A bias of T cell immunity towards type 2 (Th2) is thought to be critical for normal pregnancy. Pathological pregnancies, such as pre-eclampsia, are characterised by cell-mediated (Th1) immune dominance. The Th1/Th2 paradigm, however, is too simplistic. Normal pregnancy is associated with a systemic inflammatory response which increases throughout gestation. This inflammatory response is exaggerated in pre-eclampsia, a syndrome of the third trimester. T helper (Th) cells are considered the primary mediators of these altered immune responses, and other T cells, i.e. T cytotoxic (Tc) cells, and lymphocytes of the innate immune system, i.e. natural killer (NK) and NKT cells, have been largely disregarded. In this study, we have used novel pan type 1 (IL-18 receptor) and pan type 2 (ST2L) lymphocyte function markers in four-colour flow cytometry to broadly characterise peripheral blood lymphocyte populations from non-pregnant, normal pregnant and pre-eclamptic women. There were no changes in the Th1/Th2 or Tc1/Tc2 cell ratios between the three groups; however, the NK1/NK2 and NKT1/NKT2 cell ratios were significantly decreased in normal pregnancy compared with non-pregnant (p <0.001 and p <0.01, respectively) and pre-eclamptic women (p <0.05). These results confirm that immunoregulation occurs in pregnancy, but suggest a dominant role of the innate rather than the adaptive immune system.
There are two foetal-maternal immune interfaces in human pregnancy. In early pregnancy, interface I comprises a localised interaction in the decidua between immune cells and invasive extravillous cytotrophoblast 1. Interface II is between circulating maternal immune cells and the syncytiotrophoblast that forms the surface of the haemochorial placenta. Interface I virtually disappears in the third trimester with regression of the invasive trophoblast 2] and the degeneration of associated decidual lymphocytes 3. Interface II is activated with the onset of the uteroplacental circulation at 8–9 wk 4] and enlarges with placental growth to become the dominant maternal-foetal immune interface towards the end of pregnancy.
It has been postulated that successful pregnancy induces an immune bias towards T helper type 2 (Th2) immunity 5. The hypothesis has been useful, but in relation to current knowledge and what is known about human pregnancy, the Th1/Th2 hypothesis is increasingly incomplete 6–8. For example, normal pregnancy stimulates pro-inflammatory changes, which are further intensified in pre-eclampsia 9. In this study, we use new markers to examine the importance of circulating NK cells in the regulation of the maternal systemic immune response during the third trimester of human pregnancy.
The role of Th cells in pregnancy needs to be related to the allogenicity of the trophoblast. The human trophoblast does not express classical MHC antigens (HLA-A, -B and -D). Extravillous cytotrophoblast, the foetal tissue at immune interface I, expresses HLA-C, -E and -G 1. The syncytiotrophoblast (interface II) appears not to express membrane-bound HLA at all 10. Without the MHC molecules that enable antigen presentation, it is difficult to envisage the development of T cell-based immunity to the syncytiotrophoblast at interface II. Moreover, none has been convincingly demonstrated in human pregnancy. In addition, the shift from interface I in early pregnancy to interface II in late pregnancy has not been taken into account. Interface I involves an interaction between the HLA-expressing extravillous cytotrophoblast with a decidual immune cell population (NK cells, NKT cells, γδ T cells and macrophages) that differs substantially from that in the peripheral blood. Failure of immune adaptation at interface I leads to abortion. Many reports have addressed this topic and have highlighted the importance of NK cells in both murine and human early pregnancy 1, 11, 12.
