Transcriptome study of receptive endometrium in overweight and obese women shows important expression differences in immune response and inflammatory pathways in women who do not conceive.

Obesity and being overweight are growing worldwide health problems that also affect women of reproductive age. They impair women's fertility and are associated with lower IVF success rates. The mechanism by which increased body weight disrupts fertility has not yet been established. One possibility is that it affects the process of embryo implantation on the endometrial level. The purpose of our study was to determine the differences in enriched biological pathways in the endometrium of overweight and obese women undergoing IVF procedures. For this purpose, 14 patients (5 pregnant, 9 non-pregnant) were included in the study. Endometrial samples were obtained during the window of implantation and RNA sequencing was performed. There were no differences in general patient's and IVF cycle characteristics between pregnant and non-pregnant women. In the endometrial samples of women who did not conceive, pathways related to the immune response, inflammation, and reactive oxygen species production were over-expressed. Our findings show that the reason for implantation failure in overweight and obese women could lie in the excessive immune and inflammatory response at the endometrial level.

Introduction

Successful embryo implantation requires two conditions to be met; the presence of a high-quality embryo and a receptive endometrium. Endometrial receptivity is a limited time period during the menstrual cycle when the endometrium enables embryo implantation and is also called the window of implantation (WOI) [1]. It has been estimated that one-third of embryo implantation failures occur due to an unreceptive endometrium [2]. During in vitro fertilization (IVF) procedures the transfer of highest quality embryos remains the gold standard of care and, therefore, an implantation failure remains the unresolved issue.

Obesity is an increasing worldwide health problem with reports that more than half of reproductive-age women are overweight or obese [3]. Obesity causes impaired fertility and obese women take longer to conceive than normal weighing women [4]. The results on the impact of being overweight or obese on IVF procedures are conflicting, as both a negative impact [5, 6] as well as no impact [7, 8] have already been reported. The mechanism by which increased body weight has an effect on the success of IVF procedures is still unknown.

With the emergence of ‘omics’ technologies, numerous research groups have analyzed endometrial gene expression in different phases of the menstrual cycle [9, 10], especially during the WOI [11, 12]. Many genes have been proposed as potential biomarkers of endometrial receptivity, however, candidate genes differ between studies and a consensus on biomarkers has to date not yet been reached [13]. Low reproducibility of results is mainly the result of differences in experimental design, patient selection criteria, day and method of endometrial sampling, transcriptome platforms, and statistical methods used. Thus, biomarkers selected from these lists would perform with varying degrees of success.

Besides determining potentially differentially expressed genes, transcriptomic studies enable the analyses of enriched biological pathways and processes in tissues. A pathway analysis approach improves sensitivity and selectivity of gene expression studies as it does not focus on unrelated genes but a collection of up—and down—regulated genes obtained in the experiment [14]. This kind of an approach is advantageous, as subtle physiological changes are often not translated into large gene expression variations and many genes from the same pathway moving in a cluster are a good indication of the physiological processes that take place in the studied tissue [15]. The pathway analysis approach considers the fact that many times genes involved in the same biological process are deregulated together [15]. Furthermore, in pathway analysis minor differences in study methods are less likely to influence results and, therefore, results across different data sets are more consistent and robust [16]. Pathway analysis may, therefore, reveal novel insights about biological processes active in the examined tissue in the situation of interest [17].

Elucidating the molecular mechanisms and over-represented biological processes during the WOI is important to better understand endometrial functioning and improve success rates of IVF procedures.

The present study aimed to exhibit active biological processes in the endometrium of overweight and obese women during the WOI, when the endometrium should be receptive to allow embryo implantation. We aimed to determine the differences in endometrial transcriptome between pregnant and non-pregnant groups of overweight and obese women undergoing IVF procedures. This could characterize potential endometrial mechanisms that impact endometrial receptivity in IVF procedures. Therefore, endometrial samples were obtained during the WOI and transcriptomic analysis of endometrium using RNA sequencing was performed.

Materials and methods

All patients were informed of the study aims and signed a written informed consent form before entering the study. The study was conducted at a tertiary care university hospital in accordance with the Declaration of Helsinki. The study was approved by the National Medical Ethics Committee of the Republic of Slovenia (approval number 0120-491-2017). The study was performed at a tertiary care facility at the Department of Human Reproduction, Division of Obstetrics and Gynecology Ljubljana, Slovenia.

