,K-ATPase, and for R837K, which had a K d of 66 nM, a 3.6-fold increase.
Effects of Mutations on Na,K-ATPase Activity-The ouabain sensitive Na,K-ATPase activity was measured on SDS extracted yeast membranes. Without SDS extraction the fraction of ouabain-sensitive ATPase activity is too small to be detected in the presence of endogenous yeast ATPase activity, and SDS extraction increases the fraction of ouabain sensitive ATPase activity in the membranes by inhibiting endogenous yeast ATPase activity. After SDS extraction the ouabain-sensitive fraction is usually 30 -50% of total ATPase activity. The turnover number of wild type Na,K-ATPase expressed in yeast membranes was 3364 Ϯ 124 min Ϫ1 (S.E.). The results shown in Table I express the turnover number of the mutants as a percentage of the wild-type value. R834A, R837A, and R848A had less than 50% of wild-type ATPase activity, whereas D830A, I831A, P836A, and P839A retained 60 -80% of wildtype ATPase activity. The activities of K833A, K840A, D842A, and E847A were greater than 90% of the wild type, and the activities of both R837K and K843A were significantly greater than wild type. ATPase activity for E825A and E828A could not be measured because of their loss of ouabain binding after SDS extraction, and ATPase activity for M832A could not be accurately measured because of its low expression level.
Effects of Mutations on Na ϩ and K ϩ Affinity of Na,K-ATPase-In the presence of saturating concentrations of ATP, the apparent affinity of E 1 for Na ϩ can be measured by the Na ϩ dependence of ouabain binding (E 1 ϩ Na ϩ ϩ Mg 2ϩ ϩ ATP % EP ϩ ouabain % EP⅐ouabain). In this reaction, E 1 binds to Na ϩ with high affinity and is phosphorylated in the presence of ATP and Mg 2ϩ to EP, which binds ouabain to form a stable EP⅐ouabain complex. In the presence of Mg 2ϩ and P i , the apparent affinity of E 2 for K ϩ can be measured by K ϩ inhibition of ouabain binding (E 2 (2K) % K ϩ ϩ E 2 ϩ Mg 2ϩ ϩ P i % EP ϩ ouabain % EP⅐ouabain). In this reaction, K ϩ binds to E 2 and antagonizes ouabain binding by competing for E 2 with P i and phosphoenzyme formation. Na ϩ also inhibits Mg 2ϩ -and P i -dependent ouabain binding (E 1 Na ϩ % Na ϩ ϩ E 1 % E 2 ϩ Mg 2ϩ ϩ P i % EP ϩ ouabain % EP⅐ouabain). Both the E 1 /E 2 equilibrium and the affinity of E 1 for Na ϩ affect the amount of EP formed from P i in the presence of Mg 2ϩ and P i , and affect the amount of ouabain bound and the value of IC 50 for Na ϩ .
Most mutations did not change the apparent affinity of the pump for Na ϩ or K ϩ . For Na ϩ -dependent ouabain binding (Fig.  3), the K 0.5 value for Na ϩ ranged from 0.7-1.6 mM for wild-type and most mutants, which is close to the reported K 0.5 values of 0.7 mM and 0.6 mM measured by Na ϩ -dependent steady-state phosphoenzyme formation (11,12). The K 0.5 values of E825A and R837A for Na ϩ were increased by 2-3-fold. For K ϩ inhibition of ouabain binding (Fig. 4), the IC 50 for K ϩ of wild type and most mutants was 0.3-0.4 mM, consistent with the reported value of K 0.5 of 0.31 mM (12) and 0.77 mM (29). E825A, E828A and R837A displayed a 2-fold increase in IC 50 for K ϩ . For Na ϩ inhibition of ouabain binding, the IC 50 value of wild type for Na ϩ was 20 Ϯ 1 mM (Fig. 5). The IC 50 for Na ϩ inhibition of FIG. 1. Comparison of L6/7 sequence of human Na,K-ATPase ␣ 1 -subunit and rabbit SERCA Ca-ATPase. The residues in Na,K-ATPase changed to alanine in this study are marked with an asterisk.
