Protein Phosphatase 1c Associated with the Cardiac Sodium Calcium Exchanger 1 Regulates Its Activity by Dephosphorylating Serine 68-phosphorylated Phospholemman*

The sodium (Na+)-calcium (Ca2+) exchanger 1 (NCX1) is an important regulator of intracellular Ca2+ homeostasis. Serine 68-phosphorylated phospholemman (pSer-68-PLM) inhibits NCX1 activity. In the context of Na+/K+-ATPase (NKA) regulation, pSer-68-PLM is dephosphorylated by protein phosphatase 1 (PP1). PP1 also associates with NCX1; however, the molecular basis of this association is unknown. In this study, we aimed to analyze the mechanisms of PP1 targeting to the NCX1-pSer-68-PLM complex and hypothesized that a direct and functional NCX1-PP1 interaction is a prerequisite for pSer-68-PLM dephosphorylation. Using a variety of molecular techniques, we show that PP1 catalytic subunit (PP1c) co-localized, co-fractionated, and co-immunoprecipitated with NCX1 in rat cardiomyocytes, left ventricle lysates, and HEK293 cells. Bioinformatic analysis, immunoprecipitations, mutagenesis, pulldown experiments, and peptide arrays constrained PP1c anchoring to the K(I/V)FF motif in the first Ca2+ binding domain (CBD) 1 in NCX1. This binding site is also partially in agreement with the extended PP1-binding motif K(V/I)FF-X5–8Φ1Φ2-X8–9-R. The cytosolic loop of NCX1, containing the K(I/V)FF motif, had no effect on PP1 activity in an in vitro assay. Dephosphorylation of pSer-68-PLM in HEK293 cells was not observed when NCX1 was absent, when the K(I/V)FF motif was mutated, or when the PLM- and PP1c-binding sites were separated (mimicking calpain cleavage of NCX1). Co-expression of PLM and NCX1 inhibited NCX1 current (both modes). Moreover, co-expression of PLM with NCX1(F407P) (mutated K(I/V)FF motif) resulted in the current being completely abolished. In conclusion, NCX1 is a substrate-specifying PP1c regulator protein, indirectly regulating NCX1 activity through pSer-68-PLM dephosphorylation.

NCX1 has been shown to exist in a macromolecular complex comprising protein kinase A (PKA) and C (PKC), protein phosphatase 1 (PP1) and 2A (PP2A), as well as other anchoring and adaptor proteins (17). Direct regulation of NCX1 by kinases is controversial (18); in fact, we have shown that there is no phosphorylation of endogenous NCX1 following PKA activation, and we concluded that the identified phosphorylation site was not accessible in full-length NCX1 (19). It has been suggested that phosphorylation and dephosphorylation rather occur on accessory proteins in the NCX1 macromolecular complex, enabling fine-tuning of signals that converge on NCX1 (17). Accumulating data indicate that PLM is one of these regulatory players. This 72-amino acid transmembrane protein, belonging to the FXYD1 family of ion transporters (20), co-localizes and coimmunoprecipitates with NCX1 and has been shown to inhibit NCX1 activity when it is phosphorylated at serine 68 (pSer-68-PLM) (21)(22)(23)(24). Interestingly, pSer-68-PLM relieves inhibition of the Na ϩ /K ϩ -ATPase (NKA), causing an increase in NKA activity (25,26), suggesting that PLM may serve as a regulator of both NCX (24) and NKA (27) depending on its phosphorylation status. pSer-68-PLM is in turn regulated by PP1 (28). The latter is a ubiquitously expressed ϳ38.5-kDa serine/threonine phosphatase that counters the effects of serine/threonine kinases and has little intrinsic specificity for its substrates (29). Mammalian genomes encode four distinct catalytic subunits of PP1 as follows: PP1␣, PP1␤/␦, and the splice variants PP1c␥1 and PP1c␥2 (30). The isoforms show 85% sequence identity, but the N-and C-terminal extremities show amino acid differences (30). The catalytic subunit (PP1c) achieves substrate specificity by forming holoenzymes with more than 200 targeting proteins (31). These targeting proteins localize PP1c to specific subcellular domains and fine-tune its activity, allowing for substratespecific effects (31). More than 90% of targeting proteins interact with PP1c via a short degenerate RVXF-docking motif (32), which serves as an anchor for the initial recruitment of PP1c. Although dephosphorylation of pSer-68-PLM, by PP1c, was recently shown to modulate pSer-68-PLM regulation of NKA (28), it is unknown whether this mechanism is involved in NCX1 regulation.
In this study, we investigated whether a direct and functional NCX1-PP1c interaction is a prerequisite for pSer-68-PLM dephosphorylation. Bioinformatic and biomolecular approaches were used to investigate this hypothesis. We aimed to map the PP1ctargeting site on the NCX1 macromolecular complex and elucidate the biological effect of this interaction on NCX1 activity.

Experimental Procedures
Animal Model-Animal experiments were approved by the Norwegian Animal Research Committee (FOTS ID 3820) and conformed to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication No. 85- 23, revised 1996). Male Wistar rats (Møllergaard Breeding and Research Center, Skensved, Denmark) weighing ϳ170 g were subjected to aortic banding, as described previously (33,34). In short, anesthesia was induced in a chamber containing a mixture of 68% N 2 O, 28% O 2 , and 4% isoflurane. After endotracheal intubation, ventilation was performed with a respirator (Zoovent, Triumph Technical Services, Milton Keynes, UK), and anesthesia was maintained by administration of a mixture of 69% N 2 O, 29% O 2 , and 2% isoflurane. The chest was opened in the right second intercostal space, and the ascending aorta was carefully dissected. A significant stenosis was induced by a tight banding of the ascending aorta with a 3.0 silk suture. In sham-operated animals (SHAM), the silk suture around the ascending aorta was not tightened. Buprenorphine was administered for postoperative analgesia. After 6 weeks, echocardiography was performed with a Vevo 2100 (Fujifilm VisualSonics, Canada), and short and long axis images of the left ventricle (LV) and atrium were obtained. The flow through the mitral and aortic valve was measured. Aortic banding animals with congestive heart failure (ABHF) were selected based on echocardiographic, hemodynamic, and post-mortem analysis. The criteria for inclusion in the ABHF group were increased posterior wall diameter (Ͼ1.9 mm), increased LV weight (Ͼ0.75 g), increased lung weight (Ͼ2.5 g), and increased left atrial diameter (Ͼ5.0 mm) (34). The SHAM group served as control. All LV samples were snap-frozen in liquid nitrogen and stored at Ϫ70°C until analysis.
