Disrupted PGR-B and ESR1 signaling underlies preconceptional defective decidualization linked to severe preeclampsia

Decidualization of the uterine mucosa drives the maternal adaptation to invasion by the placenta. Appropriate depth of placental invasion is needed to support a healthy pregnancy; shallow invasion is associated with the development of severe preeclampsia (sPE). Maternal contribution to sPE through failed decidualization is an important determinant of placental phenotype. However, the molecular mechanism underlaying the in vivo defect linking decidualization to sPE is unknown. Here, we discover the footprint encoding this decidualization defect comprising of 166 genes using global gene expression profiling in decidua from women who developed sPE in a previous pregnancy. This signature allowed us to effectively segregate samples into sPE and control groups. Estrogen receptor 1 (ESR1) and progesterone receptor B (PGR-B) were found highly interconnected with the dynamic network of defective decidualization fingerprint. ESR1 and PGR-B gene expression and protein abundance were remarkably disrupted in sPE. Thus, the transcriptomic signature of impaired decidualization implicates dysregulated hormonal signaling in the decidual endometria in women who developed sPE. These findings reveal a potential footprint that may be leverage for a preconception or early prenatal screening of sPE risk, thus improving prevention and early treatments.


1
Preeclampsia (PE) is a severe complication of late pregnancy and is the second leading cause of 3 2 maternal mortality in the US, affecting 8% of first-time pregnancies (1). PE is characterized by the onset of 3 3 hypertension, proteinuria, and other signs of maternal vascular damage that contributes to neonatal mortality 3 4 and morbidity (1). Severe preeclampsia (sPE) is diagnosed based on elevated blood pressure (systolic ≥ 160 3 5 or diastolic of ≥ 100 mm Hg) or thrombocytopenia, impaired liver function, progressive renal insufficiency, 3 6 pulmonary edema, or the onset of cerebral or visual disturbances (2). sPE is a placental insufficiency 3 7 syndrome mediated by early deficient extravillous trophoblasts (EVTs) invasion of uterine decidua and 3 8 spiral arterioles, leading to incomplete endovascular invasion and altered uteroplacental perfusion (3-5). 3 9 Why shallow EVTs invasion occurs, however, remains to be determined (6). 4 0 4 1 Pregnancy health is dictated by the embryo, placenta, and the quality of the maternal decidua, where 4 2 EVTs invasion and remodeling of maternal spiral arteries occur (7,8). Accumulated evidence suggests that 4 3 the contribution of the decidua to the etiology of PE (9), sPE (10-12), and placenta accreta (13) is 4 4 significant, and cellular signaling in the decidua may determine whether these conditions develop.

5
Decidualization is the remodelling of the endometrium initiated after ovulation necessary for adequate 4 6 trophoblast invasion and subsequent placentation (14). Defective decidualization entails the inability of the 4 7 endometrial compartment to undertake tissue differentiation, leading to aberrations in placentation and 4 8 compromising pregnancy health (12). 4 9 5 0 In humans and other great apes, the formation of the decidua is a conceptus-independent process 5 1 driven by progesterone and the second messenger cyclic adenosine monophosphate (15) that stimulates 5 2 synthesis of a complex network of intracellular and secreted proteins through progesterone receptor 5 3 activation . Endometrial decidualization involves secretory transformation of uterine glands (16), influx of 5 4 specialized immune cells, vascular remodeling, and morphological (17, 18), biochemical (19, 20), and 5 5 transcriptional reprogramming of the endometrial stromal compartment (21). We recently characterized the 5 6 transcriptomics of human decidualization at single-cell resolution from secretory endometrial samples and 5 7 and validation (n=11) set of samples. The training set of samples was analyzed by RNA-seq to identify the 0 0 global transcriptomic profiling changes between control (n=12) and sPE (n=17) samples. Selection criteria 0 1 were applied to define a transcriptomic fingerprinting associated with DD detected in sPE. Finally, targeted 0 2 analysis of the DD signature was validated in the test set composed of controls (n=4) and sPE (n=7). 0 3 0 4 Human donors 0 5 Endometrial samples were collected from women aged 18-42 without any medical condition who had been 0 6 pregnant 1-8 years earlier. All participants had regular menstrual cycles (26-32 days) with no underlying 0 7 gynecological pathologic conditions and had not received hormonal therapy in the 3 months preceding 0 8 sample collection. After the inclusion criteria were applied, endometrial biopsies were obtained by pipelle 0 9 catheter (Genetics Hamont-Achel, Belgium) under sterile conditions in the late secretory phase (cycle days 1 0 Total RNA from endometrial biopsies was isolated using QIAsymphony RNA kit (Qiagen, Hilden, 1 7 Germany) following the manufacturer's protocol. RNA concentrations were quantified using a Multiskan 1 8 GO spectrophotometer (Thermo Fisher Scientific, Waltham, US) at a wavelength of 260 nm. Integrity of the 1 9 total RNA samples was evaluated using an Agilent 4200 TapeStation system (Agilent Technologies Inc., 2 0 Santa Clara, CA) and the samples with RNA integrity were used for the global RNA-seq. 2 1 2 2 Global RNA-seq library preparation and transcriptome sequencing 2 3 cDNA libraries from total RNA samples (n=40) were prepared by an Illumina TruSeq Stranded mRNA 2 4 sample prep kit (Illumina, San Diego, CA). Three micrograms of total RNA were used as the RNA input 2 5 according to the manufacturer's protocol. mRNAs were isolated from the total RNAs by purifying the poly-2 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 24, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 A containing molecules using poly-T oligo attached to magnetic beads. The RNA fragmentation, first and 2 7 second strand cDNA syntheses, end repair, single 'A' base addition, adaptor ligation, and PCR amplification 2 8 were performed according to the manufacturer's protocol. The average size of the cDNA libraries was 2 9 approximately 350 bp (including the adapters). cDNA libraries were validated for RNA integrity and 3 0 quantity using an Agilent 4200 TapeStation system (Agilent Technologies Inc., Santa Clara, CA) before 3 1 pooling the libraries. The pool concentration was quantified by qPCR using the KAPA Library 3 2 Quantification Kit (Kapa Biosystems Inc.) before sequencing in a NextSeq 500/550 cartridge of 150 cycles 3 3 (Illumina, San Diego, CA). Indexed and pooled samples were sequenced 150-bp paired-end reads by on the 3 4 Illumina NextSeq 500/550 platform according to the Illumina protocol. 3 5

