Optimization of Expansion Microscopy Protocolkth.diva-portal.org/smash/get/diva2:1238657/FULLTEXT01.pdfAbstract Expansion microscopy (ExM) enables imaging of preserved cellular or

An investigation of performing the protein-retention expansion microscopy protocol on

neuronal cells

ELIZA LINDQVIST

Department of Applied Physics

KTH Royal Institute of Technology

Supervisor: Hans Blom Examiner: Erik Lindahl

Master’s Thesis Stockholm, Sweden 2018

TRITA-SCI-GRU 2018:317

Department of Applied Physics School of Engineering Sciences

Royal Institute of Technology Stockholm SWEDEN

Examensarbete som med tillstånd av Kungliga Tekniska Högskolan främlägges till offentlig granskning för avläggande av Civilingenjörsexamen i Teknisk Fysik 4 augusti 2018 i Seminarierum Becquerel på SciLifeLab, Tomtebodavägen 23, Solna. © Eliza Lindqvist, 4 augusti 2018

Abstract Expansion microscopy (ExM) enables imaging of preserved cellular or tissue specimens with nanoscale resolution on diffraction-limited instead of super-resolution microscopes. On broad terms ExM works by physically enlarging the specimen, after having labeled it with fluorescent probes anchored to a swellable gel. In this Master Thesis work I present an investigation of the protein retention Expansion Microscopy (proExM) protocol for expansion of cultured neuronal cells. The expansion of neurological networks enables for example the ability to pinpoint small topological protein changes inside the brain, which could take affect during the development of diseases like Alzheimer’s, epilepsy and Parkinson’s disease. To evaluate the protein retention protocol I stained neuronal cells with different antibodies and I compared images of samples imaged with confocal, STED and Expansion Microscopy. I quantified the expansion factor in neurons by measuring of the distance between fixed architectural Spectrin rings. To evaluate retained protein content I varied the digestion times and anchoring treatments to study how different treatments affected the imaged intensity. Here, I show that samples anchored with Acryloyl X – SE lose a significant amount of protein with increased enzymatic digestion times. Furthermore, I show that samples anchored with Acryloyl X – SE are further affected by the digestion times as the fluorescently labelled sample lose imaged intensity over time. This is in sharp contrast to expanded samples anchored with MA-NHS which shows no imaged intensity decrease with longer enzymatic digestion times.

Preface

This is a Master’s Thesis in Engineering Physics at the Department of Cellular physics, Royal Institute of Technology (KTH), Stockholm, Sweden. The presented laboratory work was performed at Science for Life Laboratory in the Cellular Biophysics group belonging to the Department of Applied Physics, Solna, Sweden.

Acknowledgments I want to thank everyone who has helped me throughout this master’s thesis. Extra gratitude goes to my supervisor Hans Blom and examiner Erik Lindahl, for initial project idea, guidance and support. Furthermore, I would like to give my sincere gratitude to group members Steven Edwards and David Unnersjö-Jess, you have been the best and supported me tremendously, I cannot thank you enough. Huge thanks go also to Daniel Jans for general help and custom microscopy setup and William Björnstjerna for illustrating figures used in this report.

Eliza Lindqvist

2018-06-10 Stockholm, Sweden

CONTENTS

LIST OF FIGURES .................................................................................................................. 1

1. INTRODUCTION ............................................................................................................. 4

2. BACKGROUND AND THEORY ..................................................................................... 6

FLUORESCENCE ........................................................................................................................ 6 CONFOCAL MICROSCOPY.......................................................................................................... 7 SUPER-RESOLUTION IMAGING ................................................................................................... 8 EXPANSION MICROSCOPY ....................................................................................................... 10 PROCEDURE ........................................................................................................................... 10 VARIANTS OF EXM ................................................................................................................ 13 APPLICATION OF EXM TO NEUROSCIENCE ............................................................................... 15

3. EXPERIMENTAL WORK ............................................................................................. 17

PROEXM PROTOCOL ............................................................................................................... 17 SAMPLE PREPARATION TECHNIQUES ........................................................................................ 18 IMAGING ................................................................................................................................ 20 DATA ANALYSIS AND REPRESENTATIONS................................................................................. 20

4. RESULTS ........................................................................................................................ 22

STAINING ............................................................................................................................... 22 COMPARISON BETWEEN CONFOCAL, STED AND EXPANSION MICROSCOPY .............................. 24 RESOLVING THE PERIODIC EXPRESSION OF SPECTRIN ............................................................... 26 VARIATION OF THE DIGESTION TIME ........................................................................................ 27 A COMPARISON BETWEEN ACRYLOYL X -SE AND MA-NHS .................................................... 30

5. DISCUSSION .................................................................................................................. 33

6. SUPPLEMENTARY ....................................................................................................... 35

7. APPENDICES ................................................................................................................. 37

MATLAB SCRIPT .................................................................................................................. 37

8. REFERENCES ................................................................................................................ 38

1

LIST OF FIGURES Figure 1. Jablonski diagram. A Jabionski diagram demonstrating the electron transition states of the fluorophore during absorption/emission processes. (A) The energy of an incoming photon (green) is absorbed by the fluorophore molecule and the fluorophore reaches the excited state. (B) Some of the energy is dissipated as heat or other processes. (C) The fluorophore returns to the ground state and a photon of lower energy and longer wavelength is emitted (red). .................... 6 Figure 2. Confocal microscope principle. Simple illustration of the principal light pathways in a confocal microscopy. Light emitted by the laser passes through a pinhole aperture, is reflected by a dichromatic mirror and scanned across the specimen in a defined focal plane. Light emitted from the specimen (fluorescence) emanating from the focal plane passes back through the dichromatic mirror and is focused as a confocal point at the detector pinhole aperture. ................ 7 Figure 3. Confocal versus STED imaging. Non-expanded neurons treated with Spectrin. (LEFT) Confocal image. (RIGHT) STED image. ...................................................................... 9 Figure 4. Schematic of polyelectrolyte network. (A) Schematic of a collapsed polyelectrolyte network, showing crosslinker (dot) and polymer chain (line). (B) Expanded network after H2O dialysis. ..................................................................................................................................... 12 Figure 5. Schematic of microtubules and polymer network. First, the specimen is fixed and treated with compounds that bind to the biomolecules called anchoring treatment. Next, a hydrogel of densely crosslinked monomers is polymerized throughout the cells/tissue called gelation. Then, the specimen-hydrogel composite is digested and then the specimen is finally ready for the expansion by dialysis in water. (A) Schematic of microtubules (green) and polymer network (blue). (B) A label that can be anchored to the gel at site of a biomolecule, is hybridized to the oligo-bearing secondary antibody top bound via the primary to microtubules (green lines) and is incorporated into the gel (blue lines) via the methacryloyl group (red dot). ...................... 12 Figure 6. Membrane expansion and ExFISH. (LEFT) Maximum intensity projection of confocal microscopy stack following expansion of membrane labeled Brainbow3.0 neurons [2]. (RIGHT) ExFISH image with delivered probes against six RNA targets in a cultured HeLa cell, Scale bar 20 µm [9]. .................................................................................................................. 13 Figure 7. The box used for staining. ....................................................................................... 19 Figure 8. Gel chamber schematic............................................................................................ 19 Figure 9. A petri dish with a lid. A petri dish with an added lid was used as image sample holder for the gel. The added screws on top were applied as weight to prevent the gel sample from drifting while imaging. ...................................................................................................... 20 Figure 10. proExM imaging of antibodies of interest. Confocal images of expanded neuronal cells stained with (A) Alpha tubulin. (B) Spectrin. (C) Synaptotagmin. (D) Pan Neuronal marker. .................................................................................................................................................. 23 Figure 11. Comparison between confocal, STED and confocal expansion microscopy. We compared images acquired via confocal microscopy versus images acquired via STED microscopy and post-expansion confocal microscopy. All samples are stained with Spectrin. (A)

