THE ANATOMICAL RECORD 217:229-239 (1987) Morphogenetic Clonal Growth of Kidney Epithelia1 Cell Line MDCK JAMES A. McATEER, ANDREW P. EVAN, AND KENNETH D.GARDNER Department of Anatomy, Indiana University School of Medicine, Indianapolis, IN 46223 (J.A.M., A.P.E.);Department of Medicine, University of New Mexico School of Medicine, Albuquerque, NM 87106 (K.D.G.) ABSTRACT MDCK (Madin-Darby canine kidney) cells were cultured either 1) dispersed within hydrated collagen gel (HCG) or 2) seeded atop a collagen substrate and then immediately overlaid with HCG. Individual cells exhibited clonal growth in three dimensions to form spherical cysts made up of a simple epithelium enclosing a fluid-filled lumen. The cells of MDCK cysts were polarized with the basolateral surface in contact with the collagen gel and the apical surface bordering the lumen. The ultrastructure of MDCK cysts showed similarities to distal nephron. The cells bore apical microvilli and solitary cilia and had occluding junctions and a simple basolateral surface. MDCK cysts increased in size (> 800 pm diameter) with continued culture. MDCK cysts grown between layers of HCG were stripped free of the overlying collagen to give direct access to basolateral surface membrane. Unlike monolayer culture, morphogenetic clonal growth of cell line MDCK produces a polarized cell population with a true lumenal and basolateral surface. Collagen-gelcultured MDCK cysts provide a n easily manipulable in uitro cell system that may offer unique advantages for the study of renal cell structure and function. The kidney epithelial cell line MDCK (Madin-Darby canine kidney) (Madin and Darby, 1975) retains in culture many of the differentiated characteristics of a cell of the distal nephron (Rindler et al., 1979; Taub and Saier, 1979). The presumed distal nephron origin of the MDCK cell is suggested by its ultrastructure (Valentich, 1981)and pattern of hormone responsiveness (Rindler et al., 1979) and by the expression of a basolateral cell surface antigen that is specific to thin ascending limb and distal convoluted tubule in the dog (Herzlinger et al., 1982). In addition, monoclonal antibodies raised against MDCK cells have been localized to cortical and medullary collecting duct of the dog (Garcia-Perez and Smith, 1983).Although certain morphological characteristics of the MDCK cell have been shown to be dependent upon culture conditions (Valentich, 1981), in monolayer, the cells are cuboidal, bear apical microvilli and solitary cilia, produce basal lamina, and have tight junctions (Cereijido et al., 1978; Valentich, 1981). The physiology of the MDCK cell makes it a valuable model for studies of vectorial fluid and ion transport (Misfeldt et al., 1976). When grown to confluency on a permeable substrate, the cells produce a transepithelial electrical resistance (Cereijido, 1984). When raised on a n impermeable substrate, MDCK cells develop multicellular domes (hemicysts or blisters) (Leighton et al., 1969; Cereijido et al., 1981). Domes form when fluid transported from apical to basolateral surface accumulates beneath the cell sheet, lifting region of the intact cell layer away from the culture substrate (Leighton et al., 1969).Domes are transient structures that may form and then collapse (Leighton et al., 1969; Cereijido et al., 1981).It has been shown that dome formation by MDCK 0 1987 ALAN R. LISS, INC. cells is dependent upon active Naf transport (Abaza et al., 1977; Lever, 1979a) and is influenced by a variety of factors including CAMP(Lever, 1979a; Valentich et al., 1979), various inducers of mammalian cell differentiation (Lever, 1979a,b), and cell-substrate interactions (Rabit0 et al., 1980). Besides its role as a model for fluid transport by a renal cell type, MDCK has also been used to study aspects of cell-matrix interactions. Hall et al. (1982) observed that MDCK cells cultured in monolayer atop a collagen membrane exhibit a “morphogenetic” response when overlaid with collagen gel. When confronted with collagen on both basolateral and apical surfaces the cell sheet reorganizes to form a collection of tubulelike epithelial elements, each with the apical cell surface facing a n enclosed lumen. This observation complements work performed on nonrenal epithelial cell types in which mouse mammary epithelium (Bennett, 1980; Bennett et al., 1981; Yang et al., 1979, 1980) and hog thyroid cells (Chambard et al., 1981)have also been shown to exhibit histotypic reorganization and growth when cultured within collagen gel. Such culture methods, which employ collagen gel a s the substrate, are a n alternative to monolayer culture and provide a means to study the effects of a three-dimensional extracellular matrix on cell behavior. In this report we describe the formation of polarized epithelial cysts by MDCK cells suspended within collagen gel. Cyst formation under these culture conditions occurs by the successive division (clonal growth) of isoReceived June 24, 1985; accepted August 12, 1986. 