https://orcid.org/0000-0002-0039-9846
Abstract
Endometrial liquid biopsy (ELB) is a minimally invasive alternative for research and diagnosis in endometrial biology.
We sought to establish an endometrial micro ribonucleic acid (miRNA) roadmap based on ELB during the secretory phase of the menstrual cycle in both natural and hormonal replacement therapy (HRT) cycles.
Human ELB samples (n = 58) were obtained from healthy ovum donors undergoing a natural and an HRT cycle consecutively. miRNA profiles were identified using next-generation sequencing (NGS). For functional analysis, messenger ribonucleic acid targets were chosen among those reported in the endometrial receptivity analysis.
The human endometrial secretory phase is characterized by a dynamic miRNA secretion pattern that varies from the prereceptive to the receptive stages. No differences in miRNA profiles were found among natural versus HRT cycles in the same women, reinforcing the similarities in functional and clinical outcomes in natural versus medicated cycles. Bioinformatic analysis revealed 62 validated interactions and 81 predicted interactions of miRNAs differentially expressed in the HRT cycle. Annotation of these genes linked them to 51 different pathways involved in endometrial receptivity.
This NGS-based study describes the miRNA signature in human ELB during the secretory phase of natural and HRT cycles. A consistent endometrial miRNA signature was observed in the acquisition of endometrial receptivity. Interestingly, no significant differences in miRNA expression were found in natural versus HRT cycles reinforcing the functional clinical similarities between both approaches.
Liquid biopsy, or fluid biopsy, is the analysis of molecules in biological fluids. This technique is a promising alternative to diagnostic methods that utilize tissue fragments. The first liquid biopsies reported were used to diagnose breast, colon, and prostate tumour deoxyribonucleic acid (DNA) in the blood stream (1). Other compounds that can be identified within a liquid biopsy are extracellular vesicles, proteins, lipids, or micro ribonucleic acid (miRNAs), among others (2, 3).
The microenvironment that a human embryo encounters at the entrance of the uterine cavity is composed of a wide range of molecules confined to a small volume of fluid, commonly known as endometrial fluid (EF). The components of the human EF originate from the luminal and glandular epithelium through passive and active secretions. A large number of amino acids, ions, carbohydrates, lipids, proteins (including cytokines and enzymes), hormones, growth factors, transporters (4), and free or exosome-associated miRNAs are present (5,6).
miRNAs identified in blood have been used as cancer biomarkers and are found in different bodily fluids, such as tears, breast milk, saliva, seminal fluid, and urine (7,8). They play an essential role in gene regulation acting as cell-to-cell signaling molecules (9). Mature miRNAs are noncoding ribonucleic acid (RNA) molecules about 22 to 25 nt in length that regulate up to hundreds of messenger RNA (mRNA) targets through perfect or imperfect complementation with the 3’ untranslated regions of their target transcripts. Thus, depending on the degree of complementarity, the interaction may result in an inhibition of mRNA translation or promotion of its degradation leading to posttranscriptional gene silencing (10) Our lab has demonstrated the existence of miRNAs within the EF, specifically the presence of hsa-miR-30d as an exosome-associated molecule. This miRNA can be internalized in the trophectodermal cells of mouse embryos, regulating genes encoding specific molecules involved in the murine embryonic adhesion to endometrial epithelial cells including Itgb3, Itga7, and Cdh5. These findings support a model in which maternal endometrial miRNAs might act as transcriptomic modifiers of the preimplantation embryo (11).
Utilizing assisted reproductive technology, human embryos can be transferred into the mother while synchronizing the endometrium with natural cycles or hormonal replacement therapy (HRT) cycles. The natural cycle is under the control of ovarian steroid hormones and is timed after the LH peak (LH + 0) with embryo transfer scheduled at LH + 7. HRT cycles use exogenous estrogen (E2) and progesterone (P4) to mimic the natural cycle. E2 administration starts on days 1 to 3 of the menstrual cycle. After 7 days of exposure, when the endometrial thickness reaches 6 to 7 mm with a trilaminar pattern, P4 is administered for 5 days at different dosages and routes of administration (vaginal, intramuscular, or subcutaneous) for the acquisition of endometrial receptivity. The day that progesterone administration begins is considered P + 0 and embryo transfer is generally performed at P + 5.
The search for objective biomarkers to classify endometrial receptivity status has become a challenge in reproductive medicine. The transcriptomic (12,13), proteomic (14), and lipidomic (15,16) signatures of endometrial receptivity have been described, as has the miRNA profile of the human endometrium during the receptive stage (17–19). In this study, we aim to provide the secreted miRNAs roadmap present in the human endometrial liquid biopsy (ELB) during the secretory phase, both in natural and HRT cycles.
Material and Methods
Ethics statement
This study was approved by the Institutional Review Board and Ethics Committee of IVI Valencia, Spain (1304-C-117-FV). Written informed consent was obtained from each patient before ELB aspiration.
Study design
A total of 40 healthy volunteers from the ovum donation program, aged between 18 and 34 with regular menstrual cycles, normal karyotype, and body mass index between 19 and 29 kg/m2 were recruited for ELB collection. ELB was obtained in each patient in 2 cycles consecutively, first in a natural cycle guided by the determination of the LH peak and subsequently in an HRT cycle timed by the day of exogenous progesterone administration at P + 0. ELB was performed only once per cycle in each patient. Of the 80 ELBs initially performed, only 58 were included in the study because some women decided to stop participation in the study. Thus, the final number and timing of samples obtained in natural cycles was as follows: LH + 0 (n = 9), LH + 3 (n = 9), LH + 5 (n = 7) and LH + 7 (n = 2) and in HRT cycles: P + 0 (n = 9), P + 1 (n = 8), P + 3 (n = 8) and P + 5 (n = 6).
Natural cycles
A basal transvaginal ultrasound was performed the first 3 days of menstruation to ensure that the ovaries were quiescent with no cysts. Then, follow-up was performed from days 8 to 11 of the menstrual cycle monitoring follicle growth. When a dominant follicle with 16-mm of diameter was observed, patients carried out a urine LH test every 12 hours and were screened by ultrasound every day until a positive LH peak was recorded. The day of identification of the LH peak was considered LH + 0.
