The genomic analysis of Streptococcus pneumoniae strains identified the Pneumococcal adherence and virulence factor B (PavB), whose repetitive sequences, designated Streptococcal Surface REpeats (SSURE), interact with human fibronectin. Here, we showed the gene in all tested pneumococci and identified that the observed differences in the molecular mass of PavB rely on the number of repeats, ranging from five to nine SSURE. PavB interacted with fibronectin and plasminogen in a dose-dependent manner as shown by using various SSURE peptides. In addition, we identified PavB as colonization factor. Mice infected intranasally with ΔpavB pneumococci showed significantly increased survival times compared with wild-type bacteria. Importantly, the pavB-mutant showed a delay in transmigration to the lungs as observed in real-time using bioluminescent pneumococci and decreased colonization rates in a nasopharyngeal carriage model. In co-infection experiments the wild-type out-competed the pavB-mutant and infections of epithelial cells demonstrated that PavB contributes to adherence to host cell. Blocking experiments suggested a function of PavB as adhesin, which was confirmed by direct binding of SSURE peptides to host cells. Finally, PavB may represent a new vaccine candidate as SSURE peptides reacted with human sera. Taken together, PavB is a surface-exposed adhesin, which contributes to pneumococcal colonization and infections of the respiratory airways.
Streptococcus pneumoniae, more commonly known as pneumococci, are Gram-positive bacteria colonizing the human upper respiratory tract as harmless commensals. Pneumococci, however, are also responsible for a high burden of human diseases and death. In fact, these harmless bacteria can also be serious pathogens causing severe local infections including otitis media and sinusitis or even life-threatening diseases such as community-acquired pneumonia, septicaemia and meningitis (Cartwright, 2002). In particular elderly and immunocompromised individuals are at high risk of suffering serious invasive pneumococcal diseases (IPD) (Kadioglu et al., 2008). Pneumococci reside on the mucosal surface of the upper respiratory tract and use several strategies to adhere to cells of the respiratory tract. This colonization step is considered to be the initial and essential step prior to pneumococcal translocation into the lungs and bloodstream leading to IPD. The interaction between bacteria and host epithelial cells at the mucosal surface is a highly dynamic process and requires the involvement of surface-displayed bacterial adhesins and eukaryotic cellular surface receptors. Regarding the chemical nature of the bacterial adhesins, proteins are the key players in pathogen-host interactions, although carbohydrate structures are also involved in adherence. For example, human thrombospondin-1 (TSP-1) is a matricellular protein recognizing the peptidoglycan of pneumococci and other Gram-positive bacteria and this interaction facilitates bacterial adherence to host cells (Rennemeier et al., 2007). In addition, the pneumococcal cell surface contains phosphorylcholine (ChoP), which acts as an adhesin by recognizing the platelet-activating factor receptor of endothelial cells (Cundell et al., 1995). The other so far known pneumococcal adhesins are surface-displayed proteins such as lipoproteins, choline-binding proteins (CBP) and proteins with an LPxTG motif (Hammerschmidt, 2006; Kadioglu et al., 2008). CBP are non-covalently anchored to the cell wall via an interaction with ChoP, while the LPxTG proteins are covalently anchored in the cell wall after cleavage of the LPxTG sequence by a sortase (Schneewind et al., 1993; Hoskins et al., 2001; Tettelin et al., 2001). Adhesive functions were proposed for the lipoproteins PsaA, PpmA and SlrA. PsaA is part of an ABC transporter complex for manganese and thought to interact with E-cadherin (Anderton et al., 2007). SlrA, which is a peptidyl-prolyl isomerase, and PpmA contribute to colonization. However, it is assumed that PpmA and SlrA are not directly involved in adherence but accelerate folding other proteins which may directly interfere with pneumococcal adherence (Hermans et al., 2006; Cron et al., 2009). A subclass of pneumococcal strains produces pili that are encoded by pilus islet 1 or 2. Both types of pili are implicated in bacterial adherence to host cells (Barocchi et al., 2006; Bagnoli et al., 2008). A further pathogenicity-island encoded protein is PsrP. The presence of the psrP-secY2A2 island correlated positively with the ability of pneumococci to cause IPD (Obert et al., 2006; Blomberg et al., 2009). Recent studies indicated that PsrP is a protective adhesin interacting with keratin 10 on lung epithelial cells (Rose et al., 2008; Shivshankar et al., 2009). A major pneumococcal adhesin at the nasopharyngeal cavity is PspC, also referred to as CbpA or SpsA. PspC is a multifunctional CBP interacting directly and in a human-specific manner with the ectodomain of the polymeric immunoglobulin receptor (Hammerschmidt et al., 1997; 2000; Zhang et al., 2000; Elm et al., 2004). PspC binds also factor H and remarkably, this interaction promotes adherence to host cells (Dave et al., 2001; Hammerschmidt et al., 2007).
Besides direct adhesin-receptor interactions Gram-positive bacteria exploit various serum and extracellular matrix (ECM) proteins to attach to host cells (Talay, 2005). In addition to TSP-1 and factor H pneumococci were shown to interact specifically with lactoferrin, plasminogen, fibronectin and vitronectin. These interactions are implicated in pneumococcal virulence and invasion of host cells respectively (Holmes et al., 2001; Shaper et al., 2004; Bergmann et al., 2005; Pracht et al., 2005; Bergmann et al., 2009). While plasminogen- and fibronectin-binding proteins such as the surface-displayed enolase, Pce and PavA protein, respectively, were identified, the bacterial adhesin for vitronectin is not known (Holmes et al., 2001; Bergmann and Hammerschmidt, 2006; Attali et al., 2008). Fibronectin-binding activity was also reported for PfbA, which also binds plasminogen (Yamaguchi et al., 2008), and a multi-domain protein encoded by sp0082 of TIGR4 (Bumbaca et al., 2004). Both proteins contain an LPXTG motif and are suggested to be surface displayed. PfbA (encoded by spr1652 in R6) was further shown to function directly as an adhesin (Yamaguchi et al., 2008). By using the nomenclature of the fibronectin-binding protein PavA, the protein encoded by sp0082 was termed Pneumococcal adhesion and virulence factor B (PavB) (Jedrzejas, 2007), although its implication in virulence and adherence has not been explored. According to the TIGR4 genome annotation the PavB protein contains four repeats, designated as SSURE (streptococcal surface repeat) domains. The third SSURE was shown to bind to immobilized fibronectin and it is assumed that the individual SSURE bind to repetitive host fibronectin motifs (Bumbaca et al., 2004; Jedrzejas, 2007).
As the role of PavB in pneumococcal pathogenesis has not been studied, we investigated whether PavB of S. pneumoniae contributes to virulence. After conducting a detailed molecular analysis of the pavB locus in different pneumococci we demonstrated in this study, that PavB contributes to pneumococcal nasopharyngeal colonization in mice. Moreover, infections of host cells indicated that PavB expression interferes with adherence to host epithelial cells and we demonstrated that the immunogenic PavB functions directly as a pneumococcal adhesin. Due to these results we have retained the name PavB for this surface protein of S. pneumoniae.
