This protocol illustrates a chemically induced protein dimerization system to create condensates on chromatin. The formation of promyelocytic leukemia (PML) nuclear body on telomeres with chemical dimerizers is demonstrated. Droplet growth, dissolution, localization and composition are monitored with live cell imaging, immunofluorescence (IF) and fluorescence in situ hybridization (FISH).
Chromatin-associated condensates are implicated in many nuclear processes, but the underlying mechanisms remain elusive. This protocol describes a chemically-induced protein dimerization system to create condensates on telomeres. The chemical dimerizer consists of two linked ligands that can each bind to a protein: Halo ligand to Halo-enzyme and trimethoprim (TMP) to E. coli dihydrofolate reductase (eDHFR), respectively. Fusion of Halo enzyme to a telomere protein anchors dimerizers to telomeres through covalent Halo ligand-enzyme binding. Binding of TMP to eDHFR recruits eDHFR-fused phase separating proteins to telomeres and induces condensate formation. Because TMP-eDHFR interaction is non-covalent, condensation can be reversed by using excess free TMP to compete with the dimerizer for eDHFR binding. An example of inducing promyelocytic leukemia (PML) nuclear body formation on telomeres and determining condensate growth, dissolution, localization and composition is shown. This method can be easily adapted to induce condensates at other genomic locations by fusing Halo to a protein that directly binds to the local chromatin or to dCas9 that is targeted to the genomic locus with a guide RNA. By offering the temporal resolution required for single cell live imaging while maintaining phase separation in a population of cells for biochemical assays, this method is suitable for probing both the formation and function of chromatin-associated condensates.
Many proteins and nucleic acids undergo liquid-liquid phase separation (LLPS) and self-assemble into biomolecular condensates to organize biochemistry in cells1,2. LLPS of chromatin-binding proteins leads to the formation of condensates that are associated with specific genomic loci and are implicated in various local chromatin functions3. For example, LLPS of HP1 protein underlies the formation of heterochromatin domains to organize the genome4,5, LLPS of transcription factors forms transcription centers to regulate transcription6, LLPS of nascent mRNAs and multi-sex combs protein generates histone locus bodies to regulate the transcription and processing of histone mRNAs7. However, despite many examples of chromatin-associated condensates being discovered, the underlying mechanisms of condensate formation, regulation and function remain poorly understood. In particular, not all chromatin-associated condensates are formed through LLPS and careful evaluations of condensate formation in live cells are still needed8,9. For example, HP1 protein in mouse is shown to have only a weak capacity to form liquid droplets in live cells and heterochromatin foci behave as collapsed polymer globules10. Therefore, tools to induce de novo condensates on chromatin in living cells are desirable, particularly those that allow the use of live imaging and biochemical assays to monitor the kinetics of condensate formation, the physical and chemical properties of the resulting condensates, and the cellular consequences of condensate formation.
This protocol reports a chemical dimerization system to induce protein condensates on chromatin11 (Figure 1A). The dimerizer consists of two linked protein-interacting ligands: trimethoprim (TMP) and Halo ligand and can dimerize proteins fused to the cognate receptors: Escherichia coli dihydrofolate reductase (eDHFR) and a bacterial alkyldehalogenase enzyme (Halo enzyme), respectively12. The interaction between Halo ligand and Halo enzyme is covalent, allowing Halo enzyme to be used as an anchor by fusing it to a chromatin-binding protein to recruit a phase-separating protein fused to eDHFR to chromatin. After the initial recruitment, increased local concentration of the phase separating protein passes the critical concentration needed for phase separation and thus nucleates a condensate at the anchor (Figure 1B). By fusing fluorescent proteins (e.g. mCherry and eGFP) to eDHFR and Halo, nucleation and growth of condensates can be visualized in real time with fluorescence microscopy. Because the interaction between eDHFR and TMP is non-covalent, excess free TMP can be added to compete with the dimerizer for eDHFR binding. This will then release the phase separation protein from the anchor and dissolve the chromatin-associated condensate.
