Desmocollin 3 (Dsc3) is a transmembrane glycoprotein that belongs to the cadherin family of cell adhesion receptors. Together with desmoglein(s), it forms the transmembrane core of desmosomes, a multiprotein complex involved in cell adhesion, organization of the cytoskeleton, cell sorting and cell signaling. Previous reports have suggested that Dsc3 synthesis is largely restricted to stratified epithelia, and that it plays a role in the proper differentiation of these tissues during mammalian embryonic development. To test these hypotheses, we generated Dsc3-null mice. Unexpectedly, homozygous mutants show a pre-implantation lethal phenotype. In fact, most mutants die even before mature desmosomes are formed in the embryo, suggesting a new and unexpected role of Dsc3 during early development.
Desmosome-mediated cell adhesion is established by heterophilic interactions between type 1 transmembrane glycoproteins that belong to the desmocollin (Dsc) and desmoglein (Dsg) family of cadherins. Through direct and indirect interactions with the cytoplasmic proteins desmoplakin, plakoglobin and plakophilin(s), Dsc and Dsg are connected to the intermediate filament cytoskeleton (Getsios et al., 2004). It has been demonstrated that this multiprotein adhesion complex is required to maintain the mechanical integrity and function of certain tissues and organs, in particular skin and heart.
Neonatal defects in stratified epithelia and epithelial appendages have been observed in mice with mutations in desmosomal genes [Dsc1 KO mouse (Chidgey et al., 2001); Dsg3 KO mouse (Koch et al., 1997; Koch et al., 1998); Dsg4 mutants (Jahoda et al., 2004; Kljuic et al., 2003; Moss et al., 2004); for references regarding human skin and heart diseases caused by mutations in desmosomal genes see previous work (Cheng and Koch, 2004; Chidgey, 2002)]. Loss of desmosomal proteins in knockout mice can also lead to embryonic lethality around the time of implantation [Dsg2 KO mouse (Eshkind et al., 2002)], shortly after implantation [desmoplakin KO mouse (Gallicano et al., 1998; Gallicano et al., 2001)] or at mid-gestation [plakoglobin KO mouse (Bierkamp et al., 1999; Bierkamp et al., 1996; Isac et al., 1999; Ruiz et al., 1996) and plakophilin 2 KO mouse (Grossmann et al., 2004)]. Most of these null phenotypes can be explained by tissue fragility resulting from impaired desmosome function. However, it has been hypothesized that loss of cell adhesion might also lead to an interruption of short-range cell communication, e.g. by separating receptor/ligand interactions at the plasma membrane, resulting in abnormal cell differentiation (Kljuic et al., 2003). Furthermore, based on the complex expression patterns of the desmosomal cadherins during development, it is likely that the specific set of Dsc (and Dsg) proteins fine tunes the adhesive properties of cells, and thus contributes to morphogenetic movements and cell sorting during development. This hypothesis is supported by experiments showing that differential expression of desmosomal cadherins contributes to cell sorting of mammary gland epithelial cells in vitro (Runswick et al., 2001).
Very little is known about the role of desmocollins during embryonic development. Three Dsc genes (Dsc1, Dsc2, Dsc3) have been identified in mammals (Cheng and Koch, 2004). Each of these genes encodes two proteins (an `a' and a `b' form), generated by differential splicing, that differ only with respect to their C-terminal (cytoplasmic) domains. Recently, Chidgey and colleagues have demonstrated that Dsc1 is required for normal epidermal development (Chidgey et al., 2001). Newborn Dsc1-null mice develop epidermal acantholysis (loss of cell-cell adhesion) and skin barrier defects. Furthermore, these authors observed an abnormal keratinocyte differentiation pattern.
There is currently no information available regarding the role of Dsc2 and Dsc3 during embryonic development. In the present study, we focused on Dsc3. It has been reported that Dsc3 is first synthesized in stratified epithelia of late gestation-stage embryos (Chidgey et al., 1997; King et al., 1997), where it is thought to contribute to proper epidermal differentiation. Given these results, it was expected that a Dsc3-null mutation would lead to structural and functional defects restricted to stratified epithelia and their appendages, such as hair follicles. Surprisingly though, we found that Dsc3-null mutants die in utero prior to embryo implantation, long before stratified epithelia and epithelial appendages are formed. In fact, our results indicate that most homozygous mutants die before mature desmosomes are formed in the trophectoderm of pre-implantation-stage embryos at E3.5, suggesting that Dsc3 might have extra-desmosomal functions in early development.
