|Home | About | Journals | Submit | Contact Us | Français|
Cav1 (L-type) channels and calmodulin-dependent protein kinase II (CaMKII) are key regulators of Ca2+ signaling in neurons. CaMKII directly potentiates the activity of Cav1.2 and Cav1.3 channels, but the underlying molecular mechanisms are incompletely understood. Here, we report that the CaMKII-associated protein, densin, is required for Ca2+-dependent facilitation of Cav1.3 channels. While neither CaMKII nor densin independently affect Cav1.3 properties in transfected HEK293T cells, the two together augment Cav1.3 Ca2+ currents during repetitive, but not sustained, depolarizing stimuli. Facilitation requires Ca2+, CaMKII activation and its association with densin, as well as densin binding to the Cav1.3 α1 subunit C-terminal domain. Cav1.3 channels and densin are targeted to dendritic spines in neurons and form a complex with CaMKII in the brain. Our results demonstrate a novel mechanism for Ca2+-dependent facilitation that may intensify postsynaptic Ca2+ signals during high-frequency stimulation.
CaMKII is a serine-threonine protein kinase which is activated by postsynaptic elevations in Ca2+ and plays a central role in synaptic plasticity (Hudmon and Schulman, 2002; Lisman et al., 2002; Colbran and Brown, 2004; Griffith, 2004). Cav1 channels mediate L-type Ca2+ currents which can regulate CaMKII activation in dendritic spines (Lee et al., 2009), propagation in dendrites (Rose et al., 2009) and coupling to gene transcription (Wheeler et al., 2008) and synaptic plasticity (Yasuda et al., 2003; Lee et al., 2009). Functional interactions of Cav1 and CaMKII may involve tethering of CaMKII to the Cav1 channel complex. CaMKII binds to and phosphorylates the main Cav1.2 α1 (Hudmon et al., 2005) and auxiliary β subunits (Grueter et al., 2006; Grueter et al., 2008). These interactions augment cardiac Cav1.2 currents in a feedback process known as Ca2+-dependent facilitation (Dzhura et al., 2000; Wu et al., 2001; Hudmon et al., 2005; Grueter et al., 2006). CaMKII also causes voltage-dependent facilitation of Cav1.2 currents in response to strong depolarizations (Lee et al., 2006).
Due to their distinct biophysical properties (Koschak et al., 2001; Scholze et al., 2001; Xu and Lipscombe, 2001; Helton et al., 2005), Cav1.3 and Cav1.2 may play different roles in neurons (Calin-Jageman and Lee, 2008). Cav1.3 channels regulate spontaneous firing of substantia nigra dopaminergic neurons (Chan et al., 2007; Puopolo et al., 2007), upstate potentials in striatal medium spiny neurons (Olson et al., 2005), and processes underlying fear conditioning (McKinney and Murphy, 2006) and depression-like behavior (Sinnegger-Brauns et al., 2004). Therefore, Cav1.3 modulation by CaMKII and other factors may have evolved to meet the unique signaling demands of this channel in neurons. Although insulin-like growth factor 1 can potentiate Cav1.3 currents by a mechanism that requires Ca2+ release from intracellular stores and CaMKII activity (Gao et al., 2006b), the role of CaMKII in direct feedback regulation of Cav1.3 channels is unknown.
Here, we describe an unexpected mechanism for Cav1.3 modulation by CaMKII involving the PDZ-[postsynaptic density-95 (PSD-95)/Discs large/zona occludens-1 (ZO-1)]-domain containing protein, densin. Densin is a member of the leucine-rich repeat and PDZ-domain containing proteins (Bilder et al., 2000), which is localized in the postsynaptic density (Apperson et al., 1996) and binds to and is phosphorylated by CaMKII (Strack et al., 2000; Walikonis et al., 2001). Densin may scaffold other postsynaptic proteins including Shank (Quitsch et al., 2005), δ-catenin (Izawa et al., 2002), and MAGUIN (Ohtakara et al., 2002) and promotes branching of dendrites in cultured neurons (Quitsch et al., 2005). We showed previously that the related protein, erbin, enhances voltage-dependent facilitation of Cav1.3 (Calin-Jageman et al., 2007). Like erbin, densin binds to the C-terminus of the Cav1.3 α1 subunit but alone does not affect Cav1.3 properties. When coexpressed with CaMKII, densin facilitates Cav1.3 Ca2+ currents during high-frequency stimulation. Densin and Cav1.3 are targeted to dendritic spines and together associate with CaMKII in the hippocampus. Our results show that densin functionally recruits CaMKII to Cav1.3 channels, which causes frequency-dependent facilitation of Cav1.3 Ca2+ signals that may regulate neuronal excitability.
Cav1.3 channels consisted of rat α11.3 (containing exon 42, GenBank #AF3700010, in pCDNA6 from D. Lipscombe, Brown U.), β1b (GenBank #NM017346), and α2δ (GenBank #M21948). The following cDNAs were described previously: FLAG- and GST-tagged α11.3 constructs (Calin-Jageman et al., 2007); CaMKIIα, CaMKIIαT286A (Brickey et al., 1990; McNeill and Colbran, 1995); GFP-densin and GFP-densin Δ483-1377 (Jiao et al., 2008); his-densin PDZ (Fam et al., 2005); and pβA-eGFP (Obermair et al., 2004). For GFP-densin ΔPDZ, the PDZ domain-encoding region (aa 1452-1542) was deleted by PCR amplification and ligation into BglII and SacII sites of pEGFP (Clontech, BD BioSciences, Mountainview, CA).
