Neuronal G protein-gated K⁺ channels

Abstract

G protein-gated inwardly rectifying K⁺ (GIRK/Kir3) channels exert a critical inhibitory influence on neurons. Neuronal GIRK channels mediate the G protein-dependent, direct/postsynaptic inhibitory effect of many neurotransmitters including γ-aminobutyric acid (GABA), serotonin, dopamine, adenosine, somatostatin, and enkephalin. In addition to their complex regulation by G proteins, neuronal GIRK channel activity is sensitive to phosphatidylinositol 4,5-bisphosphate (PIP₂), phosphorylation, regulator of G protein signaling (RGS) proteins, intracellular Na⁺ and Ca²⁺, and cholesterol. The application of genetic and viral manipulations in rodent models, together with recent progress in the development of GIRK channel modulators, has increased our understanding of the physiological and behavioral impact of neuronal GIRK channels. Work in rodent models has also revealed that neuronal GIRK channel activity is modified, transiently or persistently, by various stimuli including exposure drugs of abuse, changes in neuronal activity patterns, and aversive experience. A growing body of preclinical and clinical evidence suggests that dysregulation of GIRK channel activity contributes to neurological diseases and disorders. The primary goals of this review are to highlight fundamental principles of neuronal GIRK channel biology, mechanisms of GIRK channel regulation and plasticity, the nascent landscape of GIRK channel pharmacology, and the potential relevance of GIRK channels to the pathophysiology and treatment of neurological diseases and disorders.

INTRODUCTION

G protein-gated inwardly rectifying K⁺ (GIRK/Kir3) channels are critical mediators of neuronal inhibition in the central nervous system (1–3). GIRK channels are best known as key mediators of the G protein-dependent postsynaptic inhibitory effect of many neurotransmitters, including the most abundant mammalian inhibitory neurotransmitter γ-aminobutyric acid (GABA) (4–6). GIRK activation by inhibitory neurotransmitters hyperpolarizes neurons and limits N-methyl-d-aspartate (NMDA) receptor-dependent activation (7). GIRK channels also contribute to neuron autoinhibition and network inhibition. In dopamine neurons of the substantia nigra (SN), for example, GIRK channels coupled to D₂ dopamine receptors (D₂R) mediate the autoinhibitory influence of activity-induced dopamine release (8). At the network level, GIRK channel activation shapes the oscillation behavior of thalamic and hippocampal networks (9–11).

GIRK channels belong to the family of K⁺ channels that exhibit inward rectification (12, 13). Members of this “Kir” family conduct K⁺ more effectively in the inward direction (into the cell) than in the physiologically relevant outward direction (out of the cell). This property is conferred by blockade of K⁺ efflux via intracellular Mg²⁺ and small polyamines. Driven largely by channel crystallization efforts, our understanding of the structural basis of GIRK channel function, including insights related to inward rectification, channel gating, and regulation by G proteins and other interactants has increased dramatically (14–21). These achievements are documented in relevant publications and previous reviews (13, 22). The goal here is to highlight fundamental and emerging principles of neuronal GIRK channel biology, with an emphasis on their regulation and pharmacology, and the relevance and therapeutic potential of GIRK channels in the context of neurological disorders and diseases.

GIRK Channel Subunit Composition

GIRK channels are found in the heart and brain, along with a limited number of other tissues including the pancreas (23–25) and pituitary gland (26). Four mammalian genes encode GIRK channel subunits: KCNJ3/GIRK1/Kir3.1, KCNJ6/GIRK2/Kir3.2, KCNJ9/GIRK3/Kir3.3, and KCNJ5/GIRK4/Kir3.4. Protein biochemistry efforts involving bovine heart tissue (27, 28), studies with linked/concatemeric subunits in expression systems (29, 30), and crystal structures of the purified recombinant GIRK2 homomeric channel (18, 20) support the contention that GIRK channels are homo- and heterotetrameric complexes. GIRK channel subunit composition and stoichiometry are best understood in the heart, where GIRK1 and GIRK4 coassemble in 1:1 stoichiometry to form the muscarinic-gated K⁺ channel known as I_KACh (27, 31). Even in this relatively simple case where only GIRK1 and GIRK4 are present, the possibility that GIRK1:GIRK4 channel subtypes of variable stoichiometries (1:3, 2:2, 3:1) coexist has been neither formally proven nor discounted. Moreover, functional GIRK4 homomers have been identified in atrial myocytes (28, 32, 33).

GIRK channel subunit composition is a more complex issue in the brain (5). GIRK channels containing GIRK1 and GIRK2 are often taken as the prototypical neuronal GIRK channel, a contention based on the broad overlapping distributions of GIRK1 and GIRK2 in the rodent brain (34, 35), the downregulation of GIRK1 in the Girk2^–/– mouse brain (36), and the results of electrophysiological investigations in GIRK subunit knockout mice. Indeed, ablation of either GIRK1 or GIRK2 in mice correlates with the complete or near-complete loss of GIRK channel activity in almost all neuron populations evaluated to date (4, 35, 37–40). GIRK3 overlaps with GIRK1 and GIRK2 throughout the rodent brain (34, 35), however, and it can form functional channels with GIRK1 or GIRK2 (41–44). GIRK4 is also found in the rodent brain, though its expression is restricted (34, 45). Alternative splicing of GIRK subunits has also been documented (46–55), and this yields multiple distinct protein variants in the case of GIRK1 and GIRK2. Midbrain dopamine neurons warrant special mention in this conversation, as they do not appear to express GIRK1 (34). The GIRK channel in rat substantia nigra dopamine neurons is formed by two GIRK2 splice variants (GIRK2a and GIRK2c) (51), whereas the GIRK channel in mouse ventral tegmental area (VTA) dopamine neurons is formed by GIRK2c and GIRK3 (43, 44). Thus, although GIRK1 and GIRK2 likely contribute to GIRK channel formation throughout the rodent CNS, there are exceptions to this rule and much remains to be learned about the precise subunit composition of neuronal GIRK channels, even in well-studied rodent models. Importantly, virtually nothing is known regarding GIRK channel subunit expression patterns, alternative splicing, and subtypes in the human brain.

Key Features of Neuronal GIRK Channel Subunits

GIRK subunits contain two transmembrane segments, a hydrophobic pore domain, and cytoplasmic amino- and carboxyl-terminal domains. The cytoplasmic domains are most divergent across the GIRK channel subunits, harboring elements that impact channel trafficking and function, and mediate protein:protein interactions. Brief descriptions of the three primary neuronal GIRK channel subunits (GIRK1, GIRK2, and GIRK3) are detailed in the following sections.

GIRK1.

GIRK1 cannot form functional homomers (31), a deficit linked to the lack of an endoplasmic reticulum (ER) export signal (52). A single-point mutation in the GIRK1 pore (F137S), however, permits surface trafficking of GIRK1 and formation of functional GIRK1 homomultimers (56). This suggests that additional structures and mechanisms supporting the surface trafficking of GIRK channels remain to be identified. Normally, GIRK1 associates with another GIRK subunit (e.g., GIRK2) to achieve surface membrane localization (57, 58). In expression systems, coexpression of GIRK1 yields substantially larger GIRK channel currents than seen with expression of GIRK2 or GIRK4 alone (31, 47, 59, 60). Relatedly, genetic ablation of GIRK1 in mice correlates with small GIRK currents in neurons, despite normal levels of residual GIRK subunit expression (61). Unique residues in the GIRK1 pore region that enhance single-channel conductance and increase mean open time, along with unique elements in the carboxyl-terminal domain, have been implicated in ability of GIRK1 to augment GIRK channel activity (61–64).

GIRK2.

GIRK2 (and GIRK4) can form functional homomeric GIRK channels (47, 52, 59), attributable at least in part to multiple forward trafficking elements, including an ER export signal and post-Golgi surface-promoting motifs (52). Alternative splicing generates multiple distinct GIRK2 variants, two of which (GIRK2a and GIRK2c) are expressed throughout the mouse brain and differ by only 11 amino acids at the carboxyl terminus (50). The longer variant (GIRK2c) possesses a PDZ domain interaction motif (ESKV). When expressed individually in CA1 pyramidal neurons in hippocampal neurons from Girk2^–/– mice, the subcellular distributions of GIRK2a and GIRK2c were overlapping but distinct; both subunits were detected in the soma and dendrites, but GIRK2c was found at higher levels that GIRK2a in secondary and tertiary dendrites (65). Although no differences in GPCR-dependent GIRK currents were seen in Girk2^–/– neurons expressing either GIRK2a or GIRK2c, the amplitude of synaptic GABA_BR-GIRK responses evoked by electrical stimulation of proximal (Schaffer collateral) or distal (perforant path) dendritic fields was tracked with the differential dendritic distribution of GIRK2a and GIRK2c.

