expression of j1 tenascin in the crypt-villus unit of ... · jl/tenascin in mouse small intestine...

Development 109, 313-321 (1990)Printed in Great Britain ©The Company of Biologists Limited 1990

313

Expression of J1 tenascin in the crypt-villus unit of adult mouse small

intestine: implications for its role in epithelial cell shedding

RA1NER PROBSTMEIER*, RUDOLF MARTINI and MELITTA SCHACHNER

Department of Ncurobiology, University of Heidelberg, Im Neuenheimer Feld 364, 69 Heidelberg, Fed. Rep. Germany

* For correspondence

Summary

The localization of the extracellular matrix recognitionmolecule Jl/tenascin was investigated in the crypt-villus unit of the adult mouse ileum by immunoelectronmicroscopic techniques. In the villus region, Jl/tenascinwas detected strongly in the extracellular matrix (ECM)between fibroblasts of the lamina propria. It was gener-ally absent in the ECM at the interface between sub-epithelial fibroblasts and intestinal epithelium, exceptfor some restricted areas along the epithelial basallamina of villi, but not of crypts. These restricted areascorresponded approximately to the basal part of oneepithelial cell. In Jl/tenascin-positive areas, epithelialcells contacted the basal lamina with numerous microvil-lus-like processes, whereas in Jl/tenascin-negative areasthe basal surface membranes of epithelial cells contactedtheir basal lamina in a smooth and continuous appo-sition. In order to characterize the functional role ofJl/tenascin in the interaction between epithelial cellsand ECM, the intestinal epithelial cell line HT-29 wastested for its ability to adhere to different ECM com-ponents. Cells adhered to substratum-immobilized fi-

bronectin, laminin and collagen types I to IV, but not toJl/tenascin. When laminin or collagen types I to FVwere mixed with Jl/tenascin, cell adhesion was aseffective as without Jl/tenascin. However, adhesion wascompletely abolished when cells were offered a mixtureof fibronectin and Jl/tenascin as substratum. The abi-lity of Jl/tenascin to reduce the adhesion of intestinalepithelial cells to their fibronectin-containing basal lam-ina suggests that Jl/tenascin may be involved in theprocess of physiological cell shedding from the villus.

Abbreviations: BSA; bovine serum albumin. DMEM;Dulbecco's modified Eagle's medium. ECM; extracellularmatrix. EDTA; ethylenediaminetetra-acetic acid. HBSS;Hank's balanced salt solution. PBS; phosphate-bufferedsaline. TBS; Tris-(hydroxymethyl)-aminomethane bufferedsaline.

Key words: epithelium, HT-29 cell line, Jl/tenascin, smallintestine.

Introduction

In the crypt-villus unit of the small intestine of adultvertebrates, epithelial cells are continuously generatedfrom stem cells near the crypt base, migrate towards thevillus tip, and are finally extruded into the intestinallumen (Cheng and Leblond, 1974). To our knowledge,no detailed information is available on the molecularbasis of this shedding process. It has been suggested byQuaroni and Trelstad (1980) that several types ofcollagen, synthesized in submaximally hydroxylatedform by intestinal epithelial cells, could influence anddecrease the binding of these cells to their basal lamina.However, no experimental systems have been devel-oped to test this hypothesis.

We have shown previously by immunohistology atthe light microscopic level that Jl/tenascin is expressedat the boundary between epithelial cells and laminapropria in an increasing gradient towards the villus tip

(Thor et al. 1987; Probstmeier et al. submitted forpublication). This expression pattern led us to hypoth-esize that Jl/tenascin may be involved in the physio-logical process of epithelial cell shedding. Here wepresent data that strongly suggest that Jl/tenascin playssuch a functional role. To this end, we have investigatedthe expression of Jl/tenascin in the crypt-villus unit byimmunoelectron microscopy and developed a cell-substratum adhesion assay to determine the adhesion ofthe intestinal epithelial cell line HT-29 to several ECMcomponents in the presence or absence of Jl/tenascin.

Materials and methods

AnimalsC57BL/6J and NMRI mice were maintained at the depart-mental animal facility.

314 R. Probstmeier, R. Martini and M. Schachner

ECM moleculesFibronectin was prepared from human plasma according tomethod A described by Miekka et al. (1982). Laminin wasisolated from the EHS mouse tumor (Timpl et al. 1979).Isolation of Jl/tenascin from cultures of embryonic mousefibroblasts has been described (Faissner et al. 1988; Probst-meier etal. 1990). Collagen type I was extracted from rat skin(Bornstein and Piez, 1966); type II from chick cartilage (vonder Mark et al. 1982); type III from fetal calf skin (Timpl et al.1981) and type IV from EHS mouse tumor or human placentaafter mild pepsin treatment, whereby the carboxyterminalNCI domain was destroyed (Timpl et al. 1981).

AntibodiesThe following antibodies were used: polyclonal antibodies tolaminin and fibronectin (Pesheva et al. 1988, 1989; a kind giftof P. Pesheva) and polyclonal antibodies to Jl/tenascin fromearly postnatal mouse brain (Faissner et al. 1988; a kind gift ofA. Faissner).