We used this tool to induce de novo promyelocytic leukemia (PML) nuclear body formation on telomeres in telomerase-negative cancer cells that use an alternative lengthening of telomeres (ALT) pathway for telomere maintenance13,14. PML nuclear bodies are membrane-less compartments involved in many nuclear processes15,16 and are uniquely localized to ALT telomeres to form APBs, for ALT telomere-associated PML bodies17,18. Telomeres cluster within APBs, presumably to provide repair templates for homology-directed telomere DNA synthesis in ALT19. Indeed, telomere DNA synthesis has been detected in APBs and APBs play essential roles in enriching DNA repair factors on telomeres20,21. However, the mechanisms underlying APB assembly and telomere clustering within APBs were unknown. Since telomere proteins in ALT cells are uniquely modified by small ubiquitin-like modifier (SUMOs)22, many APB components contain sumoylation sites 22,23,24,25 and/or SUMO-interacting motifs (SIMs)26,27 and SUMO-SIM interactions drive phase separation28, we hypothesized that sumoylation on telomeres leads to enrichment of SUMO/SIM containing proteins and SUMO-SIM interactions between those proteins lead to phase separation. PML protein, which has three sumoylation sites and one SIM site, can be recruited to sumoylated telomeres to form APBs and coalescence of liquid APBs leads to telomeres clustering. To test this hypothesis, we used the chemical dimerization system to mimic sumoylation-induced APB formation by recruiting SIM to telomeres (Figure 2A)11. GFP is fused to Haloenzyme for visualization and to the telomere-binding protein TRF1 to anchor the dimerizer to telomeres. SIM is fused to eDHFR and mCherry. Kinetics of condensate formation and droplet fusion-induced telomere clustering are followed with live cell imaging. Phase separation is reversed by adding excess free TMP to compete with eDHFR binding. Immunofluorescence (IF) and fluorescence in situ hybridization (FISH) are used to determine condensate composition and telomeric association. Recruiting SIM enriches SUMO on telomeres and the induced condensates contain PML and therefore are APBs. Recruiting a SIM mutant that cannot interact with SUMO does not enrich SUMO on telomeres or induce phase separation, indicating that the fundamental driving force for APB condensation is SUMO-SIM interaction. Agreeing with this observation, polySUMO-polySIM polymers that fused to a TRF2 binding factor RAP1 can also induce APB formation29. Compared to the polySUMO-polySIM fusion system where phase separation occurs as long as enough proteins are produced, the chemical dimerization approach presented here induces phase separation on demand and thus offers better temporal resolution to monitor the kinetics of phase separation and telomere clustering process. In addition, this chemical dimerization system permits the recruitment of other proteins to assess their ability in inducing phase separation and telomere clustering. For example, a disordered protein recruited to telomeres can also form droplets and cluster telomeres without inducing APB formation, suggesting telomere clustering is independent of APB chemistry and only relies on APB liquid property11.
1. Production of transient cell lines
2. Dimerization on telomeres
3. Immunofluorescence (IF)
4. Fluorescence in situ hybridization (FISH)
5. Live imaging
6. Fixed imaging
7. Process time-lapse images
8. Process fixed-cell images
Representative images of telomeric localization of SUMO identified by telomere DNA FISH and SUMO protein IF are shown in Figure 2. Cells with SIM recruitment enriched SUMO1 and SUMO 2/3 on telomeres compared to cells with SIM mutant recruitment. This indicates that SIM dimerization-induced SUMO enrichment on telomeres depends on SUMO-SIM interactions.
A representative time lapse movie of TRF1 and SIM after dimerization is shown in Video 1. Snapshots at four time points are shown in Figure 3A. SIM was successfully recruited to telomeres and both SIM and TRF1 foci became larger and brighter, as predicted for liquid droplet formation and growth (Figure 1B). In addition, fusion of TRF1 foci was observed (Figure 3B), which led to telomere clustering as shown in the reduced telomere number (Figure 3E) and increased telomere intensity over time (Figure 3D). In contrast, SIM mutant was recruited to telomeres after dimerization but did not induce any droplet formation or telomere clustering, as telomere intensity did not grow and telomere number did not reduce (Figure 3C,D,E, Video 2). This indicates that phase separation and thus telomere clustering is driven by SUMO-SIM interactions.
