The phenomenon of diffraction has long been thought to set an inevitable physical limit to resolution of light microscopy. According to Abbe’s diffraction limit approximation, structures smaller than half the wavelength of light cannot be resolved. The last decades however, have seen developments in fluorescence microscopy that enable to study cellular structures on the nanometer length scale.

One of these approaches has been termed single molecule localization microscopy (SMLM). In 2006, different realizations of the technique were developed simultaneously: photoactivated localization microscopy (PALM), fluorescence photoactivation localization microscopy (fPALM) and stochastic optical reconstruction microscopy (STORM). More recent approaches include direct stochastic optical reconstruction microscopy (dSTORM) and point accumulation for imaging in nanoscale topography (PAINT), including DNA-PAINT (see further reading for information about SMLM techniques).

The basic idea of all SMLM techniques is to separate the signal of individual emitters in time, which allows to determine their positions with nanometer precision. This is typically achieved by exploiting stochastic blinking or binding phenomena, leaving only a sparse subset of molecules visible at a certain time point. Hence, for image acquisition thousands of individual frames need to be recorded. The raw data is subsequently analyzed, yielding a list of localization coordinates as a final result.

In practice, however, multiple parameters affect the quality of the obtained images. For example, molecular position can only be determined to a certain precision, yielding slightly jittered images of the true biomolecular structure. In addition, not all biomolecules carry a detectable label, and hence will be missed in the analysis. Next, the label may be connected to the biomolecule of interest via a linker, and hence does not adequately represent the biomolecule’s position. Finally, the biomolecule may be labelled with multiple dye molecules, each of which may be detected multiple times, giving rise to overcounting.

Researchers have been particularly intrigued by the possibility to determine the spatial distribution of biomolecules in their natural environment, in most cases the intact cell. Application of SMLM to various plasma membrane proteins revealed the presence of nanoclusters to different degrees. One problem inherent to SMLM, however, relates to multiple detection of the very same dye molecule, rendering the detection of protein clustering challenging. As a matter of fact, the SMLM image of a single protein molecule likely contains multiple localizations, and hence may easily be misinterpreted as a protein cluster.

We have developed two potential solutions to this problem: i) a label titration approach, in which SMLM images are analyzed and compared at different degrees of labeling; ii) a two-color SMLM approach (2-CLASTA), which allows for calculating a p-value for the null hypothesis that the experimental data set corresponds to an underlying biomolecular distribution, which is not significantly different from a completely random distribution.  

Label titration SMLM 

The approach is based on deliberate variations in the labeling density of the samples and subsequent quantitative cluster analysis. Unlike common protocols, we hence obtain information about molecular distributions from a series of experiments, in which samples are prepared at varying labeling densities. In particular, our method takes advantage of a characteristic dependence between the average label density within a presumable cluster and the relative area covered by those clusters, when the degree of labeling is varied. The method circumvents the problem of clustering artifacts generated by the blinking statistics of the fluorophores used. It can be readily applied to PALM and (d)STORM experiments where either overexpressed proteins are present over a broad range of expression levels or antibody concentrations are titrated to achieve different degrees of labeling.

2-CLASTA

2-Color Localization microscopy And Significance Testing Approach (2-CLASTA) provides a parameter-free statistical framework for the qualitative analysis of two-dimensional SMLM data via significance testing methods. 2-CLASTA yields p-values for the null hypothesis of random biomolecular distributions, independent of the blinking behavior of the chosen fluorescent labels. The method is parameter-free and does not require any additional measurements nor grouping of localizations.

The idea is to target the same biomolecule of interest with two different fluorescent labels, and determine the localizations in the respective color channels. The program calculates the nearest neighbor distances between them. Next, it compares the nearest neighbor distances for the recorded data with the distances from a random distribution of biomolecules calculated from the measured data. As an output, the method provides a p-value for the null hypothesis that the experimental data set corresponds to an underlying biomolecular distribution, which is not significantly different from a completely random distribution as described by a spatial Poisson process. In this respect, 2-CLASTA differs from existing quantitative approaches, which typically aim at determining quantitative parameters before actually testing the mere presence of biomolecular clusters.

