In this study, we have focused on the functional relationship between the wnt-1 signaling protein and the connexin43 (Cx43) gap junction. Both genes coexpress along the entire dorsal neural tube, as well as the midbrain-hindbrain (MHB) junction during midgestation (Wilkinson et al.,1987; Ruangvorovat and Lo,1992; Yancey et al.,1992). Since gap junctions can regulate developmental compartments, we thought it likely that Cx43 might interact with wnt-1 at the MHB isthmus organizer and regulate the development of structures originating from the region, such as the cerebellum. In this study, we have chosen to focus on the impact of overexpression of Cx43 on rescuing the cerebellum in the wnt-1 null embryo and on patterns of gene expression in the embryonic brain as a whole and the MHB region in particular during midgestation.
The midbrain-hindbrain (MHB) junction plays a key role in the patterning of the embryonic neural tube and the formation of brain structures such as the cerebellum. The mitogen wnt-1 is critical for cerebellar development, as evidenced by the lack of MHB region and cerebellar formation in the wnt-1 null embryo. We have generated wnt-1 null embryos overexpressing the gap junction gene connexin43 by crossing wnt-1 null heterozygotes into the CMV43 mouse line. We have confirmed that these mice show an increase in gap junctional communication by dye coupling analysis. Two-thirds of wnt-1 null CMV43+ mouse embryos at E18.5 have a cerebellum. In addition, changes in the wnt-1 null phenotype in mouse embryos overexpressing connexin43 are observed as early as E9.5. At this stage, one-quarter of wnt-1 null CMV43+ embryos display extra or expanded tissue present at the MHB boundary (a wnt-1 null enlarged phenotype). In situ hybridization studies conducted on these embryos have indicated no changes in the expression of embryonic brain positional markers in this region. We conclude from these studies that overexpression of the connexin43 gap junction restores cerebellar formation by compensating for the loss of wnt-1. © 2005 Wiley-Liss, Inc.
Molecular Signaling From Isthmus Organizer
The MHB junction or isthmus is a major organizing center in the mammalian embryonic brain and key to proper cerebellar development (Marin and Puelles,1994; Joyner,1996; Liu and Joyner,2001; Wurst and Bally-Cuif,2001). Various signaling molecules involved in the patterning of either specifically midbrain or specifically hindbrain structures, such as Otx2 and Gbx2, are segregated as gradients to opposite sides of this MHB junction, and perturbation of these gradients of expression can result in a misplacement of the MHB border (Simeone et al.,1992,1993; Wassarman et al.,1997; Broccoli et al.,1999; Hidalgo-Sanchez et al.,1999; Millet et al.,1999; Simeone,2000).
Other genes, such as wnt-1 and Cx43, along with En-1, Pax2, Pax5, and FGF-8, have an expression pattern that straddles this border and may play a role in maintaining the boundary or orchestrating downstream events requiring input from both the midbrain and the hindbrain, such as cerebellar development (Crossley and Martin,1995; Joyner,1996). For example, expression of En-1 can trigger a signaling cascade leading to a partial rescue of midbrain-rostral hindbrain development in wnt-1 null embryos (Danielian and McMahon,1996).
Outside of the isthmus region, various CNS positional markers can be used to assay more widespread changes in gene expression upon specific gene disruption. In this study, forebrain, midbrain, and hindbrain markers are all used. The expression pattern of Otx2, a forebrain and midbrain marker, ends sharply at the MHB boundary, making it very useful in studying perturbations in this boundary (Simeone et al.,1992,1993; Bally-Cuif et al.,1995; Millet et al.,1996). Lim-1 is used to detect changes in gene expression in the lateral portion of the diencephalon, the ventral junction of the forebrain and midbrain, and the presumptive spinal cord (Barnes,1994; Fuji et al.,1994). Krox-20 acts as a positional marker for rhombomeres 3 and 5 of the hindbrain at E9.5, rhombomere 3 at E10.0, and the ganglia at E10.5 (Chavrier et al.,1989; Wilkinson et al.,1989). In situ hybridization studies using these markers help determine the overall impact of the wnt-1 knockout on CNS gene expression as well as the impact of overexpression of Cx43 in these wnt-1 knockout embryos.
