Abstract

Background

   Term birth (TB) and preterm birth (PTB) are characterized by uterine
   contractions, rupture of the chorioamniotic membrane, decidual
   activation, and other physiological and pathological changes. In this
   study, we hypothesize that inflammation can cause changes in mRNA
   expression and metabolic stability in the placenta, decidua,
   chorioamniotic membrane, uterus and peripheral blood, ultimately
   leading to PTB.

Methods

   To comprehensively assess the effects of inflammation on mRNA
   expression and metabolite production in different tissues of pregnancy,
   we used a mouse PTB model by intraperitoneally injecting
   lipopolysaccharide (LPS) and integrated transcriptomics and
   metabolomics studies.

Results

   Our analysis identified 152 common differentially expressed genes
   (DEGs) and 8 common differentially expressed metabolites (DEMs) in the
   placenta, decidua, chorioamniotic membrane, uterus, and peripheral
   blood, or placenta and uterus after LPS injection, respectively. Our
   bioinformatics analysis revealed significant enrichment of the NOD-like
   receptor signaling pathway (mmu04621), TNF signaling pathway
   (mmu04668), IL-17 signaling pathway (mmu04657), and NF-kappa B
   signaling pathway in the transcriptomics of different tissues, and
   Hormone synthesis, Lysosome, NOD-like receptor signaling pathway, and
   Protein digest and absorption pathway in metabolomics. Moreover, we
   found that several upstream regulators and master regulators, including
   STAT1, STAT3, and NFKB1, were altered after exposure to inflammation in
   the different tissues. Interaction network analysis of transcriptomics
   and metabolomics DEGs and DEMs also revealed functional changes in mice
   intraperitoneally injected with LPS.

Conclusions

   Overall, our study identified significant and biologically relevant
   alterations in the placenta, decidua, chorioamniotic membrane, uterus,
   peripheral blood transcriptome and the placenta and uterus metabolome
   in mice exposed to LPS. Thus, a comprehensive analysis of different
   pregnancy tissues in mice intraperitoneally injected with LPS by
   combining transcriptomics and metabolomics may help to systematically
   understand the local and systemic changes associated with PTB caused by
   inflammation.

   Keywords: maternal-fetal interface, transcriptomics, metabolomics,
   preterm birth, mice

Introduction

   Preterm birth (PTB) is a condition where delivery occurs before 37
   weeks of pregnancy. According to the World Health Organization, the
   global incidence of PTB is approximately 10%, and it is a leading cause
   of perinatal morbidity and mortality, with an estimated 15 million
   premature infants born worldwide each year ([31]1–[32]5). Although the
   withdrawal of progesterone, secretion of oxytocin, activation of
   decidual inflammation, and fetal immune response are believed to be
   involved in the onset of term labor (TB), the pathogenesis and etiology
   of PTB remain unclear ([33]3).

   PTB is widely believed to be an inflammatory process, and
   epidemiological, clinical, and animal studies have established a
   correlation between inflammation and PTB ([34]6–[35]8). The onset of
   labor in both rodents and humans involves positive feedback from the
   inflammatory cascade, driven by increased levels of pro-inflammatory
   cytokines, chemokines, and prostaglandins (PGs) at the maternal-fetal
   interface ([36]9–[37]11). The closest position between the mother and
   the fetus is the maternal-fetal interface, which comprises the
   placenta, decidua, uterus and chorioamniotic membrane. Toll-like
   receptors (TLRs) are critical upstream gatekeepers of inflammation,
   which are highly expressed at the maternal-fetal interface. Studies
   have shown that premature activation of TLR4 signaling at the
   maternal-fetal interface can lead to PTB ([38]12, [39]13). In addition,
   damage-associated molecular patterns (DAMPs) and pathogen-associated
   molecular patterns (PAMPs), the endogenous and microbe-released TLR
   ligands, respectively ([40]13), activate TLR4 to trigger the release of
   downstream pro-inflammatory cytokines and chemokines such as IL-1β,
   IL-6, IL-8, CCL5, and TNF-α, leading to leukocyte infiltration at the
   maternal-fetal interface and cervix, as well as promoting the rupture
   of membranes, cervical ripening and uterine contraction by activating
   the release of matrix metalloproteinases (MMPs) and PGs ([41]14).

   The maintenance of human pregnancy and initiation of delivery has been
   closely linked to the dynamic balance of the immune microenvironment at
   the maternal-fetal interface ([42]15, [43]16). During delivery, immune
   cells present at the maternal-fetal interface create an inflammatory
   environment that facilitates the expulsion of the fetus from the
   mother. However, disruption of the immune balance at the maternal-fetal
   interface can result in premature activation of the common pathway of
   delivery, leading to PTB ([44]15–[45]17). Besides, clinical and animal
   studies have shown a significant positive correlation between
   infection, inflammation, and PTB ([46]18). Microorganisms associated
   with PTB are typically derived from the reproductive or urogenital
   tract or blood and secrete endotoxins such as lipopolysaccharide (LPS)
   at the maternal-fetal interface, causing intense inflammation and
   ultimately resulting in premature uterine contraction, rupture of fetal
   membranes, and cervical ripening ([47]14). Therefore, maternal-fetal
   interface immune imbalance leading to defects in inflammatory factor
   release and metabolic stability may play a crucial role in the process
   of PTB. However, the pathogenesis of PTB and the role of inflammation
   in causing labor have yet to be fully elucidated, and current drug
   treatments, including antibiotics, for PTB are not entirely
   satisfactory ([48]8).

   Although the use of high-throughput omics techniques has generated a
   vast amount of omics data, it is challenging to fully understand
   complex biological processes and the regulation of biological networks
   using just one type of omics data. By integrating transcriptomic and
   metabolomic analyses, we can systematically and thoroughly examine the
   functions and regulatory mechanisms of biomolecules, ultimately
   identifying key metabolic pathways, genes, and metabolites for further
   in-depth research and application. Thus, this present study was
   designed to identify new pathways in the maternal-fetal interface that
   undergo significant changes in an acute inflammatory environment and
   provide new insights into the potential pathological mechanisms of PTB
   by exploring the role of the maternal-fetal interface in the initiation
   of labor mediated by inflammatory responses through the evaluation of
   the transcriptomes of maternal peripheral blood, placenta, uterus,
   decidua, and chorioamniotic membrane, as well as the placental and
   uterine metabolomes, in an LPS-induced PTB mouse model.

