Development 100, 653-660 (1987)Printed in Great Britain © The Company of Biologists Limited 1987

653

Trophectoderm surface expression of the cell adhesion molecule cell-

CAM 105 on rat blastocysts

PETER C. SVALANDER1, PER ODIN2, B. OVE NILSSON1 and BJORN OBRINK2

The Department of Human Anatomy^ and The Department of Medical and Physiological Chemistry2, University of Uppsala,The Biomedical Center, Box 571, S-751 23, Uppsala, Sweden

Summary

A variety of cellular interactions is involved in theprocess of implantation of the mammalian embryointo the uterine tissue. Recent discoveries havedemonstrated that intercellular recognition and ad-hesive events are governed by a class of cell surfacemolecules known as cell adhesion molecules (CAMs).In the present report, we have investigated the occur-rence of the well-characterized cell adhesion moleculecell-CAM 105 on the surface of rat pre- and peri-implantation embryos of various stages. This wascarried out by indirect immunofluorescence mi-croscopy employing affinity-purified rabbit antibodiesagainst cell-CAM 105. The embryonal stages investi-gated comprised morulae, normal day-4 blastocysts,and delayed and adhesive blastocysts obtained byusing the method of experimentally delayed implan-tation. Cell-CAM 105 was absent in the early-morulastage, but in normal day-4 blastocysts and delayed

blastocysts a specific staining for cell-CAM 105 wasseen on the entire surface. However, adhesive-stageblastocysts exhibited a marked polarity with stainingof the polar trophoblast cells. Scanning electron mi-croscopy of adhesive-stage blastocysts revealed thatthe stronger staining of the polar region was not dueto a greater number of microvilli on the polar tropho-blast cells. Thus, it seems as if cell-CAM 105 is lost ormasked from the surface of the mural trophoblastcells of adhesive-stage rat blastocysts. Since the muraltrophoblast cells are the first to adhere to the uterineluminal epithelium during the onset of implantationand subsequently invade the uterine stroma, wesuggest that the apparent downregulation of cell-CAM 105 in the mural trophoblast cells might belinked to the acquisition of trophoblast invasiveness.

Key words: cell adhesion molecule, trophectoderm, rat,cell-CAM 105, blastocyst, antibody.

Introduction

Embryonic development is governed by cellularinteractions in which cell adhesion molecules arethought to play a major role (Edelman, 1985; Obrink,1986). One key event in mammalian development isthe implantation of the blastocyst into the uterinewall, which occurs about one week after fertilizationin the rat. A variety of cellular interactions betweenthe embryonic and the uterine cells is believed to beresponsible for a proper implantation process. First,the trophoblast cells of the blastocysts must come intoclose contact with the cells of the uterine luminalepithelium (the apposition stage). Second, the blasto-cyst acquires an adhesive surface, which involvesmolecular modifications of the trophoblast cell sur-faces (the adhesion stage) (Chavez, 1986). Molecularalterations are also believed to occur in the uterine

luminal epithelium at this stage (Chavez & Anders-son, 1985). Third, the intercellular interactions in thetrophectoderm, as well as in the uterine luminalepithelium, must be modulated to allow the invasiqnof the trophoblast cells through the uterine luminalepithelium into the underlying stroma (the invasionstage). Finally, interactions between the trophoblastcells and the extracellular matrix of the uterinestroma occur. Thus, blastocyst implantation is ahighly dynamic process in terms of cell surfaceinteractions and cell adhesion events.

Based on present knowledge of the mechanisms ofcell adhesion, it seems likely that the dynamic processof implantation, at least in part, depends on themodulation of cell adhesion molecules. Today severalcell adhesion molecules are known. Among the bestcharacterized are N-CAM, L-CAM and uvomorulin(Obrink, 1986). These three CAMs appear very early

654 P. C. Svalander, P. Odin, B. O. Nilsson and B. Obrink

in avian and mammalian embryos and persist all theway to the adult stage (Edelman, 1985). Duringembryonic development, they are modulated in termsof prevalence, concentration and chemical structure(Edelman, 1985).

