We would like to thank Corina Beerli, Jerzy Samolej, Pedro Matos Pereira, Christopher Bleck, and Kathrin Scherer for their contributions to the original development and validation of VirusMapper. We would also like to thank Artur Yakimovich for his critical reading of the manuscript. This work was funded by grants from the Biotechnology and Biological Sciences Research Council (BB/M022374/1) (R.H.); core funding to the MRC Laboratory for Molecular Cell Biology, University College London (J.M.); the European Research Council (649101-UbiProPox) (J.M.); and the Medical Research Council (MR/K015826/1) (R.H. and J.M.). R.G. is funded by the Engineering and Physical Sciences Research Council (EP/M506448/1).

NOTE: This protocol and video supplement the original paper10 describing the software package in more detail. Readers are encouraged to review this carefully for additional guidance regarding the use of the software. There are three main stages: particle extraction, which segments large images into individual particles; seed selection, where common structures are identified in the data and aligned to produce seeds, which are used in the final stage; and model generation, where template matching based on these seeds aligns the extracted particles and averages a subset to produce the SPA models.
1. Setup Prior to Running the Software Package
<ol>
	<li>Prepare samples of the structure under study on a coverslip or in the relevant experimental conditions.</li>
	<li>Image the samples with super-resolution fluorescence microscopy, such as structured illumination microscopy (SIM)11 or stimulated emission depletion (STED)12 microscopy. 
	NOTE: The precise details of how to prepare and image samples depends greatly on the nature of the structure under study, so the relevant literature should be consulted. As an example, the precise method for preparing and imaging samples of vaccinia virus, such as those used here, is described in the representative results section.</li>
	<li>Create images of multiple fields of view showing a large number of separate copies of the structure or particles, preferably thousands. Image particles that are as well separated from each other as possible and ensure that the images are free of dirt or other fluorescent structures that are not of interest.</li>
	<li>Open all images containing the particles in Fiji by dragging the files into the Fiji toolbar or by selecting &#34;File&#34;&#62;&#34;Open&#34;.</li>
	<li>Select &#34;Image&#34;&#62;&#34;Stack&#34;&#62;&#34;Tools&#34;&#62;&#34;Concatenate&#34; to concatenate the images into a single stack. Then, if the resultant image is a Hyperstack, turn it into a stack by selecting &#34;Image&#34;&#62;&#34;Hyperstacks&#34;&#62; &#34;Hyperstack to Stack&#34;. 
	&#8203;NOTE: The final stack should have intercalated channels. If there are two channels, the first slice in the stack should be channel 1 from the first image, the second slice should be the corresponding channel 2, the third slice should be channel 1 from the second image, and so on.</li>
</ol>
2. Extract the Particles
<ol>
	<li>Choose the image to segment and select &#34;Extract Viral Structures&#34;. Choose where to save the extracted particles and view the &#34;Extract Viral Structures&#34; dialog (Figure 2).</li>
	<li>Assign extraction parameters with initial estimates by entering them into the &#34;Extract Viral Structures&#34; dialog as follows. Fine-tune these parameters after previewing the image segmentation.
	<ol>
		<li>Set the number of channels in the dataset as the number of different fluorescence channels that have been imaged (e.g., 2).</li>
		<li>Set the reference channel from which particles are extracted by detection of peaks in that channel by entering the number of the choice of channel. Select the most consistent channel; that is, the channel in which the most particles have the same appearance. 
		&#8203;NOTE: If possible, particles in this channel will have a central maximum.</li>
		<li>Choose whether or not to apply a pre-detection Gaussian blur. Set the pre-detection Gaussian blur to 0 to apply no blur; if this value is increased, a Gaussian blur filter of the given radius is applied before local maxima detection. Use this feature if the reference channel does not have a central maximum (e.g., ring shape); blurring induces the appearance of one. 
		&#8203;NOTE: Segmented regions of interest (ROIs) do not have the Gaussian blur applied to them, as this feature is only used to position ROIs in the reference channel.</li>
		<li>Set the ROI radius (which will be set around each local maximum) in pixels. Choose a value such that that ROIs are slightly larger than the largest particles, such as in Figure 3. For example, if the largest particles appear to have a diameter of around 30 pixels (estimated roughly by eye), then set the ROI radius to 20 pixels.</li>
		<li>Set the number of ROIs to use per frame to an initial, relatively small value below 100.</li>
		<li>Set the maximum ROI overlap. If the particles are well separated, keep this small; if particles are clustered, increase this to enable the ROIs to overlap.</li>
	</ol>
	</li>
	<li>Select &#34;show preview&#34;. 
	NOTE: When this is selected, the extracted ROIs for the reference channel associated with the current frame will appear.</li>
	<li>Adjust the ROI radius, the number of ROIs, and the maximum ROI overlap to have ROIs of suitable size around as many particles as possible, as in Figure 3.</li>
	<li>Select &#34;OK&#34; to run the segmentation. Close the image and the ROI manager. 
	&#8203;NOTE: Do not change the names of the files of the particle sets. These names must be in the format &#34;Viral particles - channelX&#34; for the following sections.</li>
</ol>
3. Select Seeds
<ol>
	<li>Select &#34;Generate Seeds&#34;, select the folder where the extracted particles are saved, and view the &#34;Generate Seeds&#34; dialog and windows (Figure 4).</li>
	<li>Assign initial seed-selection parameters by entering them into the &#34;Generate Seeds&#34; dialog, as follows.
	<ol>
		<li>Set the reference channel that will be fitted to align and center all channels. Choose a channel in which most particles have the same appearance and that has a central maximum, if possible.</li>
		<li>Select the boxes for all channels for which a seed should be generated.</li>
		<li>Select if the seeds should be rotated by 90&#176;. Use this feature to have the alignment consistent with other models</li>
		<li>Choose whether or not to apply a pre-alignment Gaussian blur. Set the blur radius for each channel to 0 to apply no blur. Increase this value to apply a Gaussian blur filter of the given radius before realignment. Use this feature if the particles do not have a central maximum; blurring induces the appearance of one. 
		&#8203;NOTE: Seeds will not have the blur applied, as this feature is only to get consistent alignment.</li>
		<li>Select whether to use shift correction to separately center seeds for non-reference channels, although not rotate them, by checking or unchecking the &#34;Shift correction&#34; boxes for each channel. 
		NOTE: Use this if the channels are not well aligned with each other; it can also be useful for aligning other channels solely according to the reference, without shift correction.</li>
	</ol>
	</li>
	<li>Choose particles to use as seeds. Search through the particle sequence to find a particle that resembles the structure that is to be modeled and record the frame number.</li>
	<li>Enter the number in the &#34;Frames to use&#34; box, and more windows will appear (Figure 5). Enter multiple frame numbers separated by commas.</li>
	<li>View the seed frames and the resultant average seeds in the windows that appear. 
	NOTE: The log will suggest seeds similar to the average.</li>
	<li>Adjust the reference channel, the Gaussian blur radius, and the shift correction options to optimize the seed selection process so that a number of seed frames found have a similar appearance. Continue adding seeds until average seeds are created that satisfactorily resemble the structure seen in the data that is to be modeled.</li>
	<li>Name the seeds and where they will be saved to and select &#34;OK&#34; to save the final seed images for later use in model generation.</li>
</ol>
4. Generate Models
<ol>
	<li>Select &#34;Generate Models Based On Seeds&#34;, select the folder where the extracted particles are saved, and view the &#34;Generate Models&#34; dialog (Figure 6).</li>
	<li>Load the seeds for each channel by using the checkboxes. 
	&#8203;NOTE: Seeds will be saved in a subfolder that was named in the &#34;Generate Seeds&#34; dialog. The default name is &#34;Analysis&#34;. Take care to select the seed Averages, not the Frames, which are also saved for reference.</li>
	<li>Assign initial model generation parameters by entering them into the &#34;Generate Models&#34; dialog, as follows. Fine-tune these parameters later.
	<ol>
		<li>Select whether to use a reference channel for alignment, which calculates translations and rotations only from the reference channel and applies them to all channels. If using this option, select the channel to use as the reference. 
		&#8203;NOTE: This option should only be chosen to specifically use one channel as a reference, such as if doing reference-based structure discovery. Otherwise, it will result in less accurate models. The channels should be well aligned if using this option.</li>
		<li>Select whether to square image intensities during template matching. This will accentuate small differences, so use this when creating a model which has particularly subtle features.</li>
		<li>Choose a minimum similarity against seed by using the &#34;Minimum similarity against seed&#34; slider or by entering a number into the box. 
		&#8203;NOTE: Only particles with a similarity to the seeds greater than this cut-off will be used. 60-80% is typically a good choice to start with.</li>
		<li>Set the maximum number of iterations to 1, optimize the minimum similarity against seed, and increase it later, as required.</li>
		<li>Select the boxes to choose the elements of the model generation process that will appear. 
		NOTE: &#34;Show seeds&#34; will display the seeds that have been loaded with the boxes at the top of the dialog. &#34;Show models&#34; will display all iterations of the models that are created from averaging the subset of particles that meets the minimum similarity against seed. &#34;Show MSE&#34; will display a mean squared error (MSE) image that highlights the areas of the model that are most variable. &#34;Show particles&#34; will display the subset of particles that are used to create the models, registered according to the highest iteration of the model that is displayed.</li>
	</ol>
	</li>
	<li>Select &#34;Show preview&#34; to generate preview models and view the results. 
	NOTE: This is the most computationally intensive step of the process. The running time for a set of a few thousand particles with diameters of a few tens of pixels on a desktop PC should be around 10 min. If computation time is an issue, users should first try the algorithm on a smaller subset of the data or use a smaller ROI radius in step 2.2.4, if possible.</li>
	<li>View the generated models and optimize the parameters, especially the minimum similarity against seed. Increase the minimum similarity against seed until only true particles of the modeled morphology are included in the model.</li>
	<li>Increase the maximum number of iterations by using the &#34;Maximum number of iterations&#34; slider or by entering a number into the box and allow the model generation process to iterate. Use a value around 10 for maximal iteration.</li>
	<li>Name models and select &#34;OK&#34; to save model evolution stacks that contain all iterations of the final model generation process. 
	NOTE: If the minimum similarity against seed is so high that no particles have this similarity, nothing will be updated. If the plugin appears to be frozen, consider the possibility that the minimum similarity is too high.</li>
</ol>

