Regulation of GABA_ARs by Phosphorylation

Abstract

γ-Aminobutyric acid type A receptors (GABA_ARs) are the principal mediators of fast synaptic inhibition in the brain as well as the low persistent extrasynaptic inhibition, both of which are fundamental to proper brain function. Thus unsurprisingly, deficits in GABA_ARs are implicated in a number of neurological disorders and diseases. The complexity of GABA_AR regulation is determined not only by the heterogeneity of these receptors but also by its posttranslational modifications, the foremost, and best characterized of which is phosphorylation. This review will explore the details of this dynamic process, our understanding of which has barely scratched the surface. GABA_ARs are regulated by a number of kinases and phosphatases, and its phosphorylation plays an important role in governing its trafficking, expression, and interaction partners. Here, we summarize the progress in understanding the role phosphorylation plays in the regulation of GABA_ARs. This includes how phosphorylation can affect the allosteric modulation of GABA_ARs, as well as signaling pathways that affect GABA_AR phosphorylation. Finally, we discuss the dysregulation of GABA_AR phosphorylation and its implication in disease processes.

1. INTRODUCTION

With kinases alone constituting about 2% of the human genome, it is perhaps unsurprising that phosphorylation plays a key role in all aspects of cellular activity and is one of the best characterized of posttranslational modifications (Manning, Whyte, Martinez, Hunter, & Sudarsanam, 2002; Ubersax & Ferrell, 2007). Phosphorylation is dynamically executed by the opposing functions of kinases and phosphatases, which lead to changes in protein conformation and function. Protein kinases contain a common catalytic domain that catalyses the transfer of γ-ATP to a serine, threonine, or tyrosine residue of the intended protein. In contrast, phosphatases remove phosphate groups from their substrates. Kinases are primarily divided into two main groups: (1) the serine/threonine kinases which include cyclic-AMP dependent protein kinase (PKA), phospholipid-dependent protein kinase C (PKC), and Ca²⁺/calmodulin-dependent protein kinase II (CamKII) and (2) the tyrosine kinases which include Src family tyrosine kinases. The protein kinase family is considerably diverse, as exemplified by PKC, which consist of multiple isoforms with distinct responses to specific activation (Song & Messing, 2005; Tanaka & Nishizuka, 1994; Taylor, Buechler, & Yonemoto, 1990; Taylor, Knighton, Zheng, Ten Eyck, & Sowadski, 1992; Ubersax & Ferrell, 2007). PKC are composed of “classical” or “conventional” cPKC subgroup (isoforms α, β, and γ) which are activated by calcium, phosphatidylserine (PS), and diacylglycerol (DAG); the “novel” nPKC (δ, ε, η, and θ) which are activated by DAG and PS; and the “atypical” aPKC (ζ and λ/ι), activated by other lipid messengers. Phorbol esters, often used to activate PKC, predominantly stimulate cPKC and nPKC (Song & Messing, 2005; Tanaka & Nishizuka, 1994; Taylor et al., 1990, 1992; Ubersax & Ferrell, 2007).

Different kinases recognize specific consensus sequences in the target polypeptide. However, as exemplified below with GABA_ARs, the presence of consensus sequences for a specific kinase does not ensure that the protein is a substrate for the kinase in vivo. Likewise, bona fide phosphorylation sites may not correspond to the consensus sequence (Ubersax & Ferrell, 2007). GABA_ARs in particular are accepted phosphoproteins and their phosphorylation governs numerous processes, including directly varying channel function, regulating receptor trafficking, affecting receptor-interacting proteins, and their sensitivity to pharmacological agents (Brandon, Jovanovic, & Moss, 2002; Houston, He, & Smart, 2009; Jacob, Moss, & Jurd, 2008; Luscher, Fuchs, & Kilpatrick, 2011). Thus, the interplay between kinases and phosphatases can dynamically regulate neuronal excitability and ultimately shape brain function.

2. THE γ-AMINOBUTYRIC ACID TYPE A RECEPTORS

GABA_ARs are GABA-gated Cl⁻-channels responsible for the majority of inhibition in the mammalian brain and the major target for many clinically relevant drugs. Deficits in GABA_AR function are increasingly implicated in numerous pathologies including anxiety (Lydiard, 2003; Rudolph & Mohler, 2004), cognitive deficits (D’Hulst & Kooy, 2007; DeLorey & Olsen, 1999; Thompson-Vest, Waldvogel, Rees, & Faull, 2003), depression (Luscher, Shen, & Sahir, 2011b), epilepsy (Benarroch, 2007; Fritschy, 2008), schizophrenia (Benes & Berretta, 2001; Charych, Liu, Moss, & Brandon, 2009), and substance abuse (Krystal et al., 2006). GABA_ARs are responsible for two forms of inhibition known as phasic (synaptic) and tonic (extrasynaptic) inhibition. Synaptic inhibition occurs via transient or “phasic” activation of GABA_ARs after the release of GABA from synaptic vesicles, whereas extrasynaptic inhibition necessitates low ambient levels of GABA for continual or “tonic” activation (Farrant & Nusser, 2005).

Structurally, GABA_ARs are heteropentameric channels (Fig. 1) composed from a pool of 19 possible subunits: α1–6, β1–3, γ1–3, δ, ε, θ, π, and ρ1–3 (Olsen & Sieghart, 2008). However, general consensus holds that the vast majority of receptors are assembled from two α, two β and one γ (or one δ) (Fig. 1). GABA_ARs composed of α1–3 and γ2 are largely synaptically located, whereas α4–6 and δ are predominantly expressed extrasynaptically (Farrant & Nusser, 2005; Luscher, Fuchs, & Kilpatrick, 2011; Nusser & Mody, 2002). In addition to the diversity of receptor subtypes, alternative splicing offers further heterogeneity. For instance, the γ2 subunits occur as short (γ2S) and long (γ2L) splice variants due to an additional eight amino acids in γ2L (Whiting, McKernan, & Iversen, 1990). Notably, differences in subunit compositions can have an impact on the physiological and pharmacological properties of these receptors (Jacob et al., 2008; Rudolph & Knoflach, 2011; Verdoorn, Draguhn, Ymer, Seeburg, & Sakmann, 1990). Moreover, specific subunits have distinct regional expression profiles which are developmentally regulated (Hortnagl et al., 2013; Laurie, Wisden, & Seeburg, 1992). Indeed, not only is GABA_AR subunit expression brain region-specific, but the expression of particular subunits is also found to be specific at a subcellular level. Finally, on top of this staggering level of complexity, there are also the added intricacies of subunit-specific phosphorylation and dephosphorylation by multiple kinases and phosphatases.

Figure 1.

Schematic representation of the pentameric GABA_AR. GABA_ARs are composed of a large extracellular N-terminal domain, four-transmembrane domains (TM1–4), and a major intracellular loop between TM3 and TM4 where phosphorylation (Ⓟ) primarily occurs. Binding of regions for benzodiazepines (BDZ) and GABA are depicted.

3. PHOSPHORYLATION SITES ON GABA_AR

The major intracellular loop between TM3 and TM4 of GABA_AR contains numerous consensus sites for phosphorylation by both serine/threonine and tyrosine protein kinases (Moss & Smart, 1996). Earlier work relied largely on glutathione S-transferase (GST) fusion proteins that encode for the large intracellular loop of specific GABA_AR subunits. Several kinases were found to phosphorylate sites in this domain, the vast majority of which lie on β and γ2 subunits (Brandon, Delmas, Hill, Smart, & Moss, 2001; McDonald & Moss, 1994; McDonald & Moss, 1997; Moss, Doherty, & Huganir, 1992; Moss, Gorrie, Amato, & Smart, 1995; Table 1b and c). For example, a conserved serine residue found in β1–3 subunits can be phosphorylated by PKA, PKC, protein kinase G (PKG), and CamKII (McDonald & Moss, 1994, 1997; Moss, Doherty, et al., 1992). Both γ2S and γ2L can be phosphorylated on serine residue 327 (S327) by PKC, whereas γ2L has an additional residue at S343 that can be phosphorylated by PKC and CamKII (McDonald & Moss, 1994, 1997; Moss, Doherty, et al., 1992).

Table 1.

Known phosphorylation sites of GABA_ARs subunits

(a) α and δ subunits
Subunit	Site	Kinase or phosphatase			Effect of phosphorylation, phosphomimetic	Selected publications
		GST	Cell line	Neuron
α1	–	☒ Akt ☒ PKA ☒ PKC ☑ Src	HEK: α1β1γ2 ☒ PKA	FB synaptosome: ☑ CamKII	☑–	HEK (Moss, Smart, et al., 1992) FB (Churn et al., 2002) GST (Moss, Doherty, et al., 1992; Vetiska et al., 2007; Wang, Liu, Pei, et al., 2003)
	Putative T375	–	–	–	Hip: Phosphomimetic ↓ gephyrin binding, ↓ synaptic clustering, and ↓ mIPSC amplitude.	(Mukherjee et al., 2011)
α4	S443	–	COS7: α4β3 ☑ PKC	Hip: ☑ PKC	COS7 (α4β3): ↑ surface, ↑ insertion, ↑ surface stability, block IGABA rundown. Hip: ↑ surface, ↑ insertion, ↑ tonic current.	(Abramian et al., 2014, 2010)
δ	–	–	COS7: ☒ PKC	–	–	(Abramian et al., 2010)

