Glutamate 779, an Intramembrane Carboxyl, Is Essential for Monovalent Cation Binding by the Na,K-ATPase*

Incubation of purified renal Na,K-ATPase with the fluorescent carboxyl-selective reagent, tl-(diazomethyl)-7-(diethylamino)-coum~n (DEAC), results in enzyme inactivation via disruption of the monovalent cation binding sites and loss of K+ and Na' binding capacity.

Asn-Ile-Pro-Glu-Ile-Thr-Pro-Phe-Leu. The length of this peptide was also examined in labeling experiments with cysteine-reactive probes which indicated that the peptide did not extend to the next carbo~l-containing amino acid residue in the @-subunit sequence (Aspso4). The site of attachment of DEAC is thus an intramembrane carboxyl residue present in all known sequences of ar-subunit isoforms of the Na,K-ATPase. This glutamate is essential for Na' and K+ binding and active transport by the sodium pump. Its location in the fifth transmembrane segment provides a way in which information about ATP binding and phosphorylation in the major cytoplasmic loop of the enzyme is transmitted to intramembrane cation sites during the reaction cycle.
The Na,K-ATPase (EC 3.6.1.37) actively transports Na+ and K+ across the plasma membrane of eukaryotic cells. This enzyme together with the gastric H,K-ATPase, the sarco(endo)plasmic reticulum Ca-ATPase, and the plasma membrane Ca-ATPase are members of a specialized class of ion pumps, termed P-type ATPases due to the phosphorylated intermediate formation in their reaction cycle. The P-type ATPases share a high homology in the amino acid sequence of their catalytic subunits (1-11, for review see Ref. 12) as well as a common catalytic mechanism (13)(14)(15). A large amount of work has been * This work was supported by Grant G~~9 5 0 0 from the National Institute of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "uduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $ To whom correspondence and reprint requests should be addressed.
done to describe the relation between hydrolytic and transport cycles of the Na,K-ATPase (14,16). However, little detail is known about the molecular mechanism which couples active cation transport to ATP hydrolysis. The primary structure of the a-and @-subunits in their different known isoforms have been determined (1-5, 11, 12, 17-21). The glycosylated @-subunit (M, = 35,000 for the protein component) has one transmembrane segment and most of its mass in the extracellular space (21,22). The a-subunit (M, = 112,000) is considered to be the catalytic subunit. Models for the a-subunit structure, containing 7, 8, or 10 transmembrane segments with a large central cytoplasmic loop, have been proposed. The cytoplasmic loop is probably primarily involved in ATP hydrolysis, while the required structure for the cation binding and occlusion seems to be restricted to the transmembrane segments of the enzyme (23)(24)(25).
Chemical modification studies have been performed in an attempt to localize functional domains of the enzyme and to identify the particular amino acids involved in the binding of various physiological ligands (for review, see Ref. 26). Amino acid residues within the cytoplasmic loop have been associated with the ATP-binding domain, including lysines, tyrosines, cysteines, and aspartates (for review, see Refs. 24 and 26). In contrast, little is known about the putative cation binding sites (24,25). Aplausible early hypothesis was that carboxyl-bearing amino acids, in or near the membrane, might be involved in cation charge neutralization and coordination at sites in the transport pathway. In the case of the Na,K-ATPase, this idea obtained some support from the well known monovalent cationprotectable inactivation of the enzyme by carbodiimides (27)(28)(29)(30)(31)(32)(33). However, complexities in the chemistry of the modification of proteins by carbodiimides (cross-linking and rearrangement subsequent to carboxyl modification) make i t difficult to confidently associate the enzyme inactivation with covalent incorporation of carbodiimide and therefore to locate the modified and essential carboxyl residues involved in cation binding (for discussion, see Refs. 26,32, and 33). Recently, Goldshleger et al. (34) have reported the Rb+-protectable modification by N , N" dicyclohexylcarbodiimide (DCCDI1 of G1u953 in the a-subunit of the Na,K-ATPase. However, their interpretation that G h P 3 is an essential part of the cation binding site (34)(35)(36) is rendered unlikely by the results of mutagenesis of a-subunit residues expressed in HeLa cells where G1u953 has been replaced and Na,K-ATPase activity is still observed (37).
