SEARCH

SEARCH BY CITATION

Keywords:

  • connexin;
  • gap junction;
  • intercellular communication;
  • differentiation;
  • placenta;
  • trophoblast;
  • skin;
  • epidermis

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The overlapping expression of gap junctional connexins in tissues has indicated that the channels may compensate for each other. During development, Cx31 and Cx43 are coexpressed in preimplantation embryos, in the spongiotrophoblast of the placenta and in the epidermis. This study shows that Cx31/Cx43 double-deficient mice exhibit the known phenotypes of the single-knockout strains but no combined effects. Thus, Cx43, coexpressed with Cx31 at midgestation in the spongiotrophoblast of the placenta, cannot be responsible for a partial rescue of the lethal Cx31 knockout phenotype, as assumed before (Plum et al. [ 2001] Dev Biol 231:334–337). It follows that both connexins have unique functions in placental development. Despite an altered expression of other epidermal connexin mRNAs, epidermal differentiation and physiology was unaltered by the absence of Cx31 and Cx43. Therefore, in epidermal and preimplantation development, gap junctional communication can probably be compensated by other isoforms coexpressed with Cx31 and Cx43. Developmental Dynamics 233:853–863, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

In recent years, it has been proven by gene targeting in mice that the different gap junctional connexins (Cx) are major players in organ development and function. Furthermore, numerous mutations in the human connexin genes have shown their importance for tissue differentiation and maintenance of organ function (for reviews, see White and Paul, 1999; Willecke et al., 2002; Sohl and Willecke, 2004). The intercellular communication mediated by the gap junction channels enables the cells to form a functional syncytium for transfer of ions, metabolites, and second messengers up to 1 kDa. Meanwhile, 20 different connexins have been identified in the mouse genome, which lead to a minimum of 20 different channels, not including the possibility of forming heteromeric (composed of different connexins) and heterotypic (composed of different connexons) channels. The variety of channels, which are characterized by different electrophysiological properties (Bukausas and Verselis, 2004), and the overlapping connexin expression pattern in different tissues raised the question if all these channels have unique function or can compensate for each other. Targeted ablation of single connexin genes revealed either specific functions or no obvious phenotype in different tissues. The latter indicates either redundancy or compensatory mechanisms for at least some channels. To address this question, several double-knockout mice and functional replacement of a channel by another connexin (knockin) have been generated (for review, see White, 2003). For example functional replacement of Cx43 with Cx40 or Cx32 rescued the malformation of the pulmonary outflow tract found in Cx43 knockout mice (Reaume et al., 1995), but the development of normal spermatogenesis was impaired in both knockins. Only heterozygous Cx43KICx32 and not the Cx43KICx40 mice were disabled to nurse their pups sufficiently (Plum et al., 2000). These results demonstrated that some of the gap junction channels investigated display both unique and shared functions.

To test synergistic or combined effects between Cx43 and Cx32 during embryonic development, we have investigated organs coexpressing both connexins in a double-mutant mouse (Houghton et al., 1999). The missing Cx32 channel did not alter the Cx43 phenotype. Thus, these connexins seem to act independently from each other and do not serve as redundant rescue channels.

In previous publications, it has been shown (Plum et al., 2001) that Cx31 but not Cx43 (Reaume et al., 1995) is essential for placental development. Cx31 expression is found in different tissues such as in the epidermal layers of the skin, in the testis (Elfgang et al., 1995), as well as in the inner ear (Xia et al., 2000). In skin, multiple other connexins are expressed such as Cx26, Cx31.1, Cx37, Cx40, and Cx45, but Cx43 is coexpressed with Cx31 in the basal and spinous layer (Kretz et al., 2003). Furthermore, Cx31, in addition to Cx43, is one of the first connexins expressed in preimplantation embryos. After implantation, Cx31 gets restricted to the extraembryonic membranes, predominantly the ectoplacental cone, which later on give rise to the placenta (Reuss et al., 1996; Dahl et al., 1996), whereas Cx43 expression characterizes the embryo proper. In the mouse placenta, Cx31 remains expressed in the spongiotrophoblast, followed by an induction of Cx43 in the same trophoblast population at day 10.5 post coitum (pc). Like the ablation of the Cx43 channel (Reaume et al., 1995), the absence of the Cx31 channel does not influence the development of the preimplantation embryos, but Cx31 is involved in placental development. Without the Cx31 channel, the trophoblast cells showed a reduced proliferation and an enhanced differentiation along the trophoblast cell lineage, resulting in a placental dysmorphogenesis at day 9.5 pc with nearly no labyrinth and spongiotrophoblast but abundant giant cells. As a consequence, around 60% of the embryos started to die from day 10.5 pc onward, whereas the placentas of the others showed a morphological recovery leading to the delivery of normal pubs (Plum et al., 2001). The reason for this placental recovery is not yet clear. Because Cx43 is induced from day 10 pc onward in the spongiotrophoblast and revealed an increased expression in the Cx31-deficient mice (Plum et al., 2001), we assume that the expression of the Cx43 channel is responsible for the placental rescue. Within the same study, no obvious morphological defects in skin or inner ear were detected and hearing was not impaired, as revealed by audiometry (Plum et al., 2001).

