1 Department of Gynecology and Obstetrics, Stanford University School of Medicine, Stanford, California 94305-5317
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
endometrium; decidualization; implantation; microarray technology
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Decidualization involves transformation of endometrial spindle-like fibroblasts into polygonal epithelial-like cells that are hypothesized to regulate placental trophoblast invasion into the endometrium during the invasive phase of implantation (53). Stromal cells undergo these characteristic morphologic changes when treated with E2/P4 and/or cAMP in vitro (6, 39, 53, 54).
Decidualization is known to be mediated by the PKA pathway (24, 54, 60) and relaxin, which acutely and permanently elevates cAMP levels, and induces prolactin expression and the morphologic changes of decidualization in human endometrial stromal cells. Also, PKA pathway members including regulatory subunit isoforms (RI, RIß, RII
, RIIß) and catalytic subunits (C
and Cß) are upregulated in stromal cells treated with relaxin (54). In progesterone-induced decidualization, progesterone acts through the PKA pathway, elevating both prolactin and intracellular cAMP levels (6). The PKA inhibitor, 8-bromoadenosine-3',5'-cyclic monophosphorothioate, significantly suppresses progesterone-dependent prolactin expression (6). In addition, cAMP is an independent mediator of decidualization (39, 53). cAMP derivatives promote the differentiation of the fibroblast-like stromal cells to the decidual phenotype and induce the expression of products characteristic of decidual cells, e.g., prolactin, IGFBP-1, desmin, hsp 27, and laminin (6, 39, 53).
Microarray analysis has been used to investigate gene expression in in vitro models of decidualization in endometrial stromal cells (39) and term decidual fibroblasts (49). In our initial cDNA microarray study with human endometrial stromal cells from nonpregnant subjects, comparison was made of genes upregulated at fixed time points, i.e., after 48 h in response to 8-Br-cAMP and after 10 days of treatment with estradiol (E2) and progesterone (39). Numerous gene families were upregulated with treatment with both cAMP and E2/P4, with concordance between the two different treatment groups, and included growth factors, neuromodulators, inflammatory cytokines, cell adhesion molecules, oncogenes, and transcription factors. A recent study on human term pregnancy decidual fibroblasts treated for 15 days with E2/P4 and cAMP revealed dynamically regulated genes with reprogramming of gene expression within functional categories, suggesting fundamental aspects of cellular differentiation (6).
Herein, we present the results of high-density oligonucleotide microarray analysis and K-means kinetic pattern grouping of genes and gene families expressed in nonpregnant human endometrial stromal cells during the first 48 h of treatment with 8-Br-cAMP. cAMP was utilized, as it is known to play a role during in vitro decidualization of the human endometrial stromal cell. Although cAMP may not be the exclusive mediator of decidualization in the human endometrial stromal cell in vitro, it has been shown to be a potential key mediator in the decidualization process as levels of PKA pathway intermediates increase with decidualization and inhibitors of the PKA pathway result in decreased expression of markers of decidualization, i.e., prolactin, IGFBP-1. The results of our array reveal genes previously known to be induced or downregulated in response to cAMP, as well as numerous newly recognized genes and gene families. In addition, up- and downregulation of genes within families and within specific functional groups supports reprogramming as an important mechanism in activation of the PKA pathway in human endometrial stromal cells.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
At each time point (0, 2, 12, 24, 36, and 48 h), medium was collected for analysis of IGFBP-1 levels by ELISA (Diagnostic Systems Labs, Webster, TX), and TRIzol was added immediately to cells to isolate total RNA. Patient samples utilized for the microarray analysis were selected based upon their similar levels of high peak concentrations of secreted IGFBP-1 at all time points, as measured by ELISA (see Fig. 2 of RESULTS).
|
Validation of Gene Expression Data
RT-PCR.
Total RNA was used in a reverse transcriptase reaction (1 µg RNA/20 µl reaction volume, 10x RT buffer, dNTPs, oligo-dT, RT enzyme) which was subsequently utilized in a PCR reaction (RT, 10x Taq buffer, MgCl2, dNTP, Taq enzyme) with primers (Table 1) specific for preprosomatostatin, IGFBP-1, IL-11, cyclin B, TIL-4, IGFBP-5, and GAPDH. PCR reaction conditions were 34 cycles of the following sequence: 3 min at 94°C, 45 s at 94°C, 45 s at 56°C, 45 s at 72°C, followed by 10 min at 72°C with subsequent cooling to 4°C.
|
Data Analysis
Data were analyzed with GeneChip Analysis Suite ver. 4.01 (Affymetrix), GeneSpring ver. 4.21 (Silicon Genetics, Redwood City, CA), and Microsoft Excel 2001 software. Expression profile data was first prepared using GeneChip Microarray Analysis Suite and subsequently exported to GeneSpring for further analysis. Within each hybridization, the 50th percentile of all measurements was used as a positive control, and the measurement for each gene was divided by this control. The bottom 10th percentile was used for background subtraction. Between different hybridization outputs, each gene was normalized to itself by making a synthetic positive control for that gene comprised of the median of the genes expression values over all samples in an experimental group and dividing the measurements for that gene by the positive control, as specified in the manufacturers instructions. Mean values were then calculated for each gene probe set among individual time points, and the fold change difference between expression at a particular time point and the expression at the time point of 0 h was calculated for each cell line studied. Data are reported in RESULTS (see Supplemental Tables A and B, available at the Physiological Genomics web site)1
for the specific genes upregulated and downregulated at individual time points, where the fold change is calculated as the ratio of the gene expression between an individual time point and the gene expression at 0 h.