For external HA epitope-tagged α11.3, the sequence for the HA tag was cloned into the extracellular loop connecting IIS5-IIS6 according to a similar strategy for α11.2 described previously (Altier et al., 2002). The insertion of the HA tag had no effects on channel properties as assessed by electrophysiological recordings of transfected HEK293T cells (data not shown). To facilitate neuronal expression HA-α11.3 was subcloned into the pβA-PL expression vector (Obermair et al., 2009) in a two-step procedure. First, a HindIII—SalI fragment (3303 bp) containing the 5′ coding sequence of α11.3 was cloned into the corresponding restriction sites of pβA-PL. Second, the BsiWI—SacII fragment (6019 bp) of HA-α11.3 was ligated together with a 25 bp SacII—SpeI linker into the BsiWI and XbaI sites of the intermediate construct, eliminating SpeI and XbaI recognition sequences, and yielding pβA-HA-α11.3. The construct was verified by sequencing prior to use (MWG-Biotech AG, Ebersberg, Germany).
Goat α11.3 antibodies (Calin-Jageman et al., 2007) and goat densin antibodies (Ab650) (Jiao et al., 2008) were characterized previously. Rabbit α11.3 antibodies (Ab144) were raised against a synthetic peptide corresponding to α11.3 N-terminal sequence (MQHQRQQQEDHANEANYARGTRKC; Covance Research Products, Denver, PA). Characterization of Ab144 specificity is described in Supplementary Fig.1. Briefly, by immunofluorescence and western blot, these antibodies labeled HEK293T cells transfected with Cav1.3 but not untransfected cells (Fig.S1A,B). In addition, Ab144 recognized a protein consistent in size with α11.3 in hippocampal lysates of wild-type but not Cav1.3 knockout mice (provided by Jörg Striessnig, U. Innsbruck; Fig.S1C). Other antibodies used were: mouse monoclonal antibodies against CaMKIIα (Affinity Bioreagents, Golden, CO), FLAG (Sigma-Aldrich, St. Louis, MO), and GFP (Santa Cruz Biotechnology, Santa Cruz, CA); and rabbit polyclonal antibodies against densin (Santa Cruz Biotechnology, Santa Cruz, CA) and hexahistidine (anti-his) antibodies (Santa Cruz Biotechnology).
For immunofluorescence of cultured neurons, the following antibodies were used: rat anti-HA (monoclonal, clone 3F10, Roche Diagnostics GmbH, Vienna, Austria, 1:100); mouse anti-PSD-95 (monoclonal, clone 6G6–1C9, Affinity Bioreagents, Inc., Golden, CO, USA, 1:1,000); rabbit polyclonal anti-GFP (1:20,000; Molecular Probes, Eugene, OR, USA); goat anti-rabbit Alexa 488 (1:2,000); goat anti-mouse Alexa 594 (1:4,000); and goat anti-rat Alexa 594 (Invitrogen, Gaithersburg, MD, 1:4,000).
Human embryonic kidney cells (HEK293) or HEK293 cells transformed with SV40 T-antigen (HEK293T) were maintained in DMEM with 10% fetal bovine serum (Invitrogen) at 37°C in a humidified atmosphere with 5% CO2. Cells were grown to ~80% confluence and transfected using Gene Porter reagent (Gene Therapy Systems, San Diego, CA) or Fugene (Roche, Brandford, CT). For pull-down assays, cells were transfected with GFP-densin (6 μg). For co-immunoprecipitation of GFP-densin and Cav1.3, cells were transfected with cDNAs encoding Cav1.3 (FLAG- α11.3 (6 μg), β1b (2 μg), and α2δ (2 μg)) with or without GFP-densin (4 μg). For co-immunoprecipitation of CaMKII and GFP-densin or Δ483-1377, cells were transfected with: GFP-densin (7 μg) or GFP- Δ483-1377 (2 μg) and CaMKIIα (1 μg). For electrophysiology, HEK293T cells were transfected with α11.3 (1.5 μg), β1b (0.5 μg), and α2δ (0.5μg) with or without GFP-tagged densin (0.5 μg) and/or CaMKII (0.5 μg).
At least 48 h after transfection, whole cell patch-clamp recordings of transfected cells were acquired with a HEKA Elektronik (Lambrecht/Pfalz, Germany) EPC-8 or EPC-9 patch-clamp amplifier. Data acquisition and leak subtraction using a P/4 protocol were performed with Pulse software (HEKA Elektronik). Extracellular recording solutions contained (in mM): 150 Tris, 1 MgCl2, and 10 CaCl2 or 10 BaCl2. Intracellular solutions contained (in mM): 140 N-methyl-D-glucamine, 10 HEPES, 2 MgCl2, 2 Mg-ATP, 5 EGTA or 10 BAPTA. The pH of the recording solutions was 7.3, adjusted with methanesulfonic acid. Electrode resistances were 3-4 MΩ in the bath solution. Series resistance was compensated up to 80%. Igor Pro software (Wavemetrics, Lake Oswego, OR) was used for data analysis. Except for Fig.7B, data analysis was restricted to Cav1.3 currents with amplitudes greater than 250 pA. All averaged data are presented as the mean ± SEM. Statistical significance was determined with either Student’s t- test or ANOVA with post-hoc analyses, as indicated (SigmaPlot; Systat Software, San Jose, CA).
