The Apical Submembrane Cytoskeleton Participates in the Organization of the Apical Pole in ...

17 Pages·1997·0.99 MB·English

The Apical Submembrane Cytoskeleton Participates in the
Organization of the Apical Pole in Epithelial Cells
Pedro J.I. Salas,* Marcelo L. Rodriguez,‡ Ana L. Viciana,§ Dora E. Vega-Salas,* and Hans-Peter Hauri(cid:105)
*Department of Cell Biology and Anatomy, and §Department of Pathology, University of Miami School of Medicine, Miami,
Florida 33101; ‡Department of Carcinogenesis, M.D. Anderson Cancer Center, University of Texas, Smithville, Texas 78957; and
CH-4056 (cid:105)Biozentrum, University of Basel, Department of Pharmacology, Basel CH-4056, Switzerland
Abstract. In a previous publication (Rodriguez, M.L., surface in cells with lower levels of cytokeratin 19. Api-
M. Brignoni, and P.J.I. Salas. 1994. J. Cell Sci. 107: cal proteins purified in the detergent phase of Triton
3145–3151), we described the existence of a terminal X-114 (typically integral membrane proteins) and those D
o
w
web-like structure in nonbrush border cells, which com- differentially extracted in Triton X-100 at 37(cid:56)C or in nlo
prises a specifically apical cytokeratin, presumably cy- n-octyl-(cid:98)-d-glycoside at 4(cid:56)C (representative of GPI- ade
d
tokeratin 19. In the present study we confirmed the api- anchored proteins), appeared partially redistributed to fro
m
cal distribution of cytokeratin 19 and expanded that the basolateral domain. A transmembrane apical pro- http
observation to other epithelial cells in tissue culture tein, sucrase isomaltase, was found mispolarized in a ://ru
p
and in vivo. In tissue culture, subconfluent cell stocks subpopulation of the cells treated with antisense oligo- re
under continuous treatment with two different 21-mer nucleotides, while the basolateral polarity of Na(cid:49)– ss.o
rg
phosphorothioate oligodeoxy nucleotides that targeted K(cid:49)ATPase was not affected. Both sucrase isomaltase /jcb
/a
cytokeratin 19 mRNA enabled us to obtain confluent and alkaline phosphatase (a GPI-anchored protein) ap- rticle
monolayers with a partial (40–70%) and transitory re- peared partially depolarized in A19 treated CACO-2 -p
d
duction in this protein. The expression of other cyto- monolayers as determined by differential biotinylation, f/1
3
7
skeletal proteins was undisturbed. This downregulation affinity purification, and immunoblot. These results /2/3
5
of cytokeratin 19 resulted in (a) decrease in the number suggest that an apical submembrane cytoskeleton of in- 9/1
2
of microvilli; (b) disorganization of the apical (but not termediate filaments is expressed in a number of epi- 70
2
0
lateral or basal) filamentous actin and abnormal apical thelia, including those without a brush border, although 4
/1
8
microtubules; and (c) depletion or redistribution of api- it may not be universal. In addition, these data indicate 8
6
6
cal membrane proteins as determined by differential that this structure is involved in the organization of the .pd
apical–basolateral biotinylation. In fact, a subset of de- apical region of the cytoplasm and the apical mem- f by g
u
tergent-insoluble proteins was not expressed on the cell brane. e
st o
n
0
6
M
a
rch
Cell polarity (asymmetry) is a broadly distributed The acquisition and maintenance of epithelial polarity is 20
2
and highly conserved feature of many different cell based on multiple interrelated mechanisms that may work 3
types, from prokaryotes to higher eukaryotes (Nel- in parallel. Although the origin of polarization depends on
son, 1992). In multicellular organisms it is more conspicu- the sorting of apical and basolateral membrane proteins at
ous in, but not restricted to, neurons and epithelial cells. In the trans-Golgi network (Simons and Wandinger-Ness,
the latter, the plasma membrane is organized in two differ- 1990), the mechanisms involved in the transport of apical
ent domains, apical and basolateral. This characteristic en- or basolateral carrier vesicles, the specific fusion of such
ables epithelia to accomplish their most specialized roles vesicles to the appropriate domain, and the retention of
including absorption and secretion and, in general, to per- membrane proteins in their correct positions are also im-
form the functions of organs with an epithelial parenchyma portant (Wollner and Nelson, 1992). Various components
such as the kidney, liver, intestine, stomach, exocrine glands, of the cytoskeleton seem to be especially involved in these
etc. (Simons and Fuller, 1985; Rodriguez-Boulan and Nel- mechanisms (Mays et al., 1994). Among them, the microtu-
son, 1989). bules, characteristically oriented in the apical–basal axis with
their minus ends facing toward the apical domain, appear
in a strategic position to transport carrier vesicles (Bacallao
et al., 1989). This orientation is largely expected because
Please address all correspondence to Pedro Salas, Department of Cell Bi-
of the apical distribution of centrioles and microtubule or-
ology and Anatomy, R-124, University of Miami School of Medicine, P.O.
Box 016960, Miami, FL 33101. Tel.: [phone]; Fax: [phone]. ganizing centers in epithelial cells (Buendia et al., 1990).
ª The Rockefeller University Press, 0021-9525/97/04/359/17 $2.00
The Journal of Cell Biology, Volume 137, Number 2, April 21, 1997 359–375 359
The molecular interactions responsible for that localiza- To assess possible functions of cytokeratin 19, we chose
tion, however, are unknown. to selectively reduce its synthesis using anti-sense phospho-
Actin is a widespread component of the membrane skel- rothioate oligodeoxy nucleotides, an extensively used ap-
eton found under apical, lateral, and basal membranes in a proach in recent years (e.g., Ferreira et al., 1992 ; Hubber
nonpolarized fashion (Drenckhahn and Dermietzel, 1988; et al., 1993; Takeuchi et al., 1994). Although we could not
Vega-Salas et al., 1988). Actin bundling into microvillus cores achieve a complete knock out, the steady-state levels of
in the presence of villin/fimbrin, on the other hand, is highly cytokeratin 19 were decreased to an extent that enabled us
polarized to the apical domain (Ezzell et al., 1989; Lou- to detect significant changes in the phenotype of CACO-2
vard et al., 1992). In fact, different isoforms of plastins de- and MCF-10A cells.
termine microvillus shape in a tissue-specific manner (Ar-
pin et al., 1994b). Why this arrangement is not found in
other actin-rich regions of the cell is unclear (Louvard et al., Materials and Methods
1992; Fath and Burgess, 1995).
Fodrin, the nonerythroid form of spectrin, underlies the Cells
basolateral domain (Nelson and Veshnock, 1987a,b) and is
MDCK cells were cultured as described before (Rodriguez et al., 1994).
