Lactobacillus rhamnosus GR-1 Stimulates Colony-Stimulating Factor 3 (Granulocyte) (CSF3) Output in Placental Trophoblast Cells in a Fetal Sex-Dependent Manner

Abstract

Bacterial vaginosis is associated with a 1.4-fold increased risk of preterm birth. We have shown previously that Lactobacillus rhamnosus GR-1 supernatant up-regulates interleukin 10 and down-regulates tumor necrosis factor-alpha output in lipopolysaccharide (LPS)-treated human primary placenta cultures in a fetal sex-dependent manner. We hypothesize that lactobacilli also exert their anti-inflammatory effect by up-regulation of colony-stimulating factor 3 (granulocyte) (CSF3), which is secreted from both immune and placental trophoblast cells, and that this activity is dependent on the sex of the fetus. Placental trophoblast cells were isolated from term elective cesarean section placentae using a Percoll gradient and separated from CD45⁺ cells using magnetic purification. Cells were treated with LPS in the presence or absence of pretreatments with L. rhamnosus GR-1 supernatant or chemical inhibitors of the intracellular signaling pathways. Phosphorylations of mitogen-activated protein kinase 14 (MAPK14, previously known as p38) and signal transducer and activator of transcription (STAT) 3 were measured by Western blot analysis, and levels of CSF3 were determined by ELISA. CSF3 output was increased only in the placental trophoblast cells of female fetuses treated with LPS, GR-1 supernatant, and a combination of both treatments. The GR-1 supernatant up-regulated the phosphorylation of STAT3 and MAPK14. CSF3 output was inhibited by both Janus kinases (JAK) and MAPK14 inhibitors. None of the treatments was able to increase CSF3 output in either the pure trophoblast or the CD45⁺ cell preparations alone. These results suggest an underlying mechanism for the sex difference in incidence of preterm birth and provide potential evidence for a therapeutic benefit of lactobacilli in reducing the risk of preterm labor.

Lactobacillus rhamnosus GR-1 increases colony-stimulating factor 3 output through activation of the members of the JAK/STAT and MAPK pathways in cultures containing both placental trophoblast and immune cells selectively from pregnancies with a female fetus.

INTRODUCTION

Cytokines are important immunoregulatory mediators at the human maternal-fetal interface. An imbalance of the pro- and anti-inflammatory cytokines is known to play a role in the mechanism whereby exposure to pathogens leads to intrauterine infection [1, 2]. Colony-stimulating factor 3 (granulocyte) (CSF3) is a protective cytokine with anti-inflammatory effects and is considered to be a key regulator of neutrophil production, essential for the clearance of bacterial pathogens as well as the modulation of inflammatory responses. CSF3 acts through its main receptor, CSF3 receptor (CSF3R), which has been detected in placentae, neurons, endothelial cells, and cardiomyocytes [3]. Inflammatory stimuli, such as endotoxins or cytokines synthesized by a group of specialized cells, including macrophages, stimulate the transcriptional and posttranscriptional mechanisms that lead to a change in CSF3 and subsequent neutrophil recruitment and activation [3]. Low levels of CSF3 have been associated with an increased incidence and severity of infection in preterm infants [4]. Studies have also shown that CSF3 is important in promoting survival of the granulocytic lineage cells and proliferation and migration of neutrophils [3] as well as trophoblast cells [5, 6].

Preterm birth complicates up to 13% of all pregnancies in the Unites States [7] and accounts for approximately 80% of neonatal mortality and morbidity [8]. Lactobacilli are the dominant urogenital microbiota in healthy females, and a reduction in their number is implicated in bacterial vaginosis (BV), a condition characterized by a vaginal pH of greater than 4.5 and the presence of pathogenic bacteria. BV is also associated with a 1.4-fold increase in the risk of preterm birth [7]. Antibiotics are the most common treatment for BV, with the aim of preventing infection-mediated preterm birth. However, this treatment is largely unsuccessful [9], and in some studies, antibiotics have actually been shown to increase the risk of preterm birth [10]. Probiotic lactobacilli may be a treatment alternative for BV [11], mainly because of their ability to replenish vaginal lactobacilli and modulate immunity [12–14]. In addition, several studies suggest that Lactobacillus rhamnosus GR-1 is able to populate the vagina and up-regulate the host's antimicrobial system after intravaginal use [15, 16]. We have shown previously that L. rhamnosus GR-1 is able to down-regulate the pro-inflammatory, and to up-regulate the anti-inflammatory, cytokines in human placental trophoblast cells [17]. This strain also up-regulates CSF3 production in mouse and human macrophages, and it is proposed that activation of the Janus kinase 2 (JAK2)/signal transducer and activator of transcription (STAT) 3 pathway by CSF3 mediates the L. rhamnosus GR-1-induced tumor necrosis factor-alpha (TNF) suppression in autocrine/paracrine routes [11]. Furthermore, incubation of the JEG-3 human trophoblastic cell line with CSF3 leads to activation of STAT3, a member of the JAK/STAT pathway, and mitogen-activated protein kinase (MAPK) 14, a member of the MAPK pathway [18]. However, the effect of L. rhamnosus GR-1 on CSF3 and its mechanism of action in human placental trophoblast cells have not been studied.