<ol>
	<li>Henriques, R., Griffiths, C., Rego, E. H., Mhlanga, M. M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=PALM+and+STORM:+Unlocking+live-cell+super-resolution.">PALM and STORM: Unlocking live-cell super-resolution.</a> Biopolymers. 95 (5), 322-331 (2011).</li><li>Gustafsson, N., Culley, S., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Fast+live-cell+conventional+fluorophore+nanoscopy+with+ImageJ+through+super-resolution+radial+fluctuations.">Fast live-cell conventional fluorophore nanoscopy with ImageJ through super-resolution radial fluctuations.</a> Nat Commun. 7, 12471 (2016).</li><li>Cheng, Y., Grigorieff, N., Penczek, P. A., Walz, T. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+Primer+to+Single-Particle+Cryo-Electron+Microscopy.">A Primer to Single-Particle Cryo-Electron Microscopy.</a> Cell. 161 (3), 438-449 (2015).</li><li>Szymborska, A., de Marco, A., Daigle, N., Cordes, V. C., Briggs, J. A. G., Ellenberg, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Nuclear+pore+scaffold+structure+analyzed+by+super-resolution+microscopy+and+particle+averaging.">Nuclear pore scaffold structure analyzed by super-resolution microscopy and particle averaging.</a> Science. 341 (6146), 655-658 (2013).</li><li>Broeken, J., Johnson, H., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Resolution+improvement+by+3D+particle+averaging+in+localization+microscopy.">Resolution improvement by 3D particle averaging in localization microscopy.</a> Methods Appl Fluoresc. 3 (1), 14003 (2015).</li><li>Burns, S., Avena, J., Unruh, J., Yu, Z. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Structured+illumination+with+particle+averaging+reveals+novel+roles+for+yeast+centrosome+components+during+duplication.">Structured illumination with particle averaging reveals novel roles for yeast centrosome components during duplication.</a> Elife. , (2015).</li><li>Lelek, M., Di Nunzio, F., Henriques, R., Charneau, P., Arhel, N., Zimmer, C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Superresolution+imaging+of+HIV+in+infected+cells+with+FlAsH-PALM.">Superresolution imaging of HIV in infected cells with FlAsH-PALM.</a> Proc Nat Acad Sci U S A. 109 (22), 8564-8569 (2012).</li><li>Laine, R. F., Albecka, A., Svan de Linde, ., Rees, E. J., Crump, C. M., Kaminski, C. F. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Structural+analysis+of+herpes+simplex+virus+by+optical+super-resolution+imaging.">Structural analysis of herpes simplex virus by optical super-resolution imaging.</a> Nat Commun. 6, 5980 (2015).</li><li>Schindelin, J., Arganda-Carreras, I., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Fiji:+an+open-source+platform+for+biological-image+analysis.">Fiji: an open-source platform for biological-image analysis.</a> Nat Methods. 9 (7), 676-682 (2012).</li><li>Gray, R. D. M., Beerli, C. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=VirusMapper:+open-source+nanoscale+mapping+of+viral+architecture+through+super-resolution+microscopy.">VirusMapper: open-source nanoscale mapping of viral architecture through super-resolution microscopy.</a> Sci Rep. 6, 29132 (2016).</li><li>Gustafsson, M. G. L. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Surpassing+the+lateral+resolution+limit+by+a+factor+of+two+using+structured+illumination+microscopy.">Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy.</a> J of Micros. 198 (2), 82-87 (2000).</li><li>Hell, S. W., Wichmann, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Breaking+the+diffraction+resolution+limit+by+stimulated+emission:+stimulated-emission-depletion+fluorescence+microscopy.">Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy.</a> Opt Let. 19 (11), 780 (1994).</li><li>Chung, C. -. S., Chen, C. -. H., Ho, M. -. Y., Huang, C. -. Y., Liao, C. -. L., Chang, W. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Vaccinia+virus+proteome:+identification+of+proteins+in+vaccinia+virus+intracellular+mature+virion+particles.">Vaccinia virus proteome: identification of proteins in vaccinia virus intracellular mature virion particles.</a> J Virol. 80 (5), 2127-2140 (2006).</li><li>Moss, B. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Poxviridae:+the+viruses+and+their+replication.">Poxviridae: the viruses and their replication.</a> Fields virology. , (2010).</li><li>Condit, R. C., Moussatche, N., Traktman, P. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=In+a+nutshell:+structure+and+assembly+of+the+vaccinia+virion.">In a nutshell: structure and assembly of the vaccinia virion.</a> Adv Virus Res. 66 (6), 31-124 (2006).</li><li>Schmidt, F. I., Bleck, C. K. E., et al. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Vaccinia+virus+entry+is+followed+by+core+activation+and+proteasome-mediated+release+of+the+immunomodulatory+effector+VH1+from+lateral+bodies.">Vaccinia virus entry is followed by core activation and proteasome-mediated release of the immunomodulatory effector VH1 from lateral bodies.</a> Cell Rep. 4 (3), 464-476 (2013).</li><li>. Nanoj-virusmapper Available from: <a target="_blank" href="https://bitbucket.org/rhenriqueslab/nanoj-virusmapper">https://bitbucket.org/rhenriqueslab/nanoj-virusmapper</a> (2016)</li><li>Fortun, D., Guichard, P., Unser, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Reconstruction+From+Multiple+Poses+in+Fluorescence+Imaging:+Proof+of+Concept.">Reconstruction From Multiple Poses in Fluorescence Imaging: Proof of Concept.</a> IEEE J Sel Topics Signal Process. 10 (1), 61-70 (2016).</li></ol>

