Uterine surgical scarring is an increasing risk factor for adverse pregnant consequences that threaten fetal-maternal health. The detailed molecular features of scar implantation remain largely unknown. We aim to study the pathologic features of uterine surgical scarring and the mechanisms of compromised pregnancy outcomes of scar implantation. We generated a mouse model of uterine surgical scarring with a uterine incision penetrating the myometrium to endometrium to examine the pathologic changes and transcriptome profiles of uterine scarring at various post-surgery (PS) time points, as well as features of the feto-maternal interface during scar implantation. We found that uterine surgical scar recovery was consistently poor at PS3 until PS90, as shown by a reduced number of endometrial glands, inhibition of myometrial smooth muscle cell growth but excessive collagen fiber deposition, and massive leukocyte infiltration. Transcriptome annotation indicated significant chronic inflammation at the scarring site. At the peri-implantation and postimplantation stages, abnormal expression of various steroid-responsive genes at the scarring site was in parallel with lumen epithelial cell hyperplasia, inappropriate luminal closure, and disorientation of the implanted embryo, restricted stromal cell proliferation, and defective decidualization. High embryonic lethality (around 70%) before E10.5 was observed, and the small amount of survival embryos at E10.5 exhibited restricted growth and aberrant placenta defects including overinvasion of trophoblast cells into the decidua and insufficient fetal blood vessel branching in the labyrinth. The findings indicate that chronic inflammation and compromised responses to steroids in uterine scar tissues are the pivotal molecular basis for adverse pregnancy consequences of scar implantation.

Materials and methods

Mouse model of uterine surgical scar

Animal experiments were approved by the Experimental Animal Welfare Committee in the Institute of Zoology, Chinese Academy of Sciences. Female virgin CD1 mice (8 weeks) were purchased from Beijing SiPeiFu Animal Center (China) and housed at 25℃ with 12/12 hours day–night light cycles. Uterine surgery was performed under sterile conditions. Briefly, the mice were anaesthetized by intraperitoneal injection of 2% 2,2,2-Tribromoethanol (100ul/10g body weight), and a low abdominal incision was made to expose uterine horns. Under stereoscope, a 1cm-in-length longitudinal incision at the mid-portion of anti-mesometrial side of uterine horn was made by using ophthalmic scissors. The incision penetrated through myometrium and endometrium of anti-mesometrial side, but not harmed tissue of mesometrial side (Supplementary Figure S1A (13)). The incision was closed with 8-0 absorbable sutures using one-layer closure technique with interrupted lock stitches. To mark the incision, both ends of the incision edge were labeled with 8-0 non-absorbable sutures. The surgery was performed on one random horn, allowing the other horn as un-surgical control. The uterus was placed back into the abdominal cavity, and the abdominal skin was closed using wound clips. The Sham control animals had operations of low abdominal incision but no uterine incision.

Following 3, 7, 20, 60 and 90 days of surgery, the mice at diestrus stage (at least 10 mice for each time point) were sacrificed and the uterine tissues from the scarring site (1cm in length) or Sham control were collected under stereoscope for histological examinations and/or RNA measurement as described below.

Following surgery, the mice were permitted to recover for 20, 60 or 90 days and mated to fertile male mice. The day on which virginal plug appeared was recorded as embryonic day 0.5 (E0.5). The pregnant mice were sacrificed at E3.5, E4.0, E5.0, E7.5, or E10.5 (at least 5 mice for each time point) to obtain uterine tissues from the scarring site (1cm in length) or Sham control for histological analysis and RNA measurement as described below.

Preparation of paraffin section and frozen section

For paraffin section, the freshly collected tissues were fixed in 4% paraformaldehyde (PFA), followed by routine gradient dehydration with ethanol and embedding in paraffin wax. Paraffin sections of 5μm in thickness were prepared using paraffin microtome (Leica, RM2245).

For frozen section, the fresh tissues were subjected to fixation in 4% PFA, dehydration with sucrose gradient, and embedding in O.C.T. compound (Sakura Finetek, Torrance). Frozen sections of 10 μm in thickness were prepared using frozen microtome (Leica, CM1950).

Histological analysis and immunohistochemistry

Paraffin sections were routinely deparaffinized and rehydrated, and subjected to Haematoxylin-eosin (H&E) staining or Masson’s trichrome staining (Leagene, DC0033 -50) according to the manufacturer’s instructions. For immunohistochemistry, the rehydrated paraffin sections were retrieved by citric acid, and incubated in 3% H₂O₂ to inactivate endogenous peroxidase. The sections were incubated with primary antibodies followed by HRP-conjugated second antibody (Table S1 (13)). Antibodies from Abcam (Cambridge, Massachusetts) included α-SMA (ab5694, Abcam)(14), CK(ab181598, Abcam)(15), CD45 (ab10558, Abcam)(16), CD3 (ab11089, Abcam)(17), Ki67 (ab15580, Abcam)(18), and MUC1 (ab15481, Abcam)(19). Antibodies from Cell Signaling Technology (Danvers, Massachusetts) included E-cadherin (3195S, CST)(20), COX2 (12282S, CST)(21) and PR (8757S, CST)(22). Antibodies from ZSGB-Bio (Beijing) included Rat IgG-HRP (PV-9004, ZSGB-Bio)(23) and goat IgG-HRP (ZB-2306, ZSGB-Bio)(24) and Rabbit IgG-HRP (PV-9001, ZSGB-Bio)(25). Antibodies from Novus Biologicals (Denver, Colorado) included F4/80 (NB600-404, Novus)(26) and NKp46/NCR1 (AF2225, RD)(27). Antibody from Santa Cruz Biotechnology (Dallas, Texas) was ERα (sc-7207, Santa Cruz)(28). Antibody from Sigma-Aldrich (Allentown, Pennsylvania) was Laminin (L9393, Sigma)(29). Followed by visualization with DAB (ZSGB-BIO, ZLI-9019) as substrate and counterstaining with hematoxylin. The results were recorded with light microscope (Olympus) and analyzed with Image-Pro Plus software (Media Cybernetics, USA).

