TLR4 Agonist and Hypoxia Synergistically Promote the Formation of TLR4/NF-κB/HIF-1α Loop in Human Epithelial Ovarian Cancer

Inflammation and hypoxia are involved in numerous cancer progressions. Reportedly, the toll-like receptor 4 (TLR4)/nuclear factor kappa B (NF-κB) pathway and hypoxia-inducible factor-1α (HIF-1α) are activated and closely related to the chemoresistance and poor prognosis of epithelial ovarian cancer (EOC). However, the potential correlation between TLR4/NF-κB and HIF-1α remains largely unknown in EOC. In our study, the possible positive correlation among TLR4, NF-κB, and HIF-1α proteins was investigated in the EOC tissues. Our in vitro results demonstrated that LPS can induce and activate HIF-1α through the TLR4/NF-κB signaling in A2780 and SKOV3 cells. Moreover, hypoxia-induced TLR4 expression and the downstream transcriptional activity of NF-κB were HIF-1α-dependent. The cross talk between the TLR4/NF-κB signaling pathway and HIF-1α was also confirmed in the nude mice xenograft model. Therefore, we first proposed the formation of a TLR4/NF-κB/HIF-1α loop in EOC. The positive feedback loop enhanced the susceptibility and responsiveness to inflammation and hypoxia, which synergistically promote the initiation and progression of EOC. The novel mechanism may act as a future therapeutic candidate for the treatment of EOC.


Introduction
Epithelial ovarian cancer (EOC) is a leading lethal cancer among the gynecological malignancies occurring worldwide [1]. In fact, 70-75% of EOC patients are detected at advanced stages, which converts into a high mortality rate. Although the popular treatment options currently available include surgery and chemotherapy, they are restricted by cases of frequent recurrence, and the five-year survival rate is approximately 30% [2,3]. Therefore, it is necessary to further elucidate the molecular mechanisms of ovarian carcinogenesis and to improve the treatment modalities for EOC.
Chronic inflammation and hypoxia are linked to the tumor microenvironment [4]. Inflammation is a crucial risk stressor that is associated with tumor survival and invasion. Toll-like receptor 4 (TLR4), a member of the TLRs family, which is the indispensable receptor for lipopolysaccharide (LPS), is detected in a variety of tumors, including in ovarian malignant tumors [5,6]. TLR4 upon recognition of its ligand LPS initiates downstream inflammatory responses. Past studies have demonstrated that the activation of TLR4 signaling generates proinflammatory cytokines and antiapoptotic proteins, which in turn contributes to the growth, metastasis, and chemoresistance in ovarian cancer [7,8]. In response to LPS, TLR4 mediates both myeloid differentiation factor 88-(MyD88-) dependent and MyD88independent signaling pathways. The MyD88-dependent pathway leads to the expression of proinflammatory factors by activating the early phase of NF-κB. The MyD88independent signaling pathway induces the late-phase activation of NF-κB, which triggers the expression of interferon-beta [9]. NF-κB, a critical regulator of inflammation, is believed to be a key link between inflammation and tumor [10]. Therefore, NF-κB has become an excellent target for the treatment of human tumors [11].
In addition to inflammation, hypoxia within tumors has also been shown to contribute to tumor progression. Hypoxia is an essential feature existing in most solid tumors [12]. Hypoxia-inducible factor-1 (HIF-1), which is a key transcription factor of hypoxia response, is a heterodimer that comprises of an inducible subunit, HIF-1α, and a constitutive subunit, HIF-1β. Under hypoxia, HIF-1α accumulates, forms a dimer with HIF-1β, and modulates more than 100 hypoxia-related genes [13].
