Subcellular compartmentalization of proximal Gαq-receptor signaling produces unique hypertrophic phenotypes in adult cardiac myocytes

G protein–coupled receptors that signal through Gαq (Gq receptors), such as α1-adrenergic receptors (α1-ARs) or angiotensin receptors, share a common proximal signaling pathway that activates phospholipase Cβ1 (PLCβ1), which cleaves phosphatidylinositol 4,5-bisphosphate (PIP2) to produce inositol 1,4,5-trisphosphate (IP3) and diacylglycerol. Despite these common proximal signaling mechanisms, Gq receptors produce distinct physiological responses, yet the mechanistic basis for this remains unclear. In the heart, Gq receptors are thought to induce myocyte hypertrophy through a mechanism termed excitation–transcription coupling, which provides a mechanistic basis for compartmentalization of calcium required for contraction versus IP3-dependent intranuclear calcium required for hypertrophy. Here, we identified subcellular compartmentalization of Gq-receptor signaling as a mechanistic basis for unique Gq receptor–induced hypertrophic phenotypes in cardiac myocytes. We show that α1-ARs co-localize with PLCβ1 and PIP2 at the nuclear membrane. Further, nuclear α1-ARs induced intranuclear PLCβ1 activity, leading to histone deacetylase 5 (HDAC5) export and a robust transcriptional response (i.e. significant up- or down-regulation of 806 genes). Conversely, we found that angiotensin receptors localize to the sarcolemma and induce sarcolemmal PLCβ1 activity, but fail to promote HDAC5 nuclear export, while producing a transcriptional response that is mostly a subset of α1-AR–induced transcription. In summary, these results link Gq-receptor compartmentalization in cardiac myocytes to unique hypertrophic transcription. They suggest a new model of excitation–transcription coupling in adult cardiac myocytes that accounts for differential Gq-receptor localization and better explains distinct physiological functions of Gq receptors.

G protein-coupled receptors that signal through G␣ q (G q receptors), such as ␣ 1 -adrenergic receptors (␣ 1 -ARs) or angiotensin receptors, share a common proximal signaling pathway that activates phospholipase C␤1 (PLC␤1), which cleaves phosphatidylinositol 4,5-bisphosphate (PIP 2 ) to produce inositol 1,4,5-trisphosphate (IP 3 ) and diacylglycerol. Despite these common proximal signaling mechanisms, G q receptors produce distinct physiological responses, yet the mechanistic basis for this remains unclear. In the heart, G q receptors are thought to induce myocyte hypertrophy through a mechanism termed excitation-transcription coupling, which provides a mechanistic basis for compartmentalization of calcium required for contraction versus IP 3 -dependent intranuclear calcium required for hypertrophy. Here, we identified subcellular compartmentalization of G q -receptor signaling as a mechanistic basis for unique G q receptor-induced hypertrophic phenotypes in cardiac myocytes. We show that ␣ 1 -ARs co-localize with PLC␤1 and PIP 2 at the nuclear membrane. Further, nuclear ␣ 1 -ARs induced intranuclear PLC␤1 activity, leading to histone deacetylase 5 (HDAC5) export and a robust transcriptional response (i.e. significant up-or down-regulation of 806 genes). Conversely, we found that angiotensin receptors localize to the sarcolemma and induce sarcolemmal PLC␤1 activity, but fail to promote HDAC5 nuclear export, while producing a transcriptional response that is mostly a subset of ␣ 1 -AR-induced transcription. In summary, these results link G q -receptor compartmentalization in cardiac myocytes to unique hypertrophic transcription. They suggest a new model of excitation-transcription coupling in adult cardiac myocytes that accounts for differential G q -receptor localization and better explains distinct physiological functions of G q receptors.
G protein-coupled receptors that signal through G␣ q (G q receptors) 2 share a common proximal signaling pathway through the activation of phospholipase C␤1 (PLC␤1), which cleaves phosphatidylinositol-4,5-bisphosphate (PIP 2 ) to produce inositol-1,4,5-trisphosphate (IP 3 ) and diacylglycerol (DAG) (1). However, in any cell, simultaneous activation of multiple proximal G q -receptor signaling events would preclude the cell's ability to process different signals and produce a unique outcome. Despite this commonality, each G q receptor mediates distinct physiologic processes, but how this specificity is achieved is unclear in many cases. Compartmentalization of signaling provides one mechanism through which a cell could handle a multitude of potentially overlapping signals. Cardiac myocytes offer several examples of compartmentalized receptor signaling (e.g. the ability to discriminate ␤-adrenergic (␤-AR)-G s -induced calcium signals that augment contractility from G q receptor-mediated calcium signals that induce hypertrophy) (2). In this case, inositol-sensitive calcium release occurs in the nucleus based on the localization of the inositol trisphosphate receptor to the inner nuclear membrane. However, this ultimately raises questions about how G q receptors, traditionally thought to localize to the cell surface, might activate a nuclear calcium signal to regulate hypertrophy.
In cardiac myocytes, G q receptors, including ␣ 1 -adrenergic, endothelin, and angiotensin receptors (␣ 1 -AR, ET-R, and AT-R, respectively), regulate vital signaling pathways controlling hypertrophy, cell survival, and inotropy that impact heart failure (HF) (3). Early studies in cell and animal models established the long-held convention that G q signaling is maladaptive and exacerbates HF (4 -6). However, several studies have challenged the convention that G q -receptor signaling universally worsens HF. Clinical studies indicate that antagonists targeting G q receptors in human HF do not uniformly improve HF outcomes. Although AT-R antagonists are standard of care (7), ␣ 1 -AR antagonists worsen HF (3). In mice, 8-fold cardiac myocyte-specific overexpression of G q induces cardiac myocyte cell death and HF (6). However, G q levels are increased only 2-fold in human HF (8,9), which in mice produces no obvious phenotype (6). Our laboratory previously demonstrated that ␣ 1 -ARs are cardioprotective, as defined by their ability to initiate adaptive hypertrophy, survival signaling, and positive inotropy (reviewed in Ref. 3). The finding that ␣ 1 -ARs are cardioprotective agrees with the clinical data indicating that ␣ 1 -antagonists worsen HF. Collectively, these data suggest that G q receptors are functionally unique, some protective, like ␣ 1 -ARs, and others maladaptive, like AT-Rs.
How can these apparently divergent data on cardiac G q -receptor function be reconciled? A potentially important clue is our finding that cardiac ␣ 1 -ARs primarily localize to and signal at the nucleus unlike other G q receptors (reviewed in Refs. 3 and 10). Using fluorescent ligands and binding assays in fractionated adult cardiac myocytes, we previously demonstrated that endogenous ␣ 1 -ARs localize primarily to the nucleus. We also found that ␣ 1 -ARs contain nuclear localization sequences (NLSs), and that whereas reconstitution of WT ␣ 1 -ARs in ␣ 1 -knockout cardiac myocytes restores ␣ 1 -signaling, reconstitution of ␣ 1 -NLS mutant receptors does not, demonstrating a requirement for ␣ 1 -nuclear localization (11). Further, we showed that ␣ 1 -ARs activate intranuclear signaling in adult cardiac myocytes based on our observations that the ␣ 1 -agonist phenylephrine activates protein kinase C in isolated nuclei and that blockade of nuclear export inhibits ␣ 1 -mediated signaling for contractile function (11). Finally, we found that organic cation transporter 3 (OCT3) mediates rapid and specific catecholamine uptake in cardiac myocytes to facilitate ␣ 1 -signaling, and others have demonstrated that OCT3 knockout mice phenocopy the small heart phenotype seen in ␣ 1 -AR knockout animals (12,13). In total, we identified an entirely novel model for nuclear ␣ 1 -cardioprotective signaling in cardiac myocytes distinct from the classical model of maladaptive G q -receptor signaling.
Based on these findings, we hypothesized that unique G qreceptor function is dictated by receptor localization. To test this hypothesis, we examined the relationship between subcellular compartmentalization of proximal G q -receptor signaling and the activation of hypertrophic signaling pathways in adult cardiac myocytes. The current model of G q -receptor hypertrophic signaling is largely based on ET-R signaling and suggests that sarcolemmal G q receptors produce IP 3 -dependent intranuclear calcium release, activation of calmodulin kinase, phosphorylation and nuclear export of histone deacetylases (HDACs), and derepression of transcription (2,14). This model is notable for explaining how cytosolic calcium transients required for contraction are segregated from IP 3 -dependent nuclear calcium signals required for hypertrophic signaling. However, it is not entirely clear how this model might reconcile the data suggesting that G q receptors induce unique physiology or our model of nuclear ␣ 1 -cardioprotective signaling. Here, we report for the first time that in adult cardiac myocytes, ␣ 1 -ARs and AT-Rs localize to and activate PLC␤1 in unique subcellular compartments. Further, we demonstrate for the first time that these compartmentalized proximal signals induce differential activation of nuclear hypertrophic signaling pathways to produce unique hypertrophic transcriptomes. Finally, these data suggest an entirely new model of excitation-transcription coupling that accounts for differential localization of G q receptors and better explains the distinct physiologic function of G q receptors in adult cardiac myocytes.

