Mechanism of PKCε regulation of cardiac GIRK channel gating

Significance

Overactivity of a potassium ion channel (GIRK1/4) in the upper chambers of the heart has been thought to contribute to atrial fibrillation (AF), a widely prevalent heart rhythm abnormality or arrhythmia in the aging population and a significant cause of mortality. The treatment of AF is largely comprised of medications that correct heart rhythm, but are inadequate and have non-specific side effects. We have identified an amino acid on GIRK1/4, that when phosphorylated by the enzyme protein kinase C-ε, leads to increased GIRK1/4 activity. We explore the mechanism underlying channel overactivity in order to provide a strategy for drug discovery to reverse overactivity, without compromising the physiological function of the channel.

Abstract

G-protein-gated inwardly rectifying potassium (GIRK) channel activity is regulated by the membrane phospholipid, phosphatidylinositol-4,5-bisphosphate (PI 4,5P₂). Constitutive activity of cardiac GIRK channels in atrial myocytes, that is implicated in atrial fibrillation (AF), is mediated via a protein kinase C-ε (PKCε)-dependent mechanism. The novel PKC isoform, PKCε, is reported to enhance the activity of cardiac GIRK channels. Here, we report that PKCε stimulation leads to activation of GIRK channels in mouse atria and in human stem cell-derived atrial cardiomyocytes (iPSCs). We identified residue GIRK4(S418) which when mutated to Ala abolished, or to Glu, mimicked the effects of PKCε on GIRK currents. PKCε strengthened the interactions of the cardiac GIRK isoforms, GIRK4 and GIRK1/4 with PIP₂, an effect that was reversed in the GIRK4(S418A) mutant. This mechanistic insight into the PKCε-mediated increase in channel activity because of GIRK4(S418) phosphorylation, provides a precise druggable target to reverse AF-related pathologies due to GIRK overactivity.

Atrial fibrillation (AF) is the most common type of cardiac arrhythmia and is projected to afflict 15 million individuals by the year 2050, in the United States alone (1). The excitability of atrial cardiomyocytes is affected by the activity of G-protein-gated inwardly rectifying potassium (GIRK) channels that pass the acetylcholine (ACh)-induced inward rectifying K⁺ current, I_KACh. ACh regulates I_KACh indirectly by stimulating muscarinic M₂ receptors and releasing the Gβγ dimer, which then activates GIRK channels (2). In chronic AF, a constitutive increase in I_KACh is observed in the absence of ACh stimulation. This promotes the shortening of the atrial action potential duration (APD) and consequently the perpetuation of AF (3, 4). GIRK channels in atrial tissue exist as GIRK1/4 (also called K_ir3.1/3.4) heteromers and GIRK4 homomers (5, 6). I_KACh is differentially modulated by members of the protein kinase C (PKC) enzyme family. The conventional PKC isoforms are reported to diminish I_KACh while the novel PKCs (nPKCs), including PKCε, enhance I_KACh (7). The phosphorylation of atrial GIRK1/4 channels by PKC and PKA (protein kinase-A) was shown by Medina et al., using neonatal rat atrial myocytes (6). Their studies reported a definitive role for phosphorylation in the case of GIRK1 but not for the GIRK4 subunit. These authors suggested a complex interplay between GIRK1/4 regulation by kinases like PKA, PKC, and phosphatases like protein phosphatase 2B. PKA has subsequently been shown to phosphorylate S385 on the GIRK1 subunit (8) while PKC, which is usually activated by canonical signaling through Gq receptors, has been widely shown to inhibit GIRK channel activity (9, 10).

Phospholipase C activation by Gq-coupled receptors results in GIRK channel inhibition through phosphatidylinositol-4,5-bisphosphate (PIP₂) depletion as well as PKC-mediated channel phosphorylation, which also diminishes channel activity (10). The extent of K⁺ channel inhibition by PKC appears to be accomplished by diminishing the PIP₂ affinity of the Kir3 channel subfamily (10). An enhancement in PKC inhibition of channel activity upon depletion of the PIP₂ content in the membrane as well as a rightward shift in the EC₅₀ for diC8 PIP₂ suggests that the nPKC, PKCδ, inhibits GIRK1/4 channels by weakening the affinity of the channel for PIP₂ (10).

In contrast, PKCε is reported to enhance atrial GIRK channel activity (7). We have also previously shown the enhancement in current of a GIRK1/4 concatemeric channel using an optogenetically activated PKCε (11). While the mechanism underlying this distinct effect has not been established, the G_q signaling mediated hydrolysis of PIP₂ and subsequent activation of PKC’s brings to the fore a delicate interplay between the phospholipid that controls GIRK channel activity and the effects of its second messenger, PKC.

Here, we establish the role of PKCε in atrial GIRK channel regulation in isolated perfused mouse hearts and human iPSC atrial cardiomyocytes and identify a critical serine residue in the GIRK4 subunit where phosphorylation of GIRK4(S418) opens the channel by strengthening channel–PIP₂ interactions.

Results

DCP-LA Alters the Atrial Action Potential in the Mouse Heart as well as in Human Atrial iPS Cardiomyocytes.