This report concerns systemic immune reactivity, in relation to the pregnancy-specific syndrome, pre-eclampsia. Pre-eclampsia is a two-stage disorder 13. The first stage involves deficient placentation during early pregnancy, resulting from shallow invasion of the placental bed by the extravillous cytotrophoblast 14, which may depend on immune reactivity affecting interface I. The second stage, which is confined to the second half of pregnancy when placentation is complete, is a decompensated systemic inflammatory response, which leads to the maternal syndrome 14. It is unclear what links the first and second stages of pre-eclampsia, but the secreted products of placental oxidative stress are currently considered to be the most likely candidates 9. All components of the systemic inflammatory network appear to be engaged in pre-eclampsia, accounting for the diversity of the pathology presented in end-stage pre-eclamptic patients. Differences between this inflammation in normal pregnancy and pre-eclampsia are of intensity, not quality 15. In this study, we have focused on the second stage of pre-eclampsia in order to investigate the roles of NK and T cells in the systemic inflammatory response.
Type 1 and type 2 immune responses are not limited to Th cells. T cytotoxic (Tc), NK and NKT cells are divisible into type 1 and type 2 subsets. Tc1 and Tc2 cells are distinguished on the basis of cytokine production profiles similar to Th1 and Th2 cells 16, although no major functional differences appear to exist and both subsets are cytotoxic 17. Peritt et al. 18] first reported NK1 and NK2 subsets that differ in their intracellular cytokine expression 19, 20 and which allow the CD56dim and CD56bright populations of peripheral blood to be further subdivided into type 1 and type 2 subsets 21. NKT lymphocytes, defined by the expression of the invariant T cell receptor Vα24 and a biased set of T cell receptor β chains, predominantly Vβ11, are also divisible into two type 1 and type 2 subsets 22. Hence, the immunoregulatory balance of pregnancy may be better considered as a type 1/type 2 issue, which may involve Tc, NK and NKT cells, in addition to Th cells.
The type 2 shift of normal human pregnancy in peripheral blood lymphocytes persists until delivery 23, 24. In pre-eclampsia, the bias is towards type 1 immunity, which is similar to that of non-pregnant women 25–27. The shifts have been detected by measurements of cytokine levels either in plasma or secreted by preparations of PBMC. However, it is not always clear which cell types produce the cytokines being measured. Investigations of lymphocyte subsets have been limited by the lack of suitable surface markers identifying human type 1 and type 2 cells. Recently, we reported stable surface markers that appear to discriminate human type 1 from type 2 cells across the T and NK cell lineages 20. IL-18 receptor (IL-18R) is selectively expressed on type 1 lymphocyte lineages, while ST2L, a homologue of IL-1R, is selectively expressed on type 2 lymphocyte lineages in mice and humans 20, 28, 29. We have previously used these markers to measure the type 2 shift in individuals suffering from AIDS 20, 30. Both ST2L and IL-18R are members of the Toll-like receptor (TLR) superfamily, most of which promote type 1 responses with activation of NF-κB. IL-18 induces IFN-γ production from Th1 cells and NK cells 31] such that the expression of IL-18R marks lymphocytes that have the potential to stimulate type 1 responses. ST2L has as yet no known ligand and does not activate NF-κB 32. In murine cells, it down-regulates type 1 responses by sequestering the adaptor molecules MyD88 and Mal 33.
These markers create new possibilities for the study of type 1/type 2 immunity in human disorders. Using antibodies to these receptors and a four-colour antibody labelling protocol, we identified type 1 and type 2 cells in each of the four lymphocyte lineages (Th, Tc, NK and NKT). We then studied peripheral blood lymphocytes from normal pregnant, pre-eclamptic and non-pregnant control women to evaluate the immune status of unstimulated cells in each population. To the best of our knowledge, this is the first report to differentiate between CD56dim and CD56bright NK cells and to include CD3+CD56+ NKT lymphocytes in the study of type 1/type 2 immunity in pre-eclampsia compared to normal third-trimester and non-pregnant women.