Cell cultures and cell-substratum adhesion assayThe epithelial cell line HT-29 derived from a human colonadenocarcinoma (Fogh and Trempe, 1975) was obtained fromE. Coudrier (Unit de Biologie des Membranes, InstitutPasteur, Paris, France). In the presence of high amounts ofglucose in the culture medium, HT-29 cells are in a morpho-logically and biochemically undifferentiated state. Whenglucose is replaced by galactose, they differentiate into cellswith enterocytic phenotype (Pinto et al. 1982; Chastre et al.1985; Zweibaum et al. 1985). HT-29 cells were grown inglucose- or galactose-containing DMEM as described (Huet etal. 1987). Glucose-containing DMEM was supplemented with10% fetal calf serum (dialyzed for three days at 4°C against0.9% NaCl), 20 mM glucose (in addition to the 5mM alreadypresent in DMEM), 10/igml"1 human transferrin, 2mMglutamine, penicillin (lOOi.u. ml"1) and streptomycin(100/igmr1). Galactose-containing DMEM (glucose- andpyruvate-free, from Serva, Heidelberg, FRG) was sup-plemented with 5mM galactose. For the adhesion assay,monolayer cultures were treated for 5 min at room tempera-ture with trypsin/EDTA (0.05 %/0.02%; w/v) in Ca2+- andMg2+-free HBSS. The cells were then collected by centrifu-gation, washed in Ca2+- and Mg2+-free HBSS and resus-pended in glucose- or galactose-containing DMEM.

Fibronectin, laminin and collagen types I to FV (at a finalconcentration of 50/igmF1 in PBS) were mixed with J l /tenascin or heat-inactivated BSA (30 to 65^gmr' in PBS),spotted (4 jA per spot) in plastic Petri dishes (5 cm in diameter,tissue culture grade, from NUNC) and incubated for 1.5 h at37°C in a humid atmosphere. Petri dishes were washed oncewith PBS containing 10% heat-inactivated BSA, blockedwith the same solution for 1 h at 37°C, and washed three timeswith glucose- or galactose-containing DMEM. Single cellsuspensions of glucose- or galactose-adapted HT29 cells wereapplied (4xlCr cells in 4 ml of the corresponding culturemedium) and maintained for 5 h at 37°C. Dishes were washedthree times with the corresponding culture medium and cellsadhering to the ECM components were monitored micro-scopically.

Determination of coating efficiency by the immunodotbinding assayCoating efficiencies to the plastic surface of Petri dishes weremeasured for the ECM component by the immunodot bindingassay. Petri dishes coated with ECM components as describedfor the cell-substratum adhesion assay were washed three

times with TBS containing 0.1% BSA (TBS/BSA) andincubated overnight at 4°C with polyclonal rabbit antibodiesto laminin, fibronectin or Jl/tenascin diluted in TBS/BSA.Dishes were washed three times with TBS/BSA, incubatedfor lh at room temperature with horseradish peroxidase-coupled antibody to rabbit immunoglobulin, and washedthree times in TBS/BSA. Bound antibodies were visualizedas described (Faissner et al. 1984). Coating efficiencies forfibronectin or laminin were not influenced by the presence ofJl/tenascin. Furthermore, coating efficiencies for Jl/tenascinwere the same in mixtures with fibronectin, laminin orcollagen types I to IV.

Immunocytology1. Immunofluorescence

Adult mice were anesthetized by intraperitoneal injection ofan aqueous solution of chloralhydrate (3%, 0.01 ml g~' bodyweight) and perfused with 4% formaldehyde in PBS. Ileawere then removed, cut into pieces, 3 mm in length, and post-fixed for 2h at 4°C. Tissue pieces were washed consecutivelyfor lh in PBS, for 24h in 10% sucrose in PBS containing0.01 % NaN3, and for 48 h in 20 % sucrose in PBS containing0.01 % NaN3, all at 4°C. Tissue pieces were then embedded inOCT (Jung, Nussloch) and shock-frozen in 2-methylbutane.Sections, 10/m\ thick, were prepared in a Frigocut 2700 (Jung,Nussloch), mounted on poly-L-lysine-coated glass coverslips,and dried for 60 min at room temperature. Sections were thenincubated for 15 min at room temperature in blocking buffer(PBS containing 1% horse serum and 0.1% BSA, pH7.4),washed once for 5 min in PBS containing 0.1% BSA, incu-bated for 20min with polyclonal antibodies to Jl/tenascin(diluted 1:100 in blocking buffer), washed three times for5 min in PBS containing 0.1% BSA, incubated with FJTC-conjugated antibodies to rabbit IgG (diluted 1:100 in blockingbuffer), washed again three times, and embedded in glycerol/PBS (1:1, v/v).

