MicroRNA-31 is Significantly Elevated in Both Human Endometrium and Serum During the Window of Implantation: A Potential Biomarker for Optimum Receptivity

Abstract

The window of implantation of human embryos into the endometrium spans Cycle Days 20–24 of the 28-day menstrual cycle. However, uterine receptivity may not be reliably replicated in infertile patients throughout this span. Thus, it is of importance to be able to determine optimal receptivity through a minimally invasive measure. We screened expression of a number of candidate micro-RNAs (miRNAs) in endometrial tissues and serum collected from a panel of fertile women during both the proliferative phase and the secretory phase of a normal menstrual cycle. We found that several miRNAs were significantly elevated in endometrial tissues in the secretory phase versus the proliferative phase. One of these, miR-31, was found to be not only detectable in serum samples but also significantly elevated in the secretory phase versus the proliferative phase. MiR-31 is known to target several immunomodulatory factors, such as FOXP3 and CXCL12. We find that both of these factors are significantly downregulated in endometrial tissues during the secretory phase. Our data suggest that miR-31 is a potential biomarker for optimal endometrial receptivity, possibly operating through an immunosuppressive mechanism.

Introduction

The human endometrium is a dynamic tissue that undergoes cyclic changes each month to prepare the uterus for embryo implantation. The window of implantation is accepted to be 5 days in length, spanning Cycle Days (CD) 20–24 in a 28-day cycle [1]. With fresh in vitro fertilization, transfer of Day 6 embryos results in decreased implantation rates when compared to transfer of Day 5 embryos [2]. However, when cryopreserved Day 5 and 6 embryos are transferred to an appropriately prepared endometrium, implantation rates are equivalent, highlighting the importance of endometrial receptivity [3].

Although the relative importance of the state of the endometrium is appreciated, definitive biomarkers that predict endometrial receptivity have remained elusive. For example, the Noyes histologic criteria were previously used to predict endometrial receptivity, yet a prospective trial evaluating endometrial histology found that histologic dating failed to discriminate fertile from infertile patients [4]. The biologic presence of pinopodes in the endometrium has also been reported to mark the receptive state [5, 6], though subsequent studies did not support this conclusion [7–9]. Expression of the αvβ3 integrin correlates with the window of implantation [10], but an invasive biopsy is required to measure levels. In addition, its predictive clinical utility has been questioned [11–13]. While circulating levels of inhibin correspond with luteal function, midluteal serum levels of inhibin did not reflect endometrial receptivity [14]. Progesterone values are generally accepted as an indicator of ovulation but are not able to differentiate a receptive from a nonreceptive endometrium. These studies collectively demonstrate that a noninvasive, accurate marker is needed both to help identify patients with implantation defects and to signal the optimal timing for embryo transfer.

Emerging literature supports an important role for micro-RNAs (miRNAs) in human embryo implantation [15, 16]; miRNAs are small noncoding RNAs, 18–24 nucleotides in length. Binding of miRNAs to the 3′-untranslated region (UTR) of target mRNAs results in repression of translation [17] through a variety of mechanisms, including mRNA cleavage and decay [18, 19]. One miRNA can have myriad target mRNAs, serving to amplify the effects of each individual miRNA [20]. In addition to expression in tissues, miRNAs have been identified in the peripheral circulation either as a complex with miRNA processing proteins or enclosed in secreted microvesicles called exosomes [21]. Exosome micro-RNAs have been shown to be extremely stable over time and are protected from RNAse degradation [22, 23]. Currently, serum miRNAs have been utilized as noninvasive biomarkers to predict survival or treatment response in cancer patients and are being investigated as markers for other diseases [24].

While the miRNA profile of the normal cycling endometrium has been characterized [25], to date no studies have examined the circulating miRNA profile during the normal menstrual cycle. The goal of this study was to investigate whether expression of select miRNAs in the normal cycling endometrium corresponds with the serum exosome miRNA profile in order to identify a noninvasive miRNA biomarker for the receptive endometrium.

