Overexpression of Cyclin D3 Improves Decidualization Defects in Hoxa-10^−/− Mice

Abstract

Uterine decidualization, a crucial process for implantation, is a tightly regulated process encompassing proliferation, differentiation, and polyploidization of uterine stromal cells. Hoxa (Homeobox A)-10, a homeobox transcription factor, is highly expressed in decidualizing stromal cells. Targeted gene deletion experiments have demonstrated marked infertility resulting from severely compromised decidualization in Hoxa-10^−/− mice. However, the underlying mechanism by which Hoxa-10 regulates stromal cell differentiation remains poorly understood. Cyclin D3, a G₁ phase cell-cycle regulatory protein involved in stromal cell proliferation and decidualization, is significantly reduced in Hoxa-10^−/− mice. The expression of cyclin D3 in the pregnant mouse uterus parallels stromal cell decidualization. Here, we show that adenovirus-driven cyclin D3 replacement in Hoxa-10^−/− mice improves stromal cell decidualization. To address our question of whether cyclin D3 replacement in Hoxa-10^−/− mice can improve decidualization, both in vitro and in vivo studies were completed after the addition of cyclin D3 or empty (control) viral vectors. Immunostaining demonstrated increased proliferation and decidualization in both in vitro and in vivo studies, and in situ hybridization confirmed increased expression of decidualization markers in vivo. Placentation was demonstrated as well in vivo in the cyclin D3-replaced animals. However, fertility was not restored in Hoxa-10^−/− mice after d 10 of pregnancy. Finally, we identified several downstream targets of cyclin D3 during decidualization in vitro via proteomics experiments, and these were confirmed using in situ hybridization in vivo. Collectively, these results demonstrate that cyclin D3 expression influences a host of genes involved in decidualization and can improve decidualization in Hoxa-10^−/− mice.

Embryo implantation is a complicated process that requires two-way signaling and interaction between the implanting blastocyst and the receptive uterus, involving several growth regulatory mechanisms (1, 2). Any aberration during this process can result in infertility, which may account for a proportion of couples that are unable to conceive. Although in vitro fertilization and embryo-transfer techniques have overcome some human fertility challenges, the implantation rate after embryo-transfer remains disappointingly low, which is likely a result of poor understanding of uterine receptivity and embryo interaction during implantation (3). The mouse model has provided valuable information for understanding the implantation process.

In mice, the implantation process is first characterized by increased uterine vasculature permeability at the blastocyst site during the attachment reaction occurring around midnight on d 4 (d 1, plug positive). This reaction leads to significant localized cell-cycle activity in the endometrial stromal bed around the implanting blastocyst. The decidualization process, initiated by blastocyst implantation, is characterized by stromal cell proliferation and differentiation, including terminal differentiation, and allows for transition into a specialized type of cells (decidual cells) with polyploidy formation. The primary decidual zone (PDZ), avascular and epithelioid in nature (4), forms adjacent to the implanting blastocyst followed by formation of the secondary decidual zone (SDZ) on d 6, where polyploid decidual cells are present (5). The SDZ is fully developed by d 7, whereas the PDZ degenerates by d 8. Placental and embryonic growth begin to slowly replace the SDZ after d 8. Several signaling molecules, including homeobox transcription factors and cell-cycle regulatory proteins, are involved in the progression of the decidualization process (6–8).

Hoxa-10 is a member of the homeobox (Hox) multigene family of transcription factors (9). Four Hox clusters (a–d) have been identified in mammals (10) and are involved in both early (genes at 3′-end cluster) and late (genes at 5′-end cluster) embryogenesis (10, 11). Abdominal B (AdbB) is the most 5′-end gene in the Drosophila homeotic complex, and several AdbB-like genes in mammals have been identified at the 5′-end of Hox a, c, and d clusters. Hoxa-10 is an AdbB-like gene expressed during genitourinary development as well as during adult pregnancy. Targeting gene deletion experiments (9) have demonstrated that Hoxa-10^−/− mice have partial oviduct transformation of the proximal uterus. However, female infertility in these mice has resulted primarily from defective implantation (failed attachment reaction) as well as aberrant decidualization (12, 13). Hoxa-10 is regulated in the uterus by progesterone (P₄) during early pregnancy (6), and expression is associated with stromal cell proliferation (9, 14, 15). Hoxa-10 expression increases in human endometrial epithelium and stromal cells during the midluteal phase of the menstrual cycle and coincides with the window of implantation, suggesting an important role for Hoxa-10 in human implantation (16, 17).

Previous studies have demonstrated that only a small number (∼40%) of Hoxa-10 null mice can initiate the implantation reaction, implicating that the structural uterine changes in these mice do not completely inhibit implantation (12). This is consistent with the finding that uterine epithelium of Hoxa-10^−/− demonstrate normal expression patterns of several implantation associated genes, including leukemia inhibitory factor, heparin-binding-epidermal growth factor, amphiregulin, and cyclooxygenase-1 (6). Hoxa-10^−/− mice also demonstrate normal epithelial proliferation in response to estradiol-17β (E₂) but have severely compromised stromal cell proliferation in response to P₄ and E₂. In addition, Hoxa-10 null mice have shown defects in decidualization with the loss of regional development of the decidual bed (mesometrial and antimesometrial) and uterine NK cell differentiation in the decidual bed (18). These results suggest that impaired decidualization is a key factor in infertility in this mouse model.