2. Immunoelectron microscopyMice were perfused with fixative (4% formaldehyde, 0.25%glutaraldehyde in Palay's buffer; Palay and Chan-Palay, 1974)as described (Martini and Schachner, 1986). Ilea wereremoved, dissected into pieces, 3 mm in length, and post-fixedin 4% formaldehyde for 12 h at 4°C. The tissue was pro-cessed, incubated with antibodies and prepared for immuno-electron microscopy as described (Martini and Schachner,1986).

Results

Immunohistological localization of Jl /tenascin in theadult mouse ileum

1. Light microscopyThe localization of Jl/tenascin was investigated incrypts and villi of cross-sections of the ileum (Fig. 1) byindirect immunofluorescence. Jl/tenascin was stronglydetectable in the lamina propria of villi, particularly atthe interface between mesenchyme and epithelium(arrows in Fig. 2A,B)- In crypts Jl/tenascin was alsoaccumulated at the epithelium-mesenchyme interface,but its expression was considerably weaker when com-pared to the villus region (Fig. 2C, D).

2. Electron microscopyTo investigate in more detail which structures express

Jl/tenascin in mouse small intestine 315

Fig. 1. Schematic diagram of the crypt-villus axis in theadult mouse small intestine. In cross-sectioned villi (A) thering-like epithelium is surrounded by the intestinal lumenand contains in its core the lamina propria. In cross-sectioned crypts (B), the ring-like epithelium is surroundedby the lamina propria and contains in its core the intestinallumen. The rectangles A and B indicate the areas fromwhich immunoelectron micrographs are shown in Figs 3Aand B. The intestinal epithelial sheath is marked by dots.EM, epithelium-mesenchyme interface; LP, lamina propria.

Jl/tenascin in the crypt-villus axis of the adult mouseileum, an immunoelectron microscopic characterizationwas earned out using pre-embedding staining pro-cedures. Representative areas of cross-sectioned crypt-and villi-regions were cut as indicated in the diagram ofFig. 1 and are shown in Fig. 3.

Jl/tenascin was scarcely detectable in the ECM ofthe lamina propria in the villus core. Also in the cryptregion, Jl/tenascin was only weakly detectable in thelamina propria. Only in the vicinity of small nerve fiberswere glial cells surrounded by strongly Jl/tenascin-positive ECM (Fig. 3A, inset). Jl/tenascin was alsostrongly expressed in the villus region, in associationwith ECM surrounding subepithelial fibroblasts of thelamina propria (Fig. 3A). Jl/tenascin was specificallylocalized, in that it was strongly detectable in the ECMbetween neighbouring fibroblasts, but was generallyabsent in the basal lamina at the interface betweenfibroblasts and epithelium. This polarized expressionwas most clearly seen when thin processes of subepi-thelial fibroblasts were oriented in parallel to the basal

lamina of the epithelium, since, here, Jl/tenascin-negative basal lamina was in close vicinity to J l /tenascin-positive areas separated only by slender fibro-blastic processes (Fig. 3A, asterisks). However, insome confined areas (approximately 3 to 5^m inlength), Jl/tenascin was detectable also in the basallamina of the epithelial cell-fibroblast interface(Fig. 3A, large arrowheads). Jl/tenascin-positive zoneswere separated from each other by a stretch of 5 to50 /zm of Jl/tenascin-free basal lamina. In approxi-mately 80% of the Jl/tenascin-positive stretches, thebasal parts of the epithelial cells differed in theirmorphology from those apposed to Jl/tenascin-negative basal lamina. Basal parts of epithelial cellstouching Jl/tenascin-negative basal lamina madesmooth, uninterrupted and non-convoluted contactswith the basal lamina (Fig. 3A, epithelial cell e). Incontrast, basal parts of epithelial cells apposed toJl/tenascin-positive basal lamina formed numerousmicrovillus-like processes (Fig. 3A; epithelial cell e').

In contrast to the villus region, no Jl/tenascin-positivity was found in the crypt region associated withthe ECM in the basal lamina at the basal parts ofepithelial cells (Fig. 3B). Basal parts of epithelial cryptcells always engaged in smooth and uninterruptedapposition to the basal lamina, never showing microvil-lus-like protrusions (Fig. 3B).

Substratum adhesion of epithelial cells in the absenceor presence of Jl/tenascinSince expression of Jl/tenascin in the epithelial basallamina was correlated with an apparently disturbedcontact between epithelial cells and basal lamina, afunctional assay system was used to investigate ad-hesion of intestinal epithelial cells to different ECMcomponents in the absence or presence of Jl/tenascin.In initial experiments, we used freshly isolated cryptand villus cells from adult mouse small intestine whichcan be separately obtained by incubation of intestinaltubes with an EDTA-containing buffer for differenttime periods (Weiser, 1973; Thor et al. 1987). Unfortu-nately, these cells did not adhere to any of the substrataoffered in our adhesion assay (collagen types I to IV,laminin and fibronectin), probably due to the limitedviability of the cells over time periods of several hours(for references, see Quaroni and May, 1980). Thus,while these cells could be successsfully used for cellaggregation assays of less than one hour (Thor et al.1987), they could not be taken for cell-substratumadhesion assays with duration of five hours. Fortuna-tely, an epithelial cell line (HT-29) is available whichcan be obtained in morphologically undifferentiatedand differentiated states. In glucose-containing me-dium, this cell line exhibits predominantly the featuresof undifferentiated epithelial cells (HT-29 glu cells),while in galactose-containing medium it differentiatesinto cells with properties of villus epithelial cells (HT-29gal cells) (Zweibaum et al. 1985).

