If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Endometrial transcriptome analysis indicates superiority of natural over artificial cycles in recurrent implantation failure patients undergoing frozen embryo transfer
Little consensus has been reached on the best protocol for endometrial preparation for frozen embryo transfer (FET). It is not known how, and to what extent, hormone supplementation in artificial cycles influences endometrial preparation for embryo implantation at a molecular level, especially in patients who have experienced recurrent implantation failure. Transcriptome analysis of 15 endometrial biopsy samples at the time of embryo implantation was used to compare two different endometrial preparation protocols, natural versus artificial cycles, for FET in women who have experienced recurrent implantation failure compared with fertile women. IPA and DAVID were used for functional analyses of differentially expressed genes. The TRANSFAC database was used to identify oestrogen and progesterone response elements upstream of differentially expressed genes. Cluster analysis demonstrated that natural cycles are associated with a better endometrial receptivity transcriptome than artificial cycles. Artificial cycles seemed to have a stronger negative effect on expression of genes and pathways crucial for endometrial receptivity, including ESR2, FSHR, LEP, and several interleukins and matrix metalloproteinases. Significant overrepresentation of oestrogen response elements among the genes with deteriorated expression in artificial cycles (P < 0.001) was found; progesterone response elements predominated in genes with amended expression with artificial cycles (P = 0.0052).
The current trend towards cryopreservation of all embryos after IVF, with transfer of a thawed embryo in a subsequent cycle now contributes more substantially than ever to the cumulative live birth rates after IVF treatment. Together with the evidence of improved endometrial receptivity in cycles without ovarian stimulation (
Evidence of impaired endometrial receptivity after ovarian stimulation for in vitro fertilization: a prospective randomized trial comparing fresh and frozen-thawed embryo transfer in normal responders.
). Furthermore, the better perinatal outcomes of preterm birth, low birth weight and being small for gestational age among singletons born after the transfer of frozen–thawed embryos compared with infants born after ovarian stimulation and IVF (
Impact of frozen-thawed single-blastocyst transfer on maternal and neonatal outcome: an analysis of 277,042 single-embryo transfer cycles from 2008 to 2010 in Japan.
Obstetric and perinatal outcomes in singleton pregnancies resulting from the transfer of frozen thawed versus fresh embryos generated through in vitro fertilization treatment: a systematic review and meta-analysis.
A crucial aspect of FET cycles is the timing and preparation of the endometrium to receive the transferred embryo(s). Protocols used in FET include natural cycle (NC-FET) and artificial cycle (AC-FET) with or without preceding pituitary down-regulation through GnRH agonist co-treatment (
A GnRH agonist and exogenous hormone stimulation protocol has a higher live-birth rate than a natural endogenous hormone protocol for frozen-thawed blastocyst-stage embryo transfer cycles: an analysis of 1391 cycles.
). Although the advantages of natural cycles FET (NC-FET) such as less medication and cheaper price somewhat counterbalance the need for more frequent ultrasonographic evaluation of the dominant follicle, the risk of unexpected ovulation and insufficient endometrial development in these cycles (
). So far, however, no clear data support one endometrial preparation method over another. In a recent systematic review and meta-analysis, it was concluded that there are no differences in the clinical pregnancy rate, ongoing pregnancy rate or live birth rate in connection with the different methods of endometrial preparation before FET (
). Furthermore, despite active investigation of the endometrial transcriptome that clearly demonstrates an unfavourable endometrial gene expression profile during hormonal stimulation in ovarian stimulation (
), there is no knowledge of how and to what extent hormonally supplemented cycles influence endometrial preparation for embryo implantation at a molecular level.
In the present study, endometrial gene expression profiles in infertile women undergoing two different endometrial preparation protocols (AC-FET and NC-FET) was compared with fertile women in a natural cycle. Our study group of women who had experienced recurrent implantation failure (RIF), also diagnosed as unexplained infertility, is especially intriguing as we have demonstrated altered endometrial receptivity in these women (
). Therefore, we aimed to clarify whether AC-FET with oestrogen and progesterone improves endometrial maturation in our study group. To answer this question, we aimed to identify two groups of genes with opposite transcriptional behaviour in patients who have experienced RIF. First, we were interested in the genes, which show abnormal expression in the natural cycle, but are amended in artificial cycles and, second, the opposite case of genes showing normal expression in natural cycles, with deteriorated gene-activity after administration of steroid hormones in artificial cycles. By analysing the genes in both categories, the aim was to arrive at conclusions about the pros and cons of using artificial cycles in FET.
Endometrial tissue is one of a few tissues that are overwhelmingly controlled by steroid hormones, oestradiol and progesterone. These steroid hormones act as transcriptional regulators via ligand-bound receptor complexes interacting with the DNA consensus sequences in target genes, referred to as hormone response elements (HRE). Although major progress has been made in deciphering the HRE for oestrogen response elements and progesterone response elements, little is known about their roles in embryo implantation. Additionally we set out to analyse the oestrogen response elements and progesterone response element sequences −50 kb upstream of genes that were differentially expressed after artificial cycles or related to infertility.
The aforementioned aims are critical in obtaining a better understanding of the mechanisms of steroid hormone involvement in endometrial maturation, and in the long term this knowledge should help to devise better hormonal regimens for FET, even for patients with the complication of RIF.
Materials and methods
Study design and endometrial biopsy sample collection
In total fifteen endometrial biopsy samples were obtained from women with unexplained RIF in NC-FET (n = 5), women with unexplained RIF in AC-FET (n = 5), and from healthy women with proven fertility in natural cycles (NC-FC) (n = 5). The characteristics of the women are presented in Table 1. Two infertile women provided biopsy samples for both NC-FET and AC-FET.
Table 1Characteristics of the study groups.
RIF, NC-FET (n = 5)
RIF, AC-FET (n = 5)
NC-FC (n = 5)
Age (years)
30.2 ± 4.3
32.4 ± 5.0
31.8 ± 3.8
Body mass index (kg/m2)
20.7 ± 1.8
20.9 ± 2.0
23.5 ± 2.1
Cycle length (days)
28.2 ± 1.7
28.0 ± 1.4
28.4 ± 0.7
Menses duration (days)
4.4 ± 0.7
4.5 ± 0.7
4.0 ± 0.2
Previous implantation failures
3.4 ± 0.9
3.4 ± 0.6
0
Parity
0
0
1.5 ± 0.2
Endometrial thickness (mm)
9.3 ± 0.9
8.5 ± 1.7
n.a.
Biopsy sample taken
LH+7
Prog+6
LH+7
Results are mean ± SD.
AC-FET, artificial cycle-frozen embryo transfer; LH+, day since the luteinizing hormone (LH) surge; na, not assessed; NC-FC, natural cycle-fertile control; NC-FET, natural cycle-frozen embryo transfer; Prog+, progesterone administration in days; RIF, recurrent implantation failure.
The fertile control women were candidates for oocyte donation, and were recruited at the Instituto Valenciano de Infertilidad, Valencia, Spain. All women signed an informed consent document approved by the local Ethics Committee (15 May 2007).
