G protein-coupled receptor kinases: More than just kinases and not only for GPCRsEugenia V. Gurevich a,, John J.G. Tesmerb, Arcady Mushegian c, Vsevolod V. GurevichaaDepartment of Pharmacology, Vanderbilt University, 2200 Pierce Avenue, Preston Research Building, Rm. 454, Nashville, TN 37232, United StatesbLife Sciences Institute and the Department of Pharmacology, University of Michigan, Ann Arbor, MI 48109, United StatescStowers Institute for Medical Research, 1000 E. 50th St. Kansas City, MO 64110, United Statesabstract arti cle i nfoKeywords:G protein-coupled receptorsG protein-coupled receptor kinasesSignalingRegulationPhosphorylationG proteinsRegulator of G protein signalingG protein-coupled receptor (GPCR) kinases (GRKs) are best known for their role in homologous desensitizationof GPCRs. GRKs phosphorylate activated receptors and promote high afnity binding of arrestins, which pre-cludes G protein coupling. GRKs have a multidomain structure, with the kinase domain inserted into a loop ofa regulator of G protein signaling homology domain. Unlike many other kinases, GRKs do not need to be phos-phorylated in their activation loop to achieve an activated state. Instead, they are directly activated by dockingwith active GPCRs. In this manner they are able to selectively phosphorylate Ser/Thr residues on only the activat-edformof the receptor, unlike relatedkinases such as proteinkinase A. GRKs also phosphorylate a variety of non-GPCR substrates and regulate several signaling pathways via direct interactions with other proteins in a phos-phorylation-independent manner. Multiple GRK subtypes are present in virtually every animal cell, with thehighest expression levels found in neurons, with their extensive and complex signal regulation. Insufcient orexcessive GRK activity was implicated in a variety of human disorders, ranging from heart failure to depressionto Parkinson's disease. As key regulators of GPCR-dependent and -independent signaling pathways, GRKs areemerging drug targets and promising molecular tools for therapy. Targeted modulation of expression and/orof activity of several GRK isoforms for therapeutic purposes was recently validated in cardiac disorders and Par-kinson's disease. 2011 Elsevier Inc. All rights reserved.Contents1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412. Structural organization of G protein-coupled receptor kinases. . . . . . . . . . . . . . . . . . . . . . . . 413. Subcellular targeting of G protein-coupled kinase isoforms. . . . . . . . . . . . . . . . . . . . . . . . . 434. Mechanism of activation of G protein-coupled receptor kinases by active G protein-coupled receptors. . . . . 435. G protein-coupled receptor kinases phosphorylate non-G protein-coupled receptor substrates . . . . . . . . 446. Proteins regulated by G protein-coupled receptor kinases in phosphorylation-independent manner . . . . . . 477. Regulation of G protein-coupled receptor kinases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 478. G protein-coupled kinase isoforms more of the same?. . . . . . . . . . . . . . . . . . . . . . . . . . 489. Physiological and pathological roles of G protein-coupled kinase isoforms. . . . . . . . . . . . . . . . . . . 4910. Conclusions and future prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49Conict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52Pharmacology & Therapeutics 133 (2012) 4069Abbreviations: AD, Alzheimer's disease; 2AR, 2-adrenergic receptor; CDK2, cyclin-dependent kinase 2; GAP, GTPase activating proteins; GIRK, G protein-coupled potassiumchannels; GIT, GRK interacting proteins; GPCR, G protein-coupled receptor; GRK, G protein-coupled receptor kinase; HDAC, histone deacetylase; IRS, insulin receptor substrate;KD, kinase domain; LBD, Lewy body disorder; LID, L-DOPA-induced dyskinesia; MS, multiple sclerosis; PD, Parkinson's disease; PDK1, phosphoinositide-dependent kinase 1; PH,pleckstrin homology; PI3K, phosphoinosidite-3-kinase; RGS, regulator of G protein signaling; RH, RGS homology; S1P, sphingosine-1-phosphate; SSRI, selective serotonin reuptakeinhibitor; Smo, Smoothened; Ptc, Patched; Hh, hedgehog. Corresponding author. Tel.: +615 936 2720; fax: +615 343 6532.E-mail address: eugenia.gurevich@vanderbilt.edu (E.V. Gurevich).495152606161610163-7258/$ see front matter 2011 Elsevier Inc. All rights reserved.doi:10.1016/j.pharmthera.2011.08.001Contents lists available at SciVerse ScienceDirectPharmacology & Therapeuticsj our nal homepage:www. el sevi er . com/ l ocat e/ phar mt her a1. IntroductionSignaling via G protein-coupled receptors (GPCR) is terminated byaremarkablyuniformtwo-stepmechanism: aGPCRkinase(GRK)phosphorylatestheactivereceptor, convertingit intoatarget forhigh afnity binding of arrestin. Bound arrestin shields the cytoplas-mic surface of the receptor, precluding G protein binding and activa-tion (Wilden, 1995; Krupnick et al., 1997).Phosphorylation of rhodopsin, a prototypical GPCR, upon its activa-tion by light wasrst described in 1972 (Bownds et al., 1972; Khn &Dreyer,1972).Soonthereafteropsinkinase (modern name GRK11),which selectively phosphorylates active rhodopsin, was identied(Weller et al., 1975). Therst clear evidence that rhodopsin phosphory-lation is necessary for its rapid deactivation was presented in 1980 andled to the hypothesis that this mechanismmay also regulate hormone re-ceptors (Liebman & Pugh, 1980). Within a few years, this idea was con-rmedfor2-adrenergicreceptor(2AR)(Stadel etal., 1983; Sibleyet al., 1985) and later for many others (reviewed in Carman & Benovic,1998). Thedemonstrationof sequencesimilaritybetweenthe2ARand rhodopsin in 1986 (Dixon et al., 1986) led to the recognition of thefamily of G protein-coupled receptors (GPCRs), of which rhodopsin is afounding member. Also in 1986, a kinase that could phosphorylate acti-vated-adrenergic receptors (ARK; modernname GRK2) was identied(Benovic et al., 1986b). This enzyme could also phosphorylate rhodopsinin a light-dependent manner (Benovic et al., 1986a). Phosphorylation ofrhodopsinfacilitatesthebindingof another proteintermedarrestin(called 48-kDa protein at the time), which physically blocks further sig-nalingbythereceptor toheterotrimeric Gproteins (Wildenet al.,1986). Demonstration that desensitization of the 2AR requires a homo-log of arrestin (Benovic et al., 1987)rmly established the paradigm oftwo-step GPCR inactivation, which was later shown to apply to the ma-jority of GPCRs (Carman & Benovic, 1998; Gurevich & Gurevich, 2004,2006b). The cloning of GRK2 in 1989 suggested that it belongs to a dis-tinct lineage of eukaryotic Ser/Thr proteinkinases (Benovic et al.,1989a)thatareasubclassoftheAGCkinasegroup(Manningetal.,2002). In rapid succession, the members of this family expanded to in-cludeARK2(GRK3) (Benovicet al., 1991), GRK4(Ambroseet al.,1992), GRK5 (Kunapuli & Benovic, 1993), and GRK6 (Benovic & Gomez,1993). Cone specic GRK7 (Hisatomi et al., 1998; Weiss et al., 1998) com-pleted the set of vertebrate GRKs.The expression of mammalian GRK1 and GRK7 is largely limited tovertebrate rod and cone photoreceptors although both are also pre-sent in pinealocytes (Somers & Klein, 1984; Zhao et al., 1997, 1999;Pugh & Lamb, 2000). Virtually every mammalian cell expresses sever-alisoformsof non-visual GRKs from early embryonic development.GRK4is expressedat highlevels only intestis (Premont et al.,1996). In addition, GRK4 expression was detected in proximal tubulecells in kidneys, where GRK4 and GRK4 variants reportedly regu-latethesignalingof D1andD3dopaminereceptors(Felderetal.,2002; Villar et al., 2009). GRK4 is also expressed in the brain (Salleseet al., 2000b) and uterus myometrium (Brenninkmeijer et al., 1999).In the rat brain, four GRK isoforms, GRKs 2, 3, 5, and 6, are found asearlyasembryonicday14(Gurevichet al., 2004). Unfortunately,the information about the cell-specic expression of GRK isoforms islimited. We mostly knowtheir distribution at the tissue level. The cel-lular complement of GRK isoforms may prove to be the most impor-tantdeterminantofspecicityinGRKfunction. Forexample, bothGRK1 and GRK2 efciently phosphorylate light-activated rhodopsin,but GRK2 does not perform this function in GRK1 knockout mice.The importance of the GRK-mediated signal shutoff is best illustrat-ed in the visual system, where the lack of GRK1 or sites for GRK phos-phorylation on rhodopsin leads to the loss of photoresponses,photoreceptor degeneration, and blindness in mice and night blindnessin humans (Chen et al., 1995; Yamamoto et al., 1997; Khani et al., 1998;Chen et al., 1999; Zhang et al., 2005; Hayashi et al., 2007; Song et al.,2009; Fan et al., 2010). In other cell types, the results are not as dramat-ic, except in development. In Drosophila, Gprk2,2an ortholog ofGRK4/5/6, is required for wing morphogenesis (Molnar et al., 2007),egg morphogenesis, and embryogenesis (Schneider & Spradling,1997). Knockdown of Grk2 in zebrash embryos induces early develop-mental arrest (Jiang et al., 2009), and knockout of GRK2 in mice is em-bryoniclethal duetoabnormal formationof theheart(Jaberetal.,1996). This lethalitystems fromgeneral, albeit undened, roleofGRK2 in embryogenesis, rather than specic role in the heart develop-ment, because mice withGRK2 ablation specic to the cardiac myocytesdevelop normally (Matkovich et al., 2006).Weknowaboutstructureandfunctionof GRKsalotlessthantheseproteins deserve, consideringthat GRKs criticallyinuencethe function of most GPCRs, which are the targets of a large percent-age of clinically used drugs (Gruber et al., 2010). Many issues are farfrom resolved.GRK specicity towards particular receptor subtypesis one important unanswered question. As mammals have onlyvenon-visual GRKs and N800 GPCRs (Gruber et al., 2010), there are hun-dreds of GPCRs per GRK. It follows that each GRK must have the abil-ity to phosphorylate many different receptors. However, neither thelevel of receptor specicity nor actual preference for particularGPCRsof non-visual GRKsisclear. Wealsoneedtofullydescribehow active GPCRs activate GRKs, which would be greatly facilitatedbyastructureof areceptorGRKcomplex. ThiswoulddenethefullreceptorfootprintontheGRKandprovidegreaterinsightintothe mechanism of kinase activation. GRKs phosphorylate many non-GPCRsubstrates, but it remainsunknownwhetherproteinsotherthan GPCRs can activate GRKs.GRKs2and5havelongbeenconsideredpromisingtherapeutictargets for cardiac diseases (Penela et al., 2006). However, there hasbeenlittleresearchregardingtheirvalueastherapeutictargetsforother conditions. We believe that GRKs, by virtue of their regulatorynature, hold a great promise for therapy of disorders involving an im-balance in GPCR signaling. However, a better understanding of theirstructureandfunctionisaprerequisitefor successful therapeuticintervention.2. Structural organization of G protein-coupled receptor kinasesBased on sequence similarity and gene structure, vertebrate GRKs areclassied into three subfamilies: GRK1 comprising GRK1 (rhodopsin ki-nase) and GRK7 (cone kinase), GRK2 comprising GRK2 and 3, and GRK4comprising GRK4, 5, and 6 (Premont et