Subcellular Localization of Hyaluronate Synthetase in Oligodendroglioma Cells*

In order to provide some insight into the mechanism of hyaluronate synthesis, the subcellular localization of the synthetase system for hyaluronate was determined in eukaryotic cells. The mouse oligodendroglioma cell line G26-24, which produces copious amounts of hyaluronate in culture, was chosen as a system for these studies. Protease treatment and ho- mogenization of cells followed by hyaluronate synthetase assay suggested that nucleotide-binding sites and trypsin-sensitive synthetase sites were not exposed at the outer membrane surface. Protease treatment fol- lowing homogenization did result in decreased activity. a plasma

In order to provide some insight into the mechanism of hyaluronate synthesis, the subcellular localization of the synthetase system for hyaluronate was determined in eukaryotic cells. The mouse oligodendroglioma cell line G26-24, which produces copious amounts of hyaluronate in culture, was chosen as a system for these studies. Protease treatment and homogenization of cells followed by hyaluronate synthetase assay suggested that nucleotide-binding sites and trypsin-sensitive synthetase sites were not exposed at the outer membrane surface. Protease treatment following homogenization did result in decreased activity. Membrane fragments, prepared by gentle homogenization in iso-and hypotonic buffers, were subjected to differential centrifugation followed by several continuous and discontinuous sucrose equilibrium and velocity gradient systems. Hyaluronate synthetase activity co-fractionated with a plasma membrane marker in all systems, including those in which Golgi markers were separable. Treatment of intact cells in culture with several hyaluronidases resulted in a marked stimulation of cell-free synthetase activity. The stimulated activity was also found exclusively in plasma membrane-enriched fractions.
Fifty years after the isolation of hyaluronate (1) and 30 years after cell-free synthesis was demonstrated (2), the mechanism of biosynthesis of hyaluronate remains enigmatic. The importance of protein synthesis and the precise molecular events in de novo synthesis, elongation, and termination have not been elucidated (3). Even the very basic information concerning the cellular localization of hyaluronate glycosyltransferase activities is not known with certainty.
In the streptococcal system the hyaluronate synthetase has been localized to the protoplast membrane (4), but in avian or mammalian cells, reports are conflicting. Ishimoto et al. (5) found that the hyaluronate synthetase activity in avian sarcoma virus-infected chick fibroblasts was high in EDTAsuspended intact cells, but reduced following trypsin treatment, and suggested that the synthetase activity might be located at the cell surface. Barland  of [3H]GlcNAc taken up by cultured human synovial cells and rapid appearance of radioactivity into the Golgi apparatus, but the intracellular radioactivity was not rigorously characterized. Chase experiments suggested that the Golgi apparatus might also serve as a site of intracellular storage. However, Bader et al. (7) found a lag of about 15 min between incorporation of [3H]GlcNAc and its appearance in hyaluronate in Rous sarcoma-transformed chick fibroblasts, and essentially no further lag in the appearance of labeled hyaluronate in the medium after synthesis. Appel et al. (8) found that approximately 80% of the hyaluronate synthetase activity sedimented between 600 X g and 10,000 X g. The possibility of localization in large membranes, which presumably included the plasma membrane, was noted.
Knowledge of the subcellular site of synthesis would provide a new approach with which to study the mechanisms of chain growth and extrusion into the extracellular space, and possibly explain the several differences between the properties and synthesis of hyaluronate and all the other glycosaminoglycans. In order to provide these answers and as a prelude to purifying the synthetase, we began a systematic study of the membranes possessing hyaluronate synthetase activity in eukaryotic cells. As some transformed cell lines produce large amounts of hyaluronate, one such line, the G26-24 mouse oligodendroglioma, was chosen for these studies in order to facilitate assay of the synthetase.

5018
Hyaluronate Synthesis in Plasma Membrane 95% humidity at 37 "C. Medium was replenished at day 3 and as needed until days 5-6, when cells reached a confluent density of 2 to 2.5 X 10' cells/cm'. Cells were harvested for replating by gently rising with a wide-mouth 10-ml pipette or further processed by collection with a rubber policeman. For metabolic labeling of glycosaminoglycans, cell cultures were incubated with fresh media containing 10 pCi/ml of [3H]acetate or [3H]glucosamine for 18 h.
