Glycosaminoglycan Sulfotransferases in Human and Animal Sera*

droitin 4/6-sulfate (80% 4-sulfate and 20% 6-sulfate), and UDP-N-acetylgalactosamine 4-sulfate were used as acceptors for the measurement of 3’-phosphoade-nylyl su1fate:glycosaminoglycan sulfotransferase ac- tivities in human serum. Chromatographic fractionation of the serum followed by determination of the sulfotransferase activities demonstrated the existence of at least four different sulfotransferases capable of introducing sulfate to 1) position 6 of the internal N-acetylgalactosamine units of chondroitin, 2) position 6 of the nonreducing terminal N-acetylgalactosamine 4-sulfate unit of chondroitin 4/6-sulfateY 3) position 2 (amino group) of the glucosamine units in heparan sulfate, and 4) the sugar units in keratan sulfate, respectively. The fourth activity was separated into two subfractions with different specificities for the structure of neighboring sugars of the sulfate-accepting sugar units. No major variations in the sulfotransferase activities on added receptors were found to occur in sera from individuals 22-41 years old. In contrast, the activities in sera of various mammalian and avian showed a species-specific

With mouse skin fibroblasts cultured in serum-free medium, preferential secretion of several sulfotransferases could be demonstrated. The results, taken together, suggest that the appearance of the sulfotransferases in serum is not a fortuitous event due to nonspecific cell death, but the result of an elaborate mechanism for enzyme secretion by a cell or tissue system.
Enzymes involved in the sulfation of growing glycosaminoglycan chains of proteoglycans are localized in the Golgi apparatus (1)(2)(3)(4). Microsomal preparations and extracts derived therefrom have been used for investigation of the sulfation process (for reviews, see Refs. [5][6][7]. When these preparations are incubated with [35S]PAPS,1 rapid incorporation of [35S]sulfate into endogenous proteoglycans or exogenous glycosaminoglycans is observed. The endogenous sulfate acceptors may represent incomplete proteoglycan molecules in all states of glycosaminoglycan chain extension which could serve as primers for both sulfation and polysaccharide polymerization (8). However, it is not possible with present meth-odology to isolate and characterize the multitude of molecular species which make up the intracellular acceptor pool. Thus, much of the work on the properties of the enzymes (glycosaminoglycan sulfotransferases) responsible for the sulfation process has been carried out with exogenous acceptors of defined structure, some of which are listed in Table I. Mostly, they consist of nonsulfated and partially sulfated glycosaminoglycans and their oligosaccharide derivatives. When a mainly sulfated molecule is an acceptor, the incorporation of sulfate could be onto either occasional nonsulfated monosaccharide units or free hydroxyl groups of sulfated monosaccharide units.
Besides these particulate or solubilized forms, naturally soluble forms of glycosaminoglycan sulfotransferases have been demonstrated in human serum (18-21), bovine fetus serum (22), and chick embryo serum (23). Despite the attention that has been given to apparent changes of the serum sulfotransferase activity during embryonic development (22) and in some connective tissue diseases (19), the serum enzymes have not been investigated in detail. How many different sulfotransferases are present in serum? Are there any differences between the properties of the serum form and tissue (or Golgi apparatus) form of sulfotransferases? Are the serum enzymes secretion products or the results of cell death and degeneration? Information about these points should throw more light on questions regarding the tissue of origin and the function, if any, of the serum sulfotransferases.
This paper describes the partial separation and characterization of four distinct sulfotransferases from human serum and the distribution of the sulfotransferase activities in various mammalian and avian sera. In the following paper (24), we will compare the properties of PAPS:GalNAc 4-sulfate 6-0-sulfotransferase preparations isolated from human serum and squid cartilage.

EXPERIMENTAL PROCEDURES
Materials-Adult human sera were prepared from fresh blood donated by volunteers, and other sera were from fresh blood specimens collected from ACI rats, Japanese rabbits, White Leghorn roosters, Japanese quails, and BALB/c mice.
[%]PAPS was synthesized from inorganic [%3]sulfate and ATP with ATP sulfurylase and adenylyl sulfate kinase preparations purified from Bakers' yeast (25). The specific activity was 3.0 X lo9 cpm/ pmol. A chondroitin sulfate-free keratan sulfate sample, prepared from fresh bovine cornea by proteolytic degradation with Pronase-P, ethanol fractionation, and digestion with chondroitinase ABC (26), was kindly donated by Dr. K. Nakazawa, Meijo University, Nagoya, Japan. Since the sample contained a significant amount of heparan sulfate, it was further treated with 0.24 M NaNOz in 1.8 M acetic acid for 80 min (27) to degrade the contaminant. The resulting mixture was dialyzed against running tap water and then mixed with 2 volumes of 95% (v/v) ethanol containing 1.3% (w/v) potassium acetate. The precipitate was washed successively with ethanol and ether and dried over Pz06 in a vacuum. The final preparation had the molar ratio of sulfate to hexosamine = 1. 13 (12,13) charides (17) The enzyme names are given to show the position of hydroxyl group (e.g. 643) or amino group (e.g. 2-N-) of the monosaccharide unit (e.g. GalNAc) in acceptor molecule to which the sulfate group is transferred.
