The amniote primitive streak (PS), equivalent of the amphibian blastopore, forms at the midline of the embryo during gastrulation. It marks the location where epithelial cells of the epiblast undergo epithelium to mesenchyme conversion and ingress to form the mesoderm and endoderm. At the end of gastrulation, the PS has reached its maximum length and begins to regress, laying in its wake the tissues that will form the embryonic body. When regression is complete, the remnant of the PS morphs into a poorly organized mass of cells located at the posterior end of the embryo, the tail bud, which generates the posterior-most regions of the body (Stern, 2004). Fate mapping studies in avian and mouse embryos have shown that at the beginning of PS regression the mesodermal trunk progenitors are found in the epiblast along the PS (Kinder et al., 1999; Psychoyos and Stern, 1996; Schoenwolf et al., 1992; Smith et al., 1994; Spratt and Condon, 1947; Tam and Beddington, 1987; Wilson and Beddington, 1996). These studies demonstrated that the anteroposterior distribution of mesodermal progenitors in the PS epiblast reflects their future medio-lateral fate. Cells within the node generate the notochord, and the most anterior PS cells produce the paraxial mesoderm, while the territories of the intermediate, lateral plate and extraembryonic mesoderm lie in progressively more posterior regions of the PS. The recent discovery of a new population of axis progenitor stem cells, the neuromesodermal progenitors (NMPs) in mouse, has led to reconsider the PS role in amniote body formation (Tzouanacou et al., 2009). These bipotent stem cells generating large clones containing both neural and mesodermal derivatives along the forming body axis were identified by a retrospective strategy, which did not allow to precisely localize them in the embryo. Intriguingly, direct reconstruction of the epiblast cell lineage from high-resolution light-sheet imaging of developing mouse embryos did not identify bipotential NMP cells in the PS region but only cells fated to one or the other lineage (McDole et al., 2018). Similar retrospective tracking experiments based on time-lapse movies of fluorescent cells in transgenic chicken embryos expressing GFP have led to identify a small population of epiblast cells that gives rise to descendants lying in neural and mesodermal territories (Wood et al., 2019). In the chicken and mouse embryo, extensive fate mappings of the PS region have been performed, but they did not reveal the existence of NMPs giving rise to paraxial mesoderm and neural tube (Brown and Storey, 2000; Iimura et al., 2007; Psychoyos and Stern, 1996; Schoenwolf et al., 1992; Selleck and Stern, 1991; Tam and Beddington, 1987; Wilson and Beddington, 1996; Wilson et al., 2009). Grafts of small territories of the epiblast adjacent to the anterior PS in avian or mouse embryos can give rise to both neural and mesodermal derivatives, suggesting that this territory could contain NMPs (Garcia-Martinez et al., 1993; Iimura et al., 2007; Wymeersch et al., 2016). In mouse, cells of this region (caudal lateral epiblast and node streak border) coexpress the neural marker Sox2 and the mesodermal marker T/Brachyury and were proposed to include the NMP population (Wymeersch et al., 2016). However, these grafting experiments do not allow to distinguish between a population of bipotential cells and a mixture of precursors committed to each lineage. To our knowledge, the only direct evidence for epiblast precursors able to give rise to both neural and mesodermal lineages in amniotes comes from single-cell injections of horseradish peroxidase in the mouse epiblast followed by embryo culture (Forlani et al., 2003). In this work, six clones with descendants in both lineages have been obtained from injections between the late streak and head fold stage. However, this study was performed before identification of NMPs by retrospective cloning and the bifated clones are not discussed in the article. Thus, the localization and fate of NMPs in amniotes remain incompletely understood.

Here, we performed direct lineage analysis of cells in the anterior PS epiblast of chicken embryos using two different approaches: labeling cells with a barcoded retroviral library and a Brainbow-derived strategy (Loulier et al., 2014). We show that single cells of the SOX2/T-positive region of the anterior PS are NMPs that can contribute to both neural tube and paraxial mesoderm and self-renew during formation of more posterior regions of the body. As most published fate mappings have been analyzed only after short term and did not study formation of the posterior regions of the embryo, cells of the anterior epiblast were considered monopotent and their bipotentiality went undetected. Thus, our findings reconcile the existence of bipotential NMPs demonstrated in mouse with the large body of amniote fate mappings described in classical developmental biology literature. We also performed single-cell RNA-sequencing (scRNAseq) analyses of the region encompassing the anterior PS epiblast and tail bud during axis elongation. This led to the identification of a cluster exhibiting two developmental trajectories leading to neural and mesodermal fates as expected for NMPs. We identified a similar cluster from published scRNAseq datasets from posterior tissues of equivalent stages of developing mouse embryos. In mouse and chicken embryos, these NMP populations maintained a conserved transcriptional identity segregating from their neural and mesodermal descendants as a single-cell cluster. During formation of the posterior body, the NMP population nevertheless underwent maturation characterized by expression of posterior Hox genes and loss of the epithelial phenotype. Moreover, using in toto imaging, we show that a posterior-to-anterior gradient of cell convergence speed toward the PS coupled to ingression results in the progressive exhaustion of the PS from its posterior end. This leads to the successive disappearance of the territories of the extraembryonic, lateral plate, intermediate mesoderm, and non-NMP paraxial mesoderm precursors (PMPs). Increased proliferation combined with limited ingression of cells ensures the self-renewal and expansion of the territory of clonally related bipotential NMPs during body formation, which constitutes the last remnant of the PS to contribute to the tail bud. Thus, our work provides a direct demonstration of the existence of NMP cells in amniotes and a mechanistic understanding for their sustained contribution to axis elongation.

NMPs are a population of stem cells that generate most of the posterior spinal cord, vertebrae, and skeletal muscles. Surprisingly, this cell population was only recently discovered using retrospective clonal analysis in mouse (Tzouanacou et al., 2009). The identification of such an important population of stem cells came much as a surprise because none of the many classical fate mapping studies of the epiblast and PS of the chicken or mouse embryo performed so far ever reported the identification of such bipotent precursors (Brown and Storey, 2000; Iimura et al., 2007; Psychoyos and Stern, 1996; Schoenwolf et al., 1992; Selleck and Stern, 1991; Tam and Beddington, 1987; Wilson and Beddington, 1996; Wilson et al., 2009). While the retrospective technique employed to first identify mouse NMPs unambiguously identified these bipotent stem cells, it could however not locate them in the embryo (Tzouanacou et al., 2009). Thus, direct identification of the amniote NMPs and their location in the embryo is still lacking. Here, we used lineage tracing and scRNAseq to identify and characterize NMPs as a population of stem cells located in the SOX2/T-expressing region of the anterior PS epiblast in the chicken embryo. We show that cells expressing the mesodermal marker T and the neural marker SOX2 are first found in the epiblast adjacent to the anterior PS and Hensen’s node region at the beginning of PS regression. This domain appears at a similar stage of mouse development and occupies a position similar to the SOX2/T domain of the node-streak border and the caudal lateral epiblast proposed to contain NMPs in mouse embryos (Wymeersch et al., 2016). In both mouse and chicken embryos, this domain contains cells fated to give rise to both neural and mesodermal derivatives, suggesting that they are functionally equivalent (Garcia-Martinez et al., 1993; Wymeersch et al., 2016). However, so far the bipotentiality of these cells has not been established at the single-cell level. We performed lineage tracing using a barcoded retroviral library and Brainbow-derived MAGIC markers (Loulier et al., 2014) to show that single cells of the SOX2/T region are bipotential and can contribute both to the neural tube and paraxial mesoderm along the trunk axis in chicken embryos. We further demonstrate clonal continuity and transcriptional homogeneity between early NMPs of the PS and late ones in the tail bud.

The significant contribution of cells of the SOX2/T territory of the anterior PS to both neural and mesodermal lineages has been missed in previous fate mapping studies of this region in chicken embryos (Brown and Storey, 2000; Fernández-Garre et al., 2002; Henrique et al., 1997; Iimura et al., 2007; Psychoyos and Stern, 1996; Schoenwolf et al., 1992; Selleck and Stern, 1991). Compared to these studies, we analyzed our lineage-tracing experiments at significantly later stages after in ovo labeling (stage 17–20HH instead of stage 10–14HH). Thus, while these fate maps indicate that cells of the anterior PS epiblast initially produce either neural or mesoderm descendants, a trend also observed in mouse clones (Tzouanacou et al., 2009), NMPs can give rise to both lineages but mostly later, in more posterior regions of the body. These observations are consistent with recent grafts of the epiblast territory in 6-somite chicken embryos showing that the territory first produces neural and then both mesodermal and neural derivatives (Kawachi et al., 2020). Direct identification of bipotential cells with a neural and mesodermal fate has been reported in zebrafish where they segregate during gastrulation and contribute to the most posterior part of the axis (Attardi et al., 2018). In zebrafish, however, only monopotent cells were found in the tail bud (Attardi et al., 2018; Kanki and Ho, 1997), while this is not the case in chicken (this report) and mouse (Tzouanacou et al., 2009).

scRNAseq analysis of the posterior embryonic region during PS to tail bud transition in chicken and mouse revealed a distinct NMP cluster lying between a cluster of PSM cells and one of neural cells. Using two different trajectory inference analysis methods (diffusion pseudotime and WOT), we identified two developmental trajectories leading from NMPs to these neural and PSM clusters, consistent with the bipotentiality of these cells. While the mouse NMP cluster shows expression of the NMP markers T, Sox2, and Nkx1.2, only T was identified in the chicken NMP cluster. SOX2 and NKX1.2 (SAX1) expression was neither found in the NMP cluster nor in neural tissue, suggesting a problem with annotation in the chicken genome. Alternatively, this might reflect a difference in sequencing depth as the chicken dataset was obtained using the inDrops pipeline, whereas most of the mouse data was obtained with the 10X platform. From our analysis, we identified a signature of genes enriched in the NMP clusters, including a majority of effectors and targets of wnt signaling, which is known to play a key role in the differentiation of this lineage (Henrique et al., 2015). Most of these signature genes are expressed in all chicken and mouse NMP clusters. Using WOT trajectory analysis, we identified cells within the NMP cluster as ancestors of the paraxial mesoderm and neural clusters in the mouse and chicken embryo. We were not able to find any bias for the NMP contribution toward the neural or mesodermal fate. This result aligns with our lineage-tracing data, which does not show any preferential enrichment toward a specific fate in our analysis timeframe. Interestingly, we identified genes associated to the NMP trajectory, including previously undescribed ones such as RARG or MNX1 in both the mouse and chicken datasets. We also identified expected transcription factors that are predictive of the fate differentiation toward the neural (PAX6) and PSM lineages (MSGN1). It will be interesting to investigate how these genes function during differentiation of these lineages. Using diffusion pseudotime and WOT analyses, we also identified a developmental trajectory within the NMP cluster, suggesting that NMP cells undergo maturation during axis formation. This is supported by the identification of clusters of early and late NMPs characterized by expression of specific genes in both chicken and mouse. As reported for mouse embryos (Wymeersch et al., 2019), late NMPs are characterized by the expression of more posterior Hox genes in both chicken and mouse. We also observed that genes associated to the epithelial state such as EpCAM, CDH1, or GJA1 are downregulated while genes associated to the mesenchymal state are upregulated at tail bud stages in both species. This is consistent with observations in mouse where tail bud progenitors were shown to undergo incomplete EMT during later stages of axial elongation (Dias et al., 2020).

