High-throughput and adaptive super-resolution fluorescence microscopy of organelles

Fluorescence super-resolution microscopy has allowed unprecedented insight into the workings of biological systems below the diffraction limit of light. Over the past decade, it has overcome several challenges to deliver 3D, multi-color and faster imaging techniques, which have found many applications in deciphering the workings and structure of organelles â specialized cellular compartments whose shape and dynamics serve their specific functions. However, most super-resolution techniques are laborious and slow, and there are existing challenges in achieving sufficient throughput for collection of large datasets and capturing dynamic processes in living cells. The aim of this thesis is to address issues of limited throughput, temporal resolution and live-cell compatibility, with application to centriole architecture and mitochondrial dynamics. The design of a novel flat-fielding module for multi-focal excitation which â once integrated into an instant structured illumination microscope â allows the capture of larger fields of view with uniform illumination and hence results in increased throughput. This flat-fielded microscope allows multi-color, 3-dimensional imaging at double the diffraction-limited resolution of up to 100 mammalian cells within 1-2 minutes. Furthermore, combined with advanced sample preparation using expansion microscopy, the setup achieves an effective resolution of ~35 nm, comparable to some of the more powerful super-resolution techniques, but at 100-1000 times increased throughput. To showcase this improvement, we study the distribution of post-translational modifications within the centriole, revealing previously unobserved molecular organization. Second, with fast, live-cell imaging, we investigate the role of membrane bending energy and tension in mitochondrial division. While many molecular factors involved in this process have been identified, little is known about its physical requirements. By analyzing mitochondrial constrictions enriched in Drp1 that divide successfully or reverse, we establish that mitochondrial constrictions with higher bending energy are more likely to divide. Furthermore, by analyzing changes in mitochondrial shape and employing a novel membrane tension sensor, we observe that conditions under which mitochondrial membrane tension is increased, result in more successful divisions. These observations are supported by a model in which membrane bending energy increases the elastic energy at the constriction site, bringing it closer to the energy barrier to fission, while membrane tension acts as a fluctuation to overcome the remaining residual energy. Finally, to overcome existing trade-offs between temporal resolution and imaging duration, we develop an acquisition framework for smart and adaptive temporal sampling. By adapting the temporal resolution is in response to events of interest, they are captured at high imaging speed, while preserving the sample during a lack thereof. Using the accumulation of Drp1 to detect mitochondrial constriction sites allowed us to adapt the imaging speed to capture mitochondrial constrictions at high temporal resolution while preserving the sample when such events are lacking. Overall, this framework provides both high temporal resolution and long imaging times when needed, decreasing resulting phototoxicity, photobleaching and amount of redundant data.

Imaging tissues and cells beyond the diffraction limit with structured illumination microscopy and Bayesian image reconstruction

Tomas Lukes, Jakub Pospisil

Background: Structured illumination microscopy (SIM) is a family of methods in optical fluorescence microscopy that can achieve both optical sectioning and super-resolution effects. SIM is a valuable method for high-resolution imaging of fixed cells or tissues labeled with conventional fluorophores, as well as for imaging the dynamics of live cells expressing fluorescent protein constructs. In SIM, one acquires a set of images with shifting illumination patterns. This set of images is subsequently treated with image analysis algorithms to produce an image with reduced out-of-focus light (optical sectioning) and/or with improved resolution (super-resolution). Findings: Five complete, freely available SIM datasets are presented including raw and analyzed data. We report methods for image acquisition and analysis using open-source software along with examples of the resulting images when processed with different methods. We processed the data using established optical sectioning SIM and super-resolution SIM methods and with newer Bayesian restoration approaches that we are developing. Conclusions: Various methods for SIM data acquisition and processing are actively being developed, but complete raw data from SIM experiments are not typically published. Publically available, high-quality raw data with examples of processed results will aid researchers when developing new methods in SIM. Biologists will also find interest in the high-resolution images of animal tissues and cells we acquired. All of the data were processed with SIMToolbox, an open-source and freely available software solution for SIM.

Oxford University Press (OUP)2019

Developing luminescent Brownian probes for near-field investigations of the intracellular environment