␣ 1 -ARs localize to the nuclei and AT-Rs localize to the sarcolemma in adult cardiac myocytes
Here, we sought to define the subcellular localization of ␣ 1 -ARs and AT-Rs in adult cardiac myocytes ( Figs. 1 and 2). Previously, we demonstrated that endogenous ␣ 1 -ARs localize to the nucleus in adult mouse ventricular myocytes (AMVM) (11,12,15). However, reagents typically employed to localize receptors are generally unreliable, especially ␣ 1 -AR subtypespecific antibodies, which lack specificity (16), or fluorescent ligands, which are no longer commercially available, but nonetheless had suboptimal binding kinetics (12). To overcome these shortcomings, we developed a novel ␣ 1 -AR ligand composed of 2-piperazinyl-4-amino-6,7-dimethoxyquinazoline, the common pharmacophore of ␣ 1 -AR antagonists such as terazosin and doxazosin, attached to a PEG linker with a terminal biotin moiety fused to a streptavidin-coated fluorescent quantum dot (QDot) with an emission wavelength of 565 nm ( Fig. 1A) (17). We validated the ␣ 1 -QDot-565 by infecting WT AMVM with a GFP-labeled ␣ 1 A-AR (␣ 1 A-GFP), incubating infected myocytes with the ␣ 1 -QDot, and co-localizing the GFP and QDot fluorescent signals as an indication of receptor binding (Fig. 1B, nuclei indicated with white arrows). In AMVM expressing ␣ 1 A-GFP, the ␣ 1 -QDot-565 fluorescent signal colocalized with the GFP fluorescent signal at the nucleus, and pretreatment with the ␣ 1 -AR antagonist prazosin diminished QDot fluorescence to nearly undetectable levels, demonstrating specificity (Fig. 1C). We employed the same method as above in uninfected WT AMVM, and the ␣ 1 -QDot-565 bound to endogenous ␣ 1 -ARs at the nuclei in WT AMVM in the absence of prazosin but was blocked by pretreatment with prazosin, replicating our previous findings (11,12,15) (Fig. 1, D and E; quantified in F). Further, the ␣ 1 -QDot-565 bound endogenous ␣ 1 -ARs in nuclei isolated from WT AMVM (Fig.  1G), and this signal could be blocked by prazosin (Fig. 1H), indicating the presence of ␣ 1 -ARs at the nuclear membrane, again replicating our previous findings (11,12,15). Therefore, the ␣ 1 -QDot-565 fluorescent ligand identified endogenous ␣ 1 -ARs at the nucleus in AMVM and outperformed prior fluorescent ␣ 1 -ligands by improving kinetics (30 min) at lower concentrations (25 nM) (12).
Similar to ␣ 1 -ARs, the lack of validated antibodies for AT-Rs led us to employ another fluorescent ligand to define the localization of endogenous AT-Rs in AMVM. In this case, we used angiotensin II (AngII), the endogenous ligand of AT-Rs, labeled with the red fluorophore tetramethylrhodamine (TAMRA). We incubated AMVM with AngII-TAMRA in the absence or presence of unlabeled AngII to demonstrate the specificity of AngII-TAMRA (Fig. 2, A-C). In AMVM, AngII-TAMRA pro-Compartmentalized myocyte G q -receptor signaling duced a distinct localization in confocal sections from the myocyte surface (Fig. 2B, left), whereas AngII-TAMRA showed much less specific binding in confocal sections from the middle of the myocyte ( Fig. 2A, left). Pretreatment with unlabeled AngII abolished the AngII-TAMRA signal, indicating specific-ity (Fig. 2C). To clarify receptor localization, slices from the top and middle of the cell were deconvolved, and 3D surface plots were created that identified AngII-TAMRA signal predominantly at the myocyte surface (Fig. 2, A versus B, center and right; arrows indicate localization). Finally, a volume rendering Compartmentalized myocyte G q -receptor signaling was created from a confocal stack of a myocyte labeled with AngII-TAMRA, demonstrating that the majority of the AngII-TAMRA signal localized to the myocyte surface (Fig. 2D).
Previously, a small population of AT-Rs was identified at the nucleus in AMVM based on immunochemical detection and functional assays, although the overall physiologic significance of this population of AT-Rs is still unclear (18). Further, this population of nuclear AT-Rs, which is activated by intracrine AngII, represents only a fraction of total AT-Rs (18). Our results with AngII-TAMRA indicate that the majority of ligand-accessible AT-Rs exist at the sarcolemma, and failure to detect nuclear AT-Rs here is probably because neither AngII nor AT-R antagonists readily cross the sarcolemma (19). In summary, AngII-TAMRA identified endogenous AT-Rs on the sarcolemma, and combined with results from ␣ 1 -QDot-565 identification of endogenous ␣ 1 -ARs at the nucleus, the data indicate that endogenous ␣ 1 -ARs and AT-Rs localize to different subcellular compartments in adult cardiac myocytes.
Based on prior demonstrations of G q -receptor localization and signaling at the nucleus, we sought to clarify whether PLC␤1 is localized to the nucleus in adult cardiac myocytes. First, we sought to validate potential PLC␤1 antibodies. Preferably, we would employ AMVM from PLC␤1 knockout mice, but these mice exhibit spontaneous seizures and high mortality around 3 weeks of age (24). Alternatively, we attempted to use siRNA technology to knock down PLC␤1 in AMVM but were unable to achieve significant PLC␤1 mRNA knockdown by 40 h, probably due to low turnover of PLC␤1 in cultured AMVM. Subsequently, we attempted to validate potential PLC␤1 antibodies using siRNA technology in the N38 embryonic mouse hypothalamic cell line due to PLC␤1 enrichment in the brain (25). To assess PLC␤1 knockdown, N38 cells were transfected with either PLC␤1 siRNA or scramble siRNA for 72 h. We observed knockdown of PLC␤1 mRNA ( Fig. 3C; quan-tified in D), and knockdown of PLC␤ protein was visualized by decrease in staining of the PLC␤1 antibody ( Fig. 3A; quantified in B). Finally, we stained WT AMVM with the PLC␤1 antibody, and our results indicate that PLC␤1 localizes to the sarcolemma, T-tubules, and nuclear envelope (Fig. 3E, nuclei indicated with white arrows). PIP 2 , the substrate for PLC␤1, localizes to the sarcolemma in AMVM (22,26,27), yet identifying a population of nuclear PIP 2 has been elusive because either 1) PIP 2 is not present at the nucleus or 2) the methods used to detect PIP 2 have been insufficient. To address this conundrum, we isolated AMVM nuclei using differential centrifugation and analyzed samples by MS to determine whether PIP 2 is present in nuclear membranes in AMVM. To detect the presence of PIP 2 species, we ran two blanks and two AMVM nuclear samples, adding one of each with commercially available 36:2 PIP 2 , PIP 2 (4Ј,5Ј)(18:1(9Z)/18: 1(9Z)). Using electrospray ionization-MS scan with precursor ion m/z 281 (carboxylated anion of oleic acid; C18:1), we were able identify PIP 2 36:2 species in all of the samples except for the blank (Fig. 4A, left). Spectra of the peak showing a parent ion m/z 1021 (Fig. 4A, right) confirmed the identity of the commer- A, visualization of siRNA-mediated PLC␤1 protein knockdown. N38 cells (embryonic mouse hypothalamic cells) were transfected with 80 pmol of either PLC␤1 siRNA or scrambled siRNA using Lipofectamine RNAiMAX for 72 h at 37°C. Cells were fixed and stained with a primary antibody against PLC␤1. Secondary antibody was conjugated to Alexa Fluor 594. Cells were imaged using confocal microscopy (ϫ20). Scale bar, 10 m. B, quantification of siRNA-mediated PLC␤1 protein knockdown. Fluorescence intensity was measured using FIJI (n ϭ 2 cultures; four optical areas were measured for scramble, and eight optical areas were measured for PLC␤1 siRNA). Data are represented as mean Ϯ S.E. (error bars). Data were analyzed by paired student's t test, and p Ͻ 0.05 was considered significant. C, siRNA-mediated PLC␤1 mRNA knockdown. N38 cells were transfected as in A. After 72 h, RNA was harvested using an RNAEasy kit, and PLC␤1 mRNA levels were measured by RT-PCR. Results of densitometry analysis of PLC␤1 mRNA levels are shown in D. D, quantification of siRNA-mediated PLC␤1 mRNA knockdown. The ratio of glyceraldehyde-3-phosphate dehydrogenase to PLC␤1 was quantified using densitometry (n ϭ 3 cultures). Data are represented as mean Ϯ S.E. E, endogenous PLC␤1 localizes to the sarcolemma and nuclear membrane in AMVM. WT AMVM were isolated, cultured for 24 h, fixed, permeabilized, and stained with the primary antibody against PLC␤1. Secondary antibody was conjugated to Alexa Fluor 488. Myocytes were imaged by confocal microscopy (ϫ60 oil immersion). Arrows, nuclear membrane. Images were cropped (white lines indicate cropped area) and horizontally aligned. Two representative images are shown. Scale bar, 10 m.
Similar to previous PIP 2 analyses (28), we used negative ion MS/MS to analyze the product ions of m/z 1045 (38:4 PIP 2 ), the most abundant PIP 2 species in the nuclear fraction sample without the PIP 2 addition (Fig. 4B). Ions at m/z 259, 283, 303, 339, and 419 correspond to IP, carboxylate anion fatty acyl chains of stearic acid, the arachidonic acid backbone, IP 2 , and IP 3 , all portions of the fragmented 38:4 PIP 2 spectrum. Our results indicate that PIP 2 localizes to membranes within the nuclei of AMVM, which to our knowledge is the first such demonstration. In total, our findings reveal that PLC␤1 and its substrate, PIP 2 , are found in the same subcellular compartments as both the ␣ 1 -AR (nuclear) and AT-R (sarcolemma), suggesting the potential for compartmentalized G q -receptor signaling in adult cardiac myocytes.

