POU-V class proteins like Oct4 are crucial for keeping cells in an undifferentiated state. An Oct4 homologue in Xenopus laevis, Oct25, peaks in expression during early gastrulation, when many cells are still uncommitted. Nevertheless, extensive morphogenesis is taking place in all germ layers at that time. Phenotypical analysis of embryos with Oct25 overexpression revealed morphogenesis defects, beginning during early gastrulation and resulting in spina-bifida-like axial defects. Analysis of marker genes and different morphogenesis assays show inhibitory effects on convergence and extension and on mesoderm internalization. On a cellular level, cell–cell adhesion is reduced. On a molecular level, Oct25 overexpression activates expression of PAPC, a functional inhibitor of the cell adhesion molecule EP/C-cadherin. Intriguingly, Oct25 effects on cell–cell adhesion can be restored by overexpression of EP/C-cadherin or by inhibition of the PAPC function. Thus, Oct25 affects morphogenesis via activation of PAPC expression and subsequent functional inhibition of EP/C-cadherin.
Stem cells are defined as undifferentiated cells that are enabled to do both, self-renewal and differentiation. Embryonic stem (ES) cells are pluripotent, i.e. they can differentiate into all cell types of the different germ layers. This pluripotency is sustained by the function of a series of stem cell factors. In the mouse model transcription factors like Oct4 and Sox2 have been found among others to be crucial for keeping cells in an undifferentiated state (Masui et al. 2007; Niwa 2007; Niwa et al. 2009). The transcription factor Oct4 is a member of the POU-V class proteins. This central molecule in the pluripotency maintenance network has three functional homologous in Xenopus laevis embryos, Oct60, Oct25 and Oct91 (Hinkley et al. 1992; Whitfield et al. 1993, 1995; Morrison & Brickman 2006). During early development, expression of Oct60 is found maternally. Oct91 and Oct25 are zygotically expressed, with Oct25 having an earlier peak of expression during gastrulation (Hinkley et al. 1992). At these stages, the expression is predominantly found in the animal region of the embryo, which contains prospective ectoderm.
The early expression maximum of Oct25 is correlated to the massive appearance of different morphogenetic movements in the amphibian embryo during early gastrulation (Vogt 1929; Davidson 2008). These morphogenetic movements result in formation of the germ layers and their arrangement in the shape of the organism. They include epiboly of the ectoderm (Keller 1978, 1980), invagination of the (suprablastoporal) endoderm (Rhumbler 1902; Lewis 1947; Keller 1981), vegetal rotation of the vegetal cell mass (Winklbauer & Schürfeld 1999; Winklbauer et al. 2001; Keller et al. 2003), involution (Shih & Keller 1994; Winklbauer & Keller 1996) and convergence and extension (Keller et al. 1985a,b) of the mesoderm. On a cellular level, involved mechanisms are differential cell–cell-adhesion, cell-substrate adhesion and cell polarity (for review see (Winklbauer 2009). These cellular properties are established during initial specification of cells, therefore often preceding differentiation events.
Before cells become committed to a certain fate (e.g. mesoderm, paraxial mesoderm, somites, muscle), they are already characterized by certain cellular behaviors, which are based on molecular properties that are not necessarily related to their fate. The cytoskeleton is arranged in structures enabling cell shape changes (e.g. epithelial versus deep cells of the animal hemisphere, unpolar versus polar cells of prospective mesoderm), adhesiveness (e.g. epithelial versus deep cells of prospective ectoderm, involuted “mesenchymal” mesoderm versus preinvoluted “epithelial” mesoderm). Since these cellular properties emerge so early during development and are often transient, they must be regulated by maternal or immediate early zygotic signals.
The Xenopus Oct4 homologues are regulators of such early pathways, namely the mesoderm inducing VegT-, activin/nodal-, and FGF-signals, the dorsalizing canonical Wnt-pathway, and the anti-dorsalizing BMP-signaling (Cao et al. 2004, 2006, 2007, 2008; Takebayashi-Suzuki et al. 2007). Inhibitory effects have been found for most of these pathways (Cao et al. 2006, 2007, 2008; Takebayashi-Suzuki et al. 2007). In general, the effects in gain of function and in loss of function experiments indicate that early inductive events and especially differentiation is inhibited. Resulting phenotypes for Oct25 overexpression show a broad diversity depending on the localization and dosage. This includes gastrulation defects, axis formation defects and loss of (differentiated) structures (e.g. eyes, brain, somites) (Cao et al. 2004, 2006, 2007).
