Elphege Nora Lab at UCSF
@elphegenoralab.bsky.social
Our laboratory seeks to understand how chromosome structure relates to genome functions
(27) We could recreate CTCF insulator bypass at the Car2 locus by introducing SRR2
🤯
> How many TAD boundaries throughout the genome get bypassed by enhancers through this synergy mechanism? 👀 👀 👀
🤯
> How many TAD boundaries throughout the genome get bypassed by enhancers through this synergy mechanism? 👀 👀 👀
November 27, 2025 at 9:59 PM
(27) We could recreate CTCF insulator bypass at the Car2 locus by introducing SRR2
🤯
> How many TAD boundaries throughout the genome get bypassed by enhancers through this synergy mechanism? 👀 👀 👀
🤯
> How many TAD boundaries throughout the genome get bypassed by enhancers through this synergy mechanism? 👀 👀 👀
(25) Putting SRR2 Car2 makes the locus less reliant on cohesin loop extrusion
> so proximal and distal enhancers seem to synergize in a pretty degenerate way (read: it's not locus-specific, you can mix and match regulatory elements, to some extent)
> so proximal and distal enhancers seem to synergize in a pretty degenerate way (read: it's not locus-specific, you can mix and match regulatory elements, to some extent)
November 27, 2025 at 9:59 PM
(25) Putting SRR2 Car2 makes the locus less reliant on cohesin loop extrusion
> so proximal and distal enhancers seem to synergize in a pretty degenerate way (read: it's not locus-specific, you can mix and match regulatory elements, to some extent)
> so proximal and distal enhancers seem to synergize in a pretty degenerate way (read: it's not locus-specific, you can mix and match regulatory elements, to some extent)
(24) Using 4C we did not find evidence regulatory element synergy works by maintaining long-range chromosome folding independently of cohesin 😦
It also does not operate LDB1 here ❌
> molecular mechanism of enhancer synergy still to be investigated (get in touch!) 🔍
It also does not operate LDB1 here ❌
> molecular mechanism of enhancer synergy still to be investigated (get in touch!) 🔍
November 27, 2025 at 9:59 PM
(24) Using 4C we did not find evidence regulatory element synergy works by maintaining long-range chromosome folding independently of cohesin 😦
It also does not operate LDB1 here ❌
> molecular mechanism of enhancer synergy still to be investigated (get in touch!) 🔍
It also does not operate LDB1 here ❌
> molecular mechanism of enhancer synergy still to be investigated (get in touch!) 🔍
(23) If you move SRR2 further than ~20kb from the Sox2 promoter, it cannot support cohesin-independent action of the distal enhancer.
~20kb was the cuttof we say at Car2 for cohesin-independent action of enhancers. Makes sense!
~20kb was the cuttof we say at Car2 for cohesin-independent action of enhancers. Makes sense!
November 27, 2025 at 9:59 PM
(23) If you move SRR2 further than ~20kb from the Sox2 promoter, it cannot support cohesin-independent action of the distal enhancer.
~20kb was the cuttof we say at Car2 for cohesin-independent action of enhancers. Makes sense!
~20kb was the cuttof we say at Car2 for cohesin-independent action of enhancers. Makes sense!
(22) Strikingly, you can replace SRR2 with other weak enhancers to restore cohesin-independence at Sox2
> SRR2 is not special
> proximal and distal elements likely generally synergize in a cohesin-independent way
(thank you @chribue.bsky.social for highlighting the Map4k3 E2 enhancer!)
> SRR2 is not special
> proximal and distal elements likely generally synergize in a cohesin-independent way
(thank you @chribue.bsky.social for highlighting the Map4k3 E2 enhancer!)
November 27, 2025 at 9:59 PM
(22) Strikingly, you can replace SRR2 with other weak enhancers to restore cohesin-independence at Sox2
> SRR2 is not special
> proximal and distal elements likely generally synergize in a cohesin-independent way
(thank you @chribue.bsky.social for highlighting the Map4k3 E2 enhancer!)