2

Non-expanded sample imaged with a confocal microscope. (B) Non-expanded sample imaged with a STED microscope. (C) Expanded sample imaged with a confocal microscope. ............... 25 Figure 13. Periodic expression of Spectrin in neurons. (A) A intensity plot of the marked area in (C). (B) Image of expanded neurons stained with Spectrin. (C) A zoomed-in area of (B) were the area of interest is marked in yellow. ..................................................................................... 26 Figure 14. Neurons with different digestion times. Neurons treated with Spectrin, Goat anti-Mouse (Atto) 594 and Acryloyl X -SE and imaged with a confocal microscope. (A) Digestion time: 15 min. (B) Digestion time: 1 hour. (C) Digestion time: 2 hours. (D) Digestion time: 4 hours. ........................................................................................................................................ 28 Figure 15. Intensity drop off with digestion time. Expanded neuronal cells stained with Spectrin, digested for different times and imaged with a confocal microscope. (A) Profile of intensity taken at different time points. (B) Confocal image of neuronal cells not digested with proteinase K. (C) Confocal image of neuronal cells digested with proteinase K for 24 hours. .... 29 Figure 16. Comparison between anchoring treatment with MA-NHS and Acryloyl X-SE. Non-expanded neurons stained with Pan Neuronal, anchored with either MA-NHS or Acryloyl X -SE and imaged with a confocal microscope. (A, B, C and D) Neurons anchored with MA-NHS and digested for (A) 0 minutes, (B) 2 hours, (C), 4 hours and (D) 24 hours. (E, F, G and H) Neurons anchored with Acryloyl X-SE and digested for (E) 0 minutes, (F) 2 hours, (G), 4 hours and (H) 24 hours. (I) Graph showing the image intensity over different digestion times in neurons anchored with MA-NHS. (J) Graph showing the average image intensity over different digestion times in neurons anchored with Acryloyl X-SE. ........................................................................ 31 Figure 17. Comparison between anchoring treatment with MA-NHS and Acryloyl X -SE. Expanded neuronal cells stained with anti-Pan-Neuronal marker, imaged with a confocal microscope and anchored with (A) MA-NHS or (B) Acryloyl X-SE. The asterisk in panel B indicates a neuronal structure which has not expanded uniformly, leaving an apparent break in an otherwise continuous axon or dendrite. ...................................................................................... 32 Figure S 20. Schematic of refractive image formation (magnified, real and inverted image), of an object placed at distance a in-front of a thin lens having focal length f. The imaged is formed at distance b and the magnification is given as M=b/a .................................................................. 35 Figure S 21. An AIRY disc diffraction pattern of a point source. The point source is here the object and it will be imaged by the optical system as a wiggly pattern with a main central peak and neighboring smaller ringing, because of the wave nature of light (in other words diffraction is scattering and inference of light waves). .................................................................................... 35 Figure S 22. Schematic of focal spot dimension of a fluorescent microscope. The excitation wavelength is used to excite fluorescent molecules pictured as small orange discs: However, none of the fluorescent molecules (point sources) will be resolved, as their distance is below the Abbe limit. They will instead all be merged into a single focal blob of width ~200 nm and height of 3-4 times this value in the very best diffraction-limited case. ........................................................... 36

3

4

1. INTRODUCTION In traditional light microscopy, fine structures are resolved by using refraction to magnify a samples topological structure. Magnification allows to for example visualizing a whole plant or animal cell with light microscopy. However, zooming-in indefinitely on optical microscopes is however not possible, as finally light magnification will be ultimately hampered and limited by diffraction. Diffraction comprises the spreading of light waves when interacting with structures within a sample. This incidence limits the ability to differentiate two objects separated by a lateral distance shorter than about half of the wavelength used to image the specimen, even when this is just a very small point. A point source diffraction pattern is referred to as an Airy disk (see Fig.S2), and the size of the disk is determined by the wavelength of the light and the aperture collection angle of the microscope objective used to image the object. In terms of resolution, the radius of the of the Airy disk in the lateral image plane is defined by the formula:

AbbeResolution-,/ = λ 2NA⁄ (Eq. 1)

where λ is illumination wavelength and the numerical aperture (NA) is the refractive index of the imaging medium multiplied by the sine of the aperture angle (i.e. NA = n sina). Consequently, minimizing λ and maximizing NA there is thus a technical lower limit below which the microscopes’ optical system cannot resolve structural details, i.e. separate two neighboring points in space [1].

However, several techniques have been developed in the past decade to circumvent the diffraction limit, and these techniques have collectively been called super-resolution microscopy (SRM). Today’s most used super-resolution techniques is Structured Illumination Microscopy (SIM), Stimulated Emission Depletion (STED), Stochastic Reconstruction Microscopy (STORM), Photo-activated Localization Microscopy (PALM) and Point Accumulation for Imaging of Nanoscale Topography (PAINT) [2] [3]. These recent techniques are focusing on increasing the resolution of the microscope by separating individual, or groups of fluorescent molecules in space and time. This spatiotemporal separation allows distinguishing neighboring entities and imaging them separately, even though they are distanced less than the Airy and Abbe diffraction limit. Yet, there are remaining complications for these SRM methods in the sense of complex hardware, high costs and need of microscopy experts for use. For such reasons, SRM imaging methods have presently not efficiently been put to use within clinical practice and they are very rarely applied to clinical samples [4]. In addition, they are limited by imaging speed and in accessible imaged volumes. Another simpler way of receiving super-resolution imaging has thus in parallel been developed in the last years.

Expansion microscopy (ExM) is a relatively new developed method, were instead of further attempting to super-resolve optically, you physically expand the sample with the help of a swellable polymer. The polymer is superabsorbent and by adding water it can expand isotopically around 4-5 in size, thus pulling the positions of labelled biomolecules apart. Due to the induced larger

5

separation between positions of investigated molecules, 3-D nanoscale resolution on low-cost diffraction-limited microscopes is made possible [5]. Expanding for example, a mouse brain, this method has enabled researches to see several tiny sub-cellular building blocks and biomolecules inside the brain and its neuronal cells. The resolved information has allowed researchers to better figure out how the brain and brain cells are organized in three dimensions. This gained knowledge has yielded a deeper understanding of the brain and how action, sensation and emotional networks and circuity might be topological wired. Expansion microscopy, through its sample induced super-resolving power, can thus also enable the ability to pinpoint small topological changes inside the brain which could increase our knowledge on the development of diseases like Alzheimer’s, epilepsy and Parkinson’s disease.

In this thesis, the focus was put on optimizing a protocol called protein retention ExM (proExM) used for Expansion Microscopy on neuronal cells. In the proExM protocol, proteins are anchored to the swellable gel allowing the subsequent use of conventional fluorescently labeled antibodies or anchoring of fluorescent proteins, to topologically investigate protein specimen structures. Numerous unsolved research questions within the neuroscience field center around the knowledge and understanding of how molecules and wiring in neuronal circuits produces behavioral functions and neurodegenerative diseases. In this thesis I stained neuronal cells with different antibodies and I structurally compared images of samples imaged with confocal, STED and Expansion Microscopy. I quantified the expansion factor by measuring of the distance between Spectrin rings, and I varied the digestion times and anchoring treatments to study how these sample treatments affected the imaged intensity.

6

2. BACKGROUND AND THEORY In this section, the background of present imaging options is presented along with the explanations of the expansion microscopy methods and previously work done within this field of research.

Fluorescence Fluorescence is a process where absorbed energy of a fluorophore molecule is emitted via electromagnetic radiation. A fluorophore is a molecule that is able to efficiently release this extra absorbed energy as light waves (photons). Normally, a fluorophore is in its relaxed state, also called the ground state, in which the molecule is stable and carrying low energy levels (see fig. 1). When light with high enough energy is transmitted onto the fluorophore, the energy from the so-called excitation light can be instantly absorbed by the fluorophore, such that the fluorophore reaches a higher energy state, also called the excited state [6]. However, after excitation fluorophores can after internal fast rotational and vibrational movements fall back to their relaxed ground state. During this relaxation process, a fluorophore can emit part of the absorbed excess energy in form of a lower energy photon; meaning that the color (i.e. wavelength) of the light transmitted onto and used to excite fluorophore with is different from the color of the light emitted. The difference in absorption/excitation wavelength and emission/fluorescence wavelength is beneficially used in fluorescence microscopy by filtering out on the emission signal from fluorophores labeling cells or tissue. Fluorophores can be anchored to specific parts of a biological sample and let the observer see desired fragments of the specimen. Fluorescent proteins, labelled peptides and antibodies enable the fluorescent visualization of structures and processes on the sub-cellular level [7].

Figure 1. Jablonski diagram. A Jablonski diagram demonstrating the electron transition states of the fluorophore during absorption/emission processes. (A) The energy of an incoming photon (green) is absorbed by the fluorophore molecule and the fluorophore reaches the excited state. (B) Some of the energy is dissipated as heat or other processes. (C) The fluorophore returns to the ground state and a photon of lower energy and longer wavelength is emitted (red).

