Dendritic cells: a family portrait at mid-gestation


E. A. Bonney, Department of Obstetrics, Gynecology and Reproductive Sciences, University of Vermont College of Medicine, Given Building Rm C-244, 89 Beaumont Avenue, Burlington, VT, USA. Email:
Senior author: Elizabeth A. Bonney


Recent advances in our understanding of dendritic cells (DCs) and their role in tolerance and immunity has fuelled study of their normal development and function within the reproductive tract. The common hypothesis that pregnancy is a state of immune suppression or deviation now includes the idea that alterations in DC phenotype and function are critical for maternal tolerance. We chose to study DCs in the uterus and lymphoid tissue in non-pregnant and pregnant mice at mid-gestation to understand what DC-related factors may be involved in premature birth. We used a mouse model where the mother’s immune system has been shown to respond to the male antigen H-Y. Observed differences among DCs in the uterus, uterine draining nodes and spleen, even in non-pregnant mice, suggest the existence of a specialized uterus-specific subset of DCs. We further found that, amongst CD45+ CD11c+ cells in the uterus and peripheral lymphoid tissue of pregnant mice, expression of major histocompatibility complex class II (MHC II) and costimulatory molecules (i.e. CD80) was similar to that in the non-pregnant state. Moreover, there was no pregnancy-related decrease in the proportion of CD11c+ cells in the uterus or in the uterine node that were CD11b CD8+. Pregnancy increased the CD11b+ subsets and the expression of chemokine (C-C motif) ligand 6 (CCL6) in DCs of the uterine draining nodes. Finally, DC subsets showed variable expression, with respect to tissue and pregnancy, of the cytokine interleukin-15, which is important in lymphoid cell homeostasis. For DCs, pregnancy is not a state of immune paralysis, but of dynamic developmental change.


The reproductive success of mammals, in which the maternal immune system is in direct contact with the semi-allogeneic fetus, depends on the ability to remain tolerant to the fetus as well as survey it for infection. The balance between active immunity and tolerance in the feto-maternal interface results in successful pregnancy.

There is evidence that during pregnancy the maternal immune system recognizes fetal antigen and the idea that the maternal immune system is suppressed has been challenged by our group1 and also by others.2 These studies demonstrate that the mother is not intrinsically tolerant to paternal antigens, and the maternal immune system can respond well to fetal antigen. Importantly, the maternal immune system can reject tissues bearing fetal antigens, such as skin grafts, while remaining tolerant to the fetus.3

There are, however, several proposed mechanisms of feto-maternal tolerance. Expression of indoleamine-2,3-dioxygenase at the feto-maternal interface,4 the presence of regulatory T cells,5 expression of non-classical major histocompatibility complex (MHC) on trophoblast, and placenta-driven T-cell apoptosis6 are among a long list of examples.

Dendritic cells (DCs) originate from bone marrow cells and belong to a sparsely distributed migratory and non-migratory group of leucocytes that is specialized for uptake, transport, processing and presentation of antigen and are present in nearly all tissues including surface epithelia, T cell-rich zones of lymphoid organs and mucosa. Encounter with antigen or activating stimulus causes these cells to initially up-regulate their ability to process antigen, move from the tissue to the draining lymph node, or within the lymph node, and simultaneously initiate a maturation programme that includes shutting down further antigen processing and completion of peptide loading into surface MHC molecules, up-regulation of costimulatory molecules,7,8 and elaboration of specific cytokines.9 The presence of DCs in the maternal decidua10,11 points to an important biological role for antigen-presenting cells in maternal–fetal interactions.

It is now held that the DC population in a given tissue may consist of three or more specialized DC types, such as those occurring at mucosal surfaces (reviewed by Shortman12), conventional migratory DCs that traffic to the draining lymph node and present antigen to T cells, lymphoid tissue resident DCs that spend their life within a tissue (i.e. in the spleen and thymus), the immediate precursors of these types, and finally DCs that arise, perhaps independently of other types, upon inflammatory or like stimuli.

Preterm delivery is a leading cause of neonatal morbidity and mortality, and there exists a growing body of evidence correlating preterm birth with inflammation and/or infection in the uterus. The critical role of DCs in preterm labour and delivery is not completely clear. There are several animal models that have led to a better understanding of the clinical problem of preterm birth. One example, meant to replicate the effects of maternal infection with Gram-negative bacteria, utilizes injection of mice with lipopolysaccharide (LPS) on day 15 of gestation.13 Such injection results in expulsion of fetuses within 24 hr. We are interested in the cellular networks involved in this phenomenon, and hope to delineate the role of DCs in this response, as has been done for other models of LPS exposure (reviewed in8). Before beginning a study of DCs in abnormal gestation, however, we first sought to examine the DC population in normal mice at day 15 of pregnancy. We used a mouse model1 where the mother’s immune system has been shown to respond to the male antigen H-Y, which is present on all male pups.

Materials and methods


All animals were treated in accordance with the Guide for Care and Use of Laboratory Animals and with the approval of the Institutional Animal Care and Use Committee of the University of Vermont. Eight- to ten-week-old C57BL/6 mice were purchased from Jackson Laboratory, Inc. (Bar Harbor, ME). The animals were allowed free access to food and water at all times and were maintained on a 12-hr light/dark cycle. C57BL/6 females were mated to same-strain males. Timed mating procedures include 3 days of seasoning before 24 hr of mating. The day on which a seminal plug was noted in mated females was considered to be day 0 (of 19) of gestation. On day 15 of pregnancy, C57BL/6 and age-matched never-mated female mice were killed. The uterus, spleen and draining lymph nodes (inguinal and para-aortic) were harvested.

