Collaboration avec des Instituts étrangers (Hammersmith Hospital)

Characterization of the endometrial decidual response in patients with persistent pregnancy failure

(a) Aims of the project

We have obtained compelling evidence that cyclic differentiation of endometrial stromal cells in preparation for pregnancy, a process termed decidualization, is grossly abnormal in patients suffering from recurrent pregnancy loss (RPL). Our overall aim is therefore to define preconception biomarkers and strategies capable of predicting and preventing subsequent pregnancy failure. The specific goals of this proposal are:

(i) To define the endometrial stromal cell epigenome and transcriptome that underpins abnormal decidual cell function in RPL.

(ii) To define the mechanism that control decidual cell responses to embryonic human chorionic gonadotrophin signalling in women with and without RPL.

(b) Work that has led up to the project

(i) Clinical background.

Compared to most mammalian species, human reproduction is remarkably inefficient. This is largely attributable to the high incidence of embryo loss, estimated to be 30% prior to implantation (pre-implantation loss) and a further 30% before 6 weeks gestation (pre-clinical/biochemical pregnancy loss). In addition, in excess of 10% of clinical pregnancies result in miscarriage, mostly prior to 12 weeks gestation, and 1-2% of couples experience recurrent pregnancy loss (RPL), defined as failure of 3 or more consecutive pregnancies. Besides the physical trauma, RPL is associated with considerable psychological morbidity with a third of patients attending specialist clinics suffering from clinical depression. A history of RPL also increases the risk of a variety of adverse obstetric outcomes in a subsequent ongoing pregnancy, including preterm delivery, premature preterm rupture of membranes, placenta praevia, low birth weight and congenital malformation.

Early pregnancy loss is widely viewed as a dichotomous disorder, attributable either to chromosomal instability in the embryo or to maternal factors. Although numerous anatomical, endocrine, immunological, thrombophilic and genetic perturbations have been invoked to explain euploidic pregnancy loss, none are specific or prevalent. Moreover, for most of these conditions, the pathological mechanisms that would lead to persistent pregnancy wastage are entirely conjectural. Chromosomal imbalances in the conceptus are estimated to account for approximately 50% of all miscarriages. These karyotypically abnormal losses are widely considered unavoidable ‘accidents’ of nature; and therefore beyond medical prevention. However, this defeatist attitude to pregnancy loss must be challenged in view of recent genome-wide array studies demonstrating that the frequency and complexity of chromosomal aberrations in pre-implantation human embryos are much higher than hitherto appreciated. In young women without a history of reproductive failure, less than 10% of cleavage-stage embryos have a normal karyotype in all blastomeres, approximately half have no normal cells at all, and the remainder are mosaic. The phenomenal incidence of gross chromosomal rearrangements in pre-implantation embryos implies that mechanisms must exist to limit implantation of severely compromised embryos and/or to promote positive selection of normal blastomeres in chromosomally mosaic embryos. Conversely, the high prevalence of aneuploidic pregnancies in RPL patients suggests that the primary defect may lie in this process of embryo selection and/or ‘repair’.

(ii) The role of endometrial decidualization in natural embryo selection

By restricting implantation to a few days in the menstrual cycle, the endometrium preferentially selects developmentally competent embryos, thus limiting subsequent pregnancy wastage. This concept is supported by clinical observations demonstrating that implantation beyond the normal period of endometrial receptivity, the so-called ‘implantation window’, is associated with a dramatic increase in subsequent pregnancy failure. Uniquely to humans and the few other menstruating species, the window of implantation during the mid-secretory phase of the cycle coincides with the differentiation of human endometrial stromal cells (HESCs) into specialized decidual cells. Using a human co-culture system, our collaborators in Utrecht found that decidualizing HESCs selectively recognize developmentally impaired embryos and respond by inhibiting the secretion of a panel of implantation mediators. Undifferentiated cells do not mount such a response, indicating that decidualization of HESCs is essential for recognition and elimination of compromised conceptions. The observation that differentiating HESCs serve as biosensors for embryo quality control further suggests that spontaneous cyclic decidualization of the endometrium, which is intricately linked to the process of menstrual shedding, may have emerged as an essential defence mechanism to limit implantation of highly invasive but chromosomally chaotic human embryos.