GIRK3.

Despite its widespread distribution in the rodent brain (34, 35), ablation of GIRK3 in mice yields modest influence on GIRK channel function (35, 38, 39, 66, 67). Nevertheless, several lines of evidence suggest that GIRK3 contributes to neuronal GIRK channel formation. For example, simultaneous ablation of GIRK2 and GIRK3 in mice correlates with a larger reduction in GIRK channel activity in neurons of the locus ceruleus (38) and hippocampus (35) than loss of GIRK2 alone. In addition, the residual GIRK channel (GIRK2c homomer, presumably) in VTA dopamine neurons from Girk3^–/– mice is more sensitive to GABA_BR activation than its wild-type (GIRK2c:GIRK3) counterpart (44). Furthermore, GIRK3 protein levels are decreased in the hippocampus and cerebellum of Girk1^–/– and Girk2^–/– mice (35, 68), and GIRK3 co-immunoprecipitates with GIRK1 and GIRK2 in the mouse cerebellum (68). Although genetic ablation of GIRK3 is not associated with overt phenotypes (38, 69), Girk3^–/– mice show altered sensitivity to drugs of abuse, several classes of analgesic compounds, and sedative hypnotics (40, 70–72), and they are deficient in some forms of GIRK channel plasticity (73, 74).

Like GIRK1, GIRK3 lacks an ER export signal and does not form functional homomers (52). Surprisingly, however, functional GIRK1:GIRK3 heteromers have been reported in expression systems (41, 75, 76), again suggesting that the surface localization of GIRK channels does not depend exclusively on known forward trafficking elements. GIRK3 does harbor a lysosomal targeting sequence (YWSI) that is absent in other subunits, which likely explains why ectopic expression of GIRK3 can suppress GIRK channel activity (52, 59, 77). Although mechanisms that dictate how and when GIRK3 plays channel-forming and/or channel-trafficking roles are unclear, the balance of evidence suggests that GIRK3 titrates GIRK channel activity in neurons.

GIRK CHANNEL REGULATION

The regulation of GIRK channel activity by G proteins is fascinating and complex (6, 78). GIRK channels are activated by inhibitory (G_i/o) G proteins. The G_i/o G protein family includes three Gα_i isoforms, two alternative splice variants of Gα_o, and Gα_z (79–81). Gα_i/o subunits assemble with Gβγ complexes formed by various combinations of four Gβ isoforms and 12 Gγ subunits (82, 83). GIRK channel gating is enhanced by direct binding of Gβγ (84–87) (Fig. 1A). GIRK channels can bind to four Gβγ subunits (20), and GIRK channel activity tracks with Gβγ occupancy of these sites (30, 88). GIRK channels can also bind to Gα_i/o, in either GDP- or GTP-bound states (6, 89–94). This interaction likely contributes to the strong functional connection between GIRK channels and G_i/o-linked GPCRs in vivo (94). Efforts in expression systems suggest that Gα_i/o isoforms exert unique allosteric influence on GPCR-GIRK signaling and basal activity (6). For example, GPCR-GIRK signaling mediated by Gα_o exhibits faster activation rates than seen with Gα_i (95). In addition, Gα_i isoforms enhance GPCR-GIRK coupling sensitivity (6, 89, 90, 93, 96–98).

Figure 1.

Depictions of G protein-gated inwardly rectifying K⁺ (GIRK) channel biology, signaling, and regulation. A: the activity of neuronal GIRK channels (e.g., the GIRK1:GIRK2 or GIRK1/2 heterotetramer) is regulated by GPCRs coupled to inhibitory G proteins (G_i/o) G proteins (G_i/o GPCRs). GIRK channel gating is enhanced by direct binding of Gβγ, which stabilizes the interaction between the channel and phosphatidylinositol 4,5-bisphosphate (PIP₂). Intracellular Na⁺ can also facilitate GIRK channel opening. Regulator of G protein signaling (RGS) proteins influence the kinetics and/or sensitivity of GPCR-GIRK signaling. In addition to Gβγ, GIRK channels can also bind to Gα_i/o, GPCRs, and RGS proteins, supporting the contention that some or many GPCR-GIRK signaling elements exist in pre-coupled states (right). B: regulation of GIRK channel activity by G_q-coupled GPCRs. Activation of G_q-coupled GPCRs leads to phospholipase C (PLC) activation, which converts PIP₂ into diacylglycerol (DAG) and inositol-1,4,5-triphosphate (IP₃). DAG activates protein kinase C (PKC), whereas IP₃ releases Ca²⁺ from internal stores (endoplasmic reticulum, ER). PIP₂ depletion, PKC activation, and increased intracellular Ca²⁺ have all been shown to suppress GIRK channel activity. Phospholipase A₂ (PLA₂) activation and arachidonic acid (AA) production may also mediate the G_q-mediated suppression of GIRK channel activity in some cases. C: schematic depiction of GIRK1-lacking GIRK channels, as found in dopamine neurons of the substantia nigra (GIRK2 homomer, left) and ventral tegmental area (GIRK2c:GIRK3 heteromer, right). GIRK3 interacts with proteins that regulate sensitivity to GPCR activation (RGS2) and channel trafficking (sorting nexin 27, or SNX27). SNX27 interacts with the PDZ interaction domain in GIRK3, targeting GIRK3-containing channels to early endosomes (EE), leading to reduced surface expression of GIRK channels by facilitating their degradation in late endosomes/lysosomes (LE-Ly), rather than recycling to the surface membrane via recycling endosomes (RE). Created with BioRender.com.

Little is known regarding the Gα and Gβγ isoforms that mediate GIRK-dependent signaling in neurons. Although the loss of Gα_o in mice prolonged the deactivation kinetics of GPCR-GIRK signaling in CA3 pyramidal neurons (99), GPCR-GIRK response amplitude was preserved. This suggests that multiple Gα_i/o isoforms shape neuronal GIRK-dependent signaling. Loss of Gγ₃ in mice correlated with a reduction in the number of neurons exhibiting a GABA_BR-GIRK current, and smaller responses in those exhibiting a response (100). Loss of Gγ₃ also correlated with reduced Gα_i3 and Gβ₂ subunit protein levels (101), suggesting that these G protein isoforms mediate GABA_BR-GIRK signaling in some neurons. Interestingly, GIRK-dependent responses to adenosine and somatostatin were preserved in Gγ₃^–/– mice, supporting the contention that different GPCR-GIRK signaling pathways in neurons are composed of distinct signaling elements.

Neuronal GIRK channels exhibit basal/tonic activity (6). Acute inhibition of neuronal GIRK channels, for example, depolarizes neurons and induces epileptiform activity in mouse hippocampal cultures (102). Basal/tonic GIRK channel activity can reflect GPCR-dependent and/or GPCR-independent channel activation. Indeed, the relatively high level of constitutive GIRK channel activity seen in dorsal (but not ventral) rat hippocampal CA1 pyramidal neurons is due to ambient adenosine and tonic activation of GIRK channel by the A₁ adenosine receptor (A₁R) (103, 104). Basal/tonic GIRK channel activity can also result from intrinsic channel gating and basal G protein turnover (105, 106), or perhaps from direct activation by other regulatory factors.

Phosphatidylinositol 4,5-Bisphosphate

Phosphatidylinositol 4,5-bisphosphate (PIP₂) is a requisite cofactor for GIRK channel regulation by Gβγ (78, 107, 108). Gβγ activates GIRK channels by stabilizing the interaction between GIRK channels and PIP₂. The structural basis of PIP₂ binding to and regulation of GIRK channels has been illuminated with the crystal structure of the GIRK2 homomer (18), along with cryo-EM (109) and molecular dynamics simulations (110, 111).

PIP₂ depletion has been implicated in the suppression of cardiac GIRK channel activity by G_q-coupled GPCRs (112, 113). Although stimulation of G_q-coupled GPCRs (e.g., muscarinic acetylcholine and group 1 metabotropic glutamate receptors, mGluR) suppressed GABA_BR-GIRK signaling in rodent CA1 pyramidal neurons (114), the muscarinic receptor-dependent suppression of GABA_BR-GIRK signaling involved activation of phospholipase C (PLC) and PKC, whereas the group 1 mGluR influence involved activation of phospholipase A₂ (PLA₂) and production of arachidonic acid. Despite these mechanistic differences, both types of inhibition culminated in the disruption of GIRK channel:PIP₂ interactions (115). Interestingly, the group 1 mGluR-mediated suppression of GABA_BR-GIRK and D₂R-GIRK signaling pathways in rat substantia nigra dopamine neurons was unaffected by PLC, PKC, or PLA₂ inhibitors. Rather, inhibition of GIRK-dependent signaling in this case involved Ca²⁺ release from intracellular stores (116). Thus, G_q-coupled GPCRs may negatively regulate neuronal GIRK channels via multiple mechanisms including PIP₂ depletion, PKC activation/phosphorylation, PLA₂ activation and arachidonic acid production, and/or increased intracellular Ca²⁺ (Fig. 1B).