HRT cycles
During menstruation, a baseline transvaginal scan was carried out to ensure that the endometrium and ovaries were in basal conditions. For all patients, a daily dose of 0.25 mg Cetrorelix (Cetrotide, 0.25 mg Merck-Serono) was administered for 5 days together with an oral administration of 6 mg estradiol valerate (Progynova; Schering Spain, Madrid, Spain). After 10 days, an ultrasound was performed to evaluate endometrial growth. When the endometrial lining had a minimum thickness of 7 mm with a typical triple layer pattern, 800 mg/day of micronized intravaginal progesterone (Progeffik; Effik Laboratories, Madrid, Spain) was added. The day of the initiation of progesterone administration was considered P + 0.
ELB technique
Endometrial fluid was aspirated with the woman lying in lithotomy position; the cervix was cleansed after insertion of the speculum and the cervical mucus was aspirated. A flexible catheter (Wallace, Smith Medical International) was gently introduced 6 cm transcervically into the uterine cavity. When the catheter was correctly positioned, a 10-mL syringe was connected to the catheter and suction was applied. To prevent contamination by cervical mucus during catheter removal, the outer sheath of the embryo transfer catheter was advanced to a depth of 4 cm from the external cervical os, followed by the cessation of suction. Approximately 10 to 50 μL of EF were obtained, snap-frozen, and stored at –80°C until analysis.
RNA extraction from ELB
RNA extraction was performed using the miRNeasy Mini Kit (Qiagen, Hilden, Germany) enabling the purification of total RNA from the EF including microRNAs. Treatment with deoxyribonuclease was carried out to avoid interference caused by the DNA. The concentration of RNA was determined by measuring the absorbance at 260 nm (A260) using a Nanodrop™ photometer (Thermo Scientific, Waltham, Massachusetts, US). The quality of the miRNA extraction was evaluated by an Agilent RNA 6000 Nano kit in an Agilent 2100 bioanalyzer (Agilent, Santa Clara, California, US)
Sequencing of the miRNAs present in the EF
Next-generation sequencing was performed using Illumina HISeq2000/100 to analyze microRNA species. Small RNAs were filtered from 200 ng-1μg of RNA sample. RNA segments of different sizes were separated by polyacrylamide gel electrophoresis (PAGE) gel; segments between 18 and 30 nt (14–30 single-stranded RNA; Ladder Marker, Takara, Kusatsu, Japan) were selected. 3’ and 5’ adaptor was ligated, and several rounds of polymerase chain reaction (PCR) amplification with a PCR primer cocktail and PCR master mix (Illumina, San Diego, California, US) were performed to enrich the complementary DNA fragments. PCR products were purified with PAGE. The final library was quantified using the Agilent 2100 bioanalyzer instrument (Agilent DNA 1000 Reagents, Santa Clara, California, US) and by real-time quantitative PCR (TaqMan Probe). The qualified libraries were amplified on cBot (Illumina, San Diego, CA) to generate the cluster on the flow cell. The amplified flow cell was sequenced in a single end mode on the HiSeq 2000 System (Illumina). Raw sequence data files are available in the Sequence Read Archive repository under the accession number: ID: PRJNA550541 (20).
Bioinformatics analysis
miRNA profiles were developed using miRDeep2 (version 2.0.07) following the default work flow (21). Briefly, the mapper module was used to align the reads to the human reference genome (GRCh37), not allowing mismatches and discarding reads shorter than 18 nucleotides. Mature and immature miRNA sequences from the 21st release of miRbase (5) were used by the quantifier module to identify miRNAs, generating an miRNA expression profile for each sample (21). After normalization of the expression profiles using the trimmed mean of M values method implemented in the edgeR package (22), the ROBPCA algorithm (23) was applied to detect outlier samples based on scores and orthogonal distances, computing three principal components for detection. Differential expression analysis was done using edgeR following the guidelines of the package (22). Comparisons were performed among the timepoints used in both treatments and among treatments within related clinical phases. miRNAs with a false discovery rate (FDR) <0.05 and absolute fold change (FC) >2 were considered significantly differentially expressed. The expression of the differentially expressed genes was represented with a heatmap, including the clustering analysis used to determine the distances among samples by the Euclidean method.
Functional analysis based on ERA genes
Targets for each differentially expressed miRNA were predefined using the miRTarBase database (release 7.0) (23) and MBSTAR software (24). A filtering of the first output was performed, selecting only those genes whose implication in the acquisition of a receptive endometrium had already been described (12). The transcriptomic signature of endometrial receptivity assay (ERA) genes was used to obtain gene expression profiles for 60 patients undergoing HRT (12). Considering their role in blocking gene expression, miRNAs that showed an inverse pattern of expression with ERA genes throughout the cycle were selected as potential candidates. The KEGG database (25) was used to identify functions in which these miRNAs could be involved.
In vitro target validation
To verify correlation between the tested miRNAs and their corresponding targets, an Ishikawa cell line (Sigma Aldrich; St. Louis, MO, US) was used as an in vitro model of endometrium. Ishikawa cells were seeded onto 24-well culture plates for 72 hours and cultured in minimum essential medium + 2 mM glutamine + 1% nonessential amino acids + 5% foetal bovine serum. They were maintained in a humidified atmosphere containing 5% carbon dioxide at 37°C and when they reached 50% confluence, they were transiently transfected with 50 nM of miRNA mimics (Syn-hsa-miR-30d-5p, Cat No. MSY0000245, 5’UGUAAACAUCCCCGACUGGAAG; hsa-miR-429 miRCURY LNA miRNA Mimic, Cat No. YM00472516-ADA, 5’UAAUACUGUCUGGUAAAACCGU; hsa-miR-141-3p miRCURY LNA miRNA Mimic, Cat No. YM00471957, 5’UAACACUGUCUGGUAAAGAUGG; hsa-miR-324-5p miRCURY LNA miRNA Mimic, Cat No. YM00473205, 5’CGCAUCCCCUAGGGCAUUGGUGU); miRNA mimics are chemically synthesized, double-stranded RNAs that mimic mature endogenous miRNAs after transfected into cells. 50 nM of miRNA inhibitors (miR-30d-5p miScript inhibitor, Cat No. MIN0000245, 5’UGUAAACAUCCCCGACUGGAAG; hsa-miR-429 miRCURY LNA miRNA Inhibitor, Cat No.; YI04101290, 5’UAAUACUGUCUGGUAAAACCGU; hsa-miR-141-3p miRCURY LNA miRNA Inhibitor, Cat No. YI04100687, 5’UAACACUGUCUGGUAAAGAUGG; hsa-miR-324-5p miRCURY LNA miRNA Inhibitor, Cat No. YI04104166, 5’CGCAUCCCCUAGGGCAUUGGUGU) were used.
