Mapping the interaction sites between AMPA receptors and TARPs reveals a role for the receptor N-terminal domain in channel gating

Abstract

AMPA-type glutamate receptors (AMPARs) mediate fast neurotransmission at excitatory synapses. The extent and fidelity of postsynaptic depolarization triggered by AMPAR activation are shaped by AMPAR auxiliary subunits, including the transmembrane AMPAR regulatory proteins (TARPs). TARPs profoundly influence gating, an effect thought to be mediated by an interaction with the AMPAR ion channel and ligand binding domain (LBD). Here, we show that the distal N-terminal domain (NTD) contributes to TARP modulation. Alterations in the NTD-LBD linker result in TARP-dependent and TARP-selective changes in AMPAR gating. Using peptide arrays, we identify a TARP interaction region on the NTD and define the path of TARP contacts along the LBD surface. Moreover, we map key binding sites on the TARP itself and show that mutation of these residues mediates gating modulation. Our data reveal a TARP-dependent allosteric role for the AMPAR NTD and suggest that TARP binding triggers a drastic reorganization of the AMPAR complex.

Graphical abstract

Figure 1

The NTD-LBD Linker Influences TARP-Dependent Changes in AMPAR Gating (A) Structure of the extracellular region of a GluA2 subunit, showing the NTD (gray), linker (blue), and LBD (yellow) (adapted from PDB: 3KG2). Sequence alignment of the rat GluA1-4 NTD linkers, red residues have been mutated in Δ link with VTxxxLPSG deleted and the two asparagines (N385 and N392 bold, underlined) mutated to Asp and Gln, respectively (analogous to PDB 3KG2). Model of a complete, stretched NTD linker is shown beneath the alignment. (B) Representative normalized current responses evoked by 100 ms glutamate application (gray bars). Outside-out patches were pulled from cells transfected with GluA2iQ WT or Δ link in the absence or presence of γ-2 and the decay of the current (−60 mV) analyzed to determine the time constant of desensitization and the magnitude of the steady state component. (Inset) Pooled data (mean ± SEM) showing the difference in charge transfer (normalized to the peak) during the 100 ms glutamate application (^∗∗∗p < 0.001, Welch t test). (C) Pooled data showing the desensitization time constants for GluA2iQ WT or Δ link expressed alone (n = 14 and 10), or with TARPs γ-2 (n = 11 and 15), γ-3 (n = 14 and 10), γ-4 (n = 12 and 17), or γ-8 (n = 20 and 15). Currents were fitted with a two-exponential function. The weighted time constant (τ_w,des) is shown ± SEM. Two-way ANOVA indicated a significant main effect of TARP (F_{4, 111} = 82.28, p = 2.66 × 10⁻³²), no significant main effect of linker mutation (F_{1, 111} = 1.73, p = 0.19) and a significant interaction between linker and TARP effects (F_{4, 111} = 10.05, p = 5.61 × 10⁻⁷). Asterisks denote significance of difference between WT and Δ link for each TARP condition (^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001; Welch t test). (D) Pooled data showing the ratio of current at the end of the 100 ms glutamate application (steady-state [SS]) to the peak response. The data are plotted and analyzed as in (C), for GluA2iQ WT or Δ link expressed alone (n = 10 each), or with γ-2 (n = 11 and 8), γ-3 (n = 14 and 9), γ-4 (n = 12 and 17), or γ-8 (n = 10 and 15). There were significant main effects of TARP subtype (F_{4, 106} = 42.74, p = 2.64 × 10⁻²¹) and linker mutation (F_{1, 106} = 5.04, p = 0.027) and a significant interaction between linker and TARP effects (F_{4, 106} = 8.89, p = 3.16 × 10⁻⁶). Asterisks are as in (C).

