Abstract
Cytotoxic efficacy of anticancer drugs has been widely studied with monolayer-cultured cancer cells. However, the drug efficacy under two-dimensional (2D) culture conditions usually differs from that of three-dimensional (3D) culture conditions. In the present study, an in vitro tumor tissue model was constructed by sandwiching a cancer cell monolayer (cancer cell sheet) between two collagen layers as a biomimicking tumor tissue model, and the in vitro cytotoxic efficacy of doxorubicin was investigated. In the biomimicking tumor tissue model, hypoxic areas were observed, and the secretion of vascular endothelial growth factor increased time-dependently. Additionally, cell growth inside the model was significantly slower than that of the 2D culture system. The in vitro cytotoxicity of doxorubicin with the 3D system decreased compared to the 2D system, suggesting that the surrounding collagen layer acted as a barrier much like cancer stroma does. Consequently, this study successfully demonstrated that the cancer cell sheet in collagen sandwich configuration provides a useful in vitro tumor tissue model.
The cytotoxicity of anticancer drugs and drug candidates has conventionally been investigated with in vitro two-dimensional (2D) systems, having monolayer-cultured cancer cells, and in vivo human tumor-bearing animal models (xenograft models) (1, 2). However, in vitro 2D cell-culture systems are known to be limited in terms of mimicking the native in situ environment of cancer (e.g. tumor stroma, or nutrient and oxygen supplies) (3). On the other hand, tumor xenograft models are a gold standard that can provide a native three-dimensional (3D) microenvironment (4). Animal experiments are usually expensive, and there are ethical concerns regarding use of animals in research (5, 6). Therefore, the development of in vitro models has been demanded instead of animal experiments. The following typical in vitro 3D tumor models have been reported: (i) tumor spheroid, (ii) natural scaffold (e.g. collagen hydrogel, hyaluronic acid, or silk protein fibers), and (iii) synthetic scaffold (e.g. polylactide, polyglycolide, or polyethylene glycol) (5, 7-12). In particular, collagen hydrogel has been frequently used for tissue engineering applications (13).
Cancer cells, such as pancreatic cancer and hepatic cancer cells, were reported to be surrounded by collagen I and collagen IV (14-16). Recently, encapsulated cancer cells with collagen gels have attracted attention as in vitro tumor tissue models for mimicking in vivo cancer environments. However, biological functions of dispersed cells are limited due to the lack of tissue-like structure and of cell–cell interactions. Therefore, we designed an in vitro tumor-tissue model using a cancer cell monolayer (cancer cell sheet), which possesses tissue-like architecture in contrast to isolated single cells (17, 18). Herein, we focused on the construction of an in vitro model of tumor tissue by sandwiching a hepatocellular carcinoma cell sheet (tissue-like form), maintaining cell–cell interactions between two collagen layers. Furthermore, the in vitro cytotoxicity of doxorubicin, as a representative anticancer drug, in the tumor tissue model was compared to that of 2D cell culture system.
Materials and Methods
Materials. Collagen type I, minimum essential medium (MEM), and Alamar blue reagent were purchased from Life Technologies (Carlsbad, CA, USA). Hypoxyprobe-1 kit was from Natural Pharmacia International (Belmont, MA, USA). Dulbecco's phosphate-buffered saline without calcium chloride and magnesium chloride (PBS) was from Iwaki Glass (Chiba, Japan). Fetal bovine serum albumin (FBS) was from Moregate Biotech (Bulimba, Australia). Antibodies against vascular endothelial growth factor (VEGF) and hypoxia-inducible factor-1α (HIF1α) were from Abcam (Cambridge, UK). Dulbecco's modified Eagle's medium (DMEM), doxorubicin hydrochloride (DOX), and other chemicals were from Wako Pure Chemical (Osaka, Japan).
Preparation of hepatocellular carcinoma (HLE) cell sheet and in vitro tumor-tissue model. A: Preparation of HLE cell sheet. Scale bar indicates 10 mm. B: Scheme of the preparation of HLE cell sheet cultured in a collagen sandwich system (in vitro tumor-tissue model). (i) Collagen solution was added to a mold and incubated at 37°C for 30 min to create a collagen layer. Then the HLE cell sheet was put on the surface of the collagen layer. (ii) The HLE cell sheet was covered with collagen solution and incubated at 37°C for 30 min. After gelling, the tumor cell-sheet tissue model was transferred onto a 60-mm petri dish and cultured in Dulbecco's modified Eagle's medium.