GST- and his-tagged fusion proteins were prepared as described (Robison et al., 2005; Calin-Jageman et al., 2007). For pull-down assays, GST-α11.3 CT was immobilized on glutathione agarose beads and incubated with lysate from GFP-densin-transfected HEK293T cells in binding buffer [Tris-buffered saline (TBS; 50 mM Tris-HCl, pH 7.5, 150 mM NaCl), 0.1% Triton X-100, and protease inhibitors (1 mg/ml each of PMSF, pepstatin, aprotinin, and leupeptin)]. Binding reactions proceeded at 4 °C for 90 min. Beads were washed three times with binding buffer (1 ml) at 4°C, and bound proteins were eluted, resolved by SDS-PAGE, and transferred to nitrocellulose. Western blotting was performed with appropriate antibodies followed by HRP-conjugated secondary antibodies and enhanced chemiluminescent detection reagents (GE Healthcare). For overlay assays, GST-α11.3 CT, GST-α11.3 CTL-A, or GST (1 μg) was run on a 4-20% SDS-polyacrylamide gel and transferred to nitrocellulose. The membrane was blocked in blocking buffer (2% milk, TBS, 0.1% Tween-20) for 30 min (4 °C) prior to incubation with His-tagged densin-PDZ domain (300 nM in blocking buffer) for 1 h at 4 °C. Bound protein was detected by western blotting with anti-his antibodies.
For coimmunoprecipitation from mouse hippocampus, a Triton-X100 soluble fraction (0.5 ml) was prepared as described previously (Abiria and Colbran, 2009) and incubated with 10 μg of either goat IgG or affinity-purified goat antibodies that recognize CaMKII, Densin, or α11.3. After 1 h, 10 μl of protein-G Sepharose (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) was added and the incubation continued for ~2 h 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 or rabbit antibodies to densin and α11.3 (Ab144).
For coimmunoprecipitation of GFP-densin and α11.3, transfected HEK293T cells were solubilized in lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate, 1 mM EDTA, and protease inhibitors), incubated at 4°C for 30 min and subjected to centrifugation at 100,000 × g (30 min) to remove insoluble material. The supernatant was incubated with 5 μg α11.3 antibodies and 50 μl of protein A-Sepharose (50% slurry) for 4 h, rotating at 4°C. After three washes with RIPA buffer (1 ml), proteins were eluted with SDS-containing sample buffer and subjected to SDS-PAGE. Coimmunoprecipitated proteins were detected by western blotting with specific antibodies as indicated.
For coimmunoprecipitation of CaMKII and densin or Δ483-1377, transfected HEK293 cells were lysed on ice with lysis buffer (2 mM Tris–HCl pH 7.5, 1% (v/v) TritonX-100, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 5 mg/L leupeptin, 20 mg/L soybean trypsin inhibitor). After sonication (2 × 5 s), lysates were incubated at 4°C for 30 min and then centrifuged for 15 min at 10 000 × g. NaCl was added into the supernatant with the final concentration as 150 mM and equal aliquots of the supernatants were incubated with 10 μg of densin Ab450 or goat IgG, or goat CaMKIIα antibody overnight at 4°C. After addition of GammaBind Plus-Sepharose (30 μl of 50% slurry; Amersham Biosciences, Piscataway, NJ), incubations were continued for 2 h at 4 °C. Beads were collected by centrifugation and washed at least 5 times with 1 ml of wash buffer containing 50 mM Tris–HCl pH 7.5, 1% (v/v) TritonX-100 and 150 mM NaCl. Immune complexes were solubilized in SDS–PAGE sample buffer prior to electrophoresis and western blotting. Interpretations of results from coimmunoprecipitation and binding assays were based on at least three independent experiments.
Low-density cultures of hippocampal neurons were prepared from 16.5-day-old embryonic BALB/c mice as described (Obermair et al., 2004). The following plasmids (1.5 μg total DNA) were transfected into neurons on day 6 using Lipofectamine 2000 transfection reagent (Invitrogen GmbH): GFP-densin (single-transfection); and pβA-eGFP and pβA-HA-α11.3 (co-transfection). Cells were immunostained and analyzed 11–14 days after transfection.
