Agarose derivatives of uridine diphosphate and N-acetylglucosamine for the purification of a galactosyltransferase.

Abstract The applicability of affinity chromatography for the purification of UDP-galactose: N-acetylglucosamine galactosyltransferase has been evaluated. Three different types of specific adsorbents have been synthesized by reaction of a ligand structurally related to portions of the substrates of the transferase with cyanogen bromide-treated agarose. The synthesis of the ligands, P1-(6-amino-1-hexyl)-P2-(5'-uridine)-pyrophosphate, 6-amino-1-hexyl-2-acetamido-2-deoxy-β-d-glucopyranoside and P1-(6-amino-1-hexyl)-P2-(β-d-galactopyranosyl)pyrophosphate is described. The galactosyltransferase from bovine milk has been found to bind to two of the adsorbents and some of the parameters influencing the binding have been evaluated. The affinity of the transferase for the UDP-hexanolamine-agarose adsorbent was enhanced by manganous ions whereas the affinity was decreased by EDTA, urea, borate, or magnesium ions. The affinity of the enzyme for N-acetylglucosamine hexanolamine-agarose was enhanced considerably by UDP or UMP and was decreased by borate, urea or N-acetylglucosamine. N-Acetylglucosamine-agarose also served as an acceptor substrate for the transferase. The galactosyl pyrophosphate-agarose was found to have a low capacity and specificity of binding for the galactosyltransferase. On the basis of the properties of the adsorbents, it has been possible to purify the galactosyltransferase from bovine milk to constant specific activity soley by affinity chromatography. The enzyme from the whey of bovine milk was first adsorbed on UDP-agarose and the eluate from this adsorbent then chromatographed on either N-acetylglucosamine-agarose or α-lactalbumin-agarose. The properties of the transferase are very similar to those reported earlier although at least three species of enzyme differing slightly in molecular weight have been found. The three species are glycoproteins and their amino acid compositions are very similar. The UDP-agarose adsorbent may be useful for purification of other UDP-glycosyltransferases as shown in preliminary studies with rabbit muscle glycogen synthetase. It is also possible that other nucleotide-agarose or glycosyl-agarose adsorbents may be used together for the purification of many nucleotide-sugar transferases.

to those reported earlier although at least three species of enzyme differing slightly in molecular weight have been found.
The three species are glycoproteins and their amino acid compositions are very similar. The UDP-agarose adsorbent may be useful for purification of other UDP-glycosyltransferases as shown in preliminary studies with rabbit muscle glycogen synthetase. It is also possible that other nucleotide-agarose or glycosylagarose adsorbents may be used together for the purification of many nucleotide-sugar transferases.
A major group of glycosyltransferases functions to transfer a glycosyl moiety from a nucleoside diphosphate glycose to an acceptor molecule that may also be a carbohydrate, releasing a nucleoside diphosphate as shown in Reaction 1.
Glycosyll-P-P-nucleoside + glycosez ---f (1) glycosyllglycose? + nucleoside diphosphat,e Such enzymes function in the synthesis of lactose, fructose, glycogen, starch, and the complex oligosaccharides of glycoproteins, glycolipids, and bacterial cell walls (1). Many of the enzymes are present in relatively small amounts in tissues and are associated with complex structures which make them dificult to purify.
In many cases membrane-bound enzymes cannot be obtained in free form.
If sufficiently specific adsorbents with functional groups resembling either glycose,, glycose2, or the nucleoside moieties of the substrates and products could be prepared, they should be useful for isolation of a specific transferase from a complrx mixture.
To test this possibility we have synthesized agarose derivatives with ligands resembling parts of the substrates of a typical galactosyltransferase and report here on their utility in the purification of the galactosyltransferase of lactose synthetasc (2, 3). ,2 preliminary rrport of these studies was made earlier (4) mine as substrate unless indicated (2,3). Glycogen synthetase columns, washed with 3 volumes of 6 M urea (deionized) and t,hen activity was assayed by a similar procedure in which the tralnsfer exhaust'ively washed with water or buffers saturated with chloroof glycosyl residues from TDP-[14C]glucose to glycogcn was form. oc-Lactalbumin-Sepharose was washed in the same way estimated as follows.
A solution (50 ~1) of 2 m&r UDP-[14C]-after use except' that 2 M instead of 6 M urea was used. The glucose, 8 mu in glucos:c G~phosphate, containing 1.3 mg of latter columns were operated as described earlier (3) except that glycogen ml-1 in 50 mni glycerol phosphate buffer, pH 8.2, coil-~T-acct.~lglucosan~ine replaced glucose dnring development of taitling 2 rn>f EDT& and 40 mnf 2-mercnptoetl~anol was incu-the column. bated with 50 ~1 of t'he enzyme solution for 30 min at 37". The Concentration of Dilute Solutions of CalactosllltransSerase-The reaction was stopped by chilling the sample in an ice bath, dilut-dilute solutions of enzyme obtained after chromatography on illg with 0.5 ml of water (5") and transferring the sample to & the specific adsorbents were concentrated by pressure dialysis small column (0.5 ml) of Dowes l-X8 (Cl-cycle) contained in a at 4" with a 15-cm Z1micon membrane (Pl%30).
During condiqjosablc Pasteur pipette. The incubation tube was rinsed centration of the enzyme care was taken to maintain t,he conwith 0.5 ml of water and the rinse applied to the column which centration of Kacetylglucosamine in the enzyme solution at was then washed with 1.0 ml of water.
Eluent and washings 0.005 M. As noted earlier (3) the enzyme is very unstable in the were collected in a sc%ltillat'ion vial and then 14 ml of scintilla-absence of its substrates but under these conditions it is stable t,ioll liuid as rrportcd earlier (3) was added. Samples were for at least 3 months when stored at 4" in the presence of c.hlorocounted at ambient temperature.