Phosphorylation

Phosphorylation has been implicated in GIRK channel biology (117–124), but little is known about kinases and substrates relevant in vivo. Phosphorylation of GIRK1 at an undetermined, protein phosphatase 2A (PP2A)-sensitive site is required for GIRK channel activation by Gβγ (119). The NMDA receptor (NMDAR)- and protein phosphatase 1 (PP1)-dependent dephosphorylation of serine-9 on GIRK2 [GIRK2(S9)] increases the surface trafficking of GIRK channels in rat hippocampal neurons (121). This process has been implicated in the depotentiation of long-term potentiation (LTP) (125). Although kinase(s) targeting GIRK2(S9) or other known or predicted phosphorylation sites on GIRK channels have not been identified, work across multiple models suggests that GIRK channels can be phosphorylated by several serine-threonine protein kinases, including PKA, PKC, and CaMKII (119, 120, 122, 123, 126, 127). Work in Xenopus oocytes suggests that tyrosine phosphorylation of GIRK channels can also regulate GIRK channel trafficking (128) or suppress GIRK channel activity (118, 129).

PKC can phosphorylate recombinant GIRK1 and GIRK2, and PKC phosphorylation suppresses GIRK channel activity (119, 120, 123, 124, 130–132). Some (but not all) of the suppression of GIRK channel activity evoked by G_q-coupled GPCRs can be mimicked by PKC activators (114, 115, 123, 130, 133), or abolished using PKC inhibitors (134). Notably, a phospho-dead variant of GIRK1(S185) diminished the negative influence of PKC on channel function, and precluded the negative regulation of GIRK channel activity by G_q-coupled GPCRs (120). As such, PKC phosphorylation likely contributes to the negative regulation of GIRK channels by G_q-coupled GPCRs (123, 135).

Regulator of G Protein Signaling Proteins

Regulator of G protein signaling (RGS) proteins negatively regulate G protein-dependent signaling (136, 137). RGS proteins enhance the intrinsic GTP hydrolysis rate of Gα, accelerating re-formation of inactive G protein heterotrimers (138). Evidence for neuronal GIRK channel regulation by RGS proteins comes primarily from studies involving two RGS subfamilies:

R7 RGS proteins.

Members of the R7 subfamily of RGS proteins (RGS6, RGS7, RGS9, and RGS11) contain a “Gγ-like” domain that promotes stable dimerization with Gβ5, the atypical/nonsignaling member of the Gβ subunit family (139). R7 RGS/Gβ5 complexes exhibit strong substrate selectivity for Gα_o (140, 141), and they are structurally similar to Gβγ (142). The structural similarity with Gβγ likely explains the ability of R7 RGS/Gβ5 complexes to bind GIRK channels (143, 144). Although the RGS7/Gβ5:GIRK channel interaction has been modeled (145), it remains unclear whether R7 RGS/Gβ5 complexes compete with Gβγ for access to the Gβγ binding sites on GIRK channels (20). As occupancy of these sites by Gβγ determines GIRK channel activity and GPCR-GIRK coupling strength (30, 88), resolving this issue is paramount to understanding the regulatory influence of R7 RGS/Gβ5 complexes on GIRK-dependent signaling. What is clear is that R7 RGS/Gβ5 complexes negatively regulate GPCR-GIRK signaling in mouse cardiomyocytes (144, 146) and neurons (143). For example, RGS7 or Gβ5 ablation in mice is correlated with prolonged synaptic GABA_BR-GIRK currents in CA1 pyramidal neurons (143, 147). GABA_BR-GIRK currents exhibited slower deactivation rates and increased GABA_BR-GIRK coupling sensitivity in cultured hippocampal neurons from RGS7^–/– or Gβ5^–/– mice (143, 148).

Adaptors for R7 RGS/Gβ5 complexes have also been identified, including the small palmitoylated protein R7BP and GPR158 (139, 149). Adaptor binding confers stability to and membrane association of R7 RGS/Gβ5 complexes, while enhancing their catalytic activity (139, 149, 150). Interestingly, RGS7 forms mutually exclusive complexes with R7BP or GPR158 in the mouse hippocampus (147). Whereas loss of R7BP diminished the regulatory influence of RGS7/Gβ5 on GABA_BR-GIRK signaling in mouse hippocampal pyramidal neurons (148), there was no impact of GPR158 ablation (147). GPR158 overexpression, however, suppressed the negative regulatory influence of RGS7/Gβ5 on GABA_BR-GIRK signaling. Thus, adaptors may be able to direct the influence of R7 RGS/Gβ5 complexes toward (R7BP) or away (GPR158) from GIRK channels in neurons.

RGS6/Gβ5 is the dominant RGS influence on cardiac GIRK-dependent signaling in the mouse (144, 146, 151), and RGS6/Gβ5 negatively regulates GABA_BR-GIRK signaling in mouse cerebellar granule cells (152). Since rodent hippocampal pyramidal neurons express RGS6 (153), it is surprising that loss of RGS6 has no impact on GABA_BR-GIRK signaling in mouse hippocampal pyramidal neurons (147, 148). In light of the finding that RGS6 negatively regulates the 5HT_1AR-dependent inhibition of adenylyl cyclase in mouse hippocampal pyramidal neurons (154), it is tempting to speculate that RGS7/Gβ5/R7BP (GIRK channels) and RGS6/Gβ5/GPR158 (adenylyl cyclase) exert negative regulatory influence on discrete branches of inhibitory GPCR signaling pathways.

R4 RGS proteins.

Two members of the R4 RGS protein subfamily—RGS2 and RGS4—have been implicated in GIRK channel regulation. The suppression of GIRK channel activity by G_q-coupled purinergic receptors was diminished by ectopic expression of RGS2 in rat sympathetic neurons (155). RGS2 is a negative regulator of GABA_BR-GIRK signaling in mouse VTA dopamine neurons (44). Ablation of RGS2 or GIRK3 in mice increased GABA_BR-GIRK coupling sensitivity in VTA dopamine neurons. There was no additive influence of RGS2 ablation on GABA_BR-GIRK coupling sensitivity in VTA dopamine neurons from Girk3^–/– mice, suggesting that GIRK3 mediates the influence of RGS2 in these neurons (Fig. 1C).

RGS4 exhibits dual specificity for Gα_i/o and Gα_q (156). RGS4 interacts with GABA_BR (157, 158), and has been shown to negatively regulate GIRK-dependent signaling in expression systems (157–163). Simultaneous genetic ablation of RGS4 and RGS6 in mice revealed an influence of RGS4 on GPCR-GIRK signaling dynamics in cardiomyocytes (151). In this setting, however, RGS4 opposed the more dominant negative regulatory influence of RGS6/Gβ5 on GPCR-GIRK signaling dynamics. This oppositional influence of RGS4 ablation on GIRK-dependent signaling in cardiomyocytes may reflect enhanced Gα_q activity and depletion of PIP₂ (107, 108).

Other GIRK Channel Regulators

Several other modes and mechanisms of GIRK channel regulation have been described. For example, GIRK2- and GIRK4-containing GIRK channels are activated by elevated intracellular Na⁺ (EC₅₀ = 30–40 mM) (47, 164). This regulatory mode may serve as a break on excessive neuronal activity, as intracellular Na⁺ entry into the cell would ultimately trigger GIRK channel activation. Like Gβγ, Na⁺ increases the affinity of GIRK channels for PIP₂ (108). The binding site for Na⁺ has been resolved in the GIRK2 homomultimer (18), and Na⁺-dependent gating of GIRK channels has been modeled by molecular dynamics simulation (111). GIRK channels are also regulated by cholesterol (165), a mode of regulation suggested by early reports of channel association with cholesterol-rich lipid rafts (35, 166). GIRK channel activity is enhanced by increases in membrane cholesterol (167–170). A diet-based increase of cholesterol levels increased GIRK channel activity in rodent hippocampal neurons, whereas cholesterol depletion or statin treatment decreased or normalized GIRK channel currents (168, 171). Recent cryo-EM studies, along with molecular simulation and mutagenesis efforts, have yielded important insights into the cholesterol-binding site in GIRK channels, and mechanisms of GIRK channel gating by cholesterol (168, 172).

Is There a GIRK “Signalosome”?