Hoxa-10 influences a host of uterine genes involved in implantation under the control of P₄ (19). We have previously reported that cyclin D3 expression is up-regulated at the time of implantation, and its expression correlates to stromal cell proliferation and differentiation during decidualization (5, 20). Cyclin D3 expression is also significantly reduced in Hoxa-10^−/− uteri at the site of implantation or after experimental stimulation for decidualization (18, 20). Together, these results suggest that impaired stromal cell proliferation and differentiation under the direction of cyclin D3 is a potential cause for aberrant decidualization, resulting in infertility in the Hoxa-10^−/− mouse. Here, we provide evidence that replacement of cyclin D3 can reverse defects in uterine stromal proliferation and decidualization in the Hoxa-10^−/− mouse.

Materials and Methods

Animals and tissue preparation

Wild-type (WT) and Hoxa-10^−/− mice on a 129/SvJ/C57BL6 background were housed in the animal care facility at Cincinnati Children's Hospital Medical Center in accordance with the National Institutes of Health and institutional guidelines for the use of laboratory animals. All protocols for the present study were reviewed and approved by the Institutional Animal Care and Use Committee (approval no. 1D05042). Hoxa-10^−/− mice were generated as previously described (9) and originally provided by Richard L. Mass (Harvard Medical School, Boston, MA). PCR analysis of tail genomic DNA was used to genotype mice. Adult females (8–10 wk of age) were mated with fertile or vasectomized males of the same strain to induce pregnancy or pseudopregnancy (d 1, vaginal plug), respectively. Mice were killed in the morning at 0900 h on each day of pregnancy or pseudopregnancy. Implantation sites (ISs) were identified grossly on d 8–12 of pregnancy and collected, recording uterine weight (grams) at the time of collection. Tissues were rapidly flash frozen and kept at −80 C for subsequent analysis or fixed in 10% formalin for paraffin blocks.

Primary culture of Hoxa-10^−/− uterine stromal cells

Isolation and culture of uterine stromal cells from Hoxa-10^−/− mice collected on d 4 of pseudopregnancy was completed as previously described with slight modifications (21). Briefly, uterine horns were isolated from surrounding fat tissues, dissected open longitudinally, and cut into small pieces (2–3 mm). Uterine pieces were put into a sterile Petri dish and washed thoroughly with Hanks' balanced salt solution (HBSS) without Ca²⁺/Mg²⁺ and phenol red but containing 100 U/ml penicillin (Life Technologies, Inc.-BRL, Carlsbad, CA), 100 μg/ml streptomycin (Life Technologies, Inc.-BRL), and 2.5 μg/ml amphotericin B (Sigma, St. Louis, MO). Tissues were then placed in 3 ml of fresh medium (HBSS + antibiotic) containing 6 mg/ml dispase (Life Technologies, Inc.-RBL) and 25 mg/ml pancreatin (Sigma), then incubated sequentially for 1 h at 4 C, 1 h at room temperature, and then 10 min at 37 C. After completion of digestion steps, tissues were diluted in HBSS (17 ml) containing 10% charcoal-stripped fetal bovine serum and mixed methodically to release the sheet of luminal epithelial cells by gentle repeated pipetting (25-ml pipette). The tissues remaining after the digestion were washed twice in fresh medium (HBSS + antibiotic) and incubated again for subsequent digestion in a fresh medium (3 ml) containing 0.5 mg/ml, collagenase at 37 C for 30 min. Digested cells (primarily composed of stromal cells) were passed through a 70-μm nylon filter to remove excess epithelial cells. The cells were then centrifuged and the pellet washed twice with HBSS before initiation of primary culture. Cells were plated at 1–5 × 10⁵ per 25-cm² dish containing phenol red-free DMEM and Ham's F-12 nutrient mixture (1:1) with 10% charcoal-stripped serum and antibiotic. After incubation for 1 h, the cells adherent to the culture dishes were transferred to fresh medium (DMEM:F-12, 1:1) containing 1% charcoal-stripped fetal bovine serum. Decidualization conditioning medium [DMEM-F-12 1:1 containing E₂ (10 nm) and P₄ (1 μm)] was applied to cells 24 h after adenoviral infection.

Construction of recombinant adenoviral vectors for cyclin D3 cDNA sequence and generation of virus particles

Replication-defective adenoviral vectors were generated as previously described (22). PCR was used to generate the full-length coding region of mouse cyclin D3 cDNA from an expression clone obtained from Charles Sherr (23). PCR primers carrying linkers for XhoI at 5′-ends were as follows: 5′-GGCTCGAGATGGAGCTGCTGTGTTGCGA-3′ (sense) and 5′-GGCTCGAGCTACAGGTGAATGGCTGTGA-3′ (antisense). The amplified DNA fragment was cloned into XhoI site of the shuttle vector, pAdTrack-CMV. The clone constructs were sequenced to confirm identity. The resultant plasmid or empty shuttle vector [expresses only green fluorescence protein (GFP)] were linearized with PmeI and cotransfected with pAdEasy-1 into Escherichia coli Bj5183. The recombinant clones, either cyclin D3 (rAd-CycD3) or empty insert (rAd-GFP), were analyzed by restriction cutting using Pac I. The viral packing of these plasmids was carried out after transfection into 293 cells as described (22). Purification of viral particles was completed through CsCl density gradient centrifugation and stored at −80 C.