al., 1999). All GRKs are multi-do-main proteins (Fig. 1) consisting of ~25-residue N-terminal region uniqueto the GRKfamily of kinases, followedby the regulator of Gproteinsignal-ing (RGS) homology domain (RH) (Siderovski et al., 1996), and a Ser/Thrproteinkinase domain(KD) that belongs to the AGCkinase family (Fig. 1).This ~500520 residue assembly is shared by all GRKs. The C-termini ofGRKs contain structural elements responsible for their membrane1In this review, systematic names of vertebrate GRK proteins are used: GRK1 (his-toricnames opsinkinase, rhodopsinkinase), GRK2(historicname -adrenorenergicreceptor kinase 1), GRK3 (historic name -adrenorenergic receptor kinase 2), GRK4,GRK5, GRK6, GRK7 (historic name cone kinase). The proteins and gene symbols for hu-man and other mammalian and non-mammalian vertebrate GRKs are used in accor-dancewithpublishedguidelines[H.M. Wain, E.A. Bruford, R.C. Lovering, M.J. Lusha,M.W. Wright &S. Povey, Guidelines for HumanGeneNomenclatureGenomics 79(2002), pp. 464470; Suggested Xenopus Gene Name Guidelines (2005), http://www.xenbase.org/gene/static/geneNomenclature.jsp, Guidelines for Nomenclature ofGenes, Genetic Markers, Alleles, andMutations inMouseandRat (2010), http://www.informatics.jax.org/mgihome/nomen/gene.shtml, ZFIN Zebrash NomenclatureGuidelines (2011), https://wiki.zn.org/display/general/ZFIN+Zebrash+Nomenclature+Guidelines.]2For invertebrateGRKs,weuse thenames employedin theoriginalpublications,withtheindicationofthevertebrateorthologs, sincetherearenostrictguidelinesforthedenotationsofproteinabbreviationsforinvertebrates[Genetic nomenclaturefor Drosophila melanogaster (2011), http://ybase.org/static_pages/docs/nomenclature/nomenclature3.html].41 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069targeting: GRK1and7carryshort C-terminal prenylationsequences,GRK2and3containpleckstrinhomology(PH)domainsthatinteractwithGprotein(G)subunits(Pitcheret al., 1992; Kochet al.,1993; DebBurman et al., 1996), GRK4 and 6 carry palmitoylation sites(Stoffel et al., 1994; Premont et al., 1996) as well as lipid-binding positive-ly charged elements (Jiang et al., 2007), whereas GRK5 relies onpositivelycharged lipid-binding elements (Pitcher et al., 1996; Thiyagarajan et al.,2004). This domain composition correlates with the degree of sequencesimilarity in the shared domains. Crystal structures of one representativefrom each subfamily were solved: GRK2 alone (Lodowski et al., 2005), incomplex with G-subunit (Lodowski et al., 2003), and with both Gqand G subunits of the heterotrimeric G protein (Tesmer et al., 2005),as well as GRK6 (Lodowski et al., 2006), and GRK1 (Singh et al., 2008).The structures suggest that GRK RH/KD core appeared as the result ofthe insertionof KDinto910loopof anancestral RHdomain(Lodowski et al., 2003, 2006), whereupon different family members ac-quiredvariousadditionalstructuralelements. Afeaturedistinguishingmostofthese GRKstructuresfromotherAGCkinasesisthat the twolobes of the KD are in an open orientation and nucleotide gate regionin the carboxyl-terminal tail (C-tail) of the kinase is disordered regardlessof the presence of an ATP analog or G protein subunits (Lodowski et al.,2006), which is atypical for an AGC kinase. This indicates that the kinaserequires aninducedrearrangement to become active, whichis apparentlyprovided by docking with active GPCRs, consistent with biochemical data(Palczewski et al.,1991; Chenet al.,1993).Inone recent structure ofGRK6, the kinase domain does adopt a relatively closed conformation,andtheuniqueN-terminusandC-tailregioncoalescewiththesmalllobe of the kinase domain to formwhat is expected to be a receptor dock-ing site (see below) (Boguth et al., 2010). This structure reveals an exten-siveat surfaceadjacent tothedockingsitewithabundant positivecharges that likely faces the membrane. This conclusion is supported bythe nding that the lipid anchors of Gproteinsubunits bound to GRK2, al-though disordered in crystal structures, are also generally localized thesame side of the kinase domain (Lodowskiet al.,2003; Tesmer et al.,2005). The GRK N-terminus, as well as several receptor-facing residuesin the kinase domain, seems to mediate allosteric activation of GRKs byFig. 1. Domain structure of GRKs. Number above the structures indicate amino acid residue numbers of human GRKs based on Lodowski et al. (2006). All GRKs have a short N-ter-minal region (green), which is implicated in GPCR binding, followed by RGS homology (RH) domain (magenta). This N-terminal region is unique to the GRK family of kinases. TheRH domain is interrupted by the catalytic domain shared by all kinases (dark yellow). These elements are shared by the GRK2/3 and GRK4/5/6 subfamilies. The dening feature ofthe GRK2/3 subfamily is a C-terminal pleckstrin homology (PH) domain (blue) implicated in binding anionic phospholipids and G. Members of GRK4/5/6 subfamily use alterna-tive mechanisms for membrane targeting, which include palmitoylation (palmitoylation sites are shown for GRK6A; Jiang et al., 2007), patches of positively charged residues (am-phipathic helix motifs (Thiyagarajan et al., 2004; Jiang et al., 2007) are shown as green boxes; N-terminal basic patches (Pitcher et al., 1996; Boguth et al., 2010) are shown as redboxes), and, in case of visual subtypes, prenylation (C-terminal prenylation sites in GRK1 and 7 are shown as red triangles). Residues Arg106 and Asp110 in GRK2/3, among others,are important for binding Gq, a function unique to this subfamily. The position of the key lysine responsible for catalysis in the kinase domain is shown. Mutations K220R in GRK2and 3, as well as K216M/K217M (Sallese et al., 2000a) in GRK4, K415 in GRK5 (Tiruppathi et al., 2000), and K215M/K216M in GRK6 (Lazari et al., 1999) yield kinase-dead GRKs. Theblue box shows the position of the nuclear localization signal (NLS) in GRK5 (residues 388395) (Johnson et al., 2004). Splice variants of GRK4 (GRK4, GRK4, and GRK4) areproduced by in-frame deletion of exon 2 (GRK4), exon 15 (GRK4), or both (GRK4) (Premont et al., 1996; Sallese et al., 1997; Premont et al., 1999) (gene structure is shownunder GRK4 protein; the exons not used in all splice variants are shown in blue) GRK6 splice variants are produced by a frame shift in the C-terminus resulting in a completelydifferent C-terminal sequence in GRK6B as compared to GRK6A and in premature transcription termination in GRK6C (Premont et al., 1999). To generate GRK6A, exon 16 startstwo nucleotides downstream, as compared to the longest variant GRK6B, resulting in a frame shift. An alternative upstream exon 16 encoding one amino acid before the stopcodonis usedtogenerateGRK6C(respective exons arelabeledExon16A, Exon16B, andExon16C). IntheC-terminiofGRK6splicevariantsamphipathichelixresiduesareshown in blue and palmitoylated cysteine in red. Note the lack of palmitoylation sites in GRK6B or GRK6C.42 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069the active GPCRs (reviewed in Huang & Tesmer, 2011). The binding ofother proteins, such as recoverin, to the N-terminus inhibits receptor-de-pendent kinase activation (Higgins et al., 2006).The GRK1 and 7 genes in humans contain 4 exons, the genes of GRK2and3have21exons, andmembersoftheGRK4subfamilyhave16exons. Despite multi-exon structure of all GRK genes, alternative splic-ing was reported only for GRK4subfamily transcripts (Fig. 1). Four splicevariants of humanGRK4were described: GRK4(578 amino acids) con-taining all 16 exons, GRK4 (546 amino acids) lacking in-frame exonXV, GRK4(532aminoacids)lackingin-frameexonII, andGRK4(500aminoacids)lackingbothN-terminal exonII andC-terminalexonXV(Premontetal., 1996;Salleseetal., 1997;Premontetal.,1999). However, orthologous mouse GRK4 protein appears to exist asonly a single 574 amino acid variant (Premont et al., 1999). Interesting-ly, the shorter 545 amino acid splice variant of GRK4 found in rat, whichlacks exon VI, has no human equivalent (Virlon et al., 1998). The longestrat (Virlon et al., 1998) and mouse (Premont et al., 1999) GRK4 variantsdisplay 76% and 77% amino acid sequence identity with human GRK4.Human GRK6 gene generates three alternative splice variants: GRK6A,B, and C. The GRK6B is the longest, with 589 amino acids. When exonXVI starts two nucleotides downstream from that in GRK6B, resultingin a frame shift, GRK6A (576 amino acids), with a different C-tail se-quence is generated. The shortest isoform GRK6C (560 amino acids) isgeneratedby utilizing analternative exon XVI upstream, whichencodesonly a single amino acid before the stop codon. In contrast to GRK4,thesesplicevariantsareconservedinmouse(Premontetal., 1999)and rat (Firsov & Elalouf, 1997).3. Subcellular targeting of G protein-coupled kinase isoformsThe various GRK subfamilies employ several distinct mechanismsthat bring themto or retain themat the membrane, where their integralmembrane substrates GPCRs are found (Fig. 1). Visual subtypes havecharacteristic CaaX motif on the C-terminus for prenylation: GRK1 isfarnesylated(Ingleseetal., 1992a), whereasGRK7isgeranylgerany-lated (Hisatomi et al., 1998). The association of GRKs 1 and 7 with themembraneismediatedbyC-terminal prenylation. Therefore, visualGRKssearchforactiverhodopsinandconeopsinsviadiffusionintwo dimensions on the membrane, which is much faster than 3D diffu-sion. This is important for the sub-second shutoff kinetics of light-in-ducedsignalinginphotoreceptors (Krispel et al., 2006). AlthoughGRK1, which is farnesylated, is rather loosely associated with the mem-brane of rod outer segments (ROS) in the dark (Khn, 1978; Anant &Fung, 1992), and its membrane association is enhanced by light expo-sure (Khn, 1978), farnesylation is important for its activity (Ingleseet al., 1992b). This importance is further underscored by delayed photo-response recovery due to impaired GRK1 transport to the rod outer seg-ment membranes in mice lacking prenyl binding protein PrBP/ (Zhanget al., 2007). Membrane targeting of GRKs 2 and 3 is signaling-depen-dent, becauseG subunits generated by theactivereceptorrecruitthese two GRKs to the locale of activated GPCRs, facilitating GPCR phos-phorylation(Haga&Haga, 1992;Pitcher etal., 1992,1995; Li et al.,2003). Palmitoylationof GRK4and6(Stoffel et al., 1994; Premontet al., 1996; Loudon & Benovic, 1997) and lipid binding by GRK5 andGRK6(Pitcher et al., 1996; Loudon&Benovic, 1997; Stoffel et al.,1998; Thiyagarajan et al., 2004; Tran et al., 2007) enhance afnity forthe lipid bilayer and the kinase activity. The C-terminus of GRK6A splicevariant contains multiple elements promoting or inhibiting membranelocalization, including palmitoylationsites (Stoffel et al., 1994), alipid-binding amphipathic helix (Jiang et al., 2007), and acidic residuesthatinhibit membranelocalization(Vatteretal., 2005; Jianget al.,2007). A non-palmitoylated form of GRK6A is not localized to the plas-ma membrane and instead is detected in the cytoplasmand the nucleus(Jiang et al., 2007). Interestingly, GRK6B and GRK6Csplice variants lack-ing palmitoylation sites still strongly localize to the plasma membranein cultured cells (Vatter et