FXYD7, mapping of functional sites involved in endoplasmic reticulum export, association with and regulation of Na,K-ATPase.

The brain-specific FXYD7 is a member of the recently defined FXYD family that associates with the alpha1-beta1 Na,K-ATPase isozyme and induces an about 2-fold decrease in its apparent K+ affinity. By using the Xenopus oocyte as an expression system, we have investigated the role of conserved and FXYD7-specific amino acids in the cellular routing of FXYD7 and in its association with and regulation of Na,K-ATPase. In contrast to FXYD2 and FXYD4, the studies on FXYD7 show that the conserved FXYD motif in the extracytoplasmic domain is not involved in the efficient association of FXYD7 with Na,K-ATPase. On the other hand, the conserved Gly40 and Gly29, located on the same face of the transmembrane helix, were found to be implicated both in the association with and the regulation of Na,K-ATPase. Mutational analysis of FXYD7-specific regions revealed the presence of an ER export signal at the end of the cytoplasmic tail. Deletion of a C-terminal valine residue in FXYD7 significantly delayed and decreased its O-glycosylation processing and retarded the rate of its cell surface expression. This result indicates that the C-terminal valine residue is involved in the rapid and selective ER export of FXYD7, which could explain the observed post-translational association of FXYD7 with Na,K-ATPase. In conclusion, our study on FXYD7 provides new information on structural determinants of general importance for FXYD protein action. Moreover, FXYD7 is identified as a new member of proteins with a regulated ER export, which suggests that, among FXYD proteins, FXYD7 has a particular regulatory function in brain.

The Na,K-ATPase uses the hydrolysis of ATP to exchange Na ϩ and K ϩ across the plasma membrane and to create and maintain the transmembrane Na ϩ and K ϩ gradients. The role of the Na,K-ATPase is essential for basic cellular homeostasis as well as for specialized tissue functions. In the kidney, for instance, the Na,K-ATPase is exclusively expressed in the basolateral membrane and thus becomes the driving force for net transepithelial Na ϩ reabsorption, which is necessary for the maintenance of the extracellular volume and blood pressure. In the nervous system, the presence of Na,K-ATPase permits a rapid restoration of the resting ionic conditions after action potentials and is thus of primary importance for a correct neuronal excitability.
At the molecular level, the Na,K-ATPase consists of an ␣and a ␤-subunit. The catalytic ␣-subunit hydrolyzes ATP and transports the cations, and it also binds cardiac glycosides with high specificity. The ␤-subunit acts as a molecular chaperone, which plays a crucial role in the structural and functional maturation of the ␣-subunit (1). Several ␣ and ␤ isoforms (4␣ and 3␤) exist that may combine to form 12 possible ␣-␤ complexes with different transport and pharmacological properties (2,3). Na,K-ATPase isozymes show a tissue-specific expression and, thus, may cope with different physiological demands.
The regulation of Na,K-ATPase activity is tight and complex and involves short and long term mechanisms mediated by hormones and neurotransmitters (for review see Ref. 4). Recently, a new regulatory mechanism has emerged, which involves association of the Na,K-ATPase with small type I membrane proteins of the newly defined FXYD protein family (5). This family contains 7 members that share a signature sequence consisting of the FXYD motif and 3 other conserved amino acid residues (for review see Ref. 6). We recently characterized FXYD7 (7), which is exclusively expressed in brain, both in neurons and glial cells. FXYD7 is subjected to Oglycosylation, which appears to be important for protein stability. Expressed in Xenopus oocytes, FXYD7 interacts with ␣1-␤1, ␣2-␤1, and ␣3-␤1 isozymes, but not with ␣-␤2 Na,K-ATPase isozymes, whereas in the brain, FXYD7 can only be co-immunoprecipitated with an antibody against the ␣1-subunit and not with antibodies against the other ␣ isoforms. FXYD7 modulates Na,K-ATPase transport properties in a different way than FXYD2 (the ␥-subunit) (8,9,10), FXYD4 (CHIF) (8,11), or FXYD1 (phospholemman) (12). After co-expression in Xenopus oocytes, FXYD7 has no effect on the apparent Na ϩ affinity of Na,K-ATPase but decreases the apparent K ϩ affinity over a wide range of membrane potentials, though it only affects ␣1-␤1 and ␣2-␤1 isozymes and not ␣3-␤1 isozymes (7).