Isolation of Glycosaminoglycans-Labeled glycosaminoglycans were isolated from the medium by a modification of the method of Dorfman and Ho (13). An aliquot of the medium, diluted with 3 volumes of 0.03 M NaCl, was mixed with 0.1 volume of 5% cetylpyridinium chloride and incubated for 10 min at room temperature. The pellet, washed with 0.1% cetylpyridinium chloride in 0.03 M NaCl, was dissolved in a solution of 5 M NaCl, methanol, and water (1:l:l) and reprecipitated by the addition of 6 volumes of ice-cold ethanol. After 10 min on ice, the pellet was collected by centrifugation, washed with ethanol, and dissolved in 0.1 M sodium acetate, pH 5. Aliquots were then incubated with and without Streptomyces hyaluronidase (5 TRU/ml, 60 "C for 18 h) followed by Sephadex G-50 chromatography to quantitate [3H]hyaluronate.
Enzyme Treatments-Cells in monolayer culture were rinsed three times with the indicated equilibrated buffer. Cells were treated with 3 ml of 0.125% sterile trypsin in calcium-and magnesium-free HBSS' buffered with 20 mM Hepes for 30 min at ambient temperature. The reaction was stopped by the addition of 1 volume of 10% fetal calf serum in media. The cells were washed again with 10% serum, then twice in buffer before harvest. Cell homogenates or isolated membranes prepared as described below were adjusted to a protein concentration of 10 mg/ml and incubated with or without trypsin at a ratio of cell protein to trypsin of 1O:l (30 min at 30 "C) by a modification of the method of Carey and Hirschberg (14). For the homogenates, the reaction was stopped by the addition of a &fold excess of the trypsin inhibitor ovomucoid. For the isolated membranes, the reaction was terminated with a 5-fold excess of ovomucoid in 10 volumes of ice-cold buffer. The membranes were collected by centrifugation at 100,000 x g for 1 h and resuspended to the original volume.
Cells were treated with testicular, Streptomyces, or leech hyaluronidase, as indicated, at the various concentrations in HBSS for 15 min at 37 "C in the presence of 10% COP. Cell layers were then washed three times with ice-cold HBSS before harvest. Control cells were treated identically in the absence of enzyme.
Cell Fractionation-Suspended cells were rinsed by hand inversion in calcium-and magnesium-free HBSS, pelleted by centrifugation at 300 X g for 10 min, cooled on ice to 4 "C, then pelleted and washed twice with 0.25 M sucrose containing 10 mM Hepes and 0.5 mM dithiothreitol, pH 7.2 (sucrose buffer).
A total membrane fraction ("crude membranes") was prepared using a Branson Sonifier (Danbury, CT) with a microtip probe at maximum output. Cells were cooled in an ice bath during three cycles of disruption for 5 s each. Large debris in the suspension was removed by Centrifugation at 600 X g for 10 min (SA-600 rotor, Dupont Instruments). The supernatant was collected and subjected to centrifugation at 105,000 x g for 1 h in the Type 65 rotor (Beckman Instruments). The pellet was resuspended in sucrose buffer and collected by centrifugation at 150,000 X g for 45 min.
In the following procedures, washed cells were disrupted by homogenization with a glass Dounce homogenizer with the tight pestle. The homogenization was monitored by phase contrast microscopy or trypan blue staining to determine the maximum point of disruption with minimum breakage of nuclei. Twenty to 30 strokes were optimal. Nuclei and large debris were removed by centrifugation at 600 X g for 10 min. The supernatant was collected and centrifuged first at 20,000 X g for 20 min and then at 100,000 X g for 1 h. The pellets were resuspended in sucrose buffer and aliquots were subjected to sonication as previously described prior to enzyme assays and protein determination.
