Human labour is associated with altered regulatory T cell function and maternal immune activation

Summary

During human pregnancy, regulatory T cell (T_reg) function is enhanced and immune activation is repressed allowing the growth and development of the feto–placental unit. Here, we have investigated whether human labour is associated with a reversal of the pregnancy-induced changes in the maternal immune system. We tested the hypothesis that human labour is associated with a decline in T_reg function, specifically their ability to modulate Toll-like receptor (TLR)-induced immune responses. We studied the changes in cell number, activation status and functional behaviour of peripheral blood, myometrial (myoMC) and cord blood mononuclear cells (CBMC) with the onset of labour. We found that T_reg function declines and that T_reg cellular targets change with labour onset. The changes in T_reg function were associated with increased activation of myoMC, assessed by their expression of major histocompatibility complex (MHC) class II molecules and CBMC inflammatory cells. The innate immune system showed increased activation, as shown by altered monocyte and neutrophil cell phenotypes, possibly to be ready to respond to microbial invasion after birth or to contribute to tissue remodelling. Our results highlight changes in the function of the adaptive and innate immune systems that may have important roles in the onset of human labour.

Graphical Abstract

We studied the effects of human labour on peripheral blood mononuclear cells obtained from term (37–41 weeks) gestation pregnancies. Our results show that labour is associated with an increasingly activated innate immune system with altered monocyte phenotype, and a change in T-helper cell targets of T_regs that favour Th1/Th17 effector subsets in response to TLR agonists. These results are reflected in mononuclear cells derived from cord blood and myometrium, with an increase in IFN-γ-producing CD4⁺ T cells and effector memory T cell proportions.

reproductive immunology, tolerance/suppression/anergy, Toll-like receptors (TLRs), regulatory T cells

Introduction

Pregnancy is associated with a series of immune adaptations designed to prevent immune-mediated rejection of the fetus while still protecting the mother and baby from invading pathogens [1,2,3,4,5]. The reversal of this immune tolerance may be a key step in the onset of parturition. The placenta is thought to initiate and maintain much of the immune modulation during pregnancy, with the syncytiotrophoblast releasing immune modulatory vesicles in pregnancy which attenuate T cell signalling, down-regulate the expression of the natural killer (NK) activating receptor NKG2D, promote CD95–CD95L-mediated apoptosis, as well as influence the local secretion and expression of tolerance-inducing transforming growth factor (TGF)-β and programmed death ligand 1 (PD-L1), respectively [6]. In fact, there is growing evidence that recognition of fetal antigen at the maternal fetal interface contributes to the induction of immune tolerance [7,8]. Extravillous trophoblasts (EVT) express non-classical human leucocyte antigen (HLA) class Ib molecules, HLA-E and HLA-G, as well as low-level polymorphic HLA-C. Whereas HLA-E and -G primarily protect against NK cell lysis, HLA-G can also induce regulatory T cell (T_reg) differentiation, HLA-C may serve as a source of allorecognition by maternal immune cells, eliciting T cell responses and potentially recruiting T_regs to the maternal–fetal interface [7,9,10]. This has been shown in humans with HLA-C mismatched, but not matched pregnancies [7]. Furthermore, murine data indicates that fetal-specific T_regs are important for fetal survival and that HLA-C may serve to present fetal antigens to maternal T cells [11]. This suggests that paternal MHC/HLA-C molecules are specifically recognized by maternal T cells and, crucially, the resultant immune response is modulated by induced T_regs. It is therefore not surprising that in mouse models fetal-specific forkhead box protein 3 (FoxP3⁺) T_regs expand in pregnancy but, interestingly, they persist following delivery and are able to rapidly accumulate with secondary mating when exposed to the same paternal antigens [12,13]. It is important to note, however, that enumeration of total T_regs alone may not reflect their functional activity. Schober et al described in humans, how alterations in T_reg-suppressive activity is associated with preterm labour [14,15].

In both humans and murine pregnancies labour is held to be an inflammatory process, potentially driven by fetal antigen exposure or innate immune system stimuli [16,17]. Data from rat pregnancies suggest that inflammation can lead to the release of mitochondrial DNA due to tissue necrosis and syncytiotrophoblast membrane microparticles, which can interact with Toll-like receptors (TLRs) [18]. In humans, these placental-derived fragments and microparticles have been shown to induce monocyte cytokine secretion, thus contributing to inflammatory processes occurring before and/or during parturition [18,19]. Interestingly, the same in-vitro human experiments demonstrated that syncytiotrophoblast membrane microparticles-induced monocyte cytokine secretion can be prevented with the use of TLR inhibitors [19]. Therefore, it is plausible that the systemic inflammatory process that occurs during labour can be amplified by bacterial or viral infections.

Infection is associated with approximately 40% of cases of preterm labour [20], mediated through a direct effect in chorioamnionitis and indirectly, in other infections, through an excessive response of the innate immune system leading to marked systemic inflammation [20]. In the pregnant mouse, infection disrupts vital immune-modulatory processes, including inducing a loss of maternal T_reg activity, which triggers a fetal-specific T cell infiltration, inflammation and pregnancy loss [21,22]. There is little evidence in the human that there is a similar infection-induced loss of maternal immune tolerance [23]. Although prospective clinical studies in humans have suggested that labour and the puerperium carry a two- to threefold increase in the risk of sepsis compared to the antenatal period [24], however, it seems much more likely that this risk is secondary to labour. Human and mouse data suggest that there is a reduction in the number of suppressive T_regs, which may accentuate the inflammatory response to infection [14,25]. This may lead to an excessive inflammatory response to sepsis, perhaps explaining why sepsis is perceived to be more severe and/or frequent in the puerperium. Animal models of acute sepsis have suggested that functional T_regs have a protective effect [26–28], consistent with the observation in mice that T_regs are activated by lipopolysaccharide (LPS) as part of the homeostatic response, protecting the animal from excessive inflammation [29].

We previously reported that pregnancy was associated with a repression of recall antigen responses and that this repression reversed with advancing gestation [30]. Other groups have demonstrated in animal models and human experiments that labour is associated with a loss of T_reg suppression of fetal antigen-specific cell-mediated immune responses. Here we investigated the hypothesis that human labour is associated with a decline in T_reg function, specifically their ability to modulate TLR ligand-induced immune responses. In a novel approach, we sought to analyse the interplay between immune cells in three pregnancy compartments that comprise the maternal and fetal circulations and the myometrium during labour.

Methods

Study approval

Subjects were recruited from Chelsea and Westminster Hospital, London, UK. Human experimentation guidelines of the authors’ institution were followed during the conduct of clinical research. Informed written consent was obtained from all participants and in all instances of ex-vivo and in-vitro work. Ethical approval was obtained from the National Research Ethics Service (NRES), London, UK committee as well as by Chelsea and Westminster NHS Trust, London, UK (Ref: 11/LO/0971).

Study design

A combination of enzyme-linked immunospot (ELISPOT), 9-parameter flow cytometry and flow cytometry-based T cell lymphoproliferative assays were developed to assess functional cellular responses and leucocyte phenotype. To ensure consistency and comparability, a standardized protocol was used for all samples. All blood and tissue samples were processed within 2 h of collection and all phenotypical and functional work was performed on fresh samples.