Patients

Clinical information and endometrial samples were collected from 14 participating infertile patients from February 2017 to February 2018. The inclusion criteria were: under 38 years of age, first or second IVF cycle, unexplained or tubal factor infertility, partner’s normal result of semen analysis. All patients underwent clinical (general and gynaecological examination), anthropometric (body mass, body mass index (BMI), waist circumference), and hormonal assessment (prolactin, LH and FSH, TSH). Blood for hormonal evaluation was drawn during the follicular phase of the menstrual cycle, i.e., between the third and the fifth day of the cycle.

Patients were stratified and analyzed according to their IVF results to a pregnancy (PG; N = 5) or non-pregnancy (non-PG; N = 9) group.

Endometrial tissue collection

Endometrial biopsy was performed using a catheter (Rampipella, RI.MOS., Mirandola, Italy) under sterile conditions between the 21st and 23rd day of the menstrual cycle preceding the IVF procedure. All samples were snap-frozen in liquid nitrogen and stored at −80°C until further processing.

IVF procedure

Short protocol (recombinant FSH in combination with GnRH antagonist) was used for controlled ovarian hyperstimulation. Vaginal ultrasound examination was performed to monitor follicular development. Final follicular maturation was induced by administering 6500 IU of human chorionic gonadotropin (hCG) when at least three follicles measured 17 mm in diameter. Ultrasound-guided transvaginal oocyte pick-up was performed 34–36 h later. Embryo transfer was performed on day 3 or 5 after oocyte pick-up. Biochemical pregnancy was determined by measuring ßhCG 14 days after embryo transfer, and clinical pregnancy was confirmed by the presence of a gestational sac and fetal heartbeat by ultrasound examination 6 weeks after embryo transfer.

Sample preparation and RNA sequencing

Total RNA was isolated from endometrial tissues using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. In short, biological samples were first lysed and homogenized in the presence of a highly denaturizing guanidine-thiocyanate–containing buffer, which immediately inactivated RNases to ensure purification of intact RNA. Ethanol was added to provide appropriate binding conditions, and the sample was then applied to a RNeasy Mini spin column, where the total RNA bound to the membrane and contaminants were washed away. RNA was then eluted in 30–100 μl water. The purity of isolated RNA was assessed using the NanoDrop 2000c spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA) and integrity was determined on Agilent 2100 Bioanalyzer using RNA 6000 Nano LabChip kit. All of the samples had RNA integrity values greater than 7.0 and were used for library preparation. Extracted RNA samples were stored at -80°C until further processing.

Library preparation was performed using TruSeq stranded total RNA sample preparation kit (Illumina), according to the manufacturer’s protocol. After Ribo-Zero rRNA depletion, the remaining RNA was purified, fragmented, and primed for first-strand cDNA synthesis with reverse transcriptase (SuperScript II) and random primers, followed by second-strand cDNA synthesis performed in the presence of dUTP. Blunt-ended double-strand DNA was 3’ adenylated and multiple indexing adapters (T-tailed) were ligated to the ends of the ds cDNA. Fragments with adapter molecules on both ends were selectively enriched with 15 cycles of PCR reaction. Prepared libraries were normalized to the final concentration of 10 nM and then pooled in equimolar concentrations. Sequencing was performed on the Illumina Hiseq 2000 sequencing system in 2x100 sequencing cycles using pair-end sequencing mode.

RNAseq data analyses

After initial read quality filtering and demultiplexing, alignments to the human reference genome (hg19) were performed. The reads were aligned to a genome and transcript splice sites were discovered with HISAT [18]. Alignments were assembled into full and partial transcripts and the expression levels of all genes and transcripts were estimated by StringTie [19]. The transcripts and expression levels from StringTie were statistically processed by Ballgown [20], which were then estimated using read counts function in the Subread package for R 5, followed by an intersample normalization by variance modeling at the observational level (voom) approach implemented in the limma Bioconductor package. The differences in gene expression were estimated using linear modeling and significance estimation procedures in the limma [21]. Quality control analysis was done with RSeQC [22].

RNA sequencing data

Functional characterization of the endometrial transcriptional profile

Functional analyses of the identified transcriptional alterations were characterized using the Ingenuity Pathway Analysis software (QIAGEN, Redwood 185 City, CA, USA). The IPA enables the analysis of functional pathway enrichment and determination of the state of upstream regulators (UR). Genes with significance values below 0.05 were included in the analyses. The remaining parameters were kept in their default setting.

