Parallin, a cerebellar granule cell protein the expression of which is developmentally regulated by Purkinje cells: evidence from mutant mice

Developmental Brain Research 104 Ž1997. 79–89

Research report

Parallin, a cerebellar granule cell protein the expression of which is developmentally regulated by Purkinje cells: evidence from mutant mice Alan M. Smith, Richard J. Mullen

)

Department of Neurobiology and Anatomy, UniÕersity of Utah School of Medicine, 401 MREB, 50 N. Medical DriÕe, Salt Lake City, UT 84132, USA Accepted 12 August 1997

Abstract In this paper we report on monoclonal antibody 3H6 with unique specificities for development of the cerebellum. Immunohistochemical studies on normal and mutant mice suggest that it is primarily located in or on granule cell parallel fibers in the cerebellum. The only other region showing immunoreactivity is a small region of the hippocampus. The antigen is detected immunohistochemically as early as postnatal day 11 in the molecular layer of the cerebellum. In adult wild-type mice parallin expression is seen in the molecular layer and to a lesser degree in the internal granular layer. In the cerebella of two neurological granule cell-deficient mutants, weaÕer Ž wÕ . and staggerer Ž sg ., parallin is not detected. However, in two Purkinje cell-deficient mutants, Purkinje cell degeneration Ž pcd . and nerÕous Ž nr ., a more complex and interesting pattern is observed. These two mutants do have granule cells and parallel fibers and 3H6 immunoreactivity is observed. However, in both of these Purkinje cell-deficient mutants the 3H6 immunoreactivity is drastically reduced in regions where Purkinje cells have degenerated. Furthermore, in nr mutants, the antigen appears to be concentrated in regions of the parallel fiber that are in close proximity to Purkinje cells, suggesting its possible association with synapses. Taken together these results suggest that parallin is a marker of granule cells and their parallel fibers, its onset correlates with the formation of granule cell synapses on developing Purkinje cells, and it requires Purkinje cells for the maintenance of expression. q 1997 Elsevier Science B.V. Keywords: Monoclonal antibody; Parallel fiber; Mutant, weaÕer; Mutant, staggerer; Mutant, Purkinje cell degeneration; Mutant, nerÕous

1. Introduction A primary goal of developmental neurobiology is to elucidate the cellular and molecular mechanisms underlying the sequential developmental processes which determine the eventual structure and function of the adult nervous system. These mechanisms appear to be epigenetic consequences of both intrinsic and extrinsic cellular and molecular events during development. To investigate mechanisms of cell interactions in an extremely complex organ like the brain it is advantageous to work with a relatively simple system which is sufficiently well characterized. The mouse cerebellum has been the focus of many researchers for this important reason. In the murine cerebellum the spatial and temporal parameters for the origin, migration, differentiation, and synaptogenesis for all types of neurons are known. In addition, all afferent inputs and

) Corresponding author. Fax: q1 Ž801. 581-4233; E-mail: [email protected]

0165-3806r97r$17.00 q 1997 Elsevier Science B.V. All rights reserved. PII S 0 1 6 5 - 3 8 0 6 Ž 9 7 . 0 0 1 4 1 - 7

efferent outputs are known and more than 20 murine neurological mutants affecting the cerebellum have been identified w9,31x making the cerebellum the system of choice for many studies. The two principal cell types in the cerebellar cortex are the Purkinje and granule cells. The Purkinje cell has an enormous dendritic tree but it is flattened in a plane transverse to the folium so that in a midline sagittal section you see the numerous branches of the tree whereas in coronal section the dendrite appears more pole-like. The granule cells are derived from the external granular layer ŽEGL. from whence they migrate across the molecular layer to form the internal granular layer ŽIGL.. In the adult, the granule cell body lies in the IGL, but its axon projects into the molecular layer where it bifurcates and the processes run in opposite directions in a plane at right angles to the plane of the Purkinje cell dendrite. Thus, in a coronal section, the granule cell axons run approximately in the plane of the section. Because of their common orientation, the axons are also called parallel fibers. Parallel fiber varicosities will form synapses on the dendritic

80

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

spines of numerous Purkinje cells as the axons pass through the dendritic arbors of the Purkinje cells. Despite what is known, most of the genetic, molecular and cellular mechanisms underlying the development of the cerebellum are still unknown. One research strategy is to study these mechanisms using the hybridoma technology to generate new molecular markers w4,16,17,24,27,37x. By using this strategy, it is possible to identify novel and important developmentally regulated molecules and to eventually determine their function in the cell. In this paper we report on the monoclonal antibody 3H6, an IgM, which recognizes a cerebellar granule cell antigen which we have named parallin. Our studies on normal and mutant mice, including weaÕer Ž wÕ ., stag-

gerer Ž sg ., Purkinje cell degeneration Ž pcd . and nerÕous Ž nr ., suggest that parallin is a specific marker of granule cells, in particular, their parallel fibers; it appears at the time granule cells are synapsing with developing Purkinje cells, and it requires these same cells for the maintenance of expression of this antigen.

