Previously we observed that stable clones of multipotent neural progenitor cells, initially isolated and propagated from the external granular layer of newborn wild-type mouse cerebellum, could participate appropriately in cerebellar development when reimplanted into the external granular layer of normal mice. Donor cells could reintegrate and differentiate into neurons (including granule cells) and/or glia consistent with their site of engraftment. These findings suggested that progenitors might be useful for cellular replacement in models of aberrant neural development or neurodegeneration. We tested this hypothesis by implanting clonally related multipotent progenitors into the external granular layer of newborn meander tail mice (gene symbol=mea). mea is an autosomal recessive mutation characterized principally by the failure of granule cells to develop in the cerebellar anterior lobe; the mechanism is unknown. We report that ~75% of progenitors transplanted into the granuloprival anterior lobe of neonatal mea mutants differentiated into granule cells, partially replacing or augmenting that largely absent neuronal population in the internal granular layer of the mature meander tail anterior lobe. (The ostensibly ‘normal’ meander tail posterior lobe also benefited from repletion of a more subtle granule cell deficiency.) Donor-derived neurons were well-integrated within the neuropil, suggesting that these progenitors’ developmental programs for granule cell differentiation were unperturbed. These observations permitted several conclusions.

(1) That exogenous progenitors could survive transplantation into affected regions of neonatal meander tail cerebellum and differentiate into the deficient cell type suggested that the microenvironment was not inimical to granule cell development. Rather it suggested that mea’s deleterious action is intrinsic to the external granular layer cell. (Any cell-extrinsic actions – albeit unlikely – had to be restricted to readily circumventable prenatal events.) This study, therefore, offers a paradigm for using progenitors to help determine the site of action of other mutant genes or to test hypotheses regarding the pathophysiology underlying other anomalies.

(2) In the regions most deficient in neurons, a neuronal phenotype was pursued in preference to other potential cell types, suggesting a ‘push’ of undifferentiated, multipotent progenitors towards compensation for granule cell dearth. These data suggested that progenitors with the potential for multiple fates might differentiate towards repletion of deficient cell types, a possible developmental mechanism with therapeutic implications. Neural progenitors (donor or endogenous) might enable cell replacement in some developmental or degenerative diseases – most obviously in cases where a defect is intrinsic to the diseased cell, but also, under certain circumstances, when extrinsic pathologic forces may exist.

The cerebellum has been a model site for studying neurogenesis in the central nervous system (CNS). The identification of genes pivotal in cerebellar development has been aided for nearly three decades by the recognition of naturally occurring mouse mutants with cerebellar defects. The meander tail (mea) mutant provides a prototype for the failure of a specific type of cerebellar neuron to develop properly. mea is an autosomal recessive mutation, which causes both a skeletal defect, resulting in a kinked tail, and a neural developmental defect, featuring a cerebellar external granular layer (EGL) that generates a markedly reduced number of granule cell (GC) interneurons within a severely attenuated, largely unformed internal granular layer (IGL) at maturity (Hollander and Waggie, 1977). This abnormality is most prominent in the anterior lobe (AL), raising speculation that mea might influence a rhombomeric-determining gene (Ross et al., 1990; Fletcher et al., 1991; Napieralski and Eisenman, 1996). However, Hamre and Goldowitz (1997), as well as data presented here, suggest that the posterior lobe (PL) is also affected. It remains uncertain, therefore, where mea exerts its effect on cerebellar histogenesis. The GC population appears to derive entirely from the EGL (Jankovski et al., 1996). It is this neuronal cell type that appears most (though not exclusively) affected by mea gene action. However, it remains unclear at what stage that action occurs and whether that action is intrinsic to the mutant GC or extrinsic to it.

We previously demonstrated that clonally related progenitors, derived from normal neonatal mouse EGL and propagated in vitro, are capable of integrating in a cytoarchitecturally appropriate fashion following transplantation into the EGLs of normal newborn mice (Snyder et al., 1992). These clones are multipotent and differentiate into various normal cerebellar neuronal and glial cell types in vivo following grafting.

In the present study, the cerebella of mea (mea = phenotype) pups were implanted at birth with clonal, multipotent, EGL-derived neural progenitors. By the completion of cerebellar development, a majority of engrafted progenitors, including within the AL, had differentiated into GCs, suggesting partial reconstitution of the mea IGL. There was a suggestion, in fact, that progenitors might shift their differentiation towards compensation for the deficient GCs, an intriguing developmental mechanism. In addition to repopulating regions of CNS as replacement therapy for certain neurodevelopmental pathologies (the first such report for progenitors), such a use of pro-genitors may also help test hypotheses regarding the pathobiology underlying a particular mutation. In the experiments described here, the survival of exogenous normal, clonal, multipotent cerebellar progenitors in affected regions of newborn mea cerebellum and their differentiation into the missing cell type suggested that the microenvironment into which they were implanted was not inimical to GC development. It was concluded, therefore, that mea’s deleterious gene action is intrinsic to the mutant GC. (A less likely possibility, is that mea operates at a stage and/or in a region other than the postnatal EGL).

Cerebellar progenitor cell transplantation into mea mouse hosts

Clonally related, multipotent cerebellar progenitor cells were propagated, maintained and transplanted as previously detailed (Snyder et al. 1992). Briefly, clone C17.2, the stable neural progenitor cell clone used in this study as well as in wild-type postnatal mouse cerebellum (Snyder et al., 1992), was originally generated by transduction of vmyc into individual primary progenitors cultured from the EGL of neonatal wild-type mouse cerebellum. The clone was subsequently transduced with a retroviral vector encoding lacZ, enabling donor progenitors to produce β-galactosidase (β-gal) constitutively. These stable β-gal+ cells are free of helper virus. They emulate, model and share properties with both primary CNS stem cells and progenitors propagated by a number of techniques (reviewed in Weiss, 1996; Gage et al., 1995; Kilpatrick et al., 1995; Martinez-Serrano and Bjorklund, 1996; McKay, 1997). Furthermore, these cells can participate in the normal development of cerebellum as well as many other CNS structures (Snyder, 1994; Snyder et al., 1997). Undifferentiated C17.2 cells were trypsinized from a 90% confluent dish and resuspended in phosphate-buffered saline at 1-4×104 cells/μl. (Progenitors that have differentiated prior to transplantation engraft poorly, if at all. Therefore, they did not constitute the donor-derived cells principally investigated in these experiments.) The cerebella of cryoanesthetized newborn mea pups and of unaffected littermates were injected with 1-2 μl of the cellular suspension via a glass micropipette placed into the EGL of the vermis and lateral hemispheres (i.e., 1-8×104 cells per injection into each of three sites).

Noninbred mea breeders were obtained initially from Jackson Lab-oratories and maintained thereafter in our laboratories. Homozygous mea mice can be reliably identified at embryonic ages and thereafter by the presence of the kinked tail. This skeletal defect correlates invariably with the cerebellar defect. Skeletal and neuroanatomic defects are never seen in heterozygous or wild-type mice.

Detection and characterization of integrated donor progenitors

At the completion of cerebellar organogenesis (≥3 weeks of age), serial 100 μm free-floating parasagittal sections from the cerebella of recipient mice were processed with X-gal histochemistry (to reveal β-gal+ donor-derived cells) and embedded in Epon-Araldite (Snyder et al., 1992, 1995, 1997). Semithin 1 μm plastic-embedded sections, counterstained with toluidine blue, were examined with bright-field optics. Ultrathin sections were examined by electron microscopy as described (Snyder et al., 1992, 1997). The X-gal reaction product forms a crystalline blue precipitate which is electron dense and non-diffusible, allowing donor-derived β-gal+ cells to be unequivocally identified and distinguished from endogenous cells at both the light microscopic and electron microscopic levels (Snyder et al., 1992, 1997). The precipitate is typically localized to the nuclear membrane, endoplasmic reticulum and other cytoplasmic organelles, and it frequently extends into cellular processes. It is best visualized without lead citrate or uranyl acetate counterstain.

