Live imaging of Alu elements with CRISPR/Cas9-based system.

(A) Schematic illustrating how genome-wide Alu elements are targeted with dCas9 constructs. (B) Fluorescence images of U2OS cells labeled with Alu elements (sgAlu) or non-target control (sgNT) sgRNAs. Scale bar, 10 μm. (C) Coefficient of variance (standard deviation normalized by mean) of fluorescence intensity in nuclear pixels from (B). Notch represents s.d., box represents quartiles (lower, Q1; center, Q2; and higher, Q3), whiskers extend to data points that lie within 1.5 IQR (interquartile range = Q3 - Q1) of the lower (Q1) and higher (Q3) quartiles. n ≥ 300 nuclei for each group. *** denotes P < 0.001 using two-sided Brunner-Munzel test with t-distribution. (D) Genomic tracks of CUT&RUN sequencing assay against dCas9 compared to Alu-element annotations (from RepeatMasker), CUT&RUN against promoter-specific epigenetic mark (H3K4me3), and A/B compartments (A compartment > 0; B compartment < 0). For dCas9 CUT&RUN tracks, data ranges were scaled to account for total read counts for each condition. sgRNAs used in dCas9 CUT&RUN: control (sgNT) or Alu-targeting (sgAlu). A/B compartments were assigned from an existing Hi-C dataset from the same cell type (Arnould et al., 2021). Genomic range shown: chr2 5-80 Mb. (E) Averaged profiles and heatmaps showing mapped CUT&RUN reads at three Alu-repeat families annotated across the hg38 genome. (F) Relationship between density of sgNT or sgAlu-targeted dCas9 CUT&RUN reads across the genome per megabase and density of Alu annotations for the same window of 1 Mb. Spearman correlation coefficient: ρ = 0.63 for sgAlu, and ρ = 0.29 for sgNT.

Optimized dCas9 cell line for targeting consensus sequence of Alu family.

Related to: Figure 1. (A) Constructs for dCas9-based Alu-imaging. (B) Clonal line expressing dCas9-SunTag system optimized for genomic imaging. Fluorescence images from the dCas9 channel are shown. Repetitive targets include telomeres (sgTelomere) and PPP1R2 (sgPPP1R2) as well as a non-target control (sgNT). (C) Alignment of Alu family sequences and sgAlu guide RNA target (green shaded). Zoomed-in: aligned sequence, and annotations for protospacer and PAM of sgAlu guide RNA design.

CUT&RUN sequencing verified the specificity of Alu targeting.

Related to: Figure 1. (A) dCas9 CUT&RUN library insert size post adapter trimming for (top) sgAlu and (bottom) sgNT. (B) U2OS Hi-C contact matrix on chromosome 2 (5-80Mb) alongside genomic tracks shown in Figure 1 D. Hi-C data obtained from an existing dataset (Arnould et al., 2021).

Alu-rich regions are depleted in heterochromatin but Alu elements are not correlated with histone density within euchromatin.

(A) Fluorescence images of Alu-labeled cells co-expressing markers of distinct subnuclear regions: HP1α for heterochromatin and SRRM1 for nuclear speckles. Images are shown in individual and merged channels (magenta: marker; green: Alu elements) as annotated. Scale bar, 10 μm. (B) Change in Pearson correlation coefficient (PCC) of pixel intensities in nucleoplasmic regions between the marker and Alu elements (Alu-targeted dCas9) channels in (A), compared to the mean of control sgRNA (sgNT). Data is shown as violin plots (estimated probability density) with median (blue solid line, middle) and first and third quartiles (dashed lines, bottom, and top, respectively) inside. Grey solid line: change = 0 (mean of PCC for sgNT). n ≥ 150 nuclei for each group. ** denotes P < 0.01 and *** P < 0.001 using two-sided Brunner-Munzel test with t-distribution to compare sgAlu and sgNT conditions for each marker. (C) Mean standardized Alu intensity in euchromatin and heterochromatin regions. Each dot corresponds to a nucleus. n = 15 nuclei. *** denotes P < 0.001 using Mann-Whitney U rank test. See Figure S3 A for Alu- and H2B-intensity distributions in an example cell. (D) Joint distribution of Alu and H2B pixel intensities within euchromatin. Pearson correlation coefficient r = −0.07.

Context-aware histone mobility analysis integrating local Alu-element intensity uncovers sequence-specific mobility across euchromatin.