The patient group of women with unexplained RIF attended the Department of Obstetrics and Gynaecology, Karolinska University Hospital Huddinge, Sweden. The Ethics Committee of Karolinska Institutet approved the study (10 March 2003, number 78/03) and signed informed consent was obtained from every patient. Unexplained RIF was diagnosed by a set of tests that included normal concentrations of thyroid-stimulating hormone (0.4–4.7 mU/L), prolactin (3–27 mg/L), oestradiol (follicular phase <600 pmol/L), LH (follicular phase 1.8–12 U/L), FSH (follicular phase 2.5–10 U/L) and progesterone (luteal phase >17 nmol/L). The infertile women had patent fallopian tubes as determined by hysterosalpingosonography, a normal mid-secretory endometrial thickness of 9.3 ± 0.5 mm (mean ± SD), and their partners had normal semen analysis results according to World Health Orgamization criteria (
One-half of the group of women who had experienced RIF underwent artificial endometrial stimulation for FET. These women received oestradiol valerate (Progynon®; Schering Nordiska, Berlin, Germany) orally, 6 mg from cycle day 1, 2 mg in the morning and 4 mg in the evening. The women were instructed to take their oestrogen tablets every 12 h, at 8 o'clock in the morning and at 8 o'clock in the evening ± 1 h. Micronized progesterone (APL, Stockholm, Sweden) was administered vaginally; 400 mg twice a day from the time the endometrium had reached a thickness of at least 7 mm, and absence of follicles with a diameter of 15 mm or more. Oral oestrogen support was simultaneously reduced to 2 mg twice a day. The embryos were not thawed and transferred in these treatments, and therefore, we cannot compare the pregnancy rates after using AC-FET or NC-FET.
Endometrial biopsy sampling
All samples were obtained during the mid-secretory phase, at the time of the “window of implantation” (WOI) (LH+7) for NC-FET patients and fertile women, and in AC-FET women on the 6th day of progesterone supplementation (Prog+6). The women with natural, non-stimulated cycles were not subject to any ovarian stimulation before entering the study. Endometrial biopsy samples were obtained from the anterior wall of the uterine cavity, without dilatation of the cervix, using a Pipelle catheter (Gynetics, Hamont-Achel, Belgium). The LH surge was detected in morning urine (Donacheck ovulación, Novalab Ibérica, S.A.L, Coslada, Madrid, Spain and Clearplan, Unipath Ltd., Bedford, UK).
The biopsy samples were prepared for histology, scanning electron microscopy (as described by
For microarray and real-time polymerase chain reaction analyses, total RNA was extracted from the endometrial biopsy samples by using an RNeasy Mini-kit (Qiagen, Venlo, the Netherlands) or the “Trizol method” according to the protocol provided by the manufacturer (Life Technologies, Inc., Gaithersburg, MD, USA). RNA quality was assessed by loading 300 ng of total RNA onto an RNA LabChip and this was followed by analysis in a 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). An RNA integrity value of over 8 was considered acceptable. Hybridization to the Whole Genome Oligo Microarray that comprises 44,000 gene targets (Agilent Technologies) was carried out as described previously (
GenePix Pro 6.0 software (Molecular Devices, Sunnyvale, CA, USA) was used for microarray image analysis and calculation of spot intensities. Replicates of the gene transcripts were merged using the average of the hybridization values. The data were normalized by the mean of the spot subtracted by the median of the background. Further, densitometry values between arrays were normalized using the quantile normalization function to remove possible sources of variation of a non-biological origin between the arrays.
), sample size calculations were made. The acceptable amount of false positives was considered to be 0.05% and desired power was set at 90%. We had three study groups with sample size five in each group. For three or more group comparisons, seven samples or over in each group have been suggested (
). An acceptable level of standard deviation in array studies is considered to be up to 0.7 (bioinformatics.mdanderson.org/MicroarraySampleSize/), but in order to strengthen the power in our study we used a standard deviation of 0.4.
Differential gene expression
R-statistical software (Free Software Foundation, Boston, USA) was used for data anlysis. Gene expression profiles were determined by comparing groups: NC-FET versus NC-FC; AC-FET versus NC-FC; and AC-FET versus NC-FET. Non-parametric tests (2 × 2 comparisons) were used. We set two criteria to define the genes that had altered mRNA abundance among the different sample sets: an absolute fold change of three or more and a proportion of false positives of less than 0.05.
Statistically significant differences between study groups were identified using the rank product non-parametric test in the Bioconductor RankProd package (Bioconductor, www.bioconductor.org). Because of the limited amount of samples, a non-parametric statistical test was conducted as a rough filter to narrow down the list of most relevant genes. In addition, we applied the rank product approach that includes a multiple hypothesis test for raw P-value correction to ascertain a false positive rate. A proportion of false postives of less than 0.05 was considered statistically significant.
Sample clustering and principal component analysis
To validate the results of a non-parametric method of analysing differentially expressed genes (DEGs), principal component analysis and hierarchical clustering were carried out using MeV 4.2.02 software (www.tm4.org) (
). Principal component analysis projects high dimensional data into a lower dimensional span, where samples with a similar gene expression level tend to cluster together on a plot. A three-dimensional scatter plot was produced for visualizing differences between sample sets based on each sample's gene expression profile. In hierarchical clustering, the data were Z-normalized by gene, the Euclidean distance was selected as the similarity measure to cluster expression profiles, and linkage was conducted with a complete-linkage hierarchical clustering algorithm method.
Functional analysis of the results
Functional analysis of differentially regulated genes was explored by using the Database for Annotation, Visualization and Integrated Discovery (DAVID, v. 6.7) (
), and Ingenuity Pathways Analysis (IPA) (Ingenuity® Systems, www.ingenuity.com). DAVID searches blocks of functionally related genes according to different criteria such as the Gene Ontology (GO) terms for biological processes, cellular locations and molecular functions. We used GO FAT search that filters the broadest terms so that they do not overshadow the more specific terms (david.abcc.ncifcrf.gov). IPA was applied for canonical pathway and molecular network analyses. A False Discovery Rate of less than 5.0 and a P-value of less tha 0.05 were considered statistically significant.
Prediction of oestrogen response elements and progesterone response elements among differentially expressed genes
We used position weight matrices (PWM), which are widely used in computational molecular biology, in order to depict the DNA binding preferences of transcription factors (including steroid hormone receptors) in target genes. An in silico motif search for steroid hormone response elements, e.g. for oestrogen and progesterone response element sequences in proximal promoter region upstream of DEGs in the endometrium was conducted using the TRANSFAC professional database (
). All promoter sequences of DEGs in the present study were extracted from the Ensembl BioMart database (Homo sapiens, GRCh38.p2). For each gene, three upstream sequences were analysed relative to the major transcription start site, as following: from −1000 bp to +150 bp; from −10,000 bp to +150 bp; and from −50,000 bp to +150 bp. We analysed 17 different PWM from the TRANSFAC database in the oestrogen and progesterone element motif search: V$ER_Q6, V$ER_Q6_02, V$ERALPHA_01, V$ESR1_01, V$ERALPHA_Q6_01, V$ERALPHA_Q6_02, V$ESR1_03, V$ESR1_04, V$ESR1_05, V$ERALPHA_Q4, V$PR_01, V$PR_02, V$PR_Q2, and V$PR_Q6. As the progesterone response element is similar to the glucocorticoid response element, PWM for glucocorticoid response element were also included in the analysis: GR V$GR_01, V$GRE_C and V$GR_Q6.
