Thread by @SerenaNikZainal

PAPER OUT!! 🥳😁

Let me share a #tweetorial
about how a nerd moment 🤓when @foad_rouhani & I were PhD students >10yrs ago (!)
has (finally) arrived @NatureGenet
with implications in using #hiPSCs in clinical studies & research modelling

1/
nature.com/articles/s4158…

A 22 yr old patient provided us with 2 skin fibroblast-derived IPSCs (F-hiPSCs) and 2 blood-derived (B-hiPSCs).

🤓: Do they differ genetically if they have differing origins?

Yep. The mutational load in F-hiPSCs was way more than in B-hiPSCs. Largely due to UV.

2/

So what’s the prevalence of UV damage in F-hiPSCs in general? Is it seen in well-known F-hiPSC repositories?

Yep. 72% of F-hiPSCs in the largest UK repository HipSci have evidence of UV damage.
And yes we’re sure it’s UV – lots of CC>TT, transcriptional strand bias.

3/

We noted heterogeneity between sister F-hiPSC clones derived from the same reprogramming experiment & same set of fibroblasts.
Resulting in marked discrepancies in mutagenesis between different clones from the same patient. 😳

So choose your clone for your studies wisely.

4/

OK so lots of mutations & UV damage in F-hiPSCs.
Does it matter?

Well we didn’t find too many “driver” mutations in general because of a heavy mutation burden.

BUT we did find an enrichment of BCOR mutations, intriguingly more so in B-hiPSCs than F-hiPSCs.
26.9% of them!
5/

Where do they come from?
Are they in starting material before reprogramming?

No we don't see them in the starting fibroblasts of affected F-hiPSCs.

6/

How about blood cells b4 reprogramming?
BCOR is a well-known driver in haematological cancer (albeit not a major driver of clonal haematopoiesis, unless patients have aplastic anaemia).

Nevertheless, we looked & did not find BCOR muts in erythroblasts / germline samples.

7/

However, we did see BCOR mutations arise in culture.

At the bottom of this image, hiPSCs from patients did not have BCOR muts in the parental IPSC. But after propagation, BCOR muts developed in (some) daughter subclones...

8/

So we probably have selection for BCOR muts in some IPSCs. Is there a functional consequence?

Global transcriptomics show greatest variance in the dataset is conferred by carrying a BCOR variant!

Volcano plot shows significantly DE genes are implicated in neurogenesis.
9/

So, to check that out, we performed directed differentiation of 6 BCOR-mutated lines and 6 BCOR-wildtype lines independently in parallel, and performed RNA-seq across the time course when the lines were hiPSCs, day6 and day12 NSCs, and day 27 neurons.

10/

Over the time course, BCOR-mut lines became more discrepant from wt lines, much less efficient at differentiation into neuronal lineages with tendency towards mesodermal lineages.
Evidence that these BCOR muts could affect behaviour in disease modelling.
11/

BTW there’s a shedload of oxidative damage that occurs in culture but we (and others) have shown that before.

12/

Bottom-line: ALL of these lines PASSED today's QC checks for deposition into cell line facilities. Customary copy number screening using arrays.
It does not have the resolution to detect the extent of single nucleotide DNA damage that is present in these cells. 😕
13/

F-hiPSCs may have lots of mutations but maybe ok if it doesn’t hit an important gene.
B-hiPSCs may be less mutated but may or may not have selected-for mutations.

I guess we don’t know if we don’t look... 🫣

14/

Given that the costs of sequencing are dropping remarkably, it would seem reasonable (if not expected) that we should characterise the genomes of our cellular models if we are going to use them for clinical studies and/or disease modelling.

End of science!
15/

ENORMOUS THANKS to @xueqing_zou who contributed heaven & earth to this project💕
@danjgaffney PetrDanacek @imartincorena AndrewBassett #HipSci & many more.
Funders: @CR_UK @NIHRresearch @UKRMP_PSEC @The_MRC #JosefSteiner

And 2 all patients that contributed samples.
🙏

/END

BTW from a human perspective, over the timeline of this project we have also produced:
- three job moves between the first & last authors
- two NHS consultants
- one pandemic wedding
- and multiple babies 👶 (none from me😝)

💕💕

AHA! @CherifBadja IS on twitter!
Just found him.
This is our IPS experimental happy-chappy who did the validation work 🙏🥳