Up-regulation of GABAB Receptor Signaling by Constitutive Assembly with the K+ Channel Tetramerization Domain-containing Protein 12 (KCTD12)

Background: KCTD12 regulates agonist potency and kinetics of GABAB receptor signaling. Results: KCTD12 stably associates with GABAB receptors as early as in the endoplasmic reticulum and reduces constitutive receptor internalization from the cell surface. Conclusion: KCTD12 increases GABAB receptor surface expression and the magnitude of receptor signaling. Significance: KCTD12 not only regulates agonist potency and kinetics but also the magnitude of GABAB receptor signaling.

GABA B receptors play an important role in regulating neuronal excitability in the mammalian central nervous system. Accordingly, GABA B receptors have been implicated in a variety of neurological and psychiatric conditions, including epilepsy, anxiety, depression, schizophrenia, addiction, and pain (1,2). Native GABA B receptors are reported to comprise principal and auxiliary subunits that assemble into molecularly and functionally distinct receptor subtypes (3)(4)(5)(6). The principal subunits GABA B1a , GABA B1b , and GABA B2 have a typical seven-transmembrane domain topology and form fully functional heteromeric GABA B(1a,2) and GABA B(1b,2) receptors that regulate Kir3-type K ϩ channels, voltage-gated Ca 2ϩ channels, and adenylate cyclase. It is well established that newly synthesized principal subunits assemble into heteromeric GABA B (1,2) receptors in the ER. 3 The masking of the ER retention motif RSRR in GABA B1 by GABA B2 ensures that only GABA B1 subunits that have assembled with GABA B2 subunits are efficiently routed out of the ER to the cell surface (1,7,8). The subunit isoform GABA B1a contains targeting motifs in its primary sequence that traffic receptors to axonal sites (3). In contrast, the subunit isoform GABA B1b traffics receptors to the somato-dendritic compartment (3). The auxiliary subunits KCTD8, -12, -12b, and -16 (named after their K ϩ channel tetramerization domain) form tetramers that bind to the C-terminal cytoplasmic domain of GABA B2 , in which mutation of a tyrosine residue to alanine (GABA B2 Y902A) completely abolishes KCTD binding (5). The KCTDs increase agonist potency at the receptor and regulate rise time and desensitization of receptormediated K ϩ and Ca 2ϩ current responses (5). GABA B receptors are examples of GPCRs that are composed of principal and auxiliary subunits. Other examples are members of the calcitonin receptor family that associate with receptor activitymodifying proteins (9).
Prototypical protein complexes that are regulated by auxiliary subunits are voltage-gated (10) and ligand-gated ion channels (11,12). Auxiliary subunits of ion channels directly and stably interact with a pore-forming subunit and modulate channel properties and/or surface trafficking of the channel complex. Auxiliary subunits of ion channels normally assemble with pore-forming subunits in the ER. Assembly in the ER is also observed with the receptor activity-modifying proteins, which promote cell surface expression of the associated GPCRs (9). Little is known about the interactions of auxiliary and principal subunits during the lifecycle of GABA B receptors. For example, it is unknown where in the biosynthetic pathway auxiliary KCTD subunits assemble with principal subunits and whether the KCTDs influence surface trafficking or internalization of receptors. Likewise, it still has to be addressed whether the KCTDs stably associate with the receptor, generally considered a criterion to qualify as an auxiliary receptor subunit (12). Here, we show that KCTD12 associates with GABA B2 while the receptor resides in the ER. Moreover, we show that KCTD12 up-regulates receptors at the cell surface, presumably by reducing constitutive receptor internalization. Finally, we found that KCTD12 remains associated with the receptor upon receptor activation and during receptor internalization. In summary, our data support that KCTD12 is a stably associated auxiliary GABA B receptor subunit that increases the magnitude of receptor signaling at the neuronal membrane.

