Characterization and developmental expression of a novel sulfotransferase for the biosynthesis of sulfoglucuronyl glycolipids in the nervous system.

Sulfoglucuronyl glycolipids (SGGLs) are temporally and spatially regulated molecules in the developing nervous system. A novel sulfotransferase (ST) from rat brain which catalyzes the terminal step in the biosynthesis in vitro of SGGLs is described. The enzyme catalyzes a transfer of sulfate from 3'-phosphoadenosine 5'-phosphosulfate to a hydroxyl group on carbon 3 of the terminal glucuronyl residue in IV3 beta-glucuronyl neolactotetraosylceramide (GlcAnLcOse4Cer) and VI3 beta-glucuronyl neolactohexaosylceramide (GlcAnLcOse6Cer) to form 3-sulfated glucuronyl glycolipids. The enzyme is highly specific for glucuronylglycolipids (GGLs) and requires the free-COOH group of the terminal glucuronic acid for reactivity. GGL:ST present in the microsomal membranes requires Mn2+ ions and a nonionic detergent, Triton X-100 for activity. The optimal pH is 7.2 with Tris-HCl buffer and Km values were 7 microM for 3'-phosphoadenosine 5'-phosphosulfate and 29 microM for GlcAnLcOse4Cer. GGL:ST was shown to be different from previously well studied galactocerebroside:sulfotransferase for the synthesis of myelin membrane-specific lipid sulfatide. This conclusion was based upon several criteria, i.e. including different requirements of incubation conditions for maximal activity, substrate competition experiments, different effects of heat, dithiothreitol, NaCl, and pyridoxal phosphate, as well as different profiles of expression of activity during development of the nervous tissues. The two enzymes were also partially resolved on a pyridoxal phosphate-ligated agarose column. Studies on the developmental expression of the GGL:ST in the rat cerebral cortex and cerebellum showed that it is not a regulatory enzyme controlling the expression of SGGLs in these neural tissues.

Sulfoglucuronyl glycolipids (SGGLs) are temporally and spatially regulated molecules in the developing nervous system. A novel sulfotransferase (ST) from rat brain which catalyzes the terminal step in the biosynthesis in vitro of SGGLs is described. The enzyme catalyzes a transfer of sulfate from 3'-phosphoadenosine 5'-phosphosulfate to a hydroxyl group on carbon 3 of the terminal glucuronyl residue in IV3j3-glucuronyl neolactotetraosylceramide (GlcAnLcOse4Cer) and V13j3-glucuronyl neolactohexaosylceramide (GlcAn-LcOsesCer) to form 3-sulfated glucuronyl glycolipids. The enzyme is highly specific for glucuronylglycolipids (GGLs) and requires the freeCOOH group of the terminal glucuronic acid for reactivity. GGL:ST present in the microsomal membranes requires Mn2* ions and a nonionic detergent, Triton X-100 for activity. The optimal pH is 7.2 with Tris-HC1 buffer and I f , values were 7 @f for 3'-phosphoadenosine 5'-phosphosulfate and 29 @ I for GlcAnLcOse4Cer. GGL:ST was shown to be different from previously well studied galactocerebroside:sulfotransferase for the synthesis of myelin membrane-specific lipid sulfatide. This conclusion was based upon several criteria, i.e. including different requirements of incubation conditions for maximal activity, substrate competition experiments, different effects of heat, dithiothreitol, NaCl, and pyridoxal phosphate, as well as different profiles of expression of activity during development of the nervous tissues. The two enzymes were also partially resolved on a pyridoxal phosphate-ligated agarose column. Studies on the developmental expression of the GGL:ST in the rat cerebral cortex and cerebellum showed that it is not a regulatory enzyme controlling the expression of SGGLs in these neural tissues.
