Tumor-associated fibroblast-conditioned medium induces CDDP resistance in HNSCC cells

Objective EMT (epithelial to mesenchymal transition) contributes to tumor progression and metastasis. We aimed to investigate the effects of EMT on CDDP resistance in HNSCC (head and neck squamous cell carcinoma)-cells. Methods EMT was induced using conditioned medium from a tumor cell/fibroblast co-culture. HNSCC cells were alternatively treated with TGF-β1. The response to CDDP was evaluated with viability and clonogenic assays. Results Treatment of SCC-25/Detroit 562 cells with conditioned medium increased viability of the tumor cells. Moreover, it doubled the IC50 of CDDP of SCC-25 cells from 6.2 μM to 13.1 μM (p < 0.001). The IC50 of CDDP of Detroit 562 cells was increased following treatment with conditioned medium from 13.1 μM to 26.8 μM (p < 0.01). Colony forming ability after treatment with 5 or 10 μM CDDP was significantly higher in HNSCC cells treated with co-culture conditioned medium than in controls (p < 0.05). Treatment with TGF-β1 had no effect on the IC50 of CDDP (p > 0.1). Conclusions Cell free medium from a co-culture was able to induce EMT in HNSCC cells. Co-culture treated HNSCC cells revealed increased viability and were less sensitive to CDDP treatment. TGF-β1 also induced a mesenchymal phenotype, but did not alter resistance to CDDP in HNSCC cells.


INTRODUCTION
Most studies on chemotherapy resistance of tumor cells focus on genetic or phenotypic alterations of the cancer cell itself. However, there is growing evidence that the stroma of solid tumors interacts with cancer cells. Tumor-stroma interaction is considered a significant determinant of disease progression and metastasis [1,2]. Stromal fibroblasts induce epithelial to mesenchymal transition (EMT) in head and neck squamous cell carcinoma (HNSCC) tumor cells [3]. EMT is a reversible cellular process mainly induced by paracrine secretion of small molecules by tumor-associated fibroblasts [4]. Among these small molecules, transforming growth factor-β1 (TGF-β1) is thought to be one of the most relevant mediators [5,6]. TGF-β1 interacts with TGF-β1 two type I and two type II transmembrane kinase receptors. These receptors activate Smad 2 and 3 signaling pathways, which form a complex with Smad 4, leading to the expression of EMT-activating transcriptional factors such as Snail/Slug [7]. Two major hallmarks of EMT in epithelial tumor cells are increased expression of the intermediate filament vimentin and decreased expression of E-cadherin [3]. We recently reported that EMT in HNSCC-cell lines results in enhanced cell proliferation [8]. In this study, we analyze if EMT increases CDDP (Cisplatin) resistance in two HNSCC cell lines.
SCC-25 cells were orignially isolated from the primary tumor of a patient with tongue carcinoma [9]. SCC-25 cells form tumors in SCID mice but not in athymic nude mice suggesting less aggressive behavior. Moreover, SCC-25 induced tumors do not develop regional or distant metastasis in mouse models [10]. In contrast, Detroit 562 cells grow tumors and develop regional and lung metastases when injected in nude mice [11]. Detroit 562 was isolated from the malignant pleural effusion of a patient with pharyngeal carcinoma [12,13].
Previously, we had induced EMT by co-cultivation of SCC-25 or Detroit 562 cells and fibroblasts [1]. The resulting co-culture conditioned medium containing EMT-promoting factors was used in this study to induce EMT in pure SCC-25 or Detroit 562 cells. In a second experimental arm, tumor cells were treated with TGF-β1. In a third experimental arm tumor cells were treated with co-culture conditioned medium and a neutralizing dose of anti-TGF-β antibody. SCC-25 and Detroit 562 cells treated with standard cell culture medium served as control in a fourth experimental arm. Changes of vimentin and E-cadherin gene expression and protein synthesis, which were evaluated by quantitative real-time polymerase chain reaction (PCR) and by western blotting were used to confirm EMT. Cell viability was determined with MTT assays and colony-forming ability was investigated using clonogenic assays [14]. Response to CDDP was assessed by cell viability and Clonogenic assays.