Past studies have demonstrated that the TLR4/NF-κB pathway can stabilize and accumulate HIF-1α under normoxic conditions. In macrophages and dendritic cells, LPS can mediate HIF-1α expression by activating NF-κB [14,15]. More recently, peroxiredoxin 1 (Prx1), a TLR4 endogenous ligand, has also been identified to increase the interaction between NF-κB and HIF-1α promoters, which contributes to the enhancement of the HIF-1α mRNA levels along with augmentation of the HIF-1 activity [16]. Therefore, HIF-1α is a key link between hypoxia and inflammation [17]. Kim and his colleagues documented that the expression level of TLR4 was upregulated through HIF-1α in macrophages under hypoxia. Furthermore, the PI3K/Akt pathway, which promotes the translocation and activation of HIF-1, contributes to hypoxia-induced TLR4 expression [18,19]. These factors suggest that the cross talk between the TLR4/NF-κB signaling pathway and HIF-1α may form a positive feedback loop that promotes the susceptibility to inflammatory signals under hypoxia. Reportedly, the interaction between the TLR4/NF-κB signaling pathway and HIF-1α contributes to tissue damage and tumor progression in pancreatic ductal adenocarcinoma, oral squamous cell carcinoma, and lung ischemia-reperfusion injury [20][21][22]. However, the presence of any association between TLR4 and HIF-1α signaling in EOC remains unclear.
In the present study, we detected a positive association among TLR4, NF-κB, and HIF-1α protein levels in the EOC tissues. Further studies demonstrated the formation of TLR4/NF-κB/HIF-1α positive feed-forward loop in EOC cells. These findings are closely associated with inflammation and hypoxia, which synergistically promote the development of ovarian cancer. We believe that our study will contribute to developing novel targeted therapies for EOC treatment.

Cell Culture.
Human epithelial ovarian cancer cell lines A2780, ES-2, SKOV3, and OVCAR3 were sourced from ATCC. Four cell lines were maintained in RPMI 1640 with 10% FBS or DMEM-H with 10% FBS, respectively. To establish a model of cellular hypoxia, we cultured the ovarian cancer cell lines in 1% oxygen (hypoxia group) instead of 21% (normoxia or control group). CoCl 2 dissolving in 1% FBS medium was used to simulate chemical hypoxia when incubated with cells, and others in cell experiments, like PDTC, were utilized similarly to cobalt chloride. The nonpolar reagents such as TAK-242 and YC-1 were dissolved in a DMSO medium to a concentrated liquid, followed by dilution to different concentrations with 1% FBS medium. In addition, the concentrations of the reagents mentioned in this paper were determined through preliminary tests.
2.3. Immunohistochemistry Analysis. The levels of TLR4, NF-κBp65, and HIF-1α in human ovarian normal tissues (n = 10), human benign ovarian tumors (n = 10), human borderline ovarian tumors (n = 10), and well-differentiated (n = 12) and poorly differentiated human EOC tissues (n = 14) were assessed by immunohistochemistry analysis. The tissue specimens were obtained from the Department of Pathology, blinded for peer review, fixed in 10% formalin buffer, and embedded in paraffin. The sections (5 μm thick) were prepared and subjected to testing by using the SP Kit (KeyGen, Nanjing, China) as per the manufacturer's instruction, followed by overnight incubation with anti-TLR4, anti-  2.5. Western Blotting. We lysed cells with lysis buffer (Key-Gen, Nanjing, China) on an ice bath, then centrifuged the resultant at 12000 rpm at 4°C for 30 min, and then used the BCA kit (Pierce Biochemicals, USA) to detect the protein concentration. 30-80 μg proteins were separated on 10-12% SDS-PAGE and then transferred onto a polyvinylidene fluoride membrane. After blocking the antigen, the membrane was incubated overnight at 4°C with primary antibodies and then washed with TBST and incubated with secondary antibodies for 1 h. We detected the target protein levels using enhanced chemiluminescence reagents (Millipore, USA Using the intratumoral injection technique, TAK-242, PDTC, or YC-1 or their common vector DMSO/saline was injected into the xenograft tumors, and LPS or saline was injected 1 h later in the same inlet. After 24 h, mice were sacrificed, and the tumors were excised, weighed, and snap-frozen for RNA and protein extractions or paraffinembedded for H&E dye (Solarbio, Science and Technology Co., Ltd.). The animal procedure was approved by the Ethics Committee.
