Pitx2c orchestrates embryonic axis extension via mesendodermal cell migration

Essential revisions:

1) A more comprehensive characterization of the gastrulation phenotype is needed. It is concluded that MZpitx2c mutants exhibit morphogenetic defects without defects in patterning or cell fate specification. Yet, in Figure 1 bilateral somitic myoD1 staining is fused in the midline, suggesting axial mesoderm deficiency. Axial/chordamesoderm markers are also presented in Figure 2 without obvious gaps. The conclusion of normal patterning in MZpitx2c requires further evidence. Can gene expression profiling shed some light on this question?

To further investigate patterning of MZpitx2c mutants, we have analyzed the expression levels of dorsal organizer genes (n=4) and mesendodermal markers (n=20) in our two datasets. In agreement with previous reports (Essner et al., 2000; Faucourt et al., 2001), overexpression of Pitx2c can affect the expression levels of many genes (e.g., dkk1a and chd). However, in MZpitx2c mutants, the expression level of the vast majority of genes was unchanged at shield stage, including those encoding key transcription factors that regulate germ layer fate (e.g., sox32, gata5) and signaling molecules (e.g., ndr1, ndr2). These data further support our interpretation that early embryonic patterning is not strongly affected in the maternal zygotic mutant embryos. We have now included a new figure with these data (Figure 7—figure supplement 2A) and have described these observations within the Results section and Discussion section.

2) Are cell death and/or proliferation contributing factors to the observed morphogenetic defects? The mutant embryo with severe CE defects presented in Figure 1e seems to exhibit cell death. And in some images, there appear to be fewer cells in the mutants. If so, in addition to defective cell movements during gastrulation, Pitx2 may regulate cell proliferation and cell death, these cell behaviors need to be examined here to rule out such possibility.

To address this possibility, we performed phospho-Histone H3 immunostaining at late gastrulation and observed a significant reduction in the number of proliferating cells in MZpitx2c mutant embryos, which may indeed contribute to the phenotype. To address cell death, we performed cleaved Caspase-3 immunostaining at late gastrulation; however, no significant changes were observed between wild-type and MZpitx2c embryos. We now have included these data in Figure 2—figure supplement 1 with the corresponding description in the text.

3) Along the same lines, Pitx2 has been involved in differential cell sorting and regulates N-cadherin expression (Plageman, 2011; Kurpios et al., 2008). It is possible that Pitx2 may be regulating differential cell adhesion in zebrafish. This possibility should be at least considered in the context of the gene expression profiling data.

To determine whether Pitx2c can regulate differential cell adhesion, we looked for any transcriptional changes in key adhesion molecules in our microarray data. Indeed, our DAVID pathway analysis identified that a subset of genes upregulated in MZpitx2c mutant embryos were categorized as ‘cell adhesion’ genes (now in revised Figure 7—figure supplement 1). We expanded our analysis for additional cell adhesion genes and have listed them in a revised figure (Figure 7—figure supplement 2C). These data have also been described in the Discussion section of the text.

4) Convergence movements' analysis needs to be properly performed. To measure the dorsal convergence of mesendoderm populations the authors photoconverted cells in embryos injected with a photoconvertible kikGR RNA and measured the anteroposterior extension and width of the photoconverted population. Whereas this strategy is appropriate for analyzing extension in zebrafish, but not to measure convergence. In zebrafish dorsal convergence is mostly driven by dorsal migration of mesendodermal cells, thus a translocation of converted population towards the dorsal midline, rather than mediolateral narrowing of the labeled population itself should be assessed.

We thank the reviewers for this critical feedback and have now re-analyzed our data and performed additional experiments. We have changed Figure 3 accordingly, and now present two graphs with the quantification of anterior extension and the angle of dorsal convergence of photoconverted lateral mesendodermal cells. Furthermore, we have updated the Materials and methods section to reflect these new analyses.

