Distinct preterm labor phenotypes have unique inflammatory signatures and contraction associated protein profiles

Abstract

Preterm labor (PTL) is the predominant cause of childhood morbidity and mortality. It has several phenotypes, each with a distinct etiology often involving inflammation. Here, in samples of reproductive tissues obtained in early PTL from women with phenotypically defined PTL, we examined the presence and distribution of inflammation and its relationship with prolabor gene expression. In chorioamnionitis (CA-PTL), cytokine protein concentrations were increased across all tissues; in idiopathic (I-PTL), the inflammatory changes were limited to the choriodecidua; inflammation was not a feature of placental abruption (PA-PTL). CA-PTL was associated with activation of p65 in the myometrium and AP-1 in the choriodecidua, and PA-PTL with CREB in the choriodecidua. In the myometrium, PGHS-2 mRNA level was increased in CA- and I-PTL; in the amnion, PGHS-2 mRNA level was higher in PA- and I-PTL, while in CA-PTL, OT, OTR mRNA, and CX-43 expression were increased. In the choriodecidua, PGHS-2 mRNA level was unchanged, but in CA and I-PTL, OT mRNA level were increased and OTR was reduced. These data show that CA-PTL is associated with widespread inflammation and prolabor gene expression. In contrast, in I-PTL, inflammation is limited to the choriodecidua, with discrete increases in PGHS-2 in the amnion and OT in the choriodecidua. Inflammation is not a feature of PA-PTL, which is associated with increased OT and OTR in the amnion.

Summary sentence: These data show distinct patterns of gene expression in different types of PTL and, consequently, have important implications for both therapy and our understanding of PTL etiology.

Introduction

Globally, spontaneous preterm labor (sPTL) is the leading cause of childhood mortality and is responsible for 20% of neonatal deaths annually [1, 2]. In singleton pregnancies, idiopathic PTL (I-PTL) is the most important cause of preterm labor (PTL) accounting for 60–70% of cases; infection, causing preterm rupture of membranes, is responsible for 25–30% of cases and placental abruption (PA-PTL) for 5–10% [2, 3]. This is consistent with a recent paper that examined a series of cases of spontaneous PTL with intact membranes and concluded that intra-amniotic inflammation is more commonly sterile [4]. Traditionally, I-PTL is thought to be the early occurrence of TL, and while the upstream pathways may differ, they share a common downstream pathway [5, 6]. Both TL and sPTL are thought to involve immune activation leading to up-regulation of cytokines and other inflammatory mediators [7], with some studies showing a dramatic up regulation of cytokines and inflammatory infiltrate in the myometrium [8, 9], amnion [10], placenta [11], and choriodecidua [12, 13]. Most of these studies have grouped cases of sPTL, independent of cause, and compared the collected group to cases of term labor. A similar approach is used clinically, with all women with sPTL receiving the same management. This simplistic approach ignores the obvious fact that sPTL is not a single entity, but rather a syndrome with multiple causes including infection, stretch, placental dysfunction, and decidual hemorrhage [14]. Clinicians currently lack the ability to predict if the cause of sPTL is due to infection, stretch, or thrombosis, but knowing this would allow us to identify those babies who need to be delivered, removing them from a hostile infective environment, and those who would benefit from remaining in-utero to avoid the complications of prematurity. Our efforts to achieve this have been hampered by a poor understanding of the underlying pathophysiology of each of the distinct sPTL phenotypes. This is particularly true of inflammation, which is known to be present in infection [15] and in stretch-induced PTL [16], but it is uncertain whether it is present in I-PTL or in PA-PTL. Further, the exact sites of inflammation in all forms of PTL and its relationship to the expression of contraction-associated proteins are yet to be elucidated. Understanding whether inflammation is involved and its tissue distribution in the different forms of PTL will suggest potential etiologies and may identify potential therapeutic targets to delay or prevent sPTL.

Our hypothesis is that inflammation is most marked and widespread in chorioamnionitis (CA-PTL), but intermediate and with a restricted tissue distribution in I-PTL and least in PA-PTL. To test our hypothesis, we obtained matched myometrial, amnion, choriodecidual, and placental samples at the time of cesarean section from women in preterm no labor (PTNL), and in early labor (≤3cms) due to CA-PTL, I-PTL, and PA-PTL, and assessed tissue inflammation using, rtPCR, cytokine and transcription factor assays, and immunohistochemistry to assess immune cell infiltration of the myometrium.

Materials and methods

Tissue collection

The study received ethics approval from the London Chelsea Ethics committee. Informed consent was obtained from all women prior to any tissue collection. Labor was defined as the presence of regular uterine contractions every 3–4 minutes. To study characteristics of the inflammatory process, each of PTL phenotype and the inclusion and exclusion criteria used is summarized in Table 1. There was no use of exogenous oxytocin or evidence of failure to progress. We used matched samples collected from women, at the time of cesarean delivery, in early labor i.e. 3 cm cervix opening or less rather than established labor. Samples were obtained from 4 groups of women: PTNL (n = 13), CA-PTL (n = 13), I-PTL (n = 8), and PA-PTL (n = 7). Table 2 summarizes the clinical demographics of the each of the PTL phenotypes.

Table 1.

Definition of distinct PTL phenotypes and inclusion and exclusion criteria.

PTL phenotype	Definition
CA-PTL	Maternal pyrexia (temperature 38°C or more, on at least one occasion) with two or more of the following: foul smelling liquor, tachycardia (>100 bpm), fetal tachycardia (>160 bpm), uterine tenderness, raised maternal peripheral white cell count (>15000), and raised C-reactive protein.
PA-PTL	Preterm delivery due to placental abruption: the diagnosis was based on clinical symptoms of vaginal bleeding accompanied by abdominal pain, uterine tenderness, or tetanic contractions with abnormal fetal heart tracing and retroplacental clot at time of delivery.
I-PTL	No evidence of infection, rupture of membranes >48 hrs, placental abruption. Exclusion criteria: fibroids, multiple pregnancy, previous cervical surgery, mid-trimester loss, previous pelvic inflammatory disease, or sexually transmitted disease.
PTNL	Women not in labor prior to 37 weeks having an elective cesarean delivery for maternal or fetal reasons.
Inclusion criteria	All women in spontaneous preterm labor in the distinct preterm labor phenotypes above without any overlap in clinical signs or symptoms.
Exclusion criteria	Uterine anomaly, maternal infectious disease (hepatitis, HIV, syphilis), polyhydramnios, multiple pregnancies, uterine fibroids, previous pelvic infection, and previous cervical surgery.

