AKAP complex regulates Ca²⁺ re-uptake into heart sarcoplasmic reticulum

AKAP18δ is present in the heart sarcoplasmic reticulum

In a search for the heart sarcoplasmic reticulum AKAP associated with PLN, we analysed sarcoplasmic reticulum fractions for the presence of PKA-RII-binding proteins by using an RII overlay assay. Rat hearts were homogenized, subjected to discontinuous sucrose density gradient fractionation and overlaid with ³²P-labelled RII in the absence or presence of the anchoring disruptor peptide Ht31 (Fig 1A). RII-binding proteins with mobilities of more than 200 and approximately 110, 90, 60 and 50 kDa were detected in the fractions containing sarcoplasmic reticulum, in addition to PKA-RII itself, owing to dimerization with low levels of monomer R in the solution. Immunoblotting of the heart fractions showed that AKAP18δ was present in fractions enriched in sarcoplasmic reticulum (Fig 1B), together with PKA subunits (Fig 1C). Furthermore, the distribution of AKAP18δ differed from that of the sarcolemmal marker, Na⁺/Ca²⁺ exchanger (NCX), which peaked in fractions 8–10, indicating that AKAP18δ is not a sarcolemmal protein (Fig 1B). We confirmed the presence of AKAP18δ in the heart by immunoprecipitation of AKAP18δ from total rat heart homogenate and analysis of immunoprecipitates by using an RII overlay (Fig 1D). The detection of AKAP18δ in the heart is also in agreement with our previous northern blot analysis (Henn et al, 2004).

Subcellular localization of AKAP18δ

The subcellular localization of AKAP18δ in rat heart tissue was examined by co-immunostaining of AKAP18δ and α-actinin, PKA-RIIα, SERCA2 and PLN (Fig 2A). The α-actinin staining identifies z-lines, allows visualization of the myofibrils and acts as a reference for the position of the sarcoplasmic reticulum (Vangheluwe et al, 2003). AKAP18δ immunofluorescence produced a striated pattern overlapping that of α-actinin, PLN and SERCA2. Furthermore, AKAP18δ colocalized with PKA-RIIα. In addition, SERCA2 colocalized with α-actinin, in agreement with earlier studies (Vangheluwe et al, 2003), and with PKA-RIIα (Fig 2A). AKAP18δ immunostaining was specific, as evident from controls with pre-immune serum and secondary antibody only (see supplementary Fig S1 online). Immunogold staining of neonatal heart tissue using specific antibodies labelled with two different sizes of gold particles allowed colocalization of AKAP18δ, PLN and SERCA2 by electron microscopy (Fig 2B). As evident from the ultrastructure, all three proteins (85–95% of grains) localized on stacks of sarcoplasmic reticulum that were interspersed with the contractile machinery. Approximately 10% of AKAP18δ and SERCA2, 12% of AKAP18δ and PLN, and 20% of PLN and SERCA2 colocalized within distances of less than 60 nm, indicating significant colocalization using this technique. Similar data were obtained by examination of adult heart tissue. The presence of AKAP18δ, PLN and SERCA2 in such close proximity suggests the possibility of a SERCA2–PLN–AKAP18δ–PKA supramolecular complex in heart tissue.

AKAP18δ interacts with PLN

To examine whether AKAP18δ forms a complex with PKA in the sarcoplasmic reticulum, a cAMP pull-down experiment from sarcoplasmic reticulum fractions using Rp-8-AHA-cAMP-agarose beads was carried out. An approximately 50-kDa protein was recognized by a specific AKAP18δ antibody in the eluate, which also contained PKA-RIIα, PKA-C and SERCA2 (Fig 3A). As SERCA2 co-purified on cAMP-agarose, we tested next whether AKAP18δ forms a complex with the PKA substrate PLN. Immunoprecipitation of AKAP18δ from sarcoplasmic reticulum fractions showed the presence of PLN in the precipitate (Fig 3B). PLN also co-precipitated with AKAP18δ from the left ventricles of adult rat hearts (data not shown). Interestingly, in the opposite immunoprecipitation experiment using anti-PLN, AKAP18δ was not detected, presumably because the epitope for the PLN antibody overlapped the AKAP18δ-binding site (Fig 3D; data not shown), but it could also be due to AKAP18δ, like SERCA2, interacting only with the monomer PLN but not the pentamer population of PLN. However, immunoprecipitation of green fluorescent protein (GFP)–PLN, but not GFP, from HEK293 cells co-transfected with AKAP18δ in the absence of SERCA2 pulled down AKAP18δ, as detected by immunoblotting (Fig 3C). Conversely, immunoprecipitation of AKAP18δ co-precipitated GFP–PLN but not GFP (Fig 3C). Collectively, these experiments indicate that AKAP18δ forms a complex with PLN and SERCA2 in situ through interaction with PLN.

