S. Della Bella, Department of Biomedical Sciences and Technologies, Laboratory of Immunology, University of Milan, Via Fratelli Cervi 93, Segrate, 20090 (MI), Italy. E-mail: firstname.lastname@example.org
Successful pregnancy relies on the adaptation of immune responses that allow the fetus to grow and develop in the uterus despite being recognized by maternal immune cells. Dendritic cells (DCs) are central to the control of immune tolerance, and their state of activation at the maternal–decidual interface is critical to the feto–maternal immunological equilibrium. So far, the involvement of circulating DCs has been investigated poorly. Therefore, in this study we investigated whether, during healthy human pregnancy, peripheral blood DCs (PBDCs) undergo changes that may be relevant to the adaptation of maternal immune responses that allow fetal tolerance. In a cross-sectional study, we analysed PBDCs by six-colour flow cytometry on whole blood samples from 47 women during healthy pregnancy progression and 24 non-pregnant controls. We demonstrated that both myeloid and plasmacytoid PBDCs undergo a state of incomplete activation, more evident in the third trimester, characterized by increased expression of co-stimulatory molecules and cytokine production but lacking human leucocyte antigen (HLA)-DR up-regulation. To investigate the contribution of soluble circulating factors to this phenomenon, we also performed culture experiments showing that sera from pregnant women added to control DCs conditioned a similar incomplete activation that was associated with reduced DC allostimulatory capacity, supporting the in vivo relevance of our findings. We also obtained evidence that the glycoprotein hormone activin-A may contribute to DC incomplete activation. We suggest that the changes of PBDCs occurring during late pregnancy may aid the comprehension of the immune mechanisms operated by the maternal immune system to maintain fetal tolerance.
Pregnancy represents a major challenge to the maternal immune system, which must tolerate fetal alloantigens encoded by paternal genes to allow survival and growth of the fetus. Multiple tolerance mechanisms have been described that operate at the feto–maternal interface and promote immune cell homeostasis during pregnancy . Initially, the contact between maternal and fetal cells is restricted to the decidua but, following extravillous trophoblast cell invasion, it is extended to the peripheral circulation . Therefore, successful pregnancy relies on the adaptation of decidual as well as peripheral immune responses that allow the fetus to grow and develop in the uterus in spite of being recognized by maternal immune cells.
Dendritic cells (DCs) are the single most central player in all immune responses. They are widely distributed potent antigen-presenting cells that bridge the arms of innate and adaptive immunity and may therefore be involved centrally in the maintenance of immune homeostasis during physiological pregnancy. Their capacity to promote either immune activation or immune tolerance relies on many factors, including their ability to capture, process and present antigens in peptide–major histocompatibility complexes (MHC) that interact with different classes of lymphocytes, their cell surface expression of co-stimulatory molecules that represent a necessary signal for the activation of lymphocytes and their ability to produce many cytokines that play a critical role in the regulation and polarization of adaptive immune responses [3–6]. All these properties of DCs, that are critical to the resulting quality and intensity of adaptive immune responses, are regulated finely by the environment where DCs encounter the antigen, their anatomical localization and their belonging to different subsets . Peripheral blood is the most accessible source of DCs, where myeloid DCs (mDCs) and plasmacytoid DCs (pDCs) represent the two major DC populations that differ from each other in origin, growth requirements, immunophenotype and function. mDCs are conventional DCs and can be activated differentially by distinct signals coming from microbes, dying cells and immune cells. pDCs are involved mainly in tolerance maintenance in their immature state, but they activate adaptive immune responses efficiently on maturation, having a primary role in anti-viral defences [7,8].
DCs in pregnant women have been characterized mainly at the maternal–decidual interface, where their state of activation has emerged as one of the key players influencing the feto–maternal immunological equilibrium , while so far PBDCs have been poorly investigated [10–12]. Therefore, in this study we performed a detailed immunophenotypic and functional analysis of PBDCs during human healthy pregnancy. To this aim, we used a six-colour flow cytometric method that we optimized recently for the multi-parametric analysis of mDCs and pDCs [13,14]. To evaluate whether the alterations of PBDCs that we observed in late pregnancy were indeed possibly related to soluble circulating factors, we incubated control PBDCs with plasma from pregnant women. To possibly support the in vivo relevance of the PBDC changes occurring in late pregnancy, we also performed parallel incubation experiments on monocyte-derived DCs (moDCs) that allow the analysis of DC allostimulatory capacity. Finally, we investigated whether activin-A (ActA), a glycoprotein hormone of the transforming growth factor (TGF)-beta family central to the endocrine physiology of human pregnancy [15,16] and demonstrated recently to be able to promote the maturation of monocyte-derived DCs (moDCs) into a tolerance-inducing phenotype , possibly contributed to DC incomplete activation.
