Loss of Function of Mouse Pax-Interacting Protein 1-Associated Glutamate Rich Protein 1a (Pagr1a) Leads to Reduced Bmp2 Expression and Defects in Chorion and Amnion Development

Abstract

Background

Human PAX-Interacting Protein 1 (PAXIP1)-associated glutamate rich protein 1 (PAGR1, also known as PA1) originally was discovered as part of a complex containing PAXIP1 and histone H3K4 methyltransferases MLL3 and MLL4, suggesting a role in epigenetic gene regulation. Further in vitro studies suggested additional functions in DNA damage repair and transcription. However, in vivo analysis of PAGR1 function has been lacking.

Results

Here we show that expression of the cognate mouse gene Pagr1a is found predominately in the extraembryonic and chorionic ectoderm from pregastrulation stages and is up-regulated within the embryo proper after gastrulation. Embryos with a germ line deletion of Pagr1a establish the anterior–posterior axis, and show normal neuroectodermal, mesodermal, and endodermal patterning, but fail to develop beyond the four- to five-somite stage or to undergo axial rotation. Pagr1a^−/− embryos also show abnormal development of extraembryonic tissues with defects seen in the amnion, chorion and visceral yolk sac. At the molecular level, Pagr1a^−/− embryos have reduced expression of BMP2, a known regulator of extraembryonic development.

Conclusions

Loss of mouse Pagr1a function leads to defective extraembryonic development, likely due at least in part to altered BMP signaling, contributing to developmental arrest.

Introduction

The mammalian embryo depends on extraembryonic tissues, including the placenta and the visceral yolk sac (VYS) and amnion, for intrauterine growth and survival. The amnion is the extraembryonic membrane closest to the embryo, defining the amniotic cavity and primarily playing a protective role. The VYS mediates the absorption of oxygen and nutrients from the maternal environment and their transfer to the embryo at early stages of development. The chorioallantoic placenta then assumes this function as the embryo grows larger and develops a circulatory system. The VYS is also the initial site of hematopoiesis and blood vessel formation.

Development of the extraembryonic membranes begins at gastrulation, accompanying the dynamic changes taking place in the embryonic region that result in germ layer formation and patterning. Before gastrulation, the mouse embryo is cylindrical in shape with embryonic and extraembryonic regions arranged along a proximal–distal axis with respect to the mesometrium. An inner epithelial layer composed of embryonic ectoderm (or epiblast) distally and extraembryonic ectoderm (ExE) proximally is arranged around a central lumen known as the proamniotic cavity and surrounded by an outer layer of visceral endoderm (VE). At embryonic day (E) 6.5, mesoderm formation commences at the embryonic/extraembryonic junction between the inner ectoderm and outer VE, and then expands distally to form the primitive streak demarcating the posterior region of the embryo proper. In the extraembryonic region, mesodermal cells from the proximal part of the primitive streak push both ExE and epiblast derived cells into the proamniotic cavity, thereby creating the amniochorionic fold (ACF) and the nascent exocoelom (Pereira et al., 2011). The continued expansion of the ACF eventually leads to fusion with the anterior ectoderm: the ExE derived component of the ACF fuses with anterior ExE to form the chorion and create the ectoplacental cavity; the epiblast derived ACF cells fuse with anterior proximal epiblast, giving rise to the amnion and closing off the amniotic cavity. These events occur at the anterior midline at a position called the anterior separation point (ASP), because this is where the chorionic and amniotic components of the ACF lose their connection through intercalation of mesoderm (Ellington, 1987; Pereira et al., 2011). After separation, the exocoelom expands as the chorion moves away from the amnion. This enlarges the VYS and eventually eliminates the ectoplacental cavity. A secondary growth of mesoderm from the embryonic/extraembryonic boundary forms the allantois, which grows into the exocoelom ultimately meeting and fusing with the chorion to form the chorioallantoic placenta.

