The human ovarian cancer cell lines, SKOV3 (ATCC, Cell Systems & cGMP Biorepository) and Kuramochi (Japanese Collection of Research Bioresources), were employed in the present study. The SKOV3 cells and Kuramochi cells were grown in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc.) and RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc.), respectively. Culture media were supplemented with 10% (v/v) fetal bovine serum (FBS) (Gibco; Thermo Fisher Scientific, Inc.) and 1% penicillin/streptomycin (Gibco; Thermo Fisher Scientific, Inc.). Cells were maintained in an incubator at 37°C with 5% CO₂. Cells at a passage of <10 and exhibiting >90% viability (measured by trypan blue staining, data not shown) were used in the experiments.

Ovarian cancer tissues were obtained from patients undergoing oophorectomy at Siriraj Hospital, Mahidol University. Ethical approval was granted by the Siriraj Institution Board (COAs no. si670/2013 and si246/2017). The patients released the informed consent for the usage of their samples for research purposes in written form. The samples were collected at Siriraj Hospital (Bangkok) during the period of 2013-2019. Tissue sections of approximately 10 mm³ were submerged in 2X antibiotic-containing DMEM with 10% FBS overnight at 4°C to reduce microbial contamination. The sample was rinsed with 1X phosphate-buffered saline (PBS) and cut into small sections of 2 mm³. Cell debris was washed out with 10% FBS-containing DMEM with 1X penicillin (100 U/ml)-streptomycin (100 µg/ml) (Gibco; Thermo Fisher Scientific Inc.), amphotericin B (5 mg/ml) (Amphotret™, Bharat Serums and Vaccines Limited Ltd.) and HEPES (20 mM) (Gibco; Thermo Fisher Scientific Inc.) (complete medium). The tissue sections were then transferred and allowed to adhere to a Petri dish at 37°C in a CO₂ incubator. After the fibroblast-like cells protruded from the underneath the tissues, adherent cells were trypsinized with 0.25% trypsin/EDTA (Gibco; Thermo Fisher Scientific Inc.) and re-plated. The complete medium was replaced every 2-3 days. Upon reaching confluency, the cells were trypsinized and re-plated. Cultures of fibroblasts at passages 4 to 17 were used in the experiments after confirming the CAF biochemical and morphological features. These cells were named OVCAFs. The normal-associated fibroblasts (OVNFs) were isolated with the same method from adjacent non-tumoral ovarian tissues of the same patient.

Immunocytochemistry of cytokeratin 19 (CK19), α-smooth muscle actin (α-SMA), vimentin (VIM) and fibroblast activation protein (FAP) was performed to verify the purity of OVCAFs and OVNFs. Briefly, cells on sterile glass coverslips were fixed in ice-cold methanol and then incubated overnight at 4°C in a humid chamber with the indicated primary antibodies as follow: 1:200 mouse anti-human CK-19 antibody (sc-6278, Santa Cruz Biotechnology Inc.), 1:500 anti-α-SMA antibody (A5228, Abcam), 1:500 anti-VIM antibody (sc-6260, Santa Cruz Biotechnology Inc.), and 1:500 rabbit anti-human FAP (ab53066, Abcam). After washing out the excess primary antibody, the coverslip was incubated for 3 h at room temperature with the appropriate secondary fluorescent antibody including goat anti-mouse IgG-Cy3 antibody (1:2,000, #115-166-071, Jackson ImmunoResearch Laboratories Inc.) or the goat anti-rabbit IgG-FITC antibody (1:2,000, ab6717, Abcam). Hoechst 33342 solution (Invitrogen; Thermo Fisher Scientific, Inc.) was added to stain the nucleus. Fluorescence was captured using an Inverted microscope model IX71 (Olympus Corporation).

OVCAFs and OVNFs were cultured in 75-cm² flasks to reach 90-95% confluency in DMEM containing 10% FBS. Cells were then washed twice with 1X PBS and twice with 1% (v/v) FBS-DMEM and incubated for 24 h at room temperature. Subsequently, the CM were collected and designated as 24 h CM. CMs were centrifuged at 2,000 × g at 4°C for 5 min to remove cell debris and the suspension stored at −80°C until use. The Bradford assay kit (Bio-Rad Laboratories, Inc.) was used to measure the amount of total protein and adjusted to be equal when supplemented to the cells.

