In search of candidate genes critically expressed in the human endometrium during the window of implantation

S. Mirkin1, M. Arslan1,4, D. Churikov1, A. Corica3, J.I. Diaz2, S. Williams5, S. Bocca1 and S. Oehninger1,6

1 The Jones Institute for Reproductive Medicine, Department of Obstetrics and Gynecology, 2 Department of Pathology and 3 Department of Microbiology and Molecular Cell Biology, Eastern Virginia Medical School, Virginia, USA, 4 Department of Obstetrics and Gynecology, Mersin University, Mersin, Turkey and 5 Pacific Gynecology Specialists, Seattle, Washington, USA

6 To whom correspondence should be addressed at: The Jones Institute for Reproductive Medicine, 601 Colley Avenue, Norfolk, VA 23507, USA. Email: oehninsc{at}evms.edu


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
BACKGROUND: In this prospective randomized blinded clinical trial, we examined gene expression profiles of the human endometrium during the early and mid-luteal phases of the natural cycle. METHODS: An endometrial biopsy was performed on day 16 (LH +3) or on day 21 (LH +8), followed by RNA extraction and microarray analysis using an Affymetrix HG-U95A microchip. Data analysis was carried out using pairwise multiple group comparison with the significance analysis of microarrays (SAM) software. RESULTS: With a false discovery rate of 0, the analysis revealed that 107 genes were significantly and differently expressed (≥2-fold) during the early versus the mid-luteal phase of the cycle. Forty-five of these genes have not been previously linked to endometrial receptivity. Validation of the microarray data was accomplished using semiquantitative RT–PCR. We demonstrated the presence of estrogen and progesterone response elements (ERE and PRE) by analysis of the 5'-flanking regions of a subset of differentially regulated genes. CONCLUSIONS: Using a strict bioinformatics approach of microarray data, we demonstrated significant changes in candidate genes during the transition of the early to the mid-luteal phase of the human endometrium that may have functional significance for the opening and maintenance of the window of implantation.

Key words: endometrium/gene expression/implantation/luteal phase/microarray


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Implantation is an early and critical event in human pregnancy. At 6–7 days post-fertilization, the free-floating blastocyst is liberated from its zona pellucida, attaches to the endometrial lining and invades the underlying stroma (Psychoyos, 1994Go). The dynamics of these spatial interactions depend on a strict temporal sequence of estradiol (E2) priming that induces endometrial proliferation followed by progesterone differentiation, resulting in the establishment of a ‘window of implantation' (Wilcox et al., 1999Go).

Although there is still not full agreement as to the exact timing of embryo implantation in the human, clinical studies suggest that the window is temporally confined to days 20–24 of a normal, ovulatory cycle (Psychoyos, 1994Go). Data from the assisted reproduction setting have demonstrated that the optimal time for embryo transfer to the uterus is ≤3 days, the so-called ‘window of receptivity' (Navot et al., 1991Go). For 2 day old IVF embryos, the window extends between the third to the fifth day following treatment with exogenous hormones (E2 and progesterone) (Bergh and Navot, 1992Go). The dynamics of the transition of the non-receptive to a receptive endometrium are poorly understood, but in humans an increased rate of spontaneous abortions was observed with human embryo transfers performed beyond the putative window of receptivity (Navot et al., 1991Go).

The correct temporal–spatial elaboration and balance of various growth factors, cytokines, lipid mediators, transcription factors and other putative molecules regulated by steroid hormones is thought to play an important role in uterine preparation for implantation (Develioglu et al., 1999Go; Brown et al., 2000Go; Carson et al., 2000Go; Lim et al., 2002Go; Giudice, 2003Go). Estrogen and progesterone actions are primarily mediated by their nuclear receptors, estrogen receptor (ER) ER{alpha} and ER{beta} and progesterone receptor (PR) PR-A and PR-B respectively (Lessey, 2004Go). The rise in progesterone during the early secretory phase is responsible for the expression of a myriad of proteins, many of which appear to be critical for normal implantation. This effect may be direct (acting through PR in the epithelial cells) or indirect by stimulation of stromal factors that, in turn, induce specific epithelial gene products (Lessey, 2004Go).

In the pre-genomic era, a ‘one-by-one' approach was adopted to investigate gene expression during the window of implantation. However, it is more realistic to view the process of normal implantation as a condition of equilibrium in the up- and down-regulation of diverse endometrial genes under control of steroid hormones and probably other local (paracrine and autocrine) regulatory factors (Reese et al., 2001Go). Altered expression of endometrial regulatory genes/proteins has been postulated as an underlying cause of infertility and recurrent pregnancy loss (Kao et al., 2003Go). A genomic-wide approach using microarray technology now allows us to investigate global gene expression during the various phases of endometrial development under physiological and pathological conditions.

Recent studies have addressed endometrial gene expression using microarray technology during the different phases of the normal menstrual cycle (Kao et al., 2002Go; Borthwick et al., 2003Go; Ponnampalam et al., 2004Go). Additional studies addressed gene expression profiles in the early versus the mid-luteal phase of the natural cycle with differential expression being reported for some subsets of genes (Carson et al., 2002Go; Riesewijk et al., 2003Go). Such an approach can be used to further understand gene expression profiles associated with the framing (initiation, maintenance and closure) of the window of implantation.

Microarray technology provides a novel sensitive technology with simultaneous analysis of the expression of thousands of genes. Issues of intra- and inter-assay variability and the complexity of expression patterns need a valid and stringent statistical assessment. A bioinformatics approach using pairwise multiple group comparison with the significance analysis of microarrays (SAM) software is estimated to be superior to conventional statistical methods for analysing microarray data (Tusher et al., 2001Go).

The objective of this prospective, randomized and blinded clinical study was to examine gene expression profiles of the endometrium during the normal menstrual cycle (natural cycle) and to compare changes occurring during the early versus the mid-luteal phase. For this purpose, normal healthy volunteers were subjected to an endometrial biopsy, following identification of the urinary LH surge on day LH +3 (early luteal phase) or on day LH +8 (mid-luteal phase, during the opening phase of the putative window of implantation). Endometrial tissue RNA was extracted and gene expression was examined by microarray technology. Gene expression of selected genes was further validated using semiquantitative RT–PCR. In addition, we analysed 5'-flanking regions of some differentially expressed genes for presence of the estrogen and progesterone response elements (ERE and PRE). By using a bioinformatics approach with stringent criteria for data analysis, our studies extended previous and related publications in the search for gene changes that might be critical for the establishment of the window of implantation in the human.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Subjects
A total of 10 female subjects were enrolled for these studies between 2002 and 2003. These were healthy women participating in our oocyte donation programme (oocyte donors), ranging in age between 24 and 32 years, with regular menstrual cycles and previously confirmed ovulation, and who were of proven fertility. All women had a normal uterus as assessed by transvaginal ultrasonography. Enrolled volunteers signed an informed consent approved by the Institutional Review Board at Eastern Virginia Medical School. Subjects were requested to use condom contraception during the preceding and study cycles.

