Nonenzymatic Rapid Control of GIRK Channel Function by a G Protein-Coupled Receptor Kinase  Adi Raveh, Ayelet Cooper, Liora Guy-David, Eitan Reuveny  Cell  Volume 143, Issue 5, Pages 750-760 (November 2010) DOI: 10.1016/j.cell.2010.10.018 Copyright © 2010 Elsevier Inc. Terms and Conditions

Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 1 GRK2 Accelerates the Desensitization of GIRK Currents Induced by A1R, but Not by mGluR2 (A) GIRK channel currents induced by the activation of A1R rapidly desensitize in the presence of GRK2. (B) GIRK channel currents induced by mGluR2 activation are insensitive to GRK2. (C) Bar plot that depicts GCD rates of cells activated with A1R or mGluR2 without or with GRK2, GRK2 shRNA, or nontarget (NT) shRNA. (D) Bar plot compares the normalized expression levels of GRK2 in silenced and NT cells as depicted from western blot for GRK2 (inset). (E) GIRK current traces induced by adenosine in control HL-1 cell (black) and of siRNA#1 silenced cell (gray). (F) Bar plot depicting GCD in HL-1 cells transfected with two independent siRNAs, NT, and siRNA#1 transfected cells rescued by the expression of silently mutated GRK2GFP (smGRK2GFP). (G) GRK2 mRNA quantification in HL-1 cells transfected with two independent siRNAs or NT control. See also Figure S1. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 2 A1R Activation Recruits GFP-Tagged GRK2 to the Membrane Simultaneously with GCD as Revealed under TIRF (A) A TIRF image of HEK293 cell transfected with GRK2-GFP. Basal membranous fluorescence can be detected before stimulation by A1R. (B) Image of the same cell in the presence of adenosine. (C) Time course of fluorescence increase seen on receptor activation. (D) A bar plot comparing the relative membrane-associated fluorescent change (ΔF/F, in %) of GRK2-GFP after activation of A1R or mGluR2. (E) Typical trace of whole-cell GIRK currents (black) and TIRF signal (green) recorded simultaneously from the same cell. (F) Bar graph depicting the similarity between GRK2-GFP translocation and GCD rates. See also Figure S2. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 3 Kinase Catalytic Activity Is Not Required for GRK2 Effect on GCD A1R-GFP or GIRK1/GIRK4-GFP plasma membrane levels are not affected by A1R stimulation. PTX treatment is not affecting basal GCD and membrane recruitment. (A) A bar graph summarizing measurements of GCD rates (τ, s) from cells cotransfected with GRK2/K220R (dnGRK2), GIRK, and A1R. (B) The relative change of membrane fluorescence under TIRF (ΔF/F, %) associated with either A1R-GFP or GIRK1/GIRK4-GFP before and during A1R activation (1 min after adenosine application). (C) Typical current traces of cells expressing GIRK and A1R (control); GIRK, A1R and PTX (+PTX); and GIRK, A1R, PTX and GRK2 (+PTX +GRK2). (D) A bar plot summarizing ΔF/F of GRK2-GFP signal after A1R activation, measured under TIRF in cells expressing GIRK, PTX, and GRK2. (E) A typical TIRF data of the membrane fluorescence change of GRK2-GFP overtime of a cell expressing PTX after A1R activation. See also Figure S3. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 4 GRK2 Mutants with Impaired Gβγ Binding Capability Fail to Accelerate GIRK Desensitization (A) A cartoon that displays the structure of the complex of GRK2 with Gβγ (Tesmer et al., 2005). The locations of the different point-mutations that were used in (B) are marked in red. (B) A bar plot summarizing the desensitization rates (τ, s) of GIRK currents, measured from cells transfected with GIRK channel, A1R, and the various GRK2 mutants. (C) A bar plot comparing the effect of myristoylated GRK2 (myr-GRK2) and myrGRK2/R587Q mutant on GIRK desensitization rate. See also Figure S4. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 5 Constituently Active, Gβγ-Independent, but Not GIRK Mutants that Have Higher Affinity to PtdIns(4,5)P2, Are Insensitive to GRK2 (A) Typical traces of GIRK1/S170P;GIRK4/S176P channel mutants, without (upper trace) or in the presence of GRK2 (lower trace). (B) A bar plot summarizing the residual current (in % of total current) after agonist application without (dark gray) and in the presence of GRK2 (light gray). (C) Typical current traces of GIRK1/M223L;GIRK4/I229L channel mutants, without (upper trace) or in the presence of GRK2 (lower trace). (D) A bar plot summarizing the residual current (in % of induced current) after agonist application from cell without (dark gray) and with GRK2 (light gray). In both case, desensitization was measure 5 s after agonist application. See also Figure S5. