|Home | About | Journals | Submit | Contact Us | Français|
Calcium/calmodulin-dependent kinase II (CaMKII) facilitates L-type Calcium Channel (LTCC) activity physiologically, but may exacerbate LTCC-dependent pathophysiology. We previously showed that CaMKII forms stable complexes with voltage-gated calcium channel β1b or β2a subunits, but not with the β3 or β4 subunits (Grueter et al 2008, Biochemistry, 47:1760–1767). CaMKII-dependent facilitation of CaV1.2 LTCCs requires Thr498 phosphorylation in the β2a subunit (Grueter et al, 2006, Mol Cell, 23:641–650), but the relationship of this modulation to CaMKII interactions with LTCC subunits is unknown. Here we show that CaMKII co-immunoprecipitates with forebrain LTCCs that contain CaV1.2α1 and β1 or β2 subunits, but is not detected in LTCC complexes containing β4 subunits.4 CaMKIIα can be specifically tethered to the I/II linker of CaV1.2α1 subunits in vitro by the β1b or β2a subunits. Efficient targeting of CaMKIIα to the full-length CaV1.2α1 subunit in transfected HEK293 cells requires CaMKII binding to the β2a subunit. Moreover, disruption of CaMKII binding substantially reduced phosphorylation of β2a at Thr498 within the LTCC complex, without altering overall phosphorylation of Cav1.2α1 and β subunits. These findings demonstrate a biochemical mechanism underlying LTCC facilitation by CaMKII.
Voltage-activated L-type calcium channels (LTCCs) generate Ca2+ signals for multiple physiological processes, including muscle contraction, neurotransmission, neuronal plasticity, dendritic arborization, gene expression and secretion (Catterall 2000). Electrophysiological and pharmacological properties of LTCCs are determined by their pore-forming α1 subunits (CaV1.1 through CaV1.4), but auxiliary subunits (β1–4, α2-δ and γ) modulate their trafficking, surface expression and biophysical properties (reviewed in (Birnbaumer et al. 1998, Catterall 2000, Dolphin 2003)). Most splice variants of the β subunits contain SH3 and Guanylate Kinase (GK)-like domains that are typical of membrane-associated guanylate kinase (MAGUK) family scaffolding proteins. The GK-like domains are responsible for primary interaction of β subunits with the Alpha Interacting Domain (AID) located in the cytosolic I/II linker region of α1 subunits (Van Petegem et al. 2004). Thus, LTCC β subunits make ideal molecular scaffolds for anchoring cytosolic regulatory proteins to fine-tune Ca2+ signals near LTCCs. These interactions contribute to the plethora of molecular interactions that have been implicated in regulating LTCC activity (reviewed in (Hidalgo & Neely 2007, Calin-Jageman & Lee 2008, Dai et al. 2009).
The multifunctional calcium/calmodulin (Ca2+/CaM)-dependent protein kinase II (CaMKII) is a key downstream effector of LTCC signals. In myocytes, CaMKII is required for normal coupling of excitation to contraction (reviewed in (Couchonnal & Anderson 2008, Bers 2008)) and transcription (Wu et al. 2006, Zhang et al. 2007, Bossuyt et al. 2008). Furthermore, CaMKII was recently shown to regulate excitation-transcription coupling in neurons (Wheeler et al. 2008). All isoforms of CaMKII are activated by Ca2+/CaM binding and undergo rapid autophosphorylation at Thr286/7 to generate a Ca2+-independent form with enhanced affinity for several CaMKII-associated proteins that may target CaMKII to specific subcellular compartments (Griffith et al. 2003, Colbran 2004, Merrill et al. 2005). In addition to these downstream actions, CaMKII doubles as a feedback regulator of LTCCs (Anderson et al. 1994, Yuan & Bers 1994, Xiao et al. 1994). CaMKII participates in several forms of CaV1.2 LTCC facilitation, including voltage-dependent facilitation induced by a single, strong depolarization, Ca2+-dependent facilitation induced during mild repeated depolarizations, and an increase in channel open probability (Anderson et al. 1994, Xiao et al. 1994, Lee et al. 2006, Dzhura et al. 2000). Hormones and neurotransmitters can also facilitate both CaV1.2 and CaV1.3 LTCCs by CaMKII-dependent mechanisms (O-Uchi et al. 2005, Gao et al. 2006).
LTCCs are localized in specific subcellular compartments that also contain CaMKII, such as at the t-tubules (z lines) of cardiomyocytes or in neuronal dendritic spines (Gao et al. 2006, Grueter et al. 2006, Lee et al. 2009, Yasuda et al. 2003, Wu et al. 1999, O-Uchi et al. 2005). CaMKII-dependent LTCC facilitation is thought to require local Ca2+-dependent activation of CaMKII at the channel because facilitation can be blocked by a “fast” Ca2+ chelator (BAPTA), but not a slow chelator (EGTA) (Xiao et al. 1994). Indeed, CaMKII associated with the cytosolic face of membrane patches excised from cardiomyocytes can regulate LTCC open probability. In addition, cytoskeletal disruption interferes with CaV1.2 LTCCs regulation by CaMKII in cardiomyocytes (Dzhura et al. 2002). However, the precise molecular basis for CaMKII targeting to and regulation of LTCCs remains unresolved.
