Overexpression of beta2-adrenergic receptors cAMP-dependent protein kinase phosphorylates and modulates slow delayed rectifier potassium channels expressed in murine heart: evidence for receptor/channel co-localization.

The cardiac slow delayed rectifier potassium channel (IKs), comprised of (KCNQ1) and beta (KCNE1) subunits, is regulated by sympathetic nervous stimulation, with activation of beta-adrenergic receptors PKA phosphorylating IKs channels. We examined the effects of 2-adrenergic receptors (beta2-AR) on IKs in cardiac ventricular myocytes from transgenic mice expressing fusion proteins of IKs subunits and hbeta2-ARs. KCNQ1 and beta2-ARs were localized to the same subcellular regions, sharing intimate localization within nanometers of each other. In IKs/B2-AR myocytes, IKs density was increased, and activation shifted in the hyperpolarizing direction; IKs was not further modulated by exposure to isoproterenol, and KCNQ1 was found to be PKA-phosphorylated. Conversely, beta2-AR overexpression did not affect L-type calcium channel current (ICaL) under basal conditions with ICaL remaining responsive to cAMP. These data indicate intimate association of KCNQ1 and beta2-ARs and that beta2-AR signaling can modulate the function of IKs channels under conditions of increased beta2-AR expression, even in the absence of exogenous beta-AR agonist.

I Ks , the slowly activating component of the human cardiac delayed rectifier K ϩ current, is a major contributor to repolarization of the cardiac action potential (1). The I Ks channel results from the co-assembly of two subunits KCNQ1 (Kv-LQT1) and KCNE1 (minK), (2,3). KCNQ1, the ␣-subunit of I Ks , shares topological homology with other voltage-gated K ϩ channels in that its 676 amino acids encode six transmembrane domains and a pore-forming region. KCNE1, the ␤ subunit of I Ks , encodes a protein containing 129 -130 amino acids consisting of a single transmembrane spanning domain (4). The contribution of I Ks to regulation of action potential duration is augmented by the sympathetic nervous system via activation of ␤-adrenergic receptors (␤-AR) 1 (5,6). Sympathetic nervous sys-tem-mediated control of cardiac function results from binding of catecholamines to G protein-coupled receptors, including the ␤ 2 -AR, resulting in activation of heterotrimeric G-proteins G ␣ and G ␤␥ and stimulation of adenylate cyclase leading to increased production of cAMP. This increased production of cAMP promotes activation of cAMP-dependent protein kinase (PKA), which in turn activates downstream effector molecules including L-type Ca 2ϩ channels, sarcoplasmic reticulum Ca 2ϩ release channels (or ryanodine receptors (RyR)) and delayed rectifier potassium channels. Sympathetic nervous system regulation of the delayed rectifier potassium channel is most marked for the slow component (I Ks ) rather than the fast component (I Kr ) (7).
Compartmentalization of the response to cAMP is achieved in part by channel-bound A kinase-anchoring proteins (8). In the case of the I Ks channel, a leucine zipper motif in the KCNQ1 C-terminal domain coordinates the binding of the A kinase-anchoring protein Yotiao (9), which in turn binds to and recruits PKA and PP1 to the channel to form a channel microsignaling domain to regulate the phosphorylation of Ser 27 in the N terminus of KCNQ1 (9,10). Compartmentalization of the response to cAMP may also be due in part to receptor-channel co-localization. Here we have examined the subcellular localization of KCNQ1 and ␤ 2 -AR in ventricular myocytes isolated from transgenic mice expressing an hKCNQ1-hKCNE1 fusion protein TG ϩ and transgenic mice expressing hKCNQ1-hKCNE1 and overexpressing ␤ 2 -AR DTG ϩ . To examine the effects of increased ␤ 2 -ARs upon I Ks , we have made use of a transgenic mouse overexpressing the ␤ 2 -AR. These transgenic mice have Ͼ100-fold increase in ␤ 2 -AR density accompanied by apparent maximal heart rate and cardiac contractility (11). Overexpression of ␤ 2 -AR has been shown to result in agonistindependent spontaneous activation (12). ␤ 2 -AR signaling has also been shown to be localized to the cell membrane, with regards to L-type Ca 2ϩ channel modulation (13). We find the localization of KCNQ1 and ␤ 2 -AR to share similar subcellular localizations and to be within a few nanometers of each other. Even in the absence of exogenous agonist, functional analysis of hKCNQ1-hKCNE1 channel activity (I Ks ) in cardiac myocytes isolated from DTG ϩ mice revealed maximal adrenergic upregulation of I Ks , and biochemical analysis indicated that the channel was PKA-phosphorylated in DTG ϩ mice. This PKA-dependent modulation of the I Ks channel was not due to global increases in intracellular cAMP by receptor expression because we found that basal L-type calcium channel current (I CaL ) was not up-regulated in these mice. Thus, our data reveal close coupling between ␤ 2 -AR signaling and the I Ks channel and indicate that, with sufficiently high level of ␤ 2 -AR expression, the channel can be PKA-phosphorylated and up-regulated even in the absence of exogenous agonist.

