Abstract
Neurons express multiple types of voltage-gated calcium (Ca2+) channels. Two subtypes of neuronal L-type Ca2+ channels are encoded by CaV1.2 and CaV1.3 pore-forming subunits. Both CaV1.2 and CaV1.3 subunits contain class I PDZ (postsynaptic density-95/Discs large/zona occludens-1) domain-binding consensus at their C termini. In yeast two-hybrid screen of rat brain cDNA library with the C-terminal bait of CaV1.3a (long C-terminal splice variant) L-type Ca2+ channel subunit, we isolated multiple clones of postsynaptic adaptor protein Shank. We demonstrated a specific association of CaV1.3a C termini, but not of CaV1.2 C termini, with Shank PDZ domain in vitro. We further demonstrated that the proline-rich region present in C termini of CaV1.3a subunit binds to Shank Src homology 3 domain. We established that CaV1.3a and Shank localized to postsynaptic locations in cultured rat hippocampal neurons. By expressing epitope-tagged recombinant CaV1.3 subunits in rat hippocampal neuronal cultures, we demonstrated that the presence of Shank-binding motifs in CaV1.3a sequence is both necessary and sufficient for synaptic clustering of CaV1.3 L-type Ca2+ channels. In experiments with dominant-negative peptides and dihydropyridine-resistant CaV1.3a mutants, we demonstrated an importance of Shank-binding motif in CaV1.3a sequence for phosphorylated cAMP response element-binding protein (pCREB) signaling in cultured hippocampal neurons. Our results directly link CaV1.3 neuronal L-type Ca2+ channels to macromolecular signaling complex formed by Shank and other modular adaptor proteins at postsynaptic density and provide novel information about the role played by CaV1.3 L-type Ca2+ channels in pCREB signaling.
- calcium channels
- PDZ domains
- protein targeting
- postsynaptic density
- CREB
- synapse
- synaptic plasticity
Introduction
Neurons express a diverse complement of voltage-gated calcium (Ca2+) channels (Catterall, 2000). N-type (CaV2.2) and P/Q-type (CaV2.1) Ca2+ channels mediate rapid Ca2+ influx into the presynaptic terminal that triggers synaptic vesicle fusion and neurotransmitter release (Dunlap et al., 1995). Neuronal L-type Ca2+ channels do not support synaptic transmission; however, recent data indicate that L-type Ca2+ channels play a critical role signaling. Ca2+ influx via postsynaptic L-type in neuronal Ca2 Ca2+ channels activates phosphorylated cAMP response element-binding protein (pCREB) (Bito et al., 1996; Dolmetsch et al., 2001; Weick et al., 2003) and nuclear factor of activated T-cells c4 (Graef et al., 1999) neuronal nuclear transcription factors. Resulting changes in protein expression may play a role in mediating long-term changes in synaptic strength (Deisseroth et al., 2003). Despite an important role of L-type Ca2+ channels in neuronal Ca2+ signaling, very little is known about molecular and cellular mechanisms involved in L-type Ca2+ channel function in neurons. Neuronal L-type Ca2+ channels are encoded by CaV1.2 (α1C) or CaV1.3 (α1D) pore-forming subunits (Hell et al., 1993; Lipscombe et al., 2004). When compared with channels encoded by CaV1.2 subunit, the channels encoded by CaV1.3 subunit are less sensitive to dihydropyridines and open at more negative membrane potentials (Koschak et al., 2001; Xu and Lipscombe, 2001; Lipscombe et al., 2004; Olson et al., 2005). Genetic ablation of CaV1.3 subunit resulted in congenital deafness and dysfunction of pancreatic β-cell and cardiac sinoatrial node (Platzer et al., 2000; Namkung et al., 2001; Zhang et al., 2002; Mangoni et al., 2003). The neuronal phenotype of CaV1.3 knock-out mice is described by Olson et al. (2005).
The CaV1.2 (α1C) subunit contains a class 1 PDZ [postsynaptic density-95 (PSD-95)/Discs large/zona occludens-1 (ZO-1)] domain binding C-terminal motif VSXL (see Fig. 1A) that has been shown to associate with PDZ domains of neuronal interleukin-16 (NIL-16) precursor protein (Kurschner and Yuzaki, 1999) and channel-interacting PDZ domain protein (CIPP) (Kurschner et al., 1998). Association with PDZ proteins plays an important role in coupling CaV1.2 L-type Ca2+ channels with phosphorylation of nuclear CREB (Weick et al., 2003). The C-terminal region of CaV1.3 subunit is alternatively spliced (Safa et al., 2001; Xu and Lipscombe, 2001). The “long” CaV1.3a splice variant also contains a class 1 PDZ domain-binding C-terminal motif ITTL (see Fig. 1A), but its binding partners are unknown. Here, we performed a yeast two-hybrid (Y2H) screen of rat brain cDNA library with the CaV1.3a C-terminal bait and isolated multiple clones of Shank postsynaptic PDZ-containing adaptor protein. In biochemical experiments, we confirmed specific association between CaV1.3a C-terminal and Shank-PDZ domain and discovered that proline-rich region in CaV1.3a C-terminal region binds to Shank-Src homology 3 (SH3) domain. In experiments with cultured hippocampal neurons, we demonstrated that the presence of Shank-binding motifs in CaV1.3a sequence is both necessary and sufficient for synaptic clustering of CaV1.3 L-type Ca2+ channels and that the same motif plays an important role in CaV1.3-mediated signaling to pCREB. Our results directly link CaV1.3 L-type Ca2+ channels to the macromolecular signaling complex formed by Shank in postsynaptic spines (Kennedy, 2000).
CaV1.3a C terminal binds to Shank-PDZ domain in yeast two-hybrid assay. A, Sequence alignment of rat CaV1.3a C-terminal region (aa 2109-2155) with the corresponding regions of human and chicken CaV1.3a subunits; with rat, human, and rabbit CaV1.2 subunits, and with the C. elegans L-type Ca2+ channel pore-forming subunit (egl-19). Sequence accession numbers are indicated; the class 1 PDZ domain-binding consensus is boxed. The LDC bait (aa 2142-2155 of rat CaV1.3a) is shown by a bar (above). B, Domain structure of rat Shank1 (2087 aa), rat Shank 2 (1474 aa), and rat Shank 3 (1740 aa). The boundaries of sequenced Shank1/3 clones isolated in the Y2H screen with LDC bait are indicated. Clones C02 and C39 correspond to Ins2a(-) Shank1 splice variant (Lim et al., 1999). C, LDC binds Shank1/3-PDZ domains. The strength of interaction between LDC bait and isolated PDZ domains of Shank1, Shank3, and other adaptor proteins was measured in Y2H liquid assay by β-galactosidase activity. D, The Shank3-PDZ domain is specific for T-X-L sequence at C termini of LDC bait. The wild-type and mutant LDC baits, LCC bait, and chimeric LDC-LCC bait were tested for the strength of association with Shank3-PDZ domain in Y2H assay by β-galactosidase activity. In C and D, the data were normalized to LDC/Shank3-PDZ pair and shown as mean ± SEM (n ≥ 3).
Materials and Methods
Plasmid construction. pLexN (bait) and pVp16-3 (prey) vectors were used for Y2H, pGEX-KG (Amersham Biosciences, Piscataway, NJ) for glutathione S-transferase (GST) fusion proteins; and pEGFP-C3 (Clontech, Palo Alto, CA) for green fluorescent protein (GFP) fusion proteins. Rat CaV1.3a (α1D-1) (Ihara et al., 1995; Xu and Lipscombe, 2001) constructs were as follows: LDC, 2142-2155; LDC2, 1640-2155; LDC2ΔC, 1640-2150; LDC5, 2121-2155; LDC5ΔC, 2121-2150. Deletions and mutations of LDC bait were generated by PCR and verified by sequencing. Rabbit CaV1.2 (α1C) (Mikami et al., 1989) construct was as follows: LCC, 2121-2171. Shank constructs were as follows: Shank3-PDZ, 566-671 of rat Shank3a; Shank3-SH3-PDZ, 473-664 of Shank3a; Shank1-PDZ, 659-764 of rat Shank1a; Shank1-SH3, 447-558 of rat Shank1B; Shank1-SH3-PDZ, 447-703 of rat Shank1B. The GFP-Shank1B construct was described previously (Sala et al., 2003). Constructs encoding PDZ domains of PSD-95, PSD-93, synapse-associated protein 102 (SAP-102), ZO-1, and Mint1 were described previously (Maximov et al., 1999). HA-CaV1.3a construct was generated by replacing NheI[5′ untranslated region (UTR)]/BsiWI (518) fragment of CaV1.3a plasmid in pcDNA6/V5-His vector (Xu and Lipscombe, 2001) with PCR-amplified 450 bp 5′ fragment of HA-CaV2.2 (HA-α1B) expression construct (Maximov and Bezprozvanny, 2002). The HA-CaV1.3a-ΔC construct (deletion of amino acids 2151CITTL 2155) was generated by PCR and verified by sequencing. The HA-CaV1.3b construct was generated by subcloning BamH1(2,751)/SacII (3′UTR) fragment from CaV1.3b plasmid (Xu and Lipscombe, 2001) into HA-CaV1.3a construct. The CaV1.3-T1033Y mutation was generated by Quick-Change kit (Stratagene, La Jolla, CA) and verified by sequencing.