2. Materials and methods 2.1. Animals and tissue preparation The strains of mice used, BALBrcByJ, C57BLr6Rora s g Ž staggerer ., C57BLr6-Kcnj6 w Õ Ž weaÕer .,

Fig. 1. 3H6 IHC in adult brain. A: sagittal section montage of adult brain showing intense staining in the cerebellar cortex. There is also a small region of the hippocampus Žboxed area. that is stained. B: higher magnification of the boxed area in ŽA. showing staining in a small region of the hippocampus, possibly CA2. C: higher magnification of the boxed area in ŽB. showing the IHC staining is most intense in the polymorphic layer above the pyramidal cells though there is also staining in the molecular layer below the pyramidal cells. Bar: 100 mm.

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

B6C3Fe-pcd Ž Purkinje cell degeneration . and C3HeBrFeJ-nr Ž nerÕous ., were bred in our animal facility. The mice were on a 12 h lightrdark cycle with food and water ad libitum. Day of birth is considered day 0 ŽP0.. Both neonatal and adult mice were perfused to improve histology and reduce non-specific staining. First, after deep Avertin anesthesia, they were perfused transcardially for 1–2 min using Heparinrsaline and then with paraformaldehyde–lysine–periodate ŽPLP. fixative w25x for approximately 8 min. The heads were immersed in fixative for 1–2 h and after dissection the whole brains were then immersed in fixative for an additional 3–4 h Žtotal fixation time: 5–6 h.. They were then rinsed overnight in phosphate-buffered sucrose Ž5%., dehydrated and embedded in polyester wax Žpolyethylene glycol 400 distearate, Ruger Chemical, Irvington, NJ. that had been treated as previously described w8x. Sections Ž8 mm. were mounted on gelatinrchrom–alum Žchromium potassium sulfate. tripled subbed slides and dried overnight before using.

81

mounted in glycerin, coverslipped and photographed. The coverslip was then removed and the slides were reacted with the anti-calbindin mAb and visualized with VECTASTAIN ABC-AP and Vector Blue substrate. The slides were then permanently mounted and re-photographed. In some regions there was mild cross-reactivity between the second set of reagents and the first such that the yellow–brown DAB staining would appear somewhat olive greenrblue in color. However, the Purkinje cells were unambiguously labeled bright blue by the calbindin staining and their appearance was clearly distinct from the 3H6 staining.

3. Results 3.1. Expression in adult The antigen was first detected with the mAb 3H6, an IgM, in an immunohistochemical ŽIHC. assay on sagittal

2.2. Hybridoma production The hybridoma producing mAb 3H6 was derived from BALBrc mice that had been immunized with brain nuclei isolated from Mus castaneus mice. The techniques for isolation of nuclei were those of Lovtrup and McEwen w23x as previously described w27x. The immunization protocol is the same as that described previously w27x except Mus castaneus was used as an immunogen, the mice were boosted only 5 times, and the splenocytes were fused with Sp2r0–Ag14 myeloma cells. 2.3. Immunohistochemistry For most studies mAb 3H6 was diluted 1:1 with 3% normal serum in Tris-buffered saline and the slides incubated overnight at 48C. Several detection methods were used to determine immunoreactivity. For the horseradish peroxidase ŽHRP. method of detection either the VECTASTAIN ABC Elite Kit ŽVector Labs, Burlingame, CA. or Hyclone ŽLogan, UT. goat anti-mouse was used with 3,3X-diaminobenzidine tetrahydrochoride ŽDAB. as the substrate which yields a brown reaction product. For the alkaline phosphatase method of detection the VECTASTAIN ABC-AP Alkaline Phosphatase Kit was used with Alkaline Phosphate Substrate Kit III ŽVector Blue. as the substrate which yields a blue reaction product. The anticalbindin–D28K, a mouse monoclonal antibody, was obtained from Sigma ŽSt. Louis, MO.. In some of the studies we used a double-labeling technique to stain both 3H6 and calbindin antigens in the same section using VECTASTAIN ABC kits according to the manufacturer’s protocol. Briefly, the slides were first reacted with mAb 3H6 and visualized with VECTASTAIN ABC–HRP and DAB substrate. The slides were then

Fig. 2. Coronal section of adult cerebellum immunostained with mAb 3H6. A: low magnification view showing specificity of staining in cerebellar cortex and absence of staining in brainstem. B: higher magnification of same section. Note striated appearance of staining in molecular layer Žm., staining in granule cell layer Žg. and absence of staining in the large Purkinje cells Žp.. Bar: ŽA. 1 mm; ŽB. 20 mm.

82

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

sections of whole adult mouse brain. The most striking feature of IHC staining is the intense labelling in the cerebellar cortex, particularly the molecular layer ŽFig. 1A.. A similar staining pattern was observed in the rat cerebellum, but no cross-reactivity was detected in the chick cerebellum. In the mouse brain we also observed lighter staining in a small region of the hippocampus ŽFig. 1A,B.. Initially, this was thought to be an artifact, but it has been a consistent observation and five other unrelated IgM monoclonal antibodies did not show staining in this area. In the hippocampus the immunoreactivity is confined to a small region midway along Ammon’s horn ŽFig. 1B.. The location suggests it may be CA2. The hippocampal staining appears in both the stratum oriens and stratum radiatum, though it is more intense in the stratum oriens, above the pyramidal cells ŽFig. 1C.. IHC labelling in the cerebellum is observed in the molecular layer and, to a lesser extent, the internal granule cell layer in the adult ŽFig. 2A.. In the wild-type cerebel-

