Inducing and monitoring liquid-liquid phase separation by type II topoisomerases

Abstract

Liquid-liquid phase separation (LLPS) is an increasingly studied property of biological macromolecules that can give rise to membrane-less compartments (“condensates”) in cells. Analogous to droplets that form after the mixing of oil and water, certain biological macromolecules are also capable of de-mixing from the bulk solution, forming condensates whose contents are partitioned from the rest of the cellular environment. Phase transitions in biology are frequently driven by weak, multivalent protein-protein or protein-nucleic acid interactions, as well as by intrinsically disordered regions (IDR’s) of proteins. Fluorescence microscopy is one method commonly used for studying the formation of biological condensates both in vitro and in cellulo. Depending on the system, imaging may be performed with protein alone or with the addition of a nucleic acid ligand or substrate to induce phase separation. Here we describe a protocol for how to produce dye-labeled samples suitable for studies of in vitro phase separation by a human type II topoisomerase using fluorescence microscopy. The approach does not rely on tags or fusion proteins (which can alter LLPS properties) and is generalizable to any protein with a free, N-terminal primary amine.

1. Introduction

Eukaryotic type II topoisomerases (Top2s) are ubiquitous nuclear enzymes that use an ATP-dependent strand passage mechanism to resolve different topological states of DNA arising from the action of natural biochemical processes such as transcription and replication [1]. All eukaryotes encode at least one Top2, with vertebrates expressing two different isoforms, TOP2A and TOP2B [2]. Studies with human Top2s have shown that functional differences between its isoforms arise in part from an intrinsically disordered C-terminal domain (CTD), which is attached to a conserved catalytic core [3].

Liquid-liquid phase separation (LLPS) is a physical process that can create membraneless compartments in cells [4]. Top2s have been reported to undergo LLPS at cellular loci where the enzyme is enriched, such as condensed chromosomal axes [5], and at different stages during the cell cycle [6]. Biochemical studies have shown that budding yeast Top2, as well as both human TOP2s, can also form condensates in vitro; the addition of DNA has been shown to stimulate phase separation by Top2 and fluidize these condensates [7]. LLPS by Top2s in general has been proposed to be driven by different kinds of multivalent interactions, in particular between the intrinsically disordered regions of different Top2 molecules, or between the protein and DNA [8].

Fluorescence microscopy is a method commonly used for studying biological condensate formation both in vitro and in cells. In this protocol, we use fluorescence imaging together with purified, Cy3-labeled TOP2A to study LLPS and characterize properties such as condensate size and the dependence of LLPS on protein, DNA, and salt-ion concentration. Visualized condensates can also be probed by fluorescence recovery after photobleaching (FRAP) to study dynamic behavior and provide insights into the exchange of material between droplets and the condensate environment. Although this protocol focuses on human TOP2A, the approaches outlined herein have also been extended for studying human TOP2B and Top2 from Saccharomyces cerevisiae [7].

2. Materials:

All buffers are passed through a 0.22 μm Millipore Steriflip sterile vacuum filter unit and stored at 4 °C.

2.1. Buffers and materials for labeling of human TOP2 protein

1. Topo storage buffer (50 mL): 500 mM KCl, 50 mM HEPES pH 8.0, 10% glycerol, 0.5 mM TCEP

2. Cyanine3 NHS Ester fluorescent dye (Lumiprobe, cat# 11020, see Note 1

3. Zeba Spin Desalting Columns – 7K MWCO (Thermo-Fisher Scientific, cat# 89883)

4. Amicon Ultra centrifugal filters – 30,000 MWCO (Millipore Sigma, REF# UFC503008)

5. 0.22 μm Millipore Sigma Steriflip sterile disposable vacuum filter units, 50 mL (Fisher Scientific, cat# SCGP00525).

6. 1.5mL centrifugal collection tubes

2.2. Buffers and materials for LLPS studies:

1. Buffer exchange solution: 600mM potassium acetate (KOAc), 50mM HEPES pH 7.5, 10% glycerol (see Note 2).

2. Dilution buffer: 50mM HEPES pH 7.5, 10% glycerol

3. Zeba Spin Desalting Columns – 7K MWCO (Thermo-Fisher Scientific, cat# 89883)

4. Amicon Ultra centrifugal filters – 30,000 MWCO (Millipore Sigma, REF# UFC503008)

5. 0.22 μm Millipore Sigma Steriflip sterile disposable vacuum filter units, 50 mL (Fisher Scientific, cat# SCGP00525).

