Abstract

   Postpartum hemorrhage (PPH) is the leading cause of maternal mortality
   worldwide. However, the mechanism underlying atonic PPH remains
   partially elucidated. Multi-omics revealed that differentially
   expressed proteins and metabolites were enriched in the
   immune-inflammation pathway in the vaginal blood of patients with
   atonic PPH. There was a pro-inflammatory immune microenvironment
   primarily activated by M1 macrophages in the decidua of the patients
   with atonic PPH, which presented as increased tumor necrosis factor
   (TNF)-α, interleukin (IL)-6, and IL-8 levels and affected the
   contraction of the uterine smooth muscle. Besides, the decidual
   macrophage of the atonic PPH group exhibited increased oxidative
   stress. The PPH decidual cell culture medium induced the polarization
   of peripheral blood monocytes towards M1 macrophages while markedly
   increasing the levels of reactive oxygen species and superoxide anion
   radical. Using hydrogen peroxide (H[2]O[2]) to stimulate decidual
   macrophages induced a similar polarization state to that in atonic PPH
   samples, and the secretion of pro-inflammatory cytokines, such as TNF-α
   and IL-8, was significantly upregulated, which markedly impacted the
   expression of contraction-associated proteins (CAPs) in the uterine
   smooth muscle cells (uSMCs). The animal model suggested that H[2]O[2]
   promoted the polarization of placental macrophages towards M1,
   affecting the levels of placental oxidative stress and inflammatory
   infiltration, and the contractility of uterine smooth muscle tissues.
   In summary, abnormal oxidative stress at the maternal-fetal interface
   induced the M1 polarization of decidual macrophages, causing the
   secretion of pro-inflammatory cytokines. TNF-α and IL-8 acted on uSMCs
   to inhibit CAP expression, inducing atonic PPH.

   Keywords: Atonic postpartum hemorrhage, Macrophage polarization,
   Oxidative stress, Cytokines, Uterine contraction-associated proteins

1. Introduction

   Postpartum hemorrhage (PPH) is a common complication in obstetrics and
   is defined as blood loss exceeding 500 mL within 24 h of vaginal
   delivery [[37]1]. PPH is the leading cause of pregnancy-related deaths
   and accounts for approximately a quarter of maternal deaths
   [[38]2,[39]3], which is correlated with the occurrence of adverse
   clinical outcomes [[40]1]. Uterine atony is the most common cause of
   PPH (approximately 70 %) [[41]4]. However, low-risk pregnant women
   experience atonic PPH in clinical settings. The vague mechanism and
   unpredictability of PPH cause delayed treatment and an increased risk
   of adverse clinical outcomes. Therefore, elucidating the mechanism of
   atonic PPH is crucial for identifying effective treatment measures.

   As the uterus remodels in preparation for delivery, the excitability
   and contractility of the uterine smooth muscle layer, the myometrium,
   increases drastically [[42]5]. At term, the increased secretion of
   pro-inflammatory cytokines (interleukin (IL)-8, tumor necrosis factor
   (TNF)-α, IL-1β, and IL-6) promote increased expression of
   cyclooxygenase-2 (COX-2) and oxytocin receptor (OTR) in myometrium;
   this results into strong and frequent uterine contractions, which
   subsequently promote the onset of labor [[43]6]. The
   immune-inflammatory reaction at the maternal–fetal interface is crucial
   in this process, with decidual immune cells playing a crucial role in
   regulating the process. Decidual macrophages are the second largest
   immune cells at the maternal-fetal interface, accounting for 10–20 %
   [[44]7]. Decidual macrophages transform into a pro-inflammatory
   phenotype at the onset of labor, inducing myometrium contraction and
   labor initiation [[45]8]. Simultaneously, the levels of
   pro-inflammatory cytokines (TNF-α, IL-6, IL-1β, and C C motif ligand 2
   (CCL-2) are positively correlated with uterine contractility [[46][9],
   [47][10], [48][11], [49][12], [50][13]].

