Abstract
The Disabled-1 protein in mouse is known to be an intercellular signaling component of the Reelin molecular pathway that subserves neuronal migration in several structures in the brain and spinal cord. The scrambler mutant mouse, which is phenotypically identical to the reeler mouse, is due to a mutation in the disabled-1gene (Howell et al., 1997; Sheldon et al., 1997). The Purkinje cells of the cerebellum express Disabled-1 and experience a massive failure of migration in the scrambler mutant mouse (Howell et al., 1997; Sheldon et al., 1997; Gallagher et al., 1998; Rice et al., 1998). We sought to define the developmental basis of this mutation by studying the Purkinje cell population in experimental mouse aggregation chimeras using a cell marker that permitted the identification of neurons derived from the mutant lineage. We found that a genetically normal component to the environment cannot assist scrambler mutant Purkinje cells in the migratory process. However, the presence of a mutant component to the environment can cause the ectopia of wild-type Purkinje cells. There appears to be a linear relationship between the percentage of the cerebellum that is genetically mutant and the number of wild-type Purkinje cells that express a mutant phenotype. These studies point to the interplay between cell-intrinsic and cell-extrinsic properties in the migration of neurons to form laminated structures during CNS development.
The manner in which populations of cells influence other cells during differentiation, termed the community effect by Gurdon (1988), has been the subject of numerous investigations in developmental biology (Gurdon et al., 1993a,b). The most complex of developing tissues, the mammalian brain, has received only limited attention in this regard, possibly because of the restricted access to important epochs of CNS development. The use of experimental mouse chimeras is one means to explore this issue (Mullen et al., 1997; Rossant and Spence, 1998). By placing cells that are deficient in a given gene in juxtaposition to cells with a normal copy of that gene, it can be determined whether there are influences of mutant or wild-type environments on the developmental processes of cells in situ. Many investigators have used the experimental chimera system to determine the intrinsic or extrinsic nature of unidentified genes in neurological mutant mice. In the near future, when all genes will be known and their function imputed from structure, the intrinsic/extrinsic issue will not be as compelling. However, the chimera situation is also ideal for examining the means by which genetic determinants in cells set the stage, or establish the environment, for subsequent developmental events. It was in this vein that we explored scrambler mouse chimeras to study cell migration in the cerebellum.
The scrambler mutant was identified by its ataxic gait and shown to have a widespread abnormality in cell positioning akin to the phenotype of the reeler mutant mouse (Sweet et al., 1996; Goldowitz et al., 1997;Gonzalez et al., 1997). The scrambler mutation affects thedisabled-1 gene that encodes an adaptor protein for tyrosine kinase signaling (Sheldon et al., 1997). Disabled-1 is expressed by Reelin-receptive targets and is downstream of Reelin signaling (Rice et al., 1998). During formation of the cerebellar cortex, Disabled-1 expression is restricted to the Purkinje cell population (Rice et al., 1998). In the scm/scm mutant, we have determined, as in the reeler cerebellum (Heckroth et al., 1989), that ∼95% of the Purkinje cells fail to migrate to the cerebellar cortex (Goldowitz et al., 1997). The cerebellar Purkinje cells offer an easily quantifiable population to assess the outcome of cell migration. A quantitative assessment of cell phenotype offers the ability to use experimental mouse chimeras to determine whether the scrambler gene acts only in a cell-autonomous manner or whether it operates in a more complex manner for its phenotypic expression. In these studies, we found that although genetically mutant Purkinje cells cannot be rescued by a wild-type milieu, genetically normal cells can express a mutant phenotype that is dependent on the percentage of mutant cells in the chimeric cerebellum.
MATERIALS AND METHODS
Mice and generation of experimental murine chimeras. Scrambler (C3HeB/FeJLe) and BALB/cJ mice (originally obtained from the Jackson Laboratory) were born and maintained in our colony at the University of Tennessee Health Science Center animal care facility. Mice were kept on a 14 hr light/10 hr dark cycle with food and waterad libitum. All of the mice were treated in accordance with the Society for Neuroscience policy on the use of animals in research.