AT-Rs, but not ␣ 1 -ARs, activate PLC␤1 at the sarcolemma in adult cardiac myocytes
Based on the observed distinct subcellular compartmentalization of ␣ 1 -ARs and AT-Rs (Figs. 1 and 2), we reasoned that AT-Rs would activate proximal signaling at the sarcolemma and ␣ 1 -ARs at the nuclei. To test this, we measured the compartmentalization of G q -receptor activation of PLC␤1 with the PLC␤1 activity sensor GFP-C1-PLC␦-PH (GFP-PHD; Fig. 5A). GFP-PHD is composed of GFP fused to the N terminus of the pleckstrin homology domain of PLC␦1, which preferentially binds PIP 2 over other membrane phosphatidylinositols in vitro (29). In general, GFP-PHD associates with PIP 2 in membranes in the basal state, and upon G q -receptor stimulation, PLC␤1 hydrolyzes PIP 2 , and as PIP 2 is depleted, GFP-PHD dissociates from the membrane, as illustrated in Fig. 5B. In AMVM expressing GFP-PHD, the probe localized to the sarcolemma and T-tubules in the basal state, in agreement with previous reports (22). More importantly, the ␣ 1 -agonist phenylephrine (PE), in the absence or presence of prazosin, produced no change in the localization of GFP-  )). The yellow peak (not detected) represents blank sample with no PIP 2 addition, the green peak represents the blank sample with PIP 2 addition, the blue peak represents the nuclear fraction with no PIP 2 addition, and the red peak represents the nuclear fraction with PIP 2 addition (left). Electrospray ionization-MS scans of the PIP 2 -added blank sample with precursor ion m/z 281 show m/z 1021, identifying the compound as 36:2 PIP2 (right). B, identification of PIP 2 species in AMVM nuclear fractions. Product ions of m/z 1045 (38:4 PIP 2 ) of the unmixed nuclear fraction sample.

