Tts, a Processive b -Glucosyltransferase of Streptococcus pneumoniae , Directs the Synthesis of the Branched Type 37 Capsular Polysaccharide in Pneumococcus and Other Gram-positive Species*

The type 37 capsule of Streptococcus pneumoniae is a homopolysaccharide built up from repeating units of [ b - D -Glc p -(1 3 2)]- b - D -Glc p -(1 3 3). The elements govern- ing the expression of the tts gene, coding for the glucosyltransferase involved in the synthesis of the type 37 pneumococcal capsular polysaccharide, have been studied. Primer extension analysis and functional tests demonstrated the presence of four new transcriptional start points upstream of the previously reported tts promoter ( ttsp ). Most interesting, three of these transcriptional start points are located in a RUP element thought to be involved in recombinational events (Oggioni, M. R., and Claverys, J. P. (1999) Microbiology 145, 2647–2653). Transformation experiments using either a recombinant plasmid containing the whole transcriptional unit of tts or chromosomal DNA from a type 37 pneumococcus showed that tts is the only gene required to drive the biosynthesis of a type 37 capsule in S. pneumoniae and other Gram-positive bacteria, namely Streptococcus oralis, Streptococcus gordonii , and Bacillus subtilis . The Tts synthase was overproduced in S. pneumoniae and puri-fied as a membrane-associated enzyme.

Streptococcus pneumoniae (pneumococcus) is an important human pathogen and a common etiological agent of community-acquired pneumonia and meningitis in adults and of acute otitis media in children. The capsular polysaccharide has been identified as the main virulence factor of S. pneumoniae (1). The capsule confers to pneumococcus the advantage to resist phagocytosis and survive in the host. Pneumococcus has evolved by diversifying its capsule, and up to 90 different capsular types synthesizing polysaccharides with different immunological properties and chemical structures have been described (2). Capsular polysaccharide biosynthesis in S. pneumoniae is usually driven by genes located in the cap/cps locus, and the capsular cluster of 13 pneumococcal types has been sequenced recently (3). In remarkable contrast, only a single gene (tts) located far apart from the cap cluster, directs the synthesis of the type 37 capsule (4). Type 37 capsular polysaccharide is the only homopolysaccharide reported in pneumococcus. Clinical isolates belonging to this serotype synthesize a conspicuous capsular envelope that is a branched polysaccharide that has a linear backbone of 33)-␤-D-Glcp-(13 repeating units with monosaccharide side chains of a ␤-D-Glc-(13 linked to C2 of each Glc residue (sophorosyl subunits) (5). Several experimental approaches demonstrated that tts is the only gene required for the synthesis of the type 37-specific capsular polysaccharide in S. pneumoniae. The tts gene encodes a putative glycosyltransferase (Tts) that exhibits significant sequence similarities with cellulose synthases of bacteria and higher plants and other ␤-glycosyltransferases (4).
Only few gene products involved in pneumococcal capsular formation have been biochemically characterized, and almost nothing is known about mechanisms as important as regulation, transport, and assembly of the polysaccharide chain subunits (3). It is generally thought that these polysaccharides are synthesized via lipid-linked repeat unit intermediates because of the biochemical complexity of the repeating oligosaccharide subunit. In types 14 and 19F, the first step of this process involves the activity of the protein coded by cps14(cps19f)E gene (6,7). This protein catalyzes the selective incorporation of Glc from UDP-Glc to a membrane lipid-linked acceptor leading to the formation of a complex where other glycosyltransferases would transfer the sugars present in the polysaccharide repeating subunit (7). However, in type 3 pneumococci, sugars are transferred directly to the growing polysaccharide chain without intervention of an anchoring lipid molecule. We have demonstrated that Cap3B, the type 3 polysaccharide synthase, is the only protein required to synthesize high molecular weight type 3 capsular polysaccharide in S. pneumoniae or Escherichia coli strains provided that UDP-Glc and UDP-GlcUA, the precursors of type 3 capsular monosaccharides, were available (8). It has also been shown that Cap3B (also designated as Cps3S) is a processive enzyme able to transfer alternated residues of Glc and GlcUA from their respective UDP-sugars to the nonreducing end of the nascent polysaccharide chain (9). Cap3B possesses a double ␤-1,3and ␤-1,4-glycosyltransferase activity in contrast to the other glycosyltransferases characterized so far among the enzymes implicated in synthesis of the pneumococcal capsule that only catalyze the transfer of a single glyco-* This work was supported in part by Grants PB96-0809 and BMC2000-1002 from the Dirección General de Investigación Científica y Técnica. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Recipient of a doctoral fellowship from Programa Sectorial de Formación de Profesorado Universitario y Personal Investigador (Ministerio de Educación y Cultura). syl residue (8). There is increasing evidence showing that this property is not so unusual as envisaged previously. Thus, the family of bacterial hyaluronan synthases (HAS) 1 like those of Streptococcus pyogenes (10), Streptococcus equisimilis (11), or Pasteurella multocida (12), and the KfiC enzyme of E. coli responsible for the synthesis of the E. coli K5 capsule (13), also provide examples of a dual enzymatic activity. It should be noted, however, that this enzymatic activity has only been demonstrated for enzymes that catalyze the formation of linear polysaccharides, whereas type 37 polysaccharide is a branched polymer.
