Mouse and Stem Cell Data Show the Brain Has Two Separate Origins

Textbooks describe the brain as a single organ unfolding from one progenitor cell type. A study from Kyle Loh's lab at Stanford, published in Nature Neuroscience on September 18, reports a different structure: two separate progenitor populations appear side by side during gastrulation, each already restricted to a different part of the brain before any brain region exists.

Two Marker Genes Split the Earliest Brain Progenitor in Two

Around embryonic day 7.5 in mice, the sheet of cells called neural ectoderm splits into two populations that never overlap. One expresses the gene Otx2 and is called anterior neural ectoderm. The other expresses Gbx2 and is called posterior neural ectoderm. According to the bioRxiv preprint version of the study, these two populations occupy the positions classical fate maps had already assigned to future forebrain/midbrain and hindbrain, respectively.

Marking a cell's future address is not the same as proving it cannot go elsewhere. To test whether the two populations were already locked in, first-author researchers Rayyan Jokhai and Carolyn Dundes used a short-pulse genetic label: a tamoxifen metabolite that marks Gbx2-expressing cells for only about 12 hours, avoiding the multi-day labeling window that had made earlier tracing experiments ambiguous. Cells labeled this way at E7.5 showed up exclusively in the hindbrain at E8.5, E9.5, and again at E18.5. None appeared in the Otx2-marked forebrain or midbrain at any of those stages.

The lab then rebuilt the same split using human pluripotent stem cells, exposing definitive ectoderm to combinations of BMP, FGF, TGFβ, WNT, and retinoic-acid signals. Posterior neural ectoderm required both FGF and retinoic acid activation, alongside WNT blockade, to form; anterior neural ectoderm formed under standard neural-induction conditions alone. When each cell type was then challenged with the signals that should have redirected it, forebrain-inducing signals failed to convert posterior neural ectoderm, and hindbrain-inducing signals failed to convert anterior neural ectoderm, even in mixed co-culture.

Clonal Tracing Shows the Split Holds by the End of Gastrulation

Pulse-labeling one gene at a time tests a hypothesis about that gene. To check the split more broadly, the team used a second method that ignores gene identity and instead tracks any Sox2-expressing neural ectoderm cell. A low dose of tamoxifen activated a four-color fluorescent reporter in scattered individual cells at E7.5, letting the researchers follow whichever colored clone each cell produced without pre-selecting for Otx2 or Gbx2 expression.

Across 16 embryos, the team recovered 494 distinct clonal clusters. Of those, 62.96% ended up entirely within the forebrain or midbrain, and 32.59% ended up entirely within the hindbrain, with the remainder split or ambiguous. Almost every clone stayed inside one compartment. That distribution, built directly from the paper's own count, is what the chart below shows.

Divergent Chromatin Locks Each Progenitor Into Its Fate

The cross-challenge failures raised a mechanistic question: what stops a posterior cell from simply switching on forebrain genes when given forebrain signals? The team's chromatin-accessibility sequencing (OmniATAC-seq) found that anterior and posterior neural ectoderm carry different sets of "open" regulatory regions from the moment they're first distinguishable. Anterior-specific open chromatin was enriched for OTX2 binding motifs. Posterior-specific open chromatin was enriched for motifs belonging to HOX genes, HNF1B, MAFB, and the downstream effectors of FGF and retinoic-acid signaling. When posterior cells were exposed to forebrain-inducing signals, their chromatin still moved toward a hindbrain-like configuration rather than opening forebrain genes.

That commitment let the researchers direct hPSC-derived posterior neural ectoderm, through ventral hindbrain progenitors, into motor neurons carrying the molecular signature of hindbrain rhombomeres 5 and 6, a subtype the field had struggled to produce. The resulting cells expressed acetylcholine-pathway genes and fired action potentials under current injection and optogenetic light stimulation. Motor neurons from this hindbrain segment control cranial nerve IX and swallowing muscles, which the authors note is relevant to studying motor-neuron loss in spinal muscular atrophy and ALS, where impaired swallowing can lead to choking or aspiration.

Kyle Loh, the study's senior author, put the practical result plainly in a Stanford Medicine release: "Our discovery means that we can now grow neurons from the back of the brain, the hindbrain, in a petri dish and study their functions."

What the Study Does Not Rule Out

The paper's own discussion section names a gap the data cannot close: a shared "pan-brain" progenitor could still exist for a brief window between the emergence of ectoderm around E6.75-E7.0 and the anterior/posterior split at E7.5. The lineage-commitment tests also rest on mouse genetics and human stem-cell differentiation rather than direct transplantation experiments inside a living embryo, which the authors describe as technically difficult in mammals.

The cross-species survey adds a separate line of support without closing that specific gap. Mutually exclusive Otx-positive and Gbx-positive neural ectoderm populations turned up in gastrulating macaque, chicken, and zebrafish embryos, and a comparable split appeared in the acorn worm, a hemichordate that shared a common ancestor with vertebrates roughly 550 million years ago. If the split is that old, it suggests the two-progenitor arrangement was already in place before the vertebrate brain itself existed in its current form, which the authors present as a hypothesis for further testing rather than a settled evolutionary claim.