Variations in cytokine mRNA expression during normal human pregnancy

Abstract

Epidemiological data provide evidence that disease activity of T cell-mediated, organ-specific autoimmune diseases is reduced during pregnancy. Although there are several experimental animal studies on the effect of pregnancy on the immune system, the situation in humans is less clear. We therefore performed a prospective analysis of cytokine mRNA expression in whole blood by a new on-line reverse transcriptase-polymerase chain reaction technique and of serum hormone levels during pregnancy in healthy women. The control group included age-matched non-pregnant healthy women. Quantitativecytokine mRNA expression revealed significantly reduced IL-18, interferon-gamma (IFN-γ), and IL-2 mRNA levels in the first and second trimester in pregnancy compared with non-pregnant women. No difference between groups was detected for tumour necrosis factor-alpha (TNF-α) mRNA. IL-4 and IL-10 mRNA were detected at low levels in only 20% of pregnant women and were reduced to a statistically significant extent in the second and third trimester compared with the control group. Changes in IL-18 mRNA expression correlated inversely with serum values for human choriogonadotropin (HCG) and IL-10 serum levels correlated with increases in serum 17β-oestradiol levels. These data indicate immunomodulatory effects of pregnancy at the cytokine level which may be related to the variations in the clinical course of organ-specific, T cell-mediated autoimmune diseases during pregnancy.

INTRODUCTION

Gestation is associated with a transient depression of maternal cell-mediated immunity to protect the semi-allogeneic embryo from rejection. The hallmark of this immune tolerance is a profound modulation of T cell responses, best characterized by a shift from a Th1 (IL-2, interferon-gamma (IFN-γ)) to a Th2 (IL-4, IL-10) type cytokine response [1–3]. It was demonstrated that a failure in the generation of Th2-type cytokine responses is associated with recurrent abortions, complications [4], and poor outcome of pregnancy. Hormonal changes during pregnancy as well as protective factors produced by the conceptus itself are regarded as important mediators of this transition [5,6], but the complex interactions between the endocrine system and immunological reactions in humans are still not well understood. Recently, it was demonstrated that IL-10 and leukaemia inhibitory factor (LIF), both Th2-type cytokines, are produced by the decidua and are required for embryo implantation and development [5,7–9].

Clinical disease activity of several T cell-mediated autoimmune disorders, such as rheumatoid arthritis, Crohn's disease or multiple sclerosis, is reduced during pregnancy but flares in the post-partum period [10–14]. As disease activity in these disorders is often associated with increased Th1-type cytokine responses in the blood, it was suggested that a cytokine shift during pregnancy may be a protective process.

There are only few ex vivo studies on cytokine production in human pregnancy, which yielded conflicting results and were unable to detect the expected early Th1/Th2 shift during pregnancy [15–17]. As these studies have applied different methodologies to measure cytokine production (serum levels versus in vitro stimulated blood cells), this could have affected the partly opposing results. In addition, time of sampling during pregnancy may be a critical parameter for the interpretation of different results [9,18,19]

The recent optimization of quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) technology [20] enabled us to perform a cross-sectional study on the expression of major cytokines involved in the Th1/Th2 differentiation process (IL-2, IL-4, IL-10, IL-18, tumour necrosis factor-alpha (TNF-α), and IFN-γ mRNA) in blood samples from normal pregnant women.

SUBJECTS AND METHODS

Study participants

This study aimed at the quantification of cytokine mRNA expression in healthy pregnant women and age-matched controls in comparison with pregnancy-associated serum hormone levels. Thirty-nine women during different stages of pregnancy and 10 healthy non-pregnant women who did not use oral contraceptives volunteered to participate in this study (Table 1). Written informed consent was obtained from each participant and the study was approved by the local ethics committee. EDTA blood (20 ml) was obtained by venepuncture between 8 and 10 a.m. and immediately processed. Aliquots of 500 μl EDTA blood were snap-frozen in liquid nitrogen within 1 h after venepuncture and stored at − 80°C until used for RNA preparation.

Table 1.

Characteristics of study participants

RNA isolation and reverse transcription

RNA was isolated as described in details elsewhere [20]. For reverse transcription RNA dissolved in 33 μl dd H₂O was mixed with 3 μl oligo dT (500 μg/ml; Pharmacia-Biotech, Freiburg, Germany), incubated for 10 min at 70°C and chilled on ice. Twelve microlitres 5 × first strand buffer (Life Technologies GmbH, Karlsruhe, Germany), 6 μl 0.1 m DTT (Life Technologies) and 3 μl dNTPs (10 mm each; Pharmacia-Biotech) were added. After incubation for 2 min at 42°C, 3 μl Superscript II reverse transcriptase (200 U/μl; Life Technologies) were added and the reverse transcription was performed for 50 min at 42°C followed by a denaturation step for 15 min at 70°C.

