INTRODUCTION

Sulfoglucuronylglycolipids (SGGLs)¹ are developmentally regulated lipids expressed mostly in specific areas of the nervous system (1). These glycolipids have been characterized as 3-sulfoglucuronyl neolactotetraosyl ceramide (SGGL-1) and 3-sulfoglucuronyl neolactohexaosyl ceramide (SGGL-2) (2, 3, 4, 5). Monoclonal antibody HNK-1 reacts with the terminal nonreducing 3-sulfoglucuronic acid of the glycolipids (3, 6). The HNK-1 carbohydrate epitope is also expressed on certain glycoproteins and proteoglycans in the nervous system (7, 8, 9). Several of these glycoproteins have been shown to be cell-adhesion molecules, such as N-CAM, L1, Ng-CAM, Nr-CAM, TAG-1, MAG, and others (10, 11, 12, 13, 14). Based on these observations, the HNK-1 reactive carbohydrate has now been considered to be a marker of cell adhesion molecules in the nervous system (15). Previous studies have suggested that HNK-1 carbohydrate is involved in the organization of early neuronal settling and axonal outgrowth patterns (7, 16, 17, 18).

The expression of SGGLs in several brain regions during development is temporally regulated. In the rat cerebral cortex, SGGLs are expressed maximally at embryonic days 18-20 and they almost disappear by postnatal day (PD) 15 (19, 20). In the rat cerebellum, the level of SGGL expression is biphasic, with an initial peak at PD 1-3, followed by a second maximum at PD 20 and continued expression in the adult (21). In the peripheral nerves, such as in the rat sciatic nerves, the levels of SGGLs increase rapidly from PD 1 to 30 and after 60 days, they slightly decline (4). The pathway for the biosynthesis of SGGL-1 is shown in Fig. 1. To understand the mechanism of differential regulation of expression of SGGLs and its physiological significance, particularly in the cerebral cortex and cerebellum, the biosynthetic enzymes involved in the synthesis of SGGLs during development were studied (22, 23, 24, 25). It was shown that the activities of the enzymes, lactotriaosylceramide:galactosyltransferase (LcOse₃Cer:Gal-Tr), neolactotetraosyl-ceramide:glucuronyltransferase (nLcOse₄Cer:GlcA-Tr), and glucuronyl glycolipid:sulfotransferase (GGL-1:SO₄-Tr) did not correlate with the expression of SGGL-1 during development. Considerable activities of these three enzymes persisted in the adult rat cerebral cortex, despite almost complete loss of SGGLs in this tissue. This suggested that these enzymes were not regulating the expression of SGGLs in the cerebral cortex. However, developmental expression of lactosylceramide:N-acetylglucosaminyltransferase (GlcNAc-Tr), both in the cerebral cortex and cerebellum correlated well with the tissue levels of SGGLs (25). In the cerebral cortex, the specific activity of GlcNAc-Tr declined sharply and was down-regulated from a maximum at embryonic day 15 to undetectable levels by PD 10. In the cerebellum, the activity of GlcNAc-Tr correlated with biphasic expression of SGGLs (25). Analysis of the levels of the product of the GlcNAc-Tr reaction, viz. LcOse₃Cer, also correlated with the enzyme activity (25). In the cerebral cortex, the level of LcOse₃Cer declined with development and it was undetectable by PD 5, whereas in the cerebellum it increased with development (25). These results suggested that GlcNAc-Tr was the key regulatory enzyme controlling the differential and stage-specific expression of SGGLs in the developing nervous system (25).

Although the embryonic expression and subsequent down-regulation of SGGLs in the cerebral cortex have been shown to be associated with the migration and maturation of the neuronal cells, the physiological significance and the biphasic expression of SGGLs in the cerebellum remain enigmatic. Our immunocytochemical and biochemical studies have shown that in the adult rodent cerebellum SGGLs were associated only with Purkinje cells and their dendrites in the molecular layer (21, 26, 27).² Whereas, the granule neurons of the internal granule cell layer and the glial cells of the white matter did not express SGGLs. However, it is not known which type of cells in the neonatal rat cerebellum express SGGLs that are subsequently down-regulated with development.

