825Journal of Cell Science 111, 825-832 (1998)Printed in Great Britain © The Company of Biologists Limited 1998JCS9716

Expression of caveolin-1 and polarized formation of invaginated caveolae in

Caco-2 and MDCK II cells

Ulla Vogel 1,*, Kirsten Sandvig 2 and Bo van Deurs 1,†

1Structural Cell Biology Unit, Department of Medical Anatomy, The Panum Institute, Blegdamsvej 3, DK-2200 Copenhagen N,Denmark2Institute for Cancer Research at the Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway*Present address: Institute of Occupational Health, Lersø Parkalle 105, DK-2100 Copenhagen Ø, Denmark †Author for correspondence (e-mail: B.v.Deurs@mai.ku.dk)

Accepted 9 January 1998: published on WWW 23 February 1998

We have studied caveolin-1 expression and the frequencyand distribution of typical invaginated caveolae as they areidentified by electron microscopy in the polarized epithelialcell lines MDCK II and Caco-2. In wild-type MDCK II cellscaveolin expression is high and more than 400 caveolae/mmfilter were observed at the basolateral membrane. Nocaveolae were found at the apical surface. By contrast,wild-type Caco-2 cells do not express caveolin-1 and haveextremely few, if any caveolae. Caco-2 cells were stablytransfected with the gene for caveolin-1 in order toinvestigate if the formation of caveolae is polarized also inthese cells. We have isolated Caco-2 clones expressingdifferent levels of caveolin-1, where the level of expressionvaries from 10-100% of the endogenous level in MDCK IIcells. Caveolin-1 expression in Caco-2 cells gives rise to a

marked immunofluorescense labeling mainly at the lateralplasma membrane. By electron microscopy an increasefrom less than 4 caveolae/mm filter in wild-type Caco-2cells to 21-76 caveolae/mm filter in Caco-2 clonestransfected with caveolin-1 was revealed and these caveolaewere exclusively localized to the basolateral membrane.Thus expression of heterologous caveolin-1 in Caco-2 cellsleads to polarized formation of caveolae, but there is a lackof correlation between the amount of caveolin expressed inthe cells and the number of caveolae, suggesting thatfactors in addition to caveolin are required for generationof caveolae.

Key words: Caveolin expression, Polarized caveolae formation,MDCK II cell, Caco-2 cell

SUMMARY

fhe

he isa

,esf

sithl.,l.,d,mech

atal.to,t

INTRODUCTION

Caveolae are small, 50-80 nm omega-shaped invaginationthe plasma membrane defined on the basis of thmorphological appearance in electron micrograp(invaginated caveolae) (Devine et al., 1971; Forbes et al., 19Gabella, 1978; Parton, 1996; Parton and Simons, 1995; Paet al., 1997). They are not present in all cell types and thabundance varies greatly among cell types (Parton, 199Despite extensive research their function is still not cleEvidence has been presented that they are involved in a numof cellular processes such as signal transduction (Lisanti et1994), calcium signalling (Fujimoto, 1993), potocytos(Anderson et al., 1992) and endocytosis (Schnitzer et 1996). Furthermore, some glycosylphosphatidylinositol (GPlinked receptors like the urokinase plasminogen receptor (Sand Mueller, 1995), the folate receptor (Mayor et al., 199Rothberg et al., 1990; Turek et al., 1993) and the membrspanning tissue factor (Sevinsky et al., 1996), a key componof the coagulation cascade, have been reported to be locato caveolae or organized in microdomains associated wcaveolae (Schnitzer et al., 1995).

A major structural component of caveolae is the membra

s ofeirhs79;rtoneir6).

ar.ber

al.,isal.,I)-tahl4;

aneent

lizedith

ne

protein caveolin. Caveolin-1, which is found in a variety otissues, was both identified as a major component of tvesicular transport system in the trans-Golgi network (TGN)(Kurzchalia et al., 1992) and as a structural component of tcaveolar coat (Dupree et al., 1993; Rothberg et al., 1992). Ita 21 kDa integral membrane protein and is found as homooligomer in the cell (Monier et al., 1995). Howevernewly synthesized caveolin-1 forms oligomers in thendoplasmic reticulum (ER) indicating that oligomerization ia characteristic of caveolin-1 rather than an indication ocaveolae localization (Monier et al., 1995). Caveolin-1 iinserted in the plasma membrane as a hair pin structure wboth the N and the C terminal facing the cytosol (Monier et a1995), and it is a cholesterol binding protein (Murata et a1995). Recently, two more caveolins have been identifiecaveolin-2 (Scherer et al., 1996) which is expressed in the sacells as caveolin-1, and caveolin-3 (Tang et al., 1996), whiis expressed in muscle.

Little is known about caveolae biogenesis, except thcaveolin and cholesterol is required (Parton, 1996). Fra et (1995) have shown that introduction of canine caveolin-1 inhuman lymphocytes, which do not normally form caveolaewas sufficient for caveolar formation. Likewise, Engelman e

826

nd

,E.sionite

he18

y

tedere

ted.

s30s oron,oranarysy as

yMn

alS-

remallThellsn

%ne

orS,atc...

e

ged a cutidsect

U. Vogel, K. Sandvig and B. van Deurs

al. (1997) recently found that transfection of human cell linwith human caveolin resulted in the formation of caveolaHowever, it is unclear whether some factors are requiredaddition to caveolin. Such factors could be present in somenot in other cell types, and caveolin expression in itself mitherefore not necessarily lead to caveolae formation. In supof this, transfection of insect cells with caveolin-1 resultedaccumulation of caveolin-1 positive vesicles without leadingmorphologically defined caveolae at the plasma membraneet al., 1996).

