DIFFERENTIAL REGULATED INTERACTIONS OF CALCIUM/CALMODULIN-DEPENDENT PROTEIN KINASE II WITH ISOFORMS OF VOLTAGE GATED CALCIUM CHANNEL BETA SUBUNITS

Abstract

Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) phosphorylates the β_2a subunit of voltage-gated Ca²⁺ channels at Thr498 to facilitate cardiac L-type Ca²⁺ channels. CaMKII colocalizes with β_2a in cardiomyocytes and also binds to a domain in β_2a that contains Thr498 and exhibits amino acid sequence similarity to the CaMKII autoinhibitory domain and to a CaMKII binding domain in the NMDA receptor NR2B subunit (Grueter et al., 2006. Mol. Cell 23:641). Here we explore the selectivity of the actions of CaMKII among Ca²⁺ channel β subunit isoforms. CaMKII phosphorylates the β_1b, β_2a, β₃ and β₄ isoforms with similar initial rates and final stoichiometries of 6–12 mole phosphate per mole protein. However, activated/autophosphorylated CaMKII binds to β_1b and β_2a with similar apparent affinity, but does not bind to β₃ or β₄. Pre-phosphorylation of β_1b and β_2a by CaMKII substantially reduces the binding of autophosphorylated CaMKII. Residues surrounding Thr498 in β_2a are highly conserved in β_1b, but are different in β₃ and β₄. Site-directed mutagenesis of this domain in β_2a showed that Thr498 phosphorylation promotes dissociation of CaMKII-β_2a complexes in vitro and reduces interactions of CaMKII with β_2a in cells. Mutagenesis of Leu493 to Ala substantially reduces CaMKII binding in vitro and in intact cells but does not interfere with β_2a phosphorylation at Thr498. In combination, these data show that phosphorylation dynamically regulates the interactions of specific isoforms of the VGCC β subunits with CaMKII.

Voltage-gated Ca²⁺ channel (VGCC) ion selectivity and responsiveness to pharmacological antagonist ligands are defined by the identity of the pore forming α₁ subunit. The biophysical properties are generally modified by differential association of auxiliary β, α2δ and γ subunits (1–3). Four genes encoding β isoforms have been identified (β_1–4) each having multiple mRNA splice variants, which differentially modulate the properties and cell surface expression of VGCC complexes (4–6). In addition, VGCC complexes are further modulated by a variety of posttranslational modifications.

The regulatory properties of Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) allow integration of signals conveyed by changes in the frequency, duration and amplitude of intracellular Ca²⁺ transients (7). This feature critically depends on Ca²⁺/calmodulin-dependent autophosphorylation of Thr286 in CaMKIIα (or Thr287 in other CaMKII isoforms) in the autoregulatory domain, which confers subsequent autonomous kinase activity (reviewed in (8, 9, 10 )). Thus, Ca²⁺ transients induce more prolonged kinase activation depending on specific parameters of Ca²⁺ signals in the local environment. Recent studies demonstrated that localization of CaMKII to specific subcellular microdomains confers distinct downstream phosphorylation events (11, 12). These studies are consistent with the emerging concept that direct interactions of signaling molecules ensures accurate and timely responses to cell stimulation (13, 14). Determining the mechanisms for CaMKII binding to target proteins, such as VGCCs, is thus an important goal for understanding the role of CaMKII in excitable cells.

CaMKII phosphorylates the α and/or β subunits of a variety of VGCCs to modulate Ca²⁺ entry. For example, CaMKII regulates T-type Ca²⁺ channels by binding to and phosphorylating the II-III intracellular loop in the Ca_V3.2 α subunit (15, 16). The EF hand motif in the C-terminal domain of the L-type Ca²⁺ channel (LTCC) Ca_V1.3 α subunit is also phosphorylated by CaMKII, causing a negative voltage shift in LTCC current activation (17). CaMKII also phosphorylates multiple sites in the LTCC Ca_V1.2 α_1c subunit to promote both Ca²⁺- and voltage-dependent facilitation in heterologous cells (18–20). In addition to these roles in feedback regulation, CaMKII is involved in crosstalk between Ca²⁺ channels: for example, LTCC activation leads to depression of R-type Ca²⁺ channels in dendritic spines via a poorly defined CaMKII-dependent mechanism (21).

We recently showed that CaMKII-dependent facilitation of cardiac Ca_V1.2 LTCCs is mediated by phosphorylation of the β_2a subunit at Thr498 in cardiomyocytes (22). Moreover, β_2a acts as a CaMKII associated protein (CaMKAP) that directly interacts with CaMKII in vitro and in intact cells. Our findings showed that VGCC regulation may be strongly enhanced or modified by association of CaMKII with the β_2a subunit. In the current study, we demonstrate that CaMKII phosphorylates all of the β subunit isoforms, but interacts with β subunits in an isoform specific manner; this CaMKII-β subunit interaction is negatively modulated by phosphorylation of the β subunit.