Studies in heterologous cells suggest that CaMKII may directly bind to and/or phosphorylate the CaV1.2α1 subunit to support multiple forms of LTCC facilitation (Hudmon et al. 2005, Lee et al. 2006, Erxleben et al. 2006). However, β subunits have been shown to play an important role in LTCC facilitation (Kamp et al. 2000, Bourinet et al. 1994, Zhang et al. 2005). We recently showed that phosphorylation of Thr498 in the β2a subunit is essential for facilitation of native LTCCs in cardiomyocytes (Grueter et al. 2006). We also showed that CaMKII can form stable complexes with β1b and β2a subunits but not with β3 or β4 subunits (Grueter et al. 2008). Thus, β1 and β2 subunits may be key loci for CaMKII-dependent regulation of LTCCs.
Here, we explored the role of β subunits in targeting CaMKII to LTCC α1 subunits. We show that CaMKII selectively associates with forebrain LTCCs containing β1 or β2 subunits, but not the β4 subunit. Studies in heterologous cells demonstrate that CaMKII interaction with the β subunit is critical for assembly with the LTCC complex and for phosphorylation of a key regulatory site in β2a (Thr498) within these complexes.
A pLenti6-V5-D-Topo (Invitrogen, Carlsbad, CA) construct containing the rabbit cardiac CaV1.2α1 open reading frame (Accession Number X15539) with a surface HA-epitope between amino acids F709 and D710 of domain II S5-H5 extracellular loop (hereafter called HA-Cav1.2α1) was a generous gift of Dr. William Thiel (Thiel et al. 2008). The I/II linker was amplified from the parent construct and inserted into a pGEX 4T-1 vector (Amersham Pharmacia Biotech, Uppsala, Sweden). The pGEX 4T-1 encoding the Cav1.2 AID was a generous gift of Dr. Rong Zhang. The open reading frames of the rat β1b, β2a, β3 and β4 (Accession Numbers X61394, M80545.1, M88751 and L02315 respectively) were generous gifts from Dr. Timothy Kamp and Dr. Ed Perez-Reyes. The pGEX4T-1 and pLenti6-V5-D-Topo vectors for expression of β subunits were described previously (Grueter et al. 2006, Grueter et al. 2008). The β subunit cDNAs were also amplified by PCR and ligated into pGEX4T-1, pRSET A (Invitrogen), pFLAG-CMV2 (Sigma-Aldrich) and pHA-CMV5 (Clontech) for expression of affinity-tagged β subunits in bacteria or HEK293 cells. The pEGFP (Clonetech, Mount View, CA) construct encoding murine CaMKII with an N-terminal EGFP tag was made by Dr. Stefan Strack. Site-directed mutagenesis was performed essentially as described in the QuikChange kit (Stratagene, La Jolla, CA).
The following procedure for preparation of Triton X-100 soluble fractions from rat or mouse brains was modified from Davare et al (Davare et al. 1999). All procedures were performed at 4°C. Forebrains from 7- to 8-week old Sprague-Dawley rats (Pel-Freeze Biologicals, Rogers, Arkansas) or from 12- to 14-week old C57B6J (Jackson Labs, West Grove, Pennsylvania) were pulverized under liquid nitrogen. The frozen powder was homogenized in Teflon-glass Potter-Elvehjem tissue grinders (Kontes, Vineland, NJ) in 5 to 10 ml of Solubilization Buffer (10 mM Tris-HCl, pH 7.5, 1% Triton X-100 (v/v), 20 mM EDTA, 10 mM EGTA, 100 μM PMSF, 500 μM benzamidine, 0.25 μM microcystin LR, 20 mM sodium β-glycerophosphate, 50 mM sodium fluoride, and protease inhibitor cocktail (Sigma, Catalog # P2714)). Homogenates were sonicated for one minute and centrifuged at 3600 × g (4000 rpm) for five minutes in a Beckman J-6B centrifuge using a JS-4.2 rotor to sediment cell/tissue debris and nuclei. Supernatants were removed and then centrifuged for 30 minutes at 250,000 × g in a Beckman L80K ultracentrifuge using a Ti70.1 rotor. The Triton-soluble supernatant was used as the input for immunoprecipitations, and the Triton-insoluble fraction was resuspended in solubilization buffer supplemented with 1% deoxycholate and centrifuged for 30 minutes at 250,000 × g. To prepare membrane-enriched fractions for Fig S1B, the pulverized brains were resuspended in Membrane Buffer (300 mM sucrose, 75 mM NaCl, 20 mM EDTA, 10 mM EGTA, 100 μM PMSF, 500 μM benzamidine, 0.25 μM microcystin LR, 20 mM sodium β-glycerophosphate, 50 mM sodium fluoride, and protease inhibitor cocktail). Extracts were made and centrifuged as described above: supernatants were considered to be the cytosolic fraction, and the membrane-enriched pellets were resuspended in Solubilization Buffer to make a membrane-enriched fraction.