7

Confocal Microscopy A confocal microscope uses fluorescence as signal to generate images with high resolution by optically imaging only a very small focal volume. The microscope is designed with an aperture placed in the optical pathway in front of the detector, were the aperture (called a pinhole) is at the same position as the microscope focus of the collected sample image (see fig. 2). Schematically, radiation from a laser system travels through a light-shaping pinhole aperture and is then reflected by a dichromatic mirror to excite the labelled specimen. The light that is emitted from the specimen (the fluorescence) will go back through lens and mirror system and finally hit a detector, in-front of which a second pinhole is placed [8]. By modifying the aperture size to match the size of the optical resolution of the objective (see Eq.1), the pinhole will block out light outside the focal plane of the objective lens and optical sectioning is made possible. Thus, emitted or scattered light that is out of focus is removed, leaving an image with highest resolution possible and a very good signal to background ratio can be achieved. This optical confocal technology offers several advantages over conventional wide-field optical microscopy since one are able to optically control the imaged depth of field, and one can eliminate background information, as well collect serial optical sections from thick specimens [9].

Figure 2. Confocal microscope principle. Simple illustration of the principal light pathways in a confocal microscopy. Light emitted by the laser passes through a pinhole aperture, is reflected by a dichromatic mirror and scanned across the specimen in a defined focal plane. Light emitted from the specimen (fluorescence) emanating from the focal plane passes back through the dichromatic mirror and is focused as a confocal point at the detector pinhole aperture.

8

Super-resolution Imaging In conventional diffraction-limited far-field optical microscopy there is a resolution limit at approximately 200 nanometers laterally when imaging with visible light (see Eq.1). This limit exists because of the laws of nature (diffraction), and a mathematical principle stated by Ernst Abbe in the late 19th century describes this limit as function of the applied wavelength ant the objective’s performance. In other words, the minimum distance (and thus highest resolution) between two objects that can be resolved, is dependent on the wavelength of light and the numerical aperture of the objective lens used [10]. Being technically limited for over hundred years by diffraction-limited microscopy, several new novel optional techniques have been developed in the last decades that can circumvent the diffraction limit. These methods are collectively known as super-resolution microscopy (SRM) and denote any light microscopy techniques that are able to image with a resolution that goes beyond the Abbe diffraction limit. In simple terms, all SRM methods circumvented the diffraction limit by switching fluorescent markers on and off between adjacent states, and one is thus capable of separate labelled entities in space (and time)having smaller distances. All of these SRM different techniques come with their pros and cons, and it is of importance to choose wisely between the different methods. The special methods can coarsely be divided into different categories, as for example with near-field methods that operate close to the sample and the opposite, far-field methods, that image samples optically at a normal distance. Within the far-field imaging category, there is an especially popular technique called stimulated emission depletion (STED) microscopy, which is well suited for imaging of thicker tissue imaging as it likes confocal microscopy have good optical section capabilities [11]. In this thesis super-resolution STED microscopy has been applied as a nanoscale imaging reference to evaluate expanded samples imaged with diffraction-limited confocal microscopy. Stimulated Emission depletion (STED) Microscopy Stimulated emission depletion microscopy (STED) uses fluorescent fluorophores (e.g. dyes) that can be switched on and off to obtain diffraction-unlimited images. Two lasers are commonly used, and they are scanning pixel by pixel over the sample. Essentially, by sequentially removing fluorescent light that is not in a nanoscaled focus area by OFF-switching higher resolution can be reached optically(see Fig. S 20). Technically this is achieved in STED by applying an intense OFF-switching laser that is sculptured to operate in the outer regions of the diffraction limited excitation/emission focus, and the high energy of this depletion laser affects the excited fluorophores so that they fall back to the ground state. This suppression of the possibility for the fluorophore to fluoresce (i.e. spontaneously emit) is obtained by stimulated emission, which occurs when a fluorophore in the excited state meets an off-switching depletion photon with energy roughly equal to the difference between the ground state and excited state (see fig.1). The excited fluorophore is through the stimulated process thus forced to falls back to the ground state, i.e. it is stimulated to relax to the ground state while actually emitting light of the same wavelength as the off-switching laser. This more red-color shifted light is however not imaged in a STED microscope (filtered out). Only the remaining exited fluorophores that is not stimulated to be turned ‘OFF’

9

into the ground state are then finally able to emit fluorescence signal from the very center of the focus. Separation in space and time by controlled switching and imaging is thus the novel principle how STED can be used to separate and better resolve fluorescently labelled structures.

In other words, selective switching after scanning from pixel to pixel thus enables features closer than the diffraction limit to be separated [12]. The novelty of STED is to realize that the OFF state is when the molecule is in its ground state, and the ON state is when the molecule is excited, and it is the controlled optical switching that gives improved (super) resolving powers. The switching of states is often controlled by co-aligning a Gaussian profiled excitation beam with a diffraction limited hollow STED beam. The first laser is often a picosecond pulsed diode laser that excites the fluorescent molecules. The excitation pulse is then followed by a red-shifted stimulated emission depletion pulse which off-switch (quench) the fluorescence from molecules in the periphery of the excitation focus. The high intensities of the stimulated emission depletion laser make sure that the periphery fluorophores switch into the ground state but is excluding the ON-signaling fluorophores remaining in the very central focal ‘zero’-intensity point. In essence STED microscopy hence, by using two laser beams allows to optically increase the spatial resolution by sculpturing the size of the OFF/ON-switching ‘zero’-region as shown theoretically and experimentally in the last decades [13].

Figure 3. Confocal versus STED imaging. Non-expanded neurons treated with Spectrin. (LEFT) Confocal image. (RIGHT) STED image.

In theory a STED microscope could with a perfect zero-region and very high stimulated emission depletion laser power image down to the spatial sizes of molecules, but practically the maximally achievable lateral resolution for cellular or tissue imaging is ~20 -50 nm [13].

However, STED microscopy has several limitations as for example the SNR (signal-to-noise-ratio) is low, because in the nanoscale focus only a few remaining or even just a single fluorescent

10

molecule contributes to the signal. SNR is affected by the efficiency of detection, of the intensities of the fluorophore signals and any background generated by the laser beams or the imaged sample structures. Measuring with STED microscopy is thus as with all SRM techniques a constant practical optimization and struggle of trying to improve signals and lower backgrounds while enhancing resolution. Accordingly, there are still lots of improvements that could advance this technique further, such as the optimization for increased sample penetration depth, or very fast temporal imaging, as well as pushing for live cell compatibility by imaging in cells, organisms or even animals [12].

Expansion Microscopy Optical super-Resolution technologies are able to achieve spatial resolution beyond the diffraction limit. However, they suffer from several technical and practical drawbacks. The methods require special fluorophores that are compatible to different ON/OFF-switching schemes, there is a risk of bleaching the samples because of the use of different high-power switching lasers, and the methods are sometimes technically very demanding both from a hardware side and software analysis side. A super-resolution method with a potentially simpler toolbox has thus also been sought for the last couple of years. By focusing on the sample and expanding it, super-resolution imaging of a blown-up sample topology is possible. This so-called expansion microscopy (ExM) approach intuitively allows one to see something invisible by making it much larger. Even though ExM is a new technique only performed for a couple of years, the method is very popular with a fast-growing use in laboratory research. The protocol is easy to follow, with relatively easy-to-get materials and subsequent imaging can be used on conventional optical microscopes. Although the method is restricted to fixed samples, the achievable nanoscale resolution provides answers to numerous of biological questions and the method is evolving to diverse scientific and clinical contexts. The crucial path to successfully apply expansion is first the tedious optimization work of finding the perfect sample protocol for the selected cell or tissue system being investigated. The second important part it have to control measurements allowing for evaluation of the expanded samples and detection of possible blown-up artifacts.