We have overcome the problems inherent to the chemistry of carbodiimides by using 4-(diazomethyl~-7-(diethyl~ino~-coumarin (DEAC) (38,391. This is a stable fluorescent derivative of DEAC, 4-~diazomethyl~-7-~diethylamino~-coumarin; PAGE, polyacryl- The abbreviations used are: DCCD, ~~-dicyclohexylcabodiimide; amide gel electrophoresis; BSA, bovine serum albumin; HPLC, high performance liquid chromatography, Tricine, ~-T r i s -( h y d~x~e t h y l )methylglycine; PWF, polyvinylidene fluoride; DMSM, 2,bdimethoxystilbene-4'-maleimide. diazomethane which reacts specifically with carboxylic residues to yield fluorescent esters (40). Treatment of the isolated Na,K-ATPase with DEAC results in enzyme inactivation due to the loss of cation binding capacity (38, 39). The inactivation rate is dramatically increased when the enzyme is treated with DEAC under conditions which produce enzyme phosphorylation (presence of Pi and Mg2+); but most importantly, inactivation can be almost completely prevented by K' and by Na', with a lower apparent affinity. Gel electrophoresis of modified protein reveals intense fluorescence labeling of the a-subunit, which is substantially reduced if treatment with DEAC is performed in the presence of K' ions. The extent of inactivation is linearly related to the amount of K'-protectable DEAC incorporation, and complete inactivation of ATPase activity occurs with 1.46 2 0.12 mol of DEAC covalently boundmol of nucleotide-binding sites. This suggests that only 1 or 2 carboxyl residueskatalytic center are involved in the modification leading to inactivation. DEAC-modified enzyme is unable to occlude K+ or Na' ions, but exhibits normal levels of high affinity ATP binding, and is able to undergo the E l e Ez conformational transitions when they are induced by ligands other than Na' or K' (38).
These results indicate that (i) there are 1 or 2 carboxyl residues which are essential for K' and Na+ binding and occlusion. (ii) These residues are probably part of the monovalent cation binding domain and occlusion site in the Na,K-ATPase a-subunit (38, 39). We describe here the identification of G~u~~~ as the carboxyl residue modified by DEAC which is protectable by both K' and Na+ and propose its involvement in the binding and occlusion of both cations in the Na,K-ATPase. A preliminary report of some of our data has been presented (41).'
Enzyme Isolation-Na,K-ATPase was purified from canine kidney outer medulla according to Jorgensen (42) with the modifications of Liang and Winter (43). After ultracentrifugation, the enzyme was washed with 25 mM imidazole, 1 mM EDTA, pH 7.5 (buffer C), before storing a t 5 'C. Protein concentration was determined by the method of Lowry et al. (44) using BSA as standard. The Na,K-ATPase activity of the enzyme used in these studies was about 17-23 pmol of Pi mg" min", assayed as described below.
Enzyme Zbeatment with DEAC-The enzyme, usually 10-20 mg (0.5 mg/ml), was treated in 50 m~ imidazole, pH 6.5, 1 mM EDTA, 3 mM MgCl,, 3 mM P,, 250 p~ DEAC, 10% dimethyl sulfoxide during 2 h a t 37 "C (38). The modified protein obtained using this procedure retained 10% of its initial activity and will be referred to as DEAC-enzyme in this paper. 100 mM KC1 was included in the treatment media when it was desired to obtain K+-protected enzyme (90% active); likewise, 100 mM NaCl was present in the media when Na+-protected enzyme (75% active) was prepared. The reaction was stopped by dilution in buffer C and ultracentrifugation or by dilution (1:lOO) in the ATPase assay medium; both methods gave the same results, Na,K-ATPase Activity Assay-The assay medium was ( m~) EGTA, 0.5; NaCI, 130; KCl, 20; MgCl,, 3; ATP, 3; and imidazole, 50, pH 7.2, 0.3 mg/ml BSA, and approximately 0.5 pg/ml enzyme protein. The assay was performed at 37 "C for 15 min and the Pi released determined (45).
The Na,K-ATPase activity was calculated from the difference between the ATP hydrolysis measured in the absence and in the presence of 0.5 m~ ouabain.
a-Subunit Isolation-The a-subunits from DEAC-enzyme, K+-protected, and Na+-protected enzymes were isolated by SDS-PAGE (46) in In the published abstract of the preliminary report (41), the essen-This was in error. There is no G I U "~~; t h e correct numbering is G~U " "~ as tial residue is identified as G~u~~~ in the fifth transmembrane segment.
reported here.
