Adv Healthc Mater. 2017 Aug;6(15). doi: 10.1002/adhm.201700057.

Desmoplastic Reaction in 3D-Pancreatic Cancer Tissues Suppresses Molecular Permeability.

Matsusaki M1,2, Komeda M1, Mura S3, Tanaka HY4, Kano MR4, Couvreur P3, Akashi M1.

1 Department of Applied Chemistry, Graduate School of Engineering, Osaka University, 2-1 Yamadaoka, Suita, Osaka, 565-0871, Japan.
2 JST, PRESTO, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan.
3 Institut Galien Paris-Sud, UMR 8612, CNRS, University of Paris-Sud, Université Paris-Saclay, Facultéde Pharmacie, 5 rue jean-Baptiste Clément, F-92296, Châtenay-Malabry, Cedex, France.
4 Department of Pharmaceutical Biomedicine, Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University, 1-1-1 Tsushima-naka, Kita-ku, Okayama-shi, Okayama, 700-8530, Japan.


The survival rate of pancreatic ductal adenocarcinoma is still the lowest among all types of cancers, primarily as a consequence of an important desmoplastic reaction. Although the presence of thick stromal tissues in pancreatic tumors has been reported, in vivo animal studies do not enable a clear understanding of the crosstalk between cancer cells and fibroblasts. Accordingly, this paper reports the design and characterization of an in vitro pancreatic cancer-stromal 3D-tissue model, which enhances the understanding of the interactions between cancer cells and fibroblasts and their influence on the secretion of extracellular matrix (ECM). 3D-tissue models comprising fibroblasts and pancreatic cancer cells (MiaPaCa-2 cell line) or colon cancer cells (HT29 cell line, used as a control) show decreased molecular permeability with increased cancer cell ratios. The 3D-MiaPaCa-2 tissues display an increase in the secretion of collagen as a function of the cancer cell ratio, whereas 3D-HT29 tissues do not show a significant difference. Notably, the secretion of ECM proteins from single fibroblasts in 3D-tissue models containing 90% MiaPaCa-2 cells is ten times higher than that under 10% cancer cell conditions. In vitro pancreatic cancer 3D-tissues will be a valuable tool to obtain information on the interactions between cancer and stromal cells.


extracellular matrix secretion; pancreatic cancer; permeability; tissue engineering

PMID: 28452178



Tumor-derived cancer cells cultured in vitro 2D monolayer are sensitive to anti-cancer drugs, but the same drug molecules often fail to suppress tumor growth in vivo. For example, in the case of two dimensional (2D) monolayer culture of pancreatic cancer cells, gemcitabine exhibits potency,[1] however the same is not true of a xenograft in vivo model,[2] may be due to the different availability of the anticancer drug. A major obstacle to deliver efficacious drug dose in vivo might is represented by the dense fibrosis surrounding cancer cells.[3,4] Thus, although therapeutic drugs loaded in drug delivery systems (DDS), may successfully attain the tumor via the the tumor blood vessels, by they have to cross a thick wall (stromal tissue) in order to reach the tumor cells. To date, there is no satisfactory in vitro modeling of drug permeability in fibrotic tissues.

In this study, we report the construction and characterization of in vitro pancreatic cancer 3D-tissues with the aim to understand the interaction between cancer cells and fibroblasts and the influence of this cross-talk on the secretion of extracellular matrix (ECM) proteins, especially type I collagen (Figure 1). The constructed 3D-tisseus composed of normal human dermal fibroblasts (NHDFs) and pancreatic cancer cells (MiaPaCa-2) or colon cancer cells (HT29, used as control) showed a decrease in the permeability of a model drug (i.e., dextran) with the increase of the cancer cell ratio. The 3D-MiaPaCa-2 tissues displayed an increase of collagen and the other ECM proteins secretion as function of the cancer cell ratio, whereas 3D-HT29 tissues did not show any significant difference. Interestingly, secretion amount of collagen or ECM/fibroblast in the 90% MiaPaCa-2/10% NHDF 3D-tissues was approximately 10-times higher than that of 10% MiaPaCa-2/90% NHDF 3D-tissuesof cancer cell condition. To the best of our knowledge, this is the first example of an in vitro engineered tumor model capable to reveal the communication between cancer cells and fibroblasts and the influence on the development of a desmoplastic reaction. Thus the pancreatic cancer-stromal 3D-tissues will be a valuable in vitro tool to understand cancer-stromal cell interaction in detail.



