Thread by @G_Devailly

I've seen a lot of tweets about this preprint by Frod et al.:

"Frequent lack of repressive capacity of promoter DNA methylation identified through genome-wide epigenomic manipulation"
biorxiv.org/content/early/…

An interesting read! Thanks for the link. Thread 🔽

Frist, DISCLAIMER:
I have done my PhD on MBD2, one of the main reader of DNA methylation.
So it might be the affected human speaking, and not the detached, objective scientist.

I like very much the fact that this paper took a fail experiment (an attempt to specificaly hyper-methylate one promoter, that was not so specific as it ended hypermethylating ten thousand of regions!) and use it to assess some very fundamental questions about DNA methylation.

For some context, citing the introduction:
"[...] abolition of DNA methyltransferase activity through chemical inhibition or genetic disruption causes global demethylation and activates numerous genes (6, 7)."

Note that the author could have cited *MANY* more publications.

(Including one of mine 🤭: ncbi.nlm.nih.gov/pmc/articles/P…)
(showing that upregulated genes upon chemical inhibition of DNMT were methylated on their promoter in the first place)

The authors rightly note that:
"However, these observations are correlative and challenging to interpret because genome-wide demethylation could have widespread and complex downstream effects upon chromatin structure and gene expression."

I would like to note that genome-wide hypermethylations are challenging to interpret because they can have widespread and complex downstream effects, too.

They authors did induce a somewhat global hypermethylation:
"In contrast, global CG methylation levels were 5.2% higher in ZF-D3A +dox cells
(68.2%) compared to MCF7-control (63%)."

So it is quite complex to interpret. Notably, following the induction, the cells might lack sufficient DNA methylation readers?

While induced hypermethylations are numerous, they are (arguably) of limited strength:
+ 0.3 mCG/CG ratio (i.e. from 5% to 35% DNA methylation, or from 55% to 85%...) (fig 2B).

I wish one day we will be able to technically induce full methylation on targeted regions, i.e. from 5% to 95% DNA methylation...
That would make interpretations easier!

The authors brilliantly show that weak hypermethylations are (globally) not maintain when the induction has stopped...

Might strong hypermethylations induce more permanent changes? We will know another day.

Know, about figure 5 A and B:
Stronger gains of DNA methylation correlates with stronger loss of expression.

I found representation of figure 5B quite misleading to be honest.

Here a more honest representation of figure 5B IMHO:

Let me tweet that again:
"Frequent​ ​lack​ of​ ​repressive​ ​capacity​ ​of​ ​ promoter​ ​DNA​ ​methylation"
Seriously ?

I know that the author talk more about long term effect.
Here, the DNMT3A fragment might still be present, and mediate some of the effect, but still.

The authors show that hypermethyalations are quickly lossed when the induction stopped.

One might wonder what would have happen if the hypermethylations were not loss? Might the effect on gene transcription be more permanent if the DNA methylation was not loss?

All in all, great paper (I would have done the analysis a bit differently, but still!), overselled by some twitter user.

You can see the glass half empty as the authors have, but I tend to see it quite full.

(I mean, have you seen that ?)

End of thread.