(b) β subunits
Subunit	Site	Kinase or phosphatase			Effect of phosphorylation	Selected publications
		GST	Cell line	Neuron
β1	T227, Y230	–	–	Hip	–	Mass spectrometry (Kang et al., 2011)
	S384	☑ CamKII	–	–	–	GST (McDonald & Moss, 1994)
	S409	☑ CamKII ☑ PKA ☑ PKC ☑ PKG	HEK:α1β1γ2 ☑ PKA; α1β1γ2S/L ☑ PKC	SCG	HEK: ☑ PKA ↓ IGABA amplitude. HEK: ☑ PKC ↓ IGABA amplitude.	GST (McDonald & Moss, 1994; Moss, Doherty, et al., 1992) HEK (Krishek et al., 1994; Moss, Smart, et al., 1992) SCG (Brandon, Jovanovic, Smart, & Moss, 2002)
β2	–	–	–	Hip, Ctx: ☒ PKC (via BDNF) PFC: ☒ PKC	Hip, Ctx (via BDNF): ☒ PKC no Ⓟ PFC (via 5HT2): ☒ PKC no change in Ⓟ	BDNF (Jovanovic et al., 2004) PFC (Feng et al., 2001)
	Y215, T439	–	–	Hip	–	Mass spectrometry (Kang et al., 2011)
	Y372/Y379	☑ Src	HEK: α1β1γ2	–	HEK (via insulin): ↑ surface.	(Vetiska et al., 2007)
	S410	☑ Akt ☑ CamKII ☑ PKA ☑ PKC ☑ PKG	HEK: α1β2γ2 ☑ Akt, ☒ PKA Oocyte: α1β2γ2S ☑ PKC HEK: α4β2δ ☑ PKC	Hip: ☑ Akt CGC: β2 KO ☒ CamKII	GST: Phosphomimetic binds gephyrin. HEK, Hip: ☑ Akt ↑ surface, ↑ mIPSC amplitude & frequency. Oocyte: ☑ PKC ↓ IGABA amplitude (with γ2 S327 for full affect). CGC β 2 KO: No affect of CamKII activation compared to WT. HEK α4β2δ: ☑ PKC ↓ tonic inhibition, ↓ surface	GST (Kowalczyk et al., 2013; McDonald & Moss, 1997) Akt (Wang, Liu, Pei, et al., 2003) Not PKA (McDonald et al., 1998) Oocyte (Kellenberger et al., 1992) β2 KO (Houston & Smart, 2006) Tonic (Bright & Smart, 2013)
β3	T282, S406	–	–	Hip	–	Mass spectrometry (Kang et al., 2011)
	S383	☑ CamKII	HEK: α1β3 ☒ CamKII NG108-15: α1β3(γ2) ☑ CamKII	Hip: ☑ CamKII	NG108-15: ☑ CamKII ↑ IGABA amplitude. Hip (via VGCCs): ☑ CamKII ↑ surface, ↑ insertion, ↑ tonic current. Hip: ☑ CamKII LTPGABA ↑ synaptic clusters of gephyrin and GABAAR.	GST (McDonald & Moss, 1997) NG108-15 (Houston et al., 2007) VGCC (Saliba et al., 2012) LTPGABA (Petrini et al., 2014)
	S408	☑ PKC ☒ CamKII	(see S408/S409)	(see S408/S409)		GST (McDonald & Moss, 1997)
	S409	☑ CamKII ☑ PKA ☑ PKC ☑ PKG	(see S408/S409)	(see S408/S409)	GST: Ⓟ-S409 ↓ binding to PKC.	GST (Brandon, Jovanovic, Smart, et al., 2002; McDonald & Moss, 1997)
	S408/S409	(As above)	HEK: α1β3γ2 ☑ PKA NG108-15: α1β3 ☒ CamKII	Ctx, Hip: RACK1-☑ PKC ☑ PP2A Ctx, Hip: PRIP-☑ PP2A Hip: ☑ PKA PRIP1-☑ PP1α	HEK: ☑ PKA ↑ IGABA. CamKII NG108-15: Mutant does not change IGABA. Ⓟ inhibits binding to AP2. Ctx, Hip (via BDNF): ☑ PKC ☑ PP2A, transient ↑ then ↓ mIPSC amplitude and Ⓟ, ↑ surface but also reported to ↑ then ↓ surface in WT and ↑ surface in PRIP dKO Hip : ☑ PKA ↑ I GABA in WT but not in PRIP1 KO. PKA activity dissociates PRIP1-PP1 α which de Ⓟ S408/S409.	HEK (McDonald et al., 1998) NG108-15 (Houston et al., 2007) AP2 (Kittler et al., 2005a) BDNF (Jovanovic et al., 2004; Kanematsu et al., 2006) PP1α (Terunuma et al., 2004)

(c) γ subunits
Subunit	Site	Kinase or phosphatase			Effect of phosphorylation	Selected publications
		GST	Cell line	Neuron
γ2S/L	–	☒ Akt ☒ PKG ☒ PKA ☑ PKC γ2L ☒ PKC γ2S	HEK: α1β1γ2 ☒ PKA	Hip, Ctx: ☒ PKC (via BDNF) PFC: RACK1-☑ PKC Hip Fyn: ☑ Fyn (KO ↓ phospho)	Hip, Ctx (via BDNF): ☒ PKC no Ⓟ PFC (via 5HT2): ☑ PKC ↓ IGABA, γ↑ 2 Ⓟ	GST (McDonald & Moss, 1994; Moss, Doherty, et al., 1992; Wang, Liu, Pei, et al., 2003; Whiting et al., 1990) HEK (Moss, Smart, et al., 1992), BDNF (Jovanovic et al., 2004). PFC (Feng et al., 2001), and Fyn (Jurd et al., 2010)
	S327	☑ PKC ☑ PKCε ☑ CaN	HEK: α1β1γ2S/L ☑ PKC ☑ PKCε Oocyte: α1β2γ2S ☑ PKC	Hip: ☑ CaN Hip, Ctx: ☑ PKCε (PKCε KO ↓ phospho)	HEK: ☑ PKC ↓ IGABA amplitude. HEK: ☑ PKCε modulates actions of benzodiazepine and ethanol Oocyte : ☑ PKC ↓ IGABA amplitude (with β2 S410 for full affect). Hip (via LTDGABA): ☑ CaN ↓ S327 Ⓟ required for LTDGABA, ↑ lateral diffusion	GST (Moss, Doherty, et al., 1992) HEK (Krishek et al., 1994), PKCe (Qi et al., 2007), oocyte (Kellenberger et al., 1992), and LTDGABA (Muir et al., 2010; Wang, Liu, Haditsch, et al., 2003)
	S348, T350	☑ CamKII	NG108-15: α1β3 ☒ CamKII		NG108-15: ☒ CamKII Ⓟ-mutant does not change IGABA.	GST (McDonald & Moss, 1994) NG108-15 (Houston et al., 2007)
	Y365/Y367	☑ Fyn ☑ Src	HEK: α1β1γ2L ☑ Src	Ctx, WB: basally Ⓟ	HEK: ☑ Src ↑ IGABA amplitude. Ⓟ Y365/Y367 binds Src. Hip: Ⓟ Y367 binds Fyn. Ⓟ Y365/Y367 ↓ AP2 binding. Y365/7F+/+ : embryonic lethality. Y365/7F+/−: ↑ surface, ↑ size inhibitory synapse, ↑ mIPSC amplitude and frequency, ↓ AP2 binding, ↓ spatial object recognition	HEK (Moss et al., 1995) GST, Ctx, and WB (Brandon et al., 2001) Fyn (Jurd et al., 2010) Y365/7F mouse (Tretter et al., 2009) AP2 (Kittler et al., 2008)
γ2L	S343	☑ CamKII ☑ PKC	HEK: α1β1γ2S/L ☑ PKC		HEK: ☑ PKC ↓ IGABA amplitude	GST (McDonald & Moss, 1994; Moss, Doherty, et al., 1992); HEK (Krishek et al., 1994)

Whole-cell GABA-activated currents (I_GABA); miniature inhibitory post synaptic current (mIPSC); Cerebellar granule cells (CGC); Cortex (Ctx); Forebrain (FB); Hippocampus (Hip); Prefrontal cortex (PFC); Superior cervical ganglion (SCG); Whole brain (WB); (↑) increase; (↓) decrease; (Ⓟ) phosphorylation; (deⓅ) dephosphor-ylation; (☑) phosphorylated/dephosphorylated by specific kinase/phosphatase; (☒) not phosphorylated/not dephosphorylated by specific kinase/phosphatase.

3.1. Phosphorylation in expression systems

The methods outlined above identified a number of sites that were later recapitulated in heterologous expression systems. Although studies of this manner may not account for the presence of proteins intimately associated with GABA_ARs, they allow for the direct examination of the effects of phosphorylation on specific residues with precise receptor combinations. These approaches have been carried out on the numerous phosphorylation sites first identified utilizing GST conjugated to the large intracellular loop of GABA_ARs (McDonald & Moss, 1994, 1997; Moss, Doherty, et al., 1992; Whiting et al., 1990). Human embryonic kidney (HEK293) cells lines have been extensively used as a vehicle for the expression of recombinant receptors for such studies, and generally these results have lent support to the in vitro studies, albeit with a few discrepancies (Table 1b and c).

3.1.1 PKA

As a specific example of such discrepancies, in vitro experiments have suggested that β1–3 subunits are phosphorylated by PKA (McDonald & Moss, 1997; Moss, Doherty, et al., 1992). In contrast, studies in HEK293 cells showed that β2-containing receptors were not phosphorylated and failed to be modulated by PKA activation (McDonald et al., 1998). However, PKA-induced phosphorylation did, in fact, differentially regulate β1 and β3 subunits expressed as α1β1/3γ2S in HEK293 cells. PKA-dependent phosphorylation of β1-containing receptors at S409 led to an inhibition of GABA-activated responses (McDonald et al., 1998; Moss, Smart, et al., 1992). In contrast, similar PKA activation in β3-containing receptors resulted in the potentiation of GABA-induced currents attributed to two adjacent serine residues at 408 and 409, both of which are phosphorylated by PKA. Moreover, mutation of S408 to an alanine (S408A) in the β3 subunit transformed the enhancement of activity to an inhibition analogous to β1 subunits that are exclusively phosphorylated on S409. The lack of β2 subunit regulation by PKA is most likely due to an inability of β2 to bind A-kinase anchoring proteins (AKAP) (see Section 4.3). These anchoring proteins are found endogenously in HEK293 cells (Gardner, Tavalin, Goehring, Scott, & Bahouth, 2006) and are crucial to subcellular targeting of PKA (Sanderson & Dell’Acqua, 2011).

Since PKA appears to exclusively phosphorylate β1 (S409) and β3 (S408/S409) subunits, it was generally expected that the regulation of GABA_AR activity by PKA should depend on whether a population of receptors is largely β1- or β3-containing. However, contrary to experiments conducted with synaptically expressed α1 receptors, extrasynaptically expressed α4β3γ2 and α4β3δ receptors tell a different story. In HEK293 cells, PKA activation caused a larger increase of spontaneous activity in cells expressing α4β3δ receptors compared to α4β3γ2 receptors; however in the same cells, PKA activation did not affect currents that were activated by 1 μM GABA (Tang, Hernandez, & Macdonald, 2010). These results suggest that PKA could regulate GABA_AR-mediated currents by an unknown process that depends on the GABA concentration.

