S TED -Microscopy of Vesicle Motion

In a first step, protocols for fluorescence staining of live cells and the general ability of the STEDmicroscope to resolve individual neurotrans-mitter vesicles in living neurons were established. Cultured hippocam-pal neurons were labeled with monoclonal mouse antibodies against the intravesicular (lumenal) domain of the synaptic vesicle protein Synap-totagmin [215, 240, 45,341]. After brief application of the antibodies to the neurons on ice (≈2^◦C), they were detected with Fab fragments, coupled to the organic fluorophore Atto 647N (Fig. 4.1). This proce-dure ensured that only vesicles fused to the plasma membrane were labeled [341], because only these vesicles were exposed to the outside space and therefore accessible for the antibodies (Fig.4.2). The low tem-perature ensured that the fused vesicles were blocked in this state and allowed the binding of the antibodies. The brief low temperature treat-ment did not seem to affect the viability of the cultures. Moreover, when the temperature was only slightly higher (≈4^◦C) the cultures are able to exocytose upon stimulation [262]. The labeled vesicle pool encompassed

Synaptotagmin

Primary antibody against Synaptotagmin Secondary Fab fragment with dye

Figure 4.2: Staining of neurotransmitter vesicles in the boutons. Offering primary antibodies and secondary Fab fragments while the neurotransmit-ter vesicles are fused to the plasma membrane enables the immunostain-ing of intra-vesicular components, here of the Synaptotagmin.

as much as 10% to 20% of all vesicles [263]. The rapid endocytosis of the label confirmed the full participation of this pool in active vesicle recycling: at room temperature, two minutes after labeling, only≈18%

of the label still resided on the surface; after 10 min only≈8% (Fig.4.3).

With the STED microscope shown in Fig. 2.4 an 1.8 µm by 2.5 µm area was imaged within 35 ms, i. e. at 28 fps; the pixel dwell time was 0.93 µs at the center, increasing to the side.

0 5 10

Time [min]

Normalized fluorescence [%]

0 20 40 60 80

Figure 4.3: The labeled vesicles are rapidly endocytosed, as shown by the reduced number of surface-exposed Synaptotagmin molecules with in-creasing time between primary and secondary antibody labeling. Normal-ized fluorescence is shown as mean ± SEM, with each data point repre-senting three to seven independent neuronal cultures. The dashed line in-dicates the background fluorescence determined from unstained prepara-tions. The insets show epifluorescence images after 0 min and after 10 min.

The imaging capabilities of the STED microscopy were compared with confocal microscopy (Fig.4.4); the STED images clearly visualize the motion of single individual vesicles or patches of clustered Synapto-tagmin within the axon, while the confocal recordings can only resolve the axon, but not the objects therein.

Figure4.5shows how STED microscopy makes the detection of sin-gle synaptic vesicles possible, in contrast to standard confocal imag-ing (The full movie is available on the website of the Science Maga-zine at http://www.sciencemag.org/content/vol0/issue2008/images/data/

1154228/DC1/1154228S1.mov). Due to the short exposure times and the confocal fluorescence detection (i. e. with a pinhole in front of the detector) the average noise levels were only≈0.1 counts per pixel in the axons and ≈0.02 counts per pixel outside (Fig. 4.5). Although a sin-gle detected photon in a pixel could be due to noise, the probability of simultaneously detecting three or more photons was several standard deviations above the mean noise level. One can therefore conclude that the detected signal originated indeed from labeled objects. This was confirmed by the comparison of consecutive frames, where the labeled objects moved only slightly (Fig.4.5, middle and right panels); random noise, by contrast, would appear at arbitrary positions. The frequency of false-positive identifications in preparations where no specific signal was expected was found to be on average around 0.1 events per frame (see Sec.4.1.7for details).

The apparent FWHM of the vesicles in the STED images was 62 nm (Fig.4.6), which means that the effective-PSF area was 18-fold reduced compared to confocal imaging. To show individual vesicle movements clearly, raw images were smoothed by convolution with an 80 nm Gaus-sian kernel (Fig.4.7).

The high spatial resolution allowed the separation of individual vesi-cles. The combination with high temporal resolution for the first time enabled automated single-particle tracking to determine the motion be-havior of many vesicles simultaneously and quantitatively. Positions of all objects in each movie frame were determined by the Local Maxima algorithm described in Sec.2.3.1. The objects were then tracked as de-scribed in Sec. 2.3.3, i. e. corresponding objects in consecutive frames were found. This allowed the calculation of the speeds (i. e. distances moved per frame time interval) of individual vesicles.

Figure 4.4: Comparison of confocal and STEDmicroscopy. (A) Successive confocal frames of stained Synaptotagmin in an axon (raw data). The axon is resolved, but not the single objects within. (B) Successive STED frames of the same area (raw data). Single objects can be seen to run along what appears to be two tracks. (C) Same data as in (A), filtered with a 65 nm spatial Gaussian filter. Some changes of the fluorescence distribution in consecutive frames are visible, but no individual objects. (D) Filtering of the STEDdata [same data as in (B)] with a 65 nm Gaussian filter reduces noise and enhances the visibility of the objects. White numbers indicate the recording time point in seconds. The colormap is indicated in the last image. Scale bar 250 nm.

Figure 4.5: Images of a short fragment of a stained axon, recorded con-focally (frame #1) and via STED microscopy (frames #2 and #3). The in-creased resolution in the STED image and the reappearance of vesicles in subsequent frames can be seen. Three arrowheads indicate relatively sta-ble vesicles; many other spots can be recognized in both frames, too. The insets in frame #1 and frame #2 indicate the color maps used, with their upper and lower end in counts per 0.93 µs.