Experimental mouse chimeras were generated in a manner that has been described previously (Goldowitz and Mullen, 1982; Goldowitz, 1989;Goldowitz et al., 2000). In brief, four- to eight-cell scrambler (scm/scm or +/scm) and BALB/cJ embryos were cultured together overnight. After successful fusion, blastocysts were transplanted into the uterine horn of pseudo-pregnant host ICR females. BALB/cJ mice were used as the wild-type component of the chimeras. BALB/cJ mice lack the nuclear envelope antigen, Ringo, which is present in neurons of other mice including scrambler (Hamre and Goldowitz, 1997; R. Mullen, personal communication). By using an antibody against Ringo, we were able to distinguish mutant from wild-type cells (see below).
Tissue preparation. Adult chimeras and control mice (scm/scm and BALB/cJ) were deeply anesthetized with Avertin and transcardially perfused with 0.1m PBS, pH 7.2, followed by paraformaldehyde–lysine–periodate fixative, pH 6.2, for 20 min (McLean and Nakane, 1974; Hamre and Goldowitz, 1997). After post-fixation for 4 hr at 4°C, cerebella were dissected from the skull, rinsed in PBS, and placed overnight in a 5% sucrose in 0.1m PB solution. The brains were then dehydrated in a series of ethanols and embedded in wax. Serial, 8 μm sagittal sections were mounted on Superfrost Plus slides (Fisher Scientific, Pittsburgh, PA).
Immunohistochemistry for detection of Purkinje cell phenotype and genotype. Slides were dewaxed and rinsed in PBS containing 0.3% Triton X-100 (PBS/T), blocked with 5% normal goat serum, and incubated overnight at room temperature with antibodies against either the cerebellar Purkinje cell marker Calbindin [a gift from M. Celio (University of Fribourg, Fribourg, Germany); used at a dilution of 1:1000], Disabled-1 [anti-B3, a gift from Brian Howell (National Institute of Neurological Disorders and Stroke, National Institutes of Health); at a dilution of 1:200), or the non-BALB/cJ nuclear antigen marker Ringo [a gift from R. Mullen (Marine Biological Laboratory, Woods Hole, MA); at a dilution of 1:2]. Slides were rinsed with PBS/T and incubated with a biotinylated secondary antibody (1:200) for 30 min at room temperature. Immunoreactivity was detected using the ABC Elite kit (Vector Laboratories, Burlingame, CA) and diaminobenzidine as the chromagen. For Disabled-1 immunohistochemistry, an additional antigen retrieval protocol was used before incubation with primary antibody (Jiao et al., 1999). Another set of slides was double immunostained for the Ringo (dilution 1:1) and Calbindin (1:200,000) antigens. Finally, a set of anti-Calbindin immunostained slides was also counterstained with the nuclear stain cresyl violet.
Quantitative analysis of chimeric tissue. Sections from chimeric and control (scm/scm and BALB/cJ) brains were quantified for cerebellar area, total number of granule cells, and total number of Purkinje cells (both ectopic and normally placed). Granule cell and Purkinje cell counts were further subdivided by genotype (Ringo− or Ringo+). To determine cerebellar area, three sections were traced using a camera lucida attachment on a Zeissmicroscope (2.5×). The areas were traced and calculated in NIH Image, and an average was determined. To estimate the total number of granule cells (as determined by their highly distinct morphology), the area of the internal granule cell layer (IGL) was measured from two nonconsecutive sections in the medial cerebellum using Scion Image software (Beta 3b, based on NIH Image for Macintosh and modified for Windows by Scion Corporation, Frederick, MD). The densities of four 2500 μm2 regions, located at an equal distance across the IGL, were then calculated and averaged. The total granule cell number was calculated by multiplying the total IGL area by the average density. The same method was applied to determine the numbers of Ringo− and Ringo+ granule cells within the IGL.
