The formation of acetylcholine receptor clusters visualized with quantum dots

Xenopus cell cultures and induction of AChR clustering

Myotomal muscle cells and spinal neurons were isolated from stage 20–22 Xenopus laevis embryos according to previously published methods [18]. Cells were plated on glass coverslips in Steinberg's solution, consisting of 60 mM NaCl, 0.67 mM KCl, 0.34 mM Ca(NO₃)₂, 0.83 mM MgSO₄ and 10 mM Hepes at pH 7.4, supplemented with 10% L-15 (Leibovitz) medium, 1% fetal bovine serum, 100 U/ml penicillin. The cultures were maintained at 23°C.

Polystyrene latex beads (10 μm diameter; Polysciences, Warrington, PA) were coated with recombinant heparan-binding growth associated molecule (HB-GAM) (kindly provided by Dr. Heikki Rauvala, University of Helsinki) and applied to muscle cells as described [19]. AChR clustering induced by the beads was examined within a few hours of bead addition. To prepare nerve-muscle cocultures, spinal neurons dissociated from neural tubes were seeded onto 3-day old muscle cells; nerve-muscle contacts were examined one day later.

Live imaging of AChR clustering

For AChR labeling, biotin-conjugated α-bungarotoxin (biotin-BTX) and Alexa 488-conjugated α-bungarotoxin (Alexa 488-BTX) (Invitrogen) were diluted into culture medium at concentrations of 0.5 nM and 25 nM respectively. Streptavidin-conjugated QD655 (Invitrogen) was diluted into culture medium at the final concentration of 2.5 nM. Cells were incubated with biotin-BTX and Alexa488-BTX (excitation wavelength 488 nm) for 5 min, washed three times over 30 min with culture medium, incubated with QD655 (maximum emission 655 nm) for 10 min, and then washed again extensively with culture medium before observation using a custom-built chamber. The 1:50 ratio of biotin-BTX and Alexa 488-BTX in the first labeling step resulted in very low density of AChRs tagged by the biotin-BTX-streptavidin QD complex. The labeling was carried at room temperature.

For labeling GM1 ganglioside, a process similar to that for AChR was used, except that 50 ng/ml biotin-conjugated cholera toxin B subunit (biotin-CTX) (Sigma-Aldrich) reconstituted in 5% BSA was applied instead of biotin-BTX.

Cells were imaged at room temperature with an Olympus IX70 inverted microscope equipped with 60× (N.A. 1.40) objective. QD655 and Alexa488-BTX were observed with fluorescence filters for rhodamine and FITC respectively. The filter combination for QD imaging was as follows: excitation filter with band-pass wavelength of 510–550 nm, emission filter with long-pass wavelength of 590 nm. Images were captured with a 500 ms exposure using a cooled digital CCD camera (Hamamatsu ORCA II-ER, C4742–98) interfaced with an electronic shutter (Sutter Instrument) under the control of Metamorph software (Universal Imaging). QDs were also tracked with 100 ms exposure time, the lower limit of our imaging system. The diffusion coefficients calculated from image stacks obtained with 500 ms and 100 ms exposure time were similar.

Receptor tracking and data analysis

Single-molecular tracking was performed using the "track point" function of Metamorph software. This feature allows single points within each image frame of a time-lapse series to be tracked and enables the measurements of paths, positions and velocities of these points. Our labeling strategy was aimed at achieving low QD density to avoid receptor crosslinking. Multiple QDs within a time-lapse series were individually analyzed by repeating the track-point function to ensure tracking accuracy. Data were then incorporated into a spreadsheet for further analysis. Because of the intrinsic blinking property of QDs, single QDs sometimes became invisible for short periods of time and reappeared later during time-lapse recordings. For short-term experiments (50 frames, 5–25 sec), only data from non-blinking periods were used.

