Structural basis for auxiliary subunit KCTD16 regulation of the GABA_B receptor

Significance

Human GABA_B receptor inhibits the activity of neurons in the brain. Malfunction of the receptor is associated with various neurological and mood disorders, including spasticity, epilepsy, addiction, and anxiety. GABA_B receptor consists of principal and auxiliary subunits. The principal subunits form the receptor core that responds to neurotransmitters, while the auxiliary subunits modify the receptor output upon stimulation. The crystal structure of an auxiliary subunit bound to a peptide fragment of the principal subunit reveals their interaction and sheds light on the assembly of GABA_B receptor signaling complex. Furthermore, the binding interface of GABA_B receptor and KCTD presents a potential site for the design of compounds for regulating receptor function and treating GABA_B-associated diseases.

Abstract

Metabotropic GABA_B receptors mediate a significant fraction of inhibitory neurotransmission in the brain. Native GABA_B receptor complexes contain the principal subunits GABA_B1 and GABA_B2, which form an obligate heterodimer, and auxiliary subunits, known as potassium channel tetramerization domain-containing proteins (KCTDs). KCTDs interact with GABA_B receptors and modify the kinetics of GABA_B receptor signaling. Little is known about the molecular mechanism governing the direct association and functional coupling of GABA_B receptors with these auxiliary proteins. Here, we describe the high-resolution structure of the KCTD16 oligomerization domain in complex with part of the GABA_B2 receptor. A single GABA_B2 C-terminal peptide is bound to the interior of an open pentamer formed by the oligomerization domain of five KCTD16 subunits. Mutation of specific amino acids identified in the structure of the GABA_B2–KCTD16 interface disrupted both the biochemical association and functional modulation of GABA_B receptors and G protein-activated inwardly rectifying K⁺ channel (GIRK) channels. These interfacial residues are conserved among KCTDs, suggesting a common mode of KCTD interaction with GABA_B receptors. Defining the binding interface of GABA_B receptor and KCTD reveals a potential regulatory site for modulating GABA_B-receptor function in the brain.

Metabotropic γ-aminobutyric acid (GABA) type B (GABA_B) receptors are implicated in various neurological and psychiatric disorders, including spasticity, epilepsy, depression, addiction, and anxiety (1–3). GABA_B receptors provide a crucial component of inhibitory neurotransmission in the nervous system (1–3), via coupling to G_i/o type G proteins that modulate three different downstream effectors: voltage-gated Ca²⁺ channels, G protein-activated inwardly rectifying K⁺ (GIRK) channels, and adenylyl cyclase (1–3). The GABA_B receptor functions as an obligatory heterodimer, consisting of the GABA_B1 and GABA_B2 subunits (4–9), whereby the GABA_B1 subunit binds orthosteric ligand (10, 11) and the GABA_B2 subunit couples to the G protein (12–18).

Recent proteomic studies led to the discovery of a family of auxiliary proteins for the GABA_B receptor, originally referred to as potassium channel tetramerization domain-containing (KCTD) proteins (19–21). A subset of KCTD proteins (numbered 8, 12, 12b, and 16) interact with the cytoplasmic tail of the GABA_B2 subunit (3, 20, 21) and convey unique functional properties to the signaling of the GABA_B receptor (3). For example, expression of each of the four KCTD proteins leads to acceleration of GABA_B receptor-dependent activation of GIRK currents, albeit to different extents (20, 22). KCTD12 and KCTD12b also promote rapid desensitization of the GABA_B receptor-induced GIRK current, by uncoupling Gβγ from GIRK channels (20, 22, 23). By contrast, KCTD8 and KCTD16 generate primarily nondesensitizing receptor responses (20, 22, 23). In genetic studies, the KCTD12 gene has been associated with major depressive disorder, bipolar disease, and schizophrenia (24, 25). Mice lacking KCTD12 or KCTD16 exhibit altered fear memory relative to wild-type mice (22, 26, 27). Thus, molecular interaction of these KCTD proteins with GABA_B receptors is important for the normal function of inhibitory brain circuits. However, the mechanism by which GABA_B receptors interact with these auxiliary proteins is poorly understood.

KCTDs are named as such primarily due to the sequence homology between their conserved N-terminal tetramerization domains (T1) and the tetramerization domains of voltage-gated potassium (Kv) channels (28). In addition to the conserved T1 domain, most KCTD proteins contain a C-terminal homology domain (H1) that exhibits greater sequence variability from protein to protein (20, 23, 29). The C-terminal domain of KCTDs can account for some of their differences in regulating GABA_B receptor function. The H1 domains of KCTD12 and KCTD12b mediate desensitization of GABA_B receptor through a sequence motif that is not found in KCTD8 and KCTD16 (23). On the other hand, KCTD8 and KCTD16 are unique in having two C-terminal homology domains, H1 and H2, and the H2 domain inhibits receptor desensitization (20, 23). Given the diversity of KCTD proteins, it is not clear how a subset of KCTDs associate specifically with GABA_B receptors.

Previously, Schwenk et al. (20) showed that the T1 domain of the KCTD proteins interacts with the C terminus of the GABA_B2 subunit. Mutagenesis studies indicated that tyrosine-902 (Y902), located in the rat GABA_B2 subunit, is important for KCTD12 binding to the receptor (20, 29). The site of interaction on the KCTD protein, however, is unknown. Interestingly, structural studies have indicated that the KCTD T1 domain adopts a diverse array of oligomeric architectures, including monomers, tetramers, and both open and closed pentamers (30–32). Thus far, there is no high-resolution structure of the KCTD protein complexed with the GABA_B receptor. Here, we present the structure of the KCTD16 oligomerization T1 domain bound to a GABA_B2-derived peptide, and discovered a binding stoichiometry of one KCTD16 pentamer for every GABA_B2. Mutational analysis of the interfacial residues further identified the molecular determinants of GABA_B–KCTD interaction and their impact on the kinetic properties of GABA_B receptor signaling.

Results

Structure of KCTD16 Oligomerization Domain.

To describe the interaction of KCTD16 with the GABA_B receptor, we first solved the structure of the T1 oligomerization domain of human KCTD16. Purified KCTD16 T1 was crystallized in two different forms (SI Appendix, Table S1). Form I crystals have the P1 space group and diffracted to a resolution of 2.25 Å. The structure was solved using an l-selenomethionine (SeMet)–substituted mutant by single isomorphous replacement with anomalous scattering (SIRAS), and refined to an R-factor of 25.4% (R_free = 27.0%) (SI Appendix, Table S1). Form II crystals belong to the P3₂ space group and diffracted to 2.80 Å. The structure was determined by molecular replacement using the form I KCTD16 T1 structure as the search model, and refined to a final R-value of 22.7% (R_free = 23.8%) (SI Appendix, Table S1).

