Zebrafish Pou5f1-dependent transcriptional networks in temporal control of early development

Abstract

• Time-resolved transcriptome analysis of early pou5f1 mutant zebrafish embryos identified groups of developmental regulators, including SoxB1 genes, that depend on Pou5f1 activity, and a large cluster of differentiation genes which are prematurely expressed.

• Pou5f1 represses differentiation genes indirectly via activation of germlayer-specific transcriptional repressor genes, including her3, which may mediate in part Pou5f1-dependent repression of neural genes.

• A dynamic mathematical model is established for Pou5f1 and SoxB1 activity-dependent temporal behaviour of downstream transcriptional regulatory networks. The model predicts that Pou5f1-dependent increase in SoxB1 activity significantly contributes to developmental timing in the early gastrula.

• Comparison to mouse Pou5f1/Oct4 reveals evolutionary conserved targets. We show that Pou5f1 developmental function is also conserved by demonstrating rescue of Pou5f1 mutant zebrafish embryos by mouse POU5F1/OCT4.

Abstract

The transcription factor Pou5f1/Oct4 controls pluripotency of mouse embryonic inner cell mass cells ( Nichols et al, 1998), and of mouse and human ES cell lines ( Boiani and Scholer, 2005). Although Pou5f1/Oct4-dependent pluripotency transcriptional circuits and many transcriptional targets have been characterized, little is known about the mechanisms by which Pou5f1/Oct4 controls early developmental events. A detailed understanding of Pou5f1/Oct4 functions during mammalian blastocyst and gastrula development as well as studies of the temporal changes in the Pou5f1/Oct4-regulated networks are precluded by the early lineage defects in pou5f1/oct4 mutant mice. To investigate Pou5f1-dependent transcriptional circuits in developmental control, we used the zebrafish ( Danio rerio) as a genetic and experimental model representing an earlier state of vertebrate evolution. Zebrafish have one pou5f1/pou2 gene ( Takeda et al, 1994) orthologous to the mammalian gene ( Niwa et al, 2008; Frankenberg et al, 2009). Both fish and mammalian orthologs are expressed broadly in tissues giving rise to the embryo proper during blastula and early gastrula stages, as well as in the neural plate ( Belting et al, 2001; Reim and Brand, 2002; Downs, 2008).

Zebrafish pou5f1 loss-of-function mutant embryos, MZ spg (abbreviated ‘MZ'), are completely devoid of maternal and zygotic Pou5f1 activity ( Lunde et al, 2004; Reim et al, 2004). MZ embryos have gastrulation abnormalities ( Lachnit et al, 2008), dorsoventral patterning defects ( Reim and Brand, 2006), and do not develop endoderm ( Lunde et al, 2004; Reim et al, 2004). In contrast to Pou5f1/ Oct4 mutant mice, which are blocked in development due to loss of inner cell mass, MZ mutant embryos are neither blocked in development nor display a general delay. Therefore, zebrafish present a good model system to identify specific transcriptional targets of Pou5f1 during development.

Our study aims to understand the structure, regulatory logic, and developmental temporal changes in the Pou5f1-dependent transcriptional network in the context of an intact embryo. Therefore, we investigated transcriptome changes in MZ compared with WT zebrafish by microarray analysis at 10 distinct time-points during development, from ovaries to late gastrulation. We identified changes in Pou5f1 target gene expression both with respect to their expression level and temporal behavior. We used correlation analysis to identify clusters of target genes enriched for genes with developmentally regulated expression profiles. This correlation analysis revealed a cluster of genes, which were not activated or were significantly delayed in MZ. Interestingly, there was also a large gene cluster with premature onset of expression in MZ.

Several targets activated by Pou5f1 encode known repressors of differentiation (RODs), of which we analyzed her3 in detail. Pou5f1 also activates several SoxB1 group transcription factors, which are known to act together with Pou5f1 in mammalian systems. Among the large group of genes prematurely activated in MZ, many genes encode developmental regulators of differentiation normally acting during organogenesis (promoters of differentiation—PODs). Our analysis of potential direct transcriptional interactions by suppression of translation of intermediate zygotic Pou5f1 or SoxB1 targets, enabled us to distinguish Sox-dependent and independent subgroups of the Pou5f1 transcriptional network. Interestingly, tissue-specific expression of Pou5f1 targets correlated with their regulation by Sox2, with Sox-dependent targets being mostly localized to ectoderm and neuroectoderm, whereas Sox-independent targets localized to mesendoderm of the developing zebrafish embryo. Further, SoxB1 independent Pou5f1 targets (for example foxD3) differ from SoxB1-dependent targets (e.g her3) in temporal dynamics of expression. Most Sox-independent direct Pou5f1 targets in WT reach maximal expression levels soon after midblastula transition (MBT) at 3–4 h postfertilization (hpf). In contrast, genes depending both on Sox2 and Pou5f1 tend to have a biphasic temporal expression curve or are activated with >2 h delay after MBT to reach maximum levels at 6–7 hpf only.