(iii) Endometrial decidualization is impaired in RPL

Based on the above observations, we speculated that impaired decidualization may not only cause euploidic pregnancy failure but also disable embryo recognition and selection, thereby facilitating implantation of developmentally compromised concepti. To test this hypothesis, the transcript levels of prokineticin-1 (PROK1, a cytokine that confers endometrial receptivity) and prolactin (PRL, a classical decidual marker) were measured in timed endometrium biopsies as well as in primary cultures from patients with and without a history of RPL. Compared with controls, RPL was associated with significantly higher PROK1 mRNA levels and approximately 100-fold lower PRL levels in endometrial biopsies. These differences were even more pronounced when primary HESCs were subjected to a decidualizing stimulus. Although PROK1 and PRL mRNA levels did not differ in undifferentiated HESCs or in cells decidualized for 48 hours, after 4 days of differentiation the rise in PRL transcript levels was several magnitudes (1000-fold) higher in the control group when compared to RPL samples. PROK1 levels continued to rise with comparable kinetics in all decidualizing cultures until day 8 when expression in the control but not RPL group declined markedly. This expression profile in cultured cells supports the notion that an impaired decidual response may lead to prolonged endometrial receptivity in RPL. Further, the pattern of expression in all samples corresponded to the clinical presentation, suggesting that defective decidualization is the primary defect in RPL.

(iv) Decidual cell responses to embryonic signals are disrupted in RPL

Because of the limited availability of surplus human embryos, it has not yet been possible to examine the decidual phenotype of individual primary cultures from RPL patients and controls upon co-culture with human embryos, although these studies are being pursued in Utrecht. We have, however, examined the effects of human chorionic gonadotropin (hCG), one of the earliest and most abundant glycoproteins secreted by embryonic trophoblast, on PRL and PROK1 expresssion. As shown in Fig. X, RPL is associated with paradoxical regulation of PRL and PROK1 expression in response to hCG signalling in decidualizing HESCs. Extrapolating to the in vivo situation, such dysregulated response will jeopardize the integrity of the feto-maternal interface as hCG, besides its luteotrophic actions in early pregnancy, directly stimulates angiogenesis and limits cell death responses in the maternal decidua.

Our data provide compelling evidence that endometrial programming in preparation of pregnancy is profoundly disrupted in RPL, characterized by impaired decidualization of resident stromal cells, prolonged endometrial receptivity, and a dysregulated maternal response to embryonic signals. This pathological pathway not only establishes a novel and unifying mechanism underpinning both chromosomal and non-chromosomal recurrent pregnancy failure but also infers that pre-pregnancy strategies can be developed to predict the quality of the decidual response, to identify women at risk of adverse pregnancy outcome, and to monitor the effectiveness of preventative medical interventions.

(c) Experimental design and methods

(i) Patient selection and primary cultures

Our expertise in establishing primary HESC cultures spans 15 years now. In the last year alone, we processed 71 and 87 endometrial samples from RPL and non-RPL patients, respectively. Patients are recruited from specialized fertility and recurrent miscarriage clinics within Imperial College Healthcare NHS trust and investigated according to standard clinic protocols. However, the outcome of these standard investigations is not routinely taken in account when recruiting participants or in subsequent data analysis. Primary HESC cultures will be established and decidualized with 0.5 mM 8-Br-cAMP (Sigma) and 10-6 µM MPA (Sigma), in the presence or absence of recombinant hCG, as previously described. In addition, we generated and characterized a novel telomerase-immortalized HESC line, St-T1b, capable of expressing a decidual phenotype. St-T1b cells are extensively used to optimize techniques. Finally, our collaborator Professor Quenby (University of Warwick) has kindly agreed to provide us with access to her endometrial bio-bank containing in excess of 600 phenotyped samples.