Procedure Simplistically, when applying tissue or cellular expansion protocols, one need to attach a small “handle” on the targeted biomolecules, a process called an anchoring treatment. By using access of monomers and different chemicals for reactions attaching, nesting and growing polymeric chains can go around and in-between the targeted biomolecules and also bind to their handles (see fig. 5). When water is later added, these polymeric chains will straighten out and bring the targeted biomolecules with them allowing a larger distance between each molecule to be induced (see fig.4). In the original expansion protocol by Boyden et al developed in 2015, targeted biomolecules were first labeled with a primary antibody, and then a secondary antibody bearing the gel-target anchoring moiety was added. A second specifically synthesized fluorescent tag (DNA primer + dye), which is able to bind to the antibody complexes, was thereafter used to be able to optically

11

target the biomolecules of interest to the polymeric gel. The last step in the initial ExM protocol was to add a swellable polyelectrolyte polymeric gel that cross-linked the sample and formed a dense meshwork that later could be stretched out (expanded ~4-5 times). The polymeric gel is densely cross-linked with a crosslinker so that after isotropic swelling, the targeted structure remains its (assumed) spatial organization relative to each expanded sample position (see fig. 4). Making ExM work properly is in large dependent on the induction and compatibility of the gel formation taking place in the sample (tissue or cells). The chemical sample and gel reactions must preserve the topology of targeted biomolecules, as well as the optical fluorescent signal used to read out positions. Simplistically speaking, expansion microscopy is chemistry, or advance gel forming chemistry applying polymers to anchor and swell sample topology. Polymers are large molecules that is made by many monomers that are joined together in a long chain. A single polymer is made by thousands of monomers and have unique properties depending on the type of molecules it is consisting of. Polymerization is a procedure were one take monomers and alter the chemical bonds that hold the molecules together by heat or pressure. Different chemical processes makes the molecules to mesh and bond in a network structure resulting in a gel polymer [14]. The polymer chains are very dense and as thin as the width of a biomolecule, and when they are absorbing water the chains move apart from each other. The whole material swells and becomes several times bigger in size. The polymerization is triggered with ammonium persulfate (APS) initiator and tetramethylethylenediamine (TEMED) accelerator, and then treated with protease to homogenize its mechanical properties and ensure an isotropic expansion of the sample. By adding water, the gel expands through osmoses. The polymer is capable to retain an enormous extent of water relative to its mass and swells to at least four times its size. The biomolecules are thus pushed away with a larger distance relative from each other, which by isotropic expanded separation remarkably decreases the impact of the diffraction limit when imaging the sample with a conventional light microscope. Another benefit is that the expanded sample is also very transparent, which allow deep tissue imaging of whole organs (e.g. a mouse brain) with an air or water index-matched objective. The expanded samples are highly transparent and water index-matched, since they contain more than 99% water, while the targeted positions of biomolecules remain covalently anchored with high yield to the polymeric network [14].

12

Figure 4. Schematic of polyelectrolyte network. (A) Schematic of a collapsed polyelectrolyte network, showing crosslinker (dot) and polymer chain (line). (B) Expanded network after H2O dialysis.

Figure 5. Schematic of microtubules and polymer network. First, the specimen is fixed and treated with compounds that bind to the biomolecules called anchoring treatment. Next, a hydrogel of densely crosslinked monomers is polymerized throughout the cells/tissue called gelation. Then, the specimen-hydrogel composite is digested and then the specimen is finally ready for the expansion by dialysis in water. (A) Schematic of microtubules (green) and polymer network (blue). (B) A label that can be anchored to the gel at site of a biomolecule, is hybridized to the oligo-bearing secondary antibody top bound via the primary to microtubules (green lines) and is incorporated into the gel (blue lines) via the methacryloyl group (red dot).

13

Variants of ExM The are many different kinds of ExM protocols invented over the past few years, optimized for different specimens and for different biological questions, The first expansion protocol was reported in 2015, when Edward S. Boyden and his colleagues at M.I.T Center for Neurobiological Engineering published a paper where they stated that they were able to expand cells and tissues five time their size and observe objects separated effectively down to a scale of 70 nanometers, through a conventional optical microscope. The researchers also performed three-color sample stretched super-resolution imaging of the mouse hippocampus with confocal microscopy [14]. Their tissue and cell expansion idea grew out from investigating research done by Toyoichi Tanaka in the 1970s, where he discovered and explored different polymeric gels that respond to stimuli such as water (swelling/shrinking) [16]. Boyden and his colleagues further explored whether dialyzing polyelectrolyte gels in water could expressed in biological samples and developed a strategy, in which they successfully made fluorescent labelling compatible with tissue expansion. They named the process of labeling, gelation, digestion and expansion for Expansion Microscopy and started to generate stunning images of for example fluorescently transferred brain cells [14].

Figure 6. Membrane expansion and ExFISH. (LEFT) Maximum intensity projection of confocal microscopy stack following expansion of membrane labeled Brainbow3.0 neurons [2]. (RIGHT) ExFISH image with delivered probes against six RNA targets in a cultured HeLa cell, Scale bar 20 µm [9].

After the first publication about ExM in 2015, numeral papers have been published about expansion microscopy. Boyden and colleagues have simultaneosly developed different methods and variants of ExM, to extend the application to diverse scientific and clinical contexts. In 2016, they published a paper describing florescent in situ hybridization imaging of RNA using ExM (ExFISH). In the ExFISH protocol, the target is to separate RNAs while still be able to support amplification of single-molecule signals, and ExM was thus refined to covalently link RNAs directly to the gel via small molecule linker [17]. Two months later another paper was published with a variant of ExM

14

in which proteins are anchored to the gel (proExM) [17] [18]. ProExM is as ExFISH an extension of the original protocol, where Boyden et al showed that it is possible to directly anchor proteins instead of the synthesized label to the gel polymerized throughout a biological sample. This latter method beneficially allows commercial secondary antibodies to be used, unlike the original protocol, where custom conjugation of secondary antibodies was needed [18].

Iterative Expansion Microscopy (iExM) and Expansion Pathology (ExPath) is two additional extensions of ExM published by Boydens group in 2017 [4] [19]. With the help of a second gelation step, they were able to expand dendritic spines in mouse brain circuitry 20-fold and entitled the process “iterative” ExM. After the first gelation- and expansion step, they prepared a second gel with another oligonucleotide bearing a fluorophore and a cleavable crosslinker. The oligonucleotide hybridizes to the oligonucleotide in the first gel, and then the first gel is dissolved by cleaving the crosslinkers before a second expansion [19]. Moreover, Boyden and colleagues furthermore extended the process for proExM on human tissues optimized for clinically samples (ExPath). This process starts with fixing the human tissues with formalin and embedding them in paraffin. Next, the samples are stained and fresh frozen and the tissues could then be used for analysis of nanoscale imaging for optical diagnosis of kidney minimal-change disease [4]. Later the same year, yet another paper was published were Boyden et al studied the function of expansion in a larval and embryonic zebrafish sample. With this work, they could resolve subsynaptic protein organization and characterize the shapes of nuclear invaginations and channels [20]. Further improved resolution was obtained when UltraExM was presented, a method in which the 9-fold symmetry of cellular centrioles can be visualized with a confocal microscope. By combining this method with optical super-resolution STED microscopy, images even revealed the chirality of the microtubule triplets of the centrioles for the first time. In the developed UltaExM sample protocol low concentrations of formaldehyde and acrylamide for fixation are used so that cross-linked macromolecules are affected as little as possible [21].

Besides the research on ExM done by Edward S Boyden and his team, other groups by them-self have also applied the concept of expansion in their experimental work. Joshua C. Vaughan and his team developed a new technique in the area of linking fluorophores to the swellable polymer. They found that treating cultured cells with MA-NHS (methacrylic acid N-hydroxy succinimidyl ester) presented a good retention of fluorescent signal after digestion and expansion. They also found that glutaraldehyde (GA) generated good fluorescence retention after digestion in combination with their sample protocol. In conventional fluorescence sample preparation protocols the use of GA often introduces a severe background signal. In the hands of Vaughan et al the cells treated with any of these compounds showed a much brighter signal than the untreated ones using ExM DNA-labeled antibodies. In their protocol they furthermore added high concentrations of organic fluorophores that are able to survive the gel polymerization step, thus adding to the protocol developments explored by Boyden et al [22]. The Vaughan group moreover very recently presented a paper that introduced ExSIM, which combines specimen expansion with optical super-resolution in this case applying structured illumination microscopy (SIM). The combined method yielded an

15

effective spatial superresolution of around ~30 nm [23]. In a follow up published paper from the Vaughn group, they further optimized the initial ExM protocol to even study Drosophila tissues, with a ~70 nm lateral resolution of the imaged biomolecular structure [24].

Furthermore, in July 2016 Chung et al introduced yet another expansion method for linear expansion of organs. The process was called MAP (Magnified Analysis of Proteome) and the protocol preserves the full content of a 3D tissue proteome and makes it possible to image undamaged biological structures for combined extraction of the molecular identity, subcellular architectures, and intracellular connectivity of various cell types. Contrasting the other listed protocols, MAP is capable to label thick tissues relatively fast since the labeling of antibodies is done post lipid removal and tissue gel permeabilization. MAP thus reverses the ExM protocol scheme (ExM à labeling, gelation, digestion and expansion) and in principle decouples labeling from other steps. This should potentially minimize influences induced on labeling by the other steps in the sample preparation protocol(s).

