Mapping Absolute DNA Density in Cell Nuclei using Single-molecule Localization Microscopy

Summary

The present protocol describes a method measuring absolute DNA densities within adherent cell nuclei using Voronoi tessellation of single-molecule localization microscopy data, known volume, genome size, and cell cycle stage.

Abstract

Within the cell nucleus, silent genes are generally located in chromatin areas of high density called heterochromatin, whereas active genes can be mostly found at the interface between chromatin and the interchromatin space called euchromatin. At present, the characterization of eu- and heterochromatin is mostly based on epigenetic modifications to histone proteins along the DNA sequence, while little is known about absolute DNA densities across the cell nucleus and their functional implications. Diagrams of the nucleus solely based on biochemical data and assumptions about the nature of chromatin as a polymer differ fundamentally from imaging data generated by high-resolution microscopy. This indicates that these methods are not well-suited for measuring density relationships in situ. We believe that spatial constraints might be involved in gene regulation and have therefore developed a method that allows the measurement of absolute DNA densities in mammalian cell nuclei by transforming super-resolution localization data into true-to-scale density maps by Voronoi tessellation.

Introduction

Since the early days of cell biology, the cell nucleus-the seat of genetic information-has fascinated biologists. After applying self-developed cytological staining methods, Emil Heitz discovered, in 1928, differently, intensely stained areas within the cell nucleus¹. He called the more intensely stained, dense areas "heterochromatin", whereas he named the less strongly stained, less dense areas "euchromatin". Over time, it became apparent that active genes are mostly located in euchromatin, while denser areas were found to be rich in repetitive elements and silent genes. The terms heterochromatin and euchromatin have survived to the present day although their definition has changed from structural to molecular properties (see below). Today we know that the cell nucleus is divided into two main phases. One liquid (the interchromatin compartment), which houses the nuclear bodies, the splicing compartment, and the nucleoplasm, and the other solid, hydrogel-like-the chromatin domain, which includes the chromosome territories and the nucleoli². At the interface to the interchromatin space, chromatin is less dense, and this is where mostly the genetically active processes such as transcription, replication, and repair take place^3,4,5, while adjacent to the lamina, around the nucleoli, and near the centromeres, chromatin shows high compaction levels and is in general transcriptionally rather inactive⁶. At the molecular level, chromatin is built from DNA wrapped around nucleosomes (an octamer of core histone proteins).

The histones possess intrinsically disordered tail domains that can be modified post-translationally by adding several chemical groups (methyl-, acetyl, phosphor-, biotin), amino-acids (arginine), or even proteins (SUMO, Ubiquitin)⁷. These modifications can be read and attract other proteins (e.g., chromatin remodeler, HP1, repair proteins, RNA) and thereby define the physiological state of the DNA sequence (transcriptionally permissive/restrictive) lying within, without directly changing it (therefore they are called "epi"genetic)⁷. The type and number of modifications around a specific sequence can differ profoundly between cell types, are reversible, and can change throughout development, senescence, and disease^8,9,10. There are modifications of histone tails that are more prevalent in highly transcribed regions and modifications that predominate in regions of the genome that have little or no transcriptional activity.

For example, highly transcribed regions that are rich in the H3K4 modifications or whose histone tails are acetylated are usually described as euchromatin, whereas areas rich in repressive histone modifications (H3K9me3, H3K27me3, or H4K20me3) are referred to as heterochromatin, which is further divided into constitutive heterochromatin (transcriptionally inactive in all cells) (e.g., H4K20me3) and facultative heterochromatin (chromatin silenced depending on cell type, e.g., H3K27me3)⁶. Although biochemical and molecular biological methods are very well suited for measuring chemical changes at the nanoscale, it is much more challenging using them to make statements about spatial properties at the mesoscale using these methods.

For example, attempts have been made using contact data from Hi-C experiments, which detect and map close proximities of DNA between sequence regions, to reconstruct spatial models of the genome¹¹. However, direct comparison with high-resolution light and electron microscopy images shows that this is only possible to a minor extent (Figure 1). Although methods such as Hi-C¹², can detect chromosome territories in interphase cells correctly as separate entities, these models are rather "near-sighted", because DNA sequence proximities alone are blind to reflect spatial relationships between more distant parts of the genome and thus are not very well suited for a proper 3D genome reconstruction. Therefore, density differences also cannot be reflected properly by contact frequency information. This leads to "spaghettified" representations of the genome in the nucleus (see Figure 1B), which appear to occur in wool-cluster-like, freely accessible loops.