Another well-characterized cell adhesion moleculeis cell-CAM 105, which was originally identified asbeing involved in intercellular adhesion of rat hepato-cytes in vitro (Ocklind & Obrink, 1982). Cell-CAM105 is a membrane-integrated cell surface glyco-protein consisting of two highly glycosylated peptidechains (Qdin, Tingstrom & Obrink, 1986). Studieswith cell-CAM 105 incorporated into liposomes haverevealed that the protein can bind to itself in acalcium-independent homophilic type of reaction(Obrink et al. 1986). Cell-CAM 105 appears ratherlate in the development of the liver and does notreach the amount seen in mature liver until threeweeks after birth (Odin & Obrink, 1986). In liverregenerating after partial hepatectomy, a transientdecrease of the concentration in the hepatocyteplasma membranes occurs (Odin & Obrink, 1986).Furthermore, in transplantable rat hepatocellularcarcinomas, cell-CAM 105 is absent or chemicallymodified (Hixson, McEntire & Obrink, 1985). Thesefindings suggest that a decreased concentration ofcell-CAM 105 facilitates cellular proliferation and/ormotility.

Since blastocyst implantation is characterized byproliferation and migration of invasive trophoblastcells into the uterine tissue (Tachi, Tachi & Lindner,1970; Schlafke & Enders, 1975), we asked whethercell-CAM 105 is expressed also on the surface of earlyrat embryos and whether the expression is downregu-lated at implantation. In this report, we demonstratethat cell-CAM 105 is indeed present, apparentlystage-specific, in the trophectoderm of the rat blasto-cyst and that it disappears from the attaching pole ofthe implanting blastocyst.

Materials and methods

Rat embryosRat preimplantation embryos of the outbred Sprague-Dawley strain (Alab, Sweden) were obtained on differentdays of gestation by flushing the uterine horns with Dul-becco's modified phosphate-buffered saline (PBS) sup-plemented with 1 % bovine serum albumin (BSA, RIA-grade, Sigma) and 0-1 % sodium azide. Morulae and early,unhatched, blastocysts were isolated at days 3 and 4 ofpregnancy, respectively. Day 1 of pregnancy is the day offinding a vaginal plug. The zonae pellucidae were removedby brief incubation in acid Tyrode's buffer, pH2-5. Afteracid treatment for removal of the zonae, the embryos werewashed several times in PBS containing 1 % BSA and 0-1 %sodium azide prior to treatment for immunocytochemistry.

Experimental delay of implantation was induced by bilat-eral ovariectomy on day 3 of pregnancy, followed by sub-cutaneous injections of 1 mg progesterone (Depo-Provera,Upjohn) every 5th day. Delayed implantation blastocysts(the zonae pellucidae of these embryos are lost in utero)were flushed from the uterus more than 7 days afterinduction of delay. For isolation of late, peri-implantation,blastocysts, activation for implantation was induced by asubcutaneous injection of 0-1 jig of estradiol 17B in propyl-eneglycol. Blastocysts in the adhesive stage were recovered18h after activation from delay (Mayer, 1963; Nilsson,1984).

AntibodiesAffinity-purified rabbit antibodies from one antiserum(anti-cell-CAM8; Odin & Obrink, 1986) against cell-CAM105 were used in all experiments reported here. The IgGfraction of both the antiserum and the preimmune serumwas isolated by affinity chromatography on protein A-Sepharose (Ocklind & Obrink, 1982). The IgG fraction ofthe antiserum was affinity-purified by chromatography onpurified cell-CAM 105 coupled to Sepharose 4B (Odin &Obrink, 1986). Bound immunoglobulins were eluted with3M-KSCN in 0-05 M-phosphate buffer, pH6-0. To test thespecificity of the antibodies, rat hepatocytes were surface-labelled with 125I (Hubbard & Cohn, 1972), solubilized inTriton X-100 and immunoprecipitated as described byOcklind & Obrink (1982). The immunoprecipitate wasanalysed by two-dimensional electrophoresis according toO'Farrell (1975). Only the A-chain (higher apparent mol-ecular weight and more basic) and the B-chain (lowerapparent molecular weight and more acidic) of cell-CAM105 were recognized by the affinity-purified antibodies(Fig. 1).

