Journal of Neuroscience Research 27:65-70 (1990)
Lurcher Purkinje Cells Express Glutamic Acid Decarboxylase and Calbindin mRNAs C.W. Wuenschell, A. Messer, and A.J. Tobin Department of Biology (C.W.W., A.J.T.), Molecular Biology Institute (A.J.T.), and Brain Research Institute (A.J.T.),University of California, Los Angcles: Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health Scicnccs, SUNY, Albany ( A . M . ) Purkinje neurons in immature Lurcher (Lcl + ) mice are destined to die as a result of a defect intrinsic to the dying cells. We have used in situ hybridization to determine whether the Lc allele interferes with the normal program of gene expression in the doomed Purkinje cells. In P21 mice, degeneration of Purkinje cells is well underway, but the surviving Purkinje cells continue to express the mRNAs for both glutamate decarboxylase and calbindin DZsK,two proteins whose expression is characteristic of normal Purkinje neurons. We conclude that the Le allele probably does not interfere with the developmental program but acts to cause cell death in already differentiated Purkinje neurons. Key words: in situ hybridization, cytodifferentiation, neurodegeneration INTRODUCTION The unfolding of the developmental program of an organism involves a series of interactions between events internal to individual cells and influences arising from sources external to those cells, The relative contributions of intrinsic and cxtrinsic influences to the cytodifferentiation of specific types of vertebrate neurons remains an open question. The role of external cues in neuronal differentiation can be investigated by methods such as transplantation and tissuc culturc, which experimentally alter the external environments of cells. The internal genetic program of a cell is much less accessible to experimentation. Mutations that are known to exert their effects only on specific cell types, however, [nay offer insight into the genetic programs of the affected cells. One such mutation, the Lurcher mutation in the mouse, specifically affects the Purkinje neurons of the cerebellum. The mouse cerebellum is an excellent region in which to study gene expression in a specific neuron type. It has a well-defined cytoarchitecture and has been extensively investigated both functionally and developmentally. Lurclzer (Lc) is an autosoma] dominant, charactcr0 1990 Wiley-Liss, Inc.
ized by a mild-to-moderate ataxic "lurching" gait (Phillips. 1960). Lxi animals show degeneration of almost 100% of their Purkinje cells during postnatal weeks 2-5, followed rapidly by the loss of nearly 905%of the cerebellar granule cells (Caddy et al., 1982). By analyzing chimeric mice, Wetts and Herrup (1982) have demonstrated that the Lc gene acts directly in the Purkinje cells. Furthermore, Lc granule cells, which normally degeneratc after thc Purkinje cells, survive in the presence of neighboring wild-type Purkinje cells in Lc-wildtype chimeras (Wetts and Herrup, 1983). We have undertaken to study the effects of the /,urchrr mutation on the genetic program of cerebellar Purkin.je neurons by means of in situ hybridization with recombinant probes for three cerebellar mRNAs. Two of thcse mKNAs, thosc cncoding glutamic acid decarboxylase (GAD) and the calcium binding protein calbindin DZHK,are normally expressed in Purkinje neurons (Wuenschell et al.. 1986; Baimbridge and Miller, 1982, Wuenschell and Tobin, 1988). The third InRNA, encoding proenkephalin mRNA, is not expressed in Purkinje neurons, but is expressed in Golgi I1 neurons (Sar et a]., 1978; Wuenschell and Tobin, 1988). In situ hybridization permits the detection of specific mRNAs at a single cell level of resolution. Since the half-lives of mRNAs are generally shorter than those of the corresponding proteins, this technique potentially gives a more direct assessment of transcriptional status than does irnmuno h i stoc hem istry.
RESULTS By the end of the 3rd postnatal week (P21), the Lc allele has had obvious effects, and the degeneration of
Kcceived November 29, IY8Y; revised April 17, 1990; accepled April 19, 1990. Address reprint requesla to A.J. Tobin, Department of Biology. U n versily ol California, I m Angela. CA 90024-1606. C . W . Wucnschcll is now at Division of Biology, California Institute of Technology, Pasadena, CA 91 125.
66
Wuenschell et al.
Fig. 1 . GAD mRNA in the Lirrchrr cerebellum. A: Darkfield photomicrograph of a coronal section of P2 I Lwcher cerebellum hybridized with GAD antisense R N A . 1,ayers: 111 = molccular laycr: g = granular layer; w = white matter. Large straight arrow indicates labeled cell at the boundary bctwcen the molecular and granular layers. Curvcd arrow indicates labeled cell in the granular laycr. Small straight arrows indicate
labeling in the molecular laycr. Bar = 200 kin.B: Brightfield photomicrograph of the same section as in A . Arrows indicate labeled cclls: Large arrows = Purkinje cells: curved arrow = Golgi 11 ccll; small arrows = stellate or basket cells. C: Brightfield photomicrograph of il control section of P2 I Ltcrclrcr- cerebellum h y b r i d i d wilh GAD scnse RNA. Arrow indicates a Purkinje cell. Bar = 50 pm.