The sulfate transfer is onto position 6 of the nonreducing GalNAc 4-sulfate end group. E The 2-0-sulfation of L-iduronic acid units appears to occur only in conjunction with the formation of such units by 5-epimerase reaction (14).
sulfate to hexosamine, = 0.99; this preparation contains non-acetylated glucosamine units, see "Results") were obtained from Seikagaku Kogyo Co., Tokyo, Japan. Chemically N-acetylated heparan sulfate was prepared from the heparan sulfate by the method of Foley and Baker (28). Chondroitin (molar ratio of sulfate to hexosamine e0.04) was prepared from squid skin by the method of Anno et al. (29), as modified by Habuchi and Miyata (30). UDP-GalNAc 4-sulfate was prepared from hen oviduct by the method of Nakanishi et al. (31).
Paper Chromatography and Paper Electrophoresis-Paper chromatography was done on 60-cm strips of Toyo No. 50 filter paper at room temperature by the descending method with n-butyric acid/0.5 M ammonia (53, by volume) as a solvent system. Paper electrophoresis was done on 60-cm strips of Toyo No. 51A filter paper at room temperature in the apparatus described by Markham and Smith (37) at a potential gradient of 25 volts/cm for 70 min. The buffer used was 0.05 M ammonium acetate/acetic acid, pH 5.0.
Enzyme Assays-The incubation mixture for the assay of serum glycosaminoglycan sulfotransferases contained 2 X lo5 cpm of ["SI PAPS, 0.1 pmol of NaF, 0.1 pmol of 5'-AMP, 100 pg of bovine serum albumin, enzyme, and acceptor, per 70 pl of 50 mM buffer. The buffers used were sodium acetate/acetic acid, pH 5.2 (when UDP-GalNAc 4sulfate was the acceptor), sodium acetate/acetic acid, pH 4.8 (when chondroitin was the acceptor), Tris-HCI, pH 8.0 (when heparan sulfate was the acceptor), Tris-HC1, pH 7.2 (when keratan sulfate was the acceptor), and Mes/NaOH, pH 6.9 (when chondroitin 4/6sulfate was the acceptor). The amounts of added acceptors were 50 nmol (UDP-GalNAc 4-sulfate), 35 pg (chondroitin and chondroitin 4/6-sulfate), and 20 pg (heparan sulfate and keratan sulfate). The NaF and 5'-AMP in the incubation mixture are included to protect the 35S-labeled nucleotide from enzymatic degradation (10,38). The bovine serum albumin in the incubation mixture is included as an enzyme stabilizer, and was omitted when unfractionated serum was used as an enzyme. When heparan sulfate was used as an acceptor, 7 pmol of CaC12 was added to the reaction mixture (for the effect of Ca2+, see "Results"). After 1 h at 37 "C, the reaction was stopped by immersing the reaction tubes in a boiling water bath for 1 min. To the solutions were added equal volumes of 0.1 M Tris-HCI, pH 8.0, and 85 pg of Pronase-P, and the mixtures were incubated at 37 "C for 1 h to degrade denatured proteins. Aliquots of the digests were then chromatographed on paper. Incorporation of [35S]sulfate into glycosaminoglycans was measured directly at the origin, as described by Suzuki and Strominger (38). Blank values obtained in the absence of exogenous acceptor were substracted. When UDP-GalNAc 4-sulfate was the acceptor, UDP-GalNAc 4,6-bissulfate was added to the Pronase digest as a reference compound. After paper chromatography, the area corresponding to the reference compound was located under ultraviolet light, cut out, and counted, as previously described (10).
Activities of glycosaminoglycan sulfotransferases in the cell layers and media of mouse dermal cell cultures were determined as above except that the reaction mixtures for heparan sulfate sulfotransferase contained 0.08% (w/w) Triton X-100.
One unit of enzyme is defined as the quantity that catalyzes the incorporation of 1 X IO' cpm (3 pmol) of sulfate/h. Analysis of p5S]Sulfate-labeled Glycosaminoglycans-Incubation mixtures for the preparation of [35S]sulfate-labeled glycosaminoglycans were the same as those used for enzyme assay except for the presence of 1 X IO6 cpm of [%]PAPS. After incubation at 37 "C for 2 h, the reaction mixtures were heated at 100 "C and then digested at 37 "C with Pronase-P as described above. After cooling in an ice bath, the digests were mixed with 3 volumes of 95% (v/v) ethanol containing 1.3% (w/v) potassium acetate to precipitate glycosaminoglycans. The precipitates were collected by centrifugation at 12,000 X g for 30 min at 2 "C and dissolved in 70 pl of water. The precipitation from water with ethanol was repeated 7 times to ensure complete elimination of low molecular weight 35S-labeled materials. When unfractionated serum was used as an enzyme, glycosaminoglycans in the Pronase digest were precipitated with acetone and washed with ether (to eliminate lipophilic substances) prior to the precipitation with ethanol.