The reversal of phase separation and telomere clustering after adding excess free TMP is shown in Video 3. Snapshots at four time points are shown in Figure 4A. Agreeing with the predicted condensate dissolution and de-clustering of telomeres, telomere number increased, and telomere intensity decreased over time (Figure 4B, C).
Representative images of APBs identified by telomere DNA FISH and PML protein IF are shown in Figure 5. Cells with SIM recruited have more APBs than cells with SIM mutant recruited, suggesting dimerization-induced condensates are indeed APBs.
The figures here show representative images. For statistical analysis with more cells, please refer to Zhang et. al., 202011.
Figure 1: Chemical dimerization to induce chromatin-associated condensates. (A) Dimerization schematic: The dimerizer consists of two linked ligands, TMP and Halo that interact with eDHFR and Haloenzyme, respectively. The phase separating protein is fused to mCherry and eDHFR, and the chromosome anchor protein is fused to Halo and GFP. (B) Before adding dimerizer (top-left nucleus), the majority of chromosome anchor proteins (green squares) are localized to the chromosomes and a small amount of anchor proteins are diffusely localized in the nucleoplasm. Phase separating proteins to be recruited (purple stars) and phase separating partners (proteins that will condense with the phase separating protein, red triangles) are diffusely localized in the nucleoplasm. After adding dimerizers (top-right nucleus), phase separating proteins are dimerized to the anchor protein on the chromosomes and in the nucleoplasm. There could be some excess phase separating proteins in the nucleoplasm, depending on the relative concentration of the anchor protein, phase separating protein and the dimerizer used. After dimerization (bottom-right nucleus), increased local concentration of the phase separating proteins at the anchor leads to phase separation and the formation of chromatin-associated condensates. Phase separating partners are enriched at the anchor because of co-condensation with the eDHFR-fused phase separating protein. Anchor proteins that are not directly bound to the chromatin can be enriched at the anchor because of dimerization to the phase separating protein. After adding excess free TMP to compete with the dimerizer for eDHFR binding (bottom-left nucleus), the phase separating protein is released from the chromatin and the condensate is dissolved. Please click here to view a larger version of this figure.
Figure 2: SUMO is enriched after recruiting SIM to telomeres with dimerizers. (A) Dimerization schematic in this experiment: SIM (or SIM mutant) is fused to mCherry and eDHFR, and TRF1 is fused to Halo and GFP. (B) A representative cell for telomere DNA FISH and SUMO1 IF after recruiting SIM. Bottom is binary layer identifying telomeres, SUMO1 and number of colocalized SUMO1 and telomere DNA foci. Scale bars: 5 µm. (C) A representative cell for telomere DNA FISH and SUMO1 IF after recruiting SIM mutant. At the bottom is the binary layer of the images used to identify the number of colocalized SUMO1 and telomere DNA foci. Scale bars: 5 µm. (D) A representative cell for telomere DNA FISH and SUMO2/3 IF after recruiting SIM. At the bottom is the binary layer identifying telomeres, SUMO2/3, and the number of colocalized SUMO2/3 and telomere DNA foci. Scale bars: 5 µm. (E) A representative cell for telomere DNA FISH and SUMO2/3 IF after recruiting SIM mutant. At the bottom is the binary layer of the images used to identify the number of colocalized SUMO2/3 and telomere DNA foci. Scale bars: 5 µm. Please click here to view a larger version of this figure.