The 20,000 X g pellet was further fractionated in a continuous sucrose density gradient. All sucrose solutions were prepared by weight in 10 mM Hepes and 0.5 mM dithiothreitol, and adjusted to a final pH of 7.2. The actual concentration was determined by the refractive index. The resuspended pellet was brought to 50% (w/w) sucrose by the addition of 70% sucrose, layered over a cushion of 60% sucrose, and equal portions of 45,40,35,30, 25,20, and 15% sucrose were sequentially overlaid. The gradient was centrifuged at 100,000 X g for 12 h in a SW 41 rotor (Beckman Instruments) and then 0.5ml fractions were collected from the bottom. For each fraction, protein and sucrose concentrations were determined, and enzymes assayed as described below.
The total postnuclear supernatant was fractionated on a discontinuous gradient system according to a modification of the procedure of Tabas and Kornfeld (15), used to isolate Golgi membranes from rat liver. The 600 X g supernatant was layered over 41% sucrose and centrifuged at 100,000 X g for 1 h. The band at the interface was recovered, brought to 34% sucrose, and layered over 36% sucrose. Sucrose solutions of 31 and 15% were layered on top, and the gradient was centrifuged for 2 h. The interfaces were collected for enzyme assays and protein determination.
In an alternate method, washed cells were resuspended under hypotonic conditions (10 mM Hepes, 0.5 mM dithiothreitol, pH 7.4) and allowed to swell for 1 h on ice. After this treatment, cells were homogenized with 5-6 strokes in the Dounce homogenizer. An equal volume of 20% sucrose was immediately added to stabilize the subcellular organelles. This approach has been discussed by DePierre and Karnovsky (16). The homogenate was centrifuged at 1000 X g for 10 min and the resulting supernatant was centrifuged at 20,000 X g for 20 min. The pellets were gently resuspended in sucrose buffer (0.25 M, as above). A discontinuous gradient was constructed with sequential layers of 50, 40, and 20% sucrose and the membrane suspension layered on top. The gradients were centrifuged for 2 h at 100,000 X g (SW 41 rotor). The interfaces were collected by aspiration, diluted with 0.25 M sucrose buffer, and collected by centrifugation at 100,000 X g for 1 h. The pellets were resuspended in sucrose buffer and assayed for enzyme activity and protein, or prepared for electron microscopy.
For treatment with trypsin, membranes were prepared by a modification of the method of Santala et al. (17). Briefly, washed cells were homogenized in 0.25 M sucrose buffer as above. The 20,000 X g (20 min) pellet, prepared from the postnuclear supernatant, was brought to 50% sucrose by the addition of 80% sucrose. Discontinuous gradients, constructed by overlaying 40, 20, and 10% sucrose solutions, were centrifuged as above. Membranes at the 20/40% interface were collected for trypsinization.
Enzyme Assays: Hyaluronate Synthetase-Hyaluronate synthetase was assayed according to the method of Appel et al. (8). In preliminary experiments, ATP strongly inhibited the reaction (50% inhibition of 2 mM ATP) and was omitted in these studies.' The standard assay contained the following components in a final volume of 0.1 ml: 5 mM dithiothreitol, 15 mM MgCl', 25 mM Hepes (pH 7.1), 2 mM UDP-GlcNAc, 0.1 mM UDP-GlcUA, 0.25 pCi of UDP-["C]GlcUA and 10 to 250 pg of enzyme protein. In early experiments, 2 mM UMP was included to inhibit the high phosphodiesterase activity recovered in some fractions, but this was found not to be a significant problem at pH near 7. Where indicated, 2.5 mg/ml heparin was included as an inhibitor of testicular hyaluronidase. Mes buffer was used in place of Hepes to determine the effect of pH in the range from pH 5 to 7.
After incubation at 37 "C, the reaction was terminated by boiling in the presence of 1% sodium dodecyl sulfate after the addition of 100 pg of unlabeled umbilical cord hyaluronate. ["C]Hyaluronate was separated by chromatography on Sephadex G-50 (0.7 X 27 cm), eluted with 0.5 M NaCl and 0.5-ml fractions were collected for determination of radioactivity. The columns were eluted with 10% pyridine adjusted to pH 6.4 with glacial acetic acid when the material in the void volume was to be lyophilized. In some experiments, Streptomyces hyaluronidase was incubated with the boiled reaction mixture and the detergent was omitted. In other experiments, the material eluting in the void volume of the G-50 column was digested with streptococcal hyaluronidase and the disaccharide products were determined as described below.
