...Gilles Crevel, Akbar Shahid, Nandini Mozumdar, @jakubtomek.bsky.social, Emelie Shepherd, and our funders including @ludwigcancer.bsky.social,
@wellcometrust.bsky.social, and BBSRC. 14/

Incredibly grateful I could work on this with this amazing team, who didn’t dismiss the crazy ideas of a then-young girl PhD student with computer science background – as what most people would probably do!
Michael McClellan, Skirmantas Kriaucionis, @dagams.bsky.social, Sue Cotterill, ... 13/

Which of the two mechanisms caused more of the mutations that we detect in cancer and healthy tissue? We don’t know – yet. Repair rates and other factors play a role in this. 12/

So, what is a bigger source of CpG>TpG mutations: 5mC deamination or replication errors? Comparison with published in vitro 5mC deamination rates suggests that the replication errors are much more frequent in replicating cells. Genomic analyses of patient data support this. 11/

Can we show this in cells? In mice? How do error rates and spectra of Pol ε P286R vs EXO-NULL vs WT differ? And are these errors because of 5-methylcytosine, or because of the CpG context? You can find the answers in the paper… 10/
www.nature.com/articles/s41...

Moreover, our in vitro measured errors of human DNA Pol ε show remarkable similarity to mutations of patients deficient in Pol ε repair, explaining (among others) the mutational signature 14. 9/

What did we find? That human DNA polymerase ε does indeed make frequent errors when replicating CpGs! Especially methylated ones. When not repaired, this causes CpG>TpG mutations. 8/

It took our team only about 6 years, one pandemic, couple of intercontinental moves, multiple births, lots of grey hair, and heroic effort of amazing Michael McCllelan in the lab! We needed to find a way to detect mismatches in single molecules, which is extremely hard. 7/

To test this, we needed to develop a new technique, Polymerase Error Sequencing (PER-seq), to measure errors made by a given polymerase. I vividly remember how someone said in that meeting “This should be fairly easy, we should have it done in three months…” 6/

Is it possible that replication errors frequently create CpG>TpG mutations? Independently of deamination!

We published the cancer patient data analysis supporting this crazy hypothesis in DNA Repair www.sciencedirect.com/science/arti...
5/

However, about 9 years ago, we noticed that patients deficient in repair of replication errors have a huge number of CpG>TpG mutations. How is this possible, if CpG>TpG were coming from spontaneous deamination? 🤔 4/

Cytosines in the CpG context are mostly in the form of 5-methylcytosine. This epigenetic modification is important for gene regulation, but it can also spontaneously deaminate, creating a thymine. This results in a CpG>TpG mutation. 3/

C to T mutations in CpGs are more than 10 times more frequent than any other mutation type. They are present in most of our cells, across different tissues, and contribute to cancer, genetic diseases, and evolution. It was thought to be pretty well understood what causes them. 2/

This is part of a joint effort to understand, predict, and prevent progression to advanced sAML together with @bethpsaila.bsky.social, Stefan Constantinescu and other LudwigOxford labs.