7.5% acrylamide gels. Large gels were used because of our preparative scale requirements (0.225 x 14 x 10 cm). They were loaded in a single gel-wide well with 2.5 mg of ATPase protein and ran 1 h a t 100 V and 4 h at 200 V. After electrophoresis, DEAC-labeled a-subunits (as well as the DEAC-labeled peptides in later gels) were detected by their fluorescence emission on illumination with a hand-held long wavelength (360 nm) W lamp and photographed using TRM film and a yellow filter. Photographic negatives were scanned in a densitometer when comparative quantification was required. The a-subunit bands were cut out of the gel, cut in 1-mm2 pieces, and homogenized with 5 volumes of 0.1% SDS, 0.1 M NH,HCO,, pH 8.5, in a glass-Teflon homogenizer. The gels were extracted during 5-8 h, and the gel pieces were kept in suspension with a rotating mixer. The elution was repeated twice, and the eluates were concentrated using Centricon 30 ultrafiltration devices (Amicon, Grace Co.). a-Subunits were then precipitated with 9 volumes of methanol at -20 "C overnight.
The isolated a-subunits were washed twice by resuspension with 1 ml of 0.1% SDS, 0.1 M NH4HC03, pH 8.5, and precipitated with 9 volumes of methanol a t -20 "C for 8-10 h. Finally, the protein was resuspended in 0.05% SDS, 20 mM phosphate buffer, pH 7. This procedure yielded 0.3-0.4 mg of a-subuniumg of initial Na,K-ATPase. Protein concentration in isolated a-subunit (as well as in later purified peptides) was determined according to Bradford (47).
Peptide Isolation-The general strategy for the isolation of DEAClabeled, K+-protectable fragments is shown in Scheme I.
Tricine-urea SDS-PAGE ofV,-treated a-subunit (50 pg/well) was performed in 16.5% polyacrylamide gels, including 6 M urea in the separating portion of the gel (0.75 mm thick 16.5% T, 6% C separating, 10 cm; 10% T, 3% C spacing, 1.5 cm; 4% T, 3% C stacking, 1 cm).The gels were aged overnight, prerun (3 h, 25 m A ) with "gel buffer" (diluted 1:3) containing 0.1 mM thioglycolate in the cathode chamber (491, the samples were run (25 mA, 16-20 h) with thioglycolate (0.1 mM) added to the running cathode buffer. After electrophoresis the gels were photographed and scanned as indicated above (a-subunit isolation). Two DEAC-labeled cation-protectable bands with approximate apparent molecular masses of 15-17 kDa (peptides I and 11) and 5-6 kDa (peptides IIIa and b) were observed in these gels. Peptides I and I1 which generally appeared as a broad band sometimes resolved as a doublet were better separated in Tris-glycine gels (see below). The band corresponding to peptides IIIa and IIIb was cut out of the gel and homogenized in 5 volumes of 0.1% SDS, 0.1 M NH4HC03, pH 8.5. The peptides were then eluted from the gel as described above for the a-subunits, the eluates concentrated in Centricon 3 filters, and the peptides precipitated with 9 volumes of acetone a t -20 "C overnight. The peptides were washed once by resuspension in 0.1 M NH,HCO,, pH 8.5, and precipitation with 9 volumes of acetone at -20 "C, 5 h. The peptides were then resuspended in 50-100 pl of 0.1 M NH,HC03, pH 8.5, and kept at -20 "C until processed for sequencing or proteolytic cleavage.
Peptides IIIa and IIIb were digested with thermolysin (l:lO, w/w) in 1 mM CaCI2, 20 m~ NH,HCO,, pH 8.5,2 h, 45 "C. The resulting peptides were separated in Tricine-urea gels, as described above. The only fluorescent band observed in this gel (peptide IV) was cut out and eluted twice with 5 volumes of 20 mM NH,HC03, pH 8.5. The eluates were concentrated using Amicon YM1 membranes (1 kDa cut-off) and then processed for sequencing.
