|Home | About | Journals | Submit | Contact Us | Français|
Streptococcus pneumoniae is a mucosal pathogen that grows in chains of variable lengths. Short-chain forms are less likely to activate complement, and as a consequence they evade opsonophagocytic clearance more effectively during invasive disease. When grown in human nasal airway surface fluid, pneumococci exhibited both short- and long-chain forms. Here, we determined whether longer chains provide an advantage during colonization when the organism is attached to the epithelial surface. Chain-forming mutants and the parental strain grown under conditions to promote chain formation showed increased adherence to human epithelial cells (A549 cells) in vitro. Additionally, adherence to A549 cells selected for longer chains within the wild-type strain. In vivo in a murine model of colonization, chain-forming mutants outcompeted the parental strain. Together, our results demonstrate that morphological heterogeneity in the pneumococcus may promote colonization of the upper respiratory tract by enhancing the ability of the organism to bind to the epithelial surface.
Streptococcus pneumoniae (the pneumococcus) is a leading human pathogen responsible for greater than 1 million deaths annually (20). Interaction with its obligate human host, however, is generally limited to asymptomatic colonization of the mucosal surface of the upper respiratory tract (3). A murine model of colonization recapitulates many of the features of carriage in the natural host. Study of the events during murine colonization has shown that the organism persists for many weeks along epithelial surfaces of the upper respiratory tract, and the effect of complement activity on the organism is minimal (4, 15, 26). In contrast, it is known that the complement system is critical for clearance during invasive disease (21). Thus, it is clear that the pneumococcus faces different selective pressures during invasive disease and asymptomatic colonization.
Initiation of colonization and stable colonization of the host requires pneumococci to adhere to the epithelial surface (3, 15). The binding of pneumococci to epithelial cells in culture has been used as a model to define the host and bacterial factors contributing to these interactions. The thick polysaccharide capsule on pneumococci is known to be necessary for virulence; however, in these assays it reduces the ability of the organism to bind to epithelial cells (6, 12, 28). Thus, during invasive disease the pneumococcus must resist clearance by the complement system, while during colonization the organism must efficiently bind to the epithelial surface and resistance to complement is less important.
Like other members of the genus, the pneumococcus grows in chains of variable lengths. Interestingly, the organism was formerly referred to as Diplococcus pneumoniae, because in infected sputum and tissue it typically appears as short diplococci (24). We recently reported that because of their small size, short chains of pneumococci more effectively evade complement deposition and subsequent uptake and killing by professional phagocytes (7). Since opsonophagocytosis is critical for an effective host defense against this encapsulated pathogen, short forms may be selected during invasive disease and account for the diplococcal morphology of pneumococci in clinical samples.
Under many conditions, however, pneumococci also exhibit longer-chain forms (1, 8). We postulated that the greater surface area of longer-chain variants could overcome the inefficient adherence of encapsulated bacteria through multivalent adhesive interactions. As a result, long-chain forms could facilitate colonization, while short-chain forms could promote invasive disease. Here, we examined whether increased chain length enhanced attachment to epithelial cells in vitro and promoted colonization in vivo.
S. pneumoniae strain TIGR4 (type 4 clinical isolate) was grown in semisynthetic casein plus yeast medium (C+Y; pH 6.8) or in tryptic soy (TS) broth. The identification of M1 and other chain-forming mutants of TIGR4 generated using the Mariner transposase was previously described (7). Strain P2266 contains the Mariner transposon in the intragenic region 5′ to SP_0480. Chain formation in the wild type (WT) was induced by the addition of choline chloride (0.125 to 2.0%, final concentration) or erythromycin (Erm; 0.025 μg/ml) to growth medium (7, 23). A comE mutant was generated by transformation of TIGR4 with genomic DNA from a previously described strain followed by serial back-transformation and confirmed by the loss of natural competence (16). The lytA mutant with an in-frame, unmarked deletion was previously described (7). The lytB mutant was generated by transformation with a construct replacing the lytB gene with the erm(B) erythromycin resistance cassette and selection on Erm (1 μg/ml). This construct was generated using overlap extension PCR with the following primers: 5′ to lytB (primer 1, 5′-TGGGAGCTTGCTATGCCTGTGTTCTAAAAG-3′; primer 2, 5′-TCACTCCTTCCATTAACCTTCTTCCTCTGTTCTTATTTA-3′), 3′ to lytB (primer 3, 5′-GTCAAAGGCAATTTCCAATTCTG-3′; primer 4, 5′-GAGGAAATAATTAGTACTATAAGTGAATATGATTTG-3′). The erm(B) gene was obtained with the primers 5′-GAAGGTTAATGGAAGGAGTGATTACATGAACAAAAA-3′ and 5′-TATAGTACTAATTATTTCCTCCCGTTAAATAATAG-3′.
