SPECIAL COMMUNICATION
Predicted location and limited accessibility of protein kinase A phosphorylation site on Na-K-ATPase

Laboratory of Membrane Biology, Neuroscience Center, Massachusetts General Hospital, Charlestown, Massachusetts 02129

	ABSTRACT

TOP ABSTRACT INTRODUCTION DATA ANALYSIS DISCUSSION REFERENCES

Regulation of Na-K-ATPase by cAMP-dependent protein kinase occurs in a variety of tissues. Phosphorylation of the enzyme's catalytic subunit at a classical phosphorylation consensus motif has been observed with purified enzyme. Demonstration of phosphorylation at the same site in normal living cells or tissues has been more difficult, however, making it uncertain that the Na-K-ATPase is a direct physiological substrate of the kinase. Recently, the structure of the homologous sarco(endo)plasmic reticulum Ca-ATPase (SERCA1a) has been determined at 2.6 Å resolution (Toyoshima C, Nakasako M, Nomura H, and Ogawa H. Nature 405: 647-655, 2000.), and the Na-K- ATPase should have the same fold. Here, the Na-K-ATPase sequence has been aligned with the Ca-ATPase structure to examine the predicted disposition of the phosphorylation site. The location is close to the membrane and partially buried by adjacent loops, and the site is unlikely to be accessible to the kinase in this conformation. Conditions that may expose the site or further bury it are discussed to highlight the issues facing future research on regulation of Na-K-ATPase by cAMP-dependent pathways.

adenosine 3',5'-cyclic monophosphate; crystal structure

	INTRODUCTION

TOP ABSTRACT INTRODUCTION DATA ANALYSIS DISCUSSION REFERENCES

THE ION GRADIENTS that are necessary for a wide range of physiological processes are maintained by Na-K- ATPase, and regulation of the Na-K-ATPase plays a role in the hormonal and homeostatic responses of many cells (recently reviewed in Ref. 55). Regulation occurs on several levels: biosynthesis and degradation; reversible recruitment to and internalization from the plasma membrane; alteration of affinity for Na; and either stimulation or inhibition of activity. The involvement of cAMP-dependent protein kinase (PKA) in acute sodium pump regulation has been documented in 20 different mammalian tissues and in lower vertebrates (55). The result can be either pump stimulation or inhibition, however, and in no case is the pathway completely understood.

The first demonstration of phosphorylation of the Na-K-ATPase catalytic subunit () by the catalytic subunit of PKA was by Mardh in 1979 (38), but little more was done until 1991, when phosphorylation was observed at stoichiometries of 0.5-1.0 phosphate per  (8, 19). Soon it was shown that the site was within 12 kDa of the COOH terminus (20), specifically at Ser-938, the only serine in a consensus motif (RRNSF) for PKA (6, 25, 27). The earliest in vitro phosphorylation studies were with enzyme treated with a mild detergent like Triton X-100, and it was observed that the detergent was actually required to obtain phosphorylation in the test tube (6, 18, 20, 25). Since then, phosphorylation by PKA in the test tube in the absence of detergent has been detected only in preparations that were mildly heat denatured (4) or in preparations that were completely solubilized with C₁₂E₈ or 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate detergent and then reconstituted with lipids (21).

It normally should be a straightforward matter to demonstrate phosphorylation in intact cells at a site like RRNSF on an abundant protein like the Na-K-ATPase, but, in fact, there is not a lot evidence for it. One lab has reported dibutyryl cAMP-stimulated or forskolin-stimulated incorporation of ³²P in  in renal cortical or medullary tubule preparations (14, 35), although the site of phosphorylation was not defined. In several cases, experiments designed to detect it have given negative results: rat sciatic nerve (12), choroid plexus (28), lung cells (10), striatal neurons (45), C6 glioma, and normal rat kidney (NRK-52E) epithelial cells (26). Evidence for the phosphorylation of Ser-938 on normally expressed (not transfected) Na-K-ATPase by endogenously activated PKA has been presented in only one paper to our knowledge. In this case, untransfected COS cells stimulated with a -adrenergic agonist, isoproterenol, showed increased phosphorylation of the endogenous monkey ₁-subunit detected with an antibody directed against the phosphorylated Ser-938 site (16). There was a 70% increase in immunoreactivity of a band at 100 kDa over a substantial background of stain in unstimulated cells. In the most recent attempt (26), an antibody prepared identically was observed to cross-react with many other phosphoproteins in cells that had been treated with phosphatase inhibitors. Although forskolin-IBMX treatment of the cells raised cAMP to very high levels, phosphorylation of endogenous Na-K-ATPase was not detected in immunoprecipitates. In cell extracts, however, addition of Triton X-100-treated exogenous Na-K-ATPase resulted in its phosphorylation, showing that PKA was present and active. The detergent seemed to be required for access to the site.

Two groups investigated the effects of mutation of Ser-938 to alanine (S938A) in transfected cells. Beguin and colleagues (5, 6) found evidence for phosphorylation in exogenous Bufo enzyme expressed in COS cells that was reduced in mutants expressing Ala-938. In studies of activity, Cheng and Fisone (16, 27) and their associates demonstrated that forskolin treatment to raise cAMP levels caused a 19% reduction in Na-K-ATPase activity in COS cells expressing wild-type rat ₁ Na-K-ATPase, whereas cells expressing S938A lacked forskolin-induced inhibition. Similar results, a 25% reduction in activity, were found for isoproterenol, a -adrenergic agonist (16). In the latter paper, the phosphorylation site-specific antibody detected an increase in Ser-938 phosphorylation in wild-type rat ₁ but not in S938A, both in stable transfectants and in transient transfectants that were overexpressing rat ₁ Na-K-ATPase by 5- to 10-fold. Oddly, however, treatment with ouabain, a specific inhibitor of the Na-K-ATPase, caused the same increase in phosphorylation of Ser-938 in the host cell -subunit, and no additional effect was seen with isoproterenol. Mutation of Ser-938 to aspartic acid was apparently not inhibitory (17).

