defective preimplantation development - PNAS

Proc. Natl. Acad. Sci. USA Vol. 92, pp. 855-859, January 1995 Genetics

A targeted mutation in the mouse E-cadherin gene results in defective preimplantation development (cell adhesion/desmosome formation/tight junction formation/morula decompaction/maternal mRNA)

DIETER RIETHMACHER*, VOLKER BRINKMANNt, AND CARMEN BIRCHMEIER*t *Max-Delbrueck-Laboratorium in der Max-Planck-Gesellschaft, Carl-von-Linne-Weg 10, 50829 Koeln, Germany; and tMax-Delbrueck-Center for Molecular Medicine, Robert-Roessle-Strasse 10, 13122 Berlin, Germany

Communicated by Michael H. Wigler, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, October 5, 1994 (received for review August 22, 1994)

ABSTRACT The Ca2+-dependent cell adhesion molecule E-cadherin functions in the establishment and maintenance of epithelial cell morphology during embryogenesis and adulthood. Downregulation or complete shut-down of E-cadherin expression and mutation of the gene are observed during the progression of tumors of epithelial origin (carcinomas) and correlate with the metastatic potential. We have introduced a targeted mutation into the E-cadherin gene by homologous recombination in mouse embryonic stem cells. The mutation removes E-cadherin sequences essential for Ca2+ binding and for adhesive function. These embryonic stem cells were used to generate mice carrying the mutation. Heterozygous mutant animals appear normal and are fertile. However, the homozygous mutation is not compatible with life: E-cadherin -/embryos show severe abnormalities before implantation. Particularly, the adhesive cells of the morula dissociate shortly after compaction has occurred, and their morphological polarization is then destroyed. Interestingly, the blastomers are still able to form desmosomes and tight junctions at sites of distorted cell-cell contact. Thus, maternal E-cadherin suffices for initial compaction of the morula but not for further preimplantation development to occur.

of the morula commences. However, it is then distributed in a nonpolar manner and is nonadhesive (10, 11). The exact mechanism that renders E-cadherin functional is not known, but phosphorylation of E-cadherin has been implicated in the regulation (11). Controlled loss of epithelial adhesion and polarity that causes mesenchymal cell morphology can occur in development, for example, during mesoderm formation. A similar but uncontrolled epithelial-mesenchymal conversion is observed in malignant tumors of epithelial origin. Malignant carcinoma cells are characterized in general by poor intercellular adhesion, loss of the differentiated epithelial morphology, and increased cellular motility. Downregulation or a complete shut-down of E-cadherin expression, mutation of the Ecadherin gene, or other mechanisms that interfere with the integrity of the adherens junctions are observed in carcinoma cells (2, 12, 13). In human tumors, the loss of E-cadherinmediated cell adhesion correlates with the loss of the epithelial morphology and with the acquisition of metastatic potential by the carcinoma cells (14). To study the role of E-cadherin genetically, we have altered functionally important gene sequences by homologous recombination in embryonal stem (ES) cells, and these modified ES cells were used to generate mice that carry the mutation. In the homozygous state, the mutation affects preimplantation development severely. A normal morphology of the mutant embryos up to and including the compaction stage of the morula is observed; consequently, the adhesive cells dissociate and appear morphologically nonpolar. Interestingly, desmosomes and tight junctions are found in the distorted embryos. The availability of mice carrying the E-cadherin mutation in a heterozygous state will allow the genetic analysis of the role of epithelial cell adhesion in tumor progression.

Cellular adherence in developing and adult tissues is mediated by cell adhesion molecules, of which one class, the cadherins, functions in a Ca2+-dependent manner (1). E-cadherin, one member of this family, is responsible for the cell adhesion of epithelial cells and also plays an essential role in generation and maintenance of epithelial cell polarity (2, 3). E-cadherin is located in the adherens junctions, which are specialized regions on the lateral side of the epithelial plasma membrane. The extracellular domain of E-cadherin binds Ca2+ and Ecadherin expressed on neighboring cells (1, 4). The cytoplasmic domain interacts with a group of proteins, the catenins, that mediate the contact with the microfilament network through unknown linkage molecules. a-Catenin has structural similarities with vinculin, which is located at the membrane attachment sites of microfilaments in fibroblasts; ,B-catenin is similar to amardillo of Drosophila; and -y-catenin is identical or closely related to plakoglobin found also at the desmosomes (5-7). Epithelial differentiation and polarization occur early in ontogeny, when in the morula stage, the embryo compacts and each cell polarizes along its apicobasal axis to generate an epithelial-like phenotype (8). E-cadherin plays an important role in the adhesion of the blastomers, since functionally interfering anti-E-cadherin antibodies cause the early embryo to decompact (9, 10). Programing of the embryo for compaction requires previous protein synthesis. Also E-cadherin is synthesized and is present in the membrane before compaction