The immune response to bacterial pathogens appears to be sex-dependent, with females being more effective in mounting a response and having a better prognosis in septic shock than males during both the neonatal and adult periods [19, 20]. Furthermore, preterm birth is reported to be more frequent in women carrying a male fetus [21, 22], and male fetuses are more likely to face intrauterine or neonatal death as well as other complications, such as long-term neurological and motor impairments [23, 24], compared to females. Previously, we showed that lipopolysaccharide (LPS) increases TNF and interleukin (IL) 10 output in human placental trophoblast cells, an effect that is more pronounced in placentae of male compared to female fetuses. Our previous work also indicates that the LPS stimulation of prostaglandin-endoperoxide synthase 2 (PTGS2) is more pronounced in placentae of male fetuses, and that this is associated with a greater abundance of toll-like receptor 4, the receptor that primarily mediates the effect of LPS, in placental trophoblast cells of male fetuses. We have also shown that up-regulation of 15-hydroxyprostaglandin dehydrogenase (PGDH), the enzyme responsible for metabolism of primary prostaglandins, by lactobacilli is more prominent in placentae of female fetuses [17].

Therefore, in the present study, we sought to investigate possible sexual dimorphisms in the output of CSF3 as a result of bacterial endotoxin as well as L. rhamnosus GR-1 treatments and to determine whether this might be associated with differences in CSF3R expression levels. We further investigated whether CSF3 mediates the down-regulation of the proinflammatory cytokines by L. rhamnosus GR-1 supernatant. Finally, we sought to identify the primary source of CSF3 production by examining the differential response of trophoblast and immune cells, both separately and together, to bacterial endotoxin as well as L. rhamnosus GR-1 supernatant treatments.

MATERIALS AND METHODS

All studies were approved by the Research Ethics Boards of Mount Sinai Hospital and Faculty of Medicine, University of Toronto, in accordance with the Canadian Tri-Council Policy Statements on Human Ethics Reviews (IRB 04–0018-U). Placental tissues were collected from women undergoing term (>37 wk of gestation) elective cesarean section at Mount Sinai Hospital. The subjects experienced a healthy pregnancy with no clinical infection and had no signs of labor before delivery. Indications for delivery by elective cesarean section included abnormal presentation of the fetus, cephalopelvic disproportion, or previous cesarean sections. Informed consent was obtained before tissue collection.

Placental Trophoblast Cell Culture

Placental trophoblast cells were isolated using established primary culture protocols [25]. Placentae with adherent membranes were collected. Briefly, placental tissue was separated from the membranes and washed with saline. Approximately 60 g of placental tissue were cut and digested in a 37°C water bath using 0.125% trypsin (Sigma) and 0.02% deoxyribonuclease-I (Sigma) in Dulbecco modified Eagle medium (DMEM; Life Technologies, Inc.) three times for 30 min each time. Supernatant was collected each time, centrifuged at 37°C and 1700 × g for 10 min, and resuspended. The pooled supernatant was then passed through a nylon gauze filter (pore size, 200 μm), loaded onto a Percoll (Sigma) gradient (5–70% at step increments of 5%), and centrifuged at 2500 × g for 20 min to separate different cell types. Cytotrophoblasts between the density markers of 1.049 and 1.062 g/ml were collected and washed using DMEM and centrifuged at 1700 × g for 10 min. The precipitate was then diluted using DMEM culture medium containing 10% fetal bovine serum (Sigma) and 1% antibiotic solution (Sigma). The cells were plated in 24-well plates at a density of 10⁶ cells/well (for ELISA measurements and immunostaining) or in 60-mm dishes at a density of 10⁷ cells/dish (for protein extraction). Cells were then cultured for 72 h at 37°C in 5% CO₂ and 95% O₂. Cells fused together to form a syncytium. The purity of the cell preparation was assessed by histochemical staining for cytokeratin (DAKO Corp.), an epithelial cell lineage marker, or vimentin (DAKO Corp.), a mesenchymal cell lineage marker. Cells were counterstained with Carrazzi hematoxylin. Immunohistochemical analysis showed that approximately 90–95% of placental cells were cytokeratin-positive, as expected, with a minority of vimentin-stained fibroblast contamination. The percentage of purity for each cell type was quantified based upon counting each cell type and expressing the value as a percentage of the total number of cells in each field of view. The percentage of purity for duplicate wells of the same treatment, with five sample field views per well, were then averaged and reported.