Mean square displacement (MSD) and diffusion coefficients were calculated according to established formula [10, 20]:

where (x_n+i, y_n+i) is the position of the QD following a time interval of nt (t is the time interval between successive measurements) after starting at position (x _i , y _i ). N is the total number of positions recorded; n ranges from 1 to N-1. QD position data at each time point were put into a spreadsheet generated by Metamorph software and then fed into software written in-house which calculated MSD automatically from data in the spreadsheet according to above formula. The statistical significance of differences was quantified with Student t-test using SigmaPlot software (SPSS, Chicago, IL).

Diffusion coefficients were calculated according to the formula:

where D is the diffusion coefficient and t is the elapsed time interval.

For QD tracking lasting more than 10 min, it was not possible to circumvent the blinking problem. To analyze tracks in which QD points were missing in three frames or less, an interpolation algorithm was developed to estimate the missing positions. First, MSD _b from points up to the blinking was calculated. Equations were then set up to calculate the x and y positions of the invisible points that minimized the difference between SD (square displacement between the invisible and visible points before and after blinking) and MSD _b . To assess the validity of this algorithm, a simulation based on a 100-point track was conducted. We randomly deleted one, two or three points from this track to simulate QD blinking and the missing points were then filled in by the algorithm. The diffusion coefficients calculated from the filled-in track showed no significant difference from that calculated from the original track.

Visualization of AChRs with QDs

In cultured muscle cells AChRs often spontaneously form clusters in the absence of synaptogenic stimulation. These pre-patterned clusters (also called hot spots) can be readily visualized by labeling cultures with fluorescent α-bungarotoxin (BTX), as shown by the example in Fig. 1a where Alexa488-BTX was used to label muscle cells. With this method, however, individual non-clustered AChRs distributed diffusely on the cell surface cannot be detected. To visualize the diffuse receptors, muscle cells were labeled with biotinylated BTX followed by streptavidin-conjugated QDs, or BBQs for b iotin-B TX/Q Ds. BBQs strongly labeled pre-patterned AChR clusters as well as non-clustered AChRs on the muscle surface (Fig. 1b). Thus, QDs enabled both aggregated and non-aggregated AChRs to be probed.

BBQ-labeling specificity was confirmed in several ways. No QD signal was present when BBQs were applied after muscle cells were pretreated with 5 μM unlabeled BTX for 30 min (Fig. 1c) or when cells were treated with QDs but without biotin-BTX (Fig. 1d), and non-muscle cells found in our cultures were not labeled by BBQs (data not shown). These results demonstrated that BBQs selectively marked muscle AChRs. We also noted that pre-patterned AChR clusters on the top surface of muscle cells were strongly labeled by QDs but clusters on the bottom facing the coverglass substratum were not, presumably because QDs were excluded by the tight cell-substratum attachment at these sites (a gap of 10–15 nm; [21, 22]). Although diffuse AChRs on cell bottom could be labeled by BBQs, in this study we focused on tracking only AChRs present on the top surface of muscle cells.

Tracking the movement of AChRs

When a saturating concentration of BBQ was used to label all diffuse AChRs, it was not possible to track individual receptors unambiguously due to their high density. A range of BBQ concentrations was thus tested to obtain the optimal labeling condition. A mixture of Alexa488-BTX and biotin-BTX at a ratio of 50:1 yielded low BBQ density at the cell surface and enabled accurate tracking of single molecules.

The high fluorescence stability of QDs allowed their observation over long periods. Under our experimental conditions cells were healthy during time-lapse recordings for more than 30 min as judged by the intactness of the characteristic cross-striations in phase optics and by diffusion coefficient measurements (described below). Furthermore, for analyses we only used data obtained during the first 10–20 min recording from cells that remained healthy at the end of the observation period. We also used muscle cells maintained in cultures for different periods of time and found that under identical BBQ labeling conditions, surface receptor density was independent of culture age up to three weeks, but the fraction of immobile AChRs increased with time (see below). Results below were all obtained using cell cultures less than one week old.

