Paracrine Signals from the Mouse Conceptus Are Not Required for the Normal Progression of Decidualization

Abstract

The purpose of this study was to determine whether the conceptus directs the formation of a tight- and adherens-dependent permeability barrier formed by the primary decidual zone and normal progression of decidual cell differentiation during embryo implantation. Four artificial models of decidualization were used, some apparently more physiological than others. The results show that both the formation of the permeability barrier and decidual cell differentiation of three of the artificial models were quite different from that of pregnant uteri. One artificial model of decidualization, namely pseudopregnant animals receiving concanavalin A-coated Sepharose bead transfers on d 2.5 of pseudopregnancy, better recapitulated the decidual changes that occur in the pregnant uterus undergoing decidualization. This included the formation of a primary decidual zone-like permeability barrier and decidual growth. This model also exhibited similar temporal changes of the expression of genes involved in decidualization that are markers of decidual cell differentiation. Overall, the results of this study indicate that some models of inducing decidualization artificially produce responses that are more similar to those occurring in the pregnant uterus, whereas others are quite different. More importantly, the results suggest that concanavalin A-coated Sepharose beads can provide an equivalent stimulus as the trophectoderm to cause the formation of the primary decidual zone permeability barrier.

A primary decidual zone and progression of decidualization is similar between pregnant mice and some, but not all, artificial deciduoma models in the absence of signals from the conceptus.

Implantation of the conceptus begins with the attachment of the conceptus to the uterine wall and culminates in the formation of the definitive placenta (1). For implantation to begin, the conceptus must have developed to a specific stage and the uterus must have gone through hormone-specific changes to become receptive (2). The blastocyst stage of the mouse conceptus begins to undergo the process of implantation in the dark phase between d 3.5 and 4.5 of pregnancy (morning of vaginal plug equals d 0.5) between 2400 and 0600 h (3). The earliest macroscopically observable sign that accompanies the onset of implantation is increased endometrial vascular permeability adjacent to the implanting conceptus (4). This is followed by the onset of the transformation of the endometrial tissue into decidual tissue in implantation segments of the endometrium, a process commonly referred to as decidualization (5,6). It is within this endometrium undergoing decidualization that the semi-allogenic conceptus receives nutrients and an environment to develop properly for a number of days. Later, the major tissue providing this support during much of the latter half of pregnancy of rodents is the placenta.

Decidualization involves the differentiation of the endometrial fibroblast cells into epithelioid-like decidual cells. In humans, this begins near the end of the menstrual cycle, and if menstruation does not occur due to pregnancy, the process continues (7). In mice and rats, decidualization occurs only if the conceptus is present and provides an implantation stimulus. In mice, decidualization initially begins in several cell layers of the endometrial stroma surrounding the implantation chamber forming the primary decidual zone (PDZ) in the antimesometrial region. This PDZ contains a high level of intercellular tight, gap, and adherens junctions and is avascular and well developed by d 5.5 of pregnancy in mice (5,6,8). Thus, it provides a permeability barrier to extracellular macromolecules and cells and is believed to help protect the developing conceptus (9,10,11). However, as decidualization continues toward the myometrium in the antimesometrial region forming the secondary decidual zone and in the mesometrial region forming the mesometrial decidua, the PDZ disappears. By d 7.5 of pregnancy, the conceptus is surrounded by cells of the antimesometrial and mesometrial decidual tissues that are highly vascular. At this time, no permeability barrier exists, and maternal cells plus macromolecules can come into contact with cells of the conceptus. Therefore, the PDZ provides only a transient permeability barrier between the mother and conceptus.

The presence of an implanting conceptus, and thus paracrine signals from it, is not absolutely required for the endometrium of rodents to undergo decidualization. This is because decidualization can be artificially induced in rodents as initially demonstrated at the beginning of the 20th century (12,13,14). To distinguish it from the decidua, the tissue that forms in response to an artificial implantation stimulus is called the deciduoma. Since it was discovered, it has periodically been noted that differences exist in the precise timing and histological detail between the formation of the decidua and deciduoma (12,15,16,17). However, most researchers in the field appreciate that how dissimilar or similar it is to the formation of the decidua depends on which artificial model is employed. Currently, the most commonly used models use pregnant, pseudopregnant, or ovariectomized hormone-treated mice and the intraluminal injection of a small amount of oil as the deciduogenic stimulus. This results in the entire uterine horn undergoing decidualization, so essentially the entire uterine horn is an implantation site. Furthermore, the use of ovariectomized animals treated with steroids to sensitize the uterus to the artificial deciduogenic stimulus also suffers from the absence of specific hormonal changes that occur in response to cervical stimulation that normally occurs in pregnant and pseudopregnant mice (18,19,20,21).