Figure 2

The Δ Link Mutation Accelerates Recovery from Desensitization in the Presence of TARPs (A) Representative traces illustrating recovery from desensitization (averages of three trials in each case). A 100 ms pulse of 10 mM L-glutamate was followed, at increasing intervals, by a 10 ms test pulse, and the recovery in the amplitude of the test response was fitted by a monoexponential function (dashed lines). Currents were normalized to the first peak, and for clarity, only selected traces are shown. (B) Summary of the data presented in (A). Relative currents at individual time points are shown ± SEM (error bars masked by the symbols). The solid lines are monoexponential fits of the average values (giving time constants of 7.3, 21.8, and 75.6 ms for GluA2 Δ link + γ-2, GluA2 WT + γ-2, and GluA2 WT + γ-8, respectively). (C) Pooled data for the time constant of recovery from desensitization (τ_rec) for GluA2iQ WT and Δ link expressed alone (n = 11 and 8, respectively) or with TARPs γ-2 (n = 17 and 9), γ-3 (n = 6 and 5), γ-4 (n = 8 each), or γ-8 (n = 7 and 8) (shown ± SEM). Two-way ANOVA showed significant main effects of TARP subtype (F_{4, 77} = 62.61, p = 1.91 × 10⁻²³) and linker mutation (F_{1, 77} = 25.58, p = 4.14 × 10⁻⁶) and a significant interaction between linker and TARP effects (F_{4, 77} = 3.14, p = 0.019). Asterisks denote significance of difference between WT and Δ link for each TARP condition (^∗p < 0.05, ^∗∗∗p < 0.001; Welch t test). See also Figure S1.

Figure 3

Gating Changes Are Mediated by Specific Structural Features of the NTD Linker (A) Sequence of the GluA2 NTD-LBD linker with the glycosylation sites and the amino acid quadruplets deleted in these experiments highlighted in red and blue. (B) Pooled data (mean ± SEM) showing the effects on desensitization (τ_w,des) of NTD-LBD linker mutations in GluA2i in presence of γ-2. LPSG denotes a construct with these four amino acids deleted; LPSG-D combines this with the N385D mutation, and LPSG-D-Q additionally includes N392Q. Following one-way ANOVA (F_{7, 25.7} = 14.69, p = 1.52 × 10⁻⁷), pairwise comparisons showed that τ_w,des was slower for LPSG (n = 11), LPSG-D (n = 6), LPSG-D-Q (n = 16), and Δ link (n = 7) compared with WT (n = 11) (^∗∗p < 0.01, ^∗∗∗p < 0.001) and was significantly slower for LPSG-D-Q compared with LPSG (#p < 0.05) (Welch t tests). There was no significant effect of the glycosylation mutations N385D (n = 8) or N392Q (n = 10) and no effect of the alternative deletion mutant, SGLE (n = 8). (C) Pooled data for the steady state-to-peak ratio (SS/peak; presented and analyzed as in B). Following one-way ANOVA (F_{7, 26.5} = 17.18, p = 2.34 × 10⁻⁸), pairwise comparisons showed that the SS/peak ratio was greater for N385D (n = 8), N392Q (n = 10), LPSG (n = 11), LPSG-D (n = 7), LPSG-D-Q (n = 16), and Δ link (n = 7) compared with WT (n = 9) (^∗∗p < 0.01, ^∗∗∗p < 0.001) and significantly greater for LPSG-D-Q compared to LPSG (#p < 0.05) (Welch t tests). Again, no effect of the alternative deletion mutant, SGLE (n = 8). (D) Pooled data (± SEM) comparing the effects of the LPSG-D mutation in GluA2 and in a chimeric construct where the NTD of GluA2 was replaced by that of GluA3. (Left) τ_w,des was increased by LPSG-D in both GluA2 (n = 10 and 6) and GluA2_A3-NTD (n = 14 and 7) (^∗∗∗p < 0.001 and ^∗p < 0.05; Welch t tests). Two-way ANOVA showed a significant interaction between NTD and linker (F_{1, 33} = 11.62, p = 0.0017). Similar results were seen for the SS/peak ratio (right); in this case, the ratio was increased by LPSG-D in GluA2 (^∗∗∗p < 0.001) but not in GluA2_A3-NTD. Again, two-way ANOVA showed a significant interaction between NTD and linker (F_{1, 33} = 13.10, p = 9.75 × 10⁻⁴), confirming that the effect of the LPSG-D linker mutation was NTD-type specific. See also Figure S2.

