Molecular analysis of the psa permease complex of Streptococcus pneumoniae
Summary
The psaBCA locus of Streptococcus pneumoniae encodes a putative ABC Mn2+-permease complex. Downstream of the operon is psaD, which may be co-transcribed and encodes a thiol peroxidase. Previously, there has been discordance concerning the phenotypic impact of mutations in the psa locus, resolution of which has been complicated by differences in mutant construction and the possibility of polar effects. Here, we constructed unmarked, in frame deletion mutants ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaD, ΔpsaBC, ΔpsaBCA and ΔpsaBCAD in S. pneumoniae D39 to examine the role of each gene within the locus in Mn2+ uptake, susceptibility to oxidative stress, virulence, nasopharyngeal colonization and chain morphology. The requirement for Mn2+ for growth and transformation was also investigated for all mutants. Inductively coupled plasma mass spectrometry (ICP-MS) analysis provided the first direct evidence that PsaBCA is indeed a Mn2+ transporter. However, this study did not substantiate previous reports that the locus plays a role in choline-binding protein pro-duction or chain morphology. We also confirmed the importance of the Psa permease in systemic virulence and resistance to superoxide and hydrogen peroxide, as well as demonstrating a role in nasopharyngeal colonization for the first time. Further evi-dence is provided to support the requirement for Mn2+ supplementation for growth and transformation of ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaBC, ΔpsaBCA and ΔpsaBCAD mutants. However, transformation, as well as growth, of the ΔpsaD mutant was not dependent upon Mn2+ supplementation. We also show that, apart from sensitivity to hydrogen peroxide, the ΔpsaD mutant exhibited essentially similar phenotypes to those of the wild type. Western blot analysis with a PsaD antiserum showed that deleting any of the genes upstream of psaD did not affect its expression. However, we found that deleting psaB resulted in decreased expression of PsaA relative to that in D39, whereas deleting both psaB and psaC resulted in at least wild-type levels of PsaA.
Introduction
Streptococcus pneumoniae continues to be an important human pathogen, causing pneumonia, meningitis and bacteraemia, as well as otitis media in young children (Paton et al., 1993; Paton, 1998). There has been much research aimed at the identification of pneumococcal virulence proteins in order to identify potential vaccine antigens and novel antimicrobial targets. One of these proteins is a 37 kDa lipoprotein, pneumococcal surface antigen A (PsaA) (Russell et al., 1990), which has been shown to be a protective immunogen in mice (Talkington et al., 1996; Briles et al., 2000a,b). psaA mutants have been shown to exhibit greatly reduced virulence in systemic as well as in respiratory tract and otitis media murine models of infection (Berry and Paton, 1996; Marra et al., 2002).
PsaA is the solute-binding component of an ATP-binding cassette (ABC) cation permease encoded by the psaBCA locus (Dintilhac et al., 1997). Analysis of the complex indicates that psaB encodes an ATP-binding protein, whereas psaC encodes a transmembrane protein (Dintilhac and Claverys, 1997). Sequence analysis places PsaA in the cluster IX family of bacterial transporters of the essential metal ions Mn2+, Zn2+ and Fe2+ (Dintilhac et al., 1997; Claverys, 2001). All the physiological evidence indicates that it is involved in the transport of Mn2+; growth of a psaA mutant showed an absolute requirement for additional Mn2+ (Dintilhac et al., 1997; Marra et al., 2002). Knocking out psaB and psaC also results in a requirement for added Mn2+ (Novak et al., 1998; Marra et al., 2002). However, in the crystal structure of PsaA, the metal binding site appears to be occupied by Zn2+ (Lawrence et al., 1998). Nevertheless, the presence of a second ABC cassette transporter for Zn2+ in S. pneumoniae (AdcABC) suggests that Zn2+ transport is unlikely to be the principal physiological function for PsaBCA (Dintilhac et al., 1997). Downstream of psaA is psaD, which encodes a thiol peroxidase (Novak et al., 1998). Although psaD has its own promoter, it is believed that there is significant readthrough from the psa promoter (Novak et al., 1998). It has also been reported that psaD mutants have a requirement for added Mn2+, despite PsaD not being part of a Mn2+ transporter (Novak et al., 1998).
Streptococcus pneumoniae is a facultative anaerobe, but it is able to adapt to conditions of high oxygen tension, thereby enabling the organism to protect itself from the harmful effects of oxygen (Auzat et al., 1999; Tseng et al., 2002). However, psaA– mutants exhibit hypersensitivity to oxidative stress. This is not surprising as Mn2+ is required for protection from reactive oxygen intermediates. For example, manganese superoxide dismutase, SodA, an antioxidant that protects the cell from superoxide, requires Mn2+ in order to function, and psaA– mutants exhibit a 30% reduction in Sod activity in the absence of supplemented Mn2+ (Yesilkaya et al., 2000; Tseng et al., 2002). However, psaA– mutants are just as sensitive to paraquat, a redox compound that generates superoxide in the cytoplasm (Hassett et al., 1987), in the presence or absence of added Mn2+. Therefore, sensitivity to superoxide does not appear to relate to the level of SodA activity. psaA– mutants are also sensitive to hydrogen peroxide even in the presence of Mn2+ supplement (Tseng et al., 2002).
