Molecular Cloning, Identification, and Sequence of the Hyaluronan Synthase Gene from Group A Streptococcus

The hyaluronan (HA) synthase of Group AStreptococci has been identified by transposon mutagenesis and deletion analysis. The genes for the HA synthase and a recently identified UDP-Glc dehydrogenase (Dougherty, B. A., and van de Rijn, I. (1993) J. Biol. Chem. 268, 7118- 7124) reside on a contiguous stretch of 3.2-kilobase pair DNA that can direct HA biosynthesis in Enterococcus faecalis and Escherichia coli as well as mutant Strep- tococcus (DeAngelis, P. L., Papaconstantinou, J., and Weigel, P. H. (1993) J. Biol. Chem. 268, 14568-14571). The synthase contains 395 residues (calculated M, = 45,063) and migrates on SDS-PAGE with a molecular mass of 42 kDa. E. coli K5, which synthesizes UDP-glucuronic acid for production of its endogenous capsular polysac- charide, can make HA if it contains a plasmid encoding the intact 42-kDa protein. E. coli SURE or x1448 cells containing the same construct, however, cannot produce HA since these strains cannot make both required sugar nucleotide precursors. The HA synthase is pre- dicted to be an integral membrane protein with four membrane-associated helices, which is consistent

ious receptors and binding proteins that modulate cellular behavior such as migration, adhesion, and wound healing (2).
Interestingly, HA is also found in the extracellular capsule of pathogenic Group A a n d C Streptococci (3). The bacterial capsule is a virulence factor that allows evasion of host defenses (4). The enzyme that polymerizes the polysaccharide is HA synthase. Although cell-free biosynthesis of HA was achieved over 3 decades ago (5) and HA synthase activity has been detergent-solubilized from plasma membranes of both eukaryotes and bacteria (6, 71, a functional enzyme has not yet been purified to homogeneity. The synthase is membrane-associated and utilizes UDP-sugar nucleotides at neutral pH in the presence of Mg2+ (5, 6).
A 52-kDa protein from Group C Streptococcus equisimilis was reported to be the synthase since antibody to it inhibited enzyme activity (8). This 52-kDa protein, however, was not shown to be biologically active, and the cloned gene did not confer HA biosynthesis to any organism (9). van

de Rijn and
Drake (10) used [P-32P15-azidoUDP-G1cA, to label proteins of 27,33, and 42 kDa in both GAS and GCS. The incorporation of radiolabel was competed with excess UDP-GlcA, and GAS mutants lacking HA synthase activity did not display the same labeling pattern. Dougherty and van de Rijn recently presented two different models of the putative Streptococcus HA biosynthesis operon based on Tn mutagenesis data. The first model (11) invoked two oppositely translated ORFs (ORF A and ORF B), while the latter (12) described three ORFs ( h a d , ha&, and h a s C ) in the same orientation; only ORF A ( h a d ) was shared between the two models. HasB was shown to be a 402-residue protein, UDP-Glc dehydrogenase, that produces one of the two sugar nucleotide precursors, UDP-GlcA, required for HA biosynthesis (12).
We recently reported that a 3.2-kb region of streptococcal DNA encoding two proteins of 42 and 45 kDa could functionally reconstitute HA biosynthesis in vivo in acapsular GAS mutants, Enterococcus faecalis or Escherichia coli (13). By deletion analysis, we showed that the 42-kDa protein was essential for HA biosynthesis activity. In this report, we identify the 42-kDa protein as HA synthase, the gene product of hasA. We also confirm that the 45-kDa protein is the UDP-Glc dehydrogenase.
EXPERIMENTAL PROCEDURES Materials and Struins-Restriction and DNA modifying enzymes were from Promega unless otherwise noted. All other reagents were of the highest grade available from Sigma unless stated otherwise. Media reagents were from Difco. Cultures to be assayed for HA were grown using the dialysate from dialyzed THY broth (i.e. nutrients <lo-14 kDa). The mucoid GAS strain, S43/192/4, was obtained from the Rockefeller Collection (14). E. coli K5 (Bi8337-41) was obtained from I. Orskov and F. Orskov (Copenhagen, Denmark; Ref. 15). All other strains and plasmids used were described previously (13).