   However, some studies revealed that prolonged exposure to
   pro-inflammatory cytokines, including TNF-α and IL-1, could cause
   impaired oxytocin signaling and depletion of calcium ions in muscle
   cells, affecting uterine muscle contraction [[51]10,[52]11]. This
   suggested that excessive activation of immune-inflammatory responses
   might inhibit myometrium contraction. Previous studies demonstrated
   that the levels of immune-inflammatory factors, such as IL-1β, IL-18,
   macrophage inflammatory protein (MIP)-1α, IL-6, and monocyte
   chemotactic protein (MCP)-1, in the peripheral blood of pregnant women
   before delivery, who had atonic PPH, were significantly higher than
   those in the control group [[53]4,[54]14]. Consequently, the abnormal
   pro-inflammatory phenotype of decidual macrophages might affect the
   contractile function of uterine smooth muscle cells (uSMCs). The
   uterine blood flow at term pregnancy can be as high as 1300 ml/min.
   When the myometrium cannot contract in a strong and sustained manner,
   it fails to timely compress the spiral arteries that supply the
   placental bed, resulting in a rapid hemorrhage from the placental bed,
   culminating in the occurrence of PPH [[55]15].

   Previous studies have shown that the macrophage phenotype was closely
   related to their oxidative stress state. Mitochondrial damage and the
   release of excess oxidative substances, such as reactive oxygen species
   (ROS), were crucial for the release of pro-inflammatory cytokines in
   monocytes and macrophages [[56]16]. Therefore, we aimed to investigate
   the relationship between immune-inflammatory and oxidative stress
   states in decidual macrophages at the maternal-fetal interface in
   atonic PPH. Similarly, we combined omics and multiplex Luminex assays
   to validate the promotion of M1 macrophage polarization and
   pro-inflammatory cytokine release through an abnormal oxidative stress
   state in decidual macrophages and the regulation of uSMC contraction to
   reveal the potential mechanism of atonic PPH which was confirmed
   through in vitro cell experiments and rat models.

2. Materials and methods

2.1. Inclusion criteria of clinical sample

   This was a nested case-control study on a prospective cohort of
   pregnant women who underwent regular prenatal checkups at Peking
   University's Third Hospital between December 2022 and December 2023.
   The inclusion criteria were natural conception, full-term pregnancy,
   and single pregnancy through vaginal delivery. The exclusion criteria
   were PPH secondary to definite causes (such as placenta accreta, soft
   birth canal lacerations, underlying coagulation disorder, previous
   autoimmune diseases, and recent acute and chronic intrauterine or
   systemic infections), obstetric complications (including preeclampsia
   and gestational diabetes mellitus (GDM) other than type A1), and
   uterine atony with definite causes (such as macrosomia, multiple
   pregnancies, and precipitous deliveries). Following the above criteria,
   30 pregnant women with PPH were included. The control (non-PPH) group
   was matched with baseline information (including maternal age,
   gestational age, and pregnancy complications, such as hypertensive
   disorders in pregnancy [HDP] and GDM) without PPH. All procedures were
   approved by the Medical Science Research Ethics Committee of Peking
   University Third Hospital (M2021685).

2.2. Clinical sample collection

   Upon entering the first stage of labor (the interval between the onset
   of labor and complete or 10 cm cervical dilation), 5 mL of peripheral
   blood was collected in an Ethylenediaminetetaracetic acid (EDTA) tube.
   At the end of the third stage of labor, after placenta delivery, 5 mL
   of vaginal blood was collected in an EDTA tube. Both blood samples were
   centrifuged at 12,000×g and 4 °C for 10 min and stored at −80 °C.

   Following the delivery of the placenta, 3 × 3 × 1 cm decidua and
   placental tissue were sectioned and fixed in 10 % neutral formalin.
   Subsequently, 5–10 g of decidua was scraped from the maternal surface
   of the placenta and washed with phosphate-buffered saline. The decidua
   tissue was then cut and digested in 100U/mL collagenase Ⅱ and Ⅳ, and
   100U/mL DNAase and placed on a constant temperature shaker at 37 °C for
   40 min. The tissue was then ground on a 100 μm sieve. The resultant
   cell suspension was collected and then centrifuged at 1500 rpm for
   5 min. After discarding the supernatant, 3–5 mL of ACK red blood cell
   (RBC) lysis buffer was added to lyse RBC at 4 °C for 10 min, and upon
   completion of lysis, 40 mL phosphate buffer saline (PBS) was added to
   terminate the process. After the decidual cells were centrifuged at
   1500 rpm for 5 min, they were collected into 24-well plates at
   3 × 10^6 cells per well for 24 h and centrifuged at 1500 rpm for 5 min.
   Finally, the supernatant, namely, the decidual cells culture medium,
   was collected.

