Altered uvomorulin expression in a noncompacting mutant cell line of F9 embryonal carcinoma cells

DE\‘ELOPMENTAL

BIOLOGY

Altered

138,338-347

(1990)

Uvomorulin Expression in a Noncompacting Cell Line of F9 Embryonal Carcinoma Cells D. ADAMSON,*”

EILEEN *La Jolla

HBL~NE

BARIBAULT,t

Cmcer Research Foundation, La Jolla. Fiir Irn munobiolo~~ie. Freib?cSq-Zah,rinyerl, Accepted

California

Federal

Nvuember

AND

ROLF

Mutant

KEMLERt

9.2037; and tMax;-Planck-Institut Republic of Germany

30, 1989

Uvomorulin (E-cadherin) is a cell adhesive molecule analogous to L-CAM in the chicken. Uvomorulin is important in the process of compaction in eight-cell mouse embryos and plays a role in F9 embryonal carcinoma cell interactions but il is not clear if iL mediates aggregation or compaclion, the closer interaction Lhat also occurs in F9 cells cultured in suspension. This paper describes the finding of reduced levels of uvomorulin in a mutant cell line of F9 (5.51 att) that is consistent with a role for uvomorulin in both aggregation and compaction. The mutant line expresses 40-50s of normal levels of uvomorulin as measured by surface radioiodination, immunoblotting, and biosynthetic labeling and immunoprecipitation with two different antisera. The mutant cell line expresses only abnormally unstable uvomorulin transcripts at very low levels. In addition, karyotypic analyses revealed an abnormal chromosome 8 on which the uvomorulin gene is located and therefore could account for aberrant uvomorulin expression. F9 5.51 att. cells aggregate loosely but do not compact (A. Grover, M. J. Rosenstraus, B. Sterman, M. E. Snook, and E. D. Anderson, 1987, Dev. BioL 119, l-1 1). The conclusion is that reduced levels of uvomorulin are sufficient for aggregat,ion but insufficient for compaction. ‘c l!wl Acadrmir Prms, 1°C INTRODTJCTION

Uvomorulin (Peyrieras et al., 1983, 1985) is the cell adhesive component that is thought to mediate the process of compaction of blastomeres at the eight-cell stage of murine embryogenesis. It is also important in the homophilic adhesion of embryonal carcinoma (EC) cells (Kemler et al., 1977) and in the intermediate junctions of epithelial cells (Boller et ah, 1985; Vestweber and Kemler, 1985). Antibodies to uvomorulin have been used to detect and locate this glycoprotein at cell surfaces and also to disrupt cell interactions mediated by uvomorulin (Kemler et al., 1977; Hyafil et al., 1980, 1981). In the case of preimplantation murine embryos, addition of uvomorulin antibody decompacts embryos (Kemler et al., 1977; Johnson et ul., 1979; Hyafil et ul., 1980; Nicolas et al, 1981; Vestweber and Kemler, 1984), but does not interfere with cell proliferation. Later the treated embryos can recover, undergo compaction, and develop normally. Tn some cases, the resulting blastocysts do not contain an inner cell mass (Shirayoshi et al., 1983). Those that do not compact eventually degenerate and die. The same antisera that inhibit compaction also inhibit endoderm formation on the inner cell mass (Richa et ah, 1985). These data suggest that uvomorulin is also important for epithelial layer formation. Uvomorulin is a cell adhesion molecule (CAM) closely ’ To whom

correspondence

should

be addressed.

homologous to L-CAM in the chicken. Uvomorulin belongs to a class of Ca’+-dependent CAMS, like canine arc-l (Behrens et ab, 1985), rr-1 (Gumbiner and Simons, 1986), and human CAM 120/80 (Damsky et ah, 1983). It is identical to E-cadherin (Yoshida-Noro e2al., 1984). It consists of a 728 amino acid polypeptide that traverses the membrane once leaving a relatively small intracellular cytoplasmic carboxyterminal portion (Ringwald et al., 1987). The extracellular domain is able to bind both Ca2+ and to self-associate in an as yet unknown way that leads to cell-cell adhesion. F9 EC cells grow readily in monolayer culture attached to either plastic or gelatin, but equally fast in petri dishes to which they adhere poorly. They adhere to each other very tightly and rapidly proliferate into large compact clusters with smooth contours. When retinoic acid (RA) is added at 5 to 50 nM to aggregate cultures of F9 cells, the smooth contour is retained, but an outer layer of epithelial cells differentiates and is seen after 3 days. The epithelial layer of this embryoid body structure is similar in structure and functional characteristics to that seen on the surface of 6-day mouse embryos. Using a variety of markers the outer layer of cells in emhryoid hodies has been identified as visceral endoderm. In the case of the embryo, this epithelial layer develops into one of the layers of the visceral yolk sac that later encircles the embryo as a protective and secretory barrier (Adamson, 1986). There are several suggestive pieces of evidence indi-