We used optical mapping of epicardial voltage to investigate whether PKCε activation leads to constitutive activation of I_KACh in the right atria of isolated, Langendorff-perfused mouse hearts. In Fig. 1, the fluorescence pictures of the right atria in A show the superior vena cava, the right atrial appendage, and the stimulation electrode. Under control conditions, the APD, measured at 60% repolarization (APD₆₀) in the stimulated right atrium, did not show appreciable prolongation after administration of the bee venom toxin, tertiapin-Q, a potent GIRK1/4 channel blocker (Fig. 1A). The bar graph (Fig. 1B) compiled results from three hearts where tertiapin-Q did not significantly prolong the APD₆₀ indicating that in DMSO (vehicle control) no significant constitutive GIRK channel activity was seen (12).

Fig. 1.

PKCε activation alters the action potential in the optical mapping of mouse hearts. A, Fluorescence images of the mapped field. SVC; superior vena cava, RAA; right atrial appendage, RV; right ventricle, Stim; bipolar stimulation electrode. Heat maps cover the RAA area only in the middle panels and represent the APD (APD₆₀) maps after 25 min of perfusion with vehicle (DMSO) (Top) or 10 μM DCP-LA (Bottom). The Right panels are the APD₆₀ maps after administration of 200 nM tertiapin-Q. B, the bar graphs show that in DMSO-treated hearts, tertiapin-Q did not significantly prolong the APD₆₀ (N.S., paired Student’s t test, N = 3 hearts). In DCP-LA-treated hearts, tertiapin-Q significantly prolonged the APD₆₀ (*P < 0.05, paired Student’s t test, N = 4 hearts). C is a quantification of the percent of APD₆₀ change after tertiapin-Q administration, which significantly increased the APD₆₀ in DCP-LA-pretreated hearts (*P < 0.05, one sample t test, N = 4), but not in DMSO pretreatment (N.S. one sample t test, N = 3).

Next, we used tertiapin-Q to study the role of GIRK channels in hearts pretreated with DCP-LA (also known as FR236924), a linoleic acid derivative that activates PKCε (13). Following pretreatment with DCP-LA, the APD₆₀ map of the stimulated right atrium showed appreciable prolongation of the atrial action potential upon tertiapin-Q administration (Fig. 1 A -bottom and B). These data indicate that the statistically significant difference between the APD₆₀ before and after tertiapin-Q treatment in the DCP-LA group can be attributed to enhanced I_KACh, indicating that DCP-LA resulted in constitutive activation of I_KACh, likely via PKCε stimulation. Fig. 1C shows the percent APD₆₀ prolongation after tertiapin-Q administration in DMSO- and DCP-LA-treated hearts. Tertiapin-Q in DMSO did not cause a significant prolongation of the APD₆₀, in contrast with its significant effect in DCP-LA-treated hearts. Altogether, these data demonstrate that in the heart, PKCε stimulation leads to constitutive activation of I_KACh. To assess the effects of selective activation of PKCε at the cellular level, we used perforated patch-clamp recordings to study action potential morphology in human-induced atrial cardiomyocytes following treatment with DCP-LA. The APD₅₀ was significantly shortened in cells treated with DCP-LA (SI Appendix, Fig. S1 and Fig. 2A). Under voltage-clamp conditions, the basal activity of atrial GIRK channels increased almost fivefold (−31.79 ± 13.08 pA/pF) upon treatment with DCP-LA, compared to the control conditions (−6.3 ± 3.8 pA/pF) (Fig. 2 B and C).

Fig. 2.

PKCε activation alters action potential parameters in human iPSC atrial cardiomyocytes. Human-induced pluripotent stem cell differentiated into atrial cardiomyocytes were treated with 100 µM DCP-LA for 90 min prior to patch-clamp electrophysiology experiments. A, Spontaneous action potentials were elicited under current-clamp conditions and APD₅₀ was measured in control and DCP-LA treated cardiomyocytes (****P < 0.00005 using Student’s unpaired t test; n = 6 to 7 cells per condition). B, Tertiapin-Q sensitive potassium currents were assessed in control and DCP-LA-treated cells under voltage-clamp conditions (**P < 0.005 using Student’s unpaired t test). C, Representative I_KACh currents from atrial cardiomyocytes. Data are current densities of n = 5 cells per condition). C, Voltage ramp protocol and representative traces of GIRK channel currents in control and DCP-LA-treated conditions are shown.

PKCε Activates Atrial GIRK Channel Assemblies.

To isolate the effect of PKCε on atrial GIRK channels, we first examined the effect of the novel PKC isoform, PKCε, on GIRK4 homomers expressed in heterologous systems. We used GIRK4(S143T) also known as GIRK4*, an active mutant of GIRK4, which has been demonstrated to produce larger G-protein-gated, inwardly rectifying currents than its wild-type counterpart (14). The catalytic subunit of PKCε, PKCε-CAT, was used throughout this study. This precludes the need for activation of PKC through G_q-coupled receptor signaling which causes PIP₂ hydrolysis prior to PKC activation, that would introduce a confounding decrease in GIRK channel activity. Test conditions comprised of cells co-expressing GIRK4, muscarinic M₂ receptor, as well as PKCε, while control cells expressed GIRK4 and the muscarinic M₂ receptor. Whole-oocyte currents of GIRK4 channels in response to high K⁺ (96 mM, referred to as HK) in two-electrode voltage-clamp (TEVC) experiments established that basal currents of the channel, in the absence of an agonist, were significantly enhanced in the presence (shown in red) compared to the absence (shown in black) of PKCε-CAT (Fig. 3 A and B). ACh was applied to elicit Gβγ-mediated signaling through the co-expressed, M₂ receptor. ACh-induced currents were doubled when PKCε-CAT was co-expressed in the oocytes (Fig. 3 A and B). To corroborate data in the literature that suggests enhanced basal GIRK channel activity in response to PKCε in primary atrial cells (7), we tested the same paradigm in a heterologous, mammalian cell system using HEK cells. Similar to the results of oocyte experiments, GIRK4WT basal current density in HEK cells co-expressing PKCε-CAT was significantly greater than that in the absence of PKCε-CAT (Fig. 3 C and D).