K-means and Ontology Analysis
Genes were selected for consideration for K-means and ontological analysis based upon their apparent expression change relative to the time 0 reference sample, where only genes demonstrating a twofold or greater change in expression at a particular time point were utilized for the analysis. This approach of only analyzing trends of the most highly regulated genes, as utilized in Aronow et al. (1), allowed detection of coordinately regulated groups of genes during the time course without the dilution effect of genes whose expression did not change significantly during treatment with cAMP. For the ontology analysis, genes demonstrating a twofold or greater change at each individual time point were utilized for our analysis. Ontological groupings were classified into biological function groupings based upon the categories listed in the GeneSpring program (Silicon Genetics). K-means analysis and tree diagrams in the GeneSpring program (Silicon Genetics) were utilized to analyze the temporal pattern of gene expression within each ontology category of gene function. For the K-means analysis of the temporal trends in gene expression, genes with regulation greater than twofold in both samples in half or more of the time points were included in the analysis. K-means was applied to the data using standard correlation analysis. Genes were clustered according to their expression pattern dynamics by subjecting the log2-transformed data set [R = log2(xt=i/xtime 0)], where R is the expression ratio for each gene, to the K-means as implemented in the GeneSpring program (Silicon Genetics). Four groups of genes with different characteristic temporal expression patterns were derived from the K-means analysis of up- and downregulated genes. Four groups were utilized to categorize the individual kinetic patterns for both the up- and downregulated genes, as a smaller number of groups did not adequately distinguish gene behavior and a larger number of groups did not display significant distinguishable trends between groups. Overall trends of temporal gene expression in the comprehensive data set were compared with those in the individual ontological groupings using K-means analysis to identify trends in gene expression by functional classification during the onset of treatment with cAMP.
Microarray data for this manuscript are available on line at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo), with accession number GSE403.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
|
|
|
|
For the downregulated genes (Table 4), a number of cancer/cell cycle regulation, enzymes, G protein receptor signaling, and extracellular matrix proteins demonstrate pattern 1. A significant subset of all major ontological categories and subcategories of downregulated genes, including cell growth, cancer/cell cycle regulation, enzymes, immune proteins, signal transduction, structural proteins, and G protein receptor signaling, is represented in pattern 2. Pattern 3 is predominantly seen in a significant subset of the cell growth, enzymes, and cancer/cell cycle regulation genes, and pattern 4 includes a significant subset of cell death, immune proteins, cancer/cell cycle regulation, enzymes and signal transduction genes. These up- and downregulated genes are described in more detail below.
Ontological Trends of Upregulated Genes
Many of the cancer and cell cycle regulation genes follow pattern 2, with an earlier first peak at 2 h (e.g., N-ras) or more commonly, peak expression at 12 h and 48 h [e.g., Von Hippel-Lindau (VHL) gene and p126 (ST5)]. Other cell cycle regulatory genes, including NF1 and trk, demonstrate a more gradual progressive increase in expression during the time course studied. Some cell growth genes can be divided into two categories based upon their pattern of temporal expression. One group, including Rb107 and carcinoembryonic antigen, follows pattern 3, and another group shows low levels of initial expression and subsequently a rapid increase in between 12 and 24 h, reaching a plateau and remaining elevated thereafter. This latter group includes IGFBP-1, VEGF receptor-2, EBAF, and TGF-ß1 binding protein, clone L5 orphan G-protein-coupled receptor, IGF-1 receptor, TGF-ß2, and inhibin ßB-subunit.
The gene subfamily which signals through the STAT protein kinase cascade (including the neuropeptides) also demonstrates a characteristic pattern of expression of a rapid increase in expression between 12 and 24 h, with a plateau of a high level of expression maintained between 24 and 48 h. Genes in this category include preprosomatostatin, somatostatin precursor I, somatostatin receptor-2 isoform (sstr2), and gene expressed in poorly metastatic human melanoma cell lines and human cerebral cortex cDNA. The coordinated pattern of regulation of three of the neuropeptide genes signaling through the STAT protein kinase cascade which follow this pattern of induction to peak levels of gene expression between 12 and 24 h and retention of high level of expression between 24 and 48 h is demonstrated graphically in Fig. 7. This well-conserved pattern of upregulation of the STAT protein kinase cascade family members implicates a coordinated regulation of these genes by cAMP.
|
Kinases displaying a pattern of negligible expression between 0 and 12 h and peaking between 12 and 24 h include Rac 1 (calmodulin binding protein), ligand for Tie2/Tek receptor tyrosine kinase and MAP kinase phosphatase. Another group of enzymes demonstrates a sharp increase in expression between 2 and 12 h, when peak levels of expression were achieved, including cytochrome P-450 nifedipine oxidase, carbonic anhydrase-like domain, and PPEF-2.
The majority of the nucleic acid binding proteins demonstrate a biphasic pattern of expression. The immediate early response gene NOT and its mouse homolog, NGFI-B/nur77 B type transcription factor, demonstrates biphasic regulation with marked increases in gene expression at 2 and 48 h. RNA helicase, SALL1, and Zinc finger protein ZN 72D expression increased in the first 12 h of cAMP treatment and subsequently leveled off.