Treatment with co-culture conditioned medium increased CDDP resistance of SCC-25 cells
Cell cultures were exposed to CDDP ranging from 0 µM to 50 µM and cell viability was assessed with MTT assays. Half-maximal inhibitory CDDP concentration was calculated using 4-parameter nonlinear logistic regression. The IC 50 for native SCC-25 cells was 6.24 μM (95% CI 5.43 μM to 7.06 μM). Treatment with conditioned medium significantly increased CDDP-chemoresistance ( Figure  3A). It more than doubled the IC 50 of SCC-25 cells to 13.05 μM (95% CI 10.35 μM to 15.76 μM; p < 0.005). In addition to viability assays, clonogenic assays following exposure to 5 µM CDDP were performed in increase of vimentin A. mRNA expression, while E-cadherin B. mRNA expression was significantly decreased. The treatment with 0.9 ng/ml TGF-β1 led to the same effect to an even greater extent with a highly significant upregulation of vimentin C. and downregulation of E-cadherin D. in SCC-25 cells. Experiments were performed in three replicates with three runs each. *p < 0.05, **p < 0.01, ***p < 0.001. www.impactjournals.com/oncotarget The treatment with co-culture conditioned medium did not increase both vimentin bands. The positive control HGF fibroblasts showed a strong band at 57 kD. E-cadherin was detected at 120 kD. SCC-25 cells at control conditions showed a strong 120 kD band, which was reduced after treatment with co-culture conditioned medium or with 1 ng/ml TGF-β1. Detroit 562 cells showed a light band at 120 kD, which was marginally reduced after treatment with co-culture conditioned medium or with 1 ng/ml TGF-β1 The control HG-fibroblasts did not not express protein at 120 kD with the vimentin antibody. Loading control was done using anti-GAPDH antibody.  TGF-ß1 treatment did not influence CDDP resistance. A neutralizing assay performed with anti-TGF-ß antibody supported this concept because it did not alter the CDDP resistance induced by co-culture conditioned medium. * p < 0.05, ** p < 0.01, *** p < 0.001 control and co-culture conditioned medium treated cells.
To support the observation that TGF-β1 is not involved in co-culture conditioned medium induced chemoresistance, a neutralizing dose of anti-TGF-β (1.5 µg/ml) was added to the co-culture medium before viability and clonogenity assays were performed. Despite that, the chemoresistance mediating effect of co-culture conditioned medium was preserved. Also with anti-TGF-β, the IC 50 of CDDP following administration of co-culture conditioned medium remained significantly higher than in control experiments (10.4 μM, 95% CI 9.2 to 11.5 μM (p < 0.05) ( Figure 3A). The effect of co-culture conditioned medium on clonogenity after treatment with 5 μM was not altered by anti-TGF-β antibody treatment (1281±20, 13% reduction, p > 0.5) ( Figure 4C).

DISCUSSION
EMT is a reversible cellular process caused by the interaction of epithelial tumor cells and tumor-associated fibroblasts. During EMT, epithelial tumor cells acquire characteristics of a mesenchymal phenotype. It is thought to be a relevant mechanism of cancer progression [5,6]. In this study, we were interested in the effect of EMT on CDDP resistance of HNSCC tumor cells in vitro. Co-cultures of epithelial and mesenchymal cells are frequently used to model EMT in vitro. Here we used a direct fibroblast-tumor cell co-culture conditioned medium to induce EMT in the two human papilloma virus-negative HNSCC cell lines SCC-25 and Detroit 562.

