Report

Diffusion Dynamics of Glycine Receptors Revealed by Single-Quantum Dot Tracking

See allHide authors and affiliations

Science  17 Oct 2003:
Vol. 302, Issue 5644, pp. 442-445
DOI: 10.1126/science.1088525

Abstract

Semiconductor quantum dots (QDs) are nanometer-sized fluorescent probes suitable for advanced biological imaging. We used QDs to track individual glycine receptors (GlyRs) and analyze their lateral dynamics in the neuronal membrane of living cells for periods ranging from milliseconds to minutes. We characterized multiple diffusion domains in relation to the synaptic, perisynaptic, or extrasynaptic GlyR localization. The entry of GlyRs into the synapse by diffusion was observed and further confirmed by electron microscopy imaging of QD-tagged receptors.

In living cells, the ability to selectively detect one molecule (or a small number of molecules) is a powerful way to understand the dynamics of cellular organization (1). So far, access to single-molecule properties in living cells has been restricted by either the size of the probe (40-nm gold nanoparticles or 500-nm latex spheres) (2) or the photobleaching of the small (1- to 4-nm) fluorescent labels (3, 4). QDs, which are intermediary in size (∼5 to 10 nm), are substantially more photostable than conventional fluorophores (5, 6), and have been vaunted as promising fluorescent probes (6, 7). In vitro and in vivo imaging with QDs has recently been demonstrated, but none of these measurements has aimed at specific cellular actors (8, 9).

GlyR is the main inhibitory neurotransmitter receptor in the adult spinal cord (10). At inhibitory synapses, GlyR clusters are stabilized by the scaffolding protein gephyrin (11, 12). The issue of lateral mobility of receptors for neurotransmitters has become central to understanding the development and plasticity of synapses (13). The membrane dynamics of GlyRs has been studied previously in transfected neurons using latex beads (14). GlyRs diffuse rapidly in the neuronal plasma membrane and transient interaction with gephyrin decreases their diffusion. Comparable results were obtained for the metabotropic- and AMPA-type glutamate receptors and their corresponding scaffolding molecules (15, 16). These measurements, however, preclude analysis of receptor dynamics in the synaptic cleft because of the use of 500-nm beads. We aimed to develop a new approach that could both access the synapse and be tracked for long periods of time.

The specific detection of endogeneous GlyR α1 subunits at the surface of spinal cultured neurons was achieved by the use of a primary antibody (mAb2b), biotinilated anti-mouse Fab fragments, and streptavidin-coated QDs (Fig. 1) (17). QD-GlyR formed numerous clusters around the soma and dendrites (Fig. 1A), similar to observations from previous immunocytochemical studies using conventional fluorophores (18). GlyRs were detected within synaptic and extrasynaptic domains (Fig. 1, B and C).

Fig. 1.

QDs as a marker for GlyR localization in neurons. (A) QD-GlyRs (red) detected over the somatodendritic compartment identified by microtubule-associated protein-2 (green). Arrows mark clusters of QD-GlyRs located on dendrites. (B and C) Relation of QD-GlyRs (red) with inhibitory synaptic boutons labeled for vesicular inhibitory amino acid transporter (green). QDs are either in front (arrows) or adjacent to (arrowheads) inhibitory synaptic boutons. The boxed region in (B) is shown as enlarged single-channel images in (C1) to (C3). Images are projections of confocal sections. Scale bars, 10 μm.

QDs were then used to study the lateral movement of individual GlyRs in living neurons. Single QDs were identified by their blinking property, i.e., the random intermittency of their fluorescence emission (5, 19). The results of these experiments were compared with those that used Cy3-coupled antibodies. Trajectories of single QD-GlyRs in the membrane could be visualized easily for at least 20 min, whereas the duration was ∼5 s for Cy3. The spots were detected with a signal-to-noise ratio of about 50 (integration time 75 ms), almost an order of magnitude higher than the signal obtained with fluorophores. Thus, the lateral resolution reached 5 to 10 nm, well below the 40 nm achieved with Cy3 dyes (20).

