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

ABSTRACT The elongating mouse anteroposterior axis is supplied by progenitors with distinct tissue fates. It is not known whether these progenitors confer anteroposterior pattern to the embryo. We have analysed the progenitor population transcriptomes in the mouse primitive streak and tail bud throughout axial elongation. Transcriptomic signatures distinguish three known progenitor types (neuromesodermal, lateral/paraxial mesoderm and notochord progenitors; NMPs, LPMPs and NotoPs). Both NMP and LPMP transcriptomes change extensively over time. In particular, NMPs upregulate Wnt, Fgf and Notch signalling components, and many Hox genes as progenitors transit from production of the trunk to the tail and expand in number. In contrast, the transcriptome of NotoPs is stable throughout axial elongation and they are required for normal axis elongation. These results suggest that NotoPs act as a progenitor niche whereas anteroposterior patterning originates within NMPs and LPMPs.

(A) Unsupervised hierarchical clustering of all genes that showed ≥1.5 fold change across E8.5 samples. Three major clusters were identified: up in RN ± other samples (pink), up in St1 ± CLE, NSB (blue) and up in St5 (yellow). In the first two clusters, we further distinguish several subpatterns (see Table S1). (B) Overview of enriched GO-terms and KEGG pathways from patterns in A. Figure S4. Comparison of LPMPs at E7.5 and E8.5. (A) Schematic overview of dissected and compared regions at E7.5 headfold stage (HF). Since E7.5 P overlaps with PP gene expression, samples were compared as follows: E7.5 P versus A (red), E7.5 PP versus A (blue) and E7.5 PP versus P (shades of yellow). Since PP is included in the P sample and thus no gene will be ON in PP and OFF in P, we used the values of a known marker of the PP region, Bmp4 (Lawson and Wilson, 2016), as the minimum cut-off for transcripts enriched in PP versus P. (B) Overlap of E7.5 sample comparisons with genes specific for E7.5 PP shown in (C) 'Included' genes, i.e. ≥BMP4 values, are most likely upregulated in PP, while some of the remainder may also be upregulated there, although less strongly.   Table S5 for gene lists). (C-D) Up-or downregulated DEGs at E8.5 and E10.5, shown in A and B compared with enriched NMP genes (versus nascent mesoderm), obtained in a parallel single cell analysis (Gouti et al., 2017).  Table S6. (C) A combined list of upregulated genes during mid-trunk formation, with the overlap showing in (D). Red, DEGs up in E10.5 CNH and down in E8.5 NSB. Blue, genes that peak at E10.5 (taken from the STEM analysis pattern in A). Yellow, genes with similar pattern to Wnt5a expression (taken from correlation analysis in Bb). (E) Overview of GO terms and KEGG signalling pathways in this combined list (see also Table S7).     (A) DiI labelling of the crown cells (red) in an E8.5 embryo: wholemount lateral (Aa), posterior view (Ab) validate the accuracy of initial labelling (n embryos =3). (B) Sox2/T immunostained section of embryo labelled with DiI in the ventral node layer after 48h culture (n embryos =2). White lines surround autofluorescent blood cells. (C) Cell death in NSB (Ca-Cb) or CLE (L2) (Cc-Cd) in electroporated embryos. After two hours ex vivo culture, GFP + cells (green) were observed at the electroporation site. Cell death, shown by DRAQ7 dye uptake (magenta), is located primarily at the ventral side of the embryo (arrowheads).

Figure S12. Fate of electroporated cells.
Cell fate was examined in CLE-electroporated embryos after 24 (A) or 48h (B) ex vivo culture. In addition to the expected fate, the hindgut was labelled in all examined embryos (arrows; n embryos =4), suggesting that also hindgut progenitors were electroporated. nt, neural tube. (C) Fate in a class II NSB-electroporated embryo. After 24h, GFP + cells were found in the neural tube (n) and paraxial mesoderm (m) of this embryo. (D) Contribution to the axial tissues and tail bud in electroporated samples.