The SKOV3 and Kuramochi cells were cultured in a 6-well plate until they reached approximately 90% confluency. The reference lines were drawn at the middle under the plate. Cells were scratched using a sterile 1,000 µl pipette tip followed by washing 3 times with serum-free medium to discard unattached cells. The cells were treated with either OVCAF-CM or OVNF-CM or with 0, 25, 250, 2,500 and 25,000 pg/ml recombinant human IL-8 (rhIL-8)/CXCL8 (11349086, ImmunoTools GmbH). Other treatments included anti-IL-8/CXCL8 neutralizing antibody (MAB208, R&D Systems, Inc.), 200 nM rapamycin (R8781, Sigma-Aldrich; Merck KGaA) and 2 mM metformin hydrochloride (PHR1084, Sigma-Aldrich; Merck KGaA), as indicated. Incubations were performed in medium supplemented with 1 or 10% FBS, as indicated. The scratched areas indicated by the reference line were recorded at the beginning and 24 h after treatment. The migration efficiency was determined as the percentage of healing of the scratched area as calculated by the following formula: % wound healing=wound space at 0 h-at the end of the experiment ×100/wound space at 0 h.

The ovarian cancer cell lines were cultured in 25-cm² flasks until they reached 80% confluency. Following trypsinization, the cells were fluorescently labeled by incubation with 5 µmol/l Green CMFDA Dye (Invitrogen; Thermo Fisher Scientific, Inc.) for 30 min following the manufacturer's protocol. The cells, 1×10⁵ cells/well for SKOV3 and 2×10⁵ cells/well for Kuramochi, were seeded onto the coated 8 µm pore Fluoroblok™ membrane insert in the upper chambers of 24-well companion plates (BD Biosciences). OVCAF-CM or OVNF-CM was added to the cells in the upper chamber and into the lower chamber. Fluoroblok™ plates were incubated in a humidified atmosphere at 37°C in 5% CO₂ in the Synergy H1 instrument (Biotek Instruments, Inc.). The measurement of the relative fluorescence unit (RFU) at an excitation of 490 nm/emission 520 nm was measured for the migrated cells. A total of 3 independent experiments were performed.

The ovarian cancer cells were seeded in a 6-well plate and cultured for 24 h in complete medium with 10% FBS as a negative control for autophagy, or 1% FBS containing medium as a positive control for autophagy, or in OVCAF-CM or OVNF-CM, as indicated. Chloroquine (CLQ; 30 µM) (Sigma-Aldrich; Merck KGaA) was added to CM 8 h prior to cell collection. The cells were harvested and washed with 1X PBS, and lysed with in-house made radioimmuno-precipitation assay (RIPA) buffer containing 150 mM NaCl, 1.0% IGEPAL^® CA-630, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris, pH 8.0 and then heated for 5 min at 95°C to denature proteins. Protein samples were separated by 15% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes by TE 70 Semi-Dry transfer units (GE Healthcare) for 1.5 h. The membranes were blocked in a blocking solution containing 5% (w/v) skim milk in 0.1% (v/v) Tween 20 in TBS-T for 1 h at room temperature. The blots were probed by overnight incubation at 4°C with rabbit anti-human LC3B antibody (1:1,000, L7543, Sigma-Aldrich; Merck KGaA) or rabbit anti-human p62 antibody (1:500, ab109012, Abcam). The blots were then washed in TBS-T before being incubated with 1:2,000 goat anti-rabbit IgG-HRP (ab6721, Abcam) for 1 h. For β-actin detection, mouse anti-human β-actin monoclonal antibody (1:5,000) (sc-47778, Santa Cruz Biotechnology Inc.) and 1:2,000 HRP-conjugated horse anti-mouse IgG antibody (7076, Cell Signaling Technology Inc.) in blocking solution were used as primary and secondary antibody, respectively. The protein bands were developed by adding Clarity™ Western ECL substrate (170-5061, Bio-Rad Laboratories, Inc.) and detected under Gel Document (Syngene^®). The band intensities were measured using ImageJ software version 1.52a. The level of β-actin was used as an internal control to ensure equal amounts of loading proteins.