To pave the way for an integrative, more realistic picture of the nucleus that combines biochemical information with biophysical and structural data, methods need to be developed that allow generating data on different aspects of nuclear organization. Until recently, it has also been difficult to determine absolute DNA densities in cell nuclei from structural data. The reasons for this were the limited resolution of conventional light microscopes, problems in electron microscopy to specifically stain DNA, and the small observation volumes in electron microscopic thin sections, which usually must be significantly thinner than a mammalian cell nucleus to be penetrable by electrons.

The resolution of conventional fluorescence microscopes is limited by the diffraction of light. The image of a point source of light is spread out due to the diffraction limit and can be described by the Point-Spread-Function (PSF). According to the PSF, the image of a fluorophore, which can be regarded in good approximation as a point source of light, occupies a volume of a certain size around its source of origin (fluorophore)¹³. When many fluorophores, located much closer to each other than the dimensions of their diffraction-limited images, are excited simultaneously, the imaged intensity distributions are superimposed, and the location of the single fluorophores cannot be resolved. Sequential, stochastic light emission from fluorophores (blinking) allows optical isolation of individual molecules and, thus, finding their exact locations by determining the intensity gravity centers of their signals.

This can be used to reconstruct structural information of samples by the accumulation of fluorophore localization data from many recorded images. This method is generally referred to as single-molecule localization microscopy (SMLM) (for further information see¹⁴). The more photons a fluorophore emits during its 'on-time,' the more precisely can the intensity gravity centers be determined. Astigmatic lenses in the beam path transform signals of fluorophores above or below the optical focal plane into ellipses, which can be used to determine their position along the optical axis. The long axes of the ellipses originating from fluorescent signals below the focal plane are rotated at 90° compared to the axes of the ellipses originating from fluorophores above the focal plane. Further, the axes-ratio of these ellipses allows determining the molecule's position along the optical axis relative to the focal plane within a range of ±300 nm¹⁵.

The quality of super-resolution images generated by the reconstruction of stochastic blinking events strongly depends on the labeling density and the number of blinking events. The latter depend on the photostability of the fluorophores and the number of blinking events before they eventually fail (number of on/off cycles). The method described here for obtaining super-resolution images of intranuclear DNA distribution is termed fBALM (DNA structure fluctuating-assisted binding-activated localization microscopy). It is based on fluorophores that transiently intercalate into nucleic acids and only fluoresce once they are bound to the DNA^16,17. Owing to the charged residues of its phosphor-diester backbone, DNA is a highly negatively charged polymer. Stabilization of the complementary DNA strands in living cells requires neutralization by positively charged proteins (e.g., histones) and ions. By lowering the pH, the stability of the complementary pairing of the bases is reduced to allow intercalating dyes to diffuse in and out^16,17.

Depending on the intercalating fluorescent dye, this state can be reached within a certain pH range. As fluorescent dyes such as YoYo-1 and SYTOX Orange only fluoresce when bound to DNA and because intercalating dyes (in contrast to dyes that bind the minor groove of the DNA such as Hoechst and 4',6-diamidino-2-phenylindole [DAPI]) usually bind in a sequence-independent manner, they are well-suited for mapping the distribution of DNA within cell-nuclei by localization microscopy.

Voronoi tessellation is a mathematical method allowing the subdivision of space into different partitions based on the location of the points. In 2D-Space, the size of the resulting tiles reflects the inverse of the density of points¹⁸. Because localization microscopy reconstructs images as a set of points that represent the location of fluorophores, Voronoi tessellation can help determinine the density of localization signals (Figure 2). Thus, using a DNA-specific dye as a fluorophore allows for measuring DNA densities.

A priori knowledge of the DNA content (number of base pairs) and the spatial dimension of the nucleus allows the transformation of the relative DNA densities into absolute DNA densities. The following protocol shows the mapping of absolute DNA densities in adherent cells using SMLM at a very high resolution and demonstrates that these densities are subject to large variations.