Concentration of solvents was performed made 9 m&r with respect to P-mercaptoethanol (2.0 ml) and 0.03 on a rotary evaporator at water aspirate pressure unless other-RI with respect to EDTA (250 ml of 0.5 M EDTA, pH 7.2). The wise specified. Optical rotations were determined with a Cary milk was then adjusted to pH 4.2 by the addition over a lo-min 60 recording spectropolnrimeter. period of 5 1\~ HC'l (50 ml). The preparation was centrifuged in a ;&ending paper chromat.ogra.phy for t,hc organic compounds Sorvall RC-3 centrifuge in a type HG-4L swinging bucket rotor was carried out on Whatman No. 1 w-it'll isobutyric acid-con-at 5250 rpm (7000 x g) at 10" for 10 min to remove most of the centrated ammonium hydroxide-water (66 : 1: 33, v/v) as Solvent casein. The pooled supernatant solution, after filtration through 1 and isopropyl alcohol-l. 3 ?\I aqueous acetate buffer pH 5.0 glass wool, was adjusted to pI-I 6.7 to 6.8 with 2 M Tris, made (i:3, v/v) as Solvent 2. Descending paper chromatography 0.12 M with respect to manganese chloride (75 g), adjusted again was performed on Whntman No. 1 with ethyl acetate-pyridineto pH 6.7 to 6.8 with 2 M Tris, and after standing for 4 to 5 water (10:1:4, v/v) as Solvent 3. Chromatograms were dehours centrifuged as described above to remove any insoluble l-eloped with a molybdate spray (6) for phosphates, ninhydrin (7) material.
The resulting characteristically amber-colored whey for amines, and silver nitrate (8) for carbohydrates.
was stored at 4" in the presence of chloroform (1.0 ml) for periods Electrophoresis was performed on 7.5% polyacrylamide gels of up to 2 weeks without any apparent loss of galactosyltransin sodium dodecyl sulfate by the method of Weber and Osborne ferase activity. (9) or by the slight modification of this method employed by Reagents-Reagent grade organic solvents dried over 4 A Schwartz et al. (10). The latter method differs principally in molecular sieves (Fisher), were used without further purificathat 6 M urea is added to the gel during polymerization.
About tion. Tri-n-octylamine, tributylamine, diphcnylchlorophos-10 pg of protein were applied to each gel. Gels were stained for phate, ethanolamine, and 1, I-carbonyldiimidazole, were purprotein with Coomassie brilliant blue and for carbohydrate with chased from Aldrich Chemical Co. N-Acetyl-D-glucosamine was a periodic acid-Schiff reagent (11).
purchased from Pfanstiehl Laboratories, nucleosides and nucleo-Protein concentrations in impure enzyme preparations were tides from P-L fiiochemicals, ethyl trifluorothiol acetate from estimated routinely by the biuret method (12) or where indi-Pierce Chemical Co., and crystalline phosphoric acid from K cated by the Lowry method (13). The concentration of highly and K Laboratories. Methyl tri-?z-octylammonium hydroxide purified galactosyltransferase was estimated from the extinction was synthesized as described earlier (14). Pancreatic a-amylase coefficient reported earlier (3).
was obtained from Sigma Chemical Co.

Chromatographic
Procedures with SpeciJc Adsorbents-All eq)criments were performed at 4" unless otherwise stated. Synthetic Procedures1 Columns were packed with a slurry of the adsorbent and allowed to settle under gravity.
Prior to application of the sample, 6-Amino-I-hezanol Phosphate (I)-6-Amino-l-hexanol (11.7 g, columns were washed with at least 3 bed volumes of the buffer 0.1 mole) was mixed with 9.8 g (0.1 mole) of crystalline phosin which the sample was contained. After use, UDP-Sepharose phoric acid in a flask attached by a connecting tube with a side alit1 Wacetylglucosamine-Sepharose' were removed from the arm to a receiver immersed in a Dry Ice-ethanol bath. The apparatus was evacuated to less than 0.1 mm Hg with a vacuum 1 The following names will be used throughout the text for the pimp and the reaction flask heated to 150 =t E? in an oil bath. compounds described here and the Sepharose derivatives: UDP-Heating under vacuum was continued for 18 hours. The reachexanolamine refers to PI-(6-amino-1-hexyl)-P2-@'-uridine)-pyrophosphate (VI). UDP-Sepharose is the derivative formed by tion mixture, an amber solid, was cooled to room temperature the reaction of compound VI with cyanogen bromide-treated and then dissolved in 150 to 200 ml of water. The solution was Sepharose 413. Galaetosyl pyrophosphate-hexanolamine refers brought to pH 10.5 with 5 N lithium hydroxide and the precipito PI-(6.amino-1-hexyl)-P2-or-D-galactopyranosyl pyrophosphate tated lithium phosphate was removed by filtration through a (VII).
acet.ic acid and passed over a column containing 150 ml of Dowes 50-X8 (pyridinium cycle; 20 to 50 mesh). The ninhydrinposit'ive eluate was concentrated and t,hen dried in vucuo for 2 to 3 hours at 100" to remove 1)yridinium acetate.
Recrystallization from water by the addition of ethyl alcohol gave material which has m.1). 245"; yield 10.0 g, 51%. The compound has pK', values of 2.0, 6.3,aiid 11.0. CaHltiNOJ? (197) Calculated: C 36. 5.i,II 8.18,N 7.10 Found: C 36.18,H 8.20,N 7.06 N-Tri$uoroacefyl6-AnGo-I-hex:nnol Phosphate (II)-To 2 g of 6-amino-l-hesanol phosphate, dissolved in 20 ml of water at O", was added 1 ml of ethyl trifluorothiol acetate. The reaction mixture was stirred vigorously wit,11 a magnetic stirrer t,o ljroduce a fiiie dispersion of the ethyl trilluorothiol acetate and t,he temperature kept below 4" with an ice bath. The reaction mixture was adjusted intermittently to pH 9.5 with 5 N lithium hydroxide and the course of the rraction was monitored by spotting an aliquot on paper, spraying with ninhydrin and heating at 100" for 5 min. After approximately 1 hour a second l-ml portion of ethyl trifluorothiol acetate was added and the reaction was judged to be complete (45 min) when a pink color wa.s obtained with ninhydrin.
The reaction mixture was adjusted to pH 5 with trifluoroacetic acid, and then concentrated to dryness at 40". The residue was dissolved in water and then concentrated several times to remove residual reagents and finally dissolved in 50 ml of Fvater, cooled, and brought to pH 1.5 with trifluoroacetic acid. This solution was passed over a column containing 25 ml of Dowex 50-X8 (II+ cycle; 20 to 50 mesh) to remove unreacted amine. The acidic &ate was concentrated to dryness under vacuum at 45" and then dried in vucuo over anhydrous calcium sulfate for 24 hours. The syrupy product (yield 95% by weight) had RF 0.68 in Solvent 1 and RF 0.58 in Solvent 2 as judged with the molybdate spray for phosphate.
