High-content 3D multicolor super-resolution localization microscopy 7 Pedro M. Pereira, Pedro Almada, Ricardo Henriques1 MRC Laboratory for Molecular Cell Biology and Department of Cell and Developmental Biology, University College London, London, UK 1Corresponding author: E-mail: r.henriques@ucl.ac.uk CHAPTER OUTLINE Introduction.............................................................................................................. 96 The Basis of an SMLM-Imaging Experiment ................................................................ 99 Hybridizing SMLM with High-Content Imaging........................................................... 101 1. Sample Preparation............................................................................................ 102 1.1 Equipment ......................................................................................... 102 1.2 Materials ............................................................................................ 102 1.3 Method .............................................................................................. 103 1.3.1 Cleaning slides and coverslips........................................................... 103 1.3.2 Labeling with primary antibodies....................................................... 104 1.3.3 Seeding the cells .............................................................................. 105 1.3.4 Immunofluorescence........................................................................ 105 1.3.5 Coverslip mounting........................................................................... 106 2. Imaging Acquisition and Image Analysis.............................................................. 107 2.1 Equipment ......................................................................................... 107 2.2 Software............................................................................................. 108 2.3 Method .............................................................................................. 108 2.3.1 Acquisition ....................................................................................... 108 2.3.2 Single-particle detection and reconstruction ...................................... 110 2.4 Important considerations ..................................................................... 110 2.4.1 Detectors used for the acquisition ..................................................... 110 2.4.2 Single-molecule detection and localization......................................... 111 2.4.3 SR drift correction and chromatic realignment................................... 112 2.4.4 SR estimation and image reconstruction ........................................... 112 CHAPTER Methods in Cell Biology, Volume 125, ISSN 0091-679X, http://dx.doi.org/10.1016/bs.mcb.2014.10.004 © 2015 Elsevier Inc. All rights reserved. 95 ----!@#$NewPage!@#$---- Conclusions and Outlook ......................................................................................... 112 Acknowledgments................................................................................................... 113 References ............................................................................................................. 113 Abstract Super-resolution (SR) methodologies permit the visualization of cellular structures at near-molecular scale (1e30 nm), enabling novel mechanistic analysis of key events in cell biology not resolvable by conventional fluorescence imaging (w300-nm resolution). When this level of detail is combined with computing power and fast and reliable analysis software, high-content screenings using SR becomes a practical option to address multiple biological questions. The importance of combining these powerful analytical techniques cannot be ignored, as they can address phenotypic changes on the molecular scale and in a statistically robust manner. In this work, we suggest an easy-to-implement protocol that can be applied to set up a high-content 3D SR experiment with user-friendly and freely available software. The protocol can be divided into two main parts: chamber and sample preparation, where a protocol to set up a direct STORM (dSTORM) sample is presented; and a second part where a protocol for image acquisition and analysis is described. We intend to take the reader step-by-step through the experimental process highlighting possible experimental bottlenecks and possible improvements based on recent de- velopments in the field. INTRODUCTION Research in cell biology has relied on different techniques to answer central ques- tions that have long puzzled life scientists. Within this realm, fluorescence light mi- croscopy (FLM) has been a critical tool, as it provides visual insights into cellular structure and molecular mechanisms. Central to FLM is its capacity to allow the visualization of cellular processes by the utilization of highly specific labeling tools. Nonetheless, the resolution of conventional FLM is limited by optical diffraction to around 200e300 nm laterally and 500e700 nm axially (Abbe, 1873). Consequently, obtaining nanoscale informationdresolving small-scale molecular assemblies or direct molecular interactiondis difficult to achieve with standard FLM. Due to the diffraction resolution limit in optical microscopy, any individually imaged fluorophore will appear as a spot with diameter equal or superior to 200e300 nm (Henriques & Mhlanga, 2009). In most fluorescence-imaging experi- ments, fluorophores cluster together at close proximity, generally at a distance less than 200e300 nm (the diffraction limit). As their spatial profile overlaps, it becomes difficult to resolve individual fluorophores or structures they compose within the blurred image. This problem has been recently overcome with the advent of several super-resolution (SR) microscopy techniques (Schermelleh, Heintzmann, & Leonhardt, 2010), such as structured illumination microscopy (SIM) (Gustafsson, 96 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- 2000; Gustafsson et al., 2008), stimulated emission depletion microscopy (STED) (Klar, Jakobs, Dyba, Egner, & Hell, 2000), and single-molecule localization micro- scopy (SMLM). One such technique, SMLM, negates this problem by locating in- dividual fluorophores on a sequence of fluorescence microscopy images (Betzig et al., 2006; Hess, Girirajan, & Mason, 2006; Rust, Bates, & Zhuang, 2006) allowing resolutions between 1 and 30 nm. This finally permits researchers to probe deeper into the mechanisms of cellular processes (Herbert, Soares, Zimmer, Henriques, & et al., 2012; Yildiz & Selvin, 2005). Several SMLM methods have been developed in recent years, such as stochastic optical reconstruction microscopy (STORM) (Rust et al., 2006), direct STORM (dSTORM) (Heilemann et al., 2008), photoacti- vated localization microscopy (PALM) (Betzig et al., 2006), and fluorescence PALM (FPALM) (Hess et al., 2006), which mainly differ on the target labeling strategy, method to induce photoswitching, and imaging protocol (Herbert et al., 2012). SMLM enables resolutions approaching those previously restricted to methods such as electron microscopy (EM) and atomic force microscopy, while retaining most of the advantages of FLM (Herbert et al., 2012). This family of methods relies on inducing fluorophore photoswitching in a manner that only allows a small random subset of fluorophores to transiently emit light (Betzig et al., 2006; Hess et al., 2006; Rust et al., 2006). If the emission profile of fluorophores does not spatiotemporally overlap extensively, it becomes possible to pinpoint each individual emitter with a precision between 1 and 30 nm (Thompson, Larson, & Webb, 2002). A typical SR microscopy experiment workflow does not differ much from a classical microscopy experiment; it requires a microscope (Figure 1(A)), sample preparation (Figure 1(B)), a support for imaging (Figure 1(C)), and data analysis (Figure 1(D)). However, these experimental