To examine the movement of QD-labeled AChRs, time-lapse recording was carried out. Taking advantage of the fact that only single QDs blink, we were able to identify AChRs linked to single BBQs (see Discussion for further details) and track their trajectories on the muscle surface (Fig. 2a). BBQ-labeled receptors underwent random movement on the cell surface, and to quantify this movement mean square displacement (MSD) was calculated. As shown in Fig. (2b and 2c), the early phase of the MSD plot lasting up to 10 min was typically linear, which is characteristic of particles undergoing free Brownian motion. Over 20 min BBQs covered an area as much as 20 × 20 μm² but the net distance traversed (between the beginning and end point) was typically much less (Fig. 2a).

From the linear portion of the MSD plot, the diffusion coefficient (D) of single BBQs was calculated to range from 10^-12 to 10^-9 cm²/s. In muscle cultures up to 6 days old, the majority of receptors moved with D on the order of 10^-10 cm²/s, while a small percentage of receptors showed faster movement (> 10^-9 cm²/s) (Fig. 3a). The proportion of fast moving receptors was highest in cultures less than 2 days old but decreased subsequently with an increase in the population of slower moving receptors, with D on the order of 10^-11 cm²/s (Fig. 3a). We also found a population of immobile or nearly immobile receptors with D less than 10^-11 cm²/s (Fig. 3a) which was rarely detected in cultures less than 2 days old but increased with culture age to reach 50% of total after 3 weeks. This suggests that diffuse AChRs become increasingly restrained in their membrane environment with development. In this study, data were collected mainly from cultures 3–6 days old.

Although the D values of single AChRs varied from cell to cell, they were remarkably constant when measured from the same cell (data not shown), and for a single AChR, D values calculated from different parts of its trajectory were also essentially the same. For example, D values calculated for a single AChR from one 17-min trajectory were 9.9 × 10^-10, 8.5 × 10^-10 and 1.2 × 10^-9 cm²/sec during the early (< 30 sec), intermediate (8–8.5 min) and late (16.5–17 min) stages respectively. During the time-lapse recording, some cells became non-viable as shown by obvious morphological changes such as sudden shrinkage or loss of cross-striation, presumably due to photo-damage. This transition was always associated with a sudden cessation of BBQ movement.

AChR movement at pre-patterned clusters

To study the behavior of single AChRs during the formation of clusters, we examined BBQ movement at or near these sites. First, pre-patterned AChR clusters in non-innervated muscle cells were studied. These prominent and structurally complex clusters were observed in nearly all cells after three days in culture and within them most BBQs were stationary or nearly immobile (with D of 10^-12 cm²/s). Fig. 4 shows BBQ movement (a'-c') within or near clusters identified by Alexa488-BTX (a-c). The BBQ images (Fig. 4a'–c') were obtained by superimposing two frames of a time-lapse recording separated by 2.5 s, with the first image pseudo-colored in green and the second in red. Immobile receptors appear as yellow dots and mobile ones as green or red dots. Most AChRs within the cluster domain were immobile (Fig. 4a, a') but those outside were mobile (arrows at the bottom of Fig. 4a') with D on the order of 10^-10 cm²/s. Movement of individual BBQs into and out of a cluster was infrequently observed.

To visualize the behavior of single AChRs at developing pre-patterned clusters, we studied muscle cultures 1–2 days old when the clusters present were much smaller than mature clusters (Fig. 4b). The clusters grew at a rate of 0.2 ± 0.07 μm²/min over 2–3 h to reach a size of 5–6 μm in diameter. Single BBQs remained mobile within and around the clusters before they reached the final size (arrows in Fig. 4b, b'), but became mostly immobile within but not outside cluster confines at the end of this process (arrows in Fig. 4c, c', same cluster as in 4b, b' at a later stage). These observations suggest that a highly localized mechanism for immobilizing diffuse AChRs is set up at the onset of cluster formation and it grows through an expansion of the membrane domain for AChR entrapment.