al., 2005). Apparently, the GRK6A C terminuscontains a string of acidic amino acids that negatively regulate mem-brane localization of the protein, since mutation of these residues toneutral or basic rescues membrane localization of non-palmitoylatedGRK6A (Jiang et al., 2007). Thus, GRK6B and GRK6C splicing variantslackingCtailofGRK6A(GRK6Bduetoaframeshift, GRK6Cduetoearlytranslationtermination)areabletolocalizetothemembranewithout the need for palmitoylation. These data suggest that palmitoy-lation works is specic structural context and may not the dominantmechanismof membrane localization of GRK6A, consistent with the re-cent crystal structure of GRK6A in which the palmitoylation sites wereshown to be relatively distant from the expected membrane-bindingsurface (Boguth et al., 2010). GRK5 and GRK6A have been detected inthenucleus(Johnsonetal., 2004; Jiangetal., 2007; Martini et al.,2008), and GRK5 contains a sequence that mediates both nuclear local-ization and DNAbinding (Johnson et al., 2004), suggesting that these ki-nases might participate in the regulation of transcription via epigeneticmechanisms. Indeed, increased accumulation of GRK5 in the nuclei ofcardiacmyocyteshasbeenshowntopromotecardiachypertrophyand early heart failure due to GRK5 acting as a class II histone deacety-lase kinase (Martini et al., 2008).Subcellular localization of GRKs has chiey been described in cul-tured cells, with attention focused on mechanisms of their membranerecruitment. However, many cells are highly compartmentalized andcontainmultipletypesofspecializedmembranesabsentinculturedcells. This is particularly obvious in neurons, which arguably performmore signaling than any other cell type. Mature neurons have very so-phisticated shape with large multi-branched dendritic trees and longaxons that often terminate at multiple pre-synapses, transmitting thesignal to many post-synaptic cells. Importantly, critical role of complexshape of the cytoplasmin the kinetics and reliability of signaling was re-cently demonstrated inrod photoreceptors (Bisegna et al., 2008; Carusoet al., 2011). Obviously, it is of prime interest howGRK isoforms are tar-geted to specialized membrane compartments in specic cell types. Dif-ferential subcellular targeting might contribute to functional specicityofGRKisoforms, evenwithoutreceptorpreferenceatthemolecularlevel. In the brain, GRK2 and 3 isoforms show somewhat different sub-cellular distribution, with GRK3 being more membrane-associated thanGRK2 (Ahmed et al., 2007; Bychkov et al., 2008), although they behavesimilarly in HEK293 cells. GRK5 and 6 preferentially localize to synapticmembranes (Ahmed et al., 2007, 2010; Bychkov et al., 2011). Currently,there are no known mechanisms that would ensure specic localizationto the synaptic as opposed to the plasma membrane. These data suggestthat there are cell type-specic mechanisms targeting GRKs to differentcompartments that need to be investigated in relevant cells, such asneurons, in vivo.4. Mechanism of activation of G protein-coupledreceptor kinases by active G protein-coupled receptorsThe ability to phosphorylate active GPCRs was therst GRK func-tion to be discovered. Receptor phosphorylation by itself can decreaseG protein coupling (Wilden, 1995) and enables high-afnity bindingof arrestin, whichstopsGprotein-mediatedsignalingbyblockingthe cytoplasmic surface of the receptors (Krupnick et al., 1997). Themost striking feature distinguishing GRKs from other kinases is thattheir activity depends on the functional state of the target: GRKs ef-fectively phosphorylate active GPCRs. However, they are clearly capa-ble of phosphorylating other targets at the membrane in response toreceptor activation, as exemplied by the so-called high-gain phos-phorylation of many more rhodopsin molecules that were light-acti-vated(Binder et al., 1990, 1996), as well as phosphorylationofrhodopsin upon activation of transgenically co-expressed cone opsinwithdistinctspectral sensitivityinrods, andviceversa(Shi etal.,2005). Anincreaseof theavailabilityof receptor phosphorylationsites upon their activation does not appear to play a role, because ac-tive GPCRs can enhance GRK phosphorylation of exogenous peptide43 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069substrates,indicating thatdocking with the active receptor directlyactivates the GRK (Palczewski et al., 1991; Chen et al., 1993). This re-ceptor-dependent activationfunctions withnon-cognate pairs, asdemonstrated by the fact that GRK2 robustly phosphorylates rhodop-sininstrictly light-dependent fashion(Benovic et al., 1986a). Ashared activation mechanism of GRKs relying on common structuralfeaturespresentedbyactiveGPCRsthusseemstoallowrelativelyfew GRKs to phosphorylate hundreds of structurally different recep-tors (Palczewski, 1997). Interestingly, GRK4 is the only GRK isoforms(GRK4 splice variant) reported to be constitutively active and capa-bleof phosphorylatingunstimulatedGPCRs(Mnardet al., 1996;Rankin et al., 2006).Current ideas about the mechanism of activation of GRKs have beenrecently reviewed (Huang & Tesmer, 2011), and thus will only be brieysummarized here. Arecent structure of GRK6 revealed for the rst time aGRK in a conformation very similar to those of other activated AGC ki-nases (Boguth et al., 2010). In this structure, the 25 N-terminal residuesare ordered and within this span form a single helix that is engagedin extensive interactions with both the small lobe and the C-tail of the ki-nase domain. These interactions are not possible when GRK6 is in a moreopen conformation (Lodowski et al., 2006). The residues involved in thecontactsbetweentheN-terminalhelixandthe restofthekinasearehighly conserved among GRKs, and site-directed mutagenesis of theseamino acids followed by kinetic analysis reveals pronounced defects inthe phosphorylation of GPCR and soluble substrates (i.e. peptides or tu-bulin), regardless of the GRK subfamily (Huang et al., 2009; Pao et al.,2009; Sterne-Marr et al., 2009; Boguth et al., 2010; Huang et al., 2011).Thus, the inter-domain bridge formed by the N-terminal helix seems tobe a critical structure for maintaining GRKs in a catalytically competentstate, consistent with previous studies indicating a critical role for theGRK N-terminal region in receptor phosphorylation (Palczewski et al.,1993). Indeed, introduction of a disulde bond that covalently attachesGRK1N-terminus toits C-tail improves catalytic efciency(Huanget al., 2011).The most recent GRK6 structure was not determined in complexwithaGPCR. Thus, althoughitseemsclearwhatneedstooccurintheGRKtoachieveanactiveconformation, itislessapparenthowan activated receptor would stabilize this state, and which elementsof theGRKcomprisethereceptordockingsite. However, thenewGRK6 structure provides some clues. Some of the residues in the N-terminal helix are highly conserved in all GRKs, yet face away fromthekinasedomainandarenotinvolvedincontactswiththesmalllobe or C-tail, interacting instead with their equivalents in a crystalcontact toformananti-parallel coiled-coil. TheauthorsreasonedthatthiscrystalcontactcouldserveasasurrogateforanactivatedGPCR, therebyallowingGRK6toassumeaclosedstate. Thismodelpredicts thatmutations of the residuesinvolved should lead to de-fects inGPCRphosphorylation, but not peptide phosphorylation.ThisindeedturnedouttobethecaseforGRK6andGRK1(Boguthet al., 2010; Huang et al., 2011), suggesting that these N-terminal res-idues form at least part of the GPCR docking site. There are thus inter-esting parallels between how heterotrimeric G proteins and GRKs areexpected to interact with active GPCRs, as they both seem to use aprotruding, amphipathichelixthatisintrinsicallydisorderedwhennot docked with a GPCR,as the primary receptor recognition motif(Huang&Tesmer, 2011). Other regions of theGRKundoubtedlyalsocontributetoGPCRinteractions, most likelyintheC-tail andthe small lobe. There have been reports of other regions outside thekinase domain that are suspected of making contributions to receptorbinding (Dhami et al., 2004; Baameur et al., 2010), but these are lessrmly established or could be receptor specic.MostGRKsrequirethepresenceofnegativelychargedlipidstophosphorylate GPCRs efciently, and they also require the cytoplas-mic surface of the active receptor to form a pocket that is not accessi-bleinthe inactive state(Choeet al., 2011;Rasmussen et al., 2011;Standfussetal., 2011)intowhichtheGRKwilldock. Thesizeandphysical properties of this pocket are expected to be highly conservedamong receptors, while the size and sequence of the more exposedcytoplasmic loops of the receptor are clearly not. Structure of opsinin complex with a peptide derived from the C-terminus of Gt dem-onstratesitspotentialtobindamphipathichelices(Scheereretal.,2008), such as those now known to be contained in GRKs and hetero-trimericGproteins. Thenegativechargeoftheinnerleaetofthelipid bilayer would in turn be recognized by a complimentary posi-tively charged surface of the GRK, such as that found immediately ad-jacent to the N-terminal helix of GRK6 (Boguth et al., 2010). Thus, aconsistentmodel forGPCR-mediated GRKactivationisonethatin-volves the activated receptor in its native lipid environment forminga surface that is complimentary in shape and charge to that of GRKsintheiractive, closedconformation, whichisapproximatedbythemost recent structure of GRK6. Sucha mechanismwouldallowGRKstorecognizeabroadarrayofGPCRsubstrates. Oncedocked,the kinase adopts a closed conformation that allows it to phosphory-late any substrate in close proximity, although the docked receptor it-self would be entropically favored. Effective phosphorylation ofmonomeric rhodopsin in nanodiscs by GRK1 has recently been dem-onstrated, which does not support the idea that GRKs dock to one re-ceptor,and phosphorylate another associated with therst, as in aGPCR oligomer (Bayburt et al., 2011). The molecular details of howGRKsinitiallyrecognizereceptorsremainamatterof speculation,but onehypothesisisthat theirintrinsicallydisorderedN-terminiformaninitial low-afnityinteractioninwhat has beentermedy-casting, which is believed to kinetically favor complex formation(Shoemaker et al., 2000; Cortese et al., 2008).5. G protein-coupled receptor kinasesphosphorylate non-G protein-coupled receptor substratesAnintriguing development inrecent years has beena discovery of theability of GRKs to interact witha variety of proteins other thanGPCRs andin many cases to phosphorylate them(Table 1). The data extend the rep-ertoire of pathways whose signaling is controlled by GRKs via phosphor-ylation of various signaling components. The list of non-GPCR substratesnowincludes single transmembrane domain tyrosine kinases (PDGFR),singletransmembranedomainserine/threoninekinases, deathrecep-tors, toll-like receptors, transcription factors and adapter proteins. It re-mains unclear whether this mode of GRKs acting on non-GPCR-linkedsignaling pathways is an exception or a rule. If indeed GRKs participatenot only in desensitization but also in signaling via such a huge varietyof targets, they might play a role inprocesses suchas