Protocol

NOTE: Figure 3 gives an overview of the workflow described in this section. See the Table of Materials for details related to the reagents, materials, equipment, and software used in this protocol. The code used for this publication can be viewed and downloaded here: https://github.com/irradiator/Mapping-absolute-DNA-density-in-cell-nuclei-using-SMLM-microscopy.

1. Cell culture

1. Cultivate C3H10T1/2, HFB, and HeLa cells in DMEM supplemented with 10% fetal bovine serum at 37 °C in a humidified atmosphere supplemented with 5% CO₂. When cells are close to confluency, subculture the cells by splitting them at a ratio of 1:3 (C3H10T1/2, HFB) and 1:10 (HeLa), respectively.
2. One day before performing the measurements, seed the cells from exponentially growing cultures onto a 35 mm gridded dish. Ensure that the observation dishes provide excellent optical quality for fluorescence microscopy (glass bottom with a thickness of 170 µm, #1.5) and possess a grid to unambiguously locate cells. Ensure also that the cells are in a single-cell suspension and that the whole area of the dish's surface is covered homogeneously with cells.
   NOTE: It is recommended to use dishes with an imprinted grid to once again find the exact cells that were assigned to a certain cell cycle stage.
3. If required by the experiment, add 100 nM Trichostatin A (TSA) to the cell culture medium and incubate the cells under conditions previously described in step 1.1.

2. Sample preparation

1. On the day of the measurement, wash the cells once with 1x DPBS before fixing them for 30 min in 4% paraformaldehyde (PFA) in 1x DPBS.
   NOTE: For reproducible results, make sure that a fresh PFA solution is prepared each time and take special care that the PFA concentration is accurate.
2. After removing the fixative, wash the cells again 2x with 1x DPBS before permeabilizing them for 20 min in 1x DPBS containing 0.4% Triton X-100.
3. Remove the permeabilization buffer with a pipette and perform an additional washing step with 1x DPBS before treating the cells with an RNase cocktail (dilution 1:1,000 in 1x DPBS). Place the cells in an incubator for 30 min at 37 °C in a humidified chamber.
4. Prepare the staining solution by diluting the fBALM-capable SYTOX Orange stock solution (5 mM in 1x DPBS at 1:10,000 dilution).
5. Remove the RNase buffer using a pipette and apply an additional washing step with 1x DPBS before adding the staining solution. Incubate the cells at room temperature in a humidified chamber for 5 min.
6. Remove the SYTOX Orange staining solution from the cells with a pipette, wash with 1x DPBS, and store the cells in 1x DPBS, protected from intense light.

3. Cytometric cell cycle determination

NOTE: For this step, an inverted high-content screening microscope was used here, but any automated, inverse widefield fluorescence microscope may be used. The intensity of the whole nucleus is a measure of its DNA content; hence, make sure to completely open the pinhole if using a confocal system. Low numerical aperture (NA) objective air lenses are preferable to oil immersion objective lenses.