Transfection was performed using HiPerfect (Qiagen, Valencia, CA, US). For protein and RNA extraction, cells were trypsinized for 5 minutes, centrifuged at 300 × g for 5 minutes and washed twice with phosphate-buffered saline (PBS) before extraction. Total RNA was extracted using the RNeasy Mini Kit (Qiagen, Hilden, Germany) from cells at 24, 48, and 72 hours posttransfection. Concurrently, protein extraction was carried out 48, 72, and 96 hours posttransfection. Experiments were performed in triplicate.
Protein extraction from cell culture
To obtain cellular lysates, cells were washed with PBS and frozen at –80°C for at least 1 hour. They were then incubated with 0.5 mL of RIPA buffer (150 mM NaCl, 1% IGEPAL CA 630, 0.5% Na-DOC, 0.1% SDS, 0.5 M EDTA, 50 mM Tris-HCl; pH 8) and supplemented with protease inhibitors (1% PMSF 0.1 M; Sigma-Aldrich, Madrid, Spain, 10% Roche mini complete; Roche, Basel, Switzerland) for 1 hour at 4ºC. Once lysis was completed, the supernatant was centrifuged at 13,000 × g for 30 minutes. The pellet was discarded, and the resulting supernatant was frozen until use. The proteins were quantified using a Bradford assay. Briefly, 5 µL of protein extract were mixed with 200 µL of a Bradford solution (150 µL distilled water + 50 µL Bradford reagent; QuickStart™ Bradford 1× dye reagent; Bio-Rad, Hercules, CA, US) in a 96-well microplate. As a standard curve, seven concentrations (2, 1.5, 1, 0.75, 0.5, 0.25, and 0.125 mg/mL) of bovine serum albumin (Sigma-Aldrich, St Louis, Mi, US) were used. The resultant mixture was incubated at RT for at least 5 minutes. Absorbance reads were performed at 595 nm in a Multiskan Go plate reader (Thermofisher Scientific, Waltham, Ma, US); 6× reducing Laemmli SDS sample buffer (Alfa Aesar, Haverhill, MA, US) was used as a loading buffer.
Western blot analysis
Samples were denatured by heating at 95°C for 5 minutes. They were analyzed by SDS-PAGE, followed by electroblotting onto polyvinylidene difluoride membranes (Bio-Rad Laboratories, Hercules, CA, US) that had been previously activated with methanol for 2 minutes. The Precision plus Protein Kaleidoscope Standard was used as a protein ladder (Ref:1610375; Bio-Rad Laboratories, Hercules, CA, US). Transfer quality was checked by Ponceau red staining. The membranes were blocked by incubation with 5% skimmed milk diluted in 1% PBS with Tween (PBST) for 1 hour at room temperature. The membranes were then incubated overnight with specific primary antibodies (1:1,000) diluted in 3% skimmed milk dissolved in 1% PBST according to the manufacturers’ specifications (KIF11 [ab37814; Abcam, Cambridge, United Kingdom] MAPK8 [ab199380; Abcam, Cambridge, United Kingdom], HOXA11 [NBP1-80228; Novus Biologicals, Centennial, CO, US], FOXO 1 [ab52857; Abcam, Cambridge, UK] and ACT [ab8227; Abcam, Cambridge, UK]). After incubation, the membranes were washed three times with 1% PBST and then incubated with the secondary antibody Goat antimouse IgG-HRP (sc-2005; Santa Cruz Biotechnology, Dallas, TX, US) in 1% PBST (1:10,000). Finally, the target proteins were detected using a SuperSignal West Femto Chemiluminescent kit (Thermo Fisher Scientific, Waltham, MA, US). Western blots of the whole lysates were performed in cuatriplicate. Relative quantification of the intensity signal was performed by means of the ImageJ software (http://rsb.info.nih.gov/ij/index.html) according to methods previously reported by the scientific community (https://lukemiller.org/index.php/2010/11/analyzing-gels-and-western-blots-with-Image-j/). ACT was used as a housekeeping control to normalize the quantifications. Uncropped gels are available from (26).