Materials and methods

Mice

   C57BL/6 inbred, timed pregnant mice were purchased from the
   GemPharmatech Co., Ltd (Jiangsu, China). Briefly, the mice were housed
   under a 12-hour light: 12-hour dark circadian cycle and female mice
   aged 8-12 weeks were mated with fertile male mice. We examined the
   females daily between 8:00 a.m. and 9:00 a.m. and considered the
   appearance of a vaginal plug as 0.5 days post coitum (dpc). After
   separation from the males, pregnancy was confirmed by a weight gain of
   ≥ 2g by 12.5dpc. All animal procedures were performed in accordance
   with the Principles of Laboratory Animal Care of the Department of
   Laboratory Animal Science at Fudan University and approved by the
   Ethics Committee of Fudan University (Protocol Number: 20211203S).

LPS model of PTB

   On 16.5 dpc, mice were intraperitoneally (i.p.) injected with either
   100μg/animal of lipopolysaccharide (LPS) (Escherichia coli O55:B5,
   Sigma-Aldrich, St. Louis, MO, L2880) in 200μL of sterile
   phosphate-buffered saline (PBS) or 200ul/animal of PBS alone as a
   control (n=4 per group). In this study, PTB was defined as delivery
   before 18.5dpc. Also, we found that a dose of 100μg of LPS was the
   lowest dose that reproducibly induced PTB without significant morbidity
   or mortality ([49]19–[50]22). The duration of labor was measured by
   recording the number of hours that passed from the time of
   intraperitoneal (i.p.) injection of LPS/PBS to the delivery of the
   first pup. [51]Figure S1 showed that the mouse model of PTB was
   successfully constructed ([52] Supplementary Figures S1A, B ). Within
   24 hours of the LPS/PBS injection, the mice were anesthetized with 1%
   pentobarbital/phenytoin immediately after the delivery of the first
   pup. Whole blood was collected using 5ml tubes (preloaded with 3ml
   TRIzol^® Reagent [Magen]) by removing the eyeball. Subsequently, the
   mice were dissected to obtain the placenta, uterus, decidua, and
   chorioamniotic membrane. The collected tissues were then placed in
   RNAlater (Thermo Fisher Scientific) and stored at -80°C until they were
   ready to be used.

RNA extraction and library preparation and sequencing

   Total RNA was extracted from the mouse placenta, decidua,
   chorioamniotic membrane, uterus, and peripheral blood using TRIzol^®
   Reagent according to the manufacturer’s instructions (Magen). The
   quality and quantity of RNA were evaluated based on the A260/A280
   absorbance ratio using a Nanodrop ND-2000 system (Thermo Scientific,
   USA), and the RNA integrity number (RIN) was determined using an
   Agilent Bioanalyzer 4150 system (Agilent Technologies, CA, USA). Only
   samples that met the quality criteria were selected for library
   construction. Paired-end libraries were prepared using an ABclonal
   mRNA-seq Lib Prep Kit (ABclonal, China) following the manufacturer’s
   instructions. Initially, mRNA was isolated from 1 μg of total RNA using
   oligo (dT) magnetic beads and fragmented using divalent cations at
   elevated temperatures in ABclonal First Strand Synthesis Reaction
   Buffer. Subsequently, first-strand cDNAs were synthesized using random
   hexamer primers and Reverse Transcriptase (RNase H) with mRNA fragments
   as templates, followed by second-strand cDNA synthesis using DNA
   polymerase I, RNAseH, buffer, and dNTPs. The synthesized
   double-stranded cDNA fragments were then adapter-ligated to prepare the
   paired-end library, and PCR amplification was performed on the
   adaptor-ligated cDNA. PCR products were purified using the AMPure XP
   system, and library quality was assessed using the Agilent Bioanalyzer
   4150 system. Finally, the library preparations were sequenced on an
   Illumina Novaseq 6000 (or MGISEQ-T7) and 150 bp paired-end reads were
   generated.

Metabolite extractions and LC-MS analysis

   To extract metabolites from mouse placenta and uterus samples, 80 mg of
   the sample was mixed with 1 ml of cold extraction solvent
   (methanol/acetonitrile/H2O, 2:2:1, v/v/v) and thoroughly vortexed. The
   lysate was then homogenized using an MP homogenizer (24×2, 6.0 M/S,
   60s, twice) and sonicated at 4°C (30min/once, twice) followed by
   centrifugation at 14,000g for 20 minutes at 4°C. The resulting
   supernatant was dried using a vacuum centrifuge at 4°C. For LC-MS
   analysis, the dried samples were re-dissolved in 100 μL of
   acetonitrile/water (1:1, v/v) solvent.

   The extracts were analyzed using a Sciex TripleTOF 6600 quadrupole
   time-of-flight mass spectrometer coupled with hydrophilic interaction
   chromatography via electrospray ionization at Shanghai Applied Protein
   Technology Co., Ltd. The LC separation was performed on an ACQUIY UPLC
   BEH Amide column (2.1 mm × 100 mm, 1.7 µm particle size, Waters,
   Ireland) using a gradient of solvent A (25 mM ammonium acetate and 25
   mM ammonium hydroxide in water) and solvent B (acetonitrile). The
   gradient started with 95% B for 1 minute and was linearly decreased to
   65% B over 11 minutes, then reduced to 40% B in 0.1 minutes and
   maintained for 4 minutes. Finally, the gradient was increased to 95% B
   in 0.1 minutes, and a 5-minute re-equilibration period was employed.
   The flow rate was 0.5 mL/minute, and the column temperature was
   maintained at 25°C. The auto-sampler was kept at 4°C, and the injection
   volume was 2 µL.