known to participate in the anchoring/retention of basolat-
MCF-10A (CRL-10317), CACO-2 (HTB-37), and CACO-2 C2BBe1 (CRL-
eral proteins (Drenckhahn et al., 1985; Nelson and Ham- 2102) cells were obtained from American Type Culture Collection (Rock-
merton, 1989). Although different groups have found spe- ville, MD). MCF-10A were maintained in DME-F12 media supplemented
cific cytoskeletal anchoring of apical membrane proteins with 5% horse serum (GIBCO BRL, Gaithersburg, MD), 100 ng/ml cholera
toxin, 1 (cid:109)g/ml insulin, 500 pg/ml EGF, and 200 ng/ml hydrocortisone. All
at the “correct” domain (Ojakian and Schwimmer, 1988; these reagents were either (cid:103)-irradiated or cell culture tested. CACO-2 and Do
Salas et al., 1988; Parry et al., 1990), no specific apical CACO-2 C2BBe1 were maintained in DME-F12 supplemented with wn
lo
counterpart of the basolateral fodrin cytoskeleton is 10% fetal bovine serum (GIBCO BRL) and 1 mM sodium pyruvate. All ad
known. This is especially puzzling since we showed that ccuelbl altinoer sa wt 3e7re(cid:56)C c.o Tnhtien ucoeulls slyto icnkcsu wbaetreed k ienp 5t %in C25O- 2o irn 7 a5 -wcmat2e rti-sjasucke ectueldt uirne- ed fro
MDCK cells can maintain apical polarity in the absence of m
flasks and harvested weekly by dissociation in 0.25% trypsin, 2 mM h
tmigehcth ajuninscmtiso anrse, oanp eirnadtiiocnaatilo fno rt hapatic ainl tmraedmombraainne r petreontetiionns EstDocTkAs f ofro rc o1n5t in(CuoAuCs Ool-i2g)o nourc l3e0o tmidien t r(eMatCmFe-n1t0 Awe) rea tm 1a:1in0t adinileudti oinn .1 Core l2l ttp://ru
p
tm((hFVoeIret sam goniasmka -kjSeeon asreo l uatcw snoan dmel .ett,p hra1o alt9.tnh,7 eae19n 9;an t8Hpe o7tiuwcaf a)ltol.lh ramekn teodemf r Smibntraitnaeenarhmlee wleiindne bi,ba ,1rt ube9 sr7fhi9idl ;abg mMeosreo dntohets re(e cIdkFeeel)slr1s-,, wbhmPtoeiAemgnflho )l s, rDr deoooe ifnubcn knaislo idai2nctm i4hsego-eislwnanm (ise n(cid:122)silLc -l ca5camolo b (cid:51)aeuvwtxel eta1pridsrd0ele 4irigs ,pi chlmsaeB soel(lsenrGs d /ctecfseoxmo.r vprFm2eda,or a,nstr ndol Mei e pxgodsAplb a et(t)soMar,s i i;m1n Co–F neFc2in so -wh1t6nse0-ef, rl Altu lohsSe r ecnoc eiccf2el yela4ns l -)lti6msni f- ( wiwmB1ce,–ei or2lPTlec i dmortpat)alus natbol;stt unewiBdr de1gei sl2ahclh-t--, ress.org/jcb/article-p
1985), although no specific protein has been identified with Clear™ filters (Corning Costar Corp., Cambridge, MA), or on 140-mm df/1
tthoi se xsttrruaccttiuornes. Tofh ea poibcasel rpvraotitoenin so fa an crheomraerdk taob lcey rtoessiksetalentcael pgcarooll yrwecnasi rscbteaolnnlsca,e tme ( TofinEltoeRrlas) ,y( weProh irincehttei cwgsra iCtsy om wrepaa.s,s Luchriveeedcrk mveidao rbAey,g Ci/tAsA gt)rC.a Iln nes -lteehpceti trchoaedsleeias ol wfe lifetihlct tearrin-- 37/2/359
preparations (Salas et al., 1988) comparable to that of in- Epithelial Volt-Ohmmeter (World Precision Instrs. Inc., Sarasota, FL). /12
7
termediate filaments, led us to the study of cytokeratins in The final measure of TER was obtained after subtracting the resistance of 02
0
polarized cells. We developed an antibody against a 53-kD the solution and the filter without cells. 4/1
8
intermediate filament protein in MDCK cells. This protein 86
was found to be distributed exclusively to the apical do- Oligonucleotides 6.pd
f b
main and to form large (2,900 S) multi-protein complexes Synthetic phosphorothioate oligodeoxy nucleotides have been used by a y g
with apical plasma membrane proteins. Internal microse- large number of researchers to reduce the synthesis of specific proteins. ue
quencing of the 53-kD protein showed very high (95– Tbohteh tohuiotsaitdee saunbds tiintusitdioen t hme ackeells ( Zthheamo emt oarle., 1p9e9r3m)e. aFboluer adnifdf emreonrte 2 1st-ambeler st on 0
100%) homology with two polypeptides in the rod domain 6
oligonucleotides, two with antisense sequences for CK19 mRNA (A19 and M
osef rcvyetdo kaenrda tpinec 1u9li a(rC iKnt1e9r;m Medoilal teet failla.,m 1e9n8t2 )p rao theiignh (lyB acodner- Ado1m9//22); ,d aensdcr tihbeeidr icno Frriegs. p1o) nwdeinreg sryanntdhoemsizizeedd i nse aq uDeNncAe ss y(rnatnhdesoimze ra n(md oradne-l arch 2
0
et al., 1986). A complete identification however, could not 394; Perkin Elmer Applied Biosystems, Foster City, CA) at the DNA 23
Core Laboratory, University of Miami. The oligonucleotides were usually
be achieved (Rodriguez et al., 1994). The present study produced on a 1 (cid:109)M scale with a 21–35% final yield and routinely purified
was undertaken to establish that identity and to determine by two consecutive cycles of ethanol precipitation and resuspension in 10
the possible functions of this apical membrane skeleton. mM Tris-Cl, 1 mM EDTA, pH 8.0, followed by a 10(cid:51) dilution in the same
Because cytokeratins have been poorly characterized in buffer and ultrafiltration in Centricon™ 3 concentrators (Amicon Inc.,
Beverly, MA) to concentrate the oligonucleotides to the original volume.
canine cells, and no cytokeratin sequences are available in
Typically, the yield of this purification procedure was 70–80%. For exper-
this species, we decided to switch from MDCK cells to two
iments, the oligonucleotides were mixed to a final concentration with the
human epithelial cell lines, CACO-2, an extensively stud- standard tissue culture media supplemented as described above and filter
ied model of epithelial polarization that differentiates in sterilized through a 0.2-(cid:109)m pore filter immediately before use. There is
culture to form brush border containing cells (Pinto et al., consensus that 21-mer phosphorothioate oligonucleotides are effective at
concentrations ranging 1–10 (cid:109)M. Because it is known that the half life of
1983), and MCF-10A (Tait et al., 1990), a nontumorigenic these oligonucleotides in tissue culture media containing sera is (cid:122)24 h
cell line derived from normal mammary epithelia, as a model (Campbell et al., 1990), we re-fed the cells every 48 h (except weekends)
of nonbrush border cells. with 10 (cid:109)M A19, A19/2, random, or random/2 oligonucleotides, aiming to
maintain a concentration always (cid:46)1 (cid:109)M. The uptake of oligonucleotides
was found to be higher in subconfluent cells. Therefore, cell stocks were dis-
1. Abbreviations used in this paper: CK, cytokeratin; F-actin, filamentous sociated as soon as they reached confluency, usually once or twice a week.
actin; IF, intermediate filament; TER, trans-epithelial electrical resis- These stocks were continuously maintained under oligonucleotide treat-
tance. ment from a minimum of two passages and up to 4 mo. For experiments,
The Journal of Cell Biology, Volume 137, 1997 360
Table I. Specific Antibodies
Antigen Host Type Nomenclature Source
Cytokeratin 19 mouse mAb A53-B/A2 Sigma Chemical Co., St. Louis, MO
Cytokeratin 19 mouse mAb K4.62 Sigma Chemical Co.
Cytokeratin 19 (nonhuman) mouse mAb MAB1675 Chemicon Int. Inc., Temecula, CA
Cytokeratin 19 mouse mAb RCK108 Accurate Chem. and Sci. Corp., Westbury, NY
Cytokeratin 19 mouse mAb 2F5/96 Biogenesis, Inc., Sandown, NH
Cytokeratin 19 rabbit polyclonal anti-53 kD (Rodriguez et al., 1994)
Cytokeratin 8 mouse mAb M20 Sigma Chemical Co.
Cytokeratin 18 mouse mAb CK5 Sigma Chemical Co.
Pan-cytokeratin mouse mAb PCK-26 Sigma Chemical Co.