The ontological subfamily of genes signaling through the NF-B pathway achieves peak levels of gene expression at various different time points. RANK and anti-colorectal carcinoma heavy chain achieve their peak levels of expression at 2 h and 48 h, whereas TRIP9 demonstrates a gradual pattern of gene induction.
With regard to genes signaling through the integrin receptor, LFA-1 -subunit precursor and disintegrin show a pattern of expression with peak levels achieved at 12 h and subsequently tapering off for the remainder of the time course. Members of the GABA R signaling family, including GABA transport protein, GABA-A receptor
-subunit and GABA-B receptor subunit gb2, demonstrate a biphasic pattern of expression.
Ontological Trends of Downregulated Genes
The cell growth genes exhibit two primary patterns of expression: pattern 2 genes include IGFBP-5, monocarboxylate transporter, p55CDC mRNA, caveolae associated protein, and Na-K-ATPase ß1-subunit mRNA. Genes with pattern 3 include human nerve growth factor (HBNF-1), monocarboxylate transporter, and human histone stem-loop binding protein (SLBP). Members of the cancer/cell cycle regulation gene family exhibit two predominant patterns of downregulation seen in the K-means analysis. Pattern 2 genes include UbcH10, proto-oncogene protein, and cyclin B2, and genes exhibiting pattern 3 include cyclin B, associated with cyclin-dependent kinase and PRAD1 mRNA for cyclin.
Most members of the cell death family of genes primarily exhibit the trend of pattern 4, including TNF ligand family, TNF type 2 receptor binding protein, and FLICE2; ICE like protease; caspase-10/b.
Kinases were found in the K-means analysis to exhibit all four predominant patterns of gene expression seen in the downregulated genes. Kinases that illustrate pattern 1 include L-kynurenine hydrolase, argininosuccinate lyase, platelet-type phosphofructokinase, arginine-tyrosine kinase, and retSDR1. Kinase genes demonstrating pattern 2 include thymidylate synthase, thymidine kinase, trypsinogen IV b-form, mitotic kinesin like protein-1, N-myristoyltransferase-2, nicotinic mononucleotide pyrophosphorylase, and Na-K-ATPase ß1-subunit. Genes demonstrating pattern 3 include phosphoenolpyruvate carboxykinase, P-13 kinase associated p85 mRNA, retinal short-chain dehydrogenase/reductase, and pre-ß-migrating plasminogen activator inhibitor. Kinases demonstrating pattern 4 include aldolase A, glyceraldehyde-3-phosphate dehydrogenase, disintegrin and metalloprotease-like domains, and nicotinate mononucleotide pyrophosphorylase.
The nucleic acid binding protein family of genes exhibits primarily pattern 1 and includes winged helix transcription factor, DNA binding protein AP-2, ERCC2 gene, hnRNAcore protein A1, and myocyte-specific enhancer factor 2A (MEF2A).
The immune gene family demonstrates three predominant patterns of expression, patterns 1, 2, and 4. Genes following pattern 1 of expression include placental protein 14 (glycodelin); genes following pattern 2 of expression include Fc--RIIA gene for IgG Fc receptor and MHC class I mRNA fragment. Genes following pattern 4 of expression include skeletal muscle 165-kDa protein and CMRF 35 mRNA.
Genes for receptors or genes playing a direct role in signal transduction followed three major patterns of expression according to the K-means analysis. Pattern 1 was exhibited by activator of NF-B, integrin-
4 subunit, arginine-tyrosine kinase, soares pregnant uterus [decidual/trophoblast PRL-related protein (d/tPRP)], and FRP (frizzled homolog). Another set of genes follows pattern 2, including GnRH receptor, keratin 18 precursor,
-interferon-inducible protein precursor, RTP gene, pregnancy-specific ß-1 glycoprotein precursor, integrin-
6, and serine/threonine kinase (BTAK). The other set of genes in the signal transduction family follows pattern 4 and includes testis-specific disintegrin and metalloprotease-like domains, and annexin homologous tetrad.
Genes of the structural protein family exhibit predominantly pattern 2 and include actin-, tropomyosin-, and calmodulin-binding protein in smooth muscle, 1 type XVI collagen (COL16A1), and preprofactor XI. Genes signaling through G protein receptors follow either pattern 1, including G-protein-coupled receptor, purinergic P2YU receptor and BLR1 gene for Burkitts lymphoma receptor, or genes following pattern 2, including endothelial differentiation protein (edg-1), type 3 inositol 1,4,5-triphosphate receptor, cytosolic thyroid hormone-binding protein, and
-interferon-inducible protein precursor. Genes in the extracellular matrix family of genes exhibit predominantly pattern 2 and include
1 collagen type XIII mRNA, keratin 18 precursor, and
1 type XVI collagen (COL16A1). Genes in the integrin receptor signaling pathway exhibit primarily pattern 2, including integrin-
6 and integrin-
4 subunits. Other genes in the integrin family exhibit pattern 1 and include eMDC II protein. Genes in the serine/threonine kinase family exhibit either pattern 1 or pattern 2. Genes following pattern 1 include cdk3 for serine/threonine protein kinase. Serine/threonine kinase family members following pattern 2 include serine/threonine kinase (BTAK) and murine pim-2 product.