As indicated earlier, GIRK channels can bind to a number of signaling elements, including Gα, Gβγ, and R7 RGS proteins. These and other data have fueled speculation that macromolecular complexes coordinate selective and efficient GPCR-GIRK signaling in neurons (173, 174). Some evidence suggests that GIRK channels bind to GABA_BR (157, 175). For example, GABA_BR is coimmunoprecipitated with GIRK channels in expression systems and in vivo (176, 177). Comprehensive proteomic analysis of GABA_BR interacting proteins, however, suggests that this interaction is indirect and/or weak (178).

Several plausible signaling elements could facilitate formation of a precoupled GABA_BR-GIRK signaling pathway, or “signalosome.” For example, given its association with GABA_BR (157, 179, 180) and scaffolding with GIRK channels (6), Gα could mediate a GABA_BR:GIRK interaction. Gβ5 is part of the GABA_BR interactome in the rodent brain (181), and since RGS7/Gβ5 binds to GIRK channels (143), RGS7/Gβ5 may serve as the bridge. Potassium (K⁺) channel tetramerization domain (KCTD) proteins are core elements of the GABA_BR interactome and they interact with Gβγ (182). Interestingly, the effect of KCTD elimination in mice on GABA_BR-GIRK signaling is reminiscent of loss of RGS control (183), suggesting that one or more KCTD family members may participate in GABA_BR-GIRK signalosome assembly in neurons. Despite evidence supporting the existence of GIRK signalosomes, it is important to note that the functional properties of reconstituted GPCR-GIRK signaling pathways (including the GABA_BR-GIRK signaling pathway), exhibit functional properties consistent with “collision coupling,” suggesting that at least one untethered or diffusible element is involved (184, 185).

GIRK CHANNEL PLASTICITY

Several examples of GIRK channel plasticity in rodent models have been described (Table 1), including an early demonstration that the same stimulus that evoked LTP in rat hippocampal pyramidal neurons also potentiated the synaptic GABA_BR-GIRK inhibitory postsynaptic current (186). Subsequently, evidence for GIRK channel plasticity in rodent models has been reported following exposure to drugs of abuse [e.g., opioids (65, 191, 192), psychostimulants (39, 74, 187, 188, 190), alcohol (193,194, 200)], neuronal activity (73, 121), aversive experience (195,196, 201), and amyloid beta (Aβ) (197–199, 202). Studies of these forms of plasticity have shed light on underlying mechanisms, and emerging themes are discussed in the following sections.

Table 1.

Plasticity of neuronal GIRK channel activity

Stimulus	Brain Region/Neuron	Plasticity	Mechanism	Refs.
Neuronal activity
NMDAR activation	HPC culture	↑ A1R-GIRK, GABABR-GIRK	Dephosphorylation/GIRK2 surface trafficking	(121, 125)
HFS	HPC slice culture	↑ GABABR-GIRK	NMDAR, Ca2+/CaMKII	(186)
Burst/depolarization	VTA/dopamine	↑ GABABR-GIRK, D2R-GIRK	NMDAR, Ca2+/CaMKII, GIRK3	(73)
Tonic/LFS	VTA/dopamine	↓ GABABR-GIRK	NMDAR, GIRK3
Drugs of abuse
Cocaine
Single injection	VTA/dopamine	↓ GABABR-GIRK, D2R-GIRK	Internalization, D2/3 R-dependent	(187)
Repeated injection	VTA/dopamine	↓ GABABR-GIRK
Single injection	VTA/GABA	↓ GABABR-GIRK		(188)
	VTA/dopamine	- GABABR-GIRK
Repeated injection	PL/layer 5/6 pyramidal	↓ GABABR-GIRK	Dephosphorylation/GABABR2 internalization	(39)
			D1/5R-dependent
	PL/layer 2/3 pyramidal	- GABABR-GIRK
	IL/layer 5/6 pyramidal	- GABABR-GIRK
Repeated injection	PL/GABA	- GABABR-GIRK		(189)
Methamphetamine
Single injection	VTA/GABA	↓ GABABR-GIRK	Dephosphorylation/GABABR2 internalization	(188)
	VTA/dopamine	↓ GABABR-GIRK
Single injection	VTA/GABA	↓ GABABR-GIRK	Dephosphorylation/GABABR2	(74)
Repeated injection	VTA/dopamine	↓ GABABR-GIRK, - D2R-GIRK	GIRK3, D1R/D2R-dependent
Self-administration	Midbrain/dopamine	↓ GABABR-GIRK, D2R-GIRK	Ca2+	(190)
Repeated injection	VTA/dopamine	↓ D2R-GIRK
Opioids
Morphine	HPC culture	↑ 5HT1AR-GIRK ↓ GABABR-GIRK	↑ GIRK2, Ca2+/CaMKII, trafficking	(65, 191)
Remifentanil	PL/layer 5/6 pyramidal	↓ GABABR-GIRK		(192)
Ethanol
Repeated injection	VTA/dopamine	↑ D2R-GIRK, - GABABR-GIRK		(193)
CIE/ethanol vapor	lateral OFC/pyramidal	↓ GPCR-GIRK, ML297-GIRK		(194)
Aversive experience
Forced swim	DRN/serotonin	↓ KOR-GIRK	p38 MAPK, GIRK1(Y12)	(195)
Foot-shock	LHb	↓ GABABR-GIRK	PP2A dephosphorylation internalization	(196)
Predator odor	LHb	↓ GABABR-GIRK
Restraint	LHb	↓ GABABR-GIRK
Foot-shock	LHb	↓ GABABR-GIRK
CUS	PL/layer 5/6 pyramidal	↓ GABABR-GIRK, ML297-GIRK
Amyloid beta (Aβ)
Aβ25–35	HPC/CA3 pyramidal	↓ GABABR-GIRK	↓ GIRK expression	(197, 198)
Aβ1–42	HPC culture	↑ GABABR-GIRK	NMDAR, trafficking, p75NTR	(199)

Examples of neuronal GIRK channel plasticity provoked by various stimuli in rodent models. CUS, chronic unpredictable stress; DRN, dorsal raphe nucleus; GABA, γ-aminobutyric acid; GIRK, G protein-gated inwardly rectifying K⁺; HFS, high-frequency stimulation; HPC, hippocampus; KOR, kappa opioid receptor; LFS, low-frequency stimulation; LHb, lateral habenula; MAPK, mitogen-activated protein kinase; NMDAR, N-methyl-d-aspartate receptor; OFC, orbitofrontal; PL, prelimbic; VTA, ventral tegmental area. Directional arrows, direction of plasticity: up arrows, strengthening; down arrows, weakening.

Subcellular Trafficking

Several forms of GIRK channel plasticity involve increased or decreased trafficking of GIRK channels to the surface membrane. For example, removal of an NMDAR antagonist following long-term incubation in cultured rat hippocampal neurons rapidly increased the surface trafficking of GIRK1 and GIRK2 (121). The increased surface trafficking required activation of protein phosphatase-1 (PP1) and subsequent dephosphorylation of GIRK2(S9), and it led to enhanced A₁R-GIRK signaling and basal activity but no change in GABA_BR-GIRK signaling.

Morphine-dependent plasticity of GIRK-dependent signaling was also reported in cultured rat hippocampal neurons (191). This adaptation is mediated by activation of pertussis toxin-insensitive G proteins and involved activation of CaMKII. Morphine treatment involved an overall increase in GIRK2 protein levels, subcompartment-dependent changes in GIRK2, and an increase in overlap between GIRK2 and the excitatory synapse protein PSD-95. Interestingly, 5HT_1AR-GIRK signaling was enhanced by chronic morphine treatment, whereas GABA_BR-GIRK signaling was suppressed. This adaptation was recapitulated in cultures from Girk2^–/– mice expressing either GIRK2a or GIRK2c (65).

Several examples of psychostimulant-induced suppression of GIRK channel activity involving GIRK channel internalization have been described in the mouse VTA and medial prefrontal cortex (39, 187, 188). For example, a single injection of cocaine transiently suppressed GABA_BR-GIRK and D₂R-GIRK signaling in mouse VTA dopamine neurons. This adaptation persisted for 3–5 days and correlated with increased GIRK2 internalization (187). Although mechanisms of plasticity were not resolved, methamphetamine self-administration in mice also triggered a suppression of GABA_BR-GIRK and D₂R-GIRK signaling in midbrain dopamine neurons (190), and a single methamphetamine injection suppressed GABA_BR-dependent signaling in mouse VTA dopamine neurons (188).

