Mapping the Interaction Sites between AMPA Receptors and TARPs Reveals a Role for the Receptor N-Terminal Domain in Channel Gating

Summary 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.

Note that inputs were comparable, while amounts of IPed TARP γ-2 varied between conditions.
There was no visible difference between GluA2i-wt and the Δlink mutant with regard to TARP γ-2 association. The lower panel show that comparable amounts of γ-2 were present in the two reactions. Figure S4. Mapping the TARP γ-2 contact region on GluA2, related to Figure 4.

A.
Regions of GluA2 and GluA3 AMPARs binding to TARP γ-2. Array membrane containing GluA2 and -A3 peptides was probed with anti-γ-2 antibody only (AB control, top panel), or incubated with purified γ-2 protein before ABprobing (lower panel) and exposed for 20 minutes. Note the absence of γ-2 signal across the GluA2 linker region (blue box). The NTD-binding region is boxed in red, the LBD region in grey and the TM segments in green. Peptide numbers are indicated on the sides. Lower panel shows the binding of TARP γ-2 to the GluA3 region flanking the NTD-LBD linker. Binding is observed for the GluA3 NTD and LBD but not for the linker peptides. Non-specific signals are indicated by the tilted white arrow.

B.
TARP γ-2 interaction sites on the GluA2 NTD. Contact sites are shown in deep-red (strong interaction) or light pink (weaker interaction). The orange footprint outlines the tetrameric interface that is formed between two NTD dimers in crystal structures. The bottom panel shows the GluA2 NTD dimer superposed on the fulllength structure (PDB: 3KG2), with the tetrameric interface (between the NTD dimers) coloured in orange. The stippled line denotes the NTD dimer interface. Upper lobe (UL), lower lobe (LL).
The relative strength of binding is indicated by the same colour gradient as in panel B.
Shown to the right are representative normalized I-V plots from patches expressing GluA2/ γ-2 wt and GluA2/ γ-2 WRT 64-66 . Fitted curves are 8 th order polynomials. Note that WRT 64-66 produced less change in RI than did γ-2 wt. This change in RI appeared to reflect changes in spermine permeation, as the effects of WRT 64-66 were evident in the outward limb of the I-V but not apparent at negative voltages, where the conductance-voltage plots were essentially indistinguishable (data not shown).
Binding sites on the NTD, LBD and LBD-TMD linkers are indicated in red. The position where L-glutamate binds is indicated with stars. The potential movement NTDs need to undergo to contact the TARPs is denoted with grey arrows.

B.
Model outlining the reorganization in the AMPAR extracellular region that may accompany TARP interaction. Left panel: Without TARP, the NTD and LBD are loosely connected. Any potential allosteric signal emanating from the NTD is not transmitted to the LBD (and the receptor), i.e. the NTD is functionally isolated (yellow circles). Right panel: The TARP 'bridges' NTD and LBD. This requires substantial receptor reconfiguration mediated by the NTD-LBD linkers. As a result the NTD and LBD are functionally connected (yellow ellipsoid).  Table S1. Sequences of peptides immobilized in the GluA2 array, related to Figure 4.

Supplemental Experimental Procedures Protein preparation
His-tagged γ-2 was produced in insect cells using a P1 baculovirus stock and a purification protocol provided by T. Nakagawa. The high-titer viral stocks were The GluA2i LBD (S1S2J R/flip construct) was sub-cloned with different tags in order to allow its detection in the peptide arrays. GluA2i LBD was first sub-cloned into the pGEX4T-2 vector (GE Healthcare), which contains an N-terminal GST-tag followed by a thrombin cleavage site. The plasmid was transformed into Escherichia coli Origami B (DE3) and grown at 37 °C to A 600 =0.9-1. Cultures were cooled to 18 °C and expression was induced by the addition of 0.4 mM IPTG. Cultures were grown at 18 °C for 20 hours. Cells were sonicated and the lysate was incubated with Glutathione Sepharose 4b beads (GE Healthcare). The protein was eluted with 10 mM reduced-Glutathione and further purified using a Superdex 200 10/300 column (GE Healthcare). Protein was flash-frozen and kept at -20 °C until used.
A Flag-tag (DYKDDDDK) was also introduced by PCR at the C-terminus of GluA2i LBD and the gene was subcloned into a modified pET22b(+) plasmid containing an The same procedure was used to add a Flag-tag to the GluK2 LBD S1S2 construct