Study groups and timing of the endometrial biopsy
Ten subjects were randomized using sealed envelopes to be biopsied in the early or the mid-luteal phase of the natural cycle. Two subjects were excluded because their biopsies were out of phase. As a consequence, the following two groups (total of eight subjects) were studied: group I (n =3) volunteers who underwent an endometrial biopsy during a defined day of the early luteal phase (day 16, representative of the pre-receptive secretory endometrium); and group II (n =5) subjects who underwent an endometrial biopsy during a defined day of the mid-luteal phase (day 21, estimated to be representative of the receptive endometrium) (Figure 1).



View larger version (21K):
[in this window]
[in a new window]
 
Figure 1. CONSORT statement flow diagram.

 
Subjects performed a daily morning assessment of the urinary LH surge beginning on day 10 of the natural cycle using a commercially available ovulation predictor kit (Ovuquik One Step; Quidel, San Diego, CA, USA). In group I, subjects underwent an endometrial biopsy on day 16 (LH +3, where LH =0 is the day of the urinary LH surge and LH +1 is the day of ovulation or day 14); subjects from group II were biopsied on day 21 (LH +8) (Develioglu et al., 1999Go; Carson et al., 2002Go; Kao et al., 2002Go; Devroey et al., 2004Go; Mirkin et al., 2004bGo). Physicians performing the endometrial biopsies remained blinded as to the menstrual cycle day.

Endometrial tissue specimens
Endometrial biopsies were performed using a Pipelle catheter (Unimar, Bridgeport, CT, USA) by an experienced investigator under sterile conditions, from the uterine fundus. Extreme care was taken to ensure that enough tissue and full depth was obtained from each biopsy. Each sample was divided into two parts: one portion was fixed in 10% formalin and processed for histological evaluation (haematoxylin–eosin, H–E) and a second portion was immediately (within a minute) frozen in liquid nitrogen for subsequent RNA isolation.

Histological evaluation
For endometrial dating, formalin-fixed 4 µm sections stained with H–E and periodic acid–Schiff stain were evaluated. All endometrial biopsies were coded and evaluated blindly according to the histopathological criteria of Noyes et al. (1950)Go. Only those biopsies where histological dating was in accordance with their temporal dating (day 16 or day 21) were further analysed.

RNA preparation/target preparation/array hybridization and scanning
For microarray analysis, each endometrial biopsy was processed individually for microarray hybridization at Genome Quebec Innovation Centre, McGill University (Montreal, QC, Canada) as previously reported (samples were not pooled) (Noyes et al., 1950Go). Total RNA was isolated using Trizol reagent (Life Technologies, Carlsbad, CA, USA) following the supplier's protocol. RNA was then further purified using the RNeasy total RNA clean-up protocol (Qiagen, Valencia, CA, USA). The integrity of the RNA samples (RNA QC procedure) was assessed by using the 2100 bioanalyzer (Agilent, Santa Clara, CA, USA), running an aliquot of the RNA samples on the RNA 6000 Nano LabChip (Agilent). The Bioanalyser results are indicative of RNA integrity based on a procedure that involves the separation of RNA into different-sized fragments by capillary electrophoresis molecular sieving, with results shown on an electropherogram. Probe synthesis, hybridization and scanning were done according to the Affymetrix protocol (Affymetrix, Sunnyvale, CA, USA).

A single round of linear amplification of RNA samples was performed by the In Vitro Transcription T7-promoter method (Van Gelder et al., 1990Go) with the Bioarray High Yield RNA transcript labelling kit (Enzo Diagnostics). The T7 primer was added to the total RNA in the very first step following the initial RNA QC procedure, prior to reverse transcription. The primer–RNA mixture was denatured for 10 min at 70°C and then chilled on ice for 2 min. Annealing of the T7 primer to RNA was performed by mixing 100 pmol/l T7-(T)24 primer [Genset, 5'-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT)24–3'] with 10 µg of RNA sample in a 10 µl volume and incubating at 37°C for 3 min. First-strand cDNA synthesis was performed using 2 µl Superscript II reverse transcriptase (Invitrogen Life Technologies, Carlsbad, CA, USA) in a 20 µl reaction volume containing 10 µmol/l dithiothreitol, 500 µmol/l each dNTP, and 1x first-strand buffer (all from Invitrogen Life Technologies) for 60 min at 42°C, resulting in the incorporation of the T7-promoter sequence (primer bound) into first strand cDNA.

Second-strand synthesis was performed by adding to the first-strand reaction mix: 40 units DNA polymerase I (Invitrogen Life Technologies), 10 IU Escherichia coli DNA ligase (Invitrogen Life Technologies) and 2 IU RNase H (MBI Fermentas, Hanover, MD) in a final reaction volume of 150 µl containing 1x second-strand buffer (Invitrogen Life Technologies). The reaction mixture was incubated at 16°C for 2 h. Ten IU of T4 DNA polymerase (MBI Fermentas) was added and incubated at 16°C for 5 min followed by the addition of 10 µl of 0.5 mol/l EDTA (Sigma, The Woodlands, TX, USA).

After second-strand synthesis, the cDNA was purified by phenol–chloroform extraction with Phase-Lock tubes (Eppendorf, Westbury, NY, USA), precipitated and redissolved in 20 µl nuclease-free water. The purified cDNA was used to generate the biotinylated cRNA probe with the Bioarray High Yield RNA transcript labeling kit (Enzo diagnostics, Farmingdale NY, USA) as indicated by the supplier. The probe synthesis reaction was performed at 37°C for 5 h with occasional agitation using T7 RNA Polymerase. The labelled cRNA was then purified using the RNeasy total RNA clean-up protocol (Qiagen), eluted in 60 µl of nuclease-free water and quantified by spectrophotometry.

An aliquot of the purified cRNA was analysed on RNA 6000 Nano LabChip (Agilent) to verify the integrity and size distribution. Twenty milligrams of cRNA was fragmented by heating at 94°C for 35 min in 1x fragmentation buffer (40 mmol/l Tris–acetate pH 8.1; 100 mmol/l KOAc; 30 mmol/l MgOAc), to reduce the probe length. The hybridization mixture was prepared by mixing 15 mg of the biotinylated probe cRNA with control oligonucleotide B2 (final concentration 50 pmol/l; Affymetrix), herring sperm DNA (final concentration 0.1 mg/ml; Research Genetics), acetylated bovine serum albumin (final concentration 0.5 mg/ml; Invitrogen Life Technologies) in a final volume of 300 µl of 1x MES [2-(N-morpholino)ethanesulphonic acid] hybridization buffer (100 mmol/l MES, 1 mol/l NaCl, 20 mmol/l, 0.01% Tween-20; all reagents from Sigma). The hybridization mixture was denatured for 10 min at 99°C, incubated for 5 min at 45°C, and centrifuged for 5 min in a benchtop microcentrifuge. The microarray was warmed to room temperature and prehybridized with 200 µl of 1x hybridization buffer for 10–20 min at 45°C. The prehybridization solution was removed and 200 µl of the hybridization mix was added to the array. The array and probe fragments were incubated at 45°C overnight (16–20 h) in a rotating oven (Affymetrix).