Sulfoglucuronyl glycolipids (SGGLs)' are specifically ex-* This study was supported by Grants R01-NS24405 and PO1-HD05505 from the National Institutes of Health and by Contract 100220023Sc from the Department of Mental Retardation of the Commonwealth of Massachusetts. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The complete biosynthesis of SGGLs is not known. Previously we have characterized a novel and specific enzyme glucuronyltransferase from embryonic chicken and rat brain, involved in the synthesis of glucuronyl neolactotetraosylceramide (GGL-1) and glucuronyl neolactohexaosylceramide (GGL-2) from their respective precursors nLcOse4Cer and nLcOsesCer (15,16). In this report we have studied the properties of a novel sulfotransferase involved in the terminal step for the synthesis of SGGLs from the GGLs. It is shown here that the GGL:sulfotransferase (GGL:ST) is different from the well studied ga1actocerebroside:sulfotransferase (Ga1Cer:ST) which is involved in the biosynthesis of sulfatides.
SGGLs are maximally expressed in the developing rodent cerebral cortex only during the pre-and perinatal period and completely disappear from the postnatal and adult cortex (3,17). However, in the rodent cerebellum and peripheral nervous system they increase with development after birth and are maximally expressed in the adult (18,19). The mechanism of regulation of such differential expression and its physiological significance in these neural areas is unclear. Studies on the developmental expression of the enzyme nLcOse4Cer:glucuronyltransferase in the rat cortex and cerebellum indicated that PAP3%, 3'-phosphoadenosine 5'-pho~pho [~~S]sulfate; CHAPS, 3-[(3cholamidopropyl)dimethylammonio]-l-propanesulfonic acid; GM3, I13NeuAcLcCer; GM1, I13NeuAcGgOse4Cer; GDI,, I13NeuAc, IV3NeuAcGgOse4Cer; GDlb, I13(NeuAc)zGgOse4Cer; GT1b, IV3NeuAc, I13(NeuAc)zGgOse4Cer. the activity of the enzyme was not correlative with the expression of SGGLs (16). Considerable activity of the glucuronyltransferase was present in the adult rat cerebral cortex, even though SGGLs were almost completely absent by postnatal day 15 (PD 15). It was therefore of importance to determine if GGL:ST was involved in the regulation of SGGLs. The developmental expression of GGL:ST in the rat cerebral cortex and cerebellum was evaluated. The results show that like the glucuronyltransferase, the sulfotransferase is not a regulatory enzyme controlling the expression of SGGLs. Part of this work has been previously reported in abstract form (20).
Enzyme Preparation-Neural tissues were homogenized in 10 volumes of 0.32 M sucrose in 10 mM Tris-HCI buffer, pH 7.4, with a Potter-Elvehjem glass homogenizer at 0-4 "C. The homogenate was centrifuged at 1,000 X g for 10 min, followed by 12,000 X g for 30 min, to remove nuclear and crude mitochondrial fractions. The microsomal fraction was yielded by centrifuging the post-mitochondrial supernatant at 100,000 X g for 60 min. Protein concentrations were determined by using the Bicinchoninic acid reagent (Pierce Chemical CO.).
Assay for GlcAnLcOse,Cer:ST and GalCerST-The complete incubation mixture contained, in a volume of 100 p l , 100-250 pg of microsomal proteins or total homogenate proteins from appropriate neural tissue, 100 mM Tris-HCI buffer, pH 7.2, 0.1% Triton X-100, 2.5 mM ATP, 10 mM MnC12, 4 pg of GlcAnLcOse,Cer, and 20 p~ PAP3% (approximately lo6 dpm, -500 dpm/pmol). The incubation was for 45 min at 37 "C. For the assay of GalCer:ST, the incubation conditions were the same, except 0.2% Triton X-100, 15 mM MnCI2, and 5 pg of GalCer instead of GGL-1 were used. After the incubation, 2 ml of chloroform/methanol (2:l) was added, and the radioactive lipid products formed were separated from the radioactive nucleotide precursor by a reversed-phase Cle-BondElut cartridge (Analytichem International, Harbor City, CA) essentially as described previously (22). A portion of the lipid products was used for radioactivity measurement, and the rest was used to separate the radioactive GalCer-SOI (sulfatide) from SGGL-1, by HPTLC with a solvent mixture of chloroform, methanol, 0.25% CaCI2 (5:4:1). When the mixed substrates (i.e. GalCer and GGLs) were present in the reaction mixture, the radioactive products were separated on the HPTLC plate and visualized by autoradiography, and the proportion of radioactive incorporation was determined by scanning the autoradiogram using a Visage 110 Computer-assisted scanner (BioImage/Millipore, Ann Arbor, MI). The HPTLC plate was also exposed to HNK-1 antibody, followed by horseradish peroxidase-conjugated goat antimouse IgM and 4-chloronaphthol to visualize the sulfoglucuronylglycolipids formed (2,19).