Evidence suggests that paracrine signals from the conceptus influence decidualization of the endometrium in rodents. For example, recent evidence showed that a functional permeability barrier similar to that of PDZ of pregnant mice does not form in artificially induced deciduomas (17). However, transfer of trophoblast vesicles into these uteri do result in the formation of a functional PDZ. Other examples that the conceptus provides paracrine signals that influence decidualization comes from low-density membrane array (22) and high-density microarray (23) experiments. These show that the expression of a number of genes is different between the decidua and deciduoma during decidualization. To our knowledge, little work has been done recently to compare the more commonly used models to see whether some are better than others. Therefore, the present study was undertaken to compare the decidualization process between several deciduoma models and pregnancy. The data suggest that a more physiological deciduoma model better recapitulates the formation of the decidua, including the formation of the permeability barrier of the PDZ. These results suggest trophoblast cells are not required for the formation of a functional PDZ.

Materials and Methods

Unless otherwise indicated, reagents used in this study were purchased from Fisher Scientific (Pittsburgh, PA).

Animals and tissue preparation

All procedures involving mice were approved by the Southern Illinois University Animal Care and Use Committee. Experiments were carried out using CD1 mice as described previously (24). The morning that a vaginal plug was detected after the female was placed with intact and vasectomized males was considered d 0.5 of pregnancy and pseudopregnancy, respectively. Four different models of artificially induced decidualization were used as shown in Fig. 1. The first model, originally described elsewhere (25), involved the use of 0.05- to 0.1-mm-diameter concanavalin A (ConA)-coated agarose beads (Sigma Chemical Co., St. Louis, MO) as the deciduogenic stimulus (Fig. 1B). Twelve beads were transferred into the lumen between 1300 and 1500 h on d 2.5 of pseudopregnancy. This pseudopregnant bead-induced deciduoma (Pseudo-BID) model is rarely used but it provides a focal implantation stimulus unlike the other models. Three additional models used hormone-treated ovariectomized oil-induced deciduoma (Ovx-OID; Fig. 1C), pregnant (Preg-OID; Fig. 1D), or pseudopregnant (Pseudo-OID; Fig. 1E) mice that received an intraluminal injection of 10–15 μl sesame oil as the deciduogenic stimulus as previously described (24,26) (Fig. 1, C–E). The first two models are commonly used, and the intraluminal injection of oil seems to be the method of choice (26) but provides a global decidual response in the uterine horn. The artificial induction of decidualization in pregnant mice was carried out in one uterine horn, whereas in the other horn, normal embryo implantation was allowed to occur. As pointed out previously, the hormonal profile of pseudopregnant mice begins to differ from pregnancy beginning at 7–8 d due to the absence of developing embryos (26). Thus, the Preg-OID model was included in this study. Intravenous injections of Evans blue dye was used to identify sites of increased uterine vascular permeability as described previously (24).

Figure 1.

Timeline of the models of decidualization used in this study. A, Normal pregnant animals; B, Pseudo-BID model; C, Ovx-OID model; D and E, Preg-OID (D) and Pseudo-OID (E) models. Solid black areas indicate periods when the lights were off, and the open areas periods when the lights were on. Days indicated are days of pregnancy (DOP) and days of pseudopregnancy (DOPP) for pregnant and pseudopregnant mice. Days for the Ovx-OID model are equivalent days of pseudopregnancy.

Immunohistochemical staining

Tissues were collected and fixed as previously described (24). Tissue sections were prepared and stained for catenin-β1 (CTNNB1) using Vector blue substrate essentially as described elsewhere (27). However, antigen recovery was accomplished by subjecting sections to 0.2% trypsin in PBS at 37 C for 10 min. Antibodies used were anti-CTNNB1 (4 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA) and anti-ZO-1 IgG (6.25 μg/ml; Zymed Laboratories, San Francisco, CA). Sections were counterstained with nuclear fast red. For negative controls, the anti-CTNNB1 or anti-tight junction protein 1 (anti-TJP1) IgGs were replaced with rabbit IgG at the same concentration. No positive staining was observed (data not shown).

