Multiplex quantitative analysis of stroma-mediated cancer cell invasion, matrix remodeling, and drug response in a 3D co-culture model of pancreatic tumor spheroids and stellate cells
Pancreatic ductal adenocarcinoma (PDAC) is a stroma-rich carcinoma, and pancreatic stellate cells (PSCs) are a major component of this dense stroma. PSCs play significant roles in metastatic progression and chemoresistance through cross-talk with cancer cells. Preclinical in vitro tumor model of invasive phenotype should incorporate three-dimensional (3D) culture of cancer cells and PSCs in extracellular matrix (ECM) for clinical relevance and predictability.
PANC-1 cells were cultured as tumor spheroids (TSs) using our previously developed minipillar chips, and co-cultured with PSCs, both embedded in collagen gels. Effects of PSC co-culture on ECM fiber network, invasive migration of cancer cells, and expression of epithelial-mesenchymal transition (EMT)-related proteins were examined. Conditioned media was also analyzed for secreted factors involved in cancer cell-PSC interactions. Inhibitory effect on cancer cell invasion was compared between gemcitabine and paclitaxel at an equitoxic concentration in PANC-1 TSs co-cultured with PSCs.
Co-culture condition was optimized for the growth of TSs, activation of PSCs, and their interaction. Increase in cancer cell invasion via ECM remodeling, invadopodia formation and EMT, as well as drug resistance was recapitulated in the TS-PSC co-culture, and appeared to be mediated by cancer cell-PSC interaction via multiple secreted factors, including IL-6, IL-8, IGF-1, EGF, TIMP-1, uPA, PAI-1, and TSP-1. Compared to gemcitabine, paclitaxel showed a greater anti-invasive activity, which was attributed to suppresion of invadopodia formation in cancer cells as well as to PSC-specific cytotoxicity abrogating its paracrine signaling.
Here, we established 3D co-culture of TSs of PANC-1 cells and PSCs using minipillar histochips as a novel tumoroid model of PDAC. Our results indicate usefulness of the present co-culture model and multiplex quantitative analysis method not only in studying the role of PSCs and their interactions with tumor cell towards metastatic progression, but also in the drug evaluation of stroma-targeting drugs.
KeywordsTumor spheroids Pancreatic stellate cell Tumor microenvironment Matrix remodeling 3D co-culture Cancer invasion Paclitaxel
Brain-derived neutrophic factor
Epithelial growth factor
Granulocyte macrophage colony-stimulating factor
Hepatocyte growth factor
Insulin like growth factor-1
Insulin like growth factor binding protein-2
Macrophage colony-stimulating factor
Membrane-type 1 matrix metalloproteinase
Plasminogen activator inhibitor-1
Pancreatic ductal adenocarcinoma
Pancreatic stellate cells
Stromal cell derived factor-1
Transforming growth factor beta-1
Tissue inhibitors of metalloproteinases
Tumor-necrosis factor super family-1B
Urokinase plasminogen activator
Urokinase plasminogen activator receptor
Vascular endothelial growth factor-A
Wingless-related integration site-2
α-Smooth muscle actin
Pancreatic ductal adenocarcinoma (PDAC) is an aggressive malignant tumor of the exocrine pancreas with a 5-year survival rate of less than 5% . Only 15–20% of the patients are candidates for curative surgical resection . In patients with unresectable or metastatic disease, gemcitabine (GEM) has been the cornerstone of treatment, despite only a small advantage in terms of survival .
PDAC is one of the most stroma-rich carcinomas, and pancreatic stellate cells (PSCs) are a major component of this dense stroma . PSCs differentiate into cancer-associated fibroblasts (CAFs) and have a strong impact on tumor progression through cross-talk with cancer cells . PSC-derived secretory factors including growth factors are released into the tumor microenvironment (TME) and promote epithelial-mesenchymal transition (EMT), which in turn induces cancer invasion and drug resistance . PSCs also participate in remodeling of the ECM  and exert a physical influence stimulating cell migration . These characteristics of PSCs highlight the importance of PSCs in tumor model for human PDAC.
In vitro tumor models may be useful for studying the effects of PSCs on tumor growth, invasion, and drug resistance. Traditional two-dimensional (2D) cultures do not take into account the in vivo conditions of the TME  and yield data with minimal clinical relevance. On the other hand, three-dimensional (3D) tumor models such as tumor spheroids (TSs) retain important 3D characteristics of in vivo tumors and TME and recapitulate the molecular signature and behavior of cancer cells [10, 11, 12]. In particular, spheroid cultures embedded in suitable matrix such as collagen allow 3D cell-matrix interactions, hence, they are increasingly used to study cell-ECM interactions , local invasion , migration , and drug resistance .