Yeast two-hybrid methods. L40 yeast strain was sequentially transformed by lithium acetate method with the pLexN-LDC bait and rat brain cDNA library (kind gift from Thomas Südhof, University of Texas Southwestern, Dallas, TX) in pVp16-3 prey vector. The primary transformation efficiency was 6 × 106. The positive clones were selected on -THUL selection plates and rescued in HB101 Escherichia coli strain. Y2H liquid assays were performed as described previously (Maximov et al., 1999).
In vitro binding assays. GST-fusion proteins were expressed in BL21 E. Coli strain and purified as described previously (Maximov et al., 1999). GFP-tagged proteins were expressed in human embryonic kidney 293 (HEK293) cells by Ca2+-phosphate method and extracted in buffer A containing 0.5% Triton X-100, 20 mm Imidasole, pH 6.8, 100 mm NaCl, 5 mm EDTA, and protease inhibitors. Protein extracts were clarified by centrifugation at 100,000 × g and used in GST pull-down experiments performed as described previously (Maximov et al., 1999).
Hippocampal neuronal cultures and immunocytochemistry. Primary embryonic day 18 (E18) rat hippocampal neuronal cultures were established and analyzed by immunostaining at 10-12 d in vitro (DIV) as described previously (Maximov and Bezprozvanny, 2002). For targeting experiments, primary hippocampal neuronal cultures at 9-10 DIV were transfected with GFP-LDC2/5 plasmids or cotransfected with HA-CaV1.3/β3/α2δ-1 plasmids using calcium phosphate method as described previously (Maximov and Bezprozvanny, 2002). Subcellular localization of GFP-LDC constructs was determined by GFP imaging 72 h after transfections using confocal microscope (Zeiss LS800; Zeiss, Thornwood, NY). Subcellular localization of HA-CaV1.3 was determined by immunostaining with anti-HA monoclonal antibody (mAb) 72 h after transfection. GFP-Shank1B plasmid was cotransfected with HA-CaV1.3 plasmids as indicated. Obtained images were collected with Olympus IX-71 microscope with 40× objective by Cascade-650 camera (Rhoper, Tucson, AZ) and analyzed using MetaFluor (Universal Imaging, West Chester, PA) and Adobe PhotoShop (Adobe Systems, San Jose, CA) software packages.
Real-time reverse transcription-PCR. Primary hippocampal neuronal cultures at 10-12 DIV were stimulated by application of KCl for 30 s, followed by 90 min incubation in the culture medium, washed twice with PBS, frozen at -80, and used for RNA extraction by RNAeasy kit (Qiagen, Valencia, CA). The RNA was DNase treated to avoid DNA contamination. Reverse transcription (RT) was done using SuperScriptIII kit (Invitrogen, Carlsbad, CA) using oligo-dT primer according to the instructions of the manufacture. The sample cDNA and control plasmids were preamplified by eight cycles of conventional PCR using a common forward primer GACAGTCATGTTCAATGCAAC (position 5112; GenBank accession number NM017298) and the reverse primers CAGCGGTTTCTCAGTATCACTTG (position 5645; GenBank accession number NM017298) for CaV1.3a (“long” form) and CCATCGTCAGCAAGTTGTGTTCAT (position 5329; GenBank accession number AF370009) for CaV1.3b (“short” form). Real-time fluorescence-based PCR assay was performed with the LightCycler (Roche Molecular Biochemicals, Indianapolis, IN) using SYBR green with nested primers. The common forward primer for both forms of CaV1.3 cDNA was TACGGACGGCTCTCAAGATCAAG (position 5147; GenBank accession number NM017298), and the reverse primer for CaV1.3a (long form) was GACGGTGGGTGGTATTGGTCTGC (position 5589); the reverse primer for CaV1.3b (short form) was GCGGTAGCTCAGGCAGACAACTC (position 4932; GenBank accession number AF370009). Triplicate runs were performed for the samples, and duplicate runs were performed for each concentration of the control plasmids. A calibration curve was obtained by amplification of CaV1.3a and CaV1.3b plasmid DNA at concentrations ranging from 10 to 10,000 mol/μl. Template controls were included in each run.
Current recordings in Xenopus oocytes. HA-CaV1.3 subunits were expressed in Xenopus oocytes as described previously (Bezprozvanny and Tsien, 1995). Briefly, GFP-Shank1B construct (Sala et al., 2003) was subcloned into pcDNA3 vector. RNA species encoding HA-CaV1.3 and HA-CaV1.3-T1033Y subunits, β3 and α2δ-1 subunits, and GFP-Shank1B were synthesized by in vitro transcription procedure with the use of bacteriophage T7 RNA polymerase. Stage V-VI Xenopus oocytes were prepared by collagenase A treatment and injected with cRNA mixtures as indicated. Currents were recorded 4-5 d after cRNA injection in 40 mm Ba2+ recording solution (40 mm Ba(OH)2, 50 mm TEA-OH, 2 mm KOH, 5 mm HEPES, adjusted to pH 7.4 with methanesulfonic acid) by two-electrode voltage-clamp amplifier (model OC-725A; Warner Instruments, Hamden, CT) controlled by pClamp6 software (Axon Instruments, Foster City, CA). Ca2+ channel openings were induced by 60 ms step depolarizations from holding potential of -80 mV to a range of test potentials as indicated. In nifedipine-blocking experiments, the holding potential was -60 mV and test potential was +10 mV. Collected data were analyzed off-line using pClamp6 software.
pCREB phosphorylation assay. Primary rat hippocampal neuronal cultures (8-10 DIV) were cotransfected with HA-CaV1.3-T1033Y or HA-CaV1.3-T1033Y-ΔC plasmids and β3 and α2δ-1 subunits. Seventy-two hours after transfection, the neurons were incubated in Tyrode solution in the presence of 1 μm TTX, 50 μm AP-5, 10 μm CNQX, 50 μm nifedipine for 3 h, stimulated for 90 s or 30 s by application of depolarizing solution in the presence of 1 μm TTX, 50 μm AP-5, 10 μm CNQX, 50 μm nifedipine [20 mm KCl in Tyrode (20K); 45 mm KCl in Tyrode (45K); 90 mm KCl in Tyrode (90K)], fixed, permeabilized, and stained with anti-pCREB polyclonal antibody (pAb) and anti-HA mAb. In electrical field stimulation (EFS) experiments, the neurons were transferred to Tyrode solution containing 50 μm AP-5, 50 μm nifedipine, and 20 μm bicuculline and stimulated at 5 Hz frequency by current pulses of 0.5 mA amplitude and 150 μs duration [World Precision Instruments (Sarasota, FL) A365 stimulator controlled by Neurolog Digitimer (Hertfordshire, UK)] applied via a pair of platinum electrodes (Bito et al., 1996). Transfected cells were identified by anti-HA mAb staining, and nuclear pCREB pAb staining was quantified by confocal imaging (Zeiss LS800). The intensity of nuclear pCREB staining in each series of experiments was normalized to mean pCREB signal in unstimulated cells and averaged across different experiments. Ca2+ imaging experiments (DeltaRAM illuminator; Photon Technology International, Lawrenceville, NJ) with primary hippocampal neuronal cultures were performed using fura-2 AM (Molecular Probes, Eugene, OR) as described previously for striatal medium spiny neurons (Tang et al., 2003).