steps increase in complexity with the increase in resolution. For example, in order to obtain an SR image, usually 1000 to 50,000 images are required, and careful sample preparation is paramount, which can result in long sample prep- aration and image acquisition times (Herbert et al., 2012). The increase in resolution from SR reveals new levels of information at the molec- ular level. Not only can structures be resolved with greater detail, but their molecular components can now be located and identified. This permits greater detail in func- tional analysis of structural and signaling components of the cell. For example, the organization of T-cell signaling nanodomains (Soares et al., 2013) was recently revealed by SMLM. SMLM was also used to unravel the location of signaling com- ponents in the brain’s chemical synapses (Dani, Huang, Bergan, Dulac, & Zhuang, 2010). In WeibelePalade bodies, STORM is the only technique so far capable of directly revealing the presence of the newly described von Willebrand Factor (vWF) quanta, the subunits from which these pro-hemostatic endothelial organelles are assembled (Ferraro et al., 2014). Additionally, SR microscopes are increasingly automated, which simplifies SR acquisition. Most commercial and custom-built sys- tems have fully automatic acquisition capabilities. An automated system is precisely one of the requirements for high-content screening (HCS) and it is immediately apparent that an increase in spatial information can be coupled to an increase in infor- mation at the population level in a straightforward manner. Such high-content assays Introduction 97 ----!@#$NewPage!@#$---- 98 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- generally entail the observation of a large number of cell populations under different conditions. On an imaging HCS, the workflow focuses on acquiring multiple fields-of- view (FOVs) to capture a statistically significant cellular sample. Hybridizing this pro- cedure with SMLM entails a protocol where, for each FOV, a large number of frames need to be captured while inducing transient photoswitching of fluorophores in the sample. Here, we further discuss the principles of how a joint SR HCS protocol can be set up, following some of the procedures shown on our recent publications. THE BASIS OF AN SMLM-IMAGING EXPERIMENT At the core of an SMLM experiment is the need to label a sample with appropriate photoswitchable fluorophores. Most fluorophores emit light continuously when illu- minated in normal conditions (either physiological conditions or in standard FIGURE 1 Schematic representation of a single-molecule localization microscopy (SMLM) experiment. (A) Schematic representation of the super-resolution (SR) microscope described. Laser light from n lasers is combined using the mirror (M) and an n number of dichroic beam-splitters (DM, DMn). The beams are modulated using an acousto-optic tunable filter (AOTF) and through a series of lenses (L1eL3) the beams are directed into the 100X high-numerical aperture objective (TIRF). The emission light from the sample goes through DMn þ 1, and a cylindrical lens (CL), which will introduce an astigmatic aberration into the image. This allows the algorithm to determine the 3D localization of the detected particles. The correspondent light is then redirected by a mirror (M4), magnified by a tube lens (L4) and filtered by an emission filter (F) before reaching the camera where the signal is detected (adapted from Henriques et al. (2010)). (B) Schematic representation of sample preparation procedure. The cells are seeded on the coverslip, fixed and labeled with a directly labeled primary antibody, as the use of primary plus secondary antibody introduces a higher linker error (exemplified by the thicker lines in the image). In order to correct for drift and chromatic shift aberrations, fiducial markers are added to the sample. (C) Schematic representation of the chamber assembly procedure. After sample preparation, the coverslip is mounted on a chamber in order to have an airtight lock between the oxygen-scavenger buffer and the sample (see “coverslip mounting” for more details). (D) Representation of the acquisition and image analysis procedure. In SMLM, the microscope acquires a sequence of diffraction- limited images where only a small subset of spatially spaced fluorophores is switched on at each frame. The first frames will have a high amount of naturally active photoswitchable fluorophores (which hinders the immediate identification of single molecules). To solve this, a bleaching step is first performed (Acquisitiondt0). We then stimulate the sample with an appropriate light source and acquire the sequence of images (Acquisitiondt1, t2.). Using localization algorithms, these frames are then analyzed, which results in an accurate fluorophore detection and localization (Analysisdt1, t2.). After drift and chromatic shift correction (the later for multicolor acquisitions), an SR reconstruction is obtained from all the localizations (AnalysisdReconstruction). Adapted from Herbert et al. (2012). = Introduction 99 ----!@#$NewPage!@#$---- imaging media). Generally, they only stop when permanently bleached, at which point they cannot emit light again. But there are certain proteins and organic fluoro- phores, which can enter transient dark states either through conformation changes (proteins) or by virtue of their photochemistry (organic fluorophores and some pro- teins). PALM techniques mostly rely on photoswitchable proteins whereas STORM techniques make use of organic fluorophores. Photoactivatable fluorescent proteins do not emit light until they are induced to change conformation when hit by a pulse of light of a certain wavelength. These include photoactivatable green fluorescent protein (PA-GFP), mEOS and Dendra2 (Lippincott-Schwartz & Patterson, 2009). The SMLM techniques of PALM and FPALM take advantage of this and work by starting the experiment with a sample tagged with inactivated proteins. By shining a very short pulse of activating light, usually UV, only a small subset of the proteins will be activated. These are imaged and then bleached permanently, enabling a new cycle to be acquired. After several thousand cycles, the images will be analyzed and an SR reconstruction will be obtained. For a nice example of PALM microscopy, see Truong Quang and Lenne (chapter 8 of this volume). STORM makes use of the fact that most organic fluorophores can be switched to a transient nonemitting state when they are illuminated with sufficiently strong laser light. This is distinct from photo- bleaching, which is permanent. The switching dynamics of the fluorophore can also be modulated with chemicals that alter the length of each cycle of emitting/dark state. Once a sample is stained with fluorophores, images can be acquired with opti- mized imaging protocols. We start by switching off the majority of the fluorophores in the sample by intensely illuminating with a wavelength compatible with the fluo- rophore excitation (e.g., 640 nm for Cy5). This step guarantees that most initial un- wanted background fluorescence in the sample is erased (bleaching phase, Figure 1(D)). Next, a second wavelength (e.g., 488 nm for Cy5) can be used at lower intensity to randomly photostimulate a small subpopulation of fluorophores into a fluorescent state. This step may be skipped since fluorophores also have a tendency to spontaneously recover from dark states under appropriate media (activation, Figure 1(D)). Imaging is done with high-intensity illumination (e.g., 640 nm for Cy5) causing fluorophores to emit for a few milliseconds and then undergo a tran- sient bleaching with seconds to minutes duration (read out, Figure 1(D)). The critical feature in this protocol is that the photoactivation (low intensity) and photoexcitation (high intensity) need to be optimized in order to minimize emitting fluorophores from overlapping, but still have a sufficiently large number of fluorophores