In polarized epithelia it is an open question whether caveoformation is polarized. Since caveolin has been reported toinvolved in transport from the TGN to the apical surface polarized MDCK cells (Kurzchalia et al., 1992) and since GPanchored proteins, which are sorted to the apical surfacpolarized epithelial cells, have been proposed to associate caveolin (Lisanti et al., 1993; Sargiacomo et al., 1993; Zurzet al., 1994), it is often implied that caveolae must be presat this surface (Joliot et al., 1997; Lisanti et al., 1993).

In the present study we have therefore asked the followquestions: (1) are caveolae localized apically, basolaterallboth in the widely used canine cell line MDCK II whicnormally expresses caveolin-1? (2) does canine caveolexpression lead, as in human lymphocytes, to generatiocaveolae in polarized human Caco-2 cells, which normallynot express caveolin and have no caveolae (Mirre et al., 19and (3) are such caveolae formed at random or localized eat the apical or basolateral surface? We show here that caveare localized basolaterally in MDCK II cells, that caveolexpression in Caco-2 cells at a level comparable to that sin MDCK II cells leads to formation of caveolae exclusively the basolateral surface and that the number of newly formcaveolae is much lower than the number of caveolae in MDII cells. Furthermore, the amount of caveolin in the transfecCaco-2 clones did not correlate with their ability to forcaveolae. We therefore suggest that formation of caveolapolarized epithelia is a basolateral phenomenon and that facin addition to caveolin are involved.

MATERIALS AND METHODS

Cell culture Caco-2 and MDCK II cells were grown in T25 flasks (Nunc, RoskildDenmark), on glass coverslips or on Transwell filters (Costar; psize 0.4 µm, diameter 24.5 mm, cells seeded at a density of 106 perfilter) in DMEM (Dulbecco’s modified Eagle’s MEM), 10% or 5%fetal calf serum, respectively, 2 mM glutamine, 100 units/penicillin, 100 µg/ml streptomycin. Caco-2 cells were furthesupplemented with nonessential amino acids. Transfected Caclones were also supplemented with 0.65 mg/ml G418 (LTechnologies).

For inhibition of protein kinase C, the Caco-2 cells were incubain normal medium containing either 1 or 10 µM phorbol-12 myristate-13 acetate or 5 or 10 µM bisindolylmalemide for 24 hours beforeprocessing for electron microscopy (EM). Caco-2 cells were incubafor 3 days in growth medium containing 1.5 mM butyric acid investigate whether differentiation could induce caveolae formatio

Construction of expression vectorscDNA encoding caveolin/VIP21 isolated from MDCK II (Kurzchaliet al., 1992) inserted into the EcoRI-XhoI sites of pBluescript was agenerous gift from K. Simons, EMBL, Heidelberg. The 890 bp EcoRI-XhoI fragment encoding the cDNA was excised and inserted i

ese. in butghtport in to (Li

lae beinI-

e ofwitholoent

ingy orhin-1n of do96)?itherolae

ineenated

CKtedme intors

e,ore

mlrco-2ife

ted

tedton.

a

nto

pUHD10-3 (obtained from H. Bujard; Gossen and Bujard, 1992) aexcised as a EcoRI-BamHI fragment of similar size. This fragmentwas inserted into pTej4 or pTej8 resulting in pUV47 and pUV51respectively (both pTej4 and pTej8 were generous gifts from T. Johansen; Johansen et al., 1990). pTej4 is an eukaryote expresvector where expression of the gene inserted in the multicloning sis driven by the strong, constitutive promotor, pUbc (Johansen et al.,1990). pTej8 is similar to pTej4, except that it also encodes tneomycin phosphotransferase gene, giving resistance to G4(Johansen et al., 1990).

Caco-2 cells were either transfected with pUV47 alone or bcotransfection of pUV51 and pSV2-neo using the calciumprecipitation method as described (Vogel et al., 1995). The transfeccells were selected using 0.65 mg/ml G418, and stable clones wisolated after 3 weeks using cloning cylinders.

Western blotsCells were scraped off in PBS (phosphate buffer saline) and pelleThe pellet was suspended in 1 volume PBS and 1 volume 2× samplebuffer (Novex, San Diego) containing 100 mM DTT. The samplewere heated to 100°C for 5 minutes, or incubated at 25°C for minutes and subjected to standard SDS-PAGE either on 12% gelon 4-12% Novex gradient gels. The proteins were immobilized Hybond™ECL™ nitrocellulose membrane (AmershamBuckinghamshire, UK) by semi-dry transfer, and processed fwestern blots. A polyclonal antibody against the N terminus of humcaveolin (sc-894, Santa Cruz Biotechnology) was used as primantibody (1:1,000 in 5% non-fat milk, 0.1% Tween-20, PBS). Asecondary antibody, HRP-linked donkey anti-rabbit antibod(Amersham life science) was used. ECL™ (Amersham) was useddetection method.

Triton X-100 insolubilityTriton X-100 insolubility assay was performed as described bDanielsen (1995). Briefly, cells were lysed and incubated in 25 mHepes, 150 mM NaCl, pH 6.5, 1% Triton X-100 for 10 minutes oice with frequent vortexing, and centrifuged at 48,000 g for 30minutes. The pellet was solubilized in sample buffer, and equvolumes of solubilized pellet and supernatant were analysed by SDPAGE and western blotting as described.