Where indicated, bacterial cultures at an optical density at 620 nm (OD620) of ~0.5 were briefly sonicated (3 s) with a 40TL probe sonicator (Ultra Sonic Power) to disrupt chains.
Nasal airway surface fluid (NASF) was collected from healthy nonsmoking, noncarrier, volunteers, without dilution or chemical stimulation, and insoluble debris was removed by centrifugation at 1,500 × g for 10 min prior to storage at −20°C. Following this treatment, human nasal airway surface fluid (hNASF) showed no growth when plated on TS blood agar. hNASF (200 μl) was inoculated with a 1/20 dilution of TS-grown bacteria (final OD620, 0.06) and incubated at 37°C in a 5% CO2 atmosphere for 6 h.
A549 cells (ATTC CCL-185) were grown in Dulbecco's modified Eagle's medium with l-glutamine, glucose, and sodium pyruvate and supplemented with 10% fetal bovine serum (FBS) along with penicillin and streptomycin (all obtained from Gibco). Cells were grown to confluence and then harvested by using trypsin-EDTA (Gibco) for seeding into 6-well plates.
Confluent monolayers of A549 epithelial cells grown on 25-mm round glass coverslips in 6-well plates were washed 3 times with phosphate-buffered saline (PBS) and maintained in minimal essential medium (with Earle salts and l-glutamine) without antibiotics until use for infection. Bacteria were grown in liquid culture to late log phase (OD620, 0.5) and added to each well at a density of 106 CFU/well after washing with PBS. The bacteria were applied to the monolayers by centrifugation (1,300 × g for 5 min) and then incubated at 37°C for 60 min to allow for adherence. A sampling of the supernatant after centrifugation demonstrated that the strains (WT and chain-forming mutants) used in this assay pelleted at comparable rates.
To determine the percentage of bacteria from the inoculum that were adherent, the monolayer was washed 5 times with PBS, and the epithelial cells and adherent bacteria were lifted off by treatment with 600 μl/well of 0.25% trypsin–1 mM EDTA. Samples were then vortexed and maintained at 4°C before serially diluting and plating on TS medium supplemented with catalase in triplicate. Adherence was calculated as the portion of the inoculum that was adherent to the target cells, as previously described (10).
For quantification of chain lengths, samples (10 μl) were prepared for microscopy on agarose pads (1% PBS) as previously described (2). Samples were analyzed on an Olympus IX81 microscope with a 1003 phase-contrast objective (Olympus UPLFLN). Images were taken using an Andor IXON 887 camera. Quantification of bacterial chain lengths was done by using Image J. The sizes (in pixels) of bacterial chains were determined as previously described (7). Because of high background of the immunofluorescence images, quantification of sizes (in microns) was carried out using the line-measuring tool found in the iVision imaging software (BioVision).
Monolayers of A549 cells infected with S. pneumoniae were washed with PBS and fixed in 2% paraformaldehyde for 30 min at 37°C. Cells were permeabilized in PBT (PBS, 0.1% bovine serum albumin, 0.2% Triton X-100) for 10 min at room temperature, and nonspecific binding was blocked with protein blocking reagent (Thermo) for 10 min at 37°C. Bacteria were detected using polyclonal type 4 antisera (Statens Serum Institut) diluted 1:500 in PBT and incubated overnight at 4°C. Primary antibody binding was detected with a Cy3-conjugated goat anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Laboratories Inc.) that was diluted 1:500 in PBT, and the mixture was incubated at 37°C for 30 min. Cells were washed in PBS and then H2O, counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1:10,000; Molecular Probes, Invitrogen), and mounted using mounting medium (KPL). Imaging was performed on a Nikon E600 Eclipse microscope equipped with a liquid crystal (Micro-Color RGB-MS-C; CRI) and a high-resolution charge-coupled-device (CCD) digital camera (CoolSnap CF; Roper Scientific) with Nomarski optics. Tissue sections from colonized mice were prepared and stained as previously described (19).
