Contribution of the ATP-Dependent Protease ClpCP to the Autolysis and Virulence of Streptococcus pneumoniae

Abstract

The ATP-dependent caseinolytic proteases (Clp) are fundamental for stress tolerance and virulence in many pathogenic bacteria. The role of ClpC in the autolysis and virulence of Streptococcus pneumoniae is controversial. In this study, we tested the role of ClpC in a number of S. pneumoniae strains and found that the contribution of ClpC to autolysis is strain dependent. ClpC is required for the release of autolysin A and pneumolysin in serotype 2 S. pneumoniae strain D39. In vivo, ClpC is required for the growth of the pneumococcus in the lungs and blood in a murine model of disease, but it does not affect the overall outcome of pneumococcal disease. We also report the requirement of ClpP for the growth at elevated temperature and virulence of serotype 4 strain TIGR4 and confirm its contribution to the thermotolerance, oxidative stress resistance, and virulence of D39.

Upon exposure to stress conditions, particularly heat shock, bacteria respond by mediating a cascade of events leading to the synthesis of a unique group of proteins called heat shock proteins. The production of these proteins represents a protective cellular response to cope with the stress-induced damage of proteins. Heat shock proteins are highly conserved in both prokaryotes and eukaryotes (8, 14, 24). Many heat shock proteins are molecular chaperones or ATP-dependent proteases and play major roles in protein folding, repair, and degradation under normal and stress conditions (13, 41).

The ATP-dependent caseinolytic protease ClpCP consists of an ATPase specificity factor, ClpC, which is a member of the HSP100/clp family (33), and a proteolytic subunit, ClpP, which by itself represents a unique family of serine proteases (25). The clpC and clpP genes are members of class III group of heat shock genes, and their expression is negatively controlled by the class three stress gene repressor CtsR (9). The presence of ClpC, ClpP, and other Clp proteases in bacterial cells is fundamental for stress survival, as mutations in some clp genes result in inability to grow under stressful conditions (7, 12, 19, 35, 36). The gram-positive pathogen Streptococcus pneumoniae is a major cause of pneumonia, meningitis, septicemia, and otitis media (37, 40). Autolysis is a characteristic mechanism that enables the pneumococcus to release and exchange its genetic material and also to release virulence factors during infection. This suicidal behavior is believed to be mediated by the action of the major autolysin, LytA, which becomes activated during the stationary phase or upon antibiotic treatment (26). There is also evidence that the pneumococcus can undergo autolysis via autolysin-independent mechanisms (2).

During the course of this study, a number of reports about the role of pneumococcal ClpC have been published, but the contribution of ClpC to the autolysis and expression of competence genes and virulence factors is still debated (6, 7, 30). ClpC has been reported to play a role in thermal tolerance, control of autolysis, and chain formation in the pneumococcus (6), while other studies suggest that the protein plays no role in these processes (7, 30). Charpentier and coworkers (6) also reported that ClpC plays a major role in processes related to the virulence of the organism, including adherence to human cells and production of known virulence factors such as pneumolysin, autolysin A, CbpA, and other choline binding proteins. This study suggests that ClpC may play a major role in the virulence of the pneumococcus. A signature-tagged mutagenesis screen identified ClpC as a virulence factor in the pneumococcus (28). However, Robertson et al. (30) showed that ClpC null mutants of the pneumococcus are not greatly affected in their virulence. We undertook the present study to define the role played by ClpC in stress response and virulence.

Recent studies have agreed on the requirement of ClpP for thermotolerance, development of competence, and virulence of Streptococcus pneumoniae (7, 21, 30). Here we report that ClpC is involved in the autolysis of Streptococcus pneumoniae. It is therefore required for the release of the major autolysin autolysin A and the toxin pneumolysin. In vivo, ClpC does not affect the overall outcome of pneumococcal disease but contributes to the ability of the pneumococcus to grow in the lungs and blood. We also report the role of ClpP in stress tolerance and virulence in both type 2 (strain D39) and type 4 (strain TIGR4) Streptococcus pneumoniae strains. The role of ClpP in TIGR4 has not been reported before.

MATERIALS AND METHODS

Bacterial strains, plasmids, primers, and growth conditions.

The bacterial strains and oligonucleotide primers used in this study are listed in Table 1. S. pneumoniae strains were grown routinely on blood agar base (BAB) number 2 (Oxoid, Basingstoke, United Kingdom) supplemented with 5% (vol/vol) defibrinated horse blood (E&O Laboratories, Bonnybridge, United Kingdom) and on brain heart infusion (BHI) at 37°C. Escherichia coli ultracompetent cells (Stratagene, Amersham Zuidoost, The Netherlands) used for cloning were grown on Luria-Bertani broth or Luria-Bertani (LB) agar plates. Where necessary, antibiotics were added to the growth media at the following concentrations: ampicillin at 50 μg/ml and erythromycin at 1 mg/ml for E. coli or 1 μg/ml for S. pneumoniae. PCR-Script plasmid (Stratagene, Amersham Zuidoost, The Netherlands) was used for cloning according to the manufacturer's recommendations.