Generation of Dsc3+/– mice
The basic strategy used to generate mutant mice was to delete part of the promoter, exon 1 (which includes the start codon) and part of intron 1 (Fig. 1). We have previously used this approach to generate a null mutation in the Dsg3 gene successfully (Koch et al., 1997; Koch et al., 1998). Of the 288 antibiotic-resistant ES cell clones that were tested by Southern blot, 32 (11%) scored positive for recombination at the Dsc3 gene locus with our 5′ probe (for details, see Materials and Methods). Two ES cell clones that had been isolated in different electroporation experiments were further characterized. Both the 5′ and the 3′ probe confirmed homologous recombination, and hybridization with a neo probe showed that no additional copy of the targeting vector had integrated into the genome of these clones (see Fig. 1E). Both clones were injected into C57Bl/6 blastocysts and the resulting male chimeras were bred with C57Bl/6 females to establish two mutant lines (line 1 and line 2). Both lines showed the same phenotype. Unless stated otherwise, all experiments described below were carried out with line 1 mice.
Dsc3+/– mice are phenotypically normal
Wild type and Dsc3+/– mice were indistinguishable with respect to birth weight, growth and fertility. It has been reported that Dsc3 is mainly synthesized in stratified epithelia (Cheng and Koch, 2004). Therefore, we carefully analyzed skin, tongue, palate, esophagus and forestomach of newborn and adult Dsc3+/– mice by conventional histology. No gross abnormalities were detected (data not shown). Next, we used a panel of antibodies against epidermal differentiation markers (keratin 6, keratin 14, keratin 10, loricrin, filaggrin, desmoplakin, Dsg1, Dsg2, Dsc1, plakoglobin, plakophilin 1, plakophilin 3) to determine whether epidermal differentiation was affected in these mice. As judged by immunofluorescence microscopy, the expression of these markers was normal in Dsc3+/– mice (data not shown). The data summarized above indicated that epidermal development and keratinocyte differentiation was not affected in heterozygous mutant mice.
Quantitative PCR (QPCR) and western blots demonstrated that Dsc3 expression was approximately halved in the epidermis of Dsc3+/– mice (Fig. 1F,G). Furthermore, we did not observe additional bands in western blots of epidermal protein extracts with our Dsc3 antibody, indicating that the recombinant Dsc3 locus does not encode a truncated Dsc3 protein. Organs and tissues that express Dsc2, but not Dsc3, showed similar Dsc2 expression levels in mutants and wild-type controls (data not shown), demonstrating that our gene-targeting experiments did not affect the Dsc2 gene locus. In summary, these results demonstrate that we have generated a true Dsc3-null mutation.
The Dsc3-null mutation is embryonic lethal
Next, we intercrossed line 1 and line 2 Dsc3+/– mice, respectively, in order to generate homozygous mutants. Surprisingly, we did not find homozygous mutants among 232 mice that were genotyped at weaning age (20-21 days of age; Table 1). We carefully monitored pups from Dsc3+/– intercrosses from birth onwards, and did not find evidence for an increase in neonatal lethality. The average litter size of Dsc3+/– females was slightly smaller than that of wild-type females on a similar genetic background. Furthermore, the ratio of Dsc3+/+ to Dsc3+/– pups was ∼1:2. Dsc3–/– mice were also absent in litters from line 1 × line 2 crosses. All of these observations are consistent with an embryonic lethal phenotype of the Dsc3-null mutation.
