Dullard/Ctdnep1 Modulates WNT Signalling Activity for the Formation of Primordial Germ Cells in the Mouse Embryo

Dullard/Ctdnep1 is a member of the serine/threonine phosphatase family of the C-terminal domain of eukaryotic RNA polymerase II. Embryos lacking Dullard activity fail to form primordial germ cells (PGCs). In the mouse, the formation of PGCs is influenced by BMP4 and WNT3 activity. Although Dullard is reputed to negatively regulate BMP receptor function, in this study we found mutations in Dullard had no detectable effect on BMP4 and p-Smad activity. Furthermore Dullard mutations did not influence the dosage-dependent inductive effect of Bmp4 in PGC formation. However, Dullard may function as a positive regulator of WNT signalling. Combined loss of one copy each of Dullard and Wnt3 had a synergistic effect on the reduction of PGC numbers in the compound heterozygous embryo. In addition, loss of Dullard function was accompanied by down-regulation of WNT/β-catenin signalling activity and a reduction in the level of Dishevelled 2 (Dvl2). Therefore, Dullard may play a role in the fine-tuning of WNT signalling activity by modulating the expression of ligands/antagonists and the availability of Dvl2 protein during specification of the germ cell lineage.


Introduction
Specification of the germ cell lineage in the mouse begins with activation of Prdm1 (Blimp1) in a subset of epiblast cells in the proximal region of the pre-gastrulation mouse embryo [1]. Analysis of germ cell formation in mutant mouse embryos has revealed the critical role of bone morphogenetic protein (BMP) signalling in the induction of primordial germ cell (PGC) precursors. The PGC population is lost or greatly reduced in embryos that are deficient for BMP activity, which is caused by losses of Bmp4 and Bmp8b in the extraembryonic ectoderm, and Bmp2 and Alk2 (encoding a Type I BMP receptor) in the visceral endoderm, or Smad1, 4 and 5 (signal transduction factors) in the embryo [2][3][4][5][6][7][8][9][10][11][12]. In addition, WNT signalling plays a role in PGC formation. PGCs are absent in embryos lacking Wnt3 activity and WNT3A is required for priming epiblast cells to respond to induction by BMP4 to differentiate into PGCs [13].
Dullard (also known as C-terminal domain nuclear envelope phosphatase 1; Ctdnep1) was identified as a gene that is expressed in the pronephros and neural tissues of Xenopus laevis embryos [14]. Dullard encodes a protein serine/threonine phosphatase with a characteristic catalytic motif, DXDX(T/V). It is a member of an emerging family of phosphatases that dephosphorylate target substrates [15,16]. This family is also known as the phosphatase family of the C-terminal domain (CTD) of eukaryotic RNA polymerase II (polII), which dynamically regulates transcription by recruiting different factors to mRNA through its multiple phosphorylation activities [17]. Other SCP/transcription factor IIF-interacting CTD phosphatases that are closely related to Dullard, e.g. small CTD phosphatases (SCPs), play a role in modulation of the expression level of specific genes. Such phosphatases silence neuronal genes in non-neuronal cells to suppress inappropriate neuronal gene expression during cell fate decision. This regulatory activity is mediated through an interaction with the repressor element 1-silencing transcription factor/neuron-restrictive silencer factor (REST/NRSF) complex [18]. Bioinformatic analyses of human DULLARD has revealed that the protein contains two potential membrane-spanning regions in the N-terminal, which may direct the localization of DULLARD to the nuclear envelope, where it dephosphorylates a nuclear membrane-associated phosphatidic acid phosphatase in human cell line cells [16]. Immunostaining further revealed punctuate localization of DULLARD in the nucleus and cytoplasm [16], suggesting that Dullard might have other target substrates that are not only associated with the nuclear envelope. Indeed, Dullard forms a protein complex with the BMP Type II receptor to promote its degradation [15], and suppression of BMP signalling may underlie its induction of neural tissue formation in the Xenopus embryo [15]. Dullard also interacts with BMP Type I receptors to repress their BMP-dependent phosphorylation. Therefore, Dullard might regulate the level of BMP signalling via its phosphatase activity to dephosphorylate BMP receptors, leading to their degradation [15]. Given the essential function of BMP signalling in the induction of PGCs, it is anticipated that factors modulating BMP signalling activity, such as Dullard, will affect germ cell specification.
In this study, we have demonstrated a critical requirement for Dullard in the formation of PGCs in the mouse embryo. However, results of the genetic study showed that loss of Dullard has no discernible effect on the expression of Bmp4 and Smads, and does not affect the dose-dependent inductive activity of BMP4 on PGC formation. Instead, Dullard functions as an agonist that modulates WNT signalling activity to facilitate the formation of PGC precursors.