Gene knockout studies in mice have shown an important role for bone morphogenetic protein (BMP) signaling in extraembryonic development (Kishigami and Mishina, 2005). BMPs are secreted ligands belonging to the transforming growth factor beta (TGF-β) superfamily that activate the intracellular effector proteins Smad1, Smad5, and Smad8 to regulate target gene expression (Massague et al., 2005). Smad1 knockout embryos show marked defects in proliferation and morphogenesis of extraembryonic tissue including over-proliferation of the chorion within the exocoelomic cavity and a shortened allantois that fails to establish connection with the placenta (Lechleider et al., 2001; Tremblay et al., 2001). Smad5 mutants have defects in the anterior amnion, which is thickened and displays characteristics of the primitive streak (Chang et al., 1999; Bosman et al., 2006; Pereira et al., 2012). Embryos lacking BMP4 show defects in gastrulation and in the allantois due to reduced mesoderm formation (Winnier et al., 1995). The initial study on Bmp2 mutant embryos showed that the ACF fails to fuse anteriorly, which was attributed to deficiencies in extraembryonic mesoderm, resulting in an open proamniotic canal (Zhang and Bradley, 1996). Subsequent analyses have pointed to a more complex, heterogeneous phenotype and provided evidence for a role in the morphogenic events that lead to enclosure of the embryo in the extraembryonic membranes and proper positioning of head and heart (Goldman et al., 2009; Madabhushi and Lacy, 2011).

Human PAX-Interacting Protein 1-associated glutamate rich protein 1 (PAGR1; also known as PA1 and GAS) was identified in a complex with PAX-Interacting Protein 1 (PAXIP1; also known as PTIP) by co-immunoprecipitation and mass spectrometry (Cho et al., 2007). Both proteins were found to be associated with MLL3/MLL4-containing histone H3K4 methyltransferase complexes (Cho et al., 2007), suggesting a role in epigenetic gene regulation. In addition, PAGR1 and PAXIP1 form a separate complex independent of MLL3/MLL4 that has a role in DNA damage repair (Gong et al., 2009). PAGR1 also acts as transcriptional co-regulator of the estrogen and glucocorticoid receptors (Liang et al., 2009; Zhang et al., 2013), a function likely independent of its association with PAXIP1 or MLL3/MLL4. Thus, PAGR1 may have multiple roles in gene regulation. Here we show that the cognate mouse gene, Pagr1a, is expressed first within the ExE and chorion and later expands to the entire embryo. Analysis of a targeted germ line deletion reveals that although Pagr1a^−/− embryos properly specify all embryonic and extraembryonic cell lineages, profound growth defects are detected as early as E8.0. The most conspicuous abnormalities are seen in extraembryonic tissues, including the accumulation of a mass of chorionic tissue at the anterior midline, within the amniotic cavity. Our analysis suggests that Pagr1a mutant embryos undergo abnormal ACF formation and separation at the ASP resulting in defects in both chorion and amnion. Importantly, Pagr1a^−/− embryos show reduced expression of Bmp2, suggesting that Pagr1a normally regulates Bmp2 gene expression and that the Pagr1a mutant phenotype is due, at least in part, to reduced BMP2 signaling.

Results and Discussion

Pagr1a Expression is Highest in the Extraembryonic and Chorionic Ectoderm

To provide insight into the in vivo function of mouse Pagr1a, we used whole-mount in situ hybridization (WMISH) with a full-length cDNA probe to analyze the expression pattern of Pagr1a mRNA in pre- and postgastrulation embryos. Expression was first detected at E5.0 to E6.0 in the extraembryonic region (Fig. 1A, arrowheads). In the prestreak (PS) to early primitive streak stage (ES), expression was readily detected in the ExE (Fig. 1B, left panel; arrowhead). The absence of any signal in control embryos following WMISH with sense strand probe (Fig. 1B, right panel) suggested that a relatively low level of Pagr1a expression may also be present within the epiblast at this stage (Fig. 1B, left panel; arrow), but not in the VE. In E7.5 embryos at the early allantoic bud stage (EB), Pagr1a mRNA was detected strongly within the chorion (Fig. 1C,D; arrowheads). Some EB-stage embryos also showed comparatively higher levels within the embryonic region (Fig. 1D, arrow), indicating an up-regulation of expression within the embryo proper occurs around this stage. By the late head fold stage (LHF) and into early somite stages, all embryos showed expression in both the embryonic and extraembryonic regions (Fig. 1E,F). Sectioning showed that Pagr1a mRNA is present in all fetal cell lineages and also within the allantois, but levels are strongest in the chorion even at these stages.

Fig. 1.

