Inactivation of the srtA Gene Affects Localization of Surface Proteins and Decreases Adhesion of Streptococcus pneumoniae to Human Pharyngeal Cells In Vitro

Abstract

Inactivation of sortase gene srtA in Streptococcus pneumoniae strain R6 caused the release of β-galactosidase and neuraminidase A (NanA) from the cell wall into the surrounding medium. Both of these surface proteins contain the LPXTG motif in the C-terminal domain. Complementation with plasmid-borne srtA reversed protein release. Deletion of murM, a gene involved in the branching of pneumococcal peptidoglycan, also caused partial release of β-galactosidase, suggesting preferential attachment of the protein to branched muropeptides in the cell wall. Inactivation of srtA caused decreased adherence to human pharyngeal cells in vitro but had no effect on the virulence of a capsular type III strain of S. pneumoniae in the mouse intraperitoneal model. The observations suggest that—as in other gram-positive bacteria—sortase-dependent display of proteins occurs in S. pneumoniae and that some of these proteins may be involved in colonization of the human host.

Many surface proteins of gram-positive bacteria are covalently anchored to the cell wall by a mechanism requiring a C-terminal anchoring motif, which consists of a conserved amino acid sequence, LPXTG (where “X” is any amino acid), that is followed by a hydrophobic domain and, in most cases, a tail of positively charged residues (11, 32). Using protein A of Staphylococcus aureus as a model, Hangovan et al. (20), Mazmanian et al. (25), and Schneewind et al. (33) determined the function of a membrane-localized cysteine protease, named sortase, which cleaves LPXTG of protein A between threonine and glycine and catalyzes the transfer of processed protein to the free amino group of pentaglycine cross bridges in the staphylococcal peptidoglycan (20, 25, 33). Sortase A mutants of S. aureus show multiple defects in pathogenesis (22, 33).

In contrast to what is seen with S. aureus, most virulence related pneumococcal proteins studied so far have been shown to be attached to the cell surface through the choline residues of teichoic acids, and pneumococci have been described as a paradigm for the display of virulence proteins through specific but noncovalent associations with the cell surface (9). The initial purpose of the experiments described here was to test whether proteins covalently linked to the pneumococcal cell wall through a sortase-dependent process may also exist in Streptococcus pneumoniae, as in S. aureus and other gram-positive bacteria. The observations described demonstrate the covalent attachment of at least two proteins to the pneumococcal surface via a sortase-dependent reaction.

MATERIALS AND METHODS

Strains and bacterial growth.

The bacterial strains and plasmids used in this study are listed in Table 1. Strains of S. pneumoniae were grown in a casein-based semisynthetic medium (C+Y medium) at 37°C without aeration as described previously (15). S. pneumoniae strains containing plasmid pLS578 (24) or its derivatives were grown in the presence of 1 μg of tetracycline/ml. Escherichia coli strains containing plasmid pJDC9 (8) or its derivatives were grown in Luria-Bertani (LB) medium in the presence of 1 mg of erythromycin (Sigma)/ml. Strains of S. pneumoniae containing pJDC9 were grown in C+Y medium in the presence of 1 μg of erythromycin/ml.

TABLE 1.

Bacterial strains and plasmids used in this study

S. pneumoniae strain or plasmid	Description	Reference or source
Strains
R36A	Penicillin-susceptible laboratory strain	1
R6	Penicillin-susceptible laboratory strain derived from R36A	39
Pen6	Penicillin-resistant transformant of R6Hex with donor DNA from penicillin-resistant strain 8249	39
Hun663 tr4	D-Ala-D-Ala branches in cell wall muropeptide	34
Hun663 tr4.tr5	Derivative of Hun663 tr4 with cell wall branches like those in R6Hex	34
R36srtA	srtA mutant of R36A	This study
R6srtA	srtA mutant of R6Hex	This study
PensrtA	srtA mutant of Pen6	This study
Hun663 tr4srtA	srtA mutant of Hun663 tr4	This study
Hun663 tr4.tr5srtA	srtA mutant of Hun663 tr4.tr5	This study
Pen6ΔmurM	Pen6 with murM deleted and replaced by erm gene from pJDC 9	10
R36ASIII	Type 3 encapsulated derivative of R36A	36
R36ASIIIsrtA	srtA mutant of R36ASIII	This study
Plasmids
pJDC9	E. coli plasmid Ermr	8
pLS578	S. pneumoniae plasmid Tetr	24
pSRTAID1	Plasmid pJDC9 carrying an internal DNA fragment of srtA	This study
pAKSRTAR64	Plasmid pLS578 carrying the srtA gene of R6	This study
pAKJDCK	Kanr derivative of pJDC9	This study
pAKSRTBT	Plasmid pAKJDCK carrying an internal DNA fragment of srtB of strain TIGR4	This study
pAKSRTDT	Plasmid pJDC9 carrying an internal DNA fragment of srtD of strain TIGR4	This study

DNA manipulations.