We report here the subcellular localization and biochemical characterization of the type 37 synthase in S. pneumoniae strains expressing the tts gene. We also show the ability of Tts to produce a type 37-specific capsule even when expressed in Gram-positive bacteria other than pneumococcus.
To construct pDLP50, chromosomal DNA prepared from the 1235/89 strain was polymerase chain reaction-amplified with oligonucleotide primers D101 and D112 and made blunt-ended with the Klenow fragment of the E. coli DNA polymerase I (PolIk). Subsequently, the DNA fragment was digested with XbaI and ligated to pLSE4 that had previously been digested with SphI, filled in with PolIk, and then treated with XbaI. The ligation mixture was used to transform S. pneumoniae M31, and a clone harboring pDLP50 was isolated by scoring the Ln R transformants for expression of the lytA gene by using a filter technique described previously (24). Plasmids pDLP48 and pDLP49 were constructed as follows: 1235/89 DNA was polymerase chain reaction-amplified with oligonucleotide primers D101 and D116, and the product was digested with either SphI (for pDLP48) or SacI (for pDLP49) and filled in with PolIk. After digestion with ClaI (restriction enzyme target included in the primer D116), the appropriate fragments were ligated to pLSE1 (previously digested with EcoRV and MspI) and used to transform competent M24 cells. Type 37-encapsulated transformants were scored among the Ln R clones, and one clone of each class, i.e., harboring either pDLP48 or pDLP49, was selected.
Preparation of Cell-free Extracts and Tts Enzymatic Activity Measurements-Exponentially growing cultures (1 liter) of S. pneumoniae M24 harboring pLSE1 or pDLP49 were chilled on ice and centrifuged (12,000 ϫ g, 20 min, 4°C), and the pellet was suspended in 10 ml of TMCa buffer (70 mM Tris-HCl, pH 7.0, 9 mM MgCl 2 , 1 mM CaCl 2 ) containing 0.2 mM phenylmethylsulfonyl fluoride (PMSF) and centrifuged again (10,000 ϫ g, 10 min, 4°C). The bacteria, resuspended in the same buffer, were disrupted by two passages of the suspension through a French pressure cell (Aminco). The homogenate was centrifuged at 12,000 ϫ g at 4°C for 15 min, and the supernatant was centrifuged again at 120,000 ϫ g at 4°C for 1 h. The pelleted membranes were homogenized in 2 ml of TMCa buffer containing 0.2 mM PMSF, distributed in 100-l aliquots, and stored at Ϫ80°C. Under these conditions, enzyme activity remained virtually unaltered for up to 1 month. Determination of protein concentration was carried out as described previously (25). Analysis of the membrane fraction for detection of Tts was carried out by 10% SDS-PAGE (26).
Unless otherwise stated, standard reaction mixtures contained 0.5 mg/ml membrane proteins, 30 M (0.1 Ci) UDP-[ 14 C]Glc (specific activity 319 mCi/mmol) (Amersham Pharmacia Biotech) in a 70 mM Tris-HCl, pH 7.0, buffer containing 9 mM MgCl 2 , 1 mM CaCl 2 , and 50 mM NaCl in a total volume of 100 l. The reactions, carried out at 30°C for 15 min, were terminated by the addition of SDS (0.5% final concentration) and were incubated at 37°C for 15 min. Afterwards, bovine serum albumin (Sigma) at a final concentration of 0.4% and 1 ml of 10% trichloroacetic acid were added. After incubation for 30 min at 0°C, the mixtures were passed through Whatman GF/A filters and extensively washed with 10% trichloroacetic acid. The filters were dried (65°C, 20 min) and counted in a 1219 Rackbeta scintillation counter (LKB Wallack). One unit of Tts activity is expressed as the amount of enzyme that catalyzed the incorporation into a macromolecular product of 1 pmol of Glc/mg of protein/min.
