Regulation of synaptic inhibition by phospho-dependent binding of the AP2 complex to a YECL motif in the GABA_A receptor γ2 subunit

A Yxxφ-Type μ2–AP2-Binding Motif Specific to the GABA_AR γ2-Subunit.

Tyrosine (Yxxφ) motifs target a variety of cargo proteins, including some ion channels for clathrin-mediated endocytosis via direct binding to a pocket within subdomain A of μ2–AP2 (11, 12, 18–20). Analysis of the amino acid sequence of the γ2-subunit ICD revealed the presence of a conserved putative classical Yxxφ motif (YECL; residues 367–370) absent in GABA_AR β-subunits. Peptides containing the Yxxφ motif from other membrane proteins can effectively bind μ2–AP2 in vitro (11, 12). To determine whether the Y³⁶⁷ECL³⁷⁰ motif in the γ2-subunit ICD can bind μ2–AP2 directly, we synthesized a peptide containing the YECL motif and upstream sequence (YECL-pep) and tested its ability to interact with [³⁵S]-labeled μ2–AP2. We immobilized either YECL-pep or a version of this peptide containing Y³⁶⁷ mutated to alanine (A³⁶⁷ECL-pep; control mutant) and looked at binding to a [³⁵S]-labeled fragment of μ2–AP2 (residues 158–435). Whereas YECL-pep exhibited robust binding to μ2–AP2, virtually no binding could be detected for the AECL mutant peptide or beads alone (Fig. 1A). We also identified the domain of μ2–AP2 that mediates binding to the GABA_AR γ2-subunit YECL motif. The removal of the carboxyl-terminal 30 amino acids of μ2–AP2 (or even the mutation of W421 to alanine) (21) is sufficient to disrupt the binding of Yxxφ-type signals to μ2–AP2. A [³⁵S]-labeled carboxyl-terminal truncation of μ2–AP2 (residues 158–407) could not bind to YECL-pep (Fig. 1A). Similar results were obtained when we examined the ability of bacterially expressed His-μ2 to bind YECL-pep beads. His-μ2 exhibited significant binding only to YECL-pep, but not to AECL-pep or beads alone (Fig. 1B). Furthermore, His-μ2 containing a W421A mutation exhibited substantially reduced interaction with the YECL-pep beads (≈15–20% binding remained).

In addition, we used surface plasmon resonance (SPR) to measure the affinity of the YECL-pep μ2–AP2 interaction as described previously for several other μ2–AP2 interaction motifs (10, 17, 22). Either YECL-pep (wild type) or AECL-pep (mutant used for the reference) was immobilized on a CM5 sensor surface, and their ability to bind recombinant His-μ2 was recorded in real time by using an SPR-based biosensor. This approach revealed that YECL-pep bound with a high nanomolar affinity to AP2 (K_d = 42.2 nM) (Fig. 1 C and D), providing compelling evidence that the γ2-subunit-specific YECL motif can mediate high-affinity μ2–AP2 binding.

A Peptide Including the YECL Motif Increases Inhibitory Synaptic Responses.

To determine the functional consequences of altering the recruitment of AP2 to the GABA_AR via the γ2-subunit-specific YECL motif, we carried out whole-cell patch-clamp electrophysiological experiments to monitor the effects on inhibitory synaptic transmission of dialyzing a YECL motif-containing peptide [(YECL-pep) which we predicted would compete with AP2 binding and block receptor internalization] into neurons via the patch pipette. The control for these experiments was the identical peptide containing Y³⁶⁷ mutated to A (AECL-pep), which displays dramatically reduced AP2 binding and would therefore not compete with γ2-subunit-dependent AP2 binding to native receptors. As shown in Fig. 2, control striatal neurons showed a stable mIPSC amplitude within 60 min from the onset of recording (Fig. 1 E and H–J). In contrast, dialysis of the YECL-pep via the patch pipette caused a sustained increase in mIPSC amplitude (but not frequency) over the same 60-min time course (YECL-pep, 20.9 ± 2.4%, n = 7; control, 2.8 ± 0.9%, n = 7) (Fig. 1 E–G and J). Therefore, we conclude that the YECL μ2–AP2-binding motif within GABA_AR γ-subunits plays a critical role in regulating the number of synaptic GABA_ARs on a relatively rapid time scale. These results also correlate with an increase in the total number of surface GABA_ARs in cultured neurons treated with a membrane-permeant YECL-pep, compared with control AECL-pep-treated neurons, as determined by using surface biotinylation [supporting information (SI) Fig. 6].

