- Email: [email protected]
Interactions Between Bladder Tumor Cells as Tumor Spheroids from the Cell Line J82 and Human Endothelial Cells in Vitro
0022-534 7/88/1393-0640$2.00/0
Vol. 139, March Printed in U.S.A.
THE JOURNAL OF UROLOGY
Copyright© 1988 by The Williams & Wilkins Co.
INTERACTIONS BETWEEN BLADDER TUMOR CELLS AS TUMOR SPHEROIDS FROM THE CELL LINE J82 AND HUMAN ENDOTHELIAL CELLS IN VITRO RUTH KNUCHEL,* J. FEICHTINGER, A. RECKTENWALD, H. G. HOLLWEG, P. FRANKE, G. JAKSE, E. RAMMAL
AND
F. HOFSTADTER
From the Department of Pathology, RWTH Aachen, Federal Republic of Germany, and the Department of Urology, University of Innsbruck, Austria
ABSTRACT
Multicellular tumor spheroids (MCTS) are a reliable model of nonvascularized tumor cell aggregates showing a well defined three-dimensional growth pattern and are comparable with small metastatic cancer cell complexes in blood circulation. In the present study we have established a co-culture system of multicellular bladder tumor spheroids with human endothelial cells on extracellular matrix (ECM) in order to investigate morphological and proliferative changes of endothelial and tumor cells within a defined time of cellcell interaction. The MCTS-endothelial cell-extracellular matrix complex was observed within coculture periods from½ to seven days. Morphological changes (light microscopy, scanning and electron microscopy) indicated that MCTS are not influenced by cocultured endothelial cells. The tumor cells invaded into the ECM after degradation of endothelial cells in the center of the contact zone. Endothelial cells, however, showed degenerative changes as well as a complex reaction in their proliferation activities. We could recognize an initial increase of proliferation of endothelial cells next to the MCTS. Later on, endothelial cells next to invading tumor cells showed changes in morphological polarity. The model system used has the advantage of using human tumor tissue. It distinguishes between basic cellular mechanisms like adherence, migration, DNA synthesis and proliferation in the study of the contact of tumor cells and vascular endothelial cells as an important event in hematogenous tumor spread. (J. Ural., 139: 640-645, 1988) Interactions between tumor cells and endothelial cells are involved in two major steps of growth and spread of neoplasms: 1) angiogenesis and neovascularization of primary and metastatic tumors; 2) tumor cell arrest and adherence to vessel wall preceding extravasation during the metastatic cascade. In both biological processes small avascular tumors are confronted with vessel structures outlined by endothelial cell layers. MCTS, grown systematically from single cells and exhibiting defined structural characteristics, are a useful in vitro model for non-vascularized tumor components. They have been used for the in vitro evaluation of antitumor drug and radiation experiments'· 2 and provide a useful tool for invasion studies. 3 A great number of invasion studies focussed on tumor properties and used organ culture systems like chicken heart, chorioallantoic membrane and bone cartilage.4- 6 Other studies have analyzed endothelial cell behavior using tumor implants in rabbit corneae or Boyden chambers with tumor cells for chemotaxis assays. 7 ' 8 A few experiments have shown the interaction of single tumor cells with bovine corneal endothelial cells, 9 and used tumor cell aggregates on a monolayer of bovine endothelial cells. 3 In the present investigation MCTS were cocultured with human endothelial cell monolayers grown on ECM. This study was designed to determine whether the interactions of human tumor cells and endothelial cells cause damage or proliferation of the respective cells and where the changes take place. Accepted for publication November 10, 1987. * Requests for reprints: Abtlg. Pathologie der RWTH Aachen, Neues Klinikum, G-5100 Aachen, Federal Republic of Germany. Supported by the Fond zur Fiirderung der Wissenschaftl. Forschung, Projektnummer 5727, Austria. 640
MATERIALS AND METHODS
Culture of MCTS. The established and characterized human bladder carcinoma cell line J82 10 was used throughout all experiments. The tumor cells were grown as monolayers in Basal Medium Eagle (BME, GIBCO) supplemented with 5% fetal calf serum and penicillin/streptomycin (100 IU /ml., 100 gm./ ml.; Sigma) at the bottom of tissue culture flasks (Falcon), To culture MCTS, J82 tumor cells were suspended with trypsin/ versene (0.05%) and vital cells were counted in a hematocytometer (Neubauer) after dilution in trypan blue. The cell suspensions were diluted adequately in medium and 4000 cells/ 200 ml. were seeded out in each well of a 96-microwell test plate (Nunc, Falcon) coated with 1 % agarose (Serva). Within four days of incubation at 37C in humid air containing 5% CO 2 the cells formed tight aggregates. Medium was changed every second day, The MCTS primarily showed an exponential increase in growth and reached a plateau phase of growth after two weeks of incubation with a maximum size of 0.3 mm. MCTS with an average diameter of 0.3 mm. were used for experiments. Production of the extracellular matrix (ECM). ECM used was prepared according to the method of Gosporodarowicz et al. (1980). 11 In short, bovine endothelial cells were grown on glass cover slips, methacrylate (Technovit), thermanox slips (LUX) or epon and induced to enhanced production of ECM by the addition of dextran to the culture medium. Eight days later cells were detached with 0.2 M NH and ECM coated slides were further used. Culture of endothelial cells. Endothelial cells were obtained from human umbilical veins. 12 Cells were washed out from the veins after pretreatment with 0.5% collagenase (Boehringer)
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
