Integrin Receptor Activation Triggers Converging Regulation of Cav1.2 Calcium Channels by c-Src and Protein Kinase A Pathways*

L-type, voltage-gated Ca2+ channels (CaL) play critical roles in brain and muscle cell excitability. Here we show that currents through heterologously expressed neuronal and smooth muscle CaL channel isoforms are acutely potentiated following α5β1 integrin activation. Only the α1C pore-forming channel subunit is critical for this process. Truncation and site-directed mutagenesis strategies reveal that regulation of Cav1.2 by α5β1 integrin requires phosphorylation of α1C C-terminal residues Ser1901 and Tyr2122. These sites are known to be phosphorylated by protein kinase A (PKA) and c-Src, respectively, and are conserved between rat neuronal (Cav1.2c) and smooth muscle (Cav1.2b) isoforms. Kinase assays are consistent with phosphorylation of these two residues by PKA and c-Src. Following α5β1 integrin activation, native CaL channels in rat arteriolar smooth muscle exhibit potentiation that is completely blocked by combined PKA and Src inhibition. Our results demonstrate that integrin-ECM interactions are a common mechanism for the acute regulation of CaL channels in brain and muscle. These findings are consistent with the growing recognition of the importance of integrin-channel interactions in cellular responses to injury and the acute control of synaptic and blood vessel function.

Voltage-gated calcium channels play critical roles in the regulation of calcium entry across the plasma membranes of excitable cells. L-type calcium channels (Ca L ), 5 which are highly expressed in brain and muscle, are heteromeric transmembrane proteins composed of a poreforming ␣ 1C (Cav1.2) subunit along with accessory ␤, ␣ 2 , ␦, and sometimes ␥ subunits (1,2). The ␣ 1C subunit contains four highly conserved repeat regions with 24 membrane-spanning domains, in addition to a variable length N terminus and relatively long, intracellular C terminus. The three ␣ 1C isoforms (neuronal, Cav1.2c; smooth muscle, Cav1.2b; cardiac, Cav1.2a) exhibit significant sequence differences in their N and C termini but all are regulated by intracellular kinases in ways that uniquely determine calcium entry and cell excitability.
The regulation of Ca L channels by serine-threonine kinases has been extensively investigated. PKG phosphorylates a conserved serine reside in the cytoplasmic I-II linker (3) of all three ␣ 1C isoforms, leading to inhibition of current. PKC phosphorylates N-terminal threonine residues in cardiac and smooth muscle isoforms (4 -6) leading in most cases to potentiation of current. PKA phosphorylates all three ␣ 1C isoforms at a conserved C-terminal serine (Ser 1901 in Cav1.2c; Ser 1928 in Cav1.2a), thereby mediating ␤-adrenergic potentiation of the calcium current in cardiac myocytes and neurons (7)(8)(9). PKA also regulates ␣ 1C in smooth muscle, but the functional consequences on calcium current are complicated by crossover activation of PKG, which is expressed at high levels in that tissue (10).
We recently demonstrated that Ca L currents in vascular smooth muscle (VSM) are acutely regulated by the integrin class of cell adhesion molecules. Ca L current is inhibited by ligands of ␣v␤3 integrin but potentiated by ligands of ␣5␤1 and ␣4␤1 integrins, including the abundant extracellular matrix (ECM) protein, fibronectin (11). ␣5␤1-mediated potentiation depends critically on the integrin-associated tyrosine kinases c-Src and focal adhesion kinase (12). This mechanism is consistent with reports that ␣ 1C isoforms are potentiated by the growth factors, insulin-like growth factor-1 and platelet-derived growth factor (13)(14)(15), and that integrin-Ca L channel interactions are required for neurotropin production (16) and acute control of synaptic function in the adult central nervous system (17)(18)(19)(20)(21)(22).
Here, we present evidence that activation of ␣5␤1 integrin acutely potentiates Ca L channels in rat brain and smooth muscle. The effect can be reproduced fully in heterologously expressed neuronal and smooth muscle Ca L isoforms and depends critically on phosphorylation of two specific C-terminal serine and tyrosine residues, indicating that integrin-ECM interactions are a common mechanism for acute regulation of Ca L channels in brain and muscle. mM): 110 CsCl, 20 tetraethylammonium chloride, 10 EGTA, 2 MgCl 2 , 10 HEPES, 1 CaCl 2 (pH 7.2 with CsOH). Perforated patch pipettes also contained 240 mg/ml amphotericin. Ba 2ϩ was used as the charge carrier to increase the size of the inward current, and to minimize calcium-dependent inactivation of current. In experiments with VSM or HEK-293 cells, the bath solution (20 Ba 2ϩ ) contained (in mM): 20 BaCl 2 , 124 choline chloride, 10 HEPES, 15 D-glucose (pH 7.4 with triethanolamine-OH). In experiments with MS/nDB neurons, the bath solution (2 Ba 2ϩ ) was similar except for containing 2 mM Ba 2ϩ .
Neuronal and VSM Cell Isolation-Basal forebrain neurons were isolated by microdissection of MS/nDB regions from coronal slices of rat (Sprague-Dawley) brain followed by incubation in trypsin and then trituration (23). Dissected segments of 1A and 2A arterioles from rat cremaster muscle were sequentially digested in low Ca 2ϩ physiological saline solutions containing papain and then collagenase/ elastase. The resulting fragments were rinsed with low Ca 2ϩ saline solution and gently triturated using a Pasteur pipette to release single, elongated VSM cells (12). All animal protocols conformed to the Public Health Service policy for the Humane Care and Use of Laboratory Animals and were approved by the respective university Animal Care Committees.
