Origin and characteristics of glycogen cells in the developing murine placenta

Authors

  • P.M. Coan,

    Corresponding author
    1. Department of Physiology, Development, and Neuroscience, University of Cambridge, Downing Street, Cambridge, United Kingdom
    • Department of Physiology, Development, and Neuroscience, University of Cambridge, Downing Street, Cambridge, CB2 3DY UK
    Search for more papers by this author
  • N. Conroy,

    1. Department of Physiology, Development, and Neuroscience, University of Cambridge, Downing Street, Cambridge, United Kingdom
    Search for more papers by this author
  • G.J. Burton,

    1. Department of Physiology, Development, and Neuroscience, University of Cambridge, Downing Street, Cambridge, United Kingdom
    Search for more papers by this author
  • A.C. Ferguson-Smith

    1. Department of Physiology, Development, and Neuroscience, University of Cambridge, Downing Street, Cambridge, United Kingdom
    Search for more papers by this author

Abstract

The junctional zone (Jz) of the mouse placenta consists of two main trophoblast populations, spongiotrophoblasts and glycogen cells (GCs), but the development and function of both cell types are unknown. We conducted a quantitative analysis of GC size, number, and invasion of cells into the decidua across gestation. Furthermore, we identified markers of GC function to investigate their possible roles in the placenta. While the spongiotrophoblast cell volume doubles, and cell number increases steadily from E12.5 to E16.5, there is a remarkable 80-fold increase in GC numbers. This finding is followed by a notable decrease by E18.5. Surprisingly, the accumulation of GCs in the decidua did not fully account for the decrease in GC number in the Jz, suggesting loss of GCs from the placenta. Glucagons were detected on GCs, suggesting a steady glucose release throughout gestation. Connexin31 staining was shown to be specific for GCs. GC migration and invasion may be facilitated by temporally regulated expression of matrix metalloproteinase 9 and the imprinted gene product, Decorin. Expression of the clearance receptor for type II insulin-like growth factor (IGF-II), IGF2R, in a short developmental window before E16.5 may be associated with regulating the growth effects of IGF-II from glycogen cells and/or labyrinthine trophoblast on the expansion of the Jz. Thus stereology and immunohistochemistry have provided useful insights into Jz development and function of the glycogen cells. Developmental Dynamics 235:3280–3294, 2006. © 2006 Wiley-Liss, Inc.

INTRODUCTION

The field of placentology has benefited greatly from studying the mouse (Rossant and Cross,2001). Much of the work encompassing molecular biology, physiology, and function of mouse placental development has recently been reviewed (Adamson et al.,2002; Cross,2005). The mouse placenta consists of two fetal compartments (labyrinthine zone [Lz] and junctional zone [Jz]), and a maternal decidual component. Of the two fetal compartments, the Jz is the least well understood (Georgiades et al.,2002). The Jz is a cellular compartment consisting of at least two trophoblast subtypes: spongiotrophoblast and glycogen cells (GCs). The cells in this zone are critically important for fetal survival as demonstrated by the Ascl2−/− lethal mutant, which lacks a Jz (Guillemot et al.,1995; Tanaka et al.,1997). Ascl2 is an imprinted gene, expressed from the maternally inherited chromosome by the ectoplacental cone (EPC) cells and has been suggested to maintain the spongiotrophoblast state and prevent differentiation into giant cells (Tanaka et al.,1997). Control of Jz growth, and of the overlying giant cell layer, may be partly regulated by Nodal, as the spongiotrophoblasts and giant cell layer expand when Nodal is disrupted (Ma et al.,2001). Nodal may be limiting Jz growth, thereby allowing Lz growth after mid-gestation (Ma et al.,2001).

One of the main functions of the Jz is to act as an endocrine compartment. To maintain progesterone secretion from the corpus luteum, the spongiotrophoblasts of the Jz and secondary giant cells produce prolactin-like hormones (PLP-M and PLP-N), lactogens and cytokines during pregnancy (Ain et al.,2003; Malassiné et al.,2003; Soares,2004). Furthermore, several other targets for prolactin-like proteins have been identified, including liver, endothelial cells, and inflammatory cells (Soares and Linzer,2001). Another suggested function of the Jz is to limit the growth of maternal endothelium into the fetal placenta. Both Flt1 and proliferin-related protein, are anti-angiogenic proteins expressed by spongiotrophoblast cells in the Jz (He et al.,1999). It has been suggested that these proteins provide protection for the fetal placenta from invading maternal endothelial cells (Adamson et al.,2002).

Over the past few years, several groups investigating placental development have reported placentomegaly often associated with expansion of the Jz (Takahashi et al.,2000; Georgiades et al.,2001; Tanaka et al.,2001; Frank et al.,2002; Yang et al.,2003; Rodriguez et al.,2004). To our knowledge, there has not been a thorough investigation of the cellular composition of the Jz. Furthermore, in cases of altered placental development, qualitative observations suggest a shift in the balance between spongiotrophoblast cells and GCs (Li and Behringer,1998; Milstone et al.,2000; Yang et al.,2003).

The GC is a trophoblast cell of unknown origin that arises in the Jz. That GCs appear to arise in the Jz after spongiotrophoblasts (Redline et al.,1993; Teesalu et al.,1998; Adamson et al.,2002) and express the spongiotrophoblast-specific gene 4311 (Lescisin et al.,1988) lead researchers to suggest they were differentiated from spongiotrophoblasts (Adamson et al.,2002; Georgiades et al.,2002). However, recent evidence may suggest that GCs are distinct from spongiotrophoblasts. A group of cells in the central part of the EPC were found to express protocadherin12 (PCDH12). This PCDH12 marker was later shown to be expressed exclusively by the GCs (Rampon et al.,2005; Bouillot et al.,2006). Apart from their transitory behavior and accumulation of glycogen, not much more is known about the function of GCs. Thus our understanding of the development of the GC is very limited. Unanswered questions include: Where do GCs originate? Where do GCs go? What are the functions of GCs in pregnancy and placental development? What is the process of GC development? How much of the cellular composition of the Jz consists of GCs throughout gestation? Can the appearance and migration of GCs out of the fetal placenta explain changes in Jz structure and volume in late gestation? To shed light on the dynamic development of the Jz and the cell populations within, we embarked on a quantitative study of cell size, number, and migration using well-established stereological techniques. Determining the dynamic changes in cell number is critically important in understanding tissue growth, differentiation, and homeostasis. Particularly important in heterogeneous tissues are the relative quantities of various differentiated cell types that together help provide the diverse array of functions that an organ such as the placenta performs. In addition, we have used immunohistochemical techniques, to elucidate the proteins involved in GC development and assign putative roles for the GC in placental function and fetal well-being.

RESULTS

Cell Size and Number

By making use of histological sections at various gestational ages and the unbiased tools illustrated in Figure 1A–D, we attempted to answer the spatial and temporal related questions of development of the Jz and the GCs. The Jz volume changes dynamically over gestation. Therefore, quantifying changes in cell size and number would lead to a deeper understanding of the dynamic development of this zone.

Figure 1.