Marker Enzymes-Phosphodiesterase I (EC 3.1.4.1) (plasma membrane) was measured by the release of p-nitrophenol from p-nitrophenyl-5'-thymidylate according to Aronson and Touster (18). ' Apparently, the production of ADP from ATP was considered a necessary condition for cell-free hyaluronate synthesis in the early literature, hence the "synthetase" nomenclature. In the interest of continuity, "synthetase" and not "synthase" is used here. phosphatase (EC 3.1.3.9) (endoplasmic reticulum) was measured according to Aronson and Touster (18) in the presence of 1 mM EDTA. The released phosphate was determined by their modification of the procedure of Fiske and SubbaRow (20). Acid phosphatase (EC 3.1.3.2) and 0-D-galactosidase (EC 3.2.1.23) (lysosomes) were assayed by the release of p-nitrophenol from p-nitrophenyl phosphate and p-nitrophenyl-B-D-galactoside at pH 4.5 in 0.1 M acetate buffer. For assays of homogenates and membrane fractions, aliquots for assay were briefly sonicated. For assay of whole cells (latency), enzymes were assayed at isotonic conditions in HBSS-Hepes, pH 7.4.
Protein Determination-Protein was determined by the dye-binding method of Bradford (21) using bovine globulin as standard. Sonication of homogenates and membrane fragments was necessary to eliminate large debris which did not react uniformly.
Electron Microscopy-Membrane fractions were prepared as described above and fixed according to the procedure of Karnovsky (22). Sections were prepared and micrographs obtained by Dr. M.
Press (University of Chicago) by the following procedure: the fractions were postfixed for 1 h in 2% osmium tetroxide, dehydrated through a graded series of alcohols and propylene oxide, and embedded in epon. Thick (1 pm) and thin (70-90 nm) sections were cut with a Porter-Blum MT-2 ultramicrotome. The thin sections were stained with uranyl acetate and lead citrate (28) and examined with a Siemens Elmiskop 101 electron microscope.
Determination of Radioactivity-Aqueous samples were brought to 0.5 ml with distilled water if needed. Samples at basic pH were neutralized with the appropriate amount of 0.1 M HCl. Samples were mixed with 4.5 ml of aqueous counting solution scintillation mixture and counted with a Packard Tri-Carb 460 CD liquid scintillation spectrometer. Paper strips were eluted with 0.5 ml of distilled water before the addition of scintillation mixture.

Assay of Hyaluronate Production by G26-24 Cells in
Culture-The G26-24 cell line has the unusual property of rendering the growth medium highly viscous after only 2-3 days in culture. Dawson and Kernes (23) found that hyaluronate accounted for 80% of the [3H]acetate incorporated into glycosaminoglycans by these cells. The total yield from the media was 35 mg/liter/48 h, approximately 10-fold higher than other neurotumor cell lines. Hypersecretion of hyaluronate appeared to be the source of this difference between cells of the G26 series and other cell types.
In our studies, production of hyaluronate was confirmed and quantitated by labeling with [3H]acetate and [3H]glucosamine. Total glycosaminoglycans labeled with [3H]acetate were recovered from the media by cetylpyridinium chloride and ethanol precipitation and chromatographed on Sephadex G-50 columns before and after digestion with Streptomyces hyaluronidase, as shown in Fig. 1. The hyaluronidase-sensitive material accounted for 89.7 f 6.7% (average of six determinations) of the radioactivity which eluted in the void volume. An aliquot of the medium from cells incubated with [3H] glucosamine was chromatographed on Sephacryl S-1000 columns. The single high molecular size peak trailed into the included volume. The material in this peak was unaffected by incubation with trypsin or pronase (not shown), but completely digested to low molecular weight products by leech hyaluronidase and by all other hyaluronidases employed in this study.
