Interaction of G-Protein βγ Complex with Chromatin Modulates GPCR-Dependent Gene Regulation

Heterotrimeric G-protein signal transduction initiated by G-protein-coupled receptors (GPCRs) in the plasma membrane is thought to propagate through protein-protein interactions of subunits, Gα and Gβγ in the cytosol. In this study, we show novel nuclear functions of Gβγ through demonstrating interaction of Gβ2 with integral components of chromatin and effects of Gβ2 depletion on global gene expression. Agonist activation of several GPCRs including the angiotensin II type 1 receptor specifically augmented Gβ2 levels in the nucleus and Gβ2 interacted with specific nucleosome core histones and transcriptional modulators. Depletion of Gβ2 repressed the basal and angiotensin II-dependent transcriptional activities of myocyte enhancer factor 2. Gβ2 interacted with a sequence motif that was present in several transcription factors, whose genome-wide binding accounted for the Gβ2-dependent regulation of approximately 2% genes. These findings suggest a wide-ranging mechanism by which direct interaction of Gβγ with specific chromatin bound transcription factors regulates functional gene networks in response to GPCR activation in cells.


Introduction
The Gb and Gc subunits form a functionally inseparable Gbc complex that generate the quiescent heterotrimeric G-proteins by associating with Ga-GDP. Current models show that G-protein activation by G-protein coupled receptors (GPCRs) occur at the plasma membrane (PM). Second messengers or protein-protein interactions leading to spatio-temporal propagation of signals initiated by Ga and Gbc to the nucleus occurs in the cytoplasm, however translocation of G-protein subunits to nucleus is not frequently considered a possibility [1]. This view is changing due to the discovery of the shuttling of Ga and Gbc subunits from the PM to cell organelles, such as the Golgi, mitochondria, endosomes, and occasionally, the nucleus [2,3]. It is possible therefore, that Ga or Gbc complex translocates to nucleus and participate in gene regulation.
Gene regulation through G-protein signaling is crucial to human adaptation and survival which reflects the enormous success of therapeutics targeting GPCRs, the largest family of receptors encoded by the human genome. The finely tuned expression of an appropriate set of genes in a cell depends on multiple transcription factors (TFs) and transcriptional co-activators. GPCRs enhance gene transcription by facilitating the interaction of histone acetyl transferases (HATs), such as p300/ CBP, to TFs on chromatin [4]. Alternatively, recruitment of histone deacetylases (HDACs) to chromatin-bound TFs, such as myocyte enhancer factor 2A (MEF2A), represses transcription, and the repression is relieved by GPCR signals [5]. Nuclear localization of b-arrestins [6], GRK5 [7] and RGS proteins [8] is reported which suggests that these proteins recruited into the nucleus upon ligand activation of GPCRs may participate in the epigenetic processes that are essential for the functioning of cells. Whether Ga or Gbc which are the primary transducers of GPCR signals, regularly enter the nucleus and directly participate in GPCR-coordinated transcriptional response remains unclear. Reports of Gb 1 or Gb 2 association with the glucocorticoid receptor [9], Gb 1 c 2 association with HDAC5 [10,11], Gb 5 association with the nuclear shuttling of the R7 family of RGS proteins [8] and Gbc 5 association with the adipocyte enhancer binding protein [12] suggest a potential broad role of Gbc in gene regulation. Therefore, we hypothesized that agonist activation of a typical GPCR such as the angiotensin II type 1 receptor (AT 1 R), changes the composition of chromatin-associated proteins which may include changes in the levels of specific G-protein subunits.
An unbiased high-throughput mass spectrometry analysis of the nuclear proteome upon activation of a GPCR led us to discover the interactions of Gb 2 c 12 with chromatin. We found that the level of Gb 2 increased in the nucleus upon activation of diverse GPCRs and that Gb 2 was essential for agonist-induced MEF2A function. Gb 2 interacted with a sequence motif present in several TFs, and this interaction accounted for the coordinated gene regulatory function of Gbc.

Nuclear and cytosolic fractionation
The nucleus and cytosol were isolated using the NUC101 nuclei isolation kit as detailed by the manufacturer (Sigma-Aldrich). The nuclear fractions were stained with DAPI, and subsequent visualization was performed using confocal microscopy to check for the integrity of nuclei. Nuclear protein was extracted using Benzonase (10 units/ml at 37uC for 60 min), which digests the nucleic acids without denaturing the proteins (chromatin proteins). The pellet was further extracted to isolate the tightly bound proteins using 0.45 N sulfuric acid (acid fractions). The purity of the fractions was determined by immunoblotting for specific cellular compartment markers, histones H1 and H2A (nuclear) and Gqa (cytosolic).

Site directed mutagenesis and plasmid construction
The amino terminus HA-tagged rat AT 1 R [13] under the control of the human cytomegalovirus (CMV) promoter was generated in the pcDNA3 plasmid. FLAG-Gb 2 was subcloned into pBudE4.1 under the EF1 promoter. Nested primers were designed to delete each of the seven WD repeats in the FLAG-Gb 2 construct. All of the subcloned plasmid constructs and the WD40 mutants were verified by DNA sequencing.
Transfection and generation of AT 1 R -expressing stable hek-293 cells Routine transient transfection of HEK-293 cells was performed with FUGENE 6 TM per the manufacturer's recommendations. The cell line stably expressing HA-AT 1 R was established by clonal isolation using geneticin (600 mg/ml) selection.

