Inhibition of G-protein-coupled receptor kinase 2 (GRK2) triggers the growth-promoting mitogen-activated protein kinase (MAPK) pathway

: Inhibition of G-protein-coupled receptor kinase 2 (GRK2) is an emerging treatment option for heart failure. Because GRK2 is also indispensable for growth and development, we analyzed the impact of GRK2 inhibition on cell growth and proliferation. Inhibition of GRK2 by the dominant-negative GRK2-K220R did not affect the proliferation of cultured cells. In contrast, upon xenograft transplantation of cells into immunodeficient mice, the dominant-negative GRK2-K220R or a GRK2-specific peptide inhibitor increased tumor mass. The enhanced tumor growth upon GRK2 inhibition was attributed to the growth-promoting MAPK pathway because dual inhibition of the GRK2 and RAF-MAPK axis by the Raf kinase inhibitor protein (RKIP) did not increase tumor mass. The MAPK cascade contributed to the cardioprotective profile of GRK2 inhibition by preventing cardiomyocyte death, whereas dual inhibition of RAF/MAPK and GRK2 by RKIP induced cardiomyocyte apoptosis, cardiac dysfunction, and signs of heart failure. Thus, cardioprotective signaling induced by GRK2 inhibition is overlapping with tumor growth promotion.

A common approach of GRK2 inhibition in vivo relies on expression of the carboxyl terminal domain of GRK2, i.e. the betaARK1ct (GRK2ct), which inhibits membrane translocation and activation of GRK2 by scavenging Gβγ subunits of heterotrimeric Gproteins (10)(11)(12). However, neutralization of Gβγ subunits by the GRK2ct may also exert GRK2independent effects, which could contribute to cardioprotection as well (13). The final "proofof-concept" for a beneficial profile of GRK2 inhibition came from studies applying mice with cardiac-specific deletion of GRK2 where GRK2 deficiency protected against myocardial damage (14) and prevented adverse remodeling after myocardial infarction (15).
While the beneficial cardiac profile of GRK2 inhibition is thus firmly established, mechanisms underlying cardio-protection are not fully understood. Because growth-regulatory pathways are essential for cardiomyocyte survival (16,17), we considered the impact of GRK2 inhibition on cell growth and proliferation. However, the role of GRK2 in cell growth and proliferation is not clear, because in addition to the above-mentioned growthpromoting activity, GRK2 can also exert growth inhibition leading to suppressed growth and proliferation of tumor cells (18,19).
To address the role of GRK2 and GRK2 inhibition in cell growth and proliferation, we performed experiments with cultured cells, or expanded cells in vivo after xenograft transplantation into immunodeficient non-obese diabetic mice homozygous for the severe combined immune deficiency spontaneous mutation (NOD.Scid mice). Systemic effects of GRK2 inhibition were also analyzed in vivo with transgenic mice expressing a GRK2-specific peptide inhibitor under control of the human cytomegalovirus (CMV) immediate-early promoter/enhancer, which directs ubiquitous expression of a transgene. Furthermore, transgenic mice with myocardium-specific expression of GRK2 inhibitors were generated to assess their cardio-protective profile. We report here that GRK2 inhibition triggered the growthpromoting MAPK pathway, which promoted tumor growth but also conferred cardioprotection by preventing cardiomyocyte death.

EXPERIMENTAL PROCEDURES
Generation of transgenic mice -To generate mice with myocardium-specific over-expression of RKIP, a transgene placing the PEBP1 cDNA under control of the α-myosin heavy chain (α-MHC) promoter (20) was constructed. For myocardium-specific expression of the GRK2specific peptide inhibitor (GRK-Inh), a DNA fragment encoding the peptide sequence, MAKFERLQTVTNYFITSE (21,22), was inserted into the Alpha-MyHC plasmid. The plasmid sequences were removed by NotI digestion, and the purified linear DNA (2ng/µl) was injected into fertilized oocytes of superovulated B6 (C57BL/6J) and FVB (FVB/N) mice. Oviduct transfer of the injected embryos into pseudopregnant CD-1 mice was performed according to standard procedures (23). Genomic DNA of the F0 generation was isolated from ear punch biopsies taken at 3-4 weeks of age and analyzed by PCR for integration of the transgene. Mice of two different transgenic lines each were born at Mendelian frequency and grew to adulthood normally.
To assess the systemic effect of GRK2 inhibition by the GRK2-specific peptide inhibitor in vivo, we generated transgenic mice with expression of the GRK2-specific peptide inhibitor under control of the CMV immediateearly promoter/enhancer, which directs ubiquitous expression of a transgene. To generate CMV-GRK-Inh-transgenic mice, the DNA fragment encoding the peptide sequence, MAKFERLQTVTNYFITSE, was inserted into the pcDNA3.1 plasmid (Invitrogen). Plasmid sequences were removed by MluI and DraIII digestion, and the purified linear DNA (2ng/µl) was injected into fertilized oocytes of superovulated B6 and FVB mice. Oviduct transfer of the injected embryos into pseudopregnant CD-1 mice and all subsequent steps were performed as detailed above.