The types of [35S]sulfate enzymatically introduced into chondroitin and chondroitin sulfate were determined by the method of Nakanishi et al. (10). Briefly, a %-labeled sample was digested with chondroitinase ABC, and the resultant degradation products were resolved by paper electrophoresis followed by paper chromatography using unlabeled ADi-4S, ADi-GS, ADi-4,6-bisS, GalNAc 4-sulfate, GalNAc 6sulfate, and GalNAc 4,6-bissulfate as reference compounds. The zone corresponding to each degradation compound was cut out and its radioactivity was determined in a liquid scintillation spectrometer. The radioactivities recovered as the monosaccharides represent the sulfate groups incorporated into the corresponding hexosamine units at the nonreducing termini. The 'radioactivities recovered as the unsaturated disaccharides, on the other hand, represent the sulfate groups incorporated into the interior repeat units.
Degradation of 35S-labeled glycosaminoglycans with HNOz was performed by reaction A (treatment with 0.24 M NaNOz in 1.8 M acetic acid for 80 min at room temperature) as described by Lindahl et al. (27). After treatment, the reaction mixture was chromatographed on a Sephadex G-50 column equilibrated and eluted with 0.2 M ammonium acetate/acetic acid, pH 6.0, to separate the degradation product (35SO!-) from undegraded polysaccharides. The radioactivity recovered as 25SO:represents the sulfate groups incorporated into the N-sulfate residues of heparan sulfate.
Digestion of both unlabeled and 35S-labeled glycosaminoglycans with keratanase (90 milliunits/mg keratan sulfate) was performed at 37 "C for 20 h, as described by Nakazawa and Suzuki (33). The digests were chromatographed on Sephadex G-50 columns equilibrated and eluted with 0.2 M ammonium acetate/acetic acid, pH 6.0, to separate the degradation products (keratan sulfate oligosaccharides) from undegraded polysaccharides. The radioactivity recovered as a disaccha-ride monosulfate, GlcNAc(6-sulfate)-Gal, represents the sulfate groups incorporated into the glucosamine units of keratan sulfate. The radioactivities recovered as higher oligosaccharides, on the other hand, should represent the sulfate groups incorporated into either the galactose units or the glucosamine units adjacent to a galactose unit already bearing a sulfate (33).
Culture of Mouse Dermal Fibroblasts-Mouse dermal fibroblasts were obtained from newborn BALB/c mouse skin by treatment with trypsin and collagenase as previously described (39). The cells were inoculated into 60-mm Falcon tissue culture dishes at a density of 6 X IO5 cells/3 ml of medium (Dulbecco's modified Eagle's medium supplemented with glucose (2 g/liter), ascorbic acid (50 mg/liter), Hepes (10 mM), fetal calf serum (lo%), penicillin (50,000 unitsfiiter), and streptomycin (50 mg/liter)). The cells were cultured for 2.5 days in a 37 "C incubator with a humidified atmosphere of 95% air/5% COS (culture medium was changed about 24 h after the inoculation). The cells were washed twice with Dulbecco's modified Eagle's medium and fed with either the serum-containing medium as above or a serum-free medium (Dulbecco's modified Eagle's medium supplemented with glucose (2 g/liter), ascorbic acid (50 mg/liter), Hepes (10 mM), epidermal growth factor (50 pg/liter), fibroblast growth factor (10 pglliter), insulin (2 mg/liter), transferrin (5 mg/liter), penicillin (50,000 units/liter), and streptomycin (50 mg/liter)). The cells were grown at 37 "C in a humidified atmosphere containing 5% COS. After approximately 1 day, culture medium was changed (the cultures at this stage are referred to as "time 0 cultures"). At the time indicated (12 and 24 h) one set of two dishes, each with the cells and either the serum-containing medium (control) or the serum-free medium, was removed and the number of cells in each dish was counted. To examine the rate of enzyme secretion, the spent medium was collected from the serum-free cultures, dialyzed against 0.02 M Tris-HC1, pH 7.2, containing 10% (w/v) glycerol and 0.5 mM dithiothreitol (solution A), concentrated to 0.1 volume by centrifugation on a Centricon 10 membrane, and assayed for glycosaminoglycan sulfotransferases (see above) and lactate dehydrogenase (for the assay method see Ref. 40). The cells, on the other hand, were washed twice with fresh medium, suspended in solution A (3 X lo6 cells/ml), and treated at 0 "C in a 10-kc sonic oscillator for 1 min. The enzyme activities in the sonicates were assayed as above.
Other Methods-Protein in the eluates from chromatography columns was monitored by the Coomassie Blue method (42). In all other cases, protein was determined by the method of Lowry et al. (41) with bovine serum albumin (Fraction V) as a standard. Hexose was determined by the anthrone-H2SO4 method (43) with D-galactose as a standard.

