Thanks for pointing this out - #3 is correct but in #7 there was a mistake: correct version is
"7/Loop extrusion-driven, long-lived encounters are longer than random collisions (up to tens of seconds vs. few seconds at most) but are much shorter than CTCF-anchored loops (which last 10-30 MIN)."

Unfortunately we don't have any data or models to answer your question. If nuclear shape changes affect levels of cofactors that determine extrusion rates (NIPBL,PDS5A/B) this could in principle affect extrusion velocity, but I can't personally point you to any data that support this scenario

The 'encounter radius' is an arbitrary physical range, which we vary systematically - and show that its particular choice doesn't affect any of the conclusions.

As you will see in the manuscript, the only condition for two loci to be 'caught' in an extrusion-driven encounter is that cohesin starts extruding approx. midway between their genomic positions (plus of course that their genomic distance is not much larger than cohesin's processivity)

Thanks for the kind words Anders - indeed this is a very much overlooked issue that also affects interpretation of DNA FISH data. Even small amounts of error make determination of 'interactions' unreliable, as your lab also pointed out earlier e.g. in www.sciencedirect.com/science/arti...

We do not impose any specific range, but rather compute (or measure) any events where the two loci spend time within an arbitrary encounter range (R_e in the manuscript), either mediated by random fluctuations or by loop extrusion...

you guys are too fast :)

19/ Read about this and much more in our preprint: www.biorxiv.org/content/10.1... Big shoutout to @laureplantard.bsky.social and Laurent Gelman from the @fmiscience.bsky.social facility for imaging & advanced microscopy!

18/ At large genomic distance, long encounters (=the important ones) are only due to loop extrusion; but at short genomic distance they can also be due to random collisions. This is why enhancers are no sensitive to loss of extrusion when they are close, but sensitive when they are far!

17/ ... and nicely predicts @elphegenoralab.bsky.social @karissalhansen.bsky.social's data at the Car2 locus www.biorxiv.org/content/10.1..., using changes in cohesin occupancy and extrusion velocity measured with @gfudenberg.bsky.social in www.biorxiv.org/content/10.1...:

16/ Finally, this simple model also predicts and explains why, at least within simple genomic locations devoid of further confounding effects, depletion of cohesin or its loading and elongation factor NIPBL affects enhancer function at large, but nor short genomic distance...

15/ This happens because by selecting longer and longer encounters, one selects more and more loop-extrusion-driven events, the probability of which is exponential as a function of genomic distance!

14/ This simple model also predicts that transcription levels should decrease exponentially as a function of genomic distance between an enhancer and promoter, exactly as we verified using data from our previous work!

13/ Strikingly, this simple hypothesis predicts that: 1) average transcription levels should increase nonlinearly as a function of enhancer-promoter contact probabilities, as we and others previously observed: www.nature.com/articles/s41... and also work from e.g. Boettiger, de Laat, Wysocka labs

12/...so the probability that such events result in transcription should thus increase when increasing the time that an enhancer and promoter spend in physical proximity. We thus asked: What if only encounters whose duration exceeds an arbitrary cutoff time are productive for transcription?

11/ But why do we care about these longer-lived encounters? Because -although we don’t know exactly what happens there molecularly- it is likely that regulatory processes at an enhancer-promoter interface take some time to occur, and require multiple proteins to be present at the same time…