Magnetic Purification of Placental Trophoblast Cells

Placental trophoblast cells were purified using a magnetic purification technique as described previously [26]. Briefly, following cell culture, the cell suspension was centrifuged at 1500 × g for 5 min, resuspended, and allowed to sit in 1 ml of magnetic cell sorting buffer (MACS; PBS, 0.25% bovine serum albumin [BSA], and 2 mM ethylenediaminetetra-acetic acid [EDTA]), 10% fetal calf serum, and 10% human serum for 5 min. Cells were then centrifuged at 1500 × g for 5 min and resuspended in 60 μl of MACS buffer for each 10 million cells. Subsequently, cells were incubated with 20 μl of antifibroblast microbeads and 20 μl of anti-CD45 microbeads (Miltenyi Biotec) for each 10 million cells for 30 min at room temperature. The column was washed on the magnet with 3 ml of MACS buffer. Five milliliters of MACS buffer were added to the tube containing the cells, and after centrifugation at 1500 × g for 5 min and resuspension in 3 ml of MACS buffer, the cell suspension was allowed to run through the column. The fibroblast/CD45-negative fraction containing pure trophoblast cells was collected, and the column was then washed twice with 3 ml of MACS buffer to complete collection. Next, the column was pulled out of the magnet and washed with 5 ml of MACS buffer to collect the fibroblast/CD45-positive cells, which included the leukocyte and fibroblast fraction. Each fraction of cells was then counted, and an equal number of pure trophoblast cells (10⁶), CD45-positive cells (10⁶), and the mixture of both (10⁶) were seeded in separate wells. Immunocytochemistry confirmed the successful removal of the CD45⁺ cells through magnetic purification. A fraction of the mixed cells was also placed in the magnet without being incubated with an antibody and allowed to flow through the column. This was then plated and used as control for the effect of the magnetic field alone on the viability of the cells. No difference was observed between the response from mixed cells plated without magnetic field influence and those placed in the magnet and then plated. This indicates that the magnetic field itself had no effect on the viability and activity of the cells. Cell viability was also tested using the trypan blue exclusion test, which showed no sign of cell lysis following magnetic purification.

Treatment of Placental and Immune Cells

Supernatant from L. rhamnosus GR-1 culture was prepared as previously described [11]. Briefly, the bacteria were grown in MRS (de Man, Rogosa, and Sharpe) broth to an optical density of 1.5 at 600 nm and centrifuged for 10 min at 6000 × g at 4°C. The supernatant was then filtered (pore size, 0.22 μm) to remove any residual bacteria. After 72 h of trophoblast cell culture, cells were washed using DMEM medium and serum starved for 12 h. Cultured cells were then divided into the following groups: 1) no treatment, 2) treatment with LPS alone (200 ng/ml for protein measurements or 100 ng/ml for cytokine measurements) starting 24 h after the beginning of cell starvation, 3) treatment with L. rhamnosus GR-1 supernatant alone (1:20 dilution) starting 12 h after the beginning of cell starvation for a duration of 12 h, and 4) pretreatment with L. rhamnosus GR-1 supernatant (1:20) starting 12 h after the beginning of cell starvation for a duration of 12 h, followed by treatment with LPS (200 or 100 ng/ml). After addition of LPS or GR-1, these treatments remained in the culture plates for the duration of the culture period. Preliminary time-course and dose-response studies established the optimal GR-1 supernatant and LPS concentration and the time of treatments. These four treatments are referred to in the present study as the basic treatments, because they were the control groups used for comparison with the effects of either a chemical inhibitor or a neutralizing antibody. The inhibitor treatments were as follows: 1) JAK inhibitor (10⁻⁵ M; CalBiochem) added 9 h after the start of cell starvation to each of the four basic treatments (3 h before lactobacilli treatment), 2) MAPK14 inhibitor (SB203580; 10⁻⁵ M; CalBiochem) added 21 h after the start of cell starvation to the four basic treatments (3 h before LPS addition), and 3) CSF3 neutralizing antibody (100 μg/ml; R&D Systems) added at the same time as L. rhamnosus GR-1 supernatant. The concentrations and time points used for the inhibitors were based on manufacturer's recommendations and preliminary studies. CSF3 neutralizing antibody (10 μg/ml) neutralizes 50% of the bioactivity of 0.125 ng/ml of recombinant human CSF3. In our cell culture model, 100 μg/ml of this antibody were able to completely neutralize the highest concentration of CSF3 without any sign of cell death or toxicity. Lactate dehydrogenase analysis revealed less than 1% cytotoxicity in the culture media of the cells treated with this antibody compared to 100% LDH release by Triton X-treated cells. Trypan blue exclusion test also showed no sign of cell lysis after treatment with the antibody. Controls for these experiments included culture medium or the inhibitors alone as well the mouse immunoglobulin (Ig) G for CSF3 neutralizing antibody. All the inhibitor and blocking agents were dissolved in dimethyl sulfoxide (DMSO). DMSO alone had no effect on the cells. No effect on the expression of cytokines or proteins was found in controls.