All routine DNA manipulations were performed by using standard methods (30). Chromosomal DNA was isolated from 5-ml cultures at an optical density at 590 nm (OD₅₉₀) of 0.6. After centrifugation, the bacterial pellet was resuspended in 0.5 ml of 1× SSC (0.15 M NaCl plus 0.015 M sodium citrate) solution containing 0.1% sodium dodecyl sulfate and 100 μg of RNase/ml, and the suspension was incubated at 37°C for 45 min. The lysed suspension was extracted twice with phenol-chloroform and once with chloroform-isoamyl alcohol. The aqueous phase was precipitated with isopropanol, and the dried DNA was resuspended in 50 μl of Tris-EDTA (pH 8.0). Plasmids were isolated as described previously (30), and PCR products were purified by using a Wizard PCR Preps DNA purification system (Promega). Oligonucleotides were purchased from GIBCO/BRL Life Technologies, and DNA sequencing was done at The Rockefeller University Protein/DNA Technology Center. Nucleotide and derived amino acid sequences were analyzed by using DNASTAR software.

srtA insertion duplication mutant and construction of a plasmid with the srtA gene.

An srtA mutant with the genetic background of strain R36A was constructed by an insertion duplication strategy. An internal 490-bp sequence of srtA was PCR amplified from R36A genomic DNA by using primers srtA-ForID-RI (CACGAATTCGTGTCACACGTTTGACTTCAC) and srtA-RevID-Bam (GCAGGATCCCTGATATTACTGTCACTGGC), which were designed on the basis of the strain R6 genome nucleotide sequence (italic type in sequences indicates restriction sites for EcoRI and BamHI, respectively). The product was restricted with EcoRI and BamHI, and the restricted and purified PCR product was ligated to pJDC9, which was also restricted with the same enzymes and purified. The ligation mixture was transformed into competent E. coli cells plated on LB agar plates containing 1 mg of erythromycin/ml. Transformants were confirmed to contain the desired size insert by plasmid restriction analysis. One of these clones releasing an insert of the desired size was sequenced and used to transform S. pneumoniae strain R36A. Transformants were selected on blood agar plates (BAP) supplemented with 1 μg of erythromycin/ml. Genomic DNA from strain R36srtA was transformed into competent cells of strains R6, Pen6, Hun663 tr4, and Hun663 tr4.tr5. Transformants were selected on BAP containing 1 μg of erythromycin/ml.

For complementation studies, genomic DNA corresponding to a 770-bp sequence lacking the native promoter, encompassing the srtA open reading frame (ORF), and preceded by a potential Shine-Dalgarno sequence was PCR amplified by using primers SrtAgRn-Bam (TGGGATCCAAAACGAAACAGAAGGTGAAGC) and SrtAgnF-HIII (TCAAAGCTTTGTATTAATAAAATTGTTTATATG), which were designed on the basis of the R6 genome nucleotide sequence (italic type in sequences indicates restriction sites for BamHI and HindIII, respectively). The PCR product and vector plasmid pLS578 were restricted with BamHI and HindIII and purified. The purified linear fragments of the vector and the insert were ligated by using a rapid ligation kit (Roche). The ligation mixture was transformed into S. pneumoniae strain R36A. Clones containing the insert were selected for tetracycline resistance (marker for pLS578) and confirmed by PCR and determination of the nucleotide sequence. One clone, pAKSRTAR64, was used in complementation studies of srtA in various genetic backgrounds.

β-Galactosidase assay.