Table 2.

Demographic table of the participants recruited in sPTL of distinct phenotype.

PTL phenotype*	PTNL	CA-PTL	I-PTL (early labor group)	I-PTL (establish labor group)**	PA-PTL
n	13	13	8	6	7
Maternal age (yrs; mean ± SD)	32.2 ± 5.7	34.2 ± 4.1	31.5 ± 7.5	33.5 ± 5.1	29.6 ± 5.2
Parity (n)
0	10	4	2	4	3
1	2	6	3	1	4
2	1	2	3
3		1		1
GA (wks; mean ± SD)	33.2 ± 5.7	29.0 ± 2.9	35.8 ± 0.9	33.6 ± 3.3	32.2 ± 3.2
Maternal BMI (mean ± SD)	26 ± 3.6	24.7 ± 5.1	24.25 ± 5.9	23.1 ± 4.3	22.7 ± 2.5
Number of women with rupture of membranes prior to cesarean delivery	N/A	12	3	3	N/A
Length of rupture of membranes prior to cesarean delivery (hrs; mean ± SD)	N/A	125.6 ± 148.7	18 ± 17.6	3.3 ± 1.7	N/A
Number of women who had antibiotic treatment prior to delivery	N/A	11	1	0	N/A
Duration of antibiotic use prior to delivery (hrs; mean ± SD)	N/A	74.6 ± 87.9	4	N/A	N/A

^*PTL phenotypes as defined in Table 1.

^**This group was only analyzed using immunohistochemistry.

All of the women received one dose of intravenous cephalosporin and metronidazole according to standard hospital policy. The women in PTNL had no history of preterm rupture of membranes (PROM), infection or signs of labor, and delivered early for pre-eclampsia, fetal growth restriction or both.

In the CA-PTL group, all of the women had PROM and clinical signs of chorioamnionitis. Five of the women were commenced on oral erythromycin and 2 women with intravenous antibiotics before delivery.

The women who presented in I-PTL did not have any clinical signs of infection and 6 of the women had PROM less than 12 hrs. Only one of the women received intravenous antibiotics before delivery for a positive vaginal Group B Streptococcus result.

From each woman, myometrium, placenta, choriodecidua, and amnion samples were collected at the time of cesarean section. The myometrial biopsies were taken from the upper aspect of the lower segment uterine incision at the time of cesarean section. The placental samples were taken randomly from sites between the umbilical cord insertion and the placental edge. They were taken >5 mm from the maternal surface to exclude any decidual contamination. The amnion and chorion were identified beyond the edge of the placenta and manually separated. The samples were taken midway between the tear-line and/or site of spontaneous rupture and the placental edge.

All the samples were collected and immediately frozen at −80°C. The placentae from the specific PTL phenotype were sent for histopathology to confirm the presence of chorioamnionitis in the CA-PTL groups, as well as to ensure no evidence of chorioamnionitis in the I-PTL and PA-PTL groups.

Ten milliliters of maternal peripheral blood was taken in an EDTA collection tube at the time of clinical diagnosis by the clinical team just prior to cesarean delivery. An optical method with laser flow cytometry (Abbott Alinity hsq) was used to calculate the total white cell, monocyte, and neutrophil count in the samples; where possible attempts were made to keep variation in samples numbers to a minimum across experiments. However, a few of the samples had to be excluded in some of the experiments as the desired lysate concentrations did not meet the requirements of the assays.

Total RNA extraction, cDNA synthesis, and rt-PCR of human samples

Myometrial, amnion, choriodecidual, and placental samples from 4 groups of women at PTNL, CA-PTL, I-PTL, and PA-PTL were homogenized in 700 μL of RNA stat using the Precellys24 Dual system (Table 3). The samples were placed in a Precellys tube type CK Mix #03961-1-00 (Scientific Laboratory Supplies LTD, UK) and spun at 5000 rpm for 20 seconds. Samples were then centrifuged on a bench top centrifuge at 4°C for 20 minutes. The supernatant was transferred to a new eppendorf and 150 μL of chloroform added and vortexed. The mixture was then centrifuged again on a bench top centrifuge at 4°C for 20 minutes. The supernatant was then transferred to a clean mini-vial for RNA extraction and purification (RNeasy mini kit, Qiagen Ltd, UK).

Table 3.

Demographic table for samples used for mRNA extraction.

PTL Phenotype	PTNL	Chorioamnionitis PTL	Idiopathic PTL	Abruption PTL
Amnion
n	9	13	8	6
GA (wks; mean ± SD)	33.8 ± 2.0	29.0 ± 2.9	35.8 ± 0.9	32.1 ± 3.8
Choriodecidua
n	7	7	8	6
GA (wks; mean ± SD)	33.7 ± 2.2	29.0 ± 3.2	35.8 ± 0.9	32.1 ± 3.8
Placenta
n	9	13	8	6
GA (wks; mean ± SD)	33.8 ± 2.0	29.0 ± 2.9	35.8 ± 0.9	32.1 ± 3.8
Myometrium
n	9	13	8	6
GA (wks; mean ± SD)	33.8 ± 2.0	29.0 ± 2.9	35.8 ± 0.9	32.1 ± 3.8