To better understand the regulatory effect on the Na,K-ATPase of FXYD proteins in general, and of FXYD7 in particular, we investigated, in the present study, the role of conserved and FXYD7-specific amino acids in FXYD7 properties, namely in its stability, its intracellular routing, its association with the Na,K-ATPase and in its functional effect on the apparent K ϩ affinity of Na,K-ATPase.

EXPERIMENTAL PROCEDURES
cDNAs-Mouse FXYD7 (7) served as a template for FXYD7 mutants (see Fig. 1). For truncated mutants (Nt-⌬15, Ct-⌬7, and ⌬V80), a series of single PCR reactions was carried out with appropriate primers. PCR products were then digested and subcloned between EcoRI and NotI restriction sites into the pSD5 vector. Point mutations were introduced by using the PCR method described by Nelson and Long (13). Nt-FLAG FXYD7 and Nt-FLAG ⌬V80 were constructed by PCR using a sense primer tailed with the appropriate sequence coding for the FLAG motif (DYKDDDDK) placed upstream of the first ATG and the appropriate antisense primer. The nucleotide sequences of all constructs were confirmed by dideoxy sequencing. cDNAs for rat ␣1and ␤1-subunits were kindly provided by J. Lingrel. All cDNAs were introduced into the pSD5 vector. cRNAs were prepared by in vitro translation (14).
Protein Expression in Xenopus Laevis Oocytes-Stage V-VI oocytes were obtained from Xenopus laevis as previously described (15). cRNAs coding for wild-type or mutant FXYD7 (2 ng/oocyte) were injected into oocytes in the presence or absence of cRNAs coding for the rat Na,K-ATPase ␣1 (10 ng/oocyte) and ␤1 (1 ng/oocyte) subunits as described in the figure legends. Under these conditions, wild-type, or mutant FXYD7 is expressed in excess of exogenous Na,K-ATPase and is not limiting for the association with Na,K-ATPase. To study protein expression and association, oocytes were incubated in modified Barth's solution (MBS) 1 in the presence of 0.8 -1 mCi/ml [ 35 S]methionine (Easy Tag Express 35 S protein labeling kit, PerkinElmer) at 19°C. Oocytes were subjected to a 6-h pulse and to 24 h and 48 h chase periods in MBS containing 10 mM cold methionine. After the pulse and chase periods, oocyte extracts were prepared as previously described (15) with a digitonin-containing buffer (0.5% w/v final concentration) and subjected to immunoprecipitations with FXYD7 antibodies (7) or Na,K-ATPase ␣-antibodies (16) under denaturing or non-denaturing conditions (15). Immunoprecipitated proteins were separated on SDS-polyacrylamide gels (5-13% gradients) or Tricine-polyacrylamide gels (16.5%) (prepared according to the manufacturer's instructions) and revealed by fluorography. Protein quantification was performed using a laser densitometer (KB Ultrascan 2202).
Electrophysiology-Electrophysiological measurements were performed 3 days after oocyte injection with rat ␣1-␤1 cRNAs alone or together with FXYD7 wild-type or mutant cRNAs by using the twoelectrode voltage clamp technique. Measurements of the apparent external K ϩ affinity at Ϫ50 mV were carried out as described (8) in the presence of 1 M ouabain, which inhibits the endogenous oocyte Na,K-ATPase but not the expressed ouabain-resistant rat Na,K-ATPase. The maximal Na,K-ATPase current and the apparent K ϩ affinity (K1 ⁄2 K ϩ ) measured in the presence of 90 mM external Na ϩ were obtained by fitting the Hill equation to the data using Hill coefficients of 1.6 (17).

125
I Anti-FLAG Binding on Intact Oocytes-Anti-FLAG M 2 IgG was iodinated using the iodo-beads iodination reagent (Pierce) and carrierfree [ 125 I]Na (Amersham Biosciences) according to the Pierce protocol. The iodinated antibody had an average specific activity of 5-20 ϫ 10 7 cpm/mol. The assay of binding to oocytes was performed 3, 6, and 48 h after injection. Oocytes were incubated on ice for 1 h in MBS containing 10% heat-inactivated calf serum. Binding was initiated by addition of 20 nM of iodinated antibody. After 1 h incubation on ice, the oocytes were washed 8 times with cold MBS containing 5% heat-inactivated calf serum and individually transferred into tubes for ␥ counting. Nonspecific binding was determined on non-injected oocytes and subtracted from the total binding determined on cRNA-injected oocytes.