Materials and methods
Study subjects and specimen collection
Forty-seven healthy pregnant women with no signs of pregnancy complications at inclusion and visiting our maternity out-patient care unit were enrolled into the study. Seven women were in the first trimester (mean age 27·8 ± 3·1 years; mean gestational age at sampling 10·9 ± 1·1 weeks), 16 in the second (mean age 33·2 ± 1·4 years; mean gestational age 22·1 ± 1·0) and 24 in the third trimester (mean age 30·8 ± 1·7 years; mean gestational age 34·0 ± 0·5). Twenty-four non-pregnant women in the follicular phase of the menstrual cycle were enrolled as controls (mean age 29·4 ± 3·2 years). None of the non-pregnant controls were using hormonal contraceptives. All subjects were known not to use cigarettes or illicit drugs, or were suffering from any medical conditions (i.e. diabetes, hypertension or metabolic diseases). Peripheral blood samples were obtained by venipuncture and collected into sodium citrate Vacutainer tubes (Becton Dickinson, San Jose, CA, USA). Ethics approval was obtained from the local Institutional Review Committee and a signed informed consent was obtained from all participants.
Counting and immunophenotypic analysis of PBDCs
Whole peripheral blood samples were analysed by six-colour flow cytometry, as described previously . Briefly, PBDCs were identified as mononuclear cells that lacked lineage cell markers (Lin–: CD3–/CD14–/CD16–/CD19–/CD20–) but expressed human leucocyte antigen (HLA)-DR; mDCs and pDCs were identified as DCs expressing CD11c or CD123, respectively. Monoclonal antibodies (mAbs) specific for lineage markers, HLA-DR, CD11c and CD123 were used in all samples to gate on mDCs and pDCs, while the other monoclonal antibodies (mAbs) [fluorescein isothiocyanate (FITC)- and phycoerythrin (PE)-conjugated] were used in varying combinations. Whole peripheral blood samples (100 µl) were incubated for 20 min with the following mouse anti-human mAbs: CD3, CD14, CD16, CD19 and CD20 peridinin chlorophyll-cyanin 5·5 (PerCP-Cy5·5), HLA-DR allophycocyanin-cyanin 7 (APC-Cy7), CD11c APC, CD123 PE-Cy7, CD80 and CD86 FITC, CD83 PE (all from BD Biosciences; San Jose, CA, USA); and CD40 PE (Serotec, Oxford, UK). Staining conditions for each mAb was determined preliminarily in titration assays. Erythrocytes were lysed by incubation with ammonium chloride for 10 min. All operations were performed at 4°C in the dark. Cells were collected and analysed using a fluorescence activated cell sorter (FACS) FACSCanto2 flow cytometer (Becton Dickinson) and data analysis was performed using FACSDiva (Becton Dickinson) and FlowJo (Tree Star, Ashland, OR, USA) software. An acquisition gate was established based on forward-scatter (FSC) versus side-scatter (SSC) that included both lymphocytes and monocytes (mononuclear cells), but excluded most granulocytes and debris . A total of 150 000 events were collected to routinely visualize and gate on this population. Samples were acquired with a lyse/wash protocol, and estimates of the absolute numbers of mDCs and pDCs were calculated from the proportion of cells recorded by flow cytometry in the mononuclear gate, multiplied by absolute mononuclear cell count measured using a standard haemacytometer. Expression of the activation/maturation markers was analysed gated on mDCs and pDCs.
