G protein-coupled receptors (GPCRs) are important regulators of various cellular functions via activation of intracellular signaling events. Active GPCR signaling is shut down by GPCR kinases (GRKs) and subsequent β-arrestin-mediated mechanisms including phosphorylation, internalization, and either receptor degradation or resensitization. The seven-member GRK family varies in their structural composition, cellular localization, function, and mechanism of action (see sect. II). Here, we focus our attention on GRKs in particular canonical and novel roles of the GRKs found in the cardiovascular system (see sects. III and IV). Paramount to overall cardiac function is GPCR-mediated signaling provided by the adrenergic system. Overstimulation of the adrenergic system has been highly implicated in various etiologies of cardiovascular disease including hypertension and heart failure. GRKs acting downstream of heightened adrenergic signaling appear to be key players in cardiac homeostasis and disease progression, and herein we review the current data on GRKs related to cardiac disease and discuss their potential in the development of novel therapeutic strategies in cardiac diseases including heart failure.
The body maintains its “physiological homeostasis” by constantly sending signals to organs, which receive and translate circulating stimuli into appropriate cellular responses. The ability of cells to communicate with each other is paramount for the maintenance of this homeostasis. This communication is achieved through molecules, also known as agonists, which bind to receptors activating downstream signal transduction pathways.
One of the major agonist-receptor interactions is that of G protein-coupled receptors (GPCRs) (155). Although there is not a solid consensus on classification, this superfamily of receptors may be considered to encompass protein members from six different classes: class A (rhodopsin-like), class B (secretin-receptor like), class C (metabotropic-glutamate/pheromone), class D (fungal mating pheromone), class E (cAMP), and frizzled-smoothened (8, 38, 77, 78, 130). Although the exact size of this superfamily is unknown, human genome sequence analysis has predicted that nearly 800 human genes are GPCRs (25). Commonly known GPCRs include: β-adrenergic receptors (β-ARs), α-adrenergic receptors (α-ARs), rhodopsin, endothelin receptors, adenosine receptors, and others (77). These receptors are integral membrane proteins characterized by seven membrane-spanning domains, which upon agonist binding change their conformation triggering heterotrimeric G proteins to bind to it in the intracellular space and subsequently activating downstream signal transduction pathways (45, 205, 206, 214) (see Figure 1). Agonist binding to GPCRs promote conformational changes (206) which allow for the dissociation of heterotrimeric G proteins into α and βγ subunits, stimulating activation of downstream proteins such as protein kinase A, Src, MEK5, phosphatidylinositol 3-kinase (PI3K), GPCR kinase (GRK), etc. While activation of signaling pathways by these receptors is extremely important, so is deactivation. When the balance of activation and deactivation is impaired, imperative cellular mechanisms become faulty, and if not corrected in a timely manner, this signaling can become irreversibly detrimental to the body, leading to diseased states.
Deactivation of GPCR-elicited signaling is mainly accomplished by GRKs, and this is termed desensitization. Termination of signaling via GPCRs is done through a two-step mechanism: first, the GRK phosphorylates the active receptor (and only this form of the receptor), leading to conformational changes that promote binding of β-arrestin; and second, phosphorylated and β-arrestin bound GPCRs are no longer able to bind to and activate G proteins (132, 228, 274). Figure 1 depicts a schematic of GPCR activation, signaling, and desensitization. Briefly, agonist binding to the receptor causes nucleotide exchange, Gα-Gβγ dissociation, and downstream signaling through effector activation. The Gα-Gβγ dissociation mobilizes GRKs to phosphorylate the activated receptor, a process which may vary depending on the GRK isoforms (see sect. IIC). Serine (Ser) and threonine (Thr) phosphorylation by GRKs at the COOH terminus of the receptor promotes β-arrestin mobilization and binding to the receptor. This may further lead to clathrin-induced endocytosis and reactivation or degradation of the receptor. Although receptor endocytosis reactivation/recycling is an extremely important and complex mechanism involving many other proteins, it is beyond the scope of this review and better detailed by Moore et al. (163). Lastly, activation of GTPases causes G protein trimer reformation and repriming of the receptor for another round of stimulation. In this review, we focus on the multifaceted functions of GRKs including their evolving non-GPCR or noncanonical activity that appears to be critical for cardiovascular homeostasis and disease.
The first description of phosphorylation-mediated deactivation of GPCRs was provided in the 1970s where the photoreceptor rhodopsin, a prototypical GPCR, was rapidly deactivated by phosphorylation by “opsin or rhodopsin kinase” (also known as GRK1) (28, 133, 270). Later, the same deactivation mechanism was proposed in many other GPCRs (reviewed in Ref. 35) including β2-ARs (21). Several studies have demonstrated that agonist-specific desensitization of the β2-AR was associated with receptor phosphorylation (19, 229, 243). GRK2 was then identified (initially named the β-AR kinase, β-ARK1) and shown to be able to phosphorylate β-ARs as well as rhodopsin in a light-dependent manner (19, 21). It was only when GRK2 was identified and characterized to possess properties similar to GRK1 that the hypothesis was put forth of a possible family of new proteins: the GRK family (16, 19).
II. G PROTEIN-COUPLED RECEPTOR KINASES
A. GRK Family Members
The GRK family is composed of seven members. This family of kinases is quite small considering that they are responsible for deactivating potentially more than 800 GPCR gene products as only a small percent of these receptors are not regulated by phosphorylation (78). Although identification of phosphorylation-induced desensitization of GPCRs by rhodopsin kinase (GRK1) (270) and GRK2 (or β-ARK1) (18, 21) opened the doors to this novel deactivation mechanism, it was only after cloning of GRK2 in 1989 that the idea of a new eukaryotic serine/threonine (Ser/Thr) protein kinase family fully emerged. Promptly thereafter, new members of this family were identified and included: GRK3 (20), GRK4 (6), GRK5 (134, 192), GRK6 (17), and GRK7 (101, 269). Based on sequence similarity and gene structure analysis, vertebrate GRKs are further divided into three subcategories: visual or rhodopsin-kinases subfamily (comprising of GRK1 and GRK7), β-AR kinases subfamily (comprising of GRK2 and GRK3), and the GRK4 subfamily (comprising of GRK4, GRK5, and GRK6) (see Figure 2).
The GRK family of proteins possesses shared structural and functional hallmarks. These proteins are mainly composed of three regions: a short 25-residue proximal NH2-terminal region unique to the GRK family of kinases, a regulator of G protein signaling (RGS) homology domain (RH), which is interrupted by a Ser/Thr kinase domain (KD). This ∼500–520 amino acid residue KD belongs to the AGC kinase family and is shared by all GRKs (reviewed in Refs. 95, 112). The entire NH2 terminus is highly conserved among GRKs though the NH2-terminal region is unique to the GRK family of kinases and implicated in GPCR binding (reviewed in Refs. 95, 112). The COOH terminus is the most diverse region among GRK subfamilies. GRK1 and GRK7 have short COOH-terminal prenylation sequences (101, 113). GRK2 and GRK3 contain pleckstrin homology (PH) domains implicated in binding anionic phospholipids and Gβγ subunits (34, 60, 128, 187). GRK2 has been recently implicated to localize to mitochondria (43, 83). GRK2 does not have a classical mitochondrial localization sequence, so it is not clear what is the required minimal sequence necessary for this novel localization. GRK4 and GRK6 have palmitoylation sites (194, 241) as well as lipid-binding capabilities (118). GRK5 membrane targeting relies on a phospholipid binding domain (residues 552–562) (196). Alternative splicing has been identified for GRK4 (4 splice variants) (194, 216) and GRK6 (4 splice variants) (75, 161, 193, 257); however, not all splice variants appear to be expressed across species. More recently, a nuclear localization sequence (NLS) has been identified within GRK5 (amino acids 388–395) (120), which happens to be a common feature in all GRK4 subfamily members (119). In addition, a nuclear export signal (NES, amino acids 259–265) has also been identified (119). Figure 2 depicts the tri-domain schematic of all members of the GRK family that include the above-described features. Catalytic activity is dependent on specific lysine residues and shown for each isoform.
B. GRK Tissue Localization
Most GRKs are ubiquitously expressed; however, there is selective tissue distribution. GRK1 and GRK7, members of the rhodopsin-kinase subfamily, are generally restricted to the retina where they primarily target rhodopsin and cone opsin, respectively. Within the retina, these two isoforms are particularly restricted to rod and cone photoreceptors, although both are also present in pinealocytes (41, 231, 283, 284). GRK4 is highly expressed in the testes (194), although it has also been shown to be expressed in proximal tubule cells in kidneys (73, 258), in the brain (217), and in the uterus myometrium (29). In particular, two GRK4 variants (GRK4α and GRK4γ) have been reported to signal through D1 and D3 dopamine receptors within the kidney (73, 258). Essentially all the remaining isoforms are ubiquitously expressed with varying degrees of expression depending on type of tissue. GRK2 primarily targets several GPCRs including β-ARs and angiotensin II type 1 receptors, and it is expressed in most tissues. GRK3 targets olfactory receptors, thrombin receptors, kappa-opioid CRF1 (57, 219), and α1- and α2–ARs (12, 246). It is found in most tissues, but most prominently in the olfactory epithelium. GRK5 can phosphorylate β-ARs, adenosine receptors, and M2 muscarinic receptors, and it is expressed in most tissues (14).
