Human Endometrial Transcriptomics: Implications for Embryonic Implantation

Endometrial gene expression profiles have been obtained for the entire menstrual cycle. These studies have led to the development of the Endometrial Receptivity Array, which can assist patients in timing embryo transfer.

Abstract

Human endometrium has been extensively investigated in the search for markers capable of predicting its receptive status. The completion of the Human Genome Project has triggered a rapid development of new fields in molecular biology, the “transcriptomics” being a major turning point in the knowledge acquisition of endometrial receptivity. Based on this, a customized Endometrial Receptivity Array (ERA) has been developed, which is capable of identifying the genomic signature of receptivity. This diagnostic tool showed that the window of implantation (WOI) is displaced in one out of four patients with implantation failure, allowing the identification of their personalized WOI. This strategy allows performing a personalized embryo transfer (pET) on the day in which the endometrium is receptive. The combination of a systems biology approach and next-generation sequencing will overcome the limitations of microarrays, and will, in the future, allow elucidation of the mechanisms involved in embryo implantation.

The endometrium, which lines the inside of the uterine cavity, undergoes cyclic changes that are regulated by ovarian steroids. It can be subdivided into the basal layer that is responsible for its regenerative capacity, and the functional layer that undergoes proliferation, secretion, and tissue degeneration every month from menarche to menopause. Its aim is to prepare the optimal moment for embryonic implantation known as the window of implantation (WOI).

The endometrial cycle comprises the menstrual, proliferative, and secretory phases. The proliferative phase, which corresponds to the ovarian follicular phase and increased production of estrogens, lasts until ovulation occurs. During this stage, the increasing estrogen levels cause the proliferation of the stromal cells as well as the elongation of the spiral arteries. After ovulation, with the appearance of progesterone secreted by the granulosa-luteal cells, the secretory phase begins. If implantation does not occur, the secretory phase ends, and the corpus luteum degenerates. Menstruation occurs owing to the drop of estrogen and progesterone, which resets the endometrium until pregnancy occurs. The secretory phase can be divided into early-, mid-, and late-secretory. The most important phase is the mid-secretory, because at this point the endometrium acquires a receptive phenotype. A peak of progesterone characterizes this period, known as the WOI. The WOI lasts between 12 h to 2 d and may vary in length from patient to patient (Fig. 1). During this phase, implantation will occur if a viable blastocyst is present and finds a receptive endometrium, and a synchronized dialogue is established between them.

Figure 1.

Displacement of the window of implantation (WOI). It has been assumed that the WOI is constant in time in all women (1). However, the genomic signature of the endometrium shows the existence of a displacement of the WOI in up to 25% of patients that can be delayed (2), advanced (3), or shorter than expected (4). P + x refers to the days after progesterone administration. (Based on Galliano et al. 2014.)

Endometrial receptivity is a widely studied process; its understanding is providing better and more comprehensive knowledge of the reproductive process. Histological criteria have been used since the 1950s to date the endometrium (Noyes et al. 1950, 1975). However, morphological criteria have major limitations for predicting endometrial receptivity, as shown in randomized studies (Coutifaris et al. 2004; Murray et al. 2004). The diagnosis obtained may vary depending on the subjective interpretation from each observer and histological variations at the moment the endometrial biopsy was obtained. In addition, these morphological markers do not properly recognize the phenotype of the receptive endometrium.

Endometrial receptivity has also been unsuccessfully investigated at the single molecular and biochemical levels. However, the identification of a search for the transcriptomic signature was a major turning point in the understanding of this function.

The completion of the Human Genome Project (Venter et al. 2001) triggered a rapid development of new fields in molecular biology that are described as the “omics revolution.” Omics refers to the application of high-throughput techniques and massive data analysis, allowing molecular profiling and changes between groups or individuals to be investigated. Among the “omics sciences”, there is the transcriptomics, in which analyses of patterns of gene expression in a specific tissue, under specific conditions, helps to diagnose physiological functions and pathological conditions.

Microarray-based gene expression technology, which allows simultaneous monitoring of the expression of thousands of genes, has been the most widely used platform for transcriptomic analysis (Altmäe et al. 2013). Data obtained are further curated using statistical analysis and exploratory methods. The most common exploratory method is clustering or principal component analysis (PCA). Other visual methods, such as heatmap representations, display the differential expression patterns among different conditions (Díaz-Gimeno et al. 2014).

