Development of a high-affinity peptide that prevents phospholemman (PLM) inhibition of the sodium/calcium exchanger 1 (NCX1)

NCX (Na⁺/Ca²⁺ exchanger) is an important regulator of Ca²⁺ homeostasis. It functions as a bidirectional antiporter, exchanging one Ca²⁺ for three Na⁺, removing either Ca²⁺ (forward mode) or Na⁺ (reverse mode) from the cytosol depending on membrane potential and local ion concentrations [1]. Three different SLC8 (solute carrier 8) genes have been identified in mammals: SLC8A1 encoding NCX1 [2], SLC8A2 encoding NCX2 [3], and SLC8A3, encoding NCX3 [4,5]. Whereas NCX2 and NCX3 are mainly expressed in brain and skeletal muscle, NCX1 is ubiquitous, being found in virtually all excitable and non-excitable tissues including brain, vascular smooth muscle, kidney and heart. NCX1 is composed of ∼970 amino acids spanning nine TMs (transmembrane segments) and a large cytoplasmic loop between TM5 and TM6. The cytoplasmic loop of NCX1 contains several regulatory domains including the inhibitory XIP (exchanger inhibitory peptide) region which binds to calmodulin [6], two Ca²⁺-binding domains (CBD1 and CBD2) [7], sites for phosphorylation by PKC (protein kinase C), Ca²⁺/calmodulin-dependent kinase and tyrosine kinase [8], and reported interaction and cleavage sites for calpain [9] and PLM (phospholemman) [10].

PLM is a 72-amino-acid phosphoprotein belonging to the FXYD1 family of ion transport regulators [11], and it is the first reported endogenous inhibitor of NCX1 [12]. It is highly expressed in brain [13] and heart [14], and has been shown to associate with NCX1 in cardiac membranes [15] and to inhibit NCX1 activity when co-expressed [12,16]. The cytoplasmic region of PLM is reported to interact with PASKT- and QKHPD-containing sequences in the cytoplasmic loop of NCX1 [17] and inhibit NCX1 when it is phosphorylated at Ser⁶⁸ (pSer⁶⁸-PLM) [18] by PKA (protein kinase A) or PKC. It is well established that PLM phosphorylation relieves its inhibitory effect on NKA (Na⁺/K⁺-ATPase) [19]. Thus an interesting mechanism is proposed whereby phosphorylation of PLM switches its inhibitory actions from NKA to NCX1 [18]. However, an absence of interaction between PLM and NCX1 has also been reported [20], and it is less clear how PLM regulation of NCX1 is integrated with PLM regulation of NKA [20–22].

A regulatory role for PLM on NCX1 has clear therapeutic potential. NCX1 expression and activity are altered during brain ischaemia, neurodegenerative disorders, aging [23,24] and HF (heart failure). In brain, it has been shown that up-regulation of NCX1 has protective effects during cerebral ischaemia [25], suggesting that inhibition of NCX1 by PLM may be detrimental. Similarly, in heart, increased NCX1 activity is believed to maintain diastolic function in diseases such as HF [26]. Constitutive overexpression of phosphorylated PLM (S68E, mimicking phosphorylated Ser⁶⁸) in mice was observed to inhibit NCX1 activity, and also resulted in arrhythmia, HF and early mortality [27]. Thus augmenting NCX1 function in disease by preventing PLM-induced inhibition may serve as a novel strategy for disease treatment in both brain and heart.

In the present study, we aimed to investigate the NCX1–PLM interaction in brain and heart, and analyse the functional consequence of this interaction. First, we mapped the reciprocal PLM–NCX1 binding sites at the amino acid level. Using two-dimensional peptide arrays, we developed further a high-affinity blocking peptide specific for the PLM–NCX1 interaction. Our data demonstrate that the optimized peptide exhibits enhanced affinity for pSer⁶⁸-PLM and can reverse phosphorylation-dependent PLM inhibition of NCX1.

The MGC mouse clone BC079673 (NCX1) was cloned into the first reading frame of pAdTrack-cytomegalovirus (CMV) shuttle vector (plasmid 16405, Addgene). Mouse PLM (AF089734) and PLM mutated at Ser⁶⁸ to an aspartic acid residue (S68D) were cloned with a C-terminal stop codon into pcDNA3.1/myc-HisA (Invitrogen) by Genscript. The fidelity of the cloning and mutation procedures was verified by sequence analysis (Genscript).

