Surgical peritoneal stress creates a pro-metastatic niche promoting resistance to apoptosis via IL-8

Cell cultures

Ovarian cancer cells lines SKOV3, OVCAR3, were purchased from ATCC and cultured following ATCC recommendations (ATCC, Manassas, VA, USA). A primary ovarian cancer cell line was derived in our laboratory from ascites of a patient with Stage III serous adenocarcinoma (APOCC) (REF papier utilize avant). The cell lines were cultured in DMEM high glucose (Hyclone, Thermo Scientific), 10% FBS (Hyclone, Thermo Scientific), 1% Penicillin–Streptomycin-Amphotericyn B solution (Sigma), 1× Non-Essential Amino-Acid (Hyclone, Thermo Scientific) and 1% l-glutamine. Cultures were incubated in humidified 5% CO₂ incubators at 37 °C and the media was replaced every 3 days.

H0 and H1 sampling

Normal peritoneal samples of patients with suspicious ovarian tumor but finally benign were used in this study. Patients included in the PELVIMASS protocol, which was accepted by the French Research Ethics Committee chair (CPP No. 2016-A01381-42), signed informed consent. All conditions required surgical treatment with laparotomy. In each patient, 4 cm² peritoneal samples were harvested at a required peritoneal incision site (broad ligaments) at the time of incision (H0 specimen) and 1 h after initiation of the procedure (H1 specimen). Peritoneal samples were incubated at 37 °C in DMEM low glucose for 6 h. Media were subsequently filtered, aliquoted and stored at − 80 °C.

Cell proliferation assay

Cells were plated at 50,000 cells per well in a 6 well plate in medium without FBS. Cells were then counted with a hemocytometer for the following 6 days every 2 days. Two wells were counted per conditions. The experiment was performed in triplicates. All functional assays were performed using conditioned media from three to five different patients.

Migration assay

Migration was assessed by wound closure assay as previously described [5]. Briefly, Cells cultured at confluence in 24-well plates were scratched with a small tip along the ruler. Cells were then cultured for 6, 24 or 48 h in starvation media with or without MPs. The distances between the edges of the scratch were measured at 0 h and 6, 24 or 48 h after scratching. Data are represented as rate of wound closure.

Tube formation assay

A Matrigel-based capillary-genesis assay was performed on cells to assess their ability to form an organized tubular network as previously described [18]. Briefly, cells were starved for 6 h prior the experiment. Then 100,000 cells were cultured on 250 μl of Matrigel (BD bioscience). The degree of tube formation was quantified at different time-points by measuring the intersection of tubes in three randomly chosen fields from each well using ImageJ.

Flow cytometry

Fluorescence (FL) was quantified on a SORP FACSAria2 (BD Biosciences). Data were processed with FACS Diva 6.3 software (BD Biosciences) as previously described [19]. Doublets were excluded by FSC-W × FSC-H and SSC-W × SSC-H analysis; calcein-AM and Annexin V were acquired with 488 nm blue laser and 510/50 nm emission, PI was acquired 488 nm blue laser and 670/14 nm emission. Charts display the median of fluorescence intensity (mfi) relative to control. Single stained channels were used for compensation and fluorophore minus one (FMO) controls were used for gating. 20,000 events were acquired per sample.

Confocal microscopy

Confocal microscopy was performed on fixed cells in 3.7% formaldehyde. Cells were stained with a 50 µg/ml AF647-conjugated phalloidin (Sigma) to label actin filaments. Slides were mounted in a mounting media SlowFade^® Gold Antifade Reagent with DAPI (Invitrogen). Imaging was performed using a Zeiss confocal Laser Scanning Microscope 710 (Carl Zeiss). Post-acquisition image analysis was performed with Zeiss LSM Image Browser Version 4.2.0.121 (Carl Zeiss).

Calcein-AM staining

Calcein-AM indicates intracellular esterase activity. Cells were washed twice with Phosphate buffer saline (PBS). Cells were next stained with the 2 μM of calcein-green-AM (Molecular Probes, Invitrogen, Leiden NL) for 45 min at 37° 5% CO₂ according to manufacturers instructions. They were then immediately analyzed by FACS on a SORP FACSAria2 (BD Bioscience, San Jose, CA) as described.