As reported for mouse embryos (Wymeersch et al., 2016; Wymeersch et al., 2019), we observe an increase in SOX2/T cell numbers during axis elongation, indicating that these cells can self-renew while giving rise to a progeny in the paraxial mesoderm and in the neural tube. Labeling experiments in mouse and chicken embryos have identified a population of epiblast cells in the region of the anterior PS and Hensen’s node, which behave as stem cells, giving rise to descendants in the paraxial mesoderm while being able to self-renew (Cambray and Wilson, 2002; Cambray and Wilson, 2007; Iimura et al., 2007; McGrew et al., 2008; Nicolas et al., 1996; Selleck and Stern, 1991; Wilson et al., 2009). These cells were proposed to contribute mostly to medial somites while lateral somitic cells are derived from more posterior areas of the PS (Iimura et al., 2007; Selleck and Stern, 1991). The anterior SOX2/T territory encompasses Hensen’s node and the epiblast adjacent to the anterior PS and approximately corresponds to the territory containing the stem cells fated to give rise to medial somites. The epiblast territory immediately posterior to the SOX2/T territory does not express SOX2 and exhibits many cells positive for the paraxial mesoderm-specific marker MSGN1 (this study). This territory likely corresponds to the prospective territory of lateral somites, which does not show a stem cell behavior, giving rise to descendants spanning only ~5–7 segments (Iimura et al., 2007; Psychoyos and Stern, 1996). Our data suggest that this territory becomes largely exhausted at the end of PS regression, resulting in posterior somites to derive mostly from the NMPs (Figure 8D). This is consistent with observations in chicken and mouse demonstrating that the selective contribution of different PS territories to medial or lateral somitic territories only applies to anterior somites (Cambray and Wilson, 2007; Psychoyos and Stern, 1996).

We observed a posterior to anterior gradient of convergence in the epiblast associated to a parallel gradient of ingression in the PS. A parallel posterior to anterior gradient of cell motility of ingressed mesodermal cells has been documented along the regressing PS (Zamir et al., 2006). These graded movements in the mesoderm are largely controlled by a gradient of Fgf8 acting as a repellent on newly ingressed mesodermal cells at this stage (Yang et al., 2002). This cellular dynamics suggests that the movement away from the posterior PS could act as a sink for the epiblastic territories generating these progenitors. These cell movements in the epiblast and the ingressing mesoderm could explain the progressive exhaustion of the PS from its posterior end first described for the extraembryonic territory (Spratt, 1947). Importantly, our data show that the precursor territories along the PS do not disappear at the same rate. We observe a sequential posterior to anterior exhaustion of the territories of the extraembryonic mesoderm, the lateral plate, and the SOX2-negative PMPs (Figure 8D). Combined to an increased proliferation in the NMP region, this could explain why the SOX2/T territory eventually remains as the major remnant of the PS in the tail bud after PS regression. Thus, while most of the PS behave as a transit zone for committed progenitors as proposed by Pasteels, 1937, its anterior-most region behaves more like a blastema, containing multipotential cells as predicted by Wetzel (Romanoff, 1960; Wetzel, 1929). The fact that the tail bud contains multipotent NMPs in addition to monopotent territories such as the precursors of the notochord or the hindgut could explain the long-standing controversy on whether the tail bud functions as a blastema or as a mosaic of committed precursors (Catala et al., 1995; Davis and Kirschner, 2000; Gont et al., 1993; Holmdahl, 1925).

Here, we show that NMPs can contribute to a single type of derivative (neural or mesodermal) for several segments followed by another type in the next segments, as reported in mouse embryos (Tzouanacou et al., 2009). This argues for a striking plasticity of NMPs and rules out simple models of asymmetric divisions for the generation of neural and mesodermal descendants. Furthermore, as reported in mouse, we also observed bifated clones that do not contribute to the tail bud (Tzouanacou et al., 2009). Plasticity of the NMPs is supported by heterotopic grafts of the NMP epiblast territory into territories fated to become neural or mesodermal, which resulted in the donor cells to adopt the fate of their new territory in chicken or mouse embryos (Garcia-Martinez et al., 1997; McGrew et al., 2008; Wymeersch et al., 2016). Our results also show that the transcriptional identity of the NMP population remains largely stable throughout axis formation. This suggests that the NMP population behaves as a unique stem cell population that remains uncommitted toward either lineage, with its descendants acquiring their identity only after entering the territory of the paraxial mesoderm or the neural tube. Such plasticity is supported by the observation that SOX2/T-positive NMP-like cells generated in vitro can be induced to differentiate to either neural or mesodermal cells by changing the culture conditions (Diaz-Cuadros et al., 2020; Edri et al., 2019a; Edri et al., 2019b; Gouti et al., 2017; Gouti et al., 2014; Henrique et al., 2015; Turner et al., 2014). The difficulty to observe NMPs in vivo contrasts with the ease with which NMP-like cells can be obtained in vitro from mouse or human pluripotent stem cells. NMPs have been generated in 2D cultures by activation of Wnt signaling at the epiblast-like stage (Diaz-Cuadros et al., 2020; Edri et al., 2019a; Edri et al., 2019b; Gouti et al., 2017; Gouti et al., 2014; Henrique et al., 2015; Turner et al., 2014). Analysis of 3D cultures such as gastruloids induced in vitro from mouse ES cells also shows that most of the cells forming these structures belong to the neural tube and paraxial mesoderm lineage, suggesting that they are largely derived from an initial NMP population (Beccari et al., 2018; Faustino Martins et al., 2020; van den Brink et al., 2020). NMP-like cells generated in vitro express both SOX2 and T and single-cell RNA-sequencing demonstrated that they form a very homogeneous population with characteristics similar to the endogenous SOX2/T cells (Diaz-Cuadros et al., 2020; Gouti et al., 2017). Our comparison of chicken, mouse, and human NMP cells identified by scRNAseq points to an significant level of conservation of the transcriptional signature of this population. Overall, our work answers an important conundrum on the existence and fate of the NMPs in vivo, an elusive stem cell population in amniotes.

Sequencing data have been deposited in GEO under accession codes GSE161905. All the raw data to generate the graphs are given in a txt file.

1. 1. Guillot C
   2. Pourquie O

   (2019) NCBI Gene Expression Omnibus

   ID GSE161905. Single cell sequencing of posterior axis development in Gallus gallus.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE161905

1. 1. Pijuan-Sala

   (2017) ArrayExpress

   ID E-MTAB-6967. Timecourse single-cell RNAseq of whole mouse embryos harvested between days 6.5 and 8.5 of development.

   https://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-6967/
2. 1. Diaz-Cuadros

   (2020) NCBI Gene Expression Omnibus

   ID GSE114186. In vitro characterization of the human segmentation clock.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE114186

1. 1. Attardi A
   2. Fulton T
   3. Florescu M
   4. Shah G
   5. Muresan L
   6. Lenz MO
   7. Lancaster C
   8. Huisken J
   9. van Oudenaarden A
   10. Steventon B

   (2018) Neuromesodermal progenitors are a conserved source of spinal cord with divergent growth dynamics

   Development 145:dev166728.

   https://doi.org/10.1242/dev.166728

   • PubMed
   • Google Scholar
2. 1. Beccari L
   2. Moris N
   3. Girgin M
   4. Turner DA
   5. Baillie-Johnson P
   6. Cossy AC
   7. Lutolf MP
   8. Duboule D
   9. Arias AM

   (2018) Multi-axial self-organization properties of mouse embryonic stem cells into gastruloids

   Nature 562:272–276.

   https://doi.org/10.1038/s41586-018-0578-0

   • PubMed
   • Google Scholar
3. 1. Beier KT
   2. Samson ME
   3. Matsuda T
   4. Cepko CL

   (2011) Conditional expression of the TVA receptor allows clonal analysis of descendents from Cre-expressing progenitor cells

   Developmental Biology 353:309–320.

   https://doi.org/10.1016/j.ydbio.2011.03.004

   • PubMed
   • Google Scholar
4. 1. Bénazéraf B
   2. Francois P
   3. Baker RE
   4. Denans N
   5. Little CD
   6. Pourquié O

   (2010) A random cell motility gradient downstream of FGF controls elongation of an amniote embryo

   Nature 466:248–252.

   https://doi.org/10.1038/nature09151

   • PubMed
   • Google Scholar
5. 1. Bénazéraf B
   2. Beaupeux M
   3. Tchernookov M
   4. Wallingford A
   5. Salisbury T
   6. Shirtz A
   7. Shirtz A
   8. Huss D
   9. Pourquié O
   10. François P
   11. Lansford R

   (2017) Multi-scale quantification of tissue behavior during amniote embryo Axis elongation

   Development 144:4462–4472.

   https://doi.org/10.1242/dev.150557

   • PubMed
   • Google Scholar
6. 1. Brown JM
   2. Storey KG

   (2000) A region of the vertebrate neural plate in which neighbouring cells can adopt neural or epidermal fates

   Current Biology 10:869–872.

   https://doi.org/10.1016/S0960-9822(00)00601-1

   • PubMed
   • Google Scholar
7. 1. Bulusu V
   2. Prior N
   3. Snaebjornsson MT
   4. Kuehne A
   5. Sonnen KF
   6. Kress J
   7. Stein F
   8. Schultz C
   9. Sauer U
   10. Aulehla A

   (2017) Spatiotemporal analysis of a glycolytic activity gradient linked to mouse embryo mesoderm development

   Developmental Cell 40:331–341.

   https://doi.org/10.1016/j.devcel.2017.01.015

   • Google Scholar
8. 1. Cambray N
   2. Wilson V

   (2002) Axial progenitors with extensive potency are localised to the mouse chordoneural hinge

   Development 129:4855–4866.

   https://doi.org/10.1242/dev.129.20.4855

   • PubMed
   • Google Scholar
9. 1. Cambray N
   2. Wilson V

   (2007) Two distinct sources for a population of maturing axial progenitors

   Development 134:2829–2840.

   https://doi.org/10.1242/dev.02877

   • PubMed
   • Google Scholar
10. 1. Catala M
    2. Teillet M-A
    3. Le Douarin NM

    (1995) Organization and development of the tail bud analyzed with the quail-chick Chimaera system

    Mechanisms of Development 51:51–65.

    https://doi.org/10.1016/0925-4773(95)00350-A

    • Google Scholar
11. 1. Cepko C
    2. Pear W

    (2001) Retrovirus infection of cells in vitro and in vivo

    Current Protocols in Molecular Biology 9:14.