and grown on the bottom of tissue culture flasks coated with 0.2% gelatine (Boehringer) in Tc 199 medium (Gibco) enriched with 15% human serum. Confluent endothelial cell layers were detached with trypsine/versene (0.005%, Sigma), resuspended in medium and plated subconfluently on ECM coated slides. Until confluency, cells were incubated at 37C in humid atmosphere and 5% CO 2 and washed every second day. MCTS-endothelial-cell-ECM-coculture. MCTS were transferred with glass pipettes from the agarose coated dishes onto confluent endothelial cell monolayers on ECM. After defined coculture periods from 12 hr. to seven days in TC 199 medium with 15% human serum the specimens were washed and fixed in adequate solutions (10% formaline, 4% glutaraldehyde in PBS (Serva), methanol/ethanol (1:1) or Karnowsky). As a control for the endothelial cell reaction sephadex beads (250 to 300 µ,) were cocultured with endothelial cells on ECM. Furthermore, MCTS were grown in parallel on ECM or plastic alone. Tissue processing and analytical methodology. For histological staining and histochemistry the MCTS-endothelial cell complexes on ECM coated methacrylate slides were dehydrated and embedded in methacrylate. The embedded specimens were cut vertically in serial sections and mounted on poly-L-lysine (Sigma) coated glass slides. Sufficient staining results of the sections were obtained with hematoxylin-eosin and acid phosphatase and the naphtyl-esterase reaction. To analyse the cellular proliferation we used the bromodeoxyuridine (BrdU)anti BrdU technique according to Gratzner (1979). 13 After defined intervals the cocultures grown on glass slides were washed and incubated with serum free medium containing 10 mol. BrdU (Boehringer) for 30 min. After fixation with methanol/ethanol the specimens were stored in PBS at 4C. The staining procedure was an indirect peroxidase technique. After destruction of the histones with warm 1 N HCL (20 min.) the specimens were incubated with the first antibody (mouse anti BrdU, Becton Dickinson) in a concentration of 1:75 for 30 min. After rinsing in buffer (PBS) a second antibody (peroxidase conjugated rabbit lgG to mouse lgG, Dakppatts) in a concentration of 1:100 was applied. Subsequent incubation with 33 diamino-benzidine tetrahydrochloride (DAB, grade III, Sigma) resulted in brown staining of the nuclei of proliferating cells. The stained samples were embedded as a whole using hollow grinded glass slides. For transmission electron microscopy the epon grown samples were fixed with buffered glutaraldehyde for 30 minutes. After postfixation in osmium, buffered in 0.1 M cacodylate, they were dehydrated and embedded in epon. Semithin and ultrathin sections were cut with an LKB ultramicrotome, mounted on copper grids, stained with uranyl acetate and lead citrate and examined with a transmission electron microscope (Phillips). For scanning electron microscopy specimens on glass slides were formalin-fixed (10% formalin in PBS), acetone dehydrated and investigated with a scanning electron microscope (Phillips).
641
HEC and an average of 18 to 22 tumor cell layers and tumor cell clumps on ECM alone. Marked proliferation of the migrated tumor cells could be seen after six days on HEC and after three days on ECM. On HEC as well as on ECM, tumor cells primarily migrated radially and showed an elongated cell shape until they reached confluence and showed a more polygonal cell shape. The histological and histochemical analyses revealed intact MCTS. Vertical sections showed spheroidal structures with an outer area of vital cells and a central necrosis (Fig. 1). Cells on the MCTS surface flattened, resembling umbrella cells. These findings were constant in the J82·MCTS in coculture after two weeks of growth in the liquid-overlay culture. The BrdU-anti-BrdU technique showed proliferating cells in the vital zone of the MCTS next to the surface. It also revealed persistant proliferation of tumor cells (32 hours after the start of the coculture) on the endothelial cell surface after primary migration. Using transmission electron microscopy, more detailed information on the single cell level is provided. The surface of the MCTS is characterized by numerous microvilli. Microvilli and cytoplasmic protrusions were numerous in the periphery of the contact area to the endothelial cells. Within the MCTS tumor cells showed a few incomplete desmosomes as well as numerous interdigitations with intercellular channels in between. We could not see any change of tumor cell morphology caused by endothelial cells. Cells in contact with endothelial cells respectively ECM showed such signs of metabolic activity as high amounts of rough endoplasmic reticulum and mitochondriae. Marked differences in cell morphology could not be found. After a coculture period of seven days tumor cells were found in direct contact with the bovine ECM and gradually penetrated the ECM (figs. 2-4). This was a commonplace observation in all specimens of that age. Beginning cytoplasmic protrusion is partly followed by complete penetration of the ECM with further expansion of tumor cell cytoplasm below the ECM. In later stages of coculture whole tumor cells can be found under the ECM and between the ECM and the epon substrate. Identical morphological pictures of tumor cells were found, when tumor cells were primarily cultured on ECM. Figs. 5. and 6 show beginning penetration of tumor cells into the ECM in a three day old coculture. Further progression occurred as described above. Growth characteristics and morphological changes of endothelial cells. The first period of coculture we were able to analyze (24 hr.) showed initial increased proliferation of endothelial cells surrounding the MCTS. This was assumed by phase microscopical observation and could be proven by BrdU anti BrdU staining of the cocultures, showing an increased amount
RESULTS
Growth characteristics of MCTS. After 24 hr. of coculture the MCTS were attached so tightly to the substrate that the complexes could be fixed and embedded. The handling of the specimens after 12 hr. showed slight attachment. The controls cocultured without endothelial cells showed a tight attachment of bladder MCTS to ECM after 12 hours. Amorphous beads did not attach to the ECM at all. Migration of tumor cells on the endothelial cell layer started after 32 hr. of coculture in comparison to an earlier onset of migration of tumor cells on ECM alone (12 hr.). After two days of coculture we could find an average of three to four tumor cell layers around MCTS on HEC and, already, nine to 11 layers on ECM. Analogously, after seven days of coculture, seven to eight tumor cell layers were found around MCTS on
FIG. 1. Vertical section of MCTS-HEC-ECM coculture. Azur II, lOOX.
642
KNUCHEL AND ASSOCIATES
FIG. 5. MCTS primarily grown on ECM. Cytoplasmic protrusions of tumor cells invade ECM. Three days of coculture. TEM, 30000X. FIG. 2. Tumor cell penetration into ECM after degeneration of endothelial cells in central zone of the MCTS-HEC-ECM coculture. MCTS cells directly on ECM, late phase of coculture. Transmission electron microscopy (TEM), 7900X.
FIG. 6. Section of fig. 5. Cytoplasmic protrusion with surrounding degraded ECM. TEM, 138000X.
FIG. 3. Cytoplasmic protrusion into ECM with degradation of smrounding ECM. Section of fig. 2. TEM, 50000X.
