- Email: [email protected]
Basement membrane components are potent promoters of rat intestinal epithelial cell differentiation in vitro
GASTROENTEROLOGY
1990;98:322-335
Basement Membrane Components Are Potent Promoters of Rat Intestinal Epithelial Cell Differentiation In Vitro URSULA HAHN, A. STALLMACH,
E. G. HAHN, and
E. 0. RIECKEN Division of Gastroenterology. of Berlin, Hindenburgdamm,
Department of Medicine, Berlin, West Germany
Basement membranes have been implicated in morphogenesis and cell differentiation. In this study, the effect of basement membrane components on intestinal epithelial cell maturation in a mesenchyme-free environment was investigated. Fetal rat small intestinal epithelial cells (from the 1&h-17th day of gestation) were exposed to basement membrane-derived proteins (laminin, collagen type IV, and a complex basement membrane-enriched extract from the Engelbreth-Holm-Swarm sarcoma) and other extracellular matrix proteins (collagen type I and fibronectin) coated onto Petri dishes. The cells attached readily only to fibronectin and basement membrane proteins. For 5 days the developing epithelial colonies were monitored in vitro, assessing morphological and functional parameters of cell maturation. Colonies grown on laminin and the basement membrane extract were larger and of greater cell density. An increase in alkaline phosphatase and lactase activity was observed after 3-4 days in these colonies which could be enhanced to yield 90%-100% positive cells by the addition of dexamethasone to the medium while no sucrase-isomaltase activity was elicited. Electron microscopy confirmed a high degree of cellular polarization illustrated by tight junctions and apical microvilli in epithelial cells grown on a basement membrane-like support. In contrast, none of the other proteins stimulated the cells to mature in vitro. The authors conclude that certain basement membrane components actively promote fetal intestinal epithelial cell differentiation.
oward the time of birth, the epithelial cells of the rodent small intestinal mucosa proceed through a rapid course of differentiation and diversification to acquire the necessary degree of maturity for postnatal food absorption. During this prenatal period, the
T
Steglitz Medical School, Free University
developmental changes within the intestinal epithelium (endoderm) require the close proximity of intestinal mesenchyme (1). Even in vitro, fetal rat small intestinal epithelial cells can also be induced to differentiate if they are cocultured with homologous mesenthyme (2,3). The molecular basis for epithelial differentiation through mesenchymal influences has not been determined. It is not clear whether the induction of epithelial differentiation depends on the contiguity of vital mesenchymal cells, i.e., cell-cell contacts, or on biosynthetic products of the mesenchymal cells, Thus, the “mesenchymal signal” has not been identified. In this context, the nature of the epithelial-mesenchymal interface should be reviewed. The epithelialmesenchymal interface within the intestinal mucosa consists of the subepithelial basement membrane and its anchoring zone facing the adjacent reticular connective tissue of the lamina propria (4,~). Over the past decade, intensive chemical and biological research of basement membranes in general has shown that this formation of the extracellular matrix is a crucial element of organogenesis, tissue remodeling, and wound repair (6-9). Major structural components of the adult and fetal intestinal basement membrane have been identified by immunofluorescence and immunoelectron microscopy, i.e., the glycoproteins laminin, fibronectin, collagen type IV, nidogen, and a proteoglycan containing heparan sulfate sidechains (16-12). These are well-characterized, highly conserved macromolecules ubiquitous in all basement membranes analyzed so far in humans and many
Abbreviations used in this paper: BME, basement membrane matrix extract; DMEM, Dulbecco’s modified Eagle’s medium; PBS, phosphate-buffered saline. 0 1990 by the American Gastroenterological Association 0016-5095/90/$3.00
PROMOTERS OF CELL DIFFERENTIATION
February 1990
other species (6,7). However, their relationship to the basal plasma membrane of epithelial cells is not fully understood. It is likely that the function of the subepithelial basement membrane of the intestinal mucosa also exceeds mere mechanical needs. In the adult mucosa, the epithelial cells mature and migrate along the crypt-villus axis while staying in permanent contact with the subepithelial basement membrane. Besides giving structural support to the epithelium and compartmentalizing the mucosa, constituents of the basement membrane could conceivably be involved in epithelial cytodiff erentiation. Few attempts have been made to identify the nature of these cell-matrix interactions. One of the problems encountered is the lack of appropriate assays to study the behavior of viable, reasonably differentiated cells in vitro. Whereas mature enterocytes freshly isolated from mammalian intestinal tissues are viable only for a short time in vitro, epithelial cells from cell lines are immortal but often transformed and do not adequately respond to stimulation. An interesting new approach has been outlined recently making use of fetal epithelial cells that are still sensitive to their local environment (2,3,13). In this study, we investigated the response of fetal epithelial cells to extracellular matrix proteins in an attempt to identify components actively supporting cellular polarization and differentiation. The effect of distinct basement membrane proteins and interstitial collagens on the development of primary colonies of fetal small intestinal epithelial cells was determined morphologically and biochemically in vitro.
Materials and Methods Preparation
of Endoderm
Fetal Wistar rats were obtained by hysterectomy precisely on the morning of day 14,15,16, or 17 of gestation. The appearance of the vaginal plug was designated day 0. The entire small intestine from the pylorus to the appendix (ca. 2 cm) was removed, freed from mesenteric vessels, and incubated in 0.03% collagenase (CLSP II, 156 U/mg; Worthington, Freehold, N.J.) in Dulbecco’s minimal essential medium (DMEM; Boehringer Mannheim. Mannheim, Federal Republic of Germany) for 90 min at 36°C. The following procedures were performed at 4°C. Under a dissecting microscope, the enzymatically loosened mesoderm was pulled off the tighter endodermal tube and discarded. Larger sections of the endodermal tube were cut into fragments 0.5 mm long. The endodermal cells were then washed extensively with 10% serum to remove collagenolytic activity and seeded onto plain or precoated plastic Petri dishes. An average of 10 dishes containing lo-20 colonies per 35-mm Petri dish was obtained per litter. The purity of the preparation (no mesenchyme contamination] was checked by phase contrast microscopy and, in the beginning, by immunostain-
323
ing with antivimentin antibodies. The preparations were LOO%mesenchyme-free (if not, they were discarded at 24 h, at which time mesenchymal cells had proliferated abundantly and were clearly’visible).
Culture Conditions All cell cultures were performed in DMEM with 4 g/L glucose. Routinely, 4 mmol glutamine, 50 pg/ml ascorbic acid and 10% fetal calf serum were added [reagents from Biochrome, West Berlin, Federal Republic of Germany]. Cell cultures were incubated at 36’C and gassed with 8% CO, in air. For some experiments, dexamethasone (Sigma, Taufkirchen, Federal Republic of Germany] was added to the routine cell culture medium at a final concentration of 100 nM.
Extracellular
Matrix Substrates
The substrates used to precoat the culture dishes included murine collagen type I, human collagen type IV from placenta [prepared by D. Schuppan and E. G. Hahn, [14,15J), mouse collagen type IV (Bethesda Research Laboratories, Gaithersburg, Md.), mouse laminin, and human plasma fibronectin (Collaborative Research Inc., Waltham, Mass.]. Bovine serum albumin (Sigma, Taufkirchen, Federal Republic of Germany] was used as a control protein. All proteins were diluted to 10 or 20 pg/ml in phosphate-buffered saline [PBS], and 1 ml of the dilution was added to each 35-mm dish (Falcon Plastics, Cockeysville, Md.) and incubated for 90 min at 36°C. The solutions were then aspirated from the dishes. The efficiency of coating had been established previously for laminin by adding trace amounts of isotopelabeled laminin to various protein dilutions and calculating the recovery rate after a 90-min incubation period. At l-2 Mug protein per cm2 in 0.1 ml medium, saturation of binding of the protein to the surface of the Petri dish was observed. Ten to 20 endodermal fragments were added immediately with 0.8 ml medium per 35-mm Petri dish. Using albumin, fibronectin, or the basement membrane matrix extract (BME) substrate at 20 pg/ml, their presence on the Petri dish surface after the coating procedure was determined by staining the dish with 1% eosin for 5 min, which stained the surface area slightly pink while saline-treated dishes did not pick up any stain at all. For preparation of the BME substrate, C 57 black mice of both sexes each received 2 subcutaneous injections of 1 million Engelbreth-Holm-Swarm (EHS) sarcoma cells (16) within 1 wk. After 3-7 wk, the tumors were harvested, minced into small pieces, and immediately placed into ice-cold 4 M NaCl, 0.2 M Tris-HCl buffer, pH 7.2, and washed with a lo-fold volume for 1 h on ice. The EHS sarcoma matrix was extracted for 2 hand then for another 16 h with 2 M guanidin, 0.5 M NaCl, 0.2 M Tris-HCl buffer, pH 7.4. The second extract was centrifuged for 30 min at 15,000 rpm and the supernatant was extensively dialyzed, first against 0.5 M NaCl, 0.2 M Tris, pH 7.4, then against PBS at 4”C, and finally against DMEM for 24 h at 4%. This preparation was called BME and used for coating. By sodium dodecyl sulfate polyacrylamide gel electrophoresis
324 HAHN ET AL.
GASTROENTEROLOGY
and determination of the total protein concentration (17), the preparation contained laminin, collagen type IV, and nilO:l:l, as well as small dogen at a ratio of approximately amounts of fibronectin. The preparation was used at a protein concentration of 20 pg/ml.
Vol. 98, No. 2
Statistics Statistical analysis was performed Whitney U test for unmatched pairs.
using the Mann-
Results Analysis
of Endodermal
Attachment
and
Proliferation The attachment rate of the endodermal fragments was calculated after 24 h as the percentage of attached and spread endodermal colonies vs. the total number of endoderma1 fragments explanted. Endodermal proliferation was defined as the increase in size of the endodermal colonies over time. The total area covered by the explants was photographed every 24 h and calculated planimetrically from the films. Phase contrast microscopy and photography was performed on an inverted microscope [Zeiss IM [Zeiss, Oberkochen, Federal Republic of Germany] carrying 3.2 x , IOX, and 20x objectives, light field and phase contrast] using HP5 (Ilford Ltd., Essex, England] and Ektachrome 50 (Kodak, Hemel Hempstead, England] films.
Brush-Border
Enzyme Activity
The brush-border enzymes alkaline phosphatase, sucrase isomaltase, and lactase were determined in unfixed fetal intestinal epithelial cell colonies histochemically according to the method of Gutschmidt et al. (18,19). As controls, unfixed cryostat sections of freshly frozen newborn and adult rat intestines in which the epithelial brush borders reacted exclusively were routinely developed in parallel. The incubation conditions and incubation time were held constant for all procedures including the controls (15 min at 36°C). At least 25 epithelial colonies were stained histochemically for each enzyme on each substrate after 24 h and 4 days in vitro. In each experimental group, the colonies with the most intensive color reaction were evaluated using the grading index. Within a colony, all stained cells were considered positive and unstained (white] cells were registered as negative. At least 150 cells per colony center were counted. A grading index was chosen as follows: -, ~5% stained cells; +, 5%-10% stained cells; + +, lo%-50% stained cells; + + + , 50% -90% stained cells; + + + +, 90% 100% stained cells. Two investigators (U.H., AS.) determined the grading index independently without prior knowledge of the specific experimental conditions.
Attachment of Small Intestinal Endodermal Explants to Different Substrates In a first set of experiments, the entire length of small intestinal epithelium from the pylorus to the cecum was harvested on the 15th day of gestation, cut into 05mm fragments (cell clusters), and explanted in primary culture. [A total of 1281 individual endoderma1 fragments were explanted onto different surfaces in 7 consecutive experiments using 10 litters of fetal rats.) After 24 h, the wells were rinsed and the attached and spread colonies in each well were counted. Fetal epithelial clusters that had not attached by 24 h were removed because they continued to float and quickly degenerated. The endodermal cell clusters explanted on plain plastic surfaces or on Petri dishes coated with albumin or matrix proteins showed significantly different attachment rates (Figure 1). On plastic and on albumin-treated dishes, only 28% and 27%, respectively, of the endodermal fragments derived from 15-day-old fetal rat small intestine adhered to these substrates within 24 h. Endoderm
, 1001
--
I
:
Transmission electron microscopy was performed on transverse sections of endodermal cell colonies grown on plain Thermonox or coated Thermonox slides (Miles Lab., Munich, Federal Republic of Germany] prepared identically as described above for Petri dishes. These cultures were fixed in 1% glutaraldehyde containing 1% tannin in 0.1 M PBS, postfixed in 1% osmium tetraoxide and embedded in Mikropal [Ferak, West Berlin, Federal Republic of Germany) by the method of Merker and Barrach (20).