We were to develop a 3D co-culture model for high-content analysis of cellular processes such as cell invasion, matrix remodeling and drug response, resulting from the interaction between cancer cells and PSCs. We used our previously developed minipillar chip designed for the culture of matrix-embedded TSs [17, 18] and culture conditions were further optimized for 3D reciprocal interactions between TSs of PANC-1 pancreatic cancer cells and PSCs. Under these conditions, TSs acquired a more invasive phenotype with a progressive EMT signature, and promoted the remodeling of the ECM. Paclitaxel (PTX), compared to GEM, had a superior inhibitory effect on the EMT signature and invasiveness of cancer cells. Overall, our minipillar co-culture model represents a novel method for evaluation of stroma-mediated cancer cell invasion, matrix remodeling and differential drug sensitivity in pancreatic tumor.
Cell culture and reagents
Human pancreatic cancer cell lines, MIAPaCa-2, AsPC-1, PANC-1, Capan-1 and BxPC-3 were purchased from American Type Culture Collection (ATCC, USA). PSC, human pancreatic stellate cell line, was obtained from ScienCell (HPaSteC, #3830, Carlsbad, CA). MIAPaCa-2 and PANC-1 cells and PSCs were maintained in high glucose DMEM (Hyclone, Logan, UT) and AsPC-1, BxPC-3 and Capan-1 cells in RPMI-1640 medium (Gibco, Grand Island, NY); both media were supplemented with 100 μg/mL streptomycin, 100 units/mL penicillin, 250 ng/mL amphotericin B and 10% fetal bovine serum (FBS, Welgene, Daegu, Korea). Cells cultures were maintained at 37 °C humidified atmosphere (5% CO2 /95% air).
Preparation of minipillar array chips
Minipillar array chip was custom-made (MBD Co., Suwon, Korea). Detailed specification and preparation of the minipillar array chip has been reported [18, 19]. Briefly, the minipillar array chip consists of a pair of reusable base chips (top and bottom chip) and disposable minipillars (Additional file 1: Figure S1-a, b). Twenty-five minipillars were arranged in 9 mm-pitch for cell culturing in 96-well plates and in 3.2 mm-pitch for preparing cryo-sections (Additional file 1: Figure S1-c). The minipillars were sterilized by boiling in 70% ethanol for 30 min followed by ultraviolet irradiation (Ultraviolet Crosslinkers, UVP®, CX-2000). Base chips were autoclaved for sterilization.
Culture and drug expose condition for co-culture of TSs and PSCs
Cancer cells and PSCs were suspended at 8 × 105 cells/mL and 4 × 104 cells/mL, respectively, in 2.33 mg/ml collagen I solution (Rat tail tendon type I collagen, BD Biosciences, San Jose, CA) . Cancer cells were loaded on the tips of minipillars at 1.6 × 103 cells/2 μL and PSCs into each well of 96-well plates in 1.6 × 103 cells/40 μL, then, both were allowed to gel before adding culture media. Co-culture was done by transferring the pillar chip containing cancer cells to 96-well plates containing PSCs. Cancer cells grew into spheroids in the collagen matrix and medium change was done every 2 days. Drug treatment was done by exposing the cells after 6-days of culture in the drug-containing media for 72 h in 96 well plates. Cell viability was determined using acid phosphatase (APH) assay and calcein AM/PIstaining (BDA-1000, BIOMAX, Seoul, Korea), according to the procedure provided by the manufacturer.
Preparation of the frozen sections of TSs
The Frozen sections of TSs grown on minipillar chips were prepared as previously described. Briefly, TSs was frozen on minipillar chips in the vapor of liquid nitrogen. To prepare frozen blocks of TSs, minipillars were rearranged from 9 mm-pitch to 3.2 mm-pitch array (Additional file 1: Figure S1-c) and embedded in OCT compound using embedding jig assembly (EM001, MBD Co.) and freezing at − 20 °C for 20 min in a cryotome chamber. After 20 min, the pillar array chip was carefully removed, leaving the TSs collagen cap array, in the frozen OCT block. The frozen TS blocks were cut into 5 μm-thick sections.