TABLE 1.

Bacterial strains and primers used in this study

Strain or primer	Description or sequence	Reference, source, or use
S. pneumoniae strains
D39 (MA)	Serotype 2, virulent strain, NCTC 7466	1
D39 (LA)	Serotype 2, avirulent strain	Laboratory stock
TIGR4	Serotype 4, clinical isolate	38
R6	Subclone of R36A, derivative of D39	34
R800	R6 derivative	23
D39 (MA) ΔclpC	D39(MA) clpC replaced with Ery cassette	This work
D39 (LA) ΔclpC	D39(LA) clpC replaced with Ery cassette	This work
R6 ΔclpC	R6 clpC replaced with Ery cassette	This work
R800 ΔclpC	R800 clpC replaced with Ery cassette	This work
TIGR4 ΔclpC	TIGR4 clpC replaced with Ery cassette	This work
SP2000	ΔclpP mutant with Kan cassette	7
D39 ΔclpP	D39(MA) clpP replaced with Kan cassette	This work
TIGR4 ΔclpP	TIGR4 clpP replaced with Kan cassette	This work
Primers
ClpC a1	ATGAACTATTCAAAAGCATTG	clpC amplification
ClpC a2	TGCAATATCAAATTTTAACTGG	clpC amplification
Clp inv1	GGCGCGCCTATCATGCAAAATCGCATAG	Deletion of clpC
Clp inv2	GGCGCGCCAACTGAAAAAAGCTTATAGACC	Deletion of clpC
Ery 3	GGCGCGCCTAACTATCGTCTTGAGTCCAACCC	Amplification of erythromycin (Ery) resistance cassette
Ery 5	GGCGCGCCGTTCATATTTATCAGAGCTCGTGC	Amplification of erythromycin cassette
MCP1.for	CATACAAGAGGTAGATAGAAAAGG	Confirmation of chromosomal insertion of interrupted clpC
MCP2.rev	TCAAACCAGTGTTTTGAGCAACCT	Confirmation of chromosomal insertion of interrupted clpC
AC94	TGACCATGGTTCCAGCTGCTAAAGTTGGC	clpP deletion (7)
AC97	GCTACCATGGCAAGCGCCACAAACGATAG	clpP deletion (7)

DNA techniques and transformation.

Chromosomal DNA was prepared as described elsewhere (3). PCR, restriction endonuclease digestions, DNA ligation, and DNA electrophoresis were performed according to standard protocols (32). Kits from Qiagen were used for DNA purification and plasmid preparations according to the manufacturer's instructions. Transformation of E. coli with plasmid DNA was carried out according to the manufacturer's instructions (Stratagene, Amersham Zuidoost, The Netherlands). Transformation of S. pneumoniae D39 (serotype 2) and S. pneumoniae TIGR4 (serotype 4) was done by a modification of the method of Lacks and Hotchkiss (22). Competence-stimulating proteins (CSP-1 for D39 and CSP-2 for TIGR4) were used for inducing competence in these different serotypes (29). The transformed cells were selected on BAB plates with appropriate antibiotic selection.

Construction of ΔclpC and ΔclpP mutants.

The whole clpC gene was amplified from the chromosomal DNA of D39 with primers ClpC a1 and ClpC a2 (Table 1) and cloned into the PCR-Script cloning vector (Stratagene, Amersham Zuidoost, The Netherlands). Internal primers Clp inv1 and Clp inv2 (Table 1) were designed to carry out inverse PCR for removing the middle region of clpC, creating AscI restriction sites in the resulting PCR product. The AscI-generated erythromycin cassette, was ligated to the inverse PCR product with Ery3 and Ery5 primers (Table 1) to be used as a selection marker. This construct was amplified out of the plasmid with primers ClpC a1 and ClpC a2 and used to transform S. pneumoniae. Mutants were selected by growth on erythromycin.

Mutation was confirmed with diagnostic primers MCP1.for and MCP2.rev. (Table 1), which amplify a 2.6-kb fragment in the D39(LA) wild type and a 1.9-kb fragment in the D39(LA) ΔclpC mutant. It was also confirmed by sequencing and the growth of the pneumococcus on erythromycin. This mutation was moved into strains R6, R800, TIGR4, and a mouse-adapted strain of D39, D39(MA). The ΔclpP mutants of TIGR4 and the mouse-adapted strain of D39 were constructed with ΔclpP mutant strain SP2000 (7) by PCR amplification of a 1.6-kb fragment corresponding to the clpP mutation carrying the kanamycin resistance cassette with primer pair AC94 and AC97 (Table 1). This PCR product was used to transform strains D39 and TIGR4. Transformed cells were selected with 250 μg of kanamycin per ml.