For E825A and E828A, ouabain K d and B max were determined using crude yeast microsomal membranes. ouabain binding for most mutants was not different from wild type, but it was increased 1.5-3-fold for E825A, E828A, and R834A, and it was decreased 3-fold for R837A. The effects of E825A, E828A, and R837A on apparent cation affinity are summarized in Table II.
Effects of Br-TITU 3ϩ and La 3ϩ on Ouabain Binding-Br-TITU 3ϩ and La 3ϩ act as competitive cation antagonists and inhibit occlusion of both Rb ϩ and Na ϩ by Na,K-ATPase (20,30). Br-TITU 3ϩ inhibits Mg 2ϩ -and P i -dependent ouabain binding to wild type Na,K-ATPase with an IC 50 of 30 M (data not shown). Fig. 6 shows the extent of inhibition of ouabain binding to wild-type Na,K-ATPase and the E825A, E828A, R834A, and R837A mutants by 50 M Br-TITU 3ϩ . Inhibition of ouabain binding was similar in the wild type and the E828A mutant, was greater than wild-type in the E825A and R837A mutants, and was less than wild type in the R834A mutant. La 3ϩ at concentrations up to 0.5 mM did not inhibit ouabain binding by wild type Na,K-ATPase, and 0.5 mM La 3ϩ had no effect on the inhibition of ouabain binding by KCl (data not shown).
Heat Inactivation of Ouabain Binding-E825A and E828A lost ouabain binding after SDS extraction suggesting that these two mutations destabilize the enzyme. To test this hypothesis, crude yeast membranes containing the Na,K-ATPase were heated at 50°C for 90 s and ouabain binding at 37°was compared in heated samples and unheated controls. As shown in Fig. 7, wild-type, D830A, M832A, R834A, and R837A lost less than 25% of ouabain binding capacity after heating, while the amount of ouabain bound by the membranes of E825A and E828A was reduced by 80 -90%.
R837A Reduced the Apparent Affinity of Na,K-ATPase for P i -Relative to wild-type, R837A exhibited a 12-fold higher K d value for ouabain in Mg 2ϩ -and P i -dependent equilibrium ouabain binding (Table I). One possible reason for the reduced ouabain binding affinity might be an effect of the mutation on the interaction between the enzyme and P i , since only phosphorylated enzyme has a high affinity for ouabain. To test whether R837A affects P i interactions, the P i concentration dependence of ouabain binding was measured. The results of these measurements showed that the K 0.5 for P i of most mutants was not significantly different from wild type (Fig. 8), whereas the mutation R837A increased K 0.5 nearly 8-fold. At a total Mg 2ϩ concentration of 4 mM, ouabain binding reached a plateau at 8 -10 mM P i . Therefore, the measurement of ouabain K d was repeated with total P i concentration increased to 10 mM and total Mg 2ϩ kept at 4 mM (2 mM free Mg 2ϩ and 8 mM free P i ). The K d for ouabain under these conditions was similar to the K d obtained in standard binding buffer, which contained 4 mM total Mg 2ϩ and 4 mM P i (3 mM free Mg 2ϩ and 3 mM free P i ). The reduced affinity of R837A for ouabain, therefore, is not due , in which x is the Na ϩ concentration, a is the maximum ouabain binding, c is K 0.5 (Na), n is the Hill coefficient. B, bars indicate mean Ϯ S.D. (n ϭ 3-6). *, p Ͻ 0.05. solely to differences in the extent of saturation of P i binding between the mutant and wild-type enzymes.
The observed increase in K 0.5 for P i activation of ouabain binding to R837A could mean that the mutation affected the intrinsic affinity of the enzyme for P i . Oxygen exchange measurements were made to test that interpretation. Table III compares averaged exchange parameters for both purified and expressed wild-type Na,K-ATPase with the estimates for selected mutants. Only the parameter including the intrinsic constant for P i dissociation from metalloenzyme (KЈ P ) is statistically different (p Ͻ 0.01). Therefore, a decrease in affinity for P i could semi-quantitatively explain why mutant R837A required higher concentrations of P i than wild type for activation of both high-affinity ouabain binding (8-fold) and catalysis of 18 O exchange between P i and H 2 O (5-fold). Phosphorylation of the enzyme by P i is an intermediate step in both reactions.