Measurement of decidual growth

Photomicrographs of sections were collected using a MZ-FLIII stereomicroscope (Leica, Wetzlar, Germany) equipped with a QImaging camera under control of Qcapture pro software (QImaging, Surrey, Canada). MacBiophotonics ImageJ software was used as instructed by the supplier to calibrate the photomicrographs and to measure endometrial cross-sectional areas (www.macbiophotonics.ca). The endometrial cross-sectional areas were measured from four to six independent samples at each time point collected. Data were analyzed using ANOVA followed by Duncan’s multiple range tests to determine differences between means at each time point using Sigmastat software (Systat Software Inc., San Jose, CA).

Permeability barrier measurements

Mice were injected iv with biotinylated BSA (0.2 mg; Sigma) in 0.2 ml PBS at 1 h before tissue collection on d 5.5 of pregnancy or pseudopregnancy. The tissues were collected and immersion fixed in 4% paraformaldehyde for 24 h and then embedded in paraffin. Sections (5 μm) were prepared and mounted onto Superfrost plus slides and stored until use. To localize the albumin, the sections were deparaffinized and hydrated and then subjected to antigen recovery using proteinase K (20 μg/ml) at 37 C for 30 min. After blocking in 2% goat serum in Tris-buffered saline (TBS) at room temperature for 30 min, the sections were incubated in TBS containing 2.5 μg/ml streptavidin-conjugated alkaline phosphatase for 30 min. After the sections were washed with TBS, they were incubated with TBS containing 0.6 mg/ml levamisole (Sigma) for 5 min to inhibit endogenous alkaline phosphatase activity. Finally, the sections were incubated with Vector blue substrate according to manufacturer’s instructions until a blue color developed and then counterstained in nuclear fast red. For negative controls, tissue sections were collected from animals that were not injected with biotinylated BSA and subjected to the same staining. This always resulted in no positive staining in the sections (data not shown).

Quantitative RT-PCR (qRT-PCR) analyses of steady-state mRNA levels

Total RNA was isolated from the uterine tissue using Trizol reagent (Invitrogen, Carlsbad, CA) as outlined by the manufacturer. After deoxyribonuclease digestion using RQ1 DNase (Promega, Madison, WI) as recommended by the manufacturer, the RNA was reextracted using Trizol. Reverse transcription was carried out using high-capacity reverse transcription kits as suggested by the manufacturer (Applied Biosystems, Foster City, CA). Real-time PCR was carried out using Bio-Rad (Hercules, CA) 2× SYBR Green Supermix as recommended by the manufacturer and a Bio-Rad CFX real-time PCR thermocyler. PCR conditions were 3 min at 94 C followed by 40 cycles of 94 C, 60–64 C (see Table 1) for 20 sec, and 72 C for 1 min for melting, annealing, and extension steps, respectively. The PCR primers (Table 1) were used at a final concentration of 400 nm. Melt-curve analyses verified the presence of single amplicons and no primer-dimer formation while sequencing of amplicons [University of Illinois, Urbana-Champaign (Urbana-Champaign, IL), Core DNA Sequencing Lab] verified target specificity. The data, obtained as cycle thresholds, were analyzed as previously described after normalization to 18 S rRNA levels (8). In all cases, the efficiencies of the PCR were linear between the mRNAs and 18 S rRNA and were greater than 85% even though the latter is highly abundant relative to the mRNA levels. Differences between mean levels of the steady-state mRNA levels were determined statistically as previously described (8).

Table 1.

