Multiplicity of expression of FXYD proteins in mammalian cells: dynamic exchange of
phospholemman and gamma subunit in response to stress.
Elena Arystarkhova 1, 3, Claudia Donnet 1, Ana Muñoz-Matta 2, Susan Specht 2, and
Kathleen J. Sweadner 1
1Laboratory of Membrane Biology, Massachusetts General Hospital/Harvard Medical
School, Boston, Massachusetts, 02114, and the 2 Department of Pharmacology and
Toxicology, University of Puerto Rico School of Medicine, San Juan, Puerto Rico 00936.
Na,K-ATPase responds to hypertonicity via FXYD proteins
Dr. Elena Arystarkhova
Laboratory Membrane Biology,
415 Thier Research Building,
Massachusetts General Hospital,
55 Fruit St., Boston, MA 02114
Phone: (617) 726-8560
FAX: (617) 726-7526
Page 1 of 60Articles in PresS. Am J Physiol Cell Physiol (October 18, 2006). doi:10.1152/ajpcell.00328.2006
Copyright © 2006 by the American Physiological Society.
Functional properties of Na,K-ATPase can be modified by association with
FXYD proteins, expressed in a tissue-specific manner. Here we show that expression of
FXYDs in cell lines does not necessarily parallel the expression pattern of FXYDs in the
tissue(s) the cells originate from. While being expressed only in lacis cells in the
juxtaglomerular apparatus and in blood vessels in kidney, FXYD1 was abundant in renal
cell lines of proximal tubule origin (NRK-52E, LLC-PK1, and OK cells). Authenticity of
FXYD1 as a part of Na,K-ATPase in NRK-52E cells was demonstrated by co-
purification, co-immunoprecipitation and co-localization. Induction of FXYD2 by
hypertonicity (500 mOsm with NaCl for 48 h or adaptation to 700 mOsm) correlated with
down-regulation of FXYD1 at mRNA and protein levels. The response to hypertonicity
was influenced by serum factors and entailed first, dephosphorylation of FXYD1 at Ser68
(1-5 hours), and second, induction of FXYD2a and decrease in FXYD1 with longer
exposure. FXYD1 was completely replaced with FXYD2a in cells adapted to 700 mOsm
and showed a significantly decreased sodium affinity. Thus dephosphorylation of
FXYD1 followed by exchange of regulatory subunits is utilized to make a smooth
transition of properties of Na,K-ATPase. We also observed expression of mRNA for
multiple FXYDs in various cell lines. The expression was dynamic and responsive to
physiologic stimuli. Moreover, we demonstrated expression of FXYD5 protein in
HEK293 and HeLa cells. The data imply that FXYDs are obligatory rather than auxiliary
components of Na,K-ATPase, and their interchangeability underlies responses of Na,K-
ATPase to cellular stress.
Key words: Na,K-ATPase; cellular stress; regulation; FXYD1; FXYD2
Page 2 of 60
The Na,K-ATPase catalyzes active efflux of Na+ and uptake of K+ ions, thus
establishing and controlling ionic gradients across the plasma membrane in virtually
every animal cell. It is the receptor for cardiac glycosides, drugs commonly used in
treatment of heart failure. Recent findings suggest that binding of ouabain to Na,K-
ATPase, besides a direct inhibition of the pump activity, initiates a signaling pathway that
ultimately leads to changes in cell hypertrophy and proliferation.
Regulation of Na,K-ATPase occurs at several levels. It may entail regulation of
gene expression of the obligatory α and β subunits (long-term regulation, which may
serve as an adaptation response) or recruitment/internalization of the active pump units
to/from plasma membrane in response to certain physiological stimuli, which can serve
as an immediate regulatory response. On top of this, multi-layered regulation of Na,K-
ATPase includes modulation of the intrinsic properties of the enzyme via physical
association with the so-called FXYD proteins [reviewed in (29,31)]. This is a family of
small (6.5-17 kDa) single-span membrane proteins with structural homology (61). The
family includes: phospholemman (PLM) (FXYD1), the major plasma membrane
substrate for PKA and PKC phosphorylation in heart (49); the γ subunit (FXYD2) (45),
which is expressed primarily in kidney in at least two splice variants (6); Mat 8
(FXYD3), a gene up-regulated in epithelial cells by oncogenic Ras and Neu and
expressed in human breast and prostate tumors (46); CHIF (corticosteroid hormone-
induced factor) (FXYD4) (7); dysadherin, or RIC (related to ion channel) (FXYD5)(27);
FXYD6 (phosphohippolin) (61); and brain-specific FXYD7. Most of the homology is
within the transmembrane segment and the signature motif PFXYD (61).