|Dsc3 mutant line .||Animals genotyped .||+/+ .||+/- .||-/- .|
|Line 1 intercrosses||126||40 (32%)||86 (68%)||0|
|Line 1 × line 2 crosses||40||10 (25%)||30 (75%)||0|
|Line 2 intercrosses||66||23 (35%)||43 (65%)||0|
|All crosses||232||73 (31%)||159 (69%)||0|
|Dsc3 mutant line .||Animals genotyped .||+/+ .||+/- .||-/- .|
|Line 1 intercrosses||126||40 (32%)||86 (68%)||0|
|Line 1 × line 2 crosses||40||10 (25%)||30 (75%)||0|
|Line 2 intercrosses||66||23 (35%)||43 (65%)||0|
|All crosses||232||73 (31%)||159 (69%)||0|
Dsc3 is expressed throughout embryogenesis
Given that Dsc3 has been described as a protein synthesized in stratified epithelia, one would have expected neonatal defects in these tissues (e.g. epidermis and mucous membranes) of Dsc3–/– mice. However, embryonic lethality was unexpected. Even a complete absence of stratified epithelia, for example in p63-null mice, does not result in embryonic death (Mills et al., 1999; Yang et al., 1999). Therefore, we decided to determine the onset of Dsc3 expression during mouse embryogenesis in order to define a time window when Dsc3–/– embryos might die. A northern blot analysis showed that Dsc3 is expressed as early as embryonic day 11.5 (E11.5; Fig. 2A), which was consistent with results reported by others at E12.5 and E13 (King et al., 1997; Chidgey et al., 1997). Interestingly, a more-sensitive RT-PCR analysis showed that Dsc3 is also expressed earlier in development (Fig. 2B). The earliest time point tested by RT-PCR was E7.5. At earlier stages of development, it becomes more difficult to prepare embryos completely free of maternal tissue (which would lead to false positive results). Tissue sections of peri-implantation embryos from E6.5 to E8 were therefore analyzed by immunohistochemistry, immunofluorescence and in-situ hybridization. Co-expression of Dsc3 and the desmosomal proteins Dp and Dsg2 was observed in the embryonic ectoderm (which will form the embryo proper) and the ectoplacental cone at E6.5-E7 (see Fig. S1 in the supplementary material). Interestingly, it appeared that Dsc3 only partially overlapped with Dp and Dsg2 in the decidua, the maternal tissue that surrounds the embryo (see Fig. S1 in the supplementary material). Further experiments will be required to elucidate the subcellular distribution and function of Dsc3 in these cells.
Next, we analyzed pre-implantation embryos at E3.5 (blastocysts). At this stage, desmosomes are present in the trophectoderm (TE), a single-layered polarized epithelium that forms the outer cell layer of the embryo (Fleming et al., 2000; Jackson et al., 1980). As shown in Fig. 2C,D, we detected Dsc3 expression in these embryos by RT-PCR and western blotting. Consistent with previous results from others (Eshkind et al., 2002; Fleming et al., 1994; Gallicano et al., 1998), we observed a punctuated staining pattern of TE cells with antibodies against desmosomal maker proteins, such as desmoplakin (Fig. 3A), in immunofluorescence microscopy. However, Dsc3 staining was linear along the cell-cell borders of TE cells, i.e. was similar to the pattern observed with antibodies against E-cadherin or β-catenin (see Fig. 3A). This suggests that at this stage of development, Dsc3 is not restricted to desmosomes. Unfortunately, we could not analyze the subcellular distribution of Dsc3 in pre-implantation embryos further, because our antibody does not stain specimens that have been subjected to the fixation procedures required for immunoelectron microscopy.
Dsc3 mRNA is also present at earlier stages of development. RT-PCR revealed the presence of Dsc3, Dsg2, desmoplakin and plakoglobin transcripts at the pre-compaction eight-cell stage (Fig. 2E). As judged by RT-PCR, Dsc3 was already present in unfertilized eggs (Fig. 2F). Immunohistochemical staining of embryos at various stages of pre-implantation development confirmed the presence of the protein (Fig. 3B). Furthermore, Dsc3 staining was already present in the cytoplasm of oocytes, as determined by staining of ovary sections.
Previous electron microscopy studies have revealed the presence of desmosome-like structures between mammalian oocytes and the surrounding follicle cells (Anderson and Albertini, 1976; Zamboni, 1974). Given the cytoplasmic staining pattern, it appears that Dsc3 is not specifically enriched in these desmosome-like cell junctions. Interestingly, Collins and colleagues have reported that Dsc2 mRNA is also present throughout pre-implantation development, with the exception of post-compaction embryos at the eight-cell stage (Collins et al., 1995). Our own RT-PCR results indicate that Dsc2 mRNA is present in blastocysts (Fig. 2C). However, we did not detect this mRNA in pre-compaction stage embryos (data not shown). It remains to be seen whether this is due to a lower sensitivity of our RT-PCR or due to differences in the Dsc2 expression profiles between our mice and those of Collins et al.
