Figure 1.
Setup of spinning disk stochastic imaging (SDSI) system and evaluation of imaging capabilities.
(A) Schematic diagram describing how the point spread function is refined in a selected axial plane by spinning disk confocal microscopy and super-resolution image processing. (B) Diagram showing the configuration of the SDSI microscope, abbreviations are as follows. EMCCD = Electron multiplied Charge Coupled Device camera. FW = Filter wheel. SD = Yokagawa CSUX1 spinning disk, M = mirror. PCF = Piezo coupled focus feedback unit. OB = Objective. (C) Workflow of the SDSI experiments used in this paper, Briefly samples are prepared with probes for either PALM or dSTORM, next samples are imaged and finally SR data is generated using the SOFI image processing algorithm.
Figure 2.
SDSI super resolution imaging using both PALM and STORM.
(A) Comparison of SD and PALM images (processed by SOFI) of Eos-Actin, (B) Comparison of SD and d-STORM images of Connexin 43, Secondary Fab fragment antibodies conjugated to AlexaFluor647 were used for dSTORM imaging, SR data was generated using 3rd order SOFI, in both images bar = 1 µm. (C) Comparison of SD and SDSI combining PALM and dSTORM imaging of Eos-Histone 2B (PALM) and Lamin A/C (dSTORM). Secondary Fab fragment antibodies conjugated to AlexaFluor647 were used for dSTORM imaging. SR data was generated using 3rd order SOFI. Arrow indicates individual histone complexes, bar = 2 µm.
Figure 3.
Comparative study of different stochastic super-resolution image processing algorithms.
(A) Reconstructed super-resolution images from: a simulated dataset of 800 frames, a 64×64 pixel, 5000 frame dSTORM dataset of actin filaments labelled with AlexaFluor647 visualised by TIRF microscopy, a 90×90 pixel 5000 frame PALM dataset of Hela cells transfected with Eos-Actin. SR images were generated using DeconSTORM, FasterStorm, QuickPALM, RainSTORM, 3rd order SOFI and GLRT implemented in the Localizer image analysis package respectively. An image of the raw, unreconstructed, data is shown for comparison. For TIRF and SD this is the average of 8 frames of data, for the simulated data a maximum intensity projection of the whole dataset is shown. (B) Graphs showing the percentage of mislocalized pixels in datasets reconstructed from a noisy background image. The noisy background image was generated by acquiring 5000 frames of images of a red fluorescent Perspex slide. Graph showing the predicted localization precision between all of the investigated algorithms. (C) Chart comparing retention of intensity information, processing speed, software interface and ease of use of the software of all the algorithms in the comparative study.
Figure 4.
Comparison between single channel SDSI super-resolution imaging and SIM imaging in a medial plane of HeLa cells.
(A) Comparison of SD, SDSI-dSTORM and SIM images of the mitotic spindle, the spindle was visualised using β-tubulin antibodies. Secondary Fab fragment antibodies conjugated to AlexaFluor647 were used for dSTORM imaging. SR data was generated using 3rd order SOFI, bar = 1 µm. (B) Comparison of SD, SDSI-dSTORM and SIM images of the mitochondria, the mitochondria were visualised by transfecting cells using the Dronpa-Mito construct for PALM imaging, SR data was generated using 3rd order SOFI, bar = 1 µm. (C) Line-scans (indicated in yellow parenthesis) through the mitotic spindle and mitochondria comparing SD resolution with SDSI and SIM.
Figure 5.
Multispectral super-resolution imaging of the nucleus.
(A) Point scanned confocal images showing orthogonal axial (x,z) view of Histone H3 (Red), LaminA/C (Green) and Nuclei (Blue). (B) Wide-field epifluorescent (WF) and spinning disk (SD) x,y and x,z images. Yellow line in x,z images indicates the plane shown in the x,y image of hetero-chromatin in the nucleus visualised using HP1α, bar = 2 µm. (C) dSTORM images of hetero-chromatin in the nucleus visualised using HP1α antibodies acquired by widefield epifluorescence (WF) and spinning disk confocal (SD). SR data from both WF and SD images was generated using 3rd order SOFI, bar = 1 µm. (D) Comparison of SD and SDSI images of hetero-chromatin visualised using HP1α antibodies and the nuclear membrane visualised using LaminA/C antibodies. Secondary FAb fragment antibodies conjugated to AlexaFluor555 and 647 were used for dSTORM imaging, SR data was generated using 3rd order SOFI, bar = 1 µm. Lower panel shows high-resolution region (indicated by orange box in upper panel) of heterochromatin, bar = 1 µm. (E) Intensity profiles, through regions indicated by yellow parenthesis in E, comparing SD resolution with SDSI.