Mechanism of Type 3 Capsular Polysaccharide Synthesis inStreptococcus pneumoniae *

The glycosidic linkages of the type 3 capsular polysaccharide of Streptococcus pneumoniae([3)-β-d-GlcUA-(1→4)-β-d-Glc-(1→] n ) are formed by the membrane-associated type 3 synthase (Cps3S), which is capable of synthesizing polymer from UDP sugar precursors. Using membrane preparations of S. pneumoniae in an in vitro assay, we observed type 3 synthase activity in the presence of either Mn2+ or Mg2+ with maximal levels seen with 10–20 mm Mn2+. High molecular weight polymer synthesized in the assay was composed of Glc and glucuronic acid and could be degraded to a low molecular weight product by a type 3-specific depolymerase from Bacillus circulans. Additionally, the polymer bound specifically to an affinity column made with a type 3 polysaccharide-specific monoclonal antibody. The polysaccharide was rapidly synthesized from smaller chains and remained associated with the enzyme-containing membrane fraction throughout its synthesis, indicating a processive mechanism of synthesis. Release of the polysaccharide was observed, however, when the level of one of the substrates became limiting. Finally, addition of sugars to the growing type 3 polysaccharide was shown to occur at the nonreducing end of the polysaccharide chain.

The capsular polysaccharides of Streptococcus pneumoniae are essential components in virulence that are necessary to resist host phagocytic mechanisms. So far, ninety different capsular polysaccharides have been identified (1). The polymers are usually composed of repeating oligosaccharide subunits containing several different monosaccharides and, in many cases, are branched structures (2). In S. pneumoniae, the genes involved in the biosynthesis of any one capsular polysaccharide are contained within a single locus on the bacterial chromosome and are termed "type-specific." Only two type 3-specific genes, cps3D and cps3S, which encode a UDP-Glc dehydrogenase and the type 3 synthase, respectively, are essential for synthesis of the type 3 polysaccharide (3), a linear structure composed of [33)-␤-D-GlcUA-(134)-␤-D-Glc- (13] repeating units (4). Two other genes, cps3U, encoding a Glc-1-P uridylyltransferase, and cps3M, encoding a homologue of phosphoglucomutases, are present in the type 3-specific locus. These sequences are not essential for type 3 capsule synthesis, however, because the functions they encode are duplicated by other genes in the pneumococcal chromosome (3,5,6). Flank-ing either side of the type-specific genes are sequences common to all capsule types (7)(8)(9)(10)(11). These common genes have been proposed to play a role in polysaccharide transport and in determination of polysaccharide chain length in some capsular polysaccharides. In type 3 strains, however, most of these sequences are mutated and are not required for polysaccharide synthesis or transport (3,5,12). No other sequences likely to be involved in transport of the type 3 polysaccharide have been identified, and this function may be performed by the type 3 synthase.
The type 3 polysaccharide synthase (Cps3S) 1 shares significant protein homology with several other glycosyltransferases from both prokaryotes and eukaryotes, including the hyaluronan synthase from S. pyogenes, the Nod factor oligosaccharide synthase (NodC) from Rhizobium meliloti, FbfA of Stigmatella aurantiaca, pDG42 of Xenopus laevis, and chitin synthases from both Saccharomyces cerevisiae and Candida albicans (3). All of these enzymes produce homopolymers or polysaccharides with a simple disaccharide repeat, and current evidence suggests that each is capable of forming all of the glycosidic linkages present in their respective polysaccharides (3,(13)(14)(15)(16)(17)(18)(19)(20)(21). Hydrophobic cluster analysis of the deduced sequences of these proteins has revealed two conserved domains that are believed to be responsible for binding the nucleotide sugars and catalyzing the formation of the glycosidic linkages (22,23). Based on this analysis, it has been proposed that these glycosyltransferases synthesize polysaccharide via a common mechanism that involves dual addition of sugars to the growing saccharide chain (23).
