SEARCH

SEARCH BY CITATION

Keywords:

  • direct development;
  • nutritional endoderm;
  • global transcriptional repression;
  • Smad2;
  • Smad4;
  • nodal signaling;
  • endoderm specification

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

The vegetal cells of a Xenopus laevis embryo commit to mesendoderm via the Nodal-signaling pathway. In the direct developing frog Eleutherodactylus coqui, mesendoderm is specified at the marginal zone of the early gastrula, and vegetal core cells transform into nutritional endoderm. Nutritional endoderm, a novel tissue, consists of transient, yolky cells that provide nutrition but remain undifferentiated. We report a dual regulation for the generation of nutritional endoderm. First, differential expressions of the Nodal-signal transducers Smad2 and Smad4 were observed during early gastrulation between the marginal zone and the vegetal core cells. Although EcSmad2 RNA as well as total and activated Smad2 protein were detected in the vegetal core, Smad4 protein was expressed less in vegetal core during early gastrulation. Only 12% and 50% of vegetal core cells were positive for nuclear Smad2 and Smad4 signals respectively compared to 100% of marginal zone cells. These results suggest a signaling disruption in the vegetal core. Second, vegetal core cells were transcriptionally repressed. At the blastula stage, both marginal zone and vegetal core cells were transcriptionally silent, but during early gastrulation, only marginal zone cells became transcriptionally active. This indicates the occurrence of a mid-blastula transition in the marginal zone by early gastrulation, but global transcriptional repression persisted in the vegetal core and its derivative, nutritional endoderm, throughout development. We have described a novel mechanism, which prevents differentiation of the vegetal core through differential Nodal-signaling and global transcriptional repression. J. Exp. Zool. (Mol. Dev. Evol.) 322B: 27–44, 2014. © 2013 Wiley Periodicals, Inc.

Abbreviations
BF

bright field

E

embryo

MZ

marginal zone

NE

nutritional endoderm

NF

Nieuwkoop and Faber stages of development

TS

Townsend and Stewart stages of development

VC

vegetal core

W

whole embryo

In the amphibian model system Xenopus laevis, mesendoderm specification is primed in the yolky vegetal cells during early embryogenesis. Cell fate determinations are specified by an orchestration between different signaling pathways and interplay among essential molecular determinants. One crucial pathway is the Nodal-signaling pathway, which includes Activin and Nodal-related proteins of TGFβ family of growth factors (Schier, 2009). VegT is an important T-box transcription factor in X. laevis, whose RNA is localized to the vegetal cortex of the oocyte. The presence of VegT RNA and protein in the vegetal region of the blastula leads to the specification of endoderm (Lustig et al., 1996; Stennard et al., 1996; Zhang and King, 1996; Horb and Thomsen, 1997; Xanthos et al., 2001).

The Nodal-signaling pathway consists of the several steps (Shen and Schier, 2000; Schier, 2003; Wu and Hill, 2009). The signaling cascade is initiated by binding of ligands to a heteromeric receptor complex of type I and type II serine-threonine kinase receptors and co-receptor EGF-CFC leading to the formation of active receptor complex. This activation leads to the phosphorylation and activation of cytoplasmic Smad2/3 proteins, which in association with Smad4 translocate into the nucleus to bind with additional transcription factors to activate the transcription of distinct but partially overlapping sets of Nodal downstream genes (Ross and Hill, 2008; Schier, 2009).

In addition to the three embryonic germ layers, Eleutherodactylus coqui, the Puerto Rican direct developing tree frog, develops nutritional endoderm (NE), a novel tissue (Buchholz et al., 2007). NE consists of transient, yolk rich cells that provide nutrition to the growing embryo but do not differentiate into any mesendodermal tissues. Unlike X. laevis, NE develops from the vegetal core (VC) of the embryo, whereas the neighboring cells of the marginal zone (MZ) become committed to endoderm/mesoderm fates (Elinson and del Pino, 2012).

By in situ hybridization, Beckham et al. (2003) demonstrated that EcVegT and EcVg1, the E. coqui orthologs of X. laevis VegT and Vg1, are expressed near the animal pole in contrast to X. laevis, where VegT and Vg1 RNAs are vegetally localized (Melton, 1987; Zhang and King, 1996; Horb and Thomsen, 1997). Recombinant analysis by Ninomiya et al. (2001) revealed the presence of strong mesoderm inducing activity in the outermost MZ cells of E. coqui early embryos, but absent in VC. More recently, Karadge and Elinson (2013) detected some inducing activity in VC, albeit delayed.

We investigated the molecular mechanism underlying the origin of NE by asking whether NE develops because of the absence of the molecular determinants of Nodal signaling in the VC in contrast to the MZ. More specifically, we characterized the expression of the Nodal signal transducers, Smad2 and Smad4, during early embryogenesis. We also addressed the functional aspect of this tissue by analyzing its global transcriptional status. Based on our results, we report a dual regulation involved in the generation of NE, first at the level of Nodal-signaling, and second, at the level of global transcriptional activity.

MATERIALS AND METHODS

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

Animals and Embryos

Adult E. coqui frogs were captured on the Big Island of Hawaii under Injurious Wildlife Export permits from the Department of Land and Natural Resources of Hawaii. Male and female frogs were kept as pairs as a reproductive colony at Duquesne University. Adults and embryos were maintained following the protocols approved by the Institutional Animal Care and Use Committee (IACUC). Clutches of embryos were collected and kept in petri dishes on a filter paper soaked in 20% Steinberg's solution. Embryos were staged according to Townsend and Stewart (1985). A TS number was used for each stage from TS3 to TS15, with TS15 denoting the hatching froglets. Embryos from stages earlier than TS3 (neural tube formation) were named according to the Nieuwkoop and Faber (1994) stages for X. laevis. Embryos with a thick blastocoel roof, small blastocoel, and absence of a dorsal lip were considered to be at NF8. Embryos at NF10 were characterized by thin blastocoel roof, a large blastocoel, and the appearance of the dorsal lip.

Ovarian oocytes was collected by removal of the ovary from a female, anesthetized in 0.1% MS222 (Tricaine methane sulfonate), pH 7.4, and killed by decapitation. Dissected ovaries were placed in 200% Steinberg's solution until use. Oocytes were defolliculated with watchmaker's forceps after incubating for 1 hr in Ca2+ free 200% Steinberg's solution (116 mM NaCl, 1.34 mM KCl, 0.8 mM MgSO4, 1 mM EGTA, 9.2 mM Tris, pH 7.4).

In order to perform experiments using molecular techniques, embryos were flooded with 20% Steinberg's solution to cause swelling of the jelly layers. Fine watchmaker forceps were used to remove the outer and middle jelly layers. Blastula (NF8) and gastrula (NF10) embryos were subjected to additional treatments to remove the inner jelly layer and the fertilization membrane. First, they were incubated in 3% cysteine, pH 8, for 8–10 min with gentle swirling to remove the inner jelly layer. After thoroughly rinsing three to four times with 20% Steinberg's solution, embryos were incubated in Hennen's solution [2.0% cysteine, 0.2% papain, and 0.2% α-chymotrypsin (Type II, from bovine pancreas), pH 8.0] (Hennen, 1973) for 45–50 sec to weaken the fertilization membrane. Embryos were immediately washed three to four times with 20% Steinberg's solution and transferred to a new petri dish containing 1% Bovine Serum Albumin (BSA) in 100% Steinberg's solution. The weakened fertilization membrane was carefully removed with fine forceps under a dissecting microscope.

