Supplementary Figure 4: dSTORM buffer efficiency over time.

(a) (Top) Heatmap representation of frame_50, the median frame at which 50% of the localized have been detected. In a perfect dSTORM regime, the median frame should be close to the middle frame (4,000 for an acquisition of 8,000 frames) as for well A1. (Bottom) Boxplots displaying the median values and IQR of 8 groups of 12 wells grouped per hour after buffer incubation (corresponding to an entire line of the plate, as color box next to the heatmap). (b) Normalized histoplots of the number of localizations per frame in wells separated by 1 hour. Even if the total number of detections and the quality of the single molecule detections are suitable during the complete 96-well plate acquisition (8 hours, see Fig. 3 and Fig. 4), we can observe that the buffer efficiency for Alexa 647 starts being affected already after 1 hour of incubation (well A10 as compared to well A1). A continuous decrease of the frame_50 metadata correlates with the buffer incubation time. (c) 3D dSTORM images taken 6 hours, 10 hours and 14 hours after the first acquisition. Images have been acquired automatically in the same plate in different wells. We can observe that, after 14 hours of buffer incubation, the total number of localizations became too small (less than 1 million in a FOV of 20.5X20.5 μm²) to properly reconstruct continuous microtubules with the buffer 200 mM of thiols at pH7.2.