Peripheral Monocyte Expression of the Chemokine Receptors CCR2, CCR5 and CXCR3 is Altered at Parturition in Healthy Women and in Women with Systemic Lupus Erythematosus

Authors


Correspondence to: U. Holmlund, PhD, Department of Molecular Biosciences, The Wenner-Gren Institute, Stockholm University, SE-106 91 Stockholm, Sweden. E-mail: ulrika.holmlund@wgi.su.se

Abstract

Monocytes are precursors of macrophages and recruited to the uterus throughout pregnancy to perform important immunological functions. In this study, we hypothesized that pregnant women have reduced peripheral monocyte expression of chemokine receptors and alterations in PBMC responses to microbial stimuli as an adaption to pregnancy and that these changes are less pronounced in women with autoimmunity. We therefore investigated the chemokine receptor expression, migratory behaviour and responses to microbial stimulation of peripheral monocytes from pregnant women at parturition (n = 13) and from non-pregnant women (n = 9). In addition, we compared healthy pregnant women with women suffering from SLE (n = 5), a condition with pronounced systemic inflammation increasing the risk for pregnancy complications. We demonstrate that peripheral monocytes are affected by pregnancy with reduced percentages of CCR2+, CCR5+ and CXCR3+ monocytes of both classical (CD16−) and inflammatory (CD16+) subsets and that the trophoblast-secreted chemokine CCL2/MCP-1 recruited monocytes of both subsets in vitro. Further, PBMCs from pregnant women had a divergent response to microbial stimulation with lower CCL5/RANTES and higher CCL2/MCP-1 secretion compared with non-pregnant women. In addition, pregnant women had lower basal PBMC-secretion of CCL5/RANTES and higher basal secretion of IL-10 and CCL2/MCP-1. Interestingly, the women with SLE responded similar to pregnancy as did healthy women with lower percentages of CCR2+, CCR5+ and CXCR3+ monocytes. However, they had increased expression of CCR5 on CD16+ monocytes and heightened PBMC-secretion of CCL5/RANTES. In conclusion, our data indicate that monocyte chemokine receptor expression and the chemokine milieu during pregnancy are tightly regulated to support pregnancy.

Introduction

During pregnancy, immunological changes occur both systemically and locally in utero. A dampening in T cell-mediated immunity is compensated by an increased innate activity in the circulation, with activated monocytes and granulocytes and an activation of the complement system [1]. Both implantation and parturition are believed to be inflammatory processes, while the second trimester of pregnancy is more anti-inflammatory [2]. Previous studies show that an autoimmune condition like systemic lupus erythematosus (SLE) can increase the risk of pregnancy complications [3]. Research in reproductive immunology has generally focused on the intra uterine environment; however, few studies have investigated how the different peripheral monocyte populations are affected by pregnancy in health and disease [4].

Peripheral blood monocytes are the circulating precursors of macrophages. Monocytes are commonly divided into two major subsets; classical monocytes (CD14+ CD16−) and inflammatory monocytes (CD14+ CD16+). The classical monocytes, which represent 80–90% of blood monocytes, express high levels of the high-affinity receptor for IgG; CD64 and the chemokine receptor CCR2 and produce interleukin (IL)-10 rather than pro-inflammatory cytokines when stimulated by microbial compounds. The inflammatory monocytes are believed to be more matured [5] and produce more IL-12 and tumour necrosis factor (TNF) and less IL-10 upon microbial stimulation [6]. Monocytes of both subsets express several toll-like receptors (TLRs) and respond to a large range of microbial stimuli including LPS, which is a TLR4-ligand [7].

Chemokines and their receptors are involved in chemotaxis of leucocytes during homoeostatic and inflammatory conditions and can be expressed constitutively or upon stimulation [8]. Further, chemokines are involved in the recruitment of immune cells, like monocytes, towards the decidua during pregnancy. In the placenta, decidual cells express several chemokine receptors and their corresponding ligands involved in the recruitment of monocytes and other leucocytes [9-11]. Trophoblasts can regulate both the recruitment of monocytes to the placenta during pregnancy by secreting exosomes and chemokines, and monocyte differentiation into macrophages [12, 13]. Decidual macrophages function as phagocytes clearing apoptotic cells and cellular debris, facilitating trophoblast invasion and protecting against infection [14]. CCR2 is a chemokine receptor mainly expressed on monocytes but is also found on other cell types like NK cells [15], immature dendritic cells and activated T cells [16]. There are several ligands to CCR2 [16] of which CCL2/MCP-1 is the chemokine primarily recruiting monocytes to tissues [17] and is produced by trophoblasts [12, 13] and decidual macrophages [18, 19]. CCR5 and CXCR3 are chemokine receptors expressed on the surface of Th1-polarized cells [20]. It is known that monocytes also express CCR5, a receptor for several ligands including CCL5/RANTES that are able to induce monocyte migration [21, 22]. The CXCR3 ligands are normally associated with inflammation and include CXCL9/MIG and CXCL-10/IP-10. They are up-regulated in pro-inflammatory milieus and are responsible for the influx of immune cells to inflammatory sites. Besides on Th1-cells, CXCR3 has been found on NK cells, dendritic cells, B-cells [21] and monocytes [23, 24].