> SRR2 is not special
> proximal and distal elements likely generally synergize in a cohesin-independent way
(thank you @chribue.bsky.social for highlighting the Map4k3 E2 enhancer!)
(17) So it’s not that complicated in the end:
If you rely on an enhancer, you will need cohesin extrusion beyond ~18kb.
UNLESS you have a promoter-proximal regulatory element that can synergize with your distal enhancer – which happens independently of
If you rely on an enhancer, you will need cohesin extrusion beyond ~18kb.
UNLESS you have a promoter-proximal regulatory element that can synergize with your distal enhancer – which happens independently of
November 27, 2025 at 9:59 PM
(17) So it’s not that complicated in the end:
If you rely on an enhancer, you will need cohesin extrusion beyond ~18kb.
UNLESS you have a promoter-proximal regulatory element that can synergize with your distal enhancer – which happens independently of
If you rely on an enhancer, you will need cohesin extrusion beyond ~18kb.
UNLESS you have a promoter-proximal regulatory element that can synergize with your distal enhancer – which happens independently of
(16) Why? Quite simply, you have now removed the two redundant axes that support long-range enhancer action at Sox2.
November 27, 2025 at 9:59 PM
(16) Why? Quite simply, you have now removed the two redundant axes that support long-range enhancer action at Sox2.
(15) That’s because SRR2 is there to support cohesin-independent regulatory communication:
Delete SRR2 when extrusion is blocked by the CTCF sites, and now they insulate very well.
Delete SRR2 when extrusion is blocked by the CTCF sites, and now they insulate very well.
November 27, 2025 at 9:59 PM
(15) That’s because SRR2 is there to support cohesin-independent regulatory communication:
Delete SRR2 when extrusion is blocked by the CTCF sites, and now they insulate very well.
Delete SRR2 when extrusion is blocked by the CTCF sites, and now they insulate very well.
(14) Others already noticed that distal enhancers can still activates Sox2 even if you put really strong CTCF sites in between.
We see that’s true even for sites that insulate *very well* at the Car2 locus.
Why is CTCF insulation not working well at Sox2?
We see that’s true even for sites that insulate *very well* at the Car2 locus.
Why is CTCF insulation not working well at Sox2?
November 27, 2025 at 9:59 PM
(14) Others already noticed that distal enhancers can still activates Sox2 even if you put really strong CTCF sites in between.
We see that’s true even for sites that insulate *very well* at the Car2 locus.
Why is CTCF insulation not working well at Sox2?
We see that’s true even for sites that insulate *very well* at the Car2 locus.
Why is CTCF insulation not working well at Sox2?
(12) So SRR2 mediates a cohesin-independent mechanism to support the communication of Sox2 with its distal enhancer 100kb away.
For Sox2, this regulatory axis and loop extrusion are redundant: that’s why you need to disable both to see a transcriptional effect.
For Sox2, this regulatory axis and loop extrusion are redundant: that’s why you need to disable both to see a transcriptional effect.
November 27, 2025 at 9:59 PM
(12) So SRR2 mediates a cohesin-independent mechanism to support the communication of Sox2 with its distal enhancer 100kb away.
For Sox2, this regulatory axis and loop extrusion are redundant: that’s why you need to disable both to see a transcriptional effect.
For Sox2, this regulatory axis and loop extrusion are redundant: that’s why you need to disable both to see a transcriptional effect.
(10) Wait. How can Sox2 work without extrusion? Its enhancer is 100kb away – that’s very far.
Loooots of detective work later 🕵️♀️ the answer is crystal clear: it’s all about an inconspicuous genetic element 3kb downstream of the promotor called SRR2.😲
Loooots of detective work later 🕵️♀️ the answer is crystal clear: it’s all about an inconspicuous genetic element 3kb downstream of the promotor called SRR2.😲
November 27, 2025 at 9:59 PM
(10) Wait. How can Sox2 work without extrusion? Its enhancer is 100kb away – that’s very far.