Co-culture conditioned medium induced EMT and increased cell viability
Co-culture conditioned medium upregulated vimentin expression and downregulated E-cadherin expression in SCC-25 cells. As expected for an epithelial cell line, baseline vimentin expression was low and baseline E-cadherin expression was high. Similar phenotypic changes were induced by treatment with 1 ng/ml TGF-β1. This TGF-β1 concentration has previously been measured in co-culture conditioned medium using enzyme-linked immunoassays (data not shown). TGF-β1 is considered a major factor for phenotypic changes in EMT [15].
Especially 46 kD vimentin protein levels increased in SCC-25 cells after treatment with co-culture conditioned medium and TGF-β1. 46 kD vimentin had been previously identified in epithelial cells under stress conditions [16]. On the protein level, co-culture conditioned medium and TGF-β1 slightly reduced the E-cadherin expression. The fact that EMT can be induced by cell free medium supports the concept that EMT mainly depends on paracrine signaling [8,17,18]. Beyond upregulation of mesenchymal and downregulation of epithelial markers, the co-culture conditioned medium increased cell viability in SCC 25 cells (p < 0.01). EMT-associated increase of cell viability is in line with several recent reports [4,19].
This effect was not mediated by TGF-β1, since TGF-β1 reduced cell viability in SCC-25 cells (p < 0.001). It was demonstrated that Smad 4 is active in SCC-25 cells, which might enable a tumor suppressive effect of TGF-β1 [20]. Apparently, other factors than TGF-β1 are responsible for increased viability in SCC-25 cells exposed to co-culture conditioned medium and these factors are even able to override TGF-β1 induced viability reduction in SCC-25 cells. Moreover, the discrepancy of co-culture conditioned medium and TGF-β1 induced effects on EMT and cell viability implies that observed changes in cell viability are not caused by the acquisition of a mesenchymal phenotype (EMT), but rather an EMT associated effect of epithelial-mesenchymal crosstalk in the tumor microenvironment.
Despite lack of phenotypic changes, coculture conditioned medium increased cell viability of Detroit 562 cells (p = 0.001), supporting the concept that EMT-induction and changes in cell

Co-culture conditioned medium induced CDDP chemoresistance
The main goal of this study was to determine if EMT is associated with increased CDDP chemoresistance. Medium from an epithelialmesenchymal co-culture more than doubled the IC 50 of CDDP in MTT assays in both cell lines (p < 0.001) (Figure 2). These results are consistent with the outcome of clonogenic assays. The anti-clonogenic effect of CDDP was significantly reduced after pretreatment with co-culture conditioned medium (p < 0.05) ( Table 2). Hazlehurst et al. observed in 2001 that tumor microenvironment is involved in cell cycle arrest and drug resistance. This so called microenvironment related drug resistance is mediated through survival pathways activated as a result of cell-cell-adhesion (cell-adhesion mediated drug resistance) and extracellular matrix derived adhesive signals, which reduce the cytotoxic effect of CDDP [21]. Niessner et al. reported that the release of paracrine signaling factors produced by carcinoma-associated fibroblasts stimulate survival pathways such as the Aktpathway and reduce cytotoxic effects of chemotherapies [22]. We treated HNSCC cell lines with a cell-free conditioned medium, suggesting the latter being responsible for the increase in CDDP resistance in our experimental setting. Similar results have been demonstrated in lung cancer cells. In these cells treatment with fibroblast-conditioned medium increased Paclitaxel resistance by restraining Paclitaxel induced apoptosis by activation of both extracellular signal-regulated kinases (Erk) 1/2 and Akt kinase. In this study, treatment with fibroblast-conditioned medium did not alter the CDDP resistance of lung cancer cells in vitro [23].

No effect of TGF-β1 on CDDP sensitivity
TGF-β1 did not increase the IC 50 of CDDP of HNSCC cells in vitro. The IC 50 of control and TGF-β1 treated cells was remarkably similar. Consistently, treatment with TGF-β1 had no influence on colony forming ability following CDDP-treatment. To support this observation, an additional experiment with co-culture conditioned medium supplemented with a neutralizing anti-TGF-β antibody was performed. Anti-TGF-β did not antagonize chemoresistance induction by co-culture conditioned medium in both cell lines. This supports the concept that increased CDDP-resistance is not caused, but merely associated with the acquisition of a mesenchymal phenotype. The factors leading to increased CDDP resistance following epithelial-mesenchymal interaction are not yet identified, but it doesn't seem to be TGF-β1. Gilbert et al. reported that IL-6 (Interleukin-6) and TIMP (Tissue Inhibitor of metalloproteinase) could be possible tumor stroma-derived factors, which promote the survival of cancer cells [24]. Other fibroblastproduced mediators which might mediate EMT-associated chemoresistance, are HGF (human hepatocyte growth factor), MED12 (mediator complex subunit 12) and PGE-2 (Prostaglandine-E2) [15].