Bioinformatic analysis of the pavB locus in pneumococci
In order to gain insight into the genomic organization of the pavB locus of pneumococcal strains, sequence alignments (blastn and blastp) were conducted using 19 available pneumococcal genome sequences (Table S1 and Fig. S1). The detailed computational analysis of the pavB-gene region, which includes the pavB ORF (e.g. TIGR4: sp0082; D39: spd0080; R6: spr0075 and G54: spn01181) as well as 1.0 kb upstream and 2.0 kb downstream flanking sequences, showed a highly conserved organization in the genomes. The inter-strain sequence alignments for this predicted surface-exposed protein confirmed the presence and conservation of a 42 amino acids signal peptide (SP) in the N-terminal region and a 108 amino acids C-terminal part, containing the hydrophobic transmembrane (TM) domain with the LPNTG peptidoglycan-anchoring motif (Fig. 1). The LPNTG is cleaved by the transpeptidase sortase and assumed to anchor PavB covalently to the cell wall. The major part of the PavB protein consists of a variable number of repetitive sequences, referred to as Streptococcal Surface REpeats (SSURE) (Bumbaca et al., 2004; Jedrzejas, 2007). Our inter-strain PavB protein and SSURE domain alignments showed that the protein variation is mainly due to a variable number of repeats, namely two, four and six for pneumococcal strains G54, TIGR4 and D39/R6 respectively (Fig. 1). According to the lengths and the sequence identities between these repeats we predicted three different kinds of repeats in the SSURE domain of PavB: the first repeat (150 amino acid residues), which is highly conserved among different strains, but differs from core SSURE and last repeat; the core repeats (each of 152 amino acid residues), which show intra- and inter-strain conservation, while their number varies among different strains; and the last repeat (136 amino acid residues), which is a C-terminally truncated SSURE but highly conserved among the strains (Fig. S1). Strikingly, the intra-strain alignment using the pavB and ssure sequences of TIGR4 identified at least one specific nucleotide for each repeat, which was employed in our molecular analysis to discriminate the ssure-sequences.
Distribution of pavB among pneumococcal strains
The genetic organization of the pavB-locus is highly conserved in the pneumococcal strains. Sequence alignments using the JVCI Comprehensive Microbial Resource (CMR) and NCBI BLAST identified SSURE homologous sequences in the three streptococcal species Streptococcus mitis (NCTC12261: SMT1642), S. agalactiae (NEM316: gbs0428) and S. gordonii (NCTC7868: SGO1182) and confirmed previous results (Jedrzejas, 2007). However, pavB orthologues were not identified in pneumococcal-related species such as S. mutans, S. pyogenes, C and G streptococci, enterococci, lactococci, and in other pathogenic microorganisms. To demonstrate the presence of the pavB gene across the pneumococcal strains and serotypes and ssure-related sequences in other streptococcal species, we isolated the genomic DNA of streptococcal species, including clinically relevant encapsulated pneumococcal serotypes, and other pathogenic bacteria (Table S2). Hybridization of chromosomal DNA with a ssure-specific DIG-labelled DNA-probe (PCR with primers Fnbrep5 and Fnbrep6) using the dot-blot technique revealed the distribution of the pavB gene and its ssure region in all pneumococcal strains tested in this study. The results further confirmed the presence of homologous ssure sequences in other streptococcal species such as S. mitis, S. agalactiae and S. gordonii and confirmed the absence in S. mutans. No signals were detected for the other tested prokaryotic species, suggesting a species-specific nucleotide/protein sequence (Fig. 2).
Molecular analysis of pavB and its ssure domain
To analyse the sequence of pavB and determine the number of ssure in pavB sequences chromosomal DNA of six pneumococcal strains including strains with annotated genomes (TIGR4, G54, D39, R6x, R800 and NCTC10319) was used as template DNA in PCR reactions and Southern blot analysis. PCR analysis of the full-length pavB gene including its flanking sequences (primers Fnbmut1/Fnbmut4; Table S3) and of the ssure region (primers Fnbrep5/Fnbrep6; Table S3) showed PCR product sizes corresponding to five ssure sequences in TIGR4 and G54, while the PCR products for strain D39 suggested the presence of seven repeats (Fig. 3A–D). To confirm these results, the chromosomal DNA of 11 different pneumococcal serotypes was digested with the restriction enzymes EcoRI and SphI and after Southern-blotting hybridized with a radiolabelled pavB DNA-probe. By using the genomic information annotated in databases for TIGR4, G54, D39 and R6x we predicted the restriction sites and calculated the theoretical size of each pavB gene. The results confirmed five ssure in TIGR4 and G54 and seven in strain D39, while pavB of pneumococcal strains 35A, R6 and R800 consist of nine ssure sequences as shown by PCR and Southern blot analysis (Fig. 3D and E). These results demonstrated a discrepancy between the actual number of repeats in the SSURE domain and the number of ssure published in the annotated genomes of TIGR4, G54, D39 and R6x. The predicted numbers of repeats for the other pneumococcal strains evaluated in this study were in the range from 2 to 6. There was no correlation between the number of SSURE and the serotype (Fig. S1 and Table S1). In addition, we aimed to determine the localization of the non-reported repeats in the annotated genome sequences of TIGR4 and G54. For that reason, ARMS-PCR analyses were conducted, using ssure-specific primers targeting the specific nucleotides previously identified for each repeat in the TIGR4-ssure region (Fig. 3A). Our results confirmed the number of five ssure for TIGR4 and G54 pneumococcal strains and demonstrated the non-reported repeat of TIGR4 between database-reported core-repeats 3 and 4. In G54, the three non-reported repeats constitute the core repeat in its ssure region, while the annotated ssure represent the first and last repeat respectively (Fig. 3F, G and H). These findings were confirmed by DNA sequencing of the complete TIGR4-ssure region cloned in plasmid pQssure1-5 (Table S2), using different primers targeting the pavB gene and/or ssure region (data not shown). The non-reported TIGR4 SSURE amino acid sequence is 99.8% similar to our SSURE-consensus sequence and 100% identical to the reported-TIGR4 SSURE3 (Fig. 3H; Fig. S1).
SSURE peptides interact with fibronectin and plasminogen
Three different TIGR4 ssure DNA sequences were designed to produce His6-tagged fusion proteins and to investigate their ability to interact with host proteins. We were able to produce one SSURE domain, representing the second SSURE of TIGR4 (SSURE2) and two SSURE domains, representing the second and third SSURE of TIGR4 (SSURE2+3). In addition a recombinant PavB protein was produced, termed SSURE1−5, which represents the mature PavB protein of TIGR4 without the membrane anchoring domain (Fig. 4A). Coomassie Brilliant Blue staining of His6-tagged fusion proteins and immunoblot analysis confirmed the calculated molecular masses of 17.5 kDa for SSURE2, 33.7 kDa for SSURE2+3 and 99.4 kDa for SSURE1−5 (Fig. S2). Polyclonal antibodies generated against SSURE2+3 and SSURE1−5, respectively, reacted with SSURE2+3 and for SSURE1−5, but showed only weak reactivity with the single repeat SSURE2, while the Penta-His™ antibody (QIAGEN) reacted equally well with all produced His6-tagged proteins (data not shown). Ligand-overlay assays with human plasma revealed recruitment of fibronectin and plasminogen by all tested SSURE peptides (Fig. 4B). The specificity of fibronectin- and plasminogen-binding by SSURE peptides was investigated by incubating immobilized host proteins with equimolar amounts of SSURE2, SSURE2+3 or SSURE1−5. All three employed SSURE peptides showed fibronectin-binding activity (Fig. 4C) as shown previously for one repeat (Bumbaca et al., 2004). In addition, results revealed also a dose-dependent interaction with plasminogen (Fig. 4C). Remarkably, the complete SSURE domain of PavB consisting of five repeats showed the highest binding to fibronectin, while this domain showed the lowest binding to plasminogen compared with the two other SSURE peptides (Fig. 4C).
Binding of SSURE to fibronectin and plasminogen analysed by surface plasmon resonance
Surface plasmon resonance (SPR) was used to analyse the interaction between His6-tagged SSURE1−5, SSURE2+3 and SSURE2 with fibronectin and plasminogen respectively. Binding of all three SSURE peptides to the immobilized host proteins fibronectin and plasminogen was concentration-dependent and elevated resonance units were measured with increasing concentrations of SSURE peptides. Representative sensorgrams for each interaction are shown in Fig. 5. The SPR analysis indicated the direct interaction of all three employed SSURE peptides with fibronectin and plasminogen. However, a qualitative assessment of these interactions revealed differences for the binding efficiencies of SSURE derivatives to fibronectin and plasminogen. SSURE2 showed the lowest binding efficiency to fibronectin and plasminogen, while SSURE1−5 showed the highest binding efficiency to the immobilized host proteins as shown in the sensorgrams (Fig. 5). In conclusion, PavB displays fibronectin and plasminogen-binding activity, which seems to be dependent on the number of repeats present in the corresponding PavB protein.