After RNA quantification, 1.0 μg was reverse transcribed with oligo dT random primers using MuLV reverse transcriptase (Applied Biosystems Ltd., Warrington, Cheshire, UK). Primer sets for genes (Table 4) were designed and obtained from Invitrogen Ltd. (Paisley, UK). Assays were validated for all primer sets by confirming that single amplicons of appropriate size and sequence were generated according to predictions. Quantitative PCR was performed in the presence of SYBR Green (Applied Biosystems Ltd.), and amplicon yield was monitored during cycling in a RotorGene Sequence Detector (Corbett Research Ltd., Mortlake, Sydney, Australia) that continually measures fluorescence caused by the binding of the dye to double-stranded DNA. Pre-PCR cycle was 10 minutes at 95°C followed by up to 45 cycles of 95°C for 20 seconds, 58–60°C for 20 seconds, and 72°C for 20 seconds followed by an extension at 72°C for 15 seconds. The final procedure involves a melt over the temperature range of 72–99°C rising by 1° steps with a wait for 15 seconds on the first step followed by a wait of 5 seconds for each subsequent step. The cycle in which fluorescence reached a pre-set threshold (cycle threshold) was used or quantitative analyses. The cycle threshold in each assay was set at a level where the exponential increase in amplicon abundance was approximately parallel between all samples. For each sample, the most stable housekeeping gene was determined by geNORM. The normalization factor was calculated as the average of the most stable housekeeping genes for each tissue. For the myometrium: average of GAPDH, CYC, and 18S; Placenta: average of GAPDH, CYC, BM2; Choriodecidua: average of GAPDH, CYC, β-actin; amnion: GAPDH. We then expressed each gene as a ratio of the normalization factor for each tissue. Due to the large number of samples, the PA-PTL samples were analyzed in a separate PCR cycle from the other PTL phenotypes.

Table 4.

All mRNA genes used for PCR including: forward and reverse primer sequences, GenBank/accession number and base pair lengths.

Genes	Forward primer sequence	Reverse primer sequence	GenBank/EMBL accession no.	Nucleotide base pairs
GAPDH	5′-TGATGACATCAAGAAGGTGGTGAAG-3′	5′-TCCTTGGAGGCCATGTAGGCCAT-3′	BC014085	239
18S	5′-AAACGGCTACCACATCCAAG-3′	5′-CCTCCAATGGATCCTCGTTA-3′	M10098	155
CYC	5′-CGGTGGCAAATTCAAGTCCT-3′	5′-TTCCTGAATTCTCCGGTGCA-3′	AF022115.1	247
IL-4	5′-TGAACAGCCTCACAGAGCAG-3′	5′-GCGAGTGTCCTTCTCATGGT-3′	M13982	151
IL-6	5′-CCTTCCAAAGATGGCTGAAA-3′	5′-AGCTCTGGCTTGTTCCTCAC-3′	NM000600	153
IL-8	5′-GCCTTCCTGATTTCTGCAGC-3′	5′-CGCAGTGTGGTCCACTCTCA-3′	NM000584	149
IL-10	5′-TGCCTTCAGCAGAGTGAAGA-3′	5′-GGTCTTGGTTCTCAGCTTGG-3′	AY029171	169
CXCL1	5′-GAAAGCTTGCCTCAATCCTG-3′	5′-GCCTCTGCAGCTGTGTCTCT-3′	NM001511	173
CXCL2	5′-CTGCTCCTGCTCCTGGTG-3′	5′-GCTTTCTGCCCATTCTTGAG-3′	NM002089	190
CCL2	5′-TCTGTGCCTGCTGCTCATAG-3′	5′-AGATCTCCTTGGCCACAATG-3′	X14768	202
CCL5	5′-CCATATTCCTCGGACACCAC-3′	5′-TGTACTCCCGAACCCATTTC-3′	NM002985	180
TNFα	5′-GAGAAGGGTGACCGACTCAG-3′	5′-GGTTGAGGGTGTCTGAAGGA-3′	NM000594	175
OTR	5_-AGAAGCACTCGCGCCTCTT-3	5_-AGGTGATGTCCCACAGCAACT-3	NM000916	101
Cx43	5′-TGGATTCAGCTTGAGTGCTG-3′	5′-GGTCGCTCTTTCCCTTAACC-3′	BC026329	205
PGHS-2	5′-TGTGCAACACTTGAGTGG CT-3′	5′-ACTTTCTGTACTGCGGGTGG-3′	AY151286	296

Bio-Plex Pro cytokine multiplex assays

Protein lysates were prepared from the frozen human tissues from women (mean gestational age ± SD in each case), at PTNL (33.8 ± 1.7 weeks, n = 11), CA-PTL (29.0 ± 3.1 weeks, n = 11), I-PTL (35.8 ± 0.8 weeks, n = 8), and PA-PTL (myometrium and amnion, 32.1 ± 3.8 weeks, n = 5; choriodecidua and placenta, 32.2 ± 3.2 weeks, n = 7). The demographic data is summarized in Table 2.

The lysates were prepared using the Bio-Plex Pro cell signaling reagent kit (BioRad, Hemel Hempstead, UK) and the Precellys24 Dual bead homogenizer system, according to the manufacturer’s instructions. Lysate concentrations were quantified by DC Protein Assay (BioRad, Hemel Hempstead, UK). About 500 μg of protein lysate was added per well to a Bio-Plex 19-plex (Table 5). Due to buffer incompatibility, an additional separate Bio-Plex CCL5 single-plex assay was performed for each tissue type from women at PTNL (33 ± 1.7 weeks, n = 10), CA-PTL (28.4 ± 3.2 weeks, n = 11), PA-PTL (32.1 ± 3.8 weeks, n = 5), and I-PTL (36.0 ± 0.8 weeks, n = 7). Samples were run in singlicate due to the number of samples available and to allow all samples to be run on the same plate, thereby eliminating interplate variation in our analyses. Appropriate standards and controls were provided with the assays and both were completed in accordance with the manufacturer’s instructions.

Table 5.

List of Bio-Plex analytes included in the custom-made Bio-Plex 19-plex assay. CCL5 (^*) was analyzed in a separate assay due to buffer incompatibility.

