iiinova — where your digital self keeps evolving

Ed Banigan @irate-physicist.bsky.social · 15/12/2025

Our (Viraat Goel, @andersshansen.bsky.social et al) paper Dynamics of microcompartment formation during the M-to-G1 transition is in this month's @natsmb.nature.com -- thanks to NSMB editors and staff for the cover space and art! www.nature.com/nsmb/volumes... See quoted posts for quick summary

December 2025 Cover of NSMB. "Mitotic chromatin microcompartments." Image is colorful tents scattered across a snowy mountainside.

0143

Ed Banigan @irate-physicist.bsky.social · 17/10/2025

Previous experiments observed a transient spike in transcription of a subset of genes during ana/telophase. We now observe that these spiking genes are associated with microcompartments that peak in interactions during ana/telo, suggesting a 3D-genome-based mechanism for this transcriptional spike!

Panel B.
Top left: Pol II ChIP at 88 TSSs across M-to-G1 transition, showing varied behavior, but clear spiking of some genes.
Bottom left: First principal component of time series shows spike in ana/telophase
Right: Colored stackup, sorted on PC1 at each time point showing normalized Pol II ChIP. prometa is very low ChIP across all genes (all blue). 33 spiking genes have high Pol II ChIP in ana/telo & associated with larger PC1 (red on top); small/negative PC1 associated with low/no ChIP signal (bluer on bottom). Less correlation in G1 (blue,red,white scattered across genes sorted by PC1).

Panel D
four plots from left to right:
1. Mean microcomp. loop strength of TSSs vs. time point. Spikes in ana/telophase, more dramatically for transcriptionally spiking TSSs
2. Scatter plot of Mean loop strengths in ana/telo vs. mid g1 showing stronger interactions in A/T
3. microcomp. strengths of PP loops vs time for spiking promoter pairs, 1 spiking 1 not, and non-spiking pairs. Txn spike associated with stronger microcomp interaciton spike
4. Same as 3 except for EP and EE pairs.

030

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

Can PDS5 therefore strengthen CTCF boundaries by limiting NIPBL-cohesin processivity (=speed x residence time)? Hi-C in cells depleted of PDS5 & now partially of NIPBL recover some of CTCF boundaries (+ also compartments)

Hi-C contact maps in control, PDS5AB-depleted, and PDS5AB+partial NIPBL -depleted cells.
Top 87 Mb window maps show checkerboard (compartments) in control, but not PDS5A/B depletion. Some recovery of compartments with partial depletion of NIPBL + PDS5 depletion.
Bottom 10 Mb window maps show corner peaks and TADs, diminished/lost in PDS5AB depletion, partially recovered with co-depletion of NIPBL

100

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

To understand this Iain did 3-color single-molecule expts w/loop-extruding NIPBL-cohesin. PDS5+NIPBL could colocalize on cohesin leading to NIPBL unbinding! Loops often released but sometimes shrunk & NIPBL returned. So PDS5 slows loop growth + speeds NIPBL unbinding by facilitated dissociation!

Images of single-molecule expts with DNA tethered at two ends with fluorescent DNA, NIPBL, and PDS5 (+cohesin in solution, not tagged)
Left: While NIPBL is on DNA, DNA fluorescence locally accumulates indicating an extruded loop. After some time, PDS5A briefly colocalizes. Shortly later NIPBL unbinds and loop is lost/released. PDS5A subsequently unbinds.
Middle: Similar sequence with PDS5B
Right: DNA loop growth with NIPBL localization on DNA, interrupted by stalling of growth when PDS5A intermittently binds and unbinds. After some time a PDS5A event occurs prior to NIPBL loss. The loop is lost, but NIPBL rebinds while PDS5A remains present and the loop begins to grow again after PDS5A unbinds

100

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

We turned to the possibility of PDS5 competing w/NIPBL for cohesin PDS5 inhibits cohesin-NIPBL ATPase+binding in vitro & suppresses NIPBL occupancy of cohesin in cells(+ PDS5&WAPL dont affect each others occupancy) iFRAP shows NIPBL chromatin residence time seems to increase a lot w/ PDS5 depletion

plot of fluorescence decay of NIPBL-eGFP in iFRAP expts. WAPL depletion shows decay over ~1.4 min similar to control. PDS5A/B or WAB show decay over ~3.5 min.

100

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

Since PDS5A/B aren't required for cohesin accumulation at CTCF, are they altering TADs etc via SMC3 acetylation? PDS5 depletion reduces acetylation but in separate expts w/ ESCO1/2 depleted we see no acetylation & TADs+CTCF contacts remain. PDS5 not influencing TADs+corner peaks only by acetylation.