Effects of Mutations in the N-terminal Extracellular Domain of FXYD7-The N-terminal extracellular domain of FXYD7
shows no homology to that of other FXYD proteins with the exception of the presence of the conserved FXYD motif (FFYD in FXYD7). To determine the role of conserved and FXYD7specific regions in the N terminus of FXYD7, we produced an AFAA mutant lacking the FXYD motif, a Nt-3T/A mutant lacking the 3 threonines, which are potential sites of O-glycosylation (7), and an N-terminally-truncated mutant (Nt-⌬15) (see Fig. 1). After expression in Xenopus oocytes, we tested the stability of these mutants, their processing, their association with the ␣1-␤1 Na,K-ATPase isozyme and their functional effect on the apparent K ϩ affinity of Na,K-ATPase.
As shown in Fig. 2, wild-type, metabolically labeled FXYD7 expressed in Xenopus oocytes without Na,K-ATPase was processed from a core protein of about 14 kDa, observed exclusively during a 6-h pulse period, to an O-glycosylated doublet of 18 -19 kDa, quite stably expressed during a 24-and 48-h chase period (Fig. 2, A, lanes 1-3, and B). The AFAA mutant was processed similar to the wild-type FXYD7 ( Fig. 2A, lanes 10 -12). As previously described (7), the Nt-3T/A mutant lacking the O-glycosylation sites Thr 3 , Thr 6 , and Thr 9 was degraded during a 48-h chase period (Fig. 2, A, lanes 4 -6, and C) indicating a role of O-glycosylation in the stability of FXYD7. Interestingly, truncation of 15 N-terminal amino acids of FXYD7 in a Nt-⌬15 mutant partially restored the stability of the FXYD7 protein during a 24-h chase period (Fig. 2, A, lanes 7-9, and C) suggesting that the 15 N-terminal amino acids may contain a degradation signal.
The ability of wild-type and mutant FXYD7 to associate with the ␣1-␤1 Na,K-ATPase isozyme was tested after co-expression in Xenopus oocytes and by immunoprecipitation under nondenaturing conditions with a FXYD7 antibody. Our results indicate that association of wild-type FXYD7 with ␣1-␤1 complexes is time-dependent and occurs post-translationally. Indeed, despite a similar expression level of the ␣-subunit after the pulse and the chase periods, as revealed by immunoprecipitation with an ␣ antibody (Fig. 3B, lanes 1-3), the amount of ␣-subunit co-immunoprecipitated with a FXYD7 antibody ( Fig. 3, A, lanes 1-3, and C) increased progressively between the 6-h pulse and the 48-h chase. Association of the AFAA mutant with the ␣-subunit proceeded similar to that of the wild-type FXYD7 (Fig. 3, A, lanes 10 -12 and C) indicating that the FXYD motif does not play a role in the efficient interaction of FXYD7 with the Na,K-ATPase. As previously shown (7), the Nt-3T/A mutant (Fig. 3A, lanes 4 -5 and C) associated with Na,K-ATPase at a level similar to wild-type FXYD7 and became stabilized, suggesting that lack of O-glycosylation does not prevent the post-translational association of FXYD7 with the Na,K-ATPase. In contrast, truncation of 15 N-terminal amino acids in the Nt-⌬15 mutant decreased significantly the amount of ␣-subunit co-immunoprecipitated with FXYD7 antibodies ( Fig. 3, A, lanes 7-9 and C) indicating that structural integrity of the N terminus is required for efficient association with Na,K-ATPase.
