Background

Pregnancy is the result of a fine immunological privilege that allows the fetus to co-habit the maternal uterus, preventing rejection of the fetal allograft [1]. There are epidemiologic and experimental data indicating that intrauterine infection is a pathological condition that disrupts this privilege [2, 3] and is the main etiological factor for the development of Premature Rupture of Membranes (PROM). This pathology complicates about 30–40% of all preterm births [4] and 10% of all pregnancies [5].

Human chorioamniotic membranes form a complex multi-laminated tissue constituted by the amnion whose epithelium is in contact with the amniotic fluid and the choriodecidua that is formed by trophoblasts inter-digitized with the maternal decidua [5, 6]. During an ascending infection, originated in the cervico-vaginal region [7–9], the choriodecidua is the first-line barrier in contact with the pathogens that can cross the membranes and infect the amnion and the amniotic fluid [10]. In response, there is an abnormal production of proinflammatory cytokines, such as interleukin (IL)-1β, tumor necrosis factor alpha (TNFα), IL-6, and IL-8 in the extra-embryonic tissues, i.e., placenta [9] and chorioamniotic membranes [11–14].

In a previous work, using an ex-vivo model, we demonstrated that in vitro stimulation with lipopolysaccharide (LPS) and Streptococcus agalactiae induced a differential response in IL-1β and TNFα secretion by the amnion and choriodecidual tissues [13]. Using the same model, the present study was aimed at: 1) Investigating the secretory profiles of IL-1β, IL-6, TNFα, IL-8, and IL-10 by human chorioamniotic membranes after stimulation with Escherichia coli, a common pathogen in cervical-vaginal infection in humans associated with pregnancy losses [8, 15], fetal cardiac dysfunction [16], neurological injury in preterm infants [17], as well as with preterm delivery and PROM [18]. 2) Disclosing whether or not a specific differential contribution by each membrane exists.

Methods

The 10 pregnant women (37–40 weeks of gestation) studied were from an urban area of Mexico City, 22–35 years old, previously normotensive, without history of diabetes mellitus, thyroid, liver or chronic renal disease, cared for at the Obstetrics Outpatient Service of the Instituto Nacional de Perinatologia. All women had uneventful pregnancies, without evidence of active labor and with neither clinical nor microbiological signs of chorioamnionitis or of lower genital tract infection. Samples were obtained after delivery by elective cesarean section.

All women provided written, informed consent, before collection of samples. The institutional Review Board approved the protocol and the collection and use of the samples and the study was conducted according to the Guidelines on the Practice of Ethical Committees in Medical Research (3rd ed) issued by the Royal College of Physicians of London.

Fetal membrane explants culture

The chorioamniotic membranes were cut at a distance of 5–6 cm from the placental disc, transported to the laboratory in sterile Dulbecco Modified Eagle Medium (DMEM; Gibco BRL, Bethesda, MD), and rinsed in sterile saline solution to remove adherent blood clots. Segments representing all zones of membranes were manually cut into 18 mm diameter discs and held together with silicone rubber rings to be placed on the upper chamber of a Transwell® system (CORNING, New York, NY) from which the original polycarbonate membrane had been previously removed. In this model the upper chamber of the Transwell® system is delimited by choriodecidual tissue and the lower chamber by amniotic tissue, which allowed testing the two compartments independently (Figure 1). A detailed description and validation of this model has been published previously [13]. One milliliter of DMEM (Gibco BRL), supplemented with 10% fetal calf serum (FCS), 1 mM sodium pyruvate, and 1× antibiotic-antimycotic solution (penicillin 100 U/mL, streptomycin 100 μg/mL) (DMEM-FCS) was added to each compartment. The mounted explants were placed in a 12-well tissue culture plate (CORNING, New York, NY) and incubated in 5% CO2 at 37°C for 24 h.

Figure 1
figure 1

Experimental model. The fetal membranes are held together with silicone rubber rings in the upper chamber of a Transwell® device. In this system, the choriodecidua region delimits the upper chamber and the amnion region, the lower chamber.

Full size image

Explants stimulation

To stabilize the explants after manipulation, they were pre-incubated for 24 h in the medium (DMEM-FCS). Subsequently, the explants were washed with saline solution to remove FCS and the remainder of the medium was changed to DMEM with 0.2% lactalbumin hydrolysate (Gibco BRL) and co-incubated with 1 × 106 colony forming units (CFU) of Escherichia coli isolated from cervicovaginal exudates.