3.1.2 CamKII and Src

Discrepancies also exist on residues that were thought to be phosphorylated by CamKII (McDonald & Moss, 1994, 1997). In these cases, it has been the choice of cell line that has appeared to underlie a particular kinase’s ability to modulate receptor function (Houston et al., 2009; Houston & Smart, 2006). CamKII did not regulate HEK293 cells transfected with α1β2 and α1β3 receptors. Contrary to this, in the neuroblastoma-glioma hybrid (NG108-15) cell line, CamKII activity enhanced GABA-evoked current amplitudes of α1β3 and α1β3γ2 receptor expressing cells, but not in cells expressing α1β2 and α1β2γ2 receptors (Houston &Smart, 2006). Moreover, similar enhancements were observed upon CamKII activation in cerebellar granule neurons from both wild-type mice and those in which the β2 subunit was specifically knocked out. This result suggests that in cerebellar granule neurons, CamKII effects are mediated primarily by the β3 subunit.

CamKII-dependent enhancement of GABA-evoked currents in α1β3-transfected NG108-15 cells (without γ2) was mediated solely by S383 on the β3 subunit (Houston, Lee, Hosie, Moss, & Smart, 2007). Other potential CamKII phosphorylation sites initially identified by GST pull-down experiments (McDonald & Moss, 1994, 1997) were not involved (Houston et al., 2007; Table 1b). Interestingly, this site not only deviates from the typical consensus site for CamKII, but, thus far, it is also a site exclusively phosphorylated by CamKII. Further, in α1β3γ2 transfected NG108-15 cells, CamKII-mediated enhancement of GABA-evoked currents were not exclusively mediated by S383 on the β3 subunit, but were also mediated by tyrosine phosphorylation sites on residues Y365/Y367 of the γ2 subunit. This suggests that activation of CamKII not only leads to the direct phosphorylation of S383 on β3 but also activates endogenous tyrosine kinases to phosphorylate sites on γ2 (Houston et al., 2007). The tyrosine kinase Src can specifically interact with β and γ2 subunits (Brandon et al., 2001). Moreover, Y365/Y367 in the γ2 subunit can be phosphorylated by Src kinase in α1β1γ2L-expressing A293 cells, which in agreement, also resulted in an enhancement of GABA-activated currents (Moss et al., 1995). These observations were attributed to the phosphorylation of γ2 residue at Y365/Y367 since mutations of these residues to phenylalanines (Y365/Y367F) ablated tyrosine phosphorylation and receptor modulation. Finally, while it is currently unknown why HEK293 cells expressing GABA_ARs lack modulation by CamKII, it has been postulated that the nonneuronal lineage of the cells might be the root. Alternatively, the lack of relevant anchoring proteins that would allow kinases to associate with GABA_ARs may also be a cause.

3.1.3 PKC

One of the first pieces of evidence that PKC modulates GABA_ARs came from experiments performed on Xenopus oocytes expressing chick brain mRNA. In this study, activation of PKC by phorbol esters decreased the amplitude of GABA_AR-mediated currents (Sigel & Baur, 1988). The reduction in current amplitude was also observed in heterologous cells expressing α1β1γ2S/L and α1β2γ2S subunits. This effect was later discovered to be mediated by phosphorylation of S409/S410 in β1/β2 subunits, S327 in γ2S/γ2L subunits, and S343 in γ2L subunits (Kellenberger, Malherbe, & Sigel, 1992; Krishek et al., 1994). Moreover, the level of modulation by a particular phosphorylation site was site-specific. In particular, S343 in γ2 displayed the greatest effect on the GABA-induced response (Kellenberger et al., 1992; Krishek et al., 1994). In contrast, application of constitutively active PKC in L929 fibroblasts expressing α1β1γ2L enhanced GABA_ARs-mediated currents. These enhancements were prevented by mutations of either β1 S409 to alanine (S409A) or, similarly, γ2L S327A and γ2L S343A (Lin, Angelotti, Dudek, Browning, & Macdonald, 1996; Lin, Browning, Dudek, & Macdonald, 1994). Although the reason for the discrepancy is unclear, there may be numerous explanations: for example, the use of different experimental systems (Houston & Smart, 2006; Mercik, Pytel, & Mozrzymas, 2003), or differential activation of specific PKC isoforms due to the use of phorbol ester versus PKC isolated from bovine brain. Notably, the same phosphorylation sites abolished the observed decreases or increases in GABA_AR currents, which perhaps indicates yet another PKC site may dictate the direction of the response.

Although the majority of studies generally concentrate on synaptic GABA_ARs, more recent evidence indicates that PKC activation may also affect extrasynaptic GABA_ARs. In COS7 cells, the activation of PKC by application of phorbol esters increases the phosphorylation of S443 on the α4 subunit and enhances surface levels of α4β3 receptors (Abramian et al., 2010). Phosphomimetic mutations revealed that the rise in surface levels were due to increased stability, concurrent with enhanced levels of insertion at the cell surface. On the contrary, activation of PKC has also been found to result in decreases in surface levels of GABA_ARs in α4β2δ-expressing HEK293 cells. In parallel, a decrease in tonic GABA_AR inhibition was observed which was dependent on the phosphorylation of S410 on the β2 subunit. Further, PKC activation did not differentially affect S443 mutants of the α4 subunit when compared to wild type (Bright & Smart, 2013). The reasons for the apparent discrepancy are unclear, although they may be explained by the differential phosphorylation and/or recruitment of kinases and phosphatases to the β2 and β3 subunits as will be discussed below. This would suggest that PKC activity could bidirectionally regulate tonic inhibition depending on the β subunit subtype found in a given GABA_AR.

3.1.4 Lessons from expression systems

The phosphorylation sites listed in Table 1a–c are by no means exhaustive; indeed, more recent work using mass spectrometric analysis has identified potential sites of phosphorylation on the β subunits of GABA_ARs (Kang, Heo, & Lubec, 2011). Moreover, the identification of the first phosphorylation site on an α subunit was more recently found on residue S443 of the α4 subunit (Abramian et al., 2010). Collectively, studies in vitro and in heterologous systems have highlighted important considerations when investigating phosphorylation. For example, the actions of kinases are not only receptor subunit specific, but also splice variants allow for additional sites of phosphorylation (McDonald & Moss, 1994; Moss, Doherty, et al., 1992; Whiting et al., 1990). Further, the determination of the existence of a phosphorylation site may depend on the experimental system utilized (Houston & Smart, 2006; Mercik et al., 2003) as well as the experimental conditions (Bright & Smart, 2013). Interestingly, there is the potential for multiple kinases to act on one site, or, alternatively, for a single kinase to be assigned to one particular site. Kinases can also phosphorylate neighboring sites to differentially regulate GABA_ARs (McDonald et al., 1998) and the phosphorylation on one site may affect phosphorylation of another site within the same subunit or on another subunit (Houston et al., 2007). Critically, these studies have made evident that subtype-specific phosphorylation of key residues can govern GABA_AR activity and trafficking (Brandon, Jovanovic, et al., 2002; Kittler & Moss, 2003).

3.2. Divergent effects of kinases and phosphatases on neuronal GABA_ARs

Due to their diverse nature, the study of GABA_ARs in their neuronal environment has been, and still is, a challenge. Nevertheless, a large body of evidence suggests that phosphorylation can affect neuronal activity through the regulation of GABA_ARs. Similar to heterologous systems, β and γ2 subunits are the primary substrates for kinases in neurons. However, the effect of kinases on GABA_ARs in neuronal preparations is complex and often yields conflicting results. Disparate effects of phosphorylation on neurons should probably be expected due to the mixed populations of GABA_ARs composed of various subunits (Pirker, Schwarzer, Wieselthaler, Sieghart, & Sperk, 2000; Poisbeau, Cheney, Browning, & Mody, 1999). Indeed, and as exemplified with PKA, a single kinase can have divergent effects that are dependent on the β-subunit subtype. Moreover, kinases have a considerable number of effects on other substrates that may influence GABA_AR activity, including the activation of signaling pathways and phosphorylation of associated proteins, which can make data interpretation challenging.

For example, PKA activation has been studied in a number of brain regions, and found to both increase and decrease GABA-evoked currents. Decreases in GABA_AR-mediated currents were observed in cultured cerebellar granule cells, hippocampal pyramidal cells, neostriatal neurons, spinal cord neurons, and superior cervical ganglia (Flores-Hernandez et al., 2000; Moss, Smart, et al., 1992; Poisbeau et al., 1999; Porter, Twyman, Uhler, & Macdonald, 1990; Robello, Amico, & Cupello, 1993). However, increases in GABA_AR activity were seen in cerebellar interneurons, hippocampal dentate granule cells and olfactory bulb granule cells (Kano & Konnerth, 1992; Nusser, Sieghart, & Mody, 1999; Poisbeau et al., 1999). Similar inconsistencies are apparent upon activation of PKC, with reports of decreased GABA_AR activity in cortical neurons, hippocampal pyramidal neurons, retinal rod bipolar cells, and thalamic neurons (Brandon et al., 2000; Bright & Smart, 2013; Chou et al., 2010; Gillette & Dacheux, 1996), increases in hippocampal dentate granule cells and hippocampal neurons (Abramian et al., 2010; Poisbeau et al., 1999), and no effect in hippocampal pyramidal cells (Poisbeau et al., 1999).

There are a plethora of reasons why such differences may be observed, from the known heterogeneity of GABA_ARs in both receptor subunit combinations and overall regional distribution to the presence of numerous isoforms of PKC that may mediate differential effects (which may or may not be directed at GABA_ARs). Indeed, studies have reported that PKA and PKC elicit differential effects even within one region—that of the hippocampus. In CA1 pyramidal cells, PKA activation enhanced miniature inhibitory postsynaptic current (mIPSC) amplitudes, whereas PKC activation showed no observable effects. On the other hand, in dentate gyrus granule cells it was PKA that was unresponsive, whereas PKC activation led to increases in peak mIPSC amplitudes (Poisbeau et al., 1999). Although it is tempting to speculate that these observed effects correspond to the prominence of specific GABA_AR subunits, the situation is unclear (Pirker et al., 2000; Wisden, Laurie, Monyer, & Seeburg, 1992), and such conclusions are perhaps even premature considering novel phosphorylation sites are still being discovered (Abramian et al., 2010; Kang et al., 2011).