Simultaneous Targeting of GABA_AR β3-Subunit and γ2-Subunit μ2–AP2 Interactions Causes an Additive Enhancement of mIPSCs.

We have previously identified a major AP2-binding site in the GABA_AR β3-subunit distinct from the YECL motif identified in γ2-subunits. Disruption of the β3–AP2 interaction [using a β3 AP2-binding motif peptide (β3-pep)] enhances the amplitude of mIPSCs in a similar fashion to what is seen here for dialysis of the YECL-pep (10), suggesting that both these μ2–AP2 interaction mechanisms are important for regulating the number of surface and synaptic αβγ-containing GABA_ARs. To investigate this idea further, we monitored the consequences, on the size of mIPSCs, of simultaneously targeting both these AP2 interaction mechanisms with the GABA_AR by using codialysis of β3-pep and YECL-pep. As shown in Fig. 2 A–D, codialysis of YECL-pep with the previously described β3-pep (10) over the same 60-min time course produced a significant additive effect on mIPSC amplitude, compared with interfering with GABA_AR–AP2 interactions individually by using β3-pep or YECL-pep alone or codialysed control peptides (β3-pep plus YECL-pep, 42.8 ± 3.2%, n = 6; β3-pep, 28.3 ± 2.3%, n = 6; YECL-pep, 21.2 ± 1.9%, n = 6; AECL plus β3-phos-pep, 3.4 ± 0.3%, n = 6). The control for β3-pep is an identical version of this peptide phosphorylated at two serine residues (β3-phos-pep) and that no longer interacts with AP2 (10).

Crystal Structure of the GABA_AR γ2-Subunit-Derived YECL-Pep Complexed with μ2–AP2.

To further examine the molecular nature of the interaction of the YECL motif and its surrounding residues with μ2–AP2, we cocrystallized a 10-aa peptide (DEEYGYECLD) corresponding to GABA_AR γ2-subunit (residues 362–371) with the cargo-binding domain of μ2–AP2 (residues 157–435; μ2Δ157). The structure of the GABA_AR YECL-pep complexed with μ2Δ157 was determined at a resolution of 2.5 Å (SI Table 1). Amino acid groups Glu2OE, Tyr6OH, Glu7O, N, and Leu9N of the YECL-pep interacted with μ2–AP2 (residues 157–435) amino acid groups Asp176OD, Lys203NZ, Lys319NZ, Lys420O, Val220O, N, Arg423NE, and NH1 as determined by electron density for peptide residues Asp-1 (D³⁶²) to Asp-10 (D³⁷¹) (see Fig. 3 A and B and SI Table 2; for stereoview of μ2Δ157 complexed with YECL-pep, see SI Fig. 7), which encompasses the canonical tyrosine Yxxφ endocytic signal ⁶YECL⁹. GABA_AR YECL-pep binds at the identical surface location identified for the binding of the FYRALM and DYQRLN hexapeptides corresponding to the canonical Yxxφ motifs of the EGF receptor (EGFR) and trans-Golgi network protein 38 (TGN-38), respectively (23), with Y³⁶⁷ and L³⁷⁰ sitting in chemically compatible pockets and playing key roles in mediating the interaction (Fig. 3B). Interestingly, the structure (Fig. 3 A and B) revealed additional interactions between μ2–AP2 and residues upstream of the canonical Y³⁶⁷ residue. Specifically, γ2-subunit residues D³⁶¹-Y³⁶⁵ (Asp-1, Glu-2, Glu-3, and Tyr-4) interact with residues Leu-316, Lys-319, Gln-318, Glu-391, Val-392, Pro-393, and Ile-425 of the μ2–AP2 subdomain A (Fig. 3 A and B). Similarly to Y³⁶⁷ (Tyr-6), a hydrophobic pocket is formed for Y³⁶⁵ (Tyr-4) (Fig. 3 A and B), with an additional salt bridge between Lys-319 and Glu-2 (see SI Tables 1 and 2). Therefore, our structural data reveal that, in addition to Y³⁶⁷ within the YECL motif, the upstream Y³⁶⁵ residue also plays a role as an additional specificity determinant of YECL-pep binding to μ2–AP2, similar to the three-pin plug mechanism reported for the association of a μ2–AP2 Yxxφ-binding peptide derived from the membrane protein P-selectin (24).