When fibronectin, laminin, collagen types I to IV, orJl/tenascin were offered as substratum for cell attach-ment (50 fig ml"1 in the solution used for coating), HT-


Fig. 2. Immunohistological localization of Jl/tenascin in cross-sections of the villus (A, B) and crypt (C, D) regions of adultmouse ileum by indirect imunofluorescence. Arrows indicate the interface between epithelium and mesenchyme. B and Dare the corresponding phase contrast micrographs to fluorescence images A and C, respectively. Bar in D represents 25 jum(A-D).

29 glu cells adhered approximately equally well tolaminin, fibronectin and collagen types I to IV, but notto Jl/tenascin (Fig. 4A,C,E,H and Table 1). Whenmixtures of Jl/tenascin (coating concentration65 /ig ml"1) with fibronectin, laminin or collagen types Ito IV (coating concentration 50/zgmP1) were offeredas substratum, adhesion of HT-29 glu cells to themixture of fibronectin with Jl/tenascin was completelyabolished (Fig. 4G), while substrata containing themixtures of Jl/tenascin with laminin or collagen types Ito IV did not affect cell adhesion (Figs4B,D andTable 1). In the latter combinations, binding was com-parable to that of mixtures where Jl/tenascin wasreplaced by heat-inactivated BSA. The inhibitory effectof the mixture of Jl/tenascin and fibronectin on theadhesion of HT-29 cells was Jl/tenascin-concentration-dependent, in that adhesion of HT-29 cells to themixture was increased with a lower concentration ofJl/tenascin (coating concentration 30 /igmF1; compareFig. 4F and G). No differences in adhesion wereobserved between HT-29 glu or HT-29 gal cells (Fig. 41to L and compare to A to H), except that HT-29 galcells bound less specifically to heat-inactivated BSAthan HT-29 glu cells.

When the RGDS-peptide was added at a concen-tration of 100 ̂ g ml" to the culture medium prior toplating of HT-29 glu or gal cells, adhesion to fibronec-tin, but not to laminin or collagen types I to IV, wasreduced (Table 1), indicating that HT-29 cells interactwith fibronectin predominantly via a RGD-dependentfibronectin receptor.

Results obtained in the cell-substratum adhesionassays are summarized in Table 1.

Discussion

Proteins and glycoproteins of the ECM have beenmainly investigated for their role as adhesive cellsubstrata. It has recently been shown, however, thatJl/tenascin is a poor substratum for fibroblasts and,when mixed with fibronectin, inhibits fibronectin-de-pendent cell spreading, while attachment of fibroblastsis not altered (Chiquet-Ehrismann et al. 1988). Theseobservations together with previous ones from ourlaboratory, showing, by immunohistology at the lightmicroscopic level, an increasing gradient of Jl/tenascinexpression from crypt base to villus tip (Thor et al. 1987;


• # - • . *

r>

^»• • »

v

3A B

Fig. 3. lmmunoelectron microscopic localization of Jl/tenascin in the villus (A) and crypt (B) regions of the adult mouseileum. lmmunoelectron microscopy was carried out by pre-embedding staining procedures using polyclonal antibodies toJl/tenascin and protein A coupled to horseradish peroxidase. (A) Jl/tenascin immunoreactivity is present in the ECMaround fibroblast-like cells of the lamina propria of the upper villus region (arrows). Note that only a small portion of thebasal lamina of the epithelium is Jl/tenascin-positive (large arrowheads). In this region, the epithelial cell (e') formsmicrovillus-like processes touching the basal lamina. Asterisks mark slender processes of fibroblasts which are in contactwith Jl/tenascin-positive ECM on one side and with the Jl/tenascin-negative basal lamina on the other (small arrowheads).Epithelial cell (e); subepithelial fibroblasts of the lamina propria (f). Inset: Small nerve in the lamina propria of the villusregion. Note strong immunoreactivity associated with ECM around the glial cell processes of the nerve (arrowhead). Axon(a); glial cell (g). (B) In the crypt region, subepithelial basal laminae (arrowheads) and collagen fibrils (arrows) do not showdetectable levels of Jl/tenascin immunoreactivity. Crypts were cut in cross-section, so that the basal parts of epithelial cells(e) of neighbouring crypts are close to each other. Subepithelial fibroblasts of the lamina propria (f). Bars: 0.5 ̂ m.


