Na+/K+-ATPase-associated FXYD3 Protein As An Intracellular Therapeutic Target in Cancer

FXYD proteins associate closely with- and protect plasmalemmal Na + /K + -ATPase against oxidative inhibition. One of them, FXYD3, is often overexpressed in cancers, including those of breast and pancreas. Down-regulation of overexpression in MCF-7 breast cancer cells with siRNA augments doxorubicin-induced cytotoxicity. Because down-regulation with siRNA is not readily translated therapeutically, we developed a peptide as an alternative for suppression of FXYD3.

Members of the mammalian FXYD protein family, numbered 1-7, associate closely with Na + /K + -ATPase and modify its activity. They are expressed in a tissue-dependent manner [1] and in normal tissues, FXYD3 is predominantly expressed in the stomach and colon. However, as cancers develop, FXYD3 often becomes overexpressed in a variety of tissues, most commonly in cancers of prostate [2], breast [3] and pancreas [4].
Overexpression of FXYD3 has led to its use as a marker of malignancy. However, studies on cultured human MCF-7 breast cancer cells indicate that high expression may also be a marker of resistance to chemotherapy and radiotherapy. Such treatments further increase expression of FXYD3. Down-regulation of overexpression with siRNA reduces treatment resistance [5].
FXYD3 might confer treatment resistance because it protects Na + /K + -ATPase against inhibition induced by oxidative stress [6]. Such protection is potentially important therapeutically because Na + /K + -ATPase function is critical for cell survival, malignant tumors have strong antioxidant defence mechanisms and cytotoxic effects of many treatments are at least in part due to the increase in oxidative stress the treatments cause [7]. Na + /K + -ATPase is considered a potential target for cancer treatments [8] We have examined if FXYD3 might affect proteins important in carcinogenesis and known to be activated by treatment-induced oxidative stress. We supressed FXYD3 by exposing cells to siRNA or to a peptide derivative of FXYD3, FXYD3-pep, that displaces the wild-type (WT) protein from Na + /K + -ATPase. We examined whether suppression of FXYD3 alters expression of glutathione-S-transferase π (GST-π) that is implicated in treatment resistance [9] and the tumor suppressor p53 that has a multitude of roles in cancer [10], including in the expression of FXYD3 [11]. We also studied activation of caspase 3/7 that can be triggered by treatment-induced oxidative stress [10]. The peptide derivative FXYD3 we developed as a tool for the study is potentially a platform for novel therapeutic compounds.

FXYD3 siRNA Transfection
A human siRNA against FXYD3 (consisting of pools of three to ve target-speci c 19-25nt siRNAs designed to knockdown gene expression, sc-60665), and the non-silencing control (control siRNA-A, sc-37007) were purchased from Santa Cruz Biotechnology. Cells were transfected using the siRNA Transfection Reagent/Medium (Santa Cruz Biotechnology) according to the manufacturer's instructions. FXYD3 mRNA expression level was quanti ed by real-time RT-PCR after exposure to FXYD3 siRNA as described previously [5]. FXYD3 protein abundance was measured by western blotting using GADPH as a loading control [5].
Cell viability and apoptosis. Cell metabolic activity was assayed as described [13] by estimating reduction of XTT (2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5 carboxyanilide), using a kit (Cell Signaling Technology). The metabolic activity was used as a surrogate for viability. For each set of experimental groups, we performed 5 experiments with 4 replicates. Caspase-3-like activity is increased through a protease cascade during the early stage of apoptosis [14] and we measured activities of Caspase 3/7 (DEVDase) using the Caspase uorogenic substrate (Calbiochem) as described [5].

Western Blot
Western blotting of cell lysates was performed as described [5]. Western blot chemiluminescence was read by a LAS-4000 image reader and quanti ed by densitometry using Multi Gauge 3.1 software (Fuji lm Life Science, Tokyo, Japan). Exposure times were adjusted to ensure that the variation in signal intensity was in the linear dynamic range [5]. Additional details are available in Supplementary Documentation.