To examine the hypothesis that Cx43 is responsible for the recovery of placental development, we generated mice missing the Cx31 as well as the Cx43 channel. Our study demonstrated that Cx31/Cx43 double-knockout mice are born with the known phenotypes of the Cx31 and the Cx43 knockout mouse but no additional defects. Thus, Cx43 is not able to replace Cx31 in placental development. We conclude that the Cx31 and Cx43 channels have unique function in placental development. In skin development, there is some evidence that the loss of Cx31 and Cx43 can be compensated by the increased expression of other connexins.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Generation of Mice and Genotype Distribution

Due to the transient placental dysmorphogenesis of the Cx31-deficient mouse, which leads to intrauterine death of 60% of the embryos (Plum et al., 2001), the number of Cx31−/−/Cx43+/− offspring was strongly decreased in the crossed strain. To generate sufficient numbers of Cx31/Cx43 double-deficient fetuses, mating of Cx31−/−/Cx43+/− males to Cx31+/−/Cx43+/− females was performed. A total number of 35 pregnant Cx31+/−/Cx43+/− females fertilized by 13 different Cx31−/−/Cx43+/− males were dissected. In addition, three pregnant Cx31−/−/Cx43+/− females fertilized by two different Cx31+/−/Cx43+/− males were used. The genotype of each fetus was analyzed by polymerase chain reaction (PCR; Fig. 1). Among the six different genotype combinations, viable Cx31/Cx43 double-deficient animals could be detected in utero at developmental stages between days 11.5 and 19.5 pc (Table 1). Cx31/Cx43 double-deficient mice were born and showed the early postnatal lethality of the Cx43-deficient mouse (Reaume et al., 1995). The absence of Cx31 and Cx43 has been confirmed by Northern blotting (Fig. 4A) and by immunofluorescence of embryonic skin (17.5 dpc, data not shown). For statistical analysis of the genotype distribution (Table 2A), the embryos between day 14.5 and 19.5 pc were used, because of the embryonic death of the Cx31-deficient mice between day 10.5 pc and 13.5 pc of development and the early postnatal lethality of the Cx43 knockout mouse. The results indicated a reduction of animals with the Cx31−/− background and an increase in number of animals with a Cx31+/− background, which has already been shown in the Cx31-deficient mouse (Plum et al., 2001). Similar to the results obtained by Plum et al. (2001) a reduction of 31.4% of Cx31−/− embryos, compared with the expected Mendelian ratios, could be detected after day 13.5 pc (Fig. 2B). The 84 resorbed embryos revealed that the lethal transient placental dysmorphogenesis of the Cx31 knockout mouse is responsible for the lack of the Cx31−/− embryos in the Cx31/Cx43 double-knockout strain. The Cx43 gene mutation did not influence the survival of the embryos (Table 2C). Each Cx43 genotype was detected at the expected Mendelian ratio. Thus, the additional inactivation of Cx43 does not further influence the survival of the Cx31 knockout embryos.

thumbnail image

Figure 1. Polymerase chain reaction (PCR) genotyping of the fetuses from one litter showing the six different genotypes. The wild-type allele of Cx31 is indicated by a 0.4-kb signal, whereas the mutated allele leads to a 0.6-kb signal. The Cx43 wild-type allele is represented by a 0.9-kb fragment and the mutated allele by a 0.7-kb fragment. Mice 1, 2, and 3 were Cx31+/−, and mice 4, 5, and 6 were Cx31−/−. In addition, mice 1 and 4 were Cx43+/+, mice 2 and 5 Cx43+/−, and mice 3 and 6 Cx43−/−. M, marker lane; mut, mutated allele; WT, wild-type allele.

Download figure to PowerPoint

Table 1. Genotypes Detected After Mating of Cx31−/−/Cx43+/− With Cx31+/−/Cx43+/− Parentsa
Stage (dpc)NlittersNanimalsGenotypes
Cx31+/−Cx31−/−
Cx43+/+Cx43+/−Cx43−/−Cx43+/+Cx43+/−Cx43−/−
NNNNNN
  • a

    dpc, days post coitum.