Cell culture and in vivo cell expansion in NOD.Scid mice -Culture of HEK and A431 cells, generation of cell clones stably expressing GRK2, GRK2-K220R, RKIP, or the GRK2specific peptide inhibitor (21,22) was performed essentially as described (24). Cells were maintained in high-glucose DMEM with 10 % fetal bovine serum and 4 mM L-glutamine at 37 °C, in a humidified atmosphere with 5 % CO 2 . The cell proliferation assay was performed essentially as detailed previously (3). As indicated, HEK cells were also cultured in growth factor-supplemented [50 ng/ml epidermal growth factor, (EGF)] DMEM and plated on mitomycin-inactivated mouse embryonic fibroblasts (MEF) as feeder cells essentially as described (25). For in vivo cell expansion, NOD.Scid mice (age 3 months) received an injection of 6-8 x 10 6 cells/200 µl PBS (24). Two weeks (A431) or four weeks (HEK) after the injection, mice were anesthetized with ketamine/xylazine (100/10 mg/kg), perfused intracardially with physiological phosphate buffer, pH 7.2, and expanded cell clones were rapidly isolated and processed for further use. In addition, cells were also isolated from NOD.Scid mouse-expanded clones and re-cultured in DMEM as detailed above. Animal experiments were performed in accordance with the NIH guidelines, and reviewed and approved by the local committee on animal care and use (University of Zurich).
Immunohistology, immunoblotting, and protein techniques -Immunohistological detection of phospho-ERK1/2 was performed with phospho-ERK1/2-specific antibodies (phosphorylated at Thr202/Tyr204 of ERK1, and Thr185/Tyr187 of ERK2; E10 mouse mAb, Cell signalling), and GRK2 was detected with GRK2-specific antibodies (raised in rabbit against full-length recombinant GRK2 protein) on cryosections of NOD.Scid mouse-expanded HEK and A431 clones, respectively, similarly as described (24,26). For detection of phospho-ERK1/2 in hearts of transgenic mice, we used paraffin-embedded sections. Nuclear fragmentation as a marker of apoptosis was determined in situ by TdT-mediated dUTPbiotin nick end labeling (TUNEL) technology (Roche Diagnostics, Germany) with paraffinembedded sections prepared from transgenic hearts as detailed previously (26,27). The TUNEL technology was also used to determine the nuclear fragmentation of neonatal mouse cardiomyocytes, which were isolated from B6 mice, Tg-RKIP and Tg-GRK-Inh mice 1-3 days after birth essentially as described (9). All sections/cells were imaged with a Leica DMI6000 microscope equipped with a DFC420 camera. Immunoblotting was used to determine the protein level of GRK2, GRK2-K220R, ERK1 (MAPK3), ERK2 (MAPK1), and phospho-ERK1/2 in cells, NOD.Scid mouseexpanded clones and organ tissue from transgenic mice as described (24,26). We also used standard immunoblotting and immunohistology techniques (9,27,28) applying RKIP-specific antibodies (affinity-purified polyclonal antibodies raised in rabbit against full-length recombinant RKIP protein) to detect the over-expressed RKIP protein in RKIP-transgenic hearts. Co-enrichment of Raf1 and GRK2 with RKIP was performed similarly as described (9). The GRK2-specific peptide inhibitor (GRK-Inh) was detected in immunoblot after Tricine-SDS-PAGE with anti-GRK-Inh antibodies (affinity-purified polyclonal antibodies raised in rabbit against the KLHcoupled GRK-Inh peptide). Nuclear phospho-ERK1/2 levels were quantified with the nuclear fraction of NOD.Scid mouse-expanded HEK and A431 clones, and organ tissue of transgenic and non-transgenic mice. The nuclear fraction was adjusted to a protein concentration of 0.2 µg/µl, solubilized by buffer (500 mM NaCl, 1 mM EDTA, 1 % Triton X100, 1 % SDS, 50 mM Tris pH 7.4) supplemented with protease/phosphatase inhibitors and assayed for phospho-ERK1/2 by immunoblotting, or a sandwich ELISA according to the instructions of the manufacturer (Pierce). Nuclear samples were normalized to histone H2B.
For subcellular fractionation, dispersed NOD.Scid mouse-expanded clones or mouse organ tissue were suspended in ice-cold buffer containing 0.25 M sucrose, 20 mM HEPES, pH 7.4, 10 mM KCl, 2 mM MgCl 2 , 1 mM DTT, supplemented with protease/phosphatase inhibitors, and homogenized on ice with a dounce homogenizer. After centrifugation for 10 min at 4 °C (750 x g), the nuclear pellet was washed twice with ice-cold buffer, and stored as the nuclear fraction at -80°C. The purity of the nuclear fraction was controlled by immunoblot detection of histone H2B and the absence of cytosolic proteins.