Dsc3–/– embryos die prior to implantation
Post-implantation embryos from Dsc3+/– intercrosses were isolated at different stages of development and genotyped. We analyzed, for example, 31 embryos at E7.5, none of which was a homozygous mutant (24 were Dsc3+/– and seven were Dsc3+/+ embryos). This suggested that Dsc3–/– embryos died either before or shortly after implantation.
Next, we genotyped embryos at the blastocyst stage at which mature desmosomes are formed in the trophectoderm. Out of 205 embryos that reached the blastocyst stage, only four (2%) were Dsc3–/– [42 (20%) were Dsc3+/+; 159 (78%) were Dsc3+/–]. Next, we isolated embryos at E2.5 (8-16 cell stage). Of these (n=63), 30% were wild type, 67% were heterozygous mutants and 3% were homozygous mutants. The ratio of wild-type to heterozygous embryos was ∼1:2, i.e. was similar to the genotype distribution observed in weaning-age mice.
A crucial step in the early development of the mammalian embryo is a process termed `compaction', which occurs at the eight-cell stage in the mouse. During this process, individual embryonic cells (blastomeres) activate cell-cell adhesion proteins, in particular E-cadherin and associated catenins (Blaschuk et al., 1990; Larue et al., 1994; Ohsugi et al., 1996; Ohsugi et al., 1997; Pauken and Capco, 1999; Riethmacher et al., 1995; Torres et al., 1997). As a consequence, blastomeres begin to polarize and assemble the first cell junctions (reviewed by Fleming et al., 2000; Gallicano, 2001). This process turns the loose aggregate of blastomeres into a compact sphere.
In the experiment described above, we had analyzed a mixture of pre- and post-compaction embryos. In order to determine whether loss of Dsc3 affects compaction, we isolated embryos at the eight-cell stage just prior to compaction and determined their genotype. The following genotype distribution was observed (n=33): 24% Dsc3+/+, 67% Dsc3+/– and 9% Dsc3–/–. These results suggest that most homozygous mutants are dead before compaction is completed.
The experiments summarized so far suggested that Dsc3–/– embryos die within the first 2 days of development. We therefore isolated embryos from Dsc3+/– intercrosses at the two-cell stage and followed their development in vitro. Individual embryos were genotyped when they reached the blastocysts stage or when they stopped developing. Embryos were defined as abnormal when they ceased to show cell divisions for 18 hours, or when they appeared to degenerate. Fig. 4 shows examples of embryos at different stages of development. Most Dsc3–/– embryos appeared to disintegrate within 2.5 days of fertilization.
Finally, we addressed the issue of whether the Dsc3-null mutation could affect the development of oocytes and sperms. We crossed Dsc3+/– males and Dsc3+/– females with wild-type mice. Normal development of Dsc3+/– oocytes and sperms should lead to a 1:1 ratio of heterozygous and wild-type embryos in these crosses. The genotype distribution of blastocysts from these crosses was indeed very close to the expected value. Thirty-two embryos derived from a cross between a Dsc3+/+ female and a Dsc3+/– male were analyzed, 15 of which were heterozygous mutants. Six out of 14 embryos from a Dsc3+/– female and a Dsc3+/– male cross were heterozygous. These results indicate that mutant oocytes and sperms are fully capable of contributing to zygote formation.