RESULTS
Sulfate Transfer Catalyzed by Human Serum-Fresh human serum was assayed at pH 7.2 for transfer of 35S from [35S]PAPS to endogenous acceptors. Paper chromatography (in solvent A) of the resultant reaction mixture indicated that a small amount of 35S was incorporated into material remaining at the origin on the paper chromatogram; the rate of incorporation was about 220 cpm/h/lO pl of serum, in the presence of 2 X lo5 cpm of [35S]PAPS (Table 11). The labeled products could be recovered directly from the reaction mixture on a preparative scale by digestion with Pronase followed by ethanol precipitation. Analysis with chondroitinase ABC, chondroitinase AC 11, nitrous acid, and keratanase indicated that about 65 and 5% of the incorporated radioactivity were accounted for by chondroitinase ABC (or AC 11)-sensitive materials (mainly chondroitin 6-sulfate) and nitrous acidsensitive materials (such as heparan sulfate and heparin), respectively. The remainder (about 30%) of the incorporated radioactivity was refractory to the action of chondroitinase ABC, keratanase, and HN02 (as ascertained by gel-chromatographic criteria), and presumably represents nonglycosaminoglycan-type sulfated polysaccharides. This resistant material was not studied further.
As Table I1 shows, the formation of radioactive material remaining at the origin was greatly stimulated by the addition of chondroitin, chondroitin 416-sulfate (80% 4-sulfate and 20% 6-sulfate), heparan sulfate, and keratan sulfate. Maximal stimulation was obtained at pH 4.8 with chondroitin (6-fold), at pH 6.9 with chondroitin sulfate (5-fold), at pH 8.0 with heparan sulfate (220-fold), and at pH 7.2 with keratan sulfate The radioactive products obtained by incubation with chondroitin and chondroitin 416-sulfate were recovered from the reaction mixtures by digestion with Pronase followed by ethanol precipitation. Analysis with chondroitinase ABC (Table 11) indicated that the serum catalyzed the formation of various types of sulfate ester bonds on the exogenous acceptors, e.g. 4-su1fatey 6-sulfate, and 4,6-bissulfate on the interior galactosamine units and 4,6-bissulfate on the nonreducing terminal galactosamine units. It is noteworthy that the type of the sulfate incorporated is largely dependent.on the chemical nature of added acceptor. Thus, when chondroitin was the acceptor, about 98.5% of the sulfate groups incorporated were onto the interior N-acetylgalactosamine units (14% 4sulfate, 82% 6-sulfate, and 2.5% 4,6-bissulfate). In contrast, when chondroitin 416-sulfate was used as the acceptor, about 30% of the sulfate groups were onto the nonreducing terminal N-acetylgalactosamine units (28% 4,6-bissulfate and 2% 4-or (2.5-fold).   Table   I). Since a small, but significant, amount of [35S]sulfate was found as N-acetylgalactosamine 4,6-bissulfate in the interior portion of chondroitin 4/6-sulfateY a 6-0-sulfotransferase for interior GalNAc 4-sulfate units might also be present in the serum. The radioactive products obtained by incubation with heparan sulfate and keratan sulfate were similarly recovered from the reaction mixtures and treated with nitrous acid and keratanase, respectively. Sephadex G-50 chromatography of the treated samples showed that over 98% of the labeled products from heparan sulfate and about 76% of those from keratan sulfate could be degraded with nitrous acid and keratanase, respectively, to yield fragments appearing in retarded fractions from the gel. The results suggest that human serum contains some heparan sulfate (heparin) and keratan sulfate sulfotransferase similar to, if not identical with, those listed in Table I. When fresh serum was kept frozen at -20 "C with occasional-.thawing and refreezing, no significant loss of the sulfotransferase activities was observed over a period of 1 year, except for the activity of GalNAc 4-0-sulfotransferase for chondroitin, which decreased gradually on repeated freezing and thawing. Comparison of the sulfotransferase activities of a fresh serum sample with those of a fresh plasma sample obtained from the same source showed no significant difference, indicating that platelets could not be a source of the sulfotransferases.
Fractionation of Serum Sulfotransferases-On the basis of the observations described above, advantage was taken in the following fractionation experiments of the fact that most, if not all, of the individual sulfotransferase activities can be monitored with UDP-GalNAc 4-sulfate, chondroitin, heparan sulfate (molar ratio of sulfate to hexosamine = 0.99), and keratan sulfate (molar ratio of sulfate to hexosamine = 1.13) as acceptors. Although chondroitin 4-sulfate may also be used as an acceptor for terminal GalNAc 4-sulfate 6-0-sulfotransferase, the use of UDP-GalNAc 4-sulfate has the advantage of simplicity in product assay (10) and is less subject to interference from other sulfotransferases which might be present; e.g. UDP-GalNAc 4-sulfate does not serve as an acceptor for GalNAc 6-O-sulfotransferase, the enzyme introducing 6-sulfate to occasional N-acetylgalactosamine units in the interior portion of chondroitin sulfate (10). Table I11 shows the summary of enzyme purification.
Fresh serum was subjected to affinity chromatography on heparin-Sepharose CL-GB (Fig. 1). Automatically dialyzed fractions from the column were assayed for sulfotransferase activities toward chondroitin, UDP-GalNAc 4-sulfate, heparan sulfate, and keratan sulfate. As Fig. 1 shows, the affinity chromatography served as an effective method in separating sulfotransferase (as a whole) from other serum proteins; over 95% of the applied proteins were recovered in the washings (panel a), whereas essentially all of the sulfotransferase activities were found in the retentates.