In this study, we hypothesized that pregnant women have a reduced peripheral monocyte expression of CCR2, CCR5 and CXCR3 in addition to alterations in PBMC responses to microbial stimuli as an adaption to pregnancy and that these changes are less pronounced in women with autoimmunity. To test this hypothesis, we have investigated peripheral monocytes from pregnant women at parturition and non-pregnant women in terms of chemokine receptor expression, responses to inflammatory stimuli and migratory capacity. In addition, we have compared healthy pregnancy to pregnancy complicated by the autoimmune disease SLE, a disease with several immunological abnormalities and pronounced systemic inflammation increasing the risk for pregnancy complications.

Materials and methods

Subjects

Healthy pregnant women (n = 13) with uncomplicated, full-term pregnancies, were recruited at the Karolinska University Hospital Solna, Stockholm, Sweden, at the time of parturition. The blood samples were taken during active delivery in all women except one, which had an elective caesarean section. Non-pregnant healthy female volunteers at reproductive age were used as controls (n = 9). In addition, a group of pregnant women with SLE (n = 5) was included to compare normal pregnancy with pregnancy complicated by an inflammatory disease known to increase the risk for pregnancy complications. The blood samples were taken during active delivery in all pregnant SLE patients, except for one woman, which had a caesarean section without labour. As controls for the pregnant SLE patients a group of non-pregnant women with SLE was included (n = 10). Four of the SLE patients were included both in the pregnant and the non-pregnant group. Blood samples were taken from the non-pregnant SLE patients 12 weeks after delivery, a time point at which pregnancy influence on immunological parameters have disappeared. All SLE patients were enrolled in a programme at Karolinska University Hospital, Solna, Stockholm, Sweden, where the women's health clinic and the rheumatology clinic collaborates in evaluation, follow-up and treatment of women with systemic autoimmunity. All SLE patients fulfilled at least four of the American College of Rheumatology classification criteria for SLE [25]. Patients were continuously evaluated by a rheumatologist and if pregnant also by an obstetrician. At each clinical evaluation, routine tests were performed including blood pressure, screening for anti-cardiolipin antibodies, β2 glycoprotein, SSA/SSB, ANA, anti-double stranded DNA as well as complement C3 and C4 levels, coagulation capacity and liver, kidney and thyroid function, C-reactive protein and blood cell counts. SLE activity was measured using the SLE disease activity index (SLEDAI). The disease was considered active if the SLEDAI score was ≥4. For demographic data of the women included in the study see Table 1. All pregnancies included in this study ended in live births.

Table 1. Demographic data of the women included in the study
GroupCases (n)Age (years)Parity (n)Gestational week at parturitionMode of delivery (vaginal/caesarian)SLEDAI ScorePrednisone dose (mg)
  1. There were no significant differences between the groups regarding age, parity, delivery mode or gestational week at delivery.

  2. a

    Median (range).

  3. b

    All patients fulfilled at least four of the American College of Rheumatology classification criteria for SLE.

Pregnant women1332 (22–42)a2 (1–4)a40 (37–42)a12/1
Non-pregnant women931 (22–41)a0 (0–2)a
Pregnant SLEb534 (30–36)a2 (1–2)a37 (34–39)a3/24 (0–14)a2.5 (0–20)a
Non-pregnant SLEb1035 (24–38)a2 (1–3)a2 (0–6)a5 (0–15)a

The study was approved by the Human Ethics Committee at Karolinska Institute, Stockholm, Sweden (Dnr 2009/740-32), and all women gave their informed written consent.

Blood sampling and isolation of PBMC

Venous blood (8–16 ml) was obtained in heparinized vacutainer tubes (BD Biosciences Pharmingen, San Jose, CA, USA). The PBMCs were isolated by Ficoll-Hypaque (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) gradient centrifugation. The cells were frozen in freezing medium [RPMI-1640 supplemented with 20 mM HEPES, penicillin (100 U/ml), streptomycin (100 μg/ml), l-glutamine (2 mM) (all from HyClone Laboratories, Inc, South Logan, UT, USA), 40% heat-inactivated FCS (Gibco, Invitrogen, Auckland, New Zealand) and 10% dimethyl sulfoxide]. Isolated cells were frozen gradually 1 °C/min to −80 °C in a freezing container (Nalgene Cryo 1 °C; Nalge Co, Rochester, NY, USA), and the samples were stored in liquid nitrogen until analysed.

In vitro activation of PBMC

PBMCs were thawed and washed twice in RPMI-1640 followed by trypan blue exclusion of non-viable cells. The cells were diluted to 106 cells/ml in culture medium [RPMI-1640 supplemented with 20 mM HEPES, penicillin (100 U/ml), streptomycin (100 μg/ml), l-glutamine (2 mM) (all from HyClone Laboratories Inc) and 10% heat-inactivated FCS (Gibco, Invitrogen)]. The cells were incubated in 48-well plates (flow cytometry analyses) or 96-well plates (cytokine/chemokine measurements) (Costar, Cambridge, UK) either with addition of culture medium only or stimulated with 1 ng/ml of LPS (Invivogen, San Diego, CA, USA) at 37 °C in 6% CO2 atmosphere for 3 h (flow cytometry) or 24 h (cytokine/chemokine measurements). After incubation, the cells were subjected to flow cytometry, and the supernatants were collected by centrifugation and stored at −80 °C until further analysed. Kinetic and titration studies were performed to determine the optimal time points and concentrations for the experiments.