We used the Affymetrix HG_U95Av2 array containing 12 686 human genes and expressed sequence tags (EST). After hybridization, the hybridization cocktail was removed from the chip and stored at –80°C. The chip was immediately placed in the Affymetrix GeneChip Fluidics Station 400 (Affymetrix). In total, 10 low-stringency washes (63 SSPE, 0.01% Tween-20, 0.005% Antifoam) and four high-stringency washes (100 mmol/l MES, 0.1 mol/l NaCl, 0.01% Tween-20, 50°C) were performed (all reagents from Sigma). The array was then incubated with SAPE (streptavidin/phycoerythrin stain; Molecular Probes), followed by 10 low stringency washes. The array was incubated with biotinylated anti-streptavidin antibody (Vector Laboratories) and washed again with 15 low stringency washes. Specifically bound probe was detected by placing the array in the Agilent GeneArray scanner 2500 (Affymetrix). The scanned images were analysed using the Microarray Analysis Suite version 5.0 (Affymetrix). Detailed information on the gene array system is available at www.affymetrix.com and www.genomequebec.mcgill.ca.

Semiquantitative expression analysis using RT–PCR
To validate microarray findings, we quantified the expression of some genes using semiquantitative RT–PCR. Three of the selected genes, ANXA4, SPP1 and FOXO1A, were up-regulated, and two genes, PIP5K1B and MSX1, were down-regulated during the receptive phase. Primers were designed using LightCycler Primer design software (Roche, Indianapolis, IN, USA) as follows: ANXA4 (GenBank accession no. M82809) forward primer 5'- TAAAACGCCTACAGCTGCCT-3', reverse primer 5'- TAAGCTTTGAAATGCAAGTACAGC-3'; SPP1 (GenBank accession no. J04765) forward primer 5'-TGAGAGCAATGAGCATTCCGATG-3', reverse primer 5'-CAGGGAGTTTCCATGAAGCCAC-3'; FOXO1A (GenBank accession no. AF032885) forward primer 5'- GCTACTCGTTTGCGCCACCAAAC-3', reverse primer 5'- CCGTGTGGGGCAGGGGACG-3'; PIP5K1B? (GenBank accession no. X92493) forward primer 5'-CACATGGATGAGACGTGAGC-3', reverse primer 5'-CAGTTCTTCACAGTTCAGCAAGC-3'; MSX1 (GenBank accession no. M97676) forward primer 5'- GGCTACAGCATGTACCACCT-3' reverse primer 5'-GTTAAAGGGAAGGCGGCTG-3'.

RT–PCR was performed using a commercially available kit (Perkin-Elmer, Norwalk, CT, USA) (Mirkin et al., 2004aGo). Briefly, 50 ng of mRNA was reverse-transcribed in 10 µl aliquots containing 1 mmol/l dNTP, 2.5 IU/µl of Moloney murine leukaemia virus reverse transcriptase, 1 IU/µl RNase inhibitor and 2.5 µmol/l random hexamers in a buffer containing 50 mmol/l KCl, 10 mmol/l Tris–HCl pH 8.3, and 5 mmol/l MgCl2 for 40 min at 42°C, followed by heating to 99°C for 5 min and finally cooling to 5°C for 5 min.

For semiquantitative PCR, ANXA4, SPP1, MSX1, FOXO1A and PIP5K1B cDNA were amplified in the same manner with cyclophylin serving as an internal control (primer upstream sequence: 5'-CCATGGTCAACCCCACCGTGTTCTT-3', and primer downstream sequence: 5'-CTGCTGTCTTTGGAACTTTGTCTGC-3').

An aliquot of each reverse transcribed product (10 µl) was added to a final volume of 80 µl of reaction mix containing 2.5 IU of Taq polymerase (Perkin Elmer, Norwalk, CT, USA) 50 mmol/l KCl, 10 mmol/l Tris–HCl pH 8.3 and 1.25 mmol/l MgCl2, the primer pairs for every gene and 1.5 µCi of [{alpha}-33] dCTP (specific activity 3000 Ci/mmol, DuPont, Boston, MA, USA). The polymerase amplification was carried out for 32 cycles for ANXA4 and SPP1, 36 for MSX1 and 38 for FOXO1A and PIP5K1B after estimating the exponential phase for each set of co-amplification reactions.

The following protocol was used for all genes with minimal differences. cDNA amplification was initiated with denaturation for 3 min at 94°C followed by 32 cycles of 1 min denaturation at 94°C, 1 min annealing at 55°C and 72°C extension cycle (1 min). Each PCR product was separated by 7% polycrylamide gel electrophoresis and visualized with ethidium bromide, excised from each gel and dissolved overnight using 0.5 N quaternary ammonium hydroxide in toluene (Soluenen 350; Packard Instruments Co., Meriden, CT, USA). After the addition of scintillation fluid, the samples were measured by scintillation spectroscopy. The mRNA results are expressed as a ratio of counts per min (cpm) of ANXA4, SPP1, MSX1, FOXO1A and PIP5K1B-I{beta} divided by cyclophylin.

Identification of ERE and PRE: in silico promoter analysis
We retrieved a 3500 bp long sequence encompassing 3000 bp upstream and 500 bp downstream of the 5' end of the first exon of each gene. This length of the promoter/regulatory region was chosen based on the positions of the functional ERE reported previously (Klinge, 2001Go). For the majority of the genes, we used the 5' end Information Extraction (FIE v.2.0) program (Chong et al., 2003Go) to extract the necessary region. In instances where FIE could not retrieve the necessary sequence, we determined it by blasting the reference mRNA sequence for the particular gene against the human genome database available at NCBI (http://www.ncbi.nlm.nih.gov/genome/seq/HsBlast.html).

Having identified the 5'-most position of the mRNA sequence on the respective genomic contig, we then manually recovered the 3500 bp sequence around this position. To search for putative ERE in the regulatory regions we used ERE Finder software ver. 2.0 (Bajic et al., 2003Go) with a sensitivity threshold set to 0.87. The presence of potential ERE was also investigated using the MATCHTM program available from the Gene Regulation website (http://www.gene-regulation.com/cgi-bin/pub/programs/match). ER (M00191) nucleotide distribution matrix (TRANSFACR 6.0 database at the same website) was used for ERE predictions with matrix similarity and core similarity set to the stringency levels optimal to reduce false positives. Due to the lack of nucleotide position matrix available for progesterone receptor in the TRANSFACR 6.0 database, the search for putative PRE was conducted using a pattern-based method implemented in the GeneQuest module of Lasergene software (DNASTAR, Inc.). The sequence pattern 5'-G G/A G/T/A AC A/G NNN TGTTCT-3' (Forman and Samuels, 1990Go) and more recent study of the progesterone receptor binding was chosen based on the widely acceptable PRE consensus AGAACA NNN TGTTCT affinity and specificity (Nelson et al., 1999Go). The sensitivity threshold for pattern search was set to 87%, which allows a maximum of two mismatches; and only forward DNA strand was investigated.