Steady-state Level of Phosphorylation-During the catalytic cycle of P-type ATPases an acid-stable phosphoenzyme intermediate is formed from ATP. In Na,K-ATPase phosphorylation occurs in the presence of Na ϩ , Mg 2ϩ , and ATP. K ϩ decreases the amount of the phosphoenzyme intermediate by stimulating dephosphorylation and pulling the enzyme to the E 2 conformation. In Ca-ATPase, mutations to Lys 819 and Arg 822 resulted in lower levels of steady-state phosphoenzyme formation from both ATP (20% of wild-type) and P i (20 -35% of wild-type) without affecting dephosphorylation of the phosphoenzyme (18). For Na,K-ATPase the steady-state level of phosphoenzyme was measured in the absence of K ϩ for wild type, R834A, R837A, and R848A. Phosphorylation with K ϩ replacing Na ϩ served as control for background phosphorylation, which was subtracted from all data points, and purified dog kidney Na,K-ATPase served as positive control. As shown in Fig. 9, the phosphoenzyme was formed in

TABLE II Effect of mutations on ion interactions with Na,K-ATPase
The effects of increasing concentrations of monovalent cations on ouabain binding were determined using yeast microsomal membranes, as described under "Experimental Procedures." From Fig. 3, Na ϩ -dependent ouabain binding. b From Fig. 5, Na ϩ inhibition of ouabain binding. c From Fig. 4, K ϩ inhibition of ouabain binding.
the presence of Na ϩ , however, it was negligible in the buffer with K ϩ . In this experiment, all samples contained equal amounts of functional Na,K-ATPase, determined as the same number of ouabain binding sites. The steady-state level of phosphoenzyme was reduced to 85, 67, and 58% relative to wild type for R834A, R837A, and R848A, respectively. Time Course of Dephosphorylation of E 2 P 3 E 2 -Phosphorylation at low concentrations of Na ϩ accumulates E 2 P (28, 31). Phosphoenzyme was formed in 20 mM NaCl, and E 2 P 3 E 2 dephosphorylation kinetics were observed by chasing the reaction to a final concentration of 0.1 mM K ϩ and 10 mM EDTA, followed by quench of the reaction at different times. Background phosphorylation measured with KCl replacing NaCl was subtracted from all data points. As shown in Fig. 10, a good fit of a monoexponential decay function to the data could be obtained. The extracted t 0.5 for wild type is 1.9 Ϯ 0.2 s, while R834A and R848A have t 0.5 values of 2.2 Ϯ 0.2 s and 2.2 Ϯ 0.1 s, respectively. R837A led to faster E 2 P 3 E 2 dephosphorylation, with a t 0.5 of 1.4 Ϯ 0.1 s, which is significantly different from that of the wild type (p Ͻ 0.05).
K ϩ -stimulated p-Nitrophenylphosphate Hydrolysis-Since R834A and R837A greatly reduced Na,K-ATPase activity, and R837A affected dephosphorylation, the effects of these mutations on the phosphatase activity were measured. As shown in Fig. 11, R837A reduced the phosphatase activity to 50% of the wild-type value, while R834A did not affect this activity.

DISCUSSION
Most mutations in L6/7 retained at least 50% of wild-type Na,K-ATPase activity, with the exception of R834A, R837A, and R848A. The ATPase activity of E825A and E828A could not be measured since they lost ouabain binding after SDS extraction. In Na,K-ATPase Arg 834 corresponds to Lys 819 (Arg 819 ) in chicken (rabbit) Ca-ATPase, Arg 837 corresponds to Arg 822 , and Arg 848 corresponds to Trp 832 (Fig. 1). Amino acid substitutions to both Lys 819 and Arg 822 in Ca-ATPase led to loss of over 50% of Ca-ATPase activity (18), similar to the effects of the R834A and R837A mutations in Na,K-ATPase. No amino acid substitutions have been reported for Trp 832 . In Ca-ATPase, the effect of these mutations on ATPase activity was similar to the inhibition of phosphoenzyme formation from ATP or P i , and it was concluded that Lys 819 and Arg 822 play an important role in determining the functional integrity of the phosphorylation domain of the enzyme (18).