The Triton-soluble fraction (1 to 1.5 ml) was precleared for 1 hour with protein-G Sepharose (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) and then incubated with 10 μg of either mouse monoclonal IgG2a LTCC β subunit antibodies (Neuromab, Davis, Ca) or an affinity-purified polyclonal goat antibody that recognizes all CaMKII isoforms (McNeill & Colbran 1995). Control precipitations were performed using 10 μg of mouse (Jackson Labs, #015-000-003) or goat (Jiao et al. 2008) IgGs of mixed subtypes. After 1 hour, 10 μl of Protein-G agarose was added and the incubation continued for ~2 hours (or overnight for IPs from rat brain) at 4°C. The resin was rinsed three times in 1 ml of solubilization buffer and bound proteins were analyzed by SDS-PAGE and western blotting with mouse antibodies to CaMKIIα (Affinity Bioreagents, Golden, Colorado) or LTCC subunits (α1, β1, and β4: NeuroMab). A rabbit polyclonal β2 antibody (a gift from Dr. Marlene Hosey) was used for the β2 blot in Figure 1A.
All samples were first separated in Tris-Glycine SDS-PAGE gels and then transferred to nitrocellulose membranes using standard methods. Membranes were blocked in 5% (w/v) milk in TTBS (50 mM Tris-HCl, pH 7.5; 150 mM NaCl, 0.1% (v/v) Triton-X-100), and then incubated with primary antibodies for 2 hours or overnight at 4°C. After washing 3 times for ~5 minutes each, membranes were incubated for 30 to 60 minutes at room temperature with horseradish peroxidase conjugated goat (Pierce, catalog number 31400), mouse (Promega, #W401B) or rabbit (Promega, #W4021) secondary antibody. Membranes were washed at least four times each for ~10 minutes and developed using enhanced chemiluminescence (Perkin-Elmer, Shelton, CT).
Glutathione S-transferase (GST) and His-tagged fusion proteins were expressed in BL21-DE3 Gold Escherichia coli bacteria and purified using glutathione-agarose (Sigma, St Louis, MI) or His-Select Nickel Affinity Gel (Sigma), according to the manufacturers’ protocol. Protein concentrations were determined by Bradford Assay (BioRad, Hercules, CA) or BCA assay (Pierce, Rockford, IL) using bovine serum albumin as a standard and validated by resolving proteins in SDS-PAGE gels and Coomassie-Blue staining.
Recombinant mouse CaMKIIα purified from baculovirus-infected Sf9 insect cells was autophosphorylated at Thr286 using ATP or [γ-32P]ATP, essentially as described previously (McNeill & Colbran 1995). Briefly, CaMKII (1 μM subunit) was incubated on ice with 50 mM Tris, pH 7.5, 10 mM magnesium acetate, 1.5 mM CaCl2, 20 μM calmodulin, 40 μM ATP, 1 mM DTT for 1.5 minutes and the reaction was stopped using 15 mM EDTA. For 32P-labeling, CaMKII (5 μM subunit) was incubated on ice with 50 mM Tris, pH 7.5, 2 mM magnesium acetate, 1.5 mM CaCl2, 20 μM calmodulin, 20 μM [γ-32P]ATP (40,000 cmp/pmol) and 2 mM DTT. After 1 minute, 5 μl of 100 mM ATP was added and the reaction was continued for an additional minute before it was stopped with 10 mM EDTA. The radioactive CaMKII was desalted on a Sephadex G-50 column and used in the plate-binding assay.
Thr286-autophosphorylated CaMKIIα was incubated with GST-I/II linker (or GST-AID) with or without His-tagged β2a for one hour at 4°C in GST pull-down buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Triton X-100; 0.4 ml total volume). Glutathione-agarose (10 μl packed resin) was added, and the incubation was continued for one hour. Resin was collected by gentle centrifugation (1 min at 1000 rpm) and washed three times in binding buffer. Bound proteins were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose membranes and detected by Ponceau-S staining. Membranes were scanned and then the levels of all proteins were quantified by densitometry using ImageJ (NIH).
GST or GST-I/II proteins (100 pmol) were diluted into 0.2 ml of plate-assay 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) and immobilized in glutathione-coated wells for 18–24 hours at 4°C. [32P-T286]CaMKII (100 pmol) was premixed with buffer (no His-β2a) or 36–100 pmol of His-β proteins. The mixture was added to the rinsed glutathione-coated wells. After a 2-hour incubation at 4°C, unbound kinase and His-β subunits were rinsed and the bound [32P-T286]CaMKII was quantified by scintillation counting. In some cases, contents of the wells were eluted into SDS-PAGE loading buffer and analyzed by electrophoresis, western blotting and autoradiography.