eating that cell-cell compaction is as important to F9 cell differentiation as it is to embryo development. First, when F9 cells are induced with a high concentration of RA (0.1 to 1 FM) in the presence of agents that increase the intracellular levels of CAMP, the cells do not remain compacted, but become separated by the secretion of large amounts of basement membrane components; this differentiated cell type has been identified as parietal endoderm. The production of these deposits may help to direct F9 differentiation irreversibly along the parietal pathway (Grover and Adamson, 1986). Second, when exogenous laminin is added during the differentiation process to embryoid bodies, an epithelial layer is not formed, but a structure resembling a clump of parietal endoderm cells is produced (Grover et al., 198313). A third way of reducing interaction of F9 cells in embryoid bodies is to add dibutyryl CAMP to the culture medium. The epithelium breaks down and parietal endoderm cells appear (Grover and Adamson, 1986). Fourth, a mutant F9 cell line that clusters in loose aggregates, but does not compact, does not form an epithelium of visceral endoderm cells, although it does differentiate almost normally into parietal endoderm (Grover et al., 1987). We have examined several cell surface components that could be involved in the inability of the mutant F9 cell line to compact. Several components are altered in mutant cells when compared with wild-type cells and some of these can be accounted for by clonal differences. The best candidate for a primary role in the defective adhesive behavior of the mutant cell is uvomorulin. This 120-kDa glycoprotein is present in lower amounts on the cells and appears to be synthesized at a similarly reduced rate. Nevertheless, even this reduced rate is higher than expected considering the severely reduced levels of uvomorulin mRNA in mutant cells. The effects of reduced amounts of uvomorulin and the possibilities of changes in its function, as well as other defects in the mutant cell line, are discussed in relation to its inability to form an eyithelium. MATERIALS

AN11

METHODS

Cells. Wild-type F9 EC cells, clone BlM, were cultured as previously described (Grover ef al., 1983a) in gelatinized tissue dishes. Mutant F9 att 5.51 cells (called 5.51 here) were derived from wild-type cells after mutagenization with ethyl methanesulfonate and were selected for severe adhesion defectiveness (Grover et al., 1987). Not only do they lack adhesiveness to plastic, but they also adhere very poorly to fibronectin-coated dishes and to feeder cell layers. They grow in suspension in petri dishes as loose clusters and strings which are readily broken up by pipetting. They grow as rapidly as wild-

type cells with a doubling time of about 11 hr and both wild-type and cell mutant lines are tumorigenic. A similar mutant cell line was produced spontaneously after another EC cell line, OC15, was cultured with 1 PLM retinoic acid for 10 days and then in normal culture medium for a further 20 days. Imr~lu?lo~uorescerrce. Two antisera to uvomorulin were used, both gave similar results. One was a polyclonal rabbit antiserum to the 84-kDa extracellular domain of uvomorulin that is released by F9 cells with trypsin in the presence of Ca’+. The other was a rat monoclonal antibody, DECMA-1; both have been described and characterized previously (Vestweber and Kemler, 1984, 1985). F9 cells were allowed to aggregate overnight or were scraped up and exposed to antiserum at 100 pg/ml total protein in PBS with 3Y0 normal goat serum as clumps, at 4°C for 2 hr. After extensive washing, cells were fixed in 3% (w/v) formaldehyde in PBS for 10 min, washed, and blocked in PBS with 0.1%) (w/v) glycine. The second antibody (diluted 1:30) was rhodamine-labeled goat anti-rabbit (or rat). Immzcnoblotti~tg. Cells were counted, extracted with 20 ~1 17rl NP-40, 0.05%’ SDS, 10 mM Tris, pH 7.5, 1 mM EDTA with proteinase inhibitors per 10” cells for 15 min, and centrifuged at 15,OOO.q for 20 min and the supernatant was boiled with l/4 vol 4~ Laemmli sample buffer. Samples of 2 X lo” to 2 X 10” cells were electrophoresed in 7% SDS gels in a Novex apparatus. Electrophoretic transfer of proteins onto nylon membranes (Biotrans, Biorad) was for 16 hr at 30 mA. Prestained markers were used in outside lanes to determine apparent MW of proteins. After washing and blocking in 5% fat-free dried milk in PBS for several hours, antibody or normal rabbit serum solutions at 20 pg/ml were incubated with membranes overnight. After three washes in 0.9% NaCl with 0.05% Tween 20 and 0.2% NaNcI, ‘“‘I-labeled protein A in the 5% milk solution was allowed to react in sealed bags for 60 min at room temperature. After washing, membranes were exposed to Kodak XAR film for 6 hr to 1 day. kfetnbolic ltrbelir~~~ crnrl irrrnc1L7lo;rrrecipitatiorL. Incorporation of I:‘“S]methionine into cells in culture was performed essentially as described earlier (Grover et oh, 1987). For phosphorglation studies, inorganic [:‘“P]phosphate was added at 200 @ml for 3-4 hr, and lysates were immunoprecipitated with rabbit anti-uvomorulin. Fluorographs were scanned using a LKB laser densitometer. cell-sur:fitce iodimtim. This was done as in Weller et al. (1987) using lactoperoxidase and glucose oxidase to catalyze the radioiodination reaction. RNA txrrnbysis CL& probes. RNA was isolated from cultured cells as described in Maniatis et al, (1982). For Northern blot analysis, 10 pg total RNA was run on