Total circulating type 1 and type 2 lymphocytes in non-pregnant and in normal pregnant and pre-eclamptic women during the third trimester
Within the total lymphocyte population (as defined by the lymphocyte gate on a forward scatter versus side scatter plot), the percentage of IL-18R+ cells was significantly decreased in normal pregnancy (P) compared to both non-pregnant (C) and pre-eclamptic (PET) women (Fig. 1A). The mean channel brightness (mean fluorescence intensity) of the IL-18R+ lymphocytes within the lymphocyte gate was also significantly decreased in normal pregnancy compared to both non-pregnancy and pre-eclampsia (Fig. 1B), suggesting that the cell surface IL-18R density is lower in normal pregnancy. There was no difference between non-pregnant and pre-eclamptic women in either percent positive type 1 lymphocytes or mean channel brightness of IL-18R+, type 1 cells. Similar shifts in type 2 ST2L+ cells or fluorescence intensity were not observed within the lymphocyte gate (data not shown).
Circulating type 1 and type 2 T lymphocytes in non-pregnant and in normal pregnant and pre-eclamptic women during the third trimester
The percentage of Th1 cells did not differ between non-pregnant, normal pregnant and pre-eclamptic groups (Fig. 2A). The percentage of Th2 cells was significantly increased in normal pregnancy compared to both non-pregnant and pre-eclamptic women (Fig. 2B). When calculated as a ratio, Th1/Th2 lymphocytes showed no difference between each patient group (Fig. 2C). Type 1 Tc cells were decreased in normal pregnancy compared to the other groups; however, this shift did not reach statistical significance (Fig. 2D). There was no change in Tc2 lymphocytes between the three patient groups (Fig. 2E). The Tc1/Tc2 ratio was also lower in normal pregnancy compared to non-pregnant and pre-eclamptic women, but did not reach statistical significance (Fig. 2F).
Circulating type 1 and type 2 NK cells in non-pregnant and in normal pregnant and pre-eclamptic women during the third trimester
The percentage of type 1 CD56dim NK cells was decreased in both normal pregnant and pre-eclamptic women when compared to non-pregnant women (Fig. 3A). Type 2 CD56dim NK cells were increased in normal pregnancy compared to the other patient groups; however, this shift only reached significance in comparison with the non-pregnant control group (Fig. 3B). The ratio of NK1/NK2 (CD56dim) was significantly decreased in both normal pregnant and pre-eclamptic women compared to non-pregnant women. In addition, the NK1/NK2 ratio was also significantly decreased in normal pregnant compared to pre-eclamptic women (Fig. 3C). Numbers of type 1 CD56bright NK cells were similar between each patient group (Fig. 3D); however, type 2 CD56bright NK cells were significantly increased in normal pregnancy compared to both non-pregnant and pre-eclamptic women (Fig. 3E). The ratio of NK1/NK2 (CD56bright) was significantly decreased in normal pregnant women compared to both non-pregnant and pre-eclamptic women (Fig. 3F). Type 1 CD3+CD56+ NKT lymphocytes did not change in normal pregnancy (Fig. 3G). However, type 2 CD3+CD56+ NKT cells were significantly increased in normal pregnancy compared to both non-pregnant and pre-eclamptic women (Fig. 3H), and the NKT1/NKT2 ratio was significantly decreased in normal pregnancy compared to both non-pregnancy and pre-eclampsia (Fig. 3I).
When all lymphocytes were considered together, our data confirmed previous studies 24–27, 34, 35 of a type 2 shift in the second half of normal pregnancy. This shift was significantly less in pre-eclampsia. With regard to the Th subset, the percentages of Th1 cells were unchanged but those of Th2 cells were significantly increased in samples from normal pregnant but not pre-eclamptic women. However, Th1/Th2 ratios were similar in all three patient groups. These results are not consistent with previous reports on Th1/Th2 immunity in pregnancy and pre-eclampsia 24, 26. Many of these studies have involved the use of PBMC stimulation in culture prior to cytokine quantification by ELISA 26. With this technique, it is impossible to identify the cell types producing the cytokines in question, and the assumption has been made that the changes are due to Th1 and Th2 cells. Other studies have used intracellular cytokine measurements 24. Although this allows the identification of the cell types producing a particular cytokine, it also requires non-specific stimulation in culture, which may trigger cytokine production by T cells that are not normally activated in vivo. Furthermore, the majority of these studies only investigated Th cells and not NK cells. In our study, cells were analysed without stimulation and immediately after isolation from peripheral blood, and shifts were consistently observed in the NK cell subsets (CD56dim and CD56bright) and in CD3+CD56+ NKT cells. For all three populations, there were significant decreases in the type 1/type 2 ratios in normal pregnancy relative to non-pregnancy, and significant increases in pre-eclamptic pregnancy relative to normal pregnancy. Although these changes were significant, their magnitude was small. In contrast, no significant changes in the T cell ratios were detectable. In all cell types, greater changes were seen in type 2 cells than in type 1 cells, suggesting that there is an enhancement of type 2 immunity during normal pregnancy rather than a suppression of type 1 immunity compared to non-pregnant and pre-eclamptic women. Whether or not these significant differences are biologically important is not demonstrated in this study.
