INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The capsular polysaccharides of Streptococcus pneumoniae are essential virulence factors that serve to protect the bacterium against opsonophagocytosis. Ninety distinct capsular serotypes, each differing in sugar composition and/or linkages, have been recognized, and the structures of approximately half of these have been determined (32, 56). A given S. pneumoniae isolate expresses only one capsular polysaccharide, and the genetic basis for capsule expression lies in the specific set of biosynthetic genes contained in that strain (reviewed in reference 57). Genes encoding the enzymes uniquely required for synthesis of a specific polysaccharide are linked in the chromosome and are flanked by genes that are common to strains of all capsular serotypes (7, 21-24, 28, 35, 48). The common sequences encode proteins that may be involved in regulation and polysaccharide transport, but none appear to have any role in the actual synthesis of the polysaccharide (28, 35). Genetic exchange of the cassettes containing the capsule genes results in replacement of the recipient's serotype-specific genes with those of the donor and subsequent expression of the donor capsular polysaccharide (7, 22, 23). Hence, all of the genes necessary for production of a given polysaccharide must either be contained in the capsule locus or be a part of the normal S. pneumoniae genetic complement.

Recent molecular characterizations of the capsule loci from several S. pneumoniae serotypes, as well as other streptococci, have found that enzymes expected to be essential for capsule production often are not encoded by genes in these loci. For example, the type 14 locus lacks genes that encode the enzymes necessary to synthesize UDP-glucose (UDP-Glc), UDP-galactose (UDP-Gal), and UDP-N-acetylglucosamine (UDP-GlcNAc), the precursors of the type 14 polysaccharide (36). Similarly, the type 1 capsule contains 2-acetamido-4- amino-2,4,6-trideoxy-D-Gal (AATGal), the type 19F capsule contains Glc, and the Streptococcus pyogenes hyaluronic acid capsule contains GlcNAc, but genes necessary for synthesis of the precursor nucleotide sugars are not present in the respective loci (18, 44, 46). In each of these instances, the sugar or intermediate is an important cellular component (e.g., AATGal in the S. pneumoniae teichoic acids and GlcNAc in peptidoglycan) and its incorporation into the capsular polysaccharide is expected to utilize existing cellular pools. In contrast to these examples, the type 3 capsule locus contains all of the genes expected to be necessary for synthesis of the glucose-glucuronic acid (Glc-GlcUA)-containing polysaccharide. Four type 3-specific genescps3DSUMare transcribed as part of an operon that begins upstream of cps3D and continues downstream past cps3M and through the common flanking gene 'plpA (16). cps3M encodes a putative phosphoglucomutase (PGM), which would convert Glc-6-P to Glc-1-P. This gene is truncated, however, and its function has not been confirmed (16). cps3U (also referred to as cap3C) encodes a Glc-1-P uridylyltransferase that converts Glc-1-P to UDP-Glc, which is then converted to UDP-GlcUA by the UDP-Glc-1-P dehydrogenase encoded by cps3D (also referred to as cap3A) (3, 4, 21, 22). Polymerization of UDP-Glc and UDP-GlcUA to form the type 3 polysaccharide is mediated by the type 3 synthase, encoded by cps3S (also referred to as cap3B) (6, 21). Despite the presence of these four genes in all type 3 loci, mutation analyses have shown that only cps3D and cps3S are absolutely required for capsule synthesis. Mutations in cps3U and cps3M have not been found to alter capsule production (16, 21, 22), and mouse virulence is not reduced in strains in which these genes have been deleted (unpublished data). These results indicate that other genes present in the S. pneumoniae chromosome can complement these mutations. Although such sequences are not apparent through hybridization analyses (16, 21, 22), the cellular Glc-1-P uridylyltransferase has recently been identified and shown to be homologous to cps3U and essential for full capsule production (43). Here, we describe a spontaneous mutant that was identified as the result of reduced type 3 capsule production. We show that the mutation causing this phenotype is located in a gene unlinked to the capsule locus and responsible for encoding the cellular PGM.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and media. The strains and plasmids used in these studies are described in Table 1. Escherichia coli derivatives were grown in L broth, on L agar, or on MacConkey agar with 1% galactose for PGM assays. S. pneumoniae strains were grown in Todd-Hewitt broth (Difco) supplemented with 0.5% yeast extract (Difco) (THY) at 37°C or on Blood Agar Base no. 2 (Difco) supplemented with 3% sheep blood at 37°C in 5% CO₂. For S. pneumoniae, we used erythromycin (EM) at 0.3 µg/ml and kanamycin at 200 µg/ml. For E. coli, we used EM at 275 µg/ml, kanamycin at 25 µg/ml, ampicillin at 100 µg/ml, and tetracycline at 10 µg/ml.

