Bioinformatic detection of E47, E2F1 and SREBP1 transcription factors as potential regulators of genes associated to acquisition of endometrial receptivity

The human endometrium is a complex tissue whose cyclic regulation is mainly driven by the changing pattern of the ovarian steroidal hormones estradiol (E₂) and progesterone (P₄) [1]. The main function of the endometrium is to provide receptive substrate at the appropriate time for blastocyst implantation. Although it is non-adhesive to embryos throughout most of the menstrual cycle [2] the action of P₄ on an E₂-primed endometrium induces a certain gene expression profile that is favorable for blastocyst adhesion during a restricted period of time known as the 'window of implantation' [3, 4]. In women, this maternally directed receptive phase appears to be of approximately 5 days' duration, from day 20 to day 24 of a 28-days menstrual cycle [5]. The molecular basis of the window of implantation in human endometrium is beginning to be unrevealed and a number of biochemical markers for uterine receptivity have been proposed [3, 6].

Microarrays analysis, an assay that is used to measure the level of mRNA expression of thousands of genes in a group of cells [7], enables discovery of genes or pathways likely to be involved in a biological process. This approach has been used to broadly characterize the molecular bases of endometrial function in women, by determining the gene expression profiles corresponding to each endometrial phase during the menstrual cycle [8‐10]. In addition, it has been used to specifically investigate the acquisition of endometrial receptivity to embryo implantation during spontaneous cycles [11‐15]. Since changes in the endometrium toward acquisition of receptivity are mainly driven by progesterone (P₄) [16, 17], two strategies have been used for gene discovery during spontaneous menstrual cycles. These are based on the comparison of the endometrial transcriptome under peak P₄ circulating levels (days 19-23, window of implantation) compared to the endometrial gene expression profiles obtained under absent (days 8-11, proliferative phase) [11, 12] or low (days 15-17, early secretory phase) [13‐15, 18, 19] serum P₄.

Although DNA microarrays are a powerful tool for gene discovery, there are several substantial sources of noise in microarray data. Intra- and inter-microarray variations limit the statistical power to discriminate the differentially expressed genes. While validation of microarray data is required to overcome this issue, most reports of endometrial gene expression analysis included validation of only a small number of differentially expressed genes (usually less than 10) by an independent mRNA quantification method (Northern blot, semi-quantitative or quantitative RT-PCR) [20]. Integration and cross-validation of data sets about endometrial gene expression profiles produced by different groups could increase confidence in gene expression results for many more genes than is tractable with classical validation [21, 22] and should provide the up- and down-regulated genes that together orchestrate the acquisition of the receptive phenotype of the endometrium. Such exploration and integration could help researchers to obtain a comprehensive view of existing data and better prioritize experimental efforts.

Transcriptional regulatory mechanisms are crucial for temporal and spatial gene expression. These mechanisms are mediated by a set of transcription factors (TFs), proteins which have the ability to bind to a specific region on the gene (known as motifs or transcription factor binding sites (TFBS)), to regulate transcription. It is thought that co-expression of genes frequently arises from transcriptional co-regulation. As co-regulated genes share some similarities in their regulatory mechanism, possibly at transcriptional level, their promoter regions may contain common motifs that are binding sites for transcription regulators [23]. Given a cluster of endometrial regulated genes with similar expression profiles, the characterization of their regulatory regions is a fundamental step toward understanding the largely unexplored networks of gene regulation in this complex tissue responsible for their coordinated behavior. Computation biology of gene regulation offers several bioinformatic tools developed for the prediction of TFBS within a specific regulatory DNA sequence [24]. Given a set of co-regulated transcripts, in silico predictions of TFBS in their regulatory regions offers a unique opportunity to identify novel components, leading to the formulation of transcriptional regulatory networks hypotheses that can be further tested in the wet laboratory.