β-Galactosidase activity was measured essentially as described by Zahner and Hakenbeck (38) with minor modifications. In brief, srtA⁺ and srtA mutant strains R36A, R6, Pen6, Hun663 tr4, Hun663 tr4.tr5, and Pen6ΔmurM were grown in C+Y medium with or without erythromycin, depending on the genetic background. Cells were grown to the logarithmic phase; growth was arrested by chilling on ice for 15 min. One milliliter of the culture was used to directly estimate total enzyme activity as follows. Triton X-100 (0.1%) was added directly to 1 ml of bacterial culture, and the mixture was incubated at 37°C for 10 to 15 min to lyse the cells. Another 1 ml of the culture was first centrifuged, the supernatant was decanted, and the pelleted bacteria were lysed by resuspension in 100 μl of sodium phosphate buffer (pH 7.0) containing 0.1% Triton X-100. After lysis, the volume of the lysate was adjusted to 1.0 ml by the addition of 900 μl of buffer (38). For the measurement of β-galactosidase activity, 0.2 ml of o-nitrophenyl-β-d-galactopyranoside solution (4 mg/ml) in phosphate buffer was added to both the lysed pellet and the supernatant. After incubation at 30°C, the reaction was arrested by the addition of 0.5 ml of 1 M Na₂CO₃ solution. Cultures were centrifuged, and enzyme activity in the supernatant was measured by determination of the OD₄₂₀. Units of β-galactosidase were calculated as described previously (38), and the percentages of β-galactosidase activity associated with the cells and released into the culture medium were calculated. All experiments were repeated at least three times. The means of enzyme units from three experiments were used to calculate the percentages of β-galactosidase activity in all three fractions: total culture, pellet, and culture supernatant.

Localization of NanA by Western blotting.

Total cell-associated proteins and proteins released into the culture medium were tested for neuraminidase A (NanA). Bacterial pellets obtained after centrifugation were resuspended and lysed in 15 mM Tris (pH 8.0) containing 0.1% sodium dodecyl sulfate. The supernatant was filtered, deoxycholate solution (final concentration, 0.02% [vol/vol]) was added and, after incubation at room temperature for 10 min, trichloroacetic acid (final concentration, 8% [vol/vol]) was added; the mixture was incubated for 15 min on ice. Precipitated proteins were centrifuged, washed with acetone, and redissolved in 15 mM Tris (pH 8.0); the pH was adjusted to 8.0; and the mixture was used for the preparation of blots. Equal volumes of pellet and supernatant proteins were loaded and separated by polyacrylamide gel electrophoresis, after which the proteins were transferred to membranes (Amersham) by electroblotting. After prehybridization with 3% skim milk powder in phosphate-buffered saline (PBS)-Tween buffer, the membranes were hybridized in PBS-Tween buffer containing skim milk powder and 1:1,000-diluted NanA-specific antibodies. The immunoblots were hybridized to secondary antibodies, and NanA distribution was analyzed after the blots were processed.

Adherence to and invasion of human pharyngeal tissue cell line Detroit 562 by S. pneumoniae.

Human pharyngeal cell line Detroit 562 (27) was inoculated into 24-well tissue culture plates. In order to allow growth, the plates were maintained at 37°C in 5% CO₂-95% air with RPMI 1640 medium without phenol red but supplied with 1 mM sodium pyruvate and 10% fetal bovine serum (FBS). Wild-type and srtA mutant cultures of S. pneumoniae strains R36A, R6, and Pen6 grown in C+Y medium to the logarithmic phase were diluted to the appropriate density (5 × 10⁶ to 1 × 10⁷ CFU/ml) in RPMI 1640 medium with 1% FBS, and 1-ml aliquots were inoculated into washed Detroit 562 monolayers in the 24-well tissue culture plates. Both the wild type and the srtA mutant grew in RPMI 1640 medium (doubling time, about 60 min) during the incubation time used for the assay. After incubation for 2 h at 37°C in 5% CO₂, the culture fluid was removed from each well and the monolayers were washed three times with PBS (pH 7.4). Pharyngeal cells were then detached from the plates by treatment with 200 μl of 0.25% trypsin-0.1% EDTA prepared in PBS (pH 7.2). Pharyngeal cells were next lysed by the addition of 800 μl of 0.025% Triton X-100, and appropriate dilutions were plated on tryptic soy agar BAP to count the numbers of bacteria adherent to and/or internalized by the pharyngeal cells. In a typical experiment with strain R6, out of the 5 × 10⁷ CFU added to a well, 2 × 10⁴ to 3 × 10⁴ bacteria were adherent and/or internalized.

