The Biosynthesis of a Disialylganglioside by Galactosyltransferase from Rat Brain Tissue

An enzymatic method for the preparation of ceramide-Glc-Gal(NeuAc)2-GalNAc (GD2) from ceramide-Glc-Gal-(NeuAc)rGalNAc-Gal (G& is described. GDz, prepared in this manner, is an acceptor of galactose from UDP-Gal when incubated with a particulate subcellular fraction from rat brain to form GDlb. This reaction was studied and compared with an analogous reaction for the synthesis of cer-amide-Glc-Gal(NeuAc)-GalNAc-Gal (GM1) using ceramide-Glc-Gal(NeuAc)-GalNAc (GM2) as acceptor of galactose. The results obtained in this comparative study suggest that GDZ is an intermediate in the pathway of synthesis of GDlb. Furthermore, the transfer of galactose in the two reactions appears to be catalyzed by the same enzyme.

Numerous reports have appeared dealing with investigations of the pathways of ganglioside synthesis in mammals and other vertebrates.
In the course of these studies, several different experimental approaches have been employed and the data derived from them have elicited conflicting interpretations. These investigations involved experiments in which labeled precursors were administered in wiuo (1,2), studies in vitro with exogenous donor compounds in the presence of endogenous acceptor materials (3), and other experiments in which both acceptor and donor molecules were exogenously supplied in vitro (4)(5)(6)(7)(8)(9)(10)(11).
Because the data obtained in these various experiments have been difficult to reconcile completely, several alternative routes have been proposed for ganglioside formation (3,6,8,9,11). One of the principal difficulties with experiments using exogenous acceptors has been the inability to obtain sufficient quantities of certain potential intermediates, especially those involved in disialylganglioside synthesis. This limitation was especially true in the case of ganglioside G& which was theorized to be the immediate precursor of GUlh (6). We recently obtained indirect evidence of the potential intermediary role of GDz by incubating GD3 in the presence of UDP-N-acetyl-[14C]galactosamine and UDP-[3H]galactose to yield sequentially GD3 and &xi, (11).
In order to examine this reaction sequence in detail, we have devised a procedure for the preparation of GDz in substrat.e amounts.
We subsequently demonstrated the ability of this material to serve as acceptor of galactose from UDP-Gal when incubated with an enzyme preparation obtained from rat brain tissue. This reaction was carefully compared with the analogous process for the synthesis of Ghll from Ghlz and UDP-galactose. Evidence is presented which suggests that the transfer of galactose to GM2 and GDz may be catalyzed by the same enzyme.

EXPERIMENTAL PROCEDURE
Materials-UDP-galactose was obtained from Calbiochem. Triton CF-54 was purchased from Sigma and Tween 80 from Nutritional Biochemicals. Gangliosides Ghll and GDla were obtained from Supelco.
Glucosyl-and galactosylceramides were gifts from Dr. A. Gal  Preparation of Ganglioside Acceptors-Cer-Glc-Gal(NeuAc)-GalNAc was obtained from brain tissue of patients with Tay-Sachs disease by the method of Gatt and Berman (13). Cer-Glc-Gal(NeuAc)z-GalNAc was prepared from Cer-Glc-GaE (NeuAc)rGalNAc-Gal by treating it with a rat liver @-galactosidase as described below.
Total bovine brain gangliosides (Sigma type III) was the starting material for obtaining GDlb which was purified by column chromatography on DEAE-cel-Issue of April 25, 1972 F. A. Cumar, J. F. Tallman, and R. 0. Brady lulose as described by Winterbourn (14). Fractions enriched in GDn, were concentrated and final purification was carried out by preparative thin layer chromatography using chloroformmethanol-10% NH40H (55:40: 10) as the developing solvent on plates of Silica Gel G (20 x 40 cm).
Conversion of GDb to GDZ-Evidence was presented in a previous communication for the conversion of Ghll to GM2 using a @-galactosidase obtained from rat liver (15). We observed that a similarly prepared aqueous extract of rat liver lysosomes catalyzes the hydrolysis of GDn, to GDz. An acetone powder was prepared from Lysosomal Pellet 2 obtained according to the procedure of Ragab et al. (16). The powder was stored overnight in vacua over PzO5 and subsequently extracted with 5 volumes (v/v) of cold distilled water.
