- Email: [email protected]
air interface supports growth of keratinocytes
EJCB
European Journal of Cell Biology 82, 549 ± 555 (2003, November) ¥ ¹ Urban & Fischer Verlag ¥ Jena http://www.urbanfischer.de/journals/ejcb
549
A basement membrane-like matrix formed by cellreleased proteins at the medium/air interface supports growth of keratinocytes Leopold Eckhart2)a, Christina Reinisch2)a, Sadayuki Inoueb, Paul Messnerc, Michael Dockala, Christoph Mayera, Erwin Tschachler1)a a b c
Department of Dermatology, University of Vienna Medical School, Vienna/Austria Department of Anatomy and Cell Biology, McGill University, Montreal/Canada Center for Ultrastructure Research, University of Natural Resources and Applied Life Sciences and Ludwig Boltzmann Institute for Molecular Nanotechnology, Vienna/Austria
Received July 15, 2003 Received in revised version September 25, 2003 Accepted September 29, 2003
Matrix ± basement membrane ± keratinocytes ± medium/air interface ± cell culture Epithelial cells require adherence to a matrix for regular growth. During standard keratinocyte cell culture in serum-free medium, we observed that cell colonies formed not only on the bottom of the culture vessels but also at the medium/air interface. Coomassie blue staining detected a protein membrane that extended up to several centimeters between the colonies of floating cells. Ultrastructural investigation of this membrane revealed structures closely resembling those of basement membranes, and immunochemical staining confirmed the presence of laminins-1 and -5 as well as collagen IV, representative components of basement membranes. Cells attached to the floating membrane proliferated and could be cultivated for up to six months. When keratinocyte-conditioned medium was filtered and transferred to a culture vessel without cells, the protein membrane at the liquid/air interface formed within one week suggesting self-assembly of cell-released proteins. Our findings provide a basis for the production of epidermal basement membranes for potential medical uses.
Introduction Under physiological conditions epithelia are attached to a basement membrane, a special sheet-like extracellular matrix which forms the interface to mesenchymal tissues. Disturban1)
Professor Dr. Erwin Tschachler, Department of Dermatology, University of Vienna Medical School, Waehringer Guertel 18 ± 20, A-1090 Vienna/Austria, e-mail: [email protected], Fax: 43 1 403 4922. 2) These authors contributed equally to this paper.
ces of the basement membrane caused by auto-immune reactions or genetic deficiencies may lead to disabling and sometimes life-threatening diseases that are difficult to treat or incurable, respectively (Eady and Dunnill, 1994; Schmidt and Zillikens, 2000). After injuries to epithelial tissues, damaged basement membrane is regenerated in the course of the wound healing process (Larjava et al., 1993). During in vitro culture of epithelial cells the plastic or glass surface of the vessel serves as matrix allowing for cell adherence (McAteer and Davis, 1994). In addition, suspended microcarriers or three-dimensional matrices of various composition can be used as culture substrates (McAteer and Davis, 1994; van Welzel, 1967). In culture systems using liquid microcarriers, i.e. emulsions of fluorocarbon droplets (Keese and Giaever, 1983; Giaever and Keese, 1983; Shiba et al., 1998), it has been proposed that a film of denatured proteins at the liquid/liquid interface provides the matrix for the proliferating cells (Giaever and Keese, 1983). Growth of epithelial cells at the liquid/ air interface in the absence of a solid substrate has been reported to occur under certain conditions (Krejci et al., 1991; Tchao, 1989), i.e. cells placed at the medium/air interface developed into colonies which could then be maintained in culture. A similar observation in our laboratories prompted us to investigate and characterize in detail the mechanism that allows keratinocytes to proliferate at the medium/air interface. Here we report the identification of a novel, basement membrane-like layer at the medium/air interface which functioned as the substrate for the culture of keratinocytes and facilitates a remarkably simple way of cultivating adherent cells.
0171-9335/03/82/11-549 $15.00/0
EJCB
550 L. Eckhart et al.
Materials and methods Cell lines and cell culture materials Normal human epidermal keratinocytes (NHEK) derived from neonatal foreskin were purchased from Clonetics (San Diego, CA). The keratinocyte cell line HaCaT (Boukamp et al., 1988) was a kind gift of Dr. Fusenig (German Cancer Research Center, Heidelberg, Germany). Both cell types were maintained in serum-free Keratinocyte Growth Medium (KGM) (Clonetics) either in absence or presence of 0.4% bovine pituitary extract as supplied in the KGM kit. Keratinocytes were grown at 37 8C in a 5% CO2/95% air environment in a humidified atmosphere. Cultures of adherent cells were split after trypsinization according to the protocol of the Clonetics keratinocyte kit. Cells growing at the medium/air interface (see Results section) were subcultured by gently pouring a fraction of the medium together with the floating cells into a new vessel containing fresh medium. Cell-free conditioned medium was prepared by filtering cell culture supernatant of HaCaT cells through a 0.45-mm membrane filter (Poly Labo, Strasbourg, France).
Enzyme treatments of floating cells For enzyme treatment of floating cell colonies, the culture medium beneath the membrane at the medium/air interface was carefully removed and replaced by 10 ml PBS to wash off medium remnants. After removing the PBS, 10 ml of one of the following enzyme solutions were pipetted beneath the membrane and incubated at 37 8C for 5 minutes: Trypsin-EDTA (0.05% trypsin, 0.53 mM EDTA ¥ 4Na in Hanks× Balanced Salt Solution without CaCl2, MgCl2 ¥ 6H2O and MgSO4 ¥ 7H2O, Gibco BRL, Gaithersburg, MD); Collagenase crude type IA (225 U/ml collagen digestion activity, 0.05% in PBS) (Sigma, St. Louis, MO). DNase I (Boehringer Mannheim, Germany) at a final concentration of 1 mg/ml was added to prevent cell clumping during protease treatments.