Here we performed experiments to analyze how morphogenesis is affected by gain of function of Oct25. Overexpression was generated in the animal hemisphere, where the Oct factors become typically activated and weaker effects on germ layer formation are induced. We then collected phenotypes, analyzed expression patterns of germ layer specific markers, and performed a set of morphogenesis assays. We found striking phenotypes developing from a defect during gastrulation. We see relatively weak effects on germ layer specific markers, but found ectopic activation of paraxial protocadherin (PAPC). PAPC is a known effector of cell adhesion and morphogenesis (Kim et al. 1998; Medina et al. 2004; Chen & Gumbiner 2006). This can serve as an explanation for the observed phenotypes. An Oct25 function on regulation of morphogenesis is further supported by the intriguing observation that some of the Oct25 overexpression effects are recovered by co-expressing the classical cell adhesion molecule EP/C-cadherin or by a functional inhibition of the ectopic PAPC using a dominant inhibitory PAPC construct.
Materials and methods
Embryos and explants
Embryos were staged according to the normal table (Nieuwkoop & Faber 1956). Embryos were kept as required at temperatures from 14 to 21°C. In vitro fertilization, microinjection, and culture of embryos and explants were carried out as previously described (Winklbauer 1990; Wacker et al. 2000). In general, explantations were performed using a pair of watchmaker forceps (Dumont No. 5) and a pair of eyebrow knives.
For mesoderm induction on animal cap (AC) cells, we used the injection method (Cooke & Smith 1989). Two hundred nanoliters of Activin A (200 ng/mL in Gurdon's injection buffer containing 0.1% bovine serum albumin [BSA]) was injected into the blastocoel of stage 8 to 9 embryos. ACs were explanted 2.5 h later at stage 10, cultivated in Modified Barth's solution (MBS) for 2 h and then transferred to 0.1x MBS to analyze elongation another 8 h later (corresponding stage 17).
Dorsal marginal zone explants
Dorsal marginal zone (DMZ) explants were explanted at stage 10–10.25. The explant contained an angle of approximately 90° of the blastopore lip and a corresponding piece of the AC up to the animal pole. Explants were placed on BSA-coated Petri dishes and covered with a slice of BSA-coated cover slip to keep the explant flat. The cover slip was attached with silicon grease. After cultivation at 14°C for 24 h, explants were analyzed by measuring the longest (l) and the shortest diameter (w). A value of l/w ≤ 1.5 is defined as “not elongated”, 1.5 < l/w < 2.5 is defined as “weak elongation”, l/w ≥ 2.5 is defined as “strong elongation”.
Animal caps were explanted at stage 11–11.5. Explants were placed in a BSA-coated Petri dish with Ca2+- and Mg2+-free dissociation buffer. Photos were made in 10 min intervals to analyze the rate of dissociation.
Reaggregation assay was modified from Brieher & Gumbiner (1994) Animal caps were explanted at stage 11. Explants were placed in BSA-coated Petri dishes with dissociation buffer. After 10 min of dissociation the epithelial layer was removed with eyebrow knives. The deep cells were further incubated and from time to time gently moved to improve dissociation. Dissociated cells were transferred to Petri dishes coated with 1% Agarose, containing MBS. To enable reaggregation, the dishes were rotated on a shaking table (45 rpm). Cells were photographed immediately after transfer, after 2 h, and after 24 h to analyze the rate of reaggregation.
In vitro transcription and microinjection
For preparation of mRNA for injection, the following plasmids were used: pCS2 + Oct-25 (Cao et al. 2004), pCS2 + MTd2EosFP (Wacker et al. 2007), pCS2Epfl(1-2699) (EP/C-Cadherin, [Wacker et al. 2000]), and pCS2 + SXPAP (secreted Xenopus PAPC, also called dnXPAPC, Kim et al. 1998). All plasmids were linearized with NotI and mRNA was synthesized with SP6 RNA polymerase. Capped mRNAs were transcribed in vitro using the mMESSAGEmMACHINE kit (Ambion) and purified on RNeasy columns (Qiagen). Stocks were stored at −80°C, aliquots of the concentrations used for injections were stored at −20°C. Ten nanoliters of the respective RNAs were injected into the animal pole region of each blastomere at stage 2, or, into the marginal zone of the two dorsal blastomeres at stage 3, respectively. The concentrations of injected mRNAs were as follows: Oct25 30 pg/nL, EP/C-cadherin 20 pg/nL, EosFP 25 pg/nL. For loss of function experiments, 10 nL of the Oct25 morpholino (Oct25 MO, concentration 200 pg/nL, Cao et al. 2006) was injected in both blastomeres at stage 2.
