Overlapping and Opposing Functions of G Protein-coupled Receptor Kinase 2 (GRK2) and GRK5 during Heart Development*

Background: GRK2 and GRK5 differentially control function and morphology of the adult heart. Results: We found that GRK2 and GRK5 distinctly govern myocardial development and function. Conclusion: GRK2 and GRK5 function during heart development. Significance: We found a differential impact of GRKs on embryonic development and adult physiology of the heart. G protein-coupled receptor kinases 2 (GRK2) and 5 (GRK5) are fundamental regulators of cardiac performance in adults but are less well characterized for their function in the hearts of embryos. GRK2 and -5 belong to different subfamilies and function as competitors in the control of certain receptors and signaling pathways. In this study, we used zebrafish to investigate whether the fish homologs of GRK2 and -5, Grk2/3 and Grk5, also have unique, complementary, or competitive roles during heart development. We found that they differentially regulate the heart rate of early embryos and equally facilitate heart function in older embryos and that both are required to develop proper cardiac morphology. A loss of Grk2/3 results in dilated atria and hypoplastic ventricles, and the hearts of embryos depleted in Grk5 present with a generalized atrophy. This Grk5 morphant phenotype was associated with an overall decrease of early cardiac progenitors as well as a reduction in the area occupied by myocardial progenitor cells. In the case of Grk2/3, the progenitor decrease was confined to a subset of precursor cells with a committed ventricular fate. We attempted to rescue the GRK loss-of-function heart phenotypes by downstream activation of Hedgehog signaling. The Grk2/3 loss-of-function embryos were rescued by this approach, but Grk5 embryos failed to respond. In summary, we found that GRK2 and GRK5 control cardiac function as well as morphogenesis during development although with different morphological outcomes.

G protein-coupled receptor kinases 2 (GRK2) and 5 (GRK5) are fundamental regulators of cardiac performance in adults but are less well characterized for their function in the hearts of embryos. GRK2 and -5 belong to different subfamilies and function as competitors in the control of certain receptors and signaling pathways. In this study, we used zebrafish to investigate whether the fish homologs of GRK2 and -5, Grk2/3 and Grk5, also have unique, complementary, or competitive roles during heart development. We found that they differentially regulate the heart rate of early embryos and equally facilitate heart function in older embryos and that both are required to develop proper cardiac morphology. A loss of Grk2/3 results in dilated atria and hypoplastic ventricles, and the hearts of embryos depleted in Grk5 present with a generalized atrophy. This Grk5 morphant phenotype was associated with an overall decrease of early cardiac progenitors as well as a reduction in the area occupied by myocardial progenitor cells. In the case of Grk2/3, the progenitor decrease was confined to a subset of precursor cells with a committed ventricular fate. We attempted to rescue the GRK loss-of-function heart phenotypes by downstream activation of Hedgehog signaling. The Grk2/3 loss-of-function embryos were rescued by this approach, but Grk5 embryos failed to respond. In summary, we found that GRK2 and GRK5 control cardiac function as well as morphogenesis during development although with different morphological outcomes.
The seven members of the G protein-coupled receptor kinase (GRK) 2 family evolved from two ancestor-like kinases.
One of those initial "GRKs" subsequently gave rise to the GRK2/GRK3 lineage, whereas the other formed the subfamily containing the GRK1/7 and GRK4/5/6 subclasses (1). This division into a GRK2-like lineage and another resembling GRK5/6 has remained on a functional level up to the present. In this work, we present evidence that members of these two distinct subfamilies produce different physiological outcomes during embryonic development.