Purkinje cells is well underway. At this time. however, cnough Purkinje cells remain to allow studies of their rnRNAs by in situ hybridization. Hybridization of GAD antisense RNA to sections of P21 Lc cerebellum revealed labeled cells sparsely
Scattered along thc boundary between the molecular and granular layers (Fig. IA,B). Labeled cells were also scattered in the granular layer and at all depths of the molecular layer. This distribution is consistent with normal expression of the GAD gene in all of the GABA-
Lurcher Yurkinje Cells
Pig. 2. mRNAs for GAD, calbindin, and proenkephalin in the Lurcher cerebellum. Darkfield photomicrographs of sections of P2 I Lurchrr cerebellum hybridized with GAD antisense RNA (A), calbindin antisense RNA (B), proenkephalin antiscnse RNA (C), or calbindin sense RNA (control. D). A: Large arrow indicates labeled cell at the boundary between the
67
molecular and granular layers. Curved arrow indicates labeled
cell in the granular layer. Small arrows indicate labeling in the molecular laycr. B: Large arrow indicates labeled cell at thc boundary between the rnolecular and granular layers. C: Curvcd arrow indicates labeled cell i n thc granular layer. Sinall arrows indicate labeling in the deep iiiolecular layer. D: Large arrow indicates boundary between the molecular and granular layers. Bar = 0.5 mm.
ergic neuron types of the cerebellum, including the remaining Purkinje cells. No specific labeling occurred in control sections hybridized with GAD sense RNA (Fig. 1C). In order to further verify the identify of the Purkinje cells, and to study the expression of additional rnKNAs, we obtained molecular probcs for the niRNAs that encode calbindin DZgKand proenkephalin (Frantz et al., 1986; Woodet al., 1988; Yoshikawact al.. 1984). In wild-type cerebella, calbindin mRNA is expressed only in Purkinje cells, and proenkephalin i s expressed in Golgi I1 cells and a population of small cells deep in the molccular layer (Wuenschell and Tobin, 1988). When neighboring sections from the cerebellum of a P21 Lurcher mouse were hybridized with intise,,se KNA for GAD, calbindin, or proenkephalin, clear differences were apparent (Figs. 2, 3 ) . As expected. the
GAD hybridization probc (Figs. 2A, 3A,B) yielded the pattern described above for the previous experiment. The calbindin probe labeled cells located only at the boundary of the molecular and granular layers (Fig. 2B). These large cells had the appearance of Purkinje cells (Fig. 3C) and were widely and randomly spaced often with long intervening gaps. The proenkephalin probc labeled cells scattered in the granular layer and small cells both in the Purkinje layer and in the deep molecular layer (Figs. 2C, 3E,F). Control sections hybridized with sense strand calbindin R N A showed no specific labeling (Figs. 2D,3D). We did not observe unlabeled cells that could be clearly identified as Purkinje neurons in sections hybridized with GAD or calbindin probes. The total numbers of surviving Purkinje cells in the Lureher sections was, however, not great, and we cannot exclude the possibility that unlabeled Purkinjc cells might exist.
68
Wuensehell et al.
Fig. 3. mRNAs for GAD, calbindin, and proenkephalin in the Lurcher cerebellum. Brightfield photomicrographs of scctions of P21 Lurcher ccrebellum hybridized with GAD antisense RNA (A,B), calbindin antisense RNA (C), calbindin sense RNA (control, D), or proenkephalin antisense KNA (E,F). A: Large arrow indicates labeled Purkin.je cell. B: Small arrow indicates small labeled cell in the molecular layer (probably a basket cell based on position). C: Large arrows indicate labeled
Purkinje cclls. D: Large arrow indicates an unlabeled Purkinje cell. (ni = molecular layer; g = granular). E: Curved arrow indicates labeled Golgi 11 cell. Small arrows indicate sniall labeled cells at the edge of the molecular layer tentatively identified as basket cells. F: Large arrow indicates unlabeled Purkinje cell. Curved arrow indicatcs labeled Golgi I1 cell. (w = white matter). Bar = SO b"m.