There have been conflicting observations from various laboratories regarding the phenotype of psa operon mutants. Novak et al. (1998; 2000) reported that the permease complex had a pleiotropic effect on pneumococcal physiology, including adherence, chain length formation, expression of choline-binding proteins and susceptibility to autolysis. This last property was of crucial importance, as resistance to autolysis confers tolerance to penicillin. However, other groups (Berry and Paton, 1996; Dintilhac et al., 1997; Claverys et al., 1999) did not notice such effects. One of the reasons put forward by Novak et al. (2000) to explain these conflicting reports was that all the psaA mutants were generated by insertion–duplication mutagenesis, but using different psaA-targeting fragments and different non-replicative plasmid vectors. This would result in truncations of PsaA of varying lengths, as well as the potential for differing polar effects on psaD. Given the potential of PsaA as a vaccine candidate for prevention of pneumococcal disease and as an antimicrobial target, it becomes evident that there is a need for a full investigation of the psa locus to determine its role in pneumococcal biology. In the present study, we have constructed unmarked in frame deletion mutants of psaB, psaC, psaA, psaD, psaBC, psaBCA and psaBCAD in the S. pneumoniae serotype 2 strain D39. The mutants were then subjected to comprehensive phenotypic characterization.
Results
Construction of psa deletion mutants
To date, the majority of psa locus mutants of S. pneumoniae have been constructed using insertion–duplication mutagenesis. However, this technique may result in the production of truncated gene products, as well as polar effects on transcription of downstream genes such as psaD. To circumvent this problem, we constructed unmarked, in frame deletion mutants ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaD, ΔpsaBC, ΔpsaBCA and ΔpsaBCAD of S. pneumoniae D39 by overlap extension polymerase chain reaction (PCR) mutagenesis, essentially as described by Ho et al. (1989) and Horton (1993). The location and extent of each deletion generated is depicted in Fig. 1. In order to distinguish deletion derivatives easily from the parent strain, an insertion–duplication mutant of psaA (psaA–; Table 1), which is resistant to erythromycin, was used as the recipient strain. This strain was constructed in our laboratory and has been used in previous studies of the psa locus (Berry and Paton, 1996). The deleted genes were used to replace the wild-type alleles by homologous recombination, and transformants were screened for loss of erythromycin resistance by replica plating on to Mn BA and Mn Ery BA plates (see Experimental procedures). The flanking regions of putative mutants were PCR amplified, and in frame deletions were confirmed by sequencing (Table 2).
Physical map of the psaBCAD locus showing the deletion resulting from each of the psa mutant strains constructed in this study. The region and extent of each deletion are depicted by dashed lines.
| Description | Source (reference) | |
|---|---|---|
| S. pneumoniae strains | ||
| D39 | Capsular serotype 2 | Avery et al. (1944) |
| psaA – | EryR, psaA insertion duplication mutant of D39 | Berry and Paton (1996) |
| ΔpsaB | psaB in frame deletion mutant of D39 | This study |
| ΔpsaC | psaC in frame deletion mutant of D39 | This study |
| ΔpsaA | psaA in frame deletion mutant of D39 | This study |
| ΔpsaD | psaD in frame deletion mutant of D39 | This study |
| ΔpsaBC | psaB and psaC in frame deletion mutant of D39 | This study |
| ΔpsaBCA | psaB, psaC and psaA in frame deletion mutant of D39 | This study |
| ΔpsaBCAD | Whole psa locus in frame deletion mutant of D39 | This study |
| DP1617 | StrR, NovR, EryR | Shoemaker et al. (1979) |
| E. coli strains | ||
| DH5α | F′/endA1 hsdR17(rK– mK+)glnV44 thi-1 recA1 gyrA (Nalr) relA1Δ(lacIZYA–argF)U169 deoR(φ80 dlacΔ(lacZ)M15) | Bethesda Research Laboratory |
| M15 | E. coli K-12 derivative/NaIS.StrS/RifS/F–/thi–/lac–/ara+/gal+/mtl+/recA+/uvr+/lon+ | Qiagen |
| Plasmid | Characteristics | Source |
|---|---|---|
| pQE31 | AmpR, His6 tag expression vector | Qiagen |
| Primers | Sequencea | Locationb,c |
|---|---|---|
| LM1 | 5′-gcg acc gaa gtt gta gaa gaa ctc-3′ | nt 1467087–1467110 |
| LM2 | 5′-cag act aat ttg cag gaa ggc aag-3′ | nt 1474227–1474204 |
| LM3 | 5′-tag gtc acc gag gtt ttc gat acg tat cat-3′ | nt 1469107–1469087, end of tail ntd 1469806–1469798 |
| LM4 | 5′-atc gaa aac ctc ggt gac cta tga ttg cag aat tta tcg-3′ | nt 1469798–1469824, end of tail nt 1469097–1469107 |
| LM5 | 5′-ttg ttt ggg agc gat tcc atc gat aaa ttc tgc aat cat-3′ | nt 1469829–1469806, end of tail complementary nt 1470615–1470601 |
| LM6 | 5′-atc gat tga atc gct ccc aaa caa cga tat ttg-3′ | nt 1470601–1470624, end of tail nt 1469821–1469829 |
| LM7 | 5′-gta gct gtc gcc ttc tcc gct agc aca tgc tac aag aat-3′ | nt 1470755–1470731, end of tail nt 1471561–1471547 |
| LM8 | 5′-agc gga gaa ggc gac agc tac tac agc-3′ | nt 1471547–1471567, end of tail nt 1470750–1470755 |
| LM9 | 5′-atc cac gta ttc aac agt tac cat agg ata ctc caa tct-3′ | nt 1471773–1471750, end of tail nt 1472172–1472158 |
| LM10 | 5′-atg gta act gtt taa tac gtg gat aat atc aat tct3′ | nt 1472158–1472184, end of tail nt 1471765–1471773 |
| LM11 | 5′-gat tgg agG atc cta tgg taa ctt-3′ | nt 1471724–1471747 (BamHI site is underlined) |
| LM12 | 5′-aga gca ggg aaA Gct ttg tga cag-3′ | nt 1472263–1472240 (HindIII site is underlined) |
| LM13 | 5′-gtc atc caa agg tag ggc agg c-3′ | nt 1468104–1468125 |
| LM14 | 5′-cag gac aag gta gag cta gca ga-3′ | nt 1473239–1473217 |
| LM15e | 5′-tcc tga aag tta tct tta gaa tct att-3′ | nt 1468998–1468972 |
| LM16 | 5′-ttg ttt ggg agc gag gtt ttc gat acg tat cat-3′ | nt 1469107–1469087, end of tail nt 1469826–1469821 |
| LM17 | 5′-atc gaa aac ctc gct ccc aaa caa cga tat ttg-3′ | nt 1470604–1470624, end of tail nt 1469097–1469107 |
| LM18 | 5′-gct gtc gcc ttc gag gtt ttc gat acg tat cat-3′ | nt 1469107–1469087, end of tail nt 1471558–1471547 |
| LM19 | 5′-atc gaa aac ctc gaa ggc gac agc tac tac agc-3′ | nt 1471547–1471567, end of tail nt 1469097–1469107 |
| LM20 | 5′-atc cac gta ttc aac gag gtt ttc gat acg tat cat-3′ | nt 1469107–1469087, end of tail nt 1472172–1472158 |
| LM21 | 5′-gaa aac ctc gtt gaa tac gtg gat aat atc aat tct-3′ | nt 1472158–1472184, end of tail nt 1469097–1469107 |
- a . For primers used in overlap extension PCR, the sequence of the region that anneals to the complementary sequence of the second primer is in bold. The region that cannot bind to DNA in the first PCR step is shown in italics.