A test of the diffusion-trap hypothesis: AChR clustering induced by beads

Although pre-patterned AChR clusters are reliably found in cultured muscle cells, their position cannot be predicted. Therefore, to further understand how single AChRs are recruited into clusters, we studied clusters induced by beads coated with heparin-binding growth-associated molecule (HB-GAM). HB-GAM-coated beads focally induce AChR clusters with high fidelity, and with their use both the onset and location of clustering can be precisely marked [19]. By fluorescent BTX labeling, clusters become detectable within two hours at bead-muscle contacts, and the density of AChRs (as reflected by fluorescence intensity) at these clusters increases until reaching saturation after overnight bead stimulation.

AChR movement at developing bead-induced clusters was followed using BBQs and time-lapse recording. Single BBQs within these clusters were individually tracked for 2 min or longer at three different time points (2, 4 and 24 hr) after the establishment of bead-muscle contacts. An example of the movement history of one such single BBQ in a bead-contact area (Fig. 5a) shows that the BBQ displayed continuous random diffusion with variable instantaneous velocity for 1 min but then suddenly stopped at the 1 min mark and remained immobile afterwards; the BBQ's mean velocity while in motion is indicated by the horizontal line. The colored lines in Fig. 5a' show other examples of BBQ movement and trapping at bead-muscle contacts, with each line indicating the mean velocity of one BBQ.

To eliminate the possibility that BBQs were immobilized at bead-induced clusters as a consequence of the tight cleft space between the bead and the cell surface, QD movement was also recorded at bead edges. Although this zone is at the periphery of bead-muscle contacts, AChRs are clustered there [23], and BBQs at these sites (Fig. 5a'; black dashed lines) moved and then suddenly stopped and remained stationary afterwards during the recording period. Thus, immobilization of BBQs is not due to simple mechanical restraint within the bead-cell cleft space.

A second subset of BBQs remained mobile within the bead-muscle contact region throughout the recording period (as shown by the velocity plot in Fig. 5b) and sometimes exited this area with time. Although these mobile AChRs were inside the area of cluster formation, they were indistinguishable in their continuous movement from receptors that were outside the bead-muscle contact (Fig. 5c). However, the fraction of immobile AChRs (D < 10^-11 cm²/s) increased over time within the bead-induced clustering domain. Sample recordings are graphically shown in Fig. 5d–h and quantified in Fig. 5i. In consistency with our previous finding that AChR clustering starts at bead-muscle contacts within minutes, we observed that the fraction of immobile AChR rose significantly (to 50%) during the first two hours of bead-stimulation (Fig. 5d, i). Thereafter, this fraction continued to increase until reaching saturation (~90%) by 24 h (Fig. 5e, f and 5i), comparable to that seen at pre-patterned AChRs (Fig. 5g, i). In contrast, the overwhelming majority of AChRs outside bead-muscle contacts remained mobile (D > 10^-11 cm²/s; Fig. 5h, i) during this period.

To determine if the bead stimulus that induces AChR clustering locally also exerts long-range effects on receptors outside the contact domain, we examined the diffusion of AChRs in zones at different distance from the beads. Receptors located within < 20 μm or at 20–40 μm and > 40 μm from the beads were tracked and their diffusion coefficients were determined through MSD analyses. Our results showed that receptors at all three distance ranges were equally mobile (Fig. 5j), with nearly identical mean diffusion coefficients. This suggests that the bead-mediated AChR cluster-stimulating signal is confined locally and does not extend beyond the immediate boundary of the contact area.

Collectively, the above results provide direct evidence at single-molecular level for the diffusion-mediated trapping of AChRs during cluster assembly in the muscle membrane. Moreover, they highlight the role of independent, diffusion-mediated movement of single receptors as a driving force for the clustering process.