Whole mount in situ hybridization and histology
For the whole mount in situ hybridization, antisense digoxigenin (DIG)-labeled probes were synthesized from the following constructs: pCS2 + Xbra (EcoRI/T7, containing the Xbra sequence [Smith et al. 1991]), pCS2 + PAPC (BglII/T7, [Kim et al. 1998]), (kind gift from A.Schambony), pCS2 + Xsox17α (ClaI/T7, [Hudson et al. 1997]), pCMV-Sport6 + Xsox3 (EcoRI/T7, [Jansen et al. 2007]), pBSK(+)+Sox2 (XbaI/T7, [Jansen et al. 2007]), pCS2 + AXPC (EcoRI/T7, [Kuroda et al. 2002]), pCS2 + Chd (HindIII/T7, [Sasai et al. 1994]), pBS+MyoD (SalI/T7, [Hopwood et al. 1989]), pBS+epidermal keratin (EcoRI/T7, containing the XK81A1 sequence [Jonas et al. 1989]).
The whole mount in situ hybridization was performed as previously described (Harland 1991; Wacker et al. 2004), except that the RNase step was omitted. BM-purple (Boehringer) served as a substrate for the alkaline phosphatase.
In some cases, embryos were bleached with hydrogen peroxide after the in situ hybridization.
For sectioning, stained embryos were embedded in gelatine blocks. They were dissected at 50 μm intervals (Vibratome 1500 Section System). The sections were collected on a glass slide with glycerol and covered with a cover slip. They were photographed with the Keyence BZ8000K microscope (2.5× lens).
Reverse transcriptase–polymerase chain reaction
Animal caps were cut at stage 11–11.5 to allow discrimination between ectopic and internal expression of PAPC. Total RNA was extracted from stage 11–11.5 embryos or explants by Trizol (Invitrogen) and phenol chloroform extraction. The extract was purified with the RNeasy columns kit (Qiagen). The first strand cDNA synthesis (Fermentas) was carried out according to the producer's instructions. The experiment was repeated at least twice for every presented marker. The following primers were used: EP/C-cadherin (f: 5′-gga gtc aga gta tgt caa-3′ and r: 5′-tcc ctc aca gat gca cac-3′), H4 (f: 5′-cat cac caa gcc cgc cat cc-3′ and r: 5′-ggc gct tga gag cgt aca cc-3′), PAPC (f: 5′-cac ctg gac aat tgt gc-3′ and r: 5′-atg gtt cgg tat gtg cag g-3′), Xbra (f: 5′-cac agt tca tag cag tga ccg-3′ and r: 5′-ttc tgt gag tgt acg gac tgg-3′), Sox2 (f: 5′-acc gct atg atg tca gtg-3′ and r: 5′-ctg agg cac tct gat agt-3′), XK81A1 (f: 5′-cac cag aac aca gag tac-3′ and r: 5′-caa cct tcc cat caa cca-3′), Xsox17α (f: 5′-gga cga gtg cca gat gat g-3′ and r: 5′-ctg gca agt aca tct gtc c-3′).
Photoconversion and imaging
Embryos were injected in the DMZ region with 500 pg of mRNA encoding the photoconvertible protein EosFP (Wacker et al. 2007) at stage 2 and cultivated in the dark. At stage 10.25, photoconversion of cells from the dorsal blastopore lip was performed on an area of approximately 250 μm in diameter. We used an inverse fluorescence microscope (DM IRB, Leica) with a 100 W Hg light source (HBO 103W/2, Osram) for irradiation with 405–425 nm light (BP430/50 excitation filter, 425 DCLP beam splitter). Imaging began within 30 min after conversion with a digital camera (C4742, Hamamatsu), using the Software “Improvision Open-Lab”. Exposure times were from 150 ms to 600 ms. The filter sets were HQ480/40 (ex), Q505LP (Beamsplitter), HQ527/30 (em) for green fluorescence and HQ545/30 (ex), Q565LP (Beamsplitter), 610/75 (em) for red fluorescence. Identical image sections of green fluorescence photographs and of red fluorescence photographs were taken and overlaid with Adobe Photoshop.