Flow cytometry

After incubation, the PBMCs were harvested and washed twice in cold PBS. The cells were pre-incubated with 10% normal human serum for 10 min to block Fc receptors and thereafter stained with titrated amounts of the following mouse anti-human antibodies: CD14-PerCP-Cy5.5 (clone: M5E2), CD16-PE (clone: 3G8), CD64-FITC (clone: 10.1), CCR2-Alexa Fluor 647 (clone: 48607), CCR5-APC (clone: 2D7/CCR5), CXCR3-APC (clone: 1C6/CXCR3) (BD Biosciences Pharmingen) and TLR4-Alexa Fluor 647 (clone: HTA125) (eBiosciences, San Diego, CA, USA). Corresponding isotype-matched antibodies or unstained cells were used as negative controls. A minimum of 5000 monocyte events based on forward and side scatter properties were acquired using a BD FACSCalibur flow cytometer (Becton Dickinson). Analysis was performed with flowjo software 9.4.11 (TreeStar, Ashland, OR, USA), gating on the living mononuclear cell population of PBMC on the basis of the forward and side scatter properties, and only CD14+ cells were used for the analysis. The results were based on percentage of positive cells, and surface expression was defined as geometrical mean fluorescence intensity (GeoMFI).

Cell migration

PBMCs were thawed, washed twice in RPMI-1640 and diluted to 3 × 106 cells/ml in culture medium [Advanced RPMI-1640 (Invitrogen) supplemented with penicillin (100 IU/ml), streptomycin (100 μg/ml), l-glutamine (2 mM) (all from HyClone Laboratories Inc) and 0.5% FCS (Gibco, Invitrogen)]. Medium only (basal migration), medium with recombinant human CCL2/MCP-1 (R&D Systems, Minneapolis, MN, USA) or medium with N-Formyl-Met-Leu-Phe (FMLP, positive control) (Sigma, Saint Louis, MO, USA) were added to 24-well low-binding plates (Costar). Transwells (5 μm pore size, Costar) were inserted into the wells, and 300,000 cells in 100 μl of medium were added to each transwell. A well without insert was used as a control for the number of monocytes added to the transwells, and all samples were run in duplicates. The cells were incubated at 37 °C in 6% CO2 atmosphere for 2 h. After incubation, the inserts were removed, and the migrated cells and control cells were harvested and washed in cold PBS. The cells were pre-incubated for 10 min with 10% normal human serum to block Fc receptors and thereafter stained with the mouse anti-human antibodies: CD14-PerCP-Cy5.5 (clone: M5E2), CD16-PE (clone: 3G8) and CCR2-Alexa Fluor 647 (clone: 48607) (BD Biosciences Pharmingen). After staining, cells were resuspended in 250 μl PBS and acquired for 90 s/tube at high rate using a BD FACSCalibur flow cytometer (Becton Dickinson) and analysed with flowjo software 9.4.11 (TreeStar). The percentage of migrating monocytes was calculated based on the number of migrated monocytes from each subset divided by the total amount of monocytes of that subset added to the transwell.

Immunohistochemical staining of placental tissue

Placental tissue was taken immediately after delivery or kept refrigerated (at 4 °C) until processed (n = 10). Placental slices were taken in a circular fashion around the umbilical cord insertion. The slices spanned from the foetal membranes to the decidua. Slices were washed in NaCl, fixed in formalin and paraffin embedded. Immunohistochemical staining for CCR2 and CD14 was performed on 4.5-μm sections using a goat anti-human CCR2 antibody (Nordic BioSite, Täby, Sweden) or a mouse anti-human CD14 antibody (Abcam, Cambridge, UK) as previously described [26]. The whole placental tissue was evaluated, and staining without the primary antibody was used as negative control.

Cytometric bead array

The concentrations of cytokines and chemokines in cell culture supernatants were measured using the Cytometric Bead Array (CBA, BD Biosciences Pharmingen) technique. The human inflammatory cytokine kit (IL-1β, IL-6, IL-10, IL-12p70, TNF, CXCL8/IL-8) and the human chemokine kit (CCL2/MCP-1, CCL5/RANTES, CXCL8/IL-8, CXCL9/MIG, CXCL10/IP-10) were used. The supernatant samples were run according to the recommendations from the manufacturer, but using an extended standard curve. The samples were acquired using a BD FACSCalibur flow cytometer and analysed with the fcap Array v2.0 software (SoftFlow, St.Louis Park, MN, USA). Calibration of the flow cytometer was performed using BD CaliBRITE Beads and BD FACSComp (BD Biosciences Pharmingen).

CCL5/RANTES ELISA

CCL5/RANTES levels in culture supernatants were determined employing a sandwich ELISA procedure (CCL5/RANTES DuoSet ELISA; R&D Systems) following the manufacturer's instructions. The optical density was determined using a microplate reader (Molecular Devices, Sunnyvale, CA, USA) set at 450 nm.