. . ..A.. .I.-
BSA
CI
BME
LAM
**** L
:
PLAS
Electron Microscopy
**'
CIV
I I
FIB
Figure 1. Attachment rates of fetal small intestinal epithelial cells on matrix protein. Rate of attachment of fetal intestinal epithelial cells resected on the 16th day of gestation and seeded onto plastic Petri dishes coated with various extracellular matrix proteins: plain plastic (PLAS),bovine serum albumin (BSA), collagen type I (CI), fibronectin (FIB),basement membrane tumor extract (BME), laminin (LAM), and collagen type IV (CIV). The attachment rate was derived from the number of attached and spread vs. floating cell clusters after 24 h of incubation at 36% by phase contrast microscopy. Bars indicate the median attachment rates (mean + 1 SD); numbers at the tops of the bars indicate the total number of explants in each group.
February 1990
explanted onto these substrates thus yielded only a few colonies. The greater variability in the attachment rates to plastic and albumin-coated dishes compared with the other substrates (SD > 50% on plastic in a total of n = 323) can be attributed to the low affinity of the endodermal colonies to plastic and albumin. Attachment was comparatively slow and weak on albumin and plastic. Even spread colonies were sometimes rinsed off before counting. A substantially higher yield of endodermal colonies was achieved by coating the Petri dishes with collagen type I, which resulted in an average attachment rate of 51% (p < 0.05) (Figure 1). In contrast, the plating efficiency of endodermal fragments on laminin, collagen type IV, and BME was much higher; 72%-75% of all fragments attached to these substrates and had spread to form single-cell monolayers within 24h (Figure 1). The attachment rate was highest on fibronectin (85%),which had a significantly higher rate than collagen type IV (p < 0.05). laminin, and BME (p < 0.01). The medium used in these experiments contained 10% fetal calf serum, but in a recently completed study using fetal intestinal epithelium in serum-free conditions, we have observed similar attachment rates (unpublished results). The increase in attachment efficiency was paralleled by an apparently stronger affinity to laminin, collagen type IV, and BME. Epithelial cells quickly spread on these substrates and could not be rinsed off with physiological buffers.
Affinity of Small Intestinal EpitheJiaJ Cells to Various Substrates as a Function of Developmental Stage It became obvious during our investigations that the developmental status of the epithelial cells played a major role in determining their affinity to the substrates. In an extra set of experiments, therefore, the affinity of epithelial cells derived from the small intestine on the 14th, 15th, 16th, and 17th days of gestation to a plastic surface was compared with their affinity to a BME-coated surface (Figure 2). Epithelial cells from day 14 of gestation attached by an average of 42% to plastic and 82% to the BME substrate by 24 h. Attachment rates were similar with epithelium derived from the 15th day of gestation (77% on BME), although a greater variation in the percentage of firmly attached cells on plastic was noted. Fetal intestinal epithelial cells harvested on day 16 displayed an attachment rate decreased to 67% on BME while an average of 28% of epithelial cell clusters bound to a plastic surface. By day 17 of gestation, only 6% of all explanted epithelial cell clusters attached to plastic within 24 h. Again, the BME substrate proved to be a significantly superior substrate although the attachment rate decreased to 44L70 of explanted clusters on
PROMOTERS
15.
OF CELL DIFFERENTIATION
16.
.17. day
325
of gestation
Figure 2. Aflinity as a function of gestational age. Attachment rates were calculated as in Figure 1 for fetal small intestinal epithelial cells resected from the small intestines of rat embryos on the 14th, ISth, 16th, and 17th days of gestation. Fetal epithelial cells were plated on BMEcoated dishes (open bars) and on albumin-coated dishes (solid bars). Bars indicate median attachment rates +1 SD. With increasing developmental age, the ability of the fetal small intestinal epithelial cells to attach to protein substrates is generally diminished, but the afenity to the BME substrate (44%) is 7 times greater than the afenity to albumin on day 17 of gestation (6%).
the 17th day of gestation (Figure 2). The ratio of attachment rates thus increases from 2:l (day 14) to 7:1 [day 17). These results also imply that the developmental gradient along the longitudinal axis of the small intestine (with the proximal intestine leading by approximately 24-48 h in terms of brush-border enzyme expression) should influence epithelial cell attachment rates. In another set of experiments, duodenal explants were indeed observed to have an equally reduced attachment rate to all substrates compared with distal explants (data not included). In the experiments described here, a random distribution of proximal and distal fragments was achieved by mixing all fragments before explantation. In summary, the development of the fetal rat small intestine between the 14th and 17th days of gestation is accompanied by a loss of plating efficiency of isolated epithelial cells to all substrates. However, the greater affinity of these cells to basement membrane proteins compared with control surfaces is enhanced at later stages of development. Proliferation and Survival of Endodermal Explants on Various Substrates To evaluate the proliferation rate of the endodermal cells as an increase in cell number over time, establishing the number of cells correctly at any time proved difficult. The total number of cells in an individual colony (estimated from composite photographs) was too small to yield appropriate DNA levels and too great to be counted morphometrically. There-
326 HAHN ET AL.
fore, we took another approach. By monitoring the explants closely by phase contrast microscopy, we had observed that all the explants expanded in the same manner. After the initial attachment and spreading phase, the colonies were composed of sessile, dividing, and migratory cells. Migratory cells were obvious at the periphery, displaying pseudopodiae and a ruffled cytoplasmic edge, while mitotic figures were found throughout the entire colony. Hence, the increase in the area covered by individual explants over the first 3 days in vitro represented the combined effect of cell migration and cell division. The incorporation of bromodeoxyuridine, a thymidine-analogue base, as a parameter of DNA synthesis correlated well with the increase in colony size (A. Stallmach et al., manuscript in preparation). Although the individual cell shape also may have determined the size of a colony, this factor was not taken into account because the cells within the largest colonies (on BME and laminin) had the smallest cell diameter. Therefore, cell shape did not additionally increase colony size but rather favored an underestimation of cell counts on the BME substrate. In conclusion, we considered the expansion of the colonies as a rough but proportional indicator of cell proliferation. Individual colonies were photographed precisely every 24 h and evaluated planimetrically (Figure 3). On all substrates the endodermal colonies reached their maximal expansion by the 3rd day in vitro (Figure 3). However, the actual colony size differed considerably. While endodermal colonies averaged 0.97 + 0.35 mm2 on albumin and 0.94 + 0.41 mm2 on collagen type IV, they expanded to 1.24+ 0.52mm2 on laminin and 2.34 + 0.73mm2 on BME. These differences were significant between BME and all other substrates (p < 0.01) and between laminin and collagen type IV or albumin. Interestingly, the expansion of endodermal colonies on fibronectin was very low and stopped after 48 h at values even below the values for the indifferent control protein albumin. After the 3rd day, more or less rapid deterioration took place (Figure 3). Cell loss from the substrate was striking on fibronectin, collagen type IV, and plastic. On laminin, morphologically intact sections of the monolayers remained in place until after the 4th day in vitro, after which all cells detached. On BME, however, colony centers stayed viable until the 6th day in vitro. The sharp decline in colony size occurring after the 3rd day on BME was caused by detachment of the most peripheral cells only. In summary, BME was the most favorable substrate for both proliferation and survival of mesenchymefree fetal intestinal epithelium in vitro leading to a 2.4-fold increase in colony size and to the survival of endodermal cells for up to 6 days in vitro. Laminin had a similar, albeit less pronounced, effect. However,
GASTROENTEROLOGYVol. 98,No.z
n q
BME LAM CIV FIB
w 0
i-
1
2
3 days in vitro
5
Figure 8. Growth of intestinal epithelial colonies. The area of individual fetal intestinal epithelial colonies cultured for 8 days on various extracellular matrix proteins--basement membrane extract (BME), laminin (LAM), collagen type IV (CIV), collagen type I (0, fihronectin (FIB), and bovine serum albumin (BSA)-was calculated planimetrically from phase contrast photographs. A minimum of 20 single colonies per substrate and per day were evaluated. Mean values are expressed in square millimeters +I SD.
collagen type IV and fibronectin did not promote colony growth or survival although the high initial attachment rates suggest specific affinity for both molecules.
Light Microscopic and Ultrastructural of Endodermal Explants
Aspect
Morphologically, the initial spreading phase of the epithelial explants was similar on all substrates. Small, circular monolayers developed, and at 24 h all explants were about the same size. They contained epithelial cells resembling those of intestinal epithelial cell lines (21). At 46 h, however, the size of the colonies and the shape of the cells within were different among the groups (Figure 4).Colonies grown on albumin (Figure 4A) and collagen type IV (Figure 4B) were small, and their cell population was homogenous. In contrast, fetal epithelial cell colonies cultured on laminin were larger with vigorously proliferating peripheral cells and a central zone of smaller, densely packed cells (Figure 4C). At the same time, fetal epithelial colonies grown on BME had the largest diameters; their central zone consisted of tightly arranged, narrow cells with intercellular gaps. Often a small area near the center
PROMOTERS OF CELL DIFFERENTIATION
February 1990
327
Figure 4. Light microscopic aspect of fetal epithelial colonies at 48 h grown on albumin, collagen type IV, laminin, and BME. Original magnification x 100. A. Epithelial colony grown on albumin. B. Epithelial colony grown on collagen type IV. In both types of outgrowth the cells are essentially homogenous. C. Center formation in a colony on laminin. Notice also the larger size. D. Densely populated center of a large, still-expanding visible by its lack of focus.
colony on the BME substrate. In the very center, a budding dome is emerging which is
was elevated above the plane of the Petri dish indicative of a budding dome [Figure 40). After 72 h, however, striking variations in the morphology of the outgrowths became apparent (Figure 5). Cultivated on plastic, albumin, collagen types I and IV, and fibronectin, fetal intestinal epithelial cells derived from 15-day-old fetal rats retained a flat, well-spread morphology. Most cells had a large cytoplasm, a well-recognizable nucleus containing two nuclei, and no granules or inclusions as shown in a colony on collagen IV in Figure 5A. There was no conspicuous difference between the center and the periphery of the colonies. In contrast, the central region of most colonies cultured on laminin and BME expanded toward the periphery. It became so tightly packed that individual cells were hardly distinguish-
able. A typical example is given in Figure 5B. On BME, many cells at the colony border contained clear cytoplasmic inclusions (Figure 5D). These inclusions were not stained by the lipid-detecting Sudan Black B stain or by the periodic-acid Schiff base, which stains complex carbohydrates (data not shown]. Therefore, there was no evidence by phase contrast microscopy of epithelial cytodifferentiation to goblet or endocrine cells. A striking feature observed exclusively in epithelial cultures on BME was domes. Domes consist of a group of cells lifted from the surface of the dish by the pressure building up from fluid secreted by the cells at their basal membranes. Tight intercellular junctions are required. In 40% of the colonies, up to 5 individual domes per colony were formed after 48-60 h in vitro.
328 HAHN ET AL.
GASTROENTEROLOGY
Vol. 98, No. 2
Figure 5. Fetal epithelial cell cultures on different matrix protein substrates after 72 h in vitro. A-D. Original magnification x 200. A. Center of an epithelial colony grown on collagen type IV, displaying uniformly flat, spread cells of equal diameters. B. Epithelial-cell colony grown on laminin. The center of the colony is shown in the left upper corner and consists of tightly packed cells. The epithelial cells surrounding it appear in a typical cobblestone pattern characteristic of a sheet of polarized epithelial cells. C. Epithelial cell colony grown on the BME substrate. The epithelial cell sheet is focused so that the 2 domes of elevated epithelial cells that have arisen in the middle of the colony appear out of focus. Domes are formed when the epithelial cells are joined by tight junctions and fluid secreted at their basal plasma membrane causes the cells to become elevated above the level of the protein substrate coating the Petri dish. D. At the periphery of a colony of fetal epithelial cells grown on the BME substrate, many migrating cells contain light, translucent inclusions. The cells at the periphery are spread out.