Validation of Gene Expression
Northern analysis and RT-PCR with RNA from endometrial stromal cells treated with and without cAMP were conducted to validate select gene expression. RT-PCR (see results in Fig. 8) was performed with cells treated with 1 mM cAMP for 48 h, where the levels of gene expression in these cells were compared with those of nontreated stromal cell RNA (time 0). The primer sets utilized for RT-PCR are shown in Table 1. Although quantitative RT-PCR was not performed, it is evident from the data shown in Fig. 8, A and B, that there is clear upregulation in the human endometrial stromal cell of preprosomatostatin, IGFBP-1, and IL-11 and downregulation of IGFBP-5, TIL-4, and cyclin B with treatment with cAMP. These data are consistent with observations from the microarray data and Northern analysis (see below). In addition, downregulation of IGFBP-5, cyclin B, and TIL-4 was validated by RT-PCR in Fig. 8, A and B.
|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Since the two cell lines studied were matched for IGFBP-1 levels achieved at each time point (0, 2, 12, 24, 36, and 48 h) (see Fig. 2), the level of differentiation in response to cAMP was not likely to contribute significantly to the variability in the levels of gene expression. Similar levels of IGFBP-1 secreted by the two patient samples have been assumed to reflect similar degrees of differentiation. Thus, although the pattern of gene expression was consistent throughout the time course, the magnitude of expression at individual time points during decidualization differed likely due to patient-to-patient variability.
The most significantly upregulated genes detected in this study include preprosomatostatin, IGFBP-1, NOT, MMP-10, orphan L5 G-protein-coupled receptor, prolactin, and human negative growth protein MYD 118. The most significantly downregulated genes include activator of NF-B, actin/troponin/tropomyosin binding protein, IGFBP-5, lipocortin III, retinal short chain dehydrogenase, and
1 type XVI collagen. Several genes and gene families found herein to be sequentially regulated during treatment with cAMP over a 48-h period were similar to those recently reported with human term pregnancy decidual fibroblasts treated for 15 days with cAMP and E2/P4 (6). The comparison of the results of the study of Brar et al. (6) and our current microarray analysis for some of the most significantly upregulated and downregulated genes is displayed in Table 5. The consistency between these data with different sources of stromal cells (pregnancy decidua vs. cycling, nonpregnant endometrium), time courses, and differing decidualization stimuli, remarkably underscores the validity of the data and the commonality of some genes expressed during development of the decidual phenotype.
|
|
K-means Analysis
K-means analysis revealed four predominant patterns of gene expression for the genes for which there was greater than twofold regulation in our time course microarray. Ontological categories of genes contained similar trends of gene regulation, indicating that there is coordinated regulation of genes by function that mediate the secretory and morphologic transformation of the human endometrial stromal cell during decidualization.
Upregulated Gene Families
Neuropeptides.
Remarkable changes in levels of expression were observed for the somatostatin receptor signaling pathway, with coordinated induction of preprosomatostatin, somatostatin R isoform 2 and the orphan G-protein-coupled receptor. Immunohistochemical studies have localized somatostatin to secretory endometrium, exclusively in the endometrial stromal cells (27), and upregulation with decidualization is consistent with a recent study (6). The somatostatin receptor sstr2 is expressed in endometrial epithelium, endothelium, and stroma throughout the menstrual cycle. Its expression varies in the epithelial cells surrounding the endometrial glands from being basal or diffuse in the proliferative and secretory phase, respectively, to being lumenal in the menstrual stage (19). Somatostatin is best known for its function in inhibiting pituitary GH release upon binding to sstr2 (44). However, it also inhibits the release of other physiologically important compounds, including insulin, glucagon, gastrin, and secretin (46). Somatostatin also functions as a promoter and an inhibitor of angiogenesis, depending upon the tissue type studied (32, 35). For example, with the pituitary tumor cell line TtT/GF the somatostatin analog, octreotide, stimulates release of VEGF (32), whereas with human umbilical endothelial cells, it inhibits basal and stimulated endothelial cell proliferation (35). Somatostatin further exhibits both inhibitory and stimulatory effects on immune cell proliferation and secretion of cytokines (22, 29), including release of IL-6 from peripheral blood monocytes (29). Somatostatin binds to target cells via a member of the seven transmembrane domain superfamily of glycoprotein receptors, leading to inhibition of adenylate cyclase and Ca2+ channel activity, stimulation of K+ and tyrosine phosphatase activity, and regulation of intracellular pH (46). We postulate that somatostatin may act in an autocrine fashion via sstr2 to inhibit proliferation and promote the differentiation of the decidualized stromal cell, in addition to acting by paracrine mechanisms to regulate angiogenesis and immune functions during endometrial cyclic changes and implantation.
Semaphorin III, upregulated during stromal cell treatment with cAMP herein, was shown in a recent report from our group to be downregulated during the window of implantation in human endometrium (28). These differences may reflect semaphorin expression in other cell types in whole tissue or paracrine interactions required for cellular expression in vivo that are not preserved in isolated cell culture systems in vitro. Semaphorins are a family of neuropeptides that act as chemoattractants or repellants, depending upon levels of cGMP (29). The finding of semaphorin expression in the endometrium and its regulation with decidualization have led us to hypothesize that chemoattractants and ion signaling may function to guide an embryo in the endometrium during implantation (28).
Immune genes.
Some of the immune genes of interest found to be sequentially upregulated with cAMP treatment are demonstrated in Table 7. The response of IL-8, IL-11, and the IL-1 receptor is consistent with work previously published (39). IL-11 upregulation with cAMP treatment is of particular interest, as IL-11 mRNA expression in endometrial stromal cells during the secretory phase precedes that of prolactin, a known marker of decidualization (10). Furthermore, in mice, a null mutation in the IL-11 Ra subunit results in an infertility phenotype, because of the absence of stromal decidualization (1). In addition, IL-11 enhances stromal cell viability and prolactin secretion from cells decidualized with 8-Br-cAMP (52). Thus IL-11 has an important role during the process of decidualization and stromal cell survival.
|
Cell growth and cell differentiation.