Bio-Plex Analytes
IL-1β	IL-10	CXCL1	CCL20
IL-2	IL-16	CXCL2	CCL17
IL-4	IFNγ	CXCL6	CCL25
IL-6	TNFα	CCL1	CCL2
IL-8	CX3CL1	CCL7	CCL5*

Transcription factor assay

Relative levels of AP-1, Phospho-p65, and Phospho-CREB in the myometrial and choriodecidual samples were measured using TransAMTM NFκB and TransAMTM AP-1 transcription factor DNA-protein binding assays (Active Motif, Carlsbad CA, USA) from women (mean gestational age ± SD in each case) at PTNL (33.8 ± 1.7 weeks, n = 13), CA-PTL (n = 11), I-PTL (n = 7), and PA-PTL (n = 7). Whole cell lysates were prepared using a Precellys24 bead homogenizer (Stretton Scientific Ltd, UK), with the Active Motif Nuclear Extraction Kit, in accordance with the manufacturer’s instructions for preparations from frozen tissues. Protein concentrations were quantified using a DC Protein Assay (Bio-Rad, Hemel Hempstead, UK). About 100 μg and 200 μg of protein lysate was added per sample well for the TransAMTM AP-1 and TransAMTM NFκB, respectively, and diluted in the appropriate individual transcription factor assay lysis buffer. Assays were completed according to the manufacturer’s instructions.

Immunohistochemistry

A subset of the myometrial biopsies (randomly chosen depending on the availability of tissue specimens) were used for histological analyses from the 3 groups of women PTNL (n = 8), I-PTL (n = 10), and CA-PTL (n = 10). Biopsies were immersion fixed in 4% paraformaldehyde, embedded in paraffin, sectioned, and stained by immunohistochemistry with antigen retrieval for CD68 macrophages as previously described (PMID26608218, 28765265) or neutrophil elastase (1:200 NP57; Dako Agilent, Santa Clara Ca). Two stained sections of the entire biopsy were scanned (Aperio scanner and ImageScope software, Leica) and analyzed for each antigen to compare resident macrophages and neutrophils among patients. Photomicrographs of 3–4 fields were taken at 40× and compared within and among groups.

Statistical analysis

All data were initially tested for normality using a Kolmogorov–Smirnoff test. Data were not normally distributed and analyzed using a Wilcoxon matched pair test for paired data and when comparing three groups or more a Friedman’s test, with a Dunn’s multiple comparisons post-hoc test. P < 0.05 was considered statistically significant.

Results

Pro-inflammatory cytokine concentrations in myometrium, amnion, choriodecidua, and placenta relative to preterm no labor phenotype

Inflammatory cytokine levels were increased in all tissues in CA-PTL only (P < 0.05–0.0001; Figure 1A–L). IL-1β levels were highest in the amnion, intermediate in the choriodecidua, and lowest in myometrium and placenta. IL-6 and TNFα levels were similar in the amnion and choriodecidua and lowest in myometrium and placenta (Figure 1E–L).

Figure 1.

TNF-α, IL-6 and IL-β cytokines. Myometrium, amnion, choriodecidual, and placental samples were obtained from four groups of women at the time of cesarean section from women (mean gestational age ± SD in each case), at preterm no labor (PTNL; 33.2 ± 1.7 weeks, n = 11), chorioamnionitis (CA-PTL; 29.0 ± 3.1 weeks, n = 11), placental abruption (PA-PTL; myometrium & amnion 32.1 ± 3.8 weeks, n = 5/choriodecidua & placenta 32.1 ± 3.8 weeks, n-7), and idiopathic (I-PTL; 35.8 ± 0.9 weeks, n = 8). After the protein content of the samples for the measurement of the cytokines was established by serial dilutions, protein from tissue homogenates were used for a multiplex assay. Tissue levels of the cytokines were quantified using Bio-plex human cytokine 19-plex-array kit (Bio-rad) and a separate assay for CCL5. The cytokine concentrations in the tissue protein lysate were calculated using an 8-point calibration curve for each individual analyte produced from manufacturer-supplied set standards of known concentration and presented as concentrations (pg/ml). The data was not normally distributed and comparison between the groups was performed using the Kruskal-Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^* is P < 0.05, ^** is P < 0.01, ^*** is P < 0.001, and ^**** is P < 0.0001 (IL-β, IL-6, and TNF-α).

Inflammatory cytokine mRNA levels were less consistent across reproductive tissues. IL-1β mRNA level was not increased in the myometrium, choriodecidua, and placenta with any type of PTL (Supplementary Figure S2). In the amnion, IL-1β mRNA level was increased in PA-PTL, but reduced CA-PTL and I-PTL (P < 0.05–0.01; Supplementary Figures S2B and S4B). For IL-6, mRNA level was increased in the myometrium for both CA-and I-PTL (P < 0.01 and 0.05 respectively; Supplementary Figure S2E) and in the choriodecidua for both CA- and PA-PTL (P < 0.05–0.01; Supplementary Figures S2G and S4B). For TNFα, mRNA level was increased in the amnion in CA- and PA-PTL (P < 0.05 and 0.001; Supplementary Figures S2J and S4C), the choriodecidua in CA- and PA = PTL (P < 0.01; Supplementary Figures S2K and S4C), but reduced in I-PTL (P < 0.01; Supplementary Figure S2J).

Anti-inflammatory cytokine mRNA and protein concentrations in myometrium, amnion, choriodecidua, and placenta relative to preterm no labor phenotype

Anti-inflammatory cytokine levels (IL-4 and IL-10) were increased in all tissues in CA-PTL, except the amnion, where IL-4 levels were reduced (P < 0.05–0.0001; Figure 2A–H). IL-4 and IL-10 levels were highest in the choriodecidua (Figure 2C and F). There was no change in IL-4 or IL-10 protein levels in other types of PTL. IL-4 mRNA level was reduced in the amnion in CA-PTL and IL-10 mRNA increased in the choriodecidua in PA-PTL (P < 0.5; Supplementary Figures S1B and S4E).

Figure 2.