A right-tailed Fisher’s exact test was used for calculating a significance score for each association between genes in the experimental dataset and a biological function. To reduce false-positive results, IPA pathways were considered significant only if they were associated with a p-value <0.05. The predicted active or inactive state was determined by Z—score (predicted activation > 2, predicted inactivation < -2).

Statistical analysis of general patient characteristics

Normal data distribution was checked with the Shapiro–Wilk test. T-test was used for the normally distributed data, and Mann-Whitney U test for non-normally distributed data. P-value of <0.05 was considered statistically significant. The results are presented as a median and interquartile range (the distance between the 25th and 75th percentiles). Statistical analysis was performed using IBM SPSS Statistics, version 24 (IBM Corp, Armonk, NY).

Results

General patient and IVF cycle characteristics

Fourteen patients were recruited and stratified according to their IVF result to PG (N = 5) or non-PG (N = 9) group.

The comparison of the anthropometric characteristics, endocrine parameters, and IVF cycle characteristics showed no significant differences between groups (Table 1).

Table 1

Comparison of anthropometric characteristics, endocrine parameters, and IVF cycle characteristics in the pregnant and non-pregnant group.

	Pregnant	Non-pregnant	P value
	(N = 5)	(N = 9)
Age (years)	34.0 (32.0, 34.0)	31.0 (29.0, 35.0)	0.36
Years of infertility	48 (24.0, 60)	36 (24.0, 48)	0.17
Body mass (kg)	80.0 (67.0, 88.0)	63.0 (60.0, 75.0)	0.62
BMI (kg/m2)	32.6 (31.5, 33.7)	36.5 (25.9, 39.3)	0.82
Abdominal circumference (cm)	90.0 (80.0, 92.0)	72.0 (72.0, 103.0)	0.76
PRL	10.3 (5.2, 13.6)	10.9 (7.2, 14.1)	0.63
FSH	6.1 (5.2, 6.6)	6.1 (4.8, 6.2)	0.48
LH	5.7 (5.5,6.1)	4,6 (4.0, 5.0)	0.47
TSH	3.6 (1.8, 4.0)	1.7 (1.2, 2.7)	0.19
Dosage of GNT (IU)	2200 (2025, 2200)	2025 (2000, 2400)	0.70
Days of stimulation	11 (9,11)	10 (10,12)	0.79
Oocytes/cycle (N)	7 (7,11)	8 (6,22)	0.74
Mature oocytes/cycle (N)	7 (6,8)	5 (4,16)	0.60
Fertilized oocytes/cycle (N)	6 (5,8)	5 (3,8)	0.9
Fertilization rate (%)	72.7 (71.4, 83.3)	50.0 (34.8, 72.7)	0.27
Blastocysts/cycle (N)	2 (1,5)	1(1,5)	0.94

Enrichment analyses revealed disturbances in pathways associated with inflammatory response and ROS production in the endometrium of non-pregnant women

Comparison of transcriptional differences between PG and non-PG women revealed that 1419 genes surpassed the nominal significance threshold of 0.05. The list of differentially expressed genes with an unadjusted p-value of < 0.05 was subjected to IPA. There were 253 canonical pathways that showed significant dysregulation between PG and non-PG endometrial samples (S1 Table).

Pathways as IL-8 signaling (p = 1.3 x 10−8, z-score– 3.8), chemokine signaling (p = 4.06 x 10−8, z-score -3.1), production of nitric-oxide and reactive oxygen species in macrophages (p = 3.8 x 10−8, z-score -2.8), role of NFAT in regulation of immune response (p = 5.2 x 10−7, z-score -3.3), NRF-2—mediated oxidative stress response (p = 9.06 x 10−5, z-score -2.4) were among the over-represented in endometrial samples of women who did not conceive (Fig 1).

Fig 1

Top 5 over-expressed pathways in non-PG endometrial samples.

According to IPA, 30.2% (76 of 252) of pathways were related to immune response and inflammation: 56 to cellular immune response, 5 to humoral immune response, and 15 to cellular stress and injury. Moreover, there was an over-expression of cellular immune response pathways (41 out of 56) in the non-PG compared to PG endometrium (S2 Table).

Discussion

In the present study, pathways related to immune response, inflammation, and production of ROS are among the over-expressed in endometrial samples of overweight and obese women who did not conceive in the IVF procedures. To the best of our knowledge, such an analysis has not yet been performed in overweight and obese infertile women. The roles of some of the overexpressed pathways in the process of embryo implantation are discussed below.