- b . Location in genome sequence of S. pneumoniae R6 strain (Hoskins et al., 2001); GenBank accession number AE007317.
- c . Capital letters in the oligonucleotide sequence indicate mismatched bases, introduced to create a restriction site.
- d . Sequence location in genome of the end of the ‘tail’ region, which does not bind to DNA in the first PCR step of overlap extension PCR (this region is indicated by italics in the sequence).
- e . Used for sequencing ΔpsaB, ΔpsaBC, ΔpsaBCA and ΔpsaBCAD.
| Strain | Primers used | Location of deleted region in R6 genomea | Position of amino acids deleted |
|---|---|---|---|
| ΔpsaB | LM1/LM3 | 1469108–1469797 | Ser-8–Gly-237 of PsaB |
| LM2/LM4 | |||
| ΔpsaC | LM13/LM5 | 1469830–1470625 | Leu-9–Leu-273 of PsaC |
| LM14/LM6 | |||
| ΔpsaA | LM1/LM7 | 1470756–1471546 | Lys-24–Lys-287 of PsaA |
| LM2/LM8 | |||
| ΔpsaD | LM1/LM9 | 1471774–1472157 | Gly-22–Tyr-149 of PsaD |
| LM2/LM10 | |||
| ΔpsaBC | LM13/LM16 | 1469108–1470625 | Ser-8 of PsaB–Leu-237 of PsaC |
| LM14/LM17 | |||
| ΔpsaBCA | LM13/LM18 | 1469108–1471546 | Ser-8 of PsaB–Lys-287 of PsaA |
| LM14/LM19 | |||
| ΔpsaBCAD | LM13/LM20 | 1469108–1472157 | Ser-8 of PsaB–Tyr-149 of PsaD |
| LM14/LM21 |
- a . Location in genome sequence of S. pneumoniae R6 strain (Hoskins et al., 2001); GenBank accession number AE007317.
psa deletion mutants require Mn 2+ supplementation for optimal growth
The growth of two separate psaA mutants has been demonstrated previously to be dependent on Mn2+ concentration (Dintilhac et al., 1997; Marra et al., 2002), although this requirement was not discovered initially because media used routinely to culture pneumococci, such as THY, SB and CAT-based media, contain sufficient Mn2+ to allow psaA– strains to grow normally. In this study, all mutants appeared to grow as well as D39 in these media. In order to verify the requirement for Mn2+ for optimal growth in the various permease mutants constructed in this study, the semi-synthetic C+Y medium originally described by Lacks and Hotchkiss (1960) was modified. Phenotypic effects of psa mutations were only evident when the concentration of yeast extract was lowered to 0.5%. Other groups have used slightly different adaptations of this recipe (e.g. 0.8% yeast extract) to enable similar phenotypic discrimination, which presumably reflects slight differences in Mn2+ content of the yeast extract used (Martin et al., 1995; Dintilhac et al., 1997; Novak et al., 1998).
For growth measurements, C+Y and C+Y+M were inoculated from frozen stocks of the various mutants such that the starting culture density was identical for all cultures. The cultures were then incubated at 37°C for 8 h, with the A600 measured every 2 h. D39 and ΔpsaD grew equally well in both media (Fig. 2). However, a marked difference in growth between the two media was clearly observable for the other mutants. By the end of the 8 h period, the culture densities of the C+Y+M cultures were two to three times that of the C+Y cultures, clearly indicating the requirement for exogenous Mn2+ to restore optimal growth in the mutants. The requirement for Mn2+ supplementation for optimal growth was also evident when the bacteria were grown for ≈ 16 h at 37°C in 95% air, 5% CO2 on BA plates supplemented with 0.4 µM Mn2+. When the mutants were grown on BA plates without added Mn2+ under the same conditions, only very small, pin-point colonies were present, except for ΔpsaD, which still formed large colonies.