visible in each image. The goal then becomes to capture a large set of images at high frame rates (10e100 Hz), which feature a population of photoswitching fluorophores large enough to represent the majority of labels in the sample (for microtubule staining in HeLa cells, this would be w10,000 frames containing millions of individually iden- tifiable particles). SR image analysis is carried out by an algorithm responsible for detecting and local- izingeachfluorophorepresentoneachframethroughpatternrecognition(particledetec- tion and localization, Figure 1(D) (Herbert et al., 2012). By listing features such as 100 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- position and intensity of all detected fluorophores, the algorithm can then plot all the particle centers in an image, creating an SR reconstruction (analysis, Figure 1(D)). HYBRIDIZING SMLM WITH HIGH-CONTENT IMAGING Merging high-content imaging with the SMLM protocol above requires automation procedures capable of acquiring SMLM image sequences in a large set of FOVs. This is achievable with most SR microscopes as they generally offer complete auto- mation of acquisition. While most SMLM systems can do HCS, not all HCS systems can do SMLM, as SR requires optimized cameras and illumination systems. Despite this, there is a large variety of choice of complete commercial systems, each with its own acquisition software which will provide the option to image multiple FOVs. In the case where the software does not offer this choice, the Micro-Manager software package (Edelstein, Amodaj, Hoover, Vale, & Stuurman, 2010) can freely control a large set of commercial microscopes currently available, enabling customized and automatic acquisition procedures. An SR HCS-specific issue is the need for appropriate imaging media. Given the need forthefluorophorestotransitionbetweenONandOFFstates(reversiblephotoswitching), it is important to control permanent bleaching of fluorophores. This can be achieved by controlling the presence of oxygen radical species in solution, known to be associated withpermanentbleaching.Thisisusuallyachievedusingeitheroxygen-scavengerbuffers (Heilemann et al., 2008) or thiol-containing buffers (e.g., b�mercaptoethylamine) (Van de Linde et al., 2011), the latter being more common due to the less demanding protocol. One other solution is the use of a permanent mounting medium such as Vecta- shield, with the trade-off that it will introduce a large background in the blue, green, and orange wavelengths, making it only suitable for red-emitting fluorophores. Themicroscopewillalsodeterminehowmanysamplescanbeacquired,astheywill either accept multiwell plate, slides with coverslips, or cell culture dishes. For HCS, multiwell plates can be used, but for SR, the cells need to be placed on a 170-mm-thick glass substrate, as this is the coverslip thickness that will provide the ideal imaging con- ditions. Unfortunately, coverslip-bottomed multiwell plates can be prohibitively expensive and perhaps unnecessary for smaller screens. In cases where fewer condi- tions will be required, there are coverslip-bottomed multiwelled culture dishes with up to 18 wells, reducing the amount of sample loading required from the experimenter. Finally, several standard coverslips per slide can be used for small screens, or even when using RNAi spotting techniques. These spotting techniques provide the option of observing multiple conditions on a single coverslip (Erfle et al., 2007). Given that the simplest SR microscopes are sold with attachments for the stan- dard combination of slide and coverslip, we will show a protocol that uses standard coverslips, but the large majority of what will be mentioned is applicable to multiw- elled dishes or plates. Aside from glass thickness, another important consideration is that the bottom be as clean as possible as dirt and fluorescent particles will cause spurious detections as well as severely degrade imaging quality. Introduction 101 ----!@#$NewPage!@#$---- A SR HCS acquisition will be considerably slower when compared to a low- resolution HCS assay. This is due to the need to acquire several thousand images for each FOV. Additionally, the traditional electron-multiplying charge-coupled de- vice (EMCCD) cameras used in SR have small FOVs, which reduces the amount of information extracted from each image. Finally, a HC screen will also require auto- mated data analysis. This is a complex topic far beyond the scope of this chapter. However, the freely available ImageJ/Fiji (Schindelin et al., 2012) software analysis package provides several plugins which can perform cell segmentation and measure- ments of batches of data in an automatic manner. It is worth noting that SR greatly increases the data size of each raw acquisition. For example, using ImageJ/Fiji, we have recently carried out a small semiautomated screen which captured 28 drug- induced conditions (8e15 cells per condition), over 12.5 days of continuous imag- ing. This screen generated >6 million untreated images and a total of 31 terabytes of data (Soares et al., 2013). 1. SAMPLE PREPARATION In this section, we will describe a detailed protocol for analyzing cellular structures by SR. Any SR-imaging experiment is similar to a classical FLM experiment. How- ever, single-molecule imaging requires the user to address several experimental de- tails that are usually overlooked when working under diffraction-limited conditions. We will start by describing how to clean your slides/coverslips, followed by a pro- tocol on how to directly label your antibodies. We will then describe how to seed the cells to be visualized and an example of how to label your targets of interest will be described. Finally, a chamber assembly strategy where your sample can be imaged by SR will be shown. These steps have a description of the possible experimental bottlenecks you may face while defining your experiment and important details that the user should consider before starting an experiment. As the sensitivity of the methodology is much higher than with classical FLM, every step should be care- fully planned, which will facilitate the analysis of the final data. 1.1 EQUIPMENT Heating plate (Corning, 6795-400D) Sonicator (Bransonic tabletop ultrasonic cleaner; Branson) Platform rocker (Thermo Scientific, M79735Q) 1.2 MATERIALS High Performance Coverslips (18 � 18 mm, 1.5H, Carl Zeiss, 474030-9000-000) Glass slides (76 � 26 mm; Scientific Inc. (VWR), 631-0113) Spec-Wipe� 5 or 7 Wipers (VWR, 21913-211 or 21913-214) Milli-Q water Phosphate-buffered saline (PBS) 102 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- Potassium Hydroxide (KOH; VWR, 26669.29) Methanol (Sigma, 34860) Acetone (Sigma, 320110) Poly-L-lysine (PLL) (Sigma, P4832) Monoclonal Anti-a-Tubulin antibody (MT) produced in mouse (Sigma, T9026) Alexa Fluor� 568 Phalloidin (Life Technologies, A12380) Sodium Bicarbonate (Sigma, S5761) Dimethyl sulfoxide (DMSO; Sigma, D8418) Succinimidyl-ester 647 nm dye, for example: Cy5 NHS ester (Lumiprobe, 13020) Slide-A-Lyzer MINI Dialysis Device, 3.5K MWCO, 0.1 mL (Thermo Scientific, 69550) Vivaspin 500 MWCO 5000 (GE Healthcare, 28-9322-23) Dulbecco’s Modified Eagle Medium, no phenol red (DMEM, Life Technologies, 31053-028) Triton� X-100 (Sigma, T8787) TWEEN� 20 (Sigma, P9416) PBS-TX (PBS 1X with 0.2% Triton-100) PBS-T (PBS 1X with 0.05% Tween 20) Fetal Bovine Serum (FBS; Sigma, F2442) Blocking buffer (10% FBS in PBS-T) TetraSpeck� Microspheres, 0.1 mm (Life Technologies, T-7279) Paraformaldehyde (PFA; Sigma, 158127) 40,6-Diamidino-2-phenylindoledihydrochloride(DAPI;LifeTechnologies,D1306) Glucose Oxidase (Sigma, G2133) Glucose (Sigma, G8270) Catalase (Sigma, C40) b-mercaptoethylamine (MEA, Sigma, M9768) Thiol-containing buffer (100 mM MEA, pH 7.4e8.4) or oxygen-scavenger buffer (0.5 mg/mL glucose oxidase, 40 mg/mL catalase, 50 mM b-MEA, and 10% W/V glucose in PBS pH 7.4) Parafilm (Alcan Packaging/VWR, PM999/52858-032) Scalpel (Fisher Scientific, 11738353) Nail polish 1.3 METHOD 1.3.1 Cleaning slides and coverslips Unclean coverslips may result in high levels of background, effectively reducing the signal-to-noise ratio (SNR) of the image capture. Also, debris can often produce un- wanted emissions in wavelengthssimilarto the fluorophores in use. While severalclean- ing protocols are available (Fischer, Jacobson, Rose, & Zeller, 2008; Joo & Ha, 2012; Sengupta, Jovanovic-Talisman, & Lippincott-Schwartz, 2013; Soares et al., 2013), the following protocol is the one we currently use in our experimental setup: 1. Sample preparation 103 ----!