Right-tailed Wilcoxon rank sum test was used in oestrogen and progesterone response element profile analysis (occurrences of respective PWM were summed) to compare the differentially expressed endometrial genes with other genes in the human genome (n = 20,710). To compare the studied subgroups (artificial cycle deteriorated, artificial cycle improved, RIF specific) Kruskal–Wallis test was used. In case of significant difference, Tukey test was used for post-hoc pairwise comparison. A standard hypergeometric test was used to assess the motif enrichment for each PWM separately. Fisher's exact test analysis was used to compare the observed proportion of occurrence of a particular PWM in the studied groups. In cases of significant difference, post-hoc analysis was applied to study the significance of pairwise differences between subgroups. To counteract the problem of multiple comparisons, Bonferroni correction was applied.
Microarray validation
Total RNA (300 ng) was reverse-transcribed using Advantage RT-for-PCR kits (Clontech, Palo Alto, CA, USA) for all samples following the protocol as described in detail previously (
). The genes HABP2, HLA-DOB, SPDEF and TRH were selected for microarray validation. Forward and reverse primer sequences for each gene were (5′–3′): AGGAAGAGAACACCAGTAGCA and TAGTAGGGAGGACTCTGGGTA for HABP2, AGGGCTCAGAAAGGATATGTGA and CTCAGAACACAGAGCTCCAGA for HLA-DOB, GTCCCGCCATGAACTACGA and CTGGAAGGTCAGAGCAGCA for SPDEF, TGGTTGCTGCTCGCTCTGGC and TCTGGGACGCGGAGTGCTCA for TRH, and CCCATCACCATCTTCCAGGA and CATCGCCCCACTTGATTTTG for the control GAPDH gene. The real-time polymerase chain reaction protocol has been described in detail elsewhere (
). Analysis of gene expression differences between the study groups was carried out by using the Mann–Whitney U-test; P < 0.05 was considered statistically significant.
Results
Evaluation of endometrial biopsies
Histological evaluation of the natural-cycle samples showed normal maturation in relation to the cycle day, according to
), in order to assess endometrial receptivity alterations among this patient group. In both infertile groups, pinopode formation differed from that in fertile women, thereby supporting the notion of aberrant endometrial receptivity among these women: in the NC-FET group, three women had no pinopodes, one had very few, and one woman presented normal formation of pinopodes; while in the AC-FET group three women had no pinopodes and two presented scarce pinopodes on the endometrial surface.
Cluster analysis of microarray data
Principal component analysis demonstrated a clear distinction in the endometrial gene expression patterns in the NC-FET, AC-FET and NC-FC groups (Supplementary Figure S1). Hierarchical clustering was then applied to the microarray data, and a similar pattern was observed: the study groups were in three clusters, where the NC-FET and NC-FC groups clustered closer than the AC-FET group (Figure 1). This clustering pattern indicates that, on the basis of whole endometrial transcriptome analysis at the time of embryo implantation, the NC-FET protocol yields a more similar endometrial pattern to fertile controls than the AC-FET protocol among women who have experienced RIF.
Figure 1Cluster analysis of Z-scored gene expression values in the endometrium at the window of implantation in healthy women with proven fertility in natural cycles (NC-FC), in infertile women undergoing a natural cyle for frozen embryo transfer (NC-FET), and infertile women undergoing an artificial cycle for frozen embryo transfer (AC-FET). Red represents genes with a positive Z score and green, genes with negative Z score.
Differential gene expression analysis between NC-FET and AC-FET groups in women who have experienced RIF, and NC-FC women
Our primary microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-3713. In total we identified: 443 up-regulated and 446 down-regulated genes (three or more fold change; proportion of false positives <0.05) in infertile NC-FET versus fertile NC-FC women; 575 up-regulated and 335 down-regulated genes in infertile AC-FET compared with fertile NC-FC women; and 502 up-regulated and 201 down-regulated genes in infertile AC-FET compared with infertile NC-FET women.
These comparisons gave the means to focus on DEGs whose expression level improved (was more similar to fertile controls) with artificial cycles (n = 620) (i.e. DEGs unique for natural cycle RIF compared with NC-FC comparison: meaning that the initial gene expression profile in NC-RIF women differed from that in fertile controls, whereas these DEGs in the AC-RIF group versus NC-FC comparison demonstrated similar values to fertile controls); DEGs whose expression profile deteriorated (differed three or more fold compared with fertile controls) with artificial cycles (n = 640) (i.e. DEGs unique for AC-RIF versus NC-FC comparison: meaning that in the AC-RIF versus natural NC-FC comparison the gene expression profile was different, whereas the comparison of gene expression profiles in natural NC-RIF versus NC-FC demonstrated similar values) and genes specific to RIF (n = 269), i.e. DEGs whose expression was significantly different in infertile women in both natural cycle and artificial cycle when compared with fertile controls (see Figure 2 for description of DEGs groups, and Supplementary Table S1 for gene lists).
Figure 2The three differentially expressed gene (DEG) groups analysed: DEGs whose expression improved with artificial cycle in infertile women (n = 620) (i.e. DEGs unique for women who have experienced recurrent implantation failure undergoing a natural cycle (NC-RIF) versus healthy women with proven fertility in natural cycles (NC-FC) comparison: meaning that the initial gene expression profile in NC-RIF women differed from that in fertile controls, whereas these DEGs in the AC-RIF (women who have experienced recurrent implantation failure undergoing an artificial cycle) group versus NC-FC comparison demonstrated similar values to fertile controls); DEGs whose expression profile deteriorated with artificial endometrium preparation in infertile women (n = 640) (i.e. DEGs unique for AC-RIF versus NC-FC comparison: meaning that in the AC-RIF versus natural NC-FC comparison the gene expression profile was different, whereas the comparison of gene expression profiles in natural NC-RIF versus NC-FC demonstrated similar values); and genes specific to RIF, n = 269 (i.e. DEGs whose expression was significantly different in infertile women in both natural cycle and artificial cycle when compared with fertile controls).
Functional profiling of differentially expressed genes that improved in artificial cycle in patients who have experienced RIF at the window of implantation or WOI
Biological functional analysis of DEGs that improved their expression pattern with artificial cycle in gene ontology terms showed that they were involved in various biological processes such as the G-protein-coupled receptor signalling pathway, defense response, potassium ion transport, cell surface receptor-mediated signal transduction, cell adhesion and the immune response; in cellular components of the extracellular matrix and region; and several molecular functions such as cytokine activity, G-protein-coupled receptor activity, ion channel activity, transmembrane transporter activity, carbohydrate binding and oxidoreductase activity (Figure 3A and Supplementary Table S2).