Compartmentalized myocyte G q -receptor signaling
PHD compared with vehicle ( Fig. 5C; quantified in D). Conversely, AngII induced a marked dissociation of GFP-PHD from the membrane compared with vehicle, which was blocked by the nonselective AT-R antagonist losartan, indicating a receptor-specific effect ( Fig. 5E; quantified in F). These results demonstrate that AT-Rs, but not ␣ 1 -ARs, activate PLC␤1 at the sarcolemma in adult cardiac myocytes, consistent with the subcellular compartmentalization of each receptor.

Compartmentalized myocyte G q -receptor signaling ␣ 1 -ARs, but not AT-Rs, activate PLC␤1 at the nuclear envelope in adult cardiac myocytes
Interestingly, GFP-PHD was not detected at the nuclear membrane in the basal state (Fig. 5). We suggest there are two explanations for this observation: 1) PIP 2 is not present in the nuclear membrane, despite our identification of PIP 2 in nuclear membranes (Fig. 4), or 2) GFP-PHD, which lacks an NLS, is unable to target the nucleus to bind nuclear PIP 2 . To clarify this and additionally determine whether G q receptor-mediated activation of PLC␤1 possibly occurs at the nucleus, we inserted an NLS sequence at the N terminus of GFP-PHD to create NLS-GFP-PHD (Fig. 6A). Insertion of an NLS promoted nuclear localization of GFP-PHD (Fig. 6, C and E), suggesting that PIP 2 is found in the nucleus, in agreement with detection of nuclear PIP 2 by MS (Fig. 4). NLS-GFP-PHD associates with PIP 2 in the nuclear membrane in the basal state, and upon nuclear G qreceptor stimulation, PLC␤1 hydrolyzes PIP 2 , and as PIP 2 is depleted, NLS-GFP-PHD dissociates from the nuclear membrane and moves into the nucleoplasm, as illustrated in Fig. 6B. Using the same experimental conditions as our experiments with GPF-PHD, the ␣ 1 -agonist PE induced a marked dissociation of NLS-GFP-PHD from the nuclear membrane, which was blocked by prazosin ( Fig. 6C; quantified in D). Conversely, AngII, in the absence or presence of losartan, produced little change in the localization of NLS-GFP-PHD ( Fig. 6E; quantified in F). These results demonstrate that ␣ 1 -ARs, but not AT-Rs, primarily activate PLC␤1 at the nuclear membrane in AMVM. The combined results from experiments using GFP-PHD and NLS-GFP-PHD indicate that ␣ 1 -AR-and AT-Rmediated proximal signaling is confined to distinct subcellular compartments, consistent with receptor localization.