Materials and Methods

Human Subjects

The study was reviewed and approved by the University of Iowa Institutional Review Board. Fifteen naturally cycling fertile females were recruited. Inclusion criteria were women age 18–40, with a live birth in the last 7 yr and regular menstrual cycles occurring every 25–32 days. Patients were excluded for any of the following: pregnancy, actively attempting pregnancy, use of hormonal contraception or intrauterine device for the preceding 3 mo, lactation in the preceding 3 mo, history or treatment of infertility, or unprotected intercourse with no pregnancy for at least 1 yr. Those women qualifying to participate agreed to a pregnancy test, two peripheral venipunctures (8 ml), two transvaginal ultrasounds, and two endometrial biopsies: one in the proliferative phase (CD 7–10) and one in the secretory phase (CD 20–24). A progesterone level was determined in the secretory-phase blood sample by the University of Iowa Diagnostic Laboratories. Patients were included if the progesterone level was > 3 ng/ml, consistent with ovulation. A total of 15 patients were enrolled in the study. Two patients withdrew from the study prior to completion, and one patient was excluded, as her progesterone value was less than 3 ng/ml. A total of 12 patients were included in the full study. The average age of participants was 33.3 yr. Patient-identified ethnicity revealed 75% Caucasian, 16.7% black, and 8.3% Hispanic. Patient and cycle characteristics are provided in Table 1.

Table 1.

Demographic and cycle characteristics of study participants.

Parameter	Value	Range
Age (yr)	33.3 ± 3.87	27–39
Body mass index	26.8 ± 7.4	19.8–40.2
Midsecretory progesterone level (ng/ml)	11.3 ± 5.7	3.2–18.8
Stripe 1 (cm)	0.55 ± 0.2	0.34–0.83
Stripe 2 (cm)	1.14 ± 0.3	0.81–1.74

Tissue Collection and RNA Preparation

Endometrial tissue collection and RNA preparation.

Endometrial biopsies were obtained using an endometrial vacuum curette (Pipelle), and the tissue was immediately flash frozen in liquid nitrogen. Total cellular RNA was purified using the Mirvana Kit according to the manufacturer's protocol (Applied Biosystems/Life Technologies). RNA yield and purity was assessed on a Nanodrop Model 1000 (Thermo Scientific).

Serum collection and RNA preparation.

For miRNA isolation from exosomes, blood was collected in a heparin-free serum separator tube (BD Biosciences) and spun at 1300 × g for 10 min. Serum was removed and stored at −80°C. Exosomes were isolated from 400 μl of serum using Exoquick (System Biosciences). Briefly, 120 μl of Exoquick solution were added to 400 μl of serum, mixed, and incubated at 4°C overnight. Following incubation, tubes were centrifuged at 1500 × g for 30 min. The supernatant was removed and the exosome-containing pellet resuspended in 100 μl PBS. Exosome RNA was isolated via a standard Trizol extraction (Life Technologies). RNA mass and concentration was determined on an Agilent Model 2100 Bioanalyzer.

miRNA Reverse Transcription and Expression Assay

Endometrium.

The miRNA-specific reverse transcriptions and expression assays were carried out on 350-ng total cellular RNA aliquots using Applied Biosystems/LifeTechnologies miR-specific TaqMan quantitative PCR primer sets and reagents. These assays are individually optimized with primer efficiencies between 0.9 and 1.0.

For the mRNA analysis, 500 ng of total RNA was reverse transcribed into cDNA using the SuperScript III kit (Invitrogen/Life Technologies). FOXP3, CXCL12, and E-selectin mRNA expression in endometrial biopsies were measured using SYBR Green real-time PCR using 18S as the reference gene. Primer sequences were validated in prior studies [26–28]:

FOXP3 forward 5′-GAGAAGCTGAGTGCCATGCA-3′ (T_m = 57.9°C), FOXP3 reverse 5′-AGGAGCCCTTGTCGGATGAT-3′ (T_m = 58.3°C); CXCL12 forward 5′-TCAGCCTGAGCACAGATGC-3′ (T_m = 56.8°C), CXCL12 reverse 5′-CTTTAGCTTCGGGTCAATGC-3′ (T_m = 54.1°C); E-SELECTIN forward 5′-GGCAGTGGACACAGCAAATC-3′ (T_m = 56.8°C), E-SELECTIN reverse 5′-TGGACAGCATCGCATCTCA-3′ (T_m = 56.6°C); 18S forward 5′-AACTTTCGATGGTAGTCGCCG-3′ (T_m = 57.3°C), 18S reverse 5′-CCTTGGATGTGGTAGCCGTTT-3′ (T_m = 57.6°C).