Immunocytochemistry

Immediately after the culture period, the cells were incubated in 1 ml of 4% paraformaldehyde (PFA) for 10 min. They were then washed twice with PBS for 5 min each time and incubated with 0.02% Triton X for 5 min. Next, cells were again washed twice with PBS for 5 min each time, after which three or four drops of DAKO Protein Block Serum were added to each well and the cells put on an orbital shaker for 30 min. The cells were incubated with CD68 (1:100; Novocastra Labs), cytokeratin (1:200; DAKO), and negative mouse IgG (1:200; Santa Cruz Biotechnology) primary antibodies separately overnight at 4°C to stain for macrophages and cytotrophoblast cells. After washing twice with PBS for 5 min each time, cells were incubated with horseradish peroxidase (HRP)-labeled anti-mouse IgG secondary antibody (1:300; DAKO) and again washed twice with PBS. Seven drops of DAKO Labeled StreptAvidin-Biotin 2 Streptavidin-HRP were added to each well for 30 min, and cells were subsequently washed twice with PBS for 5 min each time. 3,3′-Diaminobenzidine (2 ml; DAKO) was added to each well for a brief exposure. Hematoxylin (1 ml; Sigma) was added and then replaced with water. Images were assessed using a Leica DMRXE microscope (Leica Microsystems). Different fields were observed for each staining, and a representative image was photographed for each with a Sony DXC-970 MD 3CCD color video camera. The percentage of purity of each cell type was quantified based upon counting each cell type and expressing the value as a percentage of the total number of cells in each field of view. The percentage of purity for duplicate wells of the same treatment, with five sample field views per well, was then averaged and reported.

Fluorescence-Activated Cell Sorting Analysis

Fluorescence-activated cell sorting (FACS) analysis was used to identify the presence of any macrophages in the primary cell culture. After the culture, tubes of 10⁷ cells were suspended in 4% PFA and stored at 4°C until analyzed. The cells were centrifuged at 1800 × g for 5 min. DAKO Protein Block Serum (1 ml) was added to each tube, and cells were all resuspended together. Cells were then divided between four Eppendorf tubes and centrifuged at 1300 × g for 5 min. Perm solution (1:10, 500 μl) was added to each Eppendorf tube, and cells were incubated for 15 min. Cells were then centrifuged and resuspended in 200 μl of perm solution (1:10). The four Eppendorf tubes were then incubated with appropriate antibodies as follows: 1) control untreated cells, 2) CD68 antibody (1:100; Novocastra Labs), 3) yoyo antibody (a nucleic acid stain with high affinity to double-stranded DNA used as a measure of the total number of cells regardless of their viability status; Invitrogen), and 4) combination of yoyo and CD68 antibody. FACS analysis was then performed on the cells using the FACS analysis facility at the Samuel Lunenfeld Research Institute of the Mount Sinai Hospital.

Protein Extraction and Western Blot Analysis

Preliminary studies established the time point at which an optimal expression of the phosphorylated proteins occurred. Cells from 60-mm dishes were scraped off with cold radioimmunoprecipitation assay (RIPA) lysis buffer containing Mini EDTA-free protease inhibitors (Boehringer Mannheim Biochemicals) at 20 min and 8 h following LPS treatment for measurement of the phospho-STAT3, phospho-MAPK14, and CSF3R proteins, respectively. They were then mixed on a vortex shaker three times, 20 min apart, and centrifuged for 30 min at 4°C and 12 000 rpm. The supernatant was then collected. Protein concentration was measured by Bradford assay using a protein assay kit (Bio-Rad Laboratories, Inc.) with BSA as standard. The samples were then stored at −80°C until further analysis. Twenty-five micrograms of the protein extracts were mixed with specific volumes of RIPA, and 10% loading buffer was added to bring the total volume to 20 μl. Samples were then incubated at 55°C for 10 min and subjected to standard 10% SDS-PAGE. After electrophoresis, proteins were transferred to nitrocellulose membrane (Bio-Rad Laboratories, Inc). The blots were blocked overnight at 4°C with 5% nonfat milk (Nestle) in PBS with 0.1% Tween-20 (Bio-Rad Laboratories, Inc.). Membranes were incubated separately for 1 h with monoclonal rabbit anti-human phospho-STAT3 (Tyr705) antibody (1:1000; Cell Signaling Technology) to detect phospho-STAT3 Tyrosine705 (STAT3-Tyr705), STAT3 antibody (1:500; Cell Signaling Technology) to detect total STAT3, CSF3R antibody (1:500; R&D Systems) to detect CSF3R receptor expression, monoclonal mouse anti-human phospho-MAPK14 (Thr180/Tyr182) antibody (1:2000; Cell Signaling Technology) to detect MAPK14 phosphorylation, and MAPK14 antibody (1:500; Cell Signaling Technology) to detect total MAPK14. The blots were washed five times for 5 min each time with PBS plus 0.1% Tween-20 and incubated for 1 h with anti-rabbit IgG coupled to HRP secondary antibody in the case of phospho-STAT3 (Tyr705; 1:5000; Amersham Pharmacia Biotech) and with total STAT3 (1:5000) as well as total MAPK14 (1:5000) and an HRP-labeled anti-mouse IgG (Amersham Pharmacia Biotech) in the case of phospho-MAPK14 (Thr180/Tyr182) (1:5000) and CSF3R (1:3000). The antibodies revealed signals at 86, 43, and 92 kDa for phospho-STAT3/total STAT3, phospho-MAPK14/total MAPK14, and CSF3R, respectively. The intensities of protein signals were measured by scanning (6200C scanner; Hewlett Packard Co.) and analyzed by using Scion Image software (Version 4.0.2; Scion Co.). All densitometric comparisons were made in the linear range of band intensity. To standardize for protein loading, the blots were stripped and reprobed for ACTB housekeeping protein. The primary antibody was a monoclonal mouse anti-human ACTB antibody (1:7000; A-5316; Sigma), and the secondary antibody was HRP-labeled anti-mouse IgG (1:7000; Amersham Pharmacia Biotech). Blots were incubated for 1 h with each antibody and then exposed to Kodak X-OMAT blue film (PerkinElmer) for 1 min. The anti-ACTB antibody revealed a strong, 43-kDa band. Optical densities of phospho-STAT3, phospho-MAPK14, and CSF3R were divided by the optical density values for ACTB to adjust for protein loading, and these relative optical densities were then used for analysis.

