Subunit-specific role for the amino-terminal domain of AMPA receptors in synaptic targeting

Abstract

The amino-terminal domain (ATD) of AMPA receptors (AMPARs) accounts for approximately 50% of the protein, yet its functional role, if any, remains a mystery. We have discovered that the translocation of surface GluA1, but not GluA2, AMPAR subunits to the synapse requires the ATD. GluA1A2 heteromers in which the ATD of GluA1 is absent fail to translocate, establishing a critical role of the ATD of GluA1. Inserting GFP into the ATD interferes with the constitutive synaptic trafficking of GluA1, but not GluA2, mimicking the deletion of the ATD. Remarkably, long-term potentiation (LTP) can override the masking effect of the GFP tag. GluA1, but not GluA2, lacking the ATD fails to show LTP. These findings uncover a role for the ATD in subunit-specific synaptic trafficking of AMPARs, both constitutively and during plasticity. How LTP, induced postsynaptically, engages these extracellular trafficking motifs and what specific cleft proteins participate in the process remain to be elucidated.

Keywords: AMPA receptor trafficking; GluA1; LTP; amino-terminal domain.

Fig. 1.

ATD-tagged GluA1 has normal surface trafficking but impaired synaptic trafficking, unlike ATD-tagged GluA2. (A) Representative traces of synaptic rectification experiments in control (black) and WT GluA1-expressing cells (green), scaled and superimposed for comparison in the right-hand panels. Synaptic trafficking of GluA1 is independent of splice variants or basal activity. (B) Representative traces of synaptic rectification experiments in control (black) and GFP-tagged GluA1-expressing cells (green). ATD-tagged GFP GluA1(i) and GFP GluA1(o) constructs showed no inward rectification after 2 d of transfection. After 4–6 d, rectification of synaptic currents was observed. (C) Both GluA1 and GFP GluA1 showed surface inward rectification. Sample current traces from control (black) and transfected (green) outside-out patches are shown. (D) ATD-tagged AMPAR subunits showed different trafficking abilities: GFP GluA1 showed no inward rectification but GFP GluA2 did. Also, GFP A1 (ATD)-A2(Q) (CTD) chimeric receptor showed no inward rectification but GFP A2 (ATD)-A1 (CTD) did. n = 5–23 cells per condition. (Scale bars: 50 pA, 20 ms.) Error bars represent mean ± SEM; *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 2.

ATD-lacking GluA1 has normal surface trafficking but, unlike ATD-lacking GluA2, impaired synaptic trafficking. (A) Surface rectification experiments in CA1 pyramidal neurons overexpressing ΔATD GluA1 in organotypic slices. Sample current traces in control (black) and ΔATD GluA1-overexpressing (green) outside-out patches are shown. (B) Removal of the entire ATD of GluA1 (ΔATD GluA1) impaired synaptic trafficking, whereas removal of the ATD of GluA2(Q) (ΔATD GluA2(Q)) did not affect trafficking. n = 7–12 cells/condition. Error bars represent mean ± SEM; **P < 0.01 and ***P < 0.001.

Fig. S1.

ATD-lacking GluA1 has normal surface trafficking in HEK cells. (A) Sample traces of glutamate-evoked currents without (Upper) and with (Lower) cyclothiazide (CTZ) recorded in GluA1- (Left) and ΔATD GluA1- (Right) expressing HEK cells at a holding potential of −60 mV. Glutamate (1 mM) application is represented by black bars and additional CTZ (100 μM) is represented by gray bars. (B) Averaged amplitude of evoked currents in HEK cells expressing different constructs in response to glutamate, without or with CTZ. n = 7 cells per condition. (Scale bars: 1 nA, 2 s.) Error bars represent mean ± SEM; **P < 0.01.

Fig. S2.

Coexpression of CA CaMKII promotes synaptic trafficking of GFP GluA1, but not ΔATD GluA1 without altering surface receptor levels. (A₁) Timeline of the experiment. (A₂) GFP coexpression does not change synaptic trafficking of overexpressed GFP GluA1 or ΔATD GluA1. In contrast, CA CaMKII coexpression promotes synaptic trafficking (thus increased rectification) of GFP GluA1, but not ΔATD GluA1. n = 9–28 cells per condition. Error bars represent mean + SEM. (B–D) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and cells expressing CA CaMKII (B, n = 19 pairs), together with GFP GluA1 (C, n = 16 pairs) or ΔATD GluA1 (D, n = 10 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green) neurons. (Scale bars: 50 pA, 20 ms.) The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. In all cases, CA CaMKII expression results in an approximately twofold increase in AMPA EPSC size. (E–G) Scatterplots showing amplitudes of whole-cell currents evoked by fast application of 1 mM glutamate for single pairs (open circles) of control and cells expressing CA CaMKII (E, n = 7 pairs), GFP GluA1 (F, n = 10 pairs), or CA CaMKII together with GFP GluA1 (G, n = 13 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green) neurons. (Scale bars: 100 pA, 200 ms.) The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM glutamate-evoked whole-cell currents data. *P < 0.05 and **P < 0.01.