ImmunofluorescenceCells grown in T25 flasks, on glass coverslips or on filters wewashed in PBS and fixed in 2% formaldehyde in 0.1 M sodiuphosphate buffer, pH 7.2, for 1 hour at room temperature (subsequent incubations were performed at room temperature). formaldehyde was quenched with 25 mM glycine in PBS, and the cepermeabilized with 0.2% saponin in PBS for 1 hour or by 0.1% TritoX-100 for 15 minutes. The cells were blocked for 15 minutes with 5normal goat serum in PBS and incubated in the same buffer for ohour with the polyclonal anti-caveolin antibody sc-894 also used fwestern blots diluted 1:100. After 3 times 10 minute washes in PBthe cells were incubated for 30 minutes with FITC-conjugated goanti-rabbit antibody (Southern Biotechnology Associates, InBirmingham, USA), diluted 1:50 in 5% normal goat serum in PBSThe cells were then washed 4 times 10 minutes in PBS and dried

Electron microscopyConfluent cultures of MDCK II and Caco-2 cells grown on filters werrinsed with PBS and fixed with 2% formaldehyde and 0.1%glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.2. Followinfixation, the cells on filters were cut free of the plastic insert, postfixin OsO4, contrasted en bloc with 1% uranyl acetate, dehydrated ingraded series of ethanols, and embedded in epon. Sections wereperpendicular to the filter and collected on Formvar-coated mesh grand examined in a Philips CM100 electron microscope. With respto quantitative analysis, see Results.

827Caveolin-1 expression and caveolae

uldileoridsssed

nessedn in

ofelntn in

hasascts

94;aslowniersing3/2)mig.

-294laeandurhect,ral

enll

Table 1. Frequency of caveolae and relative amount ofcaveolin in the various cell lines and clones

No. caveolae at the basolateral n (number of Relative amount

Cell line/clone surface/mm filter* determinations) of caveolin†

MDCK II 433±111 4 100Wild-type Caco-2 <4 3 0‡Mock transf. Caco-2 8±5 3 0Clone 33/2 76±22 3 20Clone 33/4 39±6 3 100Clone 50/1 70 1 11Clone 50/2 27 1 38Clone 50/4 50 1 33Clone 50/5 21 1 61

The number of caveolae in each clone was determined as described in thetext. The western blot in Fig. 1 was quantified by scanning densitometryusing a Mustek Paragon MFS 12000 ex transparent scanner. The bands werequantified using a software program, Scimage (Vanck, Denmark).

*No caveolae were identified at the apical surface in any of the celllines/clones.

†The amount in MDCK II was arbitrarily set to 100%.‡Scanning value is not from the gel shown in Fig. 1.

RESULTS

Expression of caveolin-1 in Caco-2 cellsIn agreement with previous studies (Mirre et al., 1996) we conot detect any caveolin-1 in wild-type (wt) Caco-2 cells, whMDCK II cells showed strong expression (Fig. 1). Ftransfection of Caco-2 cells we used 2 different plasmexpressing caveolin-1 cDNA isolated from MDCK II cell(Kurzchalia et al., 1992). The 2 plasmids both exprescaveolin-1 cDNA from the strong, constitutive promoter pUbc(Johansen et al., 1990). Thirteen of 36 isolated Caco-2 clocould be amplified. Nine of 13 analysed clones expresdetectable levels of caveolin. Seven of these clones are showFig. 1. The different clones vary in the level of expressioncaveolin-1 from 10% of the level seen in MDCK II cells to a levsimilar to that seen in MDCK II cells (Table 1). The recombinacaveolin has the same mobility as the native caveolin seeMDCK II cells on denaturing acrylamide gels (Fig. 1).

Characterization of the caveolin-1 expressed inCaco-2 cellsCaveolin present in caveolae and in TGN transport vesiclesbeen shown to be insoluble in Triton X-100 at 0°C and it hbeen pointed out that this detergent insolubility probably reflethe biophysical properties of the protein (Fra et al., 19Kurzchalia et al., 1992; Mirre et al., 1996; Parton, 1996). It halso been shown that caveolin-1 is resistant to SDS at temperatures (25°C), whereas it dissociates upon boiling (Moet al., 1995). The caveolin-1 expressed by the highly expresCaco-2 clone (33/4) and the low expressing Caco-2 clone (3both turned out to be Triton X-100 insoluble (Fig. 2A) and for200 kDa complexes on SDS-PAGE after incubation at 25°C (F2B) thus behaving like caveolin-1 from other sources.

We determined the localization of caveolin-1 in the Cacocells by immunofluorescence microscopy using the sc-8antibody which only seems to recognize caveolin in caveo(Dupree et al., 1993). No caveolin could be detected in wt mock transfected Caco-2 cells (Fig. 3A), confirming oobservations made with western blotting (Fig. 1). In ttransfected clones, labeling with sc-894 revealed a distinsometimes fine dotted fluorescence mainly of the late

Fig. 1. Western blot of Caco-2 clones transfected with caninecaveolin-1 cDNA. Equal amounts of protein (20 µg/lane) was loadedon a 4-12% gradient SDS-polyacrylamide gel. The proteins weretransferred to a nitrocellulose membrane using semi-dry transfer.Caveolin-1 was visualised using ECL (Amersham). Lane 1: mocktransfected Caco-2; lane 2: clone 33/2; lane 3: clone 33/4; lane 4:clone 50/1; lane 5: clone 50/2; lane 6: clone 50/3; lane 7: clone 50lane 8: clone 50/5 and lane 9: MDCK II. Arrow, 21 kDa monomer.

plasma membrane (Fig. 3B). No such labeling was seen whthe focal plane was shifted to the top or bottom of the cemonolayer.