Embryos were dissected in 100% Steinberg's solution with 1% BSA. At early stages, embryos were physically manipulated with the help of a hair loop. The embryo was positioned with its animal pole upward. At NF8, a fine incision was made with forceps in a circle around the animal pole and the animal cap was removed. After this, the embryo was turned upside down bringing the vegetal pole up. Incisions were made using a hair loop to separate the thin, peripheral MZ tissues (prospective definitive endoderm or DE) of both dorsal and ventral sides, from the larger, inner VC tissue (prospective NE). The MZ and VC were separated into different microfuge tubes for further use. For NF 10 embryos, the presence of a translucent blastocoel roof and the appearance of the blastopore lip made dissection of prospective DE and NE much easier. For post-gastrulation stages when the embryonic regions have already developed, dissections became less complicated. For all post-gastrulation stages the whole embryo was dissected into the embryonic tissues and the NE.

Dissociation of MZ and VC cells was required for the immunostaining experiments. After dissection, MZ and VC tissues were incubated in Ca2+–Mg2+-free modified 100% Steinberg's solution (68 mM NaCl, 1.34 mM KCl, 4.05 mM Na2HPO4, 0.73 mM KH2PO4, 1.0 mM EGTA, pH 7.4) (modified from Ninomiya et al., 2001). A 15-min incubation was enough to dissociate respective tissues into individual cells. Cells were washed in 1× PBS and were ready to use. For embryos with definitive body structures, dissociation did not occur even after incubating for more than an hour. For dissociation of NE cells from embryos at TS6, TS8, TS11, and TS14, dissected NE tissues were incubated in Ca2+-free 200% Steinberg's solution with 1.0 mM EGTA (116.4 mM NaCl, 1.342 mM KCl, 0.812 mM MgSO4·7H2O, 4.05 mM Na2HPO4, 0.73 mM KH2PO4, 1.0 mM EGTA, pH 7.4), (modified from Shibuya and Masui, 1988) for at least an hour.

Total RNA Extraction and cDNA Synthesis

Total RNA was isolated from whole embryos as well as from dissected prospective DE and prospective NE following the Trizol RNA extraction protocol. A total of 2–5 μg RNA was used as a template to synthesize cDNA by reverse transcription with M-MLV-RT enzyme. Each reaction was accompanied by a negative RT reaction, which contained all of the reagents except M-MLV-RT enzyme. The cDNAs were purified by running the reaction mixtures through spin-50 mini-columns (1415-0600; USA Scientific, Orlando, FL, USA).

Cloning EcSmad2 cDNA

The complete Open Reading Frame (ORF) of EcSmad2, most of the 5′UTR, and the complete 3′UTR ending with a poly(A) tail, were cloned in four steps. First, degenerate primers were designed based on Smad2 ORFs from X. laevis, X. tropicalis, mice, and human. The forward primer was 5′-ACAGGCCTTTACAGCTTCTCTGAACA-3′ (EcSmad2-F2), and the reverse primer was 5′-TTABGACATGCTTGAGCADCGSACTG-3′ (EcSmad2-R1). PCR amplification was performed by denaturation at 94°C for 3 min, denaturation at 94°C for 45 sec, annealing at 52°C for 45 sec, and extension at 72°C for 90 sec for 29 cycles. The final extension was performed at 72°C for 10 min. The degenerate primer pair generated two bands of ∼1,100 bp, towards the 3′ end of the ORF. The doublet of bands indicated the presence of both isoforms as seen in other vertebrate species. An alternatively spliced variant of Smad2, known as Smad2Δexon3 was reported in X. laevis (Faure et al., 2000). The doublet of DNA bands was gel purified, cloned, and sequenced at the DNA Sequencing Facility, University of Pittsburgh, Pittsburgh, PA, using ABI 3100 Sequencer (Applied Biosystems, Inc., Foster City, CA, USA). The clones lacked the beginning 300 bp, but had the rest of the ORF ending at a stop codon. The NCBI BLAST program confirmed the identity as Smad2.

To obtain the 3′UTR of EcSmad2, we performed PCR on Custom SMART™ E. coqui ovarian cDNA Library in λTriplEx2™ (Clontech; CS1023u). Two vector specific primers, 5′ pTriplEx2 Seq Pri (forward, 5′-GCCAAGCTCCGAGATCTGGACGAGC-3′) and 3′ pTriplEx2 Seq Pri (reverse, 5′-GAATTGTAATACGACTCACTATAGGGCGAA-3′), flanking the Multiple Cloning Site (MCS) were separately used in combination with an exact EcSmad2 forward primer, EcSmad2-F18 (5′-GGGCTTTGAAGCAGTTTACCAGTTAACG-3′). As the orientation of the inserts in the vector was not known, both forward and reverse primers specific for vector sequence was tried. Sequencing of the amplified PCR product revealed the presence of the stop codon “TAA,” a polyadenylation signal “AATAAA” marking the cleavage site for the RNA, and a poly(A)-tail 13 bp past the signal.

We used a 5′RACE PCR kit (18374-058; Invitrogen, Grand Island, NY, USA) to amplify the initial missing segment of the EcSmad2 ORF along with the 5′UTR, following the Invitrogen's instructions. Two gene specific exact reverse primers - EcSmad2-RP16 GSP1 (5′-GCATATTCACAATTTTCAAT-3′) and EcSmad2-RP17 GSP2 (5′-CCAGAGACGACAGTAGATAACGTGCGGTAA-3′) were designed for use with the RACE specific forward primers from the kit. The primers were used on cDNA templates prepared from E. coqui TS3 embryos, and the PCR product was gel purified, cloned, and sequenced. Sequencing data confirmed the presence of the initial missing segments for both the isoforms.

Based on the final compiled sequences of the isoforms, two more primers were designed to PCR amplify the full length ORF from TS3 cDNA template in order to clone and make freezer stocks. Sequences for the forward and the reverse primers were 5′-ATGTCCTCCATACTGCCCTTTACACC -3′ and 5′-TTAGGACATGCTTGAGCAGCGGA-3′, respectively. Final sequences were analyzed by ClustalW program (www.ebi.ac.uk/clustalw).

Analysis of EcSmad2 Expression Using Real-Time PCR

Temporal and spatial expression of EcSmad2 was detected by qPCR using Maxima SYBR GREEN qPCR Master Mix (K0221; Fermentas, Rockford, IL, USA), and ΔΔCt was performed to quantitate the results. EcSmad2 ORF specific forward and reverse primers, EcSmad2-qPCR FP1 (5′-CCCTTTACACCTCCCGTTGTGAAACGT-3′) and EcSmad2-qPCR RP1 (5′-CCTTCTCGAGTTCGTCCAGCTGGC-3′) were used for all qPCR experiments. A pair of primers for the ribosomal protein-coding gene EcL8, EcL8-F4 (5′-GAAGGTCATCTCTTCTGCAAACAGAGC-3′) and EcL8-R5 (5′-TAAGACCAACTTTGCGACCAGCTGG-3′) served as the endogenous control. For temporal expression, cDNAs for qPCR were synthesized using 5 μg of RNAs extracted from whole E. coqui oocytes and embryos at various stages of development. For spatial expression, cDNAs for qPCR were synthesized using RNAs extracted from dissected DE and NE of embryos at various stages. In order to ensure good quality of cDNAs, a regular reverse transcriptase PCR (RT-PCR) was performed before every qPCR run.