Smith et al. (24) demonstrated that type 3 synthase activity resides within the particulate fraction of S. pneumoniae lysates. Using a type 3-specific antibody to precipitate and quantitate polysaccharide, they observed synthesis of type 3 polysaccharide using cell-free lysates, uridine nucleotide sugars, and a divalent metal cation (24). Consistent with the observed membrane location of type 3 synthase activity, the deduced amino acid sequence of Cps3S contains four hydrophobic domains potentially capable of spanning the membrane (3). The role of Cps3S as the type 3 synthase was demonstrated using isogenic strains containing insertion mutations either in or immediately downstream from cps3S. Type 3 polysaccharide could be synthesized in the in vitro system using crude membrane preparations from the latter but not from the former, thus establishing cps3S as the critical gene (3). Expression of the synthase in Escherichia coli has proven difficult, apparently because of toxicity of the membrane protein (3,13). The only reported clone expressing the enzyme produced low levels of protein, despite being contained in a high copy number expression vector (13). The possibility of mutations in the clone was not eliminated, but a small amount of type 3 polysaccha-ride was synthesized in the E. coli strain, further confirming the assignment of cps3S as the type 3 synthase. Because production of high levels of synthase from cloned products has not been possible, we chose to use membranes isolated from S. pneumoniae as a source of synthase activity for further biochemical studies. Using an assay that measures the incorporation of radiolabeled sugars into polysaccharide, we have characterized the mechanism of type 3 polysaccharide synthesis.
Bacterial Strains, Growth Conditions, and Enzyme Preparation-Membranes containing type 3 synthase were isolated from S. pneumoniae based on a previously described procedure (25). The encapsulated type 3 strain WU2 and its nonencapsulated isogenic derivative JD908, which contains an insertion mutation in cps3S, have been described (7,26). A 4-liter culture was grown at 37°C in Todd Hewitt Broth supplemented with 0.5% yeast extract (THY) to a density of 3 ϫ 10 8 colony forming units/ml. The cells were collected by centrifugation at 10,000 ϫ g for 20 min. The pellets were washed once in 2 liters of phosphate-buffered saline (137 mM NaCl, 2.7 mM KCl, 5.4 mM Na 2 HPO 4 ⅐7H 2 0, 1.8 mM KH 2 PO 4 , pH 7.4) and then suspended in 200 ml of protoplast buffer (20% sucrose, 5 mM Tris-HCl, pH 7.4, and 2.5 mM Mg 2 S0 4 ). Mutanolysin was added to a final concentration of 20 units/ml, and the mixture was incubated at room temperature overnight. Protoplast formation was checked by examining the cells using a phase contrast light microscope. The protoplasts were sedimented by centrifugation at 25,000 ϫ g for 20 min, washed once in 200 ml of protoplast buffer, and then osmotically lysed by suspension in 100 ml of sterile water containing 10 mM EDTA. Lysis was confirmed by light microscopy. The membranes were collected by centrifugation at 100,000 ϫ g for 30 min, washed three times in 40 ml of 100 mM Hepes (pH 8.0) buffer containing 10 mM sodium thioglycolate, and suspended in 15 ml of the same buffer. The final membrane preparation was adjusted to 3 mg protein/ml in 100 mM Hepes (pH 8.0) buffer containing 10 mM sodium thioglycolate and stored at Ϫ20°C.
Assay of Synthase Activity-Type 3 synthase activity was determined by the incorporation of 14 C label from either UDP-[ 14 C]Glc or UDP-[ 14 C]GlcUA into polysaccharide. Synthase assays were performed in a 100-l reaction that contained 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , UDP-Glc, UDP-GlcUA, and membranes isolated as described above. The concentrations of UDP-Glc, UDP-GlcUA, and membrane protein are indicated elsewhere in the text and in the figure legends. The reaction mixtures were incubated at 32°C for 20 min. The reaction was terminated by addition of 10 l of 12.5 M glacial acetic acid, and the reaction components were separated by ascending paper chromatography on 3 MM Whatman paper in ethanol (95%):1 M ammonium acetate (pH 7.0), 7:3 (v/v) for 16 -18 h. The chromatograms were dried, and the product retained at the origin (1 inch strip) was determined by liquid scintillation counting in Scinti Verse I.