Deletion of the PGK-Neo cassette from the genome of mutant mice
The mouse lines described above carry a neomycin-resistance cassette (which is flanked by LoxP sites, see Fig. 1) in the Dsc3-null allele. To ensure that the presence of this cassette did not contribute to the lethal phenotype, we crossed heterozygous mutant mice from line 1 with transgenic mice expressing Cre-recombinase under the control of a CMV promoter. Mice carrying both the Dsc3-null allele and the transgene were then backcrossed to C57Bl mice to separate transgene and Dsc3-null allele. Southern blots and PCR analysis confirmed that we established a line in which the PGK-Neo cassette had been deleted from the genome (data not shown). Thirty-eight pups from Dsc3+/–Δneo intercrosses were genotyped shortly after birth, none of these mice was a homozygous mutant (11 wild type and 27 heterozygous mutants were obtained). Next, we isolated blastocyst-stage embryos from heterozygous intercrosses. Again, we failed to identify Dsc3–/– embryos (eight wild type and 20 heterozygous mutants were genotyped). This demonstrated that the embryonic lethal phenotype occurred independently of the PGK-Neo cassette.
Lethality of the Dsc3-null mutation
Given the unexpected early lethal phenotype, it is necessary to demonstrate that we have generated a true Dsc3-null mutation. The Dsc3 phenotype was observed in each of the two mouse lines that were generated from independent ES cell clones. As judged by Southern blotting, both ES clones had undergone homologous recombination at the Dsc3 locus and did not contain additional copies of the targeting vector integrated randomly into their genome. Unfortunately, we could not verify the loss of Dsc3 expression in null mice, as we did not obtain a sufficient number of mutants to conduct this analysis. However, we did analyze Dsc3+/– mice: The QPCR results demonstrate that these mice express ∼50% of the Dsc3 wild-type levels. Furthermore, we only detected two specific bands, corresponding to Dsc3a and Dsc3b, in western blots of Dsc3+/– samples. As our Dsc3 antibody was raised against the C-terminal cytoplasmic domain, it is unlikely that an N-terminally truncated protein is synthesized in our mutant mice.
All Dsc genes are clustered on mouse chromosome 18 (Cheng et al., 2005). The 3′ end of the Dsc2 gene is ∼30 kb upstream of the Dsc3 gene. To exclude the possibility that our targeted mutation had affected the Dsc2 locus, we analyzed the expression of this gene in Dsc3+/– mice. Most organs (e.g. tongue and kidney) showed similar Dsc2 expression levels in wild-type and Dsc3+/– mice, confirming that targeting of the Dsc3 locus did not affect the Dsc2 gene. In summary, all of these observations are consistent with a lethal null phenotype that is linked exclusively to a null mutation in the Dsc3 locus.
Dsc3 and desmosomes in the blastocyst
Morphologically `mature' desmosomes, as determined by electron microscopy, are first observed in the trophectoderm (TE) layer of blastocysts at E3.5 (Collins et al., 1995; Ducibella et al., 1975; Fleming et al., 1991; Fleming et al., 1993; Fleming et al., 1994; Jackson et al., 1980) (for a review, see Cheng et al., 2005). At this stage of development, the mouse embryo consists of two cell types: the inner cell mass (ICM) that will later develop into the embryo (embryo proper); and the TE, which forms the outer cell layer of the blastocyst-stage embryo. The TE is the first polarized epithelium that develops in the embryo. These cells form an apical junctional complex that contains desmosomes and adherens junctions. It has been speculated that desmosomes in the TE are required to maintain the mechanical integrity of blastocysts. Two basic mechanical forces could affect this cell layer: (1) pressure from the blastocoel, a fluid-filled cavity that is formed by TE cells; and (2) the mechanical stress associated with blastocyst hatching through the zona pellucida (the protein shell that surrounds the pre-implantation embryo) in preparation for implantation. Nevertheless, two desmosomal null mutations that have been published so far, the Dsg2 and the Dp-null mutations (Eshkind et al., 2002; Gallicano et al., 1998), allow development beyond the blastula stage and do not seem to affect the ability of mutant embryos to hatch from the zona pellucida. It appears that classical desmosomes are not required for these processes, and that other adhesion proteins in these mutants are sufficient to maintain structural integrity.
Our antibody staining experiments suggest that Dsc3 only partially overlaps with desmosomal marker proteins such as Dsg2 and Dp, i.e. Dsc3 is not restricted to classical desmosomes. The Dsc3 distribution along the plasma membrane of TE cells is similar to that of E-cadherin and β-catenin (Fig. 3A). This is in clear contrast to the distribution of this protein in adult stratified epithelia (see Fig. S2 in the supplementary material). It remains to be seen whether Dsc3 functions as a classical adhesion molecule in TE cells. As most Dsc3–/– embryos (>90%) die before the blastula stage is reached, a conditional null mutation will be required to ablate Dsc3 expression specifically at the blastocyst stage.