The DEAC-labeled cation-protectable peptides with an apparent molecular mass between 15 and 17 kDa (peptides I and 11) were resolved as two separate bands in 15% polyacrylamide "is-glycine SDS-PAGE including 2 M urea in the separating portion of the gel (46). The two bands were cut out of the gel, homogenized in 0.1% SDS, 0.1 M NH,HCO,, pH 8.5, and the peptides eluted as described above. The eluates were concentrated in Centricon 10 filters and then washed by precipitation with 9 volumes of acetone, -20 "C, and resuspended in 0.1 M NH,HCO,, pH 8.5. The peptides were finally resuspended in 50-100 pl of 0.1 M NH4HC03, pH 8.5, and kept at -20 "C until processed for sequencing.
Peptide Sequencing-After isolation, the peptides were spotted directly onto polyvinylidene fluoride (PVDF) membranes (Immobilon-P, Milipore Co.) and the membranes extensively washed with water to eliminate buffer, urea, and SDS. The samples were sequenced in an Applied Biosystems model 477A Protein Sequencer with on-line model 120A phenylthiohydantoin analyzer. Automated Edman degradation was done using the manufacturer's reaction vessel cycle PRO-1, with modifications including the use of a higher reaction temperature (49 "C) and a 30% increase in the delivery time of trimethylamine vapors. In particular, for sequencing of peptide IV, the PVDF membrane was positioned on top of a Polybrene-conditioned filter disc (50). This procedure has been found necessary for similarly small decapeptides (50).

RESULTS
In our previous studies (38, 39), we described the inactivation of Na,K-ATPase upon chemical modification by DEAC. The characteristics of the reaction a s well a s those of the resulting DEAC-enzyme led us to propose that DEAC modifies 1 or 2 carboxyl residues at the cation binding site of the enzyme. We then set out to localize the residues and to determine at the same time if Na+ and K+ were indeed preventing modification of the same carboxyl groups. Fig. 1 shows the labeling of the a-subunit from DEACtreated (lanes 2 ) , Na+-protected (lanes 3), and K+-protected (lanes 4 ) enzymes. It is clear that there is a reduction of fluorescent labeling with both cations, although K+ seems to protect against the labeling to a slightly greater extent than Na+ (55-60% protection by K+ compared with 4 0 4 5 % by Na+). This corresponds with the less effective protection against inactivation provided by Na' than K' ions (38). In these Tris-glycine gels, DEAC which is non-covalently bound to proteins, runs in the SDS front and is separated from a-subunit. The a-subunits were then eluted from the gel, concentrated, and pelleted by solvent precipitation. This methodology proved to be useful in avoiding smearing of fluorescence (due to free coumarins) in the low molecular mass range (<6 kDa) of Tricine gels subsequently used to resolve proteolyzed samples of a-subunit. After testing various proteases, V8 was selected to cleave the purified a-subunits. Fig. 2 shows the patterns resulting from extensive proteolysis of DEAC-treated (lanes A, and B , ) , Na+-protected (lanes A2 and Bz), and K+-protected (lanes As and Bs) a-subunits with VH protease. The Coomassie Brilliant Blue staining of the gel (Fig. 2B) shows an identical proteolytic pattern for the three a-subunits, this allows direct comparison of the fluorescence patterns in terms of the extent of DEAC labeling and cation protection of specific peptides. The extent of fluorescence labeling of the proteolysis fragments indicates that the mixture can be resolved into both cation-protectable and non-protectable peptides. The unprotected peptides are obviously not associated with enzyme inactivation. When Na+-protected and K+-protected a-subunits are compared, a very similar distribution of DEAC labeling is observed (compareA2 andAs in Fig. 2). This supports the idea that both cations protect the same carboxyl residue against DEAC modification. Inspection of the fluorescent labeling pattern (in Fig. 2)  gether with several DEAC-labeled fragments which were also labeled in the presence of cations. The labeling in the V,-digested a-subunits maintains the same relationships observed in undigested a-subunits when protected and unprotected enzymes are compared (ie. 53 2 3%, n = 5, protection by K' and 44 2 39'0, n = 3, by Na'). Importantly, the observed cationprotectable bands contain 68 2 7%, n = 6, of the protectable fluorescence with the rest of it spread throughout the gel not associated with any particular band.
