High-throughput DNA/RNA FISH

(A) Schematic overview of the HiFISH pipeline used to simultaneously measure TAD boundary distance and gene activity at the single-cell and single-allele levels. Step 1: Design of DNA FISH probes based on Micro-C profiling and detection of DNA and nascent RNA by HiFISH. Step 2: Measurement of center-to-center TAD boundary distances and RNA signal at individual alleles by image analysis using HiTIPS(40). Step 3: Quantitative comparison of TAD boundary distances with gene activity at each allele. (B) DNA/RNA HiFISH detection of 5’ (green) and 3’ (red) MYC TAD boundaries and nascent RNA (blue) in HBEC cells. Scale bar: 10 μm. (C) Quantification of MYC nascent RNA signals using DNA/RNA HiFISH in fixed HBEC cells or a MS2-tagged MYC reporter in living HBEC cells. Bars represent means +/- SEM from two experiments. Dots indicate means from individual experiments. 166,953 cells were analyzed for MS2, and 30,137 cells for DNA/RNA HiFISH. Statistical significance was calculated using two-way ANOVA with Bonferroni correction: ns, not significant (p ≥ 0.05).

TAD boundaries interact more frequently than non-TAD regions

(A) Micro-C contact maps for EGFR and MYC TADs and adjacent regions in HFF cells, highlighting TAD boundaries and genomically equidistant non-TAD regions. Squares denote the probe positions used for 3’ (green), 5’ boundary (red), and equidistant non-TAD controls (purple). Interactions between the 5’ TAD boundaries and the 3’ TAD (yellow) or non-TAD boundaries (grey) are highlighted and total Micro-C contacts between regions are quantified, emphasizing high TAD boundary contact frequency in both EGFR and MYC TADs, as well as weaker signals in the non-TAD regions. (B-C) Representative DNA HiFISH images of EGFR and MYC TAD boundary and non-TAD regions in HFF (B) and HBEC (C). Scale bar: 10 μm. (D-E) Measurement of boundary distances. Distance distributions of EGFR and MYC TAD boundaries versus matched non-TAD regions in HFF (D) and HBEC (E) cells. Dashed line indicates 250nm threshold used to define physical interaction. Between 2,000 and 18,000 alleles were analyzed per sample. Values represent an individual dataset from a single experiment of multiple experiments. Mann-Whitney U test p-values are: *** p < 1×10⁻¹⁰⁰; ** 1×10⁻¹⁰⁰ ≤ p < 1×10⁻²⁰; * 1×10⁻²⁰ ≤ p < 0.01.

TAD boundary proximity is not related to gene activity status

(A) Representative DNA/RNA HiFISH image of EGFR nascent RNA (blue) and its associated 5’ (red) and 3’ (green) TAD boundaries in HBECs, illustrating detection of active (RNA-positive) and inactive (RNA-negative) alleles. Scale bar: 10 μm. (B–E) Comparison of TAD boundary distances for EGFR and MYC alleles based on transcriptional activity status. Histograms of allele-specific distance distributions from a representative dataset from a single experiment; Mann–Whitney U test p-values are indicated as follows: ns, not significant (p ≥ 0.05). Dot plots of the mean of median distances from multiple experiments (500–20,000 alleles per condition); Error bars represent SEM and statistical significance was calculated using two-way ANOVA with Bonferroni correction: ns, not significant (p ≥ 0.05).

Global transcription inhibition does not alter TAD boundary pairing

(A) Representative RNA HiFISH images of MYC nascent RNA in HBECs with and without 2-hour DRB treatment. Scale bars: 10 μm. (B) Quantification of TAD boundary distances for EGFR and MYC in the presence or absence of DRB. Dot plots of the mean of median distances from multiple experiments (500–20,000 alleles per condition). Error bars represent SEM. Statistical significance was calculated using two-way ANOVA with Bonferroni correction: ns, not significant (p ≥ 0.05). boundaries and non-TAD regions for the ERRFI1, FKBP5, and VARS2 TADs revealed more similar interaction frequencies between TAD boundaries compared to equidistant non-TAD control regions (Figure 5A). Accordingly, DNA FISH showed a more uniform distribution of TAD boundary distances for the Dex-inducible ERRFI1, FKBP5, and VARS2 TADs in HBEC compared to the Dex non/mildly inducible MYC and EGFR TADs

Transcription stimulation does not alter TAD boundary interactions

(A) Micro-C maps for the ERRFI1, FKBP5, and VARS2 TADs and neighboring regions, showing TAD boundaries (green, red) and equidistant non-TAD control regions (purple) in HFF; corresponding probe positions are indicated. Interactions between the 5’ TAD boundaries and the 3’ TAD (yellow) or equidistant non-TAD control regions (grey) are highlighted, and total Micro-C contacts between regions are quantified, showing prominent contact frequency between ERRFI1, VARS2, and FKBP5 TAD boundaries, as well as the non-TAD region of ERRFI1. (B) Comparison of TAD boundary and non-TAD region distances for EGFR, MYC, ERRFI1, FKBP5, and VARS2 in HBECs as measured by DNA HiFISH. Dot plots of the mean of median distances from two experiments (11,000 – 49,000 alleles per condition). Error bars represent SEM. Statistical significance was calculated using two-way ANOVA with Bonferroni correction: ****p < 0.0001; **p < 0.01; ns, not significant (p ≥ 0.05). (C) Measurement of boundary distances. Distance distributions of ERRFI1, FKBP5, and VARS2 TADs in untreated and 2-hour Dex-treated HBEC. Between 2,500 and 6,000 alleles were analyzed per condition. Values represent an individual dataset from a single experiment representative of multiple experiments. Mann-Whitney U test p-values are indicated as follows: ns, not significant (p ≥ 0.05).