Indirect immunofluorescence microscopyIndirect immunofluorescence labelling of rat embryos fordetection of cell-CAM 105 was performed according to thefollowing protocol. Embryos of a defined stage werewashed several times in PBS supplemented with 1 % BSAand 0-1% sodium azide. Dilution of antibodies and wash-ings were also carried out in this medium. The specimenswere incubated in affinity-purified anti-cell-CAM IgGor preimmune IgG (see below), at a concentration of8/igm\~\ for 2h at room temperature (identical resultswere obtained when handling of embryos and immuno-labelling were performed on ice). After washings in PBScontaining 1 % BSA and 0T % sodium azide, three timesfor lOmin, the embryos were incubated in tetramethylrho-damine isothiocyanate-conjugated swine-anti-rabbit IgG(Dakopatts), diluted 1:40, for 30min at room temperature.The specimens were washed three times for lOmin andviewed in a Nikon Diaphot inverted microscope equippedfor epifluorescence. Photographs were taken with KodakTri-X film. Two to six independent immunolabelling exper-iments were performed at each embryonic stage. Thetotal number of embryos observed in the study is given inTable 1.

Cell-CAM 105 on rat blastocysts 655

pH

IEF MrXirr3

Table 1. Immunofluorescence microscopy of ratembryos demonstrating surface staining and polarity of

the cell adhesion molecule cell-CAM 105

- 94

-67

- 43

- F

Fig. 1. Specificity of affinity-purified anti-cell-CAMantibodies. Two-dimensional electrophoresis ofdetergent-solubilized 12;>I-labelled rat hepatocytesimmunoprecipitated by affinity-purified anti-cell-CAMantibodies. In the first dimension, the components wereseparated by isoelectric focusing (IEF) in a pH gradientranging from pH3 to 8, as indicated in the top portion ofthe figure. In the second dimension, the componentswere separated by SDS-polyacrylamide gelelectrophoresis (SDS-PAGE). The migration of markerproteins of known molecular weights is indicated in theright portion of the figure. The origin of the separation isindicated by X and the front of the SDS-PAGE isindicated by F. Only the A-chain (higher apparentmolecular weight and more basic) and the Brchain (lowerapparent molecular weight and more acidic) of cell-CAM105 were recognized by the affinity-purified antibodies.

ControlsIn order to test the specificity of the immunocytochemicalmethod used for detection of cell-CAM 105, control exper-iments were carried out by: (1) incubation in preimmuneIgG instead of anti-cell-CAM IgG (at similar antibodyconcentrations) or (2) omitting the primary antibody in thefirst step of the immunolabelling protocol. At least tenembryos of each embryonic stage studied were control-immunolabelled by each of the above procedures. Theimmunocytochemical control experiments, which were car-ried out under identical conditions and in parallel to thedetection of cell-CAM 105, were all negative (see Fig. 6).

In order to test whether removal of the zona pellucida byacid treatment could influence the detection of cell-CAM105 by means of altering the immunofluorescence intensity,a separate control experiment was carried out with delayedblastocysts (zona-free in utero). The blastocysts were incu-bated in acid Tyrode's buffer (pH2-5) for 15min, washedand passed through the immunolabelling protocol used forcell-CAM 105 detection (see above). The acid treatmentdid not alter the cell-CAM 105-positive immunofluor-escence intensity compared to the intensity obtained with-out previous acid treatment of embryos (data not shown).This finding was taken as evidence for the reliability of the

Embryonic stage* Stainingt Polarity^

Morulae (40)Day-4 blastocysts (40) + —Delayed blastocysts (60) + + / —Adhesive blastocysts (60) + +

* Number of specimens in parenthesis.t + indicates a positive, and — a negative cell surface staining.t + indicates polarity, defined as a positive immuno-

fluorescence staining of one pole of the embryo, while theother (the opposite) pole is unstained. - indicates absence ofpolarity, and +/— a variable trophectoderm surface stainingwithout any consistent polarity.

immunocytochemical method used for detection of cell-CAM 105 on embryos that had had their zona pellucidaremoved by acid treatment (i.e. morulae and normal day-4blastocysts).