    https://doi.org/10.1002/0471142727.mb0914s36

    • PubMed
    • Google Scholar
12. 1. Chapman SC
    2. Collignon J
    3. Schoenwolf GC
    4. Lumsden A

    (2001) Improved method for chick whole-embryo culture using a filter paper carrier

    Developmental Dynamics 220:284–289.

    https://doi.org/10.1002/1097-0177(20010301)220:3<284::AID-DVDY1102>3.0.CO;2-5

    • PubMed
    • Google Scholar
13. 1. Davis RL
    2. Kirschner MW

    (2000) The fate of cells in the tailbud of Xenopus laevis

    Development 127:255–267.

    https://doi.org/10.1242/dev.127.2.255

    • PubMed
    • Google Scholar
14. 1. Denans N
    2. Iimura T
    3. Pourquié O

    (2015) Hox genes control vertebrate body elongation by collinear wnt repression

    eLife 4:e04379.

    https://doi.org/10.7554/eLife.04379

    • Google Scholar
15. 1. Dias A
    2. Lozovska A
    3. Wymeersch FJ
    4. Nóvoa A
    5. Binagui-Casas A
    6. Sobral D
    7. Martins GG
    8. Wilson V
    9. Mallo M

    (2020) A Tgfbr1/Snai1-dependent developmental module at the core of vertebrate axial elongation

    eLife 9:e56615.

    https://doi.org/10.7554/eLife.56615

    • PubMed
    • Google Scholar
16. 1. Diaz-Cuadros M
    2. Wagner DE
    3. Budjan C
    4. Hubaud A
    5. Tarazona OA
    6. Donelly S
    7. Michaut A
    8. Al Tanoury Z
    9. Yoshioka-Kobayashi K
    10. Niino Y
    11. Kageyama R
    12. Miyawaki A
    13. Touboul J
    14. Pourquié O

    (2020) In vitro characterization of the human segmentation clock

    Nature 580:113–118.

    https://doi.org/10.1038/s41586-019-1885-9

    • PubMed
    • Google Scholar
17. 1. Edri S
    2. Hayward P
    3. Baillie-Johnson P
    4. Steventon BJ
    5. Martinez Arias A

    (2019a) An epiblast stem cell-derived multipotent progenitor population for axial extension

    Development 146:dev168187.

    https://doi.org/10.1242/dev.168187

    • PubMed
    • Google Scholar
18. 1. Edri S
    2. Hayward P
    3. Jawaid W
    4. Martinez Arias A

    (2019b) Neuro-mesodermal progenitors (NMPs): a comparative study between pluripotent stem cells and embryo-derived populations

    Development 146:dev180190.

    https://doi.org/10.1242/dev.180190

    • PubMed
    • Google Scholar
19. 1. Faustino Martins J-M
    2. Fischer C
    3. Urzi A
    4. Vidal R
    5. Kunz S
    6. Ruffault P-L
    7. Kabuss L
    8. Hube I
    9. Gazzerro E
    10. Birchmeier C
    11. Spuler S
    12. Sauer S
    13. Gouti M

    (2020) Self-Organizing 3D human trunk neuromuscular organoids

    Cell Stem Cell 26:172–186.

    https://doi.org/10.1016/j.stem.2019.12.007

    • Google Scholar
20. 1. Fernández-Garre P
    2. Rodríguez-Gallardo L
    3. Gallego-Díaz V
    4. Alvarez IS
    5. Puelles L

    (2002) Fate map of the chicken neural plate at stage 4

    Development 129:2807–2822.

    https://doi.org/10.1242/dev.129.12.2807

    • PubMed
    • Google Scholar
21. 1. Forlani S
    2. Lawson KA
    3. Deschamps J

    (2003) Acquisition of hox codes during gastrulation and axial elongation in the mouse embryo

    Development 130:3807–3819.

    https://doi.org/10.1242/dev.00573

    • Google Scholar
22. 1. Garcia-Martinez V
    2. Alvarez IS
    3. Schoenwolf GC

    (1993) Locations of the ectodermal and nonectodermal subdivisions of the epiblast at stages 3 and 4 of avian gastrulation and neurulation

    Journal of Experimental Zoology 267:431–446.

    https://doi.org/10.1002/jez.1402670409

    • PubMed
    • Google Scholar
23. 1. Garcia-Martinez V
    2. Darnell DK
    3. Lopez-Sanchez C
    4. Sosic D
    5. Olson EN
    6. Schoenwolf GC

    (1997) State of commitment of prospective neural plate and prospective mesoderm in late gastrula/early neurula stages of avian embryos

    Developmental Biology 181:102–115.

    https://doi.org/10.1006/dbio.1996.8439

    • PubMed
    • Google Scholar
24. 1. Gibson DG
    2. Young L
    3. Chuang RY
    4. Venter JC
    5. Hutchison CA
    6. Smith HO

    (2009) Enzymatic assembly of DNA molecules up to several hundred kilobases

    Nature Methods 6:343–345.

    https://doi.org/10.1038/nmeth.1318

    • PubMed
    • Google Scholar
25. 1. Gont LK
    2. Steinbeisser H
    3. Blumberg B
    4. de Robertis EM

    (1993) Tail formation as a continuation of gastrulation: the multiple cell populations of the Xenopus tailbud derive from the late blastopore lip

    Development 119:991–1004.

    https://doi.org/10.1242/dev.119.4.991

    • PubMed
    • Google Scholar
26. 1. Gouti M
    2. Tsakiridis A
    3. Wymeersch FJ
    4. Huang Y
    5. Kleinjung J
    6. Wilson V
    7. Briscoe J

    (2014) In vitro generation of neuromesodermal progenitors reveals distinct roles for wnt signalling in the specification of spinal cord and paraxial mesoderm identity

    PLOS Biology 12:e1001937.

    https://doi.org/10.1371/journal.pbio.1001937

    • PubMed
    • Google Scholar
27. 1. Gouti M
    2. Delile J
    3. Stamataki D
    4. Wymeersch FJ
    5. Huang Y
    6. Kleinjung J
    7. Wilson V
    8. Briscoe J

    (2017) A gene regulatory network balances neural and mesoderm specification during vertebrate trunk development

    Developmental Cell 41:243–261.

    https://doi.org/10.1016/j.devcel.2017.04.002

    • Google Scholar
28. 1. Guibentif C
    2. Griffiths JA
    3. Imaz-Rosshandler I
    4. Ghazanfar S
    5. Nichols J
    6. Wilson V
    7. Göttgens B
    8. Marioni JC

    (2021) Diverse routes toward early somites in the mouse embryo

    Developmental Cell 56:141–153.

    https://doi.org/10.1016/j.devcel.2020.11.013

    • Google Scholar
29. 1. Hama H
    2. Kurokawa H
    3. Kawano H
    4. Ando R
    5. Shimogori T
    6. Noda H
    7. Fukami K
    8. Sakaue-Sawano A
    9. Miyawaki A

    (2011) Scale: a chemical approach for fluorescence imaging and reconstruction of transparent mouse brain

    Nature Neuroscience 14:1481–1488.

    https://doi.org/10.1038/nn.2928

    • PubMed
    • Google Scholar
30. 1. Hamburger V
    2. Hamilton HL

    (1992) A series of normal stages in the development of the chick embryo. 1951

    Developmental Dynamics 195:231–272.

    https://doi.org/10.1002/aja.1001950404

    • PubMed
    • Google Scholar
31. 1. Harwell CC
    2. Fuentealba LC
    3. Gonzalez-Cerrillo A
    4. Parker PR
    5. Gertz CC
    6. Mazzola E
    7. Garcia MT
    8. Alvarez-Buylla A
    9. Cepko CL
    10. Kriegstein AR

    (2015) Wide dispersion and diversity of clonally related inhibitory interneurons

    Neuron 87:999–1007.

    https://doi.org/10.1016/j.neuron.2015.07.030

    • PubMed
    • Google Scholar
32. 1. Henrique D
    2. Tyler D
    3. Kintner C
    4. Heath JK
    5. Lewis JH
    6. Ish-Horowicz D
    7. Storey KG

    (1997) cash4, a novel achaete-scute homolog induced by Hensen's node during generation of the posterior nervous system

    Genes & Development 11:603–615.

    https://doi.org/10.1101/gad.11.5.603

    • PubMed
    • Google Scholar
33. 1. Henrique D
    2. Abranches E
    3. Verrier L
    4. Storey KG

    (2015) Neuromesodermal progenitors and the making of the spinal cord

    Development 142:2864–2875.

    https://doi.org/10.1242/dev.119768

    • PubMed
    • Google Scholar
34. 1. Holmdahl DE

    (1925)

    Experimentelle untersuchungen über die lage der grenze zwischen primärer und sekundärer körperentwicklung beim huhn

    Anatomischer Anzeiger 59:393–396.

    • Google Scholar
35. 1. Hunter JD

    (2007) Matplotlib: a 2D graphics environment

    Computing in Science & Engineering 9:90–95.

    https://doi.org/10.1109/MCSE.2007.55

    • Google Scholar
36. 1. Iimura T
    2. Yang X
    3. Weijer CJ
    4. Pourquié O

    (2007) Dual mode of paraxial mesoderm formation during chick gastrulation

    PNAS 104:2744–2749.

    https://doi.org/10.1073/pnas.0610997104

    • PubMed
    • Google Scholar
37. 1. Iimura T
    2. Pourquié O

    (2008) Manipulation and electroporation of the avian segmental plate and somites in vitro

    Methods in Cell Biology 87:257–270.

    https://doi.org/10.1016/S0091-679X(08)00213-6

    • PubMed
    • Google Scholar
38. 1. Kanki JP
    2. Ho RK

    (1997) The development of the posterior body in zebrafish

    Development 124:881–893.

    https://doi.org/10.1242/dev.124.4.881

    • PubMed
    • Google Scholar
39. 1. Kawachi T
    2. Shimokita E
    3. Kudo R
    4. Tadokoro R
    5. Takahashi Y

    (2020) Neural-fated self-renewing cells regulated by Sox2 during secondary neurulation in chicken tail bud

    Developmental Biology 461:160–171.

    https://doi.org/10.1016/j.ydbio.2020.02.007

    • PubMed
    • Google Scholar
40. 1. Kinder SJ
    2. Tsang TE
    3. Quinlan GA
    4. Hadjantonakis AK
    5. Nagy A
    6. Tam PP

    (1999) The orderly allocation of mesodermal cells to the extraembryonic structures and the anteroposterior Axis during gastrulation of the mouse embryo

    Development 126:4691–4701.

    https://doi.org/10.1242/dev.126.21.4691

    • PubMed
    • Google Scholar
41. 1. Klein AM
    2. Mazutis L
    3. Akartuna I
    4. Tallapragada N
    5. Veres A
    6. Li V
    7. Peshkin L
    8. Weitz DA
    9. Kirschner MW

    (2015) Droplet barcoding for single-cell transcriptomics applied to embryonic stem cells

    Cell 161:1187–1201.

    https://doi.org/10.1016/j.cell.2015.04.044

    • PubMed
    • Google Scholar
42. 1. Landau NR
    2. Littman DR

    (1992) Packaging system for rapid production of murine leukemia virus vectors with variable tropism

    Journal of Virology 66:5110–5113.

    https://doi.org/10.1128/jvi.66.8.5110-5113.1992

    • PubMed
    • Google Scholar
43. 1. Le Douarin NM
    2. Teillet MA
    3. Catala M

    (1998)

    Neurulation in amniote vertebrates: a novel view deduced from the use of quail-chick chimeras

    The International Journal of Developmental Biology 42:909–916.

