Nuclear Effects of GRK5 on HDAC5-regulated Gene Transcription in Heart Failure

Abstract

Background

G-protein receptor kinases (GRKs) modulate cardiac β-adrenergic signaling. GRK5 is upregulated in heart failure, and a gain-of-function polymorphism substituting leucine for wild-type glutamine at amino acid 41 (GRK5-Leu41) is associated with improved outcomes in heart failure and hypertension. GRK5 is distinguished by partial nuclear localization and HDAC kinase activity that is postulated to regulate Gαq-stimulated cardiac gene expression.

Methods and Results

We used in vitro tissue culture and in vivo mouse compound genetic models to examine the effects of GRK5 on HDAC phosphorylation, nucleo-cytoplasmic HDAC transport, and Gαq-dependent transcriptional regulation. In vitro, GRK5 stimulated HDAC5 nuclear export only in the context of Gαq signaling stimulated by angiotensin II. GRK5-Gln41 and Leu41 were similar inducers of HDAC5 nucleo-cytoplasmic shuttling. In vivo, GRK5-Gln41 and-Leu41 partitioned equally to nuclear and non-nuclear myocardial fractions. GRK5 increased cardiac HDAC5 phosphorylation and reversed the increase in nuclear HDAC5 content seen with cardiomyocyte-autonomous Gαq overexpression. Deep RNA sequencing showed few changes in gene expression induced by GRK5 overexpression or ablation alone, but GRK5 overexpression normalized steady-state expression levels of 48% (96 of 200) of all Gαq down-regulated mRNAs.

Conclusions

GRK5 is a transcriptional modifier of a sub-set of Gαq-downregulated genes, acting in opposition to the pathological effects of Gαq and normalizing levels of these transcripts. This transcriptional co-regulator effect may act in concert with β-adrenergic receptor desensitization to protect against heart failure decompensation.

β-adrenergic receptor signaling plays a central role in adjusting cardiac function according to varying minute-by-minute demands. In the context of heart failure, this compensatory mechanism is co-opted in an attempt to sustain chronically diminished cardiac output, producing cardiotoxicity that accelerates functional deterioration. Accordingly, pharmacological β-adrenergic receptor blockade prolongs survival and improves ventricular function in heart failure¹.

Functional alterations in β-adrenergic signaling pathways induced by common genetic polymorphisms of adrenergic receptors or receptor kinases can alter the response to beta-blockers in heart failure or affect heart failure progression in subjects not treated with beta-blockers^1,2. Of particular interest is the glutamine (Gln) to leucine (Leu) polymorphic substitution at amino acid 41 of G-protein receptor kinase 5 (GRK5), one of two cardiac- expressed receptor kinases (the other being GRK2) that desensitize β-adrenergic receptors. The GRK5 Leu41 polymorphism is rare in European-derived populations, but is common among individuals of African descent, and is associated with more favorable outcomes in heart failure ^{3, 4} and hypertension ⁵. GRK5 Leu41 has also been associated with apical ballooning syndrome ⁶.

The Leu41 GRK5 variant enhances β1- and β2-adrenergic receptor desensitization in multiple cell and tissue types, mimicking the effects of pharmacological beta-blockade ^{3, 7}. Accordingly, its benefits in heart failure and hypertension are widely assumed to derive from desensitization of β-adrenergic signaling ^{8, 9}. However, GRK5 is unique among cardiac-expressed GRKs in that it can localize to cell nuclei ¹⁰ and, in the context of increased signaling through independent Gαq pathways, phosphorylate class II histone deacetylases (HDACs) ¹¹. As HDACs are transcriptional suppressors ¹², these actions suggest that GRK5 may regulate expression of cardiac genes.