Pan-cytokeratin mouse mAb AE1/AE3 Dako Corp., Carpinteria, CA
Actin (all isoforms) mouse mAb C4 ICN Biomedicals, Inc., Costa Mesa, CA
Alkaline phosphatase rabbit polyclonal Z 271 Dakopatt A/S, Glostrup, Denmark
Na(cid:49)–K(cid:49)ATPase rabbit polyclonal (Mays et al., 1995)
Sucrase isomaltase mouse mAb HBB2/614/88 (Hauri et al., 1985)
Tubulin ((cid:97), (cid:98)) rabbit polyclonal Polysciences, Inc., Warrington, PA
Villin mouse mAb MAB1671 Chemicon Int., Inc.
ZO-1 rat mAb anti–ZO-1 Chemicon Int., Inc.
D
o
w
n
the cells were kept under oligonucleotide treatment until fixed or ex- bated for 2 h in 1 (cid:109)M 21-mer biotinylated synthetic deoxyoligonucleotide loa
d
tracted. In all cases, the antisense (A19 or A19/2) and the control (random with the sense sequence complementary of A19 or random. In a second ed
or random/2) oligonucleotides were synthesized in tandem, using the same step, these biotinylated probes were localized with streptavidin coupled to fro
m
reagents, and applied in the same experiments. No toxic effects were ob- Texas red (Jackson ImmunoResearch Labs. Inc.). h
served with the 41 batches of oligonucleotides used in this study, except Colocalization of villin and CK19 imposed a special difficulty since ttp
one case, in which the cells died almost immediately with two of the oligo- both antibodies available to us were monoclonal (mouse). In addition, the ://ru
p
nucleotides. In preliminary experiments we could not detect any differ- anti-villin antibody was found to react only in methanol fixed cells and not re
ences in the seeding efficiency of cells kept in oligonucleotides and those after aldehyde fixations, while the morphology of IF after methanol fixa- ss.o
ofoteiftor idnuenesn ctoareenfs a dittn hen edtohs necee t rmlolesloai gotreopfd nht uhocceleol lesgloasy.mt ioDdere a lpsti naowbela.ai rtsLhiet iy kos oeetwhfa ercicsreeh ll,kes wsn k oeteow pf ontc u hinsnee dcrqka unn epodno oncsmoesist bio clwleeig aeiobrnenlte eu prcdaeliecrf---- toamintoiitnnnib, y uwloasdiatneiseghds e R tadhsC e ifnK oa 1lPnl0oBt8iw-S Cws, .Kia Tns1c hu9ues bR ucaaeCtllellKysd 1 wpw0oe8iotr hemr .f1 ATi%rosbt c gafiilnrxocdebu dupm leiinvnre f-monfrrteem tbehoe aBdtnhS ot Aphl re(,o (cid:50)5ibn02lc e0(cid:109)um(cid:56)bgCs/am),t wifloo epnr r b2oei0--f rg/jcb/article-p
formed using the Fasta algorithm (Genetics Computer Group, Madison, WI). immune goat IgG for 20 min, and with the anti-villin mAb for 20 min. The df/1
monolayers were then washed in PBS and post-fixed in 3% PFA. The rest 3
7
Antibodies and Reagents obfo dthiees s t(ewpist hw ewraes hase sd einsc rPiBbeSd baebtowveee. nT heea cohr daenrt iobro tdhye aanpdp litchaet iofonl loofw ainntgi)- /2/359
was: FITC-coupled goat anti–mouse IgG (to detect the anti-villin antibody /12
TAhlle s perciomnadrayr ysp aenctiifbico adnietsib woderiees a ufsfiendi tiyn pthuirsi fsiteudd ayn adr eh dades ncroi bcerdo sisn- rTeaabctliev Ii-. applied before the PFA fixation), 50 (cid:109)g/ml preimmune mouse IgG (to 7020
quench any remaining binding sites in the anti–mouse antibody), biotiny- 4
ties with immunoglobulins of other species other than the specific target /1
(Jackson ImmunoResearch Labs. Inc., West Grove, PA). Cross-reactivity, lated RCK108 anti-CK19 mAb, and Texas red coupled streptavidin. 886
Fluorescence on filter-grown cells was performed as described above 6
however, was routinely checked by Outcherlony agar diffusion assay be- .p
ftsEuoapsuireenegc ei eitcdfn iwo eefldar,o os Oc bmadRylr i Siz)teia hdgwtem iawo msani t kCaheen hxNpupetf2me a iarcniinctm udaarel d esnCtirtol.osu c..Ft ke(PI SdTset orC1. olL:u–4xtop0iiduh o(aianvssl o,el ioMln-/civ domOoiunle))p t a(ilhneMnad dPno oaBulenlsS ceta.uidtbl ao(cid:50)artd2 tiP0ehr(cid:56)seCo wdb. ieeBlrusee tIf inooocnrbe.s-, wfmliaonodiun rtd1 hcgte0hh edt% erhl ,o fe a run ponfsgoomtueill-alry Nob lvtlwoayhin(cid:49)it an4hy–n g lKh s- oaei,(cid:49)d nxlacAe clcosweoT hpoavPotfyei alosrt,s h snwe3lesi 0pi :a tf%s(nhi,lat t u)igegb serrlou;eny da(actblylgele);yer o( nacoclget)v,si l etl1i(sanra% ntcwniuio tgebninhrba-e.top t; T idnraohoionneeptdss y p fow(lie ldrtigr )tesmah trwlr selaea aawnptsbethteiia,eblr vies1ozi :eddwm1dii0ene o0res)ue xww(n cavteeeloesrrpldoeet/ df by guest on 06 M
Immunofluorescence and In Situ Hybridization vcoovl)e rSelodw wFitahd ea™ co (vMeroslliepc,u alanrd Parlolobwees,d I tnoc .h) awrditehn t ohvee mrnoignhotl apyreers sfeadci nwgi tuhp a, arch 2
0
50-g weight on the coverslip. Before using these prepartions for confocal 2
3
The procedures for immunofluorescence were described before (Vega- microscopy, the coverslips were glued to the glass slides with nail polish.
Salas et al., 1987a,b) and were used with the following modifications. For Standard epifluorescence was performed with a Leitz DM RB micro-
microtubule localization, the cells were permeabilized for 1 min before scope (Leica Instrs., GmbH, Wetzlar, Germany), equipped with a Leica
fixation in 0.1% saponin, 80 mM Pipes, pH 6.5, 5 mM EGTA, 2 mM MgCl, Orthomat E microphotography system, using a 63(cid:51) (1.4 NA) infinity-cor-
2
1 mM GTP, 0.1 mM AEBSF at 37(cid:56)C, and then fixed in 3% formaldehyde rected objective. Kodak TMax 400 ASA film was used for photography.
(freshly prepared from paraformaldehyde), 0.1% glutaraldehyde, 1 mM Laser confocal microscopy was performed with an Odyssey XL (Noran
MgCl, 0.1 mM CaCl in PBS, also at 37(cid:56)C. Next, standard washes in PBS Instruments, Inc., Middleton, WI) microscope, using an Omnichrome la-
2 2
and quenching of aldehyde groups in 50 mM NHCl were performed. We ser source. For colocalization experiments (FITC/Texas red), a second de-
4
checked other procedures, such as the use of sodium borohydride, but did tection channel was used with a 580-nm secondary dichroic. To increase
not find an improvement in morphology. In the case of CK19 localization, resolution in the z axis, a 15-(cid:109)m slit was routinely used. The images were
the fixation procedure depended on the mAb to be used. RCK108 and collected using Intervision software (Noran Instruments, Inc.). Each con-
K4.62 gave excellent results with formaldehyde fixations. MAB1675 and focal section was obtained as the average of 64 frames. The sections were
A53-B/A2 were best when used on methanol-fixed or -unfixed tissues. collected at 0.2 (cid:109)m intervals in the z axis with 650 dots per inch resolution
The morphology of IF, however, was best preserved after aldehyde fixa- (nearly cubic voxels) through a 63(cid:51) oil immersion objective. Usually, each
tion. In all cases, 0.1% Triton X-100 was added to all the solutions during field comprised 60–90 confocal sections. For three-dimensional recon-
the processing. Localization of filamentous (F) actin with FITC–phalloi- struction we used Intervision software. The stacks of sections were cut in 4
din was done on formaldehyde fixed cells and permeabilized with 0.1% pixel thin volumes (in the x axis), reconstructed, and rotated 90(cid:56) to obtain
Triton X-100. For in situ hybridization, oligonucleotide treated cells were a view of the apical–basal axis. The images were converted to TIFF format
fixed in 1% formaldehyde, permeabilized in 100% methanol at (cid:50)20(cid:56)C, and photographed with a 4,000 line resolution Personal LFR Plus laser
washed in PBS, and quenched in 50 mM NHCl. The cells were then incu- camera (Lasergraphics, Irvine, CA) in 35-mm Kodak TMax 100 ASA film.