To do a relative quantification, reactions were set up in MicroAmp fast optical 48- or 96-well plates (Applied Biosystems) with 1.5 µL of cDNA template (150–200 ng) in 20 µL total. Each PCR reaction was run in triplicate. The reactions were run on StepOne real-time PCR System (Applied Biosystems). A ΔΔCt method of qPCR program was set to run, and a melt curve was obtained for amplification phase each time. EcSmad2 was amplified by initial denaturation at 94°C for 10 min, followed by denaturation at 94°C for 50 sec, annealing at 56°C for 50 sec, and extension at 72°C for 75 sec for 40 cycles. The experiments were analyzed using StepOne V2.0. The data was exported as an Excel spread sheet and processed in Microsoft Excel 7.0. Ct values for EcSmad2 were normalized by subtracting the mean Ct for EcL8 from the mean Ct for EcSmad2. These produced the ΔCt values for EcSmad2 from each developmental stage. Next, the relative EcSmad2 expression was quantified using ΔΔCt, where the ΔCt value for each stage or tissue was normalized to that of a single stage or a specific tissue by subtraction (oocyte in temporal expression and MZ or E in spatial expression experiments). Finally, the relative expression or the RQ values were calculated by raising 2 to the negative value of ΔΔCt for each stage and then plotted to show the relative EcSmad2 expression.

Antibodies

Purified mouse monoclonal anti-Smad2/3 (610842; BD Transduction Laboratories, San Jose, CA, USA) was used at a dilution of 1:300 for all Western blotting. Rabbit polyclonal anti-Phospho-Smad2 (Ser465/467) antibody (3101; Cell Signaling, Danvers, MA, USA) was used at 1:150 for Western blotting and 1:25 for immunostaining of dissociated cells. Rabbit anti-Smad4 polyclonal antibody (PA1-41292; Thermo Scientific, Rockford, IL, USA) was used at 1:150 in Western blotting and 1:50 for immunostaining. To determine the transcriptional status of MZ (DE) and VC (NE) cells by immunostaining, we used three primary mouse monoclonal antibodies against the RNA Polymerase II enzyme (RNAPII). 8WG16 (MMS-126R; Covance, USA) was used at 1:50 to detect the unphosphorylated RNAPII large subunit. H14 (MMS-134R; Covance) was used at 1:50 dilution to detect the Serine 5-phosphorylation of RNAPII CTD, which indicated transcription initiation. H5 (MMS-129R; Covance) was used at 1:50 to detect Serine 2-phosphorylation of RNAPII CTD, a mark of transcription elongation. Mouse monoclonal α-tubulin antibody DM1A (ab7291; Abcam, Cambridge, MA, USA) was used at 1:40,000 for all Western blotting.

For Western blot analysis, secondary antibodies were: goat anti-mouse IgG with Peroxidase Conjugate (A2304; Sigma, St. Louis, MO, USA), was used at 1:500 against anti-Smad2, and at 1:40,000 against anti-α-tubulin; anti-rabbit IgG with Peroxidase Conjugate (A0545; Sigma) was used at 1:200 (against anti-PSmad2) and at 1:300 (against anti-Smad4). For immunocytochemistry studies, secondary antibodies, used at 1:250 dilutions, were: Alexa fluor® 488 Goat Anti-Mouse IgG (H + L) (A11029; Invitrogen™/Molecular Probes, Grand Island, NY, USA), Alexa fluor® 488 Goat Anti-Rabbit IgG (H + L) (A11034; Invitrogen™/Molecular Probes).

Protein Isolation, SDS–PAGE, and Western Blot Analysis

Preparation of the protein from yolky E. coqui embryos was difficult due to the large amounts of lipids and yolk platelets in them and the lack of existing protocols. In order to establish a standard protocol to isolate protein from E. coqui embryos, we first used some of the standard protein isolation protocols used for X. laevis embryos (Faure et al., 2000; Birsoy et al., 2005), but those did not work on E. coqui embryos. We altered the recipe several times, leading to a series of buffer recipes. The final recipe, which was modified after Callery and Elinson (1996), almost eliminated interference by yolk platelets. The buffer contained 30 mM Tris pH 7.5, 4 mM EDTA, 10 mM MgCl2·6H2O, 0.1 nM CalyculinA (C5552-10UG; Sigma), 5 mM NaF (919-25ML; Sigma), 20 µL of PMSF (93482-50ML-F; Sigma), and Protease Inhibitor Cocktail (PIC P-8340; Sigma) in deionized water.

Eight to 10 E. coqui embryos were homogenized in 150 µL of buffer for 30 sec. In the case of MZ or VC from early stages and DE or NE from post-gastrulation stages, 15–16 pieces of dissected tissues were homogenized. Extracts were centrifuged at 16,100g for 15 min at 4°C. After spinning, the yolky, white masses settled to the bottom of the tubes. The clear supernatants were transferred to new tubes and protein concentrations were determined using BCA Protein Assay Reagent kit (23225; Thermo Scientific). An equal volume of freshly made 2× Laemmli sample buffer (161-0737; BioRad, Hercules, CA, USA) was mixed thoroughly with the supernatants and boiled for 5–6 min. Finally, tubes were spun briefly at room temperature and stored at −20°C until use. Once the concentrations of the protein preparations were determined, they were equalized to the lowest one by adding 1× Laemmli sample buffer.

One hundred micrograms of each protein sample was loaded per lane of a 10% Tris–HCl Ready Gel (161-1155; BioRad), resolved at 100 V for 90 min, and transferred to Immun-Blot polyvinylidene difluoride (PVDF) membrane (162-0177; BioRad) at 150 V for 80–90 min. The membranes were incubated in TBST (10.0 mM Tris pH 8, 0.15 M NaCl, 1% BSA, 0.05% Tween-20) for 20 min. Membranes were blocked in 5% nonfat dry milk (NFDM) in TBST for 1.5 hr or overnight. Blocking of membranes was followed by incubation in specific primary antibody dilutions, usually overnight at 4°C. The next morning, each membrane was washed in TBST. In case of anti-PSmad2 antibody, the membrane was incubated after washing in Peroxidase Suppressor (35000; Thermo Scientific) for 30 min to reduce background staining. All membranes were incubated for 2 hr at RT in their respective HRP-tagged secondary antibody, made in 2% NFDM in TBST. After washing, each membrane was treated with SuperSignal® West Femto Maximum Sensitivity Substrate (34096; Thermo Scientific) for 5 min. A Typhoon 8600 Variable Mode Imager (Molecular Dynamics, Sunnyvale, CA) was used for imaging the Western blot. Scan area was selected using Scanner Control software (Amersham Biosciences, Pittsburgh, PA, USA), and resolution was selected at high with mode of scanning selected at Chemiluminiscence. Developed images were analyzed using Image Quant Version 5.0 (Molecular Dynamics) or Adobe Photoshop CS5 Extended software.

Immunocytochemistry and DAPI Staining of Dissociated Cells

Immunocytochemistry was performed on dissociated cells to detect the cellular location of selected proteins. After cells were completely dissociated, they were washed in PBS. Cells were fixed in 3.7% formaldehyde in PBS for 35 min with gently shaking. After that, cells were washed in PBS followed by permeabilization in 0.3% TritonX-100 in (07426; ICN Biomedicals, Inc., Irvine, CA, USA) for 35 min with constant gentle shaking. Cells were washed again and blocked in 10% Goat Serum (16210-064; GIBCO, Grand Island, NY, USA) in PBS for 90 min on a shaker. Cells were incubated in a specific primary antibody overnight at 4°C with gentle shaking followed by washes in 1% Goat Serum in PBS. Incubations in appropriate dilutions of specific secondary antibodies were performed for at least an hour at room temperature. Cells were washed again in 1% Goat Serum followed by DAPI staining. DAPI stock solution was made by dissolving 1 mg DAPI (D1306; Invitrogen/Molecular Probes) in 1 mL DEPC water and stored in aliquots in 1.5 mL microfuge tubes at −20°C. Cells were incubated in 1:500 to 1:1,000 dilution of the DAPI stock solution in 1% Goat Serum in PBS for 20–35 min on a shaker, followed by washing in 1% Goat Serum. Finally, cells were treated with SlowFade® Antifade kit (S2828; Invitrogen/Molecular Probes). When ready, cells were mounted in antifade medium on glass slides and observed with a Nikon Microphot-SA microscope (Nikon Corporation, Tokyo, Japan). Photographs were taken after exciting various fields with bright or fluorescent light through FITC or UV filter. A QED camera was used to take photographs, and they were processed using QCapture software and Adobe Photoshop.