Co-culture conditioned medium
For the production of co-culture conditioned medium, 4 x 10 4 /ml SCC-25 or Detroit 562 cells and 1 x 10 4 HGF cells/ml were plated in 250 ml cell culture flasks and cultured for 72 hours in 15 ml foetal bovine serumcontaining medium (1:1 mix of DMEM/F12 (PAA) and DMEM-low glucose (PAA) supplemented with 10% foetal bovine serum (FBS) (PAA), 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin). Then the cells were washed twice with Dulbecco's Phosphate-Buffered Saline (DPBS) (Biowhittaker ® , Belgium) and the serum-containing medium was replaced by 15 ml albumin-containing medium (7,5 ml DMEM/F12 (PAA) and 7,5 ml DMEM-low glucose (PAA) supplemented with bovine serum albumin (BSA, PAA) (0.4 g albumin/ 100 ml medium) replacing the protein content of 10% FBS, 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ ml streptomycin). Albumin-containing medium was left 48 hours on the co-culture allowing interacting epithelial cells and fibroblasts to secrete EMT-related factors into the medium. Afterwards, the co-culture conditioned medium was collected and cells were counted. The co-culture conditioned medium was portioned according to cell numbers as described by Hassona et al. [18]. The co-culture conditioned medium was sterile-filtered and www.impactjournals.com/oncotarget stored at -80°C.

Stimulation of SCC-25/ Detroit 562 cells with coculture conditioned medium and TGF-β1
To induce EMT in the first experimental arm, SCC-25/ Detroit 562 cells were treated with 7 ml co-culture conditioned medium per 50 ml cell culture flask for 72 hours. The medium was changed daily. To assess the effects of TGF-β1 in the second experimental arm, SCC-25/ Detroit 562 cells were cultivated in albumincontaining medium supplemented with TGF-β1 1 ng/ml (R&D Systems ® , Minneapolis, US). Exposure conditions were the same as in the first experimental group, i.e., TGF-β1 supplemented medium was used over a period of 72 hours and media were changed daily. A TGF-β1 neutralizing assay was performed in the third experimental arm. SCC-25 cells and Detroit 562 were therefore treated with anti-TGF-β 1,-2,-3 antibody (R&D Systems TM , Biomedica, Vienna; Austria) 1.5 µg/ml medium, as neutralizing dose assayed by the provider and 7 ml co-culture conditioned medium per 50 ml cell culture flask for 72 hours. Exposure conditions were the same as in the other experimental groups. In the fourth experimental arm, standard medium was used with identical media changes. At the end of the stimulation period, cells were used for RNA extraction, protein isolation, MTT assays and clonogenic assays, respectively.

RNA extraction, reverse transcription and realtime RT-PCR
A fraction of SCC-25/ Detroit 562 cells was harvested and RNA was isolated using TRIzol ® reagent following the manufacturer's instructions (Ambion ® , Life technologies TM , Thermo Fisher Scientific Inc., Waltham, MA, USA). RNA concentrations were determined by photometric measurements (BioPhotometer plus 6132, Eppendorf, Germany). Total RNA was reverse transcribed by M-MuLV Reverse Transcriptase (GeneON, Ludwigshafen, Germany) according to the manufacturer's instructions in a MyiQTM cycler (BIO-RAD Laboratories, Inc., US). Real-Time quantitative PCR (qPCR) of copy-DNA transcripts was performed in a MyiQTM cycler (BIO-RAD Laboratories Inc., Hercules, CA, US) using iTaqTM Universal SYBR ® Green Supermix (BIO-RAD Laboratories, Inc., Hercules, CA, US). β-Actin primers were purchased from Invitrogen TM (Darmstadt, Germany), while E-cadherin and vimentin primers were provided by Eurofins MWG Operon, Inc. (Ebersberg, Germany). β-Actin functioned well as a housekeeping gene and did not show significant changes across the three treatment conditions. Moreover, the size of the Real-Time PCR products was confirmed by agarose gel electrophoresis analysis.