FXYD3 derivatives
Exposure of cardiac myocytes to highly lipophilic recombinant WT FXYD3 protein displaces WT FXYD1 that is endogenous to cardiac myocytes. A Cys-mutated recombinant FXYD3 eliminates the function of WT FXYD1 to protect Na + /K + -ATPase function against oxidative inhibition [6]. The amino acid sequence of WT FXYD3 is shown in Fig. 1. Cys residues replaced with Ser in Cys-mutated derivative protein are indicated. Figure 1 also indicates a speci c Cys residue identi ed by mutational studies in Xenopus oocytes as critical for the protective effect the WT protein has against oxidative Na + /K + -ATPase inhibition Since large cytoplasmic and extracellular domains are poorly conserved across the FXYD protein family [15] we predicted peptides with most of these domains eliminated but retaining binding regions of FXYD proteins to the Na + /K + -ATPase would have effects similar to full-length Cys-mutated FXYD3. We derived a residues of the WT protein retained was used in control experiments ( Fig. 1) FXYD3-pep labelled with the uorescent dye tetramethylrhodamine (TRITC) covalently bound to the extracellular N-terminal was used to study its distribution in cells exposed to it. Customized peptides were synthesized by Mimotopes Pty Ltd, Australia.

Patch Clamping
The whole cell patch-clamp technique was used to measure electrogenic Na + /K + -ATPase pump current (I p , arising from the 3:2 ratio of intracellular Na + pumped out of cells in exchange for extracellular K + pumped in). Solutions and voltage-clamp protocols were designed to minimize non-pump currents. Patch pipette solutions that perfuse the intracellular compartment after the whole-cell voltage clamp I p was identi ed as the inward shift in holding current with exposure to K + -free extracellular solutions to eliminate Na + /K + -ATPase activation at extracellular sites as described previously for measurements in cardiac myocytes [20]. We switched to the K + -free superfusate 2-3 minutes after the whole-cell con guration had been established. Na + /K + -ATPase pump currents are small relative to other membrane currents and it is important for their accurate measurement that holding currents are stable before and after the switch from K + -containing to K + -free solutions. As for measurement of I p in cardiac myocytes [20] we used predetermined criteria for stability of holding currents before and after changing to K + -free extracellular solution. I p was normalized for cell membrane capacitance, and hence, cell size.

Fluorescent confocal microscopy
BxPC-3 cells were exposed to FXYD3-pep or to FXYD3-pep labeled with TRITC. Cells were exposed to the peptides for 2 hours, the media removed, peptide-free fresh media added, and cells then incubated for further 24 hrs. They were then xed in 3.7% paraformaldehyde, washed and mounted on non-coated color frost slides in Fluoroshield mounting medium with DAPI (ab104139, Abcam) and examined under a laser scanning confocal microscope (Leica TCS SP5, Germany). The excitation wavelength was 543 nm, and the emission wavelength was 572 nm. The uorescence images were obtained using constant settings of scanning speed, pinhole diameter and voltage gain.

Statistical analysis
Results are presented as mean ± standard deviation (SD). IC50 values for doxorubicin (Dox) in cell viability studies were calculated by GraphPad Prism. Unless otherwise indicated statistical comparisons were made with a Student's t-test or ANOVA. P < 0.05 was considered statistically signi cant.

FXYD3-siRNA transfection and doxorubicin-induced cytotoxicity on BxPC-3 cells
We examined if the enhancement of doxorubicin's (Dox) cytotoxicity with siRNA-induced suppression of overexpressed FXYD3 on MCF-7 breast cancer cells [5] is reproduced in BxPC-3 pancreatic cancer cells.

