I like this paper a lot because it’s yet another good example of decoding the functions of non-coding disease variants, and what’s more, it’s a megabase-long looping enhancer!
I’ll never forget, from my college biology class, that crazy drawing of how enhancer elements might work: a piece of DNA, anywhere from introns to intergenic deserts, is bound by some transcription factors, and somehow forms a weird loop like alligator’s jaws all the way to the proximal promoter, where the core transcription regulation occurs.
Given this possible long-range cis regulation, it is not uncommon to see DNA in one gene regulates not this gene, but another gene/genes miles away. And this causes a lot confusions in interpretation of the functions of each of these regulatory sequences, especially when you try to study a disease-associated variant.
A classic example is one of the enhancers for Sonic Hedgehog gene. This locus was initially found to be linked with preaxial polydactyly, a limb developmental disorder in which patients have more than five finger digits. This piece of DNA resides in the intron of a then unannotated gene, and people just assumed this gene to be the culprit and named it lmbr1 for limb development membrane protein 1. It turned out later that this enhancer does not regulate lmbr1 expression, but the expression of shh, which is 1 mb away.
Well, same here in this current study. The authors, by using chromosome capture techniques, found that an obesity-associated locus in the intron of FTO gene actually regulates the expression of irx3 more than 1 mb away rather than that of FTO itself. They confirmed both the physical interaction as well as expression correlation.
This all sounds good. It seems that a new gene product causal for obesity has been identified. They went ahead and showed that irx3 KO mice are leaner and more resistant to high fat diet compared to their wt littermates (they probably should also test the heterozygous animals). All great and consistent, although it’s unknown whether FTO and IRX3 work in the same pathway in the regulation of body weight (fto mRNA level does not change in IRX3 KO in hypothalamus and adipose tissue, but this certainly does not mean the two proteins work independently).
At last they tried to claim IRX3 function in hypothalamus alone is sufficient to explain the phenotypes by expressing a dominant-negative IRX3 under the control of Ins2-cre. That’s just blatantly wrong. Ins2-cre utilizes the promoter region of insulin and its expression is clearly not restricted to hypothalamus. In fact, it’s predominantly expressed in pancreatic beta cells, at least. And beta cells clearly have a role in regulating blood sugar and body weight and obesity. It’s just too bold a claim in Fig. 4 of this study. Since I can’t think on top my head a specific pan-hypothalamic cre line, hypothalamic injection of Ins2-cre in conjunction with the conditional DN IRX3 allele is going to be my first choice to address this question.