LC-tandem mass spectrometry and protein identification
For Gel C analyses [14], the gels are run to attain 50% resolution of the electrophoresed proteins. The gels were cut into three regions, and each of these regions was further cut into five equal parts and digested with trypsin. The peptides were extracted and concentrated, and the digest was analyzed by LC-tandem MS [15]. The proteins contained in the nuclear fractions were identified using a shotgun sequencing approach [16]. Relative quantitation was determined using a spectrum counting approach. The MS results were also examined by plotting mass chromatograms for the respective peptides. Data were also searched using SEQUEST (ThermoFisher, San Jose, CA) with mass tolerances set at 3.0 Da for peptides and 2.0 for fragment ions using the standard variable oxidation of methionine (+16 Da) and carbamidomethylation on cysteine (+57 Da) as fixed modification. The MASCOT program (www.matrixscience.com) was used to compare all of the CID spectra with the NCBI non-redundant database and to identify the protein. Matching peptides were verified by manual interpretation.

Isolation of ventricular myocytes from adult C57BL6 mice
The isolation procedure for ventricular cardiac myocytes from adult C57BL6 mice has been reported in detail [17]. Handling of animals used for myocyte preparation is approved by IACUC using standard protocol recommendation. Cardiac myocytes were enzymatically dispersed via Langendorff perfusion of mouse hearts [18]. Following isolation, cells were treated with 1 mM AngII for 30 min and then fixed with 3% paraformaldehyde and subjected to immunocytochemical analysis.

Measurements of [Ca 2+ ] flux
Single ventricular myocytes were incubated with 1 mM fura-2 acetoxymethyl ester (Molecular Probes) for 10 min at room temperature in the dark. [Ca 2+ ] i signaling was measured using a dual excitation spectrofluorometer (Deltascan RFK6002, Photon Technology International) at excitation wavelengths of 340 and 380 nm and an emission wavelength of 510 nm as previously described [19]. Steady-state [Ca 2+ ] i transient signals were recorded at a pacing frequency of 0.5 Hz in the absence of AngII. The stimulation was stopped and then followed by 1 mM AngII treatment. The resulting AngII-induced [Ca 2+ ] i signal was recorded as a qualitative index of the initial sarcoplasmic reticulum (SR) [Ca 2+ ] i load in the cardiomyocytes.

Immunocytochemistry and confocal microscopy
Immunolocalization using confocal microscopy were performed essentially as described previously [20]. For 3D-visualization of confocal image stacks, the confocal image slices of myocytes labeled with DAPI and FITC (0.13 mm60.13 mm60.04 mm resolution in X, Y, and Z directions, respectively) were imported into Image-Pro 6.1 (Media Cybernetics, Silver Springs, MD) as a multi-plane sequence and subsequently split into blue and green channels. Customized sequence segmentation scripts were then applied to the blue (DAPI) channel to threshold and binarize each slice. The binarized image stack was then multiplied with the green channel stack (plane-by-plane) to extract green fluorescence localized to the nucleus. Both binarized nuclear slices and their corresponding green fluorescent slices were exported into Micro-View (GE Healthcare, Piscataway, NJ), reconstructed into 3D volumes (Z-dimension resolution was increased five times to improve definition of flattened nuclei), and rendered as isosurfaces. Lastly, these iso-surfaces (DAPI and FITC) were merged together, and the opacity of the nucleus was adjusted to allow visualization of the underlying protein.

Reporter assays
The MEF2-luciferase assay (Promega) was performed as recommended by the manufacturer. Briefly, the MEF2-luciferase plasmid (1 mg) was transfected into AT 1 R-expressing cells in the presence or absence of Gb 2 to evaluate the role of Gb 2 in modulating MEF2A activity. The WD40 repeats of the N-terminal FLAG-tagged Gb 2 (FLAG-Gb 2 ) construct were sequentially deleted to create FLAG-Gb 2 DWD(1-7) mutants. The HEK-AT 1 R cells were then transfected with FLAG-Gb 2 DWD mutants, the MEF2-luciferase plasmid and 0.15 mg of the bGal plasmid (transfection control).

Immunoprecipitations with FLAG-Gb 2 and WD40 repeat mutants
The role of WD40 repeats in Gb 2 involved in the interaction with MEF2A was performed and quantified (Kodak Imager ID 3.6) as previously described [21].

Deacetylase assays
The HDAC activity (UPSTATE) was performed per the manufacturer's guidelines. For deacetylase assays, the nuclear and cytosolic fractions were obtained from AT 1 R and AT 1 R-Gb2i cells (+/230 min of 1 mM AngII). Actinin (1 mg anti-actinin-1 antibody)-associated deacetylase activity was measured in 100 mg each of the cytosolic and nuclear fractions. The samples were incubated with gentle mixing at 4uC overnight. Immunoprecipitates were collected via centrifugation and washed twice with 1 ml of ice-cold phosphate buffered saline (PBS). The resin was assayed for deacetylase activity.

RNAi-mediated knockdown of Gb 2
A DNA vector-based siRNA was designed to stably knockdown the expression of Gb 2 (gene name: GNB2) in HEK-293 cells [21]. The target sequence, ACTGGGTACCTGTCGTGTT [21], and the scrambled sequence, CGGTGTTCTACGTGGCTAT, were cloned into the pRNATU6.1/hygro plasmid under the control of the U6 promoter and a hygromycin selection marker. The cells were also co-transfected with the HA-AT 1 R-expressing plasmid that contained a neomycin selection marker. Selected clones were maintained in media containing hygromycin (100 mg/ml) and geneticin (600 mg/ml).