DISCUSSION
Purkinje layer that were intact enough to be identified as probable Purkinje cells were labeled by both of these probes, but we may not have been able to identify cells in the later stages of degeneration. We draw three provisional conclusions from these findings:
Purkinje cells are massively dying in the Lc cerebellum, yet virtually all recognizable Purkinje cells are expressing clearly detectable levels of GAD and calbindin mRNAs. Despite the ongoing degeneration, the remaining Purkinje cells are clearly identifiable by their size, shape, and position. The only possible cell type that could be confused with Purkinje cells would be ectopic Golgi I1 cells. Such cells, however, would be likely to express proenkephalin mKNA, so they do not confound our conclusions (Wuenschell and Tobin, 1988). Lc Purkinje cells thus express the genes for GAD and calbindin-at least at the mRNA level-before they degenerate. We do not know, however, whether Purkinje cells that are actually in the process of degeneration contain these mRNAs. The majority of the large cells in the
1. The genetic program of Purkinje cells is relatively stable once it is established. The GAD and calbindin genes, whose expression is characteristic of normal differentiated Purkinje neurons, continue to be expressed in Lurcher Purkinje cclls even though these cells are destined to die. 2. The Lurchcr mutation probably causes Purkinje cell death by a fairly direct or restricted path of action. If this mutation caused Purkinje cells to die by grossly disrupting the cells' program of differentiated func-
Lurcher Purkinje Cells
tions, it is unlikely that expression of genes characteristic of the differentiated phenotype would persist. 3. The i,urcher mutant gene appears to act later in time than the activation of GAD and calbindin gene expression. The available evidence indicates that the expression of GAD and calbindin genes begins early in development. At least some migrating Purkinje cells at embryonic day (E) 16 in the rat (Legrand, 1983; Wassef et al., 1985) contain immunoreactive calbindin. Furthermore, cerebellar primordia of E l 5 mouse embryos contain both GAD and calbindin mRNAs (G. Frantz and A. Tobin, unpublished results). The hybridizing cells lie above the fourth ventricle, in the region that contains precursors of Purkinje cells and neurons of the deep cerebellar nuclei. The available data thus suggest that GAL) gene expression also is activated before the end of gestation. We currently know neither the normal function of the Lc locus nor the timing of the expression of the mutant allele. Some parts of the cerebellar vcrmis show abnormalities as early as P3, and degenerating cells and cellular debris appear in the Purkinje layer at P4 (Swisher and Wilson, 1977). In the homor.ygotc LciLc, death occurs within an hour of birth, in what appear to be grossly normal newborns. Thus, there may actually be some subclinical effects of the single dose or the gene during embryogenesis. The activation of the genes for GAD and calbindin might not precede the initial action of the Lc allele. The Lc locus could regulate a cascade of developmental events, which continue in the mutant until the time of actual Purkinje cell disintegration. For example, Heckroth et al. (1990), have recently suggcstcd that the Lc defect may lead to a failure of normal Purkinje innervation by olivocerebellar fibers. The simplest interpretation of our data, however, is that, at lcast in the heterozygote, the Lc allele does not affect the unfolding of the Purkinje cell's intrinsic program of cytodifferentiation. Rather, Lcl + Purkinje cells may die only upon reaching an advanced differentiated stage. Further evidence for this interpretation comes from studies of induced hyperthyroidism in mice. Hyperthyroidism speeds up the entire program of cerebellar development, including Purkinjc cell differentiation. This acceleration also leads to faster Purkinje cell death in hyperthyroid Lurcher heterozygotes (Messer et al.. 1989). The calniodulin genes, which become active during the 1st week after birth, are also expressed at apparently normal levels in Lurcher Purkinje cells, as long as they are histologically intact (Messer et al., 1990). The pattern of cell death in Liircher mice thus shows provocative similaritics to that in Huntington's
69
disease (HT)), a neurodegenerative human disease. In Lurcher heterozygotes, a specific population of GABA projection neurons degenerate after going through a substantial portion of their developmental program. In HD, the prematurely dying cells appear to be enkephalin-containing GABA neurons that project to the lateral globus pallidus and substance P-containing GABA neurons that project to thc substantia nigra pars reticulata (Tobin, 1990; Reiner et al., 1988). In both cases, specific neurons, with established positions, connections, and patterns of gene expression, die as a result of a dominant mutation. At a genetic level, however, the Lurcher allele is unlike the HD allele. LciLc homozygotes die at birth, while the phenotype of HDiHD homozygotes appears to be indistinguishable from HDi + . Thus, the HD allele is a truc dominant and the Lc allele is not. The Lurchcr mouse may nonetheless thus provide an excellent experimental system for investigating mechanisms and specificity of cell death in inherited neurodegenerative diseases.
EXPERIMENTAL PROCEDURES In situ hybridization was performed as described in Wuenschell and Tobin (1988). Probes were transcribed with SP6 or T7 polymerase from subclones of feline GAD cDNA (Kaufman et al., 1986), murine calbindin cDNA (Wood et al., 1988), and rat proenkephalin cDNA (Yoshikawa et al.. 1984). RNA probes were labeled with 3sSs-U and partially digested with alkali to fragments 150-500 bases long. Probes were hybridized to formaldehyde-fixed cryostat sections of mid-sagittal sections of Lcl + and + / cerebella. Prior to hybridization, scctions were treated sequentially with 0.02 N HC1, 0.01% Triton N-I01 , 1 Fgirnl proteinase K, and 47c formaldehyde. Prehybridization conditions were 50% formaniide and 750 inM NaCl at 37°C. Hybridization mixtures also includcd 100 mM DTT and 5 % dextran sulfate and were performed at 50°C with 0.25-0.5 ng of probe per slide. Following hybridi7ation, sections were treated with ribonuclease A, washed to a maximum stringency of 0.1 x SSC, 10 niM Na thiosulfate, 50"C, and delipidated in xylene prior to processing for autoradiography. The counterstain was hematoxylin and eosin. The Lci mutants, on a BALBicByJ background, were bred in Dr. Messer's laboratory at the New York State Department of Health. Procedures for breeding, maintaining. and killing the mice were all approved by the Institutional Animal Care and Use Committee.