Overall, 111 atomic contacts (2–5 Å) formed between μ2Δ157 and GABA_ARYECL-pep, compared with the 18 contacts formed by FYRALM (see SI Tables 1–4). A comparison of calculated contact surface areas for binding of YECL-pep and FYRALM-pep (CCP4 program ArealMol) shows an increase of 44% in contact surface area for GABA_ARYECL-pep [505 Ų (YECL) vs. 350 Ų (FYRALM)] (SI Fig. 8 and SI Table 4). The largest increase in contact areas is contributed by Leu-316, Gln-318, Lys-319, Glu-391, Val-392, Pro-393, and Ile-425 (Fig. 3B). Only His-416 shows a larger surface contact with FYRALM-pep. In the case of GABA_ARYECL-pep, His-416 has moved out of the binding pocket and no longer contributes to the van der Waals surface.

μ2–AP2 can form dimers, which could increase the strength and specificity of binding to dimeric receptors (23). Even larger differences are observed when the binding characteristics are compared for the peptide interactions with the (μ2-YECL)₂ dimer, the monomers being related by a crystallographic C2 axis (Fig. 4 A and B and SI Tables 3 and 5; for a stereoview of the crystallographic dimer complexed with YECL peptide, see SI Fig. 9). GABA_ARYECL-pep exhibits direct molecular interactions with residues Arg-170, Phe-174, Ile-419, and Trp-421 of the other monomer within the crystallographic dimer (see Fig. 4B and SI Table 3), which form additional binding pockets around the peptide residues Tyr-6 (Y³⁶⁷) and Glu-2 (E³⁶⁸). No such interactions exist in the case of the FYRALM–μ2 complex. Furthermore, the alignment of the two GABA_AR YECL-pep dimers allows for the formation of one salt bridge (lys319NZ… Glu7OE) (SI Tables 2 and 3), which also is absent in the FYRALM–μ2 complex. Fig. 4A depicts the elongated binding pockets of (μ2–AP2 × YECL-pep)₂, one peptide shown as a wire model and the other in a surface representation. Importantly, these results suggest that γ2-subunit dimers (presumably with γ2-subunits contributed from two different heteromeric receptors) may be complexed within AP2-coated vesicles upon endocytosis, suggesting that multiple or possibly even clustered receptors may be internalized within one endocytic event.

μ2–AP2-Binding to the YECL Motif Is Regulated by Phosphorylation of Y³⁶⁵ and Y³⁶⁷.

Residues Y³⁶⁵ and Y³⁶⁷ within the γ2 subunit are major sites of tyrosine phosphorylation in the GABA_AR and are substrates of SRC family tyrosine kinases both in vitro and in vivo (4, 5). The crystallographic data predict that phosphorylation at Y³⁶⁷ within the YGYECL motif would preclude μ2-binding because there is insufficient space to accommodate a phosphate group within the binding pocket (and, in addition, D⁴⁷³ within μ2–AP2 would repel binding), as has been shown previously for other tyrosine motif-AP2 interactions (11, 12, 23). In addition, the YECL peptide–μ2–AP2 structure reveals that, because Y³⁶⁵ also contributes to μ2–AP2 binding (by hydrogen bonding with E³⁹¹), phosphorylation at this site also may negatively regulate μ2–AP2 binding because phospho-Y³⁶⁵ would repel the interaction with E³⁹¹.