   Decidua samples were collected from women with normal vaginal delivery,
   and flow cytometry was used to select CD14+macrophages from the
   decidua. The decidual tissue was digested to extract decidual
   macrophages, and RBCs were lysed as described above, with the resultant
   cell suspension being collected for subsequent analysis. Flow cytometry
   was performed in a 100 μL system with 1 μL of CD14-APC antibody/2 ∗
   10^6 cells for 20 min. The sorted cells were resuspended in a
   Dulbecco's Modified Eagle Medium (DMEM, Gibco, CAT: C11995500BT)
   containing 2 mmol/L glutamine and 25 ng/mL granulocyte-macrophage
   colony stimulating factor (GM-CSF) and collected in 48-well plates,
   specifically at 3.5∗10^5 cells per well.

2.3. Flow cytometry

2.3.1. Treatment of cells

   The decidua samples from control group (CTRL) and atonic PPH patients
   were collected and processed as outlined in the section Clinical sample
   collection. Decidual cells were sorted, including macrophages, natural
   killer (NK) cells, and T lymphocytes (T cells) from the atonic PPH and
   CTRL groups.

   Peripheral blood samples were collected from women with normal vaginal
   delivery during labor and centrifuged at 12,000×g for 10 min. The white
   membrane layer, including peripheral blood mononuclear cells, were
   collected through centrifugation with 3 mL of Ficol added. After being
   labeled with CD14 magnetic beads, CD14^+ monocytes were sorted using
   the Meitianni fully automatic magnetic bead sorter and then plated and
   cultured in 48-well plates at 4 × 10^5 cells per well for 24 h. The
   culture media were replaced with CTRL- or PPH-patient-derived decidual
   cells culture medium as outlined in the section Clinical sample
   collection. CD14+monocytes were further cultured for 48 h, and after
   centrifugation at 500×g for 5 min, the CD14+monocytes were collected
   for subsequent analysis.

   Decidua samples were collected from women with normal vaginal delivery,
   and flow cytometry was used to select CD14+macrophages from the
   decidua, which was described comprehensively in the section Clinical
   sample collection. Decidual CD14^+ macrophages were plated and cultured
   in 48-well plates at 3.5 × 10^5 cells per well for 24 h, and 0, 1, 10,
   and 100 μmol/L of H[2]O[2] were added for 1 h or cultured in the CTRL-
   or PPH-patient-derived decidual cell culture media for 48 h, followed
   by centrifugation at 500×g for 5 min. Finally, CD14+monocytes were
   collected for subsequent analysis.

2.3.2. Analyses

   The decidual cells, including macrophages, NK cells, and T cells, were
   stained for 30 min at 25 °C in the dark. The following antibodies were
   used: Brilliant Violet 421 anti-human CD45 (Biolegend, CAT: 304032,
   dilution: 1:200), APC anti-human CD14 (Biolegend, CAT: 325608,
   dilution: 1:200), PE/Cyanine5 anti-human CD3 (Biolegend, CAT: 317355,
   dilution: 1:200), FITC anti-human CD56 (Biolegend, CAT: 362546,
   dilution: 1:200), and Mito SOX (Thermo, CAT: [57]M36008, dilution:
   1:500). The samples were then centrifuged at 1500×g for 5 min, washed
   and resuspended in 300–400 μL PBS. Finally, the cells were then
   immediately analyzed using a Canto flow cytometer (Beckman Counter FACS
   Galios).

   The CD14^+ monocytes from peripheral blood samples were stained for
   30 min at 25 °C in the dark. The following dyes were used: Mito SOX
   (Invitrogen, CAT: M36007, dilution: 1:2500) and DCFH-DA (Solarbio, CAT:
   D6470, dilution: 1:500). A SONY ID7000 full-spectrum flow cytometer was
   used to detect fluorescence.

   The CD14^+ macrophages from decidual samples were stained for 1 h at
   4 °C with 50 μL of staining system. The following antibodies were used:
   CD206 (Biolegend, CAT: 321131, dilution: 1:200) and CD86 (Biolegend,
   CAT: 374216, dilution: 1:200). Finally, upon completion of staining,
   the cells were immediately analyzed using a Beckman Cytoflex flow
   cytometer.