Assay of Soluble and Membrane-associated Polysaccharide Products-Isolated membranes containing 300 g of total protein were added to reactions containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , UDP-Glc, and UDP-GlcUA in a 600-l volume. The reactions were incubated at 32°C, and, at the times indicated in the figure legends, 50-l samples were taken. These aliquots were submitted to the following protocol to determine the amount of soluble and membrane-associated polysaccharide formed under various experimental conditions. EDTA was added to the aliquots to a final concentration of 10 mM, and the samples were stored on ice. The volume was adjusted to 200 l with 100 mM Hepes (pH 8) buffer containing 10 mM sodium thioglycolate. The membranes were collected by centrifugation at 100,000 ϫ g for 30 min. The supernatants were saved, and the pellets were suspended in 200 l of the same buffer. The membranes were sedimented again by centrifugation at 100,000 ϫ g for 30 min. The resulting supernatants were combined with the previous supernatants, and the pellets were suspended in 400 l of the same buffer. Aliquots of the supernatant and pellet fractions were chromatographed on paper as described above for the synthase assay. The radioactivity at the origin, as well as that corresponding to authentic UDP-Glc, was determined by liquid scintillation counting.
Sepharose 2B Column Chromatography-Gel filtration of polysaccharide samples was performed on a 1.4 ϫ 37 cm Sepharose 2B column equilibrated with 0.2 M NaCl and 5 mM Tris acetate (pH 7.4). SDS was added to the sample to a final concentration of 2% (v/v) prior to application to the column. The column flow rate was 16 ml/h, and 1-ml fractions were collected. The void and total volumes were determined using 0.798-m diameter latex beads (Sigma) and [ 14 C]Glc, respectively. Polysaccharide that eluted between 18 and 30 ml on Sepharose 2B is referred to as high molecular weight.
Chromatography on a Monoclonal Antibody Sepharose Column-Type 3 polysaccharide-specific monoclonal antibody 16.3 (28) was coupled to Sepharose 2B as described (29). The antibody-conjugated beads (0.5-1 ml) were packed into 2-ml columns and washed with 10 column volumes of 5 mM Tris acetate buffer (pH 7.4) containing 0.2 M NaCl. Labeled polysaccharide was applied to the column, and the column was washed with the same buffer. Three 1.5-ml samples were collected, and 100 l of each fraction was chromatographed on paper as described for the assay of synthase activity. The percentage of radioactive polysaccharide that bound to the column was determined as 100 ϫ [(total counts applied Ϫ recovered counts)/total counts].
Purification of Polysaccharide on a Monoclonal Antibody Sepharose Column-Labeled polysaccharide that bound to the antibody affinity column described above was eluted with 0.05 M glycine (pH 2.5) buffer containing 0.1% Triton X-100 and 0.15 M NaCl. Seven 1-ml fractions were collected in tubes containing 100 l of Tris-HCl (pH 9.0). A portion (100 l) of each fraction was chromatographed on paper as described for the synthase assay, and column fractions that contained labeled polysaccharide were pooled.
Preparation of Type 3 Polysaccharide-specific Depolymerase-The type 3 polysaccharide-specific depolymerase was isolated from Bacillus circulans (ATCC 14175) using a modification of a previously described procedure (30). Briefly, a 5-ml culture of B. circulans was grown to mid

FIG. 1. The effect of metal ion concentration on Cps3S activity.
Membranes (3 g of total protein) from type 3 S. pneumoniae were incubated at 32°C for 20 min in a 100-l reaction containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 2 M UDP-Glc (257 mCi/ mmol), 20 M UDP-GlcUA, and increasing amounts of MnCl 2 (q), MgCl 2 (E), or CaCl 2 (f). The product was separated by paper chromatography as described under "Experimental Procedures," and the radioactivity was determined by liquid scintillation counting. The amount of radioactivity in a sample that contained no metal ion was subtracted as background. Each point is the average of duplicate samples. Activity was determined as nmol of 14 C/mg of total protein/h incorporated into polysaccharide. exponential phase in THY, diluted 1/5 in fresh THY, and incubated for 20 min at 37°C. Expression of the depolymerase by B. circulans was induced by the addition of 50 l of heat killed type 3 S. pneumoniae strain WU2, grown to late exponential phase, centrifuged, and concentrated 30-fold in THY. A second inoculum (50 l) of WU2 was added after 60 min, and the B. circulans culture was incubated at 37°C for an additional 60 min. The cells were then heat killed at 65°C for 20 min and sedimented for 2 min at 13,000 ϫ g, and the supernatant was collected and stored at Ϫ70°C.