Cytokine Measurements

Culture media were collected 8 h after LPS addition. Levels of TNF and CSF3 were measured using commercially available quantitative ELISA kits (Cayman Chemical for TNF and R&D Systems for CSF3) according to the manufacturer's instructions. The minimum detectable limits of the assays were 1.5 and 20 pg/ml for TNF and CSF3, respectively. The intra-assay coefficients of variation were less than 5% for TNF and less than 3% for CSF3. All samples from each experiment were run on the same assay plate.

Statistical Analysis

Results are expressed as the mean ± SEM. Comparisons between the data were performed using Sigma Stat (Jandel Scientific Software). The criterion for statistical significance was set at P < 0.05. Two-way ANOVA (for comparison across treatments considering the two fetal sexes) or three-way ANOVA (for comparison across treatments and between the two fetal sexes considering different cell types in magnetic purification technique), followed by Student-Newman-Keuls test, was used. One-way ANOVA (for comparison across treatments) was used in the rest of the experiments, in which one cell type and fetal sex were examined.

RESULTS

Effect of L. rhamnosus GR-1 on CSF3 Output in Human Placental Trophoblast Cells

The CSF3 concentrations measured in cultures treated with LPS alone, L. rhamnosus GR-1 supernatant alone, and the combination of both treatments were compared between pregnancies carrying a male or a female fetus. Compared to control, cultures treated with LPS, L. rhamnosus GR-1, and the combination of both treatments increased CSF3 output by 233.1-, 588.7-, and 797.1-fold, respectively, in cells from placentae of female fetuses only (P < 0.001, n = 6) (Fig. 1). The increase in CSF3 output in pregnancies with a female fetus was 47.1-, 65.6-, and 65.7-fold, respectively, more than those in pregnancies with a male fetus (P < 0.001), which did not show any detectable changes with any of the treatments (P > 0.05, n = 5).

FIG. 1.

Effect of fetal sex and L. rhamnosus GR-1 supernatant on concentration of anti-inflammatory CSF3 in LPS and L. rhamnosus GR-1-stimulated placental trophoblast cells. Histogram shows CSF3 concentration (fold-increase relative to control) for different treatment groups in media from placental trophoblast cell cultures. Results are presented as the mean ± SEM and are relative to control (n = 5 males and 6 females). Concentration of CSF3 in the media of the control cultures was 231 pg/ml. Asterisks indicate statistical difference (***P < 0.001). Different letters indicate statistical significance (P < 0.001) within gender groups (male: a; female: a′, b′, and c′) as determined by two-way ANOVA.

Effect of L. rhamnosus GR-1-Induced Up-Regulation of CSF3 on Activation of the JAK/STAT and RTK-MAPK Pathways in Human Placental Trophoblast Cells

The GR-1 treatment stimulated phosphorylation of STAT3-Tyr705 and MAPK14 in cultures from placentae of female fetuses, but CSF3 neutralizing antibody had no effect on GR-1-stimulated phosphorylation (P > 0.05, n = 4 for STAT3-Tyr705 phosphorylation and n = 5 for MAPK14 phosphorylation) (Fig. 2A). No change was found in the expression of total STAT3 and p38 across treatments. Comparisons were also made for CSF3 concentrations using ELISA between the four basic treatments and respective treatments with addition of either JAK inhibitor or MAPK14 inhibitor (SB203580). CSF3 neutralizing antibody completely absorbed and neutralized the CSF3 secreted by placental trophoblast cells of the female fetuses, which have been shown to have a larger output than males (P < 0.001, n = 7, data not shown). A significant decrease in CSF3 output was observed in cultures of placenta from female fetuses containing either the JAK inhibitor (P < 0.001, n = 5) (Fig. 2B) or the MAPK14 inhibitor (P < 0.001, n = 5) (Fig. 2B), suggesting that L. rhamnosus GR-1 supernatant induces CSF3 release in human placental trophoblast cells through both the JAK/STAT and MAPK14 pathways.