230 J.A. McATEER, A.P. EVAN, AND K.D. GARDNER lated single cells. A distinct advantage of this system is the nature of the collagen gel substrate, which is easily manipulated to provide direct physical access to the cultured cell population. Morphogenetic clonal growth of MDCK cells within collagen gel may provide a novel means to model the influence of the kidney interstitium on the growth, structure, and function of the renal tubular epithelium. MATERIALS AND METHODS Culture of MDCK Cells Suspended Within Collagen Gel MDCK cells (CCL 34, passage 53) obtained from the American Type Culture Collection (Rockville, MD) were subcultured two or more times at subconfluency in serum-free medium patterned after medium K-1 of Taub and Sat0 (Taub and Sato, 1979). Culture medium was a 1:l mixture of nutrient medium F12 and DMEM (KC Biological, Lenexa, KS) containing 2.0 mg/ml sodium bicarbonate, 10 mM HEPES, 10 nM sodium selenite, 0.1% BSA, and 100 U/ml penicillin-G and supplemented with bovine insulin, 5 pg/ml, hydrocortisone 10-7 M, PGE1, 25 ng/ml, and T3, 5 x 10-l2 M (all from Sigma Chemical Co., St. Louis, MO). The cells were trypsinized and then washed twice in medium containing 0.1% soybean trypsin inhibitor (Boehringer-Mannheim, Indianapolis, IN). Cells were resuspended in culture medium to give 2.5 x 105 cells/ml and then gently triturated through a n 18-gauge cannula and filtered (McAteer and Cavanagh, 1983) through 20-pm Nitex screen (Tetko Inc., Elmsford, NY). The dispersed cell suspension (4.3 ml) was added to a chilled collagen-medium solution. The collagen-medium solution was prepared by combining 1 ml medium FlBK(1OX) (from CaCl2-deleted powder, special order, KC Biological, Lenexa, KS), 0.5 ml 5% NaHC03, 1ml 0.1 M HEPES, 100 p1 0.2 M CaC12, and 100 pl 1 N NaOH with 3 ml of collagen stock, prepared from rat tail tendon (McAteer and Cavanagh, 1982).Collagen gels were plated in 35-mm dishes (1ml), placed in a n incubator (5%CO2 in air, 37°C) and allowed to solidify (5 minutes). Each gel was then overlaid with 1.5 ml of culture medium that was replaced every other day. Cultures were also established from MDCK cells raised in medium without supplemental hormones but containing 5%fetal bovine serum (KC Biological, Lenexa, KS). Cultures Seeded Between Layers of Collagen Gel Cultures were also established by seeding MDCK cells between layers of collagen gel. Culture dishes were prepared with a solidified base layer of collagen gel. Cells (2 x 105)were planted and allowed to settle (5 minutes); then the medium was removed. The cells were overlaid with collagen gel, which was allowed to solidify (as above) before additional medium was added. This method was used to prepare cultures to be viewed on a daily basis for documentation of clonal growth. Cultures were plated within Bionique chamber dishes (Corning, Corning, NY) assembled with a glass coverslip. The plastic lid of the chamber was replaced with a sterile glass plate prior to photomicrography by using differential interference contrast (DIC) optics. Preparation of Electron Microscopy Cultures were fixed in a mixture containing 1.25% glutaraldehyde and 1%OSOQin 0.1 M cacodylate-HC1 buffer (pH 7.41, en bloc stained in 0.25% uranyl acetate in 0.1 M sodium acetate buffer (pH 6.3), and dehydrated in a graded series of ethanols. Selected cultures were fixed in the presence of ruthenium red (5 pg/ml). Specimens for transmission electron microscopy were embedded in Epon 812. Thin sections were double-stained with uranyl acetate and lead citrate. For scanning electron microscopy, intact cultures were cryofractured from liquid nitrogen-frozen absolute ethanol (Humphreys et al., 1974). Cultures established by seeding cells between layers of HCG were fixed (as above) after removing the upper collagen layer by using fine forceps. This exposed MDCK cyst basolateral membranes for direct observation. MDCK cysts were also isolated from living cultures by digestion of the collagen matrix with 0.1% collagenase (type IV Worthington, Millipore Corp., Freehold, NJ). Isolated MDCK cysts were processed by using carriers fashioned from Nitex (Tetko, Elmsford, NY) screen (McAteer et al., 1983). Specimens for SEM were critical-point dried (Anderson, 19511, coated with gold-palladium, and viewed by using a n AMR lOOOA scanning electron microscope. Radioautography MDCK cyst cultures at day 15 were incubated for 24 hours in medium containing 1 pCi/ml [meth~l-~H-]-thymidine (specific activity 20.0 Ci/mM) (New England Nuclear, Boston, MA). The cultures were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate/HCl, pH 7.4, dehydrated in graded ethanols, and embedded in JB-4 resin (Polysciences, Warrington, PA). Sections were cut a t approximately 3 pm and placed on glass slides. Slides were dipped in Kodak NTB-2 nuclear track emulsion (Kodak Co., Rochester, NY) and then stored in light-tight boxes containing CaS04. Slide boxes were placed in a refrigerator a t 8°C for exposure periods of 2, 4, or 6 weeks. Following development in D-19, the tissue sections were counterstained with 