Western blot analysis

Western blot were carried out as previously described [7]. Immunostaining was carried out using a rabbit monoclonal caspase3, caspase9, actin and PhosphoAKT antibody (1/1000, Cells signaling) and a secondary polyclonal mouse anti-rabbit antibody HRP conjugated (1/2000, cell signalling). Blots were developed using HRP and chemiluminescent peroxidase substrate (#CPS1120, Sigma). Data were collected using Geliance CCD camera (Perkin Elmer), and analyzed using ImageJ software (NIH).

RT-PCR analysis

Chemoresistance and cell viability study (MTT assay)

Cell viability was examined with an MTT assay [21]. Briefly, 24 h after treatment with doxorubicin, 10% of MTT reagent was added to each well to a final concentration of 500 μg/ml, and the cells were incubated for 4 h at 37 °C. 100 μl of DMSO were added to each well. Optical density was read at 570 nm versus 630 with an EnVision multilabel reader (PerkinElmer, Massachusetts, USA). 3 triplicates were performed per condition.

Study population

We reviewed tumor samples from 32 patients with advanced ovarian cancer (AOC) referred to Tenon Hospital (Paris, France) from January 2004 to July 2011. This study protocol was approved by the chair of the ethics committee of Paris VI, allowing the use of tumor tissues and medical chart of patients treated for ovarian cancer in our center. They all received platinium and taxane based neoadjuvant chemotherapy, followed by interval debulking surgery. All data, including demographics, FIGO stage, histological type and grade, and treatment modalities were collected retrospectively. Completeness of cytoreduction score was used to evaluate residual disease et the end of debulking surgery [4]. During follow up, patients who relapsed within 12 months following last chemotherapy regimen or suffering from refractory disease were considered chemo resistant.

Immunohistochemistry protocol for tumor samples

Immunohistochemistry staining were performed as previously described [22]. For each patient, we selected the most relevant tumor paraffin blocks from interval debulking surgery. Paraffin-embedded sections were deparaffinized in xylene and rehydrated in graded alcohol. Immunostaining was perfomed manually, using the Dako Envision + Dual Link System-HRP kit and anti-IL8 primary antibodies (mouse monoclonal to IL8, clone 807, ref. Ab18672, Abcam, UK). Immunostaining specificity was verified with a control antibody and a positive tissue control. All slides were counterstained with hematoxylin. Microscopic analyses were performed using a Nikon Eclipse 90i microscope (Nikon, Nikon Instrument B.V., France). Representative photographs (20× magnification) of tumor immunostaining were performed using the NIS-Element BR software package (Nikon, Nikon Instrument B.V., France). Standardized quantitative analysis of IL8 immunostaining was based on Image J software (National Institutes of health, USA). Briefly, a grid delimitating 10,000 μm² squares was applied to each photograph, and 3 randomly selected squares were analyzed. We used the Image J cell counter plugin to count the stained an unstained tumor cells within each square. IL8 expression was defined as the rate of stained tumor cells.

Statistical analysis

All quantitative data were expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed using SigmaPlot 11 (Systat Software Inc., Chicago, IL). A Shapiro–Wilk normality test, with a p = 0.05 rejection value, was used to test normal distribution of data prior further analysis. All pairwise multiple comparisons were performed by one way ANOVA followed by Holm–Sidak posthoc tests for data with normal distribution or by Kruskal–Wallis analysis of variance on ranks followed by Tukey posthoc tests, in case of failed normality test. Paired comparisons were performed by Student’s t-tests or by Mann–Whitney rank sum tests in case of unequal variance or failed normality test. Clinical data were anonymized and de-identified prior to analysis. Overall survival (OS) was computed from the date of initial diagnosis. Disease free survival (DFS) was computed from the completion of first line treatment. The first-event corresponded to death of any cause for OS and to relapse or death for DFS. OS and DFS curves were achieved using Kaplan–Meier analysis. The Cox proportional hazard regression model was used for multivariate analysis. All variables associated with p < 0.10 on univariate analysis were included in the model. Statistical significance was accepted for p < 0.05 (*), p < 0.01 (**) or p < 0.001 (***). All experiments were performed in triplicates.