Effects of RAD21 and CTCF depletion on TAD boundary architecture and gene expression

(A) Representative DNA/RNA HiFISH images of EGFR nascent RNA FISH (blue) and its 3’ (green) and 5’ (red) TAD boundaries DNA FISH in HCT116-RAD21-AID1 and HCT116-CTCF-AID2 cells, respectively, in control and auxin-treated conditions. Scale bar: 10 μm. (B) TAD boundary distances and non-TAD controls after RAD21 or CTCF depletion for 3 hours. Values represent an individual dataset from a single experiment representative of multiple experiments. Between 13,000 and 127,500 alleles were analyzed per condition. Mann-Whitney U test p-values are indicated as follows: *** p < 1×10⁻¹⁰⁰; ns, not significant (p ≥ 0.05). (C) Fraction of silent (0), monoallelic (1), biallelic (2), and triallelic or more (≥3) expression of the indicated genes in individual cells after 3 hours or no auxin treatment in HCT116-RAD21-AID1 or HCT116-CTCF-AID2 cells. At least 20,000 cells were measured per experiment. Data represent values from at least two independent experiments (diamonds and circles); diamonds (DMSO control) and circles (RAD21 or CTCF-depleted) represent the mean of means and error bars indicate SD. p-values from two-way ANOVA with Bonferroni correction are shown as: ****p < 0.0001; **p < 0.01; *p < 0.05; ns, not significant (p ≥ 0.05).

Micro-C chromosome interaction maps and ChromHMM analysis of EGFR, MYC, ERRFI1, FKBP5, and VARS2 TADs in HFFc6

ChromHMM chromatin states for two different foreskin fibroblast cell lines. The cytogenic chromosome band track (grey) indicates the chromosome location of the indicated loci. The gene reference track (blue) shows all coding and non-coding genes. Diamonds denote probe interaction sites in Micro-c for both the non-TAD probes (red) and the TAD boundary probes (blue).

Sequence and location of DNA and RNA probes binding sites for DNA/RNA HiFISH

Top: schematic representation of target regions. Oligonucleotide targeting sequences are indicated in blue. Bottom: target sequence regions.

TAD boundary proximity is uncoupled from allelic gene activity in single nuclei in both 2D and 3D imaging

(A) Comparison of TAD boundary distances for EGFR and MYC alleles in HCT116 based on transcriptional activity status. Histograms of allele-specific distance distributions from a representative dataset from a single experiment. Mann–Whitney U test p-values are indicated as follows: ****p < 0.0001; ns, not significant (p ≥ 0.05). (B) Comparative analysis of TAD boundary distances between active and inactive alleles within the same nucleus for EGFR and MYC loci in HBEC and HFF cells. (C) Comparison of TAD boundary distances for MYC alleles in HBEC based on transcriptional activity status, measured using both 2D and 3D imaging. Histograms show allele-specific distance distributions from a representative dataset of a single experiment. Mann–Whitney U test p-values are indicated as follows: ns, not significant (p ≥ 0.05).

Transcriptional inhibition does not affect the spatial organization of non-TAD control regions

(A) Comparison of TAD boundary and non-TAD distances for EGFR and MYC alleles in HBEC and HFF based on transcriptional inhibition status. Histograms show allele-specific distance distributions from a representative dataset of a single experiment.

RNA levels following DEX treatment

(A) RNAseq analysis of ERRFI1, VARS2, and FKBP5 RNA levels following Dex treatment for the indicated durations in HBEC cells. Values were calculated RPKM fold-change (Dex/No_Dex) ratio. Data represent the mean of three independent experiments. (B) Histograms of the distribution of nascent RNA transcription sites per nucleus in HBECs upon Dex treatment. Data represent values from at least two independent experiments (diamonds and circles); diamonds (EtOH control) and circles (Dex 2 h) represent the mean of means, and error bars indicate SD. P-values from two-way ANOVA with Bonferroni correction are shown as: ****p < 0.0001; ***p < 0.001; **p < 0.01; *p < 0.05; ns, not significant (p ≥ 0.05).

Depletion of RAD21 and CTCF

(A) Loss of RAD21 or CTCF in HCT116-RAD21-AID1 or HCT116-CTCF-AID2 cells, respectively, following DMSO (control) or auxin treatment. RAD21 and CTCF degradation were assessed using mClover- fluorescence (green). Scale bar: 20 μm (B-C) Fraction of silent (0), monoallelic (1), biallelic (2), and triallelic or more (≥3) expression of the indicated genes in individual cells after 3 hours or no auxin treatment in HCT116-RAD21-AID1 (B) or HCT116-CTCF-AID2 (C) cells. Data represent values from at least two independent experiments (diamonds and circles); diamonds (DMSO control) and circles (RAD21 or CTCF-depleted) represent the mean of means, and error bars indicate SD. P-values from two-way ANOVA with Bonferroni correction are shown as: *p < 0.05; ns, not significant (p ≥ 0.05).