␣ 1 -ARs, but not AT-Rs, induce IP 3 -dependent nuclear export of HDAC5 in adult cardiac myocytes
G q receptors are thought to induce hypertrophy through a mechanism known as excitation-transcription coupling (2). Conventionally, sarcolemmal G q receptor-mediated production of IP 3 elicits intranuclear calcium release from the nuclear envelope, activation of calmodulin kinase type II, and phosphorylation and nuclear export of HDAC5 (2). Nuclear compartmentalization of IP 3 -dependent calcium release allows myocytes to distinguish calcium required for contraction from calcium required for transcriptional signaling. Here, we examined whether differentially localized G q receptors would have the same effect on HDAC5 export. In AMVM expressing HDAC5-GFP, PE but not AngII induced a moderate, but signif-icant, export of HDAC5 at 30 min ( Fig. 7A; quantified in B). By 1 h, PE significantly induced HDAC5 nuclear export, whereas AngII did not ( Fig. 7C; quantified in D), consistent with previous reports for PE (30). To determine whether PE-induced HDAC5 nuclear export was IP 3 -dependent, AMVM expressing HDAC5-GFP were pretreated with the IP 3 R inhibitor 2-aminoethoxydiphenyl borate (2-APB). Pretreatment with 2-APB abolished the PE-mediated HDAC5 nuclear export ( Fig. 7E; quantified in F). Taken together, these results indicate that ␣ 1 -ARs, but not AT-Rs, activate IP 3 -dependent HDAC5 nuclear export in AMVM.

␣ 1 -ARs and AT-Rs induce unique transcriptomes in adult cardiac myocytes
Physiologically, ␣ 1 -ARs and AT-Rs have diametrically opposed effects on the heart. ␣ 1 -ARs induce physiologic hypertrophy, survival signaling, and positive inotropy and are not associated with fibrosis (3). AT-Rs induce pathologic hypertrophy, myocyte cell death, and negative inotropy and are pro-fibrotic (31,32). We hypothesized that these differences in the physiologic function of cardiac ␣ 1 -ARs and AT-Rs would be revealed in their transcriptomes and reflective of their distinct subcellular localization. To evaluate ␣ 1 -AR and AT-R transcriptomes, we treated mice with vehicle, PE (30 mg/kg/day), or AngII (0.5 mg/kg/day) continually for 3 days using osmotic minipumps, at which point we isolated cardiac myocytes and performed RNA-Seq. The doses of PE and AngII that we used are known to induce hypertrophy without a concomitant increase in blood pressure (33)(34)(35)(36). Consistent with our HDAC5 results and with the fact that HDAC5 activation relieves transcriptional repression, ␣ 1 -ARs induced a larger transcriptional response, with a total of 806 genes changed (increased or decreased) 1.7-fold versus vehicle, whereas AT-Rs induced a much smaller response, with only 173 genes changed 1.7-fold versus vehicle. 1.7-fold expression over vehicle was used as the threshold because adult cardiac myocytes are postmitotic and generally do not induce large transcriptional responses. Interestingly, between ␣ 1 -ARs and AT-Rs, 155 genes were changed by both agonists, indicating that ␣ 1 -ARs induced 651 unique genes, whereas AT-Rs induced only 18 unique genes (Fig. 8A). These results suggest that the AT-R transcriptome is largely a subset of the ␣ 1 -AR transcriptome in cardiac myocytes.
To parse the RNA-Seq results further, we initially attempted to utilize the Ingenuity Pathway Analysis (IPA) software, but the majority of the database is derived from oncogenic studies The PH domain of PLC␦ has high affinity for PIP 2 , and in the basal state, GFP-PHD binds to membrane PIP 2 , and cleavage of PIP 2 by PLC␤1 releases the probe from the membrane, decreasing membrane GFP fluorescence (33). B, GFP-PHD function. In the basal state (Basal), GFP-PHD localizes to the sarcolemma bound to PIP 2 , and upon activation of PLC␤1 (Activated), PIP 2 is cleaved, and the probe moves into the cytoplasm. C and E, ␣ 1 -ARs (C) do not activate PLC␤1 at the sarcolemma, but AT-Rs do (E). WT AMVM expressing GFP-PHD were treated with vehicle (left) or PE (20 M, 5 min, 37°C) in the absence and presence of prazosin (1 mM, 30-min pretreatment, 37°C; middle and right panels, respectively) or with AngII (100 nM, 5 min, 37°C) in the absence and presence of losartan (5 M, 30-min pretreatment, 37°C, middle and right panels, respectively). Images were cropped (white lines indicate cropped area) and horizontally aligned. 3D surface images were created with FIJI to demonstrate GFP-PHD fluorescence intensity. Scale bar, 10 m. D and F, quantification of sarcolemmal PLC␤1 activity downstream of ␣ 1 -ARs (D) and AT-Rs (F). Myocytes were classified by another investigator blinded to treatment group as responders defined by GFP-PHD movement off the sarcolemma or nonresponders defined by no movement of GFP-PHD either at baseline or following agonist treatment. Data were analyzed for cells treated with vehicle or drug (PE/AngII) by 2 , and p Ͻ 0.05 was considered significant. Data are represented as mean Ϯ S.E. (error bars): vehicle (n ϭ 30 myocytes from seven hearts); PE (n ϭ 24 myocytes from four hearts); AngII (n ϭ 29 myocytes from four hearts). ns, not significant.