A: Randomly selected fields of view (A–D) from the Computer Assisted Stereological Toolbox 2.0, and (A) paraffin section of 7-μm thickness with superimposed point grid for estimating placental volume. B: Resin section with a view of spongiotrophoblasts and glycogen cells superimposed with a line grid. The line grid is required for assigning a random start point from which to make a point-sampled intercept measurement (highlighted in red) that will be used to estimate mean volume-weighted cell volume. C: Resin section with a view of a giant cell with the Vertical Nucleator superimposed. The distance (highlighted in red) from the random central position to the cell boundary is measured in four directions by the computer to estimate giant cell volume. D: Paraffin section of 20-μm thickness with a view of the junctional zone. A counting frame has been superimposed, and cell nuclei that come into focus through a 15-μm depth of the section can be counted.

At embryonic day (E) 12.5, the Jz contains predominantly spongiotrophoblast cells. A twofold increase in the size of individual spongiotrophoblast cells to ∼7,000 μm3 occurred by E16.5 (Table 1), but there was no further change with gestational age. There was also a significant increase in spongiotrophoblast cell number. We calculated an increase from ∼1.7 million at day E12.5 to 6.2 million at day E16.5, an almost fourfold rise (Table 1). No further alteration in the number of spongiotrophoblasts was observed for the remainder of gestation (Table 1).

Table 1. Mean Cell Volume (μm3) and Mean Cell Number Estimate With Gestational Age
 Stage in gestationANOVA P value
E12.5 5 litters n = 10E14.5 5 litters n = 6E16.5 6 litters n = 6E18.5 4 litters n = 8
  • a, b, c, d

    values within each row are significantly (P < 0.02) different as identified by Fisher protected least significant difference test. ± SEM. N = number of conceptuses analyzed.

  • Db glycogen cell, glycogen cells that have migrated into the decidua; Jz glycogen cell, glycogen cells in the junctional zone.

Cell size     
Spongiotrophoblast3369 ± 331a5853 ± 404b7193 ± 59c7310 ± 226c<0.0001
Glycogen cell549 ± 60a1564 ± 307b1311 ± 115b<0.0001
Secondary giant cell101387 ± 8382124478 ± 11710136548 ± 18871111409 ± 128980.0741
Cell number     
Spongiotrophoblast1.71 × 106a3.28 × 106ac6.23 × 106bd5.12 × 106cd0.0004
Jz glycogen cell2.48 × 104a1.39 × 106b4.67 × 106c1.81 × 106b0.0004
Db glycogen cell5.30 × 105a1.78 × 106b7.29 × 105a0.0019
Secondary giant cell1.57 × 1042.02 × 1042.39 × 1042.44 × 1040.1673

As there were insufficient GCs at E12.5 to verify mean cell volume statistically, volume measurements were taken from E14.5 and later. Mean GC size increased threefold to ∼1,500 μm3 in the window between E14.5 and E16.5 (Table 1). GCs were found to be roughly one-tenth the size of a spongiotrophoblast cell at E14.5, and one-fifth by E16.5 (Table 1). Next, using the optical disector method, we were able to distinguish and count those GCs at E12.5 whose cytoplasm was vacuolated. There were very few GCs identified at this stage, and not all the placentas examined had visible GCs. However, after this stage, there was an 80-fold increase in the number of GCs, reaching ∼2 million cells by E14.5 (Table 1). Between these time points more than half a million GCs had traversed the giant cell layer (whose cells do not alter in size or number) and migrated into the decidua (Table 1). Later in gestation, the number of GCs increased to almost 6.5 million at E16.5. Of these cells, we estimated the number invading the decidua had increased by over 1 million (Table 1). Strikingly, by E18.5, we found a significant reduction in GCs in both the Jz and the decidua. Less than half the GCs counted in the Jz at E16.5 remained by E18.5, but intriguingly, we noted that fewer than half the GCs observed at E16.5 in the decidua were present in that zone by E18.5 (Table 1).

GCs across gestation always cluster together. These clusters vary in size, but do not correlate with gestational age. Furthermore, these clusters may be discreet balls or “streams” of GCs spanning from the top of the Lz through the Jz to the decidua. Once in the decidua, GCs clusters appear less tight, allowing mixing with cell types in the decidua. Occasionally GCs were observed deep within the Lz, although the functional significance of this, if any, remains to be determined.

Immunohistological Analysis of GC Development and Function

To elucidate further the proteins involved in GC development and function, we selected antibodies against several protein candidates; Ki-67 and p57Kip2 were selected to investigate cellular proliferation and cell cycle inhibition; glucagon and periodic acid–Schiff (PAS) were selected to investigate glycogen storage and breakdown; matrix metalloproteinase 9 (MMP-9) and Decorin (Dcn) were selected to investigate GC invasion/migration; Connexin31 (Cx31) was selected to investigate GC differentiation and cell communication; the type II insulin-like growth factor (IGF-II) receptor was selected to investigate regulation of IGF-II produced by the Lz and Jz over gestation.

Expression of Ki-67 in the Jz

Ki-67 is a nuclear protein intrinsically linked with cell proliferation (Gerdes et al.,1983,1991; Duchrow et al.,1996). To estimate the proliferative activity of cells in the Jz, Ki-67 antiserum was applied at two gestational ages: at E12.5 to identify the possibility of pre-glycogen proliferation, and at E16.5 to identify remaining proliferative activity at maximum Jz volume.

At E12.5, the majority of cells in the Jz whose nuclei were Ki-67–positive were in patches or clusters of small cells (Fig. 2A). Those larger cells identified as spongiotrophoblasts by their morphology and relative size were Ki-67–negative (Fig. 2B). This finding may indicate that the GCs contribute more than spongiotrophoblasts toward an increase in Jz cell number. By E16.5, most of the cells in the Jz were Ki-67–negative (Fig. 2C). However, within the clusters of GCs, there were numerous Ki-67–positive cell nuclei (Fig. 2C). This finding might suggest that GCs may still have the ability to divide at E16.5, whereas the majority of spongiotrophoblasts no longer proliferate.

Figure 2.

A–L: Detection of early protein markers for glycogen cells. A–C: Ki-67 positivity identified proliferating cells. D–F: Periodic acid–Schiff (PAS) staining for presence of glycogen. G–I: Anti-p57Kip2, a known marker for glycogen cells. J–L: Anti-glucagon staining. A,B,D,G,J: At E12.5, showing pre-glycogen cells. E,H,K: At E13.5, showing emerging glycogen cells. C,F,I,L: At E18.5, showing late glycogen cells. Db, decidua basalis; gc, glycogen cell cluster; Jz, junctional zone; Lz, labyrinthine zone; mvs, maternal venous sinus; pre-gc, putative “pre-glycogen” cell; spt, spongiotrophoblast; tg, secondary giant cell. Arrows indicate where positive diaminobenzidine (DAB; for antibodies) or PAS (for glycogen) staining. Scale bar = 20 μm in A,C,D,F–I, 10 μm for B,EJ–L.

Temporal and Spatial Localization of Glycogen

PAS-positivity was observed in a few small clusters of cells in the Jz at E12.5 (Fig. 2D). The cells were morphologically similar to the majority of the PAS-negative spongiotrophoblasts present, except for being much smaller (Fig. 2D). Furthermore, analogous cells deeply stained by toluidine blue were observed in resin sections (Fig. 3A,B). Small round cells in the decidua were also found to be PAS-positive. One day later in gestation, there was a major increase in the number of small spongiotrophoblast-like cells that were staining positive for PAS. Although clusters of PAS-positive cells were observed to be profuse and random throughout the Jz, many were found surrounding the maternal venous sinusoids of the Jz (Fig. 2E). PAS-positive cell clusters also were observed near the Lz and next to the secondary giant cell layer. We suggest that these PAS-positive cells are GC precursors, for similar cells were identified by electron microscopy to contain glycogen granules (Fig. 3C).