Statistical analysis of microarray data
The statistical analysis of the microarray data (SAM) was independently performed by a bioinformatics group (Incogen, Williamsburg, VA, USA) as previously reported (Mirkin et al., 2004bGo). Multiple, pairwise group comparisons were performed using public software http://www.stat.stanford.edu/~tibs/SAM/ for the significant analysis of microarrays (Draghici and Kuklin, 2003Go). The software performed 1000 random permutations of group labels on the original data set to calculate the false discovery rate on the basis of t-statistics or ‘score' (difference between the group means in the units of SD) assuming equal group variance. Thus, significant genes were selected independent of their expression levels when their ‘score' was above the specified threshold ‘delta' for the difference between the ‘observed score' (t-statistics for true labels), and the average ‘expected score' (t-statistics for the randomly permuted data). In addition, SAM allows setting the detection threshold for the fold-change in gene expression, which is the ratio of the mean expression levels for this gene in the groups under comparison. A q-value was assigned to each detectable gene in the array. This q-value is similar to the familiar P-value, measuring the false-discovery rate (FDR) at which a gene is called significant (Hastie et al., 2001Go). A median false significant number refers to the median falsely called genes (false positive calls).

Statistical analysis of the semiquantitative RT–PCR
The statistical analysis was performed using the least significant difference (LSD) with P<0.05 indicating significant difference. The analysis was performed using SAS v.8.0 (SAS Institute Inc., Cary, NC, USA).


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Microarray results
Analysis of the microarray data obtained from the pre-receptive (LH +3) versus the receptive (LH +8) luteal phase endometrium was performed by pairwise multiple group comparison using SAM. With a false discovery rate (FDR) of 0 and a pre-defined {Delta} threshold of 2, the analysis revealed a consistent pattern of differential gene expression. In total, 107 genes were significantly differently expressed during the early versus the mid-luteal phase of the cycle. Figure 2 shows the SAM plot for this comparison, the y-axis plots the value of the ‘observed score' (distance between means in the units of SD for a particular gene) for each gene, and the x-axis plots the ‘expected score' (the ‘average random' score obtained for this gene with random permutations of group labels for all replica) for each gene; the diagonal line shows where the genes with random score (false positives statistical significance) would fall. The distance between dashed lines is ‘delta' threshold that was applied for detection of ‘false positives'. Of these 107 genes found by SAM, 49 were up-regulated and 58 were down-regulated when comparing the receptive (day 21) versus the non-receptive (day 16) endometrium.



View larger version (13K):
[in this window]
[in a new window]
 
Figure 2. Significance analysis of microarrays (SAM) plots for the comparison of gene expression during the pre-receptive (LH +3, day 16) and receptive (LH +8, day 21) luteal phase endometrium. Delta applied to this comparison: 2.00. False discovery rate (FDR) applied: 0. The y-axis plots the value of the ‘observed score’ (distance between means in the units of SD for a particular gene) for each gene. The x-axis plots the ‘expected score’ (the ‘average random’ score obtained for this gene with random permutations of group labels for all replica) for each gene. The diagonal line shows where the genes with random score (false positives statistical significance) would fall. The distance between dashed lines is the ‘delta’ threshold that was applied for detection of false positives. The expression of 107 genes/expressed sequence tags was significantly different with a fold-change ranging from –7.45 to 34.50.

 
A total of 49 genes displayed a ≥2-fold significant increase in expression during the putative window of implantation (fold-change range: 2.1–34.5) (Table I). Cell cycle regulators, ion-binding proteins and signal-transporting proteins were the major group of genes up-regulated, accounting for 38% of the total. Twenty one per cent of the genes that were up-regulated belonged to the family of immunomodulators, 19% to the cell adhesion–structural proteins, extracellular matrix components and growth factors–cytokines, 11% to enzymes, whereas genes with unknown functions accounted for 11% of the total.


View this table:
[in this window]
[in a new window]
 
Table I. Genes found to be significantly up-regulated by multiple group pairwise comparisons using the significance analysis of microarrays (SAM) software when comparing the receptive (LH +8, day 21) versus the pre-receptive (LH +3, day 16) human endometrium

 
A total of 58 genes displayed a ≥2-fold significant decrease in expression during the putative window of implantation (fold-change range: 2.01–7.50) (Table II). Genes encoding cell cycle regulators, ion-binding and signal-transporting proteins were the major group of genes down-regulated, accounting for 46% of the total. Enzymes accounted for 25% of down-regulated genes, cell adhesion and structural proteins 11%, the family of immunomodulators 3%, whereas genes with other unknown functions accounted for 16% of the total.


View this table:
[in this window]
[in a new window]
 
Table II. Genes found to be significantly down-regulated by multiple group pairwise comparisons using the SAM software when comparing the receptive (LH +8, day 21) versus the pre-receptive (LH+3, day 16) human endometrium

 
Semiquantitative RT–PCR results
Semiquantitative RT–PCR was used to verify the changes in mRNA expression levels indicated by microarray analysis. Five genes were selected for this purpose. SPP1, ANXA4 and FOXO1A were significantly up-regulated 10-, 6- and 5-fold respectively (P<0.05 for all comparisons between early and mid-luteal phase). On the other hand, MSX1 (7-fold) and PIP5K1B (3-fold) were significantly down-regulated (P<0.05), also confirming microarray results (Figure 3).



View larger version (14K):
[in this window]
[in a new window]
 
Figure 3. Validation of selected genes identified by the microarray data from natural cycles in the pre-receptive (LH +3, day 16) and receptive (LH +8, day 21) endometrium using semiquantitative RT–PCR: comparison of annexin IV (ANXA4), secreted phosphoprotein 1 (SSP1 or osteopontin), Forkhead box O1A (FOXO1A), phosphatidylinositol-4-phosphate 5-kinase type I beta (PP5K1B) and msh homeobox homolog 1 (MSX1) mRNA abundance in counts per minute relative to cyclophilin (mean ± SD).

 
In silico promoter analysis
Differential expression of many genes in the cycling endometrium is believed to occur under the control of the sexual steroid hormones, estrogen and progesterone (Lessey, 2004Go). It was our aim to verify if implantation-related genes are directly regulated by estrogen and progesterone receptors. For this purpose we analysed the promoter regions of a subset of genes for the presence of ERE and PRE. Twelve genes were identified following a careful search of previously published reports (Carson et al., 2002Go; Kao et al., 2002Go; Borthwick et al., 2003Go; Riesewijk et al., 2003Go) and our own study (Table III). All such genes were significantly and similarly regulated (≥2-fold up- or down-regulated) during the putative window of implantation in at least three of these five reports.


View this table:
[in this window]
[in a new window]
 
Table III. Endometrial genes significantly up- and down-regulated during the putative window of implantation: comparison of microarray data from four previously published studies (Kao et al., 2002Go; Carson et al., 2002Go; Borthwick et al., 2003Go; Riesewijk et al., 2003Go) and the present results

 
The candidate ERE sequences were found in the 5'-flanking regions of 10 genes among the 12 genes analysed (Table IV). Eight genes had multiple ERE-like sequences. Overall, we analysed 84 000 nt on both DNA strands (7000 nt per gene) and discovered 24 candidate ERE (21 mutually exclusive ERE). The estimated frequency of ERE occurrence in the 5'-flanking regions of these genes is 2.86x10–4 per nt. This is comparable to the frequency of ERE occurrence in the promoter regions of the known estrogen-responsive genes (3.12x10–4 per nt) and substantially higher than the average for random sequence on human chromosome 21 (1.90x10–4 per nt) analysed at the same level of sensitivity (Bajic et al., 2003Go).


