Single-molecule imaging reveals a direct role of CTCF’s zinc fingers in SA interaction and cluster-dependent RNA recruitment

Abstract CTCF is a zinc finger protein associated with transcription regulation that also acts as a barrier factor for topologically associated domains (TADs) generated by cohesin via loop extrusion. These processes require different properties of CTCF-DNA interaction, and it is still unclear how CTCF’s structural features may modulate its diverse roles. Here, we employ single-molecule imaging to study both full-length CTCF and truncation mutants. We show that CTCF enriches at CTCF binding sites (CBSs), displaying a longer lifetime than observed previously. We demonstrate that the zinc finger domains mediate CTCF clustering and that clustering enables RNA recruitment, possibly creating a scaffold for interaction with RNA-binding proteins like cohesin's subunit SA. We further reveal a direct recruitment and an increase of SA residence time by CTCF bound at CBSs, suggesting that CTCF-SA interactions are crucial for cohesin stability on chromatin at TAD borders. Furthermore, we establish a single-molecule T7 transcription assay and show that although a transcribing polymerase can remove CTCF from CBSs, transcription is impaired. Our study shows that context-dependent nucleic acid binding determines the multifaceted CTCF roles in genome organization and transcription regulation.


Figure S1 .
Figure S1.CTCF diffuses on non-CBS sites.(A) Scheme illustrating laser illumination during lifetime measurements.(B) Diffusion coefficients of AF568-CTCF WT and variants on 4x CBSs or λ-DNA at 10 nM concentration and 100 ms illumination time.For all variants, D is significantly higher on λ-DNA than CBSs.No significant difference in diffusive behavior between CTCF variants.(C) Photobleaching steps of non-diffusive (top) and diffusive (bottom) CTCF.(D) Representative kymograms showing diffusion behavior of WT CTCF.Top: White arrows indicate events where diffusive CTCF is blocked by CBSbound CTCF.No recruitment of diffusive CTCF to the binding sites occurs.Bottom: White arrows indicate events where diffusive CTCF passes CBS-bound CTCF.(E) Quantification of blocking and passing events.No significant differences were observed between WT and CTCF variants.

Figure S2 .
Figure S2.Lifetimes of AF568-CTCF variants at 10 nM concentration.(A) Lifetimes of CTCF variants at 10 and 40 s frame delay and 100 ms illumination time.(B) Photobleaching steps of CTCF variants binding to 4x CBSs.Black line: Multi-Gaussian fit.

Figure S3 .
Figure S3.SA diffusion behavior and influence of Rad21 on CTCF-SA interaction.(A) Enrichment of SA1 (purple) and SA2 (pink) on CBSs.During simultaneous load (left), both SA-LD655s are enriched significantly more on CBSs when preincubated at 100 nM concentration with 10 nM AF568-CTCF WT compared to SAs loaded alone.SA is enriched similarly in presence of all CTCF variants.When 10 nM AF568-CTCF was bound first to DNA, followed by salt enrichment and by 100 nM SA-LD655 (sequential load, right), no significant SA enrichment at CBSs was observed with any CTCF variant.(B) Representative kymograms of SAs binding static to AT-rich and diffusing randomly on GC-rich

Figure S4 .
Figure S4.In vitro transcription assay at low nucleotide concentration (A) Illustration of in-vitro transcription assays and representative kymogram for T7-Pol pushing CTCF off its site during transcription at 50 µM nucleotide concentration (cyan = Cy3-UTP labeled RNA, green = AF568-CTCF) (B) Mean transcription velocities of T7-Pol alone and T7-Pol pushing CTCF WT at 50 µM nucleotide concentration.

Figure S5 .
Figure S5.SA has a higher affinity for RNA than for DNA.(A) Left: Histogram of AF568-CTCF-WT-RNA cluster (N = 355) binding positions after salt enrichment (see Figure 5D).Right: CTCF WT-RNA clusters are significantly less enriched on CBSs than monomeric CTCF WT. (B) Histogram of CTCF WT binding positions after transcription and salt enrichment (see Figure 5G).CTCF preferentially binds to 4x CBSs but also colocalizes with RNA transcripts (CTCF, N = 292; RNA, N = 237).(C) Left: Histogram of AF568-CTCF-ΔNC-RNA cluster (N = 280) binding positions after salt enrichment.Right: ΔNC-RNA clusters are significantly less enriched on CBSs than monomeric ΔNC.(D) Histogram of photobleaching steps for ΔNC on 4x CBSs (same as Figure S2B-3) and in RNAclusters.(E) Scheme and representative kymogram of SAs being washed off from λ-DNA by the addition of RNA to the DNA-curtain.(F) More 100 nM SA-LD655 is washed off in the presence of 25 ng/µl Cy3-UTP labeled RNA than in the presence of buffer.(G) Scheme and representative kymogram of SAs colocalizing with RNA transcripts.(H) Histogram of SA1 binding positions after transcription (SA1, N = 43; RNA, N = 155).(I) SA1 and SA2 are significantly less enriched on AT-rich regions in presence of RNA transcripts on the DNA.(J) SA1 and SA2 have a significantly higher lifetime on RNA transcripts than on AT-rich or GC-rich DNA regions.All experiments were carried out at 100 ms illumination time.

Table S1 .
Number of molecules and p-values for all relevant experiments.Two tailed t-test: (t), Two tailed z-test: (z), Fisher's exact test: (f).