Repair of ΔclpC mutation in D39.

To confirm that the observed phenotype of the D39 ΔclpC mutant strain was due to the disruption of clpC and not mutations elsewhere in the chromosome, the clpC mutation was repaired. A 2,578-bp PCR product was amplified from wild-type D39 genomic DNA with primer pair MCP1 for and MCP2 rev (Table 1). This fragment contained 30 bases upstream of the start codon (ctsR gene) and an additional 115 bases downstream of the stop codon for clpC. This PCR product was purified with the Qiagen gel purification kit and used directly to transform the D39 ΔclpC strain by homologous recombination. Transformation of S. pneumoniae D39 ΔclpC was carried out as described above. The transformation reactions were screened for transformants by plating onto selection plates (with 1 μg of erythromycin per ml) and nonselection plates (without antibiotic). Colonies which lost erythromycin resistance were selected and checked by PCR and sequencing for the presence of a complete copy of clpC.

In vitro stress experiments.

To compare the effect of heat stress on ΔclpC and ΔclpP mutants to that of their wild types, the same number of viable cells (10⁶ CFU/ml) was used to inoculate BHI warmed to 30, 37, and 40°C. At 1-h intervals, samples were taken to measure the optical density at 600 nm and the viable counts. Samples were subjected to vigorous vortexing before plating for viable counting to ensure that any long chains of organisms were broken up. This was confirmed by microscopic analysis of selected samples.

The sensitivity of ΔclpC and ΔclpP mutants to H₂O₂ was tested by exposing aliquots of cultures grown to an optical density of 0.3 to 40 mM H₂O₂ for 15 min at room temperature. The viable cells were counted by plating onto blood agar plates before and after the exposure to H₂O₂ and the result was expressed as percent survival (16).

To investigate the role of ClpC in pH tolerance, equal numbers of cells of wild-type D39(LA) and the ΔclpC mutant were used to inoculate BHI adjusted to pH 4, 5, 6, 7, 8, and 9. Growth was measured after overnight incubation at 37°C (10). The penicillin-induced cell lysis of ΔclpC mutants and their wild types was also studied. Cultures grown to an optical density of 0.3 were exposed to 0.1 μg of penicillin per ml, and the effect on cell viability was recorded by plating samples of penicillin-treated cultures taken at hourly intervals on BAB (6).

Western immunoblotting.

All manipulations for protein extraction were carried out at 4°C. Cell pellets of cultures grown for 11 h were collected by centrifugation at 5,000 × g for 15 min and resuspended in 1 ml of phosphate-buffered saline. Cellular proteins were released by sonication (four times for 30 s each with 30 s of incubation on ice between each). The cell debris was removed by centrifugation at 13,000 × g for 10 min. Proteins in the culture supernatant were concentrated with ice-cold acetone. The concentration of total proteins in the cellular fraction and supernatant was determined by the Bradford assay with the use of bovine serum albumin as a standard (4). For Western blot analysis, the same amount of proteins in the cellular fraction and supernatant of wild-type and ΔclpC mutant D39 strains were separated by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis, electroblotted onto nitrocellulose membranes (Amersham), and reacted with specific antisera against pneumolysin and autolysin A by standard protocols (32).

Microarray analysis.

The pneumococcal genome microarray slides (second version) designed at the Pathogen Functional Genomics Resource Centre at TIGR (http://www.pfgrc.tigr.org/) were used in this study. The full genome array consists of amplicons representing segments of 2,131 open reading frames from S. pneumoniae reference strain TIGR4 spotted in quadruplicate on glass slides. Also, the array contains an additional 563 open reading frames from strains R6 (164) and G54 (399). The bacterial RNA was extracted from cultures grown in BHI to an OD at 600 nm of 0.6 with RNeasy Midi (Qiagen) columns according to the manufacturer's protocol. The RNA was reverse transcribed and labeled with Cy3 or Cy5 probes (Amersham Pharmacia) by the aminoallyl labeling method. Denatured probes were then hybridized to the glass slides and incubated for 18 h at 42°C. After hybridization, the slides were washed in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% sodium dodecyl sulfate for 5 min at 55°C, followed by a wash in 0.1× SSC-0.1% sodium dodecyl sulfate for 5 min and 0.1× SSC for 5 min.

Hybridized slides were scanned with the Perkin Elmer Scan Array Express. The spot intensities were defined and quantified with the Packard BioScience QuantArray Microarray analysis software. The data were further analyzed with GeneSpring 6.0 (Silicon Genetics). LOWESS intensity-dependent normalization was used to perform per-spot and per-array normalization, and the cross-gene error model was based on the replicate measurements. Statistically significant differences were defined as those with a t test P value of less than 0.0001 and a ratio change threshold of at least 2 standard deviations compared to the median ratio for each strain.

Mice and infections.