Fig. 3.

PKCε enhances the activity of GIRK4-containing channels. A, TEVC experiments with GIRK4*S143T (GIRK4*) RNA co-expressed in Xenopus oocytes along with the muscarinic M₂ receptor with (red bars/triangles) or without the catalytic subunit of PKCε (black bars/triangles). B, Representative traces. C, Whole-cell patch-clamp experiments with GIRK4WT channels expressed in HEK293T cells with or without the catalytic subunit of PKCε. D, Representative traces of patch-clamp experiments. E, TEVC experiments with GIRK1/4 RNA co-expressed in Xenopus oocytes along with the muscarinic M₂ receptor with (red bars/triangles) or without the catalytic subunit of PKCε (black bars/triangles). F, Representative traces with or without PKCε. G, Whole-cell patch-clamp experiments with GIRK1/4 channels expressed in HEK293T cells with or without the catalytic subunit of PKCε. H, Representative traces of patch-clamp experiments. Basal GIRK activity in mammalian cells was evaluated in response to 140 mM K⁺ solution. Patch-clamp data are whole-cell current densities expressed as mean ± SD (n = 5 cells per condition). Basal GIRK activity in oocytes was evaluated in response to 96 mM K⁺ solution and agonist-induced activity was evaluated in response to ACh. Data are whole-oocyte currents expressed as mean ± SD (n = 8 cells per condition). Negative currents indicate inward flow of positively charged ions; **P < 0.05, **P < 0.005, ***P < 0.0005 calculated using Students’ t test.

In addition to GIRK4 homomers, GIRK1/4 heteromers are abundantly expressed in atrial myocytes. The heteromeric channel has almost twenty-fold greater activity than the GIRK4 homomer (15), indicating that the GIRK1/4 channel is the predominant contributor to I_KACh. We hence tested the heteromeric GIRK1/4 in the same paradigm as GIRK4. Oocytes co-expressing PKCε-CAT with GIRK1/4 showed greater inward basal as well as ACh-induced currents compared to the control group (Fig. 3 E and F). In mammalian cells, current density was significantly greater in cells that co-expressed PKCε-CAT than those that expressed GIRK1/4 alone (Fig. 3 G and H).

We proceeded to evaluate the effect of PKCε on the GIRK1 subunit. Since GIRK1 is unable to form functional homomers at the cell membrane, we used the active mutant GIRK1* which has a Ser in place of Phe137 in its pore helix region(16) (17). GIRK1* was assessed in TEVC experiments, much like GIRK4*. The muscarinic M₂ receptor was co-expressed with GIRK1* in oocytes, with (in red) or without (in black) PKCε-CAT. We observed a significant increase in basal GIRK1* currents when PKCε-CAT was co-expressed compared to oocytes that did not express the kinase (SI Appendix, Fig. S2). Similarly, agonist-induced channel activity was enhanced in the presence of the kinase as compared to oocytes that did not express PKCε-CAT.

PKCε Phosphorylates a C-Terminal Serine on GIRK4.

Zhang et al. eliminated possible GIRK4 phosphorylation sites through a mutagenesis approach in the N terminus at Thr37, Thr57, Thr70, and C terminus at Ser209 and Ser233 (18). Mutation of these sites did not rescue the GIRK4 channel from PKC-mediated inhibition by phorbol 12-myristate 13-acetate (PMA). We used GIRK4* chimeras with IRK1 to narrow down the possible phosphorylation site through the loss of phosphorylation-induced channel modulation, i.e., absence of GIRK4 activation by PKCε. These have been previously used to identify the binding sites for G-proteins on GIRK channels. The activity of various GIRK4*-IRK1 chimeras has been reported by our laboratory previously (19). The constructs encompass primarily the GIRK4* subunit with certain regions replaced by homologous regions of IRK1 (K_ir2.1). This channel is insensitive to modulation by PKC (20).

The normalized current of the GIRK4*-IRK1 chimera was comparable to that of the GIRK4* channel indicating that replacement of the C-terminal end of GIRK4* with IRK1 did not significantly alter its activity in oocytes. In contrast, the augmented normalized GIRK4* current seen with PKCε-CAT (in red), which is indicative of PKCε activation of GIRK4-containing channels, was lost in the GIRK4*-IRK1 chimera (SI Appendix, Fig. S3). These results implicated the cytosolic C-terminal end of the GIRK4* channel in PKCε-mediated channel regulation.

