CXCL8 Promotes Endothelial-to-Mesenchymal Transition of Endothelial Cells and Protects Cells from Erastin-Induced Ferroptosis via CXCR2-Mediated Activation of the NF-κB Signaling Pathway

CXCL8-CXCR1/CXCR2 signaling pathways might form complex crosstalk among different cell types within the ovarian tumor microenvironment, thereby modulating the behaviors of different cells. This study aimed to investigate the expression pattern of CXCL8 in the ovarian tumor microenvironment and its impact on both endothelial-to-mesenchymal transition (EndMT) and ferroptosis of endothelial cells. The human monocytic cell line THP-1 and the human umbilical vein endothelial cell line PUMC-HUVEC-T1 were used to conduct in vitro studies. Erastin was used to induce ferroptosis. Results showed that tumor-associated macrophages are the major source of CXCL8 in the tumor microenvironment. CXCL8 treatment promoted the nucleus entrance of NF-κB p65 and p65 phosphorylation via CXCR2 in endothelial cells, suggesting activated NF-κB signaling. Via the NF-κB signaling pathway, CXCL8 enhanced TGF-β1-induced EndMT of PUMC-HUVEC-T1 cells and elevated their expression of SLC7A11 and GPX4. These trends were drastically weakened in groups with CXCR2 knockdown or SB225002 treatment. TPCA-1 reversed CXCL8-induced upregulation of SLC7A11 and GPX4. CXCL8 protected endothelial cells from erastin-induced ferroptosis. However, these protective effects were largely canceled when CXCR2 was knocked down. In summary, CXCL8 can activate the NF-κB signaling pathway in endothelial cells in a CXCR2-dependent manner. The CXCL8-CXCR2/NF-κB axis can enhance EndMT and activate SLC7A11 and GPX4 expression, protecting endothelial cells from ferroptosis.


Introduction
CXCL8 (interleukin-8, IL-8) is a pro-inflammatory chemokine that binds to and activates two G protein-coupled receptors, CXCR1 and CXCR2 [1]. It is known that different cell types in the ovarian cancer tumor microenvironment might secrete CXCL8, including tumor-associated macrophages, neutrophils, and endothelial cells [2,3]. In addition, CXCR1 and CXCR2 are also expressed in different cells, such as tumor cells, tumor-associated mesenchymal stromal cells (MSCs), and endothelial cells [4,5]. Therefore, CXCL8-CXCR1/CXCR2 signaling pathways through autocrine and paracrine mechanisms form complex crosstalk among the cells in the ovarian tumor microenvironment. For example, CXCL8 initiates the epithelial-mesenchymal transition (EMT) program and activates Wnt/beta-catenin signaling in ovarian cancer cells, promoting cancer cell invasion and metastasis [5]. CXCL8 can recruit tumor-associated neutrophils, which partly impair the

Tumor-Associated Macrophages Might Be the Major Sources of CXCL8 in the Ovarian Tumor Microenvironment
To characterize the expression profile of CXCL8 in the ovarian cancer tumor microenvironment, we checked immunohistochemistry (IHC) staining in the human protein atlas (HPA). Results showed that tumor cells usually have negative CXCL8 staining ( Figure 1). However, a small proportion of immune cells showed positive CXCL8 expression ( Figure 1, red frames). Wnt/beta-catenin signaling in ovarian cancer cells, promoting cancer cell invasion and metastasis [5]. CXCL8 can recruit tumor-associated neutrophils, which partly impair the cytotoxic effects of CD8+ T cells in a contact-dependent manner [3]. In addition, this signaling axis promotes angiogenesis, which supports tumor growth and metastasis in ovarian cancer [6,7]. Targeting CXCR1 and CXCR2 has shown promising results in preclinical studies as a potential therapeutic strategy for cancer treatment. Therefore, understanding CXCL8-CXCR1/CXCR2 signaling pathways in the ovarian tumor microenvironment could provide theoretical support for targeted therapies. The effect of CXCL8 on promoting angiogenesis in ovarian tumor tissues has been well characterized in previous studies [8,9]. Endothelial cells are a crucial component in the ovarian tumor microenvironment [10]. They are essential for angiogenesis, which supports tumor growth and metastasis [11]. Prevention of neovascularization is a critical target for ovarian cancer therapy. Although the utilization of some angiogenetic inhibitors, such as bevacizumab and tyrosine kinase inhibitors (TKIs), has shown some therapeutic benefits, acquired drug resistance will eventually lead to therapeutic failure [12]. Besides, endothelial cells can undergo endothelial-to-mesenchymal transition (EndMT), an important source of cancerassociated fibroblasts (CAFs) in the tumor microenvironment [13]. NF-κB signaling pathway is activated during EndMT [14,15] and promotes the transition [16].
Recent studies have shown that ferroptosis can have anti-angiogenic effects by inducing endothelial cell death, thereby inhibiting angiogenesis [17,18]. Ferroptosis is a form of programmed necrosis that involves the accumulation of reactive oxygen species (ROS) within cells in an iron-dependent manner [19]. This process can be induced by the small-molecule compound erastin, which inhibits the activity of the cystine-glutamate antiporter SLC7A11 (also known as system Xc−), ultimately causing a depletion of glutathione (GSH) [19]. Inhibiting CXCR2 might trigger ferroptosis in breast cancer cells [20], suggesting a potential role of the CXCL8-CXCR2 axis in ferroptosis. In this study, we further explored the expression profile of CXCL8 in the tumor microenvironment and the potential mechanisms of endothelial cell ferroptosis.