DEVELOPMENTALBIOLOGY VOLUME138,199O

340

glyoxal gels and transferred to Hybond membrane (Amersham Corp.). The filter was hybridized with a 700-bp fragment (F5) made by AvaI digestion of uvomorulin cDNA (Ringwald et al., 1987). The probe was made by random priming oligo-labeling and used at lo6 dpm/ml, at 65°C. The filter was washed at 65°C with 2~ SSC, 0.1% SDS for 15 min followed by 0.1X SSC, 0.1% SDS for 30 min. For more sensitive detection of uvomorulin transcripts, the polymerase chain reaction procedure was employed. Two oligonucleotide sequences located on different exons of the uvomorulin gene were synthesized (nucleotides 1187-1206 was named UM .36 and 1535-1554 was UM .28) that were separated by 386 bp in the mRNA. Total RNA (3 pg) was transcribed into cDNA; amplified in gelatin-containing Tris buffer, pH 8.4 (0.1 mg/ml), containing 50 mM KCl, 2 mM MgC&, 1 11 RNase inhibitor, dNTPs at 0.2 mM, 3 pg of each oligo, UM .37 and UM .28, 5 units avian-MV-reverse transcriptases (Pharmacia); and incubated at 42°C for 1 hr. Two units Taq polymerase (Amersham) was added and 30 cycles of 1 min at 90°C alternating with 5 min of synthesis at 65°C followed. Half of the reaction product was run on 2% agarose gels and transferred to a Hybond membrane to detect uvomorulin with the cDNA probe described above (F5). A control cDNA of 10 ng of circular plasmid DNA containing full length uvomorulin was run under the same conditions. Karyotype a~~alysis. The mutant F9 cells were processed as described by Eistetter et al. (1988). Analyses were performed by Dr. Werner Schemp, Department of Human Genetics, Freiburg. RESULTS

The mutant cell line 5.51 has several characteristics that differ from the parental or wild-type F9 cell line. We have already noted in our earlier work (Grover et ab, 1987) that mutant F9 cells: 1. adhere poorly to FN and fibroblast cell layers; adhesion occurs slowly and spreading does not occur; 2. have altered cell surface components that render mutant cells less sensitive to wheat germ agglutinin and more sensitive to the lectin abrin compared to wild-type F9 cells; 3. synthesize a higher basal level of laminin and type IV collagen even in the undifferentiated state; 4. can differentiate normally into parietal endoderm 4 days after stimulation of mutant cells with 1O-7 MRA, 0.2 mM dbcAMP, 0.1 mM 1BMX. The characteristic marker Troma 3 is formed and we have since confirmed that fetomodulin also appears within 4 days of treatment (Imada et ah, 1987; unpublished results); 5. upon differentiation to parietal endoderm, larger

amounts of basement membrane components are produced compared to wild-type cells. In addition, laminin A polypeptide appears as a doublet with an extra faster migrating band present in addition to the normal A chain; 6. mutant F9 cells do not differentiate into visceral endoderm, as demonstrated by a transient and small activation of the AFP gene after 2 days and absence of expression at 8 or more days of induction; 7. the mutant cell line, like wild-type, grows well in “suspension” in petri dishes. However, the mutant line aggregates poorly into clumps and strings and cannot compact. This is the major defect that clearly indicates a possible malfunctioning component, that is, uvomorulin. Reduced Steady-State

Levels of Uvomwulin

on 5.51 Cells

Since 5.51 cells cannot form smoothly compacted spherical aggregates we looked for differences in uvomorulin steady-state levels using three different procedures. 1. Immunofluorescence studies using both rabbit polyclonal and rat monoclonal antibodies demonstrated that uvomorulin stains less well on 5.51 cells compared to wild-type (Fig. 1). The staining of 5.51 cells varied from almost zero in some experiments to the readily detectable level shown in Fig. lb. After differentiation of 5.51 cells, uvomorulin staining was consistently stronger, suggesting that uvomorulin is produced at higher levels or is more ‘stable, perhaps by being trapped in the abundant extracellular matrix produced by differentiating mutant cells. However, uvomorulin produced after differentiation is ineffective in compaction, perhaps because by this time large amounts of extracellular matrix have been produced and this either physically blocks cell-cell contacts or leads differentiation along the parietal endoderm pathway. 2. Cell surface iodination and immunoprecipitation. Since immunofluorescence observations gave the impression that there were considerable amounts of uvomorulin between apposing 5.51 cell surfaces, we radioiodinated live cells and, after immunoprecipitation of lysates of equal numbers of cells, measured the amount of radiolabeled uvomorulin gp 120 kDa present on autofluorographs by densitometric scanning. There was an average of 40% of iodinated uvomorulin present on 5.51 cells compared to wild-type (Fig. 2). Note that there is no apparent difference in the molecular weights of uvomorulin in the two cell lines. 3. Immunoblotting studies were used to compare the total amounts of uvomorulin present in both intracellular and cell surface locations. Figure 3 shows that there was little difference in total amounts, averaging

341

? 116, 97,

cum

;

. __

12

FIG. 1. Immunofluorescence staining of uvomorulin. Rat monoclonal at 100 &ml protein was the primary antibody incubated with liw cell clumps (F9 cells were scraped up from plastic dishes). After fixing in formalin. secondary antibody, 130 dilution of rhodamine-laheled goat anti-rat, was added. Note that mutant FS Att-5.51 cells are weakly stained (h) compared to wild-type (a). Mutant cells after 3 days of diffc~rrntiation with 10 ’ rctinoic acid stain rather more strongly (c). ((1) Phase-contrast photograph of the cells in b. Bar y 50 pm.

52Vk in 5.51 cells compared to wild-type see no difference in the electrophoretic of uvomorulin in the two cell lines.