Peptides on cellulose membranes were synthesized using Multipep automated peptide synthesizer (standard Fmoc solid-phase peptide synthesis using the SPOT synthesis technique, INTAVIS Bioanalytical Instruments) [28,29]. Parts of the intracellular loop of rat NCX1 (EDM02743) and mouse PLM (Q9Z239) were synthesized as overlapping 20-mer peptides with three amino acid offsets. Peptides in solution were synthesized (listed below) and purified to obtain >80% purity by Genscript. For some of the peptides, a biotin tag was included at the N-terminus. The peptide sequences were: NCX1(229–248), KRYRAGKQRGMIIEHEGDRP; NCX1 (235–254), KQRGMIIEHEGDRPASKTEI; NCX1(292–311), ARILKELKQKHPDKEIEQLI; NCX1(301–320), KHPDKEIE-QLIELANYQVLS; PLM_cyt(38–72), KRCRCKFNQQQRTGEP-DEEEGTFRSSIRRLSSRRR; pSer⁶⁸-PLM_cyt, KRCRCKFNQ-QQRTGEPDEEEGTFRSSIRRLpSSRRR; NCX1(K301Y,D30-4Y), YHPYKEIEQLIELANYQVLS; NCX1(301–320) scram-bled, VYEKNKLDLLPAIEQEQSHI; NCX1 blocking peptide, CGQPVFRKVHARDHPIPST.

HEK (human embryonic kidney)-293 cells were cultured in DMEM (Dulbecco's modified Eagle's medium) (Gibco-BRL) supplemented with 10% (v/v) FBS (Gibco-BRL), 100 unit/ml penicillin, 0.1 mg/ml streptomycin (penicillin/streptomycin, Sigma–Aldrich) and 1% non-essential amino acids (Gibco-BRL) and maintained in a 37°C, 5% CO₂ humidified incubator. Plasmid DNA was transfected into HEK-293 cells using Lipofectamine™ 2000 (Invitrogen) as instructed by the manufacturer or the CaCl₂ method. After 24 h, the cells were lysed in lysis buffer containing 20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA and 0.5% Triton X-100 with Complete™ protease inhibitor cocktail tablets (Complete™ Mini EDTA-free; Roche Diagnostics). For patch-clamp experiments, the 24-h-transfected cells were transferred to poly-L-lysine (Sigma–Aldrich)-coated coverslips and incubated for an additional 24 h in a 37°C, 5% CO₂ humidified incubator.

Frozen mouse brain or LV (left ventricle) from rats was pulverized in a mortar with liquid nitrogen, followed by addition of ice-cold lysis buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA and 0.5% Triton) supplemented with 1 mM PMSF (Sigma–Aldrich) and Complete™ protease inhibitor cocktail tablets. Tissue samples were homogenized three times for 1 min on ice with a Polytron 1200 and centrifuged at 100000 g for 60 min at 4°C. Supernatants were collected and stored at −70°C.

Rat LV and cardiomyocytes were fractionated using a Compartment Protein Extraction Kit (Millipore) according to the manufacturer's instructions.

Animals were handled according to the National Regulations on Animal Experimentation in accordance with the Norwegian Animal Welfare Act. The animal experiments were approved by the Norwegian National Animal Research Committee, which conforms to the Guide for the Care and use of Laboratory Animals published by the U.S. National Institutes of Health (NIH publication number 85-23, revised 1996). Neonatal cardiomyocytes were prepared from the LV of 1–3-day-old Wistar rats as described previously [9]. The cardiomyocytes were incubated in a plating medium consisting of DMEM (Sigma–Aldrich), M-199 (Sigma–Aldrich), penicillin/streptomycin (Sigma–Aldrich), horse serum (BioWhittaker) and FBS (BioWhittaker) in a humidified incubator with 5% CO₂ at 37°C for 24 h before protein fractionation.

