New #paper out from my group, together with the Tomkova and Kriaucionis labs:
https://www.nature.com/articles/s41588-024-01945-x
In it, we show that epigenetic DNA methylation causes errors during DNA replication!
A thread:
[#]science #biology #epigenetics #cancer #genomics (boosts welcome!)
1/🧵
=> More informations about this toot | More toots from bensb@genomic.social
Cytosine methylation is an important epigenetic modification that happens mostly at CG dinucleotides (often abbreviated as "CpG"). It was noticed more than 40 years ago that in many mammalian genomes, there’s less CpG and more TpG than you’d expect by chance (in a classic paper from Adrian Bird: https://academic.oup.com/nar/article-abstract/8/7/1499/2359884)
2/🧵
=> More informations about this toot | More toots from bensb@genomic.social
This was attributed to a simple chemical reaction: "spontaneous oxidative deamination". When this happens to methyl-C, it produces T. The resulting T:G mis-pair is thought to be repaired by base-excision repair, but some T's are not repaired correctly. For decades, this has been the standard explanation for the high frequency of C>T mutations at methyl-cytosine (figure from https://aacrjournals.org/cancerres/article/55/17/3742/501331/Base-Excision-Repair-of-U-G-Mismatches-at-a)
3/🧵
=> More informations about this toot | More toots from bensb@genomic.social
C>T mutations at CpG are not just common in the germline, they are also the most common mutation type across cancer. The COSMIC database of unique "mutational signatures" in cancer calls them Signature 1, and shows how ubiquitous this process really is!
4/🧵
=> More informations about this toot | More toots from bensb@genomic.social
About 10 years ago, the first datasets of mismatch-repair deficient cancer samples were sequenced. Mismatch repair (MMR) is totally unrelated to base-excision repair. It mainly cleans up after the polymerase during DNA replication in preparation for cell division. MMR-deficient cancers have an incredibly high mutation rate. So, about 9 years ago, we asked a simple question: do the loss-of-MMR induced mutations totally outnumber the C>T mutations at methyl-C?
5/🧵
=> More informations about this toot | More toots from bensb@genomic.social
It turns out that the opposite is the case: MMR deficient cancers have more C>T mutations at CpG! Cancers that loose proofreading of the main replicative polymerase (Pol-ε) have an even higher rate of C>T at CpG. So how can this be?
6/🧵
=> More informations about this toot | More toots from bensb@genomic.social
After much wrangling of data and scratching of heads, we proposed that maybe DNA replication is itself mutagenic by occasionally mis-reading methyl-C as T, putting an A opposite it. Normally, this is repaired by proofreading or MMR, but in their absence you see a massive increase in these replication-errors.
7/🧵
=> More informations about this toot | More toots from bensb@genomic.social
As is often the case, nobody wanted to believe such a crazy hypothesis that completely overturns a paradigm that's printed in every molecular biology textbook, at least not based on just some re-analysis of public cancer genomics data. We published our results in DNA Repair, but few people ever read that paper.
https://doi.org/10.1016/j.dnarep.2017.11.005
8/🧵
=> More informations about this toot | More toots from bensb@genomic.social
Fortunately, Skirmantas Kriaucionis's group, and especially his amazing postdoc Michael McClellan, came up with an idea of how to experimentally test our hypothesis!
9/🧵
=> More informations about this toot | More toots from bensb@genomic.social
Michael set up an assay that he calls Polymerase Error Rate sequencing (PER-seq): a known region of interest (ROI) is made single-stranded, then filled in by reconstituted human Pol-ε in a cell-free setting, so that no repair or other cellular processes can interfere. Through a combination of cleanup steps, double-barcoding of single molecules and a lot of clever computational filtering, we are able to detect replication errors in single molecules!
10/🧵
=> More informations about this toot | More toots from bensb@genomic.social
By either methylating or not methylating the ROI, we can then test if Pol-ε makes more errors when copying methyl-C than C.
If you've read this far, the big reveal is not going to be too surprising: Pol-ε, even with proofreading intact, often puts an A opposite methyl-C!
11/🧵
=> More informations about this toot | More toots from bensb@genomic.social
To give some perspective: from conception to publication, this project spanned over 9 years! Marketa Tomkova, who made the initial observation, was a PhD student in my group when we started. She now leads her own group. Sometimes, you have to be very patient in science, and lucky to have enough funding from different sources to finish a project.
12/🧵
=> More informations about this toot | More toots from bensb@genomic.social
This paper wouldn't exist without the freely available mutation data from MMR deficient patients and various other open datasets. "Open Data" matters. Lastly: we could never have proven the point without working very closely with experimental scientist. This was a real team effort, everyone bringing their insights, tools and skills to bear on the problem.
Anyway, I'm incredibly proud of this paper, and so happy to finally see it out in the world.
End/🧵
=> More informations about this toot | More toots from bensb@genomic.social
@bensb This is very cool, and is something I had wondered about during my PhD when I was working with DNA methylation data (but not the mechanism you described because I didn't know enough bio at the time).
I've bookmarked your paper to read later. I'm really interested in seeing your experiments and PER-seq method
=> More informations about this toot | More toots from jrhawley@scholar.social
@jrhawley yeah, I think the observation as such was around for a while, but many people tried to squeeze it into the prevailing paradigm, so they came up with explanations like "MMR must be involved in repair of deamination" and/or "pol ε mutations are gain-of-function and magically produce exactly this pattern, but not in the WT". Let me know what you make of the paper, always curious to get other people's opinion
=> More informations about this toot | More toots from bensb@genomic.social This content has been proxied by September (3851b).Proxy Information
text/gemini