Biosynthesis of Hyaluronic Acid by Streptococcus*

Synthesis of hyaluronic acid was investigated in a cell-free system derived from a strain of Group A streptococci. Preparative procedures were improved so that an enzyme system 70 times more active than that previously reported was obtained. The hyaluronic acid synthesized could be separated into trichloroacetic acid-soluble and -insoluble fractions. On the basis of pulse-chase experiments, it was shown that the trichloroacetic acid-insoluble fraction is a precursor of the soluble fraction. The release of the trichloroacetic acid-insoluble hyaluronic acid is specifically blocked with p-chloromercuribenzoate, without inhibition of chain elongation. The addition of butanol to trichloroacetic acid resulted in solubilization of all of the hyaluronic acid. No detectable difference in molecular size was observed between the two hyaluronic acid fractions, both of which were estimated to be more than one million daltons in size. Testicular hyaluronidase digestion of either one of the two types of hyaluronic acid yielded no high molecular weight fragments, indicating that hyaluronic acid is not bound covalently to protein. However, following incubation of enzyme assay mixtures with UDP-[14C]GlcUA, even in the absence of UDP-GlcNAc, radioactive high molecular weight hyaluronic acid was obtained which suggests that the enzyme system elongates rather than initiates hyaluronic acid chains. Tunicamycin did not inhibit hyaluronic acid synthesis, indicating lack of participation of an intermediate of pyrophosphorylpolyisoprenol type. The results obtained are consistent with the hypothesis that chain elongation of hyaluronic acid proceeds by alternate addition of monosaccharides from UDP-sugars by a membrane-bound synthesizing system followed by release of completed hyaluronic acid chains.


Synthesis
of hyaluronic acid was investigated in a cell-free system derived from a strain of Group A streptococci. Preparative procedures were improved so that an enzyme system 70 times more active than that previously reported was obtained. Hyaluronic acid, a ubiquitous constituent of connective tissues, has been implicated in such diverse biological phenomena as lubrication (l), cell adhesion (2)(3)(4), cell mobility (5), and formation of aggregates with chondroitin sulfate proteoglycan (6,7). However, the mechanism of hyaluronic acid biosynthesis is imperfectly understood.
Among the reported hyaluronic acid-synthesizing systems (8-ll), the Streptococcus system has been studied most extensively (12)(13)(14)(15)(16)(17) and has several advantages including the exclusive synthesis of one glycosaminoglycan (hyaluronic acid) and the high specific activity of the enzyme preparations.
Previous studies * This investigation was supported by Grants HD-09402, HD- led to the speculation that hyaluronic acid is synthesized by alternate addition of monosaccharides (15). However, recently Turco and Heath reported the isolation of a dolichol-pyrophosphate-GlcNAc-GlcUA' from SV40-transformed human lung fibroblasts (18). Hopwood and Dorfman also reported the isolation of glycolipids containing glucuronic acid and Nacetylglucosamine from a rat fibrosarcoma (19). Although the role of these lipid intermediates is unknown, it is possible that they are involved in either heparin (or heparan sulfate) or hyaluronic acid synthesis. Takatsuki and Tamura reported the inhibition by tunicamycin of the total glycosaminoglycan synthesis in cultured embryonic chick fibroblasts (20). In the present study, the Streptococcus system was used to reinvestigate the mechanism of hyaluronic acid synthesis including the possible involvement of a lipid intermediate.   conditions. Three of the incubation mixtures were terminated after 20,40, or 60 min of incubation, when to another three of the incubations (---) were added a 7.4-fold excess (67 nmol) of nonradioactive UDP-GlcUA. The other two (p) received an equivalent volume (10 ~1) of water (control).
The incubations were terminated at times indicated and assayed for incorporation into trichloroacetic-soluble (0) and -insoluble (0) hyaluronic acid, respectively, as described under "Experimental Procedures." radioactive compounds isolated from both trichloroacetic acid-soluble and -insoluble fractions were hyaluronic acid.
The Relationship between Trichloroacetic Acid-soluble Hyaluronic Acid and Trichloroacetic Acid-insoluble Hyaluronic Acid A chase experiment was performed to examine the relationship between the soluble hyaluronic acid and the insoluble hyaluronic acid. The incubation was fast carried out with an amount of UDP-[ 14C]GlcUA that was consumed, and then an excess amount of nonradioactive UDP-GlcUA was added. As shown in Fig. 2, after 20 min of incubation, radioactive UDP-['4C]GlcUA was almost depleted, and following addition of nonradioactive UDP-GlcUA, the bound hyaluronic acid was released from the particulate enzyme fraction. The results indicate that the bound hyaluronic acid is nascent and that it is released from the particulate enzyme after elongation or completion of the chain.