Expression pattern and targeted mutation of the mouse Pagr1a locus. A: WMISH analysis at E5.0 to E6.0 showing expression of Pagr1a mRNA specifically in the extraembryonic region (arrowheads). B: The left panel shows WMISH analysis of a PS/ES-stage embryo with Pagr1a expression in the ExE (arrowhead). The right panel shows analysis of a PS/ES-stage embryo using sense strand probe. The lack of any signal suggests that the faint staining seen in the epiblast of the embryo to the left (arrow) represents low-level Pagr1a expression. C: WMISH analysis and mid-sagittal section of an EB-stage embryo showing strong Pagr1a expression in chorionic ectoderm (arrowheads). The chorion is still in contact with the amnion (brackets). D: WMISH and mid-sagittal section of another EB-stage embryo, showing strong expression of Pagr1a in the chorion (arrowheads) and also in the embryonic region (arrows). E: WMISH and mid-sagittal section of a LHF-stage embryo. Pagr1a is strongest in the chorion (arrowheads) but is also seen throughout the embryonic region and in the allantois (arrows). F: WMISH analysis of a four-somite–stage (4 som) embryo with ubiquitous Pagr1a expression. G: Schematic representation of the mouse Pagr1a genomic locus and targeted deleted allele. The Pagr1a gene contains three exons, represented by the three black rectangles. The deleted allele lacks the region containing exons 1 and 2. H: PCR genotype analysis of genomic DNA prepared from WT (+/+) heterozygous (+/−) and homozygous mutants (−/−). The arrowheads in (G) indicate location of primers used for genotyping the WT and deleted alleles. bp, base pairs. I: WMISH analysis of Pagr1a expression in a LB/EHF-stage WT embryo and Pagr1a^−/− mutant (Mut) littermate. No expression is seen in the mutant suggesting the deletion results in complete loss of function. The arrow points to the abnormal amnion and the asterisk marks tissue damage arising during dissection.

Pagr1a Loss of Function Mutation is Embryonic Lethal

To understand the consequences of loss of Pagr1a function, mice bearing a targeted deletion of the Pagr1a gene (Fig. 1G; Y-W.C., J-E.L. and K.G., manuscript in preparation) were analyzed. Pagr1a^+/− heterozygous mice, identified by polymerase chain reaction (PCR) genotyping (Fig. 1H), were found to be normal and fertile. Heterozygous animals were then mated to generate Pagr1a^−/− homozygotes. At weaning, no homozygotes were found out of a total of 135 offspring (Table 1), suggesting the mutation is embryonic lethal. We then examined embryos at different stages of development to determine the timing of embryonic death (Table 1). At the PS stage, 5 of 24 embryos examined were genotyped as Pagr1a^−/− (Fig. 1H) but none showed any abnormalities. At the ES to late streak (LS) stage, a total of 48 embryos were analyzed, of which 11 were Pagr1a homozygous mutants. None of these showed any phenotype. However, at the EB to late bud (LB) stages, 3 of a total of 24 Pagr1a^−/− embryos showed clear defects in the amnion and ACF (Fig. 1I, right panel, arrow). At the early head fold (EHF) to LHF stages, five of seven homozygotes showed ACF defects. Of interest, the number of mutants was higher at the EB/LB stages than expected from Mendelian ratios, and lower at the EHF/LHF stages. This suggests that mutants undergo developmental delay starting around the transition from allantoic bud to head fold stages. All E8.5 and E9.5 Pagr1a^−/− embryos were developmentally delayed and showed obvious defects (see below), whereas no homozygotes were found at E10.5 and later.

TABLE 1.

Phenotypic analysis of Pagr1a mutant embryos

Stage	No. examined	No. Pagr1a−/−	No. abnormal
PS	24	5 (21%)	0
ES/LS	48	11 (22%)	0
EB/LB	66	24 (37%)	3a
EHF/LHF	59	7 (11%)	5b
E8.5	111	27 (24%)	27c
E9.5	64	17 (27%)	17d
E10.5	23	0	–
E13.5	85	0	–
Pups	135	0	–

^aAmnion ruffled.

^bACF not resolved into amnion and chorion.

^c4 mutants still at LHF, 23 at 0–4 somites; all with ACF defects. WT at 3–11 somites.

^dAll mutants<5 somites, none turned; all WT with >13 somites, all turned.

Analysis using WMISH revealed a lack of expression of Pagr1a in homozygous mutant embryos (Fig. 1I, right panel). The absence of detectable mRNA corresponding to exon 3, which is retained in the deleted allele, may be due either to lack of transcription initiation or to mRNA instability, and suggests that the mutation is a complete null.