Moreover, early in 2017, a comparison between ExM and STED was presented on a wide range of different samples [25]. In connection to this, Rizzoli et al announced an enhanced expansion technique on how to modify the polymeric gel chemistry so its expansion factors can be increased approximately 10-fold [26]. To calibrate expansion factors and influences on labeled structures Scheinble and Tinnerfeld measured distances in samples before and after expansion. They did this by applying the expansion on fluorescently marked DNA nanorulers. As a final example, in March 2018 Ewers and his team combined ExM with optical super-resolution STED (calling it ExSTED) on fixed cells and could demonstrate

16

without the need for expensive SRM-techniques, or serial block-face electron microscopy micrometer volume scanning. The center of most of the questions asked in neuroscience is within the understanding of how molecules in neuronal circuits are arranged. In order to investigate these structures there is a need to map the biomolecules across the spatial extents of neurons. With ExM one can visualize synapses and synaptic proteins of many neurons in order to do neural network analyses [30]. The cytoskeleton of a neuron includes the microtubules, neurofilaments and microfilaments. The microtubules are involved in intra-cellular transport, and defects in these structures may lead to neurodevelopmental and neurodegenerative diseases. The microtubules are about 25 nanometers in diameter and thus, the processes related to these diseases take place on the nanoscale. With different versions of ExM it is now possible to resolve microtubular structures in several different sub-cellular structures[2] [28].

17

3. EXPERIMENTAL WORK The proExM experiments were performed on cell cultured neurons from the hippocampus of a mouse. All experiments were performed in accordance with animal welfare guidelines and regulations set forth by Karolinska Institutet. The samples were marked with antibodies and fluorescent markers and imaged mainly with a confocal microscope. When image comparisons were made images were taken from same or similar area of interest, and with the same laser power excitation powers; making it possible to generate image intensity curves could be made that represent the image intensity drop-offs during time. The curves were fitted and analyzed depending on the purpose of the performed experiment. As for the confocal imaging the settings were adapted according to the used dyes and the laser power was set depending on the signal strength. Most images were taken using a 40x/1.3 NA water objective or a 63x/1.4 NA oil objective. In addition to the confocal images, some images were taken using a STED microscope. This was done to obtain the best resolution possible in a sample that already had shown good resolution in the confocal microscope, or to make comparisons between different imaging techniques.

ProExM protocol Fixation. Fix a 37°C prewarmed cell sample in 1 mL 4% formaldehyde (FA) in PBS buffer for 10 minutes. To make 4% FA: dissolve 4% paraformaldehyde (PFA) in PBS in a beaker, add 1 pellet NaOH, stir and heat (80°C) until dissolved and adjust pH to 7.4. Primary antibody staining. Permeabilize cells with 0.5% Triton-X-100 in PBS for 5 minutes then block cells with blocking buffer (with 5% BSA in PBS) for 5 minutes. Incubate cells with primary antibodies diluted in blocking buffer (with 5% BSA in PBS) in desired concentration and leave for ~1 hour in for example a plastic box with a small amount of water on the bottom to keep the environment humid. Then wash in PBS for 5 minutes. Secondary antibody staining. Block cells with a blocking buffer (5% BSA in PBS) for 5 minutes and then incubate cells with secondary antibodies diluted in blocking buffer (with 5% BSA in PBS) in desired concentration and leave for ~1 hour in for example a plastic box with a small amount of water on the bottom to keep the environment humid. Wash in PBS for 20 minutes. (Optional) Secondary fixation. Treat samples with 0.25% glutaraldehyde (GA) for 10 minutes and wash 3x5 minutes in PBS buffer. Anchoring treatment with Acryloyl X-SE: Re-suspend Acryloyl X-SE (6-((acryloyl)amino)hexanoic acid, succinimidyl ester (AcX) in 500 µl anhydrous DMSO at a concentration of 10 mg/mL stock solution (keep in freezer at -20°C). Dilute AcX 1:100 (0.1 mg/mL) in PBS at a concentration of 0.1 mg/mL. Treat the sample for at least 6 hours in room temperature (may be left over night) and then wash 2x15 minutes in PBS buffer.

18

Anchoring treatment with MA-NHS: Treat cells with 1 mM MA-NHS in PBS buffer diluted from DMSO stock (0.018 g MA-NHS in 100 µl DMSO and stored at -20°C). Leave the sample for 1 hour at room temperature. Gelation. To make a 200 µl gelling solution, take 188 µl monomer solution, 4 µl TEMED (accelerator solution) that are kept as stock solution at 10%), and finally add 4 µl APS (initiator solution) that are kept as stock solution at 10% APS). Since these solutions needs to be kept at 4°C preferably mix the solutions on ice. To make the chambers for the gelation step, take a glass slide and put the cell sample in the middle of it and remove excess liquid from sample. Put three coverglasses on each side of the sample and gently dispense ~60 µl gel solution onto the cells (see Fig. 8). Then, carefully place a coverglass on top of the sample without any air bubbles and incubate in 37°C incubator for 1.5-2 hours for gelation. Digestion and expansion. Remove the top coverslip carefully. Put some digestion buffer (50 mM Tris pH 8.0, 1mM EDTA and 0.5 Triton X-100) on the gels to easier be able to remove them from the glass. Leave for 5 min and then gently remove the gels and place them in cell culture wells with digestion buffer. Proteinase K was added to the digestion buffer with 8 Units/mL. The digestion times for proteinase-K treatment may be varied from 30 minutes up to 2 hours, although the sample will lose its fluorescence if treated > 4 hours. Remove the gels from the digestion buffer and place them in deionized (DI) water to allow swelling. The gels it will expand ~4 times so make sure the wells are big enough for the expansion. Change water at least 2 times and wait for at least 1 hour before imaging the gels. Imaging. Cultured neuronal cells. Imaging was performed on a Zeiss LSM780 confocal microscope with a Argon multiline excitation laser (458, 488 and 514 nm) using a 32-Channel GaAsP detector with the 40x 1.30 NA water objective or the 63x 1.40 NA oil objective and a pinhole setting of 1 Airy unit. Some super-resolution microscopy images were taken with a Zeiss Elyra P.1 superresolution photoactivated localization microscopy (PALM) or a Leica SP8 3X STED microscope. The imaging software used to both acquire the images and do the necessary first processing steps was ZEN Black 2012.

Sample preparation techniques Hippocampal mouse neurons were fixed on high precision coverslips (No. 1.5) and kept in a 12 Well Cell Culture Cluster. The samples were blocked and stained with primary (Panorama, Alpha Tubulin, Spectrin, GFAP and Synaptotagmin) and secondary antibodies (Goat anti Mouse 488, Goat anti Mouse 594, Rat anti Mouse 568, Donkey anti Goat 488, Goat anti Guinea Pig 488 and Donkey anti Guinea pig 594). The samples were stained in a box with wet paper in the bottom (to

19

keep environment humid) and accessory fluid was removed by dipping the coverslip on paper. Around ~30 µl of dye was dropped on top of the samples to mark the neurons with the antibodies..

Figure 7. The box used for staining.

The anchoring treatment was done either by using Acryloyl X-SE or MA-NHS and was usually left overnight or more than 6 hours. The gel solution was mixed on ice and APS initiator solution was finally added. For the gelation step, chambers were made by taking a microscope slide (~76x26x1.35mm) and have the sample on the cover slip placed in the center. Thin empty glass coverslips (22x40mm) were then cut in half with help of a crystal cutter and three half glass pieces were put on top of each other on each side of the larger glass slide (see Fig. 8). Around ~70 µl of gelation solution was dropped on top of the sample and finally a whole thin coverslip was placed on top of the sample like a bridge between the cut pieces, which thus sealed the home-made sample chamber.

Figure 8. Gel chamber schematic.

The polymeric gelation step was performed in a Shake’N’Bake Hybridization Oven. The gel was later removed from the coverslip either by removing the top part and placing the bottom part in digestion buffer directly (with the risk of a quite wrinkled gel) or by adding some digestion buffer directly on the sample and gently remove it from the coverslip with a razor blade (with the risk of breaking the gel). Digestion times were varied and by the time the sample was ready to be expanded they were put in a glass bottom microwell dish (MatTek 35 mm petri dish, 14mm Microwell) and DI water was purred into the well. Occasionally a lid was placed over the sample when the sample was imaged, with a few screws as weight to avoid the gels from drifting.