2.12. Statistical Analyses. Data were analyzed with the IBM SPSS21.0 software package (SPSS, Statistical Product and Service Solution Chicago). Differences of multiple sets of measurement data were performed using one-way or twoway ANOVA. The LSD t test was used for the post hoc analysis when the variances were equal, and Dunnett's T3 test was used for unequal variances. Univariate association among clinical samples was assessed by the Chi-square test. P < 0:05 was set as statistically significant.

Positive Correlation Existed between TLR4 Level and the Progression of EOC, and the Levels of NF-κBp65 and HIF-1α
in Clinical Specimens. To confirm the presence of a correlation among TLR4, NF-κBp65, and HIF-1α, we first determined the status of the abovementioned proteins in human normal ovary tissues and the clinical specimens of EOC by immunohistochemistry (IHC). The representative images (inset in Figure 1(a)) illustrated that TLR4 was mainly expressed on the cell surface and cytoplasm, while NF-κBp65 mainly existed in the cytoplasm and HIF-1α existed in the cytoplasm or nucleus. The normal ovary tissues, benign ovarian tumors, and borderline ovarian tumors were studied to reveal that TLR4, NF-κBp65, and HIF-1α were expressed at a low level, while as cancer progressed, the proportion of the TLR4 +ve , NF-κBp65 +ve , and HIF-1α +ve EOC cells were significantly increased ( Figure 1(a)). As illustrated in Figure 1(b), the integrated optical density (IOD) value analysis of positively stained settings indicated that the increased levels of TLR4, NF-κBp65, and HIF-1α were observed in the well-or poorly differentiated EOC specimens in comparison with the normal ovarian tissues (all P < 0:05). Meanwhile, compared with the well-differentiated EOC, the IOD values of TLR4 +ve , NF-κBp65 +ve , and HIF-1α +ve cells obtained from poorly differentiated EOC were significantly higher (all P < 0:05). More importantly, further analysis revealed that, in the clinical specimens of EOC, the TLR4 level was significantly positively related to the levels of NF-κBp65 and HIF-1α, respectively (Figure 1(c)). Among the TLR4 strongly positive specimens (++/+++, IODs > 3:0 × 10 5 ), the percentage of low NF-κBp65 or HIF-1α expression (-/+, IODs ≤ 3:0 × 10 5 ) was only approximately 11.1% or 33.3%, whereas high NF-κBp65 or HIF-1α expression was approximately 88.9% or 66.7%. In addition, we also found that a positive correlation existed with the NF-κBp65 levels and HIF-1α. Based on the IHC figures in Figure 1(a) and the IOD values in Figure 1(b), we calculated the percentage of NF-κB p65 +ve cells, as well as HIF-1α +ve cells in specimens of low (-/+) or high (++/+++) TLR4 levels ( Figure 1(c)). Figure 1(c) results show the possibility of a positive correlation among TLR4, NF-κB, and HIF-1α expression in EOC.
3.2. The Constitutive Expression of TLR4/MyD88/NF-κBp65/ HIF-1α Signals in Human EOC Cell Lines. MyD88, the critical adaptor protein, contributed to the development and immune escape of various tumors. Considering the active role of MyD88, the TLR4/NF-κBp65 signaling pathway can be divided into MyD88-dependent or MyD88-independent pathways. To investigate the exact role of the TLR4/ MyD88/NF-κBp65/HIF-1α pathway in the progression of EOC, we first examined the respective constitutive mRNA and protein levels of TLR4, MyD88, NF-κBp65, and HIF-1α in A2780, SKOV3, OVCAR3, and ES-2 cell lines. As shown in Figure 2(b), the broad expressions of the abovementioned proteins were recorded in the 4 EOC cell lines, except for MyD88 in the A2780 cells. No MyD88 expression could be detected even after repeat experimentations. Notable, as compared with the A2780 cells, the SKOV3 cells showed extraordinarily high levels of MyD88, NF-κBp65, and HIF-1α. To elucidate whether the MyD88-dependent and MyD88-independent TLR4/NF-κBp65 signaling pathways had similar effects on the HIF-1α activity of EOC, we used the A2780 and SKOV3 cells as the study cell models for the subsequent analysis.