Growth of various D39 derivatives in low and high Mn2+ media. The strains were inoculated from frozen stock into C+Y and C+Y+M (C+Y supplemented with 3 µM MnSO4) to investigate the Mn2+ requirement for growth. A600 was measured every 2 h over an 8 h period. Growth measurements in C+Y medium are represented by squares, while those for C+Y+M medium are represented by triangles. Data are the mean ± standard deviation of two independent experiments.
Transformability of psa mutants
A requirement for exogenous Mn2+ for the development of competence in a psaA insertion–duplication mutant of a rough S. pneumoniae strain has been reported previously (Dintilhac et al., 1997). There has also been a report that insertion–duplication mutants of the other three psa genes are also transformation deficient, but wild-type levels of transformation were not restored by supplementation with Mn2+ (Novak et al., 1998). In this study, the level of transformation of Psa permease operon mutants (see Experimental procedures) was also found to be very low (≈ 0.00004%) when cultured in C+Y media. However, normal levels of transformation were observed when the mutants were cultured in C+Y+M; the highest level of transformants (≈ 0.015–0.025%) was obtained at a culture density of A600 = 0.08 for all strains. D39 and ΔpsaD transformed equally well in both media, consistent with their identical growth in both media (data not shown).
psa mutants are sensitive to killing by superoxide
Previously, we have shown that a psaA insertion–duplication mutant was very sensitive to killing by paraquat, which generates superoxide inside the bacterial cell (Tseng et al., 2002). Here, we examined unmarked psa deletion mutants to determine their sensitivity to superoxide. The psaB, C, A, BC, BCA and BCAD mutants were highly susceptible to killing by paraquat, especially the psaA, BC, BCA and BCAD mutants, which were killed completely after 10 min (Fig. 3). However, the psaD mutant exhibited a level of sensitivity to paraquat that was similar to that of the wild-type strain, D39 (Fig. 3).
Paraquat killing of D39 and psa mutants. Experiments were carried out with triplicate samples, and each experiment was done at least three times. The figure presented here is typical of an individual experiment. Data are the mean ± standard deviation of triplicate samples. All strains containing the psaA mutation were killed at the 10 min time point.
psa mutants are sensitive to killing by hydrogen peroxide
We have reported previously that psaA and psaD insertion–duplication mutants are very sensitive to H2O2 (Tseng et al., 2002). In order to determine whether the unmarked psa deletion mutants were sensitive to H2O2, a survival test was carried out. Compared with the wild-type D39, all the psa mutants were very sensitive to H2O2(Fig. 4).
Hydrogen peroxide survival assay of D39 and psa mutants. Experiments were carried out with triplicate samples, and each experiment was done at least three times. The figure presented here is typical of an individual experiment. Data are the mean ± standard deviation of triplicate samples.
psa mutants accumulate less Mn 2+ during growth
The amount of Mn2+ associated with cells grown in THY broth supplemented with 1 µM Mn2+ was measured by ICP-MS (see Experimental procedures). The cells were washed vigorously with PBS three times before ICP-MS analysis, so most Mn2+ measured was believed to be inside the cells rather than attached to cell surfaces. Table 3 shows that all the psa mutants, except the ΔpsaD mutant, contained approximately fivefold less Mn2+ than the wild-type D39. The ΔpsaD mutant had a similar level of Mn2+ to the wild type.
| Strains | Mn2+ conc. (ng Mn2+ g−1 dry weight) |
|---|---|
| D39 | 26 409 ± 1289 |
| ΔpsaB | 5184 ± 857 |
| ΔpsaC | 5259 ± 1005 |
| ΔpsaA | 4766 ± 726 |
| ΔpsaD | 32 699 ± 10 374 |
| ΔpsaBC | 5367 ± 894 |
| ΔpsaBCA | 5346 ± 426 |
| ΔpsaBCAD | 4380 ± 1292 |
- The amount of Mn2+ was determined by ICP-MS in two independent experiments with triplicate samples for each experiment. Data are mean ± standard deviation.
Effect of mutations on expression of PsaA, PsaD and choline-binding proteins
There have been conflicting reports regarding the effect of insertional inactivation of the psa locus on the expression of pneumococcal choline-binding proteins (CBPs) (Novak et al., 1998; Claverys et al., 1999). Western blot analysis was therefore used to assess the levels of the CBPs PspA, LytA and CbpA, as well as levels of PsaA and PsaD in the various psa deletion mutants. The results show that all mutants produced wild-type levels of PspA, LytA and CbpA when cultured in C+Y, indicating that there is no effect on CBP production (Fig. 5). Western blot analysis was also conducted on lysates of cells grown in THY and C+Y+M media, producing essentially the same results (data not shown).
Western blot analysis of various D39 derivatives for expression of PsaA, PsaD and choline-binding proteins. Whole-cell lysates of all strains were prepared in lysis buffer after growth in C+Y medium from a starting A600 = 0.04 to a final A600 = 0.15. Proteins were separated by SDS–PAGE and blotted on to nitrocellulose. Filter panels were probed separately with polyclonal antisera specific for CbpA (≈ 90 kDa), PspA (≈ 65 kDa), PsaA (37 kDa) LytA (36 kDa) or PsaD (18 kDa).
All mutants, apart from ΔpsaD and ΔpsaBCAD, produced the same level of PsaD as D39 when cultured in THY, C+Y and C+Y+M media, indicating that neither the vector inserted into psaA to create psaA– nor the deletion of upstream genes affects the expression of psaD. With regard to PsaA levels, no PsaA was detected in ΔpsaA, ΔpsaBCA and ΔpsaBCAD as expected, but a truncated derivative of PsaA (≈ 25 kDa) was detectable in psaA–, consistent with insertion of the vector pVA891 into the later portion of the gene (Novak et al., 1998). In all media, ΔpsaC had wild-type levels of PsaA, but ΔpsaB had lower levels, whereas ΔpsaBC had levels similar to that seen in the wild type.