Transgenic Mice and Isolation of Cardiac Ventricular Myocytes
Transgenic mice expressing hKCNQ1-hKCNE1 fusion protein in the heart have been described in detail (9). These mice express functional slow delayed rectifier potassium channel currents (I Ks ) normally absent from murine cardiac ventricular myocytes. Transgenic mice overexpressing ␤ 2 -AR (TG4) have also previously been described in detail (11). TG4 mice exhibit cardiac overexpression of the human ␤ 2 -AR at Ͼ100fold higher than endogenous myocardial levels. Strains of TG ϩ mice overexpressing ␤ 2 -AR and hKCNQ1-hKCNE1 were crossed to produce double TG ϩ (DTG ϩ ) mice overexpressing both ␤ 2 -AR and expressing hKCNQ1-hKCNE1. Both transgene constructs were under the control of the ␣-myosin heavy chain promoter providing cardiomyocyte specific expression. Mice were genotyped by PCR analysis with primers specific for the ␣-myosin heavy chain promoter, ␤ 2 -AR, and KCNQ1. Adult mice were humanely killed by intraperitoneal injection of a lethal dose of pentobarbital (100 mg kg Ϫ1 ), and acutely dissociated single ventricular myocytes were obtained by using previously described methods (14,15) using ϳ 10,500 units of collagenase type II (Invitrogen) and ϳ6 units of protease type XIV (Sigma). The Institutional Animal Care and Use Committee at Columbia University approved the protocols for all animal studies.

Immunohistochemistry
Ventricular myocytes were isolated as above and plated onto laminin (Sigma) coated coverslip chambers and incubated at 37°C in Dulbecco's modified Eagle's medium to allow cellular attachment. The cells were then fixed and permeabilized using 100% ethanol (15 min at Ϫ20°C). The cells were rinsed (three times) in phosphate-buffered saline and then incubated for 20 min with 5% normal goat serum in phosphatebuffered saline to reduce nonspecific binding. After overnight (5°C) incubation with primary antibodies (Ab) against KCNQ1 (Santa Cruz) or ␤ 2 -AR (Santa Cruz) cells were rinsed four times with phosphatebuffered saline, after which secondary Abs were added for 2 h at room temperature. IgGs conjugated to Alexa 488 or Alexa 660 were used to label the primary Abs. After a further four washes, coverslips (#1) were mounted onto the microscope slides along with Slowfade TM light (Molecular Probes S-7461) and anti-fade reagent in a glycerol buffer. The coverslips were sealed in place using nail varnish. To determine the degree of nonspecific binding, incubation with secondary Ab alone was conducted (minimal fluorescence was observed; results not shown). The cells were viewed using a Nikon PCM2000 fluorescence confocal laserscanning microscope. Alexa 488 was excited with 488-nm light, and fluorescence emissions were measured at greater than 505 nm. Alexa 660 was excited with 635-nm light, and fluorescence emissions were measured between 650 and 750 nm. The images were optimized by adjustment of photo multiplier gain to use the full linear range of pixel intensity, and fluorophore emissions were simultaneously recorded using separate photo multiplier tubes for each wavelength.

Co-localization
Immunofluorescence images of each protein were obtained in the same cell. For the purposes of identifying the location of each protein, fluorescence intensity was used. All regions of the cell with fluorescence intensity equal to or greater than a threshold level were identified and marked in each image as a site of the protein as long as at least three contiguous pixels were at or above the threshold. Co-localization was marked when identical regions contained both proteins. The threshold was set using pixel intensity histograms of each immunofluorescence image. The most frequent pixel intensity after the black level peak was set as the threshold. Images were discarded for this analysis when noise in the image or saturation was significant. Analysis was carried out using Velocity software by Improvision (Coventry, UK).

Acceptor Bleaching Fluorescence Resonance Energy Transfer
Immunohistochemical labeling of isolated ventricular myocytes was performed as above. Secondary Abs with one of the FRET pair Alexa 555 (donor) or Alexa 647 (acceptor) were used. This FRET pair has a Forster radius of 5.1 nm (Molecular Probes). The acceptor and donor fluorophores were excited with 543-and 635-nm light, and emissions were measured at 565-620 and 650 -750 nm, respectively. Firstly, separate images were acquired for the individual fluorophore emissions (Alexa 555 and 647). Following this, excitation of the acceptor fluorophore alone (647-nm light only) for ϳ5 min to photobleach the fluorophore was performed. After acceptor photobleaching, fluorescence emission from the donor fluorophore (Alexa 555) was recorded, using identical gains to those before acceptor photobleaching. Even at increased gains no fluorescence emission was observed from the acceptor after photobleaching. Any increase in emitted fluorescence from the donor fluorophore after acceptor photobleaching indicates FRET between donor and acceptor fluorophores was occurring prior to photobleaching. The occurrence of FRET between this particular pair of fluorophores requires a distance between them of 5.1 nm or less.