GRKs have three characteristic domains. They possess a central catalytic region that is a serine-threonine kinase and flanking N and C termini that define their interactions and cellular localization. The GRK N terminus facilitates specificity toward G protein-coupled receptors (GPCRs) (2)(3)(4). Differences between the GRK C termini provide the means to divide them into three subclasses. All GRK2/3-like kinases have a C-terminal pleckstrin homology domain that facilitates binding of a G protein ␤␥-subunit upon receptor activation (5). The interaction of cytosolic GRK2 and -3 with ␤␥-subunits also induces their translocation to the plasma membrane, where the activated receptors are located (6). Substantial amounts of GRK4 to -6, on the other hand, are basally associated with the plasma membrane. An amphipathic helix in the C terminus expresses a number of hydrophobic amino acids that bind hydrophobic lipid elements (7), and positively charged residues in this domain interact with phospholipid headgroups of the cell membrane (8). In addition, GRK4 and -6, but not GRK5, are palmitoylated (9). Thus, GRK4 to -6 do not require prior recruitment by a G protein to phosphorylate the signaling receptor (7,8). Further differences between GRK subclasses include GRK2 localization to mitochondria (10,11) and GRK5 translocation to nuclei and cilia (12)(13)(14)(15). Thus, GRKs of different subfamilies can be distinguished by their motifs and additionally by their distinct subcellular localizations.
GRKs have been predominantly recognized for their ability to desensitize GPCRs. This is achieved by phosphorylation of GPCR cytoplasmic domains and C termini that promote inter-actions between the phosphorylated GPCRs and ␤-arrestins. This interaction sterically terminates G protein-mediated signaling. The GPCR⅐␤-arrestin complex often serves as the trigger for receptor endocytosis (16,17). Interestingly, once complexed with the phosphorylated GPCR, ␤-arrestin acquires the ability to scaffold additional intracellular signaling molecules, resulting in arrestin-dependent signaling (for a review, see Ref. 18). Importantly, at least in heterologous cell systems, there appears to be a preference for GRK5 or GRK6 in arrestin-mediated signaling, as is the case for AT1A receptor activation of ERK (19). Similarly, mass spectrometry analysis demonstrated that different GRK subclasses will phosphorylate different sets of amino acids in the C-tail of the ␤-adrenergic receptor (20), and when one GRK subtype has "barcoded" the receptor, a GRK belonging to another subclass is prevented from phosphorylating it (20,21). Thus, agonist activation of a receptor can trigger different, even opposite, physiological actions if different GRKs participate.
Physiologically, GRKs are important regulators of heart function, particularly GRK2 and GRK5 (22). However, depending on the signaling pathway involved, GRK2 and -5 can have either beneficial or harmful effects on the heart. The effect is also dependent on the expression level of the respective GRK, genetic variation, the subcellular localization of the GRK, and the condition of the heart (14,(23)(24)(25). Except for the lethality of conventional GRK2 knockout observed in embryonic mice (26), most of what we know today about the differential influence of GRK2 and -5 on the heart is based on experiments in adult rodents and rabbits. Here, we demonstrate that although losses of the zebrafish homologs of mammalian GRK2 and GRK5 exert opposite effects on cardiac performance during development, they have similar negative long term consequences on cardiac health that may be caused by different modes of signal transduction.

EXPERIMENTAL PROCEDURES
Zebrafish Strains and Husbandry-Zebrafish were kept under standard conditions in a 14-h light and 10-h dark cycle. Wild-type fish of EK and AB strains were used. All procedures and experiments were carried out according to animal protocols approved by local authorities at Duke University and Ulm University.
Contractility Measurements-Zebrafish were mounted in 2.5% methylcellulose for video imaging of beating hearts. Fractional shortening of the ventricular chamber was measured with the help of zebraFS software (32).
Zebrafish Heart Sections-Zebrafish embryos were fixed in 4% paraformaldehyde and gradually dehydrated before embedding in JB-4 (Polysciences). Then 4-m sections were cut using a Leica RM2255 microtome (Leica, Wetzlar, Germany), dried, and stained with hematoxylin and eosin.
Imaging and Statistical Analysis-Zebrafish were either imaged with a Leica MZ16 dissection microscope equipped with a 4-megapixel monocamera and Openlab software (PerkinElmer Life Sciences) or using a Leica M125 equipped with a Planapo 1.0 objective and a Leica IC80HD camera. The area of the heart fields was measured using Openlab (Perkin-Elmer Life Sciences). Confocal images were taken on a Leica TCS SP5II confocal microscope.