HEK293 cells were cotransfected with cDNA encoding HA-Cav1.2α1, FLAG-β2a (WT or L493A) and EGFP-CaMKIIα using Fugene 6 (Roche, Indianapolis, IN). After a 5-minute treatment in 10 μM A23187, HA-Cav1.2α1 was immunoprecipitated from Triton-soluble cell extracts using 10 μg of anti-HA antibodies (Vanderbilt Monoclonal Antibody Core). Immune complexes were western blotted using HA, FLAG or EGFP antibodies.
In order to assess total phosphorylation of LTCC subunits, HEK293 cells (≈1 × 106 cells per well in a 6 well plate) were transfected with HA-Cav1.2α1, EGFP-CaMKIIα and FLAG-β2a (WT or L493A) cDNA for 48 hours. Intracellular phosphate pools were depleted by rinsing once in phosphate-free DMEM (GIBCO) and then incubating in the same medium for 1 hour prior to addition of 0.42 mCi per well of 32P-orthophosphate (PerkinElmer, Boston, MA) diluted in phosphate-free DMEM. After incubating for 4 hours at 37°C, radioactive medium was removed and the cells were rinsed twice in ice-cold PBS. The HA-Cav1.2α1 subunit was immunoprecipitated from cell lysates (see above) and analyzed by SDS-PAGE. TOTAL 32P-phosphorylation of LTCC subunits was detected by autoradiography, and total protein levels of each subunit were determined by western blotting with antibodies to the HA and FLAG epitopes. ImageJ was used to quantify β2a 32P-phosphorylation and protein levels, which were each corrected for background determined by similarly quantifying signals in control immune complexes isolated from cells lacking β2a. Total 32P-phosphorylation of β2a was then normalized to total β2a protein levels in the LTCC immune complex.
For specific assessment of Thr498 phosphorylation in FLAG-β2a, HA-Cav1.2α1 immune complexes were western blotted with anti-phospho-Thr286 CaMKII (Promega, Madison, WI) which we previously showed also recognizes the phosphorylated Thr498 residue in β2a (Grueter et al. 2008). Immunoblots were quantitatively analyzed using ImageJ. Signal intensities for phospho-Thr498 and FLAG-β2a were each corrected for backgrounds (estimated from control complexes isolated from cells lacking β2a), and phospho-Thr498 signals were then normalized to β2a signals in each LTCC immune complex.
Statistical analyses (t-tests or analyses of variance (ANOVA) and post-hoc tests) were performed using GraphPad Prism. P values less than 0.05 were considered significant.
LTCC α1 subunits and a heterogeneous mixture of β subunits are expressed in the brain. In order to determine whether CaMKII associates with these channels, we probed CaMKII immune complexes isolated from rat forebrain extracts for LTCC subunits (Fig. 1A). CaMKII was substantially depleted from the supernatant, with a corresponding enrichment in the immune complexes. Control IgGs failed to deplete CaMKII from the supernatant: the weak detection of CaMKII in control IgG complexes, despite extensive washing, likely reflects the abundance of CaMKIIα in forebrain. Thus, CaMKII was substantially and specifically enriched in CaMKII complexes. Initial experiments indicated that rat CaMKII complexes also contained VGCC β1 and β2 subunit variants using limited available quantities of a rabbit antibody to the β2 subunit and a mouse β1 subunit antibody (Fig. 1A). Immunoblotting the rat CaMKII complexes with a monoclonal antibody to the CaV1.2α1 subunit also specifically and reproducibly detected an ≈200 kDa protein (not shown). This ≈200 kDa protein and the β1 subunit were also specifically enriched in mouse forebrain CaMKII immune complexes (not shown). However, although the CaV1.2α1 antibody specifically detected CaV1.2α1 subunits expressed in HEK293 cells (not shown), we had no independent verification that the neuronal CaMKII-associated ≈200 kDa protein was in fact the CaV1.2α1 subunit. Nevertheless, these data together suggest that at least the LTCC β1 and β2 subunits are specifically associated with neuronal CaMKII.
In order to further explore the existence of LTCC-CaMKII complexes in brain, we first tested the specificity of commercially available monoclonal antibodies to the β1 and β4 subunits. Each antibody specifically detected the respective bacterially-expressed isoform as shown by western blotting GST fusion proteins containing the full length amino acid sequences of all four β subunit isoforms, albeit with different sensitivity (Fig S1A). The antibodies were then used to probe extracts of HEK cells transfected to express recombinant β1 or β4 and subcellular fractions of mouse brain (Fig. S1B). Both antibodies recognized the appropriate isoform in HEK cell lysates and also recognized a closely migrating series of bands that were enriched in brain particulate/membrane fractions. These proteins had apparent molecular weights appropriate for a mixture of splice variants and/or post-translationally modified forms of the respective β subunit.