Preparation of Labeled Polysaccharide for Direction of Growth Experiments-For preparation A, uniformly labeled polysaccharide membranes (300 g of total protein) were incubated in a 200-l reaction containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , 400 M UDP-GlcUA, and 400 M UDP-Glc (7 mCi/mmol) for 30 min at 32°C. The reaction was terminated by the addition of 10% SDS to a final concentration of 2% and chromatographed on Sepharose 2B as described above. For preparation B, terminally labeled polysaccharide membranes (1.5 mg of total protein) were incubated in a 1-ml reaction containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , 400 M UDP-GlcUA, and 400 M UDP-Glc for 30 min at 32°C. The reaction mixture was placed on ice, and 1 ml of ice-cold 100 mM Hepes (pH 7.5) containing 10% glycerol was added. The membranes in the reaction mixture were sedimented by centrifugation at 100,000 ϫ g for 30 min. The pelleted membranes were suspended in 2 ml of 100 mM Hepes (pH 7.5) containing 10% glycerol and sedimented as above. The membranes were washed four more times using this procedure. The washed membranes were incubated in a 1-ml reaction containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , and 2 M UDP-Glc (287 mCi/mmol) for 30 min at 32°C. The reaction was terminated by the addition of 10% SDS to a final concentration of 2% and then chromatographed on Sepharose 2B as described above. For preparation C, low molecular weight terminally labeled polysaccharide membranes (1.5 mg of total protein) were incubated in a 1-ml reaction containing 100 mM Hepes (pH 8), 10 mM sodium thioglycolate, 10 mM MnCl 2 , and 2 M UDP-Glc (287 mCi/mmol) at 32°C for 10 min. The reaction was sonicated for 1 min in a sonicating water bath at 80 watts (Ultrasonics, Plainview, NY) to release the polysaccharide from the membrane. The insoluble material was removed by centrifugation at 100,000 ϫ g for 30 min. The soluble polysaccharide was purified by chromatography on antibody-Sepharose 2B as described above.
Enzyme Treatments-Disaccharide produced by partial acid hydrolysis was treated with Type VIII-A ␤-glucuronidase (27, 800 units/ml) at 37°C for the times indicated. Digestion of polysaccharide with ␤-glucosidase was performed in a 100-l reaction containing 0.1 M sodium acetate (pH 5.0) and ␤-glucosidase (3 units/ml) at 37°C for the times specified. Experiments using the type 3 polysaccharide-specific depolymerase from B. circulans (purification described above) were performed in 20 mM MES buffer (pH 6.5) containing 0.2% sodium azide at 37°C for the length of time specified.
Protein Determinations-Protein concentrations were determined using the micro protein determination kit from Sigma.

Characterization of Type 3 Synthase Activity-Membranes
were isolated from the type 3 S. pneumoniae strain WU2 and from an isogenic derivative (JD908) that is nonencapsulated because of an insertion mutation in cps3S. Synthase activity was determined by measuring the incorporation of 14 C from UDP-[ 14 C]Glc into polysaccharide. Membranes from the parent strain incorporated 15,301 cpm, whereas membranes from the mutant strain, under identical assay conditions, incorporated 31 cpm (data not shown). These data confirmed that Cps3S is the critical enzyme responsible for the activity measured in this assay and that incorporation of label does not occur in the absence of type 3 polysaccharide synthesis.
The activity of the parent type 3 synthase was characterized by measuring the incorporation of 14   that observed when substrates were at equal concentrations.
The synthase was active in the presence of Mn 2ϩ and Mg 2ϩ , with the highest level of activity observed with 5-20 mM Mn 2ϩ (Fig. 1). The optimal pH for the synthase was between 8 and 8.5 in a reaction mixture containing either 10 mM Mn 2ϩ or 10 mM Mg 2ϩ (data not shown). The apparent K m values for both UDP-Glc and UDP-GlcUA were lower in the presence of Mn 2ϩ than with Mg 2ϩ ( Table I).  (Fig. 2B). The presence of Glc and GlcUA in the polymer was confirmed by chromatography in a second solvent containing 1-propanol/ ethyl acetate/water, 7:1:2 (data not shown). Hydrolysis in 1 N HCl of both [ 14 C]Glc-and [ 14 C]GlcUA-labeled polysaccharide resulted in the liberation of both monosaccharides, as well as a product that migrated between 6 and 11 cm from the origin, suggestive of a disaccharide (Fig. 2, A and B). Mild hydrolysis of polysaccharides containing GlcUA readily yields aldobiouronic acids because of the strong resistance of the uronidic linkage to hydrolysis with acids (31). When the putative disaccharide was digested with exo-␤-glucuronidase, 75% of the radioactivity was liberated as free glucose (Fig. 2C). These results confirmed the presence of a ␤-glucuronidic linkage and are consistent with the slower migrating product being the disaccharide GlcUA-␤-Glc.