Adenoviral infection of Hoxa-10^−/− uterine stromal cell culture

Primary uterine stromal cell culture cells grown to a level of 40–50% confluence in four-well-chambered slides were subjected to adenoviral infection with rAd-CycD3 or rAd-GFP at 10 times multiplication of infection (MOI). Infection level was confirmed by GFP expression with fluorescence microscopy.

Adenoviral infection of the Hoxa-10^−/− mice

Approximately 100–200 μl of viral solution in saline (1 × 10¹¹ virus particles) for rAd-CycD3 or rAd-GFP were injected per Hoxa-10 null mouse iv (tail vein) two times on d 5 of pregnancy at 0900 and 1800 h to achieve successful adenoviral infection in vivo. This injection schedule was used for animals killed on d 8, 10, and 12. Adenoviral injection was completed on d 5 for two reasons: 1) cyclin D3 expression increases in the mouse uterus on d 5 morning in parallel with the decidualizing process and 2) adenoviral injection before blastocyst attachment (d 4 midnight) and uterine vascular permeability results in lack of delivery of the adenoviruses to the uterus.

Western blot analysis

Western blotting was performed as previously described (5). Briefly, tissue proteins (50 μg) were run on 10% SDS-PAGE gels under reducing conditions and transferred into Immobilon membranes. Membranes were incubated overnight with primary antibody in 5% milk in Tris-buffered saline with Tween 20 in 4 C and then secondary antibody coupled to horseradish peroxidase (1:5000) in 5% milk at room temperature for 1 h. Positive signals were detected by enhanced chemiluminescence kit.

Immunohistochemical staining

Immunohistochemical staining was completed according to methods previously described (5). Staining reaction was performed using a Histostain-SP kit (Invitrogen Laboratories, Grand Island, NY).

Hybridization probes

Mouse specific cDNA clones (in pCRIITOPO vector) were generated by RT-PCR for Ang2, Bmp2, Fkbp52, Ptgs2, Tdo2, Anxa1, Cops5, Gsn, Hspa4, Prdx2, Prdx6, and Tald01. The authenticity for each of these clones was confirmed by nucleotide sequencing. For in situ hybridization, sense and antisense ³⁵S-labeled cRNA probes were generated.

In situ hybridization

Frozen sections (10 μm) were hybridized as previously described (24) with ³⁵S-labeled cRNA probes. Sections hybridized with sense probes served as negative controls and showed no positive signals.

Quantitation of mRNAs by real-time PCR

Expression levels of Bmp2, Ptgs2, and Ang2 mRNAs in the cultured stromal cells were quantitatively analyzed by real-time PCR using the ABI Prism 7700 Sequence Detector System according to the manufacturer's instructions (Applied Biosystems, Foster City, CA). All mRNA quantities were normalized against Gapdh mRNA.

Proteomic evaluation

Primary Hoxa-10 null uterine stromal cells in culture after infection with rAd-CycD3 or rAd-GFP virus particles were subjected to decidualization as above for 24 or 72 h. Cell extracts were then analyzed by two-dimensional fluorescence difference gel electrophoresis. Difference gel electrophoresis analysis, protein identification by mass spectrometry, and database interrogation have been previously described by us (25). All samples were run in triplicate.

Results

Cyclin D3 enhances Hoxa-10^−/− uterine stromal cell growth, polyploidy, and binucleation in primary stromal cell culture

Because cyclin D3 expression is significantly reduced in Hoxa-10^−/− mice and its expression is linked to stromal cell proliferation, differentiation, and polyploidization, we asked whether overexpression of this G₁ cell-cycle regulator may improve decidualization in this knockout model in vitro. We used adenoviral gene delivery system to enhance the expression of cyclin D3 in cultured Hoxa-10^−/− stromal cells. The delivery system used recombinant adenovirus carrying a cDNA for complete coding sequence of cyclin D3 and GFP reporter genes, which are regulated by two independent cytomegalovirus early promoters (rAd-CycD3). Empty adenovirus expressing only GFP controlled by the same promoter was used as a control vector (rAd-GFP). Isolated stromal cells at 40–50% confluency were infected at 10 MOI with adenoviruses. Previously, we reported that infection of stromal cells was achieved at 90–95% efficiency as demonstrated by fluorescence microscopy evaluation of GFP and immunocytochemistry staining with cyclin D3 (26).

Here, we demonstrate that adenovirus-driven overexpression of cyclin D3 in Hoxa-10^−/− primary stromal cell culture results in a significant increase of proliferation at 24 h compared with GFP control (∼8-fold over control, P < 0.001, Student's t test) (Fig. 1A). At 72 h, rAd-CycD3-cultured cells had more polyploidy formation than control; specifically when quantified, rAd-CycD3 stromal culture has a significantly higher number of binucleated cells when compared with control (∼3-fold over control, P < 0.001, Student's t test) (Fig. 1, B and C). Using the primary antibody to cyclin D3 for subsequent analysis by Western blotting, we confirmed expression of cyclin D3 in rAd-CycD3-infected cells as well as reduced expression of cyclin D3 in rAd-GFP-infected cells (Fig. 1D). The overexpression of cyclin D3 also caused induction of expression for several decidualization markers, such as Bmp2, Ptgs2, and Ang2 mRNAs, and these results are comparable with nontransfected WT decidual cells in vitro (Fig. 1E). Overall, these results suggest that adenovirus-driven cyclin D3 overexpression can induce stromal cell proliferation, differentiation, and polyploidization in Hoxa-10^−/− mice in vitro.