FIG. 4. Complete penetration of tumor cell protrusion through ECM and further expansion below ECM. TEM, 26000X.
of positively stained brown nuclei of endothelial cells next to the MCTS (figs. 7 and 8). Additionally endothelial cells, observed from the surface, looked more rounded in this zone at the early phase of coculture and also orientated in a circular fashion. This could be made visible best by scanning electron microscopy (fig. 9). Sephadex beads (controls) did not show any alterations in cell morphology and proliferation in this early state of culture (fig. 10). 12 hr. later, (32 hr.) migration and proliferation of tumor cells dominated. Endothelial cells next to the tumor cells still were orientated in a circular fashion. In this area no increased proliferation of endothelial cells could be found. Vertical sections of the coculture specimens showed less flattened endothelial cells adjacent to the migrating tumor cells, indicating endothelial cell retraction in the early phases of coculture. fo contrast, the endothelial cells directly under the MCTS as well as in the periphery of the endothelial cell layer were flattened. The phenomenon of retraction is even more evident in TEM, where we found rounded endothelial cells next to the protruding tumor cells. Some of the endothelial cells partly or completely detached from the primary ECM where tumor cells were penetrating. We could even find endothelial cells below the ECM, attached to the ECM from below with a morphological change in morphology (fig. 11). Those cells as well as many other endothelial cells showed a
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
FIG. 7. Increased proliferation of endothelial cells surrounding MCTS compared to periphery with only scattered endothelial cellsdark nuclei = nuclei of proliferated cells. Interference phase contrast, BrdU-anti-BrdU staining, lOOX.
FIG. 8. Oval regular nuclei of proliferated nuclei of endothelial cells in higher magnification. Interference phase contrast, BrdU-anti BrdU staining, 250X.
643
FIG. 9. Initial phase of MCTS-HEC-ECM coculture. Retraction and circular orientation of endothelial cells. Beginning migration of tumor cells. Scanning electron microscopy, 1 bar = 0.1 mm.
FIG. 10. Amorphous plastic bead (control) on HEC on ECM. 24 hr after beginning of coculture. Phase contrast, SOX.
lighter cytoplasmic stain compared to the tumor cells and also a lot of microvesicles. In contrast to tumor cells, cytoplasmic protrusions into the ECM were never found in endothelial cells. Endothelial cells under the MCTS showed marked degeneration after four days of coculture and no viable endothelial cell could be found there after seven days of coculture. In the periphery of the cell culture specimen endothelial cells were found in regular shape and number. After seven days endothelial cells under the control beads were also degraded (fig. 12). DISCUSSION
Metastatic spread of bladder cancer proceeds by the way of lymphatic channels and blood vessels. The hematogenous spread, however, is a comparatively rare event and manifests itself either as solitary organ metastasis or as a generalized carcinoma related to vascular embolization. 14 In an attempt to study the basic mechanisms of these observations we established MCTS from human bladder carcinoma cells. The tumor cells derive from an established tumor cell line 15 and have been recently characterized as a grade III urothelial carcinoma cell line. 10 The influence of host cells and tumorous stroma is excluded in this study to confine the observation to the interaction between endothelial cells and tumor cells alone. Additionally, we used human cells to avoid alien immunological influences. In order to imitate the vascular anatomy, we grew the endothelial cells confluently on ECM. The normal asym-
FIG. 11. Endothelial cell under ECM. Morphological change in polarity compared to endothelial cell above. Part of tumor cell on left side below ECM. TEM, lOOOOx.
metry in the display of components on the apical and the basolateral surface as described by Birdwell (1978) is found when endothelial cells grow on ECM and plays an important role in the adhesion of cells. 16
644
KNOCHEL AND ASSOCIATES
FIG. 12. Degradation of endothelial cells under amorphous plastic beads after seven days of coculture. Phase contrast, 80X.
Using this coculture system we tried to understand more of the complex interactions of endothelial cells and bladder tumor cells. Partly due to technical hindrances and partly due to changes not visible by morphologic methods, the first phases of coculture (before 24 hr.) could not be evaluated. It is interesting, however, to recognize an increased proliferation of endothelial cells after 24 hr. of coculture, soon dominated by tumor cell migration and proliferation (32 hr.). Endothelial cell proliferation has been also observed in wound healing.17 Mechanical damage however, is excluded in our system by using amorphous material as a control, that does not induce endothelial cell changes. Endothelial cell proliferation has been described as an important event in the multistep process of angiogenesis. 8 Tumor angiogenesis has been studied in detail by Auksprunk. An initial orientation of endothelial cells is followed by migration towards the tumor and later on the proliferation of endothelial cells starts next to the top of the vascular sprout. 18 We suggest the initial proliferation to be a defensive mechanism of the endothelial cells against the tumor cells. We also think of the slower attachment of tumor cells to and their delayed migration on the endothelial cell layer in comparison to tumor cells on ECM as protective features of endothelial cells. The protective properties of the endothelial cells are soon overcome by degradative changes. The degeneration of endothelial cells is most extensively expressed in the center of the contact region. An intensive contact of tumor cells to endothelial cells seems to be necessary for the induction of endothelial cell degeneration. The degeneration of cells under the control indicates that hypoxia is one unspecific reason for tumor cell necrosis. The earlier onset of necrosis under tumor cell MCTS hints on additional tumor cell specific toxic activity like proteolysis. Hypoxic changes, promoting hematogenous tumor spread are relevant in vivo, when tumor particles obstruct capillaries or when tumor clumps-partly supported by blood substances as coagulation products and platelets-get attached to venous vessel walls. Endothelial cell detachment from the ECM and morphological changes of endothelial cell polarity, as seen in our experiments, are important features of endothelial cells to be able to form vessels. The changes observed may be induced by tumor cells. A possible chemoattraction of endothelial cells to tumor cells has been described by Seppa et al. using conditioned medium of HP4 tumor cells. 7 A few authors report on endothelial cell reaction to tumor cells, using tumor cell suspensions on confluent endothelial cell layers. 18 ·19·9 The common features in the literature are confirmed by our investigations. The primary tumor cell attachment, mostly situated next to endothelial intercellular con-
tacts, 20 is followed by endothelial cell retraction. Infiltration of tumor cells starts with protrusions of tumor cells between endothelial cells and subsequent migration between endothelial cells and ECM. The network of ECM is degraded around these protrusions. Other tumor cells of the same MCTS are attached to the endothelial cell surface without any sign of interaction. This might be due to the fact that cloned cell lines represent a heterogenous cellular population with respect to metastatic capacity. 22 An additional explanation could be that invasion is impossible for single cells, because it depends on intercellular communication as proposed by Schirrmacher (1983) and observed by Zamorra (1980). 9· 3 Using confluent endothelial cells and tumor cells as small tissue components, important features of the in vivo situation are simulated. We conclude from our results, that similar ways of invasion described using other tumor cells can be found using bladder tumor cells. With the application of this coculture system, we are able to study a lot of events that have been described singularly-for example proliferation of endothelial cells and retraction or degeneration of endothelial cells- sequentially and topographically. Therefore this system provides a good tool for further comparative studies. We are encouraged to compare the reaction pattern with that of endothelial cells of different organ sites. Antigenic differences 23 and glycoprotein structures 19 of cellular membranes may influence the organotropy of metastasis. Additionally bladder tumor cell lines also possibly show different reaction patterns. REFERENCES 1. Durand, R. E.: Variable radiobiological responses of spheroids.