Analysis of engrafted cerebella

Cell counts were performed at 1500× magnification on randomly selected, noncontiguous, 1 μm semithin sections from throughout both ALs and PLs of homozygous and unaffected transplant recipients. The entire section was examined and scored for cerebellar cell types by well-established light microscopy and electron microscopy criteria (Palay and ChalPalay, 1974; Peters et al., 1991; Jones, 1991; Hamre and Goldowitz, 1996). Furthermore, we have independently validated these criteria by correlating ultrastructural, histological and immunocytochemical profiles of individual cerebellar cells as previously detailed (Snyder et al., 1992). A cell was scored as a neuron (i.e., GC) in the IGL by criteria detailed in Fig. 3 legend. Cells not meeting these criteria were scored as non-neuronal. X-gal+ cells were designated as donor-derived, and X-gal-cells as host-derived. Cell-type assignments were independently confirmed by two ‘blinded’ observers.

Individual assessments by light microscopy criteria were performed on 6,746 donor-derived cells and 19,252 host-derived cells in homozygous mutant and unaffected cerebella; in the former, 2,632 were donor-derived and 2,327 were host-derived; in the latter, 4,114 were donor-derived and 16,925 were host-derived. Various ratios were determined as detailed in Results and in Table 1. Statistical analyses to compare mean ratios in ALs with those in the PLs for the same mea mouse used a paired t-test. Comparisons between homozygous and unaffected mice employed a two-sample (unpaired) t-test (Armitage and Berry, 1987). The assumption of unequal variances was made in these calculations because Levene’s test indicated differences in variability between groups (Levene, 1960). Given the multiple comparisons and the indication of unequal variability, a conservative Bon-ferroni two-tailed P value ≤ 0.01 was considered statistically significant. The SPSS for Windows software package was used for all statistical analyses (Release 6.1, SPSS Inc., Chicago, IL).

Table 1.

Three possible models of MEA gene toxicity and the predicted behavior of engrafted exogenous wild-type multipotent neural progenitors in each

Three possible models of MEA gene toxicity and the predicted behavior of engrafted exogenous wild-type multipotent neural progenitors in each
Three possible models of MEA gene toxicity and the predicted behavior of engrafted exogenous wild-type multipotent neural progenitors in each

Integration and differentiation of donor progenitors in mea cerebellum

The mature, untransplanted cerebella of homozygous mea mice displayed the neuroanatomic features well-described for this mutation (Hollander and Waggie, 1977). There was a distinct demarcation from an ostensibly normal-appearing PL to the disorganized cytoarchitectonics of the AL which was characterized most prominently by an absent discrete IGL and a dramatic depletion of mature GC neurons (Fig. 1).

Fig. 1.

Gross appearance of mea cerebellum. Parasagittal section of (A) an unaffected mature mouse cerebellum compared with that from (B) a mea cerebellum. There is dramatic underdevelopment of the AL (arrow) with a deficient IGL. Scale bar in A, 200 μm, applies also to B.

Fig. 1.

Gross appearance of mea cerebellum. Parasagittal section of (A) an unaffected mature mouse cerebellum compared with that from (B) a mea cerebellum. There is dramatic underdevelopment of the AL (arrow) with a deficient IGL. Scale bar in A, 200 μm, applies also to B.

To determine whether the IGL of the mea AL could be reconstituted with normal GCs, the cerebella of homozygous mea (mea/mea) mice and unaffected (+/mea and +/+) litter-mates were transplanted on day of birth with EGL-derived, clonal, multipotent, neural progenitor cells (clone C17.2). Donor-derived cells were densely integrated throughout the cerebella of all mutant and unaffected mice (Fig. 2), particularly in the IGL of the AL and PL, in a manner similar to that previously reported (Snyder et al., 1992). No disruption of cytoarchitecture or tumors were seen.

Fig. 2.

Engrafted progenitors in a mature mea cerebellum. Cells derived from lacZ-expressing progenitors (implanted at birth into the EGL) are visualized under bright field following X-gal histochemistry which confers a blue color to donor-derived cells. (A) Low power overview of a representative parasagittal section through the mature cerebellum. There is dense engraftment in both the AL (arrowhead) and PL. When compared to the unmanipulated mea AL (Fig. 1B), this transplanted AL gives the visual impression of having been partially ‘reconstituted’ by donor cells. (B,C) Progressively higher power views of the AL confirm that a relatively wide IGL (igl) is densely crowded with X-gal+ cells. (Part of this 100 μm thick section was subsequently prepared for electron microscopy, D and Fig. 3). ml, molecular layer; wm, tracts of white matter. (D) A 1 μm semithin section (through the region pictured in B and C prior to electron microscopy) demonstrates that discrete donor-derived (blue) cells possessing definitive GC size, morphology and location (dgc) were integrated within the neuropil of the host AL in the IGL. Dense engraftment was similarly observed throughout the ALs and PLs of all transplanted mutant and unaffected mice. Of note, not only is a donor-derived GC (dgc) evident, with its distinctive blue perinuclear ring, but also a few, rare, unlabeled, host GCs (*). These are discussed in Results and in Figs 3, 4. pc, Purkinje cell. (Insets) Higher magnification of two donor-derived GCs from a lightly counterstained semithin section. Scale bars: (A) 250 μm; (B) 50 μm; (C) 10 μm; (D) 5 μm; insets, 1 μm).

Fig. 2.

Engrafted progenitors in a mature mea cerebellum. Cells derived from lacZ-expressing progenitors (implanted at birth into the EGL) are visualized under bright field following X-gal histochemistry which confers a blue color to donor-derived cells. (A) Low power overview of a representative parasagittal section through the mature cerebellum. There is dense engraftment in both the AL (arrowhead) and PL. When compared to the unmanipulated mea AL (Fig. 1B), this transplanted AL gives the visual impression of having been partially ‘reconstituted’ by donor cells. (B,C) Progressively higher power views of the AL confirm that a relatively wide IGL (igl) is densely crowded with X-gal+ cells. (Part of this 100 μm thick section was subsequently prepared for electron microscopy, D and Fig. 3). ml, molecular layer; wm, tracts of white matter. (D) A 1 μm semithin section (through the region pictured in B and C prior to electron microscopy) demonstrates that discrete donor-derived (blue) cells possessing definitive GC size, morphology and location (dgc) were integrated within the neuropil of the host AL in the IGL. Dense engraftment was similarly observed throughout the ALs and PLs of all transplanted mutant and unaffected mice. Of note, not only is a donor-derived GC (dgc) evident, with its distinctive blue perinuclear ring, but also a few, rare, unlabeled, host GCs (*). These are discussed in Results and in Figs 3, 4. pc, Purkinje cell. (Insets) Higher magnification of two donor-derived GCs from a lightly counterstained semithin section. Scale bars: (A) 250 μm; (B) 50 μm; (C) 10 μm; (D) 5 μm; insets, 1 μm).

Fig. 3.