FXYD3 peptide derivatives and cytotoxicity of doxorubicin
To determine if FXYD3-pep can displace WT FXYD3 from Na + /K + -ATPase we exposed BxPC-3 cells to the peptide for 2 hours. Using an antibody directed against an epitope in WT FXYD3 that is absent in the shortened peptide 1 µM FXYD3-pep reduced the co-immunoprecipitation (Co-IP) of the large catalytic Na + /K + -ATPase α1 subunit with FXYD3 by ~ 50% (Fig. 3A). The reverse Co-Ip was similarly reduced We examined if displacement of WT FXYD3 from Na + /K + -ATPase is re ected in viability of cells exposed to FXYD3-pep with and without co-exposure to Dox for 48 hours. Culture medium that contained FXYD3pep and Dox was replaced with fresh medium every 24 hours.
FXYD3-pep augmented effects of Dox on the pancreatic BxPC-3 cells that highly express FXYD3. There was a shift in IC50 for Dox from ~ 1.6 µM for cells not exposed to FXYD3-pep to ~ 0.4 µM and ~ 0.3 µM with exposure to 1 µM and 2 µM FXYD3-pep (Fig. 4A). FXYD3-pep did not augment the effect of Dox on pancreatic Panc-1 cells that do not express FXYD3 (Fig. 4B). FXYD3-pep augmented effects of Dox on the MCF-7 cell line with a shift in IC50 from ~ 3.4 µM without exposure to it to ~ 0.5 µM with exposure to 2 µM FXYD3-pep (Fig. 4C). FXYD3-pep did not augment effects of Dox on the breast cancer cell line MDA-MB-468 that expresses FXYD3 at a much lower level than the MCF-7 cells. (Fig. 4D). The peptide alone at 1 or 2 µM had no effect on viability of any of the pancreatic or breast cancer cell lines.
We also examined the effect of the peptide, FXYD3-pepC (Fig. 1C), identical to FXYD3-pep except that it retains the Cys residues of WT FXYD3 that were mutated to Ser in FXYD3-pep. In contrast to co-exposure of BxPC-3 cells to FXYD3-pep and Dox (Fig. 4A), FXYD3-pepC did not augment Dox-induced cytotoxicity (Fig. 4E). It also did not augment Dox-induced cytotoxicity of Panc-1 cells (Fig. 4F). These results implicate mutation of the Cys residues rather than elimination of cytosolic-and extracellular parts of WT FXYD3 in FXYD3-pep's augmentation of cytotoxicity.

Cellular target for FXYD3-pep
Plasmalemmal Na + /K + -ATPase is critical for cell survival and since WT FXYD3 protects Na + /K + -ATPase against inhibition induced by oxidative stress [6] we examined if inhibition of Na + /K + -ATPase augments Dox-induced cytotoxicity. We exposed MCF-7 cells to ouabain that binds to plasma membrane Na + /K + -ATPase at extracellular sites and inhibits its activity. Exposure of cells to ouabain for 48 hours, with or without co-exposure to Dox reduced cell viability in a concentration-dependent manner but did not augment effects of Dox (Fig. 5A).
Levels of the α1 Na + /K + -ATPase subunit, normalized to levels in the non-transformed human breast cell line MCF-10A, were much higher in the BxPC-3 and MCF-7 cells that were sensitized to effects of Dox by FXYD3-pep than in the MDA-MB-468 and Panc-1 cells that were not sensitized. Overexpression of FXYD3 broadly followed the same pattern as α1 Na + /K + ATPase subunit expression (Fig. 5B). We did not detect WT FXYD3 protein or FXYD3 mRNA for Panc-1 (not shown) as others also reported [4].
We examined if the ~ 15-fold higher signal for α1 Na + /K + -ATPase subunits in MCF-7 than MDA-MB-468 cells (Fig. 5B) is re ected by a difference in the cells' functional electrogenic plasmalemmal Na + /K + -ATPase current, I p . I p was measured under conditions causing near-maximal Na + /K + -ATPase turnover and hence re ecting plasmalemmal Na + /K + -ATPase abundance. I p was similar for MCF-7 and MDA-MB-468 cells (Fig. 5C).
The similar plasmalemmal near-maximal Na + /K + -ATPase pump capacity of MCF-7 and MDA-MB-468 cells despite their difference in α1 Na + /K + -ATPase subunit expression implicate the existence a cellular pool of Na + /K + -ATPase not residing in the plasmalemma in cells sensitized to Dox by FXYD3-pep.
To examine the cellular distribution of FXYD3-pep labeled with the uorescent dye TRITC. we used pancreatic BxPC-3 cells because they exhibited the strongest signal for expression of α 1 Na + /K + -ATPase subunits (Fig. 5B) that FXYD3-pep is expected to bind to. TRITC uorescence was detected in the cytosol, particularly in the perinuclear region (Fig. 5D). Fluorescence was not detected in cells exposed to unlabeled FXYD3-pep (data not shown). Fluorescence re ecting binding to plasmalemmal Na + /K + -ATPase was not selectively detected. However, a weak signal from such uorescence is likely to be undetectable with overlap of signals in multi-layered cultured cells.