For the invasion assay, monolayers were initially treated as for the adherence assay, but following the attachment of bacteria to the monolayers, 1 ml of RPMI 1640 medium with 1% FBS and containing penicillin (10 μg) and gentamicin (200 μg) was added to the monolayers in order to kill bacterial cells attached to the surfaces of pharyngeal cells. The plates were incubated for 1 h at 37°C in 5% CO₂. After this step, the monolayers were washed three times with PBS (pH 7.4) to remove the antibiotics, pharyngeal cells were released and lysed, and 100-μl aliquots of the lysates were plated on tryptic soy agar BAP to determine the numbers of bacteria internalized. In a typical experiment with strain R6, out of the 5 × 10⁷ CFU added to a well, 1.5 × 10³ to 2 × 10³ bacteria were internalized.

All experiments were run in triplicate and were repeated three to five times. The means of three experiments were used to characterize the adherence or invasion capacity of a strain.

Intraperitoneal mouse virulence model.

Strains R36ASIII and R36ASIIIsrtA were grown in a medium composed of 80% tryptic soy broth, 5% glucose, 5% Difco yeast extract, and 10% C+Y medium (and supplied with 1 μg of erythromycin for the srtA strain). Cultures were grown to an OD₅₉₀ of 0.4, centrifuged cells were resuspended in pyrogen-free 0.9% sodium chloride solution, and dilutions were prepared. Five groups (10 mice in each group) of 8-week-old CD1 mice received 500, 5 × 10³, 5 × 10⁴, 5 × 10⁵, and 5 × 10⁶ CFU of bacteria, injected into the peritoneal cavity. The survival of the mice was monitored for 10 days.

Construction of srtA, srtB, and srtD knockout mutants in the background of S. pneumoniae strain TIGR4.

Mutations were constructed by the insertion duplication strategy. Internal fragments of srtB (536 bp) and srtD (469 bp) were PCR amplified on TIGR4 genomic DNA by using primers srtBF (CCGCTGCAGTCTCGCTTGTATTATCGAG) and srtBR (CAGCAAGCTTACATAATCATGACCTGG) for srtB and primers srtDF (CCAAGTTTGGAATTCATGGAGCCGG) and srtDR (CACGGATCCACCATTGAGAGGTTGCAACAC) for srtD (italic type in sequences indicates the corresponding restriction sites). Purified PCR products were restricted with PstI-HindIII for srtB and EcoRI-BamHI for srtD, followed by purification. Vector pAKJDCK, a kanamycin-resistant derivative of pJDC9, was restricted with HindIII-PstI, and vector pJDC9 was restricted with EcoRI-BamHI. The srtB fragment was ligated to pAKJDCK, whereas the srtD fragment was ligated to pJDC9. Transformants were selected on LB agar plates supplied with kanamycin (35 μg/ml) and erythromycin (1 mg/ml) for srtB and srtD ligations, respectively. Clones pAKSRTBT and pAKSRTDT were confirmed both with restriction endonucleases and by determining the DNA nucleotide sequences of the inserts. Plasmids pAKSRTBT and pAKSRTDT were transformed into competent TIGR4 cells to obtain TIGRsrtB and TIGRsrtD mutants by using selection based on resistance to kanamycin (200 μg/ml) and erythromycin (1 μg/ml), respectively. The presence of mutations was confirmed by PCR amplification and DNA nucleotide sequence analysis.

In order to obtain srtA mutations in TIGR4, R36srtA genomic DNA was transformed into competent TIGR4 cells, and transformant TIGRsrtA was selected on BAP supplied with 1 μg of erythromycin/ml.

RESULTS

Sortase gene homologues in the genomic databases for S. pneumoniae strains R6 and TIGR4.

The genome of S. pneumoniae strain R6 contains a solitary ORF, annotated as putative sortase gene srtA (19). A nonrandom nucleotide Blast search made at http://www.ncbi.nlm.nih.gov indicated that srtA of R6 does not bear a high degree of homology to any of the sortase genes available in databases, with the exception of an ORF (SP1218) in S. pneumoniae strain TIGR4 which was annotated as a conserved hypothetical protein (CHP) gene and which was 99% identical to a substantial region of srtA of strain R6. In addition to the CHP gene, the TIGR4 genome contained a cluster of genes which were annotated as sortase-like protein genes (35) and which were recently renamed srtB, srtC, and srtD (17, 20).

