Two parallel neural ectoderm progenitors contribute to the developing brain (Nature Neuroscience, 2026)

Jokhai, Dundes (co-first authors) and colleagues, senior author kyle-loh; Nature Neuroscience, published 18 September 2026, with stanford-university as lead institution and Caltech and UCSF as collaborators. Read via a Stanford Medicine press release (med.stanford.edu, 18 Sep 2026, writer Krista Conger) — the journal page redirects to a login wall, so this page holds the release-level account, not the paper's own numbers. An earlier version circulated on bioRxiv in July 2025 with the author order reversed (Dundes first).

What was done

The team went back to mouse embryos during gastrulation, the stage when the neural tube first forms, and tracked two populations of progenitor cells by which regulatory gene they switch on: Otx2 or Gbx2. They compared the chromatin state of each population — which stretches of DNA are open and available to be read — in the front (anterior) and back (posterior) of the developing neural tissue. Then they used what they learned to push human pluripotent stem cells down the Gbx2 path in a dish, aiming for a cell type — hindbrain motor neurons — that had resisted directed differentiation before.

What was found

• The brain has two progenitor lineages, not one. Cells marked by Otx2 give rise to the forebrain and midbrain; cells marked by Gbx2 give rise to the hindbrain (brain stem). The two populations never overlap or interconvert during development. This runs against the standard model, in which a single progenitor pool patterns the whole neural tube from front to back.
• The two lineages are locked in early, at the chromatin level. Otx2 and Gbx2 cells carry fundamentally different chromatin configurations from early on, which the authors read as fixing each lineage onto its own developmental path before any brain structure exists. kyle-loh's framing: the front of the brain "arises from a totally different progenitor cell than the back of the brain," and "evolution took two existing neural systems and pushed them together spatially."
• This explains a standing lab failure. Attempts to grow human hindbrain neurons from stem cells have largely failed for years. rayyan-jokhai's reading is that those attempts "likely tried to coax forebrain and midbrain progenitors into hindbrain cells" — starting from the wrong lineage entirely, which the two-origin model says cannot work.
• Working hindbrain motor neurons, produced from human stem cells. Starting from the Gbx2 lineage instead, the team produced human hindbrain motor neurons that fired action potentials and expressed proteins typical of real hindbrain segments — described as a first for this cell type from pluripotent stem cells.
• The split looks ancient. The same two-lineage pattern shows up in chickens, zebrafish and acorn worms, species whose common ancestor with humans is roughly 550 million years back. The authors take this as evidence the two systems predate the vertebrate brain and were later joined, not evolved together from one origin.

Why it might matter

Named applications are prospective, not demonstrated in this paper: hindbrain motor neurons are implicated in spinal muscular atrophy and ALS, so a reliable way to grow them could support disease modeling for both. The release also flags hindbrain hunger circuits as relevant to how drugs like semaglutide work, again as a stated possibility, not a result of this study.

What to hold loosely

This page rests on a press release and its quotes, not the Nature Neuroscience text — the journal copy is behind a login and was not reached. Sample sizes, statistical detail and the full chromatin data are therefore absent here. The finding itself is developmental biology in mouse embryos and stem-cell culture, not a claim about two independently functioning organs in a living adult human brain, despite the press release's headline framing.

Tier T1 — peer-reviewed Nature Neuroscience paper, three named Stanford authors, cross-checked against independent reporting (Stanford Report, ScienceDaily) that agrees on the mechanism and quotes; held through a secondary because the primary is paywalled, which is a reading limit, not a tier change.

Related

kyle-loh · carolyn-dundes · rayyan-jokhai · stanford-university · synthesis