First, we used single-QD tracking (SDQT) to study the rapid lateral dynamics of GlyRs. Continuous sequences of 75-ms images were acquired for durations of ∼60 s. Individual QD-GlyRs diffusing in the neuronal membrane were either detected in extrasynaptic regions or associated with boutons identified with the amphiphilic FM4-64 dye. SQDT enabled the observation of multiple exchanges between extra-synaptic and synaptic domains, in which a GlyR alternated between free and confined diffusion states, respectively (Fig. 2A and movie S1). A GlyR, initially located at a synapse, started to diffuse rapidly (Fig. 2, A1 to A5) and, after about 30 s, stabilized close to another synaptic site (Fig. 2, A6 to A8), 4 to 5 μm away from the starting point. To quantify this observation, the instantaneous diffusion coefficients (D) were determined along the trajectory (Fig. 2B). For the 0- to 30-s period, D was ∼0.1 μm2/s, and the mean-square displacement (MSD) function varied linearly (Fig. 2C), as is expected for free Brownian diffusion (2). In the later part of the trajectory (30 to 63 s), D decreased to ∼0.02 μm2/s, and the MSD exhibited a negative curvature, characteristic of a space-confined movement (Fig. 2D).

Fig. 2.

Example of QD-GlyR motion over the neuritic surface. (A) Images extracted from a sequence of 850 frames (acquisition time: 75 ms). (A1) to (A8) correspond to frames 6, 118, 150, 267, 333, 515, 629, and 850, respectively. QD fluorescence spots (green) and FM4-64–labeled synaptic boutons (red). One QD (arrow), first located at bouton b1, diffuses in the extrasynaptic membrane [(A2) to (A5)] and associates with bouton b2 [(A6) to (A8)]. Scale bar, 2 μm. (B) Diffusion coefficient versus time. The upper line (red) denotes the frames on which the QD is fluorescent [arrowhead in (A1)] with interruptions corresponding to blinking processes. (C) MSD versus time, calculated for a continuous sequence of images between frames 54 and 161, which show the extrasynaptic motion. (D) MSD versus time, calculated for a continuous sequence of images between frames 503 and 597, when the QD is located at the periphery of bouton b2. Another QD [arrowhead in (A1)] remains associated with synaptic bouton b3 and blinks at image (A7). Error bars show mean ± SD.

The high photostability of QDs also allowed for the tracking of individual GlyRs in the same neuritic region for long durations. To avoid toxic continuous illumination of the cells, data were acquired in a time-lapse recording of one 75-ms image per second for 20 min (movie S2), a duration inaccessible when using fluorophores or even beads, which tend to stick to the cell membrane after a couple of minutes. An extrasynaptic receptor diffused freely and covered a large surface of the membrane (Fig. 3A). In contrast, some GlyRs were stable at synapses, whereas others moved but in a confined region around the bouton. These patterns, observed repetitively, led us to classify receptors as synaptic for FM4-64 overlapping spots, perisynaptic for spots with centers that were localized within two pixels of the border of the FM4-64 spot, or extrasynaptic for spots farther away from the synapse (Fig. 3B). At any given time, about 80% of the GlyRs were perisynaptic or synaptic (Fig. 3D), a proportion consistent with the results obtained with fixed neurons (18). We focused here on the properties of these receptors and plotted their location every 5 min for 40 min (Fig. 3C). Several observations deserve to be highlighted: (i) Some receptors remained synaptic or switched between perisynaptic and synaptic localization. (ii) GlyR moving from an extrasynaptic to a synaptic state and vice versa always transited by a perisynaptic state lasting for up to a few minutes. (iii) No receptor remained perisynaptic over the entire recording. The fractions of receptors in a given state did not vary much over time (Fig. 3D). Thus, the neuronal somatodendritic membrane is organized in three domains (extrasynaptic, perisynaptic, and synaptic) with distinct diffusion properties. The existence of a perisynaptic domain may be explained by the adhesion molecules present at the periphery of synapses [references in (13)] and/or by the gephyrin scaffolding molecules that overextend slightly past the limit of the synaptic complex (21).

Fig. 3.

QD-GlyRs diffusion during long recordings. (A) Time-lapse recording taken at 1Hz for 20 min of QD-GlyR trajectories (green) overlaid with FM4-64 staining (red) and bright-field image. For each pixel, the green signal corresponds to the maximum of the QD fluorescence signal over the entire stack of images. Extrasynaptic QD-GlyRs (*) explored large surfaces of the membrane, and synaptic QD-GlyRs were stable (**) or mobile (***) in a confined domain around the synapse. Scale bar, 5 μm. (B) Examples of QD-GlyRs (green) classified as synaptic (s) for FM4-64 (red) overlapping spots, perisynaptic (p) for spots within two pixels (430 nm) of the FM4-64 spot, and extrasynaptic (e) for spots farther away. Scale bar, 1 μm. (C) Localization of 17 individual QDs every 5 min for 40 min. (D) Proportions of QD-GlyRs from 0 to 10 min (1), 15 to 25 min (2), and 30 to 40 min (3). (E) Cumulative probability of diffusion coefficient of synaptic (red), perisynaptic (green), and extrasynaptic (blue) QD-GlyRs. (F) Proportions of rapid GlyRs (D > 0.1 μm2/s) measured with QDs (1), mAb2b-Cy3 (2) and Fab-Cy3 (3).