Electrophysiology
Single ventricular myocytes were isolated as above. The membrane currents were recorded at room temperature using the whole cell patch clamp technique (16). Solutions used for recording I Ks were as follows, superfusion solution 132 mM NaCl, 4.8 mM KCl, 1.2 mM MgCl 2 , 2 mM CaCl 2 , 5 mM glucose, and 10 mM HEPES, pH 7.4, plus E-4031 (5 M) to block I Kr and nisoldipine (1 M) to block I CaL . Patch pipettes of nominal resistance of 0.5-3 M⍀ were used and were filled with an internal solution of 110 mM aspartic acid (potassium salt), 5 mM ATP-K 2 , 11 mM EGTA, 10 mM HEPES, 1 mM CaCl 2 , and 1 mM MgCl 2 , pH 7.3. Whole cell L-type Ca 2ϩ current (I CaL ) were recorded using a superfusion solution containing 132 mM NaCl, 4.8 mM CsCl, 1.2 mM MgCl 2 , 2 mM CaCl 2 , 5 mM glucose, and 10 mM HEPES, pH 7.4, plus tetrodotoxin (30 M) (Calbiochem) to block I Na , and internal solution contained 50 mM aspartic acid, 5 mM K 2 ATP, 60 mM CsCl, 11 mM EGTA, 10 mM HEPES, and 1 mM CaCl 2 , pH 7.3. Access resistance was 2-10 M⍀. In experiments to study I Ks membrane potential was held at Ϫ65 mV with prepulse depolarizations from Ϫ40 mV to ϩ60 mV for 2 s imposed at 0.067 Hz before a 2-s test pulse to Ϫ40 mV. To ensure I Ks stabilization after the membrane rupture or drug application, I Ks was monitored by prepulse depolarizations (2 s) to ϩ60 mV followed by a return pulse (2 s) to Ϫ40 mV at 0.067 Hz. Unless otherwise noted, isochronal activation of I Ks was studied by analysis of deactivating tail currents recorded at Ϫ40 mV after 2-s depolarizing pulses in 20-mV increments. I Ks tail currents resulting from this test pulse were recorded. In experiments to study the effects of inhibiting cAMP dependent protein kinase, the selective and potent inhibitor of PKA, H-89 (Sigma) (5 M) (17) was included in the I Ks superfusion solution described above. In experiments to study I CaL , membrane potential was held at Ϫ40 mV to inactivate I Na . I CaL activation was measured by 150-ms test pulses to a series of potentials (10-mV increments) at 0.2 Hz from Ϫ40 mV to ϩ60 mV. To examine the effects of cAMP upon I CaL the membrane permeable cAMP analog (pCPT-cAMP) (Sigma) (0.5 mM) was added to the superfusion solution. To ensure I CaL stabilization after membrane rupture or drug application, I CaL was monitored by depolarizations (150 ms) to ϩ10 mV at 0.2 Hz. To obtain peak amplitudes, the leak currents were digitally subtracted by the P/N method (N ϭ 6). Note that the representative traces shown in the figures are raw traces without subtracting the leak currents. The activation curves were fitted with Boltzmann equations to determine the influence of ␤-AR stimulation, overexpression of ␤ 2 -AR, or effect of drugs on the voltage for which the half of channels are available (V1 ⁄2 ) and the slope factor for the curve (k).

Molecular Biology
Immunoprecipitation and PKA Back-phosphorylation of KCNQ1 Channels-Heart homogenates were prepared by homogenizing ϳ1.0 g of cardiac tissue in 1.0 ml of a buffer containing 10 mM Tris-maleate, 10 mM NaF, 1.0 mM Na 3 VO 4 , and protease inhibitors (complete tablets from Roche Applied Science), pH 7.0. The samples were centrifuged at 3,000 ϫ g for 10 min, and the supernatants were centrifuged at 12,000 ϫ g for 20 min. After determining the protein concentrations of the supernatants, the samples were aliquoted and stored at Ϫ80°C until use. The cardiac KCNQ1 channel was immunoprecipitated from heart samples by incubating 500 g of homogenate with anti-KCNQ antibody (Santa Cruz) in 0.5 ml of a modified radioimmune precipitation assay buffer (containing 50 mM Tris-HCl, 0.9% NaCl, 5.0 mM NaF, 1.0 mM Na 3 VO 4 , 0.25% Triton-X100, and protease inhibitors, pH 7.4) for 4 h at 4°C. The samples were incubated with protein A-Sepharose beads (Amersham Biosciences) at 4°C for 1 h, after which the beads were washed three times with 1ϫ kinase buffer containing 8 mM MgCl 2 , 10 mM EGTA, and 50 mM Tris/piperazine-N,NЈ-bis(2-ethanesulfonic acid), pH 6.8. After resuspending the beads in 10 l of 1.5ϫ kinase buffer containing 5 units of PKA catalytic subunit (Sigma), back-phosphorylation of the immunoprecipitated KCNQ1 was initiated with 5 l of 100 M Mg-ATP containing 10% [␥-32 P]ATP (PerkinElmer Life Sciences). The reaction was terminated after 8 min at room temperature with 5 l of stop solution (4% SDS and 0.25 M dithiothreitol). The samples were heated to 95°C, size-fractionated on 10% PAGE, and KCNQ1 radioactivity was quantified using a Molecular Dynamics phosphorimaging and ImageQuant software (Amersham Biosciences). The resulting value was divided by the amount of KCNQ1 protein and expressed as the inverse of the PKA-dependent 32 P signal.