Fig. 1.

Analysis of overexpressed cyclin D3 Hoxa-10^−/− primary stromal cell culture vs. control. Primary stromal cells grown to a level of 40–50% confluency were subjected to infection by adenovirus (at 10 MOI) carrying either sense cyclin D3 or GFP-only (control) constructs. At 24 h (A) and 72 h (B) after treatment with decidualization conditioned medium, stromal cells from rAd-CycD3 and rAd-GFP (72 h) (data not shown) were assessed using bright-field and fluorescence microscopy for GFP. Arrows show an example of binucleation. C, Percent of cells with binucleation were quantified and compared between rAd-CycD3, rAd-GFP, and WT in seven separate samples. *, P < 0.001. D, Analysis of expression by Western blotting (WB) using primary antibodies to cyclin D3 and actin (as control) at 24 h. *, Values are statistically different (P < 0.05, ANOVA followed by Newman-Keul's multiple range test) against control group. These experiments were repeated four times with independent samples and a representative blot is presented. E, Analysis of expression for decidualization markers (Bmp2, Ptgs2, and Ang2) by real-time RT-PCR against a constitutive gene (Gapdh) control. *, Values are statistically different (P < 0.05, ANOVA followed by Newman-Keul's multiple range test) against control group.

Overexpression of cyclin D3 enhances decidualization in Hoxa-10^−/− uteri on d 8 of pregnancy

As demonstrated above, we have established that adenoviral-driven cyclin D3 overexpression is effective to enhance stromal cell proliferation and polyploidy formation in Hoxa-10^−/− stromal cells in vitro. We next asked whether cyclin D3 overexpression via adenoviral infection is also effective in vivo during blastocyst-driven stromal cell decidualization. To examine the viral vector effects in vivo, Hoxa-10^−/− pregnant mice received cyclin D3 (n = 8) or control (n = 7) viruses (1 × 10¹¹ virus particles per mouse) iv by tail vein injection twice on d 5 of pregnancy at 0900 and 1800 h. Pregnant WT littermates (n = 4) without virus injections were also examined for comparison. Mice were killed on d 8 of pregnancy, and we recorded the weight and number of ISs. ISs were either flash frozen or fixed in 10% formalin for paraffin blocks. As shown in Fig. 2, A–C, rAd-CycD3 mice had both increased weight and number of ISs compared with control. Moreover, when IS weight and number were compared, rAd-Cyc3 mice had a significant increase in weight of the IS, compared with rAd-GFP (P < 0.001, ANOVA) (Fig. 2B) and increased number of IS (P < 0.001, ANOVA) (Fig. 2C). Although when compared with WT mice, both Hoxa-10^−/− groups had significantly decreased weight of the IS [P < 0.001 (rAd-GFP) and P < 0.05 (rAd-CycD3)] (Fig. 2B), but only rAd-GFP mice had a significantly reduced number of IS than WT (P < 0.001) (Fig. 2C). Again, using the primary antibody to cyclin D3 for Western blotting, we confirmed increased expression of cyclin D3 within the IS rAd-CycD3-treated mice and reduced expression in rAd-GFP-treated mice (Fig. 2D).

Fig. 2.

In vivo overexpression of cyclin D3 in Hoxa-10^−/− uterus during postimplantation period enhances decidualization at the site of implantation at d 8 of pregnancy. A, Representative photographs of d-8 IS in rAd-GFP- or rAd-CycD3-injected Hoxa-10^−/− and noninjected WT mice. B and C, Weight and number of ISs in rAd-GFP- or rAd-CycD3-injected Hoxa-10 null and noninjected WT mice. D, Analysis of expression by Western blotting using primary antibodies to cyclin D3 and actin (as control) in d-8 IS of rAd-GFP- or rAd-CycD3-injected Hoxa-10 null mice and noninjected WT mice. *, Values are statistically different (P < 0.05, ANOVA followed by Newman-Keul's multiple range test) against control group. These experiments were repeated four times with independent samples, and a representative blot is presented. ISs were pooled from adenovirus-injected animals to obtain results. E, Hematoxylin and eosin staining (at ×40) and immunostaining for GFP (×200 with the exception of insets at ×20) of paraffin-embedded d-8 IS of rAd-GFP, rAd-CycD3, or WT mice. M, Mesometrial pole; AM, antimesometrial pole; e, embryo.