    • PubMed
    • Google Scholar
44. 1. Liberzon A
    2. Subramanian A
    3. Pinchback R
    4. Thorvaldsdóttir H
    5. Tamayo P
    6. Mesirov JP

    (2011) Molecular signatures database (MSigDB) 3.0

    Bioinformatics 27:1739–1740.

    https://doi.org/10.1093/bioinformatics/btr260

    • PubMed
    • Google Scholar
45. 1. Loulier K
    2. Barry R
    3. Mahou P
    4. Le Franc Y
    5. Supatto W
    6. Matho KS
    7. Ieng S
    8. Fouquet S
    9. Dupin E
    10. Benosman R
    11. Chédotal A
    12. Beaurepaire E
    13. Morin X
    14. Livet J

    (2014) Multiplex cell and lineage tracking with combinatorial labels

    Neuron 81:505–520.

    https://doi.org/10.1016/j.neuron.2013.12.016

    • PubMed
    • Google Scholar
46. 1. McDole K
    2. Guignard L
    3. Amat F
    4. Berger A
    5. Malandain G
    6. Royer LA
    7. Turaga SC
    8. Branson K
    9. Keller PJ

    (2018) In toto imaging and reconstruction of Post-Implantation mouse development at the Single-Cell level

    Cell 175:859–876.

    https://doi.org/10.1016/j.cell.2018.09.031

    • Google Scholar
47. 1. McGrew MJ
    2. Sherman A
    3. Ellard FM
    4. Lillico SG
    5. Gilhooley HJ
    6. Kingsman AJ
    7. Mitrophanous KA
    8. Sang H

    (2004) Efficient production of germline transgenic chickens using lentiviral vectors

    EMBO Reports 5:728–733.

    https://doi.org/10.1038/sj.embor.7400171

    • PubMed
    • Google Scholar
48. 1. McGrew MJ
    2. Sherman A
    3. Lillico SG
    4. Ellard FM
    5. Radcliffe PA
    6. Gilhooley HJ
    7. Mitrophanous KA
    8. Cambray N
    9. Wilson V
    10. Sang H

    (2008) Localised axial progenitor cell populations in the avian tail bud are not committed to a posterior hox identity

    Development 135:2289–2299.

    https://doi.org/10.1242/dev.022020

    • PubMed
    • Google Scholar
49. 1. Moreau C
    2. Caldarelli P
    3. Rocancourt D
    4. Roussel J
    5. Denans N
    6. Pourquie O
    7. Gros J

    (2019) Timed collinear activation of hox genes during gastrulation controls the avian forelimb position

    Current Biology 29:35–50.

    https://doi.org/10.1016/j.cub.2018.11.009

    • Google Scholar
50. 1. Nicolas JF
    2. Mathis L
    3. Bonnerot C
    4. Saurin W

    (1996) Evidence in the mouse for self-renewing stem cells in the formation of a segmented longitudinal structure, the myotome

    Development 122:2933–2946.

    https://doi.org/10.1242/dev.122.9.2933

    • PubMed
    • Google Scholar
51. 1. Oginuma M
    2. Moncuquet P
    3. Xiong F
    4. Karoly E
    5. Chal J
    6. Guevorkian K
    7. Pourquié O

    (2017) A gradient of glycolytic activity coordinates FGF and wnt signaling during elongation of the body Axis in amniote embryos

    Developmental Cell 40:342–353.

    https://doi.org/10.1016/j.devcel.2017.02.001

    • Google Scholar
52. 1. Olivera-Martinez I
    2. Harada H
    3. Halley PA
    4. Storey KG

    (2012) Loss of FGF-Dependent mesoderm identity and rise of endogenous retinoid signalling determine cessation of body Axis elongation

    PLOS Biology 10:e1001415.

    https://doi.org/10.1371/journal.pbio.1001415

    • Google Scholar
53. 1. Ory DS
    2. Neugeboren BA
    3. Mulligan RC

    (1996) A stable human-derived packaging cell line for production of high titer retrovirus/vesicular stomatitis virus G pseudotypes

    PNAS 93:11400–11406.

    https://doi.org/10.1073/pnas.93.21.11400

    • PubMed
    • Google Scholar
54. 1. Pasteels J

    (1937)

    Etudes sur la gastrulation des vertébrés méroblastiques III Oiseaux IV Conclusions générales

    ArchBiol 48:381–488.

    • Google Scholar
55. 1. Pijuan-Sala B
    2. Griffiths JA
    3. Guibentif C
    4. Hiscock TW
    5. Jawaid W
    6. Calero-Nieto FJ
    7. Mulas C
    8. Ibarra-Soria X
    9. Tyser RCV
    10. Ho DLL
    11. Reik W
    12. Srinivas S
    13. Simons BD
    14. Nichols J
    15. Marioni JC
    16. Göttgens B

    (2019) A single-cell molecular map of mouse gastrulation and early organogenesis

    Nature 566:490–495.

    https://doi.org/10.1038/s41586-019-0933-9

    • PubMed
    • Google Scholar
56. Software

    1. Pourquie Lab

    (2021) Guillot_2021, version swh:1:rev:3d80f422bc5a675c08546cbb10ab79211e430b1e

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:rev:3d80f422bc5a675c08546cbb10ab79211e430b1e
57. 1. Psychoyos D
    2. Stern CD

    (1996) Fates and migratory routes of primitive streak cells in the chick embryo

    Development 122:1523–1534.

    https://doi.org/10.1242/dev.122.5.1523

    • PubMed
    • Google Scholar
58. Book

    1. Romanoff A

    (1960)

    The Avian Embryo: Structural and Functional Development

    The Macmillan Company.

    • Google Scholar
59. 1. Sbalzarini IF
    2. Koumoutsakos P

    (2005) Feature point tracking and trajectory analysis for video imaging in cell biology

    Journal of Structural Biology 151:182–195.

    https://doi.org/10.1016/j.jsb.2005.06.002

    • PubMed
    • Google Scholar
60. 1. Schiebinger G
    2. Shu J
    3. Tabaka M
    4. Cleary B
    5. Subramanian V
    6. Solomon A
    7. Gould J
    8. Liu S
    9. Lin S
    10. Berube P
    11. Lee L
    12. Chen J
    13. Brumbaugh J
    14. Rigollet P
    15. Hochedlinger K
    16. Jaenisch R
    17. Regev A
    18. Lander ES

    (2019) Optimal-Transport analysis of Single-Cell gene expression identifies developmental trajectories in reprogramming

    Cell 176:928–943.

    https://doi.org/10.1016/j.cell.2019.01.006

    • Google Scholar
61. 1. Schoenwolf GC
    2. Garcia-Martinez V
    3. Dias MS

    (1992) Mesoderm movement and fate during avian gastrulation and neurulation

    Developmental Dynamics 193:235–248.

    https://doi.org/10.1002/aja.1001930304

    • PubMed
    • Google Scholar
62. 1. scikit-image contributors
    2. van der Walt S
    3. Schönberger JL
    4. Nunez-Iglesias J
    5. Boulogne F
    6. Warner JD
    7. Yager N
    8. Gouillart E
    9. Yu T

    (2014) scikit-image: image processing in Python

    PeerJ 2:e453.

    https://doi.org/10.7717/peerj.453

    • PubMed
    • Google Scholar
63. 1. Selleck MA
    2. Stern CD

    (1991) Fate mapping and cell lineage analysis of Hensen's node in the chick embryo

    Development 112:615–626.

    https://doi.org/10.1242/dev.112.2.615

    • PubMed
    • Google Scholar
64. 1. Smith JL
    2. Gesteland KM
    3. Schoenwolf GC

    (1994) Prospective fate map of the mouse primitive streak at 7.5 days of gestation

    Developmental Dynamics 201:279–289.

    https://doi.org/10.1002/aja.1002010310

    • PubMed
    • Google Scholar
65. 1. Spratt NT

    (1947) Regression and shortening of the primitive streak in the explanted chick blastoderm

    Journal of Experimental Zoology 104:69–100.

    https://doi.org/10.1002/jez.1401040105

    • PubMed
    • Google Scholar
66. 1. Spratt NT
    2. Condon L

    (1947)

    Localization of prospective chordo and somite mesoderm during regression of the primitive streak in the chick blastoderm

    The Anatomical Record 99:653.

    • Google Scholar
67. 1. Stern CD

    (1979) A re-examination of mitotic activity in the early chick embryo

    Anatomy and Embryology 156:319–329.

    https://doi.org/10.1007/BF00299630

    • Google Scholar
68. Book

    1. Stern CD

    (2004)

    Gastrulation in the chick

    In: Stern C. D, editors. Gastrulation: From Cells to Embryos. Cold Spring Harbor Laboratory Press. pp. 219–232.

