Author contributions: Y.L.: collection and/or assembly of data and data analysis and interpretation; W.Y.: collection and/or assembly of data; A.J.C.: provision of study material or patients; R.J.S.: data analysis and interpretation; Y.L.: collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
first published online in STEM CELLS EXPRESS February 4, 2013.
Oct4 is the gatekeeper of stem cell pluripotency, but recent evidences also support Oct4 as a key regulator of germ layer formation and lineage commitment. How Oct4 contributes to lineage commitment is not well understood. We identified a Tcf/Lef-Oct4 composite site in the promoter of the cardiac mesoderm gene Mesp1, with a nucleotide sequence identical to the previously established Sox2-Oct4 composite site. This Tcf/Lef-Oct4 composite site mediated synergistic activation of the Mesp1 promoter by Oct4 and canonical Wnt signaling. Transcription ternary complexes were formed with Oct4 and Wnt terminal components, Lef1. Point mutations on the Tcf/Lef-Oct4 composite site impaired Oct4 and Lef1 binding and Mesp1-β-gal transgene reporter expression during mouse embryogenesis. In ZHBTc4 murine embryonic stem cells, the loss of Oct4 during differentiation impaired Mesp1 expression and the development of the cardiac program. This Tcf/Lef-Oct4 composite site appears to be a unique nodal point regulatory element that may drive pluripotency via Sox2-Oct4 and switch on lineage-related genes through Oct4's recruitment of Tcf/Lef factors. STEM Cells2013;31:1213–1217
POU family homeodomain transcription factor Oct4 is the gatekeeper of pluripotency . Together with Sox2 and Nanog, Oct4 occupies large ensembles of target genes [2–4]. Cooperation between Oct4 and Sox2 is especially intricate, involving a composite binding site composed of both Oct and Sox binding sequences separated by 0–3 nucleotides. Oct4 and Sox2 synergistically activate the transcription of downstream genes bearing this composite element, even autoregulating themselves [2, 5-11].
Interestingly, Oct4 is also involved in early cell fate determination. Choice for germ layers is directly related to Oct4 levels: 50%–150% of endogenous Oct4 maintains pluripotency, a drop in Oct4 level leads to the differentiation toward trophectoderm, and a twofold Oct4 level leads to mesoderm and endoderm . Particularly, ectopic Oct4 triggers the expression of early mesoderm and cardiac genes [13–15]. Compared to its well-characterized role in regulating pluripotency, how Oct4 contributes to lineage commitment and cell differentiation is unclear.
Mesp1 sits on the top of the cardiac regulatory hierarchy. Homologous disruption of the Mesp1 gene resulted in aberrant cardiac morphogenesis . Ectopic Mesp1 in embryonic stem cells (ESCs) leads to commitment to the cardiac program [17–19]. Here, we found that Oct4 integrates the canonical Wnt signaling to regulate Mesp1 and the underlying cardiac program, through a novel Tcf/Lef-Oct4 composite element.
This Tcf/Lef-Oct4 composite site (−31 to −17 relative to transcription starting site) was among a number of transcription factor binding sites predicted with rVista2  on a 6012-bp upstream region of the Mesp1 gene (Fig. 1A). Initially, we asked if these transcription factors (TFs) can activate the transcription of Mesp1. A constitutively active β-catenin (CA-β-Cat), the cofactor for Tcf/Lef, activated a Mesp1-Luc reporter in a dose-dependent manner (Fig. 1B). The putative Tcf/Lef site is a perfect match to the Tcf/Lef consensus sequence CTTTG[AT][AT] . Interestingly, it sits adjacent to an “octamer” putative Oct4 binding site. The sequence also fits into the consensus sequence of the previously identified Sox2-Oct4 composite site, residing on the promoters of Sox2, Oct4, Nanog, and a few other genes . In reporter assays, Oct4 only marginally transactivated the reporter but had a potent synergistic effect with CA-β-Cat (Fig. 1C). However, mutation of the octamer within the composite site blocked coactivation of the Mesp1-Luc reporter, demonstrating the requirement for Oct4 occupancy (Supporting Information Fig. S1). Hereby we name this composite site as Tcf/Lef-Oct4 site.