AChR movements at developing NMJ

In Xenopus spinal neuron-muscle cocultures, nerves induce AChR clusters where they touch muscle cells. With Alexa488-BTX-labeling new clusters along nerve-muscle contacts could be readily identified, as shown in Fig. 6, panels a and b, with the corresponding BBQ image shown in Fig. 6c. The synaptic cleft between the nerve and muscle is ~50 nm in width [24], considerably larger than the diameter of QDs used in this study. Within nerve-induced clusters most BBQs were immobile, like those in pre-patterned and bead-induced AChR clusters described above. A small sample area within the cluster domain (Fig. 6c square) is shown in time-lapse recording in Fig. 6d1–d8. In this example, an immobile BBQ can be seen (yellow arrowhead), and also seen are mobile BBQs (Fig. 6d1–d8, green arrows) whose movement was confined to an area more restricted than that of BBQs outside the postsynaptic region. The mobile receptors in postsynaptic clusters typically had D on the order of 10^-12 to 10^-11 cm²/s, lower than of those present in the extrasynaptic area, and, additionally, sudden immobilization of these BBQ-labeled receptors was often detected (Fig. 6d1–d8, red arrows; note the cessation of movement from frame d4 onwards). The mean velocity plots of BBQs in Fig 6e show the trapping of several such mobile receptors. Conversely, we also observed the escape of BBQs from clusters, seen as a sudden resumption of their movement (Fig. 6f), which suggests that receptor trapping during NMJ formation is not irreversible.

We also noted that in areas immediately adjacent to postsynaptic AChR clusters, BBQ movement was unrestrained, similar to that of BBQs far from nerve contacts or present on the surface of non-innervated muscle cells, with D values ranging from 10^-12 to 10^-9 cm²/s. Calculation of the mean diffusion coefficients of BBQs at different distances from innervation sites (Fig. 6g) clearly demonstrated that the movement of extrasynaptic receptors is independent of their distance from synapses. These results suggest that, like the bead signal, innervation generates a highly localized mechanism for AChR immobilization.

The effect of F-actin disruption on AChR movement

Previous studies have shown that local F-actin assembly is necessary for the formation of AChR clusters. In fact, dynamic actin polymerization can provide enough motive force for the translocation of entire AChR clusters [10]. Thus, it is of interest to know whether the lateral movement of AChRs at the cell surface is dependent on F-actin assembly. To this end we examined the mobility of single AChRs in cells treated with latrunculin A (LtnA), a marine sponge toxin that blocks actin polymerization by sequestering G-actin. As above, diffusion coefficients were calculated from MSD plots generated from time-lapse BBQ recordings. At a concentration of 40 μM that effectively inhibits AChR clustering [10], LtnA did not significantly alter either the distribution of diffusion coefficients or their mean as compared to the DMSO control (Fig. 3b). This showed that lateral movement of AChRs is diffusion-driven and independent of F-actin cytoskeletal assembly.

Comparison of AChR and GM1 movement

Because recent studies have suggested that AChRs in the muscle membrane are sequestered into "lipid rafts" enriched in cholesterol and sphingolipids [25, 26], we compared the behavior of AChRs and lipid molecules at the cell surface. The B subunit of cholera toxin (CTX) binds specifically to the ganglioside GM1 and can be used as a probe for glycosphingolipid in the plasma membrane [27, 28]. Here we applied biotin-conjugated CTX followed by streptavidin-conjugated QDs (B iotin-C TX/QD s, or BCQs) to track GM1 movement.

Sample trajectories of a BCQ and a BBQ are shown in Fig. 7a (note the difference in the scaling of the axes). Interestingly, like BBQ (Fig. 7b), BCQ-labeled GM1 lipid molecules also existed in mobile and immobile fractions (Fig. 7c) and the mobile BCQs exhibited diffusion-driven movement in the membrane. The diffusion coefficients of mobile BCQs, however, existed in the narrower range of 10^-10 to 10^-9 cm²/s and were generally an order of magnitude higher than that of BBQs in the wider range of 10^-11 to 10^-9 cm²/s.

To further explore the relationship between AChR and GM1, the effect of Con A on BBQ and BCQ movement was examined. Because this lectin crosslinks AChRs by binding to four mannose residues on the receptors [9], BBQ movement was completely blocked after cells were treated with 100 μg/ml Con A (Fig. 7d). In contrast, BCQ movement was unaffected by this treatment (Fig. 7e), indicating that diffuse AChRs and GM1 move independently of each other.