Statistics

The statistica 10.0 software package (StatSoft Scandinavia AB, Uppsala, Sweden) was used for statistical analysis. Groups were compared with Kruskal–Wallis anova followed by Mann–Whitney U-test. Comparison of parameters within the same individual was made with the Wilcoxon matched pairs test. A Spearman rank-order correlations test was used to test for associations between measured parameters and SLEDAI score. For statistical reasons, all undetectable cytokines and chemokines were set to a value of 0.1 pg/ml. The differences were considered significant if P < 0.05.

Results

Peripheral monocyte chemokine receptor expression is affected by pregnancy

Characterizing monocytes from pregnant women at parturition and from non-pregnant women revealed a higher percentage of classical CD14+ CD16− (CD16−) monocytes in pregnant women, while no difference was seen for the inflammatory CD14+ CD16+ (CD16+) monocyte population compared with non-pregnant women (data not shown). Further, monocytes from pregnant women had significantly lower surface expression of CD16 (P < 0.001) (Fig. S1A), higher percentage of CD16+ CD64+ monocytes (data not shown) and an increased surface expression of CD64 on CD16+ monocytes compared with non-pregnant women (P = 0.003) (Fig. S1B).

Next, we characterized the expression of the chemokine receptors CCR2, CCR5 and CXCR3 on the major monocyte subpopulations in all women and found fewer CCR2+ cells among the CD16+ monocytes compared with the CD16− population (P < 0.01), but higher percentage of CCR5+ and CXCR3+ cells (P < 0.05) (Fig. 1). The surface expression of CCR2 and CXCR3 was significantly higher on CD16+ monocytes compared with CD16− monocytes (P < 0.01), while no difference was seen for the expression of CCR5 (data not shown).

Figure 1.

Peripheral monocyte chemokine receptor expression is altered by pregnancy. PBMCs were cultured for 3 h, stained for CD14, CD16, CCR2, CCR5, CXCR3 and acquired using flow cytometry. Women at parturition had reduced percentages of (A) CCR2+, (C) CCR5+ and (E) CXCR3+ monocytes of both CD16− and CD16+ subtypes compared with non-pregnant women. Pregnant women (n = 10), non-pregnant women (n = 9). Representative histograms of monocytes from pregnant and non-pregnant women positive for (B) CCR2, (D) CCR5 and (F) CXCR3. Black line: receptor staining, grey line: isotype control. (A, C, E) Boxes cover the middle 50% of the data values, between the 25th and 75th percentiles, with the central line as median. Lines extend out to non-outlier maximum and non-outlier minimum.

To determine whether pregnancy can affect monocyte chemokine receptor expression, we evaluated the expression of CCR2, CCR5 and CXCR3 on CD16− and CD16+ monocytes from pregnant women at the time of parturition and from non-pregnant women. The percentages of CD16− CCR2+, CD16− CCR5+ and CD16− CXCR3+ monocytes were lower in pregnant women compared with non-pregnant women (P = 0.01, P = 0.003, P < 0.001, respectively) (Fig. 1). Further, the same pattern was seen for the CD16+ monocytes, where the percentages of CD16+ CCR2+, CD16+ CCR5+ and CD16+ CXCR3+ monocytes were lower in pregnant women compared with non-pregnant women (P = 0.04, P = 0.09, P < 0.001, respectively) (Fig. 1).

In addition, the surface expression of CCR2 and CXCR3 on both monocyte subpopulations was significantly lower in pregnant women compared with non-pregnant women, while no significant difference was seen for the surface expression of CCR5 between the groups (data not shown).

Monocytes from pregnant and non-pregnant women migrate towards CCL2/MCP-1

To investigate if the deviating pattern of monocyte chemokine receptor expression in pregnant women influenced migratory capacity, we compared migratory behaviour of monocytes from pregnant women at parturition and from non-pregnant women towards CCL2/MCP-1, which is a major monocyte chemoattractant.

Both CD16− and CD16+ monocytes from pregnant and non-pregnant women migrated to a significantly higher extent towards the CCR2-ligand CCL2/MCP-1 when compared to basal migration (P < 0.05) (Fig. 2A). Migration against CCL2/MCP-1 resulted in a significant decrease in the number of CCR2+ monocytes in both groups of women (P = 0.04) compared with the initial percentage of CCR2+ cells (data not shown). However, no differences could be seen in migratory capacity between pregnant and non-pregnant women. In addition, incubation of PBMCs with CCL2/MCP-1 resulted in a decreased percentage of CCR2+ monocytes, further supporting the observation seen after migration towards CCL2/MCP-1. In contrast, basal migration without chemoattractant increased the percentage of CCR2+ monocytes (data not shown).

Figure 2.

Monocytes from pregnant women at parturition and non-pregnant women migrate towards CCL2/MCP-1. PBMCs from pregnant and non-pregnant women were added to transwells and allowed to migrate basally or against CCL2/MCP-1 for 2 h. Cells were stained for CD14, CD16, CCR2 and acquired using flow cytometry. (A) Both CD16− and CD16+ monocytes from pregnant and non-pregnant women had an increased migration towards CCL2/MCP-1 compared with medium only. The percentage of migrating monocytes was calculated based on the number of migrated monocytes divided by total number of monocytes added to the transwell. Mean values, with standard deviation. *P < 0.05 comparison between basal and CCL2/MCP-1 induced migration. Pregnant women (n = 5), non-pregnant women (n = 5). (B) CCR2+ and CD14+ cells were detected within the decidual tissue (maternal part of the placenta) and in trophoblasts and chorionic villi, respectively (foetal part of the placenta). Left: CCR2+ cells, middle: CD14+ cells, right: negative control, 10× magnification. Inserts show positive cells at 100× magnification. Brown staining represents positive cells. Representative staining (n = 10).

