1/ Gather round children, I've a story to tell you. It's about how rare variants in the DNMT3A gene cause high blood pressure. In humans. The result is significant at p = 0.0000000062.

The preprint is here: https://www.medrxiv.org/content/10.1101/2021.02.10.21251503v2

2/ The story is a cool mix of genetics and epigenetics but it's quite long so make sure you are sitting comfortably. If you stick through to the end I'll tell you about a few other neat genes as well.

3/ The results come from 200,000 exome sequenced UK Biobank participants using weighted burden analysis, which just means we look at the burden of rare variants in each gene which each person has, but we give more weight to more severe variants.

4/ People often only consider the most severe variants, which cause complete loss of function of the gene, but here we include less severe ones as well, just with smaller weights. This gives us more information and hopefully more power to detect interesting things.

5/ We analyse over 20,000 genes in 66,000 cases with high blood pressure (hypertension) against 134,000 controls. One gene, DNMT3A, stands out as having more of these rare, damaging variants in cases with p value less than 0.00000001.

6/ Even after correcting for testing 20,000 genes the p value is still 0.00016 so we're pretty sure this result is real. Looking in more detail, around 1 person in a thousand has a variant which completely stops this gene working. This about doubles the risk of hypertension.

7/ Another 1 in a thousand people have a variant which means the protein coded for may not work properly and this moderately increases risk.

8/ So, why do variants in DNMT3A increase risk of hypertension?The protein it codes for is a DNA methyl transferase, which means it adds methyl groups to cytosine bases in other genes and this methylation reduces the expression of those other genes.

9/ So if DNMT3A is damaged it won't methylate those other genes properly and they will be over-expressed. Cool thing number one. This has already been shown in mice.

10/ Mice with knockdown of Dnmt3a (the mouse version of the gene) have reduced methylation of the gene for an angiotensin receptor, increased expression of that gene and get hypertension. (Angiotensin is a hormone which raises blood pressure.) But it gets better...

11/ If pregnant mice are fed a low protein diet or are given dexamethasone to simulate a stressful environment, their offspring have reduced Dnmt3a expression, increased angiotensin receptor expression and hypertension. Epigenetics in action! But it gets even better....

12/ The DNA for genes is wrapped round proteins called histones and the tails of these histones poke out and can themselves be methylated to give molecular signals to control the expression of those genes.

13/ There are 4 histones and the fourth amino acid in histone 3 is a lysine, symbolised with a K. This is referred to as H3K4 and it can be methylated or demethylated. Bear with me, it's worth it...

14/ So, DNMT3A methylates DNA but it will only do it if the H3K4 residue of the histone the DNA is wrapped around is not methylated. So methylating H3K4 stops DNMT3A working and demethylating H3K4 allows DNMT3A to function and methylate other genes, controlling their expression.

15/ Nearly there now. So what demethylates H3K4? A lysine demethylase coded for by a gene called KDM1A. And variants in this gene are associated with hypertension in humans while knockdown causes hypertension in mice.

16/ So now it all comes together. If the lysine demethylase does not work properly, H3K4 remains methylated so DNMT3A does not function, its target genes get overexpressed and the risk of hypertension increases.

17/ Looking at it another way, DNMT3A function can be impaired in any of three different ways: damaging genetic variants in the DNMT3A gene; lysine demethylase knockdown; hostile intra-uterine environment. All result in increased hypertension risk.

18/ Well I think that's a cool story. As a reward for sticking with me this far, here are some other results which are nice to know and way simpler to understand. I admit they're only significant p < 0.001 before correction but you'll still like them, I promise.

19/ Loss of function variants in either of the genes coding for subunits of soluble guanylate cyclase double hypertension risk. Soluble guanylate cyclase is responsible for detecting nitric oxide signalling to produce a hypotensive response so this makes total sense.

20/ Plus animal studies.

21/ The role for NPR1 variants in increasing hypertension risk is confirmed. These variants seem to exert their effect via the afore-mentioned guanylate cyclase. See how it all comes together?

22/ Variants which damage the gene coding for an angiotensin II receptor are associated with reduced hypertension risk. Very hard not to believe this is real.

23/ Variants which damage the gene for an enzyme which converts angiotensin II to angiotensin-(1-7) are also associated with reduced hypertension risk. Not sure why this should be but seems worth mentioning.

24/ Finally, damaging variants in the gene for zyxin are associated with reduced risk. What does zyxin do? It senses when the cells in the blood vessel walls are stretched. Could easily be something to do with blood pressure, I would think.

25/ Well, there you go. Hope you found something in here of interest. I love exomes, me. /ends