This suggests active, vectorial ion transport by the epithelial cells (Figures 5C and 6B). In fact, electron microscopy of transverse sections through the center portion of individual colonies grown on BME showed highly polarized epithelial cells with a regular brush border. The height of individual epithelial cells within domes was 8 times greater than that of the flat central cells seen on albumin or collagen types I and IV. The nucleus appeared in the lower half of the cells with an overlying golgi apparatus. Electron-dense granules indicating endocrine cells were not observed. Tight junctions and intermediate junctions were present (Figure 7B). Migratory cells at the very periphery of the colonies on BME appeared much flatter, and the brush border was irregular (Figure 6A). Similarly, fetal intestinal epithelial cells grown on albumin and colla-
gen type I retained their flat, unpolarized morphology, carried no microvilli or tight junctions, and therefore resembled the undifferentiated epithelial cells of intestinal epithelial cell lines in vitro [Figure 7A) (21). Ultrastructural analysis thus confirmed that endodermal cells grown on BME reached a high and organotypic level of differentiation within 3 days in vitro. In contrast, fetal intestinal epithelium cultured on indifferent substrates showed no such signs of morphological maturation. Activity of Brush-Border Endodermal Explants
Enzymes in
The epithelial colonies were studied histochemically for evidence of alkaline phosphatase, lactase,
February 1990
PROMOTERS OF CELL DIFFERENTIATION
329
and sucrase isomaltase. The percentage of stained vs. unstained cells was evaluated, and a grading index ranging from negative (-, ~5% cells stained] to strongly positive (+ + + +, >90% cells stained] was applied. For the experiments described in the following paragraph, we again used only colonies derived from the small intestinal epithelium of the 15th day of gestation which in situ contains no brush-border enzyme activity (221. Accordingly, frozen sections of the starting material did not produce a color reaction in any staining method. No brush-border enzyme activity was registered at all on albumin, fibronectin, or collagen type I (Figure 8A and B). On collagen type IV, only 5%-8% of all colonies were positive for alkaline phosphatase after 4 days. In the presence of 100 nmol dexamethasone, similar levels of alkaline phosphatase activity were observed in colonies grown on collagen type IV (Figure 8C and D). In contrast, epithelial colonies
Figure 6. Transverse electron microscopy of fetal epithelial cells grown on the BME substrate for 72 h. Bars = 1 pm. A. A peripheral cell at the edge of the colony is unpolarized and spread but displays an irregular brush border [original magnification x 9000). B. Cells at the foot of a dome are cuboidal and are in the process of separating from the substrate. Irregular protein deposits at the basal plasma membrane [arrows] probably represent adherent BME (original magnification x4500). C. Highly polarized epithelial cells in the center of a colony exhibit more microvilli. tight junctions, and a basally located cell nucleus. Arrows point to adherent extracellular material that is possibly BME-derived [original magnification x4500).
Figure 7. TEM of epithelial cells grown on collagen type I and BME. Original magnification x 14,660; bar = 1 pm. A. This epithelial cell of the center of a colony grown on collagen type I for 72 h shows no signs of differentiation. B. Brush border of 2 epithelial cells from the center of a colony grown on BME. The microvilli are well formed and regular. A tight junction (arrow] connects the two adjacent cells.
330 HAHN ET AL.
GASTROENTEROLOGY Vol. 98, No. 2
Figure 8. Brush-border enzyme activity (alkaline phosphatase) in epithelial cell colonies on different matrix protein substrates after 96 h in vitro. Light field optics or phase contrast; original magnification x 100. A and B. Fetal intestinal epithelial colony grown on albumin. There are no positive cells except some rounded-up colony [arrow]: light field (A] and phase contrast (B) views of the same specimen.
cells that are leaving the
C and D. Center of a colony grown on collagen type IV in the presence of dexamethasone. Throughout the colony, some cells are stained very faintly. A positive cell is marked by an arrow; light field (C) and phase contrast [D] images of same area. E and F. Fetal epithelial colonies cultured on laminin in normal medium (E) or in the presence of 100 nM dexamethasone already-strong staining pattern achieved by normal medium is further enhanced in the presence of dexamethasone. G. Colony grown on BME. More than 90% of cells carry enzyme activity throughout medium. However, randomly distributed nonreactive (white) cells remain. If. Fetal epithelial colony grown on BME in the presence of dexamethasone. cells.
the entire colony that was maintained
(F). The in normal
Abundant expression of the enzyme is seen in practically all
PROMOTERS
February 1990
grown on laminin steadily increased the expression of alkaline phosphatase until the 4th day in vitro (Figure 8E). At this point, 40%-50% of all cells within the colonies were stained intensely for alkaline phosphatase and rated grade + +. By the 4th day in vitro in normal medium, epithelial colonies on laminin also turned positive for lactase enzyme activity (grade +). Addition of dexamethasone to these cultures clearly increased the number of positive cells and the intensity of the reaction for both enzymes (Table 1). Fetal intestinal epithelial cells cultured on BME expressed the strongest activity of alkaline phosphatase. By the 4th day in vitro, 80%-90% of all endodermal cells were intensely stained for alkaline phosphatase, grade +++ (Figure 8G). Dexamethasone promoted alkaline phosphatase activity even further so that 90%-100% of all epithelial cells tested positive. The intensity of the color reaction was impressive (Figure 8H). In addition, lactase activity appeared in these cultures after 2 days in vitro and increased until up to 50% of all cells were stained the typical, bright turquoise color of the lactase reaction identical to the color product of adult intestinal brush borders. Taken together as shown in Table 1, only the explants cultured on laminin and BME expressed brush-border enzyme activities compatible with the degree of morphological maturation illustrated by electron microscopy. Fetal epithelium on plastic, albumin, fibronectin, and collagen types I and IV displayed minimal enzyme activity under these conditions, indicating low intrinsic self-differentiating properties of the endoderm. This seemed to be true for explants derived from both the duodenal and ileal endoderm. Because some preliminary experiments
Table
1.
Relative Activity of Brush-Border Enzyme Expression Substrates
OF CELL DIFFERENTIATION
331
separating proximal from ileal endoderm did not show a conspicuous acceleration of development of the proximal endoderm in vitro, we subsequently performed a separate study in which >2000 endoderma1 colonies were separated according to anatomical origin and the development of brush-border enzyme expression was investigated (U. Hahn, manuscript in preparation). Because epithelial cell differentiation of the proximal endoderm appears more complex than ileal endoderm in vitro, these results will be reported in detail in a separate study. Dexamethasone did not elicit any brush-border enzyme activity de novo in primarily negative colonies. This was true whether the medium was supplemented with hormone at the onset of the experiment or after 24-48 h. However, once brush-border enzyme expression had been initiated in colonies growing on laminin or BME, both alkaline phosphatase and lactase activity increased under the influence of dexamethasone (Table 11. Sucrase isomaltase remained undetectable even in the presence of dexamethasone.
Discussion This study demonstrates that certain extracellular matrix components can support cellular differentiation in fetal small intestinal epithelial cells in the absence of mesenchymal cells in vitro. Purified laminin, a glycoprotein found exclusively in basement membranes, and a complex, basement membrane protein-enriched extract (BME] acted as promoters. In contrast, another basement membrane-specific protein, collagen type IV, was much less effective. Fibronectin and collagen type I, both pertaining to the
in Small Intestinal
EpitheJiaJ Cells Cultured on Different
Substrate (days in vitro)
BME Enzyme Alkaline phosphatase NM DEX Lactase NM DEX Sucrase isomaltase NM DEX
1
Laminin 4
Collagen I, fibronectin, albumin, plastic
Collagen IV
1
4
1
4
1
4
+ ++
+++ ++++
+ +
++ +++
_ _
+ +
_
_ _
_ _
++ +++
_ _
+ ++
_ _
_
_
_ _
_
_ _
_ _
_ _
_ _
_ _
_ _
_
Brush-border enzyme activity was demonstrated histochemically. In the center of the endodermal colonies (comprising approximately 200 cells). all positive and negative cells were counted and the relative enzyme activity was graded according to the following index: -, ~5%; +, 5%-9%; + +, lo%-49%; + + +, 50%-89%; + + + +. 90%-100% positive, stained cells. NM, normal medium: DEX, medium containing 100 nmol dexamethasone.
332
HAHN ET AL.
GASTROENTEROLOGY
Vol. 98, No. 2
colonies was observed in response to the BME subreticular extracellular matrix rather than to the basement membrane proper, did not evoke a cellular strate and laminin. response substantially different from the control proGlucocorticoids have also been shown to enhance tein albumin. epithelial viability and brush-border enzyme activity Previous studies confined to testing the affinity of in vitro in many intestinal organ culture studies (28single intestinal epithelial cells from different species 31). The same effect was seen in epithelial-mesenchyto extracellular matrix proteins had produced diverse This has led to the assumption ma1 cocultures (1,2,13). results (22-25). Our own experience with a different that the action of the hormone is transmitted via fetal organ culture model had suggested that the mesenchymal cells (13). In contradistinction, the presence of viable mesenchymal cells was obligatory present study suggests that the epithelial cells do not for the formation of an authentic basement membrane need any mesenchymal support to respond to dexain vitro. In those studies, the intestinal epithelial cells methasone. In the complete absence of mesenchymal had not been separated from a core of underlying cells, the critical minimal degree of epithelial polarizamesenchymal cells (3). Mesenchymal and epithelial tion and intracellular organization required for horcells were shown to cooperate in the formation of a mone responsiveness was apparently induced by the basement membrane at their interface, and a concomBME complex and laminin alone. Dexamethasone itant increase in epithelial differentiation was obthus acted as a permissive promoter for alkaline served. The previous organ culture-type model, howphosphatase and lactase. Laminin and dexamethaever, was not suitable for the investigation of the effect sone complement each other with respect to brushof purified extracellular matrix proteins on epithelial border enzyme expression. Both agents have been differentiation because of the presence of mesenchyshown to stabilize cellular mRNA pools in other ma1 cells producing these proteins (and perhaps other systems (27,32). More studies are needed to determine factors as well). Therefore, for the study presented the molecular nature of such events in intestinal cells. here, endoderm and mesoderm were completely sepaIt specifically remains to be seen whether the increase rated by proteolytic and mechanical treatment of the in enzyme activity is caused by the induction or fetal gut. The reaction of purified epithelial cells to stabilization of enzyme synthesis on the transcripdifferent matrix proteins could then be monitored in tional level. Surprisingly, however, no histochemical vitro. evidence of sucrase isomaltase activity was detected We also raised the question of whether basement in any colony although this enzyme is expressed in membrane proteins could elicit cellular responses endodermal-mesenchymal cocultures of the same gesfrom intestinal epithelial cells beyond attachment and tational age after 4-6 days in vitro (2,13). Dexamethaspreading on coated surfaces. Evidence for the stimusone, which is particularly well known for its ability to latory effect of basement membrane glycoproteins on induce precocious sucrase isomaltase activity in small the differentiation of other target epithelial cells in intestinal epithelial cells, both in vivo and in vitro vitro has been reported (6,7). For example, cultured rat (28,301, failed to stimulate enzyme expression here. Sertoli cells respond to a basement membrane proteinThis may be due to a separate mode of intracellular rich gel derived from the same source, the Engelbrethregulation for sucrase isomaltase indicated by its 48-h Helm-Swarm tumor, by polarization and the acquisidelay in appearance in situ after that of alkaline tion of a cell shape virtually identical to mature Sertoli phosphatase and lactase (28). cells in vivo (26). Mammary epithelial cells display a The ultrastructural morphology of the epithelial cuboidal cell shape and maintain a high level of cells, as assessed by transmission electron microscopy, &casein messenger RNA (mRNA) level when culserved as another parameter of cellular maturation. tured on laminin or the EHS extract (27). Laminin and Intercellular junctions, cellular height, and the presthe EHS matrix seem to stabilize the mRNA pool for ence of microvilli were determined as a function of casein proteins in mammary epithelial cells, thereby the duration of the culture and the respective matrix promoting the tissue-specific cellular phenotype. This component. The BME substrate induced an s-fold particularly intriguing observation indicates the presincrease in cellular height in the colony center, reguence of intracellular pathways activated by specific lar microvilli of the apical plasma membrane, and cell-matrix interactions. We determined the exprestight junctions between adjacent cells. The degree of sion of brush-border enzyme activity as one parameter maturation induced by the BME substrate is comparaof the intestinal epithelial phenotype. Increasing activble to that induced by underlying mesenchymal cells ity of brush-border enzymes is generally considered a as shown in the earlier study (12). However, slightly good indication of enterocyte maturation (18,19,28). more elongated microvilli and occasional goblet cells Indeed, an impressive induction of alkaline phoswere observed in the coculture system (12). phatase and lactase activity in fetal epithelial cell
February 1990
The importance of junctional complexes and the cytoskeleton in determining cell polarity is well established (33-35). However, it is not known whether the formation of tight junctions and the composition of the basolateral membrane are related as in other cells (36). We observed that the plating efficiency of fetal epithelial cells from day 17 of gestation on basement membrane proteins was markedly lower than that of fetal cells from days 14-15, although their affinity to the BME substrate had increased relative to controls (Figure 2). This may indicate that the composition of the basal-cell membrane undergoes changes equally as important as those in the apical membrane during this period. In vivo, the brush border is being loaded with enzyme protein which is illustrated by a substantial increase in enzyme activity on the 19th day of gestation in vivo. Therefore, on the 19th day of gestation, an increase in brush-border enzyme expression and the acquisition of mature junctional complexes occur contemporaneously at the time when all epithelial cells have come into contact with the basement membrane (37). This corresponds precisely to the time in this model (explanation on the 15th day of gestation plus 4 days in vitro) when epithelial cells grown on BME have formed tight junctions and alkaline phosphatase is almost uniformly expressed. Thus, contact of the basal cell membrane with the basement membrane may enable the epithelial cell to organize its other membrane domains. When tight intracellular junctions are formed between epithelial cells in vitro, a coherent epithelium that presents its “luminal” side to the culture medium is established. The epithelial cells may then begin to actively absorb and secrete ions and water (38). It is consequently not surprising to detect domes (or hemicysts) in such epithelial monolayers. A dome is a small, hollow mound consisting of a group of cells lifted off the surface of the coated Petri dish through the pressure of fluid secreted by the cells at their basal plasma membrane. In the fetal epithelial colonies, domes were found exclusively in colonies grown on the BME substrate. Therefore, the tight junctions demonstrated morphologically were also found to function. High affinity of intestinal epithelial cells to laminin, fibronectin, and collagen type IV, in combination with the distinct cellular responses to these proteins, suggests the presence of specific transmembrane-binding proteins or receptors on the cell membrane. There is indeed evidence of at least 3 different types of matrix protein-binding proteins on the basal plasma membrane of enterocytes in vertebrates. A laminin receptor-like molecule, a collagen type IV-binding protein, and a well-defined fibronectin-binding protein (the so-called cell substrate attachment (CSAT) antigen in
PROMOTERS OF CELL DIFFERENTIATION
333
avian tissues or integrin complex) have been identified immunohistologically in the intestinal mucosa of different species (39-43). Integrins are known to span the plasma membrane and to connect with components of the cytoskeleton. A link is thereby established between intracellular and extracellular structural networks (44,45). On both laminin and BME, differentiation of fetal intestinal epithelial cells was preceded by proliferation. Colonies on BME tripled in size compared with fibronectin (Figures 3 and 4). It is not clear how the BME extract and laminin may promote cell proliferation (6,27,46). However, in the physiological context of wound healing, it seems desirable for the authentic basement membrane to exert such an effect. Erosions in the gastrointestinal tract in which the subepithelial basement membrane remains intact reepithelialize faster than ulcerous defects. As reported earlier, primary fetal epithelial cells and transformed epithelial cells are capable of basement membrane protein production (3,47). However, synthesis of laminin and collagen type IV have also been reported in the subepithelial mesenchyme (3,48,49). It is unresolved to what extent epithelial cells and fibroblasts participate in basement membrane formation in vivo. We are currently comparing matrix protein synthesis in mesenchyme-free epithelial cell colonies with that in diverse cocultures (unpublished results). Observations to date indicate that the very low level of endogenous matrix protein synthesis observed in freshly explanted epithelial cells is insufficient to induce differentiation and is altered by exposure of the cells to different substrates. In conclusion, we have presented evidence of the morphological and functional differentiation of fetal small intestinal epithelial cells through the action of a complex, basement membrane protein-enriched extract and laminin in vitro. We suggest that the basement membrane also plays a crucial role in the induction of epithelial differentiation during morphogenesis in utero when stratified fetal endodermal cells are organized into a monolayer of polarized enterocytes. In the adult mucosa, it is conceivable that both subepithelial myofibroblasts and epithelial cells participate in basement membrane protein production, the assembly of which occurs at the epithelial basal membrane. Whether some of these components are intrinsically linked to the process of epithelial cell differentiation along the crypt-villus axis in the adult human mucosa remains to be determined. References 1. Haffen
K, Kedinger M. Simon-Assmann P. Mesenchymedependent differentiation of epithelial progenitor cells in the gut. J Pediatr Gastroenterol Nutr 1967:6:14-23.