0.03% basic fuchsin. Nuclei were considered to be positive for 3H-thymidine uptake when they possessed 4~ the number of grains observed over the background. RESULTS Clonal Growth Within HCG MDCK cells seeded within collagen gel formed spherical, multicellular, cystlike structures (Figs. 1-4) in which a simple cuboidal epithelium surrounded a central lumen (Figs. 2-4, 5). The epithelium was polarized, with the cell apex bordering the lumen (Figs. 2,5). MDCK cysts developed by 72-96 hours following seeding. Initially, cysts were small (< 20 pm) but increased greatly in size ( > 800 pm) with continued culture. Cyst formation was not synchronous and cysts did not appear to increase in size a t the same rate. During this study we recorded multiple selected fields, using the inverted photomicroscope, at 12 hour intervals throughout the first 10 days of culture. Cells within a given field did not move from original position and prior to cyst formation remained rounded in form. Only single isolated cells appeared within the field (Fig. 3A,B; 2 days). We observed no cells with elongated or irregular shape that would suggest movement through the collagen gel. Cells sumended within collagen pel did not aggregate. MDCK kysts appeared, therifore: to form as the result of the division (clonal growth) of isolated individual cells. The series of light micrographs displayed v v MORPHOGENETIC CLONAL GROWTH OF MDCK CELLS 231 Fig. 1. Scanning electron micrograph of collagen-gelcultured MDCK cysts 12 days following seeding. This culture was established by mixing a monodisperse cell suspension within fluid collagen prior to plating. Numerous cysts (arrows) are exposed at the fractured surface of the collagen gel (HCG). ~ 4 7 . in Figure 3 shows two different fields (Fig. 3A, B) recorded at successive time points over a 10-day period. The first frame of each set, recorded at day 2, shows only single cells within the field. Subsequent frames recorded at days 4 and 10 show the formation of a t least one cyst per field derived from a single cell. These micrographs show the development of cysts from nonmigratory, isolated single cells, with no evidence of the recruitment of cells from the surrounding interstitium. In addition, this series of photographs demonstrates that not all isolated cells were capable of cyst formation. Moreover, these frames show that MDCK cells initiated cyst formation at different times following seeding and that cysts grew in diameter at different rates. The subsequent growth of MDCK cysts appeared to occur by cell division and not by the addition of cells from the interstitium. To test this, cysts were harvested from the collagen substrate by treatment with collagenase. Cultures included those in the initial period of cyst formation (72-96 hours) and older cultures of up to 1421 days. The population of cysts was washed free of single cells; then individual cysts, some less than 20 pm in diameter, were collected by using a micropipette. These isolated cysts were replanted within collagen gel that was free of dissociated cells. We observed that replanted cysts continued to grow in diameter and often grew to be several hundred micrometers in diameter. Such replanted cysts, especially those of larger diameter, often appeared to accumulate a modest number of cells within the lumen. These cells were likely sloughed from the cyst wall, possibly in response to treatment with collagenase. The morphology of isolated-replanted cysts was otherwise indistinguishable from cysts growing within the original cultures. To further demonstrate that cyst growth involved mitotic activity, cultures were incubated with tritiated thy- midine and processed for light microscopical radioautography. Figure 2B shows a n MDCK cyst bearing cells with labeled nuclei. This indicates that cells are added to the cyst wall by mitotic division. Ultrastructure of Clonal MDCK Cysts MDCK cysts were formed of a simple epithelium. The cells bore abundant apical microvilli and commonly possessed a single cilium (Figs. 2, 5). Lateral cell surfaces were usually simple, with few villi or infoldings (Figs. 2, 5). Desmosomes occurrred intermittently along the intercellular interface. Occluding junctions were present adjacent to the lumen (Figs. 2, 5). Cells typically possessed a well-developed complement of organelles, including numerous mitochondria and a n elaborate Golgi complex. Golgi stacks were commonly located on the lumenal or lateral side of the nucleus (Figs. 2, 5). The cytoplasm also contained smooth-surfaced vesicles and appeared to have only a modest amount of rough endoplasmic reticulum. Cells occasionally contained large membrane-bound vacuoles containing material of moderate density (Fig. 2). The basal surface of MDCK cysts was relatively simple. Whereas cysts isolated by collagenase (Fig. 4) appeared to possess shallow clefts at intercellular margins, specimens prepared by stripping away the overlying collagen prior to fixation (Fig. 6) showed a smooth basal surface. The uneven surface of