A variety of genes involved in promoting and regulating cell growth and differentiation were found to be upregulated herein, including VEGF, TGF-ß family members, and specific early response genes. VEGF is expressed by the stromal cells during the midsecretory phase, and both VEGF immunostaining intensity and mRNA expression are significantly increased by the administration of E2/P4 to stromal cells in vitro (2), consistent with the results of our previous cDNA microarray study (39). VEGF is postulated to play an important role in endometrial angiogenesis and implantation and in the maintenance of pregnancy. Herein, we found upregulation of EBAF, another TGF-ß family member. In fertile women, EBAF expression decreases during the window of implantation and increases in late secretory and menstrual endometrium (50). In contrast, in women with infertility and endometriosis, EBAF expression does not decrease during the window of implantation, and it has been hypothesized that it may be a marker for uterine nonreceptivity (50). Other interesting genes involved in cell growth that are upregulated with cAMP treatment include FGF-18 and FGFR2, PDGF-R, hepatocyte growth factor, ß nerve growth factor, and connective tissue growth factor related protein WISP-1 subunit precursor, and TGF-ß2 precursor, were upregulated the findings herein.
Prolactin was found to be significantly upregulated with cAMP treatment in our array (143.4-fold above time 0 at 36 h, 37.135-fold above time 0 at 48 h), which is consistent with previous studies which have found prolactin to be one of the major secreted proteins of the decidualized endometrial stromal cell (55).
Some immediate-early response genes with significant homology to steroid and thyroid hormone receptors were regulated during stromal cell treatment with cAMP. The NOT gene exhibited marked upregulation after 2 h of treatment with cAMP (239.4 and 462.5-fold above time 0), and NGFI-B/nur77, the mouse and rat homolog of the human immediate-early response gene NAK1/TK3, was expressed at 47.8 and 164.5-fold at 2 h, with a subsequent decease for the duration of our time course. NOT has previously only been detected in vivo in the brain (33). NOT and NAK1/TK3 represent a distinct group of orphan steroid receptors that function as general coactivators of gene transcription, rather than as typical steroid receptors that act to induce specific target genes (33). In vitro NOT mRNA is expressed in growth-arrested fibroblasts within 30 min of treatment with serum, and peak levels are achieved by 3 h (33). This response is not affected by cycloheximide, suggesting that NOT expression is consistent with an immediate-early response gene and does not depend on protein synthesis for expression (33). The impressive upregulation of both NOT and NAK1/TK3 within a short time frame in endometrial stromal cells in response to cAMP treatment suggests early transcriptional activation that may be important in the transition to the decidual phenotype from proliferation to production of unique extracellular matrix and secretory products (see below).
Oncogenes/cell cycle regulators.
Several genes involved in cell cycle regulation are regulated with cAMP treatment, including snoI, N-ras, NF1, Rb-related protein (p107), trk, and Kip 2; snoI is an oncogene with a proposed role in muscle gene regulation (38), and N-ras is an oncogene found in a subset of endometrial carcinomas (56). NF1 (neurofibromin 1 gene) is implicated in the pathogenesis of neurofibromatosis type 1 (30) and was also found to be upregulated in our previous cDNA microarray study (39) and in decidualization of term decidual fibroblasts (53). Its function in the endometrium is enigmatic. Retinoblastoma-related protein (p107), a known tumor suppressor gene, may have a role during development and differentiation (8). Abundantly expressed in placenta during the first trimester, it has been postulated to control trophoblast proliferation (8), and it may play a role in differentiation of the endometrial stromal cell. Trk is a high-affinity neurotrophin receptor, and TrkA and TrkC have been previously shown to be specifically expressed in secretory-phase endometrium (45). The presence of neurotransmitters, neurotropic factors, and their receptors in the endometrium requires further investigation. The CDK inhibitor Kip 2 is a gene implicated in tumorigenesis; however, there is no previous evidence of expression or tumor induction in the endometrium by Kip 2 (42). The BRCA genes are linked to a variety of reproductive tract tumors, including uterine and ovarian serous papillary carcinoma and carcinoma of the breast (23). The finding of upregulation of cell cycle regulatory genes in the decidualized stromal cell lends support to the observed cytological evidence that during the decidualization transition, endometrial stromal cells expand not only by cellular hypertrophy, but also by mitosis and endoreduplication. Proliferation of endometrial stromal cells is mediated by growth-related peptides, prostaglandins, and Ki67 (15). Roles for these cell cycle regulators in endometrial stromal cellular decidualization await further study.
Extracellular matrix.
The extracellular matrix undergoes marked transformation in the establishment of the decidua and in nonpregnancy cycles in preparation for menstrual tissue desquamation (41). Herein, MMP-10 (stromelysin-2) is markedly upregulated during cAMP treatment. Previous studies have shown stromelysin-2 mRNA is expressed in stromal cells in late secretory and menstrual endometrium (41), consistent with the current study. The pattern and temporal expression of MMP-10 suggest that it may play a key role in the decidualized stromal cell for matrix remodeling. Other genes found to be sequentially upregulated during the cAMP treatment time course include collagenase 3 and extracellular matrix protein.
Cholesterol trafficking and transport.