Scientific knowledge of how endometrial receptivity is regulated is fundamental for the understanding of the mechanisms that govern embryonic implantation. The availability of public datasets related to global endometrial gene regulation during the acquisition of the receptive phenotype, provides a tool for the analysis of regulation of gene expression using bioinformatics tools. Using DNA microarrays analysis, several approaches have been used for determining the genes of uterine receptivity assessing the endometrium in different physiological [9, 11‐15, 18, 43], pathological [44‐48] or intervened conditions [49, 50]. We here analyzed seven reports of endometrial gene expression profiling during spontaneous cycles: Carson et al. [13], Kao et al. [11], Borthwick et al. [12], Riesewijk et al. [14], Mirkin et al. [15], Talbi et al. [9] and Haouzi et al. [18]. Since the number of endometrial samples analyzed in each of these studies was limited, the question arises as to whether the groups investigated were representative of the population. This is a major concern for any statistical analysis. Therefore we considered all studies together in a research synthesis to provide a larger sample size thus consolidating the selection of actual regulated transcripts in the endometrium. A first step was to associate probes and available annotations in the reports that belong to the same UniGene cluster (i.e. with same UniGene ID), and then proceed to further comparisons to identify common transcripts that are similarly regulated during the window of implantation. Previous partial analyses [15, 43, 51‐53] found very few transcripts to be consistently regulated. In our study we found 61 transcripts regulated in the same direction in the endometrium during the window of implantation; 40 were up-regulated in at least 4 of 7 studies and 21 were down-regulated in at least 3 of 6 reports analyzed.

The relatively small number of consistently regulated transcripts identified could be explained by the differences in the study design, number of samples included and the methodology used for data analysis. However, other factors should be considered when interpreting gene expression analyses related to endometrial receptivity. Importantly, the reports included here, all used RNA extracted from whole endometrial biopsies, tissue that comprises a number of different cell types, including epithelial (luminal and glandular), stromal fibroblasts, endothelial cells, vascular smooth muscle cells and lymphoid cells. Hence the endometrial changes induced by E₂ and P₄ result from the differential response of each cell type to the same hormones. A clear example is the down regulation of the PR during the secretory phase in endometrial epithelial cells but not in the stromal compartment [54]. Microdissection of cell subpopulations (for example, with laser capture [55]) may disclose the actual gene expression profiles of each cell subpopulations within the tissue context. In addition, any biopsy sample may not represent the complete endometrium since microenvironments occur within this tissue. Nevertheless gene expression profiling of endometrial biopsies during the window of implantation is one of the most promising strategies for gene discovery related to uterine receptivity.

The intersection of gene lists performed in the present study showed that most consistently increased transcripts during the window of implantation were C4BPA, SPP1, APOD, CD55, CFD, CLDN4, DKK1, ID4, IL15 and MAP3K5 whereas OLFM1, CCNB1, CRABP2, EDN3, FGFR1, MSX1 and MSX2 were the most consistently decreased. However, correlation of transcript abundance change with changes in the corresponding protein, followed by functional testing of the biological effect of that protein, is necessary to confirm the biological significance of the microarray changes.

The functional annotations of up-regulated genes within the CERTL showed a significant association to the immune response and complement activation. Most of these genes belong to the innate immune system, which is the immunological first line of defense that provides an immediate response through its ability to distinguish between 'infectious non-self' and 'non-infectious self' [56]. Therefore, innate immunity regulation in the endometrium is of fundamental significance for establishing a microenvironment that will provide adequate tolerance to the implanting embryo [57]. Regarding complement system regulatory proteins, their possible roles and expression levels in the endometrium throughout the normal menstrual cycle have been reported [58‐62]. Most of these studies show an increase of complement-regulatory molecules during the secretory phase in human endometrium [58, 61, 62] in line with the increased mRNA levels of the complement system molecules C4b-binding protein (C4BP) and adipsin (complement component factor D, CFD) from the CERTL. It is postulated that the complement system might be conferring immunity to the uterine cavity, defending it against bacterial infection. In this sense, C4BP may provide a protective role to the embryo where an increased expression of an inhibitor of complement system activation could reduce the chance of a misdirected complement attack to the embryo (which is considered as a semiallograft). Indeed, C4BPA transcript levels are abnormally decreased in the endometrium during the receptive phase in women with endometriosis [44, 63], implantation failure [46] and unexplained recurrent abortion [64], suggesting it may have a role in embryo implantation. By contrast, adipsin may have a non-complement function in the female reproductive tract as suggested for other complement-molecules [60]. Adipsin is necessary for the production of oviduct-derived embryotrophic factor-3 (ETF-3) [65, 66] which stimulates embryo development [67, 68]. Thus up-regulation of adipsin in human endometrium may assist the embryo during the implantation process as shown for other chemokines in the endometrium [69].