Loooots of detective work later 🕵️♀️ the answer is crystal clear: it’s all about an inconspicuous genetic element 3kb downstream of the promotor called SRR2.😲
Loooots of detective work later 🕵️♀️ the answer is crystal clear: it’s all about an inconspicuous genetic element 3kb downstream of the promotor called SRR2.😲
(9) What is we drive Car2 with a crazy strong enhancer?
Let’s grab the Sox2 super-enhancer, since we know it does not really need cohesin to work at Sox2.
Same cut off: 18kb.
So your host locus, not your enhancer, decides if you need cohesin or not for distal regulation.
Let’s grab the Sox2 super-enhancer, since we know it does not really need cohesin to work at Sox2.
Same cut off: 18kb.
So your host locus, not your enhancer, decides if you need cohesin or not for distal regulation.
November 27, 2025 at 9:59 PM
(9) What is we drive Car2 with a crazy strong enhancer?
Let’s grab the Sox2 super-enhancer, since we know it does not really need cohesin to work at Sox2.
Same cut off: 18kb.
So your host locus, not your enhancer, decides if you need cohesin or not for distal regulation.
Let’s grab the Sox2 super-enhancer, since we know it does not really need cohesin to work at Sox2.
Same cut off: 18kb.
So your host locus, not your enhancer, decides if you need cohesin or not for distal regulation.
(8) By relocating enhancers closer and closer to the Car2 promoter we can render it completely resistant to inhibiting extrusion, although it is normally very dependent on extrusion
The cut off: 18kb. Above that the Car2 enhancers need extrusion to work.
The cut off: 18kb. Above that the Car2 enhancers need extrusion to work.
November 27, 2025 at 9:59 PM
(8) By relocating enhancers closer and closer to the Car2 promoter we can render it completely resistant to inhibiting extrusion, although it is normally very dependent on extrusion
The cut off: 18kb. Above that the Car2 enhancers need extrusion to work.
The cut off: 18kb. Above that the Car2 enhancers need extrusion to work.
(7) Do enhancers need cohesin to work then?
Some yes (e.g. at the Car2 locus)
Some not really (e.g. everyone’s favorite Sox2)
Why is that?
Some yes (e.g. at the Car2 locus)
Some not really (e.g. everyone’s favorite Sox2)
Why is that?
November 27, 2025 at 9:59 PM
(7) Do enhancers need cohesin to work then?
Some yes (e.g. at the Car2 locus)
Some not really (e.g. everyone’s favorite Sox2)
Why is that?
Some yes (e.g. at the Car2 locus)
Some not really (e.g. everyone’s favorite Sox2)
Why is that?
(6) Back to the gene loci that are dysregulated: why are some but not all genes messed up?
Smit Kadvani and
@gfudenberg.bsky.social
noticed that down-regulated genes lie in enhancer-rich chromosome neighborhoods
That makes sense with the tissue-specific dysregulations.
Smit Kadvani and
@gfudenberg.bsky.social
noticed that down-regulated genes lie in enhancer-rich chromosome neighborhoods
That makes sense with the tissue-specific dysregulations.
November 27, 2025 at 9:59 PM
(6) Back to the gene loci that are dysregulated: why are some but not all genes messed up?
Smit Kadvani and
@gfudenberg.bsky.social
noticed that down-regulated genes lie in enhancer-rich chromosome neighborhoods
That makes sense with the tissue-specific dysregulations.
Smit Kadvani and
@gfudenberg.bsky.social
noticed that down-regulated genes lie in enhancer-rich chromosome neighborhoods
That makes sense with the tissue-specific dysregulations.
(5) But can embryonic cell types properly emerge even when loop extrusion is prevented?
Unambiguous gastruloid experiments by
@bracciolilab.bsky.social
in
@dewitlab.bsky.social
: YES, they clearly can.
Even though morphogenesis, which is normally very stereotypical, is pretty screwed up.