Given the positive effects of the GRK5 Leu41 variant in heart failure and hypertension, we postulated that nuclear localization of GRK5 contributes to its clinical phenotypes. However, the consequences of GRK5 on cardiac transcription have not yet been examined in detail. Here, we demonstrate a requirement for GRK5 for angiotensin II-stimulated HDAC5 nuclear export in Cos7 cells, but find no difference in HDAC5 nuclear-cytoplasmic shuttling induced by GRK5 Gln41 and GRK5 Leu41. Using deep mRNA sequencing of cardiomyocyte-specific GRK5/Gαq double transgenic, or GRK5 gene knockout/Gαq overexpressing mice, we identify specific mRNAs whose Gαq-mediated regulation is reversed by GRK5. We conclude that transcriptional effects of GRK5 are significant, reversing the suppression of approximately half of cardiomyopathy-downregulated genes, but that its ability to phosphorylate HDAC5 is not affected by the polymorphic substitution of Leu41 for Gln41. Thus, the beneficial clinical phenotype conferred by the GRK5 Leu41 variant in heart failure and hypertension can be explained by its previously reported ability to enhance β-adrenergic receptor desensitization, rather than by effects on HDAC5 regulation and consequent gene transcription.

Methods

GFP-HDAC5 nuclear export studies

Cos7 cells maintained in Dulbecco’s minimal essential medium plus 10% fetal calf serum and 1% antibiotics/antimycotic were transfected using lipofection with plasmids expressing GFP-HDAC5 ¹³, human angiotensin II type 1 (AT1) receptor, and either empty pcDNA3 (vector control), pcDNA3-human GRK5-Q41, or human GRK5-L41. Twenty-four hours after transfection the culture medium was exchanged for 1% FCS. Studies were performed forty-eight hours after transfection, cells were treated with vehicle or 1 μmol/L angiotensin II for 2 hours and fixed. Cells were incubated with anti-GRK5 (sc-565, Santa Cruz), and then incubated with goat anti-rabbit Alexa Fluor 546 (Invitrogen) and examined after nuclear co-staining (DAPI) under a fluorescence microscope. 100 nmol/L PMA was used as a positive control for HDAC5 nucleo-cytoplasmic translocation.

Angiotensin II-stimulated calcium transients were assessed in identically transfected cells loaded with 5 μmol/L FURA-2-AM for 30 minutes.

Mouse models

The Gαq transgenic and human GRK5 Gln41 and Leu41 transgenic mice were generated on the FVB/N background, and have been described previously ^{3, 14}. GRK5 knockout mice on a mixed C57 background ¹⁵ were a generous gift from Richard Premont, Duke University. Echocardiographic and invasive hemodynamic studies were performed using standard methods.

Immunoblot analysis

A nuclear myocardial fraction was prepared from clarified (200xg for 20 minutes) mouse heart homogenates by centrifugation at 1000xg for 10 minutes. Nuclear and non-nuclear protein was diluted in 50 mmol/L Tris (pH7.5), 2.5 mmol/L EDTA, 25 mmol/L NaCl, 0.2% (v/v) NP-40, 10 mmol/L EGTA, 20 mmol/L sodium fluoride, 2 mmol/L sodium orthovanadate and Complete Mini Protease Inhibitor Cocktail Tablet (Roche). Nuclear protein marker was anti-lamin B (Abcam). For detection of HDAC5 in mouse hearts, 500 μg non-nuclear protein and 100 μg nuclear protein were immunoprecipitated with protein G and anti-HDAC5 (sc-133106, Santa Cruz). Proteins were subjected to SDS-PAGE, transferred to PVDF and visualized with anti-phospho-HDAC5 (PA1-14187, Thermo Scientific), anti-HDAC5 (sc-133106, Santa Cruz), anti-GRK5 (sc-565, Santa Cruz) and anti-Gαq/11 (sc-392, Santa Cruz) with HRP-coupled secondary antibodies (Santa Cruz) and chemiluminescence detection (Western Lightning, Perkin-Elmer).

Mouse cardiac mRNA profiling by RNA-sequencing

Total cardiac RNA was isolated from flash-frozen ventricular tissue using Trizol (Invitrogen) and quantified on a UV spectrometer. Preparation of cardiac cDNA libraries and DNA bar-coding was as previously described ¹⁶. Four barcoded libraries were combined in equimolar (10 nmol/L) amounts and diluted to 6 pmol/L for cluster formation on a single Illumina Genome Analyzer II flowcell lane, followed by single-end sequencing. Base calling of DNA clusters was performed using Illumina’s processing pipeline software (version 1.5) and 36-nt sequences, with quality scores, were obtained in Illumina’s SCARF text format. Cufflinks (http://cufflinks.cbcb.umd.edu/) was used with gene annotation files to calculate overall gene expression in terms of Fragments Per Kb of exon per Million mapped reads (FPKM), equivalent to RPKM (Reads Per Kb of exon per Million mapped reads). We analyzed only RNA elements with expression signals in at least 2 biological replicates.