Periodate Oxidation-Borohydride Reduction-Experiments were carried out to determine the point of attachment of the sulfate group to the acceptor GGLs. The radioactive lipid reaction products formed from PAP3%, and 10 pg of mixed GGLs were dissolved in 270 p1 of 0.2 M sodium acetate buffer, pH 4.4, and 30 p l of 0.5 M sodium metaperiodate was added at 4 "C (23). After 48 h at 0-4 "C, 50 ~1 of 10 M ethylene glycol was added to eliminate excess periodate. After standing for 3 h, the pH of the reaction mixture was adjusted to about p H 6.5 by adding 90 p l of 0.2 N NaOH, and 60 p l of 9% sodium borohydride was added at 0-4 "C (final pH 10-11). After standing overnight at 4 "C, excess borohydride was decomposed on addition of 70 p1 of 2 M acetic acid, pH 6.5. The reaction products were then passed through a reversed-phase CIS BondElut cartridge as described previously (22). The lipids isolated from the cartridge were then analyzed by HPTLC and autoradiography as described previously in the text.
Solubilization and Separation of Sulfotransferases-Microsomal membranes from 30 rat whole brains (PD 14-16) were prepared as previously described. The membranes were solubilized with 40 ml of 1% Triton X-100 in 20 mM Tris-HC1 buffer, pH 7.2, 10 mM MnCL 25% glycerol, and 1 mM dithiothreitol (solution A), overnight, at 0-4 "C. The mixture had a protein:Triton X-100 ratio of 5:l. The mixture was centrifuged at 100,000 x g for 1 h. The supernatant had approximately 95 and 80% of the total microsomal GalCer:ST and GGL:ST activity, respectively.
Aminoalkyl agarose (30 ml; Affi-Gel 102, Bio-Rad) suspended in 50 ml of water, was mixed with 0.5 g of pyridoxal 5-phosphate, and the pH was adjusted to 7.0 by adding NaOH (24). The mixture was stirred in the dark at room temperature, overnight. The resulting Schiff base was reduced by addition of 50 mg of solid sodium borohydride at 4 "C. After the reduction, the derivatized agarose was washed with 0.3 M NaCl (24).
A pyridoxal phosphate-ligated agarose column (2 X 8 cm) was preequilibrated with solution A. The microsomal membrane proteins solubilized in solution A, 40 ml, were loaded onto the column at a rate of 30 ml/h. The eluant was diluted 10-fold with solution A, but without Triton X-100 (thus changing the Triton X-100 concentration from 1 to 0.1%) and reloaded onto the column. The column was then washed with 150 ml of solution A containing 0.1% Triton X-100, and 10-ml fractions were collected. The bound proteins were then eluted with 80 ml of a linear gradient of zero to 0.2 M NaCI, and 3-ml fractions were collected. The activities of the GGL:ST and GalCer:ST were measured in the collected fractions as follows. Tris-HCI buffer, pH 7.2, 100 mM; MnCI2 10 mM; ATP, 10 mM; dithiothreitol, 1 mM; phosphoadenosine pho~pho[~~S]suIfate (1.0 pCi, 0.45 nmol); bovine brain GalCer, 5 pg; GGL-1,1 pg; Triton X-100,100 pg; bovine serum albumin, 40 pg; and phosphatidylcholine, 10 pg, and 50 p l of the eluant fraction, in a total volume of 100 p l . The mixture was incubated at 37 "C for 1 h. The radioactive lipid products were separated from PAP35S and other water soluble components by a CI8-BondElute cartridge (22). A portion of the combined radioactive lipids, sulfatide and SGGL-1, eluted from the cartridge was counted in a scintillation counter, and the rest of the fraction was further separated on a HPTLC plate developed with chloroform, methanol, 0.25% CaCIZ (5:4:1). The plate was autoradiographed with an x-ray film, and the amount of radioactivity in the two lipids was determined.