Our observation that Dsc3 is present in blastocysts appears to contradict previously published results. Collins and co-workers have reported that Dsc3 mRNA is not present in pre-implantation embryos (Collins et al., 1995; Fleming et al., 2000). Their conclusions were based on RT-PCR experiments using primers that were designed to amplify both Dsc2 and Dsc3 cDNA. Dsc3-specific primers were not used. Twelve cloned PCR products were sequenced, none of which represented the Dsc3 cDNA. Our own experiments suggest that the Dsc3 mRNA is of low abundance compared with the Dsc2 mRNA, which could easily explain this discrepancy.
Dsc3 in pre-compaction stage embryos and a possible cause of embryonic death in Dsc3-null mutants
It has been shown that E-cadherin plays a crucial role in early cleavage-stage embryos, even before compaction occurs. Using an elegant genetic system, de Vries and colleagues recently generated mouse embryos that lacked maternal E-cadherin (de Vries et al., 2004). Mutant embryos showed weakened blastomere adhesion and disaggregated when the zona pellucida was removed. However, when the zona pellucida was intact, mutant embryos persisted until the embryonic E-cadherin gene was activated. Consequently, these embryos could undergo pre-implantation development.
It appears that adhesion proteins other than E-cadherin can maintain cell-cell adhesion, and consequently mechanical integrity of the embryo, as long as the zona pellucida is intact. Given the timing of embryonic death in our Dsc3–/– mutants, and the fact that we often observed disaggregated embryos at around E2.5 (examples in Fig. 4A), it is tempting to speculate that Dsc3 might be required for blastomere adhesion of early cleavage-stage embryos.
Alternatively, loss of Dsc3 could increase the cytoplasmic pool of plakoglobin, which has been shown to bind Dsc in other tissues (Cheng and Koch, 2004). Fleming and colleagues found that plakoglobin is the first desmosomal protein synthesized in the mouse pre-implantation embryo, being present at or before the eight-cell stage (Fleming et al., 1994). Our own RT-PCR results confirm this finding (Fig. 2E). Plakoglobin has been shown to participate in nuclear signaling, mainly through the Wnt pathway (Hu et al., 2003; Shtutman et al., 2002; Simcha et al., 1998; Zhurinsky et al., 2000). It is conceivable that an increase of the plakoglobin pool could affect transcription in early cleavage-stage embryos. However, it has to be noted that currently there is no evidence that the Wnt pathway regulates gene transcription at this stage of development (Kemler et al., 2004).
In summary, this is the first report that demonstrates the requirement of a desmocollin for pre-implantation development. Our antibody-staining experiments also suggest that Dsc3 might function independently of classical desmosomes in early cleavage-stage mammalian embryos. Further experiments will be necessary to define the ultrastructural localization and function of Dsc3 in blastomeres and embryonic tissues, such as the trophectoderm layer of blastocysts.
Materials and Methods
Animal experiments were conducted in accordance with all local and federal guidelines and were approved by the `Institutional Animal Care and Use Committee' (IACUC) of Baylor College of Medicine.
Generation of Dsc3 mutants
Genomic DNA encompassing the Dsc3 gene locus was isolated from a λ-library (129/SvEV Tac; Stratagene, catalog number 946312). This library is isogenic with the ES cell line used in our experiments (see below).