Oligonucleotide primer sequences and annealing temperatures used for qRT-PCR

Target	Oligonuclotide (5′– 3′)	Annealing temperature (C)
Akp2
Sense	CTGACTGACCCTTCGCTCTC	64
Antisense	GTGGTCAATCCTGCCTCCT
Bmp2
Sense	AGATCTGTACCGCAGGCACT	64
Antisense	GTTCCTCCACGGCTTCTTC
Bmp8a
Sense	CATGACCGATGACGACGA	64
Antisense	CCAGTGTGGCTCCTGGTAG
Cebpb
Sense	AAGATGCGCAACCTGGAG	64
Antisense	CAGGGTGCTGAGCTCTCG
Fkbp5
Sense	AAACGAAGGAGCAACGGTAA	60
Antisense	TCAAATGTCCTTCCACCACA
Htra1
Sense	TCCGTTGATGCTGATGATG	64
Antisense	CATTGAAGTCATTCCTGACACC
Htra2
Sense	TGATGCTGACCCTGACTCC	60
Antisense	GTGCCAATTTCTCCCCAAT
Ifi202b
Sense	TGGCATCCTAGAGATCAATGAA	64
Antisense	TTGGGCACTTCAATAATTTGG
Il11ra1
Sense	AATACCGACCAGCACAGCAT	64
Antisense	TGACTCGTACCGCGTGTG
Il11ra2
Sense	CAAGTTCCGGTTGCAATACC	64
Antisense	TTATCACTTCCTCCAAGCCAAT
Irf8
Sense	GAGCCAGATCCTCCCTGACT	65
Antisense	GGCATATCCGGTCACCAGT
Isg12
Sense	GCTGCTACCAGGAGGACTCA	62.5
Antisense	CAATGCCTGTCCCAGTGAA
Isg15
Sense	GGGGACCAGTGTGCCTAA	64
Antisense	CCCCCATCATCTTTTATAACCA
Ptgs2
Sense	CACCTCTGCGATGCTCTTC	62
Antisense	TGGATTGGAACAGCAAGGAT
Procr
Sense	AATGCCTACAACCGGACTCG	64
Antisense	ACCAGTGATGTGTAAGAGCGA
Timp3
Sense	CACGGAAGCCTCTGAAAGTC	64
Antisense	TCCCACCTCTCCACAAAGTT
Wnt4
Sense	GCGTAGCCTTCTCACAGTCC	64
Antisense	CGCATGTGTGTCAAGATGG

Results

Model-dependent variation of artificial decidualization

Previous work has shown that adherens and tight junctions (5,9,10,11,16,28) and the proteins that form them such as TJP1, cadherin 1 (CDH1) (29), and CTNNB1 (27) are found at high levels in the PDZ or its equivalent in the deciduoma. To determine whether a PDZ forms in a similar fashion to pregnancy in the four models (Fig. 1), TJP1 and CTNNB1 were localized by immunohistochemistry in the pregnant animals or in pregnant, pseudopregnant, or ovariectomized hormone-treated mice undergoing artificially induced decidualization. On d 5.5, the PDZ showed strong staining in the endometrial stroma in the pregnant uterus in the area that surrounds the antimesometrial chamber containing the implanting conceptus (Fig. 2A). Staining was seen in the similar region on d 5.5 in uteri of Pseudo-BID and Ovx-OID mice, respectively (Fig. 2A). As shown before, little stromal staining could be seen in the uteri of Pseudo-OID mice on d 5.5 (data not shown). However, some staining for CTNNB1 could be seen on d 4.5 in the Pseudo-OID uteri (Fig. 2B). In a similar fashion, the uteri from Preg-OID mice did not exhibit much staining for TJP1 and CTNNB1 on d 5.5, but some could be seen a day earlier on d 4.5 (Fig. 2B).

Figure 2.

Decidualization in the various models used in this study. A and B, Representative photomicrographs (n = 3) of immunohistochemical localization of CTNNB1 and TJP1 in the endometrium during the early phase of decidualization in the decidua and deciduoma in mice on d 5.5 in pregnant, Pseudo-BID, and Ovx-OID models (A) or d 4.5 in Preg-OID and Pseudo-BID (B) models; C, representative photomicrographs of uterine cross-sections collected the morning of d 7.5; D, graph showing mean (±sem, n = 4–6) cross-sectional areas of endometrial tissue from d 4.5–8.5. Bars with different letters at each time point are significantly (P < 0.05) different. am, Antimesometrial side; b, bead; e, embryo. Scale bar in A, 100 μm for all panels in panels A and B.

