January 1, 2011;
Geminin cooperates with Polycomb to restrain multi-lineage commitment in the early embryo.
Transient maintenance of a pluripotent embryonic cell population followed by the onset of multi-lineage commitment is a fundamental aspect of development. However, molecular regulation of this transition is not well characterized in vivo. Here, we demonstrate that the nuclear protein Geminin
is required to restrain commitment and spatially restrict mesoderm
and non-neural ectoderm
to their proper locations in the Xenopus embryo
. We used microarray analyses to demonstrate that Geminin
overexpression represses many genes associated with cell commitment and differentiation, while elevating expression levels of genes that maintain pluripotent early and immature neurectodermal cell states. We characterized the relationship of Geminin
to cell signaling and found that Geminin
broadly represses Activin-, FGF- and BMP-mediated cell commitment. Conversely, Geminin
knockdown enhances commitment responses to growth factor signaling and causes ectopic mesodermal, endodermal and epidermal fate commitment in the embryo
. We also characterized the functional relationship of Geminin
with transcription factors that had similar activities and found that Geminin
represses commitment independent of Oct 4 ortholog (Oct25
/60) activities, but depends upon intact Polycomb repressor function. Consistent with this, chromatin immunoprecipitation assays directed at mesodermal genes demonstrate that Geminin
promotes Polycomb binding and Polycomb-mediated repressive histone modifications, while inhibiting modifications associated with gene activation. This work defines Geminin
as an essential regulator of the embryonic transition from pluripotency through early multi-lineage commitment, and demonstrates that functional cooperativity between Geminin
and Polycomb contributes to this process.
[+] show captions
Fig. S1. Cell proliferation and apoptosis in Geminin overexpressing and knockdown embryos. Embryos were injected with mRNA encoding a Geminin variant deleted for the Cdt1-binding central coiled-coil (Gemininδcoil) (Seo et al., 2005) or with a Geminin antisense morpholino oligo (MO) (Seo et al., 2005) at doses used for subsequent experiments. We have previously found that both full-length Geminin and two Cdt1 non-binding variants (the N-terminal half or Gemininδcoil) had similar effects on cell fate: overexpression of full-length Geminin at high levels in embryos was cytotoxic, but injecting lower doses (20-30 pg full-length Geminin RNA) or injecting Cdt1-non-binding Geminin variants expanded the neural plate and suppressed epidermis (Kroll et al., 1998; Seo et al., 2005). Likewise, Geminin MO-mediated knockdown could induce epidermal differentiation at the expense of neural tissue formation (Seo et al., 2005). (A) To assess whether Gemininδcoil or MO affected cell proliferation, we performed immunostaining for phosphorylated Histone H3 (PH3), which marks mitotic cells in mid-gastrula embryos (stage 11-11.5; A). Immunostaining was performed as described previously (Saka and Smith, 2001), using a 1:1000 dilution of the primary antibody (Millipore/06-570). (B) In each embryo, the number of phosphorylated Histone H3-positive cells located in the X-Gal-labeled injected region was compared with an uninjected region of equivalent size (e.g. A, top right-hand panel). Paired t-tests (brackets) showed no significant difference in PH3-positive cell numbers between injected and uninjected tissue. (C) TUNEL assays were performed as described previously (Hensey and Gautier, 1998) under the same conditions to detect apoptotic cells. (D) TUNEL-positive cells in injected embryos were counted. Most embryos had few or no TUNEL-positive cells, as expected (Hensey and Gautier, 1998); to illustrate the assay, an embryo with several TUNEL-positive cells is shown in C, bottom right-hand panel.