@#$NewPage!@#$---- • Place coverslips in a glass-slide container. • Sonicate coverslips in ethanol for 20 min. • Rinse three times with Milli-Q water. • Sonicate coverslips in acetone for 20 min. • Rinse three times with Milli-Q water. • Sonicate in KOH 1 M for 20 min, or incubate in 0.1 M KOH for 6 h. • Keep the coverslips in Milli-Q water until needed. • To bleach any remaining particles, irradiate the coverslips with UV light for 1 h. • To dry the slides, use clean gas (e.g., N2); do not air-dry. • Note: For 8-well slides (or similar), immerse for 15 min in KOH 1 M, rinse with Milli-Q water abundantly and irradiate with UV light (test milder conditions if the chamber starts to detach). We recommend a mock-imaging experiment to be performed with empty cover- slips to ensure that your cleaning procedure is working properly. 1.3.2 Labeling with primary antibodies Two components are important in the labeling strategy: (1) the spacing between the target molecule and the fluorophore, since this introduces accuracy error; (2) the flu- orophore brightness as the localization precision scales with fluorophore SNR. Several options are available for labeling in SMLM (Dempsey, Vaughan, Chen, Bates, & Zhuang, 2011; Lippincott-Schwartz & Patterson, 2009; Xu et al., 2013). However, in this chapter we will focus on antibody-based immunofluorescence. The high specificity and the possibility of using a multitude of different fluorophores make antibodies a very successful strategy. A major intrinsic limitation of this strat- egy is the size of antibodies, which introduces a 10e15 nm linker-size error in image acquisition when secondary labeling is used (Figure 1(B)). Therefore, to fully capi- talize on SR techniques, using primary labeled antibodies is extremely beneficial. There are several kits available for direct antibody labeling (e.g., Life Technologies, Bio-Rad, Pierce). We will describe a protocol that we have been successfully using in our workflow using a MT antibody but also applies to single-domain antibodies (sdAb). Commercially known as nanobodies, these have an advantage over full- sized antibodies as their small size increases penetration and resolution, and reduces spurious localizations (Ries, Kaplan, Platonova, Eghlidi, & Ewers, 2012). sdAb have specificities and affinities comparable to those of classical antibodies (except for small peptides, due to sdAb having only one variable heavy chain domain (Sundberg & Mariuzza, 2002)) and can be raised very efficiently using phage-display libraries (Harmsen & De Haard, 2007; McCoy et al., 2012; Muyldermans, 2013). • Dialyze 20e50 ml of the desired antibody (at least at 2 mg/mL, concentrate if needed) against 1L of 0.2 M sodium bicarbonate (pH 8.2) for 16 h at 4 �C with mild agitation in a mini-dialysis unit (molecular weight cutoff ¼ 3500 Da). • Solubilize a succinimidyl-ester dye of choice to a final concentration of 10 mg/ mL in DMSO. • Add the dye in five- to tenfold molar excess to antibody solution. 104 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- • Incubate the mixture for 1 h at 25 �C. • Remove excess dye via buffer exchange into PBS 1X using Vivaspin 500 col- umns (molecular weight cutoff ¼ 5000 Da) obtaining the antibody labeled with the dye of choice. • Store the labeled antibody at 4 �C. The degree of labeling can be measured by ab- sorption spectroscopy using the extinction coefficients of fluorophores and proteins (for best results, the degree of labeling should be 1 or 2 fluorophores per antibody). 1.3.3 Seeding the cells The purpose of a high-content assay is to visualize a large number of cells in different conditions. We will focus on a sample preparation and the acquisition pro- cedure, but it is outside of the scope of this chapter to address screening design. Notwithstanding, the below protocol could be easily adapted to study how different drug-induced conditions may affect microtubule structures seen by SR. • Place washed coverslips in a petri dish and add PLL 0.01%. Incubate for 16 h at 25 �C with mild agitation in a rocker. PLL does emit some autofluorescence in the infrared ranges (w700 nm); use the lowest concentration possible. Alter- natively, fibronectin can be used (Sengupta et al., 2013). • Wash coverslips 10 times with Milli-Q water (incubate slides for 2 s per wash cycle). • Wash slides with 70% ethanol. • Wash slides three times with sterile PBS 1X (in sterile conditions). • Transfer slides to a 6-well dish with sterile PBS 1X • Wash with prewarmed PBS 1X (37 �C) • Seed cells using phenol-red-free DMEM to approximately 60e70% confluency. Aim to have at least two coverslips per condition. 1.3.4 Immunofluorescence There are several published immunofluorescence methodologies for SR (Breu, Gug- genbichler, & Wollmann, 2010). Here we will describe one protocol for fixing mi- crotubules in U2OS cells, highlighting important tips that can be applied to any cell line/protocol used. Methanol fixation is used as it has shown to be the most reli- able method for microtubules in our hands, but other structures and proteins may be better preserved with other methods. General tips: • Use DMEM without phenol red. • Flat cells are optimal, as they will create low background: fibroblasts, COS7, U2OS, neurons, or 3T3, for example. • Use large coverslips, as these will minimize imaging issues due to coverslip tilt. • Use directly labeled primary antibodies to reduce linker errors. • If using genetically encoded photoactivatable fluorophores (e.g., PA-GFP), try to minimize exposure of cells to any source of blue/UV light to decrease unwanted FP photoactivation before the experiments. 1. Sample preparation 105 ----!@#$NewPage!@#$---- • If using immunofluorescence, try different antibody concentrations. If the anti- body concentration is too high, too many fluorophores will be active per frame. If it is too low, the cellular structures of interest may not be fully labeled and may appear patchy. • Image cells immediately after mounting. For dSTORM, after fixation and la- beling you may keep cells overnight in 10% Bovine Serum Albumin (BSA) and image them the next day. • You should use subdiffraction limit beads (e.g., TetraSpeck) as fiduciary land- marks. They can be used for analytical drift and chromatic shift correction after the images are acquired. Protocol for antibody labeling of microtubules in U2OS cells: • Fix cells with 1 mL ice-cold methanol for 10 min (Tanaka et al., 2010). • Incubate cells for 5 min with PBS-TX permeabilize cell membranes. • Wash 3 � 5 min with 1 mL PBS-T. • Samples may be stored at this point (no more than 48 h). • Transfer to staining box (a petri dish covered with tin foil, with wet paper on the sides to keep a humid environment, and parafilm covering the bottom is adequate for this purpose. The parafilm will keep any liquid inside the area of the coverslip). • Block for 1 h at 25 �C with 100 mL of blocking buffer. • Prepare 100 mL of desired antibody dilution (1:100 labeled MT) in blocking buffer and add 1:500 TetraSpeck microspheres 0.1 mm. • Add 100 mL of previous solution to the coverslip. • Incubate for 1 h at 25 �C (or for 16 h at 4 �C). • Wash 3 � 10 min with 200 mL PBS-T. • Repeat the incubation and washing steps for the secondary antibody. • Incubate for 10 min in PFA 2% (in PBS-T). PFA also creates autofluorescence in the green ranges (w520 nm), hence try not to use above 2%. This second fix- ation helps prevent dissociation of the antibodies. • Wash 3 � 10 min with PBS (200 mL). • Incubate 5 min with DAPI following manufacturer’s recommendations. • Wash 3 � 10 min with PBS (200 mL). • Store in blocking buffer. 