iiinova — where your digital self keeps evolving

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 06/07/2026

(8/10) With our analyses, we discovered a rare variant that disrupts OCT motif within a distal enhancer of PIK3R5 gene. Using CRISPR-Cas9 variant editing we restored PIK3R5 gene activity, validating predictions and showing variant causality for its expression.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 06/07/2026

(6/10) To this end, we used ChromBPNet deep learning models to learn TF motif syntax. We then searched for variants that disrupt TF binding sites thus explaining allele-specific accessibility. This revealed hundreds of TF binding disruptions, which we validated using ChIP-seq data.

151

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 06/07/2026

(3/10) Next, we mapped chromatin accessibility and gene expression via ATAC-seq and TT-seq. Integrating these with phased variants, we built a personal-genome pipeline to quantify and connect allele-specific open chromatin peaks to genes.

Phased variants can be used to identify allele-specific ATACseq peaks, linked to allele-specific expression data.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 06/07/2026

(2/10) To link regulatory variants to gene expression across large distances, we reconstructed chromosome-scale haplotypes using 10X linked-reads and Hi-C. This enabled high-quality phasing of variants in five individuals with diverse genetic backgrounds.

Cartoon describing how synthetic long reads and Hi-C can be used to haplotype individuals.

110

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/03/2026

To test this, we re-expressed an N-terminal truncated CTCF incapable of cohesin interaction and chromatin looping. This variant was sufficient to rescue both the expression of CTCF-bound genes and the morphogenesis defects. Looping is dispensable — at this stage. 6/8

140

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/03/2026

Early CTCF depletion severely impairs gastruloid morphogenesis. But when we profiled chromatin accessibility (ATAC-seq) and gene expression (time-resolved RNA-seq), differentiation was largely intact. Morphogenesis and differentiation can be uncoupled. 4/8

CTCF depletion disrupts morphogenesis of gastruloids

150

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 26/08/2025

Very interesting indeed! Also wondering what we see here are extrusion fountains in the RBCs?

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/08/2025

At the level of TADs we also saw intriguing differences. Below you can see an example of a region, where in one of the patient samples the TAD structure seems to be almost completely gone. We currently do no understand what causes these differences in TAD strength.

100

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/08/2025

However(!), when we looked at late stage cancer biopsies that were acquired from pleural effusions, we observed massive differences between patients. Both compartment strength and compartment identity was highly variable between patients.

100

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/08/2025

When we compared the 3D genome of primary tumors to those of metastatic samples we were once again surprised by the similarity. We also had difficulty identifying consistent changes among all the patients.

100

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/08/2025

Also at the level of compartments there seems to be limited differences in compartment scores and there do not seem to be outright switches of compartments from A to B or vice versa when we compared the primary tumor to the healthy tissue.

100

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 18/08/2025

We performed in situ Hi-C on the samples to determine the 3D genome organization. To our surprise, at the assayed resolution, the Hi-C maps look quite similar between healthy breast and primary tumor. Below this can be seen for the insulation score, which indicates the position of TAD boundaries.

scatterplots showing similarity in insulation scores

110

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 16/07/2025

Strong The Onion vibes with this one: www.nature.com/articles/d41...

040

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 01/07/2025

(8/n) Do fountains have a functional role? Our RNA-seq data revealed that their disruption resulted in downregulation of genes around fountains. This would be consistent with a model where cohesin loading at enhancers promotes enhancer-promoter communication and activates genes.

Enrichment of downregulated genes following RAD21 depletion in fountain bases. And a cartoon suggesting a mechanism.

110

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 01/07/2025

(7/n) Where do fountains originate from? We found that fountains are enriched in enhancer chromatin features, consistent with observations in other species that fountains are formed at regions of active chromatin.

ChromHMM enrichment heatmap showing that enhancer categories are enriched in fountain bases.

130

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 01/07/2025

(5/n) Are fountains then a consequence of loop extrusion? To understand this, we tagged RAD21 in the WAPL/CTCF background, making it a triple degron cell line. Sequential depletion experiments clearly demonstrated that fountains are dependent on cohesin and reduced in its absence.

Hi-C maps directly showing the role of cohesin in the formation of fountains.

130

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 01/07/2025

(4/n) What is the mechanism behind their formation? Regions making up the fountains exhibit strong cohesin binding, which is repositioned away upon WAPL and CTCF depletion. Additional analyses of WAPL and CTCF single depletions, suggests that cohesin might be loaded directly at the fountain base.

ChIPseq heatmaps suggesting that cohesin is loaded at fountain bases.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 01/07/2025

(3/n) Systematic analysis of fountains revealed that they can already be visible as a fountain-like pattern in the unperturbed cells. Fountains are extended upon depletion of WAPL and CTCF, which constrain fountain dynamics.

Hi-C maps and aggregate plots showing fountains in unperturbed and perturbed cells.

110

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 20/12/2024

In an in vitro embryo model (‘gastruloids’), we find that loss of ZFP143 reduces growth and severely hampers development. A decrease in mitochondrial membrane potential suggests that also in gastruloids mitochondrial function is disrupted.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 20/12/2024

To check this we performed TT-seq and RNAseq following depletion of ZFP143. We find that many nuclear encoded mitochondrial genes are downregulated in the absence of ZFP143. This downregulation also affected mitochondrial morphology and proliferation.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 20/12/2024

While the “real” ZNF143 peaks are diminished, those overlapping with CTCF binding sites are increased, which is exactly what you would expect in the absence of the primary antibody target.

120

Elzo de Wit lab @ NKI @dewitlab.bsky.social · 20/12/2024

We performed degron experiments for ZFP143 (mouse ZNF143) and could not detect any changes in the Hi-C maps, suggesting that this protein might have nothing to do with looping.

130