In order to rapidly and accurately assay hyaluronate synthetic activity in this investigation, a new assay procedure was developed. This assay is based on the incorporation of ['4C]glucuronic acid into high molecular weight, hyaluronidase-sensitive material. The solubilized assay mixtures were chromatographed on identical Sephadex G-50 columns in tandem. The products of the reaction were well separated and represented less than 2% of the total substrate added. Ninety f 5% of the radioactivity in the void volume was digestable by Streptomyces hyaluronidase for all enzyme fractions tested.
Only 5-10% of the macromolecular radioactivity was recovered in parallel assays in the absence of UDP-GlcNAc, indicating that the assay was indeed measuring polymerization dependent on the presence of both substrates. In Fig. 2 are shown the linearity of the reaction with time and protein concentration (left) and the pH dependence (right).
Activities with Intact Cells and Homogenates-In order to determine whether synthetase is accessible to substrates or protease at the outer cell surface, a series of experiments was performed with whole cells and homogenates derived from them. The relative activities in whole cells and homogenates of hyaluronate synthetase, phosphodiesterase, and /3-galactosidase assayed in isotonic buffer at neutral pH are shown in Table I. More than 90% of the cells excluded trypan blue before sonication. Phosphodiesterase has been found useful as a primary plasma membrane marker (18) while P-galactosidase is a lysosomal hydrolase previously described in this cell line (24). Slightly more than 10% of the hyaluronate synthetase activity of the homogenate was found in intact cells, probably due to leakiness of some cells combined with uptake of substrate over the course of the incubation. Although the two marker enzymes were not assayed under their optimal conditions, the results suggest an intracellular site  incorporation plotted against protein concentration in the assay. C : the pH activity profile of hyaluronate synthetase. The activity was measured between pH 5 and 7 in Mes buffer and between pH 6.5 and 8.5 in Hepes buffer.
' Cells were subjected to brief sonication as described under "Experimental Procedures." for UDP-sugar binding (and transfer) in hyaluronate synthesis.
In another approach to probe cell surface activities, the differential sensitivity of synthetase to trypsin treatment in whole and homogenized cells was determined. For comparison, the two other enzymes examined in the previous section were included. Hyaluronate synthetase was indeed trypsin sensitive in homogenates and isolated membranes ( Table 11). Trypsinized whole cells, however, exhibited higher activity than their untrypsinized counterparts. This is consistent with the possibility that the activity found with intact cells is not due to surface exposed synthetase, but to the partial "leakiness'' of cells to the substrates. Apparently, the cell surface was rendered more permeable to the nucleotide sugar substrates by treatment with trypsin. In contrast, the activity of the other two enzymes was not increased following trypsinization of whole cells, suggesting that they are not exposed by protease treatment.
Subcellular Fractionation-Since the preliminary experiments suggested that hyaluronate synthetase was not accessible at the external plasma membrane, attempts were initiated to subfractionate the G26-24 cells. The distribution of hyaluronate synthetase and marker activities after differential centrifugation was examined as a prelude to sucrose density fractionation. The partitions between the 20,000 X g pellet and supernatant fractions are shown in Fig. 3. Eighty per cent (+lo%) of each activity was recovered from the original homogenate in these fractions. As expected, most of the phosphodiesterase and p-galactosidase activities were recovered in the pellet. The acid phosphatase, however, showed a more even distribution; approximately 30% of the total activity did not sediment after centrifugation at 100,000 X g for 1 h. The glucose-6-phosphatase and galactosyltransferase also were more evenly distributed. However, more than 80% of the hyaluronate synthetase activity was recovered in the 20,000 x g pellet and essentially all of the remaining activity was sedimentable after centrifugation a t 100,000 x g for 1 h.
This confirmed the membrane-bound nature of the synthetase, and suggested that it was preferentially present in large particles, in good agreement with the results of Ishimoto et al. ( 5 ) and Appel et al. (8).