Hi-C maps showing ESCO1/2 depletion somewhat increases quantity+strength of corner peaks while PDS5A/B depletion decreases quantity+strength of corner peaks

110

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

What about CTCF? PDS5A/B isnt needed to recruit CTCF & it aids cohesin accumulation at CTCF. But PDS5-mediated cohesin accumulation isnt merely due to recruitment to CTCF. PDS5 depletion leads to new accumulation of cohesin elsewhere in “cohesin islands” often at sites of convergent transcription.

Top: ChIP-seq stackups in 20 kb windows for different depletions showing cohesin accumulation in "cohesin islands" emerges in PDS5A/B depletion, to a lesser extent in WAPL depletion, and very much in WAPL-PDS5A/B depletion.
Bottom: Hi-C pileups show these islands correspond to corner-peak contacts and insulation of contacts on either side of the island.

110

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

How does increasing extrusion life hinder or facilitate compartments? Competition w/chromatin conformational dynamics! Extrusion disrupts contacts. Compartmentalization results from equilibration via polymer relaxation Fast cohesin turnover->disruptions occur more quickly than they can be relaxed

Top plot: chromatin relaxation time for different segment sizes vs. residence time is non-monotonic in residence time. Relaxation time initially decreases, but once relaxation time is < residence time, it begins to increase.
Bottom plot: compartment contrast vs. residence time again. The same monotonic behavior is followed as for relaxation time.
Gray regions on both plots are identical: they show that relaxation time and compartment contrast both increase as a function of residence time once residence time is longer then relaxation time, i.e., once chromatin relaxation is fast compared to cohesin turnover

Left cartoon: equilibrated compartments at t=0 are disrupted/mixed/bridged some time later by a loop extruder. After a longer time, longer than the chromatin relaxation time, two scenarios are possible: top- cohesin residence is short, so extruder has already bound and re-bound elsewhere, causing new disruptions before chromatin polymer could relax. bottom- cohesin residence time is longer than relaxation time, so compartments requilibrate while the extruded loop remains, before new disruptions can occur.
Right cartoon: effect of velocity. Large velocities lead to faster rate of disruption of compartment contacts and larger loops (spatial extent of disruption). Small velocities lead to smaller rate (more quasistatic) of disruption of contacts and smaller loops.

120

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

Why does PDS5A/B depletion erase compartments as well as WAPL depletion despite differences in residence time? One possibility is that PDS5A/B depletion leads to faster average extrusion speed

Simulation phase diagram showing amount of compartment contrast as colored points on axes of extrusion velocity vs. residence time. Arrows indicate predicted parameter and resulting comp contrast changes with different depletions (W, AB, and WAB). Gray region indicates region of parameter space where increases in cohesin residence time increase compartment contrast. We predict PDS5A/B depletion increases cohesin speed because the depletion suppresses contrast as well as WAPL depletion despite a smaller increase in residence time.

110

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

How? In polymer sims increasing cohesin residence time can nonmonotonically alter chromatin compartmentalization. Consistently iFRAP showed residence times increase w/ WAPL & PDS5A/B depletion, dramatically so w/ triple depletion (also suggesting partially distinct PDS5 & WAPL-based cohesin release)

3 contact maps from simulations with residence times increasing from left to right and plot of compartment contrast at a particular genomic distance, s, for different residence times. In contact maps, checkerboard pattern is strong/sharp, then weaker/blurrier, then a bit stronger/sharper as residence time increases. Quantitation shows non-monotonic compartment contrast

Data from SCC1-eGFP iFRAP experiments of in control, PDS5A/B depletion, WAPL depletion, and WAPL-PDS5A/B triple depletion. Fluorescence decays rapidly (t1/2 < 20 min) in control cells, less rapidly for PDS5A/B depletion (t1/2 ~ 1 hr), even more slowly for WAPL depletion, and very slowly for triple depletion (t1/2 >> 2 hrs)

110

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

As seen previously, PDS5 or WAPL depletion suppress chromatin compartmentalization, but astonishingly, co-depletion of WAPL and PDS5A/B leads to recovery of compartments in near-cis!

