There Is More to the Picture Than Meets the Eye

Optical Imaging beyond the Classical Diffraction Limit

  • Fig. 1: The concept of single molecule-based superresolution microcopy like PALM and (d)STORM. (a) In a diffraction limited image, fluorophores that are closer than the diffraction limit cannot be resolved, because the signals of the individual fluorophores overlap. (b) Single molecule-based superresolution microscopy uses stochastic blinking of fluorophores to separate the fluorescent signals of individual molecules over time. After determining the positions of individual fluorophores, a localization map can be reconstructed.Fig. 1: The concept of single molecule-based superresolution microcopy like PALM and (d)STORM. (a) In a diffraction limited image, fluorophores that are closer than the diffraction limit cannot be resolved, because the signals of the individual fluorophores overlap. (b) Single molecule-based superresolution microscopy uses stochastic blinking of fluorophores to separate the fluorescent signals of individual molecules over time. After determining the positions of individual fluorophores, a localization map can be reconstructed.
  • Fig. 1: The concept of single molecule-based superresolution microcopy like PALM and (d)STORM. (a) In a diffraction limited image, fluorophores that are closer than the diffraction limit cannot be resolved, because the signals of the individual fluorophores overlap. (b) Single molecule-based superresolution microscopy uses stochastic blinking of fluorophores to separate the fluorescent signals of individual molecules over time. After determining the positions of individual fluorophores, a localization map can be reconstructed.
  • Fig. 2: Varying label density allows distinguishing clustered from random distributions. Fluorescently labeled antibodies are used to image molecules of interest. Each fluorescent label may be imaged multiple times, due to fluorophore blinking. (a) For randomly distributed molecules, each detected cluster of localizations is the result of repeated detections. The number of localizations per cluster thus stays constant with increasing concentration of fluorescent label. (b) For clustered molecules, the number of localizations per detected cluster increases at higher label densities.

For centuries diffraction was believed to set the fundamental limit to optical microscopy, so that structures below about half the wavelength of light cannot be resolved. In the last ten years, however, a variety of approaches has been developed, which enable optical imaging way beyond the classical diffraction limit [1].
All of them make use of peculiar properties of fluorophores, such as intensity-dependent saturation of electronic transitions in Stimulated Emission Depletion (STED) [2] or Structured Illumination Microscopy (SIM) [3], or the switching of fluorophores between bright and dark states in Photoactivated Localization Microscopy (PALM) [4] or (direct) Stochastic Optical Reconstruction Microscopy ((d)STORM) [5, 6].

Strictly speaking, diffraction only limits the ability to establish whether multiple point objects contribute to one visible feature in an obtained image, and to determine the positions of these objects. Experimentally, in fluorescence microscopy molecules of interest are visualized by attaching fluorescent dyes to them. If all fluorophores lighted up at the same time, it would be impossible to resolve molecules that are closer than ~250-300 nm due to the overlapping signals. To circumvent this problem, single molecule-based superresolution techniques such as PALM or (d)STORM rely on chemically induced stochastic blinking of fluorophores between an active on-state and a dark off-state. In a situation where most fluorophores are in the off-state and only a sparse subset is in the on-state, the active molecules can be imaged as well-separated individual spots. It is now rather easy to determine the center of each spot by fitting the signal with e.g. a Gaussian function, which can be achieved with an accuracy much below the actual width of the spot; localization precisions of 20-30 nm are routinely achieved. Repeating this process for thousands of image frames, virtually all molecules can be switched at least once to their active state. Hence, localization maps can be reconstructed from all fitted positions, yielding superresolution images with nanoscopic detail (fig. 1).

Single molecule-based superresolution is also extensively used in the authors’ research group.

Their central interest is to get a fundamental understanding of the plasma membrane of mammalian cells, in particular cells of the immune system. Superresolution microscopy has also been used by a number of other research groups to get detailed insights in the organization of plasma membrane proteins. The remarkable common finding of those studies was that virtually any studied protein seemed to be organized in nanoscopic clusters. In light of those findings, the group recently developed a new experimental strategy to probe single molecule-based superresolution microscopy data for nanoclustering [7]. The new method is insensitive to common artifacts of single molecule-based superresolution techniques. The results also challenge current paradigms of how proteins are organized on the cell plasma membrane.

