Functional Interactions of Phospholemman (PLM) (FXYD1) with Na+,K+-ATPase

Human FXYD1 (phospholemman, PLM) has been expressed in Pichia pastoris with porcine α1/His10-β1 subunits of Na+,K+-ATPase or alone. Dodecyl-β-maltoside-soluble complexes of α1/β1/PLM have been purified by metal chelate chromatography, either from membranes co-expressing α1,His10-β1, and PLM or by in vitro reconstitution of PLM with α1/His10-β1 subunits. Comparison of functional properties of purified α1/His10-β1 and α1/His10-β1/PLM complexes show that PLM lowered K0.5 for Na+ ions moderately (≈30%) but did not affect the turnover rate or Km of ATP for activating Na+,K+-ATPase activity. PLM also stabilized the α1/His10-β1 complex. In addition, PLM markedly (>3-fold) reduced the K0.5 of Na+ ions for activating Na+-ATPase activity. In membranes co-expressing α1/His10-β1 with PLM the K0.5 of Na+ ions was also reduced, compared with the control, excluding the possibility that detergent or lipid in purified complexes compromise functional interactions. When expressed in HeLa cells with rat α1, rat PLM significantly raised the K0.5 of Na+ ions, whereas for a chimeric molecule consisting of transmembranes segments of PLM and extramembrane segments of FXYD4, the K0.5 of Na+ ions was significantly reduced, compared with the control. The opposite functional effects in P. pastoris and HeLa cells are correlated with endogenous phosphorylation of PLM at Ser68 or unphosphorylated PLM, respectively, as detected with antibodies, which recognize PLM phosphorylated at Ser68 (protein kinase A site) or unphosphorylated PLM. We hypothesize that PLM interacts with α1/His10-β1 subunits at multiple locations, the different functional effects depending on the degree of phosphorylation at Ser68. We discuss the role of PLM in regulation of Na+,K+-ATPase in cardiac or skeletal muscle cells.

Six members of the family, FXYD1 (PLM), FXYD2 (␥), FXYD3 (Mat-8), FXYD4 (CHIF), FXYD5 (RIC), and FXYD7, are now known to interact specifically with the Na ϩ ,K ϩ -ATPase, and modulate the functional properties (1)(2)(3)(4). The FXYD proteins are not required for basic pump function, but they act as accessory subunits to the ␣and ␤-subunits, and modulate the kinetic properties in a tissue-specific fashion. PLM was discovered originally as a major substrate for PKC and for PKA accompanying ␣and ␤-adrenergic stimulation of heart (18 -20), and it is also a substrate of other kinases (21,22). Although expression of PLM was shown previously to induce chloride currents in Xenopus oocytes, and taurine efflux and regulatory volume decrease in mammalian cells (23)(24)(25), in regard to its physiological role in cardiac contraction, more attention has been paid recently to the functional interaction of PLM with Na ϩ ,K ϩ -ATPase.
After co-expression in Xenopus oocytes together with rat ␣-isoforms, PLM was co-immunoprecipitated with either ␣1 or ␣2 and ␣3, and co-immunoprecipitation with either ␣1 or ␣2 was also seen for native rat heart sarcolemma, albeit more weakly with ␣2 (26). In Xenopus oocytes The PLM induced an almost 2-fold decrease in apparent affinity for cytoplasmic Na ϩ , a small decrease in apparent affinity for extracellular K ϩ , and little or no effect on the maximal pump current or on ouabain binding. As mentioned above, PLM has been detected in bovine choroid plexus membranes, and an anti-PLM antibody reduces V max without affecting the apparent Na ϩ affinity (16). This result does not necessarily contradict the findings with Xenopus oocytes because the antibody can neutralize interactions only at the C terminus of PLM, whereas the effect on apparent Na ϩ affinity may be mediated by other segments of PLM. However, a recent study using sarcolemma membrane obtained from hearts of PLM knock-out mice showed a lower V max of Na ϩ ,K ϩ -ATPase, even after taking into account a ϳ20% lower expression of the ␣-subunit, but again, no difference in apparent Na ϩ affinity (27). The implication is that PLM stimulates Na ϩ ,K ϩ -ATPase activity, with no change in Na ϩ affinity. This does not fit well with the observations in Xenopus oocytes.
There is some evidence that phosphorylation of PLM affects its functional interactions. For example, Na ϩ ,K ϩ -ATPase in sarco-lemma membranes isolated from ischemic rat hearts has been found to be strongly stimulated compared with the controls, and this is accompanied by activation of PKA and PKC and phosphorylation of PLM (but not of the ␣-subunit) ( (28)). Two recent studies have concluded that, in cardiac myocytes, phosphorylation of PLM at the PKA site (Ser 68 ) mediates the stimulatory effects of ␤-adrenergic agonists on Na ϩ /K ϩ -pump function (29,30). It was proposed that PLM raises the K 0.5 of Na ϩ ion for activating Na ϩ ,K ϩ -ATPase and phosphorylation relieves this effect. Similarly in an experimental model of heart failure, phosphorylation of rabbit cardiac myocyte PLM is greatly increased (31). It has also been demonstrated recently that phosphorylation of PLM is required for translocation of PLM from the endoplasmic reticulum to the plasma membrane in mammalian cell (32).
An FXYD protein in shark rectal gland membranes, which co-purifies with the ␣and ␤-subunits, is termed PLMS because of its PLM-like phosphorylation sites, although the closest mammalian sequence homolog is in fact Mat-8 (33)(34)(35). Phosphorylation of PLMS by PKC or selective proteolysis of the C terminus, which removes the phosphorylated serine, increases Na ϩ ,K ϩ -ATPase activity. Thus, the Na ϩ ,K ϩ -ATPase activity is partially inhibited when the enzyme interacts with intact or unphosphorylated PLMS, and inhibition is relieved upon proteolysis or PKC phosphorylation (1).