The heparan sulfate-dependent activity was eluted in a single large peak with a maximum at tube 46 (panel c), whereas all the other activities eluted in peaks between tubes  50 and 70 (panels a, b, and d ) . The fractions containing most of the heparan sulfate-dependent activity (tubes 43-51) were pooled, concentrated on a PM-10 membrane by pressure ultrafiltration, and used for further studies as "Fraction I".
The major activities on UDP-GalNAc 4-sulfate (panel a) and chondroitin (panel b) were eluted in tubes 54-64 (indicated as Fraction 11). At this stage, their elution profiles overlapped each other. However, the recovered activity for chondroitin, but not the activity for UDP-GalNAc 4-sulfate, exceeded the applied one by a factor of 4.8 (Table 111), suggesting that the two activities reside in different enzyme species and that the serum contains an inhibitor of the chondroitin-dependent sulfotransferase which can be removed by the affinity chromatography.
The keratan sulfate-dependent activity (panel d ) was eluted in two peaks,' the first one appearing in the zone for UDP-GalNAc 4-sulfate/chondroitin-dependent activities (Fraction 11) and the second one in a slower eluting zone with some overlap with the first peak. As shown below, attempts to obtain a keratan sulfate sulfotransferase preparation devoid of chondroitin/chondroitin 4-sulfate sulfotransferase activities have so far been unsuccessful. In the present work, therefore, aliquots of the two peaks (tubes 54-64 and tubes 65-70) were separately pooled, dialyzed against solution A, concentrated on a PM-10 membrane by pressure ultrafiltration, and used as Fraction I1 and Fraction 111, respectively.
In an attempt to separate the activities of chondroitin, UDP-GalNAc 4-sulfate, and keratan sulfate from one another, Fraction I1 from the heparin-Sepharose CL-GB column was pooled, concentrated by pressure ultrafiltration, and loaded on a MBtrex Blue B column equilibrated with solution A. The column was washed with solution A and then eluted with an increasing salt gradient from 0 to 0.6 M NaCl in solution A (Fig. 2). The chondroitin-dependent activity was recovered almost exclusively in the washings (panel b). The UDP-GalNAc 4-sulfate-dependent activity, in contrast, was eluted in a large peak (containing 90% of the total activity) with a maximum at tube 40 plus a small peak in the washings (panel a). Very little keratan sulfate-dependent activity was recovered from the column, perhaps owing to denaturation of the enzymes during the manipulation for column chromatography. The fractions (tubes 3-12) containing most of the chondroitin-dependent activity and those (tubes 34-49) containing the main activity for UDP-GalNAc 4-sulfate were separately pooled, dialyzed against solution A, concentrated by pressure ultrafiltration, and used for further studies in "Fraction 11-a" and "Fraction 11-b", respectively.
Identification of the Products of Sulfotransferase Reactions-The following facts established that the %-labeled product formed by incubation of UDP-GalNAc 4-sulfate with [35S]PAPS in the presence of Fraction 11-b (see Fig. 2   To determine the position of the [35S]sulfate groups incorporated into added chondroitin by Fraction 11-a (see Fig. 21, the labeled product was recovered from the incubation mixture by Pronase digestion followed by ethanol precipitation. Digestion of this material with chondroitinase ABC resulted in the release of about 94% of the radiolabel as a compound with the paper chromatographic and paper electrophoretic mobilities of ADi-6s. Thus, the activity in Fraction 11-a may actually represent GalNAc 6-O-sulfotransferase, 'the known enzyme capable of introducing 6-sulfate into the interior Nacetylgalactosamine units of chondroitin. The activity of introducing 4-sulfate to the interior N-acetylgalactosamine units, which is detectable in fresh serum (Table I), was not found in Fraction 11-a. In view of its low stability (see above), it is possible that the enzyme has been denatured during the manipulation for purification.
The position of the [35S]sulfate groups incorporated into heparan sulfate by Fraction I (see Fig. 1) was determined by deaminative degradation of the 35S-labeled product that was recovered from the incubation mixture by Pronase digestion followed by ethanol precipitation. Before nitrous acid treatment, the labeled product was almost entirely excluded from a Sephadex G-25 column equilibrated and eluted with 0.2 M ammonium acetate/acetic acid, pH 6.0 (Fig. 3). Treatment of the labeled product with nitrous acid released almost all of the label as a low molecular weight compound being eluted in the position expected for inorganic sulfate (indicated by the arrow in Fig. 3). When the product of HNOz degradation was further analyzed by paper electrophoresis, a single radioactive spot was detected by autoradiography (not shown). This radioactive spot coincided with inorganic sulfate (added as internal marker) located with the BaC12-rhodizonate reagent (45). The results indicate that the sulfate residues transferred to added heparan sulfate were specifically incorporated into 2-N-sulfate groups. Thus, the activity of Fraction I on added heparan sulfate may actually represent GlcN 2-N-sulfotransferase. This conclusion was further supported by the fact that chemically N-acetylated heparan sulfate was essentially inactive as an exogenous acceptor for Fraction I; the amount of [35S]sulfate incorporated was less than 9% of that incorporated into the parent heparan sulfate sample.