6
Quality control and pre-processing data 3 7 Once sequencing was completed, we checked for outliers or any technical issue in the raw data (sequence 3 8 quality, read alignment, quantification, and reproducibility). After the initial quality assessment, pre-3 9 processing analyses were performed over the read counts data (low-count filter and normalization). All sequences were pre-processed, normalized, and analyzed comparing sPE (n=17) to controls (n=12) from 4 5 training set. We applied the trimmed mean of M-values normalization strategy to our gene expression values 4 6 (54). The Bioconductor package edgeR (54) in R software was used. The p-value adjustment method was 4 7 FDR with a cut-off of 0.05 and the minimum absolute log2-fold-change required was 1 (FC ≥ 2). A volcano 4 8 plot was created to visualize DEGs. For a better overview, we distinguished between DEGs with a high or 4 9 low fold-change. In this case, the threshold at the legend indicates: none (do not have DEGs); p-value 5 0 (DEGs, but low fold-change: smaller than 1 in absolute value); FC (high fold-change, but not DEGs) and p-5 1 value-FC (both DEGs and high fold-change: log2-fold-change greater than 1 in absolute value). 5 2 5 3 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2021. ; https://doi.org/10. 1101/2021 Transcriptomic fingerprinting definition and validation 5 4 Genes with assigned EntrezID with an FDR cut-off of 0.05 and an expression ≥ 4-fold higher in the sPE vs. 5 5 control training set samples were selected to define a fingerprint associated with DD in sPE. Targeted  5  6 analysis of fingerprinting genes was performed using the validation set of samples. PCA and unsupervised 5 7 hierarchical clustering with a Camberra distance based on gene signature were performed comparing sPE to 5 8 control specimens. 5 9 6 0 Enrichment analysis 6 1 Biological processes in which DEGs are involved were studied. In edgeR, GO analyses can be conducted 6 2 using goana. An FDR cutoff of 5% is used when extracting DE genes and for logFC, we used a cut-off 6 3 value of 1 [UP, logFC>1 and DOWN; logFC<(-1)]. The ontology domain that GO term belongs to is 6 4 biological process (BP). Because the p-values obtained are not adjusted for multiple testing, we ignored GO 6 5 terms with p-values greater than about 0.005. 6 6 6 7 Interaction network 6 8 An interaction network between proteins encoded by DD fingerprinting genes was created using the 6 9 functional analysis suite String (28). To construct the network, the interactions included were from curated 7 0 databases and included experimentally determined and predicted interactions, text mining, co-expression 7 1 information, and protein homology. The clustering algorithm k-means was applied based on the distance 7 2 matrix obtained from the String global scores. The network was visualized using Cytoscape software (29). 7 3 7 4 qRT-PCR gene validation 7 5 To validate our transcriptomic results, a selection of differentially expressed genes was validated by qRT-7 6 PCR in a subgroup of samples from the experimental cohort [controls (n=9) and sPE (n=14)]. Specific 7 7 primers for each gene are described in supplementary file 5. cDNA was generated from 400 ng of RNA 7 8 using the SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, US). Template 7 9 cDNA was diluted 5 in 20 and 1 µL was used in each PCR. Real-time PCR was performed in duplicate in 10 8 0 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2021. ; https://doi.org/10.1101/2021.07.22.21260977 doi: medRxiv preprint µL using commercially validated Kapa SYBR fast qPCR kit (Kapa biosystems Inc, Basilea, Switzerland) 8 1 and the Lightcycler 480 (Roche Molecular Systems, Inc, Pleasanton, CA) detection system. Samples were 8 2 run in duplicate along with appropriate controls (i.e., no template, no RT). Cycling conditions were as 8 3 follows: 95°C for 3 min, 40 cycles of 95°C for 10 s, 60°C for 20 s, and 72°C for 1 s. A melting curve was 8 4 done following the product specifications. Data were analyzed using the comparative Ct method (2−∆∆CT). 8 5 Data were normalized to the housekeeping gene β -actin, changes in gene expression were calculated using 8 6 the Δ Δ CT method with the control group used as the calibrator; values are illustrated relative to median in 8 7 the control group. The relative expression of PGR-A mRNA was calculated by subtracting the relative 8 8 expression of PGR-B mRNA from that of PGR total. 8 9 9 0 Immunofluorescence of tissue sections 9 1 Endometrial tissue samples were fixed in 4% paraformaldehyde and preserved in paraffin-embedded blocks. 9 2 For immunostaining, tissue sections were deparaffinated and rehydrated. Antigen retrieval was performed 9 3 with buffer citrate 1x at 100°C for 10 min. Then, non-specific reactivity was blocked by incubation in 5% 9 4 BSA/0.1% PBS-Tween 20 at room temperature for 30 min. Sections were incubated at room temperature for 9 5 1.5 h with primary antibodies (1:50 rabbit monoclonal anti-human progesterone receptor, Abcam, 9 6 Cambridge, UK) and 1:50 mouse monoclonal anti-human estrogen receptor 1 (Santa Cruz Biotechnology, 9 7 CA, USA) diluted in 3% BSA/0.1% PBS-Tween 20. Then, slides were washed two times for 10 min with 9 8 0.1% PBS-Tween 20 before they were incubated for 1 h at room temperature with AlexaFluor-conjugated 9 9 secondary antibodies diluted in 3% BSA/0.1% PBS-Tween 20. Finally, slides were washed two times in 0 0 0.1%PBS-Tween 20. To visualize nuclei, 4′,6-diamidino-2-phenylindole at 400 ng/uL was used. Tissue 0 1 sections were examined using a EVOS M5000 microscope. 0 2 0 3