Cytoplasmic cytokine expression by PBDCs
The expression of a wide range of cytokines by mDCs and pDCs was determined by six-colour flow cytometry in whole blood samples, as described previously . Briefly, whole blood samples were incubated with or without 100 ng/ml lipopolysaccharide (LPS; Sigma Chemical Co., St Louis, MO, USA) or 10 µg/ml imiquimod (IQ; InvivoGen, San Diego, CA, USA) for 5 h, in the presence of the protein transport inhibitor brefeldin-A (BFA; 10 µg/ml; Sigma) added during the last 4 h to allow intracellular accumulation of cytokines. The concentrations of Toll-like receptor (TLR) ligands and the duration of cultures were established based on previous studies [18–23]. At the end of the culture, cells were stained with mAbs specific for lineage markers, HLA-DR, CD11c and CD123, used in all samples to gate on mDCs and pDCs, as described above. Samples were then fixed, permeabilized and stained with cytokine-directed mAbs using Leucoperm Reagent (AbD Serotec), according to the manufacturer's instructions. The following anti-human mAbs were used: interleukin (IL)-10 and IL-12 FITC (Caltag Laboratoires, Burlingame, CA, USA); interferon (IFN)-α FITC (PBL Biomedical, Piscataway, NJ, USA); IL-4 and IL-6 PE (BD Biosciences); and TNF-α PE (R&D Systems, Minneapolis, MN, USA). Staining conditions for each mAb were determined preliminarily in titration assays. All operations were performed at 4°C in the dark. Data acquisition and analysis was performed as described for immunophenotypic analysis of PBDCs.
Cytokine measurement in plasma samples
The amounts of plasmatic TNF-α and IL-6 were determined by enzyme-linked immunosorbent assay (ELISA) kits from R&D Systems; those of IFN-α and IL-10 were measured by ELISA kits from Bender MedSystem (Vienna, Austria). All individual steps were performed according to the manufacturer's instructions.
Incubation of PBMCs with plasma from pregnant and non-pregnant women and ActA
Peripheral blood mononuclear cells (PBMCs) from healthy donors were obtained by Ficoll density gradient centrifugation (Cedarlane; Hornby, Canada). Plasma or sera samples were obtained under sterile conditions from single donors, who were either women in the third trimester of normal pregnancy or non-pregnant controls. PBMCs were cultured in RPMI-1640 medium (Euroclone, Wetherby, UK) in the presence of 80% human plasma. The concentration of plasma was assessed in preliminary experiments. Recombinant human ActA (R&D Systems) was added at 100 ng/ml based on previous literature  and a dose–response curve. After 1 or 5 h of culture, as indicated, PBMCs were collected, stained and analysed by flow cytometry for immunophenotypic analysis of PBDCs as described above.
In vitro culture and flow cytometric analysis of moDCs
moDCs were obtained as described previously [24–26]. Briefly, PBMCs were isolated from standard buffy coat preparations of healthy donors. Monocytes were obtained from adherent PBMCs (> 90% pure CD14+ as assessed by cytometry) and cultured in complete medium: RPMI-1640 (Euroclone) supplemented with 10% heat-inactivated fetal calf serum (Gibco, Invitrogen Co., Carlsbad, CA, USA), in the presence of 800 U/ml recombinant human granulocyte–macrophage colony-stimulating factor (rhGM-CSF; Strathmann Biotech, Hannover, Germany) and 10 ng/ml rhIL-4 (R&D Systems) at 37°C in a CO2 (5%) incubator. At day 2, one-third volume of fresh complete medium supplemented with cytokines was exchanged. After 5 days, cell differentiation into immature DCs was assessed by cell morphology, acquisition of CD1a expression and down-regulation of CD14 expression. Cell viability was > 92% in all experiments, as evaluated by trypan blue exclusion. moDCs were stimulated further for 24 h with monocyte-conditioned medium (MCM) in the presence of 10% control serum, serum from pregnant women or control serum plus 100 ng/ml ActA. MCM was added 10% v/v, based on preliminary experiments. MCM was prepared as described previously  by incubating monocytes obtained by 2 h plastic adhesion with Staphylococcus aureus Cowan strain (dilution 1:10 000; Calbiochem, Darmstadt, Germany); after 24 h the cell-free supernatants were filtered and stored at −20°C. At the end of the culture period, phenotypic analysis was carried out by flow cytometry by using the following mAbs: CD80 and CD86 FITC, CD1a PE, CD83 and CD40 APC, HLA-DR APC-Cy7 (all from BD Biosciences). Staining conditions for each mAb was determined preliminarily in titration assays. 7-Amino-actinomycin D (7-AAD, Sigma) was used to exclude non-viable cells from the analysis.