Study participants

Longitudinal analysis of immune responses and leucocyte phenotype during pregnancy and labour was undertaken using healthy pregnant patients with singleton pregnancies (n = 18), who were recruited from the antenatal clinic from April 2013 to September 2014 at Chelsea and Westminster Hospital during their first visit before 20 weeks of gestation. Once recruited, these patients were asked to provide peripheral blood samples at recruitment, 34 weeks of gestation and in labour. Cord blood was also collected from the umbilical vein immediately after birth. Samples were collected when patients were having regular contractions, but their cervical dilatation varied at sample collection.

To understand the effects of early labour on maternal and neonatal immune responses, a separate cohort of patients with singleton pregnancies who were between 37 and 41 weeks of gestation and underwent emergency caesarean sections (CS) for breech presentation (n = 6) were recruited on the labour ward at Chelsea and Westminster Hospital from November 2013 to August 2014. These patients consented to having peripheral blood samples taken in early labour, which we defined as the presence of regular painful contractions and cervical dilatation of ≤ 4 cm [31], as well as myometrial biopsies and cord blood samples that were obtained at CS. Myometrium was preferred to placenta, as it represented the maternal side of the maternal–fetal interface. A control group of patients undergoing planned CS were also recruited (n = 6).

For all cohorts, inclusion criteria included age at booking of under 40 years and singleton pregnancies. Exclusion criteria included the development of or past medical history of any pregnancy-related or -unrelated complications that affected the course of pregnancy, such as pre-eclampsia, gestational diabetes or intra-uterine growth restriction as well as any autoimmune, hypertensive or renal conditions. Patients diagnosed with serological or clinical manifestations of any of the herpesviruses cytomegalovirus (CMV), Epstein–Barr virus (EBV), herpes simplex virus (HSV) or measles, as well as influenza and tuberculosis (TB) or tetanus, were excluded.

Preparation of cells

For maternal peripheral blood leucocyte analysis, 35 ml of peripheral blood was obtained using a Vacutainer^TM system, and five 6-ml lithium heparin and one ethylenediamine tetraacetic acid (EDTA) tubes were collected (Becton Dickinson, Oxford, UK). Ten millilitres of blood from the cord vein after delivery of the placenta were obtained using the above vacutainer system and lithium heparin collection tubes. Cord blood mononuclear cells/peripheral blood mononuclear cells (CBMC/PBMC) were separated by density gradient centrifugation on Histopaque (Sigma-Aldrich, Dorset, UK), as described previously [32]. Cell viability was determined using a trypan blue exclusion test, and only samples in which this was greater than 80% were used in experiments. For functional work the cells were suspended in tissue culture medium (TCM; RPMI-1640 with penicillin and streptomycin at final concentrations of 100 IU/ml and 100 μg/ml, respectively, and L-glutamine at a final concentration of 2 mM; all from Sigma-Aldrich), and for phenotype the cells were suspended in phosphate-buffered saline (PBS) containing Ca²⁺ and Mg²⁺ (Sigma-Aldrich).

Myometrial tissue digestion

In order to obtain myometrial mononuclear cells (myoMC), 1 × 2 cm myometrial biopsies were obtained from the lower segment of the uterus at the time of CS. Where possible, serosal and decidual layers were avoided by the surgeon. The myometrial biopsies were washed with and suspended in (30 ml) sterile PBS, and then stored at +4°C until tissue digestion, which was commenced within 2 h. Enzymatic tissue digestion of myometrial samples was achieved with a mixture of Liberase™ and DNAse (Roche, Welwyn Garden City, UK) diluted in TCM, and used at final concentrations of 100 and 200 μg/ml, respectively. Prior to digestion, vascular areas of tissue were dissected out, and the remaining tissue was weighed and then chopped using two scalpel blades (Swann Morton, Sheffield, UK). The tissue was then placed into 5 ml of Liberase–DNAse mix per 0·2g of tissue and incubated for 40–50 min at 37°C. In order to ensure uniform digestion, at 15-min intervals the tissue suspension was rigorously shaken to prevent clumping. Following this, fluorescence activated cell sorter (FACS) wash buffer (FWB), containing PBS supplemented with 2% fetal bovine serum (FBS), 0·1% sodium azide and 2 mM EDTA, was added to stop the enzymatic digestion. The cell suspension was strained through a 40-μm nylon cell strainer (BD Biosciences) into a 15-ml Falcon tube™, and then centrifuged at 400 g for 5 min. Finally, the suspension was washed twice with PBS, cells were counted before the final wash and resuspended in 400 μl of FWB ready for flow cytometric analysis.

ELISPOT assay

Interferon (IFN)-γ, interleukin (IL)-10 and granzyme B ELISPOT assays were performed in order to detect recall antigen/peptide-specific T cell responses, as previously described [21]. PBMC, 1 × 10⁵/well, were cultured in 10% (heat-inactivated) male AB plasma-RPMI (200 μl/well final volume; Sigma-Aldrich) in 96-well polyvinylidene difluoride (PVDF)-backed plates (Merck Millipore, Hertfordshire, UK) that were coated with antibodies for the specific cytokines and proteases of interest (Mabtech AB, Nacka Strand, Sweden). In duplicate wells, PBMC were stimulated with 100 μl of an antigen/peptide pool obtained from NIBSC (NIBSC, Hertfordshire, UK) and Virion-Serion (Virion-Serion, Würzburg, Germany) at the manufacturer’s recommended concentrations. Stimuli included: EBV, CMV influenza A, measles and HSV whole lysates; purified protein derivative (PPD) of Mycobacterium tuberculosis tuberculin (NIBSC); purified tetanus toxoid (TTOX; NIBSC); and flu/EBV/influenza (FEC) peptide pool (NIBSC). Positive and negative controls were provided by phytohaemagglutinin (PHA, 5 μg/ml; Sigma-Aldrich) and TCM, respectively. Plates were incubated at 37°C in 5% CO₂ for 48 h. Detection of spot-forming cells (SFC) required the addition of biotinylated anti-IFN-γ, IL-10 or granzyme B (Mabtech AB) and incubation, followed by the use of a concentrated streptavidin–alkaline phosphatase conjugate (Mabtech AB). Finally, a development step was carried out using a chromogen prepared from a premixed BCIP/NBT substrate kit (BioRad Laboratories Ltd, Hertfordshire, UK). SFC reading and counting was performed using an AID ELISPOT reader (Oxford Biosystems Cadama, Oxfordshire, UK).

Magnetic bead cell separation

In order to reduce the proportion of T_regs in peripheral blood samples, two magnetic bead-based cell isolation kits were used: Dynabeads^® CD4⁺ positive isolation kit (Thermo Fisher Scientific, Paisley, UK) and Dynal^® CD4⁺CD25⁺ positive isolation kit (Thermo Fisher Scientific). Maternal PBMC were divided into two aliquots. One was left untouched. In the other, both CD4⁺ cells and CD4⁺CD25⁺ lymphocytes were selected positively sequentially and then removed. In both instances a Biomag^® 15 ml/50 ml tube separator (Polysciences Inc., Eppelheim, Germany) was used, according to the manufacturer’s instructions and previously published work [33]. However, DETACHaBEAD^® was not used. The remaining PBMC were combined to obtain a population significantly reduced of CD4⁺CD25^hi cells that included T_regs. Subsequent lymphocyte proliferation and phenotype analysis was undertaken on both T_reg-reduced and non-reduced aliquots.