   The mass spectrometer was operated in both negative and positive
   ionization modes. The ESI source conditions were set as follows: Ion
   Source Gas1 (Gas1) at 60, Ion Source Gas2 (Gas2) at 60, Curtain Gas
   (CUR) at 30, source temperature at 600°C, and IonSpray Voltage Floating
   (ISVF) at ±5500V. For MS acquisition, the instrument was set to acquire
   over the m/z range of 60-1000 Da, and the accumulation time for the TOF
   MS scan was set at 0.20s/spectra. In auto MS/MS acquisition, the
   instrument was set to acquire over the m/z range of 25-1000 Da, and the
   accumulation time for product ion scan was set at 0.05s/spectra. The
   product ion scan was obtained using information-dependent acquisition
   (IDA) with high sensitivity mode selected. The following parameters
   were used: collision energy (CE) was fixed at 35V with ±15eV;
   declustering potential (DP) at 60V (+) and −60V (-); exclude isotopes
   within 4Da, and candidate ions to monitor per cycle were set at 10.

Real-time quantitative PCR validation analysis

   To validate the transcriptome sequencing outcomes, RT-qPCR was
   performed. RNA samples were first reverse transcribed to cDNA using the
   PrimeScript RT Reagent kit (Takara, Japan) following the manufacturer’s
   instructions. qPCR was carried out using TB Green qPCR Master Mix
   (Takara, Japan) on a Q5 real-time PCR System (Life Technology, USA).
   The following thermocycling conditions were used for the PCR: an
   initial denaturation at 95°C for 30 seconds, 40 cycles of 95°C for 5
   seconds, and 60°C for 34 seconds, followed by a final extension at 95°C
   for 15 seconds, 60°C for 1 minute, and 95°C for 15 seconds. The
   relative expression of target genes was calculated using the 2^-ΔΔCt
   method, with ACTB as the internal reference gene.

Bioinformatics analysis

   The transcriptomic data generated from the Illumina platform were
   subjected to bioinformatics analysis using an in-house pipeline
   developed by Shanghai Applied Protein Technology. After quality
   control, the clean reads were aligned separately to the reference
   genome in orientation mode using HISAT2 software
   ([53]http://daehwankimlab.github.io/hisat2/) to obtain mapped reads.
   The reads numbers mapped to each gene were counted using FeatureCounts
   ([54]http://subread.sourceforge.net/), and the FPKM of each gene was
   calculated based on the gene length and read count mapped to it.
   Differential expression genes (DEGs) analysis was carried out using the
   DESeq2 package
   ([55]http://bioconductor.org/packages/release/bioc/html/DESeq2.html).
   DEGs with |log2FC| > 1 and P adj < 0.05 were considered significantly
   differentially expressed genes. Transcription factor (TF) analysis of
   DEGs was extracted directly from the AnimalTFDB
   ([56]http://bioinfo.life.hust.edu.cn/AnimalTFDB/) and PlantTFDB
   database ([57]http://planttfdb.cbi.pku.edu.cn/). Protein-protein
   interaction (PPI) analysis was performed based on the known and
   predicted interactions in the STRING database
   ([58]https://www.string-db.org/). Finally, alternative splicing (AS)
   analysis was conducted using the rMATS software
   ([59]http://rnaseq-mats.sourceforge.net/index.html).

   The metabolomic raw MS data (wiff.scan files) were first converted to
   MzXML files using ProteoWizard MSConvert and then imported into the
   freely available XCMS software. For peak picking, the following
   parameters were used: centWave m/z = 25ppm, peakwidth = c (10, 60),
   prefilter = c (10, 100). Peak grouping was conducted with bw = 5, mzwid
   = 0.025, and minfrac = 0.5. In the extracted ion features, only
   variables with more than 50% of the nonzero measurement values in at
   least one group were retained. Compound identification of metabolites
   was performed by matching MS/MS spectra with an in-house database
   established with available authentic standards. After normalization to
   total peak intensity, the processed data were uploaded into SIMCA-P
   (version 14.1, Umetrics, Umea, Sweden) for multivariate data analysis,
   including Pareto-scaled principal component analysis (PCA) and
   orthogonal partial least-squares discriminant analysis (OPLS-DA). The
   robustness of the model was evaluated using 7-fold cross-validation and
   response permutation testing. The contribution of each variable to the
   classification was determined by calculating its variable importance in
   the projection (VIP) value in the OPLS-DA model. Statistical
   significance was determined using an unpaired Student’s t-test, where
   VIP value >1 and P < 0.05 were considered statistically significant.

   Transcriptomic GO function enrichment and KEGG pathway enrichment
   analyses were performed using the clusterProfiler R software package.
   For metabolomic KEGG pathway annotation, the metabolites were blasted
   against the online Kyoto Encyclopedia of Genes and Genomes (KEGG)
   database to retrieve their COs, which were then mapped to pathways in
   KEGG. KEGG pathway enrichment analyses were conducted using Fisher’s
   exact test, with the whole metabolites of each pathway considered as
   the background dataset. GO or KEGG functions were considered
   significantly enriched when P < 0.05.

Integrated network analysis of the transcriptome and metabolome

   The Spearman statistical method was employed to analyze the correlation
   coefficient between the significant differential genes and metabolites.
   The resulting correlation matrix was visualized using R and Cytoscape
   (v3.8.0) software, which allowed exploration of the interaction between
   genes and metabolites from multiple perspectives, including matrix heat
   maps, hierarchical clustering, and correlation networks. Spearman
   correlation network analysis was performed on genes and metabolites
   with significant correlation coefficient values |r| ≥ 0.5 and P < 0.01.

Data analysis

   Statistical analyses were performed using GraphPad Prism 9 software. An
   independent samples t-test or Fisher’s exact test was used to compare
   groups, as appropriate. Adjusted P-values were calculated using the
   Benjamani-Hochberg false discovery rate (FDR), with an FDR value ≤ 0.01
   considered significant. A P-value < 0.05 was considered statistically
   significant.

Results

Identification of differentially expressed genes and metabolites

   Principal component analysis (PCA) revealed a clear distinction between
   the two treatments (LPS vs. PBS) in the placenta, chorioamniotic
   membrane, decidua, uterus, and peripheral blood samples ([60] Figures
   S2A–E ). Besides, all raw data of the samples passed quality control
   (QC) ([61] Supplementary Table S1 ). A total of 497, 2133, 522, 1549,
   and 2579 DEGs (P adj < 0.05 and |log2(FC)| > 1) were identified in
   placenta (LPS vs. PBS), chorioamniotic membrane (LPS vs. PBS), decidua
   (LPS vs. PBS), uterus (LPS vs. PBS), and peripheral blood (LPS vs.
   PBS), respectively ([62] Supplementary Table S2 ). The volcano plot of
   gene expression levels illustrates the distribution of up- and
   down-regulated genes in different tissues ([63] Supplementary Figures
   S3A–E ). Comparison of the placenta, chorioamniotic membrane, decidua,
   uterus, and peripheral blood revealed 152 significant overlapping
   differential genes (P < 0.05) ([64] Figure 1A ; [65]Table 1 ;
   [66]Supplementary Table S3 ).