Statistical analysis 0 4
Clinical data are expressed as mean ± standard error mean (SEM). Clinical data were evaluated by Student's 0 5 t-test for comparisons between sPE and control samples. Statistical significance was set at p ≤0.05. 0 6 Differential expression analysis was performed using the R package edgeR. 0 7 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

1
. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2021. ;  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2021. ;  G  a  l  l  e  g  u  i  l  l  o  s  C  ,  T  o  r  r  e  s  M  ,  S  o  v  i  n  o  H  ,  F  u  e  n  t  e  s  A  ,  B  o  r  i  c  M  A  ,  e  t  a  l  .  A  s  s  o  c  i  a  t  i  o  n  b  e  t  w  e  e  n  M  M  P  1  3  8  a  n  d  M  M  P  9  a  c  t  i  v  i  t  i  e  s  a  n  d  I  C  A  M  1  c  l  e  a  v  a  g  e  i  n  d  u  c  e  d  b  y  t  u  m  o  r  n  e  c  r  o  s  i  s  f  a  c  t  o  r  i  n  s  t  r  o  m  a  l  c  e  l  l  c  u  l  t  u  r  e  s  f  r  o  m  3  9  e  u  t  o  p  i  c  e  n  d  o  m  e  t  r  i  a  o  f  w  o  m  e  n  w  i  t  h  e  n  d  o  m  e  t  r  i  o  s  i  s  .  R  e  p  r  o  d  u  c  t  i  o  n  .  2  0  0  9  ;  1  3  8  (  5  )  :  8  3  7  -4  7  .  4  0  4  8  .  B  é  g  u  i  n  E  P  ,  v  a  n  d  e  n  E  s  h  o  f  B  L  ,  H  o  o  g  e  n  d  i  j  k  A  J  ,  N  o  t  a  B  ,  M  e  r  t  e  n  s  K  ,  M  e  i  j  e  r  A  B  ,  e  t  a  l  .  I  n  t  e  g  r  a  t  e  d  4  1  p  r  o  t  e  o  m  i  c  a  n  a  l  y  s  i  s  o  f  t  u  m  o  r  n  e  c  r  o  s  i  s  f  a  c  t  o  r  α  a  n  d  i  n  t  e  r  l  e  u  k  i  n  1  β  -i  n  d  u  c  e  d  e  n  d  o  t  h  e  l  i  a  l  i  n  f  l  a  m  m  a  t  i  o  . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 24, 2021. ; https://doi.org/10. 1101/2021