Purkinje cell number was determined, using a 25× objective and the camera lucida, from an average of four evenly spaced sections from the medial cerebellum. Purkinje cells were identified by Calbindin immunopositivity and the presence of a nucleus. The position of each Purkinje cell was assigned to one of three categories: normally placed (in the Purkinje cell layer), ectopic/intermediate (in the IGL or superficial white matter), or ectopic/deep (in the deep cerebellar white matter). In addition, the genotype (Ringo−and Ringo+) of all Purkinje cells, in each category, was assigned, and totals were calculated.
Determination of Ringo− ectopic Purkinje cells in chimeras. To ensure that Purkinje cells counted as Ringo− were not miscounted simply because of the absence of the nucleus, sections were recounted after nuclear staining. Sections double labeled with anti-Calbindin d-28K and anti-Ringo following the protocol above were analyzed and then counterstained with cresyl violet to clearly identify the nuclear envelope or nucleolus of Calbindin/Ringo double-labeled Purkinje cells.
Camera lucida drawings were made to document the position of ectopicscm/scm Purkinje cells in Calbindin/Ringo double-stained sections from five scm/scm ↔ BALB/cJ chimeras. Coverslips of double-labeled sections were then removed by immersing slides in xylenes overnight. The slides were hydrated in serial alcohols and stained in 0.03% cresyl violet solution for 30 min, and coverslips were reapplied. In these counterstained sections, Ringo-negative Purkinje cells in ectopic positions were easily identified and determined to contain or lack a nucleus.
RESULTS
Experimental mouse chimeras were generated by the aggregation of embryos produced from scm/scm female and +/scmmale matings and embryos produced from BALB/cJ female and male matings. An initial measure of chimerism in mice was indicated by the presence of agouti (+/scm or scm/scm) and albino (BALB/cJ) coat colors. The determination of the scrambler genotype (+/scm or scm/scm) was based on the appearance of ectopic Purkinje cells using antibodies against the Purkinje cell marker Calbindin. In the scm/scmcerebellum, 95% of the Purkinje cells are ectopic, whereas in adult wild-type and +/scm cerebella there are virtually no ectopic Purkinje cells (Goldowitz et al., 1997). Of the 17 chimeras examined, 9 chimeras had clearly evident populations of ectopic Purkinje cells and were thus identified as genetically scm/scm ↔ BALB/cJ chimeras. This proportion of homozygous mutant chimeras is expected from the mating scheme used to produce the scrambler component of chimeras. The percentage of scrambler chimerism of the nine chimeras, based on coat color, ranged from 4 to 99% agouti (genotypically scm/scm). Five of these nine chimeras were analyzed in detail and are summarized in Table1. Interestingly, only chimera 9, with the highest percentage of agouti coat color, demonstrated thescm/scm behavioral phenotype, which included ataxia with truncal swaying and a slightly jerky forward movement during locomotion (Goldowitz et al., 1997).
Qualitative observations of scm/scmchimeric cerebella
There was considerable variation in the gross appearance of the cerebella in the nine chimeras (Fig.1C–F). Chimeric cerebella ranged from an almost normal appearance, with only mild differences in the foliation pattern (Fig. 1C) compared with the wild-type cerebellum (Fig. 1A), to a cerebellum that was reduced in size, was lacking foliation, and was almost indistinguishable from the scm/scm cerebellum (compare chimera 9 in Fig. 1F with thescm/scm cerebellum in Fig. 1B). Along with variations in overall size and foliation, there also appeared to be considerable differences in the populations of granule and Purkinje cells among the chimeras (Fig.2E–H, Table 1).
In the wild-type cerebellum, Purkinje cells are arranged in a single layer between the molecular layer (ML) and the IGL (Fig.2A,C). In scm/scmmutants, only 5% of the Purkinje cells are found in this normal position, and the granule cell layer is greatly reduced in size (Goldowitz et al., 1997) (Fig.2B,D). The other 95% of Purkinje cells are found in one of two ectopic positions: neighboring nuclear neurons within the deep cerebellar white matter and within or below the IGL (Fig. 2B, arrowheads andarrows, respectively). These same populations of ectopic Purkinje cells were identified in all of the scm/scm ↔ BALB/cJ chimeras (Fig.2E–H).