We found that KCTD16 T1 forms an open pentamer in both crystal forms (Fig. 1A), similar to previous findings (32). The KCTD16 T1 structure adopts the expected broad-complex, tramtrack, and bric à brac fold consisting of a three-stranded β-sheet and five α-helices (Fig. 1B), similar to that found in the T1 domain of Kv channels (28, 33, 34). Five KCTD16 T1 subunits assemble in a circular fashion into an open ring that has an inner diameter of ∼25 Å. Adjacent KCTD16 T1 subunits are arranged side by side, with approximately the same N- to C-terminal orientation. Furthermore, there are multiple copies of the pentamer in each crystal form (SI Appendix, Fig. S1 A and B). The subunits within each pentamer are highly homologous, with rmsd values ranging from 0.27 to 0.49 Å (SI Appendix, Fig. S1D). All of the pentameric assemblies are highly similar in the overall curvature and arrangement of subunits (SI Appendix, Fig. S1C), suggesting that the open pentameric stoichiometry is physiologically relevant. Interestingly, the gap between I and V is ∼16 Å at its widest and 8 Å at its narrowest. By contrast, a KCTD16 T1 monomer in the plane of the pentamer ranges from 10 to 25 Å, which is too large to fit in this gap. Thus, the KCTD16 T1 structure is not compatible with the binding of a sixth subunit despite having a gap.

Fig. 1.

Crystal structure of human KCTD16 T1 at 2.3 Å. (A) Surface and ribbon representations of the KCTD16 T1 open pentamer structure in two orthogonal views. Note that the gap between I and V is ∼16 Å at its widest point. (B) Ribbon representation of KCTD16 T1 monomer structure. The color spectrum represents the order of residues in the protein sequence, and ranges from blue at the N terminus to red at the C terminus. The α-helices and β-strands within the KCTD16 T1 structure are labeled.

Structure of a GABA_B2-Derived Peptide Bound to KCTD16.

A previous mutagenesis study implicated Y902 in the cytoplasmic domain of the rat GABA_B2 receptor (Y903 in humans) in the interaction with KCTD proteins (20). Furthermore, a human GABA_B2 receptor peptide encompassing this region, ⁸⁹⁰RRLSLQLPILHHAYLPSI⁹⁰⁷, binds to KCTD12 with nanomolar affinity in isothermal titration calorimetry experiments (29). We designed a slightly shorter GABA_B2 receptor peptide for crystallization containing the amino acids ⁸⁹⁵QLPILHHAYLPSIGG⁹⁰⁹ [GABA_B2(895–909)]. To confirm the direct binding between KCTD16 T1 and GABA_B2(895–909), we examined the mobility on a native gel. As expected, the addition of Flag-tagged GABA_B2(895–909) resulted in a shift in the mobility of KCTD16 T1 (SI Appendix, Fig. S2), consistent with the peptide binding directly to KCTD16.

Next, we attempted to crystallize KCTD16 T1 in the presence of excess GABA_B2(895–909), lacking the Flag tag. GABA_B2(895–909)/KCTD16 T1 complex crystals formed in the P1 space group, and diffracted to 2.35 Å (SI Appendix, Table S1). The complex structure was solved by molecular replacement using a KCTD16 T1 pentamer as the search model, and refined to a final R_free of 24.1% (R_work = 22.2%) (SI Appendix, Table S1).

The structure of the KCTD16 T1/GABA_B2(895–909) complex reveals a 1:5 binding stoichiometry between GABA_B2(895–909) and KCTD16 T1 (Fig. 2A). Each KCTD16 T1 pentamer accommodates one GABA_B2(895–909) peptide using the inner surface of its open ring structure. Interestingly, the GABA_B2(895–909) peptide is not linear, but rather forms a loop inside the ring with the N and C termini pointing toward the N termini of the KCTD16 T1 subunits. A 3₁₀ helix containing four residues (H900–Y903) is found at the apex of the loop (Fig. 2B), illustrating how Y903 interacts with KCTD16 (20). The binding of GABA_B2(895–909) did not appear to induce any systematic conformational changes in the structure of individual KCTD16 T1 subunits, as the average rmsd was low (0.3 Å) and the overall arrangement of the pentamer did not change. These results suggest the pentameric assembly of KCTD16 T1 is relatively rigid even in the presence of the GABA_B receptor (Fig. 3A and SI Appendix, Fig. S1C). Consistent with this, the interface between neighboring KCTD16 T1 subunits (I-II, II-III, III-IV, IV-V) is largely conserved in the multiple copies of pentamers in the crystal, whether in the absence or presence of the GABA_B2(895–909) (Fig. 2C).

Fig. 2.

Crystal structure of a GABA_B2(895–909)/KCTD16 T1 complex at 2.4 Å. (A) Ribbon representation of the GABA_B2(895–909)/KCTD16 T1 complex structure. The GABA_B2(895–909) peptide is displayed as a space-filling model. (B) Surface representation of the GABA_B2(895–909)/KCTD16 T1 complex in two orthogonal views. The GABA_B2(895–909) is shown as a ribbon model in green. (C) C_α trace superposition of KCTD16 T1 pentamers in the apo form (cyan) and in complex with GABA_B2(895–909) (orange). The alignment was performed based on the subunit I structure in each pentameric assembly.

Fig. 3.

Amino acid interactions at the subunit interface in the KCTD16 T1 pentamer. (A) C_α trace superposition of all copies of KCTD16 T1 monomer within one GABA_B2(895–909)/KCTD16 T1 complex structure. Subunits II, III, IV, and V were superimposed onto subunit I. The superimposed structures are colored, and the pentamer arrangement is shown in gray. (B) Close-up view of two interacting KCTD16 T1 subunits [KCTD16 T1 (III) and KCTD16 T1 (II)] in a GABA_B2(895–909)-bound KCTD16 T1 pentamer in the direction of the red arrow as shown in A. The structural elements (β2, α3, β3′, α4′) involved in pentamer formation are labeled within the box (dashed lines). (C) Specific amino acid contacts at the pentamer interface shown in B. Note extensive salt bridge interactions. Two views of the interfacial contacts related by a 110° rotation about the vertical axis are shown. Dashed lines indicate hydrogen bonds. Size exclusion chromatography (Superdex 200) profiles (D) and SDS/PAGE gel of purified wild-type (WT) and D76R mutant KCTD16 T1 (E) are shown. (F–H) Comparison of GIRK channel activation kinetics for HA-KCTD16 WT and D76R. (F) Baclofen-induced (100 μM) GIRK currents in HEK293T cells transfected with GABA_B1b, GABA_B2, and GIRK2c alone (Control) and with full-length HA-KCTD16 WT or D76R cDNAs. (G) Bar graph shows average activation time (20–80%) for baclofen-induced GIRK currents (n = 7–9). *P < 0.05, one-way ANOVA with Bonferroni’s post hoc test vs. WT. (H) Representative immunostaining for HA-KCTD16 and HA-KCTD16-D76R using anti-HA antibodies (red) and costained with a 488 phalloidin probe (green).

KCTD16 Pentameric Interface.