Figure 3Functional enrichment analysis of the endometrial transcriptome at the window of implantation: genes – whose expression improved with artificial cycles (A), genes whose expression deteriorated with artificial cycles (B), and genes specific to recurrent implantation failure (C). Bar colour denotes different types of evidence from gene ontology (DAVID) and pathway (IPA) databases: BP, biological process (in orange); CC, cellular component (in red); CP, canonical pathways (in blue); MF, molecular function (in green). The X-axis denotes functional enrichment score, computed as −log2 of related P-values. Some abbreviations in the Y-axis are indicated in full text in Table S2, Table S4.
When investigating different biological pathways using IPA, we found that artificial endometrial stimulation in women with RIF influenced canonical pathways involved in G-protein-coupled receptor and cyclic adenosine monophosphate-mediated signalling, lipid signalling and defense responses (Figure 3A). Another analysis using DAVID indicated pathways involved in neuroactive ligand-receptor interactions (19 genes, P < 0.001), and cytokine–cytokine receptor interactions (16 genes, P = 0.023).
Analysis of the molecular relationships between DEGs that improved their expression patterns in artificial cycles (using the Ingenuity Pathways Knowledge Base) revealed one enriched gene network with high score (IPA score of 37) that united molecules involved in the inflammatory response, antigen presentation, cell-to-cell signalling and interaction where genes IL31, IL21, MMP9, MUC5AC/MUC5B, POSTN and WISP1 intertwined (Supplementary Figure S2A).
To highlight, the DEGs that improved in artificial cycles that repeatedly were identified in different functional analyses using DAVID and IPA included APOA1, AQP6, CD34, CEACAM1, CEACAM8, COMP, EDN3, HBA1/HBA2, IGFBP1, IL21, IL31, ITGA8, IFNB1, LHCGR, MMP9, MUC5B, POSTN, S100A2, SELP, SORD, TYR, WISP1, WNT16, WNT3A, WNT8B and various chemokines, collagens, G protein-coupled receptors, immunoglobulins, interferons and olfactory receptors.
Functional profiling of DEGs that deteriorated in artificl cycles in RIF patients at the implantation or WOI
Functional analyses of the DEGs that deteriorated with artificial cycles in patients who have experienced RIF showed that they were involved in various biological processes such as detection of stimuli and neurological system processes; cellular components of the extracellular region; and molecular functions such as serine-type endopeptidase activity and calcium ion binding (Figure 3B and Supplementary Table S3).
The canonical pathways that were influenced negatively by artificial cycles were shown by IPA to be calcium signalling, linoleic acid, arachidonic acid and eicosanoid metabolism and signalling and macrophage migration inhibitory factor activity (Figure 3B). DAVID analysis of the pathways that were negatively influenced by artificial cycles involved neuroactive ligand–receptor interactions (17 genes, P = 0.005), and similarly to IPA showed the involvement of linoleic acid metabolism (five genes, P = 0.05).
Molecular interaction analysis of DEGs whose expression profile altered with artificial cycles revealed one enriched network with high score (IPA score of 38) that united molecules involved in cellular movement, immune cell trafficking, cellular growth and proliferation, where genes CD4, CD22, ESR2, IL4, IL29, LEP, MMP3 and SELE interacted (Supplementary Figure S2B).
To sum up, top genes identified in all functional analyses among DEGs that were dysregulated in artificial cycles included APOA2, APOC4, CALCRL, ESR2, fibroblast growth factors FGF17, FGF8, FGFBP2; FSHR, INHBC, interleukins IL1F6, IL27, IL29, IL4, IL9R; LEP; matrix metallopeptidases MMP17, MMP27, MMP3; NEUROG1, PPARD, PTGER3, SELE, WNT8A, and kallikrein-related peptidases, olfactory receptors, phospholipases, and pregnancy-specific beta-1-glycoproteins.
Functional profiling of differentially expressed genes specific to recurrent implantation failure at the window of implantation or WOI
Biological function analysis of DEGs specific to RIF in gene ontology terms indicated involvement in various biological processes such as cell surface (G-protein coupled) receptor-linked signal transduction (Figure 3C). A significant proportion of these RIF-specific genes were located in extracellular region or plasma membrane. Regarding molecular functions, the genes were mostly involved in hormone activity. IPA analysis revealed that RIF-specific DEGs were involved in several pathways such as G-protein-coupled receptor and cyclic adenosine monophosphate-mediated signalling, metabolism of xenobiotics by cytochrome P450, corticotropin-releasing hormone signaling and sphingosine-1-phosphate signalling (Figure 3C and Supplementary Table S4).
Analysis of the molecular relationships among RIF-specific DEGs revealed one enriched gene network with high score (IPA score of 37) that united genes involved in cellular movement, cell death and lipid metabolism, where molecules ADRB2, CASP8, INHBA, LTF, MUC4, MUC5A/MUC5B, SERPINB3, TFF3, and WISP2 intertwined (Supplementary Figure S2C).
To sum up, top molecules identified in all functional analyses among genes specific to unexplained RIF included ADRB2, ADM2, CRH, CYP3A4, DRD3, HABP2, HLA-DOB, [email protected], INHBA, IL28B, ITGA10, LTF, MMP8, MUC4, SPDEF, TRH, TFF3, WISP2, G protein-coupled receptors, olfactory receptors and solute carrier family members.
Comparison of microarray data with the endometrial receptivity array gene list
A novel diagnostic transcriptomic tool concerning endometrial receptivity, the endometrial receptivity array (ERA) test, has been previously presented, where 238 genes serve as an endometrial receptivity biomarker cluster (
). This test has proved to be accurate and consistent, enabling detection of personalized timing of the window of implantation among various patient groups (
). To investigate these endometrial receptivity markers in the current study, the lists of DEGs with the ERA gene list were compared. Improved DEGs with artificial cycles shared 11 common genes with the ERA list: CALB2, COL16A1, COMP, EDN3, IGFBP1, LRRC17, OLFM4, POSTN, SLC15A1, SORD, and TMEM16A (Table 2). Deteriorated DEGs with artificial cycles shared four common genes with the ERA test: CRISP3, C14orf161, HAL and HPSE. DEGs specific to RIF shared four genes with the ERA list: HABP2, HLA-DOB, SPDEF and TRH (Table 2).
Table 2Comparison of array data with the endometrial receptivity array (ERA) gene list.
Genes whose expression improved with artificial cycles
ERA test values are fold changes obtained from comparisons of receptive vs. pre-receptive endometrium (Diaz-Gimeno et al., 2011).
NC-FET versus NC-FC
AC-FET vsersus NC-FC
HREs
HABP2
4.09
−4.48
−5.48
ERE/PRE/GRE
HLA-DOB
−11.06
5.17
12.86
ERE/PRE/GRE
SPDEF
−3.78
9.27
12.20
ERE/PRE/GRE
TRH
−21.69
7.56
9.02
ERE/PRE/GRE
AC-FET, artificial cycle-frozen embryo transfer; ERE, oestrogen response element; HRE, hormone response element; GRE, glucocorticoid response element; NC-FC, natural cycle-fertile control; NC-FET, natural cycle-frozen embryo transfer; nd, non-detectable, meaning that there was no gene expression difference between the patients who have experienced RIF and fertile controls; PRE, progesterone response element; RIF, recurrent implantation failure.
a ERA test values are fold changes obtained from comparisons of receptive vs. pre-receptive endometrium (
Analysis of hormone response elements in differentially expressed genes
In hormone response elements in silico analysis, we focused on three promoter regions (−1000 bp to +150 bp, −10,000 bp to +150 bp, and −50,000 bp to +150 bp from transcription start site) of DEGs and searched for oestrogen, progesterone and glucocorticoid response elements. Of 1529 DEGs identified via microarrays, 1273 were eligible for TRANSFAC analysis: 534 DEGs that improved with artificial cyces, 508 DEGs that deteriorated with artificial cycles, and 231 DEGs specific to RIF.