Tn Mutagenesis and Mutant Selection-Tn insertional mutagenesis
was conducted by the method of O'Connor and Cleary (16) except that ovine hyaluronidase (Type V) was added to the GAS culture (0.2 mg/ml, 1 h at 37 "C) after overnight growth and used at a higher concentration (0.1 mg/ml) in the mating plate media. The Tn916 donor, E. faecalis CGllO (17), was mated on nitrocellulose filters (88 mm, 0.45 pm, Micron Separation, Inc.) with strep' S43. The mating mixture was scraped off the filters with 0.4 ml of THY broth containing 1 mg/ml streptomycin and 5 pg/ml tetracycline. The nonmucoid mutant cells were then enriched over Percoll (Pharmacia LKB Biotechnology Inc.) step gradients (13,18). Acapsular (or hyaluronidase-treated) cells pellet through 50%

19181
Percoll, but mucoid cells float at the interface. After overnight outgrowth (50-70 pl of mating mixture/5 ml of double selective media with 5 4 serum in a 15-ml tube), the cultures were underlaid with 2 ml of 50% Percoll in water and centrifuged (3,000 x g for 10 min). The media, the cells at the interface, and most of the Percoll were removed by aspiration, and the cell pellet was then used to inoculate 5 ml of fresh double selective media. Two further rounds of outgrowth for 4-8 h (A6,,,, = 0.2-0.6) and gradient enrichment were performed. Portions of the final cell pellet were streaked on double selective plates containing 5% sheep blood and visually screened for candidate mutants of capsule biosynthesis: those with dry, discrete colonies uersus wild-type wet, spreading colonies. The mutants were streak-purified and verified to be similar to wild-type S43 with respect to vigor, B-hemolysis, DNase secretion (using DNNmethyl green agar), and production of GroupA carbohydrate Wentrescreen, Hycor). Thirteen strains did not have HA synthase activity, produce capsules, or contain HA (13); however, only one strain, S43Tn7, transduced (19) the nonmucoid phenotype.
Tn Mapping and Gene Isolation-Chromosomal DNA purified (19) from the mutants and transductants was cut with HindIII and analyzed by Southern hybridization. After electrophoresis, the agarose gels were dried directly (20) and probed with the Tn-containing EcoRI fragment of pAM118 (17) labeled by random priming (New England Biolabs kit). The hybridization was conducted overnight at 65 "C in 1 x HPB (0.5 M NaCI. 5 mM EDTA, 0.1 M Na2P04, pH 7.2) containing 1% sarcosyl, and the gel was then washed for 40 min with 20 mM "Tis-HCI, pH 8 , a t 22 "C. The 16or 18-kb chimeric Tn-tagged fragments from preparative digests (5-15 pg) of S43Tn7 were isolated from gel slices using Glass-MAX (Life Technologies, Inc.) according to the manufacturer's instructions except that the DNA was eluted from the GlassMAX unit with three sequential additions of water at 65 "C. The Sequenase method (U. S. Biochemical Corp.) for plasmids, with modifications noted below, was employed to sequence the junction at the site of Tn insertion directly from chromosomal DNA. A synthetic oligonucleotide (AAAGTGT-GATAAGTCC) based on the termini of the right arm of Tn9Z6 that reads outward into the interrupted gene was used as the primer (21). The DNA (50-100 ng) was denatured with NaOH, neutralized with sodium acetate, and quickly ethanol-precipitated in the presence of 10 pg of phenoVCH,CI-extracted glycogen. The primer (0.22 pmol) was annealed to the template by slow cooling from 65 to 30 "C. The labeling phase of the reaction was done with Mn2+ buffer, 1:15 diluted labeling mix, and [a-"SlthiodATP (Amersham Corp., 3,000 Cilmmol) for 2 min a t 20 "C. The termination phase was done for 5 min a t 37 "C with extension mix in theA and T reactions (0.6 pl) due to the AfT-rich nature of streptococcal DNA. Gels were electrophoresed, processed (22). and exposed to XAR-5 film for 1-10 days at room temperature. Typically, the sequence of the Tn terminudjunction ( f 3-10 base pairs) and 20-40 bases of the adjacent tagged streptococcal DNA were obtained.