Characterization of morphology and capsule analyses. Buoyant densities, chain lengths, and numbers of cells per colony were determined as previously described (34). Buoyant density values are the mean ± the standard error of the mean from four independent determinations. Quellung reactions and confirmations of capsular serotypes were performed using capsule type-specific antisera (Statens Serum Institut, Copenhagen, Denmark). Capsule quantitation was determined using either an inhibition enzyme-linked immunosorbent assay, performed as previously described (16), or the Stains-All assay for detection of acidic polysaccharides (51). For both assays, cultures were grown to a density of 3 × 10⁸ CFU/ml in THY and the amount of capsule produced was calculated from a standard curve generated using isolated type 3 polysaccharide (American Type Culture Collection).

DNA and RNA techniques. S. pneumoniae was transformed as previously described (60) or by induction with competence-stimulating peptide 1 (30). For the latter, cells grown to a density of 3 × 10⁸ CFU/ml were diluted 1:100 or 1:50 into competence medium (THY supplemented with 0.01% CaCl₂ and 0.2% bovine serum albumin) and synthetically derived competence-stimulating peptide 1 (Zymed Laboratories, San Francisco, Calif.) was added to a final concentration of 500 ng/ml. The cells were incubated at 37°C for 14 min, DNA was added, and the incubation was continued for an additional 4 h prior to plating on selective medium. Sources of DNA were plasmids, restriction or PCR fragments, and S. pneumoniae chromosomal DNA. Plasmids were isolated using either the alkaline lysis method of Birnboim and Doly (12) or Qiagen columns (Qiagen Inc., Valencia, Calif.). Chromosomal DNA from S. pneumoniae was used as crude lysates (60) or purified using Qiagen Genomic Tips. E. coli was transformed by electroporation. Pulsed-field gel electrophoresis was performed using a CHEF-DR II System (Bio-Rad). Cells were grown to a density of 3 × 10⁸ CFU/ml in THY and processed in accordance with the manufacturer's protocol, except that 0.25% sodium deoxycholate was used in the lysis buffer and sodium lauryl sarcosine and lysozyme were omitted. Digests were run on a 1% agarose gel in 0.5× TBE (45 mM Tris-borate, 1 mM EDTA, pH 8.0) at 14°C for 22 h at 6 V/cm. The initial switch time was 1 s, and the final switch time was 20 s. The  DNA size ladder (Bio-Rad) was used as a standard.

Cloning and expression of pgm and cps3M. To obtain the pgm region, DNA surrounding the WG44.6 insertion was cloned and sequenced by a combination of marker rescue and anchored PCR of the chromosomal DNA adjacent to the pVA891 insertion. First, a clone containing a 2-kb insert was obtained by digestion of WG44.6 chromosomal DNA with KpnI; this was followed by self-ligation, transformation into E. coli, and selection on EM. The sequence from one clone, pGH5518, was used to design primer crr-2 in order to obtain a larger clone. For that procedure, WG44.6 chromosomal DNA was digested with SphI and NsiI and ligated into pJY4163 and PCR amplification was performed using the entire ligation reaction as a template. A plasmid-specific primer, TT-2, and crr-2 were used for the PCR. The resulting 1.6-kb PCR product was cloned using the pGEM-T Easy vector system (Promega) and transformation into E. coli with selection for ampicillin resistance. The sequence of one clone, pGH5540, was homologous to that of a periplasmic binding protein for an amino acid transport operon (aatB) and to contig 130 of the S. pneumoniae type 4 genome sequence (The Institute for Genomic Research Website [http://www.tigr.org], 1999). A primer (crr-13) specific for the expected upstream sequence (aatA) was used in conjunction with an aatB-specific primer (crr-9) to generate a 1.2-kb PCR product. This fragment was initially cloned into pGEM-T Easy and then subcloned into pJY4163. Transformation of the subclone (pGH5559) into WU2 and JY1060 resulted in a plasmid insertion in the respective chromosomes. These strains (GH4511 and GH5075, respectively) were then used in marker rescue experiments to clone pgm and the adjacent downstream DNA. The 4-kb clone containing the WU2 pgm region (pGH4045) was obtained by self-ligation of SmaI/PmlI-digested GH4511 chromosomal DNA, followed by transformation into E. coli and selection on EM. A clone (pGH4061) containing part of pgm and 'aatAB of JY1060 was obtained in the same manner using EcoNI/SmaI-digested GH5075 chromosomal DNA. The sequence of JY1060 pgm was obtained from this clone and from PCR sequencing of the chromosome using primers designed from the WU2 sequence. A clone containing all of JY1060 pgm was obtained by replacing the EcoNI/BstXI fragment in pGH4045, which contains all of WU2 pgm, with the same (mutant) fragment from JY1060. Em^r transformants were isolated in E. coli W1485pgm.