Single-cell isolation

Decidual cells were isolated using a modification of a protocol previously reported by Kämmerer et al.10 Briefly, specimens were dissected free of placenta and fetus and washed in phosphate-buffered saline (PBS; Mediatech, Herndon, VA). The uterine tissue was minced into small fragments and digested for 20 min at 37° under slight agitation in PBS with 200 U/ml hyaluronidase (Sigma-Aldrich, St Louis, MO), 1 mg/ml Liberase Blendzyme 3 (Roche Applied Science, Indianapolis, IN), 0·2 mg/ml DNase I (Sigma-Aldrich) and 1 mg/ml bovine serum albumin/fraction V (Sigma-Aldrich). The cell suspension obtained was filtered through 40-μm nylon mesh (BD Falcon, Franklin Lakes, NJ) and washed twice in PBS. Spleen and lymph node single-cell suspensions were also prepared using nylon mesh. The surface expression of DC-specific molecules was determined by six-colour flow cytometry using an LSR II (BD Biosciences, San Jose, CA).

Antibodies to phenotype DC subpopulations

Blocking of non-specific binding was performed by preincubation of cell suspensions with anti-CD16/32 (Fcγ RIII/II receptor antibody, clone 2.4G2, 1 : 50). For DC phenotyping, the following antibodies were used: anti-CD11c allophycocyanin (APC; 1 : 100), anti-CD45 PerCP-Cy5.5 (1 : 100), anti-CD80 phycoerythrin (PE; 1 : 50), anti-I-A/I-E PE (clone M5/114, 1 : 400), anti-CD8a fluorescein isothiocyanate (FITC; 1 : 400), CD11b phycoerythrin Texas red (1 : 100) and CD4 APCcy7 (1 : 100). All antibodies were from BD Biosciences Pharmingen (San Diego, CA). Cells were incubated with antibodies for 30 min on ice in the dark.

Interleukin (IL)-15 intracellular staining

Uterus and lymph node single-cell suspensions were incubated with 2.4G2 (see above) in 10% goat serum (Invitrogen, Carlsbad, CA) to block non-specific binding. Cells were stained with the above antibodies to DC surface markers and fixed in 1% paraformaldeyde (Electron Microscopy Sciences, Hatfield, PA) with 0·1% bovine serum albumin (Sigma-Aldrich) in PBS. Cells were permeabilized with BD Perm/wash buffer (BD Biosciences) according to the manufacturer’s instructions and stained with biotinylated anti-IL-15 (1 μg/106 cells; R&D Systems, Minneapolis, MN). After washing in BD Perm/wash buffer, cells were incubated with streptavidin phycoerythrin Texas red (1 : 200; BD Biosciences Pharmingen). Cells from a C57Bl/6 mouse, which was injected with 30 μg of LPS intraperitoneally and killed 24 hr later, were used as a positive control for the expression of IL-15, MHC II and CD80 molecules.

Sorting of uterine DC populations

Single-cell suspensions obtained from the uterus as above were incubated antibodies against the DC surface molecules CD45, CD11c, and MHC II into relevant populations by fluorescence-activated cell sorting (FACS; BD FACS ARIA, San Jose, CA).


Sorted cells were collected in Roswell Park Memorial Institute Medium (RPMI; Invitrogen) that was supplemented with 2% fetal bovine serum (FBS; Invitrogen) and imaged with a Zeiss LSM 510 META confocal laser scanning microscope (Zeiss MicroImaging, Thornwood, NY) operating in the multitrack mode to avoid fluorescence bleed-through. The PerCP-Cy5·5 signal was excited with a 633-nm helium neon ion laser, the PE with a 488-nm argon ion laser, and the APC with a 633-nm helium neon ion laser. Images were obtained with a Plan Zeiss 63× (1·4 NA) objective lens and subsequently captured using lsm 510 meta software version 4·2 (Zeiss).

Polymerase chain reaction (PCR) array analysis of DCs

Single-cell suspensions obtained from the draining lymph nodes from pregnant and never-mated mice were incubated with the following antibodies to DC surface molecules: CD45, CD11c and MHC II CD11c+ MHC II double-positive cells were isolated by FACS (BD FACS ARIA). Sorted cells were collected into RNA lysis buffer (RNA isolation kit; Qiagen, Valencia, CA). Total RNA was extracted according to the manufacturer’s instructions. Single-stranded cDNA was synthesized from total RNA using random primers, and was probed using a mouse inflammatory cytokine and receptor PCR array kit (RT2 Profiler™ PCR Array: PAMM-011 SuperArray; Bioscience Corporation, Frederick, MD) used for real-time polymerase chain reaction (RT-PCR) on complementary DNA. The full list of genes included in the array can be found at

Statistical analysis

Data obtained by flow cytometry, including percentages and mean fluorescence intensity, were analysed using graphpad prism (Graphpad Software, Inc., San Diego, CA). A t-test with Welch’s correction was used to compare values obtained from pregnant and never-mated mice. Statistical significance was set at P < 0·05.


The uteri of both pregnant and non-pregnant mice contain DCs of a ‘mature’, i.e. CD45+ CD11c+ MHC II+ CD80+, phenotype, but pregnancy induces subtle changes

Before beginning an analysis of DCs in the pregnant as compared with the never-mated uterus, we compared the relative abundance of the lymphocyte population in this tissue. The influences of hormonal changes, increased blood flow,14,15 and the developing fetuses and placentae probably alter fluid content, support hypertrophy, and increase cellularity. Such changes may be reflected in many cellular compartments, but also in the lymphoid cell compartment, which includes natural killer (NK) cells16 and macrophages,17 which may interact with DCs. To determine the proportion of lymphocytes, uteri from day 15 pregnant and never-mated mice were isolated, and single-cell suspensions were prepared. The resulting cells were incubated with a panel of antibodies to molecules expressed on the surface of dendritic and other lymphoid cells and then analysed by flow cytometry. In analysis of forward and side scatter to determine the population of cells similar in size to lymphocytes (‘lymphocyte gate’), we did not observe qualitative differences in the lymphoid population. Additionally, in 13 pregnant and nine never-mated mice examined over three separate experiments, we observed no significant difference in the proportion of CD45+ cells (data not shown).