View this table:
[in this window]
[in a new window]
 
Table IV. Occurrence of the ERE- and PRE-like sequences in the 5’ regulatory regions of 12 genes differentially regulated during the window of implantation

 
The largest number of candidate ERE (two or more mutually exclusive) was found in SERPING1, ID4, GATA2, MAP3K5 and ANXA4 genes (Table IV). The sequence patterns of the putative ERE found in SERPING1 are also the most similar to the ERE consensus. Sequence patterns very similar to ERE consensus were also found in IL15 (transcript variant 2), C1R and GATA2 gene regulatory regions; however, ERE-like sequence in C1R may also be recognized by glucocorticoid hormone receptor (GR). Two other ERE-like sequence patterns present in CD55 and MAOA genes may also be recognized by both estrogen and glucocorticoid hormone receptors. This is not surprising, because ERE and GRE differ at only two positions in their canonical half-sites (Mader et al., 1989Go). All ERE-like sequences that we have identified in this subset of genes are imperfect palindromes with new sequence patterns (i.e. have not been observed yet in the estrogen-responsive genes). It remains to be established experimentally if any of these putative ERE are functional.

The pattern-based search for PRE in the 5' regulatory regions of the same set of 12 genes revealed PRE-like sequences in seven genes (Table IV). Four genes harboured multiple putative PRE. Since we were searching for an asymmetric PRE sequence pattern, we analysed only plus strand of DNA and recognized 14 PRE-like sequences, which had no more than two mismatches with the pattern that we searched for (see Materials and methods). The estimated frequency of PRE occurrence over all 12 genes was 3.33x10–4 per nt, which is again substantially greater than the average for a number of randomly chosen contigs on human chromosome 21 (1.90x10–4 per nt) analysed in the same way. The largest number of candidate PRE, four and three, was detected for C1R and MSX1 genes respectively; and the sequence most similar to the searched pattern (one mismatch) was detected in GATA2. All of the candidate PRE sequences that we detected were distinct from ERE-like sequences, except for one PRE in C1R, which was also recognized as ERE.


    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
We compared human endometrial gene expression profiles in defined days of the early versus the mid-luteal phase of the normal menstrual cycle. It is estimated that the window of implantation in the human is established and maintained from day 20 to day 24 (Wilcox et al., 1999Go). Consequently, our study aimed to compare the pre-receptive endometrium (day 16; LH +3) versus the early receptive endometrium (day 21; LH +8). Results showed marked changes in the gene expression profiles of the different stages of the luteal phase endometrium. Some of the differentially regulated genes that we identified may be critical for the framing (opening and maintenance) of the window of implantation in the natural cycle.

Using a stringent bioinformatics approach, we identified 107 genes differentially expressed between the early and the mid-luteal phase endometrium (Figure 2). Of these genes, 49 and 58 genes were significantly up- and down-regulated, respectively, when comparing the mid- versus the early luteal phase of the natural cycle. Cell cycle regulators, ion-binding, signalling-transporting proteins, and members of the immunomodulator family, were the major groups of genes displaying significant changes in the transition from the pre-receptive to the receptive endometrium. Such genes accounted for 38% of all genes significantly up-regulated and for 46% down-regulated.

The expression of a few of these genes was further validated using semiquantitative RT–PCR (Figure 3). One of them was ANXA4, a gene that was up-regulated 6.50-fold in the receptive endometrium compared to the pre-receptive phase. This gene was already identified in the endometrium as highly expressed during the secretory phase (Byrjalsen et al., 1995Go). Moreover Carson et al. (2002)Go and Riesewijk et al. (2003)Go in similar microarray studies observed the same pattern of regulation.

SPP1 (osteopontin) was 17-fold up-regulated in our array study in the receptive phase and results were confirmed by RT–PCR. This is in concordance with other studies that also showed the up-regulation of this gene during the implantation window (Carson et al., 2002Go; Kao et al., 2002Go; Borthwick et al., 2003Go). Recent reports have demonstrated the presence of osteopontin in uterine flushings from pregnant ewes between days 11 and 17 (Johnson et al., 2000Go), a period that corresponds to the adherence and attachment phase of early implantation in this species. Further, the presence of osteopontin mRNA in glands and protein secretion into the lumen was also demonstrated, the expression of which is regulated by progesterone (Johnson et al., 1999Go).

FOXO1A was up-regulated 5.30-fold in our array and validated by RT–PCR. It is a member of the FOXO subfamily of forkhead/winged-helix family of transcription factors. Forkhead box has been reported to participate in the decidualization of the endometrium (Christian et al., 2002Go) and to play an important role in mediating effects of insulin-like growth factor-binding protein (IGFBP-1) in endometrial cells. IGFBP-1 is a major secretory product of late secretory phase and pregnant endometrium. There are some reports indicating that forkhead box expression is enhanced by steroids in granulosa cells (Richards et al., 2002Go).

Hox family genes correlate temporally and spatially with critical morphogenesis events. One of these genes, MSX1, was 7.45-fold down-regulated in the receptive endometrium in comparison with the early luteal phase and validated by RT–PCR. MSX1 is expressed at high levels in adult mice uterine epithelium, and decreases following embryo implantation (Pavlova et al., 1994Go). This gene was also down-regulated during the window of implantation as reported by Kao et al. (2002)Go and Riesewijk et al. (2003)Go.

PIP5K1B is a gene that regulates a variety of cellular processes including proliferation, survival, vesicular trafficking and cytoskeletal organization, and it was down-regulated 5.53-fold in the array and confirmed by RT–PCR. It has been shown that the down-regulation of PIP5K1B indirectly stimulates colony-stimulated factor 1 (CSF-1) activity (Davis et al., 1997Go). This latter cytokine has been reported to play important roles in the cascade of events that leads to implantation under the influence of steroid hormones (Lindhard et al., 2002Go).

We have mined the existing literature in order to determine whether all the differentially expressed genes in our study had been previously ascribed a role in human embryo implantation. Approximately 40% of the genes showing significant up- or down-regulation in the present study (45 genes) could not be linked to implantation (Tables I and II). Therefore, our results indicate the presence of additional genes of functional significance during the transition of the pre-receptive to the receptive endometrium. Thirteen of these genes have unknown function. Among the identified genes with known function, a cell adhesion molecule (milk fat globule-EGF factor 8 protein a member of the EGF family) and multiple cell cycle regulators are being considered as candidate genes/products to be functionally validated in future studies.

The present study also confirmed and extended previous reports on the analysis of gene expression of the human endometrium during the luteal phase (Carson et al., 2002Go; Kao et al., 2002Go; Borthwick et al., 2003Go; Riesewijk et al., 2003Go). However, some differences in study design are of significance. Here, we applied stringent criteria related to: (i) a comprehensive statistical analysis using a bioinformatics approach; and (ii) timing of the biopsy restricted to a defined day of the pre-receptive (day 16; LH +3) and receptive (day 21; LH +8) stages of endometrial development.