Importantly, all DEGs that were eligible for TRANSFAC analysis had at least one HRE in their promoter regions. If all oestrogen response elements (10 PWM) and progesterone/glucocorticoid response elements (seven PMW) matrices were summarized and respective HRE profiles created, progesterone/glucocorticoid response element motifs showed significantly higher prevalence (P < 2.2e−16 from −50,000 bp to +150 bp) in promoter regions of DEGs compared with other genes in the human genome. Progesterone/glucocorticoid response element motifs showed higher frequency compared with oestrogen response elements in all promoter regions of DEGs identified in this study (Figure 4). When different subgroups were compared, significantly lower frequency (P = 0.0052) of progesterone/glucocorticoid response element sites were determined in most distal promoter region (−50,000 bp to +150 bp) of DEGs deteriorated compared with genes whose expression was improved after artificial cycles. In contrast, deteoriated genes from artificial cycles had significantly higher frequency of oestrogen response elements compared with improved genes from artificial cycles (P = 3.2e−06) and RIF-specific genes (P = 0.0045) (Figure 4).
Figure 4Tukey test for pairwise comparison of oestrogen receptor elements and progesterone receptor element profiles in promoter regions (−50,000 bp to +150 bp) of differentially expressed genes (DEGs) with artificial cycle deteriorated (green), artificial cycle improved (yellow) and recurrent implantation failure (RIF) specific (red). Box plot showing median and mean value. On the left side significant difference of the presence of oestrogen receptor element sites upstream of DEGs artificial cycle deteriorated compared with artificial cycle improved (P = 3.2e−06) and compared with RIF-specific genes (=0.0045) is shown. On the right significant difference of the presence of progesterone response element sites upstream of DEGs artificial cycle deteriorated compared with artificial cycle improved (P = 0.0052) is shown. AC, artificial cycle; ERE, oestrogen response element; HRE, hormone response element; GRE, glucocorticoid response element; PRE, progesterone response element; RIF, recurrent implantation failure.
When each PWM (17 oestrogen, progesterone and glucocorticoid response element motifs) was analysed separately, two hormone response elements motifs (PWM PR_Q6 and ESR1_01) demonstrated significant differences in promoter areas between the study groups (Figure 5). Progesterone response element motif (V$PR_Q6) was present at a significantly higher rate upstream of genes specific to RIF (P = 0.044) in the most proximal promoter region (−1,000 bp to +150 bp). For instance, progesterone response element motif was found upstream of genes as AFM, BRINP3, CNNM1, FAM151A, and IL12RB2. A significantly higher prevalence of this progesterone response elemet motif was also observed in the farther (−10,000 to +150 bp) promoter region (P = 0.033) of RIF-specific genes. In addition, DEGs whose expression improved after artificial cycle also showed a high prevalence of the same progesterone response element in the farther promoter region (−10,000 bp to +150 bp) (P = 0.040) (Supplementary Table S5). Genes whose expression improved after artificial cycles and consisted progesterone response element in promoter region were for example ADAD1, CALML5, FAM196A, IFNA5, and IL21.
Figure 5Two most prevalent hormone response elements found in promoter regions among differentially expressed genes identified in this study. Letters abbreviate the nucleotides (A, C, G, and T) in the images and are sized according to their relative occurrence. (A) ESR1 element (V$ESR1_01) was over represented upstream of differentially expressed genes (DEGs) deteriorated with artificial cycles (examples IL9R, MMP17, PTGER3, ESR2, and GATA3). This oestrogen receptor element motif in the most proximal promoter region (from −1,000 bp to +150 bp from transcription start sites) turned out to be significantly overrepresented in upstream of DEGs that deteriorated with artificial cycle compared with DEGs that improved with artificial cycle (P = 0.023); (B) progesterone response element (V$PR_Q6) was statistically more frequent in upstream of genes related to recurrent implantation failure (examples AFM, BRINP3, CNNM1, FAM151A, and IL12RB2) (P = 0.04) and among DEGs that improved with artificial cycle (examples ADAD1, CALML5, FAM196A, IFNA5, and IL21) (P = 0.04).
In contrast, DEGs that deteriorated in response to artificial cyces more frequently had oestrogen response element motif (V$ESR1_01) in their proximal promoters, but not at a significant level after applying Bonferroni correction (Supplementary Table S5). When comparing the relative frequency of hormone response elements between groups, the named oesrogen response element motif was significantly overrepresented upstream of DEGs that deteriorated with artificial cycles compared with DEGs that improved with artificial cycles (adjusted P = 0.023) in the most proximal promoter region (−1000 bp to +150 bp). For example, genes like IL9R, MMP17, PTGER3, ESR2, and GATA3 all included oestrogen response element motif in their promoter regions and their expression was deteriorated with artificial cycles. All 17 motifs used in in silico analysis and their occurrence in promoter regions of genes among DEGs are shown in Supplementary Table S5.
Comparison of differentially expressed genes with genes that responded to oestradiol and progesterone in the Ishikawa cell line
In studies with human endometrial biopsy samples, it is difficult to analyse retrospectively whether changes in the tissue are primary or secondary results of steroid hormone action. Therefore, endometrial gene expression after artificial cycles were compared with that in the recent study of ours where the oestradiol and progesterone responsive transcriptome was analysed in Ishikawa cancer cell line (
Changes in the transcriptome of the human endometrial Ishikawa cancer cell line induced by estrogen, progesterone, tamoxifen, and mifepristone (RU486) as detected by RNA-sequencing.
). The endometrial epithelial Ishikawa cell line expresses functional response for oestradiol and progesterone, and is therefore a good in-vitro model for studying the responses of the endometrial epithelium to oestradiol and progesterone (
Characterization of the functional progesterone receptor in an endometrial adenocarcinoma cell line (Ishikawa): progesterone-induced expression of the alpha1 integrin.
). Although the expression platforms of our current and previous studies were different (microarray versus RNA-sequencing), we found 50 genes for which expression was significantly changed after hormonal treatments in both studies (Figure 6). We identified 37 genes with significant regulation in response to oestradiol treatment and 42 genes were changed after progesterone treatment, where 29 genes were influenced by both hormonal treatments in our previous study. Among the DEGs that improved with artificial cycles, 18 were regulated by oestradiol and progeserone (75%), one by oestradiol (4%) and five genes solely by progesterone (21%). DEGs that were negatively influenced with artificial cycles responded more to oestradiol regulation, where nine genes were regulated by oestradiol and progesterone (47%), six genes solely by oestradiol (32%) and four genes by progesterone (21%) (Figure 6). In addition, 58% of the DEGs that improved with artificial cycles shared the same expression direction with the cell line experiments, whereas 58% of DEGs that deteriorated with artificial cycles demonstrated opposite expression direction with the Ishikawa cells (Figure 6).