Page 3 of 60
To date, 6 members of the family have been demonstrated to associate with the
Na,K-ATPase: FXYD1 (20,26); FXYD2 (6,51); FXYD3 (22), FXYD4 (10), FXYD5
(43), and FXYD7 (11). Remarkably, association of Na,K-ATPase with each of the FXYD
proteins resulted in a somewhat different modulation of the kinetic properties of the
Na,K-ATPase. FXYD4 (10,30) and FXYD5 (43) stimulated Na,K- ATPase activity
either by increasing the affinity for Na+ or Vmax, whereas FXYD3 (22), FXYD7 (23), and
FXYD2 exhibited an inhibitory effect by either decreasing the affinity for Na+ and/or K+,
or reducing the Vmax of the pump (2,42,67). Overexpression of FXYD1 reduced apparent
affinity for Na+ and K+ in heterologous expression systems (4,20) and decreased the Vmax
of Na-pump in adult rat myocytes (72). Phosphorylation of FXYD1, on the other hand,
seemed to release the inhibition with a subsequent activation of the pump (13,28).
Some of the FXYD family members exhibit a high degree of tissue specificity.
FXYD1 is highly expressed in heart and brain (cerebellum and choroid plexus) (20,26),
but in kidney it is limited to the juxtaglomerular apparatus and blood vessels (68).
Conversely, FXYD2 is only found in kidney in healthy animals (45). In stomach,
expression of FXYD3 is in mucous cells (22), while FXYD4 is expressed in distal colon
and collecting duct in kidney (18). Whether expression of FXYD6 and FXYD7 in CNS
overlaps is not known yet. Expression of FXYD5 in normal tissue is preferentially in
kidney, intestine, spleen and lung (43). In kidney FXYD5 is at the basolateral membrane
of connecting tubules and the intercalated cells of collecting duct, but it was also detected
on apical membranes in thin limb of Henle, suggesting a potential role in addition to
being part of the Na,K-ATPase complex.
Page 4 of 60
Tissue/cell-specific expression of the regulatory FXYD subunits of Na,K-ATPase
is not static, however, and may be changed to adapt to a given physiological or
pathological situation. For instance, significant up-regulation of FXYD1 was observed in
response to cardiac infarction (57) and nerve injury (19). Expression of FXYD1 is also
stimulated in muscle with exercise (54). Gene array studies revealed expression of
FXYD2 in hippocampus in response to chemical insult (38). Expression of FXYD3 may
be induced by ras- and neu-oncogenes (46), while expression of FXYD4 is regulated by
corticosteroid and K+-deprivation (66). FXYD2, which is normally limited to kidney, can
be also induced in cells of a variety of different origins in culture by exposure to different
kinds of cellular stress (hypertonicity, heavy metals, exogenous oxidation, heat shock)
Here we present evidence that expression of FXYDs is even more plastic in cells
in culture, and cells of different origin may express more than one FXYD at a time (at
protein and/or mRNA level). Moreover, we show that cellular stress (hypertonicity)
initially caused dephosphorylation with a subsequent down regulation of endogenously
expressed FXYD1 followed by its replacement with a newly synthesized FXYD2a within
the Na,K-ATPase complex in NRK-52E cells. As we showed previously (67), induction
of FXYD2a by 48 hour exposure to hypertonicity correlated with a significant reduction
of Na,K-ATPase activity. Here we observed about 20% reduction in ATPase activity
even after 5 hours of hypertonic challenge. Since no FXYD2a was synthesized yet at this
point (67), but dephosphorylation of FXYD1 was dramatic, we suggest that the latter
event is employed as a first line of cell defense to survive an apoptotic insult under stress-
related conditions. The data support the hypothesis that alterations of Na,K-ATPase
Page 5 of 60
properties entail exchanges of FXYD subunits, and this phenomenon represents an
important adaptive mechanism.
Some preliminary data have been presented (24).