Hi-C maps of 70 Mb or 15 Mb windows on chr 1 and Pearson correlation maps of 243 Mb chr 2 in control and depletion of CTCF, PDS5A/B, WAPL, WAPL-CTCF, and WAPL-PDS5A/B. Checkerboard patterns visible in control and CTCF erased in depletions of PDS5A/B and WAPL. Near-diagonal checkerboard patterns are recovered in WAPL-PDS5A/B triple depletion maps

110

Ed Banigan @irate-physicist.bsky.social · 03/09/2025

In auxin-inducible degron cells, Gordana used Hi-C to observe the loss of extrusion-mediated CTCF-CTCF contacts (corner peaks) with PDS5A/B depletion. Interestingly, while WAPL depletion increases corner peaks, co-depleting PDS5 completely suppresses them

Hi-C contact maps of control and cells depleted of CTCF or PDS5A/B or WAPL or co-depleted of WAPL+CTCF or WAPL+PDS5A/B. No corner peaks in CTCF depletion, few in PDS5A/B depletion, many and more distal peaks in WAPL depletion, none in WAPL-CTCF, none in WAPL-PDS5A/B

110

Ed Banigan @irate-physicist.bsky.social · 08/07/2025

Interestingly, the Hi-C expts display features of loop extrusion - dots of contact enrichment, stripes emanating from S compartment boundaries, and a characteristic extrusion shoulder in the contact frequency curve P(s) that is most notable in S comps

Left: Contact frequency curves for the whole chromosome, and A, B, and S comps, along with the log-derivative curves, showing an extrusion shoulder around 40-60 kb in S compartments. Right: a snippet of a contact map showing extrusion features — local contact enrichment along dots and lines that are especially visible in S compartments.

120

Ed Banigan @irate-physicist.bsky.social · 08/07/2025

But a cooperative combination of these two mechanisms can generate S compartment patterns! When loop extrusion is targeted to S compartments, we observe local contact enrichment in S, depletion of inter-S contacts, and sequestration from the rest of the chromosome

Two contact maps. Left: split map showing contacts and observed-over-expected for a simple compartment model in which loop extruders preferentially load within S compartments (indicated by purple on axes). All 3 features: i) on-diagonal contact enrichment, ii) off-diagonal inter-S depletion, and iii) depletion of contacts between S and entire chromosome are visible. Right: split map showing the agreement of the best simulation model of an experimentally observed compartment configuration and the corresponding Hi-C map from experiments.

110

Ed Banigan @irate-physicist.bsky.social · 08/07/2025

Ok it’s unusual but is this just microphase separation or loop extrusion? No!   Emily’s simulations show S compartments can't be generated by either of these mechanisms alone! Phase separation makes unwanted inter-S contacts & extrusion can't deplete inter-S contacts w/out erasing all comps

Hi-C maps from simple models of compartmentalization. Two maps in the top row show the result of weak and strong affinity-mediated microphase separation: on-diagonal S contacts are always correlated with off-diagonal inter-S compartment contacts. Two maps in the bottom row show that increasing the abundance of loop-extruding motors uniformly suppresses compartmentalization and does not give unique S compartment patterning.

110

Ed Banigan @irate-physicist.bsky.social · 08/07/2025

How are S compartments are strange? 3 key features: i) on-diagonal contacts enriched, ii) off-diagonal contact depletion bet/ S comps. (unlike usual compartments, not typical of TADs), iii) smooth depletion of S & all other parts of the chrm (extreme sequestration!)

A 7 Mb window of a Hi-C map of silkworm genome showing the newly observed “S compartments” as well as typical Hi-C features (checkerboard A/B compartment patterning and small on-diagonal TAD-like domains). Thin filled colored boxes along the axes mark A, B, and S compartments (red, green, purple). Boxes on the map outline 3 features of S compartments: i) on-diagonal contact enrichment, ii) off-diagonal contact depletion between S compartments, iii) smooth depletion of S and all other parts of the chromosome.

110

Ed Banigan @irate-physicist.bsky.social · 17/02/2025

These physical constraints have implications for force transmission into the nucleus. Upon deformation, tethered & crosslinked peripheral increases contacts with the lamina heterochromatin and reorganizes far more than unconstrained chromatin.

A+B) Plots indicating regions of chromatin fiber that are in contact with the lamina before nuclear deformation, and after. After deformation, chromatin contacts with the lamina increase, with the largest increase coming for nuclei with tethering of chromatin to the lamina. B) also shows illustration of chromatin in contact or not with the lamina, determined by proximity/spatial cutoff. F) Change in genomic contacts of heterochromatin as a function of genomic distance. With tethering and crosslinking, heterochromatin alters contacts across all genomic distances (more than nucleus without crosslinking & tethering). Less alteration within chromatin without crosslinking, but still elevated alterations at sub-Mb scales. G) With tethering and crosslinking, interior uncrosslinked euchromatin perturbed a little more across all scales. Without crosslinking, relatively little rearrangement of euchromatin contacts, as in untethered case.