As well as the aforementioned experiments performed utilizing GST fusion proteins and heterologous expression systems, PKA and PKC have both been demonstrated to phosphorylate purified neuronal GABA_ARs (Brandon et al., 2000; Moss & Smart, 1996). In neurons, β3 and γ2 are basally phosphorylated at S408/S409 and Y365/Y367, respectively (Brandon et al., 2001, 2000; Jovanovic, Thomas, Kittler, Smart, & Moss, 2004). These results support earlier studies utilizing GST fusion proteins as well as those in HEK293 cells (McDonald et al., 1998; McDonald & Moss, 1997). At β3 subunits, activation of PKC leads to increased phosphorylation at S408/S409, while inhibition of PKC led to concomitant decreases in receptor phosphorylation. Of particular note is that the enhancement of β3 phosphorylation by PKA activation in cortical neurons was only observed when PKC activity was inhibited (Brandon et al., 2000).

GABA_ARs constitutively cycle between the neuronal surface and intracellular compartments. The endocytosis of receptors from the neuronal surface is a clathrin-dependent process (Kittler et al., 2000, 2001). There is a wealth of evidence in expression systems that PKC activation also leads to a parallel loss of α1β2γ2 receptors from the surface (Chapell, Bueno, Alvarez-Hernandez, Robinson, & Leidenheimer, 1998; Connolly et al., 1999; Filippova, Sedelnikova, Zong, Fortinberry, & Weiss, 2000; Kittler et al., 2000). However, in heterologous cells, this is unlikely to be due to a direct phosphorylation of GABA_ARs since mutagenesis of all known PKC phosphorylation sites in recombinant receptors (α1, β2 S410A, γ2L S327A/S343A) did not inhibit PKC-mediated decreases in surface expression. Instead, PKC activation prevented receptors that are internalized constitutively from recycling back to the surface (Connolly et al., 1999; Filippova et al., 2000). Whether similar mechanisms are employed in PKC-mediated modifications of GABA_ARs in neurons is unclear. PKC activation resulting in decreases in GABA_AR activity has reported to show variable effects in the level of surface receptors, with observed decreases in cerebellar granule cells (Balduzzi, Cupello, & Robello, 2002; Chou et al., 2010), increases in cortical and hippocampal neurons (Jovanovic et al., 2004), or no change in cortical neurons (Brandon et al., 2000). Additionally, the constitutive cycling of neuronal receptors was observed to occur at a significantly reduced rate when compared to HEK293 cells, which suggests additional receptor regulation and anchoring at the synapse (Connolly et al., 1999). Thus, differential effects observed in neurons may be dependent on regional differences as well as the activation of specific isozymes, which then leads to the regulation of various receptor associated proteins. Indeed, PKCε was observed to form a complex with N-ethylmaleimide sensitive factor (NSF) and γ2-containing GABA_ARs. The activation of PKCε was found to result in decreases in the surface levels of GABA_AR, which was dependent upon PKCε phosphorylation of NSF and their recruitment to inhibitory synapses (Chou et al., 2010). Notably, NSF has also been shown to bind directly to β subunits of GABA_ARs at residues 395–415, which contains the major phosphorylation site conserved within β subunits (Goto et al., 2005). Whether phosphorylation of this site affects GABA_AR–NSF interactions remains to be seen. Although NSF stabilizes excitatory AMPA receptors at the surface by disrupting endocytosis (Hanley, Khatri, Hanson, & Ziff, 2002), similar characterization of NSF and GABA_ARs interactions have not been performed.

4. GABA_AR-INTERACTING PROTEINS AND PHOSPHORYLATION

GABA_AR distribution and expression is under tight subtype-specific management, governed to some degree by its interacting partners. As with the phosphorylation of these receptors, GABA_AR interactions primarily take place at the large intracellular loop, with the majority of interactions occurring through the β and γ2 subunits. These interactions affect trafficking and surface stability as well as the phosphorylation state of specific subunits (Charych et al., 2009; Chen & Olsen, 2007; Jacob et al., 2008; Kneussel & Loebrich, 2007).

4.1. Adaptor protein 2

The adaptor protein (AP2) is a heterotetrameric complex composed of α, β2, μ2, and σ2 subunits. AP2 binds membrane, cargo, and clathrin and is fundamental to clathrin-mediated endocytosis (McMahon & Boucrot, 2011). GABA_ARs β1–3, γ2, and δ subunits are directly associated with the μ2 subunit of AP2. As will be discussed below, thus far three mechanisms for the binding of AP2 to GABA_ARs have been identified (Kittler et al., 2005, 2000; Vithlani & Moss, 2009).

Binding sites important for AP2, clathrin and dynamin-mediated internalization of GABA_ARs were first identified on the β2 subunit. A dileucine motif on the β2 was reported to be required for PKC-mediated endocytosis in heterologous cells (Herring, Huang, Singh, Dillon, & Leidenheimer, 2005; Herring et al., 2003).

A second atypical AP2 binding motif on GABA_ARs was later determined as a region of highly basic amino acids, which contains a major phosphorylation site for PKA and PKC in β1 and β3 subunits, and PKC in β2 subunits (Brandon et al., 2003; Kittler et al., 2005). Phosphorylation of S408/S409 on the β3 subunit resulted in decreased association with the AP2 complex. Thus, when this site is dephosphorylated, AP2 binds GABA_ARs, thereby prompting receptor endocytosis (Fig. 2). Consequently, introducing dephosphorylated β3-derived peptides that compete for the AP2 interaction leads to an increase in mIPSC amplitude and whole-cell current (Kittler et al., 2005). Additionally, mutations of S408/S409 that decrease AP2 binding to the β3 subunit were found to result in a concomitant decrease in endocytosis and an increase in surface-expressed receptors. Interestingly, these mutations resulted in an enhancement in size and number of inhibitory synapses was observed with parallel decreases in the number of excitatory synapses (Jacob et al., 2009).

Figure 2.

Phosphoregulation of GABA_AR endocytosis. The endocytosis of GABA_ARs is regulated by the interaction of the AP2 complex with β and γ2 subunits. Phosphorylation of β3 (S408/S409) and γ2 (Y365/Y367) by PKA/PKC and Src/Fyn, respectively, prevents binding to AP2 and thus stabilizes these receptors at the cell surface. Phosphorylation of β1 and β3 subunits is facilitated by AKAP and RACK1. Dephosphorylation of β3 (S408/409) by PP1 and PP2A, and γ2 by an unknown phosphatase enables binding to AP2, triggering dynamin-dependent clathrin-mediated endocytosis. PRIP regulates dephosphorylation through binding PP1/PP2A and β subunits. The phosphorylation state of PRIP1 determines the release of active PP1α.

Finally, the third AP2 interaction motif on GABA_ARs comprises dual sites in the major intracellular domain of the γ2 subunit. A tyrosine motif Y-x-x-Φ (where x is any amino acid and Φ is hydrophobic) allows binding of AP2 to γ2-containing receptors on Y³⁶⁵GY³⁶⁷ECL. Phosphorylation of Y365/Y367 inhibits binding to AP2, leading to an enhancement in levels of surface receptors (Fig. 2). These motifs can work separately or together to decrease receptor numbers (Kittler et al., 2008). Indeed, phosphorylation within this motif (Y365/Y367) is mediated by Src family members (Brandon et al., 2001; Moss et al., 1995) including Fyn. Moreover, the phosphorylation of Y367 within the γ2 subunit facilitates the direct binding of Fyn kinase (Jurd, Tretter, Walker, Brandon, & Moss, 2010). To this end, Fyn knockout (KO) mice exhibit decreased phosphorylation at this residue in addition to altered GABA_AR function (Boehm, Peden, Harris, & Blednov, 2004; Jurd et al., 2010). Infusion of peptides that block AP2–γ2 interaction in neurons resulted in an increase in mIPSC amplitude accompanied with an increase in receptors at the surface (Kittler et al., 2008). The importance of this site in vivo is highlighted by the embryonic lethality of knock-in mice where γ2 Y365/Y367 residues were mutated to phenylalanines (Y365F/Y367F). This observed phenotype is possibly due to increased levels of GABAergic excitation during early development (Ben-Ari, Khalilov, Kahle, & Cherubini, 2012). Experiments performed with viable heterozygous mice revealed increased surface levels of γ2, enhanced levels of GABA_ARs at specific subdomains of the hippocampus as well as deficits in spatial object recognition (Tretter et al., 2009).

Notably, δ subunits bind AP2 through sites containing both the classical Y-x-x-Φ motif as found in γ2 subunits and the atypical basic motif, which is found in β subunits. Since phosphorylation sites have not currently been identified within the δ subunits, whether this association is also mediated by phosphorylation within these motifs remains to be seen (Gonzalez, Moss, & Olsen, 2012).

4.2. Gephyrin

The important role of gephyrin as a postsynaptic scaffold protein underlying GABA_AR synaptic clustering has been an area of intense study (Luscher, Fuchs, & Kilpatrick, 2011; Tretter et al., 2012; Tyagarajan & Fritschy, 2014). Gephyrin has been shown to bind GABA_AR subunits α1–α3 (Mukherjee et al., 2011; Saiepour et al., 2010; Tretter et al., 2011) and more recently to β2 and β3 subunits (Kowalczyk et al., 2013). Importantly, this interaction may be regulated by the phosphorylation of specific subunits. For example, the mutation of a putative phosphorylation site, T375, in the α1subunit to a phosphomimetic, decreases its affinity togephyrin. The decrease in gephyrin affinity subsequently leads to decreases in synaptic α1-containing clusters and reductions in the amplitude of mIPSCs (Mukherjee et al., 2011). In addition, phospho-deficient mutations of S410 in the β2 subunit lead to a decrease in affinity for gephyrin (Kowalczyk et al., 2013). Whether putative sites such as T375 are altered by kinases and phosphatases are currently unknown, but it raises the intriguing possibility that phosphorylation may dynamically regulate the clustering of GABA_ARs at inhibitory synapses.

4.3. A-kinase anchoring protein

The proper targeting of kinases and phosphatases such as PKA, PKC, and calcineurin (CaN, also known as PP2B) to their appropriate substrates at both excitatory and inhibitory synapses require a family of AKAP scaffolding proteins (Brandon et al., 2003; Klauck et al., 1996; Sanderson & Dell’Acqua, 2011).

The AKAP Yotiao has been shown to bind PKA and the type 1 protein phosphatases (PP1) (Westphal et al., 1999). Furthermore, application of dopamine D4 receptor agonists resulted in decreases in GABA_AR-mediated currents which required an intact Yotiao–PKA–PP1 complex (Wang, Zhong, & Yan, 2002). However, whether these actions are mediated by the direct phosphorylation or dephosphorylation of GABA_ARs is unknown.