Figure 3.

Images of emerging (pre-glycogen cells) and expanding glycogen cells at the light and ultrastructural level. A: Resin-embedded section illustrating the small, dark, clustered putative pre-glycogen cells at embryonic day (E) 12.5. B: Resin-embedded section illustrating late glycogen cells with their cytoplasms full of glycogen at E14.5. C: Transmission electron photomicrograph of pre-glycogen cells at E12.5. D: Transmission electron photomicrograph of expanding glycogen cells (GCs) at E16.5. Arrows indicate glycogen cells. Cyt, cytoplasm; gc, glycogen cell; gg, glycogen granules; n, nucleus. Scale bar = 5 μm.

After E13.5, GCs were easily identified morphologically by their vacuolated cytoplasm, and on adjacent histological sections, these cells alone were PAS-positive (Fig. 2F). Moreover, similar cell types were clearly visible by their affinity for toluidine blue dye (Fig. 3B). Later in gestation, using electron microscopy, cells of similar morphology packed with glycogen granules were identified (Fig. 3D). The cytoplasm of no other cell type within the Jz was observed to be PAS-positive. At this stage, morphologically similar cells within the decidua were also PAS-positive, whereas most other cells within the decidua were PAS-negative. Islets of PAS-positive GCs could also be identified within the Lz.

Expression of P57Kip2 in the Jz

The imprinted gene Cdkn1c encodes a developmentally regulated cyclin-dependent kinase inhibitor, P57Kip2. This critically important protein binds to several G1 cyclin/Cdk complexes and inhibits their kinase activity (Matsuoka et al.,1995). Thus p57Kip2 is a negative regulator of cell proliferation. P57Kip2 has been used previously to identify GCs (Georgiades et al.,2002). For these reasons, we hypothesized that P57Kip2 is required for the progression of pre-GCs to GCs by inhibiting their proliferation and promoting differentiation.

At E12.5, expression of p57Kip2 is patchy with respect to cell nuclei in the Jz and no specific subtype of trophoblast was found to have exclusive expression (Fig. 2G; Table 2). As expected, most giant cell nuclei observed demonstrated strong positive expression of p57Kip2. A day later, however, there was a clear separation of cell populations within the Jz. Strong positive nuclear staining was found in the few GCs present at this stage (Fig. 2H; Table 2). A further distinct and profuse cell type often closely associated with the small GC islets was observed to have positively stained nuclei. These cells had the morphology of a spongiotrophoblast cell but were smaller, both in size of nuclei and volume of cytoplasm (Fig. 2H). Coincidentally, on adjacent PAS-stained sections, these same positive cells were also observed to stain for glycogen (Fig. 2E). This finding suggests that cell cycle arrest is coincident with PAS-positivity in pre-GCs. Spongiotrophoblasts were completely negative for p57Kip2. Nuclei of giant cells at this stage in gestation still stained strongly for p57Kip2 (Fig. 2H; Table 2). From E14.5 onward, only glycogen and giant cells were observed to stain for p57Kip2 (Fig. 2I; Table 2). Spatial position in the placenta did not affect GC expression of p57Kip2 as we observed positively stained GC islets in the Lz. Control sections stained in the absence of primary antibody were negative throughout.

Table 2. Temporal and Spatial Expression of Selected Proteins in the Developing Chorioallantoic Mouse Placentaa
ECompartment/layerCell typeCx31DcnGluIGF2RMMP-9P57Kip2Ki67
  • a

    +, expression detected; +/−, some expression but not in all cells; −, no expression detected.

  • Cx31, connexin31; Db, decidua basalis; Dcn, decorin; FE/Me, fetal endothelium/ mesenchyme; GC, glycogen cell; Glu, glucagon; IGF2R, type II insulin-like growth factor receptor; Jz, junctional zone; Lz, labyrinthine zone; MMP-9, matrix metalloproteinase9; preGC, putative GC progenitor; TG; giant cell layer; 2'TG, secondary giant cell; Tr, labyrinthine trophoblast.

12.5Db +++
 TG2'TG+/−+/−
 JzSpt
  preGC++
 LzTr+++
  FE/Me+++
  Chorion+++ 
          
13.5–14.5Db ++++?
 TG2'TG+/−+/−+/−+/−?
 JzSpt++/−++/−+/−?
  GC++++++?
 LzTr+++/−?
  FE+++?
  Chorion +++?
          
16.5–18.5Db + ++/− 
 TG2'TG +/−+/−+/−
 JzSpt++/−
  GC++++++/−
 LzTr+++/−+/−
  FE+++/−+/−
  Chorion++++/−

Expression of Glucagon in the Jz

Glucagon is a small protein hormone produced primarily by the pancreatic α cells that has opposite effects to insulin on the metabolism of sugar (Stryer,1996). Glucagon binds to receptors on glycogen storing cells such as hepatocytes, stimulating the breakdown of stored glycogen to glucose by glycogenolysis (Lodish et al.,1998). To our knowledge, an assessment of glycogen catabolism has not been conducted in the placenta. We predicted that glucagon would be present on the GCs late in gestation when extra energy supplies may be required by the fetus, placenta, or mother to complete pregnancy successfully.

At E12.5, no specific positive staining was observed in the Jz (Fig. 2J; Table 2). However, many distinct cells in the maternal decidua stained positively, although their identification by morphology alone was difficult (Fig. 2J; Table 2). Later in gestation, we observed scant GCs that stained positively for glucagon exclusively on their cell membranes (Fig. 2K). This finding indicates that the antibody may be detecting receptor-bound glucagon. By E14.5, the strongest staining was observed on the GC membranes, although some spongiotrophoblasts also stained positive (Table 2). Late in gestation glucagon staining could be attributed exclusively to GCs in both the Jz and decidua (Fig. 2L; Table 2). Furthermore, the islets of GCs in the Lz were glucagon-positive (Table 2). No staining was present on sections treated without primary antibody. The combination of PAS positivity and glucagon staining indicates that glycogen stores in GCs are being broken down to glucose in this population of cells. The loss of staining in sections treated with saliva confirmed that the source of the PAS-positivity was glycogen.

Exclusive Expression of Cx31 by GCs

Cx31 is one of a family of proteins that together form the multi-subunit protein complexes of gap junctions. Gap junctions are one way in which the cytoplasms of adjacent cells can communicate directly with one another.

Cx31 has been found to be important for the differentiation of trophoblast stem cells (Dahl et al.,1996). Furthermore, murine Cx31 is expressed by the ectoplacental cone and its derivatives (Dahl et al.,1996; Reuss et al.,1996). This expression persists and Cx31 is found in the Jz followed by a coexpression of Cx43 at early midgestation (Kibschull et al.,2005). Cx31 might, therefore, be important for the differentiation of GCs. However, spatial and temporal expression of this protein in the mouse placenta has not been determined to our knowledge.