lum the IHC staining appears uniform throughout all areas of the molecular layer ŽFig. 2A.. There was no obvious gradient of expression in either a medial–lateral ŽFig. 2A. or cranial–caudal direction ŽFig. 1A.. Light microscopy of the molecular layer of stained coronal sections viewed at high magnification reveals transverse striations which suggests that the antigen is associated with the parallel fibers ŽFig. 2B.. Hence, we have named this granule cell marker parallin. With the light microscope it is difficult to determine with certainty the location of the staining in the granule cell layer. However, it appears to be associated with processes including processes projecting through the Purkinje layer into the molecular layer, which are likely to be fascicles of ascending processes of the granule cells. Nevertheless, we cannot rule out staining of the granule cell body or, for that matter, the glomeruli Ži.e., the granule cell dendrite.. The fact that we immunized with nuclei but only see staining of processes and no nuclear staining is perplexing

Fig. 3. Developmental regulation of 3H6 IHC in coronal sections of cerebellum. A: section from P11 pup showing ventral region of the vermis with IHC staining in the molecular layer Žm. of the nodulus. The arrows point to the pia in between adjacent lobules. Note the absence of staining in the more superior lobule. B: low magnification view of the cerebellar hemisphere of a P12 pup showing onset of immunoreactivity in several areas of cortex. A color enhancement filter ŽWratten a45 blue. was used to enhance the intensity of the staining in this section. C–F: sections showing a region near the vermisrhemisphere border on P11, P12, P13 and P14, respectively. In this region on P11 ŽC., no IHC staining is evident, but by P12 ŽD. light staining can be detected in the molecular layer. The staining intensity increases on P13 ŽE. and P14 ŽF.. Note the absence of staining in the external granule cell layer Žeg., Purkinje cell layer Žp. and internal granule cell layer, beneath the Purkinje cells.

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

but not too surprising. Our antigen may have co-purified with the nuclei or may have been a contaminant. 3.2. Expression during deÕelopment By IHC methods parallin is detected as early as P11 in the mouse. IHC staining is first observed in the developing molecular layer in ventral midline regions ŽFig. 3A.. There is no staining detected in the external or internal granule cell layers at this age. By P12, faint staining is detected in the molecular layer in many regions of the cerebellum ŽFig. 3B.. The section in Fig. 3B was photographed with a Wratten a45 blue filter to enhance the intensity of the staining. There appears to be a slight medial to lateral onset of expression and deeper parts of the lobules show more intense staining than areas near the surface of the cortex. To demonstrate the onset of expression in a common region, Fig. 3C–F shows micrographs Žwithout enhancing filters. of a region near the vermisrhemisphere border Žlobulus simplexrcrus I lobuli ansiformis. on P11, P12, P13 and P14, respectively. In this region on P11, no immunoreactivity is detected. By P12 light staining is seen in the molecular layer ŽFig. 3D. and the staining becomes more intense on P13 ŽFig. 3E. and P14 ŽFig. 3F.. Note that even at P14 there is no IHC staining detectable in the granule cell layer as is seen in adults ŽFig. 2..

83

and the granule cell degeneration is secondary w12x. Fig. 4B is from a 39-day-old homozygous sg mutant at which age virtually all granule cells have degenerated. As with wÕ mutants no detectable level of immunostaining is seen in this mutant cerebellum using mAb 3H6. In the pcd mutant, Purkinje cells in the homozygous mutant degenerate starting at about P18, continues with approximately 95% lost by P30, and by P50 is virtually complete w28x. In this mutant the granule cells undergo a secondary degeneration which is very slow with the most severe loss occurring between 3 and 12 months w38x. Also, it has been determined that in the homozygous mutants synaptogenesis appears to be qualitatively normal until P18 but soon after synaptogenesis becomes disordered

3.3. Immunohistochemical studies of neurological mutants In the homozygous wÕ mutant Žgene symbol Kcnj6 w Õ . the granule cells degenerate during the first and second weeks after birth w30x. Proliferation in the EGL appears to be relatively normal w33x but the granule cells fail to translocate inward and fail to generate bipolar cytoplasmic processes which normally differentiate into parallel fibers. In this mutant almost all granule cells die soon after their genesis. However, in the lateral hemispheres, particularly in the paraflocculus, a small population of granule cells is known to survive at least to 1 month of age w15x. No 3H6 immunoreactivity is observed in the molecular or granule cell layer of a P28 homozygous wÕ mutant cerebellum ŽFig. 4A. following the degeneration of the granule cells. We did not examine younger wÕ mice to determine if the granule cells express the 3H6 antigen before they degenerate. However, in the lateral hemispheres, most noticeably in the flocculus and paraflocculus where a few granule cells are known to survive, light staining is detected ŽFig. 4A, inset.. Granule cells also degenerate in the sg mutant Ž Rora s g . but later, between the second and fifth weeks after birth, after they have migrated to the IGL w30x. In this mutant it is thought that since the Purkinje cells fail to develop mature dendritic spines, the granule cells degenerate because they are deprived of their normal targets, the dendritic spines w19,35x. Studies of staggerer chimeras support the idea that the primary defect is in the Purkinje cell w14x

Fig. 4. Sections of cerebella from granule cell-deficient wÕ and sg mutants immunostained with mAb 3H6 and photographed with Hoffman Modulation Contrast optics so that tissue could be more readily discerned. A: sagittal section of a P28 wÕr wÕ cerebellum showing total absence of staining. The inset is from a different section showing low levels of staining in the flocculus where some granule cells do survive. B: sagittal section of P39 sg r sg with no detectable 3H6 staining. Bar: 100 mm.