It has been documented that sortases of several gram-positive bacterial species are involved in the sorting of virulence factors and that mutation of the srtA gene reduces virulence (3, 4, 13). In order to better define SrtA of S. pneumoniae strain R6, we aligned the R6 SrtA protein (consisting of 247 amino acids) with available sortase proteins from the R6 genome page at http://www.ncbi.nlm.nih.gov. Figures 1 and 2 demonstrate the degree of identity of R6 SrtA to CHP in S. pneumoniae strain TIGR4 as well as to sortases from different gram-positive bacteria, such as Streptococcus gordonii, Streptococcus suis, and Streptococcus pyogenes (Fig. 2). The TIGR4 CHP showed 99% identity to SrtA of strain R6 (Fig. 1); therefore, we propose to refer to it as SrtA of strain TIGR4. Surprisingly, the degree of identity to S. aureus SrtA was only 18% (Fig. 1). This finding contrasts with data for several streptococcal sortases, in which homology to R6 SrtA extended throughout the entire protein sequence (data not shown). A dendrogram constructed on the basis of sequence similarities indicated that the srtA genes of strains R6 and TIGR4 of S. pneumoniae may be considered members of a streptococcal sortase gene family (Fig. 2). In contrast, the sortase-like proteins SrtB, SrtC, and SrtD of strain TIGR4 are only distantly related to the streptococcal sortase protein family (Fig. 2).

FIG. 1.

Clustal V alignment of the SrtA sequences of S. aureus strain Mu50 (S.au.) and S. pneumoniae strain R6 and the CHP of S. pneumoniae strain TIGR4. Nonidentical amino acid residues are indicated in red. Broken lines indicate gaps.

FIG. 2.

Phylogenetic tree constructed from the sortase protein sequences identified in various streptococcal species and in S. aureus strain Mu50. The SrtA protein sequences of S. suis, S. gordonii, S. pyogenes, S. pneumoniae strain R6, ORF SP1218 of S. pneumoniae strain TIGR4, and S. aureus strain Mu50 as well as the SrtB, SrtC, and SrtD protein sequences of S. pneumoniae strain TIGR4 were aligned for homology by using Clustal V and megaline DNASTAR software.

Organization of the srtA gene and surface proteins containing the LPXTG motif in R6.

An analysis of downloaded databases revealed that at least one of the srtA or srtA homologues in S. gordonii, S. pyogenes, Streptococcus mutans, and Streptococcus equi is adjacent to gyrA. The closeness of gyrA and srtA was already noted for S. suis (26), and examination of the R6 genome revealed a similar structural organization for srtA-gyrA (19). For R6, the SrtA coding region starts at the third base (in italic type) of the termination codon for GyrA (TAATG), and the srtA ORF is preceded by a potential Shine-Dalgarno sequence which is included in the GyrA C-terminal coding region of the gyrA ORF, raising the possibility that these two genes may be cotranscribed.

A genome-wide search of the R6 chromosome (19) identified as many as 23 pneumococcal proteins containing the LPXTG motif whose genes are distributed over the entire genome and which include β-galactosidase and NanA. Two-thirds of these 23 proteins contain the LPXTG motif in their C-terminal regions and may therefore be involved in sortase-catalyzed processing.

Effect of srtA inactivation on the localization of β-galactosidase.

In order to test the functions of the single putative sortase in strain R6, we constructed a mutation in the srtA gene by insertion duplication mutagenesis (see Materials and Methods).

The srtA⁺ and srtA mutant strains R6 and Pen6 were grown to the logarithmic phase, and the fractions of β-galactosidase activity associated with the cells and/or released into the medium were determined. There was no detectable autolysis in any of the strains during the galactosidase assay. The results presented in Fig. 3 show that inactivation of the srtA gene in each of the strains caused a striking change in the localization of the enzyme: while 60 to 90% of the enzyme activity was cell associated in parental cells, between 60 and 90% of the enzyme was released into the growth medium in srtA mutants. The Z test performed on these data indicated that the differences observed in the β-galactosidase distributions were statistically significant (P < 0.005). The effect of the srtA mutation on the distribution of enzyme activity was particularly striking for strain Pen6.

FIG. 3.

Effect of srtA mutation on the localization of β-galactosidase at the surface of S. pneumoniae. Filled bars indicate cell pellet-associated enzyme activity; empty bars indicate activity in the culture medium. S, presence of plasmid pAKSRTAR64 carrying the srtA gene for complementation; P, presence of plasmid vector pLS578 without the srtA insert (control for complementation). Enzyme activity is expressed as a percentage of the total recovered in association with the bacteria and in the culture medium. Error bars indicate standard deviations.