Apolipoprotein E (ApoE) has previously been shown to be highly upregulated in endometrial tissue during the window of implantation (28) and herein, receptor 2 for ApoE was upregulated during decidualization. ApoE, produced locally in steroidogenic tissues, e.g., the ovary, binds to hydrophobic molecules and functions in cholesterol transport and trafficking (34). Upregulation of ApoE receptor 2, apolipoprotein A1, and apolipoprotein CII with cAMP treatment suggests that these molecules may play an important role in cholesterol transport or steroid hormone activation in the endometrium.
NF-B protein kinase cascade.
One member of the NF-B protein kinase cascade that is upregulated with cAMP treatment is the thyroid receptor interactor (TRIP9). TRIPs are ligands to the thyroid hormone receptor and are dependent upon thyroid hormone for interaction with the receptor (31). TRIPs also show similar ligand-dependent interaction with the retinoid X receptor (RXR), although they do not interact with the glucocorticoid receptor. Another member of this signaling family that is upregulated during stromal cAMP treatment is glycoprotein CANAG-50-specific IgG1-
. The NF-
B pathway is known to play a role in the baboon endometrium during decidualization (48). Current data suggest that IL-1ß activates multiple signaling pathways that either positively (in the absence of exogenous cAMP) or negatively (in presence of exogenous cAMP) regulate decidualization and IGFBP-1 gene expression in vitro, involving NF-
B activation as well as phosphorylation of p38 MAPK (48).
GABA receptor signaling.
The GABA-A receptor -subunit, upregulated with cAMP treatment, is abundantly and specifically expressed in the rat uterus (13) and was found to be upregulated in our previous microarray studies during stromal decidualization (39) and in the implantation window of human endometrium (28). The
-subunit decreases responsiveness to progesterone metabolites, such as allopregnanolone, and may inhibit uterine contractility prior to the onset of labor (13). The presence of this subunit in the decidualized stromal cell suggests a role for it during implantation, perhaps by modulating the effects of ligands such as GABA and progesterone metabolites to the GABA-A
-receptor.
Integrin receptor signaling.
LFA-1 (leukocyte function-associated antigen-1) -subunit precursor is associated with integrin receptor signaling. LFA-1 has been shown to increase in secretory endometrium, suggesting that its expression is hormonally dependent (14), consistent with the current study. LFA-1 plays a critical role in promoting NK cytolysis in peripheral blood lymphocytes, through potential mechanisms such as promoting cell-mediated target cell adhesion, lysis and apoptosis (14). It may be important in early pregnancy, as evidenced by significantly increased LFA-1 levels in decidual CD56 NK cells in women with spontaneous abortion, relative to normal pregnancy (14). Other genes upregulated with cAMP treatment, herein, include MDC2
/MKC2ß, disintegrin, and B-6 extracellular matrix receptor.
Cell adhesion.
Several cell adhesion genes were upregulated during cAMP treatment. For example, sialophorin, CD43, is a sialoglycoprotein expressed on the surface of a wide variety of blood cells including T lymphocytes. Its ligation induces proliferation and activation of human T lymphocytes (4). Intraepithelial leukocytes expressing CD43 increase from the proliferative to the late secretory phase, where higher levels of CD43-positive cells are found in the surface epithelium compared with glandular epithelium, and it is also expressed in stroma (7). It is hypothesized that during the secretory phase of the menstrual cycle and in early pregnancy, specific leukocyte recruitment to the endometrium limits the type of immune cells which gain access to the endometrium (4). One of these leukocyte recruitment molecules, neural cell adhesion molecule (NCAM), is upregulated with cAMP treatment. Previous studies have also shown that decidual infiltrating lymphocytes express NCAM, and thus it is hypothesized to play a role in endometrial lymphocyte recruitment and adhesion (4).
Cell death.
Several genes with functions in cell death were found to be upregulated with endometrial stromal cell cAMP treatment including T cell death associated protein, Bik, inhibitor of bcl-2, and BH3 interacting domain death agonist BID. BIK, a pro-apoptotic member of the bcl-2 family, is induced by p53 stimulation of apoptosis (16). BIK may act to induce cell-death by its known role in the initiation of cytochrome c release from the mitochondria and in the induction of caspase expression (16). It is unclear why cell death genes are induced in stromal cells treated with cAMP, since the majority of apoptosis during the menstrual cycle occurs in the epithelium (11).
Downregulated Genes
Some of the most markedly downregulated genes upon cAMP treatment include activator of NF-B, L-kynurenine hydrolase, actin/tropomyosin/calmodulin binding protein, cyclin B, IGFBP-5, lipocortin III, FRP (frizzled homolog), and cyclin E2. The activator of NF-
B (average downfold regulation of 369 at 48 h, relative to time 0), is a member of the cytokine-mediated IL-1R/I-
B/NF-
B activation cascade (9). It exhibits structural and functional similarities with the Drosophila Toll/Cactus/Dorsal signaling pathway, where the homologous Drosophila gene Toll (dToll) regulates dorsal-ventral polarity in the developing embryo and activates the innate immune response in the adult fly (9). Recent evidence suggests that a human homolog of the dToll protein, TIL-4 (found to be significantly downregulated, Supplemental Table A), participates in the regulation of both innate and adaptive human immunity through the activation of NF-
B and the expression of the NF-
B-controlled genes IL-1, IL-6, and IL-8 (9). The human TIL-4 (Toll/IL-1R-like-4) gene exhibits homology to both the leucine-rich repeat extracellular domains and the IL-1R-like intracellular domains of Drosophila Toll (9). Functional studies showed that TIL-4 activates NF-
B in a cell type-dependent fashion (9). As the TIL-4 gene is believed to play a similar role in immune regulation as its Drosophila homolog Toll, downregulation of TIL-4 in the decidualized stromal cell may play a key role in inhibiting the maternal immune response to the invading cytotrophoblast and the embryo during implantation, as well as in regulation of endometrial stromal cell decidualization.