The fragments with an apparent molecular mass of 15-17 kDa ran as a broad band in the Tricine gels, and in some samples they appeared as a doublet. The Coomassie staining of the gel shows very little protein associated with this particular fluorescent band, and the sequencing of peptides in this band, isolated from Tricine gels, yielded inconsistent results.
The doublet was better resolved using 15% polyacrylamide SDS-PAGE (Tris-glycine) (see Fig. 3). Peptides in both bands were eluted and sequenced, and the sequence data is displayed in Table I. The positions of these peptides in the primary structure, as well as that of others described in this paper, were assigned on the basis of their complete identity with the corresponding sequence in the a-subunit from dog.3 Both peptides start at the same amino acid; consequently, the small difference in their molecular weight is likely due to the different point of cleavage at the carboxyl end of the longer fragment. Both of these peptide sequences are produced by cleavage at predicted VR sites in the a-subunit.
The second cation-protectable band observed in Tricine gels of V,-cleaved DEAC-a-subunit (Fig. 2) had an approximate apparent molecular mass of 5-6 kDa. The fluorescent band was cut out of the gel, eluted, and spotted onto PVDF membrane for sequencing. The sequencing data (see Table I) showed the presence of two peptides (IIIa and IIIb) in this band. The starting sequence of peptide IIIa coincides with the one of the described peptides I and 11, raising the possibility that a DEAC-labeled cation-protectable carboxyl group is contained within the length of peptide IIIa. Thus, the cation-protectable DEAC-labeled carboxyl is distributed in peptides I, 11, and IIIa. We have calculated (assuming molecular masses of 112, 16, and 5.5 kDa for a-subunit and the peptides I and I1 and peptides IIIa and b, respectively) a total recovery of 67 2 5% (n = 8 )   mols of labeled fragment distributed between peptides I, 11, and IIIa, since these peptides all begin at Gly7"'.
Peptides IIIa and IIIb were too large for complete sequencing; in order to further reduce their size they were treated with thermolysin (see "Experimental Procedures"), and the digests were resolved in Tricine-urea gels (Fig.  4). A single smaller fluorescent band was observed after the thermolysin treatment. No other fluorescence bands were observed or other fluorescence at the front of the gel, indicating that all the labeling was contained in this single band. The sequencing data (see Table I) showed a fragment from peptide IIIa starting at Leu773 as the only peptide (peptide IV) in this fluorescent band. When this peptide was sequenced using standard procedures with PVDF membranes as the only immobilizing matrix, the yields of amino acids dropped quickly after the first cycle, and it was impossible to sequence this peptide with confidence beyond 5 or 6 residues (Table I). This drop in yield after the first cycle is characteristic of short peptides sequenced on PVDF. Our results agree with those obtained earlier when sequencing model decapeptides (see Table 9 (Table I). Consequently, we were able to sequence peptide IV. Since G~u~~! ' is the only carboxyl residue in this DEAC-modified peptide, it is thus identified as the target of DEAC inactivation. The presence of G I u~~" during the sequencing is due to the hydrolysis of the DEAC-carboxyl ester under the acidic sequencing conditions. The extensive hydrolysis of glutamic and aspartic esters under these conditions is well known (51,52).
The possibility that peptide IV extends beyond Leu7"" and contains another carboxyl residue which is modified by DEAC is unlikely. The next carboxyl residue in the Na,K-ATPase sequence is Asp'"". To show that peptide IV does not reach Asp'"", we took advantage of (i) the fact that Cysxo2 is the only thiol group between Leu77x and Asp'"", and (ii) the pH sensitivity of DEAC (DEAC is not fluorescent at low pH). We treated the DEAC-labeled peptide IV with the sulfhydryl reagent DMSM.