Adult Cardiomyocyte Isolation-Rats were anesthetized in a chamber filled with 95% room air and 5% isoflurane (Abbott Scandinavia Ab, Solna, Sweden) and were subsequently sacrificed by cervical dislocation. Hearts were then quickly excised and placed in ice-cold 0.15 M NaCl solution with heparin (Heparin LEO, 5000 IE/ml; Orifarm AS, Norway). The aorta was then cannulated and retrogradely perfused with a cell isolation buffer containing 130 mM NaCl, 25 mM Hepes, 22 mM D-glucose, 5.4 mM KCl, 0.5 mM MgCl 2 , 0.4 mM NaH 2 PO 4 , pH 7.4 (all chemicals from Sigma) to wash out the blood. The heart was thereafter perfused with cell isolation buffer containing 200 units/ml collagenase type II (Worthington) and 0.1 mM Ca 2ϩ . After 20 min of perfusion, the heart was cut down, and the atria and right ventricle were removed. The LV was minced and gently shaken at 37°C for 3-4 min in the same solution used in the perfusion, but with addition of 1% bovine serum albumin (BSA) (A9647, Sigma) and 0.02 units/ml deoxyribonuclease I (LS002006, Worthington). The digested ventricular tissue was then filtered (200-m nylon mesh), and cardiomyocytes were sedimented. The cardiomyocyte pellet was resuspended in cell isolation buffer with 1% BSA (A9647, Sigma) and 0.1 mM Ca 2ϩ in the solution. Isolated cardiomyocytes were kept at room temperature until used.
Total Protein Extracts-Frozen rat LV lysates were pulverized in a mortar with liquid nitrogen before transfer to lysis buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton TM X-100 (X100 -500ML, Sigma)) supplemented with 1 mM PMSF (93482, Sigma) and a Complete Mini EDTA-free tablet (11836170001, Roche Diagnostics, Basel, Switzerland). Tissue samples were homogenized three times for 1 min on ice with a Polytron 1200 and centrifuged at 100,000 ϫ g for 60 min at 4°C. Supernatants were collected and stored at Ϫ70°C until analysis.
Fractionation-LVs from rat heart and neonatal cardiomyocyte cultures were fractionated using the compartment protein extraction kit according to the manufacturer's protocol (2145, Merck Millipore Billerica, MA). The supernatants from cytosolic and/or membrane fractions were collected and stored at Ϫ70°C. Protein concentrations, where applicable, was determined by the Micro BCA protein assay kit (23235, Thermo Fisher Scientific).
Transient Transfection of HEK293 Cells-Human embryonic kidney 293 (HEK293) cells were cultured in DMEM (41965-039, Gibco, Life Technologies, Inc.) supplemented with 10% FBS (14-701F, BioWhittaker), 1% non-essential amino acids (10370-021, Gibco, Life Technologies, Inc.), 100 units/ml penicillin, and 0.1 mg/ml streptomycin (penicillin/streptomycin, P4333, Sigma) and maintained in a 37°C, 5% CO 2 -humidified incubator. Cells were cultured in medium without antibiotics 24 h prior to transfection. Cells were transfected with DNA using Lipofectamine 2000, as instructed by the manufacturer (11668-019, Invitrogen), or a solution containing CaCl 2 . For this method, two solutions were prepared: solution A (124 mM CaCl 2 and 8 g of plasmid DNA diluted in 500 l of purified water) and solution B (50 mM HEPES, 280 mM NaCl, and 1.5 mM Na 2 HPO 4 in 500 l of PBS, pH 7.0). The two solutions were then mixed together and incubated at room temperature for 30 min. After the 30 min, HEK293 cells were incubated with the mixed solution for 18 -24 h (38). After 24 h, the cells were either transferred to coverslips for patch clamp/Ca 2ϩ -dependent fluorescence experiments or lysed in immunoprecipitation (IP)buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Triton TM X-100) supplemented with a Complete Mini EDTAfree tablet (11836170001, Roche Diagnostics) and used in molecular studies. In the patch clamp/Ca 2ϩ -dependent fluorescence experiments, the glass coverslips were pre-coated with poly-L-lysine (P4707, Sigma), and the cells on the coverslips were incubated for 24 h prior to analysis. GFP was used as a positive control for the transfection.
Overlay Assay-Peptide array membranes, consisting of either the NCX1 or PP1c sequence, were activated by soaking in methanol for a few seconds and followed by three washes (three times for 10 min) with TBS-T (Tris-buffered saline with 1% Tween 20 (161-0781, Bio-Rad)). The membranes were then incubated with blocking solution (1% casein) (11921673001, Roche Diagnostics) at room temperature for 1 h. Next, the membranes were either overlaid with 1 g/ml recombinant His 6 -TF-NCX1 cyt or 1 g/ml GST-PP1␣ recombinant protein (in 1% casein) overnight at 4°C with gentle agitation. The membranes were then washed five times for 5 min with TBS-T and incubated with anti-His 6 HRP or anti-GST HRP-conjugated antibody for 1 h at room temperature. The bound protein on the membranes was detected by immunodetection. Incubation of the membranes without recombinant protein was used as a negative control. For antibody epitope mapping, incubation of the membranes without primary antibody was used as a negative control.
Pulldown Experiments-Biotinylated peptides (10 mM) were incubated with 25 l of anti-biotin-conjugated beads (A1559, Sigma) and 100 l of PBS for 2 h at 4°C under gentle rotation. The beads with bound peptides were then washed three times with PBS followed by incubation with 1 g of GST-PP1␣ recombinant protein in 100 ml of IP-buffer supplemented with 1% BSA (805095, Bio-Rad) followed by gentle rotation for 2 h at 4°C. The peptide complex was washed three times with 1 ml of IP-buffer. To elute the complex, it was boiled in 2ϫ SDS loading buffer and analyzed by immunoblotting.
Immunoprecipitation (IP)-IP was performed using 2 g of the appropriate antibody. The immunocomplexes were collected by 50 l of protein A/G PLUS-agarose beads (sc-2003, Santa Cruz Biotechnology) overnight at 4°C. They were then washed three times with 1 ml of IP-buffer followed by the boiling of the complexes in 2ϫ SDS loading buffer. Analysis was done by immunoblotting. Equal amounts of rabbit IgG (sc-2027, Santa Cruz Biotechnology) and mouse IgG (sc-2025, Santa Cruz Biotechnology) were used as negative controls. Blocking peptide (antigen, cardiac NCX1; sequence, (C)-GQPVFRKVHARDHPIPST) (Genscript Corp.) was incubated with anti-NCX1 prior to immunoprecipitation (negative control).