We next examined the structural mechanism underlying the formation of the pentamer of the KCTD16 T1 domain. We examined a superimposition of the five subunits within a KCTD16 T1 pentamer and found all to be highly homologous, with rmsd values ranging from 0.27 to 0.49 Å (Fig. 3A). The interaction between adjacent KCTD16 T1 subunits buries ∼1,600 Ų of solvent-accessible surface area. The interface also has exceptionally high shape complementarity, with an average shape-correlation statistic of 0.77, which is greater than most oligomeric interfaces and even protein–protein inhibitor interfaces (35).

The specific contacts at the pentameric interface feature predominantly salt bridges (Fig. 3 A–C). These charged interactions form the bridge between two relatively flat surfaces, one from each of the adjacent KCTD16 T1 subunits. At the center of one surface lies the α3 helix of one subunit, where it packs against two structural elements at the adjacent surface, composed of the loop between the β3 strand and α3 helix, as well as the α4 helix and its preceding loop. The α3 helix is responsible for mediating all of the salt bridges at the pentameric interface. Four intercalating acidic and basic residues from one side of the α3 helix form multiple contacts with six charged residues from the neighboring subunit. These specific interactions are distributed across the entire interfacial area, and include R83–D87′, R83–E102′, D87–R105′, R90–D76′, D91–R77′, and D91–R108′. All of these charged residues are conserved among the different KCTDs that associate with GABA_B receptors, suggesting that the pentameric assembly may be a common oligomeric arrangement of GABA_B-bound KCTDs (SI Appendix, Fig. S3).

In addition to electrostatic interactions, nonpolar associations are present at one corner of the interface (Fig. 3C). A hydrophobic patch is formed by residues V26, F37, and T38 from the β1 and β2 strands of one KCTD16 T1 molecule and by residue F74′ on the α3 helix of the adjacent subunit. Of these hydrophobic residues, F74′ is conserved among the GABA_B-related KCTDs (SI Appendix, Fig. S3).

Part of the tight association of KCTD16 T1 subunits in the pentamer may arise from the electrostatic interactions at the interface. Indeed, wild-type KCTD16 T1 migrates as an oligomer upon size exclusion chromatography (Fig. 3 D and E). To test this idea, we carried out charge reversal mutations of single amino acids involved in salt bridge formation at the interface, including D76R, R77D, D78R, and R105D. All of the charge reversal mutations reduced the expression level of KCTD16 T1 in bacteria and caused severe aggregation of the mutant proteins. However, the D76R mutant was purified to homogeneity with buffer conditions optimized for protein stabilization. Size exclusion chromatography demonstrates that the KCTD16 T1 D76R mutant exists as a monomer in solution, which has an expected molecular mass of 13.4 kDa (Fig. 3 D and E). We next examined the functional impact of the D76R mutation in KCTD16 by studying the effect of KCTDs on the coupling of GABA_B receptors with GIRK channels (20–22). In HEK293T cells expressing the GABA_B1b and GABA_B2 receptors along with GIRK2c, baclofen induces an inward K current (Fig. 3 F–H). In cells coexpressing HA-KCTD16, the rate of baclofen-induced GIRK current activation is increased almost sixfold, consistent with earlier findings (20–22) (Fig. 3 F and G). Although HA-KCTD16-D76R is expressed in HEK293T cells, as indicated by immunohistochemistry using anti-HA antibodies (Fig. 3H), it did not alter the rate of baclofen-dependent GIRK activation (Fig. 3 F and G). Taken together, these results suggest the charged interactions observed at the pentameric interface of the KCTD16 T1 structure mediate its oligomerization in solution, and support formation of the pentameric structure.

GABA_B2–KCTD16 Interface.

Having determined the structure of the GABA_B2(895–909)/KCTD16 T1 complex, we could now examine the interface between the GABA_B receptor peptide and KCTD16 T1 domain. The GABA_B2-derived peptide binds to a concave surface formed by the first four adjacent subunits (I–IV) of the KCTD16 T1 pentamer (Fig. 4 A and B). The GABA_B2-binding site is located asymmetrically, off-center, and away from the opening of the pentamer between I and V. Three independent GABA_B2(895–909)/KCTD16 T1 complexes were located in the crystal asymmetrical unit (SI Appendix, Fig. S4A). Pairwise alignment of these structures yielded rmsd values ranging from 0.48 to 0.63 Å for the entire complex, indicating that structures are substantially homologous. The full sequence of GABA_B2(895–909) is visible in only one of the complexes, while one N-terminal residue and two C-terminal residues were disordered in the other two complexes (SI Appendix, Fig. S4B). Although GABA_B2(895–909) adopts a mostly extended configuration, its structure is highly similar in the three complexes, especially in the helical region, suggesting that the peptide conformation is dictated by its interaction with KCTD16 (SI Appendix, Fig. S4B).

Fig. 4.

Identification of key amino acids at the GABA_B2(895–909)/KCTD16 T1 interface. (A) Ribbon representation showing GABA_B2(895–909) (green) interacting with four subunits of a KCTD16 T1 pentamer (cyan, pink, gray, and blue). (B) Surface representation of the GABA_B2(895–909)-binding site in KCTD16 T1, highlighting the hydrophobic grooves formed by F80 of KCTD16 T1. (C) Specific contacts at the GABA_B2(895–909)/KCTD16 T1 interface within one of the complexes in the crystal. Dashed lines indicate hydrogen bonds. KCTD16 residues that contribute extensive contacts at the interface are highlighted in circles; these include F80 from all four subunits (red ovals) and E102 from one subunit (black oval). (D) Schematic shows key contacts at the interface between GABA_B2(895–909) peptide and KCTD16 T1. The interface area is divided into three regions: 1, 2, and 3. (E–H) Coimmunoprecipitation of full-length Flag-GABA_B1b (B1), HA-GABA_B2 (B2), and HA-KCTD16 (KCTD16) from transfected HEK293T cells by anti-Flag M2 antibody. Western blots (anti-HA antibody) show the effect of single mutations in HA-KCTD16 (E and F) or GABA_B2 (G and H) on the GABA_B–KCTD16 association. Bar graphs show the mean ratio of KCTD16 to GABA_B2 coimmunoprecipitation for four experiments. *P < 0.05, one-way ANOVA with Bonferroni’s post hoc test vs. wild type (WT). (I) Baclofen-induced (100 μM) GIRK current in HEK293T cells transfected with GABA_B1b, GABA_B2, and GIRK2c alone, and with HA-KCTD16 WT or mutant cDNAs. Bar graphs show average activation time (20–80%) for baclofen-induced GIRK currents for KCTD16 mutants (J, n = 6–9) and GABA_B-receptor mutants (K, n = 7–10). *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA with Bonferroni’s multiple comparison post hoc test vs. WT.