Decidual cell differentiation in rodents involves both the proliferation and differentiation of endometrial stromal fibroblast cells into larger and polyploid or multinucleated decidual cells (5,6). Therefore, there is a great increase in tissue mass and size of the endometrium as decidualization proceeds, and this can be used as a measure of the progression of decidualization. Preliminary work showed that on the equivalent of d 7.5 of pregnancy, pseudopregnancy or its equivalent that the Pseudo-OID and Preg-OID models exhibited a greater amount of decidualization (Fig. 2C), indicating that the progression of decidualization was not normal in these models. Therefore, we analyzed the cross-sectional areas of uterine tissue for all models at various time points on d 4.5–8.5 (Fig. 2D) to see which artificial models best recapitulate changes in uterine growth associated with decidualization in pregnant uteri. The Pseudo-BID tissue looked the most identical to implantation sites in pregnant animals in three dimensions because of the focal nature of the deciduogenic stimulus. On the other hand, the use of oil as the artificial stimulus results in the decidualization of the entire uterine horn and was not focal in nature. On occasion, bumpy deciduomas resulted due to failure of injection of all the volume of oil into the lumen. Any oil-induced tissues exhibiting this were not included in these analyses. At each time point examined up to d 7.5, the mean endometrial cross-sectional areas of the Pseudo-BID and Ovx-OID sites were not significantly different (P > 0.05) to each other or to the pregnant animal endometrium undergoing normal decidualization. On d 8.5, Pseudo-BID tissue was not different from that of normal pregnant endometrium, but Ovx-OID tissue was significantly (P < 0.05) greater than both. At every time point examined, the mean cross-sectional areas of the Pseudo-OID and Preg-OID endometrial were significantly (P < 0.05) greater than the pregnant, Pseudo-BID, and Ovx-OID tissue. This difference between normal pregnant endometrial and Preg-OID plus Pseudo-BID appeared approximately 2-fold greater at each time point.

Functional PDZ-like formation in the deciduoma is model dependent

Overall, the Pseudo-BID model of obtaining decidualization better recapitulated changes, so the focus of the remaining experiments was to compare this model with that of normal pregnancies. First, we wanted to determine whether the beads cause an increase in vascular permeability on the morning of d 4.5 of pseudopregnancy in a similar fashion to pregnant animals (4,30). The increase in vascular permeability was seen in both the decidua (Fig. 3A) and Pseudo-BID (Fig. 3B), indicating the timing of the implantation stimulus of the latter is similar to normal pregnancy. Next, we tested the current hypothesis that a permeability barrier similar to the PDZ in pregnant mice is not formed in the deciduoma. Using an approach similar to previous studies on the existence of this barrier in rodents (9), we show that a permeability barrier to macromolecules does exist in the Pseudo-BID deciduomas on d 5.5. On d 5.5, abundant localization of the biotin-conjugated albumin tracer was seen in the endometrial tissue except for a region surrounding the implanting conceptus (Fig. 3, C and D) and bead (Fig. 3, E and F) in the pregnant and Pseudo-BID tissue.

Figure 3.

Comparison of the early uterine responses in pregnant and Pseudo-BID models. A and B, Representative photomicrograph (n = 3–6) of mouse uteri on the morning of d 4.5 of pregnant (A) and Pseudo-BID (B) mice showing an increase in vascular permeability (arrows pointing to blue sites) using the Evans blue dye test. C–F, Representative photomicrographs showing the mouse uterine distribution of iv injected biotinylated albumin in the pregnant decidua (C and D) and Pseudo-BID (E and F) on the morning of d 5.5 of pregnancy and pseudopregnancy, respectively. am, Antimesometrial side; b, bead; e, embryo; m, mesometrial side. Scale bars, 250 μm.

Assessment of decidualization-associated gene expression

Traditionally, one of the most used markers for decidual cell differentiation in mice and rats was the increase in alkaline phosphatase activity that accompanies decidualization (3,31,32,33,34,35). Recently, the expression of alkaline phosphatase 2 (Akp2) has been used as a marker of decidualization by measuring changes in mRNA and protein levels (36,37,38). As shown in Fig. 4A, steady-state Akp2 mRNA levels are significantly (P < 0.05) higher in implantation sites compared with nonimplantation segments of the mouse uterus on d 4.5–8.5 (Fig. 4A). Levels increased to high levels on d 6.5 and 7.5 but then decreased on d 8.5. To determine whether differences exist between the implantation and Pseudo-BID sites, the steady-state levels of Akp2 mRNA were measured on d 3.5–8.5, respectively (Fig. 4B). Although differences were not seen on the other days, Akp2 mRNA levels were greater on d 4.5 and less on d 7.5 in the decidua compared with deciduoma.