RRRRRRRRR (R9), RRRRRRRRRALPDSRSYVSNL (R9-LCC), RRRRRRRRRLADEMICITTL (R9-LDC), and RRRRRRRRRYEKLSSIESDV (R9-NR2B) peptides were chemically synthesized, coupled to FITC at the N terminus, and dissolved in PBS. In loading experiments, the R9 peptides were added to rat hippocampal neurons (>10 DIV) for 10 min at 50 μm. After loading with R9 peptides, neurons were incubated in Tyrode solution in the presence of 1 μm TTX, 50 μm AP-5, and 10 μm CNQX for 3 h before 90 s KCl depolarization. pCREB immunostaining and Ca2+ imaging were performed and analyzed as described above with omission of nifedipine from the depolarizing solution.
Antibodies. Monoclonal antibodies were as follows: anti-synapsin1 (1:1000; Chemicon, Temecula, CA), anti-GFP (1:1000; Covance, Berkeley, CA), anti-HA (1:1000; HA.11 from Covance). Anti-pCREB pAb (1:1000) was from Upstate Biotechnology (Lake Placid, NY). Anti-Shank pAb (rabbit-EK3856, 1:1000; guinea pig-EK1123, 1:200) was kindly provided by Dr. Eunjoon Kim (KAIST, Daejeon, Korea). The rabbit anti-synapsin pAb (IB2881; 1:1000) were raised against keyhole limpet hemocyanin-conjugated NYLRRRLSDSNFMANLPNGYMTDLQRPQP peptide corresponding to N terminus of Synapsin1. The rabbit anti-CaV1.3a pAb (AM9742; 1:250) were generated against GST-LDC5 and affinity purified on Sepharose-conjugated LDC6 peptide (2132-2155 of rat CaV1.3a).
Results
CaV1.3a L-type Ca2+ channel binds Shank-PDZ in yeast two-hybrid system
We noticed that the C termini of long splice variant of neuronal L-type Ca2+ channel subunit CaV1.3a (α1D-1) (Safa et al., 2001; Xu and Lipscombe, 2001) contains a class 1 PDZ domain-binding consensus ITTL (Fig. 1A). Similar consensus sequence VTTL is conserved in Caenorhabditis elegans L-type Ca2+ channels subunit (egl-19) (Fig. 1A). A class 1 PDZ domain-binding motif VSXL is also present in cardiac/neuronal CaV1.2 (α1C) subunit (Fig. 1A), but it is absent in skeletal muscle CaV1.1 (α1S) and retinal CaV1.4 (α1F) L-type Ca2+ channels subunits (data not shown). The C termini of CaV1.2 associates with the PDZ domains of NIL-16 and CIPP proteins (Kurschner et al., 1998; Kurschner and Yuzaki, 1999), but the PDZ binding partners of CaV1.3a have not been described previously. To search for a novel CaV1.3a-binding partners, we performed a Y2H screen of rat brain cDNA library with the LDC bait (aa 2142-2155 of rat CaV1.3a) (Fig. 1A) and isolated 105 positive clones. When 13 isolated clones were chosen randomly and sequenced, we identified five Shank1 clones, two Shank3 clones, two PIST (TC10-specific interacting protein) clones, one ZNF198 clone (zinc finger protein 198), and the two clones for hypothetical proteins FLJ23209 and XP_147932. Because 7 of 13 sequenced clones corresponded to the Shank family members, we focused on Shank proteins as physiologically relevant CaV1.3a-binding partners in the remainder of the study.
Shank proteins are modular adaptor proteins that play a role of “master scaffold” in the postsynaptic specializations at glutamatergic synapses (Sheng and Kim, 2000). Three isoforms of Shank proteins are expressed in the brain (Lim et al., 1999) (Fig. 1B). All sequenced Shank1/3 clones contained a PDZ domain (Fig. 1B), indicating that LDC bait most likely binds to the Shank-PDZ domain. We screened all isolated clones by PCR with the primers specific for Shank1/2/3 PDZ domain sequences and discovered 20 Shank1 clones, no Shank2 clones, and 59 Shank3 clones. Thus, 79 of 105 clones isolated in the LDC screen (75.2%) corresponded to Shank1 and Shank3 proteins. The absence of Shank2 among isolated clones may be related to under-representation of Shank2 in cDNA library used for Y2H screen or unique features of Shank2 PDZ domain. To further characterize CaV1.3a-Shank interactions, we compared the interaction of LDC bait with Shank1 and Shank3 PDZ domains and with the PDZ domains of SAP-102, PSD-93 (chapsin-110), PSD-95, Z0-1, and Mint1 proteins in liquid Y2H assay. From all domains tested, the strongest interaction of LDC bait was observed with Shank3-PDZ domain followed by Shank1-PDZ domain (Fig. 1C), in agreement with the results of Y2H screen.
PDZ domains typically bind to the most C-terminal region of the target protein (Songyang et al., 1997). In agreement with this general paradigm, LDCΔC truncation mutant lacking ITTL motif failed to bind Shank3-PDZ domain in Y2H assay (Fig. 1D). Analysis of LDC bait point mutants (Fig. 1D) led us to conclude that Shank3-PDZ domain specificity (TXL) fits with the canonical class 1 PDZ domain-binding consensus S/T-X-Φ (Songyang et al., 1997). This conclusion is in agreement with the assignment of Shank-PDZ domain to (G,H) group (Bezprozvanny and Maximov, 2001), with the previous studies of Shank-PDZ domain ligand specificity (Sheng and Kim, 2000) and with the recently determined crystal structure of Shank-PDZ domain (Im et al., 2003). The CaV1.2 L-type Ca2+ channel subunit also contains class 1 PDZ domain-binding motif (Fig. 1A). However, in contrast to LDC bait, LCC bait (aa 2121-2171 of rabbit CaV1.2) failed to interact with Shank3-PDZ domain in liquid Y2H assay (Fig. 1D). The specificity of Shank3-PDZ for CaV1.3a C terminus was determined by the last four amino acids, because LDC-LCC chimeric bait (last four amino acids of LDC bait replaced with VSNL sequence from rat CaV1.2) was only weakly associated with Shank3-PDZ domain (Fig. 1D). From the Y2H analysis, we concluded that Shank-PDZ domain binds strongly and specifically to the C termini of CaV1.3a subunit.
The CaV1.3a C termini binds Shank in vitro
To independently test for the ability of CaV1.3a C termini to bind Shank PDZ domain, we expressed GFP-LDC2 fusion protein (aa 1640-2155 of rat CaV1.3a) (Fig. 2A) in HEK293 cells and performed pull-down experiments with GST-fusion proteins of Shank3, PSD-95, and Mint1 PDZ domains expressed in bacteria. We found that GST-Shank3-PDZ, but not other GST-fusion proteins, precipitated GFP-LDC2 (Fig. 2B). In agreement with the Y2H data, the Shank3-PDZ domain did not precipitate truncated GFP-LDC2ΔC protein (Fig. 2B). In a complementary series of experiments, we expressed GFP-Shank3-PDZ domain in HEK293 cells and performed pull-down experiments with GST-LDC5 (aa 2121-2155 of rat CaV1.3a) (Fig. 2A) and GST-LCC (aa 2121-2171 of rabbit CaV1.2) (Fig. 2A) fusion proteins. We found that GST-LDC5, but not GST-LDC5ΔC or GST-LCC, precipitated GFP-Shank3-PDZ (Fig. 2C). Thus, in vitro binding experiments confirmed specific association of Shank3-PDZ domain with C termini of CaV1.3a subunit.
CaV1.3a C terminal binds to Shank in vitro. A, Diagram of GFP-LDC2, GST/GFP-LDC5, and GST-LCC fusion proteins. B, GST, GST-Mint1-PDZ1-2, GST-PSD95-PDZ1, and GST-Shank3-PDZ were used to pull down GFP-LDC2 and GFP-LDC2ΔC from HEK293 cell lysates. C, GST-LDC5, GST-LDC5-ΔC, and GST-LCC were used to pull down GFP-Shank3-PDZ from HEK293 cell lysates. The input lanes in B and C contain 1/50 of HEK293 cell lysates used for pull-down experiments. D, E, GST and GST-Shank3-SH3-PDZ were used to pull down GFP-LDC2 and GFP-LDC2ΔC (D) or GFP-LDC5 and GFP-LDC5ΔC (E) from HEK293 cell lysates. F, G, GST and GST-Shank1-SH3 were used to pull down GFP-LDC2 and GFP-LDC2ΔC (F) or GFP-LDC5 and GFP-LDC5ΔC (G) from HEK293 cell lysates. The input lane in D (GFP-LDC2ΔC) and E-G contain 1/25 and for D (GFP-LDC2) contains 1/10 of HEK293 cell lysates used for pull-down experiments. For B-G, the precipitated proteins were recognized by anti-EGFP antibodies.