For time-lapse movies, frames were grabbed at 2 min intervals for 5 to 6 h (corresponding to stages 10.25 to 14). Three independent time-lapse series were recorded for both, uninjected control and Oct25 mRNA injected embryos. The filter set for red fluorescence was used. The movie files were generated with the ImageJ software (http://rsb.info.nih.gov/ij/). In parallel to each time-lapse series, several embryos were photographed at the beginning and the end of gastrulation showing identical effects (not shown).
Embryos for the phenotype analysis, embryos from whole mount in situ hybridizations, explants and cells were photographed with the Olympus DP50-CU camera on an Olympus SZX12 Stereomicroscope and a LED dark field ring light source (DFRL-60, Optometron, Munich, Germany).
Ectopic expression of Oct25 results in morphogenesis phenotypes
After overexpression of Oct25, differential effects depending on the position of injection and the mRNA dose have been observed (Cao et al. 2004, 2006, 2007). To boost the “natural” pattern (graded from animal to vegetal, (Cao et al. 2007) we decided to inject Oct25 mRNA into the animal pole at the two-cell stage. Embryos developed regularly until the beginning of gastrulation (Fig. 1A, E, stage 9). The first effects were observed, when formation of the blastopore began (Fig. 1B, F stage 10). In injected embryos, the formation of the blastopore was delayed compared to uninjected controls. A failure of blastopore closure followed at late gastrula and early neurula stages (Fig. 1C, G, stage 11, Fig. 1D, H, stage 14). During neurulation (Fig. 1I, K, M, stage 19), the neural folds formed, but the closure of the neural tube was disturbed either in the head region resulting in a “condensed” head at tadpole stages, most obviously with reduction of the rhombencephalon (Fig. 1J, L, stage 28), or neural fold closure failed in the trunk region, showing a yolk plug at neurula stages (Fig. 1M) and a spina bifida like structure along the complete axis at tadpole stages (Fig. 1J, N, stage 28). In addition, the anterior-posterior axis was shortened in both phenotypes (Fig. 1J, L, N). Numbers for the phenotype analysis are listed in Table S1.
Overall, after an initial phase of normal developmental, defects appear during periods of extensive morphogenetic movements, namely incomplete blastopore closure during gastrulation and failure of neural tube formation, resulting in axial and head defects during tadpole stages.
To analyze, if these defects are based on a disturbed germ layer formation, we performed in situ hybridizations with a set of germ layer specific markers, i.e. mesodermal brachyury (Xbra), Snail (Xsna) and PAPC, the endodermal Xsox17α, and the ectodermal Xsox3 at midgastrula stages. Uninjected embryos (Fig. 2A–E) were compared with embryos injected either in the animal pole at stage 2 (Fig. 2F–J), or with embryos injected in the marginal zone at stage 2 (Fig. 2K–O). In general, marginal injections (i.e. in the region, where mesoderm and endoderm are formed) had stronger effects than animal injections. In accordance with the aforementioned failure of blastopore closure for both types of injection, the expression domains are less condensed. The early mesodermal markers Xbra (Fig. 2A, F, K) and Xsna (Fig. 2B, G, L) are slightly downregulated as well as the endodermal marker Xsox17α (Fig. 2D, I, N). This is in agreement with previous results showing an inhibitory effect on mesendoderm formation (Cao et al. 2006, 2007). In contrast, the expression of the paraxial mesoderm marker PAPC is not reduced, rather its expression domain has changed. The dorsal expression gap that is observed in uninjected embryos is closed after injections of Oct25 mRNA (Fig. 2C, H, M). The expression of the neurectodermal marker Sox3 is reduced after marginal injections. Both, marginally and animally injected embryos show an animal shift of the vegetal border of Sox3 expression. Overall, especially the animal injections show relatively weak effects on germ layer formation. Changes in the expression patterns of PAPC and Sox3 rather indicate a dorsoventral modification.