11.517230002
12.515201011
13.516120021
14.517312001
15.5214062231
16.515130100
17.525141264520142115
18.55367170264
19.516013011
thumbnail image

Figure 4. Expression of connexins in the fetal skin. A: Northern blot analysis of total RNA purified from the back skin of fetuses at day 17.5 of pregnancy. Total RNA from skin, liver and Hel-37 keratinocytes was used as a positive control for the connexins. The blots were hybridized with specific probes as indicated. B: Northern signals were analyzed by densitometry and referred to the particular β-actin signal. Mean value and standard deviation were calculated for each of the six offspring's genotypes.

Download figure to PowerPoint

Table 2. Mendelian Ratios of the Genotypes, Detected From 14.5 to 19.5 dpc, After Mating of Cx31−/−/Cx43+/− with Cx31+/−/Cx43+/− Parents (A)a
  • a

    The Mendelian ratios were also separately calculated for each the Cx31 (B) and the Cx43 genotype (C). dpc, days post coitum.

inline image
thumbnail image

Figure 2. Placental and fetal weight at day 17.5 post coitum (pc). A,B: Weight of placentas (A) and embryos (B) after mating of Cx31−/−/Cx43+/− with Cx31+/−/Cx43+/− parents. Wet tissues were weighed after removal of excess fluid. A total of 141 animals from 25 litters were analyzed, with a minimum of 14 animals per time point as indicated in Table 1. Weights are shown as means ± SD. Asterisks indicate genotypes for which values differed significantly (P < 0.05) compared with the Cx31+/−/Cx43+/+ genotype.

Download figure to PowerPoint

Placental Morphology and Embryonic Weight

Statistical analysis of embryonic and placental weight was carried out at 17.5 dpc. The results showed that placental weight is influenced only by the Cx31−/− but not by the Cx43 genotype (Fig. 2A). A Cx31-deficient placenta was around half of the size (53 ± 14 mg) compared with a Cx31 heterozygous placenta (98 ± 15 mg) at 17.5 dpc. The morphology of the Cx31/Cx43 double-knockout placenta was studied from days 13.5 to 17.5 pc compared with the placenta of Cx31+/−/Cx43+/+ and Cx31−/−/Cx43+/− littermates. No differences in morphology could be observed at theses stages. In detail, the Cx31/Cx43 double-knockout placenta showed the same reduction in size of the spongiotrophoblast as known from the Cx31 knockout mice (Fig. 3A–F). The number or location of trophoblast giant cells did not differ from the control placentas, and no differences in the labyrinth could be observed. Thus, the additional inactivation of the Cx43 gene did not alter the phenotype of the Cx31 knockout placenta. The weight of the embryos was dependent on their genotypes. The Cx31−/− embryos were, on average, 192 mg smaller compared with the Cx31+/− embryos. The additional Cx43 genotype did not have an influence on embryonic weight of the Cx31−/− embryos. Of interest, on the Cx31+/− background, the homozygous inactivation of Cx43 led to a statistical significant reduced weight of the embryos compared with the Cx43+/+ and Cx43+/− littermates (Fig. 2B).

thumbnail image

Figure 3. Morphology of the Cx31/Cx43 double-knockout placenta. Hematoxylin and eosin–stained paraffin sections at day 17.5 post coitum (pc). B,C,E,F: The Cx31−/−/Cx43−/− placenta (C,F) does not differ in development and placental architecture from a Cx31−/−/Cx43+/− placenta (B,E). The trophoblast populations (L, labyrinth; S, spongiotrophoblast; G, giant cells; D, decidua) were similarly developed in Cx31−/−/Cx43−/− and Cx31−/−/Cx43+/− mice. A,D: The strongly reduced spongiotrophoblast is characteristic for the Cx31-knockout genotype as indicated in comparison to a Cx31+/−/Cx43+/+ placenta. Additional inactivation of the Cx43 gene does not lead to a placental phenotype different from the Cx31−/− placenta. Scale bars = 1.57 mm in A–C, 200 μm in D–F.