In order to prove that the enhanced release of β-galactosidase into the culture supernatant was the consequence of the srtA mutation, we sought to introduce a functional copy of the srtA gene on plasmid pAKSRTAR64 into the srtA mutants. The results presented in Fig. 3 show that a functional srtA gene in trans can prevent the enhanced release of β-galactosidase from the cell surface of srtA mutants. Introduction of the vector alone had no effect (Fig. 3).

Effect of srtA inactivation on the localization of NanA.

NanA is one of the pneumococcal virulence factors (6) which contains the LPXTG motif. The distribution of NanA was studied with strains Pen6 and PensrtA by separating cell-associated proteins and proteins released into the culture medium and determining the relative amounts of NanA by Western blotting with antibodies specific for NanA. While NanA was detectable both in cell-associated form and in the growth medium, the inactivation of srtA caused most of NanA to be released into the growth medium (Fig. 4). Complementation with pAKSRTAR64 restored the normal distribution of NanA (data not shown). While the Western blotting method allowed the detection of NanA, the method was not appropriate for a more precise quantitation of this protein in cell fractions.

FIG. 4.

Effect of srtA mutation on the localization of NanA in S. pneumoniae. Bacterial cultures were centrifuged, and the pelleted cells (lanes 1 and 3) and their cell-free culture media (lanes 2 and 4) were assayed for NanA by Western blotting as described in Materials and Methods.

Branched muropeptides as preferential sites of attachment for β-galactosidase.

In S. aureus, sortase-processed proteins are attached to the amino-terminal glycine residues of branched muropeptides (25). The results presented in Fig. 3 suggest that the distribution of β-galactosidase in S. pneumoniae may also be influenced by the proportion of branched muropeptides in the cell wall. For strain Pen6, which has a highly branched cell wall, less than 10% of the enzyme activity was found in the medium of normally growing cells. In contrast, up to 40% of the enzyme activity could be detected in the culture medium of strain R36A, which has a substantially smaller proportion of branched muropeptides in its cell wall (10). In order to test the possible role of branched muropeptides in the localization of β-galactosidase, we compared strain Pen6 and its isogenic derivative Pen6ΔmurM, from which the murM gene, critical for cell wall branching, was deleted. Figure 5 shows that the proportion of β-galactosidase released into the medium was increased in Pen6ΔmurM to 23% from about 9%, a proportion which is characteristic of parental Pen6 cultures. Analysis of data with the Z test indicated that these differences in β-galacosidase distributions were highly significant.

FIG. 5.

Effect of srtA mutation and deletion of murM on the localization of β-galactosidase in strains of S. pneumoniae carrying different types of branched peptides in their cell walls. The distributions of enzyme activities associated with the bacteria and released into the culture medium are reported as described in the legend to Fig. 3. Error bars indicate standard deviations.

In order to investigate whether there was a bias for the chemical type of cell wall branch in the attachment of β-galactosidase, srtA mutants were constructed in the backgrounds of strains Pen6 and Hun663 tr4.tr5 (both strains contain high proportions of primarily seryl-alanine dipeptide branches in their cell walls) and strain Hun663 tr4 (which contains a high proportion of primarily alanyl-alanine branches in its cell wall) (10, 34). The localization of β-galactosidase in these strains and their respective srtA mutants was determined. No evidence for preferential attachment to either type of branch was apparent (Fig. 5).

Effect of srtA inactivation on adherence and invasion of a human pharyngeal cell line.

Strains R36A, R6, and Pen6 and their srtA mutants were added at several cell concentrations to monolayers of pharyngeal cell line Detroit 562 (27) grown in 24-well tissue culture plates. The numbers of bacterial cells attached to and/or internalized by the human cell line were estimated after a set of controlled washes, resuspension of the pharyngeal cells, lysis with dilute detergent, and plating of the bacteria for viable titers. The results presented in Fig. 6 show a 50 to 87% reduction in the adherence of strains with inactivated srtA compared to the respective wild-type strains.

FIG. 6.

Effect of srtA mutation on the adherence of S. pneumoniae to human pharyngeal cells. S. pneumoniae bacteria were mixed with human pharyngeal cells of the Detroit 562 cell line, and the fraction of bacteria that remained associated with the pharyngeal cells was estimated as described in Materials and Methods. Filled bars indicate adherence by the wild-type strains; empty bars indicate adherence by the corresponding srtA mutant strains. Data were normalized so that the adherence of the wild-type cells was expressed as 100%. Error bars indicate standard deviations.