Quantitative observations of scm/scmchimeric cerebella
Areal measurements of scm/scm ↔ BALB/cJ cerebella confirmed the variation in cerebellar size (Table 1). The total numbers of cerebellar Purkinje cells in the chimeras were intermediate between wild-type andscm/scm cerebella (Table 1). Likewise, the numbers of ectopic Purkinje cells were intermediate between wild-type andscm/scm cerebella. Estimates of granule cell number in the medial cerebellum of chimeras were also intermediate between wild-type and mutant cerebella. In general, as the percentage of agouti coat color (scm/scm) increased in chimeras, the cerebellar size, Purkinje cell number, and granule cell number concomitantly decreased. Conversely, the numbers of ectopic Purkinje cells increased with increasing percentages of agouti coat color (Table 1).
Genotypic identification of mutant and wild-type cells
The genotype of granule and Purkinje cells within the chimeric cerebellum was determined using the anti-Ringo antibody. As described in Materials and Methods, the Ringo antibody recognizes a nuclear antigen present in all genetically scrambler neurons (Fig.3D) and not BALB/cJ neurons (Fig. 3C). Immunopositive cells have a dark-brown ring outlining the nuclear envelope. In Ringo-immunostained material, granule cells were easily identified by their size and shape, whereas Purkinje cells were identified by double-immunostaining tissue using a highly dilute anti-Calbindin antibody solution (1:200,000) to yield a light-brown cytoplasmic reaction product. All granule and Purkinje cells in the BALB/cJ cerebellum were Ringo immunonegative and positioned normally (Fig. 3C). In thescm/scm cerebellum, all granule and Purkinje cells were Ringo immunopositive (Fig. 3D). Importantly, in the scrambler cerebellum, both ectopic and normally positioned Purkinje cells were Ringo immunopositive.
In all scm/scm ↔ BALB/cJchimeras, the granule cell population was composed of both genetically mutant and wild-type cells (Fig. 3A). As expected, these cells were located in their normal position in the IGL. The percentage of scm/scm granule cells in chimeras reflected the general chimerism seen for coat color (Tables 1,2).
The Purkinje cell population in all chimeras was also composed of both genetically mutant and wild-type cells. Wild-type and mutant Purkinje cells were found in the Purkinje cell layer (Fig. 3A). However, genetically wild-type cells greatly outnumberedscm/scm Purkinje cells within the Purkinje cell layer (Table2). In each of the scm/scm ↔ BALB/cJ chimeras, 5–8% of thescm/scm Purkinje cells were normally positioned, a percentage similar to that seen in the scrambler mutant cerebellum (Goldowitz et al., 1997). This percentage of normally placedscm/scm Purkinje cells was consistent, notwithstanding the percentage of wild-type Purkinje cells that composed the chimeric cerebellum (Fig. 3E). Thus, the same percentage of mutant Purkinje cells was found in normal positions whether there were >90% or <40% of wild-type Purkinje cells in the chimeric cerebellum. We found similar results in the analysis of two sections from the lateral cerebellum (data not shown). This finding, that a primarily wild-type environment is unable to rescue the scm/scm Purkinje cell ectopia, suggests a cell-autonomous function for thedisabled-1 gene.
The genotype of ectopic Purkinje cells was determined in anti-Ringo and anti-Calbindin double-immunostained material. Most of these ectopic Ringo+scm/scm cells were found in white matter deep in the cerebellar cortex (Fig. 3B). There was a small number of Ringo+scm/scm Purkinje cells found in the more superficial white matter and the IGL (Fig. 3A, Table 2). Most surprisingly, we found Ringo− BALB/cJPurkinje cells in both of these ectopic locations in all chimeras (Fig.3B, arrowheads). We examined sections from the cerebellar hemispheres and found very similar results (data not shown).