Human cardiac mRNA profiling by microarray

Total cardiac RNA was obtained from a previously described human heart failure cohort ¹⁷ . Poly(A)+ RNA was selected, processed and hybridized to Affymetrix HuEx v1.0 microarrays by the Multiplexed Gene Analysis core at Washington University. Further details of RNA-sequencing and microarray analyses are presented in Supplemental Methods.

Bioinformatic identification of transcription factor target genes

The MAPPER database (present at http:/bio.chip.org/mapper) was used to examine the 5 kb proximal promoter region of selected genes ¹⁸. Searches were conducted for putative cis- binding sites for the transcription factors Mef2, MyoD, myogenin, p50- and p65-NFκB, and GATA1.

Gene Ontology and IPA analysis

Biological Networks Gene Ontology (BiNGO; ¹⁹) was used to assign gene-ontology categories to each of the regulated genes. Over-representation of Gene Ontology categories was assessed at P<0.05 using a hypergeometric test with Benjamini-Hochberg false discovery rate correction ¹⁹. Ingenuity Pathway Analysis software (http:/www.ingenuity.com) was used to identify signaling networks involving subsets of Gq-regulated genes from RNA-sequencing experiments.

Statistical Methods

Paired and unpaired data were compared with Student’s t-test. Multi-group comparisons used ANOVA with post-hoc Tukey’s test. For gene expression data, false discovery rate was set at 0.05 and Partek Genomics Suite 6.4 software (Partek, St Louis, MO) was used to calculate significant differences between groups. P value of <0.05 was defined as significant unless indicated otherwise.

Results

Angiotensin II and GRK5 act synergistically to stimulate HDAC5 nucleo-cytoplasmic shuttling

The initial report of bovine GRK5 as a nuclear-localized HDAC kinase described a requirement for Gαq signaling for GRK5 nuclear localization. In those studies, Gαq signaling was induced using adenoviral infection of isolated adult cardiac myocytes with a constitutively active Gαq mutant ¹¹. Because constitutive activity of signaling pathways is artificial, we examined the effects of human GRK5 Gln41 and Leu41 on HDAC nuclear export using a physiological stimulus for Gαq signaling, angiotensin II. We chose a cell line, Cos7 cells, in which distal G-protein signal transduction pathways are intact, but that lack potentially confounding G-protein coupled receptors and receptor kinases. For this reason, Cos7 cells at baseline exhibited no cytosolic calcium transient in response to 10 μmol/L angiotensin II (not shown), whereas cells transiently transfected with a human AT1 receptor (AT1R) expressing plasmid exhibited a robust dose-dependent increase in free intracellular calcium in response to angiotensin II (Figure 1a). AT1R-mediated increases in cytosolic free calcium are transduced by Gq/phospholipase C signaling ²⁰.

Figure 1. GRK5 and angiotensin II combine to induce nuclear export of HDAC5.

A. Dose-dependent cytosolic free calcium transients induced by angiotensin II stimulation of AT1R-transfected Cos7 cells. Y-axis value is ratio of 540 nm emission fluorescence of Fura-2 loaded cells excited at 340 and 380 nm. B. Live-cell confocal studies of GFP-HDAC/AT1R co- transfected Cos7 cells 1 hour after stimulation with vehicle (veh), 1 μM angiotensin II (+Ang II), or 100 nM phorbol myristic acid (+PMA). Blue is Hoechst nuclear staining. C. Confocal microscopy of angiotensin II-stimulated fixed Cos7 cells co-transfected with GFP-HDAC (green), GRK5 (red), and AT1R, with DAPI nuclear stain (blue). Arrowhead indicates a cell expressing GRK5, arrow indicates a cell not expressing GRK5. Both cells express GFP-HDAC5. Merged image is on right. D. As in C, with GRK5-Gln41 (left panels, Q41) or GRK5- Leu41 (right panels, L41), before (upper) and 1 hour after (lower) addition of 1 μM angiotensin II. Merged images are on right of each panel. Representative of 4–6 independent experiments, each.