1. Put the dish with the cells onto an inverse fluorescence microscope equipped with a motorized stage and software that allows the reconstruction of large areas on the sample by stitching automatically recorded images together.
2. Choose an objective lens that allows recording of sufficient cells per field of view (at least 900 single nuclei in total [for details, see Roukos et al.¹⁹]) so that the image acquisition for the cell cycle determination can be performed within a reasonable time. Use a lens that does not show large differences in intensity between the center and the borders. If such a lens is not available, apply a flat-field correction to the images (see step 7). To follow this protocol, use a wide field fluorescence microscope equipped with a 20x NA 0.4 air lens with a depth of focus = 8 µm.
3. During image acquisition, cover the central area of the dish using appropriate filters for the used fluorophore. Take brightfield images in transmitted light mode.
   NOTE: Brightfield images are important for relocation of cells on the grid in step 4.6.
4. Place the dish with the cells into a refrigerator at 4 °C until step 4.4.
5. Reconstruct the scanned area from the single images by tiling together the recorded single images.
6. Transfer the image data of the fluorescent DNA stain onto a computer that runs the open source software CellProfiler4.2.1²⁰.
7. If a flat-field correction of the fluorescence images is necessary, generate a reference image of the used microscope from the recorded pictures. Open the Illumination Correction.cpproj pipeline (included as Supplemental File 1); observe that the pipeline appears as a sequence on the screen in which different steps can be clicked. Drag and drop the images into the designated field in the first step images of the pipeline.
   NOTE: Make sure to exclude all non-image data in the listed files by using the filter options in the lower part of the window.
8. In the last step of the illumination correction pipeline, click on Save Images to define a location where the reference images will be stored, and then, click on the Analyze Images button.
9. Open the pipeline CellCycleAnalysis.cpproj (included as Supplemental File 2). Drag and drop the images recorded in step 3.3 (see Figure 4A as an example), including the reference image (step 3.8), into the designated field in the first step Images of the pipeline.
   NOTE: Make sure to exclude all non-image data in the listed files by using the filter options in the lower part of the window.
10. In the CorrectIlluminationApply step of the pipeline, choose the reference image in the dropdown menu by selecting the Select the Illumination function.
11. In the last step of the cell cycle analysis pipeline, define a location where the reference image was stored in step 3.7 and click on the Analyze Images button to start bulk measuring the integrated fluorescence intensities of all nuclei within the images recorded in step 3.3.
    NOTE: CellProfiler4.2.1 will generate a text file "CellCycleAnalysis_Nuclei.txt," which contains the data of each measured nucleus (Image, ObjectID, integrated fluorescence intensity, Image coordinates) in the CSV format.
12. Make sure that the DisplayHistogram element of the pipeline is checked as active.
13. Note the integrated fluorescence intensities in the histogram for the G₁ and G₂ (see Figure 4B as an example) peaks and enter the intervals around them into the FilteredObjects elements of the pipeline. Activate the pipeline elements by clicking on the check boxes.
14. Activate the corresponding DisplayDataOnImage pipeline elements to generate an image highlighting the cells in G₁ or G₂ phase and run the pipeline again (see Figure 4C as an example).
15. Choose up to five cells representing the desired cell cycle phase from the generated image in the marked area of the observation chamber and try to relocate the cells on the SMLM microscope.
    NOTE: Since two cell lines with unknown genome size are used here, the genome sizes were determined by comparing G₁ peaks of a human fibroblast cell line to the G₁ peaks of HeLa and C3H10T1/2 cells. The histograms and the proportional relationship between their genome sizes can be viewed in Supplemental Figure S1.

4. SMLM-fBALM

NOTE: Although the net sum of oxygen in these combined biochemical reactions is zero, it is recommended to carry out the reaction in a dish that is not sealed, since limited oxygen concentrations can slow down the production of D-glucono-1,5-lactone. The increasing concentration of D-glucono-1,5-lactone gradually lowers the pH, which is necessary since an immediate lowering of the pH will alter the ultrastructure of the specimen.

1. Prepare 1 mL of the imaging buffer by mixing the following ingredients: 900 µL of glucose (1 g/mL of glucose in 1x DPBS), 50 µL of glucose oxidase (0.5 mg/mL in 1x DPBS), and 50 µL of catalase (40 µg/mL in 1x DPBS).
2. Remove the 1x DPBS from the sample using a pipette and add 1 mL of the imaging buffer into the dish containing the cells.
3. Mount the dish onto the stage of an SMLM-capable microscope. Use a high NA objective lens and a high optical magnification.
4. Locate one of the chosen nuclei (from step 3.15) using the dish's coordinate system and center it in the field of view of the camera.
   NOTE: The enzymatic reaction takes ~60-180 min until the correct pH for the fBALM method is reached and facilitates appropriate blinking. It is recommended to mount the observation dish as soon as possible to provide sufficient time for the sample to adjust to the temperature of the instrument since temperature changes can lead to axial drift. Make sure that the temperature in the room housing the SMLM microscope is stable throughout the measurement.
5. Determine the height of the nucleus by changing the focus with the microscope's z-drive and recording the position of the upper and the lower limits.
6. At different timepoints, check for proper blinking produced by the fBALM process by increasing the laser power. Once the sample shows proper stochastic blinking, begin the acquisition of the image series needed for the reconstruction of the super-resolved SMLM.
   NOTE: The microscope used for fBALM should be equipped with excitation lasers suitable to excite the used fluorophore and that can generate sufficiently high photon densities to ensure sufficient photon counts per frame. Test this in a pilot experiment before all the other experiments.
7. Once proper blinking of the fBALM process has been assured, begin the acquisition of the image series using an exposure time of 50 ms, resulting in a frame rate of ~20 fps.
8. Take a z-stack through the nucleus (around the mid-section) with 200 nm step intervals and only 500 frames per light optical section. Back at the mid-section, take an image series of 50,000 frames to collect enough blinking events for a good reconstruction of structural details at high localization precision (this will take ~45 min).
   ​NOTE: The z-stack is important to determine the mid section's fraction of the DNA signal with respect to the whole nucleus. Make sure that the effective pixel size is well suited for SMLM²² and the signal on the camera does not saturate the sensor.
9. Ensure the following requirements are met
   1. Make sure that the acquisition computer contains enough disk space as, depending on the number and the size of 16-bit images, the file sizes of the image stacks can become quite large.
   2. Ensure that the computer running the acquisition software can handle a file system on the used storage device capable of handling large file sizes (>4 GB) and that the transfer rate to the storage device is sufficiently high to write the files in real-time when directly recording image data to the storage device.
   3. When saving the data after image acquisition, make sure that the computer is equipped with sufficient RAM (e.g., 64 GB).

