Development 115, 427-437 (1992)Printed in Great Britain © The Company of Biologists Limited 1992

427

Spatial and temporal expression of an epithelial mucin, Muc-1, during

mouse development

V. M. M. BRAGA, L. F. PEMBERTON, T. DUfflG and S. J. GENDLER*

Imperial Cancer Research Fund, 44 Lincoln's Inn Fields, PO Box 123, London, WC2A 3PX, UK

•Author for correspondence

Summary

The Muc-1 mucin is found as a transmembrane proteinin the apical surface of glandular epithelia. To provideinsight into possible functions, we have assessed thetiming of expression and the distribution of the Muc-1protein during mouse embryogenesis using three differ-ent techniques: RT-PCR, northern blots and immuno-histochemistry. Our results indicate that Muc-1 ex-pression correlates with epithelial differentiation instomach, pancreas, lung, trachea, kidney and salivaryglands. Once started, Muc-1 synthesis continuallyincreases with time, mainly due to epithelial areagrowth. Our data suggest that expression of the Muc-1gene is under spatial and temporal control duringorganogenesis. Although Muc-1 is present in differentorgans, its expression is not induced systemically, butaccording to the particular onset of epithelial polariz-ation and branching morphogenesis of each individualorgan. It is of particular interest that Muc-1 protein canbe detected lining the apical surfaces of the developing

lumens when the epithelium of these organs is stillundergoing folding and branching, and glandularactivity has not yet started. We speculate that Muc-1may participate in epithelial sheet differentiation/lumenformation during early development of the organsknown to express it. This speculation is based on: (1) thedetection of Muc-1 expression early during organogen-esis, (2) the defined apical localization in differentepithelia, (3) the decrease in cell-cell interactions whenMuc-1 protein is highly expressed and (4) the possibleinteraction of its cytoplasmic tail with the actin cytoskel-eton. However, it remains to be established using in vitrosystems, whether the temporal and local expression ofthe Muc-1 gene coincident with the morphogeneticevents described here is relevant for the process.

Key words: mucin, Muc-1, PEM, epithelial polarization,mouse development, organogenesis, morphogenesis,stomach, pancreas, lung, kidney, salivary glands.

Introduction

The Muc-1 mucin is a highly glycosylated integralmembrane glycoprotein which is expressed at the apicalsurface of a wide variety of epithelial tissues. Mucins ingeneral share the common features of high molecularweights, high content of carbohydrate (50 to 90%)linked to the core protein mainly by O-glycosylationand the presence of tandem repeats, in which aminoacid number and sequence vary according to the type ofmucin molecule (Gendler et al., 1987; Gendler et al.,1990; Gum et al., 1989; Gum et al., 1990; Lan et al.,1990; Ligtenberg et al., 1990; Porchet et al., 1991;Siddiqui et al., 1988; Wreshner et al., 1990). Studieswith monoclonal antibodies to carbohydrate and coreprotein epitopes in the extracellular domain of humanMUC1 mucin (Burchell et al., 1983; Gendler et al.,1988; Siddiqui et al., 1988) revealed its expression in anumber of simple secretory epithelia present in human(Zotter et al., 1988) and transgenic mice organs (Peat etal., 1992) such as pancreas, lung, oviduct, kidney,

salivary gland, stomach, uterus and mammary gland.The presence of these same epitopes and others inmalignant mammary cell lines and adenocarcinomasrevealed that MUC1 is not only overexpressed in thesecells, but also showed differential glycosylation (Bur-chell et al., 1987). Antibody studies of MUC1 ex-pression suggest that the aberrant expression andglycosylation of this molecule can be used as differen-tiation markers of some malignant tissues. Similarevents have also been observed in studies on theexpression of Muc-1 in mouse mammary gland duringpregnancy and lactation: levels of expression, compo-sition of oligosaccharide chains and polarized distri-bution of the protein are all developmentally regulated(Parry et al., 1992).

Although mucins have been assumed to have protec-tive and/or lubrication roles in secretory epithelialtissues, the structure of the Muc-1 mucin and the factthat many adenocarcinomas express high levels of theprotein suggest additional functions. The size of thismolecule and the glycosylation may be important to its

428 V. M. M. Braga and others

function, since a transmembrane mucin with a heavilyglycosylated extracellular domain can be predicted toextend far above the plasma membrane and could effec-tively shield the surface of cells expressing it in highamounts (Jentoft, 1990). Thus, it is not surprising thatmucins have been postulated to play a role in naturalkiller cell resistance (Bharathan et al., 1990), in the es-cape of tumor cells from the immune system (Cod-ington et al., 1973), a possible involvement in inhibitionof cell growth (Shimizu et al., 1990) as well as in themetastatic potential of adenocarcinoma cells (Steck andNicolson, 1983). This latter is particularly interestingwhen the influence of MUC1 mucin on cell-cell inter-action is analysed. Stable transfection of MUC1 cDNAinto two different cell lines has shown that high levels ofexpression of this mucin reduces cellular interactions,possibly by preventing association between moleculesin adjacent cells (Ligtenberg et al., 1992).

Recent studies on apical membrane polarity, examin-ing the distribution of MUC1 in mammary epithelialcell cultures, have shown that the cytoplasmic tail of thisprotein may interact either directly or indirectly withthe actin cytoskeleton (Parry et al., 1990). Thiscytoplasmic tail of human MUC1 contains 69 aminoacids and shows the highest homology (87%) with therecently cloned mouse homologue, Muc-1 (Spicer etal., 1991). The sequence conservation and the interac-tion with actin filaments suggest that the cytoplasmictail of Muc-1 may be functionally important.

These data, taken together with the characteristicdistribution of Muc-1 protein lining the lumens ofsecretory epithelia, have led us to examine whether theexpression of this apical glycoprotein correlates withepithelial differentiation during morphogenesis. Epi-thelial formation is a continual process in embryogen-esis, and its polarization and subsequent branching playa fundamental role in organogenesis. Therefore, tounderstand more clearly the Muc-1 function in epitheliaknown to express it, we studied the local and temporalexpression of this mucin throughout mouse embryoorganogenesis.

Materials and methods

AnimalsMouse embryos were collected at different gestational ages(vaginal plug=l day) from pregnant mice (C57B1/ICRF)killed by cervical dislocation. Embryos were dissected, andplacenta and organs of interest were removed and immedi-ately frozen in liquid nitrogen or fixed in methacarn (60%methanol, 30% chloroform and 10% acetic acid). Neonatemice (2 to 8 weeks) were also dissected and their organs fixedin methacarn.