In the Dsc3 targeting vector (Fig. 1), 1.14 kb of the genomic DNA sequence immediately upstream of the start codon as well as the first coding exon and 2.65 kb of intron 1 were replaced with a neomycin resistance cassette (PGK-Neo). The total size of the genomic DNA deletion is 4.04 kb. The PGK-Neo cassette is flanked by 3.24 kb (5′ arm; SacI fragment) and 2.94 kb (3′ arm; NsiI fragment) of DNA, respectively, from the Dsc3 gene locus. The linearized targeting construct was electroporated into W4/129S6 ES cells (Taconic) and recombinant clones were selected following a positive/negative selection protocol (Mansour et al., 1988). Stem cell culture and selection of targeted clones were carried out as previously described (Koch et al., 1997). We used a 5′ probe (EcoRV/Nci fragment; Fig. 1A) and a 3′ probe (DraI/EcoRI fragment; Fig. 1A), respectively, to identify clones that had undergone homologous recombination. Southern blots with a neomycin probe verified that the ES clones had undergone a single recombination event and did not contain additional copies of the PGK-Neo cassette. Two Dsc3+/– ES clones were injected into C57/Bl 6 blastocysts (Genetically Engineered Mouse Core, Baylor College of Medicine). Chimeric males from these injections were mated to C57Bl/6 females to establish Dsc3 lines 1 and 2. Most of the experiments described below were carried out with line 1 mice. However, to confirm that the observed phenotype of Dsc3–/– mice was not due to the presence of PGK-Neo, we deleted this cassette from the mouse genome; heterozygous line 1 mutants were crossed with CMV-Cre transgenic mice (provided by Dennis Roop, Baylor College of Medicine) to remove PGK-Neo via Cre-mediated recombination. Dsc3+/–Δneo/CMV-Cre mice were then back crossed to C57Bl mice to separate the CMV-Cre transgene from the Dsc3+/–Δneo allele. As expected, Dsc3–/–Δneo embryos showed the same embryonal lethal phenotype as the original lines.
Preparation and genotyping of embryos
Mouse embryos at different stages of pre- and post-implantation development were prepared and genotyped essentially following established protocols (Nagy et al., 2003). For the analysis of Dsc3+/– intercross embryos, super-ovulated Dsc3+/– females (Nagy et al., 2003) were mated to Dsc3+/– males and checked for vaginal plug the next morning (defined as E0.5). Pre-implantation embryos were isolated at E1.5, E2.5 or E3.5 following standard procedures (Nagy et al., 2003). In some experiments, two-cell stage embryos (E1.5) were isolated and then cultured in M16 medium (Sigma) until they reached the blastocysts stage. The following Dsc3 primers were used for PCR genotyping: primer I (GCAACTGATTGAGTTTGGGGAA), primer II (ATAATTGTCCCAAGGGCTGG) and primer III (GCATGCTCCAGACTGCCTTGGGAA). Wild-type (0.537 kb) and mutant (0.46 kb) PCR products were separated in agarose gels, blotted to NyTran Supercharge membranes (Schleicher & Schuell) and then hybridized to radioactively labeled oligonucleotide H (GCGAGGATAGATTTGAAGTGCTGTCTCTGAGACC) to confirm the identity of the PCR products. Dsc3+/–Δneo mice were genotyped with a combination of primers H, II and F-R (AAACTAGCAGTCTTCCCAAGTCTG). The PCR fragment sizes were 437 bp (wild type) and 526 bp (mutant allele).
Antibodies, immunofluorescence and immunohistochemistry
The generation of polyclonal guinea pig antibodies against COOH-terminal (cytoplasmic) sequences of Dsc2 (gp2295) and Dsc3 (gp2280; see Fig. S2 in the supplementary material) has been described previously (Cheng et al., 2004). The following antibodies were used: DG3.10 [Research Diagnostics, Dsg1/Dsg2 antibody (Koch et al., 1990; Schmelz et al., 1986)], plakoglobin (PG5.1; Research Diagnostics), desmoplakin 1+2 (Research Diagnostics), TROMA I (Brulet et al., 1980; Kemler et al., 1981), normal guinea pig IgG (Fitzgerald) and biotinylated anti guinea pig IgG (H+L) (Vector Laboratories). Fluorochrome-labeled secondary antibodies (Alexa Fluor 488, Alexa Fluor 594) were purchased from Molecular Probe. Immunofluorescence labeling and immunohistochemistry, except for pre-implantation embryos, were carried out essentially as described (Koch et al., 1997; Koch et al., 1998).