Figure 6Genes regulated by oestradiol and progesterone in endometrium and Ishikawa cells. Fifty differentially expressed genes (DEGs) from the present study were also regulated by oestradiol and progesterone in our previous study using Ishikawa cells (
Changes in the transcriptome of the human endometrial Ishikawa cancer cell line induced by estrogen, progesterone, tamoxifen, and mifepristone (RU486) as detected by RNA-sequencing.
). Red denotes up-regulation and green down-regulation of gene expression in Ishikawa cells. The dotted line indicates zero and the variable line shows the expression change in response to hormone treatment (E2) – gene expression regulated by oestradiol; (P4) gene regulated by progesterone, and (E2, P4) – gene regulated by both oestradiol and progesterone. *Designates genes in the present study with an opposite expression direction versus Ishikawa cells. Red on the y-axis indicates genes whose expression improved with artificial cycle, green indicates genes whose expression deteriorated with artificial cycle, and blue highlights genes specific to recurrent implantation failure.
In addition to overlapping genes with DEGs deteriorated and improved, seven oestradiol- and progesterone-regulated RIF specific DEGs were in common with Ishikawa cell line experiments. Among these, seven genes, two genes were regulated by oestradiol and progesterone (29%), one solely by oestradiol (14%) and four by progesterone (57%) (Figure 6).
Microarray validation
Real-time polymerase chain reaction was used for validating microarray results in genes HABP2, HLA-DOB, SPDEF and TRH, which were dysregulated in infertile women in both natural and artificial cycles (Table 2), and have been shown in a previous study to be endometrial receptivity genes (
). Real-time polymerase chain reaction confirmed the array results: HLA-DOB, SPDEF and TRH were up-regulated, and HABP2 was down-regulated in patients who had experienced recurrent implanatation failure in NC-FET and AC-FET compared with NC-FC control women (Supplementary Figure S3).
Discussion
Despite the growing importance of FET in the treatment of infertility there is little consensus on the best protocol for endometrial preparation (
). To the best of our knowledge, this is the first study in which the effect of two different endometrial preparation protocols, NC-FET and AC-FET on the endometrial transcriptome among infertile patients who have experienced RIF has been analysed. We demonstrate that the whole endometrial gene expression pattern at the time of embryo implantation in the NC-FET protocol is more similar to the profile of fertile controls than is the AC-FET gene expression pattern. That means that, in this subgroup of infertile women, artificial endometrial preparation with oestradiol and progeserone alters more than improves the endometrial transcriptome, thus being disadvantageous, and the NC-FET protocol should be preferred.
Implantation failure remains an unsolved obstacle in reproductive medicine and is one of the major causes of infertility in otherwise healthy women (
). Inadequate uterine receptivity is estimated to account for two-thirds of implantation failures, whereas the embryo itself is responsible for only one-third of failures (
Specific and extensive endometrial deregulation is present before conception in IVF/ICSI repeated implantation failures (IF) or recurrent miscarriages.
Differences in the endometrial transcript profile during the receptive period between women who were refractory to implantation and those who achieved pregnancy.
). To sum up, RIF is relatively common among those seeking infertility treatment, and the management of RIF is considered as one of the most complicated issues in assisted reproduction (
In the artificial endometrial preparation cycle, oestradiol and progestrone administration is aimed at mimicking the endocrine exposure of the endometrium in normal cycles, where oestradiol initiates endometrial cellular proliferation and subsequent administration of progesterone leads to secretory changes (
). We therefore hypothesized that, among this patient group of infertile women, where aberrant endometrial maturation is suspected, artificial endometrial preparation would help to overcome or improve impaired uterine receptivity. Therefore, the result of AC-FET being disadvantageous versus NC-FET was surprising.
Our study results, in fact, demonstrate that the artificial cycle protocol favourably altered several transcripts of endometrial receptivity biomarkers implicated in previous independent transcriptome analyses on women with a fertile phenotype, such as CALB2 (
Discovery of biomarkers of endometrial receptivity through a minimally invasive approach: a validation study with implications for assisted reproduction.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Bioinformatic detection of E47, E2F1 and SREBP1 transcription factors as potential regulators of genes associated to acquisition of endometrial receptivity.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Changes in gene expression during the early to mid-luteal (receptive phase) transition in human endometrium detected by high-density microarray screening.
Bioinformatic detection of E47, E2F1 and SREBP1 transcription factors as potential regulators of genes associated to acquisition of endometrial receptivity.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Discovery of biomarkers of endometrial receptivity through a minimally invasive approach: a validation study with implications for assisted reproduction.
). The improvement of insulin-like growth factor binding protein-1 expression levels with artificial cycles is interesting, as this protein is an established biomarker of decidualization of endometrial stroma. It is involved in the implantation process through regulating insulin-like growth factor actions in endometrial cells, and cellular proliferation and differentiation required for decidualization and maintenance of early pregnancy (
). A number of pathways important in endometrial receptivity were also improved with artificial cycles, including cytokine–cytokine receptor interaction, G-protein coupled receptor and cyclic adenosine monophosphate-mediated signalling, lipid signalling and defense responses (
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
). Nevertheless, based on the whole transcriptome profile it seems that regardless of the “correction” of dysregulated genes with the AC-FET protocol, a negative effect of artificial cycles on the endometrial transcriptome prevailed.
The expression of various important genes known from previous studies to be involved in endometrial receptivity deteriorated with artificial cycles. These included CRISP3 (
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Discovery of biomarkers of endometrial receptivity through a minimally invasive approach: a validation study with implications for assisted reproduction.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Changes in gene expression during the early to mid-luteal (receptive phase) transition in human endometrium detected by high-density microarray screening.
Differences in the endometrial transcript profile during the receptive period between women who were refractory to implantation and those who achieved pregnancy.
). A number of relevant genes involved in implantation were also negatively altered with artificial cycles, including ESR2; fibroblast growth factors FGF17, FGF8, FGFBP2; FSHR; INHBC; interleukins IL1F6, IL27, IL29, IL4, IL9R; LEP; matrix metallopeptidases MMP17, MMP27, MMP3; PPARD; PTGER3, and WNT8A (
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
The effect of embryo presence on the expression of peroxisome proliferator activated receptor (PPAR) genes in the porcine reproductive system during periimplantation.
). Interestingly, the important hormone receptors, and follicle-stimulating hormone receptor were down-regulated in the endometrium with the AC-FET protocol. The crucial role of oestrogen receptor beta in the endometrium is gradually becoming more evident (summarized in
FSH Receptor (FSHR) expression in human extragonadal reproductive tissues and the developing placenta, and the impact of its deletion on pregnancy in mice.
). The most strongly artificial cyle-disrupted pathways among our patients were linoleic acid metabolism, and calcium signalling, both having important roles in endometrial receptivity (
Functional expression of large-conductance calcium-activated potassium channels in human endometrium: a novel mechanism involved in endometrial receptivity and embryo implantation.