MATERIALS AND METHODS

Generation of plasmid constructs

The open reading frames of the rat VGCC β_1b, β_2a, β₃ and β₄ subunits (Accession Numbers X61394, M80545, M88751 and L02315) (generous gifts from Dr. E. Perez-Reyes, University of Virginia), were amplified by PCR and ligated into pGEX-4T1 (Amersham Pharmacia Biotech). The β_2a subunit was also subcloned into pFLAG-CMV-2 (Sigma-Aldrich), pIRES (Clontech), and pLenti (Invitrogen). Murine CaMKIIα and rat CaMKIIδ coding sequences were inserted into pcDNA3. The cDNAs encoding β_2a were mutated essentially as described in the QuikChange kit (Stratagene). The pcDNA plasmid encoding a constitutively active T287D mutation of myc-CaMKIIδ 2 was a generous gift from Dr. E. Olson (UTSW, Dallas).

GST fusion protein expression and purification

GST fusion proteins were expressed and purified as described in (22). Protein concentrations were determined by Bradford assay (BioRad) using bovine serum albumin as standard and confirmed by resolving proteins on SDS-polyacrylamide gels followed by Coomassie-Blue staining.

CaMKII purification and autophosphorylation

Recombinant rat CaMKIIδ₂ or mouse CaMKIIα purified from baculovirus-infected Sf9 insect cells were autophosphorylated at Thr287 or Thr286, respectively, using ATP or [γ-³²P]ATP, essentially as described previously (23).

CaMKII plate binding assays

GST fusion proteins in 0.2 ml plate-binding buffer (50 mM Tris-HCl pH 7.5, 200 mM NaCl, 0.1 mM EDTA, 5 mM 2-mercaptoethanol, 0.1% (v/v) Tween-20, 5 mg/ml bovine serum albumin) were incubated for 18–24 hours at 4°C in glutathione-coated wells. After 3 washes with buffer, wells were incubated at 4°C with the indicated concentrations of ³²P-labeled, Thr287 autophosphorylated CaMKIIδ (0.2 ml) for 2 hours then washed (8 times, 0.2 ml ice-cold buffer). The bound kinase was quantified using a scintillation counter.

In order to monitor dissociation of preformed CaMKII-β_2a complexes, GST-β_2a (WT or T498A: ≈5 pmol) was immobilized in glutathione-coated multi-well plates (Pierce, Rockford, IL) and then incubated for 2 hours at 4°C with [³²P-T²⁸⁶]CaMKIIα (0.25 µM) in binding buffer. Wells were rinsed 8 times in binding buffer and immobilized complexes were then incubated with 50 mM Tris-HCl pH 7.5, 0.1 M NaCl, 0.25 mg/ml bovine serum albumin, 0.1% Triton X-100, 1 mM dithiothreitol, 10 mM magnesium acetate, with or without 0.5mM ATP. Soluble/dissociated CaMKII was removed from the wells at the indicated times and quantified by scintillation counting.

CaMKII gel overlays

GST fusion proteins (50 pmoles) were resolved by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose. Approximately equal protein loading was confirmed by staining membranes using Ponceau S. Membranes were blocked and then incubated for 2 hours at 4°C with ³²P-labeled, Thr287 autophosphorylated CaMKIIδ₂ (100 nM), essentially as described (24). After washing, bound CaMKII was quantified using a phosphoimager.

CaMKII phosphorylation of GST-β subunits

Purified GST-β subunits (or GST alone as a blank) were incubated at either 30°C or 4°C in 50 mM HEPES, pH 7.5, 10 mM magnesium acetate, 1 mg/ml bovine serum albumin, 1 mM dithiothreitol, 0.4 mM [γ-³²]ATP (≈ 500 cpm/pmol) or 0.4 mM ATP containing purified CaMKII. At the indicated times aliquots were spotted on P81 phosphocellulose papers, washed in water and counted in a scintillation counter to measure the stoichiometry of phosphorylation. Parallel aliquots were resolved by SDS-polyacrylamide gel analysis, and dried gels were analyzed by autoradiography and/or a phosphoimager followed by densitometry, as described (22).

Immunoblotting

Samples were resolved on Tris-Glycine SDS-polyacrylamide gels and transferred to nitrocellulose membranes using standard methods. Membranes were blocked in 5% (w/v) milk in TTBS (50 mM Tris-HCl, pH 7.5; 0.2 M NaCl, 0.1% (v/v) Tween-20), then incubated with primary antibodies overnight at 4°C. After washing six times for >5 minutes each, membranes were incubated for one hour at room temperature with horseradish peroxidase conjugated secondary antibody. The washed membranes were developed using enhanced chemiluminescence.