All graphs and statistics were prepared with the help of Prism4 (GraphPad Software Inc.). If not otherwise noted, data are presented as average Ϯ S.E. and analyzed using one-way analysis of variance with Bonferroni's or Tukey's multiple comparison tests.

GRK2 and GRK5 Differentially Regulate Heart Function and
Size-We used zebrafish to investigate the potential divergent physiological functions of GRK2 and GRK5 during heart development. Zebrafish embryos have the advantage of being available in large numbers. They are translucent, develop very rapidly outside the uterus, and can be easily manipulated. We previously cloned the homologs of different GRKs and developed and validated translation-blocking antisense MOs to gen- erate loss-of-function (LOF) embryos of GRK homologs (27,28). To test our hypothesis that GRKs of different subfamilies would share functions but also differ in their effect on cardiac physiology, we chose to analyze Grk2/3 and Grk5. Grk2/3 is the combined homolog of mammalian GRK2 and -3 (28) and thus nicely represents the subfamily of ␤-adrenergic receptor kinases. Grk5 is one of two GRK5 proteins in zebrafish, which we have characterized previously for its role in canonical Wnt signaling (27).
Heart rate is a simple but powerful way to measure the impact of manipulation on the heart. Knockdown of either Grk2/3 or Grk5 resulted in opposing changes in heart rate at 27 h postfertilization (hpf) of embryos. Although the loss of Grk2/3 resulted in a moderate but significant increase in heart rate compared with non-injected controls or those injected with a 5-base mismatch MO (5mis Grk2/3), Grk5 morphants displayed considerable bradycardia (Fig. 1A). One day later, however, after the zebrafish heart had compartmentalized into two chambers, we detected similar reductions in heart rate for both Grk2/3 and Grk5 morphants. Importantly, both phenotypes could be rescued by reconstitution with the respective Grk through co-injection of the MO with capped RNA (Fig.  1B).
Consistent with the reduction in cardiac function at 48 hpf, LOF embryos for both kinases developed pericardial edema that was not apparent under any of the control conditions tested (Fig. 1C). Morphological examination of the heart (Fig.  1D) revealed that knockdown of Grk2/3 produced a dilated atrium with a very small ventricle. In contrast, Grk5 LOF hearts appeared smaller overall than control hearts. To confirm that these phenotypes were not due to aberrant chamber specification, we performed WMISH for atrial and ventricular myosin heavy chain (amhc and vmhc). The staining pattern from each study showed that the heart was appropriately specified into two chambers. In addition, the data confirmed the misshaping of the hearts upon injection of either knockdown MO (Fig. 1, E and F).
The GRK Phenotypes Can Be Verified by a Second Set of MOs-In order to be sure that the observed, partially opposing phenotypes were specific, we tested whether we could reproduce our results with a second set of MOs. We injected socalled splice-blocking MOs (splMOs), which anneal to an exon-intron boundary and thereby sterically interfere with pre-mRNA splicing. SplMOs cause exon skipping and, when designed appropriately, produce premature stop codons that result in non-sense RNA. The efficiency of a certain splMO to disable splicing can be monitored through simple RT-PCR using primers, which flank the targeted exon. The corresponding PCR product of injected embryos will be shorter than in non-injected control embryos. In fact, the Grk2/3 splMO did produce the expected smaller PCR band ( Fig. 2A, left) and resulted in phenotypic embryos with U-shaped somites (data not shown). However, when we tested the splice-blocking MO for Grk5, we observed a larger rather than smaller band for the injected embryos ( Fig. 2A, right). Sequencing of the PCR product revealed that the Grk5 splMO caused retention of an intron, which resulted in a premature stop codon. We thus concluded that the splMOs knock down Grk2/3 and Grk5 expression. Not surprisingly, heart rates were altered in a similar fashion as for the translation blocking MOs (Fig. 2B). In addition, injection of both splMOs generated embryos with impaired cardiac function, resulting in pericardial edema (Fig. 2C) that was similar to what we observed for the translation-blocking MOs (Fig. 1C).