Cellular response to FXYD3-pep
Overexpression of GSTπ often confers treatment resistance, at least in part by protecting the cancer cells against treatment-induced oxidative stress [9]. Exposure of BxPC-3 cells to 1 µM FXYD3-pep for 24 hours reduced GSTπ expression by ~ 85% (Fig. 6A). A reduction consistent with this occurred with exposure to the full-length recombinant Cys-mutated FXYD3 protein (Fig. 6B). Data pooled from Fig. 6A and Fig. 6B with exposure of cells to FXYD3-pep or the full-length Cys-mutated protein is quantitated in Fig. 6C. Exposure to FXYD3 siRNA also appeared to reduce GSTπ expression and it appeared to block a Doxinduced increase in GSTπ (Fig. 6D). We did not examine effects of suppressing FXYD3 on GSTπ in MCF-7 cells because they are reported to express GSTπ at very low levels [21]. We did not detect it in the MCF-7 cells here (data not shown).
To examine if GSTπ directly facilitates glutathionylation of the β1 Na + /K + ATPase subunit we incubated Na + /K + ATPase-enriched membrane fragments with GSTπ. A clear signal for GSTπ-induced β1 subunit glutathionylation was detected in 2 experiments (Supplementary Fig S1).
GSTπ has to bind to its target protein to facilitate glutathionylation [21] and we examined if oxidative stress induced by a physiological stimulus increases association between the β 1 Na + /K + ATPase subunit and GSTπ. Exposure of isolated cardiac myocytes to angiotensin II to activate NADPH oxidase [22] appeared to increase co-immunoprecipitation of GSTπ with the β1 subunit in 4 experiments (Supplementary Fig S2).
The tumor suppressor p53 promotes both pro-and anti-apoptotic signaling in response to treatmentinduced DNA damage and oxidative stress [23]. We examined abundance of p53 after exposure of cells to FXYD3-pep. Exposure of MCF-7 and BxPC-3 cells to 1 µM FXYD3-pep for 24 hours had no effect in either cell line. We also exposed BxPC-3 cells to 0.5 µM Dox and MCF-7 cells to 2.5 µM Dox for 48 hours, with or without exposure to 1 µM FXYD3-pep. When used, FXYD3-pep was added for 24 hours before cells were also exposed to Dox. The different concentrations of Dox were chosen to re ect differences in sensitivity to cytotoxicity of Dox with exposure to 1 µM FXYD3-pep (Fig. 4). MCF-7 cells expressed low levels of p53 at baseline. Exposure to Dox alone increased expression of p53 for both cell lines and this was augmented with exposure to FXYD3-pep (Fig. 6E).
GSTπ inhibits apoptotic signalling induced by oxidative stress and the reduced expression of GSTπ FXYD3-pep causes (Fig. 6A -C) might augment apoptotic signalling. We examined effects of FXYD3-pep on caspase 3/7 activation with and without the oxidative stress caused by Dox exposure. Exposure to 1 µM FXYD3-pep for 24 hours alone had no effect on Caspase 3/7 activity in cytosolic extracts of MCF-7 and BxPC-3 cells but pre-exposure of the cells to FXYD3-pep for 24 hours followed by 48 hours coexposure to FXYD3-pep and 0.5 µM Dox for BxPC-3 cells or 2.5 µM Dox for MCF-7 cells augmented a Dox-induced increase in Caspase 3/7 activity (Fig. 6F). The net effect of FXYD3-pep is to augment Doxinduced apoptotic signaling.
In vivo uptake of FXYD3-pep Displacement of WT FXYD proteins from Na + /K + -ATPase in non-cancer cells might cause toxicity that would preclude use of a therapeutic peptide developed from FXYD-pep. In preliminary studies we examined if acute adverse effects were detectable. Fluorescence monitored in mice after intraperitoneal injection TRITC-labeled FXYD-pep indicated systemic uptake, independently con rmed by uorescence microscopy of the myocardium ( Supplementary Fig S3). There were no apparent adverse effects of the peptide.