For each of the experiments, separate sets of cells were treated with “no primary antibody, only secondary antibody” and “no secondary antibody, only primary antibody,” which served as negative controls with no signal being detected. Each experiment was repeated at least three times. Each time, a minimum of 100 cells was counted. As some of the VC or NE cells are characterized to be either anuclear or multinuclear, it was made sure to count only the cells, which contained at least one nucleus.

RESULTS

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

Cloning Smad2 From E. coqui

The role of Smad2 in germ layer specification has been well documented in X. laevis and other vertebrate species, but nothing is known for direct developing amphibians. Smad2 is an important component of the Nodal-signaling pathway and also serves as a hub connecting other essential important signaling cascades. As the genome of E. coqui is not sequenced, we cloned the Smad2 cDNA from E. coqui embryo to examine its expression.

Alignment of Smad2 sequences revealed a high degree of conservation among vertebrate species. Based on this, the size of EcSmad2 ORF was predicted to be ∼1,400 bp. Sequencing revealed the full length ORF to be 1,404 bp, which would code for 467 amino acids and a peptide with a predicted molecular weight of 52.4 kDa. The splice variant, EcSmad2▵exon3, had an ORF of 1,314 bp. The predicted peptide was 437 amino acids, and the estimated molecular weight was 49 kDa. When the full length EcSmad2 ORF nucleotide sequence was aligned with those of X. laevis and X. tropicalis, the ClustalW result showed 86% nucleotide identity shared among the three frog species. The EcSmad2 nucleotide sequence was 82% identical to that of both human and mouse. Alignment of the EcSmad2 predicted protein sequence with Smad2 of different organism revealed 99% conservation with X. laevis and X. tropicalis, and 98% with human, chicken and mouse. The GenBank accession numbers for EcSmad2 and EcSmad2Δexon3 sequences are KC986816 and KC986817, respectively.

Temporal and Spatial Expression of EcSmad2 in E. coqui Early Embryos

It was important to know whether EcSmad2 RNA was maternally contributed, whether it was expressed throughout all of the early developmental stages, and whether expression was spatially restricted. It was possible that EcSmad2 expression was spatially restricted to the marginal zone (MZ), which forms the definitive endoderm (DE), and not present in the vegetal core (VC), which forms the nutritional endoderm (NE). This idea was based on the fact that the NE does not differentiate into any adult organs; it just provides nutrient to the growing embryo. We used ΔΔCt method of real-time PCR (qPCR) to detect EcSmad2 expression. Expression of the ribosomal protein-coding gene, EcL8, was detected in both the MZ (or DE) and VC (or NE) tissues (Callery and Elinson, 2000; Ninomiya et al., 2001), so we used it as an endogenous control. Each qPCR experiment for temporal or spatial expression was performed three times independently with RNAs from different sources, and the overall results are as follows.

To detect temporal expression, RNA was isolated from whole embryos at each stage, denoted as a specific stage followed by “W.” Stages used were oocyte, blastula (NF8), gastrula (NF10), and post-gastrulation (TS3–TS8). The RQ value for oocyte was set to one, and the RQ values for other stages were expressed relative to oocyte. The results revealed a maternal contribution of EcSmad2 RNA (Fig. 1A). The level of EcSmad2 RNA stayed high at NF8 and NF10, followed by a sharp decline in post-gastrulation stages.

image

Figure 1. Temporal and spatial expression of EcSmad2. (a) Temporal expression: Relative expression (RQ) at the oocyte stage was set to 1 to serve as reference, and the corresponding RQ values of all other stages are plotted relative to oocyte. Each striped bar represents the mean of three independent experiments. There is maternal contribution of EcSmad2 RNA. RNA was also detected at lower levels in NF8 and NF10 followed by a post-gastrulation decline. (b) Spatial expression: RQ of either marginal zone (MZ) or embryo (E) from each stage was set to 1 (black bars) to serve as a reference for the corresponding RQ values of the vegetal core (VC) or nutritional endoderm (NE) samples (stripe bars). The NF8 and NF10 stages showed higher levels of EcSmad2 expression in VC, the prospective NE, whereas all the post-gastrulation stages tested showed comparatively lower levels of EcSmad2 expression in NE, compared to E.

Download figure to PowerPoint

For spatial expression, RNAs were isolated from dissected embryonic tissues from each stage. For NF8 and NF10, RNAs were extracted from dissected MZ and VC tissues. For post-gastrulation stages from TS3 to TS8, the dissected tissues were denoted by the stage followed by “E” for embryo or “NE” for nutritional endoderm. The RQ values of MZ or E from each stage were set as one, and the RQ values for VC or NE were expressed relative to MZ or E. The results indicated higher relative expression of EcSmad2 RNA in the VC at both NF8 and NF10 (Fig. 1B). This pattern changes after gastrulation. In all post-gastrulation stages, a relatively lower level of EcSmad2 RNA was present in NE.

Differential Expression of Smad2, Phospho-Smad2, and Smad4 in the Prospective Definitive Endoderm Versus Prospective Nutritional Endoderm

Although the qPCR results provide information for the presence of EcSmad2 RNA in the prospective NE, it was important to look at the expression of the total Smad2 and active, phosphorylated forms of Smad2 protein to determine whether its signaling pathway was active. Detection of Smad2 versus PSmad2 in the prospective DE versus prospective NE of early embryos was accomplished by Western blot analysis using anti-Smad2/3 and anti-PSmad2 antibodies.

Although anti-Smad2/3 antibody was raised against mouse Smad2 amino acids 142–263, it was used successfully to detect X. laevis Smad2 protein on a Western blot (Faure et al., 2000; Birsoy et al., 2005). The PSmad2 (Ser465/467) antibody was raised against a synthetic phospho-peptide and detected Smad2, when dually phosphorylated at Ser465 and Ser467. The antibody cross reacts with X. laevis PSmad2 (Faure et al., 2000; Birsoy et al., 2005). Based on EcSmad2 sequencing data, the predicted peptide sequence is a 99% match with those of X. laevis and X. tropicalis. We tested the antibody on E. coqui protein samples and found that it also cross reacts with E. coqui PSmad2. Due to the high conservation of C-terminal sequences between Smad2 and Smad3, this antibody may also detect the presence of PSmad3. In X. laevis, Smad3 provides a lesser contribution in endoderm specification (Lee et al., 2001; Whitman, 2001; Yeo and Whitman, 2001), so we will refer to the bands detected in E. coqui as Smad2 and PSmad2.