The diffusion coefficients of QD-GlyRs were determined as a function of their membrane location in about 230 trajectories (Fig. 3E). For extrasynaptic receptors, the mean D ± SEM was 0.10 ± 0.02 μm2/s (n = 83), which was about four times as large as those measured with the beads (0.029 ± 0.005 μm2/s) (14), indicating that the use of the beads significantly slowed down the receptor motion. For perisynaptic and synaptic receptors, the average diffusion coefficients were lower, at 0.023 ± 0.005 μm2/s (n = 70) and 0.015 ± 0.004 μm2/s (n = 82), respectively (22, 23). However, the mean coefficients did not reflect the wide range of movements that were observed. Therefore, we compared each receptor D to D1, where D1 = 0.01 μm2/s. This value was selected because ∼80% of the extrasynaptic movements that we measured had faster diffusion. Receptors with D > Dl were classified as rapid; those with D < D1 were classified as slow. The mean diffusion coefficients in the perisynaptic and synaptic regions were, respectively, 0.053 ± 0.009 μm2/s (n = 29) and 0.073 ± 0.013 μm2/s (n = 16) for rapid receptors and 0.0020 ± 0.0004 μm2/s (n = 41) and 0.0010 ± 0.0002 μm2/s (n = 66) for slow receptors. However, the relative fraction of rapid GlyR was significantly higher in the perisynaptic domains (41.4%) than it was in the synaptic domains (19.5%) (Fig. 3F). At synapses, rapid GlyRs are likely to be those that do not interact with gephyrin (14, 24).

Estimates of diffusion coefficient values may be altered by the size of QDs (10 to 15 nm for particles conjugated to streptavidin) and/or by their coupling to a divalent antibody that may crosslink receptors. As a control, we performed experiments with a Cy3-labeled Fab fragment of the primary antibody, a smaller (∼3 nm) and monovalent molecule. The use of Fab fragments did not significantly modify the results (25). Indeed, the antigenic determinant for mAb2b is localized in the vestibule of the GlyR channel (26, 27). Given the size of this type of receptor channel (28), the distance between the epitopes of adjacent receptors would be at least 10 nm, which does not allow for cross-linking by divalent antibodies. Our experiments could not exclude an influence of the steric hindrance of QDs on the diffusion of tagged receptors in the synaptic cleft. However, we were able to record comparable proportions of rapidly diffusing receptors with Cy3 and QDs.

The precise localization of diffusing GlyRs in the neuronal membrane was definitively established by electron microscopy (EM) images of silver-intensified QDs (Fig. 4). These images were obtained with the same labeling protocol as was used for SQDT. Extrasynaptic receptors were found over dendritic and somatic profiles (Fig. 4A). EM also provided the most direct evidence that QD-GlyRs could access the core of the synapse. GlyRs that localized on the edge of the bouton (Fig. 4B) could be easily distinguished from those inside the cleft (Fig. 4C). Perisynaptic receptors as previously defined do not necessarily equate with the ones localized on the sides of synapses in EM images. QDs were never detected intracellularly, indicating that QD-GlyRs were not internalized during the course of the experiments, in agreement with measurements of GlyR half-life in the membrane (29).

Fig. 4.

Transmission EM detection of QD-GlyRs. QD-GlyRs detected on the dendritic surface and associated with extrasynaptic membranes (arrows) (A), at the periphery of synapses (B), and within the synaptic cleft (C). d, dendrites; b, synaptic boutons. The edges of the synaptic clefts are outlined in (B) and (C). Scale bars, 500 nm.

The properties of QDs make it possible to record the mobility of individual molecules at the neuronal surface, even in confined cellular compartments. The ability to acquire both fluorescence and EM images with the same probes and labeling procedures provides access to two seemingly irreconcilable types of information: temporal dynamics and high-resolution cellular localization. QDs offer a favorable compromise between small fluorophores and large beads for single-molecule experiments in living cells and will be invaluable tools for ultrasensitive studies of the dynamics of cellular processes.

Supporting Online Material

www.sciencemag.org/cgi/content/full/302/5644/442/DC1

Materials and Methods

References

Movies S1 and S2

References and Notes

View Abstract

Navigate This Article