The data show marked overlap between groups, as do all comparable human studies (for example 25, 36, 37), but, as mentioned above, our study differs from such studies in that the lymphocytes were not stimulated in vitro. It is to be expected that stimulation would magnify the small but significant differences seen here. The changes in type 1/type 2 immunity were detected using these markers despite the fact that the women were constitutionally heterogeneous in all respects except with regard to pregnancy or not, and pre-eclampsia or not. Furthermore, pre-eclampsia itself is a heterogeneous state with multiple contributory factors 38. The significant differences that are seen are not due to outliers in any of the groups. These represent different patients in different groups and their removal does not alter the test of significance. These data confirm the previous studies 34, 37 that NK cells are involved in the type 1/type 2 shift of normal pregnancy, and extend this concept by demonstrating that different NK subsets contribute to these changes.
The participation of NK and NKT cells in the type 1/type 2 shifts of pregnancy is not surprising since they are critical to host defence through elaboration of cytokines and cytolytic activity. CD56bright NK cells comprise a small proportion of circulating NK cells, although they are the major population in early gestational decidua 39. CD56bright NK cells are a functionally distinct subset of mature NK cells, which are primarily responsible for cytokine production in response to monokines 40. CD56dim NK cells secrete significantly less pro-inflammatory cytokines but have greater cytolytic activity. NKT cells are a specialised population of TCRαβ+ T cells that co-express receptors of the NK lineage. They express a restricted TCR repertoire, with the majority expressing an invariant TCR α chain 41–43. NKT cells have the unique potential to produce large amounts of cytokines within minutes of activation 44, more per cell than NK cells or antigen-specific T cells 45, 46. NKT cells are the lymphocytes most likely to be the subset spontaneously producing cytokines in peripheral blood 47] and appear to play a dominant role in immune regulation 48. Their in vivo stimulation leads to activation of both innate and acquired immunity in humans 49.
Recently, Darmochwal-Kolarz et al. 34] examined intracellular cytokine levels in peripheral blood Th, Tc and NK cells in normal pregnant and pre-eclamptic women. They reported increased IFN-γ levels in pre-eclampsia compared to normal third-trimester pregnancy only in NK cells (identified only as CD16+ mononuclear cells), and no changes in IFN-γ levels in Th or Tc populations between the groups. Although we have not examined IFN-γ production in pre-eclampsia in this study, our results similarly show that the shift away from type 2 to type 1 in pre-eclampsia was predominantly in the NK and not the T cell populations. Th cells are also involved, but apparently to a lesser extent. If the changes in expression of these type 1 and type 2 markers reflect the patterns and relative magnitudes of cytokine production, then the evidence points to more substantial changes in type 1/2 biases in normal and pre-eclamptic pregnancies among the NK/NKT cells in peripheral blood. To show this would ideally require measurements of cytokine production from separated NK cell subsets cultured from circulating PBMC. This would not be feasible with the volumes of blood that can be ethically obtained from human subjects.
