The brain may form from two independent cell lines: research changes understanding of early nervous system development

Scientists from Stanford University and other American research centers have presented data that challenges one of the fundamental models of brain development. According to the study, published September 18, 2026, in Nature Neuroscience, the forebrain, midbrain, and hindbrain may arise not from a single common neural precursor, but from two distinct cell populations that emerge in parallel during gastrulation.

The researchers named these populations the anterior neural ectoderm and the posterior neural ectoderm. The former gives rise primarily to the forebrain and midbrain, while the latter gives rise to the hindbrain. Furthermore, experiments show that these cells acquire their own "fate" extremely early and transition poorly from one developmental pathway to another.

A popular Stanford Medicine report described the result in a striking way: the human brain can be viewed as two ancient neural structures combined into a single structure. However, the scientific article formulates its conclusion more cautiously: the authors propose considering the brain a composite organ originating from two lineage-restricted progenitors—two types of precursor cells with limited fates. Therefore, it's not a matter of literally "two brains" in humans, but rather two distinct embryonic pathways for the construction of the modern brain.

The work is not only fundamentally significant. Using a new developmental map, the scientists were able to generate motor neurons from human pluripotent stem cells in specific hindbrain regions, which had previously been particularly difficult to reliably grow in the laboratory. Such models could potentially facilitate the study of diseases affecting the brainstem and its motor neurons, including spinal muscular atrophy and amyotrophic lateral sclerosis.

The main question of the study	What the experiments showed
Is there a single precursor to the entire brain?	The data support the two parallel predecessors model.
Anterior neural ectoderm	Gives rise to the forebrain and midbrain
Posterior neural ectoderm	Gives origin mainly to the hindbrain
Key markers	OTX2 and GBX2
When division occurs	Already during gastrulation
Animal model	Genetic tracing of mouse cells
Human model	Pluripotent stem cells in vitro
Additional result	Obtaining motor neurons from rhombomeres 5-6 of the hindbrain
Evolutionary scale	It is possible that the mechanism will be preserved for at least ~550 million years.

[1]

The old model assumed a single origin for the entire brain.

The traditional, simplified model of early development assumed that a relatively unified population of neural progenitor cells initially emerges, and then, later, external molecular signals gradually separate it into cells of the forebrain, midbrain, hindbrain, and spinal cord. In other words, regional identity was considered a relatively late acquisition.

However, the early nervous system develops extremely quickly. During gastrulation, the embryo transforms from a relatively simple structure into an organism with major body axes and germ layers. Part of the ectoderm becomes neural ectoderm, from which the nervous system subsequently develops. It was at this very early stage that Kyle Loh's team decided to explore the possible divergence of cell lineages.

In a study of mouse embryos, scientists discovered two spatially and molecularly distinct groups of SOX2-positive cells. One expressed Otx2, the other Gbx2. The authors designated them, respectively, as the anterior and posterior neural ectoderm. They arose almost in parallel and occupied positions corresponding to the future anterior-midbrain and posterior regions of the brain.

Crucially, this occurred even before the emergence of well-formed brain regions. Thus, the differences between the future forebrain and hindbrain may be established much earlier than the simple model of sequential regionalization of a single neural precursor would suggest.

Model	How is development expected?
Classic simplified	One common neural progenitor → subsequent division into regions
New model	Two parallel neural lines arise already during gastrulation
OTX2 line	Forebrain + midbrain
GBX2 Line	hindbrain
The main difference	Regional fate is determined much earlier

[2]

Cell labels showed that the two lines hardly mix at all.

To test whether these cells truly have distinct fates, the researchers used genetic lineage tracing in mice. For the posterior neural ectoderm, they used a system that allows them to briefly label Gbx2-expressing cells with the fluorescent protein tdTomato and then observe the migration of their progeny as the embryo develops.

Labeled cells of the Gbx2-positive posterior neural ectoderm were subsequently detected in the hindbrain, but were virtually absent from the Otx2-positive forebrain or midbrain. This distribution persisted not only immediately after gastrulation but also at much later stages of embryonic development.