AKAP18δ binds to the cytoplasmic domain of PLN

As AKAP18δ does not have any transmembrane domains or lipid anchors, we examined next the cytoplasmic domain of PLN to identify the AKAP18δ-binding site. The PLN (1–36) sequence was synthesized as 20-mer peptides with 2-amino-acid offset on cellulose membranes and analysed for AKAP18δ binding by overlay with purified, recombinant glutathione-S-transferase (GST)–AKAP18δ protein, followed by anti-GST immunoblotting (Fig 3D). GST alone was used as a negative control. To evaluate the specificity of the assay, an AKAP18δ–PLN disruptor peptide (described below) was included in the overlay, which inhibited binding. The AKAP18δ core binding sequence was mapped to amino acids 13–20 in PLN (Fig 3D, underlined). This sequence is positioned at the end of domain IA (amino acids 1–16) and in the loop domain (amino acids 17–21), which is part of the hinge region between the two helical domains of PLN (Metcalfe et al, 2004). The identified AKAP18δ-binding site overlaps with the PKA phosphorylation site (RRAS) in PLN. To analyse whether PKA phosphorylation affected the AKAP18δ–PLN interaction, PLN was synthesized with a phosphorylated Ser 16 (pS) to mimic PKA-phosphorylated PLN and spots were overlaid with GST–AKAP18δ (Fig 3D). Introduction of the phosphorylated Ser 16 abolished AKAP18δ binding. Thus, the AKAP18δ–PLN interaction seems to be direct and possibly dynamically regulated by PKA phosphorylation of Ser 16 providing an on/off mechanism.

A two-dimensional peptide array, in which each residue in the PLN sequence from 13 to 23 was replaced with all natural amino acids, was analysed for AKAP18δ binding (see supplementary Fig S2 online). Substitutions of Arg 13, Arg 14 and Pro 21 almost totally abolished AKAP18δ binding, indicating the relevance of these amino acids for binding. The importance of Arg 13 and Arg 14 was also shown in a proline scanning array of PLN (data not shown). Interestingly, deletion of Arg 14 is associated with inherited human dilated cardiomyopathy and premature death (Haghighi et al, 2006). Conversely, the PLN-binding site in AKAP18δ was delineated by deletional mapping and interaction analysis by overlaying arrays of the cytoplasmic domain of PLN with truncated GST–AKAP18δ proteins, and by coexpression and co-immunoprecipitation analysis (see supplementary Fig S3 online).