The ovarian cancer cells were allowed adhere on sterile glass coverslips and cultured until they reached >90% confluency. The cell monolayers were scratched using a 200 µl tip and then washed several times with 1X PBS to remove the cell debris. OVCAF-CM and OVNF-CM were added to the cells for 24 h. Finally, the cells were fixed in ice-cold methanol and then incubated overnight at 4°C in a humid chamber with anti-human LC3B antibody (1:100) (L7543, Sigma-Aldrich; Merck KGaA) and anti-human E-cadherin (2Q663, sc-71008, Santa Cruz Biotechnology, Inc.) After washing out the excess of primary antibody, the coverslip was incubated for 3 h at room temperature with the appropriate fluorescence conjugated secondary antibody goat anti-rabbit IgG-FITC antibody (1:2,000, ab6717, Abcam) and goat anti-mouse IgG-CyTM3 antibody (115-166-071, Jackson ImmunoResearch Laboratories, Inc.). Antibodies were diluted in 0.1% Triton X in 1X PBS containing 10% FBS. Hoechst 33342 solution (Invitrogen; Thermo Fisher Scientific, Inc.) was then added to stain the nuclei (30 min at room temperature). Fluorescence signals were captured with a Leica DMIRE2 confocal fluorescence microscope (Leica Microsystems AG) equipped with Leica Confocal Software v. 2.61.

The cytokines/chemokines secreted by the CAFs into the CM were assayed using a ProHuman Cytokine Standard 27-Plex assay, Group I (#M50-0KCAF0Y, Bio-Rad Laboratories, Inc.), which monitors 27 human cytokines, including interleukin (IL)-1β (IL-1β), IL-1 receptor α (IL-1ra), IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-17A, eotaxin, basic fibroblast growth factor (bFGF), granulocyte-colony stimulating factor (G-CSF), granulocyte-monocyte-colony stimulating factor (GM-CSF), interferon-γ (IFN-γ), interferon-γ-induced protein 10 (IP-10), monocyte chemoattractant protein-1 (MCP-1), monocyte inhibiting protein-1α (MIP-1α), monocyte inhibiting protein-1β (MIP-1β), platelet-derived growth factor-BB (PDGF-BB), RANTES, tumor-necrosis factor-α (TNF-α) and vascular endothelial growth factor (VEGF). The detection process was performed following the instruction leaflet. Briefly, 24 h CM from OVCAFs and OVNFs was centrifuged at 1,000 × g for 10 min at 4°C and the concentration of cytokines/chemokines of each sample was determined using the Bio-Plex 200 system (Bio-Rad Laboratories, Inc.). Data were analyzed using Bio-Plex Manager™ Software Version 4.0 (Bio-Rad Laboratories, Inc.). The cytokine concentration (pg/ml) represents the mean ± SD of 3 independent experiments in triplicate.

All statistical analyses were performed using SPSS version 23 (SPSS Inc.). The values are expressed as the means ± SD. Statistical significance was determined by one-way ANOVA with the Tukey's post hoc test when comparing multiple samples. The Student's t-test was used when comparing each sample to the normal control. A P-value <0.05 was considered to indicate a statistically significant difference.

OVCAFs isolated from fresh ovarian cancer tissues from 3 patients exhibited positivity for the specific markers, including α-SMA, VIM and FAP, while they were negative for the epithelial marker, CK19 (Fig. 1). As a further confirmation, the OVCAFs, but not the normal fibroblasts isolated from counterpart benign ovary tissues of the same patients (OVNFs), expressed FAP (Fig. 1). The ability of the CM from OVCAFs and from OVNFs isolated from the 3 patients to induce the migration of fluorescently0labeled ovarian cancer SKOV3 and Kuramochi cells through a porous membrane of the Transwell was assayed (as described in the Materials and methods). As a baseline, control medium supplemented with 1% FBS was employed to maintain the level of serum factors to the minimum. Cell migration stimulated by the OVCAF-CM was evidently greater than that stimulated by the OVNF-CM counterpart, with statistical significance at all incubation times tested (Fig. 2A and B).

Given the role of autophagy in ovarian cancer cell migration (18,21), the present study assessed the level of cellular autophagy in the treated cancer cells. The cells were incubated with medium as indicated in the presence of CLQ to prevent autophagosome degradation (22), and the expression of the autophagy marker, LC3, was analyzed in the homogenates by western blot analysis. Upon the induction of autophagy, the cytosolic LC3-I isoform is converted into the autophagosomal LC3-II isoform (22). The LC3-II/LC3-I ratio, considered a marker of the induction of autophagy (22), was decreased in both ovarian cancer cell lines exposed to OVCAF-CM (from 3 patients) compared to the OVNF-CM of the respective patients (Fig. 2C and D). The downregulation of autophagy in the SKOV3 cells was significant with the CM of OVCAF-9 and of OVCAF-20, whereas in the Kuramochi cells, the CM from OVCAF-20 was more effective. Notably, the cancer cells cultured in control culture medium containing 10% FBS exhibited a lower level of autophagy as compared to the cells cultured with 1% FBS (Fig. 2C and D). Taken together, these results indicate a link between low autophagy and a high migration rate, and suggest that OVCAFs-CM induce ovarian cancer cell migration by reducing autophagy in the cancer cells. To confirm this link, the expression of vacuolar LC3 (corresponding to the autophagosomal LC3-II isoform) and of the epithelial marker E-cadherin (E-cad) was assessed by immunofluorescence staining in the ovarian cancer cells at the migration front. It is expected that E-cad, which mediates cell-cell adhesion, would be downregulated in cells that begin to migrate.