Synthesized peptide membranes were first activated by soaking membranes in methanol for a few seconds and were then washed three times for 10 min with TBS-T (TBS with 0.1% Tween 20). The membranes were then incubated with blocking solution (1% casein) (Roche Diagnostics) at room temperature. After 1 h of blocking, the membranes were incubated with 1–5 μM biotinylated peptide in 1% casein overnight at 4°C with gentle agitation. For the competition experiments, the blocking peptide [5 μM NCX1(K301Y,D304Y)] was pre-incubated with the membranes overnight at 4°C with gentle agitation, before incubation with biotinylated peptide for 2 h. The membranes were then washed three times for 10 min with TBS-T. Binding was detected by immunoblotting. The peptides are covalently linked to the membrane and should, according to the manufacturer (Intavis), remain on the membrane after washing and stripping protocols.

Each biotinylated peptide (8 μM) was incubated with 25 μl of monoclonal anti-biotin antibody-conjugated beads (A-1559, Sigma–Aldrich) in 100 μl of PBS for 2 h at 4°C with rotation. To remove unbound peptide, the beads were washed three times with PBS, followed by adding 100 μl of HEK-293 cell lysates, 0.5 μg of recombinant His–TF (trigger factor)–NCX1_cyt or 133 μM PLM_cyt peptide diluted in 150 μl of immunoprecipitation buffer containing 1% (w/v) BSA. The samples were rotated for 2 h at 4°C followed by washing the beads three times with immunoprecipitation buffer (20 mM Hepes, pH 7.5, 150 mM NaCl, 1 mM EDTA and 1% Triton X-100) before boiling in 2× SDS loading buffer. Binding was analysed by immunoblotting.

Immunoprecipitation was performed by incubating 2 μg of the appropriate antibody with 200 μl of protein sample [rat heart lysates (6.7 μg/μl), HEK-293 lysates (10 μg/μl) or brain lysate (1.5 μg/μl)] and Protein A/G PLUS–agarose (Santa Cruz Biotechnology) overnight at 4°C with rotation. The next day, samples were washed three times in immunoprecipitation buffer (or PBS for brain) and boiled in 2× SDS loading buffer before SDS/PAGE analysis. An equal amount of rabbit IgG (sc-2027, Santa Cruz Biotechnology) was used as a negative control. Blocking peptide (antigen: cardiac NCX1, sequence: CGQPVFRKVHARDHPIPST) (Genscript) was incubated with anti-NCX1 before immunoprecipitation (negative control). Usually 20 μl of the protein sample was used as an input control.

Samples from pull-down assays were analysed by SDS/PAGE on 4–15% or 15% Criterion Tris/HCl gels (Bio-Rad Laboratories) and blotted on to PVDF membranes (GE Healthcare). The PVDF membranes and peptide arrays were blocked in 5% (w/v) dried non-fat skimmed milk powder or 1% (w/v) casein in TBS-T for 60 min at room temperature and incubated overnight with primary antibody at 4°C. After incubation with primary antibody, the membranes were washed five times for 5 min in TBS-T and incubated further with an HRP (horseradish peroxidase)-conjugated secondary antibody. The membranes were developed using ECL Plus (GE Healthcare). The chemiluminescence signals were detected using a LAS 1000 or LAS 4000 instrument (Fujifilm).

Anti-NCX1 (epitope: GQPVFRKVHARDHPIPST) was custom-made by Genscript. HRP-conjugated anti-biotin (A0185) was purchased from Sigma–Aldrich. Anti-PLM (anti-FXYD1) (ab76597) was purchased from Abcam. Anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) (V-18) (sc-20357) was obtained from Santa Cruz Biotechnology. Anti-calsequestrin (PA1-913) was from Thermo Scientific. HRP-conjugated anti-rabbit IgG (NA934V, GE Healthcare) and HRP-conjugated anti-goat IgG (HAF109, R&D Systems) were used as secondary antibodies. Recombinant cytoplasmic NCX1 protein (His–TF–NCX1_cyt) with an N-terminal His and TF tag to stabilize the protein was custom-made by Genscript.