Impact of surface-displayed PavB on binding of pneumococci to fibronectin
To assess the impact of PavB on binding of pneumococci to fibronectin, pavB-mutants of non-piliated pneumococci were generated by allelic replacement. The mutants were confirmed by PCR (data not shown) and Southern blot analysis (Fig. S3). Expression of PavB in wild-type strains and its absence in isogenic mutants S.p. 35AΔpavB, D39ΔpavB and nonencapsulated D39ΔcpsΔpavB was also demonstrated by immunoblot analysis. Strikingly, the molecular masses of PavB confirmed our molecular analysis as the size of PavB proteins was in agreement with our molecular analysis and the calculated number of repeats (Fig. 3) in the corresponding pneumococcal strains (Fig. 6A). Immunogold electron microscopy using the specific mouse anti-SSURE1−5 IgG polyclonal antibodies indicated that PavB is displayed on the cell surface of wild-type pneumococci whereas the isogenic pavB-mutants showed no expression of PavB (Fig. 6B). Moreover, immunoblot analysis and electron microscopy as well as flow cytometry (data not shown) suggested a low amount of surface-displayed PavB compared with, e.g. CBP PspC. Binding assays with FITC-labelled pavB-mutants of strain S.p. 35AΔpavB and D39ΔcpsΔpavB revealed that PavB of S.p. 35AΔpavB contributes significantly to pneumococcal binding to immobilized fibronectin (Fig. 6C). However, the impact of PavB seems to be strain-dependent as we were not able to measure differences in the ability of D39Δcps and D39ΔcpsΔpavB to interact with fibronectin (Fig. 6C).
PavB contributes to pneumococcal colonization
The impact of PavB on colonization and virulence was assessed by infecting CD-1 outbred mice on different routes. In the pneumonia model, mice (n = 18) were infected intranasally with 1 × 107 wild-type or isogenic D39ΔpavB pneumococci. In mice infected with the PavB-deficient mutant the survival time of mice was significantly (P = 0.02, log rank test) prolonged (Fig. 7A). However, all mice showed severe signs of illness. With the exception of two and one mice infected with the isogenic pavB-mutant and wild-type, respectively, all mice succumbed (Fig. 7A). We also monitored pneumococcal dissemination from the nasopharynx in the lungs and blood, leading to pneumonia and septicaemia, by challenging six CD-1 mice per group with either bioluminescent wild-type D39lux or its isogenic pavB knockout D39luxΔpavB and using bioluminescent optical imaging. The infection was followed continuously in the same mice until they became moribund. Mice infected with PavB-deficient pneumococci showed a substantial delay of bacterial spread in the lungs (Fig. 7B). Thirty-six hours post-infection ΔpavB-infected mice showed weaks signs of pneumonia while mice infected with the wild-type developed severe lung infections at this time point (Fig. 7B). These results suggested that loss of function of PavB is beneficial for mice at the early stage of infection. To assess the effect of PavB on pneumococcal persistence in the nasopharynx, a nasopharyngeal carriage model was used. Groups of n = 9 mice were challenged intranasally with 1 × 105 D39 wild-type or isogenic pavB-mutant bacteria and nasopharyngeal washes were performed 3 and 7 days post-infection to determine the CFU. The results showed at both time points a significant reduction of carriage for the PavB-deficient mutant compared with the D39 wild-type (P < 0.01 and P < 0.001), confirming the role of PavB in colonization (Fig. 7C). As the optical imaging and carriage model showed a delay of pneumococcal transmigration into the lungs, we also aimed to determine the effect of PavB on colonization in another experimental setup. Groups of n = 9 mice were therefore co-infected intranasally with equal amounts (2 × 106 CFU) of both the mutant strain D39ΔpavB and its parental strain D39. The number of wild-type relative to the number of mutant bacteria was calculated 3, 6, 12 and 24 h post-infection within each individual mouse. The results demonstrated that the wild-type out-competed the PavB-deficient pneumococci. Already 3 h post-infection we have determined higher bacterial loads in the nasopharynx, airways, and lung for the wild-type strain D39 relative to the pavB-mutant strain D39ΔpavB (Fig. 7D). In contrast, PavB-deficiency did not decrease pneumococcal virulence in experimental mouse meningitis. There was no significant difference between wild-type and mutant bacterial outgrowth in cerebellum, spleen and blood 38 h after injection of 1 × 104 wild-type (n = 10) or PavB-deficient (n = 9) pneumococci into the right frontal lobe of C57BL/6 mice (Fig. S4). Furthermore, there was no difference between the two groups concerning the clinical parameters weight, clinical score and performance in the tight rope test (data not shown). Taken together, the mouse infection experiments deciphered that PavB is involved in pneumococcal colonization of the respiratory airways.
Influence of PavB on pneumococcal phagocytosis
To investigate whether PavB protects pneumococci against phagocytosis by macrophages or primary professional phagocytes, wild-type pneumococci and isogenic ΔpavB mutants were incubated for 30 min with J774 macrophages or human polymorphonuclear leucocytes (PMNs). After incubation of pneumococci with J774 macrophages extracellular bacteria were killed by antibiotic treatment and the number of internalized and recovered pneumococci was monitored by quantitative platings. The results showed no significant differences between the employed wild-type strains and their isogenic ΔpavB mutants D39ΔcpsΔpavB or S.p. 35AΔpavB (Fig. S5A). Similar, binding and uptake of PavB-deficient pneumococci by PMNs was also not enhanced compared with the isogenic parental strains D39Δcps and S.p. 35A as measured by flow cytometry (Fig. S5B). These data suggest that the impaired colonization of the lower airways by PavB-deficient pneumococci under in vivo conditions is not the result of a higher phagocytosis and killing rate by professional phagocytes.
PavB mediates adherence of pneumococci to host cells
The role of PavB in pneumococcal adherence to and invasion into host epithelial cells was elucidated by conducting infection experiments using the alveolar lung epithelial cell line A549 and the nasopharyngeal epithelial cell line Detroit 562. First, the impact of PavB on pneumococcal adherence was scored by immunofluorescence microscopy. Inactivation of the gene encoding PavB resulted in a significant reduction of adherence for S.p. 35AΔpavB compared with its isogenic wild-type (Fig. 8A and B), while adherence of D39ΔcpsΔpavB compared with D39Δcps was not impaired (Fig. 8B). Here it is important to mention that the number of adherent D39Δcps is relatively low compared with wild-type S.p. 35A (Table S4). The inability of D39 strains to adhere efficiently to host cells impedes the assessment of adhesive functions of proteins produced by D39 using cell culture infections as shown previously (Hammerschmidt et al., 2005; Pracht et al., 2005; Rennemeier et al., 2007). To measure quantitatively the effect of PavB on pneumococcal invasion, antibiotic protection assays were performed. Similar to our adherence assays the data revealed a significant reduction of recovered intracellular pneumococci for the pavB-mutant S.p. 35AΔpavB compared with its isogenic wild-type strain (Fig. 8C), suggesting that the impaired adherence of a PavB-deficient strain caused a reduction of pneumococci uptake by host cells (Fig. 8D).