The temporal expression of both Smad2 and PSmad2 was detected in protein preparations made from oocyte and embryos at stages NF10, TS3, TS5, and TS7 (Fig. 2A). Both antibodies detected two isoforms of EcSmad2. Levels of EcSmad2 expression increased with developmental stage (Fig. 2A).

image

Figure 2. Smad2 and PSmad2 in E. coqui development. (a) Temporal expression: Western blot was performed on proteins isolated from E. coqui oocyte, NF10, TS3, TS5, and TS7. The amount of protein loaded in all the lanes was 75 μg. The top panel shows anti-α-tubulin antibody staining, which served as a loading control. The middle panel represents expression of native EcSmad2 detected by purified mouse monoclonal anti-Smad2/3. The bottom panel shows the presence of active PSmad2 detected by rabbit polyclonal Phospho-Smad2 (Ser465/467) antibody. There was a general trend of increased band intensity with development. The experiment was performed three times with three independently prepared protein samples from different embryo clutches. (b) Spatial expression: Western blot of dissected MZ and VC from gastrulation stage NF10 and embryo (E) and NE from post-gastrulation stages—TS3, TS5, and TS7. Anti-α-tubulin antibody served as a loading control. The middle panel represents expression of native Smad2, and the bottom panel shows the presence of active PSmad2. Smad2 and PSmad2 were detected in MZ and VC from NF10. Spatial expression experiments were performed three times with three independently prepared protein samples from different embryo clutches. Amount of protein loaded in the lanes was 100 μg.

Download figure to PowerPoint

To detect the levels of Smad2 and PSmad2 in the prospective DE versus prospective NE, Western blotting was performed on protein preparations from dissected tissues. Smad2 as well as PSmad2 were detected in both embryonic and NE tissues from post-gastrulation stages (Fig. 2B). Levels of expression in NE for both forms of the protein were lower than in embryos. Both Smad2 and PSmad2 were detected in VC at NF10 (Fig. 2B), based on three independent experiments using embryos from different clutches. Although PSmad2 in VC was detectable only at a low level on Westerns, immunocytochemistry confirmed its presence as described in the next section.

Based on the Western analysis, both Smad2 and PSmad2 were present in VC cells of the embryo. Therefore it was necessary to look at the expression level of Smad4, the partner of PSmad2 in nuclear translocation of the protein complex (Kumar et al., 2001; Lee et al., 2001; Yeo and Whitman, 2001; Ross and Hill, 2008). Although the EcSmad4 gene was not cloned, several commercial antibodies against Smad4 were available. One antibody was rabbit anti-Smad4 polyclonal antibody (Thermo Scientific), which was raised against synthetic peptides corresponding to amino acid 186–199 and 509–523 of human Smad4. This antibody was never used for any frog samples, but based on sequence conservation, it was predicted to cross react with X. laevis. There are two isoforms of Smad4, α and β, present in X. laevis (Howell et al., 1999; LeSueus and Graff, 1999; Masuyama et al., 1999; Hill, 2001). The predicted molecular weights for X. laevis Smad4α and Smad4β are 59.8 and 61.2 kDa, respectively.

Our E. coqui Western results, both temporal (Fig. 3A) and spatial (Fig. 3B), indicated the presence of two bands for EcSmad4 around 60 kDa. In the absence of EcSmad4 sequence information, it is unclear which band corresponded to which isoform. Based on the protein size prediction in X. laevis, the higher molecular weight band was tentatively called EcSmad4β, and the lower molecular weight band was EcSmad4α. EcSmad4β showed a higher level of expression compared to EcSmad4α. Like EcSmad2 (Fig. 2A), the level of EcSmad4 increased with development (Fig. 3A). As we were most interested in early embryonic development, the spatial expression was determined for NF10. Although both isoforms were expressed in VC at NF10, the level of expression of the putative EcSmad4β was much lower than that in MZ (Fig. 3B). Interestingly, Howell et al. (1999) reported that until mid-gastrualtion in X. laevis, Smad4β, not Smad4α, carries Nodal-signaling. Moreover, it has also been reported that Smad4β plays crucial role in X. laevis mesendoderm induction via mediation of endogenous Nodal and BMP signaling (Chang et al., 2006).

image

Figure 3. Smad4 in E. coqui development. (a) Temporal expression of Smad4 isoforms. Top panel: α-tubulin was used as a loading control for the experiment. Bottom panel: Two isoforms of Smad4 were detected (arrows). Based on X. laevis sequence data, the band with higher molecular weight (MW) is tentatively Smad4β and the band with lower MW is tentatively Smad4α. One hundred micrograms of proteins from three stages were run and probed with anti-Smad4 antibody. (b) Spatial expression of Smad4 isoforms in E. coqui NF10. One hundred micrograms of protein from NF10 whole embryos (W), and dissected VC and MZ were run and probed with anti-Smad4 antibody. Two bands represent the two isoforms of the Smad4 protein (arrows). For both temporal and spatial analysis, each experiment was run three times with independently isolated proteins from different embryo clutches.

Download figure to PowerPoint

Cellular Location of PSmad2 and Smad4 in Prospective DE Versus Prospective NE

Western blots with antibodies against PSmad2 and Smad4 not only detected the presence of the respective proteins in all the tested stages of E. coqui development, but also indicated the presence of these TGF-β pathway components in VC and NE. These results led us to investigate the cellular locations of these proteins. Given that PSmad2 was detected in the VC of NF10 embryos, it was important to see whether this active protein is nuclear or not. Immunostaining experiments were carried out on dissociated MZ and VC cells from NF10 embryos. Only cells, where nuclei were clearly seen by DAPI staining, were scored. At early gastrulation (NF10), 100% of the prospective DE cells from MZ showed PSmad2 nuclear localization, but only 12% of VC cells had nuclear localization (Table 1; Fig. 4). The rest had a scattered cytoplasmic signal. As an approach to further confirm the PSmad2 immunostaining data, we injected EGFP-Smad2 capped RNA into the vegetal half of the 8- to 32-cell stage E. coqui embryos to detect the nuclear localization of PSmad2 protein. A tracer level of 10 pg mRNA injection led to nuclear EGFP signal in only 8% of the VC cells (N = 500; data not shown).

Table 1. Summary of immunostaining results with anti-PSmad2 and anti-Smad4 antibodies
 MZ CellsVC Cells
Nuclear Psmad2100 % (N = 80)12% (N = 150)
Nuclear Smad4100% (N = 100)50% (N = 150)
image

Figure 4. Nuclear localization of PSmad2 in NF10 embryos. Dissociated MZ and VC cells from NF10 embryos were stained with DAPI (b,e,h,k) to show the position of nuclei. The same cells were stained with anti-PSmad2 (Ser465/467) antibody to detect nuclear localization of PSmad2 (c,f,i,l). (a,d,g,j) represent bright field (BF) images of the same fields. MZ cells showed nuclear signal for PSmad2 overlapping their respective DAPI signals (b,c), whereas most VC cell nuclei were PSmad2 negative (e,f,h,i,k,l). Scale bar represents 5 µm.

Download figure to PowerPoint

About 50% of the VC cells were positive for nuclear localization of Smad4, whereas 100% of the MZ cells were positive (Table 1; Fig. 5). In X. laevis, the isoform Smad4α shuttles between cytoplasm and nucleus, whereas Smad4β is exclusive to the nucleus (Howell et al., 1999; Masuyama et al., 1999; Hill, 2001). The antibody used in immunostaining experiments was the same one used in Western blotting, where two bands were detected at the predicted molecular sizes. There was no way at present to know whether the nuclear Smad4 signal was α or β.

image

Figure 5. Nuclear localization of Smad4 in NF10 embryos. Dissociated MZ and VC cells from NF10 embryos were stained with anti-Smad4 antibody (c,f,i,l,o) and counter-stained with DAPI (b,e,h,k,n). (a,d,g,j,m) represents bright field (BF) images of the same fields. Scale bar represents 5 µm.