4
Salas et al. Apical Submembrane Cytoskeleton 361
Transmission Immunoelectron Microscopy and oligonucleotides as described above. The filters were then mounted on
Scanning Electron Microscopy plastic frames glued with silicone grease to separate apical and basolateral
compartments, and the monolayers were biotinylated as described above.
Immunoperoxidase for transmission electron microscopy was performed The cells were subjected to two consecutive extractions as described, and
on cells grown on Transwell™ filters following the protocol of Brown and the Triton X-100 supernatant was pooled with the Triton X-114 detergent
Farquhar (1984), slightly modified for tissue culture cells (Vega-Salas et al., phase. The biotinylated proteins were then affinity purified in batch with
1987b). For scanning electron microscopy, the cells were grown on glass 50 (cid:109)l gel/filter of streptavidin coupled to agarose beads (Pierce). After ex-
coverslips, fixed in 3.5% glutaraldehyde and 2% OsO, processed by criti- tensive washes in the Triton X-100 extraction buffer supplemented with
4
cal-point drying, gold coated, and observed with a JSM-35 (Jeol Ltd., To- 1% Triton X-100 and 600 mM KCl, the beads were eluted in 1 ml 1%
kyo, Japan) scanning microscope, using Polaroid 55 film (Polaroid, Cam- SDS, 5 M urea, 50 mM (cid:98)-mercaptoethanol in 2 mM Tris-Cl buffer for 2 h.
bridge, MA). The eluates were acetone precipitated, resuspended in SDS sample buffer,
run in SDS-PAGE, and blotted onto nitrocellulose sheets. Specific pro-
teins were detected by immunoblot with antibodies against sucrase isomalt-
PAGE and Immunoblot
ase or alkaline phosphatase and a chemiluminiscence detection system.
Cytoskeletal preparations were obtained from monolayers grown on fil- The signal intensity was measured as described above using the average
ters by extracting them in PBS supplemented with 1% Triton X-100, 2 mM pixel values in each band. Because in this case significant differences in
EGTA, 1.5 M KCl, 15 mM (cid:98)-mercaptoethanol, 1 mM AEBSF, and 2 (cid:109)g/ml the size of the bands were found, we also estimated a total signal value
aprotinin (both from Calbiochem, La Jolla CA; EB TX-100) at 4(cid:56)C for 10 (weighted signal) as pixel average (cid:51) number of pixels in the band.
min. The monolayers were gently scraped from the filter with a rubber po-
liceman, sonicated on ice for 30 s, and spun at 15,000 g for 5 min. The pel- Cryostat Sections of Organs
lets were denatured in SDS sample buffer containing 8 M urea for 3 min at
95(cid:56)C. Small aliquots from these extracts were acetone precipitated, resus- Samples of organs were obtained from biopsies (human) or from killed
pended in water, and TCA prepitated, and the protein was measured by animals (rat, mouse), immediately embedded in TBS tissue freezing me-
PTehtee rvsoolnu’ms em oof dtihfeic saatimonp leosf wthase thLeonw srlyig’sh tmlyi cardojmusettehdo tdo (sPeeedte 4rs (cid:109)ogn ,p 1ro9t7e7i)n. dCiruymos t(aEt l(e(cid:122)ct5r o(cid:109)nm M thicinro) ssceocptiyo nSsc wieenrcee sm, oFut.n tWeda sohni ngglatossn s, liPdAes) ,a nadn ds tofrroezde ant. Dow
per lane. These samples were run in SDS-PAGE (Laemmli, 1970) and (cid:50)20(cid:56)C. Some sections were immersed in 100% methanol and then pro- nlo
blotted onto nitrocellulose sheets (Towbin et al., 1979). The signal of pri- cessed, while others were directly processed for immunofluorescence. We ade
mmoaruys em IgAGb sc owuapsl erde vtoe apleerdo xuisdiansge aa nsde cao cnhdeamryi luamffiinniitsyc-epnucrei fsiyesdt egmo a(tP iaenrctie–, foouutn fdix natoi osnig wnihfiecna ntht ed isflfiedreesn wceesr ein o tbhsee rdviestdr iibmumtioend ioaft eClyK (1u9p w toit h4 ohr) wafitther- d from
Rockford, IL) on preflashed X-ray film. The ratio of signals between spe- the antibody incubations. http
cific bands was estimated by digitizing the image (Quantimet Q500 Image ://ru
Aavnearalygseis o sfy asltle pmix; eLl eviaclau)e sfr ionm th Xe -arraeya foilfm o naen bda cnodm apnadr itnhge athvee rraagteio o fo pf itxheel Results press.o
values in a similar area on the other band after subtracting the back- rg
ground measured from an irrelevant area (band OD from digitized im- A 21-mer Antisense Oligonucleotide Reduces the /jcb
athgeers )c. oWrrheecnte tdh eb ya rtehaes onfu tmwboe br aonfd sp iwxeelrse idni fefearcehn tb, athneds e(w veailguhetse wd eOreD fu (cid:53)r- Steady-state Levels of Cytokeratin 19 in Epithelial Cells /article
average pixel values (cid:51) number of pixels in the band). in Tissue Culture -pd
bouRnedp raonbtiinbgo doife sn iwtriothc e1ll0u0lo smeM sh e(cid:98)e-mts ewrcaasp tdooenteh abnyo ls, tr2i%pp inSgD Sp,r e6v2i omusMly Our previous observation of an apically distributed cyto- f/137/2
TrisCl, pH 6.7, at 55(cid:56)C for 30 min. The membranes were then extensively keratin displaying significant peptide homologies with cyto- /35
9
washed in PBS, 0.1% Tween-20 for 2 h (four washes), and reprocessed for keratin 19 was further confirmed by cross-reaction of anti- /1
2
immunoblot. This procedure resulted in slightly increased background bodies. Most mAbs against CK19 did not cross-react with 702
lsehveeelts., but could be applied for consecutive cycles to a single nitrocellulose canine cells, including the mAb against nonhuman CK19 04/18
(MAB1675). One of them (K4.62) however, did recognize 86
6
the same protein that our polyclonal antibody localized in .p
Plasma Membrane Polarity Assays d
Biotinylation of apical or basolateral membrane proteins on filter-grown tChoen avpeircsaell yd,o omuari np oolfy cMloDnCalK a ncteilblso d(Ry oadgraiignuset zt heet aalp.,i c1a9l9 c4y)-. f by gue
monolayers using the membrane impermeant biotin derivative sulfo-NHS– tokeratin in MDCK cells recognized the same band in im- st o
bTioo tsienp (aPriaetrece d) ihffaesr beneet nc adteesgcorirbieesd oefls mewehmebrer a(nReo pdrriogtueeinz-sB, tohuela cne ellts awl.e, 1re9 8e9x)-. munoblots from human cells as all of the anti-CK19 mAbs n 06 M
tcryaccletes da tt w30ic(cid:56)/e4 (cid:56)cCo;n Bseocrudtiievre, l1y9, 8t1h)e i nfi rPsBt tSi,m 2e m inM 2 E%D TTrAit,o 1n0 X m-M11 4N (Hp4uCrilf, i1e dm bMy (laTra dbilset rIi)b. uItti oconn (ssisetee nTtalyb lseh oIIwI)e din t hime smaumneo aflpuiocrael sscuebncceel luas- arch 2