Figure 1.Outline of this study. Schematic illustration of interaction between pancreatic cancer cells and stromal cells in 3D- cancer-stromal tissues constructed by cell-accumulation technique.



Figure 2.Cancer cell effect on molecular permeability. (a) Permeability assay of FD4k and 2000k molecules across 8L layered-MiaPaCa-2 and HT29 stromal tissues. The cancer cell ratio was varied at 0 (open circle), 10 (open square), 50 (open diamond), 90 (open triangle), and 100% (closed circle) conditions. (b) The relationship between cancer cell ratio and permeability of FD2000k at 24 h of incubation. The bottom shows the illustration of this relationship. Asterisks denote statistically significant difference using a two sample student’s t-test (**p < 0.05, *p < 0.01) for each comparison (n = 3~6).


To evaluate the effect of stromal fibroblasts to molecular permeability, permeability of FITC-labeled dextran (FD) with different molecular weights on 3D-cancer-stromal tissues with varied cancer cell ratio was investigated. After 2 days of incubation, two types of FD molecules, their molecular weights were 4k Da (FD4k) and 2,000k Da (FD2000k), were added to top culture media in 24 well inserts with each 3D-tissues and corrected bottom culture media at predetermined time. Hydrodynamic diameters of FD4k and 200k were about 1 and 20 nm, respectively, analyzed by dynamic laser scattering (DLS). First, we evaluated permeability of 24 well culture insert itself (without cells) until 24 hours. The FD4k indicated 75% of permeability after 24 hours of incubation, whereas 2000k showed only about the half values (36%). In the case of FD4k, all 3D-MiaPaCa-2 tissues showed almost the same approximately 50% permeability after 24 hours of incubation (Figure 2). The 3D-HT29 tissues indicated the same tendency, however only 100% HT29 tissues revealed the lowest permeability might be due to tight junction formation. Since the FD4k indicated almost the same permeability independent of existence of 3D-tissues or cancer cell type, we concluded that FD4k was not adequate to evaluate effect of tissues or cancer cell character due to tiny molecular size. On the other hand, FD2000k revealed a correlation between decrease of permeability and cancer cell ratio in both cancer cells. However, 3D-100%MiaPaCa-2 tissues only showed independent higher permeability due to the unassembled tissue structure. 3D-MiaPaCa-2 tissues indicated higher dependency on cancer cell ratio than 3D-HT29 tissues and these data clearly suggested increase of a permeability suppression factor in 3D-tissues with the increase of cancer cell ratio (Figure 2b). We speculated that the desmoplastic reaction, influence of pancreatic cancer cells to enhance secretion of ECM from fibroblasts, occurred in 3D-MiaPaCa-2 tissues. Thus, we next tried to investigate the secretion of ECM proteins in the 3D-tissues.

The secretion amount of collagen and the other ECM proteins was detected by collagen staining kit (Figure 3). 3D-MiaPaCa-2 tissues showed increase of the secreted collagen with the increase of cancer cell ratio and the non-collagen proteins were drastically secreted at over 50% of cancer cell ratio. Only 100%MiaPaCa-2 tissues indicated drastic decrement might be due to the unassembled structures. On the other hand, HT-29 tissues did not show any significant difference on both collagen and the other ECM proteins. Since type I collagen has been reported as the major collagen type in stromal tissues, we also detected the secreted amount of collagen type I using a ELISA kit. The secreted amount of collagen type I from 3D-MiaPaCa-2 tissues revealed the same increase manner, whereas HT29 tissues also showed the same trend. The secreted type I collagen was approximately one third in all secreted collagens and 1.5% in all secreted ECM proteins even at this short culture period (only 3 days).