Download figure to PowerPoint

Determination of Transcriptional Status of VC Cells

The presence of both PSmad2 and Smad4 in VC cells but their lack of differentiation to adult tissues leads to the question whether the VC cells are transcriptionally active. The activity of RNA Polymerase II (RNAPII) is determined by the phosphorylation state of the CTD heptapeptide YSPTSPS (Corden, 1990; Phatnani and Greenleaf, 2006; Venkatarama et al., 2010). There are four CTD phosphorylation states. The promoter regions are occupied normally by the unphosphorylated RNAPII. Serine 5-phosphorylation initiates transcription and Serine 2-phosphorylation leads to elongation. The presence of only Serine 2-phosphorylation without Serine 5-phosphorylation leads to termination.

To determine transcriptional status of the prospective NE cells, VC and MZ cells were dissociated from embryos at NF8 and NF10, and NE cells were dissociated from embryos after gastrulation. Cells were stained with antibodies against RNAPII CTD-Ser 2 (H5), CTD-Ser 5 (H14), and unphosphorylated RNAPII large subunit (8WG16) as a positive control.

At NF8, both MZ and VC cells were positive for unphosphorylated RNA Pol II CTD (8WG16), but failed to show any signal for transcription initiation (H14) or transcription elongation (H5) (Table 2; Fig. 6). At NF10, MZ and VC cells both showed positive signals for H14 antibody indicating the occurrence of transcription initiation. MZ cells, but not VC cells, were positive for H5, which marked transcription elongation (Table 2; Fig. 7). This indicates that MZ cells were transcriptionally active, whereas VC cells were not. This change in transcriptional activity in MZ cells could serve as the first evidence for mid-blastula transition (MBT) in E. coqui and suggested that it occurred between NF8 and NF10. MBT, a critical event, marks the onset of embryonic gene expression in X. laevis and other species (Gerhart, 1980; Heasman, 2006).

Table 2. Summary of immunostaining results with 8GW16, H14, and H5 on MZ and VC cells from NF8 and NF10, and NE cells from TS3, TS6, TS8, TS11, and TS14
StageTissue8WG16 nuclear signalH14 nuclear signalH5 nuclear signal
No. of cellsNoFaintStrongNo. of cellsNoFaintStrongNo. of cellsNoFaintFaint
NF8MZ10026921001000010010000
 VC2001071831501500015015000
NF10MZ8003771500614415002148
 VC2007019320018713020019550
TS3NE2002181802001946020019180
TS6NE2006219220019010020019460
TS8NE200681862001982020019550
TS11NE2002219620018911020019550
TS14NE20002198200185150200186140
image

Figure 6. Transcriptional status of MZ and VC at NF8. MZ and VC cells were stained with 8WG16 (c,f) and counter-stained with DAPI (b,e). (a,d) represent bright field (BF) images of the same fields. 8WG16 recognized unphosphorylated RNA Pol II large subunit and served as a positive control. MZ and VC cells were stained with H14 (i,l) and counter-stained with DAPI (h,k). (g,j) represent BF images of the same fields. H14 recognized phosphoserine-5 version of RNA Pol II, which indicated transcription initiation. Neither MZ nor VC nuclei were stained with H14. MZ and VC cells were stained with H5 (o,r) and counter-stained with DAPI (n,q). (m,p) represent BF images of the same fields. H5 recognized phosphoserine-2 version of RNA Pol II, which is the mark for transcriptional elongation. Neither VC nor MZ nuclei stained with H5. Scale bar represents 5 µm.

Download figure to PowerPoint

image

Figure 7. Transcriptional status of MZ and VC at NF10. MZ and VC cells dissociated from NF10 embryos were stained with 8WG16 (c,f) and counter-stained with DAPI (b,e). (a,d) represents BF images of the same fields. Nuclei of both cell types were positive for 8WG16. MZ and VC cells were stained with H14 (i,l) and counter-stained with DAPI (h,k). (g,j) represents BF images of the same fields. Nuclei of both MZ and VC cells were positively stained for H14. Both types of cells were stained with H5 antibody (o,r) and counter-stained with DAPI (n,q). (m,p) represents BF images of the same fields. Only MZ, and not VC nuclei were positive for H5. Scale bar represents 5 µm.

Download figure to PowerPoint

At post-gastrulation stages, the embryo was already formed, so it was hard to dissociate individual cells from the rigid embryonic tissues. For all post-gastrulation stages, undissociated embryonic tissues were stained and were positive for all three antibodies (data not shown). NE cells dissociated from TS3 embryos were positive for 8WG16, but showed very faint or no signal for either H14 or H5, indicating a transcriptionally repressed state (Table 2; Fig. 8). Similar signaling data are obtained for NE cells dissociated from all advanced stages including TS6, TS8, TS11, and TS14 (Table 2; Fig. 9). These results suggest a global transcriptional blockage for nutritional endodermal cell throughout development.

image

Figure 8. Transcriptional status of NE cells at TS3. NE cells dissociated from TS3 embryos were stained with 8WG16 (c) H14 (f), and H5 (i) antibodies and counter-stained with DAPI (b,e,h). (a,d,g) represent bright field (BF) images of the same fields. Nuclei of NE cells were positive for 8WG16. Nuclear staining for both transcription initiation (H14) and elongation (H5) were either very weak or not present. Scale bar represents 5 µm.

Download figure to PowerPoint

image

Figure 9. Transcriptional status of MZ and VC at TS14. NE cells dissociated from TS14 embryos were stained with 8WG16 (c) H14 (f) and H5 (i) and counter-stained with DAPI (b,e,h). (a,d,g) represents respective BF images. Nuclei of NE cells were positive for 8WG16. Nuclear staining for both transcription initiation (H14) and elongation (H5) were either very weak or not present. Scale bar represents 5 µm.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

We looked at the expression of Smad2, an essential transcription factor of the Nodal-signaling pathway. We showed that EcSmad2 RNA is maternally supplied, and the expression level remained high until the end of gastrulation. Moreover, during blastula and early gastrulation, the amount of EcSmad2 RNA was greater in VC cells compared to MZ cells. Not only the RNA, but also both native and the active forms of EcSmad2 protein were expressed in VC. Immunostaining of NF10 embryos, however, showed nuclear localization of PSmad2 in 100% of the MZ cells, but in only 12% of the VC cells. Nuclear Smad4 was detected in 100% of the MZ cells, but in 50% of the VC cells at NF10. Based on the Western and immunostaining results with EcSmad2 and EcSmad4, Nodal-signaling differs substantially between E. coqui VC and MZ cells, the latter being more similar to X. laevis early embryonic vegetal cells.

We also investigated the nature of VC with respect to its transcriptional activity, which led to a further understanding of why VC cells do not commit to differentiation. Immunostaining to detect the functional status of RNAPII revealed two important facts. First, we provided evidence for the occurrence of MBT by early gastrulation in E. coqui. The MZ cells of NF8 embryos are transcriptionally repressed and become active by NF10. Second, we showed that VC cells are transcriptionally silent, not only at gastrulation, but through development.

EcSmad2 RNA Is Maternally Contributed and Present in VC From Blastula and Gastrula

Work from our lab has provided evidence for VC cells expressing Nodal ligand genes like EcActivinB and EcDerriere (Karadge and Elinson, 2013) and the transcription factor EcSox17 (Buchholz et al., 2007) during early gastrulation. These results suggest that there could be some Nodal signaling in VC cells. In order to investigate this possibility, we examined the expression of EcSmad2. As the genome of E. coqui is not sequenced, our investigation of any gene begins with cloning the gene.