For double-immunolabeling (GFP-densin and PSD-95) neurons were fixed with methanol for 10 min at −20°C and rehydrated in PBS at room temperature. Fixed neurons were incubated in 5% normal goat serum in phosphate-buffered saline, 0.2% bovine serum albumin, and 0.2% Triton-X100 (PBS/BSA/Triton) for 30 min. Primary antibodies were applied in PBS/BSA/Triton at 4°C overnight and detected by fluorochrome-conjugated secondary antibodies. For staining of surface-expressed HA-α11.3, living neurons were incubated with the rat anti-HA antibody for 30 min at 37°C. Then the cultures were rinsed in Hank’s buffered saline, fixed in 4% paraformaldeyde/ 4% sucrose for 10 min, blocked with 5% normal goat serum in PBS/BSA/Triton, and incubated with the secondary antibody for 1 h (Obermair et al., 2009). Coverslips were then washed and mounted in p-phenylene-diamine-glycerol to retard photobleaching. Preparations were analyzed on an AxioImager microscope (Carl Zeiss, Inc) using a 63 ×, 1.4 NA objective. Images were recorded with a cooled CCD camera (SPOT; Diagnostic Instruments, Stirling Heights, MI, USA) and Metavue image processing software (Universal Imaging, Corp., West Chester, PA, USA). Composite images were arranged in Adobe Photoshop 9 (Adobe Systems Inc.) and linear adjustments were performed to correct black level and contrast.
HEK293T cells transfected with Cav1.3, densin, and CaMKII were loaded with Fura-2 (Invitrogen, Carlsbad, CA) via the patch pipette (100 μM), which also contained the intracellular recording solution described for electrophysiological recordings. Cells were placed in a flow-through chamber mounted on the stage of an inverted IX-71 microscope (Olympus, Japan). Fluorescence was alternately excited at 340 (12 nm bandpass) and 380 (12 nm bandpass) nm using the Polychrome IV monochromator (TILL Photonics, Germany) via a 40x oil-immersion objective (NA=1.35, Olympus, Japan). Emitted fluorescence was collected at 510 (80) nm using an IMAGO CCD camera (TILL Photonics, Germany). Pairs of 340/380 nm images were sampled at 10 Hz. Fluorescence was corrected for background, as determined in an area that did not contain a cell. Data were processed using TILLvisION 126.96.36.199 (TILL Photonics, Martinsried, Germany) and presented as a fluorescence ratio of F340/F380, where F340 and F380 are fluorescence intensities at the excitation wavelengths 340 and 380 nm, respectively. Averaged data are presented as the mean ± SEM and were statistically compared by t-test.
Because of the importance of CaMKII as both regulator and transducer of Cav1 Ca2+ signals (Dzhura et al., 2000; Wheeler et al., 2008), we tested if CaMKII directly influences Cav1.3 function. For this purpose, we compared channel properties in HEK293T cells transfected with Cav1.3 alone (α11.3, β1b, and α2δ) and those cotransfected with Cav1.3 and CaMKIIα. This isoform of CaMKII was chosen since it is one of the major isoforms of CaMKII in the brain (Colbran and Brown, 2004) and cannot be detected endogenously in HEK293T cells (Fig.1A). We found that CaMKII had no effect on voltage-dependent activation of Cav1.3 Ba2+ currents (IBa). Parameters describing current-voltage (I-V) curves were not different in cells with Cav1.3 and those with Cav1.3+CaMKII (p=0.32 for k, p=0.92 for V1/2, by t-test; Fig.1B). Expression of CaMKII also did not affect mean Cav1.3 current amplitudes (535±78 pA for Cav1.3 alone, n=18 vs. 844±227 pA for +CaMKII, n=16; p=0.18 by t-test). While CaMKII enhances voltage-dependent facilitation (VDF) of Cav1.2 (Lee et al., 2006), we did not find the same result for Cav1.3. VDF was measured as the ratio of the amplitude of IBa evoked before or after a conditioning prepulse. With this protocol, IBa underwent modest VDF that was not further affected by CaMKII (p=0.68, Fig.1C). We also did not observe any differences in the extent of IBa inactivation either during sustained or repetitive stimuli (not shown).
In contrast to our findings, previous studies of SH-SY5Y human neuroblastoma cells and cortical neurons implicated CaMKII and release of Ca2+ from intracellular stores in the potentiation of CaV1.3 currents at negative voltages following stimulation with insulin-like growth factor 1 (Gao et al., 2006b). Analogous to the role of cAMP-dependent protein kinase (PKA) anchoring proteins for PKA regulation of Cav1 channels (Hulme et al., 2004), feedback modulation of Cav1.3 by CaMKII may require additional adaptor proteins present in neurons but not HEK293T cells. Densin was considered since it binds to CaMKII (Strack et al., 2000; Walikonis et al., 2001) and contains a type I PDZ domain which could associate with the corresponding recognition site at the distal C-terminus (CT) of α11.3 (Fig.2A). Consistent with this possibility, GFP-tagged densin coimmunoprecipitated with FLAG-tagged α11.3 in HEK293T cells (Fig.2B) and bound in vitro to GST-tagged proteins containing the α11.3 CT, but not GST (Fig.2C). To test the importance of the PDZ-binding sequence of α11.3 for the interaction, we mutated the C-terminal leucine to alanine in α11.3 (CTL-A), which should prevent PDZ binding (Songyang et al., 1997; Calin-Jageman et al., 2007). Unlike for the wild-type α11.3 CT, the densin PDZ domain did not bind to CTL-A (Fig.2D). These results confirmed a direct interaction of densin with the α11.3 CT PDZ-binding sequence.
