Fluorescence microscopy with super-resolved optical sections Alexander Egner, Stefan W. Hell  Trends in Cell Biology  Volume 15, Issue 4, Pages 207-215 (April 2005) DOI: 10.1016/j.tcb.2005.02.003 Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure 1 Schematic representation of the focusing process: A laser beam propagates along the optical axis. The plane wavefront of the beam is converted by a lens into a segment of a spherical wavefront and therefore focused onto the focal point, which lies in the imaging plane. The three-dimensional extent of the focal spot results from the interference of the spherical wave with itself in the proximity of the focal point. As the focusing process is unidirectional (here from left to right), the focal spot is longer (axial direction) than wide (lateral direction). Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure 2 Exploiting counter-propagating coherent wavefronts in microscopy. The left-hand column (a1–d1) sketches the optical conditions of standing-wave microscopy (SWM), incoherent-illumination imaging microscopy (I5M), 4Pi microscopy of type A (4Pi-A) and of the type C counterpart of the latter (4Pi-C). In SWM, the object is excited with flat standing waves (blue) and the detection of fluorescence occurs in the regular epifluorescence mode (green). I5M uses flat standing waves for excitation (blue) and coherent detection (green) through both objective lenses of the spherical wavefronts emerging from the sample. In 4Pi microscopy, excitation is performed with interfering spherical wavefronts, usually in the two-photon excitation mode (red). In 4Pi microscopy of type A, the detection of the emerging fluorescence wavefront (green) is carried out without interference, usually by collecting the fluorescence with a single lens. In 4Pi microscopy of type C, the detection of the spherical fluorescence wavefronts is accomplished through both lenses so that the fluorescence light interferes at the same point of the detector. The center column (a2-d2) displays the effective point-spread function (PSF) of the corresponding microscopes, whereas the right-hand column (a3-d3) shows the corresponding amplitude of the OTF. The associated insert graphs show the profile of the PSF along the optic axes (x,y=0) as well as that of the OTF in the axial spatial frequency direction. For the 4Pi microscopes, two-photon excitation at 880nm and confocal detection at a fluorescence wavelength of 508nm have been assumed with a pinhole size of about one Airy disk. The OTFs have been calculated for a volume of 10μm thickness along the z-direction. The OTF of the SWM and the I5M feature a singularity at the frequency point kz=0μm−1and that of the SWM also at kz=± 25.75μm−1 corresponding to an excitation wavelength of 488nm. As a general rule, the smaller the side lobes of the PSFs, the larger is the resilience of the microscope against artifacts; a primary side lobe height of <0.5 is compulsory. Expressed in spatial frequencies: small lobes entail a weak modulation of the OTF. To avoid artifacts, the OTF must not consist of interrupted or only weakly connected regions. Note the >0.5 side lobe height in the PSF and the disconnection in the OTF of the SWM, indicating that the use of spherical wavefronts is mandatory for a genuine axial resolution improvement. Comparative analysis of all these systems demonstrates that the best performance is attained with the 4Pi microscope. Hence the axial resolution improvement actually results from the better coverage of the full solid focusing angle of 4π. Bars: b2, 0.25μm; b3, 25μm−1. Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure 3 Images of a pair of actin fibers in a fixed mouse fibroblast cell. While they cannot be distinguished over the whole field of view in the two-photon excitation confocal recording (a), the actin bundles appear distinct in the counterpart 4Pi recording in (b). The corresponding raw data shown in (c) produced by the two-photon 4Pi-confocal mode of type A still exhibit ghost images resulting from the side lobes. In (b) they are eliminated by a fast linear mathematical operation. Bar, 1μm. Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure 4 Surface-rendered 3D images of the GFP-labeled mitochondrial matrix of live budding yeast Saccharomyces cerevisiae: (a) wild-type and (b) fis1Δ mutants displaying mother and daughter cells. The cells were grown in a complete medium with glycerol, which results in a branched network. Bars, 1μm. Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure 5 Golgi apparatus of live Vero cells at ∼100nm 3D-resolution obtained by 4Pi imaging of (a) GalT–EGFP and (b) 2-OST–EGFP with subsequent nonlinear image restoration. The left-hand insets show regular epifluorescence images to correlate the Golgi apparatus with the nucleus. The blue Hoechst-counterstaining of the nucleus indicates that the cells were in interphase. The GalT-EGFP-transfected cell in (c) is most likely apoptotic; the corresponding 3D Golgi image pinpoints the ability of the 4Pi microscope to resolve small structures with virtually unilateral resolution. Thus, hollow cavities become apparent if the caps of the balloon-like structures are removed (d). The 4Pi microscope was operated in the two-photon excitation mode of type A with confocal detection. Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions

Figure I Trends in Cell Biology 2005 15, 207-215DOI: (10.1016/j.tcb.2005.02.003) Copyright © 2005 Elsevier Ltd Terms and Conditions