Tumor-Associated Macrophages Might Be the Major Sources of CXCL8 in the Ovarian Tumor Microenvironment
To characterize the expression profile of CXCL8 in the ovarian cancer tumor microenvironment, we checked immunohistochemistry (IHC) staining in the human protein atlas (HPA). Results showed that tumor cells usually have negative CXCL8 staining (Figure 1). However, a small proportion of immune cells showed positive CXCL8 expression ( Figure 1, red frames). It is generally accepted that high-grade serous ovarian adenocarcinoma is derived from the fallopian tube [21,22]. To identify the specific immune cells with positive CXCL8 It is generally accepted that high-grade serous ovarian adenocarcinoma is derived from the fallopian tube [21,22]. To identify the specific immune cells with positive CXCL8 expression, we analyzed single-RNA-seq data in normal human fallopian tubes and ovarian  To explore the specific macrophages with CXCL8 expression, we reviewed another recent single-cell RNA-seq dataset focusing on pan-cancer immune cells and fibroblasts [23]. Results showed that tumor-associated macrophages and antigen-presenting fibroblasts have significantly higher CXCL8 expression than cancer-associated myofibroblasts (Figure 3a). These findings imply that M2 macrophages might be the dominant source of CXCL8 in the ovarian tumor microenvironment. To validate this finding, we developed M0, M1, and M2 macrophages from THP-1 cells. Then, an ELISA was performed to quantify the concentration of CXCL8 in the conditioned medium from these cells. Results showed that the conditioned medium from both M1 and M2 macrophages had a significantly higher CXCL8 concentration than that from M0 macrophages (Figure 3b). To explore the specific macrophages with CXCL8 expression, we reviewed another recent single-cell RNA-seq dataset focusing on pan-cancer immune cells and fibroblasts [23]. Results showed that tumor-associated macrophages and antigen-presenting fibroblasts have significantly higher CXCL8 expression than cancer-associated myofibroblasts (Figure 3a). These findings imply that M2 macrophages might be the dominant source of CXCL8 in the ovarian tumor microenvironment. To validate this finding, we developed M0, M1, and M2 macrophages from THP-1 cells. Then, an ELISA was performed to quantify the concentration of CXCL8 in the conditioned medium from these cells. Results showed that the conditioned medium from both M1 and M2 macrophages had a significantly higher CXCL8 concentration than that from M0 macrophages (Figure 3b).

CXCL8 Treatment Activates the NF-κB Signaling Pathways via CXCR2 in Endothelial Cells
It is known that CXCL8 regulates tumor behavior through its receptors and downstream signaling pathways. To characterize the expression profiles of CXCR1 and CXCR2 in the tumor microenvironment of ovarian cancer tumors, we checked IHC staining in the HPA. Results showed tumor cells usually have negative CXCR1 and CXCR2 staining (Figure 4a). CXCR1 is almost undetectable by IHC in all cell types in the sections (Figure 2a, top panel). However, a small proportion of immune cells (probably neutrophils, according to previous publications [3,24]) and endothelial cells presented positive CXCR2 expression ( Figure 4a, bottom panel, red frames). We purchased commercial PUMC-HUVEC-T1 cells and confirmed CXCR2 expression on the cellular membrane by immunofluorescent staining (Figure 4b). Then, these cells were subjected to lentivirus-mediated CXCR2 knockdown (Figure 4c). CXCL8 treatment (50 ng/mL for 12 h) significantly increased the expression of p-IκBα and p-NF-κB p65 (p-p65), suggesting an activated NF-κB pathway (Figures 4d, f, g). IF staining confirmed that CXCL8 treatment increased the nucleus accumulation of NF-κB p65 (Figure 4e). Knockdown of CXCR2 or using a CXCR2 antagonist (SB225002, 30 nM, added 2 h before CXCL8 treatment) weakened or abrogated CXCL8induced activation of the NF-κB signaling pathway (Figure 4d