If they are confirmed, these findings suggest the need to change our perspective on immunoregulation in pregnancy. T cells have traditionally been though to be central to this process. However, the current and other studies suggest that the innate immune system, involving NK, NKT and possibly dendritic cells, may be dominant at interface II (in the peripheral circulation), as has been reported for interface I (the decidua) 50. The concept does not exclude the involvement of T cells but relegates it to a secondary phenomenon. We previously showed that normal human pregnancy is characterised by a systemic inflammatory response involving monocytes and granulocytes. This innate activation then leads to increased production of IL-12 15, 51 and IL-18 (Germain S. J., unpublished observation). We postulate that the inflammatory response of normal pregnancy does not lead to activation of NK or T cells, owing to the down-regulation of the IL-18R (Fig. 1).
A potential mechanism for this innate cell activation could be through TLR engagement by syncytiotrophoblast fragments, providing an alternate mechanism of activation which does not require T cell recognition of paternal MHC antigens. TLR are potential candidates since they act as sensors for cell debris and are able to stimulate strong inflammatory responses 52. Recent studies examining maternal regulatory T (Treg) cells have demonstrated an increase in numbers of circulating Treg cells during both murine 53] and human 54] pregnancy, suggesting a potential role for these cells in immunoregulation. It is possible that Treg cells present in the circulation during pregnancy may exert a suppressive effect over other T lymphocytes (Th and Tc), which then results in decreased T cell activation and, therefore, limited involvement in the type 1/type 2 shift of pregnancy.
In pre-eclampsia, the gestational inflammatory response is exaggerated. The IL-18R is more intensely expressed as it is in non-pregnancy, which would promote NK cell activation and IFN-γ production, with subsequent activation of Th and Tc cells. Thus, T cell changes in pregnancy may be secondary to changes in the innate immune system. In addition, T cells have a low resting level of IL-18R expression which is up-regulated by IL-12 and IL-18. It may be pertinent that, amongst human PBMC, IL-12 preferentially up-regulates IL-18R expression by NK cells, regardless of whether costimulants are also used, whereas on T cells, its expression is only slightly increased 55. It is not known whether the increase in the type 1/type 2 ratio in pre-eclampsia reflects a failure to achieve the normal type 2 bias at any stage of pregnancy or results from a loss of the normal type 2 bias as pre-eclampsia evolves. Only serial studies of pregnant cohorts, in which some women become pre-eclamptic, will provide this information. More detailed analysis of circulating NK cell subsets in pregnancy is also required.
Materials and methods
Pre-eclamptic women (n = 15) were recruited from those admitted to the high-risk pregnancy Maternal Medicine Unit at the John Radcliffe Hospital Women's Centre, Oxford, UK. Pre-eclampsia was characterized by new hypertension and proteinuria occurring in the third trimester of a previously normal pregnancy. Hypertension was defined as a diastolic blood pressure of ⩾90 mmHg on two separate occasions within a 24-h period and proteinuria, as defined either by ⩾500 mg protein in 24-h urine collection or ⩾2+ protein at least twice using a dipstick, according to the International Society for the Study of Hypertension in Pregnancy. The study population included twelve primiparous and three multiparous women. None of the pre-eclamptic women was affected by pre-existing clinical disorders such as chronic hypertension or renal disease before pregnancy, and none of the pregnancies were complicated by preterm labour or chorioamnionitis. As a control group, 15 healthy pregnant women, twelve primiparous and three multiparous, were recruited from antenatal clinics at the John Radcliffe Hospital and various local general practitioner surgeries in Oxford, and matched for age (±4 years), parity (0, 1–3, 4+) and gestational age (±13 days). All cases and controls were in the third trimester of pregnancy, not in labour at the time of sampling, and had singleton pregnancies, with no known foetal abnormalities. All control pregnancies progressed normally to term. Healthy, non-pregnant women (n = 15) were recruited from hospital staff and were of reproductive age, not on any medication apart from oral contraceptive and had no history of chronic inflammatory disease or allergy. Clinical characteristics of each group are presented in Table 1. There were no statistical differences between the three groups in terms of age or parity, and in the pregnant groups between gestational age, or blood pressure and proteinuria at the time of the first antenatal visit. Maximum blood pressure and maximum proteinuria were significantly increased in the pre-eclamptic group compared to the normal pregnant group. This study was approved by the Aylesbury Vale Local Research Ethics Committee, and informed consent was obtained from all subjects prior to venipuncture.