Heterologous Channel Expression-HEK-293 cells (tsA-201 line) were maintained at 30 -37°C in a 5% CO 2 incubator in supplemented Dulbecco's modified Eagle's medium, and transfected with Ca L isoforms using CaPO 4 (for electrophysiology) or Lipofectamine 2000 (for Western blotting). Rat neuronal ␣ 1C , ␤ 1b , and ␣ 2 -␦ DNA, subcloned into pcDNA3.1 vectors, were gifts from T. Snutch; rabbit SM ␣ 1C , ␤ 2a , and ␣ 2 -␦ 1 DNA were gifts from F. Hofmann and N. Klugbauer; human c-Src subcloned into SR␣ using the EcoRI site was a gift from D. Fujita. Enhanced green fluorescent protein (EGFP) was used as a co-transfectant in all electrophysiological protocols to identify successfully transfected cells. Only single cells were used for electrophysiological protocols, typically 48 -72 h after transfection.
Mutagenesis Methods-A PCR strategy was used to introduce stop codons at various positions in the Cav1.2 ␣ 1C subunit C terminus. The wild type Cav1.2 clone contained a NotI site at the 3Ј multicloning site. Thus, primer A was designed to anneal to the Cav1.2 antisense strand immediately upstream of the site intended for the stop codon. The primer included the stop codon and the sequence encoding a NotI restriction site. Primer B was designed to associate with the Cav1.2 residues upstream of both a unique SbfI restriction site and the placement of the intended stop codon. The resulting PCR product that extended from the SbfI site to the stop codon/NotI site was then sequenced, excised via SbfI and NotI, and subcloned into the original Cav1.2 construct via SbfI and NotI. Site-directed mutagenesis to eliminate phosphorylation consensus sites in the C terminus of Cav1.2 was conducted using the QuikChange site-directed mutagenesis kit (Stratagene), using the entire cDNA as a template. Subsequently, a fragment ranging from SbfI to NotI was sequenced to confirm the presence of the mutations, excised, and subcloned into the original WT Cav1.2 construct.
Immunoprecipitation, Immunoblotting, and in Vitro Kinase Assays-TsA-201 cells were lysed 48 -72 h after transfection in Tris-buffered saline solution containing 1% Triton X-100 with a mixture of protease inhibitors (4°C). Ca L channels were immunoprecipitated by overnight FIGURE 1. Potentiation of native neuronal and VSM cell Ca L currents by ␣5␤1 integrin activation. a, whole cell Ba 2ϩ current recordings from a rat MS/nDB neuron. ␣5␤1 integrin was activated by applying beads coated with anti-␣5 integrin Ab from a micropipette positioned close to the cell. Control trace was obtained 2 min after patch rupture; the ␣5-Ab trace was obtained 4 min after bead application. Current potentiation was 1.6-fold at test potential ϭ Ϫ10 mV. N-type and P/Q-type currents were blocked using 5 M -conotoxin MVIIC in the bath and V H ϭ Ϫ40 mV. The remaining current was blocked Ͼ95% by 1 M nifedipine (not shown). The horizontal line indicates zero current level. The lower panel shows current-voltage relationships obtained from the same cell using a ramp protocol (Ϫ100 to ϩ40 mV over 100 ms; V H ϭ Ϫ40 mV). Pipette, 110 Cs ϩ ; bath, 2 mM Ba 2ϩ . b, whole cell Ba 2ϩ current recordings from a rat arteriolar smooth muscle cell before and after ␣5␤1 integrin activation. Ba 2ϩ current potentiation was 2.7-fold 4 min after application of ␣5␤1 Ab beads. The lower panel shows current-voltage relationship from the same cell obtained using a ramp protocol (V H ϭ Ϫ80 mV). Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ .
incubation with ␣ 1C Ab (Chemicon) followed by addition of protein A/G beads, pelleting, washing, and resuspension in Laemmli sample buffer. For immunoblotting, lysates or immunoprecipitates (IPs) were separated by SDS-PAGE using 6% acrylamide gels, transferred to nitrocellulose, and probed with appropriate primary and secondary Ab before application of SuperSignal chemiluminescent reagent and exposure to x-ray film. c-Src or cAK-induced phosphorylation was carried out at 30°C in standard assay solution by incubation of IP complexes with 10 M [␥-32 P]ATP in the presence of purified recombinant human c-Src or cAK catalytic subunit purified from bovine heart, respectively (24). After separating the phosphorylation mixture by SDS-PAGE, the gels were exposed to x-ray film for 9 -36 h at Ϫ20°C; densitometry was performed using Scion software or, in some cases, a PhosphorImager. (Fig. 1a) were acutely dissociated from the medial septum/diagonal band nucleus (MS/nDB) of the rat. Ca L currents were measured using whole cell patch clamp techniques in the presence of 5 M -conotoxin MVIIC and a holding potential of Ϫ40 mV to block N and P/Q channels (25). ␣5␤1 integrins were activated by micropipette application of beads coated with anti-␣5 integrin Ab (␣5-Ab, a multivalent ␣5␤1 ligand). Bead attachment to the neuron whose current is shown in Fig. 1a resulted in 1.6-fold potentiation of Ca L current that peaked 3-5 min after bead application; potentiation averaged 1.43 Ϯ 0.11-fold in 4 cells. I-V relationships of whole cell currents before (control) and after attachment of ␣5 beads are illustrated at the bottom of Fig. 1a. Potentiation did not appear to be associated with a significant shift in voltage sensitivity of the channel.