A strict bioinformatics approach is critical for interpretation of microarray data. DNA microarrays contain oligonucleotides or cDNA probes for measuring the expression of thousands of genes in a single hybridization experiment. Although large amounts of data are generated, appropriate statistical methods are needed to determine whether changes in gene expression are significant. Cluster analysis of microarray data can find coherent patterns of gene expression (Eisen et al., 1998Go) but provides little information about statistical significance. Methods based on conventional t-tests provide the probability (P) that a difference in gene expression occurred by chance (Galitski et al., 1999Go; Roberts et al., 2000Go). Although P =0.01 is significant in the context of experiments designed to evaluate changes of a small number of genes, a microarray experiment for 10 000 genes would identify 100 genes by chance. This problem led us to use a statistical method adapted specifically to microarray data.

Pairwise multiple group comparison using SAM software has proved to be superior to conventional methods for analysing microarrays (Tusher et al., 2001Go). SAM identifies genes with statistically significant changes in expression by assimilating a set of gene-specific t-tests. Each gene is assigned a score on the basis of its change in gene expression relative to the SD of repeated measurements for that gene. Genes with scores greater than a {Delta}-threshold are deemed potentially significant. The percentage of such genes identified by chance is the false discovery rate (FDR). To estimate the FDR, nonsense genes are identified by analysing permutations of the measurements. The {Delta}-threshold can be adjusted to identify smaller or larger sets of genes. As {Delta} decreases, the number of genes identified as regulated significantly increased but at a cost of increasing FDR. Importantly, in our study we predetermined a {Delta}-threshold of 2 with a strict FDR of 0.

As mentioned above, the timing of the endometrial biopsy is also critical for data interpretation. A few recent publications have aimed to characterize gene expression during the putative window of implantation and have employed the same microarray chip that we used herein (HG_U95Av2). Two such reports compared endometrial gene regulation during the proliferative versus the mid-luteal phase of the natural cycle (Kao et al., 2002Go; Borthwick et al., 2003Go). As expected, the comparison of the functionally and structurally different stages of development of the proliferative and secretory endometrium demonstrated a large variability of gene expression with up- and down-regulation of multiple genes. In addition, and similar to our study, Carson et al. (2002)Go and Riesewijk et al. (2003)Go compared gene expression profiles of the endometrium during the early and mid-luteal phase. However, those studies differed from ours in several respects.

Carson et al. (2002)Go compared pooled RNA samples from three women who were biopsied on various early luteal phase days (urinary LH +2, +3 or +4) versus women who underwent a biopsy on various mid-luteal phase days (urinary LH +7, +8 or +9). Using the GeneSpring software for statistical evaluation these investigators identified 693 genes differentially expressed between these two groups. The gene expression changes identified by these investigators ranged from –100- to +45-fold. In contradistinction to that study, we analysed samples on two specific days of the luteal phase, day 16 (pre-receptive endometrium) and day 21 (receptive endometrium) following an accurate dating; we did not pool the samples and we analysed every subject independently using SAM software.

On the other hand, Riesewijk et al. (2003)Go biopsied volunteers on day 15 (LH +2) and on day 20 (LH +7) of the natural menstrual cycle. They identified 75 genes differentially expressed (>3-fold) between the early and mid-luteal phase using principal component analysis. These investigators obtained the endometrial tissue from the same menstrual cycle (i.e. subjects were biopsied 5 days apart on the same cycle), so there was a possibility of reactive inflammatory changes and subsequent gene alterations as a result of the first biopsy.

Regardless of the differences among all the above-mentioned studies (Carson et al., 2002Go; Kao et al., 2002Go; Borthwick et al., 2003Go; Riesewijk et al., 2003Go) and ours, 12 genes were significantly and similarly regulated (up- or down-regulated) during the putative window of implantation in at least three of these five reports (Table III). Four such genes (osteopontin, decay accelerating factor for complement, monoamine oxidase A, growth arrest and DNA damage inducible, alpha) were common to three or all of the four previously published reports as emphasized by Horcajadas et al. (2004)Go.

Four of these genes (marked with asterisks in Table IV) have been reported previously to respond to estrogen stimulation in the mouse model of delayed implantation (Reese et al., 2001Go). In addition, SPP1 gene, encoding osteopontin, is up-regulated by progesterone in both mouse skin (Craig and Denhardt, 1991Go) and human trophoblasts (Omigbodun et al., 1997Go). Although induction of expression by steroid hormones has been detected for the aforementioned genes, the mechanism of their transcriptional activation remains unknown. It is reasonable, however, to assume that in the endometrium, transcription of the implantation-related genes is directly regulated by nuclear estrogen and/or progesterone receptors. With this idea in mind, we analysed 5'-flanking regions of the 12 genes (Table IV) for presence of ERE- and PRE-like sequences.

Indeed, we found increased frequency of the ERE- and PRE-like sequences in the overall collection of 5'-flanking regions of these 12 genes as compared to the average for a random sequence from human genome. Putative ERE and/or PRE were detected for all the genes analysed except for GADD45A (Table IV). Several genes have multiple ERE- and/or PRE-like sequences; and some of these elements, such as those found in SERPING1, IL15, C1R, MAP3K5 and ANXA4 have sequence patterns highly similar to ERE consensus, indicating that these genes are likely candidates for direct regulation by estrogen receptor.

It is reasonable, however, to postulate that in the endometrium transcription of implantation-related genes showing steroid-responsive elements may occur either directly by estrogen or progesterone, or indirectly via paracrine mechanisms (i.e. by locally acting factors that are steroid responsive). SERPING1 was also one of the genes that responded to estrogen stimulation in delayed implanting mice (Reese et al., 2001Go). Interestingly, SERPING1 and C1R genes are linked functionally: SERPING1 encodes a plasma protein that regulates the first component of complement (C1) by inhibition of the proteolytic activity of its subcomponents (Davis et al., 1986Go) including C1r, a product of the C1R gene.

Prediction of the genes that can be potentially regulated by progesterone receptor is complicated by a common sequence pattern that is recognized by androgen, progesterone, and glucocorticoid receptors. Our search for asymmetric high-affinity–specificity sequence pattern identified previously for PR (Nelson et al., 1999Go) showed seven candidate genes that may be regulated by PR (Table IV). We did not find PRE-like sequence in the 5'-flanking region of the SPP1 gene encoding osteopontin, which is known as progesterone responsive in both mouse and human (Johnson et al., 2003Go). Similarly, Craig and Denhardt (1991)Go could not detect PRE in the murine SPP1 gene promoter. There are two possible reasons for this: first, PR can bind to non-consensus sequences (Tsuchiya et al., 2003Go); second, PR may modulate transcription through interaction with other transcription factors (e.g. Sp1) (Owen et al., 1998Go), rather than by direct binding to gene promoter. Obviously, experimental evidence is required to verify if any of the ERE and PRE predicted in this set of genes are functional transcriptional elements.