Table 1 shows tissue distribution of various GRK isoforms (reviewed in Ref. 112). Pertinent to this review is the expression of cardiac GRKs. In the human heart there is evidence for expression of GRK2, GRK3, GRK5, and GRK6 (67). The most prominent cardiac GRK isoforms are 2 and 5, although 3 and 6 have also been reported at low levels (1, 67, 162). Moreover, it is important to note that a particular GRK does not have to be the most prominently expressed GRK in a tissue to be crucial for a specific GPCR signaling regulation as its cellular localization and activity are key measures determining GPCR-GRK interactions. More recently, GRK3 and GRK4 have been described to be present in the thyroid nodules and to be upregulated in hyperfunctioning thyroids, which may also indirectly impact cardiac function (262).
C. Canonical Subcellular Targeting of GRK Isoforms
Agonist-mediated activation of GPCRs is obviously critical for cellular responses; however, sustained GPCR stimulation is deleterious in part because of receptor desensitization and eventually downregulation. If novo or de novo activation of receptors does not occur, this downregulation of receptors may not only lead to loss of responsiveness to agonists but also cellular response disarray. It is the primary job of GRKs to reverse the harmful effects of chronic GPCR stimulation. To accomplish this goal, GRKs are canonically thought to be present mainly in the cellular cytoplasm or close to the plasma membrane. Certain GRK isoforms have specific sequence trademarks that allow them to preferentially target specific subcellular compartments. In general, membrane targeting of all GRKs towards activated GPCRs involves residues and domains within the COOH terminus of these kinases (see Table 2).
GRK1 and GRK7 are localized directly to the plasma membrane after GPCR activation as both isoforms possess a short COOH-terminal prenylation sequence (101, 113), and the attached lipid molecules act as lipid anchors facilitating attachment to cell membranes. This cellular localization is advantageous for visual GRKs as they perform their function of “probing” for active rhodopsins or cone opsins. As such, visual GRKs perform a two-dimensional search instead of a three-dimensional diffusion, which is essential for their ultra-rapid signal termination. This means that because GRK1 and GRK7 attachment to the membrane is facilitated by lipid anchoring, when GPCRs become active, these specific kinases have the advantage of searching for active receptors in the x- and y-position only since they are already closely associated with the membrane. On the contrary, all other GRKs do not have this facilitated mode of attachment, having to perform an x, y, and z search for activated receptors.
The β-ARK family of kinases (GRK2 and GRK3) are membrane targeted from the cytosol through binding to the membrane-anchored Gβγ subunits of G proteins following their dissociation after GPCR activation (34, 187). Indeed, GRK2 and GRK3 were originally found to be localized in the cytoplasm and then recruited to the membrane as a response to agonist-occupied receptors (19, 21, 242). This binding occurs within the COOH terminus and includes the PH domain (34, 189). Because Gβγ subunits are generated by activated receptors, free Gβγ recruits these GRK isoforms to the locale of active GPCRs, thereby greatly enhancing GPCR phosphorylation only when needed ensuring the fidelity of the desensitization process of only agonist-occupied receptors. Nevertheless, it is important to note that the fidelity of the desensitization process is aided by the preference of GRKs for the agonist-occupied conformation of the receptor as a substrate, as agonist occupancy can increase GPCR substrate affinity manyfold. Thus, like GRK1 and GRK7, the β-ARK kinases attach to the membrane via a prenyl group; however, these kinases “pirate” the lipid anchors from Gβγ and do not have them directly attached to their C-tail.
The Gβγ binding domain within the COOH-terminal PH domain of GRK2 (Figure 2) was originally determined biochemically (128). Following the structural information of GRK2 at the atomic level, the key residues for Gβγ binding were confirmed (143). Importantly, peptides from this region can block the membrane localization of GRK2 and prevent β-AR and GPCR desensitization (128). The shortest effective GRK2-Gβγ blocking peptide is a 28-mer peptide. However, most studies over the last 20 years have used the 194-amino acid residue peptide of the entire COOH terminus of GRK2, β-ARKct (128, 129), which has been shown to be effective both in vitro and in vivo at inhibiting GRK2 activity (discussed in detail below).
Lastly, subcellular targeting of the GRK4 subfamily involves protein hallmarks and modifications also present in the COOH termini. GRK4 and GRK6 carry both palmitoylation sites as well as lipid-binding positively charged elements which promote a biased membrane localization (147, 194, 240, 241). Though lacking the palmitoylation site, the COOH terminus of GRK5 is capable of binding phospholipids, thus distinctly promoting plasma membrane targeting (251). These features likely enhance lipid bilayer affinity and kinase activity. Interestingly, GRK6A splice variant contains elements that both promote and inhibit membrane localization. While it contains the features mentioned above, it also contains acidic residues that inhibit membrane localization (118, 257). A nonpalmitoylated form of GRK6A is detected in the cytoplasm and nucleus, although nuclear function still remains unclear (118). Surprisingly, GRK6B and GRK6C, which do not contain the palmitoylation sites, are still capable of localizing to the plasma membrane in cultured cells (257). Interestingly, membrane localization is rescued when the string of acidic residues located in the COOH terminus of GRK6A is mutated to neutral or basic residues in nonpalmitoylated GRK6A. This suggests that palmitoylation is not the dominant mechanism in which GRK4 subfamily of proteins achieve their preferential membrane localization. In addition, a fourth isoform for GRK6 (GRK6D) has been proposed to lack the catalytic domain and exclusively localize to the nucleus, although its function is yet to be determined (257).
In vitro, different GRK isoforms may phosphorylate different GPCRs, and differential subcellular targeting and tissue expression levels might contribute to explaining the functional specificity observed in vivo (see Table 1). For instance, GRK2 is capable of phosphorylating rhodopsin in a light-dependent manner (18), yet GRK1 is the isoform involved in rhodopsin deactivation in vivo as GRK2 does not perform this function in GRK1 knockout mice (39). GRK2 and GRK3 are both expressed in the brain and HEK cells. In HEK cells, both isoforms behave similarly, yet in the brain, GRK3 is more membrane-associated that GRK2 (4, 33). Although enormous effort has been expended to understand GRK receptor specificity, the mechanism for receptor preference still remains unclear. Receptor phosphorylation “bar coding” has been suggested as one of the ways specificity might be determined (166), yet this concept may not entirely account for GRK selectivity. In this study, phosphorylation patterns of the receptor by GRKs lead to diverse β-arrestin-related activities. Yet, it did not explain why a specific GRK was more prone to phosphorylate a receptor over another. GRK receptor specificity has been analyzed in hybrid transgenic mouse hearts. For example, GRK2, GRK3, and GRK5 overexpressing mice were bred with cardiac-specific transgenic mice overexpressing constitutively activated α1B-ARs. The results revealed that different GRKs had varying effects on this receptor. While GRK3 desensitized α1B-ARs very well, GRK2 had no effect, and the effects of GRK5 were mixed (68).
In another study using just GRK2 and GRK5 cardiac overexpressing transgenic mice, it was found that both of these cardiac GRKs desensitize β-ARs quite well, while GRK2 could also attenuate signaling through angiotensin II receptors, and GRK5 had no effect on these receptors (212). GRK5 and GRK2 also differently effected cardiac adenosine responses, respectively desensitizing or having no effect (212). Clearly, GRKs have receptor selectivity in vivo; however, the mechanism of receptor preference still remains unknown. Different expression levels of various GRK isoforms in a cell-specific manner may be one of the ways in which GRKs accomplish different tasks in various tissues, emphasizing the importance of spatial-temporal expression. Nevertheless, relative levels of expression in a particular tissue do not necessarily explain receptor-kinase specificity. Yet, how bar coding, cell-specific expression, or structural characteristics play a role in receptor specificity still requires further investigation.
D. Mechanism of Action of GRKs
The molecular basis in which GRKs first recognize active receptors still remains a matter of speculation. The most plausible hypothesis is termed “fly casting,” in which the disordered NH2 terminus of GRKs forms an initial low-affinity interaction with the active receptor (55, 226). In the case of GRK1 and GRK7 which have a prenylation sequence thought to involve lipid anchoring to membranes, this would mean that the search is in two dimensions, increasing its chance of finding active-receptors and fulfilling its ultra-fast response requirement. The other subfamilies are thought to involve a three-dimensional search, although proximity to the membranes would also permit for these kinases to be readily available to deactivate the receptor in a timely manner.
Crystal structures of specific GRKs have proven essential in understanding possible mechanisms of action and activation processes. These concepts have been well summarized and recently reviewed in References 95 and 104. The best-described function of GRKs is the ability to phosphorylate active GPCRs. With the exception of GRK4α, which has been shown to be constitutively active and capable of phosphorylating unstimulated GPCRs (160, 203), the activity of GRKs appear to depend on the functional state of the target. In fact, docking to the active-receptor seems to directly activate GRKs (42, 172). Phosphorylation of the active receptor can in itself decrease G protein coupling (274), but most importantly, it allows for β-arrestin to bind and block the cytoplasmic surface of receptors, thus preventing interaction with effector enzymes that are necessary to initiate downstream signaling (132).
A recent structure of GRK6 in a “closed conformation” revealed an ordered NH2 terminus engaged in extensive interactions with both the small lobe and the C-tail of the kinase domain (26). These interactions are physically impossible when GRK6 is in a more open conformation (144). More interesting is the fact that the contact points of these residues in the NH2 terminus and the kinase domain are highly conserved in all GRKs and that site-directed mutagenesis revealed them to be essential for phosphorylation of active GPCRs and soluble substrates (26, 103, 105, 174, 238). In particular, mutations in hydrophobic residues in the NH2-terminal helix of the kinase abolished receptor but not peptide phosphorylation, suggesting the necessity of these residues for GPCR-GRK interactions (26). Although it seems evident how GRKs achieve an active conformation, one can only infer the mechanism in which an active receptor would stabilize and potentiate GRK activity. Recently, a structure of β-arrestin complexed to a GPCR phosphopeptide provided particular evidence of a receptor interacting interface to β-arrestin (227). Structural understanding of GRK complexed to the receptor would greatly aid in unveiling precisely how this interaction occurs. Although no such structural evidence currently exists, it is known that after a signaling event, the whole process of desensitization starts by phosphorylation of the activated receptor by GRKs. In particular, GRKs phosphorylate at one or more Ser/Thr residues located on the intracellular domains of the receptor, particularly sites located in the COOH terminus and in the intracellular loop between TM5 and TM6 of the receptor (127).