Immunofluorescent analysis was performed using frozen sections, with primary antibodies anti-Bmp2 (bs-1012R, Bioss)(30) followed by FITC-labeled secondary antibody (A21206, Life technologies)(31) (Table S1 (13)) and 4',6-diamidino-2-pheny- lindole dihydrochloride (DAPI; Sigma, 28718-90-3) for nuclear staining. The results were recorded with confocal microscope (Zeiss, LSM780) and processed by ZEN 2012 software (Carl Zeiss Microscopy GmbH, Germany).

Artificial decidualization and in vitro decidualization

Artificial decidualization was carried out as previously described(32). In brief, following 20 days of surgery, the mice were mated with vasectomized male, and the day on which virginal plug appeared was recorded as pseudopregnancy day 1 (Day 1). Mechanical stimulation by intraluminal injections of sesame oil (20µl/mouse) in one uterine horn was performed in pseudopregnancy mice on Day 4. The mice were killed on Day 6, Day 7, or Day 8 to collect uterine tissues from the scarring site (1cm in length) or Sham control for further analysis.

The stromal cells isolation and in vitro decidualization analysis were carried out as previously described(33) with slight modification. In brief, the collected uterine tissues were slit longitudinally and disaggregated with HBSS containing 6 mg/ml dispase (Gibco, 17105041), 25 mg/ml trypsin (Sigma, T8003), followed by further incubation in HBSS containing 0.5 mg/ml collagenase (Gibco, 17018-029). The supernatants were filtered through 200 mesh gauze filter and centrifuged to collect stromal cells. The stromal cells were suspended in phenol-red free DMEM/F-12 (Invitrogen, 11039021) with 10% hormone-free fetal bovine serum (BSA; SenBeiJia, Sbjser02), and seeded in 6-well plate at 4×10⁵ cells/well. Decidualization of stromal cells were induced by adding 1µM medroxyprogesterone 17-acetate (Sigma, M1629), 10 nM β-estradiol (Sigma, E2758), and 10 µM 8-bromoadenosine 3′,5′-cyclic monophosphate (8-Br-cAMP; Sigma, B5386), and the cultural media were changed every 2 days.

In situ hybridization (ISH)

ISH was performed as previously described with slight modification(34). In brief, digoxin labelled riboprobe for mouse Bmp2 gene was synthesized according to the manufacturer’s instruction (Roche, 11175025910, Indianapolis, IN, USA). Frozen sections were hybridized with digoxin labelled riboprobe, followed by incubation with AP-conjugated anti-digoxin antibody (Roche, Indianapolis, IN, USA) and visualization with BCIP/NBT (Promega, Madison, WI, USA) as substrate. Fast red for nucleus staining was performed before mounting. The results were recorded with light microscope (Olympus) and analyzed with Image-Pro Plus software (Media Cybernetics, USA).

Transcriptome profiling and bioinformatical analysis

At 3, 7, and 20 days following uterine surgery, uterine endometrium and myometrium were separately collected as described previously (35,36). Briefly, mice at diestrus stage were sacrificed and the uterine tissue from scar site or Sham control was collected and dissected under stereoscope. The endometrium specimen was obtained with rubber scraper, with the remaining part being myometrium (Figure.S1B (13)). Considering the limited amount of tissue at scaring site from one individual mouse, endometrial or myometrial tissues from 10 animals in each study group were mixed as one sample.  Total RNA was extracted using Total RNA kit (Omega Bio-Tek, Norcross, GA, USA) and treated with RNase-free DNase (TaKaRa, Otsu, Japan). An amount of 1 μg RNA per sample was subjected to sequencing library generation using NEBNext UltraTM RNA Library Prep Kit for Illumina (NEB, USA) according to the manufacturer’s instruction. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v4-cBot-HS (Illumia) according to the manufacturer’s instructions. After cluster generation, the library preparations were sequenced on an Illumina Hiseq Xten platform and paired-end reads were generated. Transcriptome sequencing data were quantitated at a protein-coding mRNA level using the RNA-seq quantitation pipeline in SeqMonk software. Differential expression genes were revealed by the DESeq R package (R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org/). The differentially expressed genes were determined by statistical analysis of FDR﹤0.05 and fold change ≥2. Heatmaps and KEGG were generated using Omicshare software (https://www.omicshare.com/tools/Home/Soft/ pathwaygsea). GO enrichment were analyzed using the AmiGO2 database (http://amigo.geneontology.org/amigo) and mapped with Cytoscape (http://www.cyto- scape.org/) to illustrate complex network regulation.

Real time quantitative PCR analysis

Real time PCR was performed as described(37). Briefly, total RNA was isolated from tissues or cells with Trizol reagent following the manufacturer’s protocol (Ambion, 15596-018). A total of 2 µg RNA was used to synthesize cDNA, and real time PCR analysis was carried out by using Roche LightCycler 480 system. The primers were used in this study listed in Table S2 (13).

Statistical analysis

Statistical analyses were performed with GraphPad Prism 6.0 software (GraphPad Software Inc, USA). Data were expressed as Mean±S.E.M. based on at least three independently repeated experiments. Statistical comparison between groups was performed using non-paired t-test, and P-values of less than 0.05 were considered statistically significant.