To identify sites on the GIRK4 channel subunit phosphorylated by PKCε, we used the group-based prediction system software, version 2.0 (17) to predict PKCε phosphorylation sites on the GIRK4 channel subunit. PKCs are cytosolic enzymes and will only have access to amino acid residues that are part of the cytosolic domain of each subunit, hence only sites in the cytosolic N and C termini were considered relevant. There were two predicted residues in the relevant regions: S241 and S418. Upon mutating Ser241 in the proximal C-terminal of GIRK4 to alanine, we observed a significant decline in channel activity, which was an impediment to studying phosphorylation-mediated changes in this mutant’s activity (SI Appendix, Fig. S4). We proceeded to test the second predicted residue, Ser418, also mutating it to alanine and performing TEVC experiments with PKCε. Basal GIRK4 channel currents were comparable to those of the GIRK4(S418A) mutant channel. When PKCε-CAT was co-expressed with wild-type GIRK4, basal channel activity predictably displayed a significant increase. The agonist-induced GIRK4 activity through the M₂ receptor was also significantly greater when PKCε-CAT was co-expressed with the channel, than GIRK4 activity without PKCε present. In contrast, the mutant GIRK4(S418A) channel, in the presence of PKCε-CAT, showed similar basal and ACh-induced activity to the oocytes that were not co-injected with PKCε-CAT mRNA (Fig. 4 A and B). These data indicate that Ser418 in GIRK4 is critical for the PKCε-mediated augmentation of channel activity. These findings are consistent with our studies using a GIRK4*-IRK1 chimera that implicated the C-terminal end of GIRK4 in the PKCε-mediated GIRK4 activation. Additionally, it appears that the alanine mutant can completely rescue the PKCε-mediated activation of GIRK4, showing that the phosphorylation of Ser418 is necessary for the activation of the GIRK4 channel by PKCε.

Fig. 4.

A C-terminal serine residue of GIRK4 is responsible for the stimulatory effects of PKCε. A, TEVC experiments with GIRK4WT and GIRK4(S148A) RNA co-expressed in Xenopus oocytes along with the muscarinic M₂ receptor with (red bars/circles) or without the catalytic subunit of PKCε (black bars/triangles). B, Representative traces of WT and mutant GIRK4 channels with and without PKCε. C, TEVC experiments with GIRK4WT and GIRK4(S418E) RNA co-expressed in Xenopus oocytes along with the muscarinic M₂ receptor with (red bars/triangles) or without the catalytic subunit of PKCε (black bars/circles). D, Representative traces of WT and mutant GIRK4 channels with and without PKCε. Basal GIRK activity in oocytes was evaluated in response to 96 mM K⁺ solution and agonist-induced activity was evaluated in response to ACh. Data are whole-oocyte currents expressed as mean ± SD (n = 8-15 cells per condition). Negative currents indicate inward flow of positively charged ions; **P < 0.05, **P < 0.005, ****P < 0.00005 calculated using one-way ANOVA with Tukey’s post hoc test for A and Dunnett’s post hoc test for C.

To further test whether S418 is a critical residue in the PKCε-mediated regulation of GIRK4, we replaced it with a glutamate residue. The negatively charged amino acid, glutamate, could serve to recapitulate the effect of a negatively charged phosphate group, hence serving as a phosphomimetic residue, in that it could mimic the electrostatic changes conferred through phosphorylation. Control TEVC experiments with oocytes expressing GIRK4 mRNA established the lower activity of the channel alone (Fig. 4 C and D) compared to oocytes co-expressing PKCε-CAT. In comparison, the phosphomimetic mutant, GIRK4(S418E), showed significantly greater basal and agonist-induced activity than GIRK4. These data indicate that a negative charge introduced in the form of an amino acid, like glutamate or a phosphate group, can up-regulate GIRK4 channel activity. Hence, phosphorylation of the serine residue at position 418 by PKCε is necessary and sufficient to cause a significant enhancement of the activity of GIRK4.

The GIRK4-active mutant, GIRK4*, contains a serine to threonine mutation in its pore helix region. This mutation enhances channel activity several-fold over the wild-type channel (14). Such mutations at homologous residues in the pore region of other GIRK subunits have been described in the literature, e.g., GIRK1(F137S) (21), GIRK2(E152D) (22). The GIRK4(S418E) mutation is unique in that it is at the penultimate residue in the cytosolic, C-terminal end of the channel. These data urged us to investigate the underlying mechanism that makes it possible for a cytosolic residue distant from the pore region—Ser418—to cause an enhancement of the activity of GIRK4-containing channels.

PKC Modulation of GIRK1/4 Channels Is Dependent on Channel–PIP₂ Interactions.

The mechanism of PKCε-mediated GIRK modulation was examined under the hypothesis that increased activity of PKCε-phosphorylated GIRK1/4 is a consequence of its increased affinity for the membrane phospholipid, PIP₂. Membrane PI(4,5)P₂ was depleted using optogenetically activated inositol-5-phosphatase (11). A schematic for optogenetic dephosphorylation of PIP₂ during patch-clamp experiments is shown in Fig. 5A.

Fig. 5.