MATERIALS and METHODS
Antibodies and Cell Lines.
Polyclonal antisera K1 or K3 (60) were used for detection of the α1 subunit of
Na,K-ATPase on blots. The VG4 monoclonal antibody (3) was employed for
immunoprecipitation of α1. The RCT-G1 polyclonal antibody raised against the C-
terminal peptide (5) was used to identify both splice variants of FXYD2. On blots, non-
phosphorylated FXYD1 (PLM) was detected with the PLM-C2 rabbit antiserum raised
against the C-terminal sequence of phospholemman (the generous gift of Dr. J. Cheung,
Milton S. Hershey Medical Center, Pennsylvania State University) (59). The form of
FXYD1 that is phosphorylated at Ser68, was detected with the CP68 antibody provided
Dr. J. R. Moorman, University of Virginia Health System)(55). For immunoprecipitation
of FXYD1, PLM-C1, an affinity-purified polyclonal rabbit antibody against the C
terminus, was employed (a kind gift of Dr. L. R. Jones, Indiana State University). The
polyclonal antibody against FXYD5 (anti-dysadherin, ab2248) was from AbCam Inc.
NRK-52E, C6, MDCK, LLC-PK1, Caco-2, OK, HEK293, HeLa, and SH-SY5Y
cell lines were purchased from ATCC and grown according to the company’s
Page 6 of 60
recommendations. mIMCD3 cells were a kind gift of Dr. J. Capasso (Colorado Health
Sciences Center). Differentiation of SH-SY5Y cells was with retinoic acid (10 µM).
Cell Culture and Hypertonicity Treatment.
The normal rat kidney epithelial cell line (NRK-52E) was grown to 85–90%
confluency in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (300
mOsm). The medium was then replaced with either control or hyperosmotic (500 mOsm
total) medium supplemented with NaCl. In the case of serum reduction, cells were
incubated overnight in a medium containing 0.5% FBS before hypertonicity treatment in
the same concentration of serum. Flasks were washed with Dulbecco’s phosphate-
buffered saline with Ca2+and Mg2+at various times and frozen at -80°C.
Adaptation of NRK-52E cells to high hypertonicity was similar to what was
described for mIMCD-3 cells (15). Briefly, hypertonicity was raised in the medium by
successive increments of 50 mOsm (with NaCl). Cells were allowed to adapt to each new
medium for 2-3 weeks before the next increase in osmolality. Passaging was performed
when cells reached confluency.
Membrane Preparations and Enzyme Purification.
Crude membrane preparations were obtained from scraped cells by
homogenization and differential centrifugation (5). Isolation of membranes from rat
kidney outer medulla was performed as described elsewhere (60). Partial purification of
Na,K-ATPase from cell membranes (1.4 mg/ml) was with SDS extraction (0.56 mg/ml)
and sedimentation on 7–30% sucrose gradients. Typical specific activity of Na,K-ATPase
was about 50–150 µmol Pi/mg of protein/h in preparations from NRK-52E cells.
Page 7 of 60
To obtain cell lysates, a buffer containing 50 mM Tris-Cl, pH 8.0, 5 mM EDTA,
1% NP-40 and protein phosphatase inhibitors cocktail I (Sigma) (1:100, v/v), was used.
Cells were scraped, triturated, and insoluble material was removed by centrifugation at
3,000 x g for 10 min (Sorvall, SS-34).
Na,K-ATPase activity was measured in SDS-purified preparations from either
control or adapted cells as a function of Na+concentration in media containing 3 mM
Tris-ATP, 4 mM MgCl2, 30 mM histidine, pH 7.4, and in the presence of 20 mM K+. All
the reactions were performed at 37 °C for 30 minutes with and without 3 mM ouabain,
and ouabain-sensitive Pi release was measured colorimetrically by using the Fiske-
Subbarow method. Data were analyzed by nonlinear regression using Sigma Plot Graph
System (Jandel Scientific). Na+-activation curves were fitted according to the Hill model
for ligand binding.
Activity in crude membranes was measured using the so-called “yellow method”,
which tolerates more protein and lipid. After quenching the ATPase reaction with acid
and molybdenum as in the Fiske-Subbarow method, the modification involves the
extraction of liberated Pi as the unreduced (yellow) phospho-molybdenum complex into
an organic phase (water-saturated n-isobutanol), and reading the absorbance at 380 nm.