Ethics Statement
All animal experiments were approved by the Animal Care and Use Committee of Kumamoto University (#A24-110). Protocols were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

Generation of Dullard Mutant and Compound Mutant Mice
Two types of embryonic stem (ES) cells harbouring modified Dullard alleles (Dullard +/2 and Dullard +/LacZ ) were generated by gene targeting (Fig. S1). Mutant mouse embryos generated from Dullard +/2 or Dullard +/LacZ ES cells displayed similar phenotypes, and were therefore designated collectively as Dullard 2/2 embryos

Expression Vector Construction and Subcellular Localization of the Dullard Fusion Protein
An expression vector containing Dullard-enhanced green fluorescent protein (EGFP; fused to the C-terminal end of Dullard) under the control of the CMV promoter (pEGFP-N1, Clontech Laboratories, Inc., Mountain View, CA, USA) was transfected into NIH-3T3 cells. Then, the localization of Dullard-EGFP signals in the endoplasmic reticulum (ER) was visualized by anticalreticulin immunostaining of the ER (Stressgen Bioreagents, Victoria, BC, Canada). Similar experiments were performed using a Dullard-FLAG fusion protein expression vector. Confocal imaging was carried out under a Leica SP2 confocal scanning system (Leica Camera AG, Wetzlar, Germany).

RNA Extraction and Real-time Quantitative (Q)-PCR
Total RNA was extracted using RNeasy Plus Micro RNA extraction kits (Qiagen, Hilden, Germany). Reverse transcription was carried out with 0.2 mg RNA using a ReverTra Ace qPCR RT Master mix with gDNA Remover (Toyobo, Osaka, Japan). For Q-PCR, a Thermal Cycler Dice TP800 (Takara Bio Inc., Shiga, Japan) and Thunderbird SYBR-green qPCR mix (Toyobo) were used. The standard curve method was used to determine the relative quantities of mRNA, and expression levels were normalized to those of Gapdh (internal control). Two to three independent samples were analyzed as technical duplicates. Amplification of was performed with primer sets described in the related publications: Brachyury, Ifitm3, Dppa3 and Arbp [29], Wnt3 [30], Axin2 [31] and Goosecoid [32]. Primers for Dullard, Bmp4, Lef1, Dkk1, Id1, Msx2, Pou5f1 and Gapdh were purchased from Takara Bio Inc. Primers for Ifitm1 amplification were: 39-acatgcctgagatctccacg-59 and 39-atggtgaagaacagggagc-59.

Microarray Analysis
Microarray analysis was performed according to the Agilent Expression Array Analysis manual (Agilent Technologies, Palo Alto, CA, USA). cDNA was amplified from 20 ng total RNA using Ovation Pico WTA system V2 (NuGEN Technologies, San Carlos, CA, USA). Cy3-labelled cDNA was hybridised to a Whole Mouse Genome 4644K v2 array (Agilent Technologies). Data normalisation and analyses were performed using the Agilent Feature Extraction program (Agilent Technologies). When the gene had multiple signal data, the average was indicated as a foldchange. The data have been deposited in the Gene Expression Omnibus database under the accession number 16596277. Construction of GST-Dullard expression vectors followed procedures described elsewhere [15,16]. The GST-pull-down assay was performed in HEK293 cells transfected with Myctagged Dvl1, Dvl2 and Dvl3 expression vectors [33]. The results were collated from two independent experiments.

Statistical Analyses
Statistical analyses were performed using the Student's t-test, with p-values of ,0.05 indicating statistically significant differences between datasets.
In NIH-3T3 cells that expressed Dullard-EGFP or Dullard-FLAG fusion protein, EGFP and FLAG signals were localized preferentially to the perinuclear domain and calreticulin-positive ER ( Fig. 1F-G99; FLAG data not shown). These results were consistent with the previously reported localization of endogenous Dullard and the transfected Dullard fusion proteins in the perinuclear domain of mammalian and yeast cells, respectively [16]. Dullard was also distributed in a punctuate pattern in the cytoplasm and nucleus (Fig. 1F, F9 and data not shown; [16]). Microarray analysis of E7.5 Dullard +/2 and Dullard 2/2 embryos revealed further defects in mesoderm development (Table S1). Anterior mesoderm markers, Gsc and Lefty1, were up-regulated in Dullard 2/2 embryos (6.9-fold and 4.5-fold, respectively), while nascent mesoderm markers Brachyury and Fgf4 were downregulated (2.3-fold and 3.1-fold, respectively). Nkx1-2 (Sax1), which is expressed in the primitive streak, was down-regulated in Dullard 2/2 embryos (2.5-fold). Increased Gsc expression and reduced Brachyury and Nkx1-2 expression were confirmed by Q-PCR (Fig. S4). Dullard activity is therefore not essential for embryonic patterning but is required for maintenance of the primitive streak and differentiation of the extraembryonic tissues.
Dullard 2/2 AP-positive cells were unable to transit from the mesoderm to the endoderm as indicated by only one Dullard 2/2 embryo showing three AP-positive cells in the endoderm. In contrast, many AP-positive cells were localized in the endoderm of Dullard +/2 embryos (Fig. 2E9). Ifitm1, which was expressed in PGC precursors and the posterior mesoderm in wild-type embryos ( Fig. 2O-O99, [39]), was not expressed in Dullard 2/2 embryos (Fig. 2P-P99). Our previous study [19,42] has shown that proper expression of Ifitm1 is essential for navigating PGCs from the mesoderm to endoderm. Disruption of Ifitm1 expression following the loss of Dullard may therefore have impeded the translocation of PGCs.