Female MF1 mice (25 to 30 g) were purchased from Harlan Olac, Bicester, United Kingdom. They were used at 9 weeks of age. For intranasal infection, mice were lightly anesthetized with 1.5% (vol/vol) halothane, and the infectious dose (10⁶ CFU/mouse) was administered in 50 μl to the nostrils of mice held vertically (17). After administering the infectious dose, mice were observed for the development of symptoms over a period of 2 weeks. For intravenous infection, the infectious dose (10⁵ CFU) was administered as a 50-μl volume injected into the lateral tail vein. A zero time point bleed was taken from a separate vein to ensure successful infection.

Bacteriological investigation.

At chosen intervals after infection, groups of mice were sacrificed by cervical dislocation, ensuring intact trachea, and a blood sample was removed via cardiac puncture. For bronchoalveolar lavage, a 16-gauge nonpyrogenic angiocath (F. Baker Scientific, Runcorn, United Kingdom) was inserted into the trachea, and the lungs were lavaged with a total volume of 2 ml of sterile phosphate-buffered saline. Lavaged lungs were homogenized in 5 ml of phosphate-buffered saline with a handheld glass tissue homogenizer (Jencons, Leighton Buzzard, United Kingdom). Viable bacteria in lung and blood samples were counted by plating out serial 10-fold dilutions on BAB (18). Because mutations in the clpC gene were found to affect chain formation during growth of the pneumococcus, we confirmed that the homogenization and vortexing procedure used prior to counting broke up long chains of organisms (data not shown).

Histological analysis.

We sacrificed mice 48 h following intranasal infection, and the lungs were inflated with 1 ml of 10% (vol/vol) formal saline prior to their removal. Following fixation, the lungs were embedded in paraffin and blocked by standard histological protocols. Lung blocks were sectioned at 5 μm prior to staining with hematoxylin and eosin (BDH Laboratory supplies, Poole, United Kingdom) (18).

Statistical analysis.

Statistical analyses were carried out with StatView 4.1 (Abacus Concept). Survival times were analyzed by nonparametric Mann-Whitney U analysis. Bacteriology results are expressed as mean ± standard error of the mean. Comparisons of bacterial loads between bacterial strains in the time course bacteriology experiment were performed with unpaired t tests. Comparisons at different times within groups were carried out with analysis of variance with the Bonferroni post hoc test. In all analyses, a P value of <0.05 was considered statistically significant.

RESULTS

Requirement of ClpC for autolysis of the pneumococcus is strain dependent.

The ability of ΔclpC mutants to grow at different temperatures was studied. The growth of all ΔclpC mutants was reduced compared to that of the wild types at 37°C as measured by viable counting. This difference was in growth rate rather than total growth, as both the mutant and wild-type strains reached the same stationary-phase viable count (Fig. 1). There was no defect in growth at 40°C, suggesting that ClpC is not involved in the heat stress tolerance of the pneumococcus. However, the ΔclpC mutants of both the laboratory-adapted strain D39(LA) and the mouse-adapted strain D39(MA) maintained cell viability after the stationary phase of growth compared to their parent strains, which tend to undergo autolysis rapidly after this phase when a certain cell density is reached (Fig. 1). Maintenance of viability was not recorded in the ΔclpC mutants of strains R6, R800, and TIGR4 (data not shown).

FIG. 1.

Representative graphs of in vitro growth of wild-type (WT) D39(LA) and D39(MA) and their ΔclpC mutants at 37 and 40°C.

The cell morphology of cultures grown at different optical densities was examined by light microscopy of the Gram-stained samples. The ΔclpC mutants of both of the D39 strains grew in long chains of bacterial cells rather than the typical diplococcus form of S. pneumoniae (Fig. 2). No chains longer than 10 organisms were observed in the wild-type cultures, while at least one chain of this length or greater was seen in 12 randomly selected fields, as shown in Fig. 2. The appearance of these chains suggests an impairment in cell separation, which is believed to be induced by the autolytic action of autolysin A (LytA) (5, 27).

FIG. 2.

Effect of clpC deletion on the morphology of the pneumococcus. Samples of the same optical density were withdrawn, Gram stained, and examined by light microscopy. Magnification, ×1,000.

To further investigate this phenomenon, penicillin-induced autolysis of ΔclpC mutants and their wild types was studied (Fig. 3). The ΔclpC mutant of D39(LA) showed a small decline in viable count over a 6-h period of exposure to penicillin, while its parent strain lost its viability and no viable cells were recorded after 6 h of exposure to the antibiotic. The ΔclpC mutant of the D39(MA) strain also maintained viability in the presence of penicillin, but surprisingly, its parent strain showed no penicillin-induced autolysis (Fig. 3). The wild-type and ΔclpC mutant TIGR4 strains were also penicillin resistant (data not shown). Similar to their parent strains, the R6 ΔclpC and R800 ΔclpC mutants underwent autolysis when treated with penicillin (data not shown).

FIG. 3.