Docking of GRKs to active receptors has been proposed to require the NH2 terminus of GRKs (26, 103, 173) as well as other regions outside the kinase domain (10, 64). In addition, in order for GRKs to phosphorylate active GPCRs efficiently, they require the presence of a cytoplasmic pocket (not present in the receptor's inactive state) as well as the presence of negatively charged lipids (46, 204, 236). The size and physical properties of this cytoplasmic pocket in GPCRs are expected to be highly conserved while the sequence and size of the cytoplasmic loops of the receptors may greatly differ. Thus the conservation of this complimentary cytoplasmic pocket within various GPCRs potentially allows GRKs to phosphorylate a broad array of GPCR substrates. The rationale behind this hypothesis lies in the fact that this pocket in the activated receptor would fit in conformation and charge to that of GRKs in their active conformation. Once docked, the kinase would adopt a closed conformation and lead to phosphorylation of substrates in close proximity (104).
The RH domain within the NH2 terminus (Figure 2) of GRKs has also been implicated in GPCR phosphorylation. Canonical RGS proteins are GTPase activating proteins (GAP), attenuating G protein-mediated signaling by promoting GTPase activity of GTP-Gα (reviewed in Ref. 102). GRK2 and GRK3 can bind to Gαq through its RH domain (36). However, RH domains of GRKs have low or no GAP activity, and their ability to halt signaling is primarily due to directly binding and sequestering activated Gαq (36) or direct hindrance of GPCRs (63, 64). In addition, GRK-RH domain-mediated signaling can also occur independently of the receptor-Gαq process. GRK2 is capable of scavenging Gαq/11 inhibiting insulin-induced glucose transport (254). Likewise, it can inhibit endothelin-1-mediated glucose transport by interfering with Gαq/11(242). Recently, GRK2 has been shown to specifically regulate insulin receptor signaling in the heart by phosphorylating IRS-1 and blocking glucose transporter membrane translocation (50), which demonstrates a physiological role for GRK2 acting towards a non-GPCR substrate.
E. Regulation of GRKs
Canonically, there is the necessity for GRKs to be regulated at the moment it phosphorylates activated receptors. Phosphorylated active receptors allow for the binding of β-arrestins, which triggers the formation of clathrin-coated pits and progression through endocytic pathways (see Figure 1). These internalized compartments may contain receptors, GRK2, β-arrestin, as well as other proteins characteristic of endocytic vesicles (reviewed in Ref. 163). Once internalized, these vesicles are sorted for degradation (desensitization) or recycling (resensitization). The presence of GRKs in these internalized active-receptor vesicles may physically prevent the kinase from performing its function on other activated GPCRs. Additionally, prior to vesicle formation and internalization, the binding of β-arrestin to the active-receptor-GRK complex leads to further recruitment of other proteins. One of these recruited proteins is Src, which has been reported to phosphorylate GRK2, diminishing its activity (177). Recently, it has been demonstrated that β-arrestin association to the phosphorylated receptor occurs in a biphasic mode. First, there is a “loose” hanging interaction between the C-tail of the receptor and the NH2-terminal domain of arrestin. Analogous to a fishing line, the receptor samples for potential interactions at high rate. Second, there is an insertion of the finger loop into the receptor core leading to longitudinal arrangement of arrestin, also known as fully engaged “tight” interaction (228). This initial “fishing” of the receptor is only possible after the receptor has been phosphorylated by GRKs. Subsequent signaling, also known as GPCR biased signaling, occurs after β-arrestin is recruited to the phosphorylated receptor, a topic which is further expanded in section IIF.
Although circulating agonists, such as catecholamines, can induce activation of GRKs mediated by GPCRs, multiple molecular mechanisms have been proposed to regulate GRKs in the cellular cytoplasm. While some regulatory mechanisms may be common to all GRKs, the fine tuning of kinase activity may be different for each GRK subtype. GRK1 has been shown to be capable of regulating its activity by autophosphorylation at Ser21, Ser488, and Thr489 (31, 173). In addition, while having no effect on GRK2, the calcium-binding protein recoverin is able to inhibit GRK1. As calcium levels change in response to changes in light, recoverin regulates GRK1 (40, 280). In HEK cells, PKA directly binds to and phosphorylates GRK2 at Ser685, enhancing Gβγ binding and promoting membrane translocation (54). PKC has been shown to modulate GRK5 and GRK2. PKC is able to phosphorylate GRK5 and inhibit its activity severalfold due to both decreased affinity and activity for the active receptor (195). PKC-mediated GRK5 phosphorylation most likely occurs at Ser572, and either Ser566 or Ser568, although PKC can also phosphorylate autoregulatory sites Ser484 and/or Thr485 (195). On the other hand, PKC phosphorylation enhances GRK2 activity two- to threefold, possibly by increasing the ability of GRK2 to bind membranes and by disinhibiting tonic calmodulin regulation (48, 131, 276). GRK5 has also been shown to be modulated by calmodulin through a calmodulin-binding domain on GRK5 (197).
GRK2 is a short-lived protein (half-life of ∼1 h in HEK, COS-7, Jurkat, and C6 glioma cells) that is polyubiquitinated and actively degraded by the proteasome proteolytic pathway with increasing degradation in response to activation of β2-ARs (177, 180). Interestingly, in HL60 cells, the half-life of GRK2 was estimated to be ∼20 h (148). With such a relative short half-life, transcriptional regulation might be an important factor for GRK regulation, although little is known about molecular mechanisms directly involved in modulating GRK mRNA levels. Initially, the GRK2 promoter was considered to be a general “housekeeping” gene (181). Studies of the GRK2 promoter region revealed several notable features, which included no evidence for TATA box, absence of CCAAT boxes in the proximal region upstream of the predicted start site, and the first 300 bases immediately upstream of the ATG are extremely GC rich (87%) (181). In addition, there are predicted secondary structures in the 5' region of the mRNA, which may serve as posttranscriptional regulators of expression (181). Nevertheless, evidence does exist on potential modulators of GRK2 mRNA content. In the heart, circulating catecholamines and β-AR activity correlate with GRK2 mRNA expression (111). Mitogenic T-cell stimulation by exposing mononuclear leukocytes to PHA led to increased GRK2 mRNA levels, while no change was observed in mRNA expression of GRK5 and GRK6 (59). More intriguing are the analysis of the human GRK2 gene promoter in aortic smooth muscle cells, where agents that led to physiological vasoconstriction or hypertrophy increased GRK2 promoter activity, and on the contrary, proinflammatory cytokines such as interleukin-1β, tumor necrosis factor-α, or interferon-γ decreased the activity of the GRK2 promoter (202).
As mentioned above, active GPCR-GRK complex recruits β-arrestins. β-Arrestins can also interact with the cytosolic tyrosine kinase, c-Src, recruiting it to the complex. c-Src has been shown to phosphorylate GRK2 on tyrosine residues, marking it for degradation (177). In addition, in vitro studies have shown that c-Src modulates α1B-AR activity through GRK2 and GRK3 (5) even though in vivo experiments show that GRK2 has no effect towards α1B-ARs (68). Moreover, GRK2 can also be phosphorylated by p38 (141) and ERK1 (188) mitogen-activated protein (MAP) kinases at Ser670. This phosphorylation results in diminished activity (less phosphorylation of the active receptor) (188) and increased degradation (71). The phosphorylation of Ser670 of GRK2 by MAP kinases has recently been shown to cause the localization of this GRK2 to mitochondria (43), which will be discussed in more detail below.
Specifically relevant to cardiovascular regulation, mechanical stretch has also been implicated in activating GRKs. In neonatal rat ventricular myocytes, mechanical stretch led to activation of GRK2 through angiotensin II type I receptor, Gαq, and PKC (153). Inflatable balloon stretch in explanted hearts activated GRK5 and GRK6 but not GRK2 (201). In addition, as detailed above, GRK2 and GRK3 contain within their COOH terminus a PH domain that is involved in Gβγ binding. Free Gβγ binds to these GRKs within this region at multiple sites (34) with high affinity and are required for GPCR phosphorylation (189). Moreover, a regulatory Gβγ-binding site has also been described at the NH2 terminus of GRK2 (70). Phospholipids have also been reported to modulate the activity of GRKs (34) through the PH domain of GRK2 and GRK3 or through amino acids 547–560 of GRK5 (34, 169, 185, 196). Another regulation at the PH domain level of GRK2 and GRK3 is the ability to bind to phosphatidylinositol 4,5-bisphosphate (PIP2) and other acidic phospholipids. PIP2 binds to the NH2 terminus of the PH domain and is able to increase by threefold GRK2 phosphorylation of membrane receptors (60, 61).