Ultrastructure of donor-derived GCs in the IGL of the AL of a mature mea mouse engrafted at birth. The blue X-gal precipitate (p), appreciated in the semithin sections (Fig. 2D), is electron dense nondiffusible and reliably distinguishes donor-derived cells from unlabeled host cells. It is visible in the nuclear membrane and occasionally overlying the nucleus (nu), in cytoplasmic organelles (e.g., endoplasmic reticulum, arrowheads), and in processes of donor-derived cells. It respects cellular and organelle membranes. Structures not meeting that rigorous standard are not assessed. (A) This labeled donor-derived GC (dgc) in the mea IGL of the AL (similar to that in Fig. 2D) demonstrated ultrastructural features characteristic of that neuron: small, round or oval soma (5-8 μm diameter), meager cytoplasm forming a rim (often asymmetric with bulges at one or both poles) around a large (>G the cell) round or oval nucleus (nu) and containing a moderate number of mitochondria (m) and short endoplasmic reticulum (arrowhead), sometimes squeezed into a small space within the cytoplasm created by dimpling of the nucleus; blocks of condensed intranuclear chromatin, often distributed along the inner side of the nuclear envelope; a small nucleolus (n), sometimes hidden within one of the chromatin blocks. (B,C) Donor-derived GC dendrites (gd), identified by X-gal precipitate (p) particles in the postsynaptic region (often enclosed in dendritic extensions of endoplasmic reticulum) receive afferent synaptic contact (curved arrows) from host mossy fibers (mf). The IGL synaptic unit is the glomerulus in which incoming mf (containing abundant presynaptic vesicles) envelope and synapse upon gds emanating from multiple surrounding GCs (Snyder et al., 1992). In cross-section, gds appear oval or elliptical, often containing mitochondria and endoplasmic reticulum-derived vacuoles. B,C are high powered views of gds within glomerular complexes near, and contributed to by, dgcs, like that pictured in A. Synaptic contact (curved arrows) by mfs upon these gds is indicated by a clustering of presynaptic vesicles in the mf in apposition to postsynaptic specializations (often appearing like thickened dark bands) in the gd. In B, the gd is in cross-section; in C, a portion of one of the gds is viewed longitudinally. (D) A labeled, donor-derived GC (dgc) is integrated between two unlabeled, residual host GCs (*). Scale bars: (A,D) 1 μm; (B,C) 250 nm).

Fig. 3.

Ultrastructure of donor-derived GCs in the IGL of the AL of a mature mea mouse engrafted at birth. The blue X-gal precipitate (p), appreciated in the semithin sections (Fig. 2D), is electron dense nondiffusible and reliably distinguishes donor-derived cells from unlabeled host cells. It is visible in the nuclear membrane and occasionally overlying the nucleus (nu), in cytoplasmic organelles (e.g., endoplasmic reticulum, arrowheads), and in processes of donor-derived cells. It respects cellular and organelle membranes. Structures not meeting that rigorous standard are not assessed. (A) This labeled donor-derived GC (dgc) in the mea IGL of the AL (similar to that in Fig. 2D) demonstrated ultrastructural features characteristic of that neuron: small, round or oval soma (5-8 μm diameter), meager cytoplasm forming a rim (often asymmetric with bulges at one or both poles) around a large (>G the cell) round or oval nucleus (nu) and containing a moderate number of mitochondria (m) and short endoplasmic reticulum (arrowhead), sometimes squeezed into a small space within the cytoplasm created by dimpling of the nucleus; blocks of condensed intranuclear chromatin, often distributed along the inner side of the nuclear envelope; a small nucleolus (n), sometimes hidden within one of the chromatin blocks. (B,C) Donor-derived GC dendrites (gd), identified by X-gal precipitate (p) particles in the postsynaptic region (often enclosed in dendritic extensions of endoplasmic reticulum) receive afferent synaptic contact (curved arrows) from host mossy fibers (mf). The IGL synaptic unit is the glomerulus in which incoming mf (containing abundant presynaptic vesicles) envelope and synapse upon gds emanating from multiple surrounding GCs (Snyder et al., 1992). In cross-section, gds appear oval or elliptical, often containing mitochondria and endoplasmic reticulum-derived vacuoles. B,C are high powered views of gds within glomerular complexes near, and contributed to by, dgcs, like that pictured in A. Synaptic contact (curved arrows) by mfs upon these gds is indicated by a clustering of presynaptic vesicles in the mf in apposition to postsynaptic specializations (often appearing like thickened dark bands) in the gd. In B, the gd is in cross-section; in C, a portion of one of the gds is viewed longitudinally. (D) A labeled, donor-derived GC (dgc) is integrated between two unlabeled, residual host GCs (*). Scale bars: (A,D) 1 μm; (B,C) 250 nm).

Because the mea neural defect has been previously characterized as a total absence of GCs in the AL (Ross et al., 1990; Napieralski and Eisenman, 1996), our initial analysis focused on detection of any donor-derived cells of that phenotype in the affected region. The AL of homozygous mea transplant recipients was crowded with X-gal+, blue, donor-derived cells, presenting an overall appearance similar to that of the engrafted PL in the same animal. At low magnification, this view created an impression that the IGL of the AL had been ‘reconstituted’ with blue cells (Fig. 2A). Under higher magnification (Fig. 2B,C) and following examination of semithin plastic-embedded sections (Fig. 2D), it was evident that discrete donor-derived cells (blue) were present which possessed definitive GC size, morphology and location, integrated among host (unlabeled) cells within the neuropil of the host AL. It, therefore, appeared that a significant percentage of engrafted progenitors had differentiated into GCs, not only in the IGL of the PL, as expected from our previous reports (Snyder et al., 1992), but also in the IGL of the AL (74.8±0.1% in the AL, discussed in detail below).

Specimens were examined by electron microscopy to confirm the assessment of differentiation state of engrafted progenitors. As seen in 1 μm semithin sections from the IGL of the AL of a mea transplant recipient, the blue X-gal precipitate forms a nuclear ring within donor-derived cells (Fig. 2D), distinguishing them from endogenous GCs, as previously described (Snyder et al., 1992). This precipitate is electron dense permitting donor-derived cells to be reliably identified and distinguished from unlabeled host cells at the ultrastructural level as well. At this level of resolution, one can appreciate that this distinctive reaction product in labeled cells is localized not only to the nuclear membrane (occasionally overlying the nucleus), but also to cytoplasmic organelles such as endoplasmic reticulum. Individual precipitate particles (often appearing crystalline) can be visualized.

Ultrastructurally confirmed, donor-derived GCs were abundant in the IGL of the engrafted mea AL, suggesting that neural progenitors could differentiate into and complement GC deficiency in that region (Fig. 3A). The dendrites of some donor-derived GCs received appropriate synaptic contacts from host mossy fibers (Fig. 3B,C). The field depicted in Fig. 3D, however, makes a surprising additional point, given the classical descriptions of the mea phenotype. A labeled, donor-derived GC neuron, integrated within the host mea neuropil, is adjacent to GCs that are unlabeled and clearly of host origin. Such host GCs are also apparent on semithin sections near donor-derived cells (Fig. 2D).

(Although recent reports have suggested abnormalities, primary or secondary, in other cerebellar cell types (Ross et al., 1990; Fletcher et al., 1991; Napieralski and Eisenman, 1996; Hamre and Goldowitz, 1996), this study was confined to an assessment of GC neurons and their potential replacement. Therefore, although donor progenitors gave rise to both neurons and glia (as in our previous report in normal mice, Snyder et al., 1992), for the purposes of this study, all cells, both donor and host, were classified as either ‘GCs’ or ‘not GCs’.)

Residual GCs in mea AL

The unexpected finding of mea GCs surviving to maturity in the AL prompted us to question whether donor cells had somehow rescued or cross-corrected host cells. Alternatively, and contrary to the classic presumption of total GC absence in the AL, do some mea GCs resist mea gene action? To help address this question, cerebella of untransplanted mature mea mice were examined under electron microscopy.

This survey revealed that some GCs are present even in the unmanipulated adult mea AL. (Fig. 4). These GCs, however, constitute <1% of the very large normal complement. The majority of these residual GCs are heterotopically located about the shaft of the Purkinje cell dendrite suggesting a failure to migrate inward from the EGL to constitute a distinct IGL. Less than 5% of these few residual mea GCs assume an orthotopic position.

Fig. 4.

Residual GCs, viewed ultrastructurally in the AL of an untransplanted, mature mea cerebellum. Most residual GCs are ectopically displaced (GCe) about the shaft of the Purkinje cell (PC) dendrite. However, ~5% of these residual GCs cells do assume an orthotopic position (GC) (as in Fig. 2D). Scale bar: 1 μm.

Fig. 4.