It contained only traces of the starting material (molybdate and ninhydrin) and could be stored in a desiccator for at least 2 months without decompositiou. On storage, crystallization occurred but attempts to find a suitable solvent for recrystallization were unsuccessful. The product was ninhydrin-negative except in concentrations above 5'; when a pink color is apparent.
Treatment of a 1 m&f solu tion in water at pH 11 .O for 30 min results in the complete con version of the N-trifluoroacetate to the starting amine as judged by chromatography in Solvents 1 and 2. ,2'-Tri$uoroacetyl-0-phosphoryl 6-Amino-i-hexanol Imidazolide (III)-To a solution of 200 mg (0.68 mmolc) of Compound II in 2 ml of acetone were added 160 mg (1.0 mmole) of 1 ,l'carbonyl diimidazole.
The reagent dissolved rapidly and a small volume of gas was evolved; the reaction was left at room temperature under a drying tube for 6 hours and then 0.015 ml (0.36 mmole) of methyl alcohol was added to hydrolyze excess diimidazole (16). After 20 min the mixture was concentrated to dryness at 45". Chromatography of the dry residue in Solvents 1 and 2 showed onIy product with RF 0.8 and 0.75, respectively.
The product is stable in dimethylformamide solution for at least 4 weeks. It is less stable in aqueous solution and is hydrolyzed to the extent of about 507; in 24 hours at pH 7.5. p1-(6-Amino-l-Zlezyl)-P2~(5'-uridine)-pyrophosphafe (VI) by Imidazolide Stefhod-The imidazolide (III) from 3.1 g (10.5 mmoles) of Compound II was dissolved in 4 ml of dimethylformamide.
To this solution was added 5 mmoles of UAIP-me&y1 trimn-octylammonium sslt in 8 ml of dimcthylformamide. The reaction vessel was stored in a desiccator over Drier& at room temperature for 24 hours at which time the solution was chromatographed as shown in Fig. 1. The adsorbance of the eluat'e fractions was measured at 262 nm by diluting lo-p1 samples to 1 .O ml. The desired product (A-triRuoroacetyl-UDYhexanolamine; Compound IV) accounted for approximately 70% of the starting adsorbencc and was eluted between tubes 70 and 110. The material in these fractions earl be used for the prep aration of the UDP-Sepharose adsorbent without further purification after incubation (10 to 20 pmoles per ml) at pH 11 for 4 hours at 25" to hydrolyze the trifluoroacet,yl derivative to the free amine (VI).
Compound IV was isolated from the column eluate and characterized as follows.
Fractions were combined and adjusted to pH 8.0 with 1 N lithium hydroxide and then concentrated to dryness at 45". The syrupy residue is extracted five to seven times by centrifugation with 50 ml of ethanol-ethyl ether (1:2, v/v) to remove lithium chloride.
The resulting residue is ninhydrin-negative but gives a strong positive reaction after incubation at pH 11 to 12 for a few minutes at 25". Its molar extinction at 262 nm is lo* based on a molecular weight of 611 for the dilithium salt. It has RF of 0.45 in Solvent 1 and 0.5 in Solvent 2. Hydrolysis with 1 x hydrochloric acid at 100" for 1 hour yields a mixture of Compound II and UMP.
Hydrolysis at pH 12 gives a product that is eluted from Dowex 1 (Cl-) as a single band in the same region as UMl' as shown in Pig. 1 but with a strong positive ninhydrin reaction. This product after acid hydrolysis yields a mixture of Compound I and IMP as determined by chromatography in Solvents 1 and 2. k E Eiz; Sr FRACTION NUMBER FIG.-1. The purification of trifluoroacetylhexanolamine-UDP and UDP-hexanolamine by chromatography on Dowex 1 (Clcycle). Trifluoroacetylhexanolamine-UDP, the reaction mixture from 1 mmole of UIVIP was applied to a column (1 X 25 cm) of Dowrex l-X2 (Cl-cycle; 200 to 400 mesh) and eluted with 250 ml of 50% aqueous methanol and then with a linear gradient of 500 ml of 0.01 N HCI as starting solvent and 500 ml of 0.01 M HCI containing 0.4 M lithium chloride as the limit solvent.
The column was operated at 25" at a flow rate of about 60 ml per hour. The elution pattern is shown by the solid line.
After 2 hours at, room temperature the mixture was dried by rotary evapora_tion and the residue extracted twice with 50 ml of ethyl ether.
The residue was thoroughly dried under vacuum and then dissolved in 5 ml of dry pyridine containing 2 mmoles of trifluoroacetylhesanolamine phosphate.
After 1 hour at 25" the pyddine was removed by rotary evaporation and dissolved in ethanol.
The solution was mixed with a 5.fold excess of Dowes 50 (Hf form) for 5 min and filtered.
The filtrate was dried, dissolved in water and the desired product purified as shown in Fig. 1.
PI-(6.Amino-i-hex&P2-galacfopyranosql Pyrophosphate (VII) -The reaction of 0.22 mmole of the methyl tri-n-octylammonium salt of a-n-galactopyranosyl phosphate with the imidazolide (III) from 150 mg of Compound II (0.45 mmole) in 3 ml of dimethylformamide was followed by chromatography in Solvent 2. cY-n-Galactopyranosyl phosphate has RF 0.50. The reaction was complete after 14 hours at room temperature.
A precipitate, which formed on mixing the dimethylformamide solutions of the reactants, redissolved when the reaction was complete. The product of this cocdensation was isolated by chromatography on Dowex 1 (CI-) as described for the UMP derivative and emerged in approximately the same position in the lithium chloride gradient.
The compound was located by phosphate analysis and had t,he following characteristics.
Methyl Tri-n-octylammonium Salts of Phosphate Esters-Sodium, potassium, or calcium salts of various phosphate esters were converted to the corresponding free acids by treating a solution or suspension with Dowex 50-X8 (II+ cycle; 20 to 50 mesh) at room temperature.
The resin was removed by filtration and washed several times with water.
To the combined filtrate and washings was added 1 molar equivalent of methyl tri-n-octylammonium hydroxide in ethanol-water solution. If necessary ethanol was added to give a homogeneous solution and then the solvents were removed in vucuo at 45". The residue was dried by evaporation from solution in dimethylformamide and stored in vucu~ over anhydrous calcium sulfate.