6. Nunc Lab-Tek II Chambered Coverglass (Thermo-Fisher Scientific, cat#155409)

7. 1.5mL centrifugal collection tubes

3. Methods:

Prior to use, thaw a dye aliquot at room temperature and dilute using topo storage buffer to the same concentration as the protein that will be used for labeling before adding to protein (see Note 3).

3.1. Labeling of Protein with Cyanine3 NHS-Ester dye

1. For expression and purification of human TOP2A, please refer to the protocol outlined in Chapter 7 of this volume [Clark and Berger]. After running the final size-exclusion column and concentrating, measure the protein concentration using spectrophotometric measurement.

2. Transfer 1/3 of the total protein volume to a fresh Eppendorf microfuge tube for fluorescent labeling. Keep this sample and a portion of the topo storage buffer on ice or at 4 °C.

3. With the remaining protein sample, dilute with cold topo storage buffer to a final concentration of 10 μM and apportion into 40 μL aliquots. To retain specific activity, we have found it useful to dilute topo II in successive 2-fold (1:1) steps until the desired final concentration is reached. Flash freeze the unlabeled protein aliquots in liquid nitrogen and store at −80 °C.

4. Take the sample that was set aside for labeling and mix at a 1:1 molar ratio with Lumiprobe Cyanine3 NHS ester dye aliquot. Leave the sample covered (in the dark) on ice at 4 °C overnight.

5. The following day, take a Zeba Spin desalting column and remove the bottom closure. Loosen the top and place the column into a fresh 1.5 mL microfuge collection tube. Centrifuge at 1,500 x g and 4 °C for 1 min to remove the storage buffer. Discard the flowthrough from the microfuge tube and place the Zeba spin column back into the collection tube. Mark where the resin bed slants upward on the side of tube. Place the column in the microfuge tube in the same orientation, facing the mark outward for all subsequent steps. Improper orientation of the column may decrease desalting efficiency.

6. Wash the Zeba spin column by adding 300 μL of the topo storage buffer and spin at 1,500 x g for 1.5 min at 4 °C. Repeat the wash three times and discard the microfuge collection tube.

7. Place the column into a fresh microfuge tube for sample collection. Add the Cy3-labeled protein to the middle of the Zeba column, being careful not to disturb the resin bed. Spin the column at 1,500 x g for 2 min and collect flowthrough. Keep flowthrough on ice and discard the Zeba column.

8. Using a NanoDrop microvolume spectrophotometer (or equivalent instrument), take three A₂₈₀ and A₅₅₀ absorbance measurements of the flowthrough to calculate protein and dye concentrations, respectively (the wavelength used for measuring fluorophore concentration should be changed accordingly if a dye other than Cy3 is being used). If using a Nanodrop, dye absorbance and concentration values can be obtained from the “protein and labels” application for some pre-defined dyes.

9. Use the average value from the three independent measurements to calculate protein concentration in units of μM. For TOP2A, this can be done by dividing the averaged A₂₈₀ absorbance measurement by its calculated extinction coefficient (ε = [163070.00 M⁻¹ cm⁻¹]):

   $$C_l=(\frac{A_{280}}{\mathrm{ℇ}\times {10}^{-6}})$$

   Where C_l = concentration of labeled protein (μM).

   This value reports on the concentration of TOP2A monomers. To determine the concentration of TOP2A dimers, divide the calculated monomer concentration value by 2.

10. Calculate the labeling efficiency (L_e) by dividing the concentration of the Cy3 dye in the labeled sample solution by the monomer protein concentration:

    $$L_e=\frac{C_d}{C_l}$$

    where C_d = concentration of dye (μM) determined in step in 3.1.8 using spectrophotometric measurement.

11. Dilute the Cy3-labeled protein using the topo storage buffer in two-fold dilution steps to a final protein concentration of 10 μM dimer (the concentration of the diluted Cy3-labeled protein should be the same as the unlabeled protein that is prepared in step 2.1.3). Apportion the labeled protein into 10 μL aliquots, flash freeze in liquid nitrogen, and store at −80 °C.