Recently, another mode of GRK2-specific regulation has been described as this kinase is capable of being S-nitrosylated, and the dynamic modification at Cys340 when nitric oxide (NO) bioavailability is high causes inhibition of GRK2 activity towards GPCRs (271). GRK2 has been found to interact with endothelial NO synthase (eNOS), and these two molecules reciprocally inhibit one another with eNOS-NO bioavailability altering the in vivo activity of GRK2 (106). Importantly, in studies using a GRK2 knock-in mouse where Cys340 was mutated to a Ser and incapable of being S-nitrosylated, GRK2 activity was higher. This led to increased myocyte cell death after ischemic injury, and these mice were resistant to NO-mediated cardioprotection (106). More recently, studies in heterozygous GRK2 knockout (KO) mice revealed that these mice are more resistant to vascular remodeling, mechanical alterations, and hypertension induced by angiotensin II, partially through the preservation of NO bioavailability since these mice have less GRK2 to inhibit eNOS (9). Therefore, the unique interaction between GRK2 and the eNOS-NO system is a powerful and dynamic mechanism for regulating this GRK isoform, especially in the cardiovascular system.
Of particular importance, GRKs have also been implicated in binding and sometimes phosphorylating other non-GPCR substrates. The first description of a non-GPCR substrate for GRKs was tubulin (37, 96, 186), and both GRK2 and GRK5 are capable of phosphorylating tubulin. Interestingly, phosphorylation of tubulin by GRK2 is enhanced in the presence of Gβγ and phospholipids (both of which promote GRK localization to the membrane) as well as in the presence of active GPCRs. It is known that active GPCRs can enhance GRK phosphorylation of peptides up to 100-fold (42, 172). Thus one may infer that any potential substrate in close proximity to GRKs might become phosphorylated. This led to the suggestion that phosphorylation of non-GPCR substrates might be related to opportunity and location, being at the right place at the right time. In support of this hypothesis is the “high gain” phosphorylation in the retina (22, 23), phosphorylation of IRS-1 by GRK2 in response to insulin (50) and GRK2 phosphorylation of p38 (183).
On the other hand, GRKs have been hypothesized to possess functions beyond its kinase capabilities. GRKs have been proposed to be modulators of cellular functions in a phosphorylation-independent manner due to their ability to interact with a variety of signaling and trafficking proteins, such as actin (79), actinin (80), tubulin (37), Gα (36), Gβγ (187), clathrin (225), PI3K (164), GIT (190), caveolin (36, 88), lipids, etc. GRKs have been placed in the center of regulatory processes involving cell migration (179), metabolism (50, 52, 254), and apoptosis (43). Further investigation is needed to address these GRK-mediated intermolecular interactions and how they link to various cellular processes. Overall, these noncanonical activities of GRKs may play key roles in physiological signaling including in the heart as detailed below.
F. GRKs Directing GPCR-Biased Signaling
GPCR signaling is rapidly attenuated by GRK phosphorylation of receptors and subsequent binding of the protein β-arrestin (191, 237). β-Arrestin binding to the activated receptor acts as a clathrin adaptor ultimately triggering the receptor for internalization (91). However, prior to receptor downregulation, activated GPCRs could be “functionally selective” or “biased” towards G protein or β-arrestin-mediated signaling. It has been shown that the β2-AR cytoplasmic end helices VI and VII adopt two major conformational states depending on its ligand. Receptor agonists primarily affect G protein specific active state of helix VI, whereas β-arrestin biased ligands predominantly affect helix VII (140).
In addition, the “pattern” of which a receptor is phosphorylated may dictate a “bar code” for β-arrestin function. For example, GRK2 phosphorylation of the β2-AR results in waning of signaling and/or receptor internalization, whereas GRK6-phosphorylation of β2-ARs induces recruitment of β-arrestin that leads to extracellular signal-regulated kinase 1/2 (ERK1/2) signaling (166). Distinct GRK2 and GRK6 phosphorylation sites have been identified on the β2-AR (166). Moreover, work in other receptors has also shown the requirement of GRK2 (and GRK3)-mediated receptor phosphorylation for GPCR internalization, and GRK5 and GRK6-mediated receptor phosphorylation for β-arrestin-mediated ERK1/2 signaling (121, 123, 126, 207, 222, 285). In support of this GRK-mediated GPCR-biased signaling are the studies using carvedilol, a β-AR antagonist that has been shown to be a weak β-arrestin-biased agonist that can activate ERK after an apparent action of GRK6 (277).
Interestingly, in myocytes, β-arrestin “biased” signaling has also been shown to involve AT1Rs, in particular stretch-mediated activation of AT1Rs signals through β-arrestin “biased” pathways (201). β-Arrestin has also been implicated in mediating transactivation of the epidermal growth factor receptor (EGFR) downstream of β1-AR signaling (167, 252). More intriguing is the fact that this β-arrestin-dependent transactivation of EGFR is cardioprotective, independent of G protein activation, and requires the actions GRK5 and GRK6 (167). Interestingly, this novel pathway shows a trend towards increase in activity when GRK2 is absent, which supports the overall hypotheses that biased, β-arrestin-dependent signaling is indeed directed by the actions of specific GRKs notion of GRK dependence of β-arrestin function (167). In fact, it appears that biased GPCR signaling is determined by which GRK phosphorylates a given receptor (166). Thus GRK phosphorylation of GPCRs is a significant cellular event as it cannot only desensitize the active receptor but also dictate downstream signaling, functionally selecting for G protein or enabling β-arrestin-mediated pathways.
G. Mouse Models of Altered GRK Levels
Undoubtedly mouse models have proven to be an essential tool in understanding in vivo GRK functions. Various mouse models have aided in understanding these multifaceted kinases and their regulation. The first GRK mouse model was presented in Science 1995 by Koch et al. (129). Here, cardiac-specific overexpression of GRK2 in vivo led to reduced functional coupling of β-ARs, dampened adenylyl cyclase activity, and attenuated isoproterenol-stimulated myocardial contractility. In the same paper, the authors showed that expression of the β-ARKct, the peptide inhibitor of GRK2, led to enhanced contractility in the presence or absence of isoproterenol (129). Many other mouse models followed this initial report as listed in Table 3. Of note, it is important to highlight that the only GRK global KO that is embryonically lethal is that of GRK2 (117). This is important to note as particularly significant since GRK2 is the predominant isoform expressed during embryogenesis (220). In fact, GRK2 transcripts are detected at 7.5 days post coitum in multipotent cells in early mesoderm and neural crest (220). Interestingly, in mice, while global GRK2 knockout (KO) is embryonically lethal, cardiac-specific ablation of GRK2 is not (158). In fact, normal heart structure, cardiac function, and gene expression were observed in cardiac-specific GRK2 KOs when Nkx2.5-Cre was used. These mice did exhibit enhanced inotropic sensitivity to isoproterenol, impaired lusitropic tachyphylaxis, and aggravated catecholamine toxicity during chronic β-AR stimulation (158). This differs significantly from the situation when GRK2 is eliminated from cardiac myocytes after birth as discussed below (199). Of note, it is important to mention recent studies crossing heterozygous GRK5 and GRK6 knockout mice previously described (84, 85) to generate double GRK5 and GRK6 knockout mice which are embryonically lethal (32).
Among genetic in vivo models of GRKs, involving manipulation of GRKs, GRK2 is the best studied isoform, with many mouse models revealing the different structural elements of this kinase (see Table 3). The majority of GRK-based genetically engineered mouse models have been targeted to cardiovascular tissues and have paved the way that proves that GRKs, especially GRK2 and GRK5, play critical roles in normal and pathophysiological cardiac regulation and function. The remaining portion of this review will focus on discussions concerning key biologically, physiologically, and translationally significant aspects of cardiovascular GRK function that have been largely gained from mouse models. Importantly, these mouse studies have also been largely complemented by human studies where applicable as well as studies in larger animal species. The understanding of GRK function and regulation provided by these mouse models and other studies provides evidence that GRKs and their targeted inhibition represent a potential new and powerful drug class that may yield new treatments for cardiovascular disorders.
III. G PROTEIN-COUPLED RECEPTOR KINASE 2 AND THE CARDIOVASCULAR SYSTEM
A. GRK2 and Heart Failure
Heart failure is the end point of many cardiac diseases which include but are not restricted to myocardial infarction, hypertension, coronary artery disease, cardiomyopathies, congenital heart diseases, and others. To prevent heart failure, it is vital to understand the molecular events that interfere with β-AR signaling homeostasis. One of the hallmarks of heart failure is increased activation of the sympathetic nervous system, leading to an increase in circulating catecholamines (reviewed in Ref. 151). In the heart, that translates into increased activation of the β-ARs. This augmentation in stimulus is thought to be initially a compensatory mechanism, to reestablish cardiac performance, increasing heart rate and contractility. Unfortunately, if this increase in signal transduction remains unabated, it can lead to the deleterious effects seen in heart failure. Data have shown that these pathological events are triggered by enhanced GRK2 activity in both the heart and the sympathetic nervous system (151, 213).
The myocardium uses positive chronotropic and ionotropic effects mediated by β1-ARs and β2-ARs through Gs coupling to accomplish the demand elicited by increased agonists. In the nonfailing heart, 75% of β-ARs are β1, with distribution patterns fairly similar among all cardiac chambers (reviewed in Ref. 14). Interestingly, chronic stimulation of β-ARs has been reported to produce opposing outcomes; stimulation of β1 and β2 would lead to deleterious and cardioprotective effects, respectively (86). Although initially thought to be only present in adipose tissue, functional β3-ARs are also upregulated in the failing myocardium (44, 62). Thus upregulation of active β-ARs increases the necessity for GRKs to turn off this increase in β-AR signal transduction. Indeed, chronic catecholamine exposure in vivo can lead to upregulated GRK2 (111). Continued activation of β1-ARs and β2-ARs on myocytes following cardiac stress/injury leads to their desensitization via enhanced GRK2 activity and chronically this leads to receptor downregulation. Importantly, a hallmark characteristic of the failing heart is the loss of β-AR-mediated inotropic reserve manifested by both a loss (∼50%) of β-AR receptor density but also overdesensitization of remaining receptors, attributed to increased GRK activity (151, 213).