ImmunostainingParaffin blocks containing the samples were sectioned (5 ^mthick) and routinely stained with hematoxylin and eosin.Immunostaining was performed as described (Bartek et al.,1985). At least three different embryos/organs were stainedseparately. The polyclonal antiserum CT1, raised to asynthetic peptide corresponding to the 17 C-terminal aminoacids in the cytoplasmic tail of human MUC1 (Pemberton et

al., 1992) was used at 1:50 dilution. The sections were eitherincubated with preimmune serum or with CT1 antiserumpreviously blocked with 5 mg/ml of the synthetic peptide (30minutes at room temperature) for negative controls. Swineanti-rabbit peroxidase-conjugated immunoglobulins (DakoImmunoglobulins a/s, Denmark) (1:50 dilution) were used assecond antibody and colour development was obtained byincubation with 1 mg/ml diaminobenzidine (DAB, 3,3',4,4'-tetraaminobiphenyl, Sigma) and 0.03% H2O2. The slideswere subsequentially counterstained with hematoxylin.

RNA extraction and processingRNA from frozen embryos and embryo organs was extractedby the acid guanidinium thiocyanate-phenol-chloroformmethod (Chomczyski and Sacchi, 1987). RNA was resus-pended in sterile water, aliquoted and kept at —70°C untilused. Concentration and purity were estimated by opticaldensity measurements (Maniatis et al., 1982).

RNA samples were fractionated in 1.5% agarose (ICNBiochemicals) formaldehyde gels and blotted to Hybond N+nylon membranes (Amersham International pic) overnightusing standard techniques (Maniatis et al., 1982). After RNAfixation to the membrane following manufacturer's instruc-tions (5 minutes in 0.1 M NaOH solution), the membrane wasblocked and treated as described (Church and Gilbert, 1984).pMuc2TR, which contains the tandem repeat sequence ofmouse Muc-1 gene (Spicer et al., 1991), was labelled with [<*-32P]dCTP (Amersham International pic, 3000 ^Ci/mmol)using the random primer technique (Feinberg and Vogelstein,1983). Unincorporated labelled nucleotides were separated byspinning in CHROMA SPIN-30 columns (Clontech Labora-tories, Inc.). Membranes were probed with labelledpMuc2TR (2 x 106 cts/minute per ml) at 42°C overnight, andwashed three times (30 minutes each) with 40 mM Na2HPO4pH 7.2, 0.1 mM EDTA and 1% SDS at 65°C. Filters wereexposed to X-Ray film (Fuji Photo Film Co.) at -70°C in asuitable cassette with intensifier screens. To check RNAloading and quality, the same membranes were stripped andreprobed with labelled /3-actin PCR fragment (see nextsection).

Reverse transcription-polymerase chain reaction (RT-PCR)The RT-PCR technique was performed as described byRapollee (Rappolee et al., 1988). Briefly, 10 ̂ g of RNA weredenatured in 10 mM Tris-HCl pH 7.5 at 95°C for 2 minutes.Each sample was split into equal volumes in two tubes: onefor the positive controls (mouse /3-actin) and the other forMuc-1. Reverse transcription was done in a 25 /A reactionvolume using the following conditions: 10 mM Tris-HCl pH8.3, 50 mM KC1, 2.5 mM MgCl2, 0.8 mM each dNTP (dATP,dTTP, dCTP and dGTP), 12.5 pmol of 3' primer, and 10 U ofMoloney Murine Leukemia Virus - Reverse Transcriptase (20units/j/1, Boehringer Mannheim GmbH). Samples wereincubated for 1 hour at 42°C. The synthesized cDNA was thenamplified by PCR, in an optimized reaction mixture of 10 mMTris-HCl pH 8.3, 50 mM KC1, 1.75 mM MgCl2, 0.2 mM eachdNTP (dATP, dTTP, dGTP, dCTP), 2% formamide, 25 pmolof each primer, and 2.5 U of Taq DNA polymerase (5 units//d,Boehringer Mannheim GmbH) in a final volume of 100 fjl.Samples were heated for 10 minutes at 95°C before theaddition of Taq polymerase. Cycles and temperatures wereempirically optimized in a thermal reactor (Hybaid) as: 1minute at 94°C, 30 seconds at 62°C, and 1 minute at 72°C for30 cycles. Primers used for Muc-1 reverse transcription andamplification were: 3' oligo, 5' CAGTCCTTCTGAGAGC-CACC 3'; 5' oligo, 5' GCAGTGTGCCAGTGCCGCCG 3'.

Developmental expression of Muc-1 mucin 429

The amplified Muc-1 fragment corresponds to the sequencebetween 2915 bp to 3532 bp in the published genomicsequence (Spicer et al., 1991), with an expected length of 427bp (without intron 6) (Fig. 1). Mouse /3-actin primers have thefollowing sequence (Rappolee et al., 1989): 3' oligo, 5'TGGCCTTAGGGTGCAGGGGG 3'; 5' oligo, 5'GTGGGCCGCTCTAGGCACCA 3' (expected amplifiedfragment length=270 bp). PCR fragments (15 u\ of thereaction) were fractionated in 2% agarose gels in TBE,stained with ethidium bromide and photographed using apositive/negative Polaroid film (Polaroid, type 55). Negativeswere scanned in a laser scanner (LKB Bromma - UltralaserScanner XL), and the area of each peak calculated. The ratiobetween the area of the Muc-1 peak and the /J-actin peak foreach sample was calculated and called relative mRNAabundance (%). Mouse embryos were assayed in at leastthree different experiments using two independent RNApreparations. Day-10, -11 and -12 embryos of the same agewere pooled, due to the low RNA content of smallerembryos. RT-PCR on RNA from embryonic organs wasperformed at least twice.

The specificity of the Muc-1 PCR fragment obtained waschecked three ways: presence of a restriction site for Mbol,reamplification using internal oligos (see Fig. 1A), andSouthern blot of the PCR fragments. Digestion of the Muc-1PCR fragment was performed using 15 jA of PCR reaction and2.5 U of Mbol (New England Biolabs, 5 units//il) in a 20 (Adigest. After incubation at 37CC for 1 hour, the samples werefractionated in 2% agarose gels and visualized by ethidiumbromide staining. Reamplification was done in 100 [A totalvolume, using 0.5 /A of the original Muc-1 PCR reactionmixture and the following primers: 5' CTCACGGACGC-TACGTGCCC 3' and 5' CCCCAGTGTCCCCCAGGGCA3', corresponding to the sequence 3012 bp to 3504 bp in theMuc-1 gene (expected fragment length=313 bp). The ream-plification reaction was performed as: 1 minute at 94°C, 30seconds at 65°C and 1 minute 72°C for 10-15 cycles. Southernblot analyses were done as follows: PCR fragments werepurified onto DEAE membranes (Schleicher and Schuell,Dassel, Germany), and 20 ng of each fragment were loadedon a 2% agarose gel and transferred to a nylon membrane(Pall Biodyne, Glen Cove, NY). Hybridization was carriedout according to manufacturer's instructions. Purified frag-ments were radiolabelled by the random primer techniqueand hybridized individually with the PCR fragments immobi-lized onto the nylon membrane.