Lymphocyte proliferation assay

Following T_reg reduction, PBMC from reduced and non-reduced samples were compared for proportions of CD4⁺CD25⁺ to ensure that there were no significant differences prior to stimulation. PBMC, 10–30 × 10⁶, were resuspended in PBS at a concentration of 10 × 10⁶ cells/ml; 1 μl/1 ml cell suspension of 1 mM stock Violet proliferation dye 450 (VPD450; BD) was added to the PBMC suspension. This mixture was incubated in the dark at 37°C in a waterbath for 15 min. Following two washing steps, the cell pellet was resuspended in TCM supplemented with 10% inactivated male AB serum (Sigma) at a concentration of 10 × 10⁶ cells/ml.

The suspension was divided into aliquots for culture and stimulation with TLR agonists (TLR-2, -4 and -8) at concentrations based on titration experiments, and PHA (Sigma-Aldrich) diluted in TCM to a concentration of 10 μg/ml to be used at a final concentration of 5 μg/ml. In earlier experiments we determined optimum concentrations of TLR agonists [2,4,8] by measuring CD4 and CD8 activation marker (CD69 and CD25) expression following stimulation of PBMC. Optimum concentrations for TLR-2 (Pam3CSK4 a TLR-1/-2 agonist) and TLR-4 (LPS Escherichia coli K12) and TLR-8 (ssRNA40/LyoVecTM) were 100 ng/ml and 1 μg/ml, respectively (all InvivoGen, Toulouse, France). Choosing an appropriate duration of stimulation to observe the effects of TLR agonists in vitro was difficult. While T cell activation is demonstratable following overnight exposure, preferential T cell apoptosis is seen after 4–6 days of culture [34,35]. Therefore, we chose a 7-day culture. For consistency, we chose the same length of stimulation for PHA. We expected to still be able to measure proliferative responses following incubation with PHA beyond 6 days [36]. TCM provided a negative control and a representation of background activity. PBMC labelled with VPD450 were cultured for 7 days in an incubator at 37°C with 5% CO₂, washed and resuspended in PBS at a concentration of 1 × 10⁷ cells/ml for surface staining. Proliferation was analysed using division index and percentage of divided cells as well as proliferation index (PI), all calculated by FlowJo version 7.65 (TreeStar Inc., Ashland, OR, USA). The PI was chosen to measure proliferation, as it represents the total number of divisions divided by the number of cells that underwent division [37,38]. Therefore, PI only takes into account responding cells. As this value only considers the fraction of responding cells, it was a more useful value to compare between samples.

Surface staining

Multi-colour colour flow cytometry was used to phenotype CD4 and CD8 T cell subsets, NK cells and dendritic cells (DC). PBMC were stained with murine, anti-human monoclonal antibodies, according to the manufacturer’s instructions; 2 × 10⁶ cells were stained per tube, incubated in the dark at room temperature for 30 min, washed with PBS and fixed with BD stabilizing fixative (BD Biosciences), before acquisition within 24 h. For T and NK cells, a minimum of 100 000 events were acquired on a three-laser flow cytometer (BD Biosciences LSR II) and subsequently gated according to corresponding isotype controls. For monocyte phenotype as well as neutrophils and NK cell counts, maternal whole blood was surface-stained using a monoclonal antibody mix. Following staining, lysis of red blood cells was performed using FACS lysing solution (BD). Cell counts were determined using AccuCheck counting beads (Invitrogen, Paisley, UK). Samples were run using a three-laser flow cytometer (CyAn ADP; Beckman Coulter, High Wycombe, UK) following compensation with normal human blood using Summit software (Cytomation Inc., Fort Collins, CO, USA).

FoxP3 staining

Following surface staining and prior to fixation, PBMC were washed and resuspended in 1 ml of freshly prepared FoxP3 fixation/permeabilization working solution, consisting of FoxP3 fixation/permeabilization concentrate (eBioscience, Hatfield, UK) diluted with FoxP3 fixation/permeabilization diluent (eBioscience) at a concentration of 1 : 3, and incubated in the dark for 30 min at +4°C. The mixture was washed in permeabilization working solution (eBioscience), resuspended in the same solution and 5 μl phycoerythrin (PE)-labelled anti-FoxP3 was added and incubated in the dark at room temperature for 30 min. After two wash steps the cells were resuspended in 200 μl of diluted stabilizing fixative and stored at 4°C until acquisition.

Intracellular cytokine staining

Cytokine staining of CBMC was optimized in earlier experiments and a 12-h incubation with the use of PMA and ionomycin, and the addition of brefeldin A for the final 5 h produced the most consistent results. CBMC were stimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin at a concentration of 100 ng/ml and 1 μg/ml, respectively (Sigma-Aldrich), and co-stimulatory anti-CD28/CD49d (BD) was added at a final concentration of 3 μg/ml. The mixture was incubated in 5% CO₂ at 37°C for 12 h in 15-ml Falcon tubes (Sigma) and the concentration of cells used was 2 × 10⁶ cells/ml. After a wash cycle, a 10 μg/ml concentration of brefeldin A (Sigma-Aldrich) was added to the mixture, vortexed and placed back in the incubator for another 5 h. The cells were washed and then surface-stained. After the incubation step, 500 μl of Cytofix/Cytoperm™ (BD) buffer solution was added to permeabilize the cells, following which they were washed with dilute Perm/Wash™ buffer (BD) and anti-cytokine fluorochrome-conjugated antibodies were added, according to the manufacturer’s recommended concentrations, and then incubated in the dark. The final pellet was resuspended in 200 μl of dilute stabilizing fixative.

Flow cytometry antibodies

T cells were identified using the following anti-human monoclonal antibodies (clones): peridinin chlorophyll protein (PerCP) Cy5.5-labelled anti-CD3 (SK7; Biolegend, London, UK); allophyocyanin (APC)-H7-conjugated anti-CD8 (SK1; Biolegend); BV421-labelled anti-IL-17A (N49-653; BD), BD Horizon V450-labelled anti-CD38 (HIT2; BD Biosciences, Oxford, UK), anti-CD127 (HIL-7R-M21; BD) and anti-CCR4 (1G1; BD); BD Horizon V500-labelled anti-HLA-DR (G46-6; BD) and anti-CD4 (RPA-T4; BD); fluorescein isothiocyanate (FITC)-labelled, anti-CD25 (2A3; BD), anti-CCR6 (53103; R&D Systems, Abingdon, UK), anti-IFN-γ (25723.11; BD) and anti-PIBF (rabbit polyclonal; Biorbyt, Cambridge, UK); phycoerythrin (PE)-conjugated anti-CCR7 (150503; R&D), anti-CCR5 (2D7; BD), anti-CCR3 (5E3; BD), anti-IL-4 (8D4-8; BD), and anti-FoxP3 (PCH101; eBioscience); APC-labelled anti-CD28 (CD28.2; BD), anti-HLA-G (87G; eBioscience), anti-CXCR3 (1C6; BD), anti-CD69 (L78; BD), anti-IL-10 (JES3-19F1;BD) and anti-HLA-DR (LN3; eBioscience); PE-cyanin 7 (Cy7)-labelled anti-CD45RA (L48; BD) and anti-CD45RO (UCHL1; BD).