+
+
ACKNOWLEDGMENTS We thank Bonnie Eisenberg for her excellent work with the mutant mice. We thank Dr. Stephen Sabol for
70
Wuenschell et al.
providing us with proenkephalin cDNA. We arc grateful to the following people for their helpful comments throughout the course of this work: Robin Fisher, Gretchen Frantz, Daniel Kaufnian, John Menkes, and Teresa Wood. This work was supported by grants from NINCDS to A.J.T. (NS 20356), and to A.M. (NS 17633). C.W.W. was supported in part by a USPHS Training Grant in Genetic Mechanisms (GM 7 104).
REFERENCES Bairnbridgc KG, Miller JJ (1982): Ir~irnunoliistocliemicallocalization of calcium-binding protein in the cerebellum. hippocampal formation and olfactory bulb of the rat. Brain Kcs 245:223-229. Caddy KWT. Patterson DL, Biscoe T J (1982): Usc of the UCHTI monoclonal antibody to explore mouse mutants and development. Nature 300:441-443. Frantz GD, Wood TL, Christakos S. Tobin AJ (1986): Cloning and expression of messenger RNA for vitamin D-dependent calcium binding protein. Soc Neuroxi Abs 12: 1457. Heckroth JA. Goldowitz D. Eisenman LM ( I 990): Olivocerebellar fiber maturation in normal and lurcher mutant micz: Defective development in lurchcr. J Comp Ncurol 291:415-430. Kaufman DL., McGinnis. JF. Krieger NR, Tobin AJ (1986): Brain glutamate decarboxylase cloned in Xgt- I I : Fusion protein produces y-aminobutyric acid. Science 232: 1138-1 140. Legrand C . Thoniasset M, Parkes CO. Greengard P, Rabie A (1983): Calcium-binding protein i n the developing rat cerebellum: An immunohistochemical study. Cell Tissue Res 233:389-402. Messer A, Eisenberg BF. Martin D ( 1989): Effects of mild hyperthyroidism in levels of amino acids in developing Lurcher cerebellum. J Neurogcnct 5:77-85. Messer A, Plumrner-Siegard J. Eiscnbcrg B (1990): Staggerer mutant mouse Purkinje cells do not contain detectable calmodulin mRNA. J Neurochem. in press.
Phillips RJS (1960): I.urcher, a new gene in linkage group XI of the house mouse. J Genet S7:35-42. Reiner A. Albin RL, AndeIson KD. D‘Amato CJ, Penney J R , Young AB (1988): Differential loss of striatal projection neurons in Huntington‘s disease. PNAS 85:5733-5737. Sar M , Stumpf WE, Miller RJ. Chang K-J, Cuatrcc Irnrnunohistochernical localization o f enkephalin in rat brain and spinal cord. J Comp Neurol 182:17-38. Swishcr DA, Wilson DB (1977): Cerebellar histogenesis in the Lurrhrr (Lc) mutant mouse. J Comp Neurol 173:205-217. Tobin A (1990): Genetic disorders of the nervous syqtem: Huntington‘s disease. ln Pearlnian AL, Collins RC (eds): “Neurobiology of Disease.” New York: Oxford University Prcsb, pp 257-275. Wassef M , Zanetta JP, Brehier A. Sotclo C [ 1985): Transient biochemical compartmentalization of Purkinje cells during early cerebellar development. Dev Biol I 11:129-137. Wetts R , IIerrup K (1982): Cerebellar Purkinje cells are descended fiom a small number or progenitors committed during early development: Quantitative analysis of Lurrlzer chimeric mice. J Neurosci 2:1494-1498. Wetts R, Herrup K (1983): Direct correlation between Purkinje and granule cell number i n the cerebella of Lurcher- chimeras and wild-type mice. Brain Kes 112:41-47. Wood TL, Kobayashi Y , Frantz G, Christakos S , Tohin 4J (1988): Expression of RNAs encoding marnmalian 28,000 M, vitamin D-dependent calcium binding protein (calbindin D281()in rodent brain and kidney. DNA 7:585-593. Wuenschell CW. Fisher RS, Kaufinan DL. Tobin AJ (1986): I n sirrr hybridiLation to localize mRNA encoding the neurotransmitter synthetic eniyme glutamatc dccarboxylase in mouse cerehellum. Proc Natl Acad Sci USA 83:6193-6197. Wuenschell C , Tobin AJ (1988):The abnormal cerebellar organiiation o r rcwrver and reeler mice does not affect the cellular distribution of three neuronal mRNAs. Neuron 1:805-815. Yoshikawa K, Williams C . Sabnl SL (1984): Rat brain proenkephalin mRNA: cDNA cloning. primary structure, arid distribution i n the central nervous system. J Biol Chem 259: 14301-143O8.