Proximity Ligation Assay (Duolink)-Isolated adult rat cardiomyocytes were plated on laminin (23017-015, Life Technologies, Inc.)-coated glass coverslips and left to adhere for 1 h. The cells were fixed in 4% paraformaldehyde (158127, Sigma), permeabilized with 0.03% Triton X-100 (X100 -500ML, Sigma), and incubated with the two primary antibodies anti-NCX1 (rabbit polyclonal antibody) and anti-PP1c (mouse monoclonal antibody) overnight at 4°C. Staining without primary antibodies and use of an anti-NCX1 blocking peptide was used as a negative control. The proximity ligation assay was performed using the Duolink kit (DUO92014, Sigma) according to the manufacturer's protocol, as described previously (15). The cells were then incubated with 600 nM SYTOX Orange (S-11368, Life Technologies, Inc.), a nucleic acid stain, for 10 min at room temperature and rinsed three times for 5 min with PBS. Imaging experiments were performed at 25°C. The waterbased Duolink In Situ Mounting Medium (refractive index 1.44), provided in the kit, was used to mount the glass coverslip to the glass slide. The fluorochromes used were the Duolink In Situ Detection Reagents Green (excitation 488 nm and emission 510 nm) and SYTOX Orange (excitation 543 nm and emission 650 nm). The cells were visualized with an inverted LSM 710 confocal microscope (Zeiss GmbH, Jena, Germany) equipped with an LD C-Apochromat ϫ40 objective (numerical aperture 1.1). Sequential optical scans were acquired using the Zeiss ZEN imaging software.
Immunoblotting-LV lysates and immunoprecipitates were resolved on either 4 -15% or 15% Criterion TM Tris-HCl precast gels (Bio-Rad) and blotted onto PVDF membranes (RPN 303F, GE Healthcare). The PVDF membranes and peptide arrays were blocked in 5% nonfat dry milk or 1% casein in TBS-T for 60 min at room temperature, followed by incubation with primary antibodies either for 1 h at room temperature or overnight at 4°C. The membranes were washed three to five times for 5 min in TBS-T and incubated with a horseradish peroxidase-conjugated secondary antibody. Blots were developed using ECL Plus or Prime (RPN 2132 or RPN 2232, GE Healthcare). The chemiluminescence signals were detected by Las-4000 (Fujifilm, Tokyo, Japan). Membranes were re-probed after stripping using the Restore Western blot Stripping buffer (21059, Thermo Scientific).
Protein Phosphatase 1 Activity Assay-Phosphatase activity was measured in vitro by a nonradioactive assay using a malachite green-based phosphatase assay kit (40) (17-128, Merck Millipore), according to the manufacturer's protocol. Briefly, 1 unit of PP1␣ recombinant protein (P0754S, New England Biolabs) was incubated with 200 M (RRApSVA) peptide substrate together with either a range of His 6 -TF-NCX1 cyt recombinant protein concentrations or biotinylated NCX1 peptides in a phosphatase buffer (50 mM HEPES, 10 mM NaCl, 2 mM DTT, 0.01% Brij 35, pH 7.5) supplemented with 1 mM MnCl 2 for 20 min at 37°C. Next, 100 ml of malachite green dye was added to the solution and incubated for 15 min at room temperature. A spectrophotometer, absorbance set at 650 nm, was used to measure the colored complex formed upon the release of free phosphate. The amount of released phosphate was calculated using a standard curve created with various concentrations of KH 2 PO 4 . PP1 inhibitor 2 was used as a control for the assay.
Protein Kinase A (PKA) Phosphorylation Assay-PKA assays were performed using cAMP-dependent protein kinase, catalytic subunit (PKA-C) (P6000S, New England Biolabs) as instructed by the manufacturer. PKA-C and 1ϫ NEBuffer for protein kinases (P6000S, New England Biolabs), supplemented with 200 M ATP, were added to the HEK293 cell lysates, fol-lowed by incubation for 30 min at 30°C. PKA-C treatment was omitted for control samples. Kinase reaction was stopped by adding sample buffer. The samples were subsequently subjected to SDS-PAGE and immunoblotting.
Structure Modeling-Atomic coordinates for human PP1 in complex with the PP1 and PDZ domains of rat spinophilin (PDB code 3EGG (41)) were used to build a model of the NCX1 tetrapeptide 405 KVFF 408 binding to PP1. The peptide was built by homology modeling using the 448 KIHF 451 segment of the spinophilin as a template. The side chain conformations of the middle Val and Phe residues were selected based on the orientation of the corresponding Ile and His residues in spinophilin, respectively. The structural models of the complexes were analyzed and visualized by PyMOL (Schrodinger LLC).
Densitometric Analysis-Densitometric analysis was performed using ImageJ (National Institutes of Health), and blots were processed in Adobe Photoshop CS2 (Adobe Systems Inc., San Jose, CA).
Statistics-All data were expressed as mean Ϯ S.E. relative to control. Comparisons between two groups were analyzed using the unpaired t test or Mann-Whitney test (GraphPad software, Prism 5.04). A p value of Ͻ0.05 was considered statistically significant.

Results
Bioinformatic Analysis to Identify PP1-binding Sites in NCX1-The Protein Pattern Find-Bioinformatics.org database was used to search for putative PP1-binding motifs in NCX1 and PLM, with the consensus sequence (R/K)(V/I)X(F/W) (see under "Experimental Procedures" for details). No PP1-binding sites were found in PLM; however, three putative binding sites were identified in human, mouse, and rat NCX1. The first site, RVFF, is localized within the cytosolic loop connecting TM3 and -4. The second site, KIFF/KVFF (human and mouse/rat), is localized in the large cytosolic loop in the first Ca 2ϩ binding domain (CBD1), the primary sensor of calcium (42). The KIFF/ KVFF site does not overlap with the sites in CBD1 that coordinate the four Ca 2ϩ ions (43). The third site, KVLF, is localized at the end of the intracellular loop (Fig. 1).
PP1 Binds Directly to NCX1-Several experiments were performed to show that PP1 interacts with NCX1. First, immunodetection with specific antibodies for NCX1 (as published previously (19)) and PP1 ( Fig. 2A, antibody epitope mapping) showed that a pool of PP1 catalytic subunit (PP1c) is localized with NCX1 in membrane fractions isolated from rat LV (Fig.  2B, left panels) and neonatal cardiomyocytes (Fig. 2B, right panels). Second, immunoprecipitation of NCX1 in LV lysate and immunoblotting with antibodies against PP1c and PLM identified a NCX1-PLM-PP1c macromolecular complex (Fig. 2C). Third, using the proximity ligation assay, we showed that NCX1 and PP1c co-localize in isolated adult rat cardiomyocytes, strongly indicating that NCX1 and PP1c are in the same molecular complex (as indicated by the green fluorescent spots in Fig. 2D, 1st panel). No signal was observed when primary antibodies were omitted (Fig. 2D, 2nd panel) or when the NCX1 antibody was pre-incubated with a specific anti-NCX1-blocking peptide (Fig. 2D, 3rd panel). Fourth, immunoprecipitation of recombinant His 6 -TF-NCX1 cyt (Fig. 2E, schematic figure) using His 6 or NCX1 antibodies showed that recombinant PP1␣ precipitated with His 6 -TF-NCX1 cyt (Fig. 2F), suggesting NCX1 and PP1␣ to be direct binding partners. Fifth, the binding kinetics of the GST-PP1␣-NCX1 interaction was analyzed by SPR. A range of concentrations of recombinant GST-PP1␣ was injected over immobilized recombinant His 6 -TF-NCX1 cyt on a CM5 chip and analyzed with the fit of a 1:1 interaction model (Langmuir). The dissociation equilibrium constant (K D ) for the interaction was 3.02 Ϯ 1.48 nM, with an association rate constant (k a ) ϭ (6.10 Ϯ 3.35) ϫ 10 4 M Ϫ1 s Ϫ1 and a dissociation rate (Fig. 2G). Conclusively, SPR data indicated that the GST-PP1␣-NCX1 interaction was strong and stable.