Next, we examined paraffin sections of the IS by histological staining (Fig. 2E) to assess IS morphology. The IS of rAd-CycD3 mice demonstrated embryo retention, which was absent in rAd-GFP sections. However, embryo development in the IS of rAd-CycD3 mice appeared retarded compared with WT, and embryo development occurred centrally within the IS rather than at the antimesometrial pole as seen in WT IS. Immunohistochemical staining to GFP was used to demonstrate the distribution pattern of adenoviral vector within the IS. Immunostaining demonstrated that in Hoxa-10^−/− mice the distribution of GFP occurred evenly throughout the IS both at the mesometrial and antimesometrial poles, and no staining was demonstrated in untreated WT mice (Fig. 2E). IS paraffin sections were then evaluated with immunostaining for the determination of immunoreactive cell number for proliferation markers, Ki-67 (antigen Ki-67) and phosphorylated histone H3 (phH3). We observed a significantly increased amount of positively stained cells per area for Ki-67 in rAd-CycD3 mice when compared with rAd-GFP mice (P < 0.001, ANOVA) (Fig. 3A) and WT mice (P < 0.05) (Fig. 3A), but staining was not significantly different for phH3 staining between any group (P = not significant) (Fig. 3B). Moreover, immunopositive cells/area labeled with cyclin D3 (Fig. 3C) and Nsbp1 (nucleosome binding protein 1) (Fig. 3D) intended to assess for decidualization and polyploidy were significantly increased in both rAd-CycD3 (P < 0.001 and P < 0.05, respectively) and WT ISs (P < 0.05 and P < 0.001, respectively) compared with rAd-GFP ISs. Cyclin D3 staining per area was also increased in rAd-CycD3 mice vs. WT (P < 0.05).

Fig. 3.

Regional distribution of proliferation and polyploidy markers during decidualization at the site of implantation. Immunohistochemical analysis as described in Materials and Methods. Localization of Ki-67, phH3, cyclin D3, and Nsbp1 are shown in cross-sections of d-8 IS at ×40. M, Mesometrial pole; AM, antimesometrial pole; e, embryo. Graphical depiction comparing number of positively stained cells per area of ISs for Ki-67, phH3, cyclin D3, and Nsbp1. Results are positive staining per area (mean ± sem). *, P < 0.05; †, P < 0.001.

Subsequently, IS frozen sections were evaluated by in situ hybridization to confirm findings of immunostaining. Bmp2, Ptgs2, Ang2, and Fkbp52 appeared to have increased expression in both rAd-CycD3 and WT IS when compared with rAd-GFP IS. Expression density of Td02 did not appear different among the three groups. However, the regional distribution pattern of expression for Bmp2, Ptgs2, Ang2, Fkbp52, and Td02 was preserved more within the rAd-CycD3 mice than rAd-GFP mice when compared with WT IS (Fig. 4). Taken together, these results demonstrate that adenoviral-driven cyclin D3 overexpression can improve decidualization at d 8 of pregnancy in Hoxa-10^−/− mice.

Fig. 4.

Regional distribution of decidualization genes at the site of implantation on d 8 of pregnancy. In situ hybridization. Expression of Ptgs2, Ang2, Bmp2, Fkbp52, and Td02 genes at the sites of embryo implantation on d 8 of pregnancy is shown in rAd-GFP, rAd-CycD3, and WT mice. Frozen sections were hybridized with ³⁵S-labeled antisense or sense riboprobes and ribonuclease-resistant hybrids detected by autoradiography. Sections were poststained with hematoxylin and eosin. Dark-field photomicrographs of representative cross-sections hybridized with antisense probes are shown at ×40. M, Mesometrial pole; AM, antimesometrial pole; e, embryo.

Overexpression of cyclin D3 prolonged pregnancy in Hoxa-10^−/− mice until d 10 of pregnancy

As we have demonstrated, cyclin D3 overexpression by adenoviral infection can improve decidualization on d 8 of pregnancy in Hoxa-10^−/− mice. Therefore, we wanted to examine whether this model could rescue Hoxa-10^−/− mice from pregnancy failure. Hoxa-10^−/− pregnant mice were injected similarly to d-8 mice (d 5, 0900 and 1800 h) with either cyclin D3 sense (n = 5) or control (n = 5) and killed on d 10. Pregnant WT mice (n = 4) without virus injections were also examined for comparison. ISs were flash frozen or fixed in 10% formalin for paraffin blocks. IS weight of rAd-Cyc3 mice was again significantly increased over rAd-GFP (ANOVA, P < 0.001) (Fig. 5, A and B), and again, both Hoxa-10^−/− groups had significantly decreased weight of the IS compared with WT (P < 0.001) (Fig. 5, A and B). Histological examination of IS (Fig. 5C) demonstrated arrested growth of rAd-GFP IS and embryo absence. Embryo development did continue in some of the ISs examined in rAD-CycD3 (best development depicted in Fig. 5C; results varied).

Fig. 5.

In vivo overexpression of cyclin D3 in Hoxa-10^−/− uterus during postimplantation period at the site of implantation at d 10 of pregnancy. A, Representative photographs of d 10 ISs in rAd-GFP, rAd-CycD3, and WT mice. B, Weight of IS in rAd-GFP, rAd-CycD3, and WT mice. Results expressed in grams (mean ± sem). *, P < 0.001. C, Hematoxylin and eosin staining of paraffin-embedded d-10 IS of rAd-GFP, rAd-CycD3, and WT mice are shown at ×40. M, Mesometrial pole; AM, antimesometrial pole; Dec, decidua; e, embryo. D, In situ hybridization. Expression of Hand1, Mash2, and Pl1 genes at the sites of embryo implantation on d 10 of pregnancy is shown in rAd-GFP, rAd-CycD3, and WT mice. Frozen sections were hybridized and developed as described in Fig. 4. Dark-field photomicrographs of representative cross-sections hybridized with antisense probes are shown at ×40. spgt; Spongiotrophoblast layer; tgc, trophoblast giant cells.