Metabolic [ 35 S]Methionine
Labeling-For immunoprecipitation of FLAG-KCTD12, COS-1 cells were plated onto collagencoated 100-mm dishes and starved in methionine-free DMEM (Sigma) for 30 min at 37°C. Cells were then incubated in 3 ml of 100 Ci of EXPRESS 35 S protein labeling mix (PerkinElmer) for 15 min at 37°C. Cells were rinsed with ice-cold PBS and harvested in 0.4 ml of lysis buffer (150 mM NaCl, 10 mM Tris-HCl, 5 mM EDTA, 1.5% Nonidet P-40) supplemented with protease inhibitors (complete Mini, Roche Applied Science). After the addition of 30 l of 50% protein A-agarose (Roche Applied Science) with 2 l of mouse anti-FLAG antibodies (Sigma), the lysate was incubated overnight at 4°C. Agarose beads were washed in radioimmune precipitation buffer (150 mM NaCl, 10 mM Tris-HCl, 5 mM EDTA, 1.0% Nonidet P-40, 0.1% SDS, 0.5% deoxycholate). Immunoprecipitated FLAG-KCTD12 and Myc-GABA B2 proteins were identified by cross-correlation of autoradiography with Western blots using rabbit anti-FLAG and rabbit anti-Myc antibodies.
Cell Surface Immunoprecipitation-COS-1 cells expressing GABA B receptors comprising Myc-tagged GABA B1 subunits were blocked with PBS containing 1% BSA and incubated with mouse anti-Myc antibodies (1:200) on ice for 3 h. After several washes with PBS, receptors were activated with 100 M baclofen for 15 min. Cells were lysed, and protein G-agarose (Roche Applied Science) was added to the lysates. Immunoprecipitated proteins were resolved by SDS-PAGE and detected on Western blots. Detection of actin with mouse antibodies (1:1000, C4, Millipore) confirmed that surface receptors were immunoprecipitated.
BRET Measurements-COS-1 transiently transfected with plasmids encoding Rluc and/or YFP fusion proteins were seeded into 96-well microplates (Greiner Bio-One) for 24 h. The Rluc substrate coelenterazine h (NanoLight Technologies) and the GABA B receptor agonist baclofen were added for 15 min at a final concentration of 5 and 100 M, respectively. Luminescence and fluorescence signals were detected sequentially using an Infinite F500 microplate reader (Tecan). The BRET ratio was calculated as the ratio of the light emitted by GABA B1 -YFP (530 -570 nm) over the light emitted by Rluc-KCTD12 or Rluc-KCTD10 (370 -470 nm). The BRET ratio was adjusted by subtracting the ratios obtained when Rluc fusion proteins were expressed alone. Total YFP fluorescence was measured with an excitation filter at 485 nm and an emission filter at 535 nm and corrected for the fluorescence measured in cells expressing Rluc fusion proteins in isolation. The results were expressed in milliBRET units, corresponding to the BRET ratio values multiplied by 1000. Each data point was obtained using quadruplicate wells of cells. Data were analyzed using GraphPad Prism 5.0 software.
Biotinylation Assay-Constitutive internalization of GABA B receptors in transiently transfected COS-1 cells was studied as described (20). For detection of cell surface proteins, hippocampal neurons on 35-mm plates were incubated with 1 mg/ml Sulfo-NHS-SS-Biotin (Pierce) in PBS for 30 min at 4°C. After quenching the biotinylation reaction with glycine and rinsing of the cells with ice-cold PBS, cells were scrapped from the plates and lysed. Biotinylated surface proteins were purified using NeutrAvidin-agarose (Pierce), washed, and resuspended in protein loading buffer. Protein samples were resolved by SDS-PAGE and Western blots carried out with rabbit anti-GABA B1 , rabbit anti-KCTD12 (1:1000 (5)), and mouse anti-␤tubulin (1:3000, Sigma) antibodies.
Electrophysiology-Transfected CHO-K1 cells and cultured hippocampal neurons were plated onto Thermanox TM plastic coverslips (Nunc) or poly-L-lysine-coated glass coverslips, respectively. Electrophysiological recordings were performed as described (5). A fast superfusion system for drug application was used (23). For recordings and data analysis, we used pClamp10 software (Molecular Devices).