Figure 4

Mapping the TARP γ-2 Contact Region on GluA2 (A) coIP of GluA2 variants with TARP γ-2. The blot was probed with polyclonal GluA2 antibody (top panel) and anti γ-2 (bottom panel). Both WT and ΔNTD protein migrated as monomer (M) and dimer (D), denoted by arrowheads. Note that while inputs were comparable, amounts of IPed GluA2 varied between conditions. (B) Schematic of the peptide array layout (right). Each peptide is spotted onto a nitrocellulose membrane (C). Peptide coverage of the rat GluA2 sequence is outlined in the left panel in color code as indicated. The four GluA2 regions—the NTD lower lobe, the NTD-LBD linkers, the LBD, and the TM segments of the channel—are highlighted. GluA2 peptide numbers covering each domain are indicated in brackets. See Table S1 for peptide sequences. (C) Regions of GluA2 binding to TARP γ-2. (Upper panel) Nonspecific signal, resulting from anti-γ-2 antibody binding to GluA2 in the absence of the γ-2 probe (“AB control”). AMPAR domains are highlighted in boxes and match the color scheme in (B). Peptide numbers are indicated on the side. The membrane was exposed to an X-ray film for 2 min; the arrow denotes nonspecific signals. (Lower panel) The same membrane was probed with full-length TARP γ-2 and detected with anti-γ-2 AB followed by a HRP-labeled secondary AB (2 min exposure). Individual AMPAR secondary structure elements, corresponding to NTD and LBD helices, are highlighted in stippled boxes on the blot (compared with D). The bottom panel shows a longer exposure for the LBD-A1 region (yellow) and the NTD-LBD linker (blue). See Figure S4A for a longer exposure of the blot. (D) TARP binding sites deduced from the peptide array in (C) are mapped onto the extracellular region of GluA2 (PDB: 3KG2). NTD interaction sites are denoted in deep red (strong interaction) and light pink (weaker interaction; see graded bar below), with alpha helices contacted by γ-2 denoted by (D), (F), and (H). LBD interaction sites are highlighted in brown (strong interaction) and yellow (weaker interaction). The three core contact regions, A–C, are denoted. Region A spans the glutamate binding cleft (interaction sites A₁ and A₂); region B encompasses the LBD-TM linker 1, and region C corresponds to the flip-flop cassette (denoted with a stippled ellipsoid). See also Figures S3, S4, and S6 and Table S1.

Figure 5

Mapping TARP γ-2 and TARP γ-8 Residues Involved in AMPAR Interaction (A) Alignment of rat type 1a (γ-2, γ-3) and type 1b (γ-4, γ-8) TARP extracellular loops, Ex1 and Ex2. Conserved residues are shaded brown, and residues highly conserved throughout the Cacng family (γ-1 to γ-8) are boxed in gray. The four cysteines are highlighted in yellow. Curly brackets above the γ-2 (green) and below γ-8 alignment (blue) indicate regions in the center of Ex1 interacting with the AMPAR extracellular domains (NTD and LBD). (B) TARP array encompassing the Ex1 and Ex2 segments of γ-2 (green box) and γ-8 (blue box) probed with the NTD. (Upper panel) Nonspecific signal, resulting from direct anti-GluA2 antibody binding to the membrane (AB control). The Ex1 and Ex2 regions for both TARPs are denoted (dashed line). (Lower panel) the same membrane was exposed to the rat GluA2 NTD followed by probing with anti-GluA2 AB. The membrane exposure time is as indicated. See Table S2 for peptide sequences. (C) γ-2 and γ-8 arrays probed with GluA2 LBD and GluK2 LBD. The negative controls with anti-FLAG AB did not show any signal and thus are not shown. Membranes were then incubated with FLAG-tagged GluA2 LBD or GluK2 LBD and probed with anti-FLAG AB. (Upper panel) GluA2 LBD interaction with γ-2 (same peptides as in Figure 5B, green box). (Central panel) GluA2 LBD interaction with γ-8 (same peptides as in B, blue box). (Lower panel) the γ-2 membrane previously probed with GluA2 LBD (top panel) was regenerated and probed with the FLAG-tagged GluK2 LBD, which produced no clear binding. Regeneration of this membrane resulted in a clear binding pattern when reprobed with the GluA2 LBD that matched the one shown in (C, top). (D) Schematic representation of TARP structure with the GluA2 NTD and LBD interacting parts of the Ex1 and Ex2 loops highlighted in orange and the highly conserved GLWR motif indicated.

