Transcription profile analysis of the endometrium revealed molecular markers of the personalized 'window of implantation' during in vitro fertilization.

Abstract To determine the most informative markers for assessing the functional state of endometrium during the ‘window of implantation’ and creating a model for assessment of the readiness of endometrium for embryo implantation. Forty-seven women with tubal infertility and a successful IVF pregnancy participated in the study. Pipelle endometrial sample was performed during the supposed ‘window of implantation’ in natural cycle with subsequent histological study, and transcriptional profile of genes GPX3, PAEP, DPP4, TAGLN, HABP2, IMPA2, AQP3, HLA-DOB, MSX1, POSTN determined by real-time quantitative polymerase chain reaction (qRT-PCR). Differences in the level of mRNA expression of all the studied genes in the receptive endometrium were found in comparison to the prereceptive one, which allowed us to classify two functional states of the endometrium. The results of histological examination responded to the stage of maturation of the endometrium in 78.7% of cases. Receptive endometrial status can be determined based on the integral evaluation of mRNA expression level of 4 PAEP, DPP4, MSX1, and HLA-DOB genes. The model for determining a personalized `window implantation' is offered for practical application in ART.


Introduction
Human implantation is a complex and multifactorial process. The success of implantation depends on the quality of the embryos and the readiness of the endometrium for embryo nidation. Recent studies of global transcriptome gene profiles have been conducted under different conditions that affect the endometrium qualitative state. The microarray-based gene expression technology of endometrium identified that the «window of implantation» not always correlates with the 19th-21th day of menstrual cycle. The «window of implantation» is displaced for 2-3 days in every fourth patient with IVF failure and this has fatal consequences for implementation of the assisted reproductive technology (ART) [1]. The main result of the studies of global transcription profiles of genes was the identification of the most informative markers of the «window of implantation», and some of them are already included in the 'gold standard' of the endometrium receptivity [2,3]. The method of quantitative reverse transcription and polymerase chain reaction (RT-PCR) can be an alternative to expensive, time-consuming, and redundant method of microarray analysis in clinical practice. To determine the personalized 'window of implantation' in women with unsuccessful attempts of IVF and ET program, it may be sufficient to study a small number of genes by real time RT-PCRa method whose sensitivity and specificity is equal to microarray analysis, but the latter is more accessible to routine practice. We were the first to carry out different versions of multivariate analysis of expression data of endometrial tissue at different stages of menstrual cycle to identify a complex of potentially informative genes. Furthermore, we tested the markers for determining the receptive status of the endometrium using the real-time quantitative polymerase chain reaction (qRT-PCR). The aim of the study is to determine the most informative markers for assessing the functional state of endometrium during the 'window of implantation' and creating a model for assessing the readiness of endometrium for embryo implantation.

Materials and methods
Patients with tubal infertility factor underwent endometrial biopsy procedure using Pipelle endometrial sampler (Pipelle de Cornier V R , Laboratorie C.C.D.», France) on LH þ7-8 days in cycle before stimulation superovulation (7-8 days after the peak of luteinizing hormone). Peak level of LH was diagnosed according to the US data and/or urine Clear Blue test (Unipath Ltd, UK). We enrolled in the study only women with successful pregnancy after IVF (n ¼ 47).
The sample was divided into two parts. One part was fixed and stored in a special transport medium for RNA stabilization in IntactRNA bioassays in accordance with the Instructions for the Kit (Eurogen, Russia). The other part was prepared for histological examination.
After appropriate standard treatment, the samples were fixed in 10% neutral formalin for 24 h and then were encapsulated in paraffin. Sections 5 microns thick were made and stained with hematoxylin and eosin. Slides were examined under the light microscope (Â50-Â400).
RNA was isolated on the columns using the RNeasy Mini Kit in accordance with the Kit Instruction Manual (Qiagen, Germany). The tissue was pre-homogenized in a solution of guanidine isothiocyanate using a TissueLyser II laboratory homogenizer and deproteinized using a QIAzol Lysis Reagent reagent (Qiagen, Germany). The quality of the isolated total RNA preparation was evaluated on a Thermo Scientific NanoDrop 2000 spectrophotometer.

Bioinformatics
Bioinformatical analysis of open access data of normal endometrium expression performed in order to search for key genes and their expression changes in the processes of endometrial regeneration. The data available in free access were analyzed. The first stage of bioinformatic analysis is the search for data suitable for studying the selected object. In this case, it was interesting for us to study the expression of genes in the normal endometrium in different phases of the menstrual cycle. For this purpose, a search was made in the database of expressions ArrayExpress [https:// www.ebi.ac.uk/arrayexpress/] [citations: https://www.ebi.ac.uk/ arrayexpress/about.html]. As search terms, we used «endometrium» and «Homo sapiens». The following datasets were selected (Table 1).
In this data set, samples of normal endometrium are available for analysis in three stages: proliferative (n ¼ 5), early secretory (n ¼ 3), and medium secretory (n ¼ 8). First of all, we loaded the data to R [https://cran.r-project.org/], the data were normalized using the method of quantile normalization implemented in the limma package [4], then samples of normal endometrium were selected, and analyzed using principal component analysis [5].