Chain formation and autolysis
All deletion mutants, D39 and psaA– were confirmed to be serotype 2 S. pneumoniae by quellung reaction. D39, the deletion mutants and psaA– displayed a normal chain length phenotype, growing as diplococci or in chains of up to eight cells in THY, C+Y and C+Y+M media (results not shown). The mutants also lysed in the presence of DOC in these media, confirming the production of active LytA (the major pneumococcal autolysin), as demonstrated in the Western blot.
Virulence studies
Previous work has indicated the importance of the Psa proteins in virulence (Berry and Paton, 1996; Marra et al., 2002). In this work, virulence studies were undertaken in order to investigate the contribution of each gene, as well as combinations of genes, to the pathogenesis of disease. The importance of the Psa proteins in systemic disease was investigated using a mouse intraperitoneal challenge model, while the significance of PsaA, PsaD, PsaBCA and PsaBCAD in nasopharyngeal colonization was determined using a mouse intranasal challenge model.
In the intraperitoneal challenge model, mice challenged with ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaBCA or ΔpsaBCAD survived the entire study, while the only death in the ΔpsaBC group appeared not to result from pneumococcal sepsis (Fig. 6). However, there was no significant difference in the median survival time of mice challenged with ΔpsaD compared with those challenged with D39.
Survival times for mice after i.p. challenge. Groups of 10 BALB/c mice were challenged i.p. with ≈ 5 × 105−1 × 106 cfu of the indicated strains. Each datum point represents one mouse. The horizontal broken lines denote the median survival time for each group.
In the nasopharyngeal colonization studies, only D39, ΔpsaA, ΔpsaD, ΔpsaBCA and ΔpsaBCAD were evaluated for their capacities to colonize the nasopharynx because of logistical considerations. The ΔpsaA, ΔpsaBCA and ΔpsaBCAD mutants were clearly defective in colonization and were not detectable (<40 cfu per nasopharynx) after 48, 24 and 24 h respectively (Fig. 7). These differences in colonization capacity among the groups were significant (P < 0.05, one-way anova). However, there was no obvious difference in colonization capacity between ΔpsaD and D39.
Bacterial recovery from the nasopharynx of CD-1 mice after i.n. challenge with ≈ 4 × 106−2 × 107 cfu of the indicated strains over a 4 day period. ‘X’ denotes that no bacteria were recovered at the indicated time points, the limit of detection being 40 cfu per nasopharynx. The data are the mean ± standard error of the mean (n = 4) for each time point.
Discussion
The increasing prevalence of antibiotic-resistant pneumococci and limitations of current polysaccharide-based vaccines has led to an intense search for virulence proteins of S. pneumoniae in recent years, with the hope of designing new vaccines and antimicrobial agents. One candidate protein of particular interest is the pneumococcal surface antigen A (PsaA). Virulence and immunization studies have shown the potential of PsaA as a vaccine antigen, particularly against mucosal infection (Berry and Paton, 1996; Talkington et al., 1996; Briles et al., 2000a,b; Marra et al., 2002). However, Novak et al. (1998) raised a concern about using PsaA as a vaccine antigen. In that paper, psa mutants were reported to have defective LytA production and, therefore, were autolysis defective, resulting in penicillin tolerance. The promotion of tolerance to penicillin and other β-lactam antibiotics is an extremely undesirable characteristic for either a vaccine or an antimicrobial agent. However, Claverys et al. (1999) disagreed with this result, as well as other observations made by Novak et al. (1998) concerning the phenotype of psa mutants. Attempts to explain discrepant findings have generally centred on the likelihood that the insertion–duplication mutagenesis technique used by the various groups to construct psa mutants might result in the production of truncated (perhaps partially active) proteins or have polar effects on downstream sequences (Claverys et al., 1999; Novak et al., 2000). In this study, we attempted to resolve these discrepancies by constructing unmarked, in frame deletion mutants for each of the psa locus genes and combinations thereof. The mutants constructed for this study were ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaD, ΔpsaBC, ΔpsaBCA and ΔpsaBCAD, and they were used to clarify the role of the psa locus in CBP expression and chain length formation, to confirm/determine the requirement for added Mn2+ for growth, transformability and resistance to oxidative stress in psa mutants and to continue the investigation into the effect of mutating psa genes on the pathogenesis of pneumococcal infection.
Effect of psa deletions on protein expression, chain length formation and autolysis
Results from this study do not support a role for the psa locus in CBP production or chain formation. All deletion mutants displayed a wild-type phenotype with regard to chain length in broth culture, and wild-type levels of the choline-binding proteins LytA, CbpA and PspA were detected by Western blot analysis. The mutants lysed in the presence of low and high concentrations of DOC, also indicating normal LytA function.
Novak and colleagues suggested that the psaA– mutant constructed by Berry and Paton (1996) was producing a truncated version of PsaA with more biological function than seen in their mutant, which had a different vector inserted at an earlier point in the psaA ORF (Novak et al., 2000). They believed that different regions of the gene might encode different functions. In the Western blot (Fig. 5), a truncated version of PsaA in the lysate of psaA– was indeed identified, the size of which (≈ 25 kDa) was consistent with its genetic structure. However, no anti-PsaA reactive species was detected in ΔpsaA as expected (Fig. 5). As these two mutants had the same phenotype with respect to chain length and CBP expression, it does not appear that the truncated version of PsaA influenced previously published results or that different regions of psaA encode different functions.