The suspension was kept at 0" for 1 hour with periodic stirring and then centrifuged at 30,000 x g for 20 min. The supernatant solution which contained all of the ganglioside /?-galactosidase activity was frozen and thawed and subsequently frozen in liquid Nz until used. The protein concentration was 11.6 mg per ml. This enzyme preparation catalyzed the hydrolysis of the terminal molecule of galactose from GDn, as well as from GM1. Some of the properties of this preparation are indicated in Table I. For preparative purposes, the conversion of 4 pmoles of GDn, to GDg was carried out in 5 ml of the P-galactosidase enzyme preparation containing 500 pmoles of potassium acetate buffer at pH 5.0 at a temperature of 37". After incubating for 5 hours, 3 ml more of the buffered enzyme were added and the incubation was continued for 4 more hours. The reaction was terminated by adding 20 volumes of chloroform-methanol (2: 1). The precipitated protein was removed by filtration through Whatman No. 1 filter paper. The filtrate was partitioned with water according to the method of Folch et al. (19), and the material in the upper phase was dialyzed against a large excess of distilled water which was changed periodically over a period of 48 hours. The retentate was lyophilized and the dry residue was extracted with chloroform-methanol-water (60 : 30 : 4.5). The redissolved material was passed over a column of Sephadex G-25 previously equilibrated with the same solvent mixture (20). Under these conditions, more than 90% of the GDn, was converted to GDz ( Fig. 1) estimated from thin layer chromatography.
The GDz was further purified by preparative thin layer chromatography as described in the previous section. The over-all yield of CD2 with this procedure was about 3% from unfractionated mixed brain gangliosides as the starting material.
The identity of the purified GDz produced by this reaction was established in the following fashion.
First, the product had the reported mobility of GDz between GDla and GDlh on thin layer chromatography using chloroform-methanol-2.5 N NHJOH as the developing solution (21  (11). Second, it is known that GD2 can be converted to GM2 with Vibrio cholerae neuraminidase (22). When the GDs prepared by the present hydrolytic procedure was treated with this neuraminidase, the lipid reaction product cochromatographed with authentic Ghlg in the following solvent systems: chloroform-methanol-lOO~O NH40H (60 : 35 :8) ; l-propanol-lbutanol-water (65 : 10 : 25) ; and chloroform-methanol-water (55 : 40 : 10). Third, free galactose was stoichiometrically released when GDn, was exhaustively treated with the rat liver lysosomal  (Table I) or N-acetylgalactosaminidase activity.
In addition, any of the other theoretically possible products such as Ghll, Gh12, GD3, or GM3 would be easily differentiated from GDs by thin layer chromatography in the solvent systems used (21). Fourth, the product GD2 is an effective acceptor of galactose from UDP-[l%]gaIactose and the product of the reaction had all of the anticipated characteristics of GDn, as shown below.
Under the conditions used for study of the galactosyltransferase, no significant incorporation of galactose occurred with any of the other individual gangliosides available for testing excepting Ghlz (cf. Gangliosides were visualized on thin layer chromatograms with resorcinol spray reagent (25). Radioactivity was detected on the thin layer chromatograms by scanning with a Berthold chromatogram scanning device (Varian-Aerograph, Palo' Alto, California).
Other radioactivity determinations were performed as described previously.  The enzyme preparation used is the "water lysosomal extract after the second freezing and storage" ( Table I). The conditions of the assay were as described under "Experimental Procedure" under "Conversion of Golt, to Gnz" but all of the components were reduced 200 times. The incubation mixtures were carried through the Sephadex column step and the dried effluents were redissolved in chloroform-methanol (2:l) and spotted on a glass plate coated with Silica Gel G. After development with chloroform-methanol-10yO ammonia (60:35:8, v/v) the plate was sprayed with the resorcinol reagent. All of the stained spots are violet.