Pagr1a Mutants Display Multiple Embryonic and Extraembryonic Defects

The abnormal phenotype detected in E8.5 and E9.5 Pagr1a mutant embryos included an overall reduction in size compared with wild-type (WT) embryos, failure to undergo axial rotation (embryo turning), limited anterior development and arrest at the four- to five-somite stage, as judged by WMISH analysis using the somite marker Uncx (Fig. 2A,B). Striking morphological abnormalities in the extraembryonic region were also apparent. In the VYS, ring-like structures were seen, giving the membrane the appearance of being cratered (Fig. 2C, arrow and inset; Fig. 2H, arrowheads). The allantois was considerably shorter in mutants, as judged by WMISH for Foxf1 (Fig. 2D, brackets). Most notably, tissue aggregations were seen at E8.5 above the developing cranial neural folds within the amniotic cavity (Fig. 2C,E; arrowheads) and exocoelom (Fig. 2E,G). Of interest, Paxip1 mutant mouse embryos also show abnormal aggregates within the amniotic cavity (Cho et al., 2003), suggesting the physical interaction of PAGR1 and PAXIP1 proteins may reflect a functional interaction in extraembryonic development. Histological examination of the tissue aggregates suggested that they consist of chorionic ectoderm (Fig. 2G). Histology also revealed a thickening of the posterior amnion in mutants (Fig. 2E, arrowhead). Examination of the VYS revealed the absence of the apical vacuoles normally seen in the endoderm layer (Fig. 2F,F′). At the LB/EHF stages, structural abnormalities in the endoderm of the VYS already were evident with the apical vacuoles appearing smaller and fewer in number (Figs. 4E,E′), suggesting the transport functions of this tissue may be impaired from an early stage. Further analysis using the TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) assay (Fig. 2I) revealed significant cell death at E8.5 in the Pagr1a^−/− VYS endodermal layer (Fig. 2J), as well as in allantois (Fig. 2K) and neuroectoderm (Fig. 2L). Although the cell death seen at these late stages may not be a direct consequence of loss of Pagr1a function, it may contribute to the limited development of these tissues in the mutant. The cell death in the VYS may also further compromise VYS-mediated gas and nutrient exchange at E8.5, thereby precipitating embryonic demise.

Fig. 2.

Pagr1a^−/− embryos exhibit severe growth defects and extraembryonic abnormalities. A: Lateral views of an E8.5 WT embryo (right) and mutant littermate (left) showing significant growth retardation. Anterior is to the left. B: Lateral views of an E9.5 WT embryo (right) and mutant littermate (left) after WMISH with a probe for the somite marker Uncx, the midline marker Foxa2 and the forebrain marker Six3. The WT littermate has 13 to 14 somites, whereas only 3 to 4 somites are detected in the mutant (curved lines). In addition, the WT embryo has undergone axial rotation while the mutant is unturned. C: Lateral views of an E8.5 WT embryo and similarly staged mutant embryo. Anterior is to the left. The mutant has an ectopic tissue mass (arrow) and abnormal VYS with crater-like structures (arrowhead). Inset shows cratering of the VYS at higher magnification. D: Posterior views of an E8.5 WT embryo and mutant littermate after WMISH with a probe for Foxf1, which marks lateral plate mesoderm and allantois (bracket). Only limited allantoic growth is detected in the mutant (bracket). E: Sagittal sections of the embryos shown in (C). The ectopic tissue mass (arrow) is within the amniotic cavity. Corresponding regions of the VYS are marked by dotted rectangles and magnified in (F) and (F′). The abnormal extension of the chorion is marked by a dotted rectangle and magnified in (G). The arrowhead marks the abnormally thickened amnion in the posterior. F: Close up of WT VYS showing the apical vacuoles of the endoderm (arrowhead). F′: Close up of mutant VYS endoderm to highlight absence of apical vacuoles (arrowhead). G: Close up of abnormal extension of the chorion seen in the mutant. H: Close up of the VYS from an extreme lateral sagittal section of the mutant shown in (C) and (E) to illustrate the crater-like structures (arrowheads) in more detail. I: TUNEL analysis of E8.5 WT and mutant embryos. Cell death is limited in the WT but extensive in the mutant. J–L: Regions marked by dotted rectangles and shown in higher magnification include the VYS (J), the allantois (K), and neural tube (L).

Fig. 4.