Aat E0B(cid:56)CS Ffo, r2 1 (cid:109)5g m/minl . aTphroet pineilnle, tasn wde 1r0e s(cid:109)pMun E in-6 t4h (eP cBoSld-E (1D5T,0A00; Cg,a 5lb mioicnh)e amn)d, the anti-CK19 mAbs in tissue culture cell lines and epithe- 023
the supernatants incubated at 30(cid:56)C for 3 min to separate the detergent lial tissues from biopsies (RCK108 and A53-B/A2, Fig. 2, g
phase of Triton X-114, which was further acetone precipitated and resus- and h, and see Fig. 11). On the basis of these experiments
pended in SDS sample buffer. The pellets from the first extraction were we confirmed that human CK19 also exhibits an apical dis-
then sonicated for 30 s in 1% Triton X-100 in PBS-EDTA and warmed to tribution in various epithelia. To assess whether these api-
37(cid:56)C for 15 min. These resuspended pellets were spun again at room tem-
cal CK19 intermediate filaments play any role in the orga-
perature (15,000 g, 5 min) and immediately resuspended in SDS sample
nization of the apical region, we attempted its knockout
buffer. The supernatants from the second extraction in Triton X-100 were
acetone precipitated and resuspended in SDS sample buffer. In some ex- using a phosphorothioate deoxyoligonucleotide with the
periments, the extraction in Triton X-100 was replaced by an extraction in antisense sequence of the first 21 bases in the open reading
60 mM n-octyl-(cid:98)-d–glycoside (Anatrace Inc., Maumee, OH) in PBS– frame of human CK19 mRNA (Eckert, 1988), hereafter
EDTA at 4(cid:56)C. The total protein in each SDS extract was measured as de-
referred to as A19 (Fig. 1). As a control, we used another
scribed in the previous section. Samples with equal amounts of total pro-
tein for each extraction procedure were run in SDS-PAGE and blotted 21-mer phosphorothioate deoxyoligonucleotide, with the
and biotinylated proteins detected with Extravidin-peroxidase™ (Sigma same bases but in a randomized sequence (Fig. 1, random).
Chemical Co.) and chemiluniscence. In these cases, a second set of biotin- In addition, another 21-mer antisense oligonucleotide, com-
ylated molecular weight standards (Sigma Chemical Co.) was used in ad- plementary to the last 11 bases in the 5(cid:57) UTR and the first
dition to the usual Coomasie blue–labeled standards.
10 of the ORF of CK19 mRNA (A19/2) was also used in
For detection of specific proteins, the cells were grown on 140-mm poly-
carbonate filters overlayed with tissue culture medium supplemented with some experiments to further confirm the specificity of the
The Journal of Cell Biology, Volume 137, 1997 362
Figure 1. Sequences of the 5(cid:57)
end of the reading frame of
CK19 cDNA (Y00503; Swiss-
prot P08727; Stasiak and
Lane, 1987; Eckert, 1988),
antisense (A19 and A19/2),
and the corresponding ran-
dom deoxyoligonucleotides.
effect and controlled with the corresponding randomized
sequence oligonucleotide (Fig. 1, random/2). All the se-
quences were searched in the GenBank database. The 32
bases in the CK19 message (to which A19 and A19/2 were
antisense sequences) showed high homologies with CK19
from other species but not with other known human
mRNAs. The only other significant homologies found in
vertebrates were proteins expressed in hippocampal neu-
rons in rat (85% identity in 20 bases; these sequence data
are available from GenBank/EMBL/DDBJ under accession
no. L26525) and in T-cells in mice (89.5% identity in 19
D
o
bases; under accession no. M16122). Therefore, CK19 is w
n
likely to be the only protein targeted by A19 or A19/2 an- loa
d
e
tisense oligonucleotides. In the case of the random oligo- d
nucleotides, no significant reverse/complemented homolo- from
h
gies were found in higher eukaryote sequences. For all the ttp
experiments, the random oligonucleotides were synthe- ://ru
p
sized at the same time with the same reagents and purified re
in parallel with the corresponding antisense oligonucle- ss.o
rg
otides for CK19. /jcb
daIynsc duibda tnioont ss hoofw c oannfylu eefnfet cmt oonn otlhaey eCrsK i1n9 Aco1n9t efnotr ian fiemw- /article
-p
munoblot experiments. The reasons for these early failures df/1
3
are most likely to be found in the long turn-over time of 7
cytokeratins (Denk et al., 1987). In brief, it was critical that FMigCuFr-e1 20.AU apntda kCeA aCndO e-2ff eccetl los.f ManCtiFse-1n0seA ( A(a1 a9n) do lbig)o annudc lCeoAtiCdOe i-n2 /2/359
tphaes scaugletus)r eins thhaed p troe sbeen cceo notfi nAu1o9u aslty c oknepcet n(tfroart iaotn lse (cid:46)as1t (cid:109)twMo ctienlulso (ucs–lyj) gwroewren ciunl teuitrheedr orann gdloamss (cao,v ce,r es,l igp,s i.) T ohre A c1e9ll s( bw, edr, ef , cho, nj)- /127020
to reduce CK19. We maintained small stocks of cells under oligonucleotides. In some cases, the cells were fixed (PFA), per- 4/1
8
continuous treatment to minimize the consumption of oli- meabilized, and processed with a biotinylated sense oligonucle- 86
6
gonucleotides. To keep the cellular mass continuously in- otide (complementary of A19, a–d; or complementary of random, .pd
creasing, these stocks were never allowed to reach conflu- e and f) followed by Texas red–conjugated streptavidin. Other f by g
ency. The uptake of oligonucleotides was assessed by in-situ monolayers were processed for indirect immunofluorescence ue
hybridization with the biotinylated sense synthetic oligo- with anti-CK19 mAb (RCK108; g–j). Some samples were ob- st o
n
nucleotide. Two human cell lines, CACO-2 (colon carcinoma served under laser confocal microscopy as a series of confocal op- 06
tical sections in the z axis (perpendicular to the plane of the M
cells that differentiate and polarize 7–14 d after reaching a
confluency; Pinto et al., 1983) and MCF-10A (nontumori- mtaoinneodl aayse ar )t,h arnede- dthimene nthsieo tnraaln rsemcoonnsotlrauycetrio sne cotfi o4n v iomxaegl eth wicaks voobl-- rch 20
genic mammary epithelium; Soule et al., 1990; Tait et al., umes (g and h). Black arrowheads point to the apical CK19 net- 23
1990) treated as described above with A19, displayed a work, and white arrowheads show the position of the basal domain.
clear signal in intracellular vesicles, presumably the endo- Other samples are shown desuper under standard epifluores-
somal compartment, and in the cytoplasm (Fig. 2, b and d) cence microscopy (i and j). Bars: (a–f, i, j) 10 (cid:109)m; (g and h) 20 (cid:109)m.