Fig. 4. Adhesion of HT-29 cells grown in glucose- (A-H) or galactose-containing medium (I-L) to different ECMcomponents and their mixtures with Jl/tenascin (Jl) coated on plastic. (A) laminin plus BSA; (B,K) laminin plus Jl;(C) collagen type III plus BSA; (D,L) collagen type III plus Jl; (E,I) fibronectin plus BSA; (F) fibronectin plus Jl(30/igml""1); (G,J) fibronectin plus Jl; (H) Jl. If not indicated otherwise, the concentrations used for coating were:50j/gml~' for laminin, fibronectin, and collagen type III and 65;;gml~' for heat-inactivated BSA and Jl/tenascin. Bar in Arepresents 500/im (A-L).

Probstmeier et al. submitted for publication), led us tostudy the expression of Jl/tenascin at the mesenchy-mal-epithelial interface at the electron microscopiclevel. In addition, we have analysed the substratumproperties of Jl/tenascin for intestinal epithelial cells.

Our study has shown that intestinal epithelial HT-29cells adhere well to laminin, fibronectin and collagentypes I to IV, but not to Jl/tenascin. In mixtures withthese ECM components, Jl/tenascin inhibited the at-tachment of HT-29 cells to fibronectin, while thepresence of Jl/tenascin in mixtures with laminin andcollagen types I to IV substrata did not influence theiradhesive properties. These observations can be in-terpreted in several ways. On the one hand, there maybe no cell surface receptor for Jl/tenascin and J l /tenascin may block the RGDS-dependent or other cell

binding sites for HT-29 cells on fibronectin, while itdoes not interfere with the cell binding site on lamininor collagen types I to IV. On the other hand, there maybe a cell surface receptor that allows recognitionbetween the cell surface and Jl/tenascin, but results inrepulsion of cells. In conjunction with fibronectin,Jl/tenascin might thus interfere with HT-29 cell bindingby allowing interaction of fibronectin with its cellsurface receptor, but annihilating fibronectin-mediatedadhesion. In this scheme, the Jl/tenascin receptorwould not annihilate laminin- or collagen-mediatedadhesion. These differential responses may result fromthe balance of opposing intracellular consequences ofrecognition between cell surface and substratum. Incombination with laminin or different types of collagen,the Jl/tenascin receptor could trigger intracellular


Table 1. Binding of HT-29 cells to Jl/tenascin or tomixtures of Jl/tenascin with other extracellular matrix

components

Protein(s) used as

Collagen type I

Collagen type II

Collagen type III

Collagen type IV

Laminin

Fibronectin

Jl/tenascinBSA

substratum

Plus BSAPlus JlPlus BSAPlus JlPlus BSAPlus JlPlus BSAPlusJlPlus BSAPlus JlPlus BSAPlus Jl

Adhesion

+ RGDS*

46411134711486251426101573371503451206031327101109513148476111478116n.d.n.d.

18110

of HT-29 cells

-RGDS

503138493159584137580167376178298161717168736173471197425116532914918182411621113

Proteins were immobilized on plastic and offered as substratumfor HT-29 glu cells. Protein concentrations used for coating areindicated in the legend to Fig. 4. For estimation of the number ofcells adhering to the different substrata, cells in micrographs ofsubstratum spots were counted in microscopic fields correspondingto 500[im2. Values represent mean valueslstandard deviation ofadherent cells in 4 to 6 different microscopic fields from 3independent experiments. Comparable results were obtained withHT-29 cells grown in glucose- or galactose-containing medium,n.d., not done.

* RGDS was present in the culture medium at a concentration of100/Jgmr1.

signals that would not override the adhesion-promotingsignals triggered by laminin or the collagens. However,when the Jl/tenascin receptor would be activated inconjunction with the fibronectin receptor, their additiveeffects would result in cell repulsion. This repulsiveaction of Jl/tenascin could be mediated by an integrin-independent cell surface receptor, but may possibly alsoact by triggering the fibronectin-dependent integrinreceptor; that is Jl/tenascin would attach to a domaindifferent from the RGDS-dependent fibronectin-bind-ing domain. That such a mechanism might exist stemsfrom the report that the hexabrachion complex, ident-ical to Jl/tenascin, binds to an integrin-like molecule(Bourdo'n and Ruoslahti, 1989). However, hexabra-chion binding to the integrin-like receptor could beshown to be RGDS-dependent, thus contradicting pre-vious reports on the RGDS-independence of tenascinaction (Chiquet-Ehrismann et al. 1988). Since tenascinhas been shown to bind directly to fibronectin, but notto laminin or the RGDS peptide (Chiquet-Ehrismann etal. 1988), it is also conceivable that the Jl/tenascin-fibronectin complex induces allosteric changes in fibro-nectin leading to an inability of fibronectin for cellbinding or to a modification of the action on thefibronectin receptor itself. Functional connections be-tween adhesion molecules and the cytoskeleton orintracellular second messenger systems have been de-scribed (Horwitz et al. 1986; Pollerbergef a/. 1986,1987;Schuch et al. 1989). The intracellular consequences ofsuch ligand-receptor interactions could vary in differ-

ent cell types, preventing the formulation of a generalscheme about influences of Jl/tenascin on the adhesi-vity of cells to different substrata. In fact, it has beenshown that cells that adhere to Jl/tenascin, such asepithelial, mesenchymal and glial cells (Bourdon andRuoslahti, 1989; Chiquet-Ehrismann et al. 1988), can bedistinguished from cells that do not, such as melanoma,F9 embryonic carcinoma (Bourdon and Ruoslahti,1989) and HT-29 cells. The combined observations urgefor a detailed analysis of the molecular mechanismsinvolved in recognition and adhesion on one hand andrecognition and de-adhesion or repulsion on the other.