Structural interactions of FXYD proteins with ␣/␤ subunits are not well understood, but information is accumulating (3). There is now good evidence that the transmembrane segments of FXYD1 (PLM), FXYD2 (␥), FXYD4 (CHIF), and probably all other FXYD proteins lie within a groove bounded by M2, M6, and M9 of the ␣-subunit (36 -39). There is also evidence for multiple interactions in the case of FXYD2, which imply that extramembrane domains also interact with the ␣-subunit and affect its function (39,40). It is an open question whether the extramembrane domains of all other FXYD proteins interact with the same or different domains of the ␣-subunit (see Ref. 41). Both FXYD2 and FXYD4 have also been shown to be in proximity to the ␤-subunit (38,39).
The present work makes use of the P. pastoris expression system for an initial characterization of functional interactions between the Na ϩ ,K ϩ -ATPase and FXYD1 (PLM). Recently, we have described expression of Na ϩ ,K ϩ -ATPase (porcine ␣ 1 ␤ 1 subunits) in the methanotrophic yeast P. pastoris (42) and purification of functional detergent-soluble ␣/␤ subunit complexes (43). This system offers an opportunity to look at interactions of FXYD proteins with the purified ␣/␤ complexes in vitro and, in principle, could allow extensive studies of functional and structural interactions of FXYD proteins under well controlled conditions. Here, we describe functional interactions detected either by co-expression of PLM with ␣/␤ subunits, or after in vitro reconstitution of separately expressed ␣/␤ subunits and PLM to form ␣/␤/PLM complexes. The results show some interesting differences from those observed in Xenopus oocytes (26). Therefore, for comparison with the P. pastoris system, we have also looked at effects of PLM expressed in HeLa cells, which have been used extensively to examine functional and structural interactions of FXYD proteins, particularly FXYD2 (␥) and FXYD4 (CHIF) (7, 38, 40, 44 -46). A preliminary account of this work has appeared in Ref. 47.
Human PLM (GenBank TM accession H23593) was subcloned either into the EcoRI site of pHIL-D2, or into the EcoRI/NotI site of pGAPZa (Invitrogen). In the pGAPZ vector PLM is constitutively expressed in P. pastoris under the GAPDH promoter.
Yeast Transformation, Selection, and Growth-SMD1165, a protease-deficient strain (his4, prb1) of P. pastoris was transformed with 10 g of NotI-linearized pHIL-D2(␣/His 10 -␤) construct. Preparation of spheroplasts and selection for Hisϩ, Mut s transformants were done as described (42,43). The selected Mut s clone showing the highest expression level of the enzyme was further transformed by electroporation with AvrII-linearized pGAPZ vector containing human PLM. Clone selection was done by Zeocin resistance and by DNA Dot Blot screening. The resulting positive clones showed methanol-inducible expression of ␣/His 10 -␤ and constitutive expression of PLM in the same cell (co-expression of ␣/His 10 -␤ with PLM). Independently, linear cDNA of PLM in pHIL-D2 (without ␣ and His 10 -␤ cDNA) was used to transform spheroplasts, similarly to the pHIL-D2(␣/His 10 -␤) vector. This clone was used to express PLM alone and reconstitute it in vitro with the ␣/His 10 -␤ complex.
Standard large scale 3-liter cultures were grown in Bellco Spinner Flasks TM with air supplement, magnetic stirring, and temperature maintained at 25°C. Growth media were prepared from YNB medium (Difco) as described (42).
DNA Dot Blot Screen-Dot blot analysis has been used to scan for maximal copy number of integrated PLM, as described (42). A radioactive probe was synthesized from PLM cDNA using a random prime labeling kit (rediprime, Amersham Biosciences, cat. no. RPN 1633).
P. pastoris Membrane Preparations-Cells were broken with glass beads, and membranes were prepared as described (43), including a step involving treatment with 2 M urea. Membranes were stored at Ϫ80°C in MOPS-Tris, 10 mM, pH 7.2; EDTA, 1 mM; glycerol, 25% with protease inhibitors. Roughly, 1 g of membrane protein was obtained per 3-liter culture. Renal Na ϩ ,K ϩ -ATPase was prepared as described (48).
Purification of Recombinant Na ϩ ,K ϩ -ATPase-Purification of ␣/ His 10 -␤ co-expressed with PLM was done as described before (43). Synthetic DOPS (cat. no. 830035) was obtained from Avanti, and stored as a chloroform solution.
In later experiments, purification was done using Co 2ϩ -chelate chromatography (BD-Talon Biosciences), which will be described fully in a forthcoming article. 4 Briefly, the DDM-solubilized membranes were incubated with BD-Talon beads in the presence of 50 M EDTA for the same period of time as with Ni 2ϩ -NTA beads and with the same beads: supernatant ratio. The beads were washed twice with NaCl, 100 mM; Tris-HCl, 20 mM, pH 7.4; DDM, 0.2 mg/ml; DOPS 0.05 mg/ml; glycerol, 10%, and imidazole, 10 mM, and eluted with imidazole, 150 mM; NaCl, 100 mM; Tris-HCl, 20 mM pH 7.4, DDM, 0.2 mg/ml; DOPS, 0.05 mg/ml; glycerol, 40%. For preparations to be used for Na ϩ ,K ϩ ATPase assay, washing and protein elution were done in the presence of choline chloride, 100 mM instead of NaCl.