Upregulating Effects of LPS on HIF-1α Activity in
Human EOC Cells Were Induced through the TLR4/NF-κB Pathway. The close association between hypoxia and inflammation is already well known. In the present study, we verified whether the TLR4/NF-κBp65 pathway is involved in mediating the upregulating effects of LPS on the expression of HIF-1α in EOC cells.
First, LPS, the inflammation inducer and the proven natural ligand of TLRs, was used as an activator of the TLR4/ NF-κB pathway in the subsequent analysis. As shown in Figure 3(a), a time-effect study by western blotting analysis showed that the HIF-1α level along with the levels of TLR4, NF-κBp65, and p-NF-κBp65 in the A2780 and SKOV3 cells was significantly enhanced by treatment with 1 μg/mL LPS for 30 min, 2 h, or 6 h, except for the p-NF-κBp65 levels in SKOV3 cells (for 5 or 15 min). Notably, the MyD88 expression in A2780 cells could not be detected either with or without LPS treatment for less than 2 h but could be induced by LPS treatment exceeding 6 h. Meanwhile, the level of MyD88 in the SKOV3 cells was slightly enhanced after LPS treatment for 2 h and 6 h. These results indicate that in EOC cells, LPS may activate the MyD88dependent or MyD88-independent TLR4/NF-κB signaling pathways and induce the expression of HIF-1α. For exploring the exact mechanisms underlying these events, the treatment time of LPS was fixed to 6 h in the subsequent analysis.
Next, the TLR4 inhibitor TAK-242 and the NF-κB inhibitor PDTC were used to determine whether the upregulating effects of LPS on the HIF-1α level in EOC cells were mediated via TLR4/NF-κB signaling. As illustrated in 4 Analytical Cellular Pathology  , the levels of TLR4, NF-κBp65, p-NF-κBp65, and HIF-1α in A2780 and SKOV3 cells treated with LPS were obviously greater than those in the control cells. Pretreatment with 25 μM TAK-242 (TLR4 inhibitor) or 25 μM PDTC (NF-κB inhibitor) for 1 h followed by LPS treatment could remarkably block the abovementioned effects, indicating that the TLR4/NF-κB signaling pathway may be involved in LPS-induced HIF-1α expression. Third, a luciferase assay was conducted to observe the HIF-1α transcriptional activity. Our results suggested that the HIF-1α transcriptional activity (indicated by the relative HRE-luc activity in Figure 3(c)) after treatment of LPS was significantly higher than the control group for A2780 and SKOV3 cells. While pretreatment with TAK-242 or PDTC could remarkably block the upregulation effects of LPS. Therefore, our cumulative findings suggest that LPS may promote the HIF-1α activity in EOC via the TLR4/NF-κB signaling pathway.

LPS and Hypoxia Stimuli
Possess the Synergistic Effects on HIF-1α Activity in Human EOC Cells. Accumulating evi-dence has demonstrated that chronic inflammation and hypoxia stimuli are the two key factors involved in tumor development [4,23,24]. After confirming the upregulating effects of LPS on the HIF-1α activity in EOC cells, we next investigated the presence of synergetic effects of LPS and hypoxia stimuli.