Activation of ␣5␤1 Integrin in Brain and VSM Potentiates Ca L Current-Neurons
In rat VSM cells, the application of anti-␣5 integrin Ab on beads also resulted in time-dependent potentiation of the whole cell Ca L current. Peak potentiation of the Ca L current was 1.8-fold for the cell shown in Fig. 1b, whereas the average potentiation was 1.64 Ϯ 0.05fold in 5 cells. This effect was specific for integrin engagement because uncoated beads or beads coated with bovine serum albumin had no significant effect on current. In addition, a non-integrin binding Ab, in soluble form or coated onto beads, was without a significant effect on current (11).
Potentiation of Heterologously Expressed Cav1.2 by Integrin Activation-To explore the mechanism of current potentiation, Ca L subunits were subsequently expressed in tsA-201 cells, which constitutively express ␣5␤1 integrin and demonstrate ␣5␤1 integrin-dependent adhesion (not shown). Transient transfection with the neuronal ␣ 1C channel subunit (Cav1.2c) alone resulted in relatively small currents that were potentiated 1.90 Ϯ 0.13-fold by application of soluble anti-␣5␤1 integrin Ab (human; Fig. 2a, left). Co-expression of ␤ 1b and ␣ 2 -␦ subunits (Fig. 2a, right) along with Cav1.2c resulted in much larger basal currents that were also potentiated by soluble ␣5␤1 integrin Ab (2.07 Ϯ 0.15fold). We also used two additional methods known to cross-link integrins (26): 1) application of microbeads coated with ␣5␤1 Ab; 2) successive application of soluble ␣5␤1 Ab (mouse anti-human) and an appropriate secondary Ab (goat anti-mouse IgG). Both methods also led to potentiation of current, 1.85-2.07-fold (Fig. 2b), either with or without the accessory channel subunits. Integrin cross-linking and activation with multivalent ECM ligands, or with the appropriate concentration of monovalent Ab, is known to initiate outside-in signaling in several other cell systems (26,27).
Potentiation of Cav1. 2 Requires the Distal C Terminus of the ␣ 1C Subunit-Compared with smooth muscle and cardiac isoforms, the neuronal channel (Cav1.2c) has 27 fewer amino acid residues on the N terminus (28), including potential PKC phosphorylation sites (6,29). The comparable potentiation of neuronal and smooth muscle isoforms argues that the N terminus is not the regulatory site for ␣5␤1 integrinmediated potentiation.
Subsequently, a series of successive, C-terminal truncation mutants of Cav1.2c were created and their responses to ␣5␤1 integrin activation tested. As shown in Fig. 4, the truncation mutant Stop 5, which terminated proximal to the canonical PKA phosphorylation site (9) at Ser 1901 (equivalent to Ser 1928 on cardiac Cav1.2a) exhibited a loss of significant potentiation by ␣5␤1 integrin Ab. In contrast, truncations distal to the PKA phosphorylation site (Stop 6 -8), including those missing a proline-rich domain (30), resulted in a significantly reduced (ϳ50%), but not completely abolished, response to integrin activation, as compared with wild-type (WT) Cav1.2c. Each of the Stop 6 -8 constructs was missing the putative tyrosine phosphorylation site at Tyr 2122 previously identified as a regulatory site for insulin-like growth factor-1 (14). These results suggest that two different regulatory sites on the C terminus of Cav1.2 are required for full potentiation following ␣5␤1 integrin activation, with one of the sites being Ser 1901 (Cav1.2c). Consistent with this conclusion is the additional observation that inclusion of PKI in the patch pipette abolished the remaining potentiation associated with integrin activation in the Stop 6 -8 constructs (Fig. 4c).
Potentiation of Cav1.2c Depends Partially on ␣ 1C Tyr 2122 -Our previous work showed that potentiation of Ca L in rat smooth muscle was partially but incompletely blocked by tyrosine kinase inhibitors (12). Because only two tyrosine residues in rat Cav1.2c are distal to Ser 1901 , we constructed single-site mutants in which each of those residues was substituted with phenylalanine (Y2122F-␣ 1C and Y2139F-␣ 1C ). After expression, current through the Y2122F-␣ 1C construct continued to be potentiated by the ␣5␤1 integrin Ab, but potentiation was only 1.39 Ϯ 0.06-fold above basal current, compared with 1.90 Ϯ 0.13-fold for WT-␣ 1C (Fig. 5, a and b). In contrast, the magnitude of potentiation of Y2139-␣ 1C was statistically indistinguishable from that for WT-␣ 1C (1.86 Ϯ 0.09versus 1.90 Ϯ 0.13-fold, respectively). To further verify that Tyr 2122 was critical for full potentiation by ␣5␤1 integrin, we recorded current through WT-␣ 1C while dialyzing the cells with a decoy peptide containing the same amino acid sequence as rat WT-␣ 1C-c . This peptide was previously found to interfere with potentiation of native Ca L channels in smooth muscle (12), and here it reduced integrin-induced potentiation of WT-␣ 1C by ϳ50% (to 1.43 Ϯ 0.05-fold above basal). As controls, neither an identical peptide containing a Y2122F substitution, nor a scrambled peptide, had any effect on integrin-induced potentiation of WT-␣ 1C (Fig. 5b). Collectively, these results suggest that Tyr 2122 is required for a significant fraction of, but not complete, potentiation of ␣ 1C-c following ␣5␤1 integrin activation.