Figure 1.

   [67]Figure 1
   [68]Open in a new tab

   Venn diagram and UpSet plot of DEGs and DEMs. (A) Venn diagram and
   UpSet plot showing the shared and unique DEGs among mouse placenta,
   decidua, chorioamniotic membrane, uterus and peripheral blood (P <
   0.05, PBS vs. LPS). (B) Venn diagram and UpSet plot showing the shared
   and unique DEMs among mouse placenta and uterus (P < 0.05, PBS vs.
   LPS). The horizontal bars display the number of DEGs of different
   tissues, while the vertical bars show the size of genes identified by
   only 1 comparison and the intersection sets. Single dots and the
   corresponding horizontal bars represent the number of genes unique to
   the data set and not shared between the other comparisons. The UpSet
   plot was performed using the OmicStudio tools at
   [69]https://www.omicstudio.cn/tool. DEGs, differentially expressed
   genes; DEMs, differentially expressed metabolites.

Table 1.

   Common DEGs and DEMs in the different tissues.
   Tissues Placenta Chorioamniotic Membrane Decidua Uterus Peripheral
   Blood
   Genes
   JUNB,SOCS3,TRIM21,STAT3,TLR3,PML,ADORA2B,RNF114,LITAF,HAP1,TAP1,DAXX,IF
   IH1,DHX58,WHAMM,MX2,OAS1B,TOR3A,PARP10,ACOD1,OAS1A,MLKL,USP18,CD274,IFI
   205,IER3,MYD88,IGTP,OAS1G,NOD2,SMOX,RASSF2,NMI,TNFRSF1A,PSMB9,CCL3,CCL4
   ,IFIT2,TNF,TIMP1,GBP3,SOCS1,CLEC4E,CCL2,CXCL1,OASL2,MITD1,MARCKS,HCAR2,
   ICAM1,OGFR,RELB,TNIP1,CSNK1A1,IRF1,CASP4,IL1B,SNX10,CCL7,RSAD2,SELP,ZBP
   1,GM4951,PTX3,NFKBIA,SLFN4,GM12250,BHLHE40,PFKFB3,OLFR56,CRISPLD2,LILR4
   B,CD14,HERC6,RND1,TRIM30D,XAF1,MS4A4C,IRF7,CCRL2,BCL2A1D,TRIM30A,IL6,LC
   N2,NFIL3,TNFAIP2,ZFP36,PSME2,RTP4,PSME2B,FPR2,SAT1,CD47,H2Q4,RNF19B,IRG
   M1,HCK,MXD1,PIK3R5,IFI209,EIF6,GBP6,BCL3,CCL12,IFITM3,PSMB8,MEFV,NFKBIB
   ,TIFA,PHF11B,SOD2,LCP2,WFDC17,BCL2A1A,TLR2,NFKBIE,SLFN8,PSMB10,HP,FCGR4
   ,IFI44,NOS2,GBP9,ACOT2,SP140,MAP3K8,OAS3,SLFN2,CD300LF,MS4A6D,GLIPR2,PH
   F11D,TNFRSF1B,MNDAL,EPSTI1,FCGR2B,GM4841,PNRC1,[70]AA467197,IL1F9,IL15R
   A,CXCL16,PNP,RHBDF2,UBE2L6,APOBEC3,SLCO3A1,GPR84,BATF2,KIFC2,IFITM1,GBP
   4
   Tissues Placenta (POS) Uterus (POS)
   Metabolites Nicotinamide N-oxide, N-acetyl-d-lactosamine, His-Asp,
   5,8-dimethylquinolin-4-ol
   Tissues Placenta (NEG) Uterus (NEG)
   Metabolites L-dihydroorotate, Trans-3’-hydroxycotinine
   o-.beta.-d-glucuronide, Humulone, Pentobarbital
   [71]Open in a new tab

   DEGs, differentially expressed genes; DEMs, differentially expressed
   metabolites; POS, positive; NEG, negative.

   Orthogonal partial least squares discriminant analysis (OPLS-DA)
   revealed significant differences between the two treatment groups (LPS
   vs. PBS) in the placenta and uterus in both positive and negative ion
   modes ([72] Figures 2A–D ). A permutation test was conducted to ensure
   model validity and stability and to prevent over-fitting ([73]
   Supplementary Figures S4A–D ). In total, 114 and 54 differentially
   expressed genes and metabolites (DEMs) were identified in the placenta
   in positive and negative ion modes, respectively, while 158 and 146
   DEMs were identified in the uterus in positive and negative ion modes,
   respectively ([74] Supplementary Table S4 ), all with OPLS-DA VIP>1 and
   P value < 0.05. When comparing the placenta with the uterus, a total of
   8 significant overlapping differential genes (P < 0.05) were identified
   in both positive and negative ion modes ([75] Figure 1B ; [76]Table 1
   ).

Figure 2.

   [77]Figure 2
   [78]Open in a new tab

   Two-dimensional scatter score plot of OPLS-DA. Scatter score plot of
   OPLS-DA in the placenta for (A) POS-LPS vs. PBS and (B) NEG-LPS vs.
   PBS, and the uterus for (C) POS-LPS vs. PBS and (D) NEG-LPS vs. PBS. An
   OPLS-DA model was generated using a seven-fold cross-validation
   procedure, with the model evaluation parameters (R^2Y, Q^2) obtained by
   7-fold cross-validation. Here, Q^2 > 0.5 indicates that the model is
   stable and reliable, 0.3 < Q^2 ≤ 0.5 indicates that the model is
   stable, and Q^2 < 0.3 indicates that the model has low reliability. The
   OPLS-DA analysis was performed in both the positive and negative ion
   modes. OPLS-DA, orthogonal partial least squares discriminant analysis;
   POS, positive; NEG, negative.