Additional evidence for the identity of the polysaccharide was obtained by digestion of the high molecular weight polysaccharide with a type 3 polysaccharide-specific depolymerase from B. circulans. This enzyme is highly specific for type 3 polysaccharide and has been shown to hydrolyze the ␤1,4 linkages of this polymer to yield oligosaccharides with an average length of a tetrasaccharide (30). Following treatment with the depolymerase for 48 h, the high molecular weight [ 14 C]Glclabeled polysaccharide was degraded to a lower molecular weight product as determined by chromatography on Sepharose 2B (Fig. 3). Identical results were obtained using [ 14 C]Gl-cUA-labeled polysaccharide (data not shown).
A monoclonal antibody specific for type 3 polysaccharide was coupled to Sepharose 2B and shown to specifically bind the high molecular weight polymer. Columns packed with 0.5 ml of the antibody-conjugated beads bound 99% of both the [ 14 C]Glclabeled and the [ 14 C]GlcUA-labeled high molecular weight products (data shown in Fig. 4 for the Glc-labeled product). The addition of unlabeled type 3 polysaccharide added along with the labeled polysaccharide resulted in a concentration-dependent inhibition of binding of the [ 14 C]Glc-labeled polymer (Fig.  4). No inhibition was observed when unlabeled type 1 polysaccharide was added. All of these results confirmed that the isolated product had the expected properties of authentic type 3 polysaccharide.
Polysaccharide Chain Elongation and Release-Pulse-chase experiments with the type 3 synthase showed that low molecular weight polysaccharide labeled in a 3-min pulse could be chased after 20 min into high molecular weight chains that eluted near the excluded volume of a Sepharose 2B column (Fig. 5). A polysaccharide of intermediate length was observed

FIG. 3. Degradation of high molecular weight product by a type 3 polysaccharide-specific depolymerase.
High molecular weight [ 14 C]Glc-labeled polysaccharide (8400 cpm), synthesized as in Fig. 2, was incubated in 20 mM MES buffer (pH 6.0) with 100 l of the depolymerase preparation for 48 h at 37°C. Untreated (q) and treated (E) samples were chromatographed on Sepharose 2B. The amount of radioactivity present in the even-numbered fractions was determined. A background of 20 cpm was subtracted from each fraction.
FIG. 4. Specific binding of product to a type 3 polysaccharidespecific monoclonal antibody column. High molecular weight [ 14 C]Glc labeled-polysaccharide (2300 cpm), synthesized as in Fig. 2, was mixed with the indicated amounts of either unlabeled type 3 polysaccharide (q) or unlabeled type 1 polysaccharide (E). The samples were applied to affinity columns made with a monoclonal antibody to type 3 polysaccharide, and the percentage of radioactivity that bound to the column was determined as described under "Experimental Procedures." after a 5-min chase. These results are suggestive of a processive biosynthetic mechanism, whereby the elongating polysaccharide chain remains associated with the enzyme-membrane complex.
To further evaluate the mechanism of the synthase reaction, we investigated the association of polysaccharide with the enzyme complex during the course of synthesis. Reaction mixtures containing 100 M UDP-Glc and 200 M UDP-GlcUA were sampled during a 60-min incubation and sedimented by centrifugation to separate the soluble polysaccharide from the membrane-associated polysaccharide. There was a steady increase in the incorporation of [ 14 C]Glc into membrane-associated polysaccharide during the first 30 min of incubation, whereas only a small fraction (12.4%) was incorporated into soluble polysaccharide (Fig. 6A). By 60 min, the fraction of radioactivity found as soluble polysaccharide had increased to 35%, whereas the amount found as membrane-associated polysaccharide had begun to decrease, suggesting that the membrane-associated product might be a precursor to the soluble polymer.