The observation that E825A and E828A cause the Na,K-ATPase to lose ouabain binding after SDS extraction or brief heating suggests that Glu 825 and Glu 828 are structurally important. Consistent with this hypothesis, the triple mutant E825A/E828A/D830A had low protein expression and no ouabain binding could be detected in yeast membranes from cells transfected with a plasmid containing this mutant. One possible explanation for these results is that the E825A/E828A/ D830A mutation might elicit the unfolded protein response in yeast resulting in proteolytic degradation. The E825A and E828A mutations probably alter the conformation of L6/7, which in turn, disturbs global protein structure. Several observations are consistent with this hypothesis. Firstly, the helical turn involving residues 816 -819 in L6/7 of Ca-ATPase was replaced by random coil after alanine substitution of Asp 813 , Asp 815 , Asp 818 , and Glu 826 in a synthetic L6/7 peptide (13, 17).
Secondly, in Ca-ATPase L6/7 interacts with phospholamban, and the mutants N810A and D813A (equivalent to E825A and E828A in Na,K-ATPase) resulted in diminished ability to interact with phospholamban (32). Thirdly, in the Ca-ATPase structure Asn 810 and Asp 813 are close enough to form hydrogen bonds with the intracellular loop connecting TM8 and TM9 (Ser 915 , Ser 917 , Asn 930 ), and TM5 (Asn 755 , Asn 756 ). Replacing Asn 810 and Asp 813 with alanine would disrupt this hydrogen bond network.
In Na,K-ATPase, it was suggested that negatively charged residues in L6/7 constitute a cytoplasmic cation entry port that controls access to the cation occlusion sites (20). In H,K-ATPase, the mutants E834Q, E837Q, and D839N (equivalent to Glu 825 , Glu 828 , and Asp 830 in Na,K-ATPase) could not form a phosphorylated intermediate from ATP (33). It was proposed that these residues might be essential for the enzyme to become phosphorylated if they were involved in H ϩ binding. In Ca-ATPase, alanine substitution of Asp 813 , Asp 815 , and Asp 818 (D813A/D818A and D813A/D815A/D818A) reduced Ca 2ϩ affinity from micromolar to millimolar as detected by Ca 2ϩ dependence of phosphoenzyme formation from ATP or P i (15,16). These observations led to the hypothesis that L6/7 functions as an initial cation binding site in Ca-ATPase. A synthetic L6/7 peptide (Gly 808 -Pro 827 ) was able to bind Ca 2ϩ and lanthanum, and this cation binding could be disrupted by alanine substitutions for all carboxylates (Asp 813 , Asp 815 , Asp 818 , and Glu 826 ) (17). In contrast, replacing aspartate by asparagine (D813N/D818N) in Ca-ATPase also reduced the Ca 2ϩ affinity, yet the ATPase activity was only 25% of the wild type even in the presence of millimolar Ca 2ϩ (19). The persistence of low ATPase activity in the presence of high concentrations of Ca 2ϩ was interpreted to indicate that L6/7 in Ca-ATPase is important in linking Ca 2ϩ binding to catalytic activation, presumably by influencing the arrangement of TM6, rather than in direct Ca 2ϩ binding.