Results

Analysis of Muc-1 mRNA expressionThe level of Muc-1 mRNA in mouse embryos is too lowto be easily detected using northern blots, since 15 ng oftotal RNA from a 15-day embryo yielded a barelydetectable signal after a 6-day exposure (data notshown). We therefore utilized the RT-PCR technique,which provides a more sensitive assay than bothnorthern blot and RNAase protection assays (Rappoleeet al., 1989). To perform this technique, oligos weredesigned to span intron 6 of the Muc-1 gene (Fig. 1A),so that contaminating genomic DNA would give aproduct ~1000 bp greater than the expected one. Thetitration of the number of PCR cycles using thedifferent samples studied here is shown in Fig. IB. Thenumber of cycles chosen, 30 cycles, is in the linear

amplification range of all samples. Using this approachand the specific oligos designed from the genomicsequence of Muc-1 (Fig. 1A), it was possible to detect aband with the predicted fragment length (427 bp) in 14-day mouse embryo and lactating mammary gland totalRNA (Fig. 2A). A faint band can occasionally bedetected when RNA from L-cells (a mouse fibroblastcell line that does not express Muc-1 protein) was used,a phenomenon previously described as illegitimatetranscription (Chelly et al., 1991). The quality andquantity of RNA used in the assay were evaluated byusing oligos specific for /3-actin message (Fig. 2B). Thespecificity of this Muc-1 PCR product obtained waschecked by two means: first, the presence of arestriction site for Mbol, which would produce twofragments of 320 bp and 107 bp (Fig. 2C), and,secondly, by reamplification using internal oligos,which would produce a 313 bp band (Fig. 2 D). The 107bp band was too faint to be reproduced photographi-cally (Fig. 2C). The low level of Muc-1 expression in L-cell samples obtained by RT-PCR was greatly enhancedafter the reamplification (Fig. 2D), even though only 15cycles were used to perform it.

The RT-PCR reaction using Muc-1 -specific oligosproduced in some samples a faint second band (400 bp),which can be better seen in the L-cell lanes. This lowerband appears consistently when RT-PCR reaction wasperformed on RNA from whole embryos and fromsome embryo tissues (salivary glands, liver, kidney,large intestine) but not others (lung, stomach, pancreasand lactating mammary gland) (data not shown). Thisband has been proved to be non-specific both bydigestion with Mbol and reamplification with internalMuc-1 oligos (Fig. 2C,D). It does not hybridize witheither the 427 bp PCR fragment specific for Muc-1 orthe 313 bp internal reamplification product (Fig. 2E).

When the expression of the Muc-1 gene was studiedthroughout mouse postimplantation development usingthe RT-PCR technique, specific Muc-1 mRNA can befirst detected in 10-day embryo RNA (data not shown).As mucin is present in neonates (see below) and adultorgans (Pemberton et al., 1992), it can be assumed thatits mRNA expression starts by day-10 gestational ageand continues through mouse infancy and maturity.

The determination of Muc-1 expression in dissectedembryo organs was performed and the message abun-dance was calculated relative to /3-actin mRNA ex-pression in each organ, as measured by RT-PCR assays(Fig. 3). A clear distinction can be observed betweenthe relative mRNA abundance of Muc-1 gene inembryo organs where this mucin is expressed (kidney,stomach, lung, pancreas and salivary gland) and theothers where it is not (liver and intestine). A cut offbetween them can be established taking as referencethe level of expression measured in L-cell samples. Thislow level of mRNA expression in cells that do notproduce detectable amounts of Muc-1 core protein canbe attributed to a basal level expression that can bedetected when a sensitive technique such as RT-PCR isused (Chelly et al., 1991). A slight reduction in theMuc-1 relative mRNA abundance is observed in all

430 V. M. M. Braga and others

PCR products313 bp

427 bp

TAGnr Mbo I

Intron 4 Intron 5 Intron 61000 bp

poly A signal

Transmembrane domain

0.9

0.8-

- . 0 . 7 -

c 0.6-

o °-5"5 0.4--5-'g 0.3-£

° 0.2-

0.2-

0si

aa

2

QQ

CO

C3

E

number of PCR cycles

Fig. 1. (A) Diagram of exons5, 6 and 7 of the genomicstructure of Muc-1 (Spicer etal., 1991), showing positions ofthe oligos used for the RT-PCR assay in mouse embryosamples with the respectivefragment lengths andrestriction site for Mbol. Darkarrows show the oligos used inthe amplification of Muc-1;white arrows show the internaloligos used for thereamplification of the firstMuc-1 PCR fragmentobtained. (B) Titration of thePCR cycles number of the RT-PCR technique, showing thatthe number of cycles chosen(30 cycles) is in the linearrange of amplification ofrepresentative RNA samples.

positive organs at the developmental age of 19-20 days.In northern blot experiments (Fig. 4A), a well definedincrease in Muc-1 mRNA level is observed in totalRNA from embryonic kidney, lung and stomach inolder embryos. /S-actin levels in the same blot are shownin Fig. 4B, where variations in hybridization signal canbe seen in some older embryo organs (Mugrauer et al.,1988). Due to the low amount of lactating mammarygland total RNA loaded (1 /ig), this sample does notshow any /3-actin hybridization signal with a 3 hourexposure (Fig. 4B); 20-day salivary gland RNA appearseither to be degraded or a smaller amount was loaded.

Analysis of Muc-1 protein expression inmorphogenesisThe distribution and appearance of Muc-1 mucin inorgans of mouse embryos and neonates was also studiedwith immunohistochemical staining using CT1 anti-serum. CT1 is a polyclonal antiserum raised to the 17 C-terminal amino acids of the cytoplasmic domain of thehuman MUC1 mucin (Pemberton et al., 1992). A verystrong and specific positive signal was obtained, located

mainly in the apical portion of epithelial cells facingluminal structures in organs expressing Muc-1 (Fig. 5)(Peat et al., 1992; Pemberton et al., 1992; Zotter et al.,1988). The specificity of the staining was evaluated byeither the absence of staining in organs such as liver(Fig. 5C) and intestine (data not shown) or by theblockage of CT1 binding by pre-incubating the antibodywith the 17 amino acid synthetic peptide to which it wasraised (Fig. 5B).