Co-immunoprecipitations from HEK293 cells

Experiments were performed as described in (22). Briefly, HEK293FT cells were transfected with FLAG-β_2a (WT, T498A, T498E or L493A), myc-CaMKIIδ (T287D), and/or flag vector alone. Cell lysates were immunoprecipitated with FLAG-coated agarose beads (40 µl: Sigma) and then immunoblotted (see above).

Statistics

Data are expressed as mean±sem. Paired comparisons were performed using the Student’s t-test. Multiple group comparisons were performed using 1-way or 2-way ANOVA with Bonferoni post-hoc testing, unless otherwise noted. The null hypothesis was rejected if p<0.05.

RESULTS

CaMKII efficiently phosphorylates β_1–4 subunits

The highly efficient phosphorylation of the β_2a subunit at Thr498 by CaMKII is critical for CaMKII effects on cardiac LTCCs (22). In order to begin to analyze potential effects of CaMKII on the other β isoforms, we assessed the phosphorylation of full-length β subunit isoforms that had been expressed in bacteria as GST fusion proteins. First, incubations were performed at 4°C in order to provide an indication of the initial rate of phosphorylation by 10 nM CaMKII (Fig. 1A). Under these conditions CaMKII preferentially phosphorylatesThr498 in GST-β_2a (22). All four isoforms were phosphorylated, although β_1b and β_2a were phosphorylated at a relative rate ≈2.5 fold faster than for β₃ or β₄. We then assessed the extent of phosphorylation by various concentrations of CaMKII using a fixed concentration of the GST-β isoforms (1 µM). We previously identified 6 CaMKII phosphorylation sites in β_2a that had been phosphorylated to a stoichiometry of ≈ 3 mol/mol (22). In the present studies, each isoform was phosphorylated with a similar CaMKII concentration-dependence (Fig. 1B), with the β_2a, β_1b, β₃ and β₄ variants attaining stoichiometries of 13.1±1.3, 10.8±0.4, 7.4±0.5 and 9.6±1.7 moles of phosphate/mole β, respectively. While ANOVA detected significant variability in the final phosphorylation stoichiometries (p=0.038), post-hoc testing revealed a single significant difference between β₃ and β_2a phosphorylation stoichiometries (p<0.05). In combination, these data show that these four β subunit isoforms are rapidly phosphorylated by CaMKII to a high stoichiometry in vitro with relatively modest differences in the phosphorylation parameters that were measured.

Figure 1. CaMKII efficiently phosphorylates all of the β subunit isoforms in vitro.

A. Initial rate of phosphorylation of GST-β isoforms by CaMKIIδ2 at 4°C. Data (mean of 2 experiments) are plotted relative to the phosphorylation of GST-β_2a after 20 min. GST-β_2a, ○. GST-β_1b, ●. GST-β₃, ▼. GST-β₄, △. B. Stoichiometry of phosphorylation of GST-β isoforms by increasing concentrations of CaMKIIδ2 after 20 minutes incubation at 30°C (mean±sem of 3 experiments). Symbols as in panel A. *: phosphorylation of GST-β₃ by 1000 nM CaMKII was significantly less than that of GST-β_2a (p<0.05). Other differences were not statistically significant.

Selective interactions of CaMKII with VGCC β subunits in vitro

Thr498 of the β_2a variant lies within a CaMKII-binding domain that is C-terminal to the SH3 and GK domains (Fig. 2A). Alignment of the amino acid sequence of the domain surrounding Thr498 in β_2a with other β subunit isoforms revealed variable conservation. A CaMKII consensus phosphorylation motif LXRXXS/T (25) is present in both β_2a and β_1b and there is additional amino acid sequence similarity outside this motif, but key residues from this motif are missing in β₃ and β₄. Based on these alignments, we hypothesized that CaMKII would bind to β_1b in a similar manner to its interaction with β_2a, but not to β₃ or β₄. To test this hypothesis we immobilized GST-β fusion proteins in glutathione coated 96-well plates and then incubated them with various concentrations of purified [³²P]-autophosphorylated CaMKII. In an initial experiment, the binding of CaMKII to GST-β_1b was indistinguishable from the binding to GST-β_2a, but binding to GST-β₃ and GST-β₄ was <1% of the binding to GST-β_2a (Fig. 2B). Binding of Thr286-autophosphorylated CaMKII to both GST-β_1b and GST-β_2a was concentration dependent and saturable (Fig. 2C). GST-β_1b exhibited a significantly higher apparent affinity for CaMKII than did GST-β_2a (apparent K_d's 35±12 and 120±21 nM CaMKII subunit, respectively. n=4. p<0.01). GST-β₃ and GST-β₄ failed to bind significant amounts of the kinase even at a concentration of 1200 nM CaMKII subunit (data not shown). Autophosphorylated CaMKII also binds to GST-β_1b and GST-β_2a, but not to GST-β₃ and GST-β₄ in glutathione agarose cosedimentation assays, but there was no significant binding of nonphosphorylated CaMKII to any GST-β subunit isoforms (Supplementary Fig. 1). Thus, CaMKII does not appear to bind significantly to the β₃ and β₄ isoforms, but interacts with the β_1b isoform with an ≈3-fold higher affinity than with β_2a following activation by Thr286/7 autophosphorylation.