NanoLC-ESI-MS/MS -To enrich proteins interacting with the GRK2-specific peptide inhibitor, GRK-Inh-expressing tumors were pulverized under liquid nitrogen. After solubilization for 30 min at 4 °C with solubilization buffer (1 % sodium deoxycholate, 0.05 % SDS, 0.05 % Tween-20 in PBS, pH 7.4, supplemented with protease inhibitors), insoluble material was removed by centrifugation. The supernatant was diluted 1:5 in PBS (supplemented with protease inhibitors) and subjected to affinity chromatography with anti-GRK-Inh-antibodies (6 mg affinity-purified IgG/ml Affigel 10). After overnight incubation at 4 °C, unbound proteins were removed by extensive washing with PBS (20 column volumes), and bound proteins were eluted with 0.25 M NH 4 OH/10% dioxane (pH 11). The pH was immediately adjusted to pH 7.4, eluted proteins were concentrated by acetone precipitation, dissolved by 8M urea and subjected to 8 % urea-containing SDS-PAGE under reducing conditions. After coomassie brilliant blue staining, the GRK2-reactive band was cut and subjected to nanoLC-ESI-MS/MS. Protein identification using nanoLC-ESI-MS/MS was performed by Proteome Factory (Proteome Factory AG, Berlin, Germany). The MS system consisted of an Agilent 1100 nanoLC system (Agilent, Boeblingen, Germany), PicoTip emitter (New Objective, Woburn, USA) and an Esquire 3000 plus ion trap MS (Bruker, Bremen, Germany). The cut protein band was in-gel digested by trypsin (Promega, Mannheim, Germany) and applied to nanoLC-ESI-MS/MS. After trapping and desalting the peptides on enrichment column (Zorbax SB C18, 0.3 x 5 mm, Agilent) using 1 % acetonitrile/0.5 % formic acid solution for five minutes, peptides were separated on Zorbax 300 SB C18, 75 µm x 150 mm column (Agilent) using an acetonitrile/0.1 % formic acid gradient from 5 % to 40 % acetonitrile within 40 min. MS spectra were automatically taken by Esquire 3000 plus according to the manufacturer`s instrument settings for nanoLC-ESI-MS/MS analyses. Proteins were identified using MS/MS ion search of Mascot search engine (Matrix Science, London, England) and nr protein database (National Center for Biotechnology Information, Bethesda, USA). Ion charge in search parameters for ions from ESI-MS/MS data acquisition were set to "1+, 2+ or 3+" according to the instrument`s and method`s common charge state distribution.
Transthoracic echocardiography -Transthoracic echocardiography was performed with a Vivid 7 echocardiograph system (GE Healthcare) and a 12 MHz linear array transducer similarly as described (27,29). Transthoracic echocardiography was used to characterize the cardiac function of transgenic and non-transgenic B6 and FVB mice without and with 4 weeks of chronic pressure overload imposed by abdominal aortic constriction, AAC, performed as detailed previously (27).
Microarray gene expression profiling -Whole genome microarray gene expression profiling of cells, NOD.Scid mouse-expanded clones and heart tissue of RKIP-transgenic, GRK-Inh-transgenic and B6 mice was performed as described (24,30). We used GeneChip Human genome U133 Plus 2.0 Arrays for human cells and Mouse genome MG430 2.0 Arrays for mouse cardiac tissue (Affymetrix). Gene ontology (GO) analyses of microarrray data were performed with GCOS/RMA processed data using GeneSpring GX software (Agilent). Data were compared between groups using the unpaired two-tailed Student`s t test. P values of <0.05 were considered significant unless specified otherwise. All microarray gene expression data were deposited to the NCBI GEO database (accession numbers GSE42753 and GSE42771).
Selected gene expression data were confirmed by real-time qRT-PCR using a LightCycler 480 (Roche). Sequences of the forward and reverse primers of genes studied were as follows: Confocal FRET imaging -The GRK2mediated interaction of enhanced yellow fluorescent protein (EYFP)-tagged β-arrestin1 with B2R-Cerulean in transfected HEK cell clones without (control) or with expression of GRK2 or the dominant-negative GRK2-K220R, respectively, was determined in the absence or presence of bradykinin (1 µM, 8 min) and quantified by confocal FRET imaging and acceptor photobleaching with a confocal laserscanning microscope (SP5-CLSM, Leica) similarly as described (21).
Plasmids -For transfection of HEK and A431 cells, we used the eukaryotic expression plasmid pcDNA3.1 (Invitrogen) and inserted the following cDNAs: ARRB1-EYFP (EYFP fused in-frame with the C-terminal amino acid codon of ARRB1); B2R-Cerulean (21), (Cerulean fused with a linker encoding GlyGlyGlyGlyGly inframe with the codon for the C-terminal amino acid 391 of BDKRB2); RKIP [cDNA encoding PEBP1 (28)]; GRK2 (cDNA encoding ADRBK1); GRK2-K220R (cDNA encoding ADRBK1 with point mutation exchanging lysine 220 for arginine); GRK-Inh [DNA encoding the sequence MAKFERLQTVTNYFITSE (21), a GRK2-specific peptide inhibitor (21,22)]. Expression of GRK2 was down regulated by RNA interference (RNAi) by POL II driven expression of a GRK2-targeting miR. The following hairpin-forming oligonucleotides were hybridized to form a 60-bp duplex and inserted into pcDNA6 Statistics -Results are presented as mean ± s. d. unless indicated otherwise. Unpaired twotailed Student`s t-test was used to calculate P values. Analysis of variance was performed with Prism (GraphPad). Statistical significance was set at a P value of < 0.05.