Fig. 6 summarizes the results obtained with theimmunostaining of mouse embryo sections. In general,the expression of Muc-1 is observed soon after the onsetof differentiation and polarization of the epitheliaduring morphogenesis (see below). The mucin proteincan be first detected at day 12 in embryonic stomach,lungs and pancreas as opposed to the detection of mucinmRNA in whole embryo total RNA at day 10 with RT-PCR. The sensitivity of the three techniques employedto study Muc-1 expression using our conditions can becompared, and it reveals the RT-PCR as the mostsensitive, followed by immunostaining and then north-ern blot. Apart from the difference in the sensitivity ofthe techniques, the data obtained with the Muc-1

o

Developmental expression of Muc-1 mucin 431

o

a.y»

5I

CMCNJC O

CDa

o

S3

EO)

O

O

EECO

ict.m

TO

-cel

l

<

zDC

oc

427 bp400 bp

B

Q.10

^

CMCMCMCCma.

emb

a*"*

-̂3O _C 33 O

^_.

3OC3

EEE

0

</>

y

1 Mil

3O «.C 33 c

427 bp400 bp320 bp

— 107 bp

- 270 bp

- 313 bp

5^c c .#% * i * ^ in fll if % ^L

PCR products: 3 S-S

aIn , • * *

Js ex

427bp-400bp-313bp-270bp-

Probes: Int. Muc-1 Lower p - actinE band

Fig. 2. RT-PCR products obtained using Muc-1 specificoligos (A) and ^-actin specific oligos (B). Specificity of theMuc-1 PCR fragment obtained in A was checked byrestriction digestion with Mbol (C) and reamplification usinginternal oligos (D). In E, a Southern blot is shown with allRT-PCR fragments probed against themselves. Exposuretimes at -7CPC were 30 minutes for ^-actin probe and 1hour for the probes Internal Muc-1 (Int. Muc-1) and Lowerband.

432 V. M. M. Braga and others

mRNA analysis correlate well with that obtained withthe protein analysis.

We have performed no analyses on the expression ofMuc-1 during the embryonic development of the

mammary gland. However, neonate mammary glanddoes not appear to express Muc-1 mucin until week 4after birth, when a weak signal can be observed in CT1-stained sections (data not shown).

140

a> 120o(0

•o 100

DCEo

JSo

40

20

ill I 1DAYS

n14 17 18 19 16 18 1*20 0 16 17 20 14 17 18 20 16 17 18 19 16 18 20 16 17 18 19 20

(Q O

on>

Q

as•2so

Fig. 3. Relative mRNAabundance of Muc-1 messagein mouse embryo organsthroughout development usingRT-PCR technique. Lactatingmammary gland total RNA(LMG, 1:10 dilution) was usedas the positive control and L-cell total RNA as the negativecontrol for Muc-1 expression.Muc-1 relative mRNAabundance was calculated inrelation to /3-actin mRNAlevels in the same samples.

OJa

oa

o&1k • E•a o =E « Eo Q.«

OGO Q Q OQQ QQQ QQQ(OOOO TTf-O) (O00O> 1 0 0 ) 0 (OCOO

2.3 Kb

•T* 2.1 Kb

Fig. 4. Northern blot of totalRNA from embryo organs (10fig) and lactating mammarygland (1 /ig) probed withlabelled Muc-1 tandem repeatprobe (pMuc2TR) (10 daysexposure) (A) or labelled /S-actin PCR fragment (3 hoursexposure) (B). Lactatingmammary gland total RNAdoes not show any /3-actinhybridization signal due to theamount loaded (1 pig). Salivarygland RNA (20 day) appearsto be either degraded or asmaller amount was loaded.

ac

• *

• * ***

y \

Fig. 5. Expression of Muc-1 mucin protein as revealed by immunostaining with CTl antiserum. Muc-1 is located in theapical borders of developing epithelial sheets in the different organs (arrows). The specificity of staining is shown by thephotos where a 15-day embryo (liver, stomach, pancreas - 40x) section was stained by CTl (C) and with blocked CTl (B).The developmental expression of Muc-1 is shown through organogenesis of the stomach (A) 12-day (200x), (C) 15-day(40x); pancreas (D) 13-day (200x), (E) 15-day (lOOx), (F) 8-week neonate (200x); salivary glands (G) 15-day (200x),(H) 18-day (40x), (I) 4-week neonate (40x); lung (J) 14-day (200x), (K) 18-day (200x), (L) 4-week neonate (200x); andkidney (M) 14-day (200x), (N) 15-day (lOOx), (O) 8-week neonate (200x). Arrowheads in 4-week neonate lung (L) showfocal expression of Muc-1 in pneumocytes. a, alveoli; ac, acini; d, duct; i, islet of Langerhans; g, glomeruli; 1, liver; m,salivary gland mucous type; p, pancreas; s, stomach; se, salivary gland serous type.

Developmental expression of Muc-1 mucin 433

Large tnL

Small Int.

Mmmm-gl.

S d l v * .

Long

Kidney

Pancreas

Stomach

Lira-

1

tzz

1 ^i 1 1 1 1

2 4 . 6 8 10

Vaginal plug

•

S31 1

EpatelHilbud /cord

Simple tube

V///W////////////A

Y/Mr///"'"'—WM0W"""""\4%y//ss"""

mw///////////////Mi i i i

12 14 16 18

Gesatlonid days

Prooephroj (Dty 9), mesoocphroi (D»y 10), mctancphro* (Day 12)

Actnc epnbcbal duTerenuMKm

Mimmary sprout growth md bnnxfamg

Muc-1 expression

Fig. 6. Coordinate expression of Muc-1 mucin proteinduring mouse organogenesis based onimmunohistochemical staining data, Rugh (1990) andSakakura (1989).

PancreasMouse pancreas and stomach start to differentiate onday 10 (Rugh, 1990). Mucin staining is first observed inthe developmental 12-day pancreas in the luminal areasof the developing tubules. The acinar structure is notclear at this stage, although it is possible to observetubules of epithelial cells being formed (Fig. 5D - 13day). The intensity of the stain is not as strong as inolder embryos/neonates nor does the glandular struc-ture resemble the mature organ (Fig. 5F). Fromgestational day 14 to 17, secretory acini with strongmucin staining in the lumen appear, but they are rathersparse and mesenchyme can be seen separating them(Fig. 5E). Although from day 18 to adulthood thepancreas gradually acquires a more compact andlobular structure, essentially no change is observed inthe Muc-1 immunostaining pattern.