Cancer cell-sheet preparation. Human hepatocellular carcinoma (HLE) cells (Health Science Research Bank, Osaka, Japan) were seeded onto an UpCell® (Cell Seed, Tokyo, Japan) thermoresponsive culture dish, at a density of 2×105 cells/cm2 and cultured with DMEM supplemented with 10 v/v% FBS at 37°C in a humidified atmosphere with 5% CO2 for 4 days. The medium was changed every 2 days. After a 4-day culture, the dish was incubated at 20°C for 30 min to detach confluent HLE cells as a contiguous cell sheet.
Preparation of the in vitro tumor-tissue model. As shown in Figure 1B, after neutralizing using 1 mol/I NaOH, a collagen solution (5 mg/ml) was mixed with 10% v/v minimum essential medium (MEM) [10-times concentrated (10×)], 10% v/v PBS (10×), and deionized water. The prepared collagen solution was then poured into a 27-mm diameter plastic mold and incubated at 37°C for 30 min to make a collagen layer. After a 30-min incubation, the HLE cell sheet was put on the surface of the collagen layer and covered with further collagen solution. The cancer cell sheet was completely sandwiched and sealed between the two collagen layers to limit the diffusion of culture medium and oxygen. The HLE cell sheet sandwiched in collagen was defined as an in vitro tumor-tissue model. The model was then transferred onto a 60-mm petri dish and cultured for up to 21 days in DMEM.
Morphology of HLE cell sheet cultured in a collagen sandwich system. Hematoxylin and eosin was used to stain the tumor cell sheet tissue model on day 7 (A) and 21 (C). Cells infiltrating the collagen layer are indicated by the yellow arrowheads. Hypoxic areas shown by staining for pimonidazole were found in the tumor-tissue model on day 7 (B) and 21 (D). Scale bars indicate 100 μm.
Immunohistochemical staining for hypoxia-inducible factor-1α (HIF1α) and vascular endothelial growth factor (VEGF), and VEGF secretion in in vitro tumor tissue model. Immunohistochemical staining for HIF1α and VEGF in the in vitro tumor tissue model were observed at day 7 (A, B) and 21 (C, D), respectively. Scale bars indicate 100 μm. HIF1α-positive areas are indicated by red arrows. VEGF-positive areas are indicated by blue arrows. E: Time-dependent VEGF release from the in vitro tumor-tissue model. Samples were collected on 1, 5, 10, and 14 days after initiating culture. Values represent the mean±S.D. (n=3). Significantly different at *p<0.05 and **p<0.01 compared to the value on day 1.
Immunohistochemical staining. In vitro tumor-tissue models were fixed with 10% v/v formalin at 4°C for 24 h and embedded in paraffin. Paraffin-embedded blocks were sliced into 8-μm-thick sections, which were dehydrated through xylene and ethanol, and stained with hematoxylin and eosin.
For detecting specific areas with a low oxygen level (less than 10 mmHg), the in vitro tumor-tissue model was immersed in 100 μmol/l pimonidazole hydrochloride solution for 2 h before fixing. Once fixed, the model was embedded in paraffin, and sliced into 8-μm-thick sections, which were treated with an antibody to pimonidazole at a dilution of 1:50 for 40 min at room temperature. After washing with PBS three times, the sections were stained with ChemMate ENVISION kit/HRP (Dako, Carpinteria, CA, USA). For immunohistochemical detection of VEGF and HIF1α, the sections were blocked with Block One Histo (Nacalai Tesque, Kyoto, Japan) at room temperature for 1 h. Then antibodies against VEGF at a dilution of 1:500 and HIF1α at a dilution of 1:200 were applied at 37°C for 2 h. After incubation, the sections were washed with PBS three times and stained with 3,3’-diaminobenzidine.
Cell growth and anticancer drug efficacy. For the 2D cell culture system, HLE cells were seeded in a 35-mm dish at a density of 8×104 cells/cm2 (day 0) and maintained at 37°C for up to 7 days. The fluorescence intensity on day 0 without DOX was defined as 100% cell viability. On days 1 and 7, fluorescence intensity was measured by the Alamar blue method as described below. The relative fluorescence intensity as a percentage for each day was calculated based on the fluorescence intensity on day 0. For measuring the anticancer efficacy in the 2D culture system, DOX was added to the culture medium. After culturing, the fluorescence intensity was measured on each day.
In the case of the in vitro tumor-tissue model, the model was transferred to a 60-mm petri dish on day 0 and was exposed to DMEM with, and without DOX (1 μmol/l), and the fluorescence intensity was measured by the Alamar blue method. The cell viability as a percentage was calculated by the same method as in the case of the 2D culture.
Alamar blue assay. Alamar blue reagent was diluted at 1:10 with phenol red-free DMEM (Life Technologies) and added to the cell culture dish. After incubation for 45 min at 37°C, the fluorescence intensity of the solution was measured with a SpectraMax M2 microplate reader (Molecular Devices, Sunnyvale, CA, USA) at an excitation and emission wavelength of 544 and 590 nm, respectively.