The authors have nothing to disclose.

With this method, researchers are equipped to combine the power of SPA and SR microscopy in order to generate high-precision, multi-channel 2D models of the protein architecture of viruses and other macromolecular complexes. However, some important considerations should be taken into account.
Seeds should be chosen to represent a structure that is consistently seen. Thus, the raw data should be inspected carefully before the seeds are chosen. This is important for preventing biased models. Choices can be validated by the examination of the minimum similarity thresholds needed to include a certain number of particles in the models. Clearly, for a choice of seed, the higher this threshold needs to be for a given number of particles, the more that structure is apparent in the data.
The template matching concept is particularly useful when there is heterogeneity in the data. All different structures that are visible should be identified and different models created for each case. By separating heterogeneous structures in one channel but simultaneously creating models in a second channel, patterns may emerge that would not have been immediately evident.
Another consideration to be aware of when using this algorithm is that the iteration procedure will maximize stochastic asymmetry. For example, when modeling a structure with two symmetric maxima, all slight asymmetries between the maxima will be aligned with each other during the iteration, and the final model will thus be maximally asymmetric. If this does not reflect a known symmetry in the structure being modeled, then this should be taken into account. Currently, the only way to avoid this maximization is to limit the number of iterations to 1, although a potential development would be for VirusMapper to incorporate axes of symmetry into the model generation process. Any new versions of VirusMapper will be available on the referenced website (see Materials Table). Users will also find a FAQ here to answer any common queries.
The software as described is applicable to any structure that can be imaged with sufficient resolution to visualize the features that the user wishes to model. Although SPA can improve resolution, it clearly will not improve the visibility of features that are otherwise not visible. This protocol is not, therefore, a method to improve the quality of data. As with any technique, careful sample preparation and optimization of imaging strategy will provide the cleanest data and the best resultant models.
The choice of SR imaging modality is also important and, in general, will depend on the sample at hand. VirusMapper has been validated to work well with SIM and STED10, and it can also be used with high-quality localization microscopy data, but care should be taken in this case, as sparse labeling could cause issues similar to those of asymmetry maximization.
Currently, VirusMapper is the only freely available algorithm for the single-particle analysis of fluorescence images and the only general-purpose 2D SPA averaging software. Other studies that have made use of the same principles4,6,8 have used custom software specialized for each particular study. General-purpose algorithms for the reconstruction of 3D data have been published5,18, although no software was provided.
When used as described in this article, VirusMapper can be used to produce precise, accurate, and robust models of the macromolecular protein architecture of viruses and other complexes. With these models, researchers can take precise measurements of the average dimensions of the structures under study, potentially allowing them to reach biological conclusions that would not have otherwise been possible.
Furthermore, with the multi-channel capabilities of this technique, it is possible to map an unlimited number of proteins and components within complexes and to discover novel protein organization. Examining changes in nanoscale structure in different biologically relevant conditions, such as different stages of a virus life cycle, has the potential to offer valuable insights into biology.