Over the last decade, the transcriptomics of the human endometrium has been widely investigated (Ruíz-Alonso et al. 2012). Several areas have been covered, from the transcriptomic expression throughout the menstrual cycle to the changes identified under different treatments or gynecological conditions. However, the main interest has been the identification of the specific transcriptomic signature that can diagnose the receptive function and improve the effectiveness of reproductive treatments.

TRANSCRIPTOMICS OF THE ENDOMETRIUM

Endometrial transcriptomic studies have covered the identification of the physiological endometrial profile throughout the menstrual cycle with special attention to the WOI (Carson et al. 2002; Kao et al. 2002; Borthwick et al. 2003; Riesewijk et al. 2003; Horcajadas et al. 2004a,b; Ponnampalam et al. 2004; Mirkin et al. 2005; Puynadeera et al. 2005; Yanaihara et al. 2005; Critchley et al. 2006; Talbi et al. 2006; Haouzi et al. 2009a; Kuokkanen et al. 2010; Tseng et al. 2010; Van Verenbergh et al. 2010; Díaz-Gimeno et al. 2011; Revel et al. 2011). Endometrial transcriptomic profiles have also been investigated under different controlled ovarian stimulation protocols (Simon et al. 2005; Horcajadas et al. 2008), or in the context of unexplained pathologies such as recurrent implantation failure (Tapia et al. 2008; Koler et al. 2009; Altmäe et al. 2010), or other endometrial conditions (Table 1) (Habermann et al. 2011; Matsuzaki 2011).

Table 1.

Original studies on endometrial transcriptomics in assisted reproductive medicine

References	Endometrial biopsy timing (in days)	Study
Carson et al. 2002	LH + (2–4) versus LH + (7–9)	ES versus MS
Kao et al. 2002	CD 8–10 versus LH + (8–10)	LP versus MS
Borthwick et al. 2003	CD 9–11 versus LH + (6–8)	LP versus MS
Riesewijck et al. 2003	LH + 2 versus LH + 7	ES versus MS
Mirkin et al. 2004	LH + 8 versus hCG + 9	Ag versus Atg versus NC
Ponnampalam et al. 2004	Complete cycle, dating by Noyes	EP versus MP versus LP versus ES versus MS versus LS versus M
Horcajadas et al. 2005	LH(+ 2; + 7) versus hCG + 7	NC versus COH
Mirkin et al. 2005	LH + 3 versus LH + 8	ES versus MS
Punyadeera et al. 2005	CD 2–5 versus CD 11–14	M versus LP
Simon et al. 2005	LH (+ 2; + 7) versus hCG (+ 2; + 7)	Ag versus Atg versus NC
Yanahaira et al. 2005	CD 9–11	Epithelial versus stromal cells in proliferative phase
Critchley et al. 2006	Dating by Noyes	MS versus LS
Talbi et al. 2006	Complete cycle, dating by Noyes	EP versus MP versus LP versus ES versus MS versus LS
Horcajadas et al. 2008	LH + (1–9) versus hCG+ (1–9)	NC versus COS
Liu et al. 2008	LH + 7 versus hCG + 7	NC versus COS
Macklon et al. 2008	LH + 5 versus hCG + 2	NC versus COS
Haouzi et al. 2009b	LH (+ 2; + 7) versus hCG + (+ 2; + 5)	NC versus COS
Haouzi et al. 2009a	LH + 2 versus LH + 7	ES versus MS
Koler et al. 2009	CD 21	Fertility versus infertility
Altmäe et al. 2010	LH + 7	Fertility versus infertility
Haouzi et al. 2011	LH (+ 2; + 7) versus hCG (+ 2; + 5)	Ag versus Atg versus NC
Tseng et al. 2010	Dating by Noyes	ES versus MS versus LS
Van Vaerenbergh et al. 2010	LH + (5–7)	MS versus pregnant
Blockeel et al. 2011	Oocyte retrieval	rFSH versus low-dose hCG
Diaz-Gimeno et al. 2011	LH + 1, + 3, + 5 versus LH + 7 LH + (1–5) versus LH + 7 versus CD 8–12	LP versus ES versus MS
Labarta et al. 2011	hCG + 7	Different serum progesterone level
Van Vaerenbergh et al. 2011	Oocyte retrieval	Different serum progesterone level
Evans et al. 2012	LH + 2 versus LH + 7	Epithelial versus stromal cells in proliferative phase
Petracco et al. 2012	CD 1–3 versus CD 5–8 versus CD 11–13	EP versus MP versus LP
Diaz-Gimeno et al. 2013	Dating by Noyes versus ERA prediction	MP versus ES versus MS versus LS
Ruíz-Alonso et al. 2013	P + 5/LH + 7 RIF versus controls	pWOI/pWOI delayed/pWOI advanced
Bermejo et al. 2014	Oocyte retrieval COS	Comparing 4 GnRH-a protocols
Haouzi et al. 2014	hCG + 2 versus hCG + 7	Different serum progesterone level
Ruíz-Alonso et al. 2014	P + 5 versus P + 7	ET versus pET