Treatment of the cells with CDDP and IC 50 determination
Following stimulation SCC-25/ Detroit 562 cells were treated with increasing doses of CDDP (0 μM, 1 μM, 2.5 μM, 5 μM, 7.5 μM, 10 μM, 20 μM, 50 μM, 100 μM) for three days. The medium was changed daily. Cell viability was plotted against CDDP-concentration and IC 50 was calculated employing a four-parameter nonlinear regression model [26]. For the clonogenic assays, 2 × 10 4 cells were plated in 250ml cell culture flasks. After stimulation they were treated with 5 μM CDDP for three days. The media were changed daily.

MTT-assay
Cell viability was evaluated by MTT-assays using the tetrazolium salt method. The MTT-assay is a quantitative colorimetric method used to determine metabolic activity [27]. After three days of CDDP treatment, 10 μl of 5 mg/ml MTT salt (in DMEM/F12 (PAA) supplemented with 10% FBS (PAA), 2 mM l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin) was administered to the cells (100 μl). Cells were incubated for 4h at 37°C and then the formazan reaction product was dissolved using 10% sodium dodecylsulphate in 10 mM HCl at 37°C for 12 hours. Absorbance at 550 nm was measured with a microtiter plate reader (Athos 2010, Salzburg, Austria). The MTT-tests were performed in four independent sets containing at least six biological repeats.

Clonogenic assay
For analysis of the anti-clonogenic effect of CDDP on the tumor cells we used the modified clonogenic assay described by Phuk and coworkers [14]. SCC-25/ Detroit 562 cells were washed with PBS and cultured in 250 ml tissue culture flasks in DMEM/F12 (PAA) supplemented with 10% FBS (PAA), 2 mM l-glutamine, 100 units/ ml penicillin, and 100 μg/ml streptomycin for 14 days. After 14 days, cultures were fixed and stained in 0.5% gentian violet dissolved in methanol. Subsequently, the stained flasks were scanned in 1200 dpi resolution using a commercial flatbed scanner. Based on the resulting micrographs, colonies were counted and occupied areas were measured semi-automatically using a macro written in imageJ/FIJI macro language [28]. Background subtraction was performed on single images using rollingball-algorithm [29]. Micrographs were subjected to color deconvolution [30] and filtered using a Fourier band-pass filter. Colonies were segmented using auto-thresholding algorithms [31][32][33][34][35]. Segmented colonies were counted and occupied areas were measured. The clonogenic assays were performed in three independent sets.

Data analysis
Data were presented as mean +/-standard deviation (SD) unless indicated otherwise. The IC 50 +/-95% confidence intervals (CI) of CDDP were calculated with four-parameter nonlinear logistic regression using CurveExpert Professional (Daniel Hyams, Hixson, TN, USA). The results of real time PCR analysis were analyzed with GraphPad Prism 4.03 (GraphPad Software Inc, San Diego, CA, USA). Mean values among groups were compared with unpaired t-tests. MTT-changes in coculture conditioned medium treated cells vs. controls were tested with unpaired t-test. For evaluation of clonogenic assays, a two-factorial analysis of variance with colony number as the response parameter was used. Cell culture medium (co-culture vs. standard) and CCDP exposure (0 µM vs 5 µM or 10 µM) served as factors. The interaction term served to indicate cell-culture mediated differences in CDDP sensitivity. For this data analysis, SPSS 22 was used (IBM Corporation, Armonk, NY, USA)