Oct25 effects on germ layer formation and early patterning
The analysis of expression patterns of a series of dorsal axial and paraxial markers confirms these dorsoventral effects. We used the markers Xbra, axial protocadherin (AXPC) and Chordin (chd) to label axial structures and PAPC and MyoD to label paraxial structures at a late gastrula stage (stage 12–12.5), when the most prominent dorsal morphogenetic movement, convergence and extension, is already quite advanced. Uninjected controls show distinct axial and paraxial expression patterns of all analyzed markers (Fig. 3A–E). Embryos injected with Oct25 mRNA display remarkable changes. The axial domain of Xbra is absent (Fig. 3F), AXPC and Chd have a shortened and broadened expression domain (Fig. 3G, H). The paraxial markers PAPC and MyoD, which normally have a small dorsal gap of expression, possess a broad gap after overexpression of Oct25 (Fig. 3I, J). Since in most injected embryos gastrulation is disturbed (as indicated by an oversized blastopore at stage 12.5), we asked, if the differences are attributed to a delay of gastrulation. Therefore, we compared the expression patterns to uninjected embryos of early gastrula stages, using the blastopore size as external criterion. Except for Xbra (Fig. 3K), the used markers show distinct differences (Fig. 3L–O) in level and shape of expression. In Oct25 injected embryos, AXPC is expressed in a restricted domain compared to uninjected embryos with the same blastopore size. Chd expression is stronger and especially around the blastopore extended. PAPC is expressed in the dorsal marginal zone (DMZ). A repression of PAPC in the axial region, which is characteristic for late gastrula stages (Kim et al. 1998), is found in injected embryos, whereas uninjected embryos with a comparable size of blastopore show a continuous expression domain in the DMZ, which is typical for an early gastrula stage (Kim et al. 1998). MyoD expression in Oct25 injected embryos shows a broader and irregular dorsal gap compared to early gastrula controls. Hence we exclude a delay of development as an exclusive cause for the Oct25 induced changes.
In summary, the changed expression domains of axial and paraxial markers after Oct25 overexpression in the animal region do not support the idea of an exclusively antidorsalizing effect of Oct25. The changed shapes of expression domains rather indicate a defect in morphogenesis, most likely by affecting the dorsal convergence and extension movements.
Overexpression of Oct25 in prospective ectoderm induces expression of PAPC without inducing mesoderm
Further examination of embryos of the whole mount in situ hybridization revealed ectopic activation of PAPC expression in the AC region at midgastrula stages (Fig. 4A, E, stage 11). One could assume that this PAPC labeling gleams across the ectoderm from the internalized mesoderm. Therefore, sections from embryos after whole mount in situ hybridization were prepared. They demonstrate that a weak, but distinct expression is found in the AC cells themselves (Fig. 4B, F). Expanded or ectopic expression was neither found for the mesodermal markers Xbra (Fig. 4C, G, compare also Fig. 2A, F) and MyoD (Fig. 3E, J), nor for the endodermal Xsox17α (Fig. 4D, H). Therefore, ectopic mesendoderm induction is unlikely. The epidermal ectoderm marker XK81A1 was repressed (Fig. 4I, K) and the neurectodermal marker Sox2 was not expanded (Fig. 4J, L).
Thus, the PAPC expression is not based on ectopic induction of mesoderm. This was confirmed by polymerase chain reaction (PCR) on AC explants (Fig. 4M). A weak, but distinct expression of PAPC resulted from the injection of Oct25 mRNA. The mesodermal marker Xbra is unchanged, as well as the expression of endodermal Xsox17α. The epidermal ectoderm marker XK81A1 was reduced and neurectodermal Sox2 was slightly increased. The expression of the classical cadherin EP/C-Cadherin was unaffected either.
According to this effect of Oct25 on PAPC expression, the animal Oct25 MO injection reduced the PAPC expression levels in whole embryos at that stage (Fig. 4N). Xbra expression remained unchanged (Fig 4N).
These experiments demonstrate that the PAPC expression is regulated by Oct25. The effects are independent of mesoderm commitment in these cells, since neither the Oct25 overexpression nor the knockdown showed effects on Xbra expression in our type of experiments.