Data Analysis-With exception of the BRET data, which are expressed as Ϯ S.D., all data are presented as mean Ϯ S.E. If not stated otherwise, Student's t test with GraphPad Prism 5.04 was used. The band intensity on Western blots was quantified on unsaturated images using luminescent image analyzer LAS-4000 (Bucher Biotec AG, Basel) and ImageJ version 1.43 software (National Institutes of Health).

KCTD12 Increases Cell Surface Expression of GABA B
Receptors-We first analyzed whether the KCTDs influence GABA B receptor surface expression. We expressed Myc-GABA B1 and HA-GABA B2 in the presence or absence of FLAGtagged KCTD8, -12, -12b, and -16 in COS-1 cells. We used the extracellular Myc tag of GABA B1 to monitor with an ELISA the amount of GABA B receptors at the cell surface. KCTD12 significantly promoted receptor surface expression (196.2 Ϯ 16.5%, p Ͻ 0.001; Fig. 1A). In contrast, KCTD8, -12b, and -16 had no significant effect on receptor surface expression. As a control, KCTD10, a KCTD protein that does not bind to GABA B receptors (5), had no effect on receptor surface expression either. Moreover, receptors without a KCTD-binding site (GABA B2 Y902A (5)) exhibited no increase in surface expression in the presence of KCTD12 (Fig. 1A). Increased receptor surface expression therefore requires direct binding of KCTD12 to GABA B2 . As expected, we observed low cell surface expression of GABA B1 in the absence of GABA B2 (21.4 Ϯ 7.3% of the surface expression obtained in association with GABA B2 , p Ͻ 0.001; Fig. 1A). Of note, GABA B receptor surface expression in the absence as well as in the presence of KCTD12 was not significantly influenced by exposure to 100 M baclofen for 15 min (data not shown), consistent with the reported lack of agonist-induced receptor internalization (20, 24 -26).
We additionally examined receptor surface expression in the presence or absence of KCTD12 using confocal microscopy (Fig. 1B). Surface and total GABA B1 expression was monitored by immunolabeling prior (green fluorescence) and after (red fluorescence) permeabilization of the cells. In agreement with the ELISA data, the ratio of the GABA B1 surface fluorescence to the GABA B1 total fluorescence was significantly increased by 49.1 Ϯ 9.1% (p Ͻ 0.001) in the presence of KCTD12.
KCTD12 Increases GABA B Receptor-mediated Kir3 Current Amplitudes-We addressed whether increased GABA B receptor surface expression in the presence of KCTD12 results in increased receptor-mediated Kir3 current amplitudes. We recorded outward K ϩ currents induced by fast application of the GABA B receptor agonist baclofen (100 M) to CHO-K1 cells expressing GABA B receptors and Kir3.1/3.2 effector channels. In agreement with earlier data (5), we observed a shorter rise time and a pronounced rapid desensitization of baclofeninduced K ϩ currents in the presence of KCTD12 ( Fig. 2A). Consistent with increased receptor surface expression in the presence of KCTD12, we observed a significantly increased K ϩ current density at the peak of the response (Fig. 2B). An increase in the peak K ϩ current density was not seen with receptors lacking the KCTD-binding site (GB1ϩGB2Y902A, Fig. 2B). Thus, electrophysiological experiments in heterologous cells support that binding of KCTD12 to GABA B receptors increases the peak receptor response, consistent with the observed increase in receptor cell surface expression.