To verify that CCR2+ and CD14+ cells could be detected within the placenta, we stained consecutive placental sections with antibodies against CCR2 and CD14. Immunoreactivity for both CCR2 and CD14 was detected within the decidua in all samples investigated (Fig. 2B).

Monocytes from pregnant and non-pregnant women have divergent responses to LPS-stimulation in terms of chemokine receptor expression and PBMC-secretion of cytokines and chemokines

We also wanted to assess if the pregnancy-induced alterations in monocyte chemokine receptor expression were maintained after microbial stimulation.

Monocytes from pregnant women at parturition and from non-pregnant women responded similarly to LPS-stimulation in terms of increased expression of CD64 on CD16− monocytes and decreased percentage of CD16− CCR2+ monocytes. In addition, the surface expression of CD16 and CXCR3 increased on the CD16+ monocytes as well as the percentage of CD16+ CXCR3+ cells (data not shown).

Monocytes from non-pregnant women had LPS-induced differences not observed in the group of pregnant women. After stimulation, non-pregnant women had significantly higher percentage of CD16+ CCR2+ monocytes (P = 0.008) and a significantly higher expression of CCR5 on the CD16+ monocytes (P = 0.02) compared with unstimulated cells (Fig. 3A,B). Further, monocytes from non-pregnant women had significantly higher surface expression of TLR4 on both CD16− and CD16+ monocytes compared with pregnant women (P = 0.004). For both groups, TLR4 expression was higher on the CD16+ monocytes compared with CD16− monocytes (P < 0.05) (Fig. 3C).

Figure 3.

Monocytes from pregnant women at parturition respond differently to microbial stimulation compared with non-pregnant women. PBMCs were cultured for 3 h with medium only or stimulated with 1 ng/ml of LPS, stained for CD14, CD16, CCR2, CCR5 and acquired using flow cytometry. (A) Monocytes from non-pregnant women respond to LPS-stimulation with an increase in the percentage of CCR2+ cells, which was not observed for monocytes from pregnant women. (B) The surface expression of CCR5 on monocytes from non-pregnant women was increased after LPS-stimulation, while no difference was observed for monocytes from pregnant women. (C) The expression of TLR4 is reduced on monocytes from pregnant women compared with non-pregnant women. Pregnant women (n = 10), non-pregnant women (n = 9). Extremes are not shown (B, C). For explanation of the box plot model, see legend to Fig. 1.

Importantly, the percentages of CCR2+, CCR5+ and CXCR3+ monocytes remained significantly lower in pregnant women compared with non-pregnant women after LPS-stimulation (CD16−: P < 0.01 all receptors, CD16+: CCR2 P < 0.05, CCR5/CXCR3 P < 0.001) (Fig. S2). The same was seen for the surface expression of CCR2 and CXCR3; however, no significant difference was seen for CCR5 expression (data not shown).

Finally, PBMCs from pregnant women released higher levels of CCL2/MCP-1 and CXCL8/IL-8 both after LPS-stimulation (P = 0.04 and P = 0.07) and with medium only (P = 0.02), but significantly lower levels of CCL5/RANTES (LPS-stimulation P = 0.01 and medium only P < 0.001). No differences were seen between the groups for secretion of CXCL9/MIG or CXCL10/IP-10. In addition, PBMCs from pregnant women tended to have a higher basal secretion of IL-1β (P = 0.07), but a lower IL-1β response to LPS-stimulation (P = 0.08). Further, IL-6 tended to be higher after LPS-stimulation of PBMCs from pregnant women compared with non-pregnant women (P = 0.05), and PBMC-cultures from pregnant women tended to release higher basal levels of IL-10 (P = 0.05). No differences were seen between the groups regarding secretion of TNF (Table 2). The concentration of IL-12p70 in culture supernatants was too low for reliable evaluation (data not shown).

Table 2. Chemokines and cytokines in cell culture supernatants
Chemokine/cytokine (pg/ml)aPregnant unstim (n = 13)Pregnant LPS (n = 13)Non-pregnant unstim (n = 9)Non-pregnant LPS (n = 9)P-valueb
  1. B, Basal secretion, S, Stimulated secretion, NS, Non-significant, ND, Not detected cytokines and chemokines were set to 0.1 pg/ml for statistical analyses.

  2. a

    Median (range).

  3. b

    Comparison between pregnant and non-pregnant women.