KEGG and GO functional enrichment analyses of DEGs

   GO functional analysis revealed that several GO terms were common among
   placenta (LPS vs. PBS), chorioamniotic membrane (LPS vs. PBS), decidua
   (LPS vs. PBS), uterus (LPS vs. PBS) and peripheral blood (LPS vs. PBS).
   These shared GO terms included biological process (BP) terms such as
   response to cytokine (GO:0034097), defense response (GO:0006952) and
   response to biotic stimulus (GO:0009607), molecular function (MF) terms
   such as signaling receptor binding (GO:0005102), cytokine activity
   (GO:0005125) and chemokine activity (GO:0008009), and cellular
   component (CC) terms such as extracellular region (GO:0005576) and cell
   surface (GO:0009986) in cellular component (CC) ([79] Figures 3A–E ;
   [80]Supplementary Table 5 ).

Figure 3.

   [81]Figure 3
   [82]Open in a new tab

   Differential gene GO annotation visualization (top 30). Bubble chart
   for the (A) placenta (LPS vs. PBS), (B) chorioamniotic membrane (LPS
   vs. PBS), (C) decidua (LPS vs. PBS), (D) uterus (LPS vs. PBS), and (E)
   peripheral blood (LPS vs. PBS). The annotated GO term is arranged from
   small to large according to log10(P value) (abscissa), and the ordinate
   represents specific functional description information. GO includes
   molecular function (MF), biological process (BP) and cellular component
   (CC) (red). Bubble size represents the number of DEGs (P adj < 0.05,
   |log2(fold change)| > 1) involved in the GO enrichment. The rich factor
   was calculated as follows: (number of DEGs in GO term)/(number of genes
   contained in GO term).

   The top 20 enriched pathways were identified through KEGG functional
   enrichment analysis of the DEGs in different mouse tissues ([83]
   Figures 4A–E ; [84]Supplementary Table S5 ). Further analysis showed
   that the NOD-like receptor signaling pathway (mmu04621), TNF signaling
   pathway (mmu04668), IL-17 signaling pathway (mmu04657) and NF-kappa B
   signaling pathway (mmu04064) were significantly enriched in all tissues
   or maternal-fetal interfaces (placenta, chorioamniotic membrane,
   decidua, and uterus) ([85] Figures 4A–E ). Besides, KEGG functional
   analysis of the 152 significant overlapping DEGs (P < 0.05) showed
   similar results ([86] Supplementary Table S6 ).

Figure 4.

   [87]Figure 4
   [88]Open in a new tab

   Bubble maps of KEGG pathway enrichment of DEGs. Bubble maps for the (A)
   placenta (LPS vs. PBS), (B) chorioamniotic membrane (LPS vs. PBS), (C)
   decidua (LPS vs. PBS), (D) uterus (LPS vs. PBS), and (E) peripheral
   blood (LPS vs. PBS). In this diagram, the degree of KEGG enrichment was
   measured using the Rich factor, P-value and the number of genes
   enriched in a given pathway. Rich factor refers to the ratio of the
   number of DEGs in the pathway to the total number of genes. P value
   refers to the significance of pathway enrichment, and the value range
   is [0,1]. The closer it is to zero, the more significant the enrichment
   is. The bubble size represents the number of DEGs (P adj < 0.05,
   |log2(fold change)| > 1) involved in the KEGG enrichment. Each point
   represents a KEGG pathway, the ordinate represents the pathway name,
   and the abscissa represents the log10(P value). The Rich factor is
   represented by color, with a darker red color indicating a higher
   significance level.

Protein-protein interaction network

   The PPI network was constructed using the online STRING database, and
   further analysis was conducted using Cytoscape software. CytoHubba was
   employed to screen for key genes, and all molecular nodes and edges
   were calculated. The top 30 key genes were obtained using the
   multiscale curvature classification algorithm. The interactions of the
   top 30 DEGs in the placenta, chorioamniotic membrane, decidua, uterus
   and peripheral blood, as well as the 152 significant overlapping DEGs
   (P < 0.05), were explored through PPI analysis ([89] Figures 5A–F ).

Figure 5.

   [90]Figure 5
   [91]Open in a new tab

   Interaction network diagram of DEGs (Top 30). Interaction network
   diagram for the (A) placenta (LPS vs. PBS), (B) chorioamniotic membrane
   (LPS vs. PBS), (C) decidua (LPS vs. PBS), (D) uterus (LPS vs. PBS), (E)
   peripheral blood (LPS vs. PBS), and (F) co-expressed DEGs in the
   placenta (LPS vs. PBS), chorioamniotic membrane (LPS vs. PBS), decidua
   (LPS vs. PBS), uterus (LPS vs. PBS) and peripheral blood (LPS vs. PBS)
   (P adj < 0.05, |log2(fold change)| > 1). Top 30 hub genes identified
   using CytoHubba. Each circle in the figure represents a gene, and the
   lines between circles represent interactions between genes. The color
   of a circle reflects the number of interactions that the corresponding
   gene has, with a darker color indicating a higher degree of
   connectivity.

GSEA, TF and AS analysis of DEGs

   Gene Set Enrichment Analysis (GSEA,
   [92]http://software.broadinstitute.org/gsea), TF analysis (TFscan
   software), and alternative splicing (AS) events analysis were used for
   further analysis of the DEGs. GSEA showed the KEGG Apoptosis pathway in
   the placenta, while the NOD-like receptor signaling pathway was
   significantly enriched in the chorioamniotic membrane, decidua, uterus
   and peripheral blood ([93] Supplementary Figures S5A–E ). The results
   of cluster analysis of all DEGs in the apoptosis pathway and NOD-like
   receptor signaling pathway are depicted in heat maps ([94] Figures 6A,
   B ). The TFs regulating DEGs caused by LPS in different mouse tissues
   were found to be different ([95] Supplementary Figures S6A–E ;
   [96]Table S5 ). AS analysis using rMATS software
   ([97]http://rnaseq-mats.sourceforge.net/index.html) revealed that the
   AS patterns induced by LPS in different mouse tissues differed ([98]
   Supplementary Figures S7A–E ). AS enrichment (top 20) showed that
   Spliceosome, Notch signaling pathway, and Mitophagy pathway were the
   most significantly enriched ([99] Supplementary Figures S8A–E ).