Discussion
Expression of the large Na + /K + -ATPase α subunits was much higher for MCF-7 breast cancer cells than MDA-MB-468 cells but electrogenic plasmalemmal Na + /K + -ATPase currents were similar suggesting the presence of a pool of Na + /K + -ATPase in MCF-7 cells that does not contribute to plasmalemmal Na + /K +transport. In agreement with this Na + /K + -ATPase α subunits are strongly expressed in the intracellular compartment of MCF-7 cells [3]. Intracellular distribution of overexpressed α subunits also occurs in cancer cells of other tissues [24,25]. We are not aware of reports of parallel intracellular expression of α/ β subunits to implicate expression of functional Na + /K + -ATPase. However, ouabain-sensitive 86 Rb + uptake mediated by Na + /K + -ATPase in the plasmalemma does not re ect variable β1 subunit expression in cancer cells [26]. This implicates the existence of non-plasmalemmal pools of β1 subunits and is in turn consistent with non-plasmalemmal intracellular pools of intact α/β complexes.
Intracellular TRITC-labeled FXYD3-pep uorescence in BxPC-3 re ects the distribution of overexpressed FXYD3 protein in resected sections of pancreatic- [4] and breast cancers [3] and the strongest uorescence in the perinuclear region is similar to the distribution of overexpressed FXYD3 and Na + /K + -ATPase α subunits in MCF-7 cells [3]. Binding of TRITC-labeled FXYD3-pep to overexpressed α subunits in BxPC-3 cells, if in the same distribution, would account for the pattern of TRITC uorescence here.
Fluorescence from a 2-hour exposure to TRITC-labeled FXYD3-pep was retained 24 hours after the peptide was washed off BxPC-3cells and since the peptide is membrane-permeable it was expected to exit cells unless bound. Consistent with this, in MCF-7 cells there is no WT FXYD3 protein that might have been freely diffusible in the cytosolic fraction while the membranous fraction contains FXYD3 in easily detectable amounts [3].
Experimentally induced synthesis of FXYD3 in non-cancer Chinese Hamster Ovary (CHO) cells suggest how enhanced synthesis of FXYD3 might drive accumulation of Na + /K + -ATPase in intracellular membranes. FXYD3 expressed in the CHO cells was distributed in the nuclear envelope and the endoplasmic reticulum (ER) but not in the plasmalemma [27]. The FXYD3 should facilitate local assembly of the α-with β Na + /K + -ATPase subunits because FXYD proteins have bonds to both [28]. Increased α/β subunit co-immunoprecipitation with induced expression of FXYD1 in Xenopus oocytes or exposure of cardiac myocytes to recombinant WT FXYD3 [6] show how abundance of FXYD proteins facilitate Na + /K + -ATPase assembly. Epitopes of FXYD3 with bonds to Na + /K + -ATPase are preserved in FXYD3-pep and these should retain the peptide intracellularly, bound to overexpressed Na + /K + -ATPase.
While functional α/β Na + /K + -ATPase is often taken to exist only in the plasmalemma of non-cancer cells it is also found in the nuclear envelope where in human embryonic kidney 293 cells it maintains a transmembrane concentration gradient for Na + . This allows Na + -Ca 2+ exchange-mediated Ca 2+ e ux from the nucleoplasm into the nucleoplasmic reticulum [29]. Ca 2+ overload of the nucleoplasm in a variety of cells directly affects chromatin organization and activates cell death signaling and apoptosis [30].
A Na + /K + -ATPase inhibitor might cause Ca 2+ overload of the nucleoplasm but here the selective inhibitor ouabain did not augment Dox's cytotoxicity. However, ouabain is poorly membrane-permeable and might not traverse the outer membrane of the nuclear envelope to access Na + /K + -ATPase in the inner membrane. In support of this it had no effect on ATP-activated Na + e ux or Na + -Ca 2+ exchange-coupled Ca 2+ e ux in cell nuclei that in contrast was almost eliminated by the membrane-permeable Na + /K + -ATPase inhibitor strophanthidin [29].
Membrane permeability of FXYD3-pep allows it to reach Na + /K + -ATPase in intracellular membranes and displacement of WT FXYD3 is expected to amplify β1 subunit glutathionylation and inhibition of Na + /K + -ATPase with Dox-induced oxidative stress. Such inhibition should alter Ca 2+ levels in intracellular compartments and, in the ER, an increase in these levels is implicated in Dox-induced apoptosis [31].
However, to our knowledge, functional Na + /K + -ATPase has not been found in ER membranes of cancer-or non-cancer cells.
Co-localization of FXYD3 with Na + /K + -ATPase α subunits in MCF-7 cells in an intracellular distribution [3] similar to that for co-localization of FXYD3 with the tyrosine kinase Src [32] implicates a mechanism independent of Na + /K + -ATPase inhibition in FXYD3-pep's augmentation of cytotoxicity. Src's SH domain interacts with FXYD3, the same domain that also binds Src to the second cytosolic domain of the Na + /K + -ATPase α1 subunit [33]. The Src family of tyrosine kinases is redox regulated [34] and the reactive Cys in FXYD3 (Fig. 1) might optimize redox balance. FXYD3 expression is promoted by the stem cell-related transcription factor SOX9 and FXYD3 in turn is critical for con ning SOX9 to the nucleus. Knock down of FXYD3 with shRNA reduces the stem cell population and markedly impairs growth of MCF-7 xenograft tumors in mice. In this scheme of a SOS9/FXYD3/Src axis [32], Na + /K + -ATPase α subunits serve to anchor FXYD3 intracellularly rather than that of a functional target for FXYD3-pep.
FXYD3-pep, full-length cysteine-mutated FXYD3 or FXYD3 siRNA almost eliminated expression of GSTπ in BxPC-3 cells. GSTπ is expressed in the cytosol, nucleus [35] and ER [36] and is widely implicated in drug resistance of cancers. The resistance has been attributed to GSTπ aiding in the export of toxins from cells, including export of chemotherapeutics. [21]. Alternative mechanisms have been proposed, including a role for GSTπ in inducing glutathionylation of proteins [21].
While glutathionylation of the β1 subunit inhibits Na + /K + -ATPase [37], glutathionylation can also protect protein cysteine residues against irreversible oxidation, preserving proteins in a reversible functionally silent state. GSTπ promotes such protection [38]. Binding of GSH to GSTπ lowers GSH's cysteine thiol pKa, forming a nucleophilic thiolate anion that GSTπ can deliver to cysteine residues in hydrophobic domains. The susceptible cysteine residue in the β1 Na + /K + -ATPase subunit is in such a domain [37] and results here suggest GSTπ can catalyze its glutathionylation.
β subunits are N-terminal glycosylated membrane proteins and MDA-MB-468 and Panc-1 cells share the p53 R273H mutation that promotes folding of such proteins in the ER [39]. The normal folding of β subunits is important for its assembly with α subunits. However, it is not apparent if this somehow relates to the low-or absent expression of FXYD3 in these cells, contrasting FXYD3 expression in the MCF-7 and BxPC-3 cells.
We are not aware of direct interdependence of FXYD3 and GSTπ that might account for FXYD3-pep's down-regulation of GSTπ but note synthesis of both FXYD3 [11] and GSTπ [40] depend on p53. Exposure to FXYD3-pep alone had no effect on p53 abundance, yet it markedly decreased GSTπ expression.
Interdependence of FXYD3 and GSTπ via common p53-dependent synthesis cannot be implicated from our data.