Scanning electron microscopyPreparation of adhesive-stage rat blastocysts for scanningelectron microscopy was performed as described previously(Nilsson, Naeslund & Curman, 1980). Briefly, the uterinehorns of rats were flushed with 2-5 % glutaraldehyde in PBS18 h after activation from delay. The blastocysts were fixedovernight at +4°C. The fixative was washed away with PBSand distilled water before the embryos were mounted onpoly-L-lysine-coated small circular glass coverslips. Theembryonic pole was identified by light microscopy of theinner cell mass. The specimens were dehydrated stepwise inan acetone scale, critical-point dried and finally coated witha thin layer of gold before examination in a Jeol U-3 or aPhilips 525 scanning electron microscope. The surfacemorphology of an adhesive-stage blastocyst is demon-strated in Fig. 7.

Results

Monospecific, affinity-purified rabbit antibodies wereused in indirect immunofluorescence microscopy tomap the surface expression of cell-CAM 105 on ratembryos of various stages. These stages includedmorulae, normal day-4 blastocysts, and delayed andadhesive blastocysts obtained by using the method ofexperimentally delayed implantation. The results ofthe immunolabelling experiments are summarized inTable 1.

On the morulae no staining for cell-CAM 105 wasdetected (Fig. 2A), but, on the blastocysts, a specificstaining of the trophectoderm was seen on all three

656 P. C. Svalander, P. Odin, B. O. Nibson and B. Obrink

stages examined. Normal day-4 blastocysts had stain-ing of the entire trophectoderm surface (Fig. 3A).The delayed blastocysts also showed staining of theentire trophectoderm surface, but the intensity of the

fluorescence varied in different portions of the troph-ectoderm (Fig. 4A). The adhesive-stage blastocystsdemonstrated a polarity with staining of the embry-onal pole (polar trophoblast cells) while the mural

B

.̂.".' X- '^J

B

--4

B

Figs 2-4. Immunofluorescence (A) and phase-contrast (B) microscopy. X300.Fig. 2. Morulae-stage embryos stained with affinity-purified anti-cell-CAM antibodies. The morula surface was negativefor cell-CAM 105 staining.Fig. 3. Normal day-4 blastocysts stained with affinity-purified anti-cell-CAM antibodies. The entire blastocyst surfaceexhibited cell-CAM 105 staining.Fig. 4. Delayed blastocysts stained with affinity-purified anti-cell-CAM antibodies. The blastocyst surface exhibited aheterogeneous cell-CAM staining with a higher concentration at the trophoblast cell-cell borders.

Cell-CAM 105 on rat blastocysts 657

B

Figs 5, 6. Immunofluorescence (A) and phase-contrast (B) microscopy. x300.Fig. 5. Adhesive-stage blastocysts stained with affinity-purified anti-cell-CAM antibodies. The blastocyst surfaceexhibited a polarity in cell-CAM staining: the embryonal pole was stained while the abembryonal pole was unstained.Fig. 6. Morula and blastocysts stained with preimmune IgG (immunocytochemical control labelling). Both morulae andblastocysts were completely negative for preimmune IgG staining. A morula-stage embryo is indicated by an arrow.

trophoblast cells were unstained (Fig. 5A). Pre-immune IgG (control antibody) stained neither mor-ulae nor blastocysts which is exemplified in Fig. 6A.