Mixed lymphocyte reaction (MLR)
The allostimulatory activity of moDCs that had been stimulated for 24 h with MCM in the presence of control serum, serum from pregnant women or control serum plus ActA was analysed in MLR by the use of carboxyfluorescein succinimidyl ester (CFSE) methodology, as described previously [25,26]. Briefly, CFSE-labelled allogeneic monocyte-depleted PBMCs (3 × 105) were cultured with 1·5 × 104 moDCs in 200 µl complete culture medium in 96-well round-bottomed plates in triplicate. After 6 days, the cells were harvested and stained with anti-CD4 PE- and anti-CD3 APC-conjugated mAbs (BD Biosciences) for 20 min at 4°C. 7-AAD was used to exclude non-viable cells from the analysis. T cell proliferation was analysed by flow cytometry and expressed as percentage of proliferating CFSElow/CD4+/CD3+ viable T cells [25,26].
All statistical analyses assumed a two-sided significance level of P < 0·05. The Mann–Whitney U-test was used for comparisons between pregnant and non-pregnant women. The Wilcoxon signed-rank test was used to analyse the effects of in vitro incubation of control PBMCs and moDCs with plasma, sera or ActA, as well as the stimulation of whole blood samples with TLR ligands. Spearman's rank test was used to describe correlations. Data analyses were performed with Openstat3 software.
Changes in PBDC count during pregnancy
We examined the number of mDCs and pDCs in the peripheral blood of healthy pregnant and non-pregnant women using a six-colour flow cytometric method that we optimized recently to allow multi-parametric analysis of PBDC subsets . As shown in Table 1, the absolute number of mDCs was significantly higher in women during the first trimester of pregnancy than in non-pregnant women and decreased progressively in the following trimesters. Women in the third trimester showed similar mDC counts as controls. The number of pDCs decreased slightly during pregnancy progression, although not significantly. These reciprocal variations resulted in a higher mDC/pDC ratio in any trimester of pregnancy compared with non-pregnancy. Similar results were observed when the frequency of mDCs and pDCs in the mononuclear cell population rather than their absolute count were considered.
Table 1. Reciprocal variations in myeloid DCs (mDCs) and plasmacytoid DCs (pDCs) along healthy pregnancy.
Data shown as mean ± standard error of the mean from 24 non-pregnant and seven, 16 and 24 pregnant women in the first, second and third trimesters, respectively. P: significance of the difference compared with non-pregnant women; n.s.: not significant.
18·68 ± 2·09
8·39 ± 0·95
2·44 ± 0·27
23·98 ± 2·83
7·91 ± 1·57
3·49 ± 0·58
21·92 ± 3·46
5·95 ± 0·74
3·83 ± 0·45
16·23 ± 1·39
5·89 ± 0·56
3·13 ± 0·34
PBDCs are activated during pregnancy but do not up-regulate HLA-DR molecules
Because the state of activation of DCs is critical in determining whether antigen presentation by DCs may result in T cell activation or tolerance , we next examined the state of activation of PBDCs in pregnant compared with non-pregnant women. The expression of each activation marker was analysed gated on mDCs and pDCs compared directly in the same tube, in parallel samples where either CD40 and CD86, or CD80 and CD83 were considered. A representative flow cytometric analysis is shown in Fig. 1. As shown in Fig. 2a, mDCs and pDCs from pregnant women showed a higher expression of CD80, CD86, CD40 and CD83. The up-regulation of these molecules, expressed as mean fluorescence intensity (MFI), tended to increase progressively during pregnancy, being more pronounced during the third trimester than in early pregnancy. Accordingly, the frequency of activated mDCs and pDCs, defined as cells with levels of CD40 and/or CD86 outlining the gate of constitutive expression, that differed between subsets as described previously , was higher in women in the third trimester of pregnancy than in controls (P < 0·001 and P < 0·01, respectively; Fig. 2b).