An additional analysis of CaV1.3a C-terminal sequence revealed a presence of upstream proline-rich region containing five putative SH3 domain-binding sites (PXXP-region) (Fig. 2A). One of these sites (1906PPTP1909) binds to SH3 domain of Rimbinding protein (Hibino et al., 2002); binding partners for other sites have not been identified. Shank proteins contain SH3 domain adjacent to PDZ domain (Fig. 1B), and we reasoned that Shank-SH3 domain may bind to CaV1.3a C-terminal PXXP-region. In agreement with this idea, we found that GST-Shank3-SH3-PDZ quantitatively precipitated GFP-LDC2 (Fig. 2D). GST-Shank3-SH3-PDZ also precipitated GFP-LDC5 and GFP-LDC2ΔC (Fig. 2D,E), albeit less efficiently than GFP-LDC2. Similar results were obtained in pull-down experiments with GST-Shank1-SH3-PDZ (data not shown). These results can be explained by association of Shank-SH3 domain with PXXP region in LDC2 sequence (Fig. 2A). In agreement with this interpretation, GST-Shank3-SH3-PDZ failed to precipitate GFP-LDC5ΔC (Fig. 2E), which lacks both PDZ and SH3 domain interaction motifs (Fig. 2A). To test this hypothesis directly, we performed pull-down experiments with GST-Shank1-SH3 domain. We discovered that GST-Shank1-SH3 domain precipitated GFP-LDC2 and GFP-LDC2ΔC (Fig. 2F) but not GFP-LDC5 and GFP-LDC5ΔC (Fig. 2G). We also found that GFP-LDC2 binds to GST-Shank1-SH3 with much lower affinity than to GST-Shank3-SH3-PDZ (Fig. 2, compare F and D), consistent with synergistic contribution of Shank SH3 and PDZ domains to association with CaV1.3a C-terminal region.
CaV1.3a and Shank colocalize at synaptic locations in hippocampal neurons
The CaV1.3 C terminal is alternatively spliced (Safa et al., 2001; Xu and Lipscombe, 2001). Only the long CaV1.3a splice variant contains Shank-SH3 and Shank-PDZ binding motifs. To determine relative abundance of CaV1.3 C-terminal splice variants, we extracted RNA from mature rat hippocampal neuronal cultures and performed a series of real-time RT-PCR experiments using CaV1.3 C-terminal splice-variant specific primers. We discovered that mRNA for long CaV1.3a C-terminal splice variant was present in ∼50-fold higher abundance in rat hippocampal neurons than short CaV1.3b form (Fig. 3A) and that 30 s depolarization of hippocampal neurons by 20K, 45K, or 90K did not induce a significant change in relative abundance of long and short CaV1.3 C-terminal splicevariants (Fig. 3A). On average, CaV1.3a/CaV1.3b concentration ratio was 53.6 ± 6.8 in control conditions, 40.8 ± 6.5 after 20K stimulation, 46.2 ± 15.2 after 45K stimulation, and 40.6 ± 6.0 after 90K stimulation (Fig. 3A).
CaV1.3a and Shank colocalize in synaptic locations in vivo. A, Real-time RT-PCR analysis of CaV1.3 C-terminal splice variant transcript abundance in rat hippocampal neuronal cultures. The means of triplicate determinations and SDs of concentrations of CaV1.3a (open bars) and CaV1.3b (filled bars) splice variant transcripts are shown for unstimulated neurons and for neurons stimulated by 30 s KCl depolarizations as indicated. Ctr, Control. B, Lysates from HEK293 cells transfected with GFP-LDC2 plasmid were analyzed by Western blotting with anti-CaV1.3a pAb (AM9742) or corresponding preimmune sera (P/S). C, D, Rat hippocampal neurons at 12 DIV were stained with anti-CaV1.3a pAb AM9742 (green) and anti-synapsin mAb (red) (C), anti-Shank pAb EK3856 (green), and anti-synapsin mAb (red) (D). Scale bars, 40 μm. Boxed sections in C and D are enlarged as indicated and shown below with corresponding profile scans.
To visualize subcellular localization of CaV1.3a subunit in hippocampal neurons, we generated anti-CaV1.3a rabbit polyclonal antibody (AM9742) against GST-LDC5 antigen. The AM9742 pAb was affinity purified on Sepharose-conjugated LDC6 peptide (2132-2155 of rat CaV1.3a) and verified in Western blotting experiments with GFP-LDC2 protein expressed in HEK293 cells (Fig. 3B). Generated anti-CaV1.3a rabbit pAb were used in double-labeling experiments with mAbs against synapsin. We found that clusters of CaV1.3a immunoreactivity were colocalized with synapsin in mature rat hippocampal neurons (Fig. 3C). The colocalization of CaV1.3a and synapsin clusters was further confirmed by profile scan (Fig. 3C). In additional double-labeling experiments, we confirmed synaptic localization of Shank in hippocampal neuronal cultures (Fig. 3D) as described previously (Naisbitt et al., 1999; Yao et al., 1999). Thus, we concluded that CaV1.3a and Shank are both clustered at synaptic locations in mature hippocampal neuronal cultures. In the accompanying study (Olson et al., 2005), we further demonstrate that CaV1.3a-Shank complex can be immunoprecipitated from rat brain synaptosomal lysates.
Targeting of recombinant CaV1.3 L-type Ca2+ channels in hippocampal neurons
To evaluate a role of Shank-binding motifs for CaV1.3 L-type Ca2+ channel targeting in neurons, we transfected cultured hippocampal neurons with GFP-LDC2 and GFP-LDC5 CaV1.3a C-terminal constructs (Fig. 4A). Subcellular localization of GFP-LDC2/5 constructs was visualized by GFP imaging and compared with MAP2 distribution determined by immunostaining. We found that GFP-LDC2 protein formed clusters in soma and dendrites of transfected neurons (Fig. 4B) but that GFP-LDC2ΔC, GFP-LDC5, and GFP-LDC5ΔC were diffusely distributed (Fig. 4C-E). Our results with GFP-LDC5 protein are consistent with the previously reported diffuse distribution of enhanced GFP (EGFP)-ITTL protein in hippocampal neurons (Weick et al., 2003). From these experiments, we concluded that the presence of both PDZ and SH3 domain-binding motifs is required to support formation of GFP-LDC2 clusters (Fig. 4B). This result is in contrast to analysis of CaV1.2 targeting (Weick et al., 2003), which demonstrated that the presence of PDZ domain-binding motif was sufficient for clustering of EGFP-VSNL protein in hippocampal neurons.
Targeting of GFP-LDC constructs in hippocampal neurons. A, Diagram of GFP-LDC2 and GFP-LDC5 fusion proteins. B-E, Subcellular localization of GFP-LDC2 (B), GFP-LDC2ΔC (C), GFP-LDC5 (D), and GFP-LDC5ΔC (E) expressed in mature hippocampal neurons (green) is compared with MAP2 (red) by confocal imaging. Scale bars, 40 μm.