Figure 6

Effects of Ex1 Mutations in TARP γ-2 (A) Sequence of the γ-2 Ex1 region surrounding the highly conserved GLWRxC₆₇ motif. Boxed regions in red identify the amino acid triplets and quadruplet mutated in these experiments. (B) Pooled data (mean ± SEM) showing the effects of Ex1 mutations in γ-2 on desensitization (τ_w,des) of GluA2. Following one-way ANOVA (F_{4, 23.27} = 28.89, p = 1.01 × 10⁻⁸), pairwise comparisons showed that γ-2 WT (n = 12) and all three γ-2 mutants (KGL_74–76, KQID_78–81, and WRT_64–66; n = 12, 8, and 14, respectively) slowed τ_w,des compared with GluA2 alone (n = 10) (^∗∗∗p < 0.001) and that the effects of KGL_74–76 and WRT_64–66 were less than those of γ-2 WT (^#p < 0.05, ^###p < 0.001) (Welch t tests). To the right are representative currents evoked by 10 mM L-glutamate (−60 mV) in patches from cells expressing GluA2/γ-2 WT and GluA2/γ-2 WRT_64–66. Also shown are the individual τ_w,des values determined from double exponential fits (blue). (C) Pooled data showing the effects of mutations in γ-2 on the steady-state/peak ratio (SS/peak). Presentation, analysis, and n numbers as in B (F_4, 22.91 = 18.26, p = 7.27 × 10⁻⁷; ^∗∗p < 0.01 and ^∗∗∗p < 0.001 compared with GluA2 alone and ^###p < 0.001 compared with γ-2 WT). To the right are representative records from GluA2/γ-2 WT and GluA2/γ-2 WRT_64–66 illustrating the current remaining at the end of the 100 ms applications of L-glutamate (10 mM) and the calculated SS/peak ratio. (D) Pooled data (±SEM) showing the effects of mutations in γ-2 on GluA2 deactivation (τ_w,deact; 1 ms, 10 mM L-glutamate, −60 mV). Following one-way ANOVA (F_{4, 17.47} = 3.06, p = 0.044), pairwise comparisons showed that only for γ-2 WT (n = 8) was τ _w,deact slowed compared with GluA2 alone (n = 8) (^∗p < 0.05). None of the mutants (KGL_74–76, KQID_78–81, and WRT_64–66; n = 11, 8, and 7, respectively) differed from GluA2 alone or GluA2/γ-2 WT. To the right are representative currents in patches taken from cells expressing GluA2/γ-2 WT and GluA2/γ-2 WRT_64–66. Also shown are the individual τ_w,deact values determined from double exponential fits (blue). (E) Pooled data (± SEM) showing the effects of mutations in γ-2 on the KA/Glu ratio. Following one-way ANOVA (F_{4, 9.85} = 44.8, p = 2.72 × 10⁻⁶), pairwise comparisons showed that γ-2 WT (n = 5) and all three γ-2 mutants (KGL_74–76, KQID_78–81, and WRT_64–66; n = 7, 5, and 8, respectively) increased kainate efficacy (KA/Glu ratio) compared with that seen with GluA2 alone (n = 4) (^∗p < 0.05, ^∗∗∗p < 0.001) and that KQID_78–81 and WRT_64–66 had less effect than γ-2 WT (^#p < 0.05, ^###p < 0.001) (Welch t tests). To the right are representative currents evoked by L-glutamate and kainate (both 500 μM, in the presence of 100 μM cyclothiazide) in patches from cells expressing GluA2/γ-2 WT and GluA2/γ-2 WRT_64–66. Also shown are the KA/Glu ratios for these representative records. See also Figure S5.