This finding was unexpected, and we felt that it was critical to confirm that the identification of a Purkinje cell as Ringo−/wild type in a given section was not because of the absence of the nucleus. As mentioned previously, Ringo labels the nuclear envelope. To confirm the presence of the nucleus in Ringo− Purkinje cells, sections immunoreacted for Calbindin and Ringo were counterstained with cresyl violet. Ringo− Purkinje cells were reconstructed in serial sections using a camera lucida. Both Ringo− and Ringo+ Purkinje cells in ectopic populations were easily identified (Fig.4A,B). We found that almost all of the Ringo− Purkinje cells in ectopic positions were also Ringo−/cresyl violet+ with clear staining of the nuclear envelope and/or nucleolus (Fig. 4A). Furthermore, when chimeric cerebella were stained with a polyclonal antisera raised to Disabled-1, there were two distinct populations of ectopic Purkinje cells: very lightly stained cells that reflected the small amount of Disabled-1 protein still expressed in the scm/scm mutant Purkinje cells (Sheldon et al., 1997) and intensely stained cells that were wild-type Purkinje cells expressing normal levels of Disabled-1 (Fig. 4C,D).
Counts of Ringo−/cresyl violet+ Purkinje cells were made for each chimera. On the basis of these counts, the overall percentages of wild-type, ectopic Purkinje cells were calculated (Table3). We found that the numbers of ectopically positioned wild-type Purkinje cells increased as the percentage of genotypically scm/scm Purkinje cells increased (Fig. 5). This finding, that the mutant environment has an affect on wild-type cells, indicates that there is a non-cell-autonomous component of mutant disabled-1 in the positioning of Purkinje cells.
DISCUSSION
Here we demonstrate that the phenotype of the scm/scmchimeric cerebellum is intermediate between wild-type and scrambler mice. With increasing numbers of scm/scm Purkinje cells in a chimera, cerebellar size, foliation, and granule cell number are diminished. Thus, our data indicate that there is a clear relationship between the percentage chimerism of mutant Purkinje cells and the severity of the mutant phenotype. This is in contrast to reeler chimeras in which equal mixtures of mutant and normal cells produced no mutant phenotype in the cerebellum (Terashima et al., 1986). In fact, only when >90% reeler cells colonized the chimeric cerebellum was there evidence of a mutant phenotype (Mullen, 1978). This demonstrates the expected cell-extrinsic nature of an extracellular matrix-like protein such as Reelin with the intrinsic property expected of a cytoplasmic adaptor molecule such as Disabled-1 (Howell et al., 1997;Sheldon et al., 1997). This cell-intrinsic mechanism is witnessed by the fact that no matter what the contribution of wild-type cells to the chimeric cerebellum, there was a consistent and low level of normally placed mutant Purkinje cells similar to what is seen in thescm/scm cerebellum. For example, in chimera 1, the cerebellum of which was almost normal, ∼95% of the mutant Purkinje cells were found in ectopic locations.
What proved to be an unexpected finding was that wild-type Purkinje cells adopted a mutant phenotype when confronted with increasing proportions of mutant Purkinje cells and/or environment. That is, as the number of mutant Purkinje cells increased (also indicative of an elevated presence of mutant cells in the overall chimeric environment), the number of wild-type Purkinje cells found in ectopic positions also increased.
The finding that cellular phenotypes are influenced in one direction but not the other, i.e., genetically mutant Purkinje cells are not aided by the presence of wild-type cells but genetically normal Purkinje cells are hampered in their migration by the presence of mutant cells, appears as an anomaly. This conclusion can be understood by considering that successful migration may be seen as requiring at least two components. First, a cell must have intact signaling pathways and cellular architecture that enable migration. Second, there are environmental conditions that are required for, or that can negatively impact, successful migration (Gao et al., 1992; Jensen et al., 1999). In essence, this is the basis of the community effect applied to a complex system such as the developing brain and can explain our findings that wild-type Purkinje cells can express an ectopic phenotype in a mutant environment.