For this purpose, we first determined the mimic doseeffect and time-effect of CoCl 2 similar to that of 1% O 2 by the MTT and western blotting methods. In Figure 4(a), low-dose CoCl 2 treatment (50 and 100 μM) for 24 h had no effect on EOC cell proliferation, similar to that of 1% O 2 . The HIF-1α level in EOC cells increased by different intensities after CoCl 2 treatment for 24 h (50, 100, 150, 200, and 300 μM), and the maximum effect of CoCl 2 occurred at the 50 μM dose for A2780 cells and 100 μM dose for SKOV3 cells. Consequently, the working concentrations of 50 μM and 100 μM CoCl 2 were, respectively, used for A2780 and SKOV3 in the subsequent experiments. Meanwhile, a time-effect study by western blotting showed that the HIF-1α expression and the TLR4, MyD88, NF-κBp65, The number of cases analyzed is shown, and the intensity of IHC staining on each section was evaluated independently by two pathologists using the 4-step grading system via IOD analysis, that is, "-, +, ++, and +++" for "negative, low, high, and extremely high," respectively (-: IODs ≤ 1:0 × 10 5 ; +: 1:0 × 10 5 < IODs ≤ 3:0 × 10 5 ; ++: 3:0 × 10 5 < IODs ≤ 5:0 × 10 5 ; and +++: IODs > 5:0 × 10 5 ). 6 Analytical Cellular Pathology and p-NF-κBp65 levels were significantly enhanced by CoCl 2 treatment for different designated durations. As shown in Figure 4(b), the obvious upregulating effects of CoCl 2 accrued mainly at the treatment time of 2 h, 6 h, and 12 h. Therefore, we used the LPS treatment time of 6 h (Figure 3(a)) as the CoCl 2 treating time in the subsequent tests to investigate the synergetic effect of LPS and hypoxia stimuli on EOC cells. (a)      Analytical Cellular Pathology Past studies have shown that hypoxia could trigger inflammation and even further aggravate the inflammatory response [18,25]. Our results confirmed that both 1% O 2 and CoCl 2 treatments could enhance the upregulating influence of LPS on the HIF-1α activity in human EOC cells via the TLR4/MyD88/NF-κB pathway. As compared to the control group, 1% O 2 alone, CoCl 2 alone, or LPS alone treatment could increase the HIF-1α expression and the TLR4/ MyD88/NF-κBp65 signaling. Meanwhile, treatment with 1% O 2 or CoCl 2 together with LPS, in comparison with treatment with LPS alone, could significantly increase the effects (Figure 5(a)). We further detected the transcriptional activity of HIF-1α after LPS or (and) hypoxia treatment. In Figure 5(b), compared with the control groups, hypoxia (1% O 2 or CoCl 2 ) or LPS alone could obviously increase the HRE luciferase activity in both A2780 and SKOV3 cells. Furthermore, treatment with LPS combined with 1% O 2 or CoCl 2 could significantly upregulate the transcriptional activity of HIF-1α in comparison with LPS alone in A2780 cells. A similar increasing trend was recorded for SKOV3 cells, albeit there was no statistical significance between the treatments with 1% O 2 together with LPS and with LPS alone. These results indicate that LPS, combined with hypoxia could synergistically enhance the TLR4/MyD88/NF-κBp65 signaling and the expression and activity of HIF-1α.

LPS and Hypoxia Stimuli Induced the Formation of TLR4/NF-κB/HIF-1α Signaling Loop in Human EOC Cell
Lines. HIF-1α plays a vital part in the initial and developing stages of tumor metastasis. As confirmed by other researchers and based on our results, both hypoxia and LPS stimuli contribute to the activation of HIF-1α via the NF-κB signaling pathway [26][27][28]. We therefore investigated whether hypoxia could regulate the activation of TLR4/NF-κB signaling via HIF-1α. First, pretreatment with YC-1 (10 μM), the commonly used specific blocker of HIF-1α, for 1 h could remarkably block the upregulating effects of 1% O 2 or CoCl 2 on the TLR4/NF-κB signals with the coconcurrent blockage of HIF-1α, indicating that hypoxia could indirectly affect the TLR4/NF-κB signals via HIF-1α ( Figure 6(a)). Moreover, pretreatment with YC-1 for 1 h could remarkably block the nucleus accumulation and transcriptional activity of NF-κB induced by 1% O 2 or CoCl 2 , respectively (Figures 6(b) and 6(c)). Second, the sense HIF-1α vector (pCMVh-HA-ssHIF-1α), small hairpin RNA targeting HIF-1α (shRNA-HIF-1α) vector, and the