Identification of the Reaction Product of the Tts Synthase-The total volume of a standard reaction mixture carried out as described above was treated with SDS and filtered through a Sepharose CL-4B column (20 ϫ 1.5 cm; Amersham Pharmacia Biotech). The products of the reaction were eluted with 20 mM Tris-HCl, pH 7.5, buffer containing 0.2 M NaCl; 0.5-ml fractions were collected, and alternate fractions were counted. The high molecular weight fractions that eluted at V 0 were pooled, dialyzed into water, lyophilized, dissolved in 2.5 M trifluoroacetic acid, and subjected to hydrolysis for 2.5 h at 120°C. Then the samples were analyzed by HPLC as indicated below or subjected to thin layer chromatography (TLC) after being repeatedly dissolved and lyophilized. The dried pellet was dissolved in 40% 2-propanol containing 5 mg/ml unlabeled carrier Glc and Gal. TLC was carried out on HPTLC silica gel 60 plates (Merck), impregnated with phosphate, and activated as described by Hansen (27) but using the solvent system 2-propanol, acetone, 0.1 M formic acid (2:2:1) (28). To visualize unlabeled sugar standards, the TLC plate was sprayed with 5% H 2 SO 4 in ethanol and heated to 100°C for 10 -30 min. The regions that contain the unlabeled sugar standards were scraped, added to water, and counted in a liquid scintillation counter.
The radioactive fractions containing the unincorporated sugar nucleotide precursors that eluted at V T were treated with 10 mM HCl at 100°C for 10 min and neutralized with NaOH. Both excluded and retained fractions were then analyzed by HPLC by using an Aminex HPX-87H column (300 ϫ 7.8 mm; Bio-Rad) and eluted at 30°C with 125 M H 2 SO 4 at 0.25 or 0.4 ml/min (see below). The elution of authentic samples of Glc and Gal was monitored with an in-line 132 refractive index detector (Gilson).
Miscellaneous Techniques-Type antisera purchased from the Statens Seruminstitut (Denmark) were used for immunological analyses. As a potential competitor in immunoprecipitation assays, we used curdlan, a linear (133)-␤-D-glucan from Alcaligenes faecalis (Sigma). This polysaccharide was suspended in water (10 mg/ml) with a glass homogenizer and centrifuged (12,000 ϫ g, 30 min, 4°C), and the insoluble pellet was discarded. The sugar content of the solution was determined by using the anthrone reagent (29). Typing by the capsular reaction (Quellung) was kindly carried out by L. Vicioso (Spanish Pneumococcal Reference Laboratory, Majadahonda, Spain). The standard assay conditions for the pneumococcal LytA amidase and the preparation of [ 3 H]choline-labeled pneumococcal cell walls have been described elsewhere (21). One unit of LytA amidase activity was defined as the 1 The abbreviations used are: HAS, HA synthase(s); Ery, erythromycin; GalUA, galacturonic acid; HA, hyaluronan, hyaluronate, or hyaluronic acid; HPLC, high performance liquid chromatography; Ln, lincomycin; ME, 2-mercaptoethanol; PAGE, polyacrylamide gel electrophoresis; pHMB, p-hydroxymercuribenzoate; PMSF, phenylmethylsulfonyl fluoride; PolIk, Klenow (large) fragment of the Escherichia coli DNA polymerase I; ttsp, promoter of the tts gene; [ ], plasmid-carrier state. amount of enzyme that catalyzed the hydrolysis (solubilization) of 1 g of pneumococcal cell wall material in 10 min.