RESULTS
Optimal Conditions for Rat Brain Sulfotransferases-A variety of conditions were tested for the optimal requirements of the enzyme GlcAnLcOse4Cer:sulfotransferase (GGL:ST) with PAP35S as the donor and GlcAnLcOse&er (GGL-1) as the acceptor. The results were also compared with the well known Ga1Cer:ST (Table I). Although the requirements of cofactors of both the sulfotransferases were similar, there was a quantitative difference in the activity of the enzyme with GGL-1 and GalCer as acceptors. GGL:ST was much more sensitive to omission of Mn2+ and Triton X-100; whereas Ga1Cer:ST was more sensitive to omission of ATP from the Complete reaction mixture for GGL:ST in 100 p l consisted of 250 pg of microsomal protein from 8-day-old rat cerebral cortex; Tris-HC1 buffer, pH 7.2, 100 mM; Triton X-100, 0.1%; ATP, 2.5 mM; MnCI2, 10 mM; GGL-1, 4 pg and PAP [35S], approximately lo6 dpm, 20 p M (-500 dpm/pmol). The incubation was for 45 min at 37 "C. For GalCer:ST the incubation conditions were the same, except Triton X-100 was 0.2%, MnCIz was 15 mM and, instead of GGL, 5 Fg of GalCer was used. * 100% activity for GGLST represents -108 pmol/mg/h and that for GalCer:ST -7.2 pmol/mg/h. The assay was done in duplicate and the results are average of the two determinations. reaction mixture. The specific activity of GGL:ST was about 15-fold higher than GalCer:ST, when 4-8-old rat brain microsomal membrane proteins were used as the enzyme source. Since GGL:ST almost absolutely required detergent Triton X-100 for the reactivity, a variety of other detergents were tested. Triton X-100 and Triton CF-54 were most effective compared to other detergents, such as CHAPS (23%) and octylglucoside ( l l % ) , however, Tween, deoxycholate, and taurodeoxycholate were ineffective. The effect of varying the concentration of Triton X-100 on the activity of Ga1Cer:ST was significantly different from that of GGL:ST (Fig. 1). GGL:ST was highly stimulated with 0.1% Triton X-100 and was inhibited with higher concentration of the detergent, whereas Ga1Cer:ST was only moderately affected with the detergent.
GGL:ST required Mn2+ for activity. The activity was completely abolished in the presence of 10 mM EDTA. Other metal ions such as M e and Caz+ were partially effective, but Cu2+, H e , and Zn2+ were completely ineffective. The optimal concentration of Mn2+ was 10 mM, a t higher concentration it was inhibitory (Fig. 2). M$+, 20-25 mM, was about 80% as effective as Mn2+. The activity profile of Ga1Cer:ST on varying the concentration of Mn2+ and M$+ ions was significantly different from that of GGL:ST (Fig. 2). The stimulation of Ga1Cer:ST was not as strong by these metal ions.
GGL:ST showed maximal activity at pH 5.8 with sodium cacodylate buffer (Fig. 3), however, at the same pH sodium  Product Identification-The radioactive products formed after the enzymatic reaction of PAP35S with GGLs, were analyzed after HPTLC, followed by autoradiography and immunoreactivity with monoclonal antibody HNK-1 (Fig. 4). The autoradiogram of the products after the reaction showed two bands which comigrated with standard SGGL-1 and SGGL-2 (Fig. 4C). These radioactive bands on the HPTLC plate reacted with monoclonal antibody HNK-1 (Fig. 4B, lane P ) indicating that the radioactive products formed were SGGLs. In the orcinol-sprayed HPTLC plate (Fig. 4A, lane P ) the products were not visualized, because the amounts formed were not sufficient enough for the orcinol reaction.