Although PBDCs are all HLA-DR-positive by definition, their levels of HLA-DR expression may vary with the state of activation. Therefore, in this study we thought it could be relevant to investigate whether HLA-DR expression may change during pregnancy. Indeed, we observed that the progressive activation of PBDCs occurring during pregnancy was not paralleled by a concomitant up-regulation of HLA-DR molecules, whose expression on pDCs was significantly lower even in pregnant women in the third trimester than in non-pregnant controls (Fig. 2c). To gain more insight into this lower expression of HLA-DR molecules in healthy pregnancy, we compared the intensity of HLA-DR expression on activated or non-activated mDCs and pDCs in women in the third trimester of pregnancy and controls. As shown in Fig. 2d, HLA-DR MFI on non-activated cells did not differ between pregnant and non-pregnant women. In non-pregnant women HLA-DR MFI was higher on activated than non-activated mDCs and pDCs, as expected; however, in pregnant women the increase of HLA-DR MFI on activated cells was less pronounced than in controls. As a result, the intensity of HLA-DR expression on activated mDCs and pDCs was significantly lower in pregnant than non-pregnant women.
PBDCs from pregnant women express increased amounts of inflammatory and regulatory cytokines
Because of the central role played by cytokines of DC origin in directing immune responses, and because of the particular immune condition establishing during pregnancy, we further asked whether cytokine production by mDCs and pDCs of pregnant women differed from non-pregnant women. To this aim, we analysed the expression of a wide range of inflammatory and regulatory intracellular cytokines in unseparated mDCs and pDCs. In unstimulated conditions mDCs from pregnant women expressed increased amounts of the proinflammatory cytokines TNF-α, IL-6 and IL-12; according to observations on PBDCs activation, in the third trimester of pregnancy the levels of expression of all these cytokines were significantly higher compared with non-pregnant women (Fig. 3a). A concomitant mild, but significant, increased expression of the regulatory cytokine IL-10 and the T helper type 2 (Th2)-polarizing cytokine IL-4 was observed. To investigate whether the increase of IL-10 expression was able to counterbalance the increased expression of proinflammatory cytokines, we calculated the ratios between MFI values of the various cytokines and observed that the TNF-α : IL-10 MFI ratio was increased in pregnant women in the third trimester compared with controls (Fig. 3b), indicating a prevalent proinflammatory pattern of cytokine production. Because the ability of DCs to respond adequately to TLR ligands is of vital importance to the capacity of the immune system to mount effective defensive responses against pathogens , we further examined cytokine expression by mDCs upon stimulation of whole blood samples with the mDC stimulator LPS. In non-pregnant women LPS induced significant up-regulation of all the cytokines studied except IL-4, as expected (Fig. 3a). mDCs from pregnant women showed a similar pattern of cytokine production, demonstrating that the response to LPS is conserved completely in physiological pregnancy. Also, pDCs (Fig. 3c) from pregnant women showed a progressive increase in basal expression of proinflammatory cytokines. Because pDCs are the main producers of IFN-α, while they do not produce IL-12 , the expression of IFN-α instead of IL-12 was considered on these cells. As for mDCs, despite a concomitant mild increase of IL-10 expression a prevalent proinflammatory pattern of cytokine production was observed, characterized by higher TNF-α : IL-10 and IFN-α : IL-10 MFI ratios in women in late pregnancy than controls (Fig. 3d). Incubation of whole blood samples with the pDC stimulator IQ induced a huge increase in the expression of TNF-α and IFN-α, as expected. pDCs from pregnant women responded to IQ similarly to controls, still demonstrating that the response to TLR ligands is unaffected in healthy pregnancy. The expression of IL-6, IL-10 and IL-4 by pDCs upon IQ stimulation was not assessed, based on a previous demonstration that in our experimental model IQ does not induce pDC production of cytokines other than TNF-α and IFN-α.
Plasma from pregnant women show a proinflammatory cytokine pattern
Because circulating cytokines that can derive from both immune and non-immune cells are potentially able to affect PBDCs, we examined the plasmatic levels of some of the cytokines that are known to be involved mainly in pregnancy, to investigate possible correlations with mDC and pDC features. Cytokine concentrations are reported in Table 2. According to Curry et al. , we observed increased levels of plasmatic IL-6 with pregnancy progression that were significantly higher than controls in the third trimester of pregnancy (P < 0·05). The frequency of subjects with detectable IL-6 also increased (from 8% to 44%). Conversely, the levels of plasmatic IL-10 showed an increase in the first trimester but decreased with pregnancy progression, reaching non-significant (n.s.) lower levels than controls in the third trimester (P = n.s.). The levels of TNF-α and IFN-α did not differ between pregnant and non-pregnant women. No correlation between single plasmatic cytokines and mDC or pDC parameters was observed.