HDAC nucleo-cytoplasmic shuttling was monitored by confocal fluorescence microscopy of green-fluorescent protein (GFP)-tagged HDAC5. GFP-HDAC5 is typically localized to the nucleus of unstimulated serum-deprived Cos7 cells ¹³, and co-expression of AT1R did not change this normal pattern (Figure 1b, left frame). Angiotensin II (1 μmol/L) failed to induce HDAC5 nuclear export in cells co-expressing GFP-HDAC5 and AT1R (Figure 1b, middle frame), whereas protein kinase (PK) C-activating phorbol ester treatment promoted rapid HDAC5 relocalization to the cytoplasm (Figure 1b, right frame), consistent with HDAC phosphorylation by PKC and/or PKD ¹³.

We determined if GRK5 interacted with angiotensin II to promote HDAC5 nuclear export by cotransfecting Cos7 cells with expression plasmids encoding GFP-HDAC5, AT1R, and human GRK5. For some studies the relative amounts of plasmid DNA were adjusted so that only a fraction of cells expressing GFP-HDAC5 and AT1R would express GRK5. Both GRK5 expression and angiotensin II signaling were essential for HDAC nuclear export: Figure 1c shows two cells expressing GFP-HDAC5 (green fluorescence; arrow and arrowhead), only one of which is expressing GRK5 (red fluorescence; arrowhead). Angiotensin II treatment results in HDAC5 cytosolic relocalization only in the cell expressing GRK5. Figure 1d shows that angiotensin II treatment is essential for GRK5-mediated HDAC export: In cells treated with vehicle (top series of micrographs), both GRK5 Gln41 (left panel; Q41) and GRK5 Leu41 (right panel; L41) are located throughout the cytosol and nucleoplasm (red fluorescence), whereas HDAC5 (green fluorescence) is strictly nuclear. By contrast, after treatment with 1 μM angiotensin II (bottom series of micrographs), GRK5 preferentially localizes to cell membranes and HDAC5 has relocalized to the cytoplasm. Notably, there were no differences in GRK5 Gln41- and Leu41-stimulated HDAC5 nucleo-cytoplasmic shuttling (Figure 1d). These results show that both GRK5-Gln41 and GRK5-Leu41 act in synergism with AT1R activation to stimulate HDAC nuclear export.

Gαq and GRK5 regulate cardiac HDAC5 phosphorylation and nuclear content

The ability of GRK5 to phosphorylate class II HDACs and induce their nuclear export (Figure 1 and ¹¹) suggests that it can regulate cardiac gene transcription. Loss of nuclear HDAC is thought to de-repress cardiac genes linked to hypertrophy ¹². Accordingly, we examined HDAC localization and phosphorylation in hearts expressing Gαq with and without GRK5. As GRK5-Gln41 and –Leu41 showed similar nuclear localization in Cos7 cells (see above) and in mouse cardiac myocytes in vivo (Supplemental Figure 1), these studies used only wild-type human GRK5-Gln41 (henceforth called GRK5). Hypertrophied Gαq transgenic hearts showed a ~3-fold greater nuclear HDAC5 content than non-transgenic controls (Figure 2a). As expected (since phosphorylation of HDACs induces their nucleo-cytoplasmic transport), this nuclear HDAC was not phosphorylated (Figure 2a). Combined expression of GRK5 with Gαq promoted HDAC5 phosphorylation and nuclear export that was not seen with expression of either factor alone (Figure 2b). To our knowledge, this is the first description of in vivo cardiac myocyte HDAC phosphorylation by GRK5.

Figure 2. GRK5 phosphorylates HDAC5 in vivo, reversing Gq-mediated nuclear HDAC5 accumulation.