To identify the point of attachment of the sulfate group in the in vitro biosynthesized radiolabeled SGGLs, periodate oxidation of the vicinal hydroxyl groups followed by sodium borohydride reduction was carried out. The radiolabeled lipids before (Fig. 5B, lane B) and after the oxidation-reduction procedure (Fig. 5B, lane A) were analyzed by HPTLC followed by autoradiography. The autoradiogram showed that both radioactive SGGLs did not change their mobility after the oxidation-reduction reaction. The HPTLC plate after spraying with orcinol reagent is shown in Fig. 5A. The major nonradioactive lipids GM1 and GDla (originating from the microsomal members) as well as the substrates GGLs (migrating with GM1 and GD~,) in the reaction mixture (Fig. 5A,  lane B ) had reacted with the periodate and were no longer present after the reaction (Fig. 5A, lane A), indicating the effectiveness of the reaction. The migration pattern of SGGLs did not change after the periodate reaction, and no other radioactive products were formed (Fig. 5B, lane A) indicating that these compounds do not have reactive vicinal hydroxyl groups in the sugar chain. Therefore, the sulfate group must as the developing solvent. The plate was subjected to autoradiography and immuno-overlay as in C and B, respectively. The lipids were visualized on an identical HPTLC plate after spraying the plate with orcinol reagent and heating, as in A. In A: Lane SI, ganglioside standards for reference; from top to bottom G M~, G M~, Gol,, GDIb, and G T~~, respectively; S2, substrates GlcAnLcOse4Cer (GGL-1, major band) and GlcAnLcOse&er (GGL-2) minor band migrating near GD~,; S3, standard SGGL-1 (major band) and SGGL-2 (minor band); P, products after incubation with PAP35S and microsomal membranes. The orcinol reactive bands migrating with G M~ and GD~. are the gangliosides from the microsomal fraction. The products formed were not visualized with orcinol due to small amounts. B, HPTLC immunoblot of GGLsulfotransferase reaction products. Lanes S3 and P were exposed to HNK-1 antibody followed by peroxidase-conjugated goat anti-mouse IgM and developing the color reaction with 4-chloronaphthol and H 2 0 2 . C, HPTLC radioautogram of the sulfotransferase reaction products. The same lane P in A and B was exposed to xray film. be linked to the 3 position of the terminal glucuronyl residue of SGGLs.
Substrate Specificity-The activity of the 4-day-old rat brain microsomal sulfotransferase was determined with different glycolipids as acceptors, under the reaction conditions as described in Table I for GGL-1 (Table 11). The maximal activity was found only with GGL-1 (GlcAnLcOse4Cer) as the acceptor. The activity was drastically reduced with 6-methylGlcAnLcOse4Cer and GlcnLcOse4Cer, as substrates, indicating that the enzyme was highly specific for the terminal GlcA and the free-COOH group was required for the reaction. Substitution of the -COOH of GlcA with either -COOCH3 or -CH*OH made the substrates ineffective as acceptors of sulfate by the sulfotransferase. In 4-day-old rat brain, little activity of sulfotransferase was seen with galactosylceramide (GalCer), lactosylceramide, asialo-G,,,,, and monogalactosyldiglyceride, whereas other glycolipid substrates were completely ineffective.

Identity GGL:ST from Other Sulfotrunsferases-Competi-
tion experiments were performed to determine the identity of G G L S T from other sulfotransferases. Rat brain homogenate

Substrate specificity of 4-days-old rat brain microsomal sulfotransferase
The assays were performed as described in Table I

Competition between acceptor substrates for the sulfotransferase activities
For experiments in ( a ) and ( b ) the acceptors were incubated together with 117 and 100 pg, respectively, of microsomal proteins from 9-day-old cerebral cortex; for ( c ) 113 pg of homogenate proteins from 10-day-old cerebral cortex were used. After the incubation, essentially as described in the legend of Table I for GGLST, the products were separated on an HPTLC plate, autoradiographed, and the radioactivity in the individual glycolipid products was determined. Each incubation was in duplicate and the values did not vary more than 12% of the average.

