To probe the endogenous TERRA, we utilized RNA-guided catalytically inactive Cas13b system (Figure 1A), which was reported to detect RNA in live cells (Yang et al., 2019). Given the variability in the TERRA sequence with tandem (UUAGGG)n repeats, there is no sequence specificity for guide RNA recognition. Therefore, we designed three guide RNAs with different lengths ranging from 22 to 30 nucleotides (nt) (Figure 1B). With the addition of guide RNA, EGFP-fused dPspCas13b indeed formed visible foci in the nucleoplasm in addition to obvious nucleolar signals in live cells (Figure 1B). Significantly, the length of guide RNA determines the RNA-labeling efficiency. We found that the shortest guide RNA with 22 nt induced more visible foci than the longer ones (Figure 1C). To verify that those are TERRA foci, we employed RNA fluorescence in situ hybridization (FISH) with TERRA FISH probes in fixed cells (Figure 1D). As expected, the dPspCas13b foci were all labeled by the TERRA FISH probe. In addition, the TERRA signal was decreased after treatment with Ribonuclease, indicating the RNA-binding specificity of the TERRA FISH probe. This suggests that the CRISPR-dCas13 system labels TERRA properly.

Although EGFP-fused dCas13b detects TERRA, the signal is weak compared to the non-specific signals in nucleoli, restricting its utility for dynamic imaging of TERRA in live cells. To improve TERRA imaging efficiency, we combined the SunTag technology with the CRISPR-dcas13 system to amplify the TERRA signal. The synthetic SunTag scaffold, including five tandem GCN4, was fused to dCas13b to recruit up to five GFP copies via scFV (Figure 2A). Additionally, we replaced EGFP with sfGFP, a form of superfolder GFP, to increase its solubility (Pédelacq et al., 2006). As visualized in Figure 2B with FISH, those visible foci indicated by dCas13b-SunTag were all TERRA positive as well (Figure 2B). Notably, the non-specific fluorescent signal in nucleoli was largely decreased with the dCas13-SunTag system. Significantly, in contrast to the original dCas13b strategy, the combination with SunTag largely increased the TERRA detection rate from around 5% to 38% (Figure 2C). Also, TERRA foci detected by dCas13b-SunTag-sfGFP were bigger and brighter than dCas13b-EGFP dots, owing to five GFP copies binding to SunTag via scFV. To demonstrate that the dCas13b signal was indeed from TERRA RNA, we analyzed the percentage of dCas13 proteins detected by the TERRA FISH probe. The data shows that 77% of dCas13-GFP and 89% of dCas13-SunTag-sfGFP are TERRA positive (Figure 2D). The results indicate that the SunTag technology with sfGFP improves TERRA labeling efficiency.

The capacity of dCas13b-SunTag-sfGFP to detect TERRA foci in live cells prompted us to monitor TERRA foci movement with single-particle tracking. The movement of many structures in the human cell nucleus, such as nanoparticles, PML nuclear bodies, and telomeres, are shown to depend on timescale, owning to particle confinement within the chromatin cages at small timescales and particle hopping between cages at large timescales (Tseng et al., 2004; Jegou et al., 2009). To determine whether TERRA foci movement is timescale dependent, we generated TERRA foci trajectories at two timescales: 0–1 s and 10–100 s (Figure 3A). Furthermore, since a subset of TERRA foci co-localizes with telomeres (Biffi et al., 2012; Yang et al., 2019; Mei et al., 2021), we aimed to determine whether telomeric localization affects TERRA foci dynamics by imaging telomeres through mCherry fused to the Shelterin component TRF1 while tracking TERRA foci (Figure 3B).

Overall, TERRA foci move heterogeneously, with some diffuse within a small area while others explore a region several times larger (Figure 3A). To quantify whether and how the heterogeneity of TERRA foci movement depends on the timescale and telomeric localization, we generated the mean-squared displacement (MSD) curves over the lag time τ for telomeric and non-telomeric TERRA foci at the two timescales (Figure 3C; Supplementary Figures S1A,B). To assess the mode of motion, i.e., whether it deviates from normal diffusion as seen for other molecules/structures in the nucleus (Woringer and Darzacq, 2018), we fitted the MSD curves of TERRA foci to the equation for anomalous diffusion, MSD = Kτ^α, where K is the generalized diffusion coefficient, τ is the lag time, and α is the anomalous exponent (α = 1 for normal diffusion, α < 1 for anomalous diffusion, and α > 1 for active diffusion). Average α for telomeric and non-telomeric TERRA foci at the 0–1 s and 10–100 s timescales are 0.34, 0.52, 0.64, and 0.7, respectively, indicative of overall anomalous diffusion (Figure 3D; Supplementary Table S1). Lower α for telomeric and non-telomeric TERRA foci at the 0–1 s timescale than their counterparts at the 10–100 s timescale agrees with reported caging of particles in the chromatin network at small timescales (Tseng et al., 2004). Lower α of telomeric TERRA foci than non-telomeric foci at the 0–1 s timescale, but not at the 10–100 s timescale, suggests that the local telomere environment confines TERRA foci more than other regions at small timescales, but attachment to telomeres does not alter TERRA foci hopping between different chromatin cages at large timescales.

