Refine
Has Fulltext
- yes (17)
Is part of the Bibliography
- yes (17)
Year of publication
Document Type
- Doctoral Thesis (8)
- Journal article (7)
- Master Thesis (1)
- Report (1)
Language
- English (17) (remove)
Keywords
- dSTORM (17) (remove)
Institute
- Theodor-Boveri-Institut für Biowissenschaften (13)
- Center for Computational and Theoretical Biology (2)
- Physikalisches Institut (2)
- Physiologisches Institut (2)
- Fakultät für Physik und Astronomie (1)
- Graduate School of Life Sciences (1)
- Institut für Klinische Neurobiologie (1)
- Institut für Theoretische Physik und Astrophysik (1)
- Klinik und Poliklinik für Unfall-, Hand-, Plastische und Wiederherstellungschirurgie (Chirurgische Klinik II) (1)
- Neurochirurgische Klinik und Poliklinik (1)
EU-Project number / Contract (GA) number
- 835102) (1)
Single molecule localization microscopy has seen a remarkable growth since its first
experimental implementations about a decade ago. Despite its technical challenges,
it is already widely used in medicine and biology and is valued as a unique tool
to gain molecular information with high specificity. However, common illumination techniques do not allow the use of single molecule sensitive super-resolution
microscopy techniques such as direct stochastic optical reconstruction microscopy
(dSTORM) for whole cell imaging. In addition, they can potentially alter the
quantitative information.
In this thesis, I combine dSTORM imaging in three dimensions with lattice lightsheet illumination to gain quantitative molecular information from cells unperturbed by the illumination and cover slip effects. Lattice light-sheet illumination
uses optical lattices for beam shaping to restrict the illumination to the detectable
volume. I describe the theoretical background needed for both techniques and detail
the experimental realization of the system as well as the software that I developed
to efficiently evaluate the data.
Eventually, I will present key datasets that demonstrate the capabilities of the
developed microscope system with and without dSTORM. My main goal here was
to use these techniques for imaging the neural cell adhesion molecule (NCAM, also
known as CD56) in whole cells. NCAM is a plasma membrane receptor known to
play a key role in biological processes such as memory and learning. Combining
dSTORM and lattice light-sheet illumination enables the collection of quantitative
data of the distribution of molecules across the whole plasma membrane, and shows
an accumulation of NCAM at cell-cell interfaces. The low phototoxicity of lattice
light-sheet illumination further allows for tracking individual NCAM dimers in living cells, showing a significant dependence of its mobility on the actin skeleton of
the cell.
Localization microscopy is a class of super-resolution fluorescence microscopy techniques. Localization microscopy methods are characterized by stochastic temporal isolation of fluorophore emission, i.e., making the fluorophores blink so rapidly that no two are
likely to be photoactive at the same time close to each other. Well-known localization microscopy methods include dSTORM}, STORM, PALM, FPALM, or GSDIM. The biological community has taken great interest in localization microscopy, since it can enhance the resolution of common fluorescence microscopy by an order of magnitude at little experimental cost.
However, localization microscopy has considerable computational cost since millions of individual stochastic emissions must be located with nanometer precision. The computational cost of this evaluation, and the organizational cost of implementing the complex algorithms, has impeded adoption of super-resolution microscopy for a long time.
In this work, I describe my algorithmic framework for evaluating localization microscopy data.
I demonstrate how my novel open-source software achieves real-time data evaluation, i.e., can evaluate data faster than the common experimental setups can capture them.
I show how this speed is attained on standard consumer-grade CPUs, removing the need for computing on expensive clusters or deploying graphics processing units.
The evaluation is performed with the widely accepted Gaussian PSF model and a Poissonian maximum-likelihood noise model.
I extend the computational model to show how robust, optimal two-color evaluation is realized, allowing correlative microscopy between multiple proteins or structures. By employing cubic B-splines, I show how the evaluation of three-dimensional samples can be made simple and robust, taking an important step towards precise imaging of micrometer-thick samples.
I uncover the behavior and limits of localization algorithms in the face of increasing emission densities.
Finally, I show up algorithms to extend localization microscopy to common biological problems.
I investigate cellular movement and motility by considering the in vitro movement of myosin-actin filaments. I show how SNAP-tag fusion proteins enable imaging with bright and stable organic fluorophores in live cells. By analyzing the internal structure of protein clusters, I show how localization microscopy can provide new quantitative approaches beyond pure imaging.