Figure 2. CaMKII association with VGCC β subunit isoforms.

A. Schematic domain structure of the β_2a subunit showing SH3 and guanylate kinase-like (GK) domains, with the C-terminal CaMKII-binding domain containing the Thr498 phosphorylation site (indicated by the “P”). The amino acid sequence of the CaMKII binding domain in β_2a is aligned with similar sequences from other β isoforms and the CaMKIIδautoregulatory domain (surrounding Thr287): identical residues are shown in gray boxes. B. GST-β subunit isoforms were immobilized in glutathione-coated multi-well plates (100 pmol/well) and incubated with ³²P-labeled Thr287 autophosphorylated CaMKIIδ2 (50 nM subunit). C. Concentration dependent binding of Thr287 autophosphorylated CaMKIIδ2 to GST-β_2a (○) and GST-β_1b (●) in glutathione-coated multi-well plates. Both panels B and C plot binding as mean±sem from 4 observations (β_2a and β_1b) or the mean of 2 observations (β₃ and β₄).

Interaction with CaMKII is modulated by phosphorylation in vitro

Thr498 and several additional serine and threonine residues lie within the CaMKII binding domain of β_2a, suggesting that the interaction with CaMKII may be modulated by phosphorylation. Therefore, GST-β isoforms were preincubated with activated CaMKII in the presence or absence of ATP and then separated from reaction components by SDS-PAGE. The electrophoretic mobility of each GST-β isoform was reduced following preincubation with ATP and CaMKII (Fig. 3: protein stain), consistent with the relatively high phosphorylation stoichiometry under these conditions (c.f., Fig. 1B). An overlay assay was then used to assess the binding of ³²P-autophosphorylated CaMKII. Non-phosphorylated GST-β_1b and GST-β_2a proteins bound substantial, but comparable, amounts of CaMKII (Fig. 3). Pre-phosphorylation significantly reduced CaMKII binding to both proteins by 70–80%. No significant interactions were seen between CaMKII and either GST-β₃ or GST-β₄, whether or not these protein were pre-phosphorylated. In combination, these data suggest that phosphorylation of the β subunit by CaMKII modulates the binding of CaMKII to the β_1b and β_2a isoforms.

Figure 3. Prephosphorylation of GST-β_2a and GST-β_1b inhibits CaMKII binding.

GST-β isoforms were preincubated with activated CaMKIIδ2 in the presence or absence of ATP, as indicated. Reactions were resolved by SDS-PAGE and transferred to nitrocellulose membrane for overlay with [³²P]-labeled Thr287-autophosphorylated CaMKIIδ2. The top panel shows a protein stain of the membrane prior to overlay and a representative autoradiograph to detect bound CaMKII. Binding was quantified using a phosphoimager and normalized to the binding detected using non-phosphorylated GST-β_2a: the mean±sem from 3 experiments is plotted. Data were analyzed by 2-way ANOVA: *: p<0.001 vs. binding to non-phosphorylated GST-β_2a. #: p<0.05 vs. binding to the corresponding non-phosphorylated protein.

The mechanism of CaMKII binding to β_2a

In order to dissect the mechanism for CaMKII binding we assessed the effect of mutating residues within the CaMKII binding domain of β_2a on the interaction with CaMKII. Mutation of Thr498 to Ala or Glu prevented or mimicked phosphorylation at this site, respectively (see above). In addition, residues homologous to Leu493 at the p-5 position relative to Thr498 in β_2a and β_1b are conserved in other high affinity CaMKII phosphorylation sites that form stable complexes with the CaMKII catalytic domain prior to phosphorylation (e.g., Ser1303 in NR2B and Thr286/7 in CaMKIIα/δ). However, hydrophobic residues at the p-5 position are not conserved in β₃ or β₄, and are not generally considered to be part of the minimal consensus phosphorylation site (25). Therefore, we also mutated Leu493 to Ala. Binding of ³²P-autophosphorylated CaMKII to GST-β_2a in glutathione coated multi-well plates was unaffected by the T498A mutation, whereas the T498E and L493A mutations significantly reduced CaMKII binding by >90% (Table 1). These data demonstrate that the region surrounding Thr498 is critical for stable binding of Thr286-autophosphorylated CaMKII to the full-length β_2a isoform.