Retrotranscription–quantitative polymerase chain reaction
Retrotranscription–quantitative PCR (RT-qPCR) was performed to determine the levels of the tested miRNAs (hsa-miR-30d, hsa-miR-429, hsa-miR-141-3p, and hsa-miR-324) and their respective mRNA targets: hsa-miR-30d (KIFF11, MAPK8, HOXA11); hsa-miR-429 (RASSF2, IGF2), hsa-miR-141-3p (RASSF2), hsa-miR-324-5p (CKB, FOXO1, KCNG1, ATP5B, TACC3) (n = 3). RT-qPCR for miRNAs were performed with miScript reverse transcription and miScript SYBR Green PCR kits (Qiagen, Hilden, Germany), respectively. To determine the relative quantity of the corresponding mRNAs targets, a SuperScript IV VILO Master Mix (Thermo Fisher Scientific, Waltham, MA, US) and the KAPA SYBR FAST qPCR universal 2× master mix (Sigma Aldrich, St. Louis, MO, US) were used. Fold changes were estimated using the –2ΔΔCt formula. Actin and SNORD96 were used as housekeeping controls for relative mRNA and miRNA quantification, respectively; the primer sequences used are detailed next. miRNAs: hsa-miR-30d, 5’UGUAAACAUCCCCGACUGGAAG; hsa-miR-429, 5’UAAUACUGUCUGGUAAAACCGU; hsa-miR-324-5p, 5’CGCAUCCCCUAGGGCAUUGGUGU; Hs_SNORD96A_11 miScript, Cat No. MS00033733. mRNAs: KIF11, Fw: 5’-GGC AGT TGA CCA ACA CAA TG-3’, Rv; 5’-TCT AGC ATG GCC TTT TGC TT-3’; MAPK8, Fw: 5’-CTGAAGCAGAAG CTCCACCA-3’, Rv; 5’-CACCTAAAGGAGAGGGCTGC-3’; HOXA 11, Fw: 5’-GGGATGCATGGAGATAGCCC-3’, Rv: 5’-CTGCAGGCTCCAAGAGTGAA; RASSF2, Fw: 5’-AAA AGAGAGGACAGCGGACG-3’, Rv: 5’-GGACTAGGGACG TTTGGTGG-3’; IGF2, Fw: 5’-CGCCGAACCAAA GTGGATTA-3’, Rv: 5’-GTGGGAGAGACAGAGTGAACG-3’; CKB, Fw: 5’-GTACATCATGACCGTGGGCT-3’, Rv: 5’-TTGA GGTCGGTCTTGTGCTC-3’; FOXO 1, Fw: 5’-GCTGCCAA GAAGAAAGCATC-3’, Rv: 5’-CATCCCCTTCTCCAAG ATCA3’, KCNG1, Fw: 5’-AGGGATGTGAAGGCCCAAAA-3’, Rv: 5’-GGTAGAACGCGCCCTTGAT-3’; ATP5B, Fw: 5’-TGCT CCCATTCATGCTGAGG-3’, Rv: 5’-CTCCAGCACCACCAA AAAGC-3’; TACC3, Fw: 5’-GGACCTGGATG CAGTGGTAA-3’, Rv: 5’-CCCCAGTTCCAGGTTCTTCC-3’; ACT, Fw: 5’-AGCACAGAGCCTCGCCTTT-3’, Rv: 5’-GATGCCTCTCTTGCTCTGGG-3’. Samples used to validate the sequencing results are from a different set of samples not related to the primary samples used to identify the differentially expressed miRNAs.
Statistics
Statistical analyses were performed using the Mann–Whitney U test for miRNA validation. The student’s t-test was used for the comparison of control, mimic, and inhibitor. Significance was defined as P < .05.
Results
miRNA profile in ELB during the secretory phase in natural and HRT cycles
A preliminary analysis of the raw data was performed to explore some basic technical features such as the number of well-covered miRNAs per sample and/or the identification of the miRNAs with the highest percentage of reads per sample. This screening showed that the 58 EF samples included in the study had an average of 888.81 ± 178.61 expressed miRNAs (Fig. 1A). Specifically, hsa-miR-10b-5p was the most abundant miRNA (as suggested by the number of reads identified) in 21 samples of the 27 analysed in natural cycles and in 18 samples of the 31 included for HRT cycles. In addition, hsa-miR10a-5p and hsa-miR-486-5p were also expressed with high prevalence in both natural and HRT cycles (Fig. 1B).
miRNA signature of the receptive endometrium in both natural and HRT cycles. (A) Number of expressed miRNAs per sample analysed. X axis reflects the different samples included in the study per phase of the cycle analysed in both natural and HRT cycles. Y axis refers to the number of miRNAs identified in every sample analysed. Colour legend: Purple (miRNAs detected in LH + 0 and P + 0), blue (miRNAs detected in LH + 3 and P + 1), dark green (miRNAs LH + 5 and P + 3), light green (miRNAs in LH + 7 and P + 5). Graph on the left represents the distribution of the number of miRNAs detected. (B) Percentage of reads of the most expressed miRNAs per sample. X axis reflects the different samples included in the study per phase of the cycle analysed in both natural and HRT cycles. Y axis refers to the percentage of reads (%) identified in every sample analysed. Color legend: Purple (% of reads detected in LH + 0 and P + 0), blue (% of reads detected in LH + 3 and P + 1), dark green (% of reads detected in LH + 5 and P + 3), light green (% of reads detected in LH + 7 and P + 5). Graph on the left reflects the number of samples with the higher number of reads for hsa-miR-10a-5p, hsa-miR-10b-5p, and hsa-miR-486-5p. (C) Heat map diagram and hierarchical clustering based on summarized intensity values of differentially expressed miRNAs between pre-receptive (LH + 3) and receptive phases (LH + 7) in natural cycles. The colour scale shows the relative expression level (log2-fold change) of a certain miRNA across all samples (red: high expression, white: moderate expression and grey: low expression). (D) Heat map diagram and hierarchical clustering based on summarized intensity values of differentially expressed miRNAs between pre-receptive (P + 3) and receptive phases (P + 5) in HRT cycles. The color scale shows the relative expression level (log2-fold change) of a certain miRNA across all samples (red: high expression, white: moderate expression and grey: low expression).
From the 58 samples examined, 12 were detected as outliers using the ROBPCA method and subsequently discarded. As a result, the final number of samples used in natural cycles were LH + 0 (n = 7), LH + 3 (n = 8), LH + 5 (n = 6), and LH + 7 (n = 2) and in HRT cycles: P + 0 (n = 6), P + 1 (n = 8), P + 3 (n = 5) and in P + 5 (n = 4). To determine the miRNA profile associated with the secretory phase, a preliminary screening was performed selecting only miRNAs presenting an absolute FC equal or higher than two (FC ≥2) and a FDR less than 0.05 (FDR <0.05).
In the natural cycle, a total of 88 differentially expressed miRNAs were identified (26). When comparing the nonreceptive (LH + 0) versus receptive (LH + 7) endometrium, 21 differentially expressed miRNAs were observed, of which 15 were downregulated and 6 were up-regulated. Similarly, 6 miRNAs were downregulated at LH + 3 versus LH + 7 (prereceptive vs. receptive status). No differences were found between LH + 5 and LH + 7.
In the HRT cycles, a total of 173 differentially expressed miRNAs were identified (26). More specifically, 61 were downregulated and 36 upregulated in P + 0 versus P + 5 (nonreceptive vs. receptive status, respectively). Interestingly, 17 differentially expressed miRNAs could be detected between P + 3 versus P + 5 (prereceptive vs. receptive endometrium, respectively). However, no significant differences could be appreciated in the subset of miRNAs compared between P + 0 versus P + 1 or in P + 1 versus P + 3.