When cancer cell ratio increased in the 3D-tissues, the ratio of fibroblasts which is the main player for ECM secretion decreased although the secreted amount of ECM was increased. The phenomena strongly suggested the increase of ECM secretion at single fibroblast with the increase of cancer cell ratio, desmoplastic reaction. Accordingly, we estimated the secreted amounts of ECM from single fibroblast cell at varied cancer cell ratio (Figure 3d-f). Surprisingly, the fibroblasts in 3D-90%MiaPaCa-2 tissues expressed significantly higher amount of collagens and non-collagen ECM proteins and the value was about 10 times higher than that of HT29 tissues. This result clearly indicates the strong influence of pancreatic cancer cells to fibroblast on ECM secretion. To understand the mechanism of this phenomenon, mRNA expression was investigated.

The mechanism of desmoplasia in pancreatic tumors is still not clear yet, but inhibition of transforming growth factor-β (TGF-β) signaling,[5] hypoxia,[6] fibroblast growth factor-2 (FGF-2),[7] and sonic hedgehog (SHH)[8] are important candidate pathways. Since TGF-β inhibition pathway is well known pathway, we here evaluated mRNA expression of SMAD7, a TGF-β inducible antagonist of TGF-β signaling, as well as COL1A1, type I collagen, in 3D-tissues with or without 90%MiaPaCa-2 condition. The relative expression of SMAD7 was predictably suppressed in 3D-MiaPaCa-2 tissues, whereas COL1A1 expression did not show statistical difference. In order to understand clearly the effect of TGF-β signaling, TGF-β inhibitor, LY364947, was added to the culture media for 3 days of incubation. Secretion of collagen type I slightly decreased at 100 nM condition, there was no significant difference by comparison with the control condition. Although the other types of inhibitor, SB-431542 also indicated no significant difference to collagen type I expression. Although our data disclaimed TGF-β inhibition pathway, we believe the remaining possibility of this mechanism because of our short-term experimental period. We compared mRNA expression just after 3 days of incubation, which is very early stage for desmoplasia of pancreatic tumor, due to the instability for long-term culture (over 1 week) by over growth of cancer cells in 3D-tissues. Such short-term incubation might not be enough to strongly induce gene expression. Of course, the other candidate pathways, hypoxia, FGF-2 and SHH, should closely relate to the desmoplastic reaction of the 3D-MiaPaCa-2 tissues. Accordingly, we are now trying to perform further experiments for long-term culture period.



Figure 3. Effect of pancreatic cancer cells on ECM secretion. The amounts of (a) collagen and (b) non-collagen proteins in 8L-cancer-stromal tissues evaluated by collagen staining kit. (d) and (e) are the estimated amounts of collagen and non-collagen proteins secreted by single cells using the kit. (c) and (f) are the secreted amounts of collagen type I from the 3D-tissues determines using the ELISA kit. Asterisks denote statistically significant difference using a two sample student’s t-test (**p < 0.05, *p < 0.01) for each comparison (n = 3).



[1]        X. Pan, T. Arumugam, T. Yamamoto, P. A. Levin, V. Ramachandran, et al., Clin. Cancer Res. 2008, 14, 8143.

[2]        E. K. Maloney, J. L. McLaughlin, N. E. Dagdigian, L. M. Garrett, K. M. Connors, et al., Cancer Res. 2003, 63, 5073.

[3]        M. R. Kano, Y. Bae, C. Iwata, Y. Morishita, M. Yashiro, et al., Proc. Natl. Acad. Sci. USA 2007, 104, 3460.

[4]        R. Kalluri, M. Zeisberg, Nat. Rev. Cancer 2006, 6, 392.

[5]        M. Pickup, S. Novitskiy, H. L. Moses, Nat. Rev. Cancer 2013, 13, 788.

[6]        A. Masamune, K. Kikuta, T. Watanabe, K. Satoh, M. Hirota, T. Shimosegawa, Am. J. Physiol. Gastrointest. Liver Physiol. 2008, 295, G709.

[7]        E. Schneider, A. Schmid-Kotsas, J. Zhao, H. Weidenbach, R. M. Schmid. A. Menke, G. Adler, J. Waltenberger, A. Grünert, M. G. Bachem, Am. J. Physiol. Cell Physiol. 2001, 281, C532.

[8]        J. M. Bailey, B. J. Swanson, T. Hamada, J. P. Eggers, P. K. Singh, T. Caffery, M. M. Ouellette, M. A. Hollingsworth, Clin. Canver Res. 2008, 14, 5995.