). To sum up, regardless of the similar number of genes whose expression improved or deteriorated in the endometrium in connection with AC-FET, endometrial stimulation with oestradiol and progesterone in artificial cycles seemed to have a stronger negative rather than an improving effect on the expression of genes and pathways crucial for endometrial receptivity.
The ovarian steroid hormones oestrogen and progesterone play pivotal roles in development of the human endometrium. With a high prevalence of steroid hormone receptors in different compartments of endometrial tissue, it is believed that gene expression of this tissue is highly controlled by respective receptors, which are master transcriptional regulators of downstream genetic activity. There are two types of oestrogen receptors, ERα and ERβ, which share approximately 97% similarity in their DNA-binding domains (
). Progesterone also has two receptor isoforms, PRA and PRB, which differ only in that PRB contains an additional 164 amino acids in the N-terminal region. Both progesterone receptors bind to the same sites of DNA.
We carried out a comprehensive analysis of oestrogen response element and progesterone response elements sequences in genes involved in endometrial receptivity. Our in-silico analysis of hormone response element sites among DEGs that were influenced by artificial cycles demonstrated that the promoter regions of all genes contained response elements for oestrogen and progesterone receptors, and can thereby be directly modulated by oestradiol and progesterone. Summarized progesterone/glucocorticoid response element motifs showed median higher frequency compared with oestrogen receptors in all DEGs identified in this study. Those DEGs whose expression was deteriorated showed significantly less progesterone/glucocorticoid response elements and more oestrogen response element sites in their promoter region compared with genes whose expression was improved after artificial cycles. In addition, we detected a significantly higher number of one particular oestrogen response elements site among DEGs that deteriorated with artificial cycles, whereas one progesterone response element site was overrepresented in the proximal promoter region upstream of genes that improved with artificial cycles. These findings indicate that DEGs that deteriorated with artficival cycles are more responsive to oestradiol, whereas DEGs that improved with artificial cycles are more responsive to progesterone. Therefore, following the current artificial cycle protocol, oestradiol could have had stronger (negative) effect on the endometrial transcriptome than progesterone. In line with this, when we compared the DEGs with those in our previous study on Ishikawa cells, which are regulated by oestradiol and progesteronoe, we found that among DEGs that deteriorated with artificial cycle 32% were solely oestradiol-regulated and 47% were regulated by oestradiol together with progesterone, whereas, among the improved transcript group, only 4% were oestradiol-regulated and 75% were regulated jointly by oestradiol and progesterone. In addition, the fact that many DEGs that changed expression with artificial cycles (58%) demonstrated an opposite expression direction in comparison with cell line experiments, gives further support to the notion that artificial cycles unfavourably influenced endometrial gene expression in this group of infertile women, where oestradiol seems to have had a higher (negative) effect than progeserone.
Oestrogen is essential for endometrial proliferation, whereas its role in the secretory phase and in implantation is less clear (
). Human endometrium seems to function normally with very low concentrations of oestrogen, and, in a recent systematic review, it was suggested that there was no overall benefit in clinical outcomes of luteal oestrogen supplementation in IVF (
The luteal phase of recombinant follicle-stimulating hormone/gonadotropin-releasing hormone antagonist in vitro fertilization cycles during supplementation with progesterone or progesterone and estradiol.
). As oestradiol levels are higher in artificial compared with natural cycles, one would expect a thicker endometrium in artificial cycles, but this has not been observed (
). In our study also, we did not detect any significant differences in endometrial thickness between the NC-FET and AC-FET patients. Elevated levels of oestradiol have been suggested to lead to alteration in endometrial features and timing of the WOI, which could result in lower pregnancy rates in artificial cycle FET (
Progesterone is absolutely necessary for endometrial receptivity. Evidence from normally ovulating women suggests that only a very small amount of progesterone is required (
). In short, only small amounts of oestradiol and progesterone seem to be required in the secretory phase for full reproductive performance among women with normal endometrial function (
). Recent research, however, has shown that too high or too low mid-luteal serum progesterone concentrations (<50 and >99 nmol/L) associate with decreased implantation rates in cryopreserved embryo transfers conducted under hormone replacement (
). Women who have experienced RIF have a high probability of endometrial dysfunction, and hormonal endometrial stimulation could help to improve aberrant endometrial maturation. Indeed, we identified several important genes (and pathways) involved in endometrial receptivity whose expression improved with artificial cycles. Therefore, our results, in line with previous studies (
), encourage reconsideration and improvement of artificial cycle protocols in order to find optimal regimens for favourably maximizing endometrial gene expression profiles for distinct patient subgroups.
An additional finding in our study was the identification of genes specific to unexplained RIF. These were a subgroup of genes that were similarly and abnormally expressed in both endometrial preparation protocols, NC-FET and AC-FET among infertile women. We identified several relevant molecules and molecular pathways in female infertility, including genes involved in G-protein-coupled receptor and cAMP-mediated signalling, metabolism of xenobiotics by cytochrome P450, hormone activity, and lipid metabolism (
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Specific and extensive endometrial deregulation is present before conception in IVF/ICSI repeated implantation failures (IF) or recurrent miscarriages.
). The top molecules detected with hormonal activity were adrenomedullin 2, corticotropin-releasing hormone, inhibin beta A, and thyrotropin-releasing hormone. Furthermore, we identified four dysregulated genes that have been reported in endometrial transcriptome studies as being window of implementation-specific genes that could serve as biomarkers of RIF: HABP2 (
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
Changes in gene expression during the early to mid-luteal (receptive phase) transition in human endometrium detected by high-density microarray screening.
Bioinformatic detection of E47, E2F1 and SREBP1 transcription factors as potential regulators of genes associated to acquisition of endometrial receptivity.
). Reduced endometrial gene expression of hyaluronan-binding protein 2 (HABP2), an interesting biomarker of RIF that was down-regulated among patients who had experienced RIF, has also been associated with unexplained female infertility and recurrent pregnancy loss (
). Another important set of genes in our study are CASK, CASP8, COG, FCGBP, GRIN3B, PPA2, and SIX1 that are part of the recently published RIF prediction model (
An additional interesting finding among RIF-specific transcripts was a significantly high number of progesterone response element sites among these genes, indicating that pogesterone can have a direct stimulatory effect on their regulation. This notion was supported by the fact that about 60% of the transcripts were solely progesterone-regulated (whereas about 30% of the genes were regulated by both oestradiol and progesterone), when we compared the microarray results with those in our previous cell-line study (
Changes in the transcriptome of the human endometrial Ishikawa cancer cell line induced by estrogen, progesterone, tamoxifen, and mifepristone (RU486) as detected by RNA-sequencing.