A. Immunoblot analysis of phospho-HDAC5 (P-HDAC5) and total HDAC5 (HDAC5) in nuclear myocardial fraction of nontransgenic (ntg) and Gαq-transgenic (Gq-TG) mice. Left, densitometric quantitation of group nuclear HDAC content (n=3; AU = arbitrary densitometric units). Right, immunoblots: Lamin B is control for nuclear protein and sample loading. Immunoreactive Gαq is also shown. B. Similar study performed on cytosolic fraction of GRK5 transgenic, Gαq transgenic, and GRK5/Gq compound transgenic mouse hearts. Left, densitometric quantitation of group cytosolic P-HDAC5/total HDAC5 ratio (n=5). Right, representative immunoblots (n=2/genotype).

Deep RNA sequencing identifies genes counter-regulated by Gαq and GRK5

The above results show that Gαq increases in vivo cardiomyocyte nuclear-associated HDAC5, which is corrected by GRK5-mediated HDAC5 phosphorylation and nuclear export. We therefore examined the consequences of these events on cardiac gene expression by crossing GRK5 transgenic and knockout mice to mice overexpressing a non-activated Gαq signaling protein (~5-fold) in cardiac myocytes ¹⁴. The functional, structural, and transcriptional characteristics of Gαq mice have been extensively characterized: they develop autonomous, load-independent cardiac hypertrophy with contractile dysfunction and almost complete absence of β-adrenergic responsiveness ^{14, 21, 22}. Gαq mice have a gene expression profile that closely reproduces that of pressure overloaded cardiac hypertrophy ^{16, 23–25}, consistent with pressure overload hypertrophy being a Gq-dependent event ²⁶. Thus, we considered that the Gαq model provides a background of pathological cardiac gene expression upon which it might be possible to identify transcriptional effects of GRK5.

Deep RNA sequencing has previously been used to identify 125 mRNAs that are the most highly regulated by cardiomyocyte-specific Gαq overexpression ¹⁶; we used these highly regulated genes in our initial assessments. The absolute and relative expression level of these Gαq-regulated cardiac genes (Supplemental Table 1) in each of our mouse heart models is illustrated in heat maps of Figures 3a and 3b, respectively: mRNA profiles of the two mouse genetic backgrounds used in these experiments, C57 (for the GRK5 knockout studies) and FVB/N (for the GRK5 transgenic studies), were similar, and simple overexpression or ablation of GRK5 had little effect on the cardiac mRNA signatures. By comparison, mRNA expression profiles of Gαq mice showed characteristic changes ¹⁶. Remarkably, co-expression of GRK5 with Gαq appeared to normalize expression of two clades of Gαq-downregulated mRNAs (Figure 3b, clades 4 and 7), which were themselves reversed by combined ablation of GRK5 with Gαq.

Figure 3. Effects of GRK5 gain and loss of function on Gq-mediated cardiac gene expression.

Heat map of raw (A) and normalized (B) quantitative RNA-sequencing results, for 125 mRNAs highly regulated by Gq overexpression, from individual nontransgenic FVB/N and C57, GRK5-transgenic (GRK5TG), Gq-transgenic (Gq-TG), GRK5/Gq compound transgenic (Gq+GRK5TG) and GRK5 knockout / Gq-transgenic (Gq+GRK5KO) hearts. Clustering was performed on rows (mRNAs) using Euclidean dissimilarity. Data are plotted using log₂ scale.

In our previous studies we have always focused upon the genes that are upregulated in the Gαq cardiomyopathy ^{14, 16, 24, 25}. Here, we further interrogated reversal of Gαq-mediated gene downregulation by GRK5 in an expanded set of 200 Gαq-downregulated mRNAs (<=-1.3-fold compared to nontransgenic, P<0.025, FDR=0.08). Forty-one of these transcripts are similarly downregulated in human heart failure (Figure 4a), supporting the clinical relevance of this model. Strikingly, downregulation of ninety-six of these mRNAs by Gαq was opposed by co-expression of GRK5 (Figure 4b, absolute expression on the left panel, normalized expression on the right panel; Supplemental Tables 2 and 3). These results are consistent with the observed patterns of HDAC regulation by Gαq and GRK5: In Gαq mouse hearts, increased nuclear HDAC can suppress expression of its target genes. GRK5 phosphorylation of nuclear HDAC in Gαq cardiac myocytes promotes its relocalization to the cytoplasm, dis-inhibiting expression of its gene targets. As previous work has implicated Mef2-regulated genes as targets of HDAC suppression in cardiac hypertrophy ^{27, 28}, we used bioinformatics to identify 23 of these genes that are regulated by Mef2 (Supplemental Table 2). Bioinformatics analysis for other transcription factors reported to be class II HDAC targets (NFκB, MyoD, myogenin and GATA1; ^29–31) did not identify additional target genes from the regulated group, suggesting that the major impact of HDAC5 in the cardiomyopathy induced by Gαq overexpression is suppression of Mef2 targets.