GGL-1 Cer-Gal Cer-Glc-Gal SGGLs Cer-Gal-SO4
Cer  (Table  111). After incubation, the radioactive products were isolated and separated on HPTLC, and the amount of incorporation was determined in each product formed. The results showed that by varying the amounts of galactosylceramide and keeping the same level of GGL-1, the incorporation into sulfatide or SGGLl was not affected (Table IIIa). Similarly, keeping the same level of GalCer and varying the levels of GGL-1 also did not affect the incorporation into SGGL-1 and sulfatide, (Table IIIb). These results suggested two separate sulfotransferases for these two substrates. However, when GalCer was incubated with varying amounts of lactosylceramide, the in-corporation into both sulfatide and sulfated lactosylceramide was affected indicating that the same enzyme may be active for the latter two substrates (Table IIIc).
Heat Inactivation Profile-13-day-old rat brain microsomal membranes were incubated at 45 "C for various time periods between 0 and 20 min and then assayed for the sulfotransferase reaction with GGL-1 and GalCer as acceptors. The Ga1Cer:ST activity was more susceptible to heat inactivation than GGLST, e.g. after 5 min a t 45 "C, about 80% of the original GGL:ST activity remained intact, whereas only 40% of the original Ga1Cer:ST activity remained.
Effects of Dithiothreitol, NaCl, and Pyridoxal Phosphate on the Sulfotransferases-The effect of dithiothreitol, a t different concentrations, on the GGL:ST and Ga1Cer:ST was studied. Dithiothreitol, 0.5 mM, stimulated GGL:ST about 3-fold but had little effect on Ga1Cer:ST. The effect of NaCl on the activities of these enzymes, in the presence and absence of dithiothreitol, is shown in Fig. 6. In the presence of dithiothreitol, 100 mM NaCl reduced the activity of GGL:ST to 25% of the original, whereas Ga1Cer:ST was not negatively affected up to 150 mM NaCl (Fig. 6A). In the absence of dithiothreitol, GGL:ST was also more negatively affected than GalCer:ST, by increasing concentration of NaCl in the incubation mixture (Fig. 6B). Pyridoxal 5-phosphate has been shown to be a potent inhibitor of Ga1Cer:ST (25). GGL:ST was also found to be inhibited by pyridoxal 5-phosphate, however, it was somewhat less susceptible to the inhibitor than Ga1Cer:ST. For example, in the presence of 0.1 mM of the inhibitor, only 10% of the original Ga1Cer:ST remained, but 37% of the original GGL:ST remained.
Partial Separation of GGL:ST from Ga1Cer:ST-In order to resolve the activity of GGL:ST from GalCer:ST, PD 15 rat brain microsomal membranes were solubilized in a solution containing 1% Triton X-100 and chromatographed on a pyridoxalphosphate-ligated agarose column (22) (Fig. 7). Under  1 mM dithiothreitol ( B ) . Microsomal membrane proteins from 14-day-old rat brains were used as the enzyme source.

FIG. 7. Partial separations of GGL:sulfotransferase form
Ga1Cer:sulfotransferase on a pyridoxal 5-phosphate-ligated aminoalkyl agarose column. Rat brain microsomal membranes, solubilized in solution A containing 1% Triton X-100, were chromatographed and analyzed as described in the text. Fraction zero represents 400 ml of unbound fraction. The column was washed with 150 ml of solution A containing 0.1% Triton X-100, and 10-ml fractions were collected (fractions [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15]. At the point marked by an arrow, the column was eluted with a linear gradient of 80 ml of zero to 0.2 M NaCI, and 3-ml fractions were collected. The activity of the GGLST and Ga1Cer:ST was measured as described in the text. the column conditions given under "Materials and Methods," the activity of Ga1Cer:ST did not readily bind to the column and was mostly eluted from the column in the loading fraction (fraction zero) and during column wash (fractions 1-15). Whereas, GGL:ST was relatively more bound to the column and was eluted with a gradient of 0-0.2 N NaCl solution. Thus, in certain fractions, e.g. in fraction 6, there was only activity of Ga1Cer:ST and no GGL:ST, whereas in several fractions eluted from the column with NaCl gradient, there was high GGL:ST activity but low levels of Ga1Cer:ST activity.