As assessed by electrophysiological means, oocytes expressing Na,K-ATPase alone or together with wild-type FXYD7, with the Nt-3T/A mutant, the Nt-⌬15 mutant, or the AFAA mutant showed similar Na,K-pump currents (Table I) indicating that these N-terminal FXYD mutants do not influence the expression of functional Na,K-ATPase at the cell surface. As previously described (7), we observed an about 50% decrease in the apparent affinity for K ϩ of the Na,K-ATPase associated with wild-type FXYD7 ( Fig 3D and Table I). A similar effect was produced by the Nt-3T/A and the AFAA mutant as well as by the less efficiently associated Nt-⌬15 mutant. This latter result can be explained by the assumption that the co-immu-noprecipitation experiment performed in the presence of digitonin does not reflect the actual proportion of complexes between Na,K-ATPase and the Nt-⌬15 and AFAA mutants in intact cells, but rather the detergent sensitivity for dissociation of these complexes, which is higher than that of complexes including wild-type FXYD7. Altogether, the results suggest that the N-terminal, extracellular domain, including the conserved FXYD motif and the sugar moiety, is not implicated in the functional effect of FXYD7 on the Na,K-ATPase. A and B, Xenopus oocytes were injected with wildtype or mutant FXYD7 cRNAs (2 ng) together with rat Na,K-ATPase ␣1 (10 ng) and ␤1 (1 ng) cRNAs. Oocytes were metabolically labeled, and digitonin extracts were prepared as described in the legend to Fig. 2. Non-denaturing immunoprecipitations were performed using an anti-FXYD7 antibody (A) or an anti-␣ antibody (B). Immunoprecipitates were separated on SDS-polyacrylamide gels. C, quantification of ␣-subunit co-immunoprecipitated with wild-type FXYD7 (squares), the Nt-3T/A (diamonds), the Nt-⌬15 mutant (circles), or the AFAA mutant (triangles) with a FXYD7 antibody. The amount of ␣-subunit co-immunoprecipitated was corrected for the expression level of the ␣-subunit shown in B. The amount of ␣-subunit co-immunoprecipitated with wild-type FXYD7 after a 6-h pulse was arbitrarily set to 1. Shown are means Ϯ S.E. of nine experiments for wild-type FXYD7, four experiments for the Nt-3T/A, three experiments for the Nt-⌬15 mutant, and five experiments for the AFAA mutant. D, three days after injection of rat ␣1 and ␤1 cRNAs in the presence or not of wild-type or mutant FXYD7, the external K ϩ activation constant (K1 ⁄2 K ϩ ) of Na,K-ATPase was determined in the presence of 90 mM external Na ϩ at a holding potential of Ϫ50 mV as described under "Experimental Procedures." Results are means of 8 -10 oocytes from 2-3 batches of oocytes.

Effects of Mutations in the Transmembrane Domain of FXYD7-
The transmembrane helix is the most conserved domain among FXYD proteins. With the exception of a conserved serine residue (Ser 46 in FXYD7), which marks the putative end of the transmembrane domain, invariant residues such as the 2 conserved glycine residues (Gly 29 and Gly 40 in FXYD7) and nearly invariant amino acids in FXYD proteins are clustered on one side of the transmembrane helix (5).
Alanine replacement of the conserved glycine residue Gly 29 or the conserved serine residue Ser 46 did not change the stability or the processing of FXYD7 expressed in oocytes (Fig. 4A, compare lanes 1-3 to lanes 4 -6 and lanes 16 -18).
Mutation of Gly 41 in FXYD2 (Gly 40 in FXYD7) into an arginine residue has recently been linked to cases of human autosomal dominant renal hypomagnesaemia (18). Since this mutation was shown to abolish the interaction of FXYD2 with the Na,K-ATPase (19), it has been speculated that magnesium wasting associated with this pathology may be indirectly caused by a loss of Na,K-ATPase regulation by FXYD2. To learn more about the role of this conserved glycine residue, we replaced Gly 40 in FXYD7 not only by alanine (G40A) but also by tryptophan, a bulky hydrophobic amino acid (G40W), or by arginine (G40R), a positively charged residue. All of these mutants showed stability and processing similar to wild-type FXYD7 when expressed alone in Xenopus oocytes (Fig. 4A,  compare lanes 1-3 with lanes 7-15).
Concerning the association with Na,K-ATPase, the S46A mutant associated as efficiently as wild type FXYD7 (Fig. 4B,  compare lanes 1-3 with lanes 16 -18). On the other hand, alanine replacement of the conserved glycines (G29A or G40A) significantly decreased the association efficiency of FXYD7. Indeed, despite a similar expression level of the ␣-subunit (Fig.  4C, lanes 1-9), the amount of ␣-subunit co-immunoprecipitated with a FXYD7 antibody was considerably lower throughout the pulse/chase periods in oocytes expressing the G40A (Fig. 4, B,  lanes 7-9 and D) or the G29A mutant (Fig. 4, B, lanes 4 -6 and D) than in oocytes expressing wild type FXYD7 (Fig. 4, B, lanes  1-3 and D). Interestingly, replacement of Gly 40 by tryptophan instead of alanine further decreased the association efficiency (Fig. 4B, lanes 13-15) and replacement by arginine completely abolished association with Na,K-ATPase (lanes 10 -12).