To compare the difference in mobilities of TERRA foci with different anomalous exponents, we calculated the mean MSD at each τ and converted it to a time-dependent diffusion coefficient, D, following MSD = 4Dτ (Figures 3E,F). Interestingly, TERRA foci diffusion coefficients decay quickly in the timescale of 0–1 s but seem to plateau at 10–100 s, similar to the behavior of other structures in the nucleus (Tseng et al., 2004). In addition, at the 0–1 s timescale, diffusion coefficients of telomeric TERRA foci are smaller than non-telomeric TERRA foci (mean 0.013 vs. 0.037 μm²/s), which suggests that the local telomere environment not only makes TERRA foci move more anomalously but also slower. However, at the 10–100 s timescale, no significant difference in diffusion coefficients for telomeric and non-telomeric foci is observed (mean 0.0033 vs. 0.0029 μm²/s; Supplementary Figure S1; Supplementary Table S1), indicating attachment to telomeres affects neither the mode nor magnitude of TERRA movement. Taken together, the dCas13b-SunTag-sfGFP system enabled us to monitor TERRA foci movement with single-particle tracking and revealed its dependence on timescale and telomeric localization.

In addition to monitoring TERRA localization and motion, we exploit our dCas13b-SunTag tool to control TERRA localization by combining it with a small molecule-mediated protein dimerization system we developed (Ballister et al., 2014; Zhang et al., 2017, Zhang et al., 2020). This system is based on two linked ligands, TMP (Trimethylolpropane), and Halo-ligand, that can interact with the protein eDHFR and Halo-tag, respectively (Figure 4A). We fused Halo-tag to the telomere binding protein TRF1 to localize the dimerizers to telomeres and eDHFR to SunTag. Adding the chemical dimerizer, TMP-NVOC (6-nitroveratryl oxycarbonyl)-Halo, would recruit dCas13b protein, and thus TERRA, to telomeres (Figure 4A). By using an antibody against telomere binding protein TRF2 to label telomeres and TERRA FISH to confirm TERRA localization, we observed a basal level of TERRA localization on telomere without dimerizers, consistent with other studies (Figure 4B) (Azzalin et al., 2007; Chu et al., 2017). After adding dimerizers, the colocalization of TERRA on telomeres increased two-fold (Figure 4C), whose effect on telomere function awaits to be determined. Meanwhile, the colocalization of dCas13b proteins on telomere is up to 80% from 6% with the dimerizer (Figure 4D), suggesting the dimerization efficiency is high. Overall, the dCas13b-SunTag is compatible with the protein dimerization system for spatiotemporal enrichment of TERRA on telomeres for functional studies.

The plasmids for expression of dPspCas13b, guide RNA and GCN4-scFv-sfGFP (SunTag) were all purchased from addgene (#132397, #103854, #60906). To construct the dCas13b-SunTag-sfGFP, GCN4-scFv-sfGFP were amplified from addgene plasmid #60906 and introduced into plasmid #132397 through in-fusion cloning (Takara Bio). All other plasmids in this study are derived from a plasmid that contains a CAG promoter for constitutive expression, obtained from E. V. Makeyev (Khandelia et al., 2011).

All experiments were performed with U2OS acceptor cells, originally obtained from E.V. Makayev (Khandelia et al., 2011). Cells were cultured in growth medium (Dulbecco’s Modified Eagle’s medium with 10% FBS and 1% penicillin–streptomycin) at 37°C in a humidified atmosphere with 5% CO₂. The constructs used in this manuscript, including mCherry-TRF1, dCas13b-EGFP, dCas13b-SunTag-sfGFP, and Halo-TRF1, were transiently expressed by transfection with Lipofectamine 3,000 (Invitrogen) 24 h prior to imaging, following the manufacturer’s protocol.