    • Google Scholar
69. 1. Subramanian V
    2. Meyer BI
    3. Gruss P

    (1995) Disruption of the murine homeobox gene Cdx1 affects axial skeletal identities by altering the mesodermal expression domains of hox genes

    Cell 83:641–653.

    https://doi.org/10.1016/0092-8674(95)90104-3

    • PubMed
    • Google Scholar
70. 1. Subramanian A
    2. Tamayo P
    3. Mootha VK
    4. Mukherjee S
    5. Ebert BL
    6. Gillette MA
    7. Paulovich A
    8. Pomeroy SL
    9. Golub TR
    10. Lander ES
    11. Mesirov JP

    (2005) Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles

    PNAS 102:15545–15550.

    https://doi.org/10.1073/pnas.0506580102

    • PubMed
    • Google Scholar
71. 1. Tam PP
    2. Beddington RS

    (1987) The formation of mesodermal tissues in the mouse embryo during gastrulation and early organogenesis

    Development 99:109–126.

    https://doi.org/10.1242/dev.99.1.109

    • PubMed
    • Google Scholar
72. 1. Turner DA
    2. Hayward PC
    3. Baillie-Johnson P
    4. Rué P
    5. Broome R
    6. Faunes F
    7. Martinez Arias A

    (2014) Wnt/β-catenin and FGF signalling direct the specification and maintenance of a neuromesodermal axial progenitor in ensembles of mouse embryonic stem cells

    Development 141:4243–4253.

    https://doi.org/10.1242/dev.112979

    • PubMed
    • Google Scholar
73. 1. Tzouanacou E
    2. Wegener A
    3. Wymeersch FJ
    4. Wilson V
    5. Nicolas JF

    (2009) Redefining the progression of lineage segregations during mammalian embryogenesis by clonal analysis

    Developmental Cell 17:365–376.

    https://doi.org/10.1016/j.devcel.2009.08.002

    • PubMed
    • Google Scholar
74. 1. van den Brink SC
    2. Alemany A
    3. van Batenburg V
    4. Moris N
    5. Blotenburg M
    6. Vivié J
    7. Baillie-Johnson P
    8. Nichols J
    9. Sonnen KF
    10. Martinez Arias A
    11. van Oudenaarden A

    (2020) Single-cell and spatial transcriptomics reveal somitogenesis in gastruloids

    Nature 582:405–409.

    https://doi.org/10.1038/s41586-020-2024-3

    • PubMed
    • Google Scholar
75. Software

    1. Veres A

    (2021) Project YAML file, version swh:1:rev:2ad4669b72ea6ba794f962ed95b778560bdfde4a

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:rev:2ad4669b72ea6ba794f962ed95b778560bdfde4a
76. 1. Wetzel R

    (1929) Untersuchungen am hühnchen die entwicklung des keims während der ersten beiden bruttage

    Wilhelm Roux' Archiv Für Entwicklungsmechanik Der Organismen 119:188–321.

    https://doi.org/10.1007/BF02111186

    • Google Scholar
77. 1. Wilson V
    2. Olivera-Martinez I
    3. Storey KG

    (2009) Stem cells, signals and vertebrate body Axis extension

    Development 136:1591–1604.

    https://doi.org/10.1242/dev.021246

    • PubMed
    • Google Scholar
78. 1. Wilson V
    2. Beddington RS

    (1996) Cell fate and morphogenetic movement in the late mouse primitive streak

    Mechanisms of Development 55:79–89.

    https://doi.org/10.1016/0925-4773(95)00493-9

    • PubMed
    • Google Scholar
79. Preprint

    1. Wood TR
    2. Kyrsting A
    3. Stegmaier J
    4. Kucinski I
    5. Kaminski CF
    6. Mikut R
    7. Voiculescu O

    (2019) Neuromesodermal progenitors separate the axial stem zones while producing few single- and dual-fated descendants

    bioRxiv.

    https://doi.org/10.1101/622571

    • Google Scholar
80. 1. Wymeersch FJ
    2. Huang Y
    3. Blin G
    4. Cambray N
    5. Wilkie R
    6. Wong FC
    7. Wilson V

    (2016) Position-dependent plasticity of distinct progenitor types in the primitive streak

    eLife 5:e10042.

    https://doi.org/10.7554/eLife.10042

    • PubMed
    • Google Scholar
81. 1. Wymeersch FJ
    2. Skylaki S
    3. Huang Y
    4. Watson JA
    5. Economou C
    6. Marek-Johnston C
    7. Tomlinson SR
    8. Wilson V

    (2019) Transcriptionally dynamic progenitor populations organised around a stable niche drive axial patterning

    Development 146:dev168161.

    https://doi.org/10.1242/dev.168161

    • PubMed
    • Google Scholar
82. 1. Yang X
    2. Dormann D
    3. Münsterberg AE
    4. Weijer CJ

    (2002) Cell movement patterns during gastrulation in the chick are controlled by positive and negative chemotaxis mediated by FGF4 and FGF8

    Developmental Cell 3:425–437.

    https://doi.org/10.1016/S1534-5807(02)00256-3

    • PubMed
    • Google Scholar
83. 1. Zamir EA
    2. Czirók A
    3. Cui C
    4. Little CD
    5. Rongish BJ

    (2006) Mesodermal cell displacements during avian gastrulation are due to both individual cell-autonomous and convective tissue movements

    PNAS 103:19806–19811.

    https://doi.org/10.1073/pnas.0606100103

    • PubMed
    • Google Scholar

Acceptance summary:

This exciting paper characterizes the cellular dynamics of a population of stem cells, recently identified, called neuro-mesodermal progenitors in the chick embryo. By conducting prospective lineage trace and transcriptomic analysis, they identify dynamic cell behaviors that explain previous 'snapshot' studies showing that neuro-mesodermal progenitors contribute over long axial distances. This manuscript makes an important contribution to the understanding of how neuro-mesodermal progenitors contribute to vertebrate axial elongation and provides novel insights into axial morphogenesis.

Decision letter after peer review:

Thank you for submitting your article "Dynamics of primitive streak regression controls the fate of neuro-mesodermal progenitor cells in the chicken embryo" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Valerie Wilson (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential Revisions

1. Many of the concerns can be addressed by text changes and minor changes to the existing data presentation as detailed in the full reviews.

2. The main problems are with the bioinformatic analysis, where both reviewers highlight issues where the analysis is too superficial and needs further work.

3. Additional imaging quantification is required to support their findings.

Reviewer #1:

The paper entitled "Dynamics of primitive streak regression controls the fate of neuro-mesodermal progenitor cells in the chicken embryo" from Guillot and colleagues aims to characterize in the context of developing chick embryos the cellular dynamics of a population of stem cells, recently identified, called Neuro-Mesodermal Progenitors (NMPs). The authors provide several lines of evidence supporting their claims that overall look interesting, however, additional imaging quantification and more comprehensive analysis of their transcriptomic datasets are required to support their findings.

Specifically, the authors generated an important scRNA-seq dataset of micro-dissect regions of the chick embryos across developmental time points from stage 5 to 35-somites. Despite the dataset represents a crucial part of the paper supporting the authors' findings of this NMP population, the analysis is not deep enough and does not report several controls and markers. In figure 2 the authors should show heatmap of the marker genes used for classification. Louvain markers genes that define each cluster should be showed in a supplementary not as an unreadable table, instead, a dotplot or heatmap would be better. It is unclear whether Louvain clustering has been used to label the data, did the authors find that each Louvain cluster correspond to a defined cell identity? The pseudotime analysis is incomplete and superficial. Diffusion pseudotime strongly depends on the root cell selection. RNA velocity analysis and/or another trajectory analysis are required to better show lineage trajectory. Figure 4 L-K, the pesudotime time is not progressively ordered and it would be advisable to perform an alternative analysis to visualize lineage trajectory. Furthermore, gene expression trends should be identified and further discussed over those trajectories. The supporting information about the classification analysis using the mouse dataset is incomplete.

– In figure 1 the authors show SOX2-T double staining. Panels A-B-C represent the first evidence of SOX2-T co-expression in stage 5 chick embryos. The merged image is not provided together with the single panels A-B and only appear in panel G of the same figure, this is confusing. Panel B has two scale bars. The authors tried to show SOX2-T co-expression by thresholding and masking SOX2-T signals in panel C and Suppl. Figure 1. The authors should clearly explain how this thresholding is obtained. Is this manually set? Is this set based on global or local signal intensity distribution? Otsu thresholding? In order to avoid artifacts due to thresholding, the authors should also display a scatter plot showing nuclear intensity values for SOX2-T. FACS analysis and quantitative assessments of the percentage of cells would further strengthen the authors' claim.

– The authors should validate SOX2-T expression in quail embryos

– The authors should better discuss the mechanical forces that are implicated in epithelium to mesenchyme, and in the posterior gradient of proliferation that counteracts ingression in the anterior PS epiblast. Is this affecting/related to WNT or hyppo pathway?

– Can the authors speculate about the conservation of this phenomena in the human context? What is the evidence that this conserved in human?

– The authors speculate about in-vitro generated NMPs, did they try transplantation of in-vitro generate cells? Can they compare the transcriptomic signature of the human in-vitro generated NMP with the chick counterpart?

– The tables in Suppl. Figure 4 should be a supplementary table.

– In Suppl.Fig5D there is not scale bar to assess classification efficacy and material and methods do not describe how this analysis has been performed.

– The authors should substantially revise scRNA-seq data.

Reviewer #2 :

In this manuscript, the authors aimed firstly to prospectively identify cells with dual neural and mesodermal fate. Previous studies in mouse had determined that such dual-fated cells existed, and had narrowed down their location to subregions of the axial progenitor zone co-expressing Sox2 and T. However, these studies had not shown that individual cells in the Sox2/T positive region were dual-fated. The authors use three approaches: live imaging, and two prospective clonal labelling studies of the Sox2/T positive region to show, for the first time, that indeed this region harbours dual-fated cells. This is an important missing piece of the current volume of work on NMPs. Whilst the rationale for the experiments are clear, it might be better to present this work more in the context of previous studies. It seems a little harsh to dismiss population fate mapping studies as 'failing to identify' NMPs rather than complementing the clonal analysis by identifying regions potentially harbouring NMPs. In addition, there are interesting parallels with the clonal analysis carried out in mouse. For example, not all NM clones contribute to the tail bud; the observation in live imaging studies that NMPs first contribute to the neural tube and later to mesoderm is also similar to the observations from mouse retrospective clonal analysis; this could perhaps be discussed more fully.

Next, the authors carry out single cell RNA-seq analyses of three stages of chick axial elongation, and compare these with existing datasets in mouse, adding data from a later stage of mouse embryos. This aspect of the study has again confirmed and extended previous studies, showing both similarities and differences between chick and mouse NMPs. Trajectory analysis suggests progression of NMPs towards neural and mesodermal lineages in both chick and mouse. This is a valuable resource for the community. However once again it would be good to present this aspect of the work in the context of previous studies showing maturation of NMPs over time and an increased mesenchymal transcriptomic signature of NMPs at late compared to early timepoints.

Finally, the authors carry out dynamic cell tracking and show that NMPs ingress later, and proliferate faster than more posteriorly-placed cells in the primitive streak, leading to exhaustion of the posterior populations and long-term retention of the NMP population. This part of the study is well-executed and contains important new information that advances our understanding of both NMPs and axial progenitors as a whole.

Suggestions for strengthening the science

1. The results presented in Figure 1 are a little difficult to interpret as there is some background fluorescence in Sox2 and Msgn1 staining. Perhaps the thresholding strategy could be described in the supplementary data to support statements about Sox2 positive time of emergence and the location of Msgn1 positive/Sox2 negative axial levels. It would be good to cite the source of the fate mapping information locating NMP/PMP/LPP/EMP in Figure 1D- is it this study or another? The section at level 5 in Figure 1E is named LPP in 1E but corresponds to a region marked 'EMP' in Fig1D. Together these two factors make it hard for the reader to judge where the posterior limit of Sox2/T positive cells is, relative to Msgn1 and to regions of differential fate. Also, the text (p5) says there are 'low levels [of TBXT] in the node' but the expected high levels in the node/ emerging notochord are visible in Fig1E sections 1-2.

2. In Figure 2, the clustering shows two datasets, (which are presumably batches?) of cells for each stage in chick (st5 1 etc). However, the colour coding does not allow the reader to distinguish these. Do the batches occupy different positions in the UMAP plot? How was batch correction done?