2.4. Vaginal proteomics

   The vaginal blood was centrifuged at 12,000×g and 4 °C for 10 min.
   Subsequently, 50 μL of blood samples were incubated in magnetic
   nanomaterials (PTM-00F13303, PTM Bio) at 1200×g and 37 °C for 1 h, and
   magnetic beads were washed thrice. Next, 70 μL of enzymatic hydrolysis
   buffer (200 mM Tris,pH8.0) was added, followed by heating at 95 °C for
   10 min. After the sample dropped to 25 °C, 20 ng/μL trypsin was added
   for hydrolysis overnight at 37 °C. To achieve a final concentration of
   5 mM, reduction was performed at 56 °C for 30 min using 1 M
   dithiothreitol. Furthermore, 550 mM iodoacetamide was added to achieve
   a final concentration of 11 mM. According to the C18 ZipTips manual,
   salt was removed, and vacuum-freezing was conducted, followed by drying
   and liquid chromatography-mass spectrometry. After enzymatic
   hydrolysis, the peptide segments were acidified with 10 %
   trifluoroacetic acid to attain pH 2–3 and centrifuged at 12,000×g for
   10 min at 25 °C, and the supernatant was collected. Similarly, 50 μL of
   activation solution, desalination solution, peptide segment, and
   desalination solution were added to StageTip sequentially, centrifuged
   at 1500×g per step for 1 min, and repeated once. Finally, 20 μL of
   eluent was added, followed by centrifugation at 750×g for 1 min,
   freezing, spin-drying, and setting aside for later use.

   The peptide segments were dissolved in liquid chromatography mobile
   phase A and separated using an EASY nLC 1200 ultra-high-performance
   liquid chromatography system. For ionization, the samples were injected
   into an NSI ion source and analyzed using an Orbitrap Exploris 480 mass
   spectrometer. Data were collected using a data-independent scanning
   program. The intensity (I) of proteins in different samples was
   normalized through a centralization transformation to obtain the
   relative quantitative values (R) in different samples. To evaluate the
   quantitative reproducibility between biological or technological
   replicates, repeatability was analyzed using three statistical methods,
   namely Pearson's correlation coefficient, principal component analysis
   (PCA), and relative standard deviation (RSD).

2.5. Vaginal blood metabolomics

   To 20 % acetonitrile methanol internal standard extraction solution,
   50 μL of the sample was added, vortexed for 3 min, and centrifuged at
   12,000×g and 4 °C for 10 min. The supernatant was stood at −20 °C for
   30 min, centrifuged at 12,000×g and 4 °C for 3 min, and transferred for
   analysis on the machine. Raw data were extracted and corrected using
   the XCMS program in mzXML format. Metabolite identification was
   performed by searching the laboratory's in-house database and
   integrating public and prediction databases, and metDNA methods.
   Finally, substances with a comprehensive identification score of ≥0.5
   and a QC sample CV of <0.3 were extracted, and the positive and
   negative modes were merged to obtain metabolite data.

   We conducted quality control analysis, PCA, cluster analysis, grouped
   principal component analysis, orthogonal partial least squares
   discriminant analysis, dynamic distribution of metabolite content
   differences, and screening of differential metabolites. Metabolites
   with a variable importance in the projection (VIP) of >1 (Student's
   t-test, P < 0.05) were further analyzed. A higher VIP value indicated
   that a metabolite played a more prominent role in distinguishing
   between different groups or in explaining experimental variables. These
   metabolites may help elucidate or distinguish the differences between
   the groups effectively. In differential metabolite analysis, the Kyoto
   Encyclopedia of the Genome (KEGG), Human Metabolome Database (HMDB),
   and MebaboAnalyst databases were used for KEGG, HMDB, and MSEA
   functional annotation and enrichment analyses, respectively.

2.6. Proteomic and metabolomic analysis

   Firstly, statistical analysis of differential proteins and metabolites
   was conducted. Specifically, the differential screening criteria for
   proteins were FC > 1.5 or FC < 1/1.5 and P < 0.05, whereas that for
   metabolites were | log2FC |>0, P < 0.05, and VIP>1. iPath analysis
   tools were used to simultaneously map differentially expressed proteins
   and metabolites to the iPath pathway map, obtain their shared pathway
   information, screen for pathways with both differential proteins and
   metabolites, and finally visualize them. Using the resultant
   quantitative outputs, the Spearman correlation analysis was performed
   on differentially expressed proteins and metabolites to screen for
   those that exhibited a significant correlation (correlation coefficient
   |r |≥0.8) and construct their co-expression relationship network.
   Finally, the Metscape plugin in Cytoscape was used to construct a
   network of differentially expressed proteins and metabolites. This
   network was built based on secondary data from various sources (i.e.,
   databases, experiments, etc.), delving into the relationship between
   metabolites and proteins. The compound network was visualized it
   displayed information on compound structures, reaction types, enzymes,
   genes, and pathways, while also being able to validate differential
   proteins at a metabolic level closer to phenotype. The above analysis
   combined methods commonly used in omics, including (1) association
   analysis based on statistical methods, (2) association analysis based
   on metabolic pathway analysis, (3) association analysis based on
   interaction, and (4) correlation analysis based on references and