SDS-PAGE, Western Blots, and Immunoprecipitation-2-5 g of recombinant enzyme and 40 g of yeast membranes were separated on 7.5%, or 10% polyacrylamide SDS-Tricine gels (49). Gels were stained with Coomassie, scanned with an imaging densitometer (GS-690, Bio-Rad), and analyzed using the Multi-analyst software (Bio-Rad). Immunoblots were blotted with anti-KETYY antibody that recognizes the C terminus of the ␣-subunit, or with anti-␤ antibodies raised against the extracellular domain of the ␤-subunit (50). Two antibodies were used to recognize PLM: anti-PLM C terminus raised against the C-terminal sequence CRSSIRRLSTRRR (26), and anti-CP-68, kindly provided by Prof. Donald Bers, which recognizes the phosphorylated Ser 68 of the cytosolic domain of PLM (51).
Immunoprecipitation of ␣, His 10 -␤, and PLM subunits from the yeast membranes dissolved in C 12 E 10 was done essentially as described (45) using a monoclonal antibody directed against the N terminus of the ␣-subunit (6H).
Biochemical Assays-Na ϩ ,K ϩ -ATPase assays were done as described (42,43). Na ϩ ,K ϩ -ATPase assays of membranes (P. pastoris or HeLa cell) were carried out at 37°C, and of purified complexes at 25°C, both in a final volume of 100 l, using [␥-32 P]ATP. P. pastoris membranes were assayed for Na ϩ ,K ϩ -ATPase activity at 0.5-0.8 mM ATP, and HeLa cells membranes at 1 mM ATP, with varying Na ϩ concentrations in the presence of 100 mM KCl. Assays using membrane preparations were done with or without 5 mM ouabain, and the Na ϩ ,K ϩ -ATPase activity was calculated from the ouabain inhibitable fraction of ATP hydrolysis.
For purified recombinant enzyme, 1 l (Ϸ0.2-0.4 g of protein) was added to 100 l of the reaction medium. DDM and lipid concentrations in the reaction medium were the same as in the elution buffer. The Na ϩ ,K ϩ -ATPase activity of purified complexes was measured at 25°C and 100 M ATP in the presence of 100 mM KCl and different Na ϩ concentrations (0 -100 mM). The ionic strength was maintained constant at 250 mM with choline chloride. Na ϩ -ATPase activity was performed at 20 nM ATP in medium containing 250 mM NaCl plus choline chloride.
Rates of ATP hydrolysis at different Na ϩ activities were fitted to the following form of the Hill equation: v ϭ V max ϫ [Na] nH /([Na] nH ϩ K 0.5 nH ) using Kaleidagraph software (Synergy Software). In this equation n H is the Hill coefficient and K 0.5 is the concentration of Na ϩ ions required for half-maximal stimulation of ATP hydrolysis. For each curve the ratio of v/V max was then calculated and, where available, values of v/V max for replicate experiments were averaged for each Na ϩ concentration. Then the best-fit average parameters K 0.5 Ϯ S.E. and n H Ϯ S.E. were recalculated.
Ouabain binding to membranes was assayed using [ 3 H]ouabain, and phosphoenzyme (EP) levels were measured as ATP-dependent phosphorylation, essentially as described (52).
PKA Phosphorylation and Dephosphorylation-50 g of P. pastoris membranes expressing PLM, or 5 g of HeLa cell membranes, or 1 g of purified ␣/␤/PLM complexes were treated with either 20 units of alkaline phosphatase (AP) for 2 h at 37°C or with 2500 units of the catalytic subunit of PKA in the presence of 10 mM MgCl 2 and 1 mM ATP, for 2 h at 30°C. Recombinant PKA (2,500,000 units/ml, cat. no. P6000L), and calf intestinal alkaline phosphatase (CIP) (10,000 units/ml, cat. no. M0290S) were obtained from Bio Labs.
Expression of PLM and PLM/CHIF Chimera in HeLa Cells-CHIF/ PLM chimera were constructed by exchanging the extracellular (Met 1 -Gln 38 ), transmembrane (Leu 39 -Leu 57 ), and cytoplasmic (Ser 58 -Thr 87 ) domains of rat CHIF with the corresponding PLM sequences. Constructs were prepared by PCR using overlapping primers and verified by sequencing. The coding regions of CHIF, PLM, and CHIF/PLM chimera were subcloned into the BamHI/BstXI site of the mammalian expression vector pIRES-hyg (Clontech). This vector has an internal initiation site which enables translation of the cloned cDNA and the hygromycin resistance gene from a single RNA species. Thus, a better correlation is achieved between hygromycin resistance and high levels of the desired protein.
HeLa cells overexpressing the rat ␣ 1 -subunit of Na ϩ -K ϩ -ATPase (HeLa-␣ 1 cells, kindly provided by Dr. J. B. Lingrel, University of Cincinnati College of Medicine) were transfected with the above cDNAs using Polyfect (Qiagen) according to the manufacturer's instructions. Colonies expressing the FXYD constructs were selected in 400 g/ml hygromycin B, and tested by Western blotting using antibodies directed to the C termini of CHIF and PLM (26,45).
Isolation of Crude Plasma Membranes from HeLa Cells-Five confluent 150-cm 2 tissue culture plates were washed twice with phosphatebuffered saline, and cells were scraped with a rubber policeman. The cells were lysed in a 20-min incubation at 4°C in 10 mM Tris plus 1 mM MgCl 2 at 4°C for 20 min, and the lysate was subjected to 20 strokes of a Teflon-glass hand homogenizer. Intact cells, debris, and nuclei were removed in a 3-min centrifugation at 3,000 ϫ g. Membranes were then pelleted at 40,000 ϫ g for 50 min and subsequently washed twice in 1 mM EDTA (pH 7.4).