(ii) Characterization of the decidual response in RPL

(iia) Decidual gene expression profiling in RPL

Rationale. The decidual response in RPL has so far only been assessed using a candidate gene approach. We and others have extensively used microarray platforms for genome-wide profiling of decidual gene expression, both in vitro and in vivo. By combining this approach with targeted gene knockdown, we also characterized decidual gene networks and cellular functions under the control of specific steroid receptors (progesterone and androgen receptors) and key cAMP-dependent transcription factors, such as FOXO1. Thus, by defining the decidual gene expression profile associated with RPL and integrating it with existing data sets, we will be able to identify informative marker genes and delineate the signalling pathways and decidual cell functions perturbed in RPL.

Experimental approach. Total RNA from primary undifferentiated HESCs and cells decidualized for 8 days will be subjected to microarray analysis (Affymetrix Gene ST Array, Almac Diagnostics). To reduce bio-variability between samples, as well as costs, we will establish 4 pools consisting of 3 different primary cultures for each of the conditions to be tested (undifferentiated/decidualized and control/RPL). The 8-day time-point is chosen to ensure that maximal differences in gene expression are captured. Real-time quantitative (RTQ)-PCR will be used to validate the expression of informative genes in independent time-course experiments. Ingenuity Pathways Analysis (IPA, Ingenuity Systems) will be used to interrogate the biological significance of differentially expressed gene networks, which in turn will be tested at functional level (e.g. cell cycle control, apoptosis, migration, cytokine expression etc.) using standard molecular approaches.

(iib) The endometrial epigenome in RPL

Rationale. As outlined above, aberrant PRL and PROK1 expression in mid-secretory biopsies from RPL patients was recapitulated, and even amplified, upon decidualization of purified HESCs maintained in prolonged culture and established from samples taken randomly in the cycle. Such cellular ‘memory’ suggests that endometrial differentiation is subject to epigenetic programming. While some of the epigenetic marks are mitotically inheritable, thus contributing to cell memory, others are extremely dynamic and respond rapidly to signalling events. Several lines of preliminary data indicate that the decidual process itself is dependent upon epigenetic reprogramming of HESCs and perturbed in RPL. First, manual mining of existing array data indicated that decidualization is associated with regulated expression of DNA methyltransferases (DNMTs), mCpG-binding proteins, and histone-lysine methyltransferases. Second, Western blot analysis of pooled RPL and non-RPL cultures demonstrated loss of expression of DNMT1, responsible for maintenance and accurate replication of methylation patterns during cell division, upon decidualization, although the kinetics of this down-regulation occurred earlier in RPL samples. We also used chromatin immunoprecipitation (ChIP) analysis to examine key histone modifications associated with either gene expression or silencing. As shown in Fig X, decidualization of control cultures was associated with loss of the repressive lysine 27 trimethylation on histone H3 (H3K27me3) on the decidua-specific PRL promoter and enrichment of histone H4 tail acetylation (acH3, an activation mark). Importantly, analysis of these bivalency marks in a RPL culture revealed a paradoxical loss of acH3 at the PRL promoter locus upon decidualization. While these observations require validation in larger sample sets, the data do suggest that aberrant gene expression in RPL is underpinned by differences in the HESC epigenome.

