**3. Discussion**

Previous studies indicated that body mass index (BMI), anemia, lower education, maternal age, primiparity, multiple pregnancy, PE/HDP in previous pregnancy, gestational diabetes mellitus, preexisting hypertension, preexisting type 2 diabetes mellitus, preexisting urinary tract infection, and a family history of hypertension, type 2 diabetes mellitus, or PE/HDP are potential risk factors for PE/HDP [39,40]. Of the risk factors, obesity is a major risk factor and is associated with an increased risk for obstetrical complications such as gestational diabetes mellitus, PE/HDP, pre-term delivery, and Cesarean section [41–45], and increased neonatal morbidity and mortality [42,46,47]. Maternal obesity has been associated with low-grade metabolic inflammation due to increased release of adipokines, which are believed to contribute to maternal glucose intolerance and insulin resistance and cardiovascular and neuroendocrine modulation associated with increased maternal BMI [48]. Increased cytokine and decreased adiponectin release from adipose tissue have been linked to the meta-inflammatory state of obesity [49,50].

In this study, we measured the mRNA levels for adipokines (*LEP*, *ADIP*, and *RETN*) in human primary adipocytes and found that they were not up-regulated in response to the addition of sera from PE/HDP patients. We also measured mRNA levels of *TNF*<sup>α</sup>, *IL-6*, and *CCL2,* which have been reported to play important roles in pathogenesis or development of PE/HDP, and found that the expression of *IL-6* and *CCL2* was elevated in response to the addition of PE/HDP sera. In addition, the mRNA levels of RAGE system members (*HMGB1*, *S100B*, and *RAGE*) were significantly elevated by the addition of PE/HDP patient sera, suggesting possible involvement of the RAGE system in the up-regulation of *IL-6* and *CCL2* in adipocytes. In order to verify this possibility, we tested whether RAGE ligands up-regulate expression of *IL-6* and *CCL2* using human SW872 adipocytes and mouse 3T3-L1 cells and found that AGE and HMGB1 but not S100B significantly up-regulated gene expression of *IL-6* and *CCL2* in SW872 cells. In contrast to SW872 cells, AGE and HMGB1 up-regulated the gene expression of *IL-6* in di fferentiated 3T3-L1 cells but not in undi fferentiated cells, and the addition of AGE, but neither HMGB1 nor S100B, up-regulated *Ccl2* expression in undi fferentiated 3T3-L1 cells but any of them up-regulated *Ccl2* in di fferentiated cells. These results indicate that RAGE ligands, especially AGE and HMGB1, stimulate adipocytes to induce gene expression of *IL-6* and *CCL2*.

IL-6 is a key player in tissue inflammation and insulin resistance, and was observed in higher serum concentrations in PE/HDP patients [51]. CCL2, also referred as monocyte chemoattractant protein-1, is a key regulator of monocyte infiltration of adipose tissue, and it plays a central role in the development and maintenance of chronic adipose tissue inflammation and insulin resistance [23,52,53]. Therefore, IL-6 and CCL2 could be key players produced from adipocytes to induce tissue damages in PE/HDP patients.

Exposure of the amino acid residues of proteins to reducing sugars, such as glucose, results in non-enzymatic glycation, which forms reversible Schi ff bases and subsequently Amadori compounds. A series of further complex molecular rearrangements including dehydration, condensation, and crosslinking, yield irreversible and heterogeneous derivatives termed AGE. AGEs are chemically heterogeneous groups of compounds. Apart from endogenously formed AGEs, exogenous AGEs from foods are absorbed in the gastrointestinal tract and reportedly constitute ~10% of total AGE in the body. In animal studies, the restriction of dietary AGE intake significantly improved insulin sensitivity and extended lifespan.