CXCL8 Treatment Activates the NF-κB Signaling Pathways via CXCR2 in Endothelial Cells
It is known that CXCL8 regulates tumor behavior through its receptors and downstream signaling pathways. To characterize the expression profiles of CXCR1 and CXCR2 in the tumor microenvironment of ovarian cancer tumors, we checked IHC staining in the HPA. Results showed tumor cells usually have negative CXCR1 and CXCR2 staining ( Figure 4a). CXCR1 is almost undetectable by IHC in all cell types in the sections (Figure 2a, top panel). However, a small proportion of immune cells (probably neutrophils, according to previous publications [3,24]) and endothelial cells presented positive CXCR2 expression ( Figure 4a, bottom panel, red frames). We purchased commercial PUMC-HUVEC-T1 cells and confirmed CXCR2 expression on the cellular membrane by immunofluorescent staining (Figure 4b). Then, these cells were subjected to lentivirus-mediated CXCR2 knockdown ( Figure 4c). CXCL8 treatment (50 ng/mL for 12 h) significantly increased the expression of p-IκBα and p-NF-κB p65 (p-p65), suggesting an activated NF-κB pathway (Figure 4d,f,g). IF staining confirmed that CXCL8 treatment increased the nucleus accumulation of NF-κB p65 ( Figure 4e). Knockdown of CXCR2 or using a CXCR2 antagonist (SB225002, 30 nM, added 2 h before CXCL8 treatment) weakened or abrogated CXCL8-induced activation of the NF-κB signaling pathway (Figure 4d

Figure 5.
CXCL8 promotes EndMT in endothelial cells via CXCR2 (a-g). qRT-PCR (a-f) and western blotting (g) were performed to assess the expression of CXCR2, CD31, CDH5, CD34, S100A4, ACTA2, and CDH2 in n PUMC-HUVEC-T1 cells treated with TGF-β1 (5 ng/mL) for 24 h alone or in combination with CXCL8 (50 ng/mL). For the indicated groups, SB225002 (30 nM, a CXCR2 antagonist) or TPCA-1 (1 μM, a potent selective inhibitor of IKK-2) were applied 2 h before CXCL8 treatment. (h,i). A tube formation assay was performed to assess the use of Matrigel (h), with treatments indicated in panel a. The number of tubule branches per view field was calculated using the Figure 5. CXCL8 promotes EndMT in endothelial cells via CXCR2 (a-g). qRT-PCR (a-f) and western blotting (g) were performed to assess the expression of CXCR2, CD31, CDH5, CD34, S100A4, ACTA2, and CDH2 in n PUMC-HUVEC-T1 cells treated with TGF-β1 (5 ng/mL) for 24 h alone or in combination with CXCL8 (50 ng/mL). For the indicated groups, SB225002 (30 nM, a CXCR2 antagonist) or TPCA-1 (1 µM, a potent selective inhibitor of IKK-2) were applied 2 h before CXCL8 treatment. (h,i). A tube formation assay was performed to assess the use of Matrigel (h), with treatments indicated in panel a. The number of tubule branches per view field was calculated using the angiogenesis analyzer of Image J software (v.1.52) (i). The data are expressed as the mean ± SD of three experiments. **, p < 0.01; ***, p < 0.001.

CXCL8 Stimulates the Expression of SLC7A11 and GPX4 via the CXCR2-Mediated NF-κB Signaling Pathway
Since NF-κB signaling has known regulatory effects on ferroptosis-related genes, such as SLC7A11 and GPX4 [28,29], we decided to investigate whether CXCL8 regulates the expression of ferroptosis-related genes via the CXCR2-mediated NF-κB signaling pathway. CXCL8 treatment significantly elevated the transcription and translation of SLC7A11 and GPX4 but not HMOX1, NRF2, or TFR2 (Figure 6a,b). However, these trends were drastically weakened in groups with CXCR2 knockdown or SB225002 treatment (Figure 6a-c). To further validate that these alterations are mediated by the NF-κB signaling pathway, we overexpressed CXCR2 in PUMC-HUVEC-T1 cells (Figure 6d). Compared to the vector control, cells with CXCR2 overexpression have increased nucleus accumulation of p65 ( Figure 6e) and higher expression of p-p65 (Figure 6f,g). However, when TPCA-1 (1 µM) was applied, CXCL8-induced p65 phosphorylation and nucleus translocation were inhibited (Figure 6e-g). In addition, TPCA-1 treatment also reversed CXCL8-induced upregulation of SLC7A11 and GPX4 (Figure 6f,h).