Detection of VEGF in the culture medium. The culture media from culture dishes of the in vitro tumor-tissue model were individually collected on days 1, 5, 10, and 14 during cell culture and immediately centrifuged at 300 × g for 5 min at 4°C to separate the debris from the supernatant. VEGF concentration was quantified using a VEGF enzyme-linked immunosorbent assay (ELISA) kit (Merck Biosciences, Darmstadt, Germany) according to the manufacturer's procedure. The color changes were quantified at 450 nm with a SpectraMax M2 microplate reader.
Statistical analysis. Data are expressed as the mean±SD. The statistical significance of differences was evaluated using Student's t-test. A probability of less than 0.05 was considered significant.
Cell proliferation and antitumor efficacy of doxorubicin (DOX). A: Cell proliferation and antitumor efficacy of DOX (1 μmol/l) under a monolayer culture condition. B: Cell proliferation and antitumor efficacy of DOX (1 μmol/l) in the in vitro tumor-tissue model. The fluorescence intensity on day 0 without DOX was defined as 100% cell viability. The relative fluorescence intensity as a percentage for each day was calculated based on the fluorescence intensity on day 0. Values represent the mean±S.D. (n=3).
Results
Preparation of cell sheet and the in vitro tumor-tissue model. By reducing the culture temperature to 20°C for 30 min, confluently cultured HLE cells peeled-off from the surface of the thermoresponsive dish as a contiguous cell sheet (Figure 1A). The diameter of the circular HLE cell sheet was 19.0±0.4 mm (n=3). The in vitro tumor-tissue model (Figure 1B) had a height and diameter of 9.8±0.3 mm and 27 mm, respectively. Cells infiltrating into the collagen layer were observed (yellow arrows in Figure 2A and C). However, no cancer cell migration outside of the gel layer was observed up to 21 days. In addition, degradation of collagen gels containing cancer cell-sheet was not observed during these 21 days. Sectional specimens of the in vitro tumor-tissue model were treated with pimonidazole and antibody against pimonidazole to clarify the internal oxygen level. As shown in Figure 2B and D, a pimonidazole antibody-positive hypoxic area (brown staining) was observed at the center of the tumor cell-sheet tissue model during the experiment.
Immunohistochemical staining and VEGF secretion. VEGF and HIF1α were detected by immunohistochemical staining with respective antibodies (Figure 3). On day 7 and day 21, HIF1α-positive areas (red arrow in Figure 3) were found to correspond to VEGF-positive areas (blue arrow in Figure 3). However, HIF-1α-positive areas and pimonidazole-positive areas did not correspond with each other (Figure 2B and 3A). HIF-1α-positive areas were located around the hypoxic regions.
The time course of VEGF concentration in culture medium was determined by an ELISA assay (Figure 3E). The concentration on days 5, 10, and 14 was approximately 3.5-, 9.2- and 8.4-fold greater, respectively, than that on day 1. A collagen layer without a cell sheet failed to secrete VEGF into the culture medium. The amount of VEGF secreted from cancer cells under 2D culture conditions at day 7 was 478.0±36.3 pg/ml, which was lower than that of the in vitro tumor-tissue model on day 1.
Comparison of cytotoxicity of doxorubicin under 2D culture conditions and that of the in vitro tumor-tissue model. Cell proliferation and the cytotoxicity of DOX (1 μmol/l) were investigated by the Alamar blue assay comparing the in vitro tumor tissue model to conventional 2D cell culture. The calibration curve of fluorescence intensity vs. HLE cell number was prepared for the assay. There was a significant positive correlation of the fluorescence intensity with the HLE cell number (r=0.98, p<0.001). The fluorescence intensity on day 0 without DOX was defined as 100% cell viability and used as a reference for that measured at other time points. Under 2D culture conditions, the cell growth on days 1 and 7 was 160% and 341%, respectively. In terms of anticancer efficacy, DOX (1 μmol/l) treatment dramatically reduced cell viability to 4%. In the case of the in vitro tumor-tissue model, the cell growth on days 1, 4, and 7 were 104%, 121%, and 145%, respectively. In addition, the anticancer effect of DOX (1 μmol/l) treatment was found to be mild.