Microarray analysis
HEK-293 (AT 1 R; 2/+AngII and AT 1 R-Gb 2 i; 2/+AngII) cells were harvested under RNase-free conditions, and RNA was isolated using the RNeasy kit (Qiagen). RNA-based probe synthesis and hybridization were performed by the Gene Expression Array Core Facility at Case Western Reserve University (www.geacf.net) as described previously [21] using high-density oligonucleotide HG-U133 Plus 2 arrays (Affymetrix, Inc.; Santa Clara, CA). Gene expression changes predicted in the Gb2i cells for proteins examined in this study were validated by western blot analysis of Gb2i cell lysates.
Functional and network analyses of gene expression data were performed using Ingenuity software (Ingenuity Systems, http:// www.ingenuity.com). The score assigned to any given gene network takes into account the total number of molecules in the data set, the size of the network, and the number of ''network eligible'' genes/molecules in the data set. The network score is based on the hypergeometric distribution and is calculated with the right-tailed Fisher's exact test. The network score is the negative log of this P-value. To identify the genes regulated by the 4 key TFs, MEF2A, NFAT, STAT1, and STAT3, we used the 'Build Networks -Expand by one group interaction' algorithm in MetaCore TM (www.genego.com). Only transcriptional regulation interaction was considered for this analysis, and the list of 658 genes (Gb2-regulated genes) was overlaid onto these networks. Only those genes that were transcriptionally regulated by one or more of these TFs as well as differentially regulated in our datasets are shown in the figure.

Molecular modeling
Reference 3D structures of Gb1c2 [PDB: 1GP2] were retrieved from the Protein Data Bank (PDB) (http://www.rcsb.org/pdb/ home/home.do) through the NCBI website (http://www.ncbi. nlm.nih.gov). For homology modeling, target and template sequences were aligned using CLUSTAL X. The alignment was then submitted electronically to the Swiss Model server (http:// www.expasy.org), which generates the homology model based on the template structure. Energy computations were performed in vacuo using the GROMOS96 implementation of the Swiss PDB Viewer (SPDBV) program (Swiss Institute of Bioinformatics). Energy minimization was carried out by 20 cycles of steepest descent, and minimization was stopped when the D energy was below 0.05 kJ/mol, as previously described. Hydrogens were added using VEGA ZZ (University of Milan, Italy; (http://www. ddl.unimi.it/vega/index2.htm). The model was then submitted for Ramachandran analysis. The structures were visualized using the PYMOL program (The PyMOL Molecular Graphics System, DeLano Scientific, Palo Alto, 2002).

Statistical analysis
All experiments were performed three or more times. For image analysis, approximately 50-100 cells were analyzed in each set, and representative images are shown. All data are expressed as the mean 6 SEM of at least three independent experiments. Each experiment was performed in triplicate unless otherwise indicated. Data were analyzed using an unpaired Student's t-test (P,0.05) using GraphPad Prism 4 software.

Gb 2 and Gc 12 traffic into the nucleus upon GPCR activation
To test the hypothesis that agonist activation of GPCRs changes the composition of chromatin-associated proteins, we examined angiotensin II type 1 receptor (AT 1 R), which is a peptide hormone GPCR. Mechanisms of AT 1 R signaling have been extensively studied in the attempt to improve AT 1 R-targeted therapies for hypertension, cardiac hypertrophy and end organ damage. Agonist (e.g., AngII)-mediated activation of AT 1 R has been reported to induce the nuclear mobilization of TFs, including GATA binding protein (GATA4), nuclear factor of activated T cells (NFAT), signal transducer and activator of transcription 3 (STAT3), nuclear factor-kappaB (NF-kB), extracellular signalregulated kinases (ERK1/2), protein kinase C and HDAC5, during the progression of cardiovascular diseases [22][23][24][25][26][27]. We used a human embryonic kidney (HEK) 293 cell clone, HEK-AT 1 R, as a surrogate model system to identify the proteins that mobilize to the nucleus and associate with chromatin ( Fig. 1). In HEK-AT 1 R cells, AngII induced G q/11 -PLC calcium signaling and pERK1/2 signaling (detailed in Fig. S1). Expression of early growth response genes was subsequently induced a result that was also found in neonatal cardiomyocytes stimulated with AngII [26,28]. The AngII effects were blocked by treatment with the AT 1 R-selective antagonist, losartan. To prepare the nuclear proteome, AT 1 R was activated for 30 min, which was determined as the time when AngII-induced pERK1/2 association with chromatin was maximal. The nuclear and cytosol subcellular fractions isolated were well separated, as indicated by the absence of Ga subunits in the chromatin preparation and the absence of the histone, H2A, in the cytosolic preparation (Fig. S2). When the nuclear proteome was queried for collision-induced dissociation (CID) spectra of peptides corresponding to abundant plasma membrane and cytosolic proteins, none corresponding to integrins, Ga, GAPDH and cytochrome b5 were detected, which further confirmed the authenticity of our nuclear proteome preparation.
The largest groups of proteins found were RNA-binding proteins, heterogeneous nuclear ribonucleoproteins, splicing factors, nucleolar proteins, ribosomal RNA-binding proteins and the proteins involved either directly in DNA binding or in cell cycle and gene regulation (Fig. 1b, Fig. S3). Most of these molecules are established nuclear proteins and/or shuttling proteins that contain a nuclear localization signal (NLS). Many signaling proteins (13.7%) without an obvious NLS present in nuclear proteome included Gb 2 , Gc 12 , and a-actinin-4 (Figs. 1b, S3, S4). The CID spectrum of the signature peptides, LLVSASQDGK for the Gb 2 isoform (Fig. 1c) and TASTNNIAQAR for the Gc 12 isoform ( Fig. 1d; see Table S1 for peptide coverage), were identified by SEQUEST (www.proteomicswiki/index.php/SEQUEST). The Gb and Gc subunits form an obligate functional monomer and translocate together. The nuclear partition coefficient estimated by WoLFPSORT (http://wolfpsort.org/) [29] for the Gb 2 c 12 complex was 20.13 (equivalent to HDAC5, which is known to localize in both the nucleus and cytoplasm), indicating the potential for Gb 2 c 12 to enter the nucleus upon GPCR activation. The nuclear partition coefficient of Gc 12 alone was 20.13. The nuclear partition coefficient of Gb 2 alone was identical to aactinin-4 (20.47), which is also known to localize both in the nucleus and cytoplasm [30], suggesting that Gb 2 most likely enters the nucleus upon GPCR activation.