Lurcher Purkinje Cells Express Glutamic Acid Decarboxylase and Calbindin mRNAs C.W. Wuenschell, A. Messer, and A.J. Tobin Department of Biology (C.W.W., A.J.T.), Molecular Biology Institute (A.J.T.), and Brain Research Institute (A.J.T.),University of California, Los Angcles: Wadsworth Center for Laboratories and Research, New York State Department of Health and School of Public Health Scicnccs, SUNY, Albany ( A . M . ) Purkinje neurons in immature Lurcher (Lcl + ) mice are destined to die as a result of a defect intrinsic to the dying cells. We have used in situ hybridization to determine whether the Lc allele interferes with the normal program of gene expression in the doomed Purkinje cells. In P21 mice, degeneration of Purkinje cells is well underway, but the surviving Purkinje cells continue to express the mRNAs for both glutamate decarboxylase and calbindin DZsK,two proteins whose expression is characteristic of normal Purkinje neurons. We conclude that the Le allele probably does not interfere with the developmental program but acts to cause cell death in already differentiated Purkinje neurons. Key words: in situ hybridization, cytodifferentiation, neurodegeneration INTRODUCTION The unfolding of the developmental program of an organism involves a series of interactions between events internal to individual cells and influences arising from sources external to those cells, The relative contributions of intrinsic and cxtrinsic influences to the cytodifferentiation of specific types of vertebrate neurons remains an open question. The role of external cues in neuronal differentiation can be investigated by methods such as transplantation and tissuc culturc, which experimentally alter the external environments of cells. The internal genetic program of a cell is much less accessible to experimentation. Mutations that are known to exert their effects only on specific cell types, however, [nay offer insight into the genetic programs of the affected cells. One such mutation, the Lurcher mutation in the mouse, specifically affects the Purkinje neurons of the cerebellum. The mouse cerebellum is an excellent region in which to study gene expression in a specific neuron type. It has a well-defined cytoarchitecture and has been extensively investigated both functionally and developmentally. Lurclzer (Lc) is an autosoma] dominant, charactcr0 1990 Wiley-Liss, Inc.
ized by a mild-to-moderate ataxic "lurching" gait (Phillips. 1960). Lxi animals show degeneration of almost 100% of their Purkinje cells during postnatal weeks 2-5, followed rapidly by the loss of nearly 905%of the cerebellar granule cells (Caddy et al., 1982). By analyzing chimeric mice, Wetts and Herrup (1982) have demonstrated that the Lc gene acts directly in the Purkinje cells. Furthermore, Lc granule cells, which normally degeneratc after thc Purkinje cells, survive in the presence of neighboring wild-type Purkinje cells in Lc-wildtype chimeras (Wetts and Herrup, 1983). We have undertaken to study the effects of the /,urchrr mutation on the genetic program of cerebellar Purkin.je neurons by means of in situ hybridization with recombinant probes for three cerebellar mRNAs. Two of thcse mKNAs, thosc cncoding glutamic acid decarboxylase (GAD) and the calcium binding protein calbindin DZHK,are normally expressed in Purkinje neurons (Wuenschell et al.. 1986; Baimbridge and Miller, 1982, Wuenschell and Tobin, 1988). The third InRNA, encoding proenkephalin mRNA, is not expressed in Purkinje neurons, but is expressed in Golgi I1 neurons (Sar et a]., 1978; Wuenschell and Tobin, 1988). In situ hybridization permits the detection of specific mRNAs at a single cell level of resolution. Since the half-lives of mRNAs are generally shorter than those of the corresponding proteins, this technique potentially gives a more direct assessment of transcriptional status than does irnmuno h i stoc hem istry.
RESULTS By the end of the 3rd postnatal week (P21), the Lc allele has had obvious effects, and the degeneration of
Kcceived November 29, IY8Y; revised April 17, 1990; accepled April 19, 1990. Address reprint requesla to A.J. Tobin, Department of Biology. U n versily ol California, I m Angela. CA 90024-1606. C . W . Wucnschcll is now at Division of Biology, California Institute of Technology, Pasadena, CA 91 125.
66
Wuenschell et al.
Fig. 1 . GAD mRNA in the Lirrchrr cerebellum. A: Darkfield photomicrograph of a coronal section of P2 I Lwcher cerebellum hybridized with GAD antisense R N A . 1,ayers: 111 = molccular laycr: g = granular layer; w = white matter. Large straight arrow indicates labeled cell at the boundary bctwcen the molecular and granular layers. Curvcd arrow indicates labeled cell in the granular laycr. Small straight arrows indicate
labeling in the molecular laycr. Bar = 200 kin.B: Brightfield photomicrograph of the same section as in A . Arrows indicate labeled cclls: Large arrows = Purkinje cells: curved arrow = Golgi 11 ccll; small arrows = stellate or basket cells. C: Brightfield photomicrograph of il control section of P2 I Ltcrclrcr- cerebellum h y b r i d i d wilh GAD scnse RNA. Arrow indicates a Purkinje cell. Bar = 50 pm.
