Interaction of granule, Purkinje and inferior olivary neurons in Lurcher chimaeric mice : I.Qualitative studies

For the past twenty years, neurological mutant mice have added major insights to our understanding of the mammalian nervous system. During development of the nervous system, numerous events occur simultaneously, and genetic mutations are a useful tool for analysing these events individually. The importance of cell-cell interactions and the specificity of synaptic connections have been revealed through the use of these mutant mice (see reviews by Sidman, 1972 and Rakic, 1976). The heterozygous lurcher (gene symbol: Lc) is one such mutant. Affected animals possess a rocking gait and impaired balance (Phillips, 1960). These behavioural abnormalities are associated with the loss of certain neurons in the central nervous system. Virtually all of the cerebellar Purkinje cells (PCs) degenerate between the second and twelfth weeks of postnatal life (Caddy & Biscoe, 1979). The major afferents to the PCs are also affected. The number of granule cells in the cerebellar cortex is decreased by 90 % (Swisher & Wilson, 1977). In the ventral brain stem, 75 % of the neurons of the inferior olivary nucleus degenerate by the twelfth postnatal week (Caddy & Biscoe, 1979). The role of the Lc gene in this pathology remains elusive. All genes are physically present in all cells, but only a specific subset of genes are expressed in any given cell type. The cell or cells in which the mutant gene is expressed is referred to as the primary site(s) of gene action (Mullen, 1977 b). These cells may or may not be morphologically affected by the intrinsic action of the mutation. Other cells, which possess the mutant gene but do not express it, can be affected indirectly as a consequence of a critical interaction with those cells which are the primary site of gene action. Although in lurcher we know that the PCs, granule cells, and olivary neurons degenerate, this does not reveal which, if any, of these cells is the primary site of gene action. For example, the Lc gene could act intrinsically within all three cell types to cause their death. It is also possible, however, that these neurons die as a secondary consequence of the Lc gene being expressed in some other cell type on which they all depend. A third hypothesis is that the Lc gene acts intrinsically within only one cell type, the PC, whereas the granule and olivary neurons are affected indirectly by the gene. This explanation was suggested by Caddy & Biscoe (1979) on the basis of their morphological observations of the mutant. In the present study, we have examined this hypothesis by producing lurcher ↔wild-type chimaeric mice.

Chimaeras are formed by aggregation of pairs of embryos to produce a single mouse. If one of the embryos is genotypically lurcher and the other wild-type, then mutant and normal cells contribute to and interact during the formation of a single organism. The nervous system of a chimaera is a fine-grained mosaic of cells each of which is descended from one of the original embryos. If the Lc gene acts intrinsically on a given cell type, then all and only those cells with the mutant gene will have the mutant phenotype. To make this correlation, we must be able to determine a cell’s genotype (mutant or wild-type) independently from its phenotype (i.e. whether the cell is present or absent). This is done by making chimaeras from two embryos each of which is homozygous for a different allele at the β-glucuronidase locus. This difference can be detected histochemically, and, thus, can be used to identify the genotype of each cell (Condamine, Custer & Mintz, 1971 ; Mullen & Herrup, 1979). The accuracy of using β-glucuronidase activity as an independent cell marker has been amply demonstrated. The staining pattern of PCs using β-glucuronidase (Mullen, 1977a) is similar to the pattern seen using either differences in β-galactosidase activity (Dewey, Gervais & Mintz, 1976) or antigenic differences between isozymes of glucosephosphate isomerase (Oster-Granite & Gearhart, 1981). In addition, β-glucuronidase activity correlated exactly with the PC phenotype in Purkinje cell degeneration (Mullen, 1977a) and staggerer chimaeras (Herrup & Mullen, 1979). Using β-glucuronidase histochemistry, we report here that the degeneration of the PCs in lurcher is an intrinsic defect while the loss of inferior olivary neurons is an extrinsic event.

Chimaeras were produced by a modification (Mullen & Whitten, 1971) of the 8-cell embryo aggregation method of Mintz (1962, 1965) and Tarkowski (1961). Heterozygous lurcher mice were obtained from the Children’s Hospital Medical Center (Boston) where they had been outcrossed. As a result some of these animals carried the agouti gene (gene symbol: A). Males from this colony were mated to C57BL/6J females. The embryos obtained from this mating were homozygous for the high-activity, Gus^b, allele at the β-glucuronidase locus, wildtype at the retinal degeneration (gene symbol: rd) locus, and were either +/ + or + /Lc. Inbred C3H/HeJ mice were obtained from the Jackson Laboratories (Bar Harbor). These embryos were A/A, rd/rd, homozygous for the low-activity, Gus^h, allele of β-glucuronidase, and wild-type at the Lc locus. Pairs of these embryos were cultured together overnight and then transplanted into the uteri of pseudopregnant females. The chimaeras finished development and were born normal in size.