Dullard does not Affect BMP4 Activity in PGC Formation
Considering the putative role of Dullard in modulation of BMP signalling activity, we examined the effect of loss of Dullard on BMP4 signalling activity. In the Dullard 2/2 embryo, Bmp4 was expressed at the appropriate level (Fig. S2A) in extraembryonic tissue (Fig. S3R-T), and the expression of BMP signal transducers, Smad1, Smad4 and Smad5, was unchanged (Table S1). p-Smads (the active form of Smad1, 5 and 8) were expressed appropriately in the proximal epiblast of E6.75 Dullard 2/2 and Dullard +/2 embryos (Fig. 3A-C). Consistent with the normal expression of Bmp4 and p- Smads, Q-PCR and in situ hybridization analyses showed appropriate expression of BMP downstream target genes, Id1 and Msx2 in E7.5 and E8.25 Dullard 2/2 embryos (Fig. S3U-W,  S4, and data not shown). In the E7.5 Bmp4 2/2 embryo, Dullard expression was unchanged (Fig. S2). Therefore, loss of Dullard does not appear to affect BMP activity in embryos at gastrulation and early organogenesis.

Loss of Dullard Affects WNT Signalling Activity
Canonical WNT signalling activity, acting in concert with BMP, has been shown to enhance the generation of the germ cell lineage from epiblast cells and ES cells in vitro [13,43]. Further analysis of the expression profiling data of the E7.5 Dullard 2/2 embryo revealed a potential role of Dullard in modulation of WNT signalling activity (Table S1). Q-PCR analyses confirmed significant down-regulation of several WNT downstream genes and WNT targets including Brachyury, a WNT3 downstream target gene [44]; Ifitm1, a WNT/b-catenin-dependent downstream gene [45,46], and Nkx1-2, which responds to WNT/b-catenin signalling and activates Brachyury expression through the suppression of Tcf3 (in P19 mouse embryonal carcinoma cells; [47]) in the E7.5 Dullard 2/2 embryo (Fig. S4). The expression of other WNT/bcatenin-dependent targets, Lef1 and Axin2/conductin, was significantly reduced as indicated by Q-PCR (Fig. S4). Reduced expression of WNT-response genes was also found by in situ hybridization and whole mount immunostaining (Brachyury, S3E-F9; Ifitm1, Fig. 2O-P99; Axin2, Fig. 4A-B99). Expression of active b-catenin was reduced in the posterior epiblast of the E7.5 Dullard 2/2 embryo, where Ifitm3-positive PGCs were found (Fig. 4C-F999). As shown by western blotting, Dullard 2/2 embryos expressed a similar amount of b-catenin as that in E7.75 control embryos, but the amount of active b-catenin was reduced in mutants (n = 2, Fig. 4G). Taken together, these findings indicate that loss of Dullard is associated with a reduction in WNT/bcatenin signalling activity.
To test whether Dullard mutation influences the effect of loss of Wnt3 on PGC formation, we studied the effect of genetic interactions in Wnt3; Dullard compound mutant embryos using WNT-response gene expression and PGC formation as experimental indicators. Reduced Dullard activity had a synergistic effect on PGC formation on a Wnt3 +/2 background. Dullard +/2 ; Wnt3 +/2 embryos contained about half of the number of Dppa3and AP-positive PGCs as that in wild-type embryos. In contrast to the Dullard 2/2 embryo, the compound heterozygous mutant embryo did not show any morphological defects ( Fig. 5K-M9, Table 1). These findings suggest that the effect on PGC formation may be unrelated to the developmental abnormalities associated with loss of Dullard, but may be the consequence of altered WNT activity. In E7.5 Dullard +/2 ; Wnt3 +/2 embryos, Dullard and Wnt3 were expressed at 70% and 35% of the wild-type level, respectively (Fig. S2A). E7.5 Wnt3 +/2 and Dullard +/2 embryos showed similar expression levels of WNT/b-catenin downstream genes, Brachyury, Ifitm1, Axin2 and Lef1 (Fig. S2B), whereas Dullard +/2 ; Wnt3 +/2 embryos displayed a significantly lower level of downstream gene activity than that in Dullard +/2 embryos (except for Ifitm1), but was higher than that in Dullard 2/2 embryos (Fig. 6).