Experimental approach. Given the perturbed decidual gene programme, aberrant DMNT1 expression and impaired regulation of histone marks such as acH3, we predict that the decidual cell epigenome will differ in RPL. A genome-wide approached will be used to identify promoters that display altered CpG methylation. DNA prepared from control and RPL HESCs, untreated or decidualized for 8 days, will be subjected to methylated DNA immunoprecipitation (MeDIP), and hybridized subjected 3x720K RefSeq Promoter Arrays (Nimblegen). These genome-wide methylation studies will be carried out in collaboration with Dr Jonathan Mill, Lecturer in Psychiatric Epigenetics (IoP, King’s College London). Bisulfite treatment followed by pyrosequencing will be used to examine the kinetics of changes in CpG methylation in promoter regions of informative genes. Complementing this approach, we will profile the expression of all DNMTs, as well as regulated mCpG-binding proteins (e.g. UHRF1, HELLS) and putative DNA methylases (e.g. GADD45A), in vitro and in timed endometrial samples, and examine their role in dynamic methylation events using overexpression and/or siRNA knockdown approaches. Because of the large number of histone modifications, a targeted approach will be employed first to identify informative marks before considering genome-wide chromatin mapping. In the context of decidualization, we will use ChIP analysis to monitor changes in bivalent modifications (e.g. H3K4me2/3, H3K9Ac, H3Ac, and H3K27me3) in promoter region of PRL and PROK1 (and possibly other highly differentially expressed genes identified in the array) upon decidualization of primary cultures from RPL and control patients. An extended screen may be necessary to uncover single or combinatory marks that are central to the cellular memory in HESCs. Several complimentary approaches will be taken to direct this screen. First, the gene array data (see above) of undifferentiated HESCs from RPL and control will be mined for the differential expression of known chromatin modifying enzymes, which in turn may point to specific informative marks. Global changes in the posttranslational histone code associated with RPL will be assessed by in primary cultures by immunofluorescence confocal microscopy or Western blot analysis of nuclear extracts, and by immunohistochemistry of tissue sections. In the absence of discriminatory global changes, promoters of non-regulated candidate genes that are differentially expressed in HESCs from RPL patients and controls will be screened for histone marks associated with transcriptionally permissive euchromatin or repressive heterochromatin. If informative, we will expand the analysis to a genome-wide scale by combining ChIP with massive parallel sequencing (ChIP-Seq). This technique is currently employed in the laboratory of Professor Malcolm Parker (IRDB, Imperial), who has kindly agreed to offer technical assistance as well bioinformatic support. Finally, the translational potential of informative epigenetic modifications at specific loci will be tested on DNA extracted from blinded, whole-tissue endometrial biopsies from RPL and non-RPL patients.

(iii) hCG signalling and endometrial responses in RPL

Rationale. Paradoxical endometrial responses to embryonic hCG signalling may constitute a major mechanism in persistent pregnancy wastage. hCG signalling is mediated by the hCG/LH (luteinizing hormone) receptor (hCG/LHR) but the downstream signalling pathways in endometrial cells are, however, largely unknown. Our preliminary data indicate that the endometrial hCG/LHR is coupled to a G-protein signalling pathway not previously reported to be activated by this receptor. hCG treatment triggered rapid activation of Gαi in both undifferentiated and decidualized HESCs, resulting in inhibition of Forskolin-induced cAMP production that is reversed by the Gαi inhibitor pertussis toxin. The hCG/LHR also activates calcium signalling in a subpopulation of HESCs, though only in undifferentiated cells. Further, several lines of evidence indicate that the hCG/LHR exhibits constitutive activity in undifferentiated cells: pertussis toxin treatment increases the unstimulated cAMP response over Forskolin treatment alone, the immuno-reactive hCG/LHR migrates at higher molecular weight on SDS-PAGE (possibly reflecting phosphorylation) and exhibits ligand-independent internalisation. This constitutive activity seems to be reversed upon decidualization of HESCs. Such a change in activity has not been observed for a GPCR in any system. Based on these observations, we propose that constitutive activation of the receptor may serve as a mechanism to prevent inappropriate expression of cAMP-responsive genes in undifferentiated cells whereas reprogramming of the cellular environment upon decidualization stabilizes the receptor on the cell surface, thereby preparing the endometrium for embryonic signals upon successful implantation. To test this hypothesis, we will perform gene expression studies to define the transcriptional responses to hCG in decidualizing HESCs from RPL and non-RPL patients, complemented by an in-depth characterization of the mechanisms and pathways involved in hCG signal transduction. This combined approach may uncover novel targets that could be exploited for the prevention of early pregnancy loss.

Experimental approach.

(a) Characterization of hCG responses in decidualizing HESCs from RPL and controls. hCG regulation of decidual marker genes, such as PRL and PROK1, in RPL and control cultures will be explored further in time-course and dose-response experiments by RTQ-PCR to determine the most informative conditions for subsequent genome-wide expression profiling. The design of microarray analysis will be similar to that outlined in section (iia). To define the importance of Gi signalling via hCG/LHR in regulating decidual genes, such as PRL and novel informative genes identified in the array, the transcriptional response to hCG in decidualizing HESCs willl be examined upon co-treatment with Pertussis toxin and upon siRNA-mediated knockdown of the hCG/LHR. Subsequent characterization of hCG/LHR and mechanism of signal transduction will be carried out first in control HESC cultures and the key findings tested in primary cells from patients with RPL.