3.2. Preparing samples for LLPS studies using fluorescence microscopy

1. Thaw unlabeled and labeled aliquots on ice. Mix the two together to obtain a labeled protein concentration between 5-10% of the total protein concentration (we generally aim for 7.5%). To calculate the volume of labeled protein to be added to 40 μL of unlabeled protein, it is important to take into account the labeling efficiency (L_e) of the dye-conjugated sample for this step. The appropriate volume (x) can be calculated as follows:

   $${x\mathrm{*}L}_e=\left(0.075\mathrm{*}\left(x+40\mu L\right)\right)$$
   As an example, if a dye-conjugated sample has a 95% labeling efficiency, 3.4 μL would be added to 40 μL of unlabeled protein to reach a final concentration of 7.5% labeled protein in the sample:

   $$0.95x=\left(0.075\mathrm{*}\left(x+40\mu L\right)\right)$$

   $$0.95x-0.075x=3\mu L$$

   $$0.875x=3\mu L$$

   $$x=3.4\mu L$$

2. Buffer exchange the sample at 4 °C. Place a Zeba spin column into a fresh 1.5 mL microfuge tube and centrifuge at 1,500 x g for 1.5 min to remove the storage buffer. Wash the Zeba column with four subsequent washes of the buffer exchange solution as detailed in steps 3.1.5.

3. Place the column into a fresh microfuge tube for sample collection. Add the protein to the middle of the Zeba column, being careful not to disturb the resin bed. Collect the flowthrough (which contains the protein sample) and discard the spin column.

4. Place a 30,000 MW Amicon centrifugal concentrator inside a fresh microfuge collection tube. Add 500 μL of the buffer exchange solution and spin at 10,600 x g for 2 min to wash membrane with buffer.

5. Discard the flowthrough and flip the concentrator upside down inside the same microfuge collection tube just used. Spin for 2 min at 10,600 x g to collect any remaining buffer and discard flowthrough.

6. Flip the concentrator back to its upright position and place inside the same collection tube. Add labeled TOP2A sample to the concentrator. Spin at 10,600 x g until approximately half of the initial volume remains (in our hands, typically 10-15 min).

7. Discard the microfuge tube with the flowthrough and place the concentrator upside down into a fresh microfuge tube to collect sample. Spin down at 10,600 x g for 3 min. Discard the centrifugal unit after use.

8. Take three A₂₈₀ absorbance measurements using a NanoDrop microvolume spectrophotometer (or equivalent) and average them together. Solve for protein concentration (in μM units), using the value of dye concentration (in μM units) obtained from the spectrophotometer measurements as detailed in steps 3.1.7 – 3.1.9. If the resultant TOP2A concentration is less than four times the concentration of the highest protein concentration being tested in the LLPS assay, concentrate further by repeating steps 3.2.4 – 3.2.7.

3.3. Inducing liquid-liquid phase separation

In the example detailed below, the final reaction volumes (50 μL) result in a series of TOP2A concentrations differing by 2-fold increments. This scheme can be readily adjusted to instead screen a range of salt or ligand concentrations (such as DNA). All samples are incubated for 30 min at room temperature prior to imaging to allow time for condensates to appear and mature (see Note 4). Since condensate number, size, and fluidity can be influenced by incubation time, it is important to image each sample in a titration series for a defined number of minutes that is held roughly constant for each (unless time is one of the parameters being varied).

1. Dilute the TOP2A sample that will be used for microscopy (i.e., 10 μM dimer in which ~7.5 % of the protein is labeled) to a concentration that is 4x higher than highest concentration used in the planned experiments (e.g., to 2 μM if the highest concentration being tested is 500 nM). Use the buffer-exchange solution for this step and dilute serially in 2-fold increments until the desired concentration is reached. The higher salt in the buffer-exchange solution will impede LLPS prior to setting up titration experiments.

2. Approximately 35-40 min prior to imaging, dilute the 4x labeled TOP2A stock to a 2x stock by mixing 1:1 with protein dilution buffer. This will dilute the potassium acetate salt in the solution to 300 mM. To ensure consistent incubation times of 30 min prior to imaging each condition in the titration series, stagger each 2x stock dilution ~ 5-10 min apart.

3. Prior to imaging, dilute the TOP2A samples down to the desired 1x concentration by mixing 1:1 with protein dilution buffer or substrate for 30 min. Stagger each dilution ~5-10 min apart from one another so that the total incubation time for each won’t vary due to the time it takes to acquire images of prior conditions.

3.4. Image Acquisition and analysis

All images taken by the authors used a Zeiss AxioObserver equipped with a LSM800 Confocal module, three GaAsp detectors, and a 40x objective lens (see Note 5). Micrographs of labeled protein samples should be taken at an appropriate excitation laser wavelength (e.g., 561 nm for Cy3) and saved in a suitable format for image analysis. Files can then be analyzed in ImageJ (FIJI) [9], to determine characteristics of condensate bodies (‘puncta’) such as area, number, and integrated density; an example macro script (‘ROI puncta measurements’) for this is provided in the supplemental materials site. Results from the script are saved as .csv file and can be further analyzed using a statistical package such as GraphPad Prism or R Studio [10].