3. The cells annotated as NMPs in chick show elevated expression of Hes5 (6-somite stage) and Dll1 (combined dataset) relative to the cells annotated as PSM or neural (Figure 3, S4,5). Also, some of the NMP signature markers are expressed in a higher percentage of the NMP population in mouse compared to chick. This suggests that the cluster annotated as NMPs in chick also includes the midline primitive streak. The absence of Sox2 from this dataset is not, in my opinion, especially troublesome as it is expressed at low levels. However, it means that the chick single cell transcriptomes cannot be independently verified as NMPs and distinguished from midline streak cells. Perhaps the mouse data can help here. Instead of using the chick as a starting point to define the NMP phenotype, if the authors can work back from cells dissected from the Sox2/T positive regions (ie from Dias et al., eLife 2020 and Gouti et al., Dev Cell 2017), as well as the NMP annotations in the mouse atlas data used here (Pijuan-Sala et al), how well do the chick cells correspond to the mouse ones? If a narrower definition of NMPs is used in chick, does this remove any of the variation seen between mouse and chick?

4. The mouse NMP trajectory analysis (Figure 3M) shows two arrows. What do they represent? if just pseudotime, it looks more like a radiation in all directions from the centre. Interestingly this would suggest NMPs also progress (upwards on the Y axis) as NMPs in pseudotime. Is this effect seen in the chick NMPs?

4. The annotations of chick versus mouse early and late clusters are confusing, In chick, there are two early clusters (coloured the same) annotated nmp-1 and nmp-2, but these do not correspond to the two earlier stages. What do they correspond to? Some discussion of the heterogeneity would be appropriate- or possibly a reanalysis of the NMP annotation in point 3 above may highlight that one of these clusters is less NMP-like. In the mouse, there is one early subcluster and two late ones. These are also annotated nmp-1 and nmp-2. A different nomenclature should be used to highlight that they are not the same as the chick ones. Unlike the chick clusters, these seem to correspond to different stages of development.

Suggestions for improving the manuscript

1. The statements that no previous studies have prospectively identified neuromesodermal-fated cells (abstract, introduction p4, discussion p10) should be modified. Forlani et al., Development 2003 10.1242/dev.00573 and Wood et al. BioRXiv https://doi.org/10.1101/622571 show dual-fated cells in mouse and chick. I think that it would be fairer in the context of previous studies to cite the population or oligoclonal studies of Iimura and Pourquié, (2006) 10.1073/pnas.0610997104, Cambray and Wilson, (2007) 10.1242/dev.02877 and Brown and Storey, (2000) 10.1016/s0960-9822(00)00601-1 as showing potential locations for cells of dual neuromesodermal fate.

2. The words 'bipotent/monopotent' used throughout the manuscript is not appropriate for the studies here, which are demonstrating fate and not potency. I would suggest 'dual fate'.

3. NMPs are presented here as 'stem cells' when the transcriptome analysis shows they mature with time. It would be good to define exactly in what sense the term is being used- it is valid since the manuscript describes an enduring progenitor producing neural and mesodermal tissue but should be used with care.

4. p5. 'We identified 7 clones containing cells expressing the same barcodes' – could this text be modified to clarify, e.g. 'expressing unique (or clone-specific) barcodes'?

5. p7 para 2 'the NMP lineage' should be changed to something not implying lineage, e.g. 'cluster'

6. Figure S7 is entitled 'NMP early and late clusters are not due to different Hox genes expression'- since the authors show differential Hox gene expression, this statement should be moderated. The text describing this observation also suggests that the genes that are differentially expressed early and late in mouse and chick are different. However, many of the genes upregulated in the later chick stage are also upregulated in mouse, eg Wnt5a, Greb1, Fgf8, Figf18, Cyp26a1 (see Wymeersch et al., 10.1242/dev.168161). It would be good to comment on this.

7. Can the authors consider whether 'persistence' of cell tracks is the optimal term? It could suggest persistence of movement rather than just their continued observation over time. Is 'longevity' a possible alternative?

Essential Revisions

1. Many of the concerns can be addressed by text changes and minor changes to the existing data presentation as detailed in the full reviews.

2. The main problems are with the bioinformatic analysis, where both reviewers highlight issues where the analysis is too superficial and needs further work.

3. Additional imaging quantification is required to support their findings.

We thank the editor for the suggestions of improvement and have now thoroughly revised the text including new analyses as suggested. These are described below.

Reviewer #1:

The paper entitled "Dynamics of primitive streak regression controls the fate of neuro-mesodermal progenitor cells in the chicken embryo" from Guillot and colleagues aims to characterize in the context of developing chick embryos the cellular dynamics of a population of stem cells, recently identified, called Neuro-Mesodermal Progenitors (NMPs). The authors provide several lines of evidence supporting their claims that overall look interesting, however, additional imaging quantification and more comprehensive analysis of their transcriptomic datasets are required to support their findings.

Specifically, the authors generated an important scRNA-seq dataset of micro-dissect regions of the chick embryos across developmental time points from stage 5 to 35-somites. Despite the dataset represents a crucial part of the paper supporting the authors' findings of this NMP population, the analysis is not deep enough and does not report several controls and markers.

We agree with the reviewer and in the revised version, we now present a thoroughly revised analysis of the single cell RNAseq dataset as described below.

In figure 2 the authors should show heatmap of the marker genes used for classification. Louvain markers genes that define each cluster should be showed in a supplementary not as an unreadable table, instead, a dotplot or heatmap would be better. It is unclear whether Louvain clustering has been used to label the data, did the authors find that each Louvain cluster correspond to a defined cell identity?

We have now redone the entire analysis using Leiden clustering algorithm which performed better than Louvain at identifying tissue-specific clusters. As requested, we now present heatmaps showing the top 10 genes differentially expressed for each cluster in Figure 3 – Supplementary Figure 1 (for chicken) and in Figure 3 – Supplementary Figure 3 (for mouse). We have moved the gene tables originally presented in supplementary figures to a Supplementary File 2 which also includes dot plots showing the expression of genes used for identification of the clusters. We also discuss more in detail how the different clusters were identified at each developmental age in the revised text.

The pseudotime analysis is incomplete and superficial. Diffusion pseudotime strongly depends on the root cell selection. RNA velocity analysis and/or another trajectory analysis are required to better show lineage trajectory. Figure 4 L-K, the pesudotime time is not progressively ordered and it would be advisable to perform an alternative analysis to visualize lineage trajectory. Furthermore, gene expression trends should be identified and further discussed over those trajectories. The supporting information about the classification analysis using the mouse dataset is incomplete.

We now present a novel analysis of the developmental trajectories using the Waddington-OT pipeline developed by the Broad Institute (Schiebieger et al., 2019). This analysis, which is based on optimal transport, shows that the cells we identify as NMPs are indeed the most likely ancestors of both the Neural and PSM cells in the datasets. By analyzing the transport maps, this allows us to show the self-renewal of the NMP cell population and its maturation during axis elongation. We also use this tool to identify transcription factors associated with these different fates. We have added novel figures (Figure 4 and Figure 4- Supplementary Figure 1) to the revised text to describe these new results.

In addition, we have performed a Gene Set Enrichment Analysis of the early and late NMP clusters which shows an enrichment in genes associated to epithelium to mesenchyme transition in the late clusters in both species (shown in the new Figure 5). We have also completed the supporting information on the mouse dataset.

– In figure 1 the authors show SOX2-T double staining. Panels A-B-C represent the first evidence of SOX2-T co-expression in stage 5 chick embryos. The merged image is not provided together with the single panels A-B and only appear in panel G of the same figure, this is confusing.

In the revised version, we now show the merge of SOX2 and T expression in Figure 1C instead of showing only the double-positive cells as in the previous version. We also show a blow-up of the sections of the NMP and PMP regions in Figure E’-F’ and E’’-F’’ to better illustrate the difference between these two regions which both contribute to the paraxial mesoderm as shown by the MSGN1 antibody labeling in the ingressed mesoderm.

Panel B has two scale bars.

This was corrected.

The authors tried to show SOX2-T co-expression by thresholding and masking SOX2-T signals in panel C and Suppl. Figure 1. The authors should clearly explain how this thresholding is obtained. Is this manually set? Is this set based on global or local signal intensity distribution? Otsu thresholding?

Thresholding of the SOX2/T cells shown in Figure 1C and Figure 1 – Supplementary Figure 1G was done manually with image J. We have now replaced the subtracted image originally shown in Figure 1C by an image showing the merge between T and SOX2 staining which clearly identifies the double positive cells without thresholding. We have nevertheless left the subtracted image with the manual thresholding performed with image J in Figure 1 – Supplementary Figure 1G in the revised version. We also explained the thresholding in the legend of this figure and material and method section.

In order to avoid artifacts due to thresholding, the authors should also display a scatter plot showing nuclear intensity values for SOX2-T.

Since all the thresholding was performed manually, we could verify that virtually all cells in the images (shown in Figure 1- Supplementary Figure 1G) were indeed expressing both signals.

FACS analysis and quantitative assessments of the percentage of cells would further strengthen the authors' claim.

While this is an interesting suggestion, this experiment would be very challenging as the number of SOX2/T double positive cells is expected to be a few hundreds at the stages discussed. Thus, getting quantitative data on such a low number might be difficult. Moreover, since both markers are nuclear, this would require sorting after fixation and labeling of the cells which is much trickier than sorting for live cells labeled with a membrane marker. In addition, SOX2 and T are expressed in overlapping opposite antero-posterior gradients in epiblast cells with significant co-expression in the epiblast adjacent to the anterior PS. The positioning of the gates for FACS analysis would be somehow arbitrary as neither marker is truly specific for the NMP population.

– The authors should validate SOX2-T expression in quail embryos

Recent work from the Benazeraf laboratory has examined the distribution of T-SOX2 cells in the quail embryo and their distribution is similar to that we observe in the chicken embryo (Romanos et al., bioRxiv; doi: https://doi.org/10.1101/2020.11.18.388611).

– The authors should better discuss the mechanical forces that are implicated in epithelium to mesenchyme, and in the posterior gradient of proliferation that counteracts ingression in the anterior PS epiblast. Is this affecting/related to WNT or hyppo pathway?

While this is indeed an interesting question, our paper is mostly focused on the lineage and fate of NMP cells in the chicken embryo. We have no experimental data addressing the mechanical properties of ingressing cells and we are not aware of such data available in the literature for the stages considered in our study. Thus, we feel this discussion is beyond the scope of the paper.

– Can the authors speculate about the conservation of this phenomena in the human context? What is the evidence that this conserved in human?

Several groups including ours have reported that SOX2/T cells showing a similar identity to NMPs can be obtained from human pluripotent cells (ES/iPS) in vitro after Wnt activation. We have reanalyzed our human dataset from SOX2/T cells differentiated in vitro from iPS cells (day1 of Diaz-Cuadros et al., 2020) and observed that the list of human SOX2/T cells signature genes was much more similar to that of chicken and mouse when looking at the genes differentially expressed with the undifferentiated iPS only rather than with all the clusters as performed in the original publication. This revised list of human NMP signature genes is shown in Supplementary File 4 together with the genes conserved with chicken and mouse NMPs. We find a significant degree of conservation among the signature NMP genes in the three species and discuss these comparisons in a new paragraph in the revised text.