Wnt-1 and Connexin43 as Putative Regulators of Cerebellum
Much attention has been given to gene regulation of cerebellar development since initial reports in 1990 that wnt-1 regulates normal development of the mouse midbrain and anterior hindbrain (McMahon and Bradley,1990; Thomas and Capecchi,1990). Both of these groups showed that mice homozygous for the wnt-1 null allele lost most of the midbrain and rostral hindbrain. Later in development, these wnt-1 null embryos lacked a cerebellum, suggesting that the cerebellum differentiates from both the midbrain and anterior hindbrain. The fully developed cerebellum coordinates muscular activity; for this reason, wnt-1 null mice either die shortly after birth or survive with ataxia, depending on the targeted disruption (McMahon and Bradley,1990; Thomas and Capecchi,1990). In addition, the swaying mouse mutant has now been identified as a wnt-1 hypomorph, expressing a truncated form of the native wnt-1 protein and showing a reduction in the MHB region leading to a smaller cerebellum (Lane,1967; Thomas et al.,1991). These mice show a perturbation in gene expression at the MHB border (Bally-Cuif et al.,1995). Clearly, wnt-1 is critical for proper cerebellar development.
Experimental manipulation of both Cx43 and wnt-1 appears to result in similar developmental brain defects. Overexpression of the Cx43 gene using a cytomegaloviral promoter (CMV43) resulted in neural tube defects such as incomplete closure in the midbrain and forebrain of the E10.5 embryo, as well as exencephaly of the midbrain and forebrain (Ewart et al.,1997). In addition, using antisense deoxyoligonucleotides (ODNs) to knock down Cx43 expression resulted in common CNS malformations such as spina bifida, anencephaly, and myeloschisis (Becker et al.,1999).
The fact that perturbation of these two genes causes similar dorsal neural tube defects may simply reflect their common location and expression pattern in the developing neural tube. On the other hand, the two genes may function in the same genetic cascade for normal development of the neural tube. Previous studies have established the interaction between wnt-1 and gap junctions in other organisms. In Xenopus embryos, increasing wnt-1 expression enhances gap junctional intercellular communication (GJIC), which leads to patterning defects during early development (Olson et al.,1991; Olson and Moon,1992). Another member of the Xenopus wnt family, Xwnt-3a, triggers the expression of the Cx30 gap junction (McGrew et al.,1999). In addition, wnt-1 specifically induces expression of Cx43 and increases GJIC in neural crest-derived PC12 cells and cardiac myocytes (van der Heyden et al.,1998; Ai et al.,2000). Ectopic expression of wnt-1 in the mouse limb bud also induces changes in Cx43 expression (Meyer et al.,1997). Based on these studies, clearly a regulatory connection exists between wnt-1 and gap junctions. However, the precise pathway still needs to be elucidated.
Gap Junctions and Embryonic Development
Cx43 is a member of the gap junction gene family of intermembrane channels involved in cell-to-cell communication (Gilula et al.,1972; Beyer et al.,1987; Bruzzone et al.,1996; Simon and Goodenough,1998). GJIC has been shown to be a critical component of embryogenesis (Lo,1996,1999; Levin,2001). In the developing mouse embryo, GJIC is progressively restricted, eventually being limited to within each germ layer in the gastrulating mouse embryo (Lo and Gilula,1979a,b; Kalimi and Lo,1988). In addition, dye flow analysis in chick neuroepithelial cells indicates that dye transfer is slowed between rhombomeres of the chick hindbrain, reflecting a decrease in GJIC between rhombomeres of the embryonic brain (Martinez et al.,1992). Studies conducted using Cx43 knockout mice and mouse embryos expressing a dominant negative form of Cx43 indicate that Cx43 is also critical for normal heart development and neural crest cell migration (Reaume et al.,1995; Huang et al.,1998; Sullivan et al.,1998). These studies have shown that GJIC compartments contribute to the formation of developmental boundaries in the embryo and play a key role in many aspects of embryonic development.
Goals of This Study
In our study, we specifically set out to look at the effect of overexpression of Cx43 on the cerebellar defect observed in wnt-1 null mice. The major focus of the study was first to examine whether there was a significant number of wnt-1 null embryos overexpressing Cx43 that had a cerebellum. We found this to be true. We then measured the specific brain vesicles of earlier-stage embryos to see if both the midbrain and hindbrain were present in these wnt-1 null embryos overexpressing Cx43. All three vesicles, forebrain, midbrain, and hindbrain, were larger in these transgenic mice compared to control mice. Finally, we characterized the expression patterns of several positional markers and also measured the dye flow via gap junctions of neuroepithelial cells of the MHB junction. No change was detected in Otx2, Krox-20, or Lim-1 expression or in the expression of the DHFR marker for the CMV43 transgene. Dye coupling studies showed that a wnt-1 null mouse with at least one copy of CMV43 displayed a higher level of GJIC than a mouse without. In all embryos carrying CMV43, there was more gap junctional activity, but this increase was greater if the embryo carried at least one copy of wild-type wnt-1. Therefore, after establishing overexpression of Cx43 and gap junctions in these wnt-1 null embryos, we suggest that this overexpression partially compensates for the mutant phenotype of the wnt-1 null embryos.