RESULTS
Transcriptional Analysis of the tts Gene-We have reported previously the identification of the tts promoter and its transcription start point (4). The ttsp promoter contains a Ϫ10 consensus sequence with an extended TtTG motif characteristic of the Ϫ16 region of S. pneumoniae (30) and transcription initiates 9 nucleotides after the Ϫ10 consensus sequence (4). However, we have now observed that unencapsulated pneumococcal cells transformed with a recombinant plasmid (pDLP49) containing the region upstream of ttsp formed colonies noticeably more mucous than those from cells transformed with pDLP48, an equivalent plasmid that only contains ttsp and the structural tts gene. To determine the promoter strength of both constructs, we compared the cell wall lytic activity (see "Experimental Procedures") expressed in a pneumococcal ⌬lytA strain (M31) transformed either with pDLP36 (4) (Fig. 1A), which contains the reporter lytA gene under the control of ttsp, or with pDLP50, a construct that also includes the upstream region of ttsp (Fig. 1A). Sonicated cell extracts prepared from M31 [pDLP50] showed 6 times more LytA activity than those from M31 [pDLP36] (Fig. 1B). In addition, M31 [pDLP50] exhibited a faster autolysis at the end of exponential phase of growth than M31 [pDLP36] (Fig. 1C). Furthermore, primer extension analysis using total RNA extracted from M31 [pDLP50] revealed the presence of at least four additional transcription start points upstream of ttsp (Fig. 2). Interestingly, three of them lie in a RUP element present in this position in the clinical type 37 strains (4) (Fig. 3). RUP elements are thought to be insertion sequence derivatives that facilitate recombinational events (31), but a promoter activity had never been described in these elements.
Expression of Tts in other Gram-positive Bacteria-According to the results described above, we used pDLP49 to transform competent cells of S. pneumoniae M24, S. oralis NCTC 11427, S. gordonii V288, and B. subtilis YB886. Ln R (or Ery R ) transformants were isolated, and selected colonies were grown in broth to test for the production of type 37 capsule. In every case, expression of tts led to agglutination of the bacterial cells when incubated in the presence of type 37-specific antiserum (Fig. 4). Immunoagglutination never occurred either when the same strains were incubated with non-type 37 antiserum or when the recipient strains harbored the vector plasmid pLSE1 and received anti-type 37 serum. These results demonstrated that only tts is required for the synthesis of a type 37 capsular polysaccharide in several Gram-positive species. Furthermore, the above immunoagglutination test using whole cells indicated that the capsular material is, at least in part, linked to the outer bacterial surface.
To determine whether a single copy of the tts gene was also sufficient to direct capsule formation in a heterologous host, we transformed competent cells of S. oralis with chromosomal DNA from the pneumococcal strain C2, a type 37 transformant carrying a single tts copy linked to the ermC resistance marker. S. oralis Ln R transformants agglutinated in the presence of type 37-specific antiserum (Fig. 4I) demonstrated that it was possible to transfer tts to this related species and that its presence in a single copy also leads to the production of detectable amounts of a capsular polysaccharide immunologically indistinguishable from the pneumococcal type 37 strains.
Subcellular Localization of the Tts Activity-To prepare a homologous system for biochemical assays, we used the type 37 pneumococcal strain M24 [pDLP49] described above. Subcellular fractions of M24 [pDLP49] were tested for incorporation of radioactivity into a macromolecular product by using UDP-[ 14 C]Glc, assuming that UDP-Glc was the natural substrate for Tts. The membrane fraction turned out to incorporate the label, whereas the soluble fraction did not (data not shown). SDS-PAGE analysis of a membrane preparation from M24 [pDLP49] revealed the presence of an overproduced protein with a molecular mass of ϳ50 kDa (Fig. 5). This protein was absent in membranes prepared from M24 [pLSE1], a strain harboring ) and pLSE4 (pDLP36 and pDLP50) to study the expression of tts. Pertinent restriction sites and oligonucleotide primers (black triangles) employed to construct the corresponding recombinant plasmids are described under "Experimental Procedures." The black and white box corresponds to the location of ttsp. B, lytic activity of sonicated extracts prepared from the ⌬lytA pneumococcal strain M31 harboring different plasmids assayed on [ 3 H]choline-labeled pneumococcal cell walls. ND, not detectable. C, growth (and lysis) curves of the S. pneumoniae M31 strain harboring plasmids pLSE4 (q), pDLP36 (E), or pDLP50 (ϫ). Cells were incubated in CϩY medium containing Ln (0.7 g/ml), and growth was followed by nephelometry (N). One N unit corresponds to about 2 ϫ 10 6 colony-forming units/ml. only the vector plasmid. Another protein band migrating faster than that of Tts could also be occasionally observed, and it might have been originated by proteolysis of Tts, although PMSF was used during the preparation of the membrane fraction.