84

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

Fig. 5. A,B: midline sagittal sections of cerebella from 7-week-old " ŽA. and pcd r pcd ŽB. litter mates immunostained with mAb 3H6. Note the reduction in staining throughout most of the molecular layer of the pcd r pcd except in the region of the nodulus. The delineated area of the nodulus in ŽB. is shown at higher power in ŽD., photographed with Hoffman Modulation Contrast optics to show surviving Purkinje cells Žarrowheads in D. in the area with high levels of 3H6 immunostaining. C: coronal section of a 30-week-old pcd r pcd showing the drastic reduction in 3H6 immunostaining throughout the molecular layer in both the vermis Žv. and hemisphere Žh. but still moderate staining in the granule cell layer in the hemisphere. Bar: ŽA–C. 1 mm; ŽD. 50 mm.

w21x. In Fig. 5B the cerebellum of a 7-week-old homozygous pcd mutant is stained with mAb 3H6. At this age virtually all Purkinje cells have degenerated and markedly reduced staining is seen throughout most of the molecular layer as compared with the litter mate control in Fig. 5A.

However, intense staining still remains in the nodulus region and in a few other regions. In Fig. 5D a region of the nodulus from Fig. 5B is shown at higher magnification, and it can be seen that there are surviving Purkinje cells in the region of intense 3H6 staining. In other pcd r pcd

Fig. 6. Coronal sections of cerebellum from a 6-month-old nr r nr mutant mouse. A: montage of a section immunostained with 3H6 visualized by HRPrDAB, coverslipped with glycerol and photographed. In the vermis Žv. region the staining appears in a banded pattern. In the hemispheres Žh., most of the staining in the molecular layer is confined to bilaterally symmetrical regions of the lobulus simplex Žls.. However, there are small foci of moderately increased staining elsewhere in the molecular layer Žarrowheads.. The delineated area in ŽA. is shown at higher magnification in ŽC.. After photographing the section, the coverslip was removed and the slide immunostained with anti-calbindin and Vector Blue to show the location of Purkinje cells as seen in montage ŽB.. The arrows point to the midline band and two bands slightly off the midline where the Purkinje cells have degenerated. Note the almost perfect correlation between the location of Purkinje cells in ŽB. and the areas with high 3H6 staining in ŽA.. Even the small foci of staining in ŽA. correspond to the location of solitary Purkinje cells in ŽB, arrowheads.. The delineated area in ŽB. is shown at higher magnification in ŽD.. The delineated area in ŽD. is shown at higher magnification in ŽF. and the distinction between the brown parallel fiber and blue Purkinje cell staining is more obvious. E: control section of nr r nr stained with just anti-calbindin and Vector Blue. G: control section of nr r nr stained with 3H6, photographed, then processed with all of the second set of reagents except the anti-calbindin antibody and re-photographed in ŽH. to show the mild cross-reaction of the second set of reagents with the first set. Slightly more cross-reactivity can be seen in the most lateral hemisphere of ŽB. where there are no Purkinje cells but the tissue has a slightly olive-green color. Bar: ŽA,B,G,H. 1 mm; ŽC–E. 100 mm; ŽF. 50 mm.

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

85

86

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

mutants in which even the Purkinje cells in the nodulus have degenerated, the intense 3H6 staining in the nodulus is not observed Ždata not shown.. We examined a total of 10 pcd r pcd varying in age from 7 to 30 weeks and, although there is some animal to animal variation, some 3H6 immunoreactivity persists even in the oldest mutants. A quite consistent observation is that in old pcd r pcd mutants the staining in the granule cell layer is more intense in the hemispheres ŽFig. 5C.. Finally, in the nr mutant, roughly 90% of the Purkinje cells slowly and selectively degenerate between P23 and P50 with much more extensive degeneration occurring in the hemispheres than in the vermis w20,32x. A fascinating feature of this degeneration is that it is not random, rather the areas of degenerated and surviving Purkinje cells form a reproducible pattern of parasagittal bands w40x. In this mutant there is only a slight loss of granule cells w20,32x and, even in the hemispheres where Purkinje cell degeneration is near total, numerous parallel fiber varicosities persist, forming synapses on interneurons or in clusters covered by glial processes w36x. Fourteen nr r nr mutants from 7 to 30 weeks of age were examined. The first sections of nr r nr mice we examined were sagittal sections and there appeared to be a blotchy staining pattern primarily in the anterior lobules of the cerebellum Ždata not shown.. Realizing that the areas with the most intense 3H6 staining were those areas where there were surviving Purkinje cells we examined coronal sections. In Fig. 6 are coronal sections of a 6-month-old nr mutant cerebellum immunostained with mAb 3H6 and double-labeled with anticalbindin along with control sections. Fig. 6A is a coronal section stained with 3H6 using HRPrDAB reagents. Note in the midline vermis region the immunoreactivity appears in distinct parasagittal bands as well as symmetrical regions of staining in the two hemispheres, particularly in the lobulus simplex. Also note that in the most lateral lobules of the hemispheres there appears to be more residual 3H6 immunoreactivity in the granule cell layer as well as the molecular layer. In Fig. 6B the same section has been stained with anti-calbindin to stain Purkinje cells using APrVector Blue reagents. The areas of most intense 3H6 staining in Fig. 6A shows a near perfect correlation with the location of Purkinje cells shown in Fig. 6B. In some regions of the left hemisphere there are small foci of moderately increased 3H6 staining in the molecular layer ŽFig. 6A. that would almost appear to be artifactual or random variations in the intensity of IHC. However, by comparing Fig. 6A and 6B it can be seen that these small foci of 3H6 staining match nearly perfectly with some scattered Purkinje cells and in some instances it appears the elevated 3H6 staining is associated with a single isolated Purkinje cell. By examining the sections at higher power we determined that none of the olive greenrblue staining in the most lateral lobules is Purkinje cell staining, rather it is artifactual staining caused by cross-reactivity of the reagents. Fig. 6C,D shows higher magnifications of the