5. Preparing and recording a dish with fluorescent beads for z-calibration of the SMLM microscope

NOTE: For proper axial calibration of the microscope's astigmatic lenses to assign correct locations along the optical axis, recording z-stacks of fluorescent 100 nm beads should be considered.

1. Dilute beads 1:10,000 in 1x DPBS and seed 100 nm beads onto the same type of dish used for the SMLM measurements.
   NOTE: Only use the best optical quality and specifications such as #1.5 coverslips.
2. Mount the calibration dish onto the microscope; record image stacks through the focal plane of the beads (~1.5 µm in total; 10 nm steps used here)¹⁵.
3. Transfer the data to the data analysis computer.
   ​NOTE: Do not store and use the calibration slides for a long time.

6. SMLM data processing

NOTE: The ImageJ²³ plug-in "ThunderSTORM"²⁴ was used for the registration of the localizations (e.g., conversion of the blinking spots in the image stack into a list containing localization coordinates, frame number, etc.). ImageJ or Fiji²⁵ runs on all major desktop operating systems (Linux/Windows/macOS) and can be downloaded and used free of charge.

1. To use ThunderSTORM in FijI, be sure to add the Hohlbein Lab's update site under HELP | Update | Manage Update Sites, check the checkbox for the Hohlbein Lab, and restart Fiji. Make sure that the program has access to sufficient memory.
   NOTE: The image data need to be in a file format that can be read by ImageJ/Fiji. When recording the data on a standard commercial system, directly open the data using the Bio-Formats plug-in for ImageJ (that automatically comes with Fiji). In this case, a MATLAB script h52tif.m (provided on the GitHub page mentioned at the beginning of the protocol) was used to convert the data recorded recorded by our custom set-up from the hdf5 format into the TIFF format.
   1. Depending on the amount of memory available on the used system and the number of frames to be recorded, split the data into several substacks to avoid running out of memory.
      NOTE: In the following, the settings that were used for extracting the data shown in the results section are being used. For a more detailed explanation of the ThunderSTORM settings, refer to the ThunderSTORM tutorials, which can be found online.
2. Optimize the settings for the detection of blinking signals by adapting the ThunderSTORM settings according to the desired recording conditions. Click on ThunderSTORM | Run Analysis, choose Camera Setup and enter the correct pixel size of the image data, the A/D count, Quantum efficiency of the used camera, base level, and EM gain (for this setup, pixel size: 65 nm, A/D count: 0.64, quantum efficiency: 0.8).
3. Next, choose the algorithm to determine the localization coordinates by fitting the blinking signals. In the Image Filtering section, use the Wavelet filter (B-Spline) with a B-Spline order 3 and a B-Spline scale 2.0. For the other settings, set Approximate localization of molecules method to Local maximum, Peak intensity threshold to 2*std(Wave.F1), and Connectivity to 8-neighborhood. In the Sub-pixel localization of molecules section, choose PSF: Integrated Gaussian, Fitting radius [px]: 3, Fitting method: Maximum likelihood, and Initial sigma [px]: 1.6.
4. Run the plug-in to create a table containing an entry for each localization detected and its properties. Save the table to the hard drive.
5. If multiple image stacks need to be analyzed or the image stacks have to be split into several stacks (e.g., to allow the analysis of large datasets or in case not much memory in the data analysis computer is available), automate localization detection by running a macro.
6. If the acquisition data had to be partitioned into multiple image stacks, concatenate the localization tables in ThunderSTORM by importing one file after another. Ensure that the Append To Current Table  option in the Import menu is activated and that the correct starting number is used to avoid overwriting localizations in the table.
7. Once the complete localization table has been generated, check the result by reconstructing the image from the locations in the results table. Press the Visualization button in the ThunderSTORM results window and wait for the reconstructed image (e.g., histogram of the localizations) to appear.