MATERIALS AND METHODS
Mice overexpressing Cx43 were generated in the laboratory of Dr. Cecilia Lo. The targeting construct and creation of the line have been described previously (Ewart et al.,1997). These mice were originally the product of SWR/J × SJL/J matings, which were then later outbred. According to Ewart et al. (1997), the same phenotypes were observed in other inbred and outbred backgrounds. The targeting construct contained a cytomegalovirus (CMV) promoter/enhancer expression vector with a 1.2 kB cDNA fragment from an identified coding region of Cx43 (Sullivan et al.,1992). In addition, a 600 bp DNA tag was attached 3′ to the cDNA fragment in an untranslated portion of the expression vector. This tag, a gene marker, came from a portion of the dihydrofolate reductase gene of Toxoplasma gondii (Roos,1993). This marker was used for PCR analysis to detect the presence of the CMV43 transgene. Dhfr was used also for in situ hybridization analysis as a gene marker for CMV43 transgene expression along the dorsal aspect of the neural tube, as well as neural crest-derived structures (Ewart et al.,1997).
CD-1 mice (Charles River, Boston, MA) heterozygous for the wnt-1 null allele were also generated in the laboratory of Dr. Cecilia Lo. The original transgenic mice were obtained from the laboratory of Dr. A.P. McMahon at Harvard University. The knockout targeting construct and creation of this mouse line have both been described previously (McMahon and Bradley,1990).
Mice were housed at the Animal Colony of the Children's Hospital of Philadelphia and in the Vivarium of the Biology Department of Villanova University and given food and water ad libitum.
Setting Up Mouse Matings
Matings were set up between mice homozygous for CMV43 and mice heterozygous for the wnt-1 null mutation. Wnt-1 null homozygotes were found to be nonviable in this wnt-1 null line (McMahon and Bradley,1990). Mice containing one copy of CMV43 and one copy of the wnt-1 null allele from this first round of mating were then used in a second round of matings. The 16 possible genotype combinations from these crosses are listed in Figure 1, whose legend also includes abbreviations used for genotypes.
There is a 3/16 probability that these mice will carry two copies of wnt-1 null and at least one copy of CMV43. There is a 6/16 probability that mice will carry one copy of wnt-1 null and at least one copy of CMV43. It is also likely that 3/16 will be wild-type for wnt-1 and carry CMV43. Finally, 2/16 should carry one copy of wnt-1 null without CMV43, 1/16 should be homozygous wnt-1 null without CMV43, and 1/16 will be completely wild-type.
In this work, we describe genotypic and phenotypic analysis of embryos produced from this second round of mating. In addition to matings listed in Figure 1, wnt-1 null heterozygous CMV43− mice were mated to wnt-1 null heterozygous CMV43+ mice for the E18.5 mouse embryo collections.
Genotyping and Embryo Collections
Genotyping of weaned mice and mouse embryos by PCR was performed as described previously (Ewart et al.,1997). Embryos were collected at E9.5/E10.5 (n = 124) and E18.5 (n = 111). The actual number of embryos at each genotype and for each sex did not differ from expected numbers (data not shown). In addition, the sex of the embryo did not depend on either the presence of the wnt-1 null transgene or CMV43 (data not shown). For in situ hybridization and morphological and morphometric analysis, embryos were placed in 2 ml of Carnoy's fixative for 3–4 hr to overnight (MacDonald and Tuan,1989). For histological analysis of E18.5 embryonic brains, sections were stained with cresyl violet as described previously (Sheehan and Hrapchak,1980). In order to preserve the structure of the cerebellum, the entire embryo head was fixed and sectioned.
Spaced serial sagittal sections were analyzed using both a digitized compensating planimeter to measure the area of the embryonic forebrain, midbrain, and hindbrain of E9.5 mouse embryos and a computerized planimeter. Total brain and cavity areas were measured, while the tissue area was calculated as the difference between these two. An example of an E18.5 sagittal section and cerebellar tracing used for morphometrics is shown in Figure 2a and c. Morphological markers were chosen to define the cranial and caudal ends of the forebrain, midbrain, and hindbrain (Fig. 2b).
Planimeter measurements (E9.5 embryos).