Overexpression of Oct25 affects convergence and extension
Paraxial protocadherin has been described as a modulator of morphogenesis. It is involved in regulation of convergence and extension and of tissue separation behavior, which are based on differential cell adhesion (Kim et al. 1998; Medina et al. 2004; Unterseher et al. 2004; Chen & Gumbiner 2006). If overexpression of Oct25 induces PAPC synthesis, one expects effects on those morphogenetic movements that are regulated by PAPC. We therefore first analyzed how the injection of Oct25 mRNA affects convergence and extension. Oct25 mRNA was injected either in the animal region or the marginal zone or the vegetal pole. At the beginning of gastrulation, approximately quadratic explants [length/width (l/w) ≈ 1] of the normally converging and extending DMZ were explanted (Fig. 5A, 0 h) and cultivated for 24 h. Elongation was then measured and classified into “no elongation” (l/w ≤ 1.5), “weak elongation” (1.5 < l/w < 2.5) and “strong elongation” (2.5 ≤ l/w) (as shown in Fig. 5A). Injection of Oct25 mRNA in the AC region resulted in a weak, not yet significant difference of elongation in DMZ explants (Fig. 5B). But injections in either the marginal zone or the vegetal pole resulted in highly significant reduction of elongation in such explants (Fig. 5B).
In AC explants, which normally do not elongate, elongation can be induced with activin A (van den Eijnden-Van Raaij et al. 1990; Ariizumi et al. 1991). It has been shown formerly that the Oct-factors inhibit this induced elongation (Cao et al. 2006). We confirmed these results using injection of activin A into the blastocoel of Oct25 mRNA-injected and uninjected embryos. The results were highly significant (Fig. 5B, C).
Such a reduction of convergence and extension could be a major factor in the formation of the aforementioned phenotypes. Convergence and extension movements were therefore analyzed in vivo in the gastrulating embryo. We labeled cells of the DMZ with the photoconvertible fluorescent protein EosFP and analyzed them with timelapse videography as formerly described (Wacker et al. 2007). In embryos without Oct25 overexpression, these cells show strong convergence and extension movements during gastrulation (Fig. 5D, Movie S1). This is correlated to closure of the blastopore. In contrast, labeled cells from Oct25 mRNA injected embryos fail to do so (Fig. 5E, Movie S2). The DMZ cells move to the blastopore and remain there at the edge of the not contracting blastopore.
Overall, these experiments demonstrate that the overexpression of Oct25 is strongly affecting convergence and extension movements. This appears at the time, when the first phenotypic effects are observed, namely during gastrulation.
Oct25 overexpression in AC inhibits cell adhesion
On a cellular level, one driving force for morphogenetic movements is cell adhesion (for review [Winklbauer 2009]). We therefore were interested, if Oct25 overexpression changes cell adhesion. Two different assays were performed. With a dissociation assay it is analyzed, how fast tissue explants are disaggregating after removal of Ca2+ and Mg2+. With a reaggregation assay it is measured, how efficient dissociated cells can rebuild aggregates after adding Ca2+ and Mg2+.
In the dissociation assay, ACs showed strong disaggregation after 30 min, when explanted from Oct25 mRNA injected embryos (Fig. 6A, explants on the right side), whereas control explants from uninjected embryos remained relatively compact for up to 1 h (Fig. 6A, left side). This disaggregation can be delayed by coinjection of the mRNA for the classical cadherin EP/C-Cadherin (Fig. 6B). Interestingly, blocking of the PAPC function by coinjection of the dominant inhibitory PAPC construct SXPAP (secreted Xenopus PAPC, also called dnPAPC, Kim et al. 1998) had a similar effect (Fig. 6C). Again, the dissociation in absence of Ca2+ and Mg2+ was delayed.
Reaggregation happened much faster in AC cells from uninjected embryos than in cells from Oct25 injected embryos. In untreated cells big aggregates were found within 120 min, which within one day had formed big balls of tissue (Fig. 7A left column). Cells from Oct25 injected embryos did almost not reaggregate at all (Fig. 7A right column). The coinjection of EP/C-cadherin and Oct25 mRNAs increased the level of reaggregation compared to cells from embryos, which were only injected with Oct25 (Fig. 7B). The dominant inhibitory SXPAP had a similar effect (Fig. 7C).
These experiments clearly indicate that the overexpression of Oct25 results in a substantial reduction of cell–cell adhesion, which may be a cause for the observed disturbance of morphogenetic movements. This can be restored by overexpression of EP/C-cadherin or by inhibition of the PAPC function.
Pluripotency maintaining Pou V factors in Xenopus laevis
The Pou V class transcription factor Oct4 is a key player in a network of transcription factors regulating the commitment and differentiation of ES cells in mammalian organisms (Nichols et al. 1998; Niwa et al. 2000; Pesce & Schöler 2001). It is crucial for maintaining pluripotency in stem cells (Nichols et al. 1998) and in the primordial germ cell lineage (Kehler et al. 2004). In lower vertebrates, embryonic cells remain pluripotent at least until the beginning of gastrulation. Some cells stay multipotent until late gastrulation. These cells cannot be considered as ES cells, since they do not self renew indefinitely. However, as in ES cells, differentiation must be prohibited temporarily.