Table 1. Disruption of CaMKII binding to GST-β_2a by site-directed mutagenesis.

Purified GST-β_2a proteins (wild type or with the indicated point mutations) or GST alone were immobilized in a glutathione-coated 96-well plate (100 pmol/well) and then incubated with ³²P-labeled Thr287-autophosphorylated CaMKIIδ₂ (100 nM subunit). After washing, bound CaMKII was quantified by scintillation counting. The data indicate the mean±sem (n = 3 experiments).

GST-β2a Protein	CaMKII bound
	(pmol subunit)
Wild type	7.9 ± 0.3
T498A	6.2 ± 0.8
T498E	0.51 ± 0.16
L493A	0.14 ± 0.03

Disruption of CaMKII binding does not affect phosphorylation of β_2a in vitro

We then explored the effects of the Thr498 and Leu493 mutations on CaMKII phosphorylation of the β_2a subunit. The T498A mutation substantially reduced the rate of β_2a phosphorylation at 4°C, as shown previously, and T498E mutation had a very similar effect. Interestingly, the rate of L493A-β_2a phosphorylation by CaMKII was indistinguishable from the rate of phosphorylation of the wild type protein (Fig. 4A), even though the L493A mutation interfered with CaMKII binding. When incubations were conducted at 30°C, CaMKII (10 nM) phosphorylated wild type β_2a to a stoichiometry of 2.2 mol/mol, consistent with stoichiometries reported in Fig. 1B using 10 nM CaMKII, but T498A mutation reduced phosphorylation to 0.3 mol/mol (Fig. 4B). In contrast, T498E and L493A mutations had no significant effect on the stoichiometry of CaMKII phosphorylation, although there was a trend for reduced phosphorylation of T498E-β_2a. These data show that two mutations that severely compromise stable binding of activated CaMKII to β_2a have no significant effect on the overall stoichiometry of CaMKII phosphorylation in vitro.

Figure 4. Phosphorylation of GST-β_2a by CaMKII is independent of stable binding.

A. Initial rate for phosphorylation of GST-β_2a (wild type, ●: T498A,▼: T498E,○: L493A, △ ) by CaMKII (10 nM) at 4°C. Data are the mean±sem of 3–4 observations normalized to the phosphorylation of the wild-type protein after 20 minutes. B. Wild type or mutated GST-β_2a proteins or GST alone were incubated for 20 min at 30°C with activated CaMKII (10 nM) and [γ-³²P]ATP. Aliquots of the reactions were resolved by SDS-PAGE and transferred to nitrocellulose membranes. Membranes were stained for total protein (bottom), and then immunoblotted using a phospho-specific antibody that was raised to phospho-Thr286 in CaMKIIα, but also detects phospho-Thr498 in β_2a (middle). Total phosphorylation of β_2a was detected by autoradiography (top gel). The stoichiometry of total phosphorylation (bar graph) was estimated after spotting parallel aliquots on phosphocellulose papers (see Methods). Data are plotted as mean±sem (n=4 experiments). *: p<0.01 vs. WT.

As noted above, Leu493 is part of a conserved sequence surrounding high-affinity phosphorylation sites in multiple CaMKII-binding proteins. In order to determine whether L493A mutation altered the phosphorylation site specificity of CaMKII, we exploited similarities in protein sequences surrounding Thr498 in β_2a and Thr287 in CaMKIIδ (Fig. 2A). Commercially available phospho-Thr287 antibodies detected the β_2a subunit following CaMKII phosphorylation, but not nonphosphorylated β_2a (Fig. 4B and data not shown). Phospho-Thr286/7 antibodies only weakly detected phosphorylated β_1b and failed to detect β₃ or β₄ before or after CaMKII phosphorylation (data not shown). Analysis of mutated β_2a proteins revealed that T498A-β_2a and T498E-β_2a could not be detected (Fig. 4B, middle blot), demonstrating that the antibody specifically recognized phospho-Thr498 in β_2a, in addition to the Thr286/287 autophosphorylation site in CaMKII. The lack of antibody cross-reactivity with T498E-β_2a is not surprising given differences in size and charge density between glutamate and phospho-threonine. Taken together, these findings validate the use of this antibody to report the phosphorylation status of Thr498 in the CaMKII binding domain of β_2a. Notably, Thr498 was phosphorylated to a similar extent in L493A-β_2a and wild type β_2a (Fig. 4B, inset). Thus, L493A mutation does not appear to significantly alter the ability of CaMKII to phosphorylate Thr498 in vitro.