RESULTS
Dominant-negative activity of the kinase-deficient GRK2-K220R mutant in HEK cells-To analyze the impact of GRK2 and GRK2 inhibition on cell growth, we used the kinase-deficient GRK2-K220R, which is reported to act as a dominant-negative mutant (31). We generated HEK cell clones stably expressing GRK2 or the kinase-deficient GRK2-K220R (Fig. 1A). Protein levels were similar because GRK2 and GRK2-K220R exerted comparable kinase-independent functions such as inhibition of a Gαq/11-mediated calcium signal stimulated by the bradykinin B2 receptor, B2R (Fig. 1B). To confirm kinase activity of GRK2, we expressed a GRK2-specific peptide inhibitor (21,22), and determined its effect on GRK2-mediated desensitization of the calcium signal stimulated by B2R, which is a kinase substrate of GRK2 (32). In agreement with kinase inhibition, GRK2-mediated desensitization of the B2R-stimulated calcium signal was partially reversed by expression of the GRK2-specific peptide inhibitor whereas the kinase inhibitor did not affect the kinase-inactive GRK2-K220R (Fig. 1B).
We next analyzed whether GRK2-K220R exerted dominant-negative activity in HEK cells and inhibited the endogenously expressed GRK2 ( Fig. 2A). As a kinase effect of GRK2, we determined the GRK2-triggered interaction of Cerulean-tagged B2R with β-arrestin1-EYFP by fluorescence resonance energy transfer (FRET) measurement. Confocal FRET imaging revealed that the kinase-deficient GRK2-K220R completely prevented the bradykinin-stimulated interaction of B2R-Cerulean with β-arrestin1-EYFP indicating dominant-negative activity (Fig. 2B, C). As a control, the expression of GRK2 did not decrease the bradykinin-enhanced FRET intensity of the B2R-Cerulean interaction with β-arrestin1-EYFP (Fig. 2B, C) Dominant-negative GRK2-K220R enhanced the growth of NOD.Scid mouseexpanded HEK clones-We determined whether GRK2 or the dominant-negative GRK2-K220R affected the proliferation of HEK cells. In agreement with previous results (3), neither GRK2 nor GRK2-K220R affected the proliferation rate of cultured HEK cells (Fig.  3A).
As GRK2-dependent growth control is active in vivo (2-4), we performed cell expansion in immunodeficient NOD.Scid mice because the xenograft transplantation model had previously been used to overcome limitations of cell culture (24). In contrast to cultured cells, GRK2 inhibition by the dominant-negative GRK2-K220R led to a strongly increased growth rate in vivo upon cell expansion in NOD.Scid mice whereas GRK2-expressing clones showed a reduced cell mass (Fig. 3B). As a control, immunoblotting and immunohistology demonstrated comparable protein levels of GRK2-K220R and GRK2, respectively (Fig. 3C, D).
We searched for the mechanism underlying the growth control by GRK2. The mitogen-activated protein kinase (MAPK) pathway triggered by RAS-dependent activation of RAF and MEK is a universal signal transduction cascade involved in cell survival, growth, and proliferation (33,34). Because MAPK pathway activation is also a common feature of many GRK2 receptor-substrates (35), we analyzed the activation of the MAPK, ERK1/2, of in vivo-expanded HEK clones. The GRK2-K220R-induced increase in HEK cell mass correlated with a high level of activated phospho-ERK1/2 (Fig. 3E). Vice versa, the reduced growth of GRK2-expressing clones was accompanied by decreased ERK1/2 phosphorylation (Fig. 3E).
Up-regulation of MAPK pathway genes in NOD.Scid mouse-expanded HEK clones expressing GRK2-K220R-To decipher the basis of the differential growth regulation of cultured cells and NOD.Scid mouse-expanded clones, we performed microarray gene expression profiling. In agreement with sustained ERK1/2 activation (cf. Fig. 3E), microarray gene expression analysis showed that MAPK pathway target genes such as FOS were specifically increased in NOD.Scid mouse-expanded clones expressing dominant-negative GRK2-K220R compared to GRK2-expressing clones ( Real-time quantitative reverse transcription PCR (real-time qRT-PCR) confirmed substantial FOS expression after NOD.Scid mouse expansion while FOS was low in cultured cells (Fig. 5B). Real-time qRT-PCR also confirmed that GRK2-K220R up-regulated FOS in NOD.Scid mouse-expanded HEK clones compared to clones expressing GRK2 (Fig. 5B, left panel). In contrast to in vivo cell expansion, GRK2 and GRK2-K220R did not affect the expression of FOS in cultured HEK cells, either before or after in vivo expansion in NOD.Scid mice (Fig. 5B, middle and right panel). These experiments further support that in vivo cell expansion in NOD.Scid mice was essential to manifest MAPK target gene modulation by GRK2 and GRK2-K220R, respectively.
Dominant-negative GRK2-K220R of NOD.Scid mouse-expanded HEK clones enhanced MAPK activation and nuclear translocation -The high expression of the nuclear MAPK target, FOS, pointed to sustained activation and nuclear translocation of the MAPkinases, ERK1/2 (34) upon GRK2 inhibition by GRK2-K220R. In agreement with that notion, immunohistology analysis with phospho-ERK1/2-specific antibodies revealed an increased total phospho-ERK1/2 level with a substantial amount of phospho-ERK1/2 in the nuclei of a NOD.Scid mouse-expanded GRK2-K220R-expressing HEK clone compared to the control or GRK2-expressing clone (Fig. 5C). Quantitative immunoblot evaluation confirmed those data and showed a high level of activated phospho-ERK1/2 in the nuclear fraction of GRK2-K220R-expressing clones whereas nuclear phospho-ERK1/2 was low in GRK2expressing clones (Fig. 5D).
