Live-cell dSTORM of cellular DNA based on direct DNA labeling.

Super-resolution (SR) fluorescence imaging relies on specially adapted microscopes and analysis software, but equally important are the fluorescent probes used to label biological proteins and molecules of interest. In point-localization SR, the on and off rates of fluorophore photoswitching must be controlled to separate the signals from individual molecules temporally. The fluorescence from each molecule is then spatially localized; this allows their positions to be obtained with nanometer precision. Molecular positions are then rendered to give a reconstructed image with a spatial resolution of up to several tens of nanometers. 8] Although fluorescent proteins can be used as labels, synthetic dyes have the advantage of higher photon yields that lead to higher attainable resolution. A special challenge for chemical biologists is to develop or identify synthetic dyes that are compatible with live-cell SR imaging as most dyes are not cell membrane-permeable. Recently, the point-localization SR method of direct stochastic optical reconstruction microscopy (dSTORM) was used to image histone H2B proteins in living cells. Proteins were labeled with rhodamine and oxazaline dyes by using the genetically encoded chemical trimethoprim or SNAP tags. In a separate study, dSTORM was applied in vitro to image purified DNA by direct labeling with the cyanine-based YoYo-1 dye. However, most DNA-associating dyes are not compatible with live-cell imaging owing to their cell impermeability and cytotoxicity. Thus, live-cell super-resolution imaging of DNA structure has never been demonstrated. We present here the imaging of DNA in living cells with dSTORM based on direct labeling with the commercially available cyanine-based Picogreen dye. Picogreen has several advantages over other DNA-specific dyes for live-cell imaging, including minimal perturbation to DNA structure and an increase in fluorescence upon binding that results in low background fluorescence. We identified a live-cell imaging medium that optimizes the reversible photoswitching of the fluorophores, and used it to resolve nuclear and mitochondrial DNA structure directly. Furthermore, due to the excellent preservation of these dyes, we were further able to perform time-lapse dSTORM imaging of directly labeled DNA. Theoretically, by choosing an appropriate reducing–oxidizing system, one can achieve the controllable reversible photoswitching required for the dSTORM of nearly any cyanine derivative. Several cyanine-based dyes have already been used in dSTORM imaging. Typically, enzymatic oxygen-scavenging buffers are used to prevent the irreversible photobleaching of dyes, and mercaptoethylamine (MEA) is used as a reducing agent to induce photoswitching or photoblinking. MEA is responsible for converting the dye into long-lived, dark, nonfluorescent states from which single molecules can spontaneously recover to the ground state upon interaction with residual oxygen; this is the photochemical basis of dSTORM. Buffer conditions determine oxygen concentration and reduction rate, and are therefore important for optimizing photoswitching rates. To perform live-cell imaging, we avoided adding the potentially toxic thiol reducing agents, such as MEA, that are commonly used to induce photoblinking. We tested buffer conditions that have worked for live-cell dSTORM in the past for other dyes. In phosphate-buffered saline (PBS) alone, we observed rapid photobleaching of Picogreen with no recovery (Figure 1A). When using a buffer containing an oxygen-scavenging (OS) system composed of 10% glucose, 0.5 mgmL 1 glucose oxidase, and 40 mgmL 1 catalase, the fluorescence intensity over time showed a much slower decay. Thus, enzymatic removal of oxygen indeed helps to avoid photobleaching of the dye. However, it was not sufficient to induce the single molecule blinking required in dSTORM. To improve the recovery and photoblinking of Picogreen, we applied ascorbic acid (AA) as a reducing agent and the OS to reduce irreversible photobleaching. We found that 1 mm ascorbic acid com-