Compartmentalized myocyte G q -receptor signaling
and lacks cardiac myocyte specific pathways. Thus, we derived our own analysis and sorted genes that were regulated by ␣ 1 -ARs alone, AT-Rs alone, or were in common between the two into gene ontologies corresponding to G q -receptor biology: hypertrophy, survival signaling, inotropy, and fibrosis ( Fig. 8B and Tables S1-S4). ␣ 1 -ARs most robustly altered genes in all Compartmentalized myocyte G q -receptor signaling categories as compared with common genes and AT-R-only genes. Whereas it is surprising that ␣ 1 -ARs altered more fibrotic genes than AT-Rs, these genes are not classically associated with alterations in the extracellular matrix leading to fibrosis (Table S4). Finally, principle component analysis was performed to determine the degree of difference between vehicle-, AngII-, and PE-treated samples (Fig. 8C). Principle component 1 (PC1) accounted for 66% of the variance and aligned with the ␣ 1 -AR transcriptomes, whereas PC2 accounted for 22% of the variance and aligned with the AT-R transcriptome. The AngII-treated samples also closely grouped with the vehicle-treated samples, indicating that AT-Rs do not induce a highly distinct transcriptome from control, consistent with our gene ontology results (Fig. 8, A and B) The top 25 genes determining PC1 and PC2 are presented in Fig. 8D. Taken together, ␣ 1 -ARs robustly activate transcription in adult cardiac myocytes, whereas AT-Rs minimally activate transcription. These results identify distinct differences in the transcriptomes induced by ␣ 1 -ARs and AT-Rs that align with their distinct subcellular localization and activation of proximal signaling to produce differential activation of nuclear hypertrophic signaling pathways.

Discussion
Here, we identified an entirely novel mechanistic explanation for unique G q -receptor function in adult cardiac myocytes predicated upon subcellular compartmentalization of proximal G q -receptor signaling and propose a novel model of excitationtranscription coupling. We found that ␣ 1 -ARs localize to the nucleus and induce intranuclear activation of PLC␤1, stimulate IP 3 -dependent nuclear export of HDAC5, and activate a robust and unique transcriptome associated with hypertrophic, survival, inotropic, and (anti-)fibrotic gene programs. Conversely, we observed that AT-Rs primarily localize to and activate PLC␤1 at the sarcolemma but have little effect on nuclear export of HDAC5 and induce a small transcriptome that is a subset of the ␣ 1 -transcriptome. More importantly, these findings are consistent with our hypothesis that G q -receptor localization dictates function by showing that compartmentalization of proximal G q -signaling is correlated with phenotypic outcome in adult cardiac myocytes.
The excitation-transcription model of G q receptormediated hypertrophic signaling in adult cardiac myocytes proposes that sarcolemmal G q receptors induce IP 3 production and IP 3 -sensitive intranuclear calcium release from perinuclear calcium stores to activate calmodulin kinase, phosphorylate and induce nuclear export of HDAC5, and thereby activate transcription (Fig. 9A). Whereas the current model of excitation-transcription coupling suggests that G q receptors induce IP 3 production at the sarcolemma leading to activation of IP 3 -dependent calcium release at the nucleus to induce HDAC5 export and promote gene transcription, it fails to explain how G q receptors might produce unique physiological function in cardiac myocytes.
Both ET-Rs and insulin-like growth factor receptors conform to the traditional excitation-transcription model (2,14,37). Our data indicate that ␣ 1 -ARs might support this model as well, but interestingly, AT-Rs do not. Although we observed AT-R-mediated activation of PLC␤1, we failed to detect AT-R-mediated nuclear export of HDAC5 and found a much smaller transcriptional response. One interpretation of this result is that close proximity to the nucleus is required for G q receptor-mediated activation of IP 3 -dependent hypertrophic signaling. In support of this interpretation, ␣ 1 -ARs localize to the inner nuclear membrane (11), and ET-Rs and insulin-like growth factor receptors localize to the bottom of T-tubules in close apposition to the nucleus in adult cardiac myocytes (37,38). Further, the failure of AT-Rs to induce nuclear export of HDAC5 suggests that AT-R-mediated activation of PLC␤1 at the sarcolemma either fails to generate enough IP 3 to reach the nucleus or that IP 3 is degraded before it reaches the nucleus. The potential degradation of IP 3 before reaching the nucleus might be analogous to the compartmentalization of cAMP signaling in cardiac myocytes (39).
Here, we propose a new model of excitation-transcription coupling that is based on compartmentalization of G q receptors that explains distinct physiologic function of G q receptors in adult cardiac myocytes. Our model suggests that G q receptorinduced IP 3 production is compartmentalized and that IP 3 produced inside the nucleus (or possibly in close proximity to the nucleus) induces HDAC5 export to promote gene transcription, whereas IP 3 produced at a distance from the nucleus (at the sarcolemma) has a different and smaller effect on transcriptional regulation (Fig. 9B). In summary, we suggest that G qreceptor compartmentalization has a large influence on the transcriptomes induced by differentially localized G q receptors, illustrating the fundamental physiologic importance of G q -receptor compartmentalization.
With regard to our model of nuclear ␣ 1 -induced HDAC5 export, previous work indicated that HDAC5 export downstream of ␣ 1 -ARs occurs mainly through activation of protein kinase D, and our results do not exclude activation of this signaling cascade (40). Nevertheless, in agreement with our Figure 6. ␣ 1 -ARs and, to a lesser extent, AT-Rs activate PLC␤1 at the nuclear membrane in adult cardiac myocytes. A, schematic representation of NLS-GFP-PHD. NLS-GFP-PHD consists of GFP-PHD tagged with an N-terminal NLS from the SV40 large T-antigen to promote localization of the probe to the nucleus. NLS-GFP-PHD is thought to operate in the same manner as GFP-PHD (Fig. 3), but at the nuclear membrane (33). B, GFP-PHD function. In the basal state (Basal), GFP-PHD-NLS localizes primarily to the nuclear membrane bound to PIP 2 , and upon activation of PLC␤1 (Activated), PIP 2 is cleaved, and the probe moves into the nucleoplasm. C and E, ␣ 1 -ARs (C) activate PLC␤1 at the nuclear membrane, but AT-Rs do not (E). WT AMVM expressing NLS-GFP-PHD were treated with either vehicle (left panels) or PE (20 M, 5 min, 37°C) in the absence or presence of prazosin (1 mM, 30-min pretreatment, 37°C; middle and right panels, respectively) or AngII (100 nM, 5 min, 37°C) in the absence and presence of losartan (5 M, 30-min pretreatment, 37°C; middle and right panels, respectively). Arrows indicate nuclear membrane activation of PLC␤1 by ␣ 1 -ARs. Images were cropped (white lines indicate cropped area) and horizontally aligned. 3D surface images were created with FIJI to demonstrate NLS-GFP-PHD fluorescence intensity. Scale bar, 10 m. D and F, quantification of nuclear PLC␤1 activity downstream of ␣ 1 -ARs (D) and AT-Rs (F). Myocytes were classified by another investigator blinded to treatment group as responders defined by NLS-GFP-PHD movement off the nuclear membrane or nonresponders defined by no movement of NLS-GFP-PHD either at baseline or following agonist treatment. Data were analyzed for cells treated with vehicle or drug (PE/AngII) by 2 , and p Ͻ 0.05 was considered significant. Data are represented as mean Ϯ S.E. (error bars): vehicle (n ϭ 30 myocytes from seven hearts); PE (n ϭ 14 myocytes from four hearts); AngII (n ϭ 16 myocytes from four hearts). ns, not significant.