RESULTS
P. pastoris cells expressing ␣and His 10 -␤-subunits, which we have described recently (43), were transfected with the pGAPZ vector containing human PLM (FXYD1), under control of the GAPDH promoter. Selection of transformed yeast is based on the dominant selectable marker, Zeocin, and transformants should constitutively express the recombinant PLM. Recombinant clones were screened by dot blot analysis for their content of PLM cDNA, as described previously (42). Three colonies with the highest signals, marked in Fig. 1A with asterisks, were then grown in small volumes, membranes prepared, and Western blots run to detect PLM and the ␣-subunit. Fig. 1B shows that all three clones expressed easily detectable amounts of PLM together with the ␣-subunit. Although the molar ratio of PLM:␣ expressed in the yeast membranes cannot be estimated from the immunoblots, the ratio appears to be quite similar to that found for PLM:␣ in native rat sarcolemma vesicle membranes (lane marked SLV). Fig. 1C shows results of an immunoprecipitation experiment. The membranes were solubilized with C 12 E 10 under conditions known to maintain ␣/␤/FXYD complexes intact and optimize co-immunoprecipitation of FXYD proteins with ␣and ␤-subunits (45). Immunoprecipitation was done with a monoclonal anti-␣ antibody (6H), and the blots were probed with anti-KETYY (anti-␣), anti-␤, and anti-PLM. The solubilized membrane protein (S) contains ␣, the two characteristic bands of the ␤-subunit, and PLM, expressed in this preparation as two bands. The immunoprecipitated protein (Co-IP) contains the ␣-subunit, the ␤-subunit, and the upper band of the PLM, demonstrating that the ␣, ␤, and the upper band of PLM interact in a detergent-soluble complex, and presumably also in the yeast cell membranes. The lower band of PLM, which was not seen in all preparations, could be a proteolytic fragment of PLM, and does not appear to interact with ␣and ␤-subunits.
For most experiments in this article, purification of n-dodecyl-␤maltoside-soluble ␣/His 10 -␤ and ␣/His 10 -␤/PLM complexes was done using Ni 2ϩ -NTA beads chromatography, as described in our recent publication (43) and under "Materials and Methods." The complex obtained is 30 -50% pure, and an additional size-exclusion HPLC step is able to raise purity to 70 -80% (43), although HPLC was not done in the present work. In later experiments of this series, we utilized an improved purification procedure employing Co 2ϩ -chelate chromatography, which provides protein of up to 90% purity even without an HPLC step. 4 As seen in Fig. 2, after elution of the protein from Ni 2ϩ -NTA beads with 250 mM imidazole, the ␣and two bands of ␤-subunits are the major proteins seen with Coomassie staining. With the amounts of protein applied to the gel in Fig. 2, PLM was not seen with Coomassie stain, but with higher amounts of applied protein the PLM could also be seen as a lightly stained band (not shown). However, the immunoblot on the right shows clearly that PLM was indeed eluted off the beads together with the ␣and ␤-subunits, indicative of the existence of the ␣/␤/PLM complex. Judging by the intensity of the chemiluminescent signals, the ratio of ␣:PLM subunit in the partially purified complex is somewhat higher than in the membranes (mem) from which the complex was prepared (or, conversely, the ratio of PLM:␣ is lower). This could indicate either that in the yeast membranes there is a molar excess of PLM compared with ␣, so that only a fraction of PLM interacts with the ␣/␤ subunits, or that a fraction of the detergent-soluble ␣/␤/PLM complex dissociates during the purification procedure. The molar ratio of PLM to ␣-subunit eluted in the complex is considered again below (see Fig. 4). Fig. 3 shows results of measurements of Na ϩ ,K ϩ -ATPase activity at 25°C, without or with co-expressed PLM, at different Na ϩ concentrations and a fixed concentration of 100 mM K ϩ ion. The curves represent best fits to the Hill equation and the fitted parameters (K 0.5 , n H , and V max ) are presented in Table 1 (row 2). Two features were observed in this type of experiment. First, the V max was higher, by 20 -30%, in the presence of PLM (upper). Second, as seen most easily in re-plots of the ratio v/V max versus Na ϩ concentration (lower), the K 0.5 for Na ϩ ions was reduced modestly but distinctly in the presence of PLM. Table 2 presents values of the turnover number of the Na ϩ ,K ϩ -ATPase without and with PLM, calculated from parallel measurements of Na ϩ ,K ϩ -ATPase activity at 25°C and covalent phosphoenzyme levels, used as an estimate of the enzyme site concentration. Evidently, the phosphoenzyme level was higher with PLM than without PLM, as found for Na ϩ ,K ϩ -ATPase activity, and there was no significant effect of PLM on the turnover numbers at 25°C. The observed effect of PLM on the V max of Na ϩ ,K ϩ -ATPase can therefore be attributed to stabilization of the eluted protein against inactivation. The K m for ATP was also determined at 25°C, without or with co-expressed PLM. PLM did not affect the K m for ATP (control, K m 5.87 Ϯ 1.41 M, average of three experiments; with PLM, K m 5.81 Ϯ 0.95 M, average of two experiments). In other experiments, no significant differences were found for K i values for inhibition of Na ϩ ,K ϩ -ATPase activity at 25°C by vanadate and oubain (K i vanadate-control: 0.081 Ϯ 0.05: ϩPLM, 0.091 Ϯ 0.05 M; K i ouabain-control, 5.79 Ϯ 0.42; ϩPLM, 7.91 Ϯ 1.0 M).
Although the effect of PLM on the K 0.5 for Na ϩ ions was observed several times, the magnitude of the effect was somewhat variable. A possible explanation is that the expression levels of ␣/␤ subunits and PLM and thus molar ratios of the PLM:␣-subunit vary in different yeast cultures. Alternatively, there could be a variable degree of dissociation of PLM from the ␣/␤/PLM complex during the preparation.