Lane 1, GM~ incubated with 0.1 M potassium acetate buffer (pH 5.0). The enzyme preparation was added at the end of the incubation time, after the chloroformmethanol.
Lane 2, GM~ as substrate incubated with the buffered enzyme preparation.
Lane 3, the same as Lane 2 but GMI was omitted in the incubation mixture.
Lane 4, the same as Lane 1 but GD~~ replaced Gilrl. Lane 5, the same as Lane 2 but GDI~ replaced GMI as substrate.
was then centrifuged at 20,000 x g for 20 min. The supernatant was decanted, and the pellet was resuspended in the same volume of 0.05 M sodium cacodylate-HCI buffer, pH 7.0, containing 0.1% 2-mercaptoethanol and then centrifuged as before. The pellet obtained in this fashion was resuspended in 0.25 M sucrose containing 0.1% 2-mercaptoethanol to yield a final protein concentration between 10 and 12 mg per ml. All of these procedures were carried out between 0 and 5". The enzyme preparation was used the same day. were pipetted into the incubation tubes in chloroform-methanol (2: 1). After shaking, the precipitated protein was removed by filtration through filter paper and the retained material and paper were washed with 2 ml of chloroform-methanol-water (60:30:4.5) (Solvent A). The filtrates were passed over 1 g of Sephadex G-25 in a column 1 cm in diameter which had previously been equilibrated with Solvent A (20). After the filtrate had penetrated into the column bed, the column was washed with 5 ml of Solvent A. Under these conditions, the labeled ganglioside products are quantitatively recovered in the effluent while the water-soluble labeled precursors remain in the Sephadex bed (5,11). The effluents were taken to dryness with warming and the radioactivity was determined.

Requirements for Galactosyltransferase
Activity-h each experiment reported in this communication, gangliosides GM2 and GDz were used in parallel as acceptors of galactose from UDP-[YJgalactose with the same enzyme preparation under identical incubation conditions. The requirements for galactosyltransferase activity are very similar with either GM2 or GDz as acceptor (Table II).
The requirement for Mn2+ with GM2 as acceptor which can be only partly replaced by Mg2+ is in agreement with previous reports (4,10). The addition of detergents increased the incorporation of radioactive galactose from 6-to la-fold in the presence of exogenous acceptors.
The addition of free galactose in concentrations from 15 to 20 times higher than the K, for UDP-[Wlgalactose did not substantially affect the incorporation of the labeled hexose. This result indicates that the sugar nucleotide was the donor substance.
In addition, when unlabeled UDP-galactose was added to the complete system at the same concentration as the UDP-[Wlgalactose, the incorporation of radioactivity was diminished by 50% (Table  III).
The reaction is essentially linear over a period of 23 hours and is proportional to the amount of protein up to 240 pg with both acceptors (Fig. 2). Substrate Spec$ic~ty-Various structurally related glycolipids were examined as potential acceptors of ['4C]galactose.
Significant incorporation was observed only with GM2 and GDz (Table  IV).
Other lipids showed only minimal incorporation or changes in radioactive background.
The sample of GDn, tested was an aliquot of the material used for the preparation of GDz. As can be seen, more than 10 times greater incorporation of galactose was obtained after the GDn, was converted to GDz by the /3galactosidase.
GM2 was about 50% more effective as an acceptor of galactose than GDg under all of the various incubation conditions employed (cf. Figs. 1 to 3). Assuming that the same enzyme catalyzes the transfer of galactose to both of these subst,rates, the presence of the second sialyl group in GDz appears to decrease the effectiveness of this compound as an acceptor for galactose.
The presence of the terminal P-N-acetylgalactosaminyl residue is required for this galactosyltransferase activity since neither of the lower ganglioside homologues GM3 or GD3 which lack this moiety were effective substrates for the reaction. The enzymes which catalyze the transfer of galactose to GM2 or Cer-Glc or Cer-Glc- Gal have been shown to differ in some respects from each other (7,26,27). In the present investigations Cer-Glc and Cer-Glc- Gal were not effective acceptors of galactose.
