In preprints: primed for patterning – how epiblast cells encode the neural map in motion

Understanding how the human brain is patterned during the earliest stages of development remains a major challenge in developmental and stem cell biology. Between gestational weeks 2 and 5, the neural plate emerges from the epiblast and undergoes complex spatial organisation to form the neural tube. This process is tightly orchestrated by gradients of key morphogens (including WNTs, SHH, FGFs and BMPs) that confer positional identity along both rostro-caudal (RC) and dorso-ventral (DV) axes. However, because this developmental window is currently inaccessible in human embryos, most of the insights come from animal models or oversimplified in vitro systems.In this preprint, Rathore and colleagues (Rathore et al., 2025 preprint) introduce a high-resolution, in vitro model of early human neural tube patterning using the MiSTR (microfluidic stem cell regionalisation) platform (Rifes et al., 2020; https://cells.ucsc.edu/?ds=neural-tube-organoids). By combining morphogen gradients with temporal single-cell and single-nucleus transcriptomics, they provide new insights into the timing, transcriptional regulation and lineage commitment of region-specific progenitors, revealing that spatial identity is specified even before neural fate commitment.The MiSTR platform enables the application of linear gradients of morphogens (WNT and SHH) across adherent stem cell monolayers, reducing variability while permitting reproducible spatial patterning. In their study, Rathore and colleagues used MiSTR to generate three axial configurations: dorsal RC, ventral RC and DV forebrain. They then sampled nine timepoints from day 0 to day 62 (over 200,000 single-cell and single-nucleus transcriptomes). The authors employed deep-learning-based label transfer and spatial transcriptomic mapping, aligning in vitro populations with reference datasets from human foetal brain and mouse gastrulation-stage embryos. These computational approaches enabled accurate reconstruction of patterning trajectories and region-specific programs across development.One of the notable findings of the study is that RC identity was established remarkably early, within 24-48 h of differentiation, which occurred before cells express canonical neural markers. Cells exposed to a WNT gradient bifurcated into rostral (OTX2+) and caudal (GBX2+) domains, while still expressing the pluripotency marker OCT4. When these cells were switched to opposite patterning conditions (day 2), their regional identity remained unchanged, indicating early fate commitment. This neural bifurcation model challenges the prevailing notion that spatial patterning follows neural fate acquisition, and suggests that axial specification occurs within a transcriptionally pluripotent state, likely priming cells for later lineage decisions.The authors then characterise the transcriptional profiles of pre-neuralised rostral and caudal cells. Rostral regions were defined by early expression of HESX1 and SHISA2, while caudal cells were defined by FST, HOXA1 and other caudal markers. These distinct gene expression pre-neural programs were validated using an independent induced pluripotent stem cell (iPSC) line (KOLF2.1J) and was further supported by mapping to mouse gastrulation-stage embryos. Together, these results suggest that transcriptional priming in the epiblast lays the foundation for subsequent neural fate commitment and regionalisation, which reinforces the concept that regional specification is an early, transcriptionally encoded event.In contrast to the rapid emergence of RC identity, ventralisation within the telencephalon occurred later. Although cells were exposed to ventralising agents SHH and purmorphamine from day 3 in the DV forebrain MiSTR model, ventral markers (such as NKX2.1 and FOXA2) were not detected until day 9. This delay contrasts with the ventral RC model, where ventral identity emerged by day 5. The data suggest that telencephalic cells have a restricted temporal window of competence to respond to ventral cues, possibly influenced by earlier WNT inhibition, highlighting the importance of developmental timing and the cell-intrinsic state in determining morphogen responsiveness.A key feature of neural tube patterning is the formation of secondary signalling centres, such as the midbrain-hindbrain boundary (MHB), which orchestrates further regionalisation. Rathore and colleagues examined the emergence of the MHB within MiSTR-patterned tissues, along with both 2D culture systems and spheroid models, and found that its formation depended not merely on the juxtaposition of OTX2+ and GBX2+ domains, as previously proposed (Hidalgo-Sanchez et al., 2022), but on precise WNT signalling during the epiblast stage. Neither WNT activation nor OTX2+/GBX2+ domain juxtaposition alone was sufficient to induce and maintain MHB identity – particularly the caudal MHB marked by FGF17. Rather, the precise temporal WNT activation during the epiblast stage triggers expression of key markers, such as PAX2, PAX5/8, EN1, WNT1 and FGF17. This temporal sensitivity was only revealed using the MiSTR platform, which allowed controlled morphogen exposure during early differentiation. In silico perturbation further identified PAX2 and MSX1 as crucial regulators, while GBX2 appeared to have minimal effect on MHB formation.Beyond basic developmental insight, the MiSTR system has translational relevance. Its ability to reproducibly generate region-specific neural progenitors could inform the optimisation of differentiation protocols for therapeutic applications. The authors note that MiSTR-based data have been used to identify surface markers, such as APCDD1, for ventral midbrain progenitors, which is relevant for Parkinson's disease cell therapies (Kirkeby et al., 2023). The precision, scalability and compatibility of this model with omics technologies make it a valuable platform for refining stem cell-based regenerative strategies.Rathore and colleagues present a detailed and systematic analysis of early human neural tube patterning, revealing that spatial identity is specified before neural fate induction and depends on early morphogen exposure during the epiblast stage. Leveraging the MiSTR platform, the authors provide a powerful model system with precise spatial control, temporal resolution and reproducibility, enabling in-depth exploration of human neural development during an otherwise inaccessible period.This work has broad implications for both developmental biology and regenerative medicine. By uncovering how early positional information is encoded, the study informs strategies for generating region-specific neuronal populations (such as midbrain dopaminergic progenitors) from human pluripotent stem cells. Moreover, the integration of deep learning-based label transfer, single-cell transcriptomics and in silico perturbation analyses demonstrates the value of combining experimental and computational tools to uncover developmental mechanisms and refine differentiation protocols.While comprehensive, there are some caveats. Ethical barriers prevent direct validation in human embryos, so conclusions rely on in vitro models and comparative datasets. MiSTR systems, although robust, isolate neural tissue from its embryonic context, omitting paracrine signals from adjacent mesoderm, notochord and surface ectoderm. Finally, while early patterning is well characterised, this study does not address long-term maturation, morphogenesis or functional integration of differentiated tissues. These considerations highlight opportunities for future refinement.These findings have implications beyond neural development. The principles of early transcriptional priming, temporal competence and organiser formation are likely applicable to other aspects of embryogenesis and tissue patterning. This raises broader questions: which extrinsic cues are truly necessary for neural patterning? Could similar gradient-based approaches improve organoid maturation? What other patterning events might be revealed by comparable in vitro systems? As stem cell models grow in sophistication, platforms such as MiSTR will become crucial for translating molecular mechanisms to developmental outcomes, bridging an important gap between experimental control and biological complexity.We thank the members of the Pellegrini lab for the helpful discussions.