Note that endometrial disorders such as cancer, endometriosis, and myomas are not considered in this table.

Based on the data published in Díaz-Gimeno et al. 2014.

Abbreviations: Ag, agonist; Atg, antagonist; CD, cycle day; COH, controlled ovarian hyperstimulation; COS, controlled ovarian stimulation; EP, early-proliferative; ERA, Endometrial Receptivity Array; ES, early-secretory; GnRH-a, Gonadotropin releasing hormone-agonist; hCG + , hCG administration + days; LH + , LH surge + days; LP, late-proliferative; LS, late-secretory; M, menstrual; MP, mid-proliferative; MS, mid-secretory; NC, natural cycle; P + , progesterone + days; pWOI, personalized window of implantation; rCG + , rCG administration + days; RIF, recurrent implantation failure.

Most of these studies investigate the transcriptomic signature in the whole endometrial tissue without separating the different compartments. However, in some studies, laser capture micro-dissection has facilitated specific compartment gene expression profiles (Yanaihara et al. 2005; Evans et al. 2012; Ulbrich et al. 2013). Even the specific profiles for stromal cells and glands at different depths in the endometrium have been reported (Gaide Chevronnay et al. 2009).

Several groups have used transcriptomics to search for the molecular diagnosis of the different phases of the human endometrium (Carson et al. 2002; Kao et al. 2002; Borthwick et al. 2003; Riesewijk et al. 2003). Ponnampalam et al. (2004) were the first to propose the transcriptomic characterization of the human endometrium throughout the menstrual cycle. Based on data extracted from samples taken at different cycle phases, they identified seven main groups of genes with a similar expression pattern throughout the cycle. Each of these groups had an expression peak in one of the seven subphases (menstrual, early-proliferative, mid-proliferative, late-proliferative, early-secretory, mid-secretory, and late-secretory). This finding was later reinforced by Talbi et al. (2006), who found groups of genes with a peak of expression in all the analyzed stages (proliferative, early-secretory, mid-secretory, and late-secretory).

TRANSCRIPTOMICS OF ENDOMETRIAL RECEPTIVITY

The most commonly used strategy to search for the human endometrial receptivity signature was to compare expression profiles during the WOI versus other stages of the menstrual cycle (Carson et al. 2002; Kao et al. 2002; Borthwick et al. 2003; Riesewijk et al. 2003; Horcajadas et al. 2004a,b; Mirkin et al. 2005; Critchley et al. 2006; Franchi et al. 2008; Kuokkanen et al. 2010; Tseng et al. 2010; Díaz-Gimeno et al. 2011).

The most frequent comparisons that have been performed are the receptive phase (mid-secretory) versus prereceptive (early-secretory) (Carson et al. 2002; Riesewijk et al. 2003; Mirkin et al. 2005; Franchi et al. 2008; Haouzi et al. 2009b; Tseng et al. 2010; Díaz-Gimeno et al. 2011); receptive versus proliferative (Kao et al. 2002; Borthwick et al. 2003; Kuokkanen et al. 2010); and receptive versus postreceptive (late-secretory) (Critchley et al. 2006; Tseng et al. 2010) (reviewed in Horcajadas et al. 2007). There was a lack of consensus on the results obtained, which was likely caused by the variability in some of the parameters such as samples taken from the same or different patients, the decision to use a pool of samples or not, the day of the cycle on which samples were taken, or the type of data analysis undertaken (Horcajadas et al. 2004a,b; Giudice 2006; Simmen and Simmen 2006; Haouzi et al. 2011).