Protein analysis and generation of anti-Cps3M antibody. Protein concentrations were determined using the Bio-Rad Protein Assay (Bio-Rad). Proteins were examined by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis and staining with Coomassie brilliant blue dye R-250. The Rainbow protein standard (Amersham) was used to estimate molecular size. Western immunoblotting was performed as previously described (59). PGM-specific mouse polyclonal antiserum was obtained by subcutaneous injection of 10 BALB/cByJ mice with 0.2 ml each of a 1:1 mixture of Freund's incomplete adjuvant and a polyacrylamide gel slice containing recombinant Cps3M. Mice were boosted by intraperitoneal injection of the same mixture after 8 days. Blood was collected 10 days after the boost, and the serum was pooled and absorbed with E. coli by incubation overnight with shaking at 4°C. E. coli cells were removed by centrifugation, and the serum was sterilized by filtration through a 0.45-µm-pore-size filter.

PGM assay. One milliliter of an E. coli culture grown overnight at 37°C was harvested by centrifugation, washed once with imidazole buffer (5 mM imidazole [pH 7.4], 1 mM MgCl₂), and suspended in the original culture volume using the same buffer. Cells were frozen at 70°C, thawed, and sonicated on ice three times for 10 s each time at 30-s intervals. Cell debris was removed by centrifugation at 10,000 × g for 15 min at 4°C. The supernatant was saved, and the protein concentration was determined. The PGM assay was performed by the method of Joshi (33) using 1-ml reaction mixtures containing 40 mM imidazole-HCl [pH 7.8], 2 mM Glc-1-P, 7.9 µM Glc 1,6-diphosphate, 5 mM MgCl₂, 0.5 mM NADP⁺, and 1 U of Glc-6-P dehydrogenase. The A₃₄₀ of samples was read at 30-s intervals for 5 min. Values for controls containing no NADP⁺ were subtracted from each sample, and PGM activity was determined as the change in micromoles of NADPH per minute per milligram of protein.

Nucleotide sequence accession number. The type 3 pgm and flanking sequences have been deposited in GenBank under accession no. AF165218.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation and initial characterization of the type 3 capsule mutant JY1060. JY1060 was isolated during transformation of type 3 strain WU2 with DNA from type 2 strain D39 in an attempt to transfer the type 2 capsule genes into the type 3 background. Four small colonies characteristic of the type 2 morphology were obtained, but none proved to be reactive with antiserum to the type 2 capsule. Instead, two retained reactivity with the type 3-specific antiserum and two were nonencapsulated. The studies described herein involve the characterization of one of the type 3 small-colony mutants, JY1060 (Fig. 1A).

View larger version (38K): [in this window] [in a new window]   FIG. 1.   Colony morphology and capsule production of WU2 and JY1060. (A) Growth at 37°C on blood agar medium. (B) Capsule quantitation. Cultures were grown in THY, and the amounts of capsule contained on washed cells and in filtered culture supernatants (super.) were determined using an inhibition enzyme-linked immunosorbent assay. Values are normalized to the optical density at 600 nm (O.D.) of the starting culture. Each sample was measured in duplicate, and the results are the means ± standard errors from six independent experiments. For both the cells and culture supernatant fluids, JY1060 was significantly different from WU2 (P < 0.0001; unpaired two-tailed t test).