Figure 4.

Graphs showing the results of qRT-PCR analyses of the steady-state Akp2 (A and B), Bmp2 (C and D), Bmp8a (E and F), Wnt2 (G and H), Ptgs2 (I and J), Fkbp5 (K and L), and Cebpb (M and N) mRNA levels in the nonimplantation and implantation sites of the pregnant uterus (graphs on left) or implantation compared with Pseudo-BID sites (graphs on right) on d 4.5–8.5 of pregnancy and pseudopregnancy. Bars represent the mean (±sem, n = 3–4). *, Denotes significant (P < 0.05) difference in mRNA level between sites on a given day.

Because slight differences were seen in Akp2 mRNA levels, we also examined the mRNA levels of other genes known to be important in decidualization in the pregnant and Pseudo-BID mice. These included the transcripts of several bone morphogenetic proteins (BMPs) (39,40) and downstream BMP2-regulated genes (41,42) and CCAAT/enhancer-binding protein β (Cebpb) (43). Although no spatial information is provided by these data, significant differences in the levels of these mRNAs would indicate major differences in the progression of the decidual cell reaction. Steady-state Bmp2 (Fig. 4C) and Bmp8a (Fig. 4E) mRNA levels were significantly greater in implantation compared with nonimplantation sites on d 4.5–8.5. In a similar fashion to Akp2, levels of Bmp2 and Bmp8a mRNA increased to maximum levels on d 6.5 and 7.5. However, on d 8.5, only Bmp2 mRNA levels decreased, whereas those of Bmp8a remained elevated. Finally, the steady-state Bmp2 (Fig. 4D) and Bmp8a (Fig. 4F) mRNA levels did not differ between the pregnant implantation and Pseudo-BID sites undergoing decidualization on each of the days examined. Of the BMP2 downstream target genes, we found that Wnt4 (Fig. 4G), Ptgs2 (Fig. 4I), and Fkbp5 (Fig. 4K) mRNA levels were significantly greater in implantation compared with nonimplantation sites on d 4.5–8.5 of pregnancy. Furthermore, their levels were virtually all similar between the pregnant implantation sites and Pseudo-BID sites undergoing decidualization on each of the days examined (Fig. 4, H, J, and L). In a similar fashion to Akp2 (Fig. 4B), the exception was the significantly lower (by ∼25%) Wnt4 mRNA levels in the implantation compared with the Pseudo-BID sites on d 7.5 (Fig. 4G). As shown in Fig. 4M, steady-state levels of Cebpb mRNA are greater in implantation sites of the uterus on d 4.5–8.5 compared with that of the nonimplantation sites. On the other hand, the levels of this mRNA are not different on each of these days between the implantation and Pseudo-BID sites (Fig. 4N). Finally, we could not detect a difference in progesterone receptor mRNA levels on each day between the implantation sites and Pseudo-BID sites (data not shown).

IL-11 receptor-α knockout mice exhibit abnormal decidual cell differentiation and have abnormally small deciduas (44) due to decreased cyclin D3 and cyclin-dependent kinase inhibitor 1A expression (45). In an effort to explain the general tendency of larger deciduomal sizes of oil-induced compared with bead-induced deciduomas, we examined IL-11 receptor-α1 (Il11ra1) and -2 (Il11ra2) mRNA levels. As shown in Fig. 5A, steady-state Il11ra1 and Il11ra2 mRNA levels are similar between implantation sites from pregnant animals and bead-induced deciduoma sites of pseudopregnant mice on d 7.5. On the other hand, mRNA levels were significantly (P < 0.05) greater in the Ovx-OID deciduoma tissue compared with the pregnant and Ovx-OID tissue. As shown in Fig. 2, the size of the deciduoma of the Ovx-OID deciduoma sites is greater than the decidua of implantation sites on the next day.

Figure 5.

Graphs showing the results of qRT-PCR analyses of the steady-state Il11ra1/Il11ra2 (A), Htra1 (B), Htra2 (C), Procr (D), Timp3 (E), Isg15 (F), Irf8 (G), Ifi202b (H), and Isg12 (I) mRNA levels between the implantation site and Pseudo-BID plus Ovx-OID deciduomal tissues on d 7.5. Bars represent the mean (±sem, n = 4–5). Bars with different letters at each time point are significantly (P < 0.05) different.