Of the few vacuolated GCs observed at E12.5, all weakly stained for Cx31 (Fig. 4A; Table 2). From E13.5 until the end of gestation, we found exclusive expression of Cx31 by the vacuolated GCs in the Jz, decidua, and in the Lz islets (Fig. 4B; Table 2). The staining was predominantly membrane-specific, demonstrating that this gap junction protein is expressed in the GC membrane, presumably in a functional gap junction (Fig. 4A–C). Positive staining was never observed in spongiotrophoblast, giant, or decidual cells. No staining was observed in the absence of primary antibody. These findings suggest the presence of Cx31-associated gap junctions solely in the differentiated GCs of the placenta.

Figure 4.

Expression of connexin 31, matrix metalloproteinase 9, decorin, and the type 2 insulin-like growth factor receptor in the junctional zone and decidua basalis over gestation. A–C: Connexin 31 embryonic day (E) 12.5 Jz (A), E13.5 Jz (B), E16.5 db (C). D–F: Matrix metalloproteinase 9 E12.5 Jz (D), E13.5 Jz (E), E16.5 Jz (F). G–I: Decorin E12.5 Jz (G), E14.5 Jz (H), E14.5 db (I). J–L: IGF2R E12.5 Jz (J), E14.5 Jz (K), (L) E16.5 Jz. db, decidua basalis; dc, decidual cells; Gc, glycogen cell cluster; Jz, junctional zone; spt, spongiotrophoblasts; tg, secondary giant cell. Scale bar = 10 μm in A–H, J–L, 50μm in I.

Expression of MMP-9 in the Jz

MMP-9 is a member of a family of membrane bound proteases that are capable of degrading the extracellular matrix of the uterine endothelium (Roth and Fisher,1999). MMP-9 has critical roles in placental development and tumor invasion (Van den Steen et al.,2002). Furthermore, MMP-9 is highly expressed by one putative human equivalent of the GC, the invasive extravillous trophoblast (Behrendtsen et al.,1992). However, its expression during mouse placental development has not been documented.

We found no expression of MMP-9 at E12.5 in the Jz or decidua of the mouse placenta (Fig. 4D; Table 2). With the emergence of GCs on the following day, we found positive staining for MMP-9 on the GCs. Expression was found in the perinuclear cytoplasmic region, possibly within Golgi bodies, and on the membranes of the GCs. Staining was not observed on spongiotrophoblasts or secondary giant cells (Fig. 4F; Table 2). At E14.5, the GCs continued to stain positively for MMP-9 (Table 2). However, at this stage, we also observed expression of MMP-9 in some clusters of spongiotrophoblasts and occasionally in the cytoplasm of giant cells (Table 2). Later in gestation, we continued to detect MMP-9 on GCs in the Jz and decidua as well as in the occasional spongiotrophoblast or giant cell (Fig. 4F; Table 2). The intensity of staining for GCs in the decidua and Jz was found to be the same. The timing and localization of MMP-9 suggests that this protease may play a role in the migration of GCs from the Jz into the maternal decidua.

Expression of Dcn in the Jz

Dcn is the protein product of an imprinted gene expressed from the maternally inherited chromosome. Dcn is a member of the small leucine-rich proteoglycan (SLRP) family, which can regulate matrix assembly and cellular growth (Iozzo,1998; Stander et al.,1999). Furthermore, the core protein has transforming growth factor-beta (TGF-β) binding sites and is able to inactivate TGF-β (Yamaguchi et al.,1990; Iozzo,1998; Stander et al.,1999). Dcn has also been shown to inhibit migration of endothelial cells (Kinsella et al.,2000; de Lange Davies et al.,2001). Hence, analysis of its expression might determine whether Dcn may be expressed by either the GCs or the maternal decidua to regulate invasion during chorioallantoic placental development in the mouse. Indeed, expression of Dcn in the decidua of the human placenta has been associated with the regulation of proliferation and invasion of extravillous trophoblasts in the first trimester of pregnancy (Xu et al.,2002).

We found no specific staining for Dcn in the Jz at E12.5, although the decidua was strongly stained (Fig. 4G; Table 2). During the following 2 days, some of the small spongiotrophoblast cells in the Jz along with the GCs stained positively for Dcn. The staining observed in both the spongiotrophoblast cells and GCs was not membranous, suggesting that this protein is at least initially cytoplasmic (Fig. 4H). By E16.5, there was more staining present in the spongiotrophoblasts and the GCs continued to stain strongly in their cytoplasmic regions. After E16.5, the spongiotrophoblast staining waned and there was a significant change in the location of the Dcn staining of the GCs (Table 2). Dcn in the E18.5 GCs was no longer cytoplasmic, but rather associated with the cell membrane. No staining was present on sections where the primary antibody was omitted. These findings indicate that Dcn is a developmentally regulated protein in the Jz and that subcellular localization may be relevant to its function.

Expression of the Type 2 IGF Receptor in the Jz

The type II IGF receptor (IGF2R) is the protein product of an imprinted gene Igf2r expressed from the maternally inherited chromosome homologue. IGF2R is identical to the mannose-6 phosphate receptor, which, using distinct binding domains, associates with both the IGF-II ligand and proteins with mannose-6-phosphate residues. IGF-II is important for placental development and fetal growth (Baker et al.,1993; Constancia et al.,2002). As IGF2R is a negative regulator of IGF-II, it was necessary to discover where and when IGF2R is expressed in the placenta.

We observed no staining in the Jz until E14.5, when there was strong staining for the IGF2R (Fig. 4J,K; Table 2). GCs were also positively stained in the small area of cytoplasm that remains adjacent to the nucleus (Fig. 4K; Table 2). Spongiotrophoblasts and some secondary giant cells stained positively in the perinuclear region of the cytoplasm. The strong distinctive IGF2R localization pattern continued until E16.5, when there was a dramatic reduction in staining everywhere in the placenta (Fig. 4L; Table 2).

An excellent in vivo negative control for the IGF2R antibody is the Thp mutant. This mutant has a large deletion on mouse chromosome 17 that includes the Igf2r gene as well as several others. When this deletion is inherited from the maternal allele, conceptuses lack IGF2R and die around E16. Wild-type siblings present in litters can act as a positive control for Igf2r expression. In the Thp mutant placentas, as expected, no positive staining could be found. These findings were identical to those of the negative controls where, when the primary antibody incubation step was omitted, no expression of IGF2R was observed. However, the wild-type sibling placentas from the Thp litters showed normal spatial expression of IGF2R (data not shown).

Expression of p57kip2 and IGF2R in the Lz

Of the proteins investigated in the development and function of the GCs, all but Cx31 were also expressed in the Lz (Table 2). Of particular interest were p57Kip2 and IGF2R. The expression pattern of p57Kip2 in the Lz may also provide support for the suggestion that p57Kip2 is not merely a marker for terminal differentiation, but that it may also act as a molecular “switch” controlling cell cycle and differentiation (see also later). Before E14.5, fetal endothelial nuclei stain strongly for p57Kip2 (Fig. 5A; Table 2). This finding may suggest that capillary angiogenesis is paused between E12.5 and E14.5, allowing time for fetal growth and sufficient increase in fetal blood volume, before permitting further angiogenesis for greater hemotrophic exchange later in gestation (Fig. 5B,C). This suggestion is supported by our stereological investigation of the Lz, confirming this temporal pattern of capillary volume growth and expansion (Coan et al.,2004).Syncytial trophoblast nuclei expressed p57Kip2 before and after E14.5 (Fig. 5A; Table 2). This finding may indicate that cytotrophoblasts that have newly fused to form syncytium are then prevented from further replicating by expressing this proliferation inhibitor. Some of the larger cytotrophoblast nuclei express p57Kip2 (Fig. 5A–C; Table 2). This expression might indicate that the cell cycle inhibitor is required to undergo endoreduplication, as we have previously suggested (Coan et al.,2005). Furthermore, Hattori et al. suggested that punctuated expression of p57Kip2 was important for the regulation of DNA replication, whereas trophoblast giant cells undergo endoreduplication (Hattori et al.,2000).