W-TriJluoroacetylhexanolamine (VIII)-To 11.7 g (0.1 mole) of 6.amino-l-hexanol dissolved in 100 ml of dioxane was added 18 ml (0.11 mole) of ethyl trifluorothiol acetate in l-ml portions during 3 hours. 'The reaction mixture was stirred continuously with a magnetic stirrer in a fume hood. The reaction was judged to be complete when a ninhydrin test of the mixture spotted on paper gave a faint pink color. After removal of the solvent and concentration from dioxane several times the residual syrup was stored over Drierite in vacua. Crystallization occurred during 24 hours. The material (m.p. 54") was obtained in quantitative yield and gave a negative ninhydrin reaction.
The mixture was stirred to dissolve the reactants and then stored in a desiccator over calcium sulfate and potassium hydroxide pellets under a slight vacuum.
After 48 hours the reaction rnixture was concentrated at 50" and the residue partitioned between methylene chloride and water.
The methglene chloride layer was extracted several times with water, dried over powdered anhydrous magnesium sulfate, and then concentrated to give 3.5 g of crystalline material.
Recrystallization from ethyl acetate (30 ml) by the addition of hexane gave several crops of crystals.
lifter 2 hours at room temperature the solution was brought to neutrality with carbon dioxide, filtered through a pad of Celite, and taken to dryness at 45". The residue melted from 149-158" with decomposition; [(Y]: = -19.5 f 1.0" (5.2 g per 100 ml of HzO).
C1,H,sN,OsF, (434.07) Calculated: C 44.26, H 6.46, N 6.73 Found : C 41.28, II 6.68, N 6.87 Preparation of Sepharose Adsorbents-The procedure described by Cuatrecasas (18) was used with modification as follows. Sepharose 4B (Pharmacia) was washed with 10 or more bed volumes of distilled water on a coarse sintered glass funnel.
The bulk of the water was removed and the firm gel (500 g) was mixed with water (300 ml).
Cyanogen bromide (from 150 to 200 mg per g of gel) was pulverized in a mortar in a hood and then added in small batches over a period of 1 to 2 min, with stirring to the gel suspension.
When all of the cyanogen bromide had been added, the mixture was brought to pII 11.0 f 0.2 by the addition of 4 N sodium hydroxide.
The temperature was kept at 20" by the addition of chipped ice. The consumption of base was rapid over a 15.min period at which time all of the cyanogen bromide was in solution.
The reaction was then stopped by the addition of 400 to 500 ml of chipped ice and the mixture of ice and gel filtered with suction on a 2-liter coarse sintered glass funnel.
The gel was washed during 3 to 4 min with 10 to 12 bed volumes of ice-cold water.
The residual ice was separated from the gel and the gel quickly mixed with a solution containing 1 to 2 mmoles of the appropriate ligand in 200 to 300 ml of water at pH 10.0. The misture was adjusted to pH 10 if necessary and the slurry stirred at 4" for 12 to 18 hours. The gel was then packed in a column and wvashed with water at room temperature.
The ninhydrin-positive or ultraviolet-absorbing washings were collected and t'he amount of unrracted ligand estimated.
The procedure gave adsorbents with 1 to 6 pmoles of ligand per ml of packed, wet gel. The concentrations of the ligands in t.he adsorbents used in this study were e&mated to be as follows. UDP-Sepharose contained 3 to 6 pmoles of UDP per ml. This was estimated on the basis of the loss of material adsorbing at 262 nm during coupling.
IJsually 70 to 80% of the ligand added is coupled.
Estimation of the extent of subst'itution by acidcatalyzed hydrolysis of the pyrophosphate bond to the adsorbent is complicated by the concomitant degradation of the Sepharose to ultraviolet-adsorbing ma,terial. Galactosyl pyrophosphate-Sepharose contained 2 pmoles of galactose per ml. This was estima,ted as the loss of organic phosphate in an aliquot of the Sepharose-free reaction mixture during coupling. Hydrolysis was performed in 0.5 1"~ hydrochloric acid at 100" for 1 hour. Similar values were obtaiued by hydrolysis of the final adsorbent under the same conditions. !\'-Acet'ylglucosamine-Sepharose contained 3 to 4 pmoles of N-acctylglucosamine per ml. This was estimated as glucosamine by ion exchange chromat'ography after hydrolysis of t'he gel with 6 x HCl at 110". The capacity of the adsorbents and the pattern of affinity chromatography will probably be a function of the amount of the lignnd per ml of a,clsorbent but this has not been examined thoroughly at present.

Synthesis of Spec$c Adsorbents
In order to test the applicability of specific adsorbents for affinity chromatography of glycosyltransferases, ligands were synthesized which were structurally related to the substrates or inhibitors of the UDPgalactose :N-acetylglucosamine galactosyltransferase of bovine milk. Three types of ligands were synthesized containing either a uridine pyrophosphate, a galactosyl pyrophosphate or an N-acetylglucosamine substituent. In each ligand one of t'hese funct.ional groups was linked to the hydroxyl group of 6-amino-1-hexanol.
Each ligand was then coupled to cyanogen bromide-treated Sepharose through the free amino group in the hexanolamine moiety.
Synthesis of UDP-hexanolamine and Galactosyl Pyrophosphate-Hexanolamine-Synthesis of these two ligands involved the formation of an unsymmetrically substituted pyrophosphate with uridine or galactose and an alkyl amine.
The latter moiety contained the amino group required for coupling to activated agarose particles.
Two methods of synthesis that appeared to have general applicability were explored. The anion displacement method (14), which involves activat'ion of a phosphate group by conversion to the diphenylpyrophosphoryl derivative as shown in Fig. 2, was utilized to synthesize N-trifluoroacetyl-UDl'-hexanolamine (IV) in 6O70 yield from UMP.
The product as isolated by chromat'ography on Dowex 1 (Cl-cycle) was contaminated with the symmetrical compound PI-!+diuridine 5'-pyrophosphate. This compound can be produced in the first step of the synthesis and its occurrence presents a major difficulty with this method.
Pure material can be obtained by removing the trifluoroacetyl group by hydrolysis above pH 11.5 for 4 hours at 25" and rechromato- The synthesis of UI>P-hexanolamine by the anion displacement method (14).
The desired product has gained a positive charge compared to Compound IV and is well separated from material that elutes in the region where simple monophosphates are found, while diuridine 5'-pyrophosphate is unchanged by the hydrolysis step. The purified material can be used for the preparation of affinity columns after the first Dowex 1 column without being isolated from the eluent.

Fractions containing
Compound IV and diuridine 5'-pyrophosphate were pooled, concentrated to a convenient volume, and adjusted to pH 12.0. After 4 hours the solution is adjusted to pH 10 and used directly for coupling.