Whole-cell patch-clamp experiments were conducted on NCX1- and PLM(S68D)-transfected HEK-293 cells. The patch electrodes were made from borosilicate glass with filament, and had a final resistance between 2 and 4 MΩ. The electrodes were connected to an Axoclamp 200B amplifier and a Digidata 1550, controlled by pClamp 10.2 software (all from Axon Instruments/Molecular Devices). The recorded signals were sampled at 20 kHz, filtered at 2 kHz using an analogue low-pass Bessel filter, and stored on the computer. The recordings were performed at 37°C in an extracellular solution containing 140 mM NaCl, 5 mM CsCl, 1.2 mM MgSO₄, 1.2 mM NaH₂PO₄, 5 mM CaCl₂, 10 mM Hepes and 10 mM glucose (pH 7.4 with CsOH) and osmolality 290 mOsm. To block K⁺-, Ca²⁺-, Cl⁻- and Na⁺/K⁺-ATPase currents, we used caesium, 20 μM nifedipine, 30 μM niflumic acid and 1 mM ouabain. The patch pipettes were filled with a solution containing 100 mM caesium glutamate, 1 mM MgCl, 10 mM Hepes, 4 mM NaCl, 2.5 mM Na₂-ATP, 10 mM EGTA and 6 mM CaCl₂ (pH 7.2) and osmolality 270 mOsm. In experiments to assess the effect of the optimized peptide we added 5 μM of either optimized or scrambled control peptide to the internal solution. Recordings were corrected for liquid junction potential (15 mV), calculated using pCLAMP 10 software (V_membrane=V_pipette−15 mV). Following whole-cell configuration, the series resistance, seal resistance and cell capacitance were monitored throughout the experiment. Only cells with a stable gigaseal and cell capacitance were used. In addition, we only used cells with series resistance <10 MΩ. The cells were clamped to −43 mV (reversal potential) for 6 min to allow sufficient intracellular diffusion of both ions and peptides. The NCX1 current was elicited by a descending voltage ramp from 120 mV to −100 mV and isolated using 5 μM Ni²⁺. The currents were normalized to cell capacitance and the current (I)–voltage (V) relationships were plotted from −100 to 100 mV.

Densitometric analysis was performed using Image Gauge version 4.0, ImageQuant TL (GE Healthcare) or ImageJ (NIH).

All data are expressed as means±S.E.M. Comparisons between two groups were analysed using unpaired Student's t test (GraphPad Prism 5.04). P<0.05 was considered statistically significant.

Several experiments were performed to investigate the PLM–NCX1 interaction. First, PLM was observed together with NCX1 in membrane fractions isolated from rat neonatal cardiomyocytes (Figure 1A, left-hand panel) and rat LV (Figure 1A, right-hand panel) using specific anti-PLM (epitope mapped in Figure 1B) and anti-NCX1 antibodies [30]. Secondly, immunoprecipitation of endogenous NCX1 using anti-NCX1 antibodies identified co-precipitation of endogenous PLM in both rat LV and brain lysate (Figures 1C and 1D respectively). Moreover, immunoprecipitation of NCX1 expressed in HEK-293 cells co-transfected with PLM revealed co-precipitation of PLM (Figure 1E). NCX1 precipitated PLM(S68D) (mimicking phosphorylated PLM) and unphosphorylated PLM (Figure 1F) at similar levels. Finally, pull-down assays with biotinylated peptides consisting of the cytoplasmic part of PLM or pSer⁶⁸-PLM (Figure 1G, schematic representation in the upper panel, immunoblots shown in the lower panel) and recombinant His–TF–NCX1_cyt were performed with anti-biotin–agarose. His–TF–NCX1_cyt precipitated with both PLM(38–72) and pSer⁶⁸-PLM(38–72) (Figure 1H), confirming that the cytoplasmic domain of PLM binds directly to the NCX1 cytoplasmic loop, and verifying that this interaction is independent of the phosphorylation status of Ser⁶⁸ on PLM. Sequence alignments show that the PLM cytoplasmic domain is identical in human, rat, mouse, pig and dog, with the exception of a serine residue at position 69 in mouse (Figure 1I).

To map PLM-binding sites in NCX1 more precisely, 20-mer NCX1 peptides were overlaid with biotin–PLM_cyt and biotin–pSer⁶⁸-PLM_cyt peptides. Both peptides bound to PASKT- and QKHPD-containing sequences in NCX1 (Figures 2A and 2B, marked in bold letters), consistent with a previous study [31]. In addition, both peptides bound to the inhibitory XIP region in NCX1 [6] (Figures 2A and 2B, underlined sequences) and to a region C-terminal of the QKHPD motif (T³⁶⁶RLMTGAGNILKRHAADQAR³⁸⁵). As peptide array screening is semi-quantitative, quantitative analysis with selected peptides in solution should be performed. Thus four peptides covering amino acids 229–248, 235–254, 292–311 and 301–320 in NCX1 (black arrows in Figures 2A and 2B) were also synthesized as purified biotinylated peptides in solution (sequences given in Figure 2C). These NCX1 peptides were analysed further for their binding affinity to PLM expressed in HEK-293 cells. Pull-down assays using anti-biotin-conjugated beads and immunoblotting with anti-PLM revealed that all four NCX1 peptides precipitated similar levels of PLM (Figure 2D).