Fig. S2. Mesodermal and endodermal gene expression changes in embryos resulting from Geminin overexpression and morpholino knockdown. (A) 120 pg Gemininδcoil mRNA or 1 ng MO was co-injected with 20 pg β-galactosidase lineage tracer mRNA into one blastomere at the eight-cell stage and the embryos were raised to stage 11-11.5. After lineage labeling and in situ hybridization as before, embryos were selected for scoring only if the lineage-labeled region overlapped the marker gene. For each gene analyzed, numbers represent additive data from three to six experiments. (B) Dose dependence of Geminin knockdown effects. The three doses of Geminin MO indicated were injected into embryos and effects on Xpo, Sox17, Mix.2 and Xbra expression were quantitated as above; the fraction of embryos with the change in expression indicated is shown in each chart, with number of embryos scored in parentheses in the legend. (C) The effects of Geminin MO knockdown (1 ng injection) on Brachyury expression was scored for embryos at stage 10-11. Examples, with lineage-labeled territory indicated (pink stain, arrowhead) are shown at right. Geminin knockdown has no effect in many embryos at early gastrulation, and induces some ectopic Brachyury expression. By contrast, Geminin knockdown strongly reduces Brachyury expression in nearly all embryos by mid-late gastrulation.
Fig. S3. Spatial and temporal aspects of gene regulation by Geminin. (A) We compared the activity of full-length (FL) Geminin to the Gemininδcoil form with respect to their ability to suppress mesodermal expression of Brachyury and Xpo and endodermal expression of Sox17. These experiments used 120 pg Gemininδcoil or 12 pg of full-length Geminin mRNA (injected into one blastomere of an eight-cell stage embryos), to avoid cytotoxicity associated with higher level overexpression of the full-length form. We also assessed the temporal aspect of the response by comparison of these mRNA-injected embryos (where Geminin is overexpressed throughout early development) with embryos injected with expression plasmids for the Geminin FL and δcoil variants, where expression is driven from the CMV promoter and restricted to the late blastula-gastrula stages. Embryos were co-injected with 120 pg of each expression plasmid plus 30 pg β-galactosidase lineage tracer mRNA, were raised to stage 11.5 and were Red-gal stained to detect the lineage tracer (oriented rightward in images) and then assayed by in situ hybridization for the markers shown. The β-galactosidase mRNA lineage tracer defines only general areas, not individual cells, when plasmid DNA is injected. As is seen for Gemininδcoil mRNA injection (Fig. 1), either FL or δcoil forms of Geminin expressed as either capped mRNAs or from expression plasmids can effectively suppress mesodermal and endodermal expression. (B) Embryos were co-injected with 120 pg Geminin mRNA or 1 ng MO and with 30 pg β-galactosidase mRNA lineage tracer as before, stained for X-gal (blue) or Red-gal (pink in Fzd7), and in situ hybridized as indicated (purple). Magnified regions of Geminin overexpressing or MO knockdown (Xpo) embryos are shown, with the embryo they derive from in the inset (or at right for Oct25/Fzd7). The repression by Geminin of Epiker, Vent2, Xpo, Brachyury and Endodermin is restricted to the Geminin-injected cells. For morpholino knockdown, ectopic Xpo expression is also confined to the MO-injected territory, although the β-galactosidase mRNA lineage tracer would not be expected to reflect MO distribution at a single cell level. In contrast to the results seen for Geminin repression of gene expression, ectopic/expanded Oct25 and Fzd7 expression in neurula stage embryos occurs broadly in both unlabeled cells and in cells labeled with the X-gal (Oct25) or Red-gal (Fzd7) lineage tracer.
Fig. S4. Geminin expression marks pluripotent ectoderm and immature neurectoderm, while being excluded from mesendoderm and ventral ectodermal regions by the onset of gastrulation. In situ hybridization for Geminin (purple or pink) and/or Brachyury (greenish-blue) from blastula (stage 9) through gastrula (stage 10-12) stages, with views as indicated (lower left). Maternal Geminin is present throughout the pluripotent ectoderm and overlaps transiently with Brachyury at blastula stages overlap between pink and greenish-blue signals results in purple signal (upper right panel). At the onset of gastrulation, the expression of Geminin becomes enriched in prospective neural ectoderm and is non-overlapping with mesodermal territories expressing Brachyury.