PKCε enhances GIRK1/4 channel activity by strengthening channel–PIP₂ interactions. A, This shows the scheme for optogenetic depletion of PIP₂ using mCherry-CRY2-5-ptase_OCRL. HEK293T cells were transfected with GIRK1, GIRK4, GFP-CIBN-CAAX, mCherry-CRY2-5-ptase_OCRL. B and C, Representative traces showing GIRK1/4 current depletion with (C) and without (B) the presence of PKCε. D, GIRK1/4 current remaining after PIP₂ depletion is greater when PKCε is present. E, The kinetics of current depletion (tau) following PIP₂ depletion are slower when PKCε is present indicating greater affinity of the phosphorylated channel for PIP₂. F, The 5-ptase_OCRL-mediated decrease in current is characterized by mono-exponential fits in the presence and absence of PKCε. Data are currents recorded from HEK293T cells using patch-clamp in whole-cell mode and are shown as mean ± SD (n = 8 to10 cells per group). Statistical significance was calculated using Students’ t test, *P < 0.05, **P < 0.005.

The affinity of GIRK1/4 channels for PIP₂, and its modulation thereof by PKC, was assessed by dephosphorylating PIP₂ optogenetically. This was accomplished in whole-cell patch-clamp experiments by illuminating HEK-293T cells expressing GIRK1/4 channels, light-activated CRY2-5ptase_OCRL, and GFP-CIBN-CAAX with and without co-expression of PKCε. CRY2-5ptase_OCRL is recruited to the cell surface and dimerizes with the membrane anchored CIBN when excited with blue light, and dephosphorylates the 5-phosphate of PI(4,5)P₂, causing GIRK channel current to decrease due to PIP₂ depletion (23). Currents recorded in cells expressing GIRK1/4 with PKCε were inhibited to a lesser extent by the phosphatase compared to cells only expressing GIRK1/4 (Fig. 5 B–D), i.e., more current remained after PIP₂ depletion in PKCε-expressing cells (Fig. 5 C and D) than in cells expressing only GIRK1/4 channels (Fig. 5 B and D), showing that PKCε-phosphorylated GIRK1/4 channels have increased sensitivity to PIP₂. Additionally, the kinetics of inhibition of GIRK1/4 activity by CRY2-5ptase_OCRL were slower for the PKCε-phosphorylated channel than the control (Fig. 5 E, F), as demonstrated by the tau (τ) value generated by fitting the current traces to the Boltzmann equation after phosphatase activation. The greater tau (τ) value in the presence of PKCε (Fig. 5 E and F) indicates that the channel, following phosphorylation by PKCε-CAT, has increased affinity for PIP₂ which explains the observed, slower kinetics of inhibition by the PIP₂ phosphatase. These experiments show that PKCε-phosphorylated GIRK1/4 channels have an increased affinity for PIP₂ which underlies their increased activity.

GIRK4(S418A) Reverses the Effects of PKCε on Channel–PIP₂ Interactions.

We harnessed the power of the phosphatase assay to compare metrics of current depletion as an indicator of altered channel function. Phosphatase activation through the illumination of HEK-293T cells with blue light revealed a comparable rate and magnitude of current inhibition in cells expressing GIRK4-WT alone (Fig. 6A) and cells expressing GIRK4(S418A) with PKCε (Fig. 6C). While the tau of inhibition in both these experimental conditions did not differ from one another, both were significantly faster (Fig. 6 E, F) in comparison with the phosphatase-induced current depletion seen in cells expressing GIRK4 with PKCε-CAT (Fig. 6B).

Fig. 6.

S418 is a critical residue in mediating the effects of PKCε on GIRK4. HEK293T cells were transfected with GIRK4 or GIRK4(S418A), GFP-CIBN-CAAX, mCherry-CRY2-5-ptase_OCRL. A–C, Representative traces showing GIRK1/4 current depletion with (B) and without (A) the presence of PKCε and C, GIRK4(S418A) with PKCε. D, GIRKWT current remaining after PIP₂ depletion is greater when PKCε is present, unlike GIRK4(S418A). E, 5-ptase_OCRL-mediated decrease in current is characterized by mono-exponential fits in the presence and absence of PKCε. F, The kinetics of current depletion (tau) following PIP₂ depletion are slower when PKCε is present indicating greater affinity of the phosphorylated channel for PIP₂. Data are currents recorded from HEK293T cells using patch-clamp in whole-cell mode and are shown as mean ± SD (n = 5 to 8 cells per group). Statistical significance was calculated using one-way ANOVA and Tukey’s post hoc test, *P < 0.05, **P < 0.005.

To emphasize the critical nature of Ser418, we co-expressed GIRK1 with GIRK4(S418A) in oocytes to evaluate the effect of PKCε on the phosphor-insensitive GIRK4-containing heteromer in optogenetic PIP₂-depletion experiments (SI Appendix, Fig. S5). We observed that the mutant-containing heteromer has greater basal activity than the wild-type GIRK1/4 (SI Appendix, Fig. S5 A and B); however, the co-expression of PKCε did not further enhance the activity of GIRK1/4(S418A). Similarly, the tau of current inhibition by the PIP₂ phosphatase of the wild-type GIRK1/4 was the shortest among all conditions, with the GIRK1/4(S418A) channel showing no change in the presence of PKCε (SI Appendix, Fig. S5 C and D). In the context of PKCε-mediated current augmentation of both GIRK1 and GIRK4 subunits alone, these data demonstrate that the effect of PKCε on GIRK4 predominates over that of GIRK1 and that Ser418 of GIRK4 is necessary and sufficient for the mediation of PKCε’s effects on the homomeric GIRK4 channel as well as the GIRK1/4 heteromer.