As a test of the hypothesis, and a way around the problem just discussed, we have developed a procedure for in vitro reconstitution of the ␣/␤/PLM complex. PLM was expressed in P. pastoris separately from the ␣and ␤-subunits and a clone expressing maximal amounts of PLM was selected for further study. Reconstitution depends on spontaneous and specific association of uncomplexed PLM with the ␣/␤ subunits bound on the beads and co-elution of a ␣/␤/PLM complex. This technique should allow experimental control of the relative amounts of ␣/␤ subunits and PLM exposed to each other in vitro, and thus maximization of the ratio of PLM:␣/␤ subunits in the eluted complex (see Fig. 4). Fig. 4A shows an immunoblot of PLM in membranes of the clone co-  expressing PLM with ␣/␤ subunits, or the clone expressing PLM alone, which appears as two bands, and in significantly higher amounts. For the reconstitution experiment in Fig. 4B membranes from the clone expressing PLM alone were dissolved in DDM, protein was concentrated by ultrafiltration, and the soluble membrane protein was incubated overnight with Ni 2ϩ -NTA beads already bound with ␣/His 10 -␤    subunits. The amount of the soluble membrane protein was varied between 1-, 2-, or 5-fold the amount of membrane protein used for binding ␣/␤ subunits to the beads. The beads were then washed twice with 60 mM imidazole to remove contaminant proteins and, finally, the ␣/His 10 -␤/PLM complex was eluted with 250 mM imidazole. The blot in Fig. 4B shows that PLM was indeed eluted off the beads together with the ␣-subunit, and the amount of eluted PLM was similar whether 1-, 2-, or 5-fold of the soluble membrane protein was used for the reconstitution. The blot of the PLM remaining in the supernatant after reconstitution shows clearly that a significant fraction of PLM remained unbound whether 1-, 2-, or 5-fold of the soluble membrane protein was used. In short, the experiment shows that the lowest amount of total PLM applied to the beads represents a large excess and suffices for optimal reconstitution. Usually a 2-fold excess of membrane protein containing PLM was used for subsequent experiments. Because the PLM is only a minor component of the added membrane protein, one obvious conclusion from the reconstitution is that PLM must interact rather specifically with the ␣/␤ subunits attached to the beads. Activation of Na ϩ ,K ϩ -ATPase by Na ϩ ions was then compared for the reconstituted and co-expressed ␣/␤/PLM complexes and the ␣/␤ complex without PLM, see Fig. 5 and Table 1, row three. The V max was elevated for both co-expressed and reconstituted complexes compared with the ␣/␤ complex. In this particular experiment, the effect of PLM on K 0.5 for Na ϩ ions for the co-expressed complex was small but, for the reconstituted complex, a distinctly lower K 0.5 for Na ϩ ions was observed. Row 4 of Table  1 presents best-fit parameters obtained from averaged data from several reconstitution experiments and shows that the difference in K 0.5 (control, 18.045 Ϯ 1.27 versus ϩPLM, 12.27 Ϯ 0.82) is highly significant (p ϭ 0.0012). The Hill coefficient n H with PLM also appears to be slightly lower than that for the control. Because the reconstitution procedure is not dependent on variations in the expression levels of ␣/␤ subunits and PLM:␣ ratio, when the subunits are expressed together, it has become the method of choice for looking at effects of the PLM. Fig. 6 demonstrates the effect of PLM on the slow Na ϩ -ATPase activity of the protein observed in the absence of K ϩ ions (53,54). By looking at the Na ϩ dependence of Na ϩ -ATPase activity a relatively direct measure of effects of PLM on interaction of Na with the enzyme is obtained and, for reasons given in the "Discussion," the activity was measured in a reaction medium of high ionic strength (250 mM choline chloride plus NaCl) and at a very low concentration of ATP (20 nM). To visualize the effect of PLM on the K 0.5 for activation by Na ϩ ions, the data were plotted as v/V max versus Na ϩ concentration, averaged for five (control) or six (ϩPLM) experiments. Evidently, under these conditions, the K 0.5 of Na ϩ ions for the ␣/␤/PLM complex was reduced more than 3-fold lower compared with the ␣/␤ control. The difference in K 0.5 values (control 3.12 Ϯ 0.49 versus ϩPLM 0.92 Ϯ 0.15) is highly significant ( p ϭ 0.0009). This finding has an interesting mechanistic implication and is significant, also, because the magnitude of the effect is much greater than that observed for activation of Na ϩ ,K ϩ -ATPase activity by Na ϩ ions (see "Discussion").