Biochemical Properties of the Type 37 Synthase-Membranes of the pneumococcal M24 [pDLP49] strain were used to evaluate the incorporation of [ 14 C]Glc from its precursor UDP-[ 14 C]Glc into a macromolecular product using different experimental conditions. Membranes prepared from S. pneumoniae M24 [pLSE1] cells were employed as a negative control. Tts activity was stimulated in the presence of 10 mM MgCl 2 or MnCl 2 . Moreover, 10 mM EDTA completely inhibited the reaction (Table I). However, Ca 2ϩ ions stimulated only slightly Tts activity when added at low concentration (1 mM) in the absence of Mg 2ϩ (data not shown). Furthermore, EGTA only produced a small inhibition of the reaction (Table I). Globally, this behavior is similar to that already described for several glycosyltransferases like cellulose synthases, HAS, or the pneumococcal type 3-specific synthase. In addition, 50 mM NaCl increased 2-fold the incorporation of [ 14 C]Glc into a macromolecular product (data not shown). Other important properties of Tts are reported in the composite Fig. 6. The Tts synthase exhibited a noticeable pH dependence, and the optimal activity was achieved between 6.8 and 7.5 (Fig. 6A). Formation of the radiolabeled macromolecular product of Tts was proportional to protein concentration and proceeded linearly with time for up to 15 min and then slowed down (Fig. 6, B and C). The enzymatic activity reached a maximum when the reaction was carried out at 30°C in the presence of the substrate UDP-[ 14 C]Glc. Tts was relatively stable when incubated at 0°C for up to 60 min, but its activity drastically decreased when preincubation was carried out at 25°C or higher temperatures (Fig. 6D).
We have also assayed the effect of sugars or sugar nucleotides on the Tts synthase activity (Table I). A strong inhibition occurred when membranes were preincubated in the presence of UTP, UDP, UMP, or TMP, whereas different sugars (Glc, Gal, GalUA, GlcUA, or Ara) did not affect noticeably the later incorporation of radiolabeled Glc. UDP-sugars like UDP-Gal, UDP-Xyl, and UDP-Man completely inhibited the reaction, whereas CDP-Glc, GDP-Glc, GMP, or CMP only exhibited a moderate inhibitory effect. These results suggest that the nucleotide moiety of the substrate, and not the sugar one, would play an important role in binding and/or activity. On the other hand, detergents were found to be powerful inhibitors of Tts FIG. 6. Some biochemical properties of the Tts synthase. A, pH dependence of the enzymatic activity of Tts. The following buffers were used: 70 mM Tris/maleic NaOH (q), 70 mM sodium phosphate (E), 70 mM Tris-HCl (f), 70 mM glycine-NaOH (OE). 14 C incorporation assays were carried out as described under "Experimental Procedures." Effect of protein concentration (B) and incubation time (C) on Tts activity. Thermal stability (D) was studied by preincubating the membranes at the indicated temperatures before adding the substrate. Aliquots were withdrawn at different times, and Tts activity was assayed as described under "Experimental Procedures." The data represent the amount of product synthesized during the assay period.  (Table I) suggesting a close association between Tts and the cell membrane. Interestingly, titration of the Tts synthase with p-hydroxymercuribenzoate (pHMB) resulted in a complete loss of enzymatic activity that could be partially prevented by addition of 2-mercaptoethanol (ME) ( Table I) indicating that there might be sulfhydryl groups implicated in the folding of the protein, in its enzymatic activity, or both. Finally, bacitracin added at concentrations of 1 or 100 g/ml to the reaction mixture did not inhibit the reaction (Table I), strongly suggesting that a lipid intermediate is not involved in the biosynthesis of the type 37 capsular polysaccharide of S. pneumoniae. Effect of UDP-Gal on the Enzymatic Activity of Tts-As reported above (Table I) UDP-Gal is a potent inhibitor of Tts synthase. Moreover, we have shown that Tts shares conserved motifs with cellulose synthases and other ␤-glucosyltransferases (4) that are presumably implicated in substrate binding (UDP-Glc) (32,33). These motifs might be specific for UDP-Glc, although we cannot rule out the possibility that they only recognize the nucleotide part of the molecule, as already suggested for the mechanism of action of this family of enzymes (34,35). If this were the case, it might account for the inhibitory effect found when adding UTP, UDP, or UMP to the reaction mixture (Table I). It is also conceivable that UDP-Gal (and perhaps any other UDP-sugar showing an inhibitory effect) may serve as substrate of the Tts synthase for polysaccharide biosynthesis. Interestingly, we were able to detect the formation of a radiolabeled high molecular weight product by gel filtration through a Sepharose CL-4B column in experiments where either UDP-[ 14 C]Glc or UDP-[ 14 C]Gal was used as substrate (Fig. 7A). The macromolecular product(s) of these reactions that eluted in the V 0 of the column was immunoprecipitated with an anti-type 37 polysaccharide serum but not with a heterologous antiserum directed against type 3 pneumococci (Table II). Interestingly, curdlan, a linear (133)-␤-D-glucan, did not preclude the recognition of the type 37 polysaccharide by its antiserum. These results demonstrated that the Ttscontaining pneumococcal membranes are capable of incorporating the 14 C label to a polymer immunologically indistinguishable from type 37 polysaccharide using either UDP-[ 14 C]Glc or UDP-[ 14 C]Gal as substrate.