StomachThe earliest Muc-1 expression in the stomach could beobserved at day 12 (Fig. 5A). From day 13 on, anincrease in the staining signal in the apical part of thestomach epithelium is observed, but it is not uniformlydistributed (Fig. 5A and C). The enhancement instaining is coincident with the appearance of glandularfunction, as suggested by the presence of vacuoles in thestomach epithelial cells (Rugh, 1990). It is possible todetect mucin in the lumen of submucosal glands afterday 16, when the formation of stained pits can also beobserved (data not shown). This stomach stainingpattern persists in neonate and mature mice, similar to

that observed with embryonic pancreas. Although it isbelieved that the mouse stomach is differentiated at 11days in the embryo (Rugh, 1990), we could be certain ofMuc-1 protein expression in the stomach only at day 12,after its expansion. The borderline between positiveand negative staining is difficult to assess in 11-day orearlier embryos because of: (1) limitations in thesensitivity of the technique; (2) edge effect observedwith general staining and (3) mucous secreted in thelumen or its border can arrest antibodies (brownishsecretion can be seen in sections stained with blockedCT1 - Fig. 5B).

Salivary glandsSubmandibular salivary glands have a later onset whencompared to the other organs studied here: bygestational day 13, there are only epithelial cords; ductsappear at day 14 (Rugh, 1990). In our hands, stainingwith CT1 revealed Muc-1 mucin focally in ducts insalivary glands of embryos aged 15 days, and sub-sequently, the presence of Muc-1 mucin was detected inthe rudimentary secretory units known as terminaltubules or primary acini (Fig. 5G,H). In developingneonate and adult salivary glands, however, we couldobserve the presence of Muc-1 mucin only in ducts andnot in acinar cells (Fig. 51). Although there are otherreports showing by immunohistochemistry the presenceof mucins in both acini and duct cells (Denny andDenny, 1982; Denny et al., 1988), the antiserum used inthese studies was directed to the whole biochemicallypurified mucin molecule. It is not clear whether thisantiserum also recognizes Muc-1. However, this possi-bility seems remote, because their antiserum recognizesa salivary gland-specific mucin and does not cross-reactwith mucins from other epithelial tissues where Muc-1is expressed (Denny and Denny, 1982).

Liver and intestineNeither small nor large intestine expressed Muc-1mucin from mouse embryogenesis to adulthood. Liveris essentially negative as well, except for the apicalsurfaces of the bile ducts (data not shown). Conse-quently, these organs were used as our negativecontrols (Fig. 5C).

Lungs and tracheaLung buds appear in 9-day mouse embryos (Rugh,1990). They grow and undergo branching morphogen-esis so that the bronchus and bronchiolus originate atdays 11 and 13, respectively. Trachea also appears atday 11. Trachea, bronchi and bronchioh have beenshown to express Muc-1 gene in adult mice (Pembertonet al., 1992). In our immunostaining experiments, aweak positive signal can be observed in the lungs at day12, and from day 13 on, this signal increases in intensityin bronchi and later in the bronchioli (Fig. 5J).Positively stained secretions can be seen inside thelarger bronchi. Pulmonary alveoli do not develop untilday 17, when the simple cuboidal epithelia differen-tiates into the alveolar type I cells and the cuboidal typeII cells in the respiratory saccules (Braukner et al.,

434 V. M. M. Braga and others

1991). Positive staining of both alveoli and the se-cretions present in the alveolar lumens is shown on day18 (Fig. 5K). However, the mucin expression in alveoliafter birth is transient: it gradually disappears so thatthe 2 week neonate shows almost no alveolar staining.Adult and juvenile alveoli lumens do not stain withCT1, but some pneumocytes express Muc-1 focally(Fig. 5L - arrowheads). Detection of Muc-1 protein inthe trachea was first possible at gestational day 15 (datanot shown).

KidneyThree distinct organs are formed sequentially duringkidney organogenesis (Rugh, 1990). The first two aretransient: the pronephros (gestational day 9) andmesonephros (day 10), the latter originating theureteric bud. Finally the metanephros appears (day 12),and its development will lead to the formation of thepermanent kidney. The metanephros is formed by thedifferentiation of fibroblast-like mesenchymal cells intoepithelial cells. This process is induced by the proximityof the branching ureteric bud (reviewed by Ekblom,1981, 1989; Klein et al., 1988). The first observation ofthe expression of mucin in the kidney by immunostain-ing with CT1 is in the day-13 embryo, in the apicalsurface of the ducts (Fig. 5M,N). A precise identifi-cation of which particular developing tubule expressesMuc-1 in early embryonic kidney (13-15 day) has beendifficult; however, both collecting ducts and distaltubules express mucin in humans (Zotter et al., 1988)and transgenic mice (MUC-1) (Peat et al., 1992) (Fig.5O). This is interesting because collecting ducts orig-inate from the branching of the epithelial ureteric bud,and distal tubules are formed by epithelial cells thathave differentiated from mesenchyme. So, irrespectiveof their origin, both tubules expresses Muc-1 mucin inadult mice. Using protein and northern blot analysis,we have observed a lower level of Muc-1 expression inthe kidney when compared to the other organs studiedhere.

Discussion

We have described in this paper the expression of theMuc-1 transmembrane mucin protein during mouseembryogenesis and have shown that its expressioncorrelates with the onset of epithelial sheet formation indifferent organs. Mouse organogenesis starts at day 8-9(Rugh, 1990) (Fig. 6). The expression of the Muc-1gene in mouse embryos can be first detected bygestational day 10, using RT-PCR technique with totalRNA from whole embryos (data not shown). Muc-1protein expression is coordinated both spatially andtemporally with epithelial differentiation in the organsknown to express it in the adult life: lungs, stomach,pancreas, salivary glands and kidney. Our RNAanalysis of Muc-1 expression correlates well with thepresence of mucin protein detected in the organs byimmunohistochemistry.

The use of the CT1 antiserum in our staining

experiments was of particular relevance. Its epitope isnot subject to blockage by differential glycosylation indifferent organs or different physiological stages of thesame organ, as has been observed with other antibodiesto mucin extracellular domains (Zotter et al., 1988;Parry et al., 1992), since CT1 was raised to the humanMUC1 cytoplasmic tail. The close homology observedbetween human and mouse cytoplasmic tail (Spicer etal., 1991) and the ability of CT1 to yield similar mucinstaining patterns in different mammalian species (Pem-berton et al., 1992) stress the probable importance ofthis intracellular domain to Muc-1 function.