Loss of Grks Does Not Affect Contractility-The contractile function of the heart greatly relies on ␤-adrenergic stimuli, which in turn are regulated by Grks, mainly GRK2. In animal models, inhibition of GRK2 results in increased ventricular contractility in response to ␤-adrenergic activation (for a review, see Ref. 34). In our zebrafish models of Grk LOF, however, we did not see differences in fractional shortening of the ventricle compared with control embryos (Fig. 3).  No Rescue by Opposing Grks-The observed phenotypes could be explained by two hypotheses. On one hand, they could simply be due to an overall loss of kinase activity, or on the other, they could be due to loss of specific types of kinase activity. In the first case, knockdown of one Grk would probably be rescued by co-injection of capped RNA encoding the other Grk. However, if Grk2/3 and Grk5 mediate opposing effects, such "rescue" would probably be ineffective. Indeed, grk5 RNA neither prevented the Grk2/3 LOF-mediated increase in heart rate at early stages (Fig. 4A) nor rescued the reduced heart rate at 2 days postfertilization (Fig. 4B). Similarly, co-application of grk2/3 RNA did not normalize the observed negative chronotropic effect of Grk5 KD (Fig. 4, A and B). Furthermore, crossrescue experiments produced no change with regard to the development of pericardial edema (Fig. 4C).
The Observed Grk Phenotype Depends on Kinase Activity-GRKs were originally identified for their ability to terminate GPCR signaling through phosphorylation of the activated receptor (35). Research in the last decade has shown that GRKs may just as well exert their physiological functions as adaptor molecules independent of their kinase activity (36). In order to test whether the effect of both GRKs on the developing heart would rely on a functional kinase domain, we repeated our rescue experiments with Grk variants carrying a point mutation in their kinase domain. Grk2/3 K220R, which resembles the kinase-dead variant of mammalian GRK2 (37, 38), could not rescue the Grk2/3 phenotype; nor would the Grk5 heart phenotype be ameliorated by Grk5 K209M, which is mutated at the lysine corresponding to Lys-215 in bovine GRK5 (Fig. 5, A-C). To be sure that the latter mutation would result in a kinase-dead or dominant negative form of Grk5, we analyzed its effect on NFB p65 transcription, which was shown to be induced by GRK5 in mice (39). Exchange of lysine 209 for methionine does in fact partly inhibit the p65 expression (Fig. 5D). We thus concluded that Grk2/3 and Grk5 do have opposing roles during heart development, which depend at least partially on their kinase activity. However, so far we have not ruled out the possibility that the observed cardiac maldevelopment could be due to Grks functioning in the heart directly.
Angiogenesis Is Not Affected by LOF of Grk2/3 or Grk5-Zebrafish embryos with a silent heart can undergo normal angiogenesis (40). On the other hand, if blood vessel development is impaired, heart morphology will be altered (41). Furthermore, a recent study associated GRK2 with angiogenesis (42). We therefore asked whether the observed changes in cardiac morphology in the Grk morphants would be caused by a defect in vascular development. WMISH for cdh5, which is expressed in all endothelial cells, indicated that angiogenesis up to this developmental stage occurred regularly (Fig. 6A). This observation could be corroborated by separate analysis of arteries (ephrinB2a) and veins (flt4) (Fig. 6, B and C). The only effect  SEPTEMBER 19, 2014 • VOLUME 289 • NUMBER 38

GRK2 and -5 in Zebrafish Heart Development
we observed was a slight enlargement of the caudal vein plexus in Grk2/3 morphants. Apart from that, vascular sprouting appeared unaffected in both Grk morphants, implying that the underlying cause for the observed phenotypes may lie within the heart itself.