Four Each experiment (n = 10) included the following set of chambers, in triplicate: Basal (n = 10), control membranes in which only the medium culture was added to the compartments; Choriodecidua (n = 10), Escherichia coli was added only to the choriodecidua side; Amnion (n = 10), E. coli was added only to the amniotic compartment; Both (n = 10), the bacterium was added simultaneously to both compartments. After 24 h of co-incubation, the medium from the amnion and choriodecidua chambers was collected and centrifuged at 5,000 rpm, 3 min at 4°C, to precipitate Escherichia coli, and the bacterium-free medium of each sample was stored at -70°C until assayed. Protein concentration in all samples was measured according to the Bradford method [19].

Cytokine assays

IL-1β, TNF-α, IL-6, and IL-10 concentrations were quantified using specific DuoSet® enzyme-linked immunosorbent sandwich assays (ELISA) (R & D Systems, Minneapolis, USA). For the IL-1β assay, a standard curve was developed from 4 to 260 pg/mL and the sensitivity was 2 pg/mL; the TNF-α assay was linear from 15 to 960 pg/mL and sensitivity was 5.0 pg/mL; for the IL-6 assay, the curve was linear from 250 to 8000 pg/mL with a sensitivity of 200 pg/mL; and for the IL-10 assay, a standard curve was developed from 1.25 to 2000 pg/mL, with a sensitivity of 1.0 pg/mL. IL-8 was measured using a commercial kit (Amersham Biosciences, Buckinghamshire, UK) according to manufacturer's instructions. The standard curve was linear from 25.6 to 1000 pg/mL and sensitivity was 5 pg/mL.

The final concentration of each cytokine was expressed per microgram of the total protein concentration of each sample. Intra- and inter-assay coefficients of variation were less than 5%. A rigorous quality control program, including external and internal standards for all cytokines, is followed in our laboratory.

Statistical analysis

Since the results did not have a normal distribution, comparisons between the experimental groups and the control were performed using the Kruskal-Wallis one-way analysis of variance on rank tests. A P < 0.05 was considered significant. The data expressed in the text and figures represent the median and the 95% confidence interval limits (95% CI).

Results

Basal secretion of IL-1β was similar in both the amniotic and the choriodecidua compartments (0.8 [0.30–2.89] and 1.73 [0.12–3.27] pg/μg protein, respectively); it increased markedly in the choriodecidua tissue (P < 0.05) regardless of whether Escherichia coli was applied directly to either the choriodecidua compartment (39.95 [26.9–55.6] pg/μg protein) or the amniotic compartment (28.9 [23.0–50.5] pg/μg protein), or simultaneously to both compartments (44.87 [21.87–50.5] pg/μg protein). However, IL-1β secreted by the amnion rose slightly only when the bacterium was directly applied to this compartment (13.9 [11.2–18.3] pg/μg protein) (Figure 2).

Figure 2
figure 2

In vitro secretion of IL-1β in amnion and choriodecidua regions after selective stimulation with Escherichia coli. IL-1β secreted to the culture medium after 24 h of infection with 1 × 106 CFU of E. coli. Data were normalized in function of protein concentration (pg/μg protein) and each bar represents the 95% confidence intervals and the median (solid line) of 10 different experiments. Significant difference between basal and stimulated values is indicated (*P < 0.05) C. Choriodecidua; A. Amnion.

Full size image

IL-6 (Figure 3) and IL-10 (Figure 4) showed a similar secretion profile, with the choriodecidua as the most active tissue after Escherichia coli stimulation whether applied directly or indirectly to this membrane (P < 0.05). In both cases, the largest response was observed when the stimulus was applied directly in the choriodecidual zone (13.45 [9.39–18.3] and 32.38 [23.6–44.2] pg/μg protein, respectively). Again, the amnion showed only a mild but significant increase in the secretion of IL-6 and IL-10 (10.8 [8.3–16.75] and 15.97 [13.02–21.49] pg/μg protein, respectively).

Figure 3
figure 3

In vitro secretion of IL-6 in amnion and choriodecidua regions after selective infection with Escherichia coli. IL-6 secreted to the culture medium after 24 h of infection with 1 × 106CFU of E. coli. Data were normalized in function of protein concentration (pg/μg protein) and each bar represents the 95% confidence intervals and the median (solid line) of 10 different experiments. Significant difference between basal and stimulated values is indicated (*P < 0.05) C. Choriodecidua; A. Amnion.