Six miRNAs were upregulated during the receptivity period in the natural cycle (LH + 7 vs. LH + 3; hsa-miR-30d-5p, hsa-miR-873, hsa-miR-345-5p, hsa-miR-30d-3p, hsa-miR-141-3p, and hsa-miR-30b-3p). When comparing the LH + 7 and LH + 3 cycle phases, the highest FC values were displayed by hsa-miR-30b-3p, hsa-miR-873, and hsa-miR-30d-5p (FC: +6.37, +4.39 and +3.88 respectively). Therefore, these 6 miRNAs allow a clear distinction between the prereceptive and receptive phases (Fig. 1C; (26)). In the HRT cycle comparing P + 5 versus P + 3, 17 miRNAs were differentially expressed, (Fig. 1D) (26), of which 3 were downregulated and 14 were upregulated (Fig. 1D). From this specific signature, the highest FCs during the window of implantation (WOI) were detected for hsa-miR-223-3p and hsa-miR-582-5p (FC: +77.28 and +6.22, respectively).
To validate this differential miRNA profile, RT-qPCRs assays were performed for both natural and HRT cycles. The validation of miRNAs enriched in the receptivity period are shown in (26). In natural cycles, an enrichment of hsa-miR-30d-5p, hsa-miR-345-5p, hsa-miR-873-3p, and hsa-miR-141-3p was identified in the LH + 7 versus LH + 3 comparison (26). Similarly, a significant overexpression (P < .05) of hsa-miR-223-3p and hsa-miR-582-5p was detected in HRT cycles at the receptive stage (26).
Finally, differential expression analysis was performed between samples from women who underwent a natural and an HRT cycle consecutively. No significantly differentially expressed miRNAs were discovered when the same phases of natural or substituted cycles were compared. However, significant differences could be appreciated for the comparison of LH + 0 versus P + 0, in which 5 miRNAs were differentially expressed (hsa-miR-3656, hsa-miR-4516, hsa-miR-3648, hsa-miR-3960, and hsa-miR-5100), and in the 1 performed for LH + 5 versus P + 3 where hsa-miR-3960 and hsa-miR-3648 were differentially present (26).
Analysis of the putative targets of the differentially expressed miRNAs in the secretory phase
Once the miRNAs with a significant prevalence in the secretory phase were identified, a search for their associated targets with an implication in the acquisition of a receptive endometrium was performed (26).
For this purpose, a preliminary screening by means of the bioinformatic tool miRTarBase was made selecting only targets previously reported as being involved in endometrial receptivity (27).
miRNA/mRNA interaction in the natural cycle
The significant miRNA signature and the respective associated target genes linked to a receptive status in a natural cycle are shown in Fig. 2a. Notably, this particular profile affects the expression of certain genes involved in progesterone production (MAPK8), polarization of the luminal epithelium (FOXO1), endometrial differentiation (HOXA11), cell adhesion (ITGB3), decidualization (RASSF2), and angiogenesis (KIF11), all of which are implicated in the essential processes of implantation and embryonic development.
Predicted targets of the differentially expressed miRNAs implicated in the acquisition of a receptive endometrium (A) miRNA signature linked to a receptive status in a natural cycle and their respective associated target genes. (B) miRNA signature linked to a receptive status in an HRT cycle and their respective associated target genes.
To validate some of the detected interactions and based on our previous experience, we focused on those associated with hsa-miR-30d (see (26)), given its role in the early crosstalk established between the endometrium and the peri-implantation embryo (11, 28). The Ishikawa cancer cell line was selected as a model as it allows the investigation of the normal endometrium by maintaining a normal hormonal responsiveness to ovarian steroids (29). Consequently, Ishikawa cells were transfected with both analogs (mimics) and inhibitors of hsa-miR-30d to evaluate their final in vitro downstream effect on the mRNA and protein levels of its respective putative targets (Fig. 3a). Scramble sequences coupled to a fluorophore (Block it) were used to evaluate the transfection efficiency. Also, they were used as negative controls to establish the basal effect of introducing random nontargeting short RNA sequences into the cells.
Validation of the miRNA/mRNA interactions predicted for a natural cycle. (A) Schematic representation of the experiment performed to validate the miRNA/mRNA interactions. (B) Time course experiment performed to determine the moment of maximum overexpression and/or downregulation of the hsa-miR-30d in the natural cycle (n = 3). 24 hours posttransfection shows the highest yield of miRNA recovery after the pre-treatment with an analogue of the hsa-miR-30d (n = 3) (Block it vs. Mimic FC: 0.98 ± 0.364 vs. 44.086 ± 35.003, P = .05). After 72 hours, the lowest levels of the miR-30d were obtained after the partial hsa-miR-30d blocking (C) FC values of the predicted targets of the hsa-miR-30d (KIF11, MAPK8, and HOXA11) after transfecting with either mimics or inhibitors (n = 3). Graphs represent mean ± SEM. Downregulation of the mRNAs were only significant after 48 hours in the case of KIF11 (C- vs. Mimic FC: 0.73 ± 0.18, P = .05; Block it vs. Mimic FC: 1.22 ± 0.12 vs. 0.73 ± 0.18, P = .02) and at 24 hours for MAPK8 (C- vs. Mimic FC: 0.77 ± 0.09, P = .002; Block it vs. Mimic FC: 1.01 ± 0.18 vs. 0.77 ± 0.09, P = .05). No significant alterations were detected for HOXA11. (D) Western blotting of the miR-30d predicted targets after performing the transfection with either mimics or inhibitors (n = 3). Left part of the panel shows the relative FC values calculated from every condition analysed. Significant changes could be observed for the following comparisons in the case of the MAPK8 after 48 hours (C- vs. Block it FC: 0.31 ± 0.03, P = 0.0014; C- vs. Inh FC: 0.62 ± 0.07, P = .0177) and 72 hours posttreatment (C- vs. Block it FC: 1.34 ± 0.04, P = .0079) and for KIF11 after 96 hours posttransfection (C- vs. Block it FC: 1.77 ± 0.15, P = .05).