In Xenopus laevis, three functional homologues of Oct4 have been described to play a crucial role in maintaining embryonic pluripotency. The maternally expressed Oct60 and the (predominantly) zygotically expressed Oct25 and Oct91 cover the early phases of development. Expression levels are highest in those cells that remain multipotent longest, namely in the ectoderm forming from the animal hemisphere (Hinkley et al. 1992). Their major function there is to antagonize different inducing signals, thereby keeping cells at an uncommitted state. Mesendoderm induction is prohibited by antagonizing the VegT activity (Cao et al. 2007), and by inhibiting the activin/nodal pathway (Cao et al. 2006, 2008). In addition further pathways are affected. Oct factors have been described as inhibitors of the β-Catenin pathway (Cao et al. 2007) and as inhibitors of the BMP pathway (Takebayashi-Suzuki et al. 2007). Actually, the commitment to an early neural fate is the only one that is enforced or at least not prevented. This has been shown for both, Oct25 (Cao et al. 2007; Takebayashi-Suzuki et al. 2007) and Oct91 (Henig et al. 1998; Snir et al. 2006; Archer et al. 2011). However, neuronal differentiation is prohibited (Cao et al. 2006; Snir et al. 2006) causing the maintenance of a neurectodermal, but undifferentiated status.
Phenotypes and gastrula morphogenesis
The ectopic expression of Oct25 delivers two very distinctive phenotypes at tadpole stages. The major attribute of the first phenotype is the extensive splitting of axial structures (Fig. 1). In contrast to the medical term “spina bifida”, this induced split axis is not restricted to a failure of neural tube closure, but also includes a gap in the underlying mesodermal tissues. The second phenotype, probably representing a milder form, is characterized by a shortened and condensed trunk and changed shape of the head (Fig. 1). Both phenotypes could be considered as a result of reduced dorsalization. This is opposed by the analysis of axial and paraxial markers, which are still present in treated embryos, but changed in their spatial expression pattern (Fig. 3).
Preceding these late phenotypes, the ectopic expression of Oct25 affects gastrulation. Formation of the blastopore is delayed in such a way as to disable its closure even during neurulation (Fig. 1). Obviously, driving forces of the morphogenesis during gastrulation are affected. Such defects quite often point on a disturbance of the dorsal convergence and extension movements. Even though injection was performed in regions that do not directly participate in these movements, life imaging of gastrulating embryos confirmed this hypothesis, since labeled dorsal mesoderm cells failed to form an elongated structure in Oct25 injected embryos (Fig. 5 and Movies S1, S2). In addition, the corresponding time-lapse movies show that the internalization of the prospective mesoderm from the marginal zone is disturbed. Labeled mesoderm cells remained in the marginal zone instead of internalizing and moving in a direction to the animal pole, as mesoderm from control embryos does (Fig. 5 and Movies S1, S2). However, we were not able to find significant effects of Oct25 on tissue separation (not shown), a cellular behavior that is crucial for the involution of mesoderm (Winklbauer & Keller 1996; Wacker et al. 2000). This indicates that mainly convergence and extension are affected. Further experiments with explants of the DMZ (i.e. cells performing convergence and extension) confirmed the effects of Oct25 overexpression on convergence and extension (Fig. 5).
Cellular and molecular effects of Oct25 overexpression
For either convergence and extension or tissue separation during mesoderm involution, PAPC has been found to be an important molecule (Kim et al. 1998; Medina et al. 2004; Unterseher et al. 2004; Wang et al. 2008). Initially regarded as cell adhesion molecule (Kim et al. 1998), evidence for signaling functions has accumulated (Medina et al. 2004; Unterseher et al. 2004; Wang et al. 2008; Köster et al. 2010). The effects on tissue separation and morphogenetic movements have been traced back to the regulation of the adhesion activity of a classical cadherin (Chen & Gumbiner 2006), which has been found to be crucial for morphogenesis (Brieher & Gumbiner 1994; Zhong et al. 1999; Wacker et al. 2000; Winklbauer 2009). The activity of EP/C-cadherins is downregulated by PAPC, thereby affecting cell adhesion, which is then resulting in changes in convergence and extension (Chen & Gumbiner 2006). In addition, PAPC has been described to interact with the Wnt-receptor Fz7 (Medina et al. 2004). Morphogenesis is affected via regulation of several noncanonical Wnt pathways. This includes the PCP pathway and pathways via small GTPases and JNK (Medina et al. 2004; Unterseher et al. 2004; Wang et al. 2008; Köster et al. 2010).