One of the potential drawbacks of the insertion–duplication mutagenesis technique is the possibility of polar effects on downstream genes. Novak et al. (2000) believed that neither termination of transcript nor readthrough, which would affect the expression of psaD, could be ruled out with the vector, pVA891, that was used by Berry and Paton (1996) to generate psaA–. However, as inserting pVA891 between psaA and psaD did not affect virulence, and mutating psaD did not reduce virulence to the extent seen with the psaA– mutant, it seemed doubtful whether pVA891 insertion affected the expression of psaD (Berry and Paton, 1996; Claverys et al., 1999). In this study, antibody to PsaD was generated for the first time and used to investigate the levels of PsaD in the deletion mutants generated in this study, as well as in the psaA– mutant generated earlier. Wild-type levels of PsaD were evident in psaA– as well as in the deletion mutants, apart from ΔpsaD and ΔpsaBCAD. Therefore, insertion–duplication mutagenesis of the psaA gene did not seem to affect the transcription of psaD. Recently, a regulator of the Sca permease operon, which is the Streptococcus gordonii equivalent of the Psa operon, was identified and shown to have no influence on the expression of a related thiol peroxidase downstream (Jakubovics et al., 2000).
Although in frame deletion of psa genes would be expected to have a much lower impact on transcription of the downstream sequences than insertion of mutagenic plasmids, some effects on downstream gene expression might still occur. Indeed, this was clearly evident in the Western blots, where ΔpsaB had lower levels of PsaA compared with D39, while the amount of PsaA expressed by ΔpsaBC was similar to that seen in the wild type. There are several explanations that may account for these observations. Deleting psaB may make the transcript less stable or psaB may contain some element that influences transcription. When both psaB and psaC are deleted, the stability of the psaA mRNA may increase.
The psa locus and Mn2+ transport
It was shown previously that mutants of psaB, psaC, psaA and psaBCA grew poorly in media with low concentrations of Mn2+ (Dintilhac et al., 1997; Marra et al., 2002). The present study also supports the role for the PsaBCA permease complex as the primary transporter of Mn2+ in S. pneumoniae, particularly as a search of the genomes of the TIGR 4 strain and R6 (an unencapsulated derivative of D39) did not reveal the presence of a homologue of the manganese ion NRAMP transporter MntH found in certain other bacteria. All mutants, except for ΔpsaD, exhibited poor growth in C+Y, compared with that seen in the same medium supplemented with 3 µM Mn2+ (C+Y+M). Presumably, at this higher concentration of Mn2+, secondary Mn2+ transporters with a lower affinity for Mn2+ are able to transport it into the cell. This notion is supported by ICP-MS data generated in this study, which shows that, with the exception of the ΔpsaD mutant, approximately fivefold less Mn2+ was accumulated inside the cells of the mutants compared with the D39 parent, even in the presence of exogenous Mn2+. It has been reported that gene replacement deletion mutants of the Psa permease genes only grow to a culture density of about 67% of that of the wild-type strain in brain–heart infusion broth supplemented with Mn2+ but have a shorter lag time (Marra et al., 2002). However, in the present study, the culture densities of the various mutants grown in C+Y+M were at least as high as that of the parent and, with the exception of ΔpsaB, ΔpsaBCA and ΔpsaBCAD mutants, the lag time in the mutants did not appear to be significantly shorter.
An insertion–duplication mutant of psaD was reported to have a slower growth rate in low Mn2+ media (Novak et al., 1998); however, the present study shows that the growth of ΔpsaD is not affected by Mn2+ concentration. As PsaD is reported to be an antioxidant, rather than a component of an ABC permease, it would not be expected to have a direct role in Mn2+ transport. Furthermore, the results from this study suggest that it does not even have an indirect role, such as involvement in a signal pathway crucial for the expression of the Psa permease operon. The requirement for Mn2+ supplementation for optimal growth was also clearly evident when the various psa mutants were grown on BA. Apart from ΔpsaD, the growth of the mutants on BA was visibly poorer compared with growth on BA supplemented with MnCl2.
Requirement for Mn2+supplementation for transformation
An insertion–duplication mutant of psaA has been shown previously to require Mn2+ supplementation for efficient transformation (Dintilhac et al., 1997). In contrast, Novak et al. (1998) found that their psaA– mutant, also constructed using insertion–duplication mutagenesis, did not have such a deficiency, even though mutants of the other three psa genes did. Moreover, the addition of Mn2+ did not restore normal transformation levels in the other psa mutants. However, the findings of the present study agree with those of Dintilhac et al. (1997) regarding the need for the addition of exogenous Mn2+ for transformation of ΔpsaA and other deletion mutants of the Psa permease operon.
A role for the psa locus in pathogenesis
Colonization of the nasopharynx is an important step in pneumococcal pathogenesis. ΔpsaA, ΔpsaBCA and ΔpsaBCAD mutants showed a statistically significant decrease in their ability to colonize the nasopharynx, whereas deleting psaD seemed to have no effect. During the course of the experiment, two mice in the D39 group and four mice in the ΔpsaD group died from pneumococcal infection, consistent with the results of a previous intranasal challenge study (Berry and Paton, 1996), but no deaths were recorded for mice in the other three groups. The results of the intraperitoneal study further highlight the importance of the Psa permease in the virulence of S. pneumoniae. All mice challenged intraperitoneally with ≈ 105 cfu of ΔpsaB, ΔpsaC, ΔpsaA, ΔpsaBCA and ΔpsaBCAD survived the 14 day experiment. There was also only one death in the ΔpsaBC group, which appeared to be unrelated to pneumococcal sepsis. In contrast, 8/10 and 9/10 mice died in the D39 and ΔpsaD groups, respectively, and there was no statistically significant difference in median survival time between these two groups. As expected from previous findings (Tseng et al., 2002), all the mutants, including ΔpsaD, were very sensitive to H2O2. However, sensitivity to H2O2 does not seem to affect the virulence of the ΔpsaD mutant. Interestingly, all these mutants, except ΔpsaD, were highly sensitive to killing by superoxide. Sensitivity of the various mutants to superoxide correlates inversely with virulence, and this suggests that superoxide-mediated killing mechanisms may be more important than H2O2 in host defence against S. pneumoniae.