MATERIALS AND METHODS Construction of a Targeting Vector. E-cadherin DNA was isolated from a 129/J genomic library; mouse cDNA corresponding to positions 953-1322 (see ref. 15) was used as a probe. The targeting vector schematically shown in Fig. 1 was constructed by standard recombinant techniques; a fragment of the E-cadherin gene that includes a part of exon 7 and the entire exon 8 (corresponding to nt 957-1211 in the cDNA; see ref. 15) was removed and replaced by a neomycin (neo)resistance gene. In addition, termination codons in all three reading frames present in the promotor of the neo-resistance gene were inserted. Also, the thymidine kinase gene from Herpes simplex virus is present as negative selection marker

(16). Targeting of ES Cells and Production of Transgenic Animals. E14-1 cells (17), a subclone of the E-14 line, were used

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: ES, embryonic stem;

neo,

neomycin.

ITo whom reprint requests should be addressed.

855

856

Proc. NatL Acad ScL USA 92

Genetics: Riethmacher et alP 6 78

9

x

0

x

x

6

X

I

9

Ino

6

X

1 v.

9

:

pB

pC

ElJ

o::1

x

x

n.

I

x

lIneo pA

x

-0

(1995)

x

l

x

C

-_

tk

exon 7

11kb neo

-4 I

kb 4.8 kb

kb 6.2 kb

b

c 1

2

3

4

_

_

kb -11

il~~~ -6.2

1

bp 439,

2

3

4

5

p

404,381 -, 242 190-' 147 -f

FIG. 1. Strategy used to mutate the murine E-cadherin locus. (a) Schematic representation of the wild-type E-cadherin allele (top line), the targeting vector (middle line), and the mutated allele (bottom line). The Xba I sites in the DNA and the size of the Xba I fragments detected by the different probes in wild-type and mutant genomic DNAs are indicated. To verify the structure of the mutant locus, Southern blot hybridizations were carried out with probes pA, pB, and pC. The primers for the PCR used to genotype embryos are indicated by arrows. (b) Southern blot analysis of genomic DNA from +/+ (lane 1) and mutant +/- (lane 2) ES cells and from +/+ (lane 3) and +/- (lane 4) animals performed with probe pA. The sizes of the hybridizing DNA fragments are indicated. (c) Genotype analysis of embryos by PCR. Lanes: 1, marker DNA; 2, +/+ embryo; 3, +/- embryo; 4, -/- embryo; 5, negative control (no DNA added). Sizes of marker fragments are indicated.

to introduce the E-cadherin targeting vector by electroporation. Targeted ES cell clones were enriched by selection with G418 (400 ,ug/ml) and gancyclovir (2 ,uM) and identified by PCR. Southern blot analysis on the ES cell clones with three probes [a probe corresponding to cDNA sequences outside of the targeting vector (pA), a probe corresponding to cDNA sequences present on the targeting vector (pC), and sequences of the neo-resistance gene (pB) (see Fig. 1)] was used to verify that a single integration event had occurred by homologous

recombination. Blastocyst injections and identification of chimeras that transmit the E-cadherin mutation were performed essentially as described (18). Routinely, the genotype of animals and embryos was identified by PCR with primer combinations CCTCTCCTTTGACAGGAACCTCCGT (complementary to the intron 6-exon 7 junction) and CGAATTCGCCAATGACAAGACGCTGG (complementary to promoter sequences of the neo-resistance gene) or CCTCTCCTTTGACAGGAACCTCCGT (complementary to the intron 6-exon

f

I

FIG. 2. Morphology of embryos with mutant E-cadherin gene at various stages of early development. Appearance of +/+ (a-f) and (a'-f) embryos as observed by phase-contrast microscopy (a-e and a'--e') and by light microscopy after semithin sectioning (f andfl). Embryos at four-cell stage (a and a'), precompacted morulae (b and b'), compacted morulae (c and c'), blastocyst (d and f), and distorted -/- embryo at comparable stage (d' and if), hatching blastocyst (e), and destorted -- embryo at comparable stage (e'). (x 165.)