Tracings of the sections were made by projecting the image of the section from a compound microscope to a white piece of paper using a series of mirrors. Every third section for controls and every other section in wnt-1 null enlarged embryos were traced and measured with the planimeter. The tissue, cavity, and total volumes were calculated by summing the areas of each embryo and multiplying this by the thickness of each section and the spacing factor (either two or three for E9.5). This number was divided by the magnification factor cubed that was determined by projecting the image of a stage micrometer at the same magnification as the tracings (Desmond and Haas,2000).
NIH Image analysis (E9.5 and E18.5).
Sections were analyzed using the NIH Image analysis computer program to measure the volume of the forebrain, midbrain, and hindbrains of E9.5 mouse embryos and the cerebellum of E18.5 mouse embryos. As with the planimeter, morphological markers were chosen to define the anterior and posterior ends of each brain cavity and the cerebellum. Sections were placed on a Nikon Optiphot-2 compound microscope connected to an MTI-CCD 72 high-resolution camera and a Zenith-386-33 data system. Using this system, sections were projected and focused on a Macintosh high-resolution monitor. These images were captured by NIH Image and saved on the computer. Tissue, cavity, and total volumes were calculated as described above, with NIH image accounting for magnification by calibrating the pixels to a stage micrometer with a given objective magnification (Desmond and Haas,2000). Thus, the areas measured are actual areas of the sample rather than magnified areas. Every fifth section was used for cerebellar measurements.
Dye Injection Studies
Dye injection studies were conducted to assess GJIC in the transgenic mouse embryos used in this study. We recorded dye transfer into dorsal neural fold cells of these embryos. Cells in the neuroepithelium just caudal to the otocyst of the embryo were impaled with a glass microelectrode containing 5% 6-carboxyfluorescein. Four to six sequential impalements of cells along the neural folds were done per embryo. Dye was ionophoretically injected into the impaled cell using 2nA hyperpolarizing current pulses of 0.5-sec duration at a frequency of once per sec. The impalements were held for 2 min followed by removal of the microelectrode. The preparation was viewed under dark-field epifluorescent illumination and the number of dye-filled cells was counted immediately after the impalement was terminated (Ewart et al.,1997).
In Situ Hybridization With Various Riboprobes
In situ hybridization, more specifically RNA-RNA in situ hybridization, was performed on sectioned embryos using a 35S-labeled antisense riboprobe as already reported (Ruangvorovat and Lo,1992). Riboprobes were made using either the T7 or T3 RNA polymerases available in an in vitro transcription kit (Promega, Madison, WI). Whole-mount in situ hybridization was carried out using a digoxygenin-labeled antisense riboprobe according to a previously published protocol (Lowe et al.,1996; Melloy et al.,1998). Sense riboprobes were used as controls.
Rescue of Cerebellum in wnt-1 Null Embryos by Overexpression of Cx43
For mouse embryos of all six genotypes, the macrostructure for each cerebellum was observed. Guidelines as to normal cerebellar structure for the E18.5 mouse embryo were set according to the structure of the wild-type E18.5 mouse embryos from these collections. All of the transgenic embryos carrying either the wnt-1 null allele and/or exogenous Cx43 were compared to the wild-type standard. Of 50 cerebella measured, 6 were wnt-1 null embryos carrying the CMV43 transgene, of which 4 had a cerebellum (Fig. 3). As expected, none of the embryos carrying two copies of wnt-1 null without exogenous Cx43 had a cerebellum. In addition, there was no detectable difference in the morphology of the cerebella of functional wnt-1 carriers and those homozygous wnt-1 null carrying CMV43 (Fig. 3).
The midbrain adjoining the cerebellum, as well as the hindbrain, appears normal as compared with cerebella from wild-type wnt-1 embryos. The tectum of the midbrain was completely formed, as contrasted with malformed tecta of the wnt-1 null embryos. The cerebellar epithelium, the small external germinal layer, and the cells comprising the differentiating field were all present in these smaller cerebella (Altman and Beyer,1995; Kaufman,1995).
As can be seen in Figure 4, the histology of the control (wild-type wnt-1, CMV43−) illustrates the three typical layers of cells in the developing cerebellum, i.e., an inner ventricular germinal layer in which cells proliferate and migrate, a middle mantle layer that has both a deep and superficial layer of cells that are differentiating into both glia and neurons, and a superficial layer from which cells migrate in a retrograde manner into the mantle layer (Miale and Sidman,1961; Jacobson,1978). The sections of the experimental cerebella (wnt-1 null, CMV43+), as illustrated in the right panel of the same figure, appear similar if not identical to those of the controls. These sections do not show the detail between cell types because they have not fully differentiated into basket cells, Purkinje cells, and granule cells, for example, and can only be identified as specific neurons and glia with certainty using antibody tagging or clonal lineage analysis (Mathis and Nicolas,2003).