In general, Muc-1 was described as being expressedonly in simple secretory epithelia (Peat et al., 1992;Zotter et al., 1988). Detection of Muc-1 mucin in 13-daylung was striking since, at this time, pulmonaryepithelia is stratified (Braukner et al., 1991). Withdevelopment, the lung epithelia changes to simplecuboidal, maintaining Muc-1 mucin at the luminalsurface of the respiratory tract. Other epithelia express-ing Muc-1 as determined by immunohistochemistry arenasal and tracheal (pseudostratified) and adult vaginaand cervix epithelia (squamous stratified) (unpublisheddata). Pulmonary alveoli have a particular feature notobserved in other organs: Muc-1 immunostaining canbe observed both in apical borders and in secretionspresent inside the lumen by gestational days 18, 19 and20. In neonates and adults, however, this alveoli-staining pattern disappears completely and mucinexpression is shown only focally in pneumocytes (Fig.5K,L). Bronchi and bronchioli continue to expressMuc-1 mucin in the adult.

We observed that in the embryonic development ofthe various organs studied, Muc-1 mucin appears liningthe luminal border of the epithelia that is still foldingand branching and, in general, continues to beexpressed throughout the life time of the mouse (Figs 5and 6). During embryogenesis, Muc-1 expressioncontinually increased with time, mainly due to growthin area (epithelial branching). An augmented concen-tration of the protein in a given lumen appears to occurin the early period of mucin expression in each organ.Pancreas and salivary gland are good examples of thisincrease in mucin expression, observed from embryonicdevelopment days 12 to 14 and from days 15 to 18,respectively.

The development of the organs studied here varies inthe time of appearance, developmental processes, andtime required to become mature (Fig. 6). Lung andkidney are able to function by day 18 (Rugh, 1990),although they grow and undergo modifications postna-tally. Pancreas and liver also attain functional activitybefore birth. On the contrary, salivary gland reachesfull maturity in 4-month-old mice (Srinivasan andChang, 1979). Therefore, the timing of Muc-1 proteinappearance in the epithelia does not correlate with thecompletion of organ maturation, since it is presentmuch before they attain full functional activity. Inaddition, the epithelial buds/cords of these organsoriginate at different times (Fig. 6). In general, it ispossible to detect Muc-1 mucin lining the lumens of the

Developmental expression of Muc-1 mucin 435

embryonic pancreas (12 day), salivary gland (15 day)and lung (12 day) soon after their epithelial buds start tobranch and differentiate (Fig. 6). By the time ofstomach expansion (12 day) and the appearance andbranching of the ureteric bud from the metanephros inthe embryonic kidney (13 day), mucin protein can alsobe observed. Our results suggest that, although theMuc-1 gene is expressed in different organs, itsexpression is not stimulated systemically, but it isinduced concomitantly with the epithelial differen-tiation in each individual organ (Fig. 6).

During embryogenesis, small epithelial buds undergoelongation, folding and branching through mesen-chyme, resulting in distinct epithelia-containing organs(Bernfield, 1978; Bernfield, 1981). Alternatively, epi-thelial cells can originate by differentiation of mes-enchymal cells (derived from mesoderm), which isinduced by close proximity of epithelial buds (Ekblom,1989; Hay, 1990). Similar Muc-1 expression patternswere observed in the various epithelia regardless oftheir origin from ectoderm or mesoderm. For instance,epithelial cells of distal tubules, which are formed bythe differentiation of fibroblast-like cells in the kidneymesenchyme, express Muc-1 as well as the collectingducts, which are derived by ramification of theprimordial ureteric bud. This might indicate thatpolarization and branching processes are similar in bothcases, and cells might respond to similar stimuli andpossess similar mechanisms of turning on and off genesto produce the epithelial sheets.

The timing and pattern of expression of Muc-1 mucinduring epithelial differentiation and its wide distri-bution make the role of this large glycoprotein duringorganogenesis interesting to study further. It is tempt-ing to speculate that, as it has been shown that highlevels of MUC1 expression decrease cell-cell interac-tions, the presence of this protein in the apical part ofthe cells might help lumen formation in the developingepithelium, by reducing adhesive associations in theapical domain (Ligtenberg et al., 1992). The character-istic structure and biochemical properties of the Muc-1protein may be responsible for preventing the interac-tions between adhesive molecules. The first feature isthe rod-like structure of the extracellular domain,extending the mucin far beyond the cell glycocalyx(Jentoft, 1990). The predicted length of this mucin is^250 nm as opposed to the expected 30 nm length ofcellular adhesion molecules (Becker et al., 1989). Inaddition, the presence of a large number of sialic acidresidues on mucin oligosaccharide chains produces anet negative charge, enhancing the steric hindranceeffect caused by the large extracellular domain (Parry etal., 1992). It is not clear how many sialic acid residuesare present in mucin protein during mouse embryogen-esis. There have been reports showing that sialic acidlevels in Muc-1 protein can be developmental^ regu-lated in the adult mouse (Parry et al., 1992). Thetemporal and local expression of the Muc-1 genecoincident with epithelial differentiation in manyorgans (stomach, pancreas, kidney, salivary glands andlung) might suggest a participation of this molecule in

the process of epithelial sheet formation. However, asMuc-1 mucin is continually expressed in adult life,additional roles must be postulated for the presence ofthis molecule after the completion of maturation ofthese organs.

Although our data have shown a correlation betweenMuc-1 expression and the onset of epithelial differen-tiation in some organs during mouse embryogenesis, itis difficult to define in vivo the relevance of thiscorrelation and hence extrapolate a possible functionfor this molecule. First of all, other epithelia-containingorgans do not express Muc-1 (e.g. intestines). Sec-ondly, Muc-1 immunoreactive epitopes have beenfound intracellularly in placenta (data not shown) andmuscle (Pemberton et al., 1992), and Muc-1 mRNAexpression in these tissues has been confirmed (thiswork and Vos et al., 1991). Thirdly, during morphogen-esis, three overlapping events occur: cell polarization,epithelial sheet formation and the process of branching,making it extremely difficult to determine the precisetiming of each event in vivo. It would be interesting tostudy the expression of Muc-1 in an in vitro systemwhere we could deal with these events separately andwhere we could manipulate the experimental conditionsin order to interfere in the process. The well-establishedin vitro systems of branching morphogenesis usingembryonic lung, kidney or salivary gland rudimentscould be utilized (Bernfield, 1981; Ekblom, 1981;Schuger et al., 1990). The process of cell polarizationcould be studied using epithelial cell lines that are ableto polarize in vitro. In these ways, it should be possibleto determine whether Muc-1 expression behaves simi-larly in vitro and to which event (if any) the expressionof this gene is correlated, e.g. epithelial cell polariz-ation, sheet formation or branching morphogenesis.Alternatively, the Muc-1 function in the mouse can bedisrupted in vivo by homologous recombination. Bothapproaches are in progress in an attempt to elucidatethe role of Muc-1.