Quantitative PCR amplifications were performed on an HT7900 Fast Real-Time PCR System (Applied Biosystems). All assays were done with three technical replicates.

Serum.

Fixed aliquots (3 μl each) of exosome RNA were reversed transcribed using the TaqMan micro-RNA reverse transcription kit (Invitrogen/Life Technologies) with the Megaplex A & B set primers pools (Human Pools v3.0, Applied Biosystems/Life Technologies). Resulting cDNAs were then preamplified using Megaplex human A & B set preamp primers (Applied Biosystems/Life Technologies). Equal aliquots of preamplification products were assayed for miRNA expression levels using the same mir-specific TaqMan quantitative PCR primer sets and reagents as described for the endometrium tissue samples. These amplifications were also performed in triplicate on an HT7900 Fast Real-Time PCR System (Applied Biosystems).

Cell Culture and RNA Extraction

Ishikawa H endometrial cancer cells (gift from Dr. Erlio Gurdipe, New York University) were grown in DMEM media with 10% fetal bovine serum and penicillin-streptomycin (Gibco). Cells were treated with 100 nM progesterone or vehicle control (DMSO) for 24 h. Total RNA was extracted using the miRvana miRNA Isolation Kit (Ambion/Life Technologies). RNA yield and purity were assessed on a NanoDrop Model 1000 spectrophotometer (ThermoScientific). The miRNA-specific reverse transcriptions and expression assays were carried out on 500-ng total RNA aliquots using miR-specific TaqMan quantitative PCR primer sets and reagents (Applied Biosystems/Life Technologies).

Data Analysis

Tissue and serum micro-RNA expression levels were assessed following normalization against endogenous controls (ΔCt). Endometrium miRNA expression was normalized against the snoRNA RNU48, while serum miRNA expression was normalized against snRNA U6 (Applied Biosystems). RNU48 normalization was selected for the endometrial tissues, as it was consistently expressed in all samples. U6 normalization was chosen for serum because it is present in both the A set and the B set Megaplex primer pools, and it displayed consistent expression in serum. Comparisons of normalized tissue and serum miRNA expression values (ΔCt) employed the conventional ΔΔCt-fold change method [29, 30]. Statistical significance was assessed via a conventional t-test with unequal variances [31].

Comparison of endometrial FOXP3 and CXCL12 mRNA expression with endometrial tissue miR-31 expression was done using a standard Pearson correlation analysis [31]. Statistical significance was assigned at P < 0.05.

Results

Differential Endometrial miRNA Expression in the Window of Implantation

Endometrial expression of several miRNAs has been reported to change during the window of implantation [32–34] or to vary in patients with implantation failure [16]. We selected five miRNAs (30b, 30d, 31, 203, and 503) because they had consistent changes in two prior studies; one performing microarray analysis [32] and the other with deep sequencing analysis on proliferative and secretory endometrial tissue [33]. Due to the reproducibility of these results, we felt these miRNAs were likely to be differentially regulated in endometrial tissue. We confirmed with miR-specific quantitative PCR that miR-30b, -30d, -31, and -203 were significantly overexpressed and that miR-503 and -145 were significantly underexpressed in endometrial tissue obtained in the midsecretory phase as compared to the proliferative phase (Table 2). Since miR-135a and -135b have been implicated as important regulators of HOX genes during implantation [35, 36], we also examined their expression but found no difference between CD 7–10 and 20–24 (Table 2).

Table 2.

Expression of endometrial miRNAs in the window of implantation relative to the proliferative phase.

miRNA	Fold change	P-value
hsa-miR-30b	2.96	0.014
hsa-miR-30d	2.74	0.002
hsa-miR-31	1.49	0.038
hsa-miR-135a	−1.15	0.582
hsa-miR-135b	−1.41	0.104
hsa-miR-145	−1.53	0.018
hsa-miR-203	2.42	<0.001
hsa-miR-503	−2.01	0.021

Relationship Between Serum and Endometrial miRNA Expression

Given that the endometrium is regenerated and shed each month, we hypothesize that the endometrium secretes exosomes into the circulatory system such that the expression of some serum miRNAs would accurately reflect endometrial tissue miRNA expression. Therefore, we purified exosomal RNAs from patient serum collected during the proliferative and midsecretory phases. With the exception of miR-503, all miRNAs detected in the endometrial tissue were also present in the serum (Fig. 1A). Furthermore, serum expression of miR-30b, -30d, and -203 demonstrated a trend toward upregulation at CD 20–24 as compared to CD 7–10, consistent with the observed changes in tissue miRNA expression. Similarly, a nonsignificant decrease in serum miR-145 paralleled its decrease in the endometrium. By contrast, miR-31 was significantly elevated in serum in the window of implantation when compared to the proliferative-phase serum (Fig. 1, A and B). These data suggest that, although the overall trends of specific miRNAs in the serum mirror those in the endometrium, miR-31 was the only miRNA with statistically significant upregulation in both tissue and serum during the window of implantation.