After comparing the cerebella of wnt-1 null CMV43+ mouse embryos to those of wnt-1 wild-type mouse embryos, we next examined total cerebellar volumes (Fig. 5, Table 1). The cerebellar volumes of all mouse embryos carrying at least one wild-type wnt-1 allele did not significantly differ from one another (Fig. 5, Table 1). In E18.5 wnt-1 null CMV43+ mouse embryos having a cerebellum, the average cerebellum size was half that of mouse embryos containing at least one wnt-1 wild-type allele (Fig. 5, Table 1). Therefore, overexpression of Cx43 can result in the rescue of the cerebellum in wnt-1 null mouse embryos. Although these cerebella are morphologically the same as their wild-type counterparts, total cerebellar volume is reduced when compared to wild-type.
|Genotype||Mean Volume (mm3)||Standard error||Number|
Analysis of wnt-1 Null Enlarged MHB Phenotype
The majority of wnt-1 null E9.5/E10.5 embryos observed had the typical wnt-1 null phenotype. This meant that these mouse embryos had a missing area of tissue at the MHB junction just as described previously (McMahon and Bradley,1990) (Fig. 6B). In addition to a wnt-1 null phenotype, one embryo displayed a collapsed hindbrain phenotype characteristic of some CMV43+ embryos (Ewart et al.,1997). However, among the wnt-1 null CMV43+ embryos observed, 27% had a phenotype that appeared to be a modification of the wnt-1 null phenotype (Fig. 6C and D). We designated this phenotype the wnt-1 null enlarged phenotype. These mutants had a missing band of tissue at the MHB junction like a typical wnt-1 null homozygote, but more tissue appeared to be present in this region (arrowheads in Fig. 6C and D vs. B). In terms of the general appearance of the embryos, the wnt-1 null embryos with the classic phenotype had a flattened neural tube roof (Fig. 6B), whereas the wnt-1 null enlarged embryos had a more rounded roof (Fig. 6C and D). This roundness was more apparent in the embryo found in Figure 6C than the embryo in Figure 6D, reflecting the range of phenotypes or gradient of rescue in the wnt-1 null enlarged embryos. Since this degree of roundedness was not uniform, identification of wnt-1 null enlarged embryos was sometimes difficult to determine without practice or other wnt-1 null littermates for comparison. Therefore, all phenotypes were verified by PCR genotyping, correlating the wnt-1 null enlarged phenotype with a subpopulation of wnt-1 null CMV43+ embryos. The nature of this enlarged phenotype became clearer on morphometric analysis.
To examine more closely the difference between wnt-1 null, wnt-1 null enlarged, and wild-type embryos, the volumes of the brains of E9.5 embryos of all six possible genotypes were measured (Fig. 7, Table 2). Total brain volumes and cavity volumes were measured. The cavity volume was then subtracted from the total volume to obtain tissue volumes. Wnt-1 wild-type embryos had approximately the same forebrain, midbrain, and hindbrain tissue volumes with and without CMV43 (controls). For wnt-1 null embryos, having the CMV43 transgene resulted in a 1.3-fold increase in forebrain tissue volume over the volume of those with wild-type Cx43 levels (Fig. 7, Table 2). In addition, overexpression of Cx43 resulted in a 1.6-fold increase in midbrain and a 0.8-fold increase in hindbrain tissue volumes in the wnt-1 null enlarged embryo when compared to wild-type controls. Even at this earlier time in development, the wnt-1 null enlarged embryos show signs of an MHB recovery.
|Genotype||Statistic||Forebrain (mm3)||Midbrain (mm3)||Hindbrain (mm3)|
|Wnt-1 null, CMV43+ (enlarged)||Mean||0.068||0.0192||0.023|
|Wnt-1 null, CMV43−/−||Mean||0.051|
|Wnt-1+/+, CMV43+ (wild-type) control||Mean||0.023||0.0074||0.0126|
|Wnt-1+/+, CMV43−/− (wild-type) control||Mean||0.023||0.004||0.008|
Functional Analysis of Gap Junctions
Dye transfer analysis studies were also conducted to assess the level of GJIC in wnt-1 null and control embryos. In the initial characterization of the CMV43 mice, it was shown that overexpression of Cx43 increased GJIC (Ewart et al.,1997). We determined the level of GJIC for mouse embryos of all three wnt-1 genotypes: wild-type, heterozygous wnt-1 null, and homozygous wnt-1 null. The level of GJIC was determined by counting the number of cells filled with dye 2 min postinjection (Table 3). These results indicated that for mouse embryos of each class of wnt-1 genotype, overexpression of Cx43 was found to enhance dye transfer and thus increase the level of GJIC within the neural tube.