as compared with controls grown in the random oligonu-
cleotide (Fig. 2, a and c). The uptake of A19 was higher in
subconfluent cells than in confluent monolayers. Likewise,
the uptake of random oligonucleotide was controlled by in immunoblot. The overall proportion of success of antisense
situ hybridization with a biotinylated complementary treatment was better observed by regular immunofluores-
probe (Fig. 2, e and f). These results are in agreement with cence. In control CACO-2 monolayers (incubated in ran-
observations in other cells in culture (Loke et al., 1989; dom oligonucleotide, Fig. 2 i) the vast majority of the cells
Noonberg et al., 1993) and demonstrated that CACO-2 displayed a network of fluorescent filaments in a focal
and MCF-10 cells do uptake 21-mer phosphorothioate oli- plane above the nucleus. The images in nontreated cells
gonucleotides, a phenomeonon that seems to be depen- were identical (not shown). Treatment with A19, on the
dent on the cell type (Crooke et al., 1995). other hand, resulted in 20–30% of the cells totally negative
The effect on the steady-state content of CK19 was de- for CK19. An additional 30–40% of the cells displayed a
termined by immunofluorescence, immunoperoxidase, and signal level clearly lower than in controls, but the remain-
Salas et al. Apical Submembrane Cytoskeleton 363
Figure 3. Immunoelectron
microscopy localization of
CK19 in MCF-10A cells. The
cells were continuously grown
in either random (a, b, and d)
or A19 (c and e) oligonucle-
otides on Transwell™ filters,
fixed, and processed for indi-
rect immunoperoxidase with
anti-CK19 mAb (K4.62). (a
and c) Apical cytoplasm; (d
and e) basal cytoplasm; (b)
lateral domain. The black
arrowheads point at peroxi-
dase positive intermediate fila-
ments. (D) White arrowheads
show desmosomes; Nu, nu-
cleus; *, microvillus. Bar: (a
and c–e) 0.52 (cid:109)m; (b) 0.35 (cid:109)m. D
o
w
n
lo
a
d
e
d
ing proportion of cells (30–40%) showed a level of CK19 The effect of the antisense oligonucleotides was also de- from
h
comparable with the controls (Fig. 2 j). If the cells were termined by immunoblot. SDS extracts of cytoskeletal pel- ttp
kept in A19 for periods of confluency (cid:46)9 d, this effect lets from CACO-2 cells incubated in A19 showed a clear ://ru
p
slowly vanished. The same was true if the cells were trans- decrease in CK19 (Fig. 4, lane B, 37% of the signal in the re
ferred back to media containing random oligonucleotide band in lane A) as compared with controls kept in random ss.o
rg
or no oligonucleotide at all (not shown), thus indicating oligonucleotide (Fig. 4, lane A). A similar effect was ob- /jcb
tcheallts cthoinst idnouwounsrleyg iunlcautiboant ewd aisn Afu1ll9y srheovweresdib al ed. eCcrAeaCsOe i-n2 sneitrrvoecde lwluitlho sAe 1s9h/2e e(tF iwg.a 4s , lsatrniepsp Ked a nadn dL )r.e Wprhoebne dth ew sitahm ea /article-p
the CK19 signal, but we could not detect loss of polarity of mAb against other cytokeratins or a mAb against all actin df/1
3
the remaining CK19 (Fig. 2 h). Monolayers of CACO-2 isoforms, no differences were observed (Fig. 4, lanes C–F, 7/2
cells observed by laser confocal microscopy in the apical- M, and N), indicating that the effects of A19 and A19/2 /35
9
basal axis displayed the typical apical distribution of CK19 were specific for CK19. Likewise, MCF-10A cells continu- /12
7
(Fig. 2, g and h, black arrowheads; white arrowheads point ously grown in A19 displayed decreased levels of CK19 02
0
at the basal domain). Other cytokeratins (8 and 18; Moll (Fig. 4, lane H) as compared with the controls in random 4/1
8
et al., 1982) localized to both apical and basal submem- oligonucleotide (Fig. 4, lane G), and a comparable effect of 86
6
brane cytoplasm in MCF-10A cells (not shown). A19/2 (Fig. 4, O and P). Reprobing the same nitrocellu- .pd
Because the fluorescence label was dim in MCF-10A lose sheet with the anti-pan cytokeratin mAb showed, again, f by g
cells, we repeated these experiments using immunoperoxi- no effect on other cytoskeletal proteins (Fig. 4, lanes I, J, ue
dase at the EM level. Control monolayers incubated in ran- Q, and R). These results were quantitatively determined st o
n
dom oligonucleotide showed bundles of CK19 filaments in by analysis of the digitized images from chemiluminescence 06
M
the apical cytoplasm (Fig. 3 a, arrowheads), at variable dis- detection (Fig. 4, legend). It must be pointed out that three arch
tances from the apical membrane. The minimum distance to four bands were detected by RCK108 mAb in MCF- 2
0
from these filaments to the apical membrane was (cid:122)50 nm 10A cells, with apparent molecular weights ranging from 23
and the maximum up to 3 (cid:109)m (sections from 22 cells). In a 44 to 53 kD, in some preparations. Interestingly, the same
few cases, some filaments were observed around and be- pattern of bands was displayed by K4.62 mAb in some
low the nucleus but never close to the basal membrane. preparations from MDCK cells. Our polyclonal Ab consis-
These filaments were never observed inside microvilli. In tently displayed affinity for the 53-kD band. We have no
some cases, CK19 positive filaments were viewed along explanation for these multiple bands, although we specu-
the lateral membrane (Fig. 3 b, arrowheads) in close vicin- late that different states of phosphorylation may be re-
ity to the desmosomes. The majority of the sections of the sponsible for this type of pattern. On the other hand, it has
lateral domain, though, were negative for CK19. The basal been shown that mammary epithelia display a complex
cytoplasm on the other hand was always negative (Fig. 3 d). pattern of cytokeratins, including at least one that is in-
Monolayers incubated in A19 displayed nearly 50% of the duced by extracellular matrix (Hall and Bissell, 1986). There-
apical cytoplasm sections without or with very few fila- fore, it is also possible that RCK108 may be recognizing an
ments (Fig. 3 c). In these cases, the filaments observed were epitope in other cytokeratins from MCF-10A that are not
found farther away from the apical membrane, with a min- present in CACO-2 cells. If that was the case, these other
imum recorded distance of 320 nm (sections from 18 cells), cytokeratins may also share a common 5(cid:57) sequence in the
and more scattered throughout the cytoplasm (Fig. 3, c reading frame of their respective mRNAs, since they were
and e). also downregulated by A19.
The Journal of Cell Biology, Volume 137, 1997 364
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
://ru
Figure 5. Scanning electron microscopy of the apical surface of pre
MCF-10A (a and b) and CACO-2 (c and d) cells continuously ss.o
grown in random (a and c) or A19 (b and d) oligonucleotides. rg
Specimens were tilted 60(cid:56) for photography. Bar, 1 (cid:109)m. /jcb
/a
rticle
-p
d
Figure 4. Immunoblot of cytoskeletal components in cells contin- Effect of Cytokeratin 19 Antisense Oligonucleotide on f/137
uously kept in random (A, C, E, G, and I), A19 (B, D, F, H, and Microvilli and the Apical F-actin Cytoskeleton /2/3
5
Jn)u, crlaenodtiodmes/.2 C(KA,C MO,- 2O ,( Aan–dF Qan),d o Kr –AN1)9 /a2n (dL M, NC, FP-,1 0aAnd (RG)– oJ liagnod- An intriguing observation in the immunoperoxidase ex- 9/127
O–R) cells were grown on 24-mm Transwell™ filters and ex- periments was the decrease in the number of microvilli in 020
tracted/scrapped from the filter in EB-TX- 100. The pellets were those cells with reduced levels of CK19 filaments (Fig. 3, c 4/18
further extracted in SDS sample buffer, and samples from equiv- versus a, *). To further explore this phenomenon, MCF- 866
alent numbers of cells were run in SDS-PAGE ( the protein in 10A (Fig. 5, a and b) and CACO-2 cells (Fig. 5, c and d) .pd
euamceh asdamjupstleed w taos s meeeda s4u (cid:109)regd o ifn p aro stmeianl li na laiqllu loatn easn;d u tshuea lslya mthpelsee v aodl-- cwoenrtei noubosuesrlvye dg rboyw nsc ainn nrianngd eolmec toror nA m1i9c roolsicgoopnyu.c lCeoontitdreosl f by gue
justments represented variations in sample volumes that were MCF-10A cells displayed a relatively modest amount of st o
(cid:44)10% of the total volume) and immunoblotted onto nitrocellu- short apical microvilli (Fig. 5 a), which was significantly n 06
lose filters. The blots were processed for indirect chemilumines- M
decreased in nearly 30% of the cells in monolayers treated a
csaemncee n uitsrioncge allnu laonseti s-CheKe1ts9 u mseAd bfo (rR AC, KB1, G08,; HA, ,K B, ,L G, O, a, nadn dH P) .w Tehree with A19 (Fig. 5 b). Control CACO-2 cells at 9 d of conflu- rch 20
then stripped in SDS/2-mercaptoethanol and reprobed with anti- ency showed a fully developed apical domain with abun- 23
Pancytokeratin mAb (AE1/AE3; C, D, I–J, M, N, and O–R). The dant long microvilli (Fig. 5 c). In treated cultures, 30–40%
same filter from CACO-2 extract was stripped and reprobed one of cells were totally depleted in microvilli (Fig. 5 d) and a
more time with anti-actin mAb (C4; E and F). Notice the de- similar proportion of cells had a decreased number of mi-
crease in CK19 bands in A19 treated cells (B and H) as compared crovilli, as compared with the control.