In conclusion, we have presented the first evidencethat the expression of the Muc-1 mucin during mouseembryogenesis is restricted to the apical surfaces ofsecretory epithelial cells in stomach, pancreas, trachea,lung, kidney, and salivary gland. The induction of Muc-1 gene expression is under spatial and temporal controlduring the epithelial differentiation in the organs wherethe protein is detected in the mouse. Muc-1 protein canbe detected lining the apical surfaces of the developinglumens when the polarized epithelium of these organs isstill folding and branching, and glandular activity hasnot properly started. Finally, although Muc-1 is presentin different organs, its expression is not inducedsystemically, but according to the particular onset ofepithelial polarization and branching morphogenesis ofeach individual organ.

We would like to thank Gillian Hutchinson for the animalwork, Christine Pike and colleagues for the histologicalpreparations, and Dr Ian Goldsmith for oligonucleotidesynthesis. We would also like to thank Drs E.-N. Lalani andS.Kimber for help in the analysis of immunohistochemistry

436 V. M. M. Braga and others

data and Drs J. Taylor-Papadimitriou, F. Watt, J. Williamsand G. Warren for critical reading of the manuscript.V.M.M.B. has a fellowship from Conselho National deDesenvolvimento Cientffico e Tecnol6gico (CNPq), Brazil.

References

Bartek, J., Durban, E. M., Hallowes, B. C. and Taylor-Papadlmltriou, J. (1985). A subclass of luminal epithelial cells inthe human mammary gland, defined by antibodies to cytokeratins./ . Cell Sci. 75, 17-33.

Becker, J. W., Erickson, H. P., Hoffman, S., Cunningham, B. A. andEdclman, G. M. (1989). Topology of cell adhesion molecules. Proc.Natl. Acad. Sci. USA 86, 1088-1092.

Bernfield, M. (1978). Mechanisms of embryonic organ formation. InAbnormal Fetal Growth: Biological Bases and Consequences (ed.Naftolin, F.), pp. 101-120, Springer-Verlag, Berlin.

Bernfield, M. (1981). Organization and remodeling of theextracellular matrix in morphogenesis. In Morphogenesis andPattern Formation, pp. 139-162, Raven Press.

Bharathan, S., Moriarty, J., Moody, C. E. and Sherblom, A. P.(1990). Effect of tunicamycin on sialomucin and natural killersusceptibility of rat mammary tumor ascites cells. Cancer Res. 50,5250-5256.

Braukner, J. H., Trautman, M. S. and Bernfield, M. (1991).Syndecan, a cell surface proteoglycan, exhibits a molecularpolymorphism during lung development. Dev. Biol. 147, 285-292.

BurcheU, J., Durbln, H. and Taylor-Papadimitriou, J. (1983).Complexity of antigenic determinants recognised by monoclonalantibodies HMFG-1 and HMFG-2, in normal and malignantmammary epithelial cells. / . Immunol. 131, 508-513.

BurcheU, J., Gendler, S., Taylor-Papadimitriou, J., Girling, A.,Lewis, A., Millls, R. and Lamport, D. (1987). Development andcharacterization of breast cancer reactive monoclonal antibodiesdirected to the core protein of the human milk mucin. Cancer Res.47, 5476-5482.

Cbelly, J., Concordet, J. P., Kaplan, J.-C. and Kahn, A. (1991).Illegitimate transcription: transcription of any gene in any cell type.Proc. Natl. Acad. Sci. USA 86, 2617-2621.

Chomczyski, P. and Sacchi, N. (1987). Single-step method of RNAisolation by acid guanidinium thiocyanate-phenol-chloroformextraction. Analyt. Biochem. 162, 156-159.

Church, G. M. and Gilbert, W. (1984). Genomic sequencing. Proc.Natl. Acad. Sci. USA 81, 1991-1995.

Codington, J. F., Sanford, B. H. and Jeanloz, R. W. (1973). Cell-surface glycoproteins of two sublines of the TA3 tumor. J. Nad.Cancer Inst. 51, 585-591.

Denny, P. A. and Denny, P. C. (1982). Localization of a mousesubmandibular sialomucin by indirect immunofluorescence.Histochem. J. 14, 403-408.

Denny, P. A., Pimprapaiporn, W., Kim, M. S. and Denny, P. C.(1988). Quantitation and localization of acinar cell-specific mucin insubmandibular glands of mice during postnatal development. CellTissue Res. 251, 381-386.

Ekblom, P. (1981). Formation of basement membrane in theembryonic kidney: an immunohistological study. / . Cell Biol. 91,1-10.

Ekblom, P. (1989). Developmentally regulated conversion ofmesenchyme to epithelium. FASEB J. 3, 2141-2150.

Felnberg, A. P. and Vogelstein, B. (1983). A technique forradiolabelling DNA restriction fragments to high specific activity.Analyt. Biochem. 132, 6-13.

Gendler, S., Taylor-Papadimitriou, J., Duhig, T., Rothbard, J. andBurchell, J. (1988). A highly immunogenic region of a humanpolymorphic epithelial mucin expressed by carcinomas is made oftandem repeats. J. Biol. Chem. 263, 12820-12823.

Gendler, S. J., BurcheU, J. M., Duhig, T., Lamport, D., White, R.,Parker, M. and Taylor-Papadlmltriou, J. (1987). Cloning of partialcDNA encoding differentiation and tumor-associated mucinglycoproteins expressed by human mammary epithelium. Proc.Natl. Acad. Sci. USA 84, 6060-6064.

Gendler, S. J., Lancaster, C , Taylor-PapadTmltriou, J., Duhig, T.,Peat, N., Peraberton, L., Lalani, E.-N. and Wilson, D. (1990).

Molecular cloning and expression of a human tumor-associatedpolymorphic epithelial mucin. J. Biol. Chem. 265, 15286-15293.

Gum, J. R., Byrd, J. C , Hicks, J. W., Toribana, N. W., Lamport, D.T. A. and Kim, Y. S. (1989). Molecular cloning of human intestinalmucin. J. Biol. Chem. 264, 6480-6487.