Inhibition of Na-K-ATPase activity by PKA has been reported in crude membrane preparations (22) and purified enzyme (8; reviewed in Ref. 15). An inhibitory effect of detergent was shown to be a complication that precluded assessment of inhibition in some systems (21, 25). Most recently, phosphorylation by PKA was observed to be dissociable from inhibition in two reports. In both cases, phosphorylation of Ser-938 was constant, while its functional consequence varied with either intracellular Ca level (15) or oxygen tension (35). In enzyme expressed in insect cells, treatment of intact cells with a cAMP analog resulted in inhibition of ₁ and ₂ by ~30%, whereas ₃ was stimulated by ~35%, all without detergent treatment (11). In this case, phosphorylation was detected, but not correlated with modulatory effects.

A simple, unified scheme of sodium pump inhibition by direct phosphorylation at the PKA site has thus not been obtained. The only consistent trend is that phosphorylation of  at the PKA site is usually detected in conditions where there might be some denatured enzyme: in detergent, after reconstitution, after mild heating, and in cells expressing exogenous -subunit from transfected DNA. Here we examine the hypothesis that the PKA site might be inaccessible to the kinase.

	DATA ANALYSIS

TOP ABSTRACT INTRODUCTION DATA ANALYSIS DISCUSSION REFERENCES

Homology of Ca-ATPase and Na-K-ATPase. The sarco(endo)plasmic reticulum Ca-ATPase (SERCA) is in the same gene family as the Na-K-ATPase and is highly homologous, showing ~30% identity, 65% similarity, and a similar total length (41). For years, it has been believed that the similarity between Ca-ATPase and Na-K-ATPase breaks down in the COOH-terminal third of the protein, where six of the transmembrane spans are found, and the predicted topology of that portion of the protein has been quite controversial. The perception that the COOH-terminal ends were different was based on the premise that hydropathy plots should show aligned blocks of largely hydrophobic transmembrane spans, while hydropathy comparisons between Ca-ATPase and Na-K-ATPase (and even between Na-K-ATPase and H-K-ATPase) showed major inconsistencies. With the 2.6 Å crystal structure of the Ca-ATPase, however, the actual transmembrane helices can be seen, as well as key structural features such as the locations of residues involved in ion binding and disruptions of helical structure to create an ion-binding pocket. It can now be seen that the hydropathy predictions failed because of the relatively high content of polar amino acids in the transmembrane spans and a relatively high content of nonpolar amino acids in the connecting loops. The underlying homology between Ca-ATPase and Na-K-ATPase is strong and is completely consistent with a shared fold. Topology studies did not always give consistent results for the Na-K-ATPase, but most biochemical and immunological evidence obtained with native enzyme supports the corresponding transmembrane spans (2, 3, 24, 29, 37, 40, 44, 53, 56). This means that overall organization of the Ca-ATPase and Na-K-ATPase -subunits should be essentially the same except for scattered loops at the surface and the NH₂- and COOH-terminal tails, where the sequences diverge the most.

Structure of the homologous site. Figure 1 shows the sequence surrounding the PKA site (RRNSF) in the rat ₁ Na-K-ATPase aligned with the corresponding segment (ENQSL) of rabbit SERCA1a. The site lies on an intracellular loop between M8 and M9, and the serine itself is conserved. Under the alignment, identical amino acids are marked with asterisks, conservative substitutions with lines, and less conservative substitutions with dots. The degree of similarity, conservatively defined, is 35%, and with the broader definition it is ~70%. In the Ca-ATPase, the Arg-Arg sequence of the PKA motif is replaced with other polar amino acids, glutamic acid and asparagine.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Alignment of Ca-ATPase and Na-K-ATPase from M8 to M9. The alignment generated by gapped BLAST for the rabbit sarco(endo)plasmic reticulum Ca-ATPase (SERCA1a) and the rat 1 Na-K-ATPase is shown for the 2 transmembrane spans flanking the Na-K-ATPase cAMP-dependent protein kinase (PKA) site and the intervening cytoplasmic loop. The positions of the membrane helices for the Ca-ATPase, derived from the crystal structure, are indicated by black bars. The arrow indicates the valine in M8 that participates in ligation of Ca ion at site 1. Below the alignment, the degree of homology is indicated as identical (*), highly conserved (|), and more distantly conserved (.) residues, as scored by gapped BLAST with a BLOSUM62 substitution matrix.

The alignment of the Na-K-ATPase sequence with the SERCA1a sequence and structure was performed with Cn3D and its associated linear utility, DDV (National Center for Biotechnology Information, www.ncbi.nlm.nih.gov). The structure file for SERCA1a is MMDB (Molecular Modeling DataBase) ID 13684 and PDB (Protein DataBase) ID 1EUL, and the Na-K-ATPase rat ₁ sequence aligned with it was from GenBank accession no. M14511. Cn3D is not a molecular modeling program and does not generate a new hypothetical structure that takes into account sequence differences between the two sequences being aligned. It determines the optimal alignment with gapped BLAST (1) and displays the original structure with optional color coding of the degree of similarity at each residue. Consequently, the pictures shown here should be viewed not as Na-K-ATPase structures but as predicted locations based on sequence homology.

Figure 2 illustrates the location of the ENQSL site homologous to RRNSF relative to the -subunit as a whole. The domains of the Ca-ATPase -subunit are shown in different colors. Helices are represented by cylinders, -strands by arrows, and loops by a string that follows the backbone of -carbons. In Fig. 2A, ENQSL is highlighted in yellow. The location of the lipid bilayer was deduced in the crystal structure from the distribution of water molecules (57) and is shown here with an assumed thickness of 35 Å. It can be seen that the highlighted site is very close to the membrane. It appears to be covered from above by the nucleotide-binding domain (N domain; red), which in this conformation is tilted back from the phosphorylation domain (P domain; blue), but, in fact, the N domain is angled somewhat behind the plane of the paper and does not block the ENQSL site completely. Figure 2B shows the transmembrane domain (aqua) from the cytoplasmic surface with the other domains removed, and 2C shows it with the other domains present. The highlighted ENQSL sequence is blocked from view from the cytoplasm by the P domain immediately above the transmembrane domain and more distantly by the N domain. Note that the predicted location of the Na-K-ATPase's protein kinase C site near the NH₂ terminus (Fig. 2, A and C) is on the opposite end of the protein and is highly exposed on the actuator domain (A domain). (Because the Ca-ATPase is shorter at the NH₂ terminus and does not have a homologous segment, no structure is highlighted.) The predicted location of the PKA site, in contrast, is not well exposed.