Our cloning results showed that both isoforms, Smad2 and Smad2Δexon3, are present in E. coqui. The qPCR analysis on the temporal expression of EcSmad2 RNA indicated a strong maternal contribution (Fig. 1A). During NF8 and NF10, EcSmad2 RNA amounts remained high, followed by a gradual decrease with development.

We compared the relative expression levels of EcSmad2 RNA between MZ and VC for NF8 and NF10, and between embryonic tissues and NE during post-gastrulation. We expected absence of EcSmad2 RNA in VC as found for EcVegT and EcVg1 RNA (Beckham et al., 2003). In contrast, there were higher relative amounts of EcSmad2 RNA in VC compared to MZ at both NF8 and NF10, and the EcSmad2 RNA level in VC was more at NF10 than at NF8 (Fig. 1B). There are two possible explanations for the higher expression of EcSmad2 RNA in NF10 VC. First, more EcSmad2 RNA may be made in VC during early gastrulation. This seems unlikely due to state of transcriptional repression present in the VC, as discussed later. Second, degradation of EcSmad2 RNA in MZ may be more than that in VC. Since the amount of EcSmad2 RNA in MZ was used as the standard and the amount in VC is a relative one, it is plausible that due to higher rate of degradation of the RNA in MZ, the relative value in VC is higher.

EcSmad2 Isoforms Are Differentially Expressed and Activated in VC Cells

Although the EcSmad2Δexon3 RNA was expressed in E. coqui oocytes, it was unclear whether the protein was maternally contributed (Fig. 2A). Following fertilization, the EcSmad2Δexon3 protein was observed in all stages (Fig. 2B). On the other hand, the RNA as well as the protein of the full-length EcSmad2 isoform was not only maternally contributed, but also expressed throughout early development (Fig. 1, 2). Western blots with anti-PSmad2 antibody, however, detected one strong band in most cases (Fig. 2). This raised the question whether the single band recognized by the anti-PSmad2 antibody (Fig. 2) was EcSmad2Δexon3 or EcSmad2.

The difference between activation of EcSmad2 versus EcSmad2Δexon3 would likely be significant. Another Smad, Smad3, binds DNA and can inhibit Smad 2 signaling (Dennler et al., 1998; Labbé et al., 1998; Faure et al., 2000; Attisano et al., 2001). These activities of Smad3 are due to the absence of exon3, a property shared with Smad2Δexon3. An inhibitory activity of EcSmad2Δexon3 could explain why VC cells do not commit to endoderm/mesoderm. The position of the size marker on the Western blot, however, suggested that the band observed with anti-PSmad2, was EcSmad2, not EcSmad2Δexon3.

Although Smad2 activation is present in both X. laevis and E. coqui, the consequences of this activation are clearly distinct in E. coqui. The vegetal half of the X. laevis embryo expressed VegT protein (Lustig et al., 1996; Stennard et al., 1996; Zhang and King, 1996; Horb and Thomsen, 1997), which in turn activated Sox17 expression. Vegetal cells were characterized by expression of Smad2, which committed these cells to become endoderm and mesoderm (Faure et al., 2000). In E. coqui, the VC cells contained Sox17 RNA but lacked EcVegT RNA (Beckham et al., 2003; Buchholz et al., 2007). Although expressions of Smad2 and PSmad2 were low in VC compared to MZ, it could be sufficient enough for Nodal-signaling to occur. Since activity of PSmad2 is based on its association with Smad4 (Lagna et al., 1996; Zhang and King, 1996), we next examined Smad4 expression.

Smad4 Isoforms Are Differentially Expressed in MZ versus VC From Gastrula

There are two isoforms of Smad4 proteins in X. laevis (Howell et al., 1999; LeSueus and Graff, 1999; Masuyama et al., 1999; Hill, 2001). The potential X. laevis ortholog of hSmad4 (human) is XSmad4α, whereas XSmad4β (a.k.a. Smad10) is a novel one with no homolog in other species (Howell et al., 1999; LeSueus and Graff, 1999; Masuyama et al., 1999). Based on the reported ORF sequences, the predicted molecular weights for XSmad4α and XSmad4β are 59.8 and 61.2 kDa, respectively (Masuyama et al., 1999; XSmad4α Genebank ID: AB022721.1 and XSmad4β Genebank ID: AB022722.1). Our Western blots (Fig. 3A,B) showed the presence of two isoforms of Smad4 in E. coqui around 60 kDa.

In E. coqui, although both Smad4 isoforms were expressed at NF8, NF10, and TS5, the expression of EcSmad4β was higher than that of EcSmad4α in MZ. In contrast, VC showed a lower expression of EcSmad4β, almost equal to that of EcSmad4α. Overall, expressions of both isoforms were much less in VC compared to MZ.

According to Howell et al. (1999), Smad4 isoforms show strikingly different temporal expression patterns in early X. laevis embryos. Adult tissues were also characterized by their expression in different ratios, which suggests their different and specific roles. First, the subcellular distributions for these two isoforms are different. Due to the lack of nuclear export signal (NES), Smad4β is exclusively nuclear. Therefore, Smad4α, which predominantly resides in cytoplasm, mediates Nodal-signaling via binding with active Smad2/3. In X. laevis, Smad4β expression is higher during blastula and early gastrulation (Howell et al., 1999; Hill, 2001). At mid-gastrulation, the ratio of the two isoforms changes, and expression of Smad4α predominates (Howell et al., 1999). Chang et al. (2006) showed that these two isoforms function differently during early frog embryogenesis, and Smad4β, not Smad4α, is the essential player in mesendoderm induction. According to some reports (Hill, 2001; Howell et al., 1999), until early gastrulation, association between PSmad2 and Smad4β can occur inside the nucleus. Complex formation inside the nucleus could also work for Smad4α, as it could be present on either side of the nuclear membrane (Hill, 2001). Also Liu et al. (1997) showed that it is not always necessary for R-Smads to bind Smad4 to accumulate inside the nucleus. These reports leave us with multiple possibilities as to which isoform is needed at a specific stage of development and where it is localized.

Our result of EcSmad4 temporal expression (Fig. 3A) indicated that the expression patterns for the individual isoforms did not switch after mid-gastrulation like in X. laevis, which could imply a disruption in Nodal-signaling. EcSmad4 spatial expression (Fig. 3B) was more informative. According to Howell et al. (1999) and Masuyama et al. (1999), in X. laevis, Smad4β mediates Nodal-signaling from MBT until mid-gastrulation. Western blot analysis of the E. coqui early gastrula indicated the predominance of EcSmad4β in MZ tissues, which will differentiate to mesendoderm. In contrast, VC cells had a very low level of EcSmad4β expression. Based on our results, we suggest that the low level of EcSmad4β expression is insufficient and prevented VC cells from differentiation.

During Gastrulation, Although 12% VC Cells Showed Nuclear PSmad2 Signal, 50% Were Positive for Smad4 Nuclear Accumulation

Although PSmad2 and Smad4 were present in VC at gastrulation, it was important to know their cellular location to determine whether Nodal-signaling was active in VC cells. Upon Nodal-stimulation, PSmad2 in combination with Smad4 should localize in the nucleus (Heldin et al., 1997; Massagué, 1998; Cobb and Goldsmith, 2000). Identification of active PSmad2 and its localization by means of immunostaining has been reported in X. laevis (Christen and Slack, 1999; Faure et al., 2000; Schohl and Fagotto, 2002). In the previous studies, immunolocalizations were obtained using whole mounts or immunofluorescence on cryosections (Fagotto and Gumbiner, 1994; Fagotto, 1999). These techniques were not ideal for studying large E. coqui embryos filled with yolk. We performed immunostaining on individual cells dissociated from the dissected MZ and VC from the E. coqui early gastrula.