Defective formation and separation of the ACF in Pagr1a^−/− embryos. A: Normal pattern of expression of Cldn3 in the ExE at PS (left panel), MS (middle panel), and LS (right panel) stages. The ACF (white arrowheads) is apparent by MS. Cldn3 is never found within the embryonic region (brackets). B: Lateral views of LB/EHF-stage WT and mutant littermates showing Cldn3 expression in the chorionic ectoderm. The ACF is undergoing separation at the ASP in the WT (left panel), whereas the ACF in the mutant is aberrantly located (right panel, arrowhead) with the distal-most portion within the embryonic region (right panel, arrow). C: Mid-frontal sections of the WT and mutant embryos shown in (B). The amnion is outlined by dotted rectangles and shown in higher magnification in (D) and (D′). The VYS at the boundary of the embryonic and extraembryonic domains also is outlined by dotted rectangles and shown at higher magnification in (E) and (E′). D: Close up of the WT amnion showing continuity with the embryonic ectoderm (arrowheads). D′: Close up of the mutant amnion showing Cldn3-positive chorionic cells (arrow) continuous with the amnionic embryonic ectoderm (arrowheads). E: Close up of the WT VYS showing prominent apical vacuoles within the endoderm layer (arrowhead). E′: Close up of the mutant VYS. The endodermal vacuoles appear smaller and fewer in number (arrowhead). F: Lateral views of LHF-stage WT and mutant littermates. The Cldn3 marked chorion (Ch) has separated from the amnion (Am) in the WT, whereas a large ectopic mass is seen within the mutant exocoelom (right panel, arrow) that is connected to the chorion (right panel, arrowhead). G: Sagittal section of the mutant embryo shown in (F). The area containing Cldn3-positive cells is outlined by the dotted rectangle and shown in higher magnification in (H). H: Close up showing that the Cldn3-positive chorionic cells maintain continuity with the amnion (arrowheads). I: Lateral views of E8.5 WT and mutant littermates after WMISH with Cldn3. A large ectopic mass is seen above the head of the mutant. This mutant was sagittally sectioned with the area defined by the dotted rectangle shown in higher magnification in (J). J: The large ectopic mass seen in the mutant in (I) is Cldn3-positive and in continuity with the amnion (arrowheads).

Pagr1a Mutants Undergo Normal Gastrulation and Germ Layer Specification

To understand the molecular basis for the Pagr1a^−/− developmental defects, we carried out WMISH analysis using gene expression markers specific for embryonic and extraembryonic lineages, examining stages both before and after the appearance of overt defects. During gastrulation, Hoxb1 marks mesoderm arising posteriorly from the primitive streak (Fig. 3A, left panel, bracket). There was no apparent difference in Hoxb1 expression in similarly staged Pagr1a^−/− embryos (Fig. 3A, right panel). Eomesodermin (Eomes), which is essential for specification of the definitive endoderm (DE) lineage, also is expressed in mesoderm arising from the primitive streak (Fig. 3B, left panel, arrow) as well as throughout the proximal ExE (Fig. 3B, left panel, bracket). We could not detect any significant difference in Eomes expression in Pagr1a^−/− embryos (Fig. 3B, right panel). The anterior visceral endoderm (AVE) is a population of extraembryonic cells that migrate from a distal location to the prospective anterior of the embryo. In LS/OB-stage embryos, Cer1 is found in the AVE (Fig. 3C, left panel, arrowhead) as well as in the nascent DE (Fig. 3C, left panel, arrow). In Pagr1a^−/− embryos, Cer1 expression was similar to WT (Fig. 3C, right panel). Nephrocan (Nepn) marks the entire midgut endoderm at early somite stages (Fig. 3D, left panel, bracket). In Pagr1a^−/− mutants, Nepn was expressed and found in the appropriate region (Fig. 3D, right panel, bracket), with the domain smaller but proportionate to that of the more developmentally advanced and larger WT littermate. Taken together these data indicate that Pagr1a^−/− embryos initiate normal gastrulation and properly specify the mesodermal and endodermal lineages.

Fig. 3.