Coomassie blue and immunostaining of floating cells Cell colonies floating at the medium/air interface were mounted on microscope slides, air-dried and fixed with acetone. Then the slides were dipped for 5 seconds into protein staining solution (aqueous solution containing 47.5% ethanol, 10% acetic acid, and 2.0 g/l Coomassie Brilliant Blue R250 (Phast Gel Blue R purchased from Pharmacia, Uppsala, Sweden). Unbound dye was removed by repeated washes with destaining solution (aqueous solution containing 50% methanol, 7.5% acetic acid). Alternatively, floating cells were stained directly by replacing the medium in the culture dish by the protein staining solution. After one minute the unbound dye was removed by repeated washes with PBS. Photographs of the floating cells with the then visualized intercellular matrix were taken with an inverted microscope. For immunostaining, floating colonies at the medium/air interface were mounted on microscopy slides, fixed with acetone and air-dried. The slides were rehydrated with PBS, and endogenous peroxidase was blocked with 0.3% H2O2 in a buffer solution containing 2% bovine serum albumin and 10% sheep or goat serum in PBS, for 30 minutes. The samples were incubated for 60 minutes at room temperature with either a rat anti-laminin-1 antibody recognizing a confirmational epitope localized on the laminin B1-B2 heterodimer and in the P1 fragment of laminin (Immunotech, Marseille, France; 4 mg/ml) or a mouse antilaminin-5 monoclonal antibody that binds to 150- and 105-kDa proteins of laminin-5 (Chemicon International, Temecula, CA; 5 mg/ml) or a mouse anti-collagen type IV antibody reacting to the human type IV collagen, a2 chains (Chemicon) or a mouse anti-collagen type VII antibody directed against the carboxyterminal peptide (Chemicon). For control, either rat IgG (Santa Cruz, Santa Cruz, California) or mouse IgG1 (Coulter, Miami, Florida) were used. After extensive washes, biotinylated sheep anti-rat IgG (Amersham, Little Chalfont, England; 1 : 200) and sheep anti-mouse IgG (Amersham, Little Chalfont, England; 1 : 200) for immunohistochemistry or Alexa Fluor 488 goat anti-mouse IgG (Molecular Probes, Eugene, Oregon) for immunofluo-
rescence were reacted to the respective first antibodies for 30 min. For immunohistochemistry, antibody binding was visualized with StreptABComplex/HRP (Dako, Glostrup, Denmark) and DAB chromogen (Dako, Carpinteria, California). Immunohistochemistry samples were counterstained with hematoxylin and immunofluorescence ones with propidium iodide. For detection of proliferating cells, adherent and floating cell populations were reacted to a monoclonal anti-Ki67 antibody (Novocastra, Newcastle upon Tyne, UK) followed by a peroxidase-labelled secondary antibody as described above. Only homogeneously decorated nuclei were regarded as positives.
Ultrastructural investigations The proteinaceous matrix at the medium/air interface both in areas without and in areas with cells adhering to it were transferred to freshly glow-discharged carbon film-coated copper grids by gentle touching of the matrix with the coated grid surface oriented towards the culture dish. Subsequently, the samples were fixed by floating the grids for 15 min on a drop of 2.5% glutaraldehyde in water before washing on three drops of water. Negative staining was performed for 1 min with 1% uranyl acetate solution in water (Hayat and Miller, 1990). The negatively stained preparations were examined in a Philips CM100 electron microscope (Philips, Eindhoven, The Netherlands) at 80 kVaccelerating voltage. The magnification was calibrated with catalase crystals. For preparation of the basement membrane of the seminiferous epithelium, seminiferous tubules were isolated, under anesthesia, from an adult Sherman rat and cut into small pieces. The tissue pieces were subjected to immersion fixation using the method of Todd and Tokito (1981). They were placed in a 3% solution of potassium permanganate in glucose-containing Krebs-Henseleit saline at 4 8C for 30 minutes and stained en bloc with 1% uranyl acetate for 45 min. Then they were dehydrated and embedded in Epon. Thin sections were counter-stained with uranyl acetate and lead citrate for observations of the seminiferous epithelium in the electron microscope. The contrast of electron micrographs of this basement membrane was reversed for comparison with the negatively stained membrane from the medium/air interface cell culture described above.
Results Cultured keratinocytes spontaneously form colonies at the medium/air interface During standard culture of the keratinocyte cell line HaCaT and of primary human keratinocytes, we repeatedly noticed the appearance of cells at the medium/air interface (Fig. 1A, arrow, focus on medium/air interface). These cells exhibited the same morphology as adherent keratinocytes (Fig. 1B, arrowhead, focus on bottom of culture vessel) and were proliferating as confirmed by staining for Ki67. Ki67 staining was found in 2.4% and 1.9% of adherent and floating cell populations, respectively. For further characterization of these cells and possible associated structures, subcultures of the floating cells were made by gently pouring them into a new flask. After this transfer, the majority of cells remained floating at the medium/ air interface and resumed proliferation. Medium changes were performed by removing and replenishing the medium beneath the floating cells with a pipette. To initiate the formation of floating cell colonies more efficiently, we modified an approach described by Tchao (1989): after trypsinization of adherent cells, medium drops containing suspended cells were placed onto the lid of an inverted culture vessel. When the vessel was subsequently reinverted to its regular position, cells formed aggregates at the lower surface of the hanging drops. After 20 to 30 minutes the bottom of the vessel was covered with medium and the vessel was tapped, leading to detachment of the hanging drops. When
EJCB
Novel basement membrane-like matrix 551
Fig. 1. Keratinocytes are able to grow at the medium/air interface. During culture of HaCaT cells, cell colonies floating at the medium/air interface (A, arrow) were regularly observed in addition to adherent
cells at the bottom of the culture vessel (B, arrowhead). Both panels depict the identical sector of the culture vessel focusing either on the medium/air interface (A) or on the bottom (B). 60.
Fig. 2. Coomassie blue staining detects a matrix at the medium/air interface of keratinocyte cultures. When keratinocyte colonies floating at the medium surface were stained with Coomassie Brilliant Blue directly in the culture vessel (A) an intercellular membrane became visible (A, arrow). This membrane covered large areas between cell
colonies. Distortions of some parts of this membrane after mounting onto microscopy slides prior to fixation (B) indicate that it could resist lateral tension caused by the preparation procedure (B, arrows). 60 (A); 150 (B).
the drops merged with the culture medium, a significant fraction of the cells remained at the medium/air interface and gave rise to floating colonies. Primary keratinocytes could be grown at the medium/air interface for several passages. HaCaT cells were cultured this way for up to six months without a significant decline in cell viability or growth rate.