AKAP79/150 has been demonstrated to directly bind to GABA_AR β1 and β3, but not β2-subunits. PKA-mediated phosphorylation of the β3 subunit was found to be AKAP dependent (Brandon et al., 2003). However, there has been disagreement as to whether this occurs at the Golgi apparatus instead of at inhibitory synapses in the hippocampus (Lilly, Alvarez, & Tietz, 2005). Nevertheless, more recently, it has been suggested that the PKA–AKAP–CaN signaling complex residing at GABAergic synapses is required for the induction of GABAergic long-term depression in dopamine (DA) neurons of the ventral tegmental area (VTA) (Dacher, Gouty, Dash, Cox, & Nugent, 2013). Unfortunately, specific phosphorylation sites on GABA_ARs have not been characterized in this study, but it is tempting to speculate that direct phosphorylation of these receptors may play a role in DA signaling required for reward-related learning.

4.4. Phospholipase C-related inactive protein

The phospholipase C-related inactive protein (PRIP) family of proteins includes the ubiquitously expressed PRIP2, and PRIP1 that is principally expressed in the CNS. PRIPs were first discovered as proteins that bound to inositol 1,4,5-triphosphate (IP₃) (Kanematsu, Mizokami, Watanabe, & Hirata, 2007; Kanematsu et al., 2000). PRIPs have multiple functions which include: (1) trafficking of γ2-containing GABA_ARs to the cell surface through the ternary binding of PRIPs to GABA_AR and the GABA_AR-associated protein (GABARAP), (2) regulating constitutive AP2- and clathrin-mediated endocytosis of GABA_AR, and (3) modulating GABA_AR phosphorylation (Kanematsu, Fujii, et al., 2007; Kanematsu, Mizokami, et al., 2007).

The ability of PRIP to regulate dephosphorylation occurs through its binding to protein phosphatases PP1α, PP2A, and the β subunits of GABA_ARs (Kanematsu, Mizokami, et al., 2007; Kanematsu et al., 2006; Terunuma et al., 2004; Yoshimura et al., 2001). The PRIP1–PP1α association renders the phosphatase catalytically inactive. However, upon PKA activation, PRIP1 phosphorylation facilitates the dissociation of the PRIP1–PP1α complex. The subsequent release of active PP1α dephosphorylates the β3 subunit specifically at the AP2 binding site, thereby resulting in GABA_AR internalization (Kanematsu et al., 2006; Terunuma et al., 2004; Fig. 2). Further, PRIP1 KO mice displayed increased PP1α activity and functional deficits in PKA-mediated modulation of GABA_ARs (Terunuma et al., 2004). In contrast to PP1α, PRIP1–PP2A association does not alter PP2A activity. Ultimately, the level of β subunit phosphorylation will depend on the balance between the rate of direct phosphorylation of β subunits and the rate of phosphorylation of PRIP1 by PKA.

4.5. Receptor for activated C-kinase

The highly conserved, multifaceted adaptor protein receptor for activated C-kinase (RACK1) binds specifically to activated PKC and enables its trafficking to membrane locales, thereby allowing for PKC phosphorylation at precise receptor sites (Adams, Ron, & Kiely, 2011; Mochly-Rosen, Khaner, & Lopez, 1991; Ron et al., 1994). In addition to binding PKC, RACK1 also interacts with the major intracellular domain of GABA_AR β1 and β3 subunits adjacent to a PKC binding site (Brandon, Jovanovic, Smart, et al., 2002; Brandon et al., 1999). The PKC isoform PKCβII binds directly to β1 and β3 subunits and phosphorylates residues S410 and S408/S409, respectively. Binding assays with GST-β1 fusion proteins revealed that RACK1 bound to residues 395–404, immediately upstream of the PKC binding site (residues 405–415) (Brandon, Jovanovic, Smart, et al., 2002; Brandon et al., 1999). Although association of RACK1 is not essential for PKC binding to β subunits, it potentiates the phosphorylation of β1 subunits at S409.

5. PHOSPHORYLATION AND ALLOSTERIC MODULATION

Pharmacological agents that target GABA_AR for therapeutics are largely positive allosteric modulators used for their anesthetic, anticonvulsant, anxiolytic, or sedative-hypnotic actions. Positive allosteric modulators bind receptors at a site separate from the agonist binding site, enhancing the response of GABA_ARs to GABA. Importantly, allosteric modulation of GABA_ARs by barbiturates, benzodiazepines, and neurosteroids can be regulated by phosphorylation in a kinase- and subunit-specific manner. Further, allosteric modulators may also regulate the phosphorylation state of GABA_ARs.

5.1. Barbiturates and benzodiazepines

Benzodiazepines increase GABA_AR current by increasing channel opening frequency, whereas barbiturates have dose-dependent differences in their effects. At low concentrations, barbiturates act to allosterically enhance the GABA response through increasing channel opening duration, while at higher concentrations they directly activate GABA_ARs (Korpi, Grunder, & Luddens, 2002; Macdonald & Olsen, 1994; MacDonald, Rogers, & Twyman, 1989).

Benzodiazepines bind the interface between α (1, 2, 3, or 5) and γ subunits and regulate GABA_ARs by increasing channel opening frequency upon GABA binding (Goldschen-Ohm, Wagner, Petrou, & Jones, 2010; Jacob et al., 2008; Macdonald & Olsen, 1994). Evidence from synaptosomal preparations suggests CamKII activation results in an increase in benzodiazepine binding to the receptor (Churn et al., 2002). Furthermore, a number of studies have reported that the effects of these drugs can be regulated via PKC activation. Pretreatment of α1β2γ2L-expressing oocytes with PKC activators resulted in increases in the ability of diazepam and pentobarbital to allosterically modulate GABA-induced currents (Leidenheimer, McQuilkin, Hahner, Whiting, & Harris, 1992). However, similar experiments performed on oocytes expressing α1β2γ2 by another group could not reproduce the effect of diazepam or pentobarbital (Ghansah & Weiss, 2001). Nonetheless, further support of PKC regulation of GABA_AR allosteric modulation, albeit in the other direction, stems from PKCε KO mice. These studies showed that mice that lacked PKCε were more sensitive to benzodiazepines and barbiturates (Harris et al., 1995; Hodge et al., 1999). PKCε has been shown to phosphorylate the GABA_AR γ2 subunit at S327. Indeed, in α1β2γ2-expressing HEK293 cells, phospho-null mutation of γ2 S327 enhances the actions of benzodiazepine (Qi et al., 2007). By comparison, PKCγ KO mice did not have altered sensitivity to barbiturates or benzodiazepines (Harris et al., 1995; Hodge et al., 1999). These studies outlined above highlight an added layer of complexity and sophistication, whereby different kinase isozymes specifically and differentially mediate the allosteric modulation of GABA_ARs, presenting a fertile topic for future studies.

5.2. Neurosteroids

Neurosteroids are steroids synthesized de novo in the brain by glia and neurons (Compagnone & Mellon, 2000; Herd, Belelli, & Lambert, 2007). Neurosteroids are potent and selective allosteric modulators of GABA_ARs that increase the channel open duration and frequency at lower concentrations and directly activate GABA_ARs at higher concentrations (Lambert, Cooper, Simmons, Weir, & Belelli, 2009). Besides their allosteric modulation of GABA_ARs, neurosteroids have also been shown to exert their effects through the potentiation of PKC phosphorylation on S443 within α4 subunits of GABA_ARs. Phosphorylation at this residue leads to a sustained up-regulation of α4-containing receptors through insertion of receptors into surface membranes, resulting in a selective enhancement of tonic current in hippocampal neurons (Abramian et al., 2014; Comenencia-Ortiz, Moss, & Davies, 2014). Although others have reported decreases in PKC-mediated tonic inhibition (Bright & Smart, 2013), this work nevertheless highlights a novel mechanism by which neurosteroids can alter neuronal inhibition.

The connection between PKC phosphorylation and neurosteroid modulation of GABA_ARs is complex. A number of studies have reported that neurosteroid modulation of GABA_ARs may be enhanced by PKC phosphorylation (Fáncsik, Linn, & Tasker, 2000; Harney, Frenguelli, & Lambert, 2003; Leidenheimer & Chapell, 1997; Vicini, Losi, & Homanics, 2002), while other studies report that the opposite is true (Hodge et al., 1999, 2002; Kia et al., 2011; Koksma et al., 2003). For example, GABA_AR δ-subunit KO mice display reduced sensitivity to neurosteroids, which could be restored by activation of PKC with phorbol esters (Vicini et al., 2002). Further, inhibition of either PKA or PKC decreased the sensitivity of GABA_ARs to neurosteroids in hippocampal CA1 pyramidal neurons, whereas, in contrast, activation of PKC had no effect (Harney et al., 2003). Interestingly, and in the same preparation, PKC activation did enhance the sensitivity of GABA_ARs to neurosteroid in dentate gyrus granule cells. This suggests that even within a specific brain region, there are likely to be neuron-specific effects that depend upon both the circulating levels of neurosteroids as well as the phosphorylation status of GABA_ARs and/or their associated proteins. As with the enhanced response to benzodiazepine and barbiturate mentioned above, PKCε-deficient mice also exhibit increased sensitivity to neurosteroids (Hodge et al., 1999, 2002). Moreover, these mice were less anxious and had reduced response to stress hormones, consistent with the anxiolytic effects of neurosteroids (Hodge et al., 2002).

Further molecular mechanisms that impact GABA_AR sensitivity to neurosteroids have emerged from studies of the hypothalamic oxytocin neurons of the supraoptic nuclei. These studies showed that substantial changes in activity around the time of parturition can be attributed to changes in circulating levels of progesterone and metabolites (Brussaard & Herbison, 2000; Concas et al., 1998). Recordings from these cells show that pregnant rats during late gestation are neurosteroid-sensitive, but that post-parturition, GABA_ARs become neurosteroid-insensitive (Brussaard, Wossink, Lodder, & Kits, 2000; Koksma et al., 2003). Moreover, these observations were ascribed to modifications in kinase and phosphatase activity. During late pregnancy, activation of PKC with phorbol esters or inhibition of phosphatases PP1 and PP2A reduced neurosteroid modulation of GABA_AR. Conversely, after giving birth, when GABA_ARs are less sensitive to neurosteroid modulation, inhibitors of PP1 and PP2A or activation of PKC rescued neurosteroid sensitivity of GABA_AR (Koksma et al., 2003).