View larger version (50K): [in this window] [in a new window]   Fig. 2.   Predicted location of the PKA site by correspondence with the Ca-ATPase. A: SERCA1a structure is shown with each domain in a different color. Pink is the NH2-terminal portion that contributes 2 -helices to the actuator (A) domain. The Na-K-ATPase protein kinase C (PKC) site is on a short NH2-terminal extension, so it is indicated close to the end of the NH2 terminus of the Ca-ATPase. The rest of the A domain is in green. The phosphorylation (P) domain is in dark blue and sits on top of the long (60 Å) M5 helix. The red nucleotide-binding (N) domain emerges from the P domain and is thought to close down over the P domain during catalysis. The transmembrane domain is shown in aqua, and bound Ca ions can be seen as pale spheres. The L6-7 and L8-9 cytoplasmic loops are labeled, and it is the L8-9 loop that contains the site ENQSL that corresponds to the Na-K-ATPase PKA site. The position of the lipid bilayer is shown as a yellow block. B: the transmembrane domain alone is shown, as viewed from the cytoplasm with the other domains cut away. The transmembrane spans are numbered. The ENQSL site is highlighted, and it can be seen that because M8 is not on the extreme periphery, the site is partially sunken between M9 and M10. C: the same view as in B, but with the cytoplasmic domains in place.

The structure of PKA itself (MMDB ID 7173, PDB ID 1FMO) is displayed next to the Ca-ATPase in Fig. 3. PKA binds to its substrate protein sequence in an extended conformation in the cleft between two domains, and the substrate's -carbon backbone is nearly linear. PKA is shown crystallized with an inhibitory peptide analog that has the sequence RRNAI in the active site (43). The inhibitory peptide is pink, but its substrate motif is highlighted in yellow. The inhibitory peptide is shown end on in this orientation but can be seen rotated 90° in Fig. 6C. The Ca-ATPase was rotated about 150° relative to what is shown in Fig. 2A to face the kinase, and the orientation of PKA was adjusted to match the presentation of the sequence ENQSL (also highlighted) in the Ca-ATPase as closely as possible. The yellow background represents the expected location of the lipid bilayer. It seems highly unlikely that PKA would be able to bind to a site in this location because of steric interference from the lipid bilayer.

View larger version (74K): [in this window] [in a new window]   Fig. 3.   PKA and Ca-ATPase. The Ca-ATPase has been rotated to make the ENQSL site face to the left, and the crystal structure of PKA complexed with an inhibitory active site peptide (pink) has been rotated to mimic the orientation it would have to adopt to approach the site on the ATPase. The active site of the kinase is in a cleft between 2 domains (green and aqua). The peptide's residues that mimic substrate have been highlighted yellow, as has ENQSL. The red and gray molecule in the PKA active site is adenosine. Steric interference from the lipid bilayer (yellow block) would make the approach of the kinase unlikely.

Figure 4 illustrates the structure of the ENQSL site on the Ca-ATPase at higher resolution and with a space-filling representation of atoms. ENQSL is highlighted in yellow. The orientation was rotated freely to optimize the visibility of the highlighted amino acids. To maximize the exposure of ENQSL, it was necessary to tilt the entire enzyme structure back. In Fig. 4A, the highlighted site is shown at low magnification from the same angle used in C and D. In Fig. 4B, the enzyme was rotated 90° to show how far backwards it was tilted. The "domain" color scheme of Cn3D was used in Fig. 4, A and B, as it was in the earlier figures. The "weighted variety" color scheme was used in Fig. 4, C and D, a feature of Cn3D that depicts how similar the aligned Na-K-ATPase residues are to the Ca-ATPase residues. The weighting entails application of a substitution probability matrix (BLOSUM62) that scores conservative substitutions higher than nonconservative ones. Redder is more conserved, bluer is less conserved, and if Na-K-ATPase sequence does not align with Ca-ATPase sequence at all (as is true of the last 5 residues of the Ca-ATPase COOH terminus), the Ca-ATPase residues are displayed in gray. What you can see is that the site is relatively sunken in surrounding residues and that the COOH terminus (gray) comes very close. The curvature of the site's -carbon backbone can be seen most clearly in Fig. 4C.

View larger version (103K): [in this window] [in a new window]   Fig. 4.   Close-up view of the location of the PKA site homologue. The Ca-ATPase has been rotated to maximize the visibility of ENQSL (yellow). A and B: the structure at low magnification to illustrate that it had to be tilted backwards to optimize the visibility of the highlighted residues. C: the site at higher magnification, illustrating the curvature of its -carbon backbone and the proximity of the Ca-ATPase COOH terminus (gray, not aligned with Na-K-ATPase). The colors in C and D reflect the degree of homology between Ca-ATPase and Na-K-ATPase. D: the structure is displayed in spacefill mode, with the residues labeled.

In Fig. 5, the accessibility of the ENQSL sequence is shown without tilting the enzyme backwards, and different orientations are displayed. When the site is viewed from different angles, the glutamic acid and asparagine are together on one face, whereas the serine is barely exposed on another face because of the curvature of the -carbon backbone. In Fig. 5, A-C, only the serine residue is highlighted, and all three panels show the same orientation. Figure 5, A and B, use the "secondary structure" color scheme of Cn3D. Helices are colored green and loops blue so that it can be seen that the serine lies at the boundary of the transmembrane helix bundle and the loops found on the cytoplasmic surface. In Fig. 5C, the weighted variety color scheme is used. From that, it can again be seen how the COOH terminus, including five unaligned amino acids (gray), comes close to the serine. This could impede the approach of another protein. In Fig 5D, ENQS is highlighted, and the structure is rotated counterclockwise relative to Fig. 5, A-C. The glutamic acid and asparagine residues are exposed under the COOH terminus, and the serine is not visible from this angle. Figure 5, E and F, illustrates crowding by the COOH terminus (gray in Fig. 5E) from another angle. The unaligned COOH-terminal amino acids were deleted from the display in Fig. 5F; the site is still crowded, principally by the phenylalanine residue from the sequence LKFIAR close to M10. In the Na-K-ATPase, the corresponding residues are RKLIIR, a degree of conservation that predicts that they will have the same fold, and leucine is a relatively large hydrophobic residue, like phenylalanine. The disposition of the rest of the COOH terminus of the Na-K-ATPase cannot be predicted at all, of course, and the gray segment could even be rotated out of the way. The COOH terminus of the Na-K-ATPase is an additional eight amino acids longer than the Ca-ATPase, however, so it may obscure the PKA site even more extensively.