To characterize this matrix, we incubated areas of the medium surface that contained colonies with a solution of Coomassie Brilliant Blue. Both cells and the area in between (Fig. 2A, arrow) were stained, confirming the proteinaceous nature of the floating matrix. Most cells were part of large cell colonies whereas others formed little islets (Fig. 2A). The uniform staining of the area between cells indicated that the thickness of the matrix did not vary significantly. Interestingly, the shape of some fragments of the matrix suggested that it could at least partly resist lateral tensions that occurred during sample preparation (Fig. 2B, arrows). To investigate the interaction of the cells with the presumable matrix, the culture medium underneath the cells was replaced with solutions of various proteolytic enzymes. Whereas the buffer solution without enzyme did not affect the behavior of floating cells, trypsin incubation caused rounding-up of cells and loss of cell-to-cell contact. Despite the contact loss these
A proteinaceous matrix extends between cells floating at the medium/air interface We noticed that the relative position of floating cell colonies at distances of up to several centimeters remained unchanged even when the medium was gently agitated by tapping on the vessel. These observations strongly suggested the presence of a floating intercellular matrix to which the cells at the medium/air interface were adhered.
552 L. Eckhart et al.
EJCB
Fig. 3. The floating matrix contains the basement membrane proteins laminin-1 and 5. Immunostaining of the floating matrix revealed the
presence of laminin-1 (A) and laminin-5 (C). Isotype controls are shown in (B) and (D), respectively. 100 (A, B); 200 (C, D).
rounded cells remained attached to the surface and did not move relative to each other. These data suggested that between the cells a matrix existed which was less trypsin sensitive than the intercellular contacts. After further incubation for several minutes and agitation of the enzyme solution cells finally sank to the bottom of the culture vessel. By contrast, after treatment with collagenase, intercellular contacts remained intact but cell colonies started to move relative to each other suggesting that the matrix between the cells had been degraded. Together, these findings suggest that the cells at the medium surface adhere to a collagenase-sensitive proteinaceous matrix via trypsin-sensitive bonds.
anchoring fibrils which connect the basement membrane with the underlying stroma was detected within floating keratinocytes but not in the membrane (not shown). Ultrastructural investigation of the floating matrix revealed that it was composed of strands which were irregular both in their course and thickness (Fig. 5A). Thin (3 nm) and thick (8 ± 10 nm) strands were often seen as being organized into network-like assemblies. For comparison, the thin-sectioned basement membrane of rat seminiferous epithelium, a typical basement membrane, is shown in Figure 5B. It is composed of a three-dimensional network of irregular strands referred to as ™cords∫ (Inoue and Leblond, 1988; Inoue, 1989; Inoue, 1994). In this basement membrane, cords vary in their thickness from 3 to 6 nm. Together with the finding of effective cell adherence and the immunodetection of laminins and collagen IV, these ultrastructural observations suggest that the floating matrix at the medium/air interface is indeed a basement membrane-like entity.
The floating matrix contains collagen IV, laminin-1 and -5 and exhibits ultrastructural features of basement membranes Since keratinocytes produce and secrete proteins that participate in the formation of the epidermal basement membrane (Rosdy et al., 1993), we hypothesized that the floating matrix at the medium/air interface might contain such proteins. Immunochemical analysis for laminin-1 (Fig. 3A), laminin-5 (Fig. 3C), and collagen IV (Fig. 4) confirmed that these constituents of basement membranes were indeed present within the matrix. By contrast, collagen VII, a component of
A proteinaceous membrane self-assembles at the liquid/air interface of HaCaT-conditioned medium in the absence of cells Since laminin-1 and collagen IV are able to self-assemble into macromolecular structures (reviewed in (Timpl and Brown,
EJCB
Novel basement membrane-like matrix 553
Fig. 4. The floating matrix contains collagen IV. Immunofluorescence staining of the floating matrix revealed the presence of collagen IV (A,
green). Nuclei were counterstained with propidium iodide (red). The isotype control is shown in (B). 200.
1996)), we hypothesized that formation of the membrane at the liquid/air interface may be independent of the contact with cells. To test this hypothesis, we filtered HaCaT-conditioned medium through a 0.45-mm filter to remove cells and membrane sheets. Subsequently the filtered conditioned medium was transferred to a cell-free culture vessel. Every 24 hours part of the medium was replaced with new filtered supernatant from the HaCaT culture. After 7 days Coomassie blue staining revealed that a membrane had formed in vessels containing conditioned medium (Fig. 6) but not in such with culture medium which had not been exposed to cells previously (not shown). Therefore, proteins released from keratinocytes into the medium are able by themselves to form a floating membrane.
epithelial cells can differentiate in the absence of extracellular matrix (Krejci et al., 1991). We believe that also in these culture systems a matrix was present but that it was then either lost during preparation or ignored. The novel matrix has several features characteristic of basement membranes. At the ultrastructural level, the network of irregular strands is strikingly similar to the cord network of typical basement membranes (Inoue, 1989; Inoue, 1994) although the thickness of a fraction of the strands in the membrane was slightly larger than that of basement membrane cords. Like basement membranes, the matrix contained laminin-1, laminin-5, and collagen IV. Since only keratinocytes were present in the culture vessels, any basement membrane components derived from mesenchymal cells in vivo were absent from the novel matrix. Also, several components, such as collagen VII produced from the epithelium, may not be integrated. Future studies aimed at the full characterization of composition of the novel matrix and elucidation of differences and similarities to basement membranes developing in vivo are currently ongoing in our laboratories. An important aspect of our findings is the assembly process of this matrix. In our culture system the formation of the newly described proteinaceous matrix was crucially dependent on cellular factors as no matrix sheets were formed in unconditioned medium. However, physical contact with cells is not necessary for matrix formation as evidenced by detection of membrane formation on the surface of conditioned medium after filtration. This finding suggests that the floating matrix sheets self-assemble from cell-secreted proteins. We believe that this process is initiated by adsorption of proteins to the liquid/air interface, a phenomenon that has been characterized extensively in cell-free systems (Jiang and Chiew, 2001; Deyme et al., 1986, 1987). In conclusion we have found that proteins released from keratinocytes in culture form basement membrane-like float-
Discussion To the best of our knowledge, this is the first description of a floating proteinaceous matrix at the liquid/air interface that supports proliferating cells. This phenomenon was observed both in keratinocyte cultures as well as in cultures of epithelial kidney cells (our unpublished results). We believe that the growth of cells at the medium surface has previously been overlooked because, during microscopic inspection of cell cultures, the focus is usually made at the bottom of the culture flask. Furthermore, failure to notice cell growth at the medium/ air interface might have resulted from prevention of this phenomenon by the presence of serum in the growth medium (not shown) and by the fact that cells at the medium surface are lost in the course of conventional medium changes. In previous studies it was shown that cell aggregates can be maintained at the medium/air interface under certain conditions (Krejci et al., 1991; Tchao, 1989). However, these authors did not detect a sheet-like extracellular matrix within their cultures (Krejci et al., 1991; Tchao, 1989) and even claimed that floating