Clearly, the impact of phosphorylation on allosteric modulation of GABA_ARs is complex, and information on the molecular mechanisms that regulate these processes is currently unavailable. Furthermore, it is still unclear from these studies whether kinases target receptors directly or one of its intimately associated proteins, and whether phosphorylation increases binding of allosteric modulators to receptors or whether it influences modulator-induced changes to channel gating. Nevertheless, it is evident that phosphorylation imparts further diversity to the interplay between GABA_ARs and allosteric modulators.

6. SIGNALING PATHWAYS THAT MODULATE GABA_AR PHOSPHORYLATION

Excitation and inhibition in the brain is tightly and dynamically balanced and is crucial for proper brain function. Since GABA_ARs are the principal mediators of inhibition in the brain, it is unsurprising that multiple signaling pathways can regulate the phosphorylation status of these receptors.

6.1. Receptor tyrosine kinases

Receptor tyrosine kinases (RTKs) are an extensive class of cell-surface receptors that are essential to numerous cellular processes (Schlessinger, 2000). In particular, two ligand-receptor pairs of the RTK class include insulin and the insulin receptor, and brain-derived neurotrophic factor (BDNF) and its associated tyrosine kinase receptor (TrkB). Insulin and BDNF have both been implicated in the regulation of GABA_ARs, and will be discussed in more detail below.

6.1.1 Brain-derived neurotrophic factor

The neurotrophin tyrosine kinase receptor 2 (TrkB) exerts its effects through activation by BDNF (Klein et al., 1991). BDNF is abundant and ubiquitously expressed in the brain, playing key roles in neurogenesis, differentiation, survival, and synaptic plasticity (Boulle et al., 2012; Geral, Angelova, & Lesieur, 2013). More specifically, BDNF has been reported to be critical for the development of GABAergic synapses and GABAergic inhibition (Seil, 2003; Vicario-Abejón, Collin, McKay, & Segal, 1998). Studies have implicated BDNF in the increase in cell-surface expression of δ subunits of GABA_ARs. Although this effect was reported to be mediated by TrkB receptors, phospholipase C (PLCγ), and PKC, the substrate for the kinase was unidentified (Joshi & Kapur, 2009). Further, the actions of BDNF on GABA_ARs seem to be developmentally regulated and require the phosphorylation of residues on the β3 and γ2 subunits, as will be discussed below.

Exogenous application of BDNF led to an early, transient potentiation followed by a lasting decrease in GABA_AR-mediated currents in cortical, hippocampal and superior colliculus neurons (Brünig, Penschuck, Berninger, Benson, & Fritschy, 2001; Henneberger, Jüttner, Rothe, & Grantyn, 2002; Jovanovic et al., 2004; Kanematsu et al., 2006). The biphasic nature of the response was attributed to the differential recruitment of PKC and RACK1 that was required for the initial rapid phosphorylation of GABA_AR β3 subunit at S408/S409 (Jovanovic et al., 2004), followed by PRIP1/2 and PP2A mediated dephosphorylation of this site (Kanematsu et al., 2006). Intriguingly, the phosphorylation of S408/S409 has been shown to reduce association of PKC, whereas it increased binding of PP2A, which may explain the transient nature of phosphorylation that has been observed at this site (Brandon, Jovanovic, Smart, et al., 2002; Jovanovic et al., 2004; Kanematsu et al., 2006). In addition, application of BDNF to neurons from PRIP1/2 double knockout (dKO) mice showed a slow consistent increase of β3 phosphorylation complemented with an increase in GABA-induced current. This demonstrated that the initial PKC-dependent phosphorylation persisted, whereas the following PRIP-mediated PP2A dephosphorylation step was abrogated (Kanematsu et al., 2006). As mentioned previously, one would expect that the dephosphorylation of β3 at Ser408/S409 by PP2A would result in AP2 binding and subsequent endocytosis of receptors (Kittler et al., 2005), which would explain the subsequent decrease in mIPSC amplitude. However, the effects of BDNF on surface levels of GABA_AR are in conflict, where both increases (Jovanovic et al., 2004; Kuczewski et al., 2011) and decreases (Kanematsu et al., 2006) have been observed. The reason for this discrepancy is unknown but may be mediated by a number of factors that include methodology, neuronal cell-type, age of neurons and the species that the neurons originated from. Nevertheless, these studies emphasize the importance of BDNF-mediated phosphorylation of GABA_ARs and bring further support to the prominence of RACK1 and PRIPs as facilitators of the phosphorylation and dephosphorylation of GABA_ARs.

In contrast to the regulation of GABA_AR by BDNF in relatively young animals/culture described above, application of BDNF to slices from the prefrontal cortex (PFC) has been shown to cause enhancements in GABA_AR-dependent inhibition to older (2–4 months) animals. BDNF led to increased phosphorylation of Y365/Y367 of γ2 subunits, a subsequent decrease in binding of AP2, and finally, an associated increase in levels of surface receptors. Further, BDNF was no longer able to modulate mIPSCs in mice where these residues had been mutated to Y365F/Y367F. Interestingly, these animals had increased antidepressant phenotype and increased neurogenesis. Although BDNF produced antidepressant-like actions and neurogenesis in wild-type animals, BDNF was unable to further modulate these effects in Y365F/Y367F mice (Vithlani et al., 2013).

Whether these mechanisms are truly different due to development, and not due to, for example, differences in experimental manipulations, deserves future attention. Indeed, BDNF application also increased surface levels of GABA_ARs (Porcher et al., 2011) in young developing cortical cultures at a stage when GABA_ARs are still excitatory (Ben-Ari et al., 2012; Owens, Boyce, Davis, & Kriegstein, 1996). Moreover, an additional complication that may be relevant to these studies is that BDNF also modulates K⁺–Cl⁻ cotransporter activity, which could affect the efficacy of inhibitory transmission (Rivera et al., 2002; Shulga et al., 2008; Wardle & Poo, 2003).

6.1.2 Insulin

Although pancreatic insulin is well recognized for its regulation of blood glucose levels, insulin is also synthesized in the brain (Havrankova, Brownstein, & Roth, 1981) where it regulates synaptic plasticity, spine and dendritic morphogenesis, and neuronal survival (Bassil, Fernagut, Bezard, & Meissner, 2014; Chiu & Cline, 2010). Insulin treatment results in a rapid increase in surface levels of GABA_ARs and an enhancement of mIPSC amplitude (Fujii et al., 2010; Vetiska et al., 2007; Wan et al., 1997; Wang, Liu, Pei et al., 2003). Work on the mechanisms regulating these effects has so far concentrated on the β2 subunit of GABA_ARs. More specifically, the presence of insulin results in the phosphorylation at residues Y372 and Y379 of the β2 subunit of GABA_ARs by an unidentified kinase and these residues were critical for the increases in surface levels of GABA_ARs upon stimulation with insulin. Furthermore, enhancement of GABA_AR-mediated currents was found to be phosphoinositide 3-kinase (PI3K) dependent and correlated with an increase in binding between PI3K p85 subunit SH2 domain and GABA_ARs specifically at these phosphorylated residues on the β2 subunit (Vetiska et al., 2007).

Elsewhere, other groups have shown that the observed increases in surface receptors required the well-documented activation by insulin of the PI3K-Akt pathway (Hemmings & Restuccia, 2012; Wan et al., 1997; Wang, Liu, Pei, et al., 2003). Activated Akt solely phosphorylated the S410 residue on the β2 subunit in α1β2γ2-containing HEK293 cells. Insulin stimulation of Akt in α1β2γ2-expressing HEK293 cells led to an enhancement in surface receptors and an increase in the amplitude of GABA_AR-mediated mIPSCs (Wan et al., 1997). Indeed, the phosphorylation of this residue by Akt was also crucial for the observed translocation of GABA_AR in cultured neurons (Wang, Liu, Pei, et al., 2003). Notably, upon insulin application, activated Akt translocated specifically to more distal dendritic locales where it colocalized with GABA_ARs. In addition, PRIP1 was required to traffic active Akt to GABA_ARs upon stimulation by insulin, since disruption of this complex in PRIP dKO mice or PRIP interference peptide in wild-type mice resulted in a dearth in phosphorylated β subunit. This was also paralleled with a lack of insulin-mediated potentiation of GABA-induced currents (Fujii et al., 2010). The S410 phosphorylation site in the β2 subunit is conserved (in β1 S409 and β3 S408/S409), and again one would presume that its phosphorylation may result in less binding to AP2, thereby increasing surface levels of receptor. Instead, inhibition of ER to Golgi trafficking by brefeldin A resulted in a decrease in GABA-evoked currents beyond that of controls. This suggests that increases in surface GABA_AR are mediated by insertion of newly synthesized receptors concurrent and to a lesser degree with a decrease in endocytosis (Fujii et al., 2010). Whether these mechanisms also apply to other β subunits has not been explored.

Thus, the insulin-mediated effects hinge on the phosphorylation of GABA_ARs by two seemingly different mechanisms. How these processes may converge will be a fascinating topic for future investigations. Moreover, these studies advocate phosphorylation of a specific site that has differential consequences and is contingent on the kinase concerned.

6.2. Glutamate receptors

The vast majority of synapses use glutamate as their excitatory neurotransmitter. Glutamate release from presynaptic terminals triggers rapid activation of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) and N-methyl-D-aspartate receptors (NMDARs). This activation leads to AMPAR and NMDAR opening to allow an influx of cations, and stimulation of the postsynaptic cell. The strength of synaptic transmission is extremely plastic and can be increased (long-term potentiation, LTP) or decreased (long-term depression, LTD) (Malenka & Bear, 2004). The polarity of this plasticity is largely defined by the kinetics and amount of Ca²⁺ influx through NMDARs. Strong, brief, high frequency stimulation (HFS) of the input neuron induces rapid and robust Ca²⁺ entry into cells, resulting in LTP. Additionally, weak, prolonged, low frequency stimulation, allowing for lower concentrations of Ca²⁺ entry, results in LTD (Bliss & Lømo, 1973; Dudek & Bear, 1992; Yang, Tang, & Zucker, 1999). The difference in the sensitivities of kinases and phosphatases to Ca²⁺ provides one of the mechanisms that underlie this bidirectional expression of plasticity (Lee, 2006). Together, LTP and LTD are thought to represent the cellular correlate for learning and memory in the CNS (Bear & Abraham, 1996; Bliss & Collingridge, 1993).