Gb 2 content in the nuclear proteome
The label-free approach of spectrum counting [31] estimated a significant increase in Gb 2 translocation into the nucleus upon AT 1 R activation (Fig. 2a). The abundance of peptides from spiked-in trypsin was comparable in AngII treated HEK-293 and HEK-AT 1 R samples. The relative abundance of the Gb 2 peptide, LLVSASQDGK, in the chromatin of AngII activated HEK-AT 1 R sample (NL1.3E6) increased <3.1 fold when compared to HEK-293 sample. This fold increase was independently corroborated through additional analysis (Fig. S5).
A variety of prohypertrophic agonists, including AngII, enhanced the nuclear translocation of Gb 2 (.1.7 fold) in human aortic smooth muscle (HASM) cells as validated by western blotting (Fig. 2b). The adrenergic receptor agonist (dobutamine) coupled to Ga s was as effective (Fig. 2b, see schematic) in the nuclear mobilization of Gb 2 as the Ga q -activating agonists (AngII and 5-HT). The activation of different GPCRs may release different Gbc isoforms, which may participate in chromatin functions with different efficacies. We envision Gb 2 as a direct mediator of the nuclear effects of activated GPCRs.

Agonist-induced nuclear translocation of Gb 2
Indirect immunofluorescence staining demonstrated an increase in Gb 2 in the nucleus of HEK-AT 1 R and HASM cells and neonatal rat ventricular myocytes (NRVMs) upon treatment with AngII (Fig. 3a). Treatment with losartan prevented the increase in Gb 2 in the nucleus (Fig. S5a). To confirm that Gb 2 translocation was physiologically relevant in cells, we isolated adult mouse ventricular myocytes (AMVMs). Pacing and AngII treatment elicited calcium transients in AMVMs after isolation (Fig. 3b). AngII treatment stimulated the translocation of Gb 2 (Fig. 3c) from the cytoplasm into the nucleus of AMVMs (<4.0-fold). Threedimensional (3-D) image reconstruction of myocyte nuclei (Fig. 3d) showed the association of Gb 2 with chromatin. Thus, using different analytical methods, a 2.5-to 4.5-fold increase in the nuclear translocation of Gb 2 was observed in different types of cells upon AngII treatment (Fig. S5b).

Association of Gb 2 with components of chromatin
Our nuclear proteome preparation was enriched in protein complexes that were associated with an AngII-activated state of the genome. In this state, Gb 2 interacted with the AngIIresponsive TF, MEF2A, the core histones, H2B and H4, the histone-modifying enzyme, HDAC5, and the calcium binding scaffold protein, a-actinin-4 (Fig. 4a). Gb 2 did not associate with histones H1, H2A and H3 or with pERK1/2. Gb 2 association with MEF2A and histones H2B and H4 suggested that Gb 2 interacted with nucleosomes at the promoters of MEF2 regulated genes. The Gb 2 interactions with a-actinin-4 and HDAC5 suggested that Gb 2 played a role in AngII-mediated remodeling of chromatin by these two proteins. Actinin-1 and -4 are isoforms ubiquitously expressed in non-muscle tissues. They are calcium-sensitive proteins that engage class II HDACs in nucleo-cytoplasmic trafficking [32]. Class II HDACs, including HDAC5, regulate gene expression through association with TFs and alteration of the histone code at gene promoters [5]. We hypothesize that Gbc is a component of the multiprotein complex at the promoters of MEF2 regulated genes that modulate transcription.