334 HAHN ET AL.
2. Stallmach A, Hahn LJ, Merker H-J. Hahn EG, Riecken E-O. Differentiation of rat intestinal epithelial cells is induced by organotypic mesenchymal cells in vitro. Gut 1989;30:959-70. 3. Hahn U, Schuppan D, Hahn EG, Merker H-J, Riecken E-O. Intestinal cells produce basement membrane proteins in vitro. Gut 1987;28(Suppl1):143-151. 4. Takahashi-Iwanaga H, Fujita T. Lamina propria of intestinal mucosa as a typical reticular tissue. Cell Tissue Res 1985;242:5766. 5. Komuro T. Fenestrations of the basal lamina of intestinal villi of the rat. Cell Tissue Res 1985;239:183-8. 6. Martin GR, Timpl R. Laminin and other basement membrane components. Annu Rev Cell Bioll987;3:57-85. 7. Timpl R, Dziadek M. Structure, development and molecular pathology of basement membranes. Int Rev Cell Bioll986;2:2749. 8. Martinez-Hernandez A, Amenta R. The basement membrane in pathology. Lab Invest 1983;48:656-77. 9. Trelstad R. Role of extracellular matrix in development. New York: Liss, 1984. 10. Laurie GW, Leblond CP, Martin GR. Localization of type IV collagen, laminin, heparansulfate proteoglycan and fibronectin to the basal lamina of basement membranes. J Cell Biol 1982;95:340-4. 11. Hahn LJ, Schuppan D, Hahn EG. Riecken E-O. Immunhistologisthe Charakterisierung interstitieller anitigener Substanzen in der Darmwand. Fortschr Gastroenterol Endosk 1985;14:198-201. 12. Martinez-Hernandez A, Chung AE. The ultrastructural localization of two basement membrane components. J Histochem Cytochem 1984;32:289-98. 13. Kedinger M, Simon-Assmann P, Alexandre E, Haffen K. Importance of a fibroblastic support for in vitro differentiation of intestinal endodermal cells and their response to glucocorticoids. Cell Differ 1987;20:171-82. 14. Schuppan D, Besser M, Schwarting R, Hahn EG. Radioimmunoassay for the carboxyterminal crosslinking domain of type IV procollagen in body fluids. J Clin Invest 1986;78:241-8. 15. Hahn EG, Wick G, Pencev D, Timpl R. Distribution of basement membrane proteins in normal and fibrotic human liver-collagen type IV, laminin and fibronectin. Gut 1980;21:63-71. 16. Orkin RW, Gehron P, McGoodwin EB, Martin GR, Valentine T, Swarm R. A murine tumor producing a matrix of basement membrane. J Exp Med 1977;145:204-20. 17. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265-75. 18. Gutschmidt S, Lange U, Riecken E-O. In situ measurements of protein contents in the brush border region along rat jejunal villi and their correlations with four enzyme activities. Histochemistry 1980;72:467-9. 19. Gutschmidt S, Kaul W, Riecken E-O. A quantitative histochemical technique for the characterization of alpha-glucosidases in the brush border membrane of rat jejunum. Histochemistry 1979:63:81-101. 20. Merker HJ, Barrach HJ. The morphology of basement membrane formation. Eur J Cell Bio11981;26:111-20. 21. Quaroni A, Wands J. Trelstad RL, Isselbacher KJ. Epithelioid cell cultures from rat small intestine. J Cell Biol 1979;80:248-65. 22. Burrill PH. Bernardini I, Kleinman H, Kretchmer N. Effect of serum, fibronectin and laminin on adhesion of rabbit intestinal epithelial cells in culture. J Supramol Struct Cell Biochem 198X16:385-92. 23. Rattner DW, Ito S. Rutten MJ. Silen W. A rapid method for culturing guinea pig gastric mucous cell monolayers. In Vitro 1985;21:453-62.
GASTROENTEROLOGY
Vol. 98, No. 2
24. Hahn U, Cho A, Schuppan D, Hahn EG, Merker HJ, Riecken EO. Intestinal epithelial cells preferentially attach to a biomatrix derived from human intestinal mucosa. Gut 1987;28(Suppl 1):153-8. 25. Hahn U. The role of extracellular matrix proteins in small intestinal cell cultures. Stand J Gastroenterol 1988;23(Suppl 151):70-8. 26. Hadley MA, Byers SW, Suarez-Quian CA, Kleinman HK, Dym M. Extracellular matrix regulates Sertoli cell differentiation, testicular cord formation and germ cell development in vitro. J Cell Bio11985;101:1511-22. 27. Li ML, Aggeler J, Farson DA, Hatier C, Hassell J, Bissell M. Influence of a reconstituted basement membrane and its components on casein gene expression. Proc Nat1 Acad Sci USA 1987;84:136-40. 28. Simon-Assmann P, Kedinger M, Grenier JF, Haffen K. Control of brush border enzymes by dexamethasone on the fetal rat intestine cultured in vitro. J Pediatr Gastroenterol Nutr 1982;l: 257-65. 29. Moog F. The differentiation and dedifferentiation of the intestinal epithelium and its brush border enzymes. In: Development of mammalian absorptive processes. Ciba Foundation series 70. Amsterdam: Elsevier, 1979:31-50. 30. Quaroni A. Crypt cell development in newborn rat small intestine. J Cell Bio11985;100:1601-10. 31. Beaulieu JF, Calvert R. Hormonal regulation of epithelial cell proliferation in the fetal mouse duodenum in vitro. Anat Ret 1987;217:250-5. 32. Ringold, GM. Steroid hormone regulation of gene expression. Annu Rev Pharmacol Toxic01 1985;25:529-66. 33. Madara J. Intestinal absorptive cell tight junctions are linked to the cytoskeleton. Am J Physiol1987;253:C171-5, 34. Madara JL, Neutra MN, Trier JS. Junctional complexes in fetal rat small intestine during morphogenesis. Dev Biol 1981;86:1708. 35. Quaroni A. Changes in keratins expression during intestinal cell differentiation. Gut 1987;28(Suppl]:282. 36. Vega-Sales DE, Salas PJI, Gunderson D, Rodriguez-Boulan E. Formation of the apical pole of epithelial cells: polarity of an apical protein is independent of tight junctions while segregation of a basolateral marker requires cell-cell interactions. J Cell Biol 1987;104:905-16. 37. Mathan M, Moxey PC, Trier JS. Morphogenesis of rat duodenal villi. Am J Anat 1976;146:73-92. 38. Misfeldt DS, Hamamoto SJ, Pitelka DR. Transepithelial transport in cell culture. Proc Nat1 Acad Sci USA 1976;73:1212-6. 39. Horan Hand P, Thor A, Schlom J, Rao CN, Liotta L. Expression of laminin receptor in normal and carcinomatous human tissues as defined by a monoclonal antibody. Cancer Res 1985;45:27139. 40. Ogle RC, Laurie GW, Kitten GT, Kandel SL, Bing JT. Isolation of a cell surface receptor for collagen type IV (abstract]. J Cell Biol 1987;105:136a. 41. Chen WT. Greve JM, Gottlieb DI, Singer SJ. Immunocytochemical localization of 14Okd cell adhesion molecules in cultured chicken fibroblasts, and in chicken smooth muscle and intestinal epithelial tissues. J Histochem Cytochem 1985;33:576-86. 42. Krotoski DM, Domingo C, Bronner-Fraser M. Distribution of a putative cell surface receptor for fibronectin and laminin in the avian embryo. J Cell Biol 1986;103:1061-71. 43. Stallmach A, Schuppan D, Dax J. Hanski C, Riecken E-O. Identification of the laminin receptor in cell membranes of a human colon adenocarcinoma cell line. Gut 1989 (in press). 44. Horwitz A, Duggan K, Greggs R, Decker C, Buck C. The CSAT antigen has properties of a receptor for fibronectin and laminin. J Cell Bio11985;101:2134-44.
February1990
45. Ruoslahti E, Pierschbacher MD. New perspectives in cell adhesion: RGD and integrins. Science 1987;238:491-7. 46. Gospodarowicz D, Delgado D, Vlodavsky I. Permissive effect of the extracellular matrix on cell proliferation. Proc Nat1 Acad Sci USA 1980;77:4094. 47.Quaroni A, Isselbacher KJ, Ruoslahti E. Fibronectin synthesis by epithelial crypt cells of rat small intestine. Proc Nat1 Acad Sci USA 1978;75:5548-52. 48. Simon-Assmann P, Bouziges F, Arnold C, Haffen K, Kedinger M. Epithelial-mesenchymal interactions in the production of basement membrane components in the gut. Development 1988;102:339-47, 49. Weiser MM, Ryzowicz S, Soroka C, Albini B. In vitro translation of rat intestinal RNA prepared from isolated villus and crypt
PROMOTERS OF CELL DIFFERENTIATION
335
cells and from the epithelium-denuded intestine. Synthesis of intestinal basement membrane. Trans Assoc Am Phys 1987;50: 316-28.
Received October 24.1988. Accepted July 21,1989. Address requests for reprints to: Ursula Hahn, M.D., Medizinisthe Klinik I, Krankenhausstrasse 12,852O Erlangen, Federal Republic of Germany. This work was supported by the Deutsche Forschungsgemeinschaft, Grant No. RI 136/12-3. The expert technical assistance of Brigitte Schneider and Doris Lazar is appreciated.