Purkinje cells is well underway. At this time. however, cnough Purkinje cells remain to allow studies of their rnRNAs by in situ hybridization. Hybridization of GAD antisense RNA to sections of P21 Lc cerebellum revealed labeled cells sparsely
Scattered along thc boundary between the molecular and granular layers (Fig. IA,B). Labeled cells were also scattered in the granular layer and at all depths of the molecular layer. This distribution is consistent with normal expression of the GAD gene in all of the GABA-
Lurcher Yurkinje Cells
Pig. 2. mRNAs for GAD, calbindin, and proenkephalin in the Lurcher cerebellum. Darkfield photomicrographs of sections of P2 I Lurchrr cerebellum hybridized with GAD antisense RNA (A), calbindin antisense RNA (B), proenkephalin antiscnse RNA (C), or calbindin sense RNA (control. D). A: Large arrow indicates labeled cell at the boundary between the
67
molecular and granular layers. Curved arrow indicates labeled
cell in the granular layer. Small arrows indicate labeling in the molecular laycr. B: Large arrow indicates labeled cell at thc boundary between the rnolecular and granular layers. C: Curvcd arrow indicates labeled cell i n thc granular layer. Sinall arrows indicate labeling in the deep iiiolecular layer. D: Large arrow indicates boundary between the molecular and granular layers. Bar = 0.5 mm.
ergic neuron types of the cerebellum, including the remaining Purkinje cells. No specific labeling occurred in control sections hybridized with GAD sense RNA (Fig. 1C). In order to further verify the identify of the Purkinje cells, and to study the expression of additional rnKNAs, we obtained molecular probcs for the niRNAs that encode calbindin DZgKand proenkephalin (Frantz et al., 1986; Woodet al., 1988; Yoshikawact al.. 1984). In wild-type cerebella, calbindin mRNA is expressed only in Purkinje cells, and proenkephalin i s expressed in Golgi I1 cells and a population of small cells deep in the molccular layer (Wuenschell and Tobin, 1988). When neighboring sections from the cerebellum of a P21 Lurcher mouse were hybridized with intise,,se KNA for GAD, calbindin, or proenkephalin, clear differences were apparent (Figs. 2, 3 ) . As expected. the
GAD hybridization probc (Figs. 2A, 3A,B) yielded the pattern described above for the previous experiment. The calbindin probe labeled cells located only at the boundary of the molecular and granular layers (Fig. 2B). These large cells had the appearance of Purkinje cells (Fig. 3C) and were widely and randomly spaced often with long intervening gaps. The proenkephalin probc labeled cells scattered in the granular layer and small cells both in the Purkinje layer and in the deep molecular layer (Figs. 2C, 3E,F). Control sections hybridized with sense strand calbindin R N A showed no specific labeling (Figs. 2D,3D). We did not observe unlabeled cells that could be clearly identified as Purkinje neurons in sections hybridized with GAD or calbindin probes. The total numbers of surviving Purkinje cells in the Lureher sections was, however, not great, and we cannot exclude the possibility that unlabeled Purkinjc cells might exist.
68
Wuensehell et al.
Fig. 3. mRNAs for GAD, calbindin, and proenkephalin in the Lurcher cerebellum. Brightfield photomicrographs of scctions of P21 Lurcher ccrebellum hybridized with GAD antisense RNA (A,B), calbindin antisense RNA (C), calbindin sense RNA (control, D), or proenkephalin antisense KNA (E,F). A: Large arrow indicates labeled Purkin.je cell. B: Small arrow indicates small labeled cell in the molecular layer (probably a basket cell based on position). C: Large arrows indicate labeled
Purkinje cclls. D: Large arrow indicates an unlabeled Purkinje cell. (ni = molecular layer; g = granular). E: Curved arrow indicates labeled Golgi 11 cell. Small arrows indicate sniall labeled cells at the edge of the molecular layer tentatively identified as basket cells. F: Large arrow indicates unlabeled Purkinje cell. Curved arrow indicatcs labeled Golgi I1 cell. (w = white matter). Bar = SO b"m.