I

3

4

FI(:. 2. Ccl1 surface iodinated uvomorulin. Equal numbers (10”) of wild-type and mutant FS cells were iodinated as described under Materials and Methods. Cells were Iysed in RIPA buffer and rabbit antiuvomorulin antitwdies were used to immunoprecipitate the labeled cell surface ylycoprotein. PAGE on 5% polyacrglamide gels was follo\vved by autoradiography to detect radioiodinated bands. Lanes 1 and 2, rabbit anti-uvomorulin; Lanes 2 and 4, control serum. Scanning densitometry was used to calculate the uvomorulin levels on 5.51 cells which were 40% of those on wild-type cells.

metabolically labeled with [“‘PIphosphate, and equal amounts of radioactivity in lysates were compared for phosphorylated uvomorulin levels after immunopreci-

cells. Again, we migration rates

We reasoned that although uvomorulin is present in 5.51 cells at 40-50s of wild type levels, a reduction in cell surface level could explain the inability of the mutant cells to compact. Possibly, the protein is synthesized at a lower rate or is more labile or more rapidly removed from the cell surface. We next measured rates of UM synthesis in the two cell lines (Fig. 4) and calculated that biosynthetic rates in 5.51 cells are 43.7% of those in wild-type cells. A reduced rate of mutant uvomorulin synthesis correlates with the reduced presence of uvomorulin on the surface and does not indicate that the mutant line may metabolize uvomorulin faster. It has been shown that integrin (fibronectin receptor) in F9 cells becomes phosphorylated in a pattern that correlates with activity (Dahl and Grabel, 1989) and therefore the same might be true for uvomorulin. To test this, cultures of wild-type and mutant F9 cells were

205, -urn 103,

123456 Frc:. 3. Immunoblotting to compare total uvomorulin present in F!) cells. Equal numbers of cells of FI) (Lanes 1, 3, and 5) and 5.51 mutant (Lanes 2, 4, and 6) were analyzed as described under Materials and Methods. IJvomorulin was detected after blotting onto nylon memhranes using rabhit anti-uvomorulin (Lanes l-4) and normal rabbit serum (Lanes 5 and ti), followed hp radioiodinated protein A and autoradiography. Lanes 1 and 2, 2.5 X l#cells; Lanes 3-6.5 X lo” cells. Densitometry revealed that the level of uvomorulin present on mutant cells was 52’/1 of that on wild-type cells. This experiment was repeated five times using, in addition, another detection method, namely, the ABC: peroxidase procedure. Identical results were obtained hut cluantitation was not attempted.

342

DEVELOPMENTAL

BIOLOGY

205 b cum

116,

a+2

97, 66,

123456 FIG. 4. Biosynthesis of uvomorulin in F9 cells. Metabolically labeled cells ([“%]methionine) were lysed and immunoprecipitated as described under Materials and Methods. Lysates were immunoprecipitated with rabbit anti-uvomorulin (Lanes 1 and 4), rat anti-uvomorulin (Lanes 2 and 5), or normal rat control (Lanes 3 and 6). After autoradiography of SDS-PAGE gels (7%), the radioactive bands and those at 120 and 90 kDa were scanned to quantify and compare levels. Average levels of bands in 5.51 cells were 43.7% of those in F9 wildtype cells.

pitation. Phosphorylation of uvomorulin varied with the growth phase, but averages of three determinations gave a ratio of 0.47 “P incorporation into mutant cell uvomorulin, compared to wild-type (Fig. 5, compare Lanes 1 and 2). Since this ratio is the same for 32S-metlabeled uvomorulin, we concluded that mutant uvomorulin is not different from wild-type in its degree of phosphorylation. This result further indicates that mutant uvomorulin protein is normal in every respect tested. The sets of proteins (So-100 kDa that coprecipitate with uvomorulin in cell lysates in RIPA buffer, using anti-uvomorulin antibodies, are not seen in immunoblots (Fig. 3) and are, therefore, antigenically dissimilar. These proteins have recently been described as cytoskeletal associated proteins that interact with the cytoplasmic domain of uvomorulin and are correlated closely with uvomorulin’s adhesive activity. It is clear from Figs, 4 and 5 that the mutant F9 line shows no visible defects in the coprecipitated set of proteins. Transcription and Mutant

of the Uvomorulin Cells

The uvomorulin mRNA species in was confirmed in bryonal carcinoma of 11 preparations

Gene

in

VOLUME

138, 1990

lin transcripts could be detected by Northern blotting (Fig. 6a, Lane 11; Fig. 6b, Lanes 3 and 6), while in two preparations a small amount of a smaller transcript at 3.5 kb was seen (Fig. 6a, Lanes l-3; Fig. 6b, Lanes 4 and 5). When 5.51 cells were treated with retinoic acid for up to 4 days, there was little change in the steady-state level of uvomorulin transcripts (Fig. 6a, Lanes l-3). We also analyzed total RNA from OC15 EC cells (Fig. 6b, Lane 1) and a mutant line derived from it which also does not self-aggregate or adhere to plastic. This adhesion- and aggregation-poor cell line also has no detectable uvomorulin mRNA when analyzed by Northern blots (Fig. 6b, Lane 2), thus strengthening the correlation between the morphological and molecular defects. Since there are detectable levels of uvomorulin protein in 5.51 cells, there must be a mRNA, even though we could not detect any by Northern blotting. To amplify the transcript signal, we utilized the polymerase chain reaction (PCR) procedure and did indeed find detectable levels of the expected 386-bp cDNA corresponding to transcripts present in the cellular RNA (Figs. ‘7a and 7b, Lane 4) similar to those found in parental cells (Figs. 7a and 7b, Lane 5). The procedure as performed was not quantitative, and hence we could not make conclusions about transcript levels in the mutant cells. It is clear, however, that the transcripts are abnormally labile and may be broken down in a series of discrete steps since sometimes a 3.5-kb fragment was detected in Northerns (Fig. 6a, Lanes 1-3; Fig. 6b, Lanes 3 and 6). Karyotypic