View larger version (93K): [in this window] [in a new window]   Fig. 5.   The site from other angles. A-C: all show the same view, and only the conserved serine residue is highlighted. A is the secondary structure view, whereas B is spacefill. It can be seen that the serine lies at the edge of the all green, all helical membrane domain (B). The colors represent degree of similarity to the Na-K-ATPase (C). The serine is surrounded by residues that were identical (red), very conserved (mauve), or somewhat similar (purple). Blue residues are not conserved, and gray (from the COOH terminus) are not aligned at all. D: the structure rotated to the right, where the exposure of residues E, N, and Q (homologous to R, R, and N) are exposed the most. E and F: a view rotated half-way between C and D. The COOH terminus blocks the site from view. In F, the unaligned (gray) COOH terminus has been cut away, but a phenylalanine (labeled F) from closer to M10 still crowds the site. The same phenylalanine is labeled in D.

Naturally, the atomic details of the Na-K-ATPase structure will be different because amino acid substitutions will result in some repositioning. Given the degree of similarity and the similar length of the M8-M9 hairpin, however, it seems unlikely that the structure would be so different from the corresponding site on the Ca-ATPase that it would be accessible to PKA. As will be discussed later, it is also possible that the Na-K-ATPase -subunit, which has no homologue in the Ca-ATPase, obscures the PKA site because of its location close to M8.

Structure of a PKA autophosphorylation site. For comparison, the structure of a known PKA site is shown in Fig. 6. This is the autophosphorylation site on PKA itself (MMDB ID 4194, PDB ID 1CDK), which is located on the activation loop near but not in the active site (31). The structure is the same as that pictured in Fig. 2B, but now the sequence RTWT*L is highlighted. The phosphate on the second threonine is shown in red. The underlying -carbon backbone is stretched out (Fig. 6A), and the site is highly exposed on the surface (6B). In Fig. 6, C and D, the structure has been rotated 90° to display the inhibitory peptide (pink) that lies in the active site. It, too, adopts a stretched-out conformation.

View larger version (95K): [in this window] [in a new window]   Fig. 6.   An autophosphorylation site on PKA. The PKA structure is the same as that shown in Fig. 3, except that the adenosine in the active site is not displayed. The inhibitory peptide is pink, and the autophosphorylation site, RTWTL, is highlighted in yellow. A and B: the autophosphorylation site in extended conformation, while the inhibitory peptide is seen end on. The red atoms are the phosphate on the threonine of the autophosphorylation site. C and D: the structure rotated 90° so that the autophosphorylation site is viewed end on; the inhibitory peptide in the active site is seen in an extended conformation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION DATA ANALYSIS DISCUSSION REFERENCES

Predicted inaccessibility of the PKA site. The structure of the Ca-ATPase loop homologous to the Na-K-ATPase PKA site suggests that the PKA site should be inaccessible to kinase. This is for several reasons: its relatively sunken position, proximity to the membrane, crowding by the nearby COOH-terminal tail, the partially overhanging N domain, and the potential for - (and -) subunits to lie nearby. The true structure of the Na-K-ATPase is, of course, not yet known, and it is possible that there are enough adjustments to the M8-M9 loop to make the site locally more prominent and to move away the residues that crowd it and the COOH terminus. The site should remain close to the membrane and tucked under the P domain, however, which would make the approach of a kinase difficult. It is notable that the arginine residues have not been observed to be digested by trypsin in membrane-bound Na-K-ATPase, although digestion did occur in denatured or chemically modified enzyme (39, 42, 48), and digestion at (K)ILIF (the last residue before the beginning of M9) has been reported in enzyme destabilized by Ca ions (52).

The PKA site is extremely well conserved within the X-K-ATPase subgroup of the P2-type ATPases, which includes the ₂-, ₃-, and ₄-isoforms, the gastric H-K-ATPase, and the colonic X-K-ATPase (49). Logic would dictate that a canonical PKA consensus motif should be used for some universally applicable regulation event, but the sequence could be conserved simply as a structural element. Epitope insertions near the PKA site have proven to be disruptive. In one case, insertion of the hemagglutinin epitope (YPYDVPDYA) at Arg-936 resulted in exposure of the M8-M9 loop on the extracellular surface, inspiring an alternative folding model with fewer transmembrane spans (13). In another case, insertion of the same epitope at Val-939 was shown to result in very low transfection and cloning efficiency, suggestive of impaired Na-K-ATPase expression or function (58).

Experimental increases in accessibility. It is clear that the Na-K-ATPase can be phosphorylated at the PKA site in the presence of Triton X-100. At the very least, the detergent should eliminate the planar lipid bilayer, leaving an annulus of detergent and, perhaps, allowing closer approach of the kinase from the side.