Radiat. Res., 81: 85, 1980. 2. Twentyman, P. R.: Response to chemotherapy of EMT6 spheroids as measured by growth delay and cell survival. Br. J. Cancer, 42: 297, 1980. 3. Zamora, P. 0., Danielson, K. G. and Hosick, Bh. C.: Invasion of endothelial cell monolayers on collagen gels by cells from mammary tumor spheroids. Cancer Res., 40: 4631, 1980. 4. Hart, I. R. and Fidler, I. J.: An in vitro quantitative assay for tumor cell invasion. Cancer Res., 38: 3218, 1978. 5. Heisel, M., Laug, W. E. and Jones, P. A.: Inhibition by bovine endothelial cells of degradation by HT-1080 fibrosarcoma cells of extracellular matrix proteins. JNCI, 71: 1183, 1983. 6. Russo, R. G., Thorgeirsson, U. and Liotta, L. A.: In vitro quantitative assay of invasion using human amnion. In: Tumor Invasion and Metastasis. Edited by Liotta, L. A. and Hart, I. R. The Hague: Martinus Nijhoff Publishers, pp. 173-187, 1982. 7. Seppa, S. T., Seppa, H. E. J., Liotta, L.A., Glaser, B. M., Martin, G. R. and Schiffmann, E.: Cultured tumor cells produce chemotactic factors specific for endothelial cells: a possible mechanism for tumor-induced angiogenesis. Invas. Metastasis, 3: 139, 1983. 8. Folkman, J.: How is blood vessel growth regulated in normal and neoplastic tissue?-G.H.A. Clowes Memorial Award Lecture. Cancer Res., 46: 467, 1986. 9. Schirrmacher, V. and Vlodawsky, I.: In vitro Interaktionen vbon Gefassendothelzellmonolayern mit Tumorzellinien von unterschiedlicher Metastasierungskapazitat. In: Struktur und Funktion endothelialer Zellen. Edited by Messmer, K. and Hammersen, F. Miinchen:Karger, pp. 34-44, 1983. 10. Masters, J. R. S. and Hepburn, P. J.: Tissue culture model of transitional cell carcinoma: characterization of 22 human urothelial cell lines. Cancer Res., 46: 3630, 1986. 11. Gosporodarowicz, D. Vlodawsky, J. and Savion, N.: The extracellular matrix and the control of proliferation of vascular endothelial and vascular smooth muscle cells. J. Supramol. Struct., 13: 339, 1980. 12. Jaffe, E. H., Nachman, R. L., Becker, D. G. and Minick, C. R.: Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and immunologic criteria. J. Clin. Invest., 52: 2745, 1973. 13. Gratzner, H. G.: Monoclonal antibody to 5- bromo- and 5- iododeoxy-uridine: a new reagent for detection of DNA replication. Science, 218: 474, 1982. 14. Tabarra, W. S., and Mehio, A. R.: Metastatic patterns of bladder carcinoma. In: Progress in Clinical and Biological Research.
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
15. 16. 17. 18. 19.
Bladder Cancer. Part A. Edited by Kiiss, R., Khoury, S., Denis, L. J., Murphy, G. P., Karr, J.P. New York: Alan R. Liss, Inc. vol. 162A, pp. 145-160, 1984. O'Toole, C., Price, Z. H., Ohnuki, J. and Unsgaard, B.: Ultrastructure, karyology and immunology of a cell line originated from human transitional cell carcinoma. Br. J. Cancer, 38: 64, 1978. Birdwell, C. R., Gospodarowicz, D. and Nicolson, G. L.: Identification, localization, and role of fibronectin in cultured bovine endothelial cells. Proc. Natl. Acad. Sci. USA, 7: 3273, 1978. Schwartz, S. M. and Haudenschild, C. C.: Endothelial regeneration. Quantitative analysis of initial stages of endothelial regeneration in rat aortic intima. Lab. Invest., 38: 568, 1978. Auksprunk, D. H. and Folkman, J.: Migration and proliferation of endothelial cells in preformed and newly formed blood vessels during tumour angiogenesis. Microvasc. Res., 14: 53, 1977. Kramer, R.H. and Nicolson, G. L.: Interactions of tumor cells with
20. 21.
22. 23.
645
vascular endothelial cell monolayers: a model for metastatic invasion. Proc. Natl. Acad. Sci., 76: 5704, 1979. Nicolson, G. A.: Metastatic tumor cell attachment and invasion assay utilising vascular endothelial cell monolayers. J. Histochem. Cytochem., 30: 214, 1982. Nicolson, G. A., Irimura, T. Nakajima, M. and Estrada, J.: Metastatic cell attachment to an invasion of vasular endothelium and its underlying basal lamina using endothelial cell monolayers. In: Cancer Invasion and Metastasis: Biologic and Therapeutic Aspects. Edited by G. L. Nicolson and L. Milas. New York: Raven Press, pp. 145-167, 1984. Fidler, I. J. and Hart, I. R.: Biological diversity in metastatic neoplasms: origins and implications. Science, 217: 998, 1982. Jones, P. A.: Construction of an artificial blood vessel wall from cultured endothelial and smooth muscle cells. Proc. Natl. Acad. Sci. USA, 76: 1882, 1979.