Fig. 3.

GluA1, but not GFP GluA1 nor ΔATD GluA1, rescues synaptic AMPAR transmission in AMPAR-null cells in slice culture. (A₁) Scheme of the inducible AMPAR replacement strategy. (A₂) Timeline of the experiment. (B–E) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and GluA1-replaced cells without DOX (B), and +DOX for 4 d (C), GFP GluA1 +DOX (D), and ΔATD GluA1 +DOX (E). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. (F) Summary of the logarithms of the ratios between transfected and control cells for every pair analyzed in each experiment. n = 15–18 pairs. (Scale bars: 50 pA, 20 ms.) *P < 0.05, **P < 0.01, and ***P < 0.001 vs. control condition. ^##P < 0.01 and ^###P < 0.001 vs. WT GluA1 condition.

Fig. S3.

Replacement of endogenous AMPARs by recombinant GluA1 does not alter NMDAR currents but results in rectified AMPAR currents. (A–D) Scatterplots showing amplitudes of NMDAR EPSCs for single pairs (open circles) of control and GluA1-replaced cells without DOX (A, n = 12 pairs), and +DOX for 4 d (B, n = 9 pairs), GFP GluA1 +DOX (C, n = 7 pairs), and ΔATD GluA1 +DOX (D, n = 19 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM NMDAR EPSC data. (E) Cells in which endogenous AMPARs had been replaced by GluA1 show significant rectification in AMPAR synaptic currents. (Scale bars: 50 pA, 20 ms.) ***P < 0.001.

Fig. S4.

The ATD is not required for GluA2 synaptic trafficking. (A₁) Timeline of the experiment. (A₂–C) Scatterplots showing amplitudes of AMPAR eEPSCs for single pairs (open circles) of control and GluA2(Q)-replaced cells without DOX (A₂, n = 17 pairs) and +DOX for 4 d (B, n = 13 pairs) and ΔATD GluA2(Q) +DOX (C, n = 12 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. (D) Summary of the logarithms of the ratios between transfected and control cells for every pair analyzed in each experiment. The inducible GluA2(Q) replacement strategy is not leaky (virtually no AMPAR current in the absence of DOX). GluA2(Q) and ΔATD GluA2(Q) show a very similar degree of synaptic AMPAR currents rescued compared with WT GluA1. (Scale bars: 50 pA, 20 ms.) *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. S5.

Replacement of endogenous AMPARs by recombinant GluA2 does not alter NMDAR currents. (A–C) Scatterplots showing amplitudes of NMDAR EPSCs for single pairs (open circles) of control and GluA2(Q)-replaced cells without DOX (A, n = 14 pairs) and +DOX for 4 d (B, n = 7 pairs) and ΔATD GluA2(Q) +DOX (D, n = 16 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM NMDAR EPSC data. (Scale bars: 50 pA, 20 ms.)

Fig. S6.

The GluA1 ATD plays a dominant role in heteromeric AMPAR synaptic trafficking. (A₁) Timeline of the experiment. (A₂ and B) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and inducible ΔATD GluA1 + constitutive GluA2(R)-replaced cells without DOX (A₂, n = 13 pairs) and +DOX for 4 d (B, n = 16 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. Synaptic trafficking of heteromeric GluA2(R)/ΔATD GluA1 receptors is impaired, indicating a functional dominance of the GluA1 ATD over GluA2. (C and D) Normal surface delivery of heteromeric ΔATD GluA1/GluA2(R) AMPARs. Averaged amplitude (C) and rectification (D) of evoked currents in control and inducible ΔATD GluA1 + constitutive GluA2(R)-replaced cells with DOX in response to glutamate, with CTZ. Sample traces of control (Upper) and experimental outside-out patches (Lower) are shown. n = 4–5 cells per condition. (Scale bars: 50 pA, 20 ms.) ***P < 0.001.

Fig. S7.

GFP GluA1, when overexpressed in WT cells for short time, does not traffic constitutively to the synapse, but it does with LTP. (A₁) Timeline of the experiment. (A₂ and B) Scatterplots showing amplitudes of AMPAR (A₂, n = 7 pairs) and NMDA (B, n = 6 pairs) EPSCs for single pairs (open circles) of control and inducible GFP GluA1 overexpressing cells after i.p. DOX treatment for 6 d. Neither AMPAR nor NMDAR currents are affected. (C) Plots showing mean ± SEM. AMPAR EPSC amplitude of control (black) and inducible GFP GluA1-expressing CA1 pyramidal neurons (red) normalized to the mean AMPAR EPSC amplitude before LTP induction (arrow). n = 11 control and 9 transfected cells, P = 0.435, min 40. Sample AMPAR EPSC current traces from control (black) and electroporated (red) neurons before and after LTP are shown to the right of each graph. (D) Baseline synaptic AMPAR currents showed normal rectification index (open bars, black for control, red for transfected cells, n = 10 control and 5 transfected cells). After LTP, control cells synaptic AMPA currents remained nonrectified, whereas GFP GluA1-overexpressing cells showed significant rectification, in agreement with previous studies. n = 9 control and 8 transfected cells. (Scale bars: 50 pA, 20 ms.) **P < 0.01.