1.3.5 Coverslip mounting There are several different sample-mounting schemes. This approach is for a fixed sample in a coverslip but other options for both fixed or live-cell imaging with a widerangeofimagingbuffersareavailable(Gunzenha¨user, Olivier, Pengo, & Manley, 2012; Jones, Shim, He, & Zhuang, 2011; Olivier, Keller, Go¨nczy, & Manley, 2013). • A schematic representation of the process is displayed in Figure 1(C). • Start with a clean, dry slide and put a square of parafilm over it, try to avoid the formation of air bubbles by gently pressing the parafilm against the slide. 106 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- • Place it on a clean heating plate at 70 �C for approximately 30 s. Once you see the parafilm becoming transparent, take out the slide from the heating plate. The parafilm should now be glued to the slide; check for air bubbles. • Cut a small square at the center of the parafilm with a scalpel (around 1 cm2) and pull out the little square (this step can be done prior to the heating if preferred). • Put approximately 25 mL of 100 mM MEA buffer (pH 7.4e8.4) inside the square (the goal is to have no air bubbles after the coverslip is attached). If using oxygen-scavenger buffer, only add the glucose to the buffer immediately before use, as the buffer may be kept stable at 4 �C for >2 months without the glucose. The oxygen-scavenger buffer is effective for a few hours, but it will lower the pH of the media over time (due to the conversion of glucose to gluconic acid), which might affect the photophysical properties of the fluorophores. • Place the coverslip over the square on the slide, cell-side down. The buffer should cover the surface of the square and the cell-side of the coverslip. • Seal the coverslip using nail polish. The nail polish should provide an airtight seal between the slide and the coverslip. • Using a swab with acetone, carefully clean the coverslip to get rid of any trace of nail polish in the area to be imaged. 2. IMAGING ACQUISITION AND IMAGE ANALYSIS 2.1 EQUIPMENT Microscope (Figure 1(A)) equipped with (Henriques et al., 2010): • Two laser illumination (spectra-physics, EXLSR-635C-60 mW, and CYAN-488- 100 mW). • Dichroic beam-splitter DM1 (Semrock, LM01-503-25). • Acousto-optic tunable filter (AA Opto-Electronic, AOTFnC-400.650). • Quad-band line dichroic mirror DM2 (Semrock FF416/500/582/657-DiO1). • Dual band pass Filter F (Semrock, FF01-577/690-25). • Cylindrical lens CL (CVI, RCX cylindrical lens 10 m focal length). • Objective (Nikon, CFI Apochromat, 100x, NA 1.49). • EMCCD camera (Andor, Ixon DV887ECS-BV). The above setup is thoroughly described in Henriques et al. (2010), nonetheless some important points should be mentioned: 1. Due to the high number of images necessary to perform an SMLM experiment, avoiding and correcting sample drift is a key factor in SR experiments. The system should be mounted on a vibration-isolated table and in a room with strict temperature control. In addition to this, fiducial markers should be used in order to correct for drift in image analysis. 2. Another important aspect is the SNR. Signal arising from background fluores- cence and detector noise should be as low as possible. This can be addressed by 2. Imaging acquisition and image analysis 107 ----!@#$NewPage!@#$---- using wide-field illumination in a total internal reflection fluorescence (TIRF) regime, permitting single-molecule fluorescence imaging with a high SNR. The TIRF evanescent field has a penetration depth of only a few hundred nano- meters, which minimizes contributions from out-of-focus light. Other illumi- nation regimes are available (Deschout et al., 2014; Herbert et al., 2012), but a discussion of their merits is outside the scope of this chapter. 3. As with the imaging regime, different strategies enable 3D acquisition of SR im- ages (Deschout et al., 2014; Herbert et al., 2012). The setup described here uses astigmatism (Huang, Wang, Bates, & Zhuang, 2008). This method relies on the introduction of a cylindrical lens in the emission optical path, introducing slight depth-dependent aberrations (Huang et al., 2008). After calibrating for these aberrations, the SR algorithm can determine the z-coordinate of the fluorophore. 4. The fluorescence is recorded on a detector, which can be an EMCCD camera or a scientific complementary metal-oxide semiconductor (sCMOS) camera directly mounted on a suitable port of the inverted microscope stand. The selection of an appropriate detector (Waters, 2009) is critical to the experiment, since it will determine the SNR, speed of acquisition, and FOV size. 2.2 SOFTWARE • mManager (Edelstein et al., 2010) • ImageJ/Fiji (Schindelin et al., 2012) • QuickPALM (Henriques et al., 2010) The details about the software used in this experimental workflow can be found in Edelstein et al. (2010), Henriques et al. (2010), and Schindelin et al. (2012). The choice of the software to create an SR image is of great importance as each software package addresses the problems imposed by SR in a different manner. In this context, we will focus on practical aspects and on the freely available software QuickPALM (Henriques et al., 2010), but an extensive overview can be found in Small & Stahlheber (2014). 2.3 METHOD 2.3.1 Acquisition Within the Micro-Manager acquisition software, the Multidimensional Acquisition interface allows one to set up a new acquisition protocol. The important steps are • Edit the FOVs of choice (under the “multiple positions” option)dplease note, Micro-Manager provides a tool to automatically generate a grid array-type position for the acquisition. • Set the number of time-points to w10,000 with no interval between frames. • Choose the channels to be acquired. To avoid channel cross talk, it is advisable to image the fluorophores with longer wavelengths first, as these will eventually be permanently bleached during the acquisition, thus minimizing the probability that they will contribute unwanted signal to the other channels. 108 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- • We advise to set up the acquisition order as “Position, Time, Channel,” since this will be the fastest modality to acquire the dataset. We frequently found the need to set up complex protocols beyond the capacity of the Multidimensional Acquisition interface. This can be achieved by direct scripting of Micro-Manager. The following example shows a similar protocol as above, but implemented in the scripting interface and with a pre-bleaching phase incorporated: • Onthe Multi-dimensional Acquisitioninterface,set upaprotocoltopre-bleacheach individual channel by setting a single frame to be strongly illuminated over an extended period of time (e.g., 1 min) but obstructing incoming light to the camera to avoid oversaturation (this can be achieved, for example, by setting a detection- path filter-wheel to a light-blocking position). Save this configuration to a file. • On the Multi-dimensional Acquisition interface, set up a protocol to acquire w10,000 frames for each channel individually, but without defining different FOVs. Save this configuration to a file. • Start the Micro-Manager script panel (Tools > Script Panel.). • Edit the following code snippet in the highlighted areas as needed: float [] xpos = new float []{0, 100, 0, 100}; float [] ypos = new float []{0, 0, 100, 100}; void run() { # Open a dialog asking the user to define the directory to save the files on DIR = ij.IJ.getDirectory(“Where do I save the data?”); if (DIR = = null) return; for (int np = 0; np Plugins > Macros > Record.). • Open a single FOV channel as a virtual stack (Fiji > File > Import > Image Sequence). • Run a QuickPALM analysis (Fiji > Analyze > QuickPALM > Analyze Parti- cles) and optimize