Vol. 139, March Printed in U.S.A.
THE JOURNAL OF UROLOGY
Copyright© 1988 by The Williams & Wilkins Co.
INTERACTIONS BETWEEN BLADDER TUMOR CELLS AS TUMOR SPHEROIDS FROM THE CELL LINE J82 AND HUMAN ENDOTHELIAL CELLS IN VITRO RUTH KNUCHEL,* J. FEICHTINGER, A. RECKTENWALD, H. G. HOLLWEG, P. FRANKE, G. JAKSE, E. RAMMAL
AND
F. HOFSTADTER
From the Department of Pathology, RWTH Aachen, Federal Republic of Germany, and the Department of Urology, University of Innsbruck, Austria
ABSTRACT
Multicellular tumor spheroids (MCTS) are a reliable model of nonvascularized tumor cell aggregates showing a well defined three-dimensional growth pattern and are comparable with small metastatic cancer cell complexes in blood circulation. In the present study we have established a co-culture system of multicellular bladder tumor spheroids with human endothelial cells on extracellular matrix (ECM) in order to investigate morphological and proliferative changes of endothelial and tumor cells within a defined time of cellcell interaction. The MCTS-endothelial cell-extracellular matrix complex was observed within coculture periods from½ to seven days. Morphological changes (light microscopy, scanning and electron microscopy) indicated that MCTS are not influenced by cocultured endothelial cells. The tumor cells invaded into the ECM after degradation of endothelial cells in the center of the contact zone. Endothelial cells, however, showed degenerative changes as well as a complex reaction in their proliferation activities. We could recognize an initial increase of proliferation of endothelial cells next to the MCTS. Later on, endothelial cells next to invading tumor cells showed changes in morphological polarity. The model system used has the advantage of using human tumor tissue. It distinguishes between basic cellular mechanisms like adherence, migration, DNA synthesis and proliferation in the study of the contact of tumor cells and vascular endothelial cells as an important event in hematogenous tumor spread. (J. Ural., 139: 640-645, 1988) Interactions between tumor cells and endothelial cells are involved in two major steps of growth and spread of neoplasms: 1) angiogenesis and neovascularization of primary and metastatic tumors; 2) tumor cell arrest and adherence to vessel wall preceding extravasation during the metastatic cascade. In both biological processes small avascular tumors are confronted with vessel structures outlined by endothelial cell layers. MCTS, grown systematically from single cells and exhibiting defined structural characteristics, are a useful in vitro model for non-vascularized tumor components. They have been used for the in vitro evaluation of antitumor drug and radiation experiments'· 2 and provide a useful tool for invasion studies. 3 A great number of invasion studies focussed on tumor properties and used organ culture systems like chicken heart, chorioallantoic membrane and bone cartilage.4- 6 Other studies have analyzed endothelial cell behavior using tumor implants in rabbit corneae or Boyden chambers with tumor cells for chemotaxis assays. 7 ' 8 A few experiments have shown the interaction of single tumor cells with bovine corneal endothelial cells, 9 and used tumor cell aggregates on a monolayer of bovine endothelial cells. 3 In the present investigation MCTS were cocultured with human endothelial cell monolayers grown on ECM. This study was designed to determine whether the interactions of human tumor cells and endothelial cells cause damage or proliferation of the respective cells and where the changes take place. Accepted for publication November 10, 1987. * Requests for reprints: Abtlg. Pathologie der RWTH Aachen, Neues Klinikum, G-5100 Aachen, Federal Republic of Germany. Supported by the Fond zur Fiirderung der Wissenschaftl. Forschung, Projektnummer 5727, Austria. 640
MATERIALS AND METHODS
Culture of MCTS. The established and characterized human bladder carcinoma cell line J82 10 was used throughout all experiments. The tumor cells were grown as monolayers in Basal Medium Eagle (BME, GIBCO) supplemented with 5% fetal calf serum and penicillin/streptomycin (100 IU /ml., 100 gm./ ml.; Sigma) at the bottom of tissue culture flasks (Falcon), To culture MCTS, J82 tumor cells were suspended with trypsin/ versene (0.05%) and vital cells were counted in a hematocytometer (Neubauer) after dilution in trypan blue. The cell suspensions were diluted adequately in medium and 4000 cells/ 200 ml. were seeded out in each well of a 96-microwell test plate (Nunc, Falcon) coated with 1 % agarose (Serva). Within four days of incubation at 37C in humid air containing 5% CO 2 the cells formed tight aggregates. Medium was changed every second day, The MCTS primarily showed an exponential increase in growth and reached a plateau phase of growth after two weeks of incubation with a maximum size of 0.3 mm. MCTS with an average diameter of 0.3 mm. were used for experiments. Production of the extracellular matrix (ECM). ECM used was prepared according to the method of Gosporodarowicz et al. (1980). 11 In short, bovine endothelial cells were grown on glass cover slips, methacrylate (Technovit), thermanox slips (LUX) or epon and induced to enhanced production of ECM by the addition of dextran to the culture medium. Eight days later cells were detached with 0.2 M NH and ECM coated slides were further used. Culture of endothelial cells. Endothelial cells were obtained from human umbilical veins. 12 Cells were washed out from the veins after pretreatment with 0.5% collagenase (Boehringer)
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