Monocytes, neutrophils and overall NK cell counts and monocyte phenotype were determined using a combination of the following antibodies: PE-labelled anti-CD115 and CD66b (9-4D2-1E4 and G10F5; Biolegend), PE-CF594-labelled anti-CD11b (ICRF44; Biolegend), PerCp-labelled anti-CD14 (HCD14; Biolegend), PeCy7-labelled anti-CD16 and anti-CD56 (HCD14 and 5.1H11; Biolegend), APC-labelled anti-triggering receptor expressed on myeloid cells (TREM)-1 (TREM-26; Biolegend), Alexa Fluor^® 647-labelled anti-HLA-DR, anti-PD-1 and anti-CD86 (L243, EH12.2H7 and IT2.2; Biolegend), APC-eFlour 780-labelled anti-CD45 (HI30; eBioscience) and FITC-labelled anti-CD11c (S-HCL-3; Biolegend).

Flow cytometric data analysis was performed using FlowJo version 7.6.5.

Statistics

For longitudinal analysis during pregnancy gestation with repeated measures and parametric data, a one-way analysis of variance (anova) with multiple group comparisons and either Tukey’s or Geisser–Greenhouse correction was used to compare group means. For non-parametric longitudinal repeated measures, data analysis was undertaken using a Friedman test with Dunn’s correction. Continuous parametric data were compared between groups using either a paired or unpaired Student’s t-test, depending on the experiment. Similarly, for non-parametric data a Wilcoxon’s matched-pairs signed rank test or Mann–Whitney U-test was used. Statistical analysis was performed on GraphPad Prism version 7.0 (GraphPad Software, San Diego, CA, USA). Data are presented as means ± standard error of the mean (s.e.m.) or median ± interquartile range (IQR), as appropriate for the distribution normality. All P-values are two-tailed and unadjusted. Significance is defined as P < 0·05.

Results

Term human labour is not associated with increased IL-10 cell-mediated immune suppression

We began by investigating the effects both pregnancy and labour have on cell-mediated immune responses. Our previous work has demonstrated that T cell activation and effector function is associated with increasing gestation in pregnancy and during birth [39]. However, in these experiments, the patients were delivered by either vaginal delivery or CS, and some of the delivery samples were taken immediately after birth rather than during labour itself. In order to first establish the effects that labour, specifically, may have on cell-mediated immune responses, we recruited low-risk singleton patients at less than 20 weeks of gestation and performed ELISPOT and flow-cytometric analysis of their peripheral blood longitudinally from recruitment, 34 weeks of gestation and in labour. However, in these initial experiments our aim was to observe the broad effects that labour may have on systemic immune responses, and so we did not control for the degree of cervical dilation during labour. In order to detect antigen-specific T cells, including memory T cells and their responses, we used the sensitive ELISPOT assay with predominantly recall antigens/peptides. A combination of viral whole lysates that require processing and presentation by APC to CD4 T cells were used to assess cell-mediated responses. The use of whole antigen or peptide pools spanning a large range of epitopes ensured that the assay was not limited by HLA restriction, and we were able to broadly define memory T cell responses. Our results showed that, overall, both IL-10 and IFN-γ responses were largely unchanged with labour (Supporting information, Fig. S1a,b). Only the IL-10 response to measles showed a significant change, but this was not reproduced with other stimuli (P < 0·01 and P < 0·05 versus 34 and < 20 weeks, respectively; Supporting information, Fig. S1a). Antigen/peptide-specific responses appeared to be relatively heterogenous, and although the overall the mean IL-10 SFC numbers increased in labour, this was not statistically significant (Supporting information, Fig S2c). When comparing these findings to memory T cell subset phenotype in the same patients, we found that although pregnancy seemed to be associated with greater numbers of T central memory (T_CM) and T effector memory (T_EM) compared to T effector memory RA+/terminally differentiated effector memory T cells (T_EMRA), there were no significant longitudinal or labour-related changes (Supporting information, Fig S2a,b). T_reg, CD38 and HLA-DR expression on CD4 and CD8 and DC subsets (myeloid and plasmacytoid) phenotype was also analysed by flow cytometry in these patients concurrently with the ELISPOT assays (Supporting information, Fig. S2c–e) but there were no labour-specific changes. Collectively, these results suggested that human labour is not associated with a change in IL-10-driven cell-mediated immune suppression, which is in contrast to others who have shown increased activation of peripheral blood T_regs in term labour [14].

Advancing gestation and delivery is associated with peripheral blood changes in inflammatory cells

Next we profiled the longitudinal changes of inflammatory cell numbers through pregnancy and in labour to establish if either the numbers of peripheral blood inflammatory cell or their markers of activation changed in our patient cohort. We assessed the phenotype of peripheral monocytes and neutrophils and, as shown in Fig. 1 (gating strategy is shown in Supporting information, Fig. S3), we found that advancing gestation and parturition were associated with an increase in intermediate monocytes (CD14⁺⁺CD16⁺⁺; P < 0·05 labour versus < 20 and 34 weeks; Fig. 1a), which are commonly expanded during inflammatory disease [40]. In addition, labour was associated with increased triggering receptor expressed on myeloid cells 1 (TREM-1) expression on all monocyte subsets, but only significantly on classical monocytes (CD14⁺⁺CD16⁻; labour versus 34 weeks, P < 0·05; Fig. 1b). TREM-1 functions to strongly enhance leucocyte activation in the presence of microbial stimulus [41]. At the same time, increasing gestation was accompanied by a fall in CD86 expression on all monocyte subsets, but only significantly on non-classical monocytes (CD14⁺CD16⁺⁺; < 20 versus 34 weeks; P < 0·05; Fig. 1c). CD86 is a co-stimulatory molecule indicative of antigen-presentation activity [42]. In addition, overall counts of neutrophils (labour versus < 20 weeks, P < 0·05) and CD11c⁺ monocytes (labour versus < 20 weeks, P < 0·05) were increased during labour compared to early gestation (Fig. 1d). Our results suggested that human labour is indeed associated with an expansion of inflammatory cells and that these have a phenotype indicating a heightened ability to respond to microbial stimuli, but not antigen presentation. Activation of monocytes by TLR agonists can modulate CD4 function and T helper (Th) polarization [43–45]. As both monocytes and neutrophils express pattern recognition receptors such as TLRs, we wondered if, during labour, TLR-mediated responses would have a similar effect on T cell proliferation or be modulated by T_regs. We have previously shown that the stage of labour impacts upon the degree of inflammation seen in reproductive tissue, where the majority of inflammation is seen in established labour (when the cervix is greater than 3 cm dilated) [46]. However, early inflammatory and migratory changes in peripheral leucocytes may occur far earlier, as demonstrated in rat pregnancies [47]. With the assumption that both systemic inflammation and the influence of T_regs as immune suppressors of inflammation would follow a similar pattern with changes in early labour, we investigated T_reg function in matched cohorts of early labouring and non-labouring patients.