PP1 Anchors to the KIFF/KVFF Motif in NCX1-CBD1-To identify PP1 binding in NCX1, mouse NCX1(FL)-GFP and a series of GFP-NCX1 deletion mutants were expressed in HEK293 cells (Fig. 3A, schematic figure). Fishing with anti-PP1c and recombinant GST-PP1␣ in the HEK293 lysates and immunoblotting with anti-GFP revealed that all NCX1 variants precipitated, except for GFP-NCX1(243-402) (Fig. 3B), which was lacking CBD1 containing the KIFF/KVFF motif. The bottom panel shows expression of the NCX1 variants and GFP (negative control) in HEK293 lysates (Fig. 3B). To map the PP1-binding site more precisely, the cytosolic loop of rat NCX1 was synthesized as 20-mer overlapping peptides on a membrane and incubated with recombinant GST-PP1␣, followed by anti-GST HRP incubation and immunodetection. Interestingly, GST-PP1␣ bound to 399 ENDPVSKVFFEQGTYQCLENCGT-VALTII 427 containing the KVFF motif (Fig. 3C, boxed sequence). Notably, some GST-PP1␣ binding was also observed to amino acids C-terminal to the KVFF motif ( 408 FEQGTYQ-CLENCGTVALTII 427 ) (Fig. 3C, underlined amino acids). Introducing Gly sequentially to the C terminus showed that GST-PP1␣ binding was lost when the asparagine (Asn-418) was deleted (Fig. 3D), suggesting additional amino acid-binding sites C-terminal to the KVFF motif to be a requisite for PP1 binding. Noteworthy, no GST-PP1␣ binding was observed for the KVLF-containing sequence, which has previously been suggested to be a putative PP1-binding site (Fig. 3C, upper panel,  lower box) (17).
Several mutation strategies were employed to test the effect of PP1 binding to the KIFF/KVFF mouse/rat, motif. Ala substitution at the second and fourth position in the KIFF/KVFF motif (KAFA mutant) has been shown to be an effective mutation strategy, as it disrupts PP1 binding (44 -46). However, upon examination of the Ca 2ϩ -bound CBD1 crystal structure of dog NCX1 (PDB code 2FWS), we noted that the Ile and Phe in the second and fourth position, localized in the ␣-1 strand in ␤-sheet 1, seem to be buried in the CBD1 structure. Thus, it seemed more sensible to mutate the Lys, at the first position, and Phe, at the third position (AIAF mutant), as these are the accessible residues. We also generated a Pro substitution mutant (F407P), and in a fourth construct the KIFF motif was deleted (GFP-NCX1(⌬399 -424)). All variants were expressed in HEK293 cells and immunoprecipitated with anti-NCX1. PP1c co-precipitated with NCX1(FL)-GFP but not with GFP-NCX1(I406A,F408A) (Fig. 4A), GFP-NCX1(K405A,F407A) (Fig. 4B), GFP-NCX1(F407P) or the deletion, GFP-NCX1-(⌬399 -424) (Fig. 4C). Interestingly, more PP1c co-precipitated with NCX1 in the presence of 3 mM CaCl 2 (Fig. 4D). In a final experiment, the rat peptide sequence 402 PVSKVFFEQGTYQ-CLENCGT 421 (identified in Fig. 3C) and corresponding sequences with KVFF substituted for KAFA or AVAF were overlaid with GST-PP1␣. Strong binding was observed for the native peptide, but not to the KAFA or AVAF mutations (Fig.  4E). Altogether, the data supports that PP1␣ anchors to the KIFF/KVFF motif in NCX1.
NCX1-KIFF/KVFF Is an Independent, Non-regulatory PP1canchoring Site-To confirm NCX1-KIFF/KVFF as an independent PP1c-binding site, we generated biotinylated NCX1-KVFF, NCX1-KAFA, and NCX1-AVAF peptides (Fig. 5A,  schematic figure). Pulldown experiments using anti-biotin-coupled agarose beads showed that GST-PP1␣ precipitated with biotin-NCX1-KVFF but not biotin-NCX1-KAFA (Fig. 5B) or biotin-NCX1-AVAF (data not shown). When the NCX1-KVFF peptide or His 6 -TF-NCX1 cyt recombinant protein was tested in an in vitro PP1 activity assay, no change in recombinant PP1 activity was observed (Fig. 5, C and D, respectively), indicating that the NCX1-KIFF/KVFF motif is a non-regulatory PP1c-anchoring site. Sequence alignments show that the NCX1-KIFF/ KVFF PP1c-anchoring site is highly conserved across human, rat, and mouse NCX1 (Fig. 5E) and that NCX2 and NCX3 harbor RIFF and KVFF motifs, respectively (Fig. 5F). Therefore,  FIGURE 2. PP1c binds directly to NCX1. A, epitope mapping was performed by overlaying an array of immobilized overlapping 20-mer rat PP1-␣ (P62138) peptides with anti-PP1 (E-9) (sc-7482, top panel). Amino acids in bold constitute the core epitope, localized in the PP1 catalytic subunit (PP1c), and are relevant for anti-PP1c binding (n ϭ 2). Immunodetection without anti-PP1c was used as a negative control (bottom panel). B, NCX1 and PP1c were analyzed in cytoplasmic and membrane fractions isolated from rat LV and neonatal cardiomyocytes using anti-NCX1 and anti-PP1c (in triplicate). Vinculin and epidermal growth factor receptor (EGFR) were used as controls for cytoplasmic and membrane fractions, respectively. C, rat LV lysates were subjected to immunoprecipitation (IP) using anti-NCX1 (in triplicate). Immunoprecipitates (right panels) and lysate (input; positive control for the immunoblot, left panels) were immunoblotted with NCX1, PP1c, and PLM antibodies. As a negative control, NCX1 antibody pre-incubated with a specific anti-NCX1 blocking peptide or non-relevant rabbit IgG was used. D, confocal images of in situ proximity ligation assay. NCX1-PP1c co-localization was analyzed in adult cardiomyocytes using anti-NCX1 and anti-PP1c (left panel, interaction indicated by green spots, see "Experimental Procedures" for details). Incubation without primary antibodies (middle panel) and pre-incubating anti-NCX1 with blocking peptide (right panel) were used as negative controls. Nuclei were stained with SYTOX orange. E, schematic figure of recombinant proteins used in this study. The cytosolic loop of NCX1 was purified, and His 6 and trigger factor (TF) tags were added to the N terminus (His 6 -TF-NCX1 cyt , top panel). Because of the size of the tags, His 6 -TF-NCX1 cyt migrates as a 150-kDa protein. Commercially available PP1␣ was used in the initial analysis of the interaction (in F) after which rat PP1-␣ (P62138) with a GST tag (GST-PP1␣) was generated (bottom panel) and used in the remainder of the study. F, recombinant proteins were subjected to immunoprecipitation using anti-NCX1 and anti-His 6 (in triplicates). Immunoprecipitates (right panels) and proteins (input panels; positive control for the immunoblot) were immunoblotted with anti-NCX1 and anti-PP1c. As a negative control, non-relevant rabbit and mouse IgG were used. G, SPR analysis was performed by immobilizing recombinant His 6 -TF-NCX1 cyt (ligand) on a CM5 chip and measuring the response when injecting a range of concentrations of GST-PP1␣ (analyte) (n ϭ 3). Fig. 5A) were also generated and used in pulldown experiments. Using equal amounts of peptides, we found strong GST-PP1c binding to biotin-NCX1(399 -424), some binding to biotin-NCX3(392-412), and weak binding to biotin-NCX2(389 -410) (Fig. 5G).