with the control with random oligonucleotide (A and G), and that Because the microvillus core contains F-actin, these re-
other cytokeratins and actin in the same samples were not sults prompted us to study the distribution of F-actin in
changed (C–F, I, and J). The average OD measures obtained from
oligonucleotide-treated cells. Confocal optical sections of
unfiltered digitized images after subtracting background from the
FITC–phalloidin stained cells at the apical level, under con-
average pixel value over the main band of each lane were as fol-
trol oligonucleotides, displayed a typical punctate pattern
lows (scale 0–255): (A) 82; (B) 12; (C) 77; (D) 100; (E) 91; (F) 81;
in all cells (Fig. 6 a). In this case, occasional “black” or
(G) 30; (H) 6; (I) 28; (J) 21; (K) 150; (L) 27; (M) 171; (N) 172; (O)
49, (P) 5; (Q) 160; (R) 174. The apparent molecular weight of darker apical domains always corresponded to taller or smal-
standards is expressed in kD. ler cells that became fully positive by adjusting the focal
plane. In A19 treated monolayers on the other hand, 50–
70% of the cells displayed negative or low signal on the
apical domains, contrasting with the still positive fluores-
cent rings at the cell–cell contacts. As an internal positive
Salas et al. Apical Submembrane Cytoskeleton 365
Figure 6. Effect of antisense
A19 oligonucleotide on the
apical distribution of apical
F-actin in CACO-2 cells. The
cells were continuously grown
in random (a, c, and e; con-
trol) or antisense A19 (b, d,
and f) oligonucleotides. For
this experiment, the cells were
plated on glass coverslips and
fixed in PFA after 9 d of con-
fluency. The monolayers were
detergent permeabilized and
processed with FITC–phal-
loidin. The monolayers were
observed under a laser con-
focal microscope. Confocal D
o
optical sections immediately wn
underneath the apical mem- loa
d
brane (a and b) or immedi- ed
ately above the basal mem- from
brane (c and d) were chosen http
from the stack of sections in ://ru
the z axis. The sections in a p
re
and c correspond to the same ss.o
field at different focal planes, rg
and the same applies to the /jcb
/a
fsieecltdi,o an st hinre eb- dainmde nds. ioInn ael arceh- rticle-p
construction section perpen- df/1
dicular to the plane of the 37
/2
monolayer is shown in e and /3
5
f. Black arrowheads point at 9/1
2
F-actin negative apical re- 70
2
gions in A19 treated cells. 04
Bars, 10 (cid:109)m. /18
8
6
6
.p
d
f b
y g
control, one cell that escaped the effect of A19 is shown on poses, cells with reduced amounts of CK19 but still show- ue
the upper right corner of Fig. 6 b. These positive cells were ing a CK19 IF network were ranked as positive for CK19. st o
n
used as landmarks to determine the appropriate location Only one of every seven cells negative to CK19 (CK19(cid:50)) 06
M
of the otherwise negative apical surfaces in the z axis. Nei- showed villin signal (Table II). On the other hand, 41% of a
ther the submembrane F-actin under the lateral domains the cells positive to CK19 ( CK19(cid:49)) were also positive for rch 2
0
(not shown in x–y sections) nor the basal network of stress villin (Table II). This experiment, in general, showed a good 23
fibers (Fig. 6, c and d) showed any noticeable changes un- correlation between the loss of CK19 IF and the absence
der the A19 treatment. Three-dimensional reconstructions of microvilli. We cannot explain the 33% of the cells that
in the x–z plane (perpendicular to the plane of the mono-
layer) showed a summary of the effect of CK19 antisense
oligonucleotide: only apical F-actin was affected (compare
Table II. Apical Expression of Villin in CACO-2 Cells Treated
Fig. 6 f, black arrowheads with control; e, white arrow-
with A19 Oligonucleotide
heads point at the basal side). It must be noted that the to-
tal cellular levels of actin did not change in A19 treated Cells with microvillum Cells without apical
images of villin image of villin
cells (Fig. 4, lanes E and F), suggesting that the differences
were in the polymerization and organization, not in the % %
expression of actin. The correlation between the decrease CK19(cid:49) 23 33
in the number of microvilli and the effect of A19 was fur- CK19(cid:50) 6 38
ther analyzed by colocalization of villin and CK19 in A19 CACO-2 cells treated with A19 oligonucleotide as described before were sequentially
treated monolayers. In this experiment we analyzed the processed for double immunofluorescence with anti-villin mAb and biotinylated anti-
CK19 mAb (RCK108; see Materials and Methods). CK19(cid:49) refers to cells with CK19.
appearence of the typical apical image of fluorescence for
Cells with lower amounts of CK19 but still showing IF were counted in this category.
villin in cells with or without CK19. For counting pur- CK19(cid:50) cells were those with no detectable CK19.
The Journal of Cell Biology, Volume 137, 1997 366
Figure 7. Effect of antisense
A19 oligonucleotides on the
distribution of tubulin in
MCF-10A cells. The cells were
continuously grown in ran-
dom (a; control), random/2
(c; control), antisense A19
(b), or A19/2 (d) oligonucle-
otides. For this experiment,
the cells were plated on lami-
nin-coated glass coverslips
and fixed in 3% PFA, 0.1%
glutaraldehyde after 5 d of
confluency. The cells were
permeabilized, fixed, and pro-
cessed for indirect immuno-
fluorescence with anti-tubu-
lin polyclonal antibody and D
o
observed under a standard wn
lo
epifluorescence microscope. a
d
e
Black arrowheads in b and d d
point to abnormal spheric from
clusters of tubulin. Bar, 10 (cid:109)m. http
://ru
p
re
ss.o
dboe sepxepcruelsast eCdK t1h9a ta nthde sseti lcl ellalcs km mayic broev dilelil,a ayletdh oinu gthh eitir m daify- t(hFiicgk. 8( a3)–.4 T h(cid:109)ims n) entweotwrko rekx teonf dmedi cfrrootmub tuhlee sle vweal so fo tbhsee rtivgehdt rg/jcb/a
ferentiation process. junctions to the cytoplasm immediately above the nucleus. rticle
At the nuclear level, the microtubules were mostly ori- -pd
Effect of Cytokeratin 19 Antisense Oligonucleotide on ented in the apical–basal axis. It must be noted that these f/13
7
the Organization of Apical Microtubules apical–basal microbutules appear in confocal sections as /2/3
5
We next examined whether cytokeratin 19 antisense oligo- small dots, giving an impression of poor preservation. 9/12
nucleotides were able to induce changes in the distribution Standard epifluorescence of these preparations showed 702
of microtubules. Indirect immunofluorescence against tu- images comparable with those in Fig. 7. Finally, the basal 04/1
bulin in MCF-10A cells permeabilized with saponin before cytoplasm showed another network of microtubules, thin- 886
fixation showed a conspicuous pattern of microtubules ner than the apical one (Fig. 8 c). These results are coinci- 6.pd
msearvneyd o fb yw hciocnhf owcearle mini carno sacpoipcya l–(nboast asl hoorwienn)t.a tIino ng,e anse orabl-, dtheenstea lc ewllist.h Tthhee doobwsenrrveagtuiolantsio onf oGf CilbKe1r9t ewti tahl .A (1199 9h1a)d ina f by gue
this pattern resembles that previously described by Bacal- striking effect on the apical microtubules. Microtubules st o
n
lao et al. (1989) in MDCK cells. Cells continuously grown were, again, mostly excluded from the apical-most cyto- 06
plasm as in the controls, but the apical network showed M
in A19 (Fig. 7 b) or in A19/2 (Fig. 7 d) oligonucleotides images of thick spherically shaped (caliper diameters 0.3– arch
schoonwtinedu otshlyic kin ctuubbuatliend cilnu sttheers ,c onrorte sopbosnedrvinegd riann tdhoem cizeellds 1.2 (cid:109)m) clusters of tubulin (Fig. 8 b, arrowheads). These 2023
images were not continuous in deeper sections, indicating
oligonucleotides (Fig. 7, a and c). This phenomenon was
that they were not thick bundles perpendicular to the plane
observed in 21 to 38% of the cells.