Download figure to PowerPoint

Skin Development in Cx31/Cx43-Deficient Mice

For a better comparison, only back skin was investigated. Cx31 and Cx43 are expressed in the epidermis of the embryonic and adult skin and show a partial coexpression in the basal and spinous layer (Butterweck et al., 1994). The expression of connexins expressed in embryonic mouse skin (Cx26, 31, 31.1, 37, 40, 43, 45; reviewed by Richard, 2000) was analyzed at the mRNA level (Fig. 4). Homozygous inactivation of Cx31 did not change expression of Cx43. The amount of Cx43 itself was dependent on the alleles, Cx43+/+ showed stronger expression of transcripts than Cx43+/− animals. Vice versa, lack of Cx43 did not influence Cx31 expression in the Cx31 heterozygous animals. The lack of Cx31, however, led to an up-regulation of Cx37 and Cx40 transcripts. Both transcripts also showed a successive reduction of expression dependent on the Cx43 alleles, but this reduction occurred similar on the Cx31+/− and Cx31−/− background. Cx26 transcripts showed a slight increase in expression when the Cx31 gene was missing. Immunostaining of Cx26 protein revealed that Cx26 is expressed in the suprabasal layers, predominantly in the granular layer of the fetal epidermis. The location of Cx26 gap junctions was not changed by the inactivation of both Cx31 and Cx43 genes (Fig. 5A,B). Cx31.1 was down-regulated by inactivation of the Cx31 gene but did not show a dependence on the Cx43 alleles. Cx45 mRNA revealed a down-regulation in Cx43+/− and Cx43−/− animals but only on the Cx31+/− background. On the Cx31−/− background, there is a nearly constant amount of Cx45 transcripts. Immunocytochemistry of Cx26, Cx31, Cx40, and Cx43 and was performed on skin sections of Cx31+/−/Cx43+/− and Cx31−/−/Cx43−/− and wild-type mice. No changes in expression of Cx26, Cx31, and Cx43 could be detected. Although Cx40 mRNA was up-regulated, the protein could only be detected in the endothelial cells of subepidermal vessels (Fig. 5C,D) but not in the epidermal layer using a specific antibody. There were no obvious alterations in vessel in morphology between Cx31/Cx43 double-knockout and control mice.

thumbnail image

Figure 5. Immunolabeling of Cx26 and Cx40. A–D: Sections of control embryos (A,C) and Cx31−/−/Cx43−/−-deficient mice (B,D). A,B: Expression of Cx26 was unaltered in double-deficient mice. C,D: Cx40 marks only the endothelium of subepidermal vessels (arrows) but not the epidermal layer. Scale bars = 25 μm in A,B, 80 μm in C,D.

Download figure to PowerPoint

Semithin section of embryonic back skin at day 17.5 pc (Fig. 6) and 18.5 pc (data not shown) did not reveal obvious alterations in Cx31/Cx43 double-knockout mice. The skin showed normal stratification and equal differentiation of the epidermal cell layers. The number of epidermal cell layers and the process of cornification were unaffected in all genotypes indicated by the same formation of keratohyalin granules.

thumbnail image

Figure 6. Semithin sections of mouse back skin. A,B,C: Comparison of Cx31−/−/Cx43−/− (C) with Cx31−/−/Cx43+/+ (B) and Cx31+/−/Cx43+/+ (A) animals showed a normal differentiation in the skin of the Cx31/Cx43 double-knockout mice. Neither inactivation of the Cx31 gene alone or both the Cx31 and Cx43 genes led to a changed organization of the keratinocytes in the epidermal layer. Scale bar = 25 μm.

Download figure to PowerPoint

To assess whether the ablation of Cx31 and Cx43 caused an altered epidermal differentiation, the expression of inducible and constitutive keratins was studied. K6, which is one of the most-sensitive indicators of altered epidermal differentiation after tissue injury, changes in proliferation and oxidative stress (Magin, 2004), was not induced in interfollicular epidermis (Fig 7E,F). It remained restricted to the hair follicle. Furthermore, the expression of K14, typical of basal keratinocytes, and of K1 and K10, which form the keratin cytoskeleton in suprabasal keratinocytes, was unchanged in both genotypes of mice (Fig. 7A–D). K17, which represents a constitutive and inducible keratin (Tong and Coulombe, 2004), was expressed in all basal keratinocytes but did not change its expression in double-deficient mice (Fig. 7G,H). The normal state of Cx31 plus Cx43 double-deficient epidermis was also reflected by the unaltered expression of the integrin α6, which was found along basolateral aspects of basal keratinocytes (Fig. 7I,J). Finally, the keratin-associated protein filaggrin and the cornified envelope constituent protein loricrin were not altered in their expression and distribution (Fig. K–N).

thumbnail image

Figure 7. Expression of integrins, epidermal keratins, and cornified envelope proteins. A,C,E,G,I,K,M: Sections of control embryos. B,D,F,H,J,L,N: Corresponding sections of double-deficient embryos. Expression of all proteins examined was unaltered in Cx31−/−/Cx43−/−-deficient mice. E,F: Note that K6, typical of altered differentiation, was absent from interfollicular epidermis; labeling of the cornified envelope in was due to nonspecific cross-reactivity of secondary antibody. Scale bar = 100 μm.