Further analysis, an example of which is seen in Fig. 1(a), of uterine cells from pregnant and non-pregnant mice showed that the proportion of CD45+ cells that were CD11c+ MHC II+ was not significantly different (pregnant 15·2 ± 3·6%, n = 13; never-mated 13·6 ± 2·8%, n = 9, P > 0·05). Finally, the expression of MHC II and the costimulatory molecule CD80, by mean fluorescence intensity of staining with specific antibodies, was observed to be similar (data not shown).

Figure 1.

 (a) CD45+ and CD45+ CD11c+ MHC II+ cells in the uterus. Uteri from mice on day 15 of gestation and from never-mated controls were isolated for preparation of single-cell suspensions that were examined by flow cytometry. Cells falling in the lymphocyte gate were then gated on CD45 (upper plots). Those cells that were positive for CD45 were examined for their expression of CD11c and major histocompatibility complex (MHC) II (lower plots). (b–d) Dendritic cell (DC) subsets in the uterus. Further study of CD45+ CD11c+ MHC II+ cells from the uterus was performed by examination of expression of CD11b, CD8 and CD4. (b, c) Delineation of CD11bhi CD8 and CD11blo/− CD8+ populations. (b) Example. (c) Proportion in pregnant (black symbols) and non-pregnant (grey) mice. Each symbol represents one uterus. The bar denotes the median. (d, e) Delineation of CD11b+ CD4 and CD11b+ CD4+ cells. (d) Example. (e) Proportion in pregnant (black symbols) and non-pregnant (grey) mice. Each symbol represents one uterus. The bar denotes the median.

Thus, for conventional markers of maturity, DCs in the uterus do not appear to be substantially different from those in non-pregnant mice. At least two possibilities may explain these findings. One is that there is another time-point in gestation when the maturational status of uterine DCs is significantly different from that found in the non-pregnant state,18 but that these differences disappear by day 15. However, it is formally possible that other more subtle changes exist within the ‘mature’ uterine DC pool during mid/late gestation.

Pregnancy increases the proportion of ‘myeloid’ DCs in the uterus

To further analyse uterine CD11c+ MHC II+ double-positive DCs, we explored the surface expression of other molecules. We chose CD11b and CD8 because they have been used to delineate subsets of ‘conventional’ DCs, albeit in lymphoid tissues: CD11bhi CD8, ‘myeloid’ DCs dependent on the transcription factor interferon regulatory factor (IRF)−4 and thought functionally to be supportive of T helper type 2 (Th2) responses, and CD11bint/lo CD8+, dependent on the transcription factor IRF-8, and thought to be supportive of Th1 responses,19 although with a slightly reduced ability to stimulate a mixed lymphocyte reaction.20 Uterine cell suspensions from day 15 pregnant and never-mated mice were prepared and incubated with antibodies to CD45, CD11c, MHC II, CD11b and CD8 and examined by flow cytometry as shown in Fig. 1(b). We observed that, as a proportion of the DCs present, those that were CD11bhi CD8 comprised a larger share in uteri of pregnant as compared with non-pregnant mice. This observation was born out in examination of several uteri as seen in Fig. 1(c) (left graph). This finding is consistent with previous results in more limited studies.18,21

By comparison, there was not a significant difference in the proportion of mature DCs that were CD11bint/lo CD8+ (Fig. 1b,c; right graph). This result is consistent with earlier suggestions22 that the maternal–fetal interface is supportive of Th2 responses. However, it is likely that there exist more levels of complexity in the immune regulation of the maternal–fetal interface.

In pregnancy, there is an increase in the proportion of CD11b+CD4 cells in the uterus

Studies of DC development suggest that the CD4 molecule delineates a potentially separate lineage or subset of DCs20 within lymphoid organs such as the spleen. Other studies have suggested that bone marrow-derived DCs can develop into lineages expressing or not expressing CD4,19 dependent on specific regulatory or transcription factors. In addition, it is suggested that CD4 is down-regulated upon activation of DCs.23 CD4 is, however, expressed in a population of highly proliferative immature human early decidual DCs, which were observed to have intimate contact with decidual NK cells.11 Because of the possible role that CD4 expression might play in phenotypic and functional differences in the uterus, we studied DCs in uteri of day 15 pregnant and non-pregnant mice as shown in Fig. 1(d). A striking result was the finding of an elevated proportion of CD4+ DCs in the uteri of non-pregnant but not pregnant mice.

Furthermore, analysis of the proportion of CD11b+ CD4 DCs in uteri from pregnant and never-mated mice (Fig. 1e, left) revealed an increased proportion in pregnant uteri (pregnant 80·1 ± 2·6%, n = 13; never-mated 63·8 ± 4·9, n = 9, P < 0·05). Summary data (Fig. 1e, lower right) confirmed that the proportion of CD11b+CD4+ cells was decreased in pregnancy. Taken together with our data on CD8 expression (Fig. 1b,c), these findings suggest that pregnancy increases the proportion of CD11b+ CD8CD4 DCs present in the uterus.