Immunohistochemical staining and analysis
Immunostaining in TSs and PSCs was done either in 96-well plates during cultivation or on cryo-sections using antibody or fluorescent dye. For in-well staining, TSs were first fixed in 3.7% formaldehyde for 20 mins and further incubated with 0.5% Triton X-100 for another 30 mins. Exposure to primary antibodies against alpha-smooth muscle actin (α-SMA) (1:100, Abcam, ab5694), type I collagen (1:300, Abcam, ab34710), transforming growth factor beta-1 (TGF-β1) (1:50, Abcam, ab92486), membrane-type 1 matrix metalloproteinase (MT1-MMP) (1:100, Abcam, ab78738), vimentin (1:500, Abcam, ab92547) and wingless-related integration site-2 (Wnt2) (1:100, Abcam, ab27794) was done at 4 °C in a humidified chamber for 2 days. After blocking non-specific binding using 10% normal goat serum for 2 h, visualization was done using secondary antibodies conjugated with either Alexa Fluor 594 (1:1000, Thermo Scientific, A27016) or 488 (1:1000, Thermo Scientific, A11034). F-actin was stained with rhodamine phalloidin (1:1000, Invitrogen, R415). Samples were counter-stained with DAPI (1:2000, Sigma-Aldrich, D9564) and subjected for confocal microscopy (LSM 800 W/Airyscan, Carl Zeiss, Oberkochen, Germany).
Immunostaining on cryo-sections of TSs was preceded by collagen removal by 20 min-incubation in cold-DPBS followed by fixation in 95% ethanol for 15 min. Incubation with primary antibodies against α-SMA (1:100, Abcam), β-catenin (1:100, Cell Signaling, 8480), epithelial-cadherin (E-cadherin) (1:500, Cell Signaling, #3195), fibronectin (1:600, Abcam, ab2413), Ki-67 (1:400, Santa Cruz, sc-15,402), TGF-β1 (1:200, Abcam), type I collagen (1:600, Abcam) and vimentin (1:600, Abcam) was done at 4 °C in a humidified chamber overnight. After blocking non-specific binding using 10% normal goat serum for 60 min, visualization was done using secondary antibody conjugated with Alexa Fluor 594 (1:2000, Thermo Scientific). Sections were then counter-stained with DAPI and mounted for confocal microscopy (Carl Zeiss).
The fluorescence intensity was determined using ZEN software (Carl Zeiss). Morphological analysis and object counting was done by using ImageJ (NIH) software. The apparent diameter (D) of the TSs was calculated from the area measured using ImageJ and the equation, D = 2 × (area/π)1/2 assuming spherical shapes of TSs. Cell aggregates with a diameter greater than 40 μm were considered as full spheroids. Surface roughness of the TSs was expressed using circularity defined as, 4π × area/perimeter2. The morphological changes of spheroids and nuclei in cells invading into surrounding matrix were analyzed by using the aspect ratio defined as ratio of major vs minor axis. Structural organization of matrix fibers was analyzed for thickness and orientation. The orientation angle of collagen fibers was set at zero to the growing axis of invadopodia. The degree of cell invasion was determined by the number of dispersed single cells in the matrix outside spheroids. The regions of interest (ROI) for image analysis were selected randomly as the area containing 30–50 spheroids on the designated sections, unless otherwise indicated.
Western blot analysis
The cultured TSs were incubated in 0.1% collagenase D (Sigma-Aldrich, 11,088,858,001) at 37 °C for 30 min and lysed using a RIPA buffer. Total protein in lysates was quantified by BCA assay (Pierce™ BCA Protein Assay Kit, Thermo Scientific, #23227), 25 μg of total protein were resolved on 6 and 10% SDS-PAGE gel under reducing conditions, and transferred onto PVDF membranes (Immun-Blot® PVDF Membrane, BIO-RAD, #1620176). The membranes were blocked with 5% skim milk in TBS with 0.05% Tween-20 (TBS-T) at room temperature, and incubated with antibodies against α-SMA (1:1000, Abcam), β-catenin (1:500, Cell Signaling), E-cadherin (1:500, Cell Signaling), TGF-β1 (1:1000, Abcam), and vimentin (1:5000, Abcam) and β-actin (1:5000, Thermo Scientific, MA5–13739) at 4 °C overnight. After exposing to horseradish peroxidase (HRP)-conjugated secondary antibody, proteins were visualized by chemiluminescent substrate (SuperSignal™ West Pico PLUS, Thermo Scientific, #34580). The band intensity was analyzed using GeneTools software (Syngene, Cambridge, UK) and expression level was expressed as the ratio of target protein relative to β-actin. The value of the mono-culture group was set to 1 and the expression ratios of the co-culture group were compared.