Fig. 1

The effect of gestation and parturition on numbers of monocyte subsets. Monocyte subsets, (a) intermediate CD14⁺⁺CD16⁺⁺, (b) classical CD14⁺⁺CD16⁻ and (c) non-classical CD14⁺CD16⁺⁺, during pregnancy and labour, and their expression of human leucocyte antigen D-related (HLA-DR), triggering receptor expressed on myeloid cells 1 (TREM-1) and CD86. (d) Overall counts of peripheral monocytes, neutrophils and CD11c⁺ monocytes during pregnancy and with delivery. Columns indicate mean and standard error of the mean (s.e.m.) or median and interquartile range (IQR) depending on the data distribution. Gestation at sampling is indicated in pregnancy: < 20 + 0 weeks (n = 9), 34 weeks (n = 13) and in labour (n = 7). Data were analysed longitudinally by one-way analysis of variance (anova) with Tukey’s correction or Friedman test with Dunn’s correction. P-values are two-tailed and significance is defined as *P < 0·05 and **P < 0·01.

Labour is associated with a loss of T_reg-modulated lymphoproliferative response

We recruited a cohort of low-risk patients at term with a singleton pregnancy, who were delivered by CS either before (nL) or after the onset of labour (tL) and with a cervix < 4 cm dilated. They were age-, body mass index (BMI)- and parity-matched (Supporting information, Table S4). Initially, we determined whether T cell lymphoproliferative responses were affected by reducing the number of T_regs. As described in the Methods section, fresh PBMC were divided into two aliquots. In one, CD4⁺ and in the other, CD4⁺CD25⁺ lymphocytes were positively selected and then removed. These aliquots were then combined to give a CD4⁺CD25⁺ reduced sample, which included the CD4⁺CD25^hi T_reg population. Therefore, the combination of depletion kits resulted in an unlabelled PBMC population reduced of T_regs. Whole PBMC and the purity of the population obtained post-T_reg removal were assessed by flow cytometry. The mean reduction in CD4⁺CD25^hi proportions, measured by flow cytometry, were 52 and 60% in nL and tL samples, respectively (both P < 0·05 versus T_reg, not reduced control samples; Fig. 2a). Our negative selection method avoided the use of antibodies targeted against cell surface proteins (lineage markers), which might inadvertently have altered cell function. T_regs were intentionally only reduced and not completely removed, as our method would also remove activated CD4 T cells (CD4⁺CD25⁺). We used the VPD450 dye for flow cytometric analysis of lymphoproliferation of viable responder T cells. Finally, we used a 7-day PHA protocol to stimulate the PBMC rather than CD3 and CD28 or irradiated cord blood, as PHA also stimulates TLR receptors [7,48,49], giving a better indication of the probable response to TLR activation. Furthermore, PHA activates T cells by direct binding to cell membrane glycoproteins such as the T cell receptor (TCR) [50]. Our results showed that following T_reg reduction in nL samples PHA increased CD4⁺ T cell proliferation compared to non-reduced samples (P < 0·01; Fig. 2b,c). In contrast, in tL samples, T_reg reduction had no effect on the CD4 T cell proliferation response to PHA (Fig. 2b,d). We compared T_reg phenotype in nL and tL patients. T_regs were gated on CD3, CD4, CD127, CD25 and FoxP3, followed by HLA-DR and CD45RA, as previously described (Fig. 2e) [14,51,52]. As shown in Fig. 2f, tL is associated with reduced proportions of activated T_regs(CD4⁺CD25⁺CD127^loFoxP3⁺CD45RA⁻HLA-DR⁺, P < 0·05), which have a greater suppressive activity on responder T cells, and increased proportions of naive T_regs (CD4⁺CD25⁺CD127^loFoxP3⁺HLA-DR⁻CD45RA⁺), which are comparatively less suppressive.

Fig. 2

The effect of regulatory T cell (T_reg) reduction on peripheral blood T cell proliferative responses. (a) Reduction in proportions of CD25^hiCD45RO⁺ T cells following magnetic bead depletion. (b) Peripheral blood CD4 lymphoproliferative responses to phytohaemagglutinin (PHA) following T_reg reduction before labour onset (nL) as well as after labour onset (tL) samples. (c) Representative flow cytometry plots demonstrating CD4 proliferative responses in (c) nL and (d) tL samples. Columns indicate mean and standard error of the mean (s.e.m.). Data were analysed using a paired Student’s t-test or Wilcoxon’s matched-pairs signed rank test depending on the data distribution. (e) Representative flow cytometry plots of T_reg subsets. (f) Proportions of forkhead box protein 3 (FoxP3⁺) T_regs and their differential expression of human leucocyte antigen D-related (HLA-DR) and CD45RA. Activated FoxP3⁺ T_regs were identified as HLA-DR⁺, whereas naive subsets were HLA-DR⁻CD45RA⁺. Columns indicate mean and s.e.m. Data were analysed using an unpaired Student’s t-test. P-values are two-tailed and significance is defined as *P < 0·05 and **P < 0·01.

Lack of lymphoproliferative responses to TLR agonists in the peripheral blood of labouring patients

We next sought to determine if TLR-induced, monocyte-driven lymphoproliferative responses are specifically suppressed in the labouring group. We used purified TLR agonists for TLR-4, -2 and -8. After T_reg reduction, in-vitro 7-day stimulation of fresh PBMC with TLR-4 and TLR-8 ligands resulted in significantly increased CD4⁺ T cell proliferation for nL (both P < 0·05) but not tL patients (Fig. 3a–d). In fact, responses in tL were more comparable to nL before T_reg reduction. TLR-2 stimulation showed a similar trend; however, this did not reach statistical significance (data not shown). It should be noted that cell counts differed between nL and tL samples post-stimulation and so they were not directly compared by statistical analysis. Our findings suggest that proliferative responses to TLR stimulation either correspond to baseline or are muted in labour irrespective of T_reg function, which was an unexpected result.

Fig. 3

The effects of regulatory T cell (T_reg) reduction on peripheral blood lymphoproliferative responses to Toll-like receptor-4 (TLR-4) and TLR-8 ligands (a) and (b) show CD4 proliferative response to TLR-4 and TLR-8 (upper panels) alongside (c) and (d) representative flow cytometry histogram plots (lower panels). Columns indicate median and interquartile range (IQR). Paired data were analysed using a Wilcoxon matched-pairs signed rank test. Before labour onset (nL) and after labour onset (tL) samples were analysed using a Mann–Whitney U-test. P-values are two-tailed and significance is defined as *P < 0·05 and **P < 0·01.