Despite the differences among all the studies performed, they all agreed on the existence of a specific transcriptomic profile during the WOI. This characteristic profile suggested that a unique transcriptional process occurs to achieve a receptive phenotype (Borthwick et al. 2003; Riesewijk et al. 2003; Horcajadas et al. 2008; Haouzi et al. 2009a,b; Díaz-Gimeno et al. 2011), being the gatekeeper is the progesterone receptor activation.

Overexpressed genes include those involved in the processes of metabolism, glandular secretion, cell differentiation, cell communication, innate immune response, response to stress, response to wounding, cell adhesion, and proteolysis regulation (reviewed at Ruíz-Alonso et al. 2012).

The implantation of the blastocyst in the endometrium activates the production of cytokines that modulate receptivity by regulating the expression of adhesion molecules in mammals (Simon et al. 1997). Leukemia inhibitory factor (LIF) is a cytokine that has been the focus of many studies (Aghajanova et al. 2008; Allegra et al. 2009; Rashid et al. 2011) because of its clear functional effect in the mouse model. Osteopontin (SPP1) has been the gene with the greatest consensus among most endometrial transcriptomic studies. The glycoprotein encoded by this gene, regulated by progesterone, is a ligand for αvβ3 integrin and mediates cellular adhesion and migration during implantation (Apparao et al. 2001). There are several studies indicating that regulation of the immune response occurs during embryonic implantation (Hannan and Salamonsen 2007; Salamonsen et al. 2007). It is consistent with the transcriptomics studies showing that during mid-secretory stage, an activation of responses to stress, defense, humoral immunity, innate immunity and injuries occurs (Díaz-Gimeno et al. 2011; Ruíz-Alonso et al. 2012). Among the genes up-regulated involved in these processes are: glycodelin, which decreases maternal immunological responses to the implanting embryo (Aghajanova et al. 2008); CXCL14, a chemokine that is thought to be the major recruitment stimulus for immune cells during the WOI (Talbi et al. 2006) as well as chemotaxis of natural killer cells to cluster around epithelial glands (Mokhtar et al. 2010); and IL15, which plays important roles in uNK cell proliferation and differentiation (Okada et al. 2000) involved in the recruitment of peripheral blood CD16(–) NK cells (Kitaya et al. 2005). Moreover, the protection of the embryo against several agents as free radicals and heavy metals is very important. Metallothioneins and GPXs (antioxidants) are also overexpressed at this stage (Talbi et al. 2006; Ruíz-Alonso et al. 2012).

It is worth highlighting the relevance of the l-selectin ligands for the conversion to the receptive phenotype. Whereas l-selectin is expressed in the blastocyst (Genbacev et al. 2003), the endometrial epithelial cells express its ligand (Wang et al. 2008).

CLINICAL TRANSLATION OF THE TRANSCRIPTOMIC SIGNATURE OF ENDOMETRIAL RECEPTIVITY

To prove the clinical applicability of endometrial transcriptomics as a diagnostic tool, it was necessary to find that samples clearly grouped according to the stage to which they belonged. During the past decade, our group has worked in this topic from basic research (Riesewijk et al. 2003) to the publication of the transcriptomic signature that identifies the expression level of 238 genes related to endometrial receptivity (Díaz-Gimeno et al. 2011). This molecular tool, named ERA, has been designed to identify the endometrial receptivity status in natural cycles at LH + 7, or in hormonal replacement cycles (HRT) 5 d after progesterone administration (P + 5) previously primed with estradiol. This customized array is coupled to a computational predictor that can identify the endometrium of a given patient regardless of its histological appearance (Díaz-Gimeno et al. 2011, 2013).

To train the computational predictor, gene expression profiles obtained from samples at different stages of the menstrual cycle (proliferative, prereceptive, receptive, and postreceptive) were used. This classification has a specificity and sensitivity of 0.8857 and 0.99758, respectively (Díaz-Gimeno et al. 2011). ERA is more accurate than histological dating and is a highly reproducible method, in the same patient using the same type of treatment even up to 40 mo apart (Díaz-Gimeno et al. 2011, 2013). For the first time, a molecular tool based on the expression of a cluster of genes has been clinically applied in reproductive medicine to assess the endometrial factor in patients with recurrent implantation failure (Ruíz-Alonso et al. 2013) with proven accuracy and consistency.