corresponding empty vectors were, respectively, used to overexpress or knock down the HIF-1α expression in A2780 and SKOV3 endogenously. As illustrated in Figure 6(d), the overexpression or knockdown of HIF-1α led to the enhancement or attenuation of the TLR4/NF-κB signaling in A2780 and     Figure 7(a), the morphological results of solid EOC tumors and HE staining suggested the successful construction of the murine EOC model. The tumor tissue taken from the nude mice showed obvious malignant tumor atypia, which is the same as the primary ovary cancer. Statistical analysis on the weight of tumor-bearing mice and the diameter of tumors showed no difference (Figure 7(b)) among the groups. Notably, western blotting analysis was conducted on protein extraction from tumor samples and revealed   Figure 5: Hypoxia stress could enhance the upregulating effects of LPS on the HIF-1α activity in human EOC cells via TLR4/NF-κB signaling. The A2780 and SKOV3 cells were treated without or with 1 μg/mL LPS for 6 h under normoxic (21% O 2 ) or hypoxic (1% O 2 or CoCl 2 ) conditions, respectively. After the treatment, western blotting and luciferase reporter assays were, respectively, used to assess the (a) protein levels and the (b) HIF-1α transcriptional activity. For (b) the HRE construct and β-gal vectors were cotransfected into the A2780 and SKOV3 cells. After treatment without or with LPS for 6 h under normoxic (21% O 2 ) or hypoxic (1% O 2 or CoCl 2 ) conditions, the values of β-gal and luciferase were measured. Data are shown as mean ± SD. The experiments shown are representative of three independent experiments with similar results. The data represent the average luciferase activity of 3 wells from one experiment. * P < 0:05 vs. the control group. + P < 0:05 vs. the LPS group. ☆ P < 0:05 vs. the 1% O 2 group. ★ P < 0:05 vs. the CoCl 2 group.

12
Analytical Cellular Pathology

Discussion
TLR4, which recognizes LPS, initiates the innate immune and inflammatory responses. The transcription factor, HIF-1α, plays a key role in hypoxia-related disease. It is well known that TLR4 and HIF-1α contribute to the progression of numerous inflammatory diseases, including cancer [18,20]. In this study, we revealed that TLR4 could induce the expression and activation of HIF-1α via NF-κB and that HIF-1α could also directly regulate the TLR4/NF-κB pathway. The cross talk between TLR4/NF-κB and HIF-1α signaling pathway forms a positive feedback loop and contributes to tumor initiation, malignant progression, and prognosis in EOC.
In recent years, enhanced TLR4 expression was identified in several tumors, including pancreatic cancer, colorectal cancer, and esophageal cancer, among others [29][30][31]. Moreover, the TLR4 expression was correlated with increased levels of NF-κB and HIF-1α, higher clinical staging, and dramatically reduced patient survival [20]. Kelly et al. [7] first reported that the expression of TLR4 in EOC cells and the chemoresistance to paclitaxel are mediated by the TLR4/MyD88 signaling. Furthermore, the level of TLR4 was positively associated with the key markers of the NF-κB signaling pathway, while the coexpression of TLR4/ MyD88/NF-κB was correlated with poor prognosis in EOC patients [32,33]. A recent investigation demonstrated that the HIF-1α level is markedly enhanced in ovarian cancer in comparison with that in normal ovarian tissues and that the enhanced expression of HIF-1α is related to markedly shorter overall patient survival. Furthermore, the expression level of HIF-1α is associated with chemoresistance in a metastatic and aggressive ovarian tumor. Thus, the high level of HIF-1α may be taken as a prognostic marker in ovarian cancer [34,35]. However, the potential associations with TLR4/NF-κB and HIF-1α remain largely unknown in EOC. In this study, we noted elevated expressions of TLR4, NF-κBp65, and HIF-1α in EOC specimens, and their expressions were significantly correlated with the histological differentiation degree of the tumor. We further demonstrated a positive correlation among TLR4, NF-κBp65, and HIF-1α, indicating a possible cross talk between TLR4 signaling and the HIF-1 pathway, which acted synergistically to promote the progression of EOC.