HMGB1 is a nuclear protein that stabilizes nucleosome formation and facilitates transcription. HMGB1 is a strong inflammatory trigger from necrotic cells as a result of passive leakage, and can be actively secreted by activated monocytes, macrophages, dendritic cells, natural killer cells, and endothelial cells, though there is no canonical signal sequence in the HMGB1 protein. It is well-known that the levels of AGE in serum such as hemoglobin A1c (HbA1c) are increased in diabetes (hyperglycemia) patients and that diabetes is a typical risk factor for PE/HDP. Elevated HMGB1 was observed in pregnan<sup>t</sup> women with other pro-inflammatory conditions as obesity and pre-term labor. It is well established that labor is associated with a pro-inflammatory systemic response. Extracellular

HMGB1 exerts its cytokine-like activity by binding to RAGE receptor. In fact, the serum HMGB1 levels were significantly increased in the PE/HDP patients (329.2 ± 93.18 ng/mL) than those in control patients (35.45 ± 25.11 ng/mL) ( *P* = 0.0473). In the managemen<sup>t</sup> of pregnan<sup>t</sup> women, monitoring of blood glucose and HbA1c are very common but HMGB1 levels in serum are rarely monitored. Although the numbers of PE/HDP patients in our study were relatively small, the increased tendency of serum HMGB1 in PE/HDP patients suggests that the serum HMGB1 measuring could be a new marker for screening of PE/HDP risk.

As AGE and HMGB1 are ligands for RAGE, it is quite possible that AGE- and HMGB1-induced up-regulation of *IL-6* and *CCL2* is mediated via RAGE. In fact, the introduction of *siRAGE* abolished the AGE- and HMGB1-mediated increases of gene expression of *IL-6* and *CCL2* in adipocytes (Figures 4 and 5), indicating involvement of AGE and/or HMGB1/RAGE system in the up-regulation of IL-6 and CCL2 in adipocytes. Among major RAGE ligands, we tested S100B, in addition to AGE and HMGB1, but S100B failed to increase gene expression of *IL-6* and *CCL2*. As most but not all the ligands for RAGE up-regulate (pro)inflammatory mediators, such as IL-6 and CCL2, some other RAGE ligands such as macrophage-1 antigen/cluster of di fferentiation molecule 11b [54], amyloid β peptide [55], β-sheet fibrils [56], advanced oxidation protein products [57], complement C3a [58], LPS [33], and phosphatidylserine on the surface of apoptotic cells [59] might increase the expression of IL-6 and CCL2, leading to PE/HDP in pregnan<sup>t</sup> women. In fact, recent reports showed that PE/HDP was also induced by LPS [32], and that RAGE mediated LPS signaling and acted as an LPS receptor [33–38]. Thus, we tested whether LPS up-regulate gene expression of IL-6 and CCL2 in human SW872 adipocytes, and found that LPS significantly up-regulated the expression of IL-6 and CCL2 in SW872 cells via RAGE (Figures 6–9).

Some soluble products of RAGE such as soluble RAGE (sRAGE) and endogenous secretory RAGE (esRAGE) are generated from *RAGE* gene and modulate the RAGE signaling [60,61]. It was previously reported that the levels of sRAGE were reduced in PE/HDP patient serum and that serum esRAGE and the esRAGE/sRAGE ratio were elevated in PE/HDP patient serum [62]. It was also reported that pregnancy induced a significant increase in RAGE protein levels in both myometrium and omental vasculature and that blood vessels from women with preeclampsia had intense staining for RAGE in both vessel beds [63]. In the present study, we showed the up-regulation of RAGE in adipocytes by PE/HDP sera (Figure 1) but did not see sRAGE and esRAGE. Reduction of sRAGE and elevation of esRAGE/sRAGE ratio could be a potential marker for screening of PE/HDP risk.

Nuclear factor κ-light-chain-enhncer of activated B cells (NF-κB) is a key transcription factor for the expression of IL-6 and CCL2 [64,65]. RAGE ligands usually activate NF-κB [66]. The RAGE-NF-κB-IL-6/CCL2 pathway might function in adipocytes stimulated by RAGE ligands (AGE, HMGB1, and LPS), resulting in the development of inflammation that may lead to PE/HDP in pregnan<sup>t</sup> women.