midline regions delineated in Fig. 6A and 6B, respectively. These figures at higher magnification more clearly show that the blue Purkinje cell staining is very distinct. Again, note the near perfect correlation between areas of intense 3H6 staining ŽFig. 6C. and surviving Purkinje cells ŽFig. 6D.. Fig. 6F is an enlargement of the area delineated in Fig. 6D to show that in some regions where there are no Purkinje cells Že.g., to the right in the figure. there is virtually no 3H6 immunoreactivity in the molecular layer while in other areas where the Purkinje cells are closer together, there is 3H6 immunoreactivity between the Purkinje cells. Fig. 6E,G,H shows nearby control sections of the same brain. Fig. 6E is a section immunostained with just anticalbindin using APrVector Blue. Fig. 6G is immunostained with 3H6 using HRPrDAB and Fig. 6H is the same section after treating with the second set of reagents, that is the APrVector Blue, etc., but omitting the anti-calbindin antibody. This control shows only minor cross-reactivity between the two sets of reagents but in some sections and regions there is somewhat more cross-reactivity such as in the lateral hemispheres in Fig. 6B, as previously mentioned. However, identification of Purkinje cells andror their dendrites is unequivocal. The parasagittal banding patterns seen in Fig. 6A,B and at high magnification in Fig. 6C,D show a remarkable correlation for both antibodies in this mutant. At high magnification it can be seen that where Purkinje cells have degenerated there is little or no detectable 3H6 staining in the region. The relative abruptness of the boundaries between stained and unstained areas with mAb 3H6 suggest that direct contact between Purkinje dendrites and parallel fibers is required for maintaining parallin expression.

4. Discussion Monoclonal antibody 3H6 is a marker for cerebellar granule cells. In the adult brain immunoreactivity is strongest in the molecular layer and appears to be associated with the parallel fibers, the axons of the granule cells. With only light microscopic observations we cannot determine whether the antigen is located within or on the surface of the parallel fiber or whether it is secreted into the extracellular space. IHC staining is also seen in the internal granular layer. The fact that it stains the parallel fibers suggests that the staining in the granular layer may be associated with ascending processes of the granule cells projecting into the molecular layer. Because the most pronounced staining is associated with cerebellar granule cells and in particular their parallel fibers, we have named this antigen parallin. With the exception of a single report by Gravel et al. w10x for mAb Q600, which appears to stain cerebellar granule cells specifically, this is the only other granule cell-specific mAb to the best of our knowledge. Several other investigators have reported monoclonal anti-

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

bodies which intensely stain cerebellar granule cells w3,7,41x. However, these other antibodies show either a different temporal and spatial staining pattern or also stain other cell types in the brain or both. A polyclonal antibody was used to isolate numerous granule cell-specific cDNA clones w18x, most of which appear to be expressed much earlier than 3H6. The only other place we detected 3H6 immunoreactivity was in a small region of the hippocampus where the staining was very light. The location of the staining suggests that it may be localized to CA2, an area frequently not identified as a distinct region in mice. Interestingly, we do not see IHC staining in rat hippocampus but did see staining in rat cerebellar molecular layer. 4.1. Significance of early expression of parallin At the time parallin is first detected in the cerebellum, P11, several major developmental events are occurring: granule cell migration, parallel fiber formation, synaptogenesis, Purkinje dendrite maturation, and the establishment of the adult cytoarchitecture. Immunoreactivity is first detected in ventral vermis, is more intense in the depths of the lobules than at the surface and shows a slight medial to lateral gradient at P12. This pattern of expression is similar to the pattern of granule cell differentiation described by Altman w2x in the rat. Most important for our study is the fact that Purkinje cell dendrite maturation and interaction with the developing parallel fibers is beginning just prior to the immunodetection of parallin. In a Golgi study of the Purkinje cell dendrite by Hendelman and Aggerwal w11x it was shown that spines appear on these dendrites between P7 and P10. Their study also found that between P11 and P14 the dendrites continue to grow and by P16 secondary and tertiary dendritic branches have appeared. Several electron microscopic analyses of postnatal histogenesis in the cerebellar cortex w19,22x have shown that parallel fibers begin to form immature synapses with Purkinje cell dendritic shafts and spines at the end of the first week but after P16 parallel fiber: Purkinje synapses in the normal mouse are present only on spines. Both the temporal and spatial expression of parallin strongly suggest that some form of transneuronal interaction between the Purkinje cell and the parallel fiber of the granule cell is involved in the induction of parallin. 4.2. Neurological mutants Four murine neurological mutants relevant for the study of interactions between Purkinje and granule cells using mAb 3H6 are weaÕer Ž Kcnj6 w Õ ., staggerer Ž Rora s g ., Purkinje cell degeneration Ž pcd . and nerÕous Ž nr .. For most of these mutants the primary site of gene action has been determined using chimeric analyses w26x. And for each mutant extensive analysis using both light and electron microscopy has been performed to identify the abnormalities in cellular and intercellular structures.