Fig. 1.

Endometrial and serum miRNA changes in the window of implantation. A) Differentially regulated endometrial tissue miRNAs (black; see also Table 2; n = 12) were assayed and compared to their level in the serum (crosshatch, n = 11). Data are reported as the fold change in expression between the midsecretory and proliferative phases as determined by ΔΔCT method (*P < 0.05). B) Distribution of serum miR-31 expression in patient samples on CD 7–10 and CD 20–24 (N = 11). C) Expression of mir-31 in Ishikawa cells after treatment with 100 nM progesterone for 24 h normalized against DMSO control exposure.

We next evaluated whether increasing progesterone levels in the midluteal phase might contribute to the increase in miR-31. Treatment of progesterone receptor-positive Ishikawa H type I endometrial cancer cells with progesterone for 24 h resulted in a nearly 85% increase in miR-31 expression (Fig. 1C), suggesting that progesterone influences expression of miR-31 in vitro.

Increased miR-31 Corresponds with Downregulation of Immune Targets

MiR-31 was the only miRNA that showed a significant change in expression in both the endometrial tissue and serum in the window of implantation. Using the list of validated miR-31 mRNA targets provided though miRBase [37], we identified three targets of miR-31 that have a plausible role in embryo implantation. These are FOXP3 [38], a transcription factor denoting T regulatory (T_reg) cells; CXCL12 [39], a chemokine for lymphocyte trafficking; and E-selectin [40], a receptor involved in lymphocyte trafficking. Using real-time quantitative PCR, we found both FOXP3 and CXCL12 to be significantly downregulated in endometrial tissue in the window of implantation as compared to the proliferative phase (Fig. 2A) Linear regression analyses demonstrate a significant inverse relationship between endometrial miR-31 and FOXP3 (R = −0.40, P = 0.05) and CXCL12 (R = −0.53, P = 0.008). These data are consistent with both FOXP3 and CXCL12 being direct targets of miR-31 as reported [38, 39]. E-selectin also trended toward downregulation in CD 20–24 but failed to reach statistical significance.

Fig. 2.

Downregulation of immunomodulatory miR-31 targets in the endometrium during the window of implantation. A) Fold change in endometrial miR-31 expression and its validated targets CXCL12, E-selectin, and FOXP3 during the window of implantation relative to the proliferative phase (n = 12). Data are reported as the fold change in expression between the midsecretory and proliferative phases as determined by ΔΔCT method (*P < 0.05, **P < 0.01). B) Schematic of dynamic changes in miR-31 and its targets in the endometrium and serum during the menstrual cycle. Our data demonstrate significant upregulation of miR-31 during the window of implantation in both the endometrium and serum (dotted blue line). Conversely, miR-31 targets of FOXP3 (red), CXCL12 (green), and E-selectin (yellow) were decreased in the endometrium (represented by dotted lines). Also depicted are findings by Arruvito et al. [43], who quantitated FOXP3-positive T_reg cells in peripheral blood throughout the menstrual cycle (red solid line).

Discussion

A receptive endometrium is a prerequisite for successful embryo implantation, but reliable, informative biomarkers have remained elusive. Using miRNA-specific q-PCR, we demonstrated significant changes in the expression of six miRNAs, miR-30b, -30d, -31, -145, -203, and -503, in the endometrium during the window of implantation. Application of these same miR-specific assays in serum from the same subjects confirmed a statistically significant increase in the expression of miR-31 in the midsecretory phase. In addition, mRNA expression of validated miR-31 targets FOXP3, a transcription factor denoting T regulatory cells, and CXCL12, a chemoattractant for uterine natural killer cells, showed a significant inverse association with changes in miR-31 levels. Our data, therefore, suggest a model (Fig. 2B) in which increased miR-31 expression may assist in producing an immune-tolerant maternal environment favoring implantation through targeting FOXP3 and CXCL12.