Pagr1a^−/− embryos undergo normal gastrulation and patterning. A: Lateral views of WT and mutant littermates at the LS/OB stage after WMISH with a Hoxb1 probe, which marks mesoderm arising at the posterior (brackets). There is no significant difference in expression. B: Lateral views of MS/LS-stage WT and mutant littermates showing no significant difference in Eomes expression in the primitive streak (arrows) or in the ExE (brackets). C: Lateral views of LS/OB-stage WT and mutant littermates showing similar Cer1 expression in the DE (arrows) and AVE (arrowheads). D: Ventral views of E8.5 WT and mutant embryos after WMISH for Nepn, which marks the midgut endoderm (brackets). The expression domain in the mutant is proportionately sized. E: Ventral views of E8.5 WT and mutant littermates with presomitic mesoderm marked by Tbx6 (brackets). The expression domain in the mutant is proportionately sized. F: Ventral views of E8.5 WT and mutant littermates after WMISH for T/Brachyury (T), Uncx and Otx2 (not visible). Two to three somites are seen in the WT (curly bracket), but none in the mutant. Expression of T in the PSM (brackets), node (arrowhead) and notochord (arrow) is similar in WT and mutant. G: Lateral view of an E8.75 WT embryo showing Hoxb1 expression in rhombomere 4 in the hindbrain (arrow) and ventral view of a mutant littermate with a similar expression domain (arrow). H: Ventral views of the anterior-most regions of the same E8.5 WT and mutant littermate in (F) showing Otx2 expression normally present in the forebrain is absent in the mutant (arrows). I: Ventral views of the anterior-most regions of an E8.5 WT and mutant littermate following WMISH for Six3 and Foxa2. Six3, normally expressed in the forebrain, is absent in the mutant (arrows) whereas Foxa2 midline expression is retained (arrowheads). J: Lateral views of EB/LB WT and mutant littermates. Sox2 expression is seen in the anterior neuroectoderm (curves) and ExE (brackets) in both. The PAC of the mutant is wider (arrows) and the amnion is irregular (arrowhead). K: Lateral views of E9.0 WT and mutant littermates showing robust Sox2 expression in the WT anterior neuroectoderm but not in the mutant (brackets). A prominent ectopic mass is seen in the mutant (arrowhead). The asterisks in (D), (F), (G), and (K) mark tissue damage arising from dissection.

Pagr1a^−/− embryos arrest by early somite stages, and fail to elongate along the anterior–posterior (AP) axis and to undergo embryo turning. Wnt and fibroblast growth factor signaling in the primitive streak and presomitic mesoderm (PSM) have been implicated in somite formation and AP axis elongation (Naiche et al., 2011; Pourquie, 2011). The T box transcription factors T/Brachyury and Tbx6 are direct targets of Wnt signaling in the primitive streak and PSM. In E8.5 Pagr1a^−/− embryos, expression levels of both markers were not significantly different from WT embryos (Fig. 3E,F; brackets). T/Brachyury expression in the node (Fig. 3F, arrowheads) and notochord (Fig. 3F, arrows) of Pagr1a^−/− embryos also was similar to WT. These results indicate normal PSM specification and midline patterning in Pagr1a^−/− embryos, and suggest that the failure to undergo axial elongation and rotation may be a consequence of concurrent developmental arrest due to extraembryonic defects.

To analyze neural development, we first looked at Hoxb1 in E8.5 embryos. WT embryos show expression of Hoxb1 in rhombomere 4 (Fig. 3G, arrow). A similar domain, proportionately sized, was seen in Pagr1a mutant embryos suggesting proper patterning of the hindbrain. However, in more anterior domains there was altered expression of forebrain markers. Expression of Otx2, which marks the anterior neuroectoderm, and of Six3, present in the most rostral neuroectoderm, was either reduced or absent in some Pagr1a^−/− embryos (Fig. 3H,I). Of interest, while the pattern of Sox2 expression was similar to WT embryos at EB/LB stages (Fig. 3J), later-stage mutants failed to show the robust anterior expression that is found in WT embryos (Fig. 3K, brackets). This may either be the consequence of the cell death occurring in the neuroectoderm of the mutant (Fig. 2L) or point to a role for Pagr1a in maintaining anterior Sox2 expression, perhaps involving epigenetic mechanisms in conjunction with MLL3/4.

Defective Formation and Resolution of the ACF in Pagr1a Mutants

In addition to being expressed in the embryo proper, Sox2 marks the ExE and chorionic ectoderm (Fig. 3J, left panel, bracket). At the EB/LB stage, Pagr1a^−/− embryos show an apparently normal pattern and level of Sox2 expression in the extraembryonic domain (Fig. 3J, right panel). This indicates that in Pagr1a mutants the ACF forms and expands toward the anterior side. However, in some mutants the proamniotic canal (PAC) appeared wider (Fig. 3J, arrows) and a distinct ruffling could be seen in the region of the forming amnion (Fig. 3J, right panel, arrowhead). To further explore these defects, we analyzed Cldn3, which is expressed specifically in the ExE and chorionic ectoderm from an early stage and marks the nascent ACF (Fig. 4A). In LB/EHF- and LHF-stage WT embryos, the chorionic membrane marked by Cldn3 has separated from the amnion and segregated away from the embryonic region (Fig. 4B,F; left panels). In mutant littermates, the Cldn3 marked chorion was found still within the exocoelom (Fig. 4B,F; right panels). Further analysis of histological sections showed that the mutant ACF consisted of Cldn3-positive ExE joined to embryonic ectoderm derived amnion (Fig. 4C, right panel; Fig. 4D′), whereas in WT embryos this region is derived solely from embryonic ectoderm (Fig. 4C, left panel; Fig. 4D). At later stages, Cldn3 was found to mark the large ectopic tissue masses seen in mutants (Fig. 4F,I; right panels). Further analysis by histological sectioning confirmed that the masses consisted of Cldn3-positive ExE cells contiguous with amnion (Figs. 4G,H,J).