Grk2/3 and Grk5 Are Expressed in the Developing Zebrafish
Heart-In mice, Sefton et al. (43) reported that GRK2 could be detected in the heart itself from embryonic day 11.5 onward. In addition, at embryonic day 7.5 transcripts were found in the mesoderm, which gives rise to cardiac progenitor cells. Similarly, in zebrafish, grk2/3 is expressed rather ubiquitously at early somitgenesis stages (28), as is grk5 (27). At later stages, when the heart has formed, the expression of either Grk has not been analyzed. Therefore, we performed WMISH for both genes. We detected transcripts of Grk2/3 (Fig. 7, A and B) and Grk5 (Fig. 7, C and D) throughout the whole heart.
Grk2/3 Regulates Proliferation in the Developing Heart-In sagittal sections, we verified the misshaping of cardiac chambers upon either Grk KD. Although NI hearts showed two nicely developed chambers, we found distinctly smaller ventricles in Grk2/3 morphants. Grk5 KD hearts appeared to be smaller overall than control hearts (Fig. 8A). However, the smaller morphology was not caused by an increase in apoptotic events because we did not find an increase in cleaved caspase-3-positive cells under either condition (Fig. 8, B and D). In contrast to this observation, we detected a small but significant  reduction in the proliferation of cells in the hearts of Grk2/3 LOF embryos (Fig. 8, C and E). However, the remaining proliferating cells were distributed throughout the entire heart. It is thus not likely that the small ventricle morphology in Grk2/3 morphants would originate from an impairment of proliferation in the two-chambered heart tube.
GRK2 and GRK5 Regulate Heart Size at the Level of Its Progenitor Cells-Because we did not find a profound vascularization defect, we hypothesized that the altered heart physiology in Grk morphants was not caused by insufficient circulation or a reduction of shear stress. Moreover, our findings that Grk KD produced large alterations in neither cardiac cell death nor proliferation suggested that both Grks may affect the late stage morphology at much earlier times (i.e. cardiac specification or differentiation). The vertebrate heart develops from two heart fields of cardiac progenitor cells lying bilaterally from the midline (44). Failure in specification of those progenitor pools can be a cause for hypo-or hyperplastic hearts at later stages. One of the earliest transcribed genes essential for cardiac specification is the homeobox transcription factor nkx2.5 (45,46). We observed weaker expression of nkx2.5, when Grk5 was knocked down in Grk LOF embryos, (Fig. 9, A and D). This effect could also be seen throughout differentiation of cardiac precursor cells. Grk5 LOF embryos exhibited a strong reduction of cmlc2positive cardiac precursor cells, whereas we did not see any significant change in Grk2/3 MO-injected embryos (Fig. 9, B  and E). Surprisingly, when we analyzed the progenitor pool of ventricular cardiomyocytes, we found a decrease in the area positive for vmhc not only in Grk5 morphants but also in Grk2/3 LOF embryos (Fig. 9, C and F). Although unexpected, the finding was consistent with our observation that the ventricle of the Grk2/3 knockdown fish looked smaller than those of controls.

Inhibition of ␤-Adrenergic Receptors Cannot Fully Rescue Grk
KD Zebrafish-The most important regulator of cardiac function in the adult is the ␤ 1 -adrenergic receptor (␤1AR). This in turn underlies a tight regulation by GRKs, most importantly GRK2 (47). We thus wondered whether at least some of the Grk KD phenotype was due to increased ␤1AR activity. To test this, we treated zebrafish embryos from the tailbud stage onward with either propranolol or, as a control, the adrenergic agonist isoproterenol. In earlier studies, it was shown that propranolol has a mild effect on the heart rate starting from 44 hpf onward (48,49). Consistent with this, MO-mediated KD of ␤1AR in zebrafish resulted in lower heart rates at 4 days postfertilization (50). Although we saw the expected, moderate reduction in heart rate in non-injected embryos using propranolol, we did not see an improvement in either Grk KD. Isoproterenol treatment, on the other hand, elevated the heart rate slightly in KD embryos, although not significantly in the case of the Grk2/3 KD (Fig. 10A). Neither blockade nor activation of ␤1ARs prevented the development of pericardial edema (Fig. 10, B and C). Thus, Grk2/3 and Grk5 do not appear to govern early heart development through ␤1ARs.