The cell-CAM 105 detected on the blastocystsurface occurred on the apical surfaces of the tropho-blast cells, but not in a homogeneous manner. Thecell-cell borders stained more intensely than theremainder of the trophoblast cell surfaces. The het-erogeneous cell surface staining was not obvious in allspecimens, since several factors such as concentrationof antibodies, plane of focus and exposure time werecritical for its detection. Prominent staining of thecell-cell borders is demonstrated in Fig. 4A.

There was a possibility that the difference instaining intensity observed at the two poles of theadhesive-stage blastocysts might be the result of adifference in number of microvilli between the twopoles. In order to investigate this possibility, weexamined adhesive-stage blastocysts by scanningelectron microscopy. As shown in Fig. 7, the twopoles of the adhesive-stage blastocyst did exhibit

different numbers of microvilli. However, a greaternumber of microvillli was found on the abembryonalpole of the blastocyst compared to the embryonalpole. Thus, since the immunofluorescence was re-stricted to the embryonal pole of the adhesive-stageblastocysts, the staining must indeed reflect a higherconcentration of cell-CAM 105 per membrane unitarea at the polar trophoblast cells.

Discussion

The molecular changes of the trophectoderm surfaceat implantation, resulting in an increased blastocystadhesiveness and invasion of trophoblast cells intothe uterine tissue, are poorly understood. However,it is reasonable to assume that significant changes inthe molecular architecture of the apical cell surfacesin both the trophectoderm and the uterine luminalepithelium are involved. One class of cell surfacemolecules that is assumed to play a key role duringdevelopment is the recently discovered cell adhesion

658 P. C. Svalander, P. Odin, B. O. Nilsson and B. Obrink

Fig. 7. Scanning electron microscopy of an adhesive-stage blastocyst. The abembryonal pole of the blastocyst, which islocated to the right, has a bulging appearance with abundant microvilli. The embryonal pole of the blastocyst, which islocated to the left, has a smooth surface with few microvilli. The prominent ridges are the trophoblast cell-cell borders.X1850.

molecules (CAMs). It has been demonstrated thatseveral CAMs appear early in avian and mammaliandevelopment and that they are modulated in a varietyof ways during the changing cellular interactionscharacteristic for embryonic development (Edelman,1985).

We examined rat morulae and blastocysts of vari-ous stages for the presence of cell surface-locatedcell-CAM 105 by indirect immunofluorescence mi-croscopy. We found that cell-CAM 105 appeared onthe trophectoderm surface of the preimplantationblastocyst and that it disappeared from the implantingblastocyst. This is the first time a cell surface antigenhas been found to be stage-specifically expressed atthe blastocyst stage and dynamically downregulatedat implantation. Further, the fact that this antigen is awell-characterized cell adhesion molecule (Obrink,1986; Obrink et al. 1986; Ocklind & Obrink, 1982;Odin et al. 1986) adds to the informative value of thisstudy.

The occurrence of cell-CAM 105 on the trophecto-derm suggests that the trophoblast cells synthesize theprotein and incorporate it in their plasma mem-branes. Since the blastocyst resides in the uterine

secretion, an alternative explanation would be thatthe cell-CAM 105 detected on the surface of thetrophectoderm might have been adsorbed from theuterine secretion. This possibility seems unlikelysince cell-CAM 105 was undetectable on morulae butdetectable on normal day-4 blastocysts. Both of thesestages are encased by the zona pellucida which wouldimpede a direct contact between the macromoleculesof the uterine secretion and the embryonic cellsurfaces. Thus, the most reasonable explanationseems to be that cell-CAM 105 is synthesized by thetrophoblast cells and becomes cell-surface expressedat the blastocyst stage.

