Primary neuronal culture
All experiments were performed with primary rat hippocampal cell cultures prepared from E18 embryos taken from time-pregnant Wistar rats (RCC, Füllinsdorf, Switzerland), as previously described 
. Cells were plated at a density of 40–50×103
cells per 18 mm glass coverslip and cultured at 37°C/5% CO2
for about 3 weeks in minimal essential medium supplemented with 2% B27, 15 mM HEPES, 0.45% glucose monohydrate, 1 mM sodium pyruvate (all from Invitrogen, Basel, Switzerland), 2 mM L-glutamine (Gibco, Basel, Switzerland), 15% Nu-serum (Becton Dickinson, Basel, Switzerland). Experiments were done at 20–22 days in vitro
Neuronal cultures were transiently transfected by magnetofection 
with mCherry-Homer1c and EGFP-gephyrin. These proteins are selective markers of glutamatergic and GABAergic postsynaptic sites 
. After transfection in primary cultures, their postsynaptic localization in GABAergic interneurons was achieved upon long term expression. Thus, both constructs were transfected at 11 div; half of the medium was replaced with fresh medium to ensure cell growth until 20–22 div.
For pharmacological experiments different drugs were bath-applied into the culture medium for defined time periods before the experiments: biotinylated α-BT (Molecular Probes, Basel, Switzerland; 125 nM, dissolved in PBS); KCl (40 mM, dissolved in H2O); methyllycaconitine citrate hydrate (MLA), a selective α7-nAChR antagonist (Sigma-Aldrich, Buchs, Switzerland; 1 µM, dissolved in H2O); PNU-282987, a selective α7-nAChR agonist (Sigma-Aldrich; 300 nM, dissolved in DMSO); tetrodotoxin (TTX) (Sigma-Aldrich; 1 µM, dissolved in H2O), latrunculin A (Sigma-Aldrich; 3 µM, dissolved in DMSO) and nocodazole (Sigma-Aldrich; 10 µM, dissolved in DMSO), which depolymerise actin and tubulin, respectively.
Post-hoc synapse labeling was performed after fixation of the cells for 10 min with 4% paraformaldehyde in 150 mM sodium phosphate buffer followed by permeabilisation for 5 min with 0.1% Triton® X-100, 10% normal goat serum (NGS; Serotec, Düsseldorf, Germany) in phosphate buffered saline (PBS). Primary rabbit polyclonal antibodies against vesicular glutamate transporter type 1 (vGluT1; Synaptic Systems, Göttingen, Germany, Nr 135 303; 1
8000) and vesicular inhibitory amino acid transporter (VIAAT; Synaptic Systems, Nr 131 003; 1
1000) were applied for 45 min in 10% NGS in PBS. Secondary goat anti-rabbit antibody coupled to Cy3 or Cy5 (Jackson ImmunoResearch, West Grove, PA; 1
1000 and 1
200) was incubated for 30 min in 10% NGS in PBS.
Images from transfected neurons were analyzed to determine the fraction of EGFP-gephyrin and mCherry-Homer1 clusters apposed to vGlut1 and VGAT-positive terminals, thereby ascertaining their postsynaptic location. Images were processed with a Gaussian filter 
(ImageJ) to amplify small and large clusters with low intensities. Then, they were converted to 1-bit images with clusters having value 1 and background value 0. The relative number of EGFP-gephyrin and mCherry-Homer1 clusters and their mean surface area was then calculated in a sample of 7 cells from two independent cultures. Next, to quantify the apposition to presynaptic terminals, EGFP-gephyrin and mCherry-Homer1c clusters were enlarged by 1 pixel using a distance map filter (ImageJ), and colocalization with vGluT1 or VGAT immunofluorescence staining determined and counted using the “analyze particles tool” of ImageJ.
Surface α7-nAChRs on living cells were labeled with α-BT-AlexaFluor® 647 (α-BT AF647; Molecular Probes; 125 nM) for 2–5 min in cell-conditioned medium at 37°C/5% CO2.
Single receptor labeling of α7-nAChRs with QDs (QDOT®s; Invitrogen) was done at 4°C to reduce internalization and unspecific staining of QDs, as follows: Living cells were incubated with biotinylated α-BT (50–125 nM) in cell-conditioned medium for 5 min at 37°C/5% CO2. Cells were rinsed 3 times with 4°C PBS. Streptavidin-coupled QD605 or QD647 were preincubated at a concentration of 1–2 nM in freshly prepared 1% BSA Fraction V (Sigma-Aldrich) in PBS for 5 min at room temperature to avoid unspecific binding and were applied in a second staining round to the cells for 2 min at 4°C. Cells were rinsed subsequently 12 times in 4°C PBS containing 100 nM biotin to block remaining streptavidin-biotin binding sites. For live imaging, cells were covered with recording medium (minimum essential medium without phenol red supplemented with 15 mM HEPES, 0.45% glucose monohydrate, 1 mM sodium pyruvate, and 2 mM L-glutamine).