During embryogenesis, Oct4 level drops to allow differentiation. Meantime, a burst of canonical Wnt signaling at gastrulation induces a number of lineage-related genes [22–24]. This expression pattern was recapitulated in embryoid body cultures, with the induction of Mesp1 overlapping the peaks of canonical Wnt components Wnt3 and β-catenin (Fig. 1D). This prompted us to ask if the Tcf/Lef-Oct4 site recruits Oct4 and canonical Wnt components to the Mesp1 promoter. In Chromatin Immunoprecipitation (ChIP)-PCR assays, Oct4 antibody enriched comparable amounts of Mesp1 immediate promoter DNA, in undifferentiated and differentiated cells (Fig. 1E). This suggests that despite a drop in Oct4 level in differentiated cells, it continues to occupy the immediate promoter region. We tested terminal Wnt component Lef1, as it patterns mesoderm formation and is therefore the likely cofactor . As the result, Lef1 antibody enriched more Mesp1 immediate promoter DNA in differentiated cells than undifferentiated cells, consistent with the notion that canonical Wnt signaling drives Mesp1 expression at the onset of differentiation (Fig. 1E). In contrast, Sox2 was not recruited to the immediate promoter in undifferentiated or differentiated cells.
We next asked whether Oct4 and Lef1 physically bind to the Tcf/Lef-Oct4 site. Oct4 and Lef1 proteins were prepared by coupled in vitro transcription/translation. In gel shift assays, the reticulocyte extracts alone produced a distinct band, which is thus labeled “non-specific” (NS, Supporting Information Fig. S2A, lane 2). Coincubation of labeled oligos and Lef1 protein produced a distinctive band, which was greatly reduced by “cold” wild-type or MtO (mutant affecting the octamer site) oligos. In contrast, “cold” MtT (mutant affecting the Tcf/Lef site) or MtTO (mutant affecting both Tcf/Lef and octamer site) competed much less (Supporting Information Fig. S2A, lanes 3–7). Moreover, Lef1 antibody abolished the oligo/Lef1 band, while SRF antibody had little effect (Supporting Information Fig. S2A, lanes 8 and 9). Similarly, the labeled oligos were shifted by Oct4 protein to produce another distinctive band, which were reduced or abolished by increasing amounts of Oct4 antibody but not SRF antibody (Supporting Information Fig. S2B, lanes 2–6). This band was reduced by wild-type or MtT oligos but not by MtTO or MtO oligos (Supporting Information Fig. S2B, lanes 7–10). These results support that Oct4 and Lef1 can individually bind to the Tcf/Lef-Oct4 site. We did not detect the binding of Sox2 to the Tcf/Lef-Oct4 site under the same condition, despite evident Sox2 expression (not shown).
In the presence of both Oct4 and Lef1, the labeled oligos were further shifted to produce a ternary complex. In Figure 2A, a third specific band appeared when both Oct4 and Lef1 were present. “Cold” wild-type oligos greatly reduced all three specific bands. “Cold” MtO oligos greatly reduced the ternary band, the Lef1 band, but not the Oct4 band. “Cold” MtT oligos greatly reduced the ternary band, the Oct4 band, but not the Lef1 band. “Cold” MtTO oligos did not preferably affect the three specific bands. Moreover, Lef1 and Oct4 antibody abolished the ternary band, and the Lef1 and Oct4 bands, respectively. These results support that Oct4 and Lef1 co-occupy the composite element.
We next studied the binding dynamics between Oct4, Lef1, and the Tcf/Lef-Oct4 site. In this experiment, either Oct4 or Lef1 protein was kept constant while the other was varied. Increasing amounts of Lef1 led to increased formation of the binary Lef1 band, and the ternary complex, while the binary Oct4 band was reduced. And vice versa for increasing amounts of Oct4 (Fig. 2B). This result suggests that binding of Lef1 or Oct4 recruits the other to the Tcf/Lef-Oct4 site.