KCTD12 Assembles with GABA B Receptors at the Cytoplasmic Side of the ER Membrane-We next determined where in the biosynthetic pathway KCTD12 assembles with GABA B receptors. We pulse-labeled COS-1 cells expressing GABA B2 and KCTD12 for 15 min with [ 35 S]methionine, at which time newly synthesized transmembrane proteins have not yet progressed beyond the ER (27,28). After 15 min of biosynthesis, immunoprecipitation with anti-FLAG antibodies not only purifies FLAG-KCTD12 but also efficiently co-precipitates 35 Slabeled Myc-GABA B2 (Fig. 3A). This shows that FLAG-KCTD12 binds to newly synthesized 35 S-labeled Myc-GABA B2 residing in the ER. The C termini of the GABA B1 and GABA B2 proteins are expected to face the cytoplasmic side of the ER membrane (7). Therefore KCTD12, a cytosolic protein, is assumed to assemble with GABA B2 at the cytoplasmic side of the ER membrane. We additionally studied assembly of KCTD12 with GABA B receptors in living cells using a YFP protein fragment complementation assay (15). We fused the YFP1 and YFP2 fragments to the C termini of GABA B2 (GABA B2 -YFP1) and KCTD12 (KCTD12-YFP2), respectively. Co-expression of GABA B2 -YFP1 with KCTD12-YFP2 in COS-1 cells produced YFP fluorescence due to reconstitution of a functional YFP protein (Fig. 3B). The YFP fluorescence fully overlapped with the immunostaining pattern for the ER marker protein calnexin, showing that YFP reconstitution already takes place at the level of the ER (Fig. 3B). Binding of KCTD12-YFP2 to GABA B2 -YFP1 was further confirmed in co-immunoprecipitation experiments (data not shown). In control experiments, we did not detect YFP fluorescence when expressing GABA B2 -YFP1 or KCTD12-YFP2 in the absence of each other (data not shown). This indicates that the YFP fragments do not generate fluorescence on their own. Moreover, we did not observe YFP reconstitution when expressing GABA B2 Y902A-YFP1, which does not bind to KCTD12, together with KCTD12-YFP2. Alto-  GABA B receptors were composed of either GABA B1 and GABA B2 (GB1ϩGB2) or GABA B1 and GABA B2 Y902 (GB1ϩGB2Y902A) subunits. KCTD12 increases the peak K ϩ current at GB1ϩGB2 but not at GB1ϩGB2Y902A receptors. Wholecell recordings were made at a holding potential of Ϫ50 mV. B, bar graph illustrating that KCTD12 enhances the K ϩ current density at the peak of the baclofen response in cells expressing GB1ϩGB2(GB1ϩGB2 ϩ KCTD12, 11.8 Ϯ 1.9 pA/pF; GB1ϩGB2, 7.12 Ϯ 0.94 pA/pF; *, p Ͻ 0.05) but not in cells expressing GB1ϩGB2Y902A (GB1ϩGB2Y902AϩKCTD12, 5.10 Ϯ 0.78 pA/pF; GB1ϩGB2Y902A, 7.30 Ϯ 1.31 pA/pF; p Ͼ 0.05). Each bar is the mean Ϯ S.E. of 19 -52 cells. ns, not significant.
gether, our data support that KCTD12 first assembles with GABA B receptors at the cytoplasmic side of the ER membrane.
KCTD12 Does Not Measurably Influence Forward Trafficking of GABA B Receptors-GABA B receptor trafficking to the plasma membrane is regulated by sequence elements in the C termini of GABA B1 (for example the ER retention signal RSRR (1,7,8)) and GABA B2 (residues between amino acids 841 and 862 (29)). In fully assembled receptors, these sequence elements are located in proximity to Tyr-902 in GABA B2 , the residue critically involved in KCTD binding (5). It is thus conceivable that binding of the KCTDs influences maturation and surface trafficking of GABA B receptors. To monitor posttranslational modification in the biosynthetic pathway, we analyzed the glycosylation patterns of GABA B1 and GABA B2 in the presence and absence of KCTD12 in transfected COS-1 cells. Endo H removes N-linked mannose-rich oligosaccharides from proteins that reside in the ER. Because all later oligosaccharide structures during biosynthesis are resistant to Endo H, Endo H cleavage identifies the fraction of protein that resides in the ER. In addition, we used N-glycosidase F, which removes N-linked oligosaccharides from both mature and immature proteins, to demonstrate that in our experiments, all GABA B receptors were accessible