Neither diuridine 5'pyrophosphate nor lithium chloride interfere with the coupling reaction.
The imidazolide method (19) provides a more convenient and more generally applicable approach to the synthesis of pyrophosphoryl ligands.
The sequence of reactions is shown in Fig. 3. This procedure allows the synthesis of N-trifluoroacetyl-6amino-1-hexanol phosphate imidazolide (III) which is apparently quite stable if protected from moisture and which reacts with phosphate or pyrophosphate esters to form the corresponding di-and triphosphates in high yield (60 to 80%).
The synthesis of the 5'-uridine diphosphate and u-galactopyranosyl pyrophosphate derivatives are presented under "Experimental Procedure." The synthesis of the corresponding adenosine, guanosine, and n-glucopyranosyl derivatives has also been accomplished2; thus, the synthetic methods employed here may provide other useful specific adsorbents.
Synthesis of 6-Amino-i-hexyl-d-acetamido-d-deoxy-P-o-glycopyranoside-This compound was synthesized as shown in Fig. 4. N-Trifluoroacetyl-6-amino-1-hexanol was condensed with 3,4,6tri-O-acetSl-2-acetamido-2-deox3r-cY-n-glucopyranosyl chloride in dimethylformamide solution in the presence of mercuric cyanide. The condensation product was a mixture of 01 and p anomers P H2N /vvc O-P-OH f HSC2 HS F3C-C-SC2 H5 which were separated by fractional crystallization from ethyl acetate-hexane mixtures. Deacylation of the /3 anomer with barium methylate in methanol afforded the desired product as its trifluoroacetyl salt. The ligand was reacted with Sepharose 413 as described under "Experimental Procedure" to give the specific adsorbent.
Aginity Chromatography of UDP-galactose:hr-Acetylglucosamine Galactosyltransferase The effectiveness of the specific adsorbents used in these studies was tested primarily with the galactosyltransferase of lactose synthetase (3). Bovine milk was used as the source of the enzyme in all studies reported here.
Chromatography on UDP-hexanolamine-Xepharose-Although the transferase can be adsorbed directly from whole or skim milk, the casein and lipid in milk markedly decreased not only the capacity but also the flow rate of columns of the adsorbent. For these reasons, the effectiveness of the adsorbent was tested with whey, which contains most of the transferase in milk but little casein or lipid.
Preliminary experiments indicated that the transferase from whey was bound weakly on UDPSepharose buffered at pH 7.4 with 0.025 M sodium cacodylate but &at manganous ions were necessary for maximum adsorption.
Once the enzyme was bound, the adsorbent could be washed with very large volumes of manganous ion-cont'aining buffers without eluting the enzyme.
The capacit,y of the UDP-Sepharose to bind the enzyme in whey cont,aining manganous ions was not altered by high concentrations of sodium chloride (0.5 to 1.0 M), indicating that the enzyme was not binding principally through an ion exchange process. Fig. 5 shows that the transferase bound to UDP-Sepharose can be eluted by several means. EDT& which was required to combine with manganous ions, and X-acctylglucosamine which stabilizes the enzyme (3) mere used under all condit'ions test'ed. Dilute sohrtions of urea (1.5 M) were most effective in eluting t'he enzyme in a small volume.
Elution with urea solutions, however, gave about half the expected yield of enzyme.
Borate buffers at pH 8.5 containing EDT& eluted t,he enzyme in better yields, but the enzyme emerged in a somewhat larger volume than that required for elution with urea. Although the exact basis for the action of borate has not been established it is possible that it forms an addition compound with the ribose moiety of UDP and thereby weakens the binding of the enzyme. Elution with buffers containing magnesium ions in the absence of EDTA was successful, presumably by competing with manganous ions, but the enzyme was eluted in a large volume.
The most satisfactory yields of enzyme were obtained on elution with cacodylate buffers containing EDT& despite the fact that the enzyme lvas eluted in a larger volume than with borate buffer or buffered solutions of urea. Ba.sed upon the results of the above preliminary studies, it was found that columns of UDP-Sepharose could be used routinely on a preparative scale to purify the enzyme from whey under the conditions shown in Fig. 6. The enzyme was adsorbed from whey containing manganous ions whereas the majority of the whey proteins passed unretarded through the column.
After washing the column with buffer, the active enzyme was eluted in buffer containing EDTA and N-acetylglucosamine.
Although an extremely dilute solution of enzyme is obtained, it can be concentrated readily by ultrafiltration or readily adsorbed on cu-lactalbumin-Sepharose on N-acetylglucosamine-Sepharose as described below. The extent of purification of the tranxferase can be judged by polyacrylamide gel clectrophoresis of appropriate fractions in sodium dodecyl sulfate.
Unfortunately, as noted earlier (3), the transferase will not give sharp, discrete bands on electrophoresis under nondenaturing conditions. Fig. 7 shows the electrophoretic patterns of whey, of the proteins in whey not adsorbed to UDP-Sepharose and the proteins adsorbed to UDP-Sepharosc.
Kechromatography of the adsorbed proteins did not result in further purification. Chromatography on N-Acetylglucosamine-Sepharase-It byas found that the galactosyltransferase in whey or in partially purified preparations from UDP-Sepharose could be effectively adsorbed to N-acetylglucosamine-Sepharose. It was also observed, however, that the binding of the transferase to this adsorbent is enhanced considerably by UDP and manganous ions, as shown in Fig. 8. Essentially identical results were obtained when UMP was used instead of UDP. Fig. 9 shows the effectiveness of N-acetylglucosamine, borate buffer and buffered solutions of urea in eluting the transferase from N-acetylglucosamine-Sepharose.
Urea and borate are much more effective than N-acetylglucosamine for elution of the enzyme, but the yields of enzyme appear to be somewhat lower. In addition, as shown by electrophoretic analyses in Fig. 10, urea and borate also appear to elute proteins which are not found on elution with N-acetylglucosamine.
Thus, the most suitable conditions for chromatography of the transferase utilize UMP and Mn+f during adsorption of the enzyme and buffers containing N-acetylglucosamine and EDTA for elution of the enzyme. The use of N-acetylglucosamine-Sepharose for purification of the transferase under these conditions is shown in Fig. 11. Appropriate fractions from the chromatogram in Fig. 5 were analyzed elcctrophoretically on gels containing sodium dodecyl sulfate by the method of Srhwartz el al. (10). Gel 1, whey; Gel 2, proteins from whey not adsorbed to UDP-Yepharose (pooled fractions between 0.0.5 and 2 liters in Fig. 5); Gel 9, proteins from whey adsorbed to UDPsepharose (pooled fractions between 2.9 and 6 liters in Fig. 5); Gel 4, the galactosyltransferase obtained by affinity chromatography of the proteins adsorbed to 'UDP-Scpharose on a column of a-tactalbumin-Sepharose.