Super-resolution (SR) microscopy has had a major impact on cell biology by providing the ability to image key molecular processes along with the molecular specific labeling crucial to understanding them. SR now enables light microscopy to approach the resolutions (20-150 nm) previously only achievable with electron microscopy (EM) while retaining the major benefits of light microscopy, such as the potential to image live cells1,2. Further, the structural conservation found at the nanoscale level permits the application of single-particle analysis (SPA) to SR data, a concept used extensively in electron microscopy3. Using SPA, many highly conserved copies of a structure can be imaged and averaged together to improve the resolution, precision, or signal-to-noise of the visualized object. When used in combination with SR, SPA has been demonstrated to be a powerful tool for the high-precision mapping of components of the nuclear pore complex4,5, centrosomes6, and viruses, such as HIV7 and HSV-18.
However, the routine combined application of SR and SPA has been challenged by a lack of available software. For this reason, we developed VirusMapper, a plugin to the popular image processing software ImageJ/Fiji9. This is the first freely available software package for generalized SPA with fluorescence images10 designed to provide fast, user-friendly, multi-channel na&#239;ve averaging of structures imaged with SR microscopy. Although designed for viruses, it can be applied to any macromolecular complex in which different molecular species can be imaged, identified, and localized.
VirusMapper can be used to produce high-precision molecular models of any known structure, allowing for the calculation of average dimensions and other parameters. The algorithm design makes it particularly useful for separating populations of structures, providing for the determination of distinct orientations or different morphological states. Additionally, multichannel imaging can be used to employ a reference channel in cases where the underlying structure is well-known, thereby allowing for reference-based structure discovery. The instructions for downloading and installing the software are provided on <a href="https://bitbucket.org/rhenriqueslab/nanoj-virusmapper">https://bitbucket.org/rhenriqueslab/nanoj-virusmapper</a>. Example data can also be found there, and users are advised to practice using the software on the example data before attempting to apply it to their own.
Here, the steps for using this plugin to produce SPA models from raw data are described. The software takes raw images containing single- or multi-labeled structures as input. It returns, subject to a number of parameters that are adjusted as the software is run, SPA models showing the average distributions of the labeled components within the imaged structures.
The goal of this protocol is to produce precise SPA models giving the average localizations of components within imaged structures according to the pipeline outlined in Figure 1. As shown in Figure 1, the software workflow is usefully divided into three stages. The first stage is to segment large images, resulting in stacks of particles for each channel. These particles are the units that will be averaged to create models and to produce seeds for model generation. The second stage is to generate seed images, which are used to register the entire set of particles in the final stage. This is done by choosing a reference channel and manually selecting particles in this channel that will contribute to the seeds. Seeds are chosen in this reference channel but can be generated for all channels. Particles are initially realigned by fitting a 2D Gaussian in this channel. All particles that have been selected and realigned are then averaged to produce a seed. For each common structure seen in the data that is to be modeled, particles should be selected as seeds that clearly and accurately represent that structure. The interface at this stage is also useful for scanning the data for such structures.
The final stage is to generate models using template matching. This is achieved through the registration of the particles originally extracted to the seed images generated in the previous section by cross-correlation. A subset of registered particles is averaged together, and the process is further iterated to reduce model mean squared error, if desired. This subset is determined by setting a minimum similarity against the seed that must be satisfied. When creating models simultaneously in multiple channels, the joint similarity, or the average of the similarities for each channel, is used. The resultant models and the registered particles that contributed to them can then be further analyzed.

<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td>Fiji</td>
			<td></td>
			<td></td>
			<td>Open-source image analysis software</td>
		</tr>
		<tr>
			<td>NanoJ-VirusMapper</td>
			<td>developed by the Henriques lab</td>
			<td></td>
			<td>Open source-Fiji plugin (https://bitbucket.org/rhenriqueslab/nanoj-virusmapper)</td>
		</tr>
		<tr>
			<td>VectaShield antifade mounting medium</td>
			<td>Vector Labs</td>
			<td>H-100</td>
			<td></td>
		</tr>
		<tr>
			<td>Elyra PS1</td>
			<td>Zeiss</td>
			<td></td>
			<td></td>
		</tr>
		<tr>
			<td>ZEN BLACK</td>
			<td>Zeiss</td>
			<td></td>
			<td>Image processing software for SIM</td>
		</tr>
		<tr>
			<td>High performance coverslip</td>
			<td>Zeiss</td>
			<td>&#160;474030-9000-000</td>
			<td></td>
		</tr>
		<tr>
			<td>TetraSpeck beads</td>
			<td>ThermoFisher</td>
			<td>T7279</td>
			<td></td>
		</tr>
	</tbody>
</table>

open-source single-particle analysis for super-resolution microscopy with virusmapper