As an additional approach to testing whether the Tyr 2122 residue of ␣ 1C-c was phosphorylated following integrin activation, we immunoprecipitated the channel after its expression in HEK-293 cells and performed in vitro phosphorylation assays with [␥-32 P]ATP in the presence or absence of purified Src tyrosine kinase. Radiolabeled phosphate incorporation into Y2122F-␣ 1C was reduced to 39 Ϯ 6% of the level in WT-␣ 1C , whereas [␥-32 P]ATP incorporation into Y2139F-␣ 1C averaged 120 Ϯ 26% of that in WT-␣ 1C (Fig. 5c, right). Phosphorylation of WT-␣ 1C and Y2139F-␣ 1C could also be detected (albeit at much lower levels) in the absence of exogenous c-Src (Fig. 5c, left), but it was nearly undetectable in Y2122F-␣ 1C under comparable conditions. These results are consistent with the conclusion that phosphorylation of the Tyr 2122 residue in ␣ 1C-c by c-Src (or other tyrosine kinases) is part of the mechanism for integrin-induced potentiation of the Ca L current.
Potentiation of Cav1.2c Is Partly Mediated by c-Src-We further explored the role of c-Src in integrin-induced potentiation of Ca L by co-expressing various c-Src constructs along with WT-␣ 1C in HEK-293 cells (Fig. 6, a and b). Co-expression of WT-Src (human) with WT-␣ 1C resulted in a significantly enhanced potentiation of current following integrin activation (2.34 Ϯ 0.13-fold) as compared with the magnitude of potentiation when WT-␣ 1C was expressed alone (1.90 Ϯ 0.13-fold). In contrast, co-expression of a kinase-dead (Kd) form of c-Src (24), along with WT-␣ 1C , resulted in ϳ50% reduction of potentiation of current after integrin activation (to 1.43 Ϯ 0.05-fold). Although we attempted various additional methods of enhancing the amount of Kd-Src expression, we were unable to completely abolish the ␣5 integrin-induced potentiation of current (not shown). As an alternative approach to Kd-Src co-expression, we tested the effects of PP2 (100 nM), a membrane permeable inhibitor of Src family kinases, which reduced potentiation of WT-␣ 1C by 40% following integrin activation (1.58 Ϯ 0.06versus 1.90 Ϯ 0.13-fold). Collectively, these results are consistent with the conclusion that phosphorylation of the channel by c-Src accounts for some, but not all, of the potentiation following ␣5␤1 integrin activation.
When the channel was immunoprecipitated and probed with an antiphosphotyrosine Ab, co-expression of c-Src with ␣ 1C significantly enhanced tyrosine phosphorylation of ␣ 1C (by ϳ5-fold; Fig. 6c). However, when Kd-Src was co-expressed instead, tyrosine phosphorylation of ␣ 1C was undetectable under comparable conditions. Although tyrosine phosphorylation of ␣ 1C by endogenous Src (Fig. 6c, lane 1) is not apparent in Fig. 6, longer film exposure times revealed detectable phosphorylation of the channel in the absence of c-Src co-transfection (not shown). To directly test this hypothesis, we constructed an ␣ 1C mutant in which the PKA site was altered to alanine (S1901A-␣ 1C ) and a double mutant in which both the PKA and Src phosphorylation sites were altered (S1901A/Y2122F-␣ 1C ). We initially checked the mutants to verify that regulation by PKA was impaired in response to the membrane-permeable cAMP analog, 8-Br-cAMP (Fig. 7, a and b, top). 8-Br-cAMP (1 mM) potentiated the WT-␣ 1C current by 1.64 Ϯ 0.12-fold and this effect was completely abolished in both S1901A-␣ 1C and S1901A/Y2122F-␣ 1C constructs. No significant potentiation by 8-Br-cAMP was observed in cells transfected with WT-␣ 1C if PKI was included in the recording pipette.