There is evidence that detergents can cause a more profound structural alteration of the Na-K-ATPase. Covalent oxidative cross-linking of cysteine residues in trypsin-digested Na-K-ATPase has been shown to result in a cross-link between the -subunit, the M1-M2 hairpin, and the M7-10 COOH-terminal tryptic fragment of the Na-K-ATPase (47, 51). To obtain a high yield of cross-linking of the -subunit to the M7-10 fragment, treatment with digitonin was required, however (50). Most recently, it was shown that  oxidatively cross-links specifically to M8, but only in digitonin (30, 47). Cys-44 of the -subunit can apparently get close enough to Cys-911 or Cys-930 of the -subunit M8 segment for cross-linking when the protein is perturbed with detergent. When the homologous M8 segment of the Ca-ATPase is viewed in the Ca-ATPase structure, the positions equivalent to Cys-911 and Cys-930 are at either end of M8 and may not be in the lipid bilayer at all (data not shown). Cys-44 of  is predicted to be near the center of the membrane span (47), so M8 would have to either ascend or descend relative to  to obtain cross-linking, and some residues from the flanking helices and loops would have to move out of the way. These motions could have consequences for the exposure of RRNSF just above the cytoplasmic end of M8.

Exposure of the site by denaturation is trivial and easily accounts for phosphorylation after mild heating and harsh reconstitution procedures. It may also account for the exposure of the site in transfected cells, however. A common practice is to transfect only the -subunit and to use a reasonably strong promoter to get good levels of expression. The endogenous Na-K-ATPase is still made, and the exogenous Na-K-ATPase must compete with it for access to the endogenous -subunit, which is required for correct folding and assembly. Most investigators do not look at the distribution of the expressed -subunit in the microscope, but Takeyasu et al. (54) showed in transfected cells that it can accumulate in intracellular compartments, as would be deduced from everything that is now known about assembly. When investigators observe regulatory effects of PKA activation on Na-K-ATPase activity, they must be detecting effects on properly assembled enzyme units. When they detect incorporation of ³²P or staining by phospho-sensitive antibody, however, the entire cell extract (or the entire population of immunoprecipitable -subunit) is examined, including both folded and misfolded enzyme. Because RRNSF is a prototypical PKA phosphorylation site, its exposure on misfolded -subunit may fortuitously serve as a good indicator of PKA activation. It could be expected that the levels of phosphorylation seen would parallel physiological pathway activation, even if the phosphorylation event is not in itself physiological. In future experiments, this problem might be solved by examining the phosphorylation state only of -subunits that can be immunoprecipitated with anti--antibody.

The implications of conformation differences. The discouraging analysis above might be incorrect if the Na-K-ATPase PKA site is exposed specifically in different enzyme conformations or by interaction with other cellular components. Interestingly, enzyme conformation appeared to play a role in the extent of Na-K-ATPase phosphorylation in the test tube, even though detergent was required (25). The Na-K-ATPase, like other P-type ATPases, is thought to cycle through conformation changes that alternately expose Na and K binding sites on opposite sides of the membrane. There is an abundance of physical and kinetic evidence for two major conformations termed E1 and E2 (33). The conformation changes are thought to entail large interdomain movements (46). These conformation changes have been long known to affect the accessibility of certain proteolytic cleavage sites (34). Experimentally, despite the presence of enough Triton X-100 to inactivate ATP hydrolysis, we observed that conditions favoring the E1 conformation resulted in higher levels of PKA phosphorylation than conditions favoring E2 (25).

The conformation of the Ca-ATPase in the crystal structure is not certain, but it is quite different from that observed by cryoelectron microscopy in the same enzyme crystallized without Ca and with decavanadate in the active site (46, 57, 60). The 2.6 Å structure has wide-spread domains, whereas the 8 Å structure is more condensed, so the 2.6 Å structure may represent E1 or a conformation related to it. The extent of PKA phosphorylation in different ligands suggests that the site should be even less exposed in E2 (25). In fact, in our hands, ouabain, which induces an E2-like conformation, completely blocked phosphorylation of pig kidney Na-K-ATPase. One could argue, however, that the rotation of the N domain closer to the A domain in E2 could conceivably expose the PKA site better, and the site could emerge more from the lipid bilayer. A salient example of a specific change in accessibility of sites is the demonstration by Lutsenko et al. (37) of changes in the accessibility of an extracellularly exposed cysteine residue, Cys-964 of sheep ₁, to reaction with maleimide derivatives. This residue was reactive when the enzyme's active site aspartate was phosphorylated but was no longer exposed after dephosphorylation and K occlusion. This residue is connected to the PKA site through M9, and it is conceivable that motion resulting in the burial of one site would cause better exposure of the other.

A caveat is that the Na-K-ATPase goes through its catalytic cycle every 6-10 ms when pumping at maximal velocity (V_max). The dwell time in any one conformation is shorter, which should reduce the likelihood of phosphorylation unless PKA is anchored in position. In principle, it is possible that inactive enzyme units could become trapped in a dead-end conformation through the action of protein kinase A. This idea goes beyond a reasonable level of speculation, however, because there are no experimental data to support it, and it does not have known physiological relevance.

Possible physiological regulatory mechanisms. It should be noted that despite all the negative evidence discussed earlier in this paper, Cheng et al. (16) and Fisone et al. (27) have reported blockage of Na-K-ATPase regulation by PKA in transfectants expressing S938A, in the latter case modulated by the intracellular concentration of Ca. They also reported the involvement of phosphorylation of Ser-938 in modulation of phosphorylation by protein kinase C in similar transfectants (17). Difficulties in demonstrating phosphorylation in purified enzyme and isolated microsomes might be caused by separation of the enzyme from important regulatory components. Protein kinase A is now known to associate with anchoring proteins (23). A long time ago, Lingham and Sen (36) reported that an intermediary protein was required in the regulation of the Na-K-ATPase by PKA.

There is evidence for physical interaction of the Na-K-ATPase -subunit with two different proteins: ankyrin, which binds to a segment between M2 and M3 (32), and an SH3 domain of phosphatidylinositol 3-kinase, which binds to a polyproline site, TPPPTPP, before M1 (59). Examination of the aligned Na-K- ATPase and Ca-ATPase sequences in Cn3D shows that the corresponding locations of both of these sites are well exposed in the Ca²⁺-ATPase, and both are a considerable distance from the PKA site (data not shown). Neither interacting protein is known to directly affect Na-K-ATPase activity, although phosphorylation by protein kinase C is thought to influence the SH3 domain site in clathrin recruitment for regulatory internalization (59). One can speculate that one of these, or another Na-K-ATPase binding protein, could alter the disposition of the PKA site enough to permit phosphorylation, resulting in a quasi-stable alteration in function.