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 6 FLIM-FRET under TIRF Reveals that A1R Activation Increases the Fraction of GRK2-Bound Gβγ (A) A cartoon showing the experimental scheme used in the FLIM-FRET experiments. (B) Time course of YFP-Gβ1 emission after the activation of A1R in the presence of GRK2-Cherry, in agreement with YFP quenching by mCherry on increase of FRET. (C) YFP-Gβ1 lifetime changes after A1R activation. Yellow symbol depicts the FRET-free YFP-Gβ1 (τ of 3.0 ns fraction) and red symbol depicts the faster lifetime component (0.9 ns) that corresponds to a FRET interaction between YFP-Gβ1 and GRK2-Cherry. This FRET interaction corresponds to a FRET efficiency of 0.7. (D) YFP-Gβ1 GRK2-Cherry binding increases after A1R activation (n = 9). Black line depicts the fitting to a monoexponential function with a τ = 4.2 ± 0.7 s. See also Figure S6. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure 7 A Cartoon Describing the Mechanism by Which GRK2 Is Negatively Regulating GIRK Channel Function On receptor stimulation by GPCR, the G protein trimer undergoes activation characterized by the exchange of GDP for GTP on the Gα subunit. This in turn leads to the dissociation of the Gβγ subunits to freely bind and activate the GIRK channel. Concomitantly, the GPCR induces the recruitment of GRK2 to the plasma membrane making it available to bind Gβγ subunits of the G protein. Due to the relative higher affinity of GRK2 for Gβγ and to the larger mass action, GRK2 is now able to effectively compete for the available pool of Gβγ with the GIRK channel, leading to a gradual removal of the Gβγ subunits and to a channel closure (desensitization), still in the presence of the receptor agonist. Channel activation precedes the action of GRK2 mainly due to the preexisting trimeric G proteins in the vicinity of the channels (Riven et al., 2006). Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S1 GRK2 Accelerates the GCD Induced by μOR, but Not by M4R in HEK293 Cells, and Desensitizing GIRK Currents Induced by Adenosine Receptors in Hippocampal Neurons, Related to Figure 1 (A) Representative traces of GIRK currents in HEK cells induced by μOR activation in the absence (upper trace) and in the presence (lower trace), of GRK2. (B) Representative traces of GIRK currents in HEK cells induced by M4R activation in the absence (upper trace) and in the presence (lower trace) of GRK2. (C) Bar plot that compares GCD rates in HEK cells without (control) or with GRK2 cotransfection. (D) Bar plot presenting GIRK desensitization and GRK2 membrane recruitment rates (τ, s) following μOR activation in HEK cells. (E) A typical trace of whole-cell currents following adenosine applications (1 min, 100 μM) in hippocampal neurons transfected with GRK2 and GRK3 siRNAs, compared to NT control transfection. (F) Bar plot that compares the ratio of two consequent GIRK activations by continuous adenosine applications (1 min, 100 μM). Current inductions shows a ratio of 100.1 ± 6.2% (n = 8) in GRK silenced cells compared to 37.1 ± 7.2% (n = 7) in NT control cells (p < 0.05). (G) A bar plot comparing GRK2 and GRK3 mRNA levels in neuronal culture cells transfected with GRK2 and GRK3 siRNAs in comparison to neuronal culture cells transfected with NT control. GRK2 and GRK3 mRNA levels were reduced by 21% and 25%, respectively, following GRK silencing. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S2 GRK2 Membrane Recruitment Is Reversible, Related to Figure 2 (A) A1R activation is followed by recruitment of GRK2-GFP to the plasma membrane, measured under TIRF. Membrane fluorescence returns to its basal level after washing the agonist, and a second application of agonist, results in similar membrane recruitment. (B) TIRF images of the cell analyzed in (A) before, during, and after second agonist application and washout. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S3 The Gs or Gq Pathways Are Not Involved in GRK2-Mediated GIRK Desensitization, Related to Figure 3 (A and B) (A) Typical current traces of cells coexpressing GIRK channels and A1R without (control) or with GRK2 (B) cotransfection of both with Gαs/α3β5;G226A;A366S (dnGαs), a robust dominant negative Gαs subunit of the G protein (Berlot, 2002). (C) A bar plot comparing GIRK current desensitization induced by A1R activation without (control) or with GRK2 in the presence of dnGαs cotransfection. dnGαs had no impact on GRK2-mediated desensitization. Comparing cell without or with GRK2 cotransfection shows τ of 24.4 ± 10.3 s, n = 4 and 3.1 ± 0.9 s, n = 4, respectively. (D) A typical trace showing the fluorescence signal from a TIRF image of the cell plasma membrane overexpressing GRK2-GFP, GIRK, A1R and dnGαs following adenosine application. No effect of dnGαs transfection on GRK2-GFP translocation to the membrane was observed (n = 7). (E) A typical calcium signal (Fluo-4AM) of non-transfected HEK293 cells shows no release of calcium from intracellular stores following adenosine application (100 μM). In contrast, activation of the endogenous muscarinic