1990;98:322-335
Basement Membrane Components Are Potent Promoters of Rat Intestinal Epithelial Cell Differentiation In Vitro URSULA HAHN, A. STALLMACH,
E. G. HAHN, and
E. 0. RIECKEN Division of Gastroenterology. of Berlin, Hindenburgdamm,
Department of Medicine, Berlin, West Germany
Basement membranes have been implicated in morphogenesis and cell differentiation. In this study, the effect of basement membrane components on intestinal epithelial cell maturation in a mesenchyme-free environment was investigated. Fetal rat small intestinal epithelial cells (from the 1&h-17th day of gestation) were exposed to basement membrane-derived proteins (laminin, collagen type IV, and a complex basement membrane-enriched extract from the Engelbreth-Holm-Swarm sarcoma) and other extracellular matrix proteins (collagen type I and fibronectin) coated onto Petri dishes. The cells attached readily only to fibronectin and basement membrane proteins. For 5 days the developing epithelial colonies were monitored in vitro, assessing morphological and functional parameters of cell maturation. Colonies grown on laminin and the basement membrane extract were larger and of greater cell density. An increase in alkaline phosphatase and lactase activity was observed after 3-4 days in these colonies which could be enhanced to yield 90%-100% positive cells by the addition of dexamethasone to the medium while no sucrase-isomaltase activity was elicited. Electron microscopy confirmed a high degree of cellular polarization illustrated by tight junctions and apical microvilli in epithelial cells grown on a basement membrane-like support. In contrast, none of the other proteins stimulated the cells to mature in vitro. The authors conclude that certain basement membrane components actively promote fetal intestinal epithelial cell differentiation.
oward the time of birth, the epithelial cells of the rodent small intestinal mucosa proceed through a rapid course of differentiation and diversification to acquire the necessary degree of maturity for postnatal food absorption. During this prenatal period, the
T
Steglitz Medical School, Free University
developmental changes within the intestinal epithelium (endoderm) require the close proximity of intestinal mesenchyme (1). Even in vitro, fetal rat small intestinal epithelial cells can also be induced to differentiate if they are cocultured with homologous mesenthyme (2,3). The molecular basis for epithelial differentiation through mesenchymal influences has not been determined. It is not clear whether the induction of epithelial differentiation depends on the contiguity of vital mesenchymal cells, i.e., cell-cell contacts, or on biosynthetic products of the mesenchymal cells, Thus, the “mesenchymal signal” has not been identified. In this context, the nature of the epithelial-mesenchymal interface should be reviewed. The epithelialmesenchymal interface within the intestinal mucosa consists of the subepithelial basement membrane and its anchoring zone facing the adjacent reticular connective tissue of the lamina propria (4,~). Over the past decade, intensive chemical and biological research of basement membranes in general has shown that this formation of the extracellular matrix is a crucial element of organogenesis, tissue remodeling, and wound repair (6-9). Major structural components of the adult and fetal intestinal basement membrane have been identified by immunofluorescence and immunoelectron microscopy, i.e., the glycoproteins laminin, fibronectin, collagen type IV, nidogen, and a proteoglycan containing heparan sulfate sidechains (16-12). These are well-characterized, highly conserved macromolecules ubiquitous in all basement membranes analyzed so far in humans and many
Abbreviations used in this paper: BME, basement membrane matrix extract; DMEM, Dulbecco’s modified Eagle’s medium; PBS, phosphate-buffered saline. 0 1990 by the American Gastroenterological Association 0016-5095/90/$3.00
PROMOTERS OF CELL DIFFERENTIATION
February 1990
other species (6,7). However, their relationship to the basal plasma membrane of epithelial cells is not fully understood. It is likely that the function of the subepithelial basement membrane of the intestinal mucosa also exceeds mere mechanical needs. In the adult mucosa, the epithelial cells mature and migrate along the crypt-villus axis while staying in permanent contact with the subepithelial basement membrane. Besides giving structural support to the epithelium and compartmentalizing the mucosa, constituents of the basement membrane could conceivably be involved in epithelial cytodiff erentiation. Few attempts have been made to identify the nature of these cell-matrix interactions. One of the problems encountered is the lack of appropriate assays to study the behavior of viable, reasonably differentiated cells in vitro. Whereas mature enterocytes freshly isolated from mammalian intestinal tissues are viable only for a short time in vitro, epithelial cells from cell lines are immortal but often transformed and do not adequately respond to stimulation. An interesting new approach has been outlined recently making use of fetal epithelial cells that are still sensitive to their local environment (2,3,13). In this study, we investigated the response of fetal epithelial cells to extracellular matrix proteins in an attempt to identify components actively supporting cellular polarization and differentiation. The effect of distinct basement membrane proteins and interstitial collagens on the development of primary colonies of fetal small intestinal epithelial cells was determined morphologically and biochemically in vitro.
Materials and Methods Preparation
of Endoderm
Fetal Wistar rats were obtained by hysterectomy precisely on the morning of day 14,15,16, or 17 of gestation. The appearance of the vaginal plug was designated day 0. The entire small intestine from the pylorus to the appendix (ca. 2 cm) was removed, freed from mesenteric vessels, and incubated in 0.03% collagenase (CLSP II, 156 U/mg; Worthington, Freehold, N.J.) in Dulbecco’s minimal essential medium (DMEM; Boehringer Mannheim. Mannheim, Federal Republic of Germany) for 90 min at 36°C. The following procedures were performed at 4°C. Under a dissecting microscope, the enzymatically loosened mesoderm was pulled off the tighter endodermal tube and discarded. Larger sections of the endodermal tube were cut into fragments 0.5 mm long. The endodermal cells were then washed extensively with 10% serum to remove collagenolytic activity and seeded onto plain or precoated plastic Petri dishes. An average of 10 dishes containing lo-20 colonies per 35-mm Petri dish was obtained per litter. The purity of the preparation (no mesenchyme contamination] was checked by phase contrast microscopy and, in the beginning, by immunostain-
323
ing with antivimentin antibodies. The preparations were LOO%mesenchyme-free (if not, they were discarded at 24 h, at which time mesenchymal cells had proliferated abundantly and were clearly’visible).
Culture Conditions All cell cultures were performed in DMEM with 4 g/L glucose. Routinely, 4 mmol glutamine, 50 pg/ml ascorbic acid and 10% fetal calf serum were added [reagents from Biochrome, West Berlin, Federal Republic of Germany]. Cell cultures were incubated at 36’C and gassed with 8% CO, in air. For some experiments, dexamethasone (Sigma, Taufkirchen, Federal Republic of Germany] was added to the routine cell culture medium at a final concentration of 100 nM.
Extracellular
Matrix Substrates
The substrates used to precoat the culture dishes included murine collagen type I, human collagen type IV from placenta [prepared by D. Schuppan and E. G. Hahn, [14,15J), mouse collagen type IV (Bethesda Research Laboratories, Gaithersburg, Md.), mouse laminin, and human plasma fibronectin (Collaborative Research Inc., Waltham, Mass.]. Bovine serum albumin (Sigma, Taufkirchen, Federal Republic of Germany] was used as a control protein. All proteins were diluted to 10 or 20 pg/ml in phosphate-buffered saline [PBS], and 1 ml of the dilution was added to each 35-mm dish (Falcon Plastics, Cockeysville, Md.) and incubated for 90 min at 36°C. The solutions were then aspirated from the dishes. The efficiency of coating had been established previously for laminin by adding trace amounts of isotopelabeled laminin to various protein dilutions and calculating the recovery rate after a 90-min incubation period. At l-2 Mug protein per cm2 in 0.1 ml medium, saturation of binding of the protein to the surface of the Petri dish was observed. Ten to 20 endodermal fragments were added immediately with 0.8 ml medium per 35-mm Petri dish. Using albumin, fibronectin, or the basement membrane matrix extract (BME) substrate at 20 pg/ml, their presence on the Petri dish surface after the coating procedure was determined by staining the dish with 1% eosin for 5 min, which stained the surface area slightly pink while saline-treated dishes did not pick up any stain at all. For preparation of the BME substrate, C 57 black mice of both sexes each received 2 subcutaneous injections of 1 million Engelbreth-Holm-Swarm (EHS) sarcoma cells (16) within 1 wk. After 3-7 wk, the tumors were harvested, minced into small pieces, and immediately placed into ice-cold 4 M NaCl, 0.2 M Tris-HCl buffer, pH 7.2, and washed with a lo-fold volume for 1 h on ice. The EHS sarcoma matrix was extracted for 2 hand then for another 16 h with 2 M guanidin, 0.5 M NaCl, 0.2 M Tris-HCl buffer, pH 7.4. The second extract was centrifuged for 30 min at 15,000 rpm and the supernatant was extensively dialyzed, first against 0.5 M NaCl, 0.2 M Tris, pH 7.4, then against PBS at 4”C, and finally against DMEM for 24 h at 4%. This preparation was called BME and used for coating. By sodium dodecyl sulfate polyacrylamide gel electrophoresis
324 HAHN ET AL.
GASTROENTEROLOGY
and determination of the total protein concentration (17), the preparation contained laminin, collagen type IV, and nilO:l:l, as well as small dogen at a ratio of approximately amounts of fibronectin. The preparation was used at a protein concentration of 20 pg/ml.
Vol. 98, No. 2
Statistics Statistical analysis was performed Whitney U test for unmatched pairs.
using the Mann-
Results Analysis
of Endodermal
Attachment
and
Proliferation The attachment rate of the endodermal fragments was calculated after 24 h as the percentage of attached and spread endodermal colonies vs. the total number of endoderma1 fragments explanted. Endodermal proliferation was defined as the increase in size of the endodermal colonies over time. The total area covered by the explants was photographed every 24 h and calculated planimetrically from the films. Phase contrast microscopy and photography was performed on an inverted microscope [Zeiss IM [Zeiss, Oberkochen, Federal Republic of Germany] carrying 3.2 x , IOX, and 20x objectives, light field and phase contrast] using HP5 (Ilford Ltd., Essex, England] and Ektachrome 50 (Kodak, Hemel Hempstead, England] films.
Brush-Border
Enzyme Activity
The brush-border enzymes alkaline phosphatase, sucrase isomaltase, and lactase were determined in unfixed fetal intestinal epithelial cell colonies histochemically according to the method of Gutschmidt et al. (18,19). As controls, unfixed cryostat sections of freshly frozen newborn and adult rat intestines in which the epithelial brush borders reacted exclusively were routinely developed in parallel. The incubation conditions and incubation time were held constant for all procedures including the controls (15 min at 36°C). At least 25 epithelial colonies were stained histochemically for each enzyme on each substrate after 24 h and 4 days in vitro. In each experimental group, the colonies with the most intensive color reaction were evaluated using the grading index. Within a colony, all stained cells were considered positive and unstained (white] cells were registered as negative. At least 150 cells per colony center were counted. A grading index was chosen as follows: -, ~5% stained cells; +, 5%-10% stained cells; + +, lo%-50% stained cells; + + + , 50% -90% stained cells; + + + +, 90% 100% stained cells. Two investigators (U.H., AS.) determined the grading index independently without prior knowledge of the specific experimental conditions.
Attachment of Small Intestinal Endodermal Explants to Different Substrates In a first set of experiments, the entire length of small intestinal epithelium from the pylorus to the cecum was harvested on the 15th day of gestation, cut into 05mm fragments (cell clusters), and explanted in primary culture. [A total of 1281 individual endoderma1 fragments were explanted onto different surfaces in 7 consecutive experiments using 10 litters of fetal rats.) After 24 h, the wells were rinsed and the attached and spread colonies in each well were counted. Fetal epithelial clusters that had not attached by 24 h were removed because they continued to float and quickly degenerated. The endodermal cell clusters explanted on plain plastic surfaces or on Petri dishes coated with albumin or matrix proteins showed significantly different attachment rates (Figure 1). On plastic and on albumin-treated dishes, only 28% and 27%, respectively, of the endodermal fragments derived from 15-day-old fetal rat small intestine adhered to these substrates within 24 h. Endoderm
, 1001
--
I
:
Transmission electron microscopy was performed on transverse sections of endodermal cell colonies grown on plain Thermonox or coated Thermonox slides (Miles Lab., Munich, Federal Republic of Germany] prepared identically as described above for Petri dishes. These cultures were fixed in 1% glutaraldehyde containing 1% tannin in 0.1 M PBS, postfixed in 1% osmium tetraoxide and embedded in Mikropal [Ferak, West Berlin, Federal Republic of Germany) by the method of Merker and Barrach (20).