IL-10, IL-4. Myometrium, amnion, choriodecidual, and placental samples were obtained from four groups of women at the time of cesarean section from women (mean gestational age ± SD in each case), at preterm no labor (PTNL; 33.2 ± 1.7 weeks, n = 11), chorioamnionitis (CA-PTL; 29.0 ± 3.1 weeks, n = 11), placental abruption (PA-PTL; myometrium & amnion 32.1 ± 3.8 weeks, n = 5/choriodecidua & placenta 32.1 ± 3.8 weeks, n-7), and idiopathic (I-PTL; 35.8 ± 0.9 weeks, n = 8). After the protein content of the samples for the measurement of the cytokines was established by serial dilutions, protein from tissue homogenates were used for a multiplex assay. Tissue levels of the cytokines were quantified using Bio-plex human cytokine 19-plex-array kit (Bio-rad) and a separate assay for CCL5. The cytokine concentrations in the tissue protein lysate were calculated using an 8-point calibration curve for each individual analyte produced from manufacturer-supplied set standards of known concentration and presented as concentrations (pg/ml). The data was not normally distributed and comparison between the groups was performed using the Kruskal-Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^* is P < 0.05, ^** is P < 0.01, ^*** is P < 0.001, and ^**** is P < 0.0001 (IL-4 and IL-10).

Chemokine mRNA and protein concentrations in myometrium, amnion, choriodecidua, and placenta relative to preterm no labor phenotype

Neutrophil chemokines (IL-8, CXCL1 and CXCL2) levels were increased in all tissues in CA-PTL (P < 0.05–0.0001; Figure 3I–T). CXCL2 was also increased in the placenta in I-PTL (P < 0.05; Figure 3P). Monocyte chemokines (CCL2 and CCL5) were less consistent, both only increased in CA-PTL again, CCL5 only in the choriodecidua (P < 0.001, Figure 3C), and CCL2 in the amnion and choriodecidua (both P < 0.0001, Figure 3F and G).

Figure 3.

IL-8, CXCl2, CXCL1, CCL2,CCL5. Myometrium, amnion, choriodecidual, and placental samples were obtained from four groups of women at the time of cesarean section from women (mean gestational age ± SD in each case), at preterm no labor (PTNL; 33.2 ± 1.7 weeks, n = 11), chorioamnionitis (CA-PTL; 29.0 ± 3.1 weeks, n = 11), placental abruption (PA-PTL; myometrium & amnion 32.1 ± 3.8 weeks, n = 5/choriodecidua & placenta 32.1 ± 3.8 weeks, n-7), and idiopathic (I-PTL; 35.8 ± 0.9 weeks, n = 8). After the protein content of the samples for the measurement of the cytokines was established by serial dilutions, protein from tissue homogenates were used for a multiplex assay. Tissue levels of the cytokines were quantified using Bio-plex human cytokine 19-plex-array kit (Bio-rad) and a separate assay for CCL5. The cytokine concentrations in the tissue protein lysate were calculated using an 8-point calibration curve for each individual analyte produced from manufacturer-supplied set standards of known concentration and presented as concentrations (pg/ml). The data was not normally distributed and comparison between the groups was performed using the Kruskal-Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^* is P < 0.05, ^** is P < 0.01, ^*** is P < 0.001, and ^**** is P < 0.0001 (IL-8, CXCL2, CXCL1, CCL2, and CCL5).

Inflammatory chemokine mRNA levels were also inconsistent across reproductive tissues. Neutrophil chemokine (IL-8, CXCL1 and CXCL2) mRNA levels were increased in CA-PTL, I-PTL PA-PTL: in CA-PTL, IL-8 was increased in the myometrium and choriodecidua (P < 0.05–0.001; Supplementary Figure S3Q and S), CXCL1 was increased in both amnion and choriodecidua (P < 0.05–0.01; Supplementary Figure S3J and L), and CXCL2 was increased in the myometrium and amnion only (P < 0.5 and P < 0.0001; Supplementary Figure S3M and N). In I-PTL, CXCL1 and CXCL2 were increased in the placenta (P < 0.05 and P < 0.01.; Supplementary Figure 3L and P). In PA-PTL, IL-8 was increased in the amnion (P < 0.05); Supplementary Figure S4J) and CXCL2 increased in the amnion, choriodecidua, and placenta (P < 0.01, P < 0.01, P < 0.1; Supplementary Figure S4I). For the monocyte chemokines, only CCL2 was increased in the myometrium in CA-PTL (P < 0.05; Supplementary Figure S3A) and reduced in the placenta in PA-PTL (P < 0.01; Supplementary Figure S4G).

Other chorioamnionitis

Other cytokines were also measured in the multiplex, the results are summarized in Figure 4A–D, where we show the multiple of the median (MoM) for each analyte as a measure of inflammation in each tissue. We found that CA-PTL was associated with an increased cytokine MoM of cytokines across all tissues (P < 0.05–0.0001; Figure 4A–D). In I-PTL, the cytokine MoM was increased in the choriodecidua (P < 0.001; Figure 4B); and in PA-PTL, the cytokine MoM was actually reduced in the amnion (P < 0.05; Figure 4B). Individual cytokines are shown in Supplementary Tables S1–S4.

Figure 4.

A–D. The MOM were used to compare the inflammatory changes in the myometrium and choriodecidua during sPTL in CA-PTL and I-PTL. The median value of PTNL was used to normalize the data across the comparison groups (CA-PTL and I-PTL). The median ranges were then used for all the 20 cytokines and comparisons between 2 or more groups were performed using the Kruskal–Wallis test with a Dunn’s Multiple Comparison post-hoc test as the distribution of the data was non-parametric. In the graphs, the level of significance is demonstrated by ^* is P < 0.05, ^** is P < 0.01, ^*** is P < 0.001, and ^*** is P < 0.0001.

Transcription factor activation in the preterm myometrium and choriodecidua

We assessed the activity of the NFκB, MAPK/AP-1, and CREB pathways in the preterm myometrial and choriodecidual tissues in the PTL phenotypes. Interestingly, the choriodecidual AP-1 binding activity was lower in CA-PTL samples (P < 0.001; Figure 5B) and there was a non-significant trend for a reduction in I-PTL too. NFκB binding activity was increased in the myometrium in CA-PTL (P < 0.05; Figure 5C) and pCREB binding activity was increased in the choriodecidua in PA-PTL (P < 0.01, Figure 5F).

Figure 5.