Studies on cerebellar murine mutants have been very valuable, since in many mutants loss of specific cell types as well as defects in histogenesis and cytoarchitectural abnormalities are well defined. Previously we have shown that SGGLs were undetectable in the cerebella of murine mutants, such as Purkinje cell degeneration (pcd/pcd), lurcher (Lc/+), and staggerer (sg/sg), where Purkinje cell loss is the primary defect (27, 28, 29). In nervous mutant (nr/nr) where Purkinje cell loss is not as severe or in leaner (tg^la/tg^la) where selective degeneration of Purkinje cells occurs, the loss of SGGLs was also limited (27, 28, 29). The loss of SGGLs and other glycolipids derived from LcOse₃Cer such as, lactoneotetraosylceramide, lactoneohexaosylceramide, Le^x antigens, and ganglioside nLD1, in these mutants was specific, since other lipids were not affected significantly. The expression of SGGLs was essentially normal in other cerebellar mutants, such as weaver, reeler, and quaking, where cells other than Purkinje cells are primarily affected (27, 28, 29).

In this study we have examined the activities of the four biosynthetic enzymes, involved in the synthesis of SGGLs, in the cerebellum of Purkinje cell abnormality mutants to evaluate which enzymes are affected due to specific loss of Purkinje cells. The activities of the enzymes were also determined in the micro-dissected layers of cryocut sections of the normal adult cerebellum to localize the site of synthesis of SGGLs in specific cellular layers of cerebellum. In addition, the activities of the enzymes were determined in isolated granule neurons at different ages to evaluate if granule neurons were capable of synthesizing SGGLs. Parts of this work were previously presented in abstract form (30, 31).

MATERIALS AND METHODS

Sprague-Dawley albino rats were purchased from Taconic Farms, Inc. (Germantown, NY). Mutant mice were reared in our animal colony (27, 28, 29). Radioactive sugar nucleotides were purchased from DuPont NEN. Non-radioactive sugar nucleotides, lactosylceramide, and other chemicals of the highest grade possible were purchased from Sigma. LcOse₃Cer, nLcOse₄Cer, GGL-1, and other glycolipids were prepared as described previously (22, 23, 24, 25).

Mutant or normal litter-mate mouse (adult) cerebella were homogenized in 10 volumes of 0.32 M sucrose in 10 mM Tris-HCl, pH 7.4, with a Polytron tissue homogenizer (Brinkmann Instruments) at 0-4 °C. Protein concentrations were determined using the bicinchoninic acid reagent (Pierce Chemical Co.).

Adult rats (~60 days old) were anesthetized with ether and killed by decapitation. Cerebella were removed immediately, placed on a microtome object disk, and frozen on dry ice for 15 min. Cerebellar sections (parasagittal, 15 µm) were cut in a cryostat at 10 °C, transferred to plastic Petri dishes, and lyophilized for 5-8 h. Freeze-dried sections were dissected under a stereo-microscope using micro tools according to the procedure of Lowry (32). Care was taken to ensure that the Purkinje cell layer was dissected into the molecular layer, whereas granular layer and white matter were collected together. Areas containing deep cerebellar nuclei were avoided. The cut layers were stored in a desiccator at 0-4 °C until used for enzyme assays. The layers were homogenized in a small Potter-Elvehjem homogenizer as described previously (22).

The activity of lactosylceramide:GlcNAc-Tr was measured in the homogenates of the mouse cerebellum or in rat cerebellar layers with UDP-[¹⁴C]GlcNAc as the donor and externally added lactosylceramide as the acceptor as described previously (25). The activity of the enzyme without the externally added acceptor lactosylceramide was also determined to see the intrinsic activity. The product of the reaction was identified after purification, by HPTLC, autoradiography, and by immunoreactivity with monoclonal antibody TE-5 (a gift from Dr. Eric Holmes) as described previously (33). The optimal conditions for the assay and product identification for the other three enzymes in the biosynthesis of SGGL-1, viz. LcOse₃Cer:Gal-Tr, nLcOse₄Cer:GlcA-Tr, and GlcAnLcOse₄Cer:SO₄-Tr, have been previously established and were used accordingly (22, 23, 24).