It is striking that, whereas the present results show that PLM lowers K 0.5 of Na ϩ ions for activation of Na ϩ ,K ϩ -ATPase activity in the puri- Interaction of FXYD1 and Na ؉ ,K ؉ -ATPase Expressed in P. pastoris fied complexes isolated from P. pastoris (Figs. 3, 5, and 6), in Xenopus oocytes PLM has the opposite effect, and raises the K 0.5 of cytoplasmic Na ϩ ions for activation of the active pump current (26). One possible explanation of the difference is that interactions of PLM with ␣/␤ subunits in P. pastoris are non-native or are not fully native. However, the following data makes this quite unlikely. One could imagine, for example, that PLM-␣/␤ interactions in purified complexes, exposed to detergent and phospholipids, are altered compared with native PLM-␣/␤ interactions in intact membranes or cells. Accordingly, activation of Na ϩ ,K ϩ -ATPase activity by Na ϩ ions was looked at in crude membranes prepared from P. pastoris expressing either ␣/␤ or ␣/␤ plus PLM. Fig. 7 presents the data as a plot of v/V max averaged for three experiments and shows that PLM reduces K 0.5 by comparison with the control (24.8 Ϯ 1.08 versus 32.1 Ϯ 0.6, p ϭ 0.019) ,similar to the result with purified detergent-soluble ␣/His 10 -␤ complexes prepared from those membranes. The somewhat higher K 0.5 values for Na ϩ , compared with those in Table 1, could be related to the fact that Na ϩ ,K ϩ -ATPase activity of these membranes was measured at 37°C and at 1 mM ATP in order to optimize the ouabain-inhibitable fraction of ATP hydrolysis. However, this does not alter the conclusion of the experiment. Another possibility we have considered, and can now exclude, is that a 2 M urea treatment during membrane preparations, described in Ref. 43, partially denatures the PLM and precludes a fully native interaction with the ␣or ␤-subunits. An essentially identical effect of PLM on the K 0.5 of Na ϩ for activating Na ϩ ,K ϩ -ATPase activity was observed in an experiment using urea-treated or untreated membranes expressing PLM for purification of the ␣/␤/PLM complex (K 0.5 of Na ϩ for control is 17.76 Ϯ 0.93 mM, plus PLM from untreated membranes K 0.5 ϭ 12.66 Ϯ 1.21 mM, plus PLM from urea-treated membranes K 0.5 ϭ 11.52 Ϯ 1.42 mM). Finally, since we are looking at an interaction of human PLM with porcine ␣/␤ subunits one could also hypothesize that some interactions are missing because of the species difference. However, recent work has shown essentially identical effects on the K 0.5 for Na ϩ ions of human PLM interacting with human ␣ 1 in ␣ 1 ␤ 1 complexes purified from P. pastoris. 4 As a second line of inquiry, one can ask if the structural and functional interactions between PLM and ␣/␤ subunits are specific for the cell type used for expression. Previously, we have used the HeLa cell expression system extensively for analyzing functional effects of FXYD proteins and chimeric proteins (45,46). Fig. 8 presents results from a series of experiments which characterize Na ϩ activation of Na ϩ ,K ϩ -ATPase using HeLa cells expressing the rat ␣1-subunit and either rat PLM, rat CHIF, or a chimeric protein in which the transmembrane segment of CHIF was replaced by that of PLM (termed CPC). The activity was measured at 37°C with a fixed K ϩ concentration of 100 mM, ATP, 1 mM, and varied Na ϩ concentrations. The individual Na ϩ activation curves were analyzed by the Hill equation to obtain V max , K 0.5 , and n H 2. 5 The values of v/V max at each Na concentration for individual experiments FIGURE 7. Effect of PLM on activation of Na ؉ ,K ؉ -ATPase activity by Na ؉ ions in P. pastoris membranes. The data show the Na ϩ ,K ϩ -ATPase activity at 37°C for membranes expressing either ␣and His 10 -␤-subunits or ␣, His 10 -␤, and PLM. Data were analyzed as in Fig. 2. The values of Na,K-ATPase activity in different experiments were in the range 0.07-0.18 mol/min/mg protein.  were calculated, and then the average values of v/V max for all experiments for EV, PLM, CHIF, and CPC, respectively. The best-fit average K 0.5 and n H values were obtained, as described under "Materials and Methods. " Fig. 8, presents a bar graph for these average best-fit K 0.5 values (ϮS.E.) for empty vector (EV), compared with PLM, CHIF, and the CPC chimera. The actual values of K 0.5 are presented in the inserted table, together with number of experiments (n) and the p values for the differences between EV and PLM, CHIF, and the CPC chimera, respectively. The differences are all highly significant. Values of the Hill coefficient n H were not different for the different conditions and have been omitted for simplicity. Evidently, PLM raised the K 0.5 for activation by Na ϩ ions, similar to the effect described for Xenopus oocytes (26) and different to that in the P. pastoris system. As reported previously, CHIF reduced the K 0.5 (45). By contrast, however, the CPC chimera showed a significantly reduced K 0.5 for Na ϩ , similar to the effect of CHIF, and to the effect of full-length PLM in the P. pastoris. In short, the experiment demonstrates that, the functional effect of PLM depends on the cell type, and the reduced K 0.5 for Na ϩ seems to be determined by the transmembrane segment of PLM.
PLM contains consensus sequences for both PKA-and PKC-dependent phosphorylation near the C terminus, and it is conceivable that the different functional effect of PLM in the different cell types is related to its phosphorylation state. Figs. 9 and 10 present data that addresses the question of whether PLM is endogenously phosphorylated at the PKA site (S68) when expressed in P. pastoris and HeLa cells. The experiments utilize two antibodies. One antibody CP68, described in Ref. 51, has been shown previously to recognize only the phosphorylated form of the PLM. The other antibody, anti-PLM raised against the C-terminal sequence CRSSIRRLSTRRR, is shown here to recognize essentially only the unphosphorylated form of PLM. Fig. 9, A and B show that the PLM expressed in crude P. pastoris membranes or in the partially purified reconstituted ␣/␤/PLM complex gives a strong response to CP68. Preincubation of the membranes or detergent-soluble ␣/␤/PLM complex with AP strongly reduces the response to CP68, and incubation with alkaline phosphatase in the presence of an inhibitor (APϩIn) prevents the reduction of the response to CP68. Qualitatively these results indicate that phosphorylated PLM is detectable in the yeast expression system. Fig. 9, C and D go a step further and provide an approximate estimate of the degree of phosphorylation of PLM at the PKA site. In Fig.  9C the detergent-soluble ␣/␤/PLM complex bound to the Ni 2ϩ -NTA beads was incubated either with alkaline phosphatase or with PKA plus ATP/Mg. The quantities of alkaline phosphatase and PKA and incubation times were assumed, somewhat arbitrarily, to suffice for full dephosphorylation or phosphorylation respectively. As seen in Fig. 9C the alkaline phosphatase indeed abolished the response to CP68, whereas incubation with PKA increased the response to CP68. The blot in Fig. 9D was probed with anti-PLM (C terminus) on samples incubated either with alkaline phosphatase or with PKA, and also after incubation with alkaline phosphatase subsequent to PKA. The result is that after incubation with alkaline phosphatase the response to anti-PLM was amplified, after incubation with PKA the response to anti-PLM was almost completely abolished, and after alkaline phosphatase treatment subsequent to PKA the response to anti-PLM was restored. Taken together the results obtained with CP68 and anti-PLM make it clear that alkaline phosphatase and PKA dephosphorylated or phosphorylated the PLM, respectively, essentially quantitatively in these conditions. By comparing the response to CP68, before and after PKA, or that to anti-PLM, before and after alkaline phosphatase, it is evident that a substantial fraction of the PLM in the P. pastoris cells, perhaps as much as half, was naturally phosphorylated at the PKA site. In other preparations the fraction of phosphorylated PLM appeared to be even higher than in Fig.  9, C and D, although this was not checked in all experiments. Fig. 10 examined phosphorylation of PLM at the PKA site in HeLa cell membranes. The experiment utilizes CP68 to detect phosphorylated PLM in the membranes, which were either not treated or were incubated with alkaline phosphatase or PKA, respectively. The data show that rat PLM expressed in HeLa cell membranes is hardly phosphorylated, if at all, at the PKA site.