(A) Chromatin MSND in euchromatin regions plotted against lag time. Data is shown as mean (thick line) over n = 15 individual nuclei (thin lines). Triangle represents slope = 0.5. (B) An H2B displacement field corresponds to the nucleus shown in (C). Displacement vectors from nucleoli and heterochromatin regions are excluded (see Methods). (C) Representative images for Alu and H2B channels. (D) Framework for context-aware chromatin mobility analysis. H2B displacement fields are spatially combined with context image(s), allowing for the analysis of chromatin mobility in different chromatin context(s). (E) H2B MSND plotted against lag time and stratified against respective chromatin context (Z-score): Alu density (top) and H2B density (bottom). Colors represent relative context image intensity. Triangles represent slope = 0.5. (F) Dependence of H2B MSND at lag time τ = 5 s on chromatin contexts (Z-score): Alu density (top) and H2B density (bottom). (G) Heat map showing H2B MSND at different combinations of Alu density (Z-score) and H2B density (Z-score), at lag time τ = 5 s. Colors represent the squared displacement in μm2. In (E), (F), and (G), n = 15 nuclei.

Relationship between Alu and H2B pixel intensities in euchromatin and heterochromatin regions.

Related to: Figure 2. (A) Mean standardized H2B (left) and Alu (right) intensity distributions in euchromatin (light) and heterochromatin (dark) regions in a single cell provided as an example. (B) Joint distribution of Alu and H2B pixel intensities within either euchromatin or heterochromatin regions. Pearson correlation coefficients: r = −0.07 for euchromatin, and r = 0.35 for heterochromatin. The Euchromatin panel is the same as Figure 1 and is shown here for direct comparison with the heterochromatin panel.

Details for PIV workflow and H2B MSND at different lag times.

Related to: Figure 3. (A) Representative images of H2B and Alu channels and the masks for nucleoli and heterochromatin regions used in MSND analyses. (B) Schematics illustrating PIV workflow to estimate chromatin network displacement field(s). (C) Dependence of H2B MSND at different lag time τ on chromatin contexts: H2B density (top) and Alu density (bottom). Lag times τ = 0.5 s, 1.0 s, 5.0 s and 11.0 s are shown. (D) Heat maps showing H2B MSND at different combinations of Alu density (sfGFP) and H2B density (miRFP670) at different lag times τ = 0.5 s, 1.0 s, 5.0 s and 11.0 s. Colors represent the squared displacement in μm2. In (C) and (D), n = 15 nuclei.

Alu-element mobility probed by single-particle tracking and PIV.

Related to: Figure 3. (A) Fluorescence image of higher resolution, compared to Figure 1 B, allowing visualization of individual Alu-element foci. Scale bar, 10 μm. (B) MSD plotted against lag time for single particle tracking of Alu-elements domains. Data shown as time-averaged MSD (grey) and their mean (blue). (C) Alu MSND plotted against lag time and stratified against respective chromatin context: Alu density (right) and H2B density (left). Colors represent relative context image intensity (Z-score). Triangles represent slope = 0.5. n = 15 nuclei. Figure 3 E is the H2B MSND equivalent. Shaded areas in (B) and (C) correspond to the same range of lag time.

Alu element-specific chromatin mobility is reduced upon Pol II transcription inhibition.

(A) Representative fluorescence images of U2OS cells treated with transcription inhibitors α-amanitin (aAM), flavopiridol (FVP), and actinomycin D (ActD) for individual channels or merged. Magenta, H2B; green, Alu elements. Scale bars, 10 μm. (B) H2B MSND euchromatin region plotted against lag time for each treatment. Triangles represent slope = 0.5. (C) Percent change in H2B MSND, at lag time τ = 5 s, in Alu-rich or Alu-poor regions after treatment. Data represented as mean ± s.e.m. ** denotes P < 0.01, *** P < 0.001, and n.s. not significant using two-sided Brunner-Munzel test with t-distribution to compare each treatment to the control (media). (D) Heat maps showing (top) H2B MSND at different combinations of Alu density and H2B density, at lag time τ = 5.0 s, after treatment, and (bottom) corresponding fold-change compared to control (media). Colors represent (top) squared displacement, in μm!, and (bottom) fold change. For (B), (C), and (D), n ≥ 10 nuclei for each condition.

H2B MSND behaviors upon Pol II transcription inhibition.

Related to: Figure 4. (A) Individual ensemble MSND traces for each nucleus are shown (light) together with the corresponding mean (heavy). Triangles represent slope = 0.5. Related to Figure 4 B. (B) (Top) H2B MSND at different degrees of Alu density (Z-score), at lag time τ = 5 s, after each treatment, and (bottom) corresponding fold-change (compared to media-only). Alu-rich (dark-shaded) or -poor (light-shaded) regions are separated at standardized Alu intensity (Z-score) = 0. Data represented as mean ± s.d. For (A) and (B), n ≥ 10 nuclei for each condition.

Distinct euchromatic sub-domains with higher mobility coupled to transcription.

Schematic illustrating the non-uniform changes in euchromatin mobility between Alu-rich and -poor areas. Mobility of Alu-rich chromatin might increase upon Pol II transcription inhibition, where transcription machinery falls off from chromatin (e.g., α-amanitin and flavopiridol), and might decrease if transcription machinery gets stalled on chromatin (e.g., actinomycin D).

Key resources

Protospacer sequences for sgRNAs used in this study