DISCUSSION
Purkinje layer that were intact enough to be identified as probable Purkinje cells were labeled by both of these probes, but we may not have been able to identify cells in the later stages of degeneration. We draw three provisional conclusions from these findings:
Purkinje cells are massively dying in the Lc cerebellum, yet virtually all recognizable Purkinje cells are expressing clearly detectable levels of GAD and calbindin mRNAs. Despite the ongoing degeneration, the remaining Purkinje cells are clearly identifiable by their size, shape, and position. The only possible cell type that could be confused with Purkinje cells would be ectopic Golgi I1 cells. Such cells, however, would be likely to express proenkephalin mKNA, so they do not confound our conclusions (Wuenschell and Tobin, 1988). Lc Purkinje cells thus express the genes for GAD and calbindin-at least at the mRNA level-before they degenerate. We do not know, however, whether Purkinje cells that are actually in the process of degeneration contain these mRNAs. The majority of the large cells in the
1. The genetic program of Purkinje cells is relatively stable once it is established. The GAD and calbindin genes, whose expression is characteristic of normal differentiated Purkinje neurons, continue to be expressed in Lurcher Purkinje cclls even though these cells are destined to die. 2. The Lurchcr mutation probably causes Purkinje cell death by a fairly direct or restricted path of action. If this mutation caused Purkinje cells to die by grossly disrupting the cells' program of differentiated func-
Lurcher Purkinje Cells
tions, it is unlikely that expression of genes characteristic of the differentiated phenotype would persist. 3. The i,urcher mutant gene appears to act later in time than the activation of GAD and calbindin gene expression. The available evidence indicates that the expression of GAD and calbindin genes begins early in development. At least some migrating Purkinje cells at embryonic day (E) 16 in the rat (Legrand, 1983; Wassef et al., 1985) contain immunoreactive calbindin. Furthermore, cerebellar primordia of E l 5 mouse embryos contain both GAD and calbindin mRNAs (G. Frantz and A. Tobin, unpublished results). The hybridizing cells lie above the fourth ventricle, in the region that contains precursors of Purkinje cells and neurons of the deep cerebellar nuclei. The available data thus suggest that GAL) gene expression also is activated before the end of gestation. We currently know neither the normal function of the Lc locus nor the timing of the expression of the mutant allele. Some parts of the cerebellar vcrmis show abnormalities as early as P3, and degenerating cells and cellular debris appear in the Purkinje layer at P4 (Swisher and Wilson, 1977). In the homor.ygotc LciLc, death occurs within an hour of birth, in what appear to be grossly normal newborns. Thus, there may actually be some subclinical effects of the single dose or the gene during embryogenesis. The activation of the genes for GAD and calbindin might not precede the initial action of the Lc allele. The Lc locus could regulate a cascade of developmental events, which continue in the mutant until the time of actual Purkinje cell disintegration. For example, Heckroth et al. (1990), have recently suggcstcd that the Lc defect may lead to a failure of normal Purkinje innervation by olivocerebellar fibers. The simplest interpretation of our data, however, is that, at lcast in the heterozygote, the Lc allele does not affect the unfolding of the Purkinje cell's intrinsic program of cytodifferentiation. Rather, Lcl + Purkinje cells may die only upon reaching an advanced differentiated stage. Further evidence for this interpretation comes from studies of induced hyperthyroidism in mice. Hyperthyroidism speeds up the entire program of cerebellar development, including Purkinjc cell differentiation. This acceleration also leads to faster Purkinje cell death in hyperthyroid Lurcher heterozygotes (Messer et al.. 1989). The calniodulin genes, which become active during the 1st week after birth, are also expressed at apparently normal levels in Lurcher Purkinje cells, as long as they are histologically intact (Messer et al., 1990). The pattern of cell death in Liircher mice thus shows provocative similaritics to that in Huntington's
69
disease (HT)), a neurodegenerative human disease. In Lurcher heterozygotes, a specific population of GABA projection neurons degenerate after going through a substantial portion of their developmental program. In HD, the prematurely dying cells appear to be enkephalin-containing GABA neurons that project to the lateral globus pallidus and substance P-containing GABA neurons that project to thc substantia nigra pars reticulata (Tobin, 1990; Reiner et al., 1988). In both cases, specific neurons, with established positions, connections, and patterns of gene expression, die as a result of a dominant mutation. At a genetic level, however, the Lurcher allele is unlike the HD allele. LciLc homozygotes die at birth, while the phenotype of HDiHD homozygotes appears to be indistinguishable from HDi + . Thus, the HD allele is a truc dominant and the Lc allele is not. The Lurchcr mouse may nonetheless thus provide an excellent experimental system for investigating mechanisms and specificity of cell death in inherited neurodegenerative diseases.
EXPERIMENTAL PROCEDURES In situ hybridization was performed as described in Wuenschell and Tobin (1988). Probes were transcribed with SP6 or T7 polymerase from subclones of feline GAD cDNA (Kaufman et al., 1986), murine calbindin cDNA (Wood et al., 1988), and rat proenkephalin cDNA (Yoshikawa et al.. 1984). RNA probes were labeled with 3sSs-U and partially digested with alkali to fragments 150-500 bases long. Probes were hybridized to formaldehyde-fixed cryostat sections of mid-sagittal sections of Lcl + and + / cerebella. Prior to hybridization, scctions were treated sequentially with 0.02 N HC1, 0.01% Triton N-I01 , 1 Fgirnl proteinase K, and 47c formaldehyde. Prehybridization conditions were 50% formaniide and 750 inM NaCl at 37°C. Hybridization mixtures also includcd 100 mM DTT and 5 % dextran sulfate and were performed at 50°C with 0.25-0.5 ng of probe per slide. Following hybridi7ation, sections were treated with ribonuclease A, washed to a maximum stringency of 0.1 x SSC, 10 niM Na thiosulfate, 50"C, and delipidated in xylene prior to processing for autoradiography. The counterstain was hematoxylin and eosin. The Lci mutants, on a BALBicByJ background, were bred in Dr. Messer's laboratory at the New York State Department of Health. Procedures for breeding, maintaining. and killing the mice were all approved by the Institutional Animal Care and Use Committee.