Table 2. Cytokine measurement in plasma sample.
Data shown as mean ± standard error of the mean from 21 non-pregnant and six, 13 and 23 pregnant women in the first, second and third trimesters, respectively. P: significance of the difference compared with non-pregnant women. IFN, interferon; IL, interleukin; TNF, tumour necrosis factor; n.s., not significant.
0·56 ± 0·06
1·38 ± 1·13
14·78 ± 6·74
0·89 ± 0·46
1·18 ± 0·39
1·04 ± 0·16
25·53 ± 11·43
1·64 ± 0·61
0·59 ± 0·07
1·11 ± 0·51
16·80 ± 4·20
2·96 ± 1·31
0·68 ± 0·18
6·51 ± 2·37
5·46 ± 1·06
2·10 ± 1·39
Plasma from pregnant women and ActA inhibit the up-regulation of HLA-DR molecules during activation of control PBDCs
To ascertain whether the incomplete activation of mDCs and pDCs characterized by high co-stimulatory molecules but low HLA-DR expression was dependent, at least in part, on the effects of soluble factors present in the plasma of pregnant women, we next incubated control PBMCs with plasma from pregnant compared with non-pregnant women. Preliminary kinetic experiments revealed that upon PBMC culture in plastic tubes mDCs underwent progressive activation that was clearly evident after 5 h (a representative experiment is shown in Fig. 4a). As shown in the same figure, up-regulation of all the activation markers was observed similarly in the presence or absence of control plasma, indicating that mDC activation was induced by culture conditions per se rather than by plasmatic factors. pDCs were far less sensitive to culture-induced activation. The addition of plasma from pregnant or non-pregnant women to the culture did not affect the expression of CD80, CD86, CD40 and CD83 differently (Figs 4a and b), suggesting that causes other than plasmatic factors are required to induce the activation of mDCs and pDCs occurring in healthy pregnancy. On the contrary, plasma from pregnant or non-pregnant women produced different effects on mDC and pDC expression of HLA-DR molecules. In particular, the addition of plasma from pregnant, but not from non-pregnant, women were able to inhibit significantly the culture-induced up-regulation of HLA-DR on both PBDC subsets. Notably, the effects of plasma from pregnant women were not evident after 1 h incubation (Fig. 4a), thus ruling out the possibility that the reduced HLA-DR expression may rely on antigen masking by plasmatic anti-HLA antibodies or other soluble factors. Because the glycoprotein hormone ActA, which increases progressively in the plasma during pregnancy, has been demonstrated similarly to inhibit MHC class II up-regulation on human moDCs , we wondered whether ActA also affected HLA-DR expression on mDCs and pDCs in our experimental model. As shown in Fig. 5a, the incubation of control PBMCs with ActA added to plasma from non-pregnant women induced a dose-dependent inhibition of HLA-DR expression on mDCs and pDCs, with the maximal effect observed at 100 ng/ml. This concentration, which was the same as used in the above study  and is relevant to the levels measured in the serum of pregnant women in the third trimester , was used for subsequent experiments. Similarly to plasma from pregnant women, ActA did not affect the expression of CD80, CD86, CD40 and CD83, while it inhibited significantly the culture-induced up-regulation of HLA-DR on both PBDC subsets (Fig. 5b), thus suggesting a contribution of this glycoprotein hormone to the state of incomplete activation of mDCs and pDCs occurring during healthy human pregnancy.