Virulence studies in mice and adherence data suggested that PavB is directly involved in pneumococcal adherence to host cells (Figs 7 and 8). To investigate the ability of exogenously added recombinant PavB to block pneumococcal adherence by interacting directly with host cellular receptors, epithelial cells were infected with pneumococci in the presence of different amounts of the His6-tagged SSURE1−5 peptides representing a mature PavB protein. The results indicated that pneumococcal adherence to A549 cells is inhibited in a dose-dependent manner by our recombinant PavB protein (Fig. 8E). In conclusion, our adherence and invasion assays suggested that PavB acts as an adhesin and modulates attachment of pneumococci to respiratory epithelial cells.
PavB interacts directly with host epithelial cells
Binding studies were performed to confirm that PavB interacts directly via SSURE domains with host epithelial cells. SSURE2+3 and SSURE1−5 were used as representative PavB derivatives and binding of Cy5-labelled PavB derivatives to A549 epithelial cells was analysed by flow cytometry. Results revealed a dose-dependent binding of SSURE2+3 and SSURE1−5 to A549 cells suggesting a specificity of this bacterial adhesin-host cell interaction (Fig. 9). When using 10 µg of the recombinant PavB derivatives, approximately 95% of the cells were positive for SSURE2+3 and 65% were positive for SSURE1−5 respectively (Fig. 9). Moreover, SSURE2+3 was coated onto 3 µM fluorescent microspheres and binding of these SSURE-beads to A549 cells was quantified by flow cytometry. Similar to Cy5-labelled SSURE peptides, SSURE-coated microspheres interacted specifically with epithelial cells (data not shown).
PavB is immunogenic
PavB is produced by all strains tested here and is displayed on the pneumococcal cell surface. To test whether PavB represents a promising candidate for vaccine development, we have analysed the immunogenicity of PavB under relevant in vivo conditions using seven convalescent-phase sera of patients suffering from pneumococcal diseases. The immunogenic CBPs PspC and PspA were used as control proteins (Kadioglu et al., 2008). Strikingly, all sera showed reactivity against our PavB derivative SSURE2+3, suggesting that PavB represents an immunodominant pneumococcal surface protein and conserved vaccine antigen (Fig. 10).
The ability of pneumococci to adhere to host epithelial cells of the respiratory tract and to evade the host immune system are prerequisites to initiate transient persistence in the nasopharynx and subsequent transmigration into the lungs or bloodstream (Hammerschmidt, 2006; Paterson and Mitchell, 2006; Kadioglu et al., 2008). Pneumococci are encased by a capsular polysaccharide (CPS) which protects against uptake by phagocytes and thus, allows pneumococci to survive and multiply during colonization and bacteraemia (Magee and Yother, 2001). However, high amounts of CPS impede efficient adherence of pneumococci to host cells, which is preferentially mediated by surface-displayed proteins (Hammerschmidt et al., 2005). Adhesive properties have been demonstrated in particular for several CBP such as the PspC, pneumococcal pili, and anchorless adhesins such as PavA (Hammerschmidt, 2006; Telford et al., 2006; Nobbs et al., 2009). Moreover, recent studies demonstrated that PsrP and PfbA act directly as adhesins. The serine-rich repeat protein PsrP was shown to bind to keratin of lung epithelial cells and the plasmin- and fibronectin-binding protein PfbA was shown to mediate pneumococcal adherence to host cells (Yamaguchi et al., 2008; Shivshankar et al., 2009).
In this study, we demonstrated that PavB is important for pneumococcal colonization and virulence, most likely by acting directly as an adhesin and thereby mediating the attachment of pneumococci to a so far unknown host cellular receptor. PavB was independently identified by screening annotated genomes of several pneumococcal strains for classical surface proteins possessing an LPxTG anchoring motif and by screening a D39 whole-genome λ-display library with sera from infected individuals respectively (Bumbaca et al., 2004; Beghetto et al., 2006). The mature protein contains a variable number of repetitive sequences and one repeat, referred to as SSURE, showed fibronectin-binding activity (Bumbaca et al., 2004).
First, we conducted a detailed molecular analysis of the pavB locus using different pneumococcal strains. The number of SSURE in pneumococcal PavB proteins whose sequences were available in databases varied from two to six SSURE. The results of our molecular analysis of pavB and its ssure region in G54, TIGR4, D39 and R6 revealed significant discrepancies between the numbers of repeats reported for pneumococcal pavB sequences and our experimental data. The initial molecular analysis of the SSURE domain of strain R6 revealed eight repeats while its progenitor D39 seemed to contain only six repeats. Importantly, the full-length pavB sequence could not be determined by primer walking or standard sequencing methods and a placeholder sequence was therefore inserted to determine the sequence (Lanie et al., 2007). By employing various molecular techniques including ARMS-PCR we demonstrated the exact number of ssure in pavB of 11 pneumococcal strains. In addition, we determined the location of non-reported ssure in pneumococcal strains TIGR4 and G54. For example, we calculated the number of repeats in D39 and R6 to be seven and nine respectively. In order to solve the discrepancies in the number of pavB-ssure between our results and those reported for D39 (Lilly and NCTC) and R6, we re-analysed the information provided in the study of Lanie and colleagues (Lanie et al. 2007). Remarkably, this initial molecular analysis is in agreement with our calculated number of repeats. They obtained PCR products of 4.7 kb for D39 and 5.6 kb for R6, using the primers III-F-036 and III-R-042 (supplemental file S1) (Lanie et al., 2007). The identification of the primer targets in our pavB-ssure model (III-F-036: 462 bp upstream of the first ssure, and III-R-042: 1109 bp downstream of the last ssure) indicated that their PCR product sizes fit perfectly with seven-times 450 bp (7 ssure) in D39 and nine-times 450 bp (9 ssure). The pneumococcal genomic region, in which the pavB gene is present, is furthermore of high interest concerning the gene conservation in pneumococci and other streptococcal species.
We further demonstrated that the pavB gene is present in all tested pneumococcal strains and confirmed the presence of pavB orthologue sequences in other closely related streptococcal species such as S. agalactiae, S. gordonii, S. mitis (Jedrzejas, 2007). Strikingly, the sequences and genes located upstream of pavB are variable and involved in diverse metabolic processes of streptococci such as transport and DNA synthesis. In contrast, the pavB or its orthologous genes correlates with a downstream-located two-component regulatory system 08 (TCS08), which is involved in pneumococcal virulence and cellobiose metabolism (Throup et al., 2000; McKessar and Hakenbeck, 2007). Our analysis revealed that in the other closely related streptococci TCS08 homologous genes were also found downstream of pavB orthologues. A loss of function of TCS08 through polar effects caused by our mutagenesis strategy was excluded by RT-PCR (Fig. S6). Although there is no experimental evidence up to now, the pavB-TCS08 gene organization suggests that TCS08 contributes to the regulation of pavB gene expression. Hence, the observed attenuation of Δtcs08-mutants in a respiratory tract infection model can also be, at least in part, due to the altered pavB gene expression (Throup et al., 2000).
A key virulence strategy of Gram-positive pathogens and pneumococci is their ability to interact with proteins of the human serum or ECM. These interactions are associated with pneumococcal immune evasion or adherence to and invasion into host cells (Hammerschmidt, 2006; Kadioglu et al., 2008; Nobbs et al., 2009). For example, pneumococci bind to host cell-bound vitronectin and exploit the vitronectin-αvβ3-integrin complex as a cellular receptor for invasion (Bergmann et al., 2009). However, pneumococci bind also to fibronectin and PavA and one SSURE of PavB were shown to adhere directly to immobilized fibronectin (van der Flier et al., 1995; Bumbaca et al., 2004). The ability of SSURE peptides to interact with fibronectin was confirmed in our binding and SPR studies. In addition, we demonstrated plasminogen-binding activity of SSURE domains. Although the three employed SSURE peptides showed fibronectin- and plasminogen-binding activities, the efficiencies varied dependent on the number of repeats in the peptides and host protein used. Our SPR data suggested that the affinity between PavB or SSURE domains and fibronectin or plasminogen increases with the number of repeats. Several prokaryotic and eukaryotic proteins are composed of repetitive structures and it is assumed that the number of repeats influences protein activity and function. Proteins carrying repeats are thought to have an advantage in adaptation processes contributing to rapid evolutionary changes when environmental conditions are altered (Alba et al., 2007).