Although the endometrial WOI has been considered standard and constant in all women, transcriptomics have helped to figure out that personalization also applies to the endometrium. Once the personalized WOI is identified, a personalized embryo transfer (pET) plan is developed to transfer the embryo according to the day in which the endometrium is receptive (Fig. 1).

The diagnostic and clinical value of the ERA test in patients with recurrent implantation failure (RIF) has been analyzed in a prospective interventional, multicenter, clinical trial (Ruíz-Alonso et al. 2013). An RIF group of 85 patients with at least three previous failed ovum donation cycles, or IVF patients younger than 40 yr old with at least three failed IVF cycles was compared with a control group (with one or no previous failed cycle). In this trial, RIF and control patients underwent ERA-based endometrial receptivity diagnosis using an endometrial biopsy obtained either on day LH + 7 in a natural cycle or on day P + 5 in an HRT cycle. Whereas the ERA test identified 12% of the patients with displaced WOI in the control group, in the RIF group 26% of patients have a displaced WOI. This means that in one out of four patients with RIF, the endometrial factor is responsible. In these patients, a second ERA test confirmed the suspected WOI displacement, and personalized embryo transfer was performed accordingly. The clinical outcome of pET shows a 50% pregnancy rate (PR) and a 38.5% implantation rate (IR), which is similar to the control group (Ruíz-Alonso et al. 2013). Now, an international prospective randomized clinical trial on the effectiveness of the ERA test in the infertility work-up in the first appointment is ongoing (NCT:954758, see http://clinicaltrials.gov/).

CLINICAL APPLICABILITY OF ENDOMETRIAL TRANSCRIPTOMICS

Clinicians used to believe that the WOI length was ∼2 d, and it was the same in all women. Published data cited in the above section shows that the WOI as diagnosed by its transcriptomic signature is displaced in one out of four RIF patients (Ruíz-Alonso et al. 2013), and in 20% of the general population. Furthermore, no ongoing pregnancy results after embryo transfer in a patient with no receptive endometrium diagnosed by ERA, whereas a personalized embryo transfer in the same group of patients leads to 60% PR and 40% IR (Ruíz-Alonso et al. 2014).

This finding implies relevant clinical implications and points out that the endometrial factor should be properly diagnosed and considered, defining the personalized WOI of each woman to improve clinical results.

NEW PERSPECTIVES: NEXT-GENERATION SEQUENCING

Microarray-based expression profiling has technical limitations because it is limited by the nature of the probes included in the given platform as well as their sensitivity and specificity (Reis-Filho 2009; Díaz-Gimeno et al. 2014). Nowadays, next generation sequencing (NGS) is an emerging technology that allows measurement of gene expression by RNA sequencing (RNA-seq). This new technology is capable of sequencing all the mRNAs present in a given sample, even the 25% of genes with low expression that remain undetected with standard microarray technologies (Mane et al. 2009; Wang et al. 2009; Díaz-Gimeno et al. 2014).

Therefore, although transcriptomics based on microarray technology has sufficiently standardized procedures to allow its clinical applicability, it is likely to be challenged by newer global gene-expression analysis technologies, as reported by the MAQC consortium (Mane et al. 2009). Massively parallel sequencing requires an availability of high-performance computing and bioinformatics support that goes way beyond many research laboratories. Furthermore, quality control and standardization of the massively parallel sequencing experiments and data reporting are important issues to consider (Reis-Filho 2009). Nevertheless, progress in this direction is rapid, and the way is paved for the implementation of NGS for routine clinical diagnosis.

CONCLUSIONS

For decades, the study of endometrial receptivity has been restricted to the use of morphological criteria under the concept of anatomical medicine. Technological advances in recent years have enabled us the transition from anatomical to molecular medicine for the diagnosis of the human endometrium discovering the endometrial transcriptome during the menstrual cycle. This strategy has enabled the development of a molecular diagnostic test (ERA) that inform us about the timing when the WOI is open in a given patient. The clinical translation of this test allows a personalized embryo transfer that, as has already been shown, increases the success of embryo implantation in patients with recurrent implantation failure of endometrial origin. Although the transcriptome has rapidly advanced, next-generation sequencing technologies is now an emerging reality that will allow analysis of not only mRNAs, but also small RNAs and noncoding RNA, providing a more comprehensive view of the transcriptomic of the endometrium.