We also tested whether the S1901A-␣ 1C mutant could be phosphorylated in vitro by the catalytic subunit of cAMP-dependent protein kinase (cAK). Compared with phosphorylation of WT-␣ 1C , phospho-  6 construct was truncated at residue 1929, just distal to the PKA site. The Stop 7 construct was truncated at residue 1999, just distal to two proline-rich domains (P1 and P2, spanning residues 1948 -1973). The Stop 8 construct was truncated at residue 2066, just proximal to two tyrosine residues near the end of the C terminus. Potentiation was significantly reduced but not abolished in all Stop 6 -8 constructs; however, PKI peptide pre-loaded into the recording pipette abolished the remaining potentiation in each case (e.g. 5th trace). The numbering schemes for both rat neuronal (␣ 1C-c ) and rat smooth muscle (␣ 1C-b ) isoforms are shown for reference (EF and DCT-i refer to an EF-hand motif and distal C terminus inhibitory domain, respectively). c, summary data for the constructs shown in comparison to basal current (prior to ␣5␤1 Ab application) and potentiation of WT ␣ 1C-c (3-5 min after ␣5␤1 Ab application). Left graph represents cells transfected with various ␣ 1C-c constructs alone; right graph represents cells transfected with the same ␣ 1C-c constructs plus wild type ␤ 1b and ␣ 2 -␦ subunits. Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ ; voltage step same as in upper left trace of panel a. *, significant difference from basal current, p Ͻ 0.05; #, significant difference from WT-␣ 1C ϩ ␣5␤1 Ab, p Ͻ 0.05. rylation of immunoprecipitated S1901A-␣ 1C by cAK was reduced to 65 Ϯ 8% (Fig. 7d). In contrast, the phosphorylation of neither Y2122F-␣ 1C nor Y2139F-␣ 1C by cAK was significantly different from that for WT-␣ 1C (88 Ϯ 14 and 108 Ϯ 11%, respectively; Fig. 7d). The remaining, background [␥-32 P]ATP incorporation into S1901A-␣ 1C likely represents phosphorylation of the channel at additional sites containing cAK/ cGK consensus sequences (3,9) that may become modified under the in vitro assay conditions. Finally, we examined the effects of ␣5␤1 integrin activation on whole cell current recorded from cells expressing channels mutated at the PKA regulatory site. Potentiation of the S1901A-␣ 1C construct following integrin activation was reduced by ϳ45% (1.90 Ϯ 0.13versus 1.47 Ϯ 0.13-fold) compared with the potentiation of WT-␣ 1C ; however, currents were still significantly higher than basal. In contrast, in cells expressing S1901A/Y2122F-␣ 1C , integrin activation did not result in any significant potentiation of current above basal levels (Fig. 7b, bottom, and c). To further test this idea, we tested five different methods of interfering with phosphorylation of ␣ 1C at Ser 1901 and Tyr 2122 : S1901A-␣ 1C plus PP2 added to the bath; S1901A-␣ 1C plus co-expression of Kd-Src; WT-␣ 1C co-expressed with Kd-Src plus PKI in the recording pipette; Y2122F-␣ 1C plus PKI in the recording pipette; and the S1901A/ Y2122F-␣ 1C double mutant. In all five cases, no significant potentiation of current was observed following integrin activation (Fig. 7c). These results strongly support the conclusion that potentiation of the Cav1.2c current by ␣5␤1 integrin activation requires dual phosphorylation of sites Ser 1901 and Tyr 2122 on ␣ 1C-c .
Potentiation of Native Ca L Channels Requires Both PKA and c-Src-The evidence for dual phosphorylation of ␣ 1C-c by PKA and c-Src underlying integrin-dependent potentiation of the Ca L current prompted a more careful examination of the magnitude and time course of native Ca L current potentiation in rat cells during inhibition of Src and PKA. Fig. 8a shows the time course of current potentiation following application of beads coated with ␣5-integrin Ab onto single vascular myocytes. Consistent with previous findings, potentiation peaked 3-4 min Potentiation was reduced by ϳ50% in Y2122F-␣ 1C-c but not significantly in Y2139F-␣ 1C-c . Potentiation was also reduced in cells transfected with WT-␣ 1C-c when an 11-amino acid decoy peptide (EDESCVYALGR; trace 4), but not a control peptide (EDESCVFALGR; trace 5), was included in the recording pipette (50 M for all). Pipette, 110 Cs ϩ ; bath, 20 Ba 2ϩ . b, summary data for peak current potentiation by ␣5␤1 integrin Ab in the various ␣ 1C-c constructs described in panel a, plus an additional control in which the peptide sequence was scrambled (LVCGAEYRDSE). c, in vitro phosphorylation of expressed WT or mutant ␣ 1C channels containing Tyr to Phe mutations. Following co-expression in HEK cells of ␣ 1C , ␤ 1b , and ␣ 2 -␦ subunits, the cells were lysed and the channels were IP using an ␣ 1C Ab and incubated with [␥-32 P]ATP (upper bands) at 37°C for 30 min in the presence or absence of purified pp60 c-Src . Proteins were separated using SDS-PAGE and transferred to nitrocellulose. Lower bands (Blot: ␣ 1C ) show the amount of immunoprecipitated ␣ 1C subunit detected in each sample after stripping and reprobing with ␣ 1C Ab. Low levels of [ 32 P]ATP incorporation into the channel were detectable in the absence of c-Src (left lanes) but the signal increased dramatically in the presence of c-Src (right lanes). Relative to control (WT-␣ 1C-c ϩ Src lane), phosphorylation of the Y2122F-␣ 1C-c and Y2139F-␣ 1C-c mutant channels was 39 Ϯ 6 and 120 Ϯ 26%, respectively (n ϭ 3). In the two lanes denoted WT-␣ 1C glu-glu, IPs from WT-␣ 1C -transfected cells were performed using a control Ab versus an irrelevant Glu-Glu epitope tag. Although this Ab did not IP any detectable ␣ 1C subunit, a nonspecific, ϳ200-kDa phosphoprotein was observed in the presence of [␥-32 P]ATP ϩ c-Src. Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ ; voltage step same as in upper left trace of panel a. *, significant difference from basal current, p Ͻ 0.05; #, significant difference from WT-␣ 1C ϩ ␣5␤1 Ab, p Ͻ 0.05. after bead application and then spontaneously declined (11) toward control levels. When Src was inhibited just before peak potentiation by PP2 (100 nM in the bath) or by a Src SH2 inhibitory peptide (2.7 M in the recording pipette), 75 and 60% less current potentiation was observed, respectively, compared with that recorded in the absence of inhibitors (Fig. 8b). PKI alone, applied by exchange of the recording pipette solution, resulted in only 17% reduction of integrin-potentiated current. PP3 (100 nM), the inactive analog of PP2, was without significant effect (not shown) on either basal or integrin-potentiated current (12). However, the combination of PP2 and PKI peptide completely blocked integrin-induced potentiation of current (Fig. 8b). These results are consistent with those from heterologously expressed channels (Figs. 4 -7) and suggest that the same dual-phosphorylation mechanisms are operating to regulate native Ca L channels.