To investigate whether Fractions 11 and I11 (see Fig. 1) catalyze the transfer of sulfate into either galactose or Nacetylglucosamine unit, or both, of added keratan sulfate, the labeled products isolated from the reaction mixtures by Pronase digestion followed by ethanol precipitation were mixed with unlabeled keratan sulfate (carrier) and the mixtures were exhaustively digested with keratanase. The resulting digests were analyzed by Sephadex G-50 chromatography (Fig. 4). While all the undigested glycosaminoglycan samples were eluted in the void volume fraction, their elution profiles were altered in a pronounced manner after keratanase digestion. Thus, when measurements were made of hexose-positive material (which represents the bulk of the oligosaccharides derived from unlabeled keratan sulfate added as a carrier), about 40% of the total hexose was recovered after keratanase digestion in a peak with the Kd value of disaccharide monosulfate, GlcNAc(6-sulfate)-Gal (panel a; see Ref. 33 for the identification of the disaccharide). When measurements were made of 35S radioactivity (which represents the oligosaccharides derived from newly sulfated keratan sulfate in uitro), it was revealed that the introduction of [35S]sulfate by Fraction I1 were isolated from the reaction mixtures by Pronase digestion followed by ethanol precipitation. Each of the labeled samples was mixed with 1 mg of unlabeled keratan sulfate (carrier) and applied to a Sephadex G-50 column (0.8 X 40 cm) before (X) and after (0) keratanase digestion. The column was equilibrated with 0.2 M ammonium acetate/acetic acid, pH 6, and elution was carried out with this buffer at a flow rate of 10 ml/h.  (panel b) and Fraction I11 (panel c) caused marked alterations in the molecular size distribution of the keratanase-derived oligosaccharide fragments. In both cases, the peaks of labeled oligosaccharide fragments were earlier than those of hexosepositive oligosaccharides. It is especially noteworthy that in no case was the label found in the zone for sulfated disaccharide. Since keratanase has been shown to hydrolyze keratan sulfate at the endogalactoside bonds of nonsulfated galactose units but not at a galactoside bond in which galactose 6sulfate participates (33), the failure of the labeled product to yield 35S-labeled GlcNAc(6-sulfate)-Gal can be interpreted as indicating 1) that the newly introduced [35S]sulfate residues were on galactose units which rendered their galactoside bonds resistant to keratanase and/or 2) that the [35S]sulfate residues were on glucosamine residues adjacent to a galactose unit already bearing a sulfate residue. As Fig. 4, b and c indicate, there is a difference between the products of Fraction I1 and I11 in the size distribution of keratanase-derived [35S] oligosaccharides. This difference may be interpreted as reflecting the fact that Fraction I1 differs from Fraction I11 in transferring sulfate more preferentially into a region of keratan sulfate where keratanase-resistant galactoside bonds (i.e. galactose 6-sulfate-participating bonds) are present in close proximity to the sulfate-accepting (nonsulfated) sugar units.
Properties of Serum Sulfotransferases-The maximum rates of sulfate incorporation into heparan sulfate (with Fraction I), keratan sulfate (with Fraction 111), chondroitin (with Fraction 11-a), and UDP-GalNAc 4-sulfate (with Fraction IIb) occurred at pH 8.0 (in 0.05 M Tris-HC1 buffer), at pH 7.2 (in 0.05 M Tris-HC1 buffer), at pH 4.8 (in 0.05 M sodium acetate/acetic acid buffer), and at pH 5.2 (in 0.05 M sodium acetate/acetic acid buffer), respectively. It should be noted that the activity toward chondroitin (optimal pH = 4.8) exhibited about 50% of the maximal activity even at pH 7.0 (the pH of the serum) and that replacement of UDP-GalNAc 4-sulfate (acceptor) by chondroitin 4/6-sulfate resulted in the shift of optimal pH of the fourth activity from pH 5.2 to 6.9 (see above and Ref. 24).
As Fig. 5 shows, Ca2+ and Mg2+ stimulated the GlcN 2-Nsulfotransferase activity of Fraction I; at the optimal concentration of Ca2+ ion (10 mM), the activity increased about 16fold. Keratan sulfate sulfotransferase (Fraction 111) and GalNAc 6-0-sulfotransferase (Fraction 11-a) were less sensitive to these cations. From the fact that MnC12, NaC1, and KC1 did not necessarily produce comparable effects on these reactions, it can be surmized that the chloride ion of itself does not account for the observed activations with calcium and magnesium salts (see Ref. 24 for the effect of cations on the terminal GalNAc 4-sulfate 6-0-sulfotransferase activity of Fraction 11-b).