A large number of genes have been identified whose expression differs between the decidua and oil-induced deciduoma (22,23,39,46,47). It is beyond the scope of the present study to go back and determine whether differential expression of all of these is also different between the decidua and Pseudo-BID deciduoma and localize the differences in such expression. However, we have analyzed the steady-state mRNA levels of a subset of these interesting genes to determine whether the differences also exist between the pregnant and pseudo-BID animals. Two proteinase genes (Htra1 and Htra2) and a proteinase receptor gene (Procr) are expressed at a higher levels in the decidua compared with Pseudo-OID deciduomas (23). However, as shown in Fig. 5B, Htra1 mRNA levels were not different between the pregnant implantation site compared with Pseudo-BID plus Ovx-OID site tissues on d 7.5. Htra2 levels, on the other hand, were significantly (P < 0.05) greater in the implantation site tissue (Fig. 5C). Procr mRNA levels were similar between pregnant implantation and Pseudo-BID site tissues, but levels were significantly (P < 0.05) lower in the Ovx-OID tissue (Fig. 5D). Tissue inhibitor of metalloproteinase-3 (Timp3) mRNA levels in the Pseudo-OID are greater than what is seen in normal implantation site tissue (23). As shown in Fig. 5E, Timp3 mRNA levels were significantly (P < 0.05) greater in the Pseudo-BID and Ovx-OID site compared with the implantation site tissues on d 7.5. Finally, several interferon-regulated genes such as Isg15 (22,23,46) plus Irf8, Ifi202b, and Isg12 (23) have been shown to be expressed at higher levels in the implantation sites of pregnant mice compared with oil-induced deciduoma sites. As shown in Fig. 5F, Isg15 mRNA levels were not different between the implantation sites and Pseudo-BID deciduoma sites. However, the results did verify differential expression between the pregnant implantation site and oil-induced deciduoma tissue. Irf8, Ifi202b, and Isg12 mRNA levels were significantly (P < 0.05) greater in the Pseudo-BID and Pseudo-OID deciduoma sites compared with implantation sites on d 7.5 (Fig. 5, G, H, and I).

Discussion

Since it was initially described (12,13,14,48), the artificially induced deciduoma model has become a useful tool to study the process of decidualization in rodents. Throughout the years, a great deal has been learned about the hormone requirements to obtain uteri receptive to a deciduogenic stimulus in that both estradiol and progesterone play key roles (30). However, unlike humans where decidualization begins in response to progesterone near the end of the menstrual cycle (7), decidualization will begin in rodents only in response to an implantation stimulus (1,5,6,49). Over the years, some studies have noted some differences between the decidua and deciduoma (12,15,17). However, during this time, the methods of deciduoma induction have changed dramatically. In the early days, quite traumatic stimuli were given such as a scratch or running a thread through the lumen or even electrical shock (12,13,14,33,50,51). However, as there was a refinement in knowledge about the precise timing and hormone requirements, less traumatic stimuli began to be used, and it was realized that a narrow window of receptivity to the less traumatic stimuli exists (30). Examples of less traumatic stimuli were luminal injections of oils, air, certain types of beads, and even saline (25,31,39,52,53). Currently, the most common artificial stimulus is the intraluminal injection of a small amount of sesame oil in pregnant, pseudopregnant, or hormonally treated ovariectomized animals (26). Another method that has not received as much attention involves the transfer of blastocyst sized ConA-coated Sepharose beads into pseudopregnant recipient mice in a similar fashion to live embryo transfer (25). As we demonstrated in this study, this last method has several advantages. Although it is technically more challenging, one advantage is that the ultimate implantation stimulus provided by the blastocyst-sized beads is focal in nature, which is similar to that provided by an implanting conceptus. This is in contrast to an injection of oil that results in decidualization of the entire uterine horn. Compared with the other models used in this study, another advantage of the Pseudo-BID model was that the progression of decidual cell reaction in response to the beads is similar to that of normal pregnancy.