554 L. Eckhart et al.
Fig. 5. Electron micrographs of the matrix formed at medium/air interface and, for comparison, the basement membrane of the rat seminiferous epithelium. Negative staining of matrix formed at the medium/air interface revealed wide (pair of short arrows) as well as less abundant, narrow (long arrow) rod-like entities that were often organized into a network structure (A). For comparison, a positively stained thin section of the basement membrane of rat seminiferous epithelium, cut slightly obliquely, is shown in (B). The contrast of this
EJCB
micrograph was reversed so that both wide (6 nm) (pair of short arrows) and narrow (3 nm) (long arrow) irregular ™cords∫ appear white. These cords are organized into a network, the main constituent of the lamina densa (D) of the basement membrane. The lamina lucida (L) appears dark. SE, seminiferous epithelium. Bars 100 nm. (Fig. 5B is modified from Fig. 5b of (Inoue and Leblond, Am. J. Anat. 181, 341 ± 358, 1988). Copyright ¹ 1988. Reprinted by permission of Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc.).
EJCB
Fig. 6. Membrane formation on the surface of conditioned medium after filtration. Cell-free supernatant of HaCaT cell cultures was filtered through a 0.45-mm membrane filter and transferred to a fresh culture dish. After incubation at 37 8C for 1 week the medium surface was touched with a microscope slide and the attached material was stained with Coomassie blue. 200.
ing matrix sheets which allow for the adherence and continuous proliferation of cells. Production of this floating matrix on a large scale might provide the opportunity to use it together with attached cells or alone for wound treatment and for covering medical devices and implants, respectively. Currently, studies are underway in our laboratories to explore the potentials of the newly discovered basement membrane-like matrix. Acknowledgements. We thank Andrea Scheberl for the help with the EM sample preparation, Chung-I Wu and J¸rgen Bach for technical support and Regina Voglauer for providing cells.
References Boukamp, P., Petrussevska, R. T., Breitkreutz, D., Hornung, J., Markham, A., Fusenig, N. E. (1988): Normal keratinization in a spontaneously immortalized aneuploid human keratinocyte cell line. J. Cell Biol. 106, 761 ± 771. Deyme, M., Baszkin, A., Proust, J. E., Perez, E., Albrecht, G., Boissonnade, M. M. (1987): Collagen at interfaces. II: Competitive adsorption of collagen against albumin and fibrinogen. J. Biomed. Mater. Res. 21, 321 ± 328. Deyme, M., Baszkin, A., Proust, J. E., Perez, E., Boissonnade, M. M. (1986): Collagen at interfaces. I. In situ collagen adsorption at
Novel basement membrane-like matrix 555
solution/air and solution/polymer interfaces. J. Biomed. Mater. Res. 20, 951 ± 962. Eady, R. A., Dunnill, M. G. (1994): Epidermolysis bullosa: hereditary skin fragility diseases as paradigms in cell biology. Arch. Dermatol. Res. 287, 2 ± 9. Giaever, I., Keese, C. R. (1983): Behaviour of cells at fluid interfaces. Proc. Natl. Acad. Sci. USA 80, 219 ± 222. Hayat, M. A., Miller, S. E. (1990): Negative staining. McGraw-Hill, New York. Inoue, S., Leblond, C. P. (1988): Three-dimensional network of cords: the main component of basement membranes. Am. J. Anat. 181, 341 ± 358. Inoue, S. (1989): Ultrastructure of basement membranes. Int. Rev. Cytol. 117, 57 ± 98. Inoue, S. (1994): Basic structure of basement membranes is a fine network of ™cords∫, irregular anastomosing strands. Microsc. Res. Techn. 28, 29 ± 47. Jiang, Q., Chiew, Y. C. (2001): Dynamics of adsorption and desorption of proteins at an air/water interface. Colloids Surf. B Biointerfaces 20, 303 ± 308. Keese, C. R., Giaever, I. (1983): Cell growth on liquid microcarriers. Science 219, 1448 ± 1449. Krejci, N. C., Smith, L., Rudd, R., Langdon, R., McGuire, J. (1991): Epithelial differentiation in the absence of extracellular matrix. In Vitro Cell. Dev. Biol. 27, 933 ± 938. Larjava, H., Salo, T., Haapasalmi, K., Kramer, R. H., Heino, J. (1993): Expression of integrins and basement membrane components by wound keratinocytes. J. Clin. Invest. 92, 1425 ± 1435. McAteer, J. A., Davis, J. (1994): Basic cell culture technique and the maintenance of cell lines. In: Davis, J. (Ed.): Basic cell culture: a practical approach. Oxford University Press, Oxford, pp. 93 ± 148. Rosdy, M., Pisani, A., Ortonne, J. P. (1993): Production of basement membrane components by a reconstructed epidermis cultured in the absence of serum and dermal factors. Br. J. Dermatol. 129, 227 ± 234. Schmidt, E., Zillikens, D. (2000): Autoimmune and inherited subepidermal blistering diseases: advances in the clinic and the laboratory. Adv. Dermatol. 16, 113 ± 157. Shiba, Y., Ohshima, T., Sato, M. (1998): Growth and morphology of anchorage-dependent animal cells in a liquid/liquid interface system. Biotechnol. Bioeng. 57, 583 ± 589. Tchao, R. (1989): Epithelial cell interaction in air-liquid interface culture. In Vitro Cell. Dev. Biol. 25, 460 ± 465. Timpl, R., Brown, J. C. (1996): Supramolecular assembly of basement membranes. Bioessays 18, 123 ± 132. Todd, M. E., Tokito, M. K. (1981): Improved ultrastructural detail in tissues fixed with potassium permanganate. Stain Technol. 56, 335 ± 342. van Welzel, A. L. (1967): Growth of cell-strains and primary cells on micro-carriers in homogeneous culture. Nature 216, 64 ± 65.