Acute exposure to methamphetamine (or cocaine) suppressed GABA_BR-GIRK responses in mouse VTA GABA neurons (74, 188). This neuroadaptation persisted for at least 7 days and involved increased internalization of GIRK2 and GABA_BR. GABA_BR-GIRK currents were rescued by the PP1/PP2A phosphatase inhibitor okadaic acid, suggesting that the internalization was sustained by phosphatase activity and reversible. Indeed, methamphetamine provoked a decrease in phosphorylation of GABA_BR2 at serine-783 in the VTA. A subsequent study involving GABA_BR2(S783A) knock-in mice, which carry a phospho-dead mutation of S783 and are thus resistant to PP2A dephosphorylation (203), did not show this plasticity (74). Notably, the coordinated internalization of GABA_BR and GIRK2 seen in this neuroadaptation, along with cases of GIRK channel plasticity specific to 1 but not to other inhibitory GPCRs in the same cell (121, 191), supports the signalosome framework for GIRK-dependent signaling.

Following a repeated cocaine injection approach commonly used to evoke locomotor sensitization in rodents, layer 5/6 pyramidal neurons in the prelimbic (PL) region of the mouse medial prefrontal cortex exhibited decreased GABA_BR-GIRK signaling (39). This form of GIRK channel plasticity was also correlated with elevated internalization of GIRK2 and GABA_BR and rescued by okadaic acid. Notably, GIRK-dependent signaling in these neurons was also suppressed in male (but not female) mice by chronic unpredictable stress (201). Moreover, remifentanil self-administration correlated with decreased GABA_BR-GIRK signaling in layer 5/6 PL pyramidal neurons in male but not female mice (192).

Aversive experience (inescapable footshock) suppressed GABA_BR-GIRK signaling in neurons of the mouse lateral habenula (196). The adaptation emerged after just 1 h and persisted for up to 14 days. This form of experience-dependent plasticity correlated with increased internalization of GABA_BR1 and GIRK2, and it was rescued by okadaic acid and the more selective and membrane-permeable PP2A inhibitor LB-100.

Phosphoregulation

As alluded to aforementioned, phosphoregulation has been implicated in several examples of GIRK channel plasticity. GIRK2(S9) activity is the best example of phosphoregulation that involves phosphorylation of GIRK channel itself (121). Dephosphorylation of this residue promotes increased surface trafficking of GIRK channels and as such, phosphorylation of GIRK2(S9) should enhance the internalization of GIRK channel.

Several examples of plasticity that involve suppression of GIRK channel activity are rescued by the PP1/PP2A inhibitor okadaic acid (39, 188, 196), or the more selective PP2A inhibitor LB-100 (196). This suggests that dephosphorylation of GIRK channels or a functionally related target(s) underlies the suppression. As treatment with phosphatase inhibitor should enhance the phosphorylation of the relevant target(s), and these forms of GIRK channel plasticity involve suppression of GIRK channel activity, they do not likely involve GIRK2(S9). Indeed, dephosphorylation of GABA_BR2(S783) was linked to the methamphetamine-induced suppression of GABA_BR-GIRK signaling in mouse VTA GABA neurons (188) and, strikingly, this adaptation was not observed in phospho-deficient GABA_BR2(S783A) knock-in mice (74). PP2A-mediated dephosphorylation of GABA_BR2(S783) has been implicated in the surface expression of GABA_BR (204), and aversive experience does impact PP2A activity in the central nervous system (CNS) (205). Moreover, a Ca²⁺-dependent association between PP2A and GABA_BR1 has been described, and this interaction enhances the dephosphorylation of GABA_BR2(S783), promoting GABA_BR internalization (206).

An alternative phosphoregulatory mechanism has been implicated in the stress-induced suppression of KOR-GIRK signaling in serotonergic neurons of the mouse dorsal raphe (195). Stress-induced suppression of KOR-GIRK signaling was lost in mice lacking p38a mitogen-activated protein kinase. A corresponding increase in tyrosine phosphorylation of GIRK1(Y12) suggests that GIRK channel phosphorylation underlies this form of stress-induced plasticity. Phosphorylation of GIRK1(Y12) suppresses GIRK channel activity (118, 129), and increased phosphorylation of GIRK1(Y12) has been reported in several aversive settings in mice including behavioral stress, neuropathic pain, and acute inflammation (195, 207, 208).

GIRK3

GABA_BR-GIRK signaling in mouse VTA dopamine neurons is subject to bi-directional plasticity triggered by neuronal activity (73). Burst firing of VTA dopamine neurons provoked a rapid increase in GABA_BR-GIRK and D₂R-GIRK signaling in mouse VTA slices. This form of plasticity was mimicked by depolarization of the VTA dopamine neurons and involved NMDAR and CaMKII activation, and was blocked by tetanus toxin, implicating increased receptor/channel insertion in the membrane. This adaptation was not seen in VTA dopamine neurons from Girk3^–/– mice. Interestingly, tonic stimulation or low-frequency stimulation of mouse dopamine neurons rapidly suppressed synaptic GABA_BR-GIRK signaling; this adaptation was also dependent upon GIRK3.

Repeated methamphetamine injections suppressed GABA_BR-GIRK but not D₂R-GIRK signaling in mouse VTA dopamine neurons (74). This form of plasticity was not seen in VTA dopamine neurons from Girk3^–/– mice, and indirect evidence suggested that GIRK channel internalization was involved. Interestingly, the methamphetamine-induced suppression of GABA_BR-GIRK signaling in VTA dopamine neurons was still observed in GABA_BR2(S783A) knock-in mice, indicating that phosphoregulation of GABA_BR2(S783A) does not underlie all forms of psychostimulant-induced GIRK channel plasticity.

The involvement of GIRK3 in plasticity driven by neuronal activity or psychostimulants may reflect the unique subunit composition of the GIRK channel in mouse VTA dopamine neurons (GIRK2c:GIRK3) and/or protein:protein interactions mediated by GIRK3 in these neurons. Indeed, GIRK3 can interact with proteins that regulate channel trafficking and/or sensitivity to GPCR activation, including RGS2 (Fig. 1C). RGS2 negatively regulates GABA_BR-GIRK coupling sensitivity in mouse VTA dopamine neurons, an influence attributable to a physical interaction with GIRK3 (44). Notably, chronic cocaine decreased RGS2 expression in the rat brain (209); a similar effect could explain the enhanced GABA_BR-GIRK coupling efficiency seen in mouse VTA dopamine neurons following repeated morphine injections (44). GIRK3 coexpression suppresses GIRK channel activity evoked by GIRK2 expression in Xenopus oocytes (59), and ectopic viral expression of GIRK3 suppresses GIRK-dependent signaling in mouse VTA dopamine neurons (210). Thus, stimuli that alter the expression or stability of GIRK3 should lead to predictable changes in GIRK-dependent signaling, at least in VTA dopamine neurons.

GIRK3 has the same carboxyl-terminal PDZ interaction domain as GIRK2c, and this domain promotes an interaction with sorting nexin 27 (SNX27) (211) (Fig. 1C). SNX27 targets GIRK3-containing channels to early endosomes, leading to reduced surface expression of GIRK channels (19, 75, 211). SNX27 mRNA levels in the rat neocortex are elevated following exposure to psychostimulants (212), suggesting a plausible mechanism for the psychostimulant-induced suppression of GIRK-dependent signaling in VTA dopamine neurons. The depolarization-induced enhancement of GIRK channel activity in mouse VTA dopamine neurons was blocked by a dominant negative peptide mimicking the GIRK3 PDZ interaction domain, supporting a role for SNX27-mediated trafficking of GIRK channels in this form of plasticity. Interestingly, the total expression and subcellular trafficking of recombinant GIRK2c homomers is unaltered by SNX27 expression in (213), indicating that other elements of GIRK3, perhaps the lysosomal targeting motif, are required for the influence of SNX27 on GIRK channel trafficking.

NMDA Receptor

NMDAR activation increases surface trafficking of GIRK channels in rat hippocampal neurons by stimulating the PP1-dependent dephosphorylation of GIRK2(S9) (121). Other forms of GIRK channel plasticity also depend upon NMDAR activation, including the high frequency stimulation-induced strengthening of the GABA_BR-GIRK inhibitory postsynaptic current in rat hippocampal neurons (186), and the bi-directional activity-dependent plasticity of GIRK-dependent signaling in mouse VTA dopamine neurons (73). These forms of plasticity also involved intracellular Ca²⁺ increases and CaMKII activation.

There is a complex inter-relationship between the strength of GIRK channel and NMDAR activity in neurons. GIRK channels normally serve as an inhibitory break on NMDAR-dependent signaling in neurons (7). Genetic ablation of NMDAR in mouse forebrain pyramidal neurons correlated with decreased GIRK2 levels in cortical synaptic membranes (214). A single injection of the NMDAR antagonist ketamine, which induces a rapid antidepressant response, negatively regulates GABA_BR-GIRK signaling in the rat spinal cord (215). This intriguing effect is mediated by stabilization of adaptor protein 14-3-3eta, which decouples GABA_BR from GIRK channels. In a social defeat model of depression in rats, GABA_BR and 14-3-3eta protein levels were decreased in the hippocampus; NMDAR antagonists increased GABA_BR and 14-3-3eta protein levels, whereas GIRK2 protein levels were decreased (216).