MyD88, a key adaptor molecule, plays critical roles in TLR4 signaling. The expression of MyD88 was detected in approximately 70% of EOC patients, and it has been

15
Analytical Cellular Pathology considered a significantly poor prognostic indicator of tumor [36,37]. In this study, we selected the MyD88(-) A2780 cells and MyD88(+) SKOV3 cells for further analysis. Our results indicated that LPS could activate the TLR4/NF-κB signaling pathway in both A2780 and SKOV3 cells. Both MyD88-dependent and MyD88-independent pathways are known to mediate the TLR4/NF-κB signaling pathway following LPS response. Although TLR4 signaling through MyD88 activates the early phase of NF-κB, the MyD88independent pathway also triggers the TLR4-mediated latephase activation of NF-κB [9]. This finding is consistent with our observation of earlier NF-κB P65 phosphorylate peaks in SKOV3 cells than in the A2780 cells. Unexpectedly, our results showed that LPS could induce the MyD88 protein expression in MyD88(-) A2780 cells by exceeding 6 h of stimulation. It has been demonstrated that MyD88, an essential signaling component of the TLR4 pathway, mediates paclitaxel (PTX) resistance [7]. PTX signaling via TLR4/MyD88 induced the NF-κB pathway activation, which has been linked to the upregulation of antiapoptotic protein expression and an increase in cellular proliferation [8]. In addition, MyD88-positive EOC cells have a functioning TLR4/MyD88 pathway and are possibly indicative of an ovarian cancer stem cell that is highly resistant to proapoptotic signaling [38]. Accordingly, we proposed that inflammation may be involved in the acquisition of chemoresistance through the MyD88 expression in ovarian cancer cells. However, further research is warranted to confirm this speculation. Meanwhile, our results indicated that in both A2780 and SKOV3 cells, the expression levels of TLR4 decreased 24 h after LPS treatment. It is possible that the endocytosis of the receptor following the ligandreceptor interaction results in lysosomal degradation, which leads to the controlled duration of TLR4 signaling [39] In addition, ubiquitinated degradation is related to the downregulation of TLR4. Triad3A, an E3 ubiquitin-protein ligase, was found to bind the cytoplasmic domain of TLR4 to promote the ubiquitylation and degradation of TLR4 [40]. We further revealed HIF-1α expression and activation in ovarian cancer cells following LPS stimulation, which also involves the TLR-induced NF-κB activation. Previous studies have indicated that LPS-induced HIF-1α stabilization is mediated by the TLR4/NF-κB signaling pathway under normoxia in immune cells and oral squamous cell carcinoma [21,27,28]. Indeed, NF-κB could directly bind with the promoter of HIF-1α located at position 197/188 upstream of the transcription start site and modulate the expression of HIF-1α. Mutation of this site resulted in a significant reduction in response to p65, confirming that this site is functional [41,42]. Within the present study, when TLR4 or NF-κB was inhibited pharmacologically, our results showed that the protein levels and transcriptional activity of HIF-1α were downregulated after LPS-induction. Therefore, the present study suggests for the first time the existence of the TLR4-NF-κB-HIF-1α signaling pathway in ovarian cancer cells, which implicates an important association between the inflammatory pathway and hypoxia-response pathway.
It is well known that hypoxia stabilizes the inducible subunit HIF-1α, which is rapidly degraded via the ubiquitin-proteasome pathway under normoxia. Interestingly, short-term hypoxia increases the HIF-1α levels not only by stabilizing HIF-1α protein but also by inducing C o n t r o l T A K -2 4 2 L P S + T A K -2 4 2 L P S P D   The protein lysates were extracted from the tumor tissues, and then western blotting was performed to assay the levels of TLR4, p-NF-κBp65, NF-κB, and HIF-1α. Three independent experiments were performed, and the representative images are shown. Data represents the mean ± SD from 3 independent experiments. 16 Analytical Cellular Pathology HIF-1α gene transcription in the pulmonary artery smooth muscle cells. This mechanism involves the activation of NF-κB and its binding to the HIF-1α promoter located at position 197/188 [26]. In contrast, as per a diverse range of studies, the predominant mode of HIF-1 regulation occurs at the level of protein stability. Hypoxia has no influence on the levels of HIF-1α mRNA. The reason for obtaining the different results on HIF-1α mRNA is related to the rapid but transient increase of HIF-1α mRNA by hypoxia. Given the transient nature of this response, the inappropriate exposure time may have prevented the detection of HIF-1 mRNA upregulation [26]. Another explanation for the controversial data was proposed that a cis-repressor acting as an antagonist on an upregulating element within the HIF-1 promoter may prevent transcription of the HIF-1 gene in a cell typespecific way [43]. In addition, the NF-κB basal activity is a prerequisite to accumulating HIF-1α protein under hypoxia in cultured cells or tissue. The complete block of the NF-κB activation pathway prevented hypoxia-induced HIF-1α [44,45]. Therefore, NF-κB may be essential for the extent and speed of HIF-1α activation following hypoxic insult. In accordance with our results, CoCl 2 treatment (a hypoxia mimetic) activated NF-κB and induced HIF-1α protein expression in both A2780 and SKOV3 ovarian cancer cells.