Nearly all of the GABA_B2(895–909) residues form extensive interactions with KCTD16 T1 (Fig. 4 A–C and SI Appendix, Fig. S4 C–E). Examining the molecular interactions at the interface revealed three regions of interaction (Fig. 4 C and D and SI Appendix, Fig. S4 C–E). In the first region, the N-terminal region of GABA_B2(895–909), before the helical turn, is anchored to the outermost subunit (I) of the KCTD16 T1 pentamer through a hydrogen bond between Q34^I and either Q895 or L896 of GABA_B2(895–909). The remaining interfacial contacts are mostly hydrophobic. GABA_B2 residues L896 and L899 fit into a hydrophobic groove lined by F80^I and F80^II from the first two KCTD16 T1 subunits.

In the second region, the 3₁₀ helix at the turning point of GABA_B2(895–909) contacts two KCTD16 T1 subunits located in the center of the interface (II and III) (Fig. 4 C and D and SI Appendix, Fig. S4 C–E). Two highly conserved interfacial hydrogen bonds are found in this region, formed by Q34^II of the second KCTD16 subunit and H901 of the GABA_B2 receptor, as well as E102^III of the third KCTD16 molecule and Y903 of GABA_B2. In addition, the two His residues H900 and H901 of GABA_B2 are sandwiched between F80^II and F80^III of KCTD16 to form an aromatic ring-stacking interaction.

The third region involves the C-terminal stretch of GABA_B2(895–909) (Fig. 4 C and D and SI Appendix, Fig. S4 C–E). The C terminus of the peptide is secured to the KCTD16 T1 surface through both direct and water-mediated hydrogen bonds. Depending on the particular complex, several GABA_B2 residues participate in hydrogen bond formation, most notably, I907 through its main-chain atoms. The KCTD16 T1 residues involved include Q34^III and its neighboring residues from the third KCTD16 T1 subunit. Additionally, hydrophobic contacts are featured prominently at this site. Residues F80^III and F80^IV from the third and fourth KCTD16 T1 subunits create a channel for the extended peptide and interact with the GABA_B2 residues L904, P905, S906, and I907.

Taken together, the GABA_B2-binding site on KCTD16 is formed by several key residues, including, most importantly, F80 from each of the first four subunits. Most prominently, the side chains of the four F80 residues protrude into the inner space of the KCTD16 ring, appearing like teeth on a gear comb, where different sections of the GABA_B2 peptide are inserted (Fig. 4B). F80 is conserved among all of the GABA_B-associated KCTD proteins, suggesting that the binding mode observed in the GABA_B2(895–909)/KCTD16 T1 structure is universal among all GABA_B-related KCTD proteins (SI Appendix, Fig. S3).

Functional Role for GABA_B2–KCTD16 Interface.

Using the high-resolution structure of the GABA_B2-KCTD16 interface, we examined the physiological importance of this interaction with full-length proteins. We first investigated the biochemical interaction of these proteins. Association of KCTD proteins with GABA_B receptors can be demonstrated by coimmunoprecipitation (20). In HEK-293 cells expressing tagged versions of full-length GABA_B1b, GABA_B2, and HA-KCTD16 proteins, anti-Flag antibodies coprecipitated Flag-GABA_B1b along with HA-GABA_B2 and HA-KCTD16, thereby confirming the coassembly of GABA_B heterodimer receptor with KCTD16 (20, 21) (Fig. 4 E and F and SI Appendix, Fig. S5 A and B). We next engineered a single mutation at the GABA_B2–KCTD16 interface, focusing on those sites buried deeply at the interface and participating in extensive interfacial contacts (Fig. 4 B–D). These interfacial residues include L896, I898, H901, Y903, and L904 of GABA_B2 and Q34, F80, P101, and E102 of KCTD16, and can be subdivided into three regions (Fig. 4D). We found that Q34A, F80A, P101S, or E102A point mutation in KCTD16 significantly reduced coimmunoprecipitation of GABA_B2 with HA-KCTD16 (Fig. 4 E and F and SI Appendix, Fig. S5 A and B). Similarly, I898S, Y903S, and L904D in the GABA_B2 also significantly reduced coimmunoprecipitation of GABA_B2 with HA-KCTD16 (Fig. 4 G and H and SI Appendix, Fig. S5 A and B).

To examine the functional relevance of this binding interaction site, we examined the effect of KCTDs on the coupling of GABA_B receptors with GIRK channels (20–22). Mutation of two key residues in KCTD16 at the GABA_B2–KCTD16 interface, F80A or E102A, significantly impaired the ability of HA-KCTD16 to accelerate GIRK channel activation (Fig. 4 I and J). We also investigated the functional effect of mutations in GABA_B2 that are predicted to interfere with association via region 1 of HA-KCTD16 (I898S) or region 3 of HA-KCTD16 (L904D). Like the previously published Y903A (region 2) (20), these mutations also disrupted the regulation of GABA_B receptors by HA-KCTD16; there was no acceleration of receptor activation (Fig. 4K). Taken together, these electrophysiological results, combined with the coimmunoprecipitation experiments, corroborate the structural identification of amino acids at the interface of the pentameric KCTD16 and the GABA_B2 peptide that are critical for mediating the association of KCTD16 and subsequent modulation of GABA_B receptors.

Conservation of GABA_B2–KCTD Interface.

Expression of KCTD12 produces rapid desensitization of GABA_B receptor-activated GIRK currents, demonstrated previously to be due to the H2 domain, which uncouples Gβγ from GIRK channels (20, 22, 23). KCTD12 has also been reported to bind to a GABA_B2 receptor peptide (29). We therefore sought to determine whether the interface between the GABA_B2 peptide and KCTD16 T1 is conserved in KCTD12. To address this, we first created a homology model of KCTD12 T1 in complex with GABA_B2(895–909) peptide, based on the structure of the GABA_B2(895–909)/KCTD16 T1 complex (Fig. 5 A and B). Since the T1-T1 interfacial residues of KCTD16 are highly conserved among GABA_B-associated KCTD proteins (SI Appendix, Fig. S3), we reasoned that a KCTD12 model with the same T1-T1 binding mode provides a likely prediction of the KCTD12 T1 structure. To test the model, we examined the effect of homologous mutations in KCTD12, Q43A, F87A, P108S, and E019A on the biochemical association of KCTD12 with the GABA_B receptor. Coimmunoprecipitation experiments confirmed that GABA_B1b coprecipitates GABA_B2 and HA-KCTD12 (20, 21) (Fig. 5 C and D and SI Appendix, Fig. S5 C and D). Each point mutation, Q43A, F87A, P108S, and E109A, significantly reduced the coprecipitation of HA-KCTD12 with the GABA_B receptor (Fig. 5 C and E). Similarly, mutations in the GABA_B2 receptor also significantly reduced coprecipitation of HA-KCTD12 with the GABA_B receptor (Fig. 5 E and F).

Fig. 5.