In an additional experiment, the scientists used the Sox2-CreER/Confetti system, which labels individual early neural cells with various fluorescent labels. A total of 494 cell clusters from 16 mouse embryos were analyzed. The clones almost always ended up in either the future forebrain/midbrain or hindbrain, supporting the existence of an early cell lineage divergence.

Lineage tracing alone reveals what happens to cells in a normal embryo, but it doesn't prove that cells are physically incapable of changing their fate. Therefore, the authors supplemented their experiments with a human pluripotent stem cell model, in which they could deliberately expose the anterior and posterior progenitors to "incorrect" signals and test whether their identity could be switched.

Method	Task
Gbx2-CreER + tdTomato	Trace the descendants of the posterior neural ectoderm
Sox2-CreER + Confetti	Tracing the fate of individual early neural cells
Analyzed	494 cell clusters from 16 embryos
Observation	The offspring remained predominantly in their anterior or posterior region
Additional verification	Altering human stem cell differentiation signals

[3]

Human stem cells help reproduce brain division in a petri dish

The next step was to simulate the first stages of human development. Pluripotent stem cells were first transformed into early ectoderm, after which, using combinations of signaling molecules, the researchers obtained anterior or posterior neural ectoderm. Within approximately two days of differentiation, cells with distinctly different molecular characteristics emerged.

To generate anterior neural ectoderm, a combination of signals suppressing several key developmental pathways was used. To generate GBX2-positive posterior neural ectoderm, the researchers discovered a particularly important combination of FGF and retinoic acid activation with simultaneous suppression of BMP, TGF-β, and WNT signaling. FGF or retinoic acid alone were insufficient for the full formation of posterior identity.

This result was unexpected given some previous laboratory protocols. Strong WNT activation has often been used to "posteriorize" neural cells, but the new work shows that such a signal may direct cells further posteriorly—toward the spinal cord or neuromesodermal progenitors—rather than shaping the desired early hindbrain program.

The anterior and posterior neural cells were then deliberately given signals that would normally cause them to adopt a different regional fate. However, the anterior neural ectoderm retained a preference for the forebrain/midbrain, while the posterior neural ectoderm favored the hindbrain. The authors therefore believe that by this point, the cells have already become lineage-committed, meaning their potential range of further development is significantly limited.

What signals direct early neural ectoderm?

cell line	Main future area	Characteristic signs
Anterior neural ectoderm	Forebrain/midbrain	OTX2
Posterior neural ectoderm	hindbrain	GBX2
For the formation of pNE	FGF + retinoid signal	Together they are necessary for the formation of the rear identity.
BMP/TGF-β	Were suppressed	Facilitate neural pathway selection
WNT in early pNE	Suppressed	Strong activation could shift cells toward a more caudal, spinal program

[4]

The "memory" of the future brain is already recorded in the chromatin structure.

The genetic fate of a cell is determined not only by which genes are active at a given moment. The state of chromatin—a complex of DNA and proteins that determines which regions of the genome are accessible to cellular machinery and which remain relatively closed—is also crucial. Two cells can contain virtually the same DNA sequence but use completely different sets of genes due to different chromatin organization.

The authors examined chromatin accessibility in the anterior and posterior neural ectoderm using OmniATAC-seq and compared the data with gene expression profiles. Even at an extremely early stage, numerous differences in the accessibility of genomic regulatory regions existed between the two cell types.

Moreover, these differences predicted the future fate of the cells. In the anterior neural ectoderm, regulatory regions associated with forebrain formation were more accessible, while the posterior cells exhibited a chromatin architecture consistent with subsequent hindbrain development. The binding motifs of transcription factors also differed.

Thus, the separation of the two lineages turned out to be deeper than simply the temporary activation of the OTX2 and GBX2 marker genes. The cells were already epigenetically prepared for different developmental pathways. The authors compare this to two parallel pathways: external signals can regulate movement within each pathway, but switching from one to the other becomes significantly more difficult.