Six- to eight-week-old C57BL/6 mice were inoculated intranasally with 1 × 107 CFU of a 1:1 mixture of mutant and WT bacteria in 10 μl of PBS. At 24 h postinoculation, mice were sacrificed and the trachea was cannulated, and 200 μl of PBS was instilled. Lavage fluid was collected from the nares and serially diluted in PBS for quantitative culture (26). The proportion of mutant and WT bacteria was determined by plating on both selective medium (with appropriate antibiotics; supporting growth of mutants) and nonselective medium (supporting growth of mutants and the WT). The input ratio was defined by the inocula, and the output ratio was defined by quantitative culture of nasal lavage fluid. The competitive index (CI) was determined by dividing the mutant/WT ratio in the output by the mutant/WT ratio in the input.
Comparisons between strains or conditions were made using a two-tailed Student's t test (Prism 4; GraphPad Software). Where indicated, the Wilcoxon signed ranks test was performed, and results were compared to an arbitrary mean value of 1.
To examine the role of chain length during colonization, we first considered whether longer-chain forms occur on the mucosal surface of the upper airway. Growth of isolate TIGR4 was characterized in hNASF to model the physiologic growth conditions of colonization. Compared to control growth in nutrient medium, growth in hNASF resulted in a more morphologically heterogeneous population (Fig. 1). Following growth in hNASF, there were both long chains (white arrow) and diplococci (black arrow).
Next, we tested whether long-chain forms, which may arise during colonization, were more adherent to human respiratory epithelial cells. Chain-forming mutants for two cell wall hydrolases, LytA and LytB, and two transposon mutants previously characterized as displaying increased chain length, M1 and P2266, were compared to the WT in bacterial adherence assays using A549 cells in culture (7, 9, 25). The size of a bacterial chains was defined as the two-dimensional area in phase-contrast images and was used as a proxy for chain length. Adherence was increased for each chain-forming mutant tested and was proportional to the average size observed (Fig. 2A, solid bars). Growth of the WT under conditions to induce chain formation, including high concentrations of choline and subinhibitory concentrations of erythromycin, also resulted in increased adherence in proportion to average size (Fig. 2A, open bars). However, the effect on adherence seemed to peak at an ~5-fold increase over the WT, despite the formation of extensive chains when TIGR4 was grown in high concentrations of choline. A TIGR4comE mutant, with an average size similar to the WT, served as a control to ensure that the transformation process used to construct mutants did not affect adherence. The TIGR4comE mutant also demonstrated that the increased adherence associated with long chains was not affected by quorum sensing, since this mutant does not respond to the peptide ComC (also known as CSP [competence-stimulating peptide]) (18). When assessed by immunofluorescence microscopy, mutants that grew in long chains also adhered to A549 cells as long chains (Fig. 2B). Moreover, adherent WT bacteria were often in long-chain forms. To further demonstrate that increased chain length enhanced adherence, bacteria were briefly sonicated to mechanically disrupt chains before application to A549 cells. Reduction in chain length following sonication was confirmed by microcopy. Mechanical disruption had a minimal effect on adherence of the WT, but it significantly reduced adherence of chain-forming mutants and the WT grown under conditions promoting chain formation (Fig. 2C, closed and open bars, respectively).
We then postulated that longer-chain variants among the WT would be more adherent than isogenic shorter forms. To test this hypothesis, after inoculating A549 cells with the WT, one set of wells was washed five times with PBS and another set of wells was left unwashed before determining the average length of adherent bacteria by microscopy and image analysis. There was a shift in the distribution of chain length, with enrichment for longer chains after washing (Fig. 3). Overall, the mean length of adherent bacteria from washed wells was significantly greater (2.4 versus 1.6 μm; P < 0.001). The greater resistance of longer-chain forms to removal from A549 cells by washing confirmed that longer-chain forms are more adherent. Taken together, these data suggest that increased chain length enhances adherence to human epithelial cells in vitro.
To extend our observations in vivo, chain-forming mutants were competed against the WT in a murine model of colonization. By 24 h postinoculation, pneumococci are found along the epithelial surfaces, and this early time point minimizes the selective pressure of innate and adaptive immune responses (15). Each of the three chain-forming mutants tested outcompeted the WT (Fig. 4A, closed symbols). The M1 mutant tested in vitro was highly attenuated in vivo and was not included in this analysis. Controls included cultures of homogenized excised nasal tissue (data not shown). These showed a similar mutant/WT ratio, demonstrating that lavage cultures reflected the tissue-associated bacteria. When tissue sections of nasal spaces of colonized mice were viewed using immunofluorescence, mutants could be seen as long-chain forms on the epithelial surface (Fig. 4B). To confirm that increased chain length was responsible for this competitive advantage, the experiment was repeated for the lytB mutant versus the WT, but prior to infection the inoculum containing both the WT and mutant was sonicated and disruption of chains was confirmed by microscopy. Mechanical disruption was sufficient to eliminate the competitive advantage of the lytB mutant, suggesting that it was the length of the bacterial chains in the lytB mutant that enhanced the ability of this mutant to colonize the upper respiratory tract (Fig. 4A, open symbols).