Phosphorylation at Thr498 disrupts CaMKII binding

In order to explore the mechanism by which phosphorylation of β_2a by CaMKII interferes with subsequent binding of activated CaMKII, we pre-phosphorylated wild type and mutated β_2a proteins and analyzed CaMKII binding using overlay assays. As seen using the glutathione plate-binding assay, T498A-β_2a bound comparable amounts of activated CaMKII to wild type β_2a. However, pre-phosphorylation of T498A-β_2a had no significant effect on binding of activated CaMKII, whereas pre-phosphorylation of wild type β_2a significantly reduced binding by ≈80% in these assays. The T498E mutation reduced CaMKII binding by ≈80% and CaMKII phosphorylation had no significant additional effect on binding to T498E-β_2a. Interestingly, L493A mutation significantly reduced binding by ≈90% in these assays, and pre-phosphorylation by CaMKII resulted in an additional significant decrease in binding. Taken together, these data suggest that pre-phosphorylation of β_2a at Thr498 substantially reduces the association of CaMKII.

We also investigated the role of Thr498 phosphorylation in the context of preformed CaMKII-β_2a complexes. Complexes of autonomously-active Thr286 autophosphorylated CaMKII bound to GST-β_2a(WT) were isolated in glutathione-coated multi-well plates and then incubated with or without addition of ATP. Addition of ATP to these complexes induced phosphorylation at Thr498 and several other sites, as reflected by immunoblotting with phospho-Thr286 CaMKII antibodies and by a substantial reduction in the electrophoretic mobility of GST-β_2a (data not shown). In the absence of ATP, CaMKII complexes with WT β_2a were remarkably stable (<5% dissociation over a 2 hour incubation), but addition of ATP induced dissociation of ≈50% of bound CaMKII. T498A mutation of β_2a had little effect on the stability of complexes with CaMKII in the absence of ATP, but substantially reduced the rate and extent of dissociation following addition of ATP as compared to WT β_2a (Fig. 6). However, ATP still enhanced dissociation of CaMKII from preformed complexes with T498A-β_2a (≈20% of in 2 hours), suggesting that phosphorylation at other residues in β_2a and/or CaMKII may have a modest effect on the stability of these complexes, at least in vitro. In combination, these data suggest that phosphorylation of β_2a at Thr498 enhances the dissociation of CaMKII from preformed complexes with β_2a in vitro.

Figure 6. Phosphorylation of β_2a at Thr498 enhances dissociation of the CaMKII-β_2a complex.

Complexes of GST-β_2a (WT or T498A) and [³²P-T²⁸⁶]CaMKIIα in glutathione-coated multi-well plates were incubated with or without ATP in a dissociation buffer (see Methods). At the indicated times, dissociated [³²P-T²⁸⁶]CaMKIIα was removed from wells and quantified by scintillation counting. Data points represent mean±sem (n=3) (n=2 for β_2a T498A in absence of ATP): error bars lie within symbols of some data points.

CaMKII interaction with β_2a is regulated in cells

Initially, we explored whether Thr498 in β_2a is phosphorylated in intact cells. Lysates of HEK293 cells expressing FLAG-β_2a (wild type or mutated) with a constitutively active CaMKIIδ₂ mutant were immunoprecipitated using FLAG antibodies. Blotting the immune complexes with FLAG antibodies revealed a relatively consistent expression and immunoprecipitation of the β_2a proteins. The phospho-Thr286 CaMKIIα antibodies detected the β_2a wild type and L493A mutant, but not T498A- or T498E-β_2a proteins (Fig. 6A). These data show that Thr498 can be phosphorylated to a similar extent in WT β_2a and L493A-β_2a in intact cells.

In order to investigate the regulation of CaMKII interaction with β_2a in intact cells by modification of Thr498 and the surrounding domain, lysates of HEK293 cells expressing constitutively active myc-tagged CaMKIIδ without or with FLAG-tagged β_2a subunits (WT or mutated) were immunoprecipitated using FLAG antibodies. CaMKII was readily detected in immune complexes containing FLAG-tagged wild-type β_2a or T498A-β_2a, but could not be detected in immune complexes formed by T498E-β_2a of L493A-β_2a (Fig. 6B). These findings demonstrate that the domain surrounding Thr498 is critical for association of CaMKII with β_2a in cells, and suggest that phosphorylation at Thr498 diminishes CaMKII binding to β_2a in cells.