5) Cell-autonomy experiments as currently presented are flawed and need to be carried out with an appropriate readout at appropriate time. The authors transplanted into wild-type hosts at 30% epiboly two populations of cells: wild-type cells and MZpitx2c mutant cells labeled with different fluorescent markers. They state, "After 3 hr post-transplant, we did not observe any differences in the two labelled populations." There are two problems with this conclusion. First, there is no quantification of these results. Second and a fundamental one, is that this assessment was done about 7 hours post fertilization (30% epiboly + 3 hours), whereas the earlier analyses of cell migration in MZpitx2c revealed no significant defects of mesendodermal cell migration during early gastrulation. Neither the authors have observed any defects with the random walk of endoderm. Rather it was the second phase of endodermal migration convergence, during late gastrulation that appeared affected. Therefore, the conclusion that "Pitx2 likely does not function cell-autonomously within the context of cell migration" is not supported by these experiments. To address the issue of cell-autonomy effectively, an assessment of these experiments should be carried out at late gastrulation, e.g. yolk plug closure. Moreover, other cellular defects, like cell shape and protrusive

activity, could be investigated as a readout in these cell autonomy studies.

We thank the reviewers for these important comments and agree that the most relevant time point had not been analysed. We have now performed live imaging of these cell transplants and analysed additional parameters of the transplantation experiments at late gastrulation (70-90% epiboly): transplanted cell protrusions, transplanted cell surface area and volume measurements, and quantification of the migration distance of transplanted cells along the anterior-posterior axis. We have also analysed these parameters at the late gastrulation to tailbud stage. These data have now been included in a revised Figure 6 and new Figure 6—figure supplement 1. These new experiments support the conclusion that the role of Pitx2c during mesendodermal migration is cell non-autonomous.

6) The proposed mechanism downstream of Pitx2c involving Cxcl12b-Cxcr4 signaling also requires additional experimental assessment. In images of sox17 expression marking endodermal and dorsal forerunner cells shown in Figure 2G, the authors note abnormal appearance of dorsal forerunner cells that appear scattered and fail to coalesce. It is surprising that the authors do not comment on the pattern of sox17 expression in endodermal cells, as abnormal anterior migration of endodermal cells and in particular their abnormal location closer to the animal pole, is a hallmark of defective Cxcl12b-Cxcr4 signaling reported by the Schilling laboratory (Science 2008). As the authors propose misregulation of Cxl12b as the main mechanism downstream of Pitx2, sox17 expression should be more closely examined. Likewise, what do endoderm derivatives look like in MZpitx2c mutants at 1dpf? Can any aspects of the phenotype be rescued by restoring normal Cxcl12b expression? This is an important point, given the above questions about sox17 expression pattern in MZpitx2c mutant. Currently, the proposed relationship between Pitx2c, Cxlcl12b and abnormal cell movements is merely a correlation.

We agree with the reviewers that the proposed relationship between pitx2c and cxcl12b is correlative at this time. We have tried to perform rescue experiments by injecting cxcl12b mRNA into MZpitx2c mutant embryos; however, these injections led to severe defects that precluded the analysis of endodermal rescue. Given that cxcl12 does not appear to be expressed ubiquitously, it is challenging to restore the wild-type expression pattern and level. We have however carried out additional analysis of endodermal derivatives in MZpitx2c mutants at 24 hpf, which revealed that the anterior endoderm remains much wider in mutants compared to wild types. Images and quantification of the width of the anterior endoderm are now shown in Figure 4.

7) The role of impaired cell division polarity is also not clear, especially the role of defective PCP signaling. The partial rescue of the orientation of cell divisions in MZpitx2c mutants by downstream target genes is not very convincing, despite the Chi-squared analysis. This is particularly true of the vangl2 rescue, shown in Figure 8F.

We have decided to remove these data from the manuscript, as we agree that they are not very convincing. We have kept the data showing that the orientation of cell division was disrupted in the MZpitx2c mutants and have discussed it in more detail in the context of current literature within the Discussion section to address this concern.