/4;

Fig. 2.Caveolin-1 expressed in Caco-2 cells is Triton X-100 resistantat 0°C (A) and forms SDS-resistant oligomers at 25°C (B). Equalamounts of protein were loaded on a 4-12% SDS-PAGE,immobilized on nitrocellulose and visualized with ECL (Amersham)using antibody sc-894 against caveolin-1. (A) Triton X-100 solubilityof caveolin-1 at 0°C. Lanes 1 and 2 represent clone 33/4 (lane 1,soluble fraction; lane 2, insoluble fraction). Lanes 3 and 4 representclone 33/2 (lane 3, soluble fraction; lane 4, insoluble fraction).Arrow: 21 kDa monomer. (B) SDS resistancy at 25°C. Cells werelysed in sample buffer containing 2% SDS and incubated either at25°C for 30 minutes or at 100°C for 10 minutes. Lanes 1 and 2:clone 33/2 (lane 1 incubated at 100°C, lane 2 at 25°C). Lanes 3 and4: clone 33/4 (lane 3 incubated at 100°C, lane 4 at 25°C). Arrow: 200kDa homooligomer. The 21 kDa monomer appears at the bottom ofthe gel.

828

of as to toon

Mellyanichre

ma

U. Vogel, K. Sandvig and B. van Deurs

Fig. 3. Immunofluorescense labeling of Caco-2 cells with thepolyclonal antibody sc-894 to caveolae-associated caveolin-1.(A) Mock transfected Caco-2 cells. (B) Caveolin transfected Caco-2clone 50/1 (focal plane in the middle of the cell monolayer). Notethat caveolin in the transfected cells primarily appears as a finedotted pattern associated with the lateral membrane. Bar, 10 µm.

Fig. 4.Low magnification electron micrograph of MDCK II cellsgrown on a filter. The length of this portion of epithelial monolayer isapprox. 110 µm, corresponding to about 12 epithelial cells. In such aportion there is on average about 50 caveolae which open towards thebasolateral surface; no caveolae are found on the apical surface. Bar,5 µm.

Taken together, these data indicate that a large fractionthe caveolin-1 expressed in the Caco-2 clones behavesendogeneous caveolae-associated caveolin-1. However,which extent the expressed caveolin-1 actually gives riseformation of caveolae can only be determined by electrmicroscopy (EM).

Frequency and distribution of caveolae in Caco-2and MDCK II cellsThe frequency and distribution of caveolae were studied by Eof MDCK II and Caco-2 cells grown on filters (Figs 4, 5). Thterm caveolae as used here is defined strictly morphologicaas characteristic 50-80 nm invaginations of the plasmmembrane which are readily identified by EM (Fig. 5) (Partoand Simons, 1995; Parton 1996). It is clear that caveolae whmay exist in a flattened (non-invaginated) form and therefocannot be distinguished from the other portions of the plas

829Caveolin-1 expression and caveolae

Fig. 5.Electron micrographs showing examples of caveolae in Caco-2 cells expressing caveolin-1 (clone 33/2, A-E), and in MDCK II cells (F).The arrows indicate various examples of caveolae opening to the cell surface as they are defined for quantification. Accordingly, profiles likethose labeled with asterisks (C and F) are not included although they may represent caveolae. Note that in Caco-2 cells the caveolae areassociated with the lateral plasma membrane (Lm) rather than with the basal plasma membrane (Bm). (A and B) Two clathrin-coated pits (Cp)to show that caveolae and clathrin coated pits are readily distinguished in our preparations. F shows the basal part of an MDCK II cell; here thecaveolae are associated with the basal plasma membrane (Bm). Fil, filter. Bar, 100 nm.

830

al

d

.,leedrys

sthersf

ef

ft

n

.yeo

ric

erinn

artlyon

e

eraldg

heal

U. Vogel, K. Sandvig and B. van Deurs

membrane (Smart et al., 1996) are not included. Moreover,quantification we only included invaginated caveolae whiclearly opened to the cell surface (Fig. 5). This is likely to gian underestimate of the total number of caveolae since smust appear as ‘free’ vesicles in a random, single sectCaveolae which do not open to the cell surface in a givsection can be accounted for by fixation with Ruthenium R(van Deurs et al., 1996), but this approach, which gave roug2.5 times higher counts, makes the precise definition caveolae more difficult and was therefore not used (data shown).

For each cell line/clone two different portions of a filtegrown, confluent epithelial monolayer cut perpendicular to tfilter were selected at low magnification (Fig. 4), and thcaveolae were identified and counted along the basal membrane facing the filter as well as at the lateral and apsurfaces at high magnification (Fig. 5). Finally the filteportions, typically 100-200 µm of filter corresponding to thedistance between the grid bars, were photographed at magnification (×700) to measure the length of filter examine(Fig. 4). The number of caveolae per mm of filter was thcalculated as the mean of the two filter portions (n=1). Thequantification revealed that there was on average 433 caveper mm filter associated with the basolateral membrane inMDCK II cells (Table 1). The caveolae were predominant(>70%) associated with the basal cell membrane facing filter, although the lateral part of the basolateral membrarepresents by far the largest surface area due to foldinginterdigitations. Moreover, the caveolae were not evendistributed but often appeared as patches with 3-10 cave(as judged from a single thin section) (Fig. 5F). Accordinglong stretches of basal plasma membrane were devoidcaveolae. Finally it should be stressed that no caveolae widentified at the apical surface.

