Ezrin mediates tethering of the gamma-aminobutyric acid transporter GAT1 to actin filaments via a C-terminal PDZ-interacting domain

Abstract

A high density of neurotransmitter transporters on axons and presynaptic boutons is required for the efficient clearance of neurotransmitters from the synapse. Therefore, regulators of transporter trafficking (insertion, retrieval, and confinement) can play an important role in maintaining the transporter density necessary for effective function. We determined the interactions that confine GAT1 at the membrane by investigating the lateral mobility of GAT1-yellow fluorescent protein-8 (YFP8) expressed in neuroblastoma 2a cells. Through fluorescence recovery after photobleaching, we found that a significant fraction ( approximately 50%) of membrane-localized GAT1 is immobile on the time scale investigated ( approximately 150 s). The mobility of the transporter can be increased by depolymerizing actin or by interrupting the GAT1 postsynaptic density 95/Discs large/zona occludens 1 (PDZ)-interacting domain. Microtubule depolymerization, in contrast, does not affect GAT1 membrane mobility. We also identified ezrin as a major GAT1 adaptor to actin. Förster resonance energy transfer suggests that GAT1-YFP8 and cyan fluorescent (CFP) tagged ezrin (ezrin-CFP) exist within a complex that has a Förster resonance energy transfer efficiency of 19% +/- 2%. This interaction can be diminished by disrupting the actin cytoskeleton. In addition, the disruption of actin results in a >3-fold increase in gamma-aminobutyric acid uptake, apparently via a mechanism distinct from the PDZ-interacting protein. Our data reveal that actin confines GAT1 to the plasma membrane via ezrin, and this interaction is mediated through the PDZ-interacting domain of GAT1.

Figure 1

Whole-footprint photobleach reveals lateral mobility. (A) Confocal images of GAT1-YFP8 localized at the cell footprint show a prephotobleached and postphotobleached region of interest representing >90% of the footprint surface area. Scale bar, 10 μm. (B) The kymograph is obtained by measuring the intensity along the red line, which is a section of the photobleached region. The line profile is plotted over time. The intensity key shows that photobleached regions are represented by “cooler” colors, and increased fluorescence is represented by “hotter” colors. (C) Diffusion model is simulated using the Axelrod-Sprague pure-diffusion model and data from GAT1-YFP8 footprint FRAP recovery. The GAT1-YFP8 recovery curve is not well-fitted by the Axelrod-Sprague pure-diffusion model. The GAT1-YFP8 footprint FRAP recovery curve does fit a double-exponential decay equation, $F\left(t\right)=F_{bo}-a{e}^{-bt}-c{e}^{-dt}$, where a, b, c, and d = 3.3, 0.30, 0.45, and 0.033, respectively.

Figure 2

Comparing GAT1 mobility with cytosolic and membrane inserted proteins. Footprint FRAP was performed on GAT1-YFP8 (not shown), (A) YFP, and (B) YFP-syntaxin. (C) Recovery curves show that YFP, a soluble cytosolic protein, has a faster recovery than either of the membrane proteins. The initial YFP recovery was faster than the resolution of detection. The membrane proteins GAT1-YFP8 and syntaxin-YFP recovered at similar rates.

Figure 3

Photobleached regions. FRAP prebleach and postbleach images of N2a cells expressing GAT1-YFP8. (A) Photobleaching is confined to cell footprint within red circle (8 μm²). Scale bar, 3 μm. (B) Photobleaching is confined to cell perimeter, within red rectangle (13 μm²). Scale bar, 4 μm. (C) GAT1-YFP8 footprint photobleach gave a recovery t_1/2 = 10 s, and a mobile fraction of 60%. GAT1-YFP8 perimeter photobleach gave a recovery t_1/2 = 20 s, and a mobile fraction of 50%.

Figure 4

GAT1 associates with actin, but does not associate with microtubules. (A, B) Microtubules are visible after treatment with TubulinTracker 488. Right-hand panels display nuclear labeling with Hoechst 33342. (B) Microtubules are disrupted by treatment with 10 μM nocodazole. Scale bars, 10 μm. (C, D) Traces display GAT1-YFP8 fluorescence recovery in cells with and without intact microtubules. Cells are photobleached at the footprint (C) and at cell perimeter (D). (E, F) Actin is revealed after N2a cells are treated with rhodamine-conjugated phalloidin. (F) Actin filaments are disrupted by treatment with latrunculin B. Scale bars, 10 μm. Traces display GAT1-YFP8 fluorescence recovery in cells with and without an intact actin network. Fluorescence is photobleached at the cell footprint (G) and at the cell perimeter (H). (I) GAT1-YFP8 mobile fraction. Disrupting actin with 5 μM latrunculin B significantly increases the amount of freely diffusing GAT1 on the plasma membrane (p < 0.001, t-test). The addition of 1 μg/mL cytochalasin D, another actin depolymerizer, also significantly increases the mobile fraction when probed with footprint photobleach (p < 0.001, t-test). Depolymerizing microtubules with 10 μM nocodazole does not affect the mobile fraction of GAT1-YFP8. (J) GAT1-YFP8 time constant. Disrupting actin with cytochalasin D or latrunculin B increases the time for recovery with perimeter photobleach (p < 0.05, Mann-Whitney test). Cytochalasin D treatment also significantly increases the time constant for recovery with footprint photobleach (p < 0.05, Mann-Whitney test). Microtubule disruption does not significantly affect the time constant. GAT1-YFP8 footprint, n = 18; GAT1-YFP8 + nocodazole footprint, n = 8; GAT1-YFP8 + latrunculin B footprint, n = 12, GAT1-YFP8 + cytochalasin D footprint, n = 10; GAT1-YFP8 perimeter, n = 12; GAT1-YFP8 + nocodazole perimeter, n = 9; GAT1-YFP8 + latrunculin B perimeter, n = 12; GAT1-YFP8 + cytochalasin D perimeter, n = 10.