. . ..A.. .I.-
BSA
CI
BME
LAM
**** L
:
PLAS
Electron Microscopy
**'
CIV
I I
FIB
Figure 1. Attachment rates of fetal small intestinal epithelial cells on matrix protein. Rate of attachment of fetal intestinal epithelial cells resected on the 16th day of gestation and seeded onto plastic Petri dishes coated with various extracellular matrix proteins: plain plastic (PLAS),bovine serum albumin (BSA), collagen type I (CI), fibronectin (FIB),basement membrane tumor extract (BME), laminin (LAM), and collagen type IV (CIV). The attachment rate was derived from the number of attached and spread vs. floating cell clusters after 24 h of incubation at 36% by phase contrast microscopy. Bars indicate the median attachment rates (mean + 1 SD); numbers at the tops of the bars indicate the total number of explants in each group.
February 1990
explanted onto these substrates thus yielded only a few colonies. The greater variability in the attachment rates to plastic and albumin-coated dishes compared with the other substrates (SD > 50% on plastic in a total of n = 323) can be attributed to the low affinity of the endodermal colonies to plastic and albumin. Attachment was comparatively slow and weak on albumin and plastic. Even spread colonies were sometimes rinsed off before counting. A substantially higher yield of endodermal colonies was achieved by coating the Petri dishes with collagen type I, which resulted in an average attachment rate of 51% (p < 0.05) (Figure 1). In contrast, the plating efficiency of endodermal fragments on laminin, collagen type IV, and BME was much higher; 72%-75% of all fragments attached to these substrates and had spread to form single-cell monolayers within 24h (Figure 1). The attachment rate was highest on fibronectin (85%),which had a significantly higher rate than collagen type IV (p < 0.05). laminin, and BME (p < 0.01). The medium used in these experiments contained 10% fetal calf serum, but in a recently completed study using fetal intestinal epithelium in serum-free conditions, we have observed similar attachment rates (unpublished results). The increase in attachment efficiency was paralleled by an apparently stronger affinity to laminin, collagen type IV, and BME. Epithelial cells quickly spread on these substrates and could not be rinsed off with physiological buffers.
Affinity of Small Intestinal EpitheJiaJ Cells to Various Substrates as a Function of Developmental Stage It became obvious during our investigations that the developmental status of the epithelial cells played a major role in determining their affinity to the substrates. In an extra set of experiments, therefore, the affinity of epithelial cells derived from the small intestine on the 14th, 15th, 16th, and 17th days of gestation to a plastic surface was compared with their affinity to a BME-coated surface (Figure 2). Epithelial cells from day 14 of gestation attached by an average of 42% to plastic and 82% to the BME substrate by 24 h. Attachment rates were similar with epithelium derived from the 15th day of gestation (77% on BME), although a greater variation in the percentage of firmly attached cells on plastic was noted. Fetal intestinal epithelial cells harvested on day 16 displayed an attachment rate decreased to 67% on BME while an average of 28% of epithelial cell clusters bound to a plastic surface. By day 17 of gestation, only 6% of all explanted epithelial cell clusters attached to plastic within 24 h. Again, the BME substrate proved to be a significantly superior substrate although the attachment rate decreased to 44L70 of explanted clusters on
PROMOTERS
15.
OF CELL DIFFERENTIATION
16.
.17. day
325
of gestation
Figure 2. Aflinity as a function of gestational age. Attachment rates were calculated as in Figure 1 for fetal small intestinal epithelial cells resected from the small intestines of rat embryos on the 14th, ISth, 16th, and 17th days of gestation. Fetal epithelial cells were plated on BMEcoated dishes (open bars) and on albumin-coated dishes (solid bars). Bars indicate median attachment rates +1 SD. With increasing developmental age, the ability of the fetal small intestinal epithelial cells to attach to protein substrates is generally diminished, but the afenity to the BME substrate (44%) is 7 times greater than the afenity to albumin on day 17 of gestation (6%).
the 17th day of gestation (Figure 2). The ratio of attachment rates thus increases from 2:l (day 14) to 7:1 [day 17). These results also imply that the developmental gradient along the longitudinal axis of the small intestine (with the proximal intestine leading by approximately 24-48 h in terms of brush-border enzyme expression) should influence epithelial cell attachment rates. In another set of experiments, duodenal explants were indeed observed to have an equally reduced attachment rate to all substrates compared with distal explants (data not included). In the experiments described here, a random distribution of proximal and distal fragments was achieved by mixing all fragments before explantation. In summary, the development of the fetal rat small intestine between the 14th and 17th days of gestation is accompanied by a loss of plating efficiency of isolated epithelial cells to all substrates. However, the greater affinity of these cells to basement membrane proteins compared with control surfaces is enhanced at later stages of development. Proliferation and Survival of Endodermal Explants on Various Substrates To evaluate the proliferation rate of the endodermal cells as an increase in cell number over time, establishing the number of cells correctly at any time proved difficult. The total number of cells in an individual colony (estimated from composite photographs) was too small to yield appropriate DNA levels and too great to be counted morphometrically. There-
326 HAHN ET AL.
fore, we took another approach. By monitoring the explants closely by phase contrast microscopy, we had observed that all the explants expanded in the same manner. After the initial attachment and spreading phase, the colonies were composed of sessile, dividing, and migratory cells. Migratory cells were obvious at the periphery, displaying pseudopodiae and a ruffled cytoplasmic edge, while mitotic figures were found throughout the entire colony. Hence, the increase in the area covered by individual explants over the first 3 days in vitro represented the combined effect of cell migration and cell division. The incorporation of bromodeoxyuridine, a thymidine-analogue base, as a parameter of DNA synthesis correlated well with the increase in colony size (A. Stallmach et al., manuscript in preparation). Although the individual cell shape also may have determined the size of a colony, this factor was not taken into account because the cells within the largest colonies (on BME and laminin) had the smallest cell diameter. Therefore, cell shape did not additionally increase colony size but rather favored an underestimation of cell counts on the BME substrate. In conclusion, we considered the expansion of the colonies as a rough but proportional indicator of cell proliferation. Individual colonies were photographed precisely every 24 h and evaluated planimetrically (Figure 3). On all substrates the endodermal colonies reached their maximal expansion by the 3rd day in vitro (Figure 3). However, the actual colony size differed considerably. While endodermal colonies averaged 0.97 + 0.35 mm2 on albumin and 0.94 + 0.41 mm2 on collagen type IV, they expanded to 1.24+ 0.52mm2 on laminin and 2.34 + 0.73mm2 on BME. These differences were significant between BME and all other substrates (p < 0.01) and between laminin and collagen type IV or albumin. Interestingly, the expansion of endodermal colonies on fibronectin was very low and stopped after 48 h at values even below the values for the indifferent control protein albumin. After the 3rd day, more or less rapid deterioration took place (Figure 3). Cell loss from the substrate was striking on fibronectin, collagen type IV, and plastic. On laminin, morphologically intact sections of the monolayers remained in place until after the 4th day in vitro, after which all cells detached. On BME, however, colony centers stayed viable until the 6th day in vitro. The sharp decline in colony size occurring after the 3rd day on BME was caused by detachment of the most peripheral cells only. In summary, BME was the most favorable substrate for both proliferation and survival of mesenchymefree fetal intestinal epithelium in vitro leading to a 2.4-fold increase in colony size and to the survival of endodermal cells for up to 6 days in vitro. Laminin had a similar, albeit less pronounced, effect. However,
GASTROENTEROLOGYVol. 98,No.z
n q
BME LAM CIV FIB
w 0
i-
1
2
3 days in vitro
5
Figure 8. Growth of intestinal epithelial colonies. The area of individual fetal intestinal epithelial colonies cultured for 8 days on various extracellular matrix proteins--basement membrane extract (BME), laminin (LAM), collagen type IV (CIV), collagen type I (0, fihronectin (FIB), and bovine serum albumin (BSA)-was calculated planimetrically from phase contrast photographs. A minimum of 20 single colonies per substrate and per day were evaluated. Mean values are expressed in square millimeters +I SD.
collagen type IV and fibronectin did not promote colony growth or survival although the high initial attachment rates suggest specific affinity for both molecules.
Light Microscopic and Ultrastructural of Endodermal Explants
Aspect
Morphologically, the initial spreading phase of the epithelial explants was similar on all substrates. Small, circular monolayers developed, and at 24 h all explants were about the same size. They contained epithelial cells resembling those of intestinal epithelial cell lines (21). At 46 h, however, the size of the colonies and the shape of the cells within were different among the groups (Figure 4).Colonies grown on albumin (Figure 4A) and collagen type IV (Figure 4B) were small, and their cell population was homogenous. In contrast, fetal epithelial cell colonies cultured on laminin were larger with vigorously proliferating peripheral cells and a central zone of smaller, densely packed cells (Figure 4C). At the same time, fetal epithelial colonies grown on BME had the largest diameters; their central zone consisted of tightly arranged, narrow cells with intercellular gaps. Often a small area near the center
PROMOTERS OF CELL DIFFERENTIATION
February 1990
327
Figure 4. Light microscopic aspect of fetal epithelial colonies at 48 h grown on albumin, collagen type IV, laminin, and BME. Original magnification x 100. A. Epithelial colony grown on albumin. B. Epithelial colony grown on collagen type IV. In both types of outgrowth the cells are essentially homogenous. C. Center formation in a colony on laminin. Notice also the larger size. D. Densely populated center of a large, still-expanding visible by its lack of focus.
colony on the BME substrate. In the very center, a budding dome is emerging which is
was elevated above the plane of the Petri dish indicative of a budding dome [Figure 40). After 72 h, however, striking variations in the morphology of the outgrowths became apparent (Figure 5). Cultivated on plastic, albumin, collagen types I and IV, and fibronectin, fetal intestinal epithelial cells derived from 15-day-old fetal rats retained a flat, well-spread morphology. Most cells had a large cytoplasm, a well-recognizable nucleus containing two nuclei, and no granules or inclusions as shown in a colony on collagen IV in Figure 5A. There was no conspicuous difference between the center and the periphery of the colonies. In contrast, the central region of most colonies cultured on laminin and BME expanded toward the periphery. It became so tightly packed that individual cells were hardly distinguish-
able. A typical example is given in Figure 5B. On BME, many cells at the colony border contained clear cytoplasmic inclusions (Figure 5D). These inclusions were not stained by the lipid-detecting Sudan Black B stain or by the periodic-acid Schiff base, which stains complex carbohydrates (data not shown]. Therefore, there was no evidence by phase contrast microscopy of epithelial cytodifferentiation to goblet or endocrine cells. A striking feature observed exclusively in epithelial cultures on BME was domes. Domes consist of a group of cells lifted from the surface of the dish by the pressure building up from fluid secreted by the cells at their basal membranes. Tight intercellular junctions are required. In 40% of the colonies, up to 5 individual domes per colony were formed after 48-60 h in vitro.
328 HAHN ET AL.
GASTROENTEROLOGY
Vol. 98, No. 2
Figure 5. Fetal epithelial cell cultures on different matrix protein substrates after 72 h in vitro. A-D. Original magnification x 200. A. Center of an epithelial colony grown on collagen type IV, displaying uniformly flat, spread cells of equal diameters. B. Epithelial-cell colony grown on laminin. The center of the colony is shown in the left upper corner and consists of tightly packed cells. The epithelial cells surrounding it appear in a typical cobblestone pattern characteristic of a sheet of polarized epithelial cells. C. Epithelial cell colony grown on the BME substrate. The epithelial cell sheet is focused so that the 2 domes of elevated epithelial cells that have arisen in the middle of the colony appear out of focus. Domes are formed when the epithelial cells are joined by tight junctions and fluid secreted at their basal plasma membrane causes the cells to become elevated above the level of the protein substrate coating the Petri dish. D. At the periphery of a colony of fetal epithelial cells grown on the BME substrate, many migrating cells contain light, translucent inclusions. The cells at the periphery are spread out.