Discussion

The cellular redistribution of PKCε appears to be the antecedent to GIRK channel dysfunction. Oxidative stress mediated by H₂O₂ has been shown to induce persistent activation of PKC enzyme isoforms. This process is no longer dependent on Ca²⁺ or lipid cofactors through the phosphorylation of a tyrosine residue in their catalytic domain (24–27). Konishi et al. reported that H₂O₂ stimulated the nPKCs, δ and ɛ, to a greater extent compared to the conventional PKCs, α, βI, and γ (24). Membrane translocation of PKCs is widely recognized as a sign of enzyme activation (28). An increase in the membrane fraction of PKCε in AF, reported by Makary et al. (7) might indicate that there is enhanced activation of this enzyme in AF. Owing to reports of a constitutive increase in I_KACh in AF (7, 12), our biophysical findings suggest a shift in the equilibrium toward PKCε-phosphorylated GIRK channels possessing enhanced activity. The increased activity of the GIRK4(S418A)-containing GIRK1/4 heteromer suggests that the integrity of the serine at the penultimate position in the GIRK4 subunit is crucial to retaining physiological activity of the heteromeric channel. This result may suggest that phosphorylation of GIRK4(S418) removes an inhibitory effect of this residue on activity thus causing current stimulation. While these effects are elicited via the introduction of mutations in vitro, they also take place at the whole heart level through PKCε-mediated phosphorylation of the GIRK1/4 channel as seen in isolated hearts perfused with the PKCε activator, DCP-LA.

GIRK channel dysfunction and fibrosis are both considered a part of remodeling in the atria (29). The former contributes to the electrophysiological remodeling that leads to shortening of the effective refractory period, while the latter contributes to the structural remodeling which leads to slowing of conduction velocity. Electrophysiological and structural remodeling result in an atrial substrate that is conducive for the formation of reentry and thus to the perpetuation of AF (29). Therefore, the electrical and structural remodeling, in tandem, create a fibrillation-prone myocardium, driving AF maintenance. Remodeling of ionic currents other than I_KACh also contributes to the shortening of the action potential in chronic AF. I_K1 is another potassium current that is increased in chronic AF (30). The channel subunits responsible for this current are Kir2.1 and Kir2.2, also known as IRK1 and IRK2. PKCε is also reported to target K_ATP that plays a role in regulating the atrial membrane potential (30).

A recent study used young and old wild type and PKCɛ knockout mice to investigate whether aging leads to the constitutive activation of I_KACh and consequently to AF facilitation (31), since aging is an important risk factor for AF (32). Knocking out PKCɛ abrogated the effects of aging on AF by preventing the development of a constitutively active I_KACh current (31). Moreover, blocking this abnormal current in the old heart reduced AF inducibility (32). This work demonstrated that in the aging heart, I_KACh is constitutively active in a PKCɛ-dependent manner, contributing to the perpetuation of AF, and supported the contention that the I_KACh/PKCɛ axis is arrhythmogenic.

Atrial GIRK channels are crucial to the control of HRV, an important parameter that endows cardiac health and adaptability to an individual and is correlated with high vagal tone (33). GIRK4 or GIRK1 knockout mice show no HRV, while high GIRK activity, as in AF, induces excessive HRV, neither of which is a good cardiac health indicator (33–35). HRV and parasympathetic regulation of the heart allow the heart to quickly transpose intervals of high activity with periods of rest. HRV decreases with advancing age and is linked to increased fatality (36). Although complete inhibition of I_KACh could be helpful for AF in theory, it could decrease heart rate variability (HRV) and compromise cardiac health.

Several GIRK1/4 channel blockers have entered clinical trials as possible AF therapies but have failed to make it to the market. Due to the significant homology between the GIRK subunits, selective drug design has proven challenging. The nuance of GIRK1/4 and GIRK1/2 channel modulation is crucial as activation of the former results in a cardiac arrhythmia while inhibition of the latter can cause neuronal hyperexcitability and result in seizures. Recent advances on the small-molecule front include ML297, a potent activator of GIRK1-containing ion channels. While it has a greater effect on the GIRK1/2 heteromer, its utility is limited by its significant activation of the GIRK1/4 channel in the atria. The small-molecule, GAT1508, was designed based on the ML297 scaffold and is highly selective for the GIRK1/2 channel over GIRK1/4 (37). Multiple and diverse modulators of activity have been found to exert their effects by allosterically controlling channel–PIP₂ interactions (38). Recent computational work has enlightened us regarding the precise nature of these allosteric effects (37–39).

Here we have shown that phosphorylation of S418 enhances PKCε-mediated GIRK4 activity by allosterically strengthening channel–PIP₂ interactions. Given that S418 lies at the very C-terminal end of GIRK4, we imagine that its negative charge upon phosphorylation may electrostatically regulate channel–PIP₂ interactions, either allosterically or directly. A C-terminal peptide of the GIRK1 subunit was shown to directly block GIRK currents (40). High-resolution structures of the phosphorylated (or the phosphor-mimetic S418E) GIRK4 C-terminus, in the context of the full channel, will be needed before the molecular mechanism can be probed computationally.