Download figure to PowerPoint

To analyze whether the epidermal barrier was formed normally in the Cx31/Cx43 double-knockout mice a dye penetration assay was performed. On day 17.5 pc, the epidermal barrier of most embryos was closed completely, but sometimes embryos of each genotype were detected that showed uniform dye penetration into the skin. As the epidermal barrier in mice is formed between days 16 and 17 pc (Hardman et al., 1998) also 18.5 dpc was examined. At this time, the epidermal barrier of all embryos was closed. Furthermore, to test whether proliferation properties of the keratinocytes in the fetal epidermis are affected by inactivation of Cx31 and Cx43, a bromodeoxyuridine (BrdU) labeling assay was performed. The number of labeled cells showed that the embryonic skin at day 17.5 pc revealed no differences in BrdU incorporation rates between the offspring genotypes (data not shown). The lack of Cx31 and Cx43 does not change proliferation properties in the fetal skin.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Our results show that the Cx31/Cx43 double-deficient mouse strain exhibits the known phenotypes of the single-knockout strains, but no combined changes in development and differentiation of the placenta and the skin, where both connexins are expressed, could be detected. Thus, Cx43 cannot serve for the recovery of the Cx31 knockout placenta. We conclude that Cx31 and Cx43 gap junctions have specific functions in skin and placenta development and function. The study, however, supports the notion that, during epidermal differentiation, the function of Cx31 and Cx43 can be replaced by other connexin isoforms.

The Cx31 knockout mouse shows a transient placental dysmorphogenesis, which leads to the death of 60% of the embryos between day 10.5 and 13.5 pc (Plum et al., 2001). The lack of Cx31 leads to a failure in trophoblast differentiation, resulting in a barely developed placental labyrinth and spongiotrophoblast but an increased number of trophoblast giant cells. Ablation of Cx43 obvious does not further alter placental development. The rescue of surviving Cx31 knockout embryos is due to placental recovery resulting in a normal placenta at birth. The assumption that the lack of a specific connexin can be partly compensated by another isoform expressed in the same cell or tissue has been widely discussed. It has been speculated by Plum et al. (2001) that the rescue starting at day 10.5 pc is due to the induction of Cx43 in the spongiotrophoblast.

Our study disproved this hypothesis by the generation of the Cx31/Cx43 double-knockout mice, because viable Cx31/Cx43 knockout animals have been born. These results demonstrate that Cx31 and Cx43, although coexpressed in the spongiotrophoblast of the placenta after day 10.5 pc, do not serve the same functions. Furthermore, the Mendelian ratios of the fetuses show that the three Cx43 genotypes are found in the expected ratios and independent from the Cx31 genotype. The lethality of the Cx31-deficient mice in our colony (31%) is not that strong as detected by Plum et al. (2001). This finding is probably due to the genetic background of the double-knockout animals generated by crossing both strains. Despite the lower lethality of the Cx31 knockout animals, placental development is more restricted by inactivation of the Cx31 gene in our strain. The size of a Cx31 knockout placenta was smaller than published by Plum et al. (2001). They were only half the size of a Cx31+/− placenta, whereas Plum et al. (2001) detected only a 20% reduction in weight. The distorted recovery of the Cx31 knockout placenta in our study can explain the reduced weight of the embryos at the end of pregnancy, which has not been detected before. Only a strong reduction of placental size leads to a decreased weight of the embryos (Kurz et al., 1999). Of interest, the Cx31+/−/Cx43−/− embryos were smaller compared with their Cx31+/−/Cx43+/− and Cx31+/−/Cx43+/+ littermates. This finding could be due to a function of Cx43 in placental physiology but also a function in embryonic development, as Cx43 is expressed in most tissues. It is known that Cx43 knockout newborns show anatomical alterations (Reaume et al., 1995), but to our knowledge, the fetal weight has not been analyzed to date. However, although Plum et al. (2001) did not observe an influence of the Cx31 genotype on the embryonic weight, it cannot be excluded that, in our study, also the Cx31+/− background had an influence on the reduced weight of the Cx31+/−/Cx43−/− embryos.

Cx31 and Cx43 are the first connexins that are most abundantly expressed during preimplantation development (Davies et al., 1996). Both connexins are coexpressed in the inner cell mass and the trophectoderm of the blastocyst. Again, it has been hypothesized that Cx31 and Cx43 could compensate for each other during preimplantation development, because there was no obvious effect in the Cx31 or Cx43 gene-deficient mouse strains during this phase (Reaume et al., 1995; Plum et al., 2001). The Cx31/Cx43 double-knockout mouse revealed that there is no necessity for at least one of these two connexins in preimplantation and embryonic mouse development. In the murine blastocyst, at least 9 connexin isoforms are expressed at the transcript level, and five of them (Cx31, Cx31.1, Cx43, Cx40, Cx45) were found to be assembled into gap junction channels (De Sousa et al., 1997; Houghton et al., 2002). Therefore, the other connexin isoforms could serve for gap junction intercellular communication during preimplantation development.