NEURONAL GIRK CHANNEL PHARMACOLOGY

Although there are relatively few selective pharmacological tools available to study neuronal GIRK channels, progress made over the last decade has offered reason for optimism; this has been the focus of several recent reviews (3, 13, 217–219).

GIRK Channel Activators

ML297 is the prototypical member of a family of small molecule modulators of GIRK1-containing GIRK channels (220). ML297 is a potent (EC₅₀ ∼100 nM) urea-based compound that shows modest selectivity for recombinant neuronal (GIRK1:GIRK2) over cardiac (GIRK1:GIRK4) GIRK channel subtypes. GIRK channel activation by ML297 is PIP₂-dependent, but Gβγ-independent (221). Unique residues in the GIRK1 pore helix and second membrane-spanning domain are critical for the ML297-induced activation of GIRK1-containing channels.

ML297 has been used in several studies exploring the physiological relevance and therapeutic potential of GIRK channel activation. Systemic ML297 postponed seizure onset in an electroshock model and prevented convulsions and death in PTZ-induced seizure model in mice (220). Systemic ML297 also decreased multiple measures of anxiety-related behavior in male mice (221–223), without impacting motor activity or depression-related behavior, or inducing a conditioned place preference (CPP) (221). Interestingly, direct infusion of ML297 into the mouse ventral hippocampus recapitulated the apparent anxiolytic effect of systemic ML297, and selective ablation of GIRK channels in CA3 and dentate gyrus subregions of the ventral hippocampus was sufficient to block the anxiolytic effect of systemic ML297 in mice (223). Coadministration of ML297 prevented the synaptic and cognitive deficits provoked by intracerebroventricular (ICV) infusion of Aβ in mice (202, 224, 225). Intrathecal administration of ML297 in rats increased the mechanical nociceptive threshold without impairing motor function (226). ML297 inhibited wake activity, preferentially prolonging nonrapid eye movement sleep without changing sleep intensity in mice (227). Systemic ML297 also rescued the deficits in cognitive flexibility evoked by chronic unpredictable stress in male mice (201).

ML297 displays low solubility and poor blood-brain barrier permeability (220, 228), properties that limit its clinical utility. Two second-generation urea-based compounds stand out following structure activity relationship optimization. VU0466551 is approximately twice as potent as ML297 and displays analgesic effects in mice when dosed alone or together with submaximal morphine (229). GAT1508 is strongly selective for recombinant GIRK1:GIRK2 channels relative to cardiac (GIRK1:GIRK4) channels, and it can serve as a both a direct activator and positive allosteric modulator of neuronal GIRK channels (230). Interestingly, GAT1508 (but not ML297) accelerated the extinction of conditioned fear responses in mice, but was without effect on motor activity and coordination, anxiety-related behavior, or cognition. VU0810464 is a non-urea-based compound with higher GIRK1:GIRK2 channel selectivity and improved brain distribution as compared with ML297 (222), but it has a shorter half-life (228). VU0810464 exhibited no anxiolytic efficacy in mice in the elevated plus maze, but it retained efficacy in the stress-induced hyperthermia test (222).

Structure-based virtual screening against the alcohol binding site in GIRK2 lead to the identification of a compound (NCATS_2) that can activate both homomeric (GIRK2) and heteromeric (GIRK1:GIRK2) GIRK channels (231). An NCATS_2 analog (NCATS_2.3) was identified as selective for GIRK1-containing GIRK channels, but with higher brain distribution than ML297. This analog, also called G protein-independent GIRK activator 1 (GiGA1), activates GIRK1:GIRK2 channels in a G-protein-independent manner. Following systemic administration in mice, GiGA1 suppressed the seizure score and reduces the time in tonic seizure in a PTZ-induced seizure model. As was noted with ML297 (but not GAT1508), GiGA1 induces a dose-dependent decrease in locomotor activity, consistent with a sedative influence.

GIRK Channel Inhibitors

Few compounds selectively block GIRK channels. Recombinant GIRK channels are inhibited by volatile anesthetics (232, 233), as well as several plant-derived compounds including ginsenoside Rf, hesperidin, and diterpenes (234–238). Tertiapin, or the stable synthetic derivative tertiapin-Q (239, 240), is a commonly used GIRK channel inhibitor. Tertiapin is a small peptide isolated from bee venom that blocks GIRK channels, as well as Kir1.1 channels (240, 241) and BK (large-conductance Ca²⁺-activated K⁺) channels (242). Tertiapin exerts proconvulsant effects (243) and ICV infusion of tertiapin in mice disrupted the induction and maintenance of LTP in the hippocampus (202, 224). ICV tertiapin in mice also impaired nonassociative and recognition hippocampal-dependent learning, without altering motor function or coordination (244).

Neuronal GIRK Channels as Secondary Drug Targets

Several drugs used in medical practice, whose primary molecular targets are not GIRK channels, nevertheless impact GIRK channel activity. This may contribute to the therapeutic efficacy of these compounds and/or their undesirable side effects. Typical antipsychotic drugs like haloperidol, pimozide, and thioridazine, as well as atypical antipsychotic drugs like clozapine, inhibit recombinant neuronal and cardiac GIRK channels (245). GIRK channel inhibition by these drugs may explain some side effects, including seizures and sinus tachycardia. Some selective serotonin reuptake inhibitors (e.g., paroxetine, fluoxetine, and sertraline), serotonin-norepinephrine reuptake inhibitors (duloxetine), and tetracyclic antidepressants (amoxapine) can inhibit recombinant neuronal and cardiac GIRK channels (246, 247). GIRK channel inhibition may explain some side effects like seizures induced by antidepressant overdose (248). Tipepidine, an antitussive drug with antidepressant effects in adolescent depression and attention deficit hyperactivity disorder (ADHD), has been shown to enhance noradrenergic and dopaminergic transmission in rodent neurons via inhibiting GIRK channels (249–251).

GIRK CHANNELS IN NEUROLOGICAL DISORDERS AND DISEASES

Genetic evidence has implicated GIRK channels in an array of human neurological disorders and diseases (3, 13, 218). For example, a genome-wide association study (GWAS) of schizophrenia identified a single-nucleotide polymorphism (SNP) in KCNJ3/GIRK1 in a Japanese population (252). A subsequent GWAS identified nine SNPs in KCNJ3/GIRK1 in a Chinese population and increased KCNJ3/GIRK1 transcript levels in the dorsolateral prefrontal cortex of postmortem brains from patients with schizophrenia or bipolar disorder (253). One SNP in KCNJ3/GIRK1 is associated with idiopathic generalized epilepsy (254). An epistatic interaction between KCNJ6/GIRK2 and CREB1 [cyclic adenosine 5′-phosphate (adenosine monophosphate)-response element-binding protein] has been linked to rumination, a cognitive abnormality of depression (255). In humans, two SNPs in KCNJ6/GIRK2 are associated with ADHD risk, and functional variants of KCNJ6/GIRK2 influence executive processes (256). The GIRK channel inhibitor tipepidene mildly improves ADHD symptoms in pediatric patients in a pilot study (251). A mutation in the Forkhead-box protein 2 (FOXP2) transcription factor underlies a severe speech and language disorder, and increased GABA_BR-GIRK signaling in cortical neurons has been implicated (257).

Work in animal models involving gain- or loss-of-function manipulations (1, 3), including recent efforts using brain-region and/or neuron-specific manipulations (Table 2), has complemented these genetic studies in humans, shedding light on potential neuron populations and circuits underlying the impact of GIRK channels on neurophysiology and behavior. What follows is a brief overview of the evidence linking neuronal GIRK channels to addiction and cognitive disorders.

Table 2.