At lower concentrations of UDP-Glc and UDP-GlcUA (2 and 20 M, respectively), the incorporation of Glc into membraneassociated polysaccharide reached a maximum by 5 min and then declined rapidly so that by 30 min approximately equal amounts of label were found in the soluble and membraneassociated fractions. Again, Glc was initially incorporated into membrane-associated polymer, which was subsequently released as a soluble form. The soluble and membrane associated polysaccharides obtained under a variety of conditions were analyzed by Sepharose 2B chromatography. Similar size profiles were observed for both fractions, indicating that release of the polysaccharide from the membrane was independent of the size of the polymer (data not shown).
Although size did not appear to be a factor in polysaccharide release, the increase in the soluble form of polysaccharide in Fig. 6 coincided with the depletion of UDP-Glc. To further examine the effect of substrate concentration on polysaccharide release, a series of reactions were performed as in Fig. 6B above, except that additional substrate was added 5 min after initiating the reaction. As shown in Fig. 7, the simultaneous The amount of radioactivity present as soluble polysaccharide (q), membrane-associated polysaccharide (E), and unincorporated UDP-[ 14 C]Glc (f) was determined after ascending paper chromatography in ethanol/1 M ammonium acetate (pH 7.0), 7:3 as a percentage of the total radioactivity. addition of both substrates prolonged the association of the polysaccharide with the membrane (Fig. 7A), whereas the separate addition of either UDP-Glc or UDP-GlcUA markedly stimulated the appearance of soluble polymer (Fig. 7, B and C). These data suggest that polysaccharide release may be enhanced when one substrate is limiting. Polysaccharide release was not due to the generation of free UDP during polysaccharide synthesis, because the addition of 1 mM UDP did not stimulate release (Fig. 7D).
Direction of Chain Growth-To determine whether sugar addition occurs at the reducing or nonreducing end of the type 3 polysaccharide, polymer was labeled with [ 14 C]Glc either uniformly (preparation A) or on the terminus of the growing end (preparation B). Both methods of preparing polysaccharide resulted in a radioactive product that eluted near the void volume of a Sepharose 2B column as well as a peak of radioactivity that corresponded to UDP-Glc (Fig. 8A). The polysaccharide product obtained by both of the methods was degraded by the type 3 polysaccharide-specific depolymerase (Fig. 8B). Digestion with the depolymerase confirmed that both labeled products were indeed type 3 polysaccharide. Digestion of the terminally labeled polysaccharide with an exo-␤-glucosidase for 24 h liberated 72.8% of the counts as [ 14 C]Glc, whereas uniformly labeled polysaccharide yielded undectable levels of [ 14 C]Glc after digestion with exo-␤-glucosidase (Fig. 8C). The degradation of the terminally labeled polysaccharide with exo-␤-glucosidase was time-dependent (Fig. 8D). Because exo-␤glucosidase removes only the terminal Glc residues from the nonreducing end of the polysaccharide, these data demonstrate that type 3 polysaccharide growth occurs from the nonreducing end.
Terminally labeled polysaccharide was also made by incubating membranes with UDP-[ 14 C]Glc in the absence of UDP-GlcUA and without any initial elongation with unlabeled substrates (preparation C). This procedure yielded labeled polysaccharide that bound to an affinity column made with monoclonal antibodies to type 3 polysaccharide. Exo-␤-glucosidase treatment of polysaccharide eluted off the affinity column released 75.4% of the counts as [ 14 C]Glc (data not shown). This result further supports growth of type 3 polysaccharide from the nonreducing end. Furthermore, the incorporation of Glc in the absence of UDP-GlcUA into a product that bound to a type 3 polysaccharide-specific antibody, supports the presence of preformed type 3 polysaccharide acceptor in S. pneumoniae membranes.