CCL2/MCP-12957 (99–15,046)27,441 (5 713–49,292)239 (31.1–3 586)8934 (528–41,365)0.02B, 0.04S
CCL5/RANTES175 (81.5–420)880 (254–1 868)508 (181–680)1750 (603–2165)<0.001B, 0.01S
CXCL8/IL-8 (ng/ml)15.1 (2.53–34)155 (111–239)5.74 (0.63–22.5)135 (69.2–159)0.02B, 0.07S
CXCL9/MIG3.9 (0.1–17.1)0.1 (0.1–22.1)2.2 (0.1–12)NDNSB,S
CXCL10/IP-1094.4 (11.8–206)37.3 (4.9–112)27.6 (2.9–157)11.9 (0.1–134)NSB,S
IL-1β6.1 (0.1–201)1240 (27–3 438)3.0 (0.1–62.5)2633 (927–5382)0.07B, 0.08S
IL-629.6 (4.9–911)35,847 (5539–65,059)10.5 (4.55–1439)25,587 (9 999–35,528)NSB, 0.05S
IL-104.3 (2.2–14.7)340 (20–1 044)2.4 (0.1–12.5)498 (99–1 077)0.05B, NSS
TNF5.7 (2.2–178)729 (21–3 658)3 (2.3–10.2)1186 (288–5 203)NSB,S

Monocytes from women with systemic lupus erythematosus and healthy women are similarly influenced by pregnancy

To study the effect of pregnancy on monocytes from women suffering from SLE – a disorder characterized by pronounced systemic inflammation and increased risk for pregnancy complications, we analysed monocytes from pregnant women with SLE at parturition in the same way as previously described and compared them to non-pregnant women with SLE, healthy pregnant women and healthy non-pregnant women.

Most of the pregnancy-induced changes seen for monocytes from healthy women were also observed for monocytes from women with SLE. Pregnant women with SLE had a reduced percentage of CD16− CCR2+, CD16− CCR5+ and CD16− CXCR3+ monocytes and a lower percentage of CD16+ CCR2+ and CD16+ CXCR3+ monocytes compared to non-pregnant women with SLE (Fig. 4). In addition, monocytes from pregnant women with SLE had a similar migratory capacity compared to non-pregnant women with SLE, healthy pregnant women and healthy non-pregnant women (data not shown).

Figure 4.

Peripheral monocyte chemokine receptor expression in women with SLE is similarly affected by pregnancy as chemokine receptor expression on monocytes from healthy women. PBMCs were cultured for 3 h, stained for CD14, CD16, CCR2, CCR5, CXCR3 and acquired using flow cytometry. Pregnant women with SLE at parturition had reduced percentage of (A) CCR2+, (C) CCR5+ and (E) CXCR3+ monocytes compared with non-pregnant women with the disease. Pregnant women with SLE (n = 5), non-pregnant women with SLE (n = 9). Extremes are not shown (A, E). Representative histograms of monocytes from pregnant and non-pregnant women with SLE positive for (B) CCR2, (D) CCR5 and (F) CXCR3. Black line: receptor staining, grey line: isotype control. For explanation of the box plot model, see legend to Fig. 1.

Both pregnant and non-pregnant women with SLE had a significantly increased expression of CCR5 on CD16+ monocytes compared with healthy women (pregnant P = 0.04 and non-pregnant P = 0.02). Noteworthy, CD16− monocytes from pregnant women with SLE had significantly higher CCR5 expression compared with CD16− monocytes from non-pregnant SLE patients (P = 0.01) and also tended towards higher surface expression of CCR5 on the CD16− monocytes compared with healthy pregnant women (P = 0.06). However, no significant difference was seen between CD16− monocytes from non-pregnant healthy women and non-pregnant women with SLE (Fig. 5A).

Figure 5.

Pregnant women with SLE have divergent monocyte surface expression of CCR5 and PBMC-secretion of CCL5/RANTES at parturition compared with healthy pregnant women. (A) PBMCs were cultured for 3 h, stained for CD14, CD16, CCR5 and acquired using flow cytometry. Monocytes (CD16− and CD16+) from pregnant women with SLE had an increased expression of CCR5 compared with monocytes from healthy pregnant women. In addition, the surface expression of CCR5 was higher on CD16+ monocytes from pregnant women with SLE compared with non-pregnant women with SLE. Extremes are not shown. (B) PBMCs were cultured for 24 h with medium only or stimulated with 1 ng/ml of LPS, and the release of CCL5/RANTES (pg/ml) was measured in the supernatants. Healthy pregnant women (n = 12), healthy non-pregnant women (n = 8), pregnant women with SLE (n = 5) and non-pregnant women with SLE (n = 10). Extreme outliers are not shown. For explanation of the box plot model, see legend to Fig. 1.

Further, PBMCs from pregnant women with SLE released higher levels of CCL5/RANTES both after LPS-stimulation and with medium only compared with healthy pregnant women (P = 0.006, P = 0.02, respectively). The levels of CCL5/RANTES in PBMC-cultures from pregnant SLE patients were more similar to levels in PBMC-cultures from both healthy non-pregnant women and non-pregnant women with SLE (Fig. 5B, Table S1). No clear associations were seen between disease activity and measured parameters.

Discussion

Monocytes are key players in innate immunity and are chemotactically recruited to the in utero environment throughout pregnancy where they are important for implantation, foetal growth and parturition [27]. Here, we demonstrate for the first time that pregnancy influences peripheral monocyte chemokine receptor expression. Our results also reveal that pregnancy affects basal and LPS-stimulated PBMC cytokine and chemokine responses. Finally, we demonstrate that in spite of an aberrant pro- and anti-inflammatory cytokine profile, which is further pronounced during pregnancy [28], women with SLE have a strikingly normal peripheral monocyte population with regard to chemokine receptor expression.