The relevance of PavB in pneumococcal binding to immobilized human fibronectin was shown in binding experiments. The results, however, suggested that the impact of the surface-exposed PavB on host protein binding depends on the pneumococcal strain. These results are in agreement with the hypothesis that in addition to PavA and PavB other fibronectin-binding proteins are produced by pneumococci (Holmes et al., 2001). Pneumococcal recruitment of plasminogen, which is the proenzyme of the serine-protease plasmin, is essential for ECM degradation and transmigration (Bergmann et al., 2005). The initial study identified eight putative plasminogen-binding proteins with a maximal molecular mass of 83 kDa and no binding protein with the molecular mass of PavB; however, the key co-factor in this scenario is the surface-displayed pneumococcal enolase (Bergmann et al., 2001). The amino acid residues critical for pneumococci-plasminogen interactions are lysine residues of the bacterial adhesins (Bergmann et al., 2005; Cork et al., 2009). Similar to other bacterial plasminogen-binding proteins, SSURE domains of PavB contain lysine-rich and charged sequences that may contribute to plasminogen binding (Bergmann and Hammerschmidt, 2007). Functional analyses of these protein–protein interactions and structural analyses of host-protein-SSURE complexes are required to elucidate these interactions in more detail. Initial structural predictions suggested a model in which the C-terminal part of one SSURE domain interacts with type III repeats located in the C-terminal region of fibronectin. This is consistent with the idea that pneumococci bind to the C-terminally located heparin-binding site of fibronectin and provides a new model as the typical signatures of fibronectin binding proteins of other Gram-positive pathogen are neither present in PavB nor in PavA (Schwarz-Linek et al., 2004; Jedrzejas, 2007).
Several pneumococcal surface proteins are involved in pneumococcal pathogenesis and the initial step of the infection, namely colonization (Bogaert et al., 2004; Bergmann and Hammerschmidt, 2006; Kadioglu et al., 2008). Mice infection experiments showed that PavB contributes to colonization of the nasopharynx and respiratory airways. Moreover, cell culture infection experiments with pneumococci confirmed a role of PavB for pneumococcal adherence to host epithelial cells. However, when comparing pneumococcal strains D39Δcps and D39ΔcpsΔpavB, we were not able to show that the presence of PavB promotes adherence of D39 under these in vitro conditions, while mice infections clearly showed that PavB expression increases the ability of the D39 strain to colonize the lower respiratory tract. These reduced amounts of PavB-deficient pneumococci in the respiratory airways is not caused by an enhanced uptake of the mutant by professional phagocytes as the phagocytosis rates of wild-type and pavB-mutants were similar. The results therefore point to the variable effects of pneumococcal virulence factors on processes such as host protein binding, colonization of mice and different strategies to adhere to eukaryotic cells (Henriques-Normark et al., 2008; Kadioglu et al., 2008). Moreover, colonization and invasion is a multifactorial event and hence, the bacterial virulence factor repertoire as well as the availability of host receptors determines the infectious process.
Blocking experiments and direct binding studies with soluble SSURE peptides and epithelial cells suggested that PavB functions as an adhesin. Hence, PavB seems to interact directly with a cellular receptor similar to the pneumococcal adhesins PspC and PsrP (Zhang et al., 2000; Shivshankar et al., 2009). To date, we cannot predict or exclude a tissue tropism of PavB as has been indicated for PspC and PsrP, as the cellular receptor is not yet known. The course of infection in mice suggests that PavB is of minor importance for the spread of pneumococci once they have breached the epithelial barrier. Contrary to PavB, the deficiency in pneumolysin or PavA attenuated pneumococci in mice infection models of bacteraemia and experimental meningitis (Wellmer et al., 2002; Pracht et al., 2005).
The presence of the gene encoding PavB was indicated in all strains tested so far and moreover, PavB and its SSURE domains are highly conserved among the strains. The SSURE domains of PavB were shown to be highly immunogenic by using convalescent-phase sera. However, it remains to be determined whether the immunogenicity of PavB is sufficient to elicit protective antibodies, which is important to consider PavB as a promising candidate for a multi-component vaccine.
In conclusion, PavB represents an important virulence determinant of pneumococci with multiple functions. In addition to its fibronectin- and plasminogen-binding activity PavB is a novel adhesin of pneumococci promoting colonization of mice and mediating directly adherence to eukaryotic cells via an interaction of SSURE domains with eukaryotic cells.
Bacterial strains, plasmids, growth conditions and transformation
Bacterial strains and plasmids used in this study are listed in Table S2. Pneumococci, other streptococci and Candida albicans were grown on blood agar or in Todd-Hewitt broth (Oxoid, Basingstoke, UK) supplemented with 1% yeast extract (THY) at 37°C and 5% CO2. Legionella were grown on active coal agar (Rennemeier et al., 2007). Escherichia coli strains were grown on Luria–Bertani (LB) agar or LB broth. Streptococcus pneumoniae strains were transformed as described previously (Hammerschmidt et al., 2007) using competence-stimulating peptide-1 and cultivated in the presence of the appropriate antibiotics: erythromycin (5 µg ml−1) and/or kanamycin (200 µg ml−1). Transformation of E. coli strains with plasmid DNA was carried out with CaCl2-treated competent cells according to standard procedures.
The pneumococcal pavB locus and its 1 kb upstream and 2 kb downstream flanking sequences from 19 available pneumococcal genome sequences were retrieved from the CMR (http://cmr.jcvi.org/tigr-scripts/CMR/shared/Genomes.cgi) of the J. Craig Venter Institute (JCVI) and/or from the GenBank (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi) of the National Center for Biotechnology Information (NCBI) genome sequence databases. A genomic pavB gene region comparison was performed by using the complete microbial genome cart available in the CMR website and the complete list of all genomic BLAST databases available in the NCBI website. The TIGR4 pavB gene (SP0082) was taken as a reference to find significant sequence similarities in other (micro-)organisms. A multiple inter- and intra-strain pavB gene and PavB protein sequences alignment was constructed by using the online multiple sequence alignment (http://bioinfo.genotoul.fr/multalin/multalin.html) programme (Corpet et al., 2000). An intensive analysis of the genomic ssure region and the protein SSURE domain was carried out by using also the multi-alignment programme to perform multiple inter and/or intra-strain SSURE nucleotide and amino acid sequences comparisons.