To compare subcellular localization of L-type Ca2+ channels formed by CaV1.3a and CaV1.3b pore-forming subunits (Safa et al., 2001; Xu and Lipscombe, 2001) (Fig. 5A), we generated epitope-tagged HA-CaV1.3a, HA-CaV1.3a-ΔC, and HA-CaV1.3b expression constructs (Fig. 5B) (see Materials and Methods for details). In control Western blotting experiments, we determined that HA-CaV1.3a, HA-CaV1.3a-ΔC, and HA-CaV1.3b subunits expressed at similar levels when transfected into HEK293 cells (Fig. 5C). To confirm functional expression of generated constructs, we coinjected HA-CaV1.3, β3, and α2δ-1 cRNA into Xenopus oocytes and performed a series of two-electrode voltage-clamp experiments using 40 mm Ba2+ as a current carrier (see Materials and Methods for details). We found that HA-CaV1.3a and HA-CaV1.3a-ΔC supported currents of similar amplitude (Fig. 5D) and displayed similar current-voltage relationships (Fig. 5E). We did not observe a hallmark shift in CaV1.3 voltage dependence in our experiments (Koschak et al., 2001; Xu and Lipscombe, 2001), because our recordings have been performed using 40 mm Ba2+ as a current carrier, and high concentrations of divalent cations are known to obscure the position of CaV1.3 voltage dependence (Xu and Lipscombe, 2001). The currents supported by HA-CaV1.3b subunit were approximately twofold larger than the currents supported by HA-CaV1.3a and HA-CaV1.3a-ΔC subunits (Fig. 5D,E). These results are consistent with the previous characterization of currents supported by untagged CaV1.3a and CaV1.3b constructs when expressed in Xenopus oocytes (Xu and Lipscombe, 2001). Does association with Shank (Figs. 1, 2) affect CaV1.3a L-type Ca2+ channel gating? To address this question, in the next series of experiments, we coinjected HA-CaV1.3a, β3, and α2δ-1 cRNA into Xenopus oocytes with or without GFP-Shank1B cRNA and analyzed resulting currents by two-electrode voltage clamp. We found that neither the shape of the current waveform nor position of the current-voltage relationship of L-type Ca2+ channels formed by HA-CaV1.3a subunit was significantly altered in the presence of GFP-Shank1B (Fig. 5F,G). Smaller size of the currents in the presence of GFP-Shank1B (Fig. 5F,G) most likely reflects lower levels of Ca2+ channel subunit expression when GFP-Shank1B cRNA is cotranslated in the same Xenopus oocyte.
HA-CaV1.3 L-type Ca2+ channel expression constructs. A, Alternative splicing of rat CaV1.3 subunit (Safa et al., 2001; Xu and Lipscombe, 2001). The position of alternative splicing (Q1637) is indicated by an arrow. B, The diagram of HA-CaV1.3 targeting constructs. Locations of SH3 and PDZ domain-binding motifs in CaV1.3a sequence are indicated in A and B. C, Analysis of HA-CaV1.3 construct expression in HEK293 cells by Western blotting with anti-HA mAb. D, Representative current traces of L-type currents supported by HA-CaV1.3a, HA-CaV 1.3a-ΔC, and HA-CaV1.3b in Xenopus oocytes. The currents were evoked by 50 ms pulse to +10 mV from a -80 mV holding potential. E, Current-voltage relationship of currents supported by HA-CaV1.3a (filled circles), HA-CaV1.3a-ΔC (filled triangles), and HA-CaV1.3b (open circles) channels coexpressed in Xenopus oocytes with β3 and α2δ-1 subunits. F, Representative current traces of HA-CaV1.3a L-type channels expressed in Xenopus oocytes with and without GFP-Shank1B. The currents were evoked by 50 ms pulse to +10 mV from a -80 mV holding potential. G, Current-voltage relationship of currents supported by HA-CaV1.3a channels coexpressed in Xenopus oocytes with β3 and α2δ-1 subunits (filled circles) or with β3 and α2δ-1 subunits and GFP-Shank 1B. The peak currents inE and G are shown as mean ± SEM (n≥3 oocytes).
In the previous study (Maximov and Bezprozvanny, 2002), we used HA-CaV2.2 constructs to dissect structural determinants of N-type Ca2+ channel targeting in neurons. In the present study, we extended the same approach to analysis of CaV1.3 L-type Ca2+ channel neuronal targeting. We coexpressed HA-CaV1.3, β3 and α2δ-1 subunits in mature hippocampal neurons cultured at high density and evaluated subcellular localization of HA-CaV1.3 channels by immunostaining with anti-HA mAb and anti-MAP2 pAb. We found that recombinant HA-CaV1.3a channels were concentrated in discrete clusters associated with MAP2-positive processes (Fig. 6A). In contrast to HA-CaV1.3a, L-type channels formed by HA-CaV1.3a-ΔC and HA-CaV1.3b subunits have not formed distinct puncta in hippocampal neurons (Fig. 6B,C), suggesting an important role of C-terminal PDZ domain-binding motif for CaV1.3a L-type Ca2+ channel clustering.
Targeting of recombinant HA-CaV1.3 L-type channels in hippocampal neurons. A-C, Subcellular localization of recombinant HA-CaV1.3 L-type Ca2+ channels in hippocampal neurons. Hippocampal neurons were transfected with HA-CaV1.3/β3/α2δ-1 plasmid combination at 10 DIV and analyzed 72 h after transfection by HA mAb (red) and MAP2 pAb (green) immunostaining. Results for HA-CaV1.3a (A; 11 cells analyzed), HA-CaV1.3a-ΔC (B; 15 cells analyzed), and HA-CaV1.3b (C; 8 cells analyzed) are shown. Scale bars, 40 μm. Magnified regions and profile scans are shown below for each panel as indicated.
To further determine an importance of CaV1.3-Shank association for L-type Ca2+ channel targeting, in the next series of experiments we coexpressed HA-tagged CaV1.3 L-type Ca2+ channels (HA-CaV1.3, β3, and α2δ-1) with GFP-Shank1B (Sala et al., 2003) in mature rat hippocampal neurons. In these experiments, subcellular localization of HA-CaV1.3 subunits is determined by immunostaining with anti-HA mAb and compared with subcellular localization of GFP-Shank1B visualized by GFP imaging. We found that HA-CaV1.3a subunits were perfectly colocalized with GFP-Shank1B puncta in distal dendrites (Fig. 7A). In contrast to HA-CaV1.3a, HA-CaV1.3a-ΔC and HA-CaV1.3b subunits did not colocalize with GFP-Shank1B puncta (Fig. 7B,C).
Targeting of recombinant HA-CaV1.3 L-type Ca2+ channels and GFP-Shank in hippocampal neurons. A-C, Hippocampal neurons were cotransfected with HA-CaV1.3/β3/α2δ-1 plasmids and GFP-Shank1B at 9 DIV and analyzed 72 h after transfection by anti-HA mAb immunostaining (red) and GFP imaging (green). Results for HA-CaV1.3a (A; 11 cells analyzed), HA-CaV1.3a-ΔC (B; 9 cells analyzed), and HA-CaV1.3b (C; 8 cells analyzed) are shown. Scale bars, 40 μm. Magnified regions and profile scans are shown for each panel as indicated.
Targeting of GFP-Shank1B to postsynaptic spines in mature hippocampal neurons has been extensively documented previously (Sala et al., 2003), and HA-CaV1.3a/GFP-Shank1B clusters observed in our experiments (Fig. 7A) most likely correspond to synaptic locations. To further test this hypothesis, we stained hippocampal neurons transfected with HA-CaV1.3a and GFP-Shank1B using anti-HA mAb and anti-synapsin pAb and found that clusters of recombinant HA-CaV1.3a and GFP-Shank1B colocalize with endogenous synapsin clusters (Fig. 8A). Colocalization of HA-CaV1.3a, GFP-Shank1B, and synapsin clusters was further confirmed by profile scan analysis (Fig. 8B). These results are consistent with the previous studies of GFP-Shank1B targeting (Sala et al., 2003) and confirm that HA-CaV1.3a/GFP-Shank1B clusters are synaptic in origin.
Recombinant HA-CaV1.3a L-type Ca2+ channels and GFP-Shank cluster at synaptic locations in hippocampal neurons. A, Hippocampal neurons were cotransfected with HA-CaV1.3/β3/α2δ-1 plasmids and GFP-Shank1B at 9 DIV and analyzed 72 h after transfection by anti-HA mAb immunostaining (red) and GFP imaging (green) and anti-synapsin pAb staining (blue). Scale bar, 40 μm. The magnified region for each channel is shown as indicated. B, Profile scan of HA-CaV1.3a (red), GFP-Shank1B (green), and synapsin (blue) signal intensities in the magnified region. Similar results have been obtained with 18 cells analyzed.