Genetics: Riethmacher et al.

Proc. Natl. Acad Sci. USA 92 (1995)

857

7 junction) and CAGCCCAGAGGTGAGCACACTGATG (complementary to exon 7 sequences) for the mutant or wild-type locus, respectively. In selected cases, the genotype of the animals was verified by Southern blot analysis. Analysis of Early Embryos. Heterozygous animals were mated and the morning past coitum was assigned as day 0.5 of development. On day 1.5 of development, the embryos were isolated by flushing the infundibulum. Further development of the embryos was observed in vitro by incubation in M16 medium for 48-96 h in a humidified C02/air chamber (19). Where indicated, the zona pellucida of the embryos was removed by incubation with acid Tyrode's solution (19). Immunohistochemistry and Electron Microscopy. Embryos were fixed in 4% (vol/vol) formaldehyde, stained with a rat anti-mouse E-cadherin IgG (DECMA-1, Sigma) and a Cy3coupled goat anti-rat IgG (Dianova, Hamburg), and viewed with a confocal laser scanning microscope (Lasersharp MRC 500, Bio-Rad) or a Zeiss Axiophot fluorescence microscope. For sectioning, the embryos were positioned in small droplets of agarose, fixed in 2.5% (vol/vol) glutaraldehyde, and postfixed in 1% OSO4. After staining with tannic acid and uranyl acetate, the tissue was dehydrated and embedded in Epon 812. Semithin sections (0.5 gm) were stained with methylene blue and examined by light microscopy; thin sections (0.1 gm) were contrasted with lead citrate and examined by electron microscopy.

RESULTS Generation of ES Cells and Mice with a Mutant E-Cadherin Gene. The targeting vector for the E-cadherin gene was constructed with 129/J genomic DNA and contained the neo-resistance gene and the Herpes simplex virus thymidine kinase gene as positive and negative selection markers, respectively (16). To introduce a null mutation, sequences of exon 7 and 8 of E-cadherin were removed (Fig. 1); the deleted fragment includes DNA encoding a Ca2+-binding site reported (20) to be essential for E-cadherin function. The targeting vector was introduced by electroporation into the E14-1 ES cells (17). Cells that had introduced the DNA by homologous recombination were enriched by selection with G418 and gancyclovir and identified by PCR; nine positive clones were found in the 250 colonies analyzed. Selected colonies were subcloned and their genomic DNA was analyzed by Southern blot hybridization to verify that a single integration event had occurred by homologous recombination (Fig. 1). Mutant ES cells derived from three subclones were injected into blastocysts of C57BL/6 animals. Germ-line chimeras were obtained from blastocysts injected with one subclone. The chimeras were used to produce animals heterozygous for the mutation that appeared normal and were fertile. However, among the offspring from matings of heterozygous animals, no viable homozygous mutant animals were found, indicating that such a genotype is not compatible with normal development. Since early development is disturbed by anti-E-cadherin antibodies (9, 10), we next analyzed preimplantation development. Early Development of E-Cadherin -I- Embryos. Embryos were isolated at the two- or four-cell stage, transferred into M16 medium, and incubated to observe early development in vitro. After the observation was completed, the individual embryos were genotyped by PCR (Fig. lc). Under these conditions, control embryos developed appropriately up to the late blastula stage. When embryos from heterozygous matings were analyzed, initially all developed normally and reached the compaction stage of the morula (at 18-24 h of incubation in vitro). However, all homozygous mutant (-/-) embryos decompacted consequently and the individual embryonal cells became morphologically depolarized (at 36-48 h of incubation in vitro) but continued to divide (Fig. 2). To assess the viability of the destorted embryos, we used the vital stain trypan blue