receptors of the Gq pathway (e.g., M1 of M3 (Leaney et al., 2004)) by carbachol application (100 μM) elicits a robust release of calcium from the stores. (F) A typical intracellular calcium signal of HEK293 cells transfected with A1R and GIRK, shows no response to the adenosine application but activation of the endogenous muscarinic receptors by carbachol is intact. (G) Intracellular calcium responses for carbachol, but not for adenosine, in the presence of GRK2. (H) Complete attenuation of the carbachol-induced intracellular calcium release by 1 hr preincubation with 2-Nitro-4-carboxyphenyl-N,N-diphenylcarbamate (NCDC, 200 μM, Apollo Scientific, UK), phospholipase C-β1 (PLC) inhibitor (Sickmann et al., 2008). These results are in agreement with Leaney et al. (Leaney et al., 2004) who did not detect calcium response following the activation of A1R in HEK293 cells with a stable expression of A1R, using Fura-2AM imaging at saturating concentration of NECA (1 μM), a full A1R agonist. (I) The desensitization rate (τ, s) of GIRK currents in cells transfected by GIRK and A1R (control), or with GRK2 with or without the pre-treatment with NCDC. GCD in GRK2 transfected cells were not significantly different with or without treatment with NCDC, with τ of 4.1 ± 1.8 s, n = 10 and 3.5 ± 1.6 s, n = 7, respectively, both showed a remarkably faster desensitization compared to the control group (18.8 ± 2.0 s, n = 40). Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S4 GRK2 Interaction with Gαq Is Not Required for GRK2-Mediated GIRK Current Desensitization, Related to Figure 4 GRK2 mutations that impair GRK2 interactions with Gβγ or with PtdIns(4,5)P2, maintain a reduced level of GRK2 membrane recruitment following A1R activation. (A) A bar plot that compares GIRK current desensitization rates (τ, s) of A1R activated GIRK currents in cells cotransfected with GIRK, A1R and mutants of GRK2 that impair its interaction with Gαq (Sterne-Marr et al., 2003). (B) A bar plot summarizing the relative fluorescence change (ΔF/F, in %) of the different GRK2-GFP mutants that were described in Figure 4. GRK2/R587Q-GFP and GRK2/K663E;K665E;K667E-GFP mutants, with impaired Gβγ binding, respond to A1R activation by 7.5 ± 1.0% (n = 7) and 4.4 ± 2.7% (n = 9), respectively. GRK2/K567E;R578E-GFP mutant, with impaired PtdIns(4,5)P2 binding, responds to A1R activation by 5.6 ± 1.8% (n = 8) and, GRK2/K567E;R578E;R587Q-GFP a triple mutant that combines mutations that impair both Gβγ and PtdIns(4,5)P2 binding, responds to A1R activation by 0.2 ± 0.5% (n = 7), compared to wt GRK2-GFP response to A1R activation of by 19.2 ± 2.2% (n = 16). Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S5 Constituently Active GIRK Mutants (GIRK1/S170P;GIRK4/S176P) Display a Robust Basal Activity in Comparison to WT Channels, Related to Figure 5 A comparison of basal and induced GIRK activity (% of total) between constituently active GIRK mutants (GIRK1 /S170P;GIRK4/S176P) and wt GIRK channels. GIRK channels were gated by A1R activation. For both wt GIRK and constitutive active mutant channels currents were normalized to current levels in the presence of the receptor agonist. Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions

Figure S6 High-efficiency FRET between YFP and Cherry and Monoexponential Lifetime Decay Observed from YFP-β1 under TIRF Configuration, Related to Figure 6 (A) Histogram of YFP fluorescence lifetime displays nearly monoexponential decay (2.61 ns). YFP-Cherry dimer displays double-exponential lifetime histogram, the first component of 2.61 ns (6.8%) and the second component of 0.59 ns (93.2%) that correspond to 77% FRET efficiency. The deviation from expected 100% FRET in the dimer molecules may stem from immature Cherry molecules in the dimer, as has been proposed earlier (Yasuda et al., 2006). Data was deconvoluted with instrument response function of 300 ps (FWHM). Black lines depict fitting to a monoexponential or double-exponential model for YFP and YFP-Cherry, respectively. (B) A fluorescence lifetime histogram of Gβ1-YFP as recorded in TIRF configuration (yellow dots). IRF was reconstructed from Gβ1-YFP monoexponential decay and is shown as blue histogram. The black line depict fitting to a monoexponential model (τ = 3.0 ns), and on the bottom the homogenous distribution of fit residuals is displayed as blue dots around the zero dotted line. Fluorescence was collected for 1 s, to assess the fitting quality to decay data collected over a 1 s time bins as done and shown in Figure 6. (C) Gβ1-YFP GRK2-Cherry binding increases following μOR activation (n = 6). Black line depicts the fitting to a monoexponential function. τ average of binding increase is 69.5 ± 15.3 s (n = 6). Cell 2010 143, 750-760DOI: (10.1016/j.cell.2010.10.018) Copyright © 2010 Elsevier Inc. Terms and Conditions