Human secretome array analysis
The expression levels of chemokines and cytokines were analyzed using a Human Cytokine Antibody Array (C5) and a customized antibody array that includes ten factors known for their roles in EMT (tissue inhibitors of metalloproteinases-1, TIMP-1; urokinase plasminogen activator, uPA; plasminogen activator inhibitor-1, PAI-1; thrombospondon-1, TSP-1; MMP-2; MMP-9; E-cadherin; neural-cadherin, N-cadherin; Latent TGF-β1; stromal cell derived factor-1, SDF-1), (RayBiotech, GA). According to the manufacturer’s instruction, antibody-embedded membranes were incubated with 1 ml of conditioned media (CM) at 4 °C overnights, followed by incubation with biotin-conjugated detection antibody cocktail and diluted HRP-streptavidin at room temperature. Proteins were then visualized using a chemiluminescent substrate reagent. The signal intensities were quantified using GeneTools software.
Optimization of the 3D co-culture conditions for TSs and PSCs
Activation of PSCs during co-culture with PANC-1 TSs
ECM remodeling around PANC-1 TSs by PSC co-culture
The effect of PSC co-culture on growth and invasion of PANC-1 TSs
Induction of EMT in PANC-1 TSs by PSC co-culture
Changes in the secretome under PSC co-culture conditions
In the analysis of the EMT-related factors in CM, four out of eight cytokines (except two cell adhesion molecules, E-cadherin and N-cadherin among 10 proteins) underwent significant upregulation under PSC co-culture conditions (Fig. 6-b). Four other factors including, MMP-2, MMP-9, Latent TGF-β1, and SDF-1 were not detectable in either condition. It is noted that TGF-β1 detected in PANC-1 TS was the mature isoform of 12 kDa fragment (Fig. 5-a), hence the latent form was not detectable by cytokine array assay (Fig. 6-b).
Paclitaxel inhibits co-cultured cancer cell invasion by targeting PSC-mediated interactions
PSC exposure to GEM resulted in decreased viability (76% of the control) and cell length (88% of the control), which are similar to the magnitude of viability reduction (70% of control) in TSs (Fig. 7-b). On the other hand, PSCs being highly sensitive to PTX (3 μM) showed over 80 and 70% reductions in viability and cell length, respectively (Fig. 7-b and Additional file 3: Figure S3-b). In PSCs, reduced amounts of α-SMA and TGF-β1 were proportional to the drug-induced changes in viability and morphology for GEM and PTX treatment; these data point to the similar mechanisms of toxicity toward PSCs induced by these two drugs. Taken together, these data indicated that PTX, not GEM, exerted a strong inhibitory effect on cancer cell invasion and on proliferation of PSCs.
Among the four cytokines upregulated in the co-culture CM (Fig. 6-b), only three cytokines, namely TIMP-1, uPA, and PAI-1, after PTX exposure underwent significant downregulation approaching the level determined in the absence of PSCs (Fig. 7-c). Despite the inhibitory effect of GEM on PSC viability, no significant change was observed in the amounts of the four cytokines. Collectively, these data indicated that PTX exerted a superior effect on cancer cell invasion as compared to GEM. In the underlying mechanism, TIMP-1, uPA, and PAI-1, but not TSP-1, may be involved (Fig. 7-d).
In this study, we established an in vitro tumor model that recapitulates the TME of PDAC. Based on our previously developed method of hydrogel-embedded 3D cell culture by means of a minipillar array chip [18, 19], the culture method was modified to facilitate simultaneous reciprocal interactions between cancer cells and PSCs. We utilized a separate (indirect) co-culture method in which respective cell populations can be monitored without using either a particular separation procedure or specific cell probes . Changes in the phenotypic and molecular signatures of TSs and PSCs were evaluated either by in-well staining followed by optical sectioning microscopy or by conventional slide staining of mechanical sections (cryo-sections and paraffin-embedded sections), for which our pitch-tunable minipillar array chip provides a most suitable platform [18, 19]. Mechanical sections are useful when either limited penetration of antibodies or light scattering hinders proper staining or 3D imaging, respectively, within a dense 3D architecture of a TS. By optimizing the culture conditions, sample preparation, and the 3D imaging protocol, we were able to analyze ECM remodeling, cell invasion, and drug resistance induced by 3D interactions between cancer cells and PSCs in an efficient and clinically relevant way.