In wt Caco-2 cells, less than four caveolae-like structures mm filter were observed (Table 1). This is in agreement wthe lack of caveolin-1 expression in these cells. However, Caco-2 clones transfected with caveolin had 21-caveolae/mm filter, and thus significantly more caveolae ththe wt Caco-2 cells (Table 1 and Fig. 5A-E). Contrexperiments showed that mock transfection of control cells not significantly change the low frequency of caveolae in Caco-2 cells.

Several intriguing observations were made in relation caveolin expression and the occurrence of caveolae in the C2 clones. First, even though the level of caveolin expressionsome Caco-2 clones was comparable to that of wt MDCKcells there was no correlation between the amount of exprecaveolin and the number of caveolae. On the contrary, it sethat the Caco-2 clones which have medium and low expresof caveolin (clones 33/2 and 50/1, Table 1), have more cavethan clones expressing high levels of caveolin (33/4 and 50Overall, the number of caveolae were six to ten times lower tin wt MDCK II cells (Table 1). Second, the caveolae in thCaco-2 clones were localized to the lateral membranes (Fig.E); with a few exceptions caveolae were not observed at basal membrane facing the filter, in contrast to the situationMDCK II cells. On the other hand, clathrin-coated pits wesometimes seen at the basal membrane. Third, caveolae not observed at the apical surface of Caco-2 clones althothere appeared to be space enough between the hi

forchveomeion.enedhlyof

not

rheencellicalr

lowden

olae wtlythene orly

olaely, ofere

periththe76an

oldidwt

toaco- in IIssedemssionolae/5).hane

5B-the inrewereughghly

differentiated microvilli for possible caveole-formation andmoreover, clathrin coated pits were frequently seen at the apicsurface of Caco-2 cells (not shown).

Since caveolin may not only be associated withplasmalemmal caveolae but also with various vesicular another membrane structures within the cytosol, including TGNelements (Joliot et al., 1997; Kurzchalia et al., 1992; Li et al1996; Parton et al., 1997), we were attentive to the possibexistence of such membrane material in particular in thvicinity of the lateral plasma membrane in the transfecteCaco-2 cells. However, no increase in vesicles and othemembrane structures was noticed in any clone. The only wato distinguish wt Caco-2 cells from transfected Caco-2 clonein blind experiments was to carefully count caveolae.

Clone 33/4 expresses a lot of caveolin, little of which thusseems to participate in formation of caveolae (Table 1). It itherefore possible that caveolae biogenesis is repressed in clone, since it has been shown that a number of factoinfluence caveolae formation. For instance, activation oprotein kinase C-α (PKC-α) has been shown to disruptcaveolae formation (Smart et al., 1994, 1995). If caveolabiogenesis is repressed by elevated PKC activity, inhibition oPKC-α might lead to caveolae formation. We have thereforetreated clone 33/4 with a PKC-α inhibitor,bisindolylmalemide, and with an activator, phorbol-12myristate-13 acetate (TPA), since prolonged stimulation oPKC by TPA has been shown to downregulate PKC (Young eal., 1987). Inhibition or downregulation of PKC did not,however, affect caveolae biogenesis. Furthermore, incubatiowith serum-free medium to avoid growth factor-inducedstimulation of PKC did not increase the number of caveolae

In some cell types, the number of caveolae is affected bdifferentation of the cells. In adipocytes the number of caveolais increased upon differentiation (Fan et al., 1983) and alscultured myoepithelial cells which are allowed to differentiatehave numerous caveolae (Petersen et al., 1989). Since butyacid induces enhanced differentiation of intestinal epithelium(Nathan et al., 1990; Stoddart et al., 1989) we tested whethbutyric acid treatment leads to increased caveolae formation a caveolin-1 expressing Caco-2 clone. However, incubatiowith butyric acid had no influence on caveolae formation.

These results clearly show that there is not necessarily direct correlation between the caveolin level and the numbeof caveolae present on the plasma membrane, and importanthe data also demonstrate that there is a polarized distributiof caveolae in the two epithelial cell lines studied.

DISCUSSION

The most important result of the present study is that thformation of invaginated caveolae is polarized both in MDCKII cells and in Caco-2 cells transfected with caveolin. InMDCK II cells such caveolae are predominantly present at thbasal membrane, although some are observed at the latemembrane. In contrast, caveolae were almost exclusively founat the lateral membrane in the Caco-2 clones expressincaveolin-1. One can speculate whether this indicates that tbasolateral membrane is really two distinct domains, the basand the lateral membrane, with different functionalcharacteristics, varying among epithelial cells.

831Caveolin-1 expression and caveolae

-2line angr

linof

oateatcee/4

nl

ylyatesbeceet

helae5)enes.eata

ir a bylesenertly,-ceInnod).hdrells

Drsnd

srcheG

The polarized distribution of caveolae is not caused by unequal distribution of cholesterol in the apical and tbasolateral membrane, since the molar percentage cholesterol in these 2 membranes is the same (30%) (Simand van Meer, 1988).

The formation of invaginated caveolae exclusively at tbasolateral side of epithelial cells studied so far is interestibecause there is a high amount of glycolipids at the apical sof epithelial cells, and glycolipid rich domains may bassociated with caveolin (Lisanti et al., 1993; Sargiacomoal., 1993; Zurzolo et al., 1994). Moreover, caveolin-1 is presin transport vesicles from the TGN which contains the apicadestined influenza haemagglutinin (Kurzchalia et al., 1992)has therefore been implied that caveolae are present atapical membrane of polarized cells although it has not beshown by EM (Joliot et al., 1997; Lisanti et al., 1994Importantly, caveolin is also found in vesicular stomatitis viruG protein-positive basolaterally destined transport vesiclesMDCK II cells (Kurzchalia et al., 1992). Whether onlycaveolin associated with these vesicles or also caveooriginally directed to the apical surface is involved ibasolateral caveolae formation is still not known. Thus,seems that caveolin-1 is transported to the apical surfacMDCK II cells without this leading to caveolae formation.