Properties of the Soluble and Bound Hyaluronic Acid
The trichloroacetic acid-insoluble radioactive material is known to be extracted with 50% pyridine or 0.5% sodium deoxycholate (15). Extraction also can be achieved by acetone/phosphate buffer as described under "Experimental Procedures." In contrast, 4% butanol, 4% butanol containing 0.01 M EDTA, 95% ethanol, chloroform/methanol (2:1), ethanol/ ether (2:1), 2 to 8 M urea, 2 M NaCl, 1 M sodium acetate, or 80% saturated (NH&SO4 did not solubilize the radioactive material (15). The mechanism by which the radioactive material is rendered soluble is not known. Attempts to extract the trichloroacetic acid-precipitable radioactive material with absolute pyridine, absolute butanol, chloroform/methanol/ water (1:6:4) or (1:1:0:3) were unsuccessful. However, extraction was accomplished with an unique solvent system, which is a mixture of a small amount of butanol and 5% trichloroacetic acid. As shown in Fig. 3, more than 80% of the insoluble radioactive material was extracted at butanol concentrations above 5%, while below 5% only 15 to 20% was extracted. It should be noted that aqueous butanol without trichloroacetic acid extracted less than 30% of the radioactive material even at butanol concentrations greater than 6%. The results suggest that there is some hydrophobic interaction resulting in binding of hyaluronic acid.
On between soluble and bound hyaluronic acid. Since both hyaluronic acid preparations were almost totally excluded from a Sepharose 2B column as shown in Fig. 4, their average molecular size was estimated to be at least 1 X 10" daltons. The peak eluted in the column volume is mainly due to the umeacted UDP-['?]GlcUA. Further investigation of the retarded peaks are described below.
p-Chloromercuribenzoate in a concentration greater than 1.0 mM inhibited hyaluronic acid biosynthesis (Table I). There is a marked difference in the extent of the inhibition of synthesis of soluble and bound hyaluronic acid. A time course experiment using 1.5 mM p-chloromercuribenzoate demonstrated (data not shown) that even after 100 min, inhibition of synthesis of insoluble hyaluronic acid was only 30%, but inhibition of incorporation of radioactivity into the soluble fraction was almost complete. These results as well as those illustrated in Fig. 2 indicated that bound hyaluronic acid chains are nascent or elongating, and soluble hyaluronic acid chains are mature and have been released from the membrane. p-Chloromercuribenzoate appears to inhibit the release of completed hyaluronic acid chains.

Binding of Insoluble Hyaluronic Acid
It was of interest to determine whether hyaluronic acid in the trichloroacetic acid-insoluble fraction is bound to some nonpolysaccharide component. In order to determine the possible nature of such a compound, the enzyme was incubated with UDP-['4C]GlcUA in the absence of UDP-GlcNAc. A phosphate buffer extract of the trichloroacetic acid precipitate from such an incubation mixture gave two radioactive peaks on Sephadex G-100 as shown in Fig. 5. One was eluted in the void volume and the other which was included had the same molecular size as authentic UDP-GlcUA on Sephadex G-50 (superfine) (data not shown). The results suggest that the acceptor for the glucuronic acid transfer from UDP-[14C]-GlcUA has a size of greater than 1 x lo5 daltons. This could either be a nonpolysaccharide acceptor or a growing hyaluronic acid chain.
The following experiments were performed to determine the possible nature of any component attached covalently to bound hyaluronic acid.
Gel Filtration of Insoluble Hyaluronic Acid under Dissociating Conditions-Insoluble hyaluronic acid was extracted by acetone/phosphate buffer and subjected to gel filtration on a G-100 column (1.2 x 107 cm) in the presence of 2 M NaCl, 2 M urea, or 0.5% sodium dodecyl sulfate. In the last case, the cases, the radioactivity emerged in the void volume (data not shown). No smaller molecular weight peak was revealed under any of those dissociating conditions, indicating that insoluble hyaluronic acid is not an aggregated form.