Analysis of the functional effect of these mutants on the Na,K-ATPase transport properties showed that the efficiently associated mutant S46A produced an effect on the K ϩ activation of the Na,K-ATPase similar to wild-type FXYD7 (Fig. 5 and Table I). On the other hand, the glycine mutants that were impeded in their association efficiency produced no (G29A) or a partial effect (G40A and G40W) on the apparent K ϩ affinity ( Fig. 5 and Table I). Interestingly, though the G40R mutant did not associate with the Na,K-ATPase (Fig. 4B, lanes 10 -12) and Digitonin extracts were prepared after the pulse and the chase periods and immunoprecipitation was performed in denaturing conditions using an anti-FXYD7 antibody. Immunoprecipitates were separated on Tricinepolyacrylamide gels. B and C, Xenopus oocytes were injected with wild-type or mutant FXYD7 cRNAs (2 ng) together with rat Na,K-ATPase ␣1 (10 ng) and ␤1 (1 ng) cRNAs. Oocytes were metabolically labeled and digitonin extracts were prepared as in Fig. 2. Non-denaturing immunoprecipitations were performed using an anti-FXYD7 antibody (B) or an anti-␣ antibody (C). D, quantification of ␣-subunit coimmunoprecipitated with wild-type FXYD7 (squares), the G29A (diamonds), or the G40A mutant (circles) with a FXYD7 antibody. The amount of ␣-subunit co-immunoprecipitated was corrected for the expression level of the ␣-subunit shown in C. The amount of ␣-subunit co-immunoprecipitated with wild-type FXYD7 after a 6-h pulse was arbitrarily set to 1. Shown are means Ϯ S.E. of nine experiments for wild-type FXYD7 and means of two experiments for the G29A and the G40A mutant. did not impede expression of stable Na,K-ATPase (Fig. 4C, lanes 10 -12), its presence led to premature death of some oocytes or to significantly reduced Na,K-pump currents in surviving oocytes. This excluded reliable measurements of the apparent K ϩ affinity. At present, we do not know the reasons for the deleterious effect on cell viability of the G40R mutant. Altogether, these results point to the importance of the transmembrane domain of FXYD7, and in particular of the conserved glycine residues, both in the association with the Na,K-ATPase and in the modulation of its transport properties.
Effects of Mutations in the C-terminal, Cytosolic Domain of FXYD7-Like the extracellular domain, the cytosolic domain of FXYD7 is very distinct from that of other FXYD proteins. For our structure-function analysis, we concentrated on two distinctive sequence motifs namely a 56 SRSES 60 sequence, which contains 3 putative phosphorylation sites as predicted by the NetPhos 2.0 program (20), and a 75 PGGGGV 80 sequence at the C-terminal end of the protein (see Fig. 1).
FXYD7 in which Ser 56 , Ser 58 ,and Ser 60 were replaced by alanine (Ct-3S/A, abolishment of the phosphorylation sites) or by aspartic acid (Ct-3S/D, mimicry of phosphorylation) showed a stability and processing similar to wild-type FXYD7 (Fig. 6A,  compare lanes 1-3 to lanes 7-12). Both mutants associated with Na,K-ATPase (Fig. 6B compare lanes 1-3 and lanes 7-12) and produced a functional effect similar to wild-type FXYD7 ( Fig. 6D and Table I). Further experiments are necessary to investigate the possible functional role of phosphorylation in FXYD7.