Penicillin-induced autolysis of ΔclpC mutants compared to their wild types (WT). Cultures grown to an optical density of 0.3 were exposed to 0.1 μg of benzylpenicillin per ml, and viable bacteria were counted on BAB plates at 1-h intervals over a 6-h period.

ClpC contributes to the release of autolysin A and pneumolysin.

It is believed that the toxin pneumolysin is released upon lysis of pneumococcal cells by the influence of autolysin A (39). The decreased rate of autolysis in the ΔclpC mutant encourages the investigation on the effect of clpC deletion on autolysin A and pneumolysin release. Proteins in the cellular fraction and those in the culture supernatant of the mouse-adapted strain D39(MA), which was chosen for the in vivo analysis, and its ΔclpC mutant grown in BHI for 11 h at 37°C were separated, transferred to nitrocellulose membranes, and reacted with antisera to autolysin A and pneumolysin. Autolysin A was not detected in the supernatant of ΔclpC mutant but found in that of the wild type and in the cellular fraction of both strains. Pneumolysin was detected in the wild type and the mutant in both the cellular fraction and the supernatant. However, the amount of the toxin was less in the supernatant of the ΔclpC mutant (Fig. 4). This again suggests a role of ClpC in the autolysis of some strains and serotypes of S. pneumoniae.

FIG. 4.

Western immunoblot analysis of autolysin A (LytA) and pneumolysin (Ply) in the D39(MA) parent strain and its ΔclpC mutant.

ClpP but not ClpC is required for oxidative stress tolerance of S. pneumoniae.

The proteolytic activity of ClpP is required by bacterial cells to prevent the accumulation of misfolded or damaged proteins resulting from different stress conditions (11, 20). To study the role of ClpC and ClpP in oxidative stress tolerance, the sensitivity of ΔclpC and ΔclpP mutants to hydrogen peroxide was compared to that of their parent strain D39(MA). As reported before (30), the ΔclpP mutant was more sensitive to peroxide than the wild type (Fig. 5). On the other hand, the response of the ΔclpC mutant to peroxide was very much similar to that of the wild type (Fig. 5), suggesting that ClpC does not play a role in pneumococcal resistance to oxidative stress. The contribution of ClpC to pH tolerance was also investigated. No differences were recorded between the wild-type and ΔclpC mutant D39 strains in growth at different pHs, suggesting that ClpC has no influence on the acid tolerance of the pneumococcus (data not shown).

FIG. 5.

Peroxide sensitivity assay for wild-type D39 (WT) and its ΔclpC and ΔclpP mutants. We added 40 mM H₂O₂ to 1-ml aliquots of culture grown to an OD at 600 nm of 0.3 and left it at room temperature for 15 min. Viable counts were performed on BAB plates before and after the addition of peroxide, and percent survival was calculated. Values expressed are the mean and standard error of the mean of three independent experiments. *, P < 0.05 for lower survival of the ΔclpP mutant than of the wild type.

ClpP is involved in the thermotolerance of S. pneumoniae strains D39 and TIGR4.

The requirement of the pneumococcal ClpP for resistance to heat stress was reported before in strains R6 and D39 (7, 30). We investigated the role of ClpP in our D39 and TIGR4 strains by growing the wild types and their ΔclpP mutants in BHI at 30, 37, and 40°C and recording the optical density at 600 nm at intervals (Fig. 6). The growth of both mutants at 37°C was very much similar to that of their wild types. However, the ΔclpP mutants of both strains were unable to grow at 40°C. The D39 ΔclpP mutant was less able to grow at 30°C than its D39 parent strain, while the ΔclpP mutant of TIGR4 and its wild-type strain were identical in their growth at this lower temperature (Fig. 6). The reduced growth rate of the D39 ΔclpP mutant at 30°C has been reported previously (30).

FIG. 6.

Representative growth curves of parent strains D39 and TIGR4 and their ΔclpP mutants at different temperatures. Values are representative of five independent experiments.

ClpC is required for the growth of S. pneumoniae in the lung and blood of a murine pneumonia model.

The effect of ClpC on the rate of autolysis and therefore the release of pneumolysin suggested that ClpC might play a role in the virulence of the pneumococcus. We used our murine models of pneumonia and bacteremia to investigate the contribution, if any, of ClpC to the pathogenesis of pneumococcal disease. Infection of mice was carried out as described in the Materials and Methods section. All mice infected intranasally with the ΔclpC mutant succumbed to the infection at the same rate and had survival times similar to those of mice infected with wild-type D39 (Fig. 7A). There were also no statistically significant differences between the D39 ΔclpC mutant and its parent strain in the ability to cause death when given intravenously to mice (data not shown).

FIG. 7.