Cell-CAM 105 was detected on all portions of thetrophectoderm surface of normal day-4 blastocystsand delayed blastocysts, but became highly polarizedon the trophectoderm of adhesive-stage blastocysts.The mural trophoblast cells, which were devoid ofcell-CAM 105, exhibited a more irregular cell surfacewith a greater number of microvilli than the polartrophoblast cells. This indicates that the observedpolarity reflects a true downregulation, or masking,of cell-CAM 105 on the surface of the mural tropho-blast cells. This observation is supported by the

Cell-CAM 105 on rat blastocysts 659

studies of Chavez (1986), who found that the polar(non-adhesive) trophectoderm, but not the mural(adhesive), of adhesive-stage blastocysts bound suc-cinylated wheat germ agglutinin (S-WGA). Since wehave found that cell-CAM 105 binds WGA (Odin etal. 1986), the downregulation of cell-CAM 105 on themural trophoblast cells might contribute to the loss ofS-WGA binding sites.

The mural and the lateral trophoblast cells of theblastocyst are the ones that first penetrate the uterineluminal epithelium and invade the uterine stroma(Alden, 1948; Chavez, 1986). This suggests that theapparent downregulation of cell-CAM 105 from themural trophoblast cells might be linked to the acqui-sition of trophoblast invasiveness. According to ourhypothesis, the disappearance of cell adhesion mol-ecules from the trophoblast cells would govern theirdissociation from the trophectoderm organization,thereby facilitating their migration into the uterinetissue during implantation. This idea is supported bythe observations of a decreased cell surface ex-pression of cell-CAM 105 on fetal hepatocytes (Odin& Obrink, 1986), regenerating hepatocytes (Odin &Obrink, 1986) and hepatocellular carcinoma cells(Hixson, McEntire & Obrink, 1985), which are allcharacterized by higher motilities and proliferationrates than normal, mature hepatocytes. Similar corre-lations between cell surface expression and cellularmigration have been observed also with other CAMs(Obrink, 1986). The most striking example is thedisappearance of L-CAM and N-CAM from theneural crest cells when they start their migration inthe embryo and the reappearance of N-CAM whenthey aggregate at their new locations (Thiery,Duband & Delouvee, 1985). Another example is thedynamic appearance of uvomorulin in the earlymouse embryo. Uvomorulin, also known as cell-CAM 120/80, is present on mouse morulae andpreimplantation blastocysts but disappears when thetrophoblastic giant cells are formed during implan-tation (Damjanov, Damjanov & Damsky, 1986).Taken together, all these findings suggest that ageneral mechanism involving modulation of cell ad-hesion molecules operates in the regulation of cellularmigration and invasion. Our results now demonstratethat cell adhesion molecules are also involved in thecomplex events of blastocyst implantation.

Blastocyst invasion into the uterine stroma not onlyrequires mobilization of the trophoblast cells, but alsopenetration of the uterine-lining epithelium. Thispenetration would be facilitated by a loosening ofthe intraepithelial cellular associations. A downregu-lation of cell adhesion molecules in the uterineepithelium, similar to that observed for cell-CAM 105in the adhesive blastocyst, would be one mechanismallowing for this. In preliminary studies, we have

found that cell-CAM 105 is present in the uterine-lining epithelium of nonpregnant rats but disappearsin pregnant rats (Odin etal. in preparation). Detailedstudies concerning the expression of cell-CAM 105both in blastocysts and in the uterus during implan-tation are now in progress in our laboratories.

We are indebted to Mrs Barbro Einarsson for her experthelp with the production of the rat embryos. This investi-gation was financially supported by the Swedish MedicalResearch Council (project nos 05200, 6686 and 00070), TheSwedish Cancer Foundation (project no. 1389), KonungGustaf V:s 80-arsfond and Magn. Bergvalls Stiftelse.Depo-Provera was a kind gift from Upjohn AB, Sweden.

References

ALDEN, R. H. (1948). Implantation of the rat egg. III.Origin and development of primary trophoblast giantcells. Am. J. Anat. 83, 143-182.

CHAVEZ, D. J. (1986). Cell surface of mouse blastocystsat the trophectoderm-uterine interface during theadhesive stage of implantation. Am. J. Anat. 176,153-158.