Labour is associated with a greater proportion of naive T_regs and suppressed Th2 effector T cell lymphoproliferation

With our results indicating muted lymphoproliferative CD4 responses in labour, we wondered if T cell polarization was also being affected and if this was mediated by T_reg function. At the same time, we wanted to determine which Th subsets the functioning T_regs in the peripheral blood samples were specifically suppressing in the nL and tL groups. Th subsets were defined by surface chemokine receptor expression using CCR4, CXCR3 and CCR6 (Fig. 4a). We chose to do this by phenotyping using surface chemokine markers, as Halim et al. has previously shown that Th subset characterization by chemokine receptor expression accurately reflects their phenotype [53]. Our results were striking; we found no significant differences before or after the onset of labour in non-T_reg reduced samples (Fig. 4b,c). However, in response to TLR-2 and -4 agonists, when T_regs were removed in nL samples we saw a relative increase in the Th1/Th17 subset versus Th2, and this was reversed in the tL samples, where we saw a relative increase in the Th2 subset when compared to the Th1/Th17 subset (both P < 0·05; Fig. 4d,e). We also compared HLA-DR, CD38 expression on CD4 and CD8 T cells, as well as proportions of memory T cell subsets, in relation to labour, but found no significant differences (Fig. 5a,b,e).

Fig. 4

Lymphoproliferative responses of effector T helper (Th) cells to Toll-like receptor (TLR)-2, -4 and -8 ligands. (a) Representative flow cytometry plots depicting gating strategy for effector Th subsets: Th1 (CCR4⁺CXCR3⁺CCR6⁻), Th1/Th17 (CCR4⁺CXCR3⁺CCR6⁺), Th17 (CCR4⁺CXCR3⁻CCR6⁺) and Th2 (CCR4⁺CXCR3⁻CCR6⁻). Isotype controls were used. Th proliferative responses to TLR-2, -4 and -8 without regulatory T cells (T_regs) reduced, (b) before labour onset (nL) and (c) after labour onset (tL). Th proliferative responses to TLR-2, -4 and -8 with T_reg reduction, (d) nL and (e) tL. Columns indicate either mean and standard error of the mean (s.e.m.) or median and interquartile range (IQR) depending on data distribution. Data were analysed using repeated-measures one-way analysis of variance (anova) with Geisser–Greenhouse correction or repeated-measures Friedman’s test with Dunn’s correction. P-values are two-tailed and significance is defined as *P < 0·05.

Fig. 5

CD4 memory T cell subsets and activation status of peripheral blood mononuclear cells (PBMC), myometrial mononuclear cells (myoMC) and cord blood mononuclear cells (CBMC). Proportions in PBMC of (a) memory T cell subsets and (b) CD38 and human leucocyte antigen D-related (HLA-DR) expression measured by MFI in nL and tL patients. Proportions in myoMC of (c) memory T cell subsets and (d) CD38 and HLA-DR expression measured by mean fluorescence intensity (MFI) in before labour onset (nL) and (c) after labour onset (tL) patients. Columns indicate median and interquartile range (IQR). Not in labour and labour samples were analysed using a Mann–Whitney U-test. P-values are two-tailed and significance is defined as **P < 0·01. Representative flow cytometry plots showing (e) PBMC memory T cells, and the predominance of effector memory T cell (T_EM) subsets in myoMC (f) in nL and (g) tL, as well as the greater expression of HLA-DR on myoMC CD4⁺ T cells from tL patients. Populations were gated using isotype controls. (h,I,j) CBMC intracellular cytokine staining of interferon (IFN)-γ, interleukin (IL)-10 and IL-17 in response to phorbol myristate acetate (PMA) and ionomycin and compare the production in samples obtained from labouring (n = 10) and non-labouring (n = 8) patients. Columns indicate mean and standard error of the mean (s.e.m.) or median and IQR, depending on the data distribution. Not in labour and labour samples were analysed using unpaired Student’s t-test or a Mann–Whitney U-test. P-values are two-tailed and significance is defined as *P < 0·05.

Labour is associated with activated maternal myoMC and increased IFN-γ expression in CBMC

With our findings in peripheral blood, we also sought to determine if T cells in contracting myometrium may be of an effector phenotype and express MHC class II molecules to enable them to interact with antigens at the maternal–fetal interface. These may include feto–placental or microbial-derived antigens. Thompson et al. has shown that the density of T lymphocytes in lower segment myometrium is significantly increased in labour [54]. Using a combination of immunohistochemistry and flow cytometric analysis of myometrial inflammatory leucocytes, we have previously shown that neutrophils obtained from human labouring patients represent intravascular cells and are not tissue-resident, whereas the macrophages are resident [55]. Therefore, the myoMC we have identified in this study may represent migrating PBMC. We expected that the majority of myoMC would be effector cells or T_EM. Our results showed that myoMC compared to peripheral blood were comprised of a greater proportion of T_EM. However, with the onset of labour, myoMC also included CD4 T cells with significantly increased expression of HLA-DR (P < 0·01; Fig. 5d,f,g).

CBMC intracellular cytokine staining was undertaken using the labour/not in labour cohort (n = 6), as well as suitable patients from the longitudinal cohort (where samples were taken at delivery from patients who were in labour with regular contractions and cervical dilatation of ≤ 4 cm; n = 2). The neonatal effect of peripherally activated myoMC-derived CD4 T cells showed increased proportions of CD4⁺ IFN-γ⁺ T cells from CBMC and in response to PMA and ionomycin measured by intracellular cytokine flow cytometric analysis (P < 0·05; Fig. 5h). However, there were no observed changes in the proportions of CD4 T cells expressing IL-10 or IL-17 (Fig. 5i,j) nor CD8 expression of IFN-γ, IL-10 or IL-17 (not shown). We also analysed CBMC activation markers (CD69, CD38 and CD25) as well as memory T cell subsets (data not shown). However, there were no significant differences between nL and tL cohorts. Unsurprisingly, the majority of T cells in CBMC were naive.