We next investigated the potential for densin and Cav1.3 to interact in neurons. We first tested if Cav1.3 channels and densin were associated with the same subcellular compartments in neurons. To restrict analysis to plasma membrane channels, we analyzed the localization of transfected hemagglutinin (HA)-tagged α11.3 in which the HA tag was inserted in an extracellular domain of α11.3. Immunofluorescence with HA antibodies applied to live neurons cotransfected with eGFP and HA-α11.3 revealed a punctate distribution for Cav1.3 along the shaft and spines throughout the dendritic arbor (Fig.3A). GFP-tagged densin showed a similar distribution, which was predominantly postsynaptic as indicated by colocalization with PSD-95 (Fig.3B).
Unfortunately, we could not determine if both GFP-densin and Cav1.3 were colocalized since cotransfection of the corresponding cDNAs was deleterious to neuronal survival (data not shown). Therefore, we tested for a physical interaction between densin and Cav1.3 by coimmunoprecipitation. The α11.3 antibodies, but not control IgG, coimmunoprecipitated densin with α11.3 from solubilized mouse hippocampal membrane extracts (Fig.3C). In the reverse approach, α11.3 was similarly brought down by densin antibodies (Fig.3D). Consistent with a role for densin in scaffolding CaMKII to the channel complex, CaMKII was coimmunoprecipitated regardless of whether antibodies against densin or α11.3 were used (Fig.3C,D). Moreover, CaMKII antibodies specifically coimmunoprecipitated both densin and α11.3 (Fig.3E), despite a weak non-specific interaction of CaMKII with the control IgG, which was most likely due to abundance of CaMKII in hippocampal extracts. Collectively, these results support the existence of a ternary complex comprised of densin, CaMKII and Cav1.3 channels in the hippocampus.
To test if densin may functionally recruit CaMKII for modulation of Cav1.3, we analyzed the effect of cotransfecting densin+CaMKII with Cav1.3 in HEK293T cells. While densin+CaMKII did not affect voltage-dependent activation or facilitation of Cav1.3 (data not shown), they significantly increased the amplitude of Cav1.3 Ca2+ currents (ICa) during trains of depolarizations (100 Hz, Fig.4A). With this voltage protocol, ICa in cells transfected with Cav1.3 alone inactivates rapidly (~40% within 50 ms, Fig.4A), due to Ca2+-dependent inactivation mediated by calmodulin (Yang et al., 2006). In cells cotransfected with densin+CaMKII, ICa inactivated significantly less, with amplitudes at the end of the 300-ms train that were ~45% greater than in cells with Cav1.3 alone (Fig.4A,C). The enhancement of ICa was independent of changes in voltage-dependent activation or peak current amplitude which were not different in cells transfected with Cav1.3 alone (k=−12.0±1.0; V1/2=−7.1±0.3; ICa amplitude at −10 mV=1298.9±304.6 pA; n=9) and Cav1.3+densin+CaMKII (k=−12.5±2.2, p=0.68; V1/2=−7.2±0.5, p=0.34; ICa amplitude at −10 mV=649.4±165.2 pA, p=0.21; n=10; by t-test).
To determine if the effect of densin+CaMKII was Ca2+-dependent, we analyzed Ba2+ currents (IBa). Because Ba2+ does not support Ca2+-dependent inactivation, the amplitude of IBa remains relatively constant throughout the train (Fig.4B). In contrast to effects on ICa, densin+CaMKII modestly increased inactivation of IBa (~11.8%; Fig.4B,C). The effects of densin+CaMKII were not seen when densin or CaMKII were singly cotransfected with Cav1.3 (Fig.4C). These results reveal that Cav1.3 currents undergo Ca2+-dependent facilitation (CDF) during repetitive stimuli, which requires both densin and CaMKII.
Since densin binds both α11.3 CT (Fig.2C,D) and CaMKII (Strack et al., 2000; Walikonis et al., 2001), we hypothesized that CDF requires densin to scaffold CaMKII to the α11.3 CT. If so, then preventing these interactions should block CDF. We tested this prediction first with α11.3 containing the L-A mutation that prevents densin binding (Fig.2D). As expected, there was no significant effect of densin+CaMKII on ICa inactivation for channels containing the L-A mutation (p=0.15; Fig.5A). We next examined the effect of deleting the PDZ domain from densin (ΔPDZ), which should also prevent binding to α11.3 CT. Unlike full-length densin, the ΔPDZ truncation did not affect ICa amplitude at the end of the train (p=0.82 compared to Cav1.3 alone; Fig.5B). Together, these data confirm the requirement for densin binding to α11.3 CT for CDF.
CaMKII directly interacts with the C-terminal domain of densin in vitro (see Fig. 2A; (Strack et al., 2000; Walikonis et al., 2001)) and this interaction is sufficient for co-immunoprecipitatation of CaMKII with densin from transfected HEK293 cells (Jiao et al., 2008). In ongoing studies to define domains in full-length densin that are necessary for interaction with CaMKII, we found that a large internal deletion (Δ483-1377) in a naturally occurring densin splice variant substantially reduces the co-immunoprecipitation of CaMKII when compared to the full-length densin (Fig. 5C). Apparently, the deleted region (Δ483-1377) is required for CaMKII binding. Since the Δ483-1377 densin variant retains the PDZ domain and can bind to the α11.3 CT in pull-down assays (data not shown), we used it to determine if weakened interactions of densin with CaMKII affect CDF. In contrast to full length densin, the Δ483-1377 variant did not support CDF of Cav1.3 ICa during repetitive stimuli (p=0.16 compared to Cav1.3 alone; Fig.5D). Thus, our results support a mechanism where densin recruits CaMKII to the α11.3 CT, which promotes CDF of Cav1.3.