DISCUSSION

A diverse family of VGCCs regulates Ca²⁺ entry into excitable and non-excitable cells. The α₁ subunits confer core biophysical and pharmacological behavior of each type of VGCC. However, a cytosolic loop between the first and second major transmembrane domains in the α subunit is generally thought to constitutively interact with a β subunit. Alternative splicing of mRNAs from four mammalian genes generates >20 distinct β subunit proteins that have divergent roles in modulating the trafficking and biophysical properties of VGCCs (6). The β subunits share highly conserved SH3 and GK domains that form a compact structure, with a hydrophobic groove in the GK domain that interacts with the α subunit (26). However, some β subunits lack substantial parts of the SH3 and GK domains, yet still modulate LTCCs (27). These observations support recent findings showing that additional SH3-GK independent modulatory interactions between the α and β subunits are important for regulating VGCC activity (28–31). The variable domains presumably account for the unique effects of β subunit variants on the properties of α subunits.

Most studies have focused on the roles of protein interactions and post-translational modifications of the α subunit in modulating VGCCs. However, the importance of β subunits in regulating VGCCs has received increasing attention. Initial studies suggested that PKA phosphorylation of the β₂ subunit plays a role in facilitating LTCCs (32), although the importance of this modification in native cells has recently been questioned (33). In addition, β subunits have been shown to serve as scaffolding proteins that bind REM GTPases to inhibit VGCCs (34, 35), and AHNAKs to link VGCCs to the actin cytoskeleton (36). Our recent work showed that CaMKII binds β_2a subunits and colocalizes with β_2a in adult cardiomyocytes. Moreover, phosphorylation of β_2a at Thr498 is required for Ca²⁺ and CaMKII-dependent facilitation of LTCCs in cardiomyocytes (22). While β subunit variants have unique direct effects on the biophysical properties and trafficking of LTCCs, the β isoform-selectivity of regulation by CaMKII and other modulators is poorly understood.

The β subunit variants tested here associate with multiple VGCC α subunits (reviewed in (2, 6)). Thus, our current results showing that activated/Thr286 autophosphorylated CaMKII forms stable complexes with β_1b and β_2a, but not with β₃ or β₄, lead us to hypothesize that association of CaMKII with VGCC complexes will depend on the identity of the associated β subunit. The β_1b and β_2a, subunits contain an LXRXXS/T motif similar to sequences surrounding phosphorylation sites in NR2B and the CaMKII autoinhibitory domain that also form stable complexes with the CaMKII catalytic domain. Mutation of Leu493 to Ala within this motif reduced CaMKII binding to β_2a by >90%, but surprisingly had little effect on the initial rate of phosphorylation at Thr498, or on the overall phosphorylation stoichiometry at multiple sites. The lack of conservation of this motif accounts for the failure to detect binding to β₃ or β₄.

Despite the apparent binding selectivity, CaMKII phosphorylates all four β subunit variants tested here with comparable relative rates and overall extent. The high maximal phosphorylation stoichiometries suggest that CaMKII efficiently phosphorylates several sites in each β isoform in vitro. Indeed, we previously identified 6 CaMKII phosphorylation sites in β_2a (22), and it will be important to identify phosphorylation sites in other β isoforms and determine their impact on the properties of VGCCs. Data presented here show that phosphorylation at Thr498 in β_2a negatively regulates CaMKII binding both in vitro and in situ. Presumably the conserved LXRXXS/T motif in β_1b is responsible for the regulated binding of CaMKII to this variant. Thus, interactions of CaMKII with β and/or β₁ variants are regulated by phosphorylation, likely playing an important role in modulating CaMKII targeting to LTCCs and/or other VGCCs.

Interestingly, β subunits are not generally thought to be important in the regulation of T-type VGCCs. However, recent studies show that CaMKII directly interacts with and phosphorylates the II-III linker of the Ca_V3.2 α subunit, shifting the current-voltage activation curve (15, 16). In addition, recent data suggest that CaMKII may also interact with multiple cytoplasmic domains in Ca_V1.2 (19). Thus, association of CaMKII with VGCC α subunits may play additional important feature allowing localized Ca²⁺ concentrations to feedback and regulate Ca²⁺ influx.

