RelB activation in anti-inflammatory decidual endothelial cells: a master plan to avoid pregnancy failure?

It is known that excessive inflammation at fetal-maternal interface is a key contributor in a compromised pregnancy. Female genital tract is constantly in contact with microorganisms and several strategies must be adopted to avoid pregnancy failure. Decidual endothelial cells (DECs) lining decidual microvascular vessels are the first cells that interact with pro-inflammatory stimuli released into the environment by microorganisms derived from gestational tissues or systemic circulation. Here, we show that DECs are hypo-responsive to LPS stimulation in terms of IL-6, CXCL8 and CCL2 production. Our results demonstrate that DECs express low levels of TLR4 and are characterized by a strong constitutive activation of the non-canonical NF-κB pathway and a low responsiveness of the canonical pathway to LPS. In conclusion, DECs show a unique hypo-responsive phenotype to the pro-inflammatory stimulus LPS in order to control the inflammatory response at feto-maternal interface.

Excessive inflammation at fetal-maternal interface is thought to be a key contributor in a compromised pregnancy. Intrauterine infections have been associated with pregnancy complications such as preterm labor, intrauterine growth restriction and pre-eclampsia 9 . Due to their location and properties, the microvascular ECs that line decidual vessels are very likely to play an important role in the control of the inflammatory response at feto-maternal interface.
In this study we investigated how decidual endothelial cells respond to LPS in terms of cytokine production, expression of TLR4 and signal transduction.

Results and Discussion
Production of pro-inflammatory cytokines by endothelial cells. We compared the production of the pro-inflammatory cytokines IL-6, CXCL8 (IL-8) and CCL2 (MCP-1) upon stimulation with TNFα or LPS in DECs, ADMECs, UtMECs and HUVECs. As expected 10 , HUVECs promptly respond to TNF-α or LPS producing IL-6, CXCL8 and CCL2 (Fig. 1). Similar results were obtained in ADMECs, and UtMECs. Interestingly, a very low production of these cytokines was observed when DECs were stimulated by LPS. DECs' low responsiveness does not seem to be time or dose dependent as assessed by time course and dose-response experiments on DECs and ADMECs (Fig. 1B). We were not surprised about this atypical behaviour of DECs upon LPS challenge, since we previously observed that LPS enhances the expression of inflammasome components and induce IL-1β secretion in trophoblasts and in decidual stromal cells but not in DEC 11 .
DECs have a lower TLR4 expression and a constitutive activation of the NF-κB non-canonical pathway. The results in Fig. 2A show that DECs express lower TLR4 mRNA levels than ADMECs.
The flow cytometric analysis in Fig. 2B confirmed the RT-qPCR data, showing that TLR4 is present on EC surface, even though its expression on DECs is very weak compared to the high expression on ADMECs. RT-qPCR experiments, summarized in Supplemental Fig. 1, show that both population of ECs, unlike macrophages used as a positive control, do not express the mRNA for CD14; interestingly DECs have also a trend to a reduced expression of the MD2, MyD88 and TRIF mRNA. These data are in agreement with Wang and colleagues who observed a significant downregulation of MyD88 and TRIF in the LPS tolerant glial cells 12 .
Since LPS induces a weak pro-inflammatory response in DECs, we aimed to investigate the activation of the NF-κ B pathways, that play a central role in modulating both pro-inflammatory and anti-inflammatory conditions 13 . The activation of the subunits p65 and p50 (canonical pathway) and of the subunits RelB, p50 and p52 (non-canonical pathway) was measured over a time-course (Fig. 2C). The p65 subunit is strongly activated by LPS in ADMECs from 45 min to 24 h. In DECs, p65 shows a certain basal activation and is mildly activated at 45 min stimulation with LPS, but not at later time points (Fig. 2C). It is evident that the canonical pathway has not a major role in DECs functions, as compared to the LPS-responsive cells.
The subunit RelB shows a retarded activation, as expected 6 . Upon LPS stimulation both cell types are characterized by a strong activation of RelB. Very interestingly, RelB shows a high basal activation in DECs which is not observed in ADMECs (Fig. 2C) and, more importantly, such basal activation is not far from its full LPS-dependent activation. In both cell types, p52 is not activated in response to LPS while p50 shows a strong basal activation, that clearly increases after LPS stimulation.
RelB basal activation and p65 low responsiveness to LPS activation seem to be the most relevant features of DECs. A similar basal activation has been observed in cancer 14 or in endotoxin tolerance. In endotoxin tolerance, RelB has a key role in silencing or inhibiting the expression of the pro-inflammatory cytokines TNF-α and IL-1β 15 and, when overexpressed, the dimer RelB/p50 inhibits the production of TNF-α in primary dendritic cells and macrophages 8 . Together with RelB activation, unstimulated DECs showed a reduced expression of Iκ Bα (Fig. 2C), the physiological inhibitor of the canonical pathway, that only slightly diminishes after LPS stimulation. This evidence would sustain the observed basal activation of p65 and its low responsiveness to LPS. In ADMECs, Iκ Bα expression is high in unstimulated cells, and rapidly decreases upon LPS activation, as occurring in LPS-responsive cells where the canonical pathway has a major role. Recently, miR-146 has been reported to regulate RelB expression in TNF-α stimulated cells 16 and to repress endothelial activation 17 . We verified that miR-146 is highly expressed in untreated DECs as compared to ADMECs (Fig. 2D) and therefore could have a role in regulating RelB activity in these cells.
To conclude, DECs are characterized by a strong constitutive activation of the NF-κB non-canonical pathway and a low responsiveness to LPS. Our finding shows a snapshot of a very atypical endothelium that, in order to avoid pregnancy failure, increases the activation threshold to LPS. This is in harmony with the concept of immune tolerance to microorganism during pregnancy 18 .