The fraction of bacteria internalized by the monolayers was also determined in the same experiments by a more direct method. Bacteria attached to pharyngeal cells were treated with penicillin and gentamicin in order to kill extracellular (but not internalized) pneumococci. After washes to remove the antibiotics, pharyngeal cells were detached and lysed, and the number of internalized bacteria was determined. Strains with inactivated srtA showed a 75 to 90% reduction in invasion compared to isogenic wild-type strains (Fig. 7).

FIG. 7.

Effect of srtA mutation on the ability of S. pneumoniae to invade human pharyngeal cells. In a follow-up of the experiment described in the legend to Fig. 6, bacteria that remained associated with the pharyngeal cells after three consecutive washes were treated with antimicrobial agents that can kill extracellular but not intracellular pneumococci as described in Materials and Methods. After removal of the antibiotics by several washes, the pharyngeal cells were lysed and the number of internalized bacteria was determined as described in Materials and Methods. Data were normalized so that invasion by the wild-type cells was expressed as 100%. Error bars indicate standard deviations.

As with the β-galactosidase distribution, the degree of reduction in adherence or invasion was most pronounced in strains (e.g., Pen6) that had a high proportion of branched muropeptides in their cell walls (e.g., a 50% reduction in adherence for strain R36A versus an 87% reduction for strain Pen6) (Fig. 6 and 7).

Effect of srtA inactivation on virulence in the intraperitoneal mouse model.

Genomic DNA from strain R36srtA was used to introduce the srtA mutation into strain R36A expressing type III capsular polysaccharides (R36ASIII). Transformants were selected for erythromycin resistance. Washed cells of R36ASIII and R36ASIIIsrtA were injected intraperitoneally into mice at a range of concentrations from 500 to 5 × 10⁶ CFU per experimental animal. The survival of mice was monitored for 10 days. The inactivation of srtA had no detectable effect on the virulence of type III capsular polysaccharide-expressing bacteria in this animal model (Fig. 8).

FIG. 8.

Effect of srtA mutation on the virulence of capsular type III S. pneumoniae in the mouse intraperitoneal model. The Kaplan-Meier plot shows the percentages of survival of mice which received 10³ CFU of R36ASIII (closed symbols) and R36ASIIIsrtA (open symbols).

Homologues of srt genes in S. pneumoniae strain TIGR4.

A large mutant library generated from strain TIGR4 and screened for mutations affecting pneumococcal virulence identified srtD as a gene important for pneumococcal colonization and pneumonia in the animal models used in the screen (17). We tested the impact of insertional inactivation of three of the four srt homologues in TIGR4 (srtA, srtB, and srtD) on the localization of β-galactosidase. In cultures of TIGR4, 75% of the enzyme activity was associated with the bacterial pellet and 25% was found in the culture medium. This distribution of enzyme activity remained unaltered in TIGR4 mutants with inactivated srtB or srtD. On the other hand, inactivation of srtA caused a redistribution of enzyme activity: 31% of β-galactosidase was retained in the bacterial pellet and 69% was released into the culture medium (Fig. 9). The effect of srtA inactivation on β-galactosidase localization in this functional assay confirms the virtual identity of this gene to srtA of S. pneumoniae strain R6. The observation that srtD does not influence the localization of β-galactosidase (present study) but has a major impact on the colonization and pneumonia model (17) suggests that different sortases of S. pneumoniae may be involved in the display of different pneumococcal surface proteins, as described for S. pyogenes (2).

FIG. 9.

Distributions of β-galactosidase activities in S. pneumoniae strain TIGR4 and its srtA, srtB, and srtD mutants. Filled bars indicate cell pellet-associated enzyme activity; empty bars indicate activity in the culture medium. Enzyme activity is expressed as a percentage of the total recovered in association with the bacteria and in the culture medium. Error bars indicate standard deviations.