The granule neurons were isolated from the Sprague-Dawley rats of three different age groups between: (I) 2 and 4 days, PD 3 group, (ii) 7 and 9 days, PD 8 group, and (iii) 13 and 15 days, PD 14 group, by a slight modification of the procedure of Hatten (34) developed for mice. Briefly, the meninges were carefully stripped off and the whole tissue was washed in Ca²⁺ and Mg²⁺-free Tyrode's solution (CMF-phosphate-buffered saline). The cells were dissociated in to single cells as described previously (34). After passing the cell suspension through a polyester 33-µm mesh size screen, the suspension was applied to a three-step gradient of Percoll comprising of 60, 35, and 20% (3.0 ml of each) in Ca²⁺ and Mg²⁺-free Tyrode's solution. The gradients were centrifuged in a IEC centra-8R centrifuge at 3300 rpm for 10 min. The three resulting bands at the interface of the gradients were removed, diluted to 10 ml with Ca²⁺ and Mg²⁺-free Tyrode's solution, and pelleted at 1000 rpm for 7 min. The granule cells were collected from the interface of 35/60% Percoll gradient, whereas astroglial cells were enriched at the interface of 20/35% Percoll. The collected cells were further washed with Dulbecco's modified phosphate buffer and resuspended in Dulbecco's modified phosphate buffer. A portion of the cells were counted with a hematocytometer. The rest were either stored at 20 °C for biochemical analyses or resuspended in complete medium supplemented with 10% newborn calf serum and 2.5% horse serum, 25 mM KCl, 2.0 mM glutamine, 0.6% glucose, and 100 µg/ml gentamicin and plated on polylysine (50 µg/ml)-coated coverslips at a density of 2 × 10⁵ cells per 15-mm diameter coverslip. The purity of the granule cell preparations was monitored by immunocytochemical staining of the cells with appropriate marker antibodies and by Western blot analyses. The presence of Purkinje cells was monitored by reaction with antibodies to calbindin-28. Antibodies (polyclonal) to SGGL-binding p30 protein was used to monitor the number of granule cells. Total number of cells and appropriate antibody-labeled fluorescent cells were counted in several randomly selected fields using the computer-assisted IBAS image analysis system (Kontron, Germany). To monitor the contamination by astroglial cells, proteins in the homogenates of granule cells were analyzed by Western blot analysis with antibodies to glial fibrillary acidic protein. The intensity of the bands on Western blots was determined by computer-assisted VISAGE image analysis system (BioImage-Millipore, MI).

RESULTS

Glycosyltransferases Synthesizing SGGL-1, in Purkinje Cell Abnormality Mouse Mutants

UDP-[¹⁴C]GlcNAc and lactosylceramide were incubated with homogenates of cerebella from two Lc/+ mutant and three normal littermate control mice and the radioactive products formed after the enzymatic reaction were isolated and separated by HPTLC (Fig. 2A). An autoradiogram of the HPTLC plate shows that with the three samples of normal cerebellum, the major radioactive product (Fig. 2A, panel A, lanes 3-5) co-migrated with standard LcOse₃Cer (LC₃ in Fig. 2A, panel C, lane 8). The radioactive product also reacted with TE-5 antibody (not shown) which is specific for terminal GlcNAc1-3Gal1-4- structure (32). However, Lc/+ mutant cerebellar homogenates failed to synthesize any radioactive product from lactosylceramide (Fig. 2A, panel A, lanes 1 and 2). The same HPTLC plate as visualized after orcinol spray is shown in Fig. 2A, panel B. The doublet bands migrating with standard lactosylceramide correspond to added acceptor lactosylceramide prepared from bovine red blood cells.

Without exogenously added acceptor lactosylceramide, normal littermate cerebellar homogenates also synthesized a trace amount of LcOse₃Cer (autoradiogram in Fig. 2B, panel A, lanes 1 and 2). However, the mutant cerebellum failed to synthesize any LcOse₃Cer under the same conditions (Fig. 2B, panel A, lane 3). The identity of the other radioactive products in Fig. 2B, panel A, is not known. It should be noted that the exposure time of the HPTLC plate to x-ray film for enzyme assays without exogenously added substrates (Fig. 2B) was 10 times longer than that with the added substrates (Fig. 2A), therefore the unidentified radioactive products seen in Fig. 2B are not seen in Fig. 2A. The same HPTLC plate after autoradiography was treated with orcinol spray and is shown in Fig. 2B, panel B. Only a trace, if any, endogenous lactosylceramide in the homogenates appears be present.