+
+
ACKNOWLEDGMENTS We thank Bonnie Eisenberg for her excellent work with the mutant mice. We thank Dr. Stephen Sabol for
70
Wuenschell et al.
providing us with proenkephalin cDNA. We arc grateful to the following people for their helpful comments throughout the course of this work: Robin Fisher, Gretchen Frantz, Daniel Kaufnian, John Menkes, and Teresa Wood. This work was supported by grants from NINCDS to A.J.T. (NS 20356), and to A.M. (NS 17633). C.W.W. was supported in part by a USPHS Training Grant in Genetic Mechanisms (GM 7 104).
REFERENCES Bairnbridgc KG, Miller JJ (1982): Ir~irnunoliistocliemicallocalization of calcium-binding protein in the cerebellum. hippocampal formation and olfactory bulb of the rat. Brain Kcs 245:223-229. Caddy KWT. Patterson DL, Biscoe T J (1982): Usc of the UCHTI monoclonal antibody to explore mouse mutants and development. Nature 300:441-443. Frantz GD, Wood TL, Christakos S. Tobin AJ (1986): Cloning and expression of messenger RNA for vitamin D-dependent calcium binding protein. Soc Neuroxi Abs 12: 1457. Heckroth JA. Goldowitz D. Eisenman LM ( I 990): Olivocerebellar fiber maturation in normal and lurcher mutant micz: Defective development in lurchcr. J Comp Ncurol 291:415-430. Kaufman DL., McGinnis. JF. Krieger NR, Tobin AJ (1986): Brain glutamate decarboxylase cloned in Xgt- I I : Fusion protein produces y-aminobutyric acid. Science 232: 1138-1 140. Legrand C . Thoniasset M, Parkes CO. Greengard P, Rabie A (1983): Calcium-binding protein i n the developing rat cerebellum: An immunohistochemical study. Cell Tissue Res 233:389-402. Messer A, Eisenberg BF. Martin D ( 1989): Effects of mild hyperthyroidism in levels of amino acids in developing Lurcher cerebellum. J Neurogcnct 5:77-85. Messer A, Plumrner-Siegard J. Eiscnbcrg B (1990): Staggerer mutant mouse Purkinje cells do not contain detectable calmodulin mRNA. J Neurochem. in press.
Phillips RJS (1960): I.urcher, a new gene in linkage group XI of the house mouse. J Genet S7:35-42. Reiner A. Albin RL, AndeIson KD. D‘Amato CJ, Penney J R , Young AB (1988): Differential loss of striatal projection neurons in Huntington‘s disease. PNAS 85:5733-5737. Sar M , Stumpf WE, Miller RJ. Chang K-J, Cuatrcc Irnrnunohistochernical localization o f enkephalin in rat brain and spinal cord. J Comp Neurol 182:17-38. Swishcr DA, Wilson DB (1977): Cerebellar histogenesis in the Lurrhrr (Lc) mutant mouse. J Comp Neurol 173:205-217. Tobin A (1990): Genetic disorders of the nervous syqtem: Huntington‘s disease. ln Pearlnian AL, Collins RC (eds): “Neurobiology of Disease.” New York: Oxford University Prcsb, pp 257-275. Wassef M , Zanetta JP, Brehier A. Sotclo C [ 1985): Transient biochemical compartmentalization of Purkinje cells during early cerebellar development. Dev Biol I 11:129-137. Wetts R , IIerrup K (1982): Cerebellar Purkinje cells are descended fiom a small number or progenitors committed during early development: Quantitative analysis of Lurrlzer chimeric mice. J Neurosci 2:1494-1498. Wetts R, Herrup K (1983): Direct correlation between Purkinje and granule cell number i n the cerebella of Lurcher- chimeras and wild-type mice. Brain Kes 112:41-47. Wood TL, Kobayashi Y , Frantz G, Christakos S , Tohin 4J (1988): Expression of RNAs encoding marnmalian 28,000 M, vitamin D-dependent calcium binding protein (calbindin D281()in rodent brain and kidney. DNA 7:585-593. Wuenschell CW. Fisher RS, Kaufinan DL. Tobin AJ (1986): I n sirrr hybridiLation to localize mRNA encoding the neurotransmitter synthetic eniyme glutamatc dccarboxylase in mouse cerehellum. Proc Natl Acad Sci USA 83:6193-6197. Wuenschell C , Tobin AJ (1988):The abnormal cerebellar organiiation o r rcwrver and reeler mice does not affect the cellular distribution of three neuronal mRNAs. Neuron 1:805-815. Yoshikawa K, Williams C . Sabnl SL (1984): Rat brain proenkephalin mRNA: cDNA cloning. primary structure, arid distribution i n the central nervous system. J Biol Chem 259: 14301-143O8.