Several pancreatic cancer cell lines are commercially available and have been characterized regarding their different mesenchymal states [25, 26]. We confirmed the differential profiles of E-cadherin and vimentin expression among five PDAC cell lines, i.e., MIAPaCa-2, AsPC-1, PANC-1, Capan-1, and BxPC-3 (Fig. 1-b). A correlation between formation or compactness of TSs and the level of E-cadherin expression has been observed for these cell lines. When growing 3D in ultra-low attachment plates, Capan-1 and BxPC-3 cells were aggregated to form spheroids, whereas MIAPaCa-2, AsPC-1, and PANC-1 cells did not (Additional file 4: Figure S4). When cultured in a collagen gel using minipillar chips, each of the five cell lines aggregated to form TSs, yet the surface roughness appeared to reflect their EMT states (Fig. 1-b). A similar relation was demonstrated for head and neck cancer cells grown by liquid overlay methods and for breast cancer cells grown embedded in a hydrogel [27, 28]. In addition, drug sensitivity has been linked to the EMT status of cells . Mesenchymal-type PDAC cells showed higher resistance to GEM as compared to the epithelial type in monolayers (MIAPaCa-2, AsPC-1, and PANC-1 cells: IC50 ≥ 10 μM; Capan-1 and BxPC-3 cells: IC50 ≤ 0.1 μM; Additional file 5: Figure S5) as well as in 3D cultures (PANC-1 TSs, IC30 = 180 μM; BxPC-3 TSs, IC30 ≤ 10 μM; Additional files 3 and 6: Figure S3-a, S6).
PSCs were activated when co-cultured with TS in our 3D model (Fig. 2). Expression level of vimentin and Wnt2 was significant, but was not affected by interaction with TS (Additional file 7: Figure S7), whereas morphologic changes to a further elongated spindle shape with stress fiber formation was observed with increased expression of α-SMA and TGF-β1 under TS co-culture conditions (Fig. 2). Activated PSCs act as CAFs to promote cancer cell proliferation and migration . CAFs induce matrix deposition and proteolytic remodeling of ECM by producing proteases and their inhibitors such as TIMP-1, TIMP-2 and uPA (Fig. 6) . CAFs also stimulate cancer cell migration via paracrine signaling of IL-6, IL-8, insulin like growth factor-1 (IGF-1), CCL2, CCL4, CCL5, CCL22, CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, TSP-1, PAI-1 osteopontin (OPN) and osteoprotegerin (OPG), all of which were shown to increase in the CM under PSC co-culture conditions (Fig. 6) [31, 32].
Due to tumor-promoting effect, targeting CAFs with Hedgehog (Hh) inhibitors has been considered a promising anti-tumor strategy . Clinical trials evaluating Hh inhibitors, however, failed to show clinical benefit in PDAC whereas two Hh inhibitors (inhibiting smoothened) , vismodegib and sonidegib have been approved to treat basal cell carcinomas . Although the role of Hh signaling pathway in PDAC and CAFs’ functional heterogeneity remains controversial [36, 37], therapeutic potential of stromal CAF targeting warrants further studies using more clinically relevant models .
Cancer cell invasion was accompanied by ECM remodeling when TSs were co-cultured with PSCs (Figs. 3 and 4). In TSs co-cultured with PSCs, formation of fibrous ECM bundles may be attributed to changes in collagen fiber orientation (Fig. 3); this notion supports the usefulness of our model for studying the plasticity of matrix and reciprocal of cell-matrix interactions . The mechanism underlying this ECM remodeling involves matrix degradation by proteases such as MMPs (Additional file 2: Figure S2-b) and ADAMs (A disintegrin and metalloproteinases) secreted by cancer cells and PSCs and internalization by collagen receptor uPARAP/Endo180 followed by deposition of the fibrillar form of collagen [40, 41]. Increased deposition of fibrous matrix proteins as part of ECM remodeling has been linked to reduced drug accumulation in tumor tissue [42, 43]. In our co-culture model, doxorubicin penetration was reduced in PANC-1 TSs by co-culturing with PSCs (Additional file 8: Figure S8), indicating that excessive ECM deposition acted as a physical barrier to diffusion or convective movement of drug molecule. Under PSC co-culture conditions, cancer cell invasion measured by dispersed single cell numbers was accompanied by increase in aspect ratio of TS (Fig. 4). The changes in the aspect ratio were not due to increase in size or cell death in either mono- and co-culture conditions (Additional file 9: Figure S9).