Small molecules that can allosterically decrease pathological GIRK activity, while leaving its baseline functions intact, are highly desirable and can be aided by this work probing the mechanism behind GIRK1/4 overactivity. The lack of structural information imposes a significant hindrance in mapping possible binding pockets for targeted drug design. Small molecules that allosterically modulate channel–PIP₂ interactions could serve as effective tools for specifically inhibiting the GIRK1/4 channel and alleviating channel overactivity while keeping the basal function of the channel intact.

Materials and Methods

Reagents.

PMA and PKCε protein were purchased from Sigma-Aldrich. The selective PKCε activator, DCP-LA, was purchased from Fisher Scientific. Tertiapin-Q was purchased from Tocris.

Molecular Biology.

Transformation.

Competent Mach1-T1 Escherichia coli. were transformed with plasmid DNA and inoculated in LB broth overnight for amplification. Plasmids were extracted using a Qiagen Plasmid Prep kit following the manufacturer’s protocol.

Linearization.

Linearization of plasmid DNA was carried out using restriction enzymes indicated in parentheses. GIRK4_pXOOM (Pme1), GIRK4(S143T) pGEMHE (Nhe1), GIRK1_pGEMHE (Nhe1), M₂R_pGEMHE (Pst1), D₂R_pXOOM (Xho1), PKCε-CAT_pXOOM (Xho1), and IRK1_pGEMHE (Nhe1). Linearization of the plasmid was verified by gel electrophoresis. DNA was purified before transcription.

RNA transcription.

cRNA was transcribed in vitro using the mMESSAGE mMACHINE T7 Transcription Kit (Thermo Fisher Scientific) according to the manufacturer’s protocol. Point mutations in GIRK1 and GIRK4 were introduced using a standard Pfu DNA polymerase-based mutagenesis technique according to the QuikChange protocol (Agilent). Mutations were verified by sequencing.

Oocyte Preparation and Injection.

Oocytes from Xenopus laevis frogs were surgically isolated, dissociated, and defolliculated using collagenase according to standard protocols. Once defolliculated, oocytes were transferred to an oocyte Ringer’s solution supplemented with Ca²⁺ and Penicillin/Streptomycin antibiotics. Stage V or VI oocytes were selected and injected with 50 nL of cRNA resuspended in DEPC water. After injection, oocytes were incubated at 17 °C before TEVC experiments.

Two-electrode Voltage-Clamp (TEVC).

Borosilicate glass electrodes were pulled using a Flaming–Brown micropipette puller (Sutter Instruments) and filled with a 3M KCl solution containing 1% agarose. Pipette resistances were between 0.2 and 1.0 MΩ. Currents were recorded 1 to 3 d after injection using a GeneClamp 500 (Molecular Devices) amplifier. Oocytes were held at 0 mV, and currents were assessed by 100-ms ramps from −80 mV to +80 mV. At the start of each TEVC experiment, oocytes were perfused directly using a multi-barrel gravity-driven perfusion apparatus with a low K⁺ solution (2 mM KCl) before assessing the basal activity of the channel using a high K⁺ (HK) solution (96 mM KCl). Currents in HK were allowed to stabilize, and agonist (5 μM ACh) dissolved in HK was perfused into the recording chamber before applying 3 mM Ba²⁺ to establish barium-sensitive GIRK activity. Current amplitudes were measured at −80 mV. Experiments with the light-activated phosphatase to dephosphorylate PIP₂ used a system that was harnessed by De Camilli’s group (23). It is comprised of two fusion proteins: mCherry-CRY2-5-ptase_OCRL and CIBN-CAAX/CIBN-GFP. mCherry-CRY2-5-ptase_OCRL contains the photolyase domain of cryptochrome 2 (CRY2) and the inositol 5-phosphatase domain of the Lowe oculocerebrorenal syndrome protein (OCRL). CIBN-CAAX or CIBN-CAAX-GFP contains the CRY2-binding domain (CIBN) and a C-terminal CAAX box for membrane targeting. Illumination with blue light between 458 and 488 nm causes CRY2 to absorb a FAD molecule and undergo a conformational change which promotes its dimerization with the membrane-anchored CIBN. This localizes the CRY2-linked 5ptase to the plasma membrane, where it dephosphorylates the 5-phosphate of PIP₂. The 5ptase_OCRL system was activated using a 460λ LED (Luminus) that was focused on the cells through the objective lens of an inverted microscope (Nikon). Cells were studied in whole-cell mode, and basal GIRK activity was assessed in HK. After currents in HK stabilized, cells were illuminated with blue light with an LED at 5 mW power to activate the phosphatase. Current depletion after phosphatase activation was allowed to stabilize before the current was blocked with 5 mM Ba²⁺ in HK. Data for % current remaining were calculated using Ba-sensitive currents. Current traces after blue-light illumination were fitted to a one-phase decay equation to extract τ values to measure the kinetics of current depletion. GIRK channel cRNA was co-expressed with CRY2–5ptase_OCRL, and the membrane-anchore CIBN-CAAX. Oocytes in TEVC experiments were illuminated with 460λ after basal currents stabilized in high K⁺ perfusion. Stable currents after blue-light illumination were recorded and blocked with 3mM Ba²⁺.