87

In the cerebellar mutants Ž wÕ and sg ., where granule cells degenerate, no detectable staining is seen in midline sections. However, in the lateral hemispheres of the wÕ mutant where a small population of granule cells is known to survive at least for the first month w15x, 3H6 immunoreactivity is observed showing that wÕ granule cells are capable of expressing the parallin if the granule cell survives. In addition to having Purkinje cells, wÕ and sg mutants also have basket and stellate cells, the other neuronal cell types in the molecular layer and these cells do make relatively normal synapses with parallel fibers, though reduced in number w19,34x. Both mutants also have climbing fibers and Golgi cells. Thus, the appearance of the 3H6 staining in the normal molecular layer and the absence of staining in these two mutants that have all neuronal cell types except granule cells strongly suggest that the parallin is restricted to the granule cell. In the neurological mutants pcd and nr, where Purkinje cells undergo degeneration after developing relatively normally, the expression of parallin is severely reduced. In the original report of pcd w28x, it was pointed out that the Purkinje cells in the nodulus either persisted or were the last ones to degenerate. We found that when Purkinje cells do survive in the nodulus region, parallin is expressed at high levels in the molecular layer. In other pcd mutants in which all Purkinje cells have degenerated the intense 3H6 staining is lost. However, even in the oldest pcd mutants examined, months after the Purkinje cells have degenerated, there is still some 3H6 immunoreactivity throughout most of the cerebellar cortex. The highest levels of this immunoreactivity appears to be in the granule cell layer in the hemispheres. In the neurological mutant nr the first morphological indication of degeneration is by P9 when the mitochondria appear abnormally rounded and by P15 all Purkinje cells exhibit these abnormally shaped mitochondria. These changes plus others affecting cell organelles culminate in cell death for many of the Purkinje cells starting at P23. In nr the elaboration of the dendritic tree is retarded during this period but parallel fiber synapses are present at P9 on normal appearing dendritic spines w20x. The degeneration of Purkinje cells in nr mutants is restricted to parasagittal bands with more cells surviving in the median region than in the lateral regions of the cerebellum. The specific banding pattern of surviving Purkinje cells seen in Fig. 6B is very similar to that observed by Wassef et al. w40x. There is a remarkable correlation between areas of the molecular layer with high 3H6 immunoreactivity and the location of surviving Purkinje cells. In some instances, high 3H6 immunoreactivity appears to be associated with the presence of solitary Purkinje cells. In the vermis of nr mutants there is a midline band and two bands to each side of the midline where the Purkinje cells have degenerated. These bands which have markedly reduced 3H6 immunoreactivity are approximately 50–250 mm wide. In mice, parallel fibers are 1500–2000 mm long Žw39x and M. Vogel, per-

88

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89

sonal communication., and therefore, it is likely that parallel fibers are traversing these bands. Thus, one interpretation of the observed staining pattern is that the parallin is not uniformly distributed along a given parallel fiber, but rather is localized to those regions of the parallel fiber in proximity to Purkinje cells. Another explanation is that the high levels of staining near Purkinje cells is associated with the ascending process of the granule cell which form synapses with Purkinje cell dendritic thorns w29x. In either case, high levels of immunoreactivity in the granule cell axon is associated with proximity to Purkinje cells. There are now more than a dozen molecular markers that exhibit a parasagittal banding pattern in the cerebellum w13x, the most well characterized being the zebrins w1,6x. In the anterior vermis of nr mice, it is the zebrinpositive bands that degenerate w5,40x so that the IHC staining we observe in nr mutants with 3H6 and calbindin appears as a negative image of the zebrin pattern. However, it should be emphasized that in normal mice, parallin is not banded but appears banded in nr mutants secondary to the loss of Purkinje cells. The IHC studies of both pcd and nr homozygous mutants showing a dramatic reduction or in some cases the absence of detectable levels of immunostaining except where there are surviving Purkinje cells strongly supports our hypothesis that Purkinje cells are involved in maintaining the expression of parallin in the granule cells. The relative abruptness of the boundaries between stained and unstained areas with mAb 3H6 suggest that direct contact between Purkinje dendrites and parallel fibers may be required for maintaining parallin expression as well as localizing it in the parallel fiber possibly to synaptic regions. If expression is regulated by the Purkinje cells, why is there residual expression in both pcd and nr months after the Purkinje cells have degenerated? One possibility is that the residual expression is associated with the parallel fibers as they contact their other normal neuronal targets in the molecular layer, that is, stellate, basket and Golgi cells. Alternatively, the gene for parallin may be constitutively expressed, and it is only the level of expression that is regulated by Purkinje cells. In pcd and nr mutants, the high residual expression in the granule cell layer in the hemispheres is puzzling. The high levels in the granule cell layer might indicate that it is being synthesized but not transported, but why it should be higher in the hemispheres is not clear. 4.3. Conclusion The results presented here strongly suggest that the antigen identified by mAb 3H6 is primarily located in cerebellar granule cells and in particular their axons, the parallel fibers. This is supported by the observed staining pattern, the time of appearance of the antigen and the absence of the antigen in the granule cell-deficient wÕ and