The miRNAs have been proposed as potential serum biomarkers for myriad conditions, especially cancer. For example, in ovarian cancer, serum micro-RNA levels were found to correspond with tumor micro-RNA levels [41]. During the window of implantation, miR-31 was significantly upregulated in both serum exosomes and endometrial tissue. This raises the possibility that serum miR-31 expression could be developed as a minimally invasive monitor of miR-31 activity in the endometrium.

Several studies suggest that miR-31 plays a role in regulating the immune system, including directly targeting the transcription factor FOXP3, a master regulator of T_reg differentiation and functional activity [39]. T_reg cells induce tolerance by inhibiting proliferation and cytokine production of lymphocytes, cytotoxic activity of natural killer cells, and maturation of dendritic cells, although the precise mechanisms are unknown [42]. Emerging evidence suggests that T_reg cells are critical for embryo implantation and pregnancy. For instance, FOXP3 expression is 43% lower in midluteal endometrial biopsies of women with unexplained infertility when compared with normal fertile controls, suggesting that a diminished T_reg cell population may compromise implantation success [26]. The number of T_reg cells in the peripheral blood increase during the follicular phase and are reduced in the luteal phase in normal cycling fertile females [43]. By contrast, the follicular-phase expansion is absent in women with recurrent spontaneous abortion (RSA) [43]. Taken together, these data indicate that successful implantation may require a preovulatory increase in T_reg cells to induce immune tolerance. Our findings extend these observations by demonstrating that serum miR-31, a negative regulator of FOXP3, is significantly upregulated during the midluteal phase (Fig. 2B). Moreover, endometrial tissue FOXP3 expression is consistent with the reported serum FOXP3 T_reg expression pattern, with a decrease in the midluteal phase when compared to the follicular phase. Since this study used endometrial biopsies, which contain numerous cell types, it is not possible to define the cellular source of FOXP3 and CXCL12 in the biopsied tissue.

In addition to FOXP3, miR-31 has also been reported to target other immunomodulatory factors, such as CXCL12 and E-selectin, that may play a role in maternal immune acceptance of an implanted embryo [44, 45]. CXCL12, a chemokine, and its receptor CXCR4 have been shown to be important in uterine natural killer cell recruitment, placentation, implantation, and embryogenesis [46–48]. We observed decreased CXCL12 mRNA levels in midluteal endometrial tissue when compared to the follicular phase. This is in contrast to a prior report that found no difference in CXCL12 mRNA expression in endometrial biopsies obtained at different times in the menstrual cycle [27]. Our CXCL12 findings may be different due to dissimilar categorizations of the biopsies as well as the larger sample size in our study during the window of implantation (CD 20–24). E-selectin, another validated target of mir-31, is exclusively expressed by the endothelium and preferentially binds Th1 lymphocytes directing lymphocyte trafficking [49]. Although currently debated, the classic paradigm is that a Th2 > Th1 cytokine response at the maternal-fetal interface supports pregnancy. We observed a trend toward downregulation of E-selectin during the midluteal-phase endometrial tissue, which may result in blocking Th1 recruitment, thereby allowing a Th2 > Th1 cytokine response.

In summary, our results identify miR-31 to be significantly overexpressed in both endometrial tissue and serum exosomes during the window of implantation of normal cycling fertile females when compared to the proliferative phase. Thus, serum miR-31 could serve as a peripheral monitor of miR-31 activity in endometrial tissue. Identification of a minimally invasive biomarker that would mark an optimally receptive endometrium has important implications not only for determining the appropriate time for embryo transfer but also for identifying patients with deficits in endometrial receptivity. In contrast to endometrial biopsy, serum collection is noninvasive and can be done during the cycle of attempted pregnancy without the concern of interfering with embryo implantation. Serum exosome miRNAs are an extremely stable, RNase-resistant, easily assayed RNA species. One limitation of this study was the relative small sample size. Despite this, we were still able to identify miRNAs that were significantly different between the proliferative and secretory phases in several different study subjects. Future validation and prospective studies to evaluate the ability of serum miRNA-31 expression levels to predict embryo implantation could deliver it as a biomarker of a receptive endometrium and, perhaps, a diagnostic tool for understanding implantation failure.