pCREB signaling mediated by recombinant CaV1.3 L-type Ca2+ channels
Activation of L-type Ca2+ channels is efficiently coupled to phosphorylation of nuclear transcription factor CREB (Bito et al., 1996; Dolmetsch et al., 2001; Weick et al., 2003). Coupling of CaV1.2 L-type Ca2+ channels to pCREB has been demonstrated in experiments with dihydropyridine-resistant (DHPR) CaV1.2 L-type Ca2+ channels containing T1039Y mutation (He et al., 1997; Dolmetsch et al., 2001; Weick et al., 2003). To evaluate a contribution of CaV1.3 Ca2+ channels to pCREB signaling, we adapted the “pharmacological knock-out” strategy from Dolmetsch et al. (2001). Primary sequence alignment reveals a high degree of sequence conservation between CaV1.2 and CaV1.3 subunits in the region surrounding CaV1.2-T1039 residue (Fig. 9A). Thus, to obtain DHPR-CaV1.3 channel, we generated an analogous T1033Y mutation in the CaV1.3 sequence (Fig. 9A). To visualize DHPR-CaV1.3 channels in neurons, we subcloned a fragment containing T1033Y mutation to HA-CaV1.3 targeting constructs (Fig. 5B) and generated HA-CaV1.3a-T1033Y and HA-CaV1.3a-T1033Y-ΔC constructs. In control Xenopus oocyte expression experiments, we found that 100 μm nifedipine efficiently inhibited L-type currents supported by the HA-CaV1.3a subunit (Fig. 9B, top) but had a much smaller effect on the currents supported by HA-CaV1.3a-T1033Y and HA-CaV1.3a-T1033Y-ΔC subunits (Fig. 9B, middle, bottom). Thus, in agreement with the sequence alignment (Fig. 9A), CaV1.3-T1033Y mutant is functionally analogous to CaV1.2-T1039Y and indeed encodes DHPR-CaV1.3 L-type Ca2+ channel. These results are consistent with the recently reported description of DHPR-CaV1.3 double mutant containing T1033Y and Q1037M substitutions (Liu et al., 2003). In additional control experiments, we demonstrated that L-type channels supported by HA-CaV1.3a-T1033Y and HA-CaV1.3a-T1033Y-ΔC subunits display similar current-voltage relationship when expressed in Xenopus oocytes (Fig. 9C).
DHP-resistant HA-CaV1.3 L-type Ca2+ channels. A, Primary sequence alignment of DHP-binding site in CaV1.2 and CaV1.3 subunits. Positions of CaV1.2-T1039Y (He et al., 1997) and CaV1.3-T1033Y mutations are indicated. B, Representative current traces of L-type currents supported by HA-CaV1.3a (top), HA-CaV1.3a-T1033Y (middle), and HA-CaV1.3a-T1033Y-ΔC (bottom) in Xenopus oocytes. The currents were evoked by a 50 ms pulse to +10 mV from a -60 mV holding potential. The recordings from the same oocytes obtained before (black traces) and 3 min after (gray traces) application of 100 μm nifedipine are shown. C, Current-voltage relationship of currents supported by HA-CaV1.3a-T1033Y (open circles) and HA-CaV1.3a-T1033Y-ΔC (filled circles) channels coexpressed in Xenopus oocytes with β3 and α2δ-1 subunits. The peak currents are shown as mean ± SEM (n = 2 oocytes). D, Hippocampal neurons were transfected with HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC plasmids with α2δ-1 and β3 auxiliary subunits and pEGFP-C3 plasmid. Forty-eight hours after transfection, neurons were loaded with fura-2 and stimulated for 60 s with 20 mm KCl (20K) or 45 mm KCl (45K) in the presence of 1 μm TTX, 50 μm nifedipine, 50 μm d-AP-5, and 10 μm CNQX. An average peak 340/380 ratio in transfected neurons (GFP positive) is shown for CaV1.3a-T1033Y (filled circles; n ≥ 7) and HA-CaV1.3a-T1033Y-ΔC (filled triangles; n ≥ 5) constructs and for nontransfected (GFP negative) neurons (open circles; n ≥ 13) as a function of stimulation.
Functional expression of L-type channels supported by HA-CaV1.3a-T1033Y and HA-CaV1.3a-T1033Y-ΔC subunits in hippocampal neurons was further evaluated by fura-2 Ca2+ imaging. In these experiments, HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC subunits were cotransfected into mature hippocampal neurons together with β3 and α2δ-1 subunits and EGFP plasmid. Transfected neurons were identified by GFP imaging, and Ca2+-imaging experiments with fura-2 were performed as we described previously for striatal medium spiny neurons (Tang et al., 2003) in the presence of 50 μm nifedipine, 50 μm d-AP-5, and 10 μm CNQX to block Ca2+ influx via endogenous L-type Ca2+ channels, NMDA, and AMPA receptors. We found that 60 s stimulation by 20K or 45K depolarizations resulted in greater Ca2+ elevation in transfected (GFP positive) neurons than in untransfected (GFP negative) neurons (Fig. 9D). In either stimulation conditions, we have not observed significant difference in total Ca2+ influx supported by HA-CaV1.3a-T1033Y and HA-CaV1.3a-T1033Y-ΔC subunits (Fig. 9D).
To evaluate the contribution of CaV1.3 L-type Ca2+ channels to pCREB signaling, we coexpressed HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC constructs with β3 and α2δ-1 subunits in mature hippocampal neurons. Transfected neurons were stimulated for 90 s by KCl-induced depolarization (20K, 45K, or 90K in the continuous presence of 50 μm nifedipine, 50 μm d-AP-5, and 10 μm CNQX) and stained with pCREB-specific pAb (Fig. 10A). The transfected neurons were identified by immunostaining with anti-HA mAb (Fig. 10A). We discovered that 20K stimulation induced significant nuclear pCREB signals in neurons transfected with HA-CaV1.3a-T1033Y but not in neurons transfected with HA-CaV1.3a-T1033Y-ΔC (Fig. 10A). In contrast, 45K (Fig. 10A) and 90K (data not shown) induced similar nuclear pCREB responses in both groups of transfected cells. The pCREB signals observed in some HA-negative neurons (Fig. 10A) most likely result from the failure to identify low-expressing transfected cells by HA immunostaining and/or from incomplete block of endogenous L-type Ca2+ channels by 50 μm nifedipine.
pCREB signaling supported by recombinant HA-DHPR-CaV1.3 L-type Ca2+ channels. A, Hippocampal neurons were transfected with HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC plasmids with α2δ-1 and β3 auxiliary subunits as indicated. Forty-eight hours after transfection, neurons were stimulated for 90 s with 20 mm KCl (20K) or 45 mm KCl (45K) in the presence of 1 μm TTX, 50 μm nifedipine, 50 μm d-AP-5, and 10 μm CNQX. Confocal images of the same field are shown for pCREB (green) and HA (red) staining as indicated. B, C, An average stimulation-induced increase in normalized pCREB nuclear staining (ΔpCREB) in neurons transfected with HA-CaV1.3a-T1033Y (filled circles; n ≥ 11) or HA-CaV1.3a-T1033Y-ΔC (filled triangles; n ≥ 12) (HA positive) and in nontransfected (HA negative) neurons (open circles; n ≥ 17) is shown as a function of stimulus intensity for 90 s KCl depolarizations (B) and for 5 Hz EFS (C). pCREB staining in neurons transfected with HA-CaV1.3a-T1033Y is significantly higher (***p < 0.05) than in neurons transfected with HA-CaV1.3a-T1033Y-ΔC at 20K 90s (B) and 45 s or 180 s EFS (C) stimulations. No stim, No stimulation.
For quantitative evaluation of pCREB signaling responses, the intensity of nuclear pCREB staining in stimulated neurons was normalized to the intensity of nuclear pCREB staining in unstimulated neurons measured in the same experiment, and the normalized pCREB values from different experiments were averaged together. An average stimulation-induced increase in nuclear pCREB staining (ΔpCREB) is shown as a function of stimulation intensity for neurons transfected with HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC constructs and for untransfected (HA negative) neurons (Fig. 10B). It is apparent that at high intensities of stimulation (45K and 90K), both constructs efficiently supported signaling to pCREB, with >150% stimulation-induced increase in pCREB nuclear staining in transfected neurons (Fig. 10B). However, at low intensity of stimulation (20K), only HA-CaV1.3a-T1033Y construct efficiently supported CREB phosphorylation, whereas HA-CaV1.3a-T1033Y-ΔC construct was much less effective (Fig. 10B). To explore these differences further, we shortened the duration of KCl depolarization to 30 s and repeated the analysis of nuclear pCREB responses. We found that at 30 s stimulation, the HA-CaV1.3a-T1033Y construct was significantly more efficient in coupling to pCREB than HA-CaV1.3a-T1033Y-ΔC construct at 20K and 45K but not at 90K stimulation (data not shown).