Pronase Digestion and Alkali Treatment of Insoluble Hyaluronic Acid-Experiments were undertaken to determine whether the insoluble hyaluronic acid is bound covalently to a protein. A radioactive insoluble hyalmonic acid preparation obtained by acetone/phosphate buffer extraction was treated with pronase or dilute alkali under conditions which cleave 0-glycosidic bonds (23). After each treatment, the sample was subjected to gel filtration on a Sepharose 4B column (1.2 x 110 cm). In both cases, no significant change in gel filtration pattern was observed since the radioactivity emerged in the void volume (data not shown). The results suggest that insoluble hyaluronic acid is not attached to a protein sufficiently large to affect the gel fdtration pattern on Sepharose 4B. Testicular Hyaluronidase Digestion-Since hyaluronic acid is a large molecule, gel filtration on Sepharose 4B may not have detected the possible slight decrease in molecular size in the experiments described above. Therefore, both soluble and insoluble hyaluronic acid fractions were digested with highly purified testicular hyaluronidase and subjected to gel filtration on a column of Sephadex G-100 (Fig. 6A). Sufflcient radioactivity was used so that if any oligosaccharide remained attached covalently to a protein, and if the addition of radioactive sugars takes place not only in the distal region but also in the proximal region of the polysaccharide it would be expected to be detected and separated from the free oligosaccharides produced by enzymic digestions. However, both digests gave only one peak which was eluted in the column volume, suggesting the absence of a protein core. Similar results were obtained by streptococcal hyaluronidase digestion followed by Sephadex G-50 chromatography (data not shown). Each radioactive peak in the column volume was further analyzed on a column of Sephadex G-50 (superfine). Both preparations showed the same chromatographic pattern, giving two major radioactive peaks, which are probably hexaand tetrasaccharide, respectively, and two minor radioactive peaks, presumably octa-and disaccharide (Fig. 6B). It seems unlikely that an additional constituent is attached covalently to only insoluble hyaluronic acid. If any, it would be very small in molecular size.

Effect of Sugars and Glycosides on Hyaluronic Acid Synthesis
Much information has been accumulating concerning the synthesis of glycosaminoglycans.
Chondroitin sulfate synthesis is known to be initiated by addition of xylose to serine residues onto an acceptor protein (26). However, nothing is known concerning chain initiation of hyaluronic acid. No substance with acceptor activity, natural or synthetic, has been found. Although Stoolmiller and Dorfman (16) tested several radioactive UDP-sugars, such as UDP-Glc, UDP-Gal, UDP-xylose, and UDP-arabinose in a streptococcal microsomal system, no measurable radioactivity was incorporated from any except from UDP-['%]Glc.
In the case of UDP-['%]Glc, radioactivity was incorporated into the particulate enzyme fraction. Although the nature of radioactive product was not determined, the addition of UDP-Glc did not stimulate the hyaluronic acid synthesis. We tested a variety of sugars and glycosides including hyaluronic acid hexasaccharide, and hyaluronic acid itself to determine whether they stimulate hyalmonic acid synthesis (Table II) (Table  III). These experiments furnished no evidence for the participation of lipid-linked intermediates of at least pyrophosphorylpolyisoprenol type.

Investigation of the Trichloroacetic Acid-soluble Fraction
When the trichloroacetic acid-soluble fraction was examined in detail by gel filtration, low molecular weight substances were found which possibly could be involved in hyaluronic acid biosynthesis.
The trichloroacetic acid-soluble fraction from a standard incubation was analyzed by gel filtration on several gels. Two peaks were obtained on Sepharose 2B, one in the void volume (containing hyaluronic acid) and the other in the column volume (Fig. 4). However, when the trichloroacetic acid-soluble fraction was analyzed on Sephadex G-50 (superfine), the retarded peak observed on the Sepharose 2B was successfully separated into two peaks (Fig. 7), one of which is UDP-GlcUA with a small shoulder of presumptive free glucuronic acid and the other is an unidentified radioactive peak, designated as Substance A. The radioactivity incorporated into Substance A accounted for approximately 1.0% of the total radioactivity added to the incubation mixture.

Properties of Substance A-Substance
A also was observed in preparations not treated with trichloroacetic acid. The high speed supernatant solution (78,000 X g for 60 min) from a standard incubation or a sodium deoxycholate extract of a standard incubation mixture gave a radioactive peak on a Sephadex G-50 column which was similar in elution position and quantity, suggesting that Substance A does not result from degradation by trichloroacetic acid. Rather, it seems to be produced by an enzymic reaction as indicated in time course experiments (Fig. 8), the rate of production is parallel to the synthesis of hyaluronic acid. On paper chromatography using Solvent B weight was estimated to be approximately 1,100 on a Bio-Gel P-2 column (Fig. 9). Interestingly, Substance A was not observed on a Sephadex G-50 column when UDP-GlcNAc was omitted from an incubation mixture, suggesting that Substance A contains both GlcUA and GlcNAc. However, it was not possible to label Substance A with UDP-[3H]GlcNAc.