Super-resolution fluorescence microscopy is currently revolutionizing cell biology research. Its capacity to break the resolution limit of around 300 nm allows for the routine imaging of nanoscale biological complexes and processes. This increase in resolution also means that methods popular in electron microscopy, such as single-particle analysis, can readily be applied to super-resolution fluorescence microscopy. By combining this analytical approach with super-resolution optical imaging, it becomes possible to take advantage of the molecule-specific labeling capacity of fluorescence microscopy to generate structural maps of molecular elements within a metastable structure. To this end, we have developed a novel algorithm &#8212; VirusMapper &#8212; packaged as an easy-to-use, high-performance, and high-throughput ImageJ plugin. This article presents an in-depth guide to this software, showcasing its ability to uncover novel structural features in biological molecular complexes. Here, we present how to assemble compatible data and provide a step-by-step protocol on how to use this algorithm to apply single-particle analysis to super-resolution images.

Here, we demonstrate the software on the model poxvirus, vaccinia virus. One of the most complex mammalian viruses, vaccinia packages around 80 different proteins within a 350 x 270 x 250 nm3 brick-shaped particle13,14. Three substructures are discernible by electron microscopy: a central core, which contains the dsDNA genome; two proteinaceous structures, called lateral bodies, which flank the core; and a single proteolipid bilayer envelope15. The large size, complex structure, and amenability to recombinant fluorescent protein tagging make vaccinia an excellent system to demonstrate the VirusMapper workflow.
Using the software as described here, the distribution of a variety of proteins on the vaccinia virion can be modeled. A protein was labeled and imaged, possibly in combination with another protein of known distribution as a reference, and the software was used as described to produce average models of the localization of that protein on the particle. In this example, two proteins were modeled, the inner core protein L4, and the major lateral body component F17.
A recombinant vaccinia virus which has F17 tagged with GFP and L4 tagged with mCherry16 was used. Purified virus was diluted in 1 mM Tris, pH 9, and bound to washed, high-performance coverslips by coating them for 30 min at room temperature. The samples were then fixed by applying 4% formaldehyde in PBS for 20 min. Coverslips were mounted immediately onto slides in antifade mounting medium. Imaging was carried out by SIM on a commercial SIM microscope. A field of view was selected containing hundreds of viruses and images were acquired using 5 phase shifts and 3 grid rotations with 561 nm (32 &#181;m grating period) and 488 nm (32 &#181;m grating period) lasers. Images were acquired using a sCMOS camera and processed using the microscope software. Channels were aligned based on a multi-colored bead slide imaged with the same image acquisition settings. After SIM reconstruction and channel alignment images were opened in Fiji and concatenated into a single image stack.
Viral particles were extracted from the images using the L4 channel as the reference and without applying any Gaussian blur, as these particles have a central maximum. Around 15,000 particles were extracted in this experiment.
Due to the geometry of vaccinia, the lateral bodies have a distinctly different appearance based on the virus orientation. We visualized two orientations in which either one or two lateral bodies could be distinguished. We referred to these orientations as frontal and sagittal, respectively.
Separate seeds for the frontal and sagittal orientations were selected by searching through the particle list at the &#34;Generate Seeds&#34; stage (Figures 4 and 5); particles that were clearly in one orientation or the other were chosen. The L4 channel was used as the reference channel to align the seeds with one another. Again, no Gaussian blur was necessary. 5 particles for each orientation were selected and were averaged to produce the seeds.
Models were generated for each orientation based on these seeds. Neither a reference channel nor squared intensity values were used. The maximum number of iterations was set initially to 1, and the minimum similarity was set to include around 1,000 particles in each case, which gave a consistent appearance for each orientation. The maximum number of iterations was then increased to allow for the convergence of the model. Models were thus generated for the two orientations in the two channels (Figure 7).
<img alt="Figure 1" src="/files/ftp_upload/55471/55471fig1.jpg" /> 
Figure 1: VirusMapper workflow. The plugin is organized into three main stages. Viral particles are extracted from large images, template images or seeds are selected semi-manually from the data, and final SPA models are generated from the data by referring to the seeds. <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig1large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 2" src="/files/ftp_upload/55471/55471fig2.jpg" /> 
Figure 2: &#34;Extract Viral Structures&#34; dialog. When selecting &#34;Extract Viral Structures&#34;, this dialog will appear. The parameters should be filled with initial estimates for optimal segmentation. &#34;Show preview&#34; can then be selected, allowing the ROIs to be previewed and the parameters to be fine-tuned. <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 3" src="/files/ftp_upload/55471/55471fig3.jpg" /> 
Figure 3: Setting extraction parameters. After previewing the ROIs that will be extracted, the ROI radius, number of ROIs, and maximum ROI overlap are adjusted to achieve a situation like this. ROIs are slightly larger than the particles, all particles are included in an ROI, and ROIs can overlap sufficiently to allow clustered particles to be separated. <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig3large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 4" src="/files/ftp_upload/55471/55471fig4.jpg" /> 
Figure 4: Generating template matching seeds. The &#34;Generate Seeds&#34; dialog (1) sets out the parameters to be assigned. The reference particles sequence (2) allows the user to scan through particles in the reference channel. When a particle is viewed in the reference particles sequence, realigned particles for all channels can be viewed in the realigned particle previews (3). <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig4large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 5" src="/files/ftp_upload/55471/55471fig5.jpg" /> 
Figure 5: Adding seed images. As seeds are added to the &#34;Frames to use&#34; box, the average of all seeds (4) and the frames involved (5) are displayed. Particles which are similar to the current average seeds are suggested in the dialog box (6). <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig5large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 6" src="/files/ftp_upload/55471/55471fig6.jpg" /> 
Figure 6: &#34;Generate Models&#34; dialog. When selecting &#34;Generate Models Based on Seeds,&#34; this dialog will appear. The parameters should be filled with initial estimates for optimal model generation, and the elements of the model generation procedure to be shown during calculation should be selected. &#34;Show preview&#34; can then be selected, allowing the model generation process to run and the parameters to be fine-tuned. <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig6large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Figure 7" src="/files/ftp_upload/55471/55471fig7.jpg" /> 
Figure 7: Models generated with VirusMapper. Vaccinia virions with the L4 core protein tagged with mCherry and the F17 lateral body protein tagged with EGFP were imaged using SIM. Models were then generated with the software, as described in the protocol. Two orientations, frontal and sagittal, are distinguished by the appearance of the lateral bodies. Scale bar = 100 nm. <a href="http://cloudflare.jove.com/files/ftp_upload/55471/55471fig7large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>