Characterization of the Polysaccharide Product of Tts Synthase-As shown above, the polymer(s) synthesized by using either UDP-[ 14 C]Glc or UDP-[ 14 C]Gal as substrate eluted in the void volume of a Sepharose CL-4B column, whereas nonincorporated radioactive UDP-sugars appeared in the V T (Fig.  7A). The excluded fractions were pooled and hydrolyzed with 2.5 M trifluoroacetic acid as described under "Experimental Procedures," and the samples were analyzed by HPLC. In addition, fractions containing the non-incorporated UDP-[ 14 C]sugars were hydrolyzed with 10 mM HCl, neutralized, and also subjected to HPLC analysis. The radioactivity found in the excluded, hydrolyzed fractions co-eluted with a Glc standard solution irrespectively of the labeled precursor used in the reaction (Fig. 7B). Identical results were obtained when the same fractions were analyzed by TLC; that is, radioactivity was detected only in the spot corresponding to Glc using either UDP-[ 14 C]Glc or UDP-[ 14 C]Gal as substrate (not shown). These results confirmed that, in both cases, Tts synthesized a polymer composed exclusively of Glc. These findings imply that UDP-[ 14 C]Gal must be epimerized to UDP-[ 14 C]Glc before incorporation into the nascent polysaccharide chain. Some authors (7,36,37) had suggested the presence of a strong UDP-Glc-4Ј-epimerase activity associated with the membrane fraction of S. pneumoniae belonging to various capsular types that did not include type 37. Here we show that this is also the case for type 37 pneumococcal membranes as fully confirmed by HPLC analysis of the hydrolyzed UDP-sugars obtained from the fractions eluted at the V T of the Sepharose CL-4B column (Fig. 7C). Independently of the radiolabeled precursor used in the assay, the presence of the pneumococcal membranes promoted the appearance of both epimers, UDP-[ 14 C]Glc and UDP-[ 14 C]Gal. DISCUSSION We have recently reported that a single gene (tts) located outside of the cap/cps locus drives the synthesis of the capsular polysaccharide in type 37 pneumococci (4). We have now found that transcription of the tts gene also initiates at four different points located upstream of the previously reported promoter ttsp ( Figs. 1 and 2). It is important to point out that three of the additional transcription start points are located inside a RUP element (Fig. 3). Several features of RUPs led to the proposal that these small (107 base pairs long) intergenic elements could be trans-mobilized by the transposase of IS630-Spn1 insertion sequence (31) and possibly promote sequence rearrangements (4,38). If this were the case, the presence of a RUP element upstream of the structural tts gene might represent a regulatory mechanism for capsule expression since transposition (or inversion) of the RUP element should lead to a variable expression of the capsular polysaccharide in type 37 pneumococci during infection. In addition, the finding that promoter activity is associated with RUP elements may have other potentially interesting implications in the physiology of this microorganism. Since up to 108 copies of this intergenic element are distributed all along the pneumococcal genome (31), it is conceivable that they could contribute to the regulation of virulence (and non-virulence) genes. Interestingly, besides the type 37 tts locus, RUP elements have been found close to genes coding for several important pathogenicity factors of S. pneumoniae such as capsular polysaccharides, neuraminidases, the hyaluronidase, etc. (31).