to the glycosidase treatment (Fig. 3, C and D). In the absence of GABA B2 , all GABA B1 protein was Endo H-sensitive, in agreement with a lack of terminal GABA B1 glycosylation due to ER retention (Fig. 3C). In contrast, the Myc-GABA B1 ASAA (GB1ASAA) protein, which escapes the ER because of mutation of the RSRR motif (13), was partially Endo H-resistant (Fig. 3C). Similarly, when GABA B1 was co-expressed with GABA B2 , a fraction of GABA B1 protein became Endo H-resistant, showing that heteromeric assembly with GABA B2 promotes GABA B1 exit from the ER (Fig. 3D). We observed that more GABA B2 than GABA B1 protein was Endo H-resistant (GABA B1 , 44.5 Ϯ 2.4% of total GABA B1 protein; GABA B2 , 74.6 Ϯ 1.8% of total GABA B2 protein; Fig. 3E). This may be explained by limiting amounts of GABA B2 protein, which may only allow a fraction of GABA B1 protein to exit the ER. In addition, the lack of an ER retention signal in GABA B2 allows GABA B2 to efficiently exit the ER in the absence of GABA B1 (30,31). Of importance for receptor maturation, we did not detect significant changes in the fraction of mature GABA B1 or GABA B2 protein in the presence of KCTD12 (GABA B1 , 45.9 Ϯ 3.1% of total GABA B1 protein; GABA B2 , 76.2 Ϯ 3.6% of total GABA B2 protein; p Ͼ 0.05 as compared with without KCTD12, nonparametric Mann-Whitney test; Fig. 3E). These biochemical data therefore support that maturation and forward trafficking of GABA B receptors are not significantly altered in the presence of KCTD12.
KCTD12 Constitutively Associates with GABA B Receptors-Next we investigated whether the interaction of KCTD12 with GABA B receptors at the cell surface is regulated by receptor activity. We used anti-Myc antibodies to immunoprecipitate We additionally studied KCTD12 assembly with GABA B receptors in intact cells using BRET. We generated BRET donor saturation curves by co-expressing increasing amounts of GABA B1 -YFP acceptor fusion protein with a fixed amount of Rluc-KCTD12 donor fusion protein (in the presence of a fixed amount of HA-GABA B2 , Fig. 4B). The BRET signal shows a hyperbolic increase with increasing GABA B1 -YFP expression (BRET max , 89.64 Ϯ 5.37 milliBRET units; BRET 50 , 0.026 Ϯ 0.005; Fig. 4B). This demonstrates a specific interaction of KCTD12 with GABA B receptors (32). The BRET saturation curve was not altered in the presence of baclofen, showing that the fusion proteins remain associated during receptor activation (Fig. 4B). We only observed unspecific BRET signals when expressing increasing amounts of GABA B1 -YFP with a fixed amount of Rluc-KCTD10, which does not interact with GABA B receptors (in the presence of a fixed amount of HA-GABA B2 ; Fig.  4B). Biochemical and BRET experiments thus support that KCTD12 remains constitutively associated with the receptor during receptor activity.
Constitutive GABA B Receptor Internalization Is Slowed in the Presence of KCTD12-Because we did not observe significant changes in GABA B receptor maturation and forward trafficking in the presence of KCTD12 (see above), we addressed whether KCTD12 possibly increases receptor surface levels by reducing constitutive endocytosis (20,26,33,34). A cell surface biotinylation assay (20) indeed reveals a significantly reduced GABA B receptor endocytosis in the presence of KCTD12 (Fig. 5A). Quantitative analysis showed that the percentage of total GABA B1 protein internalized within 120 min was significantly smaller in the presence of KCTD12 (KCTD12, 16 Ϯ 5%; without KCTD12, 33 Ϯ 6%; p Ͻ 0.05, nonparametric Mann-Whitney test; Fig. 5B). The rate of constitutive GABA B receptor internalization in the absence of KCTD12 was similar to the rate reported in an earlier study (20). We conclude that KCTD12 increases cell surface GABA B receptor expression by reducing constitutive receptor internalization. Co-immunoprecipitation experiments further reveal that GABA B1 , GABA B2 , and KCTD12 endocytose in associate with each other as a protein complex (Fig. 5A).