Protein phosphatase inhibitor was added throughout the isolation of membranes and
Immunoprecipitation. Plasma membrane-enriched fractions from NRK-52E cells (1
mg/ml) were solubilized with 3 mg/ml n-dodecyl octaethylene glycol monoether
detergent (C12E8; Calbiochem) (6) for 10 min at room temperature in buffer A, containing
Page 8 of 60
140 mM NaCl, 25 mM imidazole, and 1 mM EDTA, pH 7.3. At the end of the
incubation, the solubilized material was diluted with 2 volumes of buffer A and
sedimented by centrifugation for 30 min at 20,000 × g at 4°C. The supernatant was
incubated with primary antibodies (PLM-C1 to immunoprecipitate phospholemman or
VG4 to immunoprecipitate the α subunit) or control IgG (rabbit or mouse, respectively,
at 1-2 µg/ml) overnight at 4°C with rocking.The immune complexes were collected after
2 hr of incubation with 40 µl of secondary goat anti-rabbit or goat anti-mouse IgG
antibodiescovalently bound to agarose beads (Sigma). Immunoprecipitateswere collected
by centrifugation at 9300 × g for 10 min at 4°C and washed four times with buffer A
containing 0.05% C12E8. After the final wash, the pellet was resuspended in 1×
electrophoresis sample buffer.
Membrane fractions were resolved either on SDS-Tricine or on 4-12% MES-SDS
gels (NuPage system, Invitrogen). Proteins were transferred to nitrocellulose, incubated
with antibodies, and detection was with chemiluminescence. For positive control samples
membrane preparations from kidney or canine cardiac sarcolemma (37) were used.
Quantification was with ImageQuant TL image analysis software (Amersham).
NRK-52E cells, either wild type or adapted to 700 mOsm, were seeded into 38-
mm2 wellsof 96-well flat-bottomed plates (5 x 103/ well) in quadruplicate and allowed to
adhere overnight. Cell proliferation over the following 2-6 days was assayed based on the
cleavage of the tetrazolium salt WST-1 (Roche Diagnostics) by mitochondrial
Page 9 of 60
dehydrogenases in viable cells. Quantification was with a microtiter plate reader at 450
Immunocytochemistry was performed as described elsewhere in more detail (6).
Briefly, cells were fixed with 2% periodate/lysine/paraformaldehyde solution for 30 min
at room temperature followed by 5 minute incubation with 1% SDS in PBS, several
washes with PBS, and blocking with 5% goat serum in PBS to prevent nonspecific
binding of secondary antibodies. McK1 antibody was used for α1 detection, whereas
PLM-C1 (described above) and CP68 antibodies were used to probe non-phosphorylated
or phosphorylated FXYD1, respectively. The secondary antibodies were either Cy3-
conjugated goat anti-mouse IgG (1:300; Accurate) or fluorescein isothiocyanate-
conjugated goat anti-rabbit IgG (1:300; Jackson ImmunoResearch). Slides were
examined with a Nikon TE300 fluorescence microscope equipped with a Bio-Rad MRC
1024 scanning laser confocal system (version 3.2).
Total RNA from cells was prepared with the RNeasy system (Qiagen). The cDNA
was obtained using 1 µg of total RNA, oligo(dT) as the priming oligonucleotide, and
Super Script II reverse transcriptase (Invitrogen). For HEK293 and HeLa cells , cDNA
was obtained with the Super Script III cell Direct cDNA synthesis system (Invitrogen).
The polymerase chain reaction was performed in the presence of 3 mM MgCl2 with either
Taq polymerase (Fisher) or Platinum Taq polymerase (Invitrogen). PCR products were
separated by electrophoresis in 2% agarose gels. Primers were chosen according to
Page 10 of 60
sequences of FXYD genes from rat, mouse and human (Table 1) using software provided
FXYD1 is associated with Na,K-ATPase in NRK-52E cells.