Regional and/or cell-specific manipulation of neuronal GIRK channels

Subunit	Manipulation Brain Region/Neuron	Approach	Outcome	Refs.
GIRK1 ablation	Forebrain/pyramidal	CaMKIICre:Girk1fl/fl	↑ locomotor activity (cocaine)	(258)
GIRK1 ablation	PL/pyramidal	Viral Cre: Girk1fl/fl	↑ locomotor activity (cocaine)	(189)
			- Fear conditioning
GIRK1 ablation	PL/pyramidal	Viral Cre: Girk1fl/fl	↑ anxiety-related behavior (EPM)	(201)
			↑ depression-related behavior (FST)
			↓ cognitive flexibility
GIRK1 ablation	IL/pyramidal	Viral Cre: Girk1fl/fl	- Anxiety-related behavior (EPM)	(259)
			- Depression-related behavior (FST)
			↓ cognitive flexibility
GIRK1 ablation	Whole-brain/parvalbumin	PVCre: Girk1fl/fl	↓ anxiety-related behavior (EPM)	(260)
			↑ depression-related behavior (FST)
			- Cognitive flexibility
GIRK1 ablation	vHPC/CA3 and DG	Viral Cre: Girk1fl/fl	Abolished anxiolytic effect of ML297	(223)
GIRK2 overexpression	VTA/dopamine	DATCre, viral	- Anxiety-related behavior (EPM)	(77)
			- Depression-related behavior (FST)
			↓ locomotor activity (cocaine)
GIRK2 overexpression	VTA/dopamine	SNX27THKO, viral	Rescued cocaine sensitization	(261)
GIRK2 overexpression	dHPC/CA1 pyramidal	CaMKIICre: Girk2fl/fl, viral	rescued fear conditioning	(65)
GIRK2 ablation	Forebrain/pyramidal	CaMKIICre: Girk2fl/fl	↑ locomotor activity (cocaine)	(258)
GIRK2 ablation	Forebrain/pyramidal	CaMKIICre: Girk2fl/fl	↓ fear conditioning	(262)
			- Motor activity, hot-plate
			- Anxiety-related behavior (EPM)
			- Spatial learning (Barnes maze)
GIRK2 ablation	Whole brain/dopamine	DATCre: Girk2fl/fl	↑ locomotor activity (morphine)	(40)
GIRK2 ablation	Whole brain/dopamine	DATCre: Girk2fl/fl	↑ locomotor activity (cocaine)	(210)
			- CPP (cocaine)
			↑ self-administration (cocaine)
GIRK2 ablation	Whole brain/dopamine	DATCre: Girk2fl/fl	↓ depression-related behavior (FST)	(263)
			- Locomotor activity
GIRK2 ablation	Whole brain/GABA	GADCre: Girk2fl/fl	- Locomotor activity (morphine)	(40)
GIRK2 ablation	Whole brain/GABA	GADCre: Girk2fl/fl	- Motor activity, hot-plate	(262)
			- Anxiety-related behavior (EPM)
			- Fear conditioning (Barnes maze)
GIRK3 overexpression	VTA/dopamine	DATCre, viral	- Anxiety-related behavior (EPM)	(77)
			- Depression-related behavior (FST)
			↑ locomotor activity (cocaine)
GIRK3 overexpression	VTA	Girk3−/−, viral	Rescued morphine sensitivity	(40)

Studies employing genetic and/or viral approaches to manipulate GIRK channel activity in specific brain regions and/or neuron populations in mouse models. CPP, conditioned place preference; DG, dentate gyrus; dHPC, dorsal hippocampus; EPM, elevated plus maze; FST, forced swim test; GIRK, G protein-gated inwardly rectifying K⁺; IL, infralimbic cortex; PL, prelimbic cortex; PV, parvalbumin; vHPC, ventral hippocampus; SNX27_THKO, dopamine neuron-specific SNX27 knockout; VTA, ventral tegmental area. Directional arrows, impact of manipulation on specified behavior: up arrows, enhancement; down arrows, suppression.

Addiction

The mesocorticolimbic dopamine system is an array of interconnected brain regions tasked with reward processing that includes the VTA and medial prefrontal cortex (264). GIRK channels are expressed in key reward-related neuron populations in rodents and they have been implicated in the biological actions of many categories of drugs of abuse (265). Genetic variation in the human KCNJ6/GIRK2 gene is associated with reduced opioid sensitivity and opioid withdrawal symptoms (266–269), nicotine dependence (269, 270), and alcohol dependence and stress-related adolescent drinking (271).

Work in recombinant systems has revealed that GIRK channels are direct targets of some drugs of abuse, activated by ethanol (272, 273) and inhibited by phencyclidine (274). GIRK channels are downstream effectors of GPCRs engaged—directly or indirectly—by drugs of abuse, including D₂R (275), GABA_BR (4), and opioid receptors (MOR, DOR, and KOR) (174). Work in mouse models has shown that GIRK channel activation contributes to many facets of opioid-related physiology and behavior, including respiratory depression (276, 277), withdrawal (278), and systemic and spinal analgesia (70, 279–282). Here, we discuss the inter-relationship between GIRK channel activity and drugs of abuse, as exemplified by studies in mice involving psychostimulants and ethanol.

Psychostimulants.

Exposure to psychostimulants like cocaine weakens inhibitory G protein signaling in the mouse mesocorticolimbic system (283). As noted earlier, cocaine or methamphetamine can suppress GIRK currents in mouse VTA dopamine (187) and GABA (188) neurons, and repeated cocaine injection in mice suppresses GIRK-dependent signaling in layer 5/6 pyramidal neurons of the prelimbic cortex (PL) (39). Drug-naïve Girk1^–/– and Girk2^–/– mice exhibit enhanced glutamatergic neurotransmission in the nucleus accumbens (67), as well as increased behavioral sensitivity to drugs of abuse (40, 284, 285), similar to observations in wild-type mice following chronic drug exposure.

Genetic and viral genetic approaches in mice have been used to mimic these forms of GIRK channel plasticity, with the goal of understanding their behavioral relevance. Mice lacking GIRK channels in dopamine neurons exhibit increased sensitivity to the motor-stimulatory effect of cocaine, as well as increased responding for and intake of cocaine in a self-administration task (210). Suppression of GIRK channel activity selectively in VTA dopamine neurons, via Cre-dependent virus-driven overexpression of GIRK3 in mice, also increased the motor-stimulatory effect of cocaine, without impacting select measures of anxiety- or depression-related behavior (77). Genetic ablation of GIRK channels in mouse forebrain pyramidal neurons enhanced glutamatergic neurotransmission in D₁R-expressing medium spiny neurons in the nucleus accumbens core and increased motor-stimulatory responses to cocaine, comparable with effects seen in wild-type mice given repeated cocaine injections (258). The more selective viral Cre-dependent ablation of GIRK channels in mouse PL pyramidal neurons increased motor-stimulatory effect of cocaine (189), while also impairing working memory and cognitive flexibility in male (but not female) mice (201). Modeling these adaptations in drug-naïve mice supports the contention that psychostimulant-induced plasticity of GIRK channel in reward circuitry contributes to some of the electrophysiological and behavioral hallmarks of addiction.

Alcohol.

GIRK channels play prominent roles in key aspects of alcohol consumption, reward, and withdrawal. Genetic variation in the human KCNJ6/GIRK2 gene has been linked to alcohol dependence and stress-related adolescent drinking (271). In addition, several single-nucleotide polymorphisms in the KCNJ6/GIRK2 promoter were identified in subjects with alcohol use disorder (AUD) and offspring at risk for developing an AUD (286). Variations in Kcnj9/Girk3 expression have been linked to strain differences in key ethanol-related behaviors in mice; lower levels of Kcnj9/Girk3 expression are associated with less severe withdrawal from ethanol, pentobarbital, and zolpidem (71). Withdrawal from ethanol and sedative-hypnotics is less severe in Girk3^–/– mice (71). The Kcnj9/Girk3 gene has also been linked to ethanol drinking (287), ethanol-induced conditioned aversion (288), and acute sensitivity to alcohol (289) in mice.

Ethanol binds to and activates recombinant GIRK channels at intoxicating concentrations, independent of Gβγ or cholesterol (17, 21, 169, 272, 273). Many ethanol-induced behaviors are altered in Girk^–/– mice. For example, Girk2^–/– mice show less severe withdrawal symptoms, including handling-induced convulsions (290), and Girk2^–/– mice lack the anxiolytic effect of ethanol, as assessed in the elevated plus maze test (290). Girk2^–/– mice also lack ethanol-induced analgesia (291), and exhibit reduced conditioned place preference and conditioned taste aversion (292). Like Girk2^–/– mice, Girk3^–/– mice exhibit less severe ethanol withdrawal (71). In contrast to Girk2^–/– mice, however, Girk3^–/– mice exhibit enhanced ethanol reward and CPP (293). The ethanol-induced excitation of VTA dopamine neurons is blunted in Girk3^–/– mice (72), a physiological phenotype that correlates with elevated binge-like drinking by Girk3^–/– mice. Notably, viral re-expression of GIRK3 in the VTA normalized binge drinking behavior of Girk3^–/– mice.

Ethanol also triggers adaptations in GIRK channel activity (294). For example, significant changes in GIRK1 and GIRK3 transcript levels in the mouse nucleus accumbens were observed following chronic intermittent exposure to ethanol vapor (CIE) (200). In addition, withdrawal from repeated in vivo alcohol exposure correlated with enhanced D₂R-GIRK (but not GABA_BR-GIRK) signaling in mouse VTA dopamine neurons (193). Furthermore, following CIE treatment, the GIRK-dependent inhibition of orbitofrontal cortical neurons triggered by D₂R, 5HT_1AR, and α₂ adrenergic receptor activation was blunted in mice (194). GIRK currents evoked by ML297 were also suppressed following CIE, suggesting that this neuroadaptation involved the GIRK channel itself.