Figure 6.

   [100]Figure 6
   [101]Open in a new tab

   Gene expression heat map of significant GSEA pathway. Heat map for the
   (A) placenta (LPS vs. PBS) and (B) uterus (LPS vs. PBS) showing the
   clustered list of genes according to significant functions detected by
   GSEA with KEGG pathways. The figure uses a color scheme to represent
   the expression levels of protein-coding genes, with red indicating
   relatively high expression and blue indicating relatively low
   expression. GSEA, gene set enrichment analysis.

Validation of gene expression of the differentially expressed mRNAs by
qRT-PCR.

   To verify the accuracy of the RNA sequencing results, seven mRNA
   transcripts were randomly selected from the 152 significant overlapping
   DEGs (P < 0.05) and validated through qRT-PCR. Total RNA was extracted
   from the mouse maternal-fetal interface (placenta, chorioamniotic
   membrane, decidua and uterus) with LPS or PBS treatment, and cDNA was
   synthesized. The expression levels of the seven mRNAs were measured
   using TB Green qPCR Master Mix (Takara, Japan). The results indicated
   that the mRNA levels of ICAM1, CCL2, IL-1β, IL-6, STAT3, and IRF1 were
   significantly upregulated in the placenta, chorioamniotic membrane,
   decidua, and uterus, which was consistent with the RNA-seq data ([102]
   Figures 7A–D ). However, there was no significant difference in the
   expression of MYD88 in the chorioamniotic membrane and decidua, which
   was inconsistent with the RNA-seq results ([103] Figures 7C, D ). The
   primer sequences used in the qPCR are listed in [104]Table 2 .

Figure 7.

   [105]Figure 7
   [106]Open in a new tab

   Validation of maternal-fetal interface DEGs. The transcriptomic results
   of ICAM1, CCL2, IL1-β, STAT3, IRF1, MYD88 and IL-6 were further
   confirmed by qPCR in the (A) placenta, (B) uterus, (C) decidua and (D)
   chorioamniotic membrane respectively (LPS vs. PBS, n=4 respectively).
   The expression trends of candidate mRNAs were consistent with the
   sequencing results. Data are presented as 2^−ΔΔct values (mean ± SEM).
   *P < 0.05, **P < 0.01 and ***P < 0.001; ns, not significant.

Table 2.

   Sequences of the primers used in qRT-PCR experiments.
   mRNAs Primer    Sequence(5’ to 3’)
   ICAM1 Forward GTGATGCTCAGGTATCCATCCA
         Reverse CACAGTTCTCAAAGCACAGCG
   CCL2  Forward TTAAAAACCTGGATCGGAACCAA
         Reverse GCATTAGCTTCAGATTTACGGGT
   IL-1β Forward GCAACTGTTCCTGAACTCAACT
         Reverse ATCTTTTGGGGTCCGTCAACT
   IL-6  Forward TAGTCCTTCCTACCCCAATTTCC
         Reverse TTGGTCCTTAGCCACTCCTTC
   STAT3 Forward CAATACCATTGACCTGCCGAT
         Reverse GAGCGACTCAAACTGCCCT
   MYD88 Forward TCATGTTCTCCATACCCTTGGT
         Reverse AAACTGCGAGTGGGGTCAG
   IRF1  Forward ATGCCAATCACTCGAATGCG
         Reverse TTGTATCGGCCTGTGTGAATG
   ACTB  Forward GGCTGTATTCCCCTCCATCG
         Reverse CCAGTTGGTAACAATGCCATGT
   [107]Open in a new tab

   qPCR, real-time fluorescent quantitative PCR.

The proportions of distinct immune cell subpopulations

   To investigate immune cell infiltration in different mouse tissues
   (placenta, chorioamniotic membrane, decidua, and uterus), we analyzed
   22 immune cell phenotypes in the RNA-seq data of the CIBERSORT LM22
   file. The analysis revealed that LPS treatment increased the proportion
   of monocytes in the placenta, mast cells activated, dendritic cells
   resting, macrophages M0, and NK cells resting in blood; NK cells
   resting in chorioamniotic membrane; mast cells activated and T cells
   CD4 memory activated in decidua; and NK cells, mast cells activated,
   macrophages M1, and T cells CD4 memory activated in the uterus ([108]
   Figures 8A–E ).

Figure 8.

   [109]Figure 8
   [110]Open in a new tab

   The proportions of distinct immune cell subpopulations showed by
   CIBERSORT. (A) Placenta (LPS vs. PBS). (B) Peripheral blood (LPS vs.
   PBS). (C) Chorioamniotic membrane (LPS vs. PBS). (D) Decidua (LPS vs.
   PBS). (E) Uterus (LPS vs. PBS). Histogram showing the CIBERSORT
   estimation of immune cell infiltration based on the LM22 file.
   Different colors represent different types of immune cells.

Placenta and uterus metabolomics analysis

   Histogram and heatmap analysis of the differentially expressed
   metabolites (DEMs) showed significant changes in metabolites in mouse
   placenta and uterus after LPS intervention, including Curcumin,
   Humulone, L-dihydroorotate, PGF2α, and Nicotinamide N-oxide ([111]
   Supplementary Figures S9 , [112]S10A–D ). Further analysis showed that
   the DEMs mainly belonged to organic oxygen compounds, lipids and
   lipid-like molecules, organic acids and derivatives, and organic
   nitrogen compounds ([113] Figures 9A–D ). Additionally, a combined
   analysis of DEMs in the placenta and uterus found that Nicotinamide
   N-oxide, N-acetyl-d-lactosamine, His-Asp, 5,8-dimethylquinolin-4-ol,
   L-dihydroorotate, Trans-3’-hydroxycotinine o-.beta.-d-glucuronide,
   Humulone, and Pentobarbital showed significant changes in both mouse
   placenta and uterus after LPS intervention ([114] Table 1 ;
   [115]Supplementary Table S4 ).

Figure 9.