Because a proportion of rAd-CycD3-treated mice continued to have growth and development within examined IS, we wanted to know whether placentation could be achieved in these mice. In situ hybridization was used to assess for placentation markers in both groups of treated mice as well as untreated mice. Expression of Hand1, Mash 2, and Pl-1 (Fig. 5D) was present but diminished in rAd-CycD3 mice when compared with WT, and all markers were absent in rAd-GFP mice. Embryo development was also noted to be retarded in rAD-CycD3 mice than untreated WT mice. GFP immunostaining was also performed on paraffin sections of rAd-CycD3 and rAd-GFP mice to determine whether adenoviral expression was still present on d 10. GFP staining was absent by d 10 in adenovirus-treated animals (data not shown).

Next, we wanted to determine whether the aberrant development seen at d 10 in the rAd-CycD3 IS resulted from a delayed or failing implantation process. As described above, Hoxa-10^−/− mice treated with cyclin D3 were killed at d 12 (n = 5). At this time point, only two total ISs remained, and embryo resorption was occurring in both ISs (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). Overall, these results suggest that some ISs in rAd-CycD3-treated Hoxa-10^−/− mice can be rescued until d 10 of pregnancy. However, these ISs are abnormal at d 10, and no continued growth and development was seen at d 12.

Cyclin D3 regulates several downstream genes involved in decidualization, and they become aberrant with Hoxa-10 deficiency

Thus far, our experiments suggest that cyclin D3 is a crucial regulator of decidualization during the implantation process and that cyclin D3 replacement can further this process in our knockout model. Therefore, we wanted to determine downstream targets of cyclin D3 involved specifically in decidualization. Proteomic evaluation of triplicate samples from Hoxa-10^−/− primary stromal cell culture infected with either rAd-CycD3 or rAd-GFP (as described in Materials and Methods) were compared at 24 and 72 h after treatment with decidualization-inducing media assessing for proteins involved in stromal cell proliferation (24 h) and polyploidy (72 h). Forty-four proteins were identified for comparison between rAd-CycD3- and rAd-GFP-treated cells (a representative gel picture for 72 h after treatment analysis is shown in the Supplemental Fig. 2). Seventeen proteins were significantly different between the two groups (Table 1). Fourteen proteins had increased expression in rAd-CycD3 cells compared with rAd-GFP cells, six at 24 h and eight at 72 h. Three proteins had reduced expression in rA-CycD3 cells vs. rAd-GFP cells, one at 24 h and two at 72 h. Next, we wanted to determine whether identified gene networks involved in decidualization under cyclin D3 regulation. Using ingenuity pathway analysis, we identified two gene networks (Supplemental Fig. 3) associated with cyclin D3-driven decidualization. Based on these results, we wanted to know whether these proteins were expressed during decidualization in vivo. Seven proteins (Anxa1, Cops5, Gsn, Hspn4, Prdx2, Prdx6, and Tald01) were selected based on known function in the literature (Table 1). Results were evaluated by in situ hybridization for expression on uterine sections before decidualization on d 4 and during decidualization on d 7 in WT mice (Fig. 6A and Supplemental Fig. 4). ISs from d-7 Hoxa-10^−/− mice, which show severely reduced expression of cyclin D3 and severely reduced stromal decidualization (18, 20), were also used for comparison (Fig. 6A and Supplemental Fig. 4). Increased expression in the decidual bed was noted for all seven probes on d-7 IS in WT, compared with d-4 WT, indicating that these probes are specific to decidualizing stromal cells under normal implantation condition. Furthermore, these genes were also down-regulated with aberrant decidualization condition in Hoxa-10 knockout mice on d-7 IS, again suggesting that these genes are indeed decidualization specific (Fig. 6A). As stated above, all markers were present in d-7 WT tissue, however regional distribution of expression of these seven probes differed within the decidual bed. More specifically, Anxa1 and Prdx6 were localized within the SDZ, where the expression patterns of Gsn and Tald01 demonstrated mesometrial sparing. Prdx2 was expressed throughout the decidual bed (Fig. 6), and this pattern was most similar to our previously observed cyclin D3 expression on d 7 (20). Cops5 and Hspa4 were also distributed throughout the decidual bed but had reduced level of expression compared with the other five probes (Supplemental Fig. 4). To validate the efficacy of in vivo rescue experiments, we completed in situ hybridization studies using the above cyclin D3 target probes on d-8 IS of rAd-Cyc3, rAd-GFP, and WT mice. The expression of these targets indeed appeared to have increased in both rAd-CycD3 and WT ISs when compared with rAd-GFP IS (Fig. 6B), suggesting that cyclin D3 indeed supported the regulation of expression for several downstream targets of cyclin D3 in the defective decidual bed.

Table 1.