Trophic effects of MEF feeder cells reconstituted GRK2-dependent growth control in vitro -To reconstitute the trophic effect of NOD.Scid mice on HEK growth in vitro, growth factor-supplemented HEK cells were plated on mouse embryonic fibroblasts (MEFs) as feeder cells (25). Culture of HEK cells with MEF feeder cells revealed a GRK2-mediated reduction of HEK cell proliferation whereas dominant-negative GRK2-K220R promoted cell proliferation (Fig. 6A). The MEK inhibitor PD0325901 prevented the GRK2-mediated growth regulation indicating dependence on the MAPK pathway (Fig. 6A). In agreement with involvement of the MAPK pathway, cell proliferation control by GRK2 correlated with regulation of the nuclear phospho-ERK1/2 level (Fig. 6B). Similarly as in NOD.Scid mouseexpanded clones, GRK2 reduced the nuclear phospho-ERK1/2 level while GRK2 inhibition by GRK2-K220R significantly increased the amount of nuclear phospho-ERK1/2 of HEK cells plated on MEF feeder cells (Fig. 6B). These experiments strongly suggest that the apparent difference between in vivo and in vitro experiments could be due to the lack of sustained MAPK-activating growth-promoting stimuli under standard cell culture conditions.
The RAF-MAPK axis triggered by GRK2 inhibition promotes tumor growth-Sustained MAPK pathway activation induces and/or promotes growth of malignant tumors such as squamous-cell carcinoma (37). Therefore we used squamous cell carcinoma A431 cells, to analyze whether the GRK2-K220R-dependent enhancement of the MAPK pathway affected tumor growth. A431 clones with comparable protein levels of GRK2 and GRK2-K220R, respectively, were used (Fig. 7A, left panel). Analogous to NOD.Scid mouse-expanded HEK clones, immuno-techniques revealed a high protein level of activated phospho-ERK1/2 in the nuclei of NOD.Scid-expanded A431 tumors expressing dominant-negative GRK2-K220R whereas phospho-ERK-1/2 was low in GRK2expressing tumors (Fig. 7B,C). In agreement with growth-promoting MAPK activation, expression of the MAPK target, FOS, was induced in GRK2-K220R-expressing tumors (Fig. 7D), and A431 tumor mass increased (Fig.  7E).
To further investigate whether GRK2 inhibition triggered the growth-promoting MAPK pathway, we used a GRK2-specific peptide inhibitor (cf. Fig. 1B and ref. 21,22). Expression of the GRK2-specific peptide inhibitor (Fig. 7A, right panel) significantly increased the level of activated phospho-ERK1/2 (Fig. 7B,C), and induced the nuclear MAPK target FOS similarly as did GRK2-K220R (Fig.  7D, panel 3 vs. 2). Concomitantly, tumor growth was enhanced (Fig. 7F). Thus, GRK2 inhibition either by a dominant-negative mutant or a peptide inhibitor activated the growth-promoting MAPK pathway, induced the expression of the MAPK target, FOS and enhanced tumor growth.
The MAPK pathway is triggered by the proto-oncogene RAF.
In agreement with the requirement of the RAF/MEK/ERK pathway for enhanced tumor growth upon GRK2 inhibition, the MEK inhibitor PD0325901 significantly reduced the tumor mass of GRK2-inhibitor-expressing A431 tumors whereas the mass of RKIP-expressing A431 tumors was not affected (Fig. 7I). Moreover, MAPK activation induced by the GRK2-specific inhibitor was dependent on GRK2 because the GRK2-specific peptide inhibitor did not significantly increase the nuclear phospho-ERK1/2 level upon downregulation of GRK2 by RNA interference (Fig.  7K). As an indicator for comparable GRK2 inhibition, the target gene FN (cf. Fig. 5A) was similarly up-regulated by the three different GRK2 inhibitors whereas wild-type GRK2 reduced the expression of FN (Fig. 7L).
Together the experiments strongly suggest that the RAF-MAPK axis triggered by GRK2 inhibition promoted the growth of malignant A431 tumors.
Enrichment of GRK2 by GRK-Inhspecific immuno-affinity chromatography and identification by nano-LC-ESI-MS/MS -To analyze the interaction of the GRK2-specific peptide inhibitor (GRK-Inh) with GRK2 in vivo, we performed affinity chromatography with GRK-Inh-reactive antibodies. Proteins were immuno-affinity enriched from GRK-Inhexpressing tumor tissue and separated by SDS-PAGE. Silver staining of enriched proteins revealed a predominant protein band with an apparent molecular weight of 79 ± 3 kDa, which was not enriched by the control column (Fig.  8A). Immunoblotting demonstrated that the enriched protein co-migrated with the GRK2immunoreactive protein band (Fig. 8A). The protein band was excised from the gel, and subjected to identification by nano-LC-ESI-MS/MS. With 27 matching peptides, the Mascot search engine identified the human betaadrenergic receptor kinase 1 (GRK2) with the highest probability score (Fig. 8B and Supplemental data). Regions of identified peptides matching with the human GRK2 protein sequence are shown (Fig. 8B). Together these experiments provide strong evidence that the GRK2-specific peptide inhibitor interacts with the GRK2 protein in vivo.