Unambiguous gastruloid experiments by
@bracciolilab.bsky.social
in
@dewitlab.bsky.social
: YES, they clearly can.
Even though morphogenesis, which is normally very stereotypical, is pretty screwed up.
November 27, 2025 at 9:59 PM
(5) But can embryonic cell types properly emerge even when loop extrusion is prevented?
Unambiguous gastruloid experiments by
@bracciolilab.bsky.social
in
@dewitlab.bsky.social
: YES, they clearly can.
Even though morphogenesis, which is normally very stereotypical, is pretty screwed up.
Unambiguous gastruloid experiments by
@bracciolilab.bsky.social
in
@dewitlab.bsky.social
: YES, they clearly can.
Even though morphogenesis, which is normally very stereotypical, is pretty screwed up.
(4) What about inhibiting loop extrusion as cells are differentiating?
Lo(ooo)ts of experiments later:
transcriptional dynamics are largely preserved.
Lo(ooo)ts of experiments later:
transcriptional dynamics are largely preserved.
November 27, 2025 at 9:59 PM
(4) What about inhibiting loop extrusion as cells are differentiating?
Lo(ooo)ts of experiments later:
transcriptional dynamics are largely preserved.
Lo(ooo)ts of experiments later:
transcriptional dynamics are largely preserved.
(3) What happens to transcription after inhibiting loop extrusion for 1 day in ES Cells?
Many genes are dysregulated.
Same is true in Neural Progenitor Cells, but completely different genes are dysregulated.
> what make a given gene locus vulnerable is cell-type-specific.
Many genes are dysregulated.
Same is true in Neural Progenitor Cells, but completely different genes are dysregulated.
> what make a given gene locus vulnerable is cell-type-specific.
November 27, 2025 at 9:59 PM
(3) What happens to transcription after inhibiting loop extrusion for 1 day in ES Cells?
Many genes are dysregulated.
Same is true in Neural Progenitor Cells, but completely different genes are dysregulated.
> what make a given gene locus vulnerable is cell-type-specific.
Many genes are dysregulated.
Same is true in Neural Progenitor Cells, but completely different genes are dysregulated.
> what make a given gene locus vulnerable is cell-type-specific.
(2) Inducible degrons are great, but killing cohesin causes cell cycle fiasco before we can study enhancer dysfunction 😰
Instead, we realized that degrading NIPBL strongly inhibits loop extrusion (by lowering cohesin processivity) while avoiding cell cycle defects
Instead, we realized that degrading NIPBL strongly inhibits loop extrusion (by lowering cohesin processivity) while avoiding cell cycle defects
November 27, 2025 at 9:59 PM
(2) Inducible degrons are great, but killing cohesin causes cell cycle fiasco before we can study enhancer dysfunction 😰
Instead, we realized that degrading NIPBL strongly inhibits loop extrusion (by lowering cohesin processivity) while avoiding cell cycle defects
Instead, we realized that degrading NIPBL strongly inhibits loop extrusion (by lowering cohesin processivity) while avoiding cell cycle defects
That was >40 years ago!
So many of John's papers still read incredibly prescient to this day.
Fun fact, John is still around and scheduled to give his faculty talk to my department this Wednesday.
So many of John's papers still read incredibly prescient to this day.
Fun fact, John is still around and scheduled to give his faculty talk to my department this Wednesday.
September 21, 2025 at 10:59 PM
That was >40 years ago!
So many of John's papers still read incredibly prescient to this day.
Fun fact, John is still around and scheduled to give his faculty talk to my department this Wednesday.
So many of John's papers still read incredibly prescient to this day.
Fun fact, John is still around and scheduled to give his faculty talk to my department this Wednesday.
Not sure that's what you want but group of John Sedat was mapping polytene chromosome folding in the mid-80s.
www.nature.com/articles/308...
rupress.org/jcb/article/...
www.nature.com/articles/308...
rupress.org/jcb/article/...