enzyme-isolated cysts was likely due to partial collapse of the cysts during the isolation procedure. Indeed, many cysts were found to tear and collapse completely during isolation with collagenase. MDCK cells appeared, by routine TEM (Figs. 2, 51, to lack a well-organized, continuous basal lamina. Extracellular material occurred, instead, in irregular patches. Fixation in the presence of ruthenium red, however, Fig. 2. Structure of MDCK cysts cultured within collagen ge1.A Light micrograph of living culture (10 days) showing numerous spherical cysts of various sizes. A large cyst (arrow) near the center of the field measures approximately 800 pm in diameter. X59. B: Light microscopy radioautograph of a plastic-embedded section from a 16day culture incubated with tritiated thymidine. Several cells (arrows) show labeled nuclei. This suggests that mitotic activity was involved in the growth of this cyst. ~ 8 9 6C: . Transmission electron micrograph of a solitary MDCK cyst a t 10 days following seeding. Cells are cuboidal, bear apical microvilli (mv), and are joined by tight junctions (circle) and desmosomes (arrows). These contiguous cells form a polarized epithelium with basal cell surface in contact with the surrounding collagen matrix (HCG). X4,300. D: Scanning electron micrograph of a cryofractured 12-day culture shows a spherical cyst surrounded by collagen matrix (HCG). Apical cell surface is visible facing the lumen. Each cell appears to possess a solitary cilium (arrows). ~ 8 9 6 . MORPHOGENETIC CLONAL GROWTH OF MDCK CELLS 233 Fig. 3. MDCK cyst formation by clonal growth. These panels (A,B) demonstrate the clonal growth of solitary MDCK cells to form polarized multicellular epithelia1 cysts. Two separate fields (3A, B) were recorded by using differential interference contrast optics at days 2,4, and 10 following seeding. Frames recorded at day 2 show only single cells within the field. In both panels arrows mark the position of an individual cell at day 2 that subsequently undergoes cyst formation. It can be observed that not all single cells at day 2 underwent clonal growth. ~ 2 6 2 . demonstrated abundant flocculent electron-dense material associated with basolateral membranes (Fig. 5). Cell shape within the population of cells composing individual cysts was generally uniform. The great majority of cysts were made up of cells having cuboidal to low-cuboidal morphology. Some cysts possessed moderately attenuated cells. Some degree of cell attenuation appeared to accompany cyst growth; however, large cysts rarely possessed highly attenuated cells. Generally, cysts were spherical (Figs. 2,4). Cyst shape was however, influenced by physical conditions of the HCG matrix (see below) and by the cell population as well. Cysts fused with one another if they came in contact during growth (Fig. 6).If contact and fusion occured 234 J.A. McATEER, A.P. EVAN, AND K.D. GARDNER Fig. 4. Isolated MDCK cyst. Scanning electron micrograph of an intact single MDCK cyst isolated from culture by digestion with collagenase. This cyst is a simple sphere. Basolateral cell surface is visible. Cell boundaries appear demarcated (arrows) by shallow clefts induced during isolation and preparation for microscopy. x 1,098. between cysts of small to moderate size, the resultant fused cyst usually grew to assume a spherical form. The cysts of cultures in which cells were seeded between HCG layers were not always spherical (Fig. 6) (see below). MDCK cysts were fluid-filled. Cysts never contained collagen; however, lumina commonly contained a flocculant precipitate (Figs. 2, 5). Material within lumina was preserved by routine processing for TEM but was frequently displaced during cryofracture for SEM. Freefloating cells were often seen within cysts (Fig. 21, particularly in older cultures. The ultrastructure of MDCK cysts cultured in serumfree media was indistinguishable from cysts grown in serum-supplemented medium. MDCK Cyst Formation Between HCG Layers Cysts that developed in cultures in which cells were planted at the interface between layers of collagen gel (Fig. 6 ) often formed as moderately flattened sacs instead of spheres. Whereas many cysts appeared initially to be spherical, nearly all developed flattened morphology with continued culture. The ultrastructure of cysts in these cultures was indistinguishable from cysts raised in collagen suspension. In layered gel cultures, growing cysts often came into contact with one another. When this occurred the cysts fused and their lumina coalesced (Fig. 6). This pattern of development resulted in the formation of large irregular channels with a common interconnecting lumen and a n epithelia1 wall formed by the recruitment of numerous individual cysts. MORPHOGENETIC CLONAL GROWTH OF MDCK CELLS Fig. 5. Ultrastructure of MDCK-cyst wall. A Transmission electron micrograph shows portions of three cells. Apical surface bears microvilli (mv), while lateral and basal surfaces are relatively simple. Golgi stacks and smooth-surfaced vesicles appear polarized toward the cell apex. Basal lamina is not readily evident, although patches of apparent extracellular material are present (arrow). The lumen contains a flocculent precipitate. Collagen gel (HCG). ~ 5 , 3 3 6 B: . Scanning elec- 235 tron micrograph of cryofractured MDCK cyst shows abundant microvilli (mv) at the apical cell surface. Each cell also bears a long filamentous single cilium (arrow). X3,955. C: Transmission electron micrograph demonstrates ruthenium-red-positive extracellular material (arrows) along the basal surface and lateral interface between adjacent cells. x 57,600. 