Abstracts of the 5th European Super-Resolution User-Club Meeting

Kees Jalink, The Netherlands Cancer Institute, Amsterdam, The Netherlands

For centuries, the light microscope has played a pivotal role in biology and medicine. Without the ability to observe cells and tissues, to study their morphology, dynamics and molecular composition, we would have only a fraction of the understanding of life we have today. Yet, until recently, we failed to have the resolution to observe the building blocks of life, as the resolution of light  microscopy is about 50-fold too low. Super-resolution microscopy is rapidly changing that. We use Ground State Depletion Imaging to study the composition of the cytoskeleton and to decipher signal transduction mechanisms.

As a first step, we have invested in optimizing preparation and recording conditions. Blinking conditions have been optimized and algorithms were developed for drift correction and background subtraction. We are currently applying SR-GSD in a number of biological studies. In collaboration with the Sonnenberg lab in our institute, we have focussed on the structure of hemidesmosomes, adhesive structures within the skin. We also study keratin filament formation and plasticity in collaboration with Dr. Reinhard Windhoffer in Aachen, Germany. In my talk, I will update you on developments in our lab and focus on what will be – I think – the important directions for the near future of localization microscopy.

Luca Lanzanó, Italian Institute of Technology, Genoa, Italy

The visualization at the nanoscale level inside cells has become a fundamental need in molecular biology. Most of the super-resolution fluorescence methods developed so far provide sub-diffraction resolution by transiently switching the fluorophores between optically distinguishable states. Accordingly, the different super-resolution concepts have been broadly classified into stochastic and targeted switching methods, based on how the fluorophore states are manipulated [1]. However, sub-diffraction resolution can also be obtained with a very different strategy: by encoding spatial information into a temporal channel and decoding it after the transmission through the microscope setup [2]. The nanoscale spatial distribution of the molecules inside the detection volume of the microscope can be encoded within the fluorescence dynamics and can be decoded by resolving the signal into its dynamics components. Thus the challenge of increasing the spatial resolution of an optical microscope is translated into the spectroscopy task of resolving the signal into its dynamics components. We present here a robust and general method to spatially sort the fluorescence photons on the basis of the associated molecular dynamics [3]. In this method, the separation of dynamics components is obtained without any fitting procedure using phasor analysis [4]. In a specific implementation, we consider the gradients in fluorescence lifetime generated by stimulated emission in a CW-STED microscope [5]. We show that spatial resolution can be increased indefinitely by increasing the number of resolved components up to a maximum, predictable number, determined by the amount of noise. Interestingly the method also isolates any uncorrelated background signal. We demonstrate that this spectroscopy-based approach provides background-free nanoscale imaging of subcellular structures, opening new routes in super-resolution microscopy based on the encoding of spatial information through manipulation of molecular dynamics [3].

Imre Gaspar, Developmental Biology Unit, EMBL, Heidelberg, Germany

In vast majority of the studied animal species, body axes are determined already during oogenesis through localization and localized translation of messenger RNA molecules. These mRNAs contain specific structures (cis-acting elements) usually - but not limited to – in their 3’ untranslated region which recruit specific protein molecules (trans-acting elements) resulting in formation of functional ribonucleoprotein complexes (mRNPs) [1]. Composition of mRNPs, however, changes dynamically from translation to decay [1], making microscopic techniques with high spatial and temporal resolution invaluable for studying mRNP biogenesis.

During Drosophila oogenesis, oskar mRNPs are synthesized in the nurse cells from where they get transported into the interconnected oocyte by cytoplasmic dynein. Within the oocyte, oskar mRNA localizes to the posterior pole initially in a kinesin-1 dependent manner during mid-oogenesis [2]. To understand the formation of transport and localization competent oskar mRNPs, we chose a candidate based co-localization assay using confocal and STED microscopy. To maximize specificity and  minimize the distance of label to oskar mRNA, we used forced intercalation of thiazole orange (FIT) probes that only fluoresce when binding to target [3]. FIT probes allow performing "mild" in situ hybridization that preserves auto-fluorescence of GFP tagged proteins. Also, while the emission and excitation spectrum of TO is sufficiently different (red-shifted) from GFP to acquire bleed-through free images, its emission can be stimulated with 592 nm without any re-excitation.

We found that oskar mRNPs associate with kinesin-1 but not with dynein right after export into the nurse cell cytoplasm. Interestingly, association with kinesin-1 requires an atypical tropomyosin protein encoded by the dTropomyosin1 gene and also splicing of the first intron of oskar mRNA [4].