To probe further the relevance of CaMKII activity for Cav1.3 CDF, we used the CaMKII inhibitor, KN93, which blocks CaMKII activation by competing for the binding of Ca2+/calmodulin. Since long incubations (1-2 h) with extracellular KN93 (10 μM) nonspecifically inhibit Cav1.3 channels independent of CaMKII (Gao et al., 2006a), we included KN93 in the intracellular solution, which did not affect the amplitude or other properties of Cav1.3 currents (Supplementary Table 1). However, KN93 prevented the effect of densin and CaMKII on CDF, whereas the inactive analog, KN92, had no effect (Fig.6A). We also tested the effect of inhibiting CaMKII by directly blocking the active site with CaM-KIINtide, a peptide derived from the naturally occurring CaMKII inhibitor protein, CaM-KIIN (Chang et al., 1998). CaM-KIINtide (10 μM), introduced via the patch electrode in cells cotransfected with Cav1.3+densin+CaMKII, also inhibited Cav1.3 CDF: this group was not significantly different from cells transfected with Cav1.3 alone (Fig.6B). We also tested the effect of mutating threonine286 to alanine (T286A) in CaMKII, which prevents autonomous kinase activity after dissociation of Ca2+-calmodulin from CaMKII (Soderling, 1996). The T286A mutant, when coexpressed with densin and Cav1.3 did not support CDF (Fig.6C). These results confirm that CaMKII activation and autonomous activity is necessary for Cav1.3 CDF.
The densin-dependent positioning of CaMKII within the channel complex may allow for local activation of CaMKII by Ca2+ that emerges from the pore of individual Cav1.3 channels. If so, there should be little reliance of CDF on macroscopic current amplitude, since only the single-channel current would be relevant. Alternatively, CDF may depend on a Ca2+ microdomain due to Ca2+ influx through, and accumulation near, neighboring channels. In this case, CDF should be greater for large- than small-amplitude currents. To distinguish between these mechanisms, we separately analyzed cells with small- (<250 pA) and large- (>250 pA) amplitude currents. We found that CDF due to densin and CaMKII was significant only for the large-amplitude ICa (Fig.7A,B), which supported the importance of a Ca2+ microdomain for CDF. However, Cav β subunits potentiate peak current amplitude (Perez-Reyes et al., 1992) and influence CaMKII regulation of Cav1.2 (Grueter et al., 2006; Grueter et al., 2008; Abiria and Colbran, 2009). Thus, insensitivity of small currents to densin and CaMKII could have been related to low levels of Cav β expression. If so, small currents should show activation voltages that are positively shifted relative to large currents, since all Cav β subunits cause a negative shift in activation voltage of Cav1 channels (Perez-Reyes et al., 1992). There was no difference in voltage-dependent activation of small- and large-amplitude currents in cells with Cav1.3 alone or cotransfected with densin and CaMKII (Supplementary Table 2), which showed that small currents arose from the same Cav1.3 subunit composition as large currents.
The Ca2+-dependence of CDF was further investigated by altering intracellular Ca2+ buffering strength. For this purpose, we substituted EGTA (5 mM) in the intracelullar recording solution with BAPTA (10 mM). Due to its faster Ca2+ binding kinetics compared to EGTA (Tsien, 1980), BAPTA should more quickly nullify Ca2+ increases which support CDF. BAPTA even at 10 mM concentration, is permissive for Ca2+-dependent inactivation of Cav1.3 (Dick et al., 2008), which depends on nanodomain Ca2+ signals emanating from individual channels. If CDF depends on a similar Ca2+ nanodomain, BAPTA should spare CDF. However, BAPTA effectively blunted the effects of densin and CaMKII on Cav1.3 CDF (Fig.7C). Taken together, these results show that Cav1.3 CDF requires a Ca2+ microdomain that is supported by multiple open channels.