An appealing hypothesis is that the opposite effects of PLM on the K 0.5 for Na ϩ in P. pastoris and HeLa cells are caused by the different degrees of PLM phosphorylation in these preparations. We attempted to test this hypothesis directly by assaying Na ϩ , K ϩ -ATPase activity of ␣/␤ complexes purified with PLM that was either phosphorylated or dephosphorylated in vitro. However, the protocols involving additional manipulations and long incubations at 0°C of ␣/␤/PLM complexes bound to the beads with alkaline phosphatase or PKA, led to variable degrees of inactivation of the enzyme, variable efficiencies of phosphorylation and dephosphorylation of PLM, and inconclusive results. Thus, direct assessment of the role of PLM phosphorylation on K 0.5 for Na ϩ awaits future experiments.

DISCUSSION
The work in this article demonstrates the isolation and initial functional characterization of partially purified complexes consisting of porcine ␣/␤ subunits of Na ϩ ,K ϩ -ATPase and human FXYD1 (PLM) expressed in P. pastoris, and also a comparison with functional effects of PLM expressed in HeLa cells.  One effect of the PLM is stabilization of the ␣/␤ complex. This is seen as a 20 -30% increase in V max of Na ϩ ,K ϩ -ATPase and phosphoenzyme level with no change in the turnover number (Table 2). Because this feature is observed even in reconstitution experiments, it shows that PLM protects against partial inactivation occurring during the purification procedure.
A possible mechanism of action of PLM in the P. pastoris system is suggested by a comparison of its effects on Na ϩ ,K ϩ -ATPase (Figs. 3 and 5) and Na ϩ -ATPase activity (Fig. 6). From basic enzyme kinetic principles a PLM-induced reduction of K 0.5 of Na ϩ ions for activating Na ϩ ,K ϩ -ATPase activity, without simultaneous effects on the Na ϩ ,K ϩ -ATPase turnover rate or K m for ATP, is compatible with an increased intrinsic binding affinity of Na ϩ ions. In this respect the observed effect of PLM on activation of Na ϩ -ATPase activity by Na ϩ ions is interesting. In open membrane preparations, with access of the Na ϩ ions to both cytoplasmic and extracellular sites, the Na ϩ -ATPase activity shows a rather complex dependence on Na ϩ ion concentration (55). Activation of phosphorylation at 0 -5 mM Na ϩ at normal cytoplasmic sites is followed by partial inhibition at 5-10 mM Na ϩ and then further activation at 10 -150 mM Na ϩ , because of Na ϩ binding at extracellular sites and effects on hydrolysis of the E 2 -P intermediate (53,54,56). To avoid these complexities the Na ϩ -ATPase activity of the detergent soluble complexes was measured at a high ionic strength, which strongly stabilizes the E 1 conformation, and at a very low ATP concentration (20 nM), which should make the phosphorylation of E 1 rate-limiting. In this situation one could expect the Na ϩ dependence of the Na ϩ -ATPase to reflect primarily binding of Na ϩ ions to the cytoplasmic sites. The observed Na ϩ activation curves are indeed quite simple, reaching saturation at about 5 mM Na ϩ . The relatively large difference in the K 0.5 for Na ϩ without and with PLM, greater than 3-fold under these conditions, is consistent with an increased binding affinity of Na ϩ ions induced by PLM. By contrast, the more modest effect observed in the Na ϩ ,K ϩ -ATPase reaction conditions is expected, because the K 0.5 for Na ϩ ions is a less direct measurement, reflecting not only Na ϩ binding but also competition with K ϩ ions and rate constants of other steps in the catalytic cycle.
The effect of PLM to reduce K 0.5 of Na ϩ for activating Na ϩ ,K ϩ -ATPase in the P. pastoris system is a striking difference from that observed previously in Xenopus oocytes, in which PLM significantly raises K 0.5 for Na ϩ (26) and, as shown here, also in the HeLa cell system. By contrast, for the chimeric protein CPC, a reduced K 0.5 for Na ϩ was observed, compared with the control, similar to the effect of full-length PLM in the P. pastoris (Fig. 8). Thus, the comparison with HeLa cell experiments shows that the observed functional effect depends on the cell type. An economical explanation of the opposite effects of full-length PLM and the CPC chimera on the K 0.5 of Na ϩ in the HeLa cell membrane is the existence of at least two types of structural and functional interaction of the full-length PLM. For example, the interaction of the transmembrane segment of PLM with M2, M6, and M9 of the ␣-subunit could raise binding affinity for Na ϩ ions in the Na ϩ occlusion sites, thus reducing K 0.5 of Na ϩ for activation of ATPase activity, while interactions of the extramembrane sectors of PLM could stabilizes E 2 conformations, thus indirectly raising the K 0.5 of Na ϩ . The overall effect on the K 0.5 of Na ϩ of full-length PLM would reflect the balance of these interactions, and in the CPC chimera only the transmembrane interaction would persist. Of course, this explanation is both hypothetical and is not exclusive, but there is a convincing precedent for multiple structural and functional interactions of at least one other FXYD proteins, the ␥-subunit (FXYD2) see Ref. 57.
