Switching behaviour of dSTORM dyes in glycerol-containing buffer

To suppress optical aberrations caused by refractive index mismatch, we employ glycerol-immersion objectives in conjunction with fused silica cover glasses and imaging buffers with a high glycerol content. Here we demonstrate that the addition of glycerol to the buffer does not degrade the switching behaviour of the dyes Alexa Fluor 647 and Alexa Fluor 568 in dSTORM measurements, which shows that this approach is suitable for dSTORM. Additionally, we report evidence that sealed sample geometries as used in our experiments reduce photobleaching due to the lower influx of oxygen into the imaging buffer.


Details about Sample Preparation
The dilutions defined in Supplementary Table 1

Influence of the activation laser
We found that in our glycerol-containing-buffer, illumination with an activation laser reduced the survival fraction and duty cycle of Alexa Fluor 647. As this result differed from a previous study with aqueous buffer 4 , we measured these parameters again for a buffer that contained no glycerol. All other measurement conditions were identical. Without glycerol, we did not observe the dramatic drop of the survival fraction and the duty cycle was increased, not reduced. This shows that for some dyes, glycerol addition to the buffer indeed requires to adapt the measurement protocol. Switching properties of yellow-absorbing dyes Supplementary Table 4 shows the photon number per switching event, the mean duty cycle, the survival fraction and the total number of switches for all dyes in the yellow-absorbing range which were investigated in this study.

Dependence of photon numbers on measurement setup
While the equilibrium on-off duty cycle, the survival fraction and the mean number of switching cycles only depend on a detection of signals and recognizing fluorophores as being switched on, the calculated photon number per switch depends on the intensity of the detected signal and thus on the individual microscope. With respect to the yellow-absorbing dyes, the photon numbers are comparably low as compared to other studies 4 . Besides the difference in the calculation method of the mean photon number per switch (cf. Methods section of the main article), there is also an impact of our filters on the brightness of the detected signal. For the detection of yellow-absorbing molecules, we use a notch filter (QuadLine Rejectionband ZET405/488/561/640), AHF analysentechnik AG, Tübingen, Germany) and an additional orange filter (617/73 BrightLine HC, AHF analysentechnik AG, Tübingen, Germany). Supplementary Figure 1 illustrates for the example of Alexa Fluor 568 that the compromise of using the notch filter to filter out four different laser wavelengths has the effect that only 54.7 % of the spectrum emitted by Alexa Fluor 568 is transmitted to the detector. This impact always has to be considered if photon numbers from different setups and sample preparations shall be compared to each other.

Overview of Fluorescent Dye Analysis
For the sake of better comparability, mainly mean values and medians are used in the main article to refer to our results. In the Supplementary Figures 2 to 6, overviews about the fluorescent dye evaluations are given for getting more detailed information. In the above mentioned figures, exemplary fluorescent time traces, photon histograms over all measurements as well as the time dependent DC and the time dependent survival fraction are shown.
In the time traces, single data points that are identified as an on-switched molecule are marked with red dots. As the duration of a fluorescent blinking event is less than about 100 ms, the peaks are visible as a straight line in the chosen illustration although they are extended in time. For example, the rightmost peak in Figure Supplementary 2 a) is marked with three red dots meaning that the peak is identified as an on-switch being extended over three time steps, i.e. ≈ 100 ms.