The development of cellular life on earth is coupled to the formation of lipid-based biological membranes. Although many tools to analyze their biophysical properties already exist, their variety and number is still relatively small compared to the field of protein studies. One reason for this, is their small size and complex assembly into an asymmetric tightly packed lipid bilayer showing characteristics of a two-dimensional heterogenous fluid. Since membranes are capable to form dynamic, nanoscopic domains, enriched in sphingolipids and cholesterol, their detailed investigation is limited to techniques which access information below the diffraction limit of light. In this work, I aimed to extend, optimize and compare three different labeling approaches for sphingolipids and their subsequent analysis by the single-molecule localization microscopy (SMLM) technique direct stochastic optical reconstruction microscopy (dSTORM). First, I applied classical immunofluorescence by immunoglobulin G (IgG) antibody labeling to detect and quantify sphingolipid nanodomains in the plasma membrane of eukaryotic cells. I was able to identify and characterize ceramide-rich platforms (CRPs) with a size of ~ 75nm on the basal and apical membrane of different cell lines. Next, I used click-chemistry to characterize sphingolipid analogs in living and fixed cells. By using a combination of fluorescence microscopy and anisotropy experiments, I analyzed their accessibility and configuration in the plasma membrane, respectively. Azide-modified, short fatty acid side chains, were accessible to membrane impermeable dyes and localized outside the hydrophobic membrane core. In contrast, azide moieties at the end of longer fatty acid side chains were less accessible and conjugated dyes localized deeper within the plasma membrane. By introducing photo-crosslinkable diazirine groups or chemically addressable amine groups, I developed methods to improve their immobilization required for dSTORM. Finally, I harnessed the specific binding characteristics of non-toxic shiga toxin B subunits (STxBs) and cholera toxin B subunits (CTxBs) to label and quantify glycosphingolipid nanodomains in the context of Neisseria meningitidis infection. Under pyhsiological conditions, these glycosphingolipids were distributed homogenously in the plasma membrane but upon bacterial infection CTxB detectable gangliosides accumulated around invasive Neisseria meningitidis. I was able to highlight the importance of cell cycle dependent glycosphingolipid expression for the invasion process. Blocking membrane accessible sugar headgroups by pretreatment with CTxB significantly reduced the number of invasive bacteria which confirmed the importance of gangliosides for bacterial uptake into cells. Based on my results, it can be concluded that labeling of sphingolipids should be carefully optimized depending on the research question and applied microscopy technique. In particular, I was able to develop new tools and protocols which enable the characterization of sphingolipid nanodomains by dSTORM for all three labeling approaches.
Background
Localization-based super-resolution microscopy resolves macromolecular structures down to a few nanometers by computationally reconstructing fluorescent emitter coordinates from diffraction-limited spots. The most commonly used algorithms are based on fitting parametric models of the point spread function (PSF) to a measured photon distribution. These algorithms make assumptions about the symmetry of the PSF and thus, do not work well with irregular, non-linear PSFs that occur for example in confocal lifetime imaging, where a laser is scanned across the sample. An alternative method for reconstructing sparse emitter sets from noisy, diffraction-limited images is compressed sensing, but due to its high computational cost it has not yet been widely adopted. Deep neural network fitters have recently emerged as a new competitive method for localization microscopy. They can learn to fit arbitrary PSFs, but require extensive simulated training data and do not generalize well. A method to efficiently fit the irregular PSFs from confocal lifetime localization microscopy combining the advantages of deep learning and compressed sensing would greatly improve the acquisition speed and throughput of this method.
Results
Here we introduce ReCSAI, a compressed sensing neural network to reconstruct localizations for confocal dSTORM, together with a simulation tool to generate training data. We implemented and compared different artificial network architectures, aiming to combine the advantages of compressed sensing and deep learning. We found that a U-Net with a recursive structure inspired by iterative compressed sensing showed the best results on realistic simulated datasets with noise, as well as on real experimentally measured confocal lifetime scanning data. Adding a trainable wavelet denoising layer as prior step further improved the reconstruction quality.
Conclusions
Our deep learning approach can reach a similar reconstruction accuracy for confocal dSTORM as frame binning with traditional fitting without requiring the acquisition of multiple frames. In addition, our work offers generic insights on the reconstruction of sparse measurements from noisy experimental data by combining compressed sensing and deep learning. We provide the trained networks, the code for network training and inference as well as the simulation tool as python code and Jupyter notebooks for easy reproducibility.