Differential protein expression comparing cyclin D3-overexpressed stromal cell culture vs. control at 24 and 72 h

Gene	Location	Description: function	GFP vs. CycD3		GFP vs. CycD3
			24 ha	P value	72 ha	P value
Anxa1	PM	Annexin A1: polymorphonuclear leukocytes and intracellular vesicle trafficking, phagocytosis, cell signaling, proliferation, mediator of glucocorticoid action in inflammation	1.27	0.031	1.12	NS
C19orf10	ECS	Chromosome 19 open reading frame 10: unknown	−1.29	0.04	1.15	NS
Cops5	Nucleus	COP9 constitutive photomorphogenic homolog subunit 5: cell-cycle progression, apoptosis, DNA damage	1.02	NS	1.18	0.04
Dars	Cytoplasm	Aspartyl-tRNA synthetase: unknown	1.4	0.004	1.13	NS
Des	Cytoplasm	Desmin: antihypertophic protein, involved in muscle regulation	1.31	0.04	1.03	NS
Dpysl2	Cytoplasm	Dihydropyrimidinase-like 2: unknown	−1.06	NS	1.26	0.02
Eif4 h	Cytoplasm	Eukaryotic translation initiation factor 4H: unknown	1.55	0.03	−1.07	NS
Eno1	Cytoplasm	Enolase 1 (α): regulate differentiation and function of mast cell function	1.42	NS	1.2	0.02
Erp29	Cytoplasm	Endoplasmic reticulum protein 29: sperm-oocyte fusion, involved in acrosome reaction and epidiymal transit of sperm, carcinogenic via p53	−1.26	NS	−1.5	0.04
Gsn	ECS	Gelsolin: in vitro actin filament severing activity, hemostasis in inflammation, wound healing	−1.00	NS	1.24	0.04
Hspa4	Cytoplasm	Heat shock 70-kDa protein: τ phosphorylation and amyloid precursor protein processing, spermatogenesis	1.05	NS	1.27	0.001
Prdx2	Cytoplasm	Peroxiredoxin 2: cell proliferation and differentiation, natural killer cell function	−1.04	NS	1.21	0.004
Prdx6	Cytoplasm	Peroxiredoxin 6: oxidative stress	1.53	NS	1.31	0.004
Tald01	Cytoplasm	Transaldolase 1: lipid biosynthesis, mitochondrial structure and function	1.06	NS	1.59	0.02
Tpm3	Cytoplasm	Tropomyosin 3: actin filament system essential for many cellular functions, including shape, motility, cytokinesis, intracellular trafficking, and tissue organization	−1.24	NS	−1.11	0.03
Vil2	PM	Villin 2 (ezrin): epithelial cells organization and villus morphogenesis in the intestinal tract	1.57	0.01	1.13	NS
Yars	Cytoplasm	Tyrosyl-tRNA synthetase: unknown	1.27	0.02	1.13	NS

PM, Plasma membrane; ECS, extracellular space; NS, not significant.

^aRatio of expression of gene product for rAd-GFP vs. rAd-CycD3-infected Hoxa10^−/− stromal cells in vitro.

Fig. 6.

Analysis of expression of genes identified as downstream targets of cyclin D3. In situ hybridization for the expression of Anxa1, Gsn, Prdx2, Prdx6, and Tald01 mRNAs on uterine IS. A, d-4 WT and d-7 IS of WT and Hoxa-10^−/− mice. B, d-8 IS of rAd-GFP- or rAd-CycD3-injected Hoxa-10^−/− and noninfected WT. Frozen sections were hybridized and developed as described in Fig. 4. Dark-field photomicrographs of representative cross-sections hybridized with antisense probes are shown at ×40. le, Luminal epithelium; s, stroma; ds, decidualizing stroma; M, mesometrial pole; AM, antimesometrial pole; e, embryo.

Discussion

Decidualization is a crucial step needed for successful embryo implantation. However, the mechanisms underlying this process remain poorly understood. Previous studies have demonstrated that Hoxa-10 gene products are involved in adult pregnancy (12). Specifically, targeted gene experiments have demonstrated that the decidualization process in Hoxa-10^−/− mice is severely compromised (12, 18). Previous studies from our group have shown that within this knockout model, cyclin D3 expression, an important G₁ cell-cycle regulatory protein, is significantly reduced (18, 20). Cyclin D3 expression temporally parallels the decidualization process with cyclin D3 up-regulation occurring around d 5 of pregnancy in the murine model. In addition, cyclin D3 knockout females have shown defective decidualization when compared with WT mice but are not completely infertile (27). This knockout model demonstrates that normal blastocyst attachment on d 5, however, has severely compromised stromal cell proliferation and differentiation at certain IS on d 8. Cyclin D3 is therefore considered a crucial regulatory protein necessary for normal decidualization. Our present study demonstrates that cyclin D3 replacement via adenoviral infection can reverse some of the aberrant decidualization within the Hoxa-10 knockout mouse both in vitro and in vivo.

The cell cycle is tightly regulated during decidualization with interplay between cyclins, cyclin-dependent kinases (cdks), and cdk inhibitors controlling proliferation and differentiation. More specifically, cyclin D3, with cdk4 and cdk6, plays a role in directing decidualization (5). Expression of cyclin D3 is low on d 4 of pregnancy within murine stromal cells. However, it dramatically increases in decidualizing stromal cells after the initiation of implantation (20). Stromal cell proliferation begins to markedly increase as well on d 5 of pregnancy. In Hoxa-10^−/− mice, this proliferation response is severely compromised. Our data demonstrate that by overexpressing cyclin D3 in primary stromal cell culture and on d 5 of pregnancy, marked stromal cell proliferation can occur in this model confirmed by immunostaining (Figs. 1A and 3, A and B).