In the present study, most of the L6/7 mutations did not change the K 0.5 or IC 50 values for either Na ϩ or K ϩ (Table II). Several points suggest that the changes that were observed for the E825A, E828A, and R837A mutants are due to indirect effects of the mutations on cation binding. First, the same residues whose substitution with alanine resulted in enzyme instability or a greater than 50% reduction in Na,K-ATPase activity are also responsible for changes in cation affinity. Second, the magnitude of the effects on cation interaction with the mutants is small. E825A, E828A and R837A showed only a 1.5-3-fold decrease in apparent affinity for Na ϩ and K ϩ , while mutations to intramembrane carboxyl groups known to coordinate cations in the cation occlusion site abolished K ϩ occlusion or Na ϩ -dependent phosphoenzyme formation (9 -12). Third, the mutants are not consistently altered in their interaction with the cations. E828A had the largest increase in IC 50 for Na ϩ in Na ϩ inhibition of ouabain binding, yet showed an apparent affinity for Na ϩ similar to the wild type in the Na ϩ -and ATP-dependent ouabain binding assay. Whereas R837A showed a 2.7-fold increase in K 0.5 for Na ϩ in the Na ϩ -dependent ouabain binding reaction, the IC 50 for Na ϩ inhibition of ouabain binding was only one-third of wild type. Fourth, the apparent affinity for Na ϩ was reduced by neutralization of both basic (Arg 837 ) and acidic (Glu 825 ) amino acid side chains. L6/7 is a cytoplasmic loop and does not participate in cation occlusion. Under physiological conditions and in the forward cycling of the pump, Na ϩ binds to the enzyme from the intracellular side of the membrane and K ϩ binds to the enzyme from extracellular side. The observations that E825A, E828A, and R837A affect both Na ϩ and K ϩ affinity may be understood if L6/7 of Na,K-ATPase is involved in cation binding indirectly, possibly by affecting the cation occlusion sites through its connections with TM5 and TM6. Both TM5 and TM6 have residues involved in Na ϩ and K ϩ occlusion, and L6/7 residues are hydrogen bonded with TM5 and are directly linked with TM6.
The most compelling evidence that L6/7 is not a cytoplasmic cation binding site comes from the effects of Br-TITU 3ϩ on ouabain binding. Br-TITU 3ϩ and other cation antagonists such as Ca 2ϩ , Mg 2ϩ , La 3ϩ , mXBG 2ϩ , pXBG 2ϩ , and Br 2 -TITU 3ϩ are competitive inhibitors of both Na ϩ and Rb ϩ occlusion. The cation antagonists are not themselves occluded, however, and it has been suggested that they inhibit occlusion by competing with occluded cations for negatively charged residues on L6/7 at the entrance to the occlusion sites (20,30). As shown in Fig.  6, however, Br-TITU 3ϩ binds to wild-type Na,K-ATPase and to L6/7 mutants in which either negatively or positively charged amino acid side chains have been replaced with alanine. If Br-TITU 3ϩ were binding to negatively charged amino acids in L6/7, then replacement of these amino acids by alanine would be expected to reduce the affinity of the protein for Br-TITU 3ϩ , and, consequently, reduce the extent of Br-TITU 3ϩ inhibition of ouabain binding. For the E825A mutant, the opposite effect on ouabain binding is observed, and there is no difference in the extent of Br-TITU 3ϩ inhibition of ouabain binding by the E828A mutant and wild type. These results indicate that Br-TITU 3ϩ does not bind to charged residues in L6/7.
There is substantial evidence that arylisothiouronium derivatives and other cation antagonists inhibit occlusion by competing with Na ϩ and Rb ϩ at cytoplasmic sites (34,35). One interpretation of this result is that the antagonists and occluded cations bind to the same sites on the protein. In the two-step model for occlusion (20,35), Na ϩ and Rb ϩ bind to sites at an entrance port or gate in the cytoplasmic domain of the protein before being transferred to the sites of occlusion within the transmembrane domain. Based on the competitive inhibition of occlusion by the cation antagonists, it was proposed that the cation antagonists inhibit occlusion of Na ϩ and Rb ϩ by competing for the binding sites at the entrance port. If this is an accurate description of the mechanism of inhibition of occlusion by the antagonists, then the results reported here indicate that the entrance port does not include charged residues in L6/7. An alternative interpretation of the competition between occluded cations and the antagonists is that they bind to different sites on the protein but that binding is mutually exclusive. An entrance port or initial binding site is not required in this interpretation because binding of the antagonists at some cytoplasmic site would prevent access of Na ϩ or Rb ϩ to the occlusion sites in the transmembrane domain of the protein. The cytoplasmic binding site for the antagonists is unknown, but the results of the investigations reported here again show that it cannot include charged residues of L6/7.
Br-TITU 3ϩ binds to and stabilizes the E 1 conformation of the enzyme (30). The increased inhibition of ouabain binding by Br-TITU 3ϩ seen for the E825A and R837A mutants can be explained if these mutations stabilize the E 1 conformation, and the reduced inhibition of ouabain binding seen for the R834A mutant can be explained if this mutation stabilizes the E 2 conformation. Support for this interpretation is provided by the observation that the p-nitrophenylphosphatase activity of the R837A mutant is reduced relative to wild type, whereas the p-nitrophenylphosphatase activity of the R834A mutant is similar to wild type. p-Nitrophenylphosphate hydrolysis is catalyzed by the Na,K-ATPase in the E 2 conformation.