Analysis

Analyses of chromosome spreads of 5.51 cells such as that seen in Fig. 8 were made. Parental F9 cells have 39 aUM

1 Con.

-205

UM+

-116 -97

F9 Wild-Type

gene is transcribed into a 4.7-kb F9 cells (Ringwald et al., 1987). This Fig. 6 and, in addition, in OC15 emcells (Fig. 6b, Lane 1). However, in 9 of RNA from 5.51 cells, no uvomoru-

FIG. 5. Incorporation of phosphate into uvomorulin. Cultures were labeled with inorganic “2P-labeled phosphate for 4 hr and samples containing equal radioactivity were immunoprecipitated with rabbit anti-uvomorulin or preimmune serum. Lane 1, F9 cells; Lane 2, mutant cells which averaged 47% of the level of “‘P-labeled uvomorulin compared to wild-type.

343

a> kb

kb ,-28s

4.33.5-

28S-18s

I ,

18S-

12

34

5

6

7

8

12

91011

3

4

5

6

FI(;. 6. Northern hlot analysis of uvomorulin mRNA. Extracts of total RNA (10 gg) nere analyzed as described under Materials and Methods and probed with landom-oliyonucleotide-Iat)~,led uvomorulin cDNA. (al Lane 1, RA-treated (0.05 PM) 5.51 cells, 2 days; Lane 2, 5.51 t RA, :] days: Lane 3, ii.51 + RA, .i days; Lane 4, F9 EC cells; Lane 5, F9 + RA, 1 day; Lane 6, I’9 t RA, 2 days; Lane 7, F9 + RA, 3 days; Lant) 8, F9 + RA, 5 tl;t~-s; Law 9, FIJ + RA, X days; lane IO, PSA5E dilferentiated teratocarcinoma cells; Lane 11, 5.51 EC cells. (b) Lane I, (JCI5 EC; I,ancs 2 ()(‘Is att variant; Lanes :I and 1, two different preparations of 5.51 EC cells; Lane 5,5.51 t RA, 4 days; Lane 6, F9 + RA, 8 days. Wild-fgIw uvl)moru]in nrRNA is -1.4 kh, whilr some preparations of 5.51 cells produce a hand at 3.5 kh.

chromosomes and are X0, while 5.51 cells have 42 or 43 chromosomes. Many new abnormalities were seen compared to parental F9 cells (Silver 4 ctl., 1983), including

a)

inversions, deletions, and three fragments, one of which was duplicated. Of greatest interest was the distal portion of chromosome 8 on which uvomorulin is located

b)

C---350bp-

12345 2

3

4

5

FIG:. 7. Analysis of uvomorulin transcripts hy the polymerase chain reaction (P(:Rl procedure. After 30 cycles of amplification of cUNA reverse transcrihcd from mRNA present in F!) + RA 8 days (Lane 5) and 5.51 mutant cells (Lane 4), the expected 350 bp fragment was detected hy ethidium hromitle staining in a and hy prohing Lvith labeled uvomorulin in h. Lane 1, X phagc markers cut with EcoRI and Hi~iI11: Law 2, huffcr control; Lane 3. uvomorulin plasmid control.

344

DEVELOPMENTAL

VOL~JME 138,199O

BIOLOGY

-

FIG. 8. Karyotypeof 5.51 cells. Forty-three which carry the uvomorulin genes.

chromosomes

are visible

(Eistetter et ab, 1988). Chromosome 8 had undergone a Robertsonian translocation in which the two 8 chromosomes were fused at their centromeres. There is no indication to date that Robertsonian translocations affect gene expression. However, there is some indication that an inversion has also occurred at the distal end of one chromosome 8 in 5.51 cells.

Other Studies to Detect Diflerences Cell-Surface Proteins

in

The question that arises is, how general is the defect in cell surface proteins? Is uvomorulin only one of several glycoproteins altered in the mutant line? This is an important question since perhaps general defects may be explained by a common mechanism such as inability to synthesize complete glycopolypeptide chains (cf., laminin A chains) or inability to insert components correctly into the membrane. The interesting candidates to assay are those that concern cell-cell or cellmatrix interactions, since these could account for the properties of 5.51 cells. However, none of the matrix receptors or adhesion components that we examined were significantly different from wild-type and there-

including

a Robertsonian

translocated

Cl

fused

eight

chromosomes

(arrow)

fore were not further studied. The results of this survey are listed next. 1. In collaboration with Drs. R. Pytela and E. Ruoslahti of this Institute, we radioiodinated cell surfaces of F9 and mutant cells and lysates were passed through affinity columns of GRGDSP hexapeptide (to absorb vitronectin receptor). Subsequently, lysates were passed through Sephadex linked to a 120-kDa fibronectin fragment as described by Pytela et al. (1985a,b) to absorb fibronectin receptors. Receptors were eluted with excess hexapeptide and analyzed by PAGE. Unexpectedly, we found that 5.51 cells have levels of fibronectin receptors higher than those of F9 wild-type cells (A. Grover and E. D. Adamson, unpublished data). 2. Using the same cell surface iodinated cell extracts, we also showed that vitronectin receptors are present in 5.51 cells at levels comparable to those in parental cells. 3. We also found similar levels of mRNA coding for the 67-kDa laminin receptor (Wewer et al., 1986; unpublished studies of Drs. U. Wewer and L. Liotta). 4. An adhesive system that is known to exist in EC cells is that involving galactosyltransferase and its interaction with lactosaminyl glycans (Shur, 1983). Extracts from 5.51 mutant cells were examined by Dr.