Initially, a member of the research group decided to follow up recent reports about nanoscopic clusters of signaling proteins at the plasma membrane of T-cells, a cell type that is central to the adaptive immune system. The idea was to implement PALM and (d)STORM to get an understanding of the biological significance of molecular clustering in the T cell plasma membrane. This was based on the hypothesis that nanoclusters of signaling proteins in immune cells were an essential structural feature that was crucial for immune signaling [8]. Understanding nanoclustering of T-lymphocyte signaling proteins therefore seemed a promising goal to pursue.

Soon, however, things turned out to be a little less clear. As described above, one of the inherent features of PALM and (d)STORM is that those approaches exploit stochastic blinking of fluorophores to determine the exact positions of target molecules. Unfortunately, the induced blinking can lead to difficulties when numbers of localizations need to be correlated with underlying numbers of molecules. Exactly that, however, is necessary to make statements about molecular clustering. The central question is if clusters of localizations are derived from clusters of molecules or if they are multiple observations of one single blinking emitter.

After thorough analysis of the data, it was suspected that what had been reported as protein nano-clusters of one of the important signaling proteins in T-cells, could actually be an imaging artifact due to multiple observations of the very same fluorescent molecules. Therefore it was necessary to come up with a strategy that would circumvent overcounting artifacts. A sample-based approach was developed, where deliberate variations of the labeling density during the sample preparation yield characteristic differences in the statistics of superresolution localization maps for clustered versus random protein distributions. In case of real protein clusters, increasing the degree of fluorescent labeling leads to an increased number of localizations per clustered area. In contrast, for randomly distributed proteins each cluster of localizations originates from a blinking fluorophore that is detected multiple times. In the latter case, the number of localizations per detected cluster remains constant at increasing degrees of labeling, since it depends exclusively on the blinking statistics of the fluorophore used (fig. 2).

The new method turned out to work very robustly both in synthetic settings and on cell samples and was immediately used to probe for nanoclustering of signaling proteins on immune cells. Interestingly, most of the signaling proteins in T cells seem to be randomly distributed on the plasma membrane, albeit there are a few exceptions. Nevertheless, the hypothesis that protein nanoclustering is a general organizing principle of the plasma membrane needs to be revised.

Authors
Florian Baumgart, Andreas M. Arnold, Gerhard J. Schütz
Institute of Applied Physics, Vienna University of Technology, Wien, Austria

Contact
Dr. Florian Baumgart
Institute of Applied Physics
Vienna University of Technology
Wien, Austria
baumgart@iap.tuwien.ac.at

References
[1] L. Schermelleh, R. Heintzmann, H. Leonhardt, A guide to super-resolution fluorescence microscopy, J. Cell Biol. 190, 165-175 (2010) – DOI: 10.1083/jcb.201002018

[2] T. A. Klar et al., Fluorescence microscopy with diffraction resolution barrier broken by stimulated emission, Proc. Natl. Acad. Sci. USA 97, 8206-8210 (2000) – DOI: 10.1073/pnas.97.15.8206

[3] M. G. Gustafsson, Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution, Proc. Natl. Acad. Sci. USA 102, 13081-13086 (2005) – DOI: 10.1073/pnas.0406877102

[4] E. Betzig, et al., Imaging Intracellular Fluorescent Proteins at Nanometer Resolution, Science 313, 1642-1645 (2006) – DOI: 10.1126/science.1127344

[5] M. Rust, M. Bates, X. Zhuang, Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM), Nat. Methods 3, 793-795 (2006) – DOI: 10.1038/nmeth929

[6] M. Heilemann, et al., Subdiffraction-resolution fluorescence imaging with conventional fluorescent probes, Angew. Chem. Int. Ed. Engl. 47, 6172-6176 (2008) – DOI: 10.1002/anie.200802376

[7] F. Baumgart, et al., Varying label density allows artifact-free analysis of membrane-protein nanoclusters, Nat. Methods (2016) – DOI: 10.1038/nmeth.3897

[8] M. F. Garcia-Parajo, et al., Nanoclustering as a dominant feature of plasma membrane organization, J. Cell Sci. 127, 4995-5005 (2014) – DOI: 10.1242/jcs.146340

More articles on drug discovery: http://www.laboratory-journal.com/

Register now!

The latest information directly via newsletter.

To prevent automated spam submissions leave this field empty.