100

Ed Banigan @irate-physicist.bsky.social · 17/02/2025

All changes w/ peripheral heterochromatin tethered to the lamina w/ specific bonds. These tethers stiffen nuclei. More important, now, increasing phase-separated heterochromatin & crosslinked heterochromatin can increase nuclear stiffness. Lamina tethers are physical constraints on chromatin!

$A) an illustration of heterochromatin subunits connected by springs to the nuclear lamina to model tethering. D) A scatter plot showing that nuclear stiffness increases as the per subunit probability of heterochromatin tethering to the lamina increases. E) A bar plot showing that total heterochromatin level alters nuclear stiffness differently now. If total number of tethers is unchanged, heterochromatin level does not significantly impact nuclear stiffness. If number of tethers can change with total amount of heterochromatin, increasing heterochromatin can increase nuclear stiffness. 3 snapshots illustrate increases in peripheral localization of heterochromatin as heterochromatin fraction increases. I) Two bar plots. With peripheral tethering combined with intra-heterochromatin crosslinking, increasing heterochromatin fraction increases nuclear stiffness. This happens if total number of crosslinks increases with total heterochromatin (left) or if amount of crosslinks remains constant but tethering increases (right).$

110

Ed Banigan @irate-physicist.bsky.social · 17/02/2025

But w/ this chromatin model — widely used in 3D genome modeling — increasing heterochromatin *decreases* stiffness rather than increasing it as in expts. Why? Chromatin condensation reduces polymer osmotic pressure. Stiffening heterochromatin by crosslinking doesnt help either! What’s going on?

$A) Force-strain curves showing soft/shallow force response for small deformations (<30% strain) and stiffer response for larger deformations. But as fraction of chromatin that is heterochromatic increases, the curves show slightly softer force response. C) Bar plots quantify decreasing effective spring constant at 30% strain as heterochromatin is increased. Bottom: an illustration of springs bridging/crosslinking heterochromatin subunits. A scatter plot showing increasing crosslinking probability has no effect on nuclear stiffness. A bar plot showing that increasing amount of heterochromatin with crosslinking also decreases nuclear stiffness.$

101

Ed Banigan @irate-physicist.bsky.social · 17/02/2025

We first simulate this model with just heterochromatin (nonspecifically) adsorbed to the lamina and phase-separated from euchromatin.

An illustration of the polymer simulation model and Stephens et al.'s experiments. Springs representing micropipettes pull at opposite ends of a polymer shell filled with chromatin (gray lamina, blue constitutive heterochromatin, red facultative heterochromatin coating the interior periphery, green euchromatin interior). Simulated nucleus is deformed in response to force. Experimental schematic shows similar. An isolated nucleus is stretched by two micropipettes.

100

Ed Banigan @irate-physicist.bsky.social · 08/11/2024

Our polymer sim model suggests 3 factors govern microcompartments: 1) affinity interactions 2) loop extrusion 3) chromatin volume density/compaction Density seems to be particularly important & changes ~2-fold in M-to-G1 transition! But each factor seems to contribute to the observed dynamics

Adaptation of Fig 5. a) Illustration of model: extrusion by condensin I&II in mitosis + cohesin in interphase, affinity interactions of microcompartments + A/B compartments, polymer confinement to control density. e) Snapshots of contact maps and polymer configs from simulations

010

Ed Banigan @irate-physicist.bsky.social · 08/11/2024

Surprisingly, microcompartment strengths appear to peak during anaphase/telophase and then gradually fade during interphase. This sharply contrasts with observations of larger A/B compartments!

Figure 3. Zoom-in on example microcompartments; pile-ups of promoter-promoter, E-P, and CTCF loops; and quantifications of microcompartment/loop strengths

110

Ed Banigan @irate-physicist.bsky.social · 08/11/2024

Hi-C expts show chromosomes lose well-defined contacts during mitosis—compartments, TADs, etc…but do they really? Using region capture Micro-C to deeply resolve we find chromosomes form microcompartments of cis-regulatory elements even during mitosis! They appear as a “grid of dots” in contact maps

Adaptation of Figure 2. Snapshots of RCMC contact maps of 2 Mb region throughout the mitosis-G1 transition show time evolution of microcompartmental contacts.

110