Figure 5.

Expression of p57Kip2 and IGF2R in the labyrinthine zone. A: p57Kip2 at embryonic day (E) 12.5. B: p57Kip2 at E14.5. C: p57Kip2 at E18.5. D: IGF2R at E12.5. E: IGF2R at E14.5. F: IGF2R at E15.5. G: IGF2R at E16.5. IGF2R, type II insulin-like growth factor receptor; Ct, labyrinthine cytotrophoblast; fc, fetal capillary; fe, fetal endothelium; gc, glycogen cells; mbs, maternal blood space; syntr, syncytial trophoblast. Scale bar = 10 μm.

Trophoblast expression of IGF2R was restricted to the spongiotrophoblast and GCs of the Jz and occasionally secondary giant cells between E14.5 and E16.5 (Fig. 4K; Table 2). However, IGF2R was also expressed in the same developmental window by the fetal endothelium (Fig. 5E,F). The fetal endothelium, GCs, and spongiotrophoblasts may be providing regulatory barriers for IGF-II (see also later). This possibility is important, as IGF-II is expressed both by the GCs and by the labyrinthine trophoblast (Redline et al.,1993). Therefore, the one possible mechanism that may prevent fetal IGF-II causing Lz overgrowth might come from the expression of IGF2R by the fetal endothelium from E14.5 to E15.5 (Fig. 5E,F; Table 2). Before E14.5 and after E15.5, expression of IGF2R drops dramatically and it appears that no IGF2R protein is present outside the developmental time window (Fig. 5D,G; Table 2). Thus labyrinthine trophoblast IGFII can promote its own growth without IGF2R regulation. How the IGF-II signal is mediated, resulting in the growth effects in the placenta, is yet to be discovered, despite many transgenic and knockout mouse models investigating IGFs and their receptors (Liu et al.,1993; Lau et al.,1994; Ludwig et al.,1996; Louvi et al.,1997). One possible mediator is IGF2R and, as suggested by others, another may be an uncharacterized placenta-specific receptor XRp (Baker et al.,1993; Ludwig et al.,1996).

DISCUSSION

Cell Size, Number, and Migration Play Roles in the Dynamics of the Jz

Previous work has shown that the volume of the Jz is biphasic across gestation, expanding until E16.5, and then reducing by E18.5 (Coan et al.,2004). Here we show that alterations in cell size, proliferation, and migration are responsible for these dynamic changes in this compartment of the placenta. We estimated that maximum cell size corresponds to when the volume of the placenta is at its greatest. Furthermore, a dramatic proliferative event takes place from E12.5 to E14.5 in which an explosion in the GC population occurs. This finding may indicate that the major contributor to cell proliferation in the Jz is the GC. Owing to the combination of these changes with the increased number of spongiotrophoblasts, E16.5 marks the point in gestation for maximum cell size and number in the Jz. Furthermore, this is possibly the point when cell proliferation in the Jz ceases. This possibility is in contrast to the Lz, which continues to proliferate and develop until term as indicated by both stereological (Coan et al.,2004) and immunohistochemical (Coan, unpublished observations, data not shown) analysis.

The increase in spongiotrophoblast size and number likely relates to their function in the production of hormones and other factors necessary for a successful pregnancy. We recently showed that, in addition to increasing in size and number, spongiotrophoblast subcellular complexity also increases across gestation (Coan et al.,2005). Thus by E16.5 the spongiotrophoblasts are sufficient in number and function to provide necessary signaling and support for the remainder of gestation, although the factors produced remain unknown at present.

The subsequent decrease in the Jz volume after E16.5 is consistent with the continued migration of GCs away from the Jz. However, Waddell et al. found an increase in DNA fragmentation in the Jz of the rat placenta toward term, suggesting that apoptosis may also play a role in the reduction in this zone's volume (Waddell et al.,2000).

Identification of GCs

GCs express the spongiotrophoblast-specific gene 4311 and do not express the secondary giant cell gene Pl1, suggesting that GCs are derived from spongiotrophoblast (Adamson et al.,2002). However, Bouillot et al. recently described expression of protocadherin12 in a discrete population of ectoplacental cone cells, which were later in gestation identified as GCs (Bouillot et al.,2006).

Here, we focused on the period around the time that GCs begin to accumulate glycogen. Observations of our resin and paraffin sections around E12.5 revealed populations of small trophoblast cells arranged in clusters. We predict that these cells correspond to early pre-GCs that have not accumulated sufficient glycogen stores to exhibit the classic vacuolated appearance of GCs following fixation and embedding. Furthermore, these cells are distinct from spongiotrophoblasts, representing a discreet trophoblast lineage that may express protocadherin12 (Bouillot et al.,2006). Ultrastructural study on placental sections show pre-GCs, surrounded by extracellular matrix and beginning to accumulate glycogen granules (Fig. 3; Coan et al.,2005). Using p57Kip2 as a marker for GCs and PAS, we positively identified GCs.

P57Kip2 Marks a Transition in GC Development

The cell cycle regulator protein p57Kip2 has been used previously as a marker of GCs (Georgiades et al.,2001,2002). P57Kip2 has an important role in placental development. A transgenic mouse deficient in p57Kip2 displayed placentomegaly due to an increase in labyrinthine trophoblast and spongiotrophoblasts (Takahashi et al.,2000). We showed expression of p57Kip2 in the pre-GCs at E12.5, and vacuolated GCs until gestation ends, but not in the spongiotrophoblasts. A lack of GCs in the p57Kip2 mutant might have been expected if indeed p57Kip2 is required for the formation of GCs. Moreover, the pre-GCs deficient in p57Kip2 might adopt a path similar to spongiotrophoblast-like development, which may explain the increase in spongiotrophoblasts observed in this mutant. However, GC number in the mutant was not decreased compared with wild-type according to the counting procedure adopted by Takahashi et al. However, their counting method does not sufficiently estimate actual cell numbers. To rigorously compare cell numbers between the normal sized wild-type and oversized p57Kip2-deficient placenta, one would need to consider the reference space. This common predicament is known as the reference trap (Mayhew et al.,2003). By contrast, the methods detailed in our study provide an unbiased and accurate estimate of cell numbers, and applying them to this mutant might provide insight into the role of p57kip2 in GC formation.