European Journal of Cell Biology 82, 549 ± 555 (2003, November) ¥ ¹ Urban & Fischer Verlag ¥ Jena http://www.urbanfischer.de/journals/ejcb
549
A basement membrane-like matrix formed by cellreleased proteins at the medium/air interface supports growth of keratinocytes Leopold Eckhart2)a, Christina Reinisch2)a, Sadayuki Inoueb, Paul Messnerc, Michael Dockala, Christoph Mayera, Erwin Tschachler1)a a b c
Department of Dermatology, University of Vienna Medical School, Vienna/Austria Department of Anatomy and Cell Biology, McGill University, Montreal/Canada Center for Ultrastructure Research, University of Natural Resources and Applied Life Sciences and Ludwig Boltzmann Institute for Molecular Nanotechnology, Vienna/Austria
Received July 15, 2003 Received in revised version September 25, 2003 Accepted September 29, 2003
Matrix ± basement membrane ± keratinocytes ± medium/air interface ± cell culture Epithelial cells require adherence to a matrix for regular growth. During standard keratinocyte cell culture in serum-free medium, we observed that cell colonies formed not only on the bottom of the culture vessels but also at the medium/air interface. Coomassie blue staining detected a protein membrane that extended up to several centimeters between the colonies of floating cells. Ultrastructural investigation of this membrane revealed structures closely resembling those of basement membranes, and immunochemical staining confirmed the presence of laminins-1 and -5 as well as collagen IV, representative components of basement membranes. Cells attached to the floating membrane proliferated and could be cultivated for up to six months. When keratinocyte-conditioned medium was filtered and transferred to a culture vessel without cells, the protein membrane at the liquid/air interface formed within one week suggesting self-assembly of cell-released proteins. Our findings provide a basis for the production of epidermal basement membranes for potential medical uses.
Introduction Under physiological conditions epithelia are attached to a basement membrane, a special sheet-like extracellular matrix which forms the interface to mesenchymal tissues. Disturban1)
Professor Dr. Erwin Tschachler, Department of Dermatology, University of Vienna Medical School, Waehringer Guertel 18 ± 20, A-1090 Vienna/Austria, e-mail: [email protected], Fax: 43 1 403 4922. 2) These authors contributed equally to this paper.
ces of the basement membrane caused by auto-immune reactions or genetic deficiencies may lead to disabling and sometimes life-threatening diseases that are difficult to treat or incurable, respectively (Eady and Dunnill, 1994; Schmidt and Zillikens, 2000). After injuries to epithelial tissues, damaged basement membrane is regenerated in the course of the wound healing process (Larjava et al., 1993). During in vitro culture of epithelial cells the plastic or glass surface of the vessel serves as matrix allowing for cell adherence (McAteer and Davis, 1994). In addition, suspended microcarriers or three-dimensional matrices of various composition can be used as culture substrates (McAteer and Davis, 1994; van Welzel, 1967). In culture systems using liquid microcarriers, i.e. emulsions of fluorocarbon droplets (Keese and Giaever, 1983; Giaever and Keese, 1983; Shiba et al., 1998), it has been proposed that a film of denatured proteins at the liquid/liquid interface provides the matrix for the proliferating cells (Giaever and Keese, 1983). Growth of epithelial cells at the liquid/ air interface in the absence of a solid substrate has been reported to occur under certain conditions (Krejci et al., 1991; Tchao, 1989), i.e. cells placed at the medium/air interface developed into colonies which could then be maintained in culture. A similar observation in our laboratories prompted us to investigate and characterize in detail the mechanism that allows keratinocytes to proliferate at the medium/air interface. Here we report the identification of a novel, basement membrane-like layer at the medium/air interface which functioned as the substrate for the culture of keratinocytes and facilitates a remarkably simple way of cultivating adherent cells.
0171-9335/03/82/11-549 $15.00/0
EJCB
550 L. Eckhart et al.
Materials and methods Cell lines and cell culture materials Normal human epidermal keratinocytes (NHEK) derived from neonatal foreskin were purchased from Clonetics (San Diego, CA). The keratinocyte cell line HaCaT (Boukamp et al., 1988) was a kind gift of Dr. Fusenig (German Cancer Research Center, Heidelberg, Germany). Both cell types were maintained in serum-free Keratinocyte Growth Medium (KGM) (Clonetics) either in absence or presence of 0.4% bovine pituitary extract as supplied in the KGM kit. Keratinocytes were grown at 37 8C in a 5% CO2/95% air environment in a humidified atmosphere. Cultures of adherent cells were split after trypsinization according to the protocol of the Clonetics keratinocyte kit. Cells growing at the medium/air interface (see Results section) were subcultured by gently pouring a fraction of the medium together with the floating cells into a new vessel containing fresh medium. Cell-free conditioned medium was prepared by filtering cell culture supernatant of HaCaT cells through a 0.45-mm membrane filter (Poly Labo, Strasbourg, France).
Enzyme treatments of floating cells For enzyme treatment of floating cell colonies, the culture medium beneath the membrane at the medium/air interface was carefully removed and replaced by 10 ml PBS to wash off medium remnants. After removing the PBS, 10 ml of one of the following enzyme solutions were pipetted beneath the membrane and incubated at 37 8C for 5 minutes: Trypsin-EDTA (0.05% trypsin, 0.53 mM EDTA ¥ 4Na in Hanks× Balanced Salt Solution without CaCl2, MgCl2 ¥ 6H2O and MgSO4 ¥ 7H2O, Gibco BRL, Gaithersburg, MD); Collagenase crude type IA (225 U/ml collagen digestion activity, 0.05% in PBS) (Sigma, St. Louis, MO). DNase I (Boehringer Mannheim, Germany) at a final concentration of 1 mg/ml was added to prevent cell clumping during protease treatments.