|Non-pregnant (n = 15)||Normal Pregnancy (n = 15)||Pre-eclampsia (n = 15)||p (ANOVA)|
|Age, years||30.2 (6.5)||29.3 (6.2)||30.1 (6.2)||NSa)|
|– Gestational age, wk||–||35.2 (3.2)||35.1 (3.1)||NS|
|– Bookingb) blood pressure, mmHg||–||110/65 (± 13.0/15.0)||115/69 (± 15.4/11.7)||NS|
|– Maximum blood pressure, mmHg||–||134/86 (7.2/6.4)||174/113 (14.5/8.6)||p <0.0001|
|– Bookingb) proteinuria, mg/24 h||–||NADc)||NAD||NS|
|– Maximum proteinuria, mg/24 h||–||NAD||5266.1 (range 678–20 670)d)||p <0.01|
Preparation of PBMC
Venous blood (10 mL) was obtained from each individual using a syringe and needle, and anticoagulated with preservative-free sodium heparin (10 IU/mL blood; Sigma, St. Louis, MO). Blood samples were then diluted (1 : 1) in 10 mL PBS (Oxoid, Basingstoke, UK), and PBMC were separated by density gradient centrifugation (Ficoll-Paque; Amersham Biosciences, Uppsala, Sweden) and resuspended in PBS + 0.1% BSA (PBS/BSA; Sigma) to a standard concentration of 1 × 107 cells per mL.
Directly conjugated monoclonal antibodies were used to label surface antigen markers for Th cells (CD3+CD4+), Tc cells (CD3+CD8+), NK cells (CD16+CD56dim and CD16−CD56bright) and NKT lymphocytes (CD3+CD56+). Each of these cell types was also labelled with unconjugated antibodies to distinguish the type 1 (IL-18R+) from the type 2 (ST2L+) subset. All primary antibodies were IgG1. The four-colour antibody combinations and working concentrations (as determined by antibody titration experiments) are outlined in Table 2. Anti-IL-18R (R & D Systems, Oxford, UK) and anti-ST2L (as described 20) antibodies were amplified using goat anti-mouse IgG-biotin (Oxford Biotechnology, Oxford, UK) as a secondary antibody and PE-labelled streptavidin (Serotec, Kidlington, UK). Mouse IgG1 isotype control antibodies were either conjugated to FITC (Serotec), energy-coupled dye (ECD) or allophycocyanin (APC) (Coulter, High Wycombe, UK) or were unconjugated (Serotec). All antibodies were titrated with freshly isolated PBMC to determine saturating concentrations, and isotype control sera were used at equivalent immunoglobulin concentrations.
|Antibody dilutions||1 : 25||1 : 50||1 : 25||1 : 10|
|Negative control||Mouse IgG1b)||Mouse IgG1b)||Mouse IgG1c)||Mouse IgG1c)|
|PE-negative control||CD8b)||Mouse IgG1||CD3c)||CD4c)|
|PE-negative control||CD16b)||Mouse IgG1||CD3||CD56c)|
|Type 1 T lymphocytes||CD8||IL-18Rd)||CD3||CD4|
|Type 2 T lymphocytes||CD8||ST2Le)||CD3||CD4|
|Type 1 NK cells||CD16||IL-18R||CD3||CD56|
|Type 2 NK cells||CD16||ST2L||CD3||CD56|
Antibody labelling of PBMC
From the standard cell suspension, 1 × 106 (100 µL) cells were added to 50 µL ‘staining buffer’ (PBS + 20 mM glucose and 5% normal human serum) containing either unconjugated monoclonal antibodies to leukocyte surface antigens or negative control IgG1. Samples were incubated on ice in the dark for 30 min, washed three times with 1 mL PBS/BSA and spun down in a microcentrifuge at 13 000 × g for 7 s. Pellets were resuspended in 50 µL staining buffer containing the secondary antibody. The incubation and washing steps were repeated for both goat anti-mouse IgG-biotin antibody and streptavidin-PE. Samples were then resuspended in 170 µL of reagent-grade mouse IgG purified immunoglobulin (Sigma) diluted 1 : 6 in staining buffer and incubated on ice in the dark for 15 min, in order to block non-specific binding sites prior to labelling with directly conjugated antibodies. Finally, cells were washed and resuspended in 50 µL staining buffer containing directly conjugated antibody cocktails of anti-CD8-FITC, anti-CD3-ECD and anti-CD4-APC for T cell subsets or CD16-FITC, CD3-ECD and CD56-APC for NK cell subsets (Table 2). These cell mixtures were incubated on ice in the dark for 30 min, washed once in PBS/BSA and resuspended in 1 mL PBS/BSA for immediate analysis by flow cytometry to minimise the possibility of capping.