of the monolayer. Some of these structures were also ob-
The distribution of microtubules was slightly different in
served at the nuclear level, intercalated with normal mi-
CACO-2 cells. These cells usually display a longer apical–
crotubules. In the basal cytoplasm, the fine basal network
basal axis than the MCF-10A cells. Because the nucleus is
of microtubules was indistinguishable from the control im-
located near the basal region, they show a thicker apical cyto-
ages (Fig. 8 d).
plasm. To demarcate regions within the apical cytoplasm we
colocalized ZO-1, a tight-junction marker (Stevenson et al.,
Effect of Downregulation of Cytokeratin 19 with
1986) together with tubulin. Confocal optical sections were
Antisense on the Polarity of Plasma Membrane Proteins
collected in the green channel for ZO-1 (Fig. 8, insets) and
in the red channel for tubulin (Texas red; right hand side of The effects that the decrease of CK19 filaments had on the
each pair in Fig. 8, a–d). As in Fig. 7, the cells were briefly apical but not on the basal cytoskeleton suggested the need
permeabilized before fixation to remove the signal from for a comprehensive analysis of the polarity of plasma mem-
nonpolymerized tubulin. In control monolayers (treated with brane proteins. We first decided to assay groups of abun-
random oligonucleotide), microtubules were practically ex- dant proteins in each domain of the plasma membrane.
cluded from the apical-most region of the cytoplasm, and a The approach consisted of biotinylation of apical or baso-
Salas et al. Apical Submembrane Cytoskeleton 367
Figure 8. Effect of antisense
A19 oligonucleotide on the
apical distribution of apical
tubulin in CACO-2 cells. The
cells were continuously grown
in random (a and c; control)
or antisense A19 (b and d)
oligonucleotides on glass cov-
erslips for 9 d, saponin per-
meabilized, and fixed. The
tight junction component
ZO-1 (insets) and tubulin
were localized by indirect im-
munofluorescence using spe-
cific second antibodies cou-
pled to fluorescein and Texas
red, respectively. Laser con-
focal optical sections were
taken simultaneously in the
green (ZO-1; insets) and red
at two different levels: through
the apical cytoplasm at the D
o
deep side of the ZO-1 signal w
n
(a, b) and (cid:122)1 (cid:109)m above the loa
d
basal membrane (c, d). The ed
pairs a–c and b–d correspond fro
m
respectively to the same fields h
ttp
arot wdhifefeardesn t pfooicnatl palta nteusb. uAlirn- ://rup
calluesntte rtso. B30a.r4s ,(cid:109) 1m0 (cid:109)fomr (tehqeu iinv-- ress.org
sets). /jcb
/a
rticle
-p
d
f/1
3
lateral proteins of CACO-2 C2BBe1 cell monolayers grown (Hooper and Turner, 1988; Brown and Rose, 1992; and re- 7/2
on Transwell™ filters with the membrane-impermeant sults not shown). Finally, the remaining membrane pro- /35
9
biotin derivative sulfo-NHS–biotin in a standard fashion teins in the pellet after both extractions are more likely to /12
7
(Rodriguez-Boulan et al., 1989). We used this clone of be truly associated with the submembrane cytoskeleton 02
0
CACO-2 cells because it has been characterized as better (Salas et al., 1988; Fig. 9, lanes A–D). Among the latter, 4/1
8
polarized than its parental cell line (Peterson and Moose- the polarization was variable in control cells: at least five 86
6
ker, 1992) and also to ensure the homogeneity of the cell bands were distributed to both apical and basolateral do- .pd
monolayer. Antisense treatment of C2BBe clone cells re- mains, seven bands appeared mostly on the apical side, f by g
sulted in cytoskeletal disorganization identical to the re- and 15–17 bands on the basolateral (Fig. 9, lanes A and B). ue
sults described in the previous sections for the parental A19 treated cells showed no detectable variations in the st o
n
CACO-2 cells. No significant effect of antisense treatment basolateral cytoskeletally associated membrane proteins 06
M
dwoams o vbss. e3r9v4e d(cid:54) o2n2 0t hoeh mTE (cid:51)R c (m362 4in (cid:54) A 11992), owhhmic h(cid:51) w cemre2 sinim rialnar- (pFlaigy.e d9 oDn)ly. Tthhree ea pbiacnadl sd womitha ilne,v eolns othf ee xoptrheesrs ihoann sdim, diliasr- arch 2
0
or even larger than the values reported for CACO-2 cells to the control (220, 194, and 72 kD). Two bands showed 23
(Pinto et al., 1983), indicating that tight junctions were in- decreased levels (118 and 60 kD), while the rest were al-
tact, a result also confirmed by ZO-1 staining (Fig. 8). most undetectable (Fig. 9 C). Coincidentally, the bands
The cells were extracted at 0(cid:56)C in Triton X-114, and the that were no longer expressed on the apical domain of
proteins were acetone precipitated from the 30(cid:56)C deter- A19 treated cells are in the same molecular weight range
gent phase of the supernatant (Fig. 9, lanes I–L). These (40–86 kD) as those apical membrane proteins attached to
bands generally correspond to integral membrane proteins CK19 multiprotein complexes in MDCK cells (Rodriguez
(Bordier, 1981). The pellets from the previous extraction et al., 1994). The bands extracted by Triton X-100 at 37(cid:56)C
were further extracted in Triton X-100 at 37(cid:56)C. The pro- after a previous extraction at 0(cid:56)C showed a completely dif-
teins insoluble at 0–4(cid:56)C but solubilized at a higher temper- ferent pattern. As expected for GPI-anchored proteins
ature typically correspond to (although perhaps are not (Lisanti et al., 1988), a vast majority of them localized to
exclusively) GPI-anchored membrane proteins (Brown the apical domain in control cells (Fig. 9 E). In antisense
and Rose, 1992; Fig. 9, lanes E–H). The results with this treated cells, conversely, a number of bands appeared in
type of extraction were very similar to those obtained ex- the basolateral membrane (Fig. 9 H). Finally, the proteins
tracting the monolayers with 60 mM n-octyl-(cid:98)-d-glycoside extracted and purified in the detergent phase of Triton
at 4(cid:56)C after a Triton X-100 extraction at 0(cid:56)C, another X-114 (integral membrane proteins; Bordier, 1981) showed
procedure to selectively extract GPI-anchored proteins a more complex pattern, with a number of bands polarized
The Journal of Cell Biology, Volume 137, 1997 368