To map where the NCX1-PASKT- and NCX1-QKHPD-containing sequences bound in PLM, biotinylated peptides containing the two motifs (the three lowermost peptides in Figure 2C) were overlaid on the pSer⁶⁸-PLM cytoplasmic part synthesized as 20-mer overlapping peptides on membranes. Binding was detected by using HRP-conjugated anti-biotin antibody. The results revealed that all three NCX1 peptides bound strongest to RCKFNQQQRTGEPDEEEGTF (Figure 2E), consistent with NCX1 binding being independent of Ser⁶⁸ phosphorylation in PLM. RCKFNQQQRTGEPDEEEGTF covers the more flexible PLM region, upstream of the α-helical region (Figure 2E).

The two uppermost peptides in Figure 2(C) were excluded for optimization as they contain the autoregulatory domain of NCX1 (XIP), which is known to bind NCX1 and calmodulin [6]. The two lowermost sequences were optimized further by two-dimensional peptide arrays, where each amino acid in the native peptide sequences was systemically replaced with every possible amino acid. Immunoblotting using HRP-conjugated anti-biotin showed that substitutions within NCX1(301–320) (KHPDKEIEQLIELANYQVLS) strongly increased binding to biotin–pSer⁶⁸-PLM (Figure 3A). This effect was not observed in NCX1(292–311) (ARILKELKQKHPDKEIEQLI) (results not shown). The pSer⁶⁸-PLM binding affinity was dramatically increased when NCX1(301–320) (KHPDKEIEQLIELANYQVLS) was replaced at positions 1–11 (labelled with red circles) compared with internal control peptides (white circles). Especially, tryptophan and tyrosine substitutions highly increased the affinity for pSer⁶⁸-PLM (Figure 3A, lower bracket on the left). Replacing with negatively charged amino acids such as aspartic acid and glutamic acid were not favourable, as they mostly reduced or abolished pSer⁶⁸-PLM binding (Figure 3A, upper bracket on the left). Consistent with the disfavour for negatively charged amino acids, removal of the endogenous aspartic acid in position 4 (D4) and glutamic acid in position 6 (E6) mostly increased pSer⁶⁸-PLM binding (indicated with arrows above the array in Figure 3A).

In a second peptide membrane synthesis, residues of the native NCX1(301–320) peptide sequence were individually replaced with tyrosine at position 1 (K1), 2 (H2), 3 (P3), 5 (K5), 7 (I7) or 11 (I11). Immunoblotting using HRP-conjugated anti-biotin revealed that K1Y, H2Y, P3Y and K5Y greatly increased the pSer⁶⁸-PLM affinity (Figure 3B). In an attempt to increase the affinity to an even higher level, double substitutions in the NCX1(301–320) sequence were performed. Each of the single mutations, i.e. K1Y, H2Y, P3Y, K5Y, I7Y and I11Y, was combined with amino acid substitutions in the more flexible positions D4 and E6 (indicated by arrows above the array in Figure 3A). Most of the double-substituted peptides showed increased pSer⁶⁸-PLM binding compared with the native NCX1 peptide (average data from two independent experiments are shown only for K1Y double mutations in Figure 3B). In conclusion, the double mutation K1Y, D4Y had the highest affinity for pSer⁶⁸-PLM (Figure 3B, labelled in red circle). K1 and D4 correspond to Lys³⁰¹ and Asp³⁰⁴ in the full-length NCX1 protein.

Pull-down experiments using biotinylated peptides covering the native NCX1(301–320) sequence, the optimized sequence (Opt-pep) or a scrambled sequence (sequences are given in Figure 4A), together with an untagged PLM_cyt peptide, were performed. Immunoblotting with anti-PLM showed that Opt-pep bound more strongly to PLM_cyt than native NCX1(301–320) peptide (Figure 4B). The scrambled peptide and beads without any peptide were used as negative controls. In addition to stronger binding, overlay assays indicated that Opt-pep also bound more broadly to pSer⁶⁸-PLM sequences synthesized on membranes (Figure 4C, middle panels), compared with native NCX1(301–320) peptide (Figure 4C, left-hand panels).