– The authors speculate about in-vitro generated NMPs, did they try transplantation of in-vitro generate cells? Can they compare the transcriptomic signature of the human in-vitro generated NMP with the chick counterpart?

We have not tried such transplantations of in vitro derived NMP cells and we believe this is beyond the scope of this work. Such work is however described in the following paper: “The Chick Caudolateral Epiblast Acts as a Permissive Niche for Generating Neuromesodermal Progenitor Behaviours Peter Baillie-Johnson, Octavian Voiculescu, Penny Hayward, and Benjamin Steventon”.

As suggested and discussed above, we have now compared the transcriptomic signature of cells of the chicken and mouse NMP clusters with that of the human NMP-like cells. This is now shown in Supplementary File 4.

– The tables in Suppl. Figure 4 should be a supplementary table.

We have moved the gene tables shown in supplementary figures to the Supplementary File 2 in the revised MS.

– In Suppl. Fig5D there is not scale bar to assess classification efficacy and material and methods do not describe how this analysis has been performed.

We apologize for this mistake, and we have now added it to the figure. The details of the analysis is presented in the material and method section: “Machine-learning classification of cell states”. The classification of the different cellular states was performed using an LDA classifier as described in Diaz-Cuadros et al., in vitro characterization of the human segmentation clock. Nature 580, 113-118, doi:10.1038/s41586-019-1885-9 (2020), which we now cite in the material and methods section.

– The authors should substantially revise scRNA-seq data.

As described above, we now provide a much deeper analysis of our single cell RNAseq data in the revised MS. We improved our analysis by changing our clustering method to Leiden that allows better identification of cell identities in the data. All the figures were remade with Leiden clustering.

We also provide an extended discussion of the datasets and the genes that were used to identify the various clusters in the revised text, as well as heat maps and dotplots of the genes used to identify the clusters in Figure 3 – Supplementary Figure 1 (chicken) and 6 (mouse) and in Supplementary File 2. In Figure 3F-M, we now show Diffmap plots to better illustrate the differentiation trajectories of the NMP cells.

We also performed a new analysis of the data using the Waddington-OT pipeline (Schiebieger et al., 2019). Unlike pseudotime, this analysis does not impose to select a root for the trajectories and considers the actual developmental time. This analysis shows that the NMP cells are the most probable common ancestors to the PSM and Neural cells in both chicken and mouse datasets. It also identifies a developmental trajectory in the NMP lineage which corresponds to their maturation during axis elongation. Using transport matrices, we could also quantify the contribution of NMPs to PSM, NT and NMPs and found equivalent contributions to these three lineages. This analysis therefore strongly supports the self-renewing properties of these cells. The WOT analysis also identified a number of interesting transcription factors associated with the NMP, PSM and neural fates. Expression patterns of a selection of transcription factors enriched in the NMPs is presented in Figure 4 – Supplementary Figure 1 and the extensive lists are provided in Supplementary File 3. We now also provide the expression levels of some genes involved in the NMP differentiation toward the mesodermal and neural cellular states in the new Figure 4 and in Figure 4 – Supplementary Figure 1. The results of the WOT analysis are described in a new paragraph in the revised text and shown as a new Figure 4.

We have also performed a Gene set enrichment analysis (GSEA) to compare the NMP early and late clusters identified using Leiden algorithm. This revealed an enrichment in genes associated to epithelium to mesenchyme transition in late vs early NMP clusters in both species. This analysis is now shown in Figure 5.

These new analyses led to new figures 3, 4 and 5 and Supplementary Figures 4, 5, 6, 7, 8 and 9 as well as new Supplementary Files 2, 3, 4 and 5.

Reviewer #2:

In this manuscript, the authors aimed firstly to prospectively identify cells with dual neural and mesodermal fate. Previous studies in mouse had determined that such dual-fated cells existed, and had narrowed down their location to subregions of the axial progenitor zone co-expressing Sox2 and T. However, these studies had not shown that individual cells in the Sox2/T positive region were dual-fated. The authors use three approaches: live imaging, and two prospective clonal labelling studies of the Sox2/T positive region to show, for the first time, that indeed this region harbours dual-fated cells. This is an important missing piece of the current volume of work on NMPs. Whilst the rationale for the experiments are clear, it might be better to present this work more in the context of previous studies. It seems a little harsh to dismiss population fate mapping studies as 'failing to identify' NMPs rather than complementing the clonal analysis by identifying regions potentially harbouring NMPs. In addition, there are interesting parallels with the clonal analysis carried out in mouse. For example, not all NM clones contribute to the tail bud; the observation in live imaging studies that NMPs first contribute to the neural tube and later to mesoderm is also similar to the observations from mouse retrospective clonal analysis; this could perhaps be discussed more fully.

We agree and have revised the text along the lines suggested by the reviewer.

Next, the authors carry out single cell RNA-seq analyses of three stages of chick axial elongation, and compare these with existing datasets in mouse, adding data from a later stage of mouse embryos. This aspect of the study has again confirmed and extended previous studies, showing both similarities and differences between chick and mouse NMPs. Trajectory analysis suggests progression of NMPs towards neural and mesodermal lineages in both chick and mouse. This is a valuable resource for the community. However once again it would be good to present this aspect of the work in the context of previous studies showing maturation of NMPs over time and an increased mesenchymal transcriptomic signature of NMPs at late compared to early timepoints.

We agree and have now extended our analysis and compared the identified transcriptomic signature of the NMP clusters of the chicken and mouse datasets to a set of published NMP transcriptomic signatures from mouse (Gouti et al., 2017, Dias et al., 2020, Guibentif et al., 2021; Diaz-Cuadros et al., 2020) and human (in vitro derived) NMPs (Diaz-Cuadros et al., 2020). We also compare the signature of the early and late clusters identified in our analysis to early and late NMP signatures described by Gouti et al., (2017) and E8.0 NMPs from Dias et al., 2020. We also separately compare the signature of the early and late NMPs to the genes up and down regulated in small dissected regions of the NMP from the Node streak border (NSB, E8.5) and the Chordo neural hinge (CNH, E10.5) in mouse from Wymeersch et al., 2019. We added a paragraph describing these comparisons in the revised text and present these data in Supplementary Files 4 and 5.

Finally, the authors carry out dynamic cell tracking and show that NMPs ingress later, and proliferate faster than more posteriorly-placed cells in the primitive streak, leading to exhaustion of the posterior populations and long-term retention of the NMP population. This part of the study is well-executed and contains important new information that advances our understanding of both NMPs and axial progenitors as a whole.

Suggestions for strengthening the science

1. The results presented in Figure 1 are a little difficult to interpret as there is some background fluorescence in Sox2 and Msgn1 staining. Perhaps the thresholding strategy could be described in the supplementary data to support statements about Sox2 positive time of emergence and the location of Msgn1 positive/Sox2 negative axial levels.

We now provide higher magnification images of the T⁺/sox2⁺ and T⁺/SOX2^- and their corresponding MSGN1 regions as Figure 1 E’-F’ and E’’-F’’. Without thresholding, the images shown clearly illustrates the difference between the two paraxial mesoderm-producing regions which both exhibit MSGN1 positive nuclei in the mesodermal layer while SOX2 is only seen in the epiblast of the anterior (level 3, NMP) and not in the posterior (level 4, PMP) regions. This shows that the paraxial mesoderm progenitor domain can be subdivided in 2 progenitor domains both producing paraxial mesodermal cells (MSGN1+).

It would be good to cite the source of the fate mapping information locating NMP/PMP/LPP/EMP in Figure 1D- is it this study or another?

The fate mapping used was from Psychoyos and Stern, (Development, 1996) who generated a very detailed fate map of the PS of the chicken embryo encompassing the stages described here. We have added this information to the revised text.

The section at level 5 in Figure 1E is named LPP in 1E but corresponds to a region marked 'EMP' in Fig1D. Together these two factors make it hard for the reader to judge where the posterior limit of Sox2/T positive cells is, relative to Msgn1 and to regions of differential fate.

We apologize for the oversight. This study is focused on the fate of paraxial mesoderm and we made no attempt to distinguish between the lateral plate and extraembryonic mesoderm territories of the posterior primitive streak. We have now relabeled the posterior primitive streak territory as LPP/EMP on Figure 1E. This corresponds to the territory devoid of MSGN1 expression.

Also, as a minor point, the text (p5) says there are 'low levels [of TBXT] in the node' but the expected high levels in the node/ emerging notochord are visible in Fig1E sections 1-2.

It is indeed right that the node shows high TBXT expression but the level of TBXT in the epiblast around the Node is low as seen in our quantifications and transverse sections in Fig1E sections 1-2. We meant 'low levels [of TBXT] in epiblast around the node’ and have changed the text accordingly.

2. In Figure 2, the clustering shows two datasets, (which are presumably batches?) of cells for each stage in chick (st5 1 etc). However, the colour coding does not allow the reader to distinguish these. Do the batches occupy different positions in the UMAP plot? How was batch correction done?

The 2 datasets are technical replicates. We agree that the color coding is not properly showing the heterogeneity of the data and we now provide a Figure 3F and in Figure 3 – Supplementary Figure 1 with a new set of colors. As seen in the new figures, the 2 datasets do not occupy different parts of the UMAP. In most cases, we used the module Bbknn for batch correction, but for the WOT analysis, we used the Combat module. We have added this information to the M&Ms section.

3. The cells annotated as NMPs in chick show elevated expression of Hes5 (6-somite stage) and Dll1 (combined dataset) relative to the cells annotated as PSM or neural (Figure 3, S4,5). Also, some of the NMP signature markers are expressed in a higher percentage of the NMP population in mouse compared to chick. This suggests that the cluster annotated as NMPs in chick also includes the midline primitive streak. The absence of Sox2 from this dataset is not, in my opinion, especially troublesome as it is expressed at low levels. However, it means that the chick single cell transcriptomes cannot be independently verified as NMPs and distinguished from midline streak cells.

This is correct, as in our immunostaining we see that T-SOX2 double positive cells are also present in the anterior primitive streak (in the groove) at stage 5HH (see Figure 1C and 1E level 3). This suggests that the distribution of NMPs might be slightly different in chicken compared to mouse embryos. This is supported by the observation that we cannot identify any PS specific cluster, although we cannot rule out that this due to the limited resolution of the dataset.

Perhaps the mouse data can help here. Instead of using the chick as a starting point to define the NMP phenotype, if the authors can work back from cells dissected from the Sox2/T positive regions (ie from Dias et al., eLife 2020 and Gouti et al., Dev Cell 2017), as well as the NMP annotations in the mouse atlas data used here (Pijuan-Sala et al.,), how well do the chick cells correspond to the mouse ones? If a narrower definition of NMPs is used in chick, does this remove any of the variation seen between mouse and chick?