Figure 3: Dimerization-induced phase separation drives telomere clustering. (A) Snapshots of TRF1-GFP and SIM-mCherry before and after adding 100 nM dimerizer (final concentration). At the bottom is the telomere binary layer identified from TRF1-GFP. Scale bars: 5 µm. (B) A fusion event after recruiting SIM to telomeres. Scale bars: 2 µm. Time interval: 5 min. (C) Snapshots of TRF1-GFP and SIM mutant-mCherry before and after adding 100 nM dimerizer (final concentration). At the bottom is the telomere binary layer identified from TRF1-GFP. Scale bars: 5 µm. (D) Average telomere intensity (summarize intensity over the volume in each telomere and then average over all telomeres in a cell) over time after recruiting SIM (green, for cell in Figure 3A) and SIM mutant (blue, for cell in Figure 3C). (E) Telomere number over time after recruiting SIM (green, for cell in Figure 3A) and SIM mutant (blue, for cell in Figure 3C). Please click here to view a larger version of this figure.
Figure 4: Reversal of condensation and telomere clustering. (A) Snapshots of TRF1-GFP and SIM-mCherry after adding 100 µM TMP (final concentration) to cells with condensates formed for 3 h. At the bottom is the telomere binary layer identified from TRF1-GFP. Scale bars: 5 µm. (B) Average telomere intensity (summarize intensity over the volume in each telomere and then average over all telomeres in a cell) over time for cell in Figure 4A. (C) Telomere number over time for cell in Figure 4A. Please click here to view a larger version of this figure.
Figure 5: Dimerization-induced condensates are APBs. (A) A representative cell for telomere DNA FISH and PML IF after recruiting SIM. At the bottom is the binary layer identifying telomeres, PML bodies and the number of colocalized PML and telomere DNA foci, i.e., number of APBs. Scale bars: 5 µm. (B) A representative cell for telomere DNA FISH and PML IF after recruiting SIM mutant. At the bottom is the binary layer of the images used to identify the number of colocalized PML and telomere DNA foci, i.e., number of APBs. Scale bars: 5 µm. Please click here to view a larger version of this figure.
Video 1: Recruiting SIM with dimerizers to telomeres drives phase separation and telomere clustering. Live imaging of SIM-mCherry, TRF1-GFP, and merge channels before and after adding 100 nM dimerizer (final concentration). Scale bars: 5 µm. Time interval: 5 min. Time as shown. Please click here to download this video.
Video 2: Recruiting SIM mutant cannot drive phase separation and telomere clustering. Live imaging of SIM mutant-mCherry, TRF1-GFP and merge channels before and after adding 100 nM dimerizer (final concentration). Scale bars: 5 µm. Time interval: 5 min. Time as shown. Please click here to download this video.
Video 3: Reversal of condensation and telomere clustering. Live imaging of SIM-mCherry, TRF1-GFP and merge channels after adding 100 µM TMP (final concentration) to cells with condensates formed for 3 h. Scale bars: 5 µm. Time interval: 5 min. Time as shown. Please click here to download this video.
This protocol demonstrated the formation and dissolution of condensates on telomeres with a chemical dimerization system. Kinetics of phase separation and droplet-fusion-induced telomere clustering are monitored with live imaging. Condensate localization and composition are determined with DNA FISH and protein IF.