Together, these results suggest a specific order of events leading to the observed defects in amnion and chorion formation in Pagr1a mutants. As illustrated in the schematic shown in Figure 5A, normally the ASP represents the junction of embryonic and extraembryonic ectodermal components of the ACF to their cognate anterior tissues (Fig. 5A, left and middle panels). Subsequent separation leads to an amnion and chorion composed exclusively of embryonic ectoderm or ExE, respectively (Fig. 5A, right panel). In Pagr1a^−/− embryos, there appears to be an excess or mislocalization of the ExE component of the ACF such that it extends into the embryonic domain (Fig. 5B, left panel, brackets). Thus, at the mutant ASP there is no embryonic ectoderm, only ExE (Fig. 5B, middle panel, arrowhead). Separation from this abnormal junction appears to be incomplete in places (Fig. 2G; Fig. 4F, right panel, arrowhead), but where it does occur an abnormal membrane with both chorionic and amnionic components is formed (Fig. 5B, right panel, arrow; Fig. 4D′, arrow). It is likely that the large Cldn3-positive tissue masses seen at later stages in mutants result from continued proliferation of the chorionic component of this abnormal membrane.

Fig. 5.

A: Schematic representation of normal amnion and chorion formation from the ACF. Yellow, embryonic ectoderm; red, ExE; blue, mesoderm; green, endoderm. PAC, proamniotic canal; EPC, ectoplacental cavity; EXC, exocoelom; AMC, amniotic cavity. B: Schematic representation of amnion and chorion formation in the mutant. There is excess ExE within the ACF (brackets) such that the ASP contains only ExE (arrowhead). Following separation there is ExE derived tissue within the amnion (arrow).

Down-regulation of BMP2 in Pagr1a^−/− Mutant Embryos

Defective extraembryonic development is a prominent feature of embryos with mutations in components of the BMP signaling pathway. Defects specific to the amnion and chorion are seen in Smad1, Smad5, and Bmp2 mutant embryos. Smad1 mutants show over-proliferation of the ExE resulting in excessive folding of the chorion (Lechleider et al., 2001; Tremblay et al., 2001), somewhat similar to Pagr1a mutants. Smad5 mutants show extraembryonic tissue aggregates composed of mesodermal and ectodermal components of the amnion, thought to be in part due to ectopic BMP signaling (Bosman et al., 2006). These aggregates have a similar location as those seen in Pagr1a^−/− embryos. However, it is not clear if they also have a chorion component, as no specific ExE markers were checked. In Bmp2 null embryos, amnion formation is disrupted and can lead to failure to close the PAC and growth of the heart outside the amnion (Zhang and Bradley, 1996; Madabhushi and Lacy, 2011). These defects are more severe than seen in Pagr1a^−/− embryos, although other phenotypic features are shared with Pagr1a mutants including defects in anterior development and lack of axial rotation. Bmp2 expression is found in the mesoderm of the amnion and chorion at LS to EB stages in WT embryos (Fig. 6A,B; left panels, brackets). Of interest, in the majority of Pagr1a^−/− embryos analyzed at these stages, we found either a considerable reduction in Bmp2 expression (Fig. 6A, right panel) or almost complete down-regulation (Fig. 6B, right panel). As a control, we analyzed expression of the closely related Bmp4 gene, which is found in extraembryonic ectoderm at the PS to LS stages (Fig. 6C,D; left panels, brackets). We found no significant difference in Bmp4 expression levels in similarly staged Pagr1a mutant embryos (Figs. 6C,D; right panels). These results suggest that Pagr1a has a specific role in maintaining normal levels of Bmp2 gene expression, and indicate that the Pagr1a mutant phenotype may be due, at least in part, to reduced BMP2 signaling.

Fig. 6.