The activity of CaMKII depends on the frequency of Ca2+ transients in excitable cells (Meyer et al., 1992; Hanson et al., 1994; De Koninck and Schulman, 1998). High-frequency Ca2+ spikes limit Ca2+/calmodulin dissociation, thus enhancing CaMKII autophosphorylation, which supports autonomous enzymatic activity (De Koninck and Schulman, 1998; Hudmon and Schulman, 2002). If frequency-dependent modulation of CaMKII contributes to Cav1.3 CDF, the effects of densin and CaMKII should be reduced during sustained or low-frequency depolarizations. Alternatively, if CDF depends only on a single Ca2+ burst due to rapid opening of Cav1.3 channels, CaMKII and densin should still facilitate Cav1.3 ICa evoked by sustained or low-frequency depolarizations. To distinguish between these possibilities, we first analyzed ICa during a 300-ms step depolarization. The ratio of the residual amplitude of ICa at the end of the pulse normalized to the peak current amplitude (Ires/Ipeak, Fig.8A) provided a metric analogous to that used for assessing CDF at the end of the 100-Hz train (Fig.4A). If repetitive depolarizations are necessary for CDF, densin and CaMKII should not influence Ires/Ipeak. In cells transfected with Cav1.3 alone, Ires/Ipeak for ICa shows U-shaped dependence on test voltage due to Ca2+-dependent inactivation (Brehm and Eckert, 1978). Interestingly, densin+CaMKII actually increased inactivation at a more positive voltages (+30 mV, Ires/Ipeak=0.0.26±0.03 for Cav1.3 alone, n=7 vs. 0.15±0.03 for Cav1.3+densin+CaMKII, n=13; p<0.05; Fig.8A). This appears to result from enhanced voltage-dependent inactivation by densin+CaMKII, which was evident as increased inactivation of IBa in triply transfected cells during repetitive and sustained depolarizations (Fig.4B,C; Supplementary Fig.2). However, with all other test voltages, including the same voltage used in the 100-Hz protocol (−10 mV, Fig.Fig.44,,5),5), Ires/Ipk was not significantly different in cells transfected with Cav1.3 alone and those cotransfected with densin+CaMKII (Fig.8A). This result demonstrates that a sustained depolarization is insufficient to produce Cav1.3 CDF.
High-frequency activation of Cav1.3 channels may trigger the rapid accumulation of Ca2+ that facilitates Ca2+/calmodulin binding to CaMKII, autophosphorylation of the kinase at Thr286, and changes in channel gating that underlie CDF. Due to Ca2+ diffusion from microdomains supporting CDF, low-frequency activation of Cav1.3 channels may be less effective in promoting autonomous CaMKII activity. To test this, we characterized CDF during 50-Hz trains of depolarizations. As expected, there was no significant effect of densin and CaMKII on CDF with this voltage protocol, even for long (1-s) trains (p=0.77 compared to 100 Hz stimulation; Fig.8B). To further define the Ca2+ signal that underlies CDF, we performed simultaneous electrophysiological and optical recordings with the ratiometric Ca2+ indicator Fura-2, which was introduced into cells via the patch electrode. These experiments showed that within the 300-ms train, stimulation at 100 Hz caused a significantly greater (~32%) and faster (~39%) increase in Ca2+ compared to 50 Hz (Fig.8C,D). Taken together, these data highlight the importance of high-frequency stimulation for fast and robust increases in Ca2+ that support Cav1.3 CDF.
Our results reveal a novel feedback regulation of Cav1.3 channels that involves multivalent interactions between Cav1.3 α1 subunits, densin and CaMKII. Densin binding to the distal C-terminal domain of α11.3 permits CaMKII-dependent facilitation of Cav1.3 Ca2+ currents during high-frequency, depolarizing stimuli. This regulation depends on precise patterns of Cav1.3 Ca2+ influx, CaMKII binding to densin, and CaMKII autophosphorylation. Association of Cav1.3, densin, and CaMKII at some synapses may coordinate activity-dependent potentiation of L-type Ca2+ currents underlying alterations in synaptic efficacy and other neuronal functions.
The regulation of voltage-gated (Cav) Ca2+ channels by permeating Ca2+ ions permits fast and efficient control of Ca2+ signals in excitable cells. Ca2+-dependent inactivation (CDI) curtails Ca2+ influx and is mediated by CaM binding to the proximal C-terminal domain of the Cav α1 subunit (Halling et al., 2006). Ca2+-dependent facilitation (CDF) boosts Ca2+ entry through Cav channels, but via multiple mechanisms. For Cav1.2, CDF involves CaMKII binding to and phosphorylation of α1 and/or β subunits (Hudmon et al., 2005; Grueter et al., 2006). However, in recombinant systems, CDF of whole-cell Cav1.2 currents is not evident unless CDI is first inhibited by mutations of the CaM binding (IQ) domain (Zühlke et al., 1999; Zühlke et al., 2000). In contrast, Cav1.2 channels in cardiac myocytes undergo overt CDF (Noble and Shimoni, 1981; Marban and Tsien, 1982; Lee, 1987), suggesting that additional factors are permissive for CDF of native Cav1.2 channels. We also did not detect CDF for Cav1.3 transfected alone or cotransfected only with CaMKII in HEK293Tcells (Fig.4). Due to CDI, Cav1.3 ICa showed only inactivation during 100-Hz depolarizations (Fig.4A). However, we interpret the enhanced ICa amplitudes in cells cotransfected with densin and CaMKII (Fig.4A,C) as CDF, which overlaps temporally with CDI during the 100 Hz train. Facilitation was Ca2+-dependent since it was seen for ICa and not IBa (Fig.4B), increased with ICa amplitude (Fig.7A,B, and was inhibited by BAPTA in the intracellular recording solution (Fig.7C).