Essential role of Gb 2 in MEF2A-regulated transcription
The mechanism of gene regulation by Gb 2 was determined using small interfering RNA (siRNA) based loss-of-function approach that was similar to that used by Krumins and Gilman [33]. The Gb 2 mRNA and protein levels were specifically reduced upon stable expression of Gb 2 -targeted siRNA in cells, hereafter referred to as Gb 2 i cells, whereas mRNA levels of other Gb isoforms remained unchanged (Fig. S6, a-b). Normal Gq mediated signals upon activation of AT 1 R by AngII, such as the activation of ERK1/2 in the cytosol (Fig. S6, c-d), the accumulation of pERK1/2 in the nucleus (Fig. S6, e-f) and the mobilization of calcium from intracellular stores (Fig. S6g) remained unaltered in Gb 2 i cells.
In the Gb 2 i and control HEK-AT 1 R cell lysates, MEF2A protein levels were similar ( Fig. 4b; inset). But the basal MEF2luciferase reporter gene expression driven by the MEF2A protein was significantly reduced in the Gb 2 i cells, and AngII treatment did not increase MEF2-luciferase. In the HEK-AT 1 R cells, AngII treatment increased MEF2-luciferase, whereas the AT 1 R blockers, losartan and candesartan, antagonized the AngII-mediated increase in MEF2-luciferase ( Fig. S7a and S7b). The overexpression of Gb 2 in HEK-AT 1 R cells further increased AngII-mediated MEF2-luciferase (Fig. 4c). In the Gb 2 i cells, a-actinin-4 and the aactinin-associated HDACs were sequestered in the cytoplasm (Fig.  S8a and S8b), suggesting that these shuttling proteins preferentially remained in the cytosol when Gb 2 was knocked down. The cytoplasmic sequestration of a-actinin-4-HDAC has been shown to act as a mechanism to increase MEF2A-dependent transcription [32]. However, the reduction of basal and AngII-induced MEF2-luciferase in Gb 2 i cells when sufficient MEF2A was present  Fig. S3). (c) CID spectrum of the Gb 2 -specific tryptic peptide; peptide coverage is shown in Table S1. (d) CID spectrum of the Gc 12 -specific tryptic peptide. doi:10.1371/journal.pone.0052689.g001 indicates that Gb 2 plays a novel role in MEF2-luciferase gene transcription in normal cells.
We propose that the interaction between Gb 2 and MEF2A proteins is a GPCR-specific transcriptional cue that facilitates synergy between the MEF2A and TATA-binding protein (TBP) and transcription activating factor (TAF) complex in modulating transcription. As shown in Fig. 4d, in the presence of Gb 2 , MEF2A interacted with the TBP/TAF complex (reverse co-immunoprecipitations (co-IPs) are shown in Fig. S9a). Knockdown of Gb 2 in the Gb i cells specifically disrupted the interaction of MEF2A with TBP; however, the interaction between TBP and TAF was not affected. These results suggest that the synergy between the MEF2A and TBP/TAF complex requires Gb 2 . Hence, the knockdown of Gb 2 accounts for the decrease in basal as well as AngIIactivated expression of MEF2-luciferase. In Fig. 4e, we independently assessed the nuclear localization of myc-tagged Gc 12 in HEK-AT 1 R cells upon AngII treatment. The myc-tagged Gc 12 associated with endogenous Gb 2 , MEF2A and TBP. Thus, a novel Gb 2 c 12 -dependent multiprotein complex is formed in the nucleus and is essential for the transcriptional activation of the MEF2 promoter.
Previous studies have shown that class II HDACs directly interact with MEF2A and repress transcription through histones deacetylation [5]. The MEF2A-HDAC interaction is dynamically regulated. Our results indicate that weak basal transcription may result from the involvement of Gb 2 in a complex with TBP/TAF and MEF2A-HDAC, as the nucleosomes were deacetylated locally in this state (basal in Fig. 4f). An agonist-activated increase in Gb 2 in the nucleus is expected to generate a nascent enhancer complex in which Gb 2 interacts with TBP/TAF and MEF2A without HDAC5. This complex may facilitate recruitment of HATs, leading to local acetylation of histones and promoting AngIIstimulated transcription. The knockdown of Gb 2 weakened the synergy between TAF/TBP and MEF2A, thereby attenuating transcription. The cytoplasmic localization of a-actinin-4 and HDAC in the Gb 2 i cells suggests that HDAC shuttling regulates Gb 2 -dependent transcription in the nucleus.

WD repeat structure in Gb 2 is essential for its interaction with MEF2A
To gain insight into the molecular interaction between Gb 2 and MEF2A, we used a mutagenesis approach. Each of the 7 WD repeats was sequentially deleted to create seven DWD mutants (Fig. S9b). All WD repeat deletion mutants interacted with MEF2A (Fig. S9c), and the DWD2, DWD3, and DWD5-DWD7 mutants stimulated MEF2A function, whereas the DWD1 mutant did not (Fig. S9d). Therefore, potentiating the MEF2A function appears to require a structure that includes the WD1 repeat of Gb 2 . Rather surprisingly, the DWD4 mutant had a significant stimulatory effect, suggesting that WD4 in Gb 2 might be the site of its interaction with HDACs (which are MEF2A co-repressors). Thus, the WD1 repeat of Gb 2 is essential for promoting MEF2A function; however, all of the WD repeats contributed to interaction between Gb 2 and MEF2A. This led us to hypothesize that MEF2A makes contact with the central canal of the Gb 2 toroidal structure.

Gb 2 interacts with TFs that share an amino acid sequence motif
To localize a putative MEF2A binding site on Gb 2 c 12 , we evaluated chromatin-associated proteins that have been proven to interact with the central canal of the b-propeller proteins (see the methods). Combining molecular modeling, bioinformatics and evolutionary relationship (interactions of vertebrate b-propeller proteins in chromatin are not known) created a model for the interactions of MEF2A and histones with Gb 2 . Several proteins that bind to the b-propeller central canal, including cyclin E binding to Cdc4, use a phosphopeptide (-LLTPPG-) docking site [34][35][36]. A similar sequence motif was conserved in MEF2A, and when aligned, an extended homology with the cyclin E region was found (Fig. 5a). Molecular modeling and docking experiments (detailed in the methods) indicated that MEF2A could dock at the central canal of Gb 2 (Fig. 5b) and that the proposed MEF2A binding site should not overlap with the interaction sites for cytoplasmic effectors of Gb 2 [37]. The histone H4 peptide binds WD7 in Drosophila b-propeller protein p55 [38]. In the Gb 2 c 12 model (Fig. 5b), the binding site for histone H4 was conserved; thus, WD7 in Gb 2 may interact with the nucleosome.
Bioinformatic analysis revealed that one copy of the -LLTPPGmotif was conserved in several AngII-responsive TFs, including STAT1/3 and NFAT, but not in NFkB p65 or GATA4 (Fig. 5a). We tested this prediction by protein interaction analysis, and the data revealed that Gb 2 indeed associated specifically with NFAT and STAT1/3, but not with NFkB p65 or GATA4 (Fig. 5c). By interacting with multiple TFs via the conserved -LLTPPG-motif, Gb 2 can coordinate the expression of multiple target genes. The human genome harbors .10 5 sites for each of the Gb 2 -interacting TFs, and transcription at some of these sites must be Gb 2dependent in response to GPCR agonists. Gb 2 and other Gbc isoforms may also coordinate GPCR-dependent gene transcription (Fig. 5d). The clinical success of GPCR-targeted drugs indicates that the therapeutic benefits of these drugs potentially include modulation of Gbc-dependent chromatin remodeling. These insights led us to investigate the genome-wide transcription profile upon knockdown of Gb 2 .