Expression of proteins in cultured cells often parallels the expression pattern in
the tissues they originate from. However, expression of regulatory subunits of Na,K-
ATPase (FXYDs) does not entirely follow the same paradigm. FXYD1,
phospholemman, is abundant in muscle, brain, and heart, where it represents the major
substrate for PKA and PKC. In kidney, though, its expression is limited to afferent
arteriole and lacis cells in the juxtaglomerular apparatus (68). Nevertheless, as shown in
Fig. 1A, the FXYD1 protein was readily detectable in crude membrane preparations (P2)
from renal NRK-52E cells, which possess characteristics of proximal tubule cells.
Moreover, FXYD1 remained tightly bound to Na,K-ATPase from NRK-52E cells during
a purification procedure that involves mild SDS treatment of crude membranes and
extraction of the majority of membrane proteins other than Na,K-ATPase (5). The blots
were stained with the C2 antibody that is specific for non-phosphorylated FXYD1 and
exhibits a very weak cross-reactivity with phosphorylated FXYD1 (55). No signal was
observed when identical blots were stained with the CP68 antibody (raised against the
peptide phosphorylated at the Ser68 site) (55) (not shown). Since no protein phosphatase
inhibitors were employed here, the assumption was that any phosphorylation of FXYD1
at Ser68 site was labile and was lost during membrane preparation. Therefore, unless
otherwise specified, we used staining by C2 as a measure of the total level of FXYD1
Page 11 of 60
Interaction of the α subunit of Na,K-ATPase and FXYD1 was further
corroborated by their co-immunoprecipitation from crude membranes solubilized with
non-ionic detergent C12E8 (Fig. 1B). Antibody against FXYD1 brought down the α
subunit of Na,K-ATPase, and anti-α antibodies precipitated FXYD1 as well (not shown).
[The doublets of FXYD1 seen in some figures but not in others may have something to
do with the different gel systems employed (SDS-Tricine vs 4-12% gradient SDS-MES
gels from Invitrogen) and were not investigated further]. All together the data indicate
that FXYD1 is an authentic component of Na,K-ATPase in renal NRK-52E cells.
Interestingly, the ratio between FXYD1 and the alpha subunit was significantly
lower in the purified preparation of Na,K-ATPase compared to crude membranes (Fig.
1A) implying the existence of a pool of FXYD1 not tightly bound to Na,K-ATPase.
We used immunocytochemistry to monitor the location of FXYD1 in NRK-52E
cells. Cells were fixed with 2% paraformaldehyde and stained for FXYD1 and the α
subunit of Na,K-ATPase. Two different antibodies were employed here to distinguish
between phosphorylated (Ser68) and non-phosphorylated forms of FXYD1. As shown in
Fig. 2, subcellular location of FXYD1 was apparently affected by the phosphorylation
status of the protein. Good co-localization of FXYD1 phosphorylated at Ser68 site and α
was observed at plasma membrane (Fig. 2, A-C), while a majority of unphosphorylated
FXYD1 was seen intracellularly (Fig. 2, D-F). The data are in line with prior data (39)
suggesting that phosphorylation of FXYD1 may be prerequisite for trafficking to plasma
Thus, based on immunofluorescence data, we speculate that a “free pool” of
FXYD1 may consist of an intracellularly located non-phosphorylated FXYD1, whereas a
Page 12 of 60
major fraction of FXYD1 associated with the pump at plasma membrane consists of a
phosphorylated form of the protein.
We also detected FXYD1 in other renal cell lines originating from proximal
tubules: porcine LLC-PK1 and opossum OK cells (Fig. 3). Meanwhile no FXYD1 was
detected in canine MDCK (Fig. 3) or mouse mIMCD3 cells (not shown), both originating
from distal nephron segments, or in embryonic HEK293 cells. The antibody’s epitope is
conserved. This implies cell (tubule)-specific control of FXYD1 expression in culture.
Interestingly, in kidney, proximal tubules are positive for FXYD2a, while more distal
segments express either FXYD2b (distal convoluted tubules) or a combination of
FXYD2b and FXYD2a (inner medullary collecting duct). Although abundant in tissue,
FXYD2 is normally absent from any mammalian renal cell line (5,62). The attractive
hypothesis is that FXYD1 substitutes for FXYD2a within the Na,K-ATPase complex in
cells in culture in order to adjust intrinsic properties of the pump properly for the
Transfection with FXYD2 affects the expression of endogenous FXYD1.