The fusion of methods from several disciplines is a crucial component of scientific development. Artificial Neural Networks, based on the principle of biological neuronal networks, demonstrate how nature provides the best templates for technological advancement. These innovations can then be employed to solve the remaining mysteries of biology, including, in particular, processes that take place on microscopic scales and can only be studied with sophisticated techniques. For instance, direct Stochastic Optical Reconstruction Microscopy combines tools from chemistry, physics, and computer science to visualize biological processes at the molecular level. One of the key components is the computer-aided reconstruction of super-resolved images. Improving the corresponding algorithms increases the quality of the generated data, providing further insights into our biology. It is important, however, to ensure that the heavily processed images are still a reflection of reality and do not originate in random artefacts.
Expansion microscopy is expanding the sample by embedding it in a swellable hydrogel. The method can be combined with other super-resolution techniques to gain additional resolution. We tested this approach on microtubules, a well-known filamentous reference structure, to evaluate the performance of different protocols and labelling techniques.
We developed LineProfiler an objective tool for data collection. Instead of collecting perpendicular profiles in small areas, the software gathers line profiles from filamentous structures of the entire image. This improves data quantity, quality and prevents a biased choice of the evaluated regions. On the basis of the collected data, we deployed theoretical models of the expected intensity distribution across the filaments. This led to the conclusion that post-expansion labelling significantly reduces the labelling error and thus, improves the data quality. The software was further used to determine the expansion factor and arrangement of synaptonemal complex data.
Automated Simple Elastix uses state-of-the-art image alignment to compare pre- and post-expansion images. It corrects linear distortions occurring under isotropic expansion, calculates a structural expansion factor and highlights structural mismatches in a distortion map. We used the software to evaluate expanded fungi and NK cells. We found that the expansion factor differs for the two structures and is lower than the overall expansion of the hydrogel.
Assessing the fluorescence lifetime of emitters used for direct Stochastic Optical Reconstruction Microscopy can reveal additional information about the molecular environment or distinguish dyes emitting with a similar wavelength. The corresponding measurements require a confocal scanning of the sample in combination with the fluorescent switching of the underlying emitters. This leads to non-linear, interrupted Point Spread Functions. The software ReCSAI targets this problem by combining the classical algorithm of compressed sensing with modern methods of artificial intelligence. We evaluated several different approaches to combine these components and found, that unrolling compressed sensing into the network architecture yields the best performance in terms of reconstruction speed and accuracy.
In addition to a deep insight into the functioning and learning of artificial intelligence in combination with classical algorithms, we were able to reconstruct the described non-linearities with significantly improved resolution, in comparison to other state-of-the-art architectures.
The active zone (AZ) protein Bruchpilot (Brp) is essential for rapid glutamate release at Drosophila melanogaster neuromuscular junctions (NMJs). Quantal time course and measurements of action potential-waveform suggest that presynaptic fusion mechanisms are altered in brp null mutants (brp\(^{69}\)). This could account for their increased evoked excitatory postsynaptic current (EPSC) delay and rise time (by about 1 ms). To test the mechanism of release protraction at brp\(^{69}\) AZs, we performed knock-down of Synaptotagmin-1 (Syt) via RNAi (syt\(^{KD}\)) in wildtype (wt), brp\(^{69}\) and rab3 null mutants (rab3\(^{rup}\)), where Brp is concentrated at a small number of AZs. At wt and rab3\(^{rup}\) synapses, syt\(^{KD}\) lowered EPSC amplitude while increasing rise time and delay, consistent with the role of Syt as a release sensor. In contrast, syt\(^{KD}\) did not alter EPSC amplitude at brp\(^{69}\) synapses, but shortened delay and rise time. In fact, following syt\(^{KD}\), these kinetic properties were strikingly similar in wt and brp\(^{69}\), which supports the notion that Syt protracts release at brp\(^{69}\) synapses. To gain insight into this surprising role of Syt at brp\(^{69}\) AZs, we analyzed the structural and functional differentiation of synaptic boutons at the NMJ. At tonic type Ib motor neurons, distal boutons contain more AZs, more Brp proteins per AZ and show elevated and accelerated glutamate release compared to proximal boutons. The functional differentiation between proximal and distal boutons is Brp-dependent and reduced after syt\(^{KD}\). Notably, syt\(^{KD}\) boutons are smaller, contain fewer Brp positive AZs and these are of similar number in proximal and distal boutons. In addition, super-resolution imaging via dSTORM revealed that syt\(^{KD}\) increases the number and alters the spatial distribution of Brp molecules at AZs, while the gradient of Brp proteins per AZ is diminished. In summary, these data demonstrate that normal structural and functional differentiation of Drosophila AZs requires concerted action of Brp and Syt.