Gb 2 -dependent global gene expression pattern
Expression profiling indicated that <400 transcripts were differentially regulated by Gb 2 -dependent signals (Fig. 6a, Table   S5). The Ingenuity Pathway analysis sorted the expression data to gene networks that reflected the capacity of the gene products (i.e., receptors, enzymes, scaffold proteins, and extracellular matrix components) to influence specific cellular functions. The most prominent cellular functions that were altered are shown in Figure 6b. Each of the significantly altered cellular functions consisted of a network of .60 molecules (Fig. 6c), indicating that Gb 2 knockdown substantially altered gene regulation (see Dpvalue). Gb 2 knockdown transformed the ''cellular growth and function'' network to the ''cellular growth and function in disease'' network (Fig. S10). The network score of 41 before Gb 2 knockdown indicated that there was a 10 241 chance that these genes were randomly present in the network. The network score after the knockdown was 40. Sixteen core molecules were unaffected by Gb 2 knockdown. The functional change in the Gb 2 i cells appeared to be due to co-opted PM-resident transmembrane proteins (e.g., platelet-derived growth factor receptors and integrins) and secreted proteins (e.g., interleukin-1, serine protease inhibitors of the SERPIN gene family, insulin-like growth factor-1, and platelet-derived growth factor). Interestingly, the promoters of differentially expressed genes contained the binding sites for one or more of the TFs that associated with Gb 2 (Table S2). Analysis using Metacore TM revealed a set of genes (Fig. 6c) that were transcriptionally regulated by MEF2A, NFAT, STAT1, and STAT3. Gb 2 knockdown affected AngII-dependent transcription of these genes as confirmed by real time RTPCR, which may be the direct cause for the transformation of network function. In vivo, when the AT 1 R stimulus becomes chronic or when Gb 2 is not regulated properly, the dynamics of the signaling networks might tilt towards a disease state that can promote damage to the tissue as well as contribute to chronic disorders. We conclude that Gb 2 is a master regulator of gene expression programs in response to agonist activation of AT 1 R and likewise other the GPCRs.