Conclusions
We have for the rst time established a link between overexpression of FXYD3 and GSTπ. GSTπ is a recognized target for treatment resistance and many GSTπ inhibitors are under development [41]. Many peptide drugs are already in clinical use or under development [42] and we propose the membranepermeable FXYD-pep is a promising platform for development into a drug that selectively sensitizes cancers overexpressing cytosolic FXYD3 to chemotherapeutics, in particular when combined with a GSTπ inhibitor. Targeting FXYD3 might also address the treatment resistance that cancer stem cells confer, at least in some breast cancers [32].

Declarations
Ethics approval and consent to participate Ethics approval was not required as studies were conducted on cell lines only.

Consent for publication
Not applicable Availability of data and materials   expression. Cells were exposed to 1 μM Dox or Dox-free culture medium for 48 hours after transfection as indicated (N = 3). D. FXYD3 siRNA transfection and effect of Dox on cell viability and. Cells were exposed to Dox or Dox-free culture medium as for panel C (N = 5). E. FXYD3 siRNA transfection and effect of Dox on Caspase-3/7 activity. Cells were exposed to Dox or Dox-free culture medium as for panels C and D (N = 5).  pancreatic cancer cells. Cells were exposed to FXYD3-pep labeled with TRITC (red) for 2 hours and uorescence microscopy performed 24 hours after the peptide was washed off. DAPI was used to