PCR amplification and quantification

PCR reactions were performed as previously described [20]. β-actin was used as a standard for equal mRNA input in each assay. Optimized primer combinations for the quantification of cytokine mRNA expression were used as listed in Table 2. After reverse transcription the overall number of β-actin mRNA molecules was quantified first and samples were then diluted to contain one million β-actin mRNA molecules per aliquot and frozen for further cytokine analysis. Serial 10-fold dilutions of β-actin or cytokine plasmid standards were used as controls for quantification. The oligonucleotide sequences used for target detection and quantification are listed in Table 2. Input numbers of target sequences were quantified as molecules of cytokine mRNA per one million molecules of β-actin mRNA.

Table 2.

Oligonucleotides used for target detection and quantification

Quantification of IL-10 and soluble vascular cell adhesion molecule-1 in serum samples

For the measurement of IL-10 and soluble vascular cell adhesion molecule-1 (sVCAM-1) in serum samples we used commercially available ELISA kits (IL-10 high sensitivity; R&D Systems, Wiesbaden, Germany; VCAM-1, Biosource, Ratingen, Germany). The sensitivity of the IL-10 assay for serum samples was 4 pg/ml with an intra- and interassay variability of < 6%. The sensitivity of the sVCAM-1 assay was 500 pg/ml, the intra- and interassay variability was < 5%. Serum aliquots of 500 μl were thawed and added to precoated 96-well plates at the appropriate recommended dilution. The ELISA assays for the quantification were performed according to the instructions of the manufacturers.

Measurement of 17β-oestradiol, human choriogonadotropin and prolactin

Serum concentrations of 17β-oestradiol, human choriogonadotropin (HCG) and prolactin were quantified using two-site sandwich immunoassays and automated direct chemiluminescence technology according to the instruction of the manufacturer (Chiron Diagnostics GmbH, Fernwald, Germany) at the hormone diagnostic laboratory of the Department of Gynaecology.

Statistical analysis

Group comparison was performed by non-parametric Wilcoxon rank sum test which was informative of the intergroup homogeneity of the distributions. Correlation between individual parameters was analysed by Pearson's linear regression.

RESULTS

Whole blood cytokine mRNA expression during pregnancy

On-line quantitative RT-PCR was proved to be reproducible, reliable and very sensitive in the detection of different cytokine expression patterns. Using internal cytokine plasmid standards, the interassay variability for consecutive analyses was < 5% and the efficacy of amplification was similar for all cytokines investigated in this study (Fig. 1). We determined the cytokine mRNA expression in whole blood samples from healthy women during normal pregnancy (1st trimester (median), 10 weeks; 2nd trimester, 21 weeks; 3rd trimester, 31 weeks) in comparison with non-pregnant age-matched females (Table 1). Among the cytokines measured in the blood of all 49 study participants highest mRNA expression levels were observed for IL-4 (median 2966.5, range 104–11 501 mRNA molecules per 10⁶β-actin mRNA molecules), followed by IL-18 (median 538, range 79–1343), whereas the mRNA levels for IFN-γ, IL-2, IL-10 and TNF-α were on average much lower (Fig. 2).

Fig. 1.

Standard curves for individual cytokine reverse transcriptase-polymerase chain reaction (RT-PCR) reactions. Serial 10-fold dilutions of cytokine plasmid standards were amplified and results of the threshold cycle (Ct) are given. For each cytokine and input number of molecules the mean and s.d. from 30 independent experiments are shown. The inter- and intra-assay variability was < 5%. ▪, IL-2; •, IL-4; ▴, IL-10; □, IL-18; ○, IFN-γ; Δ, tumour necrosis factor-alpha.

Fig. 2.

Cytokine mRNA expression during the course of pregnancy. Cytokine mRNA was quantified in target samples which contained an equivalent of 10⁶ molecules mRNA of the housekeeping gene β-actin. Each graph represents the results (mean ± s.d.) from individual cytokines analysed. Abbreviations under the columns indicate: np, non-pregnant; 1.t, first trimester; pp, first week post-partum. Only highly significant differences (P < 0.01) are indicated.

Significantly lower mRNA levels of the Th1-type cytokines (IL-2 and IFN-γ) were observed during pregnancy compared with non-pregnant female controls. The mRNA levels of IL-18 were also significantly reduced during pregnancy. These effects were already prominent in women during the first trimester and continued throughout pregnancy. IL-2 and IFN-γ mRNA levels tended to increase again towards late pregnancy but decreased again immediately after delivery. No significant fluctuation during different stages of pregnancy was observed for IL-18 (Fig. 2a–c). TNF-α mRNA expression in whole blood preparations was not significantly altered during pregnancy, although levels were slightly higher compared with age-matched non-pregnant women (Fig. 2f).

There was also a slightly lower expression of IL-4 and IL-10 mRNA during pregnancy (Fig. 2d,e) but significantly decreased IL-4 mRNA levels were only observed in post-partum women (comparison with all other groups, P < 0.05). The IL-4/IFN-γ ratio (as an indicator of the Th2/Th1 balance) was significantly higher during the first and second trimesters compared with third trimester, early post-partum or with non-pregnant women. This dominant Th2 cytokine profile was even more pronounced if the IL-4/IL-2 ratio was calculated (Fig. 3).