Coomassie blue and immunostaining of floating cells Cell colonies floating at the medium/air interface were mounted on microscope slides, air-dried and fixed with acetone. Then the slides were dipped for 5 seconds into protein staining solution (aqueous solution containing 47.5% ethanol, 10% acetic acid, and 2.0 g/l Coomassie Brilliant Blue R250 (Phast Gel Blue R purchased from Pharmacia, Uppsala, Sweden). Unbound dye was removed by repeated washes with destaining solution (aqueous solution containing 50% methanol, 7.5% acetic acid). Alternatively, floating cells were stained directly by replacing the medium in the culture dish by the protein staining solution. After one minute the unbound dye was removed by repeated washes with PBS. Photographs of the floating cells with the then visualized intercellular matrix were taken with an inverted microscope. For immunostaining, floating colonies at the medium/air interface were mounted on microscopy slides, fixed with acetone and air-dried. The slides were rehydrated with PBS, and endogenous peroxidase was blocked with 0.3% H2O2 in a buffer solution containing 2% bovine serum albumin and 10% sheep or goat serum in PBS, for 30 minutes. The samples were incubated for 60 minutes at room temperature with either a rat anti-laminin-1 antibody recognizing a confirmational epitope localized on the laminin B1-B2 heterodimer and in the P1 fragment of laminin (Immunotech, Marseille, France; 4 mg/ml) or a mouse antilaminin-5 monoclonal antibody that binds to 150- and 105-kDa proteins of laminin-5 (Chemicon International, Temecula, CA; 5 mg/ml) or a mouse anti-collagen type IV antibody reacting to the human type IV collagen, a2 chains (Chemicon) or a mouse anti-collagen type VII antibody directed against the carboxyterminal peptide (Chemicon). For control, either rat IgG (Santa Cruz, Santa Cruz, California) or mouse IgG1 (Coulter, Miami, Florida) were used. After extensive washes, biotinylated sheep anti-rat IgG (Amersham, Little Chalfont, England; 1 : 200) and sheep anti-mouse IgG (Amersham, Little Chalfont, England; 1 : 200) for immunohistochemistry or Alexa Fluor 488 goat anti-mouse IgG (Molecular Probes, Eugene, Oregon) for immunofluo-
rescence were reacted to the respective first antibodies for 30 min. For immunohistochemistry, antibody binding was visualized with StreptABComplex/HRP (Dako, Glostrup, Denmark) and DAB chromogen (Dako, Carpinteria, California). Immunohistochemistry samples were counterstained with hematoxylin and immunofluorescence ones with propidium iodide. For detection of proliferating cells, adherent and floating cell populations were reacted to a monoclonal anti-Ki67 antibody (Novocastra, Newcastle upon Tyne, UK) followed by a peroxidase-labelled secondary antibody as described above. Only homogeneously decorated nuclei were regarded as positives.
Ultrastructural investigations The proteinaceous matrix at the medium/air interface both in areas without and in areas with cells adhering to it were transferred to freshly glow-discharged carbon film-coated copper grids by gentle touching of the matrix with the coated grid surface oriented towards the culture dish. Subsequently, the samples were fixed by floating the grids for 15 min on a drop of 2.5% glutaraldehyde in water before washing on three drops of water. Negative staining was performed for 1 min with 1% uranyl acetate solution in water (Hayat and Miller, 1990). The negatively stained preparations were examined in a Philips CM100 electron microscope (Philips, Eindhoven, The Netherlands) at 80 kVaccelerating voltage. The magnification was calibrated with catalase crystals. For preparation of the basement membrane of the seminiferous epithelium, seminiferous tubules were isolated, under anesthesia, from an adult Sherman rat and cut into small pieces. The tissue pieces were subjected to immersion fixation using the method of Todd and Tokito (1981). They were placed in a 3% solution of potassium permanganate in glucose-containing Krebs-Henseleit saline at 4 8C for 30 minutes and stained en bloc with 1% uranyl acetate for 45 min. Then they were dehydrated and embedded in Epon. Thin sections were counter-stained with uranyl acetate and lead citrate for observations of the seminiferous epithelium in the electron microscope. The contrast of electron micrographs of this basement membrane was reversed for comparison with the negatively stained membrane from the medium/air interface cell culture described above.
Results Cultured keratinocytes spontaneously form colonies at the medium/air interface During standard culture of the keratinocyte cell line HaCaT and of primary human keratinocytes, we repeatedly noticed the appearance of cells at the medium/air interface (Fig. 1A, arrow, focus on medium/air interface). These cells exhibited the same morphology as adherent keratinocytes (Fig. 1B, arrowhead, focus on bottom of culture vessel) and were proliferating as confirmed by staining for Ki67. Ki67 staining was found in 2.4% and 1.9% of adherent and floating cell populations, respectively. For further characterization of these cells and possible associated structures, subcultures of the floating cells were made by gently pouring them into a new flask. After this transfer, the majority of cells remained floating at the medium/ air interface and resumed proliferation. Medium changes were performed by removing and replenishing the medium beneath the floating cells with a pipette. To initiate the formation of floating cell colonies more efficiently, we modified an approach described by Tchao (1989): after trypsinization of adherent cells, medium drops containing suspended cells were placed onto the lid of an inverted culture vessel. When the vessel was subsequently reinverted to its regular position, cells formed aggregates at the lower surface of the hanging drops. After 20 to 30 minutes the bottom of the vessel was covered with medium and the vessel was tapped, leading to detachment of the hanging drops. When
EJCB
Novel basement membrane-like matrix 551
Fig. 1. Keratinocytes are able to grow at the medium/air interface. During culture of HaCaT cells, cell colonies floating at the medium/air interface (A, arrow) were regularly observed in addition to adherent
cells at the bottom of the culture vessel (B, arrowhead). Both panels depict the identical sector of the culture vessel focusing either on the medium/air interface (A) or on the bottom (B). 60.
Fig. 2. Coomassie blue staining detects a matrix at the medium/air interface of keratinocyte cultures. When keratinocyte colonies floating at the medium surface were stained with Coomassie Brilliant Blue directly in the culture vessel (A) an intercellular membrane became visible (A, arrow). This membrane covered large areas between cell
colonies. Distortions of some parts of this membrane after mounting onto microscopy slides prior to fixation (B) indicate that it could resist lateral tension caused by the preparation procedure (B, arrows). 60 (A); 150 (B).
the drops merged with the culture medium, a significant fraction of the cells remained at the medium/air interface and gave rise to floating colonies. Primary keratinocytes could be grown at the medium/air interface for several passages. HaCaT cells were cultured this way for up to six months without a significant decline in cell viability or growth rate.
