Measurement of the zero-torsion TMP baseline.

(a) Strategy for determination of the TMP signal with and without torsion. (left) Cells with torsion are crosslinked with TMP prior to library preparation for the TMP signal measurement. (right) To measure the TMP baseline signal without torsion, the native-bound protein landscape is fixed by formaldehyde, and the torsion is released via DpnII digestion of the spheroplast cells before TMP crosslinking. Both TMP crosslinking samples are then mapped using Illumina-based sequencing to generate the TMP signal (Materials and Methods).

(b) Heatmaps of TMP signals aligned at the TSSs of 5,925 genes for both the TMP signal with torsion (left, n = 2 biological replicates) and the TMP signal without torsion (right, n = 3 biological replicates) with rows sorted by the ORF size. The TMP signal was normalized against that of the purified DNA.

(c) (top) Average TMP signals with and without torsion from data in (b) aligned at the TSS (left) and aligned at the TES (right). For comparison, previous H3-ChIP-MNase-Seq data27 are also shown (bottom). To minimize signal bias from neighboring genes, signals downstream of the TES were excluded from the analysis of the left panel plots, and signals upstream of the TTS were excluded from the analysis of the right panel plots.

The twin-supercoiled domains of transcription.

(a) Average TMP signals with and without torsion using data from Fig. 1c, replotted for easy comparison, except without normalization against that of the purified DNA.

(b) The torsion signal, defined as the difference between the TMP signal with torsion and the TMP signal without torsion. The bottom cartoon shows the twin-supercoiled domain model of transcription.

(c) The torsion signal of genes sorted by transcription level and gene size, with genes grouped by percentiles of mRNA abundance and ORF size, respectively. Positions within the gene body are scaled by the gene length while preserving the torsion signal amplitude (Materials and Methods). The torsion signal is compared to previous data of Pol II (Rpb3-ChIP-Exo)27, topo I (Top1-ChIP-Exo)27, and topo II (Top2-ChIP Exo)27.

(d) Mean value of the torsion signal averaged within 20 bp of either the peak near the TSS (red) or the valley near the TES (blue) as functions of the average gene expression level (left) or gene size (right) within the percentile groups specified in (c). Error bars are standard errors of the means.

Torsion between genes.

The torsion signal plotted for three gene-pair configurations: divergent (left, 1275 pairs), convergent (middle, 1364 pairs), co-directional (right, 2426 pairs). The torsion signal is compared to previous protein mapping data of the same gene-pair configurations for cohesin (SccI-Chip)24, topo I (Top1-ChIP-Exo)27, and topo II (Top2-ChIP Exo)27, and Pol II (Rpb3-ChIP-Exo)27. TSS or TES is indicated as “T”.

Torsion impacts 3D genome structure and couples gene expression.

(a) Cohesin-ChIP seq24 and torsion signals plotted across 527 detected mitotic loops from previous MicroC-XL Data52, grouped by percentiles of the loop score. Positions between the two loop boundaries (L and R) are scaled by the DNA length between the boundaries while preserving the torsion signal amplitude (Materials and Methods). Below is a cartoon depicting a DNA loop anchored by cohesin at sites of (+) supercoiling.

(b) Average ChIP-Exo signal of Abf1 and Reb127 (top plot), and torsion plotted across subsets of divergent genes (middle and bottom plots). Gene pairs are separated by defined classes27 where both genes in the pair are of the same class as follows: unbound class (UNB; green, 421 pairs), ssTFs only (TFO; red-solid, 320 pairs), ssTFs and other factors (TFO, STM, and RP; red-dotted, 439 pairs). Positions between the two divergent promoters are scaled by DNA length between the two promoters while preserving the signal amplitude (Materials and Methods).

Psoralen binding and accessibility. Cartoon depicting the experimental method for mapping TMP cross-linking on genomic DNA. The TMP has preferential binding to naked DNA where there are no bound proteins, such as in nucleosome-free regions and on linker DNA.

Psoralen Crosslinking Conditions.

(a) Average TMP signal with torsion of all 5925 genes as in Figure 1, shown for increasing UV dosage on cells. (Porter: did you divide by the purified DNA signal?)

(b) Average TMP signal without torsion from the same genes.

(c) Average TMP signal of naked genomic DNA from the same genes. 250 ng of sheared naked genomic DNA was incubated for 2 min with the indicated concentration of psoralen and exposed to UV light for 10 s. The density of adjacent AT or TA dinucleotide cross-linking sites is plotted for comparison with the naked DNA profiles (XL site density, dark blue).

(d-f) Percent of 250 ng of input DNA remaining after two rounds of denaturation and treatment with ExoI for each respective condition in (a-c). The response in % DNA remaining is near linear with UV dosage. At low concentrations, we see that the % DNA remaining approaches the background (d-f) as does the signal from the library prep (a-c). This indicates that if the dose is too low, the signal-to-noise limit is being approached. Upon high UV exposure, the % DNA remaining indicates that there are likely multiple crosslinks/fragments (d,e), which may lead to artifacts due to the loss of small fragments during library prep. Upon a Goldilocks dose of UV exposure or psoralen concentration, signals are very similar to each other, indicating that most fragments have exactly one crosslink after Exo digestion (Fig. Supplement 3).

Optimal TMP crosslinking density. Cartoon depicting the optimal inter-strand cross-linking density is when each fragment of DNA has only one TMP cross-link after Exonuclease I digestion.