It still remains to be clarified why 40% of the Cx31 knockout placentas recover after day 10.5 pc. The Cx31/Cx43 knockout placenta indicates, that connexins are not essential to maintain proliferation of the spongiotrophoblast cells. Cx31 is potentially only essential for development of the early trophoblast cell lineage during a limited time period from implantation up to early placenta formation around day 9.5 pc. This conclusion is underlined by the observation that in rats, in contrast to mice, Cx31 expression is already turned off in the spongiotrophoblast after day 10.5 pc of development (Reuss et al., 1996). This finding leads to the assumption that factors different from gap junctions maintain the diploid trophoblast population in the placenta and obtain recovery after day 10.5 pc. The nature of these factors is unknown.

Recently, we established trophoblast stem (TS) cell lines from blastocysts of the Cx31 knockout mouse, which have the capacity to differentiate into all placental trophoblast populations (Kibschull et al., 2004). The lack of Cx31 in TS cells resulted in an enhanced differentiation into giant cells that has been indicated by the earlier induction of trophoblast marker genes such as Mash2, Tpbpa, and Pl-1 and a stop in proliferation only when differentiation of the TS cells was induced (Kibschull et al., 2004). In the presence of FGF4, which is essential to keep the undifferentiated state, no differences between Cx31-deficient TS cell lines and controls could be detected. These results indicated that the defect of the Cx31 knockout might affect very early differentiation steps, which result in a malformation of the very early placenta. The role of Cx43 in the spongiotrophoblast and in giant cell physiology remains unclear.

The results observed in this study, that connexins although coexpressed in different tissues cannot compensate for each other, are confirmed by a previous study on Cx32/Cx43 double-knockout mice (Houghton et al., 1999). No obvious effects were observed in the organs expressing both connexins during embryonic development.

In the mouse skin, Cx31 and Cx43 are coexpressed in the epidermal layer beside other connexins. Cx43 is predominantly expressed in the basal cells and shows weaker expression in the spinous cell layer. Cx31 is also expressed in basal layer but predominantly in the spinous and granular layer (Butterweck et al., 1994). Both the Cx31 and the Cx43 knockout mouse did not show alterations in skin development (Plum et al., 2001; Reaume et al., 1995). The morphology and the analysis of marker genes showed a normal state of the Cx31/Cx43 double-deficient epidermis. Furthermore, a dye penetration assay revealed no alterations in epidermal barrier formation and the lack of both connexins did not change proliferation of the keratinocytes. Obviously, the function of Cx31 and Cx43 in fetal skin development can be compensated by other connexin isoforms described in the mouse skin (reviewed by Richard, 2000). Most obvious were the up-regulation of Cx26, Cx37, and Cx40 and the down-regulation of Cx31.1 transcripts when Cx31 and Cx43 were missing. The up-regulation hints at a compensation of the missing connexins. Up-regulation of Cx26 in all epidermal cell layers occurs during hyperproliferative human skin diseases (Richard et al., 1997; Labarthe et al., 1998) and was recently found during wound healing in mice (Kretz et al., 2003). However, in contrast to these diseases and wound healing the localization of Cx26 gap junctions in the epidermis of the Cx31/Cx43 double-knockouts was not changed. In the embryonic rat skin, Cx31.1 and Cx37 have been detected in all epidermal layers beside the stratum corneum (Goliger et al., 1994). The expression pattern of Cx31.1 and Cx37 proteins during murine skin development has not yet been analyzed. Both connexins could replace the function of Cx43 and Cx31 in keratinocytes differentiation. Of interest, the changed expression of connexin transcripts in the skin of the Cx31/Cx43 double-knockout mouse was mainly dependent on the Cx31 alleles, whereas the Cx43 alleles had only weaker influence on the amount of transcripts.

Recently, a transgenic mouse lacking the C-terminal region of Cx43 was generated, showing a defective epidermal barrier in newborns, which was caused by a perturbed differentiation of keratinocytes (Maass et al., 2004). This strain demonstrates that a mutated Cx43 can change gap junction intercellular communication during skin development, leading to a malformation, whereas the complete loss of Cx43 can be compensated by other connexin isoforms as shown in this study.

The redundant and specific functions of Cx31 and Cx43 are also indicated by their channel properties. In HeLa cell transfectants, the Cx31 channels exhibited a more restricted permeability to tracer molecules than Cx43 channels: the dyes Lucifer yellow and 4,6-diamidino-2-phenylindole (DAPI) dihydrochloride passed both channels well but, in contrast to Cx43, Cx31 channels, were resistant to transfer of ethidium bromide and propidium iodide. Because the selectivity of these channels does not exactly correlate to their electrical conductance or to the molecular weight of the tracers, a connexin-specific regulation of the permeability is discussed (Elfgang et al., 1995).