MAPK pathway activation in transgenic mice with systemic expression of the GRK2specific peptide inhibitor -To investigate whether GRK2 inhibition by the GRK2-specific peptide inhibitor also regulated the MAPK pathway in non-tumor tissue, we generated transgenic mice expressing the GRK2-specific peptide inhibitor under control of the CMV immediate-early promoter/enhancer, which directs ubiquitous expression in transgenic mice (39). Two different transgenic mouse lines were identified, which showed high and low level, respectively, of the GRK2-specific peptide inhibitor, in different organs, i.e. kidney, lung, thymus and heart (Fig. 9A).
Organs with transgenic expression of the GRK2-specific peptide inhibitor also displayed an increased Fos protein level compared to nontransgenic B6 controls, suggesting enhanced activation of the MAPK pathway upon GRK2specific peptide inhibitor expression (Fig. 9B). In agreement with MAPK activation, nuclear and cytosolic protein levels of activated phospho-ERK1/2 were significantly higher in organs with GRK2-specific peptide inhibitor expression compared to B6 control tissue while total cytosolic ERK1/2 protein was not different (Fig. 9C,D). These experiments show that the GRK2-specific peptide inhibitor promoted activation of the MAPK pathway and Fos induction in vivo, not only in xenografted tumor tissue, but also in different organs of transgenic mice.
Co-enrichment studies revealed that kinase-inhibited GRK2 (inhibited by the GRK2-specific peptide inhibitor) or kinase-deficient GRK2-K220R showed an increased interaction with activated phospho-ERK1/2 compared to kinase-active GRK2 of A431 tumor tissue (Fig. 9E, left and right panel). The increased interaction of kinase-inhibited GRK2 with phospho-ERK1/2 was also detected in different organs of mice with transgenic expression of the GRK2-specific peptide inhibitor (Fig. 9F). These findings suggest that kinase-inhibited GRK2 stabilized phospho-ERK1/2.
The phospho-ERK1/2-stabilizing activity of kinase-inhibited GRK2 could contribute to sustained phospho-ERK1/2 activation required for MAPK target gene induction of A431 tumors and organs from transgenic mice.
GRK2-inhibition by transgenic RKIP or GRK-Inh expression in hearts of transgenic mice-Enhanced activation of the MAPK pathway can exert cardioprotection (16). Because heart tissue from transgenic mice with GRK2-specific peptide inhibitor expression under control of the CMV promoter showed increased MAPK pathway activation (cf. Fig. 9), we investigated the cardiac profile of the GRK2specific peptide inhibitor relative to RKIP, which is a dual-specific GRK2 and RAF/MAPK inhibitor. To this end mice with myocardiumspecific expression of RKIP or the GRK2specific peptide inhibitor, respectively, were generated (Fig. 10A). Hearts of transgenic founder lines (two different lines for each transgene) showed increased levels of the RKIP protein and the GRK2-specific peptide inhibitor, respectively (Fig. 10B). GRK2 inhibition was effective in both models as evidenced by microarray gene expression profiling: more than 60 % of regulated probe sets of RKIP-transgenic hearts showed concordant regulation with hearts expressing the GRK2-specific peptide inhibitor (Fig. 10C, left panel, and Fig. 11). Moreover, the extent of gene regulation was comparable between RKIP-transgenic and GRK-Inhtransgenic hearts (Fig. 11). These in vivo observations are compatible with equivalent in vitro potencies of RKIP and GRK-Inh in inhibiting GRK2 (9,22).
The GRK2-specific peptide inhibitor enhanced whereas RKIP inhibited the MAPK pathway in vivo -Only 3 probe sets were upregulated by the GRK2-specific peptide inhibitor and down regulated by RKIP, i.e. the MAPK target genes Fos, Egr-1, and Ctgf (Fig. 10C, right panel). Altered Fos expression was confirmed by immunoblotting, which showed increased cardiac Fos protein in transgenic hearts expressing the GRK2-specific peptide inhibitor whereas RKIP decreased Fos (Fig.  10D). The induction of Fos correlated with increased MAPK activation as demonstrated by increased (nuclear) phospho-ERK1/2 levels in transgenic hearts expressing the GRK2-specific peptide inhibitor while phospho-ERK1/2 was reduced by RKIP (Fig. 10E-G). As a control, the protein level of over-expressed RKIP was sufficient to scavenge the entire cardiac GRK2 and Raf1 pools (Fig. 10H). Together the experiments show that transgenic RKIP overexpression induced Raf1-MAPK axis inhibition whereas the GRK2-specific peptide inhibitor enhanced the MAPK pathway.
Inhibition of the MAPK pathway by transgenic RKIP over-expression triggered signs of heart failure in B6 mice-As the RAF-MAPK axis is essential for cardiomyocyte survival (16,17,40), cardiomyocyte death was assessed by TUNEL staining. There was a significant increase in the number of TUNEL-positive nuclei in RKIP-transgenic hearts whereas signs of cardiomyocyte apoptosis were low in GRK2inhibitor-expressing hearts (Fig. 12A,C).