To better understand the impact of our findings on Pou5f1/SoxB1-dependent versus Pou5f1-only regulation on developmental mechanisms, we built a small dynamic network model that links the temporal control of target genes to regulatory principles exerted by Pou5f1 and SoxB1 proteins ( Figure 6A). The model is based on ordinary differential equations, and parameters were determined by a fit to the WT and MZ gene expression data. The optimized model highlights two qualitatively different temporal expression modes of Pou5f1 downstream targets: monophasic for targets depending only on Pou5f1 ( foxd3), and biphasic for Pou5f1- and SoxB1-dependent targets ( sox2 and her3; Figure 6B). To test whether the model is also able to correctly predict a different genetic condition, we simulated the M mutant, which is lacking maternal Pou5f1, but gradually rescued by the paternal pou5f1 contribution after MBT ( Figure 6B, blue, dashed curve). The model predicts an overall shift in the developmental program. Most importantly, the sox2 and her3 expression is rescued with a delay of about 2 h. The model predictions were checked experimentally by quantitative RT–PCR ( Figure 6B, blue dots). Most predictions are in good agreement with the experimental data, for example the delayed rescue of the sox2 and her3 temporal expression profile. With respect to the ‘POD' nr2f1, the model correctly predicts the efficient downregulation by zygotic targets of Pou5f1 ( Figure 6B).

We identified an evolutionary conserved core set of Pou5f1 targets, by comparing our gene list with the lists of mouse Pou5f1/Oct4 targets ( Loh et al, 2006; Sharov et al, 2008). The evolutionary conservation suggests equivalent Pou5f1 functions during the pregastrulation and gastrulation period of vertebrate embryogenesis. Therefore, we tested whether mouse Pou5f1/Oct4 was able to rescue MZ embryos. Injection of mRNA encoding mouse Pou5f1/Oct4 into MZ embryos ( Figure 8A) was able to restore normal zebrafish development to an extent comparable with zebrafish pou5f1/ pou2 mRNA ( Figure 8B and C). The significant overlap between zebrafish and mammalian Pou5f1 targets together with the ability of mouse Pou5f1/Oct4 to functionally replace the zebrafish Pou5f1/Pou2 ( Figure 8A–C), suggests that the mammalian network may have evolved from a basal situation similar to what is observed in teleosts. We propose models that emphasize the evolution of Pou5f1-dependent transcriptional networks during development of the zebrafish ( Figure 8D) and mammals ( Figure 8E). Our representation highlights the evolutionary ancient germlayer-specific subnetworks downstream of Pou5f1, which are presumably used for controlling the timing of differentiation during gastrulation in all vertebrates ( Figure 8D and E, black arrows). As the Pou5f1 downstream regulatory nodes revealed in our zebrafish model are likely conserved across vertebrates, we envision that their knowledge will contribute to the effort of directing differentiation of pluripotent stem cells to defined cell fates.

Abstract

The transcription factor POU5f1/OCT4 controls pluripotency in mammalian ES cells, but little is known about its functions in the early embryo. We used time-resolved transcriptome analysis of zebrafish pou5f1 MZ spg mutant embryos to identify genes regulated by Pou5f1. Comparison to mammalian systems defines evolutionary conserved Pou5f1 targets. Time-series data reveal many Pou5f1 targets with delayed or advanced onset of expression. We identify two Pou5f1-dependent mechanisms controlling developmental timing. First, several Pou5f1 targets are transcriptional repressors, mediating repression of differentiation genes in distinct embryonic compartments. We analyze her3 gene regulation as example for a repressor in the neural anlagen. Second, the dynamics of SoxB1 group gene expression and Pou5f1-dependent regulation of her3 and foxD3 uncovers differential requirements for SoxB1 activity to control temporal dynamics of activation, and spatial distribution of targets in the embryo. We establish a mathematical model of the early Pou5f1 and SoxB1 gene network to demonstrate regulatory characteristics important for developmental timing. The temporospatial structure of the zebrafish Pou5f1 target networks may explain aspects of the evolution of the mammalian stem cell networks.