We recruited pregnant women attending the Princess Royal Maternity Hospital, Glasgow. Women with clinical evidence of infection or abruption or suffering from any pregnancy complications such as pregnancy-induced hypertension, systemic disease such as diabetes, or any other medical problem that would predispose them to an alteration in their immune response, and women with multiple pregnancies were all excluded. Written informed consent was obtained from each woman prior to recruitment and the study was approved by the Local Research Ethics Committee (RN05OB013).

Heparinized peripheral blood samples were taken from pregnant women in four different clinical groups as follows: preterm pregnant women at 28 to <37 weeks of gestation who were not in labour (PTNL); preterm pregnant women at 28 to <37 weeks of gestation who were in labour (PTL); term pregnant women at 37–42 weeks of gestation who were not in labour (TNL); and term pregnant women at 37–42 weeks of gestation who were in labour (TL). No woman was sampled more than once, or included in any more than one of these groups. The diagnosis of labour was made in women who had signs and symptoms of labour (regular painful contractions). Additionally, to avoid including women with these signs and symptoms but who were not actually in labour [a common scenario preterm (^{Kenyon et al., 2001})], we excluded those who did not deliver within 14 days of sampling. These women were deemed not to have been in labour, their results were not included in the analysis and their details are not presented here. Non-labouring women were recruited from antenatal clinics, and labouring women were recruited from the labour ward. Gestational age was estimated in all women from an ultrasound scan which had been performed in the first trimester. All women in the preterm labouring group had been given betamethasone 12 mg as part of their routine clinical management prior to sampling.

The absolute cell numbers of leukocyte subpopulations from whole heparinized peripheral blood were determined using a Sysmex KX-21 Haematology Analyzer which categorizes leukocytes according to cell size.

Fluorescence-activated cell sorting (FACS) with a FACSCalibur machine (Becton Dickinson, Heidelberg, Germany) was used to identify and, subsequently, quantify cell surface marker expression according to standard protocols. Briefly, 50 µl of whole blood from each subject was incubated with the appropriate monoclonal antibodies and isotype controls and 2 ml 1× Pharm Lyse lysing buffer (BD Pharmingen) for 15 min in the dark at room temperature. Samples were then centrifuged at 200 g for 5 min and washed in 2 ml FACS buffer containing phosphate-buffered saline (PBS) (Invitrogen Life Technologies, UK), 1% fetal bovine serum (Invitrogen) and 0.1% sodium azide (Sigma-Aldrich), fixed in 2% paraformaldehyde and stored at 4°C until measured by flow cytometry within 24 h. Cells were then washed and suspended in FACS flow buffer (Becton Dickinson) before analysis by flow cytometry. At least 10 000 cells were acquired for analysis for each antibody combination.

We used fluorescein isothiocyanate (FITC)- or phycoerythrin (PE)- or allophycocyanin (APC)-conjugated antihuman monoclonal antibodies to surface antigens: -CD3, -CD4, -CD8, -CD11a, -CD11b, -CD14, -CD15, -CD16, -CD19, -CD26, -CD28, -CD45, -CD54, -CD56, -CD62L and -CD64 (BD Pharmingen) and FITC-, PE- or APC-conjugated immunoglobulins of the relevant isotypes (BD Pharmingen) as controls. Each monoclonal antibody was used at saturating concentrations optimized by initial titrations for flow cytometric staining. The negative control panels were comprised of mixtures of isotype (IgG1, IgG2a and IgM) (BD Pharmingen). The data were saved for later analysis using the CellQuest software (Becton Dickinson).

We used recommended controls for the settings (^{BD Biosciences, 2002}). The FITC- or PE- or APC-conjugated isotypes were used as negative controls. Anti-CD45, -CD3, -CD4 or -CD8 was used for the settings for the leukocytes and lymphocytes, respectively, whereas CD14 and CD15 were used for the settings for monocytes and neutrophils, respectively. The same batch of these isotype controls and specific antibodies were used in every experiment. The FACSCalibur instrument was calibrated weekly using BD calibration beads (with multi-peaks). The beads were used to adjust instrument settings, set fluorescence compensation and check instrument sensitivity, as it has been shown that such bead-adjusted settings reduced day-to-day variability of measurements of fluorescence intensity (^{Waxdal et al., 1998}; ^{BD Biosciences, 2002}; ^{Korpi-Steiner et al., 2008}). The assays were optimized at the beginning of this study, and thereafter, we did not change the major settings.