Although a similar peak (Substance B) was observed on a Sephadex G-50 (superfine) column when UDP-[3H]GlcNAc was used in the presence of unlabeled UDP-GlcUA, Substance B differed from Substance A in that it had a RUDP value of 0.63 on paper electrophoresis and had a molecular weight of   1,300 (Fig. 9). In addition, Substance B was present when UDP-GlcUA was omitted from the incubation mixture. Substance B may be involved in peptidoglycan rather than hyalmonic acid synthesis.
Experiments were undertaken to determine whether radioactivity from Substance A or B is transferred to hyaluronic acid. (See "Experimental Procedures.") After incubation of the substances with the enzyme protein, less than 2% of the radioactivity was found in the trichloroacetic acid precipitate. When the trichloroacetic acid supernatant solution was subjected to gel filtration on a column of Sephadex G-50, no radioactivity was recovered in the void volume. We, therefore, could not conclude that either Substance A or B is an intermediate in hyaluronic acid synthesis. The aim of this study was to further characterize hyaluronic acid synthesis and to determine whether hyaluronic acid chain elongation proceeds via a lipid intermediate.
As previously demonstrated (15), two types of newly synthesized hyaluronic acid were observed; one soluble and the other insoluble in trichloroacetic acid. An average molecular size was estimated to be more than 1 x lo6 daltons for both types. On the basis of Adi-HA/di-HA ratio in the streptococcal hyaluronidase digests of both soluble and insoluble hyaluronic acid preparations, Stoolmiller and Dorfman (15) suggested that the latter contains growing chains. Our results from the chase experiments more clearly indicate that insoluble hyaluronic acid is still in the process of elongation and is released only after completion.
p-Chloromercuribenzoate appears to block release of completed chains preferentially.
The concurrent reduced inhibition of synthesis of insoluble chains may result secondarily from the prevention of new chain initiation. The effects of p-chloromercuribenzoate may indicate an enzymic mechanism of chain release. The action of p-chloromercuribenzoate in the presence of dithiothreitol may have resulted from partial oxidation of the latter in the buffer.
As previously demonstrated (15), incubation of the enzyme with only UDP-[%]GlcUA resulted in incorporation of [ %]GlcUA into insoluble hyaluronic acid. The molecular size of the ["'C]GlcUA-labeled material was large enough to be excluded from a column of Sephadex G-100. It seems reasonable, therefore, to assume that there should be either a protein acceptor or a primer of high molecular weight on the membrane. It is not yet certain whether hyaluronic acid is bound covalently to protein. Even after exhaustive purification of hyaluronic acid, small amounts of amino acids have been found in preparations from various sources such as synovial fluid (30)(31)(32), rooster comb (33, 34), brain (35), human umbilical cord (32), bovine vitreous (32), and streptococcus (36). No linkage region, however, has been isolated. Stoohniller and Dorfman (15) observed no inhibition by puromycin or chloramphenicol of hyaluronic acid synthesis in Streptococcus. In our studies, no evidence was obtained for the covalent linkage of hyaluronic acid to protein. Neither testicular hyaluronidase nor streptococcal hyaluronidase treatment resulted in the association of any hyaluronate fragment with a high molecular weight compound as determined by gel filtration. If, however, the enzyme system is elongating the chain rather than initiating, as presumed from the experiments described in Fig. 5, this approach might not be sensitive enough to detect an initiator.
Therefore, alkali treatment and pronase digestion also were used. Neither of these treatments of insoluble hyaluronic acid decreased the molecular size when examined on Sepharose 4B. Furthermore, none of the monosaccharides or glycosides tested for acceptor activity stimulated hyaluronic acid synthesis. On the basis of these negative results, it seems most likely that hyaluronic acid synthesis, unlike that of chondroitin sulfate, is not initiated on a protein core. It seems possible that the chain initiation takes place on a membranebound enzyme system.
The primer or bound hyaluronic acid appears to be associated with the enzyme particle through a noncovalent bond. It has been shown that insoluble hyaluronic acid is extracted with 50% pyridine, 0.5% sodium deoxycholate, or acetone/ phosphate buffer (16). It was also shown that hyaluronic acid could be extracted with butanol-containing trichloroacetic acid solution. These findings suggest that a lipid may be involved in formation of the enzyme particle or in its arrangement in the membrane. We cannot eliminate the possibility that there is a very small aglycone for hyaluronic acid molecule, which acts as a chain initiator or as an anchor inserted in the membrane.