September 21, 2025 at 10:59 PM
Not sure that's what you want but group of John Sedat was mapping polytene chromosome folding in the mid-80s.
www.nature.com/articles/308...
rupress.org/jcb/article/...
www.nature.com/articles/308...
rupress.org/jcb/article/...
29 / and this one illustrates the dramatic differences in loop dynamics after depleting
NIPBL (slow rate, normal lifetime)
WAPL (very high lifetime)
PDS5 (high lifetime, high rate)
NIPBL (slow rate, normal lifetime)
WAPL (very high lifetime)
PDS5 (high lifetime, high rate)
August 19, 2025 at 9:24 PM
29 / and this one illustrates the dramatic differences in loop dynamics after depleting
NIPBL (slow rate, normal lifetime)
WAPL (very high lifetime)
PDS5 (high lifetime, high rate)
NIPBL (slow rate, normal lifetime)
WAPL (very high lifetime)
PDS5 (high lifetime, high rate)
28/ and some movies!
This one illustrates the difference in loop dynamics between high lifetime v.s. high rate, at equal processivity.
LEFT - High lifetime: loops take time to grow but remain for very long
RIGHT - High rate: loops grow super fast but eventually dissociate to reform
This one illustrates the difference in loop dynamics between high lifetime v.s. high rate, at equal processivity.
LEFT - High lifetime: loops take time to grow but remain for very long
RIGHT - High rate: loops grow super fast but eventually dissociate to reform
August 19, 2025 at 9:21 PM
28/ and some movies!
This one illustrates the difference in loop dynamics between high lifetime v.s. high rate, at equal processivity.
LEFT - High lifetime: loops take time to grow but remain for very long
RIGHT - High rate: loops grow super fast but eventually dissociate to reform
This one illustrates the difference in loop dynamics between high lifetime v.s. high rate, at equal processivity.
LEFT - High lifetime: loops take time to grow but remain for very long
RIGHT - High rate: loops grow super fast but eventually dissociate to reform
25/ There is more in the manuscript 📖 - including
▶️ calibration of models to experiments
▶️ how to titrate & measure chromatin bound cohesin
▶️ how extrusion rate affect compartmentalization & transcription
▶️ speculations about cell types that change expression levels of cohesin co-factors 💭
▶️ calibration of models to experiments
▶️ how to titrate & measure chromatin bound cohesin
▶️ how extrusion rate affect compartmentalization & transcription
▶️ speculations about cell types that change expression levels of cohesin co-factors 💭
August 16, 2025 at 3:03 AM
25/ There is more in the manuscript 📖 - including
▶️ calibration of models to experiments
▶️ how to titrate & measure chromatin bound cohesin
▶️ how extrusion rate affect compartmentalization & transcription
▶️ speculations about cell types that change expression levels of cohesin co-factors 💭
▶️ calibration of models to experiments
▶️ how to titrate & measure chromatin bound cohesin
▶️ how extrusion rate affect compartmentalization & transcription
▶️ speculations about cell types that change expression levels of cohesin co-factors 💭
24/ On the other hand, quantitative titrations showed that extrusion rates scale with NIPBL dosage 📐
This is a whole new way to think about the molecular origin of NIPBL haploinsufficiency in Cornelia de Lange patients, who suffer from partial reduction of NIPBL levels.
This is a whole new way to think about the molecular origin of NIPBL haploinsufficiency in Cornelia de Lange patients, who suffer from partial reduction of NIPBL levels.
August 16, 2025 at 3:03 AM
24/ On the other hand, quantitative titrations showed that extrusion rates scale with NIPBL dosage 📐
This is a whole new way to think about the molecular origin of NIPBL haploinsufficiency in Cornelia de Lange patients, who suffer from partial reduction of NIPBL levels.
This is a whole new way to think about the molecular origin of NIPBL haploinsufficiency in Cornelia de Lange patients, who suffer from partial reduction of NIPBL levels.