After this marked proliferation, stromal cells begin to terminally differentiate into decidual cell with acquisition of polyploidy. Polyploid decidual cells develop via endoreplication cycle with repeated rounds of DNA replication without cytokinesis under regulation of cyclin D3 (27). These processes are impaired in the Hoxa-10^−/− mouse (3, 6). Again, using histology and RNA expression studies, we have demonstrated that by replacing cyclin D3 in Hoxa-10^−/− mice that are close to normal (WT) or supernormal levels of decidualization and polyploidy, markers can be achieved at d 8. Our in vitro studies also confirmed increased polyploidization in cyclin D3-replaced Hoxa-10^−/− primary stromal culture.

Given our findings that replaced cyclin D3 Hoxa-10^−/− mice can have increased stromal cell proliferation and differentiation, we believe that cyclin D3 regulates downstream targets in this process. We tested this hypothesis using proteomic evaluation of cyclin D3-replaced Hoxa-10^−/− stromal cell cultures vs. control (GFP) and found several up-regulated proteins in the cyclin D3-replaced culture over control during both proliferation and differentiation (with polyploidization). Ingenuity pathway analysis evaluation revealed two gene networks that are likely downstream targets of cyclin D3 (Supplemental Fig. 3). Known functions of genes in these networks include proliferation, tissue remodeling, inflammatory regulation, cell trafficking, and apoptosis; and all are processes involved in decidualization. Our in vivo comparison of mouse uterine tissue before decidualization and in a severely compromised decidualization model with uterine tissue at the peak of decidualization confirmed increased expression of these genes specifically in decidualizing stromal cells (Fig. 6 and Supplemental Fig. 4). These results confirm that cyclin D3 expression in addition to cell-cycle regulation during decidualization also regulates downstream genes to allow for successful decidualization.

Despite the success of our replacement model using adenoviral infection of cyclin D3 into Hoxa-10^−/− mice on d 8 of pregnancy, it did not reverse infertility seen in this knockout mouse. In the cyclin D3 replacement animals, expression of placentation markers were demonstrated, which are not usually expressed in Hoxa-10^−/− mice nor were found in the rAd-GFP-treated mice. However, when compared with WT tissue, placentation site and embryo development were abnormal in the rAd-CycD3-treated Hoxa-10^−/− mice on d 10. By d 12 of pregnancy, almost all embryos and ISs had resorbed. This showed that our replacement model, although successful for improving decidualization, did not continue to drive implantation after d 10.

Although our replacement model was not successful in the continuation of pregnancy in Hoxa-10^−/− mice, adenoviral infection did result in all pregnancies persisting longer than usual. We appreciate several factors they may compromise the success of our replacement model. First, cyclin D3 replacement occurred on d 5, and we know that Hoxa-10^−/− mice also have aberrant blastocyst attachment, which may be responsible for the variation in the number of successful ISs. Second, cyclin D3 was replaced on a single day of pregnancy (d-5 morning and evening) with in vivo tissue expression of the adenovirus lasting only about 3–4 d, shown by lack of GFP immunocytochemistry staining on d 10. Therefore, continued replacement may be necessary to allow sustained tissues levels of cyclin D3 for successful pregnancy outcomes. However, the adenoviral delivery model is flawed, because multiple repeated dosing results in animal sickness and most commonly death before delivery of pups. Previously, it has been shown that postimplantation injection of adenoviruses in mice primarily targets the IS in antimesometrial location of the decidual bed (26, 28). In contrast, analysis of GFP expression in rAd-CycD3 or rAd-GFP-injected Hoxa-10 null mice reveals no pole-specific pattern of GFP distribution in the IS at d 8 (Fig. 2E), and these are consistent with gene expression for Hoxa-10 null IS (18). The apparent discrepancy between the uniform GFP staining in rAd-CycD3 transfected animals and the focal expression of cyclin D3 is not clearly understood. However, it is expected that cyclin D3 staining is nuclear and GFP is cytoplasmic. The focal sparing of cyclin D3 staining occurs in the PDZ, which is avascular with cells undergoing degeneration and often lack further development. Our explanation for the discrepancy is that although seen within the cytoplasm, within the nondeveloping cells in the PDZ, the adenovirus was not incorporated into the nucleus to allow for immunocytochemistry nuclear staining. Hoxa-10^−/− mice may lack other signaling molecules, not cyclin D3, that directs regional decidual development, which seems to be aberrant still within these cyclin D3-replaced knockout animals. Finally, we must consider that other proteins (one or more) are also repressed in this knockout and may be required for normal placentation. Confirming this theory is challenging, because Hoxa-10^−/− mice typically do not reach the placentation stage of implantation. However, a comparison between the cyclin D3-supplemented Hoxa-10 null IS on d 8 or 10 against that of corresponding WT by microarray studies may be useful to identify novel compromised targets. Continued research involving decidualization and placentation regulation is necessary to further understand these processes.

The study of human implantation remains challenging due to ethical constraints on researching human blastocyst-uterine interactions. Therefore, science must rely on animal models to determine mechanisms that underlie successful implantation and decidualization. We know that human stromal cells also undergo differentiation into decidual cells during the luteal cycle. And previous research has demonstrated that HOXA-10 expression increases in human endometrial epithelium and stromal cells during the midluteal phase of the menstrual cycle, coinciding with the window of implantation (16, 17). Hoxa-10 expression has also been noted to be reduced in several fertility compromising diseases, such as endometriosis (29). Given this, our findings are important in understanding the function and mechanism of Hoxa-10 expression in adult pregnancy. By continue development of this field of knowledge, diagnosis of and possible treatment for implantation defects in humans may be possible in the future.