As per these results, we considered that hypoxia could activate the NF-κB signaling pathway and contribute to the regulation of inflammation in EOC.
In addition to activating NF-κB, hypoxia could enhance the response of susceptibility to inflammatory signals by upregulating TLR4. In macrophages, HIF-1α transcriptionally regulates the TLR4 level under hypoxia via binding to the TLR4 promoter. The mouse TLR4 promoter contains a hypoxia-responsive element (HRE) located at position -407 to -404, while the human TLR4 promoter containing at least two HREs is located at the positions -2811 to -2814 and -1185 to -1188. Therefore, hypoxia augmented the cell responsiveness to bacterial components owing to the increased TLR4 expression in macrophages [18]. In addition, further investigation demonstrated that PI3K/Akt at least partially contributes to hypoxia-induced TLR4 expression via regulating HIF-1α accumulation and transcriptional activation [19]. Our results also showed that the expression levels of TLR4 were augmented on exposure to hypoxic stress. Moreover, hypoxia treatment augmented the responsiveness of ovarian cancer cells to LPS, including the TLR4 expression level and the downstream NF-κB and HIF-1α activities. Here, we further demonstrated that the hypoxiamediated TLR4 expression and downstream NF-κB transcriptional activity were HIF-1α-dependent by blocking the HIF-1α accumulation with YC-1. Meanwhile, the ability of the HIF-1α expression plasmid to enhance the TLR4 expression and the ability of HIF-1α shRNA to inhibit the TLR4 expression provided additional evidence to sustain the role of HIF-1α on regulating the expression of TLR4 in ovarian cancer cells. In accordance with our results, recent evidences have shown that HIF-1α could mediate the expression of TLR4 in pancreatic cancer, oral squamous cell carcinoma, and hepatocellular carcinoma under hypoxia. These studies revealed the critical effect of the HIF-1α-mediated TLR4-NF-κB signaling pathway in tumor growth and progression [21,46,47].

Conclusions
In the present study, we noted a strong positive correlation among TLR4, NF-κBp65, and HIF-1α in EOC specimens, indicating the existence of a cross talk between the TLR4/ NF-κB and HIF-1α signaling pathways. Our in vitro results showed that the TLR4 signaling pathway induced the HIF-1α protein expression and enhanced the HIF-1α transcription activity via NF-κB. Moreover, HIF-1α could activate the TLR4/NF-κB signaling pathway by upregulating the expression of TLR4. The in vivo transplantation model confirmed that TLR4 and HIF-1α mutually regulated each other in EOC. Combined with the results, we believe that TLR4mediated HIF-1α activation and HIF-1α-induced TLR4 expression form a positive feedback loop that is closely associated with inflammation and hypoxia, which in turn can synergistically promote the development of EOC. Our findings may provide more promising therapeutic strategies for epithelial ovarian cancer based on the inhibition of the TLR4/NF-κB/HIF-1 loop.

Data Availability
All data in this published article are available from the corresponding author on reasonable request.

Ethical Approval
The clinical study was approved by the Ethics Committee of the Tianjin Tumor Hospital. The animal experiments were approved by The Tianjin Medical University Animal Experimentation Ethics Committee.

Disclosure
A preprint has previously been published [48].