Residual GCs, viewed ultrastructurally in the AL of an untransplanted, mature mea cerebellum. Most residual GCs are ectopically displaced (GCe) about the shaft of the Purkinje cell (PC) dendrite. However, ~5% of these residual GCs cells do assume an orthotopic position (GC) (as in Fig. 2D). Scale bar: 1 μm.

Analysis of engrafted multipotent progenitors might help determine whether mea action is intrinsic or extrinsic to GCs

The fact that the loss of GCs is not absolute and that some GCs apparently escape mea action, while the majority do not, prompted us to speculate on the possible locus of mea gene action. If mea action were intrinsic to the mea GC should not all GCs degenerate or fail to develop? Do some GCs ‘elude’ mea action, suggesting that it might be extrinsic to the GC? The engraftment of exogenous, normal, multipotent cerebellar progenitors, as presented in this study, might help address this issue.

As for most mutations, three possible models for mea action might be envisioned (Table 1). The ability of exogenous multipotent progenitors to differentiate towards ‘replacement’ of degenerated GCs could help distinguish between these three models.

The predicted behavior of neural progenitors in each of these models is summarized in the second column of Table 1. If the deleterious action of mea is intrinsic to mutant cells (model 1), then replacement of GCs with wild-type cells (GCs, or progenitors that can differentiate into GCs) should be observed. If the effects of mea are extrinsic to the mutant cells, i.e., creating an environment inimical to the survival or differentiation of normal GCs (model 2), then donor progenitors should also be affected and GC replacement should not be possible. A third possibility (model 3) is that mea action is extrinsic to the mutant cell but operates only during a particular spatial and/or temporal window of cerebellar development. In that model, one might predict that GC replacement by progenitors might be observed if the cells could somehow be introduced ‘downstream’ of the defect such that they ‘evade’ destruction (i.e., findings similar to those of model 1). Otherwise, observations consistent with model 2 would prevail. (mea action, intrinsic to the mea GC but which compromises the ability of endogenous GCs to survive various normal microenvironmental stresses, would be considered a subtype of model 1; donor cells that possess resistance to a toxic environment (either inherently or following ex vivo genetic engineering) would be considered a subtype of model 3.)

In evaluating these models, it was recognized that the degree to which cell replacement had or had not occurred entailed more than the simple observation of donor-derived GCs in the mea AL. It required quantitative comparison of the degree of progenitor cell differentiation in affected versus unaffected cerebellar regions. These contrasting regions would ideally be drawn not only from homozygous and unaffected mice transplanted at the same time with the same cellular suspension, but also from affected and unaffected regions of the same recipient mouse. The parameters by which engrafted mea cerebella were assessed are summarized in the last three columns of Table 1 and in the text below. The segregation of ratios characterizing the differentiation fate of engrafted multipotent neural progenitors helped distinguish between the three posited models and enabled a first order prediction of the site of mea action.

Comparison of AL with PL of the same mea mouse

By convention, the PL of mea has been viewed as the normal control lobe when assessing AL pathology (Ross et al., 1990; Fletcher et al., 1991; Napieralski and Eisenman, 1996). Although, as discussed below, recent findings from Hamre and Goldowitz (1997) might suggest a need to reassess this classic assumption, because the PL more closely approximates normally developed cytoarchitecture, using it as an internal control for each transplanted mea mutant remains a useful strategy for obtaining a first approximation of the differentiation fate of engrafted progenitors. Therefore, in our first set of analyses, for each transplanted mutant, its PL served as an internal control for its abnormal AL counterpart, transplanted at the same time with the same cellular suspension. We calculated ratios comparing counts of donor-derived cell types in the abnormal AL with those in the more normal PL (Table 1).

In the first ratio (ratio A), the proportion of all donor-derived cells that became GCs in the mea AL was compared with that in the mea PL. For model 1 to hold, the ratio in the AL should be at least comparable to that in the PL; in other words, in the abnormal AL, there should be no extrinsic force thwarting the expected, ‘baseline’ differentiation of progenitors into GCs as observed in the PL. Of the total population of GCs observed in the AL of model 1, however, the percentage that are donor-derived (as opposed to host-derived) (ratio B) should be vastly over-represented when compared to that proportion in the PL. In other words, because the number of host GCs in the AL is so much smaller than that seen in the PLs, it would be expected that the contribution of donor-derived GCs to the total GC population in the AL would be significantly greater than their contribution to the more nearly normal total GC number of the mea PL. Furthermore, that representation of donor-derived GCs in the AL of model 1 should occur in a manner roughly dependent on how many donor progenitors were initially implanted (‘dose-response’). The greater the number of progenitors transplanted into the AL, a proportionately greater number of donor-derived GCs should be observed.

The above-mentioned ratios, and their comparison with the PL, would be reversed if a cell extrinsic defect were broadly operative (model 2) such that all GCs either degenerate prematurely or fail to develop at all. For example, if differentiation were directed away from a GC phenotype, then, of all donor-derived cells, the percentage that were GCs in the AL would be substantially less than the baseline observed in the PL. For model 3, one might expect predictions similar to that for model 1 if donor cells were implanted ‘downstream’ of cell-extrinsic toxicity; if not, then the ratios would approximate those in model 2.

The data (Fig. 5), in fact, support the predictions for model 1 and rule out model 2; they would be consistent with model 3 under only a very narrow set of circumstances and assumptions (discussed below). Data for ratio A (Fig. 5A) demonstrated that the proportion of donor progenitors that differentiated into GCs in the AL of mea mice was not lower than that observed in their PL, suggesting that microenvironmental support for postnatal survival and differentiation was at least comparable in these two regions. In fact, ratio A appeared to be modestly but significantly greater in the AL (P<0.01), suggesting that multipotent progenitors in the AL might be ‘shifting’ their differentiation fate towards compensation for the depleted GCs.

Fig. 5.

Characterization of donor-derived GCs in mea AL compared to those in mea PL and in unaffected AL and PL. (See Results and Table 1 for details.) (A) Comparison of ratio A in ALs of transplanted mea mutants with ratio A in PLs of the same mutant mice. Ratio A was not lower in AL, as might be predicted if a cell-extrinsic defect prevailed there. In fact, it appeared to be modestly but significantly higher in AL than in PL (*P<0.01) (suggesting that differentiation towards a GC phenotype might be favored in AL). (B) Comparison of ratio B in ALs of transplanted mea mutants with ratio B in PLs of the same mutant mice. Ratio B was significantly higher in transplanted mea ALs (*P<0.005). In other words, because the mea AL is so depleted of endogenous GCs, the percentage of GCs that were donor-derived was significantly greater than in PL. (C) Ratio B was calculated in the AL of recipient mea mutants under conditions in which the number of donor progenitors implanted varied. The transplant not only augmented the total number of GCs in the abnormal AL (as suggested in Fig. 5B), but did so in a ‘dose-dependent’ manner, i.e., the more progenitors implanted, the more donorderived GCs were found compared to host GCs. (D) Comparison of ratio B in transplanted mea AL and PL with ratio B in unaffected AL and PL. Donor-derived GCs were represented significantly more in the mea AL than in other engrafted regions (**P<0.001). Interestingly, donor-derived GCs accounted for a greater percentage of GCs even in the PL of mea when compared to an unaffected PL (*P=0.01). Therefore, subtle abnormalities in the mea PL were suggested: it appeared modestly but significantly diminished in GCs. There was no significant difference in ratio B between the lobes of unaffected transplanted mice. (E) Comparison of ratio A in transplanted mea AL and PL with ratio A in unaffected AL and PL. There was significantly more neuronal differentiation in mea AL than in other engrafted regions (**P<0.001). In fact, there was significantly more neuronal differentiation in both mutant lobes: not only was there more in the mea AL compared to unaffected AL and/or PL (**P<0.001), but also more in the mea PL compared to unaffected PL (*P<0.005). These data again suggested that not only mea AL, but mea PL as well, was diminished in GCs. Furthermore, these data in E accentuated the suggestion from Fig. 5A that differentiation towards a neuronal phenotype is favored in the most GC-deficient regions of mea cerebellum. There was no significant difference in engraftment and neuronal differentiation between the lobes of unaffected transplanted mice.