Transcription of the type 3 capsule genes. To determine whether the decrease in capsule production was the result of alterations in transcription of the type 3 capsule locus, Northern and RNA dot blot analyses of the steady-state transcript levels were performed. Northern analysis of the type 3-specific genes revealed the same 6,700-nucleotide (nt) transcript (cps3D through 'plpA) in both the parent WU2 and JY1060 (Fig. 2A). Densitometric analysis of RNA dot blots performed using serial twofold dilutions of total RNA indicated that there was no difference in the amount of steady-state capsule transcripts produced by JY1060 relative to its type 3 parent (Fig. 2B).

View larger version (41K): [in this window] [in a new window]   FIG. 2.   Analysis of type 3 capsule transcripts. (A) Northern blots probed with internal fragments of each type 3-specific gene. Each lane contained 20 µg of total RNA from a culture grown in THY and harvested at an optical density at 600 nm (OD) of 0.05. pspA is a non-capsule-related gene. The probe used is indicated above each pair of samples. The RNA marker sizes (in thousands of nucleotides) are indicated to the left. The type 3 capsule transcript is 6,700 nt, and the pspA transcript is 2,100 nt. Lanes: 1, WU2; 2, JY1060. (B) Densitometry curve derived from dot blots using cps3D to probe dilutions of total RNA for capsule-specific transcripts. Mean peak optical density values for each dilution with the standard error of the mean calculated from three replicates are shown. pspA transcript levels for WU2 and JY1060 also did not differ from each other (data not shown). Symbols: , WU2; , JY1060. (C) Northern blots of type 3 and pspA transcripts throughout growth in THY. Results for WU2 are shown. Identical results were obtained with JY1060. The probe for each blot is indicated at the top. Probing with cps3S, U, or M yielded the same result as shown for cps3D. Each lane contained 20 µg of total RNA. Lanes and culture optical densities at 600 nm: 1, 0.1; 2, 0.2; 3, 0.3; 4, 0.4; 5, 0.7.

Mapping of the JY1060 mutation. (i) Analysis of potential linkage to the capsule locus. Linkage of the JY1060 mutation to the capsule locus was examined by transformation of JY1060 with DNAs from strains that contained Em^r or Km^r insertions within the capsule locus but which produced normal levels of type 3 capsule. The insertions were located either upstream of the type 3 biosynthetic genes in cps3B (KW1004A) or orf5 (JD1008) or within the type 3-specific biosynthetic gene region (JD770). Transformation and selection for the antibiotic resistance marker contained in such strains results in a high frequency of cotransformation of the type 3 capsule locus (22). Transformation of strain Rx1, which makes reduced amounts of capsule due to a point mutation in cps3D (21, 22), resulted in an Em^r-Cps⁺ cotransformation frequency of 86% (269 of 313) when using JD770 donor DNA. In contrast, transfer of the antibiotic resistance markers from JD770, KW1004, or JD1008 into JY1060 resulted in only low-frequency cotransformation of the Cps⁺ phenotype, indicating that the JY1060 mutation was not located in the type 3 locus (Table 3).

(ii) Localization of the JY1060 mutation on the chromosome. To localize the mutation on the chromosome, linkage analysis using strains containing Em^r insertions at various sites in the chromosome was performed (Table 3). Transformation with DNAs from strains WG44.6 and WG44.14 resulted in cotransformation of the Em^r marker and repair of the JY1060 mutation at frequencies of 70 and 83.7%, respectively. Subsequent Southern blot analyses of the insertions in these strains indicated that they were in the same location (data not shown). Transformation and repair of JY1060 with serial dilutions of WG44.6 DNA indicated that only a single mutation, or possibly more than one closely linked mutation, was responsible for the JY1060 phenotype. As the amount of DNA added to the transformation reaction mixture was reduced, the frequency of cotransformation of Em^r and repair of the mutation remained the same. Thus, the cotransformation was due to true genetic linkage and not to saturating levels of DNA (data not shown).