Several down-regulated genes within CERTL are associated with cell cycle regulation, including cyclin B1 (CCNB1) the most consistently down-regulated gene. CCNB1 binds to p34 (cdc2) to form the mitosis-promoting factor during G2 phase [70, 71]. In human secretory phase endometrium, CCNB1 is decreased compared to the proliferative phase [72, 73] supporting the microarray data used to construct the CERTL. Moreover, in endometrial cell cultures, P₄ decreases the expression of CCNB1, inhibits cell proliferation and induces apoptosis, suggesting that cyclin B1 may play an important role in proliferation and differentiation of the endometrial tissue under steroidal regulation.

Cellular retinol binding protein-2 (CRABP2) is a cytosolic protein that binds retinoic acid (RA) with high affinity [74]. The CRABP2 transcript has been reported to decrease from the proliferative to the secretory phase in human endometrium [75], which is in line with the microarrays reports used for constructing our CERTL. The physiological effects of RA are mediated by members of two families of nuclear receptors [76, 77] and they all have been detected by immunohistochemistry in human endometrium throughout the phases of the menstrual cycle [78] in epithelial and stromal cells. The fact that CRABP2 decreases in human endometrium at the time of embryo implantation might suggest that RA signaling is required to be silenced, since it shuttles RA to the RA receptors in the cell nucleus [74, 78]. In the mouse uterus, CRABP2 decreases around the time of embryo implantation [79] whereas P₄ induces the expression of cyp26a1, the enzyme responsible for RA catabolism in mouse uterine epithelial cells [80, 81]. Knock down of cyp26a1 in mouse uterus decreases embryo implantation rate [82]. In addition, in human secretory endometrium, cyp26a1 mRNA level is ~20 times higher than in the proliferative phase [83]. Since the action of RA is essential for endometrial stromal cell decidualization [79] silencing of RA signaling during the window of implantation might prevent precocious decidualization of stromal cells that could compromise endometrial receptivity.

The cytokine endothelin-3 (EDN3) and fibroblast growth factor receptor-1 (FGFR1) were among the transcripts consistently down-regulated in the endometrium during the window of implantation. There is abundant evidence showing that both endometrial receptivity and blastocyst implantation are regulated by cytokines and growth factors [84]. Immunoreactive pro-endothelin-3 has been described in human endometrium in luminal and glandular epithelia; however cycle-dependent regulation of this molecule is not clear [85]. Its action in the human endometrium is suggested to be in paracrine vasoactive control of the uterine vascular bed [86]. However this cytokine has many other functions such as proliferation and development of several cell types [87‐90]. In the mouse oviduct, EDN3 signaling has been associated with the regulation of transcripts related to TGFβ, IL-10, and C/EBP [91]. Its functional role in the human endometrium and the effects of its down-regulation during the window of implantation has yet to be determined. FGFR1 and its ligand FGF-2 have also been described in human endometrium [92‐95]. Immunoreactive FGFR1 and its transcript are significantly higher in proliferative that in secretory human endometrium [93, 94] supporting the down-regulation of this transcript included in the CERTL. However, not all studies have reported such endometrial regulation [95]. FGF-2 promotes endometrial stromal proliferation [94, 96] and ovarian steroid hormones modulate its synthesis and function in endometrial cells [96, 97]. The functional relevance of FGFR1 down-regulation in endometrial receptivity remains to be elucidated.

With regard to the TFs present in the CERTL, we found the inhibitor of DNA binding 4 (ID4) up-regulated and MSX-1 and -2 down-regulated. In animal models, uterine MSX-1 and -2 are down-regulated by P₄ [98] or during embryo implantation [99‐101]. Constant expression of Msx1 in the infertile Lif-/- mice uterus further supports a role for MSX-1down-regulation in endometrial receptivity [100]. ID4 TF is a member of a family of inhibitor of DNA binding proteins (Id) that has been associated with cell proliferation and differentiation [102‐105]. Its regulatory effect in human endometrium is unknown. Many other TFs associated with endometrial regulation [106‐120] have provided insights into the molecular basis of gene regulation for endometrial function in response to sex steroid hormones. We reasoned that the cluster of regulated genes derived from microarray experiments related to endometrial receptivity (i.e. CERTL) would allow a different strategy for TF discovery, namely comparative promoter analysis. This is based on the hypothesis that genes showing a common regulatory behavior may also share common regulatory mechanisms such as TFBSs in their respective promoter regions. Interestingly, we found that E47, E2F1 and SREBP1 are common TFBSs for up- and down-regulated transcripts from CERTL so it is likely that they orchestrate the changes in transcript profile for endometrial receptivity. None of these three TFs have been described in normal human endometrium in the context of their regulation during the menstrual cycle, in response to steroidal hormones or a regulatory role on uterine function. However, there is no guarantee that the revealed TFBS are indeed functional in the context of regulatory regions, hence biological verification is required.