DISCUSSION

S. pneumoniae is an exclusively human pathogen, and the natural ecological reservoir is the nasopharynx of humans, particularly in children of preschool age. At present, it is believed that S. pneumoniae can emerge from the nasopharyngeal mucosa to invade the human host and that the mechanisms of colonization and invasion involve a large number of suspected or proven bacterial factors (28, 17, 21); the latter are chemically diverse, and many of them are located at the pneumococcal surface. These surface-located virulence factors include the antiphagocytic capsular polysaccharides (90 different chemical types) (23), the choline-containing wall and membrane teichoic acids (12), and the peptidoglycan (16), all of which are structural components of the pneumococcal cell wall. Some structural features of the cell wall are recognized by the host through the Toll receptors of innate immunity, and some cell wall components are involved in complement activation (37) and induction of preinflammatory cytokines (18). In addition to these polysaccharide-based heteropolymers, an increasing number of surface proteins have also been identified and implicated as directly or indirectly interacting with the host. Until recently, most, if not all, of these proteins were shown to be attached to the pneumococcal surface by choline residues of teichoic acids through noncovalent association. Such choline binding proteins include PspA (5); PsaA (29, 31); the enzyme phosphorylcholine esterase (36); and LytA, in which a 20-amino-acid repeat was shown to recognize choline residues in wall or membrane teichoic acids (14). A surface display of proteins through such highly specific but noncovalent binding represents a modality that is unique to S. pneumoniae and that is in contrast to the covalent attachment of most surface proteins studied for other gram-positive bacteria.

The observations described in this report demonstrate that the mechanism of surface display of proteins does not depend on the choline binding paradigm alone; pneumococci also use covalent anchoring for at least some surface proteins. Data available from the genome of S. pneumoniae strain R6 (19) indicate that there are at least 23 proteins carrying the LPXTG motif, and 15 of these have this recognition sequence at the C terminus, as expected for proteins that are processed by a typical sortase. We chose two proteins from among these: a β-galactosidase of unknown function that was recently shown to be a surface protein (38) and NanA, a surface protein that was already shown to be involved in pneumococcal pathogenesis (7). The genome of S. pneumoniae R6 contains a solitary ORF that shows structural features of a sortase gene. Here we showed that inactivation of this gene, srtA, by insertion duplication mutagenesis caused drastic changes in the localizations of both β-galactosidase and NanA: the majority of these normally cell-associated proteins were released into the culture medium in srtA mutants. Complementation with plasmid-borne srtA reversed this effect. The inactivation of srtA also caused a drastic reduction in the percentage of pneumococci capable of associating—in a wash-resistant form—with human pharyngeal cells in vitro, and srtA mutants also showed a reduced capacity to invade the same cell line. On the other hand, srtA inactivation did not affect the virulence of capsular type III pneumococci in the mouse intraperitoneal model.

Examination of the database available for S. pneumoniae strain TIGR4 (35) allowed some interesting comparisons. In contrast to strain R6, strain TIGR4 has a cluster of 3 ORFs annotated as srt homologues. In addition, also present in the TIGR4 database is an ORF annotated as a CHP gene (35). The results described in this communication indicate that the CHP gene is in fact a homologue of srtA. This conclusion is based on three observations: the CHP gene shows 99% identity to srtA of strain R6, inactivation of CHP by insertional mutagenesis causes the release of a substantial fraction of β-galactosidase into the culture medium, and the CHP gene and srtA of strain R6 have similar genetic organizations in that both genes are closely linked to gyrA. On the basis of these data, we suggest the consideration of CHP of TIGR4 as srtA.

A comparison of sequence data available in databases suggests that srtA of strain R6 and srtA of strain TIGR4 belong to a family of streptococcal sortase genes which are only distantly related to srtB, srtC, and srtD of strain TIGR4. The genetic organization of srtB, srtC, and srtD is completely different from that of R6 srtA. An interesting feature of that organization is the presence of two IS1176 elements flanking the cluster of these three genes—an intact copy of the insertion sequence located upstream and a disrupted copy located downstream—raising the possibility that these genes may have heterologous origins.

In a recent publication, Hava and Camilli (17) screened a large library of mutants generated in the background of TIGR4 for determinants that affect pneumococcal virulence in a number of animal models selected to allow the identification of the specific impact of such mutations on a variety of pneumococcal diseases. In that work, srtD was shown to have a profound impact in the pneumonia model and in nasopharyngeal colonization. In our experiments described in this communication, the inactivation of srtD (and srtB) had no effect on the localization of β-galactosidase. Together, these observations suggest that, similar to what has been described for S. pyogenes (2), different sortases may be involved in the processing of functionally different pneumococcal surface proteins in S. pneumoniae as well.