Emerging data over the last couple of years suggest that mechanisms underlying CaMKII modulation of LTCCs are complex. Phosphorylation of multiple sites in the Ca_V1.2 α subunit by CaMKII appears to play distinct roles in voltage-and calcium-dependent modulation in heterologous cells (18, 20). In addition, interactions of CaMKII with multiple intracellular domains of the Ca_V1.2 α subunit have been implicated in LTCC facilitation (19). In contrast, phosphorylation at Thr498 in β_2a is required for CaMKII to increase channel open probability at the single channel level and to facilitate whole cell Ca²⁺ currents in adult cardiomyocytes (22). The present observations that phosphorylation at Thr498 appears to be relatively independent of the stable binding of CaMKII to β_2a (Fig. 4), and also inhibits CaMKII binding (Fig. 3, Fig. 5, Fig. 6), provides insights that may reconcile these seemingly disparate observations. Interestingly, phosphorylation of β_2a in preformed complexes appears to promote dissociation of CaMKII (Fig. 6). CaMKII dissociation might be required to allow protein phosphatases to act on the phospho-Thr498 site to reset channels to their basal state. Presumably, such a mechanism would be most relevant if Thr498 phosphorylation directly modulates LTCC properties, as suggested by the importance of Thr498 in CaMKII-mediated increases in LTCC open probability and LTCC current facilitation in adult cardiomyocytes (22). Secondly, CaMKII may serve a structural role to dynamically assemble complexes containing additional, as yet unidentified, proteins associated with LTCCs. Thus, Thr498 phosphorylation in β_2a may promote re-organization of these protein complexes to enable facilitation. Structural roles for CaMKII have been postulated in neuronal postsynaptic densities (23). Previous studies showed that disruption of the actin- or tubulin-based cytoskeleton essentially abrogated CaMKII-dependent facilitation of cardiac LTCCs without affecting PKA-dependent facilitation, suggesting an important role for higher orders of molecular organization close to LTCCs in mediating the effects of CaMKII (37). A final possibility is that dissociation from β_2a induced by Thr498 phosphorylation is important in allowing activated CaMKII to phosphorylate nearby regulatory sites in the α_1c or β subunits and/or other associated proteins. In other words, stable binding of activated CaMKII to β_2a may limit access of phosphorylation sites in other proteins to the CaMKII active site. Consistent with this hypothesis, dissociation of CaMKII from β_2a is substantially reduced by T498A mutation, perhaps explaining why CaMKII phosphorylates this mutant to a lower final stoichiometry than T498E-β₂ and L493A-β₂, which have much weaker interactions with CaMKII (Fig. 4B). Further studies will be needed to clarify the mechanism of CaMKII actions at VGCC complexes and, perhaps, to identify new CaMKII targets.

Figure 5. Pre-phosphorylation of β_2a at Thr498 inhibits the association of CaMKII.

A. Wild type and mutated GST-β_2a proteins were preincubated with activated CaMKIIδ2 in the presence or absence of ATP. Samples were resolved by SDS-PAGE and transferred to nitrocellulose membranes to detect total protein loaded (top) and to probe with [³²P]CaMKIIδ2 by overlay assay and autoradiography (middle). Binding was quantified using a phosphoimager and normalized to binding to non-phosphorylated wild type protein: the mean±sem from 3 experiments is plotted. #: p<0.001 vs. corresponding non-phosphorylated protein. *: p<0.001 vs. non-phosphorylated GST-β_2a wild type. Further post analyses using t tests showed that pre-phosphorylation significantly reduced CaMKII binding to the L493A mutant (†: p<0.05).

In summary, findings reported here and in other recent papers suggest that feedback regulation of Ca²⁺ influx via VGCCs is precisely controlled in specific subcellular microdomains by multiple mechanisms that allow CaMKII and other Ca²⁺-dependent signaling proteins to associate with channel subunits. The precise nature of the feedback regulation by CaMKII seems likely to depend on the identity of the β subunit associated with the complex. The regulated interaction of activated CaMKII with β₁ and β₂ variants seem likely to be important, although phosphorylation of β₃ and β₄ may also play a role in some cases. Our findings are in line with recent studies suggesting that subcellular targeting of CaMKII via its interactions with CaMKAPs modulates the specificity of its downstream actions (11, 12). These complex biochemical mechanisms for feedback regulation of Ca²⁺ influx via VGCCs presumably provide great flexibility for modulating a variety of downstream signaling events such as cardiac excitation-contraction and excitation-transcription coupling and neuronal synaptic plasticity. Moreover, alterations in the association of β subunits with VGCCs might disrupt feedback regulation and downstream signaling in heart failure and other diseases (38).

Figure 7. CaMKII interaction with β_2a is regulated by Thr498 phosphorylation in situ.

Myc-tagged T287D-CaMKIIδ2 was co-expressed with or without FLAG-tagged wild type or mutated β_2a proteins in HEK293 cells. Aliquots of the cell lysates (Inputs), FLAG immune supernatants (Super) and FLAG immune complexes (Pellets) were immunoblotted. A: Probing FLAG immune complexes using antibodies raised to phospho-Thr286 in CaMKIIα showed that WT and L493A β_2a are partially phosphorylated at Thr498 in intact cells. B: Probing FLAG immune complexes using CaMKIIδ antibodies revealed that CaMKII was associated with WT and T498A β_2a but could not be detected in the T498E and L493A β_2a immune complexes. The data are representative of >4 experiments.

ABBREVIATIONS

CaMKII

calcium/calmodulin-dependent protein kinase II

CaMKAP

CaMKII-associated protein

GK

guanylate kinase-like

GST

glutathione-S-transferase

LTCC

L-type VGCC

VGCC

voltage-gated calcium channel