Taken together, these results agree with the notion that ABC permeases specific for essential nutrients are extremely important for virulence (Quentin and Fichant, 2000; Lau et al., 2001). In fact, in a recent large-scale identification of virulence genes in S. pneumoniae using signature-tagged mutagenesis, the most abundant class of putative virulence proteins detected was transporters, the majority of which were ABC permeases (Lau et al., 2001). In addition, all mutants that could not be transformed efficiently in a low Mn2+ medium showed dramatically reduced virulence in a mouse systemic infection model. In contrast, ΔpsaD, which did not demonstrate transformation deficiency in C+Y, did not exhibit reduced virulence.
Deleting psaB, psaC, psaA or combinations of these genes had essentially the same impact on virulence. Therefore, it appears as though no individual component of the permease is more important than another. Rather, it is the presence of a complete, functioning permease that is essential for virulence, and there appears to be no cross-talk with other permeases. For example, Mn2+ cannot bind to PsaA and then enter the cell via the channel of another permease. Obviously, the Psa permease is essential in multiple stages of pneumococcal disease. Not only does mutating the genes affect colonization of the nasopharynx, bypassing this vital first step in pneumococcal disease by introducing the bacteria directly into the peritoneal cavity does not result in death of the mice. Several of the known properties of psaA mutants would be expected to have a marked impact on the capacity to colonize and proliferate within the host. psaA mutants do not grow well in environments with low levels of Mn2+, they have a reduced capacity to adhere to lung cells and are hypersensitive to oxidative stress (Berry and Paton, 1996; Dintilhac et al., 1997; Tseng et al., 2002). In this study, we establish that mutating psaB and psaC also affects sensitivity of S. pneumoniae to oxidative stress. Although the reason for this phenotype is still not fully understood, the use of microarrays could give a clearer picture of the cell components and pathways that may be associated with this phenomenon.
Experimental procedures
Bacterial strains, plasmids, growth conditions and transformation
The bacterial strains and plasmids used in this study are listed in Table 1. S. pneumoniae strains were routinely grown in Todd–Hewitt broth (Oxoid) with 1% Bacto yeast extract (THY) or on blood agar (BA) plates [39 g l−1 Columbia base agar (Oxoid), 5% (v/v) defibrinated horse blood]. To make Mn BA plates, 20 µM of MnCl2 was added to BA. Alternatively, cells were grown in serum broth (SB) 10% (v/v) donor horse serum in nutrient broth [10 g l−1 peptone (Oxoid), 10 g l−1 Laboratory Lemco powder (Oxoid) and 5 g l−1 NaCl]. Erythromycin (Ery) and gentamicin (Gent) were added to growth media where appropriate at concentrations of 0.2 µg ml−1 and 2.5 µg ml−1 respectively. Opaque and transparent opacity colony phenotypes were differentiated as described by Weiser et al. (1994). In order to confirm that bacteria were S. pneumoniae, strains were tested for optochin sensitivity by plating on Mn BA in the presence of an optochin disc (Oxoid). The production of type 2 capsule was assessed by the quellung reaction, using diagnostic pneumococcal typing serum produced by Statens Seruminstitut (Copenhagen, Denmark). Chain lengths of the pneumococci were also noted at the same time. Strains were also tested for susceptibility to complete lysis in the presence of sodium deoxycholate (DOC) at concentrations of 0.1% and 1% (v/v).
Escherichia coli strains were grown in Luria–Bertani broth [LB; 10 g l−1 tryptone-peptone (Difco), 5 g l−1 yeast extract, 5 g l−1 NaCl] or on LB agar plates (LB with 15 g of agar). Where appropriate, ampicillin (Amp) or kanamycin (Kan) was added to the growth medium at concentrations of 50 and 25 µg ml−1 respectively.
Streptococcus pneumoniae psaA – cells were grown for competence and transformation in a semi-synthetic casein hydrolysate medium supplemented with 0.5% yeast extract (C+Y medium) (Lacks and Hotchkiss, 1960) essentially as described previously (Martin et al., 1995) in the presence of 50 ng of competence stimulating peptide-1 (CSP-1; Havarstein et al., 1995) with 3 µM MnSO4 (C+Y+M) or without MnSO4 (C+Y). Transformation of E. coli K-12 with plasmid DNA was carried out with CaCl2-treated cells as described by Brown et al. (1979).
Oligonucleotide primers, DNA isolation and manipulation
Primers used in this study are listed in Table 1. S. pneumoniae chromosomal DNA was extracted, purified and analysed as described previously (Morona et al., 1999). DNA amplification was performed by high-fidelity PCR using the Expand Long Template PCR system (Roche). Overlap extension PCR products were generated from initial PCR products using the method described by Ho et al. (1989) and Horton (1993). Amplification products were purified using a Qiagen UltraClean PCR CleanUp DNA purification kit and sequenced using the Big Dye system (Applied Biosystems) on a model 3700 automated sequencer.
In vitro growth measurements and transformability assay
Frozen stocks were prepared by growing strains at 37°C in CAT medium (Porter and Guild, 1976) to an A600 of ≈ 0.2. The cultures were concentrated 20× in CAT, glycerol was added to a final concentration of 15% and then stored at −80°C. For the generation of a growth curve, cultures from frozen stock were added to C+Y and C+Y+M (0.5% yeast) media (Martin et al., 1995) and adjusted so that each strain had a starting A600 of 0.02 for both media. Cultures were then incubated at 37°C, and absorbance was measured at 2 h intervals.