Importantly, human studies have provided compelling evidence that during heart failure there is both increased catecholamine circulation as well as increased myocardial levels of GRKs (reviewed in Ref. 178). While in an aging human there are no changes in GRK activities (136), alterations in cardiac GRKs are evident in the development of human heart failure (136, 184, 253). GRK activity was reported to be elevated in end-stage heart failure patients (136). Increased myocardial and lymphocyte levels of GRK2 have been reported in human heart failure (100, 108). Myocardial biopsy samples and peripheral blood lymphocytes from 23 human transplant patients also revealed an increase in both GRK2 and GRK5, although there was no correlation between myocardial and lymphocyte levels. The authors on this study found that in myocardial tissues, the β1/β2 ratio was 57/42%, which differs from normal 80/20% ratio, and they also found significant increases in both GRK5 and GRK2 (2). In peripheral blood lymphocytes, the authors observed a substantial increase in GRK2 levels and a β1/β2 ratio of 3/96% (2). Human patients with ischemic and dilated cardiomyopathies have elevated GRK2 mRNA levels (3-fold) as well as increased activity (2-fold) (253). GRK2 levels were elevated in all chambers in human hearts of left ventricular overload disorders as well as in dilated cardiomyopathies (67). More recently, mRNA and protein levels of GRKs in left ventricular myocardium were analyzed from 31 patients with advanced heart failure undergoing transplantation. Compared with control nonfailing left ventricular samples, advanced heart failure patients possessed augmented levels of GRK2 and GRK5, but not GRK3 (3).
Animal models recapitulating human heart failure have yielded substantial and important mechanisms involving GRKs and how the increase in levels can correlate to the unfavorable prognostic cardiac pathology. These models have been extensively discussed in References 14, 95, and 178. Of note are the studies in rats subjected to surgical coronary artery occlusion. Hypertrophic failing hearts showed increased GRK2 protein and activity while no major change was observed in hypertrophic nonfailing hearts (250). Another study in hypertrophic cardiomyopathic mouse hearts showed increased GRK2 protein expression and activity compared with nontransgenic littermate controls (NLC) (81). In rats subjected to left coronary artery ligation, myocardial mRNA and protein levels of GRK2 and GRK5 were elevated in failing hearts (259). In spontaneously hypertensive heart failure rats, GRK2 increased at the intercalated disk in 6-, 12-, and 24-mo-old animals, while no change was observed in total protein expression (281). In mice, pressure overload following transverse aortic constriction also led to an increase in cytosolic and membrane GRK (47).
The most abundant circulating catecholamines are epinephrine, norepinephrine, and dopamine. Circulating epinephrine and norepinephrine primarily derive from two major sources in the body. Sympathetic nerve endings are capable of releasing their own stored norepinephrine directly onto the cardiac muscle. Alternatively, upon acetylcholine stimulation of nicotinic cholinergic receptors at the adrenal medulla, chromaffin cells release previously synthesized and stored epinephrine (mainly) and norepinephrine (reviewed in Ref. 151). This is pertinent to heart failure as sympathetic nervous system hyperactivity chronically stimulates β-ARs leading to increased GRK levels and worsens prognosis. Adrenal levels of GRK2 have been shown to be upregulated in chronic heart failure, leading to augmented catecholamine release due to downregulation of α2-ARs in chromaffin cells (149). Moreover, in rats, modulation of adrenal catecholamine release has been manipulated by overexpression of GRK2 or expression of β-ARKct (152). More important are the studies in mice specifically harboring deleted GRK2 in chromaffin cells. At 4 wk postmyocardial infarction, adrenal-targeted GRK2 KO animals showed decreased circulating catecholamines and consequently reduced sympathetic activity. This was associated with enhanced cardiac function and improved β-adrenergic reserve compared with control animals (150). Overall, the link between adrenal GRK2 and α2-AR-mediated negative feedback on catecholamine release appears important and to have great implication in adverse cardiac remodeling and progression of heart failure.
Due to the strong correlation between GRK2 levels and heart failure, GRK2 inhibition has been proposed as a potential treatment for cardiac injury. Studies in a mouse model expressing the β-ARKct as a peptide inhibitor of GRK2 has added significant evidence supporting this notion. Cardiac-targeted β-ARKct transgenic expression in mice led to a phenotype of significantly enhanced cardiac function and increased sensitivity to acute β-AR agonist exposure (129). Reciprocally, cardiac-targeted transgenic mice with enhanced levels of GRK2, at levels found in failing human myocardium, leads to a phenotype of significantly aberrant β-AR signaling and a loss of inotropic reserve (129). These mice with cardiac β-ARKct expression have also been used to rescue several models of heart failure including dilated cardiomyopathy (210) and hypertrophic cardiomyopathy (82, 97). The Muscle Lim Protein (MLP) KO mouse, which is a model of dilated cardiomyopathy, was the first heart failure mouse model rescued by β-ARKct expression as GRK2 inhibition in myocytes from birth resulted in a MLP KO mouse that never developed pathology (210). Calsequestrin (CSQ) is a sarcoplasmic reticulum calcium-binding protein. Mice overexpressing this protein have a severe hypertrophic cardiomyopathy phenotype with shortened survival (97). Surprisingly, mice expressing the β-ARKct and overexpressing CSQ have significantly enhanced survival that was synergistically potentiated by β-AR blocker treatment (86). This was the first example that GRK2 inhibition could be complementary and even more beneficial than the β-blockade, which is a standard of care for human heart failure.
The inhibition of GRK2 by β-ARKct expression is anticipated to be mixed as it will also block other Gβγ-dependent signaling processes so to ultimately prove that GRK2 inhibition is the therapeutic target of β-ARKct in heart failure. Studies have been carried out using the GRK2 KO mice. As stated, global homozygous GRK2 mice are lethal (117). However, heterozygous KO mice are viable and present with enhanced cardiac function essentially identical to β-ARKct transgenic mice (211). Interestingly, these KO mice with 50% less GRK2 expressed in their hearts could be further enhanced functionally with cardiac β-ARKct expression (211). GRK2 has also been genetically ablated only in cardiomyocytes at or after birth, and these animals are protected from heart failure development after myocardial infarction showing that GRK2 activity is indeed pathological in the heart following cardiac injury (199). In fact, induction of cardiomyocyte GRK2 gene ablation after postmyocardial infarction in mice significantly improved global cardiac function and reversed adverse ventricular remodeling (199). Overall, these studies in mice show that the primary target of β-ARKct expression is GRK2 inhibition, although there appears to be slight intricate mechanisms that are unique to β-ARKct expression versus GRK2 KO. One of these is the regulation of the L-type calcium channel that is enhanced with loss of both GRK2 and β-ARKct expression (200, 263). β-ARKct expression disinhibits the L-type current through Gβγ binding (263), while the loss of GRK2 enhances L-type currents independent of Gβγ (200). Other mechanistic differences may exist, but these models clearly show that limiting GRK2 activity in the heart after injury prevents and even reverses heart failure.
Studies beyond the mouse have solidified GRK2 inhibition via the β-ARKct as a viable therapeutic strategy for heart failure (reviewed in Ref. 260). First, adenoviral-mediated intracoronary-mediated myocardial delivery of the β-ARKct to rabbit models of cardiac dysfunction and heart failure has led to improved cardiac function (221, 248, 273). In fact, when β-ARKct was expressed before a myocardial infarction, the prevention of β-AR desensitization and downregulation was shown to prevent heart failure development (273). Recently, preclinical studies using adeno-associated virus-serotype-6 (AAV6) to express β-ARKct have shown reversal of heart failure in both rats (208) and pigs (198). These data have been submitted to the United States Food and Drug Administration, and a phase I clinical trial utilizing β-ARKct gene therapy for GRK2 inhibition is currently underway with a prospective start date in 2016.
Additional evidence in support of the potential that GRK2 inhibition has in treating heart failure patients comes from recent studies revealing that paroxetine, a drug that targets serotonin reuptake centrally and that is currently used to treat depression and central nervous system-related disorders, is capable of inhibiting GRK2 as an off-target (249). Paroxetine has been shown to increase myocyte contractility through GRK2 inhibition, and further studies will investigate whether it can improve cardiac function in heart failure models (249). Interestingly, paroxetine is known to have developmental cardiovascular side effects if given to pregnant mothers (154), and this is consistent with GRK2 KO mice (117). Although paroxetine may not be an ideal candidate for heart failure therapy, currently it is a prototype molecule that can lead medicinal chemists to develop the first selective small molecule pharmacological GRK2 inhibitor. Of note, another compound gallein, which is not a candidate for a viable human drug, has been shown recently to reverse the CSQ transgenic mouse model of heart failure (122). Gallein actually targets Gβγ binding and so is analogous to β-ARKct expression. Although other targets of Gβγ binding were found, the primary mechanisms for heart failure reversal with gallein was Gβγ-GRK2 inhibition (122). Overall, evidence strongly supports the fact that inhibition of GRK2 in treating heart failure patients can be beneficial though β-ARKct gene therapy or emerging small molecules, and this will probably become a clinical reality in the near future.