8. Check the reconstructed picture for problems such as lateral drift and other imaging artifacts. Measure and correct drift in the Drift submenu by determining the cross-correlation of summed up subsections of the data stack (used here) or by using fiducial markers in the sample. After pressing the >> menu, enter 5 bins and the magnification factor (6.5 in this protocol). Wait for a window to appear showing the x- and y-drift.
   NOTE: The first images after recording might still suffer from a strong background fluorescence (while bringing the fluorophores into the dark state); hence, it can be necessary to exclude the first couple of hundred images from the analysis (e.g., type "frame >500" into the Filter field in the ThunderSTORM main window).
9. To avoid overcounting signals that are visible over several consecutive frames, apply merging to the localization data using a radius that considers the average localization precision (20 nm here), 1 dark frame, and 0 max frames (no restriction on the maximum on-time of a molecule).
10. If only a thin light optical ultrathin section must be analyzed, exclude all localizations above and below the focal plane from the dataset. First, look for the point in z that contains the most localization by pressing Plot Histogram in the window containing the results table, and then discard the localizations 50 nm above and below this point by using the filtering command:
    z > "histogram peak" - 50 & z < "histogram peak" + 50
11. For determining the fraction of DNA in the recorded section, determine the localizations in each plane of the image stack (thickness 200 nm; 100 nm on each side) recorded in step 4.7.1. Open the first localization results table of the stack by choosing plug-ins | ThunderSTORM | Import/Export | Import Results and filter each recorded slice to 200 nm. For this, click on Apply after typing in the filter field:
    z > -100 & z < 100
12. Click on Visualization and choose the Histograms option to generate an image of each slice. Save the resulting image in a folder in the TIFF format. Close all the open images. Repeat this for all the image planes.
13. Open all slices and combine them by using Image | Stacks | Images to Stack. Choose Image | Stacks | Z-Project and choose the Sum option. Open the ROI Manager in ImageJ (Analyze | Tools | ROI-Manager); then, choose the Polygon Selection Tool from the ImageJ toolbar in the ImageJ main window and outline the nucleus. Click on Add in the ROI-Manager.
    1. Click on Analyze | Set Measurements and select Integrated Density. Click on Analyze | Measure. In ThunderSTORM, open the results table of the center plane as described above and filter it to a Thickness of 100 nm by using the following command in the filter field:
       z > -50 & z < 50
14. Generate a histogram of the filtered localizations as described above, activate the defined ROI by selecting it in the ROI Manager and click on Analyze | Measure.
15. Determine the fraction of the localizations in the 100 nm thick mid-section relative to the total number of localizations by dividing the measured integrated densities-the conversion factor needed in step 8.2 for calculating the absolute DNA densities.
    ​NOTE: The DNA content is proportional to the number of localizations. To estimate the fraction of DNA in the imaged super-resolution image (mid-section) it is therefore sufficient to determine, the total number of localizations throughout the volume of the whole nucleus and to determine the total number of localizations in the section of interest. This can be accomplished by generating 2D histograms of the localizations.

7. Z-calibration of the SMLM microscope

1. Open the image stacks recorded in section 5 in Fiji.
2. In the ThunderSTORM plugin menu, select the 3D calibration submenu | Cylindrical lens calibration.
3. Enter the step size used to record the calibration data (10 nm) and define the area of the image stack that will be used for the calibration.
4. Specify a location to save the calibration file.
5. Use the calibration data in the Run Analysis dialog window by selecting PSF: Elliptical Gaussian (3D Astigmatism) in the Sub-pixel localization of molecules panel and adjusting the file path to the calibration file created in section 5.