Discussion
PM-to-nuclear translocation of Gbc and its regulation of nuclear effectors is a novel paradigm in GPCR signaling. The most common role for Gbc may be in mediating synergy between different transcriptional regulatory complexes at gene promoters. The specific and dynamic changes that are orchestrated by Gbc could involve facilitating the interaction of the enhancer complex (MEF2A) with the TFIID complex (TBP/TAF), the stepwise dissociation of negative regulators (HDACs) from transcriptional regulatory complexes and the association of positive regulators (HATs, co-activators). Novel nuclear targets of Gbc were identified in the present work, and the ability of the Gbc complex to regulate nuclear translocation of glucocorticoid receptor [9] and HDAC5 [10,11] has been previously reported. The ability of Gbc to facilitate interactions between multiple proteins that are involved in gene regulatory complexes can explain the signaling specificity and the high-level transcriptional output by G-proteins. Many proteins in the multiprotein complex can promote gene expression individually; however, none of these components, with the exception of Gbc can function unequivocally as a GPCRspecific enhancer of gene transcription.
In specialized cells, such as cardiac and smooth muscle cells, the intracellular distribution of some GPCRs, including the AT 1 R, and G-proteins has been reported [39]. A consensus regarding how GPCRs signal in the subcellular compartments apart from image of N = 3, and in each experiment, .50 cells were scored. Scale bars = 50 mm.  coimmunoprecipitates with a-actinin-4, HDAC5, MEF2A, and the histones H2B and H4. The nuclear fractions (100 mg) prepared from HEK-AT 1 R cells treated with AngII (1 mM for 30 min) were subjected to pull-down with only ProtG (2) or with a Gb 2 antibody and ProtG (+). The immunoblot on the right shows the abundance of the respective proteins in the immunoprecipitates (2 and +) as well as input lysates for the 2 and + samples. Gel-C peptide index mining provided further supporting evidence for the provisional interactome of nuclear Gb 2 (Table S1). (b) A significant increase in MEF2-luciferase activity (*p = 0.039) when AT 1 R was exposed to AngII (bars 1 and 3 from right). The basal MEF2-luciferase activity was significantly (,50%) attenuated in AT 1 R-Gb 2 i cells when compared to wild-type AT 1 R (bars 1 and 2; **p = 0.002). The RLU is normalized to co-expressed b-gal activity in each sample. Data were further normalized to basal MEF2-luciferase activity in wild-type AT 1 R cells. Inset: No significant change was detected in MEF2A protein levels in the cell lysate. (c) A significant increase in MEF2-luciferase activity upon FLAG-Gb 2 overexpression. (d) In Gb 2positive cells, immunoprecipitation with anti-TBP antibodies revealed the interaction of TBP with MEF2A and TAF. In the absence of Gb 2 (Gb 2 i cells), TBP failed to co-immunoprecipitate MEF2A, but TAF was co-immunoprecipitated. The immunoblot on the right shows the abundance of the TAF, TBP, MEF2A and Gb 2 proteins in lysates (INPUT; Gb 2 and Gb 2 i). (e) Upon AT 1 R activation with AngII, the transiently transfected myc-Gc 12 translocated to the nucleus with endogenous Gb 2 and associated with TBP and MEF2A. (f) Model depicting the modulation of MEF2A-dependent gene transcription by Gb 2 -associated proteins (MEF2, HDAC5, a-actinin-4, TBP and TAF). Basal: In this state, Gb 2 forms a complex with MEF2A, HDAC5 and PM resident GPCRs or how intracellular and cell surface GPCR signaling coordinated is still evolving. It is possible that extracellular agonists reach the intracellular compartments, such as the nucleus and promote local G-protein signaling in specialized cells. Local G-protein activation may also regulate the nuclear targets of Gbc as well as the generalized retrograde translocation of Gbc from the PM to the nucleus.
A direct role for G-protein subunits in orchestrating gene responses to GPCRs is thought to be limited because the repertoire of conventional signaling targets of heterotrimeric G-proteins are localized in the PM and cytosol. The discovery of Gb 2 translocation to the nucleus and its role in the regulation of gene networks define Gbc as a key missing link through which GPCRs modulate gene expression. A variety of GPCR agonists promote the nuclear translocation of Gbc in physiologically relevant cells, indicating its universal significance.
WD repeat b-propeller proteins are integral components of chromatin-modifying complexes in lower eukaryotes [40]. Members of this family (.165 proteins) exhibit similar structures and remarkably, perform similar types of nuclear functions [40][41][42][43][44][45]. However, the nuclear functions of heterotrimeric Gbc proteins, which are the founding members of the b-propeller protein family, have remained elusive. Taken together with previous data [8][9][10][11][12], our findings suggest that Gbc proteins mediate chromatin remodeling, which may be an evolutionarily ancient and essential function in vertebrates. Histone gene clusters and histonemodifying enzymes were indeed modulated in Gb 2 i cells (Table  S3), similar to the regulation of histone genes by b-propeller Actinin-4. Histones are deacetylated locally. This yields the basal transcription (for instance, that of MEF2-luciferase). AngII: MEF2A forms a complex with Gb 2 , which also interacts with the TBP-TAF complex. Incoming Gb 2 displaces the existing repressor complex (i.e., the a-actinin-4-associated pHDAC is exported into the cytosol). In this state, the recruitment of HATs to the complex results in the acetylation of histones, synergy with the TBP complex, and activation of MEF2-luciferase transcription. Gb2i: In this state, the absence of Gb 2 leads to the cytosolic localization of the actinin-HDAC complex, as shown in Figure S8. In addition, MEF2A cannot interact with TBP, which leads to a lack of synergy and the attenuation of basal transcription. Note: the schematic shows no change in the TBP and RNA polymerase complex. doi:10.1371/journal.pone.0052689.g004   Table S5). Out of these, 299 transcripts were identical to the transcripts in the Gb 2 Sc control, indicating that these transcripts were regulated by Gb 2 -independent signals from AT 1 R, and the remaining <400 transcripts were specifically regulated by Gb 2 . The false discovery rate was ,3%. proteins in drosophila [42] and histone deacetylases in chicken [45]. Thus, controlling the expression of chromatin-regulating complexes may be a critical function of vertebrate Gbc.
Global gene regulation dependent on Gbc provides a mechanism for direct gene regulatory function of an activated GPCR in a variety of biological contexts. Nearly 2% of the modulated genes in Gb 2 i cells are members of the GPCR superfamily or are involved in signal transduction activated by GPCRs (Table S4), and they also include <30% of cardiac hypertrophic marker genes (Table  S4) [24]. Increased G-protein signaling is a trigger for the reactivation of the fetal gene program, which is a hallmark feature of cardiac hypertrophy and heart failure [24][25]. The extent to which deregulation of gene expression in vivo is due to extensive reconfiguration of the epigenome and/or involves Gb 2 is a critical question that remains to be elucidated. Conceivably, the Gbc pathway could be targeted pharmacologically to control physiological and pathological chromatin responses and may be particularly useful in the setting of chronic disorders [46], in which dysregulated GPCR signaling is known to play an important role. Therefore, it is essential to gain a better understanding of the role of different Gbc isoforms in epigenetic regulation.