The question now is how does the de-adhesivecomplex of Jl/tenascin with fibronectin affect thefunction of the crypt-villus unit? It is known thatintestinal epithelial cells express fibronectin receptors(Chen et al. 1985) and that fibronectin and Jl/tenascindisplay inverse gradients in immunofluorescenceintensities from the villus bottom to the villus tip, withJl/tenascin being increased and fibronectin being de-creased towards the villus tip (Thor et al. 1987; Probst-meier et al. submitted for publication). Furthermore,both molecules appear to be coexpressed in the basallamina between epithelium and mesenchyme (for fibro-nectin, see Laurie et al. 1982). The immunoelectronmicroscopic study presented in this paper shows alocally restricted expression of Jl/tenascin in the basallamina of epithelial villus cells. There was a rough,more convoluted appearance of the surface membraneof the basal parts of some villus cells towards theJl/tenascin-positive basal lamina. It is possible that thisintermittent contact between basal surface membraneand basal lamina represents de-adhesion, thus possiblyimplicating these Jl/tenascin-associated cells in theshedding process. This would mean that epithelial cellsare not shed from the villus in clusters, but as individualcells by their loss of contact with the basal lamina.Furthermore, the shedding process may be promotedby the absence of adhesion molecules at lateral cellcontact sites, such as the cell adhesion molecule LI,which is expressed by the proliferating epithelial cells inthe crypt region, but is no more detectable on epithelialcells in the villus region (Thor et al. 1987; Probstmeier etal. submitted for publication). Thus, an interplay be-tween the absence of adhesive forces contributed, forinstance, by LI, and the loss of adhesive forces con-tributed by repulsive molecules, such as thefibronectin-Jl/tenascin pair, could result in the overallshedding process. It has been observed elsewhere thatnidogen and laminin is another such pair in that nidogencan interfere with cell binding to laminin (Aumaiiley etal. 1987).

At present, it is not known how the gradient ofJl/tenascin expression is formed along the crypt-villusaxis and how only discrete patches in the basal laminabecome Jl/tenascin-positive. It is likely that J l /tenascin is synthesized by intestinal mesenchymal cells,since the intestinal mesenchyme has been shown to beinduced by intestinal epithelium to synthesize J l /tenascin, while no Jl/tenascin synthesis was detected inepithelial cells alone (Aufderheide and Ekblom, 1988).


In agreement, in intestinal explant cultures (Kondo etal. 1984), Jl/tenascin was observed exclusively in as-sociation with mesenchymal or glial cells, but neverwith epithelial cells (Probstmeier, unpublished obser-vations). The mesenchymal fibroblasts underlying theepithelial basal lamina in the small intestine form a celllayer that migrates from the crypt bottom to the villustip (Marsh and Trier, 1974a). These fibroblasts prolifer-ate in the crypt region and cease to divide in the villusregion (Marsh and Trier, 1974b). It is thus possible thatthe cessation of proliferation of these cells is one of thefactors that contributes to a more stochastic expressionof Jl/tenascin by these cells. However, Gatchalian andcoworkers (1989) have shown that fibroblasts derivedfrom perisynaptic and non-synaptic regions of dener-vated muscle express Jl/tenascin when proliferating.Thus, it is probably not a general phenomenon thatexpression of high amounts of Jl/tenascin is correlatedwith and/or implicated in cell proliferation. Theseobservations, together with others (Martini et al. 1990;Steindler et al. 1989a,b; Stern et al. 1989), suggest thatthe regulatory events underlying the expression ofJl/tenascin and its isoforms may be many fold anddepend on the cell type (Probstmeier et al. 1990), as hasbeen suggested for the intracellular signalling pathwaysthat are triggered in response to cell recognition.

The authors are grateful to Andrea Spitzauer and LuitgardFreidel for excellent technical assistance, P. Pesheva forantibodies, A. Faissner for antibodies and Jl/tenascin, K.Kiihn for collagens, E. Coudrier for HT-29 cells, andDeutsche Forschungsgemeinschaft (postdoctoral fellowshipto R.P. and SFB 317) for support. M.S. i sa member of theSwiss Federal Institute of Technology Zurich.

References

AUFDERHEIDE, E. AND EKBLOM, P. (1988). Tenascin during gutdevelopment: appearence in the mesenchyme, shift in molecularforms, and dependence on epithelial-mesenchymal interactions.J. Cell Biol. 107, 2341-2349.