Cell Culture and Transfection.

HEK293T cells were cultured in 10-cm culture dishes in DMEM media (Hyclone) supplemented with 1% Penicillin/Streptomycin and 10% Fetal Bovine Serum and maintained in 5% CO₂ at 37 °C. Cells were seeded on glass coverslips in 35-mm culture dishes at least 1 d before transfection. Cells were transiently transfected in OptiMEM using polyethyleneimine (PEI) for 2 to 6 h at ratios ranging from 1 μg of DNA:8 μL PEI to 1 μg of DNA:4 μL PEI. Cells were studied 24 to 48 h after transfection.

Whole-Cell Patch Clamp.

Whole-cell currents were recorded using an Axopatch 200B amplifier (Molecular Devices) or Tecella using WinWCP software (University of Strathclyde). Currents were filtered through a lowpass Bessel filter at 2 kHz and were digitized at 10 kHz. Borosilicate glass electrodes were pulled using a vertical puller (Narishige) and had a resistance of 2 to 5 MΩ when filled with an intracellular buffer comprised of: 140 mM KCl, 2 mM MgCl₂, 1 mM EGTA, 5 mM Na₂ATP, 0.1 mM Na₂GTP, and 5 mM HEPES buffered to pH 7.4 using KOH. Electrophysiology experiments were performed at room temperature. Cell capacitances ranged from 8 to 15 pF. Cells were co-transfected with GFP, and GFP-expressing cells were selected for analysis using a Nikon epifluorescence microscope. Cells were held at 0 mV, and currents were recorded using a repeating ramp protocol from −80 mV to +80 mV in whole-cell mode. Cells were perfused directly using a multi-barrel gravity-driven perfusion apparatus. Initially, in a physiological buffer containing 135 mM NaCl, 5 mM KCl, 1.2 mM MgCl₂, 1.5 mM CaCl₂, 8 mM glucose, and 10 mM HEPES; pH 7.4, GFP+ cells formed a Giga-Ω seal with the patch pipette. Slight suction was applied to the cells to enter whole-cell mode. GIRK channel activity was measured at −80 mV after transitioning to a high K⁺ buffer comprised of 5 mM NaCl, 140 mM KCl, 1.2 mM MgCl₂, 1.5 mM, CaCl₂, 8 mM glucose, and 10 mM HEPES; pH 7.4. Currents were blocked using 5 mM Ba²⁺ in high K⁺. For optogenetic experiments, whole-cell currents were allowed to stabilize in HK before illuminating the cell with blue light.

Human-Induced Atrial iPS Cells.

Human-induced atrial iPS cells were a gift from Axol Biosciences (ax2518). Cells were thawed, plated, and cultured according to the manufacturer’s specifications. Cells were grown at 37 °C under 5% CO₂. Culture medium was replaced every other day per the manufacturer’s protocol. Cells were validated as being atrial myocytes by eliciting tertiapin-Q-sensitive GIRK channel currents under voltage clamp. Beating cells were selected for analysis in patch-clamp experiments using an external solution in mM: NaCl 140, KCl 5.4, MgCl₂ 1, CaCl₂, 1.8, and glucose 5.5 HEPES 5. Internal solution contained in mM: K⁺ Gluconate 125, KCl 20, NaCl 5, and HEPES 5. Action potentials were elicited using the perforated patch technique with 0.22 μM Amphotericin-B. Basal channel activity in voltage clamp was elicited using a solution containing in mM: NaCl 140, KCl 50, MgCl₂ 1, CaCl₂, 1.8, glucose 5.5 HEPES 5, and Nifedipine 0.01.

Optical Mapping in Isolated Hearts.

Isolated hearts of C57BL/6 male and female mice, 4 to 6 mo of age, were retrogradely perfused with Tyrode’s solution in Langendorff mode. The preparations were maintained at 37 °C and stained with a bolus of voltage-sensitive dye (0.25 mL, 10 μM Di-4-ANEPPS, molecular probes) and imaged with a CCD camera (RedShirt Imaging), 80 × 80 pixels, 85 µm per pixel, and 1,000 frames per second (41). Excitation contraction uncoupling was achieved with 7 μM Blebbistatin (Tocris Bioscience). A bipolar, silver tip stimulation electrode was used to pace the right atrium (2.5 ms pulses, 2× diastolic threshold) at 9 Hz using an AD Instrument stimulation platform. Pretreatment with DCP-LA at 10 μM or DMSO vehicle control was done through the perfusate for 25 min, after which 1 mL of 200 nM tertiapin-Q was administered through an injection port. The APD at 60% repolarization (APD₆₀) was quantified as we have done extensively (42) after DCP-LA or DMSO, and then after tertiapin-Q.

Statistics.

All error bars represent the SD. Statistical significance was assessed using Student’s unpaired t test or ANOVA in GraphPad Prism. Statistical significance was set at P < 0.05. Data are mean ± SD. Effect sizes were adjusted to provide power of 0.8 or greater for all conclusions drawn. For τ values, curves were fitted to the mono-exponential, standard Boltzmann equation to extract information on the kinetics of current inhibition.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data required to support the conclusions of this paper are included in the article and/or SI Appendix.