Convergence of regulatory pathways in the modulation of the Na-K-ATPase has been reported in several systems (7, 9, 12, 28). This could occur by many mechanisms independent of direct phosphorylation of , however, not the least of which is regulation of protein phosphatase activity with consequences at other levels of the pathway (45).

Conclusions. On the whole, evidence for direct phosphorylation of the Na-K-ATPase at Ser-938 in the native state is weak. Examination of the location of the homologous sequence in the atomic structure of the Ca-ATPase suggests that the site should be inaccessible to the kinase. The most conservative interpretation is that much of the cAMP-dependent Na-K-ATPase regulation known to occur in cells is carried out without direct covalent phosphorylation of the enzyme. It is possible that E1-E2 conformation changes control accessibility of the site, but the likelihood is diminished in view of the rapid cycle of conformation changes at V_max. It remains possible that other cellular proteins interact with the Na-K-ATPase in vivo to improve the accessibility of the site to kinase. In this case, regulation should be complex in the sense that two regulatory pathways may have to converge to achieve phosphorylation of the Na-K-ATPase.

	ACKNOWLEDGEMENTS

This work was supported by National Institutes of Health Grants RO1NS-27653 and RO1HL-36271 (to K. J. Sweadner).

	FOOTNOTES

Address for reprint requests and other correspondence: K. J. Sweadner, Laboratory of Membrane Biology, Neuroscience Center, Massachusetts General Hospital, Charlestown, MA 02129 (E-mail: sweadner{at}helix.mgh.harvard.edu).

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 19 September 2000; accepted in final form 8 November 2000.

	REFERENCES

TOP ABSTRACT INTRODUCTION DATA ANALYSIS DISCUSSION REFERENCES

1.   Altschul, SF, Madden TL, Schaffer AA, Zhang J, Miller W, and Lipman DJ. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389-3402, 1997[Abstract/Free Full Text].

2.   Anderberg, SJ. Topological disposition of lysine 943 in native Na⁺/K⁺-transporting ATPase. Biochemistry 34: 9508-9516, 1995[Medline].

3.   Arystarkhova, E, Gasparian M, Modyanov NN, and Sweadner KJ. Na,K-ATPase extracellular surface probed with a monoclonal antibody that enhances ouabain binding. J Biol Chem 267: 13694-13701, 1992[Abstract/Free Full Text].

4.   Arystarkhova, E, Gibbons DL, and Sweadner KJ. Topology of the Na,K-ATPase: evidence for externalization of a labile transmembrane structure during heating. J Biol Chem 270: 8785-8796, 1995[Abstract/Free Full Text].

5.   Beguin, P, Beggah A, Cotecchia S, and Geering K. Adrenergic, dopaminergic, and muscarinic receptor stimulation leads to PKA phosphorylation of Na-K-ATPase. Am J Physiol Cell Physiol 270: C131-C137, 1996[Abstract/Free Full Text].

6.   Beguin, P, Beggah AT, Chibalin AV, Burgener-Kairuz P, Jaisser F, Mathews PM, Rossier BC, Cotecchia S, and Geering K. Phosphorylation of the Na,K-ATPase -subunit by protein kinases A and C in vitro and in intact cells. Identification of a novel motif for PKC-mediated phosphorylation. J Biol Chem 269: 24437-24445, 1994[Abstract/Free Full Text].

7.   Bertorello, A, and Aperia A. Inhibition of proximal tubule Na,K-ATPase activity requires simultaneous activation of DA1 and DA2 receptors. Am J Physiol Renal Fluid Electrolyte Physiol 259: F924-F928, 1990[Abstract/Free Full Text].

8.   Bertorello, AM, Aperia A, Walaas SI, Nairn AC, and Greengard P. Phosphorylation of the catalytic subunit of Na⁺,K⁺-ATPase inhibits the activity of the enzyme. Proc Natl Acad Sci USA 88: 11359-11362, 1991[Abstract/Free Full Text].

9.   Bertorello, AM, Hopfield JF, Aperia A, and Greengard P. Inhibition by dopamine of (Na⁺ + K⁺)-ATPase activity in neostriatal neurons through D1 and D2 dopamine receptor synergism. Nature 347: 386-388, 1990[Medline].

10.   Bertorello, AM, Ridge KM, Chibalin AV, Katz AI, and Sznajder JI. Isoproterenol increases Na⁺-K⁺-ATPase activity by membrane insertion of -subunits in lung alveolar cells. Am J Physiol Lung Cell Mol Physiol 276: L20-L27, 1999[Abstract/Free Full Text].

11.   Blanco, G, Sanchez G, and Mercer RW. Differential regulation of Na,K-ATPase isozymes by protein kinases and arachidonic acid. Arch Biochem Biophys 359: 139-150, 1998[Web of Science][Medline].

12.   Borghini, I, Geering K, Gjinovci A, Wollheim CB, and Pralong WF. In vivo phosphorylation of the Na,K-ATPase  subunit in sciatic nerves of control and diabetic rats: effects of protein kinase modulators. Proc Natl Acad Sci USA 91: 6211-6215, 1994[Abstract/Free Full Text].

13.   Canfield, VA, and Levenson R. Transmembrane organization of the Na,K-ATPase determined by epitope addition. Biochemistry 32: 13782-13786, 1993[Medline].

14.   Carranza, ML, Feraille E, Kiroytcheva M, Rousselot M, and Favre H. Stimulation of ouabain-sensitive ⁸⁶Rb⁺ uptake and Na,K-ATPase -subunit phosphorylation by a cAMP-dependent signalling pathway in intact cells from rat kidney cortex. FEBS Lett 396: 309-314, 1996[Web of Science][Medline].

15.   Cheng, SXJ, Aizman O, Nairn AC, Greengard P, and Aperia A. [Ca²⁺]_i determines the effects of protein kinases A and C on activity of rat renal Na⁺,K⁺-ATPase. J Physiol (Lond) 518: 37-46, 1999[Abstract/Free Full Text].