In conclusion, the Cx31/Cx43 double-knockout mouse demonstrates functional redundancy of gap junction intercellular communication in the embryonic mouse skin and preimplantation development and specific functions of both connexins in placental development. Further simultaneous connexin gene inactivation or transgenic mice carrying mutated connexin genes can give insights into the function of these proteins in development and homeostasis.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Animals

Mice were maintained in a well-controlled pathogen-free environment with regulated light/dark cycles (12 hr/12 hr) in the animal facility of the University Hospital Duisburg-Essen, Germany. Mice had access to food and water ad libitum. All experiments were carried out in accordance to German laws for animal protection and with permission of the state. Cx31-deficient mice on a mixed genetic background of 75% C57BL/6 and 25% 129/Sv (Plum et al., 2001; kindly provided by Klaus Willecke, University of Bonn) were twice backcrossed with C57BL/6 mice, before Cx31+/− animals were intercrossed to generate Cx31−/− offspring. Mice heterozygous for the null allele of the Cx43 gene (Reaume et al, 1995; purchased from The Jackson Laboratory, Bar Harbor, ME) on a mixed genetic C57BL/6 and 129/Sv background were twice backcrossed with C57BL/6 mice. To generate the double-knockout strain, Cx31−/− were crossed with Cx43+/− mice. After backcrossing Cx31−/−/Cx43+/− and Cx31+/−/Cx43+/− animals were obtained. These mice were mated to generate offspring lacking both connexins, as indicated by PCR (Fig. 1). Because of the early postnatal lethality of Cx43-deficient mice (Reaume et al., 1995), fetuses were collected by caesarean section on day 17.5 pc for analysis. Mating was performed overnight, and noon of the day of plug detection was considered as day 0.5 pc.

PCR Genotyping

Genomic DNA was isolated from mouse tails as described by Houghton et al. (1999). The presence of the wild-type and the mutated Cx43 allele was simultaneously tested by PCR using the primers Cx43-extern (5′-gattggcagcttgatgttcaagc-3′), Cx43-intern (5′-tcaacgtggagatgcacctgaag-3′), and neo3 (5′-gatattgctgaagagcttggcgg-3′). The primers Cx43-intern and Cx43-extern detected the wild-type allele, yielding a 900-bp signal. Primer neo3 bound to the neoR gene of the mutated allele and gave rise to a 700-bp signal in combination with the Cx43-extern primer. The wild-type and the mutated Cx31 allele were detected using the primers GT1 (5′-ctggactctgacatgtgcacatac-3′), GT2 (5′-ctacatgcaggatgaccagcatag-3′), and GT3 (5′-ccacagatgaaacgccgagttaac-3′) in a single PCR reaction. The 600-bp signal of the primers GT1 and GT2 indicated the presence of the wild-type allele. Primer GT3 hybridized to the mutated Cx31 allele and formed a 450-bp signal in combination with primer GT1. PCR was carried out in a final volume of 50 μl containing: 1 × PCR buffer; 1.5 mM MgCl2; 0.2 mM each dATP, dCTP, dGTP, dTTP; 50 pmol of each primer, 2–4 μg of genomic DNA and 2.5 U of Taq polymerase (Genecraft, Muenster, Germany). After an initial denaturation step at 94°C for 2 min, the DNA was amplified for 40 cycles (94°C for 1 min, 64°C for 1 min, 72°C for 1 min), completed by a final extension at 72°C for 3 min.

Preparation of Fetuses and Tissue Processing for Morphological Analysis

Pregnant mice were killed by cervical dislocation. The uteri were excised and transferred into ice-cold phosphate-buffered saline (PBS). The fetuses and the placenta were removed, dried from excessive wetness, and weighed. Tails were taken for genotyping. For histological analysis, the placenta and the back skin of the embryo were fixed in 4% formaldehyde at 4°C overnight and routinely embedded in paraffin. Paraffin sections were stained with hematoxylin and eosin. For semithin sectioning, the skin was cut to 2 × 2 mm fragments and fixed in 2.5% glutaraldehyde/0.1 M cacodylate (Fluka, Germany) buffer at 4°C for 6 hr. After three washes in cacodylate buffer (30 min, room temperature), specimens were dehydrated in increasing ethanol concentrations (1 hr, room temperature) and embedded in Epon 812. The skin was sectioned at 80–90 nm thickness using an ultramicrotome (Ultracut, Reichert-Jung, Germany); stained with methylene blue, Azure II, toluidine blue, borax (each 1%; from Merck, Darmstadt, Germany) in aqua bidest; and mounted in Histomount (Shandon, Pittsburgh, PA).