The incubation conditions of our experiment differ from those used for studying the incorporation of 2325 galactose into neutral glycolipids as far as source and preparation of the enzyme, and choice of detergent and buffer are concerned.
Kinetic Constants-The effect of varying the concentration of GM2 and GD2 on galactosyltransferase activity is shown in Fig.   3. The K, values obtained from the double reciprocal plots were 17.5 and 16 &LM for GM2 and GD2, respectively. The affinity of the galactosyltransferase for UDP-galactose with GM2 or GDz as acceptor is shown in Fig. 4. Under the conditions of these assays, the K, values for the galactosyl donor were 0.5 and 0.66 I12M with the respective substrates.
The K, for GM2 obtained in the present experiments is in good agreement with the value previously reported for rat brain enzyme preparations (10); however, it is about IO-fold lower than that obtained with chicken brain enzyme preparations (4). Conversely, the K, for UDP-galactose with GM2 as acceptor is about 5 times higher with the rat brain enzyme preparation than that obtained with the chicken brain enzyme (4). We have no ready explanation for these differences at this time.
In addition to the difference in species tested, it is necessary to keep in mind that in both cases, a particulate enzyme preparation was used as source of of Gangliosides Vol. 247, No. 8 and Gnz (A-A).
The assay conditions were those described under "Experimental Procedure" except that the incubation time was 90 min. enzyme and differences in the accessibility of the substrates to such membranous structures may play a role in the kinetics of the reaction.
However, we should like to emphasize the essentially identical results which we have obtained when GM2 and Gns were tested under the same conditions.
Experiments with Mixed Acceptors-The incorporation of galactose was examined when a single acceptor GM2 or Gnz was employed and compared with that which occurred when both acceptors were added to the same incubation flask under conditions in which the velocity of the reaction is practically independent of the substrate concentrations. The quantity of galactose incorporated in the presence of both substrates was 46% of the calculated theoretical value based on the additive incorporation expected from the results obtained when each acceptor was incubated separately (Table III).
This experimental value is very close to the figure anticipated if one assumes that the two accept.ors interact with the same site on the transferase.
However, experiments were not attempted at present to try to determine whether the inhibition is competitive because we feel that the methods are not sufficiently exact to measure the individual reaction products from experiments with both acceptors incubated together.
Ident$ication of Reaction Products-The products of the galactosyltransferase reaction with GM2 and Gnz as acceptors cochromatographed with authentic GM1 and Gnis, respectively, on thin layer chromatograms developed with chloroform-methanol-10% NHdOH (60:35:8, v/v) (Fig. 5). When these reaction products were treated with I'. cholerae neuraminidase, there was no change in the mobility of the labeled reaction product obtained with GM2 as acceptor.
This result was anticipated since the molecule of N-acetylneuraminic acid in the expected product GM1 is linked to the internal molecule of galactose in this ganglioside and is not cleaved by V. cholerae neuraminidase (12).
However, when the radioactive reaction product obtained with Gnz as acceptor was treated with neuraminidase, the resulting labeled compound migrated faster than the original product and cochromatographed with authentic GM1 (Fig. 5). The identity of the reaction products before and after treatment with neuraminidase was confirmed in two additional solvent systems using chloroform-methanol-water (55 : 40 : 10) and l-propanol-lbutanol-water (65 : 10 : 25). These findings provide additional evidence that the product of the galactosyltransferase reaction. using Gn2 as acceptor was Gnrb since the second molecule of Nacetylneuraminic acid of Grin, is susceptible to hydrolysis by V. cholerae neuraminidase (12). Further, the labeled ganglioside reaction products when the acceptors were GM2 and Gnz were hydrolyzed in 2 N HCl at 100° for 6 hours. The hydrolysates were taken to dryness under reduced pressure, and the residue was dissolved in chloroformmethanol-water (60 : 35 : 4.5) and partitioned according to Folch et al. (19). All of the radioactivity was recovered in the upper phase. The solvent was evaporated under a stream of Nz and the residue was dissolved in water.