One of the mechanisms via which TGF-β1 enhances mesenchymal traits in cells is to produce several secreted proteins including growth factors . TIMP-1, uPA, and PAI-1 are some of the mediators of TGF-β1-induced EMT in cancer [45, 46]. TIMP-1 is associated, independently of its MMP-inhibitory function, with the induction of the EMT phenotype and mediates cancer-stellate cell interactions . uPA plays an important role in ECM remodeling by converting plasminogen to plasmin for direct degradation of ECM components or activates latent TGF-β1 . PAI-1 is known as the main inhibitor of the uPA-uPA receptor (uPAR) complex but is overexpressed in CAFs and enhances cell migration in an uPA-uPAR-independent manner [49, 50]. TSP-1, a major inducer of TGF-β1 , is expressed in CAFs surrounding tumor cells and is involved in the regulation of ECM remodeling during tumor invasion [52, 53]. Changes in the EMT signature of cancer cells and PSCs were accompanied by increased expression of TGF-β1 (Figs. 2-b, 5-a). We also observed increased concentrations of four TGF-β1-related soluble factors such as TSP-1, uPA, PAI-1, and TIMP-1 in the CM under PSC co-culture conditions (Fig. 6-b). Thus, it is demonstrated that our model recapitulates the ‘reciprocal interactions’ between cancer cells and stellate cells, through which mutual activation was induced and resulted in the secretion of cytokines for enhancement of ECM remodeling and cell invasion.PTX caused significant suppression of cancer cell invasion as compared to GEM in PANC-1 TSs co-cultured with PSCs (Fig. 7-a). PTX is known to bind to and stabilize microtubules thereby inhibiting their dynamics and resulting in mitotic arrest and cell death . Nonetheless, the anti-invasive effect of PTX observed in the present study can be explained by the mitosis-independent mechanism in both PANC-1 cancer cells and PSCs. Microtubules are known to cooperate with actin and vimentin intermediate filaments for elongation of invadopodia . Hence, microtubule-stabilizing effect of PTX significantly suppresses cytoskeleton network remodeling in invadopodia maturation and subsequent cancer cell invasion (Fig. 7-a). In addition, the microtubule network negatively regulates TGF-β signaling by sequestering Smads [55, 56]. Hence, reduced mesenchymal signature in PANC-1 TSs and PSCs after PTX exposure (Fig. 7) can be explained by inhibition of TGF-β-Smad signaling via PTX-induced microtubule stabilization. PTX-induced inhibition of TGF-β signaling was reflected in the changes of secreted cytokine levels after PTX exposure, where mediators of TGF-β signaling such as TIMP-1, uPA, and PAI-1 underwent significant downregulation, but not TSP-1, a major activator of TGF-β1 , (Figs. 7-c, d). From the standpoint of non-mitotic anti-invasion activity of PTX, the mechanism behind the clinical synergism between albumin-bound PTX (nab-PTX) and GEM is worth further investigation .
We established co-culture of pancreatic TSs and PSCs by means of minipillar array chips and present it here as a novel in vitro model of a human PDAC. Our model recapitulates the 3D interactions between cancer-cancer cells, cancer cell-ECM interactions, and cancer cell-PSC interactions. Our model is demonstrated to be useful for studying the effects of PSCs on ECM remodeling and cancer cell invasion and the mechanisms involved. Drug response was successfully analyzed in a cell type-specific and quantitative manner. Overall, our co-culture model may represent a novel method not only for studying TME interactions but also for evaluating therapeutic agents targeting these interactions [58, 59, 60, 61].
This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MEST) (Nos 2016R1A2B2011832 and 2012R1A5A2047939).
Availability of data and materials
All data generated or analysed during this study are included in this published article and its supplementary information files.
Conceived and designed the experiments: HJH and H-JK Performed the experiments: HJH Analyzed the data: HJH and H-JK Contributed reagents/materials/analysis tool: HJH, M-SO, DWL and H-J K Wrote the paper: HJH and H-JK All authors contributed to the preparation of the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Consent for publication
Patent applications and registration for the minipillar array chip have been filed by the Catholic University of Korea. The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
- 49.Bagordakis E, Sawazaki-Calone I, Macedo CC, Carnielli CM, de Oliveira CE, Rodrigues PC, et al. Secretome profiling of oral squamous cell carcinoma-associated fibroblasts reveals organization and disassembly of extracellular matrix and collagen metabolic process signatures. Tumour Biol. 2016;37:9045–57.CrossRefGoogle Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.