FIG. 2.

The relationship between CSF3 and the MAPK and JAK/STAT pathways in human placental trophoblast cells of the female fetuses. A) Representative Western blots of phospho-STAT3-Tyr705 (86 kDa, n = 4), total STAT3 (86 kDa, n = 4), phospho-MAPK14 (43 kDa, n = 5), total MAPK14 (43 kDa, n = 5), and ACTB (43 kDa) expression under control, LPS, L. rhamnosus GR-1 supernatant, and LPS plus L. rhamnosus GR-1 supernatant as well as addition of a CSF3 neutralizing antibody to these four basic treatments. Western blots were normalized to ACTB. Histograms show CSF3 (B; n = 5) and TNF (C; n = 4) concentration (fold-increase relative to control) for different treatment groups in media from placental trophoblast cell cultures. Results are presented as the mean ± SEM and are relative to control. Different letters indicate statistical significance (P < 0.001) as determined by one-way ANOVA.

No significant differences were found in the level of TNF (P > 0.05, n = 4) (Fig. 2C) between cultures from placentae of the female fetuses treated with the four basic treatments and the respective treatments containing the CSF3 neutralizing antibody. The combination of LPS and CSF3 neutralizing antibody served as a positive control to demonstrate that cells treated with LPS alone or with LPS and CSF3 neutralizing antibody were able to increase TNF output to the same level, confirming that the antibody does not have any negative effect on the cells.

Effect of L. rhamnosus GR-1 on CSF3R in Human Placental Trophoblast Cells

Western blot revealed no significant difference in the expression of endogenous CSF3R (P > 0.05, n = 8 for males and n = 5 for females) (Fig. 3A) between the trophoblast cells from male and female fetuses. The combination of L. rhamnosus GR-1 with LPS, however, increased the expression of CSF3R in females by 2.7-fold (P = 0.007, n = 5) (Fig. 3, B and C). When the results from placental cultures of males and females fetuses were combined (n = 13), L. rhamnosus GR-1 and the combination of L. rhamnosus GR-1 and LPS up-regulated CSF3R expression by 1.7- and 2.0-fold, respectively (P = 0.001) (Fig. 3D).

FIG. 3.

Effect of fetal sex on CSF3R expression in placental trophoblast cells. A) Relative optical density (ROD) of CSF3R expression under control conditions in placentae of male or female fetuses normalized to ACTB (P > 0.05, n = 8 males and 5 females). B) Representative Western blot of CSF3R (92 kDa) and ACTB (43 kDa) expression under LPS, L. rhamnosus GR-1 supernatant, as well as LPS plus L. rhamnosus GR-1 supernatant treatments of placental trophoblast cells from pregnancies carrying a male or a female fetus. C and D) ROD of CSF3R Western blot normalized to ACTB for pregnancies carrying a male or female fetus (C) and total samples (D; n = 13). Results are presented as the mean ± SEM and are relative to control. Different letters indicates statistical significance within gender groups (C; male: a; female: a′ and b′; P = 0.007 by two-way ANOVA) and across treatments (D; a and b; P = 0.001 by one-way ANOVA).

Role of Immune and Trophoblast Cells in Production of Cytokines

To investigate the role of immune and trophoblast cells separately on cytokine production, CD45⁺ cells were magnetically purified from pure trophoblast cells using the technique described by Stenqvist et al [26]. Immunocytochemistry and FACS analysis both showed approximately 2% macrophage (CD68⁺) presence in the mixed culture set (Supplemental Fig. S1, available online at www.biolreprod.org), which consisted of both pure trophoblast and CD45⁺ cells. Immunocytochemistry confirmed the successful removal of these CD45⁺ cells through magnetic purification. CD45⁺ cells, pure trophoblast cells, and the mixed cells were then plated in separate wells, and the effect of different treatments was investigated in these three populations. This differed from the previous experiments, in which we examined only the mixed cell population. LPS, L. rhamnosus GR-1, and the combination of both treatments increased CSF3 output relative to control values only in the placentae of female fetuses (P = 0.048, n = 4) (Fig. 4C). This increase was only observed in the mixed culture set and was far greater than that in purified trophoblast and purified CD45⁺ cells with LPS (P = 0.015, n = 7), L. rhamnosus GR-1 treatment alone (P = 0.004, n = 7), and the combination of both treatments (P = 0.004, n = 7) when both fetal sexes were considered (Fig. 4A) or in placental trophoblasts from female fetuses alone (P < 0.001, n = 4) (Fig. 4C). A higher output of CSF3 from CD45⁺ cells compared to pure trophoblast cells (P < 0.001) also was found. This result implies that the small portion of immune cells (leukocytes) that exists in our culture system has a profound effect on the output of cytokines when interacting with trophoblast cells. No change was observed across treatments in the three culture sets of trophoblast cells from placentae of the male fetuses (P > 0.05, n = 3) (Fig. 4B).