The cardiotoxic effect of RKIP was confirmed with neonatal cardiomyocytes isolated from RKIP-transgenic hearts (Fig. 12F). RKIP over-expression induced a 9-fold increase in TUNEL-positive cardiomyocytes compared to neonatal cardiomyocytes isolated from GRK-Inh-transgenic mice (Fig. 12F). RKIP-dependent cardiomyocyte apoptosis was attributed to MAPK pathway inhibition, because a MEK inhibitor strongly increased cardiomyocyte death of B6 control cardiomyocytes but had no substantial effect on RKIP-transgenic cardiomyocytes (Fig. 12G).
Inhibition of GRK2 in FVB mice by transgenic expression of RKIP or GRK-Inh-We asked whether the genetic background of B6 mice was linked to the different phenotype of the two GRK2 inhibitors. Transgenic FVB mice with myocardium-specific RKIP expression were generated (Fig. 13A-C). Analogous to RKIP-transgenic B6 mice, the TUNEL assay revealed enhanced cardiomyocyte apoptosis of RKIP-transgenic FVB mice (Fig. 13A). Concomitantly, cardiac hypertrophy with dilatation and cardiac dysfunction developed in RKIP-transgenic FVB mice (Fig. 13A,B). Taken together these data show that dual inhibition of the GRK2-RAF/MAPK axis by RKIP could be cardiotoxic in B6 and FVB mice, independent of the genetic background of the mouse.
On the other hand, hearts of transgenic FVB mice expressing the GRK2-specific peptide inhibitor (Fig. 13D) were not dilated and showed a high left ventricular ejection fraction similarly as B6 mice with transgenic GRK-Inh expression (Fig. 13E,F vs. Fig. 12). In agreement with cardioprotective GRK2 inhibition (10)(11)(12)(13)(14)(15), the GRK2-specific peptide inhibitor retarded the development of chronic pressure overloadinduced cardiac hypertrophy with dilatation and cardiac dysfunction imposed by 4 weeks of abdominal aortic constriction (Fig. 13E,F). Thus, the GRK2-specific peptide inhibitor had a comparable cardiac profile in FVB and B6 mice and was protective against pressure overloadinduced cardiac damage.

DISCUSSION
Our study revealed a previously unrecognized role of the desensitizing kinase, GRK2, in restraining the MAPK pathway. GRK2-mediated desensitization of MAPKactivating GPCRs and non-GPCR substrates is well established (1,35,41). In addition, prevention of nuclear MAPK signaling could be mediated by the GRK2-dependent recruitment of β-arrestin, which retains MAPK in the cytosol (42)(43)(44). This mechanism is not only valid for GPCRs (42,43) but also reduces nuclear translocation of ERK1/2 upon receptor tyrosine kinases co-stimulation such as the epidermal growth factor receptor (44).
While GRK2 suppressed MAPK activation and nuclear translocation, GRK2 inhibitors enhanced the MAPK pathway, induced nuclear ERK1/2 targets, notably the proto-oncogene FOS, and promoted tumor growth. Activation of the nuclear MAPK pathway upon GRK2 inhibition was not only observed in two different xenograft cell transplantation models but also occurred in vivo, in different organs of transgenic mice upon systemic GRK2 inhibition by transgenic expression of a GRK2-specific peptide inhibitor.
Increased MAPK activation and target gene induction upon GRK2 inhibition could be due to β-arrestin scavenging/neutralization by kinase-inhibited GRK2. Thereby the formation of receptor/β-arrestin complexes is prevented as well as subsequent receptor/MAPK desensitization. In this respect, kinase-inhibited GRK2 acts as a dominant negative protein, which could differ from GRK2 knock-down because siGRK2 may trigger a shift to GRK5/6β-arrestin-mediated MAPK activation (45 (10)(11)(12)(13)(14)(15). In addition to β-arrestin neutralization, kinase-inhibited GRK2 could lead to sustained MAPK pathway activation essential for target gene induction (34), by stabilization of activated phospho-ERK1/2, which was detected by co-enrichment studies in A431 tumor tissue and various organs of transgenic mice with systemic expression of the GRK2-specific peptide inhibitor.
The tumor growth-promoting MAPK activity triggered by GRK2 inhibition overlapped with the cardioprotective activity of the MAPK cascade, which protects the myocardium against death-promoting stimuli (16,46). Notably, activation of ERK1/2 protects the heart against pressure overload-induced hypertrophic cardiomyopathy by stimulating cardiomyocyte survival and proliferation (46). Vice versa, genetic inhibition of ERK1/2 enhanced cardiomyocyte apoptosis and predisposed to the development of pressure overload-induced heart failure (47). In agreement with those data, the pro-survival function of the MAPK cascade could contribute to the beneficial cardiac profile of GRK2 inhibition because dual inhibition of the GRK2 and RAF/MAPK axis by RKIP was detrimental and induced signs of heart failure in transgenic mouse lines with two different genetic backgrounds, most likely due to cardiotoxic RAF inhibition (17,40) and lack of cardioprotective MAPK activation (16,46,48).
On the other hand, GRK2 inhibition without MAPK inhibition, by transgenic GRK2specific peptide inhibitor expression, was beneficial and retarded the development of chronic pressure overload-induced cardiac hypertrophy with dilatation and cardiac dysfunction. The cardioprotective activity of the GRK2-specific peptide inhibitor was comparable to the profile of GRK2 inhibition induced by the GRK2ct. Analogous to the GRK2-specific peptide inhibitor, transgenic expression of the GRK2ct also prevented the development of left ventricular dilatation and hypertrophy triggered by long-term chronic pressure overload (49).