For transformability assay, cultures from frozen stock were added to C+Y and C+Y+M media and adjusted so that each strain had a starting A600 of 0.02 for both media. The cultures were then incubated at 37°C. After a culture density of 0.075–0.08 was reached, 1 ml aliquots were taken every 15 min for 2 h. The A600 of each culture was measured at each time interval, and a sample was also diluted and plated for viable counts. The samples were then concentrated 10× in C+Y medium to which glycerol was added to a final concentration of 15% and stored at −80°C.
Transformability assay was performed essentially as described previously (Martin et al., 1995; Dintilhac et al., 1997). Briefly, 25 µl of frozen cells was thawed and added to 250 µl of C+Y or C+Y+M. After the cells had been incubated for 15 min at 37°C, 5 µl of genomic DNA from S. pneumoniae strain DP1617 was added. Cells were then incubated for 2 h at 37°C. Before plating, cells were serially diluted and then spotted on to Mn Ery BA plates.
Paraquat killing assay
Paraquat killing assay has been described previously (Tseng et al., 2002). This assay was carried out aerobically, i.e. the cells were exposed to atmospheric oxygen during the assay.
H2O2 survival assay
H2O2 survival assay was essentially as described previously (Tseng et al., 2002). This assay was also carried out under aerobic conditions. Briefly, 107 cells were exposed to 50 mM H2O2 at room temperature for 15 min in 100 µl of THY broth, then cells were plated for viable counts.
Inductively coupled plasma – mass spectrometry (ICP-MS)
The amount of Mn2+ taken up by S. pneumoniae cultures was determined by growing cells overnight in THY broth supplemented with 1 µM Mn2+, washing the cells three times in phosphate-buffered saline (PBS) and resuspending in MilliQ water. Half the washed cells were used to determine dry weight, and the other half were analysed using a Fisons PQ2+ ICP-MS at the Australian Centre for Queensland University Isotope Research Excellence to determine the amount of Mn2+ in the sample. Results were expressed as ng of Mn2+ g−1 dry weight of cells.
Expression and purification of PsaD
The complete PsaD open reading frame (ORF) from S. pneumoniae D39 was amplified using primers LM11 and LM12 (Table 1). The PCR product was digested with BamHI and HindIII, cloned into the corresponding restriction sites in pQE31 and transformed into E. coli K-12 expression strain M15 (Qiagen) to generate a His6 fusion protein. High-level expression of His6–PsaD was induced by the addition of 2 mM IPTG. The recombinant protein was then purified by nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography (Qiagen).
SDS–PAGE and Western blotting
Bacteria for SDS–PAGE were lysed in lysis buffer and subjected to SDS–PAGE using the method described by Laemmli (1970). Separated proteins were electroblotted on to nitrocellulose (Pall Life Sciences) as described by Towbin et al. (1979). After transfer, the membrane was probed with specific polyclonal antisera at a dilution of 1:3000 and then reacted with blotting-grade goat anti-mouse–alkaline phosphatase conjugate (Bio-Rad Laboratories).
Immunization of mice and analysis of sera
His6–PsaD antigen was formulated in aluminium hydroxide adjuvant (Alum) to a final ratio of 100 µg of antigen to 1 mg of Alum adjuvant. CD-1 mice were then immunized intraperitoneally with three doses of 10 µg of the His6–PsaD antigen formulation at 14 day intervals. Polyclonal mouse serum was collected by cardiac puncture 7 days after the final immunization and stored at 4°C until use. Mouse antisera for recombinant fragments of CbpA, PspA, LytA and PsaA were prepared in the laboratory previously using methods similar to that described for PsaD.
Challenge of mice
Strains of S. pneumoniae to be used for intraperitoneal (i.p.) challenge studies were inoculated from an overnight blood agar plate into serum broth and grown to a culture density of ≈ 1 × 108 cfu ml−1. Alternatively, for intranasal (i.n.) studies, strains were grown in THY to a similar culture density. The bacteria were then diluted to the appropriate dose in sterile SB or THY, so that 100 µl aliquots for i.p. and 10 µl aliquots for i.n. contained the required challenge dose (≈ 106 cfu and ≈ 107 cfu respectively). The actual dose administered was determined retrospectively by plating serial dilutions of the challenge inocula after administration to the mice.
After i.p. challenge, mice were monitored closely at 4 h intervals for signs of disease and time of death, which were recorded. For i.n. challenge, 10 µl of bacterial suspension, containing ≈ 107 bacteria, was slowly pipetted into the nares and involuntary inhaled. After challenge, five mice per group were sacrificed by CO2 asphyxiation at 24, 48 and 96 h. After exposure of the trachea, the nasopharynx was washed with 1 ml of buffer (0.5% trypsin, 0.02% EDTA in sterile PBS) and then excised and homogenized using a tissue homogenizer (Cat X120). Nasopharyngeal samples were serially diluted and spotted, in duplicate, on to BA supplemented with 60 µM MnSO4 and 2.5 µg ml−1 gentamicin. Plates were incubated for ≈ 16 h at 37°C in the presence of 95% air/5% CO2, and colonies were counted.
Statistical analysis
Median survival times were compared using the two-tailed Mann–Whitney U-test. The overall survival rates and numbers of infected mice were compared using the Fisher Exact test. Log10 cfus of bacteria in nasopharynx were compared using one-way analysis of variance (anova).
Acknowledgements
This work was supported by project grants 207722 and 252886 and programme grant 284214 from the National Health and Medical Research Council of Australia.