Activation of Hedgehog Signaling Rescues Grk2/3 but Not
Grk5 LOF Embryos-We have shown previously that GRK2 augments Hedgehog (Hh) signaling (51) and that loss of either GRK2 in mice or Grk2/3 in zebrafish resembles a general Hh LOF phenotype (28,52). We thus wondered whether GRK5 would share this property and tested it in zebrafish by analyzing the expression of the Hh target genes nkx2.2 and ptc1. We found that Grk5 does not alter the expression of these genes and thus most probably has no effect on Hh signaling (Fig. 11, A  and B).
In our former studies, we showed that GRK2 acts downstream of Hh at the level of Smoothened, which transduces the Hh signal, but upstream of PKA. PKA phosphorylates Gli transcription factors. This, in turn, leads to the degradation of Gli and thus reduced activation of Hh target genes (53). Therefore, we tested whether expression of a dominant negative form of PKA (dnPKA) could override the Grk2/3 heart phenotype by co-injecting it. Before the actual rescue experiments, we injected increasing amounts of dnPKA RNA to find a dose that, by itself, had no effects or only small effects on the developing embryo (Fig. 11C). This low amount of dnPKA was sufficient to   . Activation of Hedgehog signaling rescues the pericardiac edema in Grk2/3 morphants. A, WMISH for nkx2.2 upon Grk5 knockdown. B, WMISH for ptc1, another Hh target gene after loss of Grk5. C, fertilized eggs were injected with increasing doses of capped RNA encoding for dnPKA. The lowest dose, which caused only very mild developmental abnormalities, was chosen for the rescue experiments because higher amounts resulted in embryos with smaller heads, reduced eye structures, and pericardiac edema. D, live images of zebrafish embryos, which were either uninjected, controlinjected with five mismatch MOs, or injected with MOs against Grk2/3 and Grk5. Rescues were performed using capped RNA encoding for dnPKA. Arrow, pericardium, which is dilated in Grk2/3 morphants and can be rescued by downstream activation of Hedgehog signaling. Grk5 morphants were unresponsive to dnPKA. E, bar graph summarizes 4 -6 independent experiments and displays data as mean Ϯ S.E. ***, p Ͻ 0.001 (5mis Grk2/3 versus Grk2/3 MO); **, p Ͻ 0.01 (Grk2/3 MO versus Grk2/3 MO ϩ dnPKA); *, p Ͻ 0.05 (5mis Grk5 versus Grk5 MO); ns, not significant (p Ͻ 0.05) (Grk5 MO versus Grk5 MO ϩ dnPKA). n ϭ 59 -186 embryos in total. Scale bars, 100 m. generate a partial rescue of the pericardial edema phenotype in Grk2/3 LOF fish. As expected from the analysis of Hh target genes, dnPKA did not improve the Grk5 phenotype (Fig. 11, C  and D). In summary, we found that Grk2/3 and Grk5 are important regulators of vertebrate heart development, and in that capacity, they have overlapping as well as opposite functions.

DISCUSSION
In this paper, we present data obtained in zebrafish that suggest opposing functions of Grk2/3 and Grk5 during heart development. Loss of Grk2/3 generates zebrafish embryos that initially show a mild enhancement of cardiac performance, whereas depletion of Grk5 greatly reduced it. At later stages, however, loss of either kinase provokes cardiac dysfunction. Importantly, those phenotypes could be reversed neither by ectopic expression of the opposing Grk nor by kinase-dead versions of them. Our data also suggest that these are heart-autonomous effects of Grk2/3 and Grk5 because vascular sprouting, a prerequisite for proper vascularization and thus circulation and chamber ballooning, appeared to be normal in our model. Furthermore, we showed that both Grks influence the heart very early at the level of its progenitor cells. Therefore, we propose that ultimately the proper function of either Grk is required to develop a normally shaped heart.