The PDZ plays a key role in protecting the conceptus during the early phase of implantation (9,11). Recent evidence showed that a functional PDZ does not form in pseudopregnant mice that receive an intraluminal injection of oil as a deciduogenic stimulus (39). Indeed, this study also showed there is a lack or very low level of proteins involved in tight and adherens junctions in the endometrial tissue surrounding the oil-induced deciduomas in pseudopregnant and pregnant mice. However, to our surprise, the expression of TJP1 and CTNNB1 seen in the other two models used in this study were similar to that of normal pregnant mice. Furthermore, it was shown that a permeability barrier does form around the implantation stimulus in a similar fashion to the pregnant uterus for the Pseudo-BID tissue. Therefore, this study shows that the formation of a PDZ-like tissue during artificially induced decidualization is possible, but this is highly model dependent. This finding strongly supports the hypothesis that paracrine factors from cells of the conceptus are not absolutely required for the formation of a functional PDZ.

As the process of decidualization spreads throughout much of the remainder of the endometrium, there is a continued proliferation and differentiation of the mouse endometrial stromal fibroblast-like cells into decidual cells. A rapid enlargement of the endometrium accompanies this, and using this as a measure, other studies have noted that decidualization may proceed at a faster rate for some models of artificial decidualization (12,23). This study showed that decidual growth was advanced at all time points examined for the pseudopregnant and pregnant oil-induced deciduoma models, so these two models may least resemble the progression of decidualization seen in normal pregnancy. There is reason to believe that the differences for these two models may have been due to an earlier onset of decidualization due to the timing of application of the artificial stimulus because the size of the endometrium was almost double that of the pregnant uterus already on the morning of d 4.5. The third model used, the hormone-treated ovariectomized mice given oil as an artificial stimulus, exhibited fewer differences from pregnancy than the other models using the same artificial stimulus. However, this model suffers from the fact that these mice do not receive cervical stimulation that normally causes changes in prolactin secretion by the pituitary in rodents (18,19,20,21). Furthermore, as shown in this study, local IL-11 receptor-α expression is abnormal in the Ovx-OID tissue, so both the endocrine and local paracrine status of these animals is not normal. The final model used was the pseudopregnant mouse where ConA-coated Sepharose beads were transferred into the uteri on d 2.5 and was originally developed over 15 yr ago (25). Of the models tested, this last model best recapitulated the decidual growth seen in the pregnant uterus undergoing normal decidualization. Therefore, in terms of the progression of decidual growth, this final model may be the superior of the ones tested in this study. Overall, these results show that not all deciduoma models result in the similar decidual development compared with that of normal pregnancy.

Although the results of this study suggest that normal decidual growth may not require paracrine signals from the conceptus, evaluation of the expression of genes involved in decidualization still challenge this concept. Previous work has shown that Bmp2 expression, endometrial stromal cell proliferation, and decidual growth in the pseudopregnant oil-induced deciduoma formation are greater compared with normal pregnancy (23). This study supports this finding as seen by the greater sizes of the deciduomas from similar mice. Although the results of this study suggest that deciduomal growth for the Pseudo-BID tissue is similar to that of the decidua, a difference was seen for Akp2 (but not Bmp2) expression on d 7.5. The reason for this difference in Akp2 expression remains to be determined, but the lack of difference in Bmp2 and Cebpb expression suggests that the Pseudo-BID model is better than the Pseudo-OID model. Because expression of these genes in the uterus play a key role in decidualization (41,43), it was important to determine whether downstream targets of BMP2 were normal in the bead-induced deciduomas. Although the expression of other targets looked at in this study were mostly similar, Wnt4 expression in the bead-induced deciduoma was different from the decidua only on d 7.5 of pregnancy. Therefore, expression of at least one downstream target of BMP2 is not similar between the pregnant and Pseudo-BID tissues.

Recent work suggests that paracrine factors from the conceptus may indirectly or directly regulate changes in gene expression in the uterus during decidualization in rodents. Most of this work has come from comparing gene expression in the decidua compared with deciduoma using the pseudopregnant or hormone-treated ovariectomized mice with oil as an artificial stimulus (8,22,23,46,47). However, this study shows that not all models of artificially induced decidualization closely recapitulate that of the formation of the decidua. In addition, genes previously shown to be differentially expressed between the decidua and oil-induced deciduoma were found not to be differentially expressed between the decidua and bead-induced deciduomas. However, others were confirmed to be differentially expressed, supporting the hypothesis that the conceptus somehow may influence the expression of some genes in the endometrium during decidualization. The results of this study show that the pseudopregnant bead-induced deciduoma model employed in the present study might be a more preferable artificial model for further research in this area in the future.