236 J.A. McATEER, A.P. EVAN, AND K.D. GARDNER Fig. 6. Cultures established between layers of collagen gel. A Scanning electron micrograph shows MDCK cysts at the interface between the base and overlying layers of collagen gel. Cysts are not distributed throughout the matrix but, instead, occupy the zone between collagen gel layers. This specimen was prepared by stripping away the upper collagen layer prior to fixation. X276. B: Light micrograph of a living culture (10 days in vitro) demonstrates the irregular pattern of MDCK cysts that develops when cells are seeded at the interface between collagen layers. Cysts develop initially as solitary structures (arrowhead) but coalesce when they grow into contact (arrows) with one another. Continued culture results in formation of large epithelia1 channels. ~ 6 3 C.: Scanning electron micrograph of a single MDCK cyst from a specimen stripped free of its overlying collagen layer immediately before fixation. The exposed basal surface of the epithelium is smooth and regular. Cell boundaries (arrows) are tightly apposed. Collagen gel (HCG). X812. MORPHOGENETIC CLONAL GROWTH OF MDCK CELLS DISCUSSION We observe that the growth of MDCK cells dispersed within collagen gel produces clonal colonies that display the histotypic orientation of a kidney tubule. Apical cell surfaces face a n enclosed lumen while basolateral surfaces border the interstitium. This morphogenetic form of growth is likely due to multiple factors including both the physical and biochemical composition of the collagen gel substrate and intrinsic properties of the MDCK cell type. Our conclusion that the formation of MDCK cysts is due to clonal growth of individual isolated cells is based on frequent and repeated observations of living cultures with light microscopy. It can be argued that intermittent observations might not disclose cell aggregation involving rapid cell migration. However, the fact that we have not observed any migratory cells at any time during the period of cyst formation, in hundreds of cultures, indicates that cell movement does not occur, and that cyst formation must, therefore, involve cell division. Greenburg and Hay (1982) have demonstrated that certain embryonic and adult epithelia may lose polarity and exhibit migratory behavior when cultured within collagen gel. Such cells have a distinctive morphology. They are bipolar or spindle shaped and possess elongated filopodia (Green and Hay, 1982).We do not observe cells with this structure in our cultures. Instead, MDCK cells within collagen gel remain rounded. Following division the newly formed cells have a cuboidal shape, as part of the cyst wall epithelium. Thus our observations indicate that MDCK cysts form due to mitotic activity of individual cells (clonal growth). Likewise, we conclude that subsequent growth in diameter of MDCK cysts involves cell division and not cell aggregation. We observed that cysts isolated from culture and then replanted within fresh collagen gel continued to increase in size and cell number. Growth under these conditions could only be due to cell division. In addition, we demonstrated the nuclear incorporation of tritiated thymidine in cells of growing cysts. This indicates that mitotic activity occurred during cyst growth. Morphogenetic behavior is exhibited in a variety of cell systems in which hydrated collagen gel is used as the culture substrate. Mouse mammary epithelium forms ductlike tubules within collagen gel (Yang et al., 1979; Yang et al., 1980; Bennett, 1980; Bennett et al., 1981). These structures form as outgrowths from mammary tumor explants embedded within collagen gel (Bennett, 1980) or from cloned tumor cell lines cultured atop a collagen substrate (Yang et al., 1979, 1980). In addition, histotypic growth occurs when normal mouse mammary cells are dispersed within collagen gel (Bennett, 1980). Tubule formation by mammary epithelial cells has been shown to involve cell aggregation, reorganization, and subsequent growth. This seems to be the case with other morphogenetic culture systems a s well, including follicle formation by dispersed hog thyrocytes (Chambard et al., 1981).Chambard et al. (1981)demonstrated that thyroid epithelial cells seeded atop a base layer of collagen gel and then overlaid with collagen will reorganize to form follicles in which the cells show the apex-to-lumen polarity of the thyroid follicle in vivo. Similar behavior is shown by MDCK cells