Neutrophils were separated from whole blood using dextran and ficoll sedimentation. Briefly, heparinized whole blood was transferred to 15 ml conical tubes (BD), and 6% dextran T500 (Sigma-Aldrich) was added to a final concentration of 1% and gently mixed. The tubes were left standing to allow the red blood cells (RBCs) to settle for 1 h at room temperature. The cell-rich upper phase of the mixture was carefully aspirated. Hypotonic lysis was briefly applied to remove the remaining RBCs. The cells were carefully layered over Ficoll-Hypaque (Amersham Biosciences 1.077 density), then centrifuged at 400g for 30 min at room temperature. Neutrophils settled to the bottom of the Ficoll, with mononuclear cells remaining at the ficoll interface. The supernatants were discarded and the neutrophils were suspended in 15 ml of PBS buffer (Invitrogen, UK), and then centrifuged at 300g for 5 min. The cells were resuspended in PBS containing 0.05% BSA (Sigma) at a concentration of 2 × 10⁶ cells/ml and kept on ice until the cell migration assay.

Neutrophil migratory capacity was studied in response to N-formyl-methionyl-leucyl-phenylalanine (fMLP) in a 96-well disposable chemotaxis/cell migration chamber (ChemoTx, 106-5, Neuro Probe). fMLP (Sigma-Aldrich) was used at concentrations optimized by initial titrations at a range from10⁻⁵ to 10⁻⁸M. fMLP concentration at 10⁻⁷M was selected for final experimental protocols. The stock solution of fMLP was prepared at 1 × 10⁻² M in DMSO, and aliquots were stored at −70°C.

The chemotactic cell migration assay was according to the manufacturer's instructions (ChemoTx, Neuro Probe) with minor modification. Briefly, the ChemoTx microplate wells were filled with 29 µl test solutions such as positive (fMLP, 10⁻⁷M) and negative controls (absence of fMLP). The framed membrane filter was positioned over the filled ChemoTx microplate. Fifty microlitres of cell suspension was loaded on the filter top. All the test samples and negative controls were prepared in triplicates. The ChemoTx microplate and filter were covered with the lid and incubated at 37°C in humidified air with 5% CO₂ for 1 h. After the incubation, the cells on the membrane filter top were gently wiped off using cotton wool swabs. The ChemoTx microplate and filter with lid was centrifuged at 400 g for 1 min in order to remove any migrated neutrophils from the membrane filter to the wells. The migrated cells in the ChemoTx microplate wells were counted using trypan blue exclusion test. The number of cells migrating towards the fMLP (10⁻⁷M) was expressed as a ratio of those randomly migrating in the absence of fMLP.

ROS production was measured using flow cytometry. Stock solution of 2′,7′-dichlorofluorescin diacetate (DCFH-DA) (Sigma-Aldrich) was prepared at 10⁻²M in absolute ethanol, and aliquots were stored at −20°C. A concentration at 5 × 10⁻⁶ M of DCFH-DA (Sigma-Aldrich) was used for the ROS production assay.

First, 100 µl of the heparinized peripheral blood was added to 2 ml of RBC lysis buffer (containing 0.83% NH₄Cl, .0.1% KHCO₃ and 0.009% EDTA). The tube was allowed to stand for 5 min at room temperature to enable the RBCs to lyse. The cells were then centrifuged at 200g for 5 min at room temperature and washed in PBS (Invitrogen). The cells were resuspended in 1 ml of PBS containing 0.1% BSA (Sigma Aldrich) and 5 × 10⁻⁶ M of DCFH-DA and incubated at 37°C for 15 min for the basal ROS production. The samples with 100 ng/ml of PMA (phorbol-12-myristate-13-acetate, Sigma-Aldrich) for stimulation were further incubated at 37°C for 20 min. The test samples were kept on ice until flow cytometric analysis. Propidium iodide dye was used to gate for live cells and was added to each sample at a final concentration of 5 µg per ml.

ROS in the cells oxidize the membrane-permeable form of the probe 2′,7′ dichlorofluorescein (DCFH), changing the non-fluorescent intracellular DCFH into highly fluorescent dichloroflurescein (DCF). The non-polar and non-fluorescent form of DCFH is DCFH-DA. DCF data were collected in the FL1 channel with a FACSCalibur machine using the data acquisition software CellQuest (Becton Dickinson). At least 10 000 cells were examined in each sample. DCF data (from non-stimulated cells) were expressed as mean fluorescence intensity as basal ROS production, or expressed as the ratio of mean channel fluorescence intensity (MFI) data from non-stimulation/MFI data from stimulation, with gating on the appropriate cell subtype.