cell growthandpro-liferation, cell deathandmotility, immunity, cancer, anddevelopment. Assuggested above, one explanation is that only GPCR-bound active GRKsare able to phosphorylate proteins located in close proximity to thesecomplexes. Two examples of this would be high-gain phosphorylationin the retina (Binder et al., 1990, 1996) and phosphorylation of IRS-1 byGRK2 inresponse to activationof endothelinreceptors (Usui et al., 2005).Alternatively, GRKs might use their low intrinsic basal activity to phos-phorylate non-GPCR substrates. It is also conceivable that at least someof the non-receptor substrates are able to activate GRKs in a manner sim-ilar to that of active GPCRs, albeit by distinct molecular mechanisms. Thisappears to be the case for tubulin (Carman et al., 1998; Pitcher et al.,1998) and synuclein, which are phosphorylated with higher catalytic ef-ciencies than peptide substrates (Pronin et al., 2000). GRK6A, but not itsother splice variants, phosphorylates the Na+/H+exchanger regulatoryfactor by forming an interaction between its C-terminus and the PDZ do-main of the factor (Hall et al., 1999). Given the diversity of GRK non-re-ceptor substrates that have already been described, the mechanism ofGRK activation might be unique for each substrate.GRKscanphosphorylateotherclassesof cell surfacereceptors.GRKs have been shown to phosphorylate the Smoothened (Smo), aseven transmembrane domain receptor belonging to a distantgroup within the GPCR superfamily. Smo is part of the evolutionary44 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069conserved hedgehog (Hh) signaling pathway. In the absence of Hh,Smo, thesignaltransducercomponentofthepathway, isinhibitedby the Hh receptor Patched (Ptc). Hh binding to Ptc frees Smo to trav-eltotheplasmamembraneinDrosophilaortociliainvertebrates,which it must do to signal. The main signaling pathway of Smo con-sists of several signaling events ultimately leading to the regulationof Ci (in Drosophila)/Gli (in vertebrates) family of transcription fac-tors (reviewed in Huangfu & Anderson, 2006). It remains controver-sial whether G proteins play any role in Smo signaling (reviewed inAyers&Thrond, 2010), althoughrecentdataindicatethatinDro-sophila Smo can function as canonical GPCR through coupling to Gi(Ogden et al., 2008). Recently, phosphorylation of Smo by GRKs hasbeen shown to beapart of the Smosignaltransductioncascade.Aknockdownof Gprk2, a Drosophilaorthologof mammalianGRKs4/5/6, diminishes Smo signaling and induces Hh loss-of-function phe-notype (Molnar et al., 2007; Cheng et al., 2010b). Gprk1, a Drosophilaorthologous of mammalian GRKs 2/3, has also been implicated in me-diating the Smo signaling, albeit less efciently (Cheng et al., 2010b).Table 1GRK substrates.Substrate protein GRK isoform Tissue/cells Function ReferencesSeven transmembrane domain but non-GPCR receptorsSmoothened (Smo) Gprk2 Drosophila in vivo GRK-mediated phosphorylation of Smo is a partof the Smo signaling cascadeMolnar et al., 2007;Chen et al., 2010a Gprk1Grk2, Grk3 Zerbrash in vivo Philipp et al., 2008GRK2 C3H10T1/2 cells,Shh-LIGHT cellsMeloni et al., 2006Non-GPCR receptorsLow density lipoprotein-relatedprotein 6 (LRP6)GRK5&6 GRK-mediated phosphorylation of LRP6 activatesLRP6 and mediate Wnt/LRP6 signalingChen et al., 2009Platelet-derived growth factorreceptor- (PDGFR)GRK2 HEK293 cells GRK2-mediated phosphorylation inducesdesensitization of PDGFRFreedman et al., 2002;Hildreth et al., 2004GRK5 Smooth muscle cells GRK5-mediated phosphorylation inducesdesensitization of PDGFRWu et al., 2006Other membrane proteinsEpithelial Na+ channel (ENaC) GRK2 Cultured salivary ductcellsGRK2-mediated phosphorylation renders ENaCinsensitive to inhibition by the ubiquitin ligase Nedd4-2Dinudom et al., 2004Downstream regulatory elementantagonist modulator (DREAM)GRK2&6 HEK293 cells GRK-mediated phosphorylation block DREAM-mediatedmembrane expression of Kv4.2 potassium channelRuiz-Gomez et al., 2007Transcription factorsIB GRK2&5 GRK-mediated phosphorylation enhances theTNF-induced NFB activityPatial et al., 2009NFB1 p105 GRK2 Raw264.7 macrophagecells, HEK293 cellsGRK2-mediated phosphorylation of NFB1 p105 reducesthe lipopolysaccharide-induced ERK1/2 activationParameswaran et al.,2006Receptor-regulated Smads,Smad2&Smad3GRK2 Human hepatocarcinomacellsGRK2-mediated Smad phosphorylation blocksactivin/TGF-induced Smad activation, nucleartranslocation, and target gene expressionHo et al., 2005Signaling proteinsArrestin 2 GRK5 GRK5-mediated phosphorylation prevents activationof Src by 5-HT4 receptorBarthet et al., 2009Nedd4, Nedd4-2 GRK2 Cultured salivary ductcellsunknown; possibly, interference with the binding toepithelial Na+channel, which is negatively regulatedby Nedd4 and Nedd4-2Sanchez-Perez et al.,2007Synucleins (, , , and synoretin) GRK2 (& isoforms)GRK5 ( isoforms)COS-1 cells, in vitro GRK-mediate phosphorylation inhibits synuclein'sinteraction with phospholipase D2 and phospholipidsPronin et al., 2000Phosducin GRK2 In vitro GRK2-mediated phosphorylation of phosducin reducesphosducin's binding to GRuiz-Gmez et al.,2000 Subunit of cyclic nucleotidemonophosphate phosphodiesterasetype 6 (PDE)GRK2 HEK293 cells GRK2-mediated phosphorylation of PDE enhancesthe epidermal growth factor- and trombin-dependentstimulation of ERK1/2Wan et al., 2001p38 GRK2 GRK2-mediated phosphorylation of p38 at Thr123impairs p38 activation by MKK6Peregrin et al., 2006Tumor suppressor gene productadenomatous polyposis coli (APC)GRK2 HEK293 GRK2 interacts with APC via its RGS domain, whichresults in phosphorylation of a component of the -catenindestruction complex and inhibition of the canonical WntsignalingWang et al., 2009Na+/H+exchanger regulatory factor(NHERF)GRK6A HEK293 GRK6A is responsible for constitutive phosphorylation ofNHERF at Ser289Hall et al., 1999Insulin receptor substrate-1 (IRS-1) GRK2 3T3-L1 adipocytes GRK2-mediated phosphorylation of promotes degradationof IRS-1 leading to reduced insulin signalingUsui et al., 2005Nuclear proteinsClass II histon deacetylase (HDAC) GRK5 Cultured cardiomyocytes,mouse heartGRK5-mediated phosphorylation of HDAC elevatedMEF2-mediated transcription and cardiac hypertrophyMartini et al., 2008Cytoskeletal proteinsRadixin GRK2 Increased Rac1 activity, membrane protrusion, and cell motility Kahsai et al., 2010Ezrin GRK2 HEK293 GRK2-mediated phosphorylation of ezrin links the GPCRactivation to the actin cytoskeleton remodelingCant & Pitcher, 2005Tubulin GRK2&5 COS-1 cells, in vitro GRK-mediated phosphorylation of tubulin regulatesmicrotubule assembly and links the GPCR activation to thecytoskeleton remodelingCarman et al., 1998;Haga et al., 1998;Pitcher et al., 1998Ribosomal proteinsRibosomal protein P2 GRK2 HEK293 GRK2-mediated phosphorylation of P2 enhances thetranslational activityFreeman et al., 200245 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069Gprk2activates Smobyphosphorylatingit at Ser741/742, andthisphosphorylation is regulated by phosphorylation of adjacent sites byPKA and casein kinase (Chen et al., 2010a). Knockdown of Grk2/3 inzebrash leads to defective Smo signaling and developmental abnor-malities (Philipp et al., 2008), and GRK2-dependent phosphorylationof Smo has been shown to promote Smo signaling in mammalian cells(Meloni et al., 2006) suggesting a role of GRK2/3-mediated Smo phos-phorylation in Smo signaling in vertebrates analogous to that in Dro-sophila. Interestingly, Gprk2-mediated Smo phosphorylation inDrosophila (Cheng et al., 2010b) and GRK2-dependent phosphoryla-tion in mammalian cells (Chen et al., 2004) induces arrestin recruit-ment, Smo internalization and down-regulation in a wayreminiscent of normal GPCRs. However, if incanonical GPCRs,GRK/arrestin-dependent internalization and down-regulationis apart of thedesensitizationprocess, inDrosophilaGprk2-mediatedSmo trafcking does not impair signaling (Cheng et al., 2010b). Formany GPCRs, GRK- andarrestin-dependent internalizationis therststepleadingtodegradationthatoccursincontinuingagonistpresence. Incontrast, Smoexpressionis lowwithout theagonist,andHhstabilizesSmo, whereasGprk2destabilizesit(Chengetal.,2010b). Therefore, the way GRKs function in the Hh pathway is quitedistinct from that in canonical GPCR signaling. Considering the impor-tanceof theHhsignalingpathwayindevelopmental disordersandmany forms of cancer (reviewed in Jiang & Hui, 2008), thesendingsimplicating GRKs in the regulation of Smo signaling warrant further ef-fort in elucidating the role of these kinases in the Hh pathway.GRK2 and5 phosphorylate platelet-derivedgrowthfactor receptor-(PDGFR) (Freedman et al., 2002). The GRK2-dependent serine phos-phorylation of PDGFR reduces Tyr autophosphorylation and allostericactivation of Gi (Freedman et al., 2002), as well as the association ofPDGFR with the Na+/H+exchanger regulatory factor (Hildreth et al.,2004). In mouse aortic smooth muscle cells, it is GRK5 and not GRK2that phosphorylates PDGFR, thus desensitizing PDGFR-induced phos-phoinositide hydrolysis but enhancing Src activation (Wu et al., 2006).So far, however, PDGFR remains the only non-GPCR cell surface recep-tor shown to be desensitized in a GRK-dependent manner. EGFreceptor is phosphorylated by GRK2 but no detectable functional conse-quences of this phosphorylationaredocumented(Freedmanet al.,2002) (Hildreth et al., 2004).GRKs have also been reported to phosphorylate a diverse collectionof receptor-associated proteins, changing their binding to the receptorand/or their activity, which might lead to desensitization of G protein-dependent or independent signaling pathways. In most known cases,neither the mode of GRKactivation nor the functional role of GRK-medi-ated phosphorylation is well understood. It is well established that acti-vated GPCRs can enhance GRK-mediated phosphorylation of exogenouspeptides up to 100-fold (Palczewski et al., 1991; Chen et al., 1993). Thus,GRKs should be able to phosphorylate other substrates under the samecircumstances in cells. One example is arrestin-2. Serotonin 5-HT4 re-ceptors activate cAMP-dependent signaling and MAP kinase signalingvia Gs and Src, respectively, which associate with the receptor complex(Barthetetal., 2007). Upon5-HT4receptorstimulation, GRK5phos-phorylates a cluster of serinethreonine residues in the receptor C-tailand thereby promotes arrestin-2 binding. GRK5 then directly phosphor-ylates arrestin-2, which prevents activation of Src (Barthet et al., 2009).Although it is not a receptor that is phosphorylated, this clearly repre-sents a variation of desensitization. GRK2-dependent phosphorylationof p38 might also be happening largely inside signaling complexes as-sembled on active receptors, because the association of GRK2 with p38is agonist-dependent (Peregrinet al., 2006). There is at least one alterna-tive mechanism that would make GRK-dependent phosphorylation ofother proteins largely contingent on GPCR signaling. Receptor activationleads to the recruitment of GRKs and arrestins to GPCR-rich membranes(Daaka et al., 1997; Oakley et al., 2000), which greatly increases theirlocal concentration, thereby