Surgical peritoneal stress induces chemoresistance and pro-metastatic phenotype in ovarian cancer cells

We modeled peritoneal stress using peritoneal biopsies from patients undergoing staging procedures for suspicious early stage ovarian cancer with no extension to the peritoneal cavity (we kept only normal peritoneal biopsies of benign patients). During laparotomies, peritoneal biopsies were performed at the time of incision (H0 sample) and 1 h after initiation of surgery (H1 sample; Fig. 1a). A pathology examination confirmed the absence of tumoral cells within the peritoneal biopsies. The biopsies were then incubated for 6 h at 37 °C in DMEM low glucose and the obtained conditioned media used for the experimental procedures. We evaluated the effect of the peritoneal conditioned media at H0 and H1 on a primary ovarian cancer cell line derived in our laboratory from ascites of a patient with FIGO stage IIIC serous adenocarcinoma (APOCC) and two commercially available ovarian cancer cell lines (OVCAR3 and SKOV3).

To evaluate the effect of peritoneal conditioned media, we treated previously incubated ovarian cancer cells (OCC) with Taxol (0–20 µM). Both H0 and H1 induced resistance to Taxol compared to regular medium (APOCC, Fig. 1b–d and SKOV3 and OVCAR3, Additional file 1: Figure S1A, B). At high concentrations of Taxol (10 and 20 µM) H1 media was significantly more efficient than H0 to promote OCC survival (Fig. 1c, d). H1 media improved the proliferation of APOCC, OVCAR3 and SKOV3 cultured in starving condition (Fig. 1e and Additional file 1: Figure S1C). APOCC cell cycle analysis demonstrated an increase in S phase and G2/M when cultured with H1 (33.47% ± 0.35, 35.13% ± 0.23 and 48.66% ± 0.21 for control, H0 and H1 respectively p < 0.001; Fig. 1f).

To evaluate the impact on cancer cell phenotype, we performed confocal microscopy imaging of APOCC treated with H0 or H1. We observed an increase in F-actin stress fibers in the periphery of the cells (Fig. 1g). The stress fibers and filopods formation required for cancer cells invasion into tissues were observed only during H1 treatment. Accordingly, H1 media induced increased migration compared to H0 for all cell lines (Fig. 1h). We then evaluated cellular plasticity under H0 and H1 treatment by quantifying tube formation on matrigel after 24 h of culture (Fig. 1i). Tube formation was observed as early as 4 h after treatment with H0 or H1; however, (i) the number of tubes and the kinetic of tube formation were lower with H0 and (ii) the persistence of tubes at 6 and 24 h was only observed after H1 treatments. Overall, OCC incubation with peritoneal conditioned media resulted in a pro-metastatic phenotype and increased resistance to taxane therapy.

Effect of surgical peritoneal stress on cytokines secretion

To investigate factors responsible for the phenotypic effect of H1, we performed a cytokine array on H0 and H1 media from five different patients (Fig. 2a). We noticed differences in peritoneal basal inflammation status at H0 between patients. Patients 4 and 5 displayed a higher inflammatory profile at the beginning of the surgery (PH0) resulting in higher level of the cytokines at H1. This discordance could not be explained by demographic or pre-operative parameters. Among the 36 cytokines tested, many cytokines involved in the inflammatory response (Cd54, IL-6, IL-8, serpin E1 and Rantes), cell proliferation and negative regulation of cell death (G-CSF and C5/C5a) were significantly up regulated in H1 compared to H0 (p < 0.05, Fig. 2b). We then pre-treated APOCC for 24 h with human recombinant cytokines prior to Taxol therapy (20 µM) and demonstrated increased survival with IL-6, IL-8, Rantes and G-CSF (Fig. 2c). To confirm the role these different cytokines in H1 mediated resistance, we pre-incubated APOCC in H0 and H1 with specific blocking antibodies for 24 h (Fig. 2d left panel). As illustrated previously H0 and H1 increased survival of APOCC compared to controls (26.76% ± 2.6 of live cells in the control, 54.41% ± 2.2 with H0 and 73.31% ± 1.2 with H1). bAB against IL-6, IL-8, Rantes or G-CSF significantly reduced chemoresistance induced by H1 (live cells ranging from 56.39% ± 2.5 with G-CSF bAB to 38.48% ± 1.3 with IL-8 bAB; Fig. 2d right panel). Among all cytokines tested IL-8 bAB induced an apoptosis rate comparable to control conditions (27.18% ± 3.5 of apoptotic cells in the control and 20.33% ± 2.3 with IL-8 bAB).