Brief examination of the morphology of CD45+ cells obtained from the uterus

It is obvious that several types of cells exist in the pregnant and non-pregnant uterus. It is also likely that several types of DCs (tissue-specific, migratory, precursor, etc.) at differing maturational stages exist within this anatomical site. A classical method of determining lineage or maturational state is to examine cellular morphology. We thus generated single-cell suspensions from the uteri of pregnant mice, incubated them with antibodies to CD45, CD11c and MHC II, and sorted CD45+ cells into four populations based on the other markers by flow cytometry. The flow-sorted samples were placed onto microscope slides containing a drop of tissue culture medium, were cover-slipped, and within 10 min were examined by confocal microscopy. The four subpopulations were derived from the four quadrants of the cytometry plot shown in Fig. 2: (a) CD11c+ MHC II, (b) CD11c+ MHC II+, (c) CD11c MHC II, and (d) CD11c MHC II+. The cells sorted from the double-positive gate (Fig. 2b) were large and highly positive for MHC II, and some, but not all, had long dendritic extensions consistent with the description of conventional activated migratory DCs. In contrast, cells that were CD11c+, yet MHC II (Fig. 2a), were smaller and had fewer and shorter dendritic extensions. Those cells that were singly positive for MHC (Fig. 2d) were rounder. Double-negative cells (Fig. 2c) were elongated and large, but different in morphology from the other groups of cells. These last two types of cells could be comprised of a number of cells including alternative DC populations (for example ‘inflammatory DCs’ or precursor DCs,12 macrophages, and B cells). Thus, several possible subsets or lineages of DCs potentially exist within the uterus at mid gestation.

Figure 2.

 Populations of CD45+ lymphocytes in the uterus sorted by surface expression of CD11c and major histocompatibility complex (MHC) II. CD45+ lymphocytes from the uterus were flow sorted into: (a, above left) CD11c+ MHC II; (b, above right) CD11c+ MHC II+; (c, bottom left) double negative, and (d, bottom right) CD11c MHC II+ subpopulations, placed in culture medium and photographed using a confocal microscope to assess differences in morphology. Blue, CD45; green, CD11c; red, MHC II; cells positive for all markers appear white.

The lymph nodes draining the uterus in pregnant and non-pregnant mice contain DCs with higher levels of MHC II

It is now thought that, in addition to activated migratory DCs, antigen flows from tissues to their draining nodes.12 Thus at least two types of maternal DCs might be present in lymph nodes draining the uterus: those that have experienced fetal antigen in the uterus, and those that are resident in the draining lymph node and so are exposed to soluble fetal antigen that arrives in the afferent lymphatics. After observing pregnancy-regulated heterogeneity of DCs in the uterus, we next examined the draining nodes. Single-cell suspensions of para-aortic and inguinal nodes were incubated with the same antibodies used to analyse uterine DCs by flow cytometry. Most lymphocyte-sized cells from the uterine draining lymph nodes are CD45+, and relatively few of these are CD11c+ and MHC II+, i.e. cells of the mature DC phenotype. As seen in the uterus, neither the proportion of CD45+ cells nor the subset of these cells that is CD45+ CD11c+ MHC II+ is significantly different on day 15 of gestation as compared with the non-pregnant state.

Amongst mature DCs, there may be different levels of activation related to the stimuli (i.e. infection and disregulation) present in the draining node and in the related tissue.24 Thus, we examined the surface expression level of MHC II in draining node DCs. When the entire population of CD11c+ CD45+ cells was considered (Fig. 3a,b; top), the mean expression of MHC II was observed to be increased on day 15 of gestation as compared with the non-pregnant state (pregnant 6961 ± 419·1, n = 13; non-pregnant 3313 ± 451·1, n = 15, P < 0·05). As it is formally possible that cells expressing high levels of MHC II, yet low levels of costimulatory protein, could generate tolerance in naïve T cells, we also examined surface expression of the costimulatory molecule CD80 (Fig. 3a, bottom). The data shown in Fig. 3(b) (bottom) suggest that the level of surface CD80 is similar, if not slightly increased, on CD11c+ cells in the uterine draining nodes on day 15 of gestation (pregnant 1064 ± 84·54, n = 10; never-mated 814·0 ± 108·2, n = 8, P = 0·08). Thus, the proportion of mature DCs in the draining nodes of the uterus is similar.

Figure 3.

 Expression of major histocompatibility complex (MHC) II and CD80 in uterine draining node dendritic cells (DCs). Mice on day 15 of gestation and never-mated controls were killed and the nodes draining the uterus were used to prepare single-cell suspensions that were examined by flow cytometry. CD45+ lymphocytes were gated for CD11c and then examined by comparison of mean fluorescence intensity (MFI). (a) Example of the flow cytometric analysis. Left plots: expression of CD45 and CD11c. Right plots: expression of MHC II (upper) and CD80 (lower). Solid line histogram, pregnant mice; filled histogram, never-mated mice. (b) Cumulative data. The y-axis shows the mean fluorescence intensity indicative of MHC II (upper graph) or CD80 (lower graph) expression. Each symbol represents one uterus. The bar denotes the median.

Pregnancy induces subtle changes in the population of DCs present in the uterine draining nodes

‘Myeloid’ DC predominance in uterine draining lymph nodes on day 15 of pregnancy

Because of possible migration of cells from the uterus to the relevant draining nodes, we expected to observe pregnancy-related changes in the DC population of the draining lymph nodes. CD11c+ MHC II+ cells in the uterine draining nodes can be separated into CD11bhi CD8 and CD11bint/lo CD8+ populations (Fig. 4a, top), as in the uterus (Fig. 1b,c). While late pregnancy supported a larger proportion of CD11bhi CD8 cells (Fig. 4a, bottom; pregnant 35·8 ± 2·4, n = 13; never-mated 28·8 ± 1·2, n = 15, P < 0·05), there was not a significant increase in the proportion of CD45+ CD11cMHC II+ cells that were CD11bint/lo CD8+ (pregnant 6·5 ± 0·5, n = 13; never-mated 7·2 ± 0·6, n = 15, P > 0·05).

Figure 4.