Our results indicate that densin scaffolds CaMKII to the α11.3 CT, which enables local activation of CaMKII by Ca2+/calmodulin. The ability of CaMKII to respond to the frequency of Ca2+ oscillations with different levels of activity is well-established (De Koninck and Schulman, 1998) and involves Ca2+/calmodulin-stimulated autophosphorylation at Thr286 of individual subunits, 12 of which form the CaMKII holoenzyme (Hudmon and Schulman, 2002). The importance of high-frequency depolarizations and CaMKII Thr286 autophosphorylation in our experiments (Fig.(Fig.6C6C,,8)8) implies that CDF requires CaMKII catalytic activity. Potential substrates for CaMKII phosphorylation include densin (Strack et al., 2000; Walikonis et al., 2001). In addition, CaMKII may phosphorylate the Cavβ subunit (Grueter et al., 2008), which increases the open probability of cardiac Cav1.2 channels (Grueter et al., 2006). Finally, CaMKII could phosphorylate α11.3. In SHSY5Y cells and cortical neurons, CaMKII-dependent enhancement of CaV1.3 following stimulation with insulin like growth factor 1 is prevented by mutation of Ser1486 in α11.3 to Ala (Gao et al., 2006b), although it has not been shown that CaMKII directly phosphorylates Ser1486. Moreover, insulin like growth factor 1 enhanced Ba2+ currents carried by CaV1.3 channels by shifting voltage-dependent activation of the channel to more negative voltages, independent of repetitive stimulation. Thus, the CaMKII- and densin-dependent CDF in our experiments appears quite distinct from previously described modes of Cav1.3 regulation. Further studies will be necessary to dissect the molecular targets of CaMKII and their involvement in CDF and other forms of Cav1.3 modulation.
Densin interacts with a number of postsynaptic proteins, including Shank (Quitsch et al., 2005), δ-catenin (Martinez et al., 2003), and α-actinin (Walikonis et al., 2001; Robison et al., 2005), but how densin influences these proteins is generally not known. Our results indicate that densin may scaffold CaMKII to postsynaptic Cav1.3 channels, much like A-kinase anchoring proteins tether PKA (Hulme et al., 2003) and calcineurin (Oliveria et al., 2007) to Cav1.2. This mechanism allows for fast and efficient modulation of Cav1 channels, which is consistent with the millisecond time course of Cav1.3 facilitation we found in our experiments (Fig.4A). At the same time, CaMKII is also a transducer of Cav1 Ca2+ signals. For example, CaMKII responds to Cav1 channel opening by forming clusters at the neuronal plasma membrane and selectively couples Cav1 but not Cav2 channels to pCREB activation in response to moderate depolarizations (Wheeler et al., 2008). Cav2 channels are not likely to bind densin since they lack a type I PDZ-binding sequence. Association of densin and CaMKII with Cav1.3 in the brain (Fig.3C-E) may therefore contribute to the Cav1-specific nature of pCREB signaling (Zhang et al., 2006).
Although CDF due to densin and CaMKII was only seen for large-amplitude currents recorded in high concentrations of extracellular Ca2+ (10 mM) during intense depolarizing stimuli, we predict that Cav1.3 CDF due to densin and CaMKII will be physiologically relevant in neurons. Cav1.3 and densin are concentrated in dendritic spines (Fig.3A), which have a relatively small volume, particularly compared to HEK293T cells. Previous work shows that Cav1.2 channels are clustered in spines, where it was estimated that there are ~8 channels per cluster (Obermair et al., 2004). Similar clustering of Cav1.3 channels within spines, which is suggested by the immunofluorescence of HA-α11.3 (Fig.3A; see also Gao et al., 2006b), should efficiently produce the Ca2+ microdomain required for CDF even under physiological (<2 mM) extracellular Ca2+ concentrations. During stimuli that promote long-term potentiation, CaMKII activation in dendritic spines is mediated by Cav1 channels and is blocked by 20 mM BAPTA but partially spared by the same concentration of EGTA (Lee et al., 2009). In this context, densin and CaMKII may endow Cav1.3 channels with a positive feedback regulation to boost local Ca2+ signals that initiate CaMKII activation and participation in long-term synaptic plasticity.
However, densin and CaMKII may also underlie pathological changes associated with hyperactivation of Cav1.3 channels. For example, in the striatum, the excitability of striatopallidal neurons is regulated by Cav1.3 channels, which are inhibited by D2 dopamine receptors (Olson et al., 2005). Excessive Ca2+ influx via Cav1.3 channels following dopamine depletion results in a loss of dendritic spines in striatopallidal neurons, since these morphological changes are not seen in mice lacking Cav1.3 (Day et al., 2006). Intriguingly, CaMKII activity is also upregulated upon dopamine depletion (Picconi et al., 2004; Brown et al., 2005), which may further exacerbate Ca2+ overloads by promoting Cav1.3 CDF. Moreover, CaMKII inhibition alleviates defects in synaptic plasticity and motor deficits following striatal dopamine depletion (Picconi et al., 2004). Thus, elucidating the functional relationships between densin, CaMKII, and Cav1.3 at striatopallidal synapses may yield further insights into the potential therapeutic efficacy of Cav1.3 blockers for Parkinson’s disease (Chan et al., 2007; Chan et al., 2009).
The authors thank Diane Lipscombe and Randy Hall for cDNAs, Jörg Striessnig for Cav1.3 KO mice, and Irina Calin-Jageman for help in generating the HA-tagged Cav1.3. This work was supported by grants from the NIH (HL087120, DC009433 to AL; MH63232 to RJC; NS054614 to YMU; and T32 HL0071212 to CJC), the American Heart Association (AL), Israel Binational Science Foundation (AL) and Austrian Science Fund (P17807-B05 to GJO).