FIG. 3. Analysis of embryos with mutant E-cadherin gene by vital staining and by immunohistochemistry. Trypan blue staining of homozygous mutant,embryos after removal of the zona pellucida without (a) and with (d) prior treatment with sodium azide (NaN3). Note that the stain is excluded from the viable embryonal cells unless they are poisoned by pretreatment with NaN3. Immunological analysis of E-cadherin expressed on homozygous mutant (b and c) or wild-type (e and f) embryos with anti-E-cadherin antibodies (DECMA-1); embryos of comparable developmental age without the zona pellucida were used. Embryos after 60 h of in vitro incubation (b and e) or after 36 h of in vitro culture (c, f), analyzed by immunofluorescence microscopy, are shown; note that to achieve visability of the maternal E-cadherin, the photograph shown in c was exposed five times longer than the photograph shown in f. (Bars = 25 ,Lm.)

that was excluded from blastomers, whereas pretreatment with sodium' azide allowed trypan blue to enter (Fig. 3 a and d). Therefore, decompaction of the embryos is not the result of unspecific cell death. The - / - embryos never formed blastocysts with a proper blastocoel; instead, loose cell aggregates were observed (at 72-96 h of incubation in vitro; Fig. 2 d'-f ). In contrast, all wild-type (+/ +) or heterozygous mutant (+/-) embryos formed normal blastocysts containing a blastocoel (at 48-72 h of incubation in vitro). The blastocysts consequently expanded; the majority ruptured the zona pellucida and emerged from the zona in the hatching process, whereas the remaining embryos collapsed after expansion (at 72-96 h of incubation in vitro; Fig. 2 d-f). In total, we analyzed

858

Genetics: Riethmacher et al.

a;,'is"S)*~ . .

Proc. Natl. Acad Sci. USA 92

k

.

4. t?s

i:

(1995)

_'

,-, 7 ., _

+

_

tAr:0..

FIG. 4. Appearance of cell contact sites in embryos with mutant E-cadherin gene. Electron microscopic analysis of cell contact sites from distorted homozygous mutant (a) and wild-type (b) embryos at comparable developmental age. Desmosomes characterized by protein material between membranes and the attached cytoplasmic filaments (D), tight junctions characterized by the contacting membranes (T), and adherens junctions characterized by protein material between parallel membranes (A) are indicated. Note the interdigitating membranes (I) of adjacent cells in the mutant embryo. (Bars = 100 nm.)

153 embryos from heterozygous matings. Thirty-seven (24%) had a -/- genotype, and all of these embryos appeared distorted at the blastocyst stage. Seventy-six (50%) and 40 (26%) had a +/- and +/+ genotype, respectively, and all formed blastocysts with proper blastocoel. Maternal E-Cadherin Functions in the Compaction of the Morula. Antibodies against E-cadherin have been reported to decompact embryos at the morula stage, indicating that Ecadherin functions in the Ca2+-dependent adhesion and polarization of early morula cells (9, 10). Nevertheless, our embryos carrying a homozygous mutation in the E-cadherin gene are initially able to compact but do not sustain the compacted state. We studied -/- embryos in the presence of anti-E-cadherin antibodies (DECMA-1) to determine whether morula compaction occurred. Interestingly, such antibodies inhibited the compaction of -/- embryos. Immunohistological analysis of E-cadherin -/-embryos demonstrated the presence of residual E-cadherin protein, albeit only up to the morula stage (Fig. 3c). After the morphological distortion was apparent (at 60 h of incubation in vitro), E-cadherin protein could not be observed by immunohistochemistry (Fig. 3b). Thus, the functional E-cadherin protein present during the initial compaction of -/- embryos is probably derived from maternal sources (see ref. 11). Morphological Appearance of Cell Contact Sites in Mutant Embryos. Electron microscopy was used to observe the cell contact sites of the distorted embryos; these were found to be severely altered in morphology. The membranes of opposing inner and outer blastomers formed an irregular interdigitating pattern (Fig. 4). These interdigitating membranes are typical for - / - embryos and no comparable structures are present in embryos with normal external morphology (+/+ and +/embryos). Interestingly, in distorted embryos, we found desmosomes in cell contact sites of both inner and outer blastomers and tight junctions at the cell contact sites between outer blastomers (Fig. 4).