Peptide arrays
To analyze the interaction between AMPAR and TARPs we obtained peptide arrays synthesized by SPOT synthesis on Whatman 50 cellulose membranes (PepSpots TM ; JPT Peptide Technologies GmbH). AMPAR and TARP arrays contained 15-mer overlapping peptides shifted by 4 residues. The AMPAR array contained 128 peptides covering the lower lobe of GluA2 NTD, the NTD-LBD linkers with and without glycosylated residues (GluA2 and -A3 subunits), the GluA2 LBD and the four transmembrane helices ( Figure 4B, Table S1). The TARP array contained 46 peptides covering the two extracellular loops (Ex1 and Ex2) of γ-2 and γ-8 ( Figure 5, Table   S2).
Membranes were probed according to manufacturer instructions. As these membranes cannot be regenerated reliably, control experiments with the antibodies (ABs) were performed prior to incubation with the test proteins. Controls were essential and extensive to select the most adequate blocking agents, antibodies and protein tags, as some ABs showed false positives or high background in the cellulose membranes. Anti-His and anti-GST ABs showed signals in the control experiments whereas Flag M2 AB (Sigma), anti-AMPAR 2 (extracellular) AB (Alomone) or anti-Stargazin AB (Millipore) controls showed few false positives (Figures 4C and 5B).
For the protein binding assay, membranes were rinsed with methanol for 5 minutes, Membranes were washed 3 times, developed using chemiluminescence (Amersham ECL solution) and images were captured electronically with a ChemiDoc TM MP Imaging System (Biorad) or on X-ray film. Films were scanned and edited with Photoshop CS4 using auto-tone editing followed by a change to gray-scale mode.

Kinetics of AMPAR-mediated responses
Desensitization of AMPARs was examined in response to 100 ms applications of 10 mM L-glutamate. The averaged currents were fitted using a double-exponential function to calculate the weighted time constant of desensitization (τ w,des ) according to: where was determined in a similar manner, by fitting the current decay following 1 ms applications of 10 mM L-glutamate. In some cases the desensitization or deactivation time course was best described by a single exponential. The steady state-to-peak ratio (SS/peak) was determined as the current at the end of the 100 ms pulse divided by the peak current.
Recovery from desensitization was assessed by a paired-pulse protocol where a 100 ms desensitizing pulse was followed by a 10 ms pulse in increasing intervals. The relative response to the second pulse (usually the average of three consecutive runs) was then plotted against time elapsed from the first pulse and the time course fitted with a single-exponential function to obtain the time constant of recovery (τ rec ).

Relative kainate efficacy
The effects of TARPs on the efficacy of the partial agonist kainate (KA) were determined by measuring changes in KA/Glu ratios. Each patch was exposed to Lglutamate (500 μM; 15 applications of 100 ms duration, -60 mV), then KA (500 μM), and then L-glutamate again. Both control and agonist solutions contained 100 μM cyclothiazide to block AMPAR desensitization. The amplitude of the steady state current response to KA was compared to average amplitude of the steady state current responses to L-glutamate obtained before and after KA application.

Non-stationary fluctuation analysis (NSFA)
To deduce channel properties from macroscopic responses, L-glutamate (10 mM) was applied to outside-out patches (100 ms duration, 1 Hz, V hold -60mV) and the ensemble variance of all successive pairs of current responses were calculated (Conti et al., 1980). The single channel current (i) and the total number of channels in the patch (N) were determined by plotting this ensemble variance against mean current (I) and fitting with a parabolic function (Sigworth, 1980): is the background variance. The weighted-mean single-channel conductance was determined from the single-channel current and the holding potential. No correction for liquid-junction potential was used.

Current-voltage (I-V) plots and the quantification of rectification
I-V plots were generated from the peak current response to 1 ms applications of 10 mM glutamate. The voltage was stepped from -100 mV to +60 mV in 10 mV increments. Mean current amplitudes at each voltage were normalized to the peak current at -100 mV and plotted against membrane potential. The relationships were fitted with 8 th or 9 th order polynomials. The rectification index (RI) was determined from the I-V relationship of each patch as the ratio of slope conductance at positive (+40 mV to +60 mV) and negative voltages (-40 mV to -60 mV). Thus, for a completely linear I-V relationship RI would equal 1.

Analysis and statistics
Recordings were analyzed using IGOR Pro (Wavemetrics Inc.) with NeuroMatic (J. Rothman, UCL; http://www.neuromatic.thinkrandom.com). Summary data are presented in the text as the mean ± SEM from n patches and in the figures as bar plots of the group mean, with error bars denoting SEM. Comparisons involving two data sets only were performed using a two-sided Welch two-sample t test. All analyses involving data from three or more groups were performed using one-or two-way analysis of variance (Welch heteroscedastic F test) followed by pairwise comparisons using two-sided Welch two-sample t tests (with Holm's sequential Bonferroni correction for multiple comparisons). Differences were considered significant at P <