The mouse β1 and β4 subunit antibodies were then used to immunoprecipitate calcium channel complexes from a Triton-soluble fraction of mouse brain. Western blots demonstrated that the appropriate β subunit isoform was specifically enriched in each immune complex (Fig. 1B/C). The Cav1.2α1 subunit was substantially enriched in both immune complexes, though somewhat more so in β4 complexes than in β1 complexes. In contrast, CaMKIIα was specifically detected in β1 immune complexes (Fig. 1B) but not in β4 immune complexes (Fig. 1C). In combination, these data demonstrate that CaMKIIα associates with a subset of LTCCs containing β1 and β2 subunits, but not with LTCC complexes containing the β4 isoform.
We then investigated the role of direct interactions between the Cav1.2α1 and β subunits and/or CaMKII in assembly of CaMKII-LTCC complexes. We previously showed that Thr287-autophosphorylated CaMKIIδ directly interacts with the β1b and β2a subunits with similar affinities (Kd ≈100 nM CaMKII subunit), but not with β3 or β4 subunits (Grueter et al. 2008). The GK domains in β subunits bind to I/II linkers of VGCC α1 subunits with high but variable affinities (5–60 nM) (Leroy et al. 2005, Van Petegem et al. 2008). In addition, autophosphorylated CaMKII was previously shown to directly bind with similar efficacy to multiple intracellular domains of the CaV1.2α1 subunit in vitro, including the I/II linker (Hudmon et al. 2005, Lee et al. 2006), although affinities of these interactions were not determined. As a first step toward dissecting roles of these interactions in assembly of CaMKII-LTCC complexes, we investigated binding of Thr286-autophosphorylated CaMKIIα to the GST-tagged I-II linker domain of the CaV1.2α1 subunit (GST-I/II) in the absence or presence of purified hexahistidine (His)-tagged β1b, β2a and β3 subunits (we were unable to purify His-tagged β4). There was no detectable direct interaction with GST-I/II, but CaMKII associated with GST-I/II in the presence of His-β1b or His-β2a, but not in the presence of His-β3 (Fig. 2A). Interestingly, while β1b and β2a directly interact with autophosphorylated CaMKII with a similar apparent Kd (Grueter et al. 2008), His-β1b immobilized significantly less CaMKII on GST-I/II than did His-β2a. Since CaMKII immobilization on the I/II linker also depends on β subunit interactions with the I/II linker, proteins were solubilized from the glutathione wells and analyzed by immunoblotting. GST-I/II appeared to bind similar amounts of β2a and β3, but the interaction with β1b appeared to be weaker (Fig. 2B). Across the three experiments, GST-I/II binding to His-β1b and His-β3 was 45±10% and 102±37% of the binding to His-β2a (mean±sem: p=0.03 and p=0.97, respectively). Moreover, GST-I/II binding to His-β3 in GST-cosedimentation analyses was 85±15% of the binding to His-β2a (n=3; p=0.43), and CaMKIIα binding did not appear to affect the interaction of His-β2a with the I/II linker (Fig. S2 and see below). Thus, CaMKII can be tethered to the I/II linker of CaV1.2α1 subunits by specific β subunit isoforms in vitro.
Since β2a most strongly tethers CaMKII to the CaV1.2α1 I/II linker, and is essential for CaMKII-dependent regulation of native LTCCs, we used β2a to explore mechanisms underlying β subunit-dependent tethering of CaMKIIα to CaV1.2α1 subunits. Although we were unable to detect direct interactions of activated CaMKII with the 120 amino acid long I/II linker domain in GST-I/II, it seemed possible that binding of β2a to CaMKIIα might stabilize direct interactions of CaMKIIα with the I/II linker. Therefore, we performed experiments using a GST fusion protein containing the minimal 18 amino acid alpha-interacting domain (GST-AID) that is sufficient for binding to β subunits (Van Petegem et al. 2008). As expected, we failed to detect direct interaction of activated CaMKIIα with GST-AID, but His-β2a was able to tether activated CaMKII to GST-AID (Fig. S3). Mutation of Trp470 to Ala in the I/II linker was shown to severely compromise β subunit binding (Butcher et al. 2006, Leroy et al. 2005, Van Petegem et al. 2008), and the W470A mutation essentially abrogated the interactions of both His-β2a and activated CaMKII with GST-I/II (Fig. 3A). High affinity direct binding of activated CaMKIIα to β2a is disrupted by Leu493 to Ala mutation (Grueter et al. 2008). The L493A mutation significantly reduced β2a-dependent co-sedimentation of activated CaMKIIα with GST-I/II by about 60% (Fig. 3A/C) and with GST-AID by about 70% (Fig. S3). Notably, disruption of CaMKII binding with the L493A mutation had no detectable effect on the amount of β2a that co-sedimented with GST-I/II (Fig. 3B) or with GST-AID (Fig. S3), further confirming that interactions with CaMKII have no effect on the binding of His-β2a to the I/II linker.