DISCUSSION We have introduced a mutation into the mouse E-cadherin gene by homologous recombination in ES cells; the mutant ES cells were used to produce mice carrying the mutation. Since functionally essential coding sequences were removed in the E-cadherin gene (see ref. 20) and termination codons in all three reading frames were introduced, no mRNA encoding adhesive E-cadherin protein can be produced in homozygous mutant cells. Animals that carry one copy of this mutant E-cadherin gene appear healthy and are fertile. However, matings between such animals do not produce homozygous mutant offspring, demonstrating the essential function of E-cadherin for normal development. E-Cadherin Is Essential for Cell Polarity in the Early Mouse Embryo. Embryos that are homozygous for the Ecadherin mutation develop normally up to the morula stage and compact properly. The morula cells become initially adherent and polarize. However, the polarized state of the cells cannot be sustained, and consequently, the embryos appear severely distorted. Mutant (-/-) embryos do not form normal blastocysts and hatching from the zona pellucida is never observed. Since hatching is a prerequisite for implantation of the embryos, further development cannot proceed in vivo. It has been shown (9, 10) that functionally interfering antibodies against E-cadherin decompact the morula. This indicates that the early cell polarization during compaction of the morula is dependent on E-cadherin. Nevertheless, embryos homozygous for the E-cadherin mutation are initially able to compact and their cells are able to polarize. Anti-Ecadherin antibodies inhibit this initial compaction, demonstrating that E-cadherin function is not compensated by other cell adhesion molecules. It has been established (11) that maternal E-cadherin is present in the early mouse embryo. Thus, it is likely that the E-cadherin that functions in the compaction of the mutant morulae is derived from maternal sources. Indeed, we were able to demonstrate the presence of E-cadherin in the early but not the distorted mutant embryos

~~~Proc. Natl. Acad. Sci. USA 92 (1995)

Genetics: Riethmacher et al. Genetics: Riethmacher et al.

by immunofluorescence. Therefore, sufficient maternal Ecadherin protein must exist in the mutant cells to initiate but not to sustain compaction. E-Cadherin-Dependent Adhesion Is Not Essential for the Maintenance of Desmosomes and Tight Junctions in ECadherin Mutant Embryos. Electron microscopic analysis of distorted E-cadherin mutant embryos showed structurally abnormal sites of cell-cell contacts. Particularly, the two membranes of the contacting cells formed parallel interdigitated patterns. Similar membrane morphology can be observed in dense-filter cultures of MDCK epithelial cells (21). Despite the aberrant morphology, tight junctions and desmosomes were found at sites of disturbed cell-cell contacts. These were still present when the mutant embryos were

junctions

incubated for 2 days after the loss of E-cadherin-mediated cell adhesion. This is in contrast to previous observation with MDCK epithelial cells in culture that show no desmosomes and tight junctions in the absence of E-cadherin-mediated adhesion (22). Thus, E-cadherin-containing adherens junctions are not necessary or are only an initial prerequisite for the formation of desmosomes and tight junction in the early mouse

embryo.

E-Cadherin and Tumor Progression. Downregulation or missing E-cadherin expression is typical for a variety of malignant carcinomas and correlates with the invasive properties of the tumor cells in vitro and in vivo (14). In many instances, downregulation is controlled at the transcriptional level (23). In diffuse-type gastric carcinomas, mutations in the E-cadherin gene are observed in 50% of the cases. In addition, the second intact allele is often lost in such tumor cells (12). Mutation and loss of the intact second copy are typical for genes that play a causative role in the development and progression of tumors (24). Other mechanisms that interfere with the integrity of the adherens junction, for instance, mutation of a-catenin, have been reported to occur in tumor cells as well (13). In general, these changes are late events in tumor progression and are associated with the acquisition of metastatic potential of carcinoma cells. There is no evidence that loss of E-cadherin or of functional adherens junctions influences growth or genetic stability of tumor cells. So far, we