|PCR ± transgenes Wnt, dhfra||Genotype descriptionb||Number of dye-filled cells||Number of embryos||Number of cells injected with dye|
|−/−,−||Wnt KO; no Extra Cx43||1.58 ± 0.13||6||53|
|−/−, +||Wnt KO; one copy extra Cx43||1.96 ± 0.162||6||44|
|+/−, −||Wnt KO one; no extra Cx43||4.15 ± 0.164||5||34|
|+/−, +||Wnt KO one; one copy extra Cx43||5.34 ± 0.279||5||32|
|+/+, −||Wild-type two; no extra Cx43||3.97 ± 0.157||5||37|
|+/+, +||Wild-type two; one extra Cx43||6.66 ± 0.177||4||32|
After our initial observations of the increase in GJIC when Cx43 was overexpressed, we also found that dye coupling was greater in mouse embryos containing at least one wild-type copy of wnt-1. Wnt-1 null heterozygous embryos overexpressing Cx43 displayed a dye transfer rate nearly three times that of the wnt-1 null embryos overexpressing Cx43 (1.96 vs. 5.34). This coupling is even greater (6.7 cells) when two copies of wild-type wnt-1 are present and Cx43 is overexpressed (1.96 vs. 6.7). This is consistent with recent studies indicating that increasing wnt-1 expression in PC12 cells and cardiac myocytes results in an increase in dye coupling (van der Heyden et al.,1998; Ai et al.,2000). However, in the presence of wild-type wnt-1 and endogenous Cx43 alone, there is still greater coupling than in embryos with no wnt-1 and exogenous Cx43 (4.15 vs. 1.96). This suggests that there is an interaction between wnt-1 and Cx43 contributing to cell coupling and the coupling is greater in the presence of wild-type wnt-1. Nevertheless, there was a 24% increase in dye transfer seen in a wnt-1 null embryo overexpressing Cx43 vs. a wnt-1 null embryo with only endogenous Cx43 (1.96 vs. 1.58). One example of these dye transfer studies is shown in Figure 8. Based on the presence of the wnt-1 null enlarged phenotype, this 24% increase seems sufficient to compensate for the lack of wnt-1. Although changes in dye coupling were smaller without wild-type wnt-1 present, the increase in GJIC suggested that the phenotypic changes observed in the wnt-1 null embryos overexpressing Cx43 could be attributed to an increase in GJIC.
In Situ Hybridization Analysis Using Known CNS Positional Markers
To define the wnt-1 null and wnt-1 null enlarged embryo phenotypes, in situ hybridization analysis using sectioned embryos was carried out with well-known CNS positional markers. Probes to detect the gene expression of Otx2, Krox-20, dhfr, and Lim-1 were used on mutant embryos and normal littermate controls. As described above, these gene probes acted as positional markers to show whether CMV43 in combination with the wnt-1 null knockout changed any patterns of normal gene expression.
We characterized the expression patterns of Otx2, Krox-20, and Lim-1 by conducting in situ hybridization on sections (Fig. 9). No differences in the expression of any of these genes were observed between wild-type embryos, wnt-1 null CMV43+ embryos, and wnt-1 null CMV43− embryos. However, Otx2 expression was reduced in wnt-1 null embryos due to the missing tissue in the mutant embryo brain (Fig. 9b′). Therefore, the defects observed in the wnt-1 null embryo and the wnt-1 null enlarged embryo appear to be confined to the MHB border region. In addition, we did verify the overexpression of Cx43 in the wnt-1 null CMV43+ embryos by using the Dhfr marker (Fig. 9h–i′). The impact of adding CMV43 appeared to have no impact on well-established markers for various embryonic brain regions.
Whole-mount in situ hybridization studies were carried out using a probe specific to Otx2 mRNA expression (Fig. 10). A strict boundary of Otx2 expression was detected at the MHB boundary in wild-type embryos (Fig. 10a and b, arrows). However, in wnt-1 null (Fig. 8C) and wnt-1 null enlarged embryos (Fig. 10d), this sharp boundary was now obscured, and it appeared that Otx2-positive cells were now present in the hindbrain region of these embryos (Fig. 10c and d, arrows). The wnt-1 null enlarged embryo also appeared to have an expanded domain of Otx2 expression in the midbrain (Fig. 10d). These results obtained using whole-mount in situ hybridization differed from those seen in spaced serial sections where no obvious perturbations in the border of Otx2 expression were observed. Because the perturbed expression occurs in small patches of cells within the hindbrain, it is possible that a group of cells misexpressing Otx2 would be overlooked depending on the sections chosen.