Likewise, similar long-term changes in synaptic strength occur at GABAergic synapses (Gaiarsa, Caillard, & Ben-Ari, 2002). NMDAR-dependent HFS that induces LTP at excitatory synapses in the hippocampus also results in the concomitant LTD (LTD_GABA) at inhibitory synapses (Lu, Mansuy, Kandel, & Roder, 2000). The decrease in efficacy of GABAergic transmission was accompanied with the dephosphorylation at S327 in the γ2 subunit of GABA_ARs and was dependent upon the binding of the CaN catalytic domain to the intracellular loop of γ2 subunits (Wang, Liu, Haditsch, et al., 2003). Moreover, increases in excitatory activity resulting in NMDA-dependent Ca²⁺ influx lead to the fast and reversible diffusion of GABA_ARs from synaptic clusters and was found to be independent of receptor endocytosis (Bannai et al., 2009; Muir et al., 2010). The lateral diffusion of receptors was contingent upon CaN-dependent dephosphorylation of S327 on the γ2 subunit (Fig. 3) (Muir et al., 2010).

Figure 3.

Phosphoregulation of GABA_AR trafficking. GABA_AR are postsynaptically localized through its binding to gephyrin. (A) Strong NMDAR activation results in the influx of Ca²⁺. CamKII activation at excitatory synapses results in an increase of AMPARs at the cell surface. Calcineurin dependent dephosphorylation of GABA_AR subunit γ2 (S327) enables diffusion of GABA_ARs away from the synapse. (B) Weaker stimuli induced by an NMDAR-dependent chemLTD protocol results in a moderate influx of Ca²⁺ and CamKII is targeted to inhibitory synapses. CamKII phosphorylation of β3 (S383) is critical for recruitment of gephyrin to synaptic sites, immobilization of GABA_ARs at synapses and an enhancement of cell surface levels of GABA_ARs.

Conversely, NMDAR-dependent chemically induced LTD (chemLTD) of excitatory transmission results in an increase of GABA_ARs at the surface and an enhancement of inhibitory transmission (LTP_GABA). As has been found in excitatory synapses, the bidirectional regulation of GABA_ARs is dependent upon the strength of NMDAR activation and Ca²⁺ influx. Strong NMDAR activation leads to CamKIIα translocation to excitatory synapses, enhancing surface levels of AMPARs, whereas targeting of CamKIIα is inhibited at inhibitory synapses by CaN. Weaker stimuli resulted in CamKIIα localization at inhibitory synapses and an enhancement of surface levels of GABA_ARs (Marsden, Beattie, Friedenthal, & Carroll, 2007; Marsden, Shemesh, Bayer, & Carroll, 2010). More recent evidence lends further support to the role of GABA_AR phosphorylation in these mechanisms. Similar induction of chemLTD resulted in a CamKII-dependent LTP_GABA, marked by increases in surface levels of GABA_ARs and the potentiation in the mIPSC and spontaneous inhibitory synaptic current amplitude. During LTP_GABA, GABA_ARs were immobilized at the synapse and the scaffold protein gephyrin was recruited from nearby extrasynaptic sites. Moreover, the phosphorylation of β3 at S383 increased during LTP_GABA. Phosphorylation of β3 S383 was required for GABA_A Rimmobilization at synapses and the recruitment of gephyrin to synaptic sites. Thus, the phosphorylation of β3 S383 is considered critical for the expression of LTP_GABA. Finally, similar accumulation of GABA_ARs and gephyrin is observed in vivo following an experience-dependent plasticity protocol (Petrini et al., 2014; Fig. 3).

6.3. Voltage-gated Ca²⁺ channels

In cerebellar Purkinje neurons, activation of excitatory synapses induce long-lasting enhancement of GABA_AR-mediated inhibitory current. This “rebound potentiation” (RP) occurs through activation of voltage-gated Ca²⁺ channels (VGCCs) resulting in a transient Ca²⁺ influx and is dependent upon CamKII (Kano, Kano, Fukunaga, & Konnerth, 1996) and γ2 subunit association with GABARAP (Kawaguchi & Hirano, 2007). Further, RP can be suppressed by GABA type B receptor (GABA_BR) activation through decreasing levels of PKA. This reduction of PKA ultimately leads to an increase in PP1 activity through the PKA/DARPP-32/PP1 signaling pathway resulting in the inhibition of CamKII (Hirano & Kawaguchi, 2014; Kawaguchi &Hirano, 2002). Since PKA and CamKII are both kinases of the β1 and β3 subunit, and because the PKA/DARPP-32/PP1 signaling pathways have previously been reported to increase the phosphorylation of β1 and β3 subunits (Flores-Hernandez et al., 2000), it is possible that direct phosphorylation of these subunits may underlie RP in these neurons.

Activation of L-type VGCCs in hippocampal preparations results in CamKII-dependent increases in cell surface α5β3-containing receptors and an enhancement of tonic current. Moreover, these observations were contingent on the phosphorylation of S383 on the β3 subunit, which subsequently led to the insertion of new receptors to the surface without affecting endocytosis. Increases or decreases in neuronal activity also resulted in respective increases and decreases in phosphorylated β3 and surface expression of GABA_ARs. Thus, neuronal activity can bidirectionally regulate surface-expressed GABA_ARs and tonic inhibition through the phosphorylation of a single residue on the β3 subunit (Saliba, Kretschmannova, & Moss, 2012).

6.4. Dopamine

Both fast synaptic glutamatergic (excitatory) and GABAergic (inhibitory) transmission can be modulated through G-protein-coupled mechanisms or by direct interactions with the comparatively slower neurotransmission of dopaminergic receptors. DA receptors are broadly classified into two distinct groups, the D1-like (D1 and D5) and D2-like (D2–4) classes of receptors. This distinction was based on the initial observations that DA could modulate adenylyl cyclase activity resulting in increased (D1-like) or decreased (D2-like) cyclic AMP (cAMP) production required for the subsequent activation of PKA (Beaulieu & Gainetdinov, 2011; Greengard, 2001). Since dopamine receptors affect PKA activity, evidence for a role of DA-mediated phosphorylation of GABA_ARs is largely derived from studies on the β subunits (Chen, Kittler, Moss, & Yan, 2006; Flores-Hernandez et al., 2000; Goffin et al., 2010; Terunuma et al., 2004).

DA can modulate the excitability of striatal medium spiny neurons (MSNs) expressing a tonic GABA_AR current via a PKA and β3 subunit-dependent mechanism (Janssen, Ade, Fu, & Vicini, 2009; Janssen, Yasuda, & Vicini, 2011). In neostriatal neurons, DA activation of D1 and D5 receptors both decreased and enhanced GABA-evoked currents in MSNs (Flores-Hernandez et al., 2000) and cholinergic interneuron (Yan & Surmeier, 1997), respectively.

In particular, application of a D1 agonist in adult MSNs resulted in a PKA-dependent reduction in GABA-induced current. Membrane permeable analogues of cAMP mimicked the attenuation of the GABA-evoked response, which was further reduced by inhibition of PP1/PP2A. Additionally, DA was observed to increase the phosphorylation of β1 and β3 subunits (although β1 subunits were more prevalent), which was dependent upon DA- and cAMP-regulated phosphoprotein, 32 kDa (DARPP-32). Thus, the observed decrease in GABA-induced current requires activation of the PKA/DARPP-32/PP1 signaling pathway, leading to the modulation of GABA_ARs through the phosphorylation of β1 subunits (Flores-Hernandez et al., 2000).

In hippocampal cells, the activation of D1 receptors led to a PKA- and PRIP-dependent enhancement of GABA-evoked currents and a concomitant increase in phosphorylation of β3 residues S408/S409 (Terunuma et al., 2004). Thus, these studies lend support to aforementioned work in heterologous systems whereby PKA was observed to reduce GABA-induced currents by phosphorylation of S409 in the β1 subunit and increase currents by phosphorylation of two adjacent residues S408/S409 on the β3 subunit (McDonald et al., 1998).

In addition to D1-like receptors, D3 receptor activation in the MSNs of the nucleus accumbens was shown to reduce GABA_AR-mediated currents concurrent with increased internalization and a decrease in surface receptors. These effects were dependent on cAMP, PKA, the β subunit of GABA_AR as well as clathrin-dependent endocytosis (Chen et al., 2006). In this case, the activation of D3 receptors leads to reduced PKA activity resulting in the dephosphorylation of β subunits and the subsequent endocytosis of these receptors, which reduces GABA_AR function.

Activation of D2-like receptors in the VTA induces LTD of GABAergic synapses and was dependent on the clathrin-mediated endocytosis of GABA_ARs. Moreover, these observations required an IP³-dependent rise in Ca²⁺, activation of CaN, and parallel inhibition of PKA. Finally, disrupting the AKAP–PKA complex mimicked LTD_GABA (Dacher et al., 2013).

6.5. Others

In addition to its ability to suppress RP, GABA_BR activation increases α4βδ- and α6βδ-mediated tonic current in thalamocortical and dentate gyrus granule cells, respectively (Connelly, Errington, & Crunelli, 2013a). This effect was further dependent upon δ-containing receptors and PKA inhibitors mimicked GABA_BR-induced enhancements (Connelly, Errington, Di Giovanni, & Crunelli, 2013; Naylor, Liu, Niquet, & Wasterlain, 2013).

Studies of serotonergic neurotransmission in the PFC show that activation of 5-HT₂ receptors decreases GABA-evoked currents, which is dependent upon PKC–RACK1 association and results in the PKC-dependent phosphorylation of GABA_AR γ2 subunit (Feng, Cai, Zhao, & Yan, 2001). In contrast, activation of 5-HT₄ receptors resulted in the reversible, bidirectional modulation of GABA-induced currents determined by basal PKA activity (Cai, Flores-Hernandez, Feng, & Yan, 2002).

Collectively, it is clear that multiple signaling pathways can, or have the potential to, orchestrate the phosphorylation state of specific residues on GABA_ARs. Further, these signaling molecules allows for specific modification determined by intricate cues and presents an additional layer of complexity in the regulation of GABA_AR activity.

7. DYSREGULATION OF GABA_AR PHOSPHORYLATION IN DISEASE

Although compromised trafficking of GABA_ARs are thought to play key roles in a number of pathological conditions (Benarroch, 2007; DeLorey & Olsen, 1999; Krystal et al., 2006; Rudolph & Mohler, 2004; Thompson-Vest et al., 2003), evidence for the importance of phosphorylation in these studies is relatively scarce.