Materials and Methods
Tissues samples. Decidual first trimester biopsy specimens were obtained from women undergoing voluntary termination of pregnancy at 8-12 weeks' gestation. Skin samples were obtained from women of fertile age undergoing reductive plastic surgery. Endometrial tissue specimens were obtained from fertile women undergoing hysterectomy for leiomyomatosis in the mid proliferative and mid secretory phase defined according to Noyes criteria 19 . The study was approved by the institutional review board of The  Maternal-Children's Hospital (IRCCS "Burlo Garofolo", Trieste, Italy) and informed consent was obtained from all patients providing the tissue specimens. All the experiments were carried out in accordance with the approved guidelines.

Cell isolation and culture. Human umbilical vein endothelial cells (HUVECs) were isolated by collagenase treatment and cultured as previously published 20 . Decidual endothelial cells (DECs) and
Human dermal microvascular endothelial cells (ADMECs) were isolated and characterized as previously described by Bulla et al. 21 .
DECs were positively selected with Dynabeads M-450 (Life Technologies) coated with Ulex europaeus 1 lectin (Sigma-Aldrich), whereas ADMECs were further purified from a subconfluent mixed cell population with the CD31-conjugated magnetic beads from Dynabeads (Life Technologies).
DNA binding. DNA binding assay was performed as previously shown 24 . p65, RelB, p52, p50, binding to consensus sequence was assayed in nuclear lysates using the TransAm NF-κ B family kit (Active Motif), according to the manufacturer's instructions. The anti-NF-κ B family primary antibodies recognize an epitope that is accessible only when NF-κ B is activated and bound to its target DNA.
Statistical Analysis. Mean, standard deviation (SD), standard error of the mean (SEM) and statistical tests were calculated using GraphPad Prism (GraphPad Software, San Diego, CA). For the ELISA experiments, the Mann Whitney test was used to compare 2 groups of data, P value of < 0.05 was considered significant (*p < 0.05 **p < 0.001).