Discussion

Peripheral blood T_regs have been shown to have a key role as immune modulators during pregnancy and labour as well as during an inflammatory response [13,14,56]. However, their behaviour in labour and infective inflammation is likely to be very different. Understanding the changes in the maternal immune system during physiological labour is important, as they may have a role in the onset of labour and explain the observed greater susceptibility to infection [20]. Previous studies have shown that parturition is characterized by an increase in naive T_reg phenotypes and a decline in T_reg function [14]. In contrast, in severe bacterial infection, the proportion of T_regs has been shown to expand with increased immune modulatory potential, which is thought to protect against severe illness [28,56,57]. We found that the labour is associated with an increased immune responsiveness in peripheral blood samples (monocyte subset proportions and TREM expression), with diminished T_reg repression to infective and inflammatory stimuli (demonstrated by PHA and TLR responses), and in the myometrium, where CD4 T cells showed greater expression of MHC class II molecules. The decline in peripheral immune response may be due to an increase in the proportion of peripheral blood naive T_regs, as defined by a CD45RA⁺ phenotype, which are associated with a decline in T_reg function [14,58], and a shift in the repression of the Th subset of effector T cells favouring the Th1/Th17 lineage. Schober et al. investigated peripheral blood T_reg suppressive function in term and preterm labouring patients using positive T_reg selection, and co-culture with synergistic or allogenic responder T cells from pregnant donors [14]. They found no difference in the maximum suppressive activity of CD4⁺CD127^low+/−CD25⁺ T_regs or a difference in T_reg subsets obtained from patients who experienced spontaneous term delivery when compared to non-labouring patients, which is contrary to our results [14]. However, when they titrated down the proportion of T_regs in culture with responder T cells from the same patients, the ratio leading to a 15% reduction in suppression of responder T cells was significantly lower in spontaneously delivering patients, suggesting that the suppressive activity of these T_regs may be reduced [14]. They established that this is not due to less responsive effector T cells by repeating the experiments using allogenic donors and showing the same results. However, their results were not supported by IL-10 or IL-2 measurement. One weakness of the study was that it is not clear when, in the process of labour or the post-partum period, the samples were obtained. This is important, as murine work has shown differences in leucocyte infiltrates in labouring compared to immediately post-delivery myometrial samples, suggesting a change immune profile with parturition [59]. Therefore, in our experiments we took a different approach by recruiting patients in spontaneous labour (≤ 4 cm dilated) to determine whether any change in immune function could be detected at this stage; if there was no change, then the previous observations would imply that immune function was a consequence and not a cause of labour. In addition, the previous researchers used a positive isolation method to obtain T_regs that may have influenced their function, whereas we used negative selection to deplete T_regs, and so excluded any effect related to the positive isolation method. Furthermore, we used chemokine receptor profiles to determine the Th subsets of responder T cells. In the peripheral blood samples from non-labouring women, we found pregnancy-related T_reg suppression of the Th1/Th17 T effector lineage, whereas in labouring patients T_reg reduction enhanced the proliferative response of the Th2 subsets compared with Th1-like and Th17. This suggests that the pregnancy-associated T_reg suppression of Th1/Th17 lineage is lost with the onset of labour and replaced by repression of the Th2 lineage. In addition, we found that the innate arm of the immune system during labour is poised to respond to microbial stimuli and inflammation, as shown by altered monocyte phenotype.

Several studies have demonstrated that both humoral and cell-mediated immune responses are attenuated in the third trimester, accompanied by augmented IL-10 production; however, innate responses against bacteria are increased [60,61,62,63,64]. Our data show that with advancing gestation, immune tolerance declines and is replaced by a more proinflammatory functional phenotype, and this process occurs both systemically and locally at the maternal–fetal interface [16,65]. Previous studies have shown that increasing gestation is associated with increases in a range of proinflammatory cytokines, including IFN-γ, TNF-α, IL-6 and granulocyte–macrophage colony-stimulating factor (GM-CSF), among others [16,63,66,67]. In the present study, IFN-γ and IL-10 ELISPOT assays provided a good representation of antigen-specific memory cell responses in pregnancy. The stimuli we used consisted of protein antigens that are preferentially processed by APCs and presented via MHC class II to stimulate CD4 T cells. Therefore, PBMC were used rather than sorted T cells. The majority of responses detected are likely to be effector memory T cell responses [68]. A number of common recall antigens with high seroprevalence among pregnant women were used so that we could measure T cell responses over time. These included CMV, EBV, VZV and HSV, all of which have a seroprevalence ranging from 49 to 94% [69–71]. In addition, standardized UK vaccination programmes mean that most pregnant women will have received tetanus and bacillus Calmette–Guérin (BCG) vaccines in childhood, as well as influenza vaccination in pregnancy, all of which were assessed in the ELISPOT assays. The aim of the work was to assess the influence that labour may have on long-lived T cell responses using antigens and peptides to which the patients are likely to be good responders. IFN-γ responses are primarily directed towards intracellular pathogens, such as viruses, and are responsible for hyper-acute allograft rejection [64], whereas IL-10 inhibits proinflammatory cytokine and chemokine production and antigen presentation [72]. IFN-γ ELISPOT responses against recall antigens, including herpesvirus whole lysates, and peptides including CD8 epitopes, were largely unchanged throughout gestation and during parturition. Only the IL-10 response to measles was significantly increased. This suggests that labour is unlikely to be driven by cell-mediated interactions. Therefore, if T_reg function is altered in labour, our results suggest that this may not be due to cytokine production.

Next, we studied the peripheral blood monocyte and neutrophil populations. We postulated that even if our results suggest that T_reg functional responses to cell-mediated stimuli are unaffected during labour, it is plausible that inflammatory cell interactions may determine T_reg function, in addition to or even exclusively of changes in the circulating immunomodulator levels. [55,73,74]. Typically, monocyte subsets associated with inflammation are increased in pregnancy, including the intermediate monocytes [40,75]. Previous in-vitro studies have shown that peripheral blood monocytes from pregnant women are activated and have a lower surface expression of TLR-4, but have a greater capacity to migrate to the uterus during parturition [76,77]. Granulocyte subsets are thought to increase in pregnancy, and this may contribute to pregnancy-induced inflammatory changes [78]. In our study, we found that the proportions of intermediate and CD1c⁺ monocytes, as well as TREM-1 expression on classical monocytes, increased during pregnancy and labour. In addition, neutrophil numbers were increased during pregnancy and during labour. This corresponds to an increase in proinflammatory phenotypes and enhanced ability to respond to a microbial stimulus, but may also represent a loss of tolerance to feto–placental antigens [41,79]. Interestingly, the number of non-classical monocytes expressing CD86, which are thought to be involved in innate surveillance of tissues, was reduced with increasing gestation [80]. This suggests that, during pregnancy and labour, some aspects of the innate immune system display an activated phenotype with the potential to respond to both feto–placental and infective stimuli, while other components are repressed.