We then used a more quantitative glutathione-coated multi-well plate-binding assay (see Methods) to further explore mechanisms underlying His-β2a-dependent tethering of activated 32P-labeled CaMKII to GST-I/II. Consistent with co-sedimentation assays, activated CaMKII efficiently bound to GST-I/II in the presence of His-β2a, and omission His-β2a reduced binding by over 100-fold (Fig. 3D). The His-β2a-dependent binding of CaMKII to GST-I/II also was significantly compromised by L493A mutation of His-β2a (≈50% reduction in binding). Moreover, a His-β2a(410–505) protein which lacks the SH3 and GK domains and cannot bind to the AID (data not shown), but contains the CaMKII-binding domain (Grueter et al. 2008), failed to support the tethering of activated CaMKII to GST-I/II (Fig. 3D). Taken together, these findings show that the β2a subunit can serve as an adaptor protein to tether CaMKII to the I/II linker domain of the α1 subunit.
Previous co-immunoprecipitation studies showed that CaMKII associates with LTCC complexes in the presence of β2 subunits (Hudmon et al. 2005, Lee et al. 2006), but the role of the β subunit in formation of this complex was not explored. In order to investigate whether β2a serves as an adaptor protein, tethering CaMKII to LTCC complexes, we expressed EGFP-CaMKII in HEK293 cells in the absence or presence of HA-Cav1.2α1 subunits and WT or L493A mutated FLAG-β2a. LTCC complexes were isolated by immunoprecipitation using anti-HA antibodies and western blotted for all three proteins. HA-Cav1.2α1 subunits were not detected in cell extracts due to the relatively low expression levels, but were readily detected in HA-immune complexes. As expected due to known roles of β subunits in modulating surface expression and trafficking of VGCC α1 subunits, co-expression of β2a enhanced the levels of α1 subunit expression. WT and L493A mutated FLAG-β2a proteins were expressed at similar levels and induced a similar increase in α1 subunit expression. (Note: HA-immune complexes from HEK293 cells that were not transfected to express FLAG-β2a contained a weak, non-specific FLAG-immunoreactive protein of about 70 kDa). Total expression levels of EGFP-CaMKII were similar in all transfections. However, EGFP-CaMKII was only reliably detected in anti-HA immune complexes isolated from cells expressing WT FLAG-β2a. Much lower levels of EGFP-CaMKII were detected in anti-HA immune complexes isolated from cells expressing L493A-mutated FLAG-β2a or from cells that did not express FLAG-β2a (Fig. 4). These data demonstrate that high affinity binding to the β2a subunit is necessary for efficient assembly of CaMKII with the LTCC complex.
LTCCs are phosphorylated at multiple sites in both the α1 and β subunits (reviewed by (Dai et al. 2009)) and CaMKII-dependent regulation of CaV1.2 LTCCs has been linked to phosphorylation of the α1 subunit at Ser1512 and Ser1570 (Lee et al. 2006) and of the β2a subunit at Thr498 (Grueter et al. 2006). In order to test the hypothesis that CaMKII binding to β2a modulates LTCC subunit phosphorylation, HA-CaV1.2α1, EGFP-CaMKII and/or FLAG-β2a (WT or L493A) were co-expressed in HEK293 cells. After labeling cells with 32P-orthophosphate (see Methods), LTCC complexes were immunoprecipitated from cell extracts for analysis by SDS-PAGE followed by autoradiography to detect 32P-phosphorylated proteins and by western blotting, allowing normalization of 32P-phosphorylation to amounts of LTCC subunits isolated. The 32P-labeling of CaV1.2α1 in the absence of co-expressed β2a and EGFP-CaMKII indicated that endogenous kinases significantly phosphorylate CaV1.2α1 subunits under these basal incubation conditions (Fig. 5A). The co-expression of EGFP-CaMKII had no consistent effect on 32P-phosphorylation of CaV1.2α1 in the absence or presence of WT β2a. The 32P-phosphorylation of WT β2a within LTCC complexes also was not significantly altered by expression of EGFP-CaMKII (Fig. 5B). Moreover, L493A mutation of β2a had no significant impact on 32P-phosphorylation of either the CaV1.2α1 or β2a subunit (Fig. 5A/B). However, we confirmed that EGFP-CaMKII associated with LTCC complexes containing WT β2a subunits (data not shown). Thus, total 32P-phosphorylation of CaV1.2α1 and β2a subunits in LTCC complexes did not appear to be affected by CaMKII association via β2a under basal conditions.