16.   Cheng, XJ, Fisone G, Aizman O, Aizman R, Levenson R, Greengard P, and Aperia A. PKA-mediated phosphorylation and inhibition of Na,K-ATPase ₁-subunit in response to -adrenergic hormone. Am J Physiol Cell Physiol 273: C893-C901, 1997[Abstract/Free Full Text].

17.   Cheng, XJ, Hoog JO, Nairn AC, Greengard P, and Aperia A. Regulation of rat Na-K-ATPase activity by PKC is modulated by state of phosphorylation of Ser-943 by PKA. Am J Physiol Cell Physiol 273: C1981-C1986, 1997[Abstract/Free Full Text].

18.   Chibalin, AV, Lopina OD, Petrukhov SP, and Vasilets LA. Phosphorylation of the Na,K-ATPase by Ca, phospholipid-dependent and cAMP-dependent protein kinases. Mapping of the region phosphorylated by Ca, phospholipid-dependent protein kinase. J Bioenerg Biomembr 25: 61-66, 1993[Web of Science][Medline].

19.   Chibalin, AV, Lopina OD, Petukhov SP, and Vasilets LA. Phosphorylation of Na⁺,K⁺-ATPase by protein kinase C and cAMP-dependent protein kinase. Biol Membrany 8: 1440-1441, 1991.

20.   Chibalin, AV, Vasilets LA, Hennekes H, Pralong D, and Geering K. Phosphorylation of Na,K-ATPase  subunits in microsomes and in homogenates of Xenopus oocytes resulting from the stimulation of protein kinase A and protein kinase C. J Biol Chem 267: 22378-22384, 1992[Abstract/Free Full Text].

21.   Cornelius, F, and Logvinenko N. Functional regulation of reconstituted Na,K-ATPase by protein kinase A phosphorylation. FEBS Lett 380: 277-280, 1996[Web of Science][Medline].

22.   Delamere, NA, Socci RR, and King KL. Alteration of sodium, potassium-adenosine triphosphatase activity in rabbit ciliary processes by cyclic adenosine monophosphate-dependent protein kinase. Invest Ophthalmol Vis Sci 31: 2164-2170, 1990[Abstract/Free Full Text].

23.   Edwards, AS, and Scott JD. A-kinase anchoring proteins: protein kinase A and beyond. Curr Opin Cell Biol 12: 217-221, 2000[Web of Science][Medline].

24.   Felsenfeld, DP, and Sweadner KJ. Fine specificity mapping and topography of an isozyme-specific epitope of the Na,K- ATPase catalytic subunit. J Biol Chem 263: 10932-10942, 1988[Abstract/Free Full Text].

25.   Feschenko, MS, and Sweadner KJ. Conformation-dependent phosphorylation of Na,K-ATPase by protein kinase A and protein kinase C. J Biol Chem 269: 30436-30444, 1994[Abstract/Free Full Text].

26.   Feschenko, MS, and Sweadner KJ. Interaction of protein kinase C and cAMP-dependent pathways in the phosphorylation of the Na,K-ATPase. J Biol Chem 275: 34693-34700, 2000[Abstract/Free Full Text].

27.   Fisone, G, Cheng SXJ, Nairn A, Czernik A, Hemmings H, Jr, Hoog JO, Bertorello AM, Kaiser R, Bergman T, Jornvall H, Aperia A, and Greengard P. Identification of the phosphorylation site for cAMP-dependent protein kinase on Na⁺,K⁺-ATPase and effects of site-directed mutagenesis. J Biol Chem 269: 9368-9373, 1994[Abstract/Free Full Text].

28.   Fisone, G, Snyder GL, Aperia A, and Greengard P. Na,K-ATPase phosphorylation in the choroid plexus: synergistic regulation by serotonin/protein kinase C and isoproterenol/cAMP-PK/PP-1 pathways. Mol Med 4: 258-265, 1998[Web of Science][Medline].

29.   Hu, YK, and Kaplan JH. Site-directed chemical labeling of extracellular loops in a membrane protein. The topology of the Na,K-ATPase -subunit. J Biol Chem 275: 19185-19191, 2000[Abstract/Free Full Text].

30.   Ivanov, A, Zhao H, and Modyanov NN. Packing of the transmembrane helices of Na,K-ATPase: direct contact between -subunit and H8 segment of -subunit revealed by oxidative cross-linking. Biochemistry 39: 9778-9785, 2000[Medline].

31.   Johnson, LN, Noble MEM, and Owen DJ. Active and inactive protein kinases: structural basis for regulation. Cell 85: 149-158, 1996[Web of Science][Medline].

32.   Jordan, C, Puschel B, Koob R, and Drenckhahn D. Identification of a binding motif for ankyrin on the -subunit of Na⁺,K⁺-ATPase. J Biol Chem 270: 29971-29975, 1995[Abstract/Free Full Text].

33.   Jorgensen, PL, Nielsen JM, Rasmussen JH, and Pedersen PA. Structure-function relationships of E1-E2 transitions and cation binding in Na,K-pump protein. Biochim Biophys Acta 1365: 65-70, 1998[Medline].

34.   Jorgensen, PL, and Petersen J. Purification and characterization of Na,K-ATPase: differential tryptic modification of catalytic functions of the purified enzyme in presence of NaCl and KCl. Biochim Biophys Acta 466: 97-108, 1977[Medline].

35.   Kiroytcheva, M, Cheval L, Carranza ML, Martin PY, Favre H, Doucet A, and Feraille E. Effect of cAMP on the activity and the phosphorylation of Na⁺,K⁺-ATPase in rat thick ascending limb of Henle. Kidney Int 55: 1819-1831, 1999[Web of Science][Medline].

36.   Lingham, R, and Sen AK. Regulation of rat brain (Na⁺ + K⁺)-ATPase activity by cyclic AMP. Biochim Biophys Acta 688: 475-485, 1982[Medline].

37.   Lutsenko, S, Daoud S, and Kaplan JH. Identification of two conformationally sensitive cysteine residues at the extracellular surface of the Na,K-ATPase -subunit. J Biol Chem 272: 5249-5255, 1997[Abstract/Free Full Text].