Results
The results of PCA-analysis are shown in Figure 1.
On the PCA plot, we see that the samples are divided according to the stage at which they were taken. The next step is to identify genes, the expression of which is the same at different stages of the cycle (expression increases/decreases). For this purpose, Ward's hierarchical agglomerative clustering method was used [6]. Eight stable clusters of genes were identified ( Figure 2). One cluster included 2148 genes, 2-1537, 3-2830, 4-3569, 5-3517, 6-2513, 7-3611, and 8-3795 genes.
Analyzing the obtained results and comparing them with the data of the ERA study [3], 10 genes were isolated, which would allow isolate the samples of the middle secretion stage in the cohort of endometrial specimens (Table 3).
During the 'implantation window' compared to the stage of early secretion, the level of mRNA expression of 7 genes increases.
A discriminant function has been chosen on the basis of multifactor analysis that allows to classify samples according to the level of expression of 4 genes: PAEP, DPP4, MSX1, and HLA-DOB. The optimum value of the cutoff is determined using the ROC-analysis. The area under the ROC curve was AUC ¼1.0 (p ¼ 6.6 Â 10-13). The cut off value of the IRE (index receptivity endometrium) was 62. The values of cut off above 62 classified the samples as 'receptive endometrium.' The sensitivity and specificity of the proposed method for evaluating the endometrial receptivity in the range of the threshold value was 100% compared to cluster analysis.

Discussion
The study revealed changes in the expression of all the studied genes during 'implantation window.' The results are completely identical to the literature data [3]. The clustering method allowed us to distinguish two clusters of samples discriminating the receptive and pre-receptive endometrium. The results of this classification in 78.7% of cases corresponded to the histological conclusion, which corresponds to the literature data.
Each fifth female in the general population and every fourth woman with repeated unsuccessful IVF attempts is believed to have a histological conclusion that does not correspond to the receptor status of the endometrium [7].
For practical application, a classification, determined by the multivariate analysis method, can be proposed for the level of mRNA expression of 4 genes: PAEP, DPP4, MSX1, and HLA-DOB. The expression level of two of them is significantly increased (PAEP, DPP4), and the expression level of the other two (MSX1 and HLA-DOB) is lower in the receptive endometrium compared with the pre-receptive endometrium.
The gene PAEP (progestagen associated endometrial protein), also known as a pregnancy-associated endometrial alpha-2-globulin (PAEG) or placental protein 14 (PP14), encodes a glycoprotein of the lipocalin superfamily, the genes of most of which are grouped on the long arm of the chromosome 9. PAEP has three isoforms that play an important role in maintaining an intrauterine environment for providing, initiating, and elongating of pregnancy. In addition to the endometrium and amniotic fluid, high expression of this glycoprotein is found in the follicular fluid and semen plasma, where its functions are associated with providing a sequence of events necessary for fertilization [8].
The gene DPP4 (dipeptidyl peptidase 4), also known as the CD26 differentiation cluster, encodes dipeptidyl peptidase-4, which is a membrane enzyme that hydrolyzes the peptide linkage of proline at the N-terminal position. The gene product is the surface antigen of activation of CD26 T cells, is expressed on the surface of most cells of the body, participates in immune regulation, signal transduction, glucose homeostasis and in apoptosis [9].
The gene MSX1 (msh homeobox 1) or homeobox gene 7 HOX7 encodes a protein of the homeobox family. As a result of interaction with components of transcription complexes, this protein provides repression of differentiation genes during embryogenesis, plays a role in suppression of tissue growth [10].
The gene HLA-DOB (major histocompatibility complex, class II, DO beta) encodes a beta chain of a protein that performs an auxiliary function in antigens presentation. The HLA-DO protein interacting with HLA-DM suppresses the peptide loading of the molecules of the main histocompatibility complex of class II, preventing the presentation of antigens [11].
We believe that the proposed model can be used to classify the pre-receptive and receptive endometrium in the natural cycles of IVF (LH þ7) or in cycles using hormone replacement therapy (on the fifth day of progesterone administration), but it requires further validation for more quantities of samples. In addition, the issue of determining the postreceptive status of the endometrium remains urgent.
There were no samples of the 'post-receptive' endometrium in our study. Molecular classification of the three functional states of the endometrium (pre-receptive, receptive, and postresceptive)  would significantly optimize preimplantation diagnostics of the endometrium and fasten the process of determining the personalized 'window of implantation'.

Conclusion
The study of the transcriptional profiles of the GPX3, PAEP, DPP4, TAGLN, HABP2, IMPA2, AQP3, HLA-DOB, MSX1, and POSTN genes, revealed significant differences in the level of mRNA expression of these genes in the pre-receptive and receptive endometrium and it allowed to be classified the samples according to their receptive status. A proposed model for determining a personalized 'window of implantation' can be tested in future and used in ART programs.

Disclosure statement
The authors report no conflicts of interest. pre-recepƟve endometrium recepƟve endometrium