Chemokines are important for recruitment of monocytes and other cells during immune responses. CCR2, the receptor for CCL2/MCP-1 has dual roles; promoting inflammation when expressed on APCs and T cells but involved in the anti-inflammatory response when expressed on T-regulatory cells [17]. Fewer monocytes from pregnant women at parturition expressed CCR2 than monocytes from non-pregnant women (Fig. 1A,B). Further, CCR2 expression was found on both CD16− and CD16+ monocytes, which is in contrast to some earlier reports [6, 29, 30]. However, in agreement with our data, others have found expression of CCR2 on CD16+ monocytes [31]. Monocytes are a heterogeneous, constantly differentiating population, and it is likely that the expression of the CCR2 receptor varies during monocyte differentiation and is neither on nor off. This might explain discrepancies between different studies. Further, the CD16+ monocytes might also rely on CCL2/MCP-1 recruitment via CCR2 towards inflammatory sites in addition to CCR5-mediated migration. Our results demonstrate that CD16− and CD16+ monocytes had comparable migratory capacity towards CCL2/MCP-1 even though a lower percentage of monocytes within the CD16+ subset expressed CCR2 (Fig. 2A). This could be explained by the higher surface expression of CCR2 on CD16+ monocytes.

CCL2/MCP-1 is the major monocyte chemoattractant, believed to recruit monocytes to the in utero environment throughout pregnancy when secreted by trophoblasts [18] and decidual macrophages [18, 19]. In this study, we have shown that monocytes from pregnant women at parturition migrate towards CCL2/MCP-1 in vitro. Further, even though CCR2 is expressed by several cell types in the placenta, our staining clearly shows that the CD14+ cells in the decidua are also CCR2+ (Fig. 2A,B). The migration of CCR2+ monocytes to the uterus at the time of parturition might result in a reduction of CCR2+ monocytes in the circulation, which could explain the reduced percentage of CCR2+ monocytes in pregnant women seen in this study (Fig. 1A). The capacity of monocytes to migrate in response to CCL2/MCP-1 seems however to be similar in both pregnant and non-pregnant women. Further, CCL2/MCP-1 has also been indicated in polarization towards Th2 responses and secretion of IL-4 [17], which may favour the Th2 dominance in utero during the main part of pregnancy. Interestingly, PBMCs from pregnant women in our study produced more CCL2/MCP-1 compared with non-pregnant women (Table 2), which is in agreement with previously published data [32]. However, as data in this study are from women at parturition, a time point during pregnancy that might differ substantially from the rest of pregnancy, further studies are needed to elucidate if our results are linked to parturition only or holds true for pregnancy in general.

Studies have shown that an excess infiltration of monocytes to the decidua could interfere with trophoblast differentiation [33]. Women with preeclampsia have increased macrophage numbers in the decidua and activated macrophages inhibits trophoblast invasion and induce trophoblast apoptosis by production of TNF [34, 35]. Indeed, stimulation of macrophages with TNF or IL-1β increases CCL2/MCP-1 secretion. Considering that monocytes migrate to the decidua and that activated monocytes can produce TNF [36], a highly regulated balance in monocyte recruitment and activation during pregnancy is of importance and failure of this regulation could result in pregnancy complications.

As for CCR2, the percentage of CCR5+ monocytes was reduced in pregnant women at parturition compared with non-pregnant women (Fig. 1C). CCR5 is expressed on a limited number of monocytes and is mainly involved in inflammatory responses, which indicates that it may not be the main receptor for monocyte recruitment to the uterus during pregnancy. In addition, CCR2 and CCR5 have been found to form homo- and heterodimers on cells [37, 38]. It has been shown that CCR5-specific ligands can prevent binding of CCR2-ligand CCL2/MCP-1 when CCR5 and CCR2 are co-expressed on cells and that the formation of homo- or heterodimers is dependent on the expression level of respective receptor [37]. A decrease in the amount of monocytes expressing CCR5 at the time of parturition might be a regulatory mechanism maintaining a high recruitment of monocytes via CCR2 to the uterus.

Although CXCR3 is mainly expressed on T cells, monocytes have been shown to express this receptor [23, 24] and to secrete the CXCR3-ligands; CXCL9/MIG and CXCL10/IP-10 [39, 40] which are mainly involved in recruitment of Th1 cells during inflammation. Exaggerated Th1 responses have been associated with pregnancy failure [41] and an increased ratio of Th1 (CCR5, CXCR3)/Th2 (CCR3, CCR4) chemokine receptors on T cells have been linked with recurrent miscarriages [42]. The Th1-associated cytokine IFN-γ can induce secretion of CXCR3-ligands from monocytes and other cells creating an amplification loop of the Th1 immune response. Expression of CXCR3 by monocytes might also create an autocrine regulation of CXCR3-ligand production by the cells, and thereby controlling monocyte influence over T cell recruitment during inflammation. A reduction of CXCR3+ monocytes and a reduced surface expression of the receptor together with reduced levels of CXCR3-ligands as well as regulated circulating IFN-γ levels during pregnancy might dampen Th1 responses potentially deleterious for pregnancy. Indeed, we have previously shown that pregnant women have significantly lower serum levels of the CXCR3-ligands CXCL9/MIG and CXCL10/IP-10 [28], which might further dampen the inflammatory loop.