Mesp1 and other nascent mesoderm genes were dependent on canonical Wnt/β-catenin signaling [22, 23, 26, 27], which is confirmed here in β-Cat−/− cells where expression of cardiac transcription factors and structural genes were impaired (Supporting Information Fig. S3). But, was Oct4 required for Mesp1 induction during ES cell differentiation? We used ZHBTc4, an ESC line in which both alleles of Oct4 were knocked out, and engineered so that Oct4 expression was maintained by a tet-off Oct4 transgene  (Fig. 3A). ZHBTc4 cells were cultured in hanging droplets to induce differentiation and supplemented with Dox over serial time intervals. Oct4 and Mesp1 gene expression were analyzed on day 4, which coincides with the initiation of differentiation. Supplemental Dox repressed the expression of Oct4, even when given for a short interval (day 3.5–4). The loss of Oct4 expression led to impaired Mesp1 induction (Fig. 3B) and blocked the cardiac program, as shown by the reduction of spontaneous beating (Fig. 3C) and α-Actinin staining, an early marker of cardiogenesis (Fig. 3D). Thus, Oct4 is essential for Mesp1 induction and the development of the cardiac program.
How the Tcf/Lef-Oct4 site affects Mesp1 expression during embryogenesis was examined by a transient transgenic assay. The 6012-bp promoter of Mesp1 was fused to a LacZ reporter gene, and this construct was injected into the pronuclei of fertilized eggs. The resulted E7.5 embryos were analyzed for β-galactosidase expression. The wild-type promoter led to β-galactosidase expression in the primitive streak, where Mesp1 mRNA were previously shown to be expressed at peak levels . Mutation on both Tcf/Lef and octamer side of the composite site resulted in complete loss of β-galactosidase staining. Rarely did MtT (1/9) and MtO (1/7) lead to β-galactosidase expression, in contrast to the high rate (10/29) expression driven by the wild-type promoter (Fig. 3E). These results suggest that the Tcf/Lef-Oct4 site is essential for the expression of Mesp1 during embryogenesis.
In summary, our data support that Oct4 plays an active role in transactivating a cardiac lineage gene, by integrating a crucial gastrulation signal in Wnt. This explains the puzzling role of Oct4 in triggering the cardiac program and provides a novel mechanism of Oct4 in regulating the initiation of differentiation.
Supporting Information Table 1 summarizes genes reported to be regulated by an Oct4 composite site. All previously described genes, namely, Fgf4, Utf1, Sox2, Fbx15, Nanog, and Pou5f1 are subjected to the regulation of a Sox2-Oct4 tandem and play important roles in pluripotency [5–11]. Recently, the relative levels within the Sox2-Oct4 tandem were found to regulate cell fates, building a connection between pluripotency and differentiation. Here, Mesp1 is the first identified target of a Tcf/Lef-Oct4 tandem . Since about 70% of all Oct4 binding site share such a composite regulatory element , we look forward to find other genes that are subjected to the Oct4/Wnt regulation depicted in this study. Since Tcf/Lef and Sox are most closely related within the high mobility group (HMG)-containing protein family , we propose that a HMG-Oct4 composite site is a unique nodal point regulatory element that may drive pluripotency via Sox2-Oct4 and switch on lineage-related genes through recruitment of Tcf/Lef factors (Fig. 3F, a working model).
A Oct4/Wnt signaling model may be an error-proofing mechanism governing the switch from pluripotency to differentiation. It ensures that the activation of lineage-related genes such as Mesp1 is only possible in progenitor cells, where a burst of canonical Wnt signaling meets with residual Oct4 expression. In other cell types, for example, terminally differentiated cells, the ability to activate key lineage genes like Mesp1 is compromised due to the loss of Oct4 expression.
Oct4 and terminal Wnt component Tcf/Lef bind to the promoter of Mesp1 through a Tcf/Lef-Oct4 composite element. The cooperation between Oct4 and canonical Wnt signaling thus play an essential role in Mesp1 expression and the cardiac program.
We thank Randall Moon, Hitoshi Niwa, and Janet Rossant for reagents; David Stewart and Shuxing Zhang for helpful discussions; Paul Swinton for technical assistance. This research was supported by research funds from the Texas Heart Institute (R.J.S.) and the University of Houston (R.J.S.) and a grant from American Heart Association (Y.L.). Yongqing Li is currently affiliated with The Center for Heart Development, Lab of MOE for Development Biology and Protein Chemistry, College of Life Sciences, Hunan Normal University, Changsha, People's Republic of China.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.