All animals were prepared for histology as described (Mullen, 1977 a). Briefly they were perfused transcardially with cold 4 % paraformaldehyde in phosphate buffer, embedded in polyester wax, and serially sectioned in the sagittal plane. Certain sections were stained for β-glucuronidase activity by the method of Hayashi, Nakajima & Fishman (1964) with the modifications of Feder (1976). This procedure stains Gus^b/Gus^b cells with a red chromophore which appears black in the figures. The amount of background staining can vary depending on the time of incubation, the quality of the substrate, etc. Unstained Gus^h/Gus^h cells appear green due to the methyl green counterstain. Although these cells are easy to identify in colour, they contrast poorly with the background in the black-and-white photographs.

After reaching maturity, the chimaeras were each mated to a C57BL/6J mouse. Four chimaeras (designated χ11, χ7, χ13, and χ12) produced progeny which exhibited the lurcher behavioural phenotype. This confirmed that the embryos from the + /Lc × + / + mating carried the Lc gene. Hence, the genotype of three of the four mice was + /Lc a/a + Gus^b/ + Gus^b ↔+/ + A/A rd Gus^h/rd Gus^h. The genotype of the fourth chimaera is slightly different, as explained below. Despite the presence of the Lc gene, all of the chimaeras behaved normally. The age of the chimaeras at the time of sacrifice is presented in Table 1. Note that all of the chimaeras were sacrificed after the twelfth postnatal week, which is when the neuronal degeneration is complete in +/Lc animals (Caddy & Biscoe, 1979). Yet the chimaeras never displayed any behavioural abnormalities. One chimaera (χ12) is now 16 months old and has not yet been sacrificed. We plan to sacrifice this animal at an advanced age in order to study long-term changes.

Three chimaeras displayed coat mosaicism, and the percentages are given in Table 1. The fourth (χ 11) was 100% agouti and was not noticeably ataxic, yet he produced agouti, + /Lc offspring when mated to a C57BL/6J female. This animal was one of our first chimaeras, and he was produced before the A gene was eliminated from our lurcher colony. His genotype, therefore, differs from the other chimaeras in that he is A /a instead of a/a in the + /Lc component. Thus both cell genotypes carried the A gene, and so there was no mosaicism in the coat of χ11. In addition to coat mosaicism, the percentage of chimaerism was also estimated (to the nearest 10%) by a visual examination of the retina and the liver. The photoreceptor degeneration seen in rd/rd mice is expressed in chimaeric mice and is probably due to the intrinsic action of the rd gene (Wegmann, Lavail & Sidman, 1971). Little or no degeneration was observed in the retina of χ 11 suggesting the retina was largely + / + at the rd locus hence + /Lc. Liver sections were stained for β-glucuronidase activity, but no Gus^h/Gus^h cells were seen in the χ11 liver. Approximately 80 % and 30 % of the hepatocytes of χ l3 and χ 7, respectively, were Gus^h/Gus^h(Table 1) and, hence, were descended from the C3H/HeJ embryo. Despite the lack of mosaicism in the rest of the body of χ 11, wild-type cells were present in the central nervous system. Approximately 10 % of the cells of the posterior choroid plexus and various brainstem nuclei were Gus^h/Gus^h. The percentages for the other chimaeras are given in Table 1.

Upon dissection of the brains, the cerebella of the chimaeras were examined and found to be intermediate or normal in size compared to + /Lc and wildtype cerebella. This is illustrated by the sagittal sections in Fig. 1. The cortices of the chimaeric cerebella were trilaminar structures as were the + /Lc and wildtype cerebella. However, the granule cell layers and the molecular layers of χ 11 and χ 7 were intermediate in cross-sectional area. In χ 13, these layers had approximately the same area as the wild type. It is readily apparent from the reduced cerebellar size that the presence of + /Lc cells affects cerebellar development in the chimaeras.

The qualitative appearance of the PC layers was similar in all three chimaeras, and the PCs which were present appeared morphologically normal. However, there were large gaps in the PC layer, indicating a decreased number of PCs as compared to wild type. χ 11 possessed very few PCs, while χl3 had the greatest number of PCs (approximately 60% as many as wild type). For each chimaera, the number of PCs corresponded approximately with the percentage of Gus^h/ Gus^h cells in the rest of the central nervous system (Table 1). All of the PCs present were unstained by the β-glucuronidase reaction (Fig. 2). Hence, they were all Gus^h/Gus^hcells, descended from the wild-type embryo. Since the number of PCs was reduced in all chimaeras and since there were no Gus^b/Gus^b PCs observed in any of the chimaeras, we conclude that all + /Lc PCs degenerated. Thus, the Lc gene acts intrinsically on the PCs to cause their degeneration.