Flavio Maurizio Mor

Life sciences have a constantly growing need for novel methodological approaches suitable to investigate the intracellular environment with increased temporal and spatial resolutions. Recently, optical near-field probes, such as laser-irradiated pointed metal, uncoated/metal-coated tapered optical fibers, as well as nano-emitters, such as single molecules or nanoparticles, have attracted increasing attention as key components of high-resolution microscopes. In parallel, super-resolution fluorescence microscopy, operating in far-field regime and overcoming the light diffraction limit, has markedly improved imaging resolution. Although both near- and far-field optical approaches drastically improve image resolution, they still represent numerous drawbacks, such as, technical difficulties concerning probe preparation (near-field) or potential damaging through very high light intensities (far-field). The aforementioned shortcomings of high-resolution optical microscopy can be overcome to a great extent by photonic force microscopy (PFM). PFM employs a strongly focused near-infrared (NIR) laser light to hold a dielectric or metallic particle as a local probe. Such an optical trap enables one to ‘cage’ a mesoscopic particle and track its three-dimensional (3D) thermal fluctuations in the surrounding environment, e.g. a viscous liquid. Therefore, besides offering near-field 3D imaging, PFM reports also on other important quantities concerning local mechanical properties, including force and viscosity as well as dynamical properties of the surrounding medium. This additional information is derived from the careful analysis of the jittery motion (so-called Brownian motion) of the trapped single-particle probe that collides with thermally-activated surrounding molecules. In this thesis, we first study in detail the Brownian motion of a single spherical particle that is confined within the harmonic potential of the optical trap. To this end, we optimize the NIR light path and electronic noise floor of our custom-built PFM set-up for detecting and quantifying resonances in the Brownian motion. These resonances arise from the coupling between the hydrodynamic memory and strong strength of the optical trap. Due to the high sensitivity of the short-time dynamics, the size of the particle can be measured by simultaneous fitting of the velocity autocorrelation function and power spectral density of its thermal fluctuations. In order to choose the best-suitable spherical probe for a given experiment in PFM, computational modeling based on a Matlab toolbox is performed. The generalized Lorenz-Mie theory is computed using the T -matrix method for various experimental conditions, including changes in the size and refractive index of the sphere, as well as different laser polarization states. The axial equilibrium position is examined to predict its location compared with the position of the laser focus. Optical forces acting on the sphere are investigated in 3D to highlight, in particular, the limits of the perfectly harmonic potential assumption. These limits are quantified and discussed. Subsequently, we identify different types of inorganic nano/micro-sized particles that can be trapped by the NIR laser of the PFM and used simultaneously as mechanical and near-field light sources. To this end, we take advantage of the visible light emission from single probes confined in the optical trap and excited by the NIR laser light. The emphasis is on particles revealing nonlinear optical properties. In particular, nonlinear optical field enhancement resulting from excitation of surface plasmon resonance (SPR) on trapped gold particles of 200 nm is demonstrated by measuring the emission of that second-harmonic generation (SHG). Furthermore, we show, under optical trapping conditions, SHG emission from single potassium niobate (KNbO3) nano/micro-sized particles. We also report on the multicolor upconversion luminescence (UCL), at the single particle level, for randomly shaped crystals of sodium yttrium fluoride codoped with ytterbium and erbium, NaYF4:Yb,Er (UCC), when they are held in the optical trap. In parallel, 3D Brownian fluctuations of a single trapped particle are quantified. Moreover, luminescence resonance energy transfer (LRET) between a KNbO3 particle or an UCC and molecules of an organic dye (rose bengal) adsorbed on their surface is highlighted and analyzed. Finally, randomly shaped UCCs and hexagonal nanoplates are characterized in detail at the single particle level. The hexagonal β phase in the polycrystalline and monocrystalline crystals is identified, in both randomly and hexagonally shaped particles by electron microscopy. UCL emission from the trapped crystal is quantified in terms of power and found to be dependent on the laser power density, size and shape of the particle as well as its orientation within the trap. Most importantly, the different colors in the UCL spectrum do not vary with the same trend for different orientations of the randomly or hexagonally shaped crystal upon trapping. This suggests the existence of crystalline anisotropy-dependent UCL. Thereby, we should be able to design the best-suited particle probe for LRET experiments with subwavelength resolution.

EPFL2013

Exploring dynamic organellar shape using live-cell fluorescence super-resolution microscopy

Lina Carlini

Eukaryotic cells contain membrane-bound organelles, that perform specialized tasks, which control cellular fate. Organelles are dynamic and contain nanometre-scale, ultrastructural features; in fact, their shape and function are interconnected. Thus, there is a need to image with high spatial and temporal resolution while minimizing perturbations to the cell. Several fluorescence-based super-resolution (SR) techniques exist, including single molecule localization microscopy (SMLM), which offers unsurpassed spatial resolution. However, this technique is limited by factors that prevent non-perturbative, fast, high resolution, quantitative imaging of biological processes. In particular, live-cell SMLM lacks flexible, robust labeling strategies; it requires high laser irradiances and toxic chemical buffers; and 3D data is often misrepresented due to optical distortions that are apparent at such high resolutions. Here, we aim to eliminate obstacles to live SMLM. Firstly, we screen site-specific dyes that are designed to target organelles. We identify dyes with excellent photophysical properties for live SMLM. Unlike many other probes for SMLM, these work well in non-toxic buffers. Secondly, we transform rhodamine dyes into their non-fluorescent leuco-rhodamine form. This transformation is reversible and occurs spontaneously through oxidation in situ. Labelling with this leuco-rhodamine dye increases single-molecule density, which is advantageous for SMLM since higher molecular densities yield improved spatial and temporal resolutions. Furthermore, this strategy does not require buffers and performs best at relatively low laser irradiances. Thirdly, we identify a distortion present in three-dimensional (3D) SMLM, which warps structures imaged. We characterize this prevalent distortion, termed wobble, on four 3D SMLM systems and identify its source. We computationally eliminate wobble and show that live-cell, 3D structures are accurately represented post-correction. Using this correction, we also develop a strategy to register dual-color, 3D SMLM data. Finally, we apply live-cell SMLM to study mitochondrial fission. Mitochondria are dynamic and frequently undergo fusion and fission to maintain their function and structure. The dynamin related protein, Drp1 in mammals, is required for mitochondrial fission and is known to mediate this process via a constriction force. The endoplasmic reticulum and actin are also reported to play a role in mitochondrial constriction and ultimately, fission. However, a model for the final stages of fission post constriction, is still missing. With this ultimate aim, we perform live SMLM of many mitochondrial fission events. We identify a decrease in viability with our imaging conditions and thus validate our findings with an alternative, more live-cell compatible, SR technique: structured illumination microscopy (SIM). Our preliminary results with live SMLM and SIM reveal that while most constrictions proceed to fission, 5% of these events relax back to their unconstricted shape. Measurements with SMLM reveal that minimum diameters for both scission events and those that reversed were 80 nm. Also, mitochondria, which achieve a higher negative envelope curvature, are more likely to divide. By examining Drp1 location, we find that this protein is consistently on one mitochondrial end post-scission. These preliminary results are discussed in the context of the final stages of mitochondrial fission.