The radioactive products formed by the reaction of LcOse₃Cer:Gal-Tr and nLcOse₄Cer:GlcA-Tr with normal and Lc/+ mutant mouse cerebellar homogenates are shown in Fig. 3A. When exogenously added substrate LcOse₃Cer (Lc₃) was used as an acceptor (Fig. 3A, panel B, lanes 4-6) and UDP-[¹⁴C]Gal as a donor, both the normal and the mutant homogenates synthesized radioactive nLcOse₄Cer (nLc4, autoradiogram in Fig. 3A, panel A, lanes 4-6) as the major product which migrated with standard nLcOse₄Cer (Fig. 3A, panel B, lanes 1-3, lower most band, which is the exogenously added substrate for the GlcA-Tr reaction).

Without added LcOse₃Cer, normal cerebellar homogenates also synthesized trace amounts of nLcOse₄Cer doublet from endogenous LcOse₃Cer and UDP-[¹⁴C]Gal (autoradiogram in Fig. 3B, panel A, lanes 4 and 6). However, the Lc/+ mutant cerebellar homogenate synthesized barely a trace amount of nLcOse₄Cer, due to lack of endogenous LcOse₃Cer (Fig. 3B, panel A, lane 5).

Both normal and mutant cerebellar homogenates synthesized radioactive GlcA-nLcOse₄Cer (GGL-1), when incubated with exogenously added nLcOse₄Cer and UDP-[¹⁴C]GlcA (autoradiogram in Fig. 3A, panel A, lanes 1-3). Without exogenously added nLcOse₄Cer, although the normal littermate cerebellar homogenate synthesized trace levels of GlcA-nLcOse4Cer (autoradiogram in Fig. 3B, panel A, lanes 1 and 3), the mutant cerebellar homogenate failed to synthesize any product (Fig. 3B, panel A, lane 2). The latter result is interpreted as due to lack of availability of endogenous substrates nLcOse₄Cer or its precursor LcOse₃Cer in the mutant.

Normal and Lc/+ mutant cerebellar homogenate catalyzed the formation of radioactive SGGLs from exogenously added GGL-1 and PAP [³⁵S]O₄ (autoradiogram in Fig. 4A, panel A, lanes 4-8). Without exogenously added GGL-1 both the normal and mutant cerebellar homogenates failed to synthesize SGGL-1 (Fig. 4A, panel A, lanes 1-3).

The specific activities of the four enzymes, involved in the biosynthesis of SGGL-1 from lactosylceramide, in the Purkinje cell abnormality mutants, pcd/pcd, Lc/+, and tg^la/tg^la and their normal littermates' cerebella are given in Table I. The specific activities of Gal-Tr and SO₄-Tr were nearly normal or slightly elevated in all the three Purkinje cell abnormality mutants and GlcA-Tr was about 50% normal. However, GlcNAc-Tr was not detectable in pcd/pcd and Lc/+ cerebellum. In tg^la/tg^la, where Purkinje cells are not completely lost, the specific activity of GlcNAc-Tr was about 25% of normal.

Table I. Activities of enzymes involved in the biosynthesis of SGGLs in cerebellar homogenates of Purkinje cell abnormality mutants and their normal littermates The specific activities of the enzymes in the homogenates of the cerebellum from two to three individual animals were determined as previously described (22, 23, 24, 25). Each assay was done in duplicate or triplicate and the average values are given. Within individual groups, the variation in the activity was 13% or less of the average value. Mice GlcNAc-Tr Gal-Tr GlcA-Tr SO4-Tr specific activity (pmol/mg/h) Normal 6.4 993 35 72 pcd/pcd 0 1062 18 81 Normal 7.8 1020 22 73 Lc/+ 0 1225 12 80 Normal 7.0 603 45 75 tgla/tgla 1.7 754 27 90