P57Kip2 is unique among the cdk inhibitors in being essential for normal embryogenesis and placental development, for neither p21Cip1 nor p27Kip1 knockout mice present gross morphological defects (Yan et al.,1997; Zhang et al.,1997). Deciphering the roles of p57Kip2 in placental development is complex. However, in vitro work on this protein may provide clues to its role in GC development. Yokoo et al. found that, apart from p57Kip2 regulating cell cycle, it also contributes to the regulation of actin dynamics (Yokoo et al.,2003). P57Kip2 associates with LIMK-1, a LIM containing protein kinase that regulates actin polymerization–depolymerization (Yokoo et al.,2003). Thus when high levels of p57Kip2 are present in a cell, p57Kip2 binds LIMK-1 and translocates the kinase into the nucleus. This process causes cytoplasmic depletion of LIMK-1 and inactivation of downstream factors such as cofilin, thereby inducing actin depolymerization (Yokoo et al.,2003). The zinc finger-containing LIMK-1 may also have a function within the nucleus concerned within cell fate determination and growth regulation and may be important for cell differentiation (Yokoo et al.,2003). In addition, work on the Xenopus homologue of p57Kip2, has shown p27Xic1 to be involved in cell fate determination in retinal cells (Ohnuma et al.,1999). This finding is relevant when considering expression of p57Kip2 in the developing mouse placenta and suggests that, before accumulating glycogen, the pre-GCs express p57Kip2 to exit cell cycle and undergo differentiation into GCs. As part of the differentiation process, one would expect alterations in the cytoskeleton that might be necessary for the storage of glycogen and furthermore for the migration of GCs into the decidua. Moreover, Westbury and others suggested that p57Kip2 was not only expressed by terminally differentiated cells, but was present in cells undergoing shape-changing, mitosis, and cell division (Westbury et al.,2001).

GCs as a Potential Energy Source

This report is the first study to investigate temporal and spatial expression of glucagon in the mouse placenta. The hormone glucagon binds to cells, stimulating breakdown of their glycogen stores to release the energetically favorable glucose-6-phosphate. Glucagon staining was found predominantly associated with the GCs throughout their existence in the Jz and decidua. One would predict that glucagon might act in an autocrine or paracrine manner, stimulating surrounding GCs to release glucose. If autocrine, one would then expect perinuclear cytoplasmic staining for glucagon being synthesized, which we did not find.

The definitive reason for glycogen and glucose release by the GCs has eluded investigators thus far. The main source of energy for the placenta comes from glucose, which may, therefore, use the glycogen stored (Barash and Shafrir,1990). Alternatively, the glycogen stores may be destined for the fetus (Barash and Shafrir,1990). By the end of gestation, the fetus is placing high demands on maternal resources for growth, but thanks to the glycogen reserve of the GCs, this glycogen can be converted into glucose in the final stages of gestation and be released directly into the maternal blood, which eventually circulates back through the Lz. Although qualitatively it is difficult to determine whether less glycogen is present in GCs later in gestation, Lopez et al. demonstrated biochemically that total placental glycogen decreased between E15 and E18 in the normal placenta (Lopez et al.,1996).

GC-Specific Role of Cx31

Gap junctions comprising connexins are important portholes for communication between cells. Using a commercial antibody specific to Cx31 we have demonstrated unique expression of Cx31 in emerging GCs in the Jz from E12.5. Specific expression of Cx31 continues in these cells until term.

Disruption of the Cx31 gene resulted in placental dysmorphology with disruption of the labyrinthine and Jz (Plum et al.,2001). Thus Cx31 may be important in determining spongiotrophoblast cell fates (Plum et al.,2001). A further study investigating Cx31 and Cx43 in placental development suggested that connexins were not essential to maintaining spongiotrophoblast proliferation and that Cx31 is probably required only early in trophoblast lineage development, postimplantation and before E9.5 (Kibschull et al.,2005). Our data suggest that Cx31 may also play a role from E12.5 to term.

What is the function of Cx31 in GC development? Perhaps Cx31 is required to commit the trophoblast to the GC lineage, or alternatively to suppress trophoblast from following the spongiotrophoblast or giant cell path. However, expression of Cx31 by GCs may simply be a consequence of differentiation and is linked neither to spongiotrophoblast specification nor giant cell differentiation. Indeed, forming junctional complexes between GCs may be important for their function and Cx31 might play a role in this same cell–cell interaction or be required for their migration. Closer inspection of the GCs of the Cx31-deficient mouse might reveal answers to the role of this gap junction protein and GC.

GC Migration and Invasion Regulated by MMP-9 and Dcn

In order for GCs to invade, they require proteases to break down the surrounding extracellular matrix, allowing cell migration. MMPs are a good candidate for GC invasion. We have demonstrated for the first time spatial and temporal expression of MMP-9 in the mouse placenta, including expression by the GCs.

MMP-9 is likely to be involved in GC invasion, but it may also perform other roles, for example, in decidualization (Bany et al.,2000). MMP-9 expression in the decidua coincides with changes in decidual morphology around E14.5. MMP-9 from the GCs may contribute to this compositional change concurrent with remodeling of uterine arteries at the implantation site. During decidualization, MMPs are reported to be required for correct cell adhesion in the fetal and maternal layers of the mouse placenta (Solberg et al.,2003). Furthermore, inhibition of MMPs leads to malformation of hemi-desmosomes, affecting cell adhesion (Solberg et al.,2003). We speculate that expression of MMP-9 by the GCs may be required for cell adhesion in the decidua to provide extra support once the spiral arteries and blood flow are modified.

Adamson et al. suggest that remodeling of the spiral arteries occurs between E10.5 and E14.5, with much greater dilation after E12.5 (Adamson et al.,2002). Despite GCs surrounding the spiral arteries entering the implantation site (Pijnenborg et al.,1981; Adamson et al.,2002; Georgiades et al.,2002), Adamson et al. propose that these cells are not responsible for the degradation of their muscular wall, but instead, that a form of endovascular trophoblast cell is implicated (Adamson et al.,2002). We suggest that MMP-9 expressed by the GCs may play a role in the degradation of extracellular matrix surrounding the maternal arteries, facilitating subsequent dilation and contributing to increased blood flow into the placenta.

To prevent the invasion of GCs outside the implantation site, they must be inhibited. Another protein thought to be involved in regulating invasion is Dcn. This protein has been suggested to block human extravillous trophoblast (EVT) cell proliferation, migration, and invasion in a TGF-β–independent manner (Xu et al.,2002). Furthermore, there is some evidence that epidermal growth factor receptor may be associated with Dcn-mediated growth suppression (Moscatello et al.,1998; Patel et al.,1998). Human Dcn is found in the decidual ECM (Lysiak et al.,1995). This finding is probably required for Dcn to perform its anti-invasive function and thus impede excessive invasion of EVTs in to the decidua. Its collagen binding activity was proposed to contribute multifunctionally to matrix assembly and cell adhesion, migration, and proliferation (Mizuno et al.,2002). Dcn isolated from term villous tissue has also been suggested to have an antithrombotic role (Lysiak et al.,1995; Xu et al.,2002).

We have shown that decorin protein localizes to the murine decidua throughout gestation and to GCs and some spongiotrophoblast cells. In the mouse decidua, Dcn may provide a mechanism for preventing pathological invasion of GC into the mouse myometrium. The function of Dcn expressed by some spongiotrophoblasts and the GCs might well be analogous to that proposed for decorin extracted from villous tissue: an antithrombotic one. This function would be critical in a zone where blood percolates away from the Lz and back into the uterus. Furthermore, GCs clustering by maternal venous sinuses in the Jz and surrounding the maternal arteries may support a role of GCs in preventing thrombosis.