Human plasma from heparinized peripheral blood was isolated by centrifugation and stored at −70°C. Circulating cortisol (CORT), progesterone (P4) and estradiol (E2) levels in the plasma samples were measured according to the manufacturer's instructions using an Immulite Immunoassay Analyzer [Immulite 1000 cortisol (LKCO1), Immulite 1000 progesterone (LKPG1) and Immulite 1000 estradiol (LKE21) from Diagnostic Products Corporation (DPC, CA, USA)].

The leukocytes were isolated using dextran sedimentation (as detailed earlier), and total RNA was isolated according to the manufacturer's instructions using ABI PRISM 6100 Nucleic Acid PrepStation (Applied Biosystems, UK). RNA was reverse-transcribed using a High-Capacity cDNA Archive Kit (Applied Biosystems, UK) according to the manufacturer's instructions. cDNA was quantified using TaqMan technology on an ABI Prism 7900HT (Applied Biosystems). Briefly, 1.25 µl of 20× target assay or control assay mix was added to 12.5 µl of 2× Mastermix (Applied Biosystems), 10.25 µl deionized distilled water and 1 µl cDNA. The target assay mix and the 18S rRNA (4310893E) control probe were purchased from Applied Biosystems. Primers and fluorogenic probes for MCP-1 (CCL-2), ICAM-1, IL-1β, IL-6, IL-8, TLR-2 and TLR-4 were purchased from Applied Biosystems (Hs00234140_m1; Hs00277001_m1; Hs00174097_m1; Hs00174131_m1, Hs00174103_m1; Hs00152932_m1; and Hs00152939_m1). The thermal cycling conditions comprised an initial step at 50°C for 2 min and 95°C for 10 min, and then 40 cycles at 95°C for 15 s and 60°C for 1 min. Data analysis was performed using the Applied Biosystems Sequence Detection Software 2.2, which calculated the threshold cycle (CT) values. The samples tested were normalized to the housekeeping gene 18S rRNA. The relative gene expressions of the samples from labouring women at term or preterm were compared with the samples from non-labouring women at term or preterm, respectively. The expression of the target assays was normalized by subtracting the CT value of the 18S control from the CT value of the relevant target assay. The fold increase relative to the 18S control was obtained by using the formula 2^-deltaCt.

A two-way between-group analysis of variance was conducted to explore the impact of each of gestation and of labour status on the dependent variables (cell surface marker expression, cell migration, ROS production, etc). Two-way between-group analysis of variance is appropriate where there are two independent factors specifying the nature of the measurements (^{Altman, 1991}). In this case, the factors were ‘gestation’ (term or preterm) and ‘labour status’ (in labour or not in labour). Although there is no requirement for data to be normally distributed in order to test analysis of variance (^{Altman, 1991}), samples should be taken from populations of equal variances. We used Levene's test to determine equality of error variances. Where Levene's test indicated that variances of the two groups were not equal, we used a more stringent level of statistical significance as recommended (^{Pallant, 2007}).

Two-way analysis of variance was used to determine whether there was any interaction between the two independent variables (gestation and labour status). Where there was no significant interaction between gestation and labour, we report the main effects of these variables as recommended. Where there was a significant interaction between gestation and labour, we separately analysed and reported the effects of labour in the preterm and term gestation groups using either a two-sided t-test (for the majority of data which were normally distributed) or a Mann–Whitney U test (for data which were not normally distributed). For consistency, data are expressed as means ± SEM throughout.

Analysis of the subsets of circulating leukocytes (Fig. 1) showed a greater absolute number of monocytes and neutrophils in labouring compared with non-labouring group women (P < 0.01, P < 0.001, respectively) (Fig. 2; quantified using the Sysmex KX-21 Haematology Analyzer). There were no differences in lymphocyte numbers between labouring and non-labouring women, nor were there any differences between term and preterm groups for any of the leukocyte subset counts.