Expression of Glycosyltransferases in Normal Rat Cerebellar Layers

Previously, we have shown that in the adult rat the expression of SGGLs was restricted to molecular layer where Purkinje cell dendritic arborization occurs and to deep cerebellar nuclei where Purkinje cell axons terminate (21, 26, 27). SGGLs were undetectable in granule cell layer and white matter. In order to investigate if all four glycosyltransferases that are involved in the biosynthesis of SGGLs are also restricted to the same layers, we analyzed the activity of the enzymes in micro-dissected young-adult rat cerebellar layers from freeze-dried cryocut sections. In initial experiments, we ensured that each of the enzyme activities was not significantly altered due to freeze-drying as compared to fresh tissue homogenates. The expression of the four glycosyltransferases in the Purkinje cell/molecular layer versus those in the granule cell layer and white matter, as well as the activities in the total freeze-dried cryocut sections (which included all layers) are shown in Fig. 5. The specific activities of Gal-Tr, GlcA-Tr, and SO₄-Tr in the granule cell layer and white matter was approximately 66, 123, and 50% of that in the molecular layer. However, the specific activity of GlcNAc-Tr in the granule cell layer and white matter was less than 8% of that in the Purkinje cell/molecular layer. The specific activity of GlcNAc-Tr in the Purkinje cell/molecular layer was almost 2.5-fold higher than in the total freeze-dried cryocut sections, indicating enrichment of this activity and therefore localization of the enzyme mostly in the Purkinje cell/molecular layer.

Expression of Glycosyltransferases in Isolated Granule Neurons from Normal Rat Cerebellum during Development

Granule cells were isolated from PD 3, 8, and 14 rat cerebella and the purity of the isolated cells was checked by counting immunoreactive cells using antibodies to SGGL-binding p30 protein which is associated mostly with the granule neurons³ and with antibodies to calbindin-28, a marker of Purkinje cells. Purity was also checked by comparative Western blot analyses of glial fibrillary acidic protein, a marker of astroglial cells, in the homogenates of isolated cells versus that in the whole cerebellum, at the respective ages. At PD 3, 8, and 14, 10, 7, and 3% of the total number of cells counted in the granule cell preparations, respectively, were not labeled with the granule cell marker p30. Of the total cells, 0.2-0.7% of the cells were reactive with the Purkinje cell marker, calbindin-28. Similar results were obtained by Western blot analyses of proteins with the calbindin-28 antibody. Western blot analyses of proteins with glial fibrillary acidic protein antibody showed the presence of 8, 6, and 3% glial fibrillary acidic protein in cells isolated at PD 3, 8, and 14, compared to that in the homogenates of whole cerebella at the same ages. The results indicate that the isolated granule cells were virtually free of Purkinje cells and depending upon the age of isolation may contain 3-10% astroglial cells.

The activities of the four glycosyltransferases were determined in the isolated cells and also in the total homogenates of the cerebellum at the same age (Table II). The activity of the GlcNAc-Tr in the granule cells isolated from rat cerebellum at PD 14 was approximately 50% of that at PD 3 and 8. The activity of the enzyme in the homogenate, however, increased approximately 4-fold between PD 3 and PD 14. The relative GlcNAc-Tr activity in granule cells compared to that in the homogenate declined from 1.5 to 0.2, between PD 3 and 14. The activity of the enzyme Gal-Tr in the homogenate declined by approximately 35-45% from PD 3 to 14. However, the Gal-Tr activity in the granule cells did not change with age. The relative Gal-Tr activity in granule cells to homogenate was between 0.13 and 0.23. The activity of GlcA-Tr declined by approximately 50% in the homogenate, but in the granule cells it declined by about 25% at PD 14. The ratio of the activity in the granule cell versus homogenate increased by 1.7-fold from PD 3 to 14. The activity of SO₄-Tr during PD 3-14 did not change much in the granule cells and in the homogenate it declined by 45% at PD 14. The ratio of this activity thus increased by 1.9-fold.