Discussion
This study prepared an in vitro tumor-tissue model using a cancer cell sheet sandwiched between two layers of collagenous matrix and demonstrated a considerable difference in DOX cytotoxicity compared to the conventional 2D cell culture model. Recently, Nyga et al. reported developing a colorectal cancer model with collagen gel as an artificial cancer mass in order to mimic an in vivo microenvironment (19). In that study, HT29 colorectal cancer cells mixed in collagen gel became spheroid, moved to the edge of gel, and migrated to the surrounding areas on day 14 (19). In the present study, we also observed the migration of isolated HLE cells inside the collagen gel. However, cell migration was scarcely observed outside of the in vitro tumor-tissue model. We estimated that the cell sheets possess biologically intact structures maintaining cell–cell interactions unlike dispersed single cells, and thus the tissue-like structure (densely cell-packed structure with extracellular matrix) of the cancer cell sheet probably prevented cell migration.
The in vitro tumor-tissue model was able to maintain hypoxic conditions surrounding the cancer cells for a long period of time, more than 14 days and less than 21 days. In other words, cancer cells in the in vitro tumor tissue model were cultured in a hypoxic condition for up to 21 days, and the anticancer efficacy was tested under this condition. Microscopic observation of hematoxylin and eosin-stained sections revealed that the cancer cells in the in vitro tumor-tissue model increased in a time-dependent manner in the collagen gel (Figure 2A and C). By immunohistochemical observation, hypoxic regions were detected (Figure 2B and 2D).
Additionally, pimonidazole-positive areas were found next to HIF1α-positive areas. The lack of co-localization between pimonidazole-positive areas and HIF1α-positive areas was observed (Figure 2B and 3A). This result agreed with a clinical report by Jiang et al. (20). The mass of cells at the center of tumor is well known to be hypoxic because of immaturity and a tortuous vascular structure. Dysfunctional blood vessels are unable to supply sufficient nutrients and oxygen to cancer cells, resulting in hypoxic conditions (5). Hypoxic areas are divided into two groups in terms of their distance from blood vessels: (i) far areas, with a distance of approximately 70-85 μm from blood vessels and (i) farther areas, with a distance of 85-100 μm (21). The far area found in that study was HIF1α-positive and pimonidazole-negative. In contrast, the farther area was HIF1α-negative and pimonidazole-positive because of oxygen diffusion, glucose concentration, pH etc. Therefore, the in vitro tumor-tissue model had a hypoxic area in the central cancer mass, similarly to real human tumor.
The measurement of VEGF concentration in culture medium revealed that the concentration increased from day 1 to 5, peaked on day 10, and was then maintained until day 14. Previously, cancer cells were reported to frequently display different gene expression profiles under 2D and 3D conditions (e.g. VEGF and interleukin-8) due to differences in nutrient or oxygen supplies (22, 23). VEGF secretion in the in vitro tumor-tissue model increased, suggesting possibly different gene expressions under 2D culture conditions and in the in vitro tumor-tissue model.
The growth rate of HLE cancer cells in the in vitro tumor-tissue model was significantly slower than that under 2D culture conditions (Figure 4A). In addition, the anticancer drug efficacy in the 2D culture system was higher than that in the in vitro tumor-tissue model (Figure 4A). Generally, nutrient and oxygen supplies to cancer cells in a 2D culture environment are more highly sufficient than those of an in vivo or a 3D culture environment (24). Cancer cell growth under the 2D culture conditions was more aggressive than that of the in vitro tumor-tissue model. 3D culture models reportedly have slower cell proliferation rates, that are relatively similar to those of the in vivo situation than those under 2D culture conditions (19, 24). An anticancer drug in 2D culture easily accesses the cancer cells and kills them. In vivo or under 3D culture conditions, however, the efficiency of anticancer drugs decreases due to the ambient structural effect of cancer cells. In addition, cancer stroma-inhibited anticancer drug delivery to cancer cells was also reported (19, 25). Therefore, the cytotoxicity of anticancer drugs towards cancer cells cultured in a collagen sandwich system is weaker than that under 2D culture conditions. In the present study, the cytotoxic effect of DOX in the in vitro tumor tissue model was milder than that in the 2D culture system, suggesting that the surrounding collagen layers acted as a barrier to limit anticancer drug diffusion, like cancer stroma does in real tumor tissues.
In conclusion, we designed an in vitro 3D tissue model to construct tumor-mimicking material by sandwiching a contiguous cancer cell sheet in two-collagen layers. In this model with artificial cancer stromal barrier, cell growth was quite slow, and cytotoxic efficacy of anticancer drugs was significantly reduced in comparison to that under 2D culture conditions. The cancer cell sheet in collagen configuration may provide a useful in vitro tumor tissue model for applications in cancer biological studies and anticancer drug-screening tests.
Acknowledgements
This work was supported by JSPS KANENHI Grant Number 25282145, and the Global COE program, Multidisciplinary Education and Research Center for Regenerative Medicine (MERCREM), from the Ministry of Education, Culture, Sports Science, and Technology (MEXT), Japan.
- Received September 18, 2015.
- Revision received October 18, 2015.
- Accepted October 23, 2015.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.