Figure 4. Counter-regulation of Gq-downregulated mRNAs by GRK5.

A. Heat maps of raw expression data for all 200 mRNAs downregulated by Gq overexpression, in individual nontransgenic FVB/N (ntg) and Gq-transgenic (Gq-TG) mouse hearts (left), and corresponding data from nonfailing (NF) and failing (HF) human hearts (center and right; panel to the right shows the 41 mRNAs that are significantly downregulated in human heart failure). Mouse data are quantitative RNA-sequencing results, human data are from Affymetrix microarrays. B, Heat maps of RNA-sequencing results (left: raw data, right: normalized data) for all 200 mRNAs downregulated by Gq overexpression, from individual nontransgenic FVB/N (ntg), Gq-transgenic (Gq-TG) and GRK5/Gq compound transgenic (Gq+GRK5TG) hearts. Clustering was performed on rows (mRNAs) using Euclidean dissimilarity for raw data and Pearson’s dissimilarity for normalized data. Data are plotted using log₂ scale.

Gene Ontology (GO) analysis of the 200 Gαq downregulated mRNAs showed significant overrepresentation of mitochondrial and lipid metabolism genes (Figure 5a), with no difference in GO category distribution between the GRK5-opposed and GRK5-unaffected sets (not shown). By contrast, Ingenuity pathway analysis revealed numerous differences between GRK5-opposed and -unaffected mRNAs within the same signaling networks (Figure 5b, 5c). Several Gαq downregulated mRNAs were located in a signaling network focused on carbohydrate metabolism and energy production, which involved 13 of the 96 Gαq downregulated, GRK5-opposed mRNAs (Figure 5b). The central molecule depicted in this network is the transcription factor HNF4a, which is found in liver, gut, kidney and pancreatic beta cells ³². Mutations in HNF4a result in disrupted glucose homeostasis ³³. While we were not able to detect this transcription factor in heart with RNA sequencing, the DNA-binding half-sites of HNF4a ³⁴ have similarity to those recognized by retinoid X receptor alpha ³⁵, which was detectable in mouse heart and which was itself upregulated ~1.3-fold in response to Gαq overexpression. These data suggest that GRK5 opposition of mRNA downregulation by Gαq may influence metabolic pathways. Many of the other Gαq downregulated mRNAs were participants, but not central members, of a wider network composed of several signaling pathways (Figure 5c), suggesting that these Gαq downregulated mRNAs are likely to be modulators rather than key regulators of these pathways.

Figure 5. Functional classification of Gq-downregulated mRNAs according to GRK5 counter-regulation.

A, Gene Ontology analysis of the Gq-downregulated mRNAs. B, Ingenuity analysis diagrams for Gq-downregulated/GRK5-opposed and Gq-downregulated/GRK5-unaffected sets. Lines with arrowheads, molecule acts on a target; lines without arrowheads, binding only. Solid lines, direct interaction; dotted lines, indirect interaction. Green background, GRK5-opposed genes; blue background, GRK5-unaffected; white background, member of signaling network but not included in GRK5-opposed or –unaffected sets.

GRK5 effects on the Gαq-induced cardiomyopathy

The Gαq cardiomyopathy is robust and notoriously resistant to genetic “rescues” ^{22, 36, 37}. Nevertheless, we examined the combinatorial effects of Gαq and GRK5 on cardiac structure and function. As previously reported for isoproterenol stimulation ³, GRK5 transgenic mice (GRK5 TG) showed reduced inotropy (measured as left ventricular +dP/dt) at higher doses of the β1- adrenergic selective agonist, dobutamine, compared to non-transgenic controls (Figure 6a, top). In contrast, mice lacking GRK5 (GRK5 KO) showed normal responsiveness to dobutamine (Figure 6a, bottom) (attributable to continued presence of GRK2, a majorβ-adrenergic desensitizing kinase in the heart ³⁸). As previously described, cardiac size, echocardiographic function, and myocardial histology of these mice were normal (Table and not shown).