Total 32P-phosphorylation of proteins in 32P-orthophosphate labeling experiments can report the phosphorylation of multiple sites in a target protein. However, β2a can be phosphorylated at multiple serine and threonine residues by several kinases, including PKA (at Ser 458,478,479) and PKG (at Ser 496) (Gerhardstein et al. 1999, Yang et al. 2007). Therefore, we reasoned that changes in phosphorylation at a limited number of sites due to L493A mutation may be obscured by 32P-phosphorylation at other sites in β2a. We previously showed that the critical regulatory phosphorylation of β2a at Thr498 can be detected using a phospho-Thr286 CaMKIIα antibody (due to the similarity of surrounding amino acid sequences) and that L493A mutation does not affect recognition of phospho-Thr498 in β2a (Grueter et al. 2008). Therefore, we used the anti-phospho-Thr286 CaMKIIα antibody to specifically examine Thr498 phosphorylation in β2a within LTCC complexes immunoprecipitated from transfected HEK293 cells. We detected a robust signal with a phospho-Thr286 antibody within LTCC complexes containing WT β2a subunits and mutation of Thr498 in β2a to Ala eliminated the signal (Fig. S4). Moreover, a similar robust phosphorylation of WT β2a was detected when EGFP-CaMKIIα was replaced by untagged CaMKIIα (data not shown). Thus, we specifically detected Thr498 phosphorylation of β2a in the LTCC complex under these conditions, and not Thr286 phosphorylation of co-precipitating EGFP-CaMKIIα. Significantly, mutation of Leu493 to Ala to interfere with CaMKII targeting to LTCC complexes almost abolished the phosphorylation of β2a at Thr498 (Fig. 5C/D). Collectively, these data show that CaMKII binding to β2a is required for efficient phosphorylation of a key regulatory site (Thr498) in β2a without affecting overall phosphorylation of multiple sites in the CaV1.2α1 or β2a subunits.
LTCC facilitation by CaMKII can be important for physiological augmentation of cellular Ca2+ signals in response to hormones or growth factors (Gao et al. 2006, O-Uchi et al. 2005). However, excessive CaV1.2 LTCC activity, Ca2+ entry and CaMKII activation are critical events in the pathogenesis of cardiac arrhythmias and other forms of heart disease (reviewed in (Couchonnal & Anderson 2008, Bers 2008)). Similarly, hyper-activation of CaV1.2 LTCCs by CaMKII is implicated in Timothy Syndrome, a multi-organ human genetic disorder whose symptoms include mental retardation and cardiac disease (Splawski et al. 2004, Splawski et al. 2005, Erxleben et al. 2006, Thiel et al. 2008). Furthermore, the loss of dendritic spines from striatal medium spiny neurons following dopamine depletion in animal models of parkinsonism results from excessive CaV1.3 LTCC activation (Day et al. 2006), in parallel with CaMKII over-activation (Brown et al. 2005). Thus, understanding mechanistic interactions between CaMKII and LTCCs promises to provide insights into physiological and pathological processes in multiple tissues.
Subcellular compartmentalization is emerging as a key feature of Ca2+ signaling (Parekh 2008). Colocalization and functional studies suggest that a proportion of CaMKII and LTCCs exist in the same subcellular compartments in cardiomyocytes and neurons and that the association of CaMKII with the LTCC complex appears to be essential for facilitation (see Introduction). Multiple forms of β subunit- and CaMKII-dependent LTCC facilitation have been observed in heterologous cells (Lee et al. 2006, Xiao et al. 1994) or reconstituted in isolated membrane patches (Dzhura et al. 2002, Grueter et al. 2006). This implies that core elements for CaMKII regulation of CaV1.2 LTCCs in cardiomyocytes are intrinsic to the channel and must exist in neurons. CaMKII was previously shown to co-immunoprecipitate with cardiac CaV1.2 LTCC complexes. The present findings are the first to show that CaMKII associates with CaV1.2 LTCCs in forebrain.
Association of CaMKII with LTCCs in forebrain appears to be controlled by the β subunit identity. Although CaV1.2α1 subunits were similarly enriched in both β1 and β4 subunit immune complexes, CaMKII was only detected in the β1 subunit complex (Fig. 1). Our biochemical studies suggest that CaMKII will also associate with neuronal LTCCs containing the β2 subunit but not those containing the β3 subunit (Figs. 1A, ,22,,3).3). This selective association with CaV1.2 LTCCs containing the β1 and β2 subunit variants appears to be due to differential scaffolding/adaptor protein functions of the β subunits. The GK domain of the β subunits interacts with the AID of the α1 subunit, whereas a C-terminal domain conserved in β1b or β2a, but not in β3 or β4, binds to CaMKII. The number and stability of CaMKII-LTCC complexes may be influenced by additional interactions, including membrane association of β1 or the β2 subunits via N-terminal lipid modifications (reviewed in (Dolphin 2003, Hidalgo & Neely 2007)), additional interactions of β subunits with N- and C-terminal domains of the Cav1.2α1 subunit (Zhang et al. 2005, Kanevsky & Dascal 2006, Cheng et al. 2007, Lao et al. 2008) and CaMKII interaction with other Cav1.2α1 binding partners such as α-actinin (Robison et al. 2005, Malik 2009). However, the GK-AID interaction appears to be fundamental for creation of a stable scaffold that specifically targets CaMKII to LTCCs containing the β1b or β2a subunit.