Western Analysis-The proteins were size-fractionated on 10% SDS-PAGE for KCNQ1, and the immunoblots were developed using anti-KCNQ1 Ab (custom made by Zymed Laboratories Inc., San Francisco, CA) (18) using (GARRGpSAGL) as the antigenic peptide) diluted in 5% milk/Tris-buffered saline/Tween. Immunoblot signals were quantified by densitometry.

Statistics
Graphical and statistical data analysis was carried out using Origin 7.0 software (Microcal, Northampton, MA), Excel (Microsoft), and Clampfit 8.2 (Axon Instruments, Inc.). The data are presented as the mean values Ϯ S.E. Statistical significance was assessed with Student's t test for simple comparisons; differences at p Ͻ 0.05 were considered to be significant. Two-sample comparisons were performed using unpaired Student's t test.

Localization of KCNQ1 and ␤ 2 -AR in Murine Ventricular
Myocytes-Using double labeling immunohistochemical techniques combined with laser scanning confocal microscopy KCNQ1 and ␤ 2 -AR were found to be in similar subcellular localizations in ventricular myocytes. In cells isolated from the FIG. 1. Localization of KCNQ1 and ␤ 2 -AR in murine ventricular myocytes. Simultaneous labeling of KCNQ1 and ␤ 2 -AR in ventricular myocytes with primary antibodies and visualization of this localization with fluorophore conjugated secondary antibodies and laser scanning confocal microscopy was performed. Fluorophores conjugated to the secondary antibodies used were Alexa 488 or Alexa 660, eliminating any spectral overlap of fluorophores. A, subcellular localizations of KCNQ1 and ␤ 2 -AR in ventricular myocytes from hKCNQ1-hKCNE1 transgenic mice were examined using immunohistochemistry KCNQ1 is seen in the SSM, ICD and transverse (T) tubules (panels ia and ib). ␤ 2 -ARs are also seen to be localized to the SSM, ICD region, and T-tubules. B, subcellular localization of KCNQ1 and ␤ 2 -AR in ventricular myocytes of hKCNQ1-hKCNE1/␤ 2 -AR double transgenic mouse heart KCNQ1 is seen in the SSM, ICD, and transverse (T) tubules (panels ia and ib). ␤ 2 -ARs in these myocytes are also localized to the SSM, T-tubules, and ICD regions. Scale bars, 10 m. Serial sections at 0.5 m (not shown) revealed that this localization pattern was uniform throughout myocytes examined. C, co-localization of KCNQ1 and ␤ 2 -AR determined from pixel intensity distribution of digitized images. High levels of fractional overlap of KCNQ1 and ␤ 2 -AR indicate co-localization of these two proteins in isolated ventricular myocytes from both hKCNQ1-hKCNE1 (TG ϩ ) and hKCNQ1-hKCNE1/␤ 2 -AR (DTG ϩ ) mice. No significant differences in fractional overlap of KCNQ1 and ␤ 2 -AR in myocytes from TG ϩ and DTG ϩ mice are observed. hKCNQ1-hKCNE1 (TG ϩ ) mice, KCNQ1 was localized to the intercalated disc (ICD) regions, the surface sarcolemmal membrane (SSM), and the transverse (T-) tubules (see arrows A-C in Fig. 1A, panel ib). In these myocytes ␤ 2 -AR were also located in ICD, SSM, and T-tubules (Fig. 1A, panel iib). In ventricular myocytes from DTG ϩ mice, expressing hKCNQ1-hKCNE1 and overexpressing ␤ 2 -ARs, a similar localization for KCNQ1-KCNE1 and ␤ 2 -ARs to the ICD, SSM, and T-tubules was observed (Fig. 1B). KCNQ1 and ␤ 2 -AR were determined to be highly co-localized in myocytes from both TG ϩ and DTG ϩ mice. However, the resolving power of conventional confocal microscopy is limited to ϳ0.2 m. With cardiac T-tubules being up to 300 nm in diameter, it would be possible for proteins to appear co-localized without in fact having an intimate association or localization with each other. To investigate this further we made use of FRET to provide substantially increased spatial resolution.
Acceptor Bleaching FRET between Imunohistochemically Labeled KCNQ1 and ␤ 2 -AR in Murine Ventricular Myocytes-Using dual labeling immunohistochemical techniques and acceptor bleaching FRET methods combined with laser scanning confocal microscopy, an intimate localization of KCNQ1 and ␤ 2 -AR within nanometers of each other was seen. For the particular pair of fluorophores used here the distance between them must be 5.1 nm or less for FRET to occur. Any increase in donor fluorescence after acceptor photobleaching indicates the occurrence of FRET before photobleaching The increased intensity of donor fluorescence seen after acceptor photobleaching (Fig. 2, compare A and D; also see Fig. 2F) indicates there is an intimate and close localization between KCNQ1 and ␤ 2 -AR, within nanometers of each other (Fig. 2).