Fig. S7. Oct25/60 and Sox2 suppression of mesoderm and endoderm is not blocked by Geminin knockdown. Embryos were injected with the reagents indicated, plus β-galactosidase mRNA, were lineage labeled (Red-gal, pink), and were in situ hybridized to detect Xpo, Brachyury or Sox17 (see Materials and methods). 120 pg of mRNAs (Oct25, Oct60, Sox2, Ezh2), 8 ng of Ezh2MO and Suz12MO, and 1 ng Geminin MO were used. Arrowheads mark injected regions. On the right, embryos were only scored if the lineage label was visible and overlapped expression of the marker; number of embryos analyzed is in parentheses.
Fig. S10. Geminin repression of mesodermal gene expression is not sensitive to inhibition of histone deacetylase activity. Embryos were injected with Gemininδcoil mRNA (300 pg; one cell out of two cells) and half of the embryos were treated from the two-cell stage (immediately after injection) with 100 nM TSA, followed by in situ hybridization at stage 11. (A,B) Geminin repressed Brachyury expression in untreated and TSA (100 nM)-treated embryos. (C) Fraction of embryos showing the phenotypes in A and B is indicated (additive quantitation from two experiments). (D) Gemininδcoil-injected embryos were used to prepare animal cap explants and were treated with Activin as before, plus and minus TSA co-treatment. Geminin repressed Activin-induced Brachyury expression with similar efficiency in the presence and absence of TSA treatment.
Fig. 1. Geminin regulates epidermal, mesodermal and endodermal gene expression. Geminin RNA or morpholino/MO-injected regions are oriented rightwards, indicated by yellow arrowheads, and lineage is labeled by β-galactosidase RNA co-injection and Red-gal (Mix.2, Brachyury/Xbra, and Sox17 at stage 11) or X-gal staining. Embryos are stage 10.5-11, except for Epidermal keratin/Epiker and Sox17 (right), which are stage 12-13. Views are indicated: (veg, vegetal; post, posterior).
Fig. 3. Geminin suppresses mesoderm and non-neural ectoderm formation. Top: downregulated genes from microarrays were assayed by qRT-PCR for expression changes upon Geminin overexpression or knockdown in embryos (left) or stage 12 ectodermal explants (center/right). Right panel compares effects of plasmid-driven (125 pg) versus mRNA-based overexpression of full-length-FL/30 pg or δcoil/300 pg Geminin on gene expression. Relative expression normalized to EF1α is shown, control (uninjected embryo/explant) values=1.0. A representative result from at least three independent experiments is shown. Bottom: confirmation of Geminin-downregulated genes by in situ hybridization. Injected region is marked by Red-gal (Grhl1/Zic1/Zic3) or X-gal staining and yellow arrowheads. For Foxi1/Grhl1/Zic1/Zic3, stage 13/neural plate view is shown with dorsal midlines indicated (lines).
Fig. 4. Geminin promotes expression of genes that mark pluripotent ectoderm and neurectoderm. Left: in situ hybridization of Geminin-overexpressing or MO knockdown embryos with injected area marked by Red-gal (Zic2/Fzd7/Sox2-MO/Sox3-Geminin) or X-gal (yellow arrowhead). Right: genes were assayed by qRT-PCR for changes in expression in stage 12 ectoderm and comparison of Geminin FL versus δcoil after Fig. 3.
Fig. 6. Geminin suppression of mesoderm and endoderm depends on Polycomb but is independent of Oct25/60 and Sox2 activities. Geminin RNA (120 pg) with or without 8 ng of MOs (16 ng Oct60MO) + 30 pg β-galactosidase RNA was injected into one cell of eight-cell embryos, with in situ hybridization at stage 11.5. Red-gal staining and white or red arrowheads mark injected regions. Below are additive results for four experiments shown as fractions of embryos with indicated effects (embryo numbers analyzed are in parentheses).