A–F. Choriodecidual and myometrial samples were obtained from four groups of women at the time of cesarean section from women (mean gestational age ± SD in each case preterm no labor (PTNL; 33.2 ± 1.7 weeks, n = 13), chorioamnionitis (CA-PTL; 29.0 ± 3.1 weeks, n = 11), placental abruption (PA-PTL; 32.1 ± 3.8 weeks, n = 7)), and idiopathic (I-PTL; 35.8 ± 0.9 weeks, n = 7). The samples were homogenized and the levels of phospho-p65, phospho-c-Jun, and phospho-CREB were measured using TransAMTM NFκB, AP-1 and CREB transcription factor DNA-protein binding assays (Active Motif, Carlsbad CA, USA). The data were not normally distributed and analyzed using Kruskal–Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^* is P < 0.05, ^** is P < 0.01, and ^*** is P < 0.001.

Contraction-associated protein mRNA levels

Comparisons were made between mRNA levels in the PTNL group and the PTL phenotypes across tissues. In the myometrium, PGHS-2 mRNA levels were increased in CA and PA-PTL (P < 0.01 and P < 0.05 respectively, Figure 6A). In the amnion, PGHS-2 mRNA levels were increased in PA and I-PTL (both P < 0.05; Figure 6B), OT was increased in CA-PTL (P < 0.0001; Figure 6F), OTR was reduced in CA-PTL (P < 0.0001, Figure 6J–K) and there was a non-significant trend for a reduction in I-PTL too, and Cx-43 was reduced in all forms of PTL, but significantly, only in I-PTL (P < 0.0001, Figure 6N). In the choriodecidua, OT was increased in I-PTL (P < 0.05; Figure 6G) and OTR was reduced in CA-PTL and I-PTL (Figure 6K, P < 0.05). No changes in contraction-associated protein mRNA levels were found in the placenta or mRNA levels of PGDH in the choriodecidua and placenta among PTL phenotypes (Supplementary Figure S5).

Figure 6.

A–P. Myometrial, amnion, choriodecidua, and placental samples were obtained from four groups of women at the time of cesarean section and divided into the following groups (mean gestational age ± SD in each case): preterm no labor (PTNL; 33.2 ± 1.7 weeks, n = 9–10), chorioamnionitis (CA-PTL; 29.0 ± 3.1 weeks, n = 7), placental abruption (PA-PTL; 32.1 ± 3.8 weeks, n = 6), and idiopathic (I-PTL; 35.8 ± 0.9 weeks, n = 8–9). The samples were homogenized and RNA extracted and converted to cDNA. Copy numbers of PGHS-2, OTR, OT, and CX43 mRNA for PTNL, CA-PTL, PA-PTL, and I-PTL samples were measured using quantitative rtPCR. The data was not normally distributed and comparison between the groups was performed using the Kruskal–Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^* is P < 0.05 and ^** is P < 0.01.

Myometrial inflammatory cell infiltration and maternal peripheral blood white blood cell distribution

In peripheral blood samples from women in PTNL and I-PTL groups, maternal WBCs, as well as monocytes and neutrophils were consistently low (Figure 7) compared to high counts in women with CA-PTL. In sections from uterine biopsies, macrophages were clearly visible, stained brown, and associated with a methyl green-stained cell nucleus (Figure 8). Typically found around smooth muscle bundles, macrophages were sparsely and sporadically distributed in biopsies sections from women in the PTNL (7/8) and I-PTL (8/9) groups. Sections from one patient in each of these groups, contained a few clusters of macrophages around muscle bundles or near blood vessels. In these two patients, there was no evidence of membranes rupture, infection, or sepsis. By contrast, in 8 out of 10 sections from women in the CA-PTL group, macrophages were present in greater abundance both around smooth muscle bundles more clusters observed in similar tissue regions. All women in the CA-PTL group had PPROM. Moreover, the greatest prevalence of macrophages was found in a uterine biopsy section from the woman with the longest period between membrane rupture and delivery, 79 hrs. In a separate set of sections, brown-stained neutrophils were found to be very sparse and distributed at random around smooth muscle bundles and predominantly in or near blood vessels across women irrespective of PTL phenotype (data not shown).

Figure 7.

A–C. Maternal peripheral blood count of total white blood cells, monocytes and neutrophils from women in PTNL (33.2 ± 1.7 weeks, n = 8), CA-PTL (29.0 ± 3.1 weeks, n = 8), and I-PTL (35.8 ± 0.9 weeks, n = 10) just prior to delivery. The data was not normally distributed and comparison between the groups was performed using the Kruskal–Wallis test with a Dunn’s multiple comparisons post-hoc test. The data are shown as median with interquartile range. The P values are demonstrated by ^** is P < 0.01 and ^*** is P < 0.001.

Figure 8.

Photomicrographs of macrophages (CD68-stained brown) and cell nuclei counterstained with methyl green in a biopsy of uterus from a woman with PTNL, I-PTL, or CA-PTL. Inset is magnified area in box. Scale bar = 100 μm or 25 μm in inset.

Discussion

This study investigated and compared the inflammatory changes at the onset of sPTL in distinct PTL phenotypes. We found that CA-PTL was associated with increased markers of inflammation across all reproductive tissues and evidence of transcription factor activation in both myometrium and choriodecidua. Increased presence of macrophages in uterine biopsies from most women in the CA-PTL suggests a more extensive inflammation beyond fetal membranes and decidua [17, 18]. I-PTL was associated with some inflammatory changes confined to the choriodecidua and PA-PTL was not associated with any inflammatory changes. PGHS-2 mRNA levels were increased in the myometrium and amnion in both CA- and I-PTL, while OT mRNA levels were increased in the amnion in CA- and PA-PTL, and in the choriodecidua in I-PTL, suggesting that both major prolabor systems are involved in the onset of PTL, albeit in different locations, perhaps reflecting the distinct mechanisms involved. This implies that there are different mechanistic inflammatory pathways that drive PTL and this is different to term labor in which the inflammatory changes in the myometrium occur as a consequence rather than cause of labor [19]. Our findings are summarized in Figure 9.

Figure 9.

Comparison of the distinct PTL phenotypes.