Activity of GGL:ST during Development of the Rat Cerebral Cortex and Cerebellum-The
activity of the GGL:ST was measured in the homogenates of rat cerebral cortex and cerebellum from embryonic day 17 (ED 17) to postnatal day 60 (PD 60) (Fig. 8). The specific activity of GGL:ST in the cerebral cortex increased from ED 17 and reached a maximum a t around PD 5 and then declined (Fig. 8A). However, a significant activity (-30% of the maximal) still remained in the cerebral cortex from PD 30 to PD 60.
In the cerebellum, the specific activity of GGL:ST was high at ED 21, the first point measured. The activity slightly increased t o a maximum around PD 5, then declined and reached a plateau level between P D 15 and 60 (Fig. 8A).
The Ga1Cer:ST activity was also measured in the cerebral cortex and cerebellum during development (Fig. 8A). In both of these areas, the activity was very low during neonatal period and increased only after PD 10 and reached maximum around PD 15.
The total activity of GGL:ST in the cerebral cortex increased with age and reached a maximum at PD 15, remained about the same until PD 30, and then slightly declined (Fig.  8B). In the cerebellum, also the total activity of GGL:ST increased postnatally and reached a maximum at PD 15 and then slightly declined (Fig. 8B).
Expression of Sulfotransferases in Different Neural Areas-The specific activity of GGL:ST in homogenates of PD 25 rat gray matter was 2.5-fold higher than that in white matter (Table IV). The GalCer:ST, however, was 4-fold higher in the white matter than in the gray matter (Table IV)  activity was highest in the spinal chord and lowest in the cerebellum.

DISCUSSION
The expression of SGGLs in the mammalian cerebral cortex, as determined by biochemical methods, is limited to embryonic development and is no longer detectable shortly after postnatal development (3,17). In the cerebellum, however, these glycolipids are robustly expressed during neonatal development, followed by a significant decline by PD 7 and a second burst of expression near PD 20 (18). Specific glycosyltransferases and glycosidases are possibly involved in the regulation of expression of SGGLs during the neural cell differentiation stages in the cerebral cortex and cerebellum. In addition, these enzymes are possibly regulated differently in the cerebral cortex uersus cerebellum to account for the unique temporal expression of SGGLs in these areas of the nervous system. Previously, we have reported on the expression and regulation of UDP-g1ucuronate:neolactotetraosylceramide glucuronyltransferase during development of the nervous system (16). It was concluded that GlcA-transferase was not a regulatory enzyme controlling the differential expression of SGGLs in the cortex and cerebellum. Here we have characterized the activity of GGL:ST involved in the last step for the synthesis of SGGLs.
A variety of sulfotransferases have been reported in the nervous system, however, studies on sulfotransferase related to glycolipid acceptors are limited to Ga1Cer:ST. The latter enzyme has been studied in the nervous system and in kidney where it is more preponderant than in other tissues (22)(23)(24)(25)(26)(27)(28). In the nervous system, Ga1Cer:ST is mostly localized in the Golgi apparatus of oligodendrocytes in the central nervous system and in Schwann cells in the peripheral nervous system (27,28). The major function of this enzyme in the cells is to synthesize sulfatide which is a characteristic and abundant lipid of myelin membrane. The enzyme Ga1Cer:ST is expressed maximally in the central nervous system during the period of active myelination which in the rodents reaches a maximum around 30).
In this report we have described some of the properties of a novel GGL:ST, which are different from those of Ga1Cer:ST. For example, the GGL:ST had a strict requirement for a divalent metal ion, especially Mn2+, but M e and CaZ+ were also partially effective (Fig. 2). Metal ion requirement for Ga1Cer:ST was not absolute (Fig. 2), which confirmed the previous similar observations (26). The activity of GGL:ST in vitro was also dependent upon the optimal presence of a nonionic detergent such as Triton X-100 or CF-54 (0.1% for about 125 pg of protein); at higher concentration than 0.1%, Triton X-100 was inhibitory. The activity of Ga1Cer:ST was not as sensitive to the detergent concentration (Fig. 1).