Fig. 5.

Characterization of donor-derived GCs in mea AL compared to those in mea PL and in unaffected AL and PL. (See Results and Table 1 for details.) (A) Comparison of ratio A in ALs of transplanted mea mutants with ratio A in PLs of the same mutant mice. Ratio A was not lower in AL, as might be predicted if a cell-extrinsic defect prevailed there. In fact, it appeared to be modestly but significantly higher in AL than in PL (*P<0.01) (suggesting that differentiation towards a GC phenotype might be favored in AL). (B) Comparison of ratio B in ALs of transplanted mea mutants with ratio B in PLs of the same mutant mice. Ratio B was significantly higher in transplanted mea ALs (*P<0.005). In other words, because the mea AL is so depleted of endogenous GCs, the percentage of GCs that were donor-derived was significantly greater than in PL. (C) Ratio B was calculated in the AL of recipient mea mutants under conditions in which the number of donor progenitors implanted varied. The transplant not only augmented the total number of GCs in the abnormal AL (as suggested in Fig. 5B), but did so in a ‘dose-dependent’ manner, i.e., the more progenitors implanted, the more donorderived GCs were found compared to host GCs. (D) Comparison of ratio B in transplanted mea AL and PL with ratio B in unaffected AL and PL. Donor-derived GCs were represented significantly more in the mea AL than in other engrafted regions (**P<0.001). Interestingly, donor-derived GCs accounted for a greater percentage of GCs even in the PL of mea when compared to an unaffected PL (*P=0.01). Therefore, subtle abnormalities in the mea PL were suggested: it appeared modestly but significantly diminished in GCs. There was no significant difference in ratio B between the lobes of unaffected transplanted mice. (E) Comparison of ratio A in transplanted mea AL and PL with ratio A in unaffected AL and PL. There was significantly more neuronal differentiation in mea AL than in other engrafted regions (**P<0.001). In fact, there was significantly more neuronal differentiation in both mutant lobes: not only was there more in the mea AL compared to unaffected AL and/or PL (**P<0.001), but also more in the mea PL compared to unaffected PL (*P<0.005). These data again suggested that not only mea AL, but mea PL as well, was diminished in GCs. Furthermore, these data in E accentuated the suggestion from Fig. 5A that differentiation towards a neuronal phenotype is favored in the most GC-deficient regions of mea cerebellum. There was no significant difference in engraftment and neuronal differentiation between the lobes of unaffected transplanted mice.

The proportion of all GCs that were donor-derived (X-gal+ GCs) was significantly greater in the AL than in the PL of the same mea recipient (P<0.005) (ratio B, Fig. 5B), and this pro-portion tended to increase if more donor cells had been implanted (Fig. 5C).

Therefore, donor progenitors did appear to help repopulate the IGL of the AL with GCs.

Comparison of ALs and PLs of meander tail mice with those of unaffected mice

Though, as noted above, the PL of mea has conventionally been viewed as the normal control lobe when assessing AL pathology, Hamre and Goldowitz (1997), in an accompanying paper to this report, have suggested from their work with mouse chimeras that the mea PL is not entirely normal. Among other abnormalities, the complement of GCs even in the PL may be subtly reduced below that of wild-type mice. Therefore, while the PL might serve as a good internal control for some ratios within the same recipient mea mouse, it could not be the only control. Engraftment in unaffected littermates, as previously described, was robust and similar to that depicted in Fig. 2 for mutant recipients. Furthermore, overall engraftment was similar in both lobes of unaffected mice with no statistically significant differences.

Additional analyses were performed comparing GC differentiation in the AL, as well as the PL, of mea recipients with that observed in unaffected transplanted littermates. This comparison, in fact, accentuated evidence for GC repletion by progenitor transplantation into the GC-deficient mea AL. The percentage of GCs in the mea AL that were donor-derived (ratio B) was significantly greater than in the AL or PL of unaffected mice. As presented in Fig. 5D, the ratio of donor-derived GCs to host-derived GCs (ratio B) in the AL of engrafted mea mice was highly significantly greater than the ratio seen in any of the three other groups analyzed: (a) the PL of the same homozygous mutants (as per Fig. 5B), as well as (b) the AL and (c) the PL of unaffected littermate cerebella (all P<0.001). Of note, and in support of the contention by Hamre and Goldowitz (1997) that the mea PL is subtly diminished in its complement of GCs compared to an unaffected mouse PL, ratio B in the PL of mea recipients was significantly greater than that ratio in the PL of unaffected transplant recipients (P=0.01). Ratio B did not vary between the AL and PL of a given unaffected transplanted mouse.

The data presented in Fig. 5A had ruled out the hypothesis that progenitor cell differentiation toward a neuronal phenotype was compromised in the mea AL. Ratio A in the AL was not less than that in the mea’s PL. On the contrary, there was a suggestion in that data set (P<0.01) that progenitors might be differentiating in favor of GCs in the mea AL. In Fig. 5E, ratio A in both lobes of the mea was compared with that in unaffected transplanted littermates. The trend which had achieved modest statistical significance in Fig. 5A when comparing mea AL with the PL of the same mea mouse, now became quite marked in this analysis. When comparing the proportion of all donor-derived cells that differentiated into GCs in the AL or PL of mutants with that in the AL or PL of unaffected mice, there was significantly more neuronal differentiation in both mutant lobes (AL, P<0.001; PL, P<0.005). Again, Hamre and Goldowitz’s data (1997) from mouse chimeras concerning subtle abnormalities in the mea PL was confirmed by this data set: the mea PL was modestly but significantly diminished in its own complement of GCs compared to the unaffected PL.

These analyses reinforced the previously stated suggestion that multipotent neural progenitors might be shifting their differentiation fate towards compensation for the depleted neurons in the mutant cerebellar lobes.

Finally, while there was a trend towards rescue of host mea GCs in the transplanted mutants, this result was not significant (P=0.09).

These experiments did not examine the influence of progenitor transplantation on mea mouse behavior. Because even untransplanted mea mutants have only a subtle, mildly discernible ataxia, it was not possible to determine reliably whether there had been any improvement in behavior by the transplants. This study was, therefore, confined solely to cellular analysis.

In summary, these data suggested an apparent repopulation of the mea AL (and, to a more subtle extent, the mea PL) by donor-derived GCs in the newborn cerebellum. Furthermore, there was a suggestion that multipotent neural progenitors might shift their differentiation towards compensation for the deficient GCs in the developing mutant.

DISCUSSION

When clonal, multipotent neural progenitors were implanted into the EGL of newborn mea mice, the GC-deficient AL became populated, by the end of histogenesis, with numerous donor-derived GC neurons, some of which received afferent synapses, suggesting that the IGL of the mea AL had been partially ‘reconstituted’. In fact, robust, cytoarchitecturally appropriate engraftment was evident in the PL as well as the AL. Donor progenitors had the capacity to give rise to both neurons and non-neuronal cells (glia) (as in Snyder et al., 1992). Therefore, this data pool provided an opportunity not only to investigate the feasibility of cell replacement by progenitors in a mouse model of neuronal dysgenesis, but also to begin pursuing possible mechanisms that might underlie this developmental defect by directly comparing the behavior of clonally related progenitors in abnormal versus relatively ‘normal’ regions of the same mutant cerebellum as well as in corresponding normal regions of nonmutant littermates. The data further suggested that the differentiation of multipotent neural progenitors might be influenced by environments deficient in a particular cell type, a concept of potential importance for understanding both mechanisms underlying fate determination during normal CNS development as well as possible future therapeutic interventions involving neural cell replacement.