Sulfotransferase Activities in Various Serum Specimns-Ten serum specimens from healthy individuals (22-year-old men of blood type AB and blood type 0, 22-year-old women of blood type A and blood type 0, a 26-year-old man of blood type B, a 28-year-old man of blood type AB, and a 41-yearold man of blood type B) were analyzed for the sulfotransferase activities as described above. None of the enzyme activities could be demonstrated to correlate with age and blood type; variations of the specimens were within 20% of the mean values shown in Table IV. Serum specimens from seven patients of spondyloepiphyseal dysplasia3 were similarly analyzed. No significant deficiency of the sulfotransferase activities was detected in these sera.
As Table IV shows, the sulfotransferase activities in sera The serum specimens were provided by Dr. H. Iwata, Department of Orthopedic Surgery, Nagoya University School of Medicine.  of various mammalian and .avian species showed a speciesspecific variation. Apparently, the animal and avian sera tested are higher in all the enzyme activities than human sera. It remains to be explored whether the species-specific variation may reflect a quantitative variation of the same sulfotransferase as those in human serum or an existence of distinct sulfotransferases. Also to be noted is a possibility that the apparent sulfotransferase activities may reflect, at least in part, the occurrence in sera of either some sulfotransferase inhibitors (see above) or glycosaminoglycan sulfatases4 which may interfere with the enzyme assays.
Secretion of Glycosaminoglycan Sulfotransferases by Skin Fibroblusts in Culture-Available evidence indicate that glycosaminoglycan sulfotransferases are localized in the Golgi apparatus where they are involved in sulfation of growing glycosaminoglycan chains (1)(2)(3)(4). One may assume, therefore, that a source of the serum sulfotransferases is the enzymes in the Golgi apparatus. If this is so, such sulfotransferases may be synthesized in the rough endoplasmic reticulum and transported, like other secretory proteins, via the Golgi apparatus to the extracellular fluid. Thus, it was of interest to examine, as a first approach to this hypothesis, whether glycosaminoglycan sulfotransferases are selectively released into the medium when proteoglycan-synthesizing cells are cultured in uitro. For this purpose, we chose mouse dermal cells, since fibroblast-like cells from avian and mammalian skin have been shown to synthesize at least three proteoglycan species bearing chondroitin sulfate, dermatan sulfate, and heparan sulfate, respectively (46)(47)(48)(49), and because our preliminary tests have indicated that fibroblast-like cells from newborn mouse dermis may be cultured in serum-free media with no apparent cell death or degeneration.
Fibroblast-like cells, prepared from the dermis of newborn mouse by collagenase treatment, were cultured for 2.5 days in a standard medium containing 10% fetal calf serum. In one set of experiments, the cells adhered to the plate were washed once with Dulbecco's modified Eagle's medium and then fed with a serum-free medium (Dulbecco's modified Eagle's medium supplemented with epidermal growth factor, fibroblast growth factor, insulin, and transferrin; see Ref. 50). In the other set of experiments (control), the washed cells were fed with the serum-containing medium. After approximately 1 day, medium was changed (the cultures at this stage are referred to as time 0 cultures). In both serum-free and control cultures, the number of surviving cells increased from an initial value of about 6 X lo5 cells/plate on Day 0 to a value of about 3 X lo6 cells/plate on Day 4. From Day 4 to Day 4.5, the number of surviving cells per plate was nearly constant in the control culture, whereas that in the serum-free culture was increased to a value of 3.8 x lo6. Phase contrast microscopic observation of Day 1-Day 4 cultures showed no obvious morphological differences between the fibroblast-like cells grown in serum-free and serum-containing media.
The cells of time 0 culture were maintained in serum-free medium for 12 or 24 h and the spent medium and cell layers were separately assayed for glycosaminoglycan sulfotransferases, with UDP-GalNAc 4-sulfate, chondroitin, heparan sulfate, and keratan sulfate as acceptors. For comparison, the activity of lactate dehydrogenase, a marker enzyme for cytosol (51), was also measured. Fig. 6, a and b shows that the sulfotransferase activities for UDP-GalNAc 4-sulfate and chondroitin in the medium increased linearly at about 100 and 5 units/plate/l2 h, respectively. The corresponding activities in the cell layer also increased with incubation time but at far lower rates, so that after 24 h the activities for UDP-GalNAc 4-sulfate and chondroitin in the medium were about 12 and 3 times, respectively, the corresponding activities in the cell layer. The results suggest that terminal GalNAc 4-sulfate 6-0-sulfotransferase ' Under the standard assay conditions, no sulfatase activity was detected in human serum with p-nitrocatechol sulfate and UDP-GalNA~-4,6-bis-[6-~S]sulfate as substrates (H. Inoue and GalNAc 6-0-sulfotransferase are actively synthesized and secreted by the fibroblast-like cells in culture. Fig. 6c shows that the sulfotransferase for heparan sulfate also appeared in the medium at about 50 units/plate/l2 h. In this case, however, its activities were higher in the cell layer than in the medium throughout the incubation period. The sulfotransferase for heparan sulfate might keep its positions in the intracellular or cell surface membranes, with partial release into the medium. In contrast to the above three enzymes, lactate dehydrogenase, a cytosol enzyme, was at a high level in the cell layer even on time 0 and the cell-associated activity decreased very slowly with a compensatory increase of the activity in the medium (Fig. 6e). As Fig. 6d shows, the rate of appearance of keratan sulfate sulfotransferase in the medium was almost the same as that of lactate dehydrogenase release. Furthermore, the activity of keratan sulfate sulfotransferase in the medium was only 10% or less of that in the cell layer even at the end of incubation. It appears, therefore, that keratan sulfate sulfotransferase is actively synthesized in the cell but, unlike the other three sulfotransferases, remains unreleased under the culture conditions.