cultured un- 237 der similar conditions. Hall et al. (1982)observed that a confluent monolayer of MDCK cells cultured atop collagen will form multiple small cysts when overlaid with collagen. In this case lumen formation is due to reorganization of the cell sheet, not clonal growth. Cyst formation by MDCK cells suspended within a three-dimensional matrix has been reported by others. Leighton (1981) observed the development of polarized epithelial cysts by MDCK cells seeded within a plasma fibrin clot. Cell orientation of fibrin-cultured cysts was similar to cysts we observed in collagen gels. Leighton’s study (1981) did not attempt to document the growth behavior of isolated cells; therefore, it is not known if cyst formation within a fibrin matrix occurs by clonal growth or, perhaps, involves cell aggregation. The mechanism by which collagen gel promotes morphogenetic cell behavior is not known. Nor is it known why certain cell types form histotypic structures within collagen gels while others do not (Greenburg and Hay, 1982). Collagen clearly influences epithelial polarity, but again, the mechanism of this interaction is not known. Experimental evidence from several systems indicates that the polarity of a n epithelial cell is destabilized when collagen is applied to its apical surface. When thyroid epithelial cells are cultured in fluid suspension they form everted follicles in which the cell apex faces the medium and basolateral surface encloses a lumen (Mauchamp et al., 1979). These everted follicles reverse polarity when collagen is added to the medium or when the follicles are embedded in collagen (Chambard et al., 1981).The cells, therefore, appear to reverse polarity in order to avoid contact of the apical surface with the surrounding collagen. Understandably, when a monolayer of MDCK cells grown on a collagen substrate is, in turn, overlaid with collagen (Hall et al., 1982), polarity reversal will not free the apical surface from contact with collagen. Instead, the cell sheet must reorganize to create a lumenal space for apical surface to border. We observe that when MDCK cells are dispersed within collagen gel the cells remain rounded and do not migrate through the matrix. Cyst formation in this system, therefore, does not involve reaggregation but is dependent on cell division. The process of cyst formation thus involves the development of a polarized epithelium by the successive division (clonal growth) of a single cell. Our observations suggest a series of events to describe this clonal growth process. It is likely that MDCK cells dissociated from monolayer culture lose polarity when they are suspended within collagen. A cell fully surrounded by collagen is in contact over its entire surface with substrate normally bordered by basolateral membrane. Because the isolated cell cannot reverse polarity to avoid exposure of its apical surface to collagen it probably eliminates apical surface altogether. The cell must divide to reestablish polarity. We speculate that when division occurs, daughter cells remain contiguous. Their region of contact would create a collagen-free area of the cell surface and provide a n interface that each cell could polarize against. This would create a region of membrane free to develop into apical surface. Thus, with polarity established, all subsequent cell division would occur within the plane of the epithelium. The lumen would increase in diameter as cells are added by division, to the cyst wall. Tight junctions would be estab- 238 J.A. McATEER, A.P. EVAN, AND K.D. GARDNER lished between daughter cells at each division; thus, the lumen would never be continous with the surrounding interstitium. A collagen matrix is not essential to support cyst formation by MDCK cells. Other three-dimensional matrices such a s soft agar can be used. The nature of the substrate does, however, determine the polarity of the cell population. We have grown MDCK cells suspended within soft agar. As in collagen gel, the freshly isolated cells remain rounded and do not migrate. Although cysts do form in soft agar, cell polarity is opposite that of the cysts that form in collagen gel. In soft agar, the apical cell surface faces the agar medium while the basolateral surface borders the enclosed lumen. Thus, the physical support provided by a three-dimensional matrix does not, in itself, determine cell polarity. It is possible that fluid transport is involved in the growth of MDCK cysts within collagen gel. We have purposely ruptured isolated cysts before replanting them within collagen. Collapsed cysts commonly reexpand to original size within 48-72 hours, far faster than could be attributed to growth by cell division. We have not, however, determined if the reexpansion of a collapsed cyst or the growth of a n intact cyst requires active fluid transport. If unidirectional fluid transport does occur during cyst growth the polarity of the cyst wall would dictate that the direction of fluid flow must be opposite that which occurs in vivo. This likewise would