The inferior olivary nuclei of the chimaeras are shown in Fig. 3. The nuclei of χ11 and χ 13 appeared to be approximately normal in volume, although the density of cells was lower than in wild type. The nucleus of χ 7 was the most severely affected compared to the other chimaeras. The cross-sectional area appeared somewhat smaller than wild type, and the density of cells was much lower than in the other two chimaeras. The decreased density of neurons suggests that there may have been some loss of olivary neurons in the chimaeras, but the amount of loss did not seem to correlate with the percentage of + /Lc neurons in the rest of the central nervous system. It is important to point out, however, that considering both density and volume, the number of neurons was clearly greater in all of the chimaeras than in the + /Lc animal (Fig. 3A). When stained for β-glucuronidase activity, both Gus^b/Gus^b and Gus^h/Gus^h cells were seen (Fig. 4). If the degeneration of the + /Lc olivary neurons was intrinsic, one would expect to see an increased proportion of +/+ Gus^h/Gus^h cells. For example, if initially 90% of the olivary neurons in χ 11 were + /Lc, then after 75% of these cells degenerate, the percentage of +/+ Gus^h/Gus^h cells would have increased from 10 to 31 %. Preliminary counts¹ indicated that the actual percentage was 22%. Similarly, assuming again that the initial percentage of wild type olivary neurons was the same as the observed percentage of wild type PCs, one would expect 63 and 86% of the olivary neurons in χ 7 and χ 13, respectively, to have been +/+ Gus^h/Gus^h. In fact, 18 and 61% of the cells were observed to be +/+ Gus^h/Gus^h. For each of the three chimaeras, the actual proportion of wild-type cells was lower than expected if the degeneration of + /Lc olivary neurons were intrinsic. Further, the ratio of + /Lc to +/ + olivary neurons was similar to the ratio seen in the rest of the central nervous system (Table 1, lines 5 – 7). This suggests that if there was any cell loss, both + /Lc and + / + cells were equally affected. Furthermore, a substantial majority of the olivary neurons in χ11 were Gus^b/Gus^b, hence + /Lc in genotype. Since the number of olivary neurons was much greater in the chimaera than in the +/Lc animal, many + /Lc neurons survived in which would not have survived in a + /Lc animal. Therefore, we conclude that the death of the inferior olivary neurons in the + /Lc mouse is secondary to some other defect.

Our results indicate that the Lc gene acts intrinsically on the PCs to cause their degeneration. In chimaeras which do not possess the Lc gene, both Gus^h/Gus^h and Gus^b/Gus^b PCs are present (Fig. 2A). In contrast, the + /Lc Gus^b/Gus^b PCs of the lurcher chimaeras were totally absent. None was rescued by the presence of up to 60% of the wild-type number of PCs. The degeneration of PCs was observed in the chimaeras to correlate with the presence of the Lc gene. Thus, the PC is a primary site of Lc gene action. We cannot be certain whether any of the PCs which did not possess the Lc gene (the Gus^h/Gus^h cells) also degenerated. The number present corresponded roughly with the proportion of wild-type cells in the rest of the central nervous system, however, and this strongly suggests that there was no degeneration of wild-type PCs. The intrinsic action of the Lc mutation on the PCs is analogous to the conclusion of Mullen (1977 a), who studied chimaeras of another PC degeneration mutant, pcd.

Gus^h/Gus^h↔ Gus^b/Gus^b chimaeras display mosaicism in the inferior olivary nucleus (Figure 4A). The presence of the Lc gene does not alter this mosaicism. If the degeneration of olivary neurons was intrinsic to + /Lc cells, then these Cells would have also degenerated in the chimaera and the ratio of + / + to + /Lc cells would have been greatly increased. This was not observed. Hence, the results suggest that the degeneration of olivary neurons is secondary to some other defect. The loss of the PCs may be responsible, since they are the major post-synaptic target of the olivary neurons. Since the chimaeras possessed more olivary neurons than did + /Lc animals (Fig. 3), the question arises how many olivary neurons are rescued by a given number of PCs? This issue of plasticity is now being investigated by counting the number of PCs and olivary neurons in each of the chimaeras.

β-Glucuronidase is a cytoplasmic enzyme, and granule cells have very little cytoplasm, so it cannot be used to accurately identify the genotype of the granule cells (Mullen & Herrup, 1979). Since the identification of the PC as a primary site of gene action does not rule out the possibility that the Lc gene also acts intrinsically within other cell types, no conclusion concerning the granule cells can be drawn at this time. That the granule cell death is possibly an extrinsic phenomenon can be inferred from the findings in another cerebellar mutant, staggerer. In this animal the death of the granule cells is believed to be due to the lack of spines on the PC dendrites, which prevents the developing granule cells from synapsing with their major post-synaptic target (Sotelo & Changeux, 1974; Landis & Sidman, 1978). In lurcher, the granule cells may not be able to form synapses with the PCs because the latter begin to degenerate during the period of synaptogenesis. So although the extrinsic nature of the granule cell death has not yet been demonstrated in lurcher, a similar mechanism may be involved.

The survival of neurons in many systems has been shown to depend on the presence of a target (Cowan, 1973; Purves & Lichtman, 1980). Which neurons survive may be determined by the stabilization of their synapses with the target organ (Changeux & Danchin, 1976). In a similar way, PCs seem to be important to the normal development of the granule cells and the inferior olivary neurons. In contrast, the PCs may be less important to the development of their post-synaptic target, since the number of neurons in the deep cerebellar nuclei are unaffected in lurcher (Caddy & Briscoe, 1979).

We would like to thank Mr Thomas Diglio and Ms Sandra Wilczynski for their technical assistance. This work was supported by a Basil O’Connor Starter Grant from the March of Dimes and by grant HD 12213 from the NIH.