We then ran this DMSM-treated DEAC-labeled peptide IV in a non-reducing Tricine gel and lowered the pH of the gel to eliminate the fluorescence due to DEAC (Fig. 5). Under this condition, peptide IV was not fluorescent indicating that it did not Cation Site Glutamate of Na,K-ATPase

TAHI.IS I Starting sequences of cation-protectahle DEAC-laheled peptides
The peptides were isolated and sequenced a s indicated under "Experimental Procedures." The picomole yield of each amino acid is indicated in parentheses. Sequencing of peptides I and I1 was stopped after 10 cycles. Sequencing of peptides IIIa and IIIb was halted after seven cycles. Peptide IV on PVDF support was sequenced for 10 cycles, and no amino acid was detected beyond cycle 8. Peptide IV on PVDF and Polybrene was sequenced for 15 cycles; after cycle 12 no amino acid was detected. The marked reduction in yield of Ser7'" during sequencing on PVDF membranes has been reported earlier and ascribed to problems related to the sequencing chemistry on PVDF membranes (61). Gly (60) Phe (61) Gln (681 Phe (50) Asp (51) Thr (14) Asp (40) GlY5fi' contain any cysteine residues and therefore, did not reach Cys802 and hence Aspxo4. Peptides IIIa and IIIb that contained cysteines were processed as positive controls, showing fluorescence at pH 2.5 when treated with DMSM. Other data also support our conclusion that peptide IV is a small peptide with a single carboxyl residue. A modified peptide starting at and reaching Asp"04 would contain at least 32 amino acids and have a molecular mass of 3691 Da. Peptide IV runs with our 3 kDa standard, but mobility in Tricine-urea gels is unreliable for estimating the molecular mass of small peptides such as peptide IV. The anomalous mobility of such small peptides varies greatly with small changes in the electrophoresis conditions (48). However, in agreement with the sequencing data, peptide IV was not retained by microseparation devices with 3-kDa cut-off membranes (Centricon 3) but was retained by membranes with 1-kDa cut-off limit (MPS-1, Amicon, Inc.). Furthermore, the characteristics of the sequencing behavior indicate the small size of peptide IV, a 32-amino-acid peptide, would not easily wash out of the PVDF membranes during the sequencing as peptide IV did, and would have been sequenced well beyond Leu7x4. This cumulative evidence supports our assignment of G~u~~~ as the target for DEAC modification which results in K+-protectable and Na+-protectable inactivation. DISCUSSION We have previously described the inactivation of the Na,K-ATPase by DEAC (38,391. We presented evidence showing that the modification of 1 or 2 carboxyl residues in the a-subunit of enzyme was responsible for the removal of the Na' and K' binding capabilities of the enzyme with its consequent inactivation. The effects of the modification were limited to the cation binding site because DEAC-inactivated enzyme was able to bind ATP with high affinity and undergo El <--, E2 conformational transitions. The characteristics of the inactivation and the protection by Na' or K+, together with those of the DEACmodified enzyme, strongly suggested the direct involvement of the modified carboxyl groups in the cation binding and occlusion. We have now identified G~u~~~ in the a-subunit as the amino acid modified by DEAC in a monovalent cation-protectable manner. This result, together with our previous findings, suggests that G~u~~" which is localized in the putative fifth transmembrane segment of the protein (see Fig. 6) is part of the cation binding site in the Na,K-ATPase.
Isolation a n d Sequencing of Peptides-In order to design a strategy for localization of the DEAC-labeled carboxyl residues, the probe characteristics and the properties of the likely labeled fragments had to be considered. (i) Free DEAC or DEAC degradation products (38) partition into the membrane; if they are not effectively removed they produce high fluorescence backgrounds during peptide isolation. (ii) DEAC is not fluorescent at pH <6. (iii) The labeled carboxyl residues might be located in or near transmembrane segments which are extremely hydrophobic and therefore difficult to isolate and purify using high performance liquid chromatography (HPLC). (iv) The lactone ring of the DEAC molecule is unstable at pH a FIG. 6. Scheme of the a-subunit of the Na,K-ATPase. Amino acids associated with the ATP-binding site (24) and carboxyl groups probably located within the membrane are indicated.
>11 producing a nonfluorescent residue. (v) The DEAC-amino acid ester might be unstable in extremely low pH media used in peptide sequencing. Some of these characteristics prevented, for example, the isolation of DEAC-labeled peptides by HPLC or the use of tricine gels without first effectively removing free DEAC. Likewise, the resolution range of Tricine gels led us to avoid using proteases which yield large numbers of low molecular mass (<3 kDa) fragments. Bearing in mind these constraints and our previous observation that DEAC labeling was restricted to the a-subunit of the enzyme (38,39), our approach was (see Scheme I) to obtain fully labeled a-subunit free of non-covalently bound coumarines, to treat the a-subunit with a protease that yields fragments large enough to be separated by Tricine gels, and finally by comparison with digested cationprotected DEAC-labeled a-subunit, to select and isolate the cation-protectable DEAC-labeled fragments.