be opposite the direction of fluid transport that takes place during dome formation in monolayer culture (Leighton et al., 1969; Abaza et al., 1977; Lever, 1979a,b; Rabito et al., 1980; Cereijido et al., 1981). At present the literature describing cyst formation by cells dispersed within a three-dimensional matrix is somewhat limited (Leighton, 1981). Furthermore, the process of morphogenetic clonal growth that we have described for the MDCK cell has not been documented for other cell types. This may suggest that morphogenetic clonal growth is a n uncommon cell behavior; however, it may be that other cell types are indeed capable of this form of growth, but have not been studied under similar culture conditions. Not all cells of the parent MDCK cell population are capable of morphogenetic clonal growth. We observe single cells within gels after even extended culture. These cells can be harvested by filtration-separation following digestion of the collagen matrix, and will seed, with variable plating efficiency, to form a viable monolayer. It therefore may be that clonal growth under these conditions is a population selection process. Clonal growth in three dimensions offers potential methodological advantages applicable to cell selection and clonal analysis. We have demonstrated that MDCK cysts can develop from single cells. Each confirmed (see below) clonal cyst thus possesses a homogeneous cell population in contrast to the recognized heterogeneity of the parent line (Lewis and Spector, 1981; Richardson et al., 1981; Valentich, 1981). Individual cysts can be isolated and passaged to monolayer or collagen gel for continued culture. This provides a unique method for clonal selection of MDCK sublines, as in isolation of agent-resistant or sensitive strains. It must be noted that in this system MDCK cysts may also develop from clusters of cells not fully dissociated during preparation of the original cell suspension. In- deed, the degree to which true clonal growth occurs is influenced by the efficiency of preparation of a monodisperse cell population at the time of seeding. Assurance that a n individual cyst is a true clonal cell population, therefore, requires visual verification by sequential observations of the field. Further value of the cellular homogeneity of MDCK clonal cysts may become apparent should studies focus on the biology of individual cysts. It seems reasonable to suggest that investigation of a discrete clonally derived cell population will allow a more precise assessment of the functional properties of this kidney cell line. The culture system we have described is a n alternative to monolayer culture that may provide the opportunity for novel experimental design in studies on the biology of the MDCK cell. The physical characteristics of the system are unique. We are aware of no other kidney epithelial cell culture system that allows essentially unrestricted access to basolateral cell surface. Also, unlike monolayer culture, the cells of MDCK cysts are exposed to nutrient supply solely through their basolatera1 surface. This mimics the relationship of the vasculature to the basal cell surface in vivo and offers the potential for selective manipulation of the “in vitro interstitial” compartment. In addition, it is possible to alter the physical culture substrate of MDCK cysts. Not. only can the composition of the collagen gel matrix be tailored prior to seeding, but the substrate surrounding growing cysts can be removed or replaced without disrupting cell-cell interactions. Thus, the system may prove useful for study of the influence of extracellular matrix on kidney epithelial cell structure and function. ACKNOWLEDGMENTS The authors wish to thank Philip Blomgren, Bret Connors, and Ellen Vance for their valuable technical assistance and Claudette Maurer and Mary Henderson for typing the manuscript. This study was supported by grant number AM3003 from the National Institutes of Health and by the PKR Foundation (Kansas City). LITERATURE CITED Abaza, N.A., J. Leigbton, and S.G. Schultz (1977) Effects of Ouabain on the function and structure of a cell line (MDCK) derived from canine kidney. I. Light microscopic observations of monolayer growth. In Vitro, 103172-183. Anderson, T.F. (1951) Technique for the preservation of three dimensional structure in preparing specimens for the electron microscope. Trans. N.Y. Acad. Sci., 23:130-134. Bennett, D.C. (1980) Morphogenesis of branching tubules in cultures of cloned mammary epithelial cells. Nature, 285:657-659. Bennett, D.C., B.L. Armstrong, and S.M. Okada (1981) Reconstitution of branching tubules from two cloned mammary cell types in culture. Dev. Biol., 873193-199. Cereijido, M. (1984) Electrical properties of Madin-Darby canine kidney cells. Fed. Proc., 43:2230-2235. Cereijido, M., J. Ehrenfeld, S. Fernandez-Castelo, and I. Meza (1981) F l u e s , junctions, and blisters in cultured monolayers of epithelioid cells (MDCK). Ann. N.Y. Acad. Sci., 773422-441. Cereijido, M., E.S. Robbins, W.J., Dolan, C.A. Rotunno, and