7.1. Ischemia

Excessive release of glutamate in cerebral ischemia results in excitotoxicity and cell death (Lipton, 1999; Lo, Dalkara, & Moskowitz, 2003). Although the majority of research has focused on reducing the effects of the glutamatergic system, GABA_AR trafficking is also considerably modified, which may exacerbate these effects (Schwartz-Bloom & Sah, 2001). Indeed, decreases in surface levels of GABA_ARs and an enhancement of ubiquitin-dependent lysosomal degradation have been reported in the in vitro oxygen-glucose deprivation (OGD) model of ischemia (Arancibia-Carcamo & Kittler, 2009; Liu et al., 2010; Mielke & Wang, 2005). Interestingly, the association of the α1 subunit to gephyrin decreased and could be rescued with a CaN inhibitor. Inhibitors of PP1α/PP2A rescued the loss of α1 subunit by OGD. Further, ischemic insult resulted in decreased phosphorylation of S408/S409 of the β3 subunit and phosphomimetic mutants of β3 or mutants that blocked AP2 binding protected cells from neuronal death (Mele, Ribeiro, Inacio, Wieloch, & Duarte, 2014; Smith et al., 2012).

7.2. Epilepsy

Epilepsy is a common and often devastating neurological disorder based on a striking imbalance between excitatory and inhibitory activity. Prolonged, continuous seizures (status epilepticus, SE) in animal models and humans can induce the development of temporal lobe epilepsy (TLE). Modifications in GABA_AR trafficking and expression have been reported in patients and animal models of SE and TLE (Brooks-Kayal, Shumate, Jin, Rikhter, & Coulter, 1998; Goodkin, Joshi, Mtchedlishvili, Brar, & Kapur, 2008; Loup, Wieser, Yonekawa, Aguzzi, & Fritschy, 2000; Naylor, Liu, & Wasterlain, 2005; Sperk, Drexel, & Pirker, 2009; Terunuma et al., 2008). Increases and decreases in GABA_AR expression have been observed, both of which are dependent upon specific subunits and the time period of epileptogenesis studied. Loss of GABA_ARs is thought to be one mechanism that underlies pharmacoresistance to benzodiazepines (Deeb, Maguire, & Moss, 2012). Furthermore, dephosphorylation of GABA_ARs has been implicated in the loss of these receptors during SE. In particular, induction of SE by pilocarpine resulted in the decrease of PKC-mediated phosphorylation of β3 S408/S409 residues. This dephosphorylation enhanced binding to AP2, resulting in receptor endocytosis during SE (Terunuma et al., 2008). Moreover, other reports have suggested kainate and pilocarpine-induced SE results in an NMDAR-dependent increase in CaN activity and expression resulting in the dephosphorylation of β2/3 subunits of GABA_ARs. Whether this phosphorylation status also resulted in changes to surface expression was not explored (Kurz et al., 2001; Wang, Chi, Wang, Wang, & Sun, 2009).

7.3. Drug abuse

Although ligands such as benzodiazepines may have clinically important uses, there is substantial evidence for a role of GABA_ARs in the regulation of the addictive properties of drugs of abuse (Kalivas, 2007; Tan, Rudolph, & Lüscher, 2011; Trudell, Messing, Mayfield, & Harris, 2014).

Increased BDNF levels after cocaine withdrawal results in decreased GABAergic inhibition and reduced GABA_ARs surface expression, concurrent with the facilitation of activity-induced LTP. These effects were mediated by the BDNF–TrkB–PP2A signaling pathway, which also resulted in decreases in phosphorylated β3 subunits. These effects have been suggested to underlie behavioral modifications after cocaine withdrawal (Lu, Cheng, Lim, Khoshnevisrad, & Poo, 2010). Application of BDNF has reported to lead to lasting decreases in GABA_AR-mediated currents through the PRIP1/2 and PP2A mediated dephosphorylation of β3 S408/S409 (Kanematsu et al., 2006), but whether similar mechanisms apply here remains to be seen.

GABA_AR subunits composition could be altered reversibly within minutes after a single intoxicating dose of alcohol, resulting in decreased surface levels of extrasynaptic α4 and δ-containing GABA_ARs (Gonzalez et al., 2012; Liang et al., 2007; Shen et al., 2011). This may contribute to the rapid tolerance observed after acute ethanol exposure (LeBlanc, Kalant, & Gibbins, 1975; Liang et al., 2007; Ludvig, George, Tang, Gonzales, & Bungay, 2001; Wallace et al., 2007). Further changes in subunit expression were observed at later time points with increases in α4 and γ2 subunits and decreases in α1 and δ subunits (Liang et al., 2007). Moreover, these changes persisted after chronic intermittent administration of alcohol followed by withdrawal (Liang et al., 2007, 2006). The ethanol-induced decreases in δ-containing receptors were attributed to increased AP2 binding and subsequent endocytosis (Gonzalez et al., 2012). Chronic ethanol exposure was also found to lead to changes in PKCγ association with GABA_AR subunits, with decreases in α1 subunit association and enhancements in α4 subunit association. The reduction in PKCγ-α1 binding correlated with increased α1 subunit in clathrin-coated vesicles (Kumar, Kralic, O’Buckley, Grobin, & Morrow, 2003; Kumar, Sieghart, & Morrow, 2002). Further, inhibition of PKCγ (but not PKCβ) prevented ethanol-mediated modulations in α1 and α4 subunit expression (Kumar et al., 2010; Werner et al., 2011). Intriguingly, PKC phosphorylation of α4 at S443 has been shown to increase receptor numbers at the surface (Abramian et al., 2010). Whether this site can also be modulated through ethanol administration remains to be seen.

Phosphorylation of GABA_AR may well underlie some of these changes, but evidence showing direct modifications of GABA_AR phosphorylation states by ethanol comes from transgenic mice lacking specific isoforms of PKC. As well as altering the previously discussed positive allosteric effects of benzodiazepines, PKCε also modifies the allosteric effects of alcohol. As with benzodiazepines, PKCε KO mice display increased sensitivity to the effects of ethanol as well as a reduction in phosphorylated γ2 S327. In α1β2γ2 expressing HEK293 cells, phospho-null mutation of this site enhanced the effect of ethanol (Hodge et al., 1999; Qi et al., 2007). It is interesting to note that PKCε KO animals self-administered considerable less ethanol compared to their wild-type counterparts (Hodge et al., 1999; Olive, Mehmert, Messing, & Hodge, 2000). Strikingly, the modulation of ethanol sensitivity and self-administration in PKCε KO mice could be rescued to wild-type levels by suppression of the transgene (Choi, Wang, Dadgar, Chang, & Messing, 2002).

PKCδ KO mice are more resilient to the intoxicating effects of ethanol, an effect that is believed to be mediated by extrasynaptic, δ subunit-containing GABA_ARs. PKCδ KO mice showed no response to doses of ethanol which were shown to increase GABA-mediated tonic currents in thalamic and hippocampal neurons of wild-type animals (Choi et al., 2008). Although direct phosphorylation of δ subunits of GABA_ARs by PKCδ has been suggested, as yet, there is no direct evidence of δ subunit phosphorylation.

8. CONCLUSION

There has been a great deal of progress in understanding the molecular mechanisms that regulate GABA_ARs. The significance of the dynamic regulation of GABA_ARs by phosphorylation is unquestionable. Notwithstanding, there are still considerable gaps and inconsistencies in our knowledge as exemplified in the large number of conflicting results that have arisen from activation of kinases in neuronal preparations.

GABA_AR may be phosphorylated on multiple sites and on multiple subunits, but our knowledge of how the phosphorylation of a specific site may impact on other sites within the same subunit, or on different subunits, is tenuous. Additionally, it is not certain how phosphorylation on multiple sites impacts the fate of receptors. Ultimately, addressing such gaps in our knowledge will provide much-needed clarity to the many differences that are observed in the roles of phosphorylation.

GABA_AR-interacting proteins allow a further refinement of its regulation. Given the plethora of different GABA_AR subunits, it is perhaps surprising that there is such a paucity of known interacting proteins. Thus, possible interactions that may be regulated by phosphorylation remain virtually unknown. In particular, α subunits are the most diverse of the GABA_ARs subtypes, and differentiating the particular interaction partners of specific α subunits may tell us, for example, how different subunits are targeted to particular subcellular regions. Of the α subtypes, only α4 has been recently shown to be phosphorylated at a particular residue. However, the established phosphorylation sites are unlikely to be comprehensive, and with this, it is likely that more phospho-dependent interactions will emerge. Indeed, more recently discovered sites (Abramian et al., 2010; Kang et al., 2011) suggest that seemingly disparate results in the activation of kinases could simply arise from gaps in our current knowledge. Therefore, further examination into this area is necessary in order to enhance our current knowledge on this topic.

Evidence suggests that phosphorylation can increase insertion of receptors to the membrane (Abramian et al., 2014, 2010) and although the mechanisms of GABA_ARs endocytosis are becoming more apparent, our understanding of receptor recycling and insertion is still in its infancy. It will be important to determine the consequence of these internalized receptors, and the phosphorylation of GABA_ARs seems ideally poised to mediate these trafficking decisions.

Studies on the impact of ethanol clearly demonstrate the need to selectively examine specific PKC isoforms to understand their role in the phosphorylation of GABA_ARs (Song & Messing, 2005). Moreover, the different effects of kinases and phosphatases could also be due to differences in development and/or regional expression of specific GABA_AR subunits, kinases, phosphatases and/or associated proteins. Ultimately, entire pathways will have to be collectively examined to refine the full picture. Further challenges arise from the multiple signaling pathways that converge upon GABA_ARs, yielding another level of complexity. Indeed, the effects of BDNF on GABA_ARs are far from clear, probably due to developmental and region-specific expression of both proteins (Ip, Cheung, & Ip, 2001; Mizoguchi, Ishibashi, & Nabekura, 2003; Sathanoori et al., 2004; Webster, Herman, Kleinman, & Shannon, 2006).

The overwhelming complexity of GABA_ARs is multifaceted, owing to their diversity in subunit composition, differential localization and expression as well as posttranslational modifications. However, it is exactly this mechanistic complexity that makes GABA_ARs such opportune drug targets, since it favors the intricate tuning by pharmacological interventions. Accordingly, further research into this diversity through mechanisms of GABA_AR phosphorylation will surely clarify matters of contention and provide new and exciting insights to understanding neuronal function in health and disease.