B. GRK2 and Acute Cardiac Injury
In humans, neurohormonal activation such as adrenergic overdrive has been observed in patients acutely after cardiac stress/injury including after a myocardial infarction (230). In addition, increased levels of GRK2 have also been reported in other human studies of disease states (reviewed in Refs. 14, 27). In animal models of cardiac injury, GRK2 upregulation is one of the first cardiac molecular changes seen (159, 273). Of note is the observation that increased levels of GRK2 correlates with clinical severity and left ventricular stroke volume (3). In a 2-yr follow up study in humans with acute coronary syndrome, GRK2 levels increased during ST-segment elevation in myocardial infarction and correlated with decreased ejection fraction and end-systolic volume (218). In mice, upregulation of GRK2 was observed as early as 7 days after transverse aortic restriction (47), and increased levels of GRK2 appear to precede the prognosis of overt heart failure (7, 109).
Therefore, it is clear that GRK2 upregulation in myocardium occurs acutely after myocardial injury with significant involvement in chronic progression of disease. Importantly, studies have recently shown that altered GRK2 levels and activity also have profound and significant effects in the heart following acute cardiac injury (30, 72). First, when cardiac-targeted GRK2 overexpressing transgenic mice are subjected to ischemia/reperfusion (I/R) injury, they present with significantly increased injury (i.e., cardiac infarction size), and this was due to increased myocyte cell death due to apoptosis in the ischemic region (30). Interestingly, β-ARKct transgenic mice displayed significant cardioprotection after I/R injury (30). This protection against acute ischemic injury due to GRK2 inhibition with β-ARKct was confirmed in cardiac-specific GRK2 KO mice as both mice with loss of cardiomyocyte GRK2 at birth or induced loss of GRK2 in adulthood had significant cardioprotection after I/R injury (72). The mechanisms responsible for the pro-death activity of GRK2 and cardioprotection of lower GRK2 activity and expression are not entirely clear, although in the transgenic study, inhibition of eNOS-NO system via GRK2 was observed while β-ARKct increased cardiac NO levels after I/R and improved survival signaling (30). This is especially interesting due to the recent discovery of a dynamic interrelationship between GRK2 and eNOS in the myocyte (106). In the cardioprotected GRK2 KO mice, an inhibition of cytochrome c release from mitochondria was observed as well as decreased myocyte apoptosis due to lower caspase-2 and -9 activities (72). Thus it appears that acute GRK2 inhibition is a target for novel therapeutic intervention as ischemic cardioprotection without the necessity for preconditioning of the myocardium. Paroxetine, as a GRK2 inhibitor, although not ideal for the chronic treatment of heart failure, may be a candidate for acute single-dose administration around the time of reperfusion of postmyocardial infarction patients and is being considered for such preclinical testing due to potential cardioprotective properties.
The cause of increased apoptosis after acute ischemia with enhanced GRK2 activity is not entirely understood, but the above studies in GRK2 KO mice are consistent with mitochondrial-dependent cell death pathways being involved. This is especially interesting since studies by our lab (43) and others (83, 168) have shown GRK2 to be localized to mitochondria, which is a novel area of cellular localization for this GRK. This is of particular significance to I/R injury as the mitochondrion is the major source of energy to cells. However, there is still no consensus in functional role for mitochondrial GRK2. Of note, mitochondria from mice overexpressing GRK2 possess premature opening of the membrane permeability transition pore compared with nontransgenic littermate mice (43). Studies in HEK cells have argued that GRK2 overexpression is involved in increasing ATP content and I/R in quadriceps femoris muscle of Cre-deleted GRK2 mice leads to decreased ATP content compared with floxed control mice (83). On the other hand, our lab has shown that GRK2 localization to the mitochondria is detrimental to the cardiac outcome post-I/R. GRK2 has been shown to interact with Hsp90 (148) and to be phosphorylated by ERK at Ser670 (188). In cardiac cells, translocation of GRK2 to the mitochondria is mediated by ERK phosphorylation of GRK2 and Hsp90 binding (43). With the use of geldanamycin or N-acetylcysteine (Hsp90 inhibitor and antioxidant, respectively), post-I/R injury translocation of GRK2 to the mitochondria was attenuated (43). Also, the use of COOH-terminal mutations of GRK2 where Ser670 was replaced with alanine revealed a loss of mitochondrial localization following oxidative stress and, importantly, less apoptosis indicating that GRK2 targeting to the mitochondrial is a prerequisite for its pro-death activities (43). Figure 3A shows the proposed mechanism for GRK2 translocation to the mitochondria. Figure 3B shows immunogold electron micrographs from cardiac tissue of GRK2-overexpressing mice (GRK2Tg; bottom) and a nontransgenic littermate control mouse (NLC; top). This clearly shows the localization of GRK2 to the mitochondria and that overexpression of GRK2 leads to increased mitochondria GRK2 even at baseline. The functional impact of this localization is not yet clear but under active investigation.
C. GRK2 and Hypertension
Increase in blood pressure can be mediated through increase in catecholamines (norepinephrine or epinephrine), which act primarily on α-ARs (α1 and α2), potentially altering peripheral resistance. Other hypertensive neurohormones such as angiotensin II and endothelin-1 can also induce chronic GPCR activation that can lead to increased peripheral resistance. Alterations in peripheral resistance can ultimately lead to an attenuation of vasodilation, subsequently increasing cardiac afterload. Hypertension dramatically increases the risk for ischemic heart disease, which can in turn lead to stroke as well as cardiac and renal failure. Although the kidney and heart are at the core of blood pressure control, the contractile state of the vasculature is of major importance, due to the fact that blood flow is inversely proportional to the radius of the vessel to the fourth power (reviewed in Ref. 99). Interestingly, GRK2 levels are increased in young patients with hypertension compared with both normotensively young and old patients (92). In agreement are the studies in hypertensive patients showing increased peripheral blood lymphocyte GRK2 levels (116). Overall, increased GRK2 levels are correlated with increased blood pressure (94). In hypertensive rat models, there is increased GRK2 in both lymphocytes and vascular smooth muscle (93). Importantly, vascular-targeted increase of GRK2 via transgenesis is sufficient to increase mean arterial blood pressure by ∼20% compared with control NLC mice (69). The increased levels of GRK2 in vascular smooth muscle were also accompanied by thickening of the vessels and cardiac hypertrophy (69). This was attributed to the enhanced desensitization of GPCRs that relax blood vessels such as β2-ARs (69). Of course, the mechanism for how GRK2 may induce hypertension is not fully understood. Data with β-blockers would suggest it may not be entirely due to uncoupling of β-ARs, and it is important to keep in mind that the noncanonical actions of GRK2 such as MAP kinase regulation may be in play here.
GRK2-dependent changes in blood pressure and vascular homeostasis may also occur within endothelial cells where GRK2 is expressed. Consistent with results in cardiomyocytes (30, 106), lowering GRK2 levels in vascular endothelial cells increases NO bioavailability, which can lead to enhanced vasorelaxation, and this can protect against hypertension in mice (9, 245). Beneficial effects of lowering GRK2 in endothelial cells also include improved insulin signaling as recently shown in spontaneously hypertensive rats (279) and in a mouse model of type 2 diabetes (244). These data coupled with vascular smooth muscle targeting support GRK2 inhibition, not only in heart failure, but for hypertension as well. However, it must be noted that GRK2 has also been shown to be important in regulating endothelial inflammation and oxidative stress (51, 239). For example, lower GRK2 exclusively in murine endothelial cells was recently shown to be associated with increased reactive oxygen species (51). This is intriguing since lower GRK2 in endothelial cells in mice with global loss of GRK2 (heterozygous GRK2 KO mice) are protective against oxidative stress (9). Thus more research is needed to reconcile these findings.
Race is also a factor that contributes to the development of hypertension (138, 223). GRK2 levels were measured in a cohort of 133 African-American patients. Quantification of GRK2 levels in lymphocytes of hypertensive patients (systolic pressure above 130 mmHg) showed a 2-fold increase in protein expression and a greater than 40% increase in GRK2 activity (53). In addition, the same study found a direct correlation of GRK2 (but not GRK5) between plasma norepinephrine levels and systolic blood pressure in black adults. Overall, elevated levels of GRK2 in lymphocytes and vascular smooth muscle have been implicated in both human hypertension and animal models of the disease. This increase in GRK2 levels and activity seem to be very pertinent especially when hypertension takes the overt turn to heart failure (99). Many have suggested using levels of GRK2 in lymphocytes as a biomarker in heart failure at least to monitor response to treatment (108). It may also be a risk predictor of heart failure development.
IV. G PROTEIN-COUPLED RECEPTOR KINASE 5 AND CARDIAC DISEASE
A. GRK5 and Its Role in Pathological Cardiac Hypertrophy and Heart Failure
In addition to GRK2, GRK5 is found at high levels in the heart (2, 67). In fact, of all the tissues expressing GRK5 in the body, its highest levels are found in myocardium (134; also see Table 1). Interestingly, in myocardial tissue of transplanted hearts, GRK5 was shown to be the most commonly expressed GRK (2). Recently, many studies in human subjects have suggested GRK5 as a key player in the development of cardiac hypertrophy and heart failure. GRK5 levels were increased by 68% in the left ventricle of hearts with dilated cardiomyopathy and by 48% in volume-overload patients (67). In the failing myocardium, GRK5 protein was upregulated, and an increase in GRK5 protein and mRNA was observed in patients treated with β-agonists compared with those under β-blocker treatment. In this study, β-blockade led to GRK5 mRNA levels that were comparable to nonfailing hearts, although protein content only showed a decreased trend compared with failing hearts (3). Finally, in human studies comprised of ischemic, dilated, and nonischemic/nondilated cardiomyopathies, expression levels of GRK5 inversely correlated with left ventricular end-systolic and diastolic diameters (162). On the basis of these human studies, it has been proposed that GRK5 is a potential therapeutic target for the treatment of heart failure (reviewed in Ref. 107).