There are two critical steps in this protocol. The first is to determine protein and dimerizer concentration. The success in inducing local phase separation at a genomic locus relies on the increase in local concentration of the phase separating protein above the critical concentration for phase separation (Figure 1B). The global concentration of the phase separating protein needs to be high enough so that there are enough proteins to be concentrated locally. The concentration of the phase separating protein cannot be too high so that global phase separation has occurred or can be easily induced. The anchor DNA length (or size of the modified chromatin that the anchor protein binds to) and concentration of Halo-fused anchor protein determine the size of the nucleation center at maximum dimerization efficiency. The larger the anchor size the easy it is to nucleate condensates. The dimerization efficiency is affected by the amount of dimerizers relative to the amount of anchor proteins. Too few dimerizers cannot occupy all the available anchor proteins while too many dimerizers result in non-productive binding of eDHFR to the excess dimerizers rather than to the ones on the anchor protein. Dimerizer concentration, along with the anchor DNA length and concentration of Halo-fused anchor protein, can be used to determine the critical concentration required for nucleating local phase separation. A systematic approach to vary those parameters (anchor DNA length, anchor protein concentration, phase separation protein concentration and dimerizer concentration) can be used to map a multi-dimensional phase diagram. However, if the interest is not in mapping phase diagram but forming chromatin associated-condensates like demonstrated here, it is very easy to simply pick cells with bright Halo-GFP signal (larger anchor size) and cells with a wide range of brightness for mCherry-eDHFR (various phase separating protein concentration) to image with the dimerizer concentration for maximum dimerization determined in Protocol 2.3. The second critical step is to avoid photobleaching in live imaging. Different from global phase separation where droplets (bright mCherry foci labeling the phase separating protein) will emerge after phase separation, local condensation at genomic locations cannot be easily spotted by judging the presence of mCherry foci. This is because recruitment of the protein alone, without phase separation, to genomic loci will result in formation of mCherry local foci. Phase separation occurs after recruitment, so mCherry foci continue to become bigger and brighter after initial recruitment. The phase separation-induced enrichment can occur in GFP channel (the anchor protein) as well, due to the dimerization of the anchor protein to the phase separation protein. Therefore, change of physical properties (size and intensity) of the foci over time rather than the presence of foci should be used to judge phase separation. While it might be difficult to differentiate dimerization or phase separation-induced enrichment of mCherry (prey protein) foci, enrichment of GFP (anchor protein) foci only occurs if there is phase separation (Figure 3D). Therefore, the enrichment of anchor protein can be used to easily judge phase separation. Photobleaching resulted from high laser power or long exposure time during imaging makes it more difficult to judge phase separation from live imaging and therefore should be avoided as much as possible by adjusting imaging conditions. Note that the increases in foci intensity and size over time are characteristics of LLPS but cannot be used as the sole evidence for LLPS. In the case presented here, droplet fusion was used as evidence for the formation of liquid droplets, which may not occur for smaller number of anchors or fewer mobile anchors. Without droplet fusion, other methods such as diffusion of condensate components and sensitivity to small molecule perturbation can be used to further confirm condensate formation8,9,11.
Though this chemical dimerization system renders temporal resolution required for monitoring phase separation in live cells, it lacks spatial resolution at the cellular and subcellular level. Thanks to the modular design of the dimerizers, it is possible to make light-sensitive dimerizers by attaching a photocage to TMP, making the linker photosensitive or both12,32,35. By simply switching dimerizers within the same engineered cell background for different applications, high spatial and temporal control of the dimerization, reversal of dimerization, or both with light can be achieved. We envision with those light-sensitive dimerizers, it will be able to control phase separation with high spatial and temporal precision. Compared to the available optogenetic tools to control phase separation through light sensitive proteins36,37, a disadvantage of the chemical dimerization system is that it can only reverse phase separation once. However, this system can maintain sustained recruitment and thus phase separation without light, which makes it more suitable for long term live imaging applications such as to follow droplet growth or cellular consequences of phase separation. In addition, the ability to treat a population of cells without light makes it convenient for biochemical assays such as those needed to determine condensate composition or changes in genome organization.
This method can be easily adapted to induce condensates at other locations on the genome. One can simply identify a protein that binds to the genomic location of interest and fuse it to Halo to use it as an anchor (Figure 1B). Alternatively, one can combine this with CRISPR and fuse dCas9 to Halo and use guide RNAs to anchor Halo to the genomic loci of interest38. In addition, one can anchor Halo to an ectopic DNA array (e.g. LacO) integrated into the genome by fusing Halo to the targeting protein (e.g. LacI). One can then use a bottom-up approach to assess the ability of a protein to phase separate locally on chromatin, how its phase separation ability is affected by protein truncations, mutations or post-translational modifications, or how the condensate affects local functions such as chromatin modification, replication or transcription. To summarize, this chemical dimerization system can be used to induce a wide range of condensates on various chromatin locations and is particularly suitable for investigating how the material properties and chemical composition of chromatin-associated condensates contribute to chromatin functions by combining long-term live imaging with biochemical assays.
The authors have nothing to disclose.
This work was supported by US National Institutes of Health (1K22CA23763201 to H.Z., GM118510 to D.M.C.) and Charles E. Kaufman foundation to H.Z. The authors would like to thank Jason Tones for proofreading the manuscript.