20

Figure 9. A petri dish with a lid. A petri dish with an added lid was used as image sample holder for the gel. The added screws on top were applied as weight to prevent the gel sample from drifting while imaging.

Imaging The majority of the images was taken with a Zeiss LSM780 confocal microscope equipped with an Argon multiline excitation laser (458, 488 and 514 nm) and a 32-Channel GaAsP detector and two photomultiplier tubes- Fluorescence signal was collected by a 40x/1.3 NA water or a 63x/1.4 NA oil immersion objective lens and detected by a photomultiplier tube. Some super-resolution images were taken with a Zeiss Elyra P.1 superresolution photoactivated localization microscopy (PALM) and a Leica SP8 3X STED microscope with a pulsed diode laser. The software used to both acquire the images and do the necessary first processing steps was ZEN Black 2012.

Data analysis and representations To further analyze collected microscope images and data, scripts were written in the MATLAB software. The image and data analysis included making graphs and calculate mean point to point distance from imaged expanded and un-expanded labelled neuronal samples. Imaging processing was also done in Fiji (Fiji Is Just ImageJ) which is an open source image processing package based on ImageJ. Contrast and brightness in images was adjusted, some images required filter processing (median filtering) and scale bars were added using Fiji. An area of choice was selected and by the plugin “plot profile” an intensity graph of Spectrin rings could be obtained and analyzed (cf. Fig. 12). Further image managing was done with Inkscape, an open-source vector graphics editor were different images and graphs were put together into figures.

21

22

4. RESULTS In this section the results from the performed experiments are presented which shows the different outcomes from the varied sample preparation techniques. The presented images and graphs covers the results noted throughout all the experimental optimization performed in this thesis.

Staining The first step was to image neurons stained with different antibodies, in order to check that they worked well together with the proExM protocol. Antibodies used throughout the experiments were anti-Spectrin, anti-Pan-Neuronal marker, Alpha Tubulin and Synaptotagmin. Spectrin is a cytoskeletal protein forming a 2D intracellular periodic network and is crucial for support of the structure and stability of the cell membrane in axons and dendrites. Spectrin is a large heterodimeric protein with sequence motifs called “Spectrin repeats” or “Spectrin rings” and the protein interacts mainly with actin, ankyrin and lipids [33]. The Pan Neuronal marker is an antibody cocktail that binds nuclear, dendritic and axonal proteins distributed across the pan-neuronal architecture [34]. Alpha tubulin forms, together with beta tubulin, the tubulin subunit which is a component of the cytoskeleton that mediate the microtubule action such as intracellular transport. Finally, Synaptotagmin is a major calcium sensor vesicle protein involved in the regulation of neurotransmitter release [35]. As can be seen in Fig. 10, all four protein-staining preparation delivered fully decorated neurons (gray contrast) with the proExM protocol. In A individual microtubular bundles elongate within the thicker dendrites and thinner axons; in B the axon and dendrites all show their content of Spectrin; in C Synaptotagmin is mainly expressed in the thinner axons and their globules boutons; in D the Pan Neuronal marker smoothly labels the neuronal lineage in the mouse hippocampal proExM gel cell culture. In all the images resembles cell culture images rendered without applying the gel proExM protocol, showing that immunolabeling of these four selected protein-staining preparations seems to be compatible with its additional sample preparation steps.

23

Figure 10. proExM imaging of antibodies of interest. Confocal images of expanded neuronal cells stained with (A) Alpha tubulin. (B) Spectrin. (C) Synaptotagmin. (D) Pan Neuronal marker.

24

A comparison between Confocal, STED and Expansion Microscopy As previously mentioned (see background) confocal microscopy is widely used in cell and tissue imaging because of its low cost and convenient usage (e.g. can use common dyes and have down to 200 nm lateral resolution). However, because of the diffraction limit conventional confocal microscopes have a restricted lateral resolution. Super-resolution STED microscopy uses saturated de-excitation of fluorescent dyes to spatially separate individual entities to overcome the resolution limit imposed by diffraction, but this comes with a higher cost and a more technical demanding system. Here we show images of expanded neurons with a resolution similar to the STED images and with great improvement in comparison to the non-expanded image of neurons imaged with conventional confocal microscope. The neurons are all stained with anti-Spectrin and treated with the same preparation techniques (except the water swelling step applied for the expanded samples). As can be seen in Fig. 11, the comparison first shows the stunning periodic ring structures of Spectrin in the neurons. This is resolved optically with super-resolution STED imaging in Fig. 11B, using the very expensive and sophisticated Leica 3X STED microscope in our national bioimaging facility. In Fig. 11C the same stunning sub-diffraction unlimited resolving power of the periodic Spectrin rings is also visualized with expansion, just using conventional confocal microscopy. In Fig. 11A when not using expansion the periodic Spectrin structures are not possible to resolve, as their spacing are below the diffraction limit and resolving power of the confocal microscope.

25

Figure 11. Comparison between confocal, STED and confocal expansion microscopy. We compared images acquired via confocal microscopy versus images acquired via STED microscopy and post-expansion confocal microscopy. All samples are stained with Spectrin. (A) Non-expanded sample imaged with a confocal microscope. (B) Non-expanded sample imaged with a STED microscope. (C) Expanded sample imaged with a confocal microscope.

26

Resolving the periodic expression of Spectrin To evaluate and also to compare the expansion microscopy data to super-resolution imaging data published a few years ago, the periodic Spectrin rings structures was measured. A mean peak to peak distance was obtained from an image of expanded neurons stained with Spectrin, presenting the average distance between the Spectrin rings. The measured distance was 738 nanometers (see Fig. 12 below) and by comparing this result with known literature (where Spectrin rings distances are measured to be around ~180 to 190 nanometers [36]) the estimated expansion was calculated to be around 3.9 to 4.1 times (738/180 = 4.1, 738/190 = 3.9). When measuring the gel size with a ruler before and after expansion it was found that the magnification was around 3.9 to 4.2 times, which thus correlated well with the in-sample measured expansion factor of the Spectrin ring periodicity. The macroscopic gel expansion measurement also correlate with previous work in the lab, as well as expansion microscopy results published in the literature [15].

Figure 12. Periodic expression of Spectrin in neurons. (A) A intensity plot of the marked area in (C). (B) Image of expanded neurons stained with Spectrin. (C) A zoomed-in area of (B) were the area of interest is marked in yellow.

27

Variation of the digestion time To evaluate the optimal enzymatic digestion step, which in first principle only removes protein, fat and lipids but leaves the labeling structures decorating the gel untouched, different digestion times were tested. As can be seen in Fig. 13, the digestion time hade a huge influence on how well the labeling were preserved for the neuronal topology. We here imaged non-expanded neurons immunostained for Spectrin, anchored with AcX and digested with proteinase K for different durations (15 minutes, 1 hour, 2 hours and 4 hours). Digestion times over 15 minutes seems to severely deter the remaining fluorescence signal of the sample, thus pointing towards that the enzymatic digestion also removes a lot of the immunolabeling of the proExM protocol. Microscope parameters such as laser power and detector gain were kept the same for all measurements. This problem of a huge decreased fluorescence signal as function of digestion time has not been pointed out in the literature. Instead long digestion times has been applied and reported to by just beneficial. Most data has however been generated from imaging large multicellular structures. Here we have analyzed sub-cellular structures like the periodic Spectrin rings structures in dendrites. If we instead look on a multicellular level we see a somewhat slower drop in lost signal as function of digestion times (see Fig.14). However, there is as shown Figure 14, an increased intensity drop-off in preserved fluorescence intensity in these samples too with longer digestion times. To further measure the presumed intensity decrease with digestion time, we performed a real time experiment in which the sample was digested (with proteinase K) whilst being imaged in the confocal microscope. The sample was stained with anti-Spectrin and imaged every 10th minute for 6 hours. There was a 40% loss of signal intensity after ~6 hours (see Fig. 14 A). Digestion times will thus be a crucial step to control if the proExM protocol is going to be well suited for sub-cellular topological studies. Additionally, neurons stained with Pan Neuronal markers, anchored with Acryloyl X-SE (AcX) and digested with Proteinase K overnight, also lost all its fluorescence and could not be imaged. Thus pointing towards that too long digestion in general terms deter the labeled sample topology.