Figure 6. Invasive hemodynamic analysis of cardiac function in Gαq and GRK5 compound transgenic and knockout mice.

A. Left ventricular inotropic response (+dP/dt) to graded intravenous infusions of dobutamine in GRK5 transgenic (top) and GRK5 knockout (bottom) mice, compared to respective non-transgenic or wild-type controls. B. As in A., but on the Gαq transgenic background with its characteristic lack of β-adrenergic responsiveness. Each experiment is n=5–6 mice. * = P<0.05 vs non-transgenic FVB/N (panel A) or vs Gαq transgenic mouse on same background (panel B).

Table.

Echocardiographic and morphometric characteristics of Gq/GRK5 single and compound transgenic or knockout mouse hearts.

	LVEDD	LVESD	FS	HR	Vcfc	Ht/b wt
Control (FVB)	3.5 ± 0.06	1.0 ±0.02	71.4 ± 0.7	670 ± 7.2	16.6 ± 0.3	4.05±0.058
GRK5 TG	3.4 ± 0.06	1.0 ± 0.05	70.9 ± 1.5	634 ± 11.1	17.2 ± 0.1	4.29±0.058
Gq TG	4.0 ± 0.04	1.9 ± 0.16*	53.0 ± 4.0*	350 ± 19.9*	10.5 ± 0.7*	4.75±0.056*
GRK5 +Gq TG	4.2 ± 0.12*	2.1 ± 0.14*	49.5 ± 2.1*	338 ± 9.3*	9.9 ± 0.4*	4.93±0.058*
Control (C57)	3.3 ± 0.03	0.84 ± 0.02	74.0 ± 0.22	584 ± 24.2	18.24 ± 0.5	4.57±0.062
GRK5 KO	3.4 ± 0.05	0.9 ± 0.05	72.7 ± 1.4	574 ± 41.3	16.2± 1.2	4.78±0.160
GqTG	4.1 ± 0.07*	2.1 ± 0.06*	49.1 ± 0.86*	390 ± 10.7*	10.3 ± 0.35*	5.35 ± 0.10*
GRK5 KO +GqTG	4.2 ± 0.14*	2.1 ± 0.11*	50.8 ± 1.6*	403 ± 23.4*	9.2 ± 0.7*	5.04±0.130*

Studies were performed on non-anesthetized mice. KO = knockout. TG = transgenic. LVEDD = left ventricular end diastolic dimension (mm); LVESD = left ventricular end systolic dimension (mm); FS (%) = left ventricular fractional shortening, (LVEDD-LVESD)/LVEDD; HR = heart rate (beats per minute); Vcfc = velocity of left ventricular circumferential shortening corrected for heart rate. Ht/b wt = heart weight (mg)/body weight (g). For echo studies, N = 5–7 per group; for gravimetric studies, n = 8–25 per group. Mean ± s.e.m.,

^*significantly different from control (ANOVA with Tukey’s test).

We also observed the characteristic Gαq-mediated cardiac hypertrophy, left ventricular enlargement, depressed left ventricular ejection performance, and absence of β-adrenergic responsiveness (Table; Figure 6b) ^{14, 21, 22}. Cardiac overexpression of GRK5 in combination with Gαq did not alter left ventricular morphology, ejection fraction, or inotropic response at baseline or in response to dobutamine (Table; Figure 6b, top). Likewise, absence of endogenous GRK5 did not affect the cardiac hypertrophy, left ventricular dilation, or decreased ejection performance of Gαq mice (Table). However, a modest but statistically significant improvement in β1-adrenergic inotropic responsiveness was observed (Figure 2b, bottom), similar to that previously described with combined overexpression of Gαq with β2-adrenergic receptors ²².