Barry Shur and found to contain at least twice the enzyme activity compared to F9 wild-type cells. Therefore, this adhesive system was superabundant in the mutant cells. In summary, none of the obvious candidates were different, although in view of the poor ability of 5.51 cells to adhere to fibrcnectin, we suspect that the superabundant fibronectin receptors found on 5.51 cells must be malfunctioning. One cell surface receptor was found to be different in biosynthetic studies. [““SlMethionine-labeled cell lysates immunoprecipitated with rabbit anti-mouse EGF receptor showed a diffuse set of bands at 160-170 kDa compared to a tight doublet at 170 kDa in F9 parental cells and a quasivisceral endoderm teratocarcinoma cell line, PSA5E (Fig. 9). Chromosome 11 on which this gene is located in the mouse showed no sign of abnormality in karyotype analyses, however. CONCLIJSIONS

AND

DISCUSSION

Although 5.51 cells have several cell surface proteins with altered characteristics, the only one that has been recognized as important in cell-cell compaction is uvomorulin. Uvomorulin appears to be an accessory in the formation of tight and intermediate junctions between epithelial cells (Obrink, 1986), and indeed, in the electron microscope we found that junctions between mu-

205, -

EGF-R

=,

1

2

3

4

Frc. 9. Altered EGF-receptor in F9 mutant cells. a”S-labeled lysates prepared as in Fig. 5 were immunoprecipitated with a polyclonal rabbit antiserum to EGF receptors and Sfaphylococcu.~ ~UWUS as described under Materials and Methods. After analysis on ‘7’3% SDSPAGE the autoradiograph revealed labeled 170-kDa EGF receptors in F9 cells (Lane 1) and PSA5E cells (Lane S-here, one-third the total number of cpm used in Lanes 1 and 2 was used). In mutant 5.51 cells (Lane 21, a diffuse radioactive zone from 150 to 1’70 kDa was seen, indicating aberrant EGF receptors were being synthesized.

tant cells were rudimentary, compared to wild-type (our unpublished observations). We showed earlier that normal intercellular communication via gap junctions is unaltered in the mutant F9 cell line (Grover et al., 1987). The major differences observed in mutant cell uvomorulin was the reduced level of protein (40-50% of wild-type) and in the extremely low level of uvomorulin transcripts. Several explanations for these observations are possible. For instance, finding that there is 50% uvomorulin protein suggests that one uvomorulin gene could have been inactivated by the mutagen and this is consistent with the karyotype wherein a possible inversion occurs on one of the two fused number 8 chromosomes. But this does not explain why levels of mRNA are so low. It therefore seems likely that an additional event, such as an alteration in stabilization sequences has occurred. Alternatively, a mutant regulatory gene affects transcription from both alleles. In spite of the extreme lability of uvomorulin mRNA in 5.51 cells, sufficient transcripts exist to code for uvomorulin protein. This protein is not detectably different from wild-type since it reacts similarly with both a polyclonal and a monoclonal antibody and also with the catenins (see below). It is unlikely that the uvomorulin defect lies merely in the glycosylation of the protein because there is no detectable difference in molecular size. Although some lectin sensitivities are modestly altered, and laminin A chain may be underglycosylated in differentiated 5.51 cells, nevertheless, there is no global impairment since SSEA-1 and Forssman antigens (both carbohydrate epitopes) are present at normal levels (Grover et ul., 1987). It is also unlikely that the associations between the 17-kDa cytoplasmic domain of uvomorulin and the group of cytoskeletal proteins called catenins LU,/3, and y (Ozawa et nh, 1989) are defective because antibodies to uvomorulin coprecipitate these complexes in mutant cells (Figs. 4 and 5). These associations are thought to be involved in the mediation or regulation of the adhesive activity of uvomorulin. The uvomorulin in the mutant cell line complexes with a similar proportion of catenins compared to wild-type cells. It is also unlikely that the defective fibronectin receptor system accounts for the inability of 5.51 cells to compact. Defective fibronectin receptors, however, do account for the poor adhesiveness and lack of flattening and spreading of 5.51 cells on fibronectin. The addition of GRGDSP hexapeptide to wild-type F9 aggregate cultures does not affect compaction (our unpublished data) and supports the lack of involvement of the fibronectin receptor. We also found that the lactosaminyl glycan receptor enzyme, galactosyl transferase is superabundant in