1. Pipette mixed and incubated TOP2A samples into a Nunc Lab-Tek II 8-well chambered coverglass and let sit for 1 min prior to imaging to allow droplets to settle on the surface of the slide.

2. Locate the surface of the cover slip, then move the Z-axis up to a point right above the cover slip for image acquisition (this is done to keep condensates in focus and minimize background fluorescence). Acquire four images of each condition (e.g., protein concentration) at different, non-overlapping regions for each well before moving on to the next condition.

3. Next quantify condensate properties of interest using a suitable image analysis approach. This can be performed in whatever manner is desired by the experimentalist; an example macro script (‘ROI puncta measurements’) for this is included in the supplemental materials site. Working through ImageJ, the script will:

   1. Duplicate each image to preserve the original while performing particle analysis on the duplicate.
   2. Subject the duplicated image to background subtraction (using a rolling ball algorithm) and thresholding (by the MaxEntropy algorithm)
   3. Convert the duplicate to a binary image based on the thresholding to distinguish puncta from the background (Fig. 1A).
   4. Subject the binary image to watershed separation to segment overlapping puncta and count them as two individual regions of interest (ROIs).
   5. Overlay particles from the watershed image over the binary image to “map” Regions of Interests (ROIs) that correspond to individual condensates and add them to the ImageJ ROI manager.
   6. Set parameters such that any particles found to touch the border of the image or that contain fewer than 10 pixels are excluded from the following analysis.
   7. Analyze ROIs with the “Measure” command to determine the area and total fluorescence intensity contained in each condensate.
   8. Save results as .csv file (after which ROIs will be deleted to reset the environment for the next image analysis).

4. For data analysis, pool together measurements from the three best images taken for each sample condition. Condensates < 0.5 μm² in area should be filtered out before calculating the number of condensates and geometric mean area of the puncta in the image set. The resultant data can be plotted using the desired format, such as a bee-swarm plot as implemented in the ggplot2 package [11] (Fig. 1B). An example R script (‘TOP2A condensate analysis and plot’) is included in the supplemental materials site.

Figure 1. Image analysis.

A) Micrographs analysis using ImageJ (FIJI) to determine puncta fluorescence characteristics. Image particles are analyzed by background subtraction and watershed algorithms to generate regions of interest (ROIs) for each puncta. ROIs are overlayed on the original image to determine fluorescence characteristics. Top to bottom: pre-processed image, background subtracted image, watershed binary image. Left – images at 43x magnification; right – zoomed-in views of the areas outlined by dotted boxes. Red circles – magnified area that highlight the segmentation of overlapping puncta by the watershed algorithm. B) Analysis of puncta characteristic using R Studio to calculate puncta geometric mean area of all observations area for each condition. DNA is a 200 bp linear oligo. Results are visualized as a bee-swarm plot.

3.5. Fluorescence Recovery After Photobleaching (FRAP) Acquisition

Condensates can be prepared for FRAP using the protocol detailed in sections 3.2 and 3.3. Prior to imaging, samples should be incubated at room temperature for a defined period of time (e.g., 30 min) to allow for condensate formation. For time-lapse acquisition of intensity recovery, it may be useful to employ a higher magnification than for base imaging (a 63x objective lens was used for the FRAP images discussed below).

1. After preparing condensate conditions of interest, pipette the sample(s) into a Nunc Lab-Tek II 8-well chambered coverglass. Locate puncta by finding the surface of the coverglass and move ~1 μm up in Z-axis to minimize background scatter. Find a region where a substantial number of single, isolated condensates can be found. Note: it is important that the diameter of these condensates be larger than the focal diameter of the laser that will be used for photobleaching.

2. At the start of data acquisition, choose nine regions of interest. Seven should contain one isolated condensate each that will be used for photobleaching, one should have a single condensate that will be monitored but not photobleached (to correct for post-acquisition photobleaching inherent to the imaging process), and one should have no puncta present for background subtraction (Fig. 2A).

3. Center the laser so that each photobleached region is inside of the puncta and does not include any edges or the entirety of the puncta. Adjust the photobleaching intensity such that the starting intensity is reduced by ~70% of its initial value following photobleaching.

4. Monitor fluorescence recovery by taking time-lapse images (e.g., 1 per 3 s) for a sufficient period of time to see a signal plateau (e.g., 20 min) following the bleaching event (Fig. 2B). Save all micrographs in a suitable image format that preserves metadata for subsequent analysis.

Figure 2. FRAP and time-lapsed image analysis.