We have shown previously that expression of the FXYD2 splice variants can be
achieved in NRK-52E cells by transfection (2,5), although the level of expression of
FXYD2a or FXYD2b was never as high as in kidney membranes. With both FXYD2a
and FXYD2b there was a reduction in the activity of Na,K-ATPase and a reduction in the
rate of cell growth. Since FXYD1 is endogenously expressed in NRK-52E cells as an
authentic component of Na,K-ATPase, we tested whether expression of FXYD2 by
transfection would cause a replacement of one FXYD protein with another. Several
clones expressing splice variants of γ (FXYD2) have been tested, and Fig. 4 shows a
Page 13 of 60
representative Western blot analysis of crude membranes from γa-, γb-, and mock-
transfected cells. Transfection with γa correlated with only slight (if any) reduction of the
total level of FXYD1 expressed in NRK-52E: 92 +/- 13% (n = 3) expressed as the ratio
between the staining of PLM (C2) and the α subunit (K3) in transfected vs control cells.
In contrast, the FXYD1/α ratio was significantly lower in γb-transfectants compared to
γa- (18.1 +/- 4.5 %, n = 4) and mock-transfected cells (11.2 +/- 2.3 %, n = 4), even
though the amount of γ was lower in γb- compared to γa transfectants (Fig. 4). The data
imply a cross-talk in regulation of FXYD protein expression and a potential competition
between γb and phospholemman for binding to Na,K-ATPase. Since both γa- and γb-
transfectants exhibited similar slow growth compared to mock-transfected cells, the
conclusion is that introduction of FXYD2 rather than relative changes in expression of
FXYD1 had the impact on the rate of cell proliferation, which is in agreement with our
recent data on knock-down of FXYD2 in NRK-52E cells (67), where silencing restored
Na,K-ATPase activity and the normal rate of growth in the continued presence of
Reciprocal regulation of FXYD1/FXYD2a expression under stress-related conditions
As we demonstrated previously (67), FXYD2a is not expressed in renal cells in
culture under normal conditions, but can be induced by several kinds of cellular stress,
such as hypertonicity (with NaCl and sucrose, but not with urea), heat shock, exogenous
oxidants, or heavy metal treatment. The newly synthesized FXYD2a associates with the
Na,K-ATPase and inhibits its activity. Here we tested what happens with the
constitutively expressed FXYD1 during stress-mediated induction of FXYD2a.
Page 14 of 60
Fig. 5 shows a representative Western blot of crude membranes from either NRK-
52E cells (Fig. 5A) or C6 glioma cells (Fig. 5B) exposed to isotonic (ctr) or hypertonic
conditions (500 mOsm with NaCl) in the presence of 10% FBS. In both cases induction
of FXYD2a correlated with a reduction of the total level of FXYD1. (No protein
phosphatase inhibitors were added during membrane preparation). Based on densitometry
analysis, the FXYD1/α ratio was decreased by 31.8 +/- 3.5 % (n = 5) in membranes from
NRK-52E cells acutely treated with hypertonicity compared to control. Down regulation
of FXYD1 with hypertonicity was greater (at least 50% or more) in C6 cells suggesting
specificity of cellular response.
The changes in the reciprocal expression of FXYD1/FXYD2a were even more
dramatic in NRK-52E cells slowly adapted to hypertonicity. Although these cells
originate from proximal tubules (40), which do not experience high hypertonicity in vivo,
NRK-52E can be adapted to live under high salt similar to what was described for
mIMCD3 cells (15). Adapted cells (700 mOsm with NaCl in the presence of 10% FBS))
are viable, but proliferate much slower than control cells grown in 300 mOsm medium
(Fig. 6A). Western blot analysis of crude membranes showed that adaptation to high salt
caused a) an increase in protein expression of the α subunit ( 1.7 +/- 0.5 fold, n = 4) ; and
b) a reciprocal switch between FXYD proteins: induction of FXYD2a was accompanied
by a complete loss of FXYD1 from cells slowly adapted to high salt (Fig. 6B).
Functionally, there was a significant 1.5 +/- 0.4 (n = 3) fold increase in Na,K-ATPase
activity in membranes from adapted cells that correlated nicely with an increase in the
amount of α, implying no significant changes in activity per unit of protein measured
under optimal conditions. The data are in qualitative agreement with a report by Capasso
Page 15 of 60