biotinylated NCX isoform peptides (schematically illustrated in
PP1c Dephosphorylates pSer-68-PLM and Regulates NCX1 Activity through NCX1-KIFF/KVFF Anchoring-First, existence of the NCX1-PLM-PP1c macromolecular complex was confirmed in lysates from HEK293 cells co-transfected with NCX1(FL)-GFP and PLM and immunoblotting with antibodies against NCX1, PP1c, and PLM (Fig. 6A). Second, to determine whether NCX1 anchoring is required for dephosphorylation of pSer-68-PLM by PP1c, FLAG-His 6 -PP1c(1-330) and PLM were expressed with and without NCX1(FL)-GFP in HEK293. PLM has been reported to be 30 -40% phosphorylated by endogenous kinases in HEK293 (24,47), and immunodetection with a specific anti-pSer-68-PLM (Fig. 6B, antibody epitope mapping) showed that PLM was highly phosphorylated in HEK293 (Fig. 6C, lanes 4 -6, anti-pSer-68-PLM level). Importantly, the pSer-68-PLM/total PLM level was significantly decreased in PLM ϩ FLAG-His 6 -PP1c ϩ NCX1(FL)-GFP compared with PLM ϩ FLAG-His 6 -PP1c (0.4 Ϯ 0.03 versus 1.0 Ϯ 0.05). This suggests that NCX1 anchoring is a prerequisite for dephosphorylation of pSer-68-PLM by PP1c (Fig. 6C, lanes  7-9). Consistently, when the KIFF/KVFF motif was mutated full-length NCX1-GFP and a series of GFP-NCX1 deletion mutants that were generated to map the PP1c interaction site in NCX1. The numbering of NCX1 includes the N-terminal amino acid signal peptide sequence. B, GST-PP1␣ together with NCX1 deletion variants were subjected to immunoprecipitation (IP) using anti-PP1c (in triplicates). Precipitates were analyzed with immunoblotting. The asterisks in the top panel indicate the NCX1 dimer (64) and missing GFP-NCX1(243-402) fragment. Noteworthy, the GFP-NCX1(243-402) fragment was not visible when we re-ran the samples on a second gel with good separation around 50 kDa and used a light chain-specific antibody (data not shown). The middle panel shows anti-PP1c-precipitated GST-PP1␣. In the bottom panel the input lysates were immunoblotted to show the migration and corresponding molecular weight of NCX1-GFP and the deletion mutants. The asterisks in this panel indicate the NCX1 dimer, monomer, and migration of the GFP-NCX1(243-402) deletion fragment. C, identification of PP1 binding by overlaying GST-PP1␣ on 20-mer overlapping NCX1 peptides synthesized on membrane. Rat NCX1 protein (EDM02743) was used spanning amino acids 243-799. Binding was analyzed using anti-GST HRP (n ϭ 2) top panel. Boxed areas show KVFF and KVLF respectively. Underlined amino acids in the first boxed area represent the RVXF -PP1-binding motif and a putative ⌽ 1 ⌽ 2 motif, whereas amino acids in bold indicate the common sequence in the four peptide sequences. Incubation without GST-PP1␣ was used as a negative control (lower panel). D, one of the peptide sequences identified in C; 402 PVSKVFFEQGTYQCLENCGT 421 , was synthesized with a glycine spacer and was overlaid with GST-PP1␣ (right panel) to determine the effect on PP1 binding when EN, putative ⌽ 1 ⌽ 2 motif, was deleted (n ϭ 2). The peptide sequence without GST-PP1␣ served as negative control (left panel).
To exclude that the presence of NCX1 prevents PLM phosphorylation, lysate from FLAG-His 6 -PP1c(1-330) and PLM coexpressed with or without NCX1(FL)-GFP was treated with or without active PKA-C. Our results indicated that NCX1 did not block the pSer-68-PLM phosphorylation, as the increase in the pSer-68-PLM/total PLM level was similar between the two groups (Fig. 6E).
Finally, the biological role of endogenous PP1c on NCX1 activity through dephosphorylation of PLM was analyzed in HEK293 cells. First, the whole cell patch clamp technique was employed. NCX1 current was isolated by measuring Ni 2ϩ -sensitive current elicited during a slow voltage ramp from 120 to Ϫ100 mV. Co-expression of PLM with WT NCX1(FL) resulted in NCX1 inhibition (both forward and reverse mode), consistent with what was reported previously (Fig. 7F) (PLM is endogenously phosphorylated 30 -40% by kinases in HEK293) (22). Moreover, mutation of the PP1c-anchoring site in NCX1 (NCX1(F407P)) resulted in more NCX1 inhibition (Fig. 7F), indicating that endogenous PP1c was not able to dephosphorylate the pool of phosphorylated PLM. Noteworthy, the current of NCX1(F407P) alone was similar to that of WT NCX1(FL) (Fig. 7E). Current traces of the various transfections are given in Fig. 7, A-D.