Single-molecule localization microscopy (SMLM) greatly advances structural studies of diverse biological tissues. For example, presynaptic active zone (AZ) nanotopology is resolved in increasing detail. Immunofluorescence imaging of AZ proteins usually relies on epitope preservation using aldehyde-based immunocompetent fixation. Cryofixation techniques, such as high-pressure freezing (HPF) and freeze substitution (FS), are widely used for ultrastructural studies of presynaptic architecture in electron microscopy (EM). HPF/FS demonstrated nearer-to-native preservation of AZ ultrastructure, e.g., by facilitating single filamentous structures. Here, we present a protocol combining the advantages of HPF/FS and direct stochastic optical reconstruction microscopy (dSTORM) to quantify nanotopology of the AZ scaffold protein Bruchpilot (Brp) at neuromuscular junctions (NMJs) of Drosophila melanogaster. Using this standardized model, we tested for preservation of Brp clusters in different FS protocols compared to classical aldehyde fixation. In HPF/FS samples, presynaptic boutons were structurally well preserved with ~22% smaller Brp clusters that allowed quantification of subcluster topology. In summary, we established a standardized near-to-native preparation and immunohistochemistry protocol for SMLM analyses of AZ protein clusters in a defined model synapse. Our protocol could be adapted to study protein arrangements at single-molecule resolution in other intact tissue preparations.
The plasma membrane is one of the most thoroughly studied and at the same time most complex, diverse, and least understood cellular structures. Its function is determined by the molecular composition as well as the spatial arrangement of its components. Even after decades of extensive membrane research and the proposal of dozens of models and theories, the structural organization of plasma membranes remains largely unknown. Modern imaging tools such as super-resolution fluorescence microscopy are one of the most efficient techniques in life sciences and are widely used to study the spatial arrangement and quantitative behavior of biomolecules in fixed and living cells. In this work, direct stochastic optical reconstruction microscopy (dSTORM) was used to investigate the structural distribution of mem-brane components with virtually molecular resolution. Key issues are different preparation and staining strategies for membrane imaging as well as localization-based quantitative analyses of membrane molecules.
An essential precondition for the spatial and quantitative analysis of membrane components is the prevention of photoswitching artifacts in reconstructed localization microscopy images. Therefore, the impact of irradiation intensity, label density and photoswitching behavior on the distribution of plasma membrane and mitochondrial membrane proteins in dSTORM images was investigated. It is demonstrated that the combination of densely labeled plasma membranes and inappropriate photoswitching rates induces artificial membrane clusters. Moreover, inhomogeneous localization distributions induced by projections of three-dimensional membrane structures such as microvilli and vesicles are prone to generate artifacts in images of biological membranes. Alternative imaging techniques and ways to prevent artifacts in single-molecule localization microscopy are presented and extensively discussed.
Another central topic addresses the spatial organization of glycosylated components covering the cell membrane. It is shown that a bioorthogonal chemical reporter system consisting of modified monosaccharide precursors and organic fluorophores can be used for specific labeling of membrane-associated glycoproteins and –lipids. The distribution of glycans was visualized by dSTORM showing a homogeneous molecule distribution on different mammalian cell lines without the presence of clusters. An absolute number of around five million glycans per cell was estimated and the results show that the combination of metabolic labeling, click chemistry, and single-molecule localization microscopy can be efficiently used to study cell surface glycoconjugates.
In a third project, dSTORM was performed to investigate low-expressing receptors on cancer cells which can act as targets in personalized immunotherapy. Primary multiple myeloma cells derived from the bone marrow of several patients were analyzed for CD19 expression as potential target for chimeric antigen receptor (CAR)-modified T cells. Depending on the patient, 60–1,600 CD19 molecules per cell were quantified and functional in vitro tests demonstrate that the threshold for CD19 CAR T recognition is below 100 CD19 molecules per target cell. Results are compared with flow cytometry data, and the important roles of efficient labeling and appropriate control experiments are discussed.