Effect of ␤-AR Stimulation on I Ks in Ventricular Myocytes from hKCNQ1-hKCNE1 Mice-We have previously shown that I Ks channels are regulated in the hKCNQ1-hKCNE1 mouse in response to PKA stimulation (9). As indicated in Fig. 3, stimulation of ␤-AR using the ␤-AR agonist isoproterenol (ISO) increases I Ks in myocytes isolated from hKCNQ1-hKCNE1 mice, as reflected in the increase of outward current measured in response to depolarization and in the decay of outward current (tail) after termination of depolarizing pulses. I Ks tail currents in myocytes from hKCNQ1-hKCNE1 mice were significantly (p Ͻ 0.05) increased for all prepulse potentials examined. In addition to the increase in I Ks tail current, ISO induced a hyperpolarizing shift in the I Ks activation curve in myocytes from these mice (Fig. 3 and Table I). Thus, the hKCNQ1-hKCNE1 channel is regulated by agonist-induced ␤-AR stimulation in the TG ϩ mice in a manner strikingly similar to the effects of ␤-AR stimulation on I Ks in other species, such as guinea pig, where a similar channel is endogenously expressed (6,19). We next asked whether or not similar I Ks regulation could be detected in myocytes from DTG ϩ mice with overexpression of ␤ 2 -AR in addition to expression of hKCNQ1-hKCNE1 channels.
I Ks in Ventricular Myocytes from hKCNQ1-hKCNE1 and hKCNQ1-hKCNE1/␤ 2 -AR Mice-We characterized I Ks in ventricular myocytes from hKCNQ1-hKCNE1 and hKCNQ1-hKCNE1/␤ 2 -AR (DTG ϩ ) mice. We detected robust expression of I Ks in these cells and found that I Ks current density was consistently greater in DTG ϩ myocytes compared with TG ϩ (hKCNQ1-hKCNE1) myocytes. I Ks tail current density was significantly (p Ͻ 0.05) greater in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice at all prepulse potentials examined. In addition to a larger I Ks tail current, a hyperpolarizing shift in the normalized I Ks tail current was recorded in myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice ( Fig. 4 and Table II). Thus, similar to ␤-AR stimulation in hKCNQ1-hKCNE1 myocytes, overexpression of ␤ 2 -AR produces an increase in I Ks and altered channel gating even in the absence of exogenous agonist. Overexpression of ␤ 2 -AR resulted in significantly (p Ͻ 0.05) increased current density (ϳ135%), and activation was shifted in the hyperpolarizing direction in myocytes from DTG ϩ mice compared with TG ϩ mice expressing the hKCNQ1-hKCNE1 transgene alone (Table III). We found no further increase in I Ks density or change in activation voltage dependence with ␤-AR stimulation in DTG ϩ mice ( Fig. 5 and Table  III), suggesting that ␤ 2 -AR overexpression results in maximal activation of I Ks . These results suggest that overexpression of ␤ 2 -AR results in maximal up-regulation of I Ks without the application of additional ␤-AR agonist.
PKA Phosphorylation of hKCNQ1 in hKCNQ1-hKCNE1/ ␤ 2 -AR Mice-Functional data from cardiomyocytes isolated from the transgenic mice indicate that I Ks channels are maximally stimulated by ␤-ARs. These results suggest that this channel complex may be PKA-phosphorylated in the mouse model. To address this possibility we used back-phosphorylation experiments to determine the relative PKA phosphorylation of KCNQ1/KCNE1 channels in our DTG ϩ mice. In this procedure immunoprecipitated proteins are incubated with radioactively labeled ATP. The more PKA-phosphorylated a protein, the less radioactive phosphate it will be able to incorporate. Back-phosphorylation experiments revealed greater PKA phosphorylation of KCNQ1 in DTG ϩ mice overexpressing ␤ 2 -ARs compared with PKA phosphorylation of KCNQ1 in our TG ϩ mice, consistent with the predictions of our functional data (Fig. 6).