The inhibition of HLA-DR up-regulation promoted by sera from pregnant women and ActA is paralleled by lower allostimulatory activity
To investigate the potential relevance of incomplete DC activation to the maintenance of fetal tolerance we next analysed the effects of sera from pregnant women and ActA on the allostimulatory activity of DCs. To this aim we used moDCs. However, although the analysis of PBDCs directly in whole blood samples has the key advantage of providing results that directly reflect the in vivo situation, it has some limitations in the study of particular DC functions. To obtain a cytokine-induced activation, we stimulated moDCs with low-dose MCM, a commonly used DC stimulator containing a cocktail of proinflammatory cytokines produced by activated monocytes . MCM stimulation was performed in the presence of sera from non-pregnant women compared with sera from pregnant women or sera from non-pregnant women plus 100 ng/ml ActA. Similar to the effects observed on PBDCs, sera from pregnant women as well as ActA did not affect the expression of CD80, CD86, CD40 and CD83 on moDCs, but induced levels of MCM-stimulated HLA-DR expression significantly lower compared with control sera (Fig. 6a). To investigate the effects of sera from pregnant and non-pregnant women as well as ActA on the allostimulatory activity of moDCs, we evaluated the direct pathway of alloantigen presentation. We performed direct MLR by co-culturing moDCs with allogeneic monocyte-depleted CFSE-labelled PBMCs, monitored by CFSE dilution and expressed as a percentage of divided CFSElow/CD4+/CD3+ viable T cells. As shown in Fig. 6b, moDCs stimulated with MCM in the presence of pregnant sera as well as non-pregnant sera plus ActA induced the proliferation of a lower proportion of allogeneic T cells than moDCs stimulated with MCM in the presence of non-pregnant sera, indicating that the inhibition of HLA-DR up-regulation promoted by sera from pregnant women and ActA is paralleled by lower allostimulatory activity.
In this study we demonstrated that during human physiological pregnancy mDCs and pDCs undergo profound changes that probably reflect the maternal systemic reaction to the fetus and that may be relevant to the maintenance of fetal tolerance and the favourable outcome of healthy pregnancy.
The mild variations that we observed in the number of circulating PBDCs during pregnancy progression resulted in an increased mDC/pDC ratio that is consistent with the predominance of mDCs described in previous studies on PBDCs  and decidual DCs [9,30]. As observed for other leucocyte populations, it is likely that these variations in PBDC counts may reflect reciprocal changes in the levels of hormones, growth factors and cytokines that, in turn, may impact upon the generation, mobilization and peripheral localization of PBDCs. The demonstration that oestrogens promote the GM-CSF-mediated differentiation of myeloid DCs [31,32] clearly supports this hypothesis.
The most remarkable feature of PBDCs that we found in our study was their state of incomplete activation increasing with healthy pregnancy progression and characterized by up-regulation of co-stimulatory molecules and maturation markers without a concomitant up-regulation of HLA-DR molecules. The finding of DC activation per se was not surprising, as it may be reasonably sustained by the mild systemic inflammatory reaction that is known to occur in late normal pregnancy . The lack of concomitant HLA-DR up-regulation is an intriguing aspect that may be of crucial importance to the maintenance of fetus tolerance. During conventional DC activation by danger signals deriving from pathogens or endogenous cell damage, the expression of HLA-DR molecules on the DC surface is usually up-regulated together with the expression of co-stimulatory molecules and the production of cytokines, aimed to increase the number of MHC-peptide complexes that may be presented to T lymphocytes and provide full stimulatory signals . Keeping a low HLA-DR expression may therefore represent a mechanism by which DCs taper antigen presentation despite their otherwise activatory attitude.
By use of the six-colour multi-parametric approach to the study of PBDCs we could demonstrate that the lower expression of HLA-DR molecules on mDCs and pDCs during late healthy pregnancy is restricted to activated DCs, suggesting that the constitutive expression of HLA-DR molecules is unaffected while its up-regulation is inhibited partially during mDC and pDC activation. This conclusion was supported further by kinetic experiments demonstrating that incubation of control PBMCs and moDCs with plasma or sera from pregnant women does not reduce the expression of HLA-DR molecules on the PBDC surface, but rather inhibits partially the HLA-DR up-regulation induced by culture in plastic tubes or MCM stimulation. Taken together, these results indicate clearly that the lower levels of HLA-DR molecules detected on PBDCs during pregnancy cannot be ascribed to HLA-DR antigen masking by anti-HLA antibodies or other serum factors, as suggested in a previous study that also reported reduced HLA-DR expression on PBDCs in healthy pregnancy, but was accompanied by concomitant reduction of the co-stimulatory molecules CD80 and CD86 . So far, to our knowledge, no other studies have investigated PBDC activation during human healthy pregnancy.