In CA-PTL, inflammation is widespread and associated with increased PGHS-2 mRNA in the myometrium and OT in the amnion. In contrast, inflammation in I-PTL is confined to the choriodecidua and only OT mRNA levels increased in the same location, suggesting that choriodecidual inflammation may drive the onset of I-PTL. PA-PTL is not associated with inflammation, but with an increase in OT mRNA levels in the amnion only. Our current data suggest that there may be few differences in the etiology of I-PTL and TL, with similar patterns and extent of distribution of markers of inflammation. In our unpublished work, we found that there were increased numbers of NK cells in the choriodecidua from women in early labor, and we are currently examining whether this occurs in early I-PTL to determine if the both share common mechanistic pathways at the onset of labor.

Chorioamnionitis PTL

Chorioamnionitis PTL is the major cause of early preterm birth and mid-trimester pregnancy loss occurring in approximately 94.4% of women who deliver between 21–24 weeks of pregnancy [18, 20–22]. We have shown that in women presenting in sPTL with clinical signs of chorioamnionitis, there is an associated intense inflammatory reaction across all reproductive tissues including the myometrium driven by leucocyte infiltration and pro-inflammatory cytokine release. Interestingly, the increase in cytokines was most consistent in the choriodecidua, consistent with the notion that most are ascending infections, and suggesting that the infection/inflammatory stimulus of CA is most marked in this tissue. There does seem to be a counter anti-inflammatory response mediated primarily by IL-10 and to a less consistently by IL-4. Both were elevated in all tissues with the exception of the amnion, where IL-4 was reduced. Interestingly, the levels of IL-1β were highest in the amnion compared to other tissues in CA-PTL, this may reflect the proximity of the amnion to the choriodecidua, but the levels of the other cytokines were similar to other tissues. IL-4 may directly repress IL-1β synthesis and, consequently, the higher levels of IL-1β in the amnion be a direct reflection to the lower levels of IL-4. The lower levels of IL-4 also imply that there is a differential regulation of IL-4 expression in the amnion, which has not been reported before. Amnion and choriodecidua exposure to LPS increased IL-10 and a number of other cytokines, but not IL-4 [23]. Equally, human mesenchymal amnion cells increased IL-4 production in response to macrophage-activating lipoprotein-2 [24]. However, IL-4 reduces the production of most cytokines by activated monocytes, including IL-1β, IL-6, and TNF-α [25], since the production of IL-6 and TNFα was not differentially increased in the amnion, this suggests that the relative increase in IL-1β may not be related to the decrease in IL-4. In the other reproductive tissues, both IL-4 and Il-10 were increased, consistent with previous data that showed that both amnion and choriodecidual membranes secrete IL-10 in the presence of microbial infection [26]. IL-10 has also been shown to promote the anti-inflammatory ability of soluble ST2 aimed at inhibiting the production of pro-inflammatory cytokines and modulating the balance of the Th1/Th2 immune response [27].

The chemokine data suggest that CA-PTL primarily drives neutrophil and to a lesser extent monocyte recruitment. Neutrophil chemokine levels were consistently elevated across all tissues in CA-PTL, while monocyte chemokines were increased in the choriodecidua and amnion for CCL2 and only in the choriodecidua for CCL5. However, the immunohistochemistry data did not find increased numbers of neutrophils in uterine biopsies from women in the CA-PTL group compared to those from women in the PTNL group. Preterm models have shown that neutrophil activation is modulated by the microenvironment of the tissue [28] and certain bacterial infections such as Group B Streptococcus can induce neutrophil cell death within 4 hours in an in-vitro dose-dependent model [29]. It may be that in CA-PTL, at the time of inoculation, there is activation of the maternal innate immune response with increase in neutrophil chemokines across tissues. However, the activation and trafficking of neutrophils may be modulated by the type of bacteria and tissue specific. Given that we did not find an increase in the neutrophils in the myometrium, it maybe that there is trafficking in other tissues closer to the fetus such as the amnion, choriodecidua, and placenta aimed at reducing fetal infection and injury.

AP-1 activity was repressed in the choriodecidua with CA-PTL, counter to our expectation; this probably reflects earlier activation of the AP-1 system as the AP-1 system is down-regulated after activation in myometrial cells [30]. There was a similar non-significant reduction in I-PTL too, suggesting that AP-1 may have been activated in the choriodecidua earlier in the I-PTL process. NFκB binding activity was increased in the myometrium and similarly in the choriodecidua in CA-PTL, this would be consistent with the increase in PGHS-2 mRNA, as NFκB is an important regulator of PGHS-2 expression in myometrial cells [30]. CREB binding was increased in the choriodecidua of PA-PTL, with a trend present in I-PTL too. There was no associated increase in CAP mRNA in either condition in the choriodecidua, but PGHS-2 was increased in the amnion in PA-PTL and tended to increase in I-PTL, which might reflect a similar process in the amnion. In CA-PTL, multiple cytokines were consistently raised in all tissues (Figure 4) and the resultant intense inflammation has been associated with adverse perinatal outcomes such as necrotizing enterocolitis, periventricular leukomalacia, and intraventricular hemorrhage [31]. Compared to PTNL and I-PTL, women with CA-PTL had higher maternal white cell, monocyte, and neutrophil counts, and there was a greater density of monocytes but not neutrophils in the myometrium. These findings imply that CA-PTL is associated with an intense and widespread inflammatory pulse across all gestational compartments as well as the maternal circulation at the onset of PTL.

CAP gene expression was tissue dependent. CA-PTL is likely to provide the strongest stimulus to CAP gene expression, certainly in terms of inflammatory cytokines and down-stream transcription factors. For PGHS-2 gene expression, despite having greater inflammation in CA-PTL, there was no increase in the choriodecidua or amnion, rather PGHS-2 mRNA was increased in the myometrium, where we observed increased NFκB activity. In contrast, OT mRNA level were markedly increased in the amnion and lesser in the choriodecidua. OTR is reduced by exposure to oxytocin both at the protein level, by internalization [32], and in terms of gene expression [33]. Consistent with this, in the choriodecidua, OTR mRNA levels were reduced in both CA-PTL and I-PTL. In contrast, in the amnion, despite relatively greater OT mRNA levels, OTR mRNA levels were still increased. This may reflect temporal differences in OT gene expression, i.e., the increases in the amnion came after those in the choriodecidua. If the increase in OT mRNA is translated into protein, this could act on the myometrium to drive contractions as well as PGHS-2 expression [34] and hence PTL.