D.D. Sabatini (1978) Polarized monolayers formed by epithelial cells on a permeable and translucent support. J. Cell Biol., 773853-880. Chambard, M., J. Gabrion, and J. Mauchamp (1981) Influence of collagen gel on the orientation of epithelial cell polarity: Follicle formation from isolated thryoid cells and from preformed monolayers. J. Cell Biol., 91:157-166. Garcia-Perez, A., and W.L. Smith (1983) Use of monoclonal antibodies to isolate cortical collecting tubule cells: AVP induces PGE release. Am. J. Physiol., 23:C211-c220. MORPHOGENETIC CLONAL GROWTH OF MDCK CELLS Greenburg, G., and E.D. Hay (1982) Epithelia suspended in collagen gels can lose polarity and express characteristics of migrating mesenchymal cells. J. Cell Biol., 95:333-339. Hall, G.H., D.A. Farson, and M.J. Bissell (1982) Lumen formation by epithelial cell lines in response to collagen overlay: A morphogenetic model in culture. Proc. Natl. Acad. Sci. USA, 79:4672-4676. Herzlinger, D.A., T.G. Easton, and G.K. Ojakian (1982) The MDCK epithelial cell line expresses a cell surface antigen of the kidney distal tubule. J. Cell Biol., 93:269-277. Humphreys, W.J., B.O. Spurlock, and J.S. Johnson (1974)Critical point drying of ethanol-infiltrated cryofractured biological specimens for scanning electron microscopy. In: Scanning Electron Microscopy. Johari, 0. ed. IIT Research Institute, Chicago, pp. 275-282. Leighton, J. (1981) Anatomic evidence of transport function by adenocarcinomas and a suggested role in transport in the spread of cancer. Ann. N.Y. Acad. Sci., 372:455-464. Leighton, J., 2. Brada, L.W. Estes, and G. Justh (1969) Secretory activity and oncogenicity of a cell line (MDCK)derived from canine kidney. Science, 163:472-473. Lever, J.E. (1979a) Inducers of mammalian cell differentiation stimulate dome formation in a differentiated kidney epithelial cell line (MDCK). Proc. Natl. Acad. Sci., USA 76:1323-1327. Lever, J.E. (1979b)Cyclic AMP and inducers of mammalian cell differentiation stimulate dome formation in mammary and renal epithelial cell cultures. In: Hormones and Cell Culture. Cold Spring Harbor Conference on Cell Proliferation. eds. G. Sat0 and R. Ross, Cold Spring Harbor Press, New York, NY, pp. 727-738. Lewis, M.G., and A.A. Spector (1981)Differences in types of prostaglandins produced by two MDCK canine kidney cell sublines. Prostaglandins, 21: 1025-1032. .Madin, S.H., and N.B. Darby (1975) As catalogued (1958) in American Type Culture Collection Catalogue of Strains, 2:47. Mauchamp, J., A. Margotat, M. Chambard, B. Charrier, L. Remy, and M. Michel-Bechet (1979) Polarity of three-dimensional structures derived from isolated hog thyroid cells in primary culture. Cell Tissue Res., 204:417-430. McAteer, J.A., and T.J. Cavanagh (1982) Medium hydrated collagen gel as an explant support in organ culture. J. Tissue Cult. Methods, 7: 117-122. 239 McAteer, J.A., and T.J. Cavanagh (1983) Nylon screen filtration devices for crude separation of dissociated cells for primary culture. J. Tissue Cult. Methods, 8:45-47. McAteer, J.A., A.P. Evan, and W.H.J. Douglas (1983) A method for processing free-floating cultured cells for scanning electron microscopy. J. Tiss. Cult. Meth., 7:85-88. Misfeldt, D.S., S.T. Hamamoto, and D.R. Pitelka (1976)Transepithelial transport in cell culture. Proc. Natl. Acad. Sci., USA 73:1212-1216. Rabito, C.A., R. Tchao, J. Valentich, and J. Leighton (1980) Effect of cell-substratum interaction on hemicyst formation by MDCK cells. In Vitro, 16:461-468. Richardson, J.C., V. Scalera, and N.L. Simmons (1981)Identification of two strains of MDCK cells which resemble separate nephron tubule segments. Biochem. Biophys. Acta, 673~26-36. Rindler, M.J., L.M. Chuman, L. Shafter, and M.H. Saier (1979)Retention of differentiated properties in a n established dog kidney epithelial cell line (MDCK). J. Cell Biol., 81-635-647, Taub, M., and M.H. Saier (1979) An established but differentiated kidney epithelial cell line (MDCK), In: Methods in Enzymology, Vol. LVIII. Cell Culture. W.B. Jakoby and I.H. Pastan, eds. Academic Press, New York, pp. 552-560. Taub, M., and G.H. Sat0 (1979) Growth of kidney epithelial cells in hormone-supplemented serum-free medium. J. Supramol. Struct., 11:207-2 16. . Valentich, J.D. (1981) Morphological similarities between the dog kidney cell line MDCK and the mammalian cortical collecting tubule. Ann. N.Y. Acad. Sci., 372.372-384. Valentich, J.D., R. Tchao, and J. Leighton (1979) Hemicyst formation stimulated by cyclic AMP in dog kidney cell line MDCK. J. Cell Physiol., 100:291-304. Yang, J., J. Richard, P. Bowman, R. Guzman, J. Enami, K. McCormick, S. Hamamoto, D. Pitelka, and S. Nandi (1979) Sustained growth and three-dimensional organization of primary mammary tumor epithelial cells embedded in collagen gels. Proc. Natl. Acad. Sci., USA, 77:2088-2092. Yang J., J. Richard, R. Guzman, W. Imagawa, and S. Nandi (1980) Sustained growth in primary culture of normal mammary epithelial cells embedded in collagen gels. Proc. Natl. Acad. Sci. 76:34013405.