sg mutants. The results also suggest that the expression of this granule cell antigen is regulated by Purkinje cells. This is supported by the onset of expression of the antigen at the time granule cells are contacting and synapsing with Purkinje cells but most dramatically by the results with the pcd and nr mutants. The staining pattern in nr mutants suggest that the antigen is primarily located in those regions of parallel fiber in contact with Purkinje cells, in other words in the area of the parallel fiber:Purkinje cell synapse. Whether the antigen is localized to the synapse or is only in the vicinity remains to be determined. Together these findings support our working hypothesis that interaction with the Purkinje cells is necessary for the expression and maintenance of parallin in the granule cells. The mAb 3H6 has been used to screen a mouse cerebellar-specific expression library and a candidate clone of the cognate gene has been identified ŽK. Suh and N. Heintz, personal communication., indicating that the antigen is a protein. Their Northern blots with this clone correlate extremely well with our results. Future work to characterize this protein at the cellular and molecular level should help in elucidating some of the mechanisms involved in the developmental processes which result in the organization of the mammalian brain.

References w1x A.H. Ahn, S. Dziennis, R. Hawkes, K. Herrup, The cloning of zebrin II reveals its identity with alsolase C, Development 120 Ž1994. 2081–2090. w2x J. Altman, Autoradiographic and histological studies of postnatal neurogenesis: III. Dating the time of production and onset of differentiation of cerebellar microneurons in rats, J. Comp. Neurol. 136 Ž1969. 269–294. w3x P.W. Beesley, T. Paladino, I. Hill, C. Gravel, R.B. Hawkes, J.W. Gurd, Postnatal development of a granule cell-enriched, neuronespecific glycoprotein, gp50, in normal and thyroid-deficient rats, J. Neurochem. 54 Ž1990. 505–512. w4x J. Dodd, S. Morton, D. Karagogeos, M. Yamamoto, T.M. Jessell, Spatial regulation of axonal glycoprotein expression on subsets of embryonic spinal neurons, Neuron 1 Ž1988. 105–116. w5x M.A. Edwards, J.E. Crandall, N. Leclerc, M. Yamamoto, Effects of nervous mutation on Purkinje cell compartments defined by Zebrin II and 9-O-acetylated gangliosides expression, Neurosci. Res. 19 Ž1994. 167–174. w6x L.M. Eisenman, R. Hawkes, Antigenic compartmentation in the mouse cerebellar cortex: zebrin and HNK-1 reveal a complex, overlapping molecular topography, J. Comp. Neurol. 335 Ž1993. 586–605. w7x C. Faivre-Sarrailh, G. Gennarini, C. Goridis, G. Rougon, F3rF11 cell surface molecule expression in the developing mouse cerebellum is polarized at synaptic sites and within granule cells, J. Neurosci. 12 Ž1992. 257–267. w8x N. Feder, Solitary cells and enzyme exchange in tetraparental mice, Nature 262 Ž1976. 67–69. w9x D. Goldowitz and L.M. Eisenman, Genetic mutations affecting murine cerebellar structure and function. in: P. Driscoll ŽEd.., Genetically Defined Animal Models of Neurobehavioral Dysfunction, Birkhauser Press, Boston, MA, 1992, pp. 66–88.