For the pneumococcus to colonize its human host, it must successfully bind to the epithelium in the upper respiratory tract. Many studies have modeled this event by using epithelial cells in culture, and they have revealed the importance of pneumococcal surface proteins, teichoic acid, and the capsule in attachment (11). Due to the inhibitory effect of the capsule in these assays, these experiments are often performed with unencapsulated strains. In this study, we have demonstrated that pneumococci display morphological heterogeneity when grown in the physiologically relevant growth medium of human nasal airway surface fluid. Our findings with a clinically relevant encapsulated strain describe pneumococcal chain length as an important factor in adherence to epithelial cells in vitro and colonization in vivo.
The initial analysis of adherence to A549 cells was based on viable counts of adherent bacteria following inoculation with strains forming different lengths of chains. Since pneumococcal chains may dissociate to diplococci during the course of such experiments, additional methodologies were used to confirm our findings. Adherent bacteria were visualized using immunofluorescence microscopy. Although adhesive events are relatively rare and difficult to quantify by this technique, bacteria bound to A549 cells were detected and were in the form of longer chains. Moreover, within the morphologically heterogeneous and clonal population of the WT, washing increased the mean length of adherent bacterial chains. This finding provided important evidence that the increased CFU in adherence assays of long chains was not an artifact of chain breakup during the processing.
Many factors contribute to microbial morphology, including the dimensions of a single cell, the completeness of cell division, and the formation of multicellular communities, and each may impact host-pathogen interactions (29). Long filamentous forms arising from reduced septation have been shown to have a survival advantage in other biological systems (14). Filamentous forms have been reported in S. pneumoniae under severe nutrient limitation conditions in vitro during prolonged culture, but these forms do not appear to arise naturally (1). In our study, increased chain length, a consequence of incomplete cell separation, was associated with a moderate (5- to 10-fold) increase in adherence to human epithelial cells and an advantage in colonization of the nasopharynx. Longer forms may enhance adherence by providing for a greater number of adhesive events per particle (13, 29). Longer molecules have also been shown to promote more effective multivalent protein-ligand interactions (17). It appeared that an increase in size beyond >3-fold above the average of the WT conferred no additional adhesive benefit in vitro. Thus, the pneumococcus may generate forms of a length that optimizes the attachment of a subpopulation. Additionally, growth conditions were found to impact the proportion of long-chain variants within the WT. We observed that TIGR4 growth in hNASF, derived from the natural environmental niche of this organism, stimulated the presence of a subpopulation of longer-chain forms. This suggests that this environmental niche may provide the pneumococcus with a signal to stimulate the formation of long-chain variants, which would be more adherent at this anatomical site.
Pneumococcal cell wall hydrolases are important in the separation of daughter cells following replication. LytA and LytB, in particular, have been proposed as essential virulence determinants (22, 25). Our results suggest that the effect of these enzymes on bacterial chain length may confound studies looking at interactions with host cells and in models of infection. We also demonstrated that growth conditions and other genes that alter chain length might also affect virulence phenotypes. The effects of chain length are seldom considered in pathogenesis studies of bacterium-host cell interactions for streptococci. The complete array of bacterial factors regulating separation of daughter cells and chain length are not thoroughly understood for this species. Screening of a library of >7,000 genomic transposon mutants suggested that many mutations were associated with an increase in average size (7). Of note, the phenotype of chain-forming WT variants was not maintained upon serial passage (data not shown), suggesting that chain length may be controlled by a bistable switch, as has been demonstrated for other species (5).
In summary, S. pneumoniae displays morphological heterogeneity with regard to its chain length. Our results indicate that this may be a factor in the ability of this pathogen to establish both commensal and pathogenic interactions with its host by promoting attachment to epithelial cells during colonization (longer chains) and evasion of the immune response (shorter chains), respectively (27).
We thank Mark Goulian from the University of Pennsylvania for technical assistance with performing and analyzing microscopy experiments.
This work was supported by the US Public Health Service, grant number AI38446 to J.N.W. Additional support was provided by the NIH/NIDDK Center for Molecular Studies in Digestive and Liver Diseases (P30-DK050306) and its Molecular Pathology and Imaging core facilities.
Published ahead of print 23 July 2012