To test whether phosphorylation of Y³⁶⁵ and/or Y³⁶⁷ can indeed inhibit binding to μ2–AP2, we immobilized either the YECL-pep or a version of this peptide that had been chemically phosphorylated on Y³⁶⁵ and Y³⁶⁷ (YGYECL-phos) onto beads and looked at binding to [³⁵S]-labeled μ2–AP2. YECL-pep exhibited robust binding to μ2–AP2 in this assay (in agreement with the results using SPR), whereas a peptide that was chemically phosphorylated on both Y³⁶⁵ and Y³⁶⁷ bound virtually no [³⁵S]-labeled μ2–AP2 (Fig. 5A). We carried out similar experiments to test the binding of the tyrosine motif peptides in their phosphorylated or unphosphorylated forms to AP2 from brain lysate by using SDS/PAGE and detection of bound AP2 by either Coomassie blue staining followed by MALDI mass spectrometric analysis of bands from bound complexes or Western blotting with AP2 antibodies. When an unphosphorylated version of the peptide containing the YECL motif was coupled to beads and then exposed to brain lysate, Coomassie blue staining of bound protein complexes analyzed by SDS/PAGE revealed major interacting bands associated with this peptide at 50 and 115 kDa (these bands represent the correct molecular weight for the μ2- and α-subunits of AP2, respectively) (see Fig. 5B, lane 1). That these bands represented AP2 subunits was confirmed by the excision of the bands, followed by Maldi-TOF mass spectrometry, further confirming the ability of the YECL-pep to strongly interact with brain AP2. In contrast to the unphosphorylated peptide, binding of AP2 to a peptide phosphorylated on Y³⁶⁵ was reduced (as shown by much weaker Coomassie blue-stained bands at 50 and 115 kDa). A more substantial reduction in the interaction was observed when the peptide was phosphorylated on Y³⁶⁷, whereas the greatest reduction was seen with the diphosphorylated peptide (Y^365phos and Y^367phos), which showed essentially no AP2 binding (Fig. 5B). Similar results were obtained by Western blot analysis of YECL-pep brain pulldown assays by using an antibody to μ2–AP2 (Fig. 5C).

These results demonstrate that YGYECL motif binding to AP2 is regulated by phosphorylation and consequently predict that phosphorylation of Y³⁶⁵ and/or Y³⁶⁷ would block YGYECL-dependent, AP2-mediated internalization and result in an increase in surface receptor number (in agreement with earlier reports that SRC phosphorylation of Y³⁶⁵/Y³⁶⁷ enhances receptor function) (4). It has been reported that in cortical neurons blocking tyrosine phosphatase activity by orthovanadate leads to a large increase in phosphorylation at Y³⁶⁵ and Y³⁶⁷ (5). To test whether treatment of cortical neurons with orthovanadate (increasing Y³⁶⁵/Y³⁶⁷ phosphorylation) (5) also resulted in an increase in the number of surface GABA_AR as would be predicted from the above results, we used surface biotinylation of cultured neurons, which, as we have previously shown, is an effective reporter of surface receptor number (25). Orthovanadate treatment produced a statistically significant increase of 159.7 ± 11% of control (P < 0.05, n = 6) in the cell-surface number of GABA_ARs (Fig. 5 D and E). Together these results provide evidence that phosphorylation of Y³⁶⁵ and Y³⁶⁷ can directly regulate μ2–AP2-binding affinity and the number of surface GABA_ARs.

Antibodies and cDNA Constructs, Neuronal Culture, and Biotinylation Assays.

Cultures and biotinylation of cortical neurons were performed as described previously (25). Antibodies to AP2 and GABA_ARs and plasmids to the GABA_AR subunit ICDs fused to GST, and the subunits of AP2 have been described previously (10, 13).

Detection of AP-2 μ-Chain Binding to GABA_AR γ2-Subunit YECL Peptides by Affinity Pulldown and SPR Assays.

SPR, affinity purification from rat brain lysate, [³⁵S]-labeled, or purified proteins with peptides linked to CH-Sepharose beads were performed essentially as described in a number of previous studies (see SI Materials and Methods) (10, 13, 31).

Crystallography of the μ2–AP2 GABA_AR YECL-Pep Interaction.

X-ray data for μ2 YECL-pep crystals was collected at PSF beamline BL2 of Freie Universitat Berlin at BESSY/Berlin and processed by using HKL2000 (32) and scalepack. The phase problem was solved by molecular replacement with CCP4 program molrep (33) by using Mu2 Adaptin Subunit (PDB ID code 1BW8) without water and ligand atoms as model (23). After rigid body refinement, the R value was 36.5% (R_free = 39.5%) for data between 40- and 3.0-Å resolution. Subsequent cycles of isotropic B value and positional refinement to 2.51-Å resolution were performed by using Refmac5 (34). The peptide chain and the missing residues were built manually by using the model-building program, Coot (for additional details, see SI Materials and Methods) (35).

Whole-Cell Recordings.

Whole-cell recordings of mIPSCs from striatal neurons in acute slices used standard voltage-clamp techniques in the presence of 20 μM CNQX and 40 μM APV to block AMPA and NMDA receptors, respectively (see also SI Materials and Methods) (10, 25, 30).