Differential functions of β subunit isoforms in controlling the activity and membrane expression of VGCCs are widely accepted (Birnbaumer et al. 1998, Colecraft et al. 2002, Dolphin 2003). These differences are emphasized by the embryonic lethality of global knockout of β1 and β2 subunit genes in mice (Gregg et al. 1996, Weissgerber et al. 2006), whereas animals lacking the β3 and β4 subunits are viable (Murakami et al. 2002, Berggren et al. 2004). Relative expression levels of the β subunit variants are thought to play a major role in determining the specificity of β subunit association with LTCC α1 subunits (Scott et al. 1996, Pichler et al. 1997, McEnery et al. 1998, Vance et al. 1998, Dolphin 2003). In contrast to the predominant expression of LTCC β2 isoforms in cardiomyocytes, a mixture of β isoforms are differentially expressed in a complex, developmentally-regulated and cell-specific manner in the brain. The CaMKII-binding β1 and β2 subunits are expressed mainly in hippocampus, striatum, cerebellum and photoreceptors (Mermelstein et al. 1999, Ludwig et al. 1997, Vance et al. 1998, McEnery et al. 1998) and predominate during early postnatal development in rodents (Vance et al. 1998, Dolphin 2003). However, by adulthood, the β3 and β4 subunit variants together comprise about 70% of total β subunits in forebrain (Ludwig et al. 1997, Vance et al. 1998, McEnery et al. 1998). Thus, CaMKII association with neuronal CaV1.2 LTCCs may be more prevalent during postnatal development and only in certain brain regions or cell types.
It is increasingly apparent that subcellular targeting of CaMKII via protein-protein interactions promotes the phosphorylation of specific substrate targets in intact cells (Tsui et al. 2005, Tsui & Malenka 2006). We previously reported that CaMKII-dependent facilitation of CaV1.2 LTCCs requires the phosphorylation of β2a subunits at Thr498, and showed that CaMKII binds with high affinity (Kd ≈ 100 nM CaMKII subunit) to the β1b and β2a subunits, but not the β3 or β4 subunits. However, the relationship(s) between protein-protein interactions, phosphorylation and facilitation was not investigated. The primarily β-subunit-dependent association of CaMKII with LTCCs may facilitate CaMKII interaction with additional domains on the α1 subunit (Hudmon et al. 2005), and may enhance CaMKII phosphorylation of functionally relevant sites, such as Ser1512 or Ser1570 in the α1 subunit (Lee et al. 2006) or Thr498 in the β2a subunit (Grueter et al. 2006). It was initially surprising that total 32P-phosphorylation of LTCC subunits was not affected by co-expression of EGFP-CaMKII or by L493A mutation of β2a (Fig. 5). However, it is important to note that these experiments were performed under basal cell incubation conditions in which CaMKII is only partially active. The partially active CaMKII may be more selective for the preferred Thr498 site in β2a in situ relative to other sites that can be phosphorylated in vitro by fully active kinase (Grueter et al. 2006). The substantial background of LTCC subunit phosphorylation at other sites by endogenous kinases (Fig. 5) also may limit our ability to detect CaMKII-dependent changes in 32P-labeling. However, we demonstrated by immunoblotting with a phospho-specific antibody that CaMKII binding to β2a facilitates Thr498 phosphorylation within LTCC complexes isolated by immunoprecipitation. Since phosphorylation of β2a at Thr498 is required for CaMKII-dependent LTCC facilitation, the present findings provide a biochemical explanation for these actions of CaMKII.
The β subunit isoforms associate with most VGCC α1 subunit variants. Interestingly, CaMKII facilitates CaV1.3 LTCCs by phosphorylating the α1 subunit at Ser1486 (Gao et al. 2006) and antagonizes the inactivation of Cav2.1 P/Q type VGCCs by binding within the α1 C-terminal tail (Jiang et al. 2008). It will be important to investigate whether β subunits also play a role in these CaMKII-dependent effects. Unlike Cav1.3 and 2.1, CaV3.2 channels are not generally thought to associate with β subunits. Yet CaMKII enhances the activity of Cav3.2 T-type VGCCs, apparently via direct interaction of CaMKII with the II/III linker domain (Welsby et al. 2003). Thus, binding of CaMKII to the VGCC complex appears to be a common feature for CaMKII-dependent feedback regulation of these channels. The β1 or β2 subunits may have a preeminent function in some cases, but in other cases direct interactions with the α1 subunit may play an important role. Moreover, it will be interesting to investigate whether association of CaMKII with LTCCs via selected β subunit variants plays a role in CaMKII-dependent cross talk between LTCCs and R-type channels in dendritic spines (Yasuda et al. 2003, Lee et al. 2009) and/or in CaMKII activation to mediate downstream signaling to the nucleus (Wheeler et al. 2008).
We thank Martha Bass for purifying CaMKII and for additional excellent technical support. This work was supported in part by National Institutes of Health Grants RO1-MH63232 (to RJC) and American Heart Association Predoctoral Fellowship 0715137B (to SAA).