Our key publications

Verifying molecular clusters by 2-color localization microscopy and significance testing

Arnold, A. M., M. C. Schneider, C. Hüsson, R. Sablatnig, M. Brameshuber, F. Baumgart, and G. J. Schütz. 2020. Verifying molecular clusters by 2-color localization microscopy and significance testing. Scientific Reports 10(1):4230, opens an external URL in a new window.

The paper introduces a cluster detection approach based on significance testing of 2-color SMLM images. It provides an ImageJ plugin for straightforward data analysis

What we talk about when we talk about nanoclusters

Baumgart, F., A. Arnold, B. Rossboth, M. Brameshuber, and G. J. Schütz. 2019. What we talk about when we talk about nanoclusters. Methods and Applications in Fluorescence 7(1):013001, opens an external URL in a new window.

A review article put into the setting of a virtual conference, in which a group of researchers discuss various aspects of SMLM.

TCRs are randomly distributed on the plasma membrane of resting antigen-experienced T cells

Rossboth, B., A. M. Arnold, H. Ta, R. Platzer, F. Kellner, J. B. Huppa, M. Brameshuber, F. Baumgart, and G. J. Schütz. 2018. TCRs are randomly distributed on the plasma membrane of resting antigen-experienced T cells. Nature Immunology 19(8):821-827, opens an external URL in a new window.

Using both label-titration SMLM and STED microscopy we did not detect presumed clusters of the T cell receptor under non-activating conditions. 

Varying label density allows artifact-free analysis of membrane-protein nanoclusters

Baumgart, F., A. M. Arnold, K. Leskovar, K. Staszek, M. Fölser, J. Weghuber, H. Stockinger, and G. J. Schütz. 2016. Varying label density allows artifact-free analysis of membrane-protein nanoclusters. Nat Meth 13(8):661-664, opens an external URL in a new window.

We proposed to analyze and compare SMLM images obtained at different label titrations to identify scenarios of true protein clustering.

Superresolution microscopy reveals spatial separation of UCP4 and F0F1-ATP synthase in neuronal mitochondria

Klotzsch, E., A. Smorodchenko, L. Löfler, R. Moldzio, E. Parkinson, G. J. Schütz, and E. E. Pohl. 2015. Superresolution microscopy reveals spatial separation of UCP4 and F0F1-ATP synthase in neuronal mitochondria. Proc Natl Acad Sci U S A 112(1):130-135, opens an external URL in a new window.

Our first SMLM paper, in which we studied the nanoscale distribution of UCP4, a mitochondrial uncoupling protein.

Further reading

Visualizing and discovering cellular structures with super-resolution microscopy

Sigal, Y. M., R. Zhou, and X. Zhuang. 2018. Visualizing and discovering cellular structures with super-resolution microscopy. Science 361:880-887, opens an external URL in a new window.

Recent Review over the state of-the-art of SMLM. Contains a lot of examples and applications. 

Biological Insight from Super-Resolution Microscopy: What We Can Learn from Localization-Based Images

Baddeley, D., and J. Bewersdorf. 2018. Biological Insight from Super-Resolution Microscopy: What We Can Learn from Localization-Based Images. Annual Review of Biochemistry 87(1):965-989, opens an external URL in a new window.

This review article focusses on data interpretation/resolution.

Imaging Intracellular Fluorescent Proteins at Nanometer Resolution

Betzig, E., G. H. Patterson, R. Sougrat, O. W. Lindwasser, S. Olenych, J. S. Bonifacino, M. W. Davidson, J. Lippincott-Schwartz, and H. F. Hess. 2006. Imaging Intracellular Fluorescent Proteins at Nanometer Resolution. Science 313:1642-1645, opens an external URL in a new window.

One of the first three papers on SMLM, which were published quasi simultaneously.

Ultra-high resolution imaging by fluorescence photoactivation localization microscopy

Hess, S. T., T. P. Girirajan, and M. D. Mason. 2006. Ultra-high resolution imaging by fluorescence photoactivation localization microscopy. Biophys J 91(11):4258-4272, opens an external URL in a new window.

One of the first three papers on SMLM, which were published quasi simultaneously.

Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM)

Rust, M., M. Bates, and X. Zhuang. 2006. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat Methods 3(10):793-795, opens an external URL in a new window.

One of the first three papers on SMLM, which were published quasi simultaneously.