Effect of ClpC on the virulence of the pneumococcus. (A) Survival of MF1 mice infected intranasally with 10⁶ CFU/mouse of D39 or D39 ΔclpC. (B) Bacterial counts in lung tissue. (C) Bacterial counts in blood at different times after intranasal infection with 10⁶ CFU/mouse of each strain. (D) Bacterial count in blood taken by tail bleeding after intranasal infection (bacterial counts for every mouse are shown in Table 3). The dashed line represents the limit of detection of the assay. *, P < 0.05 for lower bacterial loads of the ΔclpC mutant than of wild-type D39 (WT).

We also investigated the bacterial counts in the lungs and blood of mice after intranasal infection. Both the wild-type and mutant strains were cleared from the lung airways (data not shown). However, the bacterial counts of the ΔclpC mutant in the lung tissue were significantly lower than those of wild-type D39 at 12, 18, 24, and 48 h postinfection (Fig. 7B). In the bloodstream, the wild-type organism grew to a level of about 10⁶ CFU/ml, while the ΔclpC mutant caused a transient bacteremia in mice 24 h postchallenge and then started to clear after 48 h (Fig. 7C), when the number of ΔclpC mutant bacteria was significantly lower than that of the wild type.

Because the data in Fig. 7C are derived from groups of animals sacrificed at each time point, it is not possible to determine the extent of transient bacteremia in individual animals. To measure the levels of transient bacteremia caused by the ΔclpC mutant organisms, the number of bacteria in the blood of individual mice was followed at different time points after intranasal infection by counting bacteria in blood samples taken by sequential tail bleeding (Fig. 7D). Again, the mutant organisms grew to a level of about 10⁴ CFU/ml 24 h postchallenge and then cleared after 48 h, while the wild-type organisms grew dramatically (Fig. 7D). The results from the experiment in Fig. 7C are not significantly different from those in Fig. 7D. Histological examination of lung tissue sections revealed no difference in the influx of inflammatory cells into the lungs of mice infected with either the ΔclpC mutant or its D39 parent strain 48 h after intranasal infection (data not shown).

Confirmation of the role of clpC.

In order to confirm that the construction of the clpC mutation did not have a polar effect on the expression of downstream genes, we examined the expression of these genes with microarrays. Microarray analysis revealed that the expression of genes downstream from clpC was not affected by disruption of the clpC genes (Table 2). Analysis of the levels of RNA in the region showed that the only transcript level significantly altered in this region was that of ClpC.

TABLE 2.

Expression of clpC and downstream genes in the D39 ΔclpC mutant compared to the wild type as determined by microarray

Genea	Spot Intensity ratiob	Pc	Annotationd
SP2194	0.25	4.7E-05	ATP-dependent Clp protease, ATP-binding subunit
SP2193	1.23	0.254	DNA-binding response regulator
SP2192	1.55	0.003	Sensor histidine kinase
SP2191	1.25	0.033	Conserved hypothetical protein
SP2190	1.20	0.269	Choline-binding protein A

^aGene designation of microarray ORFs which match the TIGR designations (http://www.tigr.org).

^bMutant/wild type intensity ratios as determined in microarray experiments.

^cMean value calculated from individual t-tests of intensity changes between the wild type and mutant. A P of less than 0.0001 is considered significant for up- or downregulated genes.

^dAnnotations as published for the TIGR4 genome (http://www.tigr.org).

To confirm that the phenotype observed for the D39 ΔclpC mutant strain is due to the mutation in the clpC gene and not to other mutations introduced into the chromosome, we repaired the clpC mutation by transformation with the wild-type gene. Following transformation of D39 ΔclpC with the wild-type clpC gene, we recovered several pneumococcal colonies that had lost the erythromycin-resistant phenotype. Analysis with PCR and nucleotide sequencing revealed the existence of the wild-type clpC gene in the revertant strain. The in vitro growth of the revertant strain was identical to that of the wild type (data not shown). The bacterial load of the revertant strain in a tail bleed was also very similar to that of the wild type (Table 3).

TABLE 3.

Bacterial counts in blood samples taken by tail bleeding

Mouse no.	Bacterial load (log CFU/ml) at time (h) post infection:
	D39				D39 ΔclpC				D39 clpC+
	0	18	24	48	0	18	24	48	0	18	24	48
1	NFa	2.20	4.57	7.70b	NF	NF	NF	2.96	NF	2.39	4.56	7.28b
2	NF	2.21	2.87	4.72	NF	2.76	4.98	3.02	NF	NF	NF	6.43
3	NF	4.11	5.79	7.86	NF	2.40	3.87	NF	NF	4.88	6.32	8.23b
4	NF	3.22	5.16	6.43	NF	2.22	4.99	2.52	NF	4.90	6.41	7.59b
5	NF	3.70	5.40	6.62	NF	NF	3.77	NF	NF	3.57	4.96	6.36

^aNF, none found.

^bMouse sacrificed 44 h post infection because moribund.