Inhibition of PKA in Myocytes from hKCNQ1-hKCNE1/ ␤ 2 -AR Mice-Results from the above back-phosphorylation experiments show KCNQ1 phosphorylation to be increased in the presence of overexpression of ␤ 2 AR. Consistent with functional data this increased phosphorylation of KCNQ1 is thought to result from the action of PKA, which becomes activated as a  result of spontaneous action of overexpressed ␤ 2 ARs (even in the absence of agonist). To confirm the observed increase in phosphorylation of KCNQ1 and the concomitant increase in I KS is in fact due to the action of PKA, we examined I Ks in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 AR mice in the presence of H-89, a selective inhibitor of PKA. In the presence of H-89 a significant (p ϭ 0.01) reduction (20 Ϯ 4.6%) in I Ks peak tail current density (pA/pF) was observed (Fig. 7). This clearly confirms the functional involvement of PKA phosphorylation in the maximal activation of I Ks observed in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice. Effect of Overexpressing ␤ 2 -AR on I CaL in Ventricular Myocytes-The maximal activation of I Ks seen with the overexpression of ␤ 2 -AR may simply be the result of a global increase in intracellular cAMP or alternatively a more localized and specific increase that selectively modulates the I Ks channel. To examine this possibility we recorded L-type Ca 2ϩ channel currents I CaL in ventricular myocytes isolated from mice overexpressing ␤ 2 -AR, both with and without hKCNQ1-hKCNE1, in a cardiac-specific fashion. The first indication that I Ks modulation with ␤ 2 -AR overexpression is localized and specific was the lack of modulation of I CaL seen with ␤ 2 -AR overexpression. I CaL density and the activation voltage dependence was the same for myocytes from wild type, ␤ 2 -AR overexpressing and hKCNQ1-hKCNE1/␤ 2 -AR mice (Fig. 8A). The addition of pCPT-cAMP (a membrane-permeable cAMP analog) to myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice produced a significant and characteristic increase in the magnitude of I CaL and hyperpolarizing shift (12 mV) in activation (Fig. 8B). Similar results were observed in ventricular myocytes from mice overexpressing ␤ 2 AR alone (data not shown). These results indicate the expression of I Ks does not affect modulation of I CaL and that I CaL remains fully responsive to cAMP in the presence of ␤ 2 -AR overexpression. In addition these results indicate that the modulation of I Ks as a result of ␤ 2 -AR overexpression is localized and specific and not simply the result of a global increase in cAMP. FIG. 4. I Ks in ventricular myocytes from hKCNQ1-hKCNE1 and hKCNQ1-hKCNE1/␤ 2 -AR mouse hearts. A, voltage dependence of I Ks current density (pA/ pF). I Ks tail current density in response to depolarizing step to Ϫ40 mV after prepulse steps to potentials from Ϫ40 mV to ϩ80 mV in hKCNQ1-hKCNE1 (n ϭ 8) and hKCNQ1-hKCNE1/␤ 2 -AR (n ϭ 6) myocytes. In myocytes from DTG ϩ mice I Ks is significantly (p Ͻ 0.05) greater at all potentials examined above Ϫ40 mV. Sample currents are shown in the inset. B, voltage dependence of normalized I Ks tail currents showing the voltage dependence of channel activation. In addition to greater current density in myocytes from DTG ϩ mice a negative shift in activation of I Ks is also seen, with the channel being activated at more negative potentials. Half-maximal activation (V1 ⁄2 ) for I Ks in myocytes from mice is 36.2 Ϯ 5.0 mV in contrast to 15.2 Ϯ 2.7 mV for hKCNQ1-hKCNE1/␤ 2 -AR myocytes. The asterisks indicate p Ͻ 0.05. Inset scale bars represent 1 s (abscissa) and 20 pA/pF (ordinate).

DISCUSSION
Intimate Localization of KCNQ1 and ␤ 2 -AR in TG ϩ Mice-In ventricular myocytes isolated from TG ϩ mice expressing the fusion protein hKCNQ1-hKCNE1, KCNQ1 is localized in Ttubular membranes, at the surface sarcolemmal membrane and at the intercalated discs. This distribution throughout all FIG. 5. I Ks in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice is insensitive to ␤-AR stimulation. A, voltage dependence of I Ks current density (pA/pF). I Ks tail current density in response to depolarizing step to Ϫ40 mV after prepulse steps to potentials from Ϫ40 mV to ϩ80 mV in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice in the absence (n ϭ 5) and presence (n  6. PKA phosphorylation of KCNQ1 in hKCNQ1-hKCNE1/␤ 2 -AR transgenic mice. The PKA phosphorylation of KCNQ1 was measured in hearts from transgenic mice. KCNQ1 was immunoprecipitated from 0.5 mg of cardiac homogenates, and the total amount of KCNQ1 in the immunoprecipitate was determined by Western blotting using an anti-KCNQ1 antibody. PKA phosphorylation of KCNQ1 was determined by back-phosphorylation using exogenous PKA and [␥-32 P]ATP. The strength of the 32 P signal is inversely proportional to the endogenous PKA phosphorylation of the channel. Cardiac tissue from hKCNQ1-hKCNE1/␤ 2 -AR mice showed ϳ4-fold increase in PKA-phosphorylated KCNQ1 compared with TG ϩ mice expressing hKCNQ1-hKCNE1 channels alone (n ϭ 4 in both groups, p Ͻ 0.001). In this figure Norm refers to myocytes expressing hKCNQ1-hKCNE1 alone, with X ␤ 2 AR representing myocytes expressing hKCNQ1-hKCNE1 and ␤ 2 AR. BP, back-phosphorylation; IB, immunoblot. sarcolemmal membranes at the periphery of the cell is not surprising considering the importance of I Ks in ventricular repolarization. ␤ 2 -ARs are also observed in T-tubular membranes and ICD, however, appear to be not as abundant at the surface sarcolemmal membrane. These results suggest that some KCNQ1-KCNE1 channels may be in similar subcellular regions as ␤ 2 -ARs, whereas others may be in distinct localizations. Some may be coupled to ␤ 2 -ARs, whereas others may not. In fact, recent evidence suggesting the importance of PKC phosphorylation for I Ks has been demonstrated (20), raising the possibility that there are distinct populations of I Ks modulated by different regulatory pathways. In ventricular myocytes isolated from DTG ϩ mice a subcellular localization of KCNQ1 and ␤ 2 -AR similar to that seen in TG ϩ myocytes is observed. Overexpression of ␤ 2 -AR appears to result in a slightly increased distribution of ␤ 2 -AR and possibly KCNQ1 at the surface sarcolemmal membrane compared with ICD and T-tubular membranes in these DTG ϩ myocytes, although more extensive studies are required to examine this further. In ventricular myocytes from TG ϩ mice, the occurrence of FRET between labeled KCNQ1 and endogenous ␤ 2 -AR indicates there are not only a similar subcellular localization of these proteins but also an intimate association with KCNQ1 and ␤ 2 -AR being located within nanometers of each other.