The polarized formation of caveolae is also of interest relation to the sorting of GPI-anchored proteins. For examthe urokinase plasminogen activator receptor has blocalized to caveolae (Stahl and Mueller, 1995) and is apicasorted in MDCK cells (Limongi et al., 1995). Since caveolaare present only at the basolateral membrane of polariMDCK cells, the urokinase plasminogen activator recepcannot be localized to caveolae in such cells. One can therequestion the physiological significance of the caveolocalization of the receptor. It has been reported that apsorting of other GPI-anchored proteins is also linked to thlocalization in caveolae in polarized MDCK cells (Lisanti eal., 1993; Sargiacomo et al., 1993). This now seems less likHowever, it is possible that caveolin present in the TGNinvolved in sorting, since the GPI-anchored fusion protegD1-DAF (Zurzolo et al., 1994) and the urokinase-typplasminogen activator (Canipari et al., 1992) are sortdifferently in MDCK cells, which express caveolin, and iFisher rat thyroid cells or Caco-2 cells, which do not exprecaveolin. It would certainly be interesting to study protesorting in Caco-2 cells with and without caveolin expressio

In this study we used the canine cDNA for caveolin-1 sinthis molecule has previously been used to induce caveolahuman lymphocytes (Fra et al., 1995) and since we are hcomparing the localization of caveolae in Caco-2 cells with thin the widely used canine cell line MDCK II. We find iunlikely, however, that transfection of Caco-2 cells with humcaveolin cDNA might change the distribution of caveolae sinthere are only minor differences between canine and humcaveolin-1 (Scherer et al., 1996; Tang et al., 1996).

We here show that expression of canine caveolin-1 genercaveolae in Caco-2 cells which normally do not form caveolThis is in agreement with a recent report by Fra et al. (199who found formation of caveolae upon transfection lymphocytes with caveolin-1. However, Fra et al. (1995) foua positive correlation between the level of caveolin expressand the number of caveolae upon transfection of lymphocy

anhe

ofons

heng,idee etentlly. It theen

).s in

linn ite in

inpeleenllye

zedtorforelaricaleirtely. isineednssinn.cee inereat

tancean

atesae.5),ofndiontes

with caveolin-1. In the present study, we show that in Cacoclones there is no correlation between the level of caveoexpression and the number of caveolae (Table 1). There arnumber of possible explanations for this somewhat surprisiresult. Mutations in the cDNA are unlikely, since the numbeof caveolae in the clones expressing low levels of caveocorrelates with the caveolin level, indicating the expression a fully functional protein (clone 33/2 and 50/1, Table 1).

Alternatively, a high number of caveolae may be toxic tCaco-2 cells, so only clones that either express moderamounts of caveolin or express high amounts of caveolin thdo not cause formation of caveolae could be amplified. Sincaveolin-1 has no designated function, it is hard to test if thprotein is active. However, the caveolin expressed in clone 33has the correct mobility on SDS-PAGE, forms oligomers ivivo and is Triton X-100 insoluble, like caveolin in other celtypes.

It is possible that caveolae formation in Caco-2 cells marequire other proteins in addition to caveolin-1 and that newsynthesized caveolin in the absence of such proteins associwith intracellular (non-caveolar) membranes. This seems to true for some other expression systems for caveolin-1, sinexpression of caveolin-1 in insect cells leads to a hugaccumulation of caveolin-1 positive intracellular vesicles (Li eal., 1996). Furthermore, according to these authors tcaveolin-1 positive membrane structures identified as caveoin the caveolin-1 transfected lymphocytes (Fra et al., 199were also predominantly intracellular vesicles. However, wdid not observe accumulation of vesicular and other membrastructures by EM of the caveolin-expressing Caco-2 cellMoreover, immunofluorescence labeling of caveolin-1 in thCaco-2 clones transfected with caveolin-1 indicates thcaveolin-1 is predominantly localized to the lateral plasmmembrane.

Caveolae were originally identified on the basis of thestructure as invaginations on the plasma membrane withcharacteristic shape and size. It has later been proposeddifferent groups that caveolae are able to close and form vesic(Rothberg et al., 1990; Schnitzer et al., 1996). However, whcaveolae are closed they are indistinguishable from othvesicles which happen to be of the same size. Consequencryo-immunogold labelling has been used to identify caveolinpositive membrane structures. This may be a problem sintransport vesicles from the TGN also contain caveolin-1. previous transfection studies there has apparently been differentiation between morphologically defined caveolae ancaveolin-1 positive vesicles (Joliot et al., 1997; Li et al., 1996We believe it is important for future studies to distinguisbetween caveolin-1 positive vesicles of unknown origin anfunction and morphologically well-defined caveolae if caveolafunction and biogenesis is to be elucidated. It will also binteresting to investigate to which extent other polarized cehave a polarized distribution of caveolae.

We are grateful to Ulla Hjortenberg, Mette Ohlsen, KirstenPedersen and Keld Ottosen for their excellent technical assistance.K. Simons and T. E. Johansen are thanked for providing plasmids aDr. L. Vogel for help with transfections. The present study isupported by the Danish Cancer Society, the Danish Medical ReseaCouncil, the Novo-Nordic Foundation, the Human Frontier SciencProgramme, and a NATO Collaborative Research Grant (CR900517).