A type 37 capsule was immunologically detected when several Gram-positive species were transformed with a recombinant plasmid (pDLP49) harboring the type 37 S. pneumoniae tts gene (Fig. 4). This finding demonstrates that Tts is sufficient for capsular synthesis in heterologous systems. Furthermore, a single copy of the tts gene inserted into the chromosome of S. oralis also led to capsule formation providing the first example where a polysaccharide capsule has been described in this species. This result illustrates how the commensal S. oralis might acquire the capacity to synthesize this important virulence factor in the nasopharynx, the natural habitat where many streptococci live. Similar DNA interchanges have already been reported for other pneumococcal genes, e.g. the spread of resistance to ␤-lactam antibiotics has been attributed to horizontal transfer events involving fragments of the genes coding for penicillin-binding protein(s) of pneumococcus and other related streptococcal species (39). Moreover, compelling evidence for recombination events between the galU gene of S. pneumoniae and that of several streptococcal species has also been provided recently (40).
Hydropathy analysis of Tts predicted six potential transmembrane domains and a central cytoplasmic region presumably containing the catalytic site(s) (residues 64 -346) (4). We show here that when the tts gene was overexpressed in S. pneumoniae, an ϳ50-kDa active protein was found to be associated with the membrane fraction (Fig. 5). Furthermore, both ionic and non-ionic detergents drastically affect the Tts synthase activity associated with these membranes (Table I). The M r of the overproduced Tts deduced from SDS-PAGE analysis was smaller than that predicted from sequence analysis (ϳ59 kDa), which might be due to an anomalous migration of the protein as it has been already reported for two streptococcal HAS (10,11). Tts contains five Cys residues presumably located in the cytoplasmic loop (residues at positions 105, 114, 262, 278, and 299), and one more (Cys-470) between the potential transmembrane regions V and VI. Since ME did not noticeably affect the enzymatic activity of Tts (Table I), it can be assumed that those Cys residues are not forming disulfide bonds. However, Cys residues appear to be necessary or important for Tts activity since a complete inhibition of the enzyme was obtained upon titration with the sulfhydryl-reactive agent pHMB (Table I).
The pneumococcal membranes containing Tts incorporate [ 14 C]Glc from UDP-[ 14 C]Glc into a polymer immunologically indistinguishable from that of type 37 clinical strains (Table  II). It should be emphasized that, although immunological cross-reactions have been reported among several anti-pneumococcal diagnostic sera (41), the type 37 antiserum appears to be very specific since it only recognizes the homologous polysaccharide. The only cross-reactivity reported for the type 37 capsule is a slight precipitin reaction between this polysaccharide and an antiserum raised against pneumococci of serogroup 12 (41). Types 12F and 12A contains branches of kojibiosyl residues (42). More recently, the sophorosyl unit has been demonstrated to be the main immunological determinant of type 37 capsular polysaccharide by quantitative hapten inhibition studies (43). Other disaccharides of the isomeric series of ␣and ␤-(132), -(133), -(134), and -(136) were poorly active as competitive inhibitors of antibody precipitation (43). Here we have shown (Table II) that when a linear (133)-␤-D-glucan (curdlan) was employed, no inhibition of the immunoprecipitation reaction was observed (Table II), which fully confirmed that the anti-type 37 serum preferentially recognizes the branched part of the type 37 polysaccharide. Since the type 37 polysaccharide contains two different ␤-glucosidic bonds (␤-1,3 and ␤-1,2), Tts should be responsible for the formation of both linkages according to our findings that tts is the only gene required for a type 37 capsule synthesis. The polysaccharide synthesized by Tts was composed exclusively by Glc, as revealed by HPLC analysis (Fig. 7) and TLC (not shown). Combined similar HPLC and TLC analyses and immunological tests revealed that when UDP-Gal was used in vitro as substrate, the polymer synthesized was indistinguishable from that formed by using UDP-Glc. This finding implies the presence of an epimerase that converts UDP-Gal to UDP-Glc (Fig. 7C).