Cognitive Disorders

GIRK channels are highly expressed in brain regions related to cognitive function, including the hippocampus, amygdala, and prefrontal cortex (34). Several lines of evidence indicate that dysregulation of GIRK-dependent signaling is associated with cognitive deficits in humans. Perhaps most strikingly, the sequencing results of the exomes from three unrelated human patients with Keppen–Lubinsky syndrome, a rare disease characterized by severe developmental delay and intellectual disability (295), all exhibit de novo heterozygous mutations in KCNJ6/GIRK2 that severely impair GIRK channel function (296).

Work in rodents has shown that an optimal range of GIRK channel activity is required for normal synaptic plasticity. For example, constitutive and forebrain pyramidal neuron-specific Girk2^–/– mice, as well as GIRK2 trisomy mice that exhibit elevated GIRK channel activity, are deficient in depotentiation of LTP at CA3-CA1 hippocampal synapses (125, 262). Loss-of-function GIRK1 mutant mice also display impaired depotentiation of LTP, suggesting an essential role of both GIRK1 and GIRK2 in this form of synaptic plasticity (297). Acute ICV injections of either the GIRK channel activator ML297 or inhibitor tertiapin in mice similarly transform LTP into long-term depression (LTD) following high-frequency stimulation (202, 224, 244). The threshold for induction of LTP is higher in dorsal rather than ventral CA1 pyramidal neurons in the rat due to elevated dendritic GIRK channel activity in dorsal CA1 neurons (298). Both gain- and loss-of-function approaches in mice involving GIRK channel activity disrupt associative learning (148, 262, 297, 299, 300). Girk2^–/– mice show disrupted contextual fear learning (301), a phenotype recapitulated in forebrain pyramidal neuron-specific Girk2^–/– mice (262). Viral restoration of GIRK channel activity in CA1 pyramidal neurons in the dorsal hippocampus rescued contextual fear learning in these mice (65). Mice receiving acute ICV injections of GIRK activator ML297 or tertiapin show learning impairments in open field habituation, novel object recognition, and operant conditioning tasks (244).

Down syndrome.

Down syndrome (DS) is a congenital disorder characterized by intellectual disabilities, mental impairment, and distinct facial features. DS is caused by an extra maternal copy (trisomy) of human chromosome 21 (299, 302), and KCNJ6/GIRK2 is located in the DS critical region (DSCR). A widely-used DS mouse model, Ts65Dn, is a partial trisomy model containing a region of mouse chromosome 16 orthologous to human chromosome 21 and that includes the DSCR (303). In Ts65Dn mice, GIRK2 protein levels are higher and GABA_BR-GIRK signaling is elevated (304). Ts65Dn mice show various forms of synaptic and cognitive impairments (299, 304, 305), among which several are also seen in mice with trisomy for GIRK2 alone (300). GIRK2 trisomy mice exhibit hampered depotentiation of LTP, accentuated LTD, and impaired hippocampal-dependent contextual fear recall (300). Reducing Kcnj6/Girk2 gene dosage in Ts65Dn mice abolished the impairment of synaptic plasticity and cognitive deficits (305), as well as infantile spasms induced by GABA_BR agonist (306). Trisomy of Kcnj6/Girk2 alone, however, is not sufficient to confer susceptibility to infantile spasms in this model (307). Thus, although GIRK channel upregulation is not the only mechanism underlying pathological phenotypes in this mouse model of DS, downregulation of GIRK channel activity could be a promising therapeutic approach to ameliorate the symptoms and cognitive deficits in individuals with DS.

Alzheimer's disease.

Although GIRK expression levels are altered in different rodent models of Alzheimer’s disease (AD) (197, 198, 202, 308), divergent outcomes across studies complicates easy synthesis of their role. For example, incubation of rat hippocampal slices with synthetic Aβ reduced GIRK mRNA levels (198), but GIRK1 and GIRK2 protein levels were not impacted (202). Similarly, ICV injection of Aβ in mice decreased GIRK1 protein level in the CA1 subregion of the dorsal hippocampus (202), but had no impact on GIRK1 mRNA or protein levels in the rat hippocampus as assessed after 7 days of incubation (309). Increased GIRK channel internalization was noted in the hippocampus of two transgenic mouse models of AD (308). Effects of Aβ on GIRK channel activity have also been reported and, again, results are divergent. Upregulation of GIRK channel activity was implicated in apoptotic cell death in mouse hippocampus (HPC) cultures (199); this effect was observed with 20 µM Aβ, a concentration higher than typical for Aβ neurotoxicity studies. In contrast, incubation of rat HPC slices with Aβ (1.5 µM) increased CA3 pyramidal neuron excitability, and suppression of GIRK channel activity was implicated (197). Remarkably, coadministration of ML297 rescued synaptic plasticity and memory deficits induced by acute ICV Aβ infusion in mice (202, 224, 225). ICV infusion of either Aβ or tertiapin disrupted novel environment habituation and novel object recognition in mice (202, 244, 310). Thus, GIRK channels may contribute to AD pathogenesis and GIRK channel activators could be useful for rescuing synaptic and behavioral deficits.

CONCLUDING REMARKS

It has been 30 years since the cloning of GIRK1 (311, 312), an achievement that ushered in a new era of research into the molecular mechanisms of inhibitory signaling in the nervous system. The last 15 years have seen a dramatic advance in our understanding of the structural basis of GIRK channel function and regulation. Despite this progress, important questions remain. For example, subunit composition and stoichiometry likely dictate unique aspects of GIRK channel function and regulation, and yet we know little about the precise molecular composition of GIRK channels in the nervous system, particularly in humans. Similarly, our understanding of GIRK channel structure relies heavily on insights from studies of recombinant purified GIRK2 homomers, which are likely a relatively rare channel subtype in the mammalian brain.

The impact on neuronal GIRK-dependent signaling of cholesterol, reversible phosphorylation, RGS proteins, and 14-3-3 proteins warrants rigorous examination given the physiological and/or pathophysiological relevance of these regulatory factors. Elucidating mechanisms underlying distinct forms of GIRK plasticity affords an excellent opportunity to discover additional modulatory influences. These insights may also shed important light on how GIRK-dependent signaling is organized in neurons. Available data indicate that mechanisms are in place to coalesce specific interactions between GIRK channels and discrete sets of signaling elements. As our understanding of GIRK channel regulation and compartmentalization improves, we will be in a better position to consider ways to more selectively intervene when GIRK channel activity is dysregulated.

We now have multiple distinct chemical scaffolds for small molecule GIRK channel modulators. Work with these compounds has already shown their promise in preclinical models of epilepsy, pain, anxiety, and Alzheimer’s disease. These tools afford excellent opportunities to continue building the pharmacological toolkit of GIRK channel activators, focusing on tools with improved pharmacokinetics and channel subtype-specificity. To the latter point, most GIRK channel activators characterized to date target GIRK1-containing GIRK channels. Compounds that selectively modulate GIRK1-lacking channels, like NCATS_2.3 (231), VU0529331 (313), or 3hi2one-G4 (314) are particularly intriguing tools. These compounds could lead to the development of drugs exerting a relatively selective influence on the unique GIRK1-lacking channel subtype in midbrain dopamine neurons; the potential therapeutic applications of these compounds in the context of substance use disorders and movement disorders are exciting to consider.

As we ponder the translational potential of therapeutic approaches targeting GIRK channels, it is worth underscoring the critical point that our current understanding of GIRK channel biology is based almost exclusively on work in rodent models and heterologous expression systems. The extent to which insights gleaned from these studies will align with the contribution of neuronal GIRK channels to human neurophysiology and behavior is unclear. The use of more translational models including human-induced pluripotent stem cells and human postmortem tissues, as well as nonhuman primates, may begin to help fill some of the existing translational gaps in this field. More broadly, translating knowledge gained from rodent models to efficacious and safe therapeutic approaches for human diseases and disorders has proven challenging across an array of neurological contexts (e.g., see Refs. 315 and 316). Despite this cautionary note, the efficacy of manipulations targeting neuronal GIRK channels in preclinical models provides reason for optimism as well as support for advancing efforts toward clinical trials for a variety of neurological disorders and diseases.

GRANTS

This review was completed with support from National Institutes of Health Grants DA034696 and AA027544 (to K.W.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.L. prepared figures; H.L., E.M.F.d.V., and K.W. drafted manuscript; H.L., E.M.F.d.V., and K.W. edited and revised manuscript; H.L., E.M.F.d.V., and K.W. approved final version of manuscript.