Fig. 3.

Cytokine ratios during pregnancy. The ratios of mRNA expression for IL-4/IL-2 (□) and for IL-4/IFN-γ (▪) as indicator for a Th1/Th2 shift during pregnancy are shown. Note the logarithmic scale of the ordinate. np, Non-pregnant; 1.t, 1st trimester; pp, 1st week post-partum.

We observed increased serum levels of IL-10, which correlated with pregnancy duration, and highest values were detected during the third trimester (Fig. 4a). Soluble VCAM-1 serum levels, as an indicator of immune activation and inflammation, significantly decreased during pregnancy (Fig. 4b).

Fig. 4.

Correlation between pregnancy duration and IL-10 serum levels. Each square indicates one donor. A positive correlation was detected between weeks of pregnancy and IL-10 serum concentration ((a), r² = 0.19; P < 0.018), whereas sVCAM-1 declined during pregnancy ((b), r² = 0.19; P = 0.038).

Correlation of immunological parameters with pregnancy-associated hormones

All women included in this study had a normal course of pregnancy with an expected early increase in HCG and subsequent elevation of 17β-oestradiol and prolactin post-partum. Serum levels of 17β-oestradiol steadily increased during pregnancy and significantly correlated with IL-10 serum levels, whereas no obvious association was detected with α-fetoprotein, prolactin or HCG (data not shown). The sharp decline in IL-18 mRNA expression during the first trimester was associated with an early increase in HCG (Fig. 5).

Fig. 5.

Correlation between hormones and cytokines during pregnancy. The association between human choriogonadotropin (HCG) serum concentrations and IL-18 mRNA expression in whole blood cells revealed an inverse correlation (r² = 0.19; P = 0.038).

DISCUSSION

Initial studies of cytokine production in the blood from pregnant women failed to detect an early decrease of Th1-type cytokine expression [15,16]. Using a highly sensitive on-line quantitative RT-PCR method we were able to detect significantly reduced IL-2, IL-18 and IFN-γ mRNA expression levels already during the first trimester of normal pregnancy. Surprisingly, the mRNA levels for IL-4 and IL-10 also declined during pregnancy but this reduction was only marginal. Calculating the cytokine ratios revealed a shift from a Th1 to a pronounced Th2-type response, which was at the highest level during the second trimester (Fig. 3). These data are in agreement with published results [19] and point to the important role of relative changes during pregnancy rather than absolute levels of a given cytokine. The depression of proinflammatory cytokines was associated with high HCG serum levels. The effect of HCG on cytokine production is not entirely clear [24,25] and there may be other early pregnancy signals [26] which not only have a profound effect on immune regulation at the fetal maternal interphase but also influence the cytokine expression pattern within blood cells.

In addition to the effects on cytokine expression we observed a reduction in circulating sVCAM-1 levels which are known to be associated with disease activity in several organ-specific autoimmune disorders. These observations together with the rise in serum IL-10 levels may explain the beneficial effect of pregnancy on various T cell-mediated, organ-specific autoimmune diseases, such as multiple sclerosis, rheumatoid arthritis or inflammatory bowel disease [12–14]. As the increase in serum IL-10 was not accompanied by high IL-10 mRNA concentrations within the blood cells, the most likely source for this protein was the placenta cytotrophoblasts, as has already been demonstrated by others [7]. Most data were derived from experimental animal studies [21]. The syncytiotrophoblast has been recognized as an important source for Th2-type cytokine production during pregnancy and the failure to generate Th2 cytokines is associated with poor pregnancy outcome [8]. Therefore, high IL-10 levels present in the circulation could modulate cytokine production within blood cells and may even suppress IL-10 mRNA expression at later stages of pregnancy.

It is known from the murine model that the preimplantation and implantation embryo produces and responds already to a variety of growth factors, including Th2-type cytokines [22]. Recently it was shown that supernatants from fertilized human ovum induced IL-10 mRNA production in activated lymphocytes in vitro [6]. These results point to early immunomodulatory effects of pregnancy which direct the immune response in order to tolerate the semi-allogeneic conceptus and allow proper implantation and growth of the fetus. It is not known whether these early events are restricted to the fetal–maternal interphase or whether they are already present in the circulating blood. Studies in mice demonstrated an early increase of various cytokines during the first week of gestation followed by a decrease in IFN-γ mRNA towards late pregnancy [23].

There is direct evidence that the pregnancy-associated hormones progesterone and oestradiol modify the cytokine production pattern of human antigen-specific T cells [27]. Initial data from our laboratory demonstrated that serum from pregnant women reduced the proliferation and IFN-γ production of myelin basic protein-specific T cells and stimulated IL-10 production in these cells [28]. Although several effects of pregnancy-associated molecules on the immune system have been described, it is still an unresolved question which factor is the relevant immunomodulator in vivo. Therefore, it is of importance to characterize further the potent regulatory pathways during pregnancy which affect immune reactions in pregnant women and may be of considerable therapeutic relevance in T cell-mediated autoimmune diseases.