To test whether Opt-pep was able to block NCX1 binding, untagged Opt-pep was pre-incubated with pSer⁶⁸-PLM_cyt synthesized on membranes, before overlaying with biotin–NCX1(235–254) (containing the PASKT motif) and biotin–NCX1(301–320) (containing the KHPD motif). Immunoblotting with HRP-conjugated anti-biotin showed that Opt-pep efficiently blocked binding of both NCX1 peptides to pSer⁶⁸-PLM (middle panels in Figures 4D and 4E respectively). Altogether, our data indicate that Opt-pep effectively blocks the NCX1–PLM interaction.

Sequence alignments show that the native NCX1 sequence KHPDKEIEQLIELANYQVLS and surrounding regions were almost identical in human, rat, mouse and dog, suggesting that Opt-pep is able to block the NCX1–PLM interaction across species (Figure 4F).

To study the effect of Opt-pep on NCX1 current in the presence of pSer⁶⁸-PLM, we transfected HEK-293 cells with NCX1 and PLM(S68D) (mimicking phosphorylated PLM). The currents were recorded in a whole-cell patch-clamp configuration during voltage ramps from 120 to −100 mV. Typical recordings of current traces are shown in Figures 5(A) and 5(B), with either 5 μM Opt-pep or 5 μM scrambled peptide included in the internal pipette solution respectively. Notably, the Opt-pep gradually relieved PLM induced NCX1 inhibition with the full effect observed after ∼5 min of dialysis [Figure 5A, dark grey (3 min) and black (5 min) traces]. This gradual increase in current was not observed using scrambled peptide (Figure 5B). Following 5 min of dialysis, the extracellular recording solution perfusing the cells was changed to a solution containing 5 μM Ni⁺ (Figures 5A and 5B, light grey trace) The subtracted Ni²⁺-sensitive currents, corresponding to the NCX1 current, were plotted as an I–V curve from −100 to 100 mV (Figures 5C and 5D). The assays to measure the effect of PLM(S68D) were first validated without the presence of Opt-pep. The normal NCX1 (Figure 5C, open circles) exhibited current in both forward and reverse modes. Consistent with the literature [12], co-transfecting NCX1 with PLM(S68D) inhibited both forward- and reverse-mode current (Figure 5C, open triangles). However, the presence of Opt-pep relieved the PLM inhibition (Figure 5D, closed circles) and the NCX1 current returned to the range observed in NCX1 transfected cells without PLM(S68D). The presence of the scrambled control peptide (sequence given in Figure 4A) did not relieve the inhibition of phosphorylated PLM (closed inverted triangles), strongly indicating that the reversal of NCX1 inhibition was sequence-specific.

A model of the functional interaction of pSer⁶⁸-PLM with NCX1 (inhibition) and relief of its inhibition by Opt-pep is shown in Figure 6.

Although there have been several studies suggesting that PLM is an endogenous NCX1 inhibitor in heart [32], another study has shown no direct interaction between the two proteins [20]. In the present study, we confirm that PLM interacts directly with NCX1 and that PLM phosphorylation at Ser⁶⁸ inhibits NCX1 activity. We observed that PLM precipitated with NCX1 expressed in HEK-293 cells and in rat LV lysates [12,16], and demonstrated for the first time that PLM and NCX1 also interact in the brain. We found that the cytoplasmic domain of PLM binds directly to PASKT- and QKHPD-containing sequences in the NCX1 cytoplasmic loop, as reported previously [17,31]. This interaction was observed to be independent of Ser⁶⁸ phosphorylation. Finally, we developed a blocking peptide specific for the PLM–NCX1 interaction, which reversed the inhibitory effect of PLM(S68D) on NCX1 activity in HEK-293 cells.