IGFR Is Expressed in a Short Period of Maximum Placental Remodeling

IGF2R is thought to be a negative regulator of IGF-II growth action (von Figura and Haslik,1986; Kirchhausen,2002). However, other signaling roles have been postulated relating to the stimulation of invasion of human extravillous trophoblasts by IGF-II binding of IGF2R (McKinnon et al.,2001). In the rat placenta, IGF2R was not found to be expressed in the Jz (Senior et al.,1990). However, our data suggest that IGF2R expression in the mouse placenta is distinct from the rat; maximum expression occurs in a window between E14.5 and E16.5, corresponding to the period of maximum placental growth (Coan et al.,2004). Both IGF2R and Igf2 are expressed by spongiotrophoblasts and GCs (Redline et al.,1993). The growth effects of IGF-II coming from the GCs and the Lz trophoblast are in part tightly regulated by expression of IGF2R by the Jz in a distinct developmental window. IGF2R expression may prevent further expansion of the Jz after E16.5. Migration of GCs away from the Jz may diminish the effects of IGF-II on the Jz. Furthermore, GC IGF-II may ultimately be destined for promoting labyrinthine growth in late gestation, whereas the Lz continues to expand to support the fetus (Coan et al.,2004).

IGF2R is also expressed by the endothelial cells in the Lz. The reason for this finding may be to regulate the effect of IGF-II, produced by the fetus, on the Lz. Up to E16.5 IGF-II probably has a profound affect on placental growth. Therefore, if fetal IGF-II were not negatively regulated by endothelial IGF2R, overgrowth of the labyrinthine trophoblast might take place, preventing correct remodeling of the interhemal membrane. This process may be the case in the mouse model deficient in IGF2R (Lau et al.,1994; Ludwig et al.,1996). Thus IGF2R produced by the Jz and the fetal endothelium may provide two temporal buffers for IGF-II–associated proliferation. Furthermore, down-regulation of IGF2R after E15.5 may be important in order that IGF-II fetal demand signals, demanding nutrients from the placenta, are not interfered with by IGF2R.

What Is the Ultimate Function of the GCs?

Where do the GCs go once they have migrated in to the decidua? We found a significant reduction in GC number in the Jz, without a concurrent increase in GCs that had invaded the decidua. Knowing what happens to these “lost” GCs would greatly increase our understanding as to their possible functions. One of the theories is that GCs are analogous to the human invasive extravillous trophoblast, that is, they invade the maternal spiral arteries leading to the implantation site and erode the muscular arterial walls. This process results in increased blood flow, oxygen, and nutrients to the implantation site. In the rat, GCs invade through the maternal decidua and may fuse with cells of the “metrial gland” (Davies and Glasser,1968; Croy,1999). While in the metrial gland, GCs could interact with the uterine natural killer cells (uNKs), stimulating uNKs to modify the spiral arteries (Croy et al.,2000). However, these hypotheses may be restricted to the rat, as mouse GC invasion is far more shallow and restricted to the decidua (Redline et al.,1993; Adamson et al.,2002). Alternatively, could GCs in the mouse be required for triggering parturition? Mann et al. demonstrated high levels of expression of cyclooxygenases (COX) from GCs in late gestation. These are oxidative stress markers indicative of hypoxia–reoxygenation injury, which may be part of the process of labor (Mann et al.,2003).

In conclusion, we can now begin to describe more confidently some of the elements involved in GC development and putative roles in placental function and fetal well-being: Early on around E12, a small distinct population of trophoblast, putative preglycogen cells, begins to accumulate glycogen, although their cytoplasms remain unvacuolated, after the embedding process. At the same time, p57Kip2 is expressed, preventing further proliferation of preglycogen cells and committing them to transform into GCs. P57Kip2 is then associated with cytoskeletal alterations to accommodate glycogen and allow the cells to migrate between the spongiotrophoblast to the decidua. The morphological changes that take place to the GCs at this stage result in the observed distinctive vacuolation of these fixed, embedded cells. Cx31 is also expressed and may promote the GC fate. A short while later, glycogen detected by PAS becomes available for breakdown into glucose, suggested from the presence of glucagon on the GC membranes. Both MMP-9 and Dcn are expressed shortly after the p57Kip2. MMP-9 may be one of several proteases required for GC invasion and may play roles in modifying the decidua or blood vessels entering the implantation site. Dcn may be required to prevent thrombosis in the Jz and decidua. Dcn in the decidua may also be required to regulate the extent of interstitial invasion by the GCs, which may be coordinated with the adhesion role of protocadherin12 (Bouillot et al.,2006). IGF2R expression in a short developmental window maybe critical for Jz and Lz development and may ultimately regulate placental size.

EXPERIMENTAL PROCEDURES

Animals

Timed matings of 6- to 8-week-old C57BL/6J mice (Harlan, UK) were carried out. Animals were housed under controlled laboratory conditions with a 12-hr light/dark photoperiod. Pregnant females were killed at gestational ages E12.5, E13.5, E14.5, E16.5, and E18.5 (morning of the vaginal plug E = 1.0). All experiments were carried out in accordance with the UK Government Home Office licensing procedures. One fetus and placenta was taken from each uterine mid-horn as previously described (Coan et al.,2004) from at least four litters per stage. Placentas were hemisected using a double-edged razor blade, each half weighed and immediately fixed.

Paraffin Embedding

Half-placentas were fixed by immersion in 4% paraformaldehyde (BDH) in PIPES buffer (0.1 M PIPES, 1.36 mM calcium chloride [CaCl2], 0.75 mM polyvinyl-pyrrolidone [PVP-40; Sigma Aldrich Co.], pH7.2) overnight. Tissue was dehydrated, embedded in paraffin wax, and exhaustively sectioned at either 7 μm or 20 μm using a rocking microtome (Rotary Microtome Cambridge Instruments, UK). Vertical sections selected for stereology were stained using a standard hematoxylin and eosin (H&E) protocol. For immunohistochemistry using IGF2R antiserum, placentas at E12.5, E14.5, E15.5, and E16.5 were fixed for 3 hr in Carnoy's fixative (60% C2H5OH, 30% CHCl3, 10% C2H5COOH), then transferred to absolute ethanol before embedding.

Resin Embedding

Half-placentas were fixed for 6 hr with 4% glutaraldehyde in PIPES buffer (4°C), washed in PIPES buffer, and post-fixed with 1% osmium ferricyanide for 1 hr. The post-fixed tissue was processed as previously described (Coan et al.,2004).

Estimating Placental Volumes

An Olympus BX-50 microscope with objective lenses (×1.25, ×10, ×20 and oil immersion ×40, ×60, ×100), a motorized specimen stage, and microcator was used in conjunction with the Computer Assisted Stereology Toolbox (CAST) version 2.0 program (Olympus, Denmark) to perform all measurements. The absolute volume of each placenta and its compartments (Lz, Jz, and decidua) were determined as previously described (Coan et al.,2004).

Estimating the Cellular Mass of the Jz and Decidua

Using the ×20 objective lens, randomly selected fields of view in the Jz and decidua basalis were observed. A point grid was superimposed, and interactions with the following components were recorded: cellular (spongiotrophoblasts, GCs, decidual cells grouped) and noncellular (maternal sinusoids and noncellular spaces). Volume fractions of cellular and noncellular components were converted to absolute values by referring to the combined absolute volume of the Jz and decidua basalis. The reference space used to estimate cell numbers was determined to be the total volume of Jz and decidua basalis minus the noncellular volume.