Primers and molecular techniques
Primers that were used in this study are provided in Table S3. Bacterial chromosomal DNA isolation and purification was performed using a standard Phenol/Chloroform extraction method as described previously (Hammerschmidt et al., 1997) or the QIAGEN Genomic Tip 100/G and according to the instructions of the distributor (Qiagen, Hilden, Germany). DNA amplification was performed by high-fidelity PCR using the Expand Long Template PCR system (Roche, Germany) or the AmpliTaq Gold® DNA Polymerase was used as recommended by the distributor (Perkin-Elmer). PCR amplifications were carried out in 50 µl of volumes and reactions were subjected to denaturation at 94°C, 32 cycles of 94°C, primer annealing for 1 min, and elongation at 72°C. Annealing temperature depended on the primers used and extension time on the length of the expected PCR product. PCR products were purified using the PCR DNA purification kit (Qiagen) and plasmids were extracted according to the QIAprep Spin Midi/Maxiprep Kit protocol. DNA sequencing was carried out by using the Qiagen-sequencing services. Amplification Refractory Mutation System (ARMS-PCRs) were performed by using special 3′-ended primers designed to be TIGR4 ssure-specific (Table S3 and Fig. 3A). Southern blot analysis was performed under stringent conditions in 4.8× SSC, 1× Denhardt's solution, 10 mM Tris-HCl (pH 7.5), 10% (w/v) dextran sulphate, 1% (w/v) SDS and 50% formamide at 42°C as described (Hammerschmidt et al., 1997). To generate DIG-labelled DNA probes, the PCR DIG labelling mix (Roche) was used in the PCR reactions together with primer pair Fnbmut1/Fnbmut2 which was used to amplify a region between nucleotides −308 and nucleotide +103 of the TIGR4 pavB gene to confirm the gene knockout in pavB. A radiolabelled DNA probe was used to study the presence of pavB in selected pneumococcal strains by Southern blot analysis and for hybridizations with PCR products. Primers Fnbmut1 and Fnbmut4 were used to amplify pavB including upstream (308 nt) and downstream (303 nt) regions by PCR. End-labelling of the PCR product with 5 µl of [α32P] dATP (3000 Ci/mmol; Amersham, Buckinghamshire, UK) and Klenow fragment was carried out with the Random Primers DNA Labelling System® according to the instructions (Invitrogen, Karlsruhe, Germany). To further study the distribution of pavB in pneumococcal strains and to identify pavB homologues in other microorganisms the DNA dot spot technique was used. Purified chromosomal DNA (∼1 µg) was immobilized onto a nylon membrane (Macherey & Nagel, Düren, Germany). After UV fixation (UV-Stratalinker 1800, Stratagene, La Jolla, CA) and heating at 80°C for 2 h, DNA-DNA hybridization was performed under stringent conditions using the DIG-labelled PCR product amplified with primers Fnbrep1 and Fnbrep4 (surre2+3) and TIGR4 chromosomal DNA as template. Detection of the DIG-labelled DNA hybrids was carried out according to the manufacturer's instructions (Roche).
Recombinant protein purification, SDS-PAGE, generation of anti-SSURE polyclonal antisera and immunoblotting
The TIGR4 DNA sequences coding for three representative SSURE domain fragments (Fig. 4A and Fig. S2) were PCR amplified and the products were cloned into the pGEM-T Easy vector (Promega). The three different ssure fragments (ssure1−5, ssure2+3, ssure2) were subcloned by BamHI/PstI digestion into similarly digested pQE30 expression vector and results in plasmids pQssure1−5, pQssure2+3 and pQssure2 respectively (Table S2). Integrity of the insert DNA was confirmed by DNA sequencing. For protein production these plasmids were transformed into E. coli M15[pREP4] and the expression of the recombinant His6-tagged proteins was induced with 2 mM IPTG. The His6-tagged proteins were purified by affinity chromatography using nickel-nitrilotriacetic acid resins (Macherey-Nagel, Düren, Germany) or Hi-Trap Ni-NTA columns (GE Healthcare) according to the instructions of the manufacturers. In a further purification step proteins were purified by anion exchange chromatography using a MonoQ HR5/5 column (GE Healthcare). Purity of the proteins was analysed by both Coomasie brilliant-blue (CBB) staining and immunoblotting. Detection of recombinant PavB derivatives or PavB in pneumococci was carried out by using rabbit anti-SSURE antiserum (IG-521), mouse anti-SSURE1−5 antiserum or Penta-His™ antibodies (QIAGEN; 1:500 in PBS). The polyclonal anti-SSURE antiserum, recognizing specifically one SSURE of PavB, was raised in rabbits by routine immunogenic procedures using His6-tagged-SSURE2 as antigen (Immunoglobe, Himmelstadt, Germany). In addition, we have generated antibodies against PavB in mice (female Balb/c) by using SSURE1−5 as antigen. Mice were immunized subcutaneously with 20 µg of the recombinant protein in 200 µl of 1:1 emulsion of buffer and incomplete Freund's adjuvant (Sigma). Mice were boosted at day 14 and bled after 5 weeks. Purification of rabbit anti-SSURE2 antiserum was performed by affinity chromatography using protein A-Sepharose 4B affinity chromatography followed by a His HiTrap™ chelating HP column coupled with recombinant SSURE2+3 (GE Healthcare). Purified proteins or bacterial lysates were subjected to SDS-PAGE and either stained with Coomassie Brilliant Blue or transferred by to a nitrocellulose membrane by using a semidry blotting system (Bio-Rad). The membrane was blocked with 10% skim milk (Roth) and PavB was detected by immunoblot analysis. The polyclonal anti-SSURE2 IgG was used at a dilution of 1:200 and mice-derived polyclonal anti-PavB antiserum was used at a dilution of 1:1000. As secondary antibodies, goat anti-rabbit IgG peroxidase conjugate (Dianova; 1:5000) or goat anti-mouse Ig peroxidase conjugate (Dianova; 1:5000) were used and binding activity was detected by enhanced chemiluminescence (ECL; GE Healthcare). Convalescent-phase sera were kindly provided by Gregor Zysk, University of Düsseldorf, Germany (Zysk et al., 2000).
His6-tagged SSURE1−5, SSURE2+3 and SSURE2 were spotted on nitrocellulose membranes and blocked with 5% skim milk (Roth) in PBS (10 mM) prior to incubation with human plasma obtained from healthy individuals upon informed consent. After extensive washing with PBS Tween 20 (0.05%), the membrane was incubated for 1.5 h with rabbit polyclonal anti-fibronectin antibodies (1:800; DakoCytomation) or goat anti-plasminogen antibodies (1:500; Affinity Biologicals, Ontario, Canada). Detection of fibronectin or plasminogen was performed using HRP-conjugated goat anti-rabbit IgG (1:5000; Dianova) or peroxidase-labelled anti-goat antibodies (1:1000; Sigma) followed by ECL. Binding of His6-tagged PavB derivatives SSURE1−5, SSURE2+3 or SSURE2 to purified fibronectin or plasminogen was assayed in an ELISA format. Purified human fibronectin (0.5 µg per well; Invitrogen) or human plasminogen (0.5 µg per well; Haemochrom Diagnostica, Germany) in a final volume of 50 µl of PBS was immobilized overnight at 4°C in 96-well Maxisorp microtiter plates (Nunc, Wiesbaden, Germany). After blocking with 2% bovine serum albumin (BSA) for 4 h and three washes with PBS different amounts of His6-SSURE proteins were applied for 1 h at 37°C. After washing with PBS to remove unbound SSURE proteins, wells were incubated for 1.0 h with Penta-His™ antibodies (QIAGEN; 1:600 in 1%BSA/PBS) followed by HRP-conjugated anti-mouse antiserum (1:2000; Dianova) and an incubation with the HRP substrate solution ABTS (2,2′-Azino-bis [3-ethylbenzothiazoline-6-sulfonic acid]-diammonium salt) and H2O2. Absorbance at 405 nm was measured using a multidetection microplate reader (Fluorstar Omega, BMG Labtech, Offenburg, Germany).
Surface plasmon resonance
The direct protein–protein interactions between fibronectin or plasminogen and His6-tagged SSURE1−5, SSURE2+3 or SSURE2 proteins were analysed by SPR using a Biacore T100 optical biosensor. Covalent immobilization of fibronectin and plasminogen, respectively, on a carboxymethyl dextran (CM5) sensor chip was performed by a standard amine-coupling procedure as described (Bergmann et al., 2003). Briefly, fibronectin (100 µg ml−1 in 10 mM sodiumacetate, pH 4.0) and plasminogen (10 µg ml−1 in 10 mM sodiumacetate, pH 5.5) were coupled at 10 µl min−1 onto N-hydroxysuccinimide (NHS, 0.05 M)/N-ethyl-N′-(diethylaminopropyl)carbodiimide (EDC, 0.2 M) – activated sensor chips (70 µl, 10 µl min−1). Control flow cells were prepared in the same way but without injecting the protein. Binding of the analytes was performed in HBS-EP+ (pH 7.4) at 25°C using a flow rate of 30 µl min−1 in all experiments. The affinity surface was regenerated between subsequent sample injections of SSURE1−5, SSURE2+3 proteins with 10 µl of 5 mM NaOH. Each interaction was measured at least three times. The sensorgrams show the resonance units (RU) values after subtraction of the blank run and value(s) without SSURE protein(s) from the corresponding sensorgrams.