NCX1 function was further examined by measuring Ca 2ϩdependent fluorescence in transfected HEK293 cells. Extracellular Na ϩ concentration was rapidly changed from 140 to 0 mM to induce Ca 2ϩ entry via reverse-mode NCX1 activity. In accordance with observations in the patch clamp experiments, NCX1-mediated Ca 2ϩ influx observed in WT NCX1-transfected cells was markedly reduced when PLM was co-expressed with WT NCX1(FL) (F/F 0 ϭ 1.6 Ϯ 0.08 versus 1.9 Ϯ 0.18). Mutation of the PP1c-anchoring site in NCX1 (NCX1(F407P)) resulted in a further decrease in the intracellular calcium level representing the proline substitution KIPF and GFP-NCX1(⌬399 -424) representing KIFF deletion (C) were generated and expressed in HEK293 cells after which they were used in immunoprecipitation (IP) experiments. The NCX1 antibody was used to precipitate NCX1 followed by immunoblotting with anti-PP1c for determination of PP1 association. D, rat LV lysates were subjected to immunoprecipitation using anti-NCX1, with and without the addition of 3 mM CaCl 2 . Immunoprecipitates (right panels) and lysate (input panels, positive controls for the immunoblot) were immunoblotted using relevant antibodies. E, peptide sequence 402 PVSKVFFEQGTYQCLENCGT 421 (identified in Fig. 3C) with KVFF motif was mutated to KAFA and AVAF and overlaid with GST-PP1␣ to determine the effect of mutations (n ϭ 2). GST-PP1␣ binding was analyzed by immunodetection with anti-GST HRP. All experiments were run in triplicates (A-D). FEBRUARY 26, 2016 • VOLUME 291 • NUMBER 9 when co-expressed with PLM (F/F 0 ϭ 1.2 Ϯ 0.05) (Fig. 8). Altogether, our data strongly suggest that NCX1-associated PP1c dephosphorylates pSer-68-PLM.
Structural Modeling of NCX1-KVFF into PP1-A homology model was generated of the rat NCX1 tetrapeptide 405 KVFF 408 , binding to rat/human PP1c using the 448 KIHF 451 PP1c binding segment of rat spinophilin (PDB code 3EGG) as template (Fig.  10A). It shows that Val-406 may fit into a hydrophobic pocket formed by Ile-169, Leu-243, and Leu-289 in PP1c and that Phe-407 may interact with Met-290. The peptide backbone forms hydrogen bonds with Asp-242, Leu-289, and Cys-291. Phe-257 and Arg-261 also play an important role in peptide binding. The conformation of the KVFF motif in this model is essentially similar to the conformation of the homologous KIFF motif as found in the NMR and x-ray structures of the Canis Lupus NCX1 ((PDB code 2FWS) (42) and (PDB code 2DPK (49)), the only difference being the orientation of the Lys side chain. However, large rearrangements in the NCX1 must take place upon binding to PP1c in order for the motif to be accessible for binding. To further test the model, Leu-243 and Phe-257 were single and double-mutated to Gly as follows: FLAG-His 6 -PP1c (L243G), FLAG-His 6 -PP1c (F275G), and FLAG-His 6 -PP1c (L243G, F275G). Immunoprecipitation showed that FLAG-His 6 -PP1c(1-330) co-precipitated His 6 -TF-NCX1 cyt , but no binding was observed for the PP1c mutants (Fig. 10B). This lends support to Leu-243 and Phe-257 in PP1c being important residues for NCX1 anchoring.
pSer-68-PLM-NCX1-PP1c Complex Is Increased in HF-To analyze the pSer-68-PLM-NCX1-PP1c levels in failing hearts, membrane fractions were isolated from SHAM and ABHF rats. Animal characteristics are listed in Table 1. Immunoblotting showed that the PP1c/calsequestrin level was significantly increased in ABHF hearts compared with SHAM (1.9 Ϯ 0.10 versus 1.0 Ϯ 0.03). The pSer-68-PLM/total PLM level was also significantly increased in ABHF hearts compared with SHAM (1.3 Ϯ 0.10 versus 1.0 Ϯ 0.03) (Fig. 11A). We have previously shown that the NCX1 level is up-regulated in response to pressure overload in our animal model (15), similar to what others have shown (5,50). Anti-calsequestrin and anti-total PLM were used as controls. Recently, we have identified a calpain cleavage site at Met-369 in NCX1, which resides between the PLM-anchoring site (16) and PP1c-anchoring site (this study). Wanichawan et al. (15) showed that calpain activity was greatly increased in ABHF hearts and that this correlated with increased cleavage of NCX1. To test whether calpain cleavage is responsible for dissociating PP1c away from PLM, HEK293 cells were co-transfected with PLM, FLAG-His 6 -PP1c(1-330), and NCX1(TEV). In NCX1(TEV), the calpain cleavage site is engineered into a TEV protease site. Interestingly, the pSer-68-PLM/total PLM level was significantly increased in PLM ϩ FLAG-His 6 -PP1c ϩ NCX1 (TEV) ϩ TEV/pCS2MT (HEK293 cells co-transfected with the enzyme that cleaves the TEV sequence) compared with PLM ϩ FLAG-His 6 -PP1c ϩ NCX1(TEV) (HEK293 cells are not co-transfected with the enzyme) (1.3 Ϯ 0.08 versus 1.0 Ϯ 0.05) (Fig. 11B), indicating that calpain cleavage of NCX1 at Met-369 separates the PLMand PP1c-anchoring sites. A model of PP1c anchored onto NCX1-KIFF/KVFF leading to dephosphorylation of pSer-68-PLM is summarized in Fig. 12, A and B (effect of calpain cleavage).

Discussion
In an ongoing effort to understand regulation of cardiac NCX1 activity, we have investigated the molecular basis of PP1c targeting to the NCX1 macromolecular complex, and we determined the functional consequences of this interaction.
PP1c Anchors to the NCX1-KIFF/KVFF Motif in CBD1-PP1 targeting is predominantly mediated by a short RVXF-docking motif, which is present in 90% of targeting proteins (31,51). Therefore, bioinformatic analysis was used to search for RVXF motifs on PLM and NCX1. No RVXF motifs were identified in PLM, but three putative RVXF motifs were identified in NCX1, two of which were localized to the cytosolic loop, which mediates NCX1 regulation.
Several experiments were conducted, which suggest that the KVFF motif in the rat NCX1 cytosolic loop (KIFF in human and mouse NCX1) is the PP1c-targeting site responsible for pSer-68-PLM dephosphorylation. 1) PP1c bound directly to the cytosolic loop of recombinant NCX1. 2) Peptide array data localized PP1c binding to NCX1-KVFF. Binding to other putative PP1c-binding motifs was not observed. 3) PP1c binding was abolished when the KVFF motif was deleted or mutated to KAFA or AVAF, both in the NCX1 full-length protein and in 20-mer NCX1 peptides. 4) Presence of NCX1 was a prerequisite for dephosphorylation of pSer-68-PLM by PP1c. No dephosphorylation was observed when the NCX1 fulllength protein was absent or when the KVFF motif was mutated. 5) The NCX1 cytosolic loop, containing KVFF, had no effect on PP1 activity, confirming an anchoring role of NCX1. 6) Finally, the electrophysiological experiments demonstrated that mutation of the PP1c KVFF site suppresses NCX1 activity and that this suppression is mediated through an increased pSer-68-PLM level.