In both the SKOV3 and Kuramochi cells, incubation OVCAF-CM led to a reduced positivity for vacuolar LC3 and for membrane E-cad compared to the cells incubated with the counterpart OVNF-CM (Fig. 2E and F). This pattern is characteristic of migratory cells (18,21). Serum factors are known to stimulate cell proliferation and cell migration (22) and to downregulate basal autophagy (23). Therefore, the present study assayed the expression of these proteins in cultures incubated in standard culture medium supplemented with low (1%) or high (10%) FBS as controls. A high LC3 and high E-cad expression pattern was observed in the cells cultured with 1% FBS medium and exhibiting a low invasive capability (Fig. 2A-D), whereas the opposite pattern (low LC3 and low E-cad expression) was observed in the cells cultured in 10% FBS medium exhibiting migratory features (Fig. 2E, F). These data consistently demonstrate that cancer cell migration induced by OVCAF-CM and by high serum (10% FBS) is associated with the downregulation of autophagy.

Secreted substances in OVCAF-CM and in OVNF-CM from 6 patients (case nos. 9, 20, 26, 34, 43 and 48) were assayed using a cytokine array for 27 cytokines/chemokines including IL-1β, IL-1rα, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-17A, eotaxin, bFGF, G-CSF, GM-CSF, IFN-γ, IP-10, MCP-1, MIP-1α, MIP-1β, PDGF-BB, RANTES, TNF-α and VEGF. The raw data of the cytokine levels are presented in Table I. In total, 10 cytokines, namely IL-6, IL-8, eotaxin, GM-CSF, IFN-γ, IP-10, MCP-1, MIP-1b, RANTES and VEGF, were in the detectable range. Of the 6 OVCAF-CM, only 7 from the 10 total detectable cytokines, including IL-6, IL-8, G-CSF, IFN-γ, IP-10, MCP-1 and VEGF, exhibited a significantly increased level (P-value <0.05 and <0.01) (Fig. S1). All 6 OVCAF-CM exhibited statistically significant higher levels of IL-8 (P-value <0.01) and MCP-1 (P-value <0.05) compared with the 6 pooled OVNF-CM (Fig. 3A). Since a different genetic background can influence the gene expression and the secretion of cytokines, the matched OVNF-CM and OVCAF-CM from the same patients were selected and compared. The following were included in the assay: OVCAF-9 and OVNF-9, OVCAF-20 and OVNF-20, and OVCAF-43 and OVNF-43. The results revealed a significantly higher level of IL-8 in all 3 pairings of OVCAF-CM compared to OVNF-CM. In addition, the GM-CSF, IFN-γ, IP-10 and MCP-1 levels were increased in OVCAF-43-CM compared to those in the normal counterpart with statistical significance (Fig. 3B-D).

According to the consistent finding of high IL-8 in all OVCF-CM compared to their normal counterpart CM, it was hypothesized that IL-8 may be the most relevant secreted substance in OVCAF-CM driving cell migration in association with the suppression of autophagy. To examine this hypothesis, the effect of rhIL-8 was assayed. The positive control of high basal autophagy was detected in cells cultured with 1% FBS containing medium compared to that in the cells grown in 10% FBS. CLQ was added to discriminate LC3-II accumulation arising from increased autophagosome formation from that due to impaired autophagosome degradation (24). As shown by the results of western blot analysis of the LC3-II/LC3-I ratio, it was evident that rhIL-8 significantly inhibited basal autophagy both in SKOV3 cells (at 25 pg/ml to 25,000 pg/ml) and in Kuramochi cells (at 25,000 pg/ml) (LC3, Fig. 4A and B). The p62/SQSTM1 protein, a receptor for cargo destined to be degraded by autophagy, accumulates in cells when the autophagic flux is hampered (24). In the present study, p62 protein significantly accumulated in the cells treated with high concentrations of rhIL-8 (2,500 and 25,000 pg/ml) (p62, Fig. 4A and B). Taken together, the results with both autophagy markers (LC3 and p62) support the conclusion that rhIL-8 decreases the level of basal autophagy in ovarian cancer cells.