Direct Estimate of Cell Number in the Jz

Paraffin-embedded half-placentas corresponding to those used for estimating volumes were exhaustively sectioned at 20 μm, and four equidistant sections were randomly selected. The CAST program was used to select random fields of view under the ×100 objective lens. A counting frame occupying 25% of the screen was superimposed, and cell nuclei were counted in a 15-μm depth of the section (Fig. 1D). The 15-μm depth allows a random starting depth within the total 20-μm section depth. This sample volume is referred to as an “Optical brick” (Howard and Reed,1998). Nuclei were counted if they came into focus within the 15-μm depth, and were inside or touching the green borders of the counting frame; all other nuclei were excluded. The mean number of the different cell types per Optical brick gave a number-weighted mean number density. The densities for each cell type were referenced to the absolute cellular volume, calculated from the corresponding half-placenta that was sectioned at 7 μm of the Jz and decidua to determine cell number estimates.

Estimating Giant Cell Numbers

Due to the regular arrangement within the placenta of secondary giant cells, and to both their size and scarcity during late gestation, it was not possible to perform stereological analysis of these cells simultaneously with spongiotrophoblasts and GCs to obtain a mean giant cell number estimate. Therefore, the previously analyzed 20-μm-thick paraffin sections were used. Under a ×40 lens all giant cells whose nuclei were clearly resolved within the 20-μm thickness of the section were counted. The giant cell number density for each placenta was referenced back to the absolute cellular volume of Jz and decidua basalis to arrive at a giant cell number estimate per placenta.

Estimating Mean Spongiotrophoblast and GC Volume

Using resin-embedded placental halves, 1-μm-thick sections closest to the placental midline were cut and stained with toluidine blue from at least four litters per gestational age (see above). The point sampled intercept length was measured to determine cell volume (Fig. 1B). A line grid, with random orientation, was superimposed onto randomly selected fields of view, and the ×100 objective lens was used. Approximately 100 point-sampled intercept lengths (where a line traversed a cell, the distance from one side of the cell membrane to the other along the line's length was measured) were measured per cell type per placenta. The following equation was then used to estimate volume:

equation image

Where V̄ν is the mean volume-weighted volume and l̄03 is the mean of the cubed point-sampled intercept lengths (Gundersen and Jensen,1985; Jensen and Gundersen,1989).

Estimating Mean Giant Cell Volume

Due to the regular arrangement within the placenta of secondary giant cells and due to both their size and scarcity during late gestation, it was not possible to use random sampling and a line grid to measure point sampled intercepts lengths. Therefore, the Vertical Nucleator was used to estimate giant cell volumes to obtain a mean volume estimate (Fig. 1C). Using 1-μm-thick resin sections for at least four litters at gestational ages E12.5, E14.5, E16.5, and E18.5, all secondary giant cells whose nuclei were in the plane of section were analyzed. The CAST program used the following equation to estimate volume:

equation image

where l̄n refers to the distance from the random point within the giant cell to the giant cell membrane (this is distinct from the linear intercept length; Gundersen,1988).

Immunohistochemistry

Sections were dewaxed and rehydrated. Endogenous peroxidase activity was quenched using 3% H2O2 in tap water for 15 min. Slides were heated in a microwave pressure cooker containing 0.01 M sodium citrate buffer pH 6.0 to expose antigens. Sections were permeabilized with TBS containing 0.1% Triton X-100 (TBS-TT), 10 min. Antibodies were diluted in 2% bovine serum albumin containing 5% of the appropriate animal serum and incubated in humidified chambers at 4°C, for 20 hr. Negative controls had no primary antibody incubation step. For the IGF2R antibody, a further negative control was used by obtaining placentas from mice with the hairpin-tail mutation (Thp), where the chromosomal region containing Igf2r is missing. Sections were rinsed with TBS and washed in 0.5 L TBS-TT for 1 hr with stirring. Slides were incubated with diluted secondary peroxidase–conjugated antibodies for 1 hr in a humidified chamber, and then washed in 0.5 L TBS-TT. The ABC–peroxidase staining system (Vector Laboratories, Inc., Burlingame, CA) was used followed by 3,3-diaminobenzidine (DAB) in Tris-HCl pH 7.4. Sections were counterstained with hematoxylin. Antibodies were tested on at least three sections from different gestational ages or genotypes, and experiments were repeated on three occasions to confirm observations of protein expression detected. The following antibodies were purchased and used at the manufacturer's recommended dilution: Rabbit anti-connexin31 (Alpha Diagnostics), goat anti-decorin (R&D Systems), mouse anti-glucagon (R&D Systems), anti-Ki67, goat anti-kip2p57 (Santa Cruz), rabbit anti-matrix metalloproteinase9 (Abcam). Dr. C.D. Scott (Kolling Institute of Medical Research, Sydney, Australia) kindly provided a rabbit anti-IGF2R antibody (dilution 1:2,000).

Scoring was performed for at least three placentas for each antibody at each gestational age. Expression of a protein was indicated by DAB positivity (brown staining above background), and lack of expression, indicated by DAB negativity.

PAS Stain for Glycogen

PAS is a standard histological stain for identifying the presence of glycogen in tissues. Although PAS stains glycogen, other compounds including mucin will react with the Schiff reagent to produce a characteristic positive magenta color. We used salivary amylase-treated sections to establish that specific staining was due to glycogen positivity.

Sections were dewaxed and rehydrated before incubating with either TBS or a solution of fresh saliva and TBS 1:1, for 2 hr at 40°C. After incubating, slides were placed in 1% periodic acid in 40% ethanol for 10 min at room temperature, then 50% ethanol for 1 min followed by immersion in Schiff's reagent (77 mM C20H19N3.HCl, 13 mM K2S2O5, HCl 0.13 M−1) for 7 min. After this, slides were rinsed twice for 5 min at a time in sulfurous acid rinse (4.5 mM K2S2O5, HCl 0.048 M−1). Hematoxylin was used to stain nuclei and the remaining procedure was followed as for immunohistochemistry.

Transmission Electron Microscopy

Three mid-horn placentas from gestational ages E12.5 and E16.5 were dissected from pregnant mice and immediately immersed in fixative. Tissue was processed as described previously (Coan et al.,2005). Briefly, tissue was fixed in 4% glutaraldehyde and treated with 1% osmium ferricyanide (to retain glycogen). The post-fixed tissue was dehydrated and embedded in Spurr's epoxy resin (Taab, Berkshire, UK). Sections, 1 μm thick, close to the placental midline were stained with toluidine blue and used to identify regions of interest. Thin sections (50 nm) were stained with uranyl acetate and lead citrate and were viewed in a Philips CM100 at 80 kV.

Statistical Analysis

All statistical calculations were carried out using StatView version 5.0 (SAS Institute Inc.). A mean value of results was determined for each litter, and data were subsequently analyzed on a litter basis. For normal placental development, analysis of variance (ANOVA) with age as the dependent variable followed by Fisher protected least significant difference (PLSD) post hoc test was used to test for significant differences between the different gestational ages and to identify homogenous groups. For IGF-II mutants, with genotype as the dependent variable ANOVA was performed followed by Fisher's PLSD. Mean data are expressed ± SEM.

Acknowledgements

We thank Dr. Caroline Scott for the kind gift of the IGF2R antiserum and the Multi-Imaging Centre of the School of Biological Sciences, Cambridge. This work was funded by a studentship from the Anatomical Society of Great Britain and Ireland.

Ancillary