AUMAILLEY, M . , NURCOMBE, V . , EDGAR, D . , P A U L S S O N , M . AND

TIMPL, R. (1987). The cellular interactions of laminin fragments.J. biol. Chem. 262, 11532-11538.

BORNSTEIN, P. AND PIEZ, K. (1966). The nature of intramolecularcross links in collagen. The separation and characterization ofpeptides from the alpha 1 and alpha 2 chains. Biochemistry 5,3460-3473.

BOURDON, M. A. AND RUOSLAHTI, E. (1989). Tenascin mediates cellattachment through an RGD-dependent receptor. J. Cell Biol.108, 1149-1155.

CHASTRE, E., EMAMI, S., ROSSELIN, G. AND GESPACH, C. (1985).

Vasoactive intestinal peptide receptor activity and specificityduring enterocyte-like differentiation and retrodifferentiation ofthe human colonic cancerous subclone HT29-18. FEBS (Fed,Eur. Biochem. Soc.) Lett. 188, 197-204.

CHEN, W. T., GREVE, J. M., GOTTLIEB, D. J. AND SINGES, S. J.

(1985). Immunocytochemical localization of 140 kD cell adhesionmolecules in cultured chick fibroblasts, and in chicken smoothmuscle and intestinal epithelial tissues. J. Histochem. Cytochem.33, 576-586.

CHENG, H. AND LEBLOND, C. P. (1974). Origin, differentiation andrenewal of the four main epithelial cell types in the mouse smallintestine. I. Columnar cells. Am. J. Anat. 141, 461-480.

CHIQUET-EHRISMANN, R., KALLA, P., PEARSON, C. A., BECK, K.

AND CHIQUET, M. (1988). Tenascin interferes with fibronectinaction. Cell 53, 383-390.

FAISSNER, A., KRUSE, J., CHIQUET-EHRISMANN, R. AND MACKIE, E.

R. (1988). The high-molecular weight J l glycoproteins areimmunochemically related to tenascin. Differentiation 37,104-114.

FAISSNER, A., KRUSE, J., GORIDIS, C , BOCK, E. AND SCHACHNER,

M. (1984). The neural cell adhesion molecule LI is distinct fromthe N-CAM related group of surface antigens BSP-2 and D2.EMBO J. 3, 733-737.

FOGH, J. AND TREMPE, G. (1975). New human tumor cell lines. InHuman Tumor Cells in vitro (ed. J. Fogh). Plenum PublishingCorp., New York. 115-141.

GATCHAUAN, C. L., SCHACHNER, M. AND SANES, J. R. (1989).

Fibroblasts that proliferate near denervated synaptic sites inskeletal muscle synthesize the adhesive molecules tenascin (Jl) ,N-CAM, fibronectin and a heparan sulfate proteoglycan. J. CellBiol. 108, 1873-1890.

HORWTTZ, A., DUGGEN, K., BUCK, C , BECKERLE, M. C. AND

BURRIDCE, K. (1986). Interaction of plasma membranefibronectin receptor with talin - a transmembrane linkage.Nature 320, 531-533.

HUET, C , SAHUQUILLO-MERINO, C , COUDRIER, E. AND LOUVARD,

D. (1987). Absorptive and mucus-secreting subclones isolatedfrom a multipotent intestinal cell line (HT-29) provide newmodels for cell polarity and terminal differentiation. J. Cell Biol.105, 345-357.

KONDO, Y., ROSE, I., YOUNG, G. P. AND WHITEHEAD, R. H.

(1984). Growth and differentiation of fetal rat small intestinalepithelium in tissue culture. Expl Cell Res. 153, 121-134.

LAURIE, G. W., LEBLAND, C. P. AND MARTIN, G. R. (1982).

Localization of type IV collagen, laminin, heparan sulfateproteoglycan, and fibronectin to the basal lamina of basementmembranes. J. Cell Biol. 95, 340-344.

MARSH, M. N. AND TRIER, J. S. (1974a). Morphology and cellproliferation of subepithelial fibroblasts in adult mouse jejunum.I. Structural features. Castroenterology 67, 622-635.

MARSH, M. N. AND TRIER, J. S. (19746). Morphology and cellproliferation of subepithelial fibroblasts in adult mouse jejunum.II. Radioautographic studies. Gastroenterology 67, 636-645.

MARTINI, R. AND SCHACHNER, M. (1986). Immunoelectronmicroscopic localization of neural cell adhesion molecules (LI,N-CAM, MAG) and their shared carbohydrate epitope andmyelin basic protein (MBP) in developing sciatic nerve. J. CellBiol. 103, 2439-2448.

MARTINI, R., SCHACHNER, M. AND FAISSNER, A. (1990). Enhanced

expression of the extracellular matrix molecule Jl/tenascin in theregenerating adult mouse sciatic nerve. J. Neurocytol. (in press).

MIEKKA, S. J., INGHAM, D. C. AND MENACHE, D. (1982). Rapid

methods for isolation of human plasma fibronectin. Thromb.Res. 27, 1-14.