Distinguishing between three possible models of mea gene action

Table 1 presents three possible models by which a mutant gene, such as mea, might cause neural dysgenesis in the developing CNS. Model 1 holds that the deleterious actions of the mea gene are entirely intrinsic to the GC. Model 2 holds that mea gene action is entirely extrinsic to the GC, persists throughout cerebellar organogenesis, and indirectly creates a milieu (perhaps through interaction with other genes) that compromises this susceptible cell type’s survival and/or differentiation. Model 3 holds that mea action may be extrinsic to the GC but that its effects are not exerted throughout all cerebellar development or regions; certain cells might evade its action and survive to maturity if they are implanted ‘downstream’ temporally, spatially or biochemically of mea’s point of action.

This study focused solely on GC differentiation. The mea cerebellum was regarded as a naturally occurring system in which the differentiation and/or survival of this prototypical neural cell type was regionally compromised. Therefore, all cells, host and donor, were classified as either ‘GC’ or ‘nonGC’. (Although abnormalities in other mea cerebellar cell types have been observed, authors such as Hamre and Goldowitz (1997) conclude that these defects are secondary to GC pathology.)

To summarize our findings according to the categories of Table 1, a multipotent progenitor clone, transplanted into the mea EGL, augmented the total number of GCs in the abnormal AL and did so in an apparently ‘dose’-dependent manner, i.e., the more progenitors implanted, the more donor-derived GCs were present compared to the number of host GCs. Further-more, the percentage of donor-derived cells that became GCs in the AL was no less than that in the PL. The developmental programs of donor-derived GCs, which were well-integrated within the AL neuropil, appeared unperturbed. Therefore, there was no indication that the AL milieu was detrimental to GC differentiation or survival. In fact, of the GCs present in the AL, the overwhelming majority were donor-derived. This was significantly different from the mea PL where donor-derived GCs constituted a much more modest contribution to the larger pre-existing population of host GCs. In other words, because the AL was so depleted of endogenous GCs, the percentage of GCs that were donor-derived was significantly greater than in the more nearly normal PL.

This profile, in which donor progenitors appeared to help repopulate the IGL of the mea AL, ruled out model 2 and supported model 1; it could accommodate model 3 under a very narrow set of circumstances and assumptions. We favor model 1, that the defect in mea is intrinsic to the GC, both from our own data and from the accompanying mouse chimera data of Hamre and Goldowitz (1997).

It is nevertheless instructive from both a developmental and therapeutic point of view, to keep model 3 in mind, since the failure of GCs to develop is not an all-or-none phenomenon. Some mea GCs, albeit a very small percentage, do survive to adulthood in the AL (Figs 2D,3D), as noted also by Hamre and Goldowitz (1997). They have further suggested, contrary to prior assumptions, that the PL is also slightly affected in mea. Our quite different method of analysis confirms the presence of subtle abnormalities in the PL --i.e., that the mea PL is modestly but significantly diminished in its complement of GCs compared to a wild-type PL --reinforcing the idea that GC populations in certain regions of the mea cerebellum are more profoundly affected than in others.

How might this be if, as our two laboratories’ independent methods of analysis would suggest, mea action is entirely cell intrinsic (model 1)? It is the rule, rather than the exception, for an intrinsically acting gene to affect, at a recognizable level, only a subset of the apparent target population, whether in the nervous, hematopoietic, pigmentary or other systems (Green, 1989). Susceptibility is thought to be determined by additional genetic, environmental and developmental influences that act during particular stages and in particular regions to govern cell-cell interactions influencing survival.

With these thoughts in mind, might mea, even if purely cell-autonomous, engender in the GC a variable response to environmental stresses extant at particular developmental stages, or interact with other developmentally regulated and/or spatially restricted genes, such that certain GCs are more susceptible than others to blunted developmental programs or to premature apoptotic degeneration?

Because the mea gene remains uncloned and uncharacterized, one can only speculate how such processes might occur. mea might influence the expression of GC-specific genes, particularly those that are developmentally and/or spatially restricted in their expression. To account for the obvious bias toward GC deficiency in the AL, one could entertain the possibility that mea interacts with genes that determine anterior-posterior polarity within the hindbrain in much the same way as expression of En-1 is controlled by Wnt-1. Mutagenesis of either En-1 or Wnt-1 results in a similar cerebellar dysgenesis, yet expression of the downstream En-1 gene product can rescue the dysgenesis in Wnt-1 null mouse embryos (McMahon et al., 1992; Danielian and McMahon, 1996). Similarly mea may interact with homeodomain gene products that confer apparent regional specificity to the mea mutation; exogenous progenitors which, like primary cerebellar tissue, are known to express genes such as En-1 and En-2 (Snyder et al., 1992), could conceivably rescue the phenotype by providing downstream gene products. Bypassing an upstream defect may also eliminate the ostensible regional specification of the mutation.

Other variables with which even cell-autonomous actions of mea might interact include the developmental stage-specific expression and responsiveness of GCs to neurotrophins (Segal et al., 1992; Gao et al., 1995). mea might influence the expression of receptors or other signal transduction pathways mediating GC responsiveness to trophic factors. The donor progenitors used in these experiments express phosphorylatable neurotrophin receptors (unpublished data); this might permit a response to normal trophic cues within the otherwise affected AL enabling recapitulation of the normal GC developmental program. Conversely, the paucity of mea GCs might result from the spontaneous or induced overexpression of apoptotic proteins in the EGL --perhaps from effective trophin deprivation --promoting elimination of most, but not all, nascent GCs in the developing mea AL.

The scenarios described above might explain how a cell-intrinsic genetic defect, as per model 1, might nevertheless exert a nonuniform influence over a population of mutant cells. Model 3 posits that some GC precursors might be similarly unequally influenced by even a cell-extrinsic process if that process were prevalent at only a particular time or region in the ontogeny of these cells. Fig. 6 illustrates the developmental sequence of GCs as their earliest precursors are first born in the VZ of the IVth ventricle (E13-15), then migrate over the rhombic lip to the nascent EGL (E15-P5) where they proliferate (P0-P8) and ultimately migrate inward to form the IGL (P5-P21). If mea were to create an environmental perturbation inhospitable to GC survival or differentiation only within, for example, the VZ at E14-16, then GC precursors which somehow made the journey from VZ to EGL outside this ‘window’ of gene action might survive. Similarly, if neural progenitors or GC precursors were to be placed directly into the EGL from birth (P0) onward (as in the present experiments) (Fig. 6, arrowhead), they, too, would escape mea action and allow the normal GC developmental program to proceed. (A hypothesis like this is testable with embryonic transplants.)

Fig. 6.

A hypothetical model 3 for mea. See text. Schematic of cerebellar development adapted from Altman (1982). If mea's disruption of GC development – even if extrinsic to the EGL cell – were limited to prenatal events (e.g., in the VZ (E12-E16)), then implanting wild-type progenitors at birth directly into the EGL (arrowhead) would permit the cells to evade destruction and continue differentiation towards GCs. Although the postnatal EGL (focus of present study) is typically implicated in mea pathology, this alternative hypothesis is testable with embryonic transplants.

Fig. 6.

A hypothetical model 3 for mea. See text. Schematic of cerebellar development adapted from Altman (1982). If mea's disruption of GC development – even if extrinsic to the EGL cell – were limited to prenatal events (e.g., in the VZ (E12-E16)), then implanting wild-type progenitors at birth directly into the EGL (arrowhead) would permit the cells to evade destruction and continue differentiation towards GCs. Although the postnatal EGL (focus of present study) is typically implicated in mea pathology, this alternative hypothesis is testable with embryonic transplants.

It is instructive to note, therefore (discussed below), that exogenous progenitors (depending on their mode of implantation or their preparation ex vivo prior to implantation) might be able to compensate for or evade destruction by deleterious gene action whether it be wholly cell-intrinsic or dependent on certain extrinsic factors.