Idiopathic sPTL

We did not observe the same level and tissue distribution of inflammation in I-PTL as we did in CA-PTL, although the subjects were at the same stage of labor without systemic signs of maternal infection. Tissue cytokine levels were often increased in I-PTL, but because we used an ANOVA to compare PTL subtypes, the extremely high levels in CA-PTL group often obscured these increases. When we used multiple of the median, to compare cytokine levels across PTL subtypes in different tissues, it became clear that in I-PTL, there was significant inflammation in the choriodecidua in I-PTL. This is consistent with the trend for AP-1 activation in the choriodecidua in I-PTL. While there is no doubt that CA-PTL is infection mediated, it is unclear whether I-PTL is preceded by infection, albeit of a much milder severity than CA-PTL. Alternatively, we observed inflammation in the choriodecidua in early term labor (Singh et al, unpublished observation), meaning that a similar mechanism could underlie both I-PTL and TL. We are currently investigating this possibility. Intriguingly, I-PTL showed the most consistent changes in CAP, with increased PGHS-2 in the amnion and trend to increase in the myometrium, while OT mRNA level were increased in the choriodecidua, with repression of OTR consistent with activation and down regulation as described above.

Others have also attempted to understand the inflammatory profile of women who presented with I-PTL by studying maternal blood and or cervical biopsies with some reporting no increase inflammatory mediators [35] and others a significant increase in inflammation at the time of I-PTL [36]. These inconsistencies are likely to be a consequence of the use of different methodologies, different tissues at different time points in the labor process. In this study, we found that the inflammatory pulse in I-PTL was localized to the choriodecidua, where it may contribute to the cellular senescence described in chorioamniotic membranes from women in sPTL without infection [37]. Furthermore, compared to CA-PTL we neither observed any significant increase in monocytes and neutrophils in the maternal peripheral blood nor any increase of either cell type in myometrial biopsies obtained from women in I-PTL. Earlier studies found that myometrial infiltration was present in only about 21% of sPTL cases [38]. However, these studies did not differentiate between the PTL phenotypes and considered them as a single entity. Given that there does not seem to be a widespread and intense infective inflammatory milieu, I-PTL is similar to term labor in this respect and seems to immune-meditated rather than infection induced.

Abruption sPTL

At the onset of PA-PTL we did not observe an increase in the pro-, anti-, and chemotactic cytokines. However, the multiple of the median analysis showed that inflammation was actually reduced in the amnion and tended to be lower in other tissues too. One limitation of this study is the small sample size in the abruption sPTL group, making any firm conclusion difficult. However, it has been suggested that PA-PTL is likely to be a chronic process associated with chronic inflammation induced by early pregnancy bleeding [39], and that histological examination of the placenta can reveal evidence of underlying chorioamnionitis [40]. Our data would suggest that this is not true in the cases we studied, and it seems be an acute event that drives the onset of labor independent of any inflammatory process.

Limitations and conclusion

One of the limitations is that the gestational ages differ between the PTL phenotype groups. However, our data suggest that the underlying cause of PTL is likely to exert a greater effect than any difference in gestational age, as shown by the intense inflammation in CA-PTL as compared to I-PTL. This is a relative weakness of human studies compared to animal models, where timing in pregnancy and sampling after the stimulus to parturition can be tightly defined. In contrast, human samples are subject to variation as women will present in PTL due to different etiologies, at different gestations and stages of labor.

We have used cervical dilatation to define the stage of labor. This is a potential weakness, as the duration of contractions needed to reach 3 cm dilatation will vary depending on cervical ripening, or how easily the cervix dilates. It seems more likely that the cervix will be on average less mature in women experiencing PTL than at term, equally some of the processes causing myometrial activation may also promote cervical ripening. It is well recognized by clinicians that cervical ripening does not necessarily occur at the same time as myometrial activation, even at term. It would seem likely that women in sPTL may take longer to reach 3 cm than women at term. This may mean that using the 3 cm cut off for established labor may be less reliable in sPTL, perhaps accounting for the lack of change in myometrial OTR observed in this study, which has been reported by us to decline [41] and by others to remain unchanged [42]. This may also influence transcription factor activation, mRNA level, and protein levels. We saw that AP-1 activity was reduced in CA-PTL, consistent with earlier activation [43]; similarly the lower levels of some mRNA when protein levels were high, for example in the case of IL-1β in the amnion, may reflect the existence of a negative feedback loop. These data show the need for a broad approach to the evaluation of laboring samples, looking not at one tissue, but at different points in the labor process.

In this study, the women in CA-PTL had the longest duration of rupture of membranes, prolonged antibiotic usage, and earlier gestational age compared to the I-PTL, PA-PTL, and PTNL group. As expected, the CA-PTL group had an intense inflammatory profile compared to the other PTL phenotypes in particular I-PTL despite treatment with routine broad-spectrum antibiotics. This study raises further doubts about the efficacy of broad spectrum antibiotics, such as erythromycin, which are currently widely used to treat women with preterm rupture of membranes. It may be time that for us to modify our management of CA-PTL, trying to find a more individualized approach.

Spontaneous PTL is a clinical enigma due to etiologic, pathophysiologic, and genetic heterogeneity. In this study, we observed that there was variation in location and intensity of the inflammation in the different sPTL phenotypes, and future work should be aimed at developing prediction models aimed at identifying the specific PTL phenotype and consequently inform the need to expedite (CA-PTL) or delay delivery (I-PTL).

Authors’ roles

NS and MJ made substantial contribution to the conception and design of the study. All authors contributed to the acquisition, analysis, and interpretation of data. NS, MJ, and SY drafted and critically revised the manuscript. All authors approved the final manuscript.

Conflict of interest

All authors wish to declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.