Acute and Constitutive Loss of KCTD12 in Cultured Hippocampal Neurons Reduces GABA B Receptor Surface Levels-Hippocampal neurons of mice express KCTD12 and KCTD16 protein (5,6). We used a cell surface biotinylation assay to address whether down-regulation of endogenous KCTD12 protein in cultured hippocampal neurons influences GABA B receptor expression at the plasma membrane. Neurons were infected at DIV 6 -7 with lentiviruses encoding either KCTD12 shRNA or control shRNA and analyzed at DIV 13. KCTD12 shRNA significantly reduced the expression of endogenous KCTD12 protein as compared with control shRNA (55.8 Ϯ 8.1% of control, KCTD12 protein was normalized to ␤-tubulin protein, p Ͻ 0.05; Fig. 6A). KCTD12 knockdown with shRNA resulted in a significant decrease of surface GABA B1 protein as compared with control shRNA (GABA B1a , 58.4 Ϯ 9.7% of control, p Ͻ 0.01; GABA B1b , 66.4 Ϯ 10.9%, p Ͻ 0.05; Fig. 6, A and B). In agreement with the results obtained from the cell surface biotinylation experiments, we found that knockdown of KCTD12 in hippocampal neurons significantly decreases the baclofen-induced K ϩ current density at the peak of the response (KCTD12 shRNA, 1.32 Ϯ 0.20 pA/pF; control shRNA, 2.27 Ϯ 0.21 pA/pF; p Ͻ 0.01; Fig. 6, C and D). Finally, we examined cell surface expression of GABA B receptors in cultured hippocampal neurons of Kctd12 Ϫ/Ϫ mice (6). The cell surface biotinylation assay demonstrated that Kctd12 Ϫ/Ϫ neurons express reduced amounts of GABA B1a and GABA B1b proteins at the plasma membrane in comparison with wild-type neurons (GABA B1a , 71.5 Ϯ 9.6% of WT; GABA B1b , 77.9 Ϯ 9.4% of WT, p Ͻ 0.05; Fig.  7, A and B). We additionally tested GABA B responses in Kctd12 Ϫ/Ϫ and wild-type neurons. The GABA B receptor-mediated K ϩ current density at the peak of the response was slightly but significantly decreased in Kctd12 Ϫ/Ϫ as compared with wild-type neurons (Kctd12 Ϫ/Ϫ , 1.70 Ϯ 0.07 pA/pF; WT, 1.94 Ϯ 0.07 pA/pF; p Ͻ 0.05; Fig. 7, C and D). Based on these electrophysiological results, we conclude that the presence of KCTD12 increases the maximal GABA B receptor response in cultured hippocampal neurons.

DISCUSSION
Over the past decade, the concept of principal and auxiliary subunits has been extended from voltage-gated ion channels to ligand-gated ion channels (11,12) and GPCRs (9). Signature features of auxiliary subunits are their direct and stable interaction with principal subunits and effects on channel or receptor kinetics and/or pharmacology. Additionally, auxiliary subunits often promote surface expression of channels or receptors. KCTD8, -12, -12b, and -16 subunits are examples of GPCR-associated proteins that fulfill some of the criteria of auxiliary receptor subunits (5). It is already well established that the KCTDs directly interact with the principal receptor subunit GABA B2 and that they affect kinetic and pharmacological properties of the receptor response. Our experiments establish that KCTD12 stably interacts with GABA B receptors throughout the lifecycle of the receptor, as expected for an auxiliary receptor subunit. In the biosynthetic pathway, assembly of KCTD12 with the receptor takes place as early as in the ER compartment. This association is maintained during receptor trafficking to the plasma membrane, agonist activation, and internalization. We found that assembly with KCTD12 stabilizes GABA B receptors at the cell surface. This is also supported by an earlier study showing that overexpression of KCTD12 in cultured hippocampal neurons augments axonal surface targeting of GABA B receptors (4). Our experiments indicate that the increase in surface receptors is the result of reduced constitutive receptor internalization. In this context, it is interesting to note that the C-terminal domain of GABA B2 , which comprises the KCTD12-binding site, regulates the rate of constitutive internalization, presumably through the shielding of a dileucine signal in the C terminus of GABA B1 (35). It is conceivable that KCTD12 modulates the GABA B1 /GABA B2 interface in a manner that reduces constitutive internalization. It is interesting that KCTD12 shortens the rise time and increases the magnitude of the