It has been suggested that T cell-mediated immunity changes during pregnancy [1]. CCL5/RANTES is a main chemoattractant for activated T cells and the lower secretion of CCL5/RANTES from PBMC from women at parturition in our study could reflect a general suppression of T cell-mediated responses (Table 2). In addition, CCL5/RANTES has been indicated as a chemoattractant and inducer of effector NK cells [43]. During early pregnancy, NK cells are believed to be important for implantation [11, 44]. However, recruitment of NK cells to the in utero environment later during pregnancy might not be favourable.

Monocytes from non-pregnant women had LPS-induced changes, in terms of increased percentage of CD16+ CCR2+ cells and a higher expression of CCR5 on CD16+ monocytes, which were not observed in the group of pregnant women (Fig. 3A,B). Further, the cytokine IL-1β that function as an important mediator during inflammation [45] was secreted to a lower extent by PBMCs from pregnant women at parturition after stimulation with LPS (Table 2). This might reflect a need for tighter regulation of inflammatory responses during pregnancy. These results could also be related to our finding that monocytes from pregnant women have a lower surface expression of the LPS receptor TLR4 (Fig. 3C). Further, pregnant women had higher basal secretion of the immune regulatory cytokine IL-10 compared with non-pregnant women, a difference that could not be seen after stimulation with LPS.

We have previously shown that women with the autoimmune disease SLE seem to respond to pregnancy in a similar way as healthy women, in terms of changes of serum cytokines and chemokines over the course of pregnancy [28]. However, changes in pro- and anti-inflammatory serum components during pregnancy in women with SLE, occurring on top of already more pro-inflammatory levels, might increase their risk for pregnancy complications and flares. In this study, we have in addition to healthy women investigated monocytes from women with SLE at the time of parturition to compare healthy pregnancy with pregnancy complicated by an inflammatory disorder known to increase the risk for pregnancy complications. Interestingly, and in line with our previous findings [28], the women with SLE responded to pregnancy in a similar way as healthy women with a lower percentage of CD16-CCR2+, CD16-CCR5+ and CD16-CXCR3+ monocytes and a lower percentage of CD16+ CCR2+ and CD16+ CXCR3+ monocytes at the time of parturition (Fig. 4). However, both pregnant women and non-pregnant women with SLE had an increased expression of CCR5 on CD16+ monocytes compared with healthy women, indicating that the CD16+ monocytes in women with SLE have a more activated phenotype which might be mirroring ongoing inflammatory processes attributed to their disease. Interestingly, monocytes from pregnant women with SLE also had an increased expression of CCR5 on CD16− monocytes that was not seen in non-pregnant women with SLE or in healthy pregnant and non-pregnant women (Fig. 5A). In addition, the lower secretion of CCL5/RANTES from PBMCs from healthy pregnant women in our study was not seen for pregnant women with SLE (Fig. 5B), indicating that the possible pregnancy-induced down-modulation of T cell-mediated responses seen in healthy women is less pronounced in pregnant SLE patients. Indeed, T cell dysfunction like hyperactivity is a known phenomenon in patients with SLE [46]. No clear associations were seen between disease activity and measured parameters. However, steroid treatment might mask potential associations between disease activity and the results presented in this study.

Due to the small sample size, the results from pregnant SLE patients presented in this study need to be interpreted with caution and additional studies are needed to reveal if the deviating receptor expression and chemokine secretion could contribute to the pregnancy complications known to affect these women. In addition, further studies are needed to elucidate if the changes in the peripheral monocyte population seen in both healthy women and women with SLE are linked to parturition only or holds true for pregnancy in general.

In summary, we have demonstrated that peripheral monocytes at parturition are affected by pregnancy with a lower percentage of CCR2+, CCR5+ and CXCR3+ monocytes of both classical (CD16−) and inflammatory (CD16+) subsets compared with non-pregnant women and that pregnant women at parturition have a divergent response of PBMCs to microbial stimulation. In addition, we have demonstrated that women suffering from the autoimmune disease SLE respond to pregnancy in a similar way, as do healthy women with a lower percentage of CCR2+, CCR5+ and CXCR3+ monocytes at the time of parturition. In conclusion, the data presented in this study indicate that both monocyte chemokine receptor expression and the chemokine milieu during pregnancy are tightly regulated to support pregnancy.

Acknowledgments

We would like to acknowledge the critical and valuable comments on the manuscript made by Professor Eva Sverremark-Ekström, Stockholm University. In addition, we would like to thank Siv Rödin-Andersson, Lotta Blomberg and Berit Legestam at the Women Health Clinic and Eva Andersson and Birgitta Byström at the Research Lab for Reproductive Health for blood sampling and isolation of PBMCs. Additionally, we thank Agneta Rudels-Björkman and the staff at the Delivery Unit at Karolinska University Hospital, Stockholm, Sweden, for valuable assistance in collection of blood samples and Merja Lintula for patient coordination. We are also grateful to all women participating in the study. Finally, we thank Dr Dagbjort Petursdottir, Stockholm University for valuable input regarding migration experiments. This work was financially supported by grants from Clas Groschinsky′s and Sigurd and Elsa Goljes Memorial Foundations, Samariten, Swedish Society of Medicine, Swedish Society for Medical Research, Tore Nilsson′s Foundation for Medical Research and the Åhlén Foundation.

Conflict of interest

The authors have no conflict of interest to declare.

Ancillary