Finally, to test for the presence of suppressor mutations that compensate for the lack of ClpC, we did an experiment in which organisms were recovered from the blood of animals infected with D39 ΔclpC and showing signs of disease. These organisms were then used to rechallenge a further group of animals. The phenotype observed with serial tail bleeding was very similar to that of the original D39 ΔclpC strain (data not shown), showing that the virulence profile of the D39 ΔclpC strain is not affected by serial passage. The results of the gene repair and the serial passage suggest that the phenotype observed is due to the lack of ClpC and not other mutational effects.

ΔclpP mutants of strains D39 and TIGR4 are attenuated in virulence.

ClpP-negative mutants of the type 2 strain D39 were attenuated in murine intratracheal (30) and intraperitoneal (21) models of infection. We tested the virulence of ΔclpP mutants of D39 and TIGR4 in our pneumonia model. None of the mice in the D39-infected group survived (median survival time, 51 h), while D39 ΔclpP-infected mice did not show any symptoms of illness (median survival time, 336 h) until the end point of the experiment (Fig. 8A). The virulence of the TIGR4 ΔclpP mutant was significantly reduced (median survival time, 54 h) compared to that of wild-type TIGR4 (median survival time, 28 h), although no TIGR4 ΔclpP-infected mice survived the challenge (Fig. 8B). In addition, the number of bacteria in the lung tissue and blood of mice infected with the D39 ΔclpP mutant was significantly lower than that in mice infected with wild-type D39 24 and 48 h postchallenge (Fig. 9).

FIG. 8.

Survival of MF1 mice infected intranasally with 10⁶ CFU/mouse of D39 or D39 ΔclpP (A) and TIGR4 or TIGR4 ΔclpP (B). The D39 ΔclpP mutant was completely avirulent, while the TIGR4 ΔclpP mutant was less virulent than wild-type TIGR4 (WT). *, P < 0.05 for shorter survival times for wild-type D39 and TIGR4 than for their ΔclpP mutants.

FIG. 9.

Bacterial counts of wild-type D39 (WT) and its ΔclpP mutant in the lung lavage, lung tissue homogenate, and blood 24 and 48 h after intranasal infection with 10⁶ CFU/mouse of each strain. The dashed line represents the limit of detection of the assay for blood samples. *, P < 0.05 for lower bacterial loads of the ΔclpP mutant than of the D39 parent strain.

DISCUSSION

Clp-mediated proteolysis represents a cellular regulatory mechanism that plays a role in the stress response in bacteria. Streptococcus pneumoniae contains putative orthologues of four ATPase specificity factors (ClpC, ClpE, ClpL, and ClpX) (15, 38) and a single protease subunit, ClpP. As the role of ClpC in the autolysis and virulence of S. pneumoniae is controversial (6, 7, 28, 30), we studied the role of ClpC in different pneumococcal strains. We used two versions of strain D39, one of which is mouse virulent (termed mouse-adapted), and the other is avirulent in mice and termed laboratory-adapted. The basis for the difference in the virulence of these two isolates is unknown but appears not to be related to phase variation or capsule expression, as judged by visual analysis of colonies (data not shown). These strains differed not only in their virulence for mice but also in their resistance to penicillin-induced lysis. The mouse-adapted strain was resistant to penicillin-induced autolysis, whereas the laboratory-adapted strain was not. As penicillin-induced autolysis is believed to involve the action of the major autolysin (autolysin A), these strains may differ in their basal activity of autolysin A.

The nucleotide sequence of the autolysin A gene has been determined and is not mutated compared to the D39 sequence present in the database. Total levels of autolysin A, as judged by Western blotting, are also similar in D39(MA) and D39(LA). The difference may be related to activation of the autolysin A enzyme or by some other aspect of differences in the cell wall on the strains. Whether the difference in penicillin-induced lysis in these strains is reflected in their virulence is not clear at present.

The ΔclpC mutants of all three strains studied were able to grow at high temperature in a manner similar to the parent strains. However, the ΔclpC mutant of type 2 strain D39 did not undergo autolysis after the stationary phase of growth, grew in long chains, and was resistant to penicillin-induced lysis, suggesting an impairment in the function of autolysin A. In the case of the mouse-adapted strain of D39, the parent strain was also resistant to penicillin-induced lysis. Charpentier and coworkers reported that a ΔclpC mutant of strain R6 (a derivative of D39) showed increased tolerance to high temperatures, formed long chains, and failed to undergo lysis after penicillin treatment (6). Contrary to this report, Chastanet and colleagues reported that a ΔclpC mutant of strain R348 grew as diplococcal cells, did not form chains, and was indistinguishable from the wild type in response to penicillin- or deoxycholate-induced autolysis and also in growth at different temperatures (7). A third paper by Robertson et al. (30) reported that ClpC was not involved in growth at elevated temperatures and also did not affect autolysis in strains R6 and D39.