Mutational analysis of the putative GABA_B2(895–909)/KCTD12 T1 interface. (A) Homology model of GABA_B2(895–909) (yellow) interacting with KCTD12 T1 (four subunits colored cyan, pink, gray, and blue, respectively) created based on the GABA_B2(895–909)/KCTD16 T1 complex structure. KCTD12 residues that are predicted to form extensive interfacial contacts are indicated by black spheres, and include F87 from all four subunits (red ovals in B) and E109 from one subunit (black oval in B). (B) Residues at the GABA_B2(895–909)/KCTD12 T1 interface in the homology model. Residues highlighted in circles are predicted to form substantial contacts at the interface, as in A. (C–F) Coimmunoprecipitation of Flag-GABA_B1b (B1), HA-GABA_B2 (B2), and HA-KCTD12 (KCTD12) from transfected HEK293 cells by anti-Flag M2 antibody. The conditions are the same as in Fig. 4. Bar graphs show the mean ratio of KCTD12 to GABA_B2 for mutant KCTD12 (n = 3) or GABA_B2 receptor (n = 5). *P < 0.05, one-way ANOVA with Bonferroni’s post hoc test vs. wild type (WT). (G–I) Whole-cell patch-clamp recordings show the effect of mutations at the putative GABA_B2-binding interface of HA-KCTD12 on baclofen-dependent desensitization of GIRK current. (G) Inward GIRK current plotted as a function of time (at −120 mV) for HEK293T cells transfected with GABA_B1b, GABA_B2, GIRK2c, and HA-KCTD12 WT or mutant cDNAs (gray bar; 100 μM baclofen). (H and I) Bar graphs show mean percentage of desensitization during baclofen-dependent activation (n = 10) and mean decay time (Tau) for desensitization (n = 10). *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA with Bonferroni’s multiple comparison post hoc test (black stars vs. WT, red stars vs. no KCTD).

We next examined the effect of mutations on the functional modulation of the GABA_B receptor and GIRK channels. As shown previously (20, 22, 23), expression of HA-KCTD12 dramatically increases both the rate and extent of desensitization for GABA_B receptor-mediated GIRK currents (Fig. 5 G–I). Two mutations in KCTD12 that reduced the association of KCTD12 with the GABA_B receptors, F87A and E109A, also significantly reduced both the rate and extent of agonist-induced desensitization (Fig. 5 G–I). Interestingly, the changes in desensitization with the F87A and E109A mutations were significantly different from wild-type HA-KCTD12 but were also different from cells lacking HA-KCTD12. Thus, these mutations may produce a partial disruption of the interaction with the receptor or a secondary effect of unbound KCTD12 with Gβγ subunits (Fig. 5 G–I). Overall, these functional results are consistent with coimmunoprecipitation experiments showing that F87A and E109A mutations significantly impair the binding of KCTD12 to the GABA_B2 subunit, similar to the homologous mutations in KCTD16. Taken together, these results suggest the GABA_B2 receptor uses the same interface for binding both KCTD12 and KCTD16 (Fig. 5 C–F and SI Appendix, Fig. S5 C and D).

Discussion

The class C GABA_B receptor exists as a heterodimer in a complex with a family of auxiliary subunit KCTD proteins that have been shown to modulate the functional properties of the receptor (20–22). KCTD proteins contain an N-terminal tetramerization domain (T1) (28) and a single C-terminal homology domain (H1) (20, 23, 29), or two H1 domains in the case of KCTD8 and KCTD16 (20, 23). Here, we describe the interactions that govern the direct association of KCTD16 with the GABA_B receptor, based on determination of the atomic resolution structure of the T1 domain of human KCTD16 bound to a C-terminal GABA_B2-derived peptide.

The atomic structure of the GABA_B2 peptide and KCTD16 T1 domain likely reflects the interactions of the native full-length GABA_B2 receptor and KCTD16 protein for several reasons. First, single point mutations of amino acids located at the GABA_B2(895–909)/KCTD16 T1 interface disrupt both the biochemical association and the functional modulation of GABA_B receptor-dependent activation of GIRK channels. These mutations targeted both the GABA_B2 receptor (I898, Y903, and L904) and the T1 domain of KCTD16 (Q34, F80, P101, and E102). Second, we constructed a homology model of GABA_B2(895–909) with KCTD12 T1 and found that mutation of homologous interfacial residues in full-length proteins similarly reduces the biochemical and functional association of KCTD12 with GABA_B receptors. Taken together, these findings suggest the structural arrangement of the GABA_B2–KCTD16 T1 interface may broadly apply to KCTDs that regulate GABA_B receptors. Lastly, previous studies implicated a short peptide region, including Y903 of GABA_B2, in the interaction between GABA_B2 and KCTD. Schwenk et al. (20) showed that all four KCTDs (8, 12, 12b, and 16) copurify with a rat GABA_B2 construct that is C-terminally truncated at 906, but not at 901. Furthermore, a single mutation, Y902A, in rat GABA_B2 (Y903 in humans) was sufficient to abolish its interaction with KCTD (20). Correale et al. (29) demonstrated that the cytoplasmic tail of human GABA_B2 has an increased helical content upon binding KCTD12 T1, consistent with our finding that the KCTD-binding region in GABA_B2 centers on a helix. Lastly, a human GABA_B2 peptide containing residues 890–907 binds to KCTD12 T1 with high affinity (∼400 nM K_d) (29), indicating the short GABA_B2-derived peptide is sufficient to interact with KCTD T1.

Aligning different KCTDs reveals that the GABA_B2-binding residues in KCTD16 are highly conserved in the subset of KCTDs (i.e., group D) that modulate GABA_B receptors, including KCTD8, KCTD12, KCTD12b, and KCTD16 (36) (SI Appendix, Fig. S3). Most importantly, the conserved hydrophobic residue F80 from four of the KCTD subunits forms extensive contacts with several regions of the GABA_B2 peptide (H900, H901, L904, P905, S906, I907), and provides an essential framework for anchoring the peptide. The conserved polar residues Q34 and E102 form hydrogen bonds with GABA_B2 residues H901 and Y903 (Y902 in rats) in all three complex structures in the asymmetrical unit, which has previously been shown to be important for KCTD modulation (20). The interfacial residues F80 and Q34 are unique to the subfamily of KCTDs that associate with GABA_B receptor, thus providing a plausible explanation for the lack of functional interaction of other KCTDs. Conservation of amino acids in the GABA_B2–KCTD interface is also compatible with the observation that KCTDs can form heteroligomers and functionally modulate GABA_B receptors (37).