Discussion
Tumor-associated macrophages with protumor and inflammatory characteristics (VEGFhigh/CXCL8+/IL1β+) are found in solid ovarian tumors [30,31]. The expression and production of CXCL8 by macrophages were enhanced via reactive oxygen species (ROS)

Discussion
Tumor-associated macrophages with protumor and inflammatory characteristics (VEGFhigh/CXCL8+/IL1β+) are found in solid ovarian tumors [30,31]. The expression and production of CXCL8 by macrophages were enhanced via reactive oxygen species (ROS) and NF-κB activation [32]. In this study, we explored the expression profile of CXCL8 in the tumor microenvironment using previous databases and confirmed that tumor-associated macrophages are the major source of CXCL8. In addition, we validated upregulated CXCL8 secretion from polarized macrophages. Typically, the pro-inflammatory M1 macrophages had significantly stronger CXCL8 expression and secretion than M0 and M1 macrophages. These trends were consistent with previous findings [33,34].
The overexpression of CXCR2 in tumor cells can lead to ovarian cancer progression by enhancing NF-κB activation via EGFR-transactivated Akt signaling, thereby upregulating the expression of pro-inflammatory chemokines [35]. In addition, the CXCL8-CXCR2 axis can stimulate the activation of the NF-κB pathway in THP-1 monocytes [36], which modulates GSH generation [29,37]. SB332235, a CXCR2 antagonist, can ameliorate thioacetamide-induced activation of the TNF-α and NF-κB signaling pathways and alleviate thioacetamide-induced elevation of serum nitric oxide (NO) and malondialdehyde (MDA) levels and downregulation of GSH and superoxide dismutase (SOD) levels in both brain and liver tissues of rats [38]. These findings imply that the CXCL8-CXCR2 axis might be an important upstream axis modulating the NF-κB signaling pathway. Therefore, we checked whether this is a generalized mechanism in endothelial cells. Our data confirmed that CXCL8 stimulates NF-κB p65 phosphorylation and nucleus entrance via CXCR2 in endothelial cells.
Oxidative stress is a major hallmark of cancer due to imbalanced ROS production and antioxidant defenses within the tumor microenvironment [39]. Cancer progression is also an adaptive process for cancer cells to acquire a stronger antioxidant capacity to deal with various oxidative damages [39]. EndMT generates up to 40% of CAFs in the tumor microenvironment [13,40]. CAFs can become potent supporters of ovarian carcinogenesis and promote the initiation of tumor growth, invasion, and metastasis [41,42]. In this study, we confirmed that CXCL8 could promote TGF-β1-induced EndMT via CXCR2 and the NF-κB signaling pathway. Recent studies have also linked CAFs and ferroptosis resistance to cancer [43,44]. Exosomal miR-522 from CAFs can inhibit ferroptosis in cancer cells by reducing ALOX15 expression and blocking lipid-ROS accumulation [43]. In addition, Thrombospondin-4 secreted by CAFs confers ferroptosis resistance in GBM cell lines by transcriptionally upregulating the expression of the lncRNA DLEU1 [44].
It was observed that erastin treatment at a non-lethal level could induce a ferroptosislike phenotype that promotes endothelial cell activation. Under this status, endothelial cells acquire enhanced proliferation, migration, and tube formation capabilities [45]. In addition, this ferroptosis-like phenotype promotes the formation of vascular endothelial cadherin junctional gaps and supports cancer cell adhesion to endothelial cells and transendothelial migration [45]. Therefore, although inducing ferroptosis might be a novel strategy in tumor treatment [19], acquired resistance to ferroptosis might generate more malignant phenotypes of the tumors.
An activated NF-κB signaling pathway can protect cancer cells from ferroptosis by transcriptionally activating the expression of multiple ferroptosis-related genes [28,29,37]. In this study, we found that CXCL8 could stimulate the expression of SLC7A11 and GPX4 via CXCR2-mediated activation of the NF-κB signaling pathway. Epithelial ovarian cancer patients with high co-expression of SLC7A11 and GPX4 are predicted to have unfavorable survival and platinum resistance [46]. SLC7A11 is a subunit of the cystine-glutamate antiporter that regulates cystine and glutamate exchange across the cellular membrane. This process is critical for maintaining intracellular levels of glutathione, a key antioxidant that counteracts oxidative stress [47]. GPX4 is an enzyme that utilizes GSH as a cofactor to reduce lipid hydroperoxides and phospholipid hydroperoxides to their corresponding alcohols [47]. SLC7A11 and GPX4 are particularly important for protecting cells from ferroptosis. Our functional assays confirmed that CXCL8 protects endothelial cells from erastin-induced ferroptosis by alleviating erastin-induced GSG/GSSG drop and suppressing erastin-induced ROS and lipid ROS. However, these protective effects were largely canceled when CXCR2 was knocked down.

