Sp Transcription Factors Establish the Signaling Environment in the Neuromesodermal Progenitor Niche During Axial Elongation

Reviewer #1 (Public review):

This is an important, interesting, and in-depth study examining the role of Sp5/8 transcription factors in maintaining the neuromesodermal progenitor (NMP) niche. The authors first used Sp5/8 double conditional KO mouse embryos to establish that these factors function in the NMP niche to promote trunk elongation. They then conducted extensive single-cell analyses on embryos of various genetic mutant backgrounds to unravel the complex and intricate interactions between Wnt signaling and Sp5/8. The key conclusion from these experiments is that Sp5/8 function within an autoregulatory loop crucial for maintaining the NMP niche. The authors went on to identify and characterize a novel enhancer element downstream of the Wnt3a coding sequence, which mediates the effects of Sp5/8 on Wnt3a expression. Overall, the data presented are compelling and of high quality, and the study offers a prime example of how a relatively small set of signaling pathways and transcription factors can function in concert to impart robustness to developmental processes.

Reviewer #2 (Public review):

Chalamalasetty et al. investigate the regulatory circuit of signaling molecules and transcription factors that drive the fate of neuromesodermal competent progenitors (NMCs). NMCs contribute to Sox2-positive spinal cord and Tbxt/Bra-expressing somitic mesoderm, and this choice is governed by the interplay between Wnt3a and Fgf signaling. The authors discovered that the transcription factors SP5 and SP8 participate in this process. Mouse genetics, in vivo development, and transcription factors profiling point to a model where SP5 and SP8 directly regulate Wnt3a expression to foster Tbxt-marked mesoderm formation at the expense of Sox2-marked neural ectoderm. Mechanistically, SP5/8 bind to an enhancer which the authors characterize: its activity depends on the presence of SP5, CDX2, TCF7, and TBXT binding sites, and it is activated only in primitive streak cells at E7.5, in NMP, and in caudal and somitic mesoderm, underscoring the tissue and stage-specific nature of this Wnt3a enhancer.

Moreover, the authors find that SP5/8 likely regulate the TCF7 association with the chromatin and compete for its binding to the TLE repressor.

The study is extensive, compelling, and well written. The combination of in vivo evidence with single-cell transcriptomics, transcription factors profiling, and in vitro regulatory element characterization is notable and builds a convincing picture of the action of SP5/SP8.

Here, I provide a series of comments and questions that, if addressed and clarified, could, in my opinion, improve the study.

(1) While Sp5 and Sp8 are both present in NMCs, their expression does not fully overlap. Sp5 is also detected in caudal and presomitic mesoderm, notochord and gut, while Sp8 overlaps with Sox2 in neural progenitors of the spinal cord and brain (Fig. 1D). Accordingly, Sp8 expression is also activated by the neural-promoting RA+Fgf. It is not easy for me to reconcile this non-fully overlapping expression pattern - and in particular the overlap of Sp8 and Sox2 - with the presumed redundancy (or similarity of function) described later. Sp5/8 dko NMCs show reduced Tbxt and expanded Sox2, indicating that SP8 also represses Sox2 or neural fate, an observation confirmed by Sp8 overexpression (Figure 4c). What is the explanation for this, and is the function of SP8 in Sox2-positive neural progenitors different from its Wnt3a-sustaining role in NMCs? Or what am I missing?

(2) I suggest that the authors show relevant ChIP-seq peaks in Figure 3 to lend credibility to the complicated overlapping Venn diagrams. I consider visual inspection of peak tracks as primary quality control of this type of experiment. A good choice could be the cis-regulatory elements at Sp5, Sp8, Tbxt, Cdx1, 2, 4 bound by TBXT and either CDX2, SP5, or SP8 (now referring to the Venn diagrams and the annotated peak table). On ChIP-seq visualization, in reference to Figures 5 and 7, I also suggest that the authors show the tracks of a negative control (IgG, non-related antibody, or better anti-flag in Sp5/8 dko). While I do not doubt the validity of these experiments, there are peaks in these figures bound by all factors tested that could be suspicious (even though, admittedly, they look like genuinely good TF peaks). A negative track would clearly show beyond any doubt that these are not suspect regions of positive unspecific signal caused by open chromatin, excessive cross-linking, or antibody cross-reaction.

(3) SP5 here is found as a direct inducer of Wnt3a expression, and accordingly positive regulator of Tbxt and mesoderm, caudal development. I find this in partial contradiction with a finding by the Willert group (PMID: 29044119). They show that "genes with an associated SP5 peak, such as SP5 itself, AXIN2, AMOTL2, GPR37, GSC, MIXL1, NODAL, and T, show significant upregulation in expression upon Wnt3a treatment in SP5 mutant cells". There, essentially, SP5 inhibits Wnt target genes. While the authors are aware of this and cite Huggins et al., I find that this deserves a better discussion addressing how opposite functions could be sustained in different contexts, if these really are different cellular contexts in the first place, or if this could result from different methodologies.

(4) The gastruloid experiment is nice, but I wonder whether there is any marker that the authors can use to show that other features of the gastruloids respond accordingly. For example, is the Sox2 expression domain expanded? And is there any unaffected marker to emphasize the specificity of the decreased Tbxt and Cdx2?

(5) SP5/8 seems to enhance the TCF7 occupancy at WRE. And then, SP5/8 appears to counteract the presence of TLE repressor associated with TCF7. While these two mechanisms are interesting, they are not necessarily interconnected. According to the still-established view, TCF7 should be associated with WRE even in the absence of the Wnt signal, when TLEs are also present on the locus. One could expect that SP5 competes with TLE, to decrease its presence on TCF7-bound loci, leaving the abundance of TCF7 binding unchanged. Yet, the authors also observe that the TCF7 association changes. What is the mechanism implied? Do they perhaps consider a TCF7L1 > TCF7 switch, and if so, what evidence exists for this?

(6) Along the same line as above, I wonder whether beta-catenin binding is also enhanced at these sites? Any TCF/LEF would require beta-catenin for gene upregulation.

(7) The authors write that "Small Tle peaks were identified at these WREs in WT cells, demonstrating that both repressive Tle and activating Tcf7 could be detected at active genes". However, ChIP-seq is a population assay, and it is possible - more plausible, in fact - that cells displaying TLE binding are not expressing the target genes.

Reviewer #3 (Public review):

Summary:

This is a well-done study. It shows, in a comprehensive manner, that Sp5 and Sp8 play essential roles in maintaining the complicated positive feedback circuitry needed for specification of neuromesodermal competent progenitors (NMCs) in caudal mesodermal development in murine embryos.

Strengths:

The developmental genetics, transcriptomic, and genomic survey of TF binding are all satisfactory and make a compelling story. The CRISPR deletion of the Wnt3a downstream enhancer clearly demonstrates that it plays an important role in the positive feedback circuit.

Weaknesses:

My only concerns are some of the language surrounding the mechanistic interpretation of the Wnt3a downstream enhancer and the relationship between TCF and TLE binding.