and grown on the bottom of tissue culture flasks coated with 0.2% gelatine (Boehringer) in Tc 199 medium (Gibco) enriched with 15% human serum. Confluent endothelial cell layers were detached with trypsine/versene (0.005%, Sigma), resuspended in medium and plated subconfluently on ECM coated slides. Until confluency, cells were incubated at 37C in humid atmosphere and 5% CO 2 and washed every second day. MCTS-endothelial-cell-ECM-coculture. MCTS were transferred with glass pipettes from the agarose coated dishes onto confluent endothelial cell monolayers on ECM. After defined coculture periods from 12 hr. to seven days in TC 199 medium with 15% human serum the specimens were washed and fixed in adequate solutions (10% formaline, 4% glutaraldehyde in PBS (Serva), methanol/ethanol (1:1) or Karnowsky). As a control for the endothelial cell reaction sephadex beads (250 to 300 µ,) were cocultured with endothelial cells on ECM. Furthermore, MCTS were grown in parallel on ECM or plastic alone. Tissue processing and analytical methodology. For histological staining and histochemistry the MCTS-endothelial cell complexes on ECM coated methacrylate slides were dehydrated and embedded in methacrylate. The embedded specimens were cut vertically in serial sections and mounted on poly-L-lysine (Sigma) coated glass slides. Sufficient staining results of the sections were obtained with hematoxylin-eosin and acid phosphatase and the naphtyl-esterase reaction. To analyse the cellular proliferation we used the bromodeoxyuridine (BrdU)anti BrdU technique according to Gratzner (1979). 13 After defined intervals the cocultures grown on glass slides were washed and incubated with serum free medium containing 10 mol. BrdU (Boehringer) for 30 min. After fixation with methanol/ethanol the specimens were stored in PBS at 4C. The staining procedure was an indirect peroxidase technique. After destruction of the histones with warm 1 N HCL (20 min.) the specimens were incubated with the first antibody (mouse anti BrdU, Becton Dickinson) in a concentration of 1:75 for 30 min. After rinsing in buffer (PBS) a second antibody (peroxidase conjugated rabbit lgG to mouse lgG, Dakppatts) in a concentration of 1:100 was applied. Subsequent incubation with 33 diamino-benzidine tetrahydrochloride (DAB, grade III, Sigma) resulted in brown staining of the nuclei of proliferating cells. The stained samples were embedded as a whole using hollow grinded glass slides. For transmission electron microscopy the epon grown samples were fixed with buffered glutaraldehyde for 30 minutes. After postfixation in osmium, buffered in 0.1 M cacodylate, they were dehydrated and embedded in epon. Semithin and ultrathin sections were cut with an LKB ultramicrotome, mounted on copper grids, stained with uranyl acetate and lead citrate and examined with a transmission electron microscope (Phillips). For scanning electron microscopy specimens on glass slides were formalin-fixed (10% formalin in PBS), acetone dehydrated and investigated with a scanning electron microscope (Phillips).
641
HEC and an average of 18 to 22 tumor cell layers and tumor cell clumps on ECM alone. Marked proliferation of the migrated tumor cells could be seen after six days on HEC and after three days on ECM. On HEC as well as on ECM, tumor cells primarily migrated radially and showed an elongated cell shape until they reached confluence and showed a more polygonal cell shape. The histological and histochemical analyses revealed intact MCTS. Vertical sections showed spheroidal structures with an outer area of vital cells and a central necrosis (Fig. 1). Cells on the MCTS surface flattened, resembling umbrella cells. These findings were constant in the J82·MCTS in coculture after two weeks of growth in the liquid-overlay culture. The BrdU-anti-BrdU technique showed proliferating cells in the vital zone of the MCTS next to the surface. It also revealed persistant proliferation of tumor cells (32 hours after the start of the coculture) on the endothelial cell surface after primary migration. Using transmission electron microscopy, more detailed information on the single cell level is provided. The surface of the MCTS is characterized by numerous microvilli. Microvilli and cytoplasmic protrusions were numerous in the periphery of the contact area to the endothelial cells. Within the MCTS tumor cells showed a few incomplete desmosomes as well as numerous interdigitations with intercellular channels in between. We could not see any change of tumor cell morphology caused by endothelial cells. Cells in contact with endothelial cells respectively ECM showed such signs of metabolic activity as high amounts of rough endoplasmic reticulum and mitochondriae. Marked differences in cell morphology could not be found. After a coculture period of seven days tumor cells were found in direct contact with the bovine ECM and gradually penetrated the ECM (figs. 2-4). This was a commonplace observation in all specimens of that age. Beginning cytoplasmic protrusion is partly followed by complete penetration of the ECM with further expansion of tumor cell cytoplasm below the ECM. In later stages of coculture whole tumor cells can be found under the ECM and between the ECM and the epon substrate. Identical morphological pictures of tumor cells were found, when tumor cells were primarily cultured on ECM. Figs. 5. and 6 show beginning penetration of tumor cells into the ECM in a three day old coculture. Further progression occurred as described above. Growth characteristics and morphological changes of endothelial cells. The first period of coculture we were able to analyze (24 hr.) showed initial increased proliferation of endothelial cells surrounding the MCTS. This was assumed by phase microscopical observation and could be proven by BrdU anti BrdU staining of the cocultures, showing an increased amount
RESULTS
Growth characteristics of MCTS. After 24 hr. of coculture the MCTS were attached so tightly to the substrate that the complexes could be fixed and embedded. The handling of the specimens after 12 hr. showed slight attachment. The controls cocultured without endothelial cells showed a tight attachment of bladder MCTS to ECM after 12 hours. Amorphous beads did not attach to the ECM at all. Migration of tumor cells on the endothelial cell layer started after 32 hr. of coculture in comparison to an earlier onset of migration of tumor cells on ECM alone (12 hr.). After two days of coculture we could find an average of three to four tumor cell layers around MCTS on HEC and, already, nine to 11 layers on ECM. Analogously, after seven days of coculture, seven to eight tumor cell layers were found around MCTS on
FIG. 1. Vertical section of MCTS-HEC-ECM coculture. Azur II, lOOX.
642
KNUCHEL AND ASSOCIATES
FIG. 5. MCTS primarily grown on ECM. Cytoplasmic protrusions of tumor cells invade ECM. Three days of coculture. TEM, 30000X. FIG. 2. Tumor cell penetration into ECM after degeneration of endothelial cells in central zone of the MCTS-HEC-ECM coculture. MCTS cells directly on ECM, late phase of coculture. Transmission electron microscopy (TEM), 7900X.
FIG. 6. Section of fig. 5. Cytoplasmic protrusion with surrounding degraded ECM. TEM, 138000X.
FIG. 3. Cytoplasmic protrusion into ECM with degradation of smrounding ECM. Section of fig. 2. TEM, 50000X.