The oligonucleotide derived from the chromosomal sequence determined above was used to screen two A libraries (13) to obtain the intact wild-type DNA, in which the Tn insertion had occurred in the mutant. The phage were adsorbed onto nitrocellulose filters and processed in the typical fashion (22). The filters were hybridized with end-labeled oligonucleotide (HA3; TGGCACAATATGTCAGCCC), in 1.8 x HPB (1 pmol of probe/8 ml) with 1% sarcosyl, 0.54 nonfat milk a t 42 "C for 3 h and washed with 0.5 x HPB at the same temperature for 1 h. The plaques yielding the strongest signal were replated and rescreened twice. Purified phage from a AZAF' library were converted to plasmid form by coinfection of SURE or SOLR cells (Stratagene) with the Exassist helper phage (Stratagene). One clone, pB3, was analyzed by sequencing with Sequenase using the standard protocols. The phage selected from the AGEM library using HA3 were screened with another oligonucleotide (HA16; TATGGCTTAGTGCCATTCG), corresponding to the sequence found near the end of the pB3 insert, in order to obtain DNA adjacent to pB3.
Two positively hybridizing AGEM isolates, which formed small plaques and grew poorly in liquid lysates, were obtained. Large scale plate lysates with top and bottom agarose were needed in order to prepare their DNA (22). The two clones (A1X and A2Y with 20-and 12-kb inserts. respectively) contained the same region of DNA as determined by direct sequencing of the A DNA insert using the Circumvent method (New England Biolabs) and end-labeled HA16 oligonucleotide. The sequence obtained beyond the EcoRI site of pB3 (left site; Ref. 13) was used to make another oligonucleotide (HA17; CAATCATACCAC-CAACTGC), for mapping analysis of the A clones treated with various restriction enzymes. Southern blot analysis showed that a fragment of about 7 kb could be excised from the smaller A2Y clone using the EcoRI site in the S43 DNA and the Sac1 site of the A vector (not shown). A portion of the digest was purified with a Magic minicolumn (Promega), Hyaluronan Synthase Gene and the fragments were ligated to PAT19 shuttle vector (23) digested with EcoRI and SstI (Life Technologies. Inc.). Attempts to subclone the streptococcal fragment in its entirety were thwarted by spontaneous deletions upon transformation into E. coli JM109. After using Epicurean competent SURE cells (Stratagene), using 32 "C for transformation recovery and all further growth, and restriction mapping -70 recombinant colonies, a clone designated pPD41 and containing a 6.6-kb insert was obtained that could complement the HA biosynthesis defect of mutant GAS (13).
Miscellaneous-Polypeptides encoded by plasmid genes were identified by labeling of proteins produced in minicells from E. coli x1448 (24), containing PAT19 alone or various constructs containing 543 DNA, as described previously 13). DNApurification and sequencing, A library production, nested deletion set construction. and HA synthase preparation and assay were performed as described earlier f l 3 ) . Targeted internal deletions were made by digesting pPD41A5 DNA with either EcoRV or PstI, purifying the DNA with Magic minicolumns and recircularizing by ligation. The ligation mixtures were transformed into Electrocompetent SURE cells (Stratagene) and screened for insert size. HA was quantitated using the Pharmacia HA test kit f 13). UDP-Clc dehydrogenase activity was measured as described (12) except cells (from 4.5 ml of overnight cultures in dialyzed THY broth, washed and resuspended in 0.4 ml of buffer) were disrupted by vortexing with an equal volume of washed glass beads (75-150 pm, 5 x 30 s at 4 "C with 30 s on ice between mixing). The extracts were assayed a t 30 "C for a UDP-Glc-dependent increase in A,,, corresponding to NADH production. Protein was measured by the Bradford assay 125) with a bovine serum albumin standard.