To determine the importance of CaV1.3a C-terminal PDZ domain-binding motif for pCREB signaling induced by synaptic activity, the HA-CaV1.3a-T1033Y or HA-CaV1.3a-T1033Y-ΔC transfected neurons were challenged by EFS. These experiments were performed in the continuous presence of 50 μm nifedipine and 50 μm d-AP-5 to inhibit Ca2+ influx mediated by endogenous L-type Ca2+ channels and NMDAR. Neurons were stimulated by EFS at 5 Hz frequency for 45 or 180 s and stained with pCREB-specific pAb and anti-HA mAb. Obtained results were evaluated and quantified as described above for KCl depolarization experiments. We discovered that L-type Ca2+ channels formed by HA-CaV1.3a-T1033Y subunit efficiently supported pCREB responses induced by synaptic activity (Fig. 10C). In contrast, L-type Ca2+ channels formed by HA-CaV1.3a-T1033Y-ΔC subunit were much less effective at both 45 and 180 s EFS (Fig. 10C), indicating that association of CaV1.3a subunit with PDZ proteins plays an important role in L-type-mediated synaptic signaling to pCREB.
pCREB signaling mediated by endogenous L-type Ca2+ channels
In the next series of experiments, we evaluated an importance of endogenous CaV1.3a and CaV1.2 L-type Ca2+ channel association with PDZ proteins for pCREB signaling by using dominant-negative peptides. In these experiments, we took an advantage of recently developed protein delivery technology (Schwarze et al., 2000) and used R9 signal (Wender et al., 2000; Tang and Bezprozvanny, 2004) to deliver FITC-labeled R9-LCC and R9-LDC competitive peptides and R9 control peptide into hippocampal neurons. As an additional control for specificity of observed effects, we used R9-NR2B peptide containing class 1 PDZ domain-binding motif from the NR2B NMDAR subunit. As determined by FITC fluorescence, >90% of hippocampal neurons were loaded with R9 peptides in our experiments (Fig. 11A), similar to our previous results with striatal medium spiny neurons (Tang and Bezprozvanny, 2004). Three hours after R9 peptides loading, hippocampal neurons were stimulated for 90 s by KCl depolarization (20K or 45K) in the presence of 50 μm d-AP-5 and 10 μm CNQX to block Ca2+ influx via NMDA and AMPA receptors. Using fura-2 Ca2+ imaging, we determined that total KCl-induced Ca2+ influx was not significantly different in hippocampal neurons loaded with R9, R9-LDC, and R9-LCC peptides (Fig. 11B).
Effects of R9 competitive peptides on L-type Ca2+ channels mediated pCREB signaling. A, Loading of FITC-conjugated R9, R9-LCC, R9-LDC, and R9-NR2B peptides into mature hippocampal neurons was visualized by FITC fluorescence. B, Hippocampal neurons loaded with R9, R9-LCC, and R9-LDC peptides were stimulated 3 h after loading for 60 s by 20 mm KCl (20 K) or 45 mm KCl (45 K) in the presence of 1 μm TTX, 50 μm d-AP-5, and 10 μm CNQX. An average peak 340/380 fura-2 ratio is shown as a function of stimulus intensity for neurons loaded with R9 (filled circles; n = 18), R9-LCC (open circles; n ≥ 17), and R9-LDC (filled triangles; n ≥ 15). C, The hippocampal neurons loaded with R9, R9-LCC, R9-LDC, R9-LDC plus R9-LCC, and R9-NR2B peptides were stimulated 3 h after loading for 90 s by 20 mm KCl (20 K) or 45 mm KCl (45 K) in the presence of 1 μm TTX, 50 μmd-AP-5, and 10 μm CNQX and fixed and stained with anti-pCREB pAb. Confocal images of pCREB staining in stimulated and unstimulated neurons (no stim) are shown. D, E, An average stimulation-induced increase in normalized pCREB nuclear staining (ΔpCREB) is shown as a function of stimulus intensity for control neurons (open circles; n ≥ 70), for neurons loaded with R9 (open squares; n ≥ 118), R9-NR2B (filled squares; n ≥ 55), R9-LCC (open downward triangles; n ≥ 87), R9-LDC (filled circles; n ≥ 72), and R9-LCC plus R9-LDC mixture (open upward triangles;n ≥ 78). At 20K stimulation, the increase in pCREB staining in R9-loaded neurons is significantly higher (***p < 0.05) than in R9-LCC or R9-LDC-loaded neurons. At 45K stimulation, the increase in pCREB staining in R9-loaded neurons is significantly higher than in R9-LCC-loaded neurons but not in R9-LDC-loaded neurons.
The effects of R9 competitive peptides on nuclear pCREB signaling were quantified by immunostaining with pCREB-specific pAb as described above for recombinant DHPR-CaV1.3 L-type Ca2+ channels. We found that pCREB signaling pathway was efficiently activated in R9-loaded neurons by either 20K or 45K stimulation for 90 s (Fig. 11C). To compare results from different series of experiments, ΔpCREB signal at each experiment was normalized to an average ΔpCREB signal induced by application of 45K to control neurons (no R9 peptide added) in the same experiment. Quantitative analysis revealed that pCREB signals were not significantly different in R9-loaded and control neurons in 20K and 45K stimulation conditions (Fig. 11D). However, when compared with R9-loaded neurons, pCREB responses in R9-LCC-loaded neurons were suppressed in both stimulating conditions (Fig. 11C,E). For R9-LDC-loaded neurons, the pCREB stimulation was almost completely suppressed at 20K stimulation (Fig. 11C,E) but was restored to almost full response at 45K stimulation (Fig. 11C,E). When R9-LDC and R9-LCC peptides were combined, pCREB signals were completely suppressed at both 20K and 45K stimulation conditions (Fig. 11C,E). The observed effects were specific for CaV1.3a and CaV1.2 L-type Ca2+ channels C-terminal sequences, because R9-NR2B peptide had no effect on pCREB signaling in our experimental conditions (Fig. 11C,D).
Discussion
Interactions with modular adaptor proteins and synaptic targeting of Ca2+ channels
N-type (CaV2.2) and P/Q-type (CaV2.1) voltage-gated Ca2+ channels couple membrane depolarization with Ca2+ influx and synaptic vesicle fusion in the presynaptic terminal (Dunlap et al., 1995). In contrast, L-type (CaV1.2 and CaV1.3) Ca2+ channels couple membrane depolarization with Ca2+ influx and activation of pCREB and NF-ATc4 transcription factors in the nucleus of a postsynaptic neuron (Bito et al., 1996; Graef et al., 1999; Dolmetsch et al., 2001; Deisseroth et al., 2003; Weick et al., 2003). Because of the highly compartmentalized nature of neuronal Ca2+ signaling, proper subcellular localization of Ca2+ channels is likely to be of critical importance for their function in neurons. Our previous results (Maximov et al., 1999; Maximov and Bezprozvanny, 2002), results by Weick at al. (2003), and the data described in the present report suggest that association of C termini of Ca2+ channels pore-forming subunit with modular adaptor proteins plays a key role in their neuronal localization. In the previous study (Maximov et al., 1999), we described a C-terminal association of CaV2.2 (N-type) and CaV2.1 (P/Q-type) Ca2+ channel subunits with the first PDZ domain of presynaptic adaptor protein Mint1 and with the SH3 domain of adaptor protein calcium calmodulin-associated serine/kinase (CASK). In the present report, we describe C-terminal association of CaV1.3 (L-type) Ca2+ channel subunit with PDZ and SH3 domains of Shank postsynaptic adaptor protein.