With regard to this thought, the possibility must be acknowledged that the progenitors used in this study may be inherently more tolerant of putative toxic insults in the mea microenvironment than nonpropagated progenitors, in which case a cell nonautonomous event might theoretically be masked. This possibility seems unlikely, however, given that these donor progenitors emulate, but do not outcompete, with host cells in any transplant paradigm examined to date (e.g., Snyder et al., 1992, 1995, 1997; Lacorraza et al., 1996).

Multipotent neural progenitor engraftment may help localize mutant gene action

This study illustrates the use of clonal multipotent neural progenitors to help determine whether degeneration of a given cell type is due to pathology intrinsic or extrinsic to the cell. In some respects, it is simply a refinement of classical complementation experiments. (The robust integration somewhat emulates a ‘postnatal version’ of classic chimeric mouse preparations.) That exogenous nonmutant cerebellar progenitors could survive transplantation into the affected region of the newborn mutant’s cerebellum and differentiate into the deficient neural cell type, suggested that the milieu was not inimical to normal GC development. While arguments have been offered (and the debate continues) that maintaining EGL-derived progenitors in a proliferative state in vitro, whether by an immortalizing gene or by growth factors, broadens their phenotypic potential beyond the fate observed for primary EGL cells (Gao and Hatten, 1993; Snyder, 1994; Jankovski et al., 1996), for the purposes of an experiment such as this, the ability of progenitors to pursue multiple routes of normal differentiation actually conferred an advantage to these cells. It was useful to an understanding of mea action that the donor cells be clonally related and have the ability to pursue not only a normal GC differentiation path but alternative differentiation pathways as well. Had ‘survival as a GC’ or ‘death’ been the only quantifiable outcomes, then we may have missed the opportunity to conclude that mea gene action not only failed to compromise survival of nonmutant GCs, but also failed to perturb the differentiation programs of EGL-derived progenitors. Furthermore, the use of clonally related multipotent donor cells allowed us to explore whether a microenvironment deficient in GCs might somehow influence multipotent cells to ‘shift’ their differentiation to compensate for that deficiency.

Might multipotent progenitors alter their differentiation in the face of specific neural cell-type deficiencies?

In the regions most deficient in GCs, a neuronal phenotype was selected by clonal multipotent progenitors in preference to other potential phenotypes. In other words, in examining the differentiation fate of these donor cells in the very abnormal mea AL, when compared with that in the more nearly normal mea PL (Fig. 5A) and, in more dramatic fashion, when compared with that in normal AL or PL of unaffected littermates (Fig. 5E), there was a suggestion that the microenvironment in regions deficient in a particular neural cell type (in this case GCs) might exert a ‘pressure’ on the differentiation of multipotent progenitors toward repletion of the inadequately developed or degenerating cells. The mechanism by which this ‘compensation’ might be mediated is unknown. It could conceivably involve stimulation via positive trophic signals released by residual support or neighboring cells, or perhaps dysinhibition mediated by the absence of suppressive signals from GCs ordinarily present. Alternatively, there could be a ‘progeny counting mechanism’ intrinsic to progenitors, similar to the internal ‘cell division clock’ previously postulated for O-2A glial progenitors (Temple and Raff, 1986; Wren et al., 1992).

While most studies of neural cells with stem-like features have suggested an ‘instructive’ mechanism for directing differentiation (Johe et al., 1996; Shah et al., 1996), the possibility must also be acknowledged that the mea environment might differentially select for the survival of various stochastically differentiated progeny from C17.2 clones (e.g., GCs). That the total number of donor-derived cells seems to remain constant between mutant and unaffected engrafted areas with variation solely in the proportion of a particular donor-derived cell type with negligible donor cell death, however, speaks against such a selective mechanism.

It has been an axiom of neurobiology that a system’s ontogeny is often best understood by observing its regeneration. The suggestion that multipotent progenitors might differentiate towards a particular phenotype until a requisite ‘quota’ is achieved, is an interesting hypothesis to ponder when considering fundamental neurodevelopmental mechanisms. It may also prove a principle of ultimate therapeutic value.

Strategies for cell replacement by neural progenitors

Our findings suggest that multipotent neural progenitors or stem cells may be capable of the first step required in reconstituting an abnormally developed or degenerated neural system: the replacement of dysfunctional or absent cells.

We and Hamre and Goldowitz (1997) suggest that mea exerts its deleterious action within the mea EGL cell. Alternatively, it is remotely possible from our data that mea action is extrinsic to the EGL cell but that its disruptive impact on GC development is restricted to as-yet-undefined prenatal events. Donor progenitors might have survived because they were implanted directly into the postnatal EGL (‘downstream’ temporally and spatially of where VZ cells crawl over the rhombic lip to become EGL (Fig. 6, arrowhead), or because they provided a molecule biochemically ‘downstream’ of mea action). It is instructive to keep both possibilities in mind when contemplating the broad potential for cell replacement by progenitors. Multipotent progenitor transplantation theoretically might be therapeutic for cell replacement in mutants or models of disease if (a) the defect is intrinsic to the host cell to be replaced; (b) the donor cells are somehow ‘resistant’ to a defect extrinsic to the degenerating host cell (either inherently or following ex vivo genetic manipulation); or (c) the donor cells can be implanted ‘downstream’ of a cell-extrinsic defect. The success of neural cell replacement might be predicated on some inherent biological properties of multipotent progenitors. One of these properties might be the aforementioned ability to ‘shift’ their differentiation towards the repletion of particular cell types that inadequately developed or prematurely degenerated. Another might be the relative ease with which they can be genetically manipulated. The robust expression of lacZ in the mea mutant environment attests to the feasibility of readily engineering these donor cells to express foreign genes. In fact, these cells have been used successfully to express therapeutic genes throughout the brains of both normal and mutant mice (Snyder et al., 1995; Lacorraza et al., 1996; Martinez-Serrano and Björklund, 1996).

Transplantation of multipotent neural progenitors into the AL of neonatal mea mutants could partially replenish the largely absent GC neuronal population. The GC developmental ‘program’ of these well-integrated donor-derived neurons appeared unperturbed. Regarding the etiology for GC deficiency in mea, these observations are most consistent with a cell-intrinsic defect (model 1). We could define, however, a theoretical narrow set of (unlikely) circumstances and assumptions under which a cell-extrinsic action could also be rescued: model 3 (Fig. 6). Our experiments suggest that, for other mutants, transplanting multipotent progenitors may be one strategy for testing hypotheses regarding the pathophysiology underlying a particular mutation and thereby investigating the etiology of some models of neurologic diseases. This study further suggests that progenitors with the potential for multiple fates might differentiate towards the repletion of deficient cell types, a possible developmental mechanism with therapeutic value. Exogenous progenitors (and plausibly an endogenous supply, as well, if appropriately recruited) might play a role in cell replacement for some neurodevelopmental conditions by compensating for or evading destruction by a deleterious gene--certainly in cases where that gene’s action is cell-intrinsic, but also perhaps, under certain circumstances, even if the gene mediates certain extrinsic forces (e.g., model 3). In fact, by using the survival and differentiation of progenitors as an in vivo assay for the mode of action of a mutation, one may also gain insight into the therapeutic options progenitors might enable for that abnormality.

We thank Kristin Hamre and Dan Goldowitz for sharing data with us prior to publication and for stimulating discussions and suggestions. We thank Jason Comander for some statistical analysis. This work was supported by grants to E. Y. S. from NINDS (NS34247), the American Paralysis Association, and the Paralyzed Veterans of America; by grants to R. L. S. from NINDS (NS20820); and by MRRC grant HD18655.