   [116]Figure 9
   [117]Open in a new tab

   Analysis of fold change of DEMs. (A) Placenta (POS-LPS vs. PBS). (B)
   Placenta (NEG-LPS vs. PBS). (C) Uterus (POS-LPS vs. PBS). (D) Uterus
   (NEG-LPS vs. PBS). DEMs were examined by OPLS-DA, and variable
   importance in projection (VIP > 1) and t-test (P < 0.05) were used for
   screening. Green represents downregulated DEMS, and red represents
   upregulated DEMS. Different colors of ordinate represent different
   superfamilies of DEMs. OPLS-DA, orthogonal partial least squares
   discriminant analysis; VIP, variable importance for the projection;
   POS, positive; NEG, negative; DEMs, differentially expressed
   metabolites.

   KEGG enrichment analysis found that Thyroid hormone synthesis,
   Lysosome, NOD-like receptor signaling pathway and Protein digest and
   absorption pathway were significantly enriched ([118] Supplementary
   Figure S11A, B ). In addition, molecular interaction network analysis
   was performed using the DEMs (OPLS-DA VIP>1 and P value < 0.05)
   uploaded to the online STRING database, and Cytoscape software was used
   for further analysis. CytoHubba was utilized to screen for the key
   metabolites, and the results showed that the interaction of key
   metabolites in the placenta was more obvious ([119] Supplementary
   Figures S12A–D ).

Integrative analysis of the metabolome and transcriptome

   DEGs and DEMs with significantly altered levels (P adj < 0.05 and
   |log2(FC)| > 1 and OPLS-DA VIP>1 and P value < 0.05) were analyzed, and
   Spearman correlation analysis revealed a significant correlation
   between them ([120] Supplementary Figures S13A, B ; [121]Table S7 ).
   KEGG pathway analysis showed that DEGs and DEMs were both involved in
   the Insulin resistance pathway in the placenta and the NOD-like
   receptor signaling pathway in the uterus ([122] Figures 10A, B ;
   [123]Supplementary Figures S14 , [124]S15A, B ). In addition, DEGs and
   DEMs commonly participated in the Prolactin signaling pathway, Ovarian
   steroidogenesis, and Calcium signaling pathway ([125] Supplementary
   Figures S15C, D ). Overall, there were 36 and 42 common KEGG pathways
   of DEGs and DEMs in the uterus and placenta, respectively ([126]
   Figures 10C, D ). Correlation network analysis was performed to assess
   potential interactions between DEGs and DEMs with LPS intervention, and
   the results showed that differential inflammatory-related factors
   (IL1-β and NLRP3) or chemokines (CCL3) interacted with DEMs ([127]
   Supplementary Figures S16A, B ).

Figure 10.

   [128]Figure 10
   [129]Open in a new tab

   Venn diagram and histogram of pathways involved by both DEGs and DEMs.
   (A) Placenta (LPS vs. PBS) and (B) Uterus (LPS vs. PBS). Venn diagram
   of pathways in which the DEGs and DEMs are involved. Different color
   circles represent different omics, in which blue represents
   transcriptomic and yellow represents metabolomic; the values in the
   circle represent the common/unique pathways of the two omics. (C)
   Placenta (LPS vs. PBS) and (D) Uterus (LPS vs. PBS). Histogram of
   pathways in which the DEGs and DEMs are involved (TOP10). Each column
   in the figure represents a KEGG pathway, with different colors
   representing different omics, among which blue represents
   transcriptomic, and orange represents metabolomic. The number is ranked
   from high to low. The higher the column, the more DEGs or DEMs the
   biological pathway was annotated.

Discussion

   In the present study, we comprehensively explored the tissue-dependent
   effects of acute exposure to LPS via transcriptomics and metabolomics
   analysis in mice placenta, chorioamniotic membrane, decidua, uterus and
   peripheral blood. The results revealed a total of 152 common DEGs (P <
   0.05) in mice placenta, chorioamniotic membrane, decidua, uterus and
   peripheral blood, while 8 common DEMs (P < 0.05) were found in mouse
   placenta and uterus after LPS exposure. We further investigated the
   pathways enriched in these common DEGs and DEMs and found that they
   were both enriched in the NOD-like receptor signaling pathway, while
   apoptosis, oxidative phosphorylation and hormone synthesis pathways
   showed tissue-specific correlations after LPS intervention.
   Additionally, we identified interactions between DEGs and DEMs,
   suggesting their combined role in the LPS-induced PTB model of mice. We
   also observed alterations in upstream regulators and master regulators
   important for inflammatory reactions, such as STAT1, STAT3, and NFKB1,
   in LPS-exposed tissues.

   PTB is a multifaceted syndrome with various potential underlying
   pathophysiologic mechanisms. Increased uterine contractility, cervical
   ripening and rupture of chorioamniotic membranes are the common
   pathways of both TB and PTB ([130]3, [131]14, [132]23). Placental
   insufficiency, including conditions such as placental abruption,
   chronic chorioamnionitis and twin pregnancy, appears to be responsible
   for many PTB cases, emphasizing the placenta’s significant role in PTB
   pathogenesis ([133]24). In a study by Lien et al., metabolomics and
   transcriptomics were combined to demonstrate that intrauterine
   inflammation can lead to placental dysfunction in mice by altering
   mitochondrial function and energy metabolism ([134]25). Similarly, our
   study showed significant changes in transcriptomics and metabolomics
   profiles in mice with an intraperitoneal injection of LPS. Apart from
   the inflammation-related NF-kappa B signaling pathway and TNF signaling
   pathway, we also observed significant changes in the apoptosis pathway,
   metabolic pathway and oxidative phosphorylation pathway, suggesting
   that inflammation may lead to placental dysfunction by promoting
   placental cell apoptosis, increasing placental oxidative stress and
   metabolic disorders. Mitochondrial dysfunction, caused by increased
   oxidative stress and inflammatory response, is a key factor in the
   pathophysiology of human pregnancy, including PTB, which is strongly
   associated with placental tissue rich in mitochondria ([135]26,
   [136]27).