To characterize this matrix, we incubated areas of the medium surface that contained colonies with a solution of Coomassie Brilliant Blue. Both cells and the area in between (Fig. 2A, arrow) were stained, confirming the proteinaceous nature of the floating matrix. Most cells were part of large cell colonies whereas others formed little islets (Fig. 2A). The uniform staining of the area between cells indicated that the thickness of the matrix did not vary significantly. Interestingly, the shape of some fragments of the matrix suggested that it could at least partly resist lateral tensions that occurred during sample preparation (Fig. 2B, arrows). To investigate the interaction of the cells with the presumable matrix, the culture medium underneath the cells was replaced with solutions of various proteolytic enzymes. Whereas the buffer solution without enzyme did not affect the behavior of floating cells, trypsin incubation caused rounding-up of cells and loss of cell-to-cell contact. Despite the contact loss these
A proteinaceous matrix extends between cells floating at the medium/air interface We noticed that the relative position of floating cell colonies at distances of up to several centimeters remained unchanged even when the medium was gently agitated by tapping on the vessel. These observations strongly suggested the presence of a floating intercellular matrix to which the cells at the medium/air interface were adhered.
552 L. Eckhart et al.
EJCB
Fig. 3. The floating matrix contains the basement membrane proteins laminin-1 and 5. Immunostaining of the floating matrix revealed the
presence of laminin-1 (A) and laminin-5 (C). Isotype controls are shown in (B) and (D), respectively. 100 (A, B); 200 (C, D).
rounded cells remained attached to the surface and did not move relative to each other. These data suggested that between the cells a matrix existed which was less trypsin sensitive than the intercellular contacts. After further incubation for several minutes and agitation of the enzyme solution cells finally sank to the bottom of the culture vessel. By contrast, after treatment with collagenase, intercellular contacts remained intact but cell colonies started to move relative to each other suggesting that the matrix between the cells had been degraded. Together, these findings suggest that the cells at the medium surface adhere to a collagenase-sensitive proteinaceous matrix via trypsin-sensitive bonds.
anchoring fibrils which connect the basement membrane with the underlying stroma was detected within floating keratinocytes but not in the membrane (not shown). Ultrastructural investigation of the floating matrix revealed that it was composed of strands which were irregular both in their course and thickness (Fig. 5A). Thin (3 nm) and thick (8 ± 10 nm) strands were often seen as being organized into network-like assemblies. For comparison, the thin-sectioned basement membrane of rat seminiferous epithelium, a typical basement membrane, is shown in Figure 5B. It is composed of a three-dimensional network of irregular strands referred to as ™cords∫ (Inoue and Leblond, 1988; Inoue, 1989; Inoue, 1994). In this basement membrane, cords vary in their thickness from 3 to 6 nm. Together with the finding of effective cell adherence and the immunodetection of laminins and collagen IV, these ultrastructural observations suggest that the floating matrix at the medium/air interface is indeed a basement membrane-like entity.
The floating matrix contains collagen IV, laminin-1 and -5 and exhibits ultrastructural features of basement membranes Since keratinocytes produce and secrete proteins that participate in the formation of the epidermal basement membrane (Rosdy et al., 1993), we hypothesized that the floating matrix at the medium/air interface might contain such proteins. Immunochemical analysis for laminin-1 (Fig. 3A), laminin-5 (Fig. 3C), and collagen IV (Fig. 4) confirmed that these constituents of basement membranes were indeed present within the matrix. By contrast, collagen VII, a component of
A proteinaceous membrane self-assembles at the liquid/air interface of HaCaT-conditioned medium in the absence of cells Since laminin-1 and collagen IV are able to self-assemble into macromolecular structures (reviewed in (Timpl and Brown,
EJCB
Novel basement membrane-like matrix 553
Fig. 4. The floating matrix contains collagen IV. Immunofluorescence staining of the floating matrix revealed the presence of collagen IV (A,
green). Nuclei were counterstained with propidium iodide (red). The isotype control is shown in (B). 200.
1996)), we hypothesized that formation of the membrane at the liquid/air interface may be independent of the contact with cells. To test this hypothesis, we filtered HaCaT-conditioned medium through a 0.45-mm filter to remove cells and membrane sheets. Subsequently the filtered conditioned medium was transferred to a cell-free culture vessel. Every 24 hours part of the medium was replaced with new filtered supernatant from the HaCaT culture. After 7 days Coomassie blue staining revealed that a membrane had formed in vessels containing conditioned medium (Fig. 6) but not in such with culture medium which had not been exposed to cells previously (not shown). Therefore, proteins released from keratinocytes into the medium are able by themselves to form a floating membrane.
epithelial cells can differentiate in the absence of extracellular matrix (Krejci et al., 1991). We believe that also in these culture systems a matrix was present but that it was then either lost during preparation or ignored. The novel matrix has several features characteristic of basement membranes. At the ultrastructural level, the network of irregular strands is strikingly similar to the cord network of typical basement membranes (Inoue, 1989; Inoue, 1994) although the thickness of a fraction of the strands in the membrane was slightly larger than that of basement membrane cords. Like basement membranes, the matrix contained laminin-1, laminin-5, and collagen IV. Since only keratinocytes were present in the culture vessels, any basement membrane components derived from mesenchymal cells in vivo were absent from the novel matrix. Also, several components, such as collagen VII produced from the epithelium, may not be integrated. Future studies aimed at the full characterization of composition of the novel matrix and elucidation of differences and similarities to basement membranes developing in vivo are currently ongoing in our laboratories. An important aspect of our findings is the assembly process of this matrix. In our culture system the formation of the newly described proteinaceous matrix was crucially dependent on cellular factors as no matrix sheets were formed in unconditioned medium. However, physical contact with cells is not necessary for matrix formation as evidenced by detection of membrane formation on the surface of conditioned medium after filtration. This finding suggests that the floating matrix sheets self-assemble from cell-secreted proteins. We believe that this process is initiated by adsorption of proteins to the liquid/air interface, a phenomenon that has been characterized extensively in cell-free systems (Jiang and Chiew, 2001; Deyme et al., 1986, 1987). In conclusion we have found that proteins released from keratinocytes in culture form basement membrane-like float-
Discussion To the best of our knowledge, this is the first description of a floating proteinaceous matrix at the liquid/air interface that supports proliferating cells. This phenomenon was observed both in keratinocyte cultures as well as in cultures of epithelial kidney cells (our unpublished results). We believe that the growth of cells at the medium surface has previously been overlooked because, during microscopic inspection of cell cultures, the focus is usually made at the bottom of the culture flask. Furthermore, failure to notice cell growth at the medium/ air interface might have resulted from prevention of this phenomenon by the presence of serum in the growth medium (not shown) and by the fact that cells at the medium surface are lost in the course of conventional medium changes. In previous studies it was shown that cell aggregates can be maintained at the medium/air interface under certain conditions (Krejci et al., 1991; Tchao, 1989). However, these authors did not detect a sheet-like extracellular matrix within their cultures (Krejci et al., 1991; Tchao, 1989) and even claimed that floating