(iii) PFGE mapping of the WG44.6 insertion. To localize the Em^r insertion on the WG44.6 chromosome, pulsed-field gel electrophoresis (PFGE) was performed on SmaI- and ApaI-digested DNA. Distinct band shifts indicated that the insertion was located in SmaI fragment 4 and ApaI fragment 5 of the S. pneumoniae genome. The chromosomal map for S. pneumoniae R6 was previously generated by Gasc et al. (25) using PFGE, and R6 was utilized here as a control since WG44.6 and R6 are both derivatives of type 2 strain D39. The capsule locus is at least 450 kb away from the site of the WG44.6 insertion, on SmaI fragment 3 (5).

(iv) Identification of the JY1060 mutation. A restriction map of the chromosomal region surrounding the Em^r insertion of WG44.6 was generated by Southern analysis, and the ability of WG44.6 restriction fragments containing the insertion and flanking DNA to repair JY1060 was examined. For these experiments, digested chromosomal DNA was transformed into JY1060 and the colony morphology of Em^r transformants was examined. A high frequency of Em^r transformants exhibiting normal colony morphology was obtained using BglII-digested DNA, indicating that the mutation was located within the 10-kb region contained in this restriction fragment. Because WG44.6 was derived from Rx1, a highly passaged laboratory strain descended from a type 2 isolate, further characterization of the region containing the JY1060 mutation was done using GH4511. This strain is a derivative of the type 3 parent that contains an Em^r insertion in the 10-kb BglII fragment. Restriction fragments containing the Em^r insertion and flanking DNA were used in transformation experiments to test for repair of the JY1060 mutation. The results of these experiments are summarized in Fig. 3. Taken together, they suggested that the mutation was located within or adjacent to the EcoNI/BstXI restriction fragment. These results were confirmed in repair experiments using a clone (pGH4045) encompassing the 4-kb region between the Em^r insertion of GH4511 and the PmlI restriction site. As shown in Fig. 4A, repair was obtained only with fragments containing the region between EcoNI and BstXI.

View larger version (14K): [in this window] [in a new window]   FIG. 3.   Mapping and repair of JY1060 mutation by cotransformation with restriction fragments from GH4511. The restriction map was generated by Southern analysis of GH4511 probed with the plasmid used to make the insertion. The 'aatAB sequence was used to target the insertion and is therefore duplicated. Fragments corresponding to single or double digests are shown below the map, and frequency of cotransformation (%Co-Tf) is shown to the right. Gene designations were obtained from sequence analysis of type 3 WU2 or from the type 4 genomic sequence and are based on homology. Designations: PGM, pgm; glycerol facilitator, glpF; muramidase released protein, mrpA; transposase, tnpB; bacteriocin transport-associated protein, bta; amino acid transport protein permease, aatP; amino acid transport protein ATP binding, aatA; amino acid transport binding protein, aatB; Em, point of Emr plasmid insertion. The solid block within tnpB is the site of the 90-bp repeat sequence (see text). Restriction sites: A, AatII; B, BsaHI; Bc, BclI; Bg, BglII; Bs, BstXI; Ec, EcoNI; M, MscI; N, NsiI; Pm, PmlI.

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Localization of JY1060 mutation. (A) Restriction map of pGH4045 and fragments used in repair of JY1060 mutation. PGM functional sites: , active site; , Mg2+-binding site; , substrate-binding site. Amino acid transport (aatAB) conserved sequences: , linker peptide; , Walker B ATP-binding site; , putative transcription terminator; , JY1060 mutation. The arrows above the map indicate directions and lengths of putative transcripts. Each DNA fragment used to transform JY1060 is indicated by a line below the map. Repair of the JY1060 mutation is indicated by a + or  to the right of the fragment. Gene designations and restriction sites are as indicated in Fig. 3. Additional restriction sites: E, EcoRI; P, PstI. (B) DNA and amino acid sequences of the WU2 and JY1060 region encompassing the mutation in pgm. Mutation is shown by the underlined nucleotide and the bold amino acid. Nucleotide numbering is from the beginning of the pGH4045 insertion. Amino acids are numbered from the PGM initiation site.