A second striking difference between the PLM expressed in P. pastoris and HeLa cells is that the PLM in P. pastoris is significantly phosphorylated at the PKA site (Ser 68 ) (Fig. 9) whereas that in HeLa cells is hardly phosphorylated at all (Fig. 10). Obviously, the opposite functional effects of PLM on K 0.5 for Na ϩ ions in P. pastoris and HeLa could be related to this different state of phosphorylation. In principle, one could envisage that PKA-dependent phosphorylation of Ser 68 disrupts the interaction between the cytoplasmic C terminus of PLM and ␣/␤ subunits, leaving intact the interaction between the transmembrane segments of PLM and ␣. This is similar in concept to one proposed for the structural effects of PKC-dependent phosphorylation of the PLM-like protein (PLMS) in shark rectal gland Na ϩ ,K ϩ -ATPase, (1,33,34).
How does the information in this article add to and fit in with the reported physiological effects of PLM and modulation of PLM-␣/␤ interactions by phosphorylation? The reduced Na ϩ affinity observed upon expression of PLM in HeLa cells (Fig. 8) is attributable to the non-phosphorylated form of PLM, shown in Fig. 10. The same may be true also for Xenopus oocytes (26). As argued recently in reviews (3, 57), a reduced Na ϩ affinity ensures that the pumping rate is not saturated with respect to cytoplasmic Na ϩ concentration, and may be required for regulation of pumping during basal electrical and mechanical activity in heart cells. During ␤-adrenergic activation the increased cardiac contractility is associated with an increased Na ϩ concentrations and a large transient rise in cytoplasmic Ca 2ϩ ion. The increase in cytoplasmic Na ϩ concentrations must be limited during ␤-adrenergic activity, and restored to the basal level after activation, by increased Na ϩ extrusion via the Na ϩ ,K ϩ -ATPase (58). By functional coupling with the 3Na ϩ / Ca 2ϩ exchange mechanism the Ca 2ϩ ion concentration then returns to the basal level. There are two relevant recent observations on intact ventricular myocytes. It has been shown that activation of PKA induced by forskolin (29) or isoproterenol, a ␤-adrenergic agonist, (30) is associated with phosphorylation of PLM at S68. In (29) the Na ϩ ,K ϩ -pump current was elevated and this was restricted specifically to the ␣1 and not the ␣2 isoform. In Ref. 30, it was found that in myocytes from wild-type mice the curve for activation of the pump current by Na ϩ is shifted to the left upon treatment with isoproterenol, while in myocytes obtained from PLM knock-out mice (see Ref. 27), no effect of isoproterenol was observed. This led to the suggestion that phosphorylation of PLM at Ser 68 raises the apparent affinity of Na ϩ by neutralizing the modulatory effect of non-phosphorylated PLM, which acts to reduce the apparent affinity for Na ϩ ions. The final result is activation of active Na ϩ and K ϩ transport during the ␤-adrenergic stimulation, accelerated restoration of Na ϩ and Ca 2ϩ gradients, and cardiac muscle relaxation. The present results are completely consistent with this scenario at the physiological level, but there is an important difference in interpretation at the level of PLM-␣/␤ interactions. As discussed above, the model best compatible with our finding is that phosphorylation at the PKA site abrogates one type of interaction, but leaves another interaction intact. The elevated Na affinity is the result of this interaction with phosphorylated PLM, and is not due merely to abrogation of the interaction with non-phosphorylated PLM. Although the native myocyte is obviously superior for defining the physiological role of the PLM, the different expression systems allow one to detect effects of PLM compared with the control without PLM, and thus better define the interactions with the ␣/␤ subunits.
In PLM knock-out mice the V max of the Na ϩ ,K ϩ -ATPase in isolated heart sarcolemma membranes was 50% lower than in the wild type, even allowing for lower expression levels of both ␣1 and ␣2 (27). The implication is that PLM increases the turnover rate of the Na ϩ ,K ϩ -ATPase. This is different from observations in Xenopus oocytes, mouse myo-cytes, and the P. pastoris system, which did not show effects of PLM on V max , but rather on the apparent Na ϩ affinity. Conceivably the PLM is in a different state in the wild-type mouse heart to that in the expression systems or isolated mouse myocytes. In this regard, it is important to remember that PLM has both PKA sites (at Ser 68 ) and a PKC site (Ser 63 and Ser 68 ) and is also the substrate of other kinases (21,22). The state of phosphorylation of PLM at Ser 63 in the Xenopus oocytes, or P. pastoris membranes or mouse myocyte membranes is unknown. There might be complex interactions between PLM phosphorylated at say Ser 63 and Ser 68 , and the ␣/␤ subunits, and thus multiple functional effects.
Perspective-Expression of ␣/␤ subunits and PLM in P. pastoris, combined with purification of ␣/␤ and reconstitution of ␣/␤/PLM complexes, provides a well controlled system for looking at interactions between ␣/␤ subunits and PLM. Detailed investigation of the mechanism of action of PLM, by direct measurement of cation occlusion or conformational changes detected with fluorescence probes should become possible, after upscaling of the amounts of the purified ␣/␤ and ␣/␤/PLM complexes. Upscaling is now being undertaken. 6 Eventually, structural studies may become feasible.