The putative role of the 75 PGGGGV 80 sequence was assessed in a mutant lacking the seven last C-terminal amino acids (Ct⌬7). Truncation of the PGGGGV sequence significantly delayed and decreased the processing of FXYD7 expressed in oocytes. Whereas part of the newly synthesized, wild-type FXYD7 population became O-glycosylated already during the 6-h pulse period (Fig. 6A, lane 1) and the total population after a 24-h chase period (lane 2), the Ct⌬7 mutant was not processed during the pulse (lane 4), and only part of the newly synthesized Ct⌬7 population was glycosylated after a 24-h chase period (compare lane 4 to lane 5). The remaining nonglycosylated Ct⌬7 species were degraded during the chase period (lanes 4 -6, 12-kDa band). Since O-glycosylation mainly occurs in an early Golgi compartment (for review see Ref. 21), this result indicates that the Ct⌬7 mutant exits less rapidly from the ER than wild-type FXYD7. Association of Na,K-ATPase with the Ct⌬7 mutant was somewhat less efficient than with wild-type FXYD7 (Fig. 6B, lanes 1-6). Interestingly, in oocytes co-expressing exogenous Na,K-ATPase, the proportion of O-glycosylated Ct⌬7 increased after a 24-and a 48-h chase (compare Fig. 6, A and B, lanes 4 -6) indicating that association of Ct⌬7 with Na,K-ATPase abolishes the retention effect of the mutation.
In order to define more precisely the amino acids involved in the delayed processing of Ct⌬7, we prepared a mutant in which the PGGGG motif was replaced by alanines but which still contained the terminal valine residue (Ct-P4G/A). As shown in Fig. 7A (lanes 16 -18), this mutant, when expressed alone in oocytes, exhibited a time course of processing similar to wildtype FXYD7 (lanes 1-3) indicating that the terminal valine residue and not the PGGGG motif is involved in the normal processing of FXYD7. Indeed, deletion of the terminal valine (⌬Val 80 ) led to a similar delay in FXYD7 processing (Fig. 7A,  lanes 7-9) than that observed with the Ct⌬7 mutant (Fig. 7A,  lanes 4 -6). Interestingly, replacement of Val 80 by another hy- ]methionine, and subjected to 24-and 48-h chase periods. Digitonin extracts were prepared after the pulse and chase periods, and immunoprecipitation was performed in denaturing conditions using an anti-FXYD7 antibody. Immunoprecipitates were separated on Tricine-polyacrylamide gels. B and C, Xenopus oocytes were injected with wild-type or mutant FXYD7 cRNAs (2 ng) together with rat Na,K-ATPase ␣1 (10 ng) and ␤1 (1 ng) cRNAs. Oocytes were metabolically labeled and digitonin extracts were prepared as described in the legend to Fig. 2. Non-denaturing immunoprecipitations were performed using an anti-FXYD7 antibody (B) or an anti-␣ antibody (C). One of two similar experiments is shown. D, three days after injection of rat ␣1 and ␤1 cRNAs in the presence or not of wild-type or mutant FXYD7, the external K ϩ activation constant (K1 ⁄2 K ϩ ) of Na,K-ATPase was determined in the presence of 90 mM external Na ϩ at a holding potential of Ϫ50 mV as described under "Experimental Procedures." Results are means of 8 -10 oocytes from 2-3 batches of oocytes. drophobic amino acid (V80A, lanes 10 -12) but not by a charged amino acid (V80D, lanes 13-15), restored rapid processing of FXYD7. Finally, to substantiate that Val 80 contributes to ER exit of FXYD7, we compared the kinetics of cell surface appearance of epitope-flagged wild-type FXYD7 and of the ⌬V80 mutant by radioimmunoassays with anti-FLAG antibodies on intact oocytes. As shown in Fig. 7B, 3 and 6 h after cRNA injection, the cell surface expression of the ⌬V80 mutant was significantly lower than that of wild-type FXYD7 though both proteins reached a maximal expression level after 24 h. Altogether these results indicate that the presence of a valine at the end of the intracellular domain is involved in the control of ER exit of FXYD7 and determines the rate of its cell surface expression.

DISCUSSION
The brain-specific FXYD7 belongs to the FXYD protein family and was shown to associate with and regulate Na,K-ATPase transport properties (7). FXYD7 is a small type I protein with three distinct parts, a N-terminal extracellular domain, a transmembrane domain and a C-terminal intracellular domain. In this study, we reveal structural determinants in each of these domains that are functionally relevant either for FXYD7 in particular or for FXYD proteins in general. FXYD7specific parts of the extracellular domain are important for its stability, conserved amino acids in extracellular and the transmembrane domain are involved in the association, probably of all FXYD proteins, with the Na,K-ATPase, and, finally, the intracellular domain contains FXYD7-specific determinants involved in the intracellular routing of FXYD7.