As mentioned above, loss of the NCX1-PP1c interaction was observed with both the NCX1-KAFA and NCX1-AVAF mutations. We also generated a proline substitution mutant, NCX1-(F407P), and the obtained results are in agreement with studies reporting that proline is not tolerated at position X in RVXF because it prevents ␤-strand formation (31). Other proteins that adopt a more rigid structure upon PP1-binding are spinophilin and MYPT1. Spinophilin forms a ␤-strand upon binding to PP1, which extends a ␤-strand structure in PP1 (52). Although regulator proteins adopt a distinct conformation when bound to PP1, they do not significantly change the PP1 structure.
It has been suggested that most PP1-targeting proteins combine multiple docking motifs to form a stable complex with PP1 (53). Thus, in addition to looking for the RVXF motifs, we scanned the NCX1 primary sequence for additional PP1-docking motifs. NCX1 was found to partially fulfill the extended PP1-binding motif, RVXF-X 5-8 ⌽ 1 ⌽ 2 -X 8 -9 -R (36). By introducing glycine mutations to the C terminus of KVFF in the array, loss of binding was observed with N418G, suggesting Asn-418 to be part of the ⌽ 1 ⌽ 2 motif. Using the JackHMMER (54), Peti et al. (30) constrained ⌽ 2 to amino acids FIYRHNQSC and ⌽ 1 to amino acids VIYFH. Thus, Glu-417 does not fit into FIGURE 9. Mapping of the NCX1 interaction site in PP1. A, schematic figure showing PP1c and two deletion mutants that were generated to map the NCX1 interaction site in PP1. A FLAG and His 6 tags were inserted on the N terminus. The arrow indicates NCX1 binding region. B, His 6 -TF-NCX1 cyt together with the PP1 deletion variants were subjected to immunoprecipitation using anti-FLAG (in triplicates). Precipitates were analyzed with immunoblotting. C, identification of NCX1 binding by overlaying His 6 -TF-NCX1 cyt on 20-mer overlapping rat PP1 (P62138) peptides synthesized on membrane. Binding was analyzed using anti-His 6 HRP (n ϭ 2). Boxed area shows His 6 -TF-NCX1 cyt binding to PP1c; amino acids in bold indicate the common sequence in the three peptide sequences, and underlined amino acids form part of the RVXF binding pocket. D, FLAG-His 6 -PP1c(1-330) together with a variant, where the NCX1 binding region was deleted; FLAG-His 6 -PP1c(⌬232-263) was used in immunoprecipitation experiments. The anti-FLAG antibody was used to precipitate FLAG-His 6 -PP1c variants expressed in HEK293 cells followed by immunoblotting with anti-NCX1 for determination of NCX1 association with PP1c (in triplicates). His 6 -TF-NCX1 cyt in lysate from non-transfected HEK293 cells was used as a negative control. E, residues that are important for PP1-binding of target proteins (30) is shown by the stars in the alignment of human, mouse, and rat PP1. The boxed area show His 6 -TF-NCX1 cyt binding to PP1 (identified in C).
the pure ⌽ 1 definition; however, the ⌽ 1 ⌽ 2 motif is considered to be variable. Additionally, there is a conserved Arg-428 in NCX1. Therefore, in addition to the RVXF PP1-binding motif, the partial ⌽ 1 ⌽ 2 motif and the conserved Arg-428 on NCX1 may form part of a high affinity PP1-binding site. The Arg motif in the PP1 nuclear targeting subunit (PNUTS)-PP1 holoenzyme has been shown to work as a substrate selectivity filter, by blocking access to the C-terminal binding groove (36). Therefore, dephosphorylation of pSer-68-PLM may not require binding to the C-terminal groove, and PP1 interac-tion sites such as SILK (55) and MyPhoNE (56) were not found in NCX1. Several PP1 holoenzyme complexes have been reported to be regulated by phosphorylation of amino acids flanking the RVXF motifs (31). NCX1 possesses both a serine and threonine close to the KIFF/KVFF motif, suggesting that PP1 targeting to NCX1 might be dynamic. Scansite.mit.edu software identified the flanking phosphorylation sites in NCX-KIFF/KVFF as putative casein kinase or Src kinase phosphorylation sites.
NCX1 Binds to the Hydrophobic Pocket in PP1c-PP1c deletion mutants constrained NCX1 binding to amino acids 212-330 in the N terminus of PP1c. Consistently, the peptide array assay showed very weak binding to the sequence 235 FLHKH-DLDLICRAHQVVEDGYEFFAK 260 . The Asp-242, Leu-243, and Phe-257 residues are localized in ␤-sheets 10 and 11 and form part of the RVXF binding pocket, where Asp-242 exhibits a high excess of negative charges, and Leu-243 and Phe-257 form part of the hydrophobic pocket (30,57). These different properties of the PP1 surface determine the amino acid types tolerated at the different ligand positions. To confirm that NCX1 anchoring to PP1 is mediated by the residues in the hydrophobic pocket, we generated glycine single and double mutations of FLAG-His 6 -PP1c (L243G and F257G). Our peptide docking model, with the NCX-KVFF motif in human NCX1 based on the PP1-spinophilin complex, shows that NCX1-Val-406 may fit onto a hydrophobic pocket formed by Ile-169, Leu-243, and Leu-289 in PP1. Furthermore, the NCX1-Phe-407 may interact with PP1-Met-290. The conformation of the KVFF motif in this model is similar to the conformation of the homologous KIFF motif, as found in the NMR and x-ray structures of the Canis Lupus NCX1 (Protein Data Bank code 2FWS (42) and Protein Data Bank code 2DPK (49)).