Table II. Expression of enzyme activities, involved in the biosynthesis of SGGLs, in isolated rat-cerebellar granule cells and in homogenates of cerebellum, during neonatal development Values are given as average ± S.D. The entire experiment was done three times. The specific activities of the enzymes were determined as previously described (22, 23, 24, 25) in duplicate or triplicate. Age (days) GlcNAc-TR Gal-TR GlcA-TR SO4-TR specific activity (pmol/mg/h) Granule cells   3 9.1  ± 1.1 209  ± 23 4.1  ± 0.5 16.0  ± 1.7   8 8.6  ± 1.9 205  ± 18 4.7  ± 0.7 16.4  ± 1.2   14 5.6  ± 0.6 198  ± 15 3.3  ± 0.4 15.0  ± 2.4 Cerebellum   3 5.9  ± 0.2 1630  ± 20 30.5  ± 4.0 56.0  ± 5.1   8 8.8  ± 2.5 878  ± 114 21.5  ± 1.3 48.2  ± 2.4   14 25.0  ± 2.8 1086  ± 106 15.2  ± 0.8 27.8  ± 1.6 Ratio: granulecells/cerebellum   3 1.54 0.13 0.13 0.29   8 0.98 0.23 0.22 0.34   14 0.24 0.18 0.22 0.54

DISCUSSION

Recently we have identified several neutral and acidic neolactoglycolipids in the developing rat nervous system (1, 2, 3, 4, 20, 21, 29, 35, 36). These glycolipids are expressed in the cerebral cortex mostly during embryonic stages of development and disappear postnatally. However, in the cerebellum and peripheral nervous system, the levels of neolactoglycolipids increase postnatally. This temporal and stage-specific regulation of most of the neolactoglycolipids has been ascribed to the activity of an enzyme N-acetylglucosaminyl transferase, involved in the biosynthesis of the key precursor LcOse₃Cer (1, 25).

A central problem in mammalian brain development is the identification of regulatory signals for migration of neural cells and their arrest of migration after reaching appropriate targets. Although cell-cell interactions have long been assumed to provide cues that guide neural cells and their axons, the precise nature of molecules involved in this process have not been well defined (34). Several lines of evidence indicate that carbohydrate moieties on adhesion molecules can modulate the function of their carrier molecules or can themselves participate as ligands in cell recognition and adhesion processes (10, 17, 37, 38). For example, Le^x antigen has been implicated as a signal for neural cell proliferation near the ventricular zone of the cerebral cortex of embryonic rodents (38, 39). In contrast, antibody HNK-1 reactive antigens (3SO₄GlcA-nLcOse₄-) are specifically expressed on the neural cells that have migrated and reached their targets to the molecular layer and subplate in the embryonic cerebral cortex. These antigens subsequently disappear post-natally (19). Thus the expression and down-regulation of the HNK-1 carbohydrate has been proposed as a "stop signal" for the migrating neurons.

The biphasic expression of SGGLs in the rat cerebellum appears to coincide with the cell migration and maturation processes. Thus, in the rat cerebellum, SGGLs that are expressed maximally at around PD 1-3 and which decline till PD 7 appear to correlate with the migration of the external granule neurons, guided by Bergmann glial fibers, into the internal granule cell layer which occurs maximally between PD 1 and 7. Whereas, the latter rise in the expression of SGGLs in the cerebellum between PD 7 and 20 appears to be due to the lining up of the Purkinje cells and their dendritic arborization, initiating at PD 7 and reaching a maximum at PD 20-25.

Our immunocytochemical data in adult rat and mouse as well as studies with the Purkinje cell abnormality murine mutants have shown that in the adult rodent cerebellum, SGGLs are associated with Purkinje cells and their dendrites in the molecular layer (21, 26, 27, 28, 29). Our data presented in Figs. 2, 3, 4 and Table I show that SGGLs are not synthesized in the cerebella of mutants such as pcd/pcd and Lc/+ where Purkinje cell loss is the primary event and other cells are not as severely affected. In addition, the inability of the pcd/pcd and Lc/+ mutants' cerebella to synthesize SGGLs is due to lack of activity of the key enzyme GlcNAc-Tr (Table I). About 25% remaining activity of the GlcNAc-Tr seen in the leaner mutant cerebellum is due to the survival of some Purkinje cells in this mutant, where the Purkinje cell loss is not as severe as in the other two mutants.