Supporting Information
Methods S1 Details of experimental protocol for some methods described. AngII ligation with AT 1 R mobilizes calcium from intracellular stores. (c) Immunocytochemical analysis of HEK-293 cells stably expressing HA-tagged AT 1 R (labeled green with FITC) and visualized by confocal microscopy. Under quiescent conditions, the receptors are localized at the plasma membrane. Receptor activation with 1 mM AngII caused PM ruffles (white arrows) followed by a significant increase in the immunoreactivity of pERK1/2 (labeled red) in the nucleus (blue) for up to 60 min. Note that the confocal image shown here is after 10 min of AngII stimulation. In all subsequent experiments, 30 min of stimulation was used. (TIF) Figure S2 Preparation and validation of the chromatin proteome. Nuclear fraction extraction for mass spectrometry analysis. Cytoplasmic and nuclear fractions were prepared from untransfected (UT) and AT 1 R-expressing HEK-293 cells treated with different ligands (AngII, losartan and candesartan). Fifty micrograms of protein was loaded onto 10% Nu-PAGE gels and subjected to western blot analysis. The G-protein a-subunit, Gaq was only found in the cytoplasmic fraction, whereas histone H2A was found in the nucleus, and T-ERK1/2 was present in both fractions. Note: the chromatin proteome was queried for CID spectra of peptides corresponding to plasma membrane and cytosolic marker proteins (e.g., integrins, Ga, GAPDH, bactin, and cytochrome b5). None of the peptides corresponding to the above abundant proteins were detected in the nucleus, which confirms the fractionation procedure. (TIF) Figure S3 Classification of the chromatin proteome of AT 1 R-activated cells. All peaks with at least 15 product ions in the MS/MS spectra were extracted. The peak lists from three replicate experiments were searched against mouse and rat reference sequences using search parameters for human protein tryptic fragments and allowing for standard modifications and cleavage variation (1 missed cleavage/peptide). Quantitative analysis was performed by label-free spectrum counting after applying a threshold peptide ion score of 30 for MS/MS interpretation. All peptides were manually validated. The minimum criterion for positive identification of any protein was the presence of one signature peptide with a manually validated CID spectra. A total of 173 proteins were present on the peak list, of which 137 proteins met the selection criteria applied. (TIF) Figure S4 The CID spectra of a-actinin-4 peptides. The chromatin proteome of AT 1 R-activated cells consisted of peptides (CISQEQMOXQEFR, TINEVENQILTR, FAIQDISVEET-SAK) assigned (MASCOT/NCBI non-redundant database) to aactinin-4. (TIF) Figure S5 Gb2 accumulates in the nucleus upon AT 1 R activation by AngII and is blocked by treatment with the AT 1 R antagonist, losartan. (a) The HEK-AT 1 R cells untreated or treated with 1 mM losartan and HA-AT 1 R were labeled red, and Gb2 was labeled green. The inset in the right top corner of the middle Gb2 panel shows a magnified image (10006) of a single cell (arrow in overlay). The nucleus of the cell shows green staining that corresponds to Gb2 in the nuclei. (b) Different analytical methods, including mass spectrometry (MS), immunoblot (IB) analysis and immunocytochemistry (ICC), showed equivalent fold changes in Gb2 accumulation in the nucleus upon AT 1 R activation. A pixel counting approach estimated (50 cells, n = 3) that ,30% of the Gb2 pool was localized in the nucleus when AT 1 R was activated with 1 mM AngII for 30 min in HASM and HEK-AT 1 R cells. This distribution accounts for ,2.5-4.5fold increases in Gb2 levels in the nucleus which is similar to that estimated by other methods. (TIF) Figure S6 AT 1 R-mediated cytoplasmic signaling events are unaffected upon RNAi-mediated silencing of Gb2. (a) Total lysates were prepared from untransfected HEK293 cells, dual plasmid-transfected clones expressing AT 1 R with scrambled Gb2-scrambled (Gb2Sc) and AT 1 R with a Gb2RNAi plasmid. Lysates were subjected to immunoblot analysis to detect AT 1 R expression (anti-HA), Gb (pan antibody) and b actin (loading control). Both of the cell lines exhibited equivalent levels of AT 1 R. The B max (maximal specific binding) obtained for AT 1 R-Gb2Sc was 8.7+/20.9 pmol/mg and 9.7+/20.9 pmol.mg for AT 1 R-Gb2i with a K d value of 1732.5+/2170 pM. Taken together, both cell lines expressed comparable levels of AT 1 R. (b) Table showing the Affymetrix array gene expression data from Gb2i stable cell lines compared to Gb2+ cells revealed a knockdown specifically for GNB2. (c-d) Both cell types were serum starved for a minimum of 18 hr and then exposed to vehicle (2) or 1 mM AngII (+) for 5, 10, 15, 20, 30 and 60 min. Lysates were immunoblotted for pERK1/2 and total ERK1/2 in Gb2Sc and Gb2i cells. The phosphorylation of ERK1/2 upon AngII activation of AT 1 R was preserved in the absence of Gb2. (e-f) Immunocytochemical analysis followed by confocal imaging of pERK1/2 (labeled green) localized in the nuclei (labeled blue with DAPl) in Gb2Sc and Gb2i cells upon AT 1 R activation with AngII. (g) Calcium mobilization upon AngII activation of AT 1 R was preserved in Gb2Sc and Gb2i cells (fluorescence-based assay using FLEX Station 3). (TIF) Figure S7 The AT 1 R blockers, losartan and candesartan, prevented the increase in AngII-mediated MEF2 reporter activity. (a) AngII treatment increased MEF2luciferase expression, and this increase was blocked by treatment with the AT 1 R antagonist, losartan (,54%), and (b) candesartan (,96%). Note: losartan is a less potent AT 1 R antagonist compared to candesartan. Data are expressed as % RLU normalized to the AngII response (100%) with losartan/candesartan alone as 0%.
Error bars indicate standard error of the mean (n = 3) of experiments performed in duplicate. P values were * = 0.03 and ** = 0.02 using an unpaired t-test (two-tailed with Welch's correction in GraphPad Prism software). (TIF) Figure S8 Gb2 modulates the export of the a-actinin-4-HDAC complex from the nucleus to the cytosol. (a) Immunocytochemical analysis of AT 1 R and AT 1 R-Gb2RNAi cells revealed increased cytoplasmic localization of a-actinin-4 (green) compared with control. (b) An actinin-associated HDAC activity assay on the cytosolic fraction of AT 1 R in AT 1 R-Gb2 RNAi cells (no agonist and AngII 1 mM for 30 min). There was a significant increase in actinin-associated HDAC activity upon AngII treatment of AT 1 R cells. There was a significant increase under quiescent conditions in Gb2i cells (no agonist).