IGFBP-5 mRNA is the only member of the IGFBP family with increased expression in the proliferative phase vs. the secretory phase of the menstrual cycle, consistent with our finding of downregulation of IGFBP-5 upon cAMP treatment of the stromal cell (59). IGFBP-5 demonstrates a diffuse stromal pattern of expression in the endometrium (59). The IGF system plays a fundamental role in endometrial biology, acting via autocrine and/or paracrine mechanisms, with IGF-I and IGFBP-5 being dominant in the proliferative phase, and IGF-II and the other IGFBPs predominant in the secretory phase of the menstrual cycle (59).
L-Kynurenine hydrolase, an enzyme involved in the pathway of tryptophan metabolism, was downregulated by 169.3-fold at 48 h. L-Kynurenine hydrolase activity is inhibited by natural and exogenous estrogen, as evidenced by increased urinary excretion of tryptophan metabolites upon treatment with estrogen (57). Evidence suggests that this inhibition is an estrogen-mediated decrease in the availability of vitamin B6, the coenzyme of kynureninase, although estrogen may also exhibit a direct effect on kynureninase activity (57). The roles in endometrium for this interesting gene may involve upregulation of vitamin B6 availability in this tissue during the secretory phase of the menstrual cycle.
Phospholipase C hydrolysis of phosphoinositide (PI) results in the generation of cyclic and noncyclic inositol phosphates (51). Cyclic inositol phosphohydrolase (cIPH) is a phosphodiesterase that cleaves the cyclic bond of one of the products of the phospholipase C reaction, cyclic inositol monophosphate (51). It has recently been purified from the human placenta and is identical to lipocortin III (51), a gene found to be markedly downregulated during cAMP treatment (Supplemental Table A) and during the window of implantation (28). Lipocortin III is also known as annexin III and as placental anticoagulation protein III. Regulation of PI intermediates is likely to be important during stromal decidualization and awaits further study.
Frizzled-related protein (FRP) is a member of the Wnt family and is downregulated in the endometrium during cAMP treatment (Supplemental Table A) and during the window of implantation (28). Roles of the Wnt family in endometrial stromal cell function are currently under study in our laboratory.
Cyclin B and cyclin E2, cell cycle regulatory genes, were downregulated with cAMP treatment, suggesting that decreased expression of these mediators inhibits cell division in stromal decidualization. Cyclin B expression is regulated by progesterone in a breast cancer cell line (T47D-YB) (20), where progesterone had a biphasic effect on cell growth: initially accelerating breast cancer cells through the first mitotic cycle and then subsequently arresting them in late G2 of the second cycle (20). The G1 arrest was associated with decreased levels of cyclin D1, D3, and E, and disappearance of cyclin A and B, with induction of cyclic-dependent kinase inhibitors p21 and p27 (Kip1) (20). In concordance with progesterones effects of stimulating initial cell replication and then promoting cell cycle arrest, the activity of the cell cycle-dependent protein kinase, cdk2, is regulated biphasically by progesterone: it increases initially, then decreases (20). A second treatment with progesterone cannot restart proliferation despite adequate levels of transcriptionally competent progesterone receptor. Instead, a second progesterone dose delays the fall of p21 and enhances the rise of p27 (Kip1), thereby intensifying the inhibition on cell cycle progression (20). It is postulated that the G1 arrest after progesterone treatment is accompanied by cellular changes that permit other, possibly tissue-specific, factors to influence the final proliferative or differentiative state (20). Cyclin E2 is a cell cycle regulatory gene that associates with Cdk2 in a functional kinase complex and regulates the G1/S transition (21). Overexpression of cyclin E2 in mammalian cells accelerates G1, demonstrating that cyclin E2 may be rate limiting for G1 progression (21).
Models of In Vitro Decidualization in Human Endometrial Stromal Cells
The traditional model of in vitro decidualization of the human endometrial stromal cell involves the treatment of stromal cells with progesterone after estrogen priming (1). The decidual morphologic phenotype has been described in several in vitro studies in the literature of treating endometrial stromal cells with activators of the PKA pathway, in the absence of other known decidualizing stimuli (6, 37, 39, 53). Several examples of the characteristic morphologic and secretory phenotype of the stromal cell upon treatment with cAMP in vitro (in the absence of E2/P4), have been published in the literature (6, 37, 39, 53), correlating with our current microarray study where we found evidence of the characteristic morphologic changes and secretory products (i.e., IGFBP-1) of the decidualized stromal cell after 48 h of treatment with cAMP.
Mizuno et al. (37) compared the treatment of human endometrial stromal cells with progesterone, medroxyprogesterone acetate (MPA), prostaglandin E2, or 8-Br-cAMP, and found that treatment with 8-Br-cAMP resulted in the most rapid decidualization response and the highest level of differentiation (as measured by morphology and secretion of known products of decidualized stromal cells, i.e., IGFBP-1). In this study it was found that both progesterone and MPA required more than 2 wk to induce decidualization in vitro (37). Based upon these results, the authors speculated that there are two independent signals in decidualization in vitro, a stronger and more rapid cAMP-mediated signaling pathway and a progesterone receptor-mediated signaling pathway (37).