RESULTS AND DISCUSSION
The HA synthase gene of GAS was initially identified by Tn insertional mutagenesis. The bacteriophage A25-transducing lysate (19) from one acapsular mutant (designated S43Tn7), which contained two Tn elements, transmitted the nonmucoid phenotype to three out of five transductants (Fig. 1). The nonmucoid transductants did not possess HA synthase activity or a capsule by light microscopy, but the mucoid transductants were equivalent to wild-type S43. HindIII digests of mutant S43Tn7 chromosomal DNA showed two bands migrating a t 16 and 18 kb on agarose gels that corresponded to the higher M, bands detected by a Tn-specific probe on Southern blots of all Tn916 mutants (Fig. 1). These larger species are the result of adding 10 kb of Tn DNA to the S43 HindIII fragment at the insertion site. Since the Tn-tagged DNA from S43Tn7 was well resolved from the other HindIII fragments, it could be gel-purified. The 18-kb chimeric fragment associated with the HA biosynthesis DNA does not hybridize with the probe. All the wild-type IIlndlll fragments detected with ethidium bromide (El3 panel) mivate as c 10 kb ( S ; A HindIII standards in kb). Therefore, the chimeric Tn-tagged fragments (marked with arrows) were purified and sequenced directly. An oligonucleotide probe specific for the HA biosynthetic locus was derived from t h e f r a p e n t marked with the sfar. defect was therefore used directly as a template for sequencing reactions with a Tn-specific primer that reads outward from the Tn terminus and into the interrupted gene. An oligonucleotide (HA3), corresponding to a portion of the sequence of the interrupted gene from the 18-kb chimeric fragment, was used as a hybridization probe for screening wild-type S43 genomic DNA libraries in A phage. An excised AZAF' clone, pB3, containing a 5.5-kb EcoRI fragment was selected and studied further. However, Southern analysis utilizing various oligonucleotide probes to the sequence of pB3 revealed some discrepancies between the wild-type and Tn mutant genomes (e.g. HA16 hybridized to S43 but not S43Tn7, while HA3 hybridized to both; not shown). Therefore, a larger genomic fragment spanning the Tn-induced deletion (13) was obtained from the AGEM library.

Cloning of the Streptococcus Hyaluronan Synthase
After an extensive subcloning effort and subsequent exonuclease I11 deletion, a 3.2-kb fragment of S43 DNA was identified as a locus that could direct HA biosynthesis (13). The sequence of the complementing streptococcal DNA, the insert of pPD41A5, was obtained using both the nested nuclease deletion set with the M13 vector primers and the functional plasmid with custom oligonucleotides. Two major ORFs were present (Fig. 2) in agreement with the earlier minicell analysis (13). The sequence of the first ORF, h a d , reveals the primary structure of a previously undescribed protein (Fig. 3).
The deduced polypeptide contains 395 residues with a M, = 45,063. The 42-kDa protein observed by SDS-PAGE analysis of pPD41A5 minicells is assigned to be HasA because the pPD41A7 plasmid, missing about half of the h a d gene (Fig. 2), does not produce the 42-kDa species (Fig. 4). HasA is predicted to have a PI of 8.2 and to be an integral membrane protein due to four membrane-associated regions (three predicted transmembrane segments).