settings. A detailed example can be found on the Quick- PALM Web site (https://code.google.com/p/quickpalm) in Wiki > Tutorials > Software Related > Detection and Reconstruction. • Next, save the particles table and reconstruct the SR image (Fiji > Analyze > QuickPALM > Reconstruct Dataset). Save the reconstruction. • You should now have a small macro on the ImageJ’s macro recorder similar to: run(“Image Sequence.”, “open = [/path/to/image-sequence/folder] sort use”); run(“Analyze Particles”, “minimum = 5 maximum = 4 image = 106 smart pixel = 30 accumulate = 0 update = 10 _image = imgNNNNNNNNN.tif start = 0 in = 50 _minimum = 50 local = 20 _maximum = 1000 threads = 50”); saveAs(“Results”, “/path/to/results/Results.xls”); run(“Reconstruct Dataset”, “target = 30 original = 512 original = 512 view = [2D histogram] simulate fwhm = 30 z-spacing = 50 merge = 0 merge = 0 make = 10 accumulate = 100”); saveAs(“Tiff”, “/path/to/results/Reconstruction.tif”); • Adapt the code by generating a loop that goes through each dataset automati- cally. For novice programmers, the Fiji Web site supports a tutorial that covers how to edit and generate macros with ImageJdhttp://fiji.sc/Introduction_into_ Macro_Programming. 2.4 IMPORTANT CONSIDERATIONS 2.4.1 Detectors used for the acquisition For an SR HCS, the resolution (directly proportional to the SNR collected and the number of fluorophores pinpointed), the FOV size, and the imaging speed are of 110 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- critical importance. The goal is to analyze the highest number of cells with no loss in resolution and as fast as possible. To address this issue, the choice of detector plays a significant role as several detector properties can affect the localization precision, accuracy, and speed. EMCCD cameras are the usual choice for an SR microscope (Henriques et al., 2010; Rust et al., 2006). EMCCD cameras are CCD cameras which utilize electron multiplication before reading the pixel value, which increases the sensitivity of detection while eliminating camera read noise. They are thus capable of detecting the weak fluorescence signal from single molecules. However, they are inherently CCD cameras which read values pixel by pixel, making them relatively slow if more resolution is required which may hinder the utilization of such cameras when high content and throughput is required (Long, Zeng, & Huang, 2012). The electron multiplication also introduces multiplicative noise, which effectively halves the photon collection efficiency (quantum efficiency, QE) of the camera. Alterna- tively, in sCMOS cameras, each pixel has its own detector (parallel readout) which offers low noise at extremely fast detection speeds (w100 Hz for a modern 2048 � 2048 sCMOS). On the flip side, since each pixel has an individual detector, each pixel has to have its own noise estimation (Long et al., 2012) which can intro- duce artifacts. They generally also have lower QE due to the higher number of elec- tronic components. In sum, EMCCD cameras have better sensitivity at the cost of a smaller FOV and speed, whereas sCMOS cameras offer higher resolutions with extremely high readout speeds at the cost of some sensitivity. For a detailed compar- ison of the detectors available, see Long et al. (2012) and Moomaw (2013). 2.4.2 Single-molecule detection and localization For single-molecule detection in SMLM, a large series of images is acquired so that only a small subset of fluorophores is detected and localized in each frame. For each particle that does not have any significant overlap with neighboring fluorophores on the frame, its relative position can be estimated with high precision. As mentioned before, this approach results in long acquisition and processing times, and despite the fast development of algorithms that allow image analysis and reconstruction in real time (Henriques et al., 2010; Holden et al., 2011; Mukamel et al., 2012; Ovesny et al., 2014; Wolter et al., 2012), there has been an intense focus on devel- oping algorithms that allow for more particle overlap. The best-known example is probably SR optical fluctuation imaging (SOFI), which relies on higher order statis- tical analysis of temporal fluctuations caused by fluorescence blinking/intermit- tency, allowing the identification of individual particles (Dertinger, Colyer, Iyer, Weiss, & Enderlein, 2009). Since SOFI works with high-density images, it allows for acquisition of stacks with 3000 images instead of the usual 10,000e50,000 required for SMLM methods (Dertinger et al., 2009; Henriques et al., 2010), which opens the field for faster imaging acquisition in high-throughput experiments and live-cell imaging, although at the expense of some resolution. More recently, other methodologies to obtain subdiffraction limit optical resolutions in samples with increased labeling density have been described, such as 3B (Cox et al., 2012), 2. Imaging acquisition and image analysis 111 ----!@#$NewPage!@#$---- DAOSTORM (Holden et al., 2011), or ThunderSTORM (Ovesny et al., 2014), which have the advantage of being freely available. 2.4.3 SR drift correction and chromatic realignment Even in optimal conditions, sample drift on the nanometer scale is common in most microscope setups due to vibrations and mechanical relaxation of the microscope. Whereas in conventional FLM these small aberrations are disguised within the op- tical resolution limit, in SR microscopy this drift during camera exposure affects the localization precision and accuracy effectively reducing the resolution of the ac- quired image; drift as low as a few nanometers can distort an image (Gould, Burke, Bewersdorf, & Booth, 2012). To address this, fiducial markers can be introduced in the sample (Figure 1(B) and (D)), typically gold nanoparticles, quantum dots, or fluorescent beads. As these particles do not bleach significantly during the acquisi- tion, it is possible to track the drift and correct for it in the image reconstruction step (Henriques et al., 2010). In multicolor SR imaging, besides motion artifacts caused by drift, there is the need to account for two other sources of error: chromatic aber- rations (Erdelyi et al., 2013) and wavelength-dependent magnification shifts (Clark, Dierkes, & Paluch, 2013). Both these effects are due to the light path being slightly shifted on a wavelength-dependent manner. Although modern microscope objec- tives have apochromatic characteristics that attenuate these effects, they will still contribute to the deformation of the sample at the 1e20 nm scale (Erdelyi et al., 2013). In order to correct for these effects, multicolored fiducial markers can be used, which will be detected in all channels during the acquisition. In this way, chan- nel displacement can be corrected through the application of an elastic transforma- tion during the reconstruction step (Soares et al., 2013). 