R837A greatly decreased both the affinity for ouabain in a Mg 2ϩ -and P i -dependent assay at fixed [Mg 2ϩ ] and [P i ] (Table I), and the apparent affinity for P i at fixed Mg 2ϩ and ouabain concentrations (Fig. 8). Because the ouabain K d value did not change when the concentration of P i was increased in the titration with ouabain (not shown), it is possible that the mutation altered the conformation of the ouabain binding site. Although L6/7 is cytoplasmic and ouabain binds from the extracellular side of the membrane, ouabain binding induced a conformational change in L6/7 that exposed a cytoplasmic chymotrypsin cutting site that was not accessible in the absence of ouabain (20). Similarly, conformational changes induced in L6/7 by mutations can be expected to affect the ouabain binding site. Residues affecting ouabain binding affinity have been identified in TM 5-7 and in the extracellular loops between TM 5-8 (36,37). Any effect of mutations in L6/7 on the ouabain binding site probably occur through changes in the TM6 conformation. The 18 O exchange results show unequivocally that intrinsic rate constants for P i binding and/or reaction with the protein were altered by mutation because 18 O exchange between P i and H 2 O is not ouabaindependent. The exchange parameter containing the intrinsic P i dissociation constant (KЈ P ) and the equilibrium constant between non-covalently and covalently bound P i (K H2O ) was significantly larger (5-fold) for the R837A mutant than for wild-type enzyme (Table III). An increase in KЈ P could semi-quantitatively explain both the exchange results and the decrease in apparent affinity for P i in the Mg 2ϩ -and P i -dependent ouabain binding assay. However, a remote effect of a mutation in L6/7 on phosphorylation and dephosphorylation rate constants is also possible. We have previously shown that specific effects on P i binding caused by mutating amino acids outside the active site can be detected by 18 O exchange (25).
In Ca-ATPase, mutations of Lys 819 and Arg 822 , residues equivalent to Arg 834 and Arg 837 in Na,K-ATPase, reduced steady-state phosphoenzyme levels from both ATP and P i without affecting dephosphorylation of the phosphoenzyme (18). In Na,K-ATPase, steady-state levels of phosphoenzyme formation from ATP in R834A, R837A, and R848A were also lower than the wild-type enzyme (Fig. 9). R834A and R848A do not affect the E 2 P 3 E 2 dephosphorylation rate, but R837A results in faster E 2 P 3 E 2 dephosphorylation. The steady-state level of phosphoenzyme depends on the phosphoenzyme formation rate and the dephosphorylation rate. Since the dephosphorylation rates of R834A and R848A were the same as wild type, the lower level of steady state phosphoenzyme for R834A and R848A suggests that these two mutations affect the phosphoenzyme formation rate. Attempts to measure the phosphoenzyme formation rate from ATP in these mutants directly using manual sampling were unsuccessful because the rate of phosphorylation was too fast (data not shown). In Ca-ATPase, the reduction of phosphoenzyme formation from ATP was interpreted as evidence for an important role for Lys 819 and Arg 822 in determining the functional integrity of the phosphorylation domain of Ca-ATPase, and the reduced overall Ca-ATPase activity was attributed to this effect (18). In Na,K-ATPase, R834A, R837A, and R848A inhibited overall ATPase activity to a greater extent than they inhibited phosphoenzyme formation, suggesting that the mutations interfere with other partial reaction steps as well. The observation that R837A also inhibits the K ϩ -stimulated phosphatase activity and shows an increase in the E 2 P 3 E 2 dephosphorylation rate is consistent with this suggestion. The effects of L6/7 mutations on multiple Na,K-ATPase reaction steps and protein stability, together with the small effect of these mutations on the apparent affinity for Na ϩ and K ϩ , support a role for L6/7 in maintaining the functional integrity of the phosphorylation domain and its relationship to the ion binding domain. This is similar to the role suggested for L6/7 in Ca-ATPase by Zhang et al. (18).