Computer analyses have revealed that ␤-glycosyltransferases share conserved sequences and structural features (34). The processive transferases contain a D(X) 40 -130 D(X) 90 -140 D(X) [30][31][32][33][34][35][36][37][38][39][40] QXXRW motif distributed over two domains, named "A" and "B," whereas nonprocessive enzymes lack domain B, and so have only the first two Asp residues of the motif (34,44). Both domains have also been identified in Tts since this enzyme contains the conserved motif D(X) 53 D(X) 88 D(X) 36 RXXKW (4). A classification of glycosyltransferases using nucleotide diphospho-sugars, nucleotide monophospho-sugars, and sugar phosphates (EC 2.4.1.x), and related proteins into 48 distinct sequence-based families has been proposed (45). Tts belongs to family 2 that includes, among other inverting glycosyltransferases, cellulose synthases, HAS, and ␤-1,3-glucan synthases. Although the HAS from P. multocida is currently a member of this family, it appears to be structurally distinct from other HAS (46). Experimental evidence for the role of carboxyl residues in ␤-glycan synthases comes from site-directed mutagenesis of chitin synthase 2 from Saccharomyces cerevisiae (47) and of the AcsAB cellulose synthase from Acetobacter xylinum (44) as well as from the use of amino acid-modifying reagents on a ␤-(1,3)-glucan synthase from ryegrass (48). Based on these and other results it was assumed that Asp residues are involved in the acid-base catalytic mechanism of this kind of glycosyltransferases (49). Nevertheless, the recent elucidation of the three-dimensional crystal structure of SpsA, a member of family 2 of glycosyltransferases implicated in the synthesis of the mature spore coat of B. subtilis, has allowed us to shed light on the mechanisms of this ubiquitous family of inverting glycosyltransferases (50). It has been found that the invariant Asp residues of domain A are intimately involved with UDP binding, whereas a candidate for the general base has not been identified with certainty. It should be noted, however, that the glycosyltransferase specificity of SpsA has not been characterized as yet and that this enzyme lacks the domain B characteristic of the processive transferases. Nevertheless, the observed inhibitory effects of UDP, UTP, UMP, or UDP-sugars on Tts activity (Table I) are in agreement with the involvement of the conserved Asp residues in binding to the nucleotide rather than to the sugar moiety of the UDP-sugar substrate.
To the best of our knowledge, the Tts synthase, which catalyzes both ␤-1,2 and ␤-1,3 linkages, is the first inverting glucosyltransferase able to synthesize a branched polysaccharide. Perhaps the most intriguing characteristic of Tts is that it shares sequence similarities with other enzymes that produced various types of linear polymers, either homo-or heteropolysaccharides. Therefore, only hypothetical models can be proposed for polymerization of the type 37 polysaccharide of S. pneumoniae. The formation of ␤-1,3and ␤-1,2glycosidic bonds may occur either simultaneously or consecutively. Mutational studies should be carried out in the future to determine whether the synthesis of a curdlan-like polysaccharide (␤-1,3-glucan) precedes that of sophorose or formation of ␤-1,2and ␤-1,3bonds takes place simultaneously as the polysaccharide chain grows.
Our results suggest that a lipid-linked intermediate of the type that participates in the synthesis of O-antigen and peptidoglycan is not required for type 37 capsular polysaccharide biosynthesis since bacitracin did not inhibit the Tts activity (Table I). Bacitracin inhibits the dephosphorylation of undecaprenyl pyrophosphate by forming a complex with the lipid (51); this dephosphorylation step is required to regenerate undecaprenyl pyrophosphate, the lipid carrier in peptidoglycan biosynthesis. A bacitracin-independent pathway had also been demonstrated, among others, for the synthesis of group A HAS (52) and chitin oligosaccharides from Mesorhizobium loti (53).
The observation that the tts is the only pneumococcal gene required for the synthesis of a type 37 capsule in different Gram-positive species strongly suggests that the nascent polysaccharide chain does not use specific transporters to cross the membrane. The presence of only four potential transmembrane regions at the C-terminal half and two more at the N terminus of Tts (4) might suggest that the formation by Tts synthase of a membrane pore to facilitate the extrusion of the polymer is unlikely since the presence of at least 12 transmembrane helices are apparently required to build a channel in other sugar transporters (54). However, it should be mentioned that in most of these transporters only four to eight transmembrane helices are usually involved in sugar transport since there are actually two pores, a sugar and a cation pore (55). Evidence suggesting that in the HAS from S. pyogenes only four transmembrane domains and two membrane-associated re-gions that, however, do not appear to traverse the cell membrane are required to create a pore-like structure through which a nascent HA chain can be extruded to the cell exterior has been reported recently (56). However, several alternative mechanisms might allow the transport of the hydrophilic type 37 polysaccharide across the membrane both in pneumococcus and in other Gram-positive species. These include the use of unspecific transporters, association of several Tts monomers in the membrane to conform a pore, or interaction of the synthase with membrane phospholipids, as recently proposed for HA transport (57). Additional efforts using the experimental tools developed in this work are required to determine if the current models for polymerization and transport of linear polysaccharides can be applied to the synthesis of the branched structure of type 37 polysaccharide that represents the most simplified strategy developed by pneumococcus to synthesize its main virulence factor.