All samples were analysed on a Beckman Coulter Epics Altra flow cytometer using the 488-nm line of the UV laser and the Helium-Neon (633 nm) laser. The lasers were aligned using Flow Check Beads (Coulter), and settings were adjusted using Flow Set Beads (Coulter) prior to each experiment to permit direct comparisons between individual samples. Photomultiplier tubes (PMT) collected fluorescent light at 525 nm (FITC), 575 nm (PE), 610 nm (ECD) and 675 nm (APC). Due to spectral overlap, fluorescence colour compensation was performed using single-antibody-labelled positive controls before each experiment.
For each sample analysed, 20 000 events were collected, and all data were saved in ‘listmode’ files for subsequent analysis using Beckman Coulter Expo 32 software. Lymphocytes were distinguished from monocytes by forward and side scatter, and gated initially on these physical characteristics. Percentage and mean channel brightness (mean fluorescence intensity) was determined for IL-18R+ lymphocytes within the lymphocyte gate. Subpopulations of Th, Tc, NK and NKT cells were identified within the lymphocyte gate using double-positive antigen expression. Quadrant gates were initially set at ⩽1% using the negative controls for FITC, ECD and APC. Population gates were then drawn on double-positive populations for each Th (CD3+CD4+), Tc (CD3+CD8+), NK cells (CD16+CD56+ and CD16−CD56bright) and NKT lymphocytes (CD3+CD56+; Fig. 4). Four-colour labelling enabled each T or NK cell subpopulation to be analysed simultaneously for their expression of either IL-18R or ST2L, in order to quantify type 1 and type 2 cells, respectively. Plots of PE fluorescence versus side scatter were gated on each lymphocyte subpopulation, and gates were then set within these plots at ⩽1% using the PE-specific negative controls. Type 1 (IL-18R+) and type 2 (ST2L+) positivity was then quantified in each population (Fig. 4), and data are presented as scatter plots with means for each type 1 or type 2 subset. Ratios of type 1/type 2 lymphocytes were also determined for each lymphocyte subpopulation, using numbers of percent positive cells from 1 to 100 for each type 1 and type 2 cells (Fig. 2, 3).
Comparisons between each group in percent IL-18R+ or ST2L+ lymphocytes or type 1/type 2 ratios were analysed using the Mann-Whitney U-test (Graph-Pad Prism Software, San Diego, CA), a standard non-parametric test. Clinical characteristics of matched patient groups (Table 1) were compared using ANOVA (Graph-Pad Prism Software). Differences were considered statistically significant for p values <0.05.
This study was supported by funding provided by the Nuffield Department of Obstetrics and Gynaecology, University of Oxford, UK, and awards from the Ontario Graduate Scholarship Program and The Natural Sciences and Engineering Research Council of Canada. The authors would like to thank research midwives Hazel Coburn and Carol Simms, who were supported by a grant from the Collaborative Research Projects scheme of Oxford Radcliffe Hospitals, NHS Trust, for their assistance in patient recruitment, and Dr. Richard Branton for his assistance with the flow cytometry.