346

DEVELOPMENTAI,

Bro~oc:u

F9-5.51 att- cells, and therefore, the important adhesive property that this system provides to F9 cells (Shur, 1983) is not defective. We have examined here several of the cell-cell interactive components involved in EC cell adhesion and have found only one of these components to be altered. The most striking defect in two lines of mutant cells is that the levels of uvomorulin transcripts are so low that we detected them only after amplification using reverse transcription and PCR. This very low level is sufficient to provide uvomorulin synthesis at half the normal rate. A steady-state level of uvomorulin at only 50% of that in wild-type cells appears to be insufficient for compaction of mutant cells, but sufficient for their aggregation. Perhaps uvomorulin becomes active only after the interaction (dimerization?) between individual molecules, a process that would be much less efficient when only 50% uvomorulin is present. When mutant cells differentiate by the addition of RA, the deposition of uvomorulin becomes more visible (Fig. 1). However, during differentiation, mutant cells are stimulated to synthesize and secrete large amounts of matrix components (Grover et al, 1987), which accumulate and prevent close cell-cell interactions. All our observations suggest that uvomorulin activity in binding cells close together enough to form the tight and adherens junctions needed for epithelium formation (and hence embryoid body formation) in F9 cells must take effect before the buildup of matrix. Mutant cells are unable to do this and therefore cannot make an epithelial layer (Grover et al., 1987). There is no evidence either way that the mutant uvomorulin is altered in activity or is less prominently displayed on the cell surface. An alternative explanation is that another unidentified component that plays a role in compaction together with uvomorulin is also absent. This is possible in view of the incidence of altered forms of several cell surface proteins and the abnormalities seen in the karyotype. To make a conclusive diagnosis, we intend to introduce an expression vector coding for uvomorulin into 5.51 cells to effect a possible “cure” for the inability of this cell line to compact. While this manuscript was being prepared, we were informed of another compaction-defective embryonal carcinoma cell line derived from H6 wild-type EC cells. This mutant was selected in a similar fashion to the mutant line described here and has also been shown to have undetectable levels of uvomorulin (John W. Littlefieled and Laura L. Whitehouse, 1989, unpublished observations). I am grateful to Carolyn Wang for expert technical assistance. I thank Kathleen Sweeting for secretarial assistance. This work was supported by grants from the PHS, HD 18782 and CA 28427.

VOLIIME

138. 1990 REFERENCES

AL)AMSOX, E. D. (1986). Extraembryonic tissues as sources and sinks of humoral factors in development: Teratocarcinoma model systems. It/ “Cellular Endocrinology: Hormonal Control of Embryonic and Cellular Differentiation” (G. Serrero and J. Hayashi, Eds.), pp. 159-174. A. R. Liss, New York. BEIIK~:NS, J., BIK(‘IIM~X:K, W., GOODMAN, S. L., and IMHOF, 8. A. (1985). Dissociation of Madin-Darby canine kidney epithelial cells by the monoclonal antihodg anti-Arc-l: Mechanistic aspects and identification of the antigen related to uvomorulin. J. Cell Bid. 101, 1307-1315. BOLLER, K.. VE:STWEBER, D., and KEMLXR, R. (1985). Cell-adhesion molecule uvomorulin is located in the intermediate junctions of adult intestinal epithelial cells. J. (X1. Bid. 100, 327332. DAHL, S. C., and GRABEL, I,. B. (1989). Integrin phosphorylation is modulated during the differentiation of F9 teratocarcinoma stem cells. J. (5~11. Lliol. 108, 183-190. DAMXV, C. H., R1cw.4, J., SOI,TF,R, D., KNIIUSEN, K., and BUCK, C. A. (19831. Identification and purification of a cell surface glycoprotein medial intercellular adhrsion in embryonic and adult tissue. C’eU 34, 455-566. EISTETTER, H. I(., AUOLPH, S., RIN(;WAI,I), M., SIMON-CHAZOTTI%, D., SVH(IH, R., GT@NET, J-L., and KEMIXK, R. (1988). Chromosomal mapping of the structural gcnr coding for the mouse cell adhesion molecule uvomorulin. Procl Nut/. dctrd. Sci IJSA 85, 3489-3493. GROVER, A., and ADAMSON, E. D. (1986). Evidence for the existence of an early common biochemical pathway in the differentiation of F9 cells into visceral or parietal endoderm: Modulation by cyclic AMP. Ikv. Kid. 1 14 , 492-503. I GKOVI~:K, A., ANIIKEWS, G., and AI)AMSON, E. I). (1983a). Role of laminin in (pithelium formation by F9 aggregates. J. Cell Bid. 97, 137-144. GROVEK, A., OSHIMA, R. G., and ADAMSON, E. D. (1983b). Epithelial layer formation in differentiating aggregates of F9 embryonal carcinoma cells. .J. (‘e/I. Bid. 96, 1690-1696. GKO\-EK,

A., ROSENSTRAW, M. J., STERMAN, B., SNOUK, M. E., and E. D. (1987). An adhesion-defective variant of F9 embryonal carcinoma cells fails to differentiate into visceral endoderm. /Xc>. Llio/. 119, l-11. GITMHINEK, B., and SIMONY, K. (1986). A functional assay for proteins involved in establishing an epithelial barrier: Identification of a uvomorulin-like polypeptide. J. Cell Bid. 102, 457-468. AL)AMSON,

HYAFIL, F., BABINET, early embryogenesis: (‘dl 26, 447-454.

C., and JA(:oB, A molecular

F. (1981). approach

Celllcell interactions in to the role of calcium.

HYAF’II,, F., MOKELLO, D., BABINET, C., and JACOH, F. (1980). surface glycoprotein involved in the compaction of embryonal cinoma cells and cleavage stage embryos. Cc/l 21, 927-934.