Compartmentalized myocyte G q -receptor signaling
results, Luo et al. (41) demonstrated that ␣ 1 -ARs activate IP 3dependent nuclear calcium transients in cardiac myocytes, consistent with our finding that ␣ 1 -ARs induce IP 3 -sensitive HDAC nuclear export. At this time, the discrepancy between these findings is not entirely clear.
The consensus view of cardiac G q -receptor function has been that G q receptors mediate pathologic remodeling, promoting maladaptive hypertrophy, myocyte cell death, and negative inotropic responses (4). However, in clinical trials of hypertension (Antihypertensive and Lipid-Lowering Treatment to Prevent Figure 7. ␣ 1 -ARS induce HDAC5 export, whereas AT-Rs do not. A and C, ␣ 1 -ARs activate HDAC5 nuclear export, whereas AT-Rs do not. WT AMVM expressing HDAC5-GFP were treated with either vehicle (left panels), 20 M PE (middle panels; insets of zoomed-in nuclei), or 100 nM AngII (right panels) for either 30 min (A) or 1 h (C) at 37°C. Scale bar, 10 m. B and D, quantification of HDAC5 export induced by ␣ 1 -ARs and AT-Rs. Cytoplasmic and nuclear GFP fluorescence was quantified using FIJI, and the ratio of cytoplasmic to nuclear GFP was plotted as an indication of HDAC5 nuclear export at 30 min (B) and 1 h (D). Data are represented as mean Ϯ S.E. (error bars) Data were analyzed by one-way ANOVA, and p Ͻ 0.05 was considered significant: vehicle (n ϭ 12 myocytes from six hearts); PE (n ϭ 12 myocytes from six hearts); AngII (n ϭ 12 myocytes from six hearts). E, HDAC5 export is inhibited in the presence of IP 3 R inhibitor, 2-APB. WT AMVM expressing HDAC5-GFP were treated with 2-APB (2 M, 30 min at 37°C) before treatment with PE (1 h at 37°C; middle). Scale bar, 10 m. F, quantification of HDAC5 export in the presence of IP 3 R inhibitor, 2-APB. The ratio of cytoplasmic to nuclear GFP was calculated and plotted. Data are represented as mean Ϯ S.E. (error bars). Data were analyzed by one-way ANOVA, and p Ͻ 0.05 was considered significant: vehicle (n ϭ 12 myocytes from six hearts); 2-APB (n ϭ 12 myocytes from six hearts); PE (n ϭ 12 myocytes from six hearts); PE ϩ 2-ABP (n ϭ 12 myocytes from six hearts).

Compartmentalized myocyte G q -receptor signaling
Heart Attack Trial (ALLHAT)) and HF (Vasodilator Heart Failure Trial (V-HeFT)), ␣ 1 -AR antagonists worsened outcomes (42,43). Further, our studies indicate that ␣ 1 -ARs are cardioprotective, promoting adaptive or physiologic hypertrophy, prevention of cell death, and positive inotropic effects identifying a mechanistic basis for the negative results of ␣ 1 -AR antagonists in clinical trials (3). In short, the cardioprotective nature of ␣ 1 -ARs stands in stark contrast to the consensus view of maladaptive G q -receptor function. Here, our data suggest that compartmentalization of G q receptors could explain differences in G q -receptor function. We demonstrated that although both ␣ 1 -ARs and AT-Rs activate PLC␤1, they do so in different subcellular compartments, which has a profound effect on activation of intranuclear hypertrophic signaling and transcrip- WT adult male mice were treated for 3 days with minipumps perfusing vehicle, PE, or AngII. Myocytes were then isolated, RNA was isolated, and RNA-Seq was performed. The gene list generated by RNA-Seq was filtered based on a minimum 1.7ϫ absolute -fold change and false discovery rate-corrected p Ͻ 0.05. PE treatment altered 801 transcripts (655 unique) by at least 1.7-fold, whereas AngII only altered 173 (18 unique), with 155 common between the two treatments. B, gene ontology analysis of ␣ 1 -AR-and AT-R-induced transcriptomes. G q receptors regulate hypertrophy, survival signaling, inotropy, and fibrosis. Therefore, using nonoverlapping search terms unique to these different biologic functions, genes were sorted into subclasses of hypertrophy, survival signaling, inotropy, and fibrosis. The total number of genes altered (increased and decreased) in common or by PE or AngII alone are plotted in the graph. Search terms used to sort genes associated with hypertrophy were as follows: NF-B, GATA4, MEF2, NFAT, ANF, TGF, hypertrophy, myosin heavy chain, MHC, c-myc, c-fos, cell growth, and natriuretic. Search terms for survival signaling were ERK, MAPK, apoptotic, apoptosis, survival, and cell death. Search terms for inotropy were calcium, contraction, sarcoplasmic reticulum, troponin, myosin, actin, ryanodine, protein kinase C, and sarcomere. Search terms for fibrosis were collagen, matrix metalloprotease, fibrosis, fibroblast, extracellular matrix, and matrix. C, principle component analysis of ␣ 1 -AR-and AT-Rinduced transcriptomes. Principle component analysis of fragments per kilobase of exon per million reads mapped (FPKM). PC1 determined 66% of all variance between samples. PC2 determined 22% of all variance between samples. D, top 25 genes determining PC1 and PC2. Top 25 genes for PC1 and PC2 were determined using FPKM loadings.