The structure of the GABA_B2–KCTD16 complex reveals two important features of this complex: (i) a binding stoichiometry of five KCTD16 subunits (i.e., pentamer) with one GABA_B2 peptide and (ii) an open pentameric structure. The pentameric arrangement was unexpected, given the presence of the potassium channel tetramerization domain in these KCTD proteins (19–21). In fact, pentameric assemblies have been observed for several KCTD proteins that are not associated with GABA_B receptor, including KCTD1 (31) and KCTD5 (30). Previous biophysical studies suggested that KCTD12 may exist as a tetramer in solution (20–22). However, our biochemical and functional studies of KCTD12 indicated a binding surface similar to that of KCTD16 and GABA_B2 receptor, supporting a pentameric arrangement for KCTD12. The open pentameric structure of KCTD16 T1 raises the question of whether it could adopt a hexameric organization. Although this is possible in some other crystal form, we think it is unlikely. First, the KCTD16 T1 structure is not compatible with the binding of a sixth subunit despite having a gap. The gap is ∼16 Å at its widest and 8 Å at its narrowest. By contrast, a KCTD16 T1 monomer in the plane of the pentamer ranges from 10 to 25 Å, which is too large to fit in this gap. Second, we also examined a superimposition of the five subunits within a KCTD16 T1 pentamer, and found all five to be highly homologous (rmsd range: 0.27–0.49 Å), with little difference between the apo- and peptide-bound structures (SI Appendix, Fig. S1). The structure of KCTD16 T1 is also pentameric in the absence of the GABA_B receptor peptide, indicating the peptide is not required to form the open pentamer.

Interestingly, the first four of the five KCTD16 subunits form direct contacts with the peptide in the open pentameric structure, although other binding arrangements are theoretically possible. Closer examination of the complex structure provides a possible explanation for how the GABA_B2 peptide is uniquely bound to a specific subset of KCTD16 subunits. Although structurally similar, the phenylalanines (F80) that comprise a hydrophobic belt around the center of the pentamer are not all oriented the same. Instead, the five F80 phenylalanines form a spiraling ladder with a slight screw offset (SI Appendix, Fig. S6). This screw dislocation is caused by the tilting of each KCTD16 subunit within the pentamer. We aligned protomers I–IV with the bound GABA_B2 peptide onto protomers II–V of the pentamer (SI Appendix, Fig. S7 A and B). If the GABA_B2 peptide were to bind to protomers II–V, it would be rotated within the plane of the pentamer and dislocated vertically along a pseudo-fivefold axis. Since the conformation of the GABA_B2 peptide is mostly straight up, except for the helix at the turn, this movement causes a shift in the GABA_B2 peptide–KCTD16 T1 interface. As a result, many contacts are lost, except for those maintained by F80 and Q34 of KCTD16 (SI Appendix, Fig. S7C). Most notably, Y903 of GABA_B2 retains a hydrogen bond with E102 of KCTD16 but loses all of the hydrophobic contacts with KCTD16 residues, including P101. These hydrophobic interactions are critical for maintaining the specificity and affinity of GABA_B2–KCTD binding since a rat GABA_B2 Y902F mutant retains the ability to coimmunoprecipitate KCTD proteins, but not the Y902A mutant (20). Taken together, the binding site formed by KCTD16 protomers I–IV has higher specificity and affinity for GABA_B2 peptide than that formed by protomers II–V.

The structure of the GABA_B2 peptide-bound KCTD16 T1 has broad implications for the architectural arrangement of the GABA_B receptor and KCTD at the plasma membrane. In the crystal structure, the GABA_B2 peptide loops around inside an open ring structure formed by KCTD16 T1. As a result, both the N and C termini of GABA_B2 point in the direction of the KCTD16 N terminus. Previous studies have provided evidence that KCTD8, KCTD12, and KCTD16 all constitutively bind G proteins through Gβγ (22). In addition, KCTD12 induces GABA_B receptor desensitization through a dynamic interaction between a unique motif in the H1 domain and Gβγ; the absence of the H2 domain also appears to be important for rapid desensitization (23). Based on our structural and homology models of GABA_B-associated KCTD proteins, we propose that the binding of the GABA_B receptor to the N-terminal T1 domain of KCTD proteins allows the C-terminal H1 domain to interact with G proteins and modulate the functional signaling of the GABA_B receptor (22, 23) (SI Appendix, Fig. S8).

Genetic studies have shown that KCTD12 is associated with major depressive disorder, bipolar disease, and schizophrenia (24, 25). A genome-wide association study in 1,000 bipolar I patients and 1,000 controls revealed SNPs in KCTD12, which may affect expression (25). Microarray studies of schizophrenic hippocampi revealed an increase in expression of KCTD12 (24). The atomic resolution structure of the interaction site between the GABA_B receptor and KCTDs we describe here provides an opportunity to develop novel ligands for potentially modulating the association of KCTD12 with GABA_B receptors in vivo.

Materials and Methods

Protein Expression and Purification.

The T1 domain of human KCTD16 (KCTD16 T1, residues 22–134 of human NP_065819.1) was cloned into the pSMT3 (38) vector for expression as small ubiquitin-like modifier (SUMO) fusion protein in bacteria. A hexahistidine tag was engineered at the N terminus to facilitate affinity purification. An additional construct was generated for structural analysis mutating Leu89 to Met (L89M) using the QuickChange mutagenesis system (Agilent Technologies).

SUMO-fused wild-type KCTD16 T1 protein was expressed in BL21-CodonPlus (DE3)-RIL cells, which contained extra copies of rare Arg (R), Ile (I), and Leu (L) tRNA genes. Bacterial culture was grown at 37 °C until mid-log phase. Protein production was induced with the addition of 0.5 mM isopropyl β-d-1-thiogalactopyranoside at 20 °C for 16 h. Harvested cells were lysed by sonication. The fusion protein was first isolated using Co²⁺-based immobilized metal ion affinity chromatography and then digested with ubiquitin-like protein-specific (Ulp1) protease (38) to cleave SUMO. The digested protein mixture was subjected to a second round of Co²⁺-based affinity chromatography to separate SUMO from KCTD16 T1. While His-tagged SUMO remained bound, KCTD16 T1 was contained in the flowthrough. The KCTD16 T1 protein was further purified by size exclusion chromatography (Superdex 200; GE Healthcare), followed by anion exchange chromatography (Mono Q; GE Healthcare).

For the KCTD16 T1 L89M mutant, SeMet-substituted SUMO fusion protein was prepared as described for the wild-type protein, with the following exceptions. First, bacterial culture was grown in M9 minimal media. Second, protein expression was induced in the presence of 60 mg/L SeMet, along with 50 mg/L each of the l-amino acids Lys, Phe, Thr, Ile, Leu, and Val. Finally, protein purification was carried out in the presence of the reducing reagent DTT. Mass spectrometry confirmed the incorporation of two SeMet residues in the KCTD16 T1 L89M protein, at position 54 found in the native sequence and at position 89 introduced through mutation.

To determine the effect of mutations on KCTD16 pentamer assembly, four mutant constructs were created carrying single-charge reversal mutations D76R, R77D, D78R, and R105D, respectively. Protein expression and purification were performed following a similar procedure as for the wild-type construct, with several modifications. Briefly, after the initial Co²⁺-based affinity chromatography, the SUMO-KCTD16 T1 fusion protein was purified by anion exchange chromatography (Mono Q). SUMO was then cleaved from KCTD16 T1 by Ulp1 digestion and removed from the digested protein mixture through Co²⁺-based affinity chromatography. Finally, size exclusion chromatography (Superdex 200) was used to further purify and assess the oligomeric state of each protein construct. All of the purification steps were performed in the presence of 5% glycerol for protein stabilization. In addition, the buffer pH was raised from 7.5 to 8.5 to improve solubility of the D76R mutant since it has an isoelectric point (IP) of 8.2, which is higher than the IP value of 6.3 for wild-type protein.