A cell car-

IMADA, M., IMAI)A, S., IU’ASAKI, H., KTJME, A., YAMAGIJCHI, H., and MOORE, E. E. (1987). Fetomodulin: Marker surface protein of fetal development which is modulatable by cyclic AMP. Den. Bid. 122, 483-491. JOHNSON, M. H., CHAKRABARTY, J., HANDYSIDE, A. H., WILLISON, K., and STERN, F’. (1979). The effect of prolonged decompaction on the development of the preimplantation mouse embryo. J. Embryd E.rp, Morphol. 54, 241-261. KEMLER, R., BAUINET, C., EISON, antigen in early differentiation. 4449-4452. MANIATIS, lecular ratory,

H., and JACOB, F. (1977). Proc. N&l. Acud. Sci.

Surface USA 74,

T., FRITSCH, E. F., and SAMBROOK, J. (eds). (1982). 1n “MoCloning: A Laboratory Manual.” Cold Spring Harbor LaboCold Spring Harbor, NY.

ADAMSON,

BARIRAIJLT,

ANL) KEMLER

Nrcwr,~s, J. F., KEMIZR, R., and JA(:~R, F. (1981). Effects of anti-embryonal carcinoma serum on aggregation and metabolic cooperation between teratocarcinoma cells. Del,. Biol. 81, 127-132. OHKINK, B. (1986). Epithclial cell adhesion molecules, Exp. CeZell. Res. ltiB, lm26. OZAU’A, M., BARIKAIJLT, H., and KF,MLER, H. (1989). The cytoplasmic domains of the cell adhesion molecule uvomorulin associates with three independent proteins structurally related in different species. EMBO. .J. 8, 171 l-1717. PEYRIERAS, N., HYAFIL, F., LCXJVARII, D., PLEOGH, H. L., and JACOB, F. (1983). IJvomorulin: A nonintegral membrane protein of early mouse embryo. Proc. Nutl. Accrd. Sci. lJSA 80, 62’74-6277. PEYRII~;RAS, N., L~u~ARI), D., and JACOH, F. (1985). Characterization of antigens recognized by monoclonal and polgclonal antibodies against uvomorulin. Proc. N&l. Acud. Sci (ISA 82, 8067-8071. PYTELA, R., PIERSCHI$XIII~:K, M. D., and RIJOSLAHTI, E. (1985a). Identification and isolation of a 140 kd cell surface glgcoprotein with properties expected of a fibronectin receptor. Ceil 40, 191-198. PY~:I.A, R., PIEKSCHUACIIER, M. D., and RUOSI,.~IITI, E. (1985b). A 125/115 kDa cell surface receptor specific for vitronectin interacts with the arginine-glgcine-aspartic acid adhesion sequence derived from fibronectin. Ptw. h’utl. Accrti. Sci. USA 82, 5766-5770. RI(‘H~z, J.. DAMSKY, C. H., B~JCK, C. A., KNOMUS, D. B., and SOLTER, D. (19%). Cell surface glycoproteins mediate compaction, trophoblast attachment and endoderm formation during early mouse development. I)cl*. Rio/. 108, 513-521. RINC\\‘:ZI,I), M., S(:rrr’~r, R., VESTWER~~:R, D., EISTFXTER, H., LOTTSPEI(‘H, F., EN(:EL, J., DOLZ, R., JAHNIG, F., EPPLEN, J., MAYER, S., MIII,LEK, C.. and KEMLER, R. (1987). The structure of cell adhesion molecule uvomorulin: Insights into the molecular mechanism of (‘ii” -dq)endcnt cell adhesion. E,%fRO J. 12, 3647-3653.

Alf~rd

UILWWIvrdiv

E.rpression

347

SHIR~~YOSHI, Y., OKAUA, T. S., and TAKEICHI, M. (1983). The calciumdependent cell-cell adhesion system regulates inner cell mass formation and cell surface polarization in early mouse development. Cell 35, 631-638. SIIUR, B. D. (1983). Embryonal carcinoma cell adhesion: The role of surface galactosyltransferase and its 9Ok lactosaminoglycan substrate. Dw. Riol. 99, 360-372. SILVER, L. M., MARTIN, G. R., and STRICKLANI), S. (Eds.) (1983). I?( “Teratocarcinoma Stem Cells. Cold Spring Harbor Conferences on Cell Proliferation, Vol 10.” Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. VESTU’EBER, D., and KEMLER, R. (1984). Rabbit antiserum against a purified surface glgcoprotein decompacts mouse preimplantation cmhryos and reacts with specific adult tissues. Exp Ccd/. Res. 152, 169-178. VESTWE:BER, D., and KEMLEK, R. (1985). Identification of a putative cell adhesion domain of uvomorulin. EMBO J. 4.3393-3398. WELLER, A., MEEK, J., and ADAMSON, E. D. (1987). Preparation and properties of monoclonal and polyclonal antibodies to mouse epidermal growth factor (EGF) receptors: Evidence for cryptic EGF receptors in emhryonal carcinoma cells. Lhelopw~mt 100, 351-363. WEWER, U. M., LIOTTA, L. A., JAYE, M., R~ccn, G. A., DKOIIAN, W. N., CLAYSMITH, A. P., RAO, C. N., WIRTH, P., COLIGAN, J. E., ALBRECIITSEN, R., MITDRYI, M., and SOBEL, M. E. (1986). Altered levels of laminin receptor mRNA in various carcinoma cells that have different abilities to bind laminin. Proc Not/. Arced. Sci. USA 83, 7137-7141. YOSHIDA-NORO, C., SIJZUKI, N., and TAKEICHI, M. (1984). Molecular nature of the calcium-dependent cell-cell adhesion system in mouse terato and embryonic cells studied with a monoclonal antibody. nr,: Bid. 101 7 19-27. (