Altman
,
J.
(
1982
).
Morphologic development of the rat cerebellum and some of its mechanisms
.
In The Cerebellum: New Vistas
(ed.
S. L.
Palay
and
V.
Chan-Palay
). pp.
8
49
.,
Heidelberg
:
Springer-Verlag
.
Armitage
,
P.
and
Berry
,
G.
(
1987
).
Statistical Methods in Medical Research
. ed. 2, pp.
104
112
.
Oxford
:
Blackwell Scientific
.
Danielian
,
P. S.
and
McMahon
,
A. P.
(
1996
).
Engrailed-1 as a target of the Wnt-1 signalling pathway in vertebrate midbrain development
.
Nature
383
,
332
334
.
Fletcher
,
C.
,
Norman
,
D. J.
and
Heintz
,
N.
(
1991
).
Genetic mapping of meander tail, a mouse mutation affecting cerebellar development
.
Genomics
9
,
647
655
.
Gage
,
F. H.
,
Ray
,
J.
and
Fisher
,
L. J.
(
1995
).
Isolation, characterization & use of stem cells from the CNS
.
Ann. Rev. Neurosci
.
18
,
159
192
.
Gao
,
W.-Q.
,
Zheng
,
J. L.
and
Karihaloo
,
M.
(
1995
).
Neurotrophin-4/5 (NT 4/5) and brain derived neurotrophic factor (BDNF) act at later stages of cerebellar granule cell differentiation
.
J. Neurosci
.
15
,
2656
2667
.
Gao
,
W.-Q.
and
Hatten
,
M. E.
(
1994
).
Immortalizing oncogenes subvert the establishment of granule cell identity in developing cerebellum
.
Development
120
,
1059
1070
.
Green
,
M.
(
1989
).
Catalogue of mutant genes and polymorphic loci
.
In Genetic Variants and Strains of the Laboratory Mouse
, (ed.
M. E.
Lyon
and
A. G.
Searle
). 2nd. ed., pp.
12
403
.
Oxford
:
Oxford Univeristy Press
.
Hamre
,
K. M.
and
Goldowitz
,
D.
(
1996
).
Analysis of gene action in the meander tail mutant mouse: examination of cerebellar phenotype and mitotic activity of granule cell neuroblasts
.
J. Comp. Neurol
.
368
,
304
315
.
Hamre
,
K. M.
and
Goldowitz
,
D.
(
1997
).
Meander tail acts intrinsic to granule cell precursors to disrupt cerebellar development: analysis of meander tail chimeric mice
.
Development
124
,
4201
4212
.
Hollander
,
W. F.
and
Waggie
,
K. S.
(
1977
).
Meander tail : a recessive mutant located on chromosome 4 of the mouse
.
J. Hered
.
68
,
403
406
.
Jankovski
,
A.
,
Rossi
,
F.
and
Sotelo
,
C.
(
1996
).
Neuronal precursors in the postnatal mouse cerebellum are fully committed cells: evidence from heterochronic transplantations
.
Eur. J. Neurosci
.
8
,
2308
2319
.
Johe
,
K.
,
Hazel
,
T.
,
Muller
,
T.
,
Dugich-Djordjevic
,
M.
and
McKay
R.
(
1996
).
Single factors direct the differentiation of stem cells from the fetal and adult central nervous system
.
Genes Dev
.
10
,
3129
3140
.
Jones
,
E. G.
(
1991
).
The Structural Basis Of Neurobiology
.
New York
:
Elsevier
.
Kilpatrick
,
T.
,
Richards
,
L. J.
and
Bartlett
,
P. F.
(
1995
).
The regulation of neural precursor cells within the mammalian brain
.
Molec. Cell. Neurosci
.
6
,
2
15
.
Lacorazza
,
H. D.
,
Flax
,
J. D.
,
Snyder
,
E. Y.
and
Jendoubi
,
M.
(
1996
).
Expression of human β-hexosaminidase ?-subunit gene (the gene defect of Tay-Sachs disease) in mouse brains upon engraftment of transduced progenitor cells
.
Nature Medicine
4
,
424
429
.
Levene
,
H.
(
1960
).
Robust tests for equality of variances
.
In
(ed.
I.
Olkin
) Contributions to Probablity and Statistics. pp.
278
292
.
Palo Alto
:
Stanford University Press
.
Martinez-Serrano
,
A.
and
Björklund
,
A.
(
1996
).
Gene transfer to the mammalian brain using neural stem cells: a focus on trophic factors, neuroregeneration and cholinergic neuron system
.
Clin. Neurosci
.
3
,
301
309
.
McKay
,
R. D. G.
(
1997
).
Stem cells in the central nervous system
.
Science
276
,
66
71
.
McMahon
,
A. P.
,
Joyner
,
A. L.
,
Bradley
,
A.
and
McMahon
,
J. A.
(
1992
).
The midbrain-hindbrain phenotype of Wnt-1/Wnt-1 mice results from stepwise deletion of engrailed expressing cells by 9. 5 days postcoitum
.
Cell
69
,
581
595
.
Napieralski
,
J. A.
and
Eisenman
,
L. M.
(
1996
).
Further evidence for a unique developmental compartment in the cerebellum of the meander tail mutant mouse as revealed by the quantitative analysis of Purkinje cells?
J. Comp. Neurol
.
364
,
718
728
.
Palay
,
S. L.
and
Chan-Palay
,
V.
(
1974
).
Cerebellar Cortex, Cytology and Organization
.
Heidelberg
:
Springer-Verlag
.
Peters
,
A.
,
Palay
,
S. L.
and
Webster
,
H. D.
(
1991
).
The Fine Structure Of The Nervous System, Neurons And Their Supporting Cells, ed. 3
.
Oxford
:
Oxford University Press
.
Ross
,
M. E.
,
Fletcher
,
C.
,
Mason
,
C. A.
and
Hatten
,
M. E.
(
1990
).
Meander tail reveals a discrete developmental unit in the mouse cerebellum
.
Proc. Natl. Acad. Sci. USA
87
,
4189
92
.
Segal
,
R. A.
,
Takahashi
,
H.
and
McKay
,
R. D. G.
(
1992
).
Changes in neurotrophin responsiveness during the development of cerebellar granule neurons
.
Neuron
9
,
1041
1052
.
Shah
,
N. M.
,
Groves
,
A. K.
and
Anderson
,
D. J.
(
1996
).
Alternative neural crest cell fates are instructively promoted by TGFβ superfamily members
.
Cell
85
,
331
343
.
Snyder
,
E. Y.
(
1994
).
Grafting immortalized neurons to CNS
.
Curr. Opin. Neurobiol
.
4
,
742
751
.
Snyder
,
E. Y.
,
Deitcher
,
D. L.
,
Walsh
,
C.
,
Arnold-Aldea
,
S.
,
Hartweig
,
E. A.
and
Cepko
,
C.
(
1992
).
Multipotent neural cell lines can engraft and participate in development of mouse cerebellum
.
Cell
68
,
33
51
.
Snyder
,
E. Y.
,
Taylor
,
R. M.
and
Wolfe
,
J. H.
(
1995
).
Neural progenitor cell engraftment corrects lysosomal storage throughout the MPS VII mouse brain
.
Nature
374
,
367
370
.
Snyder
,
E. Y.
,
Yoon
,
C.
,
Flax
,
J. D.
and
Macklis
,
J. D.
(
1997
).
Multipotent neural precursors can differentiate toward replacement of neurons undergoing targeted apoptotic degeneration in adult mouse neocortex
.
Proc. Natl. Acad. Sci. USA94
(in press).
Temple
,
S.
and
Raff
,
M. C.
(
1986
).
Clonal analysis of oligodendrocyte development in culture: evidence for a developmental clock that counts cell divisions
.
Cell
44
,
773
779
.
Weiss
,
S.
,
Reynolds
,
B. A.
,
Vescovi
,
A. L.
,
Morshead
,
C.
,
Craig
,
C.
and
van der Kooy
,
D.
(
1996
).
Is there a neural stem cell in the mammalian forebrain
.
Trends Neurosci
.
19
,
387
393
.
Wren
,
D.
,
Wolswijk
,
G.
and
Noble
,
M.
(
1992
).
In vitro analysis of origin and maintenance of O-2Aadult progenitor cells
,
J. Cell Biol
.
116
,
167
176
.