38.   Mardh, S. Phosphorylation by the catalytic subunit of protein kinase of a preparation of kidney Na,K-ATPase. In: Na,K-ATPase Structure and Kinetics, edited by Skou JC, and Norby JG.. London: Academic, 1979, p. 359-370.

39.   Modyanov, N, Lutsenko S, Chertova E, and Efremov R. Architecture of the sodium pump molecule: probing the folding of the hydrophobic domain. Soc Gen Physiol Ser 46: 99-115, 1991[Medline].

40.   Mohraz, M, Arystarkhova E, and Sweadner KJ. Immunoelectronmicroscopy of epitopes on Na,K-ATPase catalytic subunit. Implications for the transmembrane organization of the C-terminal domain. J Biol Chem 269: 2929-2936, 1994[Abstract/Free Full Text].

41.   Moller, JV, Juul B, and Le Maire M. Structural organization, ion transport, and energy transduction of P-type ATPases. Biochim Biophys Acta 1286: 1-51, 1996[Medline].

42.   Munson, KB, Gutierrez C, Balaji VN, Ramnarayan K, and Sachs G. Identification of an extracytoplasmic region of H⁺,K⁺-ATPase labeled by a K⁺-competitive photoaffinity inhibitor. J Biol Chem 266: 18976-18988, 1991[Abstract/Free Full Text].

43.   Narayana, N, Cox S, Shaltiel S, Taylor SS, and Xuong N. Crystal structure of a polyhistidine-tagged recombinant catalytic subunit of cAMP-dependent protein kinase complexed with the peptide inhibitor PKI(5-24) and adenosine. Biochemistry 36: 4438-4448, 1997[Medline].

44.   Ning, G, Maunsbach AB, Lee YJ, and Moller JV. Topology of Na,K-ATPase -subunit epitopes analyzed with oligopeptide-specific antibodies and double-labeling immunoelectron microscopy. FEBS Lett 336: 521-524, 1993[Web of Science][Medline].

45.   Nishi, A, Fisone G, Snyder GL, Dulubova I, Aperia A, Nairn AC, and Greengard P. Regulation of Na⁺,K⁺-ATPase isoforms in rat neostriatum by dopamine and protein kinase C. J Neurochem 73: 1492-1501, 1999[Web of Science][Medline].

46.   Ogawa, H, Stokes DL, Sasabe H, and Toyoshima C. Structure of the Ca²⁺ pump of sarcoplasmic reticulum: a view along the lipid bilayer at 9-A resolution. Biophys J 75: 41-52, 1998[Web of Science][Medline].

47.   Or, E, Goldshleger R, and Karlish SJD Characterization of disulfide cross-links between fragments of proteolyzed Na,K-ATPase. J Biol Chem 274: 2802-2809, 1999[Abstract/Free Full Text].

48.   Ovchinnikov, Y, Arzamazova NM, Arystarkhova EA, Gevondyan NM, Aldanova NA, and Modyanov NN. Detailed structural analysis of exposed domains of membrane-bound Na⁺,K⁺-ATPase. A model of transmembrane arrangement. FEBS Lett 217: 269-274, 1987[Web of Science][Medline].

49.   Palmgren, MG, and Axelsen KB. Evolution of P-type ATPases. Biochim Biophys Acta 1365: 37-45, 1998[Medline].

50.   Sarvazyan, NA, Ivanov A, Modyanov NN, and Askari A. Ligand-sensitive interactions among the transmembrane helices of Na⁺/K⁺-ATPase. J Biol Chem 272: 7855-7858, 1997[Abstract/Free Full Text].

51.   Sarvazyan, NA, Modyanov NN, and Askari A. Intersubunit and intrasubunit contact regions of Na⁺/K⁺-ATPase revealed by controlled proteolysis and chemical cross-linking. J Biol Chem 270: 26528-26532, 1995[Abstract/Free Full Text].

52.   Shainskaya, A, and Karlish SJD Evidence that the cation occlusion domain of Na/K-ATPase consists of a complex of membrane-spanning segments. J Biol Chem 269: 10780-10789, 1994[Abstract/Free Full Text].

53.   Shimon, MB, Goldshleger R, and Karlish SJD Specific Cu²⁺-catalyzed oxidative cleavage of Na,K-ATPase at the extracellular surface. J Biol Chem 273: 34190-34195, 1998[Abstract/Free Full Text].

54.   Takeyasu, K, Tamkun MM, Renaud KJ, and Fambrough DM. Ouabain-sensitive (Na⁺ + K⁺)-ATPase activity expressed in mouse L cells by transfection with DNA encoding the -subunit of an avian sodium pump. J Biol Chem 263: 4347-4354, 1988[Abstract/Free Full Text].

55.   Therien, AG, and Blostein R. Mechanisms of sodium pump regulation. Am J Physiol Cell Physiol 279: C541-C566, 2000[Abstract/Free Full Text].

56.   Thibault, D. The carboxy terminus of sodium and potassium ion transporting ATPase is located on the cytoplasmic surface of the membrane. Biochemistry 32: 2813-2821, 1993[Medline].

57.   Toyoshima, C, Nakasako M, Nomura H, and Ogawa H. Crystal structure of the calcium pump of sarcoplasmic reticulum at 2.6 A resolution. Nature 405: 647-655, 2000[Medline].

58.   Yoon, KL, and Guidotti G. Studies on the membrane topology of the (Na,K)-ATPase. J Biol Chem 269: 28249-28258, 1994[Abstract/Free Full Text].

59.   Yudowski, GA, Efendiev R, Pedemonte CH, Katz AI, Berggren PO, and Bertorello A. Phosphoinositide-3 kinase binds to a proline-rich motif in the Na⁺,K⁺-ATPase  subunit and regulates its trafficking. Proc Natl Acad Sci USA 97: 6556-6561, 2000[Abstract/Free Full Text].

60.   Zhang, P, Toyoshima C, Yonekura K, Green NM, and Stokes DL. Structure of the calcium pump from sarcoplasmic reticulum at 8-A resolution. Nature 392: 835-839, 1998[Medline].