Compartmentalized myocyte G q -receptor signaling
tional activation. Therefore, we suggest that nuclear G q -receptor signaling, typified by ␣ 1 -ARs, is cardioprotective. Aside from ␣ 1 -ARs, a small population of functional ET-Rs and AT-Rs localize to the nucleus as well (18,21), although their physiologic significance is unclear. Conversely, we observed AT-Rs primarily at the sarcolemma, but not in close proximity All G␣ q -receptors localize to the sarcolemma in adult cardiac myocytes. Upon ligand binding, sarcolemmal PLC␤1 is activated and cleaves PIP 2 into DAG and IP 3 . DAG goes on to activate protein kinase C isoforms (PKC) and induce contraction. IP 3 traverses the myocyte and binds to the IP 3 R on the inner nuclear membrane, inducing intranuclear calcium release. Calcium activates calmodulin (CaM) and calcium-calmodulin dependent protein kinase II (CaMKII), which phosphorylates HDAC5. HDAC5 phosphorylation triggers HDAC5 nuclear export and derepression of transcription. B, updated model of excitation-transcription coupling in adult cardiac myocytes. ␣ 1 -ARs (nuclear) and AT-Rs (sarcolemmal) are differentially localized in adult cardiac myocytes ( Figs. 1 and 2). Upon ligand binding to either ␣ 1 -ARs or AT-Rs, PLC␤1 is activated either at the nucleus (␣ 1 -AR; Fig. 6) or at the sarcolemma (AT-R, Fig. 5). AT-R-induced PLC␤1 activation at the sarcolemma fails to induce HDAC5 nuclear export, whereas ␣ 1 -AR-induced PLC␤1 activation at the nucleus induces HDAC5 nuclear export. Furthermore, ␣ 1 -AR-induced HDAC5 nuclear export is IP 3 -dependent (blocked by 2-APB) (Fig. 7). Consistent with ␣ 1 -induced, IP 3 -dependent HDAC5 nuclear export, ␣ 1 -ARs induce a robust transcriptional response, whereas AT-Rs, which fail to induce HDAC5 export, produce a transcriptional response that is largely a subset of ␣ 1 -AR-induced transcriptional responses (potentially suggesting a different non-HDAC dependent mechanism of transcription) (Fig. 8).

Compartmentalized myocyte G q -receptor signaling
to the nucleus, which might suggest that G q -receptor signaling at the sarcolemma is pathologic, which is supported by studies with AT-Rs (31,32). In support of this concept, adenoviral mediated expression of the PLC␤1b at the sarcolemma induces contractile dysfunction (44). In summary, our hypothetical model of compartmentalized G q -receptor signaling, where nuclear G q -receptor signaling is cardioprotective, suggests a more nuanced view of G q -receptor function in cardiac myocytes.
The concept of GPCR compartmentalization in cardiac myocytes is not without precedence. In co-cultures of sympathetic ganglionic neurons and neonatal rat cardiac myocytes, ␤ 1 -ARs localize to regions of axonal contact rich in SAP97, AKAP97, catenins, and cadherins, whereas ␤ 2 -ARs are excluded from these domains (45). In adult cardiac myocytes, ␤ 1 -ARs are distributed over the entire sarcolemma, whereas ␤ 2 -ARs are restricted deep within T-tubules (46). Finally, platelet-derived growth factor receptors, which also signal through G s , localize to caveolae in cardiac myocytes and do not induce an inotropic response (47). These examples demonstrate that cardiac myocytes compartmentalize G s -mediated GPCR signaling, analogous to our findings with G q receptors. In conclusion, our findings support a model of compartmentalized G q -receptor signaling in adult cardiac myocytes and suggest a revision of the classic model of excitation-transcription coupling. Our new model, largely based on our identification of cardioprotective nuclear ␣ 1 -AR signaling, provides a plausible mechanistic basis to explain the unique function of cardiac G q receptors. Additionally, this model suggests a reexamination of the classic paradigm of maladaptive G q signaling in cardiac myocytes in favor of a more nuanced view of compartmentalized G q -receptor signaling.

Experimental models: Mice
In this study, male and female C57BL/6J mice (10 -15 weeks of age) were used for primary adult cardiac myocytes. Male FVB/NJ mice (10 -11 weeks of age) were used for infusion of agonist to measure G q receptor-induced hypertrophic transcriptional responses. All animals were sourced from Jackson Laboratories. The use of all animals in this study conformed to the United States Public Health Service Guide for Care and Use of Laboratory Animals and was approved by the University of Minnesota institutional animal care and use committee.

Method details
Isolation and culture of AMVM-This was carried out as described previously (48).
Isolation and labeling of nuclei from AMVM-This was carried out as described previously (49). Labeling is described in the supporting information.
Synthesis of ␣ 1 -QDot-565-The synthesis of the ␣ 1 -QDot-565 was performed as described (17) with a few modifications that are described in the supporting information.
Localization of ␣ 1 -ARs, AT-Rs, and PLC␤1 in AMVM-Methods used to define the localization of ␣ 1 -ARs, AT-Rs, and PLC␤1 are described in the supporting information.
Identification of nuclear PIP 2 -PIP 2 extraction was carried out as described (52) with some modifications. The electrospray mass spectrum of each sample was analyzed on a hybrid quadrupole triple ion trap mass spectrometer (Triple TOF 5600, AB Sciex Instrument). All scans were performed in negative ionization mode and a mass-to-charge (m/z) range from 50 to 1200. PIP 2 (4Ј,5Ј)(18:1(9Z)/18:1(9Z)) was obtained from Avanti Polar Lipids.

Quantification and statistical analysis
The methods for the analysis of phospholipase C␤1 activity, HDAC5 nuclear export, and hypertrophic transcriptional profiles adenoviruses are described in the supporting information. Details and methods used for statistical analysis can be found in the supporting information.