Further fractionation of the pellets obtained after differential centrifugation was attempted with a variety of sucrose density gradient techniques. The distribution of activities from the 20,000 X g pellet in a continuous sucrose gradient was determined to provide an indication of the density distributions of the marker enzymes. The organelle marker activities utilized routinely in the rat liver system exhibited only small differences in their density distributions in the G26-24 cells (25). By this method, an assignment for hyaluronate synthetase could not be made, as the activity distribution showed some similarities to both galactosyltransferase and phosphodiesterase.
Since the marker enzyme activities of interest had shown a narrow distribution in low density fractions, a modification of the Tabas and Kornfeld (15) method, utilized originally in the purification of rat liver Golgi membranes, was employed. "Cells in monolayer culture were treated with 0.25% trypsin. Homogenate5 and membranes were treated with trypsin at a ratio of cell protein to trypsin of 101 (14).
' Cells were assayed in isotonic sucrose buffer (pH 7.2). e Prepared by brief sonication and assayed as described under "Experimental Procedures" under appropriate conditions for each enzyme.
' Per cent recovery for fraction.
The per cent recoveries and specific activities in the three interfaces of the discontinuous sucrose gradient are shown in Table 111. Although the differences are again small, the pgalactosidase showed the greatest partition into the 15/31% interface. The galactosyltransferase was differentially distributed from the phosphodiesterase activity. Hyaluronate synthetase was approximately equally distributed in the lower two interfaces, again suggesting its presence in a different membrane population than the Golgi marker. Since hyaluronate synthetase was sedimentable at intermediate forces, a procedure was devised to exploit this feature. This method consisted of allowing cells to swell in hypotonic buffer followed by brief, gentle homogenization and rapid re-

Hyaluronate Synthesis in
Plasma Membrane 502 1 equilibration to isotonic conditions. Initial centrifugation was increased from 600 X g for 10 min to 1,000 X g for 10 min. This pellet and the following one (20,000 X g for 20 min) were subjected to discontinuous sucrose gradient fractionation as shown in Fig. 4. All the enzyme activities and total protein were distributed approximately equally between the low force pellet and supernatant, perhaps indicative of the gentle homogenization conditions. Phosphodiesterase and hyaluronate synthetase under these conditions yielded strikingly similar distribution patterns that were unlike that of the galactosyltransferase activity (Fig. 4A). Electron micrographs of the various pellets and interfaces showed some differences in the ultrastructure of membrane populations (not shown). Typically, material collected from interfaces 1 and 2 (0.25 M/20% and 20/40%) contained primarily smooth vesicles and membrane fragments. In contrast, interface 3 (40/50%) contained granular vesicles and fragments indicating the presence of bound ribosomes, and also a ll I 2 3 8 n 1 2 3 8 I

FIG. 4. Distribution of activities in discontinuous sucrose gradients, control and hyaluronidase-treated cells. Top half:
cells were homogenized in hypotonic buffer (10 mM Hepes, 0.5 mM dithiothreitoi, pH 7.2) and immediateIy re-equilibrated in 0.25 M sucrose buffer. The pellets obtained after sequential centrifugation at 1,000 X g for 10 min and 20,000 X g for 20 min, resuspended in 0.25 M sucrose buffer, were layered on top of a discontinuous sucrose gradient (50, 40, and 20%, w/w). After centrifugation, the material which accumulated at each interface and the pellet was collected and assayed for protein and enzyme activity, as described under "Experimental Procedures." A : 1000 X g pellet. Top of A: the total recovered activity in each fraction; bottom: specific activity relative to that of the homogenate. Interfaces are numbered from top to bottom of the gradient. B : the same data displayed for the 20,000 X g pellet. Bottom half: Cells were treated with 10 IU/ml of testicular hyaluronidase (15 min at 37 "C) before harvest from tissue culture plates. Homogenization and construction of gradients were conducted in parallel with control and trypsin-treated cells, as described for the top half. A', top   and bottom: 1,OOO X g pellet; B, top and bottom, 20,000 X g pellet.