FIG. 4. Complete penetration of tumor cell protrusion through ECM and further expansion below ECM. TEM, 26000X.
of positively stained brown nuclei of endothelial cells next to the MCTS (figs. 7 and 8). Additionally endothelial cells, observed from the surface, looked more rounded in this zone at the early phase of coculture and also orientated in a circular fashion. This could be made visible best by scanning electron microscopy (fig. 9). Sephadex beads (controls) did not show any alterations in cell morphology and proliferation in this early state of culture (fig. 10). 12 hr. later, (32 hr.) migration and proliferation of tumor cells dominated. Endothelial cells next to the tumor cells still were orientated in a circular fashion. In this area no increased proliferation of endothelial cells could be found. Vertical sections of the coculture specimens showed less flattened endothelial cells adjacent to the migrating tumor cells, indicating endothelial cell retraction in the early phases of coculture. fo contrast, the endothelial cells directly under the MCTS as well as in the periphery of the endothelial cell layer were flattened. The phenomenon of retraction is even more evident in TEM, where we found rounded endothelial cells next to the protruding tumor cells. Some of the endothelial cells partly or completely detached from the primary ECM where tumor cells were penetrating. We could even find endothelial cells below the ECM, attached to the ECM from below with a morphological change in morphology (fig. 11). Those cells as well as many other endothelial cells showed a
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
FIG. 7. Increased proliferation of endothelial cells surrounding MCTS compared to periphery with only scattered endothelial cellsdark nuclei = nuclei of proliferated cells. Interference phase contrast, BrdU-anti-BrdU staining, lOOX.
FIG. 8. Oval regular nuclei of proliferated nuclei of endothelial cells in higher magnification. Interference phase contrast, BrdU-anti BrdU staining, 250X.
643
FIG. 9. Initial phase of MCTS-HEC-ECM coculture. Retraction and circular orientation of endothelial cells. Beginning migration of tumor cells. Scanning electron microscopy, 1 bar = 0.1 mm.
FIG. 10. Amorphous plastic bead (control) on HEC on ECM. 24 hr after beginning of coculture. Phase contrast, SOX.
lighter cytoplasmic stain compared to the tumor cells and also a lot of microvesicles. In contrast to tumor cells, cytoplasmic protrusions into the ECM were never found in endothelial cells. Endothelial cells under the MCTS showed marked degeneration after four days of coculture and no viable endothelial cell could be found there after seven days of coculture. In the periphery of the cell culture specimen endothelial cells were found in regular shape and number. After seven days endothelial cells under the control beads were also degraded (fig. 12). DISCUSSION
Metastatic spread of bladder cancer proceeds by the way of lymphatic channels and blood vessels. The hematogenous spread, however, is a comparatively rare event and manifests itself either as solitary organ metastasis or as a generalized carcinoma related to vascular embolization. 14 In an attempt to study the basic mechanisms of these observations we established MCTS from human bladder carcinoma cells. The tumor cells derive from an established tumor cell line 15 and have been recently characterized as a grade III urothelial carcinoma cell line. 10 The influence of host cells and tumorous stroma is excluded in this study to confine the observation to the interaction between endothelial cells and tumor cells alone. Additionally, we used human cells to avoid alien immunological influences. In order to imitate the vascular anatomy, we grew the endothelial cells confluently on ECM. The normal asym-
FIG. 11. Endothelial cell under ECM. Morphological change in polarity compared to endothelial cell above. Part of tumor cell on left side below ECM. TEM, lOOOOx.
metry in the display of components on the apical and the basolateral surface as described by Birdwell (1978) is found when endothelial cells grow on ECM and plays an important role in the adhesion of cells. 16
644
KNOCHEL AND ASSOCIATES
FIG. 12. Degradation of endothelial cells under amorphous plastic beads after seven days of coculture. Phase contrast, 80X.
Using this coculture system we tried to understand more of the complex interactions of endothelial cells and bladder tumor cells. Partly due to technical hindrances and partly due to changes not visible by morphologic methods, the first phases of coculture (before 24 hr.) could not be evaluated. It is interesting, however, to recognize an increased proliferation of endothelial cells after 24 hr. of coculture, soon dominated by tumor cell migration and proliferation (32 hr.). Endothelial cell proliferation has been also observed in wound healing.17 Mechanical damage however, is excluded in our system by using amorphous material as a control, that does not induce endothelial cell changes. Endothelial cell proliferation has been described as an important event in the multistep process of angiogenesis. 8 Tumor angiogenesis has been studied in detail by Auksprunk. An initial orientation of endothelial cells is followed by migration towards the tumor and later on the proliferation of endothelial cells starts next to the top of the vascular sprout. 18 We suggest the initial proliferation to be a defensive mechanism of the endothelial cells against the tumor cells. We also think of the slower attachment of tumor cells to and their delayed migration on the endothelial cell layer in comparison to tumor cells on ECM as protective features of endothelial cells. The protective properties of the endothelial cells are soon overcome by degradative changes. The degeneration of endothelial cells is most extensively expressed in the center of the contact region. An intensive contact of tumor cells to endothelial cells seems to be necessary for the induction of endothelial cell degeneration. The degeneration of cells under the control indicates that hypoxia is one unspecific reason for tumor cell necrosis. The earlier onset of necrosis under tumor cell MCTS hints on additional tumor cell specific toxic activity like proteolysis. Hypoxic changes, promoting hematogenous tumor spread are relevant in vivo, when tumor particles obstruct capillaries or when tumor clumps-partly supported by blood substances as coagulation products and platelets-get attached to venous vessel walls. Endothelial cell detachment from the ECM and morphological changes of endothelial cell polarity, as seen in our experiments, are important features of endothelial cells to be able to form vessels. The changes observed may be induced by tumor cells. A possible chemoattraction of endothelial cells to tumor cells has been described by Seppa et al. using conditioned medium of HP4 tumor cells. 7 A few authors report on endothelial cell reaction to tumor cells, using tumor cell suspensions on confluent endothelial cell layers. 18 ·19·9 The common features in the literature are confirmed by our investigations. The primary tumor cell attachment, mostly situated next to endothelial intercellular con-
tacts, 20 is followed by endothelial cell retraction. Infiltration of tumor cells starts with protrusions of tumor cells between endothelial cells and subsequent migration between endothelial cells and ECM. The network of ECM is degraded around these protrusions. Other tumor cells of the same MCTS are attached to the endothelial cell surface without any sign of interaction. This might be due to the fact that cloned cell lines represent a heterogenous cellular population with respect to metastatic capacity. 22 An additional explanation could be that invasion is impossible for single cells, because it depends on intercellular communication as proposed by Schirrmacher (1983) and observed by Zamorra (1980). 9· 3 Using confluent endothelial cells and tumor cells as small tissue components, important features of the in vivo situation are simulated. We conclude from our results, that similar ways of invasion described using other tumor cells can be found using bladder tumor cells. With the application of this coculture system, we are able to study a lot of events that have been described singularly-for example proliferation of endothelial cells and retraction or degeneration of endothelial cells- sequentially and topographically. Therefore this system provides a good tool for further comparative studies. We are encouraged to compare the reaction pattern with that of endothelial cells of different organ sites. Antigenic differences 23 and glycoprotein structures 19 of cellular membranes may influence the organotropy of metastasis. Additionally bladder tumor cell lines also possibly show different reaction patterns. REFERENCES 1. Durand, R. E.: Variable radiobiological responses of spheroids.