After 24-hour posttransfection, the maximum overexpression of the miR-30d (mimic assay) was achieved (n = 3) as it can be observed in the comparison Block it versus Mimic shown in Fig. 3b. However, the greatest downregulation (inhibitory assay) could be noticed 72 hours posttransfection. Once the Knock-in (overexpression) and Knock-out (downregulation) phenotypes were reproducible, the expression levels of KIF11, MAPK8, and HOXA11 were analyzed by RT-qPCR (Fig. 3c) and western blotting (Fig. 3d).
Pretreatment with an analogue of hsa-miR-30d resulted in a slight reduction of the mRNA levels of KIF11 and MAPK8 for practically all the timepoints analyzed (Fig. 3c). However, this downregulation was only significant after 48 hours in the case of KIF11. In turn, no noticeable alterations were detected for HOXA11. Also, no significant increases were observed in target expression after performing the hsa-miR-30d blocking. By contrast, at a protein level, slight reductions could be observed for the following comparisons in the case of the MAPK8 after 48 hours (C- vs. Block it FC: 0.31 ± 0.03, P = .0014; C- vs. Inh FC: 0.62 ± 0.07, P = .0177) and 72 hours posttreatment (C- vs. Block it FC: 1.34 ± 0.04, P = .0079) and for KIF11 after 96h posttransfection (C- vs. Block it FC: 1.77 ± 0.15, P = .05) (Fig. 3d).
miRNA/mRNA interaction in the HRT cycle
An miRNA signature linked to the receptive status in an HRT cycle is presented in Fig. 2b and the respective associated target genes are presented in (26). Most of the predicted targets have a relevant implication in decidualization (FOXO1), epithelial to mesenchymal transcription (EMT) processes (RASSF2), as well as in the initial steps of embryonic adhesion (VCAM1, ITGB1, ITGB3, CDH1) and invasion (IGF2, MMP2, MMP9) along with the early nutrition of the conceptus (SLC7A1, SLC1A1).
Since the presence of an miRNA does not necessarily imply the downregulation of its putative targets, we took advantage of the expression pattern showed by the miRNAs (previously selected as differentially expressed in the receptive state) in the different phases of the cycle to compare to the behavior of the panel of genes included in the ERA test. This tool evaluates the expression of 248 genes related to endometrial receptivity, enabling determination of whether a specific patient requires a longer or shorter duration of progesterone administration to reach receptivity (12).
Our data (26) shows the correlations established between the expression of the receptivity-linked miRNAs and their predicted ERA target genes in the different phases of the HRT cycle. Diverse trends ranging from entirely parallel behaviors to opposite tendencies were observed. To simplify the validation analysis, we preselected miRNA/mRNA interactions presenting divergent patterns. In turn, the available information in the scientific literature regarding the role of miRNAs in endometrial receptivity regulation was also considered (26). As a result, the following miRNAs and their respective targets were preselected: hsa-miR-141-3p (RASSF2), hsa-miR-429 (RASSF2 and IGF2) and hsa-miR-324-5p (CKB, FOXO1, KCNG1, ATP5B, and TACC3).
Figure 4a shows the attainment of very pronounced inductions for all the miRNAs analyzed throughout the time course (n = 3). Given the magnitude and disparity of the FC obtained, these inductions were significant at 24 hours for all the miRNAs. Similarly, the strongest inhibitions were achieved after 72 hours posttreatment in all cases.
![Validation of the miRNA/mRNA interactions predicted for an HRT cycle. A) Time course experiment performed to determine the moment of maximum overexpression and/or downregulation of the hsa-miR-429, hsa-miR-141-3p and hsa-miR-324-5p in the HRT cycle (n = 3). Inductions were significant at 24 hours for all the miRNAs (hsa-miR-429: C- vs. Mimic FC: 2235.97 ±1390.84, P = 0.05 and Block it vs. Mimic FC: 0.83 ± 0.4 vs. 2235.97 ±1390.84, P = .05; hsa-miR-141-3p: C- vs. Mimic FC: 2085.32 ±710.83, P = .007 and Block it vs. Mimic FC: 0.67 ± 0.15 vs. 2085.32 ±710.83], P = .007; hsa-miR-324-5p: C- vs. Mimic FC: 2219.38 ± 1277.83, P = .04 and Block it vs. Mimic FC: 0.78 ± 0.26 vs. 2219.38 ± 1277.83, P = .04) and at 72 hours for the hsa-miR-141-3p (C- vs. Mimic FC: 161.46 ± 23.26, P = .002 and Block it vs. Mimic FC: 0.87 ± 0.45 vs. 161.46 ± 23.26, P = .004; and hsa-miR-324-5p (C- vs. Mimic FC: 246.47 ± 114.28, P = .02) and Block it vs. Mimic FC: 0.71 ± 0.29 vs. 246.47 ± 114.28, P = .02). The strongest inhibitions were achieved after 72 hours posttreatment in all cases (hsa-miR-429: C- vs. Inh. FC: 0.0015 ± 0.0007, P < .0001) and Block it vs. Inh. FC: 0.28 ± 0.37 vs. 0.0015 ± 0.0007, P = .37; hsa-miR-141-3p: C- vs. Inh. FC: 0.24 ± 0.09, P = .0001 and Block it vs. Inh. FC: 0.87 ± 0.45 vs. 0.0.24 ± 0.09, P = .05; hsa-miR-324-5p: C- vs. Inh. FC: 0.41 ± 0.11, P = .02 and Block it vs. Inh. FC: 0.71 ± 0.29 vs. 0..41 ± 0.11, P = .02). (B) FC values of the predicted targets of the hsa-miR-429 (RASSF2 and IGF2) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. Significant reduction was obtained for RASSF2 after 72 hours postinduction: C- vs. Mimic (FC: 0.75 ± 0.11, P = .004). Conversely, supplementation with has-miR-429 inhibitors caused a significant upregulation in the RASSF2 mRNA