4.7
a Cells in monolayer culture were incubated exactly as described for hyaluronidase-treated cells, except hyaluronidase was omitted.
See "Experimental Procedures." e Cells in monolayer culture were incubated with 10 IU ml" testicular hyaluronidase as described. Stimulation of Synthesis Actiuity by Hyaluronidase Treatment-A frequent observation made while preparing cells for homogenization was that cell pellets were loose and fluffy, in contrast to tight pellets obtained with fibroblasts or other cells that did not produce copious amounts of hyaluronate. The possibility that the loose pellet was cell-associated hyaluronate (as observed in hyaluronate-producing streptococci (26)) which might be interfering with fractionation techniques was considered. Therefore, cells in monolayer were treated with hyaluronidase before harvest. Untreated and trypsinized cells were incubated and processed in parallel since it has been shown that trypsin also removes at least some hyaluronate from the surface of eukaryotic cells in culture (27). The homogenates of hyaluronidase-treated cells were unexpectedly found to have approximately 4-fold higher specific activity of hyaluronate synthetase, but all other activities were unaffected, as shown in Table IV. The activities of all enzymes from the trypsinized cells were similar to those from controls (see also Table 11).
The fractionation of the hyaluronidase-treated cells on discontinuous sucrose gradients, shown in Fig. 4, A ' and B', was conducted simultaneously with control and trypsinized cells (not shown). While the distribution of protein and other enzyme-specific activities were similar to that of controls, the specific activity of hyaluronate synthetase was elevated in each interface. However, the specific activity relative to that of the homogenate was increased by 25% in interface 2 (20/ 40%) sucrose) and decreased by 50% in the other gradient fractions. Interface 2 consistently had the highest phosphodiesterase specific activity as well, emphasizing the co-distribution of this marker enzyme and the stimulated synthetase activity.

DISCUSSION
Although production of hyaluronate by cells of neuronal origin in tissue culture was conclusively demonstrated over a decade ago (13), the present work is the first demonstration of the hyaluronate synthetase in a neuronal line. The G26-24 cell line has several features which are useful for cell fractionation studies. The cells are not particularly fastidious, and grow quickly to high densities. Since the cells do not produce a fibroblastic extracellular matrix, they are readily dislodged from tissue culture plates with minimal cell breakage, and therefore do not require treatment with protease or chelators which might perturb the cell surface. Finally, this line produces large amounts of hyaluronate but small amounts of other glycosaminoglycans.
Previous work has suggested that the hyaluronate synthetase may be located in the Golgi apparatus (6), in the manner of a typical glycosyltransferase, or that the synthetase may be located at the plasma membrane (5). While there is no precedent for an extracellular polymer to be synthesized at the latter location in mammalian systems, there does exist evidence to suggest that hyaluronate may not be synthesized in a manner similar to the other glycosaminoglycans. Synthesis of hyaluronate is unique in that it is essentially unperturbed by inhibitors of protein synthesis (3). Although the structure of hyaluronate is similar to the other glycosaminoglycans, it is uniquely not subject to a variety of postsynthetic modifications, including uronic acid epimerization, N-deacetylation, and Nand 0-sulfation, reactions that presumably occur in the Golgi and smooth membranes. Exogenously added chain initiators do not initiate hyaluronate synthesis. The use of oligosaccharides as acceptors, which was the most direct and conclusive means of elucidating the biosynthetic pathways of chondroitin sulfate and heparin, has been unsuccessful in the investigation of hyaluronate biosynthesis, despite the repeated attempts of several investigators over the last two decades (3, 29). At the least, these problems suggest that either a different synthetic mechanism or a separate compartmentation, or both, may obtain for hyaluronate.