0.25% Trypsin, 0.1% EDTA in HBSS w/o Calcium, Magnesium and Sodium Bicarbonate | Corning | MT25053CI | |
16% Formaldehyde (w/v), Methanol-free | Thermo Scientific | 28906 | Prepare 1% in 1x PBS |
6 Well Culture Plate | VWR | 10861-554 | |
Aluminum Foil | Fisher Scientific | 01-213-101 | |
Anti-mCherry antibody | Abcam | Ab183628 | |
Anti-PML antibody | Santa Cruz | sc966 | |
Anti-SUMO1 antibody | Abcam | Ab32058 | |
Anti-SUMO2/3 antibody | Cytoskeleton | Asm23 | |
Blocking Reagent | Roche | 11096176001 | |
Bovine Serum Albumin (BSA) | Fisher Scientific | BP9706100 | |
BTX Tube micro 1.5ML | VWR | 89511-258 | |
Circle Cover Slips | Thermo Scientific | 3350 | |
Confocal microscope | Nikon | MQS31000 | |
DAPI | Fisher Scientific | D1306 | |
Dimethyl Sulphoxide | Sigma-Aldrich | 472301 | |
DMEM with L-Glutamine, 4.5g/L Glucose and Sodium Pyruvate | Corning | MT10017CV | |
EMCCD Camera | iXon Life | 897 | |
Ethanol | Fisher Scientific | 4355221 | |
Fetal Bovine Serum, Qualified, USDA-approved Regions | Gibco | A4766801 | |
Formamide, Deionized | MilliporeSigma | 46-101-00ML | |
Goat anti-Mouse IgG (H+L), Recombinant Secondary Antibody, Alexa Fluor 647 | Invitrogen | A28181 | |
Goat anti-Rabbit IgG (H+L), Recombinant Secondary Antibody, Alexa Fluor 647 | Invitrogen | A32733 | |
High Precision Straight Tapered Ultra Fine Point Tweezers/Forceps | Fisher Scientific | 12-000-122 | |
Laser merge module | Nikon | NIIMHF47180 | |
Leibovitz's L-15 Medium | Gibco | 21083027 | |
Lipofectamine 2000 Transfection Reagent | Invitrogen | 11668027 | |
Figure plotting software, MATLAB | The MathWorks | ||
Microscope Slide Box | Fisher Scientific | 34487 | |
Nail Polish | Fisher Scientific | 50-949-071 | |
Imaging software, NIS-Elements | Nikon | ||
Opti-MEM Reduced Serum Media | Gibco | 51985091 | |
Parafilm | Bemis | 13-374-12 | |
PBS 10x, pH 7.4 | Fisher Scientific | 70-011-044 | |
Penicillin-Streptomycin Solution,100X | Gibco | 15140122 | |
Piezo Z-Drive | Physik Instrumente (PI) | 91985 | |
Pipet Tips | VWR | 10017 | |
Plain and Frosted Clipped Corner Microscope Slides | Fisher Scientific | 22-037-246 | |
Poly-D-Lysine solution | Sigma-Aldrich | A-003-E | |
Sodium Azide | Fisher Scientific | BP922I-500 | |
Spinning disk | Yokogawa | CSU-X1 | |
Square Cover Slips | Thermo Scientific | 3305 | |
TBS 10x solution | Fisher Scientific | BP2471500 | |
TelC-Alexa488 | PNA Bio | F1004 | |
TMP | Synthesized by Chenoweth lab | Available upon request | |
TNH | Synthesized by Chenoweth lab | Available upon request | |
Tris Solution | Fisher Scientific | 92-901-00ML | |
Triton X-100 10% Solution | MilliporeSigma | 64-846-350ML | Prepare 0.5% in 1x PBS |
U2Os cell line | From E.V. Makayev lab (Nanyang Technological University, Singapore) | HTB-96 | |
VECTASHIELD Antifade Mounting Medium | Vector Laboratories | NC9524612 |