Determinants of the Stability and Intracellular Routing of Newly Synthesized FXYD7-Our previous observation that mutations of threonines implicated in O-glycosylation leads to degradation of FXYD7 (7) indicates that O-linked sugar chains in the N terminus of FXYD7 contribute to its stabilization. Since O-glycosylation is a post-translational event that mainly occurs in different Golgi compartments (for review see Ref. 21), it is difficult to conceive how it can protect FXYD7 from the ER quality control system involved in the recognition and degradation of improperly processed or misfolded proteins (for re-view see Ref. 22). However, it may be postulated that newly synthesized FXYD7 recycles between the ER and an early Golgi compartment, and only escapes ER degradation if glycosylation occurs during its presence in the Golgi, which impedes retrograde transport to the ER and permits further routing to the plasma membrane. Significantly, association with Na,K-ATPase prevents the degradation of unglycosylated FXYD7 (Ref. 7 and this study). Since the association process most likely occurs in a post-ER compartment (see below), it may be inferred that association with Na,K-ATPase impedes retrograde transport of unglycosylated FXYD7 to the ER and thus prevents its degradation. In this study, we also show that truncation of 15 N-terminal amino acids partially protects FXYD7 from degradation. This observation may indicate that a degradation signal exists in the N-terminal part of FXYD7, which is attenuated in the truncated mutant or that this mutant has partially lost its ability to recycle to the ER.
In contrast to FXYD2, which is rapidly degraded in the ER if not associated with Na,K-ATPase (23), FXYD7 becomes stabilized by O-glycosylation in the Golgi and can be routed to the plasma membrane as an individual protein. In this study, we have identified a C-terminal valine residue which may be critically involved in the rapid exit from the ER e.g. from the site where newly synthesized and not yet O-glycosylated FXYD7 are degraded. Indeed, we observe that deletion of the most C-terminal valine residue leads to a delay in O-glycosylation and to the retention and degradation of an important part of the newly synthesized FXYD7 population in the ER. Mutation of the preceding Pro-Gly-Gly-Gly-Gly sequence into Pro-Ala-Ala-Al-Ala does not abolish the functional effect of the C-terminal valine residue indicating that it works independently. It can, however, not entirely be excluded that the Pro-Ala-Ala-Al-Ala sequence still has characteristics of a flexible spacer that may be necessary for proper functioning of the valine residue.
It has long been believed that soluble and membrane-bound proteins exit nonselectively from the ER and are packaged by default into transport vesicles (24) destined for the ER-Golgi intermediate compartment (25). Increasing experimental evidence supports an alternate mechanism (26), which predicts that at least some proteins contain ER-export motifs that selectively concentrate these proteins into vesicles during the budding process. In certain proteins, diphenylalanine or diacidic motifs were shown to be required for selective ER export or efficient ER to Golgi transport. A similar role has been attributed to a C-terminal valine residue in proTGF␣, MT1-MMP (27), and CD8 (28) with a more or less stringent effect on the ER exit process depending on the protein. By screening of databases, a single C-terminal valine residue was found in 9.8% of 488 human type I membrane proteins which is 1.7-fold higher than theoretically expected from the frequency of valine in proteins (29). Moreover, several polytopic membrane proteins such as plasma membrane Ca-ATPase, band 3 anion exchanger (30), various potassium channels etc have a putative valine signal. These results are consistent with the notion that regulation of ER export by C-terminal valine residues is a widely used mechanism.
At present, is not known how the C-terminal valine residue produces its functional effect. It has been suggested that, similar to other transport signals, C-terminal valine residues interact with COPII proteins which coat vesicles involved in ER export (29). On the other hand, C-terminal valine residues are often part of a motif, which determines interaction with PDZ binding domain containing proteins (31). Significantly, two hybrid screens using the C-terminal tail of proTGF␣ have indeed identified a specific interacting protein TACIP18, which is a PDZ protein (32). TACIP18 is located in initial compartments of the secretory pathway, in contrast to other members of the PDZ protein family, which interact with membrane proteins close or at the plasma membrane. TACIP18 interaction depends on the C-terminal valine residue in proTGF␣ but otherwise has a more relaxed specificity than other known PDZ proteins (32). Significantly, the PGGGV sequence present in FXYD7 does not belong to the classical PDZ binding motifs. Thus it remains to be established whether it represent a new class of such motifs.