PKA Phosphorylation of I Ks in Murine Hearts-PKA phosphorylation of KCNQ1 increases I Ks and alters its voltage dependence in a manner that contributes to shortening of the ventricular action potential (6,9,21), and we confirm this in I Ks measured in myocytes isolated from hearts of our DTG ϩ mice. PKA phosphorylation of KCNQ1 in failing human hearts would thus be expected to produce an increase in available repolarizing current in these hearts, which would be expected, on its own, to shorten ventricular action potentials. However, a common feature seen in heart failure is action potential duration prolongation, which, in most cases, is due to a net reduction in outward potassium currents (22). Decreases in outward current such as that reported to be due to the down-regulation of the transient outward potassium current (I to ) (22, 23) must be greater than the PKA phosphorylation-induced increase in I Ks we report here. However, if PKA phosphorylation of I Ks chan-nels were reduced in failing hearts, our data would suggest that there would be a concomitant decrease in I Ks and hence less repolarization reserve to counter the effects of down-regulation of other K channel pathways. The net result would be excessive prolongation of the ventricular response and greater increase of arrhythmia.
The present study suggests that KCNQ1 and ␤ 2 -AR are functionally coupled and that ␤ 2 -AR can contribute to the modulation of I Ks via PKA phosphorylation of KCNQ1. Moreover, data showing PKA phosphorylation of KCNQ1 in murine hearts overexpressing ␤ 2 -AR suggests that activation of I Ks during stress may be a protective mechanism to prevent cardiac arrhythmias. However, in heart failure this mechanism appears to be futile because cardiac arrhythmias are highly prevalent in failing hearts. ␤-Adrenergic receptor blockers, an approved treatment for heart failure, might be expected to inhibit PKA phosphorylation induced activation of I Ks , and this could, to some extent, attenuate the beneficial effects of this therapy. However, there is also no evidence to support this possibility, because ␤-adrenergic receptor blockers are known to decrease cardiac arrhythmias, particularly in heart failure patients. This implies that the anti-arrhythmic effects of ␤-adrenergic receptor blockers are targeted at a distinct signaling pathway (possibly the RyR2 by inhibiting PKA hyperphosphorylation of RyR2) and outweigh the potential adverse effects of inhibiting PKA phosphorylation of KCNQ1.
PKA Phosphorylation of KCNQ1 in Mice Overexpressing the ␤ 2 -AR-Here we extend the analysis of putative substrates for PKA phosphorylation that may be modulated in chronic activation of the stress response and show, for the first time, that like RyR2 (24), the KCNQ1-KCNE1 channel is also PKA-phosphorylated in a transgenic mouse in which the human KCNQ1-KCNE1 channel and h␤ 2 -AR are overexpressed. Using patch clamp analysis of myocytes isolated from DTG ϩ mice, we found that overexpression of ␤ 2 -AR increases I Ks , causing a hyperpolarizing voltage shift in channel activation, and renders the channels insensitive to further modulation by addition of external ␤-AR agonist. Thus, overexpression of ␤-ARs alone can stimulate downstream signaling that results in sustained PKA phosphorylation of substrates (hKCNQ1 channel in this case). It is inter- esting to note that the functional coupling between ␤-AR and I Ks seen in adult myocytes is fundamentally different from that seen during development when I Ks is not affected by overexpression of ␤ 2 -ARs (25), indicating that changes in ␤ 2 -AR coupling to effecter targets occur during developmental changes.
Localized and Specific Modulation of I Ks in Mice Overexpressing ␤ 2 -AR-The maximal activation of I Ks seen in myocytes overexpressing ␤ 2 -AR may simply be the result of a global increase in cAMP. To investigate this possibility we examined I CaL in ventricular myocytes from mice overexpressing ␤ 2 -ARs (both with and without hKCNQ1-hKCNE1). Using patch clamp analysis we found no basal modulation of I CaL as a result of overexpressing ␤ 2 -AR. The addition of pCTP-cAMP produced a robust increase in both the peak current and a hyperpolarizing shift in I CaL activation. This is in contrast to I Ks , which was found to be maximally activated under basal conditions and not responsive to further stimulation in myocytes from DTG ϩ . These results show the modulation of I Ks is not simply the result of a global increase in cAMP and that I Ks modulation by overexpression of ␤ 2 -AR is localized and specific. Further, we find that modulation of I Ks occurs under conditions of increased ␤ 2 -AR expression, even in the absence of exogenous ␤-AR agonist.