Blastocysts were first treated with acidic Tyrode's solution (Sigma) to remove the zona pellucida (Nagy et al., 2003). After a 1 hour recovery in M16 medium at 37°C (5% CO2, humid atmosphere), the embryos were fixed for 2 minutes in ice-cold ethanol, transferred to a glass cover slide and air dried. After re-hydration in phosphate-buffered saline (PBS), the specimens were incubated for 30 minutes in blocking solution (10% BSA, 2.5% normal goat serum in PBS). The first antibody was diluted in blocking buffer and then incubated with the specimens for 1 hour at room temperature. After three washes in 0.1% Tween 20/PBS (10 minutes each), and one 5-minute wash in PBS, the samples were incubated with secondary antibodies in blocking buffer for 1 hour at room temperature. After three final washes in PBS (10 minutes each), the samples were mounted with Fluoromount G (Southern Biotechnology). Two- to eight-cell stage embryos were stained by immunohistochemistry: after fixation for 20 minutes in 0.5% PBS-buffered formalin and 10 minutes post-fixation in ethanol at –20°C, embryos were permeabilized with 0.25% Triton X100 for 10 minutes at room temperature. Next, the embryos were treated with 1% H2O2 for 10 minutes, incubated for 30 minutes in blocking buffer (10% fetal calf serum/10% normal goat serum/PBS), washed briefly in PBS, and then incubated either with gp2280 (Dsc3 antibody) or normal guinea pig IgG in blocking buffer at 4°C overnight. After three washes in 0.1% Tween 20/PBS (10 minutes each), the specimens were incubated with biotinylated guinea pig IgG for 45 minutes and then washed twice with PBS. Antibody binding was detected with reagents from the Vectastain ABC kit (Vector Laboratories).
A Nikon Eclipse E600 microscope was used in conjunction with the MetaVue v6.1r5 imaging software (Universal Imaging) to document these experiments.
We followed the protocols described previously (Cheng et al., 2004), with the following addition: for the blastocyst analysis, we used total cell extracts from 120 embryos per lane. Signals were detected with the `Super Signal West Femto Maximum Sensitivity Substrate' (Pierce).
Total RNA isolation was done with the Cell-to-cDNA II kit (Ambion). The following primer pairs were used in conjunction with the OneStep RT-PCR system (Qiagen): Dsc2 (Dsc2-8, TCTGGGCAGACAGCTTTCAC; Dsc2-9, CGTTCACCTCTTTGCACATAC), Dsc3 (DSC3a-RT-F ACAGAGTGTGTTCTGCATGGATTCACC; DSC3a-RT-2R CTGATTACACACATGCAATTTCT), Dsg2 (DSG2-RT-1F, CCAGGGGTCACCAAGCATAGCAC; DSG2-RT-1R, CATCCCCCCAATAAATCACAGAGCAT), Dsg3 (Dsg3-1F, GCATGATACCAGCACATCAG; Dsg3-1R, GAGGCCAATGAGGCAATAG), pg (PG-1F, CAGATTACCGTAAGCGAG; PG-1R, CAGCATGTGGTCTGCAGTGC), dp (DP-1F, CTCTGGAGGAGTCAAGCCC; DP-1R, CCTTTTACCCCTTCAAAGCC) and GAPDH (Ambion). All RT-PCR products were confirmed by DNA sequence analysis.
Quantitative real-time PCR (QPCR)
Total RNA was extracted with the RNeasy Mini Kit (Qiagen). An additional DNase digestion of column-bound RNA was used to eliminate traces of DNA. cDNA syntheses were carried out with the `High-Capacity cDNA Archive Kit' from Applied Biosystems. Dsc3 cDNA was amplified using Applied Biosystem's Assay-on-Demand mix (Mm00492270_m1). QPCR with 18S cDNA primers (Hs99999901_s1) was used as an internal control. QPCR reactions were carried out with the `DNA engine Opticon 2' from MJ Research.
We thank Kusal Mihindukulasuriya for technical support, and Dennis R. Roop (Baylor College of Medicine) and Werner W. Franke (German Cancer Research Center, Heidelberg, Germany) for providing antibodies. Special thanks to Francesco DeMayo and Jie Wang from the `Genetically Engineered Mouse Core' (Baylor College of Medicine) for providing mouse blastocysts. We are also grateful to Richard Behringer (MD Anderson Cancer Center, Houston, TX) for advice regarding mouse embryo cultures. This work has been supported by a grant (AR50439) from the National Institutes of Health (NIH/NIAMS) awarded to P.J.K.