FIG. 4.

CSF3 concentration in mixed, pure trophoblast, and CD45⁺ culture sets. CSF3 concentrations (fold-increase relative to control) for control and for LPS-, GR-1-, and GR-1/LPS-treated cells from (A) both placentae of male or female fetuses (n = 7), (B) placentae of male fetuses (n = 3), and (C) placentae of female fetuses (n = 4) are shown. Results are presented as the mean ± SEM and are relative to control. Asterisks indicate statistical difference (*P = 0.015, **P = 0.004, ***P < 0.001). Different letters indicate statistical significance (mixed cells: a, b, and c; pure trophoblasts: a′; CD45⁺ cells: a″) as determined by three-way ANOVA as follows: A) relative to control, P < 0.001 for LPS and P = 0.002 for GR-1 alone as well as combined with LPS; B) P > 0.05; and C) P = 0.048.

DISCUSSION

In these experiments, we found that L. rhamnosus GR-1 caused a marked up-regulation in the output of the immunomodulatory cytokine CSF3 only in trophoblast cells of placentae carrying a female fetus and not in those carrying a male fetus. Our findings extend those of other in vitro studies in which the same L. rhamnosus GR-1 supernatant increased CSF3 output in mouse macrophages [11] as well as in the JEG-3 human trophoblastic cell line [18]. Because low levels of CSF3 have been associated with increased incidence and severity of infection in preterm infants [4], our results support the protective nature of L. rhamnosus GR-1 in the intrauterine tissues. LPS also caused an increase in the output of CSF3, as it did for IL10 in our previous studies [17], which may counter the local effects of inflammation. Here, we report sex-specific differences in the ability of L. rhamnosus GR-1 to increase the output of CSF3, with cells from placentae of female fetuses being strikingly more responsive to LPS, L. rhamnosus GR-1, and the combination of both treatments. We have previously shown that LPS increased the output of TNF more in the placental trophoblast of the male fetuses and that L. rhamnosus GR-1 supernatant enhanced the expression of protective PGDH protein in placental trophoblast cells from only female fetuses [17]. Because preterm birth is less frequent in women carrying a female fetus [21, 22], and because females are more effective at mounting an immune response than males [19, 20], we propose that lactobacilli might be able to contribute to the lower incidence of preterm birth and neonatal sepsis observed with female fetuses. Currently, the mechanism behind the sexual dimorphism is unknown. One explanation could be that human chorionic gonadotropin, which is a marker of syncytialization, is present with greater abundance in maternal serum and cord blood of women carrying female fetuses than in those carrying males [27]. Human chorionic gonadotropin inhibits LPS-induced septic shock as well as Th1 cytokine levels in mice [28]. In addition, the enzymes responsible for regulation of Th1 and Th2 cells could themselves be controlled by an X-linked gene, which could explain the sexual dimorphism in cytokine output pattern observed in our studies. Epigenetic modifications in target gene promoters, such as increased histone acetylase transferase activity, histone acetylation, and DNA demethylation, could also affect receptor regulation [29], leading to sexual dimorphism observed in cytokine output. Interferon-tau, a product of an autosomal gene known to down-regulate prostaglandins and up-regulate CSF3, is also secreted in 2-fold greater abundance by bovine blastocysts of females compared to those of males [30], providing a further possible explanation for the sexual dimorphism observed in CSF3 output during our experiments. To investigate the sexual dimorphism, promoters on the gene for CSF3 could be examined for any evidence of silencing that might explain the lack of effect in placentae of the male fetuses for this cytokine.

Previous studies suggest that the anti-inflammatory effects of CSF3 are mediated through activation of the JAK/STAT and MAPK pathways in both mouse macrophages [11] as well as the JEG-3 human trophoblastic cell line [18]. Currently, the relationship between CSF3 and the MAPK and JAK/STAT pathways in human placental trophoblast primary cell cultures is unknown. In the present study, we determined that STAT3-Tyr705 and MAPK14 phosphorylation did not change when CSF3 was blocked, implying that CSF3 is downstream of the JAK/STAT and MAPK pathways. However, L. rhamnosus GR-1-induced CSF3 output was significantly attenuated in the presence of a JAK inhibitor and a MAPK14 inhibitor (SB203580), demonstrating that the JAK/STAT and MAPK pathways are required for CSF3 output. In addition, no difference was found in the phosphorylation of STAT3-Tyr705 and MAPK14 in placental cells from the two fetal sexes (data not shown). Yet, a sex-specific difference was found in the output of CSF3 downstream of these signaling pathways. It has been shown previously that CSF3 is mainly regulated via posttranscriptional mechanisms [31]. For instance, CSF3 transcripts contain conserved AU-rich sequences in the 3′ untranslated region, which are thought to be signals for rapid degradation and explain the very short half-life of CSF3. However, different factors, such as IL17, could alter this half-life and stabilize CSF3 mRNA [32]. Therefore, even though no sexual dimorphism was observed in the phosphorylation of the factors that lead to transcription of CSF3 in our studies, other differences could exist between the two fetal sexes in the production of IL17 or other factors that help stabilize the mRNA of this cytokine posttranscriptionally and account for the differences we have observed. Interestingly, we have previously shown that these inhibitors had no effects on TNF production in response to LPS, suggesting that activation of STAT3 and MAPK14 is required for preferential output of CSF3 to TNF in female trophoblast cells and that the female fetus has anti-inflammatory preferential responses to bacterial pathogens. Many growth factors and cytokines, including IL10, have been shown to induce activation of STAT3 and MAPKs. However, the molecules and mechanisms involved in the preferential output of CSF3 in trophoblast cells from female placentae are yet to be explored.