Figure S13. Effect of ventral cell layer removal at the NSB or St3.
(A) Experimental procedure schema illustrating the removal of the endoderm layer. A sharp glass needle was inserted posteriorly and pushed anteriorly to peel away the ventral endodermal cells. This ventral layer was further trimmed, after which DiI was poured on the damaged site, labelling the cell layer above as well as the endodermal cells. (B) Fate and phenotype after ventral layer removal in the NSB (Ba-Be) or St3 (Bf-Bk) after 48h ex vivo culture. (Bb-Bc) Removal of the ventral node layer resulted in abnormal growth (similar to class II phenotypes in electroporation, n embryos =5, Fig. 8C). DiI-labelled cells were found in the notochord and neural tube (Bd-Be). Control embryos grew normally (n=3) and DiI was found in somite and gut tissues (Bg-Bh). Red dots, DiI (pink, background due to high exposure). nt, neural tube; noto, notochord; som, somites; white arrowheads, DiI label in ventral neural tube. Table S1. Full list of DEGs in the primitive streak and St5 (supplementary to Figure 2). (Tab 1) Unique DEGs for St5, not shown in Figure 2A. (Tab 2) List of unique DEGs for each region at E8.5. Columns allow for pairwise comparison between two regions (1, ≥1.5 fold upregulated; -1, ≥1.5 fold downregulated; 0, no significant change).

SUPPLEMENTARY TABLES
Click here to download Table S1 Table S2. Hierarchical clustering of DEGs ≥1.5 fold change across the primitive streak (supplementary to Figure S3). Tab 1 shows all DEGs ≥1.5 fold changed across the E8.5 sample set that were used in hierarchical clustering of Figure S3 and analysed for GO-terms/KEGG pathways. Three major clusters were identified and several sub-clusters could be distinguished. Subsequent tabs show all GO-terms and KEGG pathways for each pattern obtained from the STRING online database (accessed 10th April 2017).
Click here to download Table S2   Table S3. STEM analysis across different sample sets (supplementary to Figure 4 and S6). STEM analysis shows significant patterns in two datasets: (1) in the E7.5 embryo (A, P and PP samples) and (2) in comparable NMP-containing regions (E8.5 NSB to E9.5-E13.5 CNH, normalised to E7.5 P). Tables shows significant patterns, and the genes and enriched GO-terms within each pattern (p≤0.05, defined by permutation test in STEM).
Click here to download Table S3   Table S4. DEGs unique to and shared between LPMPs at E7.5/8.5 (supplementary to Figure 4). Full list of DEGs not shown in Figure 4C.
Click here to download Table S6   Table S7. Genes upregulated between E8.5 and E10.5 NMPs: construction of list, its annotation and comparison to other datasets (supplementary to Figure S6 and Figure S7). (Tab 1) Genes upregulated during mid-trunk formation in NMPs, with the overlap between different analyses shown. Red, DEGs that were upregulated in the E10.5 CNH, and down at E8.5 NSB. Blue, genes that peak at E9.5 (from STEM analysis). Yellow, genes with similar temporal expression pattern to Wnt5a. (Tab 2) Manual annotation of the combined list shown in Tab 1. STRING annotation (Tab 3) and GO-terms associated with the combined list (Tab 4). Tab  Click here to download Table S7   Table S8. Markers of notochord and node during axis elongation (supplementary to Figure S10). (Tab 1) Common gene expression in E8.5 RN, NSB and E9.5 to E13.5 CNH versus all other samples (dataset normalised to E7.5P). Data is separated into known and potential novel markers for the node/notochord. Red: averaged expression from different probes on the microarray. (Tab 2) Comparison between Tamplin et al., 2011. and samples containing NotoPs: enriched in E8.5 RN and NSB versus E8.5 samples ('spatial analysis'), and in E8.5 RN, NSB and E9.5-13.5 CNH ('temporal analysis', see Tab 1).  Table S9. Overview of electroporated embryos. Overview of electroporated embryos of CLE, primitive streak or NSB-targeted regions in E8.5 (2-5s) embryos. Table shows embryo number, corresponding genotype, plasmid and culture period. After 24/48h ex vivo development all embryos were assessed on their developmental features and assigned a phenotype class: N, normal; I, class I; II, class II. Table S10. Real-time qPCR primer list. Overview of the primers used in Fig. S1B.