Fig. S8.

ΔATD GluA1 does not traffic to the synapse even after long-term expression. (A₁) Timeline of the experiment. (A₂) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and long-term ΔATD GluA1-overexpressing cells (n = 9). No change in EPSC is observed, suggesting no dominant-negative effect of ΔATD GluA1. (B) Rectification of synaptic AMPAR currents is not changed upon long-term ΔATD GluA1 overexpression in vivo, indicating that synaptic trafficking of ATD-truncated GluA1 is severely impaired. n = 9 control and 11 transfected cells. (C₁) Timeline of the experiment. (C₂) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and long-term ΔATD GluA1 constitutive replacement cells (n = 13 pairs). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green) neurons. The bar graph to the right of the scatterplot is normalized to control comparing mean + SEM AMPAR EPSC data. (D) Plot showing mean ± SEM. AMPAR EPSC amplitude of control (black) and Cre + ΔATD GluA1-expressing CA1 pyramidal neurons normalized to the mean AMPAR EPSC amplitude before LTP induction (arrow). Sample AMPAR EPSC current traces from control (black) and electroporated (green) neurons before and after LTP are shown to the right of the graph. LTP was impaired in ΔATD GluA1 constitutive replacement cells (P = 0.0145, min 40). n = 7 control and 10 transfected cells. (Scale bars: 50 pA, 20 ms.) ***P < 0.001.

Fig. 4.

WT and GFP-tagged GluA1 show normal LTP but ΔATD GluA1 does not. (A₁) Timeline of the experiment. (A₂, C, E, and G) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and GluA1-replaced cells by in utero electroporation without DOX (A₂) and with i.p. DOX treatment for 6 d (C), GFP GluA1 +DOX (E), and ΔATD GluA1 +DOX (G). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. (B, D, F, and H) Plots showing mean ± SEM. AMPAR EPSC amplitude of control (black) and Cre + inducible GluA1-expressing CA1 pyramidal neurons normalized to the mean AMPAR EPSC amplitude before LTP induction (arrow). Experimental cells were transfected with GluA1 −DOX (B, P = 0.04, min 40) or +DOX (D, P = 0.48, min 40), GFP GluA1 +DOX (F, P = 0.70, min 40), or ΔATD GluA1 +DOX (H, P = 0.01, min 40). Sample AMPAR EPSC current traces from control (black) and electroporated (green, −DOX and red, +DOX) neurons before and after LTP are shown to the right of each graph. n = 9–14 pairs in baseline experiments and 6–13 cells per condition in LTP experiments. (Scale bars: 50 pA, 20 ms.) **P < 0.01 and ***P < 0.001.

Fig. 5.

WT and ΔATD GluA2 show normal LTP. (A₁) Timeline of the experiment. (A₂, C, and E) Scatterplots showing amplitudes of AMPAR EPSCs for single pairs (open circles) of control and GluA2(Q)-replaced cells by in utero electroporation without DOX (A₂), and with DOX treatment for 6 d (C) and ΔATD GluA2(Q) +DOX (E). Filled circles represent mean ± SEM. Insets show sample current traces from control (black) and transfected (green, −DOX and red, +DOX) neurons. The bar graphs to the right of the scatterplots are normalized to control comparing mean + SEM AMPAR EPSC data. (B, D, and F) Plots showing mean ± SEM. AMPAR EPSC amplitude of control (black) and Cre + inducible GluA2(Q)-expressing CA1 pyramidal neurons normalized to the mean AMPAR EPSC amplitude before LTP induction (arrow). Experimental cells were transfected with GluA2(Q) −DOX (B, P = 0.02, min 40) or +DOX (D, P = 0.84, min 40) or ΔATD GluA2(Q) +DOX (F, P = 0.72, min 40). Sample AMPAR EPSC current traces from control (black) and electroporated (green, −DOX and red, +DOX) neurons before and after LTP are shown to the right of each graph. n = 6–14 pairs in baseline experiments and 4–11 cells per condition in LTP experiments. (Scale bars: 50 pA, 20 ms.) **P < 0.01.

Fig. S9.

Inducible AMPAR replacement results in fully rectified synaptic currents. (A–E) Inducible AMPAR replacement with recombinant GluA1, GFP GluA1, ΔATD GluA1, GluA2(Q), and ΔATD GluA2(Q) resulted in very significantly rectified synaptic AMPAR currents (n = 3–7 cells per condition). *P < 0.05, **P < 0.01, and ***P < 0.001.