Level of differences	Anterior and posterior neural ectoderm
Marker genes	OTX2 vs. GBX2
Gene expression	Different transcriptional profiles
Chromatin	Different accessibility of regulatory sites
Transcription factors	Various enriched motifs
Reaction to signals	Limited ability to transition to the opposite fate
Result	Regional identity is established very early

[5]

A new scheme has made it possible to reliably obtain specific neurons in the hindbrain for the first time.

The practical value of the discovery became particularly apparent when the researchers attempted to use the discovered route to obtain mature human neurons. Previously, scientists had been relatively successful in creating many types of forebrain and midbrain neurons in culture, but some hindbrain cells were poorly produced or had mixed regional identities.

The authors hypothesized that the problem might lie at the very beginning. If a researcher first forms an anterior neural progenitor and then attempts to convert it into a posterior one with late signals, the cell might already be too tightly locked into an anterior program. Therefore, the team began directly with the posterior neural ectoderm.

Thus, the scientists succeeded in sequentially obtaining hindbrain cells and then motor neurons from rhombomeres 5 and 6. Rhombomeres are temporary segments of the embryonic hindbrain from which certain nuclei and neural circuits of the brainstem are formed. The resulting cells expressed characteristic regional proteins and demonstrated electrical action potentials, indicating functional maturation of the neurons.

These areas are particularly interesting for studying neurons that control facial and pharyngeal muscles, and other motor functions. The ability to obtain relatively well-defined human hindbrain cells in the laboratory could improve cell-based models of diseases in which the corresponding neurons are damaged or destroyed.

What was obtained from human stem cells?

Stage	The resulting cells
Pluripotent stem cell	Initial state
Early ectoderm	General intermediate stage
Posterior neural ectoderm	GBX2-positive hindbrain progenitor
Hindbrain progenitor	Regionally defined precursor
Rhombometers 5-6	A specific area of the hindbrain
The final result	Functional motor neurons
Function check	Generation of action potentials

[6]

Why the result is important for the study of ALS and spinal muscular atrophy

The hindbrain and brainstem structures provide functions essential to life: regulation of respiration, cardiac activity, sleep, and a number of other automatic processes. Its motor nuclei also participate in the control of swallowing, tongue, facial, and pharyngeal muscles.

Spinal muscular atrophy and amyotrophic lateral sclerosis can affect motor neurons, including those involved in swallowing and breathing. In the late stages, this becomes one of the most dangerous components of the disease. However, studying living human brainstem neurons directly in patients is virtually impossible.

Therefore, cell models based on pluripotent stem cells allow us to study the development of pathological processes in the laboratory: for example, comparing cells with and without a mutation, analyzing changes in gene expression, or testing potential drug treatments. The more accurately a laboratory cell corresponds to a real neuron in a specific brain region, the more informative such a model potentially becomes.

However, the current study does not represent a treatment for ALS or spinal muscular atrophy, nor does it demonstrate transplantation of the resulting neurons into patients. Its immediate achievement is a more precise way to generate specific hindbrain cells in vitro. Clinical applications, if they occur, will require significantly more research.

Possible application	What does the new job offer?
BASS	More accurate models of vulnerable hindbrain neurons
Spinal muscular atrophy	The possibility of studying motor neurons of human origin
Swallowing disorders	Model of the corresponding motor circuits
Development research	Possibility to study the front and back lines separately
Drug testing	Potential cellular platform
Treatment of patients	Not yet demonstrated

The division may be older than vertebrates

Another unusual result of the study concerns evolution. The authors examined early embryonic structures in several very distant animal groups and found evidence of a comparable anterior-posterior ectoderm division in mice, birds, fish, and hemichordates. Specifically, they analyzed the acorn worm, a marine animal extremely distant from humans in its external structure.

Hemichordates and vertebrates share a common ancestor that lived over half a billion years ago. Therefore, the similar organization of early anterior and posterior cell populations suggests that this separation principle arose approximately 550-600 million years ago, before the emergence of vertebrates and possibly even before the origin of chordates.

The authors propose an intriguing evolutionary interpretation: the modern forebrain and posterior brain regions may have evolved from ancient neural systems or developmental programs that were initially much more independent and later became spatially integrated during evolution. This hypothesis is based on comparative developmental biology, rather than direct observation of how two ancient organs literally "merged."