Moreover, inhibition of GRK2 by the Gβγscavenging GRK2ct, also preserved the activated MAPK pathway under aortic constriction (50).
In addition to the cardio-protective MAPK pathway (48), GRK2 inhibition could enhance cardiomyocyte survival by increasing AKT-mediated induction of nitric oxide (51). Such a mechanism is active in vivo and contributes to cardio-protection of GRK2cttransgenic mice against acute myocardial ischemia/reperfusion injury (51). Since prosurvival signaling by MAPK and AKT is integrated in the heart by specific scaffold proteins (52), future studies will have to determine the impact of the GRK2-specific peptide inhibitor on synergistic enhancement of MAPK-and AKT-mediated pro-survival pathways.
While we found analogous results of the GRK2-specific peptide inhibitor and the dominant-negative GRK2-K220R mutant regarding MAPK activation and tumor cell proliferation, the cardiac phenotype of GRK2-K220R still needs to be determined. Since GRK2-K220R is capable to inhibit GRK2 as a dominant-negative protein and by scavenging of Gβγ subunits similarly as does GRK2ct, GRK2-K220R is also expected to exert cardioprotection.
Taken together, our study investigated the in vivo profile of GRK2 inhibition by a GRK2-specific peptide inhibitor and the dominant-negative GRK2-K220R mutant. Our study provides strong evidence that GRK2 inhibition induces activation of the growthpromoting MAPK pathway in vivo, which seems critical for cardiomyocyte survival. In view of the well-established cell proliferation and cell growth-stimulating activity of the MAPK pathway, future efforts will have to balance the cardiomyocyte survival profile of GRK2 inhibitors with the inherent risk of tumor cell growth. New approaches could focus on the development of dual-specific inhibitors, which stimulate GRK2-dependent cardiomyocyte survival but inhibit tumor cell proliferation.    , and of re-cultured cells after NOD.Scid mouse expansion (Ex-Scid; right panels). Gene expression data from GRK2-expressing (GRK-1, GRK-2) and GRK2-K220R-expressing HEK clones (K220R-1, K220R-2) are shown. Microarray probe sets were selected according to the following criteria: (i) significant difference (P≤0.01) between NOD.Scid mouse-expanded clones expressing GRK2-K220R (Scid-K220R-1; Scid-K220R-2) and GRK2 (Scid-GRK-1; Scid-GRK-2), (ii) >1.6-fold up-regulation in GRK2-K220R-expressing clones relative to GRK2-expressing clones, and (iii) involvement in the MAPK pathway according to GO analysis. Probe sets with significant difference (K220R-relative to GRK2expressing HEK cells/clones) are marked in bold. The following genes were identified: FOS, v-fos FBJ murine osteosarcoma viral oncogene homolog; KLF2, Kruppel-like factor 2; S100A6, S100 calcium-binding protein A6, calcyclin; TRIB1, tribbles homolog 1; DCBLD2, discoidin, CUB and LCCL domain containing 2; CRYAB, crystallin, alpha B; PDGFB, platelet-derived growth factor beta polypeptide; EGR1, early growth response 1; DUSP1, dual specificity phosphatase-1. Signal intensities of probe sets detecting the TGFβ target fibronectin (FN) were significantly lower in all GRK2-expressing HEK cells/clones indicative of TGFβ-antagonistic activity of GRK2. Selected data of two gene chips are presented for each group (two different clones/cell preparations per gene chip). The probe set detecting GAPDH is shown as a control. . Expression data were normalized to β-actin (FOS/actin) and represent mean ± s.d., n=3. C. Detection of phospho-ERK1/2 by immunohistology with phospho-ERK1/2-specific antibodies in sections of a NOD.Scid mouse-expanded control HEK clone (Cont.), and HEK clones expressing GRK2 (GRK2) or GRK2-K220R (K220R). Nuclei were stained with hematoxylin (HE), bar: 50 µm. D. Nuclear phospho-ERK1/2 levels of NOD.Scid mouseexpanded HEK clones without (Cont., set to 100 %) or with expression of GRK2 or GRK2-K220R (all samples were normalized to histone H2B). Data are mean ± s.d., n=3. The right panels show a representative immunoblot experiment. E. Increased interaction of β-arrestin with GRK2-K220R compared to GRK2 as determined by immuno-enrichment of GRK2 (IP) from GRK2-expressing or GRK2-K220R-expressing HEK clones, respectively, followed by immunoblot (IB) detection of coenriched β-arrestin and enriched GRK2. Bars represent mean ± s.d., n=3. The right panels show a representative experiment (middle, lower panels) and a control immunoblot detecting β-arrestin (upper panel). All results are representative of 2-3 different HEK cell clones each.    Inh-transgenic hearts (Tg-GRK-Inh) with concordant up-regulation (upper panel) or down-regulation (lower panel) compared to B6 control mice (-fold change relative to B6: ≥2 or ≤-2, and P≤0.01). Bars represent mean of two different gene chips (three mice/gene chip). All data were normalized to beta actin.