Bioinformatic Analysis
To evaluate gene expression at the individual cell level, two RNA-seq datasets were analyzed, one from the Human Protein Atlas (HPA) [48] and another recently published dataset (Luo 2022) that focuses on cancer-associated fibroblasts and macrophages [14]. Additionally, protein-level gene expression was assessed by examining immunohistochemistry data from the HPA [49].
To generate M0 macrophages, THP-1 cells were exposed to 10 ng/mL of 12-Otetradecanoylphorbol-13-acetate (TPA) for 24 h. Following this, the medium containing TPA was replaced with fresh RPMI-1640 medium supplemented with 10% FBS and 20 ng/mL of human recombinant M-CSF. The cells were then incubated for an additional 48 h to obtain M0 macrophages. For the induction of M1 macrophages, M0 macrophages were stimulated with 20 ng/mL of human recombinant IFN-γ and 100 ng/mL of lipopolysaccharide (LPS) for an additional 24 h. On the other hand, to induce M2 macrophages, M0 macrophages were stimulated with 20 ng/mL of human recombinant IL-4 and 20 ng/mL of human recombinant IL-13 for an additional 24 h.

Tube Formation Assay
In total, 300 µL of the Matrigel (Cat. No. 356231; BD Biosciences, San Diego, CA, USA) solution was added to each well of a 24-well plate. PUMC-HUVEC-T1 cells (3 × 104 cells/well) were then seeded on the Matrigel in the 24-well plate with TGF-β1 (5 ng/mL) for 24 h, either alone or in combination with CXCL8 (50 ng/mL). For specific groups, SB225002 (30 nM) or TPCA-1 (1 µM) were applied 2 h before CXCL8 treatment by the researchers. Tubular structures were examined using an inverted microscope (Nikon) and the number of tube branches in three fields.

Detection of Lipid Peroxidation
C11-BODIPY 581/591 (Maokang Biotechnology, Shanghai, China)) was used to monitor lipid peroxidation. PUMC-HUVEC-T1 cells were cultured in a six-well plate. After 24 h of treatment with erastin (10 µM) alone or in combination with CXCL8 (50 ng/mL), cells were washed and then treated with 10 µM of C11-BODIPY 581/591 for 30 min at 37 • C and washed with PBS. The mean fluorescence intensity (MFI) was measured using a flow cytometer (LSRFortessa; BD, Franklin Lakes, NJ, USA) by recording BODIPY emission on channels FL1-H at 530 nm and FL2-H at 585 nm. Data were collected from at least 10,000 cells and analyzed using NovoExpress (v1.5.4, Agilent, Santa Clara, CA, USA). The experiment was conducted three times independently.

Statistical Analysis
GraphPad Prism 9.5.1 software (GraphPad Software, San Diego, CA, USA) was utilized for statistical analysis. The results were expressed as the mean ± SD from at least three technical replicates of three independent experiments. To perform multiple comparisons, a one-way analysis of variance (ANOVA) with Tukey correction was employed. An unpaired Welch's t-test was used for comparisons between the two groups. The significance level was established at p < 0.05.

Conclusions
CXCL8, mainly secreted by tumor-associated macrophages, can activate the NF-κB signaling pathway in endothelial cells via a CXCR2-dependent manner. The CXCL8-CXCR2/NF-κB axis can enhance EndMT and activate SLC7A11 and GPX4 expression, protecting endothelial cells from ferroptosis. This mechanism helps expand our understanding of the acquired resistance to ferroptosis in the ovarian tumor microenvironment.  Institutional Review Board Statement: Ethical review and approval were waived for this study since no primary human or animal tissues were collected.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data is contained within the article.