Chemokine signaling

Chemokines are a family of chemotactic cytokines that recruit and activate leukocytes from the periphery to the site of inflammation [23]. In the uterus, chemokines act as chemo-attractants for different leukocytes–uterine natural killer cells (uNK), T-cells, and macrophages [24]. Although inflammation with on-site angiogenesis, tissue remodeling, and recruitment of macrophages represents a necessary step in embryo implantation, an excessive inflammatory immune response acts as an important factor in the pathophysiology of impaired embryo implantation [25]. The cellular immune response that is mediated by uNK and T-cells represents the crucial step in the process of embryo implantation. Excessive or inappropriate recruitment of peripheral blood NK cells to the uterus, however, causes cytotoxic environment in the uterus with disrupted trophoblast proliferation and differentiation [25]. Additionally, uNK cells play a major role in the development and remodeling of uterine spiral arteries, a crucial step for pregnancy [26]. An increased number of uNK cells were found at the implantation site of women with recurrent pregnancy loss as compared with pregnant women with uneventful pregnancies [27]. This can cause abnormal development of spiral arteries and inadequate blood flow to the feto-placental unit, which leads to oxidative stress to the developing trophoblast and implantation failure or miscarriage [28].

Besides chemotactic properties, chemokines also alter the expression of cellular adhesion molecules and extracellular matrix (ECM) components that allow cell migration and it has been shown that they have functional effects on trophoblast cell attachment and migration [24]. Genes SDF1(CXCL12) and its receptor CXCR4, both part of chemokine signaling pathway and up-regulated in non-PG endometrium in the present study, are known to play a role in the cross-talk between trophoblast and endometrium, recruitment of lymphocytes into the uterus of pregnant females, and vascularization [29]. It is possible that besides the aberrant immune response and spiral arteries development, the adhesion of the embryo to the endometrium was disrupted in the present study due to pathological activation of chemokines.

Taken together, chemokines are crucial for the establishment of the placenta and pregnancy. Up-regulation of the chemokine signaling pathway in non-PG endometrial samples in the present study has again shown that aberrant activity of this pathway has a negative impact on the establishment of pregnancy.

Th1 pathway, Th1 and Th2 activation pathway

T–helper cells play a central role in modulating the immune response during pregnancy [30]. There are two classes of T-helper lymphocytes; Th1, which are considered to be pro-inflammatory, and Th2, which counteract Th1 cells and are characterized as anti-inflammatory. Th1 immunity and impaired Th2 response are related to unexplained recurrent pregnancy loss [31]. Women with recurrent pregnancy loss and implantation failure have increased Th1 immune response [32] as well as increased NK cell levels and NK cytotoxicity [33]. Considering all of the above, we assume that the response of pro-inflammatory Th1 cells is increased in non-PG endometrial samples in our study which leads to unsuccessful embryo implantation.

In line with this hypothesis also lies the finding that in non-PG endometrial samples the pathway role of NFAT in the regulation of the immune response was up-regulated. The nuclear factor of activated T cells (NFAT) family of transcription factors is believed to play a pivotal role in T cell activation and proliferation and T-cell dependent regulation of cytokine genes [34]. In a study by Porter and Clipstone [35], sustained NFAT signaling acted as a promoter of Th1-like pattern of gene expression with an increase in the number of cells expressing the prototypical Th1 cytokines (e.g. IFN-gamma, PSGL-1).

Production of nitric oxide and reactive oxygen species in macrophages

Macrophages represent 1–2% of cells within the endometrium during the proliferative phase of the menstrual cycle. The percentage increases to 3–5% during the secretory phase and peaks at 6–15% during the menstrual phase [36]. In early pregnancy, macrophages represent approximately 10–20% of the leukocytes at the maternal–fetal interface [37]. They are broadly divided into two categories: M1 and M2 [38]. Activation of M1 macrophages involves antigen presentation, reactive oxygen species (ROS) production, and Th1—type response characterized by maximizing cytotoxic and inflammatory capabilities of immune cells responding to viral infections and grafts [39]. On the other hand, M2 macrophages have immunosuppressive characteristics and are involved in Th-2 response [34]. During the embryo implantation period and pregnancy, uterine macrophages have roles in maternal uterine adaptation including establishing local immune balance, trophoblast invasion, and spiral arteries remodeling [40, 41]. Aberrantly activated macrophages are capable of producing several molecules and when activated by type 1 stimuli, macrophages increase the production of inducible nitric oxide synthase (iNOS) and subsequently NO. Macrophage-derived NO has been shown to significantly reduce trophoblast invasiveness in vitro [42].