(b) Characterization of hCG/LHR in HESCs. Western blot analysis indicated that the predominant hCG/LHR form expressed in HESCs differs from the mature cell surface receptor expressed in other cell types, which in turn suggests either differential glycosylation of the receptor or expression of a splice variant. To discriminate between these possibilities, we will first treat HESC lysates with the endoglycosidases EndoH and PNGase F that differentiate between endoplasmic reticulum and cell surface receptor, respectively. Four different hCG/LHR splice variants have been described, including variants missing exons encoding for the N-terminal domain of the receptor and severely truncated forms, missing part of the hinge region, transmembrane domains, and C-terminal tail. Expression of hCG/LHR splice variants at the mRNA level will be determined by RT-PCR. The cDNA encoding a receptor variant will be cloned into an expression vector (pCDNA3.1), transfected in a heterologous cell system (e.g. HEK293) and functionally assessed, either alone or in combination with the wild-type receptor to determine possible dominant-negative function of the variant.

(C) Characterization of hCG/LHR signal complex in HESCs. To characterize the interacting molecular machinery that govern hCG/LHR activity in HESCs, we will use a proteomic approach. Endogenous receptor will be immunoprecipitated (IP) from undifferentiated and decidualizing HESCs. If insufficient yield, we will express a FLAG-tagged endometrial hCG/LHR and IP with anti-FLAG antibodies. Both 1D and 2D gel approaches will be utilised to identify changes in the proteins associated with the LHR. Bands/spots of interest will be sent to the Glycotric/Mass Spec Facility (Imperial College) for analysis by both MALDI-ToF and LC/MS/MS (Q-ToF) mass spectrometry. Proteins interactions will be confirmed using co-IP and bioluminescence resonance energy transfer (BRET). BRET measures real-time interactions between donor-tagged (Renilla Luciferase) proteins with acceptor-tagged (YFP) proteins and is more sensitive than FRET in detecting low level and/or transient interactions. The proteomics screen will complemented by functional analysis of known hCG/LHR interacting partners, such as the PDZ protein GIPC. PDZ proteins provide a cell plasma membrane scaffolding function and may act to stabilise the hCG/LHR at the surface in decidualizing cells. The cellular distribution of GIPC in relation to the hCG/LHR will be visualized by confocal microscopy in undifferentiated and decidualized cells. Functionally, the effect of siRNA-mediated knockdown of GIPC on hCG/LHR activity will be assessed by measurement of hCG-induced Gi signalling (EIA cAMP kit, Assay Designs) and on constitutive receptor trafficking.

(d) hCG/LHR trafficking in HESCs. Membrane trafficking of GPCRs critically defines cellular signalling patterns. Agonist-induced internalisation contributes to rapid signal desensitisation, resensitisation or recovery of hormonal signalling, sorting of receptors to the lysosome for downregulation and even activation of non-G protein signalling pathways from the endosome. The lack of hCG-induced internalisation in decidualizing HESCs suggests that receptor may be resistant to desensitisation, perhaps as a means to maintain hCG-induced signalling in the early stages of pregnancy and/or to define a pattern of signalling that regulates downstream gene expression. The ability of hCG to induce rapid receptor desensitisation will be measured by its ability to signal following repeated challenges of hCG. Furthermore, the ability of hCG to induce long-term downregulation in HESCs under chronic hCG treatment will be determined by Western blot analysis. Resistance to internalisation, and perhaps down-regulation, in decidualizing cells suggest that liganded receptor does not recruit endocytic adaptors such as arrestin. This can be examined by measuring expression levels of arrestin and the GPCR kinases (GRKs) in undifferentiated and decidualizing HESCs and by monitoring the effect of overexpression of these proteins on hCG/LHR desensitisation/internalisation and signalling.

Taken together these experiments will represent the first in-depth analysis of hCG signalling in HESCs and the comparison of hCG/LHR function in RPL versus non-RPL endometrial cells may uncover treatment targets for the prevention of early pregnancy loss.