DISCUSSION
Our results demonstrate that neuronal and smooth muscle Ca L channels are regulated by ␣5␤1 integrin through an intracellular signaling pathway involving phosphorylation of the ␣ 1C channel subunit by PKA and c-Src. Integrin activation produces up to 2-fold potentiation of Ca 2ϩ current over a relatively short time frame (1-5 min). Given the central role of Ca L channels in nerve and smooth muscle excitability, the ECM-integrin-Ca L signaling pathway is likely to have major effects on both short-and long-term control of [Ca 2ϩ ] i and Ca 2ϩ -dependent processes in these respective cell types.
Ion Channel Regulation by Integrins-A growing body of literature supports our conclusion that ion channels are acutely regulated by integrin-ECM interactions (31). ␤1 integrins are known to modulate excitatory synaptic transmission through an as yet defined role of c-Src (16,22,32), observations that are consistent with the well known regulation of ligand-gated channels by tyrosine kinases (33). Recently, ␤1 integrins have also been implicated in the regulation of several different K ϩ channels, with evidence for both physical (34,35) and functional (36 -38) associations between the channels and integrins.
Several independent lines of evidence support the idea that integrins also regulate Ca L channels. For example, bidirectional regulation of native rat smooth muscle Ca L channels is observed following engagement of ␣5␤1, ␣4␤1, and ␣v␤3 integrins with ECM proteins or integrin antibodies (11,39). In the case of ␣5␤1 integrin, selective activation by multivalent integrin ligands results in ϳ70% potentiation of Ca L current FIGURE 6. Role of c-Src in integrin-induced potentiation of Cav1.2c. a, representative traces showing potentiation of Ba 2ϩ current following ␣5␤1 integrin activation in cells transfected with ␤ 1b , ␣ 2 -␦ subunits plus WT-␣ 1C-c alone, or WT-␣ 1C-c co-expressed with c-Src or kinase-dead (Kd) c-Src. Potentiation by ␣5␤1 integrin activation was modestly increased (from 1.90 to 2.34-fold) with co-expression of c-Src, but reduced (from 1.90 to 1.43-fold) in the presence of Kd-Src, which presumably competed with endogenous c-Src. PP2 (100 nM) treatment resembled the effects of Kd-Src expression. b, summary data showing peak current potentiation by the treatments described in a. c, following IP from HEK-293 cells, a Western blot of the WT-␣ 1C subunit using an anti-phosphotyrosine (P-Tyr) Ab (4G10) revealed very little Tyr phosphorylation of channel expressed alone, whereas Tyr phosphorylation of the ␣ 1C subunit was dramatically increased in cells co-expressing WT c-Src, but not Kd c-Src. The ϳ60-kDa phosphotyrosinecontaining protein detected in immunoprecipitates from co-transfected cells likely represents c-Src, and the decreased intensity of this band for Kd-Src expressing cells could reflect the lack of autophosphorylation of Src. The lowest bands (Blot: ␣ 1C ) show the amount of ␣ 1C protein recovered in each IP sample; detection was performed after stripping and reprobing with ␣ 1C Ab. Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ ; voltage step same as in the upper left trace of panel a. *, significant difference from basal current, p Ͻ 0.05; #, significant difference from WT-␣ 1C ϩ ␣5␤1 Ab, p Ͻ 0.05. above basal levels and this is substantially reduced by inhibition of focal adhesion kinase and c-Src tyrosine kinases (12). In Lymnaea, soluble fibronectin potentiates neuronal HVA current, enhances the [Ca 2ϩ ] i response to KCl, and leads to an increase in neuronal excitability (40). The specific Lymnaea integrins responsible have not yet been identified but cRGD peptide, which interacts with multiple integrins, produces dose-dependent, biphasic changes in HVA current (composed of both N-and L-type channels), suggesting that the responses are mediated by multiple integrins and/or multiple Ca 2ϩ channel types. The regulation of mammalian cardiac Ca L channels by muscarinic and adrenergic receptors is substantially influenced by the adhesion of cardiac myocytes to their substratum. Specifically, Ca L responses to ␤2 adrenergic agonists are potentiated in cells adhered to laminin or ␤1 integrin Ab (compared with adhesion on glass), whereas responses to acetylcholine are attenuated (41,42). However, another study suggests that adhesion-dependent potentiation of cardiac Ca L is not specific to integrins (43). In cardiomyocytes derived from embryonic stem cells, muscarinic inhibition of Ca L current is absent in ␤1-integrin Ϫ/Ϫ cells (44), although the effect is at least partly due to defective coupling of G␣ i subsequent to ␤1 integrin knockout. In this context it is interesting that G␣ 1 co-localizes extensively with the focal adhesion proteins talin and vinculin (45) known to be critically involved in integrin-mediated signaling (46). In E63 skeletal muscle cells, engagement of ␣7␤1 integrin by laminin or ␣7-integrin Ab triggers both Ca 2ϩ release and influx (27). Interestingly, the extracellular domain of ␣7␤1 integrin associates with the dihydropyridine receptor/channel (␣ 1S ), which shares substantial sequence homology with Cav1.2. Collectively, these findings support our contention that the pathways elucidated in the present study may be common mechanisms for regulation of Ca 2ϩ entry in many nerve and muscle cells.