When we analysed the frequencies of the different subsets of peripheral blood lymphocytes and monocytes in gestation-matched women, we observed significantly lower proportions of CD3⁺ T cells (P < 0.01) and significantly greater proportions of CD14⁺ monocytes (P < 0.05) and CD16/CD56⁺ NK cells (P < 0.05) in labouring compared with not labouring women (representative plots in Fig. 3, graph of frequencies in Fig. 4A, D and E). We did not observe any significant differences in the percentage distribution of CD19⁺ B cells with labour (data not shown). Analysis of CD4⁺ T cells showed a significant interaction between labour and gestation, further exploration of the differences showing a significantly lower proportion of CD4⁺ T cells and a greater proportion of CD8⁺ T cells in preterm labouring versus non-labouring women (P < 0.01 and 0.05, respectively, t-test) but no differences between labouring and non-labouring women at term (Fig. 4B and C). There were no significant effects of gestation on CD3⁺ T cells, CD14⁺ monocytes or CD16/CD56⁺ NK cells (Fig. 4).

After incubation at 37°C for 1 h, neutrophil migration in response to fMLP (10⁻⁷ μM) expressed as a ratio of those randomly migrating in the absence of fMLP was significantly greater in labouring compared with non-labouring women (P < 0.001) (Fig. 5A).

The basal ROS values in selectively gated neutrophils or monocytes were not significantly different in labouring compared with non-labouring women (data not shown). There were no effects of labour or gestation on stimulated ROS (oxidative burst generated in response to 100 ng/ml PMA stimulation) in neutrophils. In contrast, in monocytes, there was an interaction between gestation and labour on oxidative burst, with PMA stimulation being significantly greater in preterm, but not in term, labour, compared with gestation-matched women (P < 0.001, t-test) (Fig. 5B).

In labouring women, we observed a significant increase in the proportion of monocytes expressing each of the cell surface markers CD64 (P < 0.01), CD11b (P < 0.001) and CD62L (P < 0.05) compared with monocytes from non-labouring women (Fig. 6A, B and C). There were no differences in the percentage of monocytes expressing CD54 (data not shown).

There was no effect of labour on the percentage of T cells expressing CD54 (data not shown) or CD62L (Fig. 6F). A significantly greater number of T cells from labouring women expressed CD28 (P < 0.01). There was an interaction between the effect of labour and gestation in the proportion of T cells expressing CD26: further analysis showed an increase in CD26 in preterm, but not in term, labouring women compared with gestation-matched non-labouring women (P = 0.001, PTL/PTNL, t-test) (Fig. 6D).

Over 90% of neutrophils expressed each of the cell surface markers CD11a, CD 11b and CD62L, and there were no significant differences in the numbers of expressing cells between labouring and non-labouring women (data not shown).

Analysis of the MFI of surface activation markers and adhesion molecules is shown in Fig. 7. Neutrophils from labouring women showed a significantly higher surface expression of CD11a and CD11b (P = 0.001) and CD62L (P < 0.02) compared with the non-labouring women. Monocytes from labouring women showed significantly higher surface expression of CD11b (P = 0.001) and CD64 (P < 0.05) when compared with monocytes from non-labouring women. There was an interaction between the effect of labour and gestation in MFI of CD62L on the surface of monocytes: monocytes from preterm, but not from term, labouring women showed significantly higher expression of CD62L compared with gestation-matched non-labouring women (P < 0.05, PTL/PTNL, t-test). T lymphocytes from labouring women showed significantly higher expression of CD26 (P < 0.001) and CD62L (P < 0.01) compared with non-labouring women. There was a significant interaction between the effect of labour and gestation in MFI of CD28 on the surface of T cells with preterm, but not with term, labouring women showing significantly higher expression. However, analysis within each gestation group also found higher expression of CD28 in both preterm P < 0.01 (Mann–Whitney U) and term P < 0.05 (t-test) labouring women versus gestation-matched controls.

Levels of estradiol and progesterone were significantly greater in term versus preterm pregnant women (P = 0.01) (Fig. 8A and B). There was an interaction between the effect of labour and gestation in plasma cortisol levels: further analysis showed a significant increase in cortisol in labour at term (t-test P < 0.001) but not at preterm. There was no effect of labour on progesterone or estradiol levels.

The quantification of mRNA gene expression in maternal circulating leukocytes is shown in Fig. 9. Expression levels of MCP-1 (CCL-2), IL-1β and IL-8 (CXCL8) were significantly greater in labouring women compared with non-labouring women (P < 0.01) (Fig. 9A, B and D). There was no effect of labour on IL-6 or ICAM-1 expression (data not shown). Messenger RNA expression of the innate immune receptor TLR2 was significantly greater in labouring compared with non-labouring women (P = 0.001) (Fig. 9E). There was no effect of gestation on any expression of any of these genes.