Generation of pneumococcal mutants
Pneumococcal mutants in D39, D39Δcps (Rennemeier et al., 2007) and NCTC10319 were generated by insertion deletion mutagenesis. Briefly, the 5′end (a 411 bp DNA fragment: nt −308 to nt +103 of pavB) and the 3′end (a 417 bp DNA fragment: 114 bp of pavB and 303 bp downstream) of pavB were amplified by PCR with primer pairs Fnbmut1/Fnbmut2 and Fnbmut3/Fnbmut4, respectively, in which restriction sites were incorporated (Table S3). PCR products were separately cloned into pGEM-T Easy resulting in pGEM315 and pGEM316. DNA-inserts were subcloned sequentially into pQE30 (Qiagen) and the erythromycin resistance gene cassette (ermB) amplified by PCR and digested with PstI was cloned into the PstI restriction sites, which was generated between 5′end and 3′end of pavB during our cloning procedure. The integrity of the ermB was verified by sequence analysis. Transformation of pneumococci was performed as described previously and erythromycin (5 µg ml−1) was added to select the mutants, which were verified by Southern blot analysis (Fig. 6A). This strategy resulted in PavB-deficient but TCS08 positive pneumococci as verified by immunoblot analysis using a mouse anti-TCS08 response regulator-specific polyclonal antiserum (data not shown). To obtain bioluminescent wild-type D39 or D39ΔpavB, pneumococci were transformed with the genomic DNA from the bioluminescent strain Xen10 (A66.1 serotype 3) which was purchased from Xenogen (Xenogen Corporation/Caliper Life Sciences, CA, USA). S. pneumonaie-Xen10 possesses a stable integration of the pAUL-A Tn4001 luxABCDE Kmr at a single integration site in the chromosome and colonies show a bright bioluminescence. Transformation of D39 or its isogenic pavB-mutant with the Xen10 genomic DNA resulted in stable bioluminescent derivatives bearing the pAUL-A Tn4001 luxABCDE Kmr at the same integration sites as in Xen10. The bioluminescent derivatives D39lux and D39luxΔpavB, respectively, showed no distinct growth behaviour or properties compared with the parental strains (data not shown).
Binding of pneumococci to immobilized fibronectin
Human fibronectin (0.5 µg per well; Invitrogen) was applied to a 96-well microtiter plate (polystyrene surface; Nunc) at 4°C overnight. The surfaces of the wells were subsequently blocked with 2% BSA for at least 3 h at room temperature. Labelling of the bacteria with fluorescein-isothiocyanate (FITC) was performed as described previously (Rennemeier et al., 2007). Extensively washed FITC-labelled pneumococci (2.0 × 108) were added to the wells and incubated for 1 h at 37°C for binding. Fluorescence was measured at 485 nm/520 nm (excitation/emission) using a multidetection microplate reader (Fluorstar Omega, BMG Labtech, Offenburg, Germany). Measurements were done prior to washing steps with PBS (total fluorescence of inoculum) and after three washing steps with PBS (fluorescence of bound bacteria).
Mouse models of infection and carriage
Female outbred CD1 mice (Harlan Winkelmann, Borchen, Germany; Charles River, Sulzfeld, Germany) were used in infection experiments. Mice were 8–10 weeks old when infected and weighted 26–30 g. Pneumococci were grown in THY (Todd-Hewitt broth supplemented with 0.5% yeast extract) to A600 = 0.35, washed once with PBS and diluted in PBS to get infection doses of 105, 106 or 107 CFU per 25 µl. Before intranasal infection, mice were lightly anesthetized by intraperitoneal injection of ketamine (Ketanest S; Pfizer Pharma, Karslruhe, Germany) and xylazine (Rompun®; Provet AG, Lyssach, Germany). Once anesthetized the animals were scuffed, with the nose held upright, and the bacterial suspension of 25 µl was administered intranasally by adding a series of small droplets into the nostrils for the mice to involuntarily inhale. The infection dose was confirmed by viable count after plating out on blood agar plates following infections. Determination of CFU in the nasopharynx, bronchi, lung, blood or brain was performed after intranasal infection at pre-chosen time intervals (6 h, 12 h, 24 h and 36 h) post-infection and as described previously (Hermans et al., 2006). Briefly, mice were sacrificed and the nasopharynx or lung was washed after dissection of the trachea. One millilitre of sterile PBS was passed through the nasopharynx or inserted into the lung with the inserted trachea cannula, recovered, and viable counts were determined by serial dilution in sterile PBS and plating onto blood agar plates (Oxoid, Basingstoke, UK). Lung, brain and spleen tissue sampling were carried out as previously described (Kerr et al., 2002). Organs were removed separately into 2.5 ml of sterile PBS, weighed, and then homogenized via an Ultra-Turrax® T10 homogeniser (IKA, Germany). Blood samples were obtained by cardiac puncture. Viable counts in homogenates or blood were determined by serial dilution in sterile PBS and plating onto blood agar plates (Oxoid, Basingstoke, UK). For competition infection experiments, wild-type and mutant bacteria were mixed at a 1:1 ratio each with 2 × 106 CFU. Sampling was performed as described above, and the output of mutant versus wild-type bacteria was determined by selection on antibiotic blood agar plates containing kanamycin and/or erythromycin. The competitive index (CI) was calculated as the ratio of mutant to wild-type output CFU divided by the mutant to wild-type input CFU. A denominator of 1 indicates identical input CFU of wild-type and mutant bacteria and hence, CI values lower than 1 indicated a higher output of wild-type bacteria. The experimental mouse meningitis model was performed as described (Wellmer et al., 2002). Briefly, C57BL/6 mice were anesthetized with ketamine (100 mg kg−1 of body weight) and xylazine (10 mg kg−1 of body weight) and infected with 10 µl of 0.9% NaCl containing 104 CFU of bacteria in the right frontal lobe. Mice were followed up at 12, 24, 32 and 36 h after infection by weighing, tight rope test and a clinical score (Wellmer et al., 2000). To determine bacterial in vivo growth rates, mice were sacrificed by decapitation 38 h after infection. Blood was collected, cerebellum and the ventral half of the spleen were homogenized in 0.9% NaCl. Bacterial titres were determined by quantitative plating. All experiments were approved by the Animal Care Committees of the Universities of Wuerzburg, Munich, and Goettingen and by the District Governments of Lower Franconia, Wuerzburg, Upper Bavaria, Munich, and Lower Saxony, Braunschweig, Germany.
Bioluminescent image analysis of pneumococcal infections
Dissemination of pneumococci after intranasal infection of female outbred CD1 mice was observed in real time using the IVIS® Lumina Imaging System (Xenogen Corporation; part of Caliper Life Sciences). Six mice per group were anesthetized and infected with a dose of 1 × 107 bacteria as described above. At pre-chosen time intervals mice were imaged for 1 min post infection to monitor dissemination of pneumococci into the lungs. In addition, bioluminescent intensity (BLI) was determined by quantification of the total photon emission using the LivingImage IgorPro 4.0 software package (Xenogen Corporation).