Two cation-protectable bands with apparent molecular masses 5 4 and 15-17 kDa were obtained. Isolation of these fragments and sequencing showed the presence of four peptides in these bands (I, 11, IIIa, and IIIb; Figs. 2 and 3; Table I).
Considering that these peptides account for most of the cationprotectable fluorescence (68%) and their good recovery (67%), we are confident that these peptides contain the major cationprotectable DEAC-labeled site.
Peptides IIIa and IIIb comigrated in Tricine-urea gels. Peptide IIIa, starting at Gly758, contains the putative fifth transmembrane segment, while peptide IIIb, extending from G l~~~l , consists of part of the cytoplasmic loop of the enzyme (see Fig.  6). The treatment of peptides IIIa and IIIb with thermolysin yielded a single fluorescent band. Sequencing of the material extracted from this band showed a single peptide (peptide IV) starting at (Table I), which indicated that peptide IV is part of peptide IIIa and, therefore, peptide IIIb is not esterified by DEAC. Sequencing of peptide IV revealed a small peptide extending from to with a single carboxyl residue, G~u~~' . Furthermore, the experiment treating peptide IV with the sulfhydryl probe DMSM indicated that the peptide did not reach Aspso4, and, thus, G~u~~~ is the amino acid modified by DEAC.
Peptides I, 11, and IIIa start at Gly758 and probably are the result of differences in the extent of proteolytic cleavage yielding different carboxyl termini. These three peptides have at least one common carboxyl group esterified by DEAC, and we have demonstrated that peptide IIIa is only esterified at G~u~~' .
However, peptides I and 11, extending perhaps to Glugo8and Glusg2, respectively, have a number of additional carboxyl residues that might also be modified by DEAC or protected by Na+ or K' , in addition to G~u~~~. At this time, this issue is not fully resolved, but all our results point to ( 2 1~7~9 as the only DEAC-labeled cation-protectable residue. Our previous studies (38) suggested that the inactivation was likely caused by a single residue.
Monovalent Cation Protection-We have shown that Na' is slightly less efficient than K+ in protecting enzyme against inactivation by DEAC (38, 39) and DEAC labeling (Fig. 1); furthermore, after modification the loss of Na' binding (21% of control) was also slightly smaller than the loss of K+ binding (9% of control) (38,39). Consequently, an important point to be addressed in the localization studies was whether or not Na+ protects the labeling of the same carboxyls as K+ and, as seemed to be indicated by our previous results, Na' and K' both utilize this carboxyl in their interaction with the enzyme. There were no detectable differences in the labeling pattern of proteolytic digests of a-subunits coming from Na+-or K+-protected enzymes. This indicates that Na' and K+ protect the same peptides against modification with DEAC. Furthermore, since we have observed a single residue modified by DEAC, G~u~~' , it is protectable by both cations. This supports the idea that residues involved in Na+ binding and transport are also involved in K' binding and transport. The lack of sensitivity of phosphoenzyme to K' ions, we reported earlier (38, 391, also suggests that K+ binding to the phosphoenzyme involves the same residues.