 Dendritic cell (DC) subsets in the uterine draining nodes. Further study of CD45+ CD11c+ MHC II+ (see Fig. 1 for an example) cells from this tissue included examination of expression of CD11b, CD8 and CD4. (a) Delineation of CD11bhi CD8 and CD11blo/− CD8+ populations. Top: example. Bottom: proportion in pregnant (black symbols) and non-pregnant (grey) mice. (b) Delineation of CD11b+ CD4, CD11b+ CD4+ and CD11b CD4+ cells. Top: example. Bottom: proportion in pregnant (black symbols) and non-pregnant (grey) mice.

In pregnancy, there is an increase in the proportion of CD11b+ CD4 cells in the uterine draining nodes

The lymph nodes draining the uterus contain CD11b+ cells that can also be subdivided according to their expression of CD4 (Fig. 4b, top). In the nodes draining the day 15 uterus, there is a population of CD11b+ cells that appears to be partially or completely down-regulated in surface expression of CD4. The CD4-regulating population appears to be present in the nodes of the non-pregnant uterus, but has not yet down-regulated CD4. When the proportion of CD11b+ DCs that have no CD4 on the surface is analysed, it appears that more of them are present in the nodes draining the pregnant uterus (Fig. 4b, bottom left; pregnant 14·6 ± 0·6, n = 13; never-mated 11·9 ± 0·8, n = 15, P < 0·05). This is consistent with the finding in the uterus (above), suggesting an increase in the ‘myeloid’ DC population on day 15 of pregnancy. In contrast, as was seen in the uterus, the proportion of CD11b+ CD4+ DCs was not different on day 15 of gestation as compared with the non-pregnant state (Fig. 4b, bottom middle; pregnant 37·2 ± 3·0, n = 13; never-mated 33·4 ± 2·0, n = 15, P < 0·05).

Decreased CD11b CD4+ DCs are found in the uterine draining nodes of pregnant mice

A brief comparison of the uterus and the uterine draining nodes, regardless of pregnancy, suggests an increase in the number of DCs that are negative for (or very low expressers of) CD11b in the uterine draining nodes. CD11b DCs may be indicative of a plasmacytoid DC population.25 Few of these cells appear to express CD8 (see Fig. 4a). When we analysed the DCs in the uterine draining nodes for expression of CD11b and CD4, we found that a significant proportion of these were CD11b yet CD4+. However (Fig. 4b, bottom right), the proportion present in the nodes draining the pregnant uterus was decreased (pregnant 20·4 ± 1·0, n = 13; never-mated 28·6 ± 1·9, n = 15, P < 0·05). Although CD11b CD4 DCs may occupy a variable niche in the uterine draining nodes, we did not observe an increase on day 15 of pregnancy (summary data not shown). In a separate analysis, the presence of CD11b, but not the absence of CD4, correlated positively with the surface expression of MHC II (data not shown), and this was not affected by pregnancy.

DCs from the lymph nodes draining the non-pregnant and day 15 uterus express similar patterns of inflammatory cytokines and molecules

It has been asserted that pregnancy is a pro-inflammatory state.26 Yet, it is also asserted that DCs related to the maternal–fetal interface should support the development of regulatory T cells27 To continue to address this issue, CD45+ CD11c+ MHC II+ cells were sorted from the pooled (at least three mice) uterine draining nodes of non-pregnant and pregnant mice. The RNA extracted from the sorted populations was analysed in a PCR Array for over 70 inflammatory cytokines and chemokines (see Materials and methods, above). Table 1 shows the genes expressed at detectable levels in sorted CD45+ CD11c+ MHC II+ cells. Other genes in the array were below the level of detection in the assay. Most of the genes detected were known to be expressed in DCs of other tissues.9,28–33 However, in our results chemokine (C-C motif) ligand 6 (CCL6) was more highly expressed in pregnant versus never-mated mice (4·7-fold, P < 0·05). CCL6 is a chemoattractant for CD4+ T cells and B cells. It is also thought to be important in inflammatory responses,34 as it is a chemoattractant for macrophages and eosinophils.

Table 1.   Representative genes expressed by dendritic cells in the lymph nodes draining the pregnant uterus
  1. *Expressed more in the pregnant than in the never-mated uterine draining lymph node.

  2. CD45+ CD11c+ MHC II+ cells (MHC, major histocompatibility complex) were sorted from the lymph nodes of pregnant mice and used for isolation of RNA that was later assayed with the RT2 Profiler™ PCR Array PAMM-011 SuperArray to determine gene expression. All genes listed amplified robustly with an expected amplification curve relative to a panel of housekeeping genes. The expression of only one gene (*), CCL6, was significantly elevated in uterine draining node dendritic cells from pregnant as opposed to non-pregnant mice.

Bcl6B-cell leukemia/lymphoma 6
Casp1Caspase 1
CCR7Chemokine (C-C motif) receptor 7
CCR5Chemokine (C-C motif) receptor 5
CCL22Chemokine (C-C motif) ligand 22
CCL6*Chemokine (C-C motif) ligand 6
CXCL10Chemokine (C-X-C motif) ligand 10
MIFMacrophage migration inhibitory factor
TGFβ1Transforming growth factor, beta 1

Proportional decrease in spleen DCs during pregnancy

Although distal from the uterus, the spleen undergoes marked changes during pregnancy,35 possibly related to systemic changes in blood flow or hormone excretion. Moreover, the immune response to fetal1,35 and viral36 antigen appears to be preserved during pregnancy. To gain insight into a population of DCs distal from the uterus, yet possibly affected by systemic regulation during pregnancy, we isolated DCs from the spleens of pregnant and never-mated mice. As in our examination of other tissues, we first observed the proportion of cells in the lymphocyte gate that were CD45+ (Fig. 5a). The proportion of CD45+ cells in the spleen was significantly decreased by pregnancy (Fig. 5b, top) (pregnant 90·6 ± 1·5, n = 13; never-mated 97·7 ± 0·5, n = 15, P < 0·05). Because the proportion of CD45+ cells that were also CD11c+ MHC II+ was similar (Fig. 5b, bottom), we observed that pregnancy was associated with an overall decrease in the proportion of lymphocyte-sized cells that fit the mature DC phenotype. However, DCs present in the pregnant spleen expressed similar levels of both MHC II and CD80 to DCs in the non-pregnant spleen, suggesting equivalent activation (data not shown).

