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!

thanks! reading your paper (I see now accepted, congrats) really inspired the approach that made sense of the physics for me

I didn’t really complete the thought earlier: these papers point toward some of the key principles governing compartmentalization-extrusion interplay and are definitely worth reading!

Postscript: for a long time I was very confused about the PDS5-WAPL depletion expts+sims & how extrusion alters compartments. I was fortunate to eventually stumble across @ranjithpa.bsky.social‬'s preprint www.biorxiv.org/content/10.1... & later this by Chan/Rubinstein www.pnas.org/doi/abs/10.1...

I also want to point out another interesting and relevant recent preprint by @elphegenoralab.bsky.social and @gfudenberg.bsky.social

Thanks to our other co-authors who had essential contributions: Ryotaro Kawasumi, Roman Stocsits, Wen Tang, Kota Nagasaka, Lorenzo Costantino, Ralf Jansen, Kouji Hirota, Dana Branzei

Summary
-PDS5 facilitates NIPBL dissociation from cohesin by invading NIPBL binding site
-PDS5 strengthns CTCF boundaries by limiting NIPBL-cohesin life
-PDS5 lowers avg extrusion speed/depletion ups processivity
-Compartmentalization governed by competition of extrusion &polymer relaxation dynamics

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)

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!

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

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.

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.

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

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

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)

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!

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