It is here that the popular "two-brain" concept requires the greatest caution. The study convincingly demonstrates two distinct early developmental lineages, but the question of whether they truly once existed as two functionally distinct organs in a common ancestor is a matter of evolutionary reconstruction. The authors themselves formulate this as a model and believe that the two lineages may have been evolutionarily conserved for approximately 550 million years.

Evolutionary data

Organism/group	Why is it important for research?
Mouse	Genetic tracing of cell lines
Human	Pluripotent stem cells
Chicken	Comparative analysis of embryonic development
Danio rerio	Checking the integrity of the mechanism in fish
Acorn worm	Representative of the hemichordates
Evolutionary interpretation	The division may exist for ≥550 million years

[7]

What the study actually proved—and what it hasn't yet proven

The most powerful part of the study was the genetic tracing of cells in mice. The researchers directly labeled early cells and observed where their descendants ended up. This approach provides significantly more reliable information about tissue origins than simply comparing the expression of a few genes. Additional experiments with stem cells supported the idea of early cell fate commitment.

However, direct lineage tracing of the developing human embryo was not conducted in the study. Conclusions about human development are based primarily on the differentiation of human pluripotent stem cells and the comparison of their molecular programs with embryonic models. Therefore, it remains to be determined how accurately the in vitro system reproduces all the events of early human brain development.

Furthermore, the inability to convert one type of precursor into another under the conditions used does not imply the absolute biological impossibility of such a transition. Theoretically, a different combination of factors or experimental intervention could be capable of reprogramming cells. The authors therefore speak of lineage restriction and commitment within the context of the data obtained, rather than irreversibility under all possible conditions.

Finally, Stanford University noted filed patent applications related to neural differentiation technologies, with several authors of the study listed as inventors. This does not refute the study's findings, but it is an important disclosure of potential conflicts of interest when evaluating future commercial applications of the technology.

Which is shown quite convincingly	What remains a hypothesis or requires confirmation
Mice have two early neural lineages.	The precise organization of these lines in vivo in humans
Gbx2 precursors give rise to the hindbrain	It is absolutely impossible to reprogram them.
Human PSCs can reproduce two programs	How closely does the culture replicate the embryo?
The two lines have different chromatin	All molecular causes of division
It is possible to obtain hindbrain motor neurons	Clinical efficacy of such cells
A similar principle is found in distant species.	The literal existence of two separate ancient "brains"

What does this job change?

Until now, the creation of a specific neuron from a stem cell has often been viewed as a matter of selecting the right factors for the final stage of differentiation. New research shows that errors can arise much earlier: if the cell has already passed through the "wrong" early neural progenitor, later exposure to the right factors may prove insufficient.

This makes early embryology of practical importance for cell biology. To reliably obtain a specific hindbrain cell type, researchers may need to first reproduce the posterior neural ectoderm, rather than creating a universal "neural precursor" and only then attempting to impart the desired regional identity.

At a fundamental level, the study offers a different picture of brain formation: its main anterior and posterior regions begin to diverge along their trajectories almost from the very beginning of neural development. Differences are visible simultaneously in cell fate, gene activity, and the structure of accessible chromatin.

Therefore, the main conclusion of the study is best formulated not as a sensational "humans have two brains," but more accurately: a single adult brain likely assembles from two extremely ancient, distinct cellular programs already in their early stages of development. If these results are confirmed by further research on human development, this model could become an important part of a new understanding of how the central nervous system develops.

News source

Jokhai RT, Dundes CE, Ahsan HS, Kang RS, Salomon-Shulman REA, Rajan A, et al. Two parallel neural ectoderm progenitors contribute to the developing brain. Nature Neuroscience. Published September 18, 2026. DOI: 10.1038/s41593-026-02433-7.

The study was peer-reviewed; the article was received by the editors on November 7, 2025, accepted on July 30, 2026, and published as a version of the record on September 18, 2026. The authors have deposited the raw RNA-seq, single-cell RNA-seq, and OmniATAC-seq data in the Gene Expression Omnibus, and made the analysis code available separately.