2.4.4 SR estimation and image reconstruction The final resolution of an SR image is an important aspect of the image obtained, and can vary a lot not only depending on the methodology used but also throughout a sin- gle image. Several factors influence this parameter, such as fluorophore orientation, local refractive index variations, quality of the camera, local aberrations, statistical selection bias, fluorophore density, and the SNR (Schermelleh et al., 2010). Ulti- mately, the choice of the software to use for the reconstruction of SR image and the actual resolution of the image will depend on several factors and on the biolog- ical question in hand. Although the rapid evolution of the SR field could make this challenging, there are a number of very good reviews covering virtually every aspect of the experimental process from the hardware setup, sample preparation, choice of fluorophores, and image analysis and have been referenced throughout this protocol. CONCLUSIONS AND OUTLOOK Although SMLM is severely hampered by a slow acquisition speed, recent strides have been made aiming to tackle this problem by accelerating the analysis (Cox 112 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- et al., 2012; Henriques et al., 2010; Huang et al., 2013b). The emergence of sCMOS cameras, which can now acquire 100 frames per second at full-chip resolution, will soon accelerate the pace of SMLM imaging. These have high QE and low noise, the features necessary for SR as well as being faster than any CCD currently available. Importantly for HC, their large resolutions can acquire much larger areas. They seem to check all the boxes for the perfect camera for SR HCS, but have one problem: they have nonuniform noise. This can introduce artifacts in SR and while corrections for this can be made, it is still in the domain of experimental SR labs. Once this has been made available to the general scientific community in a user-friendly manner, SR HCS will benefit from a speed increase of at least an order of magnitude, since more information can be acquired in a single image and each image will be up to 3x faster than a conventional EMCCD. Nevertheless, it will still take a considerable time until SR high-content assays approach the speed possible in low-resolution screens. Another issue with SR is live imaging. Most HCS can be done with live cells, but SMLM and STED techniques are speed limited and highly damaging to the cell, and as such has been mostly applied to fixed samples. SIM can be used for live imaging (Shao, Kner, Rego, & Gustafsson, 2011) but offers only a moderate increase in resolution. STED has also been used to image live cells (Na¨gerl, Willig, Hein, Hell, & Bonhoeffer, 2008) but is inherently limited in speed, which limits use- fulness for HCS. Recent advances in scientific cameras, SMLM algorithms, and la- beling strategies have enabled very fast SR of microtubules in live cells (Huang et al., 2013a). In the future, we hope to see the dissemination of these developments enabling not only faster (and therefore, with higher content) HCSs but also SMLM screens being applied on live cells. Here we show that for small screens it is quite feasible to bring some high- content traits to the SR imaging field. It is becoming evident that there are consider- able molecular-scale phenotypes only perceived by achieving resolutions below the diffraction limit (Dani et al., 2010; Soares et al., 2013), making SR an essential tool that needs to be brought into the HCS field. ACKNOWLEDGMENTS We thank Joe Grove (Royal Free Hospital, UCL) for helpful suggestions and discussions. REFERENCES Abbe, E. (1873). Beitra¨ge zur Theorie des Mikroskops und der mikroskopischen Wahrnehmung. Archiv Fu¨r Mikroskopische Anatomie, 9(1), 413e418. http://dx.doi.org/ 10.1007/BF02956173. Betzig, E., Patterson, G. H., Sougrat, R., Lindwasser, O. W., Olenych, S., Bonifacino, J. S., et al. (2006). Imaging intracellular fluorescent proteins at nanometer resolution. Science, 313(5793), 1642e1645. http://dx.doi.org/10.1126/science.1127344. References 113 ----!@#$NewPage!@#$---- Breu, F., Guggenbichler, S., & Wollmann, J. (2010). Immunocytochemical methods and pro- tocols. In C. Oliver, & M. C. Jamur (Eds.), Vasa (Vol. 588). Totowa, NJ: Humana Press. http://dx.doi.org/10.1007/978-1-59745-324-0. Clark, A. G., Dierkes, K., & Paluch, E. K. (2013). Monitoring actin cortex thickness in live cells. Biophysical Journal, 105(3), 570e580. http://dx.doi.org/10.1016/j.bpj.2013.05.057. Cox, S., Rosten, E., Monypenny, J., Jovanovic-Talisman, T., Burnette, D. T., Lippincott- schwartz, J., et al. (2012). Bayesian localization microscopy reveals nanoscale podosome dynamics. Nature Methods, 9(2), 195e200. http://dx.doi.org/10.1038/nmeth.1812. Dani, A., Huang, B., Bergan, J., Dulac, C., & Zhuang, X. (2010). Superresolution imaging of chemical synapses in the brain. Neuron, 68(5), 843e856. http://dx.doi.org/10.1016/ j.neuron.2010.11.021. Dempsey, G. T., Vaughan, J. C., Chen, K. H., Bates, M., & Zhuang, X. (2011). Evaluation of fluorophores for optimal performance in localization-based super-resolution imaging. Na- ture Methods, 1e14. http://dx.doi.org/10.1038/nmeth.1768. advance on(nOveMbeR). Dertinger, T., Colyer, R., Iyer, G., Weiss, S., & Enderlein, J. (2009). Fast, background-free, 3D super-resolution optical fluctuation imaging (SOFI). Proceedings of the National Acad- emy of Sciences of the United States of America, 106(52), 22287e22292. http:// dx.doi.org/10.1073/pnas.0907866106. Deschout, H., Zanacchi, F. C., Mlodzianoski, M., Diaspro, A., Bewersdorf, J., Hess, S. T., et al. (2014). Precisely and accurately localizing single emitters in fluorescence microscopy. Nature Methods, 11(3), 253e266. http://dx.doi.org/10.1038/nmeth.2843. Edelstein, A., Amodaj, N., Hoover, K., Vale, R., & Stuurman, N. (2010). Computer Control of Microscopes Using mManager. Current Protocols in Molecular Biology, 92, 14.20: 14.20.1e14.20.17. Erdelyi, M., Rees, E., Metcalf, D., Schierle, G. S. K., Dudas, L., Sinko, J., et al. (2013). Cor- recting chromatic offset in multicolor super-resolution localization microscopy. Optics Express, 21(9), 10978. http://dx.doi.org/10.1364/OE.21.010978. Erfle, H., Neumann, B., Liebel, U., Rogers, P., Held, M., Walter, T., et al. (2007). Reverse transfection on cell arrays for high content screening microscopy. Nat. Protocols, 2(2), 392e399. Ferraro, F., Kriston-Vizi, J., Metcalf, D. J., Martin-Martin, B., Freeman, J., Burden, J. J., et al. (2014). A two-tier Golgi-based control of organelle size underpins the functional plas- ticity of endothelial cells. Developmental Cell, 29(3), 292e304. http://dx.doi.org/ 10.1016/j.devcel.2014.03.021. Fischer, A. H., Jacobson, K. A., Rose, J., & Zeller, R. (2008). Preparation of slides and cov- erslips for microscopy. Cold Spring Harbor Protocols, 2008(5). http://dx.doi.org/10.1101/ pdb.prot4988. Gould, T. J., Burke, D., Bewersdorf, J., & Booth, M. J. (2012). Adaptive optics enables 3D STED microscopy in aberrating specimens. Optics Express, 20(19), 20998e21009. Retrieved from http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid¼3635694& tool¼pmcentrez&rendertype¼abstract. Gunzenha¨user, J., Olivier, N., Pengo, T., & Manley, S. (2012). Quantitative super-resolution imaging reveals protein stoichiometry and nanoscale morphology of assembling HIV-Gag virions. Nano Letters, 12(9), 4705e4710. http://dx.doi.org/10.1021/nl3021076. Gustafsson, M. G. L. (2000). Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy. Short communication. Journal of Microscopy, 198(2), 82e87. http://dx.doi.org/10.1046/j.1365-2818.2000.00710.x. 114 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- Gustafsson, M. G. L., Shao, L., Carlton, P. M., Wang, C. J. R., Golubovskaya, I. N., Cande, W. Z., et al. (2008). Three-dimensional resolution doubling in wide-field fluores- cence microscopy by structured illumination. Biophysical Journal, 94(12), 4957e4970. http://dx.doi.org/10.1529/biophysj.107.120345. Harmsen, M. M., & De Haard, H. J. (2007). Properties, production, and applications of cam- elid single-domain antibody fragments. Applied Microbiology and Biotechnology, 77(1), 13e22. http://dx.doi.org/10.1007/s00253-007-1142-2. Heilemann, M., van de Linde, S., Schu¨ttpelz, M., Kasper, R., Seefeldt, B., Mukherjee, A., et al. (2008). Subdiffraction-resolution fluorescence imaging with conventional fluores- cent probes. Angewandte Chemie (International Ed. in English), 47(33), 6172e6176. http://dx.doi.org/10.1002/anie.200802376. Henriques, R., Lelek, M., Fornasiero, E. F., Valtorta, F., Zimmer, C., & Mhlanga, M. M. (2010). QuickPALM: 3D real-time photoactivation nanoscopy image processing in ImageJ. Nature Methods, 7(5), 339e340. http://dx.doi.org/10.1038/nmeth0510-339. Henriques, R., & Mhlanga, M. M. (2009). PALM and STORM: what hides beyond the Rayleigh limit? Biotechnology Journal, 4(6), 846e857. http://dx.doi.org/10.1002/biot.200900024. Herbert, S., Soares, H., Zimmer, C., & Henriques, R. (2012). Single-molecule localization super-resolution microscopy: deeper and faster. Microscopy