The PP1␣-NCX1 interaction was strong (K d of 3.0 nM), in a similar range to that of PNUTS-PP1␣/␤/␥ (9.3 nM) (36) and  spinophilin-PP1␣ (8.7 nM) (41). Two-thirds of all PP1-targeting proteins have a PP1 interaction domain that is predicted to be intrinsically disordered. These interaction domains adopt a defined fold upon binding of PP1. X-ray crystallography and NMR studies have shown a substantial loss of structural integrity in the Ca 2ϩ -binding sites of NCX1-CBD1 in the absence of Ca 2ϩ , but upon Ca 2ϩ binding NCX1-CBD1 is arranged into antiparallel ␤-strands (58). Interestingly, Ile-406 and Phe-408 in canine NCX1, which are part of the identified NCX1-KIFF/ KVFF motif, have been suggested to be important for stability of the protein structure (58). We speculate that the PP1-Asp-242 might form a salt bridge with the NCX1-Lys-408, facilitating high affinity binding, as shown in our SPR analysis results. Importantly, NCX1 was able to bind PP1-Phe-257 on the peptide array, which has been shown to be an important anchoring site in PP1, and mutation of this residue to alanine resulted in loss of PP1-binding (48). PP1c Anchored to NCX1-KIFF/KVFF, Dephosphorylates pSer-68-PLM, and Regulates NCX1 Activity in HEK293-We confirmed that PLM co-precipitates with NCX1 in lysates from rat cardiomyocytes and HEK293, as reported previously (22). Immunoblotting with a specific anti-pSer-68-PLM antibody showed that PLM was highly phosphorylated in HEK293, consistent with reports that PLM is endogenously phosphorylated (30 -40%) by kinases in HEK293 (24,47). Noteworthy, both PLM and non-phosphorylatable PLM bind to NCX1 (22); however, it is pSer-68-PLM that exerts an inhibitory effect on NCX1 activity (24). The pSer-68-PLM level was significantly reduced in the presence of NCX1(FL)-GFP, suggesting that NCX1 anchoring is a prerequisite for dephosphorylation of pSer-68-PLM by PP1c. Co-expression of PLM with WT NCX1(FL) reduced the current in both forward and reverse mode. Moreover, NCX1 function was almost completely suppressed when HEK293 cells were co-transfected with PLM and NCX1(F407P) (mutated PP1-anchoring site). These results suggest that PP1 plays a significant role in relieving NCX1 inhibition, by dephosphorylating pSer-68-PLM. Reverse mode NCX1 was also observed to be inhibited in Ca 2ϩ fluorescence measurements. This result is in agreement with Ahlers et al. (22), who used a radioactive tracer uptake method in HEK293 cells. They found FIGURE 11. Analysis of NCX1, PP1c, and PLM in SHAM and ABHF rats. A, PP1c, pSer-68-PLM and PLM levels were analyzed in isolated cardiac membrane fractions by immunoblotting using relevant antibodies (n ϭ 5). Anti-calsequestrin and total PLM were used as loading controls. Bar graph show relative PP1c/calsequestrin level and pSer-68-PLM/total PLM level quantified by densitometry analysis. Differences were tested using unpaired t test (*, p Ͻ 0.05; ***, p Ͻ 0.001). B, FLAG-His 6 -PP1c(1-330), PLM, and NCX1 (TEV) were expressed with or without TEV protease followed by immunoblotting using relevant antibodies (in triplicates). Bar graph show relative pSer-68-PLM/total PLM level quantified by densitometry analysis. Differences were tested using unpaired t test (*, p Ͻ 0.05). that co-expressing NCX1 and PLM resulted in 15% reduction of 45 Ca 2ϩ uptake after cells were exposed to a solution containing zero Na ϩ .
PLM-NCX1-PP1c Macro-molecular Complex in HF-Consistent with previous findings (15), we found an up-regulation of NCX1 expression in our pressure overload model. However, evaluation of NCX1 function is complex, as an increase in NCX1 expression does not imply an increase in NCX1 activity. Indeed, there have been some models showing a decreased or unchanged NCX1 activity (9,59). Regulator proteins of NCX1 might also be altered in HF, which also affects NCX1 activity. The PLM level may vary in different models of hypertrophy or HF, as indicated in the study by El-Armouche et al. (28), where they showed a down-regulation of PLM phosphorylation in human heart failure. Bossuyt et al. (60) found a reduction in total PLM expression in HF rabbit myocytes, LV homogenates, and in human HF but an up-regulation in the pSer-68-PLM. The existence of different pools of PLM in the heart could account for the different phosphorylation outcomes for the NKA-associated and NCX1-associated PLM. Such pools are known to exist in myocytes (61). Global PP1c activity and expression is increased in human and experimental HF and is suggested to be a unique characteristic of HF (62). Mice overexpressing the catalytic subunit of PP1␣ to similar levels as observed in HF had depressed cardiac function, dilated cardiomyopathy, and premature mortality (63). We found up-regulation of both PP1c/calsequestrin levels and pSer-68-PLM/ total PLM levels in our pressure overload model, suggesting that PP1 is not able to dephosphorylate pSer-68-PLM. PLM interacts with two distinct regions in the intracellular loop of NCX1 (amino acids 238 -270 and 300 -328) (16) and inhibits the activity of exchanger. Recently, we showed that NCX1 was cleaved by calpain in HF and mapped a calpain cleavage site to Met-369 in NCX1, which separates the PLM-and PP1c (this study)-anchoring sites. Cleavage at this site led to a reduction in NCX1 activity (both forward and reverse mode) compared with WT NCX1(FL) (15). Interestingly, cleavage at this site in HEK293 up-regulated pSer-68-PLM/total PLM levels ( Fig. 10B; this study), suggesting that dephosphorylation of pSer-68-PLM by PP1c is inhibited. This finding could account for why it is reported that pSer-68-PLM and PP1c levels are up-regulated in HF.

JOURNAL OF BIOLOGICAL CHEMISTRY 4577
Author Contributions-T. L. H. conceived and coordinated the study, performed and analyzed the experiments shown in Figs. 1, 2, A  and C-F, 3, A-C, 4, A-C, 5, A-D, 5G, 6, A-D, 8, A and B, 9, A-D, 10B, 11A, and 12, wrote the paper, reviewed and approved the final version of the manuscript. K. H. performed and analyzed the experiments shown in Fig. 7, reviewed the results, and approved the final version of the manuscript. P. W. performed experiments in Figs. 2B and 6E, reviewed the results, and approved the final version of the manuscript. J. M. A. performed the surgery on the animals and did the echocardiographic, hemodynamic, and post-mortem analysis in Table 1, reviewed the results, and approved the final version of the manuscript. B. D. made the structure model shown in Fig. 10A, was involved in the planning of experiments shown in Figs. 3D and 10B, reviewed the results, and approved the final version of the manuscript. P. K. L. was involved in the planning of the experiment shown in Fig. 8 Fig. 6B and optimization conditions, reviewed the results, and approved the final version of the manuscript. I. S. performed the surgery on the animals and did the echocardiographic, hemodynamic, and post-mortem analysis in Table 1, reviewed the results, and approved the final version of the manuscript. W. E. L. conceived, coordinated, and analyzed the experiment shown in Fig. 8 and reviewed the results, and approved the final version of the manuscript. O. M. S. conceived the study, reviewed the results, and approved the final version of the manuscript. C. R. C. conceived and coordinated the study, performed the experiments in Figs. 2G, 5, E and F, and 9E, analyzed the experiments, supervised the writing of the paper, reviewed and approved the final version of the manuscript.