have not observed any increase of spontaneous tumor formation in our heterozygous animals, the oldest being 10 months of age at the time this manuscript was submitted. However, it will be interesting to analyze tumor progression after a controlled induction of tumors in such animals, for instance by oncogene expression or the administration of carcinogens.

Note Added in Proof. After submitting our manuscript, we learned of a similar publication by Larue et aL. (25). are extremely grateful to Drs. Klaus Rajewsky, Werner Mueller, Ralph Kuehn (Institute of Genetics, Universitaet zu Koeln) for a

We

and

859

on blastocyst injections and ES cell culture. We thank Dr. Walter Birchmeier for helpful discussions and critically reading the manuscript, Dr. Georg Kreimer (Botanisches Institut der Universitaet zu Koeln) for help with the laser scanning microscopy, Dr. Juergen Behrens for advice on the interpretation of the electron microscopy data, Dr. Andrew Smith (Medical Research Council, Cambridge) for a gift of the 129/J genomic library, and Udo Ringeisen for preparing the figures. This work was supported by Grant 0316150a from the Bundesministerium fuer Forschung und Technolo-

gift of E14-1 cells and advice

gie. 1 . Takeichi, M. (1991) Science 251, 1451-1455. 2. Behrens, J., Mareel, M., Van Roy, F. & Birchmeier, W. (1989) J.1 Cell Biol. 108, 2435-2447. 3. McNeill, H., Ozawa, M., Kemler, R. & Nelson, W. (1990) Cell 62, 309-316. 4. Kemler, R. (1992) Semin. Cell Biol. 3, 149-155. 5. McCrea, P., Turck, C. & Gumbiner, B. (1991) Science 254, 1359-1361. 6. Herrenknecht, K., Ozawa, M., Eckerskorn, C., Lottspeich, F., Lenter, M. & Kemler, R. (1991) Proc. Nati. Acad. Sci. USA 88, 9156-9160. 7. Knudsen, K. & Wheelock, M. (1992) 1. Cell Biol. 118, 671-679. 8. Fleming, T., Hay, M. & Javed, Q. (1992) Development Suppi. (Cambridge, U. K) 17, 105-113. 9. Hyafil, F., Morello, D., Babinet, C. & Jacob, F. (1980) Cell 21, 927-934. 10. Vestweber, D. & Kemler, R. (1984) Exp. Cell Res. 152, 169-178. 11. Sefton, M., Johnson, M. & Clayton, L. (1992) Development (Cambridge, U.K) 115, 313-318. 12. Becker, K., Atkinson, M., Reich, U., Huang, H., Nekarda, H., Siepert, J. & Hoefler, H. (1994) Cancer Res. 54, 3845-3852. 13. Hirano, S., Kimoto, N., Shimoyama, Y., Hirohashi, S. & Takeichi, M. (1992) Cell 70, 293-301. 14. Birchmeier, W. & Behrens, J. (1994) Biochim. Biophys. Acta 1198, 11-26. 15. Nagafuchi, A., Shiraoshi, Y., Okazari, K., Yasuda, K. & Takeichi, M. (1987) Nature (London) 329, 341-343. 16. Mansour, S., Thomas, K. & Capecchi, M. (1988) Nature (London) 336, 348-352. 17. Kuehn, R., Rajewsky, K. & Mueller, W. (1991) Science 254, 707-710. 18. Bradley, A. (1987) in Teratocarcinimas and Embryonic Stem Cells: A Practical Approach, ed. Robertson, E. (IRL, Oxford), pp. 113-157. 19. Hogan, B., Constantini, F. & Lacy, E. (1986) Manipulating the Mouse Embryo (Cold Spring Harbor Lab. Press, Plainview, NY). 20. Ozawa, M., Engel, J. & Kemler, R. (1990) Cell 63, 1033-1038. 21. Braendli, A., Parton, R. & Simons, K. (1990) J. Cell Biol. 111, 2909-2921. 22. Gumbiner, B., Stevenson, B. & Grimaldi, A. (1988) J. Cell Biol. 107, 1575-1587. 23. Behrens, J., Loewrick, 0., Klein-Hitpass, L. & Birchmeier, W. (1991) Proc. Natl. Acad. Sci. USA 88, 11495-11499. 24. Knudson, A. (1986) Annu. Rev. Genet. 20, 231-251. 25. Larue, L., Ohsugi, M., Hirchenhain, J. & Kemler, R. (1994) Proc. Natl. Acad. Sci. USA 91, 8263-8267.