Figure 5

GAT1-YFP8 and GAT1₀-GFP schematics, FRAP, and GABA uptake. (A) Schematics of GAT1-YFP8 and GAT1₀-GFP. GAT1₀-GFP is mostly found in vesicles near the plasma membrane (8). This faulty trafficking arises because the addition of linker and GFP moiety interrupts the PDZ-binding motif. GAT1-YFP8 traffics properly due to the addition of eight amino acids after the fluorescent protein, the final three amino acids being a consensus PDZ-binding motif, AYI-CO₂⁻. (B) GAT1₀-GFP has a higher mobile fraction than GAT1-YFP8 (p < 0.05, t-test). Disrupting actin does not significantly affect the mobile fraction of GAT1₀-GFP, compared with GAT1₀-GFP mobility in cells with intact actin. GAT1₀-GFP footprint, n = 9; GAT1₀-GFP + latrunculin B footprint, n = 9; GAT1₀-GFP perimeter, n = 12; GAT1₀-GFP + latrunculin B perimeter, n = 10. (C) Disrupting actin filaments through 1-h treatment with 5 μM latrunculin B significantly increases GABA uptake by GAT1-YFP8 and GAT1₀-GFP, relative to nontreated cells (p < 0.001, Mann-Whitney test).

Figure 6

Determination of expression of ezrin and Pals1 in N2a cells by qRT-PCR, and an interaction between ezrin and GAT1-YFP8 by FRET. (A) mRNA levels of β-actin, γ-actin, ezrin, and Pals1, normalized to β-actin expression. One-step qRT-PCR shows that ezrin is expressed in N2a cells at levels similar to those of PDZ protein Pals1. (B–D) FRET results. Prebleaching and postbleaching images of respective CFP and YFP fused proteins. Images are processed using Image J software with the enhance-image plugin (Rasband, W.S.; US National Institutes of Health, Bethesda, MD). Scale bar, 5 μm. (E) GAT1-YFP8 and ezrin-CFP interact, as represented by the 25% ± 3% increase in ezrin-CFP fluorescence that accompanies photodestruction of GAT1-YFP8. The disruption of actin through the addition of 5 μm latrunculin B significantly decreased FRET between ezrin-CFP and GAT1-YFP8. As a positive control, FRET between YFP-ezrin and ezrin-CFP was performed, resulting in a 26% ± 3% increase in ezrin-CFP fluorescence that accompanied the photodestruction of YFP-ezrin. (F) FRET efficiency for ezrin-CFP/GAT1-YFP8 is 19% ± 2%, for ezrin-CFP/GAT1-YFP8 + latrunculin B it is 7% ± 5%, and for ezrin-CFP/YFP-ezrin it is 20% ± 2%. Values are represented as the mean ± SE of 23 replicates. Significance was determined by one-way analysis of variance with Tukey's post hoc test (p < 0.05). (G, H) As a negative control, experiments analogous to those in B were performed with GAT1₀-YFP. No detectable increase in ezrin-CFP fluorescence accompanied the photodestruction of GAT1₀-YFP. Scale bar, 5 μm.

Figure 7

Schematic of GAT1 interactions with Pals1, ezrin, and actin. (A) Schematic of GAT1-YFP8 interaction with actin via ezrin and Pals1. The additional PDZ domain after YFP moiety restores the interaction between GAT1 and the PDZ protein. (B) The addition of 5 μM latrunculin B disrupts actin, and reduces the interaction between ezrin and GAT1 either by (B1) pulling away the GAT1-PDZ interaction, or (B2) reducing the interaction between ezrin and the PDZ protein. Altogether, this increases the mobility of GAT1 on the membrane. (C) GAT1₀-GFP, a GAT1 with a disrupted PDZ domain, is more mobile on the membrane, indicating that this domain may also stabilize the GAT1-ezrin-actin interactions.