levels, specially 24 hours posttreatment treatment (C- vs. Inh. FC: 1.84 ± 0.51, P = .02 and Block it vs. Inh. FC: 0.97 ± 0.22 vs. 1.84 ± 0.51, P = .02). (C) FC values of the predicted targets of the hsa-miR-141-3p (RASSF2) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. (D) FC values of the predicted targets of the hsa-miR-324-5p (CKB, FOXO1, KCNG1, ATP5B, and TACC3) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. Significant downregulations were obtained for FOXO1 (C- vs. Mimic FC: 0.75 ± 0.08, P = .0009) and ATP5B (C- vs. Mimic FC: 0.84 ± 0.13, P = .05) after 72 hours postinduction. (E) Western blotting of the hsa-miR-324-5p predicted targets after performing the transfection with either mimics or inhibitors (n = 3). Left part of the panel shows the relative FC values calculated from every condition analysed. A significant suppression of FOXO1 could be observed 72 hours posttreatment coinciding with that observed at a mRNA level (C- vs. Mimic FC: 0.6 ± 0.2, P = .019).](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jcem/105/3/10.1210_clinem_dgz146/6/m_jcem_105_3_877_f4.jpeg?Expires=1747943697&Signature=5IN2xyrh5KK4bW7yoMyI82S~owSvLcckPkmS~WiFbeFcXHlo1Czb8ZLBKwgVzXNnRxYQAUQgLUZMFrZ3RvdPb2VTUPGaf62Ub-Vs-SCfrMN2WW0MueImIaoHlf4~5Rkqj5SfW5JifftyA2~J37h97ktGRafT3Pv3HlnfOtZAUSIffcMQCFqcXiFZDarFQw7~R1P5vhG-bHpqgD~H7-kbsGlE5L-LYJEf7J2S9ab5jbDeIe1~DEd-VPnGGFIs92ZnFk-E9aN8H41g9q-L4OKhIJdN425IAjCsQtj6fUVxkPKDtELevBIEICB0OphNJrJb0puCklW04J1ykh70e73m6Q__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Validation of the miRNA/mRNA interactions predicted for an HRT cycle. A) Time course experiment performed to determine the moment of maximum overexpression and/or downregulation of the hsa-miR-429, hsa-miR-141-3p and hsa-miR-324-5p in the HRT cycle (n = 3). Inductions were significant at 24 hours for all the miRNAs (hsa-miR-429: C- vs. Mimic FC: 2235.97 ±1390.84, P = 0.05 and Block it vs. Mimic FC: 0.83 ± 0.4 vs. 2235.97 ±1390.84, P = .05; hsa-miR-141-3p: C- vs. Mimic FC: 2085.32 ±710.83, P = .007 and Block it vs. Mimic FC: 0.67 ± 0.15 vs. 2085.32 ±710.83], P = .007; hsa-miR-324-5p: C- vs. Mimic FC: 2219.38 ± 1277.83, P = .04 and Block it vs. Mimic FC: 0.78 ± 0.26 vs. 2219.38 ± 1277.83, P = .04) and at 72 hours for the hsa-miR-141-3p (C- vs. Mimic FC: 161.46 ± 23.26, P = .002 and Block it vs. Mimic FC: 0.87 ± 0.45 vs. 161.46 ± 23.26, P = .004; and hsa-miR-324-5p (C- vs. Mimic FC: 246.47 ± 114.28, P = .02) and Block it vs. Mimic FC: 0.71 ± 0.29 vs. 246.47 ± 114.28, P = .02). The strongest inhibitions were achieved after 72 hours posttreatment in all cases (hsa-miR-429: C- vs. Inh. FC: 0.0015 ± 0.0007, P < .0001) and Block it vs. Inh. FC: 0.28 ± 0.37 vs. 0.0015 ± 0.0007, P = .37; hsa-miR-141-3p: C- vs. Inh. FC: 0.24 ± 0.09, P = .0001 and Block it vs. Inh. FC: 0.87 ± 0.45 vs. 0.0.24 ± 0.09, P = .05; hsa-miR-324-5p: C- vs. Inh. FC: 0.41 ± 0.11, P = .02 and Block it vs. Inh. FC: 0.71 ± 0.29 vs. 0..41 ± 0.11, P = .02). (B) FC values of the predicted targets of the hsa-miR-429 (RASSF2 and IGF2) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. Significant reduction was obtained for RASSF2 after 72 hours postinduction: C- vs. Mimic (FC: 0.75 ± 0.11, P = .004). Conversely, supplementation with has-miR-429 inhibitors caused a significant upregulation in the RASSF2 mRNA levels, specially 24 hours posttreatment treatment (C- vs. Inh. FC: 1.84 ± 0.51, P = .02 and Block it vs. Inh. FC: 0.97 ± 0.22 vs. 1.84 ± 0.51, P = .02). (C) FC values of the predicted targets of the hsa-miR-141-3p (RASSF2) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. (D) FC values of the predicted targets of the hsa-miR-324-5p (CKB, FOXO1, KCNG1, ATP5B, and TACC3) after transfecting with either mimics or inhibitors (n = 3). Bars represent mean ± SEM. Significant downregulations were obtained for FOXO1 (C- vs. Mimic FC: 0.75 ± 0.08, P = .0009) and ATP5B (C- vs. Mimic FC: 0.84 ± 0.13, P = .05) after 72 hours postinduction. (E) Western blotting of the hsa-miR-324-5p predicted targets after performing the transfection with either mimics or inhibitors (n = 3). Left part of the panel shows the relative FC values calculated from every condition analysed. A significant suppression of FOXO1 could be observed 72 hours posttreatment coinciding with that observed at a mRNA level (C- vs. Mimic FC: 0.6 ± 0.2, P = .019).
Expression of the different putative targets were analyzed through RT-qPCR and western blotting. From all the mRNAs analyzed, significant reductions of RASSF2 (Fig. 4b), FOXO1, and ATP5B (Fig. 4d) were obtained after 72 hours postinduction of their respective associated miRNAs. Interestingly, the downregulation of hsa-miR-429 promoted by the inhibitor’s supplementation allowed us to detect a significant upregulation in the RASSF2 mRNA levels, specially 24 hours posttreatment. However, no statistically significant results were obtained for the variation in RASSF2 expression when the miRNA was hsa-miR-141-3p (Fig. 4c). Finally, at a protein level, a significant suppression of FOXO1 could be observed 72 hours posttreatment coinciding with that observed at the mRNA level (Fig. 4e). Therefore, these results strengthen the existence of regulation between hsa-miR-429 and RASSF2 as well as between hsa-miR-324-5p with FOXO1 and/or ATP5B.