The importance of T_regs in pregnancy is well established [15,22]. Fetal-specific T_regs expand in pregnancy, contributing to the maintenance of immune tolerance, and these cells have immunological memory [12,13]. In contrast, labour has been associated with a decline in T_reg function [14]. Consistent with these observations, we found that T_reg reduction enhanced the CD4 T cell proliferative response to PHA in non-labouring samples. A protocol using PHA was used to stimulate PBMC rather than CD3 and CD28 or irradiated cord blood, as PHA also stimulates TLR receptors [48,49,81] giving a better indication of the probable response to TLR activation. Furthermore, PHA activates T cells by direct binding to cell membrane glycoproteins, such as the TCR [50]. Other suitable lectin-based mitogens would have been concanavalin A, pokeweed or Triticum vulgaris [49]. However, PHA provides the best lymphocyte stimulation in vitro when compared to other mitogens [82]. As one of the proposed mechanisms of TLR effects on lymphocyte function is through co-stimulation, a bead-based stimulation protocol would have provided a better way of assessing antigen-specific T cell lymphoproliferative responses, and this is a study limitation. Next, we studied the T_reg interaction with TLR agonists. Previous studies found that TLR agonists activate memory T cells in vitro [34], that TLR receptors are present on T_regs [83–85] and mediate T_reg activation and proliferation, enhancing their suppressive effects [86,87]. Their mechanism of action is not well understood, but TLR ligands may directly modulate T cell activation and proliferation via TLR-mediated co-stimulation of T cells and/or selective apoptosis [88–90]. In our experiments, T_reg reduction enhanced the CD4 T cell proliferation in response to TLR-4 and TLR-8, but not TLR-2 agonists. Intriguingly, proliferative responses to TLR agonists in labour samples in the presence or absence of T_regs, despite good PHA responses, were comparable to baseline in non-labour samples. Phenotyping the T_regs showed increased proportions of (CD45RA⁺) naive and reduced proportions of (CD45RA⁻HLA-DR⁺) activated T_regs in samples from patients in labour. The CD45RA⁺ T_reg phenotype is associated with poor suppressive function [14,58], whereas the CD45RA⁻ subset is a more effective suppressor of responder T cells and has greater potential to differentiate into Th17 cells [58]. This suggested that labour is associated with less potent T cell proliferative responses to direct TLR stimulation. However, there were no differences in proportions of naive or memory cells nor any changes in activation marker expression associated with labour. We show that, in the absence of labour, there is preferential suppression of Th1 and Th17 subsets. This is reversed in labour where our observations show that, with T_reg reduction, there is a significant increase in proliferation of effector Th2 cells compared to proinflammatory subsets such as Th1 and Th17. During periods of inflammation the phenotype of Th subsets is thought to reflect the immune response, with T_regs co-localizing and modulating specific Th populations [91]. As a result, these T_regs undergo functional specialization and resemble Th-like cells as defined by their chemokine receptor expression [91]. In particular, Th2 T_regs have enhanced chemotactic capacity and are able to inhibit Th1–Th17–Th1/17 effector lineages [53]. Our results suggest that this phenotype predominates prior to the onset of labour. However, with labour onset peripheral blood T_regs bear a naive phenotype, and the dominant Th-like T_reg subset appears to support the Th1/Th17 lineage. We used a chemokine receptor profile, which correlates well with transcription marker expression and has been effectively used by other researchers, to define our T helper subsets [53,92]. This was an unanticipated result as we expected labour to be associated with enhanced CD4 activation and a Th1/Th17 skewed response, particularly as we knew that the CD4⁺ cells had the capacity to proliferate, as shown by PHA stimulation. We initially thought that our findings were a consequence of the methods used to deplete the T_regs, as the process may have also removed activated CD4 effector T cells, or due to an inherent loss of activated cells in response to labour. However, when we compared the expression of activation marker CD25 (not shown) in the suppression assay, as well as HLA-DR, CD38 and proportions of memory subsets prior to T_reg reduction, we found no labour-related effects. However, we acknowledge that to confirm the Th lineage, transcription factor assessment is required. Any future experiments will need to incorporate transcription and intracellular cytokine analysis.

It is possible, however, that the presentation of antigen at the maternal–fetal interface may serve to reinforce the activation of placenta-specific T cells that were first exposed to feto–placental antigen elsewhere [93]. We hypothesized that local modulation of maternal immune responses may also help to explain the peripheral T_reg responses we observed with parturition. Therefore, myometrial samples were obtained, as they represent the maternal side of the maternal–fetal interface and may include migrating leucocytes. Flow cytometric analysis was used to analyse these biopsies following tissue digestion. Compared to peripheral blood, the proportions of T_EM were significantly increased and naive cells reduced, although there were no differences between labouring and non-labouring samples. However, myoMC CD4 MHC class II molecule expression was increased in labour. HLA-DR⁺ CD4 T cells are able to present antigen to other T cells, and induce cytotoxicity in resting CD8 T cells [94]. Therefore, in the myometrium, where T cells are largely T_EM, parturition is associated with a greater ability to interact with antigen at the maternal–fetal interface and to effect, regulate and balance immune responses. It should be noted that nL and tL samples provided very different numbers of cells, with a greater number of lymphocytes in the nL samples (for CD4: tL median 1651 IQR = 528–5934; nL median 2392 IQR = 986–3965). Background staining was also variable in both cohorts, so we used isotype controls for each sample to aid gating. Blocking the Fc-receptor may have helped to reduce background staining. Immunohistochemistry may have helped to confirm our flow cytometric data. Our previous reports have shown that inflammatory cells, measured by flow cytometry, in contracting myometrium are increased [55]. Here, we used immunohistochemistry of these myometrial sections to show that some of these cells were intravascular and marginated [55]. Other reports have shown that decidua contains differentiated T_EM cells, and that fetal-specific T cells respond to placental antigens in draining LNs, but not decidua [93,95,96,97]. Therefore, our observed, muted peripheral TLR responses in labour may be a protective adaptation to prevent overtly increased responses to feto–placental antigens during labour. The heightened susceptibility to antigen stimulation and therefore inflammation in labour samples was further demonstrated in the cord blood of neonates. CBMC from labouring mothers were rich in CD4⁺IFN-γ⁺ T cells. Frascoli et al. has previously investigated the influence of ‘maternal antigen rejection’ by T cells originating from the cord blood of preterm infants leading to fetal inflammation [98]. In contrast to our group, they compared preterm with term pregnancies, whereas we have compared labouring and non-labouring groups at term. Compared to the CBMC from term infants, their preterm cohort, of which 40% of placentas showed histological chorioamnionitis, had increased DC activation, a predominance of CD4 T_CM that were Th1-like and significant HLA-DR maternal microchimerism [98]. However, they did not observe any differences in T_reg proportions or HLA-mismatch between mother and baby [98]. In CBMC, these changes resulted in significantly increased proinflammatory cytokines, specifically IFN-γ, and proliferative CD4 and CD8 responses against maternal alloantigen [98]. They concluded that the inflammatory environment in the preterm infant may be a consequence of fetal rejection of maternal antigens, but this may also be due to a loss of cord blood T cell regulation [98,99].

One of the notable differences between the nL and tL cohorts is the difference in gestation that may have influenced the results obtained. Overall, the spread of gestations overlapped in both groups, such that the range in nL and tL were 37–39 and 38–42 weeks of gestation, respectively. We have previously shown that gestation may influence Th subset proportions, T cell CD38 expression and NK subsets, but labour/delivery has a much greater influence on immune responses [39]. In the current study, our longitudinal cohort data show that gestation did not affect proportions of T_regs, activated CD4 and CD8 T cells or subsets of DC. Therefore, it is unlikely that a difference of 14 days will impact upon immune function between nL and tL groups, but differentiating between gestation-specific and labour-specific responses would be difficult.

Collectively, our results show that parturition is unique, in that it is accompanied by a targeted adaptation of T_reg-mediated immune tolerance and T cell responses to inflammation. Moreover, labour is characterized by an enhanced capacity for innate responses to inflammatory stimuli and a shift towards Th1/Th17 responses. Our findings suggest that during labour the maternal immune system transforms into a more reactive phenotype, which may have a role in the onset of labour but may also contribute to an increased risk of excessive inflammatory responses and risk of progression to sepsis and septic shock.

Acknowledgements

This work was funded by grants from Borne (charity number 1167073), and infrastructure support was provided by the National Institute for Health Research (NIHR) Imperial Biomedical Research Centre (BRC). The authors also wish to thank the patients and staff at Chelsea and Westminster Hospital who participated in this study.

Disclosures

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Author contributions

N. M. S., L. E., N. I. and M. R. J. made a substantial contribution to the conception and design of the project and its interpretation; were responsible for the acquisition, analysis and interpretation of the data; and drafted the work. N. M. S. and L. E. performed the experiments. All authors contributed to revision of the manuscript and have approved the final version. All authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.