A Rare Adult Hypothalamic State at the rs1421085 Ortholog

The obesity-associated non-coding variant rs1421085 has established regulatory effects in adipocyte and other contexts. But the molecular state connecting the exact homologous T>C substitution to increased Irx3 in the adult male posterior hypothalamus remains unresolved. The question is not whether the broader FTO–IRX regulatory landscape is functional. It is whether the exact variant-bearing interval participates in a rare adult hypothalamic state that can be measured independently of broad locus proxies.

The unresolved edge

Existing work supports a downstream adult neuronal chain after the exact homologous T>C edit: increased Irx3 in the posterior hypothalamus, altered activity of IRX3-positive neurons, and effects on feeding and body weight. What remains missing is the regulatory mediator between sequence and adult hypothalamic transcription.

Coordinate identity came first

CAUSAL-DNA resolves the published mouse edit sequence uniquely at mm10 chr8:91,374,372 (1-based; T>C). An independent assembly cross-check places the corresponding GRCm39 base at chr8:92,101,000, inside Fto.

This exact site is not the recently characterized Fto-Irx::hibE1 element. It lies approximately 13.9 kb upstream. That distinction prevents a neighboring functional enhancer from being mistaken for evidence about the SNP-bearing interval itself.

Rare accessibility across all eight intended hypothalamus samples

In GSE246791, a large adult mouse-brain single-nucleus ATAC atlas, the ortholog-containing 500-bp tile is absent from the thresholded union candidate-enhancer catalogue. Yet the raw tile matrix shows low-frequency accessibility in all eight intended male hypothalamus samples.

The correct description is therefore neither “strongly open” nor “closed.” It is a third regime:

Official annotations reveal neuronal structure

All 154 target-positive barcodes were joined to the atlas authors’ official metadata rather than re-annotated from sparse markers. The signal is not explained by the largest glial populations.

The most reproducible prioritized neuronal subclass is LHA-AHN-PVH Otp Trh Glut:

A second subclass, DMH-LHA Vgll2 Glut, shows a larger pooled enrichment (~5.67×) but appears in only 3/8 samples. It is therefore treated as a secondary signal rather than promoted because of effect size alone.

Large glial subclasses are depleted relative to the pooled target rate: Astro-NT is ~0.48× and Oligo ~0.34×. The target counts therefore do not simply follow population abundance.

An independent Multiome dataset adds sparse same-nucleus support

GSE226277 provides paired hypothalamic RNA and ATAC measurements in the same nuclei. Across four verified male wild-type pairs, 2/4 contain a filtered nucleus with an ATAC fragment overlapping the target interval and detectable Irx3 RNA. Only two such nuclei were observed in total.

Both rare locus+ / Irx3+ nuclei also contain detectable transcripts for Arid5b, Cux1, Tet1 and Kdm2b; one contains Mecp2. This is availability, not occupancy. The narrow value is orthogonal: the exact target interval can be accessible in an adult hypothalamic nucleus in which Irx3 is transcribed.

What changed in the hypothesis space

The data weaken two simple stories: that the exact site is effectively closed throughout adulthood, and that it behaves like a broad, constitutively active adult enhancer. The surviving high-information model is more conditional:

What this article does not show

The next falsifiable boundary

The next decisive question is exact-site rather than TAD-wide:

In parallel, exact-site ARID5B/CUX1 occupancy, methylation-state dependence and allele-specific accessibility remain load-bearing discriminators. A positive wild-type contact would strengthen a 3D-contact model but would still not demonstrate allele dependence.

Why this qualifies as a narrow computational discovery candidate

The novelty claim is deliberately smaller than “a new obesity mechanism.” Public adult hypothalamic data support rare accessibility specifically at the exact rs1421085 ortholog-containing interval, with reproducible official neuronal-subclass structure, while an independent paired Multiome dataset supplies sparse same-nucleus target-interval ATAC / Irx3 RNA support.

That is enough to narrow the unknown molecular state M. It is not enough to fill it.

The broad FTO–IRX landscape is known to be functional. The exact rs1421085 site remains an open causal edge — but it is now a smaller, more testable one.