Watch this Scientific Journal Video about Open-source Single-particle Analysis for Super-resolution Microscopy with VirusMapper at JoVE.com

Open-source Single-particle Analysis for Super-resolution Microscopy with VirusMapper

This manuscript uses the Fiji-based open-source software package VirusMapper to apply single-particle analysis to super-resolution microscopy images in order to generate precise models of nanoscale structure.

Super-resolution fluorescence microscopy is currently revolutionizing cell biology research. Its capacity to break the resolution limit of around 300 nm ...

open-source-single-particle-analysis-for-super-resolution-microscopy

Research

JoVE Journal

Biology

10.0K Views.  University College London. This manuscript uses the Fiji-based open-source software package VirusMapper to apply single-particle analysis to super-resolution microscopy images in order to generate precise models of nanoscale structure.

Video: Open-source Single-particle Analysis for Super-resolution Microscopy with VirusMapper

Super-resolution Imaging of the Bacterial Division Machinery

Bacterial cell division requires the coordinated assembly of more than ten essential proteins at midcell1,2. Central to this process is the formation of a ring-like suprastructure (Z-ring) by the FtsZ protein at the division plan3,4. The Z-ring consists of multiple single-stranded FtsZ protofilaments, and understanding the arrangement of the protofilaments inside the Z-ring will provide insight into the mechanism of Z-ring assembly and its function as a force generator5,6. This information has remained elusive due to current limitations in conventional fluorescence microscopy and electron microscopy. Conventional fluorescence microscopy is unable to provide a high-resolution image of the Z-ring due to the diffraction limit of light (~200 nm). Electron cryotomographic imaging has detected scattered FtsZ protofilaments in small C. crescentus cells7, but is difficult to apply to larger cells such as E. coli or B. subtilis. Here we describe the application of a super-resolution fluorescence microscopy method, Photoactivated Localization Microscopy (PALM), to quantitatively characterize the structural organization of the E. coli Z-ring8.
	PALM imaging offers both high spatial resolution (~35 nm) and specific labeling to enable unambiguous identification of target proteins. We labeled FtsZ with the photoactivatable fluorescent protein mEos2, which switches from green fluorescence (excitation = 488 nm) to red fluorescence (excitation = 561 nm) upon activation at 405 nm9. During a PALM experiment, single FtsZ-mEos2 molecules are stochastically activated and the corresponding centroid positions of the single molecules are determined with &lt;20 nm precision. A super-resolution image of the Z-ring is then reconstructed by superimposing the centroid positions of all detected FtsZ-mEos2 molecules.
	Using this method, we found that the Z-ring has a fixed width of ~100 nm and is composed of a loose bundle of FtsZ protofilaments that overlap with each other in three dimensions. These data provide a springboard for further investigations of the cell cycle dependent changes of the Z-ring10 and can be applied to other proteins of interest.

We describe a super-resolution imaging method to probe the structural organization of the bacterial FtsZ-ring, an essential apparatus for cell division. This method is based on quantitative analyses of photoactivated localization microscopy (PALM) images and can be applied to other bacterial cytoskeletal proteins.

Bacterial cell division requires the coordinated  assembly of more than ten essential proteins at midcell1,2. Central  to this process is the formation of ...

Test Samples for Optimizing STORM Super-Resolution Microscopy

STORM is a recently developed super-resolution microscopy technique with up to 10 times better resolution than standard fluorescence microscopy techniques. However, as the image is acquired in a very different way than normal, by building up an image molecule-by-molecule, there are some significant challenges for users in trying to optimize their image acquisition. In order to aid this process and gain more insight into how STORM works we present the preparation of 3 test samples and the methodology of acquiring and processing STORM super-resolution images with typical resolutions of between 30-50 nm. By combining the test samples with the use of the freely available rainSTORM processing software it is possible to obtain a great deal of information about image quality and resolution. Using these metrics it is then possible to optimize the imaging procedure from the optics, to sample preparation, dye choice, buffer conditions, and image acquisition settings. We also show examples of some common problems that result in poor image quality, such as lateral drift, where the sample moves during image acquisition and density related problems resulting in the 'mislocalization' phenomenon.

We describe the preparation of three test samples and how they can be used to optimize and assess the performance of STORM microscopes. Using these examples we show how to acquire raw data and then process it to acquire super-resolution images in cells of approximately 30-50 nm resolution.

STORM is a recently developed super-resolution microscopy technique with up to 10 times better resolution than standard fluorescence microscopy ...

Sample Preparation for Single Virion Atomic Force Microscopy and Super-resolution Fluorescence Imaging

Immobilization of virions to glass surfaces is a critical step in single virion imaging. Here we present a technique&#160;adopted from single molecule imaging assays which allows adhesion of single virions to glass surfaces with specificity. This preparation is based on grafting the surface of the glass with a mixture of PLL-g-PEG and PLL-g-PEG-Biotin, adding a layer of avidin, and finally creating virion anchors through attachment of biotinylated virus specific antibodies. We have applied this technique across a range of experiments including atomic force microscopy (AFM) and super-resolution fluorescence imaging. This sample preparation method results in a control adhesion of the virions to the surface.