Mechanism of Ca L Regulation by ␣5␤1 Integrin-Clues to the regulation of Ca L by integrins may lie in the structure of the ␣ 1C C terminus. Our finding of comparable integrin-mediated potentiation of both smooth muscle and neuron channels suggest that the N terminus, which is 27 amino acids shorter in the neuronal isoform (6,28), is not critical for this regulation. The ␣ 1C C terminus is composed of ϳ550 amino acid residues known to regulate channel function in a highly complex manner. The C terminus contains domains that mediate Ca 2ϩ -induced inactivation, calmodulin binding, and interactions with the ␤ subunit (47)(48)(49)(50). The terminal ϳ150 amino acids (2021-2171 in Cav1.2a) contain an inhibitory domain that interacts internally with more proximal (1733-1900) residues (51). ␣ 1C-a constructs missing substantial portions of the C FIGURE 7. Integrin-mediated potentiation of Cav1.2 is prevented by blocking channel phosphorylation at two C-terminal sites. a and b, application of the membranepermeable PKA activator 8-Br-cAMP (1 mM) from a picospritzer pipette produced 1.64 Ϯ 0.12-fold potentiation of WT-␣ 1C-c expressed with ␤ 1b and ␣ 2 -␦ subunits. Potentiation was completely blocked in the S1901A-␣ 1C-c mutant (second trace, top) and in the S1901A/Y2122F-␣ 1C-c double mutant (third trace, top). Potentiation was prevented in WT-␣ 1C-c with PKI peptide (2.7 M) in the recording pipette. c, summary data showing effects of S1901A and S1901A/Y2122F mutations on current potentiation in response to ␣5␤1 integrin activation. About 50% potentiation by ␣5␤1 Ab remained in S1901A-␣ 1C-c (panel b, bottom left trace): significantly less than in WT-␣ 1C-c but significantly greater than basal current. PKI peptide also blocked ϳ50% of ␣5␤1 Ab-mediated potentiation of WT-␣ 1C-c current. In contrast, five different methods of blocking phosphorylation of both Ser 1901 and Tyr 2122 sites, including the S1901A/Y2122F-␣ 1C-c double mutant (panel b, bottom right trace) resulted in no significant current potentiation following ␣5␤1 integrin activation. d, in vitro kinase assays comparing phosphorylation of immunoprecipitated WT and mutant channels by the purified catalytic subunit of cAK in the presence of [␥-32 P]ATP. Phosphorylation of neither mutant (Y2122F or Y2139F) was significantly different than for WT-␣ 1C-c (88 Ϯ 14 and 108 Ϯ 11%, respectively; n ϭ 3); however, phosphorylation of S1901-␣ 1C-c was reduced to 65 Ϯ 8% (n ϭ 3) of that for WT-␣ 1C-c . Glu-Glu Ab did not immunoprecipitate the channel and served as a control. Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ ; voltage step same as in upper left trace of panel b. *, significant difference from basal current, p Ͻ 0.05; #, significant difference from WT-␣ 1C ϩ 8-Br-cAMP (panel a) or from WT-␣ 1C ϩ ␣5␤1 Ab (panel b), p Ͻ 0.05.
terminus exhibit up to a 10-fold increase in basal current (30,51,52), which we observed to a lesser extent with our ␣ 1C-c truncation mutants. Thus, it is possible that potentiation of Ca L following phosphorylation of the C terminus is mediated by partial relief of tonic inhibition.
A number of studies present evidence that Cav1.2 is assembled in and regulated by a multiprotein complex at the plasma membrane. Neuronal Cav1.2 co-localizes with PKA, PP2A, and adenylate cyclase, along with the ␤ 2 -adrenergic receptor and at least one heterodimeric G-protein in what appears to be a signaling complex that might facilitate rapid and specific [Ca 2ϩ ] i signaling (53). This complex is assembled on a scaffold of membrane-bound proteins including AKAP 79 (54,55), AKAP 15 (56), MAP2B (57), and possibly cytoskeletal proteins such as ␣-actinin (58) and actin (59). The Cav1.2a II-II linker region and C terminus both contain proline-rich, class I and class II, Src homology 3 (SH3) domains that mediate binding of c-Src and Src family kinases to the cardiac Ca L channel (60). Residues containing the C-terminal SH3 domain (1974 -2000) are important for tethering ␣ 1C-a to the plasma membrane (30) and the ␤2 channel subunit (48). Importantly, both cardiac and smooth muscle channel ␣ 1C isoforms have been shown to physically associate with and be phosphorylated by c-Src (13,30,51,60,61), observations that are consistent with our present in vitro phosphorylation studies (Figs. 5 and 6). Likewise, the growth factor PDGF is also known to potentiate native rat smooth muscle channels through an unknown mechanism involving c-Src (13,62).