One of the shortcomings of studies of this kind is the possibility that products expressed in a given cell type (i.e. epithelium, glands or stroma) may undergo changes that are of insufficient amplitude to be detected in the entire tissue (Carson et al., 2002Go). Although all biopsies studied were obtained using a similar technique and H–E confirmed a full endometrial depth, we are now using laser capture microdissection techniques in order to examine gene expression profiles of the compartmentalized endometrium.

Our study provides new information on genes of potentially critical function for the acquisition of endometrial receptivity. Such information can establish the background for future assessment of gene expression and protein profiling of the human endometrium that may lead to the diagnosis of specific aetiologies in cases of reproductive failure. Furthermore, as described in our previous report (Mirkin et al., 2004bGo), microarray data can provide the basis for analysng the controversial impact of controlled ovarian stimulation protocols (COS) on implantation. In this regard, we have identified gene expression changes when comparing natural and COS cycles for IVF, with profound differences observed when comparing protocols involving the use of GnRH agonists versus antagonists.

In conclusion, using a strict bioinformatics approach of microarray data, we demonstrated significant changes in gene expression in the transition of the early to the mid-luteal phase endometrium. We identified a number of genes that may have functional significance for the opening and maintenance of the window of receptivity. We confirmed the expression of several other genes that also appear to be important for implantation as reported by others (Carson et al., 2002Go; Kao et al., 2002Go; Borthwick et al., 2003Go; Riesewijk et al., 2003Go). Moreover, we identified a number of putative ERE and PRE in a subset of differentially expressed genes. Further studies assessing the role of these genes and their products could lead to new approaches for improving the management of early recurrent pregnancy loss, optimizing embryo implantation in IVF therapy, and could also be the basis for the development of new contraceptive strategies.


    Acknowledgements
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
We are thankful for the support of Organon Inc., USA.


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Acknowledgements
 References
 
Bajic VB, Tan SL, Chong A, Tang S, Strom A, Gustafsson JA, Lin CY and Liu ET (2003) Dragon ERE Finder version 2: A tool for accurate detection and analysis of estrogen response elements in vertebrate genomes. Nucleic Acids Res 31, 3605–3607.[Abstract/Free Full Text]

Bergh PA and Navot D (1992) The impact of embryonic development and endometrial maturity on the timing of implantation. Fertil Steril 58, 537–542.[ISI][Medline]

Borthwick JM, Charnock-Jones DS, Tom BD, Hull ML, Teirney R, Phillips SC and Smith SK (2003) Determination of the transcript profile of human endometrium. Mol Hum Reprod 9, 19–33.[Abstract/Free Full Text]

Brown SE, Mandelin E, Oehninger S, Toner JP, Seppala M and Jones HW Jr (2000) Endometrial glycodelin-A expression in the luteal phase of stimulated ovarian cycles. Fertil Steril 74, 130–133.[CrossRef][ISI][Medline]

Byrjalsen I, Larsen PM, Fey SJ and Christiansen C (1995) Human endometrial proteins with cyclic changes in the expression during the normal menstrual cycle: characterization by protein sequence analysis. Hum Reprod 10, 2760–2766.[Abstract]

Carson DD, Bagchi I, Dey SK, Enders AC, Fazleabas AT, Lessey BA and Yoshinaga K (2000) Embryo implantation. Dev Biol 223, 217–237.[CrossRef][ISI][Medline]

Carson DD, Lagow E, Thathiah A, Al-Shami R, Farach-Carson MC, Vernon M, Yuan L, Fritz MA and Lessey B (2002) Changes in gene expression during the early to mid-luteal (receptive phase) transition in human endometrium detected by high-density microarray screening. Mol Hum Reprod 8, 871–879.[Abstract/Free Full Text]

Chong A, Zhang G and Bajic VB (2003) FIE2: A program for the extraction of genomic DNA sequences around the start and translation initiation site of human genes. Nucleic Acids Res 31, 3546–3553.[Abstract/Free Full Text]

Christian M, Zhang X, Schneider-Merck T, Unterman TG, Gellersen B, White JO and Brosens JJ (2002) Cyclic AMP-induced forkhead transcription factor, FKHR, cooperates with CCAAT/enhancer-binding protein beta in differentiating human endometrial stromal cells. J Biol Chem 277, 20825–20832.[Abstract/Free Full Text]

Craig AM and Denhardt DT (1991) The murine gene encoding secreted phosphoprotein 1 (osteopontin): promoter structure, activity, and induction in vivo by estrogen and progesterone. Gene 100, 163–171.[CrossRef][ISI][Medline]

Davis AE 3rd, Whitehead A, Harrison RA, Dauphinais A, Bruns GA, Cicardi M and Rosen FS (1986) Human inhibitor of the first component of complement, C1: characterization of cDNA clones and localization of the gene to chromosome 11. Proc Natl Acad Sci USA 83, 3161–3165.[Abstract/Free Full Text]

Davis JN, Rock CO, Cheng M, Watson JB, Ashmun RA, Kirk H, Kay RJ and Roussel MF (1997) Complementation of growth factor receptor-dependent mitogenic signaling by a truncated type I phosphatidylinositol 4-phosphate 5-kinase. Mol Cell Biol 17, 7398–7406.[Abstract]

Develioglu OH, Hsiu JG, Nikas G, Toner JP, Oehninger S and Jones HW Jr (1999) Endometrial estrogen and progesterone receptor and pinopode expression in stimulated cycles of oocyte donors. Fertil Steril 71, 1040–1047.[CrossRef][ISI][Medline]

Devroey P, Bourgain C, Macklon NS and Fauser BC (2004) Reproductive biology and IVF: ovarian stimulation and endometrial receptivity. Trends Endocrinol Metab 15, 84–90.[CrossRef][ISI][Medline]

Draghici S and Kuklin A (2003) Data Analysis Tools for DNA Microarrays, 1st edn. Chapman & Hall, New York.

Eisen M, Spellman P, Brown P and Botstein D (1998) Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA 95, 14863–14868.[Abstract/Free Full Text]

Forman BM and Samuels HH (1990) Interactions among a subfamily of nuclear hormone receptors: the regulatory zipper model. Mol Endocrinol 4, 1293–1301.[Abstract]

Galitski T, Saldanha A, Styles C, Lander E and Fink G (1999) Science 285, 251–254.[Abstract/Free Full Text]

Giudice LC (2003) Elucidating endometrial function in the post-genomic era. Hum Reprod Update 9, 223–235.[Abstract/Free Full Text]

Hastie T, Tibshirani R and Friedman JH (2001) The Elements of Statistical Learning, 1st edn. Springer-Verlag, New York.