A GABA_B2-derived peptide [GABA_B2(895–909)] that contains human GABA_B2 (NP_005449.5) residues 895–909 (⁸⁹⁵QLPILHHAYLPSIGG⁹⁰⁹) was synthesized. In addition, a tagged version of GABA_B2(895–909) includes a Flag tag attached at the N terminus [Flag-GABA_B2(895–909), DYKDDDDK⁸⁹⁵QLPILHHAYLPSIGG⁹⁰⁹]. The binding of each GABA_B2-derived peptide to purified KCTD16 T1 was assayed on a 12% native gel.

Protein Crystallization.

KCTD16 T1 was crystallized in two different forms at 20 °C. Form I crystals of wild-type KCTD16 T1 were grown in 20% ethanol and 0.1 M Tris (pH 7.0). The crystals were transferred stepwise to a cryoprotectant solution containing the precipitants for crystallization and 20% PEG 400 in 5% increments of PEG 400. The crystals were then flash-cooled with liquid nitrogen. Diffraction data were measured to 2.25 Å at the 24ID-E beamline of the Advanced Photon Source (APS).

Form I crystals of SeMet-incorporated KCTD16 T1 L89M mutant were obtained from 12% ethanol, 4% PEG 400, and 0.1 M Tris (pH 7.0). The crystals were flash-cooled in a cryoprotectant solution containing 12% ethanol, 20% PEG 400, and 0.1 M Tris (pH 7.0). Diffraction data were collected to 2.5 Å at the APS 24ID-C beamline. Form I crystals of both wild-type and mutant KCTD16 T1 belong to the P1 space group, with 15 molecules per asymmetrical unit. Data were integrated and scaled with HKL2000 (39).

Form II crystals of wild-type KCTD16 T1 were grown in 6% PEG 6000, 2% glycerol, and 0.1 M sodium cacodylate trihydrate (pH 6.0). The crystals were transferred stepwise to a cryoprotectant solution containing 6% PEG 6000, 20% glycerol, and 0.1 M sodium cacodylate trihydrate (pH 5.6) in 5% increments of glycerol, and then flash-cooled with liquid nitrogen. Diffraction data were collected to 2.80 Å at the X4A beamline of the National Synchrotron Light Source. The form II crystals have a P3₂ space group with 10 molecules in the asymmetrical unit. Data were integrated using XDS (40) and scaled with AIMLESS (41).

The GABA_B2(895–909) peptide was solubilized in DMSO, and mixed with purified wild-type KCTD16 T1 at a molar ratio of 1.5:5 for crystallization. Crystals of the GABA_B2(895–909)/KCTD16 T1 complex were obtained from 2% PEG 6000, 2% glycerol, 8% PEG 400, and 0.1 M sodium cacodylate trihydrate (pH 5.6). The complex crystals were flash-cooled in a cryoprotectant solution containing the precipitants for crystallization and 20% glycerol. Diffraction data were collected to 2.35 Å at the APS 24ID-C beamline. These crystals belong to the P1 space group with 15 molecules per asymmetrical unit. Data were integrated using XDS (40) and scaled with AIMLESS (41).

Structure Determination.

The structure of KCTD16 T1 in crystal form I was solved by the SIRAS method. Data were collected from a SeMet-substituted KCTD16 T1 L89M crystal at the absorption edge of Se (λ = 0.9292 Å) to obtain experimental phases. The positions of 30 Se atoms in the crystal were located with the Phenix software suite (42) using the anomalous differences of Se atoms. These correspond to the locations of two Met residues (Met-54 and Met-89) in each of the 15 KCTD16 T1 L89M molecules in the asymmetrical unit. Refinement of Se atom parameters, phase calculation, and density modification were all performed with Phenix (42). The resulting SIRAS electron density maps allowed us to build an initial model in Coot (43) that contained 15 KCTD16 T1 molecules in three pentameric assemblies. This model was refined against the native data to 2.35 Å using BUSTER (44), applying noncrystallographic symmetry restraints between individual KCTD monomers throughout refinement. In the final structure, the 15 KCTD16 T1 molecules in the asymmetrical unit had slightly different stretches of ordered residues. One protomer contained residues 22–123. In the other 14 protomers, the loop region around residues 58–64 was not visible. The C-terminal region around residues 125–134 was disordered in all of the KCTD16 T1 molecules.

The structure of KCTD16 T1 in crystal form II was solved by molecular replacement with Phaser (45) using a monomer structure in crystal form I as the search model. Each asymmetrical unit consisted of 10 copies of KCTD16 T1 forming two pentameric assemblies. Structural refinement was carried out using BUSTER (44). In the final model, most of the KCTD16 T1 protomers in the asymmetrical unit contained residues 22–57 and 64–124. The loop region around residues 58–63 and C-terminal tail around residues 125–134 were not visible in the electron density map of all of the protomers.

The structure of the GABA_B2(895–909)/KCTD16 T1 complex was solved by molecular replacement with Phaser (45) using a pentamer structure of KCTD16 T1 in crystal form I as the search model. After initial refinement, GABA_B2(895–909) was modeled into the residual electron density map. Structural refinement was completed with BUSTER (44). The conformation and geometry of the bound peptide were unambiguously supported by examination of omit maps. The final refined structure contained three GABA_B2(895–909)/KCTD16 T1 complexes within the asymmetrical unit. Each complex consisted of a single GABA_B2(895–909) molecule bound to a KCTD16 T1 pentamer. GABA_B2 residues 895–909 were visible in one of the GABA_B2(895–909) molecules, while residues 896–907 were ordered in the other two. The KCTD16 T1 residues 23–57 and 65–123 were visible in most KCTD16 T1 protomers in the asymmetrical unit. The loop region around residues 58–64 and C-terminal region around residues 124–134 were disordered in the electron density map of all of the KCTD16 T1 protomers.

Ramachandran analysis of each structure was performed with MolProbity (46).

Structural Analysis and Homology Modeling.

Pairwise structural comparison was performed using LSQMAN (47). Shape correlation statistics were quantified using the program sc in the CCP4 package (35). The homology model of a GABA_B2(895–909)/KCTD12 T1 complex was generated based on the structure of the GABA_B2(895–909)/KCTD16 T1 complex using SWISS-MODEL (48). Figures were produced in PyMOL (https://www.pymol.org/2/). Software installation support was provided by SBGrid (49).

Coimmunoprecipitation.