Gum, J. R., Hicks, J. W., SwaUow, D. M., Lagace, R. L., Byrd, J. C ,Lamport, D. T. A., Slddiki, B. and Kim, Y. S. (1990). Molecularcloning of a novel human intestinal mucin gene. Biochem. Biophys.Res. Comm. 171, 407-415.

Hay, E. D. (1990). Role of cell-matrix contacts in cell migration andepithelial-mescnchymal transformation. Cell Diff. Dev. 32, 367-376.

Jentoft, N. (1990). Why are proteins O-glycosylated? TIBS 15, 291-294.

Klein, G., Langegger, M., Tlmpl, R. and Ekblom, P. (1988). Role oflaminin A chain in the development of epithelial cell polarity. Cell55, 331-341.

Lan, M. S., Batra, S. K., Ql, W.-N., Metzgar, R. S. andHoUlngsworth, M. A. (1990). Cloning and sequencing of a humanpancreatic tumor mucin cDNA. J. Biol. Chem. 265, 15294-15299.

Ligtenberg, M. J. L., Buijs, F., Vos, H. L. and Hllkens, J. (1992).Suppression of cellular aggregation by high levels of episialin.Cancer Res. In press.

Ligtenberg, M. J. L., Vos, H. L., Gennlssen, A. M. C. and Hllkens, J.(1990). Episialin, a carcinoma associated mucin, is generated by apolymorphic gene encoding splice variants with alternative aminotermini. J. Biol. Chem. 265, 5573-5578.

Manlatis, T., Fritsch, E. F. and Sambrook, J. (1982). MolecularCloning: a Laboratory Manual, Cold Spring Harbour Laboratories,Cold Spring Harbour.

Mugrauer, G., Alt, F. W. and Ekblom, P. (1988). N-myc proto-oncogene expression dunng organogenesis in the developingmouse as revealed by in situ hybridization. J. Cell Biol. 107, 1325-1335.

Parry, G., Beck, J. C , Moss, L., Bartley, J. and Ojakian, G. K.(1990). Determination of apical membrane polarity in mammaryepithelial cell cultures: the role of cell-cell, cell-substratum, andmembrane-cytoskeleton interactions. Exp. Cell. Res. 188, 302-311.

Parry, G., Li, J., Blssell, M., Schmldhauser, C , Splcer, A. P. andGendler, S. J. (1992). Studies of Muc-1 mucin expression andpolarity in the mouse mammary gland demonstrate developmentalregulation of Muc-1 glycosylation and establish the hormonal basisfor mRNA expression. J. Cell Sci. 101, 191-199.

Peat, N., Gendler, S. J., Lalani, E.-N., Duhig, T. and Taylor-Papadimitriou, J. (1992). Tissue specific expression of a humanpolymorphic epithelial mucin (MUC1) in transgenic mice. CancerRes. In press.

Pemberton, L., Taylor-Papadimitriou, J. and Gendler, S. G. (1992).Antibodies to the cytoplasmic tail domain of the MUC1 mucinshow conservation throughout mammals. Biochem. Biophys. Res.Comm. 185, In press.

Porchet, N., Van Cong, N., Dufosse, J., Audie, J. P., Guyonnet-Duperat, V., Gross, M. S., Denis, C , Degand, P., Bernheim, A.and Aubert, J. P. (1991). Molecular cloning and chromosomallocalization of a novel human tracheo-bronchial mucin cDNAcontaining tandemly repeated sequences of 48 base pairs. Biochem.Biophys. Res. Comm. 175, 414-422.

Rappolee, D. A., Brenner, C. A., SchuKz, R., Mark, D. and Werb, Z.(1988). Developmental expression of PDGF, TGF-a-, and TGF-/3genes in preimplantation mouse embryos. Science TAX, 1823-1825.

Rappolee, D. A., Wang, A., Mark, D. and Werk, Z. (1989). Novelmethods for studying mRNA phenotypes in single or smallnumbers of cells. J. Cell. Biochem. 39, 1-11.

Rogh, R. (1990). The Mouse: Its Reproduction and Development,Oxford University Press.

Schuger, L., O'Shea, K. S., Nelson, B. B. and Varanl, J. (1990).Organotypic arrangement of mouse embryonic lung cells on abasement membrane extract: involvement of laminin. Development110, 1091-1099.

Shimizu, M., Tanimoto, H., Azuma, N. and Yamauchl, K. (1990).Growth inhibition of BALB/c 3T3 cells by a high-molecular-weightmucin like glycoprotein of human milk fat globule membrane.Biochem. Intern. 20, 147-154.

Developmental expression of Muc-1 mucin ATI

Siddiqui, J., Abe, M., Hayes, D., Shani, E., Ynnls, E. and Kufe, D.(1988). Isolation and sequencing of a cDNA coding for the humanDF3 breast carcinoma-associated antigen. Proc. Natl. Acad. Sci.USA 85, 2320-2330.

Spicer, A. P., Parry, G., Patton, S. and Gendler, S. J. (1991).Molecular cloning of the mouse homologue of the tumor-associatedmucin, Muc 1, reveals conservation of potential O-glycosylationsites, transmembrane and cytoplasmic domains and a loss ofminisatellite-like polymorphism. J. Biol. Chem. 266, 15099-15109.

Srtnivasan, R. and Chang, W. W. L. (1979). The postnataldevelopment of the submandibular gland of the mouse. Cell TissueRes. 198, 363-371.

Steck, P. A. and Nicolson, G. L. (1983). Cell surface glycoprotein of13762NF mammary adenocarcinoma clones of differing metastasispotentials. Exp. Cell. Res. 147, 255-267.

Vos, H. L., de Vries, Y. and Hilkens, J. (1991). The mouse epsialin(Muc-1) gene and its promoter: rapid evolution of the repetitivedomain in the protein. Biochem. Biophys. Res. Comm. 181, 121-130.

Wreshner, D. H., Hareuvenl, M., Tsarfaty, I., Smorodlnsky, N.,Horev, J., Zaretsky, J., Kotkes, P., Weiss, M., Lathe, R., Dion, A.and Keydar, I. (1990). Human epithelial tumor antigen cDNAsequences. Differential splicing may generate multiple proteinforms. Eur. J. Biochem. 189, 463-473.

Zotter, S., Hageman, P. C , Lossnltzer, A., Mooi, W. J. and Hilgers,J. (1988). Tissue and tumor distribution of human polymorphicepithelial mucin. Cancer Rev. 11-12, 55-101.

{Accepted 3 March 1992)