). These findings highlight the fact that the genes dysregulated among our RIF patients respond to progesterone, reflecting dysregulation of progesterone signalling among these women. It is known from previous studies that compromised progesterone signalling can lead to impaired endometrial function (
). Indeed, aberrant expression of progesterone-regulated genes in the endometrium has been implicated in several gynaecological disorders, such as endometriosis, polycystic ovary syndrome and endometrial hyperplasia (
). Furthermore, a recent endometrial transcriptome study demonstrated that women refractory to embryo implantation have compromised progesterone signaling (
). Several signalling pathways are believed to be implicated in the pathogenesis of RIF, and they are of interest as regards identifying potential therapeutic targets and developing new therapies. Our findings provide additional information concerning the complex molecular regulation of endometrial receptivity in RIF patients, where dysregulation of progesterone signalling could have an important role.
To sum up, in a recent systematic review and meta-analysis, it was concluded that “it is not possible, based on the current published literature, to recommend one endometrial preparation method in FET over another” (
). Our results suggest that NC-FET is preferable to AC-FET in the subgroup of women who have experienced RIF. Furthermore, they also indicate that oestradiol might have a stronger (negative) effect on the endometrial transcriptome than progesterone in artificial cycles, thus encouraging reconsideration of the doses of administered steroid hormones and thereby improving protocols in artificial cycles. Despite the relatively small patient group, we believe that our transcriptome data provide valuable insights into the potential biomarkers and molecular mechanisms related to endometrial receptivity and RIF, but most importantly we hope our data will provide a step forward towards personalized medicine in the subgroup of infertile patients who have experienced recurrent implantation failure.
Funding
The study was funded by grants from the European Science Foundation (ESF) for the activity 'Frontiers of Functional Genomics', a Marie Curie post-doctoral fellowship (FP7, number 329812, NutriOmics), Enterprise Estonia (grant EU30020 and EU48695), Estonian Ministry of Education and Research (grant IUT34-16), the EU-FP7 Eurostars Program (grant NOTED, EU41564) and the EU-FP7 IAPP Project (grant SARM, EU324509).
Acknowledgements
We are grateful to all participants and study personnel who took part in this work. We thank Nick S Macklon from University of Southampton, UK for constructive comments and discussion, and Nicholas John Bolton for language revision.
Appendix. Supplementary material
The following is the supplementary data to this article:
List of genes whose expression improved with artificial cycle (AC) in patients who have experienced recurrent implantation failure (RIF) (column A), genes whose expression deteriorated with artificial cycles in patients who have experienced RIF (column B), and genes specific to RIF (column C). Down-regulation of RIF-specific genes is indicated with hyphen.
Biological function analysis in gene ontology terms of endometrial gene expression that artificial cycle improved in patients who have experienced recurrent implantation failure (RIF) using the Database for Annotation, Visualization and Integrated Discovery (DAVID). Canonical pathways were obtained using the Ingenuity Pathway Analysis. P < 0.05 was considered statistically significant.
Biological function analysis in gene ontology terms of endometrial gene expression that deteriorated after artificial cycle in patients who have experienced recurrent implantation failure using the Database for Annotation, Visualization and Integrated Discovery (DAVID). Canonical pathways were obtained using the Ingenuity Pathway Analysis. P < 0.05 was considered statistically significant.
Biological function analysis in gene ontology terms of infertility specific genes expressed in the endometrium of women who have experienced recurrent implantation failure using the Database for Annotation, Visualization and Integrated Discovery (DAVID). Canonical pathways were obtained using the Ingenuity Pathway Analysis. P < 0.05 was considered statistically significant.
List of all 17 position weight matrices used for the in-silico motif search and their occurrence in upstream of differentially expressed genes (DEG): DEG that improved with artificial cycle; DEG that deteriorated with artificial cycle, and DEG specific to patients who have experienced recurrent implantation failure.
Principal component analysis of full gene expression profiles at “window of implantation” in infertile women who have experienced recurrent implantation failure in natural cycles for frozen embryo transfer (NC-FET) (designated as I), in hormone replacement therapy cycles for FET (AC-FET) (designated as IS), and in NC-fertile control women (NC-FC) (designated as C).
Research resource: interactome of human embryo implantation: identification of gene expression pathways, regulation, and integrated regulatory networks.
The effect of embryo presence on the expression of peroxisome proliferator activated receptor (PPAR) genes in the porcine reproductive system during periimplantation.
Changes in gene expression during the early to mid-luteal (receptive phase) transition in human endometrium detected by high-density microarray screening.
Discovery of biomarkers of endometrial receptivity through a minimally invasive approach: a validation study with implications for assisted reproduction.
The luteal phase of recombinant follicle-stimulating hormone/gonadotropin-releasing hormone antagonist in vitro fertilization cycles during supplementation with progesterone or progesterone and estradiol.
A GnRH agonist and exogenous hormone stimulation protocol has a higher live-birth rate than a natural endogenous hormone protocol for frozen-thawed blastocyst-stage embryo transfer cycles: an analysis of 1391 cycles.
Impact of frozen-thawed single-blastocyst transfer on maternal and neonatal outcome: an analysis of 277,042 single-embryo transfer cycles from 2008 to 2010 in Japan.
Specific and extensive endometrial deregulation is present before conception in IVF/ICSI repeated implantation failures (IF) or recurrent miscarriages.
Characterization of the functional progesterone receptor in an endometrial adenocarcinoma cell line (Ishikawa): progesterone-induced expression of the alpha1 integrin.
Obstetric and perinatal outcomes in singleton pregnancies resulting from the transfer of frozen thawed versus fresh embryos generated through in vitro fertilization treatment: a systematic review and meta-analysis.
Evidence of impaired endometrial receptivity after ovarian stimulation for in vitro fertilization: a prospective randomized trial comparing fresh and frozen-thawed embryo transfer in normal responders.
FSH Receptor (FSHR) expression in human extragonadal reproductive tissues and the developing placenta, and the impact of its deletion on pregnancy in mice.
Changes in the transcriptome of the human endometrial Ishikawa cancer cell line induced by estrogen, progesterone, tamoxifen, and mifepristone (RU486) as detected by RNA-sequencing.
Differences in the endometrial transcript profile during the receptive period between women who were refractory to implantation and those who achieved pregnancy.
Bioinformatic detection of E47, E2F1 and SREBP1 transcription factors as potential regulators of genes associated to acquisition of endometrial receptivity.
Functional expression of large-conductance calcium-activated potassium channels in human endometrium: a novel mechanism involved in endometrial receptivity and embryo implantation.
Signe Altmäe is a Marie-Curie Post-Doctoral Fellow at the School of Medicine, University of Granada, Spain and a researcher at the Competence Centre on Health Technologies, Estonia. Her research interests are female and male infertility, endometrial receptivity and early pregnancy establishment.
Article info
Publication history
Published online: March 28, 2016
Accepted:
March 15,
2016
Received in revised form:
March 14,
2016
Received:
November 26,
2015
Declaration: The authors report no financial or commercial conflicts of interest.
30028-1&id=rbmo1528-fig-0001.jpg){kind=link}
30028-1&id=rbmo1528-fig-0002.jpg){kind=link}
30028-1&id=rbmo1528-fig-0003.jpg){kind=link}
30028-1&id=rbmo1528-fig-0004.jpg){kind=link}
30028-1&id=rbmo1528-fig-0005.jpg){kind=link}
30028-1&id=rbmo1528-fig-0006.jpg){kind=link}