The attachment of virions to a surface is a requirement for single virion imaging by Super-resolution fluorescence imaging or atomic force microscopy (AFM). Here we demonstrate a sample preparation method for controlled adhesion of virions to glass surfaces suitable for use in AFM and super-resolution fluorescence imaging.

Immobilization of virions to glass surfaces is a critical step in single virion imaging. Here we present a technique&#160;adopted from single molecule ...

Tracking and Quantifying Developmental Processes in C. elegans Using Open-source Tools

Quantitatively capturing developmental processes is crucial to derive mechanistic models and key to identify and describe mutant phenotypes. Here protocols are presented for preparing embryos and adult C. elegans animals for short- and long-term time-lapse microscopy and methods for tracking and quantification of developmental processes. The methods presented are all based on C. elegans strains available from the Caenorhabditis Genetics Center and on open-source software that can be easily implemented in any laboratory independently of the microscopy system used. A reconstruction of a 3D cell-shape model using the modelling software IMOD, manual tracking of fluorescently-labeled subcellular structures using the multi-purpose image analysis program Endrov, and an analysis of cortical contractile flow using PIVlab (Time-Resolved Digital Particle Image Velocimetry Tool for MATLAB) are shown. It is discussed how these methods can also be deployed to quantitatively capture other developmental processes in different models, e.g., cell tracking and lineage tracing, tracking of vesicle flow.

Tracking and Quantifying Developmental Processes in C. elegans Using Open-source Tools

Here it is shown how to track and quantify developmental processes in C. elegans. The methods presented are based on open-source tools that can be easily implemented. It is demonstrated how to reconstruct 3D cell-shape models, how to manually track subcellular structures, and how to analyze cortical contractile flow.

Quantitatively capturing developmental processes is crucial to derive mechanistic models and key to identify and describe mutant phenotypes. Here ...

Three-dimensional Super Resolution Microscopy of F-actin Filaments by Interferometric PhotoActivated Localization Microscopy (iPALM)

Fluorescence microscopy enables direct visualization of specific biomolecules within cells. However, for conventional fluorescence microscopy, the spatial resolution is restricted by diffraction to ~ 200 nm within the image plane and &#62; 500 nm along the optical axis. As a result, fluorescence microscopy has long been severely limited in the observation of ultrastructural features within cells. The recent development of super resolution microscopy methods has overcome this limitation. In particular, the advent of photoswitchable fluorophores enables localization-based super resolution microscopy, which provides resolving power approaching the molecular-length scale. Here, we describe the application of a three-dimensional super resolution microscopy method based on single-molecule localization microscopy and multiphase interferometry, called interferometric PhotoActivated Localization Microscopy (iPALM). This method provides nearly isotropic resolution on the order of 20 nm in all three dimensions. Protocols for visualizing the filamentous actin cytoskeleton, including specimen preparation and operation of the iPALM instrument, are described here. These protocols are also readily adaptable and instructive for the study of other ultrastructural features in cells.

Three-dimensional Super Resolution Microscopy of F-actin Filaments by Interferometric PhotoActivated Localization Microscopy iPALM

We present a protocol for the application of interferometric PhotoActivated Localization Microscopy (iPALM), a 3-dimensional single-molecule localization super resolution microscopy method, to the imaging of the actin cytoskeleton in adherent mammalian cells. This approach allows light-based visualization of nanoscale structural features that would otherwise remain unresolved by conventional diffraction-limited optical microscopy.

Fluorescence microscopy enables direct visualization of specific biomolecules within cells. However, for conventional fluorescence microscopy, the spatial ...

Open Source High Content Analysis Utilizing Automated Fluorescence Lifetime Imaging Microscopy

We present an open source high content analysis instrument utilizing automated fluorescence lifetime imaging (FLIM) for assaying protein interactions using F&#246;rster resonance energy transfer (FRET) based readouts of fixed or live cells in multiwell plates. This provides a means to screen for cell signaling processes read out using intramolecular FRET biosensors or intermolecular FRET of protein interactions such as oligomerization or heterodimerization, which can be used to identify binding partners. We describe here the functionality of this automated multiwell plate FLIM instrumentation and present exemplar data from our studies of HIV Gag protein oligomerization and a time course of a FRET biosensor in live cells. A detailed description of the practical implementation is then provided with reference to a list of hardware components and a description of the open source data acquisition software written in &#181;Manager. The application of FLIMfit, an open source MATLAB-based client for the OMERO platform, to analyze arrays of multiwell plate FLIM data is also presented. The protocols for imaging fixed and live cells are outlined and a demonstration of an automated multiwell plate FLIM experiment using cells expressing fluorescent protein-based FRET constructs is presented. This is complemented by a walk-through of the data analysis for this specific FLIM FRET data set.

We present an open source high content analysis (HCA) instrument utilizing automated fluorescence lifetime imaging (FLIM) for assaying protein interactions using F&#246;rster resonance energy transfer (FRET) based readouts. Data acquisition for this openFLIM-HCA instrument is controlled by software written in &#181;Manager and data analysis is undertaken in FLIMfit.

We present an open source high content analysis instrument utilizing automated fluorescence lifetime imaging (FLIM) for assaying protein interactions ...