We find that PAPC is activated in the prospective ectoderm by overexpression of Oct25 (Fig. 4). It is implausible that this activation results from common pathways of mesoderm induction. Oct25 overexpression has been reported to antagonize mesoderm induction (Cao et al. 2006, 2007, 2008). Experiments here have confirmed this by the absence of mesodermal markers in the injected regions (Fig. 4). In addition, since Oct25 overexpression at the same time inhibits Wnt pathways (Cao et al. 2006, 2007; Takebayashi-Suzuki et al. 2007), we postulate that the prevention of convergence extension and mesoderm internalization may directly result from changed cell–cell adhesion. This could happen via the classical cell adhesion molecule EP/C-cadherin, which is functionally downregulated by PAPC (Chen & Gumbiner 2006). This hypothesis is supported by experiments showing that cell adhesion is reduced after the overexpression of Oct25, an effect, which is recovered by coinjection of EP/C-cadherin mRNA or by inhibition of PAPC (Figs 6 and 7). These effects on cell adhesion may cause the changed convergence and extension behavior.
Oct25 as morphogenesis regulator
The overexpression of Oct25 represses several signals involved in commitment of cells (including mesoderm induction, see above). In parallel, it causes an activation of the usually mesodermal PAPC expression in the uncommitted cells of the AC. Based on our results and a model published by Chen & Gumbiner (2006), we postulate that, as a consequence of Oct25 overexpression, cell adhesion via PAPC and the classical EP/C-cadherin is functionally downregulated (Fig. 7D). Such a modulation of cell adhesion is a major mechanism in regulation of morphogenesis, as shown e.g. for convergence and extension (Zhong et al. 1999).
While for the regulation of cell commitment opposite effects are often expected and found for “loss of function” and “gain of function” experiments (e.g. for the Oct genes see Cao et al. 2007, 2008, 2006), the regulation of morphogenesis may function differently. Convergence and extension need a spatially and temporally exact pattern of cell adhesion changes to allow axis elongation (Keller & Danilchik 1988). Any disturbance to either low or to high adhesion should therefore inhibit convergence and extension, which finally results in similar instead of opposite phenotypes. This can be seen e.g. in the classical DAI series (Kao & Elinson 1988). Either hyperdorsalized embryos (trying to perform convergence and extension all around the marginal zone at the same time) or hyperventralized embryos (with a general reduction of convergence and extension) show a shortened axis in comparison to “normal” embryos.
Cao et al. (2006) have described morphological phenotypes resulting from “Oct loss of function” approaches using morpholinos against Oct-25 and Oct-60. Two major effects on morphogenesis resulted from their experiments. First, morpholino injected embryos showed a delayed closure of the blastopore or no closure of the blastopore at all, indicating an arrest of gastrulation. Second, the axial structures of treated embryos were strongly reduced (Cao et al. 2006). Morpholino experiments in zebrafish also displayed an arrest of gastrulation (Burgess et al. 2002). A mutation of the zebrafish Spg/Pou2/Oct4 locus has been described to arrest gastrulation as well. In the MZspg mutation (maternal and zygotic spg), series of experiments proved that this arrest results from effects on cell shape, cell adhesion and cell behavior during early gastrulation (Lachnit et al. 2008).
Here we find comparable phenotypical effects after gain of function experiments with Oct25 mRNA. Similar phenotypes were observed, cell adhesion and cell behavior during early gastrulation were changed. Therefore, we conclude that, apart from its role as a transcriptional regulator of cell commitment, Oct25 might directly act on morphogenesis, namely affecting dynamic cell adhesion via PAPC and EP/C-cadherin. This delivers an explantation for the achieved phenotypes and especially for the seemingly paradoxical observation that both, functional knockdown and overexpression have very similar phenotypical effects.
We thank Nicole Heymann and Cornelia Donow for excellent technical assistance. This study has been aided by a grant from the Deutsche Forschungsgemeinschaft to W.K. (SFB497/A1).