Effect of disrupting PLN–AKAP18δ interaction

To evaluate the significance of AKAP18δ in the coordination of the PKA-mediated phosphorylation of PLN, we made a short peptide from PLN covering the AKAP18δ-binding domain to compete with and displace the AKAP18δ–PLN interaction. Plasma membrane-permeable PLN derivatives were generated by the coupling of 9 or 11 arginine residues to the amino or carboxyl terminus of the peptide (Arg 9–11-PLN or PLN-Arg 9–11). N- or C-terminal coupling, or length of the poly-arginine sequence between 9 and 11 residues did not affect biological activity, as evident from in vitro and in situ testing (data not shown). We then used rat neonatal cardiac myocytes, which have been shown to contain AKAP18δ (Fig 4A), for further functional experiments. The active peptide abolished the striated distribution pattern of AKAP18δ detected by immunofluorescence microscopy, indicating that the peptide disrupts the interaction of the two binding partners, whereas the control peptide did not seem to influence the distribution of AKAP18δ (Fig 4B). Neither peptide affected the distribution of PLN or α-actinin (data not shown). The isoproterenol-induced phosphorylation of PLN-Ser 16 was analysed in the presence or absence of the disruptor peptide (Fig 4C). Neonatal cardiac myocytes were incubated with or without the Arg 9-PLN peptide (Arg 9-RRASTIEMPQQ) for 30 min and then stimulated with isoproterenol. The phosphorylation of PLN at Ser 16 increased fivefold by β-adrenergic stimulation. The presence of the PLN peptide inhibited the increase in phosphorylation by almost 50%, indicating that AKAP18δ is necessary for the recruitment of PKA to its target, PLN. As Ser 16-phosphorylated PLN does not seem to bind to AKAP18δ, we used Arg 9-pSer 16-PLN as a negative control. This had no influence on the phosphorylation level of PLN-Ser 16 after stimulation with isoproterenol (Fig 4C). By contrast, a peptide in which Ser 16 was substituted with cysteine to control for substrate competition was equally effective as the PLN-derived peptide (data not shown). Furthermore, we examined the consequence of disrupting the AKAP18δ–PLN complex on Ca²⁺ re-uptake into the sarcoplasmic reticulum. Neonatal cardiac myocytes were transfected with the fluorescence resonance energy transfer (FRET)-based Ca²⁺ sensor Cameleon D1ER (Palmer et al, 2004) targeted to the sarcoplasmic reticulum, and the response to a 10 mM caffeine pulse was recorded in the presence or absence of 10 μM norepinephrine in control cells or cells pretreated with 25 μM PLN-Arg 11 for 40 min (Fig 4D). Ca²⁺ re-uptake showed a recovery time of approximately 1 min, which is consistent with the kinetics by the Kasai et al (2004) using the same sensor, and reflects simultaneous re-uptake of Ca²⁺ through SERCA2, the release of Ca²⁺ through RYR owing to elevated cytosolic Ca²⁺ and reassociation with Ca²⁺ buffers in the sarcoplasmic reticulum after the depletion. Neonatal cardiac myocytes treated with PLN-Arg 11 showed significantly reduced Ca²⁺ re-uptake into the sarcoplasmic reticulum, both at the basal level and after treatment with norepinephrine (Fig 4D).

To confirm the involvement of AKAP18δ, we knocked down AKAP18δ using short interfering RNA (siRNA; see blot in Fig 5 (right) for siRNA efficacy tested in HaCaT cells) and measured Ca²⁺ re-uptake. Cy3-labelled siRNA was transfected into cardiomyocytes together with the FRET-based Ca²⁺ sensor D1ER. Sarcoplasmic reticulum was Ca²⁺-depleted by blocking SERCA2 with 2,5-di-tert-butylhydroquinone (BHQ, a reversible inhibitor) in a Ca²⁺-free solution, BHQ was washed out, Ca²⁺ was added and Ca²⁺ re-uptake was measured in Cy3-positive cells with the Ca²⁺ sensor in the absence or presence of norepinephrine. As shown in Fig 5, AKAP18δ siRNA oligonucleotides abolished the effect of norepinephrine on Ca²⁺ re-uptake in the sarcoplasmic reticulum, whereas control siRNA did not interfere with the action of norepinephrine.

Our results indicate that AKAP18δ recruitment of PKA to a supramolecular complex containing PLN and SERCA2 is important to discretely regulate PKA phosphorylation of PLN at Ser 16, and thereby the PLN inhibitory effect on SERCA2 and Ca²⁺ re-uptake into heart sarcoplasmic reticulum. Moreover, our data indicate that the β-adrenoceptor/PKA-dependent phosphorylation of PLN requires the interaction of AKAP18δ with PLN. Our results provide a mechanism for the precise spatiotemporal control of PLN phosphorylation through the interaction with AKAP18δ and a pharmacologic tool that can specifically target PKA phosphorylation of a particular substrate and determine its role in the response of failing myocardium to catecholamines and to catecholamine antagonists. Further work will be necessary to determine how and to what extent the manipulation of the SERCA2–PLN–AKAP18δ–PKA complex can be targeted to regulate PLN/SERCA2 function under pathophysiological conditions.