832

on

n

s

m

a

ed

ll.

is

an

d

tina

al.

f

U. Vogel, K. Sandvig and B. van Deurs

REFERENCES

Anderson, R. G. W., Kamen, B. A., Rothberg, K. G. and Lacey, S. W.(1992). Potocytosis: Sequestration and transport of small moleculescaveolae. Science255, 410-411.

Canipari, R., Zurzolo, C., Polistina, C., Garbi, C., Aloj, L., Cali, G.,Gentile, R. and Nitsch, L. (1992). Polarized secretion of plasminogeactivators by epithelial cell monolayers. Biochem. Biophys. Acta1175, 1-6.

Danielsen, E. M.(1995). Involvement of detergent-insoluble complexes in thintracellular transport of intestinal brush border enzymes. Biochemistry34,1596-1605.

Devine, C. E., Simpson, F. O. and Bertaud, W. S.(1971). Surface featuresof smooth muscle cells from the mesenteric artery and vas deferens. J. CellSci.8, 427-443.

Dupree, P., Parton, R. G., Raposo, G., Kurzchalia, T. V. and Simons, K.(1993). Caveolae and sorting in the trans-Golgi network of epithelial ceEMBO J.12, 1597-1605.

Engelman, J. A., Wykoff, C. C., Yasuhara, S., Song, K. S., Okamoto, T.and Lisanti, M. P. (1997). Recombinant expression of caveolin-1 ioncogenically transformed cells abrogates anchorage-independent groJ. Biol. Chem.272, 16374-16381.

Fan, J. Y., Carpentier, J.-L., Obberghen, E. v., Grunfeld, C., Gorden, P.and Orci, L. (1983). Morphological changes of the 3T3-L1 fibroblasplasma membrane upon differentiation to the adipocyte form. J. Cell Sci.61, 219-230.

Forbes, M. S., Rennels, M. L. and Nelson, E.(1979). Caveolar system andsarcoplasmic reticulum in conorary smooth muscle cells of the mouseJ.Ultrastruct. Res.67, 325-339.

Fra, A. M., Williamson, E., Simons, K. and Parton, R.(1994). Detergent-insoluble glycolipid microdomains in lymphocytes in the absence caveolae. J. Biol. Chem.269, 30745-30748.

Fra, A. M., Williamson, E., Simons, K. and Parton, R. G.(1995). De novoformation of caveolae in lymphocytes by expression of VIP21-caveolProc. Nat. Acad. Sci. USA92, 8655-8659.

Fujimoto, T. (1993). Calcium pump of the plasma membrane is localizedcaveolae. J. Cell Biol.120, 1147-1157.

Gabella, G. (1978). Inpocketings of the cell membrane (caveolae) in the myocardium. J. Ultrastruct. Res.65, 135-147.

Gossen, M. and Bujard, H. (1992). Tight control of gene expression inmammalian cells by tetracycline-responsive promoters. Proc. Nat. Acad. Sci.USA89, 5547-5551.

Johansen, T. E., Scholler, M. S., Tolstoy, S. and Schwartz, T. W.(1990).Biosynthesis of peptide precursors and protease inhibitors using nconstitutive and inducible eukaryotic expression vectors. FEBS Lett. 267,289-294.

Joliot, A., Trembleau, A., Raposo, G., Calvet, S., Volovitch, M. andProchiantz, A. (1997). Association of engrailed homeoproteins witvesicles presenting caveolae-like properties. Development124, 1865-1875.

Kurzchalia, T. V., Dupree, P., Parton, R. G., Kellner, R., Virta, H., Lehnert,M. and Simons, K. (1992). VIP21, a 21-kD membrane protein is an integrcomponent of trans-Golgi-network-derived transport vesicles. J. Cell Biol.118, 1003-1014.

Li, S., Song, K. S., Koh, S. S., Kikuchi, A. and Lisanti, M. P.(1996).Baculovirus-based expression of mammalian caveolin in Sf21 insect ceJ. Biol.Chem.271, 28647-28654.

Limongi, P., Resnati, M., Hernandez-Marrero, L., Cremona, O., Blasi, F.and Fazioli, F.(1995). Biosynthesis and apical localization of the urokinareceptor in polarized MDCK epithelial cells. FEBS Lett. 369, 207-211.

Lisanti, M. P., Tang, Z. and Sargiacomo, M.(1993). Caveolin forms a hetero-oligomeric protein complex that interacts with an apical GPI-linked proteimplications for the biogenesis of caveolae. J. Cell Biol.123, 595-604.

Lisanti, M. P., Scherer, P. E., Tang, Z. L. and Sargiacomo, M.(1994).Caveolae, caveolin and caveolin-rich membrane domains: a signalhypothesis. Trends Cell Biol.4, 321-235.

Mayor, S., Rothberg, K. and Maxfield, F. (1994). Sequestration of GPI-anchored proteins in caveolae triggered by crosslinking. Science264, 1948-1951.

Mirre, C., Monlauzeur, L., Garcia, M., Delgrossi, M. H. and Bivic, A. L.(1996). Detergent-resistant membrane microdomains from Caco-2 cellsnot contain caveolin. Am. J. Physiol.271, C887-894.

Monier, S., Parton, R. G., Vogel, F., Behlke, J., Henske, A. and Kurzchalia,T. V. (1995). VIP21-caveolin, a membrane protein constituent of tcaveolar coat, oligomerizes in vivo and in vitro. Mol. Biol. Cell6, 911-927.