554 L. Eckhart et al.
Fig. 5. Electron micrographs of the matrix formed at medium/air interface and, for comparison, the basement membrane of the rat seminiferous epithelium. Negative staining of matrix formed at the medium/air interface revealed wide (pair of short arrows) as well as less abundant, narrow (long arrow) rod-like entities that were often organized into a network structure (A). For comparison, a positively stained thin section of the basement membrane of rat seminiferous epithelium, cut slightly obliquely, is shown in (B). The contrast of this
EJCB
micrograph was reversed so that both wide (6 nm) (pair of short arrows) and narrow (3 nm) (long arrow) irregular ™cords∫ appear white. These cords are organized into a network, the main constituent of the lamina densa (D) of the basement membrane. The lamina lucida (L) appears dark. SE, seminiferous epithelium. Bars 100 nm. (Fig. 5B is modified from Fig. 5b of (Inoue and Leblond, Am. J. Anat. 181, 341 ± 358, 1988). Copyright ¹ 1988. Reprinted by permission of Wiley-Liss, Inc., a subsidiary of John Wiley & Sons, Inc.).
EJCB
Fig. 6. Membrane formation on the surface of conditioned medium after filtration. Cell-free supernatant of HaCaT cell cultures was filtered through a 0.45-mm membrane filter and transferred to a fresh culture dish. After incubation at 37 8C for 1 week the medium surface was touched with a microscope slide and the attached material was stained with Coomassie blue. 200.
ing matrix sheets which allow for the adherence and continuous proliferation of cells. Production of this floating matrix on a large scale might provide the opportunity to use it together with attached cells or alone for wound treatment and for covering medical devices and implants, respectively. Currently, studies are underway in our laboratories to explore the potentials of the newly discovered basement membrane-like matrix. Acknowledgements. We thank Andrea Scheberl for the help with the EM sample preparation, Chung-I Wu and J¸rgen Bach for technical support and Regina Voglauer for providing cells.
References Boukamp, P., Petrussevska, R. T., Breitkreutz, D., Hornung, J., Markham, A., Fusenig, N. E. (1988): Normal keratinization in a spontaneously immortalized aneuploid human keratinocyte cell line. J. Cell Biol. 106, 761 ± 771. Deyme, M., Baszkin, A., Proust, J. E., Perez, E., Albrecht, G., Boissonnade, M. M. (1987): Collagen at interfaces. II: Competitive adsorption of collagen against albumin and fibrinogen. J. Biomed. Mater. Res. 21, 321 ± 328. Deyme, M., Baszkin, A., Proust, J. E., Perez, E., Boissonnade, M. M. (1986): Collagen at interfaces. I. In situ collagen adsorption at
Novel basement membrane-like matrix 555
solution/air and solution/polymer interfaces. J. Biomed. Mater. Res. 20, 951 ± 962. Eady, R. A., Dunnill, M. G. (1994): Epidermolysis bullosa: hereditary skin fragility diseases as paradigms in cell biology. Arch. Dermatol. Res. 287, 2 ± 9. Giaever, I., Keese, C. R. (1983): Behaviour of cells at fluid interfaces. Proc. Natl. Acad. Sci. USA 80, 219 ± 222. Hayat, M. A., Miller, S. E. (1990): Negative staining. McGraw-Hill, New York. Inoue, S., Leblond, C. P. (1988): Three-dimensional network of cords: the main component of basement membranes. Am. J. Anat. 181, 341 ± 358. Inoue, S. (1989): Ultrastructure of basement membranes. Int. Rev. Cytol. 117, 57 ± 98. Inoue, S. (1994): Basic structure of basement membranes is a fine network of ™cords∫, irregular anastomosing strands. Microsc. Res. Techn. 28, 29 ± 47. Jiang, Q., Chiew, Y. C. (2001): Dynamics of adsorption and desorption of proteins at an air/water interface. Colloids Surf. B Biointerfaces 20, 303 ± 308. Keese, C. R., Giaever, I. (1983): Cell growth on liquid microcarriers. Science 219, 1448 ± 1449. Krejci, N. C., Smith, L., Rudd, R., Langdon, R., McGuire, J. (1991): Epithelial differentiation in the absence of extracellular matrix. In Vitro Cell. Dev. Biol. 27, 933 ± 938. Larjava, H., Salo, T., Haapasalmi, K., Kramer, R. H., Heino, J. (1993): Expression of integrins and basement membrane components by wound keratinocytes. J. Clin. Invest. 92, 1425 ± 1435. McAteer, J. A., Davis, J. (1994): Basic cell culture technique and the maintenance of cell lines. In: Davis, J. (Ed.): Basic cell culture: a practical approach. Oxford University Press, Oxford, pp. 93 ± 148. Rosdy, M., Pisani, A., Ortonne, J. P. (1993): Production of basement membrane components by a reconstructed epidermis cultured in the absence of serum and dermal factors. Br. J. Dermatol. 129, 227 ± 234. Schmidt, E., Zillikens, D. (2000): Autoimmune and inherited subepidermal blistering diseases: advances in the clinic and the laboratory. Adv. Dermatol. 16, 113 ± 157. Shiba, Y., Ohshima, T., Sato, M. (1998): Growth and morphology of anchorage-dependent animal cells in a liquid/liquid interface system. Biotechnol. Bioeng. 57, 583 ± 589. Tchao, R. (1989): Epithelial cell interaction in air-liquid interface culture. In Vitro Cell. Dev. Biol. 25, 460 ± 465. Timpl, R., Brown, J. C. (1996): Supramolecular assembly of basement membranes. Bioessays 18, 123 ± 132. Todd, M. E., Tokito, M. K. (1981): Improved ultrastructural detail in tissues fixed with potassium permanganate. Stain Technol. 56, 335 ± 342. van Welzel, A. L. (1967): Growth of cell-strains and primary cells on micro-carriers in homogeneous culture. Nature 216, 64 ± 65.