There are at least three environmental conditions that may bring about the ectopia of wild-type Purkinje cells. First, thedisabled-1 mutation in scrambler may affect afferent input and impede migration guided by incoming axons. It has been hypothesized that early afferents may provide a migratory substrate for Purkinje cells in the developing cerebellum (Altman and Bayer, 1997), and furthermore it has been shown that these fibers are Disabled-1 positive (Howell et al., 1997; Rice et al., 1998). Second, mutations in the Reelin signaling pathway may hinder the ability of Purkinje cells to detach from radial glia on which they are believed to migrate (Yuasa et al., 1993). There is reasonable evidence to suggest that the mode of action of the Reelin signaling pathway is to enable migrating cells to leave the radial glial guide (Pinto-Lord et al., 1982). In the scrambler chimeric cerebellum, scm/scm Purkinje cells may remain attached to the radial glia and prevent wild-type Purkinje cells from proceeding further. However, we found wild-type ectopic Purkinje cells that do not have neighboring mutant Purkinje cells. This suggests an influence of a possible “group effect” that operates at some distance from the affected cell. Thus, a third possibility is that a defect in signaling from mutant Purkinje cells to the radial glia (or some other cell type) may disrupt important instructive events for successful migration.
Our finding of a non-cell-autonomous component in Purkinje cell migration demonstrates the complexity of this signaling pathway. It is only through careful quantitative analysis that the non-cell-autonomous role of Disabled has come to light. This finding highlights the power of the cerebellum as a model system to examine the function of Disabled-1 and other molecules involved in cell migration. Over the past few years, most of the studies designed to explore the function of these molecules have focused on the cortex. However, because of the large number of cell types and overall complexity of the cortex, phenotypic analysis is restricted to qualitative observations. For example, the recent work by Hammond et al. (2001), who analyzed the cortex from blastocyst injection chimeras of disabled-1 null mice, found that wild-type cells could traverse mutant territory to establish an apparently normal “super cortex.” This finding is qualitatively similar to our analysis of wild-type Purkinje cells and dramatically illustrates the cell-autonomous nature of thescrambler mutation, but as we found for Purkinje cells,Hammond and colleagues (2001) also found some wild-type cells in ectopic positions. However, the quantitative relationship between chimerism and wild-type cells expressing a mutant phenotype in the chimeric cortex is not easily ascertainable. Such a quantitative relationship as shown in Figure 5 can provide critical insights into cellular interactions in neural development.
Finally, it appears that what we found in cerebellum and Hammond et al. (2001) reported in cortex can be generalized to other regions affected by mutations in the disabled-1 gene. Preliminary examination of the hippocampus in our scrambler chimeras revealed a clear segregation of genotypically mutant and normal cells (our unpublished observations). As we found for the cerebellum, wild-type cells in the hippocampus are segregated in the normal pyramidal cell layer, whereas mutant cells occupy a ragged bilayer that is typical of the scrambler condition. Importantly, in chimeras with increased percentages of mutant cells, we began to find wild-type cells in ectopic positions.
Thus the cellular simplicity of the cerebellum offers a model system in which one can quantify the effects of single molecular differences on cell behaviors such as migration. The chimeric cerebellum is an ideal assay system for the sensitive quantitation of migratory behavior in the brain and demonstrates in the present study that Disabled-1 works in both a cell-intrinsic and cell-extrinsic manner to guide Purkinje cell migration.
Footnotes
This research was supported by a University of Tennessee Health Science Center Neuroscience postdoctoral fellowship (H.Y.) and a grant from the Human Frontiers Science Program (D.G.). We thank Richard Cushing for technical assistance.
Correspondence should be addressed to Dan Goldowitz, Department of Anatomy and Neurobiology, University of Tennessee Health Science Center, 855 Monroe Avenue, Memphis, TN 38163. E-mail:dgold{at}nb.utmem.edu.