and Microanalysis: The Offi- cial Journal of Microscopy Society of America, Microbeam Analysis Society, Microscop- ical Society of Canada, 18(6), 1419e1429. http://dx.doi.org/10.1017/S1431927 612013347. Hess, S. T., Girirajan, T. P. K., & Mason, M. D. (2006). Ultra-high resolution imaging by fluo- rescence photoactivation localization microscopy. Biophysical Journal, 91(11), 4258e4272. http://dx.doi.org/10.1529/biophysj.106.091116. Holden, S. J., Uphoff, S., & Kapanidis, A. N. (2011). DAOSTORM: an algorithm for high- density super-resolution microscopy. Nature Methods, 8(4), 279e280. http://dx.doi.org/ 10.1038/nmeth0411-279. Huang, F., Hartwich, T. M. P., Rivera-Molina, F. E., Lin, Y., Duim, W. C., Long, J. J., et al. (2013). Video-rate nanoscopy using sCMOS camera-specific single-molecule localization algorithms. Nature Methods, 10(7), 653e658. http://dx.doi.org/10.1038/nmeth.2488. Huang, F., Hartwich, T. M. P., Rivera-Molina, F. E., Lin, Y., Duim, W. C., Long, J. J., et al. (2013). Video-rate nanoscopy using sCMOS cameraespecific single-molecule localiza- tion algorithms. Nature Methods, 10(7), 653e658. http://dx.doi.org/10.1038/nmeth.2488. Huang, B., Wang, W., Bates, M., & Zhuang, X. (2008). Three-dimensional super-resolution imaging by stochastic optical reconstruction microscopy. Science (New York, NY), 319(5864), 810e813. http://dx.doi.org/10.1126/science.1153529. Jones, S. A., Shim, S.-H., He, J., & Zhuang, X. (2011). Fast, three-dimensional super-resolu- tion imaging of live cells. Nature Methods, 8(6), 499e508. http://dx.doi.org/10.1038/ nmeth.1605. Joo, C., & Ha, T. (2012). Labeling DNA (or RNA) for single-molecule FRET. Cold Spring Harbor Protocols, 2012(9), 1005e1008. http://dx.doi.org/10.1101/pdb.prot071027. Klar, T. A., Jakobs, S., Dyba, M., Egner, A., & Hell, S. W. (2000). Fluorescence microscopy with diffraction resolution barrier broken by stimulated emission. Proceedings of the National Academy of Sciences, 97(15), 8206e8210. http://dx.doi.org/10.1073/pnas.97.15.8206. Lippincott-Schwartz, J., & Patterson, G. H. (2009). Photoactivatable fluorescent proteins for diffraction-limited and super-resolution imaging. Trends in Cell Biology, 19(11), 555e565. http://dx.doi.org/10.1016/j.tcb.2009.09.003. References 115 ----!@#$NewPage!@#$---- Long, F., Zeng, S., & Huang, Z.-L. (2012). Localization-based super-resolution microscopy with an sCMOS camera part II: experimental methodology for comparing sCMOS with EMCCD cameras. Optics Express, 20(16), 17741e17759. Retrieved fromhttp://www. ncbi.nlm.nih.gov/pubmed/23038326. McCoy, L. E., Quigley, A. F., Strokappe, N. M., Bulmer-Thomas, B., Seaman, M. S., Mortier, D., et al. (2012). Potent and broad neutralization of HIV-1 by a llama antibody elicited by immunization. Journal of Experimental Medicine, 209(6), 1091e1103. http://dx.doi.org/10.1084/jem.20112655. Moomaw, B. (2013). Camera technologies for low light imaging: Overview and relative ad- vantages (4th ed.). In Methods in cell biology (4th ed.), (Vol. 114). Elsevier Inc. http:// dx.doi.org/10.1016/B978-0-12-407761-4.00011-7. pp. 243e83. Mukamel, E.a, Babcock, H., & Zhuang, X. (2012). Statistical deconvolution for superresolu- tion fluorescence microscopy. Biophysical Journal, 102(10), 2391e2400. http:// dx.doi.org/10.1016/j.bpj.2012.03.070. Muyldermans, S. (2013). Nanobodies: natural single-domain antibodies. Annual Review of Biochemistry, 82, 775e797. http://dx.doi.org/10.1146/annurev-biochem-063011-092449. Na¨gerl, U. V., Willig, K. I., Hein, B., Hell, S. W., & Bonhoeffer, T. (2008). Live-cell imaging of dendritic spines by STED microscopy. Proceedings of the National Academy of Sci- ences of the United States of America, 105(48), 18982e18987. http://dx.doi.org/ 10.1073/pnas.0810028105. Olivier, N., Keller, D., Go¨nczy, P., & Manley, S. (2013). Resolution doubling in 3D-STORM imaging through improved buffers. PloS One, 8(7), e69004. http://dx.doi.org/10.1371/ journal.pone.0069004. Ovesny, M., K�ri�zek, P., Borkovec, J., Svindrych, Z., & Hagen, G. M. (2014). Thunder- STORM: a comprehensive ImageJ plugin for PALM and STORM data analysis and super-resolution imaging. Bioinformatics (Oxford, England), 30(16), 2389e2390. http://dx.doi.org/10.1093/bioinformatics/btu202. Ries, J., Kaplan, C., Platonova, E., Eghlidi, H., & Ewers, H. (2012). A simple, versatile method for GFP-based super-resolution microscopy via nanobodies. Nature Methods, 9(6), 582e584. http://dx.doi.org/10.1038/nmeth.1991. Rust, M., Bates, M., & Zhuang, X. (2006). Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nature Methods, 3(10), 793e795. http:// dx.doi.org/10.1038/NMETH929. Schermelleh, L., Heintzmann, R., & Leonhardt, H. (2010). A guide to super-resolution fluo- rescence microscopy. Journal of Cell Biology, 190(2), 165e175. http://dx.doi.org/ 10.1083/jcb.201002018. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., et al. (2012). Fiji: an open-source platform for biological-image analysis. Nature Methods, 9(7), 676e682. http://dx.doi.org/10.1038/nmeth.2019. Sengupta, P., Jovanovic-Talisman, T., & Lippincott-Schwartz, J. (2013). Quantifying spatial organization in point-localization superresolution images using pair correlation analysis. Nature Protocols, 8(2), 345e354. http://dx.doi.org/10.1038/nprot.2013.005. Shao, L., Kner, P., Rego, E. H., & Gustafsson, M. G. L. (2011). Super-resolution 3D micro- scopy of live whole cells using structured illumination. Nature Methods, 8(octobeR), 1044e1046. http://dx.doi.org/10.1038/nmeth.1734. Small, A., & Stahlheber, S. (2014). Fluorophore localization algorithms for super-resolution microscopy. Nature Methods, 11(3), 267e279. http://dx.doi.org/10.1038/nmeth.2844. 116 CHAPTER 7 High-content 3D multicolor super-resolution ----!@#$NewPage!@#$---- Soares, H., Henriques, R., Sachse, M., Ventimiglia, L., Alonso, M.a, Zimmer, C., et al. (2013). Regulated vesicle fusion generates signaling nanoterritories that control T cell activation at the immunological synapse. The Journal of Experimental Medicine, 210(11), 2415e2433. http://dx.doi.org/10.1084/jem.20130150. Sundberg, E. J., & Mariuzza, R. A. (2002). Molecular recognition in antibody-antigen complexes. Advances in Protein Chemistry, 61, 119e160. Retrieved fromhttp://www. ncbi.nlm.nih.gov/pubmed/12461823. Tanaka, K. A. K., Suzuki, K. G. N., Shirai, Y. M., Shibutani, S. T., Miyahara, M. S. H., Tsuboi, H., et al. (2010). Membrane molecules mobile even after chemical fixation. Na- ture Methods, 7(11), 865e866. http://dx.doi.org/10.1038/nmeth.f.314. Thompson, R. E., Larson, D. R., & Webb, W. W. (2002). Precise nanometer localization anal- ysis for individual fluorescent probes. Biophysical Journal, 82(5), 2775e2783. http:// dx.doi.org/10.1016/S0006-3495(02)75618-X. Van de Linde, S., Krsti�c, I., Prisner, T., Doose, S., Heilemann, M., & Sauer, M. (2011). Photo- induced formation of reversible dye radicals and their impact on super-resolution imaging. Photochemical & Photobiological Sciences: Official Journal of the European Photochem- istry Association and the European Society for Photobiology, 10(4), 499e506. http:// dx.doi.org/10.1039/c0pp00317d. Waters, J. C. (2009). Accuracy and precision in quantitative fluorescence microscopy. Journal of Cell Biology, 185(7), 1135e1148. http://dx.doi.org/10.1083/jcb.200903097. Wolter, S., Lo¨schberger, A., Holm, T., Aufmkolk, S., Dabauvalle, M.-C., van de Linde, S., et al. (2012). rapidSTORM: accurate, fast open-source software for localization microscopy. Nature Methods, 9(11), 1040e1041. http://dx.doi.org/10.1038/nmeth.2224. Xu, J., Chang, J., Yan, Q., Dertinger, T., Bruchez, M., & Weiss, S. (2013). Labeling cytosolic targets in live cells with blinking probes. Journal of Physical Chemistry Letters, 4(13), 2138e2146. http://dx.doi.org/10.1021/jz400682m. Yildiz, A., & Selvin, P. R. (2005). Fluorescence imaging with one nanometer accuracy: appli- cation to molecular motors. Accounts of Chemical Research, 38(7), 574e582. http:// dx.doi.org/10.1021/ar040136s. References 117