28

Figure 13. Neurons with different digestion times. Neurons treated with Spectrin, Goat anti-Mouse (Atto) 594 and Acryloyl X -SE and imaged with a confocal microscope. (A) Digestion time: 15 min. (B) Digestion time: 1 hour. (C) Digestion time: 2 hours. (D) Digestion time: 4 hours.

29

Figure 14. Intensity drop off with digestion time. Expanded neuronal cells stained with Spectrin, digested for different times and imaged with a confocal microscope. (A) Profile of intensity taken at different time points. (B) Confocal image of neuronal cells not digested with proteinase K. (C) Confocal image of neuronal cells digested with proteinase K for 24 hours.

30

A comparison between Acryloyl X -SE and MA-NHS To evaluate if different anchoring treatments could influence the decrease seen in fluorescence signal with longer digestion time, we next compared images of samples anchored with MA-NHS to images of samples anchored with Acryloyl X -SE. Figure 15 shows the comparison when we stained the neuronal cells with anti-Pan-Neuronal and digested for 0 minutes, 2 hours, 4 hours and 24 hours. By measuring the mean intensity of the images over time, it was clear that the samples anchored with MA-NHS showed much greater intensity stability over time in comparison with the Acryloyl X -SE anchoring samples, where there was again a clear decrease in signal intensity over time. This last result thus points towards that MA-NHS anchors sample labeling much more firmly. In addition, when comparing the non-digested samples anchored with MA-NHS to non-digested samples anchored with Acryloyl X -SE, we found that there were more inhomogeneities in the sample anchored with Acryloyl X -SE (see Fig. 16). All these sample preparation artefacts needs to be more studied and better controlled, in order to find the best protocols when applying expansion microscopy for sub-cellular topological trustworthy imaging. This is no news in biological imaging as all preparation steps need to preserve the studied structures in order to be of use.

31

Figure 15. Comparison between anchoring treatment with MA-NHS and Acryloyl X-SE. Non-expanded neurons stained with Pan Neuronal, anchored with either MA-NHS or Acryloyl X -SE and imaged with a confocal microscope. (A, B, C and D) Neurons anchored with MA-NHS and digested for (A) 0 minutes, (B) 2 hours, (C), 4 hours and (D) 24 hours. (E, F, G and H) Neurons anchored with Acryloyl X-SE and digested for (E) 0 minutes, (F) 2 hours, (G), 4 hours and (H) 24 hours. (I) Graph showing the image intensity over different digestion times in neurons anchored with MA-NHS. (J) Graph showing the average image intensity over different digestion times in neurons anchored with Acryloyl X-SE.

32

Figure 16. Comparison between anchoring treatment with MA-NHS and Acryloyl X -SE. Expanded neuronal cells stained with anti-Pan-Neuronal marker, imaged with a confocal microscope and anchored with (A) MA-NHS or (B) Acryloyl X-SE. The asterisk in panel B indicates a neuronal structure which has not expanded uniformly, leaving an apparent break in an otherwise continuous axon or dendrite.

33

5. DISCUSSION Expansion Microscopy provides an additional option to overcome the diffraction limit to generate super-resolved images, and the method is relatively easy to use compared to other super-resolution techniques. In this report we investigated the proExM protocol for imaging cultured neuronal cells. We first examined the labeling of Spectrin, Pan-Neuronal marker, Alpha Tubulin and Synaptotagmin. All of these antibodies yielded good image quality (Fig. 10). We then compared Confocal, STED and Expansion Microscopy and we obtained similar image quality in the STED and Expansion Microscopy, with an obvious improvement over the image quality obtained in the conventional confocal microscopy image with higher signal of the labeled structures for example (Fig. 11). To quantify the expansion factor of the gel, we expanded neuronal cells stained with Spectrin and measured the mean peak-to-peak distance and compared the result with known measurements from literature on non-expanded neurons stained with Spectrin. This result showed an expansion factor of 3.9 to 4.1 times which yielded a good approximation of the physical expansion and was consistent with the results obtained by measuring macroscopic swelling by means of a ruler. However, depending on the chosen area of interest, this result might differ slightly and there is no comparison done with the Spectrin rings distances before expansion in the same sample. STED microscopy however allows such a ground truth comparison for a non-expanded sample, before swelling take place and confocal (non-STED) imaging is done. The next step of this thesis was to evaluate how the digestion time affected the image intensity. First, we found that in different samples digested for different time scales there was a decrease in image intensity. However, this result was not trustworthy since the intensity differences could arise because of different staining quality in different samples or because there were different areas of the samples imaged. To do further investigations we imaged the same sample and the same region of interest over time when digested with Proteinase K. Again, we found that there was a significant reduction in image intensity and from these studies we draw the conclusion that samples anchored with AcX should not be digested for more than 1-2 hours in order to keep necessary intensity for imaging. One can argue that the intensity decrease also depends on the fact that the fluorescence is bleached by every image that was taken. However, one imaging procedure lasted for about 5 seconds and was taken with a 1% laser power which should generate such severe difference for the image intensity decrease. Also, because we had a lid on the sample (see Fig. 9) and used a smaller amount of digestion buffer to enable imaging of the sample, it could have affected the time for Proteinase K to diffuse into the whole sample. It could also have affected the digestion compared to a sample not digested with a lid. The samples were, according to this fact, not treated exactly as it would have been treated in a regular well. In addition, the stability of the microscope and the temperature in the sample could also have affected the digestion procedure.

34

To further investigate the intensity changes with digestion, we also anchored neurons with MA-NHS instead of AcX. We did a comparison between MA-NHS and AcX anchoring treatments and imaged the samples with different digestion times. We showed that samples anchored with MA-NHS were less affected by the digestion time and did not lose at much intensity at all over time. Also, we noticed that we could not find topological breaks in non-digested samples anchored with MA-NHS in comparison with the non-digested samples treated with AcX (see Fig. 16). Apart from all the previously mentioned advantages with ExM, there are plenty of difficulties with the technique that needs to be considered. The method is still dependent on perfect staining and sample preparation, plus the reliability of the isotropic expansion process at a molecular level remain modest. As with most new technologies, one might need to confront potential problems to organize expansion microscopy for the desired application. Depending on the purpose of the experiment there are some factors that might need to be changed, as for example the digestion and denaturation step. Overcoming drift is another problem that needs to be solved and in this report, this problem has been managed by attaching the samples by using a lid. In summary, proExM has shown to be useful as it physically magnifies samples homogeneously and have many applications in imaging of biological samples with nanoscale precision. The technique has demonstrated its usefulness in many different neuroscientific contexts. Using our findings as a connection to previously published work we suggest that MA-NHS is used for anchoring treatment and in cases where AcX is used, there should not be any longer digestion times than tens of minutes in order to keep the majority of the fluorescence signal. To push the ExM technique to the next level one need to overcome the retention of signal intensity. The results summarized in this thesis are a small addition to the optimization of the proExM protocol, but can hopefully be used in future research. The complex anchoring chemistry and gel mesh polymerization steps, with added enzymatic digestion steps, before or after expansion and lebeling is a complex parameter space to optimize. There are thus plenty of unanswered question related to this work that needs to be further investigated.

35

6. SUPPLEMENTARY

Figure S 17. Schematic of refractive image formation (magnified, real and inverted image), of an object placed at distance a in-front of a thin lens having focal length f. The imaged is formed at distance b and the magnification is given as M=b/a

Figure S 18. An AIRY disc diffraction pattern of a point source. The point source is here the object and it will be imaged by the optical system as a wiggly pattern with a main central peak and neighboring smaller ringing, because of the wave nature of light (in other words diffraction is scattering and inference of light waves).

36

Figure S19. Schematic of focal spot dimension of a fluorescent microscope. The excitation wavelength is used to excite fluorescent molecules pictured as small orange discs: However, none of the fluorescent molecules (point sources) will be resolved, as their distance is below the Abbe limit. They will instead all be merged into a single focal blob of width ~200 nm and height of 3-4 times this value in the very best diffraction-limited case.

Figure S 20. Schematic of STED microscopy. (Left) Fluorescent molecules are excited with the blue laser focus. Overlaying the excitation focus with the red STED focus allows one to optically switch excited fluorescent molecules into their ground state by stimulated emission. In-center molecules then deliver their fluorescence (green) at the targeted position. (Right) Real image of one fluorescent molecule attached to a glass surface localized and resolved with STED compared to confocal imaging (i.e. no STED). Scale bar: 200 nm.