Dullard Interacts with Dishevelled
Results of the genetic study suggested that Dullard may act as a positive regulator of WNT signalling activity. To test whether Dullard acts at another juncture in the WNT cascade, a GST-pull down assay was performed in HEK293 cells to identify Dullardinteracting proteins that are involved in signal transduction. A GST-Dullard recombinant protein (lacking the first 30 amino acids; [16]) interacted with Myc-tagged Dvl1, Dvl2 and Dvl3 (Fig. 7A), which are members of the intracellular WNT signal transducer family. A GST-Dullard protein harbouring the D69E mutation in the catalytic domain, which abolishes its phosphatase activity [15,16], also displayed a binding activity with Myc-tagged Dvl2 similar to that of the control counterpart (Fig. 7B). Phosphatase activity of Dullard is therefore not required for the interaction with Dvl2.
Less Dvl2 protein was present in E7.75 Dullard 2/2 embryos compared with Dullard +/2 embryos (n = 2, Fig. 7C), while expression of Dvl1-3 mRNA was unchanged (Table S1). There- fore, loss of Dullard, which may act as a Dvl-interacting protein, may reduce the availability of Dvl2 protein and negatively affect WNT signal transduction activity (Fig. 8A) in conjunction with its potential role in balancing agonist and antagonist activities (Fig. 8B).