Ground State Depletion Super-resolution Imaging in Mammalian Cells

Advances in fluorescent microscopy and cell biology are intimately correlated, with the enhanced ability to visualize cellular events often leading to dramatic leaps in our understanding of how cells function. The development and availability of super-resolution microscopy has considerably extended the limits of optical resolution from ~250-20 nm. Biologists are no longer limited to describing molecular interactions in terms of colocalization within a diffraction limited area, rather it is now possible to visualize the dynamic interactions of individual molecules. Here, we outline a protocol for the visualization and quantification of cellular proteins by ground-state depletion microscopy&#160;for fixed cell imaging. We provide examples from two different membrane proteins, an element of the endoplasmic reticulum translocon, sec61&#946;, and a plasma membrane-localized voltage-gated L-type Ca2+ channel (CaV1.2). Discussed are the specific microscope parameters, fixation methods, photo-switching buffer formulation, and pitfalls and challenges of image processing.

This article outlines a protocol for the detection of one or more plasma and/or intracellular membrane proteins using Ground State Depletion (GSD) super-resolution microscopy in mammalian cells. Here, we discuss the benefits and considerations of using such approaches for the visualization and quantification of cellular proteins.

Advances in fluorescent microscopy and cell biology are intimately correlated, with the enhanced ability to visualize cellular events often leading to ...

Application of High-speed Super-resolution SPEED Microscopy in Live Primary Cilium

The primary cilium is a microtubule-based protrusion on the surface of many eukaryotic cells and contains a unique complement of proteins that function critically in cell motility and signaling. Since cilia are incapable of synthesizing their own protein, nearly 200 unique ciliary proteins need to be trafficked between the cytosol and primary cilia. However, it is still a technical challenge to map three-dimensional (3D) locations of transport pathways for these proteins in live primary cilia due to the limitations of currently existing techniques. To conquer the challenge, recently we have developed and employed a high-speed virtual 3D super-resolution microscopy, termed single-point edge-excitation sub-diffraction (SPEED) microscopy, to determine the 3D spatial location of transport pathways for both cytosolic and membrane proteins in primary cilia of live cells. In this article, we will demonstrate the detailed setup of SPEED microscopy, the preparation of cells expressing fluorescence-protein-labeled ciliary proteins, the real-time single-molecule tracking of individual proteins in live cilium and the achievement of 3D spatial probability density maps of transport routes for ciliary proteins.

Recently we mapped the three-dimensional (3D) spatial locations of transport routes for various proteins translocating inside primary cilia in live cells. Here this paper details the experimental setup, the process of biological samples and the data analyses for the 3D super-resolution fluorescence imaging approach newly applied in live primary cilia.

The primary cilium is a microtubule-based protrusion on the surface of many eukaryotic cells and contains a unique complement of proteins that function ...

Isolation and Fluorescence Imaging for Single-particle Reconstruction of Chlamydomonas Centrioles

Centrioles are large macromolecular assemblies important for the proper execution of fundamental cell biological processes such as cell division, cell motility, or cell signaling. The green algae Chlamydomonas reinhardtii has proven to be an insightful model in the study of centriole architecture, function, and protein composition. Despite great advances toward understanding centriolar architecture, one of the current challenges is to determine the precise localization of centriolar components within structural regions of the centriole in order to better understand their role in centriole biogenesis. A major limitation lies in the resolution of fluorescence microscopy, which complicates the interpretation of protein localization in this organelle with dimensions close to the diffraction limit. To tackle this question, we are providing a method to purify and image a large number of C. reinhardtii centrioles with different orientations using super-resolution microscopy. This technique allows further processing of data through fluorescent single-particle averaging (Fluo-SPA) owing to the large number of centrioles acquired. Fluo-SPA generates averages of stained C. reinhardtii centrioles in different orientations, thus facilitating the localization of distinct proteins in centriolar sub-regions. Importantly, this method can be applied to image centrioles from other species or other large macromolecular assemblies.

Isolation and Fluorescence Imaging for Single-particle Reconstruction of Chlamydomonas Centrioles

We have developed a strategy to purify and image a large number of centrioles in different orientations amenable for super-resolution microscopy and single-particle averaging.

Centrioles are large macromolecular assemblies important for the proper execution of fundamental cell biological processes such as cell division, cell ...

Super-Resolution Live Cell Imaging of Subcellular Structures

There has long been a crucial tradeoff between spatial and temporal resolution in imaging. Imaging beyond the diffraction limit of light has traditionally been restricted to be used only on fixed samples or live cells outside of tissue labeled with strong fluorescent signal. Current super-resolution live cell imaging techniques require the use of special fluorescence probes, high illumination, multiple image acquisitions with post-acquisition processing, or often a combination of these processes. These prerequisites significantly limit the biological samples and contexts that this technique can be applied to.
Here we describe a method to perform super-resolution (~140 nm XY-resolution) time-lapse fluorescence live cell imaging in situ. This technique is also compatible with low fluorescent intensity, for example, EGFP or mCherry endogenously tagged at lowly expressed genes. As a proof-of-principle, we have used this method to visualize multiple subcellular structures in the Drosophila testis. During tissue preparation, both the cellular structure and tissue morphology are maintained within the dissected testis. Here, we use this technique to image microtubule dynamics, the interactions between microtubules and the nuclear membrane, as well as the attachment of microtubules to centromeres.
This technique requires special procedures in sample preparation, sample mounting and immobilizing of specimens. Additionally, the specimens must be maintained for several hours after dissection without compromising cellular function and activity. While we have optimized the conditions for live super-resolution imaging specifically in Drosophila male germline stem cells (GSCs) and progenitor germ cells in dissected testis tissue, this technique is broadly applicable to a variety of different cell types. The ability to observe cells under their physiological conditions without sacrificing either spatial or temporal resolution will serve as an invaluable tool to researchers seeking to address crucial questions in cell biology.

Presented here is a protocol for super-resolution live-cell imaging in intact tissue. We have standardized the conditions for imaging a highly sensitive adult stem cell population in its native tissue environment. This technique involves balancing temporal and spatial resolution to allow for the direct observation of biological phenomena in live tissue.

There has long been a crucial tradeoff between spatial and temporal resolution in imaging. Imaging beyond the diffraction limit of light has traditionally ...