CHAVEZ, D. J. & ANDERSON, T. L. (1986). The glycocalyxof the mouse uterine luminal epithelium during estrus,early pregnancy, the peri-implantation period anddelayed implantation. 1. Acquisition of Ricinuscommunis 1 binding site's during pregnancy. Biol.Reprod. 32, 1135-1142.

DAMJANOV, I., DAMJANOV, A. & DAMSKY, C. H. (1986).Developmentally regulated expression of the cell-celladhesion glycoprotein cell-CAM 120/80 in peri-implantation mouse embryos and extraembryonicmembranes. Devi Biol. 116, 194-202.

EDELMAN, G. M. (1985). Expression of cell adhesionmolecules during embryogenesis and regeneration. ExplCell Res. 161, 1-16.

HIXSON, D. C , MCENTIRE, K. D. & OBRINK, B. (1985).Alterations in the expression of a hepatocyte celladhesion molecule by transplantable rat hepatocellularcarcinomas. Cancer Res. 45, 3742-3749.

HUBBARD, A. L. & COHN, Z. A. (1972). The enzymaticiodination of the red cell membrane. J. Cell Biol. 55,390-405.

MAYER, G. (1963). Delayed nidation in rats: A method ofexploring the mechanisms of ovo-implantation. InDelayed Implantation (ed. A. C. Enders), pp. 213-231.Chicago, USA: The University of Chicago Press.

NILSSON, B. O. (1984). The fine structure and function ofmouse and rat trophoblast during delayed implantation.In Ultrastructure of Reproduction (ed. J. Van Blerkom &P. M. Motta), pp. 260-266. The Hague: MartinusNijhoff Publ.

NILSSON, B. O., NAESLUND, G. & CURMAN, B. (1980).Polar differences of delayed and implanting mouseblastocysts in binding of Alcian Blue and ConcanavalinA. J. exp. Zool. 214, 177-180.

OBRINK, B. (1986). Epithelial cell adhesion molecules.Expl Cell Res. 163, 1-21.

660 P. C. Svalander, P. Odin, B. 0. Nilsson and B. Obrink

OBRINK, B., ODIN, P., TINGSTROM, A., HANSSON, M.,RUBIN, K. & BLIKSTAD, I. (1986). Cell adhesionmolecules involved in cell-cell adhesion phenomena.Structure and function of cell-CAM 105. In Biology andPathology of Platelet-Vessel Wall Interactions (ed. G.Jolles, Y. Legrand & A. T. Nurden), pp. 161-178.London: Academic Press Inc.

OCKLIND, C. & OBRINK, B. (1982). Intercellular adhesionof rat hepatocytes. Identification of a cell surfaceglycoprotein involved in the initial adhesion process. J.biol. Chem. 257, 6788-6795.

ODIN, P. & OBRINK, B. (1986). Dynamic expression ofthe cell adhesion molecule cell-CAM 105 in fetal andregenerating rat liver. Expl Cell Res. 164, 103-114.

ODIN, P., TINGSTROM, A. & OBRINK, R. (1986). Chemicalcharacterization of cell-CAM 105, a cell-adhesionmolecule isolated from rat liver membranes. Biochem.J. 236, 559-568.

O'FARRELL, P. (1975). High resolution two-dimensionalelectrophoresis of proteins. J. biol. Chem. 250,4007-4021.

SCHLAFKE, S. & ENDERS, A. C. (1975). Cellular basis ofinteraction between trophoblast and uterus atimplantation. Biol. Reprod. 12, 41-65.

TACHI, S., TACHI, C. & LINDNER, H. R. (1970).Ultrastructural features of blastocyst attachment andtrophoblast invasion in the rat. J. Reprod. Fert. 21,37-56.

THIERY, J. P., DUBAND, J. L. & DELOUVEE, A. (1985).The role of cell adhesion in morphogenetic movementsduring early embryogenesis. In The Cell in Contact.Adhesions and Junctions as Morphogenetic Determinants(ed. G. M. Edelman & J. P. Thiery), pp. 169-196.New York, USA: John Wiley & Sons.

{Accepted 14 April 1987)