Although little is known concerning the mechanism of chain elongation of hyaluronic acid in mammalian systems, previous studies of the Streptococcus system furnished no evidence of involvement of a lipid intermediate and suggested alternative addition of monosaccharide units (15). The following evidence has been presented: (a) no radioactive intermediates were detected by organic solvent extraction; (b) a major reaction product from UDP-sugars in hyaluronic acid synthesis is not UMP, but UDP; (c) bacitracin, an inhibitor for dephosphorylation of lipid pyrophosphate in peptidoglycan synthesis, does not inhibit hyaluronic acid synthesis.
Recently, Turco and Heath (18) reported the isolation of GlcUA-(1 + 4)-GlcNAc-P-P-dolichol from SV40-transformed human lung fibroblasts, and suggested that it might be involved in the biosynthesis of heparin, heparan sulfate, or both. Hopwood and Dorfman (19) also reported the isolation of lipid-linked oligosaccharides containing both glucuronic acid and N-acetylglucosamine, which might be lipid intermediates for heparin (heparan sulfate) or hyaluronic acid biosynthesis. Although it has not been demonstrated that those lipid components can be transferred to glycosaminoglycans, these findings induced a reinvestigation of the involvement of a lipid intermediate.
We repeated the lipid extraction with butanol, which was used to isolate the lipid components from a rat fibrosarcoma (19). However, no measurable radioactivity appeared in the extracts.
Tunicamycin is known to specifically inhibit the formation of N-acetylglucosaminyl pyrophosphorylpolyisoprenol (27)(28)(29), the lipid intermediate for biosynthesis of N-glycosidically linked glycoproteins (37). If indeed such a lipid intermediate is involved in hyaluronic acid biosynthesis, tunicamycin would be expected to inhibit hyaluronic acid synthesis. Recently, Takatsuki and Tamura (20) have reported that the drug inhibits the biosynthesis of total glycosaminoglycans in cultures of chick embryo fibroblasts. No inhibition, however, by tunicamycin of hyaluronic acid synthesis in Streptococcus was found by us. A similar observation in this laboratory using cultured rat glial cells4 was made. We concluded that a pyrophosphorylpolyisoprenol type of lipid intermediate was not involved in hyaluronic acid synthesis. In the course of preparation of this manuscript, Hart and Lennarz (38) reported no inhibition of hyaluronic acid synthesis by tunicamycin in embryonic chick cornea. It has been recently reported by Yamamori et al. (39) that a tunicamycin-like antibiotic 24010 inhibits the formation of N-acetylglucosaminyl pyrophosphorylundecaprenol, but not of some N-acetylglucosaminyl phosphorylundecaprenol.
Therefore, the possibility of the involvement of phosphorylpolyisoprenol can not be ruled out. Formation of a radioactive high molecular weight compound on incubation of the enzyme with UDP-["'C]GlcUA in the absence of UDP-GlcNAc also seems to support the hypothesis that chain elongation may proceed by alternate addition of monosaccharides rather than a disaccharide unit. However, the possibility that endogenous UDP-GlcNAc might have given some disaccharide unit-linked intermediate cannot be eliminated.
On examination of the trichloroacetic acid-soluble fraction from an incubation, Substance A was found which appeared to be enzymically synthesized from UDP-[i4C]GlcUA. Since it was not produced when UDP-GlcNAc was omitted from the incubation mixture, it seems to contain both glucuronic acid and N-acetylglucosamine.
Although the Substance A was not labeled with UDP-[3H]GlcNAc, it is presumably because of the fact that the Michaelis constant for UDP-GlcNAc of the enzyme system is 10 times greater than that for UDP-GlcUA (15), and that as a result hyaluronic acid is not synthesized effectively at a very low concentration of UDP-[3H]GlcNAc. The addition of Substance A to an incubation mixture did not result in its transfer to hyaluronic acid. Furthermore, under none of the inhibitory conditions used for hyaluronic acid synthesis, was Substance A accumulated.
Those findings do not support its role as an intermediate for hyaluronic acid synthesis. Interestingly, the production of Substance A seems to parallel hyaluronic acid synthesis as observed in the time course experiments and the inhibition studies. Although it is not feasible to elucidate the structure of Substance A due to the very small amounts of the material, it could be a UDPoligosaccharide.
UDP-oligosaccharides have been isolated from Group A streptococci (40) and from several other sources, such as milk (41,42) and hen oviduct (43)(44)(45)(46), although their role is not yet known. On the basis of the results presented here, we can conclude with a fairly high degree of certainty that the chain elongation of hyaluronic acid synthesis in Streptococcus proceeds by alternative addition of monosaccharide units, but not via a lipid intermediate.

Acknowledgment-We
wish to thank Dr. Allen L. Horwitz (University of Chicago) for helpful discussions during the course of this work.