receptor response while at the same time promoting desensitization. Assembly with KCTD12 may therefore provide the means to increase the temporal precision of GABA B receptor signaling. An increase in cell surface stability of the receptors is also mediated by the sushi domains in the extracellular domain of GABA B1a (36). This demonstrates that both intracellular and extracellular mechanisms have evolved to stabilize GABA B receptors at the cell surface. It is intriguing that only KCTD12 influences GABA B receptor surface expression, whereas the KCTD8, -12b, and -16 subunits do not. The reason for this is unknown. Because the amino acid sequence identity between the KCTD8, -12, -12b, and -16 proteins is only between 33 and 54%, it may be argued that consti- A, GABA B receptor internalization in the presence or absence of KCTD12 was investigated using a biotinylation assay. Cell were biotinylated on ice for 15 min, washed, and then either incubated on ice for 120 min to prevent internalization of cell surface proteins (Total surface) or incubated at 37°C for 120 min to allow internalization (Internalized). Biotinylated total surface and internalized proteins were purified with NeutrAvidin-Sepharose. GABA B1 (GB1) and KCTD12 protein was revealed on Western blots. For purification of biotinylated internalized protein, the biotin label at the cell surface was first removed by incubation with reduced glutathione (GSH). Biotinylated total surface protein cleaved by GSH (GSH cleavage) demonstrates that GSH efficiently removes the biotin label from surface proteins. Internalized GB1 and KCTD12 proteins (Internalized) were always analyzed in duplicates. B, quantification of GABA B receptor internalization in the absence and presence of KCTD12. The percentage of the total GB1 protein that was internalized was calculated after subtraction of uncleaved GB1 protein after GSH cleavage at the cell surface. Bars are the means Ϯ S.E. of 3 separate experiments. *, p Ͻ 0.05. FIGURE 6. Knockdown of KCTD12 reduces GABA B receptor surface expression in hippocampal neurons. A, cell surface biotinylation of mouse hippocampal neurons infected with KCTD12 shRNA or control shRNA. GABA B1 (GB1a and GB1b) and KCTD12 proteins in neuronal lysates (total) and NeutrAvidin-purified cell surface proteins (surf) were revealed on Western blots using specific antibodies. Tubulin (tub) was visualized to control for loading of the SDS-PAGE. B, graph illustrating the decrease of surface GB1a and GB1b protein in neurons after KCTD12 knockdown. Data are means Ϯ S.E. from 5-8 experiments. *, p Ͻ 0.05, **, p Ͻ 0.01. C, representative GABA B receptor-mediated K ϩ current responses recorded at Ϫ50 mV from cultured hippocampal neurons infected with control (gray trace) or KCTD12 shRNA (black trace). D, the bar graph shows that the K ϩ current density at the peak of the response is significantly reduced in neurons infected with KCTD12 shRNA as compared with neurons infected with control shRNA. Data are the means Ϯ S.E., n ϭ 10 -17, **, p Ͻ 0.01. tutive internalization at the GABA B1 /GABA B2 interface is influenced by sequence elements that are unique to KCTD12.
We found that knock-out or knockdown of KCTD12 in cultured hippocampal neurons, which express high amounts of KCTD12 protein (6), decreases surface GABA B receptor expression and the peak amplitude of the baclofen-induced K ϩ current. Acute knockdown of KCTD12 had a more pronounced effect in reducing surface receptors than the constitutive loss of KCTD12 in knock-out mice. Acute and dynamic changes in KCTD12 expression may influence surface GABA B receptor expression more strongly because of the lack of compensatory changes. Regulation of GABA B receptors via KCTD12 may be of relevance to disease. A genome-wide association study found that a polymorphism in the promoter region of the KCTD12 gene is associated with bipolar I disorder (37). Likewise, KCTD12 has been associated with depressive disorders (38) and schizophrenia (39). We speculate that KCTD12 is necessary for the precise timing of GABA B receptor-mediated inhibitory effects on network activity (40), perturbation of which would manifest as psychiatric disorders.