Identification of the gene affected by the JY1060 mutation. Sequence analysis of pGH4045 revealed five complete or partial open reading frames (ORFs) within the 4-kb insertion (Fig. 4A). The JY1060 mutation was located in a 1,716-nt ORF (pgm) encoding a predicted protein of 572 amino acids with an expected molecular size of 62.7 kDa. A putative promoter containing a 10 region identical to consensus E. coli ⁷⁰ promoters and a 35 sequence containing two mismatches with the consensus E. coli sequence (31) were present upstream of the ORF. The highest observed sequence similarity was with Cps3M, the PGM homologue contained in the S. pneumoniae type 3 capsule locus. At the amino acid level, these sequences were 81% identical, and at the nucleotide level, they were 74% identical. Comparison with the type 4 S. pneumoniae genomic sequence revealed 99% nucleotide and amino acid identity with the sequence contained in that strain (http://www.tigr.org). Sequence similarity for the predicted protein was also observed with PGMs and phosphomannomutases (PMM) from numerous organisms, including the PGMs from E. coli and yeast (20 to 25% amino acid identity and 34 to 44% similarity) and a putative PMM from Bacillus subtilis (45% identity and 61% similarity). The S. pneumoniae sequence contained three conserved functional sites present in each of these proven or putative phosphomutases: a substrate-binding site, an Mg²⁺-binding site, and an active site (Fig. 4A). The JY1060 sequence differed from that of the type 3 parent only at position 381, where the mutation resulted in a lysine changing to a threonine (K381T) (Fig. 4B). This residue is analogous to K-359 of rabbit muscle PGM (49). K359 is not part of the conserved active or binding sites. However, based on the crystal structure of rabbit muscle PGM, it is a surface residue contained within the active-site pocket and it is expected to be important for interaction with the substrate (20, 39). Formation of the active-site pocket involves four domains. The last of these domains is deleted in Cps3M, which truncates immediately after the substrate-binding site (16).

Function of PGM. Recombinant clones expressing Cps3M and the putative S. pneumoniae PGM were constructed to determine if these proteins could function as PGMs. E. coli clones that expressed Cps3M, which is truncated at the C-terminal end relative to other phosphomutases, yielded proteins of the expected size (43 kDa), but no PGM or PMM activity could be demonstrated in either enzymatic or complementation assays (data not shown). Nonetheless, recombinant Cps3M was reactive with anti-yeast PGM antibody, and polyclonal antibody raised against the recombinant Cps3M (anti-Cps3M) reacted with rabbit muscle PGM (Fig. 5A). Sequence analyses indicated that the clones were intact; hence, the lack of activity may be due to the C-terminal truncation. Neither the anti-yeast PGM nor the anti-Cps3M antibody detected Cps3M or PGM expressed in S. pneumoniae (data not shown). In addition, insertional inactivation of cps3M in JY1060 did not further reduce the amount of type 3 capsule produced by this strain (data not shown).

View larger version (43K): [in this window] [in a new window]   FIG. 5.   Recombinant PGM and Cps3M. (A) Expression in E. coli. Whole cultures were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. cps3M expression was induced with IPTG. pgm was expressed from its natural promoter. Lanes 1, 2, 5, and 6 are Coomassie stained, and lanes 3, 4, and 7 to 9 are Western blots reacted with anti-Cps3M antibody. Equal numbers of cells were loaded in all lanes. Lanes: 1, E. coli BL21(DE3)/pET-21a; 2 and 3, MC4033 [recombinant WU2 Cps3M in E. coli BL21(DE3)]; 4, rabbit muscle PGM; 5 and 7, GH4078 (E. coli W1485pgm::tet, vector only); 6 and 8, GH4080 (recombinant WU2 PGM in E. coli W1485pgm::tet); 9, GH4104 (recombinant JY1060 PGM in E. coli W1485pgm::tet). The values on the left are molecular sizes in kilodaltons. (B) PGM assay of E. coli sonicates. +, E. coli 1485 (PGM+ parent strain); , GH4078 (E. coli W1485pgm::tet, vector only); WU2, GH4080 (recombinant WU2 PGM in E. coli W1485pgm::tet); JY1060, GH4104 (recombinant JY1060 PGM in E. coli W1485pgm::tet).