In view of these considerations, an examination of the subcellular location of the synthetase was undertaken. The possibility that hyaluronate synthetase might in fact belong to the class of ectoenzymes with functional groups located on the outer surface of the cell membrane, as suggested by Ishimoto et al. (5) was explored. Minimal synthetase activity was observed when nucleotide sugars were incubated with intact cells, as opposed to homogenates or membrane preparations. Trypsin treatment of intact cells did not destroy activity, as did treatment of homogenates. Prehm (29) has also reported trypsin sensitivity of the synthetase in membrane preparations in a teratocarcinoma cell line. Unfortunately, we have not been able to identify a nonpenetrating reagent which preferentially inactivates the synthetase in whole cells, in contrast to the stimulation seen after hyaluronidase treatment of cells in culture. Thus, the results of homogenization and trypsin treatment of whole cells suggest that both the nucleotide sugar-binding sites and proteasesensitive sites of the synthetase are located intracellularly.
Upon homogenization of cells, essentially all of the hyaluronate synthetase activity was sedimentable by ultracentrifugation or at moderate forces (20,000 x g for 20 min), confirming its membrane-bound nature. The partition of the synthetase between pellet and supernatant fractions after centrifugation correlated with a plasma membrane marker and not with a glycosyltransferase Golgi marker, although optimal separation of the various membrane marker activities in sucrose gradients proved problematic. Even in continuous sucrose gradients where the several marker enzymes had similar equilibrium densities concentrated at 1.14 g/ml (30 to 36% sucrose), the distribution of hyaluronate synthetase was most similar to that of alkaline phosphodiesterase. The narrow distributions may have been due to the high amounts of glycosphingolipids, particularly galactosyI-and glucosylceramide, sulfatide, and ganglioside (GMl) produced by these cells, which is characteristic of their neuronal origin (24).
Better separation was achieved by nonequilibrium velocity sedimentation in discontinuous sucrose gradients. In these experiments, the most striking separations were obtained when the cells were subject to very gentle homogenization in hypotonic buffer. One thousand x g pellets were prepared to take best advantage of the previous observation that both hyaluronate synthetase and phosphodiesterase activities were found to sediment preferentially at low to intermediate forces, presumably in large membrane fragments. Intact cells, nuclei, and very large cellular debris were expected to sediment at higher densities in the gradient than liberated membranes. Analysis of these gradients showed quantitative similarity between the hyaluronate synthetase and phosphodiesterase activity in smooth membranes recovered in the 20/40% sucrose interface. Thus, by several different methods, hyaluronate synthetase was recovered in plasma membrane-enriched fractions.
Treatment of cells in monolayer culture with hyaluronidases produced an immediate and sustained increase in the activity of the synthetase, suggesting direct communication of cell surface conditions to the enzyme. This modified enzyme allowed an additional criterion with which to probe the localization of hyaluronate synthesis. To our knowledge, this effect of hyaluronidase treatment on intact cells has not previously been reported. Further analysis of the hyaluronidase effect and a novel kinetic model for hyaluronate synthesis will be presented ~ubsequently.~ In conclusion, evidence has been presented for the localization of hyaluronate synthetase at the inner surface of the plasma membrane. The presence of the synthetic apparatus for a glycosaminoglycan in the plasma membrane is novel, but does help explain several of the enigmas concerning the properties and biosynthesis of hyaluronate. If the hyaluronate molecule is not an intracellular product, then the lack of involvement of protein synthesis and any of the postsynthetic modification reactions common to the other glycosaminoglycans is immediately explained. Synthesis and extrusion directly from the plasma membrane also obviate the necessity for a packaging or secretion mechanism for these extremely large (>IO6 Da) and voluminous molecules. Furthermore, in those cells and tissues where hyaluronate is involved in aggregated structures, such as with chondroitin sulfate proteoglycan and link protein in cartilage, the obligate association interactions may begin to occur near the site of synthesis and extrusion of hyaluronate on the membrane surface.
Whether this mechanism represents an "alternative" synthetic pathway as suggested by Goldberg and Toole (30) is unclear. Their finding that the ionophore monensin inhibits extracellular accumulation of hyaluronate in rat fibrosarcoma and 3T3 cells, but not in human articular chondrocytes or rat chondrosarcoma cells (31) may be easily explained by the interruption of transport of the synthetase, not the hyaluronate, to the cell surface. Different turnover rates of the synthetase in the various cell types would unify the apparently discrepant results. The investigation of this possibility is underway.