More experimental work is needed to determine the mechanism of the ER export control of FXYD7 dependent on the C-terminal valine residue, and to identify its functional importance. Significantly, only FXYD7 but no other member of the FXYD family has an N-terminal valine residue suggesting that the role of an ER export control may be linked to the specific expression of FXYD7 in neurons and glial cells and/or to the particular requirements of the regulation of Na,K-ATPase expression and function in the brain.
Determinants in FXYD7 That Are Involved in the Interaction with and the Regulation of Na,K-ATPase Expression and Function-Up to now, the only functional role ascribed to FXYD7 is the isozyme-specific regulation of the cerebral Na,K-ATPase. As other Na,K-ATPase regulators of the FXYD family, FXYD7 mediates this function by direct and specific interactions with the ␣-␤ complex. The precise domain(s) or amino acids in FXYD proteins that are directly involved in the interaction with Na,K-ATPase are still unknown. Recent studies suggest that, both in FXYD2 and FXYD4, the transmembrane domain is mainly involved in the stable interaction with Na,K-ATPase (19,33) though in FXYD2 the cytoplasmic C terminus may also play a role in this process (8). In the present study, we confirm recent observations (19) that the conserved Gly 40 , which is associated with a form of renal hypomagnesemia when mutated into an arginine residue in FXYD2 (18), is involved in Na,K-ATPase-FXYD protein interaction. Indeed, replacement of Gly 40 in FXYD7 by Ala, Trp, or Arg progressively decreases its association with Na,K-ATPase. We also show that the second conserved Gly 29 in the transmembrane domain, which is located on the same face of the TM helix (5), plays a critical role for efficient association. Finally, our mutational analysis of FXYD7 reveals that, in contrast to our previous observations made with FXYD2 and FXYD4 (8), the FXYD motif in the extracellular domain of FXYD7 does not affect the stability and/or the efficiency of the interaction with Na,K-ATPase.
As suggested by recent experimental evidence (33), the amino acids in FXYD proteins that are involved in the stable interaction with the Na,K-ATPase may not be the same as those involved in the functional effect, though they are also concentrated in the transmembrane domain. In our study, this could be reflected by the observation that some FXYD7 mutants affecting the N-terminal domain indeed decrease association efficiency in detergent but still produce the same functional effect on the K ϩ kinetics of Na,K-ATPase as wild type FXYD7. Nevertheless, a clear distinction between interaction sites involved in the stable association or in a specific functional effect is not obvious in all cases. Based on experiments with mimetic peptides, it has indeed been proposed that Gly 41 (corresponding to Gly 40 in FXYD7) is not primarily involved in Na,K-ATPase interaction but rather mediates the effect on the Na ϩ kinetics of FXYD2 (34). Knowing that FXYD7 has no effect on the apparent Na ϩ affinity of Na,K-ATPase, this result is difficult to reconcile with our results which show that a G40A mutant of FXYD7, decreases the association efficiency and completely abolishes the K ϩ effect. Obviously, interactions of Na,K-ATPase with single, conserved amino acids cannot explain the diverse functional effects of FXYD proteins.
In addition to structural information, our study also provides insight into dynamic aspects of the association process of FXYD7 with Na,K-ATPase. First, our results indicate that similar to phospholemman (12), efficient association of FXYD7 occurs post-translationally. At present it is not known in which compartment of the secretory pathway association takes place. Since we observe very inefficient interaction of Na,K-ATPase with newly synthesized non-glycosylated FXYD7, we can exclude the ER as an important cellular site of interaction. On the other hand, we observe that FXYD7, mutated in the glycosylation sites, becomes rapidly stabilized by the interaction with Na,K-ATPase. This is consistent with an association compartment close to the ER, for instance an early Golgi compartment. We cannot, however, exclude that association with Na,K-ATPase could occur even in later transport compartments or at the plasma membrane. What is the reason for the post-translational association of FXYD7 with Na,K-ATPase? In view of the presence of an ER-export signal in FXYD7, it is conceivable that Na,K-ATPase and FXYD7 are sorted into different transport vesicles in the ER and therefore exit the ER separately and with different rates.
In conclusion, we have further characterized one of the FXYD proteins, which act as tissue-specific regulators of Na,K-ATPase. Our results on FXYD7 provide information on structural determinants that are relevant for the mechanism of action of all FXYD proteins but also on FXYD7-specific structural and functional properties. Future experiments will be directed to show the physiological relevance of these FXYD7specific characteristics in the functional context of cells in which FXYD7 is expressed.