We appreciate the suggestion which in principle could be a good way to present the data. The problem we are facing here, which we did not sufficiently highlight in the previous version of the paper, is that chicken cells were sequenced with Indrops, which provides limited depth, whereas mouse cells were sequenced with SMARTseq (Gouti et al.,) or 10x (Pijuan-Sala, Dias et al.,) which allows for more sequencing depth. This is probably the main reason why we do not identify Sox2 or Nkx1.2 in our chicken dataset. This is also why it is difficult to identify the chicken NMPs based on the mouse NMP signature genes. However, the most stringent definition of NMPs is not based on expression of marker genes (in fact, we don’t even know whether SOX2/T cells are NMPs) but is based on their ability to give rise to both neural tube and PSM. This is the criterion we use to define NMPs in this analysis. Our new WOT analysis strongly argues that in both chicken and mouse, cells of the identified NMP clusters are ancestors to the PSM and Neural Tube.

We now present a comparison between the NMP signature we identified in chicken and mouse embryos and lists of genes identified as NMP signatures in mouse embryos and human NMP-like cells differentiated in vitro in the following papers: Gouti et al., Dev Cell 2017; Wymeersch et al., 2019; Dias et al., eLife 2020; Guibentif et al., Dev Cell, 2021, Diaz-Cuadros et al., Nature 2020. This analysis shows significant conservation of genes between these different lists. While only 38 genes are conserved in 4 or 5 of the 6 datasets compared. The most striking characteristic of this list is the strong enrichment in genes of the Wnt pathway, consistent with the requirement for Wnt signaling for NMP differentiation in vivo and in vitro. We also find an enrichment of genes involved in glycolysis in line with recent papers describing higher glycolytic activity required for Wnt signaling in the tail bud (Oginuma et al., 2017, 2020; Bulusu et al., 2017). This new analysis is detailed in a new paragraph in the revised text and described in Supplementary File 4 and 5.

4. The mouse NMP trajectory analysis (Figure 3M) shows two arrows. What do they represent? if just pseudotime, it looks more like a radiation in all directions from the centre. Interestingly this would suggest NMPs also progress (upwards on the Y axis) as NMPs in pseudotime. Is this effect seen in the chick NMPs?

As suggested, we now present diffusion maps plots in Figure 3 that better show the pseudotime trajectories toward the neural and mesodermal fates in chicken and mouse cells. Indeed, NMP also exhibit changes in pseudotime that parallel developmental time in both species although this is more subtle in chick than in mouse probably because of sequencing depth. We also performed a new analysis of the mouse and chicken datasets using the Waddington OT pipeline (Schiebieger et al., 2019). This analysis shows that the NMPs are the most probable ancestors of the PSM and neural cells. It also identifies a trajectory in the NMP cluster (confirming the trajectories observed in pseudotime) correlating with the age of the samples, suggesting maturation of these cells. We now describe these new analyses in detail and have added a new figure 4 to present this data.

4. The annotations of chick versus mouse early and late clusters are confusing, In chick, there are two early clusters (coloured the same) annotated nmp-1 and nmp-2, but these do not correspond to the two earlier stages. What do they correspond to? Some discussion of the heterogeneity would be appropriate- or possibly a reanalysis of the NMP annotation in point 3 above may highlight that one of these clusters is less NMP-like. In the mouse, there is one early subcluster and two late ones. These are also annotated nmp-1 and nmp-2. A different nomenclature should be used to highlight that they are not the same as the chick ones. Unlike the chick clusters, these seem to correspond to different stages of development.

The analysis of the early and late NMP clusters has been revisited using Leiden clustering (instead of Louvain initially). We now only find one early and late clusters in both species. We perform a GSEA analysis using the gene set “Epithelium to mesenchyme transition” which shows enrichment in the late clusters in both species. We also compared these early and late clusters to other similar datasets published in the literature (Gouti et al., 2017; Wymeersch et al., 2019; Dias et al., 2020; Dias-Cuadros et al., 2020). This part has been modified in the revised text and a new figure 5, Supplementary File 4 and Figure 5 – supplementary Figure 1 have been generated with the new data.

Suggestions for improving the manuscript

1. The statements that no previous studies have prospectively identified neuromesodermal-fated cells (abstract, introduction p4, discussion p10) should be modified. Forlani et al., Development 2003 10.1242/dev.00573 and Wood et al., BioRXiv https://doi.org/10.1101/622571 show dual-fated cells in mouse and chick.

We are well aware of the Forlani paper, where they perform single cell injections of HRP in epiblast cells. However, this study was intended for a different purpose (comparing the anterior boundaries of the clones with the timing of Hox gene activation). It is true that they show a few bipotent clones in their Figure 7 but this work was done before the identification of the NMPs in 2009 and in no place in the text do they refer to bi-potential precursors. We nevertheless introduced and discussed this reference in the text.

With respect to the Wood preprint, I don’t think they demonstrate the existence of dual-fated cells. Their data arguing for the dual fate is shown in Figure 5a-c which shows cell fate tracked for only 15 hours (ie up to the 5-somite stage approximately). Their designation of the neuro-mesodermal region is quite arbitrary. The main argument in support of a dual fate of these cells is that their descendants span neural and mesodermal territories after 15 hours when backtracked from time lapse movies (white bars on Figure 5c). However, at these stages, these two territories are not clearly separated into neural tube and paraxial mesoderm and thus the designation of the final fate is not supported by any anatomical or molecular landmark. This study is extremely preliminary and lacking critical details to evaluate the findings. I don’t think it would be fair to consider that this work (which is essentially contemporary to ours) first identified dual-fated cells in the chicken embryo. We nevertheless discuss this reference in the revised version of the manuscript.

I think that it would be fairer in the context of previous studies to cite the population or oligoclonal studies of Iimura and Pourquié, (2006) 10.1073/pnas.0610997104, Cambray and Wilson, (2007) 10.1242/dev.02877 and Brown and Storey, (2000) 10.1016/s0960-9822(00)00601-1 as showing potential locations for cells of dual neuromesodermal fate.

We were already discussing these studies in the previous version of this paper but have rewritten this part. We were not trying to diminish the merit of these oligoclonal studies, which are very elegant and compelling. Nevertheless, we believe it is fair to say that due to their design, these studies cannot prospectively identify bipotential cells, only territories.

2. The words 'bipotent/monopotent' used throughout the manuscript is not appropriate for the studies here, which are demonstrating fate and not potency. I would suggest 'dual fate'.

We respectfully disagree about using dual fated to describe NMP cells as dual fated means that the two fates can be observed. Thus, we believe using this term would only be appropriate for the description of a subset of the cells identified by the lineage analysis (and we do use it in some places) as in these experiments we can visualize the fate of the cells. However, in many cases, only one of the fates is realized even though the precursors (the NMPs) appear to be bipotent at the population level. For the scRNAseq analysis, the cells of the NMP cluster cannot be called dual-fated as the analysis does not reveal the actual fate of the cells. We believe the cells identified as NMP using this approach should be called bipotent.

3. NMPs are presented here as 'stem cells' when the transcriptome analysis shows they mature with time. It would be good to define exactly in what sense the term is being used- it is valid since the manuscript describes an enduring progenitor producing neural and mesodermal tissue but should be used with care.

We define NMPs as stem cells because they can give rise to differentiating progeny and self-renew. We now add further evidence for a self-renewing behavior with the WOT analysis transport maps which show that NMP contribute significantly to the NMP population itself over time. This is now shown in the revised Figure 4C-D. The fact that NMPs can also mature is a situation frequently observed for other stem cell types (fetal vs adult satellite cells or hematopoietic cells for instance). We have however removed stem cell from the title of the revised paper.

4. p5. 'We identified 7 clones containing cells expressing the same barcodes' – could this text be modified to clarify, e.g. 'expressing unique (or clone-specific) barcodes'?

This has been modified accordingly.

5. p7 para 2 'the NMP lineage' should be changed to something not implying lineage, e.g. 'cluster'

We agree and have removed NMP lineage from the revised text and replaced it by the appropriate clusters.

6. Figure S7 is entitled 'NMP early and late clusters are not due to different Hox genes expression'- since the authors show differential Hox gene expression, this statement should be moderated. The text describing this observation also suggests that the genes that are differentially expressed early and late in mouse and chick are different. However, many of the genes upregulated in the later chick stage are also upregulated in mouse, eg Wnt5a, Greb1, Fgf8, Figf18, Cyp26a1 (see Wymeersch et al., 10.1242/dev.168161). It would be good to comment on this.

We agree and have changed the title of what is now Figure 5 – supplementary Figure 1 with “Characterization of the early and late NMP clusters”. We have also performed a series of comparisons between the differentially expressed genes from the early and late NMP clusters identified in our chicken and mouse datasets and with the datasets from Gouti et al., 2017, Wymeersch et al., 2019, Dias et al., 2020 and Dias-Cuadros et al., 2020. We do find similarities between the gene signatures identified in our mouse and chicken early and late NMPs clusters and those identified in these studies and the results are now shown in the Supplementary File 4.

7. Can the authors consider whether 'persistence' of cell tracks is the optimal term? It could suggest persistence of movement rather than just their continued observation over time. Is 'longevity' a possible alternative?

We have made the corresponding change.

Author details

1. Charlene Guillot

   1. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
   2. Department of Genetics, Harvard Medical School, Boston, United States
   3. Harvard Stem Cell Institute, Boston, United States

   Contribution

   Conceptualization, Resources, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   charlene_guillot@hms.harvard.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9214-7509

2. Yannis Djeffal

   1. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
   2. Department of Genetics, Harvard Medical School, Boston, United States
   3. Harvard Stem Cell Institute, Boston, United States

   Contribution

   Formal analysis, YD performed the trajectory analysis in silico with CG and OP

   Competing interests

   No competing interests declared

3. Arthur Michaut

   1. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
   2. Department of Genetics, Harvard Medical School, Boston, United States
   3. Harvard Stem Cell Institute, Boston, United States

   Contribution

   Software, AM wrote the code to analyze the cell trajectories and their speed

   Competing interests

   No competing interests declared

4. Brian Rabe

   1. Department of Genetics, Harvard Medical School, Boston, United States
   2. Howard Hughes Medical Institute, Boston, United States

   Contribution

   Resources, BR constructed the viral library and prepared the viral solutions

   Competing interests

   No competing interests declared

5. Olivier Pourquié

   1. Department of Pathology, Brigham and Women’s Hospital, Boston, United States
   2. Department of Genetics, Harvard Medical School, Boston, United States
   3. Harvard Stem Cell Institute, Boston, United States

   Contribution

   Supervision, Funding acquisition, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   pourquie@genetics.med.harvard.edu

   Competing interests

   No competing interests declared

• 3,996

  Page views

• 441

  Downloads

• 15

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.