Full size image
Figure 4
figure 4

In vitro secretion of IL-10 in amnion and choriodecidua regions after selective infection with Escherichia coli. IL-10 secreted to the culture medium after 24 h of infection with 1 × 106CFU of E. coli. Data were normalized in function of protein concentration (pg/μg protein) and each bar represents the 95% confidence intervals and the median (solid line) of 10 different experiments. Significant difference between basal and stimulated values is indicated (*P < 0.05) C. Choriodecidua; A. Amnion.

Full size image

The basal secretion of TNFα by the amnion and choriodecidua was similar (Figure 5). However, the secretion by the choriodecidua tissue showed a 22-fold increase (53.15 [40.0–94.2] pg/μg protein) only when the Escherichia coli stimulus was applied simultaneously to both membranes (P < 0.05), while the amnion showed a 25-fold increase (P < 0.05) (29.2 [14.5–35.3] pg/μg protein) (Figure 5).

Figure 5
figure 5

In vitro secretion of TNF-α in amnion and choriodecidua regions after selective infection with Escherichia coli. TNF-α secreted to the culture medium after 24 h of infection with 1 × 106 CFU of E. coli. Data were normalized in function of protein concentration (pg/μg protein) and each bar represents the 95% confidence intervals and the median (solid line) of 10 different experiments. Significant difference between basal and stimulated values is indicated (*P < 0.05) C. Choriodecidua; A. Amnion.

Full size image

The secretion profile of IL-8 was different from that of the other cytokines. Regardless of the primary side of stimulation with Escherichia coli, the secretion was markedly increased in both compartments; however, the amnion was the most active region with a 19-fold increase (P < 0.05) in comparison with its basal concentration (Figure 6).

Figure 6
figure 6

In vitro secretion of IL-8 in amnion and choriodecidua regions after selective infection with Escherichia coli. IL-8 secreted to the culture medium after 24 h of infection with 1 × 106CFU of E. coli. Data were normalized in function of protein concentration (pg/μg protein) and each bar represents the 95% confidence intervals and the median (solid line) of 10 different experiments. Significant difference between basal and stimulated values is indicated (*P < 0.05) C. Choriodecidua; A. Amnion.

Full size image

Discussion

In the present study we used a previously published ex-vivo model that resembles the presence of two independent compartments with a fully functional chorioamnion; its upper chamber is delimited by the choriodecidua and the lower chamber by the amnion, allowing us to test both compartments simultaneously [13]. We included three different conditions of membrane stimulation with Escherichia coli, resembling clinical scenarios. 1) Microorganisms arrive to the choriodecidua through ascendant colonization (choriodecidua stimulation), 2) bacteria reach the amnion side by iatrogenic introduction during amniocentesis (amnion stimulation), and 3) bacteria are in contact with both sides of the membranes at late stages of colonization, which enables them to cross the membranes.

With this system, we were able to show that the basal secretions of IL-1β, IL-6, IL-10, and IL-8 by the choriodecidual tissue increased significantly (P < 0.05) in response to Escherichia coli stimulation, regardless of whether the stimulus was applied directly to this region, the amnion, or to both regions simultaneously. In addition, TNFα increased markedly in both the choriodecidua and the amnion sides when the bacterium was applied to both membranes at the same time. IL-1β rose mildly when the amnion was directly stimulated by Escherichia coli whereas IL-8 secretion rose markedly in the amnion regardless of the primary side stimulated. A marked rise in TNFα secretion in the amnion side was also observed but only when the bacterium was simultaneously applied to the amnion and choriodecidua tissues.

Previous works have evidenced that inoculation of the amniotic cavity with Escherichia coli induces a toxic response, characterized by the overproduction of pro-inflammatory cytokines [20, 21], such as IL-1β, a potent immunomodulator able to induce preterm labor after its experimental infusion in pregnant rhesus monkeys [22]. However, the used animal models do not allow characterizing the particular and specific contribution of each of the chorioamniotic membranes in this response and cannot be compared with our results. In addition, we demonstrated that in vitro stimulation of chorioamniotic membranes with lipopolysaccharide (LPS) and Streptococcus agalactiae induces IL-1β increase in the choriodecidua region [13].