Reactive oxygen species are generated during processes of oxygen consumption [43] and they play an important role in several physiological processes in female reproduction (oocyte maturation, ovarian steroidogenesis, physiological follicular atresia, ovulation, fertilization, luteal regression, and luteal maintenance in pregnancy) [44, 45]. In physiological states, antioxidants maintain a careful oxidant/antioxidant balance and, therefore, intracellular homeostasis [43]. However, oxidative stress occurs when this balance is disrupted. It has been shown that oxidative stress negatively impacts female reproductive processes. It causes oocyte meiotic spindle damage and chromosomal changes that lead to lower oocyte quality [46], embryo development arrest [47], embryo fragmentation [48], and recurrent implantation failure [49, 50].

Our results show that the uterine microenvironment in the non-PG endometrial samples is in the state of aberrant M1 macrophage activation and consequently, oxidative stress, which negatively impacts embryo implantation.

NRF-2—Mediated oxidative stress response

This pathway was also over-represented in non-PG endometrial samples in the present study. Nrf2 has a key role in eliminating oxidative stress by regulating antioxidant enzymes and thus protecting cell function [51]. Our findings are in line with a study of Ponsuksili et al. [52], where microarray analysis of bovine endometrium on day 7 of estrous cycle revealed that Nrf2—mediated oxidative stress response pathway is enriched in a non-receptive endometrium. Altogether, these findings show that oxidative stress causes a hostile environment in the endometrium and, therefore, prevents embryo implantation.

Considering the fact that we analyzed the endometrium of over-weight and obese women in both study groups, we presume that bodyweight itself is not the reason for the over-expression of inflammatory and immune response pathways. The reason for over-activity of these pathways could lie in the impaired inflammatory state of the endometrium, a condition that includes chronic endometritis (CE), currently considered as being caused by a microbial agent, and endometrial inflammation caused by other agents (pollutants, mechanical trauma, hypoxia, ischemia) [53]. Chronic endometritis is an asymptomatic, subtle, and continuous endometrial inflammation characterized by the presence of plasma cells in the endometrial stroma, endometrial edema, increased stromal cell density, and dissociated maturation between epithelial cells and stromal fibroblasts [54, 55]. The presence of CE can alter endometrial receptivity, thus creating an inadequate microenvironment that causes implantation failure and recurrent pregnancy loss [56]. It has been shown that chronic endometrial inflammation affects the production of pro-inflammatory cytokines, involved in NK recruitment [55], which negatively impacts endometrial decidualization and consequently endometrial receptivity [54]. When considering non-infectious etiology for chronic endometrial inflammation, a higher prevalence of chronic inflammation was correlated with endometriosis, a history of prolonged menstrual bleeding, infertility, miscarriage, and fallopian tube obstruction [53]. Studies have shown that women with endometriosis have higher lead (Pb) levels in the urine and follicular fluid as compared to other causes of infertility [57, 58]. Furthermore, non-pregnant women in IVF cycles had higher intrafollicular Pb levels as compared to pregnant women [58]. It is possible that higher Pb levels also influence endometrium in terms of an impaired endometrial inflammatory state which causes lower IVF success rates.

Conclusions

Despite the efforts aiming to improve IVF success rates, embryo implantation remains the most important limiting step in IVF procedures. Several different strategies and therapies, such as preimplantation genetic testing [59], endometrial receptivity analysis [60], and intentional endometrial scratch injury [61, 62], have been proposed for increasing the chances of embryo implantation.

The present study has shown that in the endometrium of overweight and obese women who do not conceive pathways associated with immune response and inflammation are overactive during the window of implantation. Considering all studies that have already proposed different genetic biomarkers of endometrial receptivity, it is becoming more or less obvious that the analysis of one or only a few genes will not be enough to determine whether the endometrium is able to provide a suitable environment for embryo implantation during the WOI. Unraveling the unbalanced pathways during the WOI will perhaps enable the correction of defects that lead to implantation failure and thus improving IVF success rates.

Further studies are needed to evaluate which one of the proposed procedures for the increased chances of embryo implantation will truly have a significant impact on IVF success rates.

List of over-expressed canonical pathways in non-PG endometrial samples.

(DOCX)

Click here for additional data file.

List of pathways related to immune response over-expressed in non-PG endometrial samples.

(DOCX)

Click here for additional data file.