Suppressor mutations and the effect of insertional inactivation of pgm. During transformation reactions with JY1060, we noted an increased frequency of isolates exhibiting nearly normal colony morphology. Under standard growth conditions, such colonies were rarely observed and reversion was estimated to occur in fewer than 1 in 10⁵ cells. Following a transformation reaction, however, large-colony isolates were noted at a frequency of 1.3 × 10³. This number was calculated from 17 independent control transformations of JY1060 to which no DNA was added. A similar frequency was observed in the presence of exogenous DNA. The DNA sequence of the region containing the JY1060 mutation was determined for two such isolates. One (GH5087) was obtained from a no-DNA control transformation, and the other (GH5089) was obtained during transformation with the 350-bp PCR fragment from WU2 that repaired the JY1060 mutation. As shown in Table 4, the JY1060 mutation was not corrected in either of these isolates and the amount of capsule produced was intermediate between those of WU2 and JY1060. Second-site suppressor mutations were also obtained when pgm was insertionally inactivated in WU2. Unlike the JY1060 PGM, which retained its full length and was partially functional, the PGM in the insertion mutants had lost both the C-terminal domain and the substrate-binding site. These mutants exhibited a small-colony morphology. Determination of their true capsule phenotype was complicated, however, by an apparently high frequency of pseudorevertants. Inoculation of THY with cells derived from a single small colony often resulted in as many as 10 to 50% of the isolates exhibiting a large-colony morphology following growth in the liquid medium. The majority (>90%) of these isolates retained the insertion in pgm, indicating the presence of suppressor mutations. Cultures that retained the small-colony phenotype had extended doubling times (at least 150 min, compared to 60 min for WU2 and JY1060), whereas those containing large-colony variants demonstrated normal doubling times after extended incubation. Microscopic examination of THY-cultured cells reacted with type 3 antiserum (Quellung reaction) showed that those retaining the small-colony phenotype uniformly produced a small amount of capsule. In contrast, those containing large-colony variants had a mixed population that contained nonencapsulated cells, as well as ones of minimal and high capsule levels. The amount of capsule produced by four independent pgm insertion mutants (GH4531, GH4532, GH4533, and GH4534) was consistently determined to be less than 30% of that produced by JY1060 and less than 10% of the parental level. Insertions downstream of pgm did not show similar effects; hence, the phenotype is expected to be due to loss of PGM activity.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, we identified a mutation affecting type 3 capsule production that was unlinked to the capsule locus and localized within the gene encoding the cellular PGM. PGM converts Glc-6-P to Glc-1-P, which is needed to funnel glucose into the production of cell wall teichoic acids and capsules that contain glucose. The reverse reaction is also necessary for the metabolism of galactose. The type 3 locus also contains a PGM homologue (Cps3M), but this protein is truncated just past the substrate-binding site. We have not identified any requirement for Cps3M in either capsule production or virulence (16, 21, 22; unpublished data), and from the present study, it is not apparent that it has PGM activity. Thus, the cellular PGM provides most of the enzymatic activity necessary for this step in capsule synthesis. A similar situation exists with Cps3U, the Glc-1-P uridylyltransferase contained in the type 3 locus. Although enzymatic function has been demonstrated (3), like Cps3M, it is not required for capsule production or virulence (21; unpublished data) and Glc-1-P uridylyltransferase activity is derived from cellular GalU (43).

UDP-Glc, the ultimate end product of the PGM and GalU activities, is an essential intermediate in a number of pathways, including biosynthesis of the Glc-containing teichoic acids. In the JY1060 mutant, which contains a partially functional PGM, we did not observe alterations in growth or teichoic acid production (unpublished data). However, the lack of a functional PGM, as obtained with pgm insertion mutations, appeared to place a severe metabolic drain on the cell. In addition to reductions in growth rate and capsule production, second-site mutations that either eliminated or increased the level of capsule production (and presumably other cellular functions) were readily enriched in the population due to their more rapid growth. Although the locations of these mutations have not yet been determined, it is clear that they can occur in sequences outside pgm, as the suppressor mutants maintained the insertions that inactivated pgm. The fact that pgm insertion mutants were readily obtained suggests that this is not an essential gene and that other means of synthesizing Glc-1-P and UDP-Glc are present.