Like GRK2, the canonical function of GRK5 is to terminate GPCR signaling. Indeed, cardiac-targeted GRK5 overexpressing transgenic mice have severely attenuated β-AR signaling and myocardial in vivo responsiveness, demonstrating that, like GRK2, this kinase can act as a β-AR kinase in vivo (212). Cardiac-targeted overexpression of GRK2 or GRK5 in mice has been shown to result in different regulation of GPCR signaling. For instance, adenosine receptors are targeted by the overexpression of GRK5 but not GRK2, while angiotensin II receptors are targeted by the overexpression of GRK2 and not GRK5 (212). Of importance to cardiac function, recent studies have revealed functions for GRK5 beyond its canonical role of regulating GPCRs. GRK5 can bind to calmodulin with an affinity 40 times higher than GRK2 (215). The binding to calmodulin prevents membrane-mediated signaling, although phosphorylation of soluble substrates is unchanged (49, 79). This binding of calmodulin has been shown to be essential for the non-GPCR activity of GRK5 in the heart as this causes the nuclear accumulation of GRK5 (90). GRK5 has also been shown to bind to DNA and to have an active nuclear localization signal (NLS) corresponding to amino acids 388–395 in the catalytic domain (see Figure 2). Altering the NLS to keep GRK5 out of the nucleus does not appear to significantly affect GPCR kinase activity (120). More recently, GRK5 has been shown to possess a nuclear export signal (NES) corresponding to amino acids 259–265 (see Figure 2) (119). Of note, the NLS within the catalytic domain appears to be a common feature of GRK4 subfamily members (119). Following cardiac hypertrophic stimulation, which causes calcium-calmodulin activation, GRK5 translocates to the nucleus where it has noncanonical activity as a class II histone deactylase (HDAC) kinase and induces gene transcription through the derepression of MEF2 (157, 282). This has been shown in transgenic GRK5 overexpressing mice to lead to exaggerated cardiac hypertrophy and accelerated heart failure (157). Importantly, transgenic mice overexpressing a GRK5 with its NLS mutated have normal responses to hypertrophic stimulation showing that it is this novel activity of GRK5 responsible for the pathology (157, 282). To confirm the importance of GRK5 to cardiac pathology when there is abnormal hypertrophic stress, GRK5 KO mice (both global and cardiac-specific) have significantly less hypertrophy and improved function after ventricular pressure overload (89). These data demonstrate the first known pathophysiological role for a noncanonical activity of a GRK, and a schematic of this mechanism of GRK5 in pathological cardiac hypertrophy is shown in Figure 4A. Additionally, studies in myocytes of spontaneously hypertensive heart failure rats show GRK5 protein localization to the nucleus (Figure 4B), further supporting the notion of GRK5 redistribution in cardiac diseased states.
Genome-wide RNA interference in Drosophila identified G protein-coupled receptor kinase (Gprk)2, which in other organisms is homologous to GRK5, to be a conserved regulator of NFκB (255). Further studies have also confirmed GRK5 as a mediator of NFκB signaling in mammals, with potential implications in cardiac hypertrophy and other cardiovascular pathologies, although the mechanism of action is still not fully understood (114, 170, 232). NFκB is a highly regulated dimeric transcription factor present as heterodimers (p50:p65) or homodimers (p50:p50 or p65:p65) (87). Extracellular stimulus induces the activation of IκB kinase (IKK)-mediated phosphorylation and degradation of NFκB repressor, IκBα, causing NFκB to be free to shuttle to the nucleus (13, 66). However, this recent view has been challenged by the findings that NFκB:IκBα complexes are able to continuously shuttle between cytoplasm and nucleus by nuclear transporters (24, 87). With that in mind, studies in HEK and BAEC cells overexpressing GRK5 showed nuclear accumulation of IκBα, which was linked to inhibition of NFκB signaling. The mechanism suggested in this study was due to a physical interaction between GRK5-RH domain (residues 50–176) and IκBα NH2 terminus rather than phosphorylative events (232). In addition, in support of this mechanism, intracardiac injection of a NH2-terminal peptide of GRK5 (residues 1–176 containing the RH domain) reduced left ventricular hypertrophy by inhibiting NFκB-mediated hypertrophic gene expression (233). A schematic view of this mechanism is shown in Figure 5A. Nevertheless, in neonatal myocytes exposed to the hypertrophic agonist, phorbol 12-myristate 13-acetate (PMA), there were increased levels of GRK5, p50, and p65. Expression of dominant negative IκBα, usage of N-acetylcysteine, or silencing of p65 showed decreased levels of GRK5. Activation of NFκB led to binding of p50 and p65 to the GRK5 promoter and subsequent increase in GRK5 expression after hypertrophic stimuli (115). Conversely to the mechanism proposed in Figure 5A, overexpression of GRK5 increased levels of p50 and p65, and silencing of GRK5 led to decreased levels of p50 and p65 and loss of NFκB DNA binding activity (114). A schematic of the latter mechanism is shown in Figure 5B.
Moreover, in other studies, GRK5 has also been implicated in mediating NFκB activation in macrophages, and these cells are important in injured myocardium (175, 176). Overexpression of GRK5 in Raw 264.7 macrophages enhanced TNF-α-mediated NFκB activity, and knockdown of GRK5 led to diminished NFκB activity, in agreement with the mechanism in Figure 5B. Moreover, in macrophages, GRK2 and GRK5 interacted with IκBα NH2 terminus (in agreement with Figure 5A), and GRK5 (but not GRK2) phosphorylated IκBα at the same residues as IκBα kinase does (Ser32 and Ser36) (175). Due to the role of NFκB signaling in inflammation, studies have also looked at how GRK5 levels modulate inflammation. In a sepsis model, GRK5 KO mice showed less inflammation than its control, with diminished expression of NFκB-driven genes and decreased phosphorylation of IκBα (170, 176). In agreement, lipopolysaccharide (LPS)-induced IκBα phosphorylation was significantly reduced in macrophages from GRK5 KO mice. LPS-induced p65 translocation to the nucleus and NFκB DNA binding was significantly diminished in the GRK5 KO mice (176).
B. GRK5 Polymorphism and Its Role in Cardiac Disease
Genetic polymorphisms have been suggested to be involved in inter-individual differences in response to β-AR activation and its outcome to disease. GRK5 nonsynonymous polymorphisms have been reported in humans while there is none in GRK2 (137). In particular, many studies have looked into GRK5 residue 41 glutamine to leucine polymorphism due to its high prevalence in the population from African genetic background. GRK5-L41 has been reported to represent a gain-of-function polymorphism by enhancing β2-ARs uncoupling during persistent agonist exposure (266). This polymorphism has been suggested to present a greater risk for loss of bronchodilatory function during β-AR agonist stimulation in asthma patients (266). Studies in transgenic mice bearing cardiac GRK5-L41 showed an enhanced uncoupling of isoproterenol-stimulated β-adrenergic signaling compared with GRK5-Q41 or nontransgenic littermate mice (137). Interestingly, GRK5-L41 and GRK5-Q41 show similar nuclear localization and ability to stimulate HDAC nuclear export (282) (see Figure 4). In this study, diminished β-AR-signaling was viewed to be beneficial as it was interpreted as mimicking β-blockade. In fact, African-American GRK5-L41 heart failure patients showed prolonged survival or prolonged time for cardiac transplantation (137). In addition, in a cohort of 2,673 acute coronary syndrome patients, improved outcome was observed in African-American patients harboring GRK5-L41 (56). Moreover, analysis of human patients carrying GRK5-L41 was found to be protective in treated hypertensive patients with coronary artery disease, although no influence in blood pressure response to antihypertensives was seen (142). Yet, another study, looking at patients with left ventricular apical ballooning syndrome showed an increased prevalence of GRK5-L41 polymorphism (235), although no link to prognosis was drawn in this study. Overall, it is important to point out that while nuclear events may be an important factor observed in these polymorphisms, canonical GPCR phosphorylation and desensitization as well as noncanonical induction of β-arrestin-biased signaling due to altered GRK5 activity may be a contributing factor to what has been observed with this GRK5 polymorphism.
More recently, intronic short tandem repeat (STR) (CAn) polymorphisms of GRK5 have been linked to type 2 diabetes mellitus (T2DM) (278). A cohort of 1,164 patients was classified in four groups (normal fasting glucose, impaired fasting glucose, impaired glucose tolerance, and T2DM), and five intronic CAn alleles in GRK5 were identified. CA16 was identified as the only GRK5 STR to be associated with abnormal blood glucose (278), yet the mechanism in which this would alter transcription is still unknown. Overall, GRK5 polymorphisms represent a potential mechanism in which genetic variability could associate with cardiac disease susceptibility and risk factor. Nevertheless, further investigation needs to address mechanism of action, polymorphisms that are silent versus nonsilent, and which cardiac conditions GRK5 polymorphisms would mostly influence, as there are also reports of GRK5 polymorphisms not associating to certain diseases (74, 247).