Figure 5.

 CD45+ and CD45+ CD11c+ MHC II+ cells in the spleen. Spleens from mice on day 15 of gestation and from never-mated controls were treated as above. (a) Example of flow cytometry data. Upper panel: pregnant mouse. Lower panel: never-mated mouse. Cells falling in the lymphocyte gate (upper left plots) were then gated on CD45. Those cells that were positive for CD45 were then examined for their expression of CD11c and major histocompatibility complex (MHC) II. (b) Cumulative data. Black symbols, pregnant mice; grey symbols, never-mated mice. Each symbol represents one spleen. Bars indicate the median. Upper panel: comparison of the percentage of cells in the lymphocyte gate that are CD45+. The y-axis shows the percentage of positive cells. Lower panel: percentage of CD45+ cells that were CD11c+ MHC II+.

Decreased pregnancy-induced heterogeneity in spleen DCs

In contrast to what was observed in both the uterus and the uterine draining nodes, there was a similar distribution between the DC subsets examined in the spleens of pregnant and never-mated mice. When we examined the expression of CD11b and CD8 in CD11c+ MHC+ CD45+ cells (Fig. 6a), we did not observe a significant difference in the proportion of CD11bhi CD8 and CD11bint/lo CD8+ populations. When spleen DCs were further examined for expression of CD4 (Fig. 6b, top), there was a striking contrast with either the uterus or uterine draining nodes, regardless of pregnancy, as the spleen contained less CD11b+ CD4 and comparatively very few CD11b+ CD4+ cells. The presence of DCs that were CD11b CD4+ and presumed to be plasmacytoid in nature was not significantly altered (Fig. 6b, bottom right).

Figure 6.

 Dendritic cell (DC) subsets in the spleen. CD45+ CD11c+ MHC II+ spleen cells were further studied by examination of CD11b, CD8 and CD4 expression. (a) Delineation of CD11bhi CD8 and CD11blo/− CD8+ populations. Top: example. Bottom: proportion in pregnant (black symbols) and non-pregnant (grey) mice. (b) Delineation of CD11b+ CD4 and CD11b CD4+ cells. Top: example. Bottom: proportion in pregnant (black symbols) and non-pregnant (grey) mice. Each symbol represents one spleen. The bar denotes the median.

IL-15 in the mature DCs of the uterus, uterine draining nodes, and spleen: the influence of pregnancy

Recent data have underlined the expression and importance of IL-15 in the decidua16,37 as an important regulator of decidual NK cells, and a constituent of decidual DCs. We devised a protocol to stain DCs intracellularly, using a commercially available enzyme-linked immunosorbent assay (ELISA) antibody. The upper left panel of Fig. 7 shows an example of the analysis. Single-cell suspensions of uterine cells were incubated with antibodies to CD45, CD11c and MHC II as above. Cells positive for all three markers were examined for intracellular expression of IL-15. In the uterus (Fig. 7, upper right), the proportion of DCs that expressed detectable IL-15 was observed to be much lower on day 15 of pregnancy, as compared with the non-pregnant state (pregnant 2·1 ± 0·4, n = 5; never-mated 7·9 ± 0·9, n = 3, P < 0·05). While this may be reflective of a sudden increase in secretion amongst this DC population and therefore decreased intracellular stores of IL-15, the similarity in mean fluorescence intensity (data not shown) of the populations of DCs in the pregnant and non-pregnant mice argues against this idea. However, another possibility is that IL-15-expressing DCs move from the pregnant uterus to the lymph nodes draining this tissue, perhaps reflective of a decreased demand by a shrinking uterine NK cell population.38

Figure 7.

 Expression of interleukin (IL)-15 in dendritic cells (DCs) from the uterus, uterine draining nodes and spleen. Single-cell suspensions were prepared from the uterus, uterine draining lymph nodes (DLN) and spleen. CD45+ CD11c+ MHC II+ cells (DCs) falling in the lymphocyte gate were stained with a biotinylated primary antibody specific for IL-15 and then streptavidin phycoerythrin Texas red and examined by flow cytometry. Upper left plot: an example. Contour lines belong to the test sample. Grey dots are the control without anti-IL-15 antibody. Upper right and lower plots: data from five pregnant mice (black symbols) and three never-mated mice (grey symbols). Each symbol represents a mouse. The y-axis shows the percentage of CD45+ CD11c+ MHC II+ cells that were positive. Upper right plot: DCs from the uterus. Lower left plot: draining lymph node DCs. Lower right plot: spleen DCs.

When we examined DCs from the uterine draining lymph nodes of day 15 pregnant and non-pregnant mice (Fig. 7, bottom left), we found no difference in the proportion of DCs that were IL-15 expressing (pregnant 7·23 ± 0·7; never-mated 6·8 ± 0·7, P > 0·05). This is consistent with our array data (see above) where we found that, while DCs from uterine draining nodes expressed IL-15, the level of expression was not significantly different on day 15 of pregnancy as compared with the non-pregnant state.

An important function of IL-15 is support of lymphocyte proliferation. One lymphoid tissue that increases in size during pregnancy is the spleen.35 Moreover, it has been shown that spleen T cells specific and non-specific (to a lesser extent) for fetal antigen cycle significantly between days 13 and 15 of gestation (Erlbacher39 and M Norton, KA Fornter and EA Bonney, unpublished data). We thus thought it important to begin to understand which cells in the spleen might contribute to the splenic IL-15 content. Examination of CD11c+ MHC II+ cells in the spleen (Fig. 7, bottom right) suggested an increased proportion of cells expressing IL-15 in the spleen from day 15 pregnant mice (pregnant 1·1 ± 0·2; never-mated 0·5 ± 0·1, P < 0·05). Thus, DCs may contribute to lymphocyte proliferation in the pregnant spleen through expression of IL-15.