Significance and Role of G l~~~~--S o m e idea of the functional significance of a particular amino acid residue may be provided by its conservation in different isoforms of the enzyme, among various species, and by its presence in the sequences of structurally and functionally related enzymes. Glutamate residues corresponding to G~u~~~ are conserved in all the sequences we examined for Na,K-ATPase, gastric H,K-ATPase, and the sarco(endo)plasmic reticulum Ca-ATPase, but it is not present in other P-type ATPases such as plasma membrane Ca pump, yeast H pump, or Escherichia coli K pump. For comparison, Table I1 shows a n alignment of partial sequences surrounding G~u~~' in the Na,K-ATPase (different isoforms and various species) and highly homologous areas from the gastric H,K-ATPase, and the sarco(endo)plasmic Ca-ATPase. It is clear that this particular carboxyl is in a highly conserved region of the Na,K-ATPase and also the H,K-ATPase. In all of the pumps which possess a residue corresponding to G~u~~' , it appears in the fifth transmembrane segment of models based on hydropathy analysis. The sarcoplasmic reticulum calcium pump also has this residue, although the surrounding sequence shows little identity with the corresponding Na,K-ATPase primary structure. The different residues surrounding in the rabbit SERCA2) compared with the Na,K-ATPase sequence may explain our observation that DEAC inhibits the sarcoplasmic reticulum Ca-ATPase to only a limited extent, and this inactivation is not protectable by Ca2+ ions.4 However, G l~~~l i n the Ca2+ pump (SERCA2) has been the target of mutagenesis studies, and the products of those mutations ( G l~~~l ---f Gln and G~u~~~ + Asp) are unable to support Ca2+-dependent functions of the enzyme (53, 54). The conservation shown in Table I1 together with this latter finding supports the idea that G~u~ may play an important role in cation binding in the Na' pump. It is interesting to note that G~u '~~ previously identified from chemical modification studies to be at the K binding (35,36) site (but see Ref. 37), is not conserved in the a-subunit sequences of Drosophila (31, Artemia (4), or Hydra (5).
Chemical modification of carboxyl residues in the Na,K-ATPase has been previously performed using carbodiimides (27)(28)(29)(30)(31)(32)(33)(34). Goldshleger et al. (34) have shown, under conditions of very low inactivation (necessary to reduce high nonspecific labeling), Rb+-preventable DCCD labeling of and other (unident~fied~ carboxyl residues together with ~b+-preventable cross-linking and concluded that was at the cation binding site. However, recent data from mutagenesis studies of heterologously expressed enzyme in HeLa cells have shown that neither G~u~~~ nor its neighbor GluS5* are essential for enzyme activity and hence monovalent cation binding (37). Using a different experimental system where a large proportion o f the extramembranous portions of the a-subunit have been removed proteolytically to produce so-called "19 kDa membranes,'' Karlish et al. (35) also found monovalent cation-preventable incorporation of DCCD in an electrophoresis band containing the transmembrane fragments M1 + M2 and M3 + M4 (see Fig. 6) and in the 19-kDa carboxyl end of the a-subunit starting at It is interesting that the peptide starting at Gly737 (M, = 8,000) which contains the G~u~~~ was not modified by DCCD (35). Furthermore, the peptides labeled with DEAC (I, 11, and IIIa) do not coincide with any of the residues labeled with DCCD. Our identification of G1u779 as the target for DEAC inactivation emphasizes the importance of those transmembrane segments in the native enzyme and in post-t~ptic residues which form part of the cation occlusion site but are not contained in the C-terminal 19-kDa fragment.
Cation Pump Mechanism-Analysis of the primary structure of the a-subunit of the Na,K-ATPase using one of several available forms of hydropathy analysis has led to models for its topological arrangement in the membrane containing either 7, 8, or 10 transmembrane segments (1,23). Subsequent studies utilizing antibodies or selective proteolytic cleavage demonstrating the cytoplasmic location of the carboxyl terminus (55-57) have reduced the possibilities to models containing either 8 or 10 transmembrane segments. Fig. 6 shows one such model containing 10 transmembrane segments; this is based on a composite of studies on the Na,K-ATPase and alignments with sequence from other P-type ATPases. the residue we have identified and postulate is directly involved in monovalent cation binding, is close to the middle of the fifth transmembrane segment. In other models with eight transmembrane segments, this residue is at the beginning (the cytoplasmic side) of the fifth transmembrane segment. This segment is directly connected to the major cytoplasmic loop which contains the phosphorylation domain, including Asp369 and the ATPbinding domain. "he involvement of Glu779 in cation binding makes clear how binding of ATP (or phosphorylation) could affect cation binding and vice versa. Alterations in the cytoplasmic loop confo~ation would be directly transmitted to transmembrane segment 5 and thus to G~u~~~ and the cation binding site. Thus, the antagonistic effects of K+ on high affinity ATP binding, and the ATP-or Pi-stimulated deocclusion of K+ ions are readily explained. These types of interactions could also account for the observed effects of Na' or K+ ions on the reac-tivity of lysine 501 toward isothiocyanate reagents (58-60) and the increased reactivity of G I U~'~ toward DEAC when enzyme is phosphorylated (38).