The GGL:ST transferred radioactive 35S04 from PAPS specifically to the hydroxyl group on carbon 3 of the terminal GlcA residue of GGL-1 and GGL-2. The activity was drastically reduced when other structural analogs of GGL-1, such as 6-methylGlcAnLcOse4Cer or GlcnLcOse4Cer, were used as acceptors. This showed requirement of a free -COOH group in the terminal GlcA residue, for the transfer of SO,. No other side products were formed from GGLs by the GGL:ST. Little sulfotransferase reaction activity was seen with other lipid acceptors, when microsomal enzyme from 4-day-old rat brain was used as the enzyme source (Table 11). This suggested that at this age, GGL:ST was the major sulfotransferase in the cerebral cortex and other lipid acceptors were not utilized by the enzyme, under the conditions of the assay. Competition experiments with GalCer also showed that GGL:ST activity i n vitro was not affected by the presence of GalCer, similarly Ga1Cer:ST activity was not affected by the presence of GGL-1 (Table 111, a and b), indicating two separate sulfotransferases. However, Ga1Cer:ST activity was affected by the presence of Cer-Glc-Gal (ceramide lactoside) in the reaction mixture (Table IIIc), indicating Ga1Cer:ST also catalyzed the transfer of SOa to ceramide lactoside. It has been reported, with purified preparations of Ga1Cer:ST from rat brain and kidney, that GalCer and ceramide lactoside were equally good acceptors, but little activity was seen with psychosine and monogalactosyldiglyceride (26). Studies on heat inactivation and effects of sulfahydral reagent dithiothreitol, NaCl (Fig.  7), and pyridoxal phosphate on the activities of GGL:ST and Ga1Cer:ST clearly showed that these two enzyme activities are most probably due to two different enzyme proteins. Both of these sulfotransferase activities were partially resolved on a phosphopyridoxyiagarose column, leading to the conclusion that these lipid sulfotransferases are two different proteins.
The expression profiles of GGL:ST and Ga1Cer:ST in the rat cerebral cortex and cerebellum during development are remarkably different (Fig. 8). As previously reported by others (29,30), the specific activity of Ga1Cer:ST was negligible neonatally and reached a maximum near PD 15-20 (Fig. 8). The specific activity of GGL:ST in the cerebral cortex increased sharply during embryonic and neonatal development and reached a maximum at PD 5. The specific activity in the cerebellum also increased neonatally but not as sharply as in the cerebral cortex and reached a maximum around PD 5. The total and specific activities of GGL:ST do not correspond to the amount of SGGLs in the rat cerebral cortex and cerebellum during development of these tissues (3, 18). SGGLs almost completely disappeared from the cerebral cortex by PD 20, however, a significant level of GGL:ST activity remained in the cortex. In the cerebellum, the amount of SGGLs increased during postnatal development, but the specific activity of GGL:ST declined. These results suggest that GGL:ST is not a regulatory enzyme which controls the expression of SGGLs in these neural tissues. We have previously reported similar results with the expression of UDP-glucuronate:neolactotetraosylceramide glucuronyl transferase involved in the synthesis of GGLs (16). In fact, analyses of the availability of the precursors nLcOse&er and nLcOse&er in these tissues during development correlated much better with the expression of SGGLs, suggesting that enzymes involved in the synthesis of these precursors control the differential expression of SGGLs in these tissues (16).
Available evidence indicates that SGGLs are the products of and primarily localized in the neuronal cells rather than in glial cells at least in the central nervous system (17). Immunocytochemical localization and analysis of SGGLs in Purkinje cell abnormality murine mutants clearly showed their expression only in Purkinje cells in the cerebellum (18, 31, 32). Comparisons of the activity of GGL:ST and Ga1Cer:ST in the gray matter, which is enriched in neuronal cells, and in white matter, which is enriched in oligodendroglial cells, indicate that GGL:ST is enriched in gray matter, whereas Ga1Cer:ST is enriched in white matter. Van der Pal et al. (30) have reported that the activity Ga1Cer:ST was 5-fold higher in the spinal cord than in cerebral cortex at PD 15. We confirm these results, however, about &fold higher GGL:ST activity in the gray matter than in the spinal cord corroborates the conclusion that these two sulfotransferases are different.
Previously, it has been reported that the levels of SGGLs in spinal cord are extremely low.