DISCUSSION
Earlier studies showed that mammalian and avian sera were able to catalyze the transfer of sulfate from added PAPS to exogenous chondroitin (19, 22), chondroitin sulfate (18), dermatan sulfate (18), and keratan sulfate (23), as well as endogenous chondroitin sulfate proteoglycans (22). We now show that human serum contains four distinct sulfotransferases, each having a high degree of specificity for the structure of glycosaminoglycan acceptor molecule. These include 1) a sulfotransferase sulfating position 6 of the N-acetylgalactosamine units in the interior portion of chondroitin (designated GalNAc 6-O-sulfotransferase), 2) a sulfotransferase introducing a second sulfate to position 6 of the nonreducing terminal N-acetylgalactosamine 4-sulfate unit of chondroitin sulfate (designated terminal GalNAc 4-sulfate 6-O-sulfotransferase), 3) a sulfotransferase sulfating position 2 (amino group) of the glucosamine units in heparan sulfate (designated GlcN 2-Nsulfotransferase), and 4) a sulfotransferase introducing sulfate to the sugar units in keratan sulfate (designated keratan sulfate sulfotransferase). The second and third enzymes can be separated from each other as well as from the other enzymatic activities by sequential affinity chromatography on heparin-Sepharose CL-GB (Fig. 1) and Mgtrex Blue B (Fig.  2). The keratan sulfate sulfotransferase has been resolved into two peak fractions overlapped with those of the GalNAc 6-0-sulfotransferase and terminal GalNAc 4-sulfate 6-0-sulfotransferase ( Fig. 1). After Mstrex Blue B chromatography, however, little or no keratan sulfate sulfotransferase activity was detected in any fractions from the column, perhaps owing to denaturation during the manipulation. The results, taken together, indicate that the four enzymic activities represent different enzyme species.
The specificity of the serum GalNAc 6-0-sulfotransferase appears to be the same as that of the enzymes described in chick embryo cartilage (9) and quail oviduct (10). The pH and divalent cation optima and K, values are also similar for all three sources of GalNAc 6-0-sulfotransferase. Likewise, the specificities and properties of the other three serum sulfotransferases are similar to those found with the terminal GalNAc 4-sulfate 6-0-sulfotransferase from quail oviduct (lo), the GlcN 2-N-sulfotransferase from calf arterial wall (13) and mouse mast cell tumor (12), and the Gal 6-0sulfotransferase from bovine cornea (17), respectively. The similarity between the serum and tissue forms may reflect a relatively low species specificity of the enzymes and further suggests that the sulfotransferases found in serum are derived from such tissues as those synthesizing and secreting sulfated proteoglycans. However, the lack of some of the sulfotransferase found in tissues (e.g. heparan sulfate O-sulfotransferases, see Table I) suggests that the appearance of the four sulfotransferase species in human serum is not a fortuitous event due to nonspecific cell death and degeneration, but the result of an elaborate mechanism for enzyme secretion by a cell or tissue system. Studies in vivo through pulse-chase electron microscopy autoradiography with radioactive sulfate (3,4) and studies in vitro through subcellular fractionation (1,2) have shown that sulfation of nascent or fully elongated glycosaminoglychan chains is accomplished by membrane-bound sulfotransferases located in the Golgi apparatus. The secretory route from this area is considered to be discontinuous in that the proteoglycans must be packaged into vesicles to be conveyed to the next compartment, i.e. vesicles fusing with the plasma membrane (52). It may therefore be postulated that the membrane of the secretory granules bear on their inner surface some of the sulfotransferases. This suggests that when a vesicle bearing such enzymes fuses with the plasma membrane, the enzymes can be released, presumably in combination with sulfated proteoglycans, outside the cell. Our results obtained with mouse dermal cell cultures as a model system are consistent with this hypothesis and show that, whereas keratan sulfate transferase and lactate dehydrogenase (cytosol marker) remain unreleased, the sulfotransferases toward UDP-GalNAc 4-su1fate7 chondroitin, and heparan sulfate are released into medium at characteristic rates. The evidence presented here for the appearance of specific sulfotransferase species in sera and the results of other investigators indicating that there are glycosyltransferases appearing in sera (53) suggest the existence of a specific group of enzymes for glycoconjugate biosynthesis which are destined to be discharged into the blood stream. Further isolation and characterization of such enzymes and comparison between secretory and nonsecretory sulfotransferases should result in a better understanding of their topography and function. It is of note in this respect that all of the human serum sulfotransferases described here exhibit significant activities at the pH of the serum and that human serum contains chondroitin sulfate proteoglycans (54,55). This suggests a role in serum-dependent sulfation, but what this is remains unknown.