making an encounter between a GRK andany receptor-associated or membrane-enriched substrate more likely.GRKs are also able to phosphorylate and regulate nuclear proteins,such as class II histone deacetylases (HDACs) (Martini et al., 2008) andmultiple transcription factors. In the case of HDAC, GRK5, which has a nu-clear localization signal, is implicated (Johnson et al., 2004). GRK2, whichlacks recognizable nuclear localizationsignal and is usually excludedfromthe nucleus in most cells was strongly implicated in the phosphorylationof transcription factors (Ho et al., 2005). In most known cases, phosphor-ylation occurs in response to stimulation of specic membrane receptors,and thus it appears that transcription factors would need to be phosphor-ylated in the cytosol and travel to the nucleus afterwards. Indeed, GRK-mediated phosphorylation affects nuclear translocation of transcriptionfactors as well as their activation(Ho et al., 2005; Patial et al., 2010). Phos-phorylationof IBandnuclear translocationof NFBp65following stim-ulationofToll-likereceptor-4bylipopolysaccharidewereimpairedinprimary macrophages derived from GRK5 knockout mice, along with re-duced inammatory response to lipopolysaccharide (Patial et al., 2010).These data establish GRK5 as a positive regulator of the Toll-receptor-4-linked NFB pathway. Conversely, GRK2 negatively regulates the trans-forminggrowthfactor(TNF)/activinsignalingpathway(Hoetal.,2005). Prolongedexposure to activinupregulates GRK2, whichphosphor-ylates Smad2 and 3 preventing their phosphorylation by type I TNF re-ceptor. The phosphorylation of these receptor-regulated Smads atcarboxy-terminal serine residues by type I receptor kinase is necessaryfor their oligomerization and nuclear translocation (Moustakas & Heldin,2009). As a result, GRK-dependent phosphorylationof Smads antagonizesTNF-mediatedtargetgeneexpressionandcellulareffects(Hoetal.,2005). Puried GRK2 phosphorylates GST-Smad2 and 3, but in cells theformationof GRK2-Smadcomplexwasactivin-dependent (Hoet al.,2005). In other examples, GRK2 and 5 mediate tumor necrosis factor (TNF)induced activation of the NFkB transcription factor in mouse mi-crophages via phosphorylation of IkB (Patial et al., 2009). Conversely,GRK5-mediated phosphorylation of p105 stabilizes p105 and negativelyregulates ERK1/2 activation via toll-like receptor 4 (Parameswaran et al.,2006).Cytoskeletal targets of GRKs are also evident. Tubulin was therstnon-GPCR substrate of GRKs described in 1998 (Carman et al., 1998;Haga et al., 1998; Pitcher et al., 1998). GRK2 and 5 phosphorylate tubu-lin, with the preference for -tubulin and, under certain conditions, forthe III-isotype (Carman et al., 1998). Thending that tubulin phos-phorylation by GRK2 is enhanced by the addition of G and phospho-lipids (Carman et al., 1998; Haga et al., 1998), both of which promoteGRK2 association with the plasma membrane, and by agonist-boundGPCRs (Haga et al., 1998; Pitcher et al., 1998) suggest once again thatincells GRK2 activated by GPCRs couldbe largely responsible for tubulinphosphorylation, which would localize this process to the vicinity of re-ceptor-rich membranes. GRK2 also phosphorylates two other cytoskel-etal proteins, ezrin and radixin, both members of ezrinradixinmoesin(ERM) protein family (Fehon et al., 2010). GRK2 phosphorylates ezrin ata single Thr567 residue (Cant & Pitcher, 2005), which is important formaintainingezrininanactiveconformationwithboththeplasmamembrane and F-actin binding domains accessible (Fehon et al.,2010). Similarly, GRK2 phosphorylates radixin at critical Thr564 residue(Fehon et al., 2010). GRK2 phosphorylation of ezrin is PIP2- and G-dependent, suggesting that it is performed by the membrane-associat-ed, possiblyreceptor-activatedkinase. Furthermore, GRK2-mediatedphosphorylationofezrinisrequiredformuscarinicM1receptor-in-ducedrufeformationinHep2cells(Cant &Pitcher, 2005), againstrongly suggesting the involvement of GPCR in the GRK2 activation.In contrast to ezrin, GRK2 phosphorylation of radixin was unaffectedby PIP2 and G (Kahsai et al., 2010). Based on this limited data, it isreasonabletoconcludethatGRKsmighttransmitGPCRsignalingtothe cytoskeleton. Synucleins are another interesting class of proteinsshown to be phosphorylated by GRKs. GRK2 and 5 phosphorylate synu-cleins very efciently, with- and -synucleinbeing the best substratesfor GRK2 and -synuclein for GRK5 (Pronin et al., 2000). Lipids stimu-latedsynucleinphosphorylationbybothGRKsandlipidsplusG46 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069strongly enhanced GRK2-dependent synuclein phosphorylation, whichsuggests that GPCRs might regulate synuclein phosphorylation by GRKsincells. Althoughthefunctionof synucleinsremainsunclear, theyare of great medical interest because -synuclein is therst gene tobe implicated as a cause of a rare familial form of Parkinson's disease(Polymeropoulos et al., 1997).6. Proteins regulated by G protein-coupledreceptor kinases in phosphorylation-independent mannerGRKs have been reported to regulate several signaling proteins via di-rect interaction that does not require kinase activity (Table 2). It is notunusual for enzymes to performscaffolding functions inadditionto or in-stead of their enzymatic activity. GRKs, particularly GRK2 and 3, are fairlylarge multidomain proteins, and it is conceivable that they could interactwith a multitude of proteins via different domains (Pronin et al., 1997;Lodowski et al., 2003; Tesmer et al., 2005). As the number of proteinsGRKs scaffold proliferate (Ribas et al., 2007) (as happened before witharrestins; Gurevich & Gurevich, 2006a), it becomes particularly impor-tant to verify these interactions in physiological settings, elucidate howall these multiple interactions are organized in cells, whether GRKs scaf-fold proteins complexes in a receptor-dependent manner or are able todo so independently of receptor stimulation, and most importantly, todetermine the biological roles of these interactions.6.1. Receptor desensitization via the R domainIdenticationof a regionwithinGRKN-terminus witha homology toRGS proteins (the RH domain) by in silico methods (Siderovski et al.,1996) led to discovery of a new way for GRK to suppress G protein-de-pendent signaling. Subsequent studies showed that the RH domain ofGRK2 and 3 binds and sequesters active Gqsubunit (Table 2) (Carmanet al., 1999; Usui et al., 2000; Tesmer et al., 2005). Canonical RGS pro-teins act as GTPase activating proteins (GAP) promoting GTPase activityof GTP-liganded -subunit of a G protein and thus attenuating G pro-tein-dependent signaling (reviewed in Hollinger & Hepler, 2002). RHdomains of GRKs, on the other hand, possess weak or no GAP activity,and their ability to attenuate signaling is primarily due to sequesteringactivated Gq/11(Carman et al., 1999) or by direct blockade of the GPCR(Dhami et al., 2002). Another difference between RGS proteins and RHdomains of GRKs is that most RGS proteins interact with Gi/o as wellaswithGq, withvaryingselectivityamongRGSfamilies(Ross&Wilkie, 2000; Hollinger & Hepler, 2002), whereas GRK RH is selectivefor Gq (Carman et al., 1999; Sallese et al., 2000a) and is even able todiscriminateamongdifferentmembersofGqfamily, bindingG11and G14, but not G16 (Day et al., 2003). Although all GRKs possessRH domain, in members of the GRK1 or GRK4 subfamilies it does notseem to be able to bind any G (Carman et al., 1999; Picascia et al.,2004). Mutagenesis studies identied eight residues in the N-terminalregion of GRK2 important for the Gq binding, six of which are con-served in GRK3 but not in other members of GRK family (Sterne-Marret al., 2003), which likely explains why only GRK2 and 3 bind Gq.RH-mediated dampening of signaling via Gq-coupled GPCRs is an im-portant regulatory mechanism, acting in concert with phosphorylation-dependent receptor desensitizationinprimary cells (Willets et al.,2004), oronendogenousreceptorsignalingincertainculturedcells(Luo et al., 2008). GRKs acting via RH can also interfere with GPCR-inde-pendentGq-dependentprocesses. Inonesuchcase, insulinstimulatesglucose transport acting via insulin receptor-dependent phosphorylationof Gq/11(Imamura et al., 1999). GRK2 acting as a scavenger of Gq/11in-hibitsinsulin-inducedglucosetransport(Usui etal., 2004). Similarly,GRK2 mediates an inhibitory effect of chronic treatment with endothe-lin-1 on glucose transport by interfering with the Gq/11-dependent sig-nalinginaphosphorylation-independentmannerviaitsRHdomain.Table 2Phosphorylation-independent regulation by GRKs.Binding protein GRKisoformTissue/cells Function ReferencesGPCR receptors via RGS-like domainmGluR5 GRK2 Striatal culturedneuronsAttenuates mGluR5 signaling Ribeiro et al., 2009M3 muscarinic receptor GRK2 HEK293(endogenousreceptors)Attenuates carbachol-induced Ca2+mobilization Luo et al., 2008Transcription factorsIB GRK5 Bovine aortaendothelial culturedcells,GRK5 interaction with IB viaRGS-like domain inducesnuclear accumulation of IB and inhibition of NFB activitySorriento et al., 2008Signaling proteinsPTCH1 GRK2 Zebrash Interaction of GRK2 with PTCH1 disrupts PTCH1-mediateinhibition of cyclin1 nuclear translocation and promotescell proliferationJiang et al., 2009(cyclin B1 regulator pathched homolog 1)PI3K GRK2 HEK293 GRK2-mediated recruitment of PI3K to the plasma membraneand promotes receptor internalizationNada Prasad et al.,2001; Naga Prasad etal., 2002(phosphoinositide-3-kinase)GIT1 GRK2 HEK293 GRK2 interaction with GIT1 may recruit GIT1 to the membrane;GIT1 recruitment slows down receptor internalization andresensitizationPremont et al., 1998(G protein-coupled receptorkinase-interacting protein, a GTPase-activatingprotein for the ADP ribosylation factor (ARF))GRK2 HEK293, COS7, HeLa,GRK2+/+ orGRK2+/ MEFsGRK2 interaction with GIT1 facilitates GIT1-dependent ERK1/2activation and focal adhesion turnover; important for theGRK2-dependent regulation of cell motilityPenela et al., 2008MEK GRK2 HEK293 GRK2 interaction with reduces the chemokine-inducedMAPK activationJimnez-Sainz et al.,2006Clathrin GRK2 HEK293 GRK2 interaction with clathrin promotes GPCR internalization Shiina et al., 2001;Mangmool et al., 2006Akt GRK2 Sinusoidalendothelial cells,in vivoAkt interaction with GRK2 inhibits the Akt activity Liu et al., 2005Epac (exchange protein activated by cAMP) GRK2 HEK293 Epac interaction with GRK2 inhibits activation of theEpac-Rap1 cascadeEijkelkamp et al., 2010b47 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069Additionally, GRK2, in response to endothelin-1 treatment, phosphory-lates insulin receptor substrate (IRS)-1 promoting its degradation and re-ducing insulin-dependent signaling (Usui et al., 2005). Therefore, bothmodes of GRK2 action contribute to endothelin-1-induced insulinresistance.The GRK1 and GRK4 subfamilies also have structurally similar RHdomains (Lodowski et al., 2003,2006), which have not been reportedtobindanyGprotein-subunits(Carmanetal., 1999;Dayetal.,2004; Sterne-Marr et al., 2004).However, these domains are foundinall GRKskinases, suggestingthattheyhavebiological functionsthat remain to be