Discussion

These studies show that GRK5 stimulates HDAC nuclear export under conditions of AT1R activation in vitro and Gαq signaling in vivo. Increased GRK5, which is characteristic of heart failure ^39–41, reversed nuclear HDAC5 accumulation and mRNA downregulation caused by cardiomyocyte-autonomous activation of Gαq signaling pathways. Thus, HDAC kinase activity of GRK5 can oppose transcriptional repression of a subset of hypertrophy/heart failure regulated genes. Because nuclear effects of GRK5-Gln41 and –Leu41 were indistinguishable, we conclude that the transcriptional co-regulatory actions of GRK5 do not contribute to clinical benefits in heart failure and hypertension that are attributed to the Leu41 polymorphic variant ^3–5.

β-adrenergic stimulation is the primary means of acutely increasing cardiac contractility. When adrenergic receptor stimulation is persistent, it can be toxic to cardiomyocytes and contribute to maladaptive cardiac remodeling and development of heart failure ¹. Accordingly, mechanisms have evolved to restrain catecholamine-mediated adrenergic signaling through receptor desensitization by G-protein receptor kinases (GRKs) ². In mammalian hearts, GRK2 and GRK5 are both expressed at high levels, and forced myocardial expression of either depresses β-adrenergic-stimulated cardiac function ^{42, 43}. GRK2 and GRK5 are reportedly upregulated in clinical and experimental heart failure ^{39–41, 44}. However, GRK2 is primarily cytosolic and upon β-receptor activation translocates to cell membranes, whereas GRK5 is constitutively membrane bound and localizes in part to the nucleus ¹⁰. These data suggest that myocardial GRK5 does not simply duplicate the functions of GRK2, but also has unique effects.

The association of the human GRK5 Leu41 polymorphism with positive outcomes in hypertension and heart failure is attributed to heightened desensitization of β-adrenergic receptors ^{3, 7}. The mechanism by which GRK5 Leu41 more effectively desensitizes β-adrenergic receptors is not known, but enhanced catalytic kinase activity has been ruled out based on normal phosphorylation of a rhodopsin substrate ³. It is not surprising, therefore, that HDAC nuclear export, which is dependent upon HDAC phosphorylation by GRK5, is also the same for GRK5 Leu41 and Gln 41.

The original observation that cardiomyocyte GRK5 can affect in vivo cardiomyocyte HDAC activity utilized a transgenic GRK5 overexpression model expressing very high (30-fold) levels of bovine GRK5 ¹¹. Because this model spontaneously develops heart failure ⁴⁵, the pathophysiological consequence of the GRK5-HDAC interaction was unclear. By comparison, our mouse model of 4-fold human GRK5 overexpression does not develop heart failure ³. This allowed us to define the consequences of GRK5 on Gαq-mediated gene expression without obfuscation by regulated genes secondarily induced by heart failure. Furthermore, by using Cos7 cells expressing human AT1R, we show that GRK5 can interact with physiological activation of Gq signaling to regulate HDAC5 nuclear export.

These studies show that GRK5 opposes, not reinforces, changes in cardiac gene expression induced by Gαq. GRK5 is an HDAC kinase. Gαq activates PKC that (directly or indirectly) phosphorylates class II HDACs ^{27, 28}. Since both factors can increase HDAC phosphorylation, we expected they would act synergistically to regulate HDAC target gene expression. However, nuclear HDAC was actually increased in Gαq-transgenic hearts. In approximately 100 papers that have used the Gαq model, we could find no published data examining HDAC localization or Mef2 activity. Increased nuclear HDAC may be the result of PKC downregulation from chronic diacylglycerol production in Gαq mice ¹⁴, or to ERK5 or calcineurin-HDAC interactions that are activated by Gq signaling ^{46, 47}. By whatever mechanism nuclear HDAC is increased, GRK5 reversed this by inducing HDAC phosphorylation and cytoplasmic localization, thus de-repressing a substantial fraction of Gαq-downregulated mRNAs. This is compelling evidence for a relevant in vivo effect of GRK5 on pathological cardiac gene expression. Reversal of part of the pathological gene profile induced by Gαq may contribute, along with its more conventional effect of “genetic β-blockade”, to the benefits attributed to GRK5 in the human condition.