Luc Reymond, Laboratory of Protein Engineering, École polytechnique fédérale, Lausanne, Switzerland

The ideal fluorescent probe for super-resolution imaging is bright, absorbs at long wavelengths and can be implemented flexibly in living cells and in vivo. However, the design of synthetic probes that combine all of these properties has proved to be extremely difficult. Here, we introduce a suite of silicon rhodamine based probes that combine far-red emission, fluorogenicity, minimal cytotoxicity, excellent brightness and photostability for fluorescence imaging of cellular structures in living cells.  Applied in stimulated emission depletion (STED) nanoscopy, they allowed imaging of cellular structures in living cells at an unprecedented resolution. The current status of the probes development will be presented together with their applications in live cell imaging.

Leila Nahidi Azar, Netherlands Cancer Institute, Amsterdam, The Netherlands

Hemidesmosomes have been extensively studied by immunofluorescence microscopy, but due to its limited resolution, the precise organization of hemidesmosomes remained poorly understood. We studied hemidesmosome organization in cultured keratinocytes by highly corrected 2- and 3-color Super-Resolution microscopy. We observed that in the cell periphery, nascent hemidesmosomes are associated with individual keratin filaments. By quantifying molecular distances, we demonstrate that both linker proteins BP230 and plectin interact asymmetrically with keratin. Furthermore, we show that BP180 and BP230 have a characteristic arrangement within these hemidesmosomes with BP180 molecules surrounding a central core of BP230 molecules. In my presentation I will emphasize how novel image analysis methods have to be developed to replace concepts such as colocalization, which are of limited value for SR imaging.

Robert Nieuwenhuizen, Delft University of Technology, Department of Imaging Physics, Quantitative Imaging Group, Delft, The Netherlands

Cytoskeletal protein networks serve many crucial roles in living cells. To understand how the different filament networks interact and collaborate to perform these functions, it is important to obtain insight into the nanoscale spatial arrangements of these networks. Super-resolution microscopy techniques now make it possible to resolve these networks. This introduces the need for new quantitative tools to interrogate the organization of and mutual interrelations between the different cytoskeletal elements.

We present a new quantitative framework in which we determine how the orientational alignment of filaments affects their colocalization (as measured by the pair cross-correlation function). The simultaneous orientational alignment and colocalization of filaments will be referred to as co-orientation. We show how the strength of  the co-orientation can be quantified locally by an anisotropic generalization of Ripley’s K-coefficient and we propose a test for its statistical significance. We demonstrate our methods on simulated localization microscopy data of filament structures, as well as multi-color GSDIM images of filamentous structures.

Daniela Leyton Puig, University of Manchester, Manchester Collaborative Centre for Inflammation Research, Manchester, UK

Cell surface receptors relay information from extracellular messenger molecules to the cell interior. Following activation, many receptors can rapidly form into small clusters that, in most cases, eventually are being internalized by the cells. Some receptor types are known to continue signaling intracellularly, but for most receptors, internalization uncouples them from the second messenger cascades. I have studied receptor clustering and the clathrin-mediated uptake of receptors by the cell using GSDIM. After a general introduction, I will describe my experiments and theoretical work aiming at quantification of receptor clustering at the plasma membrane.

David Williamson, University of Manchester, Manchester Collaborative Centre for Inflammation Research, Manchester, UK

Single-molecule localisation microscopy (SMLM) images present unique challenges compared to conventional pixel-based microscopy images. Much work has been directed to approaches and algorithms used to generate SMLM images. However, there has been less focus on the subsequent extraction and interpretation of useful information from these images, yet it is these data that are crucial to drawing biologically meaningful conclusions from SMLM images.

Here we show how information on the spatial arrangement of proteins in the plasma membrane can be routinely quantified by examining the density of molecules at different scales, formally described by Ripley's K function and the related Getis & Franklin's L function. These methods are extended to deliver information on the "co-clustering" of different proteins at high resolution and to 3D images. A novel method currently under development utilises the distances to a molecule's n nearest neighbours to derive a "clustering" value and avoids some of the problems inherent in other clustering approaches.

Jonas Ries, CBB, EMBL, Heidelberg, Germany

Single-molecule localization based super-resolution microscopy (SMLM) nowadays reaches a resolution sufficient to determine structures of protein assemblies in the cellular context. It is therefore a technique complementary to classical structural tech-niques such as x-ray crystallography or electron microscopy to investigate, how molecular machines are organized.

In this talk, I will introduce a new method for isotropic 3D SMLM, I will present two alternative labelling schemes for SMLM and I will report on our progress towards resolving a fundamental multi-protein machinery on the nanometer scale, namely the endocytotic machinery in S. cerevisiae.