The E2F1 TF belongs to the E2F family [121] and displays properties of both an oncogene (induction of proliferation) and tumor suppressor (induction of apoptosis) [122, 123]. E47 is a TF that belongs to the class I bHLH proteins, also known as E proteins [124] which form homo- or hetero-dimers and bind to specific DNA sequences [125]. Sterol regulatory element-binding protein 1 (SREBP1) is a membrane-bound TFs that belongs to a family of basic helix-loop-helix-leucine zipper (bHLHLZ) TFs [126]. Upon activation, SREBP1 translocates into the nucleus where it binds to sterol regulatory sites located in the promoter regions of genes involved in cholesterol homeostasis and transport [127, 128] such as the steroidogenic acute regulatory protein (StAR), a key regulator of steroidogenesis [129]. Function of bHLH TFs such as E47 can be blocked by Inhibitor of DNA binding (Id) TFs [130, 131]. In addition, SREBP1 as a member of bHLHLZ family, may also be subjected to regulation by Id proteins [132]. In the CERTL ID4 transcript was up-regulated in the receptive endometrium: as a consequence E47 and SREBP-1 TFs may be less available for binding to DNA in target sequences and direct co-regulated transcripts. Interestingly, the TF E2F1 is involved in the transcriptional control of id4 gene expression [133], supporting our bioinformatics findings of overrepresented TFBSs.

It is well known that P₄ is essential for the establishment and maintenance of pregnancy in the women and in this sense the study of its actions in the uterus has been focused on changes in gene expression [134, 135]. Responses to P₄ in reproductive tissues occur by the activation of classical nuclear P₄ receptors (PRA and B), which upon binding with their ligand, function as TFs regulating gene expression [136]. In addition, many transcriptional actions of P₄ require interactions with corepressors and coactivators [137‐139]. However, P₄ may also act in the uterus through at least two families of nonclassical membrane progestin receptors [140, 141]. Hence the genomic and non-genomic pathways may interact and integrate to ultimately affect endometrial gene expression. Interestingly, two of the endometrial transcripts more consistently up-regulated during the mid-secretory phase, APOD and SPP1, do not display progesterone response elements in their cis-regulatory sequences [12, 15] suggesting that P₄ induction is not directly mediated by the ligand-bound PR. Interestingly both APOD and SPP1 genes display TFBS for E2F1 in their upstream regulatory sequences. In breast cancer cells, P₄ up-regulates the expression of E2F1 and hence indirectly affects transcription of classic E2F1 target genes [115]. Such regulation of E2F1 induced by progestins has been shown to be multimodal since ligand-bound PR can regulate its transcription directly but also indirectly through other molecules to achieve further progestin-mediated regulation of E2F1 expression [142]. Whether E2F1 along with E47 and SRBP1 are also mediating the P₄ transcriptional regulation in the endometrium for acquisition of receptivity has yet to be determined.

Identification of the CERTL and the possible regulatory TFs in the present research synthesis should not be viewed as an end in itself. Their real value increases only as these results move through to biological validation, ranging from the numerical verification of expression levels with alternative techniques, to ascertaining the actual regulatory role of the TFs in the endometrial transcriptional networks. Finally, for several transcripts contained in the CERTL, biological knowledge is completely lacking in relation to endometrial physiology, so extensive research is required to better understand the mechanisms underlying endometrial receptivity.

In conclusion, a CERTL comprised of 61 transcripts consistently regulated in human endometrium during the receptive period for embryo implantation has been identified in this study. These transcripts are mainly involved in immune response, complement activation and cell cycle regulation; suggesting that these biological process are associated with the acquisition of the receptive phenotype. Finally, TFBS for E47, SREBP1 and E2F1 were over-represented in the regulatory region of genes from CERTL, suggesting that they may be mediating the effects of the ovarian steroidal hormones in the endometrial transcriptional regulation. Biological validation of such bioinformatic predictions will shed light on the transcriptional networks associated to uterine receptivity for embryo implantation. Moreover, this knowledge can potentially be applied to improve fertility in infertile patients.