Fluorescence labeling of difficult to access protein sites, e.g., in confined compartments, requires small fluorescent labels that can be covalently tethered at well-defined positions with high efficiency. Here, we report site-specific labeling of the extracellular domain of γ-aminobutyric acid type A (GABA-A) receptor subunits by genetic code expansion (GCE) with unnatural amino acids (ncAA) combined with bioorthogonal click-chemistry labeling with tetrazine dyes in HEK-293-T cells and primary cultured neurons. After optimization of GABA-A receptor expression and labeling efficiency, most effective variants were selected for super-resolution microscopy and functionality testing by whole-cell patch clamp. Our results show that GCE with ncAA and bioorthogonal click labeling with small tetrazine dyes represents a versatile method for highly efficient site-specific fluorescence labeling of proteins in a crowded environment, e.g., extracellular protein domains in confined compartments such as the synaptic cleft.
Cellular responses to outer stimuli are the basis for all biological processes. Signal integration is achieved by protein cascades, recognizing and processing molecules from the environment. Factors released by pathogens or inflammation usually induce an inflammatory response, a signal often transduced by Tumour Necrosis Factor alpha (TNF). TNFα receptors TNF-R1 and TNF-R2 can in turn lead to apoptosis or proliferation via NF-B. These processes are closely regulated by membrane compartimentalization, protein interactions and trafficking. Fluorescence microscopy offers a reliable and non-invasive method to probe these cellular events. However, some processes on a native membrane are not resolvable, as they are well below the diffraction limit of microscopy. The recent development of super-resolution fluorescence microscopy methods enables the observation of these cellular players well below this limit: by localizing, tracking and counting molecules with high spatial and temporal resolution, these new fluorescence microscopy methods offer a previously unknown insight into protein interactions at the near-molecular level. Direct stochastic optical reconstruction microscopy (dSTORM) utilizes the reversible, stochastic blinking events of small commercially available fluorescent dyes, while photoactivated localization microscopy (PALM) utilizes phototransformation of genetically encoded fluorescent proteins. By photoactivating only a small fraction of the present fluorophores in each observation interval, single emitters can be localized with high precision and a super-resolved image can be reconstructed. Quantum Dot Triexciton imaging (QDTI) utilizes the three-photon absorption (triexcitonic) properties of quantum dots (QD) and to achieve a twofold resolution increase using conventional confocal microscopes. In this thesis, experimental approaches were implemented to achieve super-resolution microscopy in fixed and live-cells to study the spatial and temporal dynamics of TNF and other cellular signaling events. We introduce QDTI to study the three-dimensional cellular distribution of biological targets, offering an easy method to achieve resolution enhancement in combination with optical sectioning, allowing the preliminary quantification of labeled proteins. As QDs are electron dense, QDTI can be used for correlative fluorescence and transmission electron microscopy, proving the versatility of QD probes. Utilizing the phototransformation properties of fluorescent proteins, single-receptor tracking on live cells was achieved, applying the concept of single particle tracking PALM (sptPALM) to track the dynamics of a TNF-R1-tdEos chimera on the membrane. Lateral receptor dynamics can be tracked with high precision and the influences of ligand addition or lipid disruption on TNF-R1 mobility was observed. The results reveal complex receptor dynamics, implying internalization processes in response to TNFα stimulation and a role for membrane domains with reduced fluidity, so-called lipid raft domains, in TNF-R1 compartimentalization prior or post ligand induction. Comparisons with previously published FCS data show a good accordance, but stressing the increased data depth available in sptPALM experiments. Additionally, the active transport of NF-κB-tdEos fusions was observed in live neurons under chemical stimulation and/or inhibition. Contrary to phototransformable proteins that need no special buffers to exhibit photoconversion or photoactivation, dSTORM has previously been unsuitable for in vivo applications, as organic dyes relied on introducing the probes via immunostaining in concert with a reductive, oxygen-free medium for proper photoswitching behaviour. ATTO655 had been previously shown to be suitable for live-cell applications, as its switching behavior can be catalyzed by the reductive environment of the cytoplasm. By introducing the cell-permeant organic dye via a chemical tag system, a high specificity and low background was achieved. Here, the labeled histone H2B complex and thus single nucleosome movements in a live cell can be observed over long time periods and with ~20 nm resolution. Implementing these new approaches for imaging biological processes with high temporal and spatial resolution provides new insights into the dynamics and spatial heterogeneities of proteins, further elucidating their function in the organism and revealing properties that are usually only detectable in vitro.