By extensive use of peptide array technology, we mapped PLM binding to two regions in NCX1, containing the PASKT and QKHPD motifs, reported previously to be involved in PLM binding [17]. PLM also bound to the XIP domain, which is in close proximity to PASKT. The XIP domain, containing a calmodulin-binding site [33], has an autoregulatory role in regulating NCX1 activity [6] and is involved in Na⁺-dependent inactivation [34]. However, PLM also inhibits NCX1 when the XIP region is deleted [17]. Importantly, the PLM-binding region in NCX1 overlapped with the YFP cloning site in the FRET study finding no PLM–CFP–NCX1–YFP interaction [20].

NCX-PASKT and NCX1-QKHPD bound to R⁴¹CKFN-QQQRTGEPDEEEGTF⁶⁰ in PLM_cyt. Thus NCX1 binding resided outside Ser⁶³, Ser⁶⁸ and Thr⁶⁹ in PLM, confirming that the PLM–NCX1 interaction is not dependent on PLM phosphorylation. The finding of identical binding motifs suggests that NCX-PASKT and NCX1-QKHPD might bind to each monomer in the PLM dimer. The two motifs are important for NCX1 regulation, because alanine mutations in these two motifs are reported to abolish or reduce NCX1 inhibition. Alanine mutation of lysine in the QKHPD motif (QAHPD) removes PLM inhibition of NCX1 [31]. Consistently, pSer⁶⁸-PLM was not able to bind to NCX1(301–320), when the lysine residue at position 301 was mutated to alanine (AHPDKEIEQLIELANYQVLS) (Figure 3A). Importantly, the NCX1-binding site in PLM (R⁴¹CKFNQQQRTGEPDEEEGTF⁶⁰) overlapped with the PLM antibody epitope (ab76597) (Figure 1B). Thus the antibody will only precipitate PLM without NCX1 bound, causing a false negative result.

The NCX-PASKT and NCX1-QKHPD motifs are located within an unstructured domain of NCX1 (CLD or catenin-like domain). The CLD is suggested to be important for transducing signals from CBD1 and CBD2 in NCX1 to the TMs upon Ca²⁺ binding [35]. It is less likely that PLM inhibits Ca²⁺-dependent activation of NCX1, since PLM inhibits NCX1 even in the absence of the CBDs [10], and also inhibits the NCX1-G503P mutant which lacks Ca²⁺-dependent activation of NCX1 [36]. Rather, it has been suggested that PLM regulates NCX1 activity by altering the oligomerization state of NCX1 [31]. Experiments using chemical cross-linking [37] and FRET [38] have suggested that NCX1 exists as dimers in the cell membrane of insect and Xenopus oocytes. Further experiments are required to investigate whether the NCX-PASKT and NCX1-QKHPD motifs are held distant or in close proximity in the NCX1 monomer or in a possible dimer. The stoichiometry of the PLM–NCX1 interaction is unknown.

The blocking peptide was derived from a NCX1 sequence containing the partial QKHPD motif [31]. Optimization of the NCX1-derived sequence by two-dimensional peptide arrays identified amino acids important for increased pSer⁶⁸-PLM_cyt binding. These specific amino acids were therefore replaced in the peptide sequence. The final optimized peptide (Opt-pep), where tyrosine substituted for lysine and aspartic acid (YHPYKEIEQLIELANYQVLS), had an 8-fold higher affinity for pSer⁶⁸-PLM_cyt compared with the native NCX1 sequence. Opt-pep blocked binding of NCX1-PASKT and NCX1-QKHPD to PLM and reversed the inhibitory effect of PLM(S68D) on NCX1 activity (both forward and reverse modes) in HEK-293 cells. The reversal was specific because a scrambled control peptide had no effect.

PLM regulates NCX1 and NKA differently. The cytoplasmic domain in the PLM dimer binds to NCX1, whereas the transmembrane domain in the PLM monomer binds to the NKA α-subunit. Contrary to NCX1 regulation, phosphorylation at Ser⁶⁸ relieves its inhibitory effect on NKAα [32], although pSer⁶⁸-PLM still remains physically associated [20,39]. Two PLM–NKAα interaction sites have been reported [40]. PLM-Phe²⁸ and NKA-Glu⁹⁶⁰ have been reported to be critical for the inhibitory interaction. A second non-regulatory NKA interaction site is suggested in the N-terminus (extracellular) or in the transmembrane domain of PLM. Thus, since Opt-pep binds to the cytoplasmic part of PLM, it is less likely that Opt-pep will interfere with the PLM–NKA interaction. PLM has also been suggested to modulate the activity of L-type Ca²⁺ channels in heterologous expression systems [41,42] and in adult cardiac myocytes [43]. Interestingly, only the extracellular and the transmembrane domains of PLM were found to be crucial for modulation of L-type Ca²⁺ channel, suggesting that Opt-pep will not interfere with the L-type Ca²⁺ channel activity.