The incomplete activation of PBDCs during healthy pregnancy may reveal a further mechanism operated by the immune system to maintain tolerance despite DC activation. Peripheral DCs are known to be central to the establishment and maintenance of peripheral tolerance [3–6] and, in this respect, a reduced expression of HLA-DR molecules on DC surface may affect T cell activation by reducing the strength of the specific signal provided to T cells. Our evidence on moDCs demonstrating that the inhibition of HLA-DR up-regulation promoted by sera from pregnant women is paralleled by lower allostimulatory activity seems to support the in vivo relevance of incomplete DC activation to the maintenance of fetal tolerance. Accordingly, serum from pregnant mouse has been reported to down-regulate MHC class II molecules on cultured DCs and this, in turn, to condition a reduction in their allostimulatory capacity in vitro. Moreover, it has been demonstrated recently that peripheral DCs recirculate to the thymus, thus contributing to the induction of acquired thymic tolerance [35–37]. In this respect, the possibility that during normal pregnancy antigen transport to the thymus by incompletely activated PBDCs may promote thymic tolerance to fetal antigens represents a fascinating hypothesis that may merit further investigatory studies.
Several factors may contribute to promote the incomplete activation of mDCs and pDCs during normal pregnancy progression. Among them, we focused on the glycoprotein hormone ActA because it plays a pivotal role in the endocrine physiology of human pregnancy by promoting the production and release of other key hormones such as progesterone and human chorionic gonadotrophin (hCG) [15,38]. The serum levels of ActA, produced mainly by the placenta, increase progressively throughout pregnancy and they are higher in the third trimester of pregnancy [28,39,40], just when the changes that we observed in PBDCs are more marked. Because it belongs to the TGF-beta family, ActA has been demonstrated recently to prevent the up-regulation of HLA-DR molecules during cytokine-induced moDC maturation conditioning a lower allostimulatory capacity of DCs . For all these reasons, we wondered whether ActA could be responsible, at least in part, for the inhibition of HLA-DR expression on PBDCs and found that this was indeed the case. Moreover, the prevention of HLA-DR up-regulation was paralleled by a lower allostimulatory function of moDCs, as expected . hCG and inhibin, which share many functional activities with ActA, [11,17,41], may also contribute to prevent the up-regulation of HLA-DR molecules on mDCs and pDCs during normal pregnancy progression. Because ActA did not affect the expression of co-stimulatory molecules, mediators other than ActA should be considered to be responsible for DC activation. Proinflammatory cytokines such as IL-6 and TNF-α, placental microparticles, complement factors and thrombin are all examples of well-known circulating factors able to induce the expression of DC co-stimulatory molecules that show a definite increase during healthy pregnancy [27,31,42–45]. Indeed, in our experiments plasma from pregnant women did not up-regulate co-stimulatory molecules on mDCs and pDCs to a higher extent than control plasma. To explain this finding, it is possible that soluble factors may require a longer incubation time to induce PBDC activation, or that non-soluble factors may also contribute to PBDC activation. In vivo, it is likely that the combination of all the above stimulatory factors with other factors circulating at higher concentrations in pregnancy but endowed with immunoregulatory functions, such as placental growth factor, IL-10, soluble HLA-G and glucocorticoid hormones [46–50], together control the quality of DC maturation, which is of critical importance to the activation and regulation of adaptive immune responses.
Taken together, our results indicate clearly that during the third trimester of healthy human pregnancy both mDCs and pDCs undergo a peculiar pattern of incomplete activation, characterized by increased expression of co-stimulatory molecules and production of cytokines but lacking HLA-DR up-regulation. These findings may provide new insight into the comprehension of the immune mechanisms operated by the maternal immune system to maintain fetal tolerance.
This work was supported by grants from MIUR (PRIN 2006065944-002) to S.D.B; Fondazione Cariplo (1055/104878-2005) to M.L.V and I.C.; Università degli Studi di Milano (FIRST 2007) to S.D.B. S.G and M.C. were supported by a fellowship from the Doctorate School of Molecular Medicine, University of Milan, 20090 Milan, Italy. The authors thank Dr Vincenzo Toschi and the staff of the Transfusional Centre of Ospedale S. Carlo Borromeo of Milan for kindly supplying buffy coats used to obtain moDCs.