A) Left image: Cy3 TOP2A condensates prior to time-lapse imaging. Middle image: Red box highlights magnified area of whole image for visualization. Right image: representative ROIs inside of the control and photobleached condensates, and outside of any condensate (background). B) Magnified photobleached puncta from that shown in panel (A) during (T= 0 min) and after (T = 10, 20 min) a photobleaching event. C) ImageJ analysis of photobleached regions inside condensates using the multi-measure tool in the ROI manager. Results are saved as .csv file and later converted into an excel file. D) Fluorescence measurements extracted from ImageJ analysis are background subtracted and corrected for photobleaching that occurs during image acquisition using a statistical analysis package. Traces of individual photobleached condensates can be plotted singly (left graph) or together (right graph). The time axis begins at T= 0, which is defined as the beginning of the photobleaching event.

3.6. FRAP Image Analysis

1. Using the approach described in section 3.4, measure total fluorescence intensities for each of the seven photobleached regions defined in section 3.5, as well as the unbleached and control regions, as a function of time. If photobleached regions cannot be automatically detected using a tool such as ImageJ, areas of interest can be manually traced using the selection tool and added to the ROI manager.

2. From the ROI manager, select all ROI’s being analyzed, choose the “multi measure” option, and save the results as a .csv file (Fig. 2C).

3. Use a statistical analysis package to import and analyze the results from the ImageJ measurements. In the example below, R studio was employed. An example R script (‘FRAP Single Trace’) provided in the supplemental materials site will accomplish the following (see Note 6):

   1. Perform a background subtraction for the photobleached and control (unbleached) condensates by subtracting the integrated value of the background region (D_b) from that of each condensate (D_c) for each time frame:

      $$F_b(t)=D_c(t)-D_b$$

      where F_b(t) is the background corrected fluorescence intensity value of a given condensate at time t, D_c (t) is the (raw) integrated density of a given condensate at time t, and D_b = integrated fluorescence measured for the background ROI.
   2. Calculate a correction factor that accounts for laser-induced signal decay or other image quality fluctuations over the course of the experiment using the control (unbleached) puncta:

      $$Cor(t)=\frac{F_{con,max}}{F_{con}(t)}$$

      where Cor is the correction factor, F_{con, max} is the maximum value of the integrated density of the control puncta after background subtraction, and F_con is the integrated density values of the control puncta after background subtraction.
   3. Apply the correction factor to all background subtracted values obtained in step 3 for each photobleached condensate as follows:

      $$F_{b,cor}\left(t\right)=F_b\left(t\right)\mathrm{*}Cor(t)$$
   4. Normalize F_b,cor(t) for each photobleached condensate by dividing these values by the maximum value of the background subtraction (F_{b, max}) and multiplying by 100 to rescale images in a way that represents the maximum fluorescence as 100% :

      $$F_{norm}(t)=\frac{F_{b,cor}(t)}{F_{b,max}}\mathrm{*}100$$
   5. Analyze F_norm(t) to find the minimum value (F_norm,min) in every recovery time course for each bleached condensate. This value will then be subtracted from all F_norm (t) values to set all intensities above zero.

      $$F_{norm1}(t)=F_{norm}(t)-F_{norm,min}$$
   6. Since this time course now becomes a new trace, it will be normalized again to the maximum value as done in step 2.6.3d to obtain the final values.

      $$F_{norm,fin}(t)=\frac{F_{norm1}(t)}{F_{norm1,max}}$$

4. Following data acquisition and quantification, graph fluorescence recovery traces individually for each photobleached condensate, or collectively, using a package such as R Studio and ggplot [11] to assess properties such as condensate fluidity (Fig. 2D).

4. Notes

1. Cy3 dye is dissolved in 100% DMSO to a concentration of 10 mM and dispensed into 10 μL aliquots before storing in the dark at −20°C.

2. Potassium and acetate are used for LLPS studies as these are more physiological ions for intracellular proteins than sodium or chloride.

3. A high efficiency of dye conjugation to TOP2A (>90%) is desirable in the labeling reaction. If this level of labeling is not initially achieved by the protocol below, increase the dye concentration to 2x-5x of the TOP2A concentration and repeat the procedure. If the labeling efficiency is > 100%, repassage the sample through a fresh de-salting column to remove any possible excess dye, or alternatively relabel the sample for a shorter duration of time and/or with a lower dye concentration to avoid derivatizing lysines in addition to the N-terminal amine.

4. After buffer exchanging and concentrating protein sample, all preparation steps for inducing LLPS in section 3.3 are done at room temperature.

5. Images can be acquired using any commercially available fluorescence microscope.

6. The FRAP Single Trace R script follows analysis steps discussed in reference [12].