elucidated. One possibility is to modulate the con-formation of the GRK kinase domain, as the RH domain bridges the ki-nasesmall andlargelobes, therebycontributingtothelowbasalactivity of the kinase (Lodowski et al., 2003). Indeed, in the relativelyclosed GRK6 structure the contacts between the RH domain and thelarge lobe are severed (Boguth et al., 2010). A second global functionmay be to x the C-terminus of the C-tail of the kinase domain so thatthe so-called hydrophobic motif, common to all AGC kinases (Pearceet al., 2010), remains docked to the small lobe. As a result, an addi-tionalphosphorylationeventisnotneededtoactivateGRKsinre-sponse to the activation of GPCRs.6.2. Regulation via other mechanismsGRK2and3alsohavetheGbindingsiteintheirC-terminus(Kochetal., 1993; Touharaetal., 1994; Lodowski etal., 2003). Ithaslongbeenconsideredapartofthemechanismresponsiblefortransient recruitment of these kinases to the membrane upon GPCRTable 3Proteins regulating the GRK functions.Protein GRKisoformTissue/Cells Function ReferencesVia phosphorylationPKA GRK2 HEK293 PKA-mediated phosphorylation of GRK2 at Ser685increases G binding to GRK2, GRK2 translocationto the membrane and receptor phosphorylationCong et al., 2001GRK1,GRK7HEK293, in vitro PKA-mediated phosphorylation of GRKs 1 and 7reduced their ability to phosphorylate rhodopsin in vitroHorner et al., 2005GRK1,GRK7Mouse rod Photoreceptors,Xenopus laevis conesPKA-mediated phosphorylation of GRKs 1 and 7 issignicantly elevated in the dark-adapted retinas andis sharply decreased upon exposure to lightOsawa et al., 2008; Osawa et al.,2011PKC GRK5 COS-1 PKC-mediated phosphorylation of the GRK5 C-terminusreduces its catalytic activity towards both receptors andnon-receptorPronin and Benovic, 1997GRK2 CHO, HEK293 PKC-mediated phosphorylation of the GRK2 C-terminusincreased its catalytic activity towards the membrane-boundreceptor but soluble substratesWinstel et al., 1996GRK2 HEK293, COS-1 PKC-dependent phosphorylation of GRK2 at Ser29 relievescalmodulin-induced inhibition of the GRK2 activityKrasel et al., 2001ERK1/2 GRK2 HEK293 ERK1/2-mediated phosphorylation of GRK2 at Ser670reduces G binding and GRK2 activityElorza et al., 2000Platelet-derived growthfactor receptor- (PDGFR)GRK2 HEK293 PDGFR phosphorylates and activates GRK2 to enableGRK2 to phosphorylate the receptorWu et al., 2005GRK5 smooth muscle cells PDGFR phosphorylates and activates GRK5 Cai et al., 2009c-Src GRK2 COS-1 cells, in vitro agonist-dependent tyrosine phosphorylation (Tyr13,86,92)of GRK2 by c-Src increases the GRK2 kinase activitySarnago et al., 1999GRK2 HEK293 cells GRK2 phosphorylation by c-Src promotes GRK2 degradation Penela et al., 2001GRK2 HEK293 cells GRK2 phosphorylation by c-Src increases the GRK2interaction with GqMariggi et al., 2006CDK2-Cyclin A GRK2 HEK293, HeLa CDK2-mediated phosphorylation of GRK2 at Ser670 is requiredfor GRK2 downregulaiton in G2 phase and cell cycle progressionPenela et al., 2010Via interactionCaveolin-1&3 GRK2 A431, NIH-3T3, COS-1 cells GRK-caveolin interaction inhibits GRK-mediatedreceptor phosphorylationCarman et al., 1999GRK1&5 In vitroCaveolin-1 GRK4 Cultured human renalproximal tubule cellsGRK4-caveolin-1 interaction inhibits GRK4-mediatedphosphorylation of the D1 dopamine receptorGildea et al., 2009Calmodulin GRK5NGRK6NNGRK2NNCOS-1 cells, in vitro GRK interaction with calmodulin inhibits GRK activity viareduced ability of the kinase to bind to receptors and lipids;calmodulin activates GRK5 autophosphorylation, which inhibitsChuang et al., 1996; Haga et al.,1997; Pronin et al., 1997GRK1 GRK5 interaction with receptorsRecoverin GRK1 In vitro recoverin interaction inhibits the GRK1 activity Chen et al., 1995; Klenchin et al.,1995Actin GRK5 In vitro Actin inhibits GRK5 activity Freeman et al., 1998Heat shock protein 90 (HSP90) GRK2 HL60 cells, COS-1 cells Interaction with Hsp90 regulates GRK2 maturaiton Luo and Benovic, 2003Raf kinase inhibitorprotein (RKIP)GRK2 HEK293, in vitro RKIP interaction with GRK2 blocks the kinase activity Lorenz et al., 2003Nitric oxide synthase (NOS);S-nitrosothiols (SNOs)GRK2 HEK293 cells, U2-osteosarcoma cells,S-nitrosylation of GRK2 by NOS or SNOs inhibits thekinase activityWhalen et al., 2007in vivoUbiquitin ligase Mdm2 GRK2 HEK293 cells, MCF-7 cells,MEFsMdm2-mediated ubiquitination of GRK2 promotesGRK2 degradationSalcedo et al., 2006Pin1 GRK2 HEK293, HeLa Binding of Pin1 to CDK2-phosphorylated GRK2 promotesGRK2 degradationPenela et al., 201048 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069activation (Haga & Haga, 1992; Pitcher et al., 1992; Li et al., 2003). Re-cently, a different function of the G-binding pleckstrin homologydomain came to light, which is quite similar to the function of RH do-main: desensitization of GPCR signaling by sequestering a G proteinsubunit, inthis caseG. Gprotein-coupledpotassiumchannels(GIRK) are activated by Pertussis toxin-sensitive Gi/o-coupled GPCRsvia G subunits (for review see Sadjaa et al., 2003). GRK2, and likelyGRK3, competitively binds G with high enough afnity to keep itaway fromGIRK, thereby reducing channel activationby GPCRs(Raveh et al., 2010). This is a novel mechanism of GPCR inactivationbyGRKat thelevel of theeffector, andit might beapplicabletoother G-dependent processes in cells (Raveh et al., 2010).The discovery of GRK interacting proteins (GIT) GIT1 and GIT2 wasthe result of a search for proteins that might bind GRKs and serve aseffectors (Premont et al., 1998) mediatingGPCRsignaling. GITswere therst proteins identied that bind GRKs without serving assubstrates. GITs are multidomain scaffolding proteins that bindmany partners, including ARFs, other small GTPases, and various ki-nases (Hoefen & Berk,2006).The main function of GITs appears tobe modulation of cytoskeletal dynamics during cell attachment andmigration. It is signicant that the rst proteinfoundtointeractwithGITs (after GRKs) was paxillin, animportant component offocal adhesions (Turner et al., 1999). GITs bind another major compo-nent of focal adhesions, PIX, recruiting active Rac to its downstreameffector PAK via PAK binding to PIX in GIT-PIX oligomers (Hoefen &Berk, 2006). Additional functions of GITs include their role in receptorinternalization and membrane trafcking and scaffolding of signalingcascades (Hoefen & Berk, 2006). The ability of GRKs to interact withGITs might give GRKs asay in all these functions, including cyto-skeletal remodeling. Recently it has been shown that GRK2 promotesintegrin-dependent migration of epithelial cells towardsbronectinvia interaction with GIT1 (Penela et al., 2008). The migration of epi-thelial cells was promoted by a lipid messenger sphingosine-1-phosphate (S1P) acting via Gi-coupled S1P1/3 receptors, and the effectof GRK2 was dependent on the function of these receptors. The phys-iological importance of the GRK2-dependent coordination of integrin-andGPCR-directedcell migrationis underscoredby the delayedwound healing in Grk2 hemizygous mice (Penela et al., 2008). Inter-estingly, GRK2 acted as a positive modulator of the S1P1/3 receptor-dependentsignalingviatheMEK/ERKpathway, suggestingthatinthis case it does act as an effector actually mediating rather than desen-sitizing the GPCR signaling. Importantly, GRK2 serves as an effector notinitskinasecapacity, but asascaffoldingproteinrecruitingmajorplayers in regulating cell migration to the focal adhesion points.GRKshavebeenshowntoscaffoldothersignalingproteinswithfunctional consequences. GRK2 negatively regulates ERK activation bychemokine receptor CCR2B via binding to MEK or MEK-containing sig-naling complexes (Jimnez-Sainz et al., 2006; Luo et al., 2008). GRK2phosphorylates CCR2B upon activation and desensitizes chemokine-in-ducedcalciumsignaling (Aragay etal., 1998). However,the effectofGRK2 on ERKdid not require kinase activity or Gqbinding via the RHdo-main. Similareffectsof GRK2onchemokine-inducedERKactivationwere observed with heterologously expressed chemokine CXCR4 recep-tor andin splenocytes, which express endogenous CXCR4 receptor, fromGRK2 hemizygous mice (Jimnez-Sainz et al., 2006). GRK2 can directlybind and inhibit another signaling molecule, a serinethreonine kinaseAkt. GRK2boundbothactiveandinactiveAkt, andthebindingsitewas localized to the C-terminus. GRK2 binding to Akt reduced Akt activ-ity and its ability to phosphorylate and activate nitric oxide synthase insinusoidal endothelial liver cells (Liu et al., 2005). Nitric oxide synthaseis critical to maintain vascular homeostasis, and a defect in its activityseen in injured sinusoidal endothelial cells leads to intra-hepatic portalhypertension(Rockey&Chung, 1998; Shahet al., 1999; Yuet al.,2000). The expression of GRK2 in injured sinusoidal endothelial cells in-creased proportionally to the severity of the injury, and GRK2 knock-down increased phosphorylation of Akt, the Akt-mediated nitric oxideproduction, andamelioratedportal hypertension(Liuet al., 2005).GRK2 was reported to directly interact with phosphoinosidite-3-kinase(PI3K) promoting its membrane localization, phosphoinositide produc-tion, AP-2 recruitment to the receptor, and receptor endocytosis (NadaPrasad et al., 2001; Naga Prasad et al., 2002). Thus, GRK2-mediated re-cruitment of PI3K to agonist-stimulated -adrenoreceptors that facili-tates receptor endocytosis is a contributing factor in the heart failure(Perrino et al., 2005a,b). PI3K is an upstream kinase in the Akt pathway(Manning &Cantley, 2007). Phosphoinositides produced by PI3K recruitphosphoinositide-dependentkinase1(PDK1)andAkt tothemem-brane, and PDK1 phosphorylates Akt at the regulatory residue Thr308(Chan et al., 1999; Manning & Cantley, 2007). It would be of interest todetermine whether the association of GRK2 with a PI3K signaling com-plex (Nada Prasad et al., 2001; Naga Prasad et al., 2002) affects PI3K ac-tivity and the downstream events such as Akt activity. GRK2-mediatedrecruitment of PI3K to the membrane should be expected to activatethe Akt pathway, whereas direct GRK2Akt interaction inhibits the Aktactivity. It is unclear how this complex interplay of GRK2-mediated sig-naling events inuences the control of physiological functions by thePI3K-Akt signaling pathway.7. Regulation of G protein-coupled receptor kinasesThe expression level, as well as activity of most enzymes in the cellis tightly regulated. GRKs are no exception. As described above, thebest-known mechanism of GRK regulation is via direct binding to ac-tive GPCRs. However, this is just one of several established regulatorymechanisms.7.1. Regulation of G protein-coupled receptor kinase activityLike many kinases, GRKs are regulated by phosphorylation and pro-teinprotein interactions (Table 3). G subunits stimulate GRK2 and 3viatheirPHdomains1012-foldinvitro(Haga&Haga, 1992; Kimet al., 1993), likely due to enhanced recruitment of kinases to active re-ceptors (Pitcher et al., 1992; Kim et al., 1993; Pitcher et al., 1995), al-thoughit isclearthat G-bindingalsoaffectstheconformationoftheGRK2kinasedomainallosterically(Lodowski etal., 2005). GRK2and 3 are also stimulated by anionic phospholipids such as