On the other hand, the IL-6 concentration in the amniotic fluid is considered a marker of intra-amniotic inflammation and is frequently associated with an infectious process in either the amniotic fluid or the chorioamniotic space [23, 24]. Previous evidence has demonstrated that the concentration of this cytokine is increased in human decidua [25] and chorion cells [26] after treatment with IL-1β and TNFα.

Similar findings have been obtained with IL-10, an anti-inflammatory cytokine, to which a role as therapeutic factor in preterm labor has been ascribed [27] and whose concentration increases in patients with preterm labor associated with intrauterine infection [28].

The differential secretion capacities of pro-inflammatory cytokines by the amnion and the choriodecidua after Escherichia coli stimulation here reported suggest the existence of a complex interactive regulation network. There is evidence indicating that IL-1β [29], TNFα [30], IL-6 [31], and IL-10 [32], in different gestational tissues, are able to induce biosynthesis and secretion of prostaglandin E2 (PGE2) and PGF2α; these uterotonic factors play key roles in the onset and progression of labor [33] in normal and pathological conditions. In a previous work, we demonstrated that the amnion's epithelium is the main source of PGE2 secretion after stimulation with Candida albicans, yeast that has been associated with cervico-vaginal infections [14].

Under the experimental conditions used in our study, it was clear that IL-1β, IL-6, and IL-10 were mainly secreted to the choriodecidual compartment regardless of the primary zone stimulated with Escherichia coli. Interestingly, when the amnion was primarily stimulated, its contribution to increase the secretion of theses cytokines was small if any; whereas the secretion of theses cytokines by the choriodecidual region was significantly increased. This observation suggests the existence of a communication or "cross-talk" between both regions and, thus, it is possible that the amnion's epithelium might be an important indirect factor in the whole cytokine response.

On the other hand, TNFα, whose negative effects on the course of pregnancy have been characterized [34, 35], increases in the amniotic fluid of women with preterm labor and intra-amniotic infection [30]. There is also experimental evidence demonstrating that the administration of a TNFα bolus to pregnant animals causes profuse hemorrhage and pregnancy termination [36, 37]. In our model, the increase in TNFα secretion in both the amnion and choriodecidua compartments was significant (P < 0.05) only after simultaneous stimulation of both membrane sides. A possible interpretation of the biological significance of these findings is that, since TNFα is a pro-inflammatory cytokine with major immuno-toxic properties, the chorioamniotic membrane coordinates its secretion in response to a very complicated infectious scenario, as represented by chorioamnionitis, in which both the fetal and maternal sides are insulted by an infectious agent.

It is possible that the secretion of IL-1β, IL-6, IL-10, and TNFα in the choriodecidua region after infection with Escherichia coli would favor their trans-membranal translocation to the amnion [38] and thereby exert their effect on the whole membrane.

The IL-8 secretion pattern was also interesting, because both the amnion and the choriodecidua were active in the secretion of this chemokine. This ubiquitous production might play a key role in the recruitment and activation of professional cells of the immune system, such as neutrophils whose migration toward the cervix, the placenta, and chorioamniotic membranes is a clinical/histological characteristic of infection/inflammation [39, 40]. The present results show that the IL-8 response in the amnion was mild as compared to the choriodecidua region, which is the first tissue to be colonized by the microbial pathogen during an ascending intrauterine infection and is the main barrier to progression of infection into the amniotic cavity. Therefore, the tissue-specific capacities could be important factors in determining the severity of the inflammation in fetal membranes infected with Escherichia coli.

It is tempting to hypothesize that Escherichia coli infection of the chorioamniotic membrane may induce a precocious onset of the overproduction of pro-inflammatory cytokines (an "anticipation" of the normal parturition cascade?), leading to PROM and preterm labor.

Conclusion

Our results demonstrated that fetal membranes respond differentially to Escherichia coli infection. The amnion and choriodecidual cellular populations display a cooperative and bidirectional communication to secret different immunologic modulators, such as Il-1β, TNFα, IL-6, IL-8, and IL-10.

The choriodecidua is the most responsive region to the infection, as it is the first tissue to be colonized by the microbial pathogen during an ascending intrauterine infection and it is the main barrier to progression of the infection into the amniotic cavity. Therefore, the tissue-specific capacities of this region to secrete different proinflammatory cytokines are crucial factors for determining the severity of the inflammation process of fetal membranes.