PALAY, S. L. AND CHAN-PALAY, V. (1974). Cerebellar Cortex.Cvtology and Organization. Springer Verlag, New York,Heidelberg, Berlin. 348 pp.

PESHEVA, P., JUUANO, R. L. AND SCHACHNER, M. (1988).

Expression and localization of the fibronectin receptor in themouse nervous system. J. Neurosci. Res. 20, 420-430.

PESHEVA, P., SPIESS, E. AND SCHACHNER, M. (1989). Jl-160 and

Jl-180 are oligodendrocyte-secreted nonpermissive substrates forcell adhesion. J. Cell Biol. 109, 1765-1778.

PINTO, M., APPAY, M. D., SIMON-ASSMANN, P., CHEVALIER, G.,

DRACOPOU, N., FOGH, J. AND ZWEIBAUM, A. (1982). Enterocytic

differentiation of cultured human colon cancer cells byreplacement of glucose by galactose in the medium. Biol. Cell.44, 193-196.

POLLERBERG, E . , BURRIDGE, K . , KREBS, K . , GOODMAN, S. AND

SCHACHNER, M. (1987). The 180 kD component of the neural celladhesion molecule N-CAM is involved in cell-cell contacts andcytoskeleton-membrane interactions. Cell Tiss. Res. 250,227-236.

POLLERBERG, E . , SCHACHNER, M . AND D A V O U S T , J. ( 1 9 8 6 ) .

Differentiation-state dependent surface mobilities of two formsof the neural cell adhesion molecule. Nature 324, 462-465.

PROBSTMEIER, R., KOHN, K. AND SCHACHNER, M. (1990).

Fibroblast-derived J l adhesion glycoproteins show binding


properties to extracellular matrix constituents different fromthose of central nervous system origin. J. Neurochem. 54,1016-1019.

QUARONI, A. AND MAY, R. J. (1980). Establishment andcharacterization of intestinal epithelial cell cultures. Meth. CellBiol. 21,403-427.

QUARONI, A. AND TRELSTAD, R. (1980). Biochemicalcharacterization of collagens synthesized by intestinal epithelialcell cultures. J. biol. Chem. 255, 8351-8361.

SCHUCH, U., LOHSE, M. J. AND SCHACHNER, M. (1989). Neural celladhesion molecules influence second messenger systems. Neuron3, 13-20.

STEINDLER, D. A., COOPER, N. G. F., FAISSNER, A. AND

SCHACHNER, M. (1989a). Boundaries defined by adhesionmolecules during development of the cerebral cortex: TheJl/tenascin glycoprotein in the mouse somatosensory corticalbarrel field. Devi Biol. 131, 243-260.

STEINDLER, D. A., FAISSNER, A. AND SCHACHNER, M. (1989ft).

Brain 'cordones': Transient boundaries of glia and adhesionmolecules that define developing functional units. Comments devlBiol. 1,29-60.

STERN, C. D., NORRIS, W. E., BRONNER-FRASER, M., CARLSON, G.

J., FAISSNER, A., KEYNES, R. J. AND SCHACHNER, M. (1989).

Jl/tenascin-related molecules are not responsible for thesegmented pattern of neural crest cells or motor axons in thechick embryo. Development 107, 1-11.

THOR, G., PROBSTMEIER, R. AND SCHACHNER, M. (1987).

Characterization of the cell adhesion molecules LI, N-CAM andJl in the mouse intestine. EMBO J. 6, 2581-2586.

TlMPL, R . , ROHDE, H . , ROBEY, P . G . , RENNARD, S. J . , FOIDART, J.M. AND MARTIN, G. R. (1979). Laminin - a glycoprotein frombasement membranes. J. biol. Chem. 254, 9933-9937.

TlMPL, R., WlEDEMANN, H., VAN DELDEN, V. , FuRTHMAYR, H. AND

KOHN, K. (1981). A network model for the organization of typeIV collagen in basement membranes. Eur. J. Biochem. 120,203-211.

VON DER MARK, K., VAN MENXEL, M. AND WIEDEMANN, H. (1982).

Isolation and characterization of new collagens from chickcartilage. Eur. J. Biochem. 124, 57-62.

WEISER, M. M. (1973). Intestinal epithelial cell surface membraneglycoprotein synthesis. I. An indicator of cellular differentiation.J. biol. Chem. 248, 2536-2541.

ZWEIBAUM, A., PINTO, M., CHEVAKIER, G., DUSSAUIX, E.,

TRIADOU, N., LACROIX, B., HAFFEN, K., BRUN, J. L. AND

ROUSSET, M. (1985). Enterocytic differentiation of asubpopulation of the human tumor cell line HT-29 selected forgrowth in sugar-free medium and its inhibition by glucose. J. CellPhysiol. 122, 21-29.

{Accepted 2 March 1990)

expression of j1 tenascin in the crypt-villus unit of ... · jl/tenascin in mouse small intestine...

Documents