phosphoino-sidides(Debburmanetal., 1995b;Onoratoetal., 1995;Pitcheretal.,1995; DebBurman et al., 1996). In fact, the ability of G to stimulateGRK2 and recruit it to the membrane depends on the presence of anionicphospholipids (DebBurman et al., 1996; Pitcher et al., 1996). There iseven a certain specicity of GRK interaction with G subunit: GRK2binds G containing G1 or G2 but not G3, whereas GRK3 binds allthreeisoformsequallywell(Daakaetal., 1997). Incontrast, GRK5isnot activated by G subunits (Kunapuli & Benovic, 1993) but it is acti-vatedbybindingtophospholipidssuchasphosphatidylinositol 4, 5-bisphosphate(PIP2), asareothermembersofthesubfamily, GRKs4and 6 (Pitcher et al., 1996). However, in contrast to the GRKs 2 and 3,whichbindPIP2andotherlipidsviatheirPHdomain, themembersof the GRK4 subfamily seem to interact with PIP2 via the N-terminus(Pitcher et al., 1995, 1996), presumably at the basic region close to theN-terminal helix in the GRK6 structure (Boguth et al., 2010).GRK1undergoes intramolecular autophosphorylationat Ser488and Thr489(Buczyko et al., 1991; Palczewski et al., 1992). Autopho-sphorylation reduces the GRK1 activity towards phosphorylatedlight activated rhodopsin (Buczyko et al., 1991; Pulvermller et al.,1993; Palczewski et al., 1995) thereby serving as a negative feedbackmechanismlimitingthe rateandextentofrhodopsin phosphoryla-tion. GRK1 (at Ser21) and GRK7 (at Ser23and Ser36) are phosphorylat-ed by PKA in vitro and in cultured cells, which reduces their activitytowards rhodopsin (Horner et al., 2005). GRK7 was found phosphor-ylatedat the conservedPKAphosphorylationsiteSer36indark-adapted cone photoreceptors in various species (Osawa et al.,2008). Using the Xenopus laevis retina as a model, it was shown that49 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069thelevelofthePKA-mediatedphosphorylationofgrk7wasdimin-ishedbylightexposure (Osawaetal., 2008). Similarly, thelevelofGRK1 phosphorylation was higher in the dark-adapted mouse retinaas comparedto the light-adaptedanimals (Osawa et al., 2011).These data suggest that PKA phosphorylation of GRKs 1 and 7 serveas a feedback mechanism: dephosphorylation increases kinase activ-ity upon light exposure when rapid opsin deactivation is required. In-terestingly, phototransduction does not seem to be required for thelight-induceddephosphorylationof GRK1, sinceit ispreservedinmice lacking transducin -subunit (Osawa et al., 2011).GRK5isregulatedbyarapidphospholipid-stimulatedautopho-sphorylation in its C-tail region at Ser484and Thr485, which enhancesits ability to phosphorylate receptors (Kunapuli et al., 1994a). This isanalogous to what happens at the so-called turn motif in other AGCkinases (Pearce et al., 2010). GRK5 is also phosphorylated in the C-ter-minusandinhibitedbyproteinkinaseC(PKC)(Pronin&Benovic,1997). In contrast, GRK2 phosphorylated by PKC has higher activity to-wards rhodopsin (Chuang et al., 1995; Winstel et al., 1996). GRK2 is alsophosphorylated by protein kinase A at Ser685, which promotes its inter-action with G and membrane recruitment (Cong et al., 2001). Simi-larly, GRK2 phosphorylationby cSrc, which ispromotedby receptorstimulation, enhances the GRK2 activity towards receptors as well asnon-receptor soluble substrates (Sarnago et al., 1999) and its ability tointeract with Gq (Mariggi et al., 2006). Thus, GRK2 phosphorylationby cSrc acts as a negative feedback augmenting receptor desensitizationand reducing signaling. Conversely, agonist-induced GRK2 phosphory-lation by ERK1/2 reduces the GRK2 activation by G and the abilitytophosphorylatereceptors(Pitcheretal., 1999;Elorzaetal., 2000)functioning as a positive feedback loop enhancing G protein-mediatedsignaling. On the other hand, arrestin-mediated ERK activation, whichis strongly enhanced by arrestin recruitment to active GPCRs (Luttrellet al., 2001), would be dampened. GRKs 2 and 5 are known to phos-phorylate and desensitize a tyrosine kinase receptor PDGFR (Freemanet al., 2002; Hildrethet al., 2004; Wuet al., 2006). Inturn, PDGFRphos-phorylates and activates GRK2 in PDGF-dependent manner, which thenphosphorylates and deactivates the receptor (Wu et al., 2005). A similararrangement exists in case of GRK5 (Cai et al., 2009). The data on GRKregulation by phosphorylation appear to be incomplete: GRK2 and 5are phosphorylated by multiple kinases, whereas no phosphorylationof GRK3, 4, and 6 is reported. In most studies, GRK2 and 5 served as rep-resentatives of their respective subfamilies, and there is noreason to ex-pect that they are the only isoforms regulated via phosphorylation.Calmodulin, caveolin-1, actin, and several less ubiquitous proteinswere shown to affect GRK activity via direct binding. Mechanistically,the best understood case is inhibition of GRK1 and GRK7 by calcium-loaded recoverin, which directly competes with opsins for the kinaseN-terminus (Chenet al., 1995; Klenchinet al., 1995; Ames et al.,2006). An alternative mechanism of recoverin inhibition of GRK1 hasbeen proposed involving a blockade by recoverin of a conformationalchange in GRK1 induced by active rhodopsin rather than direct compe-tition for the binding site (Komolov et al., 2009). However, the concen-tration of Ca2+in light-exposed rod outer segments seems insufcientto support complex formation, and recoverin actually translocates outof ROS in the light (Strissel et al., 2005), whichcasts doubts on the phys-iological relevance of this interaction. Although the phenotype of reco-verinknockout miceisconsistent withitsroleinCa2+-dependentinhibitionofGRK1(Makinoetal., 2004;Sampathetal., 2005;Chenet al., 2010b), alternative explanations cannot be excluded. Inhibitionof GRK7 by recoverin is much less studied than that of GRK1. A recentstudy has demonstrated that in the carp retina a non-mammalian reco-verinhomolog expressedincones,visinin, provides amuch broaderrangeof Ca2+-dependent regulationof theGRK7activitythantherange of the GRK1 regulation in rods (Arinobu et al., 2010) suggestingthat recoverinGRK7 interaction might be particularly functionally sig-nicant. Calmodulin binds to all GRK isoforms, but has a preference forGRK5(Proninet al., 1997). Calmodulinbindinginhibits theGRK5activity via reductionof its binding to lipids and stimulationof its autop-hosphorylation, which inhibits GRK5 interaction with the receptor. Cal-modulin-dependent GRK5 autophosphorylation occurs at sitesdistinct from the two major residues Ser484and Thr485for phospholip-id-stimulated autophosphorylation, which, in contrast to calmodulin-dependent autophosphorylation, activates the kinase (Kunapuli et al.,1994a;Premont etal., 1994). CalmodulinbindstoastretchofbasicandhydrophobicresiduesintheN-terminal domain(Proninet al.,1997), which is highly conserved in the subfamily suggesting that cal-modulin might similarly regulate GRKs 4 and 6.GRK2, inaddition to being regulated by phosphorylation by multiplekinases, is S-nitrosylated at Cys340by S-nitrosothiols and nitric oxidesynthase following activation by multiple GPCRs (Whalen et al.,2007). S-nitrosylation inhibits the GRK2-mediated phosphorylation ofAR, recruitment of arrestins to the receptors, and receptor internaliza-tion and downregulation. This is a novel mode of regulation of the re-ceptordesensitization, whichmayalsobeapplicabletoGRK3, sinceCys340is conserved in that isoform.7.2. Regulation of expression of G protein-coupled receptor kinasesThe expression level of GRKs is known to be regulated by various fac-tors and to be altered in pathological conditions. Chronic or even acuteadministration of GPCR agonists can increase the level of GRKs in thebrain, which may lead to tolerance to drugs (Hurle, 2001; Diaz et al.,2002; Fan et al., 2002; Rubino et al., 2006; Schroeder et al., 2009). Ad-ministrationofGPCRantagonistsorremovalofendogenousagonistsalso can affect the GRK concentration (Hurle, 2001; Diaz et al., 2002;Bezard et al., 2005; Ahmed et al., 2007, 2008, 2010). The concentrationof GRKs in vivo and in cultured cells is responsive to various conditionssuch as stress (Taneja et al., in press), neonatal ventral hippocampal le-sion (Bychkov et al., 2010), cell cycle progression (Penela et al., 2010),and drug treatment (Salim & Eikenburg, 2007). Human diseases alterGRK levels. The best-known case is the upregulation of GRK2 in the fail-ing heart (Ungerer et al., 1993, 1994), but other pathological conditionsalso affect GRK expression (Garca-Sevilla et al., 1999; Grange-Midroitet al., 2003; Suo et al., 2004; Bychkov et al., 2008, 2011). In some cases,transcriptional regulation is involved, whereas in others alterations intheproteinconcentrationarenot accompaniedbychanges inthemRNA levels, suggesting the regulation at posttranscriptional levels.Littleisknownabout theregulationof GRKtranscription. Epi-nephrineactingsimultaneouslyvia2-and2ARsupregulatesthelevel of GRK3 mRNA and protein in neuronal cell lines (Salim et al.,2007). Theactionof epinephrineismediatedbytheactivationofERK1/2 and possibly transcription factors Sp-1 and Ap-2. Interactionof GRK2 and 3 with the heat shock protein Hsp90 is critical for themaintenanceof stablekinaselevelsincells(Luo&Benovic, 2003;Salim&Eikenburg, 2007). GRK2Hsp90interactionalsoseemstoplay a role in the proper folding and maturation of the newly synthe-sizedGRK2. Intheneuronal cell lineBE(2)-C, interactionof GRK3with Hsp90 aids in the kinase folding as well as increases its stabilityby suppressing proteosomal degradation (Salim & Eikenburg, 2007).The interaction with Hsp90 seems to be critical for folding and matu-ration of GRK5 and 6 as well (Luo & Benovic, 2003).Relative abundance of mRNAs and proteins suggests that the half-lives of GRKs are widely different. GRK2 has the highest mRNA:proteinratio, suggesting that it is short-lived, followed by GRK3 (Gurevichet al.,2004; Bezard et al., 2005; Ahmed et al., 2007). Indeed, the half-life ofGRK2 has been estimated at 60 min in HEK293, COS-7, Jurkat, and C6glioma cells due to rapid degradation (Penela et al., 1998, 2001). Inter-estingly, inHL60 cells the GRK2half-life was estimated at ~20 h and wasreduced to 2 h by geldanamycin, an inhibitor of GRK2 interaction withHsp90 (Luo & Benovic, 2003). Thus, the half-life of the same GRK iso-formandpresumablythemechanismsof itsregulationindifferentcell types vary widely.50 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 4069The regulation of GRK2 degradation is the most extensively stud-ied, but little is known about other GRK subtypes. GRK2 is rapidly de-graded via the proteasome pathway. Its degradation is facilitated byGPCRactivation(Penelaet al., 1998)andrequireskinaseactivity(Penela et al., 2001). GRK2 is phosphorylated by cSrc in vitro and incells (Sarnago et al., 1999). GRK2 phosphorylation by cSrc is facilitat-ed by receptor activation (Sarnago et al., 1999) and promotes GRK2degradation(Penelaet al., 2001). cSrcphosphorylationsiteshavebeen localized to Tyr13, 86, and 92, and the Y13/86/92F GRK2 mutantis resistant to both cSrc-mediated phosphorylation and degradation(Penelaetal., 2001). GRK2isalsophosphorylatedbyERK(Pitcheret al., 1999; Elorza et al., 2000), and ERK-mediated