An aliquot was run by descending paper chromatography in butanol-pyridine-water (3 : 1:1.5, v/v, upper phase plus 1 volume of pyridine) (28). The radioactive material from the two samples cochromatographed with galactose which was added as internal standard. Two other aliquots from each hydrolyzed product were analyzed by thin layer chromatography using as developing solvent l-propanol-water (7: 1, v/v) and methylacetate-isopropyl alcohol-water (18 : I : I., v/v) as described by Gal (29). Again, radioactivity was detected only in the area of the galactose standard.
Standard samples of glucose were well differentiated from galactose in the three systems. DISCUSSION The method we describe for the preparation of Gnz by the conversion of Grin to Gn2 with the rat liver lysosomal fi-galactosidase is probably the most feasible procedure currently available for the preparation of the latter ganglioside. Gnz is a minor ganglioside component in brain.
The only procedure previously described for isolating this compound is that by Klenk and Naoi by countercurrent extraction from human brain gangliosides (22). The yield with their procedure is about 0.07% based on t.otal mixed gangliosides a,s starting material.
In the present procedure, the yield is about 3%.
In a previous communication, we obtained evidence which suggested that Gn2 was an intermediate in the biosynthesis of Gull, according to the following sequence of reactions: Gna -----+ [Grit] -----+ Grin (11). This scheme was based on data obtained from the conversion of Gn3 to Gnz in the presence of UDP-Nacetylgalactosamine and the two-step conversion of Gn3 to Gnii, in the presence of TJDP-N-acetylgalactosamine and UDPgalactose.
In that report, a comparative study was made of the corresponding analogous reactions Gh13 -+ GM2 and GM3 + [C,21-GM.
Since Gh13 had been considered the precursor of GM2 (3,6), the strong similarities in the kinetic parameters for the two reactions GM3 ---f Gh12 and Gn3 --) Gnz allowed us to postulate that Gn3 is t,he precursor of Gnz. Difficulty was encountered in quantitatively evaluating the synthesis of Gnrb by these consecutive reactions.
The principal obstacle was the dependence of the rate of the second reaction on the rate of formation of the intermediate Gns.
Because of this, the concentration of the suspected iutermediate [Gnz] was unknown and varied in the course of the incubation.
In addition, the requirements for the individual reactions could not be examined. For this reason, a detailed analysis of this process was deferred. We chose for the present the alternative procedure of examining the reaction in a stepwise fashion.
The results obtained in the present undertaking support the proposed reaction scheme. Agail similarities were observed regarding reaction rates and substrate affinities for t.he two analogous reactions Ghlz + GM1 and G,,z + Gnu,. Since GM2 has been shown to be the precursor of G&ri (3, 6), we thiuk it is likely that Grit is a precursor of c XDll, .
The result obtained from the experiment with mixed acceptors and the constant ratio of incorporation of galactose when Gh12 and GD2 were used as acceptors under various experimental conditions suggest that. the same enzyme catalyzes these two reactions.
The essentially identical K, values of the sugar nucleotide with the two acceptors and requirements for the processes are also consistent with this postulate.
The present experiments provide further evidence for the biosynthetic pathway of gangliosides as postulated by Kaufman et al. (6) employing a similar experimental approach, e.g. using exogenous acceptors in. vitro. However, these investigators did not examine the reactivity of the gangliosides directly involved in the synthesis of Grin,. On the other hand, Arce ef al.
(3) ob-tained data which suggest that GM1 is the immediate precursor of Gnlb in experiments in vitro designed to approach more physiological conditions by avoiding the use of detergents and using endogenous acceptors. More recently the same investigators have studied the biosynthesis of gangliosides in viva using labeled precursors (2). In order to reconcile the results obtained in studies of ganglioside synthesis using endogenous acceptors in vitro and synthesis in vivo they postulated that there is compartmentalization of multienzyme-substrate complexes in rat brain. If such an organization exists in brain tissue, we feel that only cautious extrapolation of our observations should be attempted with regard to the exact nature of the processes involved in ganglioside formation in viva.