To further support this contention, neutralizing anti-IL8 antibody was applied to the culture medium supplemented with FBS or rhIL-8. The addition of anti-IL8 to the culture medium restored the autophagic flux, as evidenced by the high LC3-II/LC3-I ratio and the reduced accumulation of p62, both in the SKOV3 and Kuramochi cells cultured in the presence of rhIL-8 or 10% FBS (Fig. 4C and D). Immunofluorescence staining of LC3 further confirmed that IL-8 effectively inhibited autophagy in ovarian cancer cells (Fig. 4E). Of note, the fact that anti-IL-8 restored autophagy in the cells cultured with 10% FBS to the level of that in control cells cultured with 1% FBS suggests that FBS contains some IL-8 (Fig. 4C and D, lane 3 vs. lane 2). In a wound-healing assay, the addition of rhIL-8 markedly promoted wound closure, much alike in the culture with 10% FBS, supporting the hypothesis that FBS contains IL-8 (Fig. 4F and H; with respective quantification data shown in Fig. 4G and I). The data also demonstrated that the SKOV3 cells were more responsive than the Kuramochi cells to IL-8. Indeed, the Kuramochi cells exhibited less motility and required a longer time (up to 48 h) for wound healing in the presence of a high FBS or IL-8 concentration compared to the SKOV3 cells.

Finally, a wound-healing assay was performed using cells incubated with rhIL-8 in the presence of rapamycin or metformin, two drugs known to induce autophagy via mTOR inhibition and AMPK activation, respectively. Both autophagy inducers neutralized the migratory capability of the SKOV3 (Fig. 5A and B) and Kuramochi (Fig. 5C and D) ovarian cancer cells exposed to IL-8. These findings confirm that the pro-migratory effect of IL-8 is mechanistically linked to the downregulation of autophagy.

The possible involvement of IL-8 in the modulation of autophagy by FBS and the CM from normal and CAFs was investigated further. To this end, the cellular levels of LC3-II and p62 in the SKOV-3 cells (Fig. 6A-C) and in Kuramochi cells (Fig. 6D-F) cultured in 10 or 1% FBS or in CM from OVNFs or OVCAFs in the presence of the neutralizing anti-IL-8 antibody was assayed. CLQ was added to prevent autophagosome degradation. Consistent with the data in Fig. 4C and D, in both ovarian cancer cell lines the anti-IL-8 antibody restored autophagy in the cancer cells cultured in 10% FBS medium to the level found in the control (no anti-IL-8) cells cultured in 1% FBS medium (Fig. 6A and B, and D and E; lane 3 vs. lane 2), and this effect was statistically significant. Moreover, anti-IL-8 abrogated the negative effect of CAF-CM on autophagy modulation (reduced autophagy), maintaining the LC3-II/LC3-I levels at values comparable to those observed in cells incubated with OVNF-CM or in control 1% FBS medium (Fig. 6A and B, and D and E; lanes 5-10 compared to lane 2). A decreased accumulation of p62 was observed in the OVCAF-CM (cases no 20 and 43)-treated cells upon the addition of anti-IL-8 antibody compared to that in the control cells incubated in 10% FBS medium (Fig. 6A-C, and D-F; lanes 5-10 compared to lane 2), indicating that anti-IL-8 antibody restored the autophagic flux, reverting the block induced by OVCAF-CM.

Finally, to link this observation with the functional effect on cell migration, the expression levels of LC3 and E-cad in the cells at the migration front in the cancer cells cultured as described above were assayed. Neutralizing anti-IL-8 antibody attenuated the migration of both ovarian cancer cells cultured in 10% FBS medium (Fig. 6G and H). Anti-IL8 antibody increased the expression of LC3 and of E-cad in the cells incubated in 1% FBS, while this effect was less prominent in the cells cultured in 10% FBS. Similarly, anti-IL8 antibody blocked the cell migration induced by OVCAFs-CM concomitantly, with the upregulation of LC3 expression (indicative of autophagy) and the downregulation of E-cad expression (indicative of migratory cells) in the cells at the migration front (Fig. 6G and H). From these data, it can be concluded that OVCAF-CM reduces autophagy and mediates cancer cell migration through IL-8.