A specialized population of DCs in the uterus?

It is currently thought that different tissues may contain different populations of DCs. Consistent with that idea, we found subtle differences in the MHC II+ CD11c+ DC population in the uterus, uterine draining lymph node and spleen, even in the non-pregnant state. An example is the appearance of CD8+ DCs expressing very low levels of CD11b in the draining node and spleen (compare Figs 1a–4a and 6a). According to some authors,12 there may be T-cell lineage-related precursor DCs in lymphoid tissues, and this may be the source of the cells we have observed. These DCs may also be a subset of DCs that is thought to circulate only within lymphoid tissue. Another example of a difference between uterine and lymphoid tissues is the presence of CD11b CD4+ cells in the draining lymph node and spleen, but not the uterus. Such cells may represent a subset of plasmacytoid DCs,40 cells that produce type I interferons upon stimulation.

In some tissues, for example the skin,41 there exists a specialized population(s) of DCs whose major duty is to sample and process antigens found in the tissue and then move to draining nodes to present these antigens to T cells. We hypothesize that the presence of CD11b+ CD4+ DCs in the uterus and draining lymph node, but not in the spleen (Figs 1b, 4b and 6b), may indicate such a specialized DC population. In support of this idea, studies in human decidua11 have documented the presence of a highly proliferative immature population of CD4-expressing DCs that were conspicuously associated with uterine NK cells. In our studies, we do not see a large population of pregnancy-induced uterine DCs that were CD4 positive. In contrast we found that, on day 15 of gestation, the uterine population of CD11b+ CD4+ DCs was decreased as compared with the non-pregnant state. Although the disparate observations may be related to differences in mice and humans, it is formally possible that these DCs normally migrate to the uterine draining nodes or down-regulate their CD4 molecules with activation or as pregnancy proceeds into mid to late gestation in mice.

Another candidate for such a specialized DC population is those DCs that are CD11b+ CD8, and that represent nearly 60% of the DCs in the uterus and 30–40% in the draining node, yet only 5% of the DCs in the spleen. Other studies have also suggested that the level of CD8 expression in uterine DCs is low.18 The level of CD11b decreased the further away these DCs were from the uterus, suggesting the involvement of local (uterine) regulators of development. Moreover, the proportion of DCs that were CD11b+ CD8 was increased on day 15 of gestation in the uterus and uterine draining node, but not in the spleen. Others21 have found a decrease in the proportion of all CD11c+ cells that are CD11b+, suggesting possible decreased detection of CD11b on MHC II cells.

The mechanism(s) by which the uterus develops and maintains its population of DCs is unclear. While developmental changes including proliferation and death may play an important role in this process, trafficking is also likely to be important. For example, the increased presence of various populations of CD11b+ DCs in the uterus during pregnancy may reflect circulation and retention based on increased expression of CD11b ligands such as members of the intercellular adhesion molecule (ICAM) family in response to local (trophoblast-induced) stimuli.42

DCs: a critical tool for pregnancy-related immune regulation?

Dendritic cells are thought to be involved in both initiation and regulation of immune responses. The developing paradigm is that for most tissues there is heterogeneity in the DC population based on lineage, phenotype and function. This heterogeneity is driven by possibly systemic and also by tissue- and microenvironment-specific growth and other factors. Current thinking also suggests that within a tissue the balance between immunity and tolerance is based on the functional subtypes of DCs present. The idea persists in the literature that pregnancy exists as a state of obligate immune regulation either systemically or at the maternal–fetal interface. We therefore sought to find at least phenotypic evidence of shifts in DC populations in pregnancy as compared with the non-pregnant state. Contrary to current thinking, the only tissue that seemed to contain a lower proportion of CD45+ cells in pregnancy was the spleen. In addition, we did not find decreased proportions of CD11c+ cells in the uterus, uterine draining node or spleen that were positive for MHC II. Moreover, there was not a significant decrease overall in the level of expression of MHC or CD80. This argues against DC-based suppression, as measured by decreased maturity, associated with pregnancy. A caveat to this is the possibility of down-regulation of MHC, or an influx of immature DCs at a discrete time-point early after implantation.43 Although we could have missed a discrete window when DCs express an ‘immature phenotype’, our studies are focused on the homeostasis achieved at mid gestation (after placental remodelling) which is probably critical to continued pregnancy and prevention of premature birth. There may be other developmentally important time-points related to pregnancy, including peri-implantation, parturition, involution and weaning, when DCs of the reproductive tract also deserve to be examined closely, and these will be the focus of future studies.

In comparison with the uterus, the CD11c+ population in the uterine draining nodes expressed higher levels of both MHC and CD80. For us, this finding is consistent with the idea that trauma at the maternal–fetal interface might lead to uptake of fetal cells or cellular debris that could then be processed and presented by uterine node DCs.1,39 It is also consistent with the idea that pregnancy does not obligately depress DC activation or maturation. Thus quiet, but not suppressed DCs are usually in the uterus, and these can be activated and move to the draining node to present antigen to T cells. In these studies, one antigen that could be presented is H-Y,44,45 as it is present on male fetuses as early as the blastocyst stage. However, other developmentally (fetal) specific antigens could also be presented and drive T-cell responses. These antigens might be important in shaping the T-cell response even during syngeneic pregnancies in mice strains, such as BALB/c, not responsive to H-Y.