Radiat. Res., 81: 85, 1980. 2. Twentyman, P. R.: Response to chemotherapy of EMT6 spheroids as measured by growth delay and cell survival. Br. J. Cancer, 42: 297, 1980. 3. Zamora, P. 0., Danielson, K. G. and Hosick, Bh. C.: Invasion of endothelial cell monolayers on collagen gels by cells from mammary tumor spheroids. Cancer Res., 40: 4631, 1980. 4. Hart, I. R. and Fidler, I. J.: An in vitro quantitative assay for tumor cell invasion. Cancer Res., 38: 3218, 1978. 5. Heisel, M., Laug, W. E. and Jones, P. A.: Inhibition by bovine endothelial cells of degradation by HT-1080 fibrosarcoma cells of extracellular matrix proteins. JNCI, 71: 1183, 1983. 6. Russo, R. G., Thorgeirsson, U. and Liotta, L. A.: In vitro quantitative assay of invasion using human amnion. In: Tumor Invasion and Metastasis. Edited by Liotta, L. A. and Hart, I. R. The Hague: Martinus Nijhoff Publishers, pp. 173-187, 1982. 7. Seppa, S. T., Seppa, H. E. J., Liotta, L.A., Glaser, B. M., Martin, G. R. and Schiffmann, E.: Cultured tumor cells produce chemotactic factors specific for endothelial cells: a possible mechanism for tumor-induced angiogenesis. Invas. Metastasis, 3: 139, 1983. 8. Folkman, J.: How is blood vessel growth regulated in normal and neoplastic tissue?-G.H.A. Clowes Memorial Award Lecture. Cancer Res., 46: 467, 1986. 9. Schirrmacher, V. and Vlodawsky, I.: In vitro Interaktionen vbon Gefassendothelzellmonolayern mit Tumorzellinien von unterschiedlicher Metastasierungskapazitat. In: Struktur und Funktion endothelialer Zellen. Edited by Messmer, K. and Hammersen, F. Miinchen:Karger, pp. 34-44, 1983. 10. Masters, J. R. S. and Hepburn, P. J.: Tissue culture model of transitional cell carcinoma: characterization of 22 human urothelial cell lines. Cancer Res., 46: 3630, 1986. 11. Gosporodarowicz, D. Vlodawsky, J. and Savion, N.: The extracellular matrix and the control of proliferation of vascular endothelial and vascular smooth muscle cells. J. Supramol. Struct., 13: 339, 1980. 12. Jaffe, E. H., Nachman, R. L., Becker, D. G. and Minick, C. R.: Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and immunologic criteria. J. Clin. Invest., 52: 2745, 1973. 13. Gratzner, H. G.: Monoclonal antibody to 5- bromo- and 5- iododeoxy-uridine: a new reagent for detection of DNA replication. Science, 218: 474, 1982. 14. Tabarra, W. S., and Mehio, A. R.: Metastatic patterns of bladder carcinoma. In: Progress in Clinical and Biological Research.
INTERACTIONS BETWEEN BLADDER TUMOR CELL LINES AND HUMAN ENDOTHELIAL CELLS
15. 16. 17. 18. 19.
Bladder Cancer. Part A. Edited by Kiiss, R., Khoury, S., Denis, L. J., Murphy, G. P., Karr, J.P. New York: Alan R. Liss, Inc. vol. 162A, pp. 145-160, 1984. O'Toole, C., Price, Z. H., Ohnuki, J. and Unsgaard, B.: Ultrastructure, karyology and immunology of a cell line originated from human transitional cell carcinoma. Br. J. Cancer, 38: 64, 1978. Birdwell, C. R., Gospodarowicz, D. and Nicolson, G. L.: Identification, localization, and role of fibronectin in cultured bovine endothelial cells. Proc. Natl. Acad. Sci. USA, 7: 3273, 1978. Schwartz, S. M. and Haudenschild, C. C.: Endothelial regeneration. Quantitative analysis of initial stages of endothelial regeneration in rat aortic intima. Lab. Invest., 38: 568, 1978. Auksprunk, D. H. and Folkman, J.: Migration and proliferation of endothelial cells in preformed and newly formed blood vessels during tumour angiogenesis. Microvasc. Res., 14: 53, 1977. Kramer, R.H. and Nicolson, G. L.: Interactions of tumor cells with
20. 21.
22. 23.
645
vascular endothelial cell monolayers: a model for metastatic invasion. Proc. Natl. Acad. Sci., 76: 5704, 1979. Nicolson, G. A.: Metastatic tumor cell attachment and invasion assay utilising vascular endothelial cell monolayers. J. Histochem. Cytochem., 30: 214, 1982. Nicolson, G. A., Irimura, T. Nakajima, M. and Estrada, J.: Metastatic cell attachment to an invasion of vasular endothelium and its underlying basal lamina using endothelial cell monolayers. In: Cancer Invasion and Metastasis: Biologic and Therapeutic Aspects. Edited by G. L. Nicolson and L. Milas. New York: Raven Press, pp. 145-167, 1984. Fidler, I. J. and Hart, I. R.: Biological diversity in metastatic neoplasms: origins and implications. Science, 217: 998, 1982. Jones, P. A.: Construction of an artificial blood vessel wall from cultured endothelial and smooth muscle cells. Proc. Natl. Acad. Sci. USA, 76: 1882, 1979.