Mechanisms by which type 3 synthesis is regulated are not known. Genes potentially involved in the regulation of other capsule types are mutated in the type 3 locus and are not expected to be functional (3, 16). The high level of DNA sequence identity between pgm and cps3M, along with the fact that these are the most closely related among known sequences, may indicate the occurrence of a gene duplication event and subsequent divergence. The same may also be true for galU and cps3U, which have 72% nucleotide sequence identity (43). The presence of cps3M and cps3U in the type 3 capsule locus could serve to allow distinct regulation of capsule production and to avoid depletion of cellular pools of Glc intermediates under conditions of enhanced capsule synthesis. However, situations in which these genes are needed have not been identified, despite the fact that both are maintained in all of the type 3 strains examined (16, 21). In Neisseria meningitidis, transcription enhances mismatch repair and may thus preserve transcriptionally active regions (37). Such an effect in S. pneumoniae could explain the retention of cps3UM and downstream sequences, which are contained on the cps3DS transcript. Continued capsule production by the pgm insertion mutants, despite its detrimental effect on cell growth, indicates that a low level of precursor molecules also does not serve to completely shut off capsule synthesis.

Our finding of reduced capsule transcripts in late exponential and stationary phases is similar to that reported for the hyaluronic acid capsule of S. pyogenes (19). There, both transcription and capsule production are eliminated through the action of a two component regulatory system which controls gene expression (9, 19, 38). In S. pneumoniae type 3, however, capsule production is not lost and it is not yet clear whether transcription ceases or if the transcripts are more readily degraded. In contrast to pspA transcripts, which were easily isolated and remained stable throughout the growth phases, transcripts from the capsule locus frequently appeared degraded. Other than the cps3DSUM-tnpA-plpA transcript that initiates at a promoter upstream of cps3D, we have not identified specific transcripts or active promoters within the type 3 locus. Thus, the large proportion of type 3-specific RNA found in lower-molecular-weight bands may represent degradation products and effects on both transcription and transcript stability are potential mechanisms involved in the regulation of type 3 capsule production.

During transformation reactions with JY1060, an apparent increase in the frequency of pseudorevertants was observed. Previous studies have reported an increased mutation rate in S. pneumoniae cells undergoing transformation when homologous DNA was present in the reaction mixture (27). However, we noted an increase whether or not DNA was present. It is not clear why PGM mutants would exhibit an increased spontaneous mutation frequency under these conditions, but our results suggest that this is the case. Alternatively, or perhaps in addition, most protein synthesis in competent cells is directed toward proteins involved in this reaction (45) and cell growth is consequently reduced during this time (27). Hence, if the pseudorevertants fail to respond appropriately to competence induction, their continued accelerated growth during this period would give the appearance of an increased frequency of occurrence. In contrast to the type 3 parent, JY1060 cultures increase in cell number during competence induction (unpublished data). Although this increase alone is not sufficient to explain the high frequency of pseudorevertants, it may be one contributing factor.

It is clear that genes outside the capsule locus encode components essential for the synthesis of capsular polysaccharides. Thus, one of the keys to understanding many aspects of capsule production, including its regulation, lies in identifying the basic cellular metabolic pathways to which it is intimately linked. As would be expected for a mutant reduced in capsule production, JY1060 exhibits significantly reduced virulence in mice (unpublished data). The previous finding of sequences required for capsule production and common to the capsule loci of most S. pneumoniae strains, as well as other streptococci and staphylococci, suggested a shared mechanism for capsule synthesis, as well as potential common targets for therapeutic intervention (reviewed in reference 57). Similarly, as suggested for GalU (43), cellular enzymes such as PGM that are required for capsule synthesis provide potential targets for disruption of multiple pathways essential for both cell growth and virulence. However, because second-site suppressor mutations that result in increased levels of capsule occur with high frequency in both pgm and galU mutants (unpublished data), there may be a high likelihood of rapid selection for resistant isolates. The virulence of such mutants and pseudorevertants is thus an important consideration and will be described for the pgm mutants in a separate communication (unpublished data).