This suggests active, vectorial ion transport by the epithelial cells (Figures 5C and 6B). In fact, electron microscopy of transverse sections through the center portion of individual colonies grown on BME showed highly polarized epithelial cells with a regular brush border. The height of individual epithelial cells within domes was 8 times greater than that of the flat central cells seen on albumin or collagen types I and IV. The nucleus appeared in the lower half of the cells with an overlying golgi apparatus. Electron-dense granules indicating endocrine cells were not observed. Tight junctions and intermediate junctions were present (Figure 7B). Migratory cells at the very periphery of the colonies on BME appeared much flatter, and the brush border was irregular (Figure 6A). Similarly, fetal intestinal epithelial cells grown on albumin and colla-
gen type I retained their flat, unpolarized morphology, carried no microvilli or tight junctions, and therefore resembled the undifferentiated epithelial cells of intestinal epithelial cell lines in vitro [Figure 7A) (21). Ultrastructural analysis thus confirmed that endodermal cells grown on BME reached a high and organotypic level of differentiation within 3 days in vitro. In contrast, fetal intestinal epithelium cultured on indifferent substrates showed no such signs of morphological maturation. Activity of Brush-Border Endodermal Explants
Enzymes in
The epithelial colonies were studied histochemically for evidence of alkaline phosphatase, lactase,
February 1990
PROMOTERS OF CELL DIFFERENTIATION
329
and sucrase isomaltase. The percentage of stained vs. unstained cells was evaluated, and a grading index ranging from negative (-, ~5% cells stained] to strongly positive (+ + + +, >90% cells stained] was applied. For the experiments described in the following paragraph, we again used only colonies derived from the small intestinal epithelium of the 15th day of gestation which in situ contains no brush-border enzyme activity (221. Accordingly, frozen sections of the starting material did not produce a color reaction in any staining method. No brush-border enzyme activity was registered at all on albumin, fibronectin, or collagen type I (Figure 8A and B). On collagen type IV, only 5%-8% of all colonies were positive for alkaline phosphatase after 4 days. In the presence of 100 nmol dexamethasone, similar levels of alkaline phosphatase activity were observed in colonies grown on collagen type IV (Figure 8C and D). In contrast, epithelial colonies
Figure 6. Transverse electron microscopy of fetal epithelial cells grown on the BME substrate for 72 h. Bars = 1 pm. A. A peripheral cell at the edge of the colony is unpolarized and spread but displays an irregular brush border [original magnification x 9000). B. Cells at the foot of a dome are cuboidal and are in the process of separating from the substrate. Irregular protein deposits at the basal plasma membrane [arrows] probably represent adherent BME (original magnification x4500). C. Highly polarized epithelial cells in the center of a colony exhibit more microvilli. tight junctions, and a basally located cell nucleus. Arrows point to adherent extracellular material that is possibly BME-derived [original magnification x4500).
Figure 7. TEM of epithelial cells grown on collagen type I and BME. Original magnification x 14,660; bar = 1 pm. A. This epithelial cell of the center of a colony grown on collagen type I for 72 h shows no signs of differentiation. B. Brush border of 2 epithelial cells from the center of a colony grown on BME. The microvilli are well formed and regular. A tight junction (arrow] connects the two adjacent cells.
330 HAHN ET AL.
GASTROENTEROLOGY Vol. 98, No. 2
Figure 8. Brush-border enzyme activity (alkaline phosphatase) in epithelial cell colonies on different matrix protein substrates after 96 h in vitro. Light field optics or phase contrast; original magnification x 100. A and B. Fetal intestinal epithelial colony grown on albumin. There are no positive cells except some rounded-up colony [arrow]: light field (A] and phase contrast (B) views of the same specimen.
cells that are leaving the
C and D. Center of a colony grown on collagen type IV in the presence of dexamethasone. Throughout the colony, some cells are stained very faintly. A positive cell is marked by an arrow; light field (C) and phase contrast [D] images of same area. E and F. Fetal epithelial colonies cultured on laminin in normal medium (E) or in the presence of 100 nM dexamethasone already-strong staining pattern achieved by normal medium is further enhanced in the presence of dexamethasone. G. Colony grown on BME. More than 90% of cells carry enzyme activity throughout medium. However, randomly distributed nonreactive (white) cells remain. If. Fetal epithelial colony grown on BME in the presence of dexamethasone. cells.
the entire colony that was maintained
(F). The in normal
Abundant expression of the enzyme is seen in practically all
PROMOTERS
February 1990
grown on laminin steadily increased the expression of alkaline phosphatase until the 4th day in vitro (Figure 8E). At this point, 40%-50% of all cells within the colonies were stained intensely for alkaline phosphatase and rated grade + +. By the 4th day in vitro in normal medium, epithelial colonies on laminin also turned positive for lactase enzyme activity (grade +). Addition of dexamethasone to these cultures clearly increased the number of positive cells and the intensity of the reaction for both enzymes (Table 1). Fetal intestinal epithelial cells cultured on BME expressed the strongest activity of alkaline phosphatase. By the 4th day in vitro, 80%-90% of all endodermal cells were intensely stained for alkaline phosphatase, grade +++ (Figure 8G). Dexamethasone promoted alkaline phosphatase activity even further so that 90%-100% of all epithelial cells tested positive. The intensity of the color reaction was impressive (Figure 8H). In addition, lactase activity appeared in these cultures after 2 days in vitro and increased until up to 50% of all cells were stained the typical, bright turquoise color of the lactase reaction identical to the color product of adult intestinal brush borders. Taken together as shown in Table 1, only the explants cultured on laminin and BME expressed brush-border enzyme activities compatible with the degree of morphological maturation illustrated by electron microscopy. Fetal epithelium on plastic, albumin, fibronectin, and collagen types I and IV displayed minimal enzyme activity under these conditions, indicating low intrinsic self-differentiating properties of the endoderm. This seemed to be true for explants derived from both the duodenal and ileal endoderm. Because some preliminary experiments
Table
1.
Relative Activity of Brush-Border Enzyme Expression Substrates
OF CELL DIFFERENTIATION
331
separating proximal from ileal endoderm did not show a conspicuous acceleration of development of the proximal endoderm in vitro, we subsequently performed a separate study in which >2000 endoderma1 colonies were separated according to anatomical origin and the development of brush-border enzyme expression was investigated (U. Hahn, manuscript in preparation). Because epithelial cell differentiation of the proximal endoderm appears more complex than ileal endoderm in vitro, these results will be reported in detail in a separate study. Dexamethasone did not elicit any brush-border enzyme activity de novo in primarily negative colonies. This was true whether the medium was supplemented with hormone at the onset of the experiment or after 24-48 h. However, once brush-border enzyme expression had been initiated in colonies growing on laminin or BME, both alkaline phosphatase and lactase activity increased under the influence of dexamethasone (Table 11. Sucrase isomaltase remained undetectable even in the presence of dexamethasone.
Discussion This study demonstrates that certain extracellular matrix components can support cellular differentiation in fetal small intestinal epithelial cells in the absence of mesenchymal cells in vitro. Purified laminin, a glycoprotein found exclusively in basement membranes, and a complex, basement membrane protein-enriched extract (BME] acted as promoters. In contrast, another basement membrane-specific protein, collagen type IV, was much less effective. Fibronectin and collagen type I, both pertaining to the
in Small Intestinal
EpitheJiaJ Cells Cultured on Different
Substrate (days in vitro)
BME Enzyme Alkaline phosphatase NM DEX Lactase NM DEX Sucrase isomaltase NM DEX
1
Laminin 4
Collagen I, fibronectin, albumin, plastic
Collagen IV
1
4
1
4
1
4
+ ++
+++ ++++
+ +
++ +++
_ _
+ +
_
_ _
_ _
++ +++
_ _
+ ++
_ _
_
_
_ _
_
_ _
_ _
_ _
_ _
_ _
_ _
_
Brush-border enzyme activity was demonstrated histochemically. In the center of the endodermal colonies (comprising approximately 200 cells). all positive and negative cells were counted and the relative enzyme activity was graded according to the following index: -, ~5%; +, 5%-9%; + +, lo%-49%; + + +, 50%-89%; + + + +. 90%-100% positive, stained cells. NM, normal medium: DEX, medium containing 100 nmol dexamethasone.
332
HAHN ET AL.
GASTROENTEROLOGY
Vol. 98, No. 2
colonies was observed in response to the BME subreticular extracellular matrix rather than to the basement membrane proper, did not evoke a cellular strate and laminin. response substantially different from the control proGlucocorticoids have also been shown to enhance tein albumin. epithelial viability and brush-border enzyme activity Previous studies confined to testing the affinity of in vitro in many intestinal organ culture studies (28single intestinal epithelial cells from different species 31). The same effect was seen in epithelial-mesenchyto extracellular matrix proteins had produced diverse This has led to the assumption ma1 cocultures (1,2,13). results (22-25). Our own experience with a different that the action of the hormone is transmitted via fetal organ culture model had suggested that the mesenchymal cells (13). In contradistinction, the presence of viable mesenchymal cells was obligatory present study suggests that the epithelial cells do not for the formation of an authentic basement membrane need any mesenchymal support to respond to dexain vitro. In those studies, the intestinal epithelial cells methasone. In the complete absence of mesenchymal had not been separated from a core of underlying cells, the critical minimal degree of epithelial polarizamesenchymal cells (3). Mesenchymal and epithelial tion and intracellular organization required for horcells were shown to cooperate in the formation of a mone responsiveness was apparently induced by the basement membrane at their interface, and a concomBME complex and laminin alone. Dexamethasone itant increase in epithelial differentiation was obthus acted as a permissive promoter for alkaline served. The previous organ culture-type model, howphosphatase and lactase. Laminin and dexamethaever, was not suitable for the investigation of the effect sone complement each other with respect to brushof purified extracellular matrix proteins on epithelial border enzyme expression. Both agents have been differentiation because of the presence of mesenchyshown to stabilize cellular mRNA pools in other ma1 cells producing these proteins (and perhaps other systems (27,32). More studies are needed to determine factors as well). Therefore, for the study presented the molecular nature of such events in intestinal cells. here, endoderm and mesoderm were completely sepaIt specifically remains to be seen whether the increase rated by proteolytic and mechanical treatment of the in enzyme activity is caused by the induction or fetal gut. The reaction of purified epithelial cells to stabilization of enzyme synthesis on the transcripdifferent matrix proteins could then be monitored in tional level. Surprisingly, however, no histochemical vitro. evidence of sucrase isomaltase activity was detected We also raised the question of whether basement in any colony although this enzyme is expressed in membrane proteins could elicit cellular responses endodermal-mesenchymal cocultures of the same gesfrom intestinal epithelial cells beyond attachment and tational age after 4-6 days in vitro (2,13). Dexamethaspreading on coated surfaces. Evidence for the stimusone, which is particularly well known for its ability to latory effect of basement membrane glycoproteins on induce precocious sucrase isomaltase activity in small the differentiation of other target epithelial cells in intestinal epithelial cells, both in vivo and in vitro vitro has been reported (6,7). For example, cultured rat (28,301, failed to stimulate enzyme expression here. Sertoli cells respond to a basement membrane proteinThis may be due to a separate mode of intracellular rich gel derived from the same source, the Engelbrethregulation for sucrase isomaltase indicated by its 48-h Helm-Swarm tumor, by polarization and the acquisidelay in appearance in situ after that of alkaline tion of a cell shape virtually identical to mature Sertoli phosphatase and lactase (28). cells in vivo (26). Mammary epithelial cells display a The ultrastructural morphology of the epithelial cuboidal cell shape and maintain a high level of cells, as assessed by transmission electron microscopy, &casein messenger RNA (mRNA) level when culserved as another parameter of cellular maturation. tured on laminin or the EHS extract (27). Laminin and Intercellular junctions, cellular height, and the presthe EHS matrix seem to stabilize the mRNA pool for ence of microvilli were determined as a function of casein proteins in mammary epithelial cells, thereby the duration of the culture and the respective matrix promoting the tissue-specific cellular phenotype. This component. The BME substrate induced an s-fold particularly intriguing observation indicates the presincrease in cellular height in the colony center, reguence of intracellular pathways activated by specific lar microvilli of the apical plasma membrane, and cell-matrix interactions. We determined the exprestight junctions between adjacent cells. The degree of sion of brush-border enzyme activity as one parameter maturation induced by the BME substrate is comparaof the intestinal epithelial phenotype. Increasing activble to that induced by underlying mesenchymal cells ity of brush-border enzymes is generally considered a as shown in the earlier study (12). However, slightly good indication of enterocyte maturation (18,19,28). more elongated microvilli and occasional goblet cells Indeed, an impressive induction of alkaline phoswere observed in the coculture system (12). phatase and lactase activity in fetal epithelial cell
February 1990
The importance of junctional complexes and the cytoskeleton in determining cell polarity is well established (33-35). However, it is not known whether the formation of tight junctions and the composition of the basolateral membrane are related as in other cells (36). We observed that the plating efficiency of fetal epithelial cells from day 17 of gestation on basement membrane proteins was markedly lower than that of fetal cells from days 14-15, although their affinity to the BME substrate had increased relative to controls (Figure 2). This may indicate that the composition of the basal-cell membrane undergoes changes equally as important as those in the apical membrane during this period. In vivo, the brush border is being loaded with enzyme protein which is illustrated by a substantial increase in enzyme activity on the 19th day of gestation in vivo. Therefore, on the 19th day of gestation, an increase in brush-border enzyme expression and the acquisition of mature junctional complexes occur contemporaneously at the time when all epithelial cells have come into contact with the basement membrane (37). This corresponds precisely to the time in this model (explanation on the 15th day of gestation plus 4 days in vitro) when epithelial cells grown on BME have formed tight junctions and alkaline phosphatase is almost uniformly expressed. Thus, contact of the basal cell membrane with the basement membrane may enable the epithelial cell to organize its other membrane domains. When tight intracellular junctions are formed between epithelial cells in vitro, a coherent epithelium that presents its “luminal” side to the culture medium is established. The epithelial cells may then begin to actively absorb and secrete ions and water (38). It is consequently not surprising to detect domes (or hemicysts) in such epithelial monolayers. A dome is a small, hollow mound consisting of a group of cells lifted off the surface of the coated Petri dish through the pressure of fluid secreted by the cells at their basal plasma membrane. In the fetal epithelial colonies, domes were found exclusively in colonies grown on the BME substrate. Therefore, the tight junctions demonstrated morphologically were also found to function. High affinity of intestinal epithelial cells to laminin, fibronectin, and collagen type IV, in combination with the distinct cellular responses to these proteins, suggests the presence of specific transmembrane-binding proteins or receptors on the cell membrane. There is indeed evidence of at least 3 different types of matrix protein-binding proteins on the basal plasma membrane of enterocytes in vertebrates. A laminin receptor-like molecule, a collagen type IV-binding protein, and a well-defined fibronectin-binding protein (the so-called cell substrate attachment (CSAT) antigen in
PROMOTERS OF CELL DIFFERENTIATION
333
avian tissues or integrin complex) have been identified immunohistologically in the intestinal mucosa of different species (39-43). Integrins are known to span the plasma membrane and to connect with components of the cytoskeleton. A link is thereby established between intracellular and extracellular structural networks (44,45). On both laminin and BME, differentiation of fetal intestinal epithelial cells was preceded by proliferation. Colonies on BME tripled in size compared with fibronectin (Figures 3 and 4). It is not clear how the BME extract and laminin may promote cell proliferation (6,27,46). However, in the physiological context of wound healing, it seems desirable for the authentic basement membrane to exert such an effect. Erosions in the gastrointestinal tract in which the subepithelial basement membrane remains intact reepithelialize faster than ulcerous defects. As reported earlier, primary fetal epithelial cells and transformed epithelial cells are capable of basement membrane protein production (3,47). However, synthesis of laminin and collagen type IV have also been reported in the subepithelial mesenchyme (3,48,49). It is unresolved to what extent epithelial cells and fibroblasts participate in basement membrane formation in vivo. We are currently comparing matrix protein synthesis in mesenchyme-free epithelial cell colonies with that in diverse cocultures (unpublished results). Observations to date indicate that the very low level of endogenous matrix protein synthesis observed in freshly explanted epithelial cells is insufficient to induce differentiation and is altered by exposure of the cells to different substrates. In conclusion, we have presented evidence of the morphological and functional differentiation of fetal small intestinal epithelial cells through the action of a complex, basement membrane protein-enriched extract and laminin in vitro. We suggest that the basement membrane also plays a crucial role in the induction of epithelial differentiation during morphogenesis in utero when stratified fetal endodermal cells are organized into a monolayer of polarized enterocytes. In the adult mucosa, it is conceivable that both subepithelial myofibroblasts and epithelial cells participate in basement membrane protein production, the assembly of which occurs at the epithelial basal membrane. Whether some of these components are intrinsically linked to the process of epithelial cell differentiation along the crypt-villus axis in the adult human mucosa remains to be determined. References 1. Haffen
K, Kedinger M. Simon-Assmann P. Mesenchymedependent differentiation of epithelial progenitor cells in the gut. J Pediatr Gastroenterol Nutr 1967:6:14-23.