It is interesting to compare CaV1.3 (L-type) association with Shank and CaV2.2 (N-type) association with Mint and CASK. The association of CaV2.2 subunit with Mint-PDZ1 domain is mediated by a novel PDZ domain-binding motif DHWC (Maximov et al., 1999; Bezprozvanny and Maximov, 2001). In contrast, the association of CaV1.3 subunit with Shank-PDZ domain is mediated by a “canonical” class 1 PDZ domain-binding consensus ITTL (Figs. 1, 2). Upstream of PDZ domain-binding motif, the C termini of both CaV2.2 and CaV1.3 subunits contain proline-rich (PXXP) region, which in the case of CaV2.2 binds to CASK-SH3 (Maximov et al., 1999) and in the case of CaV1.3 binds to Shank-SH3 (Fig. 2D,F). Interestingly, Mint1 and CASK bind to each other (Butz et al., 1998) and therefore, CaV2.2 C terminal binds to Mint1/CASK complex, whereas the CaV1.3 C terminal binds to adjacent PDZ and SH3 domains of Shank. Both CaV2.2 and CaV1.3 subunits are alternatively spliced at their C termini, resulting in long (CaV2.2a and CaV1.3a) and short (CaV2.2b and CaV1.3b) isoforms (Williams et al., 1992; Safa et al., 2001; Xu and Lipscombe, 2001). Long, but not short, isoforms of CaV2.2 and CaV1.3 subunits contain PDZ and SH3 domain-binding motifs (Maximov et al., 1999) (Fig. 5A). In targeting experiments with HA-CaV2.2 and HA-CaV1.3 subunits, we found that long (CaV2.2a and CaV1.3a), but not short (CaV2.2b and CaV1.3b), isoforms formed synaptic clusters when expressed in hippocampal neurons (Maximov and Bezprozvanny, 2002) (Figs. 6, 7, 8). In previous experiments, we demonstrated that the presence of either Mint-PDZ1 and CASK-SH3 binding motifs is sufficient to support synaptic clustering of HA-CaV2.2 in hippocampal neurons (Maximov and Bezprozvanny, 2002). Now, we found that the presence of both Shank-PDZ and Shank-SH3 domain-binding motifs is necessary for synaptic clustering of HA-CaV1.3 in hippocampal neurons (Figs. 4, 6, 7, 8).
Interactions with modular adaptor proteins and L-type Ca2+ channels signaling to pCREB
Our data indicate that association of CaV1.3 L-type Ca2+ channels with Shank is important for pCREB signaling. At low levels of KCl stimulation (20K for 90 s, 20K or 45K for 30 s) and at 5 Hz EFS, truncated CaV1.3a-ΔC construct was significantly less potent in supporting pCREB signaling than the full-length CaV1.3 construct (Fig. 10) (data not shown), although both constructs expressed in HEK293 cells and in Xenopus oocytes at similar levels (Figs. 5C-E, 9B,C) and supported similar Ca2+ influx in hippocampal neurons (Fig. 9D). Consistent with these findings, pCREB signaling was significantly suppressed in neurons loaded with R9-LDC peptide at low intensity of stimulation (20K for 90 s) (Fig. 11C,E) without affecting total Ca2+ influx (Fig. 11B). In contrast to results obtained at low levels of stimulation, CaV1.3a-ΔC and CaV1.3 constructs displayed a similar ability to support pCREB signaling at high levels of KCl stimulation (45K or 90K for 90 s, 90K for 30 s) (Fig. 10) (data not shown). Also, pCREB signaling was not significantly affected in the presence of R9-LDC peptide at 45K 90 s stimulation (Fig. 11C,E). These results are consistent with the idea that association of CaV1.3a L-type Ca2+ channels with Shank plays an important role in pCREB signaling at low levels of stimulation. At high-stimulation intensity, compartmentalization is lost, presumably because of larger Ca2+ flux via CaV1.3 L-type channels.
It is interesting to compare our findings regarding pCREB signaling supported by CaV1.3 L-type Ca2+ channels with the recent analysis of pCREB signaling supported by CaV1.2 L-type Ca2+ channels (Weick et al., 2003). Similar to our findings with CaV1.3, deletion of PDZ domain-binding motif attenuated CaV1.2-mediated pCREB signaling at low intensities of stimulation (20K for 3 min) (Weick et al., 2003). In the same report, it was demonstrated that expression of EGFP-VSNL dominant-negative construct attenuated pCREB signaling at low (20K for 3 min) and high (90K for 3 min) stimulation intensities (Weick et al., 2003). In agreement with these findings, loading of hippocampal neurons with R9-LCC peptide suppressed pCREB signaling in our experiments (Fig. 11C,E) without affecting total Ca2+ influx in the same stimulation conditions (Fig. 11B). Thus, for both L-type Ca2+ channel isoforms, association with PDZ domain proteins (NIL-16 for CaV1.2 and Shank for CaV1.3a) plays an important role in mediating efficient signaling to pCREB in hippocampal neurons. Consistent with this conclusion, simultaneous delivery of R9-LCC and R9-LDC competitive peptides uncoupled KCl depolarization from L-type Ca2+ channel-mediated pCREB responses (Fig. 11C,E). When compared with CaV1.2 L-type Ca2+ channels, CaV1.3 L-type Ca2+ channels open at more negative membrane potentials (Koschak et al., 2001; Xu and Lipscombe, 2001; Lipscombe et al., 2004; Olson et al., 2005) and therefore CaV1.3 L-type Ca2+ channels are more likely to play a predominant role in pCREB signaling at low levels of depolarization, when CaV1.3 association with Shank is of particular importance.
CaV1.3 L-type Ca2+ channels and macromolecular postsynaptic Ca2+ signaling complex
In conclusion, here, we show that the long splice variant of neuronal L-type Ca2+ channels pore-forming subunit CaV1.3a is specifically associated with Shank postsynaptic modular adaptor protein. CaV1.3a-Shank association is mediated via PDZ and SH3 domains of Shank and C-terminal region of CaV1.3a (Figs. 1, 2). Shank proteins (Fig. 1B) have been proposed to play a role of master scaffold in the postsynaptic specialization (Sheng and Kim, 2000). Via proline-rich region, Shank proteins bind to adaptor protein Homer (Tu et al., 1999), which in turn binds to InsP3R1 and mGluR1α/5 (Tu et al., 1998). Via PDZ domain, Shank is associated with GKAP-PSD95-NR2B (Boeckers et al., 1999; Naisbitt et al., 1999; Tu et al., 1999; Yao et al., 1999). SAM and PDZ domains allow Shank to multimerize (Sheng and Kim, 2000; Im et al., 2003), which may lead to formation of extended polymeric structures in postsynaptic specialization. Interactions listed above bring together a number of channels and signaling molecules involved in the control of intracellular Ca2+ concentration in postsynaptic spines of glutamatergic synapses (Kennedy, 2000). Our results indicate the following: (1) CaV1.3a L-type channels and Shank colocalize in postsynaptic locations in hippocampal neurons (Figs. 3, 7, 8); (2) Shank-binding motif in CaV1.3a sequence plays a critical role in synaptic clustering of L-type Ca2+ channels in hippocampal neurons (Figs. 4, 6, 7, 8); and (3) Shank-binding motif in CaV1.3a sequence is necessary for an efficient signaling from CaV1.3 L-type Ca2+ channels to pCREB in hippocampal neurons (Figs. 10, 11). Thus, results in our study demonstrate that CaV1.3 L-type Ca2+ channels is a part of macromolecular Ca2+ signaling complex organized by Shank in postsynaptic spines in hippocampal neurons. This conclusion is consistent with the emerging role of L-type voltage-gated Ca2+ channels in Ca2+ signaling in hippocampal postsynaptic spines (Sabatini and Svoboda, 2000; Hoogland and Saggau, 2004). In the accompanying study (Olson et al., 2005), we further examine an importance of CaV1.3-Shank association for Ca2+ signaling mediated by L-type Ca2+ channels in medium spiny striatal neurons.
Footnotes
This work was supported by the Robert A. Welch Foundation and National Institutes of Health Grants R01 NS39552 (I.B.), NS34696 (D.J.S.), and DA12958 (D.J.S.). We thank Yin Wang, Masaya Okamoto, and Shuza Sugita for advice with yeast two-hybrid screen; Thomas C. Südhof for the gift of rat brain cDNA library; Eunjoon Kim for anti-Shank antibodies; Susumu Seino, Diana Lipscombe, and Weifeng Xu for the CaV1.3 clones; Morgan Sheng and Carlo Sala for GFP-Shank1B plasmid; Richard W. Tsien for the CaV1.2, β3 and α2δ-1 clones; Haruhiko Bito and Paul Mermelstein for advice with pCREB experiments; Nan Wang, Xiaohong Shen, Tianhua Lei, and Linda Patterson for expert technical and administrative assistance; and Paul Mermelstein, Eunjoon Kim, and Alexander Petrenko for comments on this manuscript.
Correspondence should be addressed to Dr. Ilya Bezprozvanny, Department of Physiology, University of Texas Southwestern Medical Center at Dallas, Dallas, Texas 75390. E-mail: Ilya.Bezprozvanny{at}UTSouthwestern.edu.
Copyright © 2005 Society for Neuroscience 0270-6474/05/251037-13$15.00/0