Discussion
In this study, we have shown that loss of Dullard function in the mouse embryo phenocopies the PGC deficiency associated with the loss of BMP4 and WNT3 signalling activities. Dullard, which encodes a protein with phosphatase activity, is reputed to negatively regulate BMP signalling by dephosphorylation that inhibits the function of BMP receptors. Therefore, loss of Dullard is anticipated to result in enhanced BMP receptor activity, which may counteract the effect of absent or reduced BMP signalling. However, the results of our genetic study showed that Dullard has neither a synergistic nor an opposing effect on BMP4 activity in PGC formation. However, a possible role of Dullard in BMP signalling for PGC formation cannot be completely ruled out, and Dullard functions may intersect with BMP signalling in other developmental processes.
Similar to the effect of loss of Wnt3 on PGC formation, as shown in our study and a previous study [13], PGCs are absent in Dullardnull mutant embryos, even though BMP signalling activity is unabated and the mutant cells remain responsive to BMP4 induction. It has been proposed that WNT3 may influence the competency of epiblast cells at E5.5 to respond to BMP4 induction of Prdm1-positive germ cell progenitors [13]. Although WNT alone is insufficient to induce the differentiation of epiblast cells into PGCs, induction of PGCs by BMP can be enhanced by WNT3A [13]. A combination of WNT and BMP signalling activity also induces the generation of PGCs from ES cells and induced pluripotent stem cells [43]. These findings suggest that WNT signalling acts to render epiblast cells competent to respond to BMP induction. The results of this study show that Dullard may function as an agonist of WNT activity. Loss of Dullard activity down-regulates WNT-response gene expression and the level of active b-catenin, which may disrupt the competency of epiblast cells to generate PGCs.
The induction of PGCs by WNT and BMP appears to require an intermediate step of differentiation into epiblast-like cells [43]. In vivo, the generation of Prdm1-positive cells from epiblast cells may continue through E6.5 to E7.25 [50]. In addition, distal epiblast cells in E6.5 embryos at the early primitive streak stage, which would not have experienced strong WNT or BMP signals, can give rise to the AP-positive PGCs when transplanted heterotopically into the proximal region [51]. This observation suggests that epiblast cells are still responsive to WNT and BMP signalling well after the initial phase of induction of PGCs. Therefore, WNT signalling activity may have a continuous influence on PGC formation beyond E5.5 to after the onset of gastrulation.
The action of Dullard on components of the WNT signalling cascade may be related to regulation of signalling activity through intricate positive and negative feedback regulatory loops [52][53][54][55][56]. Loss of Dullard resulted in reduced expression of some WNT ligands (Wnt3A, Wnt8b, and Wnt10a) and antagonists (Dkk1, Sfrp1, and Sfrp5), but paradoxically led to up-regulation of Wnt3, the functionally dominant ligand that influences early post-implantation embryonic development [26,48]. The gain in WNT3 activity in the Dullard 2/2 embryo may be nullified by the more robust activity of the antagonist, Dkk1, which is a direct transcriptional target of WNT/b-catenin signals and negatively modulates signalling activity through its interaction with the receptor complex. The activation of this regulatory feedback loop may offset the effect of Wnt3 up-regulation, thereby recapitulating a reduced overall WNT activity, such as that under a partial ''Wnt3null'' (hypomorphic) condition. In this regard, reduction of the gene dosage of Wnt3 in the Dullard +/2 embryo may result in more balanced Wnt3 and Dkk1 activities, which still results in an overall decrease in signalling activity, as revealed by the decreased activation of WNT downstream genes (intermediate), but allows partial restoration of PGC formation (Fig. 8B). This proposed scenario highlights that the fine-tuning of WNT/b -catenin signalling activity by Dullard is crucial for PGC formation. . Dkk1 antagonizes WNT3 signalling activity by interacting with the Lrp5/6 co-receptor upstream of the signal transducing pathway involving Dishevelled. Dullard may interact with a repressor molecule that negatively regulates the expression of Wnt3, which signals through the b-catenin/Tcf dependent cascade (via a multi-step process: arrows) to activate Dkk1 expression. (B) Profiles of Wnt3 (agonist) and Dkk1 (antagonist) expression as well as overall signalling activity for PGC formation. The fine-tuning of WNT/b-catenin signalling activity by Dullard is crucial for PGC formation. Loss of Dullard function produced high Wnt3 activity (175% of the heterozygous level) that may be effectively nullified by higher Dkk1 activity (267%). This occurrence would lead to the overall reduced (low) WNT signalling activity that was accompanied by loss of PGCs. In the Dullard +/2 ; Wnt3 +/ 2 embryo, moderate Wnt3 activity (52% of the heterozygous level: low) in the presence of lower (43%) Dkk1 activity may establish an overall intermediate level of WNT activity that permits the formation of a smaller number of PGCs (partly restored PGC numbers). doi:10.1371/journal.pone.0057428.g008 It is possible that Dullard may regulate WNT signal-related gene expression (Fig. 8A) in cooperation with other transcriptional partner(s). Other Dullard-related phosphatases that contain the DXDX(T/V) amino acid signature of the SCP/FCP family, such as the eyes absent homolog protein, can regulate gene expression through the formation of complexes with sine oculis-related homeobox protein homolog (Six) DNA binding proteins in developing organs including the eye, ear, muscle and kidney [57]. In yeast cells, nuclear envelope morphology protein 1 (Nem1) is required for protein complex formation with Sporulationspecific protein 7 (Spo7p) to dephosphorylate the target substrate, nuclear membrane-associated phosphatidic acid Smp2p [58]. Human DULLARD is able to rescue the aberrant nuclear envelope phenotype of Nem1-deficient yeast cells by dephosphorylation of Smp2p [16]. Recombinant human DULLARD is unable to dephosphorylate Lipin, a predicted mammalian ortholog of Smp2p, whereas DULLARD dephosphorylates mouse Lipin 1b only when it is over-expressed in BHK cells, but not in HeLa or HEK293A cells, suggesting that the mammalian counterpart of yeast Spo7p might be expressed in a cell type-specific manner [16]. In the Dullard 2/2 mouse embryo, the WNT signalling activity, as revealed by active b-catenin localization, was downregulated most robustly in the posterior region of the embryo, where PGCs and their precursors are localized (Fig. 4C-F999). This result raises the possibility that the action of Dullard depends on its interaction with stage/site/cell-type specific partner protein(s) for targeting to the appropriate substrate for dephosphorylation.
The action of Dullard on other components of the WNT signalling cascade was revealed by the physical interaction between Dullard and Dvl proteins of the signal transduction cascade (Fig. 8A). The precise effect of the Dullard-Dvl interaction is presently unknown, but the amount of Dvl2 protein was reduced in Dullard 2/2 embryos. Dvl protein turnover is regulated by proteasomal and lysosomal degradation. Several Dvl-interacting proteins have been reported to facilitate Dvl poly-ubiquitination [59]. It is possible that binding to Dullard stabilizes the Dvl2 protein by inhibiting the binding of other Dvl-interacting protein(s) that facilitate ubiquitination and degradation. Therefore, the interaction with Dullard may ensure the availability of Dvl proteins for the transduction of WNT signals to effectors.