334 HAHN ET AL.
2. Stallmach A, Hahn LJ, Merker H-J. Hahn EG, Riecken E-O. Differentiation of rat intestinal epithelial cells is induced by organotypic mesenchymal cells in vitro. Gut 1989;30:959-70. 3. Hahn U, Schuppan D, Hahn EG, Merker H-J, Riecken E-O. Intestinal cells produce basement membrane proteins in vitro. Gut 1987;28(Suppl1):143-151. 4. Takahashi-Iwanaga H, Fujita T. Lamina propria of intestinal mucosa as a typical reticular tissue. Cell Tissue Res 1985;242:5766. 5. Komuro T. Fenestrations of the basal lamina of intestinal villi of the rat. Cell Tissue Res 1985;239:183-8. 6. Martin GR, Timpl R. Laminin and other basement membrane components. Annu Rev Cell Bioll987;3:57-85. 7. Timpl R, Dziadek M. Structure, development and molecular pathology of basement membranes. Int Rev Cell Bioll986;2:2749. 8. Martinez-Hernandez A, Amenta R. The basement membrane in pathology. Lab Invest 1983;48:656-77. 9. Trelstad R. Role of extracellular matrix in development. New York: Liss, 1984. 10. Laurie GW, Leblond CP, Martin GR. Localization of type IV collagen, laminin, heparansulfate proteoglycan and fibronectin to the basal lamina of basement membranes. J Cell Biol 1982;95:340-4. 11. Hahn LJ, Schuppan D, Hahn EG. Riecken E-O. Immunhistologisthe Charakterisierung interstitieller anitigener Substanzen in der Darmwand. Fortschr Gastroenterol Endosk 1985;14:198-201. 12. Martinez-Hernandez A, Chung AE. The ultrastructural localization of two basement membrane components. J Histochem Cytochem 1984;32:289-98. 13. Kedinger M, Simon-Assmann P, Alexandre E, Haffen K. Importance of a fibroblastic support for in vitro differentiation of intestinal endodermal cells and their response to glucocorticoids. Cell Differ 1987;20:171-82. 14. Schuppan D, Besser M, Schwarting R, Hahn EG. Radioimmunoassay for the carboxyterminal crosslinking domain of type IV procollagen in body fluids. J Clin Invest 1986;78:241-8. 15. Hahn EG, Wick G, Pencev D, Timpl R. Distribution of basement membrane proteins in normal and fibrotic human liver-collagen type IV, laminin and fibronectin. Gut 1980;21:63-71. 16. Orkin RW, Gehron P, McGoodwin EB, Martin GR, Valentine T, Swarm R. A murine tumor producing a matrix of basement membrane. J Exp Med 1977;145:204-20. 17. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement with the Folin phenol reagent. J Biol Chem 1951;193:265-75. 18. Gutschmidt S, Lange U, Riecken E-O. In situ measurements of protein contents in the brush border region along rat jejunal villi and their correlations with four enzyme activities. Histochemistry 1980;72:467-9. 19. Gutschmidt S, Kaul W, Riecken E-O. A quantitative histochemical technique for the characterization of alpha-glucosidases in the brush border membrane of rat jejunum. Histochemistry 1979:63:81-101. 20. Merker HJ, Barrach HJ. The morphology of basement membrane formation. Eur J Cell Bio11981;26:111-20. 21. Quaroni A, Wands J. Trelstad RL, Isselbacher KJ. Epithelioid cell cultures from rat small intestine. J Cell Biol 1979;80:248-65. 22. Burrill PH. Bernardini I, Kleinman H, Kretchmer N. Effect of serum, fibronectin and laminin on adhesion of rabbit intestinal epithelial cells in culture. J Supramol Struct Cell Biochem 198X16:385-92. 23. Rattner DW, Ito S. Rutten MJ. Silen W. A rapid method for culturing guinea pig gastric mucous cell monolayers. In Vitro 1985;21:453-62.
GASTROENTEROLOGY
Vol. 98, No. 2
24. Hahn U, Cho A, Schuppan D, Hahn EG, Merker HJ, Riecken EO. Intestinal epithelial cells preferentially attach to a biomatrix derived from human intestinal mucosa. Gut 1987;28(Suppl 1):153-8. 25. Hahn U. The role of extracellular matrix proteins in small intestinal cell cultures. Stand J Gastroenterol 1988;23(Suppl 151):70-8. 26. Hadley MA, Byers SW, Suarez-Quian CA, Kleinman HK, Dym M. Extracellular matrix regulates Sertoli cell differentiation, testicular cord formation and germ cell development in vitro. J Cell Bio11985;101:1511-22. 27. Li ML, Aggeler J, Farson DA, Hatier C, Hassell J, Bissell M. Influence of a reconstituted basement membrane and its components on casein gene expression. Proc Nat1 Acad Sci USA 1987;84:136-40. 28. Simon-Assmann P, Kedinger M, Grenier JF, Haffen K. Control of brush border enzymes by dexamethasone on the fetal rat intestine cultured in vitro. J Pediatr Gastroenterol Nutr 1982;l: 257-65. 29. Moog F. The differentiation and dedifferentiation of the intestinal epithelium and its brush border enzymes. In: Development of mammalian absorptive processes. Ciba Foundation series 70. Amsterdam: Elsevier, 1979:31-50. 30. Quaroni A. Crypt cell development in newborn rat small intestine. J Cell Bio11985;100:1601-10. 31. Beaulieu JF, Calvert R. Hormonal regulation of epithelial cell proliferation in the fetal mouse duodenum in vitro. Anat Ret 1987;217:250-5. 32. Ringold, GM. Steroid hormone regulation of gene expression. Annu Rev Pharmacol Toxic01 1985;25:529-66. 33. Madara J. Intestinal absorptive cell tight junctions are linked to the cytoskeleton. Am J Physiol1987;253:C171-5, 34. Madara JL, Neutra MN, Trier JS. Junctional complexes in fetal rat small intestine during morphogenesis. Dev Biol 1981;86:1708. 35. Quaroni A. Changes in keratins expression during intestinal cell differentiation. Gut 1987;28(Suppl]:282. 36. Vega-Sales DE, Salas PJI, Gunderson D, Rodriguez-Boulan E. Formation of the apical pole of epithelial cells: polarity of an apical protein is independent of tight junctions while segregation of a basolateral marker requires cell-cell interactions. J Cell Biol 1987;104:905-16. 37. Mathan M, Moxey PC, Trier JS. Morphogenesis of rat duodenal villi. Am J Anat 1976;146:73-92. 38. Misfeldt DS, Hamamoto SJ, Pitelka DR. Transepithelial transport in cell culture. Proc Nat1 Acad Sci USA 1976;73:1212-6. 39. Horan Hand P, Thor A, Schlom J, Rao CN, Liotta L. Expression of laminin receptor in normal and carcinomatous human tissues as defined by a monoclonal antibody. Cancer Res 1985;45:27139. 40. Ogle RC, Laurie GW, Kitten GT, Kandel SL, Bing JT. Isolation of a cell surface receptor for collagen type IV (abstract]. J Cell Biol 1987;105:136a. 41. Chen WT. Greve JM, Gottlieb DI, Singer SJ. Immunocytochemical localization of 14Okd cell adhesion molecules in cultured chicken fibroblasts, and in chicken smooth muscle and intestinal epithelial tissues. J Histochem Cytochem 1985;33:576-86. 42. Krotoski DM, Domingo C, Bronner-Fraser M. Distribution of a putative cell surface receptor for fibronectin and laminin in the avian embryo. J Cell Biol 1986;103:1061-71. 43. Stallmach A, Schuppan D, Dax J. Hanski C, Riecken E-O. Identification of the laminin receptor in cell membranes of a human colon adenocarcinoma cell line. Gut 1989 (in press). 44. Horwitz A, Duggan K, Greggs R, Decker C, Buck C. The CSAT antigen has properties of a receptor for fibronectin and laminin. J Cell Bio11985;101:2134-44.
February1990
45. Ruoslahti E, Pierschbacher MD. New perspectives in cell adhesion: RGD and integrins. Science 1987;238:491-7. 46. Gospodarowicz D, Delgado D, Vlodavsky I. Permissive effect of the extracellular matrix on cell proliferation. Proc Nat1 Acad Sci USA 1980;77:4094. 47.Quaroni A, Isselbacher KJ, Ruoslahti E. Fibronectin synthesis by epithelial crypt cells of rat small intestine. Proc Nat1 Acad Sci USA 1978;75:5548-52. 48. Simon-Assmann P, Bouziges F, Arnold C, Haffen K, Kedinger M. Epithelial-mesenchymal interactions in the production of basement membrane components in the gut. Development 1988;102:339-47, 49. Weiser MM, Ryzowicz S, Soroka C, Albini B. In vitro translation of rat intestinal RNA prepared from isolated villus and crypt
PROMOTERS OF CELL DIFFERENTIATION
335
cells and from the epithelium-denuded intestine. Synthesis of intestinal basement membrane. Trans Assoc Am Phys 1987;50: 316-28.
Received October 24.1988. Accepted July 21,1989. Address requests for reprints to: Ursula Hahn, M.D., Medizinisthe Klinik I, Krankenhausstrasse 12,852O Erlangen, Federal Republic of Germany. This work was supported by the Deutsche Forschungsgemeinschaft, Grant No. RI 136/12-3. The expert technical assistance of Brigitte Schneider and Doris Lazar is appreciated.