A.M. Smith, R.J. Mullenr DeÕelopmental Brain Research 104 (1997) 79–89 w10x C. Gravel, N. Leclerc, J. Rafrafi, R. Sasseville, L. Thivierge, R. Hawkes, Monoclonal antibodies reveal the global organization of the cerebellar cortex, J. Neurosci. Methods 21 Ž1987. 145–157. w11x W.J. Hendelman, A.S. Aggerwal, The Purkinje neuron: I. A Golgi study of its development in the mouse and in culture, J. Comp. Neurol. 193 Ž1980. 1063–1079. w12x K. Herrup, Role of the staggerer gene in determining cell number in the cerebellar cortex: I. Granule cell death is an indirect consequence of staggerer gene action, Dev. Brain Res. 11 Ž1983. 267–274. w13x K. Herrup, B. Kuemerle, The compartmentalization of the cerebellum, Annu. Rev. Neurosci. 20 Ž1997. 61–90. w14x K. Herrup, R.J. Mullen, Staggerer chimeras: intrinsic nature of Purkinje cell defects and implications for normal cerebellar development, Brain Res. 178 Ž1979. 443–457. w15x K. Herrup, E. Trenkner, Regional differences in cytoarchitecture of the weaver cerebellum suggest a new model for weaver gene action, Neuroscience 23 Ž1987. 871–885. w16x S. Hockfield, R.D.G. McKay, Identification of major cell classes in the developing mammalian nervous system, J. Neurosci. 5 Ž1985. 3310–3328. w17x F. Keller, K. Rimvall, M.F. Barbe, P. Levitt, A membrane glycoprotein associated with the limbic system mediates the formation of the septo-hippocampal pathway in vitro, Neuron 3 Ž1989. 551–561. w18x S.G. Kuhar, L. Feng, S. Vidan, M.E. Ross, M.E. Hatten, N. Heintz, Changing patterns of gene expression define four stages of cerebellar granule neuron differentiation, Development 117 Ž1993. 97–104. w19x D.M. Landis, R.L. Sidman, Electron microscopic analysis of postnatal histogenesis in the cerebellar cortex of staggerer mutant mice, J. Comp. Neurol. 179 Ž1978. 831–864. w20x S.C. Landis, Ultrastructural changes in the mitochondria of cerebellar Purkinje cells of nervous mutant mice, J. Cell Biol. 57 Ž1973. 782–797. w21x S.C. Landis, R.J. Mullen, The development and degeneration of Purkinje cells in pcd mutant mice, J. Comp. Neurol. 177 Ž1978. 125–144. w22x L.M.H. Larramendi, Analysis of synaptogenesis in the cerebellum of the mouse. in: R. Llinas ŽEd.., Neurobiology of Cerebellar Evolution and Development, AMA, Chicago, IL, 1969, pp. 803–843. w23x H. Lovtrup-Rein, B.S. McEwen, Isolation and fractionation of rat brain nuclei, J. Cell Biol. 30 Ž1966. 405–415. w24x R.D.G. McKay, S.J. Hockfield, Monoclonal antibodies distinguish antigenically discrete neuronal types in the vertebrate central nervous system, Proc. Natl. Acad. Sci. USA 79 Ž1982. 6747–6751. w25x I.W. McLean, P.K. Nakane, Periodate-lysine-paraformaldehyde fixative, a fixative for immunoelectron microscopy, J. Histochem. Cytochem. 22 Ž1974. 1077–1083. w26x R.J. Mullen, Ontogeny and genetics of the mammalian nervous system. in: N. LeDouarin and A. McLaren ŽEds.., Chimeras in Developmental Biology, Academic Press, London, 1984, pp. 353– 368.

89

w27x R.J. Mullen, C.R. Buck, A.M. Smith, NeuN, a neuronal specific nuclear protein in vertebrates, Development 116 Ž1992. 201–211. w28x R.J. Mullen, E.M. Eicher, R.L. Sidman, Purkinje cell degeneration, a new neurological mutation in the mouse, Proc. Natl. Acad. Sci. USA 73 Ž1976. 208–212. w29x S.L. Palay and V. Chan-Palay, Cerebellar Cortex: Cytology and Organization, Springer, New York, 1974, 348 pp. w30x R.L. Sidman, Development of interneuronal connections in brains of mutant mice. in: F.D. Carlson ŽEd.., Physiological and Biochemical Aspects of Nervous Integration, Prentice-Hall, Inglewood Cliffs, NJ, 1968, pp. 165–193. w31x R.L. Sidman, Neuropathology, mutant studies. in: G. Adelman ŽEd.., Encyclopedia of Neuroscience, Vol. II, Birkhauser, Boston, MA, 1987, pp. 828–832. w32x R.L. Sidman and M.C. Green, ‘Nervous’, a new mutant mouse with cerebellar disease. in: M. Sabourdy ŽEd.., Les Mutants Pathologiques chez l’Animal, Editions du Centre National de la Recherche Scientifique, CNRS, Paris, 1970, pp. 69–79. w33x R.J. Smeyne, D. Goldowitz, Development and death of external granular layer in the weaver mouse cerebellum: a quantitative study, J. Neurosci. 9 Ž1989. 1608–1620. w34x C. Sotelo, Anatomical, physiological and biochemical studies of the cerebellum from mutant mice: II. Morphological study of cerebellar cortical neurons and circuits in the weaver mouse, Brain Res. 94 Ž1975. 19–44. w35x C. Sotelo, J.P. Changeux, Transsynaptic degeneration ‘en cascade’ in the cerebellar cortex of staggerer mutant mice, Brain Res. 67 Ž1974. 519–526. w36x C. Sotelo, A. Triller, Fate of presynaptic afferents to Purkinje cells in the adult nervous mutant mouse: a model to study presynaptic stabilization, Brain Res. 175 Ž1979. 11–36. w37x T. Suzue, Z. Kaprielian, P.H. Patterson, A monoclonal antibody that defines rostrocaudal gradients in the mammalian nervous system, Neuron 5 Ž1990. 421–431. w38x L.C. Triarhou, J. Norton, C. Alyea, B. Ghetti, A quantitative study of the granule cells in the Purkinje cell degeneration mutant, Ann. Neurol. 18 Ž1985. 146. w39x M.W. Vogel, J. Soha, J. Prittie, J.E. Crandall, Analysis of parallel fiber elongation in the wild type and mutant mouse cerebellum, Soc. Neurosci. Abstr. 19 Ž1993. 1089. w40x M. Wassef, C. Sotelo, B. Cholley, A. Brehier, M. Thomasset, Cerebellar mutations affecting the postnatal survival of Purkinje cells in the mouse disclose a longitudinal pattern of differentially sensitive cells, Dev. Biol. 124 Ž1987. 379–389. w41x P.L. Woodhams, M. Webb, A developmentally regulated axonal glycoprotein Ž7-8D2 antigen. with a restricted distribution in mature rat brain, Neuroscience 32 Ž1989. 417–434.