IL-8 in H1 induces resistance to apoptosis

Cells incubated with H1 and treated by taxol displayed less DNA fragmentation as shown by DAPI staining (85% are fragmented in the control compare to 5% only when treated with PH1; Fig. 3a). When cells incubated with H1 were treated with IL8-bAB Taxol induced DNA fragmentation was restored (70%). We examined the expression levels of anti-apoptosic proteins Bcl2, Bcl-XL and p53 in APOCC incubated with H0 or H1 and treated with IL-8-bAB (Fig. 3b). While Bcl2 and Bcl-XL were both up regulated in APOCC exposed to H1 compared to H0, IL-8-bAB inhibited the increased expression of these anti-apoptotic proteins. Concordantly H1 treatment decreased p53 expression compared to H0 and controls (40% decrease compared to 15%, respectively; p < 0.01). P53 plays a pro-apoptotic role in Taxol-induced cell death. Therefore p53 decreased expression with H1 treatment seems to be concordant with induced resistance to apoptosis [23]. This effect was inhibited by IL-8-bAB treatment.

To investigate the downstream effectors of P53 and bcl-2 family we investigated the activation of caspases under Taxol and H1 treatment (Fig. 3c). APOCC cells pre-incubated with H1 for 24 h were treated by Taxol or anti-Fas receptor (CD95) monoclonal antibody (mAb) as positive control for apoptosis. Western blot for caspase 3 and 9 showed cleaved caspase 3 and 9 in positive controls as well as cells treated with Taxol. Pre-incubation with H1 prevented caspase 3 and 9 cleavage. The effect of H1 was inhibited by IL-8-bAB.

H1 IL-8 induces resistance to apoptosis through AKT

IL8 effect is mediated by different pathways including the Pi3-K/akt pathway. H1 incubated OCCs displayed an AKT phosphorylation which could be blocked using concomitant IL-8-bAB treatment (Fig. 4a). When we inhibited the AKT pathway (LY294002) we showed a decrease in H1 induced resistance to apoptosis (Fig. 4b). Concordantly Akt inhibition reversed the effect of H1 on Bcl2, Bcl-XL and p53 expressions (Fig. 4c).

As illustrated previously in other models we wondered if H1 was able to induce an IL-8 autocrine loop in OCC that could participate to a pro-tumoral niche. We showed an increase expression of IL-8 receptor α and β as well as IL-8 secretion in APOCC, treated with H1 suggesting the induction of a tumor autonomous IL8 autocrine loop by H1 cytokines. IL-8 blockade or AKT inhibition using LY294002 resulted in the inhibition of this autocrine loop (Fig. 4d).

IL8 overexpression in ovarian cancer is associated with chemioresistance in vivo and impacts survival

We reviewed tumor samples from 32 patients with advanced ovarian cancer referred to Tenon Hospital from January 2004 to July 2011. They all received platinium and taxane based neoadjuvant chemotherapy. The mean overall survival (OS) and disease free survival (DFS) of our study population were 54.8 and 22 months, respectively.

Sixteen patients displayed chemoresistance as defined by a recurrence within 6 months after initial treatment. No difference was observed between chemoresistant and chemosensitive patients regarding demographic and clinical parameters. Tumoral expression of IL8 was significantly higher in chemoresistant women (31% versus 11%, respectively; p = 0.003, Table 1). On multivariate analysis, IL8 expression was the only independent risk factor for chemoresistance (HR = 1.1, p = 0.016). We determined an optimal cut-off of 40% (p = 0.005) for IL8 staining in tumor samples retrieved during debulking surgery, using iterative log-rank test to maximize its prognostic value.