The ability of cAMP to mediate in vitro decidualization independently of progesterone was demonstrated by Tang et al. (53). They demonstrated that 8-Br-cAMP induces prolactin secretion in human endometrial stromal cells in addition to provoking the differentiation of the fibroblast-like stromal cells to the decidualized phenotype, as evidenced by both morphologic changes and by the expression of the products characteristic of decidualized cells, e.g., IGFBP-1, desmin, hsp 27, and laminin (53).
A manuscript by Brar et al. (6) demonstrated the importance of cAMP for decidualization and that the effects of progesterone on decidualization may be mediated via the cAMP signaling pathway. Progesterone treatment of endometrial stromal cells resulted in increased cAMP levels that positively correlated with levels of prolactin secreted into the conditioned medium (53). In addition, all-trans retinoic acid, which attenuates progesterone-dependent decidualization, significantly decreased both cAMP levels and the secretion of prolactin by endometrial stromal cells treated with progesterone, possibly implicating that the effects of progesterone on decidualization are mediated via the cAMP pathway (54). In addition, the PKA inhibitor, 8-bromoadenosine-3',5'-cyclic monophosphorothioate, significantly suppressed progesterone-dependent prolactin expression, also suggesting that progesterones effects on decidualization are mediated through cAMP (55).
A recent manuscript by Yoshino et al. (58) describes progesterones inhibition of the phosphorylation of AKT/PKB and that H-89, a PKA inhibitor, inhibits this effect, further supporting the PKA pathway as a mediator of progesterones action in endometrial stromal cells.
Despite this strong evidence of the PKA pathway mediating decidualization of human endometrial stromal cells in vitro, this signaling pathway may not be the sole mechanism mediating progesterone-induced decidualization. While experimental evidence suggests that progesterone exhibits its effects on decidualization through the cAMP pathway (12, 53, 54, 58), progesterone may mediate its effects on decidualization through additional signal transduction pathways than cAMP. In our previous microarray study (8) on endometrial stromal cell decidualization in vitro in response to cAMP or progesterone (after estradiol priming), most genes were coordinately regulated by both treatments, strongly suggesting PKA as a mediator of progesterone action in the stromal cell. Although the PKA pathway may be a mediator of progesterones induction of decidualization in the stromal cell, it may not be the only mechanism of action.
Phenotypic Changes upon Treatment with cAMP
The decidualized human endometrial stromal cell phenotype has been observed in several in vitro studies of treating endometrial stromal cells with activators of the PKA pathway (6, 37, 39, 53) in the absence of other decidualizing stimuli. These studies correlate with our current microarray study where we found evidence of the characteristic morphologic changes and secretory products (i.e., IGFBP-1, prolactin, hsp 27, laminin) of the human endometrial decidualized stromal cell after 48 h of treatment with cAMP. After 24 h of treatment with cAMP, it was noted that the stromal cells began a transformation from elongated spindle-like fibroblasts into polygonal epithelial-like cells. In parallel with the change in the morphologic phenotype of the stromal cells observed at 24 h of treatment with cAMP, the levels of IGFBP-1 protein measured to be secreted from the stromal cells increased markedly between 12 and 24 h. At 48 h, the time point of maximal IGFBP-1 secretion over the time course studied, we found a predominance of the cells treated with cAMP which had transformed into polygonal epithelial-like cells, characteristic of decidualized cell morphology. Many of the genes we found in our microarray which play a known or speculated role in mediating the morphologic or secretory changes during decidualization of the human endometrial stromal cell (i.e., IGFBP-1, prolactin, activin subunits, VEGF, IL-11, EBAF, and other TGF-ß family members) also increased in expression markedly between 24 and 48 h of treatment with cAMP, correlating with the timing of the phenotypic changes observed during our time course (Table 8).
|
|
Although the current study opens new opportunities to explore the function of a diverse group of gene families regulated by cAMP and in the events underlying decidualization, it should be noted that in vitro endometrial stromal cells may exhibit different patterns of behavior and gene expression under in vivo conditions where endocrine factors and/or paracrine interactions may affect decidual gene programming. Comparing patterns of gene expression in vitro and in vivo is critically important to validate these data, and additional validation with animal models is essential to provide insight into mechanisms underlying endometrial stromal cell decidualization and functions of the decidualized endometrial stromal cell during implantation as it communicates with the conceptus, immune cells, and other cell populations present during implantation. Derived functions should provide insight into endometrial disorders that have a major impact on womens reproductive health, normal pregnancy, and pregnancy disorders, as well as fetal growth and development. Candidate genes necessary for decidualization may also be utilized as diagnostic screens and targets for drug discovery in women with endometrial-based infertility, and identification of the molecular mechanisms underlying decidualization and implantation can lead to the development of novel contraceptives (Fig. 11).
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: L. C. Giudice, Division of Reproductive Endocrinology and Infertility, Dept. of Gynecology and Obstetrics, Stanford Univ. School of Medicine, 300 Pasteur Drive, Rm. HH-333, Stanford, CA, 94305-5317 (E-mail: giudice{at}stanford.edu).
10.1152/physiolgenomics.00066.2003.
1 The Supplementary Material for this article (Tables A and B) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00066.2003/DC1.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|