Grk2/3 and Grk5 are the close homologs of mammalian GRK2 and -5, which are highly abundant in the adult heart. Here they modulate greatly the signaling of ␤-adrenergic receptors as well as angiotensin receptors (23,24,54,55) and, through those, the performance of the heart. In line with that, ␤1AR stimulation slightly raised the heart rate in Grk5 LOF embryos. A growing number of studies suggest further that GRK2 and -5 can influence the heart through other signaling pathways. GRK2, for instance, can facilitate apoptotic events by translocation to mitochondria (10). GRK5, on the other hand, was shown to initiate hypertrophic responses after pressure overload through the activation of Mef2 upon phosphorylation of a histone deacetylase (14,56). Additionally, GRK5 mediates hypertrophy by the activation of the NFB signaling pathway (25). Thus, based on these data obtained in adult rodent models and because of the observation that GRK2 and -5 are up-regulated in failing hearts (57,58), therapeutic targeting of either GRK was concluded to be instrumental for prevention of cardiac hypertrophy and eventually heart failure. In contrast, we found in an embryonic model that loss of either GRK subtype would be detrimental for the developing heart. This phenomenon may be explained by our finding that both GRKs partially influence different subsets of myocardial progenitor cells before a heart tube has formed. Our results are particularly interesting because Khan et al. (59) have convincingly shown that GRK2 negatively regulates survival and proliferation of cardiac progenitor cells isolated from adult hearts. Based on our data in embryos, however, we conclude that GRK2 functions differently, depending on the kind of cardiac progenitor cell and additionally whether this cardiac progenitor cell has an embryonic or adult origin. Apart from that, we believe that embryonic GRK2 in particular mediates its effect on myocardial progenitor number, most likely by facilitating Hh signaling and only in part through ␤-adrenergic receptors, as in the study by Khan et al. (59), which analyzed the effect on adult progenitors. Hh positively regulates heart progenitor proliferation (60), whereas Grk2/3 in turn augments Hh signaling (28). Thus, we propose that the observed Grk2/3 heart phenotype is at least partially due to a loss of Hh pathway activity, although we cannot exclude the possibility that it is mediated through additional Hh-independent signaling pathways too. The Hh LOF hypothesis was further strengthened by the reduction in pericardial edema of Grk2/3 morphants upon downstream activation of Hh signaling. Because LOF of Grk5 did not result in a reduction of Hh target genes and also failed to respond to pathway activation, it remains to be elucidated in a follow-up study by which molecular mechanism GRK5 would determine the size of the heart field and the later heart morphology.
A recurring theme in the course of our study has been the fact that Grk2/3 and Grk5 appear to function similarly in some aspects but in an opposing way in others. This is an observation that has been made in cells before and may be partially due to differential dependence of G protein and arrestin biased receptor signaling on certain GRKs. So is classical G protein-mediated signaling of GPCRs reportedly more regulated by GRKs of the ␤-adrenergic receptor kinase type (61,62)? GRKs of the GRK4 to -6 family have been shown to provoke non-canonical GPCR signal transduction through arrestins, as demonstrated for a number of GPCRs, including the angiotensin receptor 1a, the ␤-adrenergic receptor 1, the vasopressin receptor 2, the histamine 1 receptor, and the follicle-stimulating hormone receptor (62)(63)(64)(65)(66). Moreover, GRKs of the two subfamilies are able to compete with each other for the phosphorylation of an activated receptor (20). This functional competition of GRKs was first shown by the Lefkowitz group (62,64). However, in certain cases, opposing effects may also be explained by potentially differing availability of different GRKs. The different subcellular localization of both GRK types may further expedite the highly variable impact of a GRK on a particular signaling cascade or physiological function. Finally, the different functions may have been embedded in GRK2 and GRK5 already in the earliest ancestors of metazoan life. Thus, despite evolution, GRK2 and GRK5 remain highly related kinases of different makeup and performance.