Immuno labelling for electron microscopy
Pneumococci (50 ml of cultures) were grown in THY and subsequently centrifuged, washed with PBS and the resulting pellet was redissolved in 1 ml of PBS. 100 µl of this suspension was taken, centrifuged, resuspended in a 1:20 dilution of the polyclonal mouse anti-PavB antibodies and incubated for 1 h at 30°C with occasional shaking. Samples were then washed twice with PBS and incubated with 15 nm gold-nanoparticles coated with protein A/G for 30 min at 30°C, washed twice with PBS and subsequently fixed with 1% formaldehyde. The fixed samples were placed onto butvar plastic films on 300 mesh copper grids, allowed to settle for 5 min, washed with TE buffer (10 mM TRIS, 2 mM EDTA, pH 6.9), distilled water and air-dried. Samples were examined in a Zeiss EM910 at an acceleration voltage of 80 kV at calibrated magnifications. Images were recorded digitally with a Slow-Scan CCD-Camera (ProScan, 1024 × 1024, Scheuring, Germany) with ITEM-Software (Olympus Soft Imaging Solutions, Münster, Germany). Images were corrected for brightness and contrast applying Adobe Photoshop CS 3.
Phagocytosis experiments were performed with human polymorphonuclear leucocytes (PMNs) or murine J774 macrophages. Isolation of PMNs and phagocytosis experiments with PMNs were conducted as described recently (Rennemeier et al., 2007). Briefly, 2 × 105 PMNs were incubated for 30 min in 100 µl of PBS/1% FBS Gold (PAA Laboratories, Coelbe, Germany) with 1.5 × 106 FITC-labelled pneumococci at 37°C. Attachment to and uptake of pneumococci by phagocytes was stopped by a 5 min incubation on ice and subsequent centrifugation. The PMNs were fixed with 200 µl of PBS/1% FBS/1% paraformaldehyde. Binding and uptake of FITC-labelled bacteria was assessed by flow cytometry using a FACSCanto™ (Becton Dickinson). The mean fluorescence intensity (MFI) of the entire PMN population was recorded as a measure for bacterial binding and/or uptake.
The number of viable intracellular pneumococci after phagocytosis by the murine J744 macrophages was quantified by the antibiotic protection assay as described previously (Hermans et al., 2006). Briefly, confluent monolayers in 96-well cell culture plates were infected with pneumococci, and 30-min post-incubation, wells were washed three times with PBS to remove unbound pneumococci. To kill any extracellular bacteria, cells were incubated with 50 µl per well of DMEM (4.5 g l−1 glucose; PAA Laboratories) with 1% FBS containing gentamicin (100 µg ml−1) and penicillin G (100 units per millilitre) for 1 h. Intracellular pneumococci were recovered after washing of the cells by a saponin-mediated lysis (1% w/v) for 10 min at 37°C in 5% CO2. Serial 10-fold dilutions were plated onto blood agar plates to determine released intracellular pneumococci. All experiments were performed at least in triplicate.
Cell culture adherence assays and immune fluorescence microscopy
Adherences assays were conducted with human epithelial cell lines A549 (ATCC CCl-185; type II pneumocytes), and Detroit 562 (ATCC CCL 138; human pharynx carcinoma). These cell lines were cultured and infected as described recently (Bergmann et al., 2009). Briefly, epithelial cells were seeded on glass coverslips (diameter 12 mM) or directly in the wells of a 24-well plate (Cellstar, Greiner, Germany) at a density of 5 × 104 cells per well and cultivated to confluent cell layers with approximately 2 × 105 cells per well. The cells were washed three times with Dulbecco's modified Eagle's medium containing HEPES (DMEM-HEPES, PAA Laboratories) supplemented with 1% fetal calf serum (FCS) and then infected with pneumococci. In a standardized assay we used a multiplicity of infection (MOI) of 25 bacteria per host cell and infections were carried out for 3 h at 37°C and 5% CO2. Thereafter, unbound bacteria were removed by rinsing three times with DMEM-HEPES with 1% FBS. The number of host cell attached pneumococci was calculated by immunofluorescence microscopy. Therefore, the infected host cells were fixed on the glass coverslips with 3% paraformaldehyde and host cell-bound bacteria were stained using a polyclonal anti-pneumococcal antiserum and secondary goat anti-rabbit IgG coupled Alexa-Fluor-488 (green) (Invitrogen). At least 25 cells were counted using a fluorescence microscope (Nikon Eclipse TS100-F) and image acquisition was performed using a confocal laser scanning microscope (Zeiss LSM510 META) and the LSM software. Each bar in the images represents 10 µM. In blocking experiments, the His6-tagged SSURE peptides were added to the infection experiments. All experiments were performed at least three times with two or more replicate wells tested for each experimental setup.
Antibiotic protection assays
Antibiotic protection assays were performed to quantify the total number of ingested pneumococci after infecting the host cells. Briefly, epithelial cells were infected with pneumococci (MOI of 25) and thereafter, the infected and washed host cells were incubated for 1 h with DMEM containing 100 µg ml−1 gentamicin and 100 U ml−1 penicillin G at 37°C and 5.0% CO2 to kill extracellular and non-adherent pneumococci. Invasive and viable pneumococci were recovered from the intracellular compartments of the host cells by a saponin-mediated host cell lysis (1.0% w/v) and the total number of invasive pneumococci was monitored after plating sample aliquots on blood agar plates, followed by colony formation and enumeration. Each experiment was repeated at least three times and results were expressed as mean ± SD.
PavB binding to host epithelial cells
Binding of Cy5-labelled His6-tagged SSURE2+3 and SSURE1−5 peptides was measured by flow cytometry. Cy5-labelled BSA was used as control protein. Cy5-labelling of SSURE peptides (1 mg ml−1) and BSA was performed in 0.1 mM sodium carbonate buffer (pH 9.2) using the Cy5 Mono-Reactive Dye Pack (GE Healthcare) according to the instructions of the manufacturer. Fluorescence of proteins was confirmed via a fluorescence scan of a SDS-PAGE and by measuring the absorbance at 280 nm for protein and 650 nm for the Cy5-dye (NanoDrop 1000). A549 cells were cultured in a six-well cell culture dish to a density of 2–3 × 105 cells and fixed for 1 h with 2% paraformaldehyde in PBS/0.5%FBS. After two washing steps with PBS, cells were incubated for 1 h at room temperature with 0, 5, 10, 25 and 50 µg ml−1 of SSURE peptides or BSA and then washed again four times. The samples were then analysed by flow cytometry using a FACSCaliburΣ (Becton Dickinson). Data acquisition was conducted using the CellQuestPro Software 6.0 and for the analysis the FACSDiva Software 5.0.3 was used. The results of protein binding to host cells are shown as the total fluorescence (geometric mean fluorescence intensity (GMFI) multiplied with percentage of positive gated events).
All data are reported as mean ± SD unless otherwise noted. Results were statistically analysed using the unpaired two-tailed Student's test. Kaplan-Meier survival curves were compared by the log rank test. P-values for bioluminescence measurements were calculated using the unpaired, one-tailed t-test for differences between groups, while differences of one group between days were analysed by the paired t-test. Statistical significance was confirmed by ANOVA analysis with Bonferroni's Multiple Comparison post-hoc test. Nonparametric data of co-infection studies were analysed by the two-tailed Mann–Whitney U-test. A P-value < 0.05 was considered statistically significant.
Sequence data for the full-length pavB gene and PavB protein of TIGR4 are available in the EMBL/GenBank databases under accession number FN547057.
This work was supported in part by the Deutsche Forschungsgemeinschaft (Ha 3125/4-1 and GRK840) and EU FP7 CAREPNEUMO. Gustavo Gámez was partially supported by the German Academic Exchange Service (DAAD) and University of Antioquia (U de A), Medellín, Colombia. We are also grateful to Drs Ivo Steinmetz, Thomas Pribyl and Gerhard Burchhardt (Greifswald) for critical discussions and Dr Uwe Bierfreund (GE Healthcare) for helpful suggestions in our Biacore studies.