TERRA FISH assay was performed as previously described (Flynn et al., 2011). Briefly, cells were washed twice with cold PBS and treated with cytobuffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 10 mM PIPES pH 7, 0.1% Triton X-100, and 10 mM vanadyl ribonucleoside complex) for 7 min on ice. Cells were rinsed with cytobuffer (100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 10 mM PIPES pH 7, 0.1% Triton X-100, and 10 mM vanadyl ribonucleoside complex) for 7 min at 4°C, fixed in 4% formaldehyde for 10 min at room temperature, followed by permeabilization in 0.5% Triton X-100 for 10 min. For the control group, cells were then digested with 200 mg/ml RNaseA in PBS for 30 min at 37°C and were washed twice with PBS for 5 min. After incubation with blocking solution containing 1% BSA for 1 h, cells were then dehydrated in a series of ethanol washes 70, 85, and 100% for 5 min each at room temperature, and the coverslips were dried at room temperature. 20 nM Telo Miniprobe SCy3 short probe in hybridization buffer (50% formamide, 2x SSC, 2 mg/ml BSA, 10% dextran sulfate) was added to coverslips and then placed in a humidified chamber at 39°C overnight. The following day, the coverslips were washed in 2x SSC with 50% formamide three times at 39°C for 5 min each, three times in 2x SSC at 39°C for 5 min each, and finally one time in 2x SSC at room temperature for 10 min. The coverslips were mounted on glass microscope slides with Vectashield mounting medium containing DAPI and analyzed with microscopy.

For live imaging, cells were seeded on 22 × 22 mm glass coverslips (no. 1.5; Fisher Scientific) coated with poly-D-lysine (Sigma-Aldrich) in single wells of a 6-well plate. When ready for imaging, coverslips were mounted in magnetic chambers (Chamlide CM-S22-1, LCI) with cells maintained in L-15 medium without phenol red (Invitrogen) supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C on a heated stage in an environmental chamber (TOKAI HIT Co., Ltd.). Images were acquired with a microscope (ECLIPSE Ti2) with a 100 × 1.4 NA objective, a 16 XY Piezo-Z stage (Nikon instruments Inc.), a spinning disk (Yokogawa), an electron multiplier charge-coupled device camera (IXON-L-897), and a laser merge module equipped with 488 nm, 561 nm, 594 nm, and 630 nm lasers controlled by NIS-Elements Advanced Research. For fixed cells, images were taken with 0.5 μm spacing between Z slices, for a total of 8 μm. For single-particle tracking, GFP images were taken at two time intervals, 30 ms and 5 s, to generate tracks for 0–1 s timescale and 10–100 s timescale. For 5 s time interval, both GFP and mCherry images were taken during the time course to accurately identify TERRA telomeric localization. For 30 ms interval, taking a mCherry image before and after tracking the GFP channel was sufficient to determine co-localization between TERRA foci and telomeres.

Dimerization on telomeres was performed as previously described (Zhao et al., 2021). Briefly, dimerizers were added to growth medium to a final working concentration of 100 nM in a dark room with a dim red-light lamp. Cells were incubated with the dimerizers-containing medium for 2 h, followed by immunofluorescence (IF) or fluorescence in situ hybridization (FISH).

Images were processed and analyzed using NIS Elements Software. Maximum projections were created from z stacks, and thresholds were applied to the resulting 2D images to segment and identify TERRA foci as binaries. For colocalization quantification of two fluorescent labels, fixed images were analyzed by NIS-Elements AR to determine if the different labels were located in the same area of the cell.

NIS Elements tracking module was used to generate tracks for the TERRA binaries and Mean Square Displacements (MSDs) were calculated as: MSD ( τ ) = < ( x t + τ − x t ) 2 + ( y t + τ − y t ) 2 >

Where x_t and y_t are the foci coordinates at time t while x_{t + τ} and y_{t + τ} are the foci coordinates after a lag time of τ. MATLAB was used for plotting and curve-fitting MSD vs. Lag Time log-log plots to evaluate the anomalous exponent α from the anomalous diffusion model:

MSD = Kτ ^α

Where K is the generalized diffusion coefficients, τ is the lag time, and α is the anomalous exponent. The time-dependent diffusion coefficient, D, is calculated from mean MSD at each τ from MSD = 4Dτ.

All p values were generated with a two-sample t-test in MATLAB with function ttest2.