phosphorylationof GRK2 not only reduces the GRK2 activity but also promotes its deg-radation (Elorza et al., 2003). Although cSrc- and ERK-mediated phos-phorylationcantarget GRK2for degradationindependently, ERKpreferentially phosphorylates GRK2 previously phosphorylated on ty-rosine residues by cSrc (Elorza et al., 2003). GRK2 associates with andisubiquitinatedbyMdm2, whichalsofacilitatesGRK2degradation(Salcedoet al., 2006). Mdm2-dependent GRK2degradationuponGPCRactivationisfacilitatedbypreviousGRK2phosphorylationatSer370 by MAP kinases but not by tyrosine phosphorylation (Noguset al., 2011). Arrestins are known to recruit cSrc (Luttrell et al., 1999),ERK (Luttrell et al., 2001), and Mdm2 (Shenoy et al., 2001) to activeGPCRs. Arrestinscaffolding activity is requiredfor cSrc- (Penelaet al., 2001), ERK- (Elorza et al., 2003), and Mdm2-mediated (Salcedoet al., 2006; Nogus et al., 2011) GRK2 degradation induced by GPCRactivation. However, in the absence of GPCR activity, arrestins do notparticipate in the Mdm2-mediated regulation of the GRK2 basal turn-over, but instead compete with GRK2 for Mdm2 and suppress basalGRK2degradation(Nogusetal., 2011). Arrestin-dependentphos-phorylation of GRK2 by cSrc, on the other hand, in addition to primingGRK2 for the phosphorylation by MAP kinase and subsequent Mdm2-dependent degradation upon GPCR activation (Sarnago et al., 1999),is able to drive Mdm2-independent basal GRK2 degradation in the ab-sence of GPCR activity (Nogus et al., 2011). Thus, arrestins play a co-ordinating role recruiting kinases and/or ubiquitin ligases to GRK2 inthebasal conditionoruponactivationof GPCRs, regulatingGRK2turnover via different pathways.Dynamic regulation of the GRK2 expression during cell cycle pro-gression depends on its phosphorylation at Ser370by cyclin-depen-dent kinase 2 (CDK2) followed by the binding of a prolyl isomerasePin1 to phosphorylated GRK2 (Penela et al., 2010). This sequence ofeventsiscritical fortheGRK2degradationanditsdownregulationduring G2 phase of the cell cycle. In contrast to other modes of regu-lation, CDK2-Pin1-dependentregulationof GRK2degradationdoesnot require arrestins (Penela et al., 2010).The picture of the regulation of the GRKexpression remains in-complete. TherecentintriguingndingthatfemalesexpressmoreGRK3 and 5 in the brain than males, whereas the levels of GRK2 and6 are essentially the same (Bychkov et al., 2010), indicates that sex-specic regulatory mechanisms might be involved.8. G protein-coupled kinase isoforms more of the same?No review of GRKs would be complete without discussion of recep-tor specicity of GRK isoforms. The situation with GRKs resembles thatwith arrestins (Gurevich & Gurevich, 2006a) in that there are too fewofthem to be strictly receptor specic. Thus, it is often assumed that GRKisoforms are nonspecic towards GPCRs. However, the issue is not assimple as it appears. There are just too many GRK isoforms preservedover millions of years of vertebrate and mammalianevolutionto believethat they are promiscuous and entirely interchangeable. Even if we ig-nore highly specialized visual isoforms (GRKs 1 and 7), there are stilltwo subfamilies, GRK2/3 and GRK4, comprising two and three mem-bers, respectively, not counting splice variants, so there must be a func-tional reason for their existence.There are several aspects to the issue of receptor specicity. Firstquestion is whether any GRK is capable of phosphorylating any recep-tor. Such question is answered by testing receptor phosphorylation invitro with puried GRKs and puried receptors in natural or articialmembranes. Due to easy availability, by far the most popular prepara-tion has been puried rhodopsin in disk membranes as a substrate.The answer to the initial question seems to beyes, as all GRKs arebiochemically capable of phosphorylating any receptor, although notnecessarily with the same efcacy. Rhodopsin is a non-cognate recep-tor for all GRKs except GRK1, and yet all GRK isoforms phosphorylateit, indicating some degree of biochemical promiscuity. GRK6 is an iso-formthat acts as very weak kinase for rhodopsin (5% activity of GRK2;Benovic & Gomez, 1993). GRK6 phosphorylates 2AR somewhat bet-ter (35% of the GRK2 activity; Benovic & Gomez, 1993). GRK5, a mem-ber of the same GRK4/5/6 subfamily, is signicantly less activetowardsrhodopsin, 2AR, andmuscarinicM2receptorthanGRK2,but it is a much stronger kinase for rhodopsin and a slightly better ki-nase for 2AR than GRK6 (Kunapuli et al., 1994b). The deletion of theC-terminal sequence in GRK6C splice variant results in a signicantlyhigher kinase activity towards light-activated rhodopsin than that ofthe longer variant GRK6A (which was originally cloned and most ex-tensively studied) and the longest GRK6B (Vatter et al., 2005). It re-mains to be seen whether GRK6 has generally lower kinase activitythan other isoforms or simply stricter receptor requirements that arestill undiscovered. Some receptor preferences evident in the in vitrostudies with puried proteins are common for GRK subfamilies, suchas the ability of GRK2 and 3 to phosphorylate muscarinic M3 receptor,which is not signicantly phosphorylated by GRKs 5 or 6 (Debburmanet al., 1995a). The repertoire of GPCRs available in puried form for invitro phosphorylation studies has been so far quite limited. A widerarray of receptors might yet reveal unexpected biochemical receptorspecicity of GRK isoforms.Second question is how different GRKs operate in living cells. Eachcell expresses many GPCRs and more than one GRK isoform. In and of it-self, it does not mean that any GRK has a chance to interact with any re-ceptor. A cell is a highly compartmentalized environment and,consequently, someGRKisoforms andGPCRs subtypes maynevermeet. Furthermore, there might be conditions in a particular cell type,such as the presence of GRK modulating proteins, discouraging or pre-cluding GRK interaction with particular GPCRs. Another considerationisthecellularcomplementof GRKisoformsandrelativeexpressionlevels. Just because a GRK can phosphorylate a receptor does not meanit wouldever have a chance, simply because another more abundant iso-form would always get thererst. Thesener points are hard to appre-ciate in experiments involving heterologously expressed, andoverexpressed, receptors and kinases. Receptors or GRKs may not local-ize to proper cellular compartments due to unnaturally high level of ex-pressionand, if they are not native to the cells they are expressedin, maynot be subject to proper regulatory inuences. An overexpressed kinasedue to its abundance would always have an edge over endogenous pro-teins. Recent studies of endogenously expressed receptors and kinasesproved that GRK isoforms could be quite selective towards receptors.These studies employed knockdown of endogenous GRKs or used dom-inant-negative constructs to interfere with the kinase functions. HEK293cells have a full complement of non-visual GRK isoforms. Of these, GRKs2, 3, and 6 apparently participate in desensitization of endogenous mus-carinic M3 receptors, since knockdown of these isoforms enhanced car-bachol-induced calcium mobilization, whereas GRK5 knockdown wasineffective (Luo et al., 2008). In HEK293 cells, GRK2 and 6 were chieyresponsibleforarrestinrecruitmenttotheheterologouslyexpressed2 adrenergic receptor (Violin et al., 2006). In contrast, in U2-OS osteo-sarcoma cells, which express little or no GRK6, GRKs 2 and 3 were themost efcacious. These data stress the contribution of the cellular com-plement of GRKs to the apparent receptor specicity. In humanneuro-blastoma SH-SY5Y cells, GRK6, but not GRKs 2, 3, or 5, phosphorylatedanddesensitizedendogenousM3muscarinicreceptor(Willetsetal.,51 E.V. Gurevich et al. / Pharmacology & Therapeutics 133 (2012) 40692002, 2003). Interestingly, in the in vitro experiments, neither GRK6 norGRK5 phosphorylated puried M3 receptor, whereas both GRK2 and 3phosphorylated it (Debburman et al., 1995a). One has to draw inevita-ble, althoughdiscouraging, conclusionthat thereceptor preferencefound in vitro may not be the same a GRK would exhibit in cells.Inadult cultured cardiac myocytes, GRK3 effectively desensitized endoge-nousendothelin-1and1-adrenergicreceptors, whereasGRK2wasmuch less effective (Vinge et al., 2007). Conversely, both kinases desen-sitized the 1 adrenergic receptor with similar efcacy, although the ef-cacyof GRK3at the1receptor was 20-foldlower thanat theendothelin-1 receptor. GRK3 also seems responsible for desensitizationof 1-adrenergic receptors in the heart of living mice (Eckhart et al.,2000; Vinge et al., 2008), whereas GRKs 2 and 5 regulate -adrenergicreceptor-mediated responses (Koch et al., 1995; Rockman et al., 1996;Iaccarino et al., 1998). These are interesting observations demonstratingthat even such closely related kinases as GRK2 and 3 may act quite dif-ferently in their native milieu. One possible mechanismcould be the dif-ferential preference of GRK2 and 3 not for the receptors but for Gisoformsnecessarytorecruit thesekinasestoactiveGPCRs(Daakaet al., 1997). In that study, GRK2 was found to form a complex withG following stimulation of endogenous 2AR and lysophosphatidicacidreceptorsbutnotthrombinreceptors, whereasGRK3interactedwith G after stimulation of all three GPCRs (Daaka et al., 1997). If dif-ferent receptors engage different G isoforms, then a preference of aGRK for particular G would result in some receptors not recruiting itand, therefore, not being regulated by it.Studies with GRK knockout mice contributed their share of com-plications to the question of receptor specicity of GRKs. For suppos-edly promiscuous kinases, knockout of individual GRK isoformsresultedinmild, yet surprisinglydistinct phenotypes. Apparently,mammalian GRKs have partially overlapping receptor specicity, sothat remaining GRK isoforms successfully compensate for the missingone in most, but not all, cases. The only exception was GRK2, knock-out of which in mice is embryonic lethal due to abnormal formationof theheart (Jaber et al., 1996). Apparently, this lethalitystemsfrom general, albeit undened, role of GRK2 in embryogenesis, ratherthan specic role in the heart development, since mice with GRK2 ab-lation specic to the cardiac myocytes develop normally (Matkovichet al., 2006). OtherGRKknockout linesdisplaydecitsinspecicfunctions. KnockoutofGRK3, theclosestrelativeofGRK2, resultedin complete loss of olfactory function (Peppel et al., 1997), which islikely explained not by particular specicity of GRK3 for odorant re-ceptors but by the fact that GRK3 is the major, perhaps theonly GRK expressed in the olfactory epithelium (Peppel et al., 1997;Gurevich et al., 2004). Mice lacking GRK3 are resistant to opioid-in-duced dysphoria (Bruchas et al., 2007) and showreduced tolerance toopioid analgesics (McLaughlin et al., 2004; Terman et al., 2004). Themice with GRK5 deletion are functionally supersensitive to muscarin-ic, but not dopaminergic, stimulation (Gainetdinov et al., 1999) andhave reduced hippocampal acetylcholine release due to impaired de-sensitization of M2/M4 autoreceptors (Liu et al., 2009).Conversely,mice la