Full-length human GABA_B1b and GABA_B2 were cloned into a pcDNA3.1(+) vector (Life Technologies) with a Flag tag and an HA tag, respectively, inserted after their signal peptides. Full-length human KCTD16 (NP_065819.1) and KCTD12 (NP_612453.1) were cloned into a pIRES-EGFP vector (Clontech) with an HA tag added after the initial Met residue of each protein. Mutants of GABA_B2, KCTD16, and KCTD12 were constructed using the QuikChange mutagenesis system. We selectively mutated residues that are buried deeply at the GABA_B2(895–909)/KCTD16 T1 interface and form extensive interfacial contacts. The single mutations include L896A, I898S, H901A, Y903S, and L904D of GABA_B2 and Q34A, F80A, P101S, and E102A of KCTD16. Each mutation was designed to remove interfacial contact between GABA_B2(895–909) and KCTD16 T1, and the introduction of polar and charged residues was specifically aimed at disrupting hydrophobic interactions. Similarly, the mutations Q43A, F87A, P108S, and E109A of KCTD12 were designed to interfere with the potential interaction between GABA_B2(895–909) and KCTD12 T1.

HEK293 T/17 cells (American Type Culture Collection) were cultured in DMEM (Gibco) supplemented with 10% FBS at 37 °C in 5% CO₂. Cells were transiently transfected with the GABA_B1b, GABA_B2, and KCTD plasmids using Lipofectamine 3000 (Life Technologies). Control experiments involved cotransfection of GABA_B1b and KCTD or GABA_B1b and GABA_B2.

Cells were harvested 72 h posttransfection; washed in PBS; and lysed in a lysis buffer containing 50 mM Tris [pH 7.5 (KCTD12) or pH 8.5 (KCTD16)], 300 mM NaCl, 1% Nonidet P-40, and protease inhibitor mixture (Roche). The cell extract was incubated with anti-Flag affinity gel at 4 °C overnight to capture GABA_B1b. The resin was washed extensively with the lysis buffer to remove any protein not precipitated on the beads. The bound GABA_B1b and associated proteins were then eluted with 0.2 mg/mL Flag peptide. The samples were analyzed by SDS/PAGE and Western blot. Mouse anti-Flag M2 antibody (Sigma) was used to detect GABA_B1b. Mouse anti-HA antibody HA.11 clone 16B12 (BioLegend) was used to probe GABA_B2 and KCTD. Each assay was replicated in a total of three experiments.

The results of coimmunoprecipitation were quantified using ImageJ (NIH) and the “Gels” function. Since HA-KCTD and HA-GABA_B2 were visualized on the same blot by anti-HA antibody but had different molecular weights, we measured the optical density for the KCTD band and for the GABA_B2 band, calculated the ratio of HA-KCTD/HA-GABA_B2, and averaged over three to five separate experiments.

Electrophysiology.

For whole-cell patch-clamp electrophysiology, HEK293T cells were cultured in DMEM (Sigma–Aldrich) supplemented with 10% (vol/vol) FBS, 100 units/mL penicillin, 100 μg/mL streptomycin, and 1× Glutamax (Thermo Fisher Scientific) at 37 °C and in a 5% CO₂-humidified atmosphere. Cells were plated onto poly-d-lysine (100 μg/mL; Sigma–Aldrich)–coated 12-mm glass coverslips in 24-well plates and transiently transfected with cDNA using Lipofectamine 2000 (Thermo Fisher Scientific). HEK293T cells were transiently transfected with human GABA_B1b (0.2 μg/mL), GABA_B2 (0.2 μg/mL), and mouse GIRK2c (0.2 μg/mL) cDNAs alone or with human HA-KCTD [KCTD12 or KCTD16 wild-type (WT) or mutant; 0.2–0.6 μg/mL] cDNAs. Enhanced YFP cDNA (0.02 μg/mL) was included for identifying transfected cells. YFP⁺ cells were recorded within 48 h following transfection. The same mutant KCTDs generated for coimmunoprecipitation experiments were used here.

Whole-cell patch-clamp experiments were performed as previously described (50). Briefly, currents were recorded with an Axopatch 200B amplifier (Molecular Devices), digitized with a Digidata 1320a (Molecular Devices) at 5 kHz, filtered at 10 kHz, and stored on a computer. Borosilicate glass electrodes (Warner Instruments) of 3–5 MΩ were filled with an intracellular solution containing 130 mM KCl, 20 mM NaCl, 5 mM EGTA, 5.46 mM MgCl₂, 2.56 mM K₂ATP, 0.3 mM Li₂GTP, and 10 mM Hepes (pH 7.4, ∼313 mOsm). The extracellular “20K” solution contained 20 mM KCl, 140 mM NaCl, 0.5 mM CaCl₂, 2 mM MgCl₂, and 10 mM Hepes (pH 7.4, ∼318 mOsm). Extracellular BaCl₂ (1 mM) was added to the 20K solution to monitor the agonist-independent basal and leakage currents. (±)-Baclofen (Sigma–Aldrich) was diluted from a stock to 100 μM in the 20K solution, and applied locally through a gravity-fed perfusion manifold (MM8; Warner Instruments).

For KCTD16 experiments, K⁺ currents were recorded continuously at −100 mV (20 kHz) during perfusion with 20K solution in the absence and presence of baclofen (100 μM). The activation time was measured as the rise time between 20% and 80% of baclofen-activated peak current. For KCTD12 experiments, K⁺ currents were elicited at 0.5 Hz by a voltage step from a holding potential of −40 to −120 mV, followed by a ramp from −120 to +50 mV to monitor inward rectification. Currents were measured at −120 mV (5 kHz). GABA_B-mediated responses were evoked with the addition of 100 μM baclofen in the extracellular solution. The extent of current desensitization was calculated using the equation, percentage desensitization = (I_ss − I_p)/I_p × 100%, where I_p represents the amplitude of the baclofen-activated peak current and I_ss is the steady-state current. The desensitization rate of decay (Tau) was derived from a single-exponential fit between I_p and I_ss for only experiments with KCTD12.

Immunostaining.

HEK293T cells were transiently transfected with GABA_B1b, GABA_B2, GIRK2c, and/or either HA-tagged KCTD16 WT or KCTD16 D76R cDNAs. Cells were plated onto poly-d-lysine–coated glass coverslips in 24-well plates. After 24 h, cells were washed with PBS and fixed with 2% paraformaldehyde in PBS for 10 min. Cells were then permeabilized with 0.1% Triton X-100 in PBS for 5 min and blocked with 1% BSA in PBS at room temperature for 50 min. Subsequently, cells were incubated with 5 μg/mL rat anti-HA antibody (Thermo Fisher Scientific). Cy3-conjugated secondary antibody (Abcam) was used to detect HA tag, Alexa Fluor 488 phalloidin was used to label actin, and DAPI was used to label the nucleus. Images were taken on a confocal microscope using the same acquisition settings for all transfections.

Statistical Analyses.

Statistically significant differences were assessed by one-way ANOVA with Bonferroni’s multiple comparison post hoc test for significance (P < 0.05). Values are reported as mean ± SEM for n > 3 and mean ± SD for n = 3.