Exogenous Cx43 Can Compensate for Loss of wnt-1 in Cerebellum Formation
In summary, overexpression of Cx43 appears to rescue the cerebellar defect in wnt-1 null embryos. This rescue is visible at E18.5, with the cerebella of wnt-1 null CMV43+ mice being approximately half the size of those of wild-type mice. At midgestation, one-quarter of wnt-1 null CMV43+ embryos display a wnt-1 null enlarged phenotype. Mouse embryos with this phenotype appear to have a larger MHB and have been determined morphometrically to show increased tissue volume in the forebrain, midbrain, and hindbrain. Overexpression of Cx43 in these mouse embryos also resulted in an increase in gap junctional intercellular communication. In wnt-1 null CMV43+ midgestational mouse embryos used for in situ hybridization studies, no perturbations in gene expression were seen with several positional markers, with the exception of a fuzzy MHB border of expression for Otx2. This contrasted with the discrete border of Otx2 expression seen in wild-type embryos.
This compensation of the wnt-1 null phenotype by overexpression of Cx43 suggests that endogenous Cx43 may be directly involved in the organizer activity at the MHB junction. In addition, it is possible that cerebellar formation can be induced without normal wnt-1 expression, and that a number of gene pathways may potentially activate formation of the cerebellum. Since the cerebella of wnt-1 null CMV43+ mice are only half the size of their wild-type counterparts, and wnt-1 null enlarged embryos still display a fuzzy Otx2 expression border in the MHB border region, this suggests that there is only partial compensation of the cerebellar developmental defect in these mouse embryos. One explanation for this partial compensation could be the limitations of the CMV43 promoter. These limitations were revealed in previous studies when overexpression of Cx43 by the CMV promoter was shown to rescue the phenotype of the Cx43 knockout mouse only partially (Ewart et al.,1997). Another possible explanation for the partial rescue of the wnt-1 null phenotype by CMV43 may be that this rescue depends on the dose of the CMV43 transgene. Since it is not known what factors are affected by overexpression of Cx43, it is also possible that the partial rescue by exogenous Cx43 may simply reflect the maximum that Cx43 can do without its interaction with wnt-1.
Importance of Gap Junctional Intercellular Communication in Cerebellar Development
We know that overexpression of Cx43 translates into increased cell coupling or increased GJIC based on previous studies (Ewart et al.,1997). In addition, our dye coupling analysis indicated an increase in GJIC in all classes of embryos overexpressing Cx43, establishing a direct connection between overexpression of Cx43 and increased GJIC in our mouse background as well. Moreover, since the cerebellum itself depends on signaling between the midbrain and anterior hindbrain for its formation, GJIC may play a critical part in communication between these two compartments (Hallonet et al.,1990; Millet et al.,1996). As has been mentioned, differential GJIC is important in setting up communication compartments during patterning in embryogenesis (Weir and Lo,1982,1984,1985; Martinez et al.,1992; Levin,2001). Gap junctions can transfer many types of signaling molecules necessary for establishing particular cell and tissue types in the embryo (Bruzzone et al.,1996; Levin,2001). Changes in GJIC have also been correlated with changes in conveyance of signaling molecules involved in cell growth. Transformed cells typically lose gap junctional connections, and transformed cells can be restricted in growth when coupled with normal cells (Mehta et al.,1986; Ruch,1994). More germane to our study is the finding that both calcium ions and cAMP, known to be transferred via gap junctions, are essential for normal brain expansion during early development of the chick (Desmond,1993). GJIC may act on several different levels to affect proper brain expansion and thus cerebellar development. First, GJIC could indirectly upregulate control of apical-to-basal fluid transport. It has been shown that fluid transport is controlled by both hydrostatic pressure and sodium-potassium pumps in the chick embryonic brain (Li and Desmond,1991). Since molecules of low molecular weight (< 1,000 D) can pass between cells through gap junctions, these channels may be involved in transferring signaling molecules necessary to increase fluid transport, leading to an increase in cavity in the size of the cavity of the embryonic brain. Increase in the size directly increases the internal pressure, and increased pressure upregulates cell proliferation of the neuroepithelium surrounding the brain cavity in chick embryos (Desmond and Jacobson,1977; Desmond,1985). This increased pressure has been shown to lead to an increase in mitotic activity (Desmond and Levitan,2004).