Collectively, this evidence points to the possibility that Ca L channels co-localize with integrins and/or focal adhesion proteins in excitable cells and that Ca L -integrin-focal adhesion signaling complexes potentially overlap with Ca L -PKA-AKAP complexes. The activation of PKA and subsequent phosphorylation of Cav1.2 channels following integrin activation observed in our study are consistent with activation of the cAMP signaling cascade in endothelial cells following application of mechanical stress through RGD-containing integrins (63); indeed, the activation of the cAMP cascade required both integrin receptor occupation and intact G␣ signaling. Thus, it is possible that scaffolding proteins such as WAVE1, which associate with both PKA and integrin-linked tyrosine kinases (64), will be found to play critical roles in the regulation of Ca L channels. The possible crossover between these signaling pathways will be an important issue to resolve with respect to the regulation of Ca L in excitable cells.
It is interesting that the heterologously expressed rat neuronal isoform (Cav1.2c) showed a similar degree of potentiation as native rat neuronal and VSM Ca L channels following ␣5␤1 integrin activation (compare Fig. 1 with Figs. 2 and 3). Similar physiological responses might be predicted based on the fact that there is Ͼ99% sequence homology between the rat neuronal (Cav1.2c) and rat smooth muscle (Cav1.2b) isoforms. For other species, there is less (ϳ80%) homology in the regulatory portion (distal 2/3) of the C terminus. Notably, rabbit Cav1.2b lacks the critical tyrosine residue corresponding to Tyr 2122 on Cav1.2c that mediates ϳ50% of integrin-induced potentiation and that has also been implicated in regulation of rat Cav1.2c by the growth factor IGF-1 (14). However, rabbit Cav1.2b is also potentiated to a comparable degree by ␣5␤1 integrin activation (Fig. 3). Whether potentiation of rabbit Cav1.2b depends both on phosphorylation by PKA at Ser 1928 and c-Src at an alternate tyrosine residue are important questions that need to be answered. However, it is also known that rabbit smooth muscle Ca L channels may be regulated by c-Src through more indirect mechanisms (4,13,65).
ECM-Integrin-Cav1.2 Signaling Axis in Cellular Responses to Injury-The regulation of Ca L channels by integrin-ECM interactions can potentially play a role in a number of important physiological processes in both the central nervous system and the cardiovascular system. In the brain, integrins are concentrated at sites of synaptic contact (66) and are critical for the formation, maturation, and maintenance of synaptic structure (67). For example, laminin, along with its receptor, ␣7␤1 integrin, plays a role in the clustering of acetylcholine receptors at the neuromuscular junction (68). Integrin engagement is required under some conditions for neurotrophic signaling (16) and integrin clustering can initiate receptor tyrosine kinase signaling in the absence of neurotrophins, as has been documented for other growth factors in other cell systems (69,70). In this context, integrin-Ca L channel interactions . When ␣5-bead application was followed by introduction into the patch pipette of a peptide (10 M) that selectively inhibits Src binding through its SH2 domain, beginning at 6 min, there was a significant reversal of potentiation (closed squares). b, summary of data from protocols as shown in panel a. Currents are normalized to unstimulated basal current (i.e. open circles in a) or to the degree of current potentiation by ␣5-beads alone (i.e. open squares in a). Src SH2 peptide alone (in pipette) inhibited ϳ60% of potentiation following integrin activation by ␣5-Ab on beads; PP2 (in bath) inhibited ϳ75% of potentiation; PKI peptide (in pipette) inhibited potentiation by 17%. In contrast, no significant potentiation remained when the combination of PKI peptide (in pipette) and PP2 (in bath) was used. All recordings were made in the conventional whole cell mode to allow peptide access and therefore exhibited current run-up and run-down. Pipette, 110 Cs ϩ ; bath, 20 mM Ba 2ϩ ; V H ϭ Ϫ80 mV; test potential ϭ ϩ30 mV. *, significant difference from control, p Ͻ 0.05. appear to be critical for the up-regulation of brain-derived neurotrophic factor mRNA in hippocampal neurons (16). Finally, a number of studies have now confirmed that integrins are instrumental in modulating long term potentiation and therefore in controlling synaptic function in the adult central nervous system (17)(18)(19)(20)(21)(22). However, whether their role involves changes in cellular adhesion or modulation of intracellular signaling has not yet been defined.
Integrin-Ca L channel interactions could also play a major role in the responses of neurons and blood vessels to injury and repair. Both neuronal activity and vascular reactivity are modulated by integrin signaling through the generation or exposure of new integrin ligands from limited degradation of extracellular matrix and/or turnover of new integrins (71)(72)(73). For example, successful axonal regeneration is highly correlated with the induction of integrins on the surface of peripheral neurons; therefore, peptides derived from ECM proteins have the potential to act as therapeutic agents for neuronal regeneration (74). In blood vessels, it is well established that VSM sensitivity to agonists and mechanical forces is altered following ischemic injury and during hypertrophic and eutrophic remodeling (75,76). Many factors contribute to these processes, but among them are changes in the number and composition of VSM integrins, changes in ECM components, and changes in the density of Ca L channels. Thus the interactions between Ca L channels and integrins are likely to play a major role in the ability of neuronal and vascular cells to recover from tissue injury.