Live microscopy and QD imaging
Live microscopy was performed on a Leica DMI6000b inverted microscope equipped with a 63× objective (NA 1.3). Cells were mounted in a metal chamber (Life Imaging Services, Basel, Switzerland) covered with recording medium, and kept at 5%CO2 and 37°C. Dyes were illuminated by a mercury lamp (EL6000; Leica). QD excitation and emission was controlled by specific filters (AHF, Tübingen, Germany). For detection an EM-CCD camera (C9100; Hamamatsu, Solothurn, Switzerland) was used (106 pixels; pixel size, 0.125 µm×0.125 µm). Acquisition of images and movies was performed with the software Velocity (Improvision, Coventry, UK).
Prior to QD imaging, pictures of differential interference contrast (DIC), α-BT AF647, and transfected constructs were taken. Subsequently a movie of the QD labeled α7-nAChRs was recorded for 30–40 s with 50 ms exposure time at a rate of 20Hz. To limit phototoxicity cells were illuminated with the lowest possible light intensity and sessions lasted maximally 45 min.
Single particle tracking and analysis
For better handling, recorded movies were converted from RAW to AVI format using ImageJ (Rasband, 1997–2008). Single QDs were identified by their on-off blinking behavior. Trajectories of single QDs were tracked by using the ImageJ plugin Particle Tracker 
. Trajectory interruptions due to off phases of the QDs were interpolated and subtrajectories were linked if the phase did not exceed more than 10 consecutive frames. The maximal displacement of a QD from frame to frame was set to 3 pixels (375 nm). Trajectories consisting of less than 100 frames were excluded from further analysis.
Trajectory analysis was done with custom software written in Excel Visual Basic. The mean square displacement (MSD) of a trajectory was calculated according to the following equation 
) and y
) is the particle position of a trajectory with N
frames at frame i
with a frame time interval of dt
. By fitting the first 5 data points of the MSD versus time (t
) curve the diffusion coefficients (D
) was calculated with the equation:
. Instantaneous diffusion coefficients were derived accordingly over contiguous trajectory stretches of 20 frames.
To distinguish between random movements and confined behavior of receptors diffusing within a limited area, the confinement index L was calculated as described 
. L is a probability level reflecting the tendency for non-random confinement. When averaged over time, it can be used as a confinement index. As shown in the references above 
, L>3.16 for 2.5 sec has a likelihood of >99.3% to reflect confined mobility. This threshold was used here for identifying episodes of confined diffusion. Finally, the confinement surface area L2
was estimated by fitting the MSD plot with the following equation 
To determine at any time point of its trajectory the distance of a QD to the nearest postsynaptic site, clusters of EGFP-gephyrin or mCherry-Homer1c were binarized as described above, and the distance to the nearest cluster was determined at every position of the image by applying a distance map filter (ImageJ). From these data, the distribution of instantaneous locations as a function of a distance from the PSD was calculated. This distribution was normalized to the surface area occupied by image pixels around PSDs. The PSD was considered as being circular with a mean radius of 0.1 µm (as determined from their average size). The surface area of concentric circles around the PSD, spaced by 0.125 µm (pixel size), was then used for calculating the number of subtrajectories present around the PSD as a function of distance.
Next, we arbitrarily defined EGFP-gephyrin and mCherry-Homer1 clusters as representing a synaptic zone, and, based on the distribution curves of the instantaneous location of QDs, a radius of 4 pixels (<500 nm) around these clusters as representing a perisynaptic zone. Membrane areas further than 500 nm from the edge of clusters were defined as extrasynaptic.
Measuring the distance of QDs from clusters thus allowed to split and classify trajectories into synaptic, perisynaptic and extrasynaptic parts, and to determine the instantaneous diffusion coefficient and dwell time in these compartments. Calculation of distance from a cluster vs. diffusion coefficient and confinement vs. diffusion coefficient was enabled by taking the instantaneous diffusion coefficient into account.
Data and statistical analysis
Data are presented as mean±SEM. Statistical analyses were done using Prism® 4 (GraphPad Software, La Jolla, CA) and SPSS 11.5 (SPSS Inc., Zürich, Switzerland).