DISCUSSION
The type 3 synthase of S. pneumoniae belongs to a family of processive ␤-glycosyltransferases whose members include the hyaluronic acid synthases from S. pyogenes and X. laevis, the Nod factor synthase from Rhizobium sp., chitin synthases from C. albicans and S. cerevisiae, RfbB-0:54 from Salmonella enterica, and the cellulose synthases from bacteria and plants. All of these enzymes catalyze the formation of all the glycosidic linkages of their respective polysaccharides and are thought to function via a similar mechanism (22,23). The ␤-glycosidic linkages of the polysaccharides synthesized by these enzymes are derived from UDP sugar precursors, which are linked in the ␣ configuration. By analogy with the extensively characterized glycosyl hydrolase systems, these enzymes would be expected to provide for an inverting mechanism during polymer formation (32). A model has recently been proposed for a mechanism of polymerization for the family of processive ␤-glycosyltransferases that would allow for the simultaneous or consecutive formation of two ␤-glycosidic linkages (23). Hydrophobic clustering analysis of the enzymes in this family showed two conserved domains, each of which is believed to be capable of binding a nucleotide sugar and catalyzing the formation of a glycosidic linkage (22,23). Because polysaccharides like chitin, cellulose, and hyaluronic acid adopt a 2-fold screw axis (33)(34)(35), this model proposed that the binding sites for each of the nucleotide sugars are oriented 180°in regard to one another, thus allowing such polysaccharides to be generated without rotating either the enzyme or the polysaccharide (23). The loss of two UDP molecules from the catalytic site after addition of the two monosaccharides has been proposed to provide the necessary energy to translocate the polymer and allow two more nucleotide sugars to bind (23).
We have shown here that the type 3 synthase in S. pneumoniae membrane preparations is optimally active in the presence of Mn 2ϩ . Furthermore we have shown that the apparent K m values for both UDP-Glc and UDP-GlcUA are lower in the presence of Mn 2ϩ than in Mg 2ϩ . In our standard assay, no significant incorporation of Glc from UDP-Glc occurred in the FIG. 7. The effect of substrate on polysaccharide release. Reactions were performed as described for Fig. 6B, except the following additions were made after 5 min of incubation (see arrow). A, 400 M of both UDP-Glc and UDP-GlcUA; B, 400 M UDP-GlcUA; C, 400 M UDP-Glc; D, 1 mM UDP. The radioactivity present as soluble (q) and membrane-associated (E) polysaccharide was determined as described under "Experimental Procedures." The rate of UDP-Glc depletion was similar to that shown in Fig. 6B. absence of UDP-GlcUA, and no incorporation of GlcUA from UDP-GlcUA occurred in the absence of UDP-Glc. These data are in agreement with the expected catalytic properties for the formation of a polymer containing alternating glucosyl and glucuronosyl residues. The product was shown to be composed of Glc and GlcUA, and could be degraded by a type 3 polysaccharide specific-depolymerase. Furthermore, the polymer bound specifically to an affinity column made with a monoclonal antibody to type 3 polysaccharide, and no activity was observed with membranes from strains containing an insertion mutation in cps3S. We confirmed the presence of ␤-linkages by demonstrating sensitivity to both an exo-␤-glucosidase and an exo-␤-glucuronidase. All of these properties are consistent with the product being type 3 polysaccharide.
The type 3 synthase is capable of rapidly forming both glycosidic linkages of type 3 polysaccharide to polymerize UDP-Glc and UDP-GlcUA into high molecular weight polysaccharide. We have shown that the polysaccharide remained associated with the membrane-enzyme complex during elongation, indicating a processive mechanism. Theoretically the addition of Glc and GlcUA to type 3 polysaccharide could occur from either the reducing or nonreducing end. The direction of chain elongation catalyzed by hyaluronan synthases has been debated for a number of years. Stoolmiller and Dorfman (36) demonstrated convincingly in 1969 that the direction of hyaluronic acid chain growth in S. pyogenes occurs by the addition of monosaccharide units to the nonreducing end of endogenous polysaccharide. The results of this investigation were, however, largely obscured by the subsequent finding of dolichol-linked disaccharides and oligosaccharides containing GlcUA and Glc-NAc (37,38) and particularly by the report that hyaluronate chain growth occurs at the reducing end in teratocarcinoma cells (39,40). Correspondingly, the model of Saxena et al. (23) indicated that growth of hyaluronic acid, as well as several other ␤-glycans, occurred from the reducing end. We have shown, however, that approximately 75% of labeled Glc added to the growing end of type 3 polysaccharide chains can be removed by an exo-␤-glucosidase, indicating that type 3 polysaccharide grows from the nonreducing end. Recent reports on the direction of chain growth for cellulose in Cladaphora and Acetobacter (41) and hyaluronate in P. multocida (42) have also indicated that growth occurs from the nonreducing end. Whether all the members of the family of processive ␤-glycosyltransferases synthesize their polysaccharides from the nonreducing end, as suggested by Koyama et al. (41), remains to be determined.