To further investigate the source of CSF3, we investigated whether the effect of L. rhamnosus GR-1 supernatant on cytokine production and the subsequent response was from the trophoblast cells, immune cells, or the combination of both, through magnetic purification of the trophoblast cells. Most immune cells that are positive for CD45 are washed away through our culture process, but macrophages tend to adhere to the bottom of the culture dish. By measuring CSF3, we observed that the effect of LPS, L. rhamnosus GR-1, and the combination of both treatments provide an additive effect requiring both pure trophoblast cells as well as macrophages. Therefore, in an in vivo setting, both cell types act together at the maternal-fetal interface to respond to bacterial endotoxins or probiotic organisms. In particular, regarding CSF3, pure trophoblast cells showed minimum output of this cytokine, and macrophages played a more important role in its release. We have also found that the sexual dimorphism pattern in cytokine output only exists in our mixed culture sets. Upon removal of the immune cells, the sexual dimorphism in CSF3 production, which is secreted in significant amounts from macrophages, was abolished. The sex difference in cytokine output pattern observed in our mixed culture set could result from the fact that macrophages express receptors for gonadal hormones (androgens and estrogen). Estrogen usually has Th2 cell activity, whereas androgens foster Th1 cell activity [33]. Although primary cultures impose a limitation in terms of possible cell contamination and decreasing cell viability, this model serves as a better representation of the true intrauterine environment than cell lines.

No difference was found in the expression of endogenous CSF3R between placental trophoblast cells of male and female fetuses in control cultures; therefore, the only sexual dimorphism was observed in the actual output of the substrate of these receptors (CSF3R), which would affect subsequent signaling pathways. The combination of L. rhamnosus GR-1 with LPS stimulated CSF3R expression in placental trophoblast cells of female fetuses. This could be indicative of increased binding of CSF3 to its receptor, facilitating activation by L. rhamnosus GR-1 of anti-inflammatory responses to infection. An alternative explanation for the up-regulation of CSF3R in placental cell cultures is that L. rhamnosus GR-1-induced CSF3 may act to increase its own receptor. This effect has been reported in murine bone marrow cells, in which treatment with CSF3 led to an increase in CSF3R mRNA [34].

To our knowledge, the present study is the first to investigate the effect of L. rhamnosus GR-1 supernatant on CSF3 output in both placental trophoblast as well as immune cells. Based on our previous [17] and current study, we propose that L. rhamnosus GR-1 supernatant acts through the JAK/STAT and MAPK pathways to increase the anti-inflammatory IL10 and CSF3. GR-1 supernatant also independently down-regulates TNF, which is known to play a key role in infection-mediated preterm birth. Proinflammatory cytokines such as TNF increase prostaglandin synthesis and matrix metalloproteinase-9 expression, leading to rupture of membranes, uterine contractions, and preterm labor. Down-regulation of TNF output by lactobacilli will have a protective role and, therefore, significantly interfere with these processes [35]. The increase in the production of CSF3 with L. rhamnosus GR-1 supernatant is only observed in pregnancies carrying a female fetus and, potentially, could provide a mechanism for the lower incidence of preterm births in these pregnancies. We also conclude that both trophoblast and immune cells are necessary for the initiation of an anti-inflammatory cascade in the intrauterine environment. The exact nature of the active agent of lactobacilli is still unknown, although studies have fractionated the supernatant of a different strain (Lactobacillus plantarum) to protein and polysaccharide components and demonstrated that only the peptide fraction with the molecular weight of 8.7 kDa contributes to stimulation of the anti-inflammatory response and down-regulation of the proinflammatory cytokine TNF [36]. Up-regulation of the anti-inflammatory cytokines could be an innate protective mechanism in gestational tissues implemented by the placenta in a healthy urogenital environment to protect against infection. L. rhamnosus GR-1 has been shown to have beneficial effects for the vaginal microflora [37] and, therefore, may serve as a suitable alternative to antibiotics in preventing infection/inflammation-mediated preterm birth.