Supercritical angle localization microscopy (SALM) is a 3D SMLM technique based on the principle of surface-generated fluorescence.

This near-field fluorescence is strongly dependent on the distance of fluorophores from the coverslip and can therefore be used to estimate their axial positions. We established a robust and simple implementation of supercritical angle fluorescence detection for single-molecule localization microscopy, calibrated it using fluorescent bead samples, validated the method with DNA origami tetrahedra, and present proof-of-principle data on biological samples.

Localization microscopy requires a high degree of labelling with bright and switchable dyes. Until now however, this required special fluorescent proteins to be cloned or high-affinity antibodies to be generated for specific labelling. On the other hand, many laboratories will have most of their constructs in GFP form and entire genomes are available as functional GFP-fusion proteins.

Here, we report a method that makes all these constructs available for superresolution microscopy by targeting GFP with tiny, high-affinity antibodies coupled to blinking dyes. It thus combines the molecular specificity of genetic tagging with the high photon yield of organic dyes and minimal linkage error. We show that in combination with GFP-libraries, virtually any known protein can immediately be used in super-resolution microscopy and that high-throughput super-resolution imaging using simplified labelling schemes is possible.

As an alternative to using photo-switchable fluorophores, we introduce binding-activated localization microscopy (BALM), which employs fluorescence enhancement of fluorogenic dyes upon binding to target structures for super-resolution microscopy.

We used this approach to study DNA structures and a-synuclein amyloids and could demonstrate a superb labelling density combined with a very high resolution.

Endocytosis is a highly intricate cellular process, which in yeast involves the ordered recruitment and disassembly of around 60 proteins. Our current efforts focus on understanding the intermediate and late coat assembly preceding scission. Here, we were able to reveal sub-diffraction features regarding shape and structure of endocytic coat proteins that were previously inaccessible. By visualizing many proteins pairs with dual-color super-resolution microscopy, we are pursuing to obtain a comprehensive structural picture of the endocytic proteome.

Jürgen J. Schmied, Technical University of Braunschweig, Braunschweig, Germany

The field of fluorescence microscopy achieved tremendous improvements in resolution by the emergence of novel techniques breaking the diffraction limit, summarized under the term super-resolution microscopy. Now resolutions down to 20 nm and less can be achieved but validating and testing those super-resolution microscopes remains challenging due to the lack of precise and nano-scaled fluorescence standards. The gap between top-down and bottom-up techniques – ranging from a few to a few hundred nanometers – could be bridged by means of DNA nanotechnology allowing to specifically design molecular structures with programmed shapes and functionality. The technique behind is called DNA origami and offers the possibility to precisely attach fluorescent dyes at preassigned positions and subsequently establish defined point-to-point distances of fluorophores [1, 2]. The flexibility and versatility in the design of DNA origami-based microscopy standards makes them ideally suited for the broad variety of emerging super-resolution microscopy methods. As DNA origami structures are durable and portable, they can become a universally available specimen to check the everyday functionality of a microscope. For instance, nanoscopic rulers qualify for the quantitative determination of resolution in a statistically firm and reproducible way and allow for the comparison of the resolution of different super-resolution techniques. Rulers are developed for different applications such as STED and SIM as well as localization-based super-resolution microscopy in 2D and 3D.

Ivan Michel Antolovic, TU Delft, Dept. of Microelectronics, Circuits & Systems, Delft, The Netherlands

Localization super resolution microscopy (GSDIM, STORM, and other techniques) is based on accurately determining the center of sparsely activated point spread functions of single fluorophores. Since the number of detected photons determines image resolution, the cameras of choice have been EMCCDs because of their high quantum efficiency and built-in electron amplification.

Alternatively, single-photon avalanche diode (SPAD) imagers can provide even faster frame rates and zero readout noise. We used a 1-bit 512 × 128 SPAD imager, called SwissSPAD, which enables a frame of 6.4 μs [1]. Variable sequences of 1-bit frames can be used to form gray level images of programmable resolution, while keeping the same time resolution information. The sensitivity of the imager, characterized as photon detection efficiency (PDE) has been boosted 12x using a microlens array mounted directly on the sensor [2].

We measured multicolor intensity florescence with SwissSPAD and compared its performance to a commercial EMCCD camera. Furthermore, the higher sensitivity enabled us to detect single fluorophores as required by GSDIM and to reconstruct super resolution images, while the 6.4 μs time resolution led to unprecedented accuracy in analysing the blinking effects of the fluorophore independently in each pixel.