Immunofluorescence

Fetal back skin was embedded in Tissue-Tek (Sakura, Zoeterwoude, Netherlands), frozen in liquid propane, and stored at −80°C. Indirect immunofluorescence was carried out on methanol/acetone-fixed cryostat sections as described (Winterhager et al., 1991) using antibodies directed to a C-terminal polypeptide of mouse connexins26, 31, 40, and 43 (Butterweck et al., 1994; dilution 1:100). An Alexa Fluor 488-conjugated donkey anti-rabbit secondary antibody (MoBiTech, Goettingen, Germany) was used for visualization. Antibodies against keratins K6, K14, K17, integrin, loricrin, and filaggrin were diluted as described before (Reichelt et al., 2001; Reichelt and Magin, 2002). For the simultaneous detection of K1 and K10, the monoclonal antibody KG8.60 was used. Secondary antibodies were used as described before (Reichelt et al., 2001). Sections were mounted in Mowiol (Sigma, Munich, Germany) and photos were taken with a confocal laserscan microscope (Zeiss, Germany).

Northern Blot Analysis

Total RNA from mouse back skin was isolated using Trizol reagent (Invitrogen). Northern blotting was performed using a standard protocol (Gabriel et al., 2001). cDNA probes for specific connexins were labeled with α-32P-dCTP using the random primed labeling system (Amersham). The densitometric analysis of the Northern blot signals was performed using a Gel Imager System (INTAS, Germany) and the GELSCAN Pro, version 4 software (BioSciTec, Germany). The expression of each signal was referred to the according β-actin signal. The following probes were used: Cx26, Cx31, Cx31.1, Cx43, and β-actin (Kibschull et al., 2004). The cDNA for each Cx37, Cx40, and Cx45 was amplified by reverse transcriptase-PCR from mouse trophoblast stem cell RNA, cloned into the pCRII-vector, and controlled by sequencing as described by Kibschull et al. (2004). The following primers were used: Cx37-forward (5′-ggctgcaccaacgtctgctatgac-3′) with Cx37-reverse (5′-ctgacacaccgacacagcaggtga-3′), according to data base accession no. NM008120; Cx40-forward (5′-aagaagccaactccagggaggagg-3′; X61675) with Cx40-reverse (5′-tgacttgccaaagcgctgtcggat-3′; NM008121); and Cx45-forward (5′-catacagtctatacacctggcgccagggaa-3′) with Cx45-reverse (5′- gtgagtctcgaattgtcccaaaccctaagt-3′); accession no. X63100.

BrdU Labeling of Embryonic Mouse Skin

Pregnant mice (17.5 dpc) were intraperitoneally injected with 1 ml BrdU solution (10 mg/ml BrdU in 0.9% NaCl; Sigma) per 100 g weight and killed 2 hr later. The back skin of the fetuses was fixed in 4% formaldehyde/PBS and embedded as described above. Paraffin sections were rehydrated, and antigen retrieval was performed using 10 mM sodium citrate buffer (pH 6.0) for 20 min at sub-boiling temperatures in the microwave oven. Sections were treated for 30 min with 2 N HCl at 37°C followed by incubation in 0.1% trypsin (Invitrogen, Karlsruhe, Germany) in PBS at 37°C. After washing in PBS, the antigen was detected by incubation with an anti-BrdU primary antibody (BU-33, 1:1,000, Sigma) for 2 hr. Immunohistological visualization of the anti-BrdU antibody was performed using the “Super Sensitive Detection Kit” and AEC-solution (both BioGenex, San Ramon, CA) according to the manufacturer's instructions. Sections were counterstained with hematoxylin and mounted in Mowiol (Sigma). For evaluation of the proliferation rate, three sections of five Cx31/Cx43 double-knockout fetuses compared with controls were used. Per section, a minimum of 1,000 cells of the basal epidermal layer were counted for BrdU incorporation.

Dye Penetration Assay

The assay was carried out as described by Hardman et al. (1998). Briefly, embryos at day 17.5 or 18.5 dpc were killed by ether inhalation and fixed in a methanol series (25, 50, 75, 100% methanol, 1 min each). After 1 min equilibration in PBS, animals were stained in 0.2% toluidine blue (Serva, Munich) in aqua bidest for 5 min and subsequently three times rinsed in 90% ethanol.

Statistics

Data are presented as means ± SD. A one-way analysis of variance test was used for comparison of placental or embryonic weights of the different offspring genotypes. Probability values of P < 0.05 were considered significant.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Natalie Knipp, Ulrike Laub, Georgia Rauter, Uschi Reuter, and Gabriele Sehn for excellent technical assistance. E.W. was supported by grants from the NIH and the Deutsche Forschungsgemeinschaft.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES