MAP3K15 and type 2 diabetes
The story of MAP3K15 as a potential diabetes drug target, initially discovered in the UK Biobank, continues.
A new preprint by 23andme scientists reminded me of an interesting rare variant discovery from the UK Biobank. Let’s talk about MAP3K15.
When the UK Biobank whole exome sequencing was completed in 2020, the Pharma companies who funded the sequencing (and so had exclusive early access to the data) were in the race to discover all the low-hanging fruits (the associations that can be found just by analyzing the UK Biobank without adding more data), particularly the protective associations that point to potential drug targets.
One of the major diseases endpoints of interest was obviously type 2 diabetes (T2D), and I think, there are at least 4 to 5 different versions of exome-wide association studies (ExWAS) of T2D based on the UK Biobank published to date by different research groups (both industry and academic researchers).
One interesting discovery that came up in the T2D ExWAS studies was the protective association between MAP3K15 and T2D, reported by both Regeneron and AstraZeneca scientists independently as part of their flagship papers. AstraZeneca scientists discovered the association in their analysis of the early release of the UKB exomes, published in Nature in 2021. Regeneron scientists found the same in their analysis of the full release of the UKB exomes, published in Nature in 2021 (I wrote a detailed thread on this flagship paper). Later, AstraZeneca scientists published a separate paper on the rare variant associations of T2D in 2022 in which they focussed specially on MAP3K15.
So, what was the actual finding? The exome analyses revealed that individuals who carried rare deleterious coding variants (either predicted loss of function (pLOF) or deleterious missense variants) in MAP3K15 were at lower risk of having T2D. The findings were replicated in independent cohorts. Regeneron scientists replicated the findings in their in-house Geisinger Health System (GHS) dataset and AstraZeneca scientists replicated the findings in the Mexico City Prospective City (MCPS) and Finngen biobanks.
The most interesting aspect of the discovery is that MAP3K15 is an X-linked gene and the AstraZeneca scientists found that the association is strongest in the recessive model. Heterozygous females for MAP3K15 pLOFs showed an 18% reduction in T2D risk whereas hemizygous males (who are obligatory knockouts) showed a 35% reduction in T2D risk (homozygous females are rare to make meaningful estimates). The protective association was also observed for blood glucose and HbA1c levels.
So, what MAP3K15 does? Nothing much is known about this gene except that it encodes an enzyme protein (mitogen-activated kinase) that might be playing some key role in signal transduction related to insulin or glucose metabolism. The protein is expressed in many tissues including pancreatic alpha and beta cells. The mechanism of MAP3K15’s role in T2D is yet to be understood. But it’s an attractive drug target as the encoded protein is a kinase which makes it more amenable to targeting via small molecules (perhaps, that should explain AstraZeneca’s interest in this finding).
It looks like 23andMe also has developed an interest in MAP3K15 as a drug target. In a recent preprint, their scientists have specifically looked into the MAP3K15 association with T2D in the 23andMe massive database. Note that, the 23andMe genetic database is not based on sequencing but on genotyping. Yet, the massiveness of their sample size would mean that certain rare variants even when imputed with not a very high level of confidence would offer substantial power to make rare variant discoveries. Proving that point, the 23andMe scientists were able to replicate the MAP3K15 rare variant burden association with T2D using their genetic database. They even found a novel protective association with blood cholesterol levels (perhaps adding more clues to understanding the MAP3K15 mechanism of action).
What stood out for me is that 23andMe made special efforts to do a genotype-by-recall study. There is a pLOF (rs148312150) in the MAP3K15 that is most common in Europeans and 23andMe scientists identified nearly more than 5000 recessive carriers (hemizygous males mostly, but also a few homozygous females) and tried contacting them for more detailed analysis. Interestingly, the response rate was extremely low with only 71 participants (1.26%) answering research questions and giving blood samples. Expectedly, no major insights came out of the callback study. But the fact that 23andMe made this effort is itself very interesting, and their concluding line in the abstract “Our findings demonstrate the discovery power of the 23andMe database, including the feasibility of consented participant recruitment to inform therapeutic discovery and development” might be a hint that we might see more of drug target discovery and validation work from 23andMe in the future.
Anyway, MAP3K15 is an interesting, relatively new discovery where much of the story is yet to unfold. Given it’s a potential target, functional studies might swiftly follow. And if that happens, I’ll update you on the new developments.