   Uterine contraction is a defining characteristic of labor, and most
   drugs, such as progesterone, betamimetics, calcium channel blockers,
   magnesium sulfate, oxytocin receptor antagonists, nitric oxide donors,
   and others, used in clinical practice aim to inhibit uterine
   contraction. However, these drugs have limitations in terms of safety
   and clinical application ([137]28, [138]29). Humans and animals (i.e.,
   mice) share a high degree of similarity in terms of uterine
   contraction. Through transcriptomic and metabolomic analysis, we found
   significant changes in the cytokine-cytokine receptor interaction,
   calcium signaling pathway, metabolic pathways and steroid hormone
   biosynthesis pathway in the mice uterus following intraperitoneal
   injection of LPS. Additionally, activation of the decidua is an
   important feature of both the TB and PTB process ([139]30). The decidua
   contains a large number of immune cells that can produce cytokines.
   While the number of decidual macrophages increased in both TB and PTB,
   the number of neutrophils increased significantly only in patients with
   PTB ([140]10). Prostaglandin E2 and F2 are also mainly synthesized in
   the decidua and are essential in promoting uterine contraction
   ([141]31). Our study found that intraperitoneal injection of LPS can
   cause decidual activation. The TNF signaling pathway, NOD-like receptor
   signaling pathway, Cytokine-cytokine receptor interaction, and
   Chemokine signaling pathway were significantly enriched. In addition to
   the decidua, the chorioamniotic membrane also plays a crucial role in
   TB or PTB, particularly during preterm premature rupture of membranes
   (PPROM) ([142]3). Studies have shown that both infectious and sterile
   inflammation can cause rupture of the chorioamniotic membrane and lead
   to PTB ([143]32). Moreover, chorioamniotic membrane senescence is also
   considered one of the possible signals for the onset of labor
   ([144]33). Although our study found that the chorioamniotic membrane
   was least affected by intraperitoneal injection of LPS, with only 898
   DEGs (P < 0.05), most of these DEGs were related to inflammatory
   response, including the TNF signaling pathway and NF-kappa B signaling
   pathway.

   Our study confirmed that inflammation induced by intraperitoneal
   injection of LPS could lead to PTB in mice. Several inflammatory
   factors, such as IL1β and IL6, as well as chemokines, such as CXCL1 and
   CCL2, exhibited significant differences in various tissues.
   Additionally, some transcription factors, such as IRF1 and STAT3,
   showed similar changes. Notably, the significance of the identified 152
   co-expressed DEGs in the placenta, chorioamniotic membrane, decidua,
   uterus and peripheral blood requires further investigation. Meanwhile,
   the involvement of genes, such as NLRP3, in the common enriched
   NOD-like receptor signaling pathway and their role in PTB has been
   previously reported ([145]34). Furthermore, our study found that
   intraperitoneal injection of LPS in mice led to significantly different
   changes in placental and uterine metabolites. However, several
   metabolites, including Nicotinamide N-oxide, N-acetyl-d-lactosamine,
   His-Asp, 5,8-dimethylquinolin-4-ol, L-dihydroorotate,
   Trans-3’-hydroxycotinine o-.beta.-d-glucuronide, Humulone and
   Pentobarbital, showed similar changes. Nicotinamide N-oxide, a major
   nicotinamide catabolite in mice, may help inhibit the excessive
   production of inflammatory cytokines ([146]35). N-acetyl-d-lactosamine
   is a Galectin-3 inhibitor that plays a role in immunosuppression
   mediated by Gal-3 and prostate cancer stem cell-like cells ([147]36).
   On the other hand, His-Asp is a downstream product of the cytokine
   signaling pathway, which may regulate the function of genes associated
   with plant cytokine signaling pathway ([148]37). L-dihydroorotate is an
   intermediate product of pyrimidine metabolism and a substrate of
   dihydroorotate dehydrogenase in mitochondria. Interestingly, Humulone
   can inhibit COX-2 expression through the NF-kappa B pathway and A-P1
   pathway after drug-induced mouse skin damage ([149]38). Metabolomics
   can detect the processes that have occurred and help us to further
   investigate our findings.

   Although the PTB model induced by intraperitoneal injection of LPS in
   late pregnancy of mice is a widely used method in research ([150]19,
   [151]39, [152]40), there were limitations to this study that should be
   acknowledged. First, our experiment began after the first neonatal
   mouse was delivered, and although the interval time was shorter than 3
   hours, we cannot rule out the effect of LPS action time on our
   experimental results. Second, considering that mice are multiparous,
   different pregnancy sites might have also interfered with our
   experimental results and the sample size was relatively small. Third,
   our study utilized a single dose of LPS for intraperitoneal injection
   to induce a mouse model of PTB, lacking exploration of different doses
   and durations of LPS exposure, and the method of intraperitoneal LPS
   injection for inducing PTB is less commonly used compared to
   ultrasound-guided intrauterine or intra-amniotic LPS injection. Lastly,
   there are significant differences between human and mouse pregnancy
   tissues and labor processes, and we must be cautious when referencing
   these results to human studies.

   In conclusion, this study presents a thorough investigation of
   differences in mRNAs and metabolites among various tissues in a mouse
   model of LPS-induced PTB. By integrating metabolomics and
   transcriptomics analyses, we have provided comprehensive evidence of
   the distinct molecular characteristics of the placenta, chorioamniotic
   membrane, decidua, uterus, and peripheral blood in this model. Our
   findings support the idea that PTB is an inflammatory syndrome, as most
   co-expressed DEGs and DEMs are linked to inflammation. A comprehensive
   analysis of these molecular changes could help identify common pathways
   and shed light on the underlying mechanisms of PTB initiation and labor
   in mice treated with intraperitoneal LPS. However, it is important to
   note that the specific role of these co-expressed DEGs and DEMs in
   triggering the inflammatory process associated with PTB or labor
   remains unclear, and further research is needed to elucidate their
   functions and potential clinical applications.

Data availability statement

   The datasets presented in this study can be found in online
   repositories. The names of the repository/repositories and accession
   number(s) can be found below: PRJNA917515 (SRA) and MTBLS6775
   (Metabolights).

Ethics statement

   The animal study was approved by Ethics Committee of Fudan University
   (Protocol Number: 20211203S). The study was conducted in accordance
   with the local legislation and institutional requirements.

Author contributions

   XC is responsible for the experimental part and literature writing. XZ
   is responsible for auxiliary revision. SC and CX are responsible for
   reviewing the revised manuscript. All authors contributed to the
   article and approved the submitted version.

Acknowledgments