Chronic PKA Phosphorylation of Targets in Failing Human Hearts-Heart failure has been described as a maladaptive activation of the classic "fight or flight" stress response that occurs as a consequence of a systemic response to maintain cardiac output in the face of decreasing cardiac contractile performance (26). One consequence of the chronic maladaptive response has been shown to be PKA hyperphosphorylation of FIG. 8. L-type Ca 2؉ current (I CaL ) in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mouse hearts is sensitive to cAMP stimulation. A, comparison of I CaL activation between myocytes from wild type (WT) (open square), ␤ 2 -AR overexpressing (filled circle), and hKCNQ1-hKCNE1/␤ 2 -AR (filled square) mice. Peak current-voltage relationships (left) measured during depolarizing pluses (150 ms) from Ϫ40 mV to ϩ60 mV (10-mV increments) from a holding potential of Ϫ40 mV were not changed by ␤ 2 -AR overexpression or expression of hKCNQ1-hKCNE1. Representative raw traces (middle panel) recorded at 0 mV test potential show similar inactivation kinetics for I CaL in all three types of myocytes. Scale bars indicate 5 pA/pF and 50 ms. The voltage dependence of I CaL activation (right panel) was plotted by normalizing the current-voltage relationship to driving force. Half-maximal activation (V1 ⁄2 ) is Ϫ1.0 Ϯ 2.7 mV (n ϭ 10) for WT, Ϫ1.0 Ϯ 1.6 mV (n ϭ 16) for ␤ 2 -AR, and Ϫ2.0 Ϯ 1.1 mV (n ϭ 7) for hKCNQ1-hKCNE1/␤ 2 -AR myocytes (p ϭ not significant). B, effect of cAMP stimulation on I CaL in ventricular myocytes from hKCNQ1-hKCNE1/␤ 2 -AR mice. Peak current-voltage relationships (left panel) were obtained from seven cells as described in A before (control, open circle) and after (pCPT-cAMP, filled circle) a 5-min external application of pCPT-cAMP (0.5 mM). With the addition of pCPT-cAMP, peak I CaL amplitudes over depolarizing pulses (Ϫ30 to ϩ60 mV) were increased (left panel). Representative raw traces (middle panel) recorded at 0 mV test potential are superimposed before and after the application of pCPT-cAMP. Capacitative currents greater than 1 pA/pF are truncated. Scale bars indicate 5 pA/pF and 50 ms. Voltage dependence of I CaL activation (right panel) shows a hyperpolarizing shift in the presence of pCPT-cAMP. Half-maximal activation (V1 ⁄2 ) for control is Ϫ2.0 Ϯ 1.1 mV and Ϫ14 Ϯ 2.5 mV in the presence of pCPT-cAMP. The asterisks indicate p Ͻ 0.05 versus control. the cardiac ryanodine receptor, RyR2, which results in altered calcium homeostasis (an aberrant diastolic sarcoplasmic reticulum calcium leak) in the heart (24,27,28). Selective downregulation of ␤ 1 -AR results in a relative increase in the abundance of ␤ 2 -ARs (29). PKA phosphorylation of RyRs have been observed in heart failure (26). Moreover, RyRs are seen to be PKA-phosphorylated in ␤ 2 -AR overexpressing mice (24). Our results complement those previously obtained for the RyR and show that a second PKA substrate, the I Ks channel, is also chronically PKA-phosphorylated in the ␤ 2 -AR overexpressing mouse. It will be of interest to determine whether, like the RyR, the I Ks channel is also PKA-phosphorylated in failing human heart where, if so, it would act to counter the documented action potential prolongation in this disorder. This observation in addition to the fact that a relative increase in the abundance of ␤ 2 -ARs is seen in heart failure prompts the question: are other end effectors of ␤-AR stimulation, namely the KCNQ1/ KCNE1 channel, PKA-phosphorylated under these conditions?
In summary, KCNQ1 and ␤ 2 -AR are localized to the intercalated discs, surface sarcolemma, and transverse tubules of isolated ventricular myocytes. In these subcellular regions KCNQ1 and ␤ 2 -AR are closely associated, lying within nanometers of each other. This intimate association results in a functional coupling as shown by overexpression of ␤ 2 -AR in the heart, which results in PKA phosphorylation of KCNQ1 and maximal activation of I Ks . Conversely, ␤ 2 -AR overexpression did not affect I CaL under basal conditions with I CaL remaining responsive to cAMP. These data indicate intimate association of KCNQ1 and ␤ 2 -AR and that ␤ 2 -AR signaling can modulate the function of KCNQ1 in a localized and specific fashion with up-regulation of the slow delayed rectifier potassium channel current (I Ks ) with chronic activation of ␤ 2 -AR, suggesting functional coupling between I Ks channels and ␤ 2 ARs with modulation of I Ks occurring under conditions of increased ␤ 2 -AR expression, even in the absence of exogenous ␤-AR agonist. As such, cardiomyocytes overexpressing ␤ 2 -AR may provide a good system to examine effects on ion channel function under conditions of chronic activation of ␤-AR. Maximal activation of I Ks seen under these conditions may act as a physiological brake to counter action potential duration prolongation during stress.