Our data suggest that ClpC is not involved in thermal tolerance but can play a role in autolysis, cell separation, and resistance to penicillin-induced lysis. However, our findings with the two variants of D39 also show that this phenotype can vary according to the history of the strain. This presumably reflects the accumulation of other mutations that affect the phenotype of ClpC mutants. Differences also occur between strains, as shown by the effect of the clpC mutation in the TIGR4 strain reported here. Because we wished to address the role of these proteins in pathogenesis, further in vivo analysis was confined to the mouse-adapted strain of D39.

In a Western blot analysis, the quantity of released pneumolysin was less in the ΔclpC mutant than in the wild type, and no release of autolysin A into the culture supernatant was detected for the mutant. Autolysin is activated during the stationary phase or upon penicillin treatment to cause lysis of bacteria and the release of cytoplasmic contents including pneumolysin and other virulence factors (26, 31). Thus, it appears that ClpC is involved in the control of this autolytic process and in the release of pneumococcal virulence factors. However, it should be noted that the extracellular release of pneumolysin from the pneumococcal strain WU2 (type 3) can also be independent of the action of autolysin A (2). The lack of autolysis was not observed for strains R6, R800, and TIGR4, suggesting that the contribution of ClpC to the autolysis of the pneumococcus is strain dependent. We also tested the contribution of ClpC to pH and oxidative stresses and found that ClpC seems to have no effect on pneumococcal acid tolerance or resistance to peroxide.

ClpC was identified as a virulence factor in a signature-tagged mutagenesis screen (28). It was suggested to play a role in the virulence of S. pneumoniae by affecting expression of choline-binding proteins (Cbps) and other virulence factors, such as pneumolysin (6), and the virulence of D39 ΔclpC was reported to be marginally attenuated in an intratracheal model of infection (30). In our pneumonia and bacteremia models, ClpC did not affect the overall outcome of the pneumococcal disease, as judged by survival times. However, growth in the lungs and blood after intranasal infection was dependent on ClpC. The number of ΔclpC mutant cells recovered from the lungs declined from approximately 10⁵ to less than 10⁴ whereas the wild type increased to more than 10⁶.

The appearance of bacteria in the bloodstream following intranasal challenge was similar for the wild type and ΔclpC strains in that bacteria appeared in the bloodstream at a similar time and grew to about 10⁴ per ml of blood by 24 h (significantly higher counts than at time zero for both groups). Therefore, although ΔclpC is less able to grow at 37°C in vitro, it is able to grow normally for the first 24 h in mouse blood. However, the wild-type bacteria then continued to grow to a level of approximately 10⁶ by 48 h, whereas the number of the mutant was reduced to a level that was no longer significantly different from the counts at time zero. The effect of disruption of the clpC gene on bacteriology is therefore dramatic, with very small numbers of the mutant present in the lung and blood at 48 h, when animals begin to show signs of disease. This suggests that the lethal event is triggered early in the infection or that small numbers of bacteria are able to cause the effect. In fact, analysis of serial tail bleed from the same animal (Table 3) shows that the mouse can actually clear transient bacteremia to levels below detection but still go on to die from the infection, suggesting that the former is more likely. It also remains a possibility that the level of bacteremia in ΔclpC infected mice increases again between 48 h and the time of death. The event triggered by the pneumococcus that results in death is still not known.

ClpP proteolysis is indispensable for survival under conditions of stress. We report herein that ClpP is necessary for the growth of both type 2 and type 4 strains at elevated temperatures. We also found that ClpP is required for the growth of D39 at lower temperature (30°C) but not for TIGR4 (Fig. 6). The defect in growth at 30°C for a ClpP mutant of D39 has been reported previously (30). The requirement of ClpP for the survival of TIGR4 under heat shock has not been reported before but again highlights strain differences in the role of these molecules. We also confirm the involvement of ClpP in the resistance of S. pneumoniae strain D39 to oxidative stress, which was reported by Robertson and coworkers (30).

Clp-controlled proteolysis is also essential for disease progression and virulence of bacterial pathogens, favoring survival in the host organism or modulating the activity of virulence factors. In vivo, the D39 ΔclpP mutant was completely avirulent in our pneumonia model of infection, while the TIGR4 ΔclpP mutant was less virulent than its wild type (mirroring the less pronounced effect of the ΔclpP mutation on in vitro growth) (Fig. 6). A ΔclpP mutant of strain D39 was reported to be strongly attenuated for virulence in murine lung and sepsis infection models (30). Despite the agreement on the contribution of ClpP to the virulence of the pneumococcus, it is not completely understood whether this contribution is due to its proteolytic effect on misfolded or stress-damaged proteins or to regulation of other virulence factors. Recently, Kwon and coworkers reported that expression of virulence genes such as pneumolysin and pneumococcal surface antigen A is modulated by ClpP protease (21).

The difference in virulence phenotype between the ΔclpC and ΔclpP strains shows that other ATPase specificity factors interact with ClpP or that ClpP has roles independent of the activity of these factors. We are currently investigating the role of other ATPase specificity factors in the virulence process.