Relationship Between wnt-1 and Cx43
One of the reasons for undertaking the study that we report here is that wnt-1 and Cx43 exhibit colocalized patterns of gene expression along the dorsal neural tube (Wilkinson et al.,1987; Ruangvorovat and Lo,1992; Yancey et al.,1992). This similarity may be due to independent induction of expression by other factors, or some type of functional relationship in their gene expression. We know from cell culture studies that increased expression of wnt-1 increases dye coupling (van der Heyden et al.,1998; Ai et al.,2000). This suggests that a direct or indirect relationship is present between wnt-1 and gap junctions. Researchers have also suggested that there is a relationship between expression of Cx43 and β-catenin, a downstream member of the wnt-1 pathway (Ai et al.,2000). Based on our data showing changes in the wnt-1 null phenotype on Cx43 overexpression, Cx43 and wnt-1 expression may be interdependent. Perhaps Cx43 in turn can also affect expression of wnt-1 or other members of the wnt-1 cascade. This may be a way that events occurring at points of cell-to-cell communication, where both connexins and β-catenin are found, can have a direct or indirect effect on gene regulation. Future studies can address the relationship between gap junctions and other wnt-1 pathway members, as well as other related wnt family members.
Significance of This Work
Two major conclusions can be drawn from this study. Increased gap junctional communication in the wnt-1 null embryo can result in the recovery of the cerebellum. In addition, the effect of this increased GJIC is first visible at midgestation, when wnt-1 null enlarged embryos are observed.
There are some important possible extensions to this study. For example, older neonates with the rescued cerebellum phenotype could be examined to determine whether or not their cerebella can perform normal physiological activities such as respiration. Usually, wnt-1 null mice die shortly after birth due to a failure of the diaphragm, which is under the control of the autonomic nervous system (McMahon and Bradley,1990). Future studies could also elaborate on the cellular mechanisms at the isthmus during early embryonic stages of development. In addition, one could examine more developmental stages to follow the wnt-1 null enlarged embryos to the point of cerebellar development.
Similar Neural Tube Defects Result From Different Experimental Approaches
The loss of wnt-1 causes a prominent neural tube defect (NTD), i.e., the complete loss of the midbrain and hindbrain in the mouse. Similarly, overexpression of the Cx43 gene using a cytomegaloviral promoter (CMV43) sometimes resulted in embryonic NTDs (Ewart et al.,1997). While the NTDs observed in the CMV43 embryos were less severe (holes in the roof plate) compared with those of the wnt-1 knockout (absence of entire vesicles), the existence of the NTDs in both cases suggest that their etiologies may be the same or similar.
Specific NTDs observed in heterozygous CMV43 mice included openings along the dorsal midline in both the mesencephalon and telencephalon of the E10.5 embryo as well as exencephaly of the telencephalon and mesencephalon. The neuroepithelium of the brains of these embryos with openings exhibited collapsed and excessive folding similar to the phenotype of chick embryos in which artificial holes were made via microsurgical intubation (Desmond and Jacobson,1977). The foldings of the neuroepithelium can result in excessive expansion of brain vesicles (hydrocephalus) rostral to the obstruction caused by the foldings. Openings in the neural tube that occur spontaneously, e.g., spina bifida, vent the fluid out of the neural tube, resulting in obstructive foldings. Such openings of the developing neural tube to the outside, plus folding of the neuroepithelium as a consequence of venting of the cerebrospinal fluid, were recently proposed as a general cause of the Chiari II malformation in a unified theory for the pancerebral abnormalities associated with this malformation (McLone and Dias,2003). The cause of this NTD has been debated for years and the specific types of malformations vary. However, being able to link NTDs caused microsurgically with those caused by overexpression (Cx43) or underexpression of a gene (wnt-1) suggests the plausibility of using experimental gene strategies to understand the natural etiology of such NTDs.
The Desmond laboratory is grateful to the following undergraduate students who did the daily tasks of maintaining the mouse colony, genotyping the embryos, preparing sections, and measuring embryo sections: Jessica Bazan, Marissa Fankhanel, Danielle Garneau, Rachel Kobos, Michael Ladrigan, Suzanne Maahs, Jessica Mellilo, Regina McInerney, Chemine Richa, Jennifer Smoral, Joseph Spina, and Danielle Waldrop. The authors also acknowledge Matthew Cohen of the Lo laboratory and are grateful for the help in making the photographs publication ready by Mr. Kevin Donahue, Center for Instructional Technologies, Villanova University. Supported by the National Institute of Neurological Disorders and Stroke (NINDS) (grant 24136 to M.E.D.), the National Institute of Child Health and Human Development (NICHHD) (grant 24710 to M.E.D.), the National Institutes of Health (grant Z01-HL005701 to C.W.L.), a Villanova Graduate Research Fellowship (to P.G.M.), and a Summer Research Grant (to M.E.D.).