To identify the role of the two genes on the complementing streptococcal DNA, two constructs were made that substantially truncated either HasA or HasB (Fig. 2). One plasmid, pPDAEcoRV, should produce the intact 45-kDa protein, HasB. The other, pPDAPstI, should make the intact 42-kDa protein, HasA. The pPDAEcoRV construct, in which the truncated h a d gene produced a new 27-kDa species2 (instead of the 42-kDa protein) as determined in minicells (Fig. 41, did not confer the TTTAATGGAAACACAATTTTATTAAAAATATCTCTATATCTAGTTGACATTATTTCTTAT TTATATTATAATATTGAGGTCCTTTCTTTCAAGGAAATTAA"iGAAAGAGGTGTAATT QTQCCTATTTTTAAAAAAACTTTAATTGTTTTATCCTTTATTTTTTTGATATCTATCTTQ 1

CCACATGACTATAAAGTTGCTGCTGTAATTCCTTCTTATAATGAAGATGCCGAGTCATTA P H D Y K V A A V I P S Y N E D A E S L
TTAGAAACACTTAAAAGTGTGTTAGCACAGACCTATCCGTTATCAGAAATTTATATTGTT ability to produce HA in any host (Table I). Minicells containing pPDAPstI produced two nonvector-derived proteins: the intact 42-kDa protein and a 29-kDa truncated version of HasB (Fig.  4). The deleted hasB gene product is predicted to be 23 kDa based on the sequence. When transformed into SURE or x1448 cells, pPDAPstI could not direct HA synthesis (Table I). On the other hand, E. coli K5 transformed with pPDAPstI could produce HA (Table I). This observation should be the result of the endogenous UDP-Glc dehydrogenase, which is responsible for producing UDP-GlcA needed for K5 capsular polysaccharide synthesis, substituting for the nonfunctional streptococcal enzyme. To verify this, we assayed strains with the various plasmid constructs for UDP-Glc dehydrogenase activity (Table I). Indeed, all K5 cultures, including those with vector PAT19 alone, demonstrated this activity. SURE or x1448 cells with plasmids encoding an intact 45-kDa protein possessed elevated enzyme activity, whereas cells with the pPDAPstI plasmid possessed levels similar to host cells alone.
These above results demonstrate that the hasA gene product, the 42-kDa protein, is the HA synthase, and that the 45-kDa protein derived from hasB, is the UDP-Glc dehydrogenase (12). Furthermore, experiments in progress confirm that the 42-kDa protein has both UDP-GlcNAc and UDP-GlcA glycosyltransferase activities. Crude membranes from the various E. coli constructs show HA synthase activity only in cells with the intact hasA gene (not shown). UDP-[l4C1GlcA or UDP-[3H]GlcNAc are incorporated into hyaluronidase-sensitive product only in the presence of UDP-GlcNAc or UDP-GlcA, respectively. This incorporation is decreased by >98% if UDP-GalNAc or UDP-Glc are substituted for UDP-GlcNAc or if UDP-Glc or UDP-GalA are substituted for UDP-GlcA. Dougherty and van de Rijn (12) proposed in their later model that three ORFs (hasA, ha&, and hasC) are involved in HA biosynthesis. We find that the 543 strain hasB is 99.8% identical at the nucleotide level to their GAS strain hasB sequence; there was perfect conservation at the protein level (not shown). The region containing the hasA and hasB genes (12) possesses a restriction map consistent with the two ORFs we find in pPD41A5 (Fig. 2). We find that a putative hasC gene, however, is not present in S43 and is not required for HA capsule biosynthesis.
HA synthase possesses significant homology with the nodC gene product of Rhizobium. NodC is a membrane enzyme that synthesizes chitin-like (poly-p-1,4-GlcNAc backbone) oligomers (27), which is an activity analogous to that of streptococcal HA FIG. 5. Sequence homology of HasA and NodC. The most conserved regions of the two proteins are shown with identical residues in boldface. These residues may be essential for activity. Several conservative substitutions are also present (e.g. D E , W R , or SI").
synthase. NodC possesses several stretches of residues that are identical or similar to the HA synthase (Fig. 5). Overall the two proteins are 30.6% identical. The hydropathy plots of the two proteins are comparable, including three predicted transmembrane segments in the same location near the carboxyl terminus (not shown). Other proteins with homology to HA synthase include DG42 from Xenopus Zaeuis, yeast chitin synthase 11, and an associated protein CSH2 (28). The 52-kDa protein described by Prehm and co-workers (8,9) is not homologous to HasA.
HasA and hasB are the only exogenous genes required to direct HA biosynthesis in most bacteria, due to the presence of one of the sugar nucleotide precursors of HA, UDP-GlcNAc, which is necessary for cell wall formation. In cells that make both UDP-GlcNAc and UDP-GlcA only HA synthase, the gene product of hasA, is needed to polymerize the HA polysaccharide.