PLM is considered to be a stress protein [32]. Under resting conditions, PLM is functionally quiescent, and the in vivo myocardial contractility is similar between wild-type and PLM-knockout mice [32,44]. However, during catecholamine stress, when PLM is phosphorylated, one of its major physiological functions is to suppress Na⁺ overload through NKA activation. Low intracellular Na⁺ levels favour Ca²⁺ extrusion by NCX1, its ‘forward’ mode of operation. However, such indirect effects of catecholaminergic stress on Ca²⁺ extrusion appear to be tempered by direct inhibition of NCX1 activity upon binding of pSer⁶⁸-PLM. Thus the net effects of PLM on cardiomyocyte Ca²⁺ handling and contraction/relaxation are complex, particularly since experimental data indicate that NKA sets local gradients of Na⁺, which regulate NCX1 activity [45,46]. This insight suggests that therapeutic modulation of PLM levels may be less appropriate than modulation of PLM interaction with either NKA or NCX. Our data indicate that blockade of the interaction between pSer⁶⁸-PLM and NCX1 increases NCX1 activity. Since NCX1 competes with SERCA (sarcoplasmic/endoplasmic reticulum Ca²⁺-ATPase), an increase in NCX1 activity has been shown to reduce sarcoplasmic reticulum Ca²⁺ content and thus reduce contractility [47]. Thus inhibition of the pSer⁶⁸-PLM–NCX1 interaction would not be expected to be beneficial in the setting of systolic HF where contractility is depressed. However, during diastolic HF [HFPEF (heart failure with preserved ejection fraction)], reducing the Ca²⁺ store is of less concern and enhancing Ca²⁺ extrusion by NCX1 may favourably improve diastolic function in these patients.

We also identified a PLM–NCX1 interaction in brain tissue, which may have important consequences for brain ischaemia. Previous study has confirmed the neuroprotective function of NCX, as knockout of NCX1, NCX2 and NCX3 resulted in augmented brain damage during ischaemia [48]. Furthermore, up-regulation of NCX1 and NCX3 have been shown to be neuroprotective in ischaemic pre-conditioning [49]. Similarly, in the heart, maintenance of low Ca²⁺ during ischaemia/reperfusion is known to be protective through forward-mode NCX1. Our present strategy of activation of NCX1 function without interference with NKA function may therefore be well suited for treating brain and cardiac ischaemia.

Pimthanya Wanichawan and Cathrine Rein Carlson conceived and co-ordinated the study and wrote the paper. Pimthanya Wanichawan, Tandekile Lubelwana Hafver, Marianne Lunde, Marita Martinsen and Cathrine Rein Carlson designed, performed and analysed the experiments shown in Figure 1–4 and Figure 6. Kjetil Hodne designed, performed and analysed the experiments shown in Figure 5. William Edward Louch, Ole Mathias Sejersted and Cathrine Rein Carlson provided technical assistance and contributed to the preparation of the Figures. All authors reviewed the results and approved the final version of the paper.

Pimthanya Wanichawan, Kjetil Hodne, Ole Mathias Sejersted and Cathrine Rein Carlson are partners on a patent application.

This work was supported by the Research Council of Norway, Norwegian National Health Association, Stiftelsen Kristian Gerhard Jebsen, Anders Jahre's Fund for the promotion of Science, the South-Eastern Norway Regional Health Authority [grant number 2016095], and the Simon Fougner Hartmanns family Fund, Denmark.

CBD

calcium-binding domain

CLD

catenin-like domain

DMEM

Dulbecco's modified Eagle's medium

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

HF

heart failure

NCX

Na⁺/Ca²⁺ exchanger

HRP

horseradish peroxidase

LV

left ventricle

NKA

Na⁺/K⁺-ATPase

PKC

protein kinase C

PLM

phospholemman

SLC8

solute carrier 8

TBS-T

TBS with 0.1% Tween 20

TF

trigger factor

TM

transmembrane segment

XIP

exchanger inhibitory peptide