+ =

Exc. STED no STED STED

37

7. APPENDICES MATLAB script

ThemeasurementofthepeaktopeakdistancescriptwritteninMATLAB.Itrequiresatxtfileinput.

profile = dlmread('Lineplot1.txt','', 1,0) figure; [pks,locs,w,p] = findpeaks(profile(:,2), profile(:,1)) plot(profile(:,1), profile(:,2),locs, pks, 'o') ylabel('Intensity [AU]','FontSize', 15) xlabel('Distance [\mum]', 'FontSize', 15) pp = diff(locs); meanpp = mean(pp)

38

8. REFERENCES [1] J. S. Silfies, S. A. Schwartz and M. W. D. W., "Super Resolution: The Diffraction Barrier

in Optical Microscopy," Nikon Instruments Inc., 2018. [Online]. Available: https://www.microscopyu.com.

[2] V. v. Batenburg, "Mapping the Neuronal Cytoskeleton with Expansion Microscopy," 2017. [3] "Beyond the diffraction limit," Nature Photonics, July 2009. [4] Y. Zhao, O. Bucur, H. Irshad, F. Chen, A. Weins, A. L. Stancu, E.-Y. Oh, M. DiStasio, V.

Torous, B. Glass, I. E. Stillman, S. J. Schnitt, A. H. Beck and E. S. Boyden, "Nanoscale imaging of clinical specimens using pathology-optimized expansion microscopy," nature biotechnology, 17 July 2017.

[5] R. Gao, S. M. Asano and E. S. Boyden, "Q&A: Expansion Microscopy," BMC Biology, 2017.

[6] Thermo Fisher Scientific, "Process of Fluorescence," Thermo Fisher Scientific, 2018. [Online]. Available: http://www.thermofisher.com/.

[7] Chemistry and Light School Didactic Kit, "Fluorescence," Chemistry and Light, 2018. [Online]. Available: http://www.chemistryandlight.eu/.

[8] T. J. Fellers and M. W. Davidson, "Introduction to Confocal Microscopy," Olympus Corp., 2018. [Online]. Available: https://www.olympus-lifescience.com.

[9] S. W. Paddock, T. J. Fellers and M. W. Davidson , "Introductory Confocal Concepts," Nikon Intruments Inc., 2018. [Online]. Available: https://www.microscopyu.com.

[10] E. F. Fornasiero and F. Opazo, "Super-resolution imaging for cell biologists," Bioessays, 2015.

[11] "Beyond the diffraction limit," Nature Photonics, July 2009. [12] G. Vicidomini, P. Bianchini and A. Diaspro, "STED super-resolved microscopy," Nature

Methods, 29 January 2018. [13] "Life Science: Stimulated Emission Depletion Microscopy (STED)," 2018. [Online].

Available: https://www.picoquant.com/. [14] A. Bradford, "What is a Polymer?," Purch Copyright Agent, 13 October 2017. [Online].

Available: https://www.livescience.com/. [15] F. Chen, P. W. Tillberg and E. S. Boyden, "Expansion Microscopy," Research, vol. 347,

no. 6221, 30 January 2015. [16] J. Markoff, "Expansion Microscopy Stretches Limits of Conventional Microscopes," 19

January 2015.

39

[17] F. Chen, A. T. Wassie, A. J. Cote, A. Sinha, S. Alon, S. Asano, E. R. Daugharthy, J.-B. Chang, A. Marblestone, G. M. Church, A. Raj and E. S. Boyden, "Nanoscale imaging of RNA with expansion microscopy," Nature Methods, 4 July 2016.

[18] P. W. Tillberg, F. Chen, K. D. Piatkevich, Y. Zhao, C.-C. Yu, B. P. English, L. Gao, A. Martorell, H.-J. Suk, F. Yoshida, E. M. DeGennaro, D. H. Roossien, L. Gong, U. Seneviratne and S. Tannenbaum, "Protein-retention expansion microscopy of cells and tissues labeled using standard fluorescent proteins and antibodies," Nature Biotechnology, 4 July 2016.

[19] J.-B. Chang, F. Chen, Y.-G. Yoon, H. Babcock, J. S. Kang, S. Asano, H.-J. Suk, N. Pak, P. W. Tillberg, A. T. Wassie, D. Cai and E. S. Boyden, "Iterative expansion microscopy," Nature Methods, 17 April 2017.

[20] L. Freifeld, I. Odstrcil, D. Förster, A. Ramirez, J. A. Gagnon, O. Randlett, E. K. Costa, S. Asano, O. T. Celiker, R. Gao, D. A. Martin-Alarcon, P. Reginato, C. Dick, L. Chen, D. Schoppik and Enge, "Expansion microscopy of zebrafish for neuroscience and developmental biology studies," PNAS, 21 November 2017.

[21] D. Gambarotto, F. U. Zwettler, M. Cernohorska, D. Fortun, S. Borgers, J. Heine, J. G. Schloetel, M. Reuss, E. S. Boyden, M. Sauer, V. Hamel and P. Guichard, "Imaging beyond the super-resolution limits using ultrastructure expansion microscopy (UltraEx;)," bioRxiv, 25 April 2018.

[22] T. J. Chozinkski, A. R. Halpern, H. Okawa, H.-j. Kim, G. J. Tremel, R. O. Wong and J. C. Vaughan, "Expansion Microscopy with conventional Antibodies and fluorescent proteins," Nature Methods, 1 June 2016.

[23] A. R. Halpern, G. C. Alas, T. J. Chozinski, A. R. Paredez and J. C. Vaughan, "Hybrid Structured Illumination Expansion Microscopy Reveals Microbial Cytoskeleton Organization," ACS Nano, 22 November 2017.

[24] N. Jiang, H.-J. Kim, T. J. Chozinski, J. E. Azpurua, B. A. Eaton and J. Z. Parrish, "Super-resolution imaging of Drosophila tissues using expansion microscopy," Mol Biol Cell, 24 April 2018.

[25] L. Pesce, I. Cainero, M. Oneto, M. Cozzolino, L. Lanzano, A. Diaspro and P. Bianchini, "Expansion and STED nanoscopy a new tool for pushing the resolution at the limit, the fluorescent label," 2017.

[26] S. Truckenbrodt, M. Maidorn, D. Crzan, H. Wildhagen, S. Kabatas and S. O. Rizzoli, "X10 Expansion Microscopy Enables 25 nm Resolution on Conventional Microscopes," bioRxiv, 11 August 2017.

[27] M. Gao, R. Maraspini, O. Beutel, A. Zehtabian, B. Eickholt, A. Honigmann and H. Ewers, "Expansion stimulated emission depletion microscopy (ExSTED)," bioRxiv, 8 March 2018.

[28] "A new way to study the brain's invisible secret," TEDx Talks, 2016. [Online].

40

[29] E. D. Karagiannis and E. S. Boyden, "Expansion microscopy: development and neuroscience applications," ScienceDirect, 2018.

[30] L. L. Kirkpatrick and S. T. Brady, "Molecular Components of the Neuronal Cytoskeleton," Basic Naurochemistry: Molecular, Cellular and Medical Aspects, 1999.

[31] R. Zhang, C. Zhang, Q. Zhao and D. Li, "Spectrin: Structure, function and disease," Science China, December 2013.

[32] N. Stefanakis, I. Carrera and O. Hobert, "Regulatory logic of pan-neuronal gene expression in C. elegans," Neuron, 2015.

[33] Y. Lai, J. Diao and Y.-K. Shin, "Molecular origins of synaptotagmin activities on vesicle docking and fusion pore opening," Scienfific Reports, March 2015.

[34] K. Xu, G. Zhong and X. Zhuang, "Actin, spectrin and associated proteins form a periodic cytoskeletal structure in axons," Science, November 2013.

[35] D. Unnersjö-Jess, L. Scott, S. Zambrano Sevilla, J. Patrakka, H. Blom and H. Brismar, "Confocal super-resolution imaging of the glomerualr filtration barrier enabled bu tissue expansion," Kidney International, 17 August 2017.

[36] M. B. Scheible and P. Tinnefeld, "Quantifying Expansion Microscopy with DNA Origami Expansion Nanorulers," bioRxiv, 14 February 2018.

[37] "Glomerular Disease," UNC Kidney Center, 2018. [Online]. Available: https://unckidneycenter.org/kidneyhealthlibrary/glomerular-disease/.

[38] Khan Academy, "Renal physiology: Glomerular filtration," 2018. [Online]. Available: https://www.khanacademy.org/test-prep/mcat/organ-systems/the-renal-system/a/renal-physiology-glomerular-filtration.