8. Voronoi Tessellation

NOTE: Once the localization table has been generated and groomed, proceed to the last step of the image analysis-the Voronoi tessellation. Use MATLAB 2021 for this step; the MATLAB scripts are part of the "Localization Analyzer for Nanoscale Distributions" (LAND) software package and can be downloaded from the links in the Table of Materials. Similar to the recording of the data and the generation of the localization table, it is important to use a system that is well-equipped with memory and processing power. The system used to generate the images for this publication was equipped with 128 GB of RAM and a 9-core Intel i9 CPU.

1. Convert the ThunderSTORM localization table from the .csv format into the Orte format by running the MATLAB script "TS2Orte.m", which transforms the localization table into a MATLAB matrix and saves it in the ".mat" format.
2. Once the data are in the correct format, start the preparations for the Voronoi tessellation and DNA density calculation. Calculate the conversion factor by dividing the number of base pairs in the nucleus by the relative number of localizations in the imaged section with respect to the volume (see step 6.15). Go into the LAND-Voronoi folder and in the /coreAlgorithm subfolder, open the file voronoiCluster.m and adjust the conversion factor for the absolute density calculations on line 13.
   NOTE: See Figure 5 for an example of the G₁ HeLa nucleus; a conversion factor of 265 was measured.
3. Edit the script that starts the analysis "VonoRoi.m". Adjust the file path to the Orte localization data and the output folder for the result files. In the coordinate list, specify the coordinates defining the area in which the tessellation should be done. As the Voronoi tessellation can take a long time to compute (depending on the specs of the machine used), define multiple areas to compute within the same input dataset.
4. After running the script, look for the image showing the absolute DNA densities together with other files containing graphs showing the histogram of the area distribution and a "densities.mat" file containing the densities of every single calculated Voronoi-cell in the output folder.

Results

The HeLa nucleus shown in Figure 5 was chosen from the table generated by CellProfiler4.2.1 in section 3 to have an integrated fluorescence intensity close to the first peak in the histogram shown in Figure 5 that represents nuclei in the G₁ phase. Given its small size and pronounced inner structure, it is likely to be an early G₁ nucleus that is still in the process of decondensing its chromatin after mitosis. Being in G₁ means ...

Discussion

This article outlines how to measure absolute DNA densities in mammalian cell nuclei using SMLM. In addition, we have demonstrated how to determine the cell cycle stage of cultured cells and how to use this information to estimate the amount of DNA present in a light optical ultra-thin section. Also described in detail are the preparation of adherent cells for fBALM SMLM microscopy and how to process SMLM data to generate super-resolved microscopic images of genomic DNA in cell nuclei. Finally, the protocol shows how the...

Materials

Name	Company	Catalog Number	Comments
Cell Culture
µ-Dish 35 mm, high Grid-500 Glass Bottom	Ibidi	81168
C3H 10T1/2	IMB (Niehrs Lab)
DMEM	ThermoFisher	12320032
dPBS	ThermoFisher	14190144
FBS	Life Technologies	16000-044
HeLa	Microscopy Core Facility (IMB)
HFB	IMB (Kubben Lab)
L-Glutamine	Sigma-Aldrich	G7513
Nonessential Aminoacids and vitamins for HFB
Sodium Pyruvate	S8636
Sample Prep
Catalase	Merck	2593710
Glucose	ThermoFisher	241922500
Glucose Oxidase	Merck	49180
Paraformaldehyde	Sigma-Aldrich	158127
RNase Cocktail	ThermoFisher	AM2286
SYTOX Orange	ThermoFisher	S11368
TetraSpeck Fluorescent Microspheres Sampler Kit	ThermoFisher	T7284
Triton X-100	ThermoFisher	327372500
Software
BioFormats	OpenMicroscopy.org		open source software https://www.openmicroscopy.org/bio-formats/
CellProfiler v4.2.1	CellProfiler.org		open source software https://cellprofiler.org
FiJI	nih.gov		open source software https://imagej.net/software/fiji/?Downloads
LAND	nih.gov		open source software https://github.com/Jan-NM/LAND
MatLab 2021	Math Works		commercial software - requires "Image Processing Toolbox"
R v.4..0.3	r-project.org		open source software https://www.r-project.org
ThunderSTORM v1.3			open source software https://zitmen.github.io/thunderstorm/
Microscopes:
AF 7000	Leica
Leica GSD	Leica
SMLM Microscope	Cremer Lab		custom-built by Dr. S-Y. Chen