Unlike MZ, only 12% of VC cells were positive for PSmad2 (Table 1; Fig. 4). The remaining cells showed either scattered cytoplasmic label or no label. Although there was nuclear accumulation of PSmad2, 12% of the VC cell population may not be high enough to carry out Nodal signaling for differentiation to endoderm. Furthermore, MZ and VC cells from NF10 were also stained with the anti-Smad4 antibody. Although all the MZ cells were positive for nuclear EcSmad4, but only 50% of VC cells were positive (Table 1; Fig. 5).

The 12% and 50% of VC cells, which showed nuclear PSmad2 and Smad4 signals respectively, provide several possibilities for the lack of endodermal development. First, it may be that PSmad2 and Smad4 translocate inside the nucleus of only small numbers of VC cells, and this is quantitatively not enough to undergo mesendoderm specification. Second, the overall expression of EcSmad4 may not be sufficient to carry on Nodal-signaling. Although, 50% of the VC cells showed nuclear Smad4 signal (Fig. 5), the level of Smad4 expression according to the Western blot was low. Moreover, when Smad4β is required during mesendoderm specification in X. laevis, its expression was negligible in E. coqui VC. Third, Smad4 can act as a transcriptional activator in a Smad2-independent manner. SMIF is a protein that translocates into the nucleus along with Smad4 without involving Smad2/3 (Bai et al., 2002). Smad4 can also form complex with Smad1 and translocate inside the nucleus (Lagna et al., 1996). Continuous nucleocytoplasmic shuttling of Smad4 on its own has also been suggested (Pierreus et al., 2000; Watanabe et al., 2000; Shi and Massagué, 2003). Hence, the nuclear signal for EcSmad4 in those 50% VC cells could be due to Smad4α, which is brought inside the nucleus along with some unidentified protein rather than Smad2/3. In addition to the differential expression of Nodal-signaling, all these possibilities suggest the presence of another regulatory block, further downstream, which ultimately determines the fate of VC cells.

Global Transcriptional Repression Prevented VC Cells From Differentiation

Although MZ cells showed transcription initiation and elongation at NF10, VC cells did not. Moreover, little or no sign of transcription initiation or elongation was detected in dissociated NE cells from post-gastrulation stages TS3, TS6, TS8, TS11, and TS14, a stage right before hatching (Table 2; Figs. 8 and 9).

A few NE cells from each stage showed a faint signal for transcription initiation or elongation. This may indicate that there are a few genes, which are actively transcribed in NE cells. The E. coqui thyroid hormone receptor gene (EcTRβ) is one such gene, since its RNA level was upregulated by TS14 (Singamsetty and Elinson, 2010). Similar global transcriptional repression events have been reported in X. laevis PGC formation (Venkatarama et al., 2010). Cells containing germ plasm are transcriptionally repressed while neighboring cells of endoderm lineage are transcriptionally active. Global transcriptional repression is also well known in Caenorhabditis elegans (Guven-Ozkan et al., 2008) and Drosophila melanogaster (Hanyu-Nakamura et al., 2008). In mouse, PGCs are formed after gastrulation, and they do not require germ plasm. Seki et al. (2007) showed that the migrating mouse PGCs were negative for P-Ser5 or P-Ser2, also suggesting no transcription initiation or elongation.

MBT in E. coqui Is Likely Established by NF10

During early embryogenesis, the mid-blastula transition (MBT) is a landmark event. Before MBT in X. laevis, early embryonic cells are transcriptionally silent, and survive using maternally supplied stockpiles of proteins and RNAs (Newport and Kirshner, 1982; Kisielewska and Blow, 2012). Although MBT is a complex mechanism, onset of zygotic/embryonic gene expression is considered as one strong parameter in identifying it (Gerhart, 1980; Heasman, 2006; Skirkanich et al., 2011; Shiokawa, 2012). In X. laevis, MBT occurs at NF8. In E. coqui, the timing of MBT was not defined, but the global transcriptional status of MZ cells provided strong evidence for MBT in E. coqui. At NF8, MZ cells showed no sign of transcriptional initiation or elongation (Fig. 7). At NF10, MZ cells were transcriptionally active as they showed both transcription initiation and elongation (Fig. 8).

Proposed Model of Differentiation in the MZ Versus VC Cells

Taking these findings into consideration, we propose a model for the differences in the state of differentiation of MZ and VC cells (Fig. 10). In E. coqui, Nodal-signaling could account for endoderm and mesoderm specification in MZ cells as occurs in X. laevis. VC cells, on the other hand, differed at two levels: first, by differential expression of Nodal-signaling components like Smad2 and Smad4, and second, by global transcriptional repression. Our model (Fig. 10) incorporates these two levels to explain the generation of NE from VC cells. In the first level, we propose two possibilities. According to scenario “A,” only the alternatively spliced variant, EcSmad2Δexon3 is activated and undergoes nuclear translocation before association with Smad4β, or after associating with Smad4α in the cytoplasm. Once inside the nucleus, EcSmad2Δexon3 then exerts an inhibitory effect, as speculated for X. laevis, and blocks the Nodal-pathway. In Scenario “B,” the full length EcSmad2 is activated and translocates into the nucleus in association with poorly expressed Smad4α (cytoplasmic) or Smad4β (nuclear). This event then leads to mesendoderm specification. According to our PSmad2 and Smad4 immunostaining data, only a small population of VC cells can undergo such a specification event, which does not occur because all the VC cells are then subjected to a second level of regulation. In the absence of P-Ser2 at CTD of RNAPII, a global transcriptional repression prevails in vegetal half blocking any effects of Nodal-signaling.

image

Figure 10. Two-level regulation of the fate of VC cells. At NF10, both Smad4 isoforms are expressed weakly (represented by dashed boxes). Smad4α resides in cytoplasm and Smad4β localizes exclusively in the nucleus. Level 1: Regulation of Nodal-signaling. There are two possibilities. (A) If the spliced variant, EcSmad2Δexon3 is the only form that is phosphorylated, its association with either of the Smad4 isoforms would lead to inhibition of Nodal-signaling. (B) If it is EcSmad2, which is phosphorylated, it could associate with Smad4 isoforms and may induce Nodal-signaling in a small subpopulation of VC cells. Level 2: Global transcriptional repression. Repression occurs via absence of P-Ser2 at CTD of RNAPII, leading to no elongation. If an active PSmad2-Smad4 complex were present inside the nucleus of VC cells, it would not turn on the transcription machinery. This leads to the development of NE instead of DE.

Download figure to PowerPoint

This repression mechanism can also explain why EcSox17, whose RNA was present in VC cells (Buchholz et al., 2007), did not cause endodermal differentiation. Therefore even if EcSox17 RNA were translated, this transcription factor would not activate downstream endodermal genes.

Our findings suggest the probable mechanism through which NE development occurs. The VC was not devoid of molecular signaling as suggested by the low levels of EcVegT RNA (Beckham et al., 2003) and by lack of mesoderm-inducing activity (Ninomiya et al., 2001). Rather, the VC contains some of the Nodal signal transducers, including PSmad2. Whether or not there was sufficient Nodal-signaling for mesendoderm specification, the large, yolky VC cells stayed transcriptionally repressed throughout development and served only as nutritional reserves.

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED

The authors acknowledge Dr. Gary Ten Eyck (Department of Pharmaceutical Sciences, University of Hawaii, Hilo HI) for collecting and sending adult E. coqui to us.

LITERATURE CITED

  1. Top of page
  2. ABSTRACT
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. ACKNOWLEDGEMENTS
  7. LITERATURE CITED