A potential limitation of these studies is that they were not performed in multiparous C57BL/6 mice, especially in light of older data46,47 suggesting that multiparity induces systemic tolerance to H-Y. In contrast, more recent data1 suggest that multiparity produces immune activation, and this is consistent with the current examination of lymphoid DCs in pregnant mice. The discrepancy in the reported effect of multiparity may be attributable to the number of pregnancies, with tolerance produced in mothers who have had a larger number (8–11) and activation in those who have had a smaller number (2–3), making comparison of DCs and T cells in the two types of mothers potentially interesting. However, it is also possible that this discrepancy is not related to a change in ‘professional’ DC phenotype, but instead related to an increased depot of chimeric non-DC fetal cells that express male antigen, but cause maternal T-cell depletion as a result of insufficient or inappropriate costimulation. The mechanism of tolerance in grand multiparas would then be similar to that found in non-pregnant females given large quantities of resting B cells.48

An important arising assertion is that the DC population in the uterus and uterine draining nodes, while able to process and present antigen at ‘steady state’, does so with the tendency to inhibit T-cell responsiveness and to generate functional tolerance to fetal antigens.5,24 This idea has been somewhat supported by the finding that specific modulation of DCs adoptively transferred to the uterus49 leads to better pregnancy outcome in a model of abnormal pregnancy.

In our model of normal pregnancy in which the mother is exposed to a common fetal antigen, H-Y, we did not specifically examine the presence of so-called ‘plasmacytoid’ DCs, i.e. DCs that are thought, as a result of low levels of MHC and costimulatory molecules, or through activation of regulatory T cells,50 to support immune tolerance. We did, however, note that a possible subset of these cells, expressing CD4 but not CD11b, was actually not expressed in the uterus, and was decreased in the uterine draining nodes at mid gestation. Further, there was no difference in this subset in the spleens of pregnant versus non-pregnant mice. This argues against the idea of such a sentinel tolerizing DC as a critical mechanism to support normal pregnancy.

We also did not observe a decreased (or an increased) proportion of DCs expressing CD8 but not CD11b, which are thought to support Th1 responses (reviewed for example in51), consistent with another study based on careful immunohistochemistry.21 Moreover, we observed a pregnancy-related increase in DC expression of the chemokine CCL6, which is thought to support inflammatory responses.34 Finally, DC activation is apparent when pregnant mice undergo LPS-induced preterm delivery (P Bizargity, M Phillippe, R Del rio Guerra and EA Bonney, unpublished data).

Classic self–non-self models of immune activation would suggest that, unless specifically suppressed or deviated, the immune system must respond to that which is non-self. Such models would demand comparative study of DCs from the reproductive tracts of female mice mated to genetically similar (for example syngeneic) and dissimilar males, or, in this study, compel examination of tissue surrounding male versus female concepti. Unfortunately, the molecular nature of `non-self', or how DCs might specifically detect it, has only partially been determined with regard to infectious agents.52 It remains unclear how a DC, in the process of presenting a particular fetal antigen, could determine that the antigen came from a syngeneic or allogeneic fetal cell.

However, there exists an alternative model, where metabolic dysregulation or other dysfunction,53 detected at a molecular level,54 is the primary driver of immune system activation. This model, in the context of pregnancy,55 supports study of the steady state of any healthy pregnancy, such as presented herein, where fetal antigen can be recognized.

We did, however, observe an increase in the proportion of DCs that express CD11b, and the majority of this increase appeared to be in CD11b+ cells that are CD4 negative. While the increase in the proportion of cells expressing CD11b suggests an increased capacity to support Th2 responses,51 the implication of the lack of CD4 is unclear and may have to do with increased activation or trafficking. Other studies have also suggested an increased expression of IL-10 over IL-12 in DCs of the uterus;18 however, this was found in pregnant and in non-pregnant mice, suggesting that this ratio, to an extent, may support normal functioning of the uterus apart from pregnancy.56 Finally, in light of other studies,3,57 these data do not suggest that successful pregnancy depends critically on a shift to this class of response.

DCs and tissue homeostasis

Current thinking about DCs suggests that they comprise a diverse, highly plastic and multifunctional group of cells that are at once sensitive to, and eminently capable of shaping, their environments. If this is the case, the functions of DCs may have implications outside that of immune regulation, particularly in a tissue or an animal undergoing significant developmental change. Our observation that uterine DCs can express IL-15 may be an example, although our studies may have missed an early-pregnancy surge in IL-15 expression. In the mouse uterus, NK cells, through expression of interferon (IFN)-γ and other factors, orchestrate several processes including modification of decidual arteries and movement of trophoblast. The function of NK cells in this regard is dependent on IL-15.37,58 As another example, NK cell activity at the maternal–fetal interface may include elaboration of IL-10 which can regulate angiogenesis56 as well as apoptosis,59 and recent data (for example60) suggest that the interaction between NK cells and DCs is important in expression of this cytokine by NK cells. Thus, not only immune function, but also anatomic development at the maternal–fetal interface may be linked to collaboration between NK cells and DCs.

An expanded view, then, is that DCs probably interact, according to a complex scheme, within the uterus, within the draining nodes and systemically in order to support both the development and immune protection of the fetus. The details of the complex signals produced by DCs and their changing capacity to receive signals from their environment deserve wider consideration other than as part of obligate mechanisms of immune tolerance.


This work was supported by NIH R01 HD043185, the University of Vermont Department of Obstetrics, Gynecology and Reproductive Sciences, and the Vermont Center for Immunology and Infectious Diseases. We thank Matt Poynter for critical reading of this manuscript, Colette Charland for assistance with flow cytometry, Marilyn Wadsworth of the University of Vermont Microscopy Imaging Facility for assistance with confocal microscopy, and Tim Hunter and Scott Tighe from the University of Vermont Microarray facility for advice. Thanks also to all members of our laboratory for useful discussions and support.