Discussion
ELB is a minimally invasive procedure for the study of human endometrial biology. We aimed to determine the miRNA profile of the secretory phase leading into endometrial receptivity using natural and medicated cycles among the same women. This allowed us to identify miRNA potential candidates to identify endometrial receptivity and to investigate differences in miRNA expression derived from two different protocols for the preparation of the endometrium intended for embryo transfer.
The number of miRNAs expressed per endometrial sample was similar to other body fluids (30–32) and hsa-miR-10a and hsa-miR-10b were the most abundantly expressed. Six differentially expressed miRNAs (hsa-miR-30b-3p, has-miR-873-3p, hsa-miR-30d-5p, hsa-miR-141-3p, hsa-miR-30d-3p, and hsa-miR-345-5p) presented a FC ≥2 and a FDR ≥0.05. Of these, hsa-mir-30d-5p and hsa-miR-30b-3p have established roles in the regulation of endometrial receptivity (19, 33). Hsa-mir-30d-5p is an inducer of adhesive molecule expression in trophectoderm cells during the crosstalk established at the initial stages of implantation (11). In turn, downregulation of miR-30b-3p in women undergoing assisted reproductive treatment is associated with poor prognosis (34).
Downregulation of hsa-miR-873-3p results in a higher probability of ectopic pregnancies (13). Hsa-miR-141-3p belongs to the miR-200 family, which are markedly downregulated in endometrial stromal cells after ovulation and prior to the prereceptive phase (35). MiR-141-3p downregulation facilitates the expression of PTEN, which in turn influences cell proliferation and apoptosis during decidualization (36,37). Finally, hsa-miR-345-5p favors embryo implantation (38) due to the protective function exerted in the luminal cell layer through the regulation of cell adhesion molecule pathways (39).
Concurrently, miRNA candidates related to endometrial receptivity were identified in the HRT cycle in the comparison P + 3 versus P + 5 (hsa-miR-598-3p, hsa-miR-500a-5p, hsa-miR-141-3p, hsa-miR-210-3p, hsa-miR-223-3p, hsa-miR-324-5p, hsa-miR-338-3p, hsa-miR-34b-5p, hsa-miR-378i, hsa-miR-429, hsa-miR-582-3p, hsa-miR-582-5p, hsa-miR-671-3p, hsa-miR-1271-5p, hsa-miR-7974, hsa-miR-34c-5p, and hsa-miR-326). Among these, we validated the expression of hsa-miR-223-3p and hsa-miR-582-5p. Interestingly, hsa-miR-223-3p has a putative binding site on the 3’untranslated regions of leukaemia inhibitory factor (LIF), an established marker of endometrial receptivity (40). Expression of LIF is increased during the secretory phase of the menstrual cycle and is detected in the first-trimester decidua (41). Hsa-miR-582-5p is significantly upregulated in the mid-secretory phase (17).
Once the differential signature of miRNA was obtained for each of the studied cycles, targets that could have an impact on the acquisition of a receptive endometrium were identified. In the case of the natural cycle, the miR-30 family is associated with genes involved in the regulation of the window of implantation (WOI). Functional analyses predicted a greater overexpression of crucial genes such as HOXA11, MAPK8, LIFR, KIF11, HBEGF, or ITGB3; ITGB3 plays important roles mediating the adhesion of trophoblast cells via the LIF pathway. In turn, HOXA11 deregulation results in defects in blastocyst attachment and implantation (42,43). Combined with FOXO1, it regulates prolactin expression in decidualized endometrial stromal cells (44). Likewise, miRNAs that are significantly expressed in the receptive phase in hormonal replacement therapies regulate genes included in what is currently known as the meta-signature of human endometrial receptivity (i.e., MMP2, MMP9, RASSF2, FOXO1, ITGB3, SLC7A1, SLC1A1, ARIDB5B, ITB1) (45). Interestingly, the in vitro models used for the analysis of the predicted miRNA/mRNA interactions suggest a direct regulation of KIF11 and MAPK8 by hsa-miR-30d, as well as a direct regulation between hsa-miR-429/RASSF2 and hsa-miR-324-5p with FOXO1 and ATP5B.
There is an ongoing controversy regarding whether to perform endometrial preparation for embryo transfer. Some studies support the idea that the natural cycle is associated with better outcomes compared to the use of HRT (46,47). Conversely, others find no significant differences in live birth rates between frozen embryo transfers in natural or HRT cycles (48). Likewise, no noticeable differences between the 2 options have been described in terms of pregnancy and miscarriage rates (49), implantation, pregnancy, or live birth rates per cycle or embryo transferred (50). However, HRT is often proposed as a better option since it allows doctors and patients to program the day of embryo transfer (51). This is not possible in the natural cycle due to the variability of follicular phases among patients and spontaneous ovulation (51,52).
This study is a starting point toward understanding the dynamic environment contributing to the preparation of the endometrium. Also, the similarities we found in the natural cycle and HRT from the same patients in equivalent phases (26) supports the similar endometrial status for embryo transfer typically observed in both types of cycles, at least at the miRNA profile. However, further studies are necessary to evaluate if similar miRNA profiles lead to equivalent states of endometrial receptivity and improve the diagnostic power and accuracy of miRNA detection to develop a minimally invasive test intended to predict the receptive status of the endometrium.
Acknowledgments
We thank Sheila M. Cherry, PhD, ELS, President and Senior Editor from Fresh Eyes Editing, LLC, for her work editing this manuscript.
Financial Support: This work was supported by Santiago Grísolía fellowship GRISOLIA/2014/002 to A. G; I. M. and J. J. were supported by a Torres-Quevedo grant (I. M. PTQ-13-06133; J. J. PTQ-14-06758) from the Spanish Ministry of Economy and Competitiveness, a MINECO/FEDER Grant SAF2015-67154-R and GFI 2013 to C. S; a Miguel Servet Program Type II of ISCIII [CPII18/00020]; and a FIS project [PI18/00957] to F. V.
Additional Information
Disclosure Summary: The authors have no competing financial interests.
Data Availability: All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.
References
Author notes
These authors contributed equally to this work.