In our experiments determining the direction of growth, we observed that S. pneumoniae membranes contain nascent polysaccharide that can be labeled with Glc from UDP-Glc without the addition of UDP-GlcUA. The presence of type 3 polysaccharide in membrane preparations of S. pneumoniae (3) and our results showing that polysaccharide terminally labeled with Glc specifically binds to an affinity column made with monoclonal antibodies to type 3 polysaccharide indicate that pre-existing type 3 polysaccharide serves as an acceptor in these experiments. Because our membrane preparations contain preformed polysaccharide acceptor, we are unable to determine whether the type 3 synthase is capable of de novo synthesis from the nucleotide sugars or whether some of form of primer is required. However, the recent expression of a related glycosyltransferase in S. cerevisiae (20) could provide a means to answer this question because this organism does not make UDP-GlcUA.
The growth of type 3 polysaccharide at the nonreducing end and the synthesis by a processive mechanism suggests that the growing polymer is bound to the enzyme via a site that recognizes the terminal sugar(s) of the nonreducing end. Examination of the association of type 3 polysaccharide with the membrane-enzyme complex indicated that the interaction of the polysaccharide chain with the enzyme is affected by the UDP sugar concentrations. We found that when both substrates are in excess, the polysaccharide chain remained associated with the membrane-enzyme complex. However, a portion of the polysaccharide was released from the membrane-enzyme complex when the level of one of the substrates became limiting and was stimulated when one UDP substrate was added in excess. That this effect occurred at the point of substrate depletion suggests that the presence of a single substrate may stimulate release. Prehm (40) also observed a slow but distinct shedding of pulselabeled hyaluronate when one obligatory component, such as MgCl 2 , UDP-GlcNAc, or UDP-GlcUA, was omitted during the chase period. This release of hyaluronan was shown to be FIG. 8. Direction of type 3 polysaccharide growth. Polysaccharide was labeled either uniformly (E, preparation A) or terminally (q, preparation B) with [ 14 C]Glc as described under "Experimental Procedures." A, the labeled polysaccharides were chromatographed on Sepharose 2B, and the amount of radioactivity in the even-numbered fractions was determined. B, fractions (18 -30 ml) containing the high molecular weight polysaccharide were pooled and concentrated 10-fold by filtration on Amicon YM10 ultrafiltration membranes followed by ethanol precipitation. An aliqout of each polysaccharide was treated with 100 l of depolymerase for 72 h and then chromatographed on Sepharose 2B. The amount of radioactivity found in the even numbered fractions was determined. C, samples of both polysaccharides were treated for 24 h with 1.5 units of exo-␤-glucosidase. The digestions were chromatographed in ethanol (95%)/1 M ammonium acetate (pH 7.0), 7:3, and the amount of radioactivity present in successive 1 cm strips was determined as a percentage of the total radioactivity present in all the strips. A background of 20 cpm was subtracted from each strip. The location of Glc on the chromatogram is indicated on the graph. D, samples of both polysaccharides were treated with 1.5 units of exo-␤-glucosidase for the times indicated on the graph. The digestions were chromatographed as in C, and the amount of radioactivity present as Glc was determined as a percentage of the total radioactivity found at the origin and as Glc. independent of size (40), an observation that we have also made for type 3 polysaccharide. The selective inhibition of hyaluronate chain release, but not elongation, by p-chloromercuribenzoate prompted the suggestion that the release process might be an enzymatic mechanism (43). Current experiments examining the effect of a single substrate on type 3 polysaccharide release have shown that the release is dependent on time, temperature, and the concentration of the UDP-monomer, 2 also suggesting an enzymatic mechanism.
In summary, the data presented here provide an in-depth characterization of the type 3 synthase involved in the synthesis of pneumococcal capsule. The demonstration of growth of type 3 polysaccharide from the nonreducing end, as well as a potential mechanism of polysaccharide release, provides new information on the mechanism of type 3 polysaccharide synthesis as well as other ␤-glycans synthesized by related enzymes.