Down-regulation of Bmp2 gene expression in Pagr1a^−/− embryos. A: Lateral views of LS/OB-stage WT and mutant littermates showing Bmp2 expression in the mesoderm of the amnion and chorion (bracket) is significantly reduced in the mutant. B: Frontal views of OB/EB-stage WT and mutant littermates showing loss of Bmp2 expression in the mutant. C: Lateral views of PS/ES-stage WT and mutant littermates showing no significant difference in expression of Bmp4 in the proximal ExE (bracket). D: Lateral views of MS/LS-stage WT and mutant littermates also showing no apparent change in Bmp4 expression (bracket).

In conclusion, Pagr1a^−/− embryos undergo normal germ layer specification and developmental patterning but show limited anterior development and arrest by E8.5, likely due at least in part to defects in extraembryonic tissues that compromise their support functions. Consistent with initial Pagr1a expression being highest in the ExE, the first overt defects are found in the developing chorion and amnion, which derive from the ExE. Pagr1a is then broadly expressed in the embryonic and extraembryonic domains, and it will be an important avenue of further research to use tissue specific conditional inactivation to assess how loss of Pagr1a function in each lineage contributes to the phenotypic defects and embryonic arrest. Although our data shows that loss of Pagr1a affects Bmp2 transcription, further studies will be required to determine if regulation is direct. For instance, it will be of interest to assess whether this involves recruitment of MLL3/MLL4 complexes to the Bmp2 locus, or some other transcriptional co-regulatory function, as seen for ER target genes (Liang et al., 2009). It is also possible that Pagr1a regulates other, still unidentified target genes that act in parallel with the BMP pathway in extraembryonic development.

Experimental Procedures

Mouse and Embryo Analysis

The Pagr1a deleted allele (Fig. 1G) was PCR genotyped using forward primer A located 500 bp upstream of exon 1 (5′-CAGGACCAAAGTTCCCAGCTGCAGG-3′) and reverse primer C located in exon 3 (5′-GGCCCTCTGAGTCTAAGCTGAAG-3′), which results in a 297 bp product (Fig. 1H). The WT allele was PCR genotyped using forward primer B located in intron 2 (5′-TCTCGCAGCC GGCCAAGCCAACAG-3′) and reverse primer C, which results in a 530 bp product (Fig. 1H). Pagr1a heterozygous animals on mixed 129S6 and C57Bl6 background were intercrossed and the day of plug detection was assigned as E0.5. Embryos were dissected in phosphate buffered saline with 0.1% Tween (PBS-T) and fixed in 4% paraformaldehyde (PFA) overnight at 4°C. After washing twice with PBS, they were dehydrated and stored in 100% methanol at −20°C. To make the Pagr1a probe for WMISH analysis, a full-length cDNA was obtained from Open Biosystems (clone ID 3485982) and transferred into the pCMV-SPORT6 vector. Anti-sense probe was made by restriction digestion with EcoRI followed by transcription with T7 polymerase. Sense strand probe was made by restriction digestion with HindIII, and transcription with SP6 polymerase. WMISH to detect expression of Pagr1a and other genes was done as described previously (Kumar et al., 2008). Following WMISH, selected embryos were transferred into 80% glycerol in PBS and imaged using a Leica MZ16 stereomicroscope and Zeiss AxioCam MRc5 camera with AxioVision software. Embryos were PCR genotyped postanalysis by lysing in a small volume (20 μl for E6.5, 50 μl for E7.5 and 200 μl for E8.5) of 1× GeneAmp PCR Buffer II (Applied Biosystems, N8080010) containing 0.45% NP40 and 100 μg/ml of Proteinase K, incubating at 56°C for 4 hr and then heating at 95°C for 10 min. 1 μl of the lysate was used for PCR with AmpliTaq DNA polymerase (Applied Biosystems, N8080161) using the primers and conditions for amplification described above. Selected embryos were paraffin embedded, sectioned at 5 μm and counterstained with nuclear fast red or neutral red. High-resolution images of selected sections were obtained using the Aperio Scanscope slide scanner at 40× and analyzed using Aperio Imagescope software. For immunohistochemistry, fixed embryos were dehydrated in an ethanol series, embedded in paraffin and sectioned at 8 μm. TUNEL assay was done using Apoptag peroxidase in situ apoptosis detection kit (Millipore).

Ethics Statement

All experiments involving the use of animals were carried out on approved protocols in accordance with the policies and procedures set forth by the Animal Care and Use Committees of the NCI-Frederick and the NIDDK, NIH.