Horcajadas JA, Riesewijk A, Martin J, Cervero A, Mosselman S, Pellicer A and Simon C (2004) Global gene expression profiling of human endometrial receptivity. J Reprod Immunol 63, 41–49.[CrossRef][ISI][Medline]

Johnson GA, Burghardt RC, Spencer TE, Newton GR, Ott TL and Bazer FW (1999) Ovine osteopontin: II. Osteopontin and alpha(v)beta(3) integrin expression in the uterus and conceptus during the periimplantation period. Biol Reprod 61, 892–899.[Abstract/Free Full Text]

Johnson GA, Spencer TE, Burghardt RC, Taylor KM, Gray CA and Bazer FW (2000) Progesterone modulation of osteopontin gene expression in the ovine uterus. Biol Reprod 62, 1315–1321.[Abstract/Free Full Text]

Johnson GA, Burghardt RC, Bazer FW and Spencer TE (2003) Osteopontin: roles in implantation and placentation. Biol Reprod 69, 1458–1471.[Abstract/Free Full Text]

Kao LC, Tulac S, Lobo S, Imani B, Yang JP, Germeyer A, Osteen K, Taylor RN, Lessey BA and Giudice LC (2002) Global gene profiling in human endometrium during the window of implantation. Endocrinology 143, 2119–2138.[Abstract/Free Full Text]

Kao LC, Germeyer A, Tulac S, Lobo S, Yang JP, Taylor RN, Osteen K, Lessey BA and Giudice LC (2003) Expression profiling of endometrium from women with endometriosis reveals candidate genes for disease-based implantation failure and infertility. Endocrinology 144, 2870–2881.[Abstract/Free Full Text]

Klein-Hitpass L, Schorpp M, Wagner U and Ryffel GU (1986) An estrogen-responsive element derived from the 5' flanking region of the Xenopus vitellogenin A2 gene functions in transfected human cells. Cell 46, 1053–1061.[CrossRef][ISI][Medline]

Klinge CM (2001) Estrogen receptor interaction with estrogen response elements. Nucleic Acids Res 29, 2905–2919.[Abstract/Free Full Text]

Lessey BA (2004) Endometrial responsiveness to steroid hormones: a moving target. J Soc Gynecol Invest 11, 61–62.[CrossRef][ISI][Medline]

Lim H, Song H, Paria BC, Reese J, Das SK and Dey SK (2002) Molecules in blastocyst implantation: uterine and embryonic perspectives. Vitam Horm 64, 43–76.[CrossRef][ISI][Medline]

Lindhard A, Bentin-Ley U, Ravn V, Islin H, Hviid T, Rex S, Bangsboll S and Sorensen S (2002) Biochemical evaluation of endometrial function at the time of implantation. Fertil Steril 78, 221–233.[CrossRef][ISI][Medline]

Mader S, Kumar V, de Verneuil H and Chambon P (1989) Three amino acids of the oestrogen receptor are essential to its ability to distinguish an oestrogen from a glucocorticoid-responsive element. Nature 338, 271–274.[CrossRef][ISI][Medline]

Mirkin S, Mahony MC and Archer DF (2004a) Effect of tibolone and its metabolites on vascular endothelial growth factor isoforms 121 and 165 and thrombospondin-1 mRNA in Ishikawa cells. Menopause 11, 82–88.[CrossRef][ISI][Medline]

Mirkin S, Nikas G, Hsiu J, Díaz JI and Oehninger S (2004b) Gene expression profiles and structural/functional features of the peri-implantation endometrium in natural and gonadotropin-stimulated cycles. J Clin Endocrinol Metab 89, 5742–5752.[Abstract/Free Full Text]

Navot D, Scott RT, Droesch K, Veeck LL, Liu HC and Rosenwaks Z (1991) The window of embryo transfer and the efficiency of human conception in vitro. Fertil Steril 55, 114–118.[ISI][Medline]

Nelson CC, Hendy SC, Shukin RJ, Cheng H, Bruchovsky N, Koop BF and Rennie PS (1999) Determinants of DNA sequence specificity of the androgen, progesterone, and glucocorticoid receptors: evidence for differential steroid receptor response elements. Mol Endocrinol 13, 2090–2107.[Abstract/Free Full Text]

Noyes RW, Hertig AT and Rock J (1950) Dating the endometrial biopsy. Fertil Steril 1, 3–25.[ISI][Medline]

Omigbodun A, Ziolkiewicz P, Tessler C, Hoyer JR and Coutifaris C (1997) Progesterone regulates osteopontin expression in human trophoblasts: a model of paracrine control in the placenta? Endocrinology 138, 4308–4315.[Abstract/Free Full Text]

Owen GI, Richer JK, Tung L, Takimoto G and Horwitz KB (1998) Progesterone regulates transcription of the p21(WAF1) cyclin-dependent kinase inhibitor gene through Sp1 and CBP/p300. J Biol Chem 273, 10696–10701.[Abstract/Free Full Text]

Pavlova A, Boutin E, Cunha G and Sassoon D (1994) Msx1 (Hox-7.1) in the adult mouse uterus: cellular interactions underlying regulation of expression. Development 120, 335–345.[Abstract/Free Full Text]

Ponnampalam AP, Weston GC, Trajstman AC, Susil B and Rogers PA (2004) Molecular classification of human endometrial cycle stages by transcriptional profiling. Mol Hum Reprod 10, 879–893.[Abstract/Free Full Text]

Psychoyos A (1994) The implantation window; basic and clinical aspects. In Mori et al. (eds) Perspectives on Assisted Reproduction. Ares-Serono Symposium 4, Roma, 1994, pp 57–63.

Reese J, Das SK, Paria BC, Lim H, Song H, Matsumoto H, Knudtson KL, DuBois RN and Dey SK (2001) Global gene expression analysis to identify molecular markers of uterine receptivity and embryo implantation. J Biol Chem 276, 44137–44145.[Abstract/Free Full Text]

Richards JS, Sharma SC, Falender AE and Lo YH (2002) Expression of FKHR, FKHRL1, and AFX genes in the rodent ovary: evidence for regulation by IGF-I, estrogen, and the gonadotropins. Mol Endocrinol 16, 580–599.[Abstract/Free Full Text]

Riesewijk A, Martin J, van Os R, Horcajadas JA, Polman J, Pellicer A, Mosselman S and Simon C (2003) Gene expression profiling of human endometrial receptivity on days LH+2 versus LH+7 by microarray technology. Mol Hum Reprod 9, 253–264.[Abstract/Free Full Text]

Roberts C, Nelson B, Marton M et al. (2000) Science 287, 873–880.[Abstract/Free Full Text]

Tsuchiya S, Tanaka S, Sugimoto Y, Katsuyama M, Ikegami R and Ichikawa A (2003) Identification and characterization of a novel progesterone receptor-binding element in the mouse prostaglandin E receptor subtype EP2 gene. Genes Cells 8, 747–758.[Abstract/Free Full Text]

Tusher VG, Tibshirani R and Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98, 5116–5121.[Abstract/Free Full Text]

Van Gelder RN, von Zastrow ME, Yool A, Dement WC, Barchas JD and Eberwine JH (1990) Amplified RNA synthesized from limited quantities of heterogeneous cDNA. Proc Natl Acad Sci USA 87, 1663–1667.[Abstract/Free Full Text]

Wilcox AJ, Baird DD and Weinberg CR (1999) Time of implantation of the conceptus and loss of pregnancy. New Engl J Med 340, 1796–1799.[Abstract/Free Full Text]

Submitted on December 9, 2004; resubmitted on March 7, 2005; accepted on April 5, 2005.