Murata, M., Paranen, J., Schreiner, R., Wieland, F., Kurzchalia, T. and

by

n

e

lls.

nwth.

t

.

of

in.

in

rat

ew

h

al

lls.

se

in:

ling

do

he

Simons, K. (1995). VIP21/caveolin is a cholesterol-binding protein. Proc.Nat. Acad. Sci. USA92, 10339-10343.

Nathan, D. F., Burkhart, S. R. and Morin, M. J. (1990). Increased cellsurface EGF receptor expression during the butyrate-induced differentiatiof human HCT-116 colon tumor cell clones. Exp. Cell Res.190, 76-84.

Parton, R. G. and Simons, K.(1995). Digging into caveolae. Science269,1398-1399.

Parton, R. G.(1996). Caveolae and caveolins. Curr. Opin. Cell Biol.8, 542-548.Parton, R. G., Way, M., Zorzi, N. and Stang, E.(1997). Caveolin-3

associates with developing t-tubules during muscle differentiation. J. CellBiol. 136, 137-154.

Petersen, O. W., Hansen, S. H., Laursen, I. and van Deurs, B.(1989). Effectof insulin on growth and expression of smooth muscle isoactin in humabreast gland myoepithelial cells in a chemically defined culture system. Eur.J. Cell Biol.50, 500-509.

Rothberg, K. G., Ying, Y., Kolhouse, J. F., Kamen, B. A. and Anderson,R. G. W. (1990). The glycophospholipid-linked folate receptor internalizefolate without endtering the clathrin-coated pit endocytic pathway. J. CellBiol. 110, 637-649.

Rothberg, K. G., Heuser, J. E., Donzell, W. C., Ying, Y.-S., Glenney, J. R.and Anderson, R. G. W.(1992). Caveolin, a protein component of caveolaemembrane coats. Cell 68, 673-682.

Sargiacomo, M., Sudol, M., Tang, Z. and Lisanti, M. P.(1993). Signaltransducing molecules and glycosyl-phosphatidylinositol-linked proteins fora caveolin-rich insoluble complex in MDCK cells. J. Cell Biol.122, 789-807.

Scherer, P. E., Okamoto, T., Chun, M., Lodish, H. F. and Lisanti, M. P.(1996). Identification, sequence, and expression of caveolin-2 definescaveolin gene family. Proc. Nat. Acad..Sci. USA93, 131-135.

Schnitzer, J. E., McIntosh, D. P., Dvorak, A. M., Liu, J. and Oh, P.(1995).Separation of caveolae from associated microdomains of GPI-anchorproteins. Science269, 1435-1439.

Schnitzer, J. E., Oh, P. and MacIntosh, D. P.(1996). Role of GTP hydrolysisin fission of caveolae directly from plasma membranes. Science274, 239-242.

Sevinsky, J. R., Rao, L. V. M. and Ruf, W.(1996). Ligand-induced proteasereceptor translocation into caveolae: A mechanism for regulating cesurface proteolysis of the tissue factor-dependent coagulation pathwayJ.Cell Biol. 133, 293-304.

Simons, K. and van Meer, G.(1988). Lipid sorting in epithelial cells.Biochemistry27, 6197-6202.

Smart, E. J., Foster, D. C., Ying, Y.-S., Kamen, B. A. and Anderson, R. G.W. (1994). Protein kinase C activators inhibit receptor-mediated potocytosby preventing internalization of caveolae. J. Cell Biol.124, 307-313.

Smart, E. J., Ying, Y.-S. and Anderson, R. G. W.(1995). Hormonalregulation of caveolae internalization. J. Cell Biol.131, 929-938.

Smart, E. J., Mineo, C. and Anderson, R. G. W.(1996). Clustered folatereceptors deliver 5-methyltetrahydrofolate to cytoplasm of MA104 cells. J.Cell Biol. 134, 1169-1177.

Stahl, A. and Mueller, B. M. (1995). The urokinase-type plasminogenactivator receptor, a GPI-linked protein, is localized in caveolae. J. Cell Biol.129, 335-344.

Stoddart, J. H., Lane, M. A. and Niles, R. M.(1989). Sodium butyratesupresses the transforming activity of an activated N-ras oncogene in humcolon carcinoma cells. Exp. Cell Res.184, 16-27.

Tang, Z., Scherer, P. E., Okamoto, T., Song, K., Chu, C., Kohtz, D. S.,Nishimoto, I., Lodish, H. F. and Lisanti, M. P.(1996). Molecular cloningof caveolin-3, a novel member of the caveolin gene family expressepredominantly in muscle. J. Biol. Chem.271, 2255-2261.

Turek, J. J., Leamon, C. P. and Low, P. S.(1993). Endocytosis of folate-protein conjugates: ultrastructural localization in KB cells. J. Cell Sci.106,423-430.

van Deurs, B., von Bulow, F., Vilhardt, F., Holm, P. K. and Sandvig, K.(1996). Destabilization of plasma membrane structure by prevention of acpolymerization. Microtubule-dependent tubulation of the plasmmembrame. J. Cell Sci.109, 1655-1665.

Vogel, L. K., Noren, O. and Sjostrom, H. (1995). Transcytosis ofaminopeptidase N in Caco-2 cells is mediated by a non-cytoplasmic signJ. Biol. Chem.270, 22933-22938.

Young, S., Parker, P. J., Ullrich, A. and Stabel, S.(1987). Down-regulationof protein kinase C is due to an increased rate of degradation. Biochem. J.244, 775-779.

Zurzolo, C., van´t Hof, W., van Meer, G. and Rodriguez-Boulan, E.(1994).VIP21/caveolin, glycosphingolipid clusters and the sorting oglycosylphosphatidylinositol-anchored proteins in epithelial cells. EMBO J.13, 42-53.