V. OTHER GRKs AND THE CARDIOVASCULAR SYSTEM
A. GRK3 and the Heart
GRK3 has been shown to be expressed in the human heart although to a much lesser extent than GRK2 and GRK5 (2, 67, 162). Clinical studies have shown that decreased heart performance inversely correlates with levels of GRK3 (162). Other studies, however, have reported no change in GRK3 levels in heart failure patients (3). Nevertheless, discrepancies such as these may be explained, at least in part, by varying heart failure etiologies, such as the augmented levels of GRK3 in volume overload patients and unchanged levels of GRK3 in dilated cardiomyopathy hearts (67). Studies in mice have been particularly helpful in unveiling the potential functional role of cardiac GRK3 (see Table 3). GRK3 KO mice developed normally, albeit displaying impairment in odorant-induced desensitization of cAMP responses (182). Cardiac-restricted expression of GRK3 peptide inhibitor (GRK3ct, analogous to the β-ARKct of GRK2) in mice led to hypertension derived from increased cardiac output, hyperkinetic myocardium, and increased α1-AR responsiveness (261). No alterations in cardiac mass or left ventricular dimensions were observed in GRK3ct mice at baseline. Nevertheless, after chronic ventricular pressure-overload, preserved cardiac function was observed only in GRK3ct transgenic mice compared with NLC control mice (264). This preservation in function after pressure overload was observed even though there was similar induction of hypertrophic gene expression as seen in control mice (264). Based on this, during development of heart failure, it can be inferred that GRK3 may play a role in receptor responsiveness without modulating hypertrophic responses (as in the case of GRK5). In addition, mice overexpressing GRK3 in the heart have no hypertrophy (68) and possessed normal β-AR-signaling both biochemically and functionally, although altered thrombin-mediated signaling was observed (110). In hybrid mice studies, GRK3 overexpression has been highly implicated in diminishing α1-adrenergic receptor signaling (68). While studies in mice have provided clues as to potential cardiac responses driven by GRK3, further studies still need to elucidate the role of GRK3 in human heart failure.
B. GRK4 and GRK6 in Hypertension
To a much lesser extent, GRK4 and GRK6 have also been implicated in cardiac dysfunction. Since the levels of GRK4 and GRK6 in the human heart are rather low (67), it has been hypothesized that most of the cardiac impact by these two isoforms is through an indirect mechanism, possibly through altered blood pressure. GRK4 is the main GRK isoform in the testes, although it is also expressed in the kidneys and brain, and at minor levels in other organs (see Table 1) (194, 256). GRK4 has been reported to possess four alternative splice variants (194). Furthermore, GRK4 single nucleotide polymorphisms have been previously reported and include R65L, A142V, and A486V. GRK4-A486V has been associated with hypertension in Italian patients (15), in Chinese patients (267), and in an Australian cohort of patients (234). Although in these three studies no association between hypertension and GRK4-R65L or GRK4-A142V variant was seen, increasing copies of GRK4-R65L and GRK4-A142V were associated with reduced atenolol-induced diastolic blood pressure lowering (256). In fact, mice harboring the GRK4-A142V variant had increased arterial blood pressure compared with control animals (73). GRK4-A142V mice have defective coupling between D1 dopamine receptors and GRK4 in the renal proximal tubule; thus the kidneys are unable to properly excrete sodium, leading to hypertension (73). In addition to the above-mentioned human studies, no single site GRK4 polymorphism showed significant association with hypertension in a population from Ghana (275). Yet, in another study, GRK4-A486V has also been implicated with hypertension, as has R65L, but only when in conjunction with other polymorphisms (156). To clarify discrepancies, meta-analysis of all five human studies was recently conducted and subgroup analysis by ethnicity revealed that GRK4-A486V inversely correlated with hypertension among East Asians, while it positively correlates among Europeans. In addition, in Europeans GRK4-R65L also correlated with hypertension (139). Overall, GRK4 polymorphisms may have an impact in the heart by promoting hypertensive phenotypes; however, inconsistent results in human studies have yet to provide solid evidence in this matter.
GRK6 has four alternative splice variants (75, 161, 193, 257), and its association with hypertension has been studied to a much lesser extent. GRK6 mRNA has been reported to be barely detectable in human hearts (67). Due to its close relationship with GRK5 and due to its minor expression in the heart, efforts have attempted to determine whether hypertension-mediated heart failure induces changes in GRK6 expression. In a rat model of spontaneous hypertensive heart failure (SHHF), GRK6 expression was increased in 12- and 24-mo-old animals compared with age-matched controls with an increased cytoplasmic-to-membrane ratio as seen by confocal imaging (281). Increased localization to the membrane and kinase activity are proposed to occur because of palmitoylation of GRK6 (241). GRK6-A contains three cysteines in the COOH terminus known to serve as substrates for palmitoylation. GRK6-B (slice variant-B) lacks those palmitoylation sites but has a site for protein kinase phosphorylation, while GRK6-C has none of the these two motifs, and GRK6-D lacks catalytic activity (161, 193). While GRK6 A-C differ in their COOH terminus, which is involved in membrane targeting and receptor interaction, in COS-6 cells GRK6-D is exclusively localized to the nucleus (257). Importantly, three of the GRK6 variants identified in mouse are likely to exist in humans and rats (257). The link between GRK6 levels, GRK6 variants, and hypertension-mediated heart failure still needs further investigation.
Immune response and inflammation are important cellular mechanisms that may also impact the heart. Important to note are the studies reporting the involvement of GRK6 in lymphocyte chemotaxis and inflammation (76, 124). Lymphocytes from GRK6 KO mice possess impaired chemotactic ability to respond to CXCL12, which is more pronounced in T cells than B cells (76). This effect is specific to GRK6 as other GRK isoforms were not able to rescue the impaired migratory behavior. Polymorphonuclear neutrophils (PMN) are involved in defense against infection, and leukotriene (LT) B4 is one of the most potent chemoattractants of PMN. Studies in GRK6 KO mice revealed increased PMN response towards LTB4 and stromal cell-derived factor 1 (SDF-1) (124). Loss of GRK6 in mice reduces and enhances chemotaxis in lymphocytes and PMN, respectively, suggesting a cell-specific role of GRK6. Acute inflammation response was also increased in GRK6 KO mice (124), and GRK6 deficiency in mice has been implicated with autoimmune disease related to impaired apoptotic cell clearance (165). Nevertheless, GRK6-mediated immune response and inflammation impact on the heart still remain to be fully elucidated.
GPCR signaling is a dynamic process that can quickly adapt to meet cellular demands. This signaling highly depends on proteins that regulate its activity. Two protein families that are intricately linked in the regulation of the activated receptors are the GRKs and β-arrestins. GRKs have a wide array of physiological impact, and alterations in GRKs have been linked to many noncardiac pathologies such as bipolar disease (11), Alzheimer's disease (168), rheumatoid arthritis (145), multiple sclerosis (265), Parkinson's disease (33), and others. This review focused on GRKs, and in particular, the role of cardiac GRKs in maintaining myocardial function in diseased and nondiseased conditions. GRK2 and GRK5 are the main cardiac isoforms. While both isoforms have been shown to mediate canonical GPCR receptor signaling, unique non-receptor-dependent regulatory roles have been recently uncovered. For example, GRK2 has been reported to be involved in mitochondrial homeostasis and GRK5 in nuclear transduction signaling. The noncanonical localization of these GRKs opens new possibilities and exciting potential, as it provides a new avenue in which GRKs could mechanistically exert function in the myocardium.
β-Blockers are one of the most widely used drugs in the treatment of hypertension and congestive heart failure. Like β-blockers, the GRKs are also involved in terminating β-AR signaling. Changes in GRK levels or inactivation of GRKs may reduce or enhance β-AR-mediated signaling, subsequently altering downstream signaling pathways. Human studies suggest that GRK2 and GRK5 are increased in expression and activity in various cardiac diseases. Inhibition of GRK2, as seen by the expression of the β-ARKct as well as the targeted KO of GRK2 in myocytes, has been shown to improve cardiac function in several animal models of cardiomyopathy. It is well established that alterations in GRK levels lead to changes in signaling pathways, which may regulate apoptosis, inflammation, hypertrophy, and/or transcription. While canonical signaling of GRKs in the heart exacerbates progression to heart failure, novel, noncanonical molecular mechanisms of action for GRK2 and GRK5 in the mitochondria and nucleus, respectively, hold promise for future drug development to limit the pathology evoked by upregulation of these kinases after cardiac stress/injury. With the elucidation of novel GPCR biased signaling mediated via β-arrestins have led to more importance for GRK-mediated phosphorylation as GRKs enable this type of signaling. In fact, which GRK can phosphorylate a given receptor appears to control β-arrestin independent signaling.
In addition, new evidence on the role of nonabundantly expressed cardiac GRK isoforms (GRK3, GRK4, and GRK6) is intriguing. These isoforms, though expressed at low levels at baseline, may be important and relevant in diseased states. Finally, it is worth emphasizing that the GRKs are expressed at different levels in various tissues, and the challenge remains to understand the role that GRKs play in various cardiac conditions. The overall correlation between GRK levels to different cardiac disease etiologies still remains to be determined, and further investigation will no doubt continue to uncover important translational aspects that GRKs present in the cardiovascular system.
W. J. Koch is the W. W. Smith Chair in Cardiovascular Medicine and is funded by National Heart, Lung, and Blood Institute Grants R37 HL061690, R01 HL085503, P01 HL091799, P01 HL75443 (Project 2), and P01 HL108806 (Project 3). P. Y. Sato is supported by a postdoctoral fellowship from the Great Rivers Affiliate of the American Heart Association.
No conflicts of interest, financial or otherwise, are declared by the authors.
Present address of M. Schwartz: Advanced Institutes of Convergence Technology, Suwon, South Korea.
Address for reprint requests and other correspondence: W. J. Koch, W. W. Smith Chair in Cardiovascular Medicine, Professor and Chair, Dept. of Pharmacology, Director, Center for Translational Medicine, Temple University School of Medicine, 3500 N. Broad St., 940 MERB, Philadelphia, PA 19140 (e-mail:).
- Copyright © 2015 the American Physiological Society