These results show that in the adult cerebellum, GlcNAc-Tr is only associated with Purkinje cells. Other cell types do not have this enzyme activity. Deficiency of GlcNAc-Tr in the Purkinje cell abnormality mutants would explain severe reduction of all neolactoglycolipids derived from LcOse₃Cer in the mutants (1, 27, 28, 29). Although levels of LcOse₃Cer were not measured in the mutants' cerebella, analysis of its immediate product nLcOse₄Cer showed that it was only 4, <1, and 2.6% of the normal (~3 ng/mg dry weight) in Lc/+, tg^la/tg^la, and pcd/pcd mutants, respectively (27, 29). Previously we have reported the levels of LcOse₃Cer in normal rat cerebellum during development (25). In the adult rat cerebellum, the level of LcOse₃Cer was about 6 ng/mg dry weight (25). The other three enzymes involved in the synthesis of SGGLs, viz. the Gal-Tr, GlcA-Tr, and SO₄-Tr, in the mutants were nearly normal or only slightly reduced (Table I). The latter data indicate that in the adult cerebellum these enzymes are not exclusively associated with Purkinje cells but are also localized in other major cell types, viz. the granule cells and glial cells.

The regulatory role of GlcNAc-Tr in the expression of neolactoglycolipids in specific cell types is not limited to neural cells. Among human leukocyte subclasses, only myeloid cells express neolactoglycolipids due to the presence of GlcNAc-Tr in these cells. Lymphoid cells are devoid of the enzyme and therefore lack the neolactoglycolipids (40). Similarly, many human colonic and gastric adenocarcinomas accumulate large quantities of lacto- and neolactoglycolipids which are regulated and expressed due to activation of GlcNAc-Tr. This enzyme is undetectable in the normal colonic epithelial cells and therefore they lack these glycolipids, although subsequent enzymes involved in the synthesis of these glycolipids are expressed in the normal tissue (41).

The localization of GlcNAc-Tr in Purkinje cells was further confirmed by the direct analyses of the enzymes in the different cell layers of the normal adult cerebellum (Fig. 5). The results showed that less than 8% of the GlcNAc-Tr activity was associated with the granule cell layer and glial cells in the white matter compared to that in the Purkinje cell/molecular layer. Whereas, the other three enzymes expressed about 37-80% activity in the granule cell layer and white matter compared to those found in the Purkinje cell/molecular layer. The results thus show that in the adult rat cerebellum, only Purkinje cells continue to synthesize SGGLs, whereas the synthesis in the other cell types is arrested due to down-regulation of GlcNAc-Tr in these cells with development, even though the other enzymes subsequent to the latter in the pathway of synthesis are available.

Since our results showed that during cerebellar development SGGLs are expressed maximally around PD 1-3 and decline till PD 7, we propose the hypothesis that immature and migrating granule cells of the external granule cell layer express SGGLs and after reaching their targets in the internal granule cells, they down-regulate the synthesis of SGGLs, similar to neuronal cells in the cerebral cortex during development. To evaluate this hypothesis, we analyzed the specific activities of the biosynthetic enzymes in the isolated and enriched granule cells at different ages and compared them with the enzyme activities in the cerebellar homogenates of the same age. The results presented in Table II show that bulk isolated granule cells at early ages do posses GlcNAc-Tr as well as other enzyme activities necessary for the synthesis of SGGLs. The relative proportion of the GlcNAc-Tr activity in the granule cells at PD 3 was 1.5-fold higher than that in the homogenate, indicating that at PD 3 a large portion of the total activity of this enzyme was associated with the immature granule cells. At PD 14, the specific activity of GlcNAc-Tr rapidly declined to only about 20% of the activity in the homogenate. Beyond PD 14, the isolation of granule cells from cerebellum was difficult due to increases in number of other cell types, synapse formation, and myelination. Therefore, it was not possible to determine the activity of the enzymes in isolated granule cells beyond PD 14. During age 3-14 days the relative activities of the other enzymes in the granule cells were not significantly reduced, indicating that the synthesis of SGGLs in the migrating granule cells was regulated by the GlcNAc-Tr. GlcNAc-Tr in these cells is down-regulated with maturation similar to the maturing neuronal cells in the cerebral cortex. It is therefore suggested that expression of SGGLs in the migrating granule cells have a role in guiding the cells to their targets in the internal granule cell layer. Once they reach their destination, the expression of the SGGLs is down-regulated via the down-regulation of the enzyme GlcNAc-Tr. This perhaps is a stop signal as well as a maturation signal for these cells. The continuous expression of SGGLs in the Purkinje cells and their dendrites is perhaps relevant to the plasticity of these cells in forming dendritic arbors even at mature ages.