The latter column was operated as described in Table I Two columns (0.6 X 3 cm) of N-acetylglucosamine-Sepharosc at 5" were eqnilibrated with 0.025 u sodium cacodylate buffer, pH 7.4, containing 0.0% Y manganous chloride and 0.001 hi mercaptoethanol.
A solution (30 ml) of the galactosyltransferase, which was purified partially on a UDP-Scpharose column as shown in Fig. 3, and which had been dialyzed against the equilibration buffer. was Frc:. 9. l*:lntion of galactosyltransfer:~sc from N-acetylglnrosamine-Sepharose.
Three columns (2.2 X 10 cm) of N-acetylglucosaminc-Sepharose were equilibrated with 0.025 Y sodium cacodylate, pH 7.4, containing 0.001 M mercaptoethanol, 0.02.i \r manganous chloride, and 3 X 10-4 M UMP. A solut.ion (200 ml) of partially purified enzyme from a IJDP-Sepharose column as described in Fig. 5 and dialyzed against. the equilibration buffer was applied to each column at .i". After the enzyme wa9 applied, the colnmns were washed with 200 ml of the equilibration bnlfcr at .4 (nrrow) and then with one of three buffers at B (arrolu Ii-hey by .i@il~ Chromatography-On the basis of the behavior of the galactosyltransferasc with the specific adsorbents described here it was possible to devise a purification of the transfcrase from whey based solely on affinity chromatography. This procedure is summarized in Table I. The enzyme from whey was adsorbed onto, and rluted from a column of I-DP-Sepharose (Step 3) as described in Fig. 6. Ikcau~e it has a much larger capacity for the enzyme than the othrr adsorbent+ lYI)l'-Sepharose sccnwd best suited for adsorbing the enzymr frorn whey, although it is less specific than the other adsorhents.
a-Lactalbumin-Sepharose and N-acetylglucosaminc-Sepharosr are also capable of ndsorbing enzyme directly frorn whey but are best used at later stages of the purification procedure.
hs shown in Table I, the enzyme from the UDP-Sepharose column can he purified further on a column of cr-lactalburnin-Sepharose.
The enzyme obtained at this stq has a constant specific activity on rechromatography on any of the specific adsorbents and gave the gel pattern shown in Fig. 7. It is noteworthy that this transferast preparation from a-lartalbumin-Sepharo*l columns with a constant qwific activity gives two bands on the sodium dodccyl sulfate gels prcthen applied to each column. One enzyme solution also contained 2.5 X IO-4 M Ul)P as indicated in the figure by the closed circles.
After the enzyme was applied t.he columns were washed with buffered urea solutions (ar,ozc) as described in Fig. 9. Fractions (I ml) were collected and assayed for transferase activity. Enzyme applied in the absence and presence of UDP are indicated by the open and closed circles, respectively. pared by the method of Schwartz et al. (10). The major band gives an average molecular weight of about 51,000 and the minor band a molecular weight of about 43,000. In contrast, the same preparation gives three distinct bands as noted below (Fig. 12) when analyzed by the method of Weber and Osborne (9).
N-Acetylglucosamine-Sepharose can also be used on a preparative scale to purify the transferase. This is indicated by the results in Table II which summarizes the purification of the enzyme from whey. After the enzyme was partially purified with UDP-Sepharose (Step 3) it was then applied to a column of N-acetylglucosamine-Sepharose ( Step 4) as shown in Fig. 11.
Clearly, this specific adsorbent is not as effective as a-lactal-  10. Electrophoretic patterns on polyacrylamide gels in sodium dodecyl sulfate of galactosyltransferase eluted from N-acet,ylglucosamine-Sepharose as shown in Fig. 9. Gel 1, enzyme eluted with buffer containing 1.5 M urea, EDTA, and N-acetylglucosamine.
Gel 2, enzyme eluted with borate buffer containing EDTA and Iv-acetylglucosamine.
Gel S, enzyme eluted with buffer containing EDTA and N-acetylglucosamine.
The molecular weights (M.W.) are those estimated by comparison with standards of known molecular weight.
The gels were prepared and operated by the method of Schwartz et al. (10).
It is noteworthy that either ar-la&albumin-Sepharose or Nacetylglucosaminc-Scpharose can adsorb the enzyme from very dilute solutions, such as those obtained after elution of the enzyme from UDP-Sepharose (Fig. 6), however, better adsorption is obtained if the dilute solutions are concentrated about 5-fold before being applied to these specific adsorbents. The binding constants for the enzyme to the two adsorbents have not been determined and the relationship between concentration of enzyme and binding to these adsorbents cannot be stated quantitatively at present.

Properties of Galactosyltransjerase Purdfed by A$nity
Chromatography Repeated chromatography of the galactosyltransferase on ar-lactalbumin-Sepharose as described in Table I  A solution of transferase (2000 ml), which had been partially purified on UDP-Sepharose as described in Fig. 5, was concentrated to a volume of about 400 ml, dialyzed against 0.025 M sodium cacodylate, pH 7.4, containing 0.01 M mercaptoethanol, 0.025 M manganous chloride, and 5 X 10-' UMP, and then applied to a column (2.2 X 10 cm) of N-acetylglucosamine-Sepharose at 5" as described in Fig. 9. The column was washed with equilibration buffer (400 ml) at A (arrow) and then at B (arrow) with the buffer used with Column 3 in Fig. 9. The flow rate was about 50 ml per hour. The protein concentration as judged by the Lowry method (0) and enzymatic activity (0) of the fractions (10 ml) from the column were measured. TABLE I Purification of galactosyltransferase by aflnity chromatography with UDP-Sepharose and a-lactalbumin-Sepharose Whey, which was prepared from raw skim milk as described under "Experimental Procedure" was applied to and eluted from a UDP-Sepharose column (7.8 X 30 cm) as indicated in Fig. 5 to give the preparation shown at Step 3. The partially purified enzyme from Step 3 was applied to a column of or-lactalbumin-Sepharose (3 X 15 cm) and the absorbed enzyme eluted as described earlier (3), except that the enzyme was applied in 0.005 M N-acetylglucosamine rather than glucose. crease the specific activity of the transferase. The specific activity of different preparations from a-lactalbumin-Sepharose columns varies from about 11 to 15 units per mg (3) but the preparations are indistinguishable electrophoretically and in amino acid composition. It appears, however, that the transferase purified solely by affinity chromatography (Tables I and  II) is heterogeneous in size as shown by the electrophoretic analyses in sodium dodecyl sulfate (Fig. 7). The gel patterns given by preparations purified on either cr-la&albumin-Sepharose or N-acetylglucosamine-Sepharose are compared in Fig. 12  Gel 1, transferase prepared by the method of Trayer and Hill (3). Gel 2, transferase prepared solely by affinity chromatography first with UDP-Sepharose and then with N-acetylglucosamine-Sepharose (Table II). Gel 3, transferase prepared solely by affinity chromatography first with UDP-Sepharose and then with a-lactalbumin-Sepharose.
The molecular weights (M.W.) are those estimated by comparison with standards of known molecular weight.
The gels were prepared and operated as described by Weber and Osborne (9). The major bands on each gel were detected with Coomassie blue but each band also stained for carbohydrate when the gels were developed with the Schiff -periodate reagent (11).
those of the enzyme prepared by the method of Trayer and Hill (3). These gels show that the enzyme prepared solely by affinity chromatography contains at least three different species with approximate molecular weights of 54,000, 49,000, or 43,000. Three species are also noted in the enzyme prepared by N-acetylglucosamine-Sepharose affinity chromatography, but the species with a molecular weight of 54,000 appears to be present in larger amounts. Chromatography of this preparation on cY-lactalbumin-Sepharose columns reduced the amount of the higher molecular weight species to give a pattern essentially identical to that on Gel S in Fig. 12 and the specific activity increased slightly. In contrast, a preparation of enzyme prepared by the method of Trayer and Hill (3) gave only one major band with a molecular weight of about 43,000, and three bands in minor amount which do not correspond to other species of the transferase.
The basis for the differences among these species is unknown, but the amino acid composition of the enzyme prepared as described in Table I and containing three components is indistinguishable from that reported earlier (3) for a species of 42,000 molecular weight. In addition, these compositions are indistinguishable from those for each of the bands shown in Fig. 12 (these compositions were obtained by analysis of acid hydrolysates of gel slices containing the appropriate band as described earlier (10)). It is possible that the three species differ in carbohydrate content although each species on a gel gives stains positively for carbohydrate with the periodic acid-Schiff reagent (11).

Conversion of N-Acetylglucosamine-Sepharose
to N-Acetyllactosamine-Sepharose by Galactosyltransjerase N-Acetylglucosamine-Sepharose proved to be an acceptor substrate for the galactosyltransferase. The adsorbent (1.5 ml) was saturated with enzyme as shown in Fig. 9, washed with 0.025 M cacodylate buffer, pH 7.4, containing 0.02 M manganous chloride and 0.02 M mercaptoethanol, and then incubated with 25 pmoles of UDP-[14C]galactose (175,000 cpm total) in a volume of 2.5 ml at 37" for 45 min. After the incubation, the supernatant solution contained 142,000 cpm, which is equivalent to 20 hmoles of UDP-[14C]galactose. The supernatant also contained synthetase activity approximately 4 times that in the supernatant before incubation. The adsorbent was packed in a small Purification 0 A unit is 1 pmole of galactose incorporated per min. b Total activity is activity times volume and is expressed in units. c Specific activity is expressed in units per mg of protein. d One-third of the active fraction from Step 4 was applied to a second NAG-Sepharose column, and one-third to an or-lactalbumin-Sepharose column, as described in Table I. NAG is an abbreviation for N-acetylglucosamine column and washed with 200 bed volumes of water.
The eluent, was checked for radioactivity and none was found after 20 bed volumes had passed through the column.
The adsorbent was then dispersed in an equal volume of 1 N hydrochloric acid and heated at 140" for 30 min to give a slightly amber solution. Aliquots were counted and the whole sample was calculated to contain approximately 37,000 cpm, equivalent to 5.0 pmoles of UDP-[14C]galactose, or 3.3 pmoles of galactose incorporated per ml of resin. This value is in close agreement with the amount of N-acetylglucosamine estimated to be contained in the adsorbent.
When N-acetylglucosamine and UDP-[Wlgalactose are both added to the enzyme adsorbed to N-acetylglucosamine-Sepharose, the incorporation of [14C]galactose into the adsorbent and the release of enzyme into the medium are both inhibited.
The conversion of the unbound N-acetylglucosamine to N-acetyllactosamine was obtained in good yield.

UDP-Sepharose for Pur$cation of Glycogen Synthetase
In order to test whether another UDP-glycosyltransfernse could be purified with the aid of UDP-Sepharose, studies were performed with glycogen synthetase.
The partially purified enzyme (44-fold) for these experiments was obtained from rabbit muscle by differential centrifugation as described earlier (20). This was the glycogen-rich fraction containing 70 to 80% of the synthetase activity of the muscle. The impure preparation was treated with pancreatic cu-amylase to reduce the glycogen content and then chromatographed as shown in Fig. 13. A small amount of synthetase emerged unretarded from this column and was associated with opalescent material.
Inactive protein was washed from the column after the sample was applied with buffer containing EDTR.
Synthetase activity did not appear during washing but was eluted only when glycogen was added to the wash buffer. In a separate experiment, the synthetase activity was not removed from the column on washing with buffer containing 10 m&f UDP.
Analysis of the synthetase from the col- FIG. 13. The glycogen-rich fraction containing glycogen synthetase (20) from 600 g of rabbit muscle was incubated with pancreatic a-amylase (2 mg) in 50  umn by gel electrophoresis in sodium dodecyl sulfate revealed one major band with an apparent molecular weight of about 90,000. No attempts were made to purify further the synthetase on this column or to assess its chemical and physica. properties. The synthetase was, however, free of glpcogen phosphorylase activity.

DISCUSSION
The synthesis of agarose derivatives useful in enzyme purification can frequently be accomplished by reacting a suitable ligand such as a substrate or inhibitor containing an amino function with cyanogen bromide activated agarose as described by Porath and Axon (21), and Cuatrecasas (18). In some instances a ligand attached directly to the agarose matrix is ineffective and it is necessary to allow it to extend from the matrix through a linear side chain and thereby enhance the affinity of t,he enzyme (22). Ligands without amino functions can bc attached by a variety of methods (18) but in all cases the derivatization involves reaction of agarose with an amine as an initial event. Of the ligands of interest in this study only some of the nucleoside bases have an amino function capable of reaction with activated agarose and reaction of these would not only mask groups that may be essential for binding the enzyme but also place ligands very close to the agarose matrix.
The syntheses described here were designed to generate ligands with a side chain containing a terminal amino group suitable for direct condensation with cyanogen bromide-activated agarose and of sufficient length to decrease steric interference for binding of proteins. Direct coupling of a previously formed ligand of suitable structure also assures that the agarose derivative possessed a single kind of functional group. Synthesis of adsorbents by reaction of a ligand with a preformed arm on the agarose may lead to mixed function adsorbents because of incomplete reactions. Such adsorbents would have less specificity because of nonspcrific adsorption, particula,rly by ionic interactions.
The synthesis of UDP-hesanolamine (VI) was accomplished successfully by two different means. The anion displacement method (14) (Fig. 2) gave good yields but in preliminary studies2 with other nucleosides (ARIP, GMP) this procedure gave large amounts of symmetrical diphosphates and the desired product was difficult to purify. This method could be made more generally applicable if the diyhenylphosphoryl derivative of N-trifluoroacctyl-6-amino-1-hexanol phosphate could be synthesized and then coupled with UDP or other appropriate phosphates. Unfortunately, attempts to synthesize UDP-hexanolamine by this means were unsuccessful.
The imidazolide method (19) also gave adequate yields of both UDP-hexanolamine and t.he galactosylpyrophosphate-hexanolamine and appears to bc the more generally applicable method. The N-trifluoroacctpl-6amino-l-hexanol phosphate imidazolide in principle can react with many other phosphat,e derivatives and has been used successfully for synthesis of GDP-and ADP-hexanolamine.2 It is anticipated that the specific adsorbents of the type used in this study can be applitd generally for the purification of a wide variety of enzymes.
In principle, any enzyme with a uridine phosphate derivative as a substrate or a product could be purified with the aid of UDP-Sepharose.
Many enzymes of this type have K, or Kl values for uridine derivatives of t,he same order of magnit.ude of those for the galactosyltransferase studied here. UDP and UDl-'-glucose, two inhibitors of the transferase, have KI values of 7 x lo-" M (23,24) and 10e4 RI (23), respectively, and the K, for UDP-galactose, its normal substrate, is about 6 x 10-j M (23,24). Thus, especially low K, or K1 values are not necessary for effective binding.
It should be noted, however, that the galactosyltransferase is not bound by the galactosyl pyrophosphate-Sepharose.
This suggests that the major binding energy of I'DP-galactose is contributed by the UDP moiety. In addition, the enzyme is poorly bound to N-acetylglucosamine-Scpharose, in accord with the observation that the iY, for N-acetylglucosamine is only about 5 X lop3 M. It has been shown, however, that the K, for N-acetylglucosamine is decreased by IJDP-galactose (23), and as expected, the transferase is adsorbed effectively by Nacetylglucosamine-Sepharose in the presence of UDP or UMP. Thus, it would appear possible in the case of glycosyltransferases to effect considerable purification, first on a specific adsorbent with a ligand resembling the nucleotide portion of the substrate and then on another adsorbent containing a ligand resembling the acceptor substrate.
Studies to test this possibility for other glycosyltransferases arc now in progress. The preliminary studies with glycogen synthetase reported here indicate that in some cases successful purificat)ion can be achieved.
The fact that N-acetylglucosamine-Sepharose serves as an acceptor substrate for the gnlactosyltransferase is of interest in three major respects.
First, the N-acetyllactosamine-Sephnrose that is formed on reaction of the enzyme saturated adsorbent with UDP-galactose, should serve as an effective specific adsorbent for other transferases which use N-acetyllactosamine as a substrate. Secondly, it should be possible to synthesize polysaccharides of known sequence beginning with an appropriate monosaccharide-Sepharosc derivative if other glycosyltransferases are available to add the appropriate sugars. In this manner several very specific ndsorbents could be synthesized and, indeed, if some selective means were available to cleave t,he cnzgmically synthesized polysaccharide from the Sepha.rose, a solid phase type of synthesis of polysaccharides would be theoretically possible.
Finally, prelimina~ry studies (25) have shown that the N-acetylglucosnminc-Sepharose derivative may be useful in detecting t,he galactosyltraiisferase on cell surfaces. In this respect the specific adsorbcnts may be useful in evaluating t.he role of int.ercellular glycosylation in cellular adhesion (26).
Further insight into the nature of the galactosyltransferase of lactose synthetase has also been obtained from the studies presented here. Although it is &dent that the specific adsorbents can be employed to obtain a ltighly purified transferase it is of interest that the propertics of the adsorbents reflect the order of binding as well as the interactions among the transferase substrates. Earlier studies (23,24) have shown that the transferase acts by an ordered mechanism and that it must bind its substrates in the sequence manganous ion, UDP-galactose, and Nacctylglucosamine for catalysis to occur. The enhancement of the binding of enzyme to 1'1)PSepharose by mangnnous ions supports this mechanism.
In addition, the fact that 1:DP and one major species with a molecular weight of about 42,000. In contrast, at Ieast three species with apparent molecular weights of about 54,000, 49,000, and 42,000, respectively, are clearly observed on purifying the enzyme solely by affinity chromatography.
It is probable that species with molecular weights greater than 42,000 were removed on purification by the methods reported earlier.
In this respect, it is of interest that the transferase prepared by methods which do not employ affinity chromatography is also reported to be homogeneous with a molecular weight of about 44,000 but containing only about 5% carbohydrate (27). Each species isolated as described here appears to be closely related since each has an amino acid composition very similar to one another as well as to that reported earlier (3) for the species of 42,000 molecular weight.
In addition, the species of 42,000 molecular weight has essentially the same specific activity and K, for UDP-galactose and N-acetylglucosamine as preparations which contain three species. Magee et al. (28) have